Thermoelectric Conversion of Concentrated Solar Radiation and Geothermal Energygeothermal Energy PDF

Diss. ETH No.
20337
THERMOELECTRIC CONVERSION OF
CONCENTRATED SOLAR RADIATION AND
GEOTHERMAL ENERGY
A dissertation submitted to
ETH ZURICH
for the degree of
Doctor of Sciences
presented by
CLEMENS SUTER
MSc ETH ME
born January 30, 1985
citizen of Zrich (ZH)
accepted on the recommendation of

Prof. Dr. Aldo Steinfeld, examiner
Prof. Dr. Anke Weidenkaff, co-examiner
2012
Abstract
In this thesis the thermoelectric conversion of concentrated solar radiation
and geothermal energy for decentralized power generation is investigated. The
report is divided into two sections. In the first part, the conversion of hightemperature heat from concentrated solar radiation is considered. Hightemperature stable perovskite-type oxide demonstrator modules (proof-ofconcept) consisting of two pairs p-type La1.98Sr0.02CuO4 and n-type
CaMn0.98Nb0.02O3 thermoelement legs and sandwiched between two ceramic
Al2O3 plates with surface 30 x 30 mm2 are investigated. The not yet optimized
figure-of-merit is approximately 0.05. Single modules are exposed to direct
concentrated solar radiation with peak radiative flux intensities of 300 kW m-2
and operated at 900 K on the hot side. The maximum solar-to-electricity
efficiency is 0.081%. A heat transfer model coupling conduction, convection
and radiation is implemented and validated in terms of experimentally
measured open-circuit voltages and solar-to-electricity efficiencies. The heat
transfer model shows that 60% of the incident solar radiation is lost due to reradiation and only 20% is conducted through the thermoelement legs. In a next
step, a 1 kWth solar cavity-receiver prototype packed with 18 modules is
developed to capture efficiently concentrated solar radiation and minimize reradiation losses. The solar cavity-receiver was subjected to peak solar
concentration ratios exceeding 600 suns in the aperture and the modules were
operated at maximally 900 K on the hot side. The measured maximum solar-toelectricity efficiency is 0.12%. A heat transfer model is formulated to simulate
the solar cavity-receiver and validated in terms of experimentally measured
open-circuit voltages. Vis--vis directly irradiated modules the cavity
configuration enables a reduction of the re-radiation losses from 60% to 4% of
the solar radiative power input and an increase of the heat conduction through
the thermoelement legs from 20% to 70%. In order to show the potential of the
cavity design, a hypothetical 50 kWth solar cavity-receiver packed with efficient
cascaded dual-stage modules is considered. The cavity is operated with a solar
ii
Abstract
concentration ratio of 1500 suns in the aperture and in a temperature range of

900 1200 K. The cascaded modules consist of three Al2O3 plates with surface
10 x 10 cm2 sandwiching a low- and a high-temperature stage with 2822
thermoelement legs per stage. The thermoelement legs of the low- and the hightemperature stage are based on Bi-Te with a figure-of-merit of 1 and
perovskite-type oxides with a figure-of-merit in the range of 0.36 1.7,
respectively. This yields a maximum module efficiency in the range of 11.7
26%. The heat transfer model for the 1 kWth cavity-receiver prototype is
extended and applied to find the optimum cavity geometry, which is
characterized by a width of 60 cm, a height of 50 cm and an aperture diameter
of 20.6 cm, and has a number of 156 modules. Furthermore, the model is used
to investigate the effect of the maximum module efficiency and the operated
maximum cavity temperature on the solar-to-electricity efficiency. The solarto-electricity efficiency ranges from 7.4% to 20.8%, showing a strong
dependency on the maximum module efficiency.
In the second part of the report, a stack comprising an array of Bi-Tebased thermoelectric converter modules is considered for geothermal energy
conversion. Each module consists of 127 (Bi0.2Sb0.8)2Te3/Bi2(Te0.96Se0.04)3 p/ntype thermoelement pairs, fastened by 30 x 30 mm2 Al2O3 plates. The
thermoelement pairs have a leg cross-section of 1.05 x 1.05 mm2, a figure-ofmerit equal to 1, and a theoretical heat-to-electricity conversion efficiency of
approximately 5%, when the module is operated at the maximum temperature
difference of 200 K. A temperature gradient across the thermoelement legs
within an array is imposed via a Cu parallel-plate heat exchanger adhering to
the Al2O3 plates and operating hot and cold water in counter-flow
configuration. A heat transfer model coupling conduction through the
thermoelement legs with convection to and from the Al2O3 plates is formulated
to investigate the performance of the stack as a function of the following
parameters: hot water inlet and outlet temperatures (313 413 K and 303
393 K, respectively), stack length (300 1500 mm), thermoelement leg length
(0.5 4 mm) and hot/cold channel heights (0.2 2 mm). The simulated opencircuit voltages are compared to those resulting from the temperature
differences computed via CFD. The heat transfer model predicts for an
optimized 1 kWel stack with hot water inlet and outlet temperatures of 413 K
and 313 K, respectively, either a maximum heat-to-electricity efficiency of
4.2% or a minimum volume of 0.0021 m3.
Zusammenfassung
In der vorliegenden Arbeit wird die thermoelektrische Stromerzeugung
von Solarstrahlung und geothermischer Wrme fr dezentrale Anwendungen
untersucht. Die Arbeit ist in zwei Teile gegliedert. Im ersten Teil wird die
Umwandlung von Hochtemperatur-Wrme aus konzentrierter Solarstrahlung
betrachtet. Es werden hochtemperaturfeste Demonstrationsmodule untersucht.
Jedes Modul enthlt 4 Perovskite-artige Oxid-Thermoelemente, welche
zwischen zwei kalten Al2O3-Platte mit einer Flche von 30 x 30 mm2 montiert
sind. Von den 4 Schenkeln bestehen jeweils 2 aus p-dotiertem La1.98Sr0.02CuO4
und 2 aus n-dotiertem CaMn0.98Nb0.02O3. Die noch nicht optimierte Figure-ofMerit ist circa 0.05. Einzelne Module werden direkter konzentrierter
Sonnenstrahlung mit maximalen Strahlungsflssen von 300 kW m-2 ausgesetzt
und mit 900 K auf der heissen Seite betrieben. Der maximale Solar-zu-Strom
Wirkungsgrad ist 0.081%. Ein Wrmebergangsmodell, welches
Wrmeleitung, Konvektion und Strahlung koppelt, wird implementiert und mit
Hilfe experimentell gemessener Leerlaufspannungen und Solar-zu-StromWirkungsgraden validiert. Das Wrmebergangsmodell zeigt, dass 60% der
einfallenden Solarstrahlung infolge Rckstrahlung verloren gehen und nur 20%
durch die Thermoelemente geleitet werden. In einem nchsten Schritt wird ein
solarer 1 kWth cavity-receiver-Prototyp (im Folgenden nur Kavitt
genannt) entwickelt, um die konzentrierte Solarstrahlung effizient zu
absorbieren und die Rckstrahlungsverluste zu minimieren. Die Kavitt wird
Sonnenkonzentrationen von mehr als 600 in der Apertur ausgesetzt und ist mit
18 Modulen bestckt, welche wiederum auf der heissen Seite mit 900 K
betrieben werden. Der gemessene Solar-zu-Strom Wirkungsgrad ist 0.12%. Ein
Wrmebergangsmodell wird implementiert, um die Kavitt zu simulieren,
welches mit Hilfe gemessener Leerlaufspannungen validiert wird. Im Vergleich
zu den direkt bestrahlten Modulen ermglicht das Kavitt-Design eine
Reduktion der Rckstrahlungsverluste von 60% auf 4% der solaren
Eingangsstrahlung, whrend die Wrmeleitung durch die Thermoelemente von
iv
Zusammenfassung
20% auf 70% erhht wird. Um das Potenzial des Kavitt-Designs zu zeigen,
wird eine 50 kWth Kavitt betrachtet, welche mit effizienten, zweistufig
kaskadierten Modulen bestckt ist. Die Kavitt wird mit einer
Sonnenkonzentration von 1500 in der Apertur und in einem maximalen
Temperaturbereich von 900 1200 K betrieben. Die kaskadierten Module
bestehen aus drei Al2O3-Platten mit einer Flche von 10 x 10 cm2 und einer
Nieder- und Hochtemperatur-Stufe mit 2822 Thermoelementen pro Stufe. Die
Nieder- und Hochtemperatur-Stufe besteht aus Bi-Te-Thermoelementen mit
einer Figure-of-Merit von 1 beziehungsweise Perovskite-artigen OxidThermoelementen mit einer Figure-of-Merit im Bereich von 0.36 1.7, was zu
einem maximalen Modulwirkungsgrad im Bereich von 11.7 26% fhrt. Das
Wrmebergangsmodell der 1 kWth Kavitt wird erweitert und angewendet, um
die optimale Kavittsgeometrie zu bestimmen, welche eine Breite von 60 cm,
eine Hhe von 50 cm und einen Aperturdurchmesser von 20.6 cm hat, und mit
156 Modulen bestckt ist. Ausserdem wird der Effekt des maximalen
Modulwirkungsgrades und der maximalen Kavittstemperatur auf den Solarzu-Strom-Wirkungsgrad untersucht. Der Solar-zu-Strom Wirkungsgrad variiert
im Bereich von 7.4 20.8%, und zeigt eine starke Abhngigkeit vom
maximalen Modulwirkungsgrad ab.
Im zweiten Teil der Arbeit wird die Umwandlung von geothermischer
Wrme mittels eines thermoelektrischen Stacks betrachtet, welcher aus Bi-TeModulreihen zusammengesetzt ist. Jedes Modul besteht aus 127
(Bi0.2Sb0.8)2Te3 / Bi2(Te0.96Se0.04)3 p/n-dotierten Thermoelement-Paaren, welche
zwischen 30 x 30 mm2-Al2O3-Platten verbaut sind. Die Thermoelemente haben
eine Querschnittsflche von 1.05 x 1.05 mm2 und eine Figure-of-Merit von 1
mit einem theoretischen Wirkungsgrad von 5%, wenn das Modul mit dem
maximalen Temperaturunterschied von 200 K betrieben wird. Der
Temperaturgradient entlang der Thermoelemente wird durch einen
Wrmetauscher auferlegt, dessen Heiss- und Kaltwasserkanle im
Gegenstromprinzip funktionieren und dessen parallelen Kupferplatten die
Modulplatten berhren. Ein Wrmebergangsmodell wird implementiert,
welches Wrmeleitung in den Thermoelementen mit Konvektion von und zu
den Al2O3-Platten koppelt, um die Leistung des Stacks als Funktion folgender
Parameter zu untersuchen: Heisswasser Einlass- und Auslasstemperatur (313
413 K beziehungsweise 303 393 K), Stack-Lnge (300 1500 mm),
Thermoelement-Lnge (0.5 4 mm) und heisse/kalte Kanalhhe (0.2 2 mm).
v
Die simulierten Leerlaufspannungen werden mit denjenigen verglichen, welche
aus
den
mittleren
Temperaturdifferenzen
von
numerischen
Strmungssimulationen (CFD) resultieren. Fr einen optimierten 1 kWel Stack
sagt das Modell entweder einen maximierten Wirkungsgrad von 4.2% oder ein
minimiertes Volumen von 0.0021 m3 voraus, wenn die HeisswasserTemperatur am Einlass 413 K und am Auslass 303 K betrgt.
Acknowledgements
First of all I thank Prof. Aldo Steinfeld, head of the Professorship of
Renewable Energy Carriers (PRE) at ETH Zrich and Solar Technology
Laboratory at the Paul Scherrer Institute, for giving me the possibility to work
in a highly exciting and interesting research field, for his supervision and
suggestions. I further thank Prof. Anke Weidenkaff, head of the Solid State
Chemistry Group at EMPA and professor at the Department of Chemistry and
Biochemistry at the University of Bern, for her collaboration in the project and
inspiring support. Special thanks go to Petr Tome for the collaboration, the
help in synthesizing the thermoelectric material and fabricating the
thermoelectric modules and the fruitful scientific discussions, to Philipp
Haueter and Laurenz Schlumpf for their technical support in the design and
construction of the experimental setups, to Matthias Trottmann, Oliver Brunko,
Dimas Surya and Alexandros Hmmerli for the help in synthesizing the
thermoelectric material and the support in the experimental campaigns, to
Zoran Jovanovic for the general help and suggestions, Hansmartin Friess for
the proofreading of this thesis, to Wojciech Lipiski and Matthew Roesle for
the help in modeling tasks, to all bachelor and master students being involved
in the project, i.e. Dominik Zimmermann and Simon Ackermann. Last but not
least I thank my family, my girlfriend Eveline and all my friends for their
consistent support and enthusiasm.
Contents
Abstract .............................................................................................................. i
Zusammenfassung ........................................................................................... iii
Acknowledgements ......................................................................................... vii
Contents ............................................................................................................ ix
Nomenclature ................................................................................................. xiii
1
Introduction ............................................................................................. 1
1.1
Electricity from renewable energy sources..................................... 1
1.2
Electricity from solar energy .......................................................... 1

Concentrating solar power ................................................ 2
1.2.2
Photovoltaics ..................................................................... 2
1.3
Geothermal energy ......................................................................... 3
1.4
Decentralized thermoelectric power generation ............................. 4
1.5
2
1.2.1
1.4.1
Solar thermoelectric conversion........................................ 4
1.4.2
Geothermal thermoelectric conversion ............................. 5
Thesis outline ................................................................................. 5
Thermoelectric basics .............................................................................. 7

2.1
History ............................................................................................ 7
2.2
Seebeck effect................................................................................. 8
2.3
Schematic of a thermoelectric converter module ........................... 8
2.4
Peltier and Thomson effect ........................................................... 10
2.5
Prospective of material research ................................................... 10
2.6
State-of-the-art application ........................................................... 11
Contents
Directly irradiated modules .................................................................. 13

3.1
4-leg perovskite-type oxide modules ........................................... 13

3.1.1
p-type La1.98Sr0.02CuO4 and n-type CaMn0.98Nb0.02O3 .... 14
3.1.2
Module assembly ............................................................ 15
3.2
Experimental setup ....................................................................... 15
3.3
Experimental results ..................................................................... 17
3.4
Heat transfer model ...................................................................... 21

3.4.1
Heat conduction .............................................................. 22
3.4.2
Radiative heat transfer .................................................... 23
3.4.3
Natural heat convection .................................................. 25
3.4.4
Maximum power output and solar-to-electricity efficiency

........................................................................................ 25
3.5
3.4.5
Boundary conditions and model parameters ................... 26
3.4.6
Numerical solution ......................................................... 26
Comparison between semi-transparent and opaque absorber plate

..................................................................................................... 27
3.6
Model validation .......................................................................... 28
3.7
Heat transfer analysis ................................................................... 29
3.8
Geometric analyses ...................................................................... 31
3.9
4
3.8.1
Variation of leg length .................................................... 32
3.8.2
Packing of legs ............................................................... 32
Summary and conclusion ............................................................. 35
A 1 kWth solar cavity-receiver prototype ............................................ 37

4.1
Advantage of cavity design .......................................................... 38
4.2
Experimental setup ....................................................................... 38
4.3
Experimental results ..................................................................... 41
4.4

4.4.1
Heat conduction .............................................................. 44
4.4.2
xi
4.4.3
Natural heat convection .................................................. 47
4.4.4
4.4.5
Numerical solution .......................................................... 49
4.5
Model validation........................................................................... 49
4.6
Heat transfer analysis ................................................................... 51
4.7
Summary and conclusion ............................................................. 52
Towards a 50 kWth solar cavity-receiver ............................................. 53

5.1
Idea of cascading .......................................................................... 53
5.2
Cascaded dual-stage modules ....................................................... 54
5.3

5.3.1
Heat conduction .............................................................. 57
5.3.2
5.3.3

........................................................................................ 58
5.4
5.5
6
5.3.4
5.3.5
Numerical solution .......................................................... 59
Simulation results ......................................................................... 60

5.4.1
Geometrical optimization and heat transfer analysis ...... 61
5.4.2
Solar-to-electricity efficiency ......................................... 63
Summary and conclusions ............................................................ 64
A 1 kWel geothermal stack .................................................................... 67

6.1
Schematic of thermoelectric stack ................................................ 67
6.2
Heat Transfer Model .................................................................... 68

6.2.1
Heat conduction .............................................................. 69
6.2.2
Forced heat convection ................................................... 70
6.2.3

........................................................................................ 71
6.2.4
xii
Contents
6.2.5
Numerical solution ......................................................... 73
6.3
Comparison with the CFD-based model ...................................... 73
6.4
Simulation results ......................................................................... 74

6.4.1
Maximized heat-to-electricity efficiency ........................ 75
6.4.2
Minimum stack volume .................................................. 76
6.4.3
Summary and conclusions .............................................. 78
Summary and conclusions .................................................................... 79

7.1
7.2
7.3
Solar energy conversion ............................................................... 79

7.1.1
A 50 kWth solar cavity-receiver ...................................... 79
7.1.2
Comparison to other technologies .................................. 80
Geothermal energy conversion .................................................... 80

7.2.1
A 1 kWel geothermal stack ............................................. 80
7.2.2
Comparison to other technologies .................................. 81
Comparison of thermoelectric conversion of concentrated solar

radiation and geothermal energy .................................................. 82
Outlook ................................................................................................... 85
8.1
Material research .......................................................................... 85
8.2
Analysis of geothermal energy use for small-scale applications .. 85
8.3
Fabrication and testing of solar cavity and geothermal stack ....... 86
8.4
Further Optimization .................................................................... 86
Appendix A ..................................................................................................... 87
Appendix B ..................................................................................................... 91
List of Figures ................................................................................................. 95
List of Tables ................................................................................................ 101
References ..................................................................................................... 103
Curriculum Vitae ......................................................................................... 111
List of publications ....................................................................................... 113
Nomenclature
Latin characters
a
A
A D
A
B
cp
d
d*
e
f
Fk-j
g
h
H
I
j
aperture diameter/width, m
surface area, m2
fitting constants
control surface, m2
domain boundary, m2
heat capacity, J kg-1 K-1
solar concentration ratio, suns
distance between legs, m
adjusted distance between legs, m
charge of electron/hole, A s
friction factor
view factor from surface k to j
gravitational acceleration, m s-2
convective heat transfer coefficient, W m-2 K-1
height, m
direct normal irradiation, W m-2
electric current density, A m-2
J
l
K
L
m
M
N
Nrays
Nsurfaces
n
p
electric current, A
leg length, m
number of stages
cavity/stack length, m
water mass flow, kg s-1
number of thermoelement legs
number of modules
number of Monte Carlo rays
number of surfaces
normal vector
pressure, Pa
xiv
Nomenclature
P
qcc
electric power, W
heat flux, W m-2
qccc
volumetric heat source, W m-3
Q
R
R2
S
t
t
T
heat transferred, W
electrical resistance,
coefficient of determination
Seebeck coefficient, V K-1
thickness, m
time, s
temperature, K
T
T
u
U
V
V
w
W
x,y
X
mid temperature, K
temperature difference, K
velocity, m s-1
voltage, V
volume, m3
control volume, m3
leg width, m
plate width, m
cartesian coordinates, m
cavity width, m
Greek characters
apparent
,
kj
thermal diffusivity, m2 s-1

total absorptivity
total apparent absorptance
correction factors
extinction coefficient, m-1
volumetric thermal expansion coefficient, K-1
iteration step
Kronecker delta
error
total emissivity
thermal conductivity, W m-1 K-1
adjusted thermal conductivity, W m-1 K-1
xv
efficiency
chemical potential, J mol-1
dynamic viscosity, kg m-1 s-1
water density, kg m-3
electrical resistivity, m
electrical contact resistivity, m2
Stephan-Boltzmann constant, W m-2 K-4

kinematic viscosity, m2 s-1
scattering albedo
Subscripts
0
abs
c
conv
dual
el
h
high
i,j,k
ins
int
low
m
max
mod
m,n
leg
OC
ray
seg
th
w
at normal conditions (T0 = 273.15 K, p0 = 101.325 Pa)

absorber plate
cold
convection
dual-stage
electric
hot
high-temperature stage
indices
insulation
intermediate
low-temperature stage
middle
maximum
module
number of finite volumes
thermoelement leg
open-circuit
single ray in Monte Carlo
stack segment
thermal
water
surroundings
xvi
Nomenclature
Dimensionless groups
Gr
Nu
Pr
Ra
Re
ZT
Grashof number
Nusselt number
Prandlt number
Rayleigh number
Reynolds number
figure-of-merit
Abbreviations
CFD
CPC
CSP
DNI
EMPA
ETH
Exp
FV
HFSS
MC
PDE
PB
PV
Sim
SOR
TEC
XRD
computational fluid dynamics

compound parabolic concentrator
concentrating solar power
direct normal irradiation
Eidgenssische Materialprfungs- und Forschungsanstalt (Swiss
Federal Laboratories for Materials Science and Technology),
Dbendorf, Switzerland
Eidgenssische Technische Hochschule (Swiss Federal Institute
of Technology), Zrich, Switzerland
experiment
finite volume
High Flux Solar Simulator
Monte Carlo
partial differential equation
periodic boundaries
photovoltaic
simulation
successive over-relaxation
thermoelectric converter
X-ray diffraction
1 Introduction
1.1 Electricity from renewable energy sources
In 2008, the world net electricity generation was 19.1 PWh el and is
predicted by the International Energy Outlook 2011 to rise to 25.5 PWhel by
2035, which represents an increase of 34%. From 1990 to 2008, the growth in
net electricity generation was higher than the growth in total energy
consumption (3.0% per year and 1.8% per year, respectively). The world
demand for electricity is predicted to increases by 2.3% per year from 2008 to
2035, which exceeds the expected growth in total energy consumption of 1.4%
per year [1].
In order to meet the future fast-growing electricity demand and the
environmental issues of using fossils fuels, big efforts have to be done in the
development of renewable energy sources including hydroelectric, wind,
geothermal, and solar sources. They are predicted to be the fastest growing
sources of electricity generation. Their contribution to the worldwide mix is
estimated to grow from 19% in 2008 to 23% in 2035. More than 82% of
increase is expected in hydroelectric and wind power in the same period of
time. In view of the more and more stringent need to substitute fossil and
nuclear fuels, it is particularly important to promote also other renewable
energy sources like solar and geothermal energy, which possess a promising
potential, but have been confined so far to a niche existence [1].
1.2 Electricity from solar energy

Solar energy is the worlds primary source of energy and theoretically it
would only take 2% of Saharas land area to cover the worlds electricity
demand. Solar energy is virtually unlimited, freely available and has no impacts
on the ecology. However, its drawbacks are high dilution, intermittency and
Introduction
unequal distribution over the earth [2]. In 2010, the main use of solar energy to
produce power is photovoltaics (PV) and concentrating solar power (CSP).
1.2.1
Concentrating solar power
CSP systems use mirrors to concentrate solar radiation and produce hightemperature heat, which is then converted into electricity by e.g. a steam/gas
turbine or a Stirling engine. The three main technologies are:
(1) Parabolic troughs
(2) Solar towers
(3) Parabolic dishes
Parabolic troughs have an operating temperature up to 500C and commonly
drive a Rankine-cycle (steam turbine). They are commercially proven and
reliable. The peak solar-to-electricity efficiency reaches 21%. Solar towers can
be operated at >1000C and are able to drive a Brayton-cycle (gas turbine) or
even a combined Brayton-Rankine-cycle. They are commercially still in proof.
The peak efficiency reaches 23%. Both parabolic troughs and solar towers are
used in large-scale power plants with installed electric powers in the order of
magnitude of 10 MWel and have high potential to become economically
competitive with other non-solar electricity generation technologies. Parabolic
dishes usually drive a Stirling engine and theoretically reach a solar-toelectricity efficiency up to 29%. They can be operated in single units for
decentralized power generation (10 kWel) or in parks (1 MWel). However they
have not found commercial application yet [2, 3]. In 2010, the worldwide
installed power for CSP is ~1 GWel with an expected huge capacity rise to 147
GWel by 2020 [4].
1.2.2
Photovoltaics
PV cells consist of doped semiconductors, which directly convert solar

light into electricity. It has no moving components and is silent, totally scalable
and reliable. Simple amorphous Si cells, which are standard modules on the
market, achieve a solar-to-electricity efficiency of about 7%. Newer multi-layer
3
thin-film cells or crystalline Si cells can reach solar-to-electricity efficiencies
up to 20%, but at the expense of much higher material and fabrication costs.
The latest trend is towards concentrated PV. The intended benefit is to reduce
the required cell area and to increase the solar-to-electricity efficiency. Systems
having solar concentration ratios of 360 and using III/V-semiconductor cells
have already achieved a solar-to-electricity efficiency of up to 42% [5].
However, concentrated PV is still expensive and therefore essentially restricted
to small-scale and decentralized power generation. In 2011, the worldwide
installed power is 40 GWel with a predicted annual growth rate of 71% in the
next years [6].
1.3 Geothermal energy

Geothermal energy is a sustainable heat source, independent of climate,
seasons and day/night cycles, which can be used for power and heat generation.
The worldwide mean temperature gradient in the earths crust is 33C / 1kmdepth with a strong geographic and geologic dependency. The temperature
ranges are classified by the following three categories:
(1) High-temperature heat (> 100C)
(2) Mid-temperature heat (40 100C)
(3) Low-temperature heat (~15C)
High-temperature heat is mainly used for combined power (Rankine-cycle) and

heat generation. Due to the high investment cost of a Rankine-cycle power
plant and the cost-intensive deep (>4km) borehole drilling, high-temperature
recovery and conversion is only profitable in large scale plants with an installed
power of >1 MWel. Mid- and low-temperature heat is commonly used only for
heat production. The mid-temperature heat delivers direct heat, e.g. for district
heating, whereas the low-temperature heat is used for driving heat pumps. The
use of mid- and low-temperature heat is used in all plant sizes, i.e. from
decentralized small-scale to large scale plants [7, 8]. In 2010, the globally
installed thermal power (mid- and low-temperature heat) was 50.6 GWth with
an annual growth rate of 12.3%, and the electric power (high-temperature heat)
was 10.7 GWel with an annual growth rate of 3.7% [9].
Introduction
1.4 Decentralized thermoelectric power generation

The figures for the installed electric power of solar and geothermal energy
have shown the increasing importance of these energy sources. Additionally, it
is expected that decentralized power generation from renewable energy sources
will be of increasing importance [10, 11]. Apart from well-established systems
like PV etc., thermoelectric converters (TEC) offer a new and attractive
alternative for decentralized power generation. A TEC is a solid-state heat
engine, in which the electron gas serves as the working fluid and converts a
flow of heat into electricity. It has no moving components and is silent, totally
scalable, extremely reliable and able to convert any heat at any temperature
level [12, 13].
1.4.1
Solar thermoelectric conversion
Different systems to convert solar radiation by TEC modules have been

investigated and tested. A TEC module using a flat-panel spectrally selective
absorber, which is also a thermal concentrator, has been proposed achieving a
peak efficiency of 4.6% [14]. A parabolic trough concentrator combined with
TEC modules and placed in the receiver tube has been investigated. The
thermal efficiency of the concentrator/receiver system was found to be <40%
[15]. A thermodynamic analysis of a parabolic dish combined with a Bi-Te
module predicted a solar-to-electricity efficiency of 2.81% at the temperature of
280C [16]. The design of a solar cavity-receiver for supplying hightemperature heat to a thermionic/thermoelectric system has been proposed and
temperature distributions were measured on a prototype made of graphite [17].
With the advent of novel functional thermoelectric materials withstanding hightemperatures (>900 K), the thermoelectric conversion of highly concentrated
solar radiation has been considered [18]. The theoretical solar-to-electricity
efficiency of Si-Ge alloys operated at a temperature of 1000 K is 12%, showing
the advantage of achieving higher efficiencies by using high temperatures [19].
The direct conversion of highly concentrated solar radiation has been
experimentally demonstrated by directly irradiated TEC modules operating at
900 K on the hot side and achieving solar-to-electricity efficiencies <1% [20,
21].
5
1.4.2
Geothermal thermoelectric conversion
As of recent, the thermoelectric conversion of geothermal energy has been

considered [22]. Stacks composed of TEC module arrays and sandwiched
between parallel-, counter- and cross-flow water heat exchangers have been
analyzed. The counter-flow configuration has shown linear temperature profiles
in the channels and a constant temperature difference across the TEC modules
[23-26]. The effect of the channel dimensions and the geometry of the TEC
modules on the heat-to-electricity efficiency and the volume has been analyzed
in a stack with counter-flow heat exchangers, showing that the stack can either
be optimized for maximum efficiency or minimum volume [27, 28].
1.5 Thesis outline

In this work, the decentralized conversion of solar and geothermal energy
is investigated by thermoelectric converters. The goal is to develop, analyze
and optimize a TEC system that can be operated in a 50 kW th solar dish, which
delivers highly concentrated solar radiation for high temperatures (>900 K),
and a TEC stack converting geothermal energy and delivering 1 kW el power for
small remote buildings. The thesis was carried out in the framework of a joint
project of ETH Zrich and EMPA Dbendorf.
Chapter 2 gives an introduction to the basics of thermoelectricity and
outlines the state-of-the art applications in the field of thermoelectricity.
Chapter 3 presents the investigations on high-temperature stable TEC modules
directly irradiated by concentrated solar radiation. Several modules have been
tested at ETHs High-Flux Solar Simulator (HFSS). A combined radiationconvection-conduction heat transfer model was formulated and experimentally
validated. It was applied to make a heat transfer analysis of single modules.
Chapter 4 describes the design and testing of a 1 kWth solar cavity-receiver
prototype. Experimentation was carried out again at ETHs HFSS. A heat
transfer model coupling conduction, convection and radiation is implemented
and experimentally validated. It is then used to analyze the heat transfer in the 1
kWth cavity-receiver. In chapter 5, a 50 kW th cavity-receiver packed with
efficient cascaded modules is considered. The model for the 1 kW th cavityreceiver is extended and applied to simulate the 50 kWth cavity-receiver, to
optimize its geometry and to analyze the effect of the maximum module
Introduction
efficiency and the operated maximum cavity temperature on the solar-toelectricity efficiency. In chapter 6, the conversion of geothermal energy by a 1
kWel thermoelectric stack is considered. A heat transfer model coupling
conduction in the modules with convection to and from the modules plates is
formulated and compared to CFD simulations. The model is then applied to
optimize the stack either for maximum heat-to-electricity efficiency or
minimum stack volume for a hot water inlet and outlet temperature range of
313 413 K and 303 393 K, respectively. Chapter 7 gives a summary and
comparison with other technologies. In chapter 8, finally, an outlook to further
investigations is presented.
2 Thermoelectric basics
2.1 History
In 1821, the first of three thermoelectric effects was discovered by T.
Seebeck. He showed that an electromotive force can be produced by heating the
junction between two electric conductors made of different metals. In 1834, J.
Peltier, a French watchmaker, observed the second thermoelectric effect. He
found that the passage of an electric current through a thermocouple produces a
small heating or cooling effect depending on its direction. It seems that it was
not immediately realized that the Seebeck and Peltier phenomena are closely
related with each other. However, in 1855, this relationship was recognized by
W. Thomson (who later became Lord Kelvin). By applying the theory of
thermodynamics to the problem, he was able to establish a relationship between
the coefficients describing the Seebeck and Peltier effects. His theory also
showed that there must be a third thermoelectric effect, which exists in a
homogeneous conductor. This effect, known as the Thomson effect, consists of
reversible heating or cooling when there is both a flow of electric current and a
temperature gradient [29, 30].
However, the relatively low heat-to-electricity efficiency (typically around
5%) of thermoelectric converters has restricted their use in the past to
specialized applications only. In the 1950s and 1960s the main application
area was space exploration by the USA and the former USSR where
autonomous long-life power generators were required. At the time the only
existing technology were thermoelectric generators powered by radioisotope
heat sources [31-33]. In recent years, an increasing public awareness of
environmental issues, in particular global warming, has resulted in broad based
research for alternative commercial methods of generating electric power. With
advances in the material research, thermoelectric generators have emerged as a
promising new alternative. Today, thermoelectric power generation has
attracted increasing attention as a green and flexible source of electricity able to
meet a wide range of power requirements [34, 35].
Thermoelectric basics
2.2 Seebeck effect

An n-type semiconductor bar is connected between a hot and a cold
temperature reservoir Th and Tc, respectively, yielding the temperature
difference T, as shown in Figure 2-1. The temperature gradient in the
semiconductor bar induces a gradient of the electron concentration, and a
voltage difference U across the bar is established. This phenomenon is called
Seebeck effect [36]. The voltage difference can be expressed in terms of the
temperature T and the temperature-dependent and material-specific Seebeck
coefficient S by:
'U
Th
Tc
S T dT
(2.1)
Figure 2-1: n-type semiconductor bar between temperature

difference T showing voltage difference U.
2.3 Schematic of a thermoelectric converter module

A thermoelectric converter (TEC) module comprises p-type and n-type
semiconductor legs sandwiched between a hot and a cold plate at hot (Th) and
cold (Tc) temperature reservoirs, respectively, and connected thermally in
parallel and electrically in series, as shown in Figure 2-2 for a single p/n-type
leg pair. Then the Seebeck effect induces a voltage difference U between the
cold terminals. When an external load is connected, an electric current (denoted
by J) flows and power can be extracted.
Figure 2-2: Schematic of a 2-leg (one p/n-type leg pair) thermoelectric converter
operated at temperature difference T and powering an external load.
The performance of the thermoelectric material is characterized by its

figure-of-merit defined as:
ZT
S2 T
U N
(2.2)
where S is the Seebeck coefficient, the electric resistivity, N the thermal

conductivity and T the mid temperature between the hot and cold temperature,
given by T 0.5 Th Tc . Then the maximum heat-to-electricity efficiency of
a TEC module can be expressed in terms of the hot temperature Th, the cold
temperature Tc, and the figure-of-merit ZT by:
Kmod
Th Tc
1 ZT 1
T
Th
1 ZT c
Th
(2.3)
The first factor of Equation (2.3) represents the Carnot efficiency, which
depends only on the temperature difference of the two temperature reservoirs Th
and Tc, and the second factor embodies the figure-of-merit ZT . The higher
ZT , the higher is the maximum module efficiency, approaching the Carnot

efficiency for ZT o f [29, 30, 36, 37].
As an example, a Bi-Te-based module having ZT = 1 and operating at 200
K temperature difference reaches the maximum module efficiency of 5% [38].
10
2.4 Peltier and Thomson effect

A TEC module as shown in Figure 2-2 can be operated in the opposite
mode as cooler or heater. In absence of a temperature difference across the
module an external current source is connected, which drives an electric current
(denoted by J) through the module. Then the current flow induces a heat
transport in the thermoelement legs, which will cause a temperature difference
T between the hot (Th) and cold (Tc) temperature on the upper and lower side,
respectively. This effect is called Peltier effect. The heat transport in a
thermoelement leg is the product of the Seebeck coefficient S, the
thermoelement leg temperature T and the electric current J:
(2.4)
QPeltier S T J
If the direction of the current flow is reversed the direction of the heat transport
also changes and the hot (Th) and cold (Tc) temperature are interchanged.
The third effect is the Thomson effect, which is a volumetric effect similar
to Joule heating. It is given in terms of T, S and J by:
dS
ccc
qPeltier
T
J T
(2.5)
dT
It is reversible, and depending on the sign of equation (2.5) it becomes positive
or negative, which means that the Thomson effect is either a heat source or a
heat sink.
The three thermoelectric effects (Seebeck, Peltier and Thomson effect) are
permanently present and superimpose each other.
2.5 Prospective of material research

The primary goal of the material research is to increase the figure-of-merit
ZT . Figure 2-3(a) shows the electric conductivity , the Seebeck coefficient S

and the electric power factor S 2 V as a function of the carrier concentration
ranging from insulators to metals. The electric conductivity increases with the
carrier concentration, while the Seebeck coefficient decreases. The electric
power factor has a peak at a carrier concentration of around 1019/cm3, which
lies in the semiconductor range. In general, the total thermal conductivity has
a contribution of atomic lattice thermal conduction L and electron thermal
conduction e. The electron thermal conduction contributes 1/3 and the lattice
11
thermal conduction 2/3 of the total thermal conduction for semiconductors with
a carrier concentration of 1019/cm3, as shown in Figure 2-3(b). One goal is to
minimize the thermal conductivity. As the electric conductivity and the electron
thermal conductivity are correlated, only the lattice thermal conductivity can be
decreased without badly affecting the electric conductivity. One approach uses
so-called phonon-glass-electronic-crystals, in which crystal structures contain
weakly bound atoms or molecules that rattle within an atomic cage and
conduct heat like a glass but conduct electricity like a crystal. Candidate
materials receiving considerable attention are the filled skutterudites and the
clathrates [30, 39].
(a)
(b)
Figure 2-3: Schematic dependence of (a) the electric conductivity , Seebeck

coefficient S, power factor S2; and (b) the thermal conductivity on the carrier
concentration. Extracted from Ref. [30] and modified by author.
2.6 State-of-the-art application

Today, the standard TEC modules are based on Bi-Te semiconductors, e.g.
p/n-type (Bi0.2Sb0.8)2Te3 / Bi2(Te0.96Se0.04)3 thermoelement legs. These modules
12
can be operated with a maximum temperature difference of 200C, i.e. with

225C and 25C on the hot and cold side, respectively. The figure-of-merit is 1
and the maximum module conversion efficiency around 5% [40].
A fairly well-established application is the conversion of waste heat. Vast
quantities of waste heat are released to the environment at temperatures which
are too low in order to recover the heat using conventional electric power
generators [12]. Waste heat powered thermoelectric Bi-Te generators find
application in various areas. Examples are: heat recovery from the exhaust
gases in automobiles (~700 Wel) and large scale power plants (~1 MW el) [41,
42].
3 Directly irradiated modules 1

In this chapter, the direct conversion of concentrated solar radiation with
high-temperature modules is experimentally investigated with respect to the
influence of the thermoelement leg length on the solar-to-electricity efficiency.
Further, a heat transfer model based on finite volumes (FV) coupling radiation,
conduction and convection is developed and experimentally validated in terms
of voltage/power output measurements. The model is then applied to analyze
the heat transfer through the TEC modules and further geometric effects on the
solar-to-electricity efficiency.
3.1 4-leg perovskite-type oxide modules

Figure 3-1 depicts a schematic of a 4-leg high-temperature module
fabricated at EMPA and used in the experimental runs. The dimensions of the
hot and cold plates are W x W x t = 30 30 0.25 mm3. The plates are made of
Al2O3. The hot plate serves as absorber of the solar irradiation and is
additionally coated with graphite to augment its absorptivity (total absorptance
graphite = 0.95) [44]. Maximum operating temperature is 900 K, at which
graphite becomes unstable and starts to burn [45]. Six demonstrator modules
were manufactured with leg lengths l = 4, 5, and 10 mm (2 modules for each
leg length) [20]. Each leg has a quadratic cross section of width w = 4.3 mm
and a distance d = 10.7 mm from the neighboring legs. The thermoelement legs
are made of perovskite-type oxides; the p- and n-type legs are La1.98Sr0.02CuO4
and CaMn0.98Nb0.02O3, respectively [46, 47].
Material from this chapter has been published in [43]: C. Suter, P. Tome, A. Weidenkaff,
and A. Steinfeld, "Heat transfer analysis and geometrical optimization of thermoelectric converters
driven by concentrated solar radiation," Materials, vol. 3, pp. 2735-2752, 2010.
14
Directly irradiated modules
Figure 3-1: Top and front view of 4-leg module. Indicated are the plate width W, the leg
width w, the distance from the neighbouring legs d, the leg length l and the plate
thickness t.
3.1.1
p-type La 1.98Sr0.02 CuO4 and n-type CaMn0.98Nb 0.02 O3
Perovskite-type materials exhibit chemical and mechanical stability at

high temperatures (> 1000 K) but at the expense of having a figure-of-merit ZT
~ 0.07, as compared to Bi-Te with ZT = 1.0, but Bi-Te is stable only up to about
500 K [38, 48, 49]. General advantages of thermoelectric oxide materials are
the high-temperature stability in air, non-toxicity and the low material costs
[50-54]. Perovskite-type oxides have a high potential for improvement as it has
been shown that materials with ZT > 0.3 can be synthesized [48]. The p-type
La1.98Sr0.02CuO4 and n-type CaMn0.98Nb0.02O3 materials used in the present
work are not yet optimized for a high ZT but serve for demonstrator modules
(proof-of-concept).
The p-type La1.98Sr0.02CuO4 and n-type CaMn0.98Nb0.02O3 materials are
prepared by a chimie-douce synthesis whose main advantage is high
homogeneity and purity of the synthesized product [55]. The calcined powders
are characterized by X-ray diffraction (XRD) [46, 47]. The particle size is
reduced to 1 6 m by ball milling. Then the powders are hydrostatically
pressed to cylindrical pellets with a diameter of 32 mm at a pressure of 200
kPa. The p- and n-type pellets are sintered for 5 h at 1373 K and 1523 K,
respectively [20]. Finally, the pellets are machined into the thermoelement legs.
The electric resistivity leg and the Seebeck coefficient Sleg were measured
with an RZ2001i Ozawa Science measurement system. The thermal
conductivity Nleg was evaluated indirectly by measurements of the thermal
diffusivity (Netzsch LFA apparatus) and the specific heat (Netzsch DSC
15
apparatus). The electric and thermal transport properties were measured in the
temperature range of 300 K < T < 900 K. A detailed description of the
thermoelectric measurements is reported in Ref [56]. The thermoelectric
properties Nleg, leg, and Sleg are shown in Appendix A.1, Figures A-1, A-2 and
A-3, respectively.
3.1.2
Module assembly
The module plates are made of Al2O3 (Rubalit 708 S, CeramTec GmbH,
Plochingen, Germany). The assembly schematic is shown in Figure 3-2. In a
first step, a dual-layer screenprinting through a stainless steel stencil is applied
in order to metallize the electric contacts of the thermoelement legs and the
Al2O3 plates [57]. The first layer is printed with AuPtPd conductor paste (4597
AuPtPd, DuPont, Wilmington, USA), dried at 423 K for 15 min to evaporate
the solvents, and followed by annealing at 1223 K for 15 min to induce the
diffusion. The second layer is printed with AgPd conductor paste (DuPont,
Wilmington, USA), and again dried at 423 K for 15 min and annealed at 1223
K for 15 min. In a second step, Ag sheets serving as electric contacts are placed
between the thermoelement legs and the Al2O3 plates using a mask for
positioning depicted in Figure 3-2. Finally, the assembled modules were heated
at 1223 K for 1 h in order to solder the electric contact layers, which also serve
as mechanical linkage between the thermoelement legs and the Al2O3 plates,
and in order to ensure and test the mechanical stability of the module [58].
3.2 Experimental setup

Experimentation was carried out at the ETHs High Flux Solar Simulator
(HFSS): a high-pressure Argon arc enclosed in a 27 mm-diameter, 200 mmlength water-cooled quartz envelope and close-coupled to an elliptical reflector
that delivers an external source of intense thermal radiation to simulate the heat
transfer characteristics of highly concentrating solar systems [59]. The solar
flux concentration is characterized by the mean concentration ratio C , which is
defined as C
Qsolar
l / I A , where Qsolar is the solar power intercepted by a
16
target of area A. The ratio C is often expressed in units of suns when

normalized to I = 1 kW/m2.
Figure 3-2: Assembly schematic of a 4-leg module, extracted from Ref. [58]. Indicated
are the p/n-type thermoelement legs, the Al2O3 plates, the dual-layer metallization
(AuPtPd and AgPd), the electrical contacts, the electrodes and the positioning mask.
The experimental set-up is schematically shown in Figure 3-3. Incident

radiative fluxes were measured by a thermogage (denoted by F, Thermogage
Circular Foil Flux Transducer TG1000-1, accuracy 3% [60]) placed
symmetrically to the TEC module at the focal plane. The 4-leg module was
exposed to a maximum mean solar concentration ratio of 300 suns. K-type
thermocouples (denoted by T, tip = 0.5 mm, spatial accuracy = 0.25 mm)
were used to measure temperatures of the plates and of the hot end, middle, and
cold end of the thermoelement legs. Terminals (denoted by U) were provided at
the cold ends for measuring the output voltage U of the module. The cold plate
was attached to a water-circuit cooler at room temperature by a screw fixation.
17
(a)
(b)
Figure 3-3: Schematic of the experimental setup at ETHs HFSS. (a) The 4-leg module
is placed at the HFSSs focal plane; incident solar radiative fluxes measured by a
thermogage (F). (b) Position of K-type thermocouples (T) used to measure temperatures
of the plates and of the hot end, middle, and cold end of the legs; terminals (U) provided
at the cold ends for measuring the output voltage of the module. The cold plate was
attached to a water-circuit cooler (denoted by screw fixation).
3.3 Experimental results

The hot (Th) and cold (Tc) plate temperature and incident solar radiative
cc as a function of time t are shown in Figure 3-4 for a representative
flux qsolar
experimental run using a module with leg length l = 4 mm. The solar radiative
flux was increased stepwise and held at constant level for intervals of 3 to 5
minutes. Due to the low thermal inertia and fast temperature response, steadystate conditions are assumed for each time interval. Maximum hot temperature
cc
was Th = 900 K, at which graphite starts to burn [45]. Tc increased with qsolar
18
due to the incorporation of the screw fixation (see Figure 3-3(b)) causing
insufficient cooling rates between the cold plate and the cooler.
For the same module (l = 4 mm), Figure 3-5 shows the measured power
output P as a function of the voltage output U for external loads with
resistances in the range Rload = 0.1 3.5 and for incident solar radiative
cc = 1.8 10 W cm-2. A parabola that corresponds to the
fluxes in the range qsolar
behavior of an ideal voltage source with an internal resistance is fitted through
each data point set. The maximum power output is the maximum of the
parabola, which is achieved when the external load resistance equals the
internal resistance (matched load) [61]. In this case it is Pmax = 0.006, 0.015,
cc = 1.8, 2.9, 4.1, 5.4, 8.2 and
0.023, 0.031, 0.038 and 0.046 W for qsolar
10 W cm-2, respectively.
The measured temperature distribution for two tested modules with leg
cc = 6 W cm-2. The x-axis error
length l = 10 mm is shown in Figure 3-6 for qsolar
bars indicate the spatial accuracy (0.25 mm) of the thermocouple placing. As
expected, the quasi linear profile (fitted straight lines) indicates a predominant
heat transfer by conduction across the legs. The abnormal behavior of 83 K
temperature difference between the p-type leg of module 1 and the one of
module 2 at the cold side is presumably due to the incorporation of the screw
fixation (see Figure 3-3(b)) causing different heat transfer rates.
1000
17.5
12.5
700
10
qcc
600
7.5
solar
solar
800
[W cm-2]
15
Th
qcc
T [K]
900
500
400
2.5
Tc
300
500
1000
1500
t [s]
2000
2500
Figure 3-4: Temperature of hot and cold plates and solar radiative flux as a function of
time during a representative experimental run for a 4-leg module with leg length l = 4
mm.
19
0.05
-2
q''
solar
= 10 W cm
0.045
measured
fitted (parabola)
-2
q''solar = 8.2 W cm
0.04
0.035
q''solar = 5.4 W cm-2
P [W]
0.03
-2
q''solar = 4.1 W cm
0.025
0.02
-2
q''solar = 2.9 W cm
0.015
0.01
q''solar = 1.8 W cm-2
0.005
0
0.05
0.1
0.15
0.2
U [V]
0.25
0.3
0.35
0.4
Figure 3-5: Fitted and measured power output as a function of the voltage output for
cc = 1.8 10 W cm-2, and for external

incident solar radiative fluxes in the range qsolar
loads with resistance in the range Rload = 0.1 3.5 for a 4-leg module with leg length
l = 4 mm. The coefficient of determination for each fitted parabola is R2 > 0.997.
900
Module 1, p-type leg
Module 1, n-type leg
Module 2, p-type leg
Module 2, n-type leg
fitted (linear)
800
T [K]
700
600
500
400
300
10
x [mm]
Figure 3-6: Temperature distribution along the p- and n-type legs for two modules with
leg length l = 10 mm. The x-axis error bars indicate spatial accuracy (0.25 mm) of
thermocouple placing. The coefficient of determination for each fitted straight line is
R2 > 0.993.
20
The solar-to-electricity efficiency of a TEC module is defined as the ratio

of the maximum power output of the module (Pmax) to the incident solar power
(Qsolar):
Ksolar
Pmax
Qsolar
(3.1)
cc and the
where Qsolar is obtained from the measured solar radiative flux qsolar
absorber plate surface Aplate:
Qsolar
cc
Aplate qsolar
(3.2)
For modules with leg lengths l = 4, 5, and 10 mm the maximal power

cc = 9.9, 9.7, and 5.7 W cm-1,
outputs are Pmax = 45.6, 51.6 and 42.2 mW for qsolar
respectively. The solar-to-electricity efficiency solar as a function of the solar
cc is shown in Figure 3-7 for the leg lengths l = 4, 5, and 10
radiative flux qsolar
cc ; the y-axis error

mm. The x-axis error bars are due to the uncertainty of qsolar
bars are due to the averaging over 2 4-leg modules (2 modules for each leg
cc = 9.9, 9.7 and 5.7
length). The curves are plotted up to the maximum qsolar
W cm-1, respectively, for which the maximum hot plate temperature Th = 900 K
is reached. Higher solar fluxes resulted in the burning of the graphite coating.
cc as a result of the higher
The solar-to-electricity efficiency increases with qsolar
temperature difference across the legs, which in turn corresponds to a higher
Carnot limitation [62]. In contrast, solar decreases with Th as re-radiation losses
cc , which
are proportional to Th4 . Thus, for each l there is an optimum qsolar
cc = 4, 8 and 4
maximizes solar. For l = 4, 5 and 10 mm the optimum flux is qsolar
W cm-2 and the corresponding efficiency is solar,max = 0.065, 0.06 and 0.083%,
respectively. The modules with l = 10 mm have a higher efficiency than the
modules with l = 4 or 5 mm. This results from the contact resistance whose
contribution to the total resistance of the module becomes less for longer leg
lengths [63].
21
0.1
l = 10 mm, Exp
l = 5 mm, Exp
l = 4 mm, Exp
0.09
Ksolar [%]
0.08
0.07
0.06
0.05
0.04
0.03
6
qccsolar [W cm-2]
10
12
Figure 3-7: Experimentally measured solar-to-electricity efficiency as a function of

solar radiative flux for the leg lengths l = 4, 5, and 10 mm. Error bars indicating
cc and of solar.
uncertainty of qsolar
3.4 Heat transfer model

A 2D steady-state heat transfer model is formulated. A cross section of the
model domain, divided into m x n cells, is depicted in Figure 3-8. It contains
the three major components: the absorber plate, one p- and one n-type
thermoelement leg, and the space in between. The domain is assumed to be
infinitely long in y-direction; therefore, periodic boundaries conditions are
imposed at the sides, denoted by the dashed-dotted lines (PB). The heat transfer
modes considered are: (1) conduction in the complete domain, and (2) radiative
heat transfer among all surfaces with two approaches assuming: (a) a semitransparent absorber plate; (b) an opaque absorber plate. It is further assumed:
(i) the p/n-type thermoelement legs are opaque, gray and diffusely scattering;
(ii) the gas phase is radiatively non-participating, and its refractive index is
equal to unity; (iiia) for approach (a), the absorber plate is radiatively
participating with isotropic scattering and with temperature and wavelength
independent extinction coefficient abs and albedo abs; (iiib) for approach (b),
the absorber plate is opaque, gray and exhibits diffuse scattering; (iv)
convection is only considered from the surface on top of the hot plate; (v) only
open-circuit voltage (J = 0) is considered.
22
Figure 3-8: Schematic of the model domain (divided into m x n cells). Indicated are the
incoming solar radiation (Qsolar), the re-radiation and convective heat losses
(Qreradiation+concection), the outgoing radiation and conduction (Qconduction+radiation,out), the cold
plate temperature (Tc) and the open-circuit condition (J = 0).
3.4.1
Heat conduction
The general steady-state energy conservation equation in differential form

applied to the absorber plate for approach (a) is given by:
ccc
N plate T qradiation
(3.3)
ccc
where plate is the thermal conductivity of the absorber plate and qradiation
the
radiative volumetric heat source. For approach (b), Equation (3.3) is used with
ccc
= 0.
qradiation
for the thermoelement legs (p/n-type domain) including Fouriers heat
conduction, Joule heating and the Thomson effect is given by:
23
N leg Tleg U leg j 2 Tleg
dSleg
dTleg
j Tleg
(3.4)
where leg is the thermal conductivity, Tleg the temperature in the leg, leg the
electrical resistivity, and Sleg the Seebeck coefficient. The current density j
satisfies conservation of electrical charge:
j 0
(3.5)
and is given in terms of the voltage Uleg and the leg temperature Tleg by [62]:
Pleg
U leg Sleg Tleg

(3.6)

U leg eleg
Note that the ratio of the chemical potential leg and the charge of
electrons/holes eleg is assumed constant. Thus, the gradient of this ratio
vanishes. Considering open-circuit (j = 0), Equation (3.4) simplifies to:
j
N leg Tleg 0
(3.7)
0 Uleg Sleg Tleg
(3.8)
and Equation (3.6) to:
As only a 2D geometry is considered the thermal conductivity in the

*
in order to compensate the reduced
direction along the plate is adjusted as N plate
heat conduction in 2D.

Thermoelectric properties of the thermoelement legs are interpolated from
measured properties, see appendix A.1. The thermal conductivity of the Al 2O3
has been approximated by a polynomial, see Appendix B.5, Table B-5.
3.4.2
Radiative heat transfer
For approach (a), the radiative heat transfer within the absorber plate is
determined by the collision-based Monte Carlo (MC) method [64]. The
ccc
radiative source term qradiation,i
in volume cell i is approximated by:
ccc
qradiation,
i |
N rays,abs,i qray
'Vi
4
2 E plate 1 Zplate V Tplate,
i
(3.9)
24
where qray is the power carried by a single ray, Tplate,i the discrete temperature of
cell i, Nray,abs,i the number of rays absorbed, Vi the control volume of cell i,
plate the extinction coefficient of the absorber plate, and plate its albedo.
cc
The net radiative flux qradiation,i
for inner radiating surfaces of surface cell i
is calculated as:
cc
qradiation,
i |
N ray,abs qray
'Ai
4
H surface V Tsurface,
i
(3.10)
where qray is the power carried by a single ray, Tsurface,i the temperature of the
surface cell i, Nray,abs the number of rays absorbed, Ai the control surface of
cell i, and surface the emissivity.
cc
For approach (b), the net radiative flux qradiation,i
from inner radiating
surface elements is computed by means of the radiosity method (enclosure
theory) [65], assuming the p/n-type surfaces with emissivities p-type and n-type,
respectively, and the inner surface of the cold and absorber plate with
emissivity Al2O3. The corresponding system of equations is [64]:
G kj
1 H j
cc
Fk - j
qradiation,
j
H
H
j 1
j
j
for k 1...2(mP/N nspace )
2(mP/N +nspace )
2(mP/N +nspace )
kj
j 1
4
Fk - j V Tsurface,
j
(3.11)
where kj is the Kronecker function, mP/N the number of p/n-type elements in xdirection and nspace the number of elements in y-direction. The view factors Fk-j
are calculated by applying reciprocity relations (AjFj-k = AkFk-j), enclosure
N
criterion ( Fk j
1 ), and tabulated view factors [66].
j 1
Re-radiation from the surface of top of the absorber plate is calculated by

MC for approach (a). For approach (b) it is given for a surface cell i by:
cc
qreradiation,
i
4
H graphite V Tsurface,
i
(3.12)
where graphite is the emissivity of the graphite-coated absorber plate.

The outgoing heat flux on the cold side contains radiation losses Qradiation,out
through the space to the cold plate as well as conduction Qconduction through the
thermoelement legs to the cold plate. Qradiation,out is either calculated by MC for
approach (a), or by the radiosity method for approach (b).
25
3.4.3
Natural heat convection
Free convection from the top of the absorber plate to the environment is
calculated using a Nusselt correlation for a horizontal flat plate [67]:
Nu W
h W / N air
0.54 Ra1/4
W
10
d RaW d 107
(3.13)
g E Tsurface Tf W 3
Ra W GrW Pr
Q D
where NuW, RaW, GrW, and Pr are the Nusselt, Rayleigh, Grashof, and Prandtl
numbers, h the convective heat transfer coefficient, air the thermal conductivity
of air, g = 9.81 m s-2 the gravitational acceleration, the volumetric thermal
expansion coefficient approximated by 1/Tsurface, the kinematic viscosity of
air, its thermal diffusity, Tsurface the mean surface temperature of the absorber
plate, T the surroundings temperature and W the width of the absorber plate.
Then the free convective heat flux is given by:
cc
qconvection
h Tsurface Tf
(3.14)
Physical properties of air were estimated for 400 K [68].
3.4.4
As only open-circuit voltage (J = 0) is considered the maximum power

output of the 4-leg module Pmax is calculated based on the matched load
assumption in terms of the open-circuit voltage UOC, the resistance of the
thermoelement legs Rlegs, and the contact resistance between thermoelement
legs and conduction strips Rcontact given by:
Pmax
2
U OC
1
4 Rlegs Rcontact
(3.15)
The leg resistance for one leg is calculated according to:
'xleg
mleg
U leg,i
(3.16)
w2 i 1
where leg,i is the legs temperature dependent electric resistivity, x the cell
length in the leg in x-direction, w the leg width and mleg the number of leg
elements in x-direction. The mean contact resistance is set to Rcontact = 0.53
0.13 , which has been determined for the six demonstrator modules [20].
Rleg
26
Then the solar-to-electricity efficiency solar is determined according to

Equations (3.1) and (3.2).
3.4.5
Boundary conditions and model parameters
The incident solar radiative flux from the ETHs HFSS is assumed to be
cc = 1.46
distributed uniformly and diffuse, and it was varied in the range qsolar
20 W cm-2. The cold plate temperature is set to Tc = 300 K. The geometric
parameter ranges were varied as follows: (1) leg length l = 4 15 mm, (2) leg
width w = 3 6 mm, and (3) distance between neighboring legs d = 1 10.7
mm. As the total absorber surface per leg must be the same for 3D (module)
and 2D (model) geometries, the distance d between the legs for 3D is
transformed to d* for 2D. For approach (a), the extinction coefficient was
varied in the range abs = 102 105 and the albedo abs = 0 0.5. The general
model parameters used for the simulations are listed in Table 3-1.
3.4.6
Numerical solution
The FV technique is applied to discretize the coupled governing Equations

(3.3) to (3.16) and to solve the partial differential equation (PDE) system
iteratively using the successive over-relaxation (SOR) method implemented in
FORTRAN [69, 70]. The algorithm is repeated until the convergence criterion
Th,J i , j Th,J i,1j

J
Th,i , j
d 106
(3.17)
for all hot plate temperatures Th,i,j after iterations as well as the overall energy
balance
Qsolar Qreradiation Qconvection Qconduction Qradiation,out

Qsolar
d 103
(3.18)
are satisfied. A convergence study indicated an optimal trade-off between

accuracy and computational time for a grid containing 425 elements.
27
Parameter
Value
1.46 20
Unit
W cm-2
Source
measured / model parameter
graphite
Al2O3
p-type
n-type
abs
abs
4 - 15
3-6
1 10.7
2.167 48.026
8 30
0.95
0.95
0.3
0.7
0.7
1e2 1e5
0.0 - 0.5
mm
mm
mm
mm
mm
m-1
-

model parameter
[44]
[44]
[44]
assumed
assumed
model parameter
model parameter
N plate
8.626 34.451
W m-1 K-1
Appendix B.5, Table B-5
*
N plate
250
W m-1 K-1
assumed
1 2.5
1.75 3
0.025 - 0.05
0.02 0.036
120 260
170 -230
300
300
0.40-0.66
0.032954
2.5810-5
3.6610-5
W m-1 K-1
W m-1 K-1
cm
cm
V K-1
V K-1
K
K
W m-1 K-1
m2 s-1
m2 s-1
Appendix A.1, Figure A-1

assumed
assumed
[20] / assumed
[68]
[68]
[68]
cc
qsolar
l
w
d
d*
W
Dgraphite
kp-type
kn-type
p-type
n-type
Sp-type
Sn-type
T
Tc
Rcontact
kair
Qair
Dair
Table 3-1: General model parameters for simulations of directly irradiated modules.
3.5 Comparison between semi-transparent and opaque absorber

plate
The difference between the open-circuit voltage UOC as a function of the
cc calculated by the two model approaches (a) and (b) is
solar radiative flux qsolar
shown in Figure 3-9 for leg length l = 10 mm. Different radiation properties
(plate, plate) of the absorber plate have been tested. For plate and plate0,
no incoming radiation is transmitted through the absorber plate, and the
28
solution obtained by approach (a) converges to that for an opaque absorber

plate obtained by approach (b). Since the absorber plate used in the
measurements can be well approximated by an opaque surface a fact to be
shown in the model validation in the next subchapter only approach (b) is
applied in the following simulations.
0.45
0.4
0.35
oc
0.3
0.25
0.2
Eplate = 102, Zplate = 0.5
0.15

0.1
0.05
1.5
opaque
2.5
3.5
4
qccsolar [W cm-2]
4.5
5.5
Figure 3-9: Simulated open-circuit voltage as a function of solar radiative flux for leg
length l = 10 mm for the approach (a) with plate = 102, plate = 0.5; plate = 103,
plate = 0.5; plate = 104, plate = 0.1; plate = 105, plate = 0.0; and for the opaque
approach (b).
3.6 Model validation

The experimentally measured and simulated open-circuit voltages UOC as a
cc are shown in Figure 3-10 for leg
function of the solar radiative flux qsolar
lengths l = 4, 5, and 10 mm. The x-axis error bars of the experimentally
measured date points represent the uncertainty of the determination of the
incident solar radiative flux. A reasonably good agreement is observed, except
cc > 8 W cm-2), where the model predicts 15%
for l = 4 mm and high fluxes ( qsolar
higher values. This discrepancy can be attributed to the insufficient cooling of
the cold plate at high fluxes, as evidenced by a rise of its temperature (see
Figure 3-4), which in turn caused higher absorber plate temperatures and,
29
consequently, higher re-radiation losses. Thus, the temperature difference
across the legs is shifted towards higher temperatures and reduced due to the
higher re-radiation.
The simulated solar-to-power efficiencies solar as a function of the solar
cc are shown in Figure 3-11 (together with the experimentally
determined solar shown in Figure 3-7). The simulations reflect the uncertainty
of the contact resistance (Rcontact = 0.530.13 ) for the highest (Rcontact = 0.66
) and lowest contact resistance (Rcontact = 0.4 ). The simulated values are in
the same range as the measured ones, except for l = 4 mm and high fluxes
cc >8 W cm-2), where the discrepancy results from an overestimation of UOC
( qsolar
(see Figure 3-10).
0.5
0.45
0.4
0.35
0.25
oc
[V]
0.3
0.2
l = 10mm, Exp
l = 10mm, Sim
l = 5mm, Exp
l = 5mm, Sim
l = 4mm, Exp
l = 4mm, Sim
0.15
0.1
0.05
0
6
-2
qccsolar [W cm ]
10
12
Figure 3-10: Simulated and experimental open-circuit voltages as a function of solar

radiative flux for leg lengths l = 4, 5, 10 mm.
3.7 Heat transfer analysis

The percentage of incident solar radiation Qsolar transferred by the different
heat transfer modes is shown in Figure 3-12 for two cases where the maximum
temperature on the surface on top of the absorber plate was 900 K: (1) solar
cc = 6 W cm-2 and leg length l = 10 mm, and (2) qsolar
cc = 10
W cm-2 and l = 5 mm. In both cases, the heat losses by re-radiation and free
convection from the absorber plate represent more than 70% of Qsolar. About
30
20% of Qsolar is transferred by conduction through the legs, and <10% is lost by
radiation to the cold plate.
0.1
Rcontact = 0.4 :
0.09
Rcontact = 0.4 :
0.08
K solar [%]
l = 10mm, Exp
l = 10mm, Sim
l = 5mm, Exp
l = 5mm, Sim
l = 4mm, Exp
l = 4mm, Sim
0.07
Rcontact = 0.4 :
Rcontact = 0.66 :
Rcontact = 0.66 :
0.06
Rcontact = 0.66 :
0.05
0.04
0.03
6
7
qccsolar [W cm-2]
10
11
Figure 3-11: Simulated and experimentally measured solar-to-electricity efficiencies as

a function of the solar radiative flux for leg lengths l = 4, 5, 10 mm (experimental data
from Figure 3-7). Simulations for contact resistances Rcontact = 0.4 and 0.66 .
Figure 3-13 shows the simulated temperature distribution in a p-type

thermoelement leg of length l = 10 mm obtained for a solar radiative flux
cc = 6 W cm-2, where the module is operated at the maximum temperature
qsolar
difference of 600 K. A similar distribution is obtained for an n-type leg and
other leg lengths. Along the leg (x-axis), the profile is linear, as corroborated by
the experimental data (see Figure 3-6). Perpendicularly (y-axis), the
temperature is almost uniform, with a slightly higher temperature at the surface
because of the radiative exchange among legs and plates. The small
temperature gradient indicates that this radiative heat exchange is not
predominant, which is confirmed by the fact that <10% of Qsolar is lost by
radiation to the cold plate (see Figure 3-12). Further, the simulations have
shown that the absorber plates are quasi isothermal as the difference between
maximum and minimum temperature in the absorber plate was <10% of the
maximum temperature.
31
Figure 3-12: Percentage of incident solar radiation transferred by the different heat
transfer modes: conduction through the thermoelement legs, re-radiation and convective
cc = 6 cm-2,
losses from the hot plate, and radiation to the cold plate; for two cases: qsolar
cc = 10 cm-2, l = 5 mm.
l = 10 mm and qsolar
Figure 3-13: Simulated 2D temperature profile in p-type thermoelement leg for l = 10

cc = 6 cm-2, and temperature difference of 600 K.
mm at solar radiative flux qsolar
3.8 Geometric analyses

The heat transfer model is used to investigate the solar-to-electricity as a
function (1) of the leg length, and (2) of the packing of the legs. For the
following simulations the black coating of the absorber plate (Dgraphite = 0.95) is
32
assumed to be stable for all temperatures. The contact resistance is set to

Rcontact = 0.55 .
3.8.1
Variation of leg length
The dimensions of the 4-leg modules are defined by a leg width w = 4.5
mm and a plate width W = 30 mm. The leg length was varied in the range
cc
l = 5 - 15 mm and the solar radiative fluxes in the range qsolar
2 - 10 W cm2 .
The values of the model parameters used for these simulations are shown in
Table 3-2. The remaining model parameters are listed in Table 3-1.
Parameter
cc
qsolar
l
w
W
d
d*
Rcontact
Value
2 10
5 15
4.5
30
10.5
45.5
0.55
Unit
W cm-2
mm
mm
mm
mm
mm
Table 3-2: Model parameters for simulations of directly irradiated modules:

variation of length.
Figure 3-14 shows the simulated solar-to-electricity efficiency solar as a

cc for leg lengths l = 5, 7.5, 10, 12.5 and
function of the solar radiative flux qsolar
cc = 4
15 mm. The highest solar = 0.081% is obtained for l = 7.5 mm at qsolar
W cm-2. Note that l = 7.5 mm is not optimal in the whole solar radiative flux
cc < 3 W cm-2, l = 10 mm is most efficient, whereas for qsolar
cc > 7
range. For qsolar
W cm-2, l = 5 mm is most efficient. Thus, for increasing solar radiative fluxes,
the optimal leg length decreases.
3.8.2
Packing of legs
For practical manufacturing purposes, it is assumed that the minimal leg

width is w = 3 mm and the minimal distance between neighboring legs d = 1
33
mm. The simulated dimensions of the modules are w = 3 6 mm and d = 1 3
mm for a leg length l = 7.5, yielding plate widths W = 8 18 mm. The solar
cc = 2 20 W cm-2. The values of the
radiative fluxes are varied in the range qsolar
model parameters used for these simulations are shown in Table 3-3. The
remaining model parameters are listed in Table 3-1.
0.085
l = 7.5 mm
0.08
l = 5 mm
Ksolar [%]
0.075
l = 10 mm
0.07
l = 12.5 mm
0.065
l = 15 mm
0.06
0.055
0.05
6
cc
-2
qsolar [W cm ]
10
Figure 3-14: Simulated solar-to-electricity efficiency as a function of solar radiative

flux for leg lengths l = 5, 7.5, 10, 12.5 and 15 mm.
Parameter
cc
qsolar
l
w
d
d*
W
Rcontact
Value
2 20
7.5
36
13
2.167 9
8 18
0.55
Unit
W cm-2
mm
mm
mm
mm
mm
Table 3-3: Model parameters used in simulation varying the leg width and the distance
between neighboring legs.
In Figure 3-15, the simulated solar-to-electricity efficiency solar is plotted

cc : (a) for d = 1 mm and w = 3, 4.5
as a function of the solar radiative flux qsolar
and 6 mm, and (b) for w = 3 mm and d = 1, 2 and 3 mm. The highest solar is
34
obtained (a) for w = 3 mm in the whole solar flux range; and (b) for d = 1 mm
cc = 8 20 W cm-2. The peak solar = 0.375% at qsolar
cc = 20
in the range qsolar
W cm-2 is obtained for w = 3 mm and d = 1 mm i.e. for the densest packing of
the thermoelement legs, which is the smallest leg width and distance between
neighboring legs.
(a)
0.4
0.35
w = 3 mm
0.3
w = 4.5 mm
Ksolar [%]
0.25
0.2
w = 6 mm
0.15
0.1
0.05
0
(b)
10
12
qccsolar [W cm-2]
14
16
18
20
18
20
0.4
0.35
d = 1 mm
d = 2 mm
Ksolar [%]
0.3
d = 3 mm
0.25
0.2
0.15
0.1
0.05
10
12
qcc
[W cm-2]
14
16
solar
Figure 3-15: Simulated solar-to-electricity efficiency as a function of the solar radiative

flux with l = 7.5 mm: (a) for d = 1 mm and w = 3, 4.5 and 6 mm, and (b) for w = 3 mm
and d = 1, 2 and 3 mm.
35
3.9 Summary and conclusion

The conversion of solar high-temperature heat was demonstrated using a
set of non-optimized 4-leg demonstrator modules with figure-of-merit ZT ~
0.05 exposed to concentrated solar radiation. A 2D numerical heat transfer
model coupling conduction, convection and radiation has been implemented
and validated based on experimental data. Two different approaches of
modeling the radiation have been applied, namely the Monte Carlo method
considering a semi-transparent absorber plate and the radiosity method
considering an opaque absorber plate. The absorber plate used in the
measurements can be well approximated by an opaque plate. The heat transfer
analysis for 4-leg modules operated at a maximum hot plate temperature of 900
K and with leg lengths l = 4 10 mm and absorber plates of W W t = 30
30 0.25 mm3 indicated that more than 70% of the incident solar power is lost
due to re-radiation and free convection from the surface on top of the absorber
plate, while 20% is conducted through the legs and <10% is lost by radiation to
the cold plate. Quasi 1D heat conduction is the predominant mode of heat
transfer through the legs, as corroborated by the measured and simulated linear
temperature profiles in the legs. The highest experimentally measured solar-toelectricity efficiency was solar = 0.083% for the leg length l = 10 mm and solar
cc = 4 W cm-2. The optimum leg length of the 4-leg module with leg
flux qsolar
width w = 4.5 mm is l = 7.5 mm, which results in a maximum solar-tocc = 4 W cm-2. Denser packing of
electricity efficiency of solar = 0.081% at qsolar
the thermoelement legs, namely a 4-leg module with leg width w = 3 mm,
distance d = 1 mm and leg length l = 7.5 mm (plate width W = 8 mm) leads to
solar = 0.4%. For smaller dimensions than simulated here, solar can be expected
to reach 0.5%.
4 A 1 kWth solar cavity-receiver prototype2

The previous chapter has shown the high re-radiation losses (60% of
incoming power) of TEC modules at high temperatures (900 K) on the hot side.
For systems operating with highly concentrated solar radiation at hightemperatures, a cavity-receiver design is a convenient method to reduce reradiation losses significantly [72]. A cavity-receiver is a well-insulated
enclosure with a small opening, the aperture, to let radiation in. Because of
multiple reflections among the inner walls, the fraction of the incoming energy
absorbed by the cavity exceeds the surface absorptance of the inner walls [73].
In this chapter, a solar cavity-receiver is packed with TEC modules which
are directly exposed to concentrated solar radiation. As it will be shown in the
following analysis, the advantages of the proposed design are twofold: (1) the
geometrical configuration allows for efficient capture of concentrated solar
radiation and significant reduction of the re-radiation losses; (2) the direct
irradiation of the TEC modules allows efficient heat transfer to the site,
bypassing the limitations associated with conduction heat transfer through the
walls of an opaque solar absorber (i.e., limitations imposed by the materials
with regards to maximum operating temperature, thermal conductivity, and
resistance to thermal shocks). A 1 kW th solar cavity-receiver is fabricated as a
proof-of-concept of the proposed configuration. A heat transfer model is
formulated and validated in terms of open-circuit voltages measured as a
function of the mean concentration ratio over the aperture, and further applied
to analyze the heat transfer.
Material from this chapter has been published in [71]: C. Suter, P. Tome, A. Weidenkaff,
and A. Steinfeld, "A solar cavity-receiver packed with an array of thermoelectric converter
modules," Solar Energy, vol. 85, pp. 1511-1518, 2011.
38
A 1 kWth solar cavity-receiver prototype
4.1 Advantage of cavity design

As mentioned above, the cavitys apparent absorptance 3 Dapparent exceeds
the inner surface absorptivity D and, consequently, increases its ability to
absorb incoming radiation. Dapparent has been calculated for cylindrical, conical,
and spherical geometries having diffusely and specularly reflecting inner walls
[65, 74, 75]. Smaller apertures reduce re-radiation losses but intercept less
sunlight. Consequently, the optimum aperture size becomes a compromise
between maximizing radiation capture and minimizing radiation losses [76]. To
some extent, the aperture size may be reduced with the help of a non-imaging
secondary concentrator, e.g. a compound parabolic concentrator (CPC), placed
at the aperture in tandem with the primary concentrating system [77]. Examples
of relevant studies include a cavity containing an array of tubular absorbers
[78] and two cavities in series [79, 80]. Figure 4-1 shows Dapparent as a function
of D for incoming radiation uniformly distributed over the aperture for a 2D
cavity with a width and a height of X = 56 mm and an aperture width a = 7.1,
14.1 and 28.2 mm. For D > 0.8 and a < 14 mm, Dapparent > 0.98 and the cavity
approaches a blackbody absorber.
4.2 Experimental setup

The same 4-leg modules as described in subchapter 3.1 and schematically
shown in Figure 3-1 are used in the 1 kWth solar cavity-receiver prototype. The
dimensions of the modules have been characterized [58]. In this chapter, mean
dimensions have been used. The plates have the dimensions W x W x t = 25
25 0.5 mm3. The legs have a length of l = 5 mm, a width of w = 5.5 mm, and
a distance from the neighboring legs of d = 7 mm.
A schematic of the cavity-receiver is shown in Figure 4-2. It comprises a
water-cooled 68 x 62 x 81 mm3rectangular parallelepiped (box) enclosure,
with a windowless rectangular aperture (Aaperture = 14.1 x 81 mm2) for the access
of concentrated solar radiation. A water-cooled 2D trough CPC, of
20 x 81 mm2-rectangular entrance and 45 half-acceptance angle, is
3
Fraction of energy flux emitted by a blackbody surface stretched across the aperture that is
absorbed by the cavity walls.
39
incorporated in the aperture to augment the solar concentration ratio by a factor
of 1.4. The CPC further creates a more uniform radiation distribution on the 4leg modules, which is crucial for having a uniform temperature over each
module. The cavity contains an array of 18 modules arranged on two lateral
rows and one central row, with each row containing 2 bottom modules, 2 lower
modules, and 2 upper modules. The modules are arranged with 2 mm spacing
to neighboring modules for constructive reasons. The lateral side walls were
insulated with Al2O3. The designed maximum mean solar concentration ratio at
the CPC entrance is C = 600 suns for a solar power input of 1 kW and a hot
plate temperature of about 900 K.
1
0.9
D apparent [-]
0.8
0.7
0.6
a = 28.2 mm
a = 14.1 mm
a = 7.07 mm
0.5
0.4
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
D [-]
Figure 4-1: Apparent absorptance of a 2D cavity as a function of the absorptivity of its

inner walls for incoming radiation uniformly distributed over the aperture. Baseline
dimension is: cavity width/height X = 56 mm. The parameter is the aperture width:
a = 7.1, 14.1 and 28.2 mm.
Experimentation was carried out at the ETHs HFSS. A description of the

HFSS is given in subchapter 3.2. The peak radiative flux was measured as a
function of the arc current by a thermogage (Thermogage Circular Foil Flux
Transducer TG1000-1, accuracy 3% [60]) placed at the focal point. The
radiative flux distribution was measured optically on an Al2O3-plasma-coated
Lambertian target by a CCD camera equipped with neutral filters and calibrated
by a Kendall radiometer (error 8%) and the thermogage. The solar cavityreceiver was positioned with the CPC entrance at the focal plane of the HFSS.
40
Measurements include (1) the open-circuit voltage of each 4-leg module, and
(2) the power outputs of all 4-leg modules connected electrically in series
measured by a source-sink (Hcherl & Hackl source NL series 100 880
Watt). The hot plate temperatures of all TEC modules were measured and
controlled not to exceed the maximum operating temperature of 900 K.
(a)
(b)
Figure 4-2: Schematic of the experimental setup at ETHs HFSS. (a) cross-section of
the 1 kWth solar cavity-receiver prototype with CPC, cooler, and bottom, lower and
upper 4-leg modules. (b) longitudinal section with CPC, cooler, and central and lateral
4-leg modules. The dimensions of the cavity-receiver are: length L = 81 mm,
width/height X = 56 mm and aperture a = 14.1 mm.
41
4.3 Experimental results

Figure 4-3 shows the measured radiative flux distribution at the focal
plane normalized by the peak radiative flux, which varied between 128 and 657
suns. The 81 x 20 mm2-dotted rectangle indicates the entrance of the CPC.
Integration of the radiative flux over the CPC entrance yields the total solar
radiative power input, Qsolar. Accounting for 2.5mm positioning accuracy of
the CPC and the error of the calibrated CCD camera, the error in Qsolar is
11.8%.
15
0.6
0.6
0.9
-15
-50
0.8
0.9
0.9
-5
-10
0.6
0.8
5
0
0.6
0.8
0.9
y [mm]
10
0.9
0.8
0.9
0.8
0.6
0.6
0.4
-40
-30
-20
0.8
0.6
-10
10
0.6
20
30
40
50
x [mm]
Figure 4-3: Solar radiative flux distribution measured at the HFSS focal plane,
normalized by the peak radiative flux which varied between 128 and 657 suns. The 81 x
20 mm2-dotted rectangle indicates the entrance of the 2D trough CPC.
The distribution of the incoming solar radiation in the cavity is calculated

by the 3D Monte Carlo (MC) ray-tracing technique [81, 82]. The MC model
simulated the HFSS, the CPC, and the cavity as depicted in Figure 4-4. The
following assumption were made: (1) directional distribution of the Ar arc tube;
(2) a total reflectivity of 0.8 for HFSS and CPC mirrors with an average angular
error of 6 mrad for specular reflection and a probability for diffuse reflection of
0.05; (3) a total absorptivity of 1 for the inner cavity surfaces; and (4)
neglecting the spacing between the absorber plates of the modules.
The losses in the CPC were determined to be ~33.3%. The distribution of
incoming radiation within the cavity was calculated and approximated
uniformly and with 50%, 26% and 24% of the incident radiation on the bottom,
lower and upper modules, respectively. 15% of the incoming radiation Qsolar
(passing the CPC aperture and going into the cavity) was assumed to be lost
through the spacing between the modules.
42

0.5
0.4
z [m]
0.3
0.2
0.1
-0.1
-0.3
-0.2
-0.1
0
x [m]
0.1
0.2
0.3
Figure 4-4: Cross-section of the HFSS with the solar cavity-receiver and CPC placed at
the focal plane (z = 0 m).
The measured open-circuit voltage UOC as a function of the mean solar

concentration ratio C in the aperture is shown in Figure 4-5 for all regions of
interest, i.e. central/lateral bottom, central/lateral lower and central/lateral upper
modules. The x-axis error bars are due to the uncertainty of incoming solar
radiative fluxes; the y-axis error bars are due to averaging over 2 experimental
runs and at least 2 modules. For all cases, UOC increases monotonically with
C . The difference between lateral and central modules is marginal, which

correlates with the relatively flat radiative flux map along the x-axis (see Figure
4-3). In contrast, the bottom modules exhibit the highest UOC, whereas the
upper modules exhibit the lowest UOC, which is in accordance with the
radiation distribution simulated by MC ray-tracing.
The solar-to-electricity efficiencies solar are calculated in terms of the
maximum power output Pmax and the solar radiation input Qsolar according to
Equation (3.1). Qsolar is defined as:
Qsolar
C I Aaapertur
aperture
(4.1)
where C is the mean concentration ratio in the aperture, I = 1 kW m-2 the direct
normal irradiation and Aaperture = 14.1 x 81 mm2 the aperture area. The
experimentally measured Pmax and solar as a function of C for all 18 modules
connected electrically in series are shown in Figure 4-6. Pmax and solar increase
43
monotonically with C . The peak efficiency is solar = 0.12 % for Pmax = 0.86 W
at C = 620 suns (Qsolar = 710 W) and a maximum hot plate temperature of Th =
900 K on the bottom modules. Remember that the measured solar was only
0.06% for a single directly irradiated 4-leg module (leg length l = 5 mm)
exposed to a mean solar radiative flux of 8 W cm-2, as shown in Figure 3-7.
Thus, the cavity-receiver configuration was able to augment the solar-toelectricity efficiency by a factor of 2.16 vis--vis that obtained for a single
directly irradiated 4-leg module.
450
400
350
OC
[mV]
300
250
200
central bottom
lateral bottom
central lower
lateral lower
central upper
lateral upper
150
100
50
0
100
200
300
400
500
600
700
Figure 4-5: Experimentally measured open-circuit voltage as a function of the mean

solar concentration ratio in the aperture for central/lateral bottom modules,
central/lateral lower modules, and central/lateral upper modules.

Since the experimental results from the previous subchapter revealed only
marginal differences between lateral and central modules, a quasi 2D treatment
of the problem is justified. Accordingly, the heat transfer model considers only
one module row (i.e. 2 bottom, 2 lower and 2 upper modules) and assumes
symmetry planes at the lateral sides. The solar cavity-receiver configuration
with a width and a height of X = 56 mm and a length of L = 27 mm (1/3 of the
experimental cavity-receiver with L = 81 mm) is shown schematically in Figure
44
0.9
0.12
0.75
0.1
0.6
0.08
0.45
0.06
0.3
max
[W]
0.14
K solar [%]
4-7. The aperture width is a = 14.1 mm. The cavity contains N = 6 modules. In
the previous chapter, the heat transfer analysis for single directly irradiated
4-leg modules has shown 1D temperature profiles in the thermoelement legs
and negligible radiation exchange between the hot and cold plate (see
subchapters 3.3 and 3.7). Thus, the considered heat transfer modes are 3D
radiative and 2D convective exchange within the cavity, 1D conduction through
the thermoelement legs of the modules, and 2D convective heat losses out of
the cavity. Further, it is assumed: (i) the graphite-coated Al2O3 absorber plates
are opaque, gray and diffuse [44], and have a uniform temperature Th,i; (ii) the
inner CPC insulation surface has the uniform temperature Tins; (iii) the cold
plates and the outer CPC insulation surface have the uniform cold temperature
Tc, (iv) the gas phase is radiatively non-participating and has a refractive index
equal to unity; (v) the radiative heat transfer between hot/cold plates and
thermoelement legs is neglected; and (vi) only open-circuit (J = 0) voltages are
simulated.
Ksolar
P
0.04
0.02
100
200
300
400
500
600
max
0.15
0
700
Figure 4-6: Experimentally measured maximal power output and efficiency as a

function of the mean solar concentration ratio in the aperture for all 18 4-leg modules
connected electrically in series.
4.4.1
Heat conduction
The general steady-state energy conservation equations in differential

form for the thermoelement legs are given in Equations (3.4) and (3.5). With
the assumption of 1D heat conduction in the thermoelement legs, linear
45
dependency of the thermoelectric material properties on temperature and opencircuit (J = 0), these equations can be algebraically solved.
Figure 4-7: Schematic of the model domain: cross-section of a rectangular

parallelepiped (box) cavity-receiver with rectangular aperture and insulated 2D (trough)
CPC. The dimensions are: cavity width/height X = 56 mm and length L = 27 mm
(perpendicular to the cross-section plane), aperture width a = 14.1 mm. The cavity
contains an array of N = 6 modules. Indicated are: incoming solar radiation (Qsolar), reradiation and convective losses (Qloss), radiation exchange (Qradiosity), heat convection
(Qconvection), heat conduction (Qconduction,i) and open-circuit voltage (UOC,i) per module i,
hot plate temperatures (Th,i), inner CPC insulation surface temperature (Tins) and cold
temperatures (Tc) of cold plates and outer CPC insulation surface.
Then the 1D heat conduction in a thermoelement leg for module i in terms

of the hot (Th,i) and cold (Tc,i) plate temperatures, the thermal conductivity
Nleg,m,i evaluated at the middle leg temperature Tm,i
h,i
Tc,i / 2 , the leg
length l, and the cross-section area of the legs Aleg = w2 reduces to:
Qleg,i
Ei Aleg
N leg,m,i
l
Th,i Tc,i
(4.2)
46
where
dN
eg,m,i
l
(4.3)
Th,i Tc,i /12

Ei 1
N leg,m,i dTleg,m,i
is the correction factor accounting for the linear dependency of the material
properties on temperature [83].
Further the open-circuit voltage UOC,leg,i across a leg of module i is given
by:
2
UOC,leg,i
Sleg,m,i Th,i Tc,i
(4.4)
where Sleg,m,i is the Seebeck coefficient evaluated at Tm,i. Finally, the opencircuit voltage of the cavity-receiver UOC is expressed by:
N
U OC
M U OC,leg,i
(4.5)
i =1
where M is the number of legs per module and N the number of TEC modules.
Thermoelectric properties of the thermoelement legs have been
approximated by linear correlations, see appendix B.2, Table B.2.
4.4.2
The problem of radiation heat exchange is solved by the radiosity method

(enclosure theory) [65], yielding a system of equations in terms of the net
cc
radiative flux qradiation,j
and temperature Tsurface,j for surface element j:
N 2
G kj
1 H j
4
cc
F
q

radiation,j G kj Fk - j V Tsurface,j
k- j
(4.6)
Hj
j 1 Hj
j 1
for k 1... N 2
where kj is the Kronecker function, N+2 the number of surface elements where
the N hot plates, the insulation and the aperture are participating surface
elements (spacing between modules is neglected). The view factors Fk-j are
calculated by applying reciprocity relations (AjFj-k = AkFk-j), the enclosure
N 2
criterion ( Fk j
1 ), and tabulated view factors for parallel and
j 1
perpendicular plates [66].
47
4.4.3
Natural heat convection
Natural convection has been considered in previous studies for large

cubical cavities [84] and for spherical, hemispherical, and cylindrical cavityreceivers [78, 85-87]. Here, transient 3D natural convective heat transfer
simulations have been performed for an upward-facing parallelepiped cavity
using the computational fluid dynamics (CFD) software ANSYS CFX 12.0
[88]. Steady-state simulations are inadequate due to the unsteady buoyancy
flow in cavities [89].
The simulated air domain is shown in Figure 4-8 and includes the cavity
volume and an environment on top of the cavity. The size of the environment
was increased until a further increase had no significant effect on the
convective heat transfer. The final size of the environment was 100 mm in xdirection, 55 mm in y-direction and 50 mm in z-direction. A time step of 0.1s
was found to be sufficient for producing accurate results at reasonable
computational times. In each time step a convergence criterion of 10 4 was
imposed on the residuals of the continuity, momentum, energy and mass
equations. The grid consisted of 250,000 elements, which is a trade-off between
computational time and accuracy. The boundary conditions were set as follows:
(1) The inner cavity walls were assumed to have a constant temperature, which
was varied in the range 400 K < Tsurface < 1000 K. (2) The CPC walls, the top of
the CPC and the gas boundaries of the environment were set to Tc = 300 K and
T = 300 K, respectively. (3) All walls had no-slip conditions and the gas
boundaries had opening conditions with p = 1 bar. As initial condition the air
temperature was set to 300 K.
A period of 5s was considered in the simulation. Quasi steady-state
convective heat transfer (Qconvection) was attained after 2s, yielding the Nusselt
correlation at the absorber plates via Equation (3.14):
0.11
Nu W h W / N air 0.001 Ra surface

(4.7)
Tf
where h is the convective heat transfer coefficient, W the width of the cavity,
Nair the thermal conductivity of air, and RaW the Rayleigh number. Equation
(4.7) is valid for 3.6106 < RaW < 8.0106 and consistent with studies of natural
convection in similar cavities [78, 85]. Air properties were calculated at
Tfilm 0.5 (Tsurface Tf ) by using correlations, see Appendix B.3, Table B-3.
0.76
W
48
Figure 4-8: Cross-section of simulated air domain including cavity with surface
temperature Tsurface, CPC with cold temperature Tc and environment with opening
boundary at temperature and pressure T and p , respectively.
Figure 4-9 shows the parity plot of the Nusselt numbers Nu for the values
calculated by the correlation and those extracted from the CFD simulations.
The high coefficient of determination in this parity plot (R2 = 0.911) implies
good agreement between the two methods, further supported by the mean
relative error of 5% [90].
4.4.4
The solar radiative flux entering through the aperture is distributed over
the bottom, lower and upper modules according to the fractions 50%, 26%, and
24%, respectively. These fractions have previously been extracted from MC ray
tracing (see subchapter 4.3). The mean solar concentration in the aperture is
varied in the range C = 121 620 suns. The temperatures of the cold plates
and the outer CPC insulation surface was set to Tc = 300 K. The model
parameters used are listed in Table 4-1.
49
4
Nu (Correlation)
3.5
2.5
1.5
1.5
2.5
3
Nu (CFD)
3.5
Figure 4-9: Parity plot of the Nusselt number for the natural convective heat transfer for
the values calculated by the correlation and those extracted from CFD simulations. The
coefficient of determination is R2 = 0.911 and the mean relative error 5%.
4.4.5
Numerical solution
The coupled governing Equations (4.2) to (4.7) were implemented in a

FORTRAN code [70], and solved iteratively until the convergence criterion:
Th,J i Th,J i1

J
Th,i
d 106
(4.8)
for all hot temperatures Th,i after iterations as well as the overall energy
balance
Qsolar Qreradiation Qconvection Qconduction Qloss,CPC+walls

Qsolar
d 103
(4.9)
are satisfied.
4.5 Model validation

Model validation is accomplished in terms of the measured open-circuit
voltage UOC, since this quantity can be measured most reliably and is directly
proportional to the mean temperature difference across the thermoelement legs.
50
The experimentally measured and averaged UOC as well as the simulated UOC
for bottom, lower and upper modules are shown in Figure 4-10 as a function of
the mean solar concentration ratio C in the aperture. The x-axis error bars are
due to the uncertainty of incoming solar radiative fluxes; the y-axis error bars
are due to averaging over 2 experimental runs and at least 2 modules. A slight
under-prediction for low radiative fluxes is observed, which can be attributed to
the linear temperature dependence of the material properties used in the model.
In contrast a slight over-prediction is observed for high radiative fluxes as the
cold plate is insufficiently cooled, resulting in a decrease of the temperature
difference and, consequently, a decrease in UOC. The same phenomenon could
be observed for single directly irradiated 4-leg modules (see subchapter 3.6).
Parameter
cc
qsolar
l
w
W
t
N
M
a
X
L
Dgraphite
graphite
ins
kins
kp-type
kn-type
p-type
n-type
Sp-type
Sn-type
T
Tc
Value
121 620
Unit
kW m-2
Source
model parameter
5.045
5.546
25
0.5
6
4
14.1
56
27
0.95
0.95
0.3
0.2
1 2.5
1.75 3
0.025 - 0.05
0.02 0.036
120 260
-170 -230
300
300
mm
mm
mm
mm
mm
mm
mm
mm
W m-1 K-1
W m-1 K-1
W m-1 K-1
cm
cm
V K-1
V K-1
K
K
measured
measured
measured
measured
measured
measured
measured
measured
measured
[44]
[44]
assumed
assumed
Appendix B1, Table B-1
assumed
assumed
Table 4-1: General model parameters for 1 kWth cavity-receiver simulations.
51
450
400
350
OC
[mV]
300
250
200
bottom, Exp
bottom, Sim
lower, Exp
lower, Sim
upper, Exp
upper, Sim
150
100
50
0
100
200
300
400
500
600
700
Figure 4-10: Simulated and experimentally measured open-circuit voltage as a function

of the mean solar concentration ratio in the aperture for bottom, lower and upper
modules.
4.6 Heat transfer analysis

For the incoming solar radiation Qsolar = 710 W ( C = 620 suns), the
bottom, lower and upper modules reached 847, 716 and 678 K, respectively.
The corresponding percentages of the solar radiation input transferred by the
different heat transfer modes are shown in Figure 4-11. 71% of Qsolar is
conducted through the thermoelement legs. 23% of Qsolar is lost by conduction
through the cavity and CPC walls, including losses to water-cooled surfaces.
Only 4% and 2% of Qsolar is lost by re-radiation and natural convection through
the aperture, respectively. Comparison with the heat transfer analysis of the
single directly irradiated 4-leg modules (see subchapter 3.7) indicates that the
cavity design reduces the re-radiation losses from 60% to 4% of Qsolar, while
increasing the useful heat conduction through the thermoelement legs from
20% to 71% of Qsolar.
52
Figure 4-11: Percentage of solar radiation input Qsolar transferred by different heat
transfer modes: conduction through the thermoelement legs, re-radiation and convective
losses through the aperture, and conduction losses through the cavity and CPC walls.
Qsolar = 710 W and C = 620 suns.
4.7 Summary and conclusion

A 1 kWth solar cavity-receiver prototype containing an array of 18 4-leg
modules with figure-of-merit ZT ~ 0.05 was designed, fabricated and
experimentally tested under simulated concentrated solar radiation. The
modules had a leg length and a width of l = 5.5 mm and w = 5 mm,
respectively, and absorber plate dimensions of W W t = 25 25 0.5 mm3.
The peak solar-to-electricity efficiency was solar = 0.12% for Pmax = 0.86 W at
C = 620 suns (Qsolar = 710 W) and a maximum hot plate temperature of

Th = 900 K. The measured solar-to-electricity efficiency of a single directly
irradiated 4-leg module (leg length l = 5 mm), also operated at the maximum
Th = 900 K, was only solar = 0.06%, which means that the cavity design could
double solar. A 2D heat transfer model of the cavity was developed and
validated in terms of open-circuit voltages experimentally measured as a
function of the mean solar concentration ratio in the aperture. For
Qsolar = 710 W ( C = 620 suns, Th,max = 900 K), the heat transfer analysis
indicated that 71% of the incoming solar power input was conducted through
the thermoelement legs, as compared to 20% for single directly irradiated 4-leg
modules. Re-radiation losses were reduced from 60% to 4% of the solar power
input, showing the significant advantage of the cavity design.
5 Towards a 50 kWth solar cavity-receiver

In the previous chapter, the heat transfer analysis of the 1 kWth solar
cavity-receiver has shown that the re-radiation losses were reduced to 4% of the
incoming solar radiation compared to 60% for single directly irradiated 4-leg
modules, showing the high potential of the cavity design. However, the
measured maximum solar-to-electricity efficiency was only 0.12% due to the
low figure-of-merit of ~0.05 of the perovskite-type oxide thermoelectric
materials used in the 4-leg modules. An approach to increase the figure-ofmerit is to use cascaded modules when a large temperature difference is present
[91]. In this chapter, a cubic 50 kWth solar cavity-receiver packed with
cascaded modules is considered. The heat transfer model of the 1 kWth solar
cavity-receiver from previous chapter is extended, reformulated and applied to
optimize the geometry of the 50 kWth cavity-receiver. Furthermore, it is used to
investigate the effect of the maximum module efficiency of and the maximum
cavity temperature on the solar-to-electricity efficiency.
5.1 Idea of cascading

In order to achieve high efficiencies, large temperature differences as well
as a high figure-of-merits are required. Figure 5-1 shows the figure-of-merit as
a function of the temperature for several n-type materials showing the strong
temperature dependency. Thus, it is not desirable or even possible to use the
same material throughout an entire, large temperature difference [91]. Ideally,
thermoelement legs with sandwiched materials would be used, each material
operating in its most efficient temperature range. One approach is to cascade
modules as depicted schematically in Figure 5-2. The module is operated
between temperatures TK+1 and T1 and sandwiched with K module units called
stages connected thermally in series [92]. Each stage k consists of
thermoelement legs with different materials and has plates at temperatures Tk+1
and Tk. The cascading design allows operating each stage k with an independent
54
Towards a 50 kWth solar cavity-receiver
electric circuit, avoiding incompatibility between different thermoelectric

materials, which would reduce the efficiency of the module significantly [93].
However, cascading is much more difficult to implement than other separation
methods, e.g. segmentation [92].
Figure 5-1: Figure-of-merit as a function of the temperature for several n-type

materials, extracted from Ref. [92].
Figure 5-2: Schematic of a cascaded module consisting of K stages and operated

between temperature TK+1 and T1, with intermediate temperatures Tk.
5.2 Cascaded dual-stage modules

For the 50 kWth solar cavity-receiver, cascaded modules with two
thermoelement leg stages (K = 2) are considered. They are represented by three
Al2O3 plates with surface W x W = 10 x 10 cm2 sandwiching a low- and a high
temperature stage, as depicted in Figure 5-3. The thermoelement legs of the
low- and high-temperature stage consist of Bi-Te alloys having a figure-of-
55
merit ZT = 1 [38] and perovskite-type oxides having a ZT in the range of 0.36
1.7, respectively. Perovskite-type oxide materials with ZT > 0.3 have already
been synthesized with high potential for improvement [48]. In this study it is
assumed that the upper limit will be ZT = 1.7. Each stage contains M = 2822
thermoelement legs (1411 p/n-type thermoelement leg pairs), each leg having a
cross-section of Aleg = 1.05 x 1.05 mm2. The dimensions of the legs crosssection area and packing density (number of legs per plate area) are common in
standard Bi-Te modules [40]. The thermoelement leg lengths llow and lhigh of the
low- and high-temperature stage, respectively, are allowed to vary as model
parameters. The cascaded modules are operated between the hot plate
temperature (Th) and the cold plate temperature (Tc) with the intermediate plate
temperature (Tint) between the two stages. The top of the hot plate is blackcoated to augment the absorptivity of solar radiation (total absorptance
black = 0.95).
The maximum module efficiency is given by [92]:
Kmod,dual Kmod,high (1 Kmod,high ) Kmod,low
(5.1)
where mod,high and mod,low are the maximum module efficiencies of the highand low-temperature stage, respectively, calculated according to Equation (2.3).
Figure 5-3: Schematic of the cascaded dual-stage module: three Al2O3 plates
sandwiching a low- and a high-temperature stage with variable thermoelement leg
length llow and lhigh, respectively. The cascaded module is operated between the
hot plate temperature Th and the cold plate temperature Tc with the intermediate
plate temperature Tint.

The 50 kWth solar cavity-receiver configuration is similar to the 1 kWth
cavity-receiver configuration (see subchapter 4.4) as shown schematically in
Figure 5-4. It consists of a square parallelepiped (box) cavity-receiver with a
56
windowless circular aperture for the access of concentrated solar radiation. The
dimensions are given by the cavity width X, the height H, and aperture diameter
a. The cavity-receiver is packed with N cascaded dual-stage modules. The heat
transfer model for the 1 kWth cavity-receiver (see subchapter 4.4) is extended as
follows: (1) incorporating the dual-stage approach of the cascaded modules, (2)
adjusting the geometry to the 3D cavity design, (3) considering electric currents
with independent circuits for each stage, and (4) using constant thermoelectric
properties. Further the convective heat losses are neglected as the heat transfer
analysis of the 1 kWth solar cavity-receiver has shown convective heat losses of
only 2% of the solar power input (see subchapter 4.6).
Figure 5-4: Schematic of the model domain: cross-section of the square parallelepiped
(box) cavity-receiver with circular aperture and insulated CPC. The dimensions are:
cavity width X, height H, and aperture diameter a. The cavity-receiver contains an array
of N cascaded dual-stage modules. Indicated are: incoming solar radiation (Qsolar), reradiation losses (Qloss), radiation exchange (Qradiosity), hot plate temperatures (Th,i), inner
insulation surface temperature (Tins) and cold temperatures (Tc) of cold plates and outer
CPC insulation surface.
57
5.3.1
Heat conduction
The general steady-state energy conservation equations in differential

form for the thermoelement legs are given in Equations (3.4) and (3.5). The
heat transfer equations will only be illustrated for the high-temperature stage of
module i due to the complete analogy with the low-temperature stage. With the
consideration of the dual-stage approach of the cascaded modules, the
assumption of 1D heat conduction in the thermoelement legs, constant
thermoelectric material properties and consideration of electric currents, these
equations can be algebraically solved. Then the 1D heat conduction in a
thermoelement leg in the high-temperature stage for module i at the hot and
cold plate reduces to [83]:
Uhigh
N high
Qhigh,leg,h,i Shigh Th,i J high lhigh
J high 2 / 2 Aleg
(Th,i Tint,i ) (5.2)
Aleg
lhigh
and
Qhigh,leg,c,i
Shigh Tint,i J high lhigh
Uhigh
Aleg
J high 2 / 2 Aleg
N high
lhigh
(Th,i Tint,i ) (5.3)
where Th,i is the hot plate temperature, Tint,i the intermediate plate temperature,
and Jhigh the high-temperature stage current. As aforementioned, the heat
transfer equations for the low-temperature stage are analogous to the ones
presented above.
5.3.2
Radiation heat exchange is solved by the radiosity method (enclosure

theory) [65], yielding a system of equations in terms of the net radiative fluxes
cc
and surface temperature Tsurface,j for element j:
qradiation,j
N 2
G kj
1 H j
4
cc
G kj Fk - j V Tsurface,
Fk - j
qradiation,

j
j
(5.4)
Hj
j 1 Hj
j 1
for k 1... N 2
where kj is the Kronecker function, N+2 the number of surface elements where
the N hot plates, the insulation and the aperture are participating surface
elements.
N 2
58
The view factors Fk-j are calculated by MC ray-tracing using Nrays = 109
rays [64]. They are approximated by:
Fk - j |
N rays,absorbed, k ,j
(5.5)
N rays,emitted,k
where Nrays,emitted,k is the number of rays emitted from surface k and Nrays,absorbed,k,j
is the number of rays emitted from surface k and absorbed by surface j.
5.3.3
The equations for voltages, resistances and power outputs will only be
illustrated for the high-temperature stages due to the complete analogy with the
low-temperature stage. The electric current is given by:
U high
(5.6)
J high
Rlegs,high Rcontact,high Rload,high
The voltage Uhigh is given by:
N
U high
M Shigh (Th,i Tint,i )
(5.7)
i =1
where M is the number of thermoelement legs per module. The material

resistance Rlegs,high is determined by:
Rlegs,high
N M U high
lhigh
Aleg
(5.8)
The contact resistance is calculated as:
Rcontact,high
2 N M
U*
Aleg
(5.9)
where * is the specific contact resistance. The matched load is given by:
Rload,high Rcontact,high Rlegs,high
(5.10)
Then the maximum power output of all high-temperature stages is calculated
by:
Pmax,high
2
J high
Rload,high
(5.11)
As aforementioned, the equations for voltages, resistances and power outputs

for the low-temperature stages are analogous to those presented above.
Evidently, the maximum power output of the cavity-receiver is:
59
Pmax
Pmax,high Pmax,low
(5.12)
Finally, the solar-to-electricity efficiency solar in terms of Pmax and the solar
radiation input Qsolar is given according to Equation (3.1).
5.3.4
The diffuse solar radiation input was set to Qsolar = 50 kW with a uniform
flux distribution and a mean concentration C = 1500 over the aperture. Thus,
the aperture area is Aaperture = 0.033 m2 and the aperture diameter a = 20.6 cm
according to Equation (4.1). The maximum hot plate temperature was varied in
the range Th,max = 900 1200 K; the maximum intermediate plate temperature
was assumed to be Tint,max = 500 K. The temperature of the cold plates and the
outer CPC insulation surface was set to Tc = 300 K. The geometric parameters
were varied as follows: (1) cavity height H = 30 90 cm and (2) width X = 40
80 cm. The thermoelement leg lengths of the low- and high-temperature stage
lhigh and llow, respectively, were adjusted to satisfy the set of prescribed
maximum hot and intermediate plate temperatures. The Seebeck coefficient of
the high-temperature stages is varied in the range Shigh = 90 196 V K-1. The
general model parameters are listed in Table 5-1.
5.3.5
Numerical solution
The coupled governing Equations (5.2) to (5.12) are implemented in a

FORTRAN code [70], and solved iteratively until the convergence criterion
(Th,i Th,-1
i )
h,i
d 109
(5.13)
for all hot plate temperatures Th,i after iterations as well as the overall energy
balance
Qsolar Qreradiation Qconduction Pmax

d 103
Qsolar
are satisfied.
(5.14)
60

Parameter
Qsolar
Value
50
1500
Unit
kW
-
Source
set
set
H
X
a
N
M
Th,max
Tint,max
Tc
llow
llow
lhigh
Nrays
low
high
low
high
Slow
Shigh
*
black
ins
ins
lins
30 90
40 80
20.6
80 320
2822
900 1200
500
300
2 20
2 20
1 10
1109
1.25
1.25
13.5
13.5
200
90 196
510-9
0.95
0.2
0.2
10
cm
cm
cm
K
K
K
mm
mm
mm
W m-1 K-1
W m-1 K-1
m
m
V K-1
V K-1
m2
W m-1 K-1
cm
model parameter
model parameter
set
model parameter
set
model parameter
set
set
model parameter
model parameter
model parameter
set
assumed
assumed
assumed
assumed
assumed
model parameter
assumed [94]
assumed
assumed [44]
assumed
set
Table 5-1: General model parameters of the 50 kWth cavity-receiver simulations.
5.4 Simulation results

The heat transfer model was applied to optimize the geometry of the 50
kWth cavity-receiver within the investigated geometric parameter range, to
perform a heat transfer analysis in the cavity, and to investigate the effect of the
maximum module efficiency and the maximum hot plate temperatures on the
solar-to-electricity efficiency and the total thermoelement leg length.
61
5.4.1
Geometrical optimization and heat transfer analysis
The simulated cavity dimensions were varied as follows: height H = 30

90 cm and width X = 40 80 cm. The maximum hot plate temperature range
was Th,max = 900 1200 K. The Seebeck coefficient for the high-temperature
stage was held at Shigh = 90 V/K. The values of the model parameters used for
these simulations are shown in Table 5-2. The remaining model parameters are
listed in Table 5-1.
Parameter
H
X
Th,max
Shigh
Value
30 90
40 80
900 1200
90
Unit
cm
cm
K
V K-1
Source
model parameter
model parameter
model parameter
model parameter
Table 5-2: Model parameters of the 50 kWth cavity-receiver simulations

for geometrical optimizations.
Figure 5-5(a) shows the solar-to-electricity efficiency solar and the number
of modules N as a function of the cavity width X and height H, which were
varied simultaneously, for the maximum hot plate temperatures Th,max = 900
and 1200 K. The curves monotonically increase with X and H. A good trade-off
between high solar and low N is achieved for X = H = 60 cm where the model
predicts solar = 7.3% and 11.1% for Th,max = 900 and 1200 K, respectively, and
N = 180 modules. Figure 5-5(b) shows solar and N as a function of H for X = 60
cm for Th,max = 900 and 1200 K. N increases linearly with H, whereas solar has a
maximum at H = 50 cm for Th,max = 900 and at H = 60 cm for Th,max = 1200 K.
Thus, the optimum is realized in the case of H = 50 cm where less modules are
required than for H = 60 cm.
For the optimized cavity geometry of X = 60 cm, H = 50 cm and N = 156
modules, the model predictions are summarized in Table 5-3 for Th,max = 900 K
and Th,max = 1200 K. The model predicts the highest re-radiation losses of 6.7%
for Th,max = 1200 K, which is comparable to the 1 kWth solar cavity-receiver
prototype with re-radiation losses of 4%.
62

(a)
12
360
Th,max = 1200 K
240
Th,max = 900 K
180
120
2
40
(b)
300
Ksolar [%]
10
45
50
55
60
65
X and H [cm]
70
75
12
280
Th,max = 1200 K
10
240
Th,max = 900 K
200
N
Ksolar [%]
60
80
160
120
2
30
40
50
60
H [cm]
70
80
80
90
Figure 5-5: Solar-to-electricity efficiency and number of modules as a function of: (a)
the cavity width X and height H, and (b) the cavity height H for width X = 60 cm; for
maximum hot plate temperature Th,max = 900 and 1200 K.
Th,max Th,,min solar Qreradiation/Qsolar lhigh

lhigh
[K]
[K]
[%] [%]
[mm] [mm]
900 584
7.4 1.6
2.6
4.3
1200 1058 11
6.7
3.7
10.4
Table 5-3: Main results for the optimum 50 kWth cavity-receiver geometry:
X = 60 cm, H = 50 cm and N = 156.
63
5.4.2
Solar-to-electricity efficiency
In the following simulations the optimized cavity geometry was used

(W = 60 cm, H = 50 cm and N = 156). The Seebeck coefficient of the hightemperature stage and the maximum hot plate temperatures were varied in the
ranges Shigh = 90 196 V/K and Th,max = 900 1200 K, respectively. The
values of the model parameters used for these simulations are shown in Table
5-4. The remaining model parameters are listed in Table 5-1. For the
considered temperature and Seebeck coefficient ranges, the maximum module
efficiency is in the range mod,dual = 11.7 26% according to Equation (5.1).
Parameter
Value
Unit
Source
-1
Shigh
90 196
V K
model parameter
Th,max
900 1200
K
model parameter
H
50
cm
model parameter
X
60
cm
model parameter
Table 5-4: Model parameters of 50 kWth cavity-receiver simulations
for the analysis of the solar-to-electricity efficiency.
Figure 5-6 shows the solar-to-electricity efficiency solar as a function of

the maximum module efficiency mod,dual for Th,max = 900 1200 K. For all
cases, solar increases linearly with mod,dual and monotonically with Th,max,
implying a stronger dependency on the former. The model predicts a solar-toelectricity efficiency ranging from solar = 7.4% with the maximum power
output Pmax = 3.7 kW for mod,dual = 11.7% and Th,max = 900 K; to solar = 20.8%
with Pmax = 10.4 kW for mod,dual = 26% and Th,max = 1200 K.
Figure 5-7 shows the total thermoelement leg length (ltotal = lhigh + llow) as a
function of the solar-to-electricity efficiency solar for Th,max = 900 1200 K.
For all cases, ltotal increases monotonically with solar and Th,max. The
dependency on solar is attributed to the required increase of leg length in order
to compensate the increase of the Peltier heat flux. The total thermoelement leg
length ranges from ltotal = 7 mm for solar = 7.4% and Th,max = 900 K to
ltotal = 21.7 mm for solar = 20.8% and Th,max = 1200 K.
64

22
20
18
Ksolar [%]
16
14
12
Th,max = 900 K
10
Th,max = 1000 K
Th,max = 1100 K
Th,max = 1200 K
6
10
12
14
16
18
20
22
24
26
Kmod,dual [%]
Figure 5-6: Solar-to-electricity efficiency as a function of the maximum module

efficiency for maximum hot plate temperatures in the range Th,max = 900 1200 K.
22
Th,max = 900 K
20
Th,max = 1000 K
Th,max = 1100 K
18
Th,max = 1200 K
14
total
[mm]
16
12
10
8
6
10
12
14
16
18
20
22
Ksolar [%]
Figure 5-7: Total thermoelement leg length as a function of the solar-to-electricity

efficiency for maximum hot plate temperatures in the range Th,max = 900 1200 K.
5.5 Summary and conclusions

A 50 kWth solar cavity-receiver with a circular aperture at a solar
concentration of 1500 and packed with cascaded modules has been investigated
65
for optimizing the geometry and investigating the effect of the maximum
module efficiency of the modules (mod,dual) and the maximum hot plate
temperatures (Th,max) on the solar-to-electricity efficiency (solar). The cascaded
modules consist of three Al2O3 plates with a surface of 10 x 10 cm2
sandwiching a low- and a high-temperature thermoelement stage. The
thermoelement legs of the low- and the high-temperature stage consist of Bi-Te
with figure-of-merit ZT = 1 and perovskite-type oxides with ZT = 0.36 1.7,
respectively, each stage having 2822 thermoelement legs with a cross-section
area of 1.05 x 1.05 mm2. The heat transfer model of the 1 kWth cavity-receiver
prototype, which couples radiation exchange within the cavity and conduction
in the thermoelement legs, was extended and used to simulate the 50 kW th solar
cavity-receiver. The optimum geometry has been found as a trade-off between
high solar and low number of modules (N). It is characterized by a width of
X = 60 cm, a height of H = 50 cm and an aperture diameter of a = 20.6 cm,
packed with N = 156 modules. Further, the model predicts the highest reradiation losses of 6.7% of Qsolar for Th,max = 1200 K, which is comparable to
the 1 kWth solar cavity-receiver prototype with re-radiation losses of 4%.
The effect of mod,dual and Th,max on solar is summarized in Table 5-5,
implying a strong dependency of solar on mod,dual.
Th,max
mod,dual
solar
Pmax
[K]
[%]
[%]
[kW]
900
11.7
7.4
3.7
900
19.5
13.4
6.7
1200
14.2
11
5.5
1200
26
20.8
10.4
Table 5-5: Summarized results of solar-to-electricity analysis of
the 50 kWth cavity-receiver.
The 50 kWth cavity-receiver packed with today available thermoelectric

materials (low-temperature Bi-Te with ZT = 1 and high-temperature perovskitetype oxide with ZT = 0.36, mod,dual = 11.7) already reaches an efficiency of
solar = 7.4%. In the future new efficient high-temperature materials are
expected, which would boost solar up to 20.8%.
6 A 1 kWel geothermal stack4

This chapter presents a numerical analysis of the performance of a 1 kW el
stack comprising arrays of TEC modules with counter-flow heat exchangers.
The analysis focuses on the influence of hot water inlet and outlet temperatures
on the heat-to-electricity efficiency (stack) and the stack volume (Vstack). It
complements a previous study in which a similar concept has been investigated
from the standpoint of the stack geometry only [28]. Optimization is
accomplished for either (a) maximum heat-to-electricity efficiency or (b)
minimum size. In order to perform the stack simulations, a heat transfer model
is formulated to couple conduction in the TEC modules with convection to and
from the modules plates. The calculated open-circuit voltages resulting from
the temperature differences are validated with those temperature differences
computed via CFD.
6.1 Schematic of thermoelectric stack

The modular stack configuration is schematically shown in Figure 6-1.
The stack comprises lateral arrays of TEC modules. Each module is represented
by a pair of Al2O3 plates fastening 127 pairs of p/n-type (Bi0.2Sb0.8)2Te3 /
Bi2(Te0.96Se0.04)3 thermoelement legs (M = 254) [40]. The surface of the Al2O3
plates and the cross-section of the thermoelement legs are Aplate = 30 x 30 mm2
and Aleg = 1.05 x 1.05 mm2, respectively; the length of the stack (L) and the
length of the thermoelement legs (l) are allowed to vary as model parameters. A
temperature gradient across the thermoelement legs is imposed via a parallelplate heat exchanger separating the thermoelectric arrays and operating hot and
cold water in the counter-flow configuration. The heat exchanger is made of 1
mm thick Cu plates adhering to the 0.5 mm thick Al2O3 plates.
4
Material from this chapter has been submitted for publication in [95]: C. Suter, Z. Jovanovic,
and A. Steinfeld, "A 1 kWel Thermoelectric Stack for Geothermal Power Generation - Modeling
and Geometrical Optimization," Applied Energy, submitted.
68
A 1 kWel geothermal stack
Figure 6-1: Schematic of the thermoelectric stack with arrays of TEC modules
separated by Cu plate heat exchangers in counter-flow configuration. The dashed line
outlines a stack segment for which the heat transfer model has been developed.
6.2 Heat Transfer Model

The heat transfer model has been developed for a stack segment indicated
by the dashed line in Figure 6-1. A more detailed schematic of the segment is
shown by Figure 6-2, representing the segment as an array of Narray modules
having the total length L, the width W = 30 mm and hot and cold water halfchannels with thickness th and tc, respectively, resulting in the total height
H l th tc 2 tplate 2 tCu . After neglecting the thermal resistance for the
heat conduction through the Cu and Al2O3 plates, the heat transfer is assumed
to be based on two mechanisms: (1) convection between water and the
thermoelement legs, and (2) 1D conduction through the thermoelement legs.
Furthermore, it is assumed that the hot and cold sides of the legs within a
module have uniform temperatures. The model only considers steady-state.
69
Figure 6-2: Schematic of the modeled stack segment: an array of stack length L with
Narray modules having a leg length l between hot and cold water half-channels of
thickness th and tc., respectively, resulting in the height H. Each channel is discretized
into Narray cells (dotted boxes), having the uniform water temperatures Tw,h,i and Tw,c,i
within the cell i, respectively. The dashed-dotted lines indicate symmetry boundaries.
6.2.1
Heat conduction

for the thermoelement legs are given in Equations (3.4) and (3.5). The heat
conduction equations will be illustrated for the hot side only due to the
complete analogy with the cold side. With the assumption of 1D heat
conduction in the thermoelement legs, linear dependency of the thermoelectric
material properties on temperature and the overall current J, these equations
can be algebraically solved. Thus, the 1D heat conduction at the hot side of a
thermoelement leg in cell i in terms of the hot (Tleg,h,i) and cold (Tleg,c,i) side leg
temperature, and the overall current J reduces to:
Uleg,m,i 2
N leg,m,i
Qleg,h,i Sleg,m,i Tleg,h,i J D h,i l
J / 2 Eh,i Aleg
Tleg,h,i Tleg,c,i (6.1)
Aleg
l
where
D h,i
d U leg Tleg,h,i Tleg,c,i
1
1

4
U
dT
leg,m,i
leg
(6.2)
and
d N leg
E h,i 1
Tleg,h,i Tleg,c,i / 12
(6.3)
N
dT
leg
leg,m,i
are the correction factors accounting for the linear dependency of the material
properties on temperature. The thermoelectric properties, namely the Seebeck
coefficient Sleg,m,i, the electric resistivity leg,m,i, and the thermal conductivity
2
70
Nleg,m,i are evaluated at the middle leg temperature Tleg,m,i
leg,h,i
Tleg,c,i / 2
[83]. As aforementioned, the heat conduction equations for the cold side are
analogous to those presented above.
Thermoelectric properties of the thermoelements are estimated by linear
correlations, see Appendix B.2, Table B-2.
6.2.2
Forced heat convection
The heat convection equations will be illustrated for the hot side only due
to the complete analogy with the cold side. The general energy conservation
equation in integral form for the hot water channel considering heat transport
by the water flow and heat convection through walls is given in terms of the
water density w, the water heat capacity cp,w, the hot water temperature Tw,h,
the velocity of the water flow uh , the normal vector of the channel domain nh ,
the convective heat flux qconv,h and the channel domain boundary B by [96]:
cp,w Tw,h uh nh dB
qconv,h nh dB
(6.4)
The hot water channel is discretized into Narray rectangular cells (control
volumina, see dotted boxes in Figure 6-2). Each cell i has a uniform discrete
hot water temperature Tw,h,i. For the discretization, the following relations are
used: (1) The hot mass flow is given by:
(6.5)
mseg,h Uw uh Achannel
where Achannel,h
W th is the hot water channel cross-section area. (2) The
convective heat flux is given by:
qconv,h
hh,i Tw,h,i Tleg,h,i
(6.6)
where hh,i is the convective (hot water)-to-(thermoelement leg) heat transfer

coefficient and Tleg,h,i the hot leg side temperature in the cell i. Then the
discretized equation in cell i has the form [69]:
T
T T
T
cp,w,i mseg,h w,h,i -1 w,h,i w,h,i w,h,i +1 hh,i Aplate Tw,h,i Tleg,h,i (6.7)
2
2
The heat transfer coefficients hh,i have been determined from Nusselt
correlations for laminar, transitional and turbulent flows between parallel flat
71
plates [97-100]. As aforementioned, the heat convection equations for the cold
side are analogous to the ones presented above.
Physical properties of water are estimated by correlations, see Appendix
B.4, Table B-4.
6.2.3
The electric current is given by:
U array
Rarray Rcontact Rload
(6.8)
The voltage Uarray is given by:

N array
M Sleg,m,i 'Tleg,i
U array
(6.9)
i =1
where M is the number of legs. The electrical resistance of the material is

calculated by:
l
M
Aleg
Rarray
N array
leg,m,i
(6.10)
i =1
and the contact resistance by:
Rcontact
2 N array M
U*
Aleg
(6.11)
where * is the electric contact resistivity. The matched load resistance is given
by:
(6.12)
Rload Rcontact Rarray
The maximum power output of the module array is calculated in terms of Rload
and J by:
Parray
Rload J 2
and the useful power output of the stack segment is given by:
Pseg Parray Ppump
(6.13)
(6.14)
where the pump work Ppump is estimated in terms of the pressure drop and the
flow velocity in the water channels. The pressure drop in the hot channel is
estimated in terms of the friction factor fh and the flow velocity uh by:
72
'ph
fh
L1
2
U w uh
th 2
(6.15)
An analogous equation holds for the cold channel. The friction factor has been
determined for laminar, transitional, and turbulent flow between parallel flat
plates [97, 99]. Then the pumping work in the hot channel is calculated by:
(6.16)
Ppump,h 'ph uh Achannel,h
Evidently, the total pump work is Ppump = Ppump,h + Ppump,c. Finally, the heat-toelectricity efficiency of the stack segment seg is defined as the ratio of the
electrical power output over the heat delivered to the stack segment by:
Kseg
Pseg
mseg,h cp,w 'Tw,h
(6.17)
The volume of the stack segment is the product of the stack length L, the stack
segment height H and the stack segment width W:
(6.18)
Vseg L H W
Finally for the total stack, the magnitudes stack, Vstack, mstack,h and Nstack are
calculated in terms of the ratio of the power outputs of the stack Pstack and stack
segment Pseg by the Equations (6.19) (6.22):
(6.19)
Kstack Kseg
Vstack
6.2.4
Vseg
Pstack
Pseg
(6.20)
mstack,h
mseg,h
Pstack
Pseg
(6.21)
N stack
N array
Pstack
Pseg
(6.22)
The power output of the stack was set to Pstack = 1 kW. The hot water inlet
(Tw,h,in) and outlet (Tw,h,out) temperatures were varied in the range Tw,h,in = 313
413 K and Tw,h,out = 303 393 K, respectively, yielding hot water temperature
differences (Tw,h) in the range Tw,h = 10 110 K. The cold water inlet (Tw,c,in)
and outlet (Tw,c,out) temperatures were set to Tw,c,in = 293 K and Tw,c,out = 298 K,
73
respectively, yielding the cold water temperature difference Tw,c = 5 K. The
mass flow rates of hot (mstack,h) and cold water (mstack,c) were adjusted to satisfy
the prescribed hot/cold water outlet temperatures. The dashed-dotted line in
Figure 6-2 in the middle of the water channels was considered as an axis of
symmetry. The pressure at the channel inlets is 5 bars. The geometric
parameters were ranging in the following intervals:
(a) stack segment length L = 300 1500 mm, i.e., Narray = 10 50
(b) thermoelement leg length l = 0.5 4 mm and
(c) hot/cold half-channel thicknesses ^th , tc ` = 0.1 1 mm.
The model parameters are shown in Table 6-1.
6.2.5
Numerical solution
The coupled governing Equations (6.1) to (6.22) were implemented in a

FORTRAN code [70] and solved iteratively until the convergence criterion
-1
(Tw,h,
i Tw,h,i )
w,h,i
d 106
(6.23)
for all hot water temperatures Tw,h,i in elements i after iterations as well as the
overall energy balance
'Qh 'Qc Pseg

'Qh
were satisfied, where 'Qh
d 104
mseg,h cp,w 'Tw,h and 'Qc
(6.24)
mseg,c cp,w 'Tw,c .
6.3 Comparison with the CFD-based model

The steady-state momentum and heat balance equations for the stack
segment have been solved using the CFD software ANSYS CFX 13.0 [101] in
order to obtain the rigorous temperature profiles of cold and hot sides of the
legs along a segment. Open-circuit voltages were then calculated using the
mean temperature differences across the TEC modules. They are compared in
Figure 6-3 with those resulting from the heat transfer model described in this
74
chapter. The high coefficient of determination in this parity plot (R2 = 0.997)
implies good agreement between the two methods, further supported by the
mean relative error of 3% [102].
Parameter
Pstack
Tw,h,in
Value
1
313 413
Unit
kW
K
Source
set
model parameter
Tw,h,out
Aleg
mstack,h
mstack,c
Aplate
l
L
Narray
M
tplate
tCu
tc
th
W
p-type
Sp-type
p-type
n-type
Sn-type
n-type
*
303 393
1.05 x 1.05
0.1 5
1 10
30 x 30
0.5 4
0.3 1.5
10 50
254
0.5
1
0.1 1
0.1 1
30
1.13 1.36
204 235
10.4 16.3
1.26 1.38
-215 -207
11 15.5
510-9
K
mm2
kg s-1
kg s-1
mm2
mm
m
mm
mm
mm
mm
mm
W m-1 K-1
V K-1
m
W m-1 K-1
V K-1
m
m2
model parameter
assumed
assumed
model parameter
model parameter
model parameter
set
assumed
assumed
model parameter
model parameter
assumed [94]
Table 6-1: General model parameters for 1 kWel geothermal stack simulations.
6.4 Simulation results

A stack delivering the useful power output Pstack = 1 kW has been
optimized by varying geometric parameters with a given set of hot water inlet
(Tw,h,in) and outlet (Tw,h,out) temperatures. The optimization was performed either
(1) with respect to maximum heat-to-electricity efficiency, or (2) with respect
to minimum size. The used model parameters are shown in Table 6-1.
75
200
120
80
array,OC
[V] (Model)
160
40
40
80
120
Uarray,OC [V] (CFD)
160
200
Figure 6-3: Parity plot of the stack segments open-circuit voltage Uarray,OC simulated by
the heat transfer model and extracted from CFD simulations. The coefficient of
determination is R2 = 0.997. The mean relative error is 3%.
6.4.1
Maximized heat-to-electricity efficiency
The heat-to-electricity efficiency stack as a function of Tw,h,in and Tw,h,out is

shown in Figure 6-4(a). It increases almost linearly with Tw,h,in and Tw,h,out,
implying a stronger dependency on Tw,h,out. The calculated values range from
stack = 0.5%, obtained for Tw,h,in = 313 K and Tw,h,out = 303 K, to stack = 4.2%
for Tw,h,in = 413 K and Tw,h,out = 393 K. Figure 6-4(b) presents the influence of
Tw,h,in and Tw,h,out on the hot water mass flow mstack,h. mstack,h monotonically
decreases with Tw,h,in but increases with Tw,h,out. The values range from
mstack,h = 4.85 kg/s, calculated for Tw,h,in = 313 K and Tw,h,out = 303 K, to
mstack,h = 0.28 kg/s, calculated with Tw,h,in = 413 K and Tw,h,out = 303 K. One
should notice that an increase in Tw,h,in leads to an increase in stack but a
decrease in mstack,h, while both stack and mstack,h increase with increasing Tw,h,out.
The corresponding thermoelement leg length for all data points is l = 4 mm.
76

(a)
4.5
T
4
w,h,out
= 393 K
383 K
3.5
373 K
363 K
Kstack [%]
353 K
2.5
343 K
2
333 K
323 K
1.5
313 K
1
303 K
0.5
0
300
(b)
10
320
340
360
Tw,h,in [K]
380
400
420
w,h,out
= 303 K
313 K
333 K
343 K
10
353 K
363 K
373 K 383 K
stack,h
[kg/s]
323 K
10
393 K
-1
300
320
340
360
Tw,h,in [K]
380
400
420
Figure 6-4: a) The heat-to-electricity efficiency and b) the hot mass flow of the stack as
a function of the hot water inlet and outlet temperature for the case of maximized heatto-electricity efficiency.
6.4.2
Minimum stack volume
The stack volume Vstack is shown in Figure 6-5(a) as a function of Tw,h,in

and Tw,h,out. It monotonically decreases with Tw,h,in and Tw,h,out. The stack
volumes range from Vstack = 0.29 m3, calculated for Tw,h,in = 313 K and
Tw,h,out = 303 K, to Vstack = 0.0021 m3 obtained with Tw,h,in = 413 K and
77
Tw,h,out = 393 K. Figure 6-5(b) shows Vstack as a function of the number of
modules Nstack for all simulated hot inlet and outlet temperatures. The data
points reveal a linear relationship (correlation factor of 0.9947). The number of
modules ranges from Nstack = 550 to Nstack = 79740 as calculated with
Vstack = 0.0021 m3 and Vstack = 0.29 m3, respectively. The corresponding
thermoelement leg length for all data points is l = 0.5 mm.
(a)
10
303 K
-1
313 K
[m3]
10
stack
323 K
333 K
10
-2
343 K
353 K
363 K
373 K
383 K
Tw,h,in = 393 K
10
(b)
10
300
320
340
360
Tw,h,in [K]
380
400
420
-1
stack
[m3]
10
-3
10
10
-2
-3
10
10
10
10
Nstack [-]
Figure 6-5: a) The stack volume as a function of the hot water inlet and outlet
temperature; and b) the stack volume as a function of the number of TEC modules
in the stack for the case of minimized stack volume.
78
6.4.3
Summary and conclusions
A stack comprising arrays of Bi-Te modules with counter-flow water heat

exchangers has been considered. A heat transfer model coupling 1D conduction
through the thermoelement legs with convection to and from the legs has been
implemented in order to investigate the performance of the thermoelectric stack
as a function of operating parameters and stack geometry. The simulated opencircuit voltages have been found to agree with those calculated via CFD within
a mean relative error of 3%.
A stack delivering the useful power output Pstack = 1 kW has been
optimized by varying geometric parameters with a given set of hot water inlet
(Tw,h,in) and outlet (Tw,h,out) temperatures. The optimization was performed either
(1) with respect to maximum heat-to-electricity efficiency, or (2) with respect
to minimum size. The key results are summarized in Table 6-2. They imply a
strong dependency of both stack and Vstack on Tw,h,in and Tw,h,out.
(a) maximized stack
(b) minimized Vstack Comparison

stack,(a)/ Vstack,(a)/
Tw,h,in Tw,h,out stack
Vstack
stack Vstack
stack,(b)
Vstack,(b)
3
3
[K]
[K]
[%]
[m ]
[%]
[m ]
[-]
[-]
313
303
0.5
0.9075
0.16 0.2871
3.13
3.16
413
303
1.9
0.0611
0.78 0.0146
2.44
4.18
413
393
4.2
0.0135
2.1
0.0021
2.00
6.41
Table 6-2: Summarized results for maximized stack and minimized Vstack of
the 1 kWel geothermal stack.
7 Summary and conclusions

This thesis has considered the thermoelectric conversion of concentrated
solar radiation and geothermal energy for decentralized power generation.
7.1 Solar energy conversion

7.1.1
A 50 kW th solar cavity-receiver
In the first phase the behavior of 4-leg perovskite-type oxide modules with
figure-of-merit of 0.05 and operated at a maximum hot plate temperature of 900
K was investigated. The heat transfer analysis of directly irradiated modules
indicated that 60% of the incident solar irradiation is lost due to re-radiation. In
order to reduce the re-radiation, a 1 kWth solar cavity-receiver prototype
containing an array of 18 modules was designed, fabricated and tested. The
measured peak solar-to-electricity efficiency was solar = 0.12 % at a solar
concentration ratio C = 620 suns (Qsolar = 710 W). The heat transfer analysis has
revealed a reduction of the re-radiation losses to 4% compared to 60% for the
directly irradiated modules, showing the significant advantage of the cavity
design. In order to show the potential of the cavity design, a cubic 50 kWth
solar cavity-receiver was packed with cascaded dual-stage modules, which
consist of 10 x 10 cm2 plates sandwiching a low-temperature stage (Bi-Te with
figure-of-merit ZTlow = 1) and a high-temperature stage (perovskite-type oxides
with ZThigh = 0.36 1.7). The cavity has a circular aperture and is operated at
C = 1500. The optimum geometry of the cavity-receiver has been found to be

characterized by a width of W = 60 cm, a height of H = 50 cm and an aperture
diameter of a = 20.6 cm, packed with N = 156 TEC modules. For today
available material (ZThigh = 0.36) and Th,max = 900 K, the cavity-receiver reaches
an efficiency of solar = 7.4%; and for future material (ZThigh = 1.7) and
Th,max = 1200 K an efficiency of solar = 20.8%.
80
7.1.2
Comparison to other technologies
In Table 7-1, the 50 kWth cavity-receiver packed with today available

material (solar = 7.4%) is compared to other technologies also used for
decentralized conversion of concentrated solar radiation in solar dishes, which
are stirling engines and concentrated PV cells.
cavity-receiver
Stirling engine
Concentrated PV
a
solar
7.4%
29%
42%b
maintenance
low
higha
lowb
fabrication cost
unknown
higha
highb
Table 7-1: Comparison of cavity-receiver, stirling engine and concentrated PV for
today. aRef. [3], bRef. [5].
The solar-to-electricity efficiency of the cavity-receiver is significantly

lower than the alternative technologies, but has high potential for improvement
with the advent of more efficient material in the future. A major advantage of
the cavity-receiver is its low maintenance costs, which also applies for
concentrated PV as they have no moving components [5, 13]. Fabrication costs
are not yet known, as no efficient and high-temperature stable thermoelectric
material is yet on the market. However, they are expected to be low [52]. In
contrast, high-efficient Stirling engines and sophisticated concentrated PV cells
presumably have high costs [3, 5].
The fabrication and testing of a cavity-receiver as presented in chapter 0 is
necessary for proving the concept. Furthermore, economic analyses are
required to check the competitiveness of cavity-receivers with other
technologies.
7.2 Geothermal energy conversion

7.2.1
A 1 kW el geothermal stack
In the second part, the thermoelectric conversion of geothermal lowtemperature heat by a stack comprising arrays of Bi-Te modules with counterflow water heat exchangers has been studied. A stack delivering the maximum
power output Pstack = 1 kW has been optimized by varying geometric
81
parameters for a given set of hot water inlet (Tw,h,in) and outlet (Tw,h,out)
temperatures having as optimization goal either (a) maximum solar-toelectricity efficiency stack or (b) minimum volume Vstack. The key results are
summarized in Table 6-2, implying strong dependency of both stack and Vstack
on Tw,h,in and Tw,h,out.
(a) maximized stack
(b) minimized Vstack Comparison

stack,(a)/ Vstack,(a)/
Tw,h,in Tw,h,out stack
Vstack
stack Vstack
stack,(b)
Vstack,(b)
3
3
[K]
[K]
[%]
[m ]
[%]
[m ]
[-]
[-]
313
303
0.5
0.9075
0.16 0.2871
3.13
3.16
413
393
4.2
0.0135
2.1
0.0021
2.00
6.41
Table 7-2: Summarized results for maximized stack and minimized Vstack.
7.2.2
Comparison to other technologies
A 1 kWel stack optimized for maximum heat-to-electricity efficiency and

operating at Tw,h,in = 413 and Tw,h,out = 353, yielding stack = 3.3%, is the
reference case whose hot outlet water could be used for heating purposes,
which is not further considered here. In Table 7-3, the stack is compared to
standard amorphous Si PV cells, which are commonly used on rooftops.
geothermal stack
PV
conversion efficiency
3.3%
10 15%b
fabrication/installation costs
higha
lowc
temporal availability of source permanenta
intermittentc
regional availability of source globala
sunny regionsc
Table 7-3: Comparison of TEC stack and standard PV for today. aRef. [7], bRef. [5],
c
Ref. [2].
The conversion efficiency of the PV cells (solar = 10 15%) is clearly

higher than the one of the geothermal stack. The manufacturing cost of standard
PV cells has lowered significantly in the last years [5]. The same trend can be
expected for TEC modules in the next years. Therefore, the fabrication costs of
the geothermal stack will presumably be low. However, the geothermal energy
recovery is still cost-intensive [7]. Therefore the installation is expected to be
very expensive. The geothermal stack can be operated continuously, since
geothermal energy is permanently available, whereas PV cells are affected by
82
the intermittency of the sun. In order to provide electricity permanently, a

storage device e.g. batteries would be required. Geothermal energy is available
all over the world whereas the use of PV cells is reasonable only in sunny
regions [5, 7].
The fabrication and testing of a TEC stack as presented in chapter 6 is
necessary in order to prove the concept. Cost and feasibility analyses are
required in order to evaluate the possibilities of geothermal energy recovery for
small-scale applications (1 kWel). Furthermore, economic analyses have to be
done to estimate the fabrication costs of the geothermal stack. Finally, the
competitiveness of thermoelectric conversion of geothermal energy compared
to PV cells and other technologies have to be evaluated, and the configuration
and operating parameter range can be defined that are optimal from a
commercial standpoint.
7.3 Comparison of thermoelectric conversion of concentrated

solar radiation and geothermal energy
Both the solar and the geothermal concepts, presented in chapter 0 and 6,
respectively, are compared in Table 7-4.
solar system
geothermal system
7.4%
3.3%
a,b
fabrication/installation costs
low
highd
temporal availability of source intermittentc
permanentd
regional availability of source sunny regionsc
globald
Table 7-4: Comparison of solar and geothermal system. aRef. [3]. bRef. [52], cRef. [2],
d
Ref. [7]
The conversion efficiency of solar radiation is more than twice that of

geothermal energy and has a high potential for improvement, whereas the
geothermal energy conversion efficiency is limited due to the low temperature
range and the already efficient thermoelectric materials. Installation costs of the
geothermal TEC system are huge whereas the manufacturing costs of a solar
dish with a TEC cavity are moderate compared to the geothermal system [3, 7,
52]. The geothermal system can count on permanent and global availability,
83
whereas the solar system is limited to sunny regions with intermittent power
production.
In-depth cost and feasibility analyses of both systems will show which
system under what conditions is reasonable and competitive.
8 Outlook
This chapter gives suggestions for future research in the field of
thermoelectricity for the conversion of solar and geothermal energy in smallscale applications (decentralized power generation).
8.1 Material research

The research for the already efficient low-temperature materials (figure-ofmerit of 1) should continue to focus on low-cost materials and TEC module
fabrication. The goal should be to compensate the rather low conversion
efficiency by low costs and reliability.
In the high-temperature applications, the most crucial point is the
thermoelectric material, which has a strong influence on the conversion
efficiency. The research should be concentrated on maximizing the figure-ofmerit as there is still a high potential for improvement. Further, the research
should also be focused on minimizing fabrication costs of TEC modules in
order to compensate the low conversion efficiency and to make the
thermoelectric conversion competitive to other technologies.
8.2 Analysis of geothermal

applications
energy
use
for
small-scale
The technical and economic feasibility of borehole drilling and pump/tube

system installations has to be evaluated in combination with localizing
favorable regions throughout world. The development of the learning curve of
this yet young technology shall be considered to estimate if and when
geothermal energy recovery could become economic.
86
Outlook
8.3 Fabrication and testing of solar cavity and geothermal stack

Both a geothermal 1 kWel stack and a 50 kWth solar cavity-receiver in
combination with a solar dish should be fabricated and tested in order to prove
the concepts and to validate the performances predicted by the simulations. The
stability in long-time runs and the durability in on/off cycles have to be tested
and proven. Furthermore, the experimentally collected data can then be used to
refine the simulation models and experimentally validate them.
8.4 Further Optimization

With the proven concepts and the validated models, the existing setups can
be further improved. Additionally, economic considerations have to be
incorporated in the optimization process. Finally, the competitiveness with
other technologies can be evaluated for today or the future.
Appendix A
Material Properties
A.1 Properties of p-type
CaMn0.98Nb0.02O3
La1.98Sr0.02CuO 4
and
n-type
The thermoelectric properties Nleg, leg, and Sleg are shown in the Figures
A-1, A-2 and A-3, respectively.
3.6
p-type
n-type
3.4
3.2
-1
Nleg [W m ]
3
2.8
2.6
2.4
2.2
2
1.8
1.6
200
400
600
800
T [K]
1000
1200
1400
Figure A-1: Measured thermal conductivity Nleg as a function of the temperature T for
p-type La1.98Sr0.02CuO4 and n-type CaMn0.98Nb0.02O3.
88
Appendix A
400
Uleg [P: m]
350
300
250
200
p-type
n-type
150
200
400
600
800
T [K]
1000
1200
1400
Figure A-2: Measured electric resistivity leg as a function of the temperature T for
p-type La1.98Sr0.02CuO4 and n-type CaMn0.98Nb0.02O3.
260
p-type
n-type
240
|Sleg| [PV K-1]
220
200
180
160
140
120
200
400
600
800
T [K]
1000
1200
1400
Figure A-3: Measured absolute values of the Seebeck coefficient Sleg as a function of
the temperature T for p-type La1.98Sr0.02CuO4 (positive) and n-type CaMn0.98Nb0.02O3
(negative).
89
A.2 Properties
of
p-type
Bi2(Te0.96Se0.04)3
(Bi 0.2Sb0.8)2Te3
and
n-type
The thermoelectric properties Nleg, leg, and Sleg were extracted from Ref.
[40], and are shown in the Figures A-4, A-5 and A-6, respectively.
1.6
p-type
n-type
1.55
-1
Nleg [W m ]
1.5
1.45
1.4
1.35
1.3
1.25
260
280
300
320
340
T [K]
360
380
400
420
Figure A-4: Thermal conductivity Nleg as a function of the temperature T for

p-type (Bi0.2Sb0.8)2Te3 and n-type Bi2(Te0.96Se0.04)3, extracted from Ref. [40].
15
14
Uleg [P: m]
13
12
11
10
9
8
260
p-type
n-type
280
300
320
340
T [K]
360
380
400
420
Figure A-5: Electric resistivity leg as a function of the temperature T for

p-type (Bi0.2Sb0.8)2Te3 and n-type Bi2(Te0.96Se0.04)3, extracted from Ref. [40].
90
Appendix A
230
225
|Sleg| [PV K-1]
220
215
210
205
200
195
190
260
p-type
n-type
280
300
320
340
T [K]
360
380
400
420
Figure A-6: Absolute values of the Seebeck coefficient Sleg as a function of the
temperature T for p-type (Bi0.2Sb0.8)2Te3 and n-type Bi2(Te0.96Se0.04)3, extracted from
Ref. [40].
A.3 Thermal conductivity of Al 2O3

The data for the thermal conductivity of Al2O3 have been extracted from
Ref. [103].
180
160
140
2 3
-1
NAl O [W m ]
120
100
80
60
40
20
0
200
400
600
T [K]
800
1000
1200
Figure A-7: Thermal conductivity NAl2O3 as a function of the temperature T for Al2O3,
extracted from Ref. [103].
Appendix B
Material property correlations
B.1 p-type La1.98Sr0.02CuO4 and the n-type CaMn0.98Nb0.02O3
The linear correlations of the thermoelectric properties of p-type
La1.98Sr0.02CuO4 and n-type CaMn0.98Nb0.02O3 are shown in Table B-1.
Parameter Linear correlation
Range (K)
Unit
Confidence
Source
Nn-type
-1.579 10 T 3.164
3
298.5 T 1123.3
Wm K
R = 0.858
Fig. A-1
Np-type
-1.818 10 T 3.847
3
323.1 T 1272.7
W m-1 K-1
R2 = 0.975
Fig. A-1
n-type
3.251 10 T 9.793 10
p-type
-1.740 10 T 1.418 10
Sn-type
-1.172 10 T 1.350 10
Sp-type
-9.786 10 T 2.331 10
-1
-1
7
5
319.8 T 945
R2 = 0.997
Fig. A-2
7
4
352.2 T 1247.7
R2 = 0.960
Fig. A-2
7
4
319.8 T 945
V m-1 K-1
R2 = 0.923
Fig. A-3
8
4
352.2 T 1247.7
V m-1 K-1
R2 = 0.751
Fig. A-3
Table B-1: Linear material correlations for the p-type La1.98Sr0.02CuO4 and n-type
CaMn0.98Nb0.02O3.
B.2 p-type (Bi 0.2Sb0.8)2Te3 and n-type Bi 2(Te0.96Se0.04)3

The linear correlations of the thermoelectric properties of p-type
(Bi0.2Sb0.8)2Te3 and n-type Bi2(Te0.96Se0.04)3 are shown in Table B-2.
92
Appendix B
Parameter Polynomial
Range (K)
Unit
Confidence
Source
Nn-type
-8.246 10
4
T 1.629
273 T 413
W m-1 K-1
R2 = 0.561
Fig. A-4
Np-type
-1.786 10
3
T 1.899
273 T 413
W m-1 K-1
R2 = 0.799
Fig. A-4
n-type
3.748 10
7
273 T 413
R2 = 0.974
Fig. A-5
p-type
6
273 T 413
R2 = 0.894
Fig. A-5
4
273 T 413
V m-1 K-1
R2 = 0.444
Fig. A-6
4
273 T 413
V m-1 K-1
R2 = 0.950
Fig. A-6
8
T 3.103 10
4.613 10
8
T 3.466 10
Sn-type
5.324 10
8
T 1.907 10
Sp-type
2.573 10
7
T 1.260 10
Table B-2: Polynomial correlations for the p-type (Bi0.2Sb0.8)2Te3 and n-type
Bi2(Te0.96Se0.04)3.
B.3 Air
The correlations of the air properties are shown in Table B-3.
Parameter
Polynomial
E
T
T
A B
D
cosh E
sinh C
T
T
28.966
2
cp,air
Range (K)
Unit
Confidence
Source
50T1500
J kg-1 K-1
< 5%
[104]
70T2000
W m-1 K-1
< 5%
[104]
80T2000
kg m-1 s-1
< 5%
[104]
none
kg m-3
none
[68]
A = 28958, B = 9390, C = 3012,

D = 7580, E = 1484
A T
Nair
1 C
D
-4
A = 3.141710 , B = 0.7786,
C = -0.7116, D = 2121.7
A T
air
1 C
T
-6
A = 1.42510 , B = 0.5039, C =
108.3
air
U air,0
T0
T
air,0 = 1.293, T0 = 273.15
Table B-3: Correlations for air properties.
93
B.4 Water
The correlations of the water properties are shown in Table B-4.
cp,w
18
A = 276370, B = -2090.1, C = 8.125,
D = -0.014116, E = 9.370110-6
3
A BT CT DT E T
A = -0.432, B = 0.0057255,
C = -8.07810-6, D = 1.86110-9
Nw
exp A
C ln T D T
T
A = -52.843, B = 3703.6, C = 5.866,
D = -5.879E-29, E = 10
A B t 0.35 C t 2/3 D t E t 4/3 18

where t = 1 T
A = 17.863, B = 58.606, C = -95.396,
D = 213.89, E = -141.26
Unit
Confidence
Source
273.16 T
533.1
J kg-1 K-1
< 1%
[104]
273.16 T
633.15
W m-1 K-1
< 1%
[104]
273.16 T
646.15
kg m-1 s-1
< 3%
[104]
273.15 T
647.1
J kg-1 K-1
< 1%
[104]
A B T C T 2 D T 3 E T 4
2
Range (K)
Table B-4: Correlations for water properties.
B.5 Al2O3
The polynomial correlation of the thermal conductivity of Al2O3 is shown
in Table B-5.
NAl2O3
1.78 10
11
4
8
Unit
Confidence
Source
300 T 900
W m-1 K-1
R2 = 0.9991
Fig. A-7
T 9.79 10 T
2
Range (K)
2.02 10 T 0.19 T 75.77
Table B-5: Polynomial correlations for Al2O3.
List of Figures
Figure 2-1: n-type semiconductor bar between temperature difference T
showing voltage difference U. ........................................................... 8
Figure 2-2: Schematic of a 2-leg (one p/n-type leg pair) thermoelectric
converter operated at temperature difference T and powering an
external load. ........................................................................................ 9
Figure 2-3: Schematic dependence of (a) the electric conductivity , Seebeck
coefficient S, power factor S2; and (b) the thermal conductivity on
the carrier concentration. Extracted from Ref. [30] and modified by
author. ................................................................................................. 11
Figure 3-1: Top and front view of 4-leg module. Indicated are the plate width
W, the leg width w, the distance from the neighbouring legs d, the leg
length l and the plate thickness t......................................................... 14
Figure 3-2: Assembly schematic of a 4-leg module, extracted from Ref. [58].
Indicated are the p/n-type thermoelement legs, the Al2O3 plates, the
dual-layer metallization (AuPtPd and AgPd), the electrical contacts,
the electrodes and the positioning mask. ............................................ 16
Figure 3-3: Schematic of the experimental setup at ETHs HFSS. (a) The 4-leg
module is placed at the HFSSs focal plane; incident solar radiative
fluxes measured by a thermogage (F). (b) Position of K-type
thermocouples (T) used to measure temperatures of the plates and of
the hot end, middle, and cold end of the legs; terminals (U) provided at
the cold ends for measuring the output voltage of the module. The cold
plate was attached to a water-circuit cooler (denoted by screw
fixation). ............................................................................................. 17
Figure 3-4: Temperature of hot and cold plates and solar radiative flux as a
function of time during a representative experimental run for a 4-leg
module with leg length l = 4 mm. ....................................................... 18
Figure 3-5: Fitted and measured power output as a function of the voltage
cc = 1.8 10
output for incident solar radiative fluxes in the range qsolar
96
List of Figures
W cm-2, and for external loads with resistance in the range Rload = 0.1
3.5 for a 4-leg module with leg length l = 4 mm. The coefficient of
determination for each fitted parabola is R2 > 0.997. ......................... 19
Figure 3-6: Temperature distribution along the p- and n-type legs for two
modules with leg length l = 10 mm. The x-axis error bars indicate
spatial accuracy (0.25 mm) of thermocouple placing. The coefficient
of determination for each fitted straight line is R2 > 0.993. ............... 19
Figure 3-7: Experimentally measured solar-to-electricity efficiency as a
function of solar radiative flux for the leg lengths l = 4, 5, and 10 mm.
cc and of solar. ...................... 21
Error bars indicating uncertainty of qsolar
Figure 3-8: Schematic of the model domain (divided into m x n cells).
Indicated are the incoming solar radiation (Qsolar), the re-radiation and
convective heat losses (Qreradiation+concection), the outgoing radiation and
conduction (Qconduction+radiation,out), the cold plate temperature (Tc) and
the open-circuit condition (J = 0). ...................................................... 22
Figure 3-9: Simulated open-circuit voltage as a function of solar radiative flux
for leg length l = 10 mm for the approach (a) with plate = 102, plate =
0.5; plate = 103, plate = 0.5; plate = 104, plate = 0.1; plate = 105, plate =
0.0; and for the opaque approach (b). ................................................ 28
Figure 3-10: Simulated and experimental open-circuit voltages as a function of
solar radiative flux for leg lengths l = 4, 5, 10 mm. ........................... 29
Figure 3-11: Simulated and experimentally measured solar-to-electricity
efficiencies as a function of the solar radiative flux for leg lengths l =
4, 5, 10 mm (experimental data from Figure 3-7). Simulations for
contact resistances Rcontact = 0.4 and 0.66 . ...................................... 30
Figure 3-12: Percentage of incident solar radiation transferred by the different
heat transfer modes: conduction through the thermoelement legs, reradiation and convective losses from the hot plate, and radiation to the
cc
cc
cold plate; for two cases: qsolar
= 6 cm-2, l = 10 mm and qsolar
= 10 cm2
, l = 5 mm.......................................................................................... 31
Figure 3-13: Simulated 2D temperature profile in p-type thermoelement leg for
cc
l = 10 mm at solar radiative flux qsolar
= 6 cm-2, and temperature
difference of 600 K. ........................................................................... 31
Figure 3-14: Simulated solar-to-electricity efficiency as a function of solar
radiative flux for leg lengths l = 5, 7.5, 10, 12.5 and 15 mm. ............ 33
97
Figure 3-15: Simulated solar-to-electricity efficiency as a function of the solar
radiative flux with l = 7.5 mm: (a) for d = 1 mm and w = 3, 4.5 and 6
mm, and (b) for w = 3 mm and d = 1, 2 and 3 mm. ............................ 34
Figure 4-1: Apparent absorptance of a 2D cavity as a function of the
absorptivity of its inner walls for incoming radiation uniformly
distributed over the aperture. Baseline dimension is: cavity
width/height X = 56 mm. The parameter is the aperture width: a = 7.1,
14.1 and 28.2 mm. .............................................................................. 39
Figure 4-2: Schematic of the experimental setup at ETHs HFSS. (a) crosssection of the 1 kWth solar cavity-receiver prototype with CPC, cooler,
and bottom, lower and upper 4-leg modules. (b) longitudinal section
with CPC, cooler, and central and lateral 4-leg modules. The
dimensions of the cavity-receiver are: length L = 81 mm, width/height
X = 56 mm and aperture a = 14.1 mm. ............................................... 40
Figure 4-3: Solar radiative flux distribution measured at the HFSS focal plane,
normalized by the peak radiative flux which varied between 128 and
657 suns. The 81 x 20 mm2-dotted rectangle indicates the entrance of
the 2D trough CPC. ............................................................................ 41
Figure 4-4: Cross-section of the HFSS with the solar cavity-receiver and CPC
placed at the focal plane (z = 0 m). .................................................... 42
Figure 4-5: Experimentally measured open-circuit voltage as a function of the
mean solar concentration ratio in the aperture for central/lateral bottom
modules, central/lateral lower modules, and central/lateral upper
modules. ............................................................................................. 43
Figure 4-6: Experimentally measured maximal power output and efficiency as
a function of the mean solar concentration ratio in the aperture for all
18 4-leg modules connected electrically in series. ............................. 44
Figure 4-7: Schematic of the model domain: cross-section of a rectangular
parallelepiped (box) cavity-receiver with rectangular aperture and
insulated 2D (trough) CPC. The dimensions are: cavity width/height X
= 56 mm and length L = 27 mm (perpendicular to the cross-section
plane), aperture width a = 14.1 mm. The cavity contains an array of N
= 6 modules. Indicated are: incoming solar radiation (Qsolar), reradiation and convective losses (Qloss), radiation exchange (Qradiosity),
heat convection (Qconvection), heat conduction (Qconduction,i) and opencircuit voltage (UOC,i) per module i, hot plate temperatures (Th,i), inner
98
Figure
Figure
Figure
Figure
Figure
Figure
Figure
Figure
List of Figures
CPC insulation surface temperature (Tins) and cold temperatures (Tc) of
cold plates and outer CPC insulation surface. .................................... 45
4-8: Cross-section of simulated air domain including cavity with
surface temperature Tsurface, CPC with cold temperature Tc and
environment with opening boundary at temperature and pressure T
and p , respectively. .......................................................................... 48
4-9: Parity plot of the Nusselt number for the natural convective heat
transfer for the values calculated by the correlation and those extracted
from CFD simulations. The coefficient of determination is R2 = 0.911
and the mean relative error 5%. ......................................................... 49
4-10: Simulated and experimentally measured open-circuit voltage as a
function of the mean solar concentration ratio in the aperture for
bottom, lower and upper modules. ..................................................... 51
4-11: Percentage of solar radiation input Qsolar transferred by different
heat transfer modes: conduction through the thermoelement legs, reradiation and convective losses through the aperture, and conduction
losses through the cavity and CPC walls. Qsolar = 710 W and C = 620
suns. ................................................................................................... 52
5-1: Figure-of-merit as a function of the temperature for several n-type
materials, extracted from Ref. [92]. ................................................... 54
5-2: Schematic of a cascaded module consisting of K stages and
operated between temperature TK+1 and T1, with intermediate
temperatures Tk................................................................................... 54
5-3: Schematic of the cascaded dual-stage module: three Al2O3 plates
sandwiching a low- and a high-temperature stage with variable
thermoelement leg length llow and lhigh, respectively. The cascaded
module is operated between the hot plate temperature Th and the cold
plate temperature Tc with the intermediate plate temperature Tint. ..... 55
5-4: Schematic of the model domain: cross-section of the square
parallelepiped (box) cavity-receiver with circular aperture and
insulated CPC. The dimensions are: cavity width X, height H, and
aperture diameter a. The cavity-receiver contains an array of N
cascaded dual-stage modules. Indicated are: incoming solar radiation
(Qsolar), re-radiation losses (Qloss), radiation exchange (Qradiosity), hot
plate temperatures (Th,i), inner insulation surface temperature (Tins) and
99
cold temperatures (Tc) of cold plates and outer CPC insulation surface.
............................................................................................................ 56
Figure 5-5: Solar-to-electricity efficiency and number of modules as a function
of: (a) the cavity width X and height H, and (b) the cavity height H for
width X = 60 cm; for maximum hot plate temperature Th,max = 900 and
1200 K. ............................................................................................... 62
Figure 5-6: Solar-to-electricity efficiency as a function of the maximum
module efficiency for maximum hot plate temperatures in the range
Th,max = 900 1200 K. ........................................................................ 64
Figure 5-7: Total thermoelement leg length as a function of the solar-toelectricity efficiency for maximum hot plate temperatures in the range
Th,max = 900 1200 K. ........................................................................ 64
Figure 6-1: Schematic of the thermoelectric stack with arrays of TEC modules
separated by Cu plate heat exchangers in counter-flow configuration.
The dashed line outlines a stack segment for which the heat transfer
model has been developed. ................................................................. 68
Figure 6-2: Schematic of the modeled stack segment: an array of stack length L
with Narray modules having a leg length l between hot and cold water
half-channels of thickness th and tc., respectively, resulting in the
height H. Each channel is discretized into Narray cells (dotted boxes),
having the uniform water temperatures Tw,h,i and Tw,c,i within the cell i,
respectively. The dashed-dotted lines indicate symmetry boundaries. 69
Figure 6-3: Parity plot of the stack segments open-circuit voltage Uarray,OC
simulated by the heat transfer model and extracted from CFD
simulations. The coefficient of determination is R2 = 0.997. The mean
relative error is 3%. ............................................................................ 75
Figure 6-4: a) The heat-to-electricity efficiency and b) the hot mass flow of the
stack as a function of the hot water inlet and outlet temperature for the
case of maximized heat-to-electricity efficiency. ............................... 76
Figure 6-5: a) The stack volume as a function of the hot water inlet and outlet
temperature; and b) the stack volume as a function of the number of
TEC modules in the stack for the case of minimized stack volume. .. 77
List of Tables
Table 3-1: General model parameters for simulations of directly irradiated
modules. ............................................................................................. 27
Table 3-2: Model parameters for simulations of directly irradiated modules:
variation of length. ............................................................................. 32
Table 3-3: Model parameters used in simulation varying the leg width and the
distance between neighboring legs. .................................................... 33
Table 4-1: General model parameters for 1 kWth cavity-receiver simulations . 50
Table 5-1: General model parameters of the 50 kWth cavity-receiver
simulations. ........................................................................................ 60
Table 5-2: Model parameters of the 50 kWth cavity-receiver simulations for
geometrical optimizations. ................................................................. 61
Table 5-3: Main results for the optimum 50 kW th cavity-receiver geometry:
X = 60 cm, H = 50 cm and N = 156. ................................................... 62
Table 5-4: Model parameters of 50 kWth cavity-receiver simulations for the
analysis of the solar-to-electricity efficiency. ..................................... 63
Table 5-5: Summarized results of solar-to-electricity analysis of the 50 kWth
cavity-receiver. ................................................................................... 65
Table 6-1: General model parameters for 1 kWel geothermal stack simulations.
............................................................................................................ 74
Table 6-2: Summarized results for maximized stack and minimized Vstack of the
1 kWel geothermal stack. .................................................................... 78
Table 7-1: Comparison of cavity-receiver, stirling engine and concentrated PV
for today. aRef. [3], bRef. [5]. ............................................................. 80
Table 7-2: Summarized results for maximized stack and minimized Vstack ....... 81
Table 7-3: Comparison of TEC stack and standard PV for today. aRef. [7], bRef.
[5], cRef. [2]........................................................................................ 81
Table 7-4: Comparison of solar and geothermal system. aRef. [3]. bRef. [52],
c
Ref. [2], dRef. [7]............................................................................... 82
102
List of Tables
Table B-1: Linear material correlations for the p-type La1.98Sr0.02CuO4 and
n-type CaMn0.98Nb0.02O3. ................................................................... 94
Table B-2: Polynomial correlations for the p-type (Bi0.2Sb0.8)2Te3 and n-type
Bi2(Te0.96Se0.04)3. ................................................................................ 95
Table B-3: Correlations for air properties. ...................................................... 95
Table B-4: Correlations for water properties. .................................................. 96
Table B-5: Polynomial correlations for Al2O3................................................. 96
References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
International Energy Outlook 2011. Available:

http://www.eia.gov/forecasts/ieo/, last checked January 2012.
"Concentrating Solar Power Global Outlook 09," Greenpeace
International, SolarPACES, and ESTELA, 2009.
"Concentrating Solar Power in 2001," SolarPACES, 2001.
"Technology Roadmap: Concentrating Solar Power," International
Energy Agency, 2010.
K. Mertens, Photovoltaik: Lehrbuch zu Grundlagen, Technologie und
Praxis, 1st ed. Mnchen: Fachbuchverlag Leipzig im Carl Hanser
Verlag, 2011.
"Global Market Outlook for Photovoltaics until 2015," European
Photovoltaic Industry Association, 2011.
Geothermie Schweiz. Available: http://www.geothermie.ch, last
checked January 2012.
Geothermie Deutschland. Available: http://www.geothermie.de, last
checked January 2012.
J. W. Lund, D. H. Freeston, and T. L. Boyd, "Direct application of
geothermal energy: 2005 Worldwide review," Geothermics, vol. 34,
pp. 691727, 2005.
M. Grubb, Renewable Energy Strategies for Europe, Foundations and
Context vol. 1. London, UK: The Royal Institute of International
Affairs, 1995.
T. Ackermann, G. Andersson, and L. Sder, "Distributed generation: a
definition," Electric Power Systems Research, vol. 57, pp. 195204,
2001.
D. M. Rowe, "Review: Thermoelectric Waste Heat Recovery as a
Renewable Energy Source," International Journal of Innovations in
Energy Systems and Power, vol. 1, pp. 13 -23, 2006.
S. B. Riffat and X. Ma, "Thermoelectrics: a review of present and
potential applications," Applied Thermal Engineering, vol. 23, pp.
913-935, 2003.
D. Kraemer, B. Poudel, and G. Chen, "High-performance flat-panel
solar thermoelectric generators with high thermal concentration,"
Nature Materials, vol. 10, pp. 532538, 2011.
104
References
[15]
S. A. Omera and D. G. Inield, "Design and thermal analysis of a two

stage solar concentrator for combined heat and thermoelectric power
generation," Energy Conversion & Management, vol. 41, pp. 737-756,
2000.
M. Eswararmoorthy and S. Shanmugam, "Thermodynamic analysis of
solar parabolic dish thermoelectric generator," Int. J. Renewable
Energy Technology, vol. 1, pp. 348-360, 2010.
H. Naito, Y. Kohsaka, D. Cooke, and H. Arashi, "Development of a
Solar Receiver for a High-Efficiency Thermionic/Thermoelectric
Conversion System," Solar Energy, vol. 58, pp. 191-195, 1996.
A. Weidenkaff, R. Robert, M. H. Aguirre, L. Bocher, T. Lippert, and
S. Canulescu, "Development of Thermoelectric Oxides for Renewable
Energy Conversion Technologies," Renewable Energy, vol. 33, pp.
342-347, 2008.
D. M. Rowe, "A High Performance Solar Powered Thermoelectric
Generator" Applied Energy, vol. 8, pp. 269-273, 1981.
P. Tome, M. Trottmann, A. Weidenkaff, C. Suter, P. Haueter, and A.
Steinfeld, "Thermoelectric oxide modules (TOMs) applied in direct
conversion of simulated solar radiation into electrical energy,"
Materials, vol. 3, pp. 2801-2814, 2010.
P. Tome, R. Robert, L. Bocher, M. Trottmann, M. H. Aguirre, A.
Weidenkaff, P. Haueter, A. Steinfeld, and J. Hejtmnek, "Direct
conversion of simulated solar radiation into electrical energy by a
perovskite thermoelectric oxide module (TOM)," Proc. Materials
Science and Technology Conference and Exhibition, MS&T08, vol. 1,
pp. 429-435, 2008.
C. Eisenhut and A. Bitschi, "Thermoelectric Conversion System based
on Geothermal and Solar Heat " in 25th International Conference
onThermoelectrics (ICT), 2006., Vienna, Austria, August 6 - 10, 2006,
pp. 510-515.
J. Yu and H. Zhao, "A numerical model for thermoelectric generator
with the parallel-plate heat exchanger," Journal of Power Sources, vol.
172, pp. 428434, 2007.
J. Esartea, G. Minb, and D. M. Roweb, "Modelling heat exchangers
for thermoelectric generators," Journal of Power Sources, vol. 93, pp.
72-76, 2001.
D. T. Crane and G. S. Jackson, "Optimization of cross flow heat
exchangers for thermoelectric waste heat recovery," Energy
Conversion and Management, vol. 45, pp. 15651582, 2004.
S. Blanger and L. Gosselin, "Thermoelectric generator sandwiched in
a crossflow heat exchanger with optimal connectivity between
modules," Energy Conversion and Management, vol. 52, pp. 2911
2918, 2011.
[16]
[17]
[18]
[19]
[20]
[21]
[22]
[23]
[24]
[25]
[26]
105
[27]
[28]
[29]
[30]
[31]
[32]
[33]
[34]
[35]
[36]
[37]
[38]
[39]
[40]
[41]
[42]
A. Bethancourt, R. Echigo, and H. Yoshida, "Thermoelectric

Conversion Analysis in a Counter-flow Heat Exchanger," presented at
the The thirteenth international conference on thermoelectrics, 1994.
A. Bitschi, "Modelling of thermoelectric devices for electric power
generation," PhD thesis, ETH Zurich, 2009.
H. J. Goldschmid, Introduction to Thermoelectricity, 1st ed. Berlin:
Springer, 2009.
D. M. Rowe, "General Principles and Basic Considerations," in
Thermoelectrics Handbook: Macro to Nano, Boca Raton, USA: CRC
Press, 2006.
R. D. Abelson, "Space Missions and Applications," in Thermoelectrics
Handbook: Macro to Nano, Boca Raton, USA: CRC Press, 2006.
D. M. Rowe, "United States Thermoelectric Activities in Space," in
Proceedings VIIIth Int Conf on Thermoelectric Energy Conversion,
Nancy, France, 10-13 July 1989, pp. 133-142.
G. Bennett, "Space Applications," in CRC Handbook of
Thermoelectrics, Boca Raton, USA: CRC Press, 1995.
D. M. Rowe, "Thermoelectrics, an environmentally-friendly source of
electrical power," Renewable Energy, vol. 16, pp. 12511256, 1999.
L. H. Dubois, "An introduction to the darpa program in advanced
thermoelectric materials and devices," in 18th International
Conference on Thermoelectrics, Baltimore, USA, 1999, pp. 1-4.
D. K. C. MacDonald, Thermoelectricity: an introduction to the
principles: Wiley, 1962.
G. Min, "Thermoelectric Module Design Theories," in
Thermoelectrics Handbook: Macro to Nano, Boca Raton, USA: CRC
Press, 2006.
O. Yamashita and S. Sugihara, "High-performance bismuth-telluride
compounds with highly stable thermoelectric figure of merit," Journal
of Materials Science, vol. 40, pp. 64396444, 2005.
J. S. Tse and D. D. Klug, "Recent Trends for the Design and
Optimization of Thermoelectric Materials A Theoretical
Perspective," in Thermoelectrics Handbook: Macro to Nano, Boca
Raton, USA: CRC Press, 2006.
Quick-Ohm. Available: http://www.quick-ohm.de/, last checked
January 2012.
J. Yang, "Potential applications of thermoelectric waste heat recovery
in the automotive industry," in International Conference on
Thermoelectrics, Clemson, USA, 2005.
K. Nagaol, A. Nagail, I. Fujiil, and T. Sakurai, "Design of
Thermoelectric Generation System Utilizing The Exhaust Gas of
Internal-Combustion Power Plant," in 17th international Conference
on Thermoelectrics, Nagoya, Japan, 1998, pp. 468-472.
106
References
[43]
C. Suter, P. Tome, A. Weidenkaff, and A. Steinfeld, "Heat transfer

analysis and geometrical optimization of thermoelectric converters
driven by concentrated solar radiation," Materials, vol. 3, pp. 27352752, 2010.
Y. S. Touloukian, Thermal radiative properties. New York, USA:
IFI/Plenum, 1972.
Graphit. Available: http://www.graphit.de, last checked January 2012.
S. Zhou, J. Zhao, S. Chu, and L. Shi, "Synthesis, characterization and
magnetic properties of lightly doped La2-xSrxCuO4 (x = 0.04)
nanoparticles," Physica C Superconductivity, vol. 451, pp. 38-43,
2007.
L. Bocher, R. Robert, M. H. Aguirre, S. Malo, S. Hbert, A. Maignan,
and A. Weidenkaff, "Thermoelectric and magnetic properties of
perovskite-type manganate phases synthesised by ultrasonic spray
combustion (USC)," Solid State Sciences, vol. 10, pp. 496501, 2008.
L. Bocher, M. H. Aguirre, D. Logvinovich, A. Shkabko, R. Robert, M.
Trottmann, A. Weidenkaff, and A. Quelle, "CaMn1-x NbxO3
(x 0.08) perovskite-type phases as promising new high-temperature
n-type thermoelectric materials," Inorg. Chem., pp. 8077-8085, 2008.
S. S. Kim, F. Yin, and Y. Kagawa, "Thermoelectricity for
crystallographic anisotropy controlled Bi-Te based alloys and p-n
modules," Journal of Alloys and Compounds, vol. 419, pp. 306-311,
2006.
E. S. Reddy, J. G. Noudem, S. Hebert, and C. Goupil, "Fabrication and
properties of four-leg oxide thermoelectric modules " Journal of
Physics D: Applied Physics, vol. 38, pp. 3751-3755, 2005.
W. Shin, N. Murayama, K. Ikeda, and S. Sago, "Thermoelectric power
generation using Li-doped NiO and (Ba, Sr)PbO3 module," Journal of
Power Sources, vol. 103, pp. 80-85, 2001.
R. Funahashi, M. Mikami, T. Mihara, S. Urata, and N. Ando, "A
portable thermoelectric-power-generating module composed of oxide
devices," Journal of Applied Physics, vol. 99, pp. 066117-1/3, 2006.
R. Funahashi, I. Matsubara, H. Ikuta, T. Takeuchi, U. Mizutani, and S.
Sodeoka, "Oxide single crystal with high thermoelectric performance
in air," Japan J. Applied Physics, vol. 39, pp. 1127-1129, 2000.
R. Funahashi, S. Urata, K. Mizuno, T. Kouuchi, and M. Mikami,
"Ca2.7Bi0.3Co4O9La0.9Bi0.1NiO3 thermoelectric devices with high
output power density," Applied Physics Letters, vol. 85, pp. 10361038, 2004.
A. Weidenkaff, "Preparation and application of nanostructured
perovskite phases," Adv. Eng. Mater, vol. 6, pp. 709-714, 2004.
M. H. Aguirre, S. Canulescu, R. Robert, N. Homazava, D.
Logvinovich, L. Bocher, T. Lippert, M. Dbeli, and A. Weidenkaff,
[44]
[45]
[46]
[47]
[48]
[49]
[50]
[51]
[52]
[53]
[54]
[55]
[56]
107
[57]
[58]
[59]
[60]
[61]
[62]
[63]
[64]
[65]
[66]
[67]
[68]
[69]
[70]
[71]
"Structure, microstructure, and high-temperature transport properties

of La1xCaxMnO3 thin films and polycrystalline bulk materials,"
Journal of Applied Physics, vol. 103, pp. 013703-1/6, 2008.
D. D. L. Chung, Composite Material: Science and Applications
(Engineering Materials and Processes), 2nd ed. London, England:
Springer-Verlag, 2010.
P. Tome, C. Suter, M. Trottmann, A. Steinfeld, and A. Weidenkaff,
"Thermoelectric oxide modules tested in a solar cavity-receiver,"
Journal of Material Research, vol. 26, pp. 1975-1982, 2011.
D. Hirsch, P. v. Zedtwitz, and T. Osinga, "A New 75 kW High-Flux
Solar Simulator for High-Temperature Thermal and Thermochemical
Research," Journal of Solar Energy Engineering, vol. 125, pp. 117120, 2003.
Thermogage Vatell, 1000 Series. Available:
http://www.vatell.com/thermogage.htm, last checked January 2012
G. Hagmann, Grundlagen der Elektrotechnik. Wiebelsheim,
Deutschland: Aula-Verlag, 2009.
T. P. Hogan and T. Shih, "Modeling and Characterization of Power
Generation Modules Based on Bulk Materials," in Thermoelectrics
G. Min and D. M. Rowe, "Recent Concepts in Thermoelectric Power
Generation," presented at the 21st International Conference on
Thermoelectronics, 2002.
M. F. Modest, Radiative Heat Transfer, second ed. San Diego, USA:
Academic Press, 2003.
R. Siegel and J. Howell, Thermal Radiation Heat Transfer, 4th ed.
New York, USA: Taylor & Francis Inc., 2002.
A catalog of radiation heat transfer configuration factors. Available:
http://www.me.utexas.edu/~howell/tablecon.html, last checked
January 2012.
F. P. Incropera and D. P. DeWitt, Fundamentals of Heat and Mass
Transfer. Hoboken, USA: John Wiley & Sons Inc., 2002.
Fluid Property Calculator. Available:
http://www.mhtl.uwaterloo.ca/old/onlinetools/airprop/airprop.html,
last checked January 2012.
S. Patankar, Numerical Heat Transfer and Fluid Flow. New York,
USA: Taylor & Francis, 1980.
S. J. Chapman, Fortran 95/2003: For Scientists and Engineers, third
ed. Boston, USA: McGraw-Hill, 2008.
C. Suter, P. Tome, A. Weidenkaff, and A. Steinfeld, "A solar cavityreceiver packed with an array of thermoelectric converter modules,"
Solar Energy, vol. 85, pp. 1511-1518, 2011.
108
References
[72]
R. B. Diver, "Receiver/Reactor concepts for thermochemical transport

of solar energy," Solar Energy Engineering, vol. 109, pp. 199-204,
1987.
E. M. Sparrow and R. D. Cess, Radiation heat transfer. Belmont,
California, USA: Wadsworth Publishing Company, 1966.
S. Lin and E. Sparrow, "Radiant Interchange among Curved
Specularly Reflecting Surfaces Application to Cylindrical and
Conical Cavities," J. Heat Transfer, vol. 87, pp. 299-307, 1965.
A. Steinfeld, "Apparent Absorptance for Diffusely and Specularly
Reflecting Spherical Cavities," J. Heat Mass Transfer, vol. 34, pp.
1895-1897, 1991.
A. Steinfeld and M. Schubnell, "Optimum Aperture Size and
Operating Temperature of a Solar Cavity-Receiver," Sol. Energy, vol.
50, pp. 19-25, 1993.
W. T. Welford and R. Winston, High Collection Nonimaging Optics.
San Diego: Academic Press, 1989.
T. Melchior, C. Perkins, A. W. Weimer, and A. Steinfeld, "A CavityReceiver Containing a Tubular Absorber for High-Temperature
Thermochemical Processing Using Concentrated Solar Energy," Int. J.
Thermal Sci., vol. 47, pp. 1469-1503, 2008.
C. Wieckert, A. Meier, and A. Steinfeld, "Indirectly Irradiated Solar
Receiver-Reactors for High-Temperature Thermochemical Processes,"
J. Sol. Energy Eng., vol. 125, pp. 120-123, 2003.
A. ZGraggen and A. Steinfeld, "Radiative Exchange within a TwoCavity Configuration With a Spectrally Selective Window," J. Sol.
Energy Eng., vol. 126, pp. 819-822, 2004.
J. Petrasch, "Thermal Modeling of Solar Chemical Reactors: Transient
Behavior, Radiative Transfer," Master Thesis, ETH Zurich, 2002.
J. Petrasch, "A Free and Open Source Monte Carlo Ray-Tracing
Program for Concentrating Solar Energy Research," in ES2010,
Phoenix, USA, May 17-22, 2010.
O. Yamashita, "Effect of linear temperature dependence of
thermoelectric properties on energy conversion efficiency," Energy
Conversion and Management, vol. 49, pp. 3136-3169, 2008.
A. M. Clausing, J. M. Waldvogel, and L. D. Lister, "Natural
convection from isothermal cubical cavities with a variety of side
facing apertures," J. Heat Transfer, vol. 109, pp. 407-412, 1987.
U. Leibfried and J. Ortjohann, "Convective heat loss from upward and
downward-facing cavity solar receivers: measurements and
calculations," J. Sol. Energy Eng., vol. 117, pp. 75-84, 1995.
T. Taumoefolau, S. Paitoonsurikarn, G. Hughes, and K. Lovegrove,
"Experimental investigation of natural convection heat loss from a
[73]
[74]
[75]
[76]
[77]
[78]
[79]
[80]
[81]
[82]
[83]
[84]
[85]
[86]
109
[87]
[88]
[89]
[90]
[91]
[92]
[93]
[94]
[95]
[96]
[97]
[98]
[99]
[100]
[101]
[102]
[103]
model solar concentrator cavity receiver," J. Sol Energy Eng., vol.

126, pp. 801-807, 2004.
N. S. Kumar and K. S. Reddy, "Numerical investigation of natural
convection heat loss in modified cavity receiver for fuzzy focal solar
dish concentrator," Solar Energy, vol. 81, pp. 846-855, 2007.
"ANSYS Workbench, Academic Teaching Advanced, Release
12.0."
D. Hess, "CFD modeling of convective heat loss from an upwardfacing cavity solar receiver," ETH Zrich, Semester thesis, 2007.
D. Zimmermann, "CFD simulations of a thermoelectric solar cavityreceiver," Bachelor Thesis, ETH Zrich, Bachelor thesis, 2010.
G. J. Snyder, "Application of the compatibility factor to the design of
segmented and cascaded thermoelectric generators," Applied Physics
Letters, vol. 84, 2004.
G. J. Snyder, "Thermoelectric Power Generation: Efficiency and
Compatibility," in Thermoelectrics Handbook: Macro to Nano, Boca
Raton, USA: CRC Press, 2006.
G. J. Snyder and T. S. Ursell, "Thermoelectric Efficiency and
Compatibility," Physical Review Letters, vol. 91, 2003.
V. L. Kuznetsov and M. V. Vedernikov, "Thermocouple with a
Passive High Temperature Superconductor Leg," in Thermoelectrics
C. Suter, Z. Jovanovic, and A. Steinfeld, "A 1 kWel Thermoelectric
Stack for Geothermal Power Generation - Modeling and Geometrical
Optimization," Applied Energy, submitted.
P. K. Kundu and I. M. Cohen, Fluid Mechanics, 4th ed. San Diego:
Academic Press, 2007.
A. Bejan, Convection heat transfer, 2nd ed.: J. Wiley, 1995.
R. K. Shah and A. L. London, Laminar flow forced convection in
ducts. New York: Academic Press, 1978.
S. Kakac and R. K. Shah, Handbook of single-phase convective heat
transfer. New York: Wiley & Sons, 1987.
V. Gnielinski, "Ein neues Berechnungsverfahren fr die
Wrmebertragung im bergangsbereich zwischen laminarer und
turbulenter Rohrstrmung," Forschung im Ingenieurwesen, vol. 61,
pp. 240-248, 1995.
"ANSYS Workbench, Academic Teaching Advanced, Release
13.0."
S. Ackermann, "Thermoelectric Conversion of Geothermal Heat: CFD
Simulations of a Thermoelectric Stack," Semester thesis, ETH Zrich,
Semester thesis, 2011.
"MPDB 5.50, JAHM Software, Inc.," North Reading, USA, 1999.
110
[104]
References
D. I. f. P. Properties, DIPPR Project 801 - Full Version: Design
Institute for Physical Property Research/AIChE, 2010.
Curriculum Vitae
Name
Clemens Suter
Nationality
Swiss
Citizen of
Zrich ZH
Date of birth
January 30, 1985
2006 2008
Master studies at ETH Zurich in Mechanical

Engineering; specialization in renewable energies
2006 2006
Internship at
(Switzerland)
2003 2006
Bachelor studies in Mechanical Engineering at ETH

Zurich; focus in energy, flows & processes
1997 2003
Matura at Gymnasium Freudenberg, Zurich
Paul
Scherrer
Institut,
Villigen
List of publications
Journal Papers:
Suter C., Jovanovic Z., and Steinfeld A., A 1 kW e Thermoelectric Stack for
Geothermal Power Generation Modeling and Geometric Optimization,
Applied Energy, submitted 2012.
Suter C., Tome P., Weidenkaff A., and Steinfeld A., 2011, A solar cavityreceiver packed with an array of thermoelectric converter modules, Solar
Energy, 85, pp. 1511-1518.
Tome P., Suter C., Trottmann M., Steinfeld A., and Weidenkaff A., 2011,
Thermoelectric oxide modules (TOMs) tested in a solar cavity-receiver,
Materials Research, 26, pp. 1975-1982.
Suter C., Tome P., Weidenkaff A., and Steinfeld A., 2010, Heat Transfer and
Geometrical Analysis of Thermoelectric Converters Driven by Concentrated
Solar Radiation, Materials, 3, pp. 2735-2752.
Tome P., Trottmann M., Suter C., Aguirre M.H., Steinfeld A., Haueter Ph.,
and Weidenkaff A., 2010, Thermoelectric Oxide Modules (TOMs) for the
Direct Conversion of Simulated Solar Radiation into Electrical Energy,
Materials, 3, pp. 2801-2814.
Conference Proceedings:
Suter C., Jovanovic Z., and Steinfeld A., 2011, A 1 kWel Thermoelectric Stack
for Geothermal Power Generation Modeling and Geometrical
114
List of publications
Optimization, Proceedings of the 9th European Conference on Thermoelectrics

ECT 2011, September 28 30, Thessaloniki, Greece.
Suter C., Tome P., Weidenkaff A., and Steinfeld A., 2010, Thermoelectric
Converters Driven by Concentrated Solar Irradiation: Heat Transfer Analysis
and Cavity-Receiver Design, Proceedings of the 8th European Conference on
Thermoelectrics ECT 2010, September 22 24, Como, Italy.

Thermoelectric Conversion of Concentrated Solar Radiation and Geothermal Energygeothermal Energy PDF

Diunggah oleh

Informasi Dokumen

Deskripsi Asli:

Judul Asli

Hak Cipta

Format Tersedia

Bagikan dokumen Ini

Bagikan atau Tanam Dokumen

Opsi Berbagi

Apakah menurut Anda dokumen ini bermanfaat?

Apakah konten ini tidak pantas?

Hak Cipta:

Format Tersedia

Thermoelectric Conversion of Concentrated Solar Radiation and Geothermal Energygeothermal Energy PDF

Diunggah oleh

Hak Cipta:

Format Tersedia

Diss. ETH No.

accepted on the recommendation of

concentration ratio of 1500 suns in the aperture and in a temperature range of

Electricity from renewable energy sources..................................... 1

Electricity from solar energy .......................................................... 1

Geothermal energy ......................................................................... 3

Decentralized thermoelectric power generation ............................. 4

Solar thermoelectric conversion........................................ 4

Geothermal thermoelectric conversion ............................. 5

Thesis outline ................................................................................. 5

Thermoelectric basics .............................................................................. 7

Schematic of a thermoelectric converter module ........................... 8

Peltier and Thomson effect ........................................................... 10

Prospective of material research ................................................... 10

State-of-the-art application ........................................................... 11

Directly irradiated modules .................................................................. 13

4-leg perovskite-type oxide modules ........................................... 13

p-type La1.98Sr0.02CuO4 and n-type CaMn0.98Nb0.02O3 .... 14

Module assembly ............................................................ 15

Experimental setup ....................................................................... 15

Experimental results ..................................................................... 17

Heat transfer model ...................................................................... 21

Heat conduction .............................................................. 22

Radiative heat transfer .................................................... 23

Natural heat convection .................................................. 25

Maximum power output and solar-to-electricity efficiency

Boundary conditions and model parameters ................... 26

Numerical solution ......................................................... 26

Comparison between semi-transparent and opaque absorber plate

Model validation .......................................................................... 28

Heat transfer analysis ................................................................... 29

Geometric analyses ...................................................................... 31

Variation of leg length .................................................... 32

Packing of legs ............................................................... 32

Summary and conclusion ............................................................. 35

A 1 kWth solar cavity-receiver prototype ............................................ 37

Advantage of cavity design .......................................................... 38

Experimental setup ....................................................................... 38

Experimental results ..................................................................... 41

Heat transfer model ...................................................................... 43

Heat conduction .............................................................. 44

Radiative heat transfer .................................................... 46

Natural heat convection .................................................. 47

Boundary conditions and model parameters ................... 48

Numerical solution .......................................................... 49

Heat transfer analysis ................................................................... 51

Summary and conclusion ............................................................. 52

Towards a 50 kWth solar cavity-receiver ............................................. 53

Idea of cascading .......................................................................... 53

Cascaded dual-stage modules ....................................................... 54

Heat transfer model ...................................................................... 55

Heat conduction .............................................................. 57

Radiative heat transfer .................................................... 57

Maximum power output and solar-to-electricity efficiency

Boundary conditions and model parameters ................... 59

Numerical solution .......................................................... 59

Simulation results ......................................................................... 60

Geometrical optimization and heat transfer analysis ...... 61

Solar-to-electricity efficiency ......................................... 63

Summary and conclusions ............................................................ 64

A 1 kWel geothermal stack .................................................................... 67

Schematic of thermoelectric stack ................................................ 67