You are on page 1of 19

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/261727418

Dimensioning and Evaluating a Multi-Channel Thermoelectric Generator


Using a Costumized Simulation Architecture

Article · January 2013

CITATIONS READS

4 114

5 authors, including:

Roland Kühn Hans-Fridtjof Pernau


Technische Universität Berlin Fraunhofer Institute for Physical Measurement Techniques IPM
21 PUBLICATIONS   33 CITATIONS    44 PUBLICATIONS   289 CITATIONS   

SEE PROFILE SEE PROFILE

Daniel Jänsch
IAV
19 PUBLICATIONS   15 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

NanoCaTe View project

novel working pairs for sorption chillers View project

All content following this page was uploaded by Roland Kühn on 10 August 2015.

The user has requested enhancement of the downloaded file.


18 Dimensioning and Evaluating a Multi-Channel
Thermoelectric Generator Using a Customized Simulation
Architecture
Jens Kitte, Roland Kühn, Hans-Fridtjof Pernau, Kristof Littmann,
Daniel Jänsch

Abstract
Thermoelectrics provide a way of recovering thermal energy in the vehicle. However,
the combination of classic heat exchanger and thermoelectric material integrated into
the vehicle environment results in high demands on dimensioning the thermoelectric
generator system. Without detailed analysis, this prevents the thermoelectric system
from providing any sufficient benefit. This paper discusses simulation results
generated using a simulation structure adapted to suit thermoelectrics. The
investigations carried out are presented on the basis of multi-channel thermo-
generators for a gasoline-engined vehicle.

Three abstraction levels exist within the overall model environment that are used as
required during the dimensioning process. To begin with, a rough configuration is
defined in a series of steps on the basis of a continuous model. Although this
abstraction level provides the capability of computing many different variations, it
proceeds from a highly simplified basis.

CFD simulation and FE simulation represent the levels with the highest degree of
detailing. They serve as verification of the results from the higher abstraction levels.
A simulation environment exists between the abstraction levels presented that is a
compromise between the environments described. This models the system fairly
accurately while requiring computing times that still permit an analysis of different
variants.

The design process finally produces three thermogenerator variants that perform in
the targeted way. These are examined and assessed under various aspects at the
end of the paper. The summary shows the most favorable system under the specified
boundary conditions.

207
1. Introduction

Thermoelectrics provide an attractive concept for recovering and converting thermal


energy in the vehicle. Direct conversion into electric energy and the good dynamic
behavior produce remarkable advantages over alternative concepts, such as Rankine
cycle [1]. To extract maximum potential from the overall thermoelectric system, it is
expedient to combine analysis and synthesis within the design process. Here, the
interplay of thermoelectric material and heat transfer system is crucial to the way in
which the system ultimately performs.
Section 2 presents the virtual model environment in terms of its basic approach and
objectives. It provides the means for an integrated analysis of the thermoelectric
system from heat exchanger to material level.
Approaches to configuring a TEG entirely on a virtual basis are presented in [2][3].
There the thermoelectric leg pair represents the smallest unit. The results are
exported into a map model that is used within the process of simulating the heat
exchanger. As a result, the virtual design of the planar thermogenerator up to leg-pair
level must be regarded as state of the art. However, the environment presented here
permits analysis of an individual leg segment.
Section 3 uses the example of a planar architecture to discuss dimensioning a multi-
channel thermoelectric heat exchanger. The planar architecture of a thermoelectric
system belongs to the state of the art and is presented, for example, in [2]. This
section discusses the relationships between the system’s design and the boundary
conditions resulting from the vehicle environment and derives appropriate structures.
Attention also turns to the properties of various fin and channel options and their
influence on thermoelectric efficiency.
Section 4 compares and assesses the resultant channel variants. Among other
aspects, the evaluation criteria concentrate on the maximum power output capable of
being generated as well as on power density, system efficiency, power variation for
different operating points and on sensitivity. The system variables, such as gas-side
pressure loss and transferred heat flow that are responsible for the main interactions
within the vehicle environment, are also compared.

2. Model and Simulation Environment


The thermoelectric generator is synthesized and analyzed using a specifically aligned
simulation environment. Simulation at various detailing levels provides the ideal basis
for comprehensively analyzing the thermoelectric system and subsequently ensuring
that it is dimensioned in the optimum way (Figure 1).
From simulating the overall system to detailed analysis of selected zones in the
thermoelectric configuration, the simulation environment provides the capability of
dimensioning all aspects of the thermogenerator. The virtual model environment of
the thermogenerator covers the three following detailing levels:

1. Steady-state continuous model


2. Dynamic quasi-two-dimensional model
3. CFD (Computational Fluid Dynamics) and FE model (Finite
Element Method)

208
Semi-empiric heat
Continous
exchanger
TE model
Utilize: Optimization model
Computing time: < 0,2 s
Contraints Verifikation Contraints Verifikation

Quasi-2D Quasi-2D heat


TE model Utilize: holistic TEG design exchanger model
Real time
Contraints Verifikation Contraints Verifikation

FE simulation CFD simulation


Utilize: localized TEG design
Computing time: < 1 h

Figure 1: Simulation environment for assessing and optimizing thermoelectric generators

Proceeding from the steady-state continuous model, boundary conditions are


successively determined for the next higher degree of detailing. A high-speed, quasi-
steady-state continuous and a dynamic model are available for simulating the
thermoelectric heat exchanger. Both models provide the framework for dimensioning
the generator concept in a holistic way. Besides there are some boundary conditions
given by the overall system without thermoelectric generator which are the maximum
permissible pressure loss and the maximum electric power output.
CFD simulation (Computational Fluid Dynamics) and FE (Finite Element) simulation
represent the level with the lowest degree of abstraction and the highest level of
model quality. However, the high amount of input involved in modeling and high
computing speed only permit examination of a few selected concept variants and
operating points.
The model levels are verified in reverse order so that errors, e.g. in simplifying the
higher abstraction levels, can be quickly identified and eliminated.

2.1 Continuous Model

The continuous model represents the concept with the highest level of abstraction
and the shortest computing time. Simulation takes place from material to heat-
exchanger level. In general, the thermoelectric system is modeled as a 2 or 3-
dimensional matrix. All variables affecting the thermoelectric system are presented in
matrix form. A new approach transforms discrete analysis of the system into a
continuous analysis across the entire heat exchanger along exhaust-gas flow
direction x.
The heat exchanger can be viewed in segmented fashion along the flow direction
(coordinate x). A segment section is presented as a thermally and electrically self-
contained unit with corresponding electric load resistance.
Cumulating individual power segments produces the stepped curve profile for the
total electric power indicated (Figure 2). Linearization between individual steps
provides the basis for differential analysis of the thermoelectric system.

209
Using a natural exponential function, electric power can thus be described as a
closed solution. The model is linked with a semi-empirical heat exchanger model for
computing heat transfer behavior on the gas side. The temperature of the heat sink is
assumed to be constant.
The gas temperature profile between heat exchanger inlet and outlet can also be
described using a natural exponential function.

Pel Approximated continuous


el. power Pel

Gas temperature Peel Electrical power Pel


Coolant temperature Discrete power Pel Heat flow Q’
Gas mass flow results from matrix model System efficiency

x Flow direction x

Pel dPel

Current

Pel I ( x) dU x ( x)

Differential
voltage

Geometry of
heat exchanger
Figure 2: Transforming the discrete model into a continuous model

This means the thermoelectric heat exchanger is not treated on a discrete basis and
the equations can be solved explicitly. The model computes the system’s final steady
state value that is produced for the input variables.
This high-speed model is used, for instance, to find the optimum leg geometry within
different heat exchanger configurations. Besides the CAD design, the results provide
the geometric input variables for the dynamic quasi-two-dimensional model. Selected
model results were verified with results from CFD and FE simulation.

2.2 Dynamic Quasi-Two-Dimensional Thermogenerator Model

Proceeding from the boundary conditions determined from the continuous model, the
TEG is further dimensioned using the dynamic quasi-two-dimensional model.
The thermoelectric system is embedded in a finite heat source and heat sink with
flowing work media. The difference in temperature between source and sink, and
thus between the work media entering and leaving, has a major influence on system
performance and is therefore modeled as accurately as possible.
Figure 3 shows the schematic simulation structure of the dynamic quasi-two-dimen-
sional thermogenerator model.

210
Figure 3: Simulation structure of the dynamic quasi-two-dimensional thermogenerator model

The system’s geometry as well as temperature and pressure-related substance


functions are defined at the start of simulation, with the input variables that change
over time, i.e. temperature, pressure and mass flow of the exhaust gas and of the
cooling water, then being transferred to the model. The pressure loss over running
length, the temperature field in the exhaust gas flow and cooling water in flow
direction as well as the electric power generated and other electric quantities form the
output variables.

Figure 4: System diagram of the transient quasi-two-dimensional thermogenerator model

The system diagram in Figure 4 shows the structure of the model’s control volumes.
The exhaust gas (control volume 1) flows in the opposite direction to the cooling flow
(control volume 8). The thermoelectric material lies between the two flows (control
volumes 3-6) that is separated from the fluids by the cold or hot wall (control volumes
2 and 7). The non-linear temperature relationship between the thermoelectric
material properties makes it imperative to divide the legs into four separate control
volumes (3 to 6), permitting quasi-differential analysis in z direction. This step was
selected to determine temperature distribution in the two dimensions of x und z while
still permitting simulation with the greatest possible real-time capability.
The parasitic heat flows between the TE legs (control volume II) and between the
modules (control volume III) are taken into account.
The model is used in globally dimensioning the system of the thermoelectric heat
exchanger and also in dynamic vehicle simulation. The model was verified with
results from CFD and FE simulation.
211
3. System Dimensioning of a Multi-Channel System

System dimensioning is based on the example of a multi-channel system with three


different numbers of channels. The boundary conditions to be considered within the
design process are explained step by step.

3.1 Changeable Variables

Designing the thermoelectric generator involves responding to different boundary


conditions. To adapt the thermogenerator to the different boundary conditions, its
configuration can be modified in numerous ways.
The main variables to be included in modifying the thermoelectric generator are:

1. Leg edge length


2. Gap between leg pairs
3. Leg height
4. Fin dimensioning
(spacing, thickness, height)
5. Fin material
6. Channel dimensioning

The thermoelectric material used is lead telluride and belongs to the state of the art
[4]. The thermoelectric leg height is selected as the analytical reference variable. It
influences both thermal and electric performance and can be quickly adapted in
terms of manufacturing.

3.2 Boundary Conditions

Analysis of the heat exchanger is geared towards the “top-down” design process.
Appropriate constraints and boundary conditions that influence the thermoelectric
generator’s configuration are derived from the familiar application environment. The
aim is to determine the generator variants that achieve the absolute power output
demanded. Here, the influence of the constraints and boundary conditions defined in
Figure 5 are taken into consideration one after the other, with the relevant changes
being made to the thermogenerator.

Consequently, the thermogenerator is analyzed in the following five steps (Figure 5):

1. Definition of thermal and volumetric boundary conditions


2. Definition of finite fin structure while taking account of fluid-
dynamic boundary conditions
3. Definition of fin structures that can be realistically manufactured
4. Definition of material constraints
5. Definition of geometric constraints

212
"Constraints for top-down design process"

Maximum temperature and mass


Thermal flow

Volumetric Maximum length, height


and width

Fluid mechanics Maximum pressure loss

Manufacturing Design of fins

Heat conduc- Material of fins


tance of fins

Constraints betw.
Geometric height of legs and
Geometric
height of channel

Figure 5: Design process – Analysis of heat exchanger

3.2.1 Thermal and Volumetric Boundary Conditions

Among the various boundary conditions for the thermoelectric system are the thermal
boundary conditions that focus on the operating range of the TEG in the vehicle. For
example, these include a typical urban, interurban and highway drive. The position at
which the thermogenerator is installed in the exhaust system is downstream of the
gasoline engine’s main catalyst. In the standard-size car selected, operating points
prevail with exhaust gas temperatures of between 450°C and 800°C and exhaust gas
mass flows of between 50 kg/h and approximately 200 kg/h. The design point is
defined at an exhaust gas mass flow of 110 kg/h and an exhaust gas temperature of
607°C.
The engine cooling circuit with a water temperature of 90°C and an independent low-
temperature circuit at 50°C are available for cooling the TEG. The volumetric
boundary conditions include the maximum available package in the vehicle’s exhaust
system, thus defining the thermoelectric system’s maximum length, width and height.

3.2.2 Production-Related Boundary Conditions

In the further course, it is also necessary to define production-related boundary


conditions. These set limits for fin thickness and fin density.
In this context, fin density describes the share accounted for by the sum of all fin foot
widths in relation to channel width. Two fin versions are proposed

1. Weight-saving fin (0.2 mm thick, fin density 10%)


2. Power-maximizing fin (1 mm thick, fin density 50%)

213
Both versions are used in configuring the heat exchanger.

3.2.3 Geometric Boundary Conditions

The given volumetric boundary conditions produce geometric constraints because as


channel height increases less space is available for the thermoelectric material, the
cooling water channels and the electronics.
Increasing the number of channels results in a disproportionate reduction in available
gas channel height because both the TE material as well as the electronics and the
cooling water channels always take up a certain amount of package per channel. As
a result, the number of channels has a direct influence on fin height and thus on heat
transfer behavior.
This is seen against the behavior of the thermally embedded thermoelectric material.
The poorer the heat transfer behavior, i.e. the smaller the channel is the higher
optimum leg height will be [5], as can be seen in Figure 6.
Absolute electrical power in W
0.010
600 Design point
Geometric constraints
Multichannel TEG
0.009
Fin: variant 2

0.008
Optimum Thermoelectric optimum
Maximum
0.007 power
Height of legs in m

0.006
630 650
0.005 640

0.004
600

0.003
550

0.002
500
450
400
40
0.001 350
0.030 0.032 0.034 0.036 0.038 0.040 0.042 0.044 0.046 0.048
Height of channel in m
Figure 6: Relationship between thermoelectric optimum and geometric constraint

Figure 6 shows the relationship between channel height and TE leg height using the
example of a multi-channel system. Maximum power would be achievable at the
greatest possible channel height (48 mm), but leaving just 1 mm height for TE
material and not 6 mm, if the volumetric boundary conditions should be met.
However, with the above-described geometric boundary conditions, it is only possible
to realize thermogenerators with a leg and channel height that results in points in the
gray shaded area. In this case, maximum power is always produced at the right edge
of the gray area. In the figure, the real geometry of maximum power is marked and
occurs in the example at a leg height of 6 mm and a channel height of 38 mm. Under
the given geometric boundary conditions, it is not possible to find any point delivering
maximum power at the thermoelectric optimum. This means that the maximum power
achieved here does not coincide with maximum efficiency.

214
3.2.4 Material Constraints – Influence of Fin Material

In the same way as the ratio between width and height, the fin material’s thermal
conductivity plays a crucial part in determining heat transfer.
Thermal resistance, which is also governed by fin geometry, can be influenced by
changing fin material. The thermal conductivity of a fin with a large ratio between
height and width has a far greater influence on the thermoelectric system than it does
on a system with short or wide fins. This is illustrated in Figure 7 using the example of
a multichannel thermogenerator.

Absolute electrical power in W


Steel Nickel Copper Steel Nickel Copper
200 W 390 W 470 W 327 W 500 W 540 W
58% 39%
10 10
Design point Design point
Multichannel TEG Multichannel TEG
9 9 400
Fin: variant 1 Fin: variant 2
8 8

7 7
Height of legs in m

Height of legs in m

460
6 6

5 400 5 500
440
470
4 460 4 530
350 539
3 3

280
2 2

200
1 1
0 50 100 150 200 250 300 350 400 0 50 100 150 200 250 300 350 400
Heat conductance of fin material Heat conductance of fin material
in W/m/K in W/m/K
Figure 7: Contour lines for electric power as a function of thermal conductivity and leg height (1-
channel TEG)

Figure 7 permits a comparison between the sheet-metal fin with a large height-to-
width ratio and a milled fin with a small ratio between height and width. The power
difference between a thermogenerator with a steel fin and one with a copper fin
amounts to 39% for the milled version. The system with sheet-metal fin shows a
power drop of approximately 58%. This makes the choice of material particularly
important for thin-walled fins.

215
3.3 Determining an Appropriate Number of Channels

The boundary conditions presented provide the basis for deciding on how many
channels the thermoelectric generator should have. Figure 8 shows selected variants
from the analyses in the design process. All electric power outputs are normalized to
the ideal thermoelectric generator that is only restricted by the thermal and volumetric
boundary conditions.
The fluid-mechanical boundary conditions for a finite fin density already accounts for
a power drop of at least 37%. Using fins that can actually be manufactured causes
the system’s power output to fall further. Changing the channel height, and thus the
possible fin length and transferrable thermal output, leads to a further significant
reduction in electric power output on account of the geometric boundary conditions.

0.82
Design point + Fluid mechanic constraints ( 50 mbar)
0.76 + Manufacturing constraints
+ Geometric and material constraints
0.70
Rel. electrical power

0.64 0.63

0.58

0.52
Fin: variant 2 (Nickel)
0.46
Target power
0.40
0 1 2 3 4 5 6 7 8 9 10
Number of channels
Figure 8: Power profile for an n-channel thermogenerator taking into account all previous boundary
conditions

The power curve bends more sharply as more boundary conditions are taken into
account. After allowing for all boundary conditions discussed, the two-channel
thermogenerator is ultimately shown to be capable of generating the maximum
electric power. The system undergoes the highest power loss as a result of the fluid-
mechanical boundary condition. Figure 8 shows that generable electric power falls
rapidly with more than three channels in the thermogenerator which is why the
maximum number of channels is set at three. The three non-optimized variants of the
thermogenerator can be seen in Figure 9.

216
Channels

Figure 9: Selected TEG variants

4. Comparison and Assessment of the Channel Variants


Produced

Overall system analysis and flexibility within the simulation environment presented
permit assessment of a number of factors that are important in respect of integrating
a thermogenerator into the vehicle environment. All three channel variants are
compared below – after they have been optimized.

4.1 Absolute Electric Power

A target value to be reached is set for the absolute electric power output. Both the
one and two-channel thermogenerator reach the target power output. Given its
simplicity, the single-channel TEG is to serve as a weight-optimized solution and is
therefore equipped with the lighter-weight sheet-metal fins. The multichannel variants
are designed for the highest possible electric power output and are therefore
equipped with the heavier fins milled from a block. The three-channel TEG only
reaches the set target power output with the milled fins anyway.
The total power output from the channel variants selected is compared in Figure 10.

217
1-channel TEG
1.6
2-channel TEG
3-channel TEG
1.4
Relative power

1.2

1.0

0.8

0.6
Urban Interurban drive Highway
drive (design point) drive

Figure 10: Absolute power output compared with single-channel TEG

The single-channel thermogenerator provides the reference at each operating point


and is therefore set at 100%. The two-channel thermogenerator exhibits the highest
electric power output during urban driving and at the design point. The three-channel
TEG attains the highest power output on the highway. This is explained in greater
detail at the end of the next section.

4.2 Difference in Power Output

It is not only the absolute electric power output at the design point that is important in
assessing the thermoelectric generator but also the way in which the power output
behaves in response to a change in the load point. Figure 11 shows the change in
power output for the three channel variants in the typical operating range from urban
to highway driving.

218
3.5
1-channel TEG
2-channel TEG
3.0
3-channel TEG

2.5
1.55
Relative power

1.79
2.0

1.5 0.92

1.0

0.5

0.0
Urban Interurban drive Highway
drive (design point) drive

Figure 11: Difference in power output in the operating range

Whereas the three variants hardly differ for urban driving, the multichannel variants
deliver a far higher power output at the design point than the single-channel
thermogenerator. On the highway drive, the power output from the three-channel
TEG then increases once again, and more than it does for the other variants.
This can be explained in the diagram shown in Figure 12.

Thermoelectric optimum Thermoelectric optimum


(3- channel TEG design point) (1- channel TEG design point)
Thermoelectric optimum Thermoelectric optimum
(3- channel TEG maximum point) (1- channel TEG maximum point)
0.010

0.009
Geometric Configuration – 3-channel TEG
0.008
constraints at design point
Height of legs in m

Optimum
0.007
is shifted
0.006

0.005 Configuration – 1-channel TEG


at design point
0.004
Optimum
0.003 is shifted

0.002

0.001
0 1
Height of channel
Figure 12: Shift in optimum

219
All three variants are optimized for the design point presented in section 3.3. Figure
12 indicates the geometric design points for the single- and three-channel TEG.
The thermoelectric optimum line can also be seen for two operating points that
indicate maximum power output for the given channel height. Increasing mass flow
shifts the optimum line down. This results in an optimum line moving away from the
set point of the single-channel TEG whereas the optimum line of the three-channel
variant approaches its geometric design point.

4.3 Power Density and System Efficiency

In addition to absolute electric power, a further key assessment variable is efficiency.


It is formed from the ratio between electrically generated power output and thermal
power taken up from the exhaust gas.
Figure 13 shows that system efficiency is greatest for the single-channel TEG and
that efficiency falls as the number of channels rises. The single-channel TEG
furthermore exhibits the highest level of material utilization. This is because at most
load points the single-channel TEG’s long fins and small heat exchanger basic
surface area produce a far greater difference in temperature between the cold and
hot side of the thermoelectric material than is the case for the other variants.

8.0 1.0
1-channel TEG
(with regards to TE-material)

2-channel TEG
7.0 3-channel TEG
0.8
Relative power density
System efficiency

6.0
0.6
5.0
0.4
4.0

0.2
3.0

2.0 0.0
Urban Interurban drive Highway Urban Interurban drive Highway
drive (design point) drive drive (design point) drive

Figure 13: System efficiency and mass-specific electric power output of the TEGs

The heat exchanger behavior described for the single-channel system permits
sparing use of the thermoelectric material. In relation to the TE material, the power
density of the two and three-channel thermogenerators is less than 50% of that of the
single-channel system. Even taking into consideration the 25% higher power output
from the multichannel variants, they require a far greater quantity of thermoelectric
material. If all variants were to generate the same total power output, the three-
channel system would need about five times as much thermoelectric material as the
single-channel system.
These aspects give the single-channel thermogenerator a significant advantage over
the other variants.

220
4.4 Sensitivity

The different thermoelectric generator variants respond differently to geometric and


manufacturing changes. The sensitivity to such changes in the overall project context
is important insofar as it indicates the extent to which errors in simulation can lead to
deviations from experiment-based results.
An example is now examined to illustrate the system’s sensitivity to using different
materials for the fins. This not only represents a production-related change but also
makes it possible to get an idea about how the changes during long-term use affect
the quality of the thermogenerator. While the thermoelectric generator is operating,
for example, heat transfer can be restrained by corrosion or clogging of the spaces
between the fins. Figure 14 shows the effects of changed material on the basis of
absolute electric power output from a system using copper fins.
The single-channel thermogenerator demonstrates the greatest drop in electric power
from using a poorly conducting fin material. As already explained, this is because of
the large ratio between fin length and fin thickness of the sheet-metal fins. Heat must
be fed over a long distance through a narrow cross section, reducing the heat flow
that can be transferred. The single-channel TEG shows a power loss of over 40% for
a reduction in the coefficient of thermal conductivity from 400 W/(mK) to 10 W/(mK).

1.00
1-channel TEG Design point
2-channel TEG
0.90 3-channel TEG
Relative power

0.80

0.70

0.60

0.50
100 10 100 10 100 10
Heat conductance of fin in W/m/K
Figure 14: Influence of the coefficient of thermal conductivity on power output

By comparison, the three-channel thermogenerator with milled fins suffers a loss of


8% for a reduction on the same scale. This makes the material used less relevant to
the three-channel TEG. The shorter fins of the three-channel TEG also give it an
advantage over the two-channel system.

4.5 Pressure Loss and Heat Flow

Variables influencing the overall system environment play a key part in the design
process. These include pressure loss behavior on the gas side and the heat flow

221
transferred. The channel variants exhibit different characteristics here, as shown in
Figure 15.

All systems analyzed exhibit a laminar-type flow over the entire operating range. This
means that heat transfer behavior and pressure loss between gas and wall are
essentially defined by fin architecture and flow velocity.
As the narrow flow cross sections prevent any flow turbulence, pressure loss does
not rise in line with flow velocity but slightly faster.

Even at low heat transfer coefficients, the long fins of the single-channel
thermogenerator give it a high level of transferred heat output. As the flow-through
area is larger than that of the other two variants, a high level of heat transfer can be
achieved on the gas side while still keeping pressure loss low.

30 22
20
25
Pressure loss in mbar

18
Heat flow in kW

20 16
14
15
12
10 10
8
5
6
0 4
Urban Interurban drive Highway Urban Interurban drive Highway
drive (design point) drive drive (design point) drive

Figure 15: Comparison of pressure loss and transferred heat flow

The heat flow transferred to the heat sink is a further important assessment criterion.
As the cooling system must accommodate this additional load, cooling water
temperature may rise. Within certain limits, a higher-dimensioned cooling concept
provides the capability of taking up an additional quantity of heat. On exceeding a
heat quantity limit, however, measures must be provided to shut down the
thermoelectric generator.

The three-channel thermogenerator transfers the greatest flow of heat although it has
the smallest specific output per unit area because it also exhibits by far the largest
total heat exchanger surface area. Accordingly, the temperature gradient between
inlet and outlet is greatest. In terms of transferred heat quantity and pressure loss,
this makes the single-channel thermogenerator the most favorable variant for vehicle
integration.

222
5. Summary and Outlook

The process of designing a thermoelectric system differs significantly from that of


configuring a classic heat exchanger. Optimizing the design demands accompanying
support from virtual analysis. The virtual development environment was explained
using the example of a multichannel system. Analysis of the thermogenerator
provided the basis for designing a suitable planar heat exchanger geometry under
specific constraints and boundary conditions. The single, two and three-channel
thermogenerators were regarded here as being suitable for achieving the goal that
was set.

On concluding the design phase, all three variants were compared with regard to the
assessment criteria selected.
The results show that the single-channel system exhibits the highest power density
while at the same time providing the greatest thermoelectric system efficiency with
the smallest quantity of thermoelectric material. This means the single-channel TEG
has the highest utilization factor regarding the thermoelectric material. This system
also comes with the lowest pressure loss and puts the least strain on the vehicle’s
cooling system. This provides advantages in implementing it in the vehicle.

The drawback of the single-channel system lies in its high sensitivity to change in
boundary conditions and geometry. Using the example of reducing the fins’ thermal
conductivity, it was possible to show that the three-channel system is least sensitive.

Finally, the single-channel thermoelectric generator represents the variant that is


most favorable in economic terms for implementation in the vehicle under the
assessment criteria considered. However, the highest power output at the design
point is generated by the two-channel thermogenerator.

Proceeding from the potential for reducing consumption, future investigations will
involve comparing and assessing scalability and, in particular, system costs. Here,
simulating and evaluating entire driving cycles will play a part in future.

223
Reference

[1] R. Freymann, Der Turbosteamer der zweiten Generator, BMW, MTZ 2012
[2] B. Mazar, Gesamtsystemoptimierung eines thermoelekrischen Generators
für eine Fahrzeuggruppe, Disseration (BMW), (2010),
[3] B. Mazar, Die gesamthafte Auslegung eines TEGs vom thermoelektrischen
Material bis zum Kraftfahrzeug, Thermoelectrics Goes Automotive, HdT
Conference, Berlin 2010
[4] J. LaGrandeur1, Automotive Waste Heat Conversion to Electric Power using
Skutterudite, TAGS, PbTe and BiTe, ICT '06. 25th International Conference
on Thermoelectrics
[5] K. Matsubara, M. Matsuura, CRC Thermoelectrics Handbook (Boca
Raton,FL:CRC,2006), pp. 52-4

Acknowledgements

The authors want to thank the Bundesministerium für Bildung und Forschung and all
project partners fort the highly appreciated support and cooperation.

224

View publication stats

You might also like