Thermoelectric Generation from Solar Water Heater Excess Heat Tyron John Ellul
Institute for Sustainable Energy University of Malta
A dissertation submitted to the Institute for Sustainable Energy In partial fulfilment of the requirements for the degree of Master of Science in Sustainable Energy September 2014
i
University of Malta Library – Electronic Thesis & Dissertations (ETD) Repository The copyright of this thesis/dissertation belongs to the author. The author’s rights in respect of this work are as defined by the Copyright Act (Chapter 415) of the Laws of Malta or as modified by any successive legislation. Users may access this full-text thesis/dissertation and can make use of the information contained in accordance with the Copyright Act provided that the author must be properly acknowledged. Further distribution or reproduction in any format is prohibited without the prior permission of the copyright holder.
STATEMENT OF AUTHENTICITY No portion of the work referred to in the dissertation has been submitted in support of an application for another degree or qualification of this or any other university or other institute of learning
Tyron John Ellul
ii
COPYRIGHT NOTICE Copyright in text of this dissertation rests with the Author. Copies (by any process) either in full, or of extracts, may be made only in accordance with regulations held by the Library of the University of Malta. Details may be obtained from the Librarian. This page must form part of any such copies made. Further copies (by any process) of copies made in accordance with such instructions may not be made without the permission (in writing) of the Author. Ownership of the rights over any original intellectual property which may be contained in, or derived from, this dissertation is vested in the University of Malta and may not be made available for use by third parties without the written permission of the University, which will prescribe the terms and conditions of any such agreement.
iii
ABSTRACT Solar water heater installations in hot climate zones such as Malta produce significant amount of excess heat such that some suppliers suggest to partial shading of the collectors to avoid operation at the upper limit of the system. Thermoelectric generation could be the ideal candidate to recover some of this heat and avoid having to shade the solar collector. A literature review on thermoelectric generation was carried out to better understand the principal of operation and how to improve the performance at the design stage. Thermoelectric conversion occurs when a thermoelectric module is subject to a temperature difference across its surfaces and heat is allowed to flow across. As a result an electromotive force is generated across its terminals. A detailed explanation of the process is presented in the literature review and theory section. A thermoelectric cooler module operates in reverse mode, i.e. when voltage is applied across its terminals, current flows through and as a result heat flows from one side of the module to the other. Thermoelectric generator modules are specifically built for power generation but alternatively thermoelectric cooling can be used in reverse operation. The advantage of using cooling modules for power generation is a reduction of 90% in modules cost. This figure is based on the purchase cost of the modules used in this project. Lower temperature applications are less common due to reduced thermoelectric conversion efficiency. However higher temperature studies were analyzed to assess the conclusions presented and acquire data to compare with the results of this project. Five different setups were built to increase knowledge on thermoelectric generation. Based on this data a final setup of 4 thermoelectric modules was built and connected to a flat-plate solar water heater. Performance comparison between cooler and generator modules is presented. Data from different setups was compared to each other and to results presented in the literature review. The results of all setups were superior or equivalent to those present in the literature review. However this does not necessarily mean that this project produced better results. The performance of a thermoelectric setup is assessed by the generated power against temperature
iv
difference. It is explained that the difference in performance may be related to the ambiguity of temperature measurement. The ambiguity is in how temperature sensing is done, that is, at the cooling medium and hot source entry point or directly over the module surfaces.
v
ACKNOWLEDGEMENTS I would like to express my deepest appreciation to my tutor Ing. Robert. N. Farrugia B.Eng.(Hons)(Melit.),M.Phil.(Melit.) and co-supervisor Dr. Ing. Mario Farrugia B.Eng.(Hons.),M.Sc.(Hull), Ph.D.(Oakland) for their help and guidance throughout this thesis project.
vi
TABLE OF CONTENTS STATEMENT OF AUTHENTICITY...................................................................................ii COPYRIGHT NOTICE .................................................................................................... iii ABSTRACT .....................................................................................................................iv ACKNOWLEDGEMENTS ...............................................................................................vi LIST OF FIGURES .......................................................................................................... x LIST OF TABLES ........................................................................................................... xii LIST OF GRAPHS......................................................................................................... xiii GLOSSARY OF SYMBOLS .......................................................................................... xiv LIST OF ABBREVIATIONS ........................................................................................... xvi Chapter 1 –INTRODUCTION .......................................................................................... 1 1.1 General information ............................................................................................... 1 1.2Objectives of project ............................................................................................... 1 Chapter 2 – LITERATURE REVIEW ............................................................................... 3 2.1 Thermoelectric generators ..................................................................................... 3 2.2 The Seebeck effect ................................................................................................ 4 2.3 Segmented generators .......................................................................................... 6 2.4 Cascaded Thermoelectric generators .................................................................... 8 2.5 Thermoelectric modules ......................................................................................... 9 2.5.2 Thermoelectric systems ................................................................................ 11 2.6 Chapter conclusion .............................................................................................. 23 Chapter 3 – THEORY.................................................................................................... 25 3.1 Thermoelectric generator parameters .................................................................. 25 Chapter 4 – CONSTRUCTION, DEVELOPMENT AND TESTING ................................ 30 4.1 Thermoelectric Generator .................................................................................... 30 4.2 The Thermoelectric Cooler................................................................................... 31 4.3 Thermoelectric Module Assembly ........................................................................ 32 vii
4.4 System Design Considerations ............................................................................ 34 4.5 Setup 1................................................................................................................. 37 4.6 Data logger .......................................................................................................... 40 4.6.1 Temperature measurement ........................................................................... 42 4.6.2 Voltage measurement ................................................................................... 48 4.6.3 Current measurement ................................................................................... 49 4.7Boiler assembly..................................................................................................... 50 4.8 Temperature controller ......................................................................................... 54 4.9 Setup 2................................................................................................................. 55 4.9.1 Experiment procedure ................................................................................... 59 4.10 Setup 3............................................................................................................... 60 4.11 Setup 4............................................................................................................... 62 4.12 Setup 5............................................................................................................... 63 4.13 Setup 6............................................................................................................... 63 4.10 Chapter Conclusion ........................................................................................... 67 Chapter 5 – RESULTS AND DISCUSSIONS ................................................................ 68 5.1 Setup 1................................................................................................................. 68 5.2 Setup 2................................................................................................................. 68 5.3 Setup 3................................................................................................................. 72 5.4 Setup 4................................................................................................................. 75 5.5 Setup 5................................................................................................................. 78 5.6 Setup 6................................................................................................................. 79 Chapter 6 – CONCLUSION AND FUTURE WORK ...................................................... 84 REFERENCES .............................................................................................................. 87 BIBLIOGRAPHY ........................................................................................................... 89 APPENDIX 1 ................................................................................................................. 90
viii
ix
LIST OF FIGURES Figure 1: Semiconductor TEM. Left: Graphite on ceramic. Right: ceramic only .............. 2 Figure 2: TEG and TEC setups ....................................................................................... 4 Figure 3: Charge carrier movement ................................................................................ 4 Figure 4: Thermoelectric module cross-section ............................................................... 5 Figure 5: Carnot cycle ..................................................................................................... 6 Figure 6: Segmented thermoelectric module................................................................... 7 Figure 7: Figure of merit and compatibility plots [1],[4] .................................................... 8 Figure 8: Cascaded thermoelectric module ..................................................................... 9 Figure 9: Thermoelectric cooler assembly cross section [20] ........................................ 33 Figure 10: Insulated thermoelectric cooler assembly cross section [20]........................ 34 Figure 11: Direct and indirect SWH setup (Source: http://www.renuholdings.co.za) ..... 35 Figure 12 Direct vs Indirect SWH .................................................................................. 36 Figure 13: Generator hot water storage tank................................................................. 36 Figure 14: Setup 1 ......................................................................................................... 37 Figure 15: Cross section of setup 1 mounting system ................................................... 38 Figure 16: Data logger module - iCP12-V1.0 ................................................................ 40 Figure 17: Data logger PCB layout ................................................................................ 41 Figure 18: Data logger photograph ............................................................................... 41 Figure 19: Live data view – SmartDAQ v1.3 screenshot ............................................... 42 Figure 20: Cavity for temperature sensing by thermocouple ......................................... 43 Figure 21: Front end crcuitry that uses AN8497 for temperature sensing ..................... 44 Figure 22: Cold and hot side temperature measurment front end circuitry .................... 46 Figure 23: Voltage sensing amplifier ............................................................................. 49 Figure 24: Current sensing amplifier ............................................................................. 50 Figure 25: Boiler cross section ...................................................................................... 51 Figure 26: boiler support system ................................................................................... 52 Figure 27: Thermocouple .............................................................................................. 53 Figure 28: Top part of container .................................................................................... 54 Figure 29: Temperature controller and solid state relay ................................................ 55 Figure 30: Setup 2 ......................................................................................................... 56 Figure 31: Setup 3 ......................................................................................................... 61 Figure 32: Setup 4 ......................................................................................................... 62 x
Figure 33: Heatechanger of 4 modules ......................................................................... 64 Figure 34: Solar heat exchanger ................................................................................... 65 Figure 35: Solar water heater installation ...................................................................... 66
xi
LIST OF TABLES Table 1: Seeebeck coefficient various of materials. Source [2] ....................................... 3 Table 2: Thermoelectric generator comparison ............................................................. 30 Table 3: Thermoelectric comparison ............................................................................. 31 Table 4: Setup 2 parameters ......................................................................................... 58 Table 5: Setup 3 parameters ......................................................................................... 61 Table 6: Setup 4 parameters ......................................................................................... 63 Table 7: Setup 5 parameters ......................................................................................... 67 Table 8: Setup 2 results ................................................................................................ 70 Table 9: Comparison to other researches ..................................................................... 71 Table 10: Updated comparison (1) ................................................................................ 71 Table 11: Setup 3 results .............................................................................................. 73 Table 12: Updated Comparison (2) ............................................................................... 75 Table 13: Setup 4 results .............................................................................................. 75 Table 14: Setup 5 results .............................................................................................. 78 Table 15: Setup 6 results (per module) ......................................................................... 80
xii
LIST OF GRAPHS Graph 1: MPPT comparision ......................................................................................... 18 Graph 2: Thermistor resistance vs temperature ............................................................ 47 Graph 3: Thermistor voltage vs temperature ................................................................. 48 Graph 4: TEG2-07025HTSS internal reistance vs temperature difference.................... 59 Graph 5: Load current vs hot side temperature (TEC1) ................................................ 69 Graph 6: Output power vs temperature difference of setup 1 up to 170°C T hot ............. 70 Graph 7: Output power vs temperature difference of setup 1 up to 110°C T hot ............. 72 Graph 8: Temperature comparison ............................................................................... 72 Graph 9: Setup 3 performance ...................................................................................... 74 Graph 10: Setup 4 performance .................................................................................... 76 Graph 11: Comparison of setups 2, 3 and 4 TEC ......................................................... 77 Graph 12: Comparison of setups 2, 3 and 4 TEG ......................................................... 77 Graph 13: Natural to forced cooling comparison ........................................................... 78 Graph 14: Temperature difference vs time on September 1st ....................................... 79 Graph 15: Power vs load resistance.............................................................................. 80 Graph 16: Typical power vs temperature difference plot for a single module ................ 81 Graph 17: Comparison of solar generator, forced, natural cooling ................................ 82 Graph 18: Voltage degradation after 135 days .............................................................. 82 Graph 19: Current degradation after 135 days .............................................................. 83
xiii
GLOSSARY OF SYMBOLS A
Area, m2
An
Area of cross section of n-type semiconductor, m2
Ap
Area of cross section of p-type semiconductor, m2
C
torque coefficient
D
diameter
g
Length to cross sectional area ratio
I
Current, A
kl
Thermal conductivity of interconnecting metal, W/°C
kn
Thermal conductivity of n-type semiconductor, W/°C
kp
Thermal conductivity of p-type semiconductor, W/°C
Ln
Length of n-type semiconductor, m
Lp
Length of p-type semiconductor, m
m
Resistance load ratio
mopt
Resistance load ratio optimum
n
Number of thermocouples
N
Number of bolts
P
Power, W
P
Pressure, Nm
Qc
Energy released in cold sink, J
Qh
Energy withdrawn from hot sink, J
R
Resistance, Ω
Rl
Resistance of interconnecting metal, Ω
RL
Resistance of load, Ω
T
Torque, Nm
Tc
Temperature of cold side, °C
Th
Temperature of hot side, °C
V
Voltage, V
W
Electrical energy through load, J
Z
Figure of merit
αn
Seebeck coefficient of n-type semiconductor, V/°C
αp
Seebeck coefficient of p-type semiconductor, V/°C
xiv
ΔT
Temperature difference, °C
η
Efficiency
ρn
Resistivity of n-type semiconductor, Ωm
ρp
Resistivity of p-type semiconductor, Ωm
xv
LIST OF ABBREVIATIONS COP
Coefficient of performance
DAQ
Data acquisition
EGR
Exhaust gas recirculation
EMF
Electromotive force
ESC
Extremum Seeking Control
FOM
Figure of merit
IC
Integrated circuit
MPP
Maximum power point
MPPT
Maximum power point tracking
OPAMP Operational amplifier P&O
Perturb and Observe
PCB
Printed circuit board
SWH
Solar water heater
TEC
thermoelectric cooler
TEG
Thermoelectric generator
TEM
Thermoelectric module
USB
Universal serial bus
xvi
Chapter 1 – INTRODUCTION
Chapter 1 –INTRODUCTION 1.1 General information Energy efficiency, energy recovery and renewable energy resources have been researched for quite some time now. These three are the principles of green energy. Green energy is possible with technology based on both modern and old discoveries. The Seebeck effect, discovered by Thomson Seebeck in 1821, is one of the old discoveries that has not yet penetrated the green energy market. The Seebeck effect occurs when two dissimilar materials are joined at one end and heat flows from the joined ends to the loose ends. An electromotive force (EMF) is generated and current flows when the loose ends are connected to an electric load. A number of these dissimilar materials, called thermocouples, are connected in series to increase the generated EMF. This setup is known a thermoelectric generator (TEG). The reverse occurs when an EMF is applied to the loose ends. Current flows through the dissimilar materials and heat will flow from the loose ends to the joined ends. This is known as the Peltier effect which was discovered in 1834 by Jean Charles Athanase Peltier. A number of these dissimilar materials are connected electrically in series and thermally in parallel to form a thermoelectric cooler (TEC). Thermoelectric modules (TEMs) are commercially used for accurate temperature measurement, temperature control of sensitive circuits and for other cooling purposes. TEGs are not yet commercially used due to their low efficiency and cost when compared to other proven technologies for energy recovery. However TEGs are used for specific applications, mainly in military and aerospace applications due to their reliability, robustness, silent operation, long life and solid state technology.
1.2 Objectives of project This thesis project aims to quantify and analyze the energy that can be recovered from waste heat in a solar water heater (SWH). An evacuated tube solar water heater can reach high temperatures, up to 80°C, in the coldest months of the year. This means that during the warmer months and periods of low system utilization, there is a lot of excess heat that can be recovered using TEMs. Such recovery has the added benefit of 1
Chapter 1 – INTRODUCTION
extracting waste energy, instead of shading solar water heater to avoid overheating during the warmer months. In this write-up, TEMs can refer to both TEGs and TECs. Figure 1 shows how a typical semiconductor TEM looks like.
Figure 1: Semiconductor TEM. Left: Graphite on ceramic [21]. Right: ceramic only
2
Chapter 2 – LITERATURE REVIEW
Chapter 2 – LITERATURE REVIEW In this chapter the thermoelectric process is explained. Different types of thermoelectric modules are also presented. A review of a number of projects was done to learn from other studies and build more knowledge on the subject. The review studies include: waste heat recovery from motor vehicles, optimization of heat exchanger design, analysis of maximum power point tracking, performance analysis of series connected thermoelectric modules and a low temperature application using hot water as the energy source.
2.1 Thermoelectric generators Initially thermoelectric electric devices were made of metals but nowadays semiconductor is being used since it has a higher Seebeck coefficient which results in higher energy density. Table 1 is a list of Seebeck coefficients of the most common materials. It can be observed that Bismuth (Bi) and Tellurium (Te) can make the best thermocouple since they have the highest Seebeck coefficients. The Bi 2Te3 thermocouple has the best figure of merit (FOM) for low temperature applications [1].
Material
Seebeck Coefficient (µV/°C)
Material
Seebeck Coefficient (µV/°C)
Material
Seebeck Coefficient (µV/°C)
Aluminum Antimony Bismuth Cadmium Carbon Constantan Copper Germanium
3.5 47.0 -72.0 7.5 3.0 -3.5 6.5 300.0
Gold Iron Lead Mercury Nichrome Nickel Platinum Potassium
6.5 19.0 4.0 0.6 25.0 -15.0 0.0 -9.0
Rhodium Selenium Silicon Silver Sodium Tantalum Tellurium Tungsten
6.0 900.0 440.0 6.5 -2.0 4.5 500.0 7.5
Table 1: Seeebeck coefficient of various materials. Source [2]
Figure 2 shows a simplified cross-section diagram of the basic setup to operate a TEM as a generator or temperature controller. The setups are very similar since one operation is the reverse of the other. A further explanation of the Seebeck effect is given in the next section.
3
Chapter 2 – LITERATURE REVIEW
HEAT OUTPUT
COLD SIDE
HOT SIDE
HEAT OUTPUT
HEAT INPUT
HOT SIDE
COLD SIDE
HEAT INPUT
GENERATOR
HEAT FLOW
COOLER
Figure 2: TEG and TEC setups
2.2 The Seebeck effect Thermoelectric semiconductor materials have free holes and electrons that can carry both charge and heat. A p-type semiconductor has free holes that flow from a point of high thermal energy to one of low thermal energy when a temperature gradient is applied across the material. The movement of charge carriers to the cold end of the material results in a charge density gradient across the material. This charge density will repel back charge carriers to the hot end. At steady state, the temperature gradient will balance the charge density gradient. The resultant electric potential across the material is called the Seebeck voltage. The generated voltage is proportional to the temperature difference by the Seebeck coefficient [1]. i
Hot
+
-
-
n-type
p-type
e-
h+
e- e-
h+ h+
Cold
+
Figure 3: Charge carrier movement
4
Chapter 2 – LITERATURE REVIEW
The same effect occurs in an n-type semiconductor but the charge and heat carriers are electrons instead of holes. The cold end will be negatively charged with respect to the hot end in n-type semiconductors and vice versa for p-type semiconductor as shown in Figure 3. If the hot ends are electrically linked and the cold ends are connected together through a load, current flows through the load. Increasing the temperature difference will increase the output voltage and increasing the flow of heat from the hot to the cold end will increase the current flow. The output voltage may be increased by connecting more n-p type pairs in series to form a thermoelectric module as shown in Figure 4. Hot Side
N
P
N
P
Cold Side
Load
Electrical insulator Electrical conductor
Figure 4: Thermoelectric module cross-section
The thermoelectric module can be considered as a solid state heat engine. Like any other heat engine the maximum operating efficiency is given by the Carnot cycle. (1)
Where
is the cold sink temperature,
energy released in the cold sink,
is the hot source temperature,
is the
is the energy taken from the hot source and,
5
is
Chapter 2 – LITERATURE REVIEW
the electrical energy through the load. The actual efficiency of the thermoelectric generator depends on the thermal and electrical conductivity of the semiconductor, the Seebeck coefficient and temperature difference, as will be explained in Chapter 3. Both the thermal and electrical conductivity are temperature dependent, so to maximize efficiency, an optimal semiconductor material must have variable conductivities along the temperature gradient to which it is exposed. This is achieved by segmented generators.
Hot Source
Qh
W
QC
Cold Sink
Figure 5: The Carnot cycle
2.3 Segmented generators Segmented generators are made up of multiple layers of different materials instead of a single core. Each layer is designed to operate at its own maximum efficiency along the temperature gradient across the generator. A thermoelectric compatibility factor is used to select the best matching materials in terms of thermal and electric conductivity for optimum efficiency at the desired hot side temperature. Segmented thermoelectric generators can be exposed to high temperatures such that higher efficiency is obtained. Figure 6 shows a cross section of a segmented thermoelectric generator that was tested by Caillat et al. [3]. The theoretical efficiency of this particular segmented
6
Chapter 2 – LITERATURE REVIEW
thermoelectric generator is 15.5% while that of the non-segmented structure is 12.5% at a hot side temperature of 600°C.
Hot Side 702°C
N
P
402°C 202°C 27°C
Cold Side
Load Electrical insulator
Bi0.4Sb1.6Te3.6
Electrical conductor CeFe4Sb12
CoSb3 Bi2Te2.95Se0.005
Zn4Sb3
Figure 6: Segmented thermoelectric module
The selection of the n-type materials for the thermoelectric generator of figure 6 was done by selecting the materials that have matched compatibility factors at key temperatures. The best option is to have two segments with a changeover at 200°C. Bi2Te2.95Se0.005 is placed on the colder side since its best conversion efficiency occurs at low temperature. The changeover thickness is obtained from the temperature gradient estimation. Observing the plots of Figure 7, one can notice that the p-type semiconductor can be segmented in three stages for best efficiency. The changeover points are at 200°C and 400°C as shown in Figure 7.
7
Chapter 2 – LITERATURE REVIEW
Figure 7: Figure of merit and compatibility plots [1],[4]
2.4 Cascaded Thermoelectric generators Cascaded thermoelectric generators have high and low temperature regions like segmented thermoelectric generators. These thermoelectric generators are constructed by
unicouple
stages
of
different
operating
temperatures,
unlike
segmented
thermoelectric generators that are made of a single unicouple operating across the full temperature range. Figure 8 shows a cross-section of a cascaded thermoelectric module. The thermal conductivity of the cascade materials needs to be matched; otherwise heat flow will be limited by the weakest one along the cascade. Similarly, current flows though all unicouples in series, so the electrical conductivity needs to be matched too. However matching cascaded thermoelectric generators is easier since this may be adjusted by the number of unicouples connected electrically in series and thermally in parallel. Like segmented thermoelectric generators, cascades can be operated at high temperature with higher efficiency [4].
8
Chapter 2 – LITERATURE REVIEW
Hot Side N
N
P
P
N
P
N
P
Cold Side
Load Electrical insulator
P-type 2 N-type 1
Electrical conductor P-type 1
N-type 2
Figure 8: Cascaded thermoelectric module
2.5 Thermoelectric modules In the energy sector, renewable resources exploitation has been the major research area for the past few years. Amongst the main topics one can find wind and solar energy, fuel cells, bio fuels and so on. These sources are intensively studied since they offer a high energy yield due to the high efficiency of the respective converters and the availability of the primary energy source in large quantities. Unlike these, energy recovery from waste heat has not been studied extensively due to the application in specific scenarios. There are different methods to recover energy from waste heat depending on the type of energy output required, the waste heat temperature and available capacity. For instance, when high temperature heat is available, one can opt to power a gas turbine, but this will only be feasible if the heat available is constant and in adequate quantity. When the available energy is at a lower temperature one may consider the use of the Organic Rankine cycle, or thermoelectric energy conversion. Again, the choice depends on the waste energy profile, investment cost and useful energy output. Many experimental studies on thermoelectric energy conversion were 9
Chapter 2 – LITERATURE REVIEW
based on the higher temperature range of thermoelectric modules due to the higher efficiency in this range. A thermoelectric module can convert heat at 200°C directly to electricity at an approximate efficiency of 5% [5]. This figure may seem poor but there is a good potential for research when considering that this output is converted from zero cost waste heat using a relatively simple setup. The present semiconductor thermoelectric modules can be compared to the first photovoltaic modules that appeared on the market in the 1980 with a conversion efficiency of 5%. Continuous research by semiconductor manufacturers has brought the efficiency of photovoltaic modules up to 22%. Thermoelectric conversion was discovered in the 1821 and its theory has been studied and written since the beginning of the 1900s but more intensive research has only picked up momentum in the last three decades with the mass production of the semiconductor. There are publications by Universities, particularly from Asian countries and by thermoelectric module producers such as Komatsu and TECTEG. Other companies are experimenting with the implementation of thermoelectric generation from waste heat to enhance the efficiency of their products. Car manufacturer BMW has been running tests to recover energy from engine exhaust using high temperature thermoelectric modules since 2008 [6]. Exhaust gases passing through the exhaust pipe and the exhaust gas recirculation system (EGR) are used to power a thermoelectric generator. The target power output of the production models is 1000W but the present prototype at the time of publication is capable of generating 600W from the exhaust pipe emissions and 250W from the EGR. These values boost the vehicle efficiency by a total of 7% which makes it only 3% short of the 10% target. The improvement in efficiency is attributed to less engine load required by the alternator to power the electrical systems. The improvement in system efficiency depends on two main aspects [7]. One is the enhancement of the conversion efficiency of the thermoelectric device itself and the other is to improve system implementation by reducing losses.
10
Chapter 2 – LITERATURE REVIEW
2.5.2 Thermoelectric systems As already mentioned, thermoelectric generation for automobile heat recovery is a commonly researched application since the conditions of operation are favorable for the thermoelectric generator. Also this technology is not intrusive on the vehicle’s weight and space constraints. With a maximum exhaust temperature of about 700°C to 800°C for gasoline engines and about 400°C to 500°C for a diesel engine, there is enough potential to extract some of the 40% loss of input fuel energy [8]. Studies with lower operating temperature have also been published. Wang et al. [9] carried out a theoretical analysis of thermoelectric generation from vehicle exhaust. It was found that the output power and efficiency increase asymptotically with cooling medium flow rate. When the cooling medium was actively powered by either fan or pumping, there was a cooling flow rate which gives the optimal net power output. This optimal condition depends on the heat flow rate across thermoelectric modules, since this affects the cooling rate required at the cold side. Furthermore, the increase in heat flow rate also requires an increase in generator surface area such that there is enough time for heat to be absorbed. The exhaust flow rate is continuously changing, depending on the driving conditions, so there is an optimal generator size that has the widest power output bandwidth. Heat absorption by the thermoelectric modules depends on the convection heat transfer coefficient the hot side of the module. Increasing
at
will increase the hot side temperature and
thus increase the temperature gradient across the thermometric module. A higher temperature gradient will enable more heat flow across the module, so output power increases. The temperature gradient across the device can also be increased by increasing the cold side convection heat transfer coefficient
. Changes in
and
create an equivalent change in power output, but beyond a certain point, an increase in
generates a higher increase in output power and efficiency compared to
an equivalent change in
. This is attributed to the increase in the Seebeck
coefficient as the hot side temperature increases. The increase in the hot side temperate is limited by the temperature of the waste heat available and the upper limit of operation of the device. However, in many applications it is much cheaper and
11
Chapter 2 – LITERATURE REVIEW
simpler to decrease the cold side temperature than to increase the hot side temperature by, for instance, increasing the heat sink size or changing the cooling medium. According to this theoretical study, the output power improves by 72% if the system is cooled by the engine cooling water at 80°C rather than by air. The authors also explain that there is a limit to the increase of
without reducing the engine efficiency due to
the increase in exhaust gas back pressure. One major improvement that is suggested is to use a closed loop high temperature phase change medium that is heated by a heat exchanger in the exhaust system instead of the exhaust heating the thermoelectric modules directly. Such a setup is expected to increase the system power output by 10 times. The authors present results that show that thermoelectric device selection is important to optimize power output and efficiency at the available conditions. A parameter that needs attention when selecting a device is the semiconductor thickness since the conversion efficiency and power output vary independently with device height. Both depend on three main parameters; the Seebeck coefficient, thermal and electric conductivity. Each of these three parameters peaks at different semiconductor height, so there is an optimal height which gives the best combination of all. At this height the power peaks. On the other hand the efficiency is always increasing with increase of device height. Wang et al. [9] explain that electrical power sources have their own internal resistance which limits the output current that can be withdrawn for them. The power delivered to the load increases as current increases and peaks when the load resistance matches the internal resistance of the load. This point is known as the maximum power point (MPP). Beyond this point the power delivery drops again since the source will be overloaded and the voltage drops as current increases. Like convectional sources, thermoelectric generators have an internal resistance, but MPP occurs when the external resistance in approximately 1.5 times the internal resistance. This is the net result of 5 dependent parameters, the Seebeck coefficient, Seebeck voltage, current flow, temperature and the internal resistance of the device. Other publications, contradict this hypothesis as will be shown later in this section. 12
Chapter 2 – LITERATURE REVIEW
Tatarinov et al. [10] investigated the electrical, thermal and thermoelectric degradation properties of a thermoelectric module (with part number TG126-26-34) when applied to automotive applications. One of the analyzed parameters was the change in module internal resistance when operating at a temperature difference of 50°C to 225°C using 4 different types of heat transfer compound. The internal resistance of the modules increases as the temperature difference across it increases, irrespective of the type of heat transfer compound used. The resistance values when using the different compounds converge to 1.25Ω at a temperature difference of 225°C. A linear representation of results shows a change in resistance from 0.85Ω to 1.25Ω, which is equivalent to a change of 30%. The corresponding change in power output due to this 30% change in internal resistance was equal to 4% when a fixed resistive load was used. This result suggests that MPP tracking to optimize for change in internal resistance due to changes in operating temperature is not critical. However matching the load resistance ratio within the 30% range is still critical. This is proven by another set of results presented by the authors in the same study. As shown by other studies, there was a sharp increase in power output as the resistance ratio increased until peak power was reached, then the power decreased gradually as the ratio increased. The rate of decrease in power beyond the MPP, increased as the temperature difference increased. An output of 0.25W was obtained at a temperature difference of 50°C followed by an approximately linear increase in power from 2W at 100°C up to 10.25W at 250°C. These results, obtained from a single 0.25cm2 module, suggest that no significant power can be obtained when operating below at a temperature difference 50°C. Another property that was investigated by Tatarinov et al. [10] was the clamping pressure applied to the thermoelectric module. When the clamping pressure increases the power output increases with a similar profile for all 4 types of heat transfer compounds; LP10, LP3, graphite foil and CuSi grease. The highest power output was obtained using LP3 followed by LP10, Copper silicon and graphite foil. The power output difference between the best and worst compound was 4W which is equivalent to 40%. The approximate change in power output with contact pressure for graphite foil was 16% while that of LP3 was 5.6%. The results presented show that LP10 and LP3 13
Chapter 2 – LITERATURE REVIEW
have the highest and very similar heat conduction coefficients except in the temperature difference range of 25°C to 75°C. LP10 was 7.5% better at a temperature difference of 25°C. Despite this equivalent thermal property of LP10 and LP3, a thermoelectric setup using LP3 generates 5.6% more power than one using LP10 as previously mentioned. This suggests that there are other characteristics within these materials that contribute to their performance. The worst of the 4 tested compounds was graphite foil which was approximately 35% inferior to LP10 at a temperature difference of 25°C. Module degradation was investigated by Tatarinov et al. [10] by periodically changing the systems’ operating temperature from 50°C to 250°C instantly for 340 cycles within 60 hours. A drop of 11% in power output was observed while the efficiency remained constant. This suggests that the degradation lies in the heat transfer compound, which was only tested for LP3. So there is a limitation when using LP3 in motor vehicles due to continuously changing operating conditions. Kumar et al. [11] built a prototype thermoelectric generator and installed it on an 800cc petrol engine. This experiment is particularly interesting because it gives an insight into how much power can be generated from a small engine. Other experiments, such as those conducted by BMW, are based on a 3000cc engine which may not be the preferred engine size for the majority of the commuters. The authors ran a number of simulations using different software to come up with the best heat exchanger design for the thermoelectric generator. It was discovered that the square section heat exchanger is the best option since it gives a uniform heat distribution with a peak temperature within the capabilities of the thermoelectric modules. Internal fins were introduced to increase the heat transfer coefficient. However, this resulted in decreased engine performance due to exhaust back pressure. The heat exchanger was composed of three main components; an iron frame, cold plates and hot plates. The hot plates were made of aluminum since it is very light, yet it has good thermal conductivity. However, the cold plates were made of copper, since it has twice the thermal conductivity of that of aluminum. Similar to Wang et als.’ [9] conclusions, it is necessary to have the highest possible heat transfer coefficient at the cold side to increase the power output rather than on the hot side, since the latter is limited by the peak temperature that the modules 14
Chapter 2 – LITERATURE REVIEW
can operate at. The drawback of copper is that it is 3.5 times denser than aluminum. The authors opted to use a counter flow heat exchanger with engine coolant as the cooling medium. The generator setup with 18 thermoelectric modules had a total weight of 14.5kg. The results presented by the team show that the thermoelectric generator has an insignificant effect on engine performance, hydrocarbon and carbon monoxide emissions. At full load the thermoelectric generator only uses 2.1% of the water cooling system flow to generate 70W of electrical power. The exhaust temperature at the maximum engine power of 24kW was 700°C. The results presented confirm that the electrical power increases with the square of temperature as claimed by thermoelectric theory. The system efficiency at 10kW engine power was 1.5% and 3% at 25kW. The authors fail to specify the resistance ratio that they used and one cannot confirm if the system was operating at MPP. They suggested decreasing the cold side temperature to increase the system performance. Yu et al. [12] presented a numerical model to predict the performance of a parallel heat exchanger thermoelectric generator setup. The model consisted of a thermoelectric module sandwiched between a layer of hot water at the top and a layer of cold water at the bottom. Several assumptions were taken to simplify the model such as, a matched resistance ratio of 1. The analysis focused on any differences in performance between parallel and counter flow in the heat exchanger. Results show that the temperature difference along the counter flow setup was negligible, with an average of 92.5°C, while that in the parallel flow case varied linearly by approximately 10°C, with an average of 92.5°C too. Both models used water at an inlet temperature of 150°C at the hot side and cold water at 32°C on the cold side. The authors suggested that since the average temperature difference along both setups was equal, then the output from all the thermoelectric modules should be identical in both setups. However in his study, Hsu et al. [13] showed that the output of his system was limited by the coolest thermoelectric module in the setup unless an MPP tracker is connected to each individual thermoelectric module. In the model of Yu et al., the coolest device was located at the entry of cold water in the counter flow exchanger. Further results show that the power 15
Chapter 2 – LITERATURE REVIEW
output of the system was more dependent on the hot water temperature then on its flow rate. This result matches that presented by Wang et al. [9], even though the load resistance ratio used by the authors was different, as previously mentioned. Unfortunately this study does not analyze the effect on the power output by changes to the cold water inlet temperature in order to compare the results to those of other research. The theoretical study by Yu et al. [12] was followed by an experimental study a year later [14]. A test jig was built to analyze the performance of thermoelectric modules and verify the theoretical results. The jig consisted of fifty six thermoelectric modules powered by a five layer parallel plate heat exchanger. The top, middle and bottom layers were used for closed loop cold water flow and the two intermediate layers were used for closed loop hot water flow. Both loops were temperature controlled using a fan and a heating element. Temperature, pressure and flow measurements were taken at key points in the system. The thermoelectric modules used in this experiment had a cooling power of 42.8W but the authors did not specify their model number. No results that compare counter and parallel flow are presented. This contradicts the emphasis that they made in their earlier theoretical study, which focused on a comparison of flow types. However, an interesting summary of other operating conditions is presented. Results show that load resistance matching had a significant effect until the maximum power was reached due to a sharp increase in power until MPP. Power decreased gradually when the load resistance was increased beyond MPP. Furthermore, the rate of decay of power for an equivalent load increased as the temperature difference increased, i.e. load matching had more effect at high temperatures. A maximum power of 146.5W was reached at the maximum hot side temperature of 150°C and a cold side temperature of 30°C. This is approximately equal to 2.61W per module. The estimated overall conversion efficiency of the system at this condition was 4.44%. Both the efficiency and power increased asymptotically with increasing heat exchanger flow rate. There was no significant increase in power output beyond a critical flow rate of 0.3m3/h. A similar behavior was reported by Wang et al. [9], however the two differ on which side it is best to increase the flow rate for power optimization. The hot side fluid flow rate had greater effect on power output in the water cooled setup of [12] while the cold side flow 16
Chapter 2 – LITERATURE REVIEW
rate had more effect in the air cooled setup of [9]. The theoretical model presented earlier by the authors overestimated the power output of the system and they concluded that this was due to the assumptions taken to simplify the theoretical analysis. MPP may be achieved using a controlled direct current to direct current (DC-DC) converter instead of adjusting the load resistance. The converter continuously adjusts the terminal voltage of the thermoelectric module by adjusting current such that the maximum current-voltage product is maintained. As the current drawn from the modules increases, the voltage across its terminal decreases and vice versa. The DC-DC converter is also used to precisely control the output voltage of the system, that is, the voltage on the load side. For instance, if the system is used on a motor vehicle to charge its 12V battery, the charging voltage should be in the range of 14.3V to 14.5V. Phillip et al. [15] built a simulation model to analyze the performance of a thermoelectric generator when operated with fixed duty cycle and two different modes on MPP control, Perturb and Observe (P&O) and Extremum Seeking Control (ESC). Graph 1 shows the results as plotted from the results table presented by the authors. It can be seen that the maximum change between the ideal and fixed load output was 16% unlike the results presented by Tatarinov where the change was of only 4%. The major difference between the two research studies is that one was a simulation while the other was hardware based. Furthermore the simulation was conducted for a much higher temperature than the operating temperature of the test jig.
17
Chapter 2 – LITERATURE REVIEW
100 98
Power ratio (%)
96 94
Fixed P&0
92
ESC 90 88 86 84 325
375
425
475
525
575
625
T (°C)
Graph 1: MPPT comparision
A test jig to analyze the effect of MPP tracking on the performance of thermoelectric modules was built by Vadstrup et al. [16]. This study was based on TEG1-127-4.5-2.0250 by Guangdong Fuxin Electric. The jig had 4 thermoelectric modules that can be connected either in series or in parallel. The temperature difference across each module can be independently controlled. The aim of this research was to analyze the effect on the power output when series and parallel thermoelectric modules are subject to different operating conditions. Each module was connected to an individual MPP tracking buck boost converter controlled by a perturb and observe algorithm. The results show that the module internal resistance changed linearly from 0.15Ω to 0.24Ω with a change in temperature difference from 0°C to 150°C. This is equivalent to approximately 42%. In a similar setup using a different thermoelectric module, Tatarinov et al. [10] showed that for an equivalent temperature range, the change was of approximately 25%. An interesting result presented by the authors shows that the power output from the parallel connected setup was more susceptible to temperature difference between the modules than for series connected ones. In other words, parallel connected modules require MPP tracking more than series connected ones do if the 18
Chapter 2 – LITERATURE REVIEW
modules are to operate at different temperatures. The decrease in output efficiency of four modules connected in series and operating at 60°C module-to-module temperature difference was 8.4% while that for parallel connected modules under the same operating conditions was of 16.7%. A plot of output power against module-to-module temperature shows a significant increase in output power if the modules are operated on an individual bases. However this setup would be very expensive to build when compared to a single, series or parallel MPP tracker. Low temperature applications are less attractive than high temperature ones because of lower thermoelectric module efficiency. However, a study by Chen et al. [17] shows that if thermoelectric cooler modules are used in reverse operation instead of thermoelectric generator modules, equivalent power output can be extracted with 86% reduction in thermoelectric device cost [18]. This study was based on a test jig of four thermoelectric modules heated by an electric heater and cooled by four water heat sinks. The thermoelectric modules were operated in different power conditions by using a multiple channel electronic load. The results presented include the effect of cooling water flow pattern and flow rate, performance analysis of series connected modules and a comparison of thermoelectric coolers to generators. The team analyzed the effect of water cooling flow pattern by machining two different heat sinks, one had four changes in water flow direction and the other had six. The test results show that the hot side and cold side temperatures of both setups at steady state were nearly identical. It was noted that steady state occurs after ten minutes of running time. A number of tests were also carried out using different water flow rates. It can be noted that irrespective of the flow rate, the resultant temperature difference across the module was of 97.4°C when using a heating temperature of 150°C and a cooling water temperature of 30°C. The minimum flow rate to achieve such operating condition was 0.4 liter per minute. A flow rate of 1.6 liters per minute will only increase the temperature difference to 99.6°C. The thermoelectric module used in this study was the 40mm by 40mm TECCP-24-001 with a cooling capacity of 94.2W. The maximum allowable temperature difference across the module is 66°C while the maximum allowed temperature on the hot side is 19
Chapter 2 – LITERATURE REVIEW
150°C. The thermoelectric generator was the TG-12-8-01L with approximately the same footprint and a maximum allowable hot side temperature of 250°C. When the thermoelectric cooler was operated at a hot side temperature of 110°C, a power output of 0.74W was supplied to the load at MPP. A 25% increase in the hot side temperature to 150°C generated an approximate increase of 42% in output power to 1.75W. The stated power figures are an average of the four thermoelectric modules when operated individually. It was observed that the peak power and the matched load resistance of each thermoelectric module are different, with an approximate peak power difference of 18%. The thermoelectric generator performance lies within the range of the 4 thermoelectric modules however, it is not specified whether the quoted figures are an average of a number of generators or the results of only one particular generator. In other words, it is not clear whether the performance of the generators varies from module to module. As previously mentioned, thermoelectric modules are preferred over generator since at the time of this research the cooler modules cost US $7 while the generators cost US $49. The power output of the four series-connected modules was higher than the summation of individual power at a particular operating temperature. This was an unexpected result as the series current was expected to be limited by the poorest performing module. The authors explain that this phenomenon can be related to the increase in module series resistance for an equivalent series current. If the modules’ series resistance increases for a given series current, the Peltier effect, which was always ignored in the consulted literature of this literature review, is reduced and the temperature difference across the modules increases. So, the power output of series connected modules cannot be predicted with accuracy. A low temperature test was also carried out by Gou at el [19] using ten TEC103180T125 thermoelectric coolers. The modules were sandwiched between a heat sink and a liquid heat exchanger. Unlike other reviewed studies, this one is not water cooled, but either naturally or forced air cooled using a 6.6W fan. The maximum hot side temperature at which the system can be operated was limited by the heat exchanger to 80°C. At this temperature the open circuit voltage was approximately at 1.4V when the system was naturally cooled and at 10.5V when forced cooling was used. This significant difference was attributed to the temperature difference between the two sides 20
Chapter 2 – LITERATURE REVIEW
of the module. A 10°C increase in hot side temperature increased the temperature difference across the module by 1°C when naturally cooled and by 3°C when using forced cooling. In these conditions the open circuit voltage increases at a rate of 0.3V/K and 0.7V/K respectively. The thermoelectric modules were connected to a variable load resistor to operate at MPP. The estimated internal resistance of the ten modules is 30Ω, which makes it the highest electrical resistance per module encountered during this literature review. The setup allowed temperature control of the hot side while the cold side can only be operated with or without forced cooling. So, the experimental analysis was based on the effect of the hot side temperature and on the difference between natural and forced cooling. The maximum power delivered to the load occurred at the highest hot side temperature for both types of cooling. A power output of approximately 15mW and 0.85W were delivered using natural and forced cooling respectively. These figures are surprisingly low when compared to the water cooled figures. However this is also dependent on the type of thermoelectric modules used. A theoretical analysis carried out by the authors investigated the effect of the heat transfer coefficient on the system performance. It was discovered that the cold side heat transfer coefficient has a major effect on the electrical power output. As expected, the graphical envelope representing power output against the cold side heat transfer coefficient and heat sink surface area are similar and quasi-exponential due to the dependency on each other. An air cooled thermoelectric generator was also built by Hsu et al. [13] to recover waste energy from the exhaust of a 2000cc petrol engine. Two models were built for two different studies; a 1st generation one with eight thermoelectric modules and a 2nd generation one with twenty four modules a year later. The generators were composed of a square section container with funnel ends for the inlet and outlet of the exhaust gases. The funnel ends were intended to reduce the sudden expansion in exhaust flow which would otherwise lead to uneven heat distribution of the square container. Thermodynamic simulation of the heat exchanger showed that the funneled ends were not sufficient to distribute heat evenly, so diffusers were introduced at the exhaust entry. 21
Chapter 2 – LITERATURE REVIEW
The square section container was made of two large aluminum heat sinks with finned sides facing each other and the flat surfaces being the outer sides of the container. So the exhaust passes through aluminum channels inside the heat exchanges to improve heat transfer to the outside in a uniform pattern. The thermoelectric modules, TMH400302055 by Wise Life Technology Taiwan, are installed on both the top and bottom of the container. The space around the thermoelectric modules was covered with Bakelite to reduce heat loss from the container and useless heating of the heat sink. Similar to the setup by Kumar et al. [11], the modules were cooled by copper heat sinks, to increase the cold side heat transfer coefficient. The upper temperature limit of this test jig is at 300°C due to the thermoelectric module specification. Initially the authors carried out tests on a single thermoelectric module to analyze its characteristics, such as internal resistance and power output. The single module test jig had no operational limitations expect for the module itself. The setup used water cooling and heating though a copper heat exchanger to maximize control over the temperature difference across the module. The open circuit voltage and power at various temperature differences were recorded. According to circuit theory, at MPP, for each temperature difference, the output voltage is half the open circuit voltage at the same temperature conditions. So the power capabilities of the module at difference temperature conditions were estimated by changing the load resistance until the output voltage was half the corresponding open circuit voltage. The results presented show that the internal resistance varied between 1Ω to 1.25Ω with a corresponding variation of 5°C to 30°C in temperature difference. This corresponds to a change of 23% in resistance with the maximum hot side temperature being 130°C. A similar test by Tatarinov et al showed a 30% change in internal resistance when the module was operated at a temperature difference of 0°C to 220°C with a maximum hot side temperature of 250°C. A power output of approximately 0.45W was delivered from the single module at a temperature difference of 25°C, with the hot side at 110°C. At the same hot side temperature, the setup by Chen et al. [17] using water cooling, had a temperature difference of 63.7°C, and delivered 0.74W to the load. Despite this higher power output, 22
Chapter 2 – LITERATURE REVIEW
both setups have an approximately equal power to temperature difference ratio of 0.0117°W/C at this particular hot side temperature. This means that if the setup by Chen et al. [17] uses the thermoelectric modules that Hsu et al. [18] used, the output power will be significantly higher due to the higher operating temperature difference. A maximum power output of 44.13W was recorded when the 1st generation model with 8 thermoelectric modules was operated at an average temperature difference of 88.3°C. The 2nd generation model with 24 thermoelectric modules operated at a much lower average temperature difference of 38.6°C. In this condition the power generation was 12.41W, which is equivalent to a reduction of 70% with an increase of three times in system sizes. Both setups were subject to the same engine exhaust conditions, so the authors suggest that the difference is due to the thermodynamic design of the heat exchanger. All 24 modules are connected electrically in series, so similar to what other researchers concluded, the power was limited by the coolest modules in the system. A comparison of results of this research to that by Kumar et al. [11] shows, that even though this setup was air cooled, it has higher power generation at 5.5W per thermoelectric module against 3.9W per module. However it is not clear whether the fan power consumption is reduced for the generated power of 5.5W.
2.6 Chapter conclusion In this chapter the principal of a thermoelectric module was explained together with the different types of modules. A number of publication on the subject were reviewed to build further knowledge on the subject. The most important points that were mentioned in the literature review include; the importance of uniform heat and temperature distribution over a single module and multiple module systems:
The advantage of using maximum power point tracking over fixed load systems or alternatively load matching within a window of 0% to +30% of matched load
The load voltage is half the open circuit voltage at maximum power point
Maximum power point is dependent on hot side temperature and the temperate difference across the module
23
Chapter 2 – LITERATURE REVIEW
The effect of improving the cold side and hot side heat transfer coefficient independently
Analysis of different heat transfer compounds
The effect of adding a spacer block in a system and other equally important suggestions.
This knowledge was used to build a number of setups for thermoelectric module testing and ultimately a small scale solar installation, as will be explained later in Chapter 4.
24
Chapter 3 – THEORY
Chapter 3 – THEORY This chapter presents the basic parameters that affect the generating capacity of a thermoelectric module. These include the electrical and thermal resistance, the highest efficiency and power that can be drawn from a module and the heat flow across it. The data sources are [4], [2] and bibliography.
3.1 Thermoelectric generator parameters The open circuit voltage of a thermoelectric generator is dependent on the Seebeck coefficient, given in µV/°C, and on the temperature difference across the hot and cold junction. However the Seebeck coefficient itself is temperature dependent so, accurate calculation of the open circuit voltage has to be done using: (2)
Otherwise the open circuit voltage can be estimated by neglecting the Seebeck coefficients change with temperature, then equation 2, can be simplified to: (3) (4)
where
is the number thermocouples making the thermoelectric generator,
Seebeck coefficient of the n-type semiconductor, type semiconductor,
is the
is the Seebeck coefficient of the p-
is the hot-side temperature and
is the cold side temperature.
The electrical resistance of the thermoelectric generator is dependent on the series resistance of the n-type, p-type and metal interconnects as shown in Figure 4. Again, the resistivity of these materials is temperature dependent, but estimation can be made using: (5)
25
Chapter 3 – THEORY
where
is the resistivity of the n-type semiconductor,
semiconductor,
is the length of the semiconductor,
the n-type semiconductor, and
is the resistivity of the p-type is the cross-sectional area of
is the cross-sectional area of the p-type semiconductor
is the resistance of the interconnecting metal. The electrical resistance of the
generator is an important parameter to match a load either for maximum power point or for maximum efficiency. The thermal conductance of a thermoelectric generator is the summation of three paths of heat flow through it as shown by equation 6. The first term represents the conductance through the n-type semiconductor, the middle term represents that in the p-type semiconductor and the last represents the thermal loss by conduction, radiation and other losses. (6)
where
is the thermal conductivity of n-type material,
is the thermal conductivity of
p-type material. The thermal conductivity of the thermoelectric generator is required to estimate the heat flow from the hot source to the cold sink such that accurate sizing can be made. The heat input at the hot junction is given by the sum of the heat by conduction heat, the Joule heat and the Peltier heat. It is assumed that the Joule loss will be split in two, half will be lost through the cold sink and the other half through the hot sink, and thus only half of the Joule loss is being absorbed by the hot side of the generator. (7)
where
is the thermal conductivity of the generator (equation 6),
difference across the hot and cold junction,
is the temperature
is the current through the thermocouple,
is the electrical resistance of the thermocouple and
is the resultant Peltier
coefficient of the thermocouple. The Peltier coefficient and the Seebeck coefficient are proportional to each other by the hot side temperature , then; 26
Chapter 3 – THEORY
(8)
Then the heat absorbed at the hot junction can be rewritten as follows: (9)
The electrical power output delivered to the load by the thermoelectric generator is: (10)
or the Joule loss through the load (11)
The efficiency of the thermoelectric generator is given by the ratio of heat input
to
electrical output (12)
The current
is dependent on the internal resistance of the thermoelectric generator
and the load resistance as shown by: (13)
Let
be the ratio of load resistance to thermoelectric generator internal resistance, so
that: (14)
Then, the efficiency becomes: (15)
To maximize efficiency, the
product must be minimized. Recalling equation 5 and 6,
R and K are dependent on the length to area ratio of the semiconductor material. Decreasing this ratio will decrease K but will also increase R and vice versa. The optimum is found by differentiating RK with respect to the length to area ratio, designated by g as follows: 27
Chapter 3 – THEORY
(16)
(17)
(18)
The result of equation 18 is an optimal ratio equal to:
(19)
or (20)
The efficiency using the optimum
product becomes: (21)
Where:
(22)
The term Z is known as the figure of merit. Increasing Z, will increase the efficiency of the device as can be deduced from equation 21. The optimum load resistance the resistance ratio,
is found by differentiating equation 21 with respect to
. The result is shown the following equation:
(23)
Using the optimum resistance ratio the maximum efficiency becomes:
28
Chapter 3 – THEORY
(24)
The power output from the device when operating at the maximum efficiency is given by equation 25 (25)
Ignoring the maximum efficiency, the maximum power that can be extracted from the device occurs when the load resistance is matched with the internal resistance of the device. So equation 10 can be rearranged to; (26)
The device efficiency in this condition is given by equation 27
(27)
29
Chapter 4 – CONSTRUCTION, DEVELOPMENT AND TESTING
Chapter 4 – CONSTRUCTION, DEVELOPMENT AND TESTING The aim of this thesis is to quantify the electrical power output from a thermoelectric module powered by a solar water heater. The output power has to be compared to some reference to ensure that the full potential of the thermoelectric module is exploited. This can be done comparing the results with those of other researches and with the modules’ datasheet. However only a few research studies at low temperatures could be found and the datasheet of the thermoelectric cooler does not specify the generating capacity. Only the thermoelectric generator module datasheet provides a trend of output power against temperature difference. To solve this issue a set of bench tests were conducted to analyze the characteristics of the thermoelectric cooler module, thermoelectric generator module, the heat source and the heat sink. These tests were done in 6 different setups which evolved as data from the experiments was analyzed. Each setup, from 1 to 6, is explained in this chapter. Each test was carried out with both thermoelectric cooler and generator with multiple test runs. Repeated readings using equivalent thermoelectric coolers from the different suppliers were carried out, however this was not possible with the generator module as only one company that produces low temperature generators was found.
4.1 Thermoelectric Generator The majority of the thermoelectric generators available on the market are designed to operate at high temperatures. Table 2 is a short list of 4 thermoelectric generators. Brand
Komatsu
Part No.
TEC
TEC
MARLOW
TEG2-07025HT
TEG1-12611-6.0
TG12-8
ΔTmax (°C)
250
180
270
153
Thot max (°C)
280
190
300
230
Powermax (W)
24
6.8
14.6
7.95
N/A
0.59
1.2
3.46
Area (mm ) Price (€)
2500
1600
3136
1600
215
30
80
36.6
Power / Price ratio
0.11
0.23
0.18
0.22
Power / Area ratio
0.11
0.23
0.18
0.22
Resistance (Ω) 2
Table 2: Thermoelectric generator comparison
30
Chapter 4 – CONSTRUCTION, DEVELOPMENT AND TESTING
One can notice that the TEG2-07025HT-SS, produced by TECTEG, has the highest power to area and power to cost ratio. Furthermore this module has the highest power output at low temperatures, that is, from room temperature up to water boiling point. Two of these 40mm X 40mm modules were purchased to carry out the experiments
4.2 The Thermoelectric Cooler The TEC1-12710 is a generic thermoelectric cooler produced by various manufacturers. Its main use is to cool Central Processing Units (CPUs), as a small cooler and so on. The maximum temperature difference that this device can produce is 75°C and has a maximum power consumption of 85W. It can operate at a maximum hot side temperature of 138°C. A number of these modules were bought for less than 3€ a piece in retail, so the price at wholesale level is expected to be much lower. There are modules with higher power dissipation or cooling capacity but their size will not match that of the thermoelectric generator module. For comparison reasons both the thermoelectric cooler and thermoelectric generator are 40mm X 40mm X 3.3mm. Tests with modules of lower cooling capacity, but of equal dimensions were not carried out because it is expected that their generating capacity will be less. Table 3 is a summary of five different modules. One can notice that the TEC1-12710 has the highest power to cost ratio but the least power to area ratio. A maximum power consumption of 85W does not mean that the generating capacity is equivalent. Brand Part No.
Laird Tech
Laird Tech
Generic
Ams Tech
Ams Tech
UT15
UT11
TEC1-12710
TB-199-2.0-0.9
TB-99-1.4-0.8
ΔTmax (°C)
68
67
66
69
69
Thot max (°C)
N/A
N/A
138
N/A
N/A
340.6
95.4
85
310
86
Powermax (W) Resistance (Ω)
1.97
1.2
1.08
0.86
0.8
Area (mm2) Price (€)
2704
1088
1600
3844
800
98
35
3
73
22
Power / Price ratio
3.48
2.73
28.33
4.25
3.91
Power / Area ratio
0.13
0.09
0.05
0.08
0.11
Table 3: Thermoelectric comparison
31
Chapter 4 – CONSTRUCTION, DEVELOPMENT AND TESTING
4.3 Thermoelectric Module Assembly Thermoelectric modules are assembled between two thermal surfaces. The power output from the module is dependent on the heat flow from one thermal surface to the other through the module. The thermal conduction losses through each surface, the module and contact between the module and thermal surfaces has to be minimized. However, thermal conduction between the thermal surfaces and through the air gap between them has to be maximized as this will result in a loss of heat from the hot surface and a corresponding effortless gain in the cold sink. Also, heat flow between the two surfaces through mounting bolts has to be minimized. Thermoelectric modules are brittle due to their construction of semiconductor matrix sandwiched between two ceramic plates. However they can take some degree of compression and high G-forces when assembled between rigid parallel surfaces. Laird Technologies, a leading thermoelectric module manufacturer, suggests that the mechanical tolerances of the hot and cold surfaces interfacing with a thermoelectric module should not exceed 0.025mm/mm with a summed error not exceeding 0.076mm [21]. At these tolerances the forces on the module during and after tightening will be equally distributed on its surfaces and thus there are no high pressure points which would otherwise damage it. When the modules are assembled in arrays, the height tolerance of the surfaces and the modules should not exceed 0.025mm. Thermal conductive grease or a graphite pad has to be used on both sides of the thermoelectric module to fill irregularities in the surface finish and thus improve the heat transfer to and from the module. Figure 9 shows a typical thermoelectric assembly used for cooling as suggested by Laird Technologies [20]. Heat is transferred from the cold plate to the heat sink, at some coefficient of performance (COP), by passing a current through the thermoelectric module. The mounting screws are placed on the centreline of the setup and parallel to the heat sink fins to prevent distortion on the surfaces. Three washers per screw are used; the top one is made of fibre to reduce heat flow between the surfaces, the middle is a flat steel washer to protect the fibre washer against the spring loading of the bottom washer. Spring loading is required to always keep the module under pressure otherwise 32
Chapter 4 – CONSTRUCTION, DEVELOPMENT AND TESTING
it may become loose if the thermal grease spreades out of the thermoelectric module border. A thermal grease coating of 0.025mm is enough if the mechanical tolerances and surface roughness are properly prepared. Thicker layers lead to reduced system efficiency due to increased thermal resistance and grease spread out.
Figure 9: Thermoelectric cooler assembly cross section [20]
The efficiency of the setup can be improved by reducing the heat flow between the hot and cold surfaces though the air gap as previously explained. This can be done by increasing the distance between the two surfaces and filling it with some thermal insulation as shown in figure 10. The insulation serves also as a barrier for foreign particles, such as dirt, to enter between the plates and reduce the efficiency even further. The boundary of the insulation is also coated with epoxy sealant to prevent moisture absorption by the insulation. This is essential since moist thermal insulation become a heat conductor. Ideally the module wire leads do not go through the insulation but are mounted on an electrically insulated standoff. Typically the system would only need a pair of these for every thermoelectric module array. A rubber seal has to be added below the fibre washer to complete the systems’ sealing. Laird technologies suggest that the spacer block should not be higher than 6mm if it is made of Aluminium and 4mm if it is made of copper. The spacer block is ideally placed at the heat sink side since this is the area with least heat flux.
33
Chapter 4 – CONSTRUCTION, DEVELOPMENT AND TESTING
Figure 10: Insulated thermoelectric cooler assembly cross section [20]
Laird technologies suggest a clamping pressure of 10 to 21 kilograms per square centimetre. Using the average value in the range, this is equivalent to 150,000 kg/m2 or a 1500,000 N/m2. The screw tightening torque to achieve this pressure is found by equation 28. (28)
where
is the torque coefficient (typically equal to 0.2 for dry surfaces),
diameter of the screws in meters, modules in m2 and
is the pressure in Nm,
is the
is the total surface area of
is the total number of screws.
4.4 System Design Considerations The system design is based on maximizing output power from waste solar water heater heat with the least cost without affecting the performance of the solar water heater itself. Solar water heaters are split in two main categories, direct and indirect types. Direct systems are sometimes referred to as open-loop type and indirect ones as closed loop type. Both types of solar heaters use a storage tank, usually horizontal and mounted at the highest part of the system. The tank is specifically positioned on top such that hot water from the top of the collector rises by convection to heat the storage tank. This water displacement forces cold water out of the bottom of the storage tank into the bottom of the collector. This water circulation stops when the temperature in the storage
34
Chapter 4 – CONSTRUCTION, DEVELOPMENT AND TESTING
tank reaches the temperature of the collector, and thus convection flow cannot occur. At this point stagnation is said to occur.
Figure 11: Direct and indirect SWH setup (Source: http://www.renuholdings.co.za)
In a direct type solar water heater, the water that will be used in the hot water faucets itself circulates between the collector and tank as its name suggests. On the contrary in indirect type heaters a heat transfer fluid is used. This fluid circulates through the collector and a heat exchanger inside the storage tank in a closed loop. The working fluid, typically a mixture of glycol and corrosion inhibiter, should never mix with the enduse water. Figure 11 shows the main components of the direct and indirect type solar water heaters. Some direct type solar water heaters use evacuated tube collectors that are individually, directly connected to the storage tank. Therefore there is no hot water flow out of the collector that can be tampered to power a thermoelectric generator. This major difference between different types of solar water heaters makes it a challenge to have one standard heat recovery system. Figure 12 (a), shows a flat plate solar water heater with an external closed loop cycle that can be tampered with. Figure 12 (b), shows an evacuated tube heater which does allow access to hot water unless forced circulation is done by a circulation pump.
35
Chapter 4 – CONSTRUCTION, DEVELOPMENT AND TESTING
(a) (b) Figure 12: Direct vs Indirect SWH (Source: Google Sktechup onine library)
The disadvantage of forced circulation is that some power output from the generator will be used to power the pump. One possible option to avoid pumping is to install the thermoelectric generator around the storage tank as shown by figure 13.
Figure 13: Generator hot water storage tank (source: own drawing)
The following is a list of advantages associated with installing the generators in contact with the storage tank as shown in figure 13:
The system design is very simple and thus it is expected to be cheaper to build
The hot water in the storage tank can be used as a small energy storage. This energy can be extracted without pumping.
36
Chapter 4 – CONSTRUCTION, DEVELOPMENT AND TESTING
The following is a list of disadvantages associated with installing the generators in contact with the storage tank as shown in figure 13:
The storage tank has to be specifically designed for such a setup as it is not possible to retro fit the system
Heat will be lost though the heat sink even when no power is drawn from the generator. So, the heat sink has to be insulated during the night to keep the stored water hot enough for domestic use.
Water stratification will create imbalance between the thermoelectric generators at different water levels.
The system will not start generating power until all storage water heats up.
A test jig, referred to as setup 1, was built to analyze the output from a single thermoelectric module mounted at the bottom of a hot water storage tank.
4.5 Setup 1 The setup consisted of an insulated water container with an aluminum bottom mounted on top of a heat sink, with a thermoelectric generator in between as shown in figure 14.
Figure 14: Setup 1
It was anticipated that the heat sink would be a limiting factor in maintaining the maximum temperature difference between the two surfaces of the thermoelectric 37
Chapter 4 – CONSTRUCTION, DEVELOPMENT AND TESTING
generator. To overcome this limitation, an oversized heat sink was used since this experiment was intended to test the thermoelectric generator and water stratification and not the cold sink requirements. Also, the container was bolted to the heat sink in such a way that no heat flows from the container to the heat sink through the mountings. There is no off the shelf container that matches the required properties, so a custom one had to be fabricated. A150 X 50mm PVC pipe was prepared and an aluminum cap was machined to form a bottom. The aluminum was machined to be as thin as possible such that the temperature difference across it would be minimized. The bottom is stepped such that the internal wall of the pipe fits tightly on the smaller diameter of the bottom and its larger diameter was kept 2mm over the outer diameter of the pipe. A nylon ring was machined to fit through the pipe but not through the 2mm over size of the aluminum bottom as shown in figure 15. Two holes were drilled through the nylon ring and a matching pair was drilled and tapped through the heat sink. This ring is used to hold the container to the heat sink and to apply adequate pressure on the thermoelectric module. A 2.5mm hole was drilled from the finned side of the heat sink, up to 2mm below the polished surface of the heat sink. This hole was filled with thermal conductive grease using a needle syringe and a thermocouple was inserted through to measure the cold side temperature. Thermal conductive grease was applied to both surfaces of the thermoelectric module. A nylon lid was machined with three 3mm holes. One was used as an air vent to bleed pressure off the container and the other two were used to enable temperature readings at the top and bottom of the container. 50mm glass wool insulation was wrapped round the container and the semicircular pieces were cut to insulate the lid of the container.
Figure 15: Cross section of setup 1 mounting system
38
Chapter 4 – CONSTRUCTION, DEVELOPMENT AND TESTING
The internal resistance of the thermoelectric cooler when used for power generation is unknown however, the datasheet of TEC1-12710 specifies a value at two different hot side temperatures. When the hot side is at 25°C, the internal resistance is claimed to be 1.08
and when it is at 50°C it is 1.24 . Resistivity is inversely proportional to
temperature but, in order to get a rough estimate of the resistance at 100°C, it was assumed to be linear. The two values were interpolated up to 100°C. A power resistor of 1.2Ω, 10W was used as an electrical load for this experiment. The following is a list of instruments that were used to measure the required data.
Voltage - 12 bit oscilloscope on channel 1
Current - Chauvin Arnoux E3N current clamp on channel 2 with multiple current loops to increase the resolution of clamp and oscilloscope
Bottom temperature – 10
NTC connected to a Fluke 175 multimeter
Surface temperature – 10
NTC connected to a Fluke 175 multimeter
Stopwatch
250ml of boiling water was poured into the container and the lid was closed. Temperature at the top and bottom of the container, voltage and current output were recorded every minute. An initial peak in power output was observed but it quickly decreased to tens of milliwatts. This was attributed to water stratification, since the top of the container was at 70°C while the bottom was at 30°C. Further details are presented in the results section. It was concluded that a test jig with the thermoelectric device on the top was needed. To meet this requirement a boiler had to be built such that the contained steam will heat the thermoelectric module to the required temperature. It was anticipated that temperature control of such a setup would be very unstable. When heat is applied, the temperature at the top will not increase until enough pressure is built and when heat is removed, the temperature of the boiler is expected to continue to increase due to steam pressure build up. It was concluded that with a significant heat loss to the environment through the boiler wall and a PID temperature controller, the system can be made to run in a damped control mode. Also, an automated data logger was required as the logging process was expected to take several minutes due to the system’s inertia and the need for higher accuracy. 39
Chapter 4 – CONSTRUCTION, DEVELOPMENT AND TESTING
4.6 Data logger A customized data logger was built to measure all the different readings simultaneously at desired intervals for unlimited time frames. The logger was designed to fit the allocated budget without compromising performance. A basic yet powerful logger module was found through an internet search from www.picircuit.com/shop. This is a hobbyist online shop that sells electronic circuits based on Microchip micro-controllers. The advantage of this Malaysian company is that it provides free computer software to interface with their hardware, which includes live data plotting as shown in figure 19. A data logger module, with reference number iCP12-V1.0, made up of a 66mm by 23mm printed circuit board (PCB) was chosen. This module comes populated with 3 components only: an 18F2550 Microchip micro-controller, a Universal Serial Bus (USB) connector and a set of input/output pin headers. The data logger module has a resolution of 10bit over a range of 5V, which is equivalent to a resolution of 4.88mV. A table with percentage error is presented at the end of section 4.6. No front end circuitry is present on the module, so a larger PCB had to be designed for signal conditioning of the four inputs. The logger module was then mounted on the main data logger PCB via pin headers. Further details on the logger design are presented next. Figure 16 shows the logger module as delivered from the manufacturer.
Figure 16: Data logger module - iCP12-V1.0
40
Chapter 4 – CONSTRUCTION, DEVELOPMENT AND TESTING
Figures 17 and 18 show the PCB layout of the complete data logger.
Figure 17: Data logger PCB layout
Figure 18: Data logger photograph
41
Chapter 4 – CONSTRUCTION, DEVELOPMENT AND TESTING
Figure 19: Live data view – SmartDAQ v1.3 screenshot
4.6.1 Temperature measurement The thermocouple is the preferred temperature sensing device for this setup since it is very small in size. Its thermal inertia is negligible compared to that of the thermoelectric module. Also, it will not affect the heat transfer to and from the heat sinks when it is placed in a cavity just below their surface as shown in figure 20. It is required to measure the temperature at the thermoelectric module-to-heat sink boundary and heat source-to-thermoelectric module boundary to know exactly at what the temperatures are on both sides of the module. The actual mounting location for each setup is described in the respective section. A k-Type thermocouple was chosen since it has the highest voltage to temperature ratio, at an average of 40.96µV/°C. The advantage of having a high ratio is improved resolution and accuracy with less electronic gain from the front end circuitry of the data logger. By using lower gain, noise amplification is also reduced. The setup we will run at a temperature range that is well within the k-type range.
42
Chapter 4 – CONSTRUCTION, DEVELOPMENT AND TESTING
Figure 20: Cavity for temperature sensing by thermocouple
Two options were considered for the thermocouple amplifier circuitry, an integrated circuit (IC) manufactured by Analogue Devices, AN8497, that is specifically built to interface to k-type thermocouples or a customized amplifier built from cascaded general purpose operational amplifiers OPAMS. The voltage generated by the thermocouple is nonlinear so it has to be accurately modeled by a polynomial equation using the standard thermocouple tables. However a thermocouple voltage of 40.96µV/°C can be used to estimate the voltage gain required. The hot side is expected to be subject to a temperature range of 30°C to 170°C and the cold side to 30°C - 55°C. The voltage gain of the two thermocouple inputs has to be different such that sensing on both hot and cold sides is done over the full scale of the ADC. It was decided to design for both options since the availability of the AN8497 had a lead time of 12 weeks. The circuit selection can be done by manual signal jumpers. The AN8497 does not have a fixed voltage gain because it has an integrated compensation
for
thermocouple
voltage
non-linearity
and
also
cold
junction
compensation. However it has a fixed output of 5mV/°C over a user selected voltage offset. When the AN8497 is operated on single supply without an offset, the minimum sensing temperature is 5°C since the output stage can only go down to 25mV. Also, the offset voltage cannot be set below 0V such that the full output voltage swing is dedicated for sensing between 30°C to 170°C. This can be solved by using a dual supply, however this was not available since the data logger is powered by 5V via the USB connection. The development of a system that runs on a clean dual supply requires a major investment in both time and money, so it was discarded as the 43
Chapter 4 – CONSTRUCTION, DEVELOPMENT AND TESTING
accuracy is affected more by other issues rather than by this. No negative offset was used but the circuitry for a positive offset was included should it be needed during testing. In this configuration the amplified thermocouple voltage will have a range of 25mV to 925mV for a corresponding range of 5°C to 170°C. This means that the output of the AN8497 has to be amplified again by an operational amplifier (OPAMP) to increase the output voltage swing. The MCP604 rail-to-rail OPAMP, manufactured by Microchip, was used for this purpose. A voltage gain of five was used such that maximum output will be at 4.75V. This is intentionally made at 0.25V below the maximum voltage sampling of the ADC to allow for some over temperature overshooting and to avoid clipping.
Figure 21: Front end crcuitry that uses AN8497 for temperature sensing
A similar line of thought was applied to the cold side temperature sensing using the appropriate voltage gain. As previously mentioned this circuitry was included in the PCB
44
Chapter 4 – CONSTRUCTION, DEVELOPMENT AND TESTING
design and etching but was never soldered as the lead time of the AN8497 increased to 28 weeks during the course of the project. The second amplifier option was built from a cascade of three MCP619 OPAMPs manufactured by Microchip. The circuit design is based on an application note AN844 by Microchip. The first stage is a differential amplifier with a gain of 10 followed by the second stage with a gain of 24. The first two stages are common for both hot and cold side temperature measurement at a gain of 240, but the gain of the third stage is different. As previously explained, the output of the amplifier has to be designed to make use of the full output swing. At a temperature of 170°C the output of the second stage is expected to be 40.96µV multiplied by 170 and a gain of 240, that is, 1.82V. So the gain of the last stage was set at approximately 2.5 such that the amplified voltage is 4.55V. This gives a head room of 0.35V since the maximum output voltage of the MCP619 is 4.9V at a supply voltage of 5V. The total voltage gain of the hardware setup had to be accurately measured such that the input voltage can be calculated by dividing the amplifier output by the gain. This was done using 0.01% precision resistors in a potential divider topology. Five evenly spaced voltage levels were applied to the input and the output was recorded each time. Linear fitting was done on the data to come up with a gain value. A similar line of thought was applied to the cold side temperature measurement amplifier to optimize for accuracy. When using the second amplifier option, cold junction compensation and thermocouple nonlinearity are not compensated for automatically, so this has to be done analytically for every temperature value. The process starts by converting the cold junction temperature from °C to the equivalent thermocouple voltage in micro volts. The cold junction temperature has to be logged at the same interval as the hot and cold side temperatures. Logging can be done using an LM335 connected to a spare channel on the data logger. The LM335 is an IC that regulates a voltage output at 10mV/°C. Such a signal can be accurately measured by the Microcontroller without the need for amplification. Otherwise, an ambient temperature sensor that can log data at the same interval as the data logger can be used. Conversion between thermocouple voltage and temperature was done using polynomial equations provided by OMEGA Inc. Readings 45
Chapter 4 – CONSTRUCTION, DEVELOPMENT AND TESTING
stored in the data logger were converter to thermocouple voltages by dividing them by the respective voltage gain. The cold junction temperature was then deduced from the thermocouple voltage and the result converted back to temperature using another polynomial equation. This process was done for every measurement interval using Microsoft Excel Marcos.
Figure 22: Cold and hot side temperature measurment front end circuitry
A fourth temperature measurement was done on the heat sink surface. This was done using an NTC10k thermistor connected in potential divider topology with another fixed 10kΩ, 0.25W carbon resistor. The potential divider is fed from the common 5V USB supply. At 25°C, the thermistor has a resistance of 10kΩ, so at a temperature of 25°C, the data logger reading is at 2.5V. The thermistor resistance was calculated at every sampling interval using the following equation:
46
Chapter 4 – CONSTRUCTION, DEVELOPMENT AND TESTING
(29)
The thermistor manufacturer provides a look up table of resistance and temperature values at 1°C interval. As shown on graph 2, the relationship is not linear but using the following equation, provided by the manufacturer, the temperature can be calculated. (30)
A unique probe type thermistor was used in all setups to reduce errors. The actual mounting location for each setup is described in the respective section. 160 140 120 T / °C
100 80 60 40 20 0 0
5,000
10,000
15,000
20,000
25,000
30,000
R/Ω
Graph 2: Thermistor resistance vs temperature
The heat sink surface temperature is expected to range between 15°C and 55°C. As shown by graph 3, the corresponding voltage range is between 1V and 3V. Ideally this range is amplified to use the full scale of the ADC for accurate measurement. However, accuracy is not critical since the purpose of the heat sink surface temperature measurement is to verify that the system has reached steady state condition. Melting Ice and boiling water were used to verify the thermistor readings.
47
Chapter 4 – CONSTRUCTION, DEVELOPMENT AND TESTING
100 90 Temperature / °C
80 70 60 50 40 30 20 10 0 0
0.5
1
1.5
2
2.5
3
3.5
4
Volatge / V
Graph 3: Thermistor voltage vs temperature
4.6.2 Voltage measurement As discovered from the literature review, the voltage generated by the thermoelectric modules is dependent on the temperature difference, type of module and the load resistance ratio. So building a logger that records data accurately for all setups and conditions requires an adjustable gain amplifier. However the design effort and cost do not justify such an approach. It was therefore decided to build a fixed gain OPAMP using an MCP604 in non-inverting topology as shown in Figure 23. The gain can still be adjusted by manually changing the feedback resistor R3. This resistor can be easily replaced as it is screwed into a screw terminal. The amplifier input is protected against voltage spikes via a suppressor D2 and protected against current inrush via resistor R5. R34 is an optional pull-up resistor and R21 is a current limiter for the amplifier output to protect against short circuit. Each time the amplifier gain is changed, the circuit has to be recalibrated. Calibration is done by applying an accurate voltage input to the amplifier at the lower and upper end of the ADC range. The amplifier is assumed to be linear, so linear fitting is done between the two points. Similar to what has been done before, the calibration voltage input is obtained through a potential divider circuit using 0.01% precision resistors. The calibration result is an equation that returns the actual terminal voltage at the thermoelectric module from the voltage logged by the data logger. Each experimental reading is then subject to this relationship. 48
Chapter 4 – CONSTRUCTION, DEVELOPMENT AND TESTING
Figure 23: Voltage sensing amplifier
4.6.3 Current measurement Current measurement is done by a Chauvin Arnoux E3N current to voltage transducer. This instrument outputs a voltage of 100mV/A at an accuracy of 3%, with a maximum output of 1V equivalent to 10A. To increase accuracy multiple current loops were used. The number of loops was also used to change the maximum current measurement without changing the amplifier gain. Voltage amplification is required since the maximum transducer output is only at 20% of full ADC scale. The amplifier topology and calibration setup are identical to those explained in the previous section 4.6.2. However, the calibration results in an equation that returns the current flowing through the load from the voltage measured by the data logger.
49
Chapter 4 – CONSTRUCTION, DEVELOPMENT AND TESTING
Figure 24: Current sensing amplifier
4.7 Boiler assembly The assembly consists of a closed water container that can withstand a pressure of 14 Bar such that the water can be heated up to 170°C. The container is made of a 5mm thick steel tube with a top plug to transfer heat to the thermoelectric module and a bottom flange to mount an electric immersion heater. Ideally the top surface of the container, which is in contact with the thermoelectric module, shall have the same footprint as the module such that heat flow to the heat sink happens only through the module. However if this setup is a small scale model of a fully developed generator array, the mating surface should replicate its operating conditions. Thus the hot surface
50
Chapter 4 – CONSTRUCTION, DEVELOPMENT AND TESTING
overshoots the thermoelectric module as shown in figure 25. As learnt from the literature review, the best material for the hot side surface is copper. However a fully developed generator is more likely to be made of some aluminum alloy due to weight and cost limitations. So, an aluminum round bar was machined and screwed to the top of the container. The top surface was polished to improve thermal conduction to the thermoelectric module and to obtain uniform clamping pressure over the module. Failing to do so may result in module damage, particularly at higher temperatures as aluminum expands. Heat at the perimeter of the thermoelectric module is likely to be lost to the ambient and through the heat sink more than it does at the center, so the bottom surface of the plug was tapered to obtain uniform heat transfer to the top surface in contact with the thermoelectric module. High temperature thread sealant was used to seal and lock the aluminum plug in place. The thickness of the aluminum plug is less critical than that of Setup 1 since heat is continuously supplied and the hot side temperature measurement is taken 4mm away from the thermoelectric module. Further details on temperature measurement are presented later in this section.
Figure 25: Boiler cross section
The bottom flange is made of 6mm thick steel with a central thread to enable screwing on a 1.5kW immersion heater. The flange was welded to the container and a pressure test at 80 bar using hydraulic oil and a manual hand pump was performed. The container was then cleaned and the immersion heater was screwed on and secured with high temperature thread sealant. Two eyelets were welded at the bottom of the container to mount the container to the aluminum heat sink by two long 6mm threaded bars. The mount supports were intentionally placed at the bottom of the container to 51
Chapter 4 – CONSTRUCTION, DEVELOPMENT AND TESTING
reduce heat transfer to the heat sink through the thread bars. However an unexpected side effect to his was discovered during testing of the setup. As the water temperature increased beyond 140°C, a crackling sound was heard and the output for the thermoelectric module dropped to zero. The module was then removed and inspected but visually it appeared to be intact. It was concluded that the module had failed internally due to increased clamping pressure as the container expanded but the cooler threaded bars did not. So rubber dampers were introduced at the bottom to cater for the expansion and to keep the clamping pressure constant.
Figure 26: boiler support system
As explained in the data logger section, temperature measurements were done using ktype thermocouples. The thermocouple tip is enclosed in a thin end cap of stainless steel and crimped over the braded screen that covers the thermocouple wires. The end cap, with a swollen tip, runs through a hollow M6 stainless steel bolt such that it can be screwed in at the measurement point and made tight enough for good thermal contact.
52
Chapter 4 – CONSTRUCTION, DEVELOPMENT AND TESTING
Figure 27: Thermocouple
A 5mm diameter hole was drilled through the steel container and an aluminum plug to a depth of 11mm as shown in figure 25. At this depth the thermocouple tip is exposed to the same temperature as the corner of the thermoelectric module. If the holes are drilled deeper, the temperature measurement would be more accurate; however heat flow to the thermoelectric modules would then be affected. A second hole was drilled at 90° to the first one with identical dimensions. Both holes were tapped (M6) and two thermocouples were screwed into both. No sealant was required to seal the threads since the holes are drilled thought a solid cross-section of steel and aluminum. One thermocouple was used to measure the hot side temperature and the other to control the temperature by means of a PID controller. Further details on the operation of the setup are explained later on in this section. A third hole was drilled through the container just below the top aluminum plug. The round surface of the container was filed flat so an M6 bolt with a copper washer would seal the hole as shown in figure 28. This serves as the water filling point of the container.
53
Chapter 4 – CONSTRUCTION, DEVELOPMENT AND TESTING
Figure 28: Top part of container
Water was poured to fill 75% of the container; i.e. just enough to cover the immersion heater, and the remaining 25% were left empty to cater for water expansion. If the container is filled to the top, the container would explode under the hydraulic pressure of the expanding water. As long as the water level is carefully checked by volume and double checked by weight, there is no risk of bursting the boiler. As previously mentioned, the container was tested up to 80 Bar while steam at 170°C exerts a pressure of only 14Bar on the container. So the possibility of introducing a pressure relief valve was discarded. However, all experiments were done with the container isolated in a room and personnel in a separate room. Experiments were only monitored via live data from the data logger and a USB web cam.
4.8 Temperature controller A standard, off the shelf, PID temperature controller meets the requirements of this setup. A temperature control accuracy of ±2°C is adequate enough for the scope of this project since the hot side temperature is continuously being logged independently of control. If the temperature increases further than a specific set-point, a higher hot side temperature will only result in a higher thermoelectric generation. Accuracy is only required not to damage the thermoelectric module due to large overshoots in temperature.
54
Chapter 4 – CONSTRUCTION, DEVELOPMENT AND TESTING
Figure 29: Temperature controller and solid state relay
A temperature controller that uses a K-type thermocouple and comes complete with a solid state relay output was bought. The controller, with model number REX-C100FK02V, is manufacture by RKC Industrial Instrument Inc.
4.9 Setup 2 The purpose of this setup was to quantify the electrical power output of a thermoelectric module with controlled heat supply and an oversized heat sink. The results of this experiment are to be taken as a reference to compare other experimental results with. The boiler assembly was mounted on to the large aluminum heat sink as shown in figure 30 and supported on a stand at the far ends of the heat sink (photos of the actual setup are found in the appendix). So the boiler was hanging down from the heat sink on the threaded bars whilst being pressed up against the heat sink by the rubber buffers. The heating element was connected to the temperature controller and all sensors were connected to the data logger. A fan was used to move air at 5m/s along the heat sink fins and boiler. It was necessary to keep air moving over the boiler to blow away hot air otherwise it will rise and heat up the heat sink. Also, as previously mentioned, cooling the boiler is necessary to dampen the system response to the heat input. Another scope of this setup was to compare the modules’ performance to that of other researchers and datasheets. The results are also required to justify the use of a 55
Chapter 4 – CONSTRUCTION, DEVELOPMENT AND TESTING
thermoelectric generator module over a thermoelectric cooler module for heat recovery systems.
Figure 30: Setup 2
Table 4 is a summary of the operating conditions of the setup using a thermoelectric cooler and generator modules. The modules were exposed to the same operating condition such that a performance comparison can be made. The only difference between the two is the heat transfer compound. The thermoelectric cooler uses thermal conductive paste while the manufacturer of the generator suggests the use of graphite film for optimum performance.
56
Chapter 4 – CONSTRUCTION, DEVELOPMENT AND TESTING
The thermoelectric generator module has a maximum operating temperature of 170°C while the cooler module is suggested for application below 100°C. This module is sold with temperature ratings of 138°C and 227°C. However, only the 138°C module was available for sale in small quantities. Nevertheless it was operated up to 170°C to discover more on the effects of operation outside stipulated conditions. The cooler module is very cheap compared to the generator, so all experiments were first tested on the cooler then, if no damage was observed, the same procedure was applied to the generator types.
57
Chapter 4 – CONSTRUCTION, DEVELOPMENT AND TESTING
TEG2-07025HTSS
TEC1-12710
Ambient temperature (°C)
28
28
Source temperature (°C)
30-170
30-170
40 x 40 x 3.4
40 x 40 x 3.4
0.5
1.24
Thermoelectric module size (mm) Approx. Internal resistance of module (Ω) Module cooling power( W)
N/A
96
Aluminum
Aluminum
Heat sink surface area (m )
0.7
0.7
Heat sink weight (kg)
7.8
7.8
Heat sink material 2
Heat sink no. of fins
16
16
Heat sink fin thickness (mm)
4.2
4.2
Heat sink base thickness (mm)
9.2
9.2
Aluminum
Aluminum
Heat Source to module area ratio
2.4
2.4
Heat Source to sink gap (mm)
3.4
3.4
Heat exchanger material
Heat transfer compound
Graphite
Cooling type Electric load
Air at 5ms 1.1Ω
PT10 -1
Air at 5ms 1.1Ω
-1
Table 4: Setup 2 parameters
The thermoelectric cooler and generator have a different matched load resistance, so the load is ideally changed accordingly. It would have been even better to change the load resistance as the temperature changes such that the system continuously tracks the MPP. However, an electronic load for such an operation was not available and not enough time was available to build one. As discovered in the literature review, the matched load resistance variation is expected to be approximately equal to 30%. However, the power only changes by 4%. It was decided that a fixed load will be used during all experiments. The modules were operated at the suggest load conditions by the manufacturer. Graph 4 is a plot of the matched load resistance against temperature difference across the thermoelectric generator as provided by the manufacturer. The average internal resistance of the module is 0.55Ω, however this figure is only an estimate since the true resistance value is not only dependent on the temperature difference but also on the absolute value of the cold and hot side temperatures, as discovered in the literature review. This figure lies within the range of the analyzed results in the literature review.
58
Chapter 4 – CONSTRUCTION, DEVELOPMENT AND TESTING
Internal resistance (Ω)
0.62 0.60 0.58 0.56 0.54
0.52 0.50 0.48 0
50
100
150
200
Temperature differnce (°C)
Graph 4: TEG2-07025HTSS internal reistance vs temperature difference (Source: Manufacturer)
The internal resistance of the thermoelectric cooler is specified for cooling purposes and only in two conditions, 1.08Ω at a hot side of 25°C and 1.24Ω at a hot side of 50°C. The match load resistance had to be found through this experiment by taking repeated reading using different load resistances. Repeated experiments were done using different but identical modules to ensure repeatability of results. The TEC1-12710 thermoelectric cooler is manufactured by a number of companies, so a set of modules was purchased from different suppliers to verify equivalent performance of each. On the other hand the TEG2-07025 is a specific module designed by TECTEG as previously mentioned.
4.9.1 Experimental procedure The following experimental procedure was applied to all experiments.
The data logger set at one minute intervals where each interval is an average of 1000 samples.
The PID temperature controller set to the specified operating temperature.
The data logger and live trend started.
The heating element started.
59
Chapter 4 – CONSTRUCTION, DEVELOPMENT AND TESTING
Live trend continuously monitored to detect abnormal operation and steady state condition. Steady state happens when the heat sink surface temperature is stable.
Heating element powered off at steady state.
Data logger stopped when temperature decreases below 30°C.
Thermoelectric module changed and process repeated.
4.10 Setup 3 The scope of this setup is to get an insight into the power generation capabilities of both module types when using a realistic module to heat sink area ratio as shown in figure 31. Forced cooling is still used for the previously mentioned reasons. As seen in the summary of operating condition of table 5, this setup was operated up to 110°C since the working fluid temperature of the solar water heat is not expected to be higher and to avoid damaging the thermoelectric cooler modules. The cold side temperature of the thermoelectric modules is expected to be higher than that of the previous setup for an equivalent hot side temperature since the heat sink is smaller. So, the temperature sensing amplifier was adjusted accordingly. The heat sink was supported at the far ends by thermal insulators such that no heat flowed from the heat sink to the supporting structure. The heat sink base was intentionally machined 5mm wider than the thermoelectric module to allow for proper cooling of the module’s edge and thus a uniform temperature was obtained over the module surface. It was assumed that 5mm is half of the spacing used between modules in array systems. So, the results can be extrapolated to a number of adjacent modules with 10mm spacing. The heat sink surface temperature was measured by the NTC10k as previously explained, half way along the heat sink fins and the cold side temperature was measured by k-type thermocouple. This was inserted with heat transfer compound in a 2mm hole, 2mm above the bottom of the heat sink base. The experiment procedure listed in section 4.9.1 was followed.
60
Chapter 4 – CONSTRUCTION, DEVELOPMENT AND TESTING
Figure 31: Setup 3
TEG2-07025HTSS
TEC1-12710
Ambient temperature (°C)
28
28
Source temperature (°C)
30-110
30-110
40 x 40 x 3.4
40 x 40 x 3.4
Approx. Internal resistance of module (Ω)
0.5
1.24
Module cooling power( W)
N/A
96
Aluminum
Aluminum
Heat sink surface area (m )
0.12
0.12
Heat sink weight (kg)
0.32
0.32
Heat sink no. of fins
10
10
Heat sink fin thickness (mm)
1.3
1.3
Heat sink base thickness (mm)
14
14
Thermoelectric module size (mm)
Heat sink material 2
Heat exchanger material
Aluminum
Aluminum
Heat Source to module area ratio
2.4
2.4
Heat Source to sink gap (mm)
3.4
3.4
Heat transfer compound
Graphite
Cooling type Electric load
Air at 5ms 1.1Ω Table 5: Setup 3 parameters
61
PT10 -1
Air at 5ms 1.1Ω
-1
Chapter 4 – CONSTRUCTION, DEVELOPMENT AND TESTING
4.11 Setup 4 Setup 4 is very similar to Setup 3 but with an even smaller heat sink as shown in figure 32. Table 6 summarizes the operating conditions of this setup. The scope of this setup was to understand the effect of a small heating on the system performance.
Figure 32: Setup 4
Ambient temperature (°C) 62
TEG2-07025HTSS
TEC1-12710
28
28
Chapter 4 – CONSTRUCTION, DEVELOPMENT AND TESTING
Source temperature (°C) Thermoelectric module size (mm)
30-110
30-110
40 x 40 x 3.4
40 x 40 x 3.4
Approx. Internal resistance of module (Ω)
0.5
1.24
Module cooling power( W)
N/A
96
Aluminum
Aluminum
Heat sink surface area (m )
0.014
0.014
Heat sink weight (kg)
0.07
0.07
Heat sink no. of fins
8
8
Heat sink fin thickness (mm)
2
2
Heat sink material 2
Heat sink base thickness (mm)
4
4
Aluminum
Aluminum
Heat Source to module area ratio
2.4
2.4
Heat Source to sink gap (mm)
3.4
3.4
Heat exchanger material
Heat transfer compound
Graphite
Cooling type Electric load
Air at 5ms 1.1Ω
PT10 -1
Air at 5ms 1.1Ω
-1
Table 6: Setup 4 parameters
4.12 Setup 5 It was required to quantify the power output when the system is naturally cooled. To do this test, the jig had to be turned on its side such that the boiler was laid horizontal. By doing so, the rising heat from the boiler would not be transferred directly to the heat sink without going through the thermoelectric module. In a proper working product the heat source would be insulated to solve this problem as previously mentioned in section 4.4. The heat sink fins were placed vertical to improve convection cooling. This setup is identical to Setup 3 except for the cooling type and boiler orientation. The smallest heat sink, used in Setup 4, was never tested with natural cooling since the performance with forced cooling was already poor.
4.13 Setup 6 As mentioned in the introduction of this chapter, the aim of this project is to quantify the electrical power that can be drawn from a thermoelectric module powered by a solar water heater. A generator setup of 4 thermoelectric modules was built and connected to the closed loop circuit of a solar water heater. The heat exchanger was made out of a solid aluminum block. The block was machined to form a 50mm cube with finely
63
Chapter 4 – CONSTRUCTION, DEVELOPMENT AND TESTING
polished surfaces. A 20mm hole was drill through two opposite sides and a ¾” BPST thread tapped at both ends to enable pipe fitting.
Figure 33: Heatechanger of 4 modules
A thermoelectric module was fixed to each of the two pairs of opposite sides. So no insulation was required to reduce energy losses since all four sides of the cube were covered with thermoelectric modules and the other two had pipe fittings. 22 mm thick insulation was used to sleeve the pipes and fittings and reduce heat loss. Four heat sinks with a 50mm square base were cut out of a large heat sink. As previously explained, these dimensions give a 5mm oversize on each side of the thermoelectric module to ensure uniform temperature distribution. The heat sinks were positioned with their fins vertical, to enhance heat transfer. Another advantage of having opposite heat sinks is that the mountings are bolted from heat sink to heat sink. Thus no heat flows from the hot surface to the cold surface though the bolts. Four angle brackets were screwed to each corner of each heat sink such that they are facing each other when heat sinks are mounted opposite to each other as shown in figure 34. Sets of 3mm threaded rod were cut to lengths to mount all four heat sinks. Thin rubber buffers were added to each end of the threaded bars to cater for heat exchanger expansion
64
Chapter 4 – CONSTRUCTION, DEVELOPMENT AND TESTING
Figure 34: Solar heat exchanger
The heat exchanger was assembled vertically on the outlet of a solar collector since this is the hottest point in the system. This point is also the first to reach the maximum working temperature because it is immediately prior to the entry point to the hot water storage tank. If the generator is permanently installed, a high temperature motorized valve has to be included to bypass the generator until the domestic water reaches the desired temperature. Alternatively, the valve can be omitted if the generator is positioned on the re-entry line to the solar collector. However this point does not heat up until the domestic water heats up. The advantage of installing the generator on the reentry to the solar collector is that it is the least intrusive on the thermo syphon cycle. A 2mm by 8mm deep cavity was drilled half way along one of the heat exchanger block corners, pointing towards the center. This was filled with heat transfer compound and a k-type thermocouple was inserted to measure the average hot side temperature. Another 2mm hole was drilled in one of the heat sink bases to be used for cold side temperature measurement. The thermoelectric modules were connected in series to a load resistor and to the data logger for power measurement as explained in Section 4.6. The closed loop of the SWH was filled again with glycol and water and allowed to settle for 24 hours to have any air pockets removed. It was topped up early in the morning 65
Chapter 4 – CONSTRUCTION, DEVELOPMENT AND TESTING
before sunrise and the logger was started. The system was left logging data for 12 hour of sunshine for a number of days. Each day a different load condition was set. Table 7 is a summary of the solar water heater and generator setup properties.
Figure 35: Solar water heater installation
TEC1-12710 Number of modules
4
Electrical Connection
Series
Ambient temperature (°C)
28
Source temperature (°C)
30-110
Thermoelectric module size (mm) Approx. Internal resistance of module (Ω) Module cooling power( W)
40 x 40 x 3.4 1.24 96
Heat sink material
Aluminum 2
Heat sink surface area (m )
0.4
Heat sink weight (kg)
1.28
Heat sink no. of fins
36
Heat sink fin thickness (mm)
1.3
Heat sink base thickness (mm)
14
Heat exchanger material
Aluminum
Heat Source to module area ratio
66
1.25
Chapter 4 – CONSTRUCTION, DEVELOPMENT AND TESTING
Heat Source to sink gap (mm)
3.4
Heat transfer compound
PT10 -1
Cooling type Electric load Ω
Air at 5ms 1.1 – 8.2 Table 7: Setup 5 parameters
4.10 Chapter Conclusion The construction and testing procedure of six different setups was presented in this chapter. The design of a custom made data logger was also presented. Minor changes were applied to the operating temperature mentioned in this chapter as will be explained in the next chapter to avoid damaging the thermoelectric modules.
67
Chapter 5 – RESULTS AND DISCUSSIONS
Chapter 5 – RESULTS AND DISCUSSIONS This Chapter is a review of the results from each of the setups presented in Chapter 5. A comparison to other research studies and datasheets is done although this is subject to interpretation. The generated power is mainly dependent on the cold and hot side temperatures, so measurement accuracy is critical. The ambiguity of interpretation lies in how the measurement is done. Some research studies may have measured the cooling medium temperature and others may have measured the actual thermoelectric generator temperature. As explained in Chapter 5, temperature readings in this research were taken 4mm below the hot side surface of the module and 2 mm above the cold side surface of the module. The power against temperature difference plots that are presented in this section were obtained by applying trend lines using Microsoft Excel. It was ensured that the R2 value was always better than 0.98. The trend lines are indicated in the graph legend by poly. (Series Name)
5.1 Setup 1 The setup consisted of a large heat sink at the bottom with a thermoelectric module and a hot water container on top. When boiling water was poured in, an instantaneous power output of 300mW was logged however, this decayed to 0.6mW in two minutes. Such behavior was observed with both thermoelectric cooler and generator module. It was concluded that the reason for such poor performance was related to water stratification in the container. The bottom water temperature inside the container dropped to 30°C while the top was still at 70°C.
5.2 Setup 2 The setup consisted of a boiler hanging down from a large heat sink with a thermoelectric module in between. The boiler was heated up to 170°C while a stream of air was kept moving over the heat sink at 5m/s. This setup was first used to compare identical modules manufactured by different suppliers up to a temperature of 100°C. Results show that the performance of all modules was within a 25% range with no
68
Chapter 5 – RESULTS AND DISCUSSIONS
pattern of hierarchy in performance. So, it was decided to run each of the six setups with two thermoelectric coolers and two thermoelectric generators. Repeated readings with both module types were taken. One of thermoelectric coolers failed at an approximate temperature of 145°C but it was discovered that the module itself was intact except for the leads that detached. The leads were trimmed and original solder was cleaned from the module. The leads were soldered again using higher temperature solder. The module was then operated up to 170°C without further failures. The other cooler module never failed. However it was observed that at approximately 150°C, both current and voltage of both cooler modules dropped sharply by 4% and then continued to increase at the previous rate as shown by Graph 5. It is not clear whether this phenomenon does permanent damage to the modules. Nonetheless these two modules were only used for the highest temperature test, i.e. Setup 1 and then discarded. The thermoelectric generator modules never showed such behavior.
2.5
Power / P
2 1.5 1 0.5 0 30
50
70
90
110
130
150
170
Temperature / °C Graph 5: Load current vs hot side temperature (TEC1)
Table 8 is a summary of the operating conditions that generated the highest power output of the best cooler and generator module.
69
Chapter 5 – RESULTS AND DISCUSSIONS
Unit °C
TEC 174.31
TEG 172.63
T cold at Thot max
°C
53.31
52.33
ΔT at Thot max
°C
121
120.3
Tsink at Tmax
°C
35
34
P at Thot max
W
4.88
4.61
Thot max
Table 8: Setup 2 results
Surprisingly, the thermoelectric cooler is the best performing module but only by a margin of 4% at 170°C. Furthermore the output from the thermoelectric generator is better than that specified by the module datasheet. The black trend line on Graph 6 is plotted from the module data sheet. The difference may be related to the temperature sensing points of the setup, as previously explained; however there is a significant difference in the envelope of the two plots. The difference between equivalent cooler and generator modules at a temperature difference 120°C is 8% and 10% respectively. 5 4.5 4
Power / W
3.5 3 2.5 2
Reference
1.5
Poly. (TEG1)
1
Poly. (TEG2) Poly. (TEC1)
0.5
Poly. (TEC2)
0 0
20
40
60 80 100 Temperature difference / °C
120
140
Graph 6: Output power vs temperature difference of setup 1 up to 170°C Thot
Another set of readings was taken with higher data logger resolution, by lowering the maximum temperature to 110°C. A new set of thermoelectric coolers was used. As previously mentioned, there is the possibility that the previous set modules got degraded due to high temperature exposure. Results confirm that all four modules have nearly equivalent power for equivalent temperature differences as shown in Graph 7. It was observed that the modules’ performance during the heating, steady state and the 70
Chapter 5 – RESULTS AND DISCUSSIONS
cooling cycles are nearly equivalent too. In the coming section it will be shown that this is not like so for setups with smaller heat sinks. ΔT °C
Review Power (W)
This project Power (W)
Yu et al. [14]
120
2.60
4.88
Chen et al. [17]
120
1.75
4.88
Tatarinov et al. [10]
Reference
100
2.00
3.60
Chen et al. [17]
80
0.74
2.40
Tatarinov et al. [10]
50
0.25
1.45
Hsu et al. [13]
25
0.45
0.45
Gou et al. [19]
18
0.09
0.25
Table 9: Comparison to other researches
Table 9 shows a comparison of this project to other researches in terms of power output at specific temperature differences. The experiments by Hsu et al. [13] produced equivalent results to this project however the other research studies had inferior outputs. This data can be used to prove that the ambiguity in performance comparison is related to the temperature sensor locations rather than the module itself. The only research with matched output is the air cooled one by Hsu et al. [13]. In the reviewed water cooled setups, the cold side temperature was taken as the water inlet temperature not the actual module temperature. So Table 9 was updated to Table 10 using the heat sink temperature as the cold side temperature. ΔT °C
Review Power (W)
This project Power (W) Tsink Reference
This project Power (W) Tcold Reference
Yu et al. [14]
120
2.60
3.83
4.88
Chen et al. [17]
120
1.75
3.83
4.88
Tatarinov et al. [10]
100
2.00
2.75
3.60
Chen et al. [17]
80
0.74
1.63
2.40
Tatarinov et al. [10]
50
0.25
0.78
1.45
Hsu et al. [13]
25
0.45
0.25
0.45
Gou et al. [19]
18
0.09
0.14
0.25
Reference
Table 10: Updated comparison (1)
Results in the updated comparison table are still better than the water cooled experiments but the air cooled one is now inferior.
71
Chapter 5 – RESULTS AND DISCUSSIONS
1.8 1.6
Power / W
1.4 1.2 1 0.8 Poly. (TEG1)
0.6
Poly. (TEG2)
0.4
Poly. (TEC1)
0.2
Poly. (TEC2)
0 0
10
20
30
40
50
60
Temperature differnce / °C Graph 7: Output power vs temperature difference of setup 1 up to 110°C T hot
The effectiveness of the oversized heating can be confirmed by graph 8. The cold sink temperature was very stable and thus the results of Setup 1 were used as a reference, as planned. 100 90
Temperature / °C
80 70 60 50 40 30
Thot
20
ΔT
10
Tcold
0 30
40
50
60
70
80
90
100
Temperature / °C
Graph 8: Temperature comparison
5.3 Setup 3 This setup is very similar to setup 2 but with a smaller heat sink. It was expected that the performance of the modules with a smaller heat sink would be slightly less than that 72
Chapter 5 – RESULTS AND DISCUSSIONS
of Setup1. The maximum temperature difference across the module was expected to be reduced due to a higher increase in the cold side temperature. It was planned to run the setup up to 110°C but during testing it was observed that the cold side temperature was increased significantly, so the parameters were reduced to 90°C to avoid damaging the modules. It was required to use the same unique set of modules though all experiments for comparison reasons. Table 11 is a summary of the operating conditions that generated the highest power output of the best cooler and generator modules. Unit °C
TEC 90.00
TEG 90.00
T cold at Thot max
°C
50.00
50.80
ΔT at Thot max
°C
40.00
39.20
Tsink at Tmax
°C
35.00
33.00
P at Thot max
W
1.01
1.00
Thot max
Table 11: Setup 3 results
It was observed that the heat sink temperature was higher for the thermoelectric cooler. This suggestes that the cooler module has a higher thermal conductivity. The thermoelectric cooler uses heat transfer compound while the generator uses a graphite film. So, either the module or compound may have higher thermal conductivity but only by a small margin. A study by Tatarinov et al. [10] shows that LP10 thermal grease does have higher thermal conductivity at lower temperature. However this may degrade quickly with time as reviewed in Chapter 2.
73
Chapter 5 – RESULTS AND DISCUSSIONS
1.00
Power / P
0.80 0.60 0.40
Poly. (TEG1) Poly. (TEG2)
0.20
Poly. (TEC1) Poly. (TEC2)
0.00 0.0
5.0
10.0
15.0
20.0
25.0
30.0
35.0
40.0
Temperature difference / °C Graph 9: Setup 3 performance
The performance of three of the modules was very similar but one thermoelectric generator module had a consistent poorer performance. It was concluded that this may be the result of improper installation of the module since repeated readings were taken from a single installation. Thus it was learnt that ideally repeated readings are taken after removing and reinstalling the module. The module performed better in subsequent tests. As expected, for an equivalent hot side temperature, Setup 3, with a smaller heat sink, produced less temperature difference across the module than Setup 2. Similarly, the generated power was reduced for an equivalent temperature difference. This may be attributed to reduced heat transfer by the smaller heat sink and thus reduced current through the load. Two comparison graphs are presented in section 6.4. Table 12 is a comparison of Setup 3 to other research studies. The entries with N/A are not available since the setup was never operated in such conditions. The discrepancy between the performance of this project and other studies was reduced due to the reduction in heat sink size.
74
Chapter 5 – RESULTS AND DISCUSSIONS
ΔT °C
Review Power (W)
This project Power (W) Tsink Reference
This project Power (W) Tcold Reference
Yu et al. [14]
120
2.60
N/A
N/A
Chen et al. [17]
120
1.75
N/A
N/A
Tatarinov et al. [10]
100
2.00
N/A
N/A
Chen et al. [17]
80
0.74
N/A
N/A
Tatarinov et al. [10]
50
0.25
0.75
N/A
Hsu et al. [13]
25
0.45
0.23
0.42
Gou et al. [19]
18
0.09
0.13
0.23
Reference
Table 12: Updated Comparison (2)
5.4 Setup 4 This setup is very similar to Setup 2 and 3 but with an even smaller heat sink. The maximum hot side temperature was limited to approximately 71°C such that the maximum cold side temperature was in the range of 50°C as was done in the previous setups. Table 13 is a comparison of the best performing thermoelectric cooler and generator. The heat sink temperature of the generator module was higher than that of the cooler module at an equivalent cold side temperature. This contradicts the explanation given earlier for Setup 3. It was also observed that both generator modules performed worse than they did in setup 3 for equivalent temperature difference. These two observations can be linked together to justify both. Since the heat sink temperature was higher, less heat was being transfered away from the heat sink, so less current flowed through the load. Thus, less power was generated. However the reason for a warmer heat sink is not clear. Unit °C
TEC 67.76
TEG 68.88
T cold at Thot max
°C
51.46
51.28
ΔT at Thot max
°C
16.30
17.60
Tsink at Tmax
°C
37.00
41.00
P at Thot max
W
0.24
0.19
Thot max
Table 13: Setup 4 results
The two generator modules and the one cooler module had matched performance but one cooler module was approximately 35% superior. This may be related to the thickness of the heat transfer compound or the clamping pressure. As previously 75
Chapter 5 – RESULTS AND DISCUSSIONS
explained, repeated readings were taken from a unique installation, so ideally the module is freshly installed for each test run. Figure 36 shows the generated power against temperature difference for each module.
0.20
Power / P
0.15
0.10 Poly. (TEG1) Poly. (TEG2)
0.05
Poly. (TEC1) Poly. (TEC2)
0.00 0.0
2.0
4.0
6.0
8.0
10.0
12.0
14.0
16.0
18.0
Temperature difference / °C Graph 10: Setup 4 performance
Graphs 11 and 12 show a performance comparison of Setups 2, 3 and 4 using the best thermoelectric cooler and generator respectively. It is clear that the performance of both module types is equivalent, but it is not known whether either of the modules will be affected by long term use. The best performing generator module was unique throughout testing but the best cooler module was not. The following list is a summary of the possible reasons to try to explain why the thermoelectric cooler modules’ performance was inconsistent.
Difference in the applied thickness of heat transfer compound to the TEC while the graphite film of the TEG is factory made with precision. Increased compound thickness reduces the thermal conductivity to and from the module.
Tightening torque affects the spread out of heat transfer compound. So different setups may have been subject to different spreads.
Thickness uniformity of heat transfer compound. Semiconductor manufacturers suggest allowing 24 hours for the heat transfer compound to spread out before operating a device. However this was not possible in the allowed time frame. 76
Chapter 5 – RESULTS AND DISCUSSIONS
The setup consisted of a long boiler with a small diameter. So there is the possibility that the top face of the boiler was not always perfectly parallel to modules. This is likely to occur with TEC since the heat transfer compound reduces the surface friction and thus the boiler may slip during tightening.
0.30
Power / P
0.25 0.20 0.15 0.10
Poly. (Setup 1) Poly. (Setup 2)
0.05
Poly. (Setup 3)
0.00 0.0
2.0
4.0
6.0
8.0
10.0
12.0
14.0
16.0
18.0
Temperature differecne / °C Graph 11: Comparison of setups 2, 3 and 4 TEC
0.35
Power / W
0.30 0.25 0.20 0.15 0.10
Poly. (Setup 2) Poly. (Setup 3)
0.05
Poly. (Setup 4)
0.00 0.0
2.0
4.0
6.0
8.0
10.0
12.0
14.0
Temperature difference / °C Graph 12: Comparison of setups 2, 3 and 4 TEG
77
16.0
18.0
20.0
Chapter 5 – RESULTS AND DISCUSSIONS
5.5 Setup 5 This setup is identical to Setup 3 but is naturally cooled. The scope was to have a reference with which to compare the solar water heater installation results. Table 14 is a summary of the results from this setup. Unit °C
TEC 67.30
T cold at Thot max
°C
44.90
ΔT at Thot max
°C
22.40
Tsink at Tmax
°C
35.00
P at Thot max
W
0.20
Thot max
Table 14: Setup 5 results
Natural cooling reduced the performance of the thermoelectric cooler by approximately 50% when compared to an equivalent setup with forced cooling as shown in Graph 13. After reviewing the data of the first five setups it was concluded that the best option is to use the medium size heat sinks for the solar setup. Thermoelectric cooler modules were used with natural cooling since this option is mostly likely the cheapest one for a full scale installation. However it does not necessarily mean that it is the most feasible. 0.50 0.45 0.40
Power / P
0.35 0.30 0.25 0.20 0.15 0.10
Poly. (Natural cooling)
0.05
Poly. (Forced cooling)
0.00 0.0
5.0
10.0
15.0
Temperature difference / °C Graph 13: Natural to forced cooling comparison
78
20.0
Chapter 5 – RESULTS AND DISCUSSIONS
5.6 Setup 6 The setup consisted of a generator made of four thermoelectric coolers. A simple design was adopted with a thermoelectric cooler and heat sink on each side of a square section heat exchanger block. The heat exchanger was placed in line with the closed loop of working fluid of a flat plat collector. Graph 14 is a plot of the hot side temperature measured on 15th of May and 1st of September during the hours of sunshine. 75
May Septemeber
Temperature / °C
65 55 45 35 25 15 0
100
200
300
400
500
600
Time / minutes Graph 14: Temperature difference vs time on May15th and September 1st
79
700
800
Chapter 5 – RESULTS AND DISCUSSIONS
Five different load resistances were used to find the highest power point of the system. As mentioned in Chapter 4, ideal constant power point tracking is done to compensate for the changes in the internal resistance of the modules as the temperature changes. However, as discovered in the literature review the change in performance should be in the range of only 4%. 0.9 5°C
0.8
10°C
0.7
15°C
Power / W
0.6 0.5 0.4 0.3 0.2 0.1 0 0
1
2
3
4
5
6
7
8
9
Load resistance / Ω Graph 15: Power vs load resistance
Graph 15 shows the power output of all four series-connected modules, at 3 different levels of temperature difference under different load conditions. Results confirm that the power increases sharply until MPP and then it decreases gradually. Also, the rate of increase of power increases as temperature difference increases. Unit °C
TEC 72.10
T cold at Thot max
°C
52.80
ΔT at Thot max
°C
19.30
Tsink at Tmax
°C
44.00
P at Thot max
W
0.28
Thot max
Table 15: Setup 6 results (per module)
All the power output against temperature difference plots of this setup were observed to be significantly scattered. Graph 16 is one particular example. Unlike this setup, Setups 3 to 5 resulted in a neater plot with practically no scatter. It was concluded that this is 80
Chapter 5 – RESULTS AND DISCUSSIONS
the result of changes in wind speed and direction. The importance of this observation is to understand that continuous MPPT is required to optimize for power output when the system is naturally cooled in an outside environment. A maximum change in power of 25% was observed. This figure is much higher than 4% as previously mentioned. This result suggests that MPPT is required for naturally cooled systems. 0.3 y = 0.0002x2 + 0.0087x + 0.0166 R² = 0.9844
0.25
Power / P
0.2 0.15 0.1 0.05 0 0.0
2.0
4.0
6.0
8.0
10.0
12.0
14.0
16.0
18.0
20.0
Temperature difference / °C Graph 16: Typical power vs temperature difference plot for a single module
The performance of Setup 6 was compared to Setup 3 and 5. As expected the power output of Setup 6 lies between that of Setup 3, which was force cooled, and Setup 5 which was naturally cooled. Graph 17 shows a comparison between the setups.
81
Chapter 5 – RESULTS AND DISCUSSIONS
0.2
Power / P
0.15
0.1 Poly. (Setup 5)
0.05
Poly. (Setup 6) Poly. (Setup 3)
0 0
2
4
6
8
10
12
14
Temperature / °C Graph 17: Comparison of solar generator, forced, natural cooling
The power generation from the solar installation degraded by an average of 34% in 135 days. Results show that the voltage degradation and current are at 11% and 26% respectively as shown in graphs 18 and 19. The high current degradation suggests that the heat flux to thermoelectric modules deteriorated. This may be attributed to the degradation of the heat transfer compound due to the long exposure to high temperature. Such behavior was also reported by Tatarinov et al. [10]. 0.3
Volatge / V
0.25 0.2 0.15 Poly. (Day 1) Poly. (Day 135)
0.1 0.05 0 0.0
2.0
4.0
6.0
8.0
10.0
Temperature / °C Graph 18: Voltage degradation after 135 days
82
12.0
14.0
Chapter 5 – RESULTS AND DISCUSSIONS
0.50 0.45 0.40
Current / A
0.35 0.30 0.25 Poly. (Day 1)
0.20
Poly. (Day 135)
0.15 0.10 0.05 0.00 0.0
2.0
4.0
6.0
8.0
10.0
Temperature difference / °C Graph 19: Current degradation after 135 days
83
12.0
14.0
Chapter 6 – CONCLUSIONS AND FUTURE WORK
Chapter 6 – CONCLUSION AND FUTURE WORK The aim of this thesis was to quantify the electrical power output from a thermoelectric module powered by a solar water heater. A literature review was done to learn from other researchers’’ experience and to foresee any challenges that may be encountered. Literature of a similar setup was not found, however the knowledge from the reviewed results was used to build a better working model of the system. The Literature review was started by learning the basics of the thermoelectric conversion process. The parameters that affect the conversion efficiency and performance were explained in further detail in the theory section. Studies on low temperature system are much less common since thermoelectric conversion efficiency increases with the square of temperature across a thermoelectric module. It was learnt that thermoelectric cooler modules can be used in reverse operation to generate power just as thermoelectric generator modules do. So by applying a temperature difference across the module, heat will flow through the module and an electromotive force will be generated. The advantage of the cooler modules is that they are much cheaper. The thermoelectric coolers used in the project were bought at 3€ each while the generator modules were bought at 30€ each. The most important points that were mentioned in the literature review include; the importance of uniform heat and temperature distribution over a single module and multiple module systems; the advantage of using maximum power point tracking over fixed load systems or alternatively load matching within a window of 0% to +30% of matched load; the load voltage is half the open circuit voltage at maximum power point; maximum power point is dependent on hot side temperature and the temperate difference across the module; the effect of improving the cold side and hot side heat transfer coefficient independently; analysis of different heat transfer compounds; the effect of adding a spacer block in a system and other equally important suggestions. Five different setups were built to collect data and more knowledge on thermoelectric generation. The knowledge was then used to build a simple generator to be operated with the solar water heater. The collected data was also used as a reference to cross 84
Chapter 6 – CONCLUSIONS AND FUTURE WORK
check the data collected from the solar water heater setup. A customized data logger was built to measure voltage, current, hot, cold and heat sink temperatures. The first setup consisted of a heat sink with a thermoelectric module and a hot water container on top. The output from this setup was very poor and thus the initial idea to install the thermoelectric modules around the hot water storage tank of a solar water heater was discarded. The second setup consisted of an oversized heat sink on top and a custom built boiler hanging down from it with a thermoelectric module in between. The main purpose of this setup was to test the modules in an environment where the only limitation is the module itself. The results were considered as the best performance that can be drawn for the modules. Furthermore the data was compared to other studies and the manufacturer’s datasheet. It was pointed out that comparison to the datasheet can only be done for the generator modules since the cooler modules are not intended for power generation. The third and fourth setups were similar to the second, but with a progressively smaller heat sink. The results from each setup were compared to each other and to other studies mentioned in the literature review. The results of all setups were superior or equivalent to those present in the literature review; however this does not necessarily mean that this project produced better results. The performance of a thermoelectric setup is done by plotting the power output against temperature difference. It was explained that the difference in performance may be related to the ambiguity of temperature measurement. The ambiguity lies in how temperature sensing is done, that is, at the cooling medium and hot source entry point or directly over the module surfaces. The fifth setup was identical to the third but using natural cooling. The sixth setup was the actual installation of a small scale thermoelectric generator of 4 modules. This setup was tested only with cooling modules since it was concluded that generator and cooler modules have matched performance provided that they are carefully assembled. A simple design was adopted with a thermoelectric module and heat sink on each side of a square section heat exchanger block. The hollow heat exchanger was placed in line with the closed loop working fluid of a flat plat collector. Full days test runs were done with different load resistances. It was concluded that a
85
Chapter 6 – CONCLUSIONS AND FUTURE WORK
maximum power point tracker is required due to continuously changing temperature conditions in a windy, open environment. A simple degradation test was done by operating a thermoelectric module for 135 day between May and September. This resulted in a decay of 35% in power output. It was concluded that this decay in power was due to deterioration of the heat transfer compound since the major degradation was in load current not voltage. This result was reported by at least one other study [10]. A maximum temperature difference of 19.3°C was recorded when the hot side temperature was 72.1°C. The power generation in this condition was 0.275W per thermoelectric module. It would be interesting to know the net power output of a force cooled solar water heat installation. Testing is ideally done over 12 months such that a year round performance profile is obtained to judge the feasibility of a full installation. A larger scale test can be done with an increased number of modules such that a significant temperature difference before and after the generator is obtained. Such data can be used to assess the energy that can be recovered without compromising the scope of the solar water heater, i.e. heating domestic water. However increasing the number of modules introduces the challenge of uniform temperature and heat distribution. The effect of a spacer block between the heat sink and the thermoelectric module is another interesting aspect that can be investigated.
86
REFERENCES [1]Webpage: California Institute of Technology. The Science of Thermoelectric Materials. Thermoelectrics Caltech Materials Science. February 12th 2014. http://thermoelectrics.caltech.edu/thermoelectrics/index.html [2] Webpage: California Institute of Technology. “Engineering Fundamentals, Thermoelectric Effect”, Thermoelectrics Caltech Materials Science, February 12th 2014. http://www.efunda.com/designstandards/sensors/thermocouples/thmcple_theory.cfm [3] T. Caillat, J. -P. Fleurial, G. J. Snyder, and A. Borshchevsky. “Development of High Efficiency Segmented Thermoelectric Unicouples” Thermoelectrics, 2001. Proceedings ICT 2001. XX International Conference, pg282 – 285. [4] Webpage: California Institute of Technology. “Thermoelectric Engineering”. Thermoelectrics Caltech Materials Science. February 12th 2014. http://thermoelectrics.caltech.edu/thermoelectrics/engineering.html [5] Riffat SB, Xiaoli MA. Thermoelectronics: “A review of present and potential application” Applied Thermal Engineering, 2003, Vol23:913-35. [6] Webpage: Peter XX, Green Car Congress, “BMW provides an update on waste heat recovery projects, Turbostreamer and the Thermoelectric Generator”, 16 February 2014. http://www.greencarcongress.com/2011/08/bmwthermal-20110830.html [7] Bell LE. “Cooling, heating, generating power, and recovering waste heat with thermoelectric systems” Science, 2008, 321:1457-61. [8] L.I. Anatychuk, O.J Luste, R.V. Kuz, “Theoretical and experimental study of thermoelectric generators for vehicles”, Journal of Electronic Materials, Vol.40, No.5, 2011. [9] Yuchao Wang, Chuanshan Dai, Shixue Wang, “Theoretical analysis of a thermoelectric generator using exhaust gas of vehicles as heat source”, Applied Energy, 112:1171-1180, 2013.
87
[10] Dimitri Tatarinov, Daniel Wallig, Georf Bastian, “Optimization characterization of thermoelectric generators”, Journal of Electronic Materials, Vol. 41, No. 6, 2012. [11] C. Ramesh Kumar, Ankit Sonthalia, Rahul Goel, “Experimental study on waste heat recovery from an internal combustion engine using thermoelectric technology”, Thermal Science, 2011, Vol. 15, No. 4, pp. 1011-1022. [12] Jianlin Yu, Hua Zhao, “A numerical model for thermoelectric generator with the parallel-plate heat exchanger”, Journal of Power Sources 172:428-434, 2007. [13] Cheng-Ting Hsu, Da-Jeng Yao, Ke-Jyun Ye, Ben Yu, “Renewable energy of waste heat recovery system for automobile”, Journal of Renewable and Sustainable Energy 2, 013105, 2010. [14] Xing Niu, Jianlin Yu, Shzhomg Wang, “Experimental study on low-temperature waste heat thermoelectric generator”, Journal of Power Sources 188:621-626, 2009. [15] Navneesh Phillip, Othman Maganga, Keith J. Burnham, Marka A.Ellis, Simon Robinson, Julian Dunn, “Investigation of maximum power point tracking”, Journal of Electronic Materials, Vol. 42, No. 7, 201. [16] Casper Vadstrup, Erik Schaltz, Min Chen, “Individual module maximum power point tracking for thermoelectric generator systems”, Journal of Electronic Materials, Vol. 42, No. 7, 2013. [17] Chen-Ting-Hsu, Gia-Yeh Huang, Hsu-Shen Chu, Ben Yu, Da-Jeng Yao, “Experimental and simulations on low-temperature waste heat harvesting system by thermoelectric power generators”, Applied Energy 88:1291-1297, 2011. [18] Wei-Hsin Chen, Chen-Yeh Liao, Chen-I Hung, Wei-Lun Huang, “Experimental study on thermoelectric modules for power generation at various operating conditions”, Energy 45:871-881, 2012. [19] Xiaolong Gou, Heng Xiao, Suwen Yang, “Modeling, experimental study and optimization on low-temperature waste heat thermoelectric generator system”, Applied Energy 87:3131-3136, 2010. 88
[20] Laird Technologies, Thermoelectric handbook, November 2010. Accessible from: http://www.lairdtech.com/ [21] Webpage: TEG Thermoelectric GenCell Technology, September 2014, http://www.tecteg.com/
BIBLIOGRAPHY Manfred Altman, Elements of Solid-State Energy Conversion, University Series in Basic Engineering, American Book Company, 1969. George W. Sutton, Direct Energy Conversion, Inter-University Electronics Series, Vol.3, McGraw-Hill Book Company, 1966.
89
APPENDIX 1
90
