Research Article Thermoelectric Energy Harvesting for ...downloads.hindawi.com/journals/ijdsn/2013/232438.pdf · e main contribution of this paper lays in the power management circuit

Hindawi Publishing CorporationInternational Journal of Distributed Sensor NetworksVolume 2013 Article ID 232438 14 pageshttpdxdoiorg1011552013232438

Research ArticleThermoelectric Energy Harvesting for Building EnergyManagement Wireless Sensor Networks

Wensi Wang Victor Cionca Ningning Wang Mike Hayes Brendan OrsquoFlynnand Cian OrsquoMathunaMicrosystems Group Tyndall National Institute Cork Ireland

Correspondence should be addressed to Wensi Wang wensiwangtyndallie

Received 12 March 2013 Accepted 14 May 2013

Academic Editor Al-Sakib Khan Pathan

Copyright copy 2013 Wensi Wang et al This is an open access article distributed under the Creative Commons Attribution Licensewhich permits unrestricted use distribution and reproduction in any medium provided the original work is properly cited

A thermoelectric energy harvester powered wireless sensor networks (WSNs) module designed for building energy management(BEM) applications is built and tested in this work An analytic thermoelectric generator (TEG) electrical model is built andverified based on parameters given in manufacturer data sheets of Bismuth Telluride TEGs A charge pumpswitching regulatortwo-stage ultra-low voltage step-up DCDC converter design is presented in this work to boost the lt05 V output voltage of TEGto usable voltage level for WSN (33 V) The design concept device simulation circuits schematic and the measurement resultsare presented in detail The prototype device test results show 25 end-to-end conversion efficiency in a wide range of inputtemperaturesvoltages Further tests demonstrate that the proposed thermoelectric generator design can effectively power WSNmodule which operates with a 17 duty cycle (58 seconds measurement time interval) when the prototype is placed on a typicalwall-mount heater (60∘C surface temperature) The thermoelectric energy harvesting powered WSN demonstrates duty cyclessignificantly higher than the required duty cycle for BEMWSN applications

1 Introduction

Ubiquitous computing and short range wireless communi-cation have been developing rapidly since 1990s Evolvingfrom basic point-to-point radio frequency communicationwireless communication technology now provides reliabledata links within ad hoc networks In the similar period oftime the developments of advanced and low cost sensorssignificantly increase the applications of modern sensingsystems The combination of wireless communication andsensor technologies has made it possible to transform theconventional large scale monitoring equipment into smartwireless sensor node within ad hoc networks

Wireless sensor networks (WSNs) technologies havebecome a research focus in recent years for wide range ofapplications One important application ofWSN is within thebuilding energy management (BEM) area Energy consump-tion of commercial and residential buildings is responsiblefor 40 of energy usage in the USA since mid-2000s [1]Many have suggested that the utilization of building energymanagement systems potentially leads to 25ndash40 energy

savings from smart control of heating ventilation air con-ditioning (HVAC) and lighting [2ndash4] Applications of WSNfor building energymanagement system have been addressedin [5ndash8]

In these studies one issue that has been frequentlydiscussed for BEMWSN development is the limited lifetimeof wireless sensor modules (also known as ldquomoterdquo) due tobattery energy capacity With increased number of motes inthe WSN the mandatory requirement to replace battery forseveral hundreds even thousands ofWSNmotes significantlyincreases themaintenance cost and reduces system reliability

Energy harvesting methods have been proposed to scav-enge ambient energy in order to ldquoself-powerrdquo WSN systemsEnergy harvesting technologies present a solution to supplyWSN mote with infinite ambient energy instead of locallystored battery energy Various types of ambient energysources have been proposed for energy harvesting such asindoor and outdoor photovoltaic energy [9 10] mechanicalvibration energy [11 12] and thermoelectric energy [13ndash16]

In BEM applications a large number of heating devicesin the HVAC units boiler water heaters and hot water

2 International Journal of Distributed Sensor Networks

pipes exist in most commercialresidential buildings Thesepotential heat sources with 50ndash100∘C temperature provideideal energy sources for building energy monitoring systems

Utilization of these thermoelectric energy sources maylead to the highly desired ldquodeploy-and-forgetrdquo WSN thatis once the energy harvesting powered mote is deployedonto the heat source the WSN becomes self-powered andachieves power autonomy However the implementation ofthermoelectrical energy harvesting in order to achieve thecontinuous and maintenance-free WSN mote operation issignificantly constrained by the ultra-low voltage (less than05 V) small power (1mW or sub-1mW) and device sizelimitation of thermoelectric generator (TEG)

Despite these challenges thermoelectric energy harvest-ing technologies are developing at fast speed in recentyears Many works addressed the high ZT figure of meritthermoelectric generator using MEMS or nanotechnologyfabrication processes [17ndash19] Ultra-low voltage DCDC con-verter for thermoelectric generator applications has alsobeen proposed in [14 16] However thermoelectric energyharvester system level design for wireless sensor module isan area less addressed Although the thermoelectric energyharvester systems presented in [20ndash22] give a well-definedgeneral review and analysis these works lack the importantdetails on powermanagement circuit design and componentsselection

The purpose of this paper is to demonstrate a practicalmethodology concerning the thermoelectric energy har-vester design for WSNmote in real-world BEM applicationsThis method firstly characterizes the power consumptionprofile of WSN mote The thermoelectric energy harvestermodel is then used to estimate the TEG device size andconfigurations for the WSN power consumption Associatedpower management circuit is then designed towards highconversion efficiency with TEG configuration This method-ology builds and optimizes energy harvester based on real-world WSN mote power consumption This application-oriented design concept incorporates more realistic designconsiderations when the device is deployed in real-worldconditions

This paper introduces a thermoelectric generator elec-trical characteristics model for low temperature applications(hot side temperature lt100∘C cold side temperature isroom temperature with passive heat sink cooling) and itsresults verification based on a custommanufactured BismuthTelluride (Bi2Te3) TEG module

The main contribution of this paper lays in the powermanagement circuit design for ultra-low voltage DCDCconversion a two-stage DCDC converter circuit with ultra-low start-up voltage charge pump and switched-mode boostconverter is proposed to step up the minimal 250mV TEGvoltage output The detailed circuit design and componentsselection are presented Energy storage unit and associatedoutput power regulator circuit are introduced to completethe thermoelectric energy harvester design The power con-sumption of Tyndall Zigbee WSN mote is also presentedand compared to the power generated from thermoelectricenergy harvester Finally the thermoelectric energy harvest-ing poweredWSN prototype device and its evaluation results

are presented The energy flow and conversion efficiencyin each stage of power conversion are presented based onmeasurements made on the prototype

The rest of the paper is organized as follows In Section 2background and related works in the area of thermoelectricenergy harvesting are introduced and compared Section 3shows the proposed system architecture of thermoelectricenergy harvester Section 4 addresses the detailed designissues of thermoelectric energy harvesting powered WSNmote system Section 41 presents the energy consumption ofWSN mote designed in Tyndall and its typical power profileSection 42 shows the proposed TEG simulation model forroom temperature applications and its verifications Sec-tion 43 discusses the power management circuits designand implementation Section 43 also introduces the energystorage unit and the output power regulator circuits Section 5presents the TEG prototype device and the experimentalresults Section 6 summarizes the conclusions of this paper

2 Background and Related Works

Thermoelectric energy harvesting is based on the Seebeckeffect which directly converts temperature difference intoelectricity The structure of a typical vertical TEG moduleis illustrated in Figure 1 When a temperature difference isapplied cross the P-N types of TEG couples voltage potentialis generated on the the TEG couple The thermoelectricmaterial performance is usually measured in thermoelectricfigure of merit ZT [13]

Currently a typical ldquohighrdquo energy conversion efficiencycommercially off-the-shelf TEG module features a ZTaround 06-07 in room temperature This was achieved byBi119909Sb2minus119909

Te3 type material in the late 1960s [24] MEMSbased TEGs have been proposed in recent years in [1719] The MEMS based TEG although it adopts the similarmaterial used in conventionalmachinedTEG due to the largenumbers of the thermo-couples the power output of MEMSbased TEG increases significantly

Designated for powering miniaturized wireless sensornodes the TEGmodule and its heat sink are inherently smallin form factor The temperature difference available on aconventional heat sink in such condition is also small (lt10∘Cfrom a lt100∘C heat source) With relatively low temperaturedifference and limited thermocouples size further powerconditioning is essential to the energy harvesting systemdesign

Dalola et al presented a TEG powered sensor and trans-mitter circuit in their paper [25] Charge pump converter isused as the only DCDC conversion circuit Ramadass andChandrakasan proposed an advanced thermoelectric powermanagement solution [26] A DCDC converter with mini-mum start-up voltage at 35mV is designed and implementedwith a 013120583m CMOS process A manual controlled switchis required to start up the DCDC converter Carlson et aldemonstrates a 20mV start-up voltage DCDC converterin [14] Although this DCDC converter achieves a 75maximum converter efficiency it requires an external voltagesource to initialize the start-up circuit

International Journal of Distributed Sensor Networks 3

Heat sink

Top substrate

Bottom substrateHeat

Copper contact

Hot side

Cold side

Seal material

Thermo-element

P N PN PN PN PN

Figure 1 Seebeck effect in thermoelectric energy harvesting

IS ID

ILVL

LD

C R

S

+

V+ S1 S2

S3 S4

C1

C2

Step-upDCDC converter

Energy storage unit

Output power regulation WSN mote

Thermal energy

SuperCap

Buck-boost converter Tyndall mote

TEG module

Heat sink

Impedance match

Input stage Storage stage Output stage

State-of-charge (SOC) monitor

PN

Figure 2 Block diagram of thermoelectric energy harvesting powered wireless sensor network

Linear Technology released an energy harvesting powermanagement chip LTC3108 [27] It achieves top efficiencyat 40 using a 1 100 transformer setup However forinput voltage at 02 V or higher the conversion efficiencydecreases to less than 10 when output voltage is set to 45 V(1 100 transformer setup) Texas Instruments also releasedan energy harvesting power management chip BQ25504 forthermoelectric energy harvesting [28] It features a mini-mal start-up voltage of 40mV BQ25504 requires a start-up current of several mA in order to obtain 18 V powersupply voltage on the storage capacitor During this start-upphase this converter operates in cold start mode with lowconversion efficiency For many 1mW or sub-1mW energyharvesting applications it is possible that the harvested powerin the cold start phase is not sufficient to charge the storagecapacitor to its required voltage threshold which potentiallyleads to unsuccessful start-up and the system cannot enterthe normal operation mode

3 Thermoelectric Energy Harvesting PoweredWSN System Architecture

A complete thermoelectric energy harvesting powered WSNsystem consists of five subsystems as shown in Figure 2

(1) Thermoelectric generator (TEG) main design issuesare type of material number size and connectionconfiguration of thermocouples

(2) Ultra-low voltage step-up DCDC converter theoutput voltage of most of TEGs is less than 500mVConventional boost converter and charge pump can-not provide the low start-up voltage required bythe TEGs DCDC converter with lower minimalstart-up voltage is essential for thermoelectric energyharvesting

(3) Energy storage unit the conventional rechargeablebattery which requires relatively complex chargingcircuit and has a limited lifetime and charge cyclesis not suitable for most low power energy harvestingapplications Electrical double layer capacitors (alsoknown as supercapacitors or SuperCaps) which onlyrequire a simple charging circuitry and have a longoperation lifetime (gt10 years) are considered to bethe preferred energy storage unit (ESU) in energyharvesting

(4) Output power regulation output voltage regulator isused to extract ldquousablerdquo energy from energy storageunit For SuperCap type ESU with a changeable out-put voltage the output power regulation is importantto obtain a constant voltage to the WSN mote Inaddition with a buck-boost converter topology theoutput voltage regulator also increases the amount ofstored energy that can be used for WSN mote

(5) WSN mote the energy harvester design is largelybased on the power consumption and other electrical

4 International Journal of Distributed Sensor Networks

interface

MCU ADC

MCU module

Zigbee EM2420 RF module

Battery voltage monitor

Battery energy harvester

Power supply module

Pow

er m

anag

emen

t Sensirion SHT temperature and

humidity sensor

Avago APDSanalog light sensor

Murata IRA PIR sensor

PT-100 thermistor

Sensor module

I2CMCU

Atmega1281core

Figure 3 Wireless sensor node for building energy management application [23]

characteristics of WSN mote The power consump-tion analysis of WSN mote is an essential step beforeenergy harvester design

In addition these building blocks of thermoelectricenergy harvester are not isolated components but inter-related subsystems Several design considerations and chal-lenges need to be addressed in the practical design andimplementation of thermoelectric energy harvesting pow-ered mote

(1) The energy equilibrium between mote power con-sumption and thermoelectric energy harvester powergenerationmust be achieved to ensure a continuouslyoperating WSN mote

(2) The impedance matching between TEG and DCDCconversion has direct impact on the system powerconversion efficiency

(3) The energy storage unit and power consumption ofWSN mote will determine the lifetime of mote whenthermal energy source is temporarily unavailable

Detailed design and analysis of each subsystem and therelationships between the aforementioned subsystems arepresented in the following sections

4 Design and Optimization of ThermoelectricEnergy Harvesting Powered WSN Mote

41 Power Consumption of Wireless Sensor Module A typicalwireless sensor network for BEM applications consists of anumber of wireless sensor nodes (motes) For each moteit normally features (1) microcontroller unit (MCU) (2)wireless communication unit (3) sensors with digitalanaloginterfaces and (4) power supply

The Tyndall mote [29] layout and implementation forBEM application is shown in Figure 3 The sensor interfacesare 1198682119862 bus for digital sensors and 10-bit analog-to-digitalconverter (ADC) for analog sensors The microcontrolleradopted in the design is an Atmel1281 and the RF moduleis CC2420 radio chip

WSN mote features high power consumption (10ndash100mW) in active mode and low power consumption in

sleep mode (10ndash50 120583W) Thus WSN mote often operates inactivesleep duty cycles in order to reduce average powerconsumption During operating cycles each mode of oper-ation has an intrinsic power consumption value The mainoperating modes are

(i) initialization and clear channel assessment (CCA)mode

(ii) sensing mode (sensors are active and being sampledthrough the ADC)

(iii) preamble and payload Tx (RF transmit) mode(iv) acknowledgement (ACK) Rx (receiving) mode and(v) sleep mode

In addition to the power consumption profile the averagepower consumption of the WSN mote is also determined bythe active mode duty cycle 119863 The average power consump-tion 119875Avg can be expressed as

119875Avg = 119875act sdot 119863 + 119875slp sdot (1 minus 119863) (1)

The power consumption values for the Tyndall moteshown in Figure 4 are listed in Table 1

In order for aWSNmote to operate indefinitely the aver-age power consumption needs to be lower than the averageharvested power and this can be controlled by adjusting theduty cycle(s)119863 of the application Furthermore the capacityof the power source (battery or supercapacitor) needs tobe large enough to support peak power consumption forexample the power consumption during RF transmission

Based on the power consumption and duty cycle char-acteristics of WSN mote thermoelectric energy harvester isdesigned for Tyndall mote to achieve power autonomy withlarge duty cycle

42 Thermoelectric Generator Module Design and Electri-cal Model The basic thermoelectric effect Seebeck effectdescribes a phenomenon that generates voltage differencewhen temperature gradient is applied across two seriesconnected dissimilar materials as shown in Figure 1 Theequivalent circuit for the thermoelectric voltage generationis illustrated in Figure 5

International Journal of Distributed Sensor Networks 5

Sleep mode

Sensing mode

Preambleand payload

Tx mode

ACK Rx mode

Initializationand CCA mode

0030

0025

0020

0015

0010

0005

0000

0 20 40 60 80 100 120 140 160 180Time (ms)

Curr

ent (

A)

minus20

Sleep Sl

ensing mode

PreamblePreambleand payloadp y

Tx mode

ACK Rx ACK Rxmode

alizationlizationCA modeCA mode

Figure 4 Measured current consumption of main operating modes(Tyndall WSN mote Tx transmitting power = 0 dBm)

minus

+

TEG

R

RL

V = 120572pminusn(TH minus TC)

Figure 5 Basic equivalent circuit of thermoelectric generator

The voltage generated from the single TEG pair is

119881119871= 120572119901minus119899Δ119879 times

119877119871

119877119871+ 119877

(2)

where 119881119871

is the load voltage and 120572119901minus119899

is the Seebeckcoefficient it is the difference between positive and negativeSeebeck coefficient in p-type and n-type materials 119877 and119877119871are the internal resistance of TEG and load resistance

respectively Δ119879 is the temperature difference between thetwo sides of TEGs For this single pair TEG the output power119875119871can be expressed as

119875119871= 119877119871sdot (

120572119901minus119899Δ119879

119877119871+ 119877

)

2

(3)

Single pair of thermocouples can only generate very limitedvoltage and power In most cases thermoelectric moduleconsists of large number of thermocouples to increase thevoltage and power output With increased number (normallyseveral hundreds) of thermo-couples the interconnectionsare mostly facilitated using copper contact In addition

Table 1 The power consumption of a Tyndall BEM mote at 33 V(sleepmode is 300 seconds that is 5-minutemeasurement interval)

Mode Symbol Power (mW) Time (mSec) Energy (mJ)Init amp CCA 119875IC 271 27 073Sensing 119875

119878

825 5 041RF Tx 119875Tx 732 42 307RF Rx 119875Rx 585 13 076Active modetotal 119875Act 571 87 497

Sleep 119875slp 0033 300000 99

ceramic substrates are used to physically support the TEGsThe thermoelectric generator layout and key parameters isshown in Figure 6

A typical TEG module [30] with precision machinedthermo-couples is shown in Figure 7

The Bi2Te3 based thermocouple has a thermal conductiv-ity 120582 at approximately 15W sdotmminus1Kminus1 The ceramic substratethermal conductivity 120582

119878is 120ndash150 times larger at 180W sdot

mminus1Kminus1 The thickness of substrate is around 03ndash05mmWith both upper and lower substrates the total thicknessis 06ndash1mm The impact on heat transfer from the thicksubstrates is no longer negligible Considering the substratethermal conductivity the actual temperature difference Δ1198791015840on the thermo-couple is less than the measured temperaturedifference Δ119879

The temperature difference across the thermo-couples isillustrated in Figure 8 The actual temperature difference onthe thermo-couples is expressed as

Δ1198791015840

Δ119879

=

1

1 + 2 (120582120582119878) sdot (119871119878119871)

(4)

where 119871 and 119871119878are the thermo-couple length and substrate

height respectively It is worth noting that (4) only considersthe heat transfer within thermoelectric materials but theheat transfers at material interfaces are not considered in thisequation

Based on this heat transfer model and temperaturedifference Δ1198791015840 the voltage output of 119873 pairs of thermo-couples module can be expressed as

119881 =

119873 sdot 120572119901minus119899sdot Δ1198791015840

119877 + 119877119871

sdot 119877119871 (5)

With increased number of thermo-couples the internalresistance 119877 also increases due to series connection oflarge number of thermo-couples and copper contacts Theresistance of the TEGmodulewith119873 pairs of thermo-couplesis shown as

119877 = 119873 sdot (

120588 sdot 119871

119860

+

2120588119862119871119862

119860119862

) (6)

where 120588 is thermo-couple electrical resistivity 120588119862

is the(copper) contact electrical resistivity119860 is the thermo-couplescross-section area and 119860

119862is the contact (vertical) cross-

section area Derived from (5) and (6) the analytic model

6 International Journal of Distributed Sensor Networks

Cross-section Area A

Length of thermoelement L

Contact cross-section

Thermocouple

Thermocouple

Temperature difference on thermocouple

Thermocouple

Height of ceramicsubstrate LS

middot middot middot

TEG module

(Coppergold) contacts

Substrate coldside temp TC

Substrate hotside temp TH

TH minus TC = ΔT

T998400H minus T998400

C = ΔT998400

Length of contact LC

cold side temp TC998400

hot side temp TH998400

area AC

NP

Number of thermocouplespairs N

Figure 6 Thermoelectric generator (TEG) module and thermo-couple structures

Copper contact

ThermocouplesTEG module

50mm

50

mm

095mm16mm

095mm

+

minus

Figure 7 Thermoelectric generator layout thermo-couple and thermoelectric module pictures

for maximum output power simulation at matched loadcondition (119877 = 119877

119871) is

119875max =119873 sdot (120572

119901minus119899Δ119879)

2

(120588119871119860 + 2 (120588119862119871119862119860119862)) sdot (1 + 2 (120582119871

119878120582119878119871))2

(7)

The thermoelectric module used to verify this modelis provided by thermonamic [30] This TEG module is acustom-designedmodule for low power generation It adoptsBi2Te3 thermo-couples with ZT figure of merit around 07 atroom temperature For each thermo-couple the cross-sectionarea is 091mm2 A thin layer of ceramic substrate and a layerof heat-conductive foamed carbon thermal pad are appliedon each side of the module The carbon thermal pad is usedto increase the heat transfer from the heat source to themodule The total thickness of the module is 34mm while

the thickness of ceramic layer and copper contacts is 09mmon each side The height of the thermo-couple is 16mmIn this basic unit the total number of thermo-couple pairsis 127 (16 times 8 array with 1 thermo-couple pair removed toaccommodate contact leads) This custom-designed modulecan bemanufactured into configurationswith 16 pairstimes 119873colwhere119873col is the number of column and119873col is a multiple of2The power factor for bothN type and P type thermo-coupleis approximately 36 120583WcmK2 The main parameters of thismodule are summarized in Table 2

A series of tests were conducted in order to verifythis analytic electrical simulation model The test setup isillustrated in Figure 9 The heat source is a temperaturecontrolled hot plate The TEG module is cooled by a passiveldquofinrdquo type heat sink A PicoTech ADC-1112 data acquisitionsystem is set up to monitor the temperature on the outer

International Journal of Distributed Sensor Networks 7

Table 2 Conventional machined thermoelectric module parame-ters summary

Symbol Definition Value119860 Thermocouple cross-section area 091mm2

119873 Number of thermocouples pairs 127119871 Length of thermo-couples 16mm119871119878

Height of ceramic substrate 09mm

119871119862

Length of contacts (betweenthermo-couples)

17mm times 2 (upperand lower substrate)

119860119862

Contacts cross-section area 01mm2

120582Thermocouple thermal

conductivity 15Wmminus1Kminus1

120582119878

Substrate thermal conductivity 180Wmminus1Kminus1

120588119862

Contacts electrical resistivity 16 times 10minus8Ωsdotm1205722

N120588N Power factormdashN type 36 120583WcmK2

1205722

P120588P Power factormdashP type 36 120583WcmK2

side of upperlower substrates by using PT-100 temperaturesensors The I-V characteristics of the TEG module aremeasured by oscilloscopes and multimeters

The room temperature during the experiments is approx-imately 20∘C Four hot side temperatures are applied 50∘C60∘C 70∘C and 80∘C The measured I-V characteristics ofthermonamic TEG module are shown in Figure 10 Themeasured power-voltage characteristics are illustrated inFigure 11

The load resistance tested in the characterization isbetween 1Ω and 1 KΩ The maximum power is obtainedwhen the load resistance is 85Ω The analytic model andthe measured results are compared in Table 3The simulationin the thermonamic TEG shows a high level of consistencywith the measured TEG electrical characteristics The powersimulation errors are less than 5 of themeasurement values

This analytic model is then realized in MatLab and usedto simulate the output voltage and power in this work It isalso used to calculate the internal resistance of TEG

43 Thermoelectric Energy Harvesting Power ManagementCircuit Design and Implementation From the device char-acterization one main issue discovered for thermoelectricenergy harvesting is that the voltage of the TEG output is oneorder of magnitude lower than theWSN operating voltage Avoltage step-up circuit is required to boost the 100ndash500mVinput voltage to 25ndash45 V output voltage This problem leadsto two types of proposed power management methods thefirst one uses ultra-low voltage boost converter with a largeconversion ratio transformer the second method uses lowvoltage charge pump and boost converter a two-stage step-up design

As introduced in [14] ultra-low voltage boost converterwith a transformer conversion ratio 1 100 can step up inputvoltage as low as 20mV to 20ndash45 VThemain concern in thistype of design is the inherent low conversion efficiency forhigh ratio voltage step up

Temperature differenceon substrate

Temperature differenceon TEG incl substrate

Temperature differenceon TEG excl substrate

Length (120583m)

Cold side

Thermoelement

Substrate thermalconductivity 120582S

Heat transfer along thermoelement

N

P

T998400H

TC

T998400C

ΔT

TH

120582

120582S

120582S

LSLS L

ΔT998400

Subs

trat

e

Subs

trat

e

Hot side

T (∘C)

Thermal conductivity 120582

Figure 8 Heat transfer within thermoelectric generator

THTC

T

RL

VL

IL Oscilloscope

Temperature controlledhot plate

Heat sink

Ceramic substrates

Copper contactsThermoelements

PT-1000 RTD

Temperature dataacquisition

Figure 9 Thermoelectric generator electrical characterizations testconfiguration

Mea

sure

d lo

ad v

olta

ge (V

)

0 5 10 15 20 25 30 35 4000

01

02

03

04

Measured current 50 (mA)

50∘C60∘C

70∘C80∘C

R = 16Ω

R = 8Ω

R = 4Ω

R = 2Ω

R = 1Ω

R = 12Ω

R = 20ΩR = 25ΩR = 50Ω

R = 200Ω

Pmax

R = 1KΩ

Figure 10 Measured I-V (current-voltage) characteristics of ther-monamic TEG module

8 International Journal of Distributed Sensor Networks

Table 3 Thermonamic TEG characterizations results and simula-tion results at matched load

Heat source temperature (∘C) 50 60 70 80Module temperature difference (∘C) 25 40 55 75Measured voltage (V) 0053 0084 0116 0158Simulated voltage (V) 0055 0088 0121 0167Measured power (mW) 0346 0886 1676 3117Simulated power (mW) 0353 0920 1722 3265Power simulation error () +18 +37 +27 +46

This work adopts the second type of DCDC conversionthe low voltage charge pump and boost converter two-stage step-up design The start-up DCDC converter is aSeiko Instruments S882Z-18 ultra-low voltage charge pumpwith a minimal start-up voltage at 025Vndash03V The mainDCDC converter is Texas Instruments TPS61020 with a09V minimal start-up voltage

In addition to the multiple stage power conversionimpedance matching is also considered in this designPrevious study confirmed that Bi2Te3 materials have smalltemperature coefficient of electrical conductivity within 50ndash100∘C temperature range [31] By revisiting the character-ization of TEG power-voltage characteristics illustrated inFigure 11 it can be clearly seen thatwhen temperature changesfrom 50∘C to 80∘C the matched load resistance (equals tointernal resistance) only changes less than 3 The internalresistance of TEG is mainly determined by the thermo-couples configuration When the configuration is finalizedthe TEG will have a near constant source resistance

Changes on the duty cycle of the boost converter caneffectively adjust the input resistance of the power manage-ment module By matching the input resistance of powermanagement circuit 119877IN and the TEG source resistance 119879TEGas shown in Figure 12 the energy transfer from TEG moduleto power management circuit is at maximum efficiency Asintroduced in the last section TEG source resistance canbe accurately simulated based on the aforementioned TEGanalytic electrical model

Another design issue related to the thermoelectric energyharvesting power management is the energy storage unitand its output power regulation In this work supercapacitoris used as the energy storage unit The porous structureof electrode material in supercapacitor effectively separatedby electrochemical property of the electrolyte instead ofthick physical dielectric layer ensures a large capacitance ofseveral Farads However one issues that has not been fullyaddressed in the previous literature is the leakage currentof supercapacitor Due to the small power consumption ofWSN mote the 10ndash100 120583A level leakage current is no longernegligible

To investigate the leakage current characteristics self-discharge tests were conducted on 4 different supercapaci-tors (Table 4) All the super-capacitors were precharged tothe same voltage level They were then isolated and thevoltage drops were monitored periodically by using a PicoTechnologies ADC-1112 data acquisition device The dataacquisition device has an output end impedance of 1MΩ

Load resistance (Ohm)

Out

put p

ower

(mW

)

1 10 100 1000

00

05

10

15

20

25

30

35

50∘C60∘C

70∘C80∘C

Figure 11 Measured power-voltage characteristics of thermonamicTEG module

RTEG

VTEG Rin Rout Cout RL

DCDC powermanagement

+

minus

Figure 12 TEG and power management circuit impedance match-ing

Table 4 Supercapacitor average leakage current and leakage corre-lation 120588leak

Mfr Capacity119862119904

(F)Average leakagecurrent 119868leak (120583A)

Leakage correlation120588leak = 119868leak119862119904119881

(120583AVsdotF)Maxwell 500 356 354EPCOS 410 197 241PanasonicGoldCap 022 202 450

AVXBestCap 010 038 191

during measurements and a 10MΩ impedance in idle modewhich effectively eliminated the current flow through theprobe The voltage drop is therefore only related to the self-discharge of the super-capacitors The 24-hour results of theself-discharge tests are presented in Figure 13

These results confirm that the self-discharge rates (SDR)of the super-capacitors are considerably higher than those ofrechargeable batteries (5ndash10 monthly self-discharge rate)The super-capacitors have SDR ranging from 45 to 15every 24 hoursTherefore without an intermittently available

International Journal of Distributed Sensor Networks 9

Table 5 Component selection for thermoelectric energy harvester

Component name Value119862IN 47120583F119862CP 1120583F119862SC (SuperCap) 25 F119862OUT 47120583F1198712

47120583H1198772

390 kΩ1198774

270 kΩ1198776

180 kΩ119862SU 10120583F1198621

47120583F1198622

10120583F1198711

22 120583H1198771

510 kΩ1198773

16MΩ1198775

1MΩ1198777

= 1198778 1MΩ

energy source every few days using the super-capacitorsalone as a long-term storage solution is not feasible

I-V characterizations were conducted to investigate theinput resistance 119877IN of power management module

Based on these measurement results a near linear cor-relation is found between leakage current 119868leak(119905) and theproduct of 119862

119904times 119881(119905) at time 119905 during the first 24 hours

of self-discharge The self-discharge mechanism in the earlyphase of this experiment is dominated by Faradic redox reac-tions generated ionic species concentration near the carbonsurfaces [32] The self-discharge due to this phenomenonsignificantly decreases after first 8ndash24 hours [33]The leakagecorrelation 120588leak is between 191 to 45 in various capacitorsFor large supercapacitor (gt5 F) the leakage current is of thesame order as the current consumption of WSN modulesClearly the leakage characteristics of the super-capacitor havea significant impact on the operation time of the WSN mote

Based on these design considerations the completeschematic of the thermoelectric energy harvester is shown inFigure 14

In order to extract most of the energy from thesuper-capacitor a buck-boost converter Texas InstrumentsTPS61220 is used for the output voltage regulator The inputvoltage range is between 07V and 55 V The output voltageis programmed to 33 V for Tyndall WSN mote The detailedcomponent selection and value are given in Table 5

In this work a super-capacitor array 119862SC with totalcapacitance of 25 F is usedThe fully charged capacitor has anenergy capacity of 3125 Joules or 868mWh Becouse designrequires larger capacity super-capacitor with higher capacityor thin film battery [34] can also be usedThe super-capacitorvoltage is monitored by Atmega1281 micro-controller ADCthrough R7R8 voltage divider

The results of the characterization are shown in Figure 15For input voltage (TEG output voltage) between 025V

and 07V the input resistance 119877IN is within a small range

Time (h)

Maxwell 50 FEPCOS 41 F

Panasonic 220 mFAVX 100 mF

0 5 10 15 20 2510

15

20

25

Volta

ge (V

)

Figure 13 Supercapacitors self-discharge test over 24 hours (Con-ditions room temperature 10∘C all samples have been fully chargedand discharged for 100 times before the test)

between 8Ω and 122Ω By applying the TEG analytic elec-trical characteristics model the output voltage and internalresistance of TEG module are simulated In the BEM appli-cations the target size of the TEG is approximately 50mm times50mm A 36 times 32 thermo-element (one of the two piles in athermo-couple) array with 54mmtimes 48mmdimensionmeetsthe form factor requirement In terms of thermo-couples theproposed TEG has 576 thermo-couples These 576 thermo-couples can be arranged in several different configurationsThe simulated TEG internal resistance 119877TEG in differentconfigurations is compared to input resistance 119877IN of thethermoelectric energy harvester power management modulein Figure 15

For 2 parallel-connected 288 thermo-couple pairs 119877TEGis 99Ω 119877TEG in this configuration matches with powermanagement module input resistance 119877IN with a less than20 error in the 025Vndash075V input voltage range Theimpedance match error is less than 10 when the inputvoltage is between 025V and 038V The verification of thismatched impedance is presented in the implementation andexperimental results section

Figure 16 shows the TEG module output voltage simu-lation based on the analytic model The simulation resultsshow that for 2 parallel-connected 288 thermo-couple pairsconfiguration the TEGoutput voltage is between 280mVand450mV when the temperature difference on the substratesis 3ndash5∘C This output voltage is higher than the minimalstart-up voltage of the charge pump and can start up thecharge pump and the other circuits in the powermanagementmodule

5 Thermoelectric Energy HarvesterImplementation and Experimental Results

Based on the proposed TEGmodule and power managementcircuit design the device manufacturing and assembling

10 International Journal of Distributed Sensor Networks

Table 6 TEG power management energy transfer

Stage Voltage(V)

Current(mA)

Power(mW)

Efficiency()

TEG output 025 170 408Charge pump Seiko S-882Z 148 0797 118 289Boost converter TI TPS61020 325 0338 1097 929Buckboost converter TI TPS61220 330 0312 1029 938End-to-end conversion 025 rarr 33 V 170 rarr 0312mA 408 rarr 1029mW 252

C in

L1

VinputCSU

Vin

SW

Out

ENGNDR3

R4

Vout

FBC1

R1 R2

LBIVM VSSCSC

L2

L

GND

InEN

R5

R6

Vout

FB

CoutC2

To microcontroller ADC(SuperCap voltage monitoring)

PS

R8 R7

To WSNmote VccFrom TEG

Input stage charge pump Input stage boost converter SuperCap Output stage buckboost converter

CCP

Vbat

+

minus

Texas instrumentsTPS61020

Boost converter

Seiko instruments

Charge pump

Texas instrumentsTPS61220

Buckboost converter

S-882Z18

CPout

Figure 14 Schematic of thermoelectric energy harvester

were conducted Four thermonamic TEG modules are usedto assemble the TEG module into 288 times 2 thermo-coupleconfiguration The complete prototype is shown in Figure 17The form factor of the thermoelectric energy harvesterpowered WSN mote is 6 cm (L) times 5 cm (W) times 7 cm (H)

The prototype design also took the thermal dissipation onthe PCB layer into considerationThe heat sink of the TEG isfastened to the PCB through a set of four long screws Thisconfiguration allows typical indoor airflow to further coolingof the heat sink and generates higher temperature difference

The viability of the TEG design and the application onWSNwere tested through a set of experimentsThe prototypewas placed on hotplate in various temperatures to test thestart-up performance continuous operation efficiency andenergy storage charge timeThe experiment result of the TEGstart up at 60∘C is shown in Figure 18

The charge pump S882-Z starts at 025ndash03V and thevoltage on storage capacitor of charge pump 119862SU and thevoltage on output capacitor of the charge pump 119862CP startto increase The charge pump is moving towards the targetvoltage 18 V When 119862CP voltage reaches 095V (09V min-imal start up voltage + 50mV hysteresis) the main boostconverter TI TPS61020 starts to operate and steps up the095V input voltage to charge the super-capacitorThe outputvoltage regulator TI TPS61220 starts to operate when super-capacitor voltage reaches 07 V and steps up the input voltageto 33 V output voltage

At 60∘C hot side temperature when the energy harvesterreaches thermal static state the output voltage of TEGmodule is measured at 025V as shown in Figure 18 Theopen circuit voltage of TEG before power regulation is at047V The output voltage 025V is close to the theoreti-cal maximum power point at half of open circuit voltage

0235V (047V times 05) This proves the concept of impedancematching between the TEG source resistance and the inputresistance of power management proposed in this work

Further tests were conducted to investigate the conver-sion efficiency of this proposed thermoelectric energy har-vester The I-V characterization of the TEG module outputpower and the step-up DCDC converter output power arepresented in Figure 19

Figure 20 shows the power-voltage characterizations ofTEG output power and the step-up DCDC converter outputpower The matched load TEG maximum output power ismeasured at 408mW The output power of step up DCDCconverter (charge pump and switching regulator) thermo-electric energy harvester is measured at 11mW

Table 6 summarizes the energy transfer in the powermanagement circuit when the hot side temperature is 60∘CWith a 289 conversion efficiency most of the power lossis due to the conversion energy loss in the ultra-low voltagecharge pump Both the boost converter (second stage of thestep-up DCDC conversion) and buckboost converter (out-put voltage regulator) achieved conversion efficiency higherthan 90 The system end-to-end conversion efficiency is252

It is worth noting that the super-capacitor(s) are not con-sidered in this conversion efficiency summary The thermo-electric energy harvester performance with super-capacitorsis characterized separately The super-capacitor chargingresults from 3 difference hot side temperatures are presentedin Figure 21

When charging the capacitive load the average chargingpower 119875avg during the complete charging phase (charged

International Journal of Distributed Sensor Networks 11

122

102 92 89 87 85 835 82 81 8 795 79

0225 0325 0425 0525 0625 0725

4

2

68

101214161820

40

Power management module

Voltage (V)

TEG internal

TEG config 576lowast1

TEG config 288lowast2

TEG config 192lowast3

RTEG = 40Ω

RTEG = 99Ω

RTEG = 41Ω

Input resistance Rin

resistance RTEG

RTEG

Rin

Resis

tanc

e (Ω

)

Figure 15 Thermoelectric energy harvester power management module input resistance and TEG internal resistance impedance matching

Number of series connected

thermoelements

[thermocouple pairs 2]

config

54

32

1 100 200 300400 500

600576

100

0

200

300

400

500

288lowast2

Load

mat

ched

out

put v

olta

ge (m

V)

Temperature difference (ΔT)

Figure 16 TEG module output voltage simulation

from 0V to the target voltage 119881target 33 V) can be calculatedas

119875avg =119862SC sdot 119881

2

target

2 sdot 119879chrg (8)

where 119862SC is the super-capacitor capacitance 119879chrg is thetotal charging time Based on the measured result shown inFigure 21 the average charging power is 095mW 21mWand435mW on 60∘C 70∘C and 80∘C hot surfaces respectively

The thermoelectric energy harvester performance withWSN mote is then evaluated based on WSN power con-sumption and energy storage leakage power consumptionWSN mote can be programmed with different time intervals120591 between two measurements (active mode) Since the activemode time 119879act is generally constant for certain application

(90mSec in this application) the measurement time interval120591 will determine the active mode duty cycle119863

119863 =

119879act119879act + 120591

times 100 (9)

Once the duty cycle 119863 is programmed the average powerconsumption can be calculated using (1) The WSN averagepower consumption and the energy storage leakage power aremeasured and illustrated in Figure 22 In order to continu-ously operate WSN mote from harvested power the averagepower consumption must be higher than the thermoelectricenergy harvester output power For each thermoelectricenergy harvester operating temperature the harvested powerhas a minimal WSN mote measurements time interval anda maximum active mode duty cycle The thermoelectricenergy harvester powered WSN operation limits are shownin Figure 22

12 International Journal of Distributed Sensor Networks

WSN mote

Heat sink

Low ESR SuperCap

Power management module

TEG module

6 cm

7 cm

5 cm

SuperCap array

Figure 17 Thermoelectric energy harverting and power manage-ment module prototype for BEM applications

005

115

225

335

445

0 50 100 150 200 250

Volta

ge (V

)

Time (s)

Charge pump

Boost converter

starts at 025Vndash03V

starts at 095V

Thermal static 025V

Power management output (test pin J12)TEG voltage output (test pin J1)Charge pump output (test pin 2)

Figure 18 Thermoelectric energy harvester start-up test results

005

115

225

335

4

001 01 1 10 100

Volta

ge (V

)

Current (mA)

Regulated voltagecurrent output

TEG voltagecurrent output

Figure 19 I-V characterization of the TEG module output powerand the step-up DCDC converter output power (hot side tempera-ture 60∘C)

005

115

225

335

445

001 01 1 10 100

Pow

er (m

W)

Current (mA)

TEG max power

Regulated power output

Figure 20 Power-voltage characterizations of the TEG moduleoutput power and step-up DCDC converter output power (hot sidetemperature 60∘C)

ΔT = 40∘C

ΔT = 55∘C

Heat source temperature T = 80∘CHeat source temperature T = 70∘CHeat source temperature T = 60∘C

ΔT = 75∘C

0 50 100 150 20000

05

10

15

20

25

30

35

Time (min)

Volta

ge (V

)

Figure 21 Thermoelectric energy harvester prototype charging

For 60∘C hot side temperature the minimal measure-ment time interval is 58 seconds that is when placingthe thermoelectric energy harvester on a 60∘C heat sourcethe generated power allows WSN mote to make a BEMmeasurement and transmit the data every 58 seconds Forhigher hot side temperatures theminimalmeasurement timeintervals are shorter Most BEM applications (light intensitytemperature relative humidity etc) require measurementstime interval between 1 and 10 minutes The thermoelectricenergy harvester proposed in this work can effectively pro-vide a ldquopower-autonomousrdquo power supply for BEM WSNmotes when thermal energy is available

When compared with recent work in the thermoelec-tric energy harvesting this bulk TEG based thermoelectricenergy harvester proposed in this work achieved the self-start capability suggested in [35] and can be deployed forlong-term operation The TEG generated power is measuredat 095mW with a 4∘C temperature difference in this workwhilst in [35] 02mW is generated with a 35∘C temperaturedifference In addition to the higher output power the bulk

International Journal of Distributed Sensor Networks 13

266

1734 84946

241101

053 028013 012

007005

000500100015002000250030003500400045005000

0001

001

01

1

10

01 1 10 100

WSN mote average powerconsumption and energy storage

leakage power (mW)

Duty cycle ()

Dut

y cy

cle D

()

Harvested thermoelectric power (mW)

Pow

er (m

W)

50

1205918012059170

12059160

T = 80∘C P = 435mWD = 75 12059180 = 13 sT = 70∘C P = 21mWD = 36 12059170 = 28 s

T = 60∘C P = 095mWD = 17 12059160 = 58 s

Time interval between two WSN mote measurements 120591 (s)

Figure 22 Thermoelectric energy harvester prototype chargingsupercapacitor

TEG can bemanufactured at lower cost than theMEMS TEGutilized in [35]

6 Conclusion

In this paper low temperature thermoelectric energy har-vester is investigated for the use of powering wirelesssensor network The main focus is on the energy con-version efficiency improvement from the prospective ofpower management and TEG design This study charac-terized the Bi2Te3 based thermoelectric generator with hotside temperature between 50∘C and 100∘C A TEG outputpowervoltageresistance analytic model is created to sim-ulate the device performance in low temperature energyharvesting This simulation model shows high level of con-sistency with the measured results with maximum error lessthan 5 in the experiments

Based on the characterizations and the simulation of theTEGs power management circuits with emphasis on lowvoltage step up are investigated in the second half of thispaper In order to obtain a regulated output voltage fromthe less than 05 V low input voltage multiple-stages voltageregulation is considered in this workThepowermanagementadopted the charge pumpswitching regulator two stagesdesign to obtain lower conversion ratio on each stage Thecharge pump starts up the voltage regulation when the TEGvoltage is higher than 250mV whilst the boost converterstarts up at 095V

It has been noticed that by adjusting the configura-tions of TEG the source resistance and output voltagecan be modified to match the power management circuitinput impedance Based on this method the TEG was re-configured based on the analytic simulation model with a99Ω source resistance whilst the power management circuitinput resistance is matched with the source resistance withless than 10 error (025Vndash038V input voltage) The deviceimplementation consists of two 160 thermo-couple pairsTEG modules configured in the parallel-connected layoutAt 60∘C the output voltage of TEG (input voltage of powermanagement) is measured at 025V close to the 0235Vmaximum power voltage (half of the open circuit voltage047V)

Although the charge pump runs in a relatively low con-version efficiency of 28ndash30 the high conversion efficiency(approximately 93) of the second stage voltage regulatorTPS61020 (from 10V to 33 V) and output voltage regulatorTPS61220 (approximately 94) allows the entire powermanagement circuit to operate at an overall efficiency closeto 25

Several experiments were conducted to measure theperformance of the TEG in different temperatures For testswith heat source temperature higher than 60∘C in roomtemperature environment the power generated regulatedand supplied from the prototype device is sufficient to operateWSNmote running in low duty cycles and reached the targetof power autonomous operation with thermoelectric energyharvesting

Acknowledgment

Thiswork was supported by ENIACJU project 2010 (270722-2) ERG ldquoEnergy for a Green Society from sustainable har-vesting to smart distribution Equipments materials designsolution and their applicationsrdquo

References

[1] L Perez-Lombard J Ortiz and C Pout ldquoA review on buildingsenergy consumption informationrdquo Energy and Buildings vol40 no 3 pp 394ndash398 2008

[2] Y-J Wen and A M Agogino ldquoPersonalized dynamic design ofnetworked lighting for energy-efficiency in open-plan officesrdquoEnergy and Buildings vol 43 no 8 pp 1919ndash1924 2011

[3] B Roisin M Bodart A Deneyer and P DrsquoHerdt ldquoLightingenergy savings in offices using different control systems andtheir real consumptionrdquo Energy and Buildings vol 40 no 4 pp514ndash523 2008

[4] E Mills ldquoWhy were here the 230 billion global lighting energybillrdquo in Proceedings from the 5th European Conference on EnergyEfficient Lighting Citeseer 2002

[5] Y-J Wen and A M Agogino ldquoControl of wireless-networkedlighting in open-plan officesrdquo Lighting Research and Technologyvol 43 no 2 pp 235ndash248 2011

[6] G G Mueller I A Lys K J Dowling et al ldquoWireless lightingcontrol methods and apparatusrdquo US Patent 7659674 2010

[7] H Park J Burke and M B Srivastava ldquoIntelligent lightingcontrol using wireless sensor networks for media productionrdquoKSII Transactions on Internet and Information Systems vol 3no 5 pp 423ndash443 2009

[8] X Cao J Chen Y Xiao and Y Sun ldquoBuilding-environmentcontrol with wireless sensor and actuator networks centralizedversus distributedrdquo IEEE Transactions on Industrial Electronicsvol 57 no 11 pp 3596ndash3605 2010

[9] W Wang T OrsquoDonnell N Wang M Hayes B OrsquoFlynn andC OrsquoMathuna ldquoDesign considerations of sub-mw indoor lightenergy harvesting for wireless sensor systemsrdquoACM Journal onEmerging Technologies inComputing Systems vol 6 no 2 article6 2010

[10] W Wang N Wang M Hayes B OrsquoFlynn and C OrsquoMathunaldquoPower management for sub-mw energy harvester with adap-tive hybrid energy storagerdquo Journal of Intelligent MaterialSystems and Structures 2012

14 International Journal of Distributed Sensor Networks

[11] R H Bhuiyan R A Dougal and M Ali ldquoA miniature energyharvesting device for wireless sensors in electric power systemrdquoIEEE Sensors Journal vol 10 no 7 pp 1249ndash1258 2010

[12] S P Beeby M J Tudor and N M White ldquoEnergy harvestingvibration sources for microsystems applicationsrdquoMeasurementScience and Technology vol 17 no 12 article R175 2006

[13] G J Snyder ldquoThermoelectric energy harvestingrdquo in EnergyHarvesting Technologies pp 325ndash336 2009

[14] E J Carlson K Strunz and B P Otis ldquoA 20 mV input boostconverterwith efficient digital control for thermoelectric energyharvestingrdquo IEEE Journal of Solid-State Circuits vol 45 no 4pp 741ndash750 2010

[15] G Sebald D Guyomar and A Agbossou ldquoOn thermoelectricand pyroelectric energy harvestingrdquo SmartMaterials and Struc-tures vol 18 no 12 Article ID 125006 2009

[16] J P Carmo LM Goncalves and J H Correia ldquoThermoelectricmicroconverter for energy harvesting systemsrdquo IEEE Transac-tions on Industrial Electronics vol 57 no 3 pp 861ndash867 2010

[17] H Bottner D G Ebling A Jacquot J Konig L Kirste andJ Schmidt ldquoStructural and mechanical properties of SparkPlasma sintered n- and p-type bismuth telluride alloysrdquo PhysicaStatus Solidi vol 1 no 6 pp 235ndash237 2007

[18] A I Hochbaum R Chen R D Delgado et al ldquoEnhanced ther-moelectric performance of rough silicon nanowiresrdquo Naturevol 451 no 7175 pp 163ndash167 2008

[19] M Strasser R Aigner C Lauterbach T F Sturm M Franoschand G K M Wachutka ldquoMicromachined CMOS thermoelec-tric generators as on-chip power supplyrdquo Sensors and Actuatorsvol 114 no 2-3 pp 362ndash370 2004

[20] D Samson M Kluge T Becker and U Schmid ldquoWirelesssensor node powered by aircraft specific thermoelectric energyharvestingrdquo Sensors and Actuators vol 172 no 1 pp 240ndash2442011

[21] C Lu S P Park V Raghunathan and K Roy ldquoAnalysis anddesign of ultra low power thermoelectric energy harvestingsystemsrdquo in Proceedings from the 16th ACMIEEE InternationalSymposium on Low-Power Electronics and Design (ISLPED rsquo10)pp 183ndash188 IEEE August 2010

[22] V Leonov ldquoHuman machine and thermoelectric energy scav-enging for wearable devicesrdquo ISRN Renewable Energy vol 2011Article ID 785380 11 pages 2011

[23] W Wang R OrsquoKeeffe N Wang M Hayes B OrsquoFlynn andC OrsquoMathuna ldquoPractical wireless sensor networks power con-sumption metrics for building energy management applica-tionsrdquo in Proceedings of the 23rd European Conference ForumBauinformatik 2011 Construction Informatics Cork IrelandSeptember 2011

[24] G Nolas J Sharp and H Goldsmid Thermoelectrics BasicPrinciples and New Materials Developments vol 45 Springer2001

[25] S Dalola M Ferrari V Ferrari M Guizzetti DMarioli and ATaroni ldquoCharacterization of thermoelectricmodules for power-ing autonomous sensorsrdquo IEEETransactions on Instrumentationand Measurement vol 58 no 1 pp 99ndash107 2009

[26] YK Ramadass andA P Chandrakasan ldquoAbattery-less thermo-electric energy harvesting interface circuit with 35 mV startupvoltagerdquo IEEE Journal of Solid-State Circuits vol 46 no 1 pp333ndash341 2011

[27] L Datasheet Linear Technology Corporation Milpitas CalifUSA 2010

[28] M Pinuela D Yates S Lucyszyn and P Mitcheson ldquoCurrentstate of research at imperial college london in rf harvesting andinductive power transferrdquo in Proceedings of PowerMEMS pp41ndash44 2010

[29] K Menzel D Pesch B OrsquoFlynn M Keane and C OrsquoMathunaldquoTowards a wireless sensor platform for energy efficient build-ing operationrdquo Tsinghua Science and Technology vol 13 no 1pp 381ndash386 2008

[30] Thermonamic ldquoModules for power generationrdquo 2009 httpwwwthermonamiccom

[31] H J Goldsmid ldquoThe electrical conductivity and thermoelectricpower of bismuth telluriderdquo Proceedings of the Physical Societyvol 71 no 4 article 312 pp 633ndash646 1958

[32] BW Ricketts and C Ton-That ldquoSelf-discharge of carbon-basedsupercapacitors with organic electrolytesrdquo Journal of PowerSources vol 89 no 1 pp 64ndash69 2000

[33] M J Guan and W H Liao ldquoCharacteristics of energy storagedevices in piezoelectric energy harvesting systemsrdquo Journal ofIntelligentMaterial Systems and Structures vol 19 no 6 pp 671ndash680 2008

[34] J B Bates N J Dudney BNeudecker AUeda andCD EvansldquoThin-film lithium and lithium-ion batteriesrdquo Solid State Ionicsvol 135 no 1ndash4 pp 33ndash45 2000

[35] Q Huang C Lu M Shaurette and R Cox ldquoEnvironmentalthermal energy scavenging poweredwireless sensor network forbuilding monitoringrdquo in Proceedings of the 28th InternationalSymposiumonAutomation andRobotics inConstruction (ISARCrsquo11) pp 1376ndash1380 Seoul Korea June-July 2011

International Journal of

AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

RoboticsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Active and Passive Electronic Components

Control Scienceand Engineering

Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

International Journal of

RotatingMachinery

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporation httpwwwhindawicom

Journal ofEngineeringVolume 2014

Submit your manuscripts athttpwwwhindawicom

VLSI Design

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Shock and Vibration

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Civil EngineeringAdvances in

Acoustics and VibrationAdvances in

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Electrical and Computer Engineering

Journal of

Advances inOptoElectronics

Hindawi Publishing Corporation httpwwwhindawicom

Volume 2014

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

SensorsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Chemical EngineeringInternational Journal of Antennas and

Propagation

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Navigation and Observation

International Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

DistributedSensor Networks

International Journal of