Citation
Electrical energy harvesting from microbial fuel cell

Material Information

Title:
Electrical energy harvesting from microbial fuel cell
Creator:
Alaraj, Muhannad A. ( author )
Place of Publication:
Denver, CO
Publisher:
University of Colorado Denver
Publication Date:
Language:
English
Physical Description:
1 electronic file (10 pages) : ;

Subjects

Subjects / Keywords:
Microbial fuel cells ( lcsh )
Electric current converters ( lcsh )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Review:
Microbial fuel cells (MFC) generate electricity using bacteria and they have been investigated as promising renewable energy source. Although they generate low power compared to other fuel cells, they have been used to replace batteries in powering remote wireless sensors. Energy harvesting circuits are used to collect energy from MFCs, therefore, they need to be able to store the harvested energy and be efficient in order to use most of the MFC available power. Power electronics converters can be used to harvest MFC energy efficiency and the harvested energy is typically stored in a capacitor or a battery. Power converters can be also used to control the operating point of the MFC reactor to harvest the maximum available power at the operating point, maximum power point (MPP), that gives the maximum available power. This dissertation is focused on improving the use of power electronics converters in MFC energy harvesting in terms of reliability and essence. It also investigates the eect of using power electronics converters on the reactor performance and conditions. The reliability of the energy harvesting circuit was increased by proposing an algorithm that can detect and avoid voltage overshoot in MFCs. Voltage overshoot happens during energy harvesting, where the reactor terminal voltage collabses because of high currents, which significantly aect MFC energy harvesting. The proposed algorithm is based on extremum seeking (ES) algorithm, where it can track MPP in the normal conditions and can detect voltage overshoot once it happens. Then, the algorithm tries to nd an operating point that is far from the voltage overshoot region. Energy harvesting from MFC using power electronics converters imposes current ripple on MFC reactor because of their switching behavior, and such current ripple can have an eect on other fuel cells such as PEMFC. The experimental results showed that there is no significant eect of power electronics converters current ripple on the MFC reactor performance in terms of voltage, power, and longevity. It was also shown that the conditions of the reactor such as pH, dissolved oxygen (DO), electrical conductivity (EC), and oxidation reduction potential (ORP) are not affected by that current ripple. Finally, a self-powered energy harvesting system (EHS) is proposed. That system has a microcontroller that can be used to track MPP using a proposed power estimation method that saves substantial power that is normally consumed to measure the power. EHS is designed to use only fraction of the MFC harvested power in order to make a self-sustainable system. The experimental results show that the essence of the proposed system is up to 59.4% with a microcontroller power consumption of 8.67W with a 119W MFC reactor.
Thesis:
Thesis (Ph.D.)--University of Colorado Denver
Bibliography:
Includes bibliographical references.
System Details:
System requirements: Adobe Reader.
Statement of Responsibility:
by Muhannad A. Alaraj.

Record Information

Source Institution:
University of Colorado Denver
Holding Location:
Auraria Library
Rights Management:
Copyright Muhannad A. Alaraj. Permission granted to University of Colorado Denver to digitize and display this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Resource Identifier:
on10859 ( NOTIS )
1085902990 ( OCLC )
on1085902990

Downloads

This item is only available as the following downloads:


Full Text

PAGE 1

ELECTRICALENERGYHARVESTINGFROMMICROBIALFUELCELL by MUHANNADAALARAJ M.S.,UniversityofColoradoDenver,2013 Adissertationsubmittedtothe FacultyoftheGraduateSchoolofthe UniversityofColoradoinpartialfulllment oftherequirementsforthedegreeof DoctorofPhilosophy EngineeringandAppliedScienceProgram 2018

PAGE 2

ThisdissertationfortheDoctorofPhilosophydegreeby MuhannadAAlaraj hasbeenapprovedforthe EngineeringandAppliedScienceProgram by MilojeRadenkovic,Chair Jae-DoPark,Advisor SatadruDey TimbereleyRoane Jung-JaeLee Date:May12,2018 ii

PAGE 3

Alaraj,MuhannadAPh.D.,EngineeringandAppliedScienceProgram ElectricalEnergyHarvestingfromMicrobialFuelCell DissertationdirectedbyAssociateProfessorJae-DoPark ABSTRACT MicrobialfuelcellsMFCgenerateelectricityusingbacteriaandtheyhavebeen investigatedaspromisingrenewableenergysource.Althoughtheygeneratelow powercomparedtootherfuelcells,theyhavebeenusedtoreplacebatteriesin poweringremotewirelesssensors.Energyharvestingcircuitsareusedtocollect energyfromMFCs,therefore,theyneedtobeabletostoretheharvestedenergy andbeecientinordertousemostoftheMFCavailablepower.Powerelectronics converterscanbeusedtoharvestMFCenergyecientlyandtheharvestedenergy istypicallystoredinacapacitororabattery.Powerconverterscanbealsousedto controltheoperatingpointoftheMFCreactortoharvestthemaximumavailable powerattheoperatingpoint,maximumpowerpointMPP,thatgivesthe maximumavailablepower.Thisdissertationisfocusedonimprovingtheuseof powerelectronicsconvertersinMFCenergyharvestingintermsofreliabilityand eciency.Italsoinvestigatestheeectofusingpowerelectronicsconvertersonthe reactorperformanceandconditions.Thereliabilityoftheenergyharvestingcircuit wasincreasedbyproposinganalgorithmthatcandetectandavoidvoltage overshootinMFCs.Voltageovershoothappensduringenergyharvesting,wherethe reactorterminalvoltagecollabsesbecauseofhighcurrents,whichsignicantlyaect MFCenergyharvesting.Theproposedalgorithmisbasedonextremumseeking ESalgorithm,whereitcantrackMPPinthenormalconditionsandcandetect voltageovershootonceithappens.Then,thealgorithmtriestondanoperating pointthatisfarfromthevoltageovershootregion.EnergyharvestingfromMFC iii

PAGE 4

usingpowerelectronicsconvertersimposescurrentrippleonMFCreactorbecause oftheirswitchingbehavior,andsuchcurrentripplecanhaveaneectonotherfuel cellssuchasPEMFC.Theexperimentalresultsshowedthatthereisnosignicant eectofpowerelectronicsconverterscurrentrippleontheMFCreactor performanceintermsofvoltage,power,andlongevity.Itwasalsoshownthatthe conditionsofthereactorsuchaspH,dissolvedoxygenDO,electricalconductivity EC,andoxidationreductionpotentialORParenotaectedbythatcurrent ripple.Finally,aself-poweredenergyharvestingsystemEHSisproposed.That systemhasamicrocontrollerthatcanbeusedtotrackMPPusingaproposed powerestimationmethodthatsavessubstantialpowerthatisnormallyconsumedto measurethepower.EHSisdesignedtouseonlyfractionoftheMFCharvested powerinordertomakeaself-sustainablesystem.Theexperimentalresultsshow thattheeciencyoftheproposedsystemisupto59.4%withamicrocontroller powerconsumptionof8.67 Wwitha119 WMFCreactor. Theformandcontentofthisabstractareapproved.Irecommenditspublication. Approved:Jae-DoPark iv

PAGE 5

DEDICATION Thisdissertationisdedicatedtomymotherandfather,whorsttaughtthe valueofeducation. Ialsodedicatethisdissertationtomywifewhosupportedmetopursuemy dreamsandnishmydissertation. v

PAGE 6

ACKNOWLEDGMENTS Thisthesiswouldnotbepossiblewithouttheguidanceandsupportofmy advisor,Dr.Park.Thankyousomuchforallyoudo.Icanremembertimeswhen I'dfranticallyemailyouaquestionandtenminuteslaterI'dhaveananswer.Iam sofortunatetobeabletohaveyourhelpandjustwanttoletyouknowthatI appreciateallyoudo! AlsoitwouldnothavebeenpossiblewithoutthegeneroussupportofQassim University. vi

PAGE 7

TABLEOFCONTENTS TABLES......................................xi FIGURES......................................xii CHAPTER I.INTRODUCTION...............................1 1.1MicrobialFuelCells...........................2 1.1.1MFCElectricalCharacteristics.................3 1.1.1.1 EquivalentCircuit ....................3 1.1.1.2 PolarizationCurve ....................4 1.2PassiveEnergyHarvesting.......................6 1.2.1Resistors.............................6 1.2.2Supercapacitors..........................8 1.2.3ChargePumps..........................8 1.3ActiveEnergyHarvesting:PowerEelctronicsConverters.......9 1.3.1BoostConverter.........................9 1.3.1.1 OperationofBoostConverter ..............9 1.3.1.2 ControlofBoostConverter ...............12 1.3.1.3 ParametersDesign ....................12 1.4MaximumPowerPointTrackingMPPT...............13 1.4.1MaximumPowerPointTrackingAlgorithms..........14 1.5MFCVoltageOvershootandPowerGeneration............14 1.6EectofPowerElectronicsConvertersonMFCPowerGeneration.15 II.INTELLIGENTENERGYHARVESTINGSCHEMEFORMICROBIAL FUELCELLS:MAXIMUMPOWERPOINTTRACKINGANDVOLTAGEOVERSHOOTAVOIDANCE......................16 2.1Introduction...............................16 2.1.1MaximumPowerPointTrackingMPPT...........16 vii

PAGE 8

2.1.2VoltageOvershootPhenomenon.................17 2.2ExtremumSeekingAlgorithmESA..................18 2.2.1MaximumPowerPointTrackingMPPTUsingESA.....21 2.2.2ESAComputerSimulation...................23 2.2.2.1 SimulationModel .....................23 2.2.2.2 SimulationResults ....................26 2.3VoltageOvershootAvoidanceAlgorithm................27 2.4ExperimentSetup............................30 2.4.1HardwareSetup.........................30 2.4.1.1 BoostConverter ......................31 2.4.1.2 Microcontroller ......................31 2.4.1.3 VoltageandCurrentMeasurements ...........32 2.4.1.4 OverallSystem ......................35 2.4.2ExperimentalResults......................35 2.4.2.1 ESMPPT .........................36 2.4.2.2 VoltageOvershootAvoidance ...............38 III.EFFECTOFPOWERSHAPEONENERGYEXTRACTIONFROMMICROBIALFUELCELL............................44 3.1Introduction...............................44 3.1.1PowerElectronicsConverterCurrentRipple..........44 3.1.2Physical-ChemicalParameters..................44 3.1.3CurrentRippleTypes......................45 3.2ExperimentSetup............................45 3.2.1Physical-ChemicalSensorProbes................48 3.2.2CurrentRippleSimulators....................48 3.2.3Microcontroller..........................53 3.3PowerandEnergyCalculations.....................53 viii

PAGE 9

3.4ExperimentalResults..........................56 3.4.1PowerandEnergy........................57 3.4.2Physical-ChemicalParameters..................63 3.5Conclusion................................66 IV.NETPOWERPOSITIVEMAXIMUMPOWERPOINTTRACKING POWERMANAGEMENTSYSTEMFORMICROBIALFUELCELL..67 4.1Introduction...............................67 4.2MicrobialFuelCells...........................69 4.2.1EquivalentCircuit........................69 4.2.2Steady-StateAnalysis......................72 4.2.3TransientAnalysis........................73 4.3MaximumPowerPointTracking....................74 4.3.1ProposedAlgorithm.......................75 4.3.2CurrentEstimation........................75 4.3.3PowerCalculationandMPPT..................78 4.4EnergyHarvestingSystem.......................79 4.4.1BoostConverter.........................80 4.4.2MicrocontrollerandControlPowerConsumption.......81 4.4.2.1 OperatingMode ......................81 4.4.2.2 SamplingTime ......................82 4.4.2.3 SystemClockFrequency .................82 4.4.2.4 SupplyVoltage ......................82 4.4.2.5 Built-InAnalogCircuit ..................83 4.4.2.6 ReferenceVoltage .....................83 4.5ExperimentalValidation........................84 4.5.1MFCReactor...........................84 4.5.2MPPT...............................85 ix

PAGE 10

4.5.3Eciency.............................86 4.6Conclusion................................87 V.SUMMARYANDFUTUREWORK.....................88 5.1Summary.................................88 5.2FutureWork...............................90 REFERENCES...................................91 x

PAGE 11

TABLES TABLE 2.1Perturbationanddemodulationsignalsparameters.............21 4.1ResultsofMFCtests.............................72 4.2Resultsofidentiedequivalentcircuitparameters..............72 xi

PAGE 12

FIGURES FIGURE 1.1MFCconceptualdiagram...........................3 1.2MFCequivalentcircuits............................5 1.3Polarizationcurveexample..........................6 1.4Schematicdiagramofboostconverter....................10 1.5Inductorvoltageandcurrentwaveforms...................11 2.1Extremumseekingblockdiagram.......................19 2.2GraphicalrepresentationofextremumseekingMPPT...........23 2.3Simulinkmodelscreenshots:aOverallsystembESalgorithmblock.24 2.4MFCsimpleequivalentcircuit........................25 2.5ESMPPTsimulationresults.........................26 2.6MFCpolarizationcurvewithavoltageovershoot..............28 2.7MFCpolarizationcurvewithavoltageovershootthatrecovers......28 2.8Flowchartoftheproposedvoltageavoidancealgorithm..........29 2.9Experimrntsetuppicture...........................31 2.10Measurementcircuitsschematicdiagrams:aVoltagebCurrent....33 2.11Theoverallsystemschematicdiagram....................35 2.12ESMPPTexperimentalresults:voltageandpower.............37 2.13ESMPPTexperimentalresults:boostconverter'sdutycycle.......38 2.14Voltageavoidancealgorithmexperimentalresult:MFCvoltage......39 2.15Voltageavoidancealgorithmexperimentalresult:MFCcurrent......40 2.16Voltageavoidancealgorithmexperimentalresult:MFCpower.......41 2.17Voltageavoidancealgorithmexperimentalresult:MFCenergy.......42 2.18Voltageavoidancealgorithmexperimentalresult:dutycycle........42 3.1MFCreactorpictureincludingphysical-chemicalsenors..........46 3.2Microcontrollerandcircuitsetup......................47 xii

PAGE 13

3.3Continuouscurrentcircuitschematicdiagram................49 3.4Square-shapecurrentripplesimulatorcircuitschematicdiagram.....50 3.5Triangular-shapecurrentripplesimulatorcircuitschematicdiagram...51 3.6Expectedcurrentandvoltagewaveforms...................54 3.7Experimentalwaveformsofcontinuouscurrentextraction.........57 3.8Experimentalwaveformsoftriangular-shapecurrentextraction......58 3.9Experimentalwaveformsofsqaure-shapecurrentextraction........59 3.10MFCvoltageexperimentalresults......................60 3.11MFCpowerexperimentalresults.......................61 3.12Harvestedenergyexperimentalresults....................61 3.13AnodechamberpHexperimentalresult...................63 3.14Anodechamberlectricalconductivityexperimentalresult.........64 3.15Anodechamberoxidationreductionpotentialexperimentalresult.....65 3.16Anodechamberdissolevedoxygenexperimentalresult...........65 4.1MFCconceptualdiagram...........................69 4.2MFCelectricalequivalentcircuit[26].....................71 4.3FlowchartoftheproposedMPPTalgorithm................76 4.4Graphicalrepresentationof V t and V c measurementprocess: V t ismeasuredjustbeforeopeningthecircuit, V C ismeasuredrightafterthecircuit isopened, t timerequiredtomeasure V C ,and T s isthesamplingtime.77 4.5Thecalculatedcapacitorvoltagewhentheterminalvoltagewasincreased from220mVto240mVbasedontheequivalentcircuitmodel.......79 4.6SchematicdiagramoftheoverallEHSsystem................80 4.7Experimentsetuppicture:theMFCreactorandtheusedcircuit.....84 4.8MPPTresults:circlesrepresenttheoperatingpointsofthealgorithm..86 xiii

PAGE 14

CHAPTERI INTRODUCTION Ultralowpowero-the-gridsystemssuchaswirelesssensors,cameras,and wearableshavebeenwidelyusedformonitoring,control,ormeasurementsinmany dierentapplications.Thosesystemshavebeenpoweredwithbatteries.However, thehighinitialandmaintenancecost[1]andthesafetyconcerninsome applications[2]arebigdisadvantagesofusingbatteries.Othernaturalsourcessuch aslight,heat,andvibrationshavebeenconsideredtopowerthosesystems. Collectingelectricalpowerfromthesenaturalenergyresourcesrequiresusingenergy harvestingsystemsthataredesignedbasedonthecharacteristicsofeachsourceto ecientlyharvesttheavailableenergyanddeliveritattherequiredvoltage.Recent studieshavefocusedonimprovingenergyharvestingsystemsforvariousenergy resources[2{6]. ThermoelectricgeneratorsTEGareoneofthewidelyinvestigatedrenewable energysources[3,4].Ithasbeenshownthatsmallamountofwasteheatcanbe usedtopowerwirelesssensornodeswithhighreliabilityandpredictability[3].Also microsacleTEGwasusedtopowerwearabledeviceswithapowerof10mW[4]. Anotherexampleofambientrenewablepowersourcethathasbeenconsideredisthe piezoelectricpower[2],wheretheyusemovementsofhumanbodyorstructureto supplytherequiredpower.Vibrationenergyhavebeenusedasaelectricalpower sourceforportabledevices[5,6],wherein[6],thevibrationsthatresultfrom vehiclespassingovertracowratesensor,areusedconvertedtoelectricalpower andusedtoenergizethatsensor. MicrobialfuelcellsMFCshavebeenalsousedasapowersourceforremote underwaterwirelesssensors[7{12].Eventhoughpowergenerationisatthelower endofthefuelcellelectricalpowergeneration,where6.9 W=m 2 isthehighestpower densitythatwasreported[13].Studieshaveproposedvariousenergyharvesting 1

PAGE 15

systemsforMFCs,yetthatareaneedsmoreresearchtoincreasethesystem reliability,eciency,andpredictability. 1.1MicrobialFuelCells MicrobialfuelcellsMFCusesanaerobicbacteriasuchas Shewanella and Geobacter togenerateelectricitywhileconsumingorganicmatter.Electricalcurrent generationbybacteriawasinitiallyobservedin1911[14],however,theadvancesin thatareawereverysmallevenafter50yearsofthatobservation[15].Thefocuson MFCselectricalgenerationstartedtoincreaseinthe1990swhenfuelcellsingeneral becamemoreinterestingforscientists[16].OneofthedicultiesthatMFCsfaced intheearlydesignsisthattheyneededchemicalmediatorsorelectronshuttlesto createapathfortheelectronsbetweenthebacteriaandtheelectrodes,itwas resolvedwhenthemediator-lessreactorsweredeveloped,whichisconsideredtobea breakthroughinMFCsresearch[17]. InanMFCreactor,anaerobicbacteriaconsumetheorganicsubstratesinsidethe anodechamberofthereactorandtransportextracellularelectronstotheanode electrode.Connectingaloadbetweentheanodeandcathodeelectrodesclosesthe circuitandallowstheelectronstobetransferredtothecathodethroughthe externalloadandthecurrentstartsowing.Asaresult,adirectcurrentDC powerisdeliveredtotheload[18{20]. AconceptualdiagramofMFCisshowningure1.1.Itshowsaschematic diagramforasinglechamberMFCreactor,whereithasonlyonechamber,thatis theanodechamber,wherethebacterialive.Theanodeelectrodeisplacedinside theanodechambertocollectelectronsthatisgeneratedbythebacteria.Theanode chamberneedstobeoxygenfreebecauseofusinganaerobicbacteriasincethe presenceofoxygeninhibitselectricitygeneration.However,theprotonsinsidethe anodechamberneedtobetransferredtothecathodewheretheycombinewith electronscomingfromexternalcircuitandoxygentoproducewater.Andthiscan 2

PAGE 16

Figure1.1:MFCconceptualdiagram. beachievedbyplacinganionexchangemembranethatkeepstheoxygenoutofthe anodechamberandatthesametimeallowschargetransfertothecathode[17].The cathodeelectrodeisplacedsuchthatittouchestheionexchangemembranetoallow thechargetransfer.Then,thetwoelectrodesareexternallyconnectthroughthe load,whichcreatesanexternalpathforelectrons,andtheyaretransferredfromthe anodeelectrodetothecathodeelectrodecreatingacurrentthatowsthroughthe load. 1.1.1MFCElectricalCharacteristics 1.1.1.1 EquivalentCircuit TheimpedanceofMFCconsistsofcontributionsofactivationkinetics,charge transfer,doublelayertransfercharging,masstransporteect,andionic resistance[21],whichresultsinaninternalresistanceandadoublelayercapacitance. TheinternalresistanceofMFCcanbedividedintotwoparts:ohmicresistanceand polarizationresistance.Theohmicresistanceistheresistanceoftheliquidsinside thereactorplustheresistanceoftheelectrodes[22],andthepolarizationresistance 3

PAGE 17

includeschargetransferresistanceanddiusionresistance[21].Theohmic resistancecontributiontotheinternalresistanceismorethanthechargetransfer anddiusionresistances[21,23],andsincethatohmicresistanceismostlythe resistanceoftheliquidsinsidethereactor,thesizingofthereactorhaveaneecton theinternalresistanceofMFC.Forexample,usingsinglechamberMFC,whichhas lessliquidspace,resultsinincreasingtheoutputpower6timescomparedtotwo chambersMFCandthatisduetothedecreasedinternalresistance[24]. VariousequivalentcircuitshavebeenusedtomodelMFCs[21,25{27].Figure 1.2showsthreedierentequivalentcircuitsforMFC.Theequivalentcircuitsin gures1.2aand1.2baremoredetailedequivalentcircuitthantheequivalentcircuit ingure1.2cbecausetheyhavethecapacitance,whichrepresentsthedynamicof theMFCelectricalbehavior,wherethecapacitorvoltageneedstimetochangefrom oneleveltoanother.However,insteadystate,thesimpliedmodelingure1.2c canbeusedtomodeltheMFCbecausethecapacitanceisconsideredtobeopen circuitindirectcurrentDCpowersystemssteadystate. 1.1.1.2 PolarizationCurve Thepolarizationcurveisdenedastherelationshipbetweenvoltageandcurrent offuelcells.Thisrelationshipisimportantbecauseitshowsthedierentoperating pointsoffuelcells.ForanMFC,atypicalpolarizationcurveisshowningure1.3. Togetthispolarizationcurve,dierentresistorsconnectedbetweentheterminalsof MFC,andthevoltagewasmeasuredandrecordedforeachresistance.Thenthe currentwascalculatedusingOhm'slaw I MFC = V MFC R .Thepolarizationcurveof MFCcanbeexplainedusingthesimpleequivalentcircuitingure1.2c.Inthe beginning,theopencircuitvoltageOCVappearsattheterminalsoftheMFC whentheterminalsareopen,meaningthatnocurrentisowingoutoftheMFC. Whenaloadisconnected,thecurrentstartstoowtotheloadthroughtheinternal resistance.Thiswillcauseavoltagedropacrosstheinternalresistanceandthe 4

PAGE 18

a b c Figure1.2:MFCequivalentcircuits. 5

PAGE 19

Figure1.3:Polarizationcurveexample. terminalvoltagewouldbelowerthanOCV.Increasingtheloadcurrentreducesthe terminalvoltagebecausehigherloadcurrentmeanshighervoltagedropacrossthe internalresistance,whichmeanslowerterminalvoltage. Thepowerforeachoperatingpointvoltage/currentinthepolarizationcurve canbecalculatedbymultiplyingthevoltageandcurrentateachpoint.The resultedpowerisshowningure1.3anditisclearthatthepowerhasaquadratic shape,wheretheoutputpowercanbemaximizedatspecicoperatingpoint.This operatingpointisknownasthemaximumpowerpointMPP. 1.2PassiveEnergyHarvesting DierentpassiveenergyharvestingcircuitshavebeenusedtocollectMFC power.Thesepassiveenergyharvesterscanbedividedintothreecategories: resistor-based[28{30],supercapacitor-based[7{9],andchargepump-based[12,31]. 1.2.1Resistors UsingexternalresistanceattheterminalsoftheMFChasbeenwidely used[28{30],anditisconsideredtobethemostbasicenergyharvestingmethod. 6

PAGE 20

WhenaresistorisconnectedtotheterminalsofMFC,currentstartsowingoutof thereactorandinthiscase,theterminalvoltageofthereactorwouldbe V MFC = I MFC R ext .1 where V MFC istheMFCterminalvoltage, I istheMFCcurrent,and R ext isthe resistanceoftheexternalresistor.Then,theMFCpowerthatwasdeliveredtothe externalresistorisgivenby P MFC = V MFC I MFC = I 2 MFC R ext .2 Theharvestedenergyiscalculatedbyintegratingthepowerovertime E MFC = Z P MFC dt .3 ChoosingthevalueoftheexternalresistanceiscriticalbecausetheMFCpower canbemaximizedwhentheexternalresistanceisequaltotheinternalresisance. Thiscanbeshowninthefollowinganalysisusingthesimpleequivalentcircuitin gure1.2c.Thecurrentcanbegivenas I MFC = V MFC R int + R ext .4 Then,theMFCpowerbecomesfrom1.2, P MFC = V 2 MFC R int + R ext 2 R ext .5 Assuming R ext isvariable,thispowerequationisquadraticandshouldhavea maximumpointat dP MFC dR ext =0.6 Fromthatderivative,wegetthatthemaximumpowerpointhappenswhen R ext = R int .7 7

PAGE 21

Thismeansthatusinghigherorlowerloadresistancewouldresultingettinglower powerfromtheMFC. Thevalueoftheinternalresistancecouldvarywithtimeduetooperating conditionschanges[28].Forthisreason,andinordertooperateatMPP,avariable loadresistancewasusedtotrackthevaryingMPPusingperturbandobserve P&Oalgorithm[28].Thatalgorithmwasabletosetthevariableloadresistance suchthatthemaximumpowerisextracted,orinotherwords,itwasabletosetthe loadvariableresistancetobeequaltotheinternalresistance.However,usinga resistorasenergyharvesterisnotpracticalsincealltheharvestedenergywillbe burnedasheatacrossthatresistorinsteadofbeingusedtorunsomeload. 1.2.2Supercapacitors SupercapacitorshavebeenwidelyusedtocollectMFCenergy[7{9].Using supercapacitorsismorepracticalthanusingresistorssincecapacitorscanstorethe harvestedenergy,sothattheenergycanbeusedtorunanyload.However,their disadvantageisthattheydonotprovideanycontrollabilitywhentheyare connecteddirectlytotheMFC.Whenasupercapacitorisconnectedtothe terminalsoftheMFC,theMFCterminalvoltagewillgodownuntilitbecomes equaltothesupercapacitorvoltage.Then,MFCcurrentwillstartowingintothe supercapacitoranditwillstartchargingit.Chargingthesupercapacitorwill increaseitsvoltage,whichwillincreaseMFCterminalvoltageanddecreaseMFC current.Thiswillhappenuntilthevoltageofthesupercapacitorbecomesequalto theOCVofthereactorandthen,thecurrentwillstopowing,whichmeansthat theenergyharvestingwillstop.Hence,theoperatingpointwillchangeasload conditionchanges,andthemaximumvoltagecannotexceedOCV. 1.2.3ChargePumps UsingchargepumpstoharvestMFCenergyisusefulsincetheharvestedenergy canbestoredintheoutputcapacitor[12,31].Furthermore,theadvantageofusing 8

PAGE 22

chargepumpsisthattheiroutputvoltageishigherthantheirinputvoltage,which advantageoussinceMFCterminalvoltageislowanditneedstobeincreasedin ordertobeusedtorunanypracticalload.However,chargepumpscanprovide limitedcontrollabilityonthereactoroperatingpoint.Inaddition,theeciencyof chargepumpsisverylowinMFCenergyharvesting,whereitwasreportedtobe between16.6%and24.4%in[31]. 1.3ActiveEnergyHarvesting:PowerEelctronicsConverters EnergyharvestingisoneofthemostimportantchallengesforMFCstobeused fortheelectricitygenerationbecauseoftheirlowterminalvoltage,anddicultiesin series/paralleloperation[32,33].Thebestelectricalmethodtoharvestenergyfrom lowvoltagesourcessuchasMFCisusingpowerelectronicsconverters[13,27,31,34] duetotheirhigheciencyandcontrollability. 1.3.1BoostConverter Theboostconverterisacommonconvertertopologythathavebeenwidelyused toharvestenergyfromlowvoltagesources[35,36].Theboostconverterbooststhe inputvoltagetoahigheroutputvoltagelevelandalsostorestheharvestedenergy intheoutputcapacitororbattery.Theschematicdiagramoftheboostconverteris showningure1.4.Itconsistsofoneinductor,onecapacitor,andtwoswitches. Notethatthetwoswitchesarecomplimentary,andthatmeanswhenoneofthemis closed,theotherisopen. 1.3.1.1 OperationofBoostConverter Theoperationofboostconverterscanbedividedintotwomodesofoperation: continuousanddiscontinuousconsuctionmode.Inthecontinuousconduction CConmode,thecurrentcontinuouslyowsfromthesource,andinthe discontinuousconductionDConmode,thecurrentfallstozeroforsometime duringtheoperation.Thecontinuousoperationmodehappenswithheavyloads 9

PAGE 23

Figure1.4:Schematicdiagramofboostconverter. whichaectstheoperationofboostconverters.However,forMFCenergy harvesting,theboostconverterneedstobedesignedtooperateatthecontinuous modetokeepextractingcurrentenergyallthetimewhichisthegoalforenergy harvesting.Tooperateinthecontinuousoperationmode,currentrippleneedstobe controlledsuchthatthecurrentdoesnotfalltozeroanytimeduringtheoperation. Theinductorsizeandtheswitchingfrequencyareimportantparametersthatneeds tobeconsideredinordertocontrolthecurrentripple.Thecontinuousoperation modeisdividedintotwooperationstates:On-stateandO-state.Theoperationof theboostconverterisbasedonswitchingbetweenthosetwostatesinahigh frequency. TheOn-stateiswhentheswitch S 1 ingure1.4isclosedand S 2 isopen.Inthis state,theinductorisconnectedacrosstheterminalsofthesource,andthecurrent willstartowingthroughtheinductorsatisfyingthefollowinginductorequation V L = V in = L dI L dt .8 where L istheinductanceinHenryand I L isthecurrentowingthroughthe inductor.From.8,andsince V L equalstotheinputsource,thecurrentwillstart increasingasshowningure1.5.Inthisstate,energywillbeextractedfromthe sourceandwillbestoredintheinductor'smagneticeld.Thevalueoftheinductor 10

PAGE 24

Figure1.5:Inductorvoltageandcurrentwaveforms. storedenergycanbecalculatedusing.9. E L = 1 2 LI 2 L .9 TheO-stateiswhentheswitch S 1 ingure.4isopenedand S 2 isclosed.In thisstate,thecurrentwillowthroughtheswitch S 2 totheoutputcapacitor,and consideringnovoltagedropacrosstheswitch S 2 ,thatcurrentwillfollowthe followingequation V L = V in )]TJ/F20 11.9552 Tf 11.955 0 Td [(V o = L dI L dt .10 where V o istheoutputcapacitorvoltage.From.10,andsincetheoutput capacitorvoltage V o ishigherthantheinputvoltage V in boostconverter,the derivative dI L dt willhaveanegativevaluewhichwillcausethecurrenttodecreaseas showningure1.5.Thisdecreaseintheinductorcurrentshowsthattheenergyis beingtransferredfromtheinductortothecapacitorasdescribedby.9.Notice thatfrom.8,thedecreaseininductorcurrentmeansareversedinductorvoltage polarity,andthisisshowninthewaveformsingure1.5.Whenthepolarityis reversed,theinductorvoltageisaddedtotheinputvoltageandresultsinahigher 11

PAGE 25

outputvoltageatthisoperationstate.Notethatwhentheboostconverterreturns totheOn-stateoperation,theoutputcapacitormaintainstheoutputvoltagelevel andsuppliesthecurrenttotheload. 1.3.1.2 ControlofBoostConverter Controllingtheboostconverterisimportanttoachievetherequiredfunctionof theconverter.Thecontrolhandleoftheboostconverteristhegatesignalwhich controlstheswitchstate,andtherearetwoparameterstobeconsideredin controllingthegatesignal.Therstparameteristheswitchingfrequencywhichis thenumberofcompletecyclesOn-andO-statespersecond.Theswitching frequencycanalsobecalculatedusingthetimeperiodforonecompletecycleas follows F sw = 1 T S .11 where T S isthetimeforonecompletecycle T On + T Off asshowningure1.5. Thesecondparameteristhedutycycledoftheswitchwhichistheratioofthe On-stateperiodoveracompletecycleperiod. d = T On T S .12 Increasingthedutycycleresultsinincreasingtheinputcurrentbecausethe converterspendsmoretimeintheOn-statewherethecurrentincreasesfollowing .8.Theabilitytocontroltheinputcurrentusingthedutycycleisimportant becausethedutycyclecanbeusedtocontroltheoperatingpointoftheMFC reactor. 1.3.1.3 ParametersDesign Designingtheparametersoftheboostconverterisimportanttoachievethe desiredoperation.Thetwoparametersthatneedstobedesignedaretheinductor andthecapacitor. 12

PAGE 26

Theinductorcanbedesignedtocontroltheinputcurrentripple,andthiscanbe shownusingtheinductorequation V L = L dI L dt = L I L t .13 where I L isthecurrentchangein t time.Choosing t tobetheOn-time dT S theinductorequationcanberewrittenasfollows L = V L dT S I L .14 where d isthedutycycle.Sincetheswitchingfrequency F SW isequalto 1 T S ,the equationissimpliedtobe L = V L d I L F SW .15 Fromthisequationitisclearthatthevalueoftheinductancecanbecalculated basedonthedesiredcurrentrippleusingtheinputvoltage,dutycycle,andthe switchingfrequency.Choosingthecurrentrippleshouldbebasedonthevalueof theinputcurrent,becausechoosingacurrentripplethatishighcomparedtothe inputcurrentmightresultinoperatinginthediscontinuousmode. Theoutputcapacitorcanbedesignedtocontroltheoutputvoltagelevelbased ontheloadcurrent.Thesameanalysiscanbeobtainedforthecapacitancedesign usingthecapacitorequation I C = C dV C dt .16 Itshouldbenotedthat,fortheMFCenergyharvesting,theoutputcapacitoris usuallychosentobeasupercapacitorwithaveryhighcapacitanceinordertostore theharvestedenergy. 1.4MaximumPowerPointTrackingMPPT TheamountofpowerthatcanbeextractedfromMFCdependsontheoperating pointoftheMFCreactorvoltage/current.TheMFCpowercanbemaximizedat MPP. 13

PAGE 27

HarvestingmostoftheMFC'savailableenergyisthegoalforallenergy harvesters;hence,operatingattheMPPisveryimportant;itreducesthestart-up timeoftheMFC[37,38],resultsinhighercolumbicandwastewatertreatment eciencies[28,39],andcouldresolvethevoltagereversalphenomenawhenMFCs areconnectedinseries[40].However,operatingattheMPPischallengingbecause theMPPcouldchangeovertimeduetovariousfactors. 1.4.1MaximumPowerPointTrackingAlgorithms Maximumpowerpointtrackingalgorithmscanbeusedtotrackthevarying MPP.ThemostbasicMPPTalgorithmistheperturbandobserveP&O algorithm.Thisalgorithmappliesasmallperturbationtothevoltageorcurrent andobservesthechangeinpower.Dependingonthedirectionoftheperturbation +/-andtheobservedpowerchange,thedirectiontowardMPPcanbe determined.ThisprocesscantracktheMPPifitchangesduringtheoperation. Anotheralgorithmthatcanbeusedistheextremumseekingbasedcontrol.The extremumseekingalgorithmusesasinusoidalperturbationtothevoltageorthe current,andthatshouldresultinasinusoidalpowerperturbationthatcanbeused totracktheMPP.Thisalgorithmisexplainedindetailsinchapter2. 1.5MFCVoltageOvershootandPowerGeneration AcommonfailurefortheenergyharvestingfromMFCiscausedbythevoltage overshootphenomenon[41{43].Thevoltageovershoothappensduringtheenergy harvestingwhenahighcurrentisextractedfromMFCwhichchangesthepolarity oftheanodepotentialfromnegativeinnormalcasestobepositiveorcloseto zero[43].Thischangeinpolarityresultsinreductionoftheterminalvoltageand theMFCoutputpower.ThevoltageovershootcouldaecttheoperationofMPP trackingalgorithmsbecauseoncethecurrentreachesthevoltageovershootvalue, thevoltagecollapsesandthealgorithmcannolongertracktheMPPbecauseofthe changeinthevoltage/currentrelationship.Afulldescriptionofthevoltage 14

PAGE 28

overshootphenomenaanditseectonthevoltage/currentrelationshipcanbefound inchapter2,aswellasaMPPTalgorithmthatcandetectandavoidthevoltage overshootregion. 1.6EectofPowerElectronicsConvertersonMFCPowerGeneration PowerelectronicsconvertershavebeenwidelyusedtoharvestenergyfromMFC becausetheyintroducemanyadvantagessuchascontrollability,higheciency,and theabilitytostoretheharvestedenergyattheoutputcapacitor.However,they introducecurrentripplewhichcouldhaveanimpactontheMFCreactoravailable poweranditslifespan.Currentripplefrompowerelectronicsconvertersis practicallytriangular-shape,whenusingboostconverter[13]oryback converter[44].TheeectofsuchcurrentrippleonMFChavenotbeeninvestigated yet.Inchapter3,theeectofpowerelectronicscurrentrippleonthepower,energy, andlifespanisinvestigatedinordertoevaluatetheirimpactasenergyharvesters forMFC. 15

PAGE 29

CHAPTERII INTELLIGENTENERGYHARVESTINGSCHEMEFORMICROBIAL FUELCELLS:MAXIMUMPOWERPOINTTRACKINGAND VOLTAGEOVERSHOOTAVOIDANCE 2.1Introduction MaximumpowerpointMPPisimportantinordertoextractthemaximum availablepowerfromMFC.TheMPPasdescribedinchapterI,istheoperating pointwherethemaximumpowercanbeextractedfromthereactor.Thatoperating pointisachievedbyextractingaspeciccurrentfromthereactor I MPP ,orby controllingtheterminalvoltagetobeequalto V MPP .Itshouldbenotedthat extracting I MPP fromthereactormeansthattheterminalvoltageshouldbeequal to V MPP .Thismeansthateitherthevoltageotthecurrentneedstobemonitored andcontrolledinordertoachieveMPPoperation. 2.1.1MaximumPowerPointTrackingMPPT ThebasicalgorithmthathavebeenusedtotrackMPPinMFCsistheperturb andobserveP&O[13,19,45,46].ThisalgorithmtrackstheMPPbyapplyinga stepperturbingtotheMFCvoltageorcurrentandobservingthenewpowerthat resultsfromthatperturbation.Bycomparingthepre-perturbandpost-perturb powers,thealgorithmndsthedirectiontowardMPP.Thisalgorithmcanbe implementedusingvariableresistors[19,45],wherethevoltageperturbationis achievedbychangingtheresistancevalue.However,usingresistorsisnotpractical becausetheharvestedenergycannotbeusedbutinsteadisburnedacrossthe resistance.ThesamealgorithmP&Owasimplementedusingpowerelectronics converters[13,46],wherethecurrentperturbationisachievedbychangingtheduty cycleofthepowerelectronicsconverter. TheextremumseekingESalgorithmisusedinthischapteralongwithpower electronicconverterstotrackMPPinMFCs.ESalgorithmusessinusoidal 16

PAGE 30

perturbationtondthedirectionofMPP.Thisisachievedbyapplyingasinusoidal perturbationtotheMFCvoltageorcurrent,whichwillresultinasinusoidal varyingpower,andthatcanbeusedtondthedirectionofMPP.TheES algorithmissimilartotheP&Oalgorithm,wherebothalgorithmsuse voltage/currentperturbationstondthedirectionofMPP,butthetypeof perturbationisdierent,wheretheESusessinusoidalperturbationcomparedto stepperturbationinthecaseofP&O. 2.1.2VoltageOvershootPhenomenon Thevoltageovershootphenomena[41{43]happenswhenahighcurrentis extractedfromMFC,whichchangesthepolarityoftheanodepotentialfrom negativetopositive,andthischangeinthepolarityresultsinasignicantreduction intheMFCterminalvoltage,whichreducestheMFCavailablepower[43].Aswell asreducingtheMFCavailablepower,thevoltageovershootcanmisleadbasic MPPTalgorithmstoarun-awaycondition,whichresultsinharvestingzeropower fromthereactor.Researchershavebeenfocusingonunderstandingthevoltage overshootphenomena[41{43,47,48]ratherthantheoperationinthepresenceof voltageovershoot. Inthenormaloperation,reducingorincreasingtheMFCvoltageincreasesor decreasesthecurrent,respectively.ThisbehaviorhasbeenexplainedinchapterI anditwasshownusingthepolarizationtest.However,whenthevoltageovershoot happens,theMFCvoltageisreducedaswellasthecurrentasshowningures2.7. Ingure2.7,theslopeofthevoltage/currentrelationshipchangedfromnegativeto positivewhenthevoltageovershoothappened.Thischangeintheslopemisleads thebasicMPPTalgorithmsbecausewhenthevoltageovershoothappens,the algorithmwilltrytoincreasetheMFCcurrent,byincreasingtheboostconverter's dutycycle,toincreasetheMFCpower,sincetheslopeispositiveasshowningure 2.7.IncreasingtheMFCcurrentmakesthevoltageovershootworse,sinceit 17

PAGE 31

happenedintherstplacebecauseofhighcurrent,andthepowerkeepsdecreasing. Asaresult,theMPPTalgorithmwillkeepincreasingtheMFCcurrentandthe powerkeepsdecreasinguntiltheMPPTalgorithmshortstheterminalsoftheMFC. Itshouldbenotedthatsometimesthereactorrecoversfromthevoltageovershoot athighcurrentsasshowningure2.6,butoncethereactorrecovers,thealgorithm ndsaverylowpoweroperatingpointandcannotgobacktotheMPP.Inthis chapter,anintelligentvoltageavoidancealgorithmisproposed.Theproposed algorithmcantrackMPPinnormalconditionsanditcandetectthevoltage overshootwhenithappens.Afterdetectingthevoltageovershoot,thealgorithm triestooperateatanoperatingpointrightbeforetheoperatingpointwherethe voltageovershoothappens. 2.2ExtremumSeekingAlgorithmESA Operatingthesystematitsmaximumorminimumvalueissometimesrequired insomeengineeringsystems,andthatcanbeachievedbycontrollingthesystem inputandmaintainingitataspecicvaluesuchthatthedesiredoperationis achieved.Toperformthatcontrol,extremumcontrolalgorithmsorextremum seekingESalgorithmscanbeused.ThehistoryanddevelopmentsoftheES controlwithalargenumberofreferenceshavebeenreportedin[49]. Giventhatthenonlinearsteadystateinput-outputmapis y t = f )]TJ/F20 11.9552 Tf 5.48 -9.684 Td [(u t ;u t > 0.1 where y t istheoutput, u t istheinputcontrolhandle,and u istheunknown parameter.Itisassumedthatthefunction f ; achievesitsextreme y when u t = u ,andthismeansthattheequation.1becomes y = f u ;u whenitis controlledtoitsextreme,i.e.,theinput u t isequalto u TheblockdiagramoftheEScontrolisshowningure2.1.Thesamplerblock representstheconversionoftheplant'soutput y t intoadiscretetimesampled 18

PAGE 32

Figure2.1:Extremumseekingblockdiagram. signal y k ,where k =0 ; 1 ;::: ,thatwillbeusedintheEScontrol.Ingure2.1, d k and p k arethedemodulationandperturbationsignals,respectively,and q isa forwardshiftoperator,i.e., qu k = u k +1 .Inthealgorithm,thevalueof p k is modulatedbytheslopeofthenonlinearfunction f ; ,andthedemodulation signal d k producesanestimateofthegradientof f ; bythedemodulationofthe plant'soutputsignal y k .Then,theestimatedgradientisintegratedandthatgives thevalueof^ u k ,whichisanestimationoftheoptimizer u .Thisestimatedvalueis addedtotheperturbation p k andthatgivesthedesiredsysteminput u k .TheZOH blockingure2.1representsthezeroorderholdoperationwhichconvertsthe discretetimesampledsignal u k intoananalogsignal u t ,anditisdenedbythe followingequation u t = u k fort 2 kT; k +1 T .2 where T isthesamplingperiodand k> 0. Inthischapter,sinusoidallyshapedperturbationanddemodulationsignalsare 19

PAGE 33

usedasfollows p k = Acos 0 k d k = Acos 0 k + .3 where A> 0, > 0,and isanarbitraryphaseshiftangle.Itshouldbenotedthat theperturbationfrequency 0 mustbewithinthebandwidthof f ; ifitisapart ofsomesystemdynamics,andhigherperturbationfrequency 0 resultsinabetter estimationof f ; gradient[50,51].Ifthefunction f ; isaquadraticfunction, f u t ;u = f 0 + f 1 u t )]TJ/F20 11.9552 Tf 11.956 0 Td [(u 2 ,where f 1 < 0,theinitial^ u 0 canbeanynumber withintheinputlimitsofthesystem,andthealgorithmshouldreachthecritical pointandmaintaintheoperationinthatpoint.Noticethatwhenthedesired operatingpointisreached,thesinusoidalperturbation p k in.3isstillappliedto theinputandthatwillresultsinanoscillationaroundthethatcriticalpoint.Then, from.3itisclearthatsmallerperturbationamplitude A resultsinsmaller oscillationaroundthecriticalpoint,butithasbeenshownin[50]thatsmaller A reducesthespeedofconvergenceof^ u k .Thismeansthatthechoiceof A isa trade-obetweentheerrorsize u k )]TJ/F20 11.9552 Tf 11.955 0 Td [(u andthespeedofconvergenceof^ u k Inthischapter,theboostconverterisusedtocontroltheenergyharvestingfrom MFC.Forthatreason,theMFCinstantaneouspower p t isconsideredtobethe systemoutput y t ,andthedutycycleoftheboostconverter D t isconsideredto betheinput u t ofthesystem.Thealgorithmthenwillapplythesinusoidal perturbation p k totheboostconverter'sdutycycle D t .Noticethatapplyinga sinusoidalperturbationtothedutycycleoftheboostconverterwillresultin sinusoidallychangingvoltageandcurrentandthatmeanssinusoidallychanging MFCpower.Theparametersin.3thatwasusedfortheexperimentsinthis chapterarelistedintable2.1.Noticethatthevalueoftheperturbationamplitude A is0 : 05,andsincetherangeofthedutycycle u t inthischapteris[0 1], A is settobe5%ofthemaximumdutycycle. 20

PAGE 34

Table2.1:Perturbationanddemodulationsignalsparameters. A 0 T 0 : 05 88 : 8rad/s 18 106 0 0 : 33ms 2.2.1MaximumPowerPointTrackingMPPTUsingESA Thetypicalquadraticpowerthatresultsfromthepolarizationtest,assuming linearcurrent-voltagerelationship,isshowningure2.2.Itisclearthatthereisa pointcurrentvaluewherethepowerismaximized.Thismaximumpowerpoint MPPcanbetrackedusingtheextremumseekingalgorithm.Thesinusoidal perturbationischosentobeusedwithextremumseekingalgorithm,wherethe algorithmsimplyappliesasmallsinusoidalperturbationtotheMFCcurrentas follows I MFC = I DC + p k p k = i sin !t .4 where k =0 ; 1 ; 2 ;::: I MFC istheMFCcurrent, I DC istheDCcomponentofthe MFCcurrent,and p k istheappliedperturbation.Whenapplyingsuchsinusoidal perturbationtotheMFCcurrent,thepowerwillhaveasinusoidalcomponentas well.Thosetwosinusoidalcomponentsareusedbytheextremumseekingalgorithm totrackMPP. Inregion#1ofgure2.2,thepowerthatresultsfromperturbingthecurrent canbedescribedasfollows,assuminglinearrelationshipinregion#1 P MFC = P DC + p sin !t .5 Whentheperurbationisappliedtothecurrent,MFCpowerismeasuredandthat measurementispassedthroughahighpasslterandthathighpasslteronly passesthesinusoidalcomponentofthepower. P filtered = p sin !t .6 21

PAGE 35

Itshouldbenotedthatthehighpassltermightimposeanamplitudegainor phaseshifttothesinusoidalcomponentofthepower,butthoseeectsaresmalland canbeneglected.Thesinusoidalcomponentofthepowergivestheinformation aboutthedirectiontowardMPP.ThedirectionofMPPcanbesimplyfoundby multiplyingthesinusoidalcomponentofthepowerwithademodulationsignaland integratingtheresult.Thedemodulationsignalisdenedasfollows d k = i sin !t + .7 where > 0, isanarbitraryphaseangleshift.Thedemodulationsignalissimilar tothecurrentperturbation.4withtwoextraterms.Thetwoparameters nad areincludedinthedemodulationsignaltohavesomecontroloverthespeedof convergence.However,fortheexplanationoftheoperationofextremumseeking algorithm,theyareneglected =1, =0.ThedirectionofMPPcanbefound byintegratingtheresultofthemultiplicationofthedemodulationsignal.7and thesinusoidalcomponentofthepower.6. Z P filtered d k dt = Z p sin !t i sin !t = Z p i 2 + cos 2 !t > 0 .8 ThisintegrationisalwayshigherthanzeroandthatmeanstheMPPisinthe positivedirectionofthecurrent,whichmeansthatthecurrentneedstobeincreased inordertomovetowardMPP. Thesameanalysiscanbeobtainedforregion#2butthepowerwouldbe180 phaseshiftedfromthecurrentperturbation.This180 phaseshiftwillresultina negativeresultoftheintegrationin.8whichmeansthattheMPPisinthe negativedirectionofthecurrent,andthecurrentneedstobedecreasedinorderto movetowardMPP. Inthischapter,thedutycycleoftheboostconverterisusedasacontrolhandle, anditisknownthatincreasingordecreasingthedutycycleincreasesordecreases 22

PAGE 36

Figure2.2:GraphicalrepresentationofextremumseekingMPPT. thecurrent,respectively.Then,feedingtheresultoftheintegrationtothecontrol handledutycycleshouldmovetheoperatingpointtowardMPP.Agraphical representationoftheextremumseekingcontrolispresentedingure2.2. 2.2.2ESAComputerSimulation 2.2.2.1 SimulationModel TheESalgorithmwassimulatedinthecomputerbeforebuildingthehardware tomakesureitworksasexpected.TheMatlab/Simulinksoftwarewasusedforthe simulationoftheESalgorithmandtheoverallsystem.Theoverallsimulation systemisshowningure2.3.ItconsistsofMFC,boostconverter,andES algorithmblock.ThedetailsoftheESalgorithmblockareshowningure2.3b.In thissection,thesimulationmodelandtheresultsaredescribedandanalyzed. TheMFCwasmodeledusingthesimpleelectricalequivalentcircuitthatis showningure2.4.Thisequivalentcircuitconsistsofavoltagesourceandaseries resistance.Thevaluesofthevoltagesourcevoltageandtheseriesresistancewas chosesuchthatthemodelgivesapowerthatcomparabletotheactualMFCpower. Inthecurrentsimulation,thevoltagesourcewassetto0 : 7 V andtheseries 23

PAGE 37

a b Figure2.3:Simulinkmodelscreenshots:aOverallsystembESalgorithmblock. 24

PAGE 38

Figure2.4:MFCsimpleequivalentcircuit. resistancewassetto150,whichmeansthatthemaximumpowerthatcanbe harvestedfromthecircuitis816 W Theboostconverterwasusedinthesimulationasenergyharvester,andatthe sametime,itwasusedtocontroltheoperatingpointoftheMFC.Theboost converterharveststheenergyfromMFCandstoresitintheoutputcapacitor,and thatcapacitorneedstobeveryhighsupercapacitorinordertostorethe harvestedenergy.ControllingtheoperatingpointoftheMFCcanbeachievedby controllingthedutycycleoftheboostconverter,asexplainedindetailsinchapterI. Theinductanceoftheboostconverterwassetto200 mH andtheoutputcapacitor wassetto0 : 5 F TheESalgorithmblock,asshowningure2.3b,usestheMFCterminalvoltage andcurrentasinputs,andthepoweristhencalculated,anditisconsideredtobe thesystemoutputthatneedstobemaximized.TheoutputoftheESalgorithm blockisthedutycycleoftheboostconverter,anditisclearthattheblock"Sine wave1"isaddingasinusoidalperturbationtothecalculateddutycycle,whichisone ofthecharacteristicsoftheESalgorithm.However,thedutycycleoftheboost 25

PAGE 39

Figure2.5:ESMPPTsimulationresults. converterisconsideredtobethesysteminput.ThismeansthattheESalgorithm willperturbsinusoidalperturbationtheboostconverter'sdutycyclesystem inputandwillobservetheMFCpowersystemoutput,andwillchangetheduty cyclesuchthattheMFCpowerismaximized.ThevaluesoftheESalgorithm parameters,thatisusedin.3,wassettobeequaltothevaluesintable2.1. 2.2.2.2 SimulationResults ThesimulatedmodelwasabletosuccessfullyndandoperateattheMPP.The modelwasrunfor30secondswithaninitialdutycycleof0 : 1.Theresultsofthe MFCpower,voltage,andcurrentareshowningure2.5.Itisclearthatthe algorithmwasabletondtheMPP W inlessthan10seconds,andthe algorithmwasabletomaintaintheoperationatthatpoint.TheMFCvoltageis alsoshowningure2.5,anditshowsthattheoperationstartedatavoltagelower 26

PAGE 40

thantheOCV : 7 V andthatisexplainedbythe0 : 1initialdutycycle.The voltageresultshowsthatthealgorithmstartedtodecreasethevoltageuntilit reachedtheMPPvoltage,whichishalfoftheOCV.Ontheotherhand,the simulationstartedwithacurrentthatishigherthanzeroandthisisduetothe0 : 1 initialdutycycle.Thecurrentincreaseduntilitreachedthecurrentvaluethat generatesthemaximumpower.ItshouldbenotedthatevenafterndingtheMPP, thealgorithmkeepsperturbingsinusoidalperturbationthesysteminputduty cycleandthismeansthatthesinusoidalperturbationwillalwaysappearinthe power,voltage,andcurrentwaveforms.Thissinusoidalperturbationisshowninthe magniedcurrentwaveformingure2.5,anditisclearthatisstillexistsevenafter reachingtheMPP. 2.3VoltageOvershootAvoidanceAlgorithm Thevoltageovershootchangesthecurrent-voltagerelationshipoftheMFCas canbeseeningure2.7.Afterthevoltageovershoot,increasingordecreasingthe currentwouldincreaseordecreasethevoltage,respectively.Thischangeinthe current-voltagerelationshipcanmisleadtheESMPPTalgorithmbecauseitchanges thecurrent-powerrelationshipaswell,ascanbeseeningure2.7.Thevoltage overshootingure2.7happenedinregion#2ofgure2.2,anditisclearthatthe voltageovershotchangedthecurrent-powerrelationshipinregion#2tobesimilar tothatinregion#1.Becauseofthischangeinthecurrent-powerrelationship,the ESMPPTalgorithmwillreactasiftheoperatingpointisinregion#1.The operationinregion#1requiresincreasingthecurrenttomovetowardtheMPP, andthatisachievedbyincreasingthedutyratiooftheconverter.Asaresult,the ESMPPTkeepsincreasingthedutyratiointhevoltageovershootregion,which willdecreasethevoltageaswellasthecurrentandpowerandthisbecomesa runawaycondition.Detectingthevoltageovershootisimportantbecauseithelpin avoidingtherunawaycondition.Thevoltageovershootcanbedetectedbychecking 27

PAGE 41

Figure2.6:MFCpolarizationcurvewithavoltageovershoot. Figure2.7:MFCpolarizationcurvewithavoltageovershootthatrecovers. 28

PAGE 42

Figure2.8:Flowchartoftheproposedvoltageavoidancealgorithm. 29

PAGE 43

thecurrent-voltagerelationshipduringtheoperationofESMPPT.Checkingthe current-voltagerelationshipcanbeachievedbyusingthevoltageandcurrent measurementstocheckthefollowinginequalities. V k )]TJ/F20 11.9552 Tf 11.955 0 Td [(V k )]TJ/F20 11.9552 Tf 11.955 0 Td [(n <" 1 I k )]TJ/F20 11.9552 Tf 11.955 0 Td [(I k )]TJ/F20 11.9552 Tf 11.955 0 Td [(n <" 2 .9 wherekrepresentsthemostrecentmeasurement, n> 0representsanold measurement,and 1 and 2 aresmallpositivenumbers,ideallyzero.Inthenormal operation,onlyoneinequalityin.9willbesatisedbecausetheincreaseinthe voltagewillcauseadecreaseinthecurrent.Ontheotherhand,whenthevoltage overshoothappens,theincreaseinthevoltagewillcauseanincreaseinthecurrent, whichwillsatisfybothinequalitiesin.9.Thevoltageovershootisthendetected whentheinequalities.9aresatisedforNconsecutivetimes.Thereasonfor including 1 and 2 in.9istoconsiderthemeasurementerrors. Oncethevoltageovershootisdetected,theESMPPTisdisabled,dutyratiois recorded,andthecircuitisopened.Oncethecircuitisopened,theterminalMFC voltagewillstarttoincreasetowardOCV.WhentheMFCvoltagereachesOCV, theESMPPTstartedagainwithadutyratiolimitthatisequalto0.95ofthe recordeddutyratio.Thismeansthatwhenthedutyratioreachesthelimit,theES MPPTwillnotbeabletoincreasethedutyratiobecauseitmightleadtovoltage overshoot.Ifthevoltageovershootisdetectedwiththenewdutyratiolimit,the processisrepeatedandthedutyratiolimitisreducedagainuntilthealgorithm ndsadutyratiolimitthatavoidsthevoltageovershoot.Theowchartofthe proposedvoltageovershootavoidanceVOAalgorithmisshowningure2.8. 2.4ExperimentSetup 2.4.1HardwareSetup Tobeabletotesttheproposedalgorithms,theboostconverterwasbuiltinthe labaswellastherequiredcontrols.Thepictureoftheexperimentsetupisshown 30

PAGE 44

Figure2.9:Experimrntsetuppicture. ingure2.9.Inthissection,thedetailsofeachcomponentoftheexperimentsetup arepresentedandexplained. 2.4.1.1 BoostConverter TheoperationoftheboostconverterisexplainedindetailsinchapterI.The boostconvertersimplyconsistsofaninductor,aMOSFET,adiode,andanoutput capacitor.Forthisexperiment,aninductorof28mHat1kHzCST206-1A,Triad Magnetics,CA,theMOSFETNG,ONSemiconductors,AZ,whichhaslow onresistance,thediodeBAT46,VishayElectronicGmbH,andanoutputsuper capacitorof1Fwereusedtobuildtheboostconverter.Theboostconverteris showningure2.9insidethedashedbox. 2.4.1.2 Microcontroller Tocontroltheboostconverterandtoimplementtheproposedalgorithms,the microcontrollerMSP430f5529,TexasInstruments,TXwasused.This microcontrollercanbeprogrammedusingTexasInstrumentssoftwareCode ComposerStudio,TexasInstruments,TXusingC/C ++ language.The microcontrollercaneasilygeneratethePWMtoMOSFETgate,withthedesired 31

PAGE 45

switchingfrequencyanddutycycle,usingthebuiltintimerandcounter.Itshould benotedthatthevoltagesthatisgeneratedforbythetimerandcounterare enoughtoturnontheusedMOSFET,thereforethereisnoneedforagatedriver betweenthemicrocontrollerandtheusedMOSFET.Fortheproposedalgorithms, thedesireddutycycleistheoutputanditisfedtothetimerregisterstoconvert thatintogatepulses.FortheinputoftheproposedalgorithmsMFCpower,the analogtodigitalconverterADCwasused.TheADCsimplymeasuresvoltages throughthemicrocontrollerpinsandconvertthatvoltagemeasurementintoa digitalnumberthatcanbeusedinthecode.Thedetailsaboutvoltage/current measurementsaredescribedinthefollowingsection. Theimplementedcodetriggersthevoltage/currentmeasurementevery0.1s samplingtime T S ,andthepoweriscalculatedbysimplymultiplyingthecurrent timesthevoltage.Thepoweristhenfedtothealgorithmandthenewdesiredduty cycleisgeneratedbythealgorithmsandfedtomicrocontrollertimerandcounterto generatetheMOSFETgatepulses.Itshouldbenotedthattheusedmicrocontroller isaxedpointmicrocontroller,whichmeansthatittakeslongtimetocalculate oatpointcalculation,andsincethecalculationsareoatingpointcalculation,the MSP430IQlibrarywasusedtoreducethecalculationtime.Forthedetailsabout programmingthetimerandtheADC,checkthemicrocontrollerdatasheetandthe MSP430x5xxfamilyuser'sguide. 2.4.1.3 VoltageandCurrentMeasurements ThemeasurementsofthevoltageandcurrentareneededtocalculatetheMFC power,whichistheoutputofthesystem,thatneedstobemaximized.The measurementofthevoltageissimplebecausethemicrocontrollercanmeasure voltagesthroughthebuildinADC.However,becausetheMFCvoltageislow comparedtothemicrocontrollervoltage,themeasurementmightbeinaccurate.For thatreason,theMFCvoltageisampliedusinganoperationalamplierOP-AMP 32

PAGE 46

a b Figure2.10:Measurementcircuitsschematicdiagrams:aVoltagebCurrent. 33

PAGE 47

LM358,TexasInstruments,TX.Theinvertingtopologywhichisshowningure 2.10aisusedbecauseofitssimplicity.However,thattopologygeneratesnegative voltagebecauseofitsinvertingbehavior,butthemicrocontrollerADCcannotread negativevoltages.Tore-invertthevoltageandgenerateapositivevoltagethatis readablebythemicrocontrollerADC,anotherOP-AMPisusedandtheoutputof thatOP-AMPisfedtothemicrocontrollerADC.Toincreasetheaccuracyofthe voltagemeasurement,alowpasslter,thatremovesthenoiseofthevoltageis addedtothesecondOP-AMP.Itshouldbenotedthatthepassbandofthelowpass lterneedstobedesignedsuchthatitpassestheESsinusoidalperturbation.The schematicdiagramoftheusedvoltagemeasurementcircuitisshowningure2.10. Thecurrentmeasurementismorecomplexthanthevoltagemeasurementcircuit becausethemicrocontrollerADCcanonlyreadvoltages.Inordertoreadthe current,ashuntresistance R shunt needstobeinsertedinthecurrentpath.Inthe currentsetup,the R shunt isinstalledinserieswiththeboostconverter'sinductor. Thecurrentisthencanbemeasuredbyreadingthevoltageacrossthatshunt resistance,andbyknowingthe R shunt value,thecurrentcanbesimplycalculatedby dividingthe R shunt voltagemeasurementby R shunt value.Insertingaresistancein thepathofthepathofMFCcurrentresultsinapowerloss,butthatresistance shouldbesmalltominimizethepowerloss.However,usingsmall R shunt meansthat thevoltageacrossthatresistanceislowanditneedstobeampliedinordertobe readablebythemicrcontrollerADC.Toamplifytheshuntresistancevoltage,the instrumentationamplierINA122,Burr-Brown,AZ,whichisdesignedforcurrent measurementswasused.Thatinstrumentationamplierampliestheinputvoltage R shunt voltagebyanamplicationfactorthatdependson R G .Thatamplication factorneedstobeconsideredwhencalculatingthecurrentaftermeasuringthe voltagebythemicrocontrollerADC.Toincreasetheaccuracyofthecurrent measurement,similarlowpassler,thatwasaddedtothevoltagemeasurement, 34

PAGE 48

Figure2.11:Theoverallsystemschematicdiagram. wasaddedattheoutputoftheINA122.Theschematicdiagramoftheusedcurrent measurementcircuitisshowningure2.10b. 2.4.1.4 OverallSystem Theoverallsystemschematicdiagram,thatincludestheboostconverter,the microcontroller,andthevoltage/currentmeasurements,isshowningure2.11. Thissystemwasusedtoimplementtheproposedalgorithmsasshowninthe experimentsetuppictureingure2.9. 2.4.2ExperimentalResults Theproposedalgorithmswereimplementedandtestedusingthedesigned hardwaresetup,andinordertocollecttheresults,themicrocontroller'sserial communication,UARTmode,wasused.Theserialcommunicationwassettocreate acommunicationchannelwiththemicrocontrollerusingthecomputer.The communicationisachievedbyconnectingRS232cablebetweenthemicrocontroller's communicationpinsandthecomputerCOMterminal.Then,dierentcommands canbedenedinthemicrocontrollercode,andoncethemicrocontrollerreceivesa 35

PAGE 49

commandfromthecomputer,throughserialcommunication,itexecutesaspecic functionrelatedtothatcommand.Forexample,oneofthefunctionsthatwere implementedwas"ADC",andsending"ADC"fromthecomputertothe microcontroller,makesthemicrocontrollermeasureMFCvoltageandcurrentand replieswiththosemeasurementstothecomputer.Thesoftwarethatwasusedto sendandreceivecommandsfromandtothecomputerwastheHyperTerminal. Fortheexperimentsresults,oncethealgorithmstartsoperating,the microcontrollerisprogrammedtosendallmeasurementsfrequently,every0.1s,to thecomputerthroughserialcommunication.ThosemeasurementsincludeMFC voltage,MFCcurrent,MFCpower,boostconverter'sdutycycle,andboost converter'soutputvoltage.Then,themeasurementswerecapturedandsavedintoa textleandimportedtoMatlabtobeplotted. 2.4.2.1 ESMPPT Inthissection,thesinusoidalbasedextremumseekingalgorithmwas implementedandusedtotrackMPP.However,theMFCreactorthatisavailablein thelabhasavoltageovershootthathappensbeforetheMPPascanbeseenin gures2.7and2.6.Forthatreason,thefunctionalityoftheESMPPTalgorithm cannotbeshownusingthatreactorbecauseofthevoltageovershoot.Forthat reason,andinordertoevaluatetheESMPPTalgorithm,thesimpleequivalent circuit,thatisshowningure2.4,wasusedtomimicthepowercharacteristicsof theMFC.TheDCvoltagesourceGPC-30300,GWINSTEKwassetto700mV whichisclosetotheopencircuitvoltageOCVoftypicalMFC,andaresistanceof 150waschosenbecausethatvalueresultsinanoutputpowerthatiscomparable totheMFCoutputpower.Thefollowingequationcanbeusedtocalculatethe theoreticalpolarizationcurveofthesimpleequivalentcircuit P output = V OCV )]TJ/F20 11.9552 Tf 11.955 0 Td [(iR int i .10 36

PAGE 50

Figure2.12:ESMPPTexperimentalresults:voltageandpower. where V OCV istheDCsupplyvoltage, R int istheinternalresistance,and i isthe outputcurrent.Byinsertingdierentcurrentvalues,thepolarizationcurvewas calculatedandplottedingure2.12.Thistheoreticalpolarizationcurveshowsthe operatingpointoftheusedcircuitanditwasusedevaluatethefunctionalityofthe ESMPPTalgorithm. Thealgorithmwasoperatedfor3minutesandtheresultisshowningure2.12. Thedottedlineshowsthetheoreticalpolarizationcurvethatwascalculatedusing 2.10.Theblueemptydotsrepresentthevoltagemeasurementsandtheredlled dotsrepresentthecalculatedpowerbasedonthevoltage/currentmeasurement.Itis clearthattheESMPPTalgorithmwasabletoreachandoperateattheMPP.The operationstartedfromavoltagethatisclosetotheopencircuitvoltageOCVand withasmallinitialcurrent.Then,thealgorithmstartedtodecreasethevoltageby increasingthecurrent,usingtheboostconverter'sdutycycle,andthevoltagekept decreasingandthecurrentkeptincreasingaswellasthepoweruntiltheMPPwas 37

PAGE 51

Figure2.13:ESMPPTexperimentalresults:boostconverter'sdutycycle. reached. Thecontrolhandlewhichistheboostconverter'sdutycyclevalueduringthe experimentisshowningure2.13.Itisclearthatthedutycyclestartedfromalow initialvalueandstartedincreasingtoincreasethepoweruntilMPPwasreached. NoticethatthedutycyclekeptincreasingevenafterndingtheMPPandthisis becausetheboostconverter'soutputcapacitorvoltage.Storingtheharvested energyintheoutputcapacitorresultsinhighercapacitorvoltageandforthat reason,higherdutycycleisrequiredtomaintainthecurrentthatgivesthe maximumpowerMPPcurrent. 2.4.2.2 VoltageOvershootAvoidance Theresultsoftestingthevoltageovershootavoidancealgorithmisshownin gures2.14,2.15,2.16,2.17and2.18.TheexistingMFCreactorthathasavoltage overshootwasusedtotestthevoltageavoidancealgorithm.Toshowthe functionalityofthevoltageovershootavoidancealgorithm,twoexperimentswere conducted.TherstexperimentwasconductedusingtheESMPPTalgorithm alone.Inthesecondexperiment,thevoltageavoidancealgorithm,whichisbasedon 38

PAGE 52

Figure2.14:Voltageavoidancealgorithmexperimentalresult:MFCvoltage. theESalgorithm,wasusedtoharvestenergyfromtheMFC.Theextremumseeking algorithmwasnotabletooperatewiththeexistenceofthevoltageovershoot,where theovershootcausedthealgorithmtoarunawayconditionthatresultedinshorting theterminalsoftheMFCdutyratio > 1after0.5minasshowningures2.14, 2.15,2.16,2.17and2.18.Ontheotherhand,thevoltageavoidancealgorithmwas abletodetectthevoltageovershootthathappenedaround0.4minandeventually foundanoperatingpointwherethevoltageovershootdidnothappen. TheMFCvoltageduringthetwoexperimentsisshowningure2.14.Itisclear thattheESMPPTalgorithmstartedfromtheOCVanddecreaseduntilitreached zerowhentheterminalsoftheMFCwasshortedbecauseofthevoltageovershoot. Ontheotherhand,thevoltageavoidancealgorithmstartedfromtheOCVand decreaseduntilthevoltageovershootwasdetected,thentheterminalswasopenand thealgorithmuntiltheMFCvoltagereachedavoltagethatisclosetheOCV mV.Thealgorithmstartedagainwithadutycyclelimit,butthevoltageovershoot happenedagain,andtheterminalwasopenforthesecondtimeandthelimitwas decreased.Inthethirdtime,theovershootdidnothappenandthevoltagereached around330 mV andthealgorithmkeptoperatingatthatvoltagelevel. 39

PAGE 53

Figure2.15:Voltageavoidancealgorithmexperimentalresult:MFCcurrent. TheMFCcurrentduringthetwoexperimentsisshowningure2.15.Itisclear thattheESMPPTalgorithmstartedfromsmallcurrentandstartedtoincrease, andwhenthevoltageovershoothappened,itstarteddecreasinguntilisgoesdown tozeroinaround0.4min.Thevoltageovershootavoidancealgorithm,startedfrom asmallcurrentandincreaseduntilthevoltageovershoothappenedandthecircuit wasopenedandthecurrentwentdowntozero.Whenthealgorithmstartedagain, thesamethinghappenedbuttheovershoothappenedwithasmallercurrent.On thethirdtime,thealgorithmwasabletondadutycyclelimitwherethevoltage overshootdidnothappen,andacurrentofaround1 : 2 mA wascontinuously extractedfromMFC. Figure2.16showstheMFCpowerduringthetwoexperiments.Itisshownthat theESMPPTalgorithmpowergoesdowndowntozeroataround0.4minandthe powerwaszeroallthetime.Ontheotherhand,thevoltageavoidancealgorithm wasabletondanoperatingpointwithapowerofaround400 W afteropening thecircuittwotimesbecauseofthevoltageovershoot.Forthecomparisonofthe twoalgorithms,theharvestedenergyisshowningure2.17,anditshowsthatafter 10minutesofoperationthattheESMPPTalgorithmwasabletoharvest17 : 7 mJ 40

PAGE 54

Figure2.16:Voltageavoidancealgorithmexperimentalresult:MFCpower. comparedto197 mJ harvestedbythevoltageavoidancealgorithm.Noticethatthe lowenergyfromtheESMPPTalgorithmisbecausetheESMPPTalgorithm harvestedzeropowerafter0 : 4 s becauseofthevoltageovershoot,butthevoltage avoidancealgorithmkeptharvestingenergywiththeexitanceofvoltageovershoot. Theoperationofthevoltageovershootavoidancealgorithmisclearwhenlooking atthecontrolhandlewhichistheboostconverter'sdutycycleingure2.18.Itis clearthatthedutycyclestartedfromasmallinitialvalueandincreaseduntilthe voltageovershootwasdetected,andadutycyclelimitwassetbasedonthelast dutycyclewherethevoltageovershoothappened.Then,thealgorithmdecreases thatdutycyclelimitagainwhenthevoltageovershootwasdetectedforthesecond time.Inthirdtime,thedutycycledidnotexceedthedutycyclelimitandthat preventedthevoltageovershootfromhappening.ThedutycycleinthecaseofES MPPTalgorithmshowsthatthedutycycleexceeded1afterthevoltageovershoot. Inconclusion,Althoughthevoltageovershootavoidancealgorithmoperatedata lowerpoweroperatingpointcomparedtoMPP,itwasabletoharvestmoreenergy fromMFCbecausetheESalgorithmstoppedharvestingenergyafter0.4min.The ESalgorithmwasabletoreach1 mW powerbutthatwasforashortperiodoftime 41

PAGE 55

Figure2.17:Voltageavoidancealgorithmexperimentalresult:MFCenergy. Figure2.18:Voltageavoidancealgorithmexperimentalresult:dutycycle. 42

PAGE 56

beforethevoltageovershootmisleadit.Ontheotherhand,thevoltageovershoot avoidancealgorithmwasabletooperateat400 W forlongperiodoftime. 43

PAGE 57

CHAPTERIII EFFECTOFPOWERSHAPEONENERGYEXTRACTIONFROM MICROBIALFUELCELL 3.1Introduction 3.1.1PowerElectronicsConverterCurrentRipple Powerelectronicsconvertersintroducecurrentripplestothesourcebecauseof theirswitchingbehaviourasdescribedinchapter1.Thiscurrentripplecouldhave animpactontheMFCreactorintermsofitsenergyproductionandlifespan.The impactofsuchcurrentrippleonMFCshasnotbeeninvestigatedyetalthough powerelectronicsconvertershavebeenwidelyusedtoharvestenergyfromMFC. Theeectofdierenttypesofcurrentrippleshavebeenreportedfor hydrogen-basedfuelcells[52{54],whereitwasconcludedthatthelowfrequency currentripplecouldreducetheprotonmembraneexchangefuelcellsPEMFC outputpowerbymorethan9%[53].Ontheotherhand,thecurrentrippledidnot haveanysignicanteectonsolidoxidefuelcellsSOFC[54].Theeectof dierentcurrentripplesonthesource,thatisproducedbypowerelectronics converters,ontheMFCpowergenerationneedstobeinvestigatedconsideringtheir usageintheMFCenergyharvestingcircuits. 3.1.2Physical-ChemicalParameters MFCelectricalpowergenerationisknowntobeaectedbythereactor physical-chemical-electricalconditions[55,56].Suchconditionscouldhaveaneect onthebacteriainsidethereactorwhichcouldaectthepowergenerationofMFC. StudiesshowedthattheMFCpowergenerationcanbeaectedbythe oxidation-reductionpotentialORP[57,58],andotherphysical-chemical parameters,includingthedissolvedoxygenDO[59],pH[60{62].Theelectrical conductivityECisanotherphysical-chemicalparameterthatisknowntoinuence themaximumpowerthatcanbeextractedfromMFC[63],becausehigherEC 44

PAGE 58

meanslessinternalresistanceandthisincreasestheMFCoutputpower.Therefore, theeectofdierentcurrentripplesonthoseparametersneedstobeinvestigatedin ordertoinvestigatetheeectofpowerelectronicconvertersontheMFCreactor. 3.1.3CurrentRippleTypes PowerelectronicsconvertersthathavebeenusedtoharvestenergyfromMFC canbedividedintotwocategoriesintermsofthecurrentrippletype:converters thatproducetriangular-shaperipplesuchasboostconverters[13]andconverters thatproducesquare-shapecurrentripplesuchasybackconverters[44].Evaluating theeectofsuchripplesonMFCelectricitygenerationisimportanttoevaluatethe usageofpowerelectronicsconvertersasenergyharvesterforMFC. 3.2ExperimentSetup Inthischapter,theeectofdierenttypesofcurrentripplesonMFCpower, energy,andlifespanaswellastheeectofsuchcurrentripplesonthephysical chemicalparametersofMFCreactorareinvestigated.Duringtheexperiments, threedierentcurrentshapescontinuous,triangularshape,andsquareshapewere usedtoharvestenergyfromMFCandtheresultsfromthethreecurrentshapes werecomparedtondtheeectifthereisany.Thecontinuouscurrentexperiment wasusedasareferencesincethecontinuouscurrenthasnoripple,anditis consideredtobeabasicenergyharvestingmethod.Theresultsfromtheother currentshapeswereusedtondtheeectofcurrentripplesonMFC.Ineach experiment,thesolutionofthereactorwasreplacedandonecurrentshapewas extractedfromthereactorforacompletebatchuntilthevoltagegoesdownto around20 mV .Duringtheexperiments,voltage,current,andother physical-chemicalparameterswererecordedandstoredinthecomputer.Inorderto validatetheexperimentalresults,eachexperimentwasrepeatedtwiceandthe conditionswerekeptascloseaspossibletobeabletomakeafaircomparison. 3.2.1Physical-ChemicalSensorProbes 45

PAGE 59

Figure3.1:MFCreactorpictureincludingphysical-chemicalsenors. 46

PAGE 60

Figure3.2:Microcontrollerandcircuitsetup. 47

PAGE 61

Measuringthephysical-chemicalparametersinsidethereactorduringthe experimentswasachievedinreal-timeusingsensorprobes.TheMFCreactorwas designedsuchthatithasfourholesintheanodechamberandfourreal-timesensors ENV-40-PH,ENV-40-DO,ENV-40-EC,andENV-40-ORP,AtlasScientic,NYto measurePH,DO,EC,andORP,respectively,wereinstalledinthosefourholesas showningure3.1.Thosesensorscomeswithacontrolcircuitthatcancanbe controlledwithserialcommunicationcommands.Whenthemeasurementcommand itsenttotheprobecontrolcircuit,itcapturesthemeasurementandsenditthrough thebuilt-inserialcommunicationcircuit.Then,thismeasurementcanbereceived asadigitalnumber.Thecommunicationsendingandreceivingcommandswith thesensorprobeswasachievedusingamicrontrollerTMS320F28335,Texas Instruments,TX,gure3.2usingRS-232.Themeasurementsarethensenttothe computerthroughanotherchannelofserialcommunicationandsavedinatextle tobeusedintheanalysis. 3.2.2CurrentRippleSimulators TostudytheeectofpowerelectronicsconvertersonMFC,dierenttypesof powerelectronicsconvertersneedstousedtoharvestenergyfromtheMFC,andin ordertobeabletocomparetheresults,thoseconvertersneedstoberunningatthe sameoperatingpointMFCvoltageandcurrent.Controllingtheoperatingpointof powerelectronicsconvertersisnoteasybecausetheiroperatingpointchanges dependingontheoutputcapacitorvoltage,consideringxeddutycycle.Forthat reason,andinordertomaintainxedoperatingpoint,thedutycycleofpower electronicsconvertersneedstoberepeatedlychangeddependingontheoutput capacitorvoltage.Forthatreason,asimpleresistor-basedcircuits,thatmimicsthe powerelectronicsconvertersoperation,wereused.Thoseresistor-basedcircuitscan beeasilydesignedtooperateataspecicoperatingpointthatisbasedonthe valuesoftheusedresistors. 48

PAGE 62

Figure3.3:Continuouscurrentcircuitschematicdiagram. HavingrippleonMFCcurrentmeansthattheoperatingpointchanges continuouslydependingontheinstantaneousvalueoftheMFCcurrent.Therefore, forthecomparisonofdierentcurrentripples,theaveragecurrentvalueisusedto designtheresistor-basedcircuits.Havingsimilaraveragecurrentinallexperiments wouldmakeafaircomparisonbetweendierentcurrentripples.Inthese experiments,thereactorwasoperatedatanoperatingpointthatisclosetothe maximumpowerpointMPP,whichisthepointweretheoperationisnormally desired.Continuouscurrentextractionisusedasabaseforthecomparisonsince usingresistorsacrosstheterminalsofMFCisthemostbasicenergyharvesting circuit.HavingaresistorattheterminalsofMFCwillextractacontinuouscurrent thathasnoripple,consideringconstantMFCvoltage,anditsvaluecanbe calculatedby I cont = V MFC R .1 where R istheusedresistanceand V MFC istheMFCvoltage.Aresistanceof375 wasusedinthisexperimentbecausethisresistancekeepstheoperationofthe reactorclosetoMPP.Theschematicdiagramoftheusedcircuitisshowningure 3.3. Generatingthesquare-shapecurrentripple,thatresultswhenusingyback converter,wasachievedbyusingoneresistorandoneMOSFETinseriesasshown 49

PAGE 63

Figure3.4:Square-shapecurrentripplesimulatorcircuitschematicdiagram. ingure3.4.Inthatcircuit,whentheMOSFETisturnedon,thecurrentwillow throughtheresistorandthevalueofthatcurrent,consideringconstantMFC voltage,is V MFC R sq .2 where R sq isthevalueoftheresistanceand V MFC istheMFCvoltage.Whenthe MOSFETisturnedo,currentwillstopowingbecausethecircuitisopened.This meansthat,turningtheMOSFETonandowillresultinsquare-shapecurrent thatissimilartotheybackconvertercurrentripple.Theybackconvertercurrent ripplehappensathighfrequencybecauseofthenatureofpowerelectronics converters.Forthatreason,theusedcircuitneedstobeoperatedatsimilar frequenciesinordertogeneratesimilarwaveforms.Theratioofturningthe MOSFETon T on andonecompletecycle T on + T off isknownasthedutycycle oftheMOSFET,andinthisexperiment,50%dutycyclewasusedwhichmeans thatthecurrentwillowthroughtheresistanceonlyhalfofthetimeineachcycle. Then,theaveragecurrentthatisowingthroughtheresistorcanbecalculatedby I sq ave = V MFC 2 R sq = V MFC R sq eq .3 where R sq eq istheequivalentresistanceand V MFC istheMFCvoltage.This averagecurrentneedstobeequaltothecontinuouscurrentin.1tomaintain 50

PAGE 64

Figure3.5:Triangular-shapecurrentripplesimulatorcircuitschematicdiagram. sameoperatingpoint.Assumingsimilarinputvoltage,thetwocurrentswillbe equalonlyif R sq eq isequalto375,andthatiswhen R sq = 375 2 I sq ave = I cont = V MFC 375 .4 Generatingthetriangular-shapecurrentripple,thatresultsfromusingboost converters,requiresusingoneMOSFET,tworesistors,andoneinductorasshownin gure3.5.Tounderstandtheoperationofthatcircuit,considerthecircuitingure 3.5withouttheinductor.WhentheMOSFETisturnedon,thecurrentwillbypass theresistor R 2 andowonlythroughtheresistor R 1 becausetheMOSFEThas muchlowerresistance.Thevalueofthatcurrentcanbecalculatedby I 1 = V MFC R 1 .5 However,closingtheMOSFETwillforcethecurrenttopassthroughbothresistors becausetheareconnectedinseries.Inthiscase,andsincetheresistanceishigher, thecurrentwillbelower,assumingconstantinputvoltage.Thenewcurrentcanbe calculatedby I 2 = V MFC R 1 + R 2 .6 WhentheMOSFETiscontinuouslyturnedonando,Thecurrentwillswitch betweenthesetwovaluesofcurrent I 1 and I 2 .Includingtheinductanceinthe 51

PAGE 65

circuitchangesthatsquare-shapecurrentintotriangular-shapecurrentbecausethe inductorresistssuddenchangesinthecurrentvalueandthatbehaviorisdescribed bythederivativeintheinductorequation v L t = L dit d t .7 where v L t isthevoltageacrosstheinductor, L istheinductanceinHenry,and i t isthecurrentthroughtheinductor.However,thevalueoftheinductancehaveno eectonthecurrentvalueanditonlycontrolstherateofchangeofthecurrent. Then,theaveragecurrentcanbecalculatedusingthevaluesofthetworesistorsand ignoringtheinductanceinthecircuit.Intheexperiments,thedutycycleofthe MOSFETwassetto50%,andthismeansthatthecurrentwillbe I 1 halfthetime and I 2 theotherhalfofthetime.Then,theaveragecurrentcanbecalculatedfrom .5and.6asfollows I tri ave = I 1 + I 2 2 = V MFC R 1 + V MFC R 1 + R 2 2 = MFC R 1 + R 2 2 R 1 R 1 + R 2 V MFC I tri ave = R tri ave =2 R 2 1 + R 1 R 2 2 R 1 + R 2 .8 where R tri eq istheequivalentresistanceofthecircuit.Theresistors R 1 and R 2 needstobechosesuchthattheaveragecurrentisequaltothecontinuouscurrentin .1.Intheexperiments,theresistors R 1 and R 2 areselectedtobe250and 500,respectively,whichmakes R tri eq =375andthatshouldproducesimilar averagecurrentascontinuousandsquare-shapecurrentsin.1and.3 I sq ave = I cont = I tri ave = V MFC 375 .9 3.2.3Microcontroller 52

PAGE 66

Tocontroltheusedcircuits,amicrocontrollerTMS320F28335,Texas Instruments,TX,gure3.2wasused.Thismicrocontrollerwasusedgeneratethe gatesignalsfortheMOSFETinthesimulatorcircuitsusingthebuilt-intimerthat cangeneratePWMsignals.Anotherfunctionofthemicrocontrolleristomeasure theMFCaveragevoltagethroughouttheexperimentsusingthebuilt-inanalogto digitalconverterADC.ItshouldbenotedthatsincetheMFCvoltagevalueis smallcomparedtotheVCCofthemicrocontroller,anamplifyingcircuitwasused wasusedtoamplifyMFCvoltagetoanappropriatelevel.Alowpasslterwasalso includedintheamplifyingcircuitandthatlowpasslterremovestheripplesand givestheaveragevoltage.Thedetailsoftheusedcircuitwasdescribedinchapter2. Themeasurementsoftheaveragevoltagewascapturedevery3-4secondsanditwas senttothecomputerthroughserialcommunication. Anothertaskforthemicrocontrolleristocommunicatewiththe physical-chemicalsensorsandcollectthemeasurementsfromtheprobes.Those measurementsweresenttothecomputeralongwiththeaveragevoltage measurement.Therealtimeforeachmeasurementwasalsorecordedbythe microcontrolleranditwassenttothecomputerwhichwasusedtoplottheresults. 3.3PowerandEnergyCalculations Usingresistor-basedcurrentripplesimulatorstosimulatetheeectofpower electronicsconvertersiseasiersincepowerelectronicsconvertersarenoteasytobe controlled.However,thepowerandenergycalculationswhensuchcircuitsareused isnotstraightforwardanditneedstobecarefullyaddressed.Theexpected waveformsofthevoltageandcurrentwithtriangularandsquareshapecurrent ripplesareshowningure3.6.Itisclearthatthevaluesofthevoltageandcurrent areinstantaneouslychangingwhichneedstobeconsideredwhencalculatingpower andenergysincetheaveragevaluesweremeasuredduringtheexperiments. Inthecontinuouscurrentextraction,thevoltageaswellasthecurrentare 53

PAGE 67

Figure3.6:Expectedcurrentandvoltagewaveforms. constantbecausetheoutputresistanceisconstant.Becauseofthatreason,the averagevoltageandcurrentareequaltotheinstantaneousvoltageandcurrent. Then,usingtheaveragevoltage,thepowercanbecalculatedby P cont = V 2 MFC R .10 where V MFC istheterminalvoltageand R istheresistance. Theterminalvoltageofthetriangular-shapecurrentextractionisvariableand changesdependingontheinstantaneousvalueofthecurrentasshowninthe expectedwaveformsingure3.6.However,sincethecurrentiscontinuouslyowing outoftheMFC,theaveragepowerthatisharvestedfromthereactorcanbe calculatedusingtheaveragevoltageandtheequivalentresistance R tri eq bythe 54

PAGE 68

followingequation P tri ave = V 2 MFC ave R eq tri .11 where V MFC ave istheaverageterminalvoltageinthecaseoftriangular-shape currentrippleand R eq tri istheequivalentresistance5. Theexpectedvoltageandcurrentwaveformsinthecaseofthesquare-shape currentextractionareshowningure3.6.Thevoltages V 1 and V 2 arethevoltages whentheswitchisoandon,respectively.Whentheswitchiso,thecurrentis zeroandthismeansthatthepowerisalsozeroeventhoughthevoltageis V 1 .On theotherhand,whentheswitchison,currentstartsowingthroughtheresistance andthepowerisnotzero.Thatpowercanbecalculatedusing V 2 andthe R sq as follows P sq = V 2 2 R sq .12 Then,theaveragepowerthatisharvestedfromthereactor,assuming%50duty cycle,canbecalculatedasfollows P sq ave = V 2 2 2 R sq .13 Noticethatusingtheaveragevoltage,whichis V 1 + V 2 2 ,tocalculatetheaverage powerwouldresultinapowerthatishigherthantheactualpowerbecausethe powerisdeliveredonlyhalfofthetimewhentheMOSFETison.Thisisshownin thefollowinganalysis V ave = V 1 + V 2 2 .14 V ave = V 1 + V 2 2 P = V 2 ave 2 R sq = V 2 2 2 R sq + V 2 1 +2 V 1 V 2 2 R sq .15 55

PAGE 69

Itisclearthatthepowerin.15thatresultsfromusingtheaveragevoltageis higherthantheactualpowerthatisdeliveredtotheloadin.13.Forthatreason, andsinceonlytheaverageterminalvoltagewasmeasuredduringtheexperiments, thevoltage V 2 needstobeestimatedinordertocalculatetheactualpower.Tond theestimationof V 2 fromtheaverageterminalvoltage,thevoltagedierence V thatisshowningure.6wasusedtoestimatethevoltage V 2 asfollows V 2 = V ave )]TJ/F15 11.9552 Tf 13.15 8.088 Td [( V 2 .16 Thevalueof V wasmonitoredusingtheoscilliscopeduringexperiments,andit wasapproximatelyconstant 100 mV allthetime.Thevoltagethatresultsfrom .16,whichisanapproximationofthevoltage V 2 ,wasusedtocaculatethe averagepowerusing.13. Theenergyforallthreetypesofcurrentshapescanbecalculatedbyintegrating theaveragepowerin.10,.11,and.13overtimeasfollows E t = Z P t dt .17 where P t istheaveragepowerattime t .Thisintegrationwasimplemented numerically,wheretherealtimeforeachmeasurementwascapturedbythe microcontrolleranditwasstorednexttoeachmeasurementandfromthatreal time,thetimedierencet betweeneachtwomeasurementswascalculated.Then, thepowerforeachmeasurementwasmultipliedbytheits dt andthetotalenergy wascalculatedbysummingthosemultiplications. 3.4ExperimentalResults Theproposedcurrentripplesimulatorswerebuiltinthelabandusedtoharvest energyfromMFCforacompletebatch 2days.Duringtheexperiments,the averagevoltagewasmeasuredandrecordedevery3to4susingthemicrocontroller. ThegatesignalsfortheMOSFETsinthecurrentripplesimulatorswererunningat 56

PAGE 70

Figure3.7:Experimentalwaveformsofcontinuouscurrentextraction. 1.25kHzfrequencywhichisconsideredtobealowswitchingfrequencyfromthe viewpointofpowerelectronicsconverters.Thereasonforchoosingsuchlow frequencyistoincreasetheimpactofcurrentripplesonmicrobiologyandchemistry ofthereactor. 3.4.1PowerandEnergy Thecurrentandvoltagewaveformsthatresultedfromusingcurrentripple simulatorstoharvestpowerfromMFCwerecapturedasscreenshotsfromthe oscilloscopeTPS2014,Tektronix,OR.Thosewaveformsareshowningures3.7, 3.8,and3.9forcontinuous,triangular-shape,andsquare-shapecurrents, respectively.Itisclearthatthesewaveformsaresimilartotheexpectedwaveforms ingure.6whichmeansthatthepowercalculationanalysis,thatwasdiscussed insection3.3,canbeusedtocalculatetheMFCpowerandenergy. Theaverageterminalvoltagethatwasrecordedduringexperimentsisshownin gure3.10,whereitshowstheresultsforallsixexperimentsthreedierenttypes ofcurrentrippleandtwoexperimentsforeachcurrentrippletype.Thevoltage startedfromalowvoltagerightafterreplacingthereactorsolutionandstartedto 57

PAGE 71

Figure3.8:Experimentalwaveformsoftriangular-shapecurrentextraction. 58

PAGE 72

Figure3.9:Experimentalwaveformsofsqaure-shapecurrentextraction. 59

PAGE 73

Figure3.10:MFCvoltageexperimentalresults. increaseuntilitnallystaysatalmostconstantvoltage.Thatconstantvoltagewas similarforallexperimentsexceptforthesquare-shapecurrentripple,wherethe voltagewashigherthantheothers.Thevoltagewashigherinthecaseof square-shapecurrentripplebecausethatvoltageistheaveragevoltagein.14, anditwasshownthatthisaveragevoltageisexpectedtobehigherthanthe voltagesofothercurrentrippleshapes.Thisvoltageneedstobeadjustedinorder tocalculatethecorrectpowerasdiscussedinsection3.3.Adjustingtheaverage voltageofthesquare-shapecurrentrippleexperimentwasachievedbymonitoring thevoltagedierence V ingure3.6usingtheoscilloscopethroughoutthe experiments.Thevoltagedierence V wasapproximatelyconstantataround 100 mV andthatisclearintheosilloscopescreenshotingure3.9 Theharvestedpowerwascalculatedforallthreecurrentshapesusing.10, .11,and.13andtheresultsareshowningure3.11.Itisclearthatthe generatedpowerfromcontinuousandtriangular-shapecurrentexperimentsis similar,butsquare-shapecurrentexperimentsgeneratedlesspower.Thethree experimentsareexpectedtogeneratesimilarpowerasanalysedbefore,butthiswas 60

PAGE 74

Figure3.11:MFCpowerexperimentalresults. Figure3.12:Harvestedenergyexperimentalresults. 61

PAGE 75

notthecasewiththeexperiments.Thereasonbehindthatchangeinthepower levelisthattheanalysiswasbasedontheassumptionthattheterminalvoltageis constant.Thisassumptionignoresthevoltagedropacrosstheinternalresistanceof MFCwhenthecurrentstartsowing.Thisinternalvoltagedropresultedin operatingatadierentoperatingpointterminalvoltage,andthatinternal resistancevoltagedrophavemoreeectonthesquare-shapecurrentexperiment becausethevalueofthecurrentishigher,whentheswitchison,thantheothertwo experiments.Figure3.9showstheoscilloscoperesultsanditisclearthatwhenthe switchison,theterminalvoltagewasaround220mVcomparedto245mVinthe continuouscurrentexperiment.Thismeansthattheoperatingpointterminal voltageinthecaseofsquare-shapecurrentrippleisdierentthantheoperating pointterminalvoltageoftheothertwocurrentshapes.Havingdierentoperating pointmeansthatthepowerwillbedierentaswell,andsincethesquare-shape currentrippleexperimentwasrunningatalowerpoweroperatingpoint,itspower waslowerthantheothertwocurrentshapes.Fromthisanalysis,itisclearthatthe currentrippledidnothaveasignicanteectontheharvestedMFCpower. Theharvestedenergywascalculatedforallthreecurrentshapesusing.17 andtheresultsareshowningure3.12.Notethatthesquare-shapecurrentenergy waslowerthantheothersbecausetheharvestedpowerisless,andthismeansthat theharvestedenergymustbeless,andthatismainlybecauseofrunningatalower poweroperatingpoint.TherunningtimeoftheMFCreactorlongevityisanother factorthatcouldaecttheamountofharvestedenergy.Ifonecurrent-shapecaused thereactortorunforshortertimebatchtime,thenthatcurrent-shapehasa negativeeectonthereactorperformance.Theresultsingure3.12showsthatall experimentsrunforasimilartimeperiodsexceptthesecondexperimentof continuousandtriangular-shapecurrents,wherethereactorwasrunningfor15% moreandlessperiodsoftime,respectively.Forthatreason,theharvestedenergy 62

PAGE 76

Figure3.13:AnodechamberpHexperimentalresult. wasdierentinthosetwoexperimentsthanotherexperimentsascanbeseenin gure3.12.Sinceonlyoneofthecontinuouscurrentexperimentsandoneofthe triangular-shapecurrentrippleexperimentshavedierentoperationlongevity,itis expectedthatdierentoperationalparameters,suchasdierentsodiumacetate levelsorganicmatter,causedthatdierenceinrunningtime. 3.4.2Physical-ChemicalParameters TheresultsfromthepHsensorprobeforallexperimentsareshowningure 3.13.Ingure3.13,theupperthreecurvesrepresenttheresultsfromtherst experimentsetcontinuous,square-shape,andtriangular-shape.Itisclearthatall thesethreecurvesaresimilar,wheretheystartwithapHvalueofaround7.5and theystartincreasingwhenenergyisharvestedfromthereactor.Ontheotherhand, thelowersetofcurvesrepresentsthesecondexperimentset.Inthatset,allthree curvesaresimilarandtheyhavesametrendastherstsetofexperiments.Notice thatthevaluesofthepHinthesecondexperimentsetareslightlydierentthanthe rstexperimentsetandthatcouldbebecauseofdierentreactorconditionatthe timeoftheexperiments.Inconclusion,thecurrentripplehasnosignicanteecton 63

PAGE 77

Figure3.14:Anodechamberlectricalconductivityexperimentalresult. thepHvalueoftheMFCreactor. TheelectricalconductivityECproberesultsareshowningure3.14.Itis clearthatallexperimentsshowsimilartrend.However,thevalueofECinthe square-shapecurrentextractionislowerthanotherexperiments,butingeneral,the ECshowsnodierencewhendierentcurrentshapesareextractedfromthe reactor.Thismeansthatthecurrentripplehavenosignicanteectonthe reactor'selectricalconductivity. TheoxidationreductionpotentialORPresultsingure3.15showsthatthe valueofORPstartedfromavalueofaround-0.1andincreasedto0.1oncethe energyharvestingstartedanddecreasedbacktoaround-0.5duringtheexperiments. Theresultswereconsistentinallexperiments,whichmeansthatthecurrentripple hasnosignicanteectontheoxidationreductionpotentialoftheMFCreactor. Thelastphysical-chemicalparameterthatwasmeasuredduringtheexperiments isthedissolvedoxygenDOandtheresultsforthatareshowningure3.16.The resultsshowthattheDOincreasedtoalevelofaround3inthebeginningofthe experimentsandthendecreasedtoalevelof0.5throughouttheexperiments.From 64

PAGE 78

Figure3.15:Anodechamberoxidationreductionpotentialexperimentalresult. Figure3.16:Anodechamberdissolevedoxygenexperimentalresult. 65

PAGE 79

theresultsingure3.16,itisclearthatthecurrentripplehasnosignicanteect onthedissolvedoxygenoftheMFCrector. 3.5Conclusion Fromtheresultsoftheexperiments,itwasshownthatthecurrentripplethatis imposedbypowerelectronicsconvertersshouldnothaveasignicanteectonthe MFCpower,energy,andoperationlongevity.Inaddition,thephysical-chemical probesresultsshowedthatpowerelectronicsconverter'scurrentripplehaveno signicanteectonthereactor'spH,EC,ORP,andDO.Inconclusion,usingpower electronicsconvertersasenergyharvestersforMFChasnosignicanteectonthe reactorperformance. 66

PAGE 80

CHAPTERIV NETPOWERPOSITIVEMAXIMUMPOWERPOINTTRACKING POWERMANAGEMENTSYSTEMFORMICROBIALFUELCELL 4.1Introduction MostrenewableenergysourcesrequireusingenergyharvestingsystemEHSin ordertoharvesttheavailableenergy.TheEHSmustbeveryecientinordertobe abletousemostoftheavailableenergy.Buildingaself-poweredEHSisimportant tomakeanetpowerpositiveenergysource.IncreasingtheeciencyofEHScanbe achievedbyminimizingthenumberofcomponentsandonlyusingecientones.In addition,theenergyharvestingalgorithm,thatisusedtocontroltheEHS,playsan importantroleintheeciencyofthesystemanditneedstobewelldesignedto reducethelosses. MicrobialfuelcellMFCisoneoftherenewableenergysources,wherethe bacteriainsidethereactorconsumeorganicmatterandgenerateelectricity.The amountofpowerthatisextractedfromMFCdependsontheoperatingpoint;there isaspecicoperatingpoint,maximumpowerpointMPP,inwhichthemaximum powercanbeextracted[18].ThisMPPcanchangeduringoperationdependingon variousreactorconditions,hence,itneedstobetrackedtokeepextractingthe maximumpossiblepowerfromthegivenreactorconditions[18].variousMPP trackingMPPTalgorithmshavebeenused[28,38,39,64{66].However,previous worktypicallydidnotconsiderthepowerconsumedbythecontrolcircuits;for example,apersonalcomputer[28,38,39,64]andpowerfulmicrocontrollers[65,66] wereusedtodotherequiredcalculations,butnotincludedintheoveralleciency calculation.Botharenotpracticalbecausetheyrequiresubstantialpower consideringwhatMFCcouldgenerate.ThepowerconsumptionofMPPTalgorithm mustbelowenoughfortheoverallsystemtobenetpowerpositive. Designingself-poweredEHSforMFCisachallengesincetheavailableMFC powerissolowthatusingsomeofthatpowertoenergizetheEHScontrolcircuit 67

PAGE 81

signicantlyreducestheoveralleciency,evendowntoanimpracticallevel. VariousMFCself-poweredEHShavebeenreported[31,67{69];however,theyshare theideaofchargingasupercapacitoruntilcertainvoltageleveleitherdirectlyor usingachargepumpbeforeadc-dcconverterisenabledtoboostthevoltage.The disadvantageofthisapproachisthatthereisnocontrolovertheoperatingpointof MFC,hence,thesustainedoperationatMPPisnotpossible,whichresultsin harvestinglesspowerthantheMFCcansupply.Self-poweredEHSsthatcanbe controlledtooperateatMPPwerereportedin[70,71],butthemethodthatwas usedtondMPPisnotpracticalsinceitassumedthattheopencircuitvoltage OCVoftheMFCisxedat720mV.TheOCVcouldchangedependingthe reactorconditionduringtheoperation[26]. Anideaofminimizingtheenergyconsumptionofthecontrolcircuitwas discussedin[72],whichisbasedonincreasingthesamplingtimeandspendingmore timeinthe sleepmode toreducepowerconsumptionwhenusinglowpower microcontrollers.Thealgorithmwastestedusingpersonalcomputer,butthe samplingtimeisperiodicallycalculatedusingafunctionthathasanexponential termandthisisdiculttobeimplementedinalowpowermicrocontroller. Inthispaper,aself-poweredEHSthatusesanMPPTalgorithmisproposed. TheexperimentalresultsshowthattheproposedMPPTalgorithmwasableto reachandmaintaintheoperationaroundtheMPPwitheciencyupto59.4%with 119 WMFCpower.TheproposedMPPTalgorithmusestheMFCelectrical characteristicstoestimatethepowerbyonlymeasuringvoltages,whichsavesthe powerthatisconsumedbycurrentmeasurementcomponents.Thatalgorithmis implementedinanultra-lowpowermicrocontrollerusestheMFCpower. Thisworkisorganizedasfollows.AgeneraldescriptionofMFCsandtheir electricalcharacteristicsarediscussedinsection4.2.Thedesignandconsiderations fortheproposedMPPTalgorithmaredescribedindetailsinsection4.3.The 68

PAGE 82

Figure4.1:MFCconceptualdiagram. proposedself-poweredEHSisdescribedinsection4.4andtheexperimentalresults arepresentedinsection4.5.Finallyaconclusionisprovidedinsection4.6. 4.2MicrobialFuelCells ElectricitygenerationinMFCsisachievedusinganaerobicbacteriasuchas Geobacter and Shewanella ,wheretheyconsumeorganicmatterinsidetheanode chamberoftheMFCreactorandgenerateextracellularelectronsthatcanbe transferredfromtheanodeelectrodetothecathodeelectrode.Whenanexternal loadisconnectedbetweentheanodeandcathodeelectrodes,itresultsindelivering DCpowertotheload[18{20].AconceptualdiagramofMFCisshowningure4.1. TheamountofpowerthatisavailablefromMFCislow;thehighestpowerdensity reportedsofaris6.9W/m 2 [13].However,theyhavebeenconsideredasapower sourceforvariousapplicationssuchasunderwaterremotewireless sensors[10,73,74]. 4.2.1EquivalentCircuit TheMFCequivalentcircuitthatwasproposedin[26]isshowningure4.2,and itisusedtomodeltheelectricalcharacteristicsofMFCinthispaper.This 69

PAGE 83

equivalentcircuitrepresentstheMFCvoltagebehaviorafteropeningthecircuit withashuntcapacitor;becauseittakestimefortheterminalvoltagetochange, ThisvoltagebehaviorisimportantintrackingMPP. Theinternalvoltage V int oftheequivalentcircuitcanbefoundbyopeningthe terminalsoftheMFCforlongenoughtime,whichiscalledopencircuitvoltage OCV.For R 2 ,aloadneedstobeconnectedtotheterminalsoftheMFCandthe terminalvoltage V )]TJ/F21 7.9701 Tf -2.601 -7.645 Td [(t andoutputcurrent I )]TJ/F15 11.9552 Tf 10.986 -4.338 Td [(aremeasured.Then,thecircuitis openedandterminalvoltage V t ismeasuredagain,whichisthevoltage V C .Itis assumedthatthecapacitanceissohighthattheoutputterminalvoltageis reasonablycloseto V C eveniftheMFCcurrentstartschargingthecapacitorwhen thecircuitopens.Theresistance R 2 canbecalculatedas R 2 = V C )]TJ/F20 11.9552 Tf 11.955 0 Td [(V )]TJ/F21 7.9701 Tf -2.601 -7.645 Td [(t I )]TJ/F15 11.9552 Tf 177.735 4.747 Td [(.1 Thevalueoftheresistance R 1 canbecalculatedbysubtractingthevalueof R 2 from thattotalinternalresistance,whichcanbefoundfromthepolarizationcurve consideringthattheexternalresistanceisequaltotheinternalresistanceatthe MFCMPP.Then,thetotalinternalresistancecanbecalculatedusing R int = V MPP I MPP .2 where V MPP and I MPP arethevoltageandcurrentattheMPP,respectively.Then, R 1 canbefoundusing R 1 = R int )]TJ/F20 11.9552 Tf 11.955 0 Td [(R 2 .3 Nowthatthevaluesof V int R 1 ,and R 2 arecalculated,thecapacitanceoftheshunt capacitorcanbefoundusingthosevaluesalongwiththeopencircuittestresults. Assumethatwestartfromaninitialcapacitorvoltage V C )]TJ/F15 11.9552 Tf 7.084 -4.339 Td [(whenaloadwas connectedtotheterminalsofMFC.Whenthecircuitisopened,thecapacitor voltagewillstartincreasingbecauseofthedynamicsofthecapacitorvoltagethatis describedasfollows. 70

PAGE 84

Figure4.2:MFCelectricalequivalentcircuit[26]. C dV C t dt = V int )]TJ/F20 11.9552 Tf 11.955 0 Td [(V C t R 1 .4 dV C t dt + V C CR 1 = V int CR 1 .5 Convertedtofrequencydomainwhichbecomes sV C s )]TJ/F20 11.9552 Tf 11.955 0 Td [(V C )]TJ/F15 11.9552 Tf 7.085 -4.936 Td [(+ V C s CR 1 = V int CR 1 1 s .6 V C s s 2 + 1 CR 1 s = V int CR 1 + V C )]TJ/F15 11.9552 Tf 7.084 -4.936 Td [( s .7 V C s = V int CR 1 + V C )]TJ/F15 11.9552 Tf 7.085 -4.339 Td [( s s 2 + s CR 1 .8 andcanbesimpliedas V C s = V int s + V C )]TJ/F15 11.9552 Tf 7.085 -4.339 Td [( )]TJ/F20 11.9552 Tf 11.955 0 Td [(V int s + 1 CR 1 .9 Byconvertingbacktothetimedomain, V C t = V int + V C )]TJ/F15 11.9552 Tf 7.085 -4.936 Td [( e )]TJ/F18 7.9701 Tf 6.587 0 Td [( 1 CR 1 t )]TJ/F20 11.9552 Tf 11.955 0 Td [(V int e )]TJ/F18 7.9701 Tf 6.586 0 Td [( 1 CR 1 t .10 Thiscapacitorvoltageequationcanbeusedtocalculatetheshuntcapacitancefrom theopencircuittestresults.AloadshouldbeconnectedtotheterminalsofMFC tohavecurrentowingbeforetheopencircuittest.Oncethecircuitisopened, V C 71

PAGE 85

Table4.1:ResultsofMFCtests. Polarizationcurveresults Shortcircuittestresults Opencircuittestresults V OCV V MPP I MPP V t V C I V C )]TJ/F15 11.9552 Tf 7.085 -4.339 Td [( V C t t 425mV212.5mV0.5539mA 176mV208mV0.38mA 331.1mV411mV40s Table4.2:Resultsofidentiedequivalentcircuitparameters. Equivalentcircuitparameters R int R 2 R 1 C 378.6584.21294.4471.4mF ismeasuredtwotimes;therstvoltageisconsideredtobetheinitialcapacitor voltage V C )]TJ/F15 11.9552 Tf 7.085 -4.338 Td [(,andthesecondvoltage,whichismeasuredafter t seconds,is V C t Thosemeasurementscanbeusedin.10alongwith V OCV R 1 ,and R 2 tondthe valueof C Theresultsofthereactortestsandidentiedparametersareshownintable4.1. 4.2.2Steady-StateAnalysis Theinternalcapacitorvoltage V C dependsontheterminalvoltage V t andthe parametersoftheequivalentcircuitingure4.2.Thevalueof V C insteadystate, assumingconstant V t ,canbefoundas V C = V int )]TJ/F20 11.9552 Tf 11.955 0 Td [(IR 1 .11 where I isthecurrentthatisowingthroughtheresistors R 1 and R 2 I = V int )]TJ/F20 11.9552 Tf 11.955 0 Td [(V t R 1 + R 2 .12 Hencethecapacitorvoltagecanbegivenas V C = V int )]TJ/F20 11.9552 Tf 13.15 8.088 Td [(V int )]TJ/F20 11.9552 Tf 11.955 0 Td [(V t R 1 + R 2 R 1 .13 72

PAGE 86

V C = V int R 2 + V t R 1 R 1 + R 2 .14 Noticethatthisvoltagedependsonthevalueoftheterminalvoltage,where increasingordecreasingtheterminalvoltageincreasesordecreasesthecapacitor voltage,respectively,assumingconstantequivalentcircuitparameters. 4.2.3TransientAnalysis Thetransientofthecapacitorvoltagewhentheterminalvoltageischangedfrom aninitialvalueof V t 0 toavalueof V t canbeanalyzedusingthecapacitorvoltage equation C dV C dt = V int )]TJ/F20 11.9552 Tf 11.955 0 Td [(V C R 1 )]TJ/F20 11.9552 Tf 13.151 8.088 Td [(V C )]TJ/F20 11.9552 Tf 11.955 0 Td [(V t R 2 .15 whichcanberearrangedtobe dV C dt + R 1 + R 2 CR 1 R 2 V C = V int R 2 + V t R 1 CR 1 R 2 .16 ThisisarstorderdierentialequationthatcanbesolvedusingLaplace transformation.TheLaplacetransformof.16is sV C s )]TJ/F20 11.9552 Tf 11.955 0 Td [(V C )]TJ/F15 11.9552 Tf 7.084 -4.936 Td [(+ R 1 + R 2 CR 1 R 2 V C s = V int R 2 + V t R 1 sCR 1 R 2 .17 s 2 V C s + R 1 + R 2 CR 1 R 2 sV C s = sV C )]TJ/F15 11.9552 Tf 7.084 -4.936 Td [(+ V int R 2 + V t R 1 CR 1 R 2 .18 V C s s 2 + R 1 + R 2 CR 1 R 2 s = sV C )]TJ/F15 11.9552 Tf 7.084 -4.936 Td [(+ V int R 2 + V t R 1 CR 1 R 2 .19 V C s s s + R 1 + R 2 CR 1 R 2 = sV C )]TJ/F15 11.9552 Tf 7.085 -4.936 Td [(+ V int R 2 + V t R 1 CR 1 R 2 .20 Theinitialcondition V c )]TJ/F15 11.9552 Tf 7.085 -4.339 Td [(canbecalculatedusing.14when V t = V t 0 V c )]TJ/F15 11.9552 Tf 7.085 -4.936 Td [(= V int R 2 + V t 0 R 1 R 1 + R 2 .21 Thecapacitorvoltage V C s isgivenas V C s = V int R 2 + V t 0 R 1 R 1 + R 2 s + V int R 2 + V t R 1 CR 1 R 2 s )]TJ/F20 11.9552 Tf 5.479 -9.684 Td [(s + R 1 + R 2 CR 1 R 2 .22 = A s + B s + R 1 + R 2 CR 1 R 2 .23 73

PAGE 87

where A = V int R 2 + V t R 1 R 1 + R 2 ;B = )]TJ/F15 11.9552 Tf 14.934 8.087 Td [( VR 1 R 1 + R 2 .24 and V = V t )]TJ/F20 11.9552 Tf 11.955 0 Td [(V t 0 .25 Then, V C s becomes V C s = V int R 2 + V t R 1 R 1 + R 2 1 s )]TJ/F18 7.9701 Tf 25.941 14.705 Td [( VR 1 R 1 + R 2 s + R 1 + R 2 CR 1 R 2 .26 andinverseLaplacetransformationof.26yields V C t = V int R 2 + V t R 1 R 1 + R 2 | {z } steadystatevalue )]TJ/F15 11.9552 Tf 16.927 8.088 Td [( VR 1 R 1 + R 2 e )]TJ/F26 11.9552 Tf 6.586 9.683 Td [()]TJ/F22 5.9776 Tf 6.728 -5.359 Td [(R 1 + R 2 CR 1 R 2 t | {z } decaystozeroas t !1 .27 Noticethatthersttermisthesteadystatevalueof V C t thatwasshownin.14 andthesecondtermisanexponentialtermthatdecaystozeroaftersometime. 4.3MaximumPowerPointTracking TypicalMPPTtrackingalgorithmsrequirethecalculationofMFCpowerfor eachoperatingpoint,fromMFCvoltageandcurrentmeasurements.Measuring MFCvoltageisstraightforwardsincemicrocontrollerscandirectlymeasurevoltages, whilecurrentmeasurementisnot.Itrequiresashuntresistortomeasurethevoltage acrossit.Asmallresistanceisrecommendedforshuntresistorinordertoreducethe powerlossandvoltagedrop,anampliermightbeneededtomeasurethelowshunt voltage,whichcanconsumerelativelyhighpowercomparedtoMFCpower.In[66], theinstrumentalamplierINA122,TexasInstruments,TXwasused,whichtakes 60 Aoperatingat2.2VwhiletheMFCgeneratesabout400 W.Forthisreason, anMFCoutputcurrentestimationmethodusingthecharacteristicsofMFCis proposedinthischaptertominimizethepowerconsumptionofthecontrolcircuit. 4.3.1ProposedAlgorithm 74

PAGE 88

Ithasbeenshownthatmovingfromoneoperatingpointtoanothertakestime becauseofthehighinternalcapacitance.Thistimeshouldtobeconsideredwhen trackingMPPbecausemeasuringMFCcurrentbeforethecapacitorvoltagereaches steadystatewillaectthepowercalculationandmisleadtheMPPTalgorithm. Therefore,theproposedMPPTalgorithmwasdesignedconsideringthedynamic behavioroftheMFCinternalcapacitor. TheproposedMPPTalgorithminthispaperisbasedontheperturband observeP&Oalgorithm[28,45,64].Itmeasures V t and V C atoneoperatingpoint andcalculatestheproportionalestimatedpower.Thentheoperatingpointis changed,andaftersometime T s V t and V C ofthenewoperatingpointare measuredandtheproportionalestimatedpoweriscalculated.Bycomparingthe powerofthetwooperatingpoints,theoperatingpointischangedinthedirection thatgiveshigherpower.Aowchartthatdescribestheoperationoftheproposed algorithmisshowningure4.3. ThecontributionoftheproposedalgorithmontheoriginalP&Oalgorithmis thepowerestimationwhichisbasedonthecurrentestimation,thederivationof minimumsamplingtime T s thatguaranteesthemeasurementofthesteadystate power,andimprovmentoftheestimationreliability. 4.3.2CurrentEstimation Inthispaper,estimatedcurrentisusedinsteadofactualcurrentmeasurementin ordertoincreasetheoveralleciencyofthecontrolcircuitbecausemeasuringthe actualcurrentrequiresadditionalcomponentsandpower.Theproposedcurrent estimationmethodusestheequivalentcircuitmodelingure4.2.Theloadcurrent isgivenas I MFC = V C )]TJ/F20 11.9552 Tf 11.955 0 Td [(V t R 2 .28 andthecurrentcanbeestimatedwith V t and V C fromtheopencircuittest assumingthattheresistance R 2 isconstant,whichisareasonableassumption[26]. 75

PAGE 89

Figure4.3:FlowchartoftheproposedMPPTalgorithm. 76

PAGE 90

Figure4.4:Graphicalrepresentationof V t and V c measurementprocess: V t ismeasuredjustbeforeopeningthecircuit, V C ismeasuredrightafterthecircuitisopened, t timerequiredtomeasure V C ,and T s isthesamplingtime. Agraphicalrepresentationthatshowstheprocessofmeasuring V t and V C isshown ingure4.4. Usingthismethodforcurrentestimationreducestheenergyconsumptionofthe controlcircuitbecauseonlythesimplevoltagemeasurementsarerequired. AlthoughtheMFCenergywillnotbeharvestedduringtheopencircuitperiod t thatenergylossispracticallynegligibleconsideringthatitisveryshortaround 15msinthisworkcomparedtothesamplingtime T s s. WhentheMFCterminalvoltageisreducedbytheMPPTalgorithm,the capacitorvoltagewilldecreasetothenewsteadystatevalue,asshownin.14. TheMFCcurrentshouldincreaseascanbeseeninthepolarizationcurve,butthe outputcurrentwillbehigherthanthesteadystatecurrentbecausetheenergy storedinthecapacitorsuppliestheoutputcurrentduringthetransient.This continuesuntilthecapacitorvoltagereachesthesteadystatevalue.Ontheother hand,whentheMFCterminalvoltageisincreased,thecurrentisreducedbeforethe capacitorvoltagereachesthesteadystatevalue.Duringthistransient,someof 77

PAGE 91

MFCenergyisstoredinthecapacitorandthisexplainstheloweroutputcurrent. TheeectofthecapacitorvoltagetransientonMFCcurrentisimportant becausemeasuringtheMFCcurrentbeforethecapacitorvoltagereachesthesteady statevaluewillresultinpowermeasurementerror.Forthatreason,thesampling time T s shouldbelargeenoughtoallowthecapacitorvoltagetoreachthesteady statevalue. Theminimumsamplingtimeforthecapacitorvoltagetoreachitssteadystate valuecanbeextractedfrom.27.Thetimeconstantofthecapacitortransient equationis = CR 1 R 2 R 1 + R 2 .29 Then,inordertomeasurethesteadystatepower,thesamplingtime T s needstobe T s 5 .30 wherethecapacitorvoltageneedsvetransienttimeconstantstoreach99%ofthe steadystatevalue. FortheusedMFCreactorinthisworktheminimumsamplingtimeis determinedas23.31sbasedontheequivalentcircuitparameters.gure4.5shows thecalculatedcapacitorvoltage,using.27,whentheterminalvoltagewas increasedfrom220mVto240mV.Notethattheinitialandsteadystatevoltages arehigherthan220mVand240mV,respectively,becauseoftheloadcurrent.It canbeseenthatthecapacitorvoltagereachesthesteadystatevalueafteraround 23s,whichisclosetotheminimum T s in.30. 4.3.3PowerCalculationandMPPT Basedonthecalculatedcurrent,theMFCpowercanbeobtainedby P est = V c )]TJ/F20 11.9552 Tf 11.955 0 Td [(V t R 2 V t .31 However,theactualvalueofthepowerisnotnecessaryfortheoperationofthe P&Oalgorithmiftherelativeamountofthepoweroftwooperatingpointshigher 78

PAGE 92

Figure4.5:Thecalculatedcapacitorvoltagewhentheterminalvoltagewasincreased from220mVto240mVbasedontheequivalentcircuitmodel. orlowerisknown.Therefore,thevalueof R 2 canberemovedfrom.31,which resultsinaproportionalestimatedpower ~ P est = V c )]TJ/F20 11.9552 Tf 11.955 0 Td [(V t V t .32 ThealgorithmwillconvergetoMPPevenif R 2 changes,assumingthatitdoesnot changebetweentwoconsecutivemeasurements,whichisareasonable assumption[26]. 4.4EnergyHarvestingSystem Intheproposedself-poweredEHS,theMFCpowerishandledusingcommercial o-the-shelfboostconverterBQ25504,TexasInstruments,TX.Thisconverteris ecientandcanbeusedtoharvestenergyfromMFCsatlowvoltage.Controlling theboostconverterisachievedusingamicrocontrollerMSP430L092,Texas Instruments,TXwiththeproposedMPPTalgorithm.Thismicrocontroller consumesverylowpower,whichisappropriatefortheimplementationofthe 79

PAGE 93

Figure4.6:SchematicdiagramoftheoverallEHSsystem. proposedself-poweredEHS.TheoverallschematicdiagramoftheproposedEHSis showningure4.6. 4.4.1BoostConverter BQ25504isdesignedtooperateatlowinputvoltagesaslowas100mV.This voltagerangecoverstypicalMFCvoltage,whichmakesitagoodchoiceforanEHS aswellasitshigheciency.Theboostconverterhasabuilt-inMPPTfunctionthat ndstheOCVandsetstheoperatingpointtohalfoftheOCV.However,theOCV ismeasuredafteropeningthecircuitfor256ms,whichisnotenoughforanMFCto reachitsOCV.Ithasbeenshownthatapplyingthisbuilt-inMPPTfunctiontoan MFCresultsinanoperatingpointfarfromtheMPPofthereactor[72].Theinput voltageofBQ25504i.e.,MFCoutputvoltagecanbecontrolledatareference voltage V ref bythebuilt-inhysteresiscontroller.Thisreferencevoltagewillbe providedbytheproposedMPPTalgorithmandthemicrocontrollertocontrolthe operatingpointofMFC. SincetheproposedMPPTalgorithmopensthecircuitinordertomeasurethe internalshuntcapacitorvoltage,ahighsideloadswitchTPS22860,Texas 80

PAGE 94

Instruments,TX.isusedattheboostconverter'sinput,whichiscontrolledbya microcontroller'sGPIO.Theswitchisonlyopenedatthetimeof V C measurement otherwiseitisclosed.Thisloadswitchhasaverylowon-resistanceandpowerloss, henceitdoesnotsignicantlyaecttheoveralleciency. BQ25504requiresanoutputvoltageofatleast1.8Vtooperate,therefore,two rechargeablebatteriesUltraLastNi-Mh,1.25Vconnectedinseriesareusedatthe outputoftheboostconvertertomaintaintheoutputvoltageatanominalvoltage of2.5V. 4.4.2MicrocontrollerandControlPowerConsumption ThedesignedcontrolsystemusesMSP430L092tomonitorandcontrolthe operatingpointoftheMFC,aswellascontrollingthepowercircuitintheEHS. MSP430L092isselectedbecauseofitsverylowpowerconsumptionandlow operatingvoltage.Ithasabuilt-inanalogpoolthatcanconguredtoimplement ADCtomeasureMFCvoltageandtimer/captureforpulsewidthmodulation PWMandsamplingtimecontrol. TheanodeoftheMFCisconnectedtothemicrocontroller'sADCtomeasure V t and V C ,andalowpasslterisconnectedtoatimerchanneltogenerate V ref forthe boostconverter.Thebottomoneofthetwobatteriesattheboostconverteroutput isusedtofeedthemicrocontroller. 4.4.2.1 OperatingMode Themainpowerconsumerinthecontrolcircuitisthemicrocontrollerandits powerconsumptioncanbereducedbystayinginthe sleepmode andchangingto the activemode onlywhenneeded[72].Accordingtoourlabexperiments, MSP430L092consumesupto78 A inthe activemode whenoperatingat1.65V and1MHzcomparedtoonly8 A inthe sleepmode 4.4.2.2 SamplingTime 81

PAGE 95

Thetimebetweentwopowermeasurementsisimportantbecauseitaectsthe powerconsumptionofthemicrocontroller[72].Thisisbecausethemicrocontroller needstobeinthe activemode atthetimeofmeasurement.Highersamplingtime meanslowerenergyconsumption,butfrequentsamplinggivesfasterconvergence. Therefore,acompromisebetweenenergyconsumptionandoperationperformance needstobemade.However,thesamplingtimeneedstosatisfytherequirementin .30.Inthispaper,asamplingtime T s of30swasused. 4.4.2.3 SystemClockFrequency TheenergyconsumptionofMSP430L092dependsonitsclockfrequency,where itconsumesupto101 A whenoperatingat1.65Vand1MHzcomparedtoonly 42 A at125kHz[75].OperatingwithsuchhighfrequenciesMHzand125kHZ requirethehighfrequencyoscillatorHFO,whichconsumesupto22 A [75].In MSP430L092's sleepmode ,theHFOisturnedoandthelowfrequencyoscillator typically20kHzthatonlyconsumes0.6 A isusedtoreducepowerconsumption. Althoughthehighfrequencyclockisneededforfastmicrocontrollerperformancein theactivemode,thelowfrequencyoscillatorLFOisusedinthispaperinboth modesofoperationtominimizepowerconsumption. 4.4.2.4 SupplyVoltage Thesupplyvoltage V CC isanotherfactorofthemicrocontrollerenergy consumption.MSP430L092operatesatarangeofsupplyvoltages.9-2.2Vand usinghighersupplyvoltageincreasesitspowerintakeandthatreducestheoverall eciencyoftheEHS.Theadvantageofusinghigher V CC isthatHFOandLFOcan generatehigherclockfrequenciestoenhancetheperformance,butlowersupply voltageisrecommendedtoreducethepowerconsumptionofthemicrocontroller. However,themicrocontrollerrequiresaminimumsupplyvoltageofabout1.25Vto bestarted,andoncestarteditcanoperatewithaslowas0.9V.Therefore,the bottombatteryattheboostconverteroutputisusedtosupplythemicrocontroller, 82

PAGE 96

whichhasanominalvoltageof1.25Vandthatishighenoughforthe microcontrollertooperate. 4.4.2.5 Built-InAnalogCircuit Anotherpowerconsumerinthemicrocontrolleristhebuilt-inanalogpoolthat canbeusedtoimplementdigital-to-analogconverterDACandtheADC.Ourlab experimentssuggestthatthemicrocontrollerconsumes35 A at1.3Vwhenthe analogpoolisconguredasADCturnedonallthetime,andanADCmeasurement takesupto40 A .Ontheotherhand,atthesameconditions,themicrocontroller onlyconsumesabout5.5 A whentheanalogpoolwasturnedo.Therefore,using theanalogpoolneedstobeminimizedinordertoreducethemicrocontroller's powerconsumption.Consideringthatthesamplingtimeis30sinthiswork,having theanalogpoolconguredasADCandturnedononlyatthetimeofthe measurementsubstantiallyreducestheenergyconsumption. 4.4.2.6 ReferenceVoltage TheproposedMPPTalgorithmcalculatesthedesiredMFCvoltagebasedonthe powercalculations.Then,themicrocontrollershouldgenerateacommandvoltage V ref fortheboostconverter.GeneratingsuchvoltagewouldbeachievedusingDAC, butcontinuouslygenerating V ref usingDACrequirestheanalogpooltobeturned onallthetime,whichisnotecientbecauseofitshighpowerconsumption. Therefore,aPWMreferencegeneratorisused.ThetimergeneratesaPWMsignal basedonitsinputclocksignal,andsincethetimercangettheLFOclocksignal duringboth active and sleepmode ,itcangeneratethePWMsignalallthetime. ThePWMgenerationinourlabexperimentsdidnotshowasignicantimpacton themicrocontroller'spowerconsumption.ThevalueoftheDCcomponentinthe PWMsignal b V ref dependsonthedutycycle D ofthePWMsignalandthe V CC of themicrocontroller. b V ref = DV CC .33 83

PAGE 97

Figure4.7:Experimentsetuppicture:theMFCreactorandtheusedcircuit. 4.5ExperimentalValidation Theexperimentsetupisshowningure4.7,whereitshowstheproposedEHS andusedMFCreactor.TheMFCvoltageandcurrentweremonitoredusingtwo digitalmultimetersA,B&KPrecisionduringtheexperiments.Havinga samplingtimeof30smakesiteasytomanuallyrecordthemeasurementsforeach operatingpoint.OnemultimeterwasdirectlyconnectedtoMFCterminalsto measuretheterminalvoltageforeachoperatingpoint.Theothermultimeterwas usedtomeasurethevoltageacrossashuntresistortocalculatethecurrent.Since theshuntresistorisonlyneededforthemonitoringandnotrequiredforthesystem operation,thepowerlossacrosstheshuntresistorisnotconsideredintheeciency calculation. 4.5.1MFCReactor Asingle-chamberair-cathodeMFCreactorwasusedintheexperimentsofthis paper.The60mLanodechamberwasbuiltusingacustom-built cm 5cm 3.5cm,4.5cminternalcylinderdiameterandcommercialo-the-shelf plasticblockscm 5cm 2cm,3cminternalcylinderdiameter,Physichemi,Hong 84

PAGE 98

Kong.Acarbonbrushona4-cmtitaniumrodMillrose,OHwasusedasanode electrode.Thecathodeelectrodeismadeofacircularcarbonpapermm diameterthatcontains0.5mg/cm 2 ofPt%ofPt/Ccatalyst,30%wet proong.Thecathodeelectrodecurrentcollectorismadeofasmallpieceof1-mm titaniumwire45485-BY,AlfaAesar,MAthattouchesthecarbonpaper.To connectallcomponents,four10-24stainlesssteelboltsandnutswereusedtohold theblocksandtheelectrodestogether.Thereactorsolutionleakagewasavoidedby installingrubbergasketsbetweenblocks. Thereactorwasinoculatedusingasludgethatwascollectedfromalocalwaste watertreatmentplant,withasodiumacetate2g/LCH 3 COONaandsome vitaminsina50mMphosphatebuersolutionPBS,CH3COONa,0.31g/LNH 4 Cl, 0.13g/LKCl,3.321g/LNaH 2 PO 4 2H 2 O,and10.317g/LNa 2 HPO 4 12H 2 O.The inoculationwasrepeateduntilthereactorterminalvoltagereached350mVwith1 kload.Then,thereactorwasmaintainedwitha2g/LsodiumacetateinaPBS. 4.5.2MPPT TheresultoftheMPPTalgorithmisshowningure4.8.Thestraightlines representthecurrent-voltageI-Vrelationshipi.e.,polarizationcurve.Those curveswereplottedbasedonthemeasurementsofthevoltageandcurrentduring theoperationoftheMPPTalgorithm,andthecurrent-powercurveswerecalculated andplottedbasedI-Vcurves.ThecirclesrepresenttheoperatingpointsoftheMFC commandedbytheMPPTalgorithmandthelinesbetweenthecirclesshowthe directionoftransitions.Notethatthealgorithmstays30sineachoperatingpoint duetothesamplingtime.Itisclearfromgure4.8thattheMPPTalgorithmwas abletoreachandmaintaintheoperationaroundMPPstartingfromaninitial voltageof379mV.Itisalsoclearthatthealgorithmwasablekeepoperatingatthe newMPPwhenthatpointMPPhaschangedaftersometimeduetothechanges ofthereactorconditions.Thisvalidatestheperformanceoftheproposedrealtime 85

PAGE 99

Figure4.8:MPPTresults:circlesrepresenttheoperatingpointsofthealgorithm. MPPTalgorithm. 4.5.3Eciency Theeciencyofthesystemwascalculatedbymeasuringtheboostconverter inputandoutputpowersaswellasthemicrocontroller'spowerconsumption.The boostconverter'sinputistheMFCoutputpowerwhichwascalculatedby multiplyingtheMFCvoltageandcurrent,andtheoutputpowerwascalculatedby measuringbatteryvoltageandconverteroutputcurrent.Themicrocontroller's powerwascalculatedbymeasuringthecurrentthroughashuntresistorand measuringitssupplyvoltage V CC .Itshouldbenotedthatthepoweracrossthe shuntresistorswasremovedfromtheoveralleciencycalculationsincethose resistorareonlyneededforeciencycalculationandhavenoeectonthesystem performance.Fortheinputshuntresistor,thelossneedstobesubtractedfromthe inputpower,andfortheothertworesistors,theirpowerlossneedstobeaddedto theoutputpower. 86

PAGE 100

Theboostconvertereciencywas66.7%withaninputpowerof119 Wandan outputpowerof79.4 Woperatingataninputvoltageof213mVandoutput voltageof2.62V,andthemicrocontrollertakesabout8.67 Woperatingat1.3V. TheoveralleciencyoftheEHSwasabout59.4%.However,theoveralleciencyis mainlylimitedbytheboostconverter'seciency;hence,usingmoreecient convertershouldincreasetheoveralleciencyconsideringthatthemicrocontroller powerconsumptionisxedwhenusing1.3Vpowersupply. 4.6Conclusion Inthispaper,anecientself-poweredenergyharvestingsystemEHSwas proposed.ThesystemtracksMFCMPPusingP&Oalgorithmbasedonthe proposedcurrentestimationmethod.Thepowermeasurementsamplingtimewas increasedtoallowtheinternalshuntcapacitorvoltagetoreachitssteadystate value,otherwiseitstransientwillhaveaneectonpowermeasurement.The proposedMFCcurrentestimationmethodwasabletoestimateMFCcurrent withoutactualcurrentmeasurements.Usingthatcurrentestimationmethodand themeasurementsamplingtime T s ,theproposedMPPTalgorithmwasableto estimatethesteadystateMFCpoweranditsuccessfullytrackedandoperatedat MFCMPPevenafteritchangedduetochangingoperatingconditions.Inaddition, theproposedEHSwasabletoharvestMFCpoweratupto59.4%eciencyusing 119 WMFCwithoutanyexternalpowersupply.Acommercialo-the-shelf microcontrollerwasconguredtoconsumeonly8.67 Wfortheimplementationof theproposedMPPTalgorithm. 87

PAGE 101

CHAPTERV SUMMARYANDFUTUREWORK 5.1Summary Thisdissertationisfocusedonimprovingelectricalenergyharvestingfrom microbialfuelcellsMFCs.Thepresentedworkshowsanimprovementinthe reliabilityandeciencyofMFCenergyharvestingcircuits.Italsostudiestheeect ofusingpowerelectronicsconvertersonthereactorperformanceandconditions. Inchapter1,ageneraldescriptionaboutlowpowerrenewableenergysources suchaspiezoelectricandthermoelectricgenerationispresented.Thehistoryof MFCsandhowtheybecameinterestingforelectricalgenerationisalsopresentedin theintroductionaswellasadetaileddescriptionofMFCsandtheiroperation.In addition,dierentkindsofenergyharvestingcircuitsandtechniques,thathavebeen usedwithMFCs,ispresentedwithadetailsabouttheiradvantagesand disadvantages. TheextremumseekingESmaximumpowerpointtrackingMPPTalgorithm wasusedtotrackthemaximumpowerpointofMFCinchapter2.TheES algorithmusessinusoidalperturbationtondthedirectionoftheMPP.The simulationandexperimentalresultsshowedthattheESMPPTalgorithmwasable tondandoperateattheMPP.However,duetothevoltageovershootphenomena thatsometimeshappentoMFC,MPPTalgorithmsincludingESMPPTcannot keepharvestingenergyfromMFCbecausethevoltageovershootmisleadsthem, whichresultsinarunawayconditionandeventuallynopowerisharvestedbecause ofthat.Toavoidthat,avoltageovershootavoidancealgorithmisproposed.That proposedalgorithmdetectsthevoltageovershootwhenithappensandoperatesat anoperatingpointjustbeforetheoperatingpointwherethevoltageovershoot happens.Theexperimentalresultsshowedthattheproposedalgorithmwasableto detectthevoltageovershootandwasabletokeepharvestingenergywiththe existenceofvoltageovershoot. 88

PAGE 102

PowerelectronicsconvertershavebeenusedtoharvestenergyfromMFC. However,theyimposecurrentrippleonthesourceMFCbecauseoftheirswitching behavior.Suchcurrentripplewasshowntohaveanimpactonotherfuelcellssuch asprotonmembraneexchangefuelcellsPEMFCandsolidoxidefuelcellsSOFC. Inchapter3,theeectofthecurrentrippleonMFCreactorhavebeeninvestigated. TheperformanceoftheMFCreactorwasmonitoredduringenergyharvestingwith twotypesofcurrentshapessquare-shapeandtriangular-shapeforacomplete batch 2daysandtheresultswerecomparedwiththecontinuouscurrentenergy harvesting.Thesquare-shapeandtriangular-shapecurrentswereusedbecausethey representthepossiblecurrentripplesthatcouldbeimposedbypowerelectronics converters.MonitoringtheMFCreactorperformancewasachievedbymeasuring andrecordingtheaveragevoltage,whichcanbeusedtocalculatetheMFCpower andtheharvestedenergy.Inaddition,otherphysical-chemicalparametersincluding pH,EC,ORP,andDOoftheanodechamberweremeasuredusingrealtimesensor probesandrecordedinordertochecktheeectofcurrentrippleonsuch parameters.Theexperimentalresultsshowedthatpowerelectronicsconverters currentripplehavenosignicanteectonMFCreactorperformance. Inchapter4,asimplecurrentestimationmethodwasproposed.Thismethodis basedonthecharacteristicsoftheMFC,wheretheMFCequivalentcircuitmodel wasusedtoestimatethecurrentfromthemeasurementoftheterminalvoltageand theinternalcapacitorvoltagewhilethecurrentisowing.Fromthosetwo measurements,thecurrentiscalculatedusingohmslawassumingtheresistance R 2 isknown.However,theproposedMPPTalgorithm,thatisbasedonP&O algorithm,ndstheMPPbythecomparisonofthepowerbetweentwooperating pointsandsincetheactualvalueofthepowerisnotrequiredforthiscomparison, thevalueoftheresistance R 2 isremovedfromthecurrentestimationmethodwhich resultsincalculatingtheproportionalpower ~ P est thatisdierentthantheestimated 89

PAGE 103

power P est .Using ~ P est intheMPPTalgorithmshouldhavenoeectsincethe relationshipbetweenanytwopowerswouldremainthesameoneishigherorlower thantheother.Inaddition,asamplingtime T s ,thatisbasedontheparametersof theMFCequivalentcircuit,isaddedbeforemeasuringthepowerwhenmovingtoa newoperatingpointtoguaranteethemeasurementofthesteadystatepower.The proposedcurrentestimationandMPPTalgorithmwasimplementedina microcontrollerthatispoweredfromtheMFCpowerinordertobuildaself poweredpowermanagementsystemPMS.Theexperimentalresultsshowsthat theproposedMPPTalgorithmwasabletoreachandoperateatMPPusingthe proposedcurrentestimationmethodwithamicrocontrollerpowerconsumptionof 8.67 W.TheoveralleciencyofharvestingMFCpowerusingtheselfpowered PMSwasupto59.4%includingthepowerthatisrequiredtorunthecontrolcircuit. 5.2FutureWork FutureworkneedstofocusedonimprovingMFCenergyharvestingcircuits. ThisincludesdierenttopologiesthatttheneedsofMFCanddierentalgorithms thatcanimprovetheoperationoftheenergyharvestingcircuit.Inaddition,the voltageovershootphenomenaneedstobemoreinvestigatedfromthepointofview ofelectricalengineeringbecauseitcouldbeexplainedbytheMFCdynamic behavior,thatispresentedintheequivalentcircuitmodel[26]. 90

PAGE 104

REFERENCES [1]A.Bindra,Energyharvesting:Asustainableambientsourcetoultralowpower devices," IEEEPowerElectronicsMagazine ,vol.4,pp.4{8,2017. [2]E.Iranmanesh,A.Rasheed,W.Li,andK.Wang,AWearablePiezoelectric EnergyHarvesterRectiedbyaDual-GateThin-FilmTransistor," IEEE TransactionsonElectronDevices ,vol.65,no.2,pp.542{546,2018. [3]B.Haug,Wirelesssensornodescanbepoweredbytemperaturegradients;no batteriesneeded:Harvestingenergyfromthermoelectricgenerators," IEEE PowerElectronicsMagazine ,vol.4,pp.24{32,2017. [4]T.Kondo,N.Chiwaki,andS.Sugahara,Designandperformanceofthin-lm tEGmodulesforwearabledeviceapplications," 2017IEEEElectronDevices TechnologyandManufacturingConference,EDTM2017-Proceedings pp.201{203,2017. [5]C.WeiandX.Jing,Acomprehensivereviewonvibrationenergyharvesting: Modellingandrealization," RenewableandSustainableEnergyReviews ,vol.74, pp.1{18,2017. [6]K.VijayaraghavanandR.Rajamani,Novelbatterylesswirelesssensorfor trac-owmeasurement," IEEETransactionsonVehicularTechnology vol.59,no.7,pp.3249{3260,2010. [7]A.Shantaram,H.Beyenal,R.Raajan,A.Veluchamy,andZ.Lewandowski, Wirelesssensorspoweredbymicrobialfuelcells.," EnvironmentalScienceand Technology ,vol.39,no.13,pp.5037{5042,2005. [8]F.Zhang,L.Tian,andZ.He,Poweringawirelesstemperaturesensorusing sedimentmicrobialfuelcellswithverticalarrangementofelectrodes," Journal ofPowerSources ,vol.196,no.22,pp.9568{9573,2011. [9]C.Donovan,A.Dewan,D.Heo,andH.Beyenal,Batteryless,wirelesssensor poweredbyasedimentmicrobialfuelcell," EnvironmentalScienceand Technology ,vol.42,no.22,pp.8591{8596,2008. [10]C.Donovan,A.Dewan,H.Peng,D.Heo,andH.Beyenal,Powermanagement systemfora2.5Wremotesensorpoweredbyasedimentmicrobialfuelcell," JournalofPowerSources ,vol.196,no.3,pp.1171{1177,2011. [11]Y.Gong,S.E.Radachowsky,M.Wolf,M.E.Nielsen,P.R.Girguis,andC.E. Reimers,Benthicmicrobialfuelcellasdirectpowersourceforanacoustic modemandseawateroxygen/temperaturesensorsystem," Environmental ScienceandTechnology ,vol.45,no.11,pp.5047{5053,2011. 91

PAGE 105

[12]Q.Zheng,L.Xiong,B.Mo,W.Lu,S.Kim,andZ.Wang,Temperatureand HumiditySensorPoweredbyanIndividualMicrobialFuelCellinaPower ManagementSystem," Sensors ,vol.15,no.9,pp.23126{23144,2015. [13]H.Wang,J.D.Park,andZ.Ren,Activeenergyharvestingfrommicrobial fuelcellsatthemaximumpowerpointwithoutusingresistors," Environmental ScienceandTechnology ,vol.46,no.9,pp.5247{5252,2012. [14]M.C.Potter,Electricaleectsaccompanyingthedecompositionoforganic compounds," ProceedingsoftheRoyalSocietyofLondonB:BiologicalSciences vol.84,no.571,pp.260{276,1911. [15]K.Lewis,SymposiumonBioelectrochemistryofMicroorganisms,IV. BiochemicalFuelCells," BacteriologicalReview ,vol.30,no.1,pp.101{113, 1966. [16]R.M.AllenandH.P.Bennetto,Microbialfuelcells:Electricityproduction fromcarbohydrates," AppliedBiochemistryandBiotechnology ,vol.39, pp.27{40,1993. [17]L.Bruce, MicrobialFuelCells .Hoboken,NewJersey,USA:JohnWiley& Sons,Inc.,2008. [18]B.E.Logan,B.Hamelers,R.Rozendal,U.Schroder,J.Keller,S.Freguia, P.Aelterman,W.Verstraete,andK.Rabaey,Microbialfuelcells: Methodologyandtechnology," EnvironmentalScienceandTechnology ,vol.40, no.17,pp.5181{5192,2006. [19]Z.Du,H.Li,andT.Gu,Astateoftheartreviewonmicrobialfuelcells:A promisingtechnologyforwastewatertreatmentandbioenergy," Biotechnology Advances ,vol.25,no.5,pp.464{482,2007. [20]H.LiuandB.E.,Electricitygenerationusinganair-cathodesinglechamber microbialfuelcellinthepresenceandabsenceofaprotonexchange membrane.," Environmentalscience&technology ,vol.38,no.14, pp.4040{4046,2004. [21]N.S.RamarajaPRamasamy,ElectrochemicalImpedanceSpectroscopyfor MicrobialFuelCellCharacterization," JournalofMicrobial&Biochemical Technology ,2013. [22]C.Ogugbue,E.Ebode,andS.Leera,ElectricityGenerationFromSwine WastewaterUsingMicrobialFuelCell," JournalofEcologicalEngineering vol.16,no.5,pp.26{33,2015. [23]Z.HeandF.Mansfeld,Exploringtheuseofelectrochemicalimpedance spectroscopyEISinmicrobialfuelcellstudies," Energy&Environmental Science ,vol.2,no.2,p.215,2009. 92

PAGE 106

[24]A.K.Manohar,O.Bretschger,K.H.Nealson,andF.Mansfeld,Theuseof electrochemicalimpedancespectroscopyEISintheevaluationofthe electrochemicalpropertiesofamicrobialfuelcell," Bioelectrochemistry ,vol.72, no.2,pp.149{154,2008. [25]J.Coronado,B.Tartakovsky,andM.Perrier,On-linemonitoringofmicrobial fuelcellsoperatedwithpulse-widthmodulatedelectricalload," Journalof ProcessControl ,vol.35,pp.59{64,2015. [26]J.-D.Park,T.Roane,Z.Ren,andM.Alaraj,Dynamicmodelingofa microbialfuelcellconsideringanodicelectronowandelectricalcharge storage," AppliedEnergy ,vol.193,2017. [27]J.-d.ParkandZ.Ren,EcientEnergyHarvesterforMicrobialFuelCells usingDC/DCConverters," IEEEEnergyConversionCongressand Exposition ,pp.3852{3858,2011. [28]R.P.Pinto,B.Srinivasan,S.R.Uiot,andB.Tartakovsky,Theeectof real-timeexternalresistanceoptimizationonmicrobialfuelcellperformance," WaterResearch ,vol.45,no.4,pp.1571{1578,2011. [29]a.Rhoads,H.Beyenal,andZ.Lewandowski,Microbialfuelcellusing anaerobicrespirationasananodicreactionandbiomineraliedmanganeseasa cathodicreactant," EnvironmentalScience&Technology ,vol.39,no.12, pp.4666{4671,2005. [30]H.Liu,S.Cheng,andB.E.Logan,Productionofelectricityfromacetateor butyrateusingasingle-chambermicrobialfuelcell," EnvironmentalScience andTechnology ,vol.39,no.2,pp.658{662,2005. [31]A.Meehan,H.Gao,andZ.Lewandowski,Energyharvestingwithmicrobial fuelcellandpowermanagementsystem," IEEETransactionsonPower Electronics ,vol.26,no.1,pp.176{181,2011. [32]P.Aelterman,K.Rabaey,H.T.Pham,N.Boon,andW.Verstraete, Continuouselectricitygenerationathighvoltagesandcurrentsusingstacked microbialfuelcells," EnvironmentalScienceandTechnology ,vol.40,no.10, pp.3388{3394,2006. [33]A.Dewan,H.Beyenal,andZ.Lewandowski,Scalingupmicrobialfuelcells," EnvironmentalScienceandTechnology ,vol.42,no.20,pp.7643{7648,2008. [34]E.Carlson,K.Strunz,andB.Otis,mVinputboostconverterfor thermoelectricenergyharvesting," 2009SymposiumonVLSICircuits ,vol.2, pp.162{163,2009. [35]E.J.Carlson,K.Strunz,andB.P.Otis,A20mvinputboostconverterwith ecientdigitalcontrolforthermoelectricenergyharvesting," IEEEJournalof Solid-StateCircuits ,vol.45,pp.741{750,April2010. 93

PAGE 107

[36]J.D.ParkandZ.Ren,Ecientenergyharvesterformicrobialfuelcellsusing dc/dcconverters,"in 2011IEEEEnergyConversionCongressandExposition pp.3852{3858,Sept2011. [37]H.C.Boghani,J.R.Kim,R.M.Dinsdale,A.J.Guwy,andG.C.Premier, Controlofpowersourcedfromamicrobialfuelcellreducesitsstart-uptime andincreasesbioelectrochemicalactivity," BioresourceTechnology ,vol.140, pp.277{285,2013. [38]D.Molognoni,S.Puig,M.D.Balaguer,A.Liberale,A.G.Capodaglio, A.Callegari,andJ.Colprim,Reducingstart-uptimeandminimizingenergy lossesofMicrobialFuelCellsusingMaximumPowerPointTrackingstrategy," JournalofPowerSources ,vol.269,pp.403{411,2014. [39]D.Molognoni,S.Puig,M.D.Balaguer,A.G.Capodaglio,A.Callegari,and J.Colprim,Multiparametriccontrolforenhancedbiolmselectionin microbialfuelcells," JournalofChemicalTechnologyandBiotechnology vol.91,no.6,pp.1720{1727,2016. [40]H.C.Boghani,G.Papaharalabos,I.Michie,K.R.Fradler,R.M.Dinsdale, A.J.Guwy,I.Ieropoulos,J.Greenman,andG.C.Premier,Controllingfor peakpowerextractionfrommicrobialfuelcellscanincreasestackvoltageand avoidcellreversal," JournalofPowerSources ,vol.269,pp.363{369,2014. [41]P.C.Nien,C.Y.Lee,K.C.Ho,S.S.Adav,L.Liu,A.Wang,N.Ren,and D.J.Lee,Powerovershootintwo-chamberedmicrobialfuelcellMFC," BioresourceTechnology ,vol.102,no.7,pp.4742{4746,2011. [42]J.Wineld,I.Ieropoulos,J.Greenman,andJ.Dennis,Theovershoot phenomenonasafunctionofinternalresistanceinmicrobialfuelcells," Bioelectrochemistry ,vol.81,no.1,pp.22{27,2011. [43]V.J.WatsonandB.E.Logan,Analysisofpolarizationmethodsfor eliminationofpowerovershootinmicrobialfuelcells," Electrochemistry Communications ,vol.13,no.1,pp.54{56,2011. [44]M.Alaraj,Z.J.Ren,andJ.-D.Park,Microbialfuelcellenergyharvesting usingsynchronousybackconverter," JournalofPowerSources ,vol.247, pp.636{642,2014. [45]N.Degrenne,F.Buret,B.Allard,andP.Bevilacqua,Electricalenergy generationfromalargenumberofmicrobialfuelcellsoperatingatmaximum powerpointelectricalload," JournalofPowerSources ,vol.205,pp.188{193, 2012. [46]J.D.ParkandZ.Ren,Hysteresis-controller-basedenergyharvestingscheme formicrobialfuelcellswithparalleloperationcapability," IEEETransactions onEnergyConversion ,vol.27,no.3,pp.715{724,2012. 94

PAGE 108

[47]X.Zhu,J.C.Tokash,Y.Hong,andB.E.Logan,Controllingtheoccurrence ofpowerovershootbyadaptingmicrobialfuelcellstohighanodepotentials," Bioelectrochemistry ,vol.90,pp.30{35,2013. [48]Y.Hong,D.F.Call,C.M.Werner,andB.E.Logan,Adaptationtohigh currentusinglowexternalresistanceseliminatespowerovershootinmicrobial fuelcells," BiosensorsandBioelectronics ,vol.28,no.1,pp.71{76,2011. [49]Y.Tan,W.H.Moase,C.Manzie,D.Nesic,andI.M.Y.Mareels,Extremum seekingfrom1922to2010,"in Proceedingsofthe29thChineseControl Conference ,pp.14{26,jul2010. [50]M.Krstic,Performanceimprovementandlimitationsinextremumseeking control," Systems&ControlLetters ,vol.39,no.5,pp.313{326,2000. [51]K.B.AriyuK.B.Ariyur,M.KrsticandM.Krstic, RealTimeOptimizationby ExtremumSeekingControl .NewYork,NY,USA:JohnWiley&Sons,Inc., 2003. [52]W.Choi,G.Joung,P.N.Enjeti,andJ.W.Howze,AnExperimental EvaluationoftheEectsofRippleCurrentGeneratedbythePower ConditioningStageonaProtonExchangeMembraneFuelCellStack," Journal ofMaterialsEngineeringandPerformance ,vol.13,no.June,pp.257{264,2004. [53]W.Choi,J.W.Howze,andP.Enjeti,Developmentofanequivalentcircuit modelofafuelcelltoevaluatetheeectsofinverterripplecurrent," Journalof PowerSources ,vol.158,no.2SPEC.ISS.,pp.1324{1332,2006. [54]S.K.Mazumder,K.Acharya,C.L.Haynes,R.Williams,M.R.von Spakovsky,D.J.Nelson,D.F.Rancruel,J.Hartvigsen,andR.S.Gemmen, Solid-oxide-fuel-cellperformanceanddurability:Resolutionoftheeectsof power-conditioningsystemsandapplicationloads," IEEETransactionson PowerElectronics ,vol.19,no.5,pp.1263{1278,2004. [55]S.K.Butti,G.Velvizhi,M.L.K.Sulonen,J.M.Haavisto,E.OguzKoroglu, A.YusufCetinkaya,S.Singh,D.Arya,J.AnnieModestra,K.VamsiKrishna, A.Verma,B.Ozkaya,A.M.Lakaniemi,J.A.Puhakka,andS.VenkataMohan, Microbialelectrochemicaltechnologieswiththeperspectiveofharnessing bioenergy:Maneuveringtowardsupscaling," RenewableandSustainableEnergy Reviews ,vol.53,pp.462{476,2016. [56]S.VenkataMohan,G.Velvizhi,J.AnnieModestra,andS.Srikanth,Microbial fuelcell:Criticalfactorsregulatingbio-catalyzedelectrochemicalprocessand recentadvancements," RenewableandSustainableEnergyReviews ,vol.40, pp.779{797,2014. [57]Y.AhnandB.E.Logan,Eectivenessofdomesticwastewatertreatment usingmicrobialfuelcellsatambientandmesophilictemperatures," Bioresource Technology ,vol.101,no.2,pp.469{475,2010. 95

PAGE 109

[58]T.Shimoyama,S.Komukai,A.Yamazawa,Y.Ueno,B.E.Logan,and K.Watanabe,Electricitygenerationfrommodelorganicwastewaterina cassette-electrodemicrobialfuelcell," AppliedMicrobiologyandBiotechnology vol.80,no.2,pp.325{330,2008. [59]B.Min,S.Cheng,andB.E.Logan,Electricitygenerationusingmembrane andsaltbridgemicrobialfuelcells," WaterResearch ,vol.39,pp.1675{1686, 2005. [60]Z.He,Y.Huang,A.K.Manohar,andF.Mansfeld,EectofelectrolytepHon therateoftheanodicandcathodicreactionsinanair-cathodemicrobialfuel cell," Bioelectrochemistry ,vol.74,no.1,pp.78{82,2008. [61]R.a.Rozendal,H.V.M.Hamelers,andC.J.N.Buisman,Eectsof MembraneCationTransportonpHandMicrobialFuel," Environmental science&technology ,vol.40,no.17,pp.5206{5211,2006. [62]G.-C.Gil,I.-S.Chang,B.H.Kim,M.Kim,J.-K.Jang,H.S.Park,andH.J. Kim,Operationalparametersaectingtheperformannceofamediator-less microbialfuelcell," BiosensorsandBioelectronics ,vol.18,no.4,pp.327{334, 2003. [63]J.Y.Nam,H.W.Kim,K.H.Lim,H.S.Shin,andB.E.Logan,Variationof powergenerationatdierentbuertypesandconductivitiesinsinglechamber microbialfuelcells," BiosensorsandBioelectronics ,vol.25,no.5, pp.1155{1159,2010. [64]L.Woodward,M.Perrier,B.Srinivasan,R.P.Pinto,andTartakovsky, ComparisonofReal-TimeMethodsforMaximizingPowerOutputin MicrobialFuelCells," AIChEJournal ,vol.56,no.10,pp.2742{2750,2010. [65]J.-D.ParkandZ.Ren,Hysteresiscontrollerbasedmaximumpowerpoint trackingenergyharvestingsystemformicrobialfuelcells," JournalofPower Sources ,vol.205,pp.151{156,2012. [66]M.Alaraj,M.Radenkovic,andJ.D.Park,Intelligentenergyharvesting schemeformicrobialfuelcells:Maximumpowerpointtrackingandvoltage overshootavoidance," JournalofPowerSources ,vol.342,pp.726{732,2017. [67]U.Karra,E.Muto,R.Umaz,M.Kolln,C.Santoro,L.Wang,andB.Li, Performanceevaluationofactivatedcarbon-basedelectrodeswithnovelpower managementsystemforlong-termbenthicmicrobialfuelcells," International JournalofHydrogenEnergy ,vol.39,no.36,pp.21847{21856,2014. [68]N.Tang,W.Hong,T.Ewing,H.Beyenal,andJ.-h.Kim,ASelf-Sustainable PowerManagementSystemforReliablePowerScalingUpofSediment MicrobialFuelCells," IEEETransactionsonPowerElectronics ,vol.30,no.9, pp.4626{4632,2015. 96

PAGE 110

[69]E.Dallago,A.L.Barnabei,A.Liberale,G.Torelli,andG.Venchi,A300-mV low-powermanagementsystemforenergyharvestingapplications," IEEE TransactionsonPowerElectronics ,vol.31,no.3,pp.2273{2281,2016. [70]S.Carreon-Bautista,C.Erbay,A.Han,andE.Sanchez-Sinencio,Power ManagementSystemWithIntegratedMaximumPowerExtractionAlgorithm forMicrobialFuelCells," IEEETransactionsonEnergyConversion ,vol.15, no.1,pp.262{,2015. [71]S.Carreon-bautista,S.Member,C.Erbay,andS.Member,AnInductorless DC-DCConverterforanEnergyAwarePowerManagementUnitAimedat MicrobialFuelCellArrays," IEEEJournalofEmergingandSelectedTopicsin PowerElectronics ,vol.3,no.4,pp.1109{1121,2015. [72]H.C.Boghani,R.M.Dinsdale,A.J.Guwy,andG.C.Premier,Sampled-time controlofamicrobialfuelcellstack," JournalofPowerSources ,2017. [73]F.Yang,K.-C.Wang,andY.Huang,Energy-NeutralCommunication ProtocolforVeryLowPowerMicrobialFuelCellBasedWirelessSensor Network," IEEESensorsJournal ,vol.15,no.4,pp.2306{2315,2015. [74]I.Lee,G.Kim,S.Bang,A.Wolfe,R.Bell,S.Jeong,Y.Kim,J.Kagan, M.Arias-Thode,B.Chadwick,D.Sylvester,D.Blaauw,andY.Lee, System-On-Mud:Ultra-LowPowerOceanicSensingPlatformPoweredby Small-ScaleBenthicMicrobialFuelCells," IEEETransactionsonCircuitsand SystemsI:RegularPapers ,vol.62,no.4,pp.1126{1135,2015. [75]TexasInstruments,TX, MSP430L092datasheet ,2010. 97