Professional Documents
Culture Documents
BY
IOANNIS D. SPYROU
AND SUPERVISED BY
UNIVERSITY OF SOUTHAMPTON
ABSTRACT
FACULTY OF ENGINEERING, SCIENCE AND MATHEMATICS
SCHOOL OF ENGINEERING SCIENCES
Master of Science
SOLAR HYDROGEN PRODUCTION USING WATER ELECTROLYSIS
DESIGN STUDY OF A STAND-ALONE HYDROGEN PRODUCTION SYSTEM POWERED BY
PHOTOVOLTAICS AND ASSISTED BY BATTERIES WITH THE AIM OF ELECTROLYSER
SUSTAINABLE OPERATION
By Ioannis D. Spyrou
CONTENTS
CONTENTS.........................................................................................................................1
LISTOFFIGURES................................................................................................................4
LISTOFTABLES..................................................................................................................6
LISTOFABBREVIATIONS....................................................................................................7
ACKNOWLEDGEMENTS...................................................................................................10
1.
INTRODUCTION.......................................................................................................11
2.
PROJECTFEATURES.................................................................................................13
2.1. GENERALDESCRIPTIONOFTHESYSTEM..........................................................13
2.2. AIMS&OBJECTIVESOFTHEPROJECT..............................................................14
2.3. THENOVELTYOFTHEPROJECT........................................................................15
3.
LITERATUREREVIEW...............................................................................................16
3.1. HYDROGENFUELSTATIONSSIMULATIONSTUDIES&ANAPPLICATION......16
3.2. DIRECTCOUPLINGOFPVANDELECTROLYSERMODELING&EXPERIMENTAL
RESULTS.......................................................................................................................20
3.3. THELIMITATIONSOFCOUPLINGPVANDELECTROLYSER...............................24
3.4. WAYSTOOVERCOMECOUPLINGLIMITATIONS..............................................26
3.5. SUMMARY........................................................................................................28
4.
MODELDESCRIPTION..............................................................................................30
4.1. INTRODUCTION................................................................................................30
4.2. DATAINPUTSUBSYSTEMSMODELINGDESCRIPTION...................................32
4.2.1.
TECHNICAL................................................................................................32
4.2.1.1.
PHOTOVOLTAICARRAY......................................................................33
4.2.1.2.
HYDROGENFUELDEMAND................................................................35
4.2.1.3.
BATTERIES..........................................................................................36
4.2.1.4.
ELECTROLYSER...................................................................................41
4.2.1.5.
HYDROGENSTORAGETANK..............................................................43
4.2.2.
ECONOMICS..............................................................................................44
1
4.2.2.1.
PHOTOVOLTAICARRAY......................................................................45
4.2.2.2.
BATTERIES..........................................................................................46
4.2.2.3.
ELECTROLYSER...................................................................................46
4.2.2.4.
HYDROGENSTORAGETANK..............................................................48
4.2.2.5.
SYSTEM..............................................................................................49
4.3. POWERMANAGEMENTSTRATEGY..................................................................49
4.3.1.
ELECTROLYSEROPERATIONCONSTRAINTS..............................................52
4.3.2.
ENERGYPRODUCTIONMANAGEMENT.....................................................54
4.4. RESULTSOUTPUT.............................................................................................58
5.
RESULTS...................................................................................................................61
5.1. CASESTUDY......................................................................................................61
5.2. SINGLEOBJECTIVEOPTIMIZATION...................................................................66
5.3. PARAMETRICSTUDIES......................................................................................69
5.3.1.
LOCATION..................................................................................................69
5.3.2.
RELIABILTY.................................................................................................71
5.3.3.
NUMBEROFCARSREFUELED....................................................................74
5.3.4.
HYDROGENAUTONOMY...........................................................................75
5.3.5.
CAPITALCOSTREDUCTION.......................................................................77
5.4. DOUBLEOBJECTIVEOPTIMIZATION.................................................................79
6.
5.4.1.
ELECTROLYSERMINIMUMOPERATIONHOURSENERGYAUTONOMY..80
5.4.2.
ELECTROLYSERCYCLESENERGYAYTONOMY.........................................81
5.4.3.
PVELECTROLYSERCAPACITYFACTORS..................................................82
DISCUSSION.............................................................................................................84
6.1. OPERATION.......................................................................................................84
6.2. SIZING...............................................................................................................85
6.3. PERFORMANCE.................................................................................................88
6.4. HYDROGENPRODUCTIONCOST.......................................................................90
7.
CONCLUSIONS.........................................................................................................93
APPENDIX........................................................................................................................95
A. PVFUNCTION......................................................................................................95
B. BATTERIESFUNCTION.........................................................................................96
C.
ELECTROLYSERFUNCTION..................................................................................98
D. STORAGETANKFUNCTION.................................................................................99
E.
POWERMANAGEMENTSTRATEGYFUNCTION.................................................100
F.
ECONOMICSFUNCTION....................................................................................104
G. FUELDEMANDFUNCTION................................................................................105
H. SYSTEMFUNCTION............................................................................................106
REFERENCES..................................................................................................................114
LISTOFFIGURES
FIGURE1SIMPLIFIEDSOLARHYDROGENSYSTEMSCHEME.....................................................................13
FIGURE2SCHEMATICDIAGRAMOFTHEENERGYANDMATERIALSFLOWINHOMESOLARHYDROGEN
PRODUCTIONSYSTEM[17]........................................................................................................17
FIGURE3THELEVELIZEDSOLARHYDROGENCOSTVERSUSTHEAVERAGEYEARLYTOTALSOLARENERGYONTHE
HORIZONTALSURFACEFORFIXEDANDTRACKINGPANELSSYSTEMS[13]...........................................18
FIGURE4ACORRELATIONFORHYDROGENCOSTFORFIXEDANDSUNTRACKINGPANELS[12].....................19
FIGURE5CHARACTERISTICOFELECTROLYSERANDMAXIMUMPOWERPOINTSOFPVFOROPTIMAL
CONFIGURATIONOFPEMCELLSANDPVPANELS[18]..................................................................20
FIGURE6ELECTROLYZEREFFICIENCYVERSUSELECTROLYZERCELLTEMPERATUREFORARELATIVELYCONSTANT
CURRENT(125130A)MEASUREDOVERTHE14DAYSTUDYPERIOD[5].........................................21
FIGURE7AVERAGESOLARHYDROGENPRODUCTIONEFFICIENCYUSINGPVELECTROLYSERSYSTEMSWITHDC
DCCONVERTERSFOROPTIMIZATIONVERSUSDIRECTCONNECTION[15]...........................................22
FIGURE8MEASUREDSTACKVIANDEFFICIENCYINITIALLY(LEFT)ANDOVER5DATAPERIOD(RIGHT)FORA13
CELLSTACK[6].......................................................................................................................23
FIGURE9ELECTROLYSERCYCLESVSHYSTERESISANDSTATEOFCHARGEOFBATTERY[9]............................26
FIGURE10MODELOVERVIEW........................................................................................................31
FIGURE11FLOWCHARTOFBATTERIESSUBROUTINE...........................................................................37
FIGURE12ELECTROLYSERCOMPONENTCOSTSOBTAINEDFROM10MAJORSUPPLIERSINEUROPE[35].......47
FIGURE13PRICE/CAPACITYRELATIONSFORCOMMERCIALLYAVAILABLECOMPRESSEDSTORAGEOPTIONSFOR
HYDROGENINTHEMEDIUMTOLARGESCALERANGE[35]............................................................48
FIGURE14POWERMANAGEMENTSTRATEGYFLOWCHART..................................................................50
FIGURE15CONSUMPTIONSOURCESFORPVPOWERPRODUCTION(WINTERWEEK)................................62
FIGURE16ELECTROLYSERPOWERINGSOURCES(WINTERWEEK)...........................................................62
FIGURE17BATTERIESANDSTORAGETANKSTATUSVARIATION(WINTERWEEK).......................................63
FIGURE18CONSUMPTIONSOURCESFORPVPOWERPRODUCTION(SUMMERWEEK)...............................63
FIGURE19ELECTROLYSERPOWERINGSOURCES(SUMMERWEEK)..........................................................64
FIGURE20BATTERIESANDSTORAGETANKSTATUSVARIATION(SUMMERWEEK)......................................64
FIGURE21BATTERIESANDSTORAGETANKSTATUSVARIATION(ANNUAL)...............................................65
FIGURE22CONVERGENCEINSINGLEOBJECTIVEOPTIMIZATION............................................................67
FIGURE23CAPITAL(LEFT)/MAINTENANCE(RIGHT)COSTSHAREINTHEOPTIMUMSYSTEM.....................68
FIGURE24PRODUCTIONCOSTVARIATIONREGARDINGTHEANNUALSOLARENERGYDENSITY....................70
FIGURE25ELECTROLYSERCYCLESVSANNUALSOLARENERGYDENSITYINOPTIMALECONOMICSYSTEMS.....71
FIGURE26EFFECTOFSYSTEMRELIABILITYONSPECIFICPRODUCTIONCOSTOFOPTIMUMECONOMICSYSTEM72
FIGURE27RATIOOFPVTOELECTROLYSERINSTALLEDPOWERREQUIREDFORVARIOUSLEVELSOFRELIABILITY
............................................................................................................................................72
FIGURE28PERFORMANCEINDICATORSOFOPTIMUMECONOMICSYSTEMSWITHDIFFERENTRELIABILITY.....73
FIGURE29PV/ELECTROLYSERINSTALLEDPOWERINOPTIMALECONOMICSYSTEMSFORDIFFERENTFUEL
DEMAND(CARSREFUELED).......................................................................................................74
FIGURE30OPTIMALPRODUCTIONCOSTVARIATIONWITHTHENUMBEROFCARSREFUELLEDBYSTATION....75
FIGURE31HOURSOFAUTONOMYREQUIREDBYBATTERIESVSDAYSOFAUTONOMYPROVIDEDBYSTORAGE
TANKINOPTIMALECONOMICSYSTEMS.......................................................................................76
FIGURE32OPTIMALPRODUCTIONCOSTFORDIFFERENTDAYSOFAUTONOMY(STORAGETANK)REQUIRED..77
FIGURE33OPTIMALHYDROGENPRODUCTIONCOSTFORDIFFERENTFUELDEMAND(CARSREFUELLED)UNDER
THREESCENARIOSOFPVANDELECTROLYSERCAPITALCOSTREDUCTION...........................................78
FIGURE34PERFORMANCEINDICATORSFOROPTIMALECONOMICSYSTEMSABLETOREFUEL2CARSUNDER
DIFFERENTCAPITALCOSTREDUCTIONSCENARIOS.........................................................................78
FIGURE35ENERGYAUTONOMYVSMINIMUMELECTROLYSEROPERATIONFOR99.5%RELIABLESYSTEM.....80
FIGURE36ENERGYAUTONOMYVSELECTROLYSERCYCLESFORA99.5%RELIABLESYSTEM.......................81
FIGURE37RELATIONSHIPBETWEENOPTIMALPVANDELECTROLYSERCAPACITYFACTORS.........................82
LISTOFTABLES
TABLE1MODELVARIABLES............................................................................................................33
TABLE2PVARRAYSUBROUTINESUMMARY&CONSTANTDATA......................................................34
TABLE3FUELDEMANDSUBROUTINESUMMARY&CONSTANTDATA................................................36
TABLE4BATTERIESSUBROUTINESUMMARY&CONSTANTDATA.......................................................38
TABLE5ELECTROLYSERSUBROUTINESUMMARY&CONSTANTDATA.................................................41
TABLE6STORAGETANKSUBROUTINESUMMARY&CONSTANTDATA................................................43
TABLE7ECONOMICSSUBROUTINESUMMARY&CONSTANTDATA...................................................44
TABLE8STRATEGYSUBROUTINESUMMARY&CONSTANTDATA......................................................51
TABLE9CASESTUDYSYSTEMPARAMETERS......................................................................................61
TABLE10CASESTUDYRESULTS......................................................................................................65
TABLE11SINGLEOBJECTIVEOPTIMIZATIONSPECIFICPRODUCTIONCOSTPARAMETERSVALUESRANGE......66
TABLE12SINGLEOBJECTIVEOPTIMIZATIONSPECIFICPRODUCTIONCOSTOPTIMUMSYSTEMPARAMETERS 67
TABLE13SINGLEOBJECTIVEOPTIMIZATIONSPECIFICPRODUCTIONCOSTOPTIMUMSYSTEMRESULTS........68
TABLE14PARAMETRICSTUDIESPARAMETERSANDCONSTRAINTDEFINITION.........................................69
TABLE15DOUBLEOBJECTIVEOPTIMIZATIONSPARAMETERSANDCONSTRAINTDEFINITION....................79
LISTOFABBREVIATIONS
Description
Units
Energyautonomy
hours
Hydrogenautonomy
days
EffectiveareaofPVpanels
m2
BatteryCapacity
kWh
Batteriescapitalcost
Electrolysercapitalcost
PVcapitalcost
Batteriesspecificcapitalcost
PVspecificcapitalcost
SystemCapitalcost
storagetankcapitalcost
ElectrolyserSpecificEnergyConsumption
kWh/Nm3
Electrolysercapacityfactor
PVcapacityfactor
HydrogenStorageTankcapacity
Nm3
Actualenergythatbatteryischargedordischargedwith
kWh
Symbol
Batterychargingcapability
kWh
Maximumenergythatcanbestoredinbattery
kWh
Batterydischargingcapability
kWh
Maximumenergythatbatterycanprovide
kWh
Energyavailable/requiredforcharging/discharging,
kWh
Energyrejectedbybattery
kWh
Energyavailable/demandedforcharging/discharging
kWh
Energyconsumedbyelectrolyser
kWh
Energyrejectedbyelectrolyser
kWh
Minimumcontinuouselectrolyseroperation
hours
ElectrolyserOnoffswitchingcycles
cycles
_
_
ElectrolysermaximumONOFFswitchingcycles
ElectrolyserMaximumoperationpoint
cycles
%
ElectrolyserMinimumoperationpoint
Energyproducedbyphotovoltaicpanels
kWh
TotalEnergyrejection
Averagedailyhydrogenfuelconsumptionpercar
kg
Dailyhydrogenfueldemand
Nm3
DailyUnsatisfiedfueldemand
Nm3
Batterieslifetime
years
Electrolyserlifetime
Years
PVlifetime
Years
StorageTanklifetime
Years
Maintenancecostforbatteries
Totalmaintenancecostfortheelectrolyser
MaintenancecostforthePVpanels
Batteriesspecificmaintenancecost
_
_
Electrolyserspecificmaintenancecost
PVspecificmaintenancecost
SystemMaintenancecost
Storagetankmaintenancecost
Storagetankspecificmaintenancecost
Powerconsumedbyelectrolyser
kW
Electrolyserinstalledpower
kW
Powerrejectedbyelectrolyser
kW
ElectrolyserMaximumoperationlimit
kW
ElectrolyserMinimumoperationlimit
kW
PVinstalledpeakpower
kW
PVpowerproduced
kW
Poweravailableforelectrolyserpowering
kW
Electrolysernominalproductionrate
Nm3/h
ElectrolyserHydrogenproduction
Nm3
HydrogenFlowtotank
Nm3
BatteryMaximumcharge/dischargerate
kW
BatteryMaximumstateofcharge
Batteryminimumstateofcharge
TankMaximumStateoffulfillment
TankMinimumStateoffulfillment
Hydrogenspecificproductioncost
/Nm3
Ambienttemperature
Celltemperature
Celltemperatureatareferenceambienttemperature
TankHydrogenvolumeavailableforrefueling
Nm3
TankVolumeavailableforstorage
Nm3
Solartohydrogenconversionefficiency
Efficiencyofthepowerconditioningdevices
BatteryChargingenergyefficiency
DegradationfactorofthePVarray
BatteryDischargingenergyefficiency
Solarcellefficiency
Solarcellefficiencyatnetradiation
WiringefficiencyofthePVsystem
Annualpayment
Temperaturecoefficientofthesolarcell
C1
BatteryDepthofDischarge
Globalsolarradiation
kWh/m2
Higherheatingvalue
kWh/Nm3
Numberofcarsrefueledbythestation
cars
Nominaloperatingcelltemperature
Reliability
StateofCharge,
Stateoffulfillment
Iteration
Systemlifetime
Years
Discountrate
Timestep
hour
ACKNOWLEDGEMENTS
Iwouldliketothankmyfamilyandfriends,withouttheirsupport,studyingandliving
abroadwouldnotbefeasibleforme.SpecialthankstoProf.MarkvartandDr.Ponce
deLeonforthesupervisionofthethesis.
Wordscount:19,018
10
1. INTRODUCTION
Thecontinuingincreaseofthepriceofconventionalfuelsusedintransportsectorand
the global demand for reducing the resulted CO2 emissions, have set the gradual
replacement of the fossil fuels as urgent. Hydrogen can play significant role to this
transitionduetothefactthatitisconsideredasapotentiallycleanfuel.Nevertheless,
the method applied for the hydrogen production has influence on this feature [1].
Currently, about the 80% of the global hydrogen production comes from steam
methane reforming (SMR) of natural gas [2]. However, the natural gas demand and
pollutants emitted during the production via SMR do not contribute to a sustainable
nature of hydrogen. On the other hand, hydrogen can be produced with zero
emissionsbywaterelectrolysisincasethatofpoweringbyrenewableenergysources
[3].Tothisdirection,photovoltaics(PV)canbeservedasacandidatetechnologytobe
integratedinhydrogenproductionsystems[4].Atpresent,electrolysissystemsrelied
solely on PV have not applied widely due to the resulted high hydrogen specific
productioncostsandthelimitationsthatintermittentoperationofPVcause.
However,researchinteresthasbeenattractedbytheareaasbothtechnologies(PV,
electrolyser)areunderdevelopmentnotonlyintermsoftheirperformancebutalsoof
their investment costs. Reviewing the literature, the way of coupling PV and
electrolyser unit is the major concern of researches. The integration of power
regulation devices between the two technologies in conventional systems ensures
theircompatibilitybutresultstolowoverallsystemefficiencies[5].Toeliminatesuch
intermediate devices, several studies have approached the direct coupling of PV and
electrolyserexploitingtheirmodularnature.Besidestheirpromisingresults,notonly
system durability was not proved but also electrolyser performance showed
degradation after time extended experiment [6]. The variable nature of PV has
negativeimpacttoelectrolyserperformanceandresultstofrequentonoffswitching
and periods of low operating points. Not only alkaline but also proton exchange
membrane(PEM)electrolysers,arerecommendedtooperateattheirdesignpointfor
relatively long time periods, so to achieve the best possible values of efficiency and
prolong their operation without degradation and malfunctions[7,8,9]. To satisfy this
requirement,batteriesincorporationhasbeensuggestedinsystemswherehydrogen
11
Thepresentthesishasasgeneralsubjectthestudyofastandaloneelectrolysissystem
powered by PV and which uses batteries as shortterm energy storage. In brief, this
projectproposesthebatteryincorporationinstandalonePVhydrogensystemsforthe
effectiveandreliableelectrolyseroperation.Tobemoreprecise,theenergysimulation
andtheoptimaldesignofanautonomoushydrogenrefuelingstationconstitutetwoof
the project aims. For this scope, a numerical code is developed able to simulate the
system operation in energy terms. Furthermore, the optimum sizing of the system
components is resulted by the code optimization having not only technical but also
economical objectives. Project focuses on the cooperation among the subsystems
developing a power management strategy with respect to electrolyser operation
requirements.Theinvestigationoftheeffectthatsuchconstraintshaveonthesystem
technoeconomicviabilityisalsoanissue.Inaddition,theconditions(location,system
size, capital cost reduction) under which the system will be economically viable are
studied.
Regardingthestructureofthethesis,sixmainchaptersfollowthisintroduction.Firstly,
project objectives and diversity are stated at the second chapter where the system
description also takes place. The third chapter is devoted to the review of the
literature concerning with autonomous PV electrolysis systems. Both simulation and
experimental studies are discussed. However, the electrolyser operation limitations
andthewaythattheyareconfrontedincompletesolarhydrogensystemsarepointed
out.Atthefourthchapter,thefeaturesofthemodeldevelopedarepresented.Data
input, model variables, power management strategy and potential output are the
groupsofmodelfeatures.Furthermore,theresultsobtainedbythemodelapplication
arepresentedatthefifthchapter.Simulationresultsincludeapresentationofacase
study and a series of optimizations with, either single or multi, technoeconomical
objectives. Then, results are discussed thoroughly at the sixth chapter. Finally, thesis
finalchapterstatestheconclusionsextractedfromthestudy.
12
2. PROJECTFEATURES
2.1. GENERALDESCRIPTIONOFTHESYSTEM
Figure1Simplifiedsolarhydrogensystemscheme
Asmentionedintheintroduction,thestudyofasystem(Figure1)capabletoproduce
hydrogen fuel via water electrolysis powered by PV is the general subject of the
project. In addition, leadacid batteries, as shortterm energy storage system, are
included in the system having an assistive role in the powering of the electrolyser
maintaining its operation stable. Even though that such a system is noted as
conventional in the literature [8,15], the fact that electrolysers face performance
degradation under intermittent operation makes the presence of storage devices in
standalone systems required [7,16]. In addition, hydrogen storage tanks are also a
part of the system as the distribution of the fuel produced is not supposed to be
instant. Oxygen is also produced by electrolyser but its exploitation is not subject of
thepresentstudy.
Tonoticethateventhoughthatthemaincomponentsofthesystemarepresentedin
Figure1asblackboxes,suchasystemismuchmorecomplex.However,thepresent
projectfacesthesubsystemsasboxessimulatingthecooperationoftheminenergy
termsanddoesntfocusoninternaloperationofthem.
13
The described system can be proposed as a small hydrogen refueling station able to
coverthedemandofsomehydrogenfueledvehicles.Suchdistributedstationsthatcan
be installed also in every home or neighborhood may have greater specific capital
costs but provide domestic independency. Hence, variation of the fuel purchase, as
petrolisfeaturedby,isavoidedbecausethecustomerisalsotheproduceratthesame
time. To this direction the project also deals with the investigation of how
economically viable is the application of a small system at present and in the future
undertheexpectedtechnologiescostreduction.
2.2. AIMS&OBJECTIVESOFTHEPROJECT
The general aim of the project is the energy simulation and the technoeconomic
optimizationofastandaloneelectrolysissystempoweredbyPVandbatteries.Forthis
scope, a numerical code will be developed which using short time step and having
solar radiation data series and fuel demand as inputs will be able to export techno
economic magnitudes for the specific dimensioned system evaluation. Significant
attentionisgoingtobepaidonthepowermanagementstrategythatsystemshould
followforbestperformanceandeconomicalresultswithrespecttotherequirementof
electrolyser for steady operation. The satisfaction of subsystems operational
requirementsineconomicallyviablesystemsservesanaimandwillbeinvestigatedby
theoptimizationofthenumericalcode.Finally,thesensitivityofthesystemfeatures
tofuturepossibleimprovementsincostandperformancecanbestudied.
Concerningtheobjectivesthatcanbeachievedbythecompletionoftheproject,they
arestatedbelow:
Thedevelopmentofamodelabletosimulatethesystemoperationinenergy
termsandtobeusedfortechnoeconomicaloptimizationofit.
14
The development of a power control strategy for the effective and reliable
operationoftheelectrolyser.
2.3. THENOVELTYOFTHEPROJECT
Followingtheconclusionsoftheliteraturereview,thegeneralfieldswheretheproject
issupposedtoprovideanalternativeapproachofstudyare:
Theenergysimulationofastandalonesolarhydrogensystem,whichseemsto
be more applicable at present in contrast to innovative but not durable yet
directlycoupledsystems.
Theinvestigationofthereliabilityoffuelingstationwithspecifiedfueldemand.
Theinvolvementofpowermanagementstrategyinthemodelingofthesystem
aimingtotheelectrolysersustainableoperationandtheenergyproducedbest
exploitation.
Mostoftheabovestatementsareverycommoninotherresearchareas,buthavent
beenincludedindesignstudiesforautonomoussolarhydrogenfuelstations.
15
3. LITERATUREREVIEW
Atthischapter,adiscussionaboutstudiesrelatedtotheprojecttopictakesplace.At
first,somedesignstudieshavebeentakenintoaccount[12,13,14,17].However,itwas
obviousthatthemostrecentresearchpublicationsapproachexperimentallythedirect
couplingofPVwithelectrolyser[5,6,18,15,19].However,theperformancedegradation
oftheelectrolysersundervariableoperatingconditionsisanissue,astheirdurability
hasnotbeenprovedyet[7].Thegeneralabsenceofexperimentalresultsregardingthe
performanceoftheelectrolyserthroughalongtimeperiodisnoticeable[6].Forthis
scope,thewaysthatpowerinputtotheelectrolyserisconditionedinlargersystems
(completehydrogencyclefuelcellintegrated)usedforelectricityproductionarealso
presented[7,9,10,11].
3.1. HYDROGENFUELSTATIONSSIMULATIONSTUDIES&AN
APPLICATION
A few simulation studies were pointed in the literature developing models for the
technoeconomical evaluation of complete solar hydrogen systems. All the systems
simulated were consisted of PV coupled with electrolyser via power conditioning
devices, while in some of them batteries used for the auxiliary systems powering
[12,13,14,17].
Tobeginwith,theperformanceofahomesolarhydrogenproductioninstallationwas
studied by Hollmuller et al [17]. The system is mainly consisted of PV panels, an
alkaline electrolyser, batteries used only for powering auxiliary devices and storage
tanks(Figure 2).Also,systemisconnectedtothegridforelectrolyserregulation.The
controlunitpermitsdifferentoperationmodesconcerningthecooperationamongPV,
batteries and grid to be applied manually. After measurements taken for 3 days,
features of each subsystem were extracted. Electrolyser operating temperature took
about 1h and 15 min after the first startup of the day to be stabilized. It should be
pointedthatduringthemeasurementsperiod,electrolyserswitchedoffandrecovered
inaveryshorttimeduetostrongcurrentvariations,butthereisnomentionaboutthe
16
Figure2Schematicdiagramoftheenergyandmaterialsflowinhomesolarhydrogen
productionsystem[17].
Moreover,astudyhasbeenconducteddevelopingamodelfortheoptimizationofthe
performance of large scale PV electrolyser systems [13]. The paper provides details
concerning the modeling of the system while the procedure for the system cost
estimation is also described. To comment, however, that the cost for the means of
hydrogen storage seems not to be taken into account. PV and electrolyser are
assumedtobeconnectedthroughMaximumPowerPointTracker(MPPT)andDC/DC
17
Figure3Thelevelizedsolarhydrogencostversustheaverageyearlytotalsolarenergyonthe
horizontalsurfaceforfixedandtrackingpanelssystems[13].
Thesameauthor,asfollowtothepreviouspublication,developedasimplifiedmodel
for a PVelectrolyser system. Concerning the numerical code developed, it was
consistedbyfourpartsofcalculationsaboutthesolarradiationreceived,theelectric
energy produced by PV, the hydrogen produced by electrolyser and the system
economics. It was resulted to correlations between couples of magnitudes such as
annualelectricalenergyproductionannualsolarradiation,hydrogenenergyannual
solarradiation,hydrogenproductioncostratioofPVpanelsandelectrolysercostto
18
annualsolarradiation(Figure4)etc[12].Moreover,itwasprovedthatelectrolysersize
increasesalmostproportionallywiththeannualtotalradiationreceived. Studyalso
focusedespeciallyonthecostofhydrogenproductionconductingtwocasestudiesfor
presentandfuturecostconditions.Resultsshowedthattheoverallsystemefficiency
variesfrom9.34%to8.64%whileaconclusionwasthatfurthercapitalcostreduction
isneededforviabilityofthesystem.
Figure4Acorrelationforhydrogencostforfixedandsuntrackingpanels[12].
Ontheotherhand,J.M.Vidueiraetal.presentedtheenergysimulationforacomplete
solarhydrogenproductionsystemabletorefueltwofuelcellbuses[14].Tonoticethat
abatterysystemisincluded,however,justforthepoweringofotherthanelectrolyser
devices. The sizing of the system was based on reasonable sequence of assumptions
whileaneconomicassessmentwasalsoconducted.Asimulationofthesystemcarried
outanditwasreachedtothefactthatthesizingwassuccessfulasthereliabilityofthe
system took values at the expected range. Electrolyser was allowed to operate in a
widerangeofpowerinput.Furthermore,differentalternativesofsystemdesignsuch
astheconnectionofPVtothegridwereinvestigated.Eventhoughthatthecalculation
of the total system cost is not described, it was found that the hydrogen production
costforanautonomousstationarysystemis1.06/Nm3.Thestudyconcludedthatat
that time (2003) in order the investment to be even more attractive, government
subsidies, performance improvement and cost reduction of the technologies are
required.
19
To sum up, simulation studies dealing exclusively with hydrogen production systems
powered by PV are very limited and not recent. Also, sizing methods based on
numerical optimizations of the system were not noticed. To point out, that special
considerations about the electrolyser operation have not been taken during the
modelsdevelopment.Finally,concernsaboutthereliabilityofthesystemwerefound
inonlyonestudy[14],buttheeffectthathasonthesizingandperformancefeatures
ofthesystemwasnotinvestigated.
TheconceptofdirectcouplingofanelectrolyserandPVhasbeendiscussedbymany
authors recently [5,6,18,15,19,20,21]. Such an approach is proposed for the
elimination of electronic devices. These devices are used for the conditioning of the
power sent to the electrolyser butthey are responsible for energy losses. Also, their
installation adds to the system capital cost. Experimental studies have showed that
directcouplingofsolarcellsandelectrolysercellsappropriatelydesignedisachievable
anddrivestoanincreasedoverallconversionefficiency.However,suchstudieslackof
timeextendedexperimentalresultssotoprovethedurabilityofthesystem.
Figure5CharacteristicofelectrolyserandMaximumpowerpointsofPVforoptimal
configurationofPEMcellsandPVpanels[18].
20
PaulandAndrewsinvestigatedthepossibilityofdirectlycouplingbetweenaPVarray
andproton exchangemembrane(PEM)electrolyser[18].Studyingthecharacteristics
ofcommercialdevicesofthetwotechnologiesandseveralcombinations,theyended
uptoanoptimalconfigurationofseriesparallelconnectionnotonlyofPVpanelsbut
alsoPEMcells.Theresultwas,asshownin Figure5,theasmuchaspossiblematching
of their IV characteristics in order to achieve the almost maximum power transfer.
Couplingtheoptimalcombination,someinitialexperimentsshowedthata95%ofthe
theoretical maximum energy transfer can be achieved. Contrary to MPPT electronic
devices that have typical losses of about 10%, direct coupling seems to be superior.
Also, such an approach advances due to the fact that thecosts of converterscanbe
avoidedsotoresulttoloweroverallcapitalcost.Authorsnoticethatexperimentslast
for a short period so these findings cannot be confirmed by the study. Finally, they
state their prospect to investigate in a future work whether the configuration that
resultstomaximumpowertransferisalsotheoptimalineconomicterms.
Figure6Electrolyzerefficiencyversuselectrolyzercelltemperatureforarelativelyconstant
current(125130A)measuredoverthe14daystudyperiod[5].
SignificantevolutionontheareaispointedbyateamofGeneralMotorsR&Dcenter
who have published recently a series of experimental and modeling studies
approaching the direct connection in the system [5,15,19]. The first time study
conducted presented the results from the operation of a PVelectrolysis system [5].
Even though electrolyser and PV were directly connected, electrolyser was also
21
powered by the grid and batteries. An average daily solar to hydrogen energy
efficiencyof8.5%over14daystestingwasachieved.Specialnoticewasshownatthe
effect that electrolyser temperature has on the electrolyser temperature. It was
pointed that operating at high temperature close to the designed value, drove to
currentandconsequentlytoelectrolyserefficiencyincrease(Figure6).Suchafactadds
an advantage of constant electrolyser operation. However, wide variability of the
electrolyser operation due to changing weather conditions didnt result to serious
performancedegradationduringthese14daysoftesting.Authorsclaimedthatfurther
experimental results under longer testing period and wider range of operating
conditionsareneededfordurabilityconfirmation.
Figure7AveragesolarhydrogenproductionefficiencyusingPVelectrolysersystemswithDC
DCconvertersforoptimizationversusdirectconnection[15].
Then,GibsonandKellycomparedadirectlycoupledsystemwithasystemthatusesa
DC/DC converter for matching the PV voltage to electrolyser voltage [15]. Several
combinations of configurations and different models of panels were tested and
resulted to maximum efficiencies of 12.4% and 10.6% for the direct and indirect
couplingrespectively(Figure7).Regardingthedirectcoupling,successoftheapproach
is relied on the selection of PV modules which should match with the electrolyser
operating voltage and have a good electric efficiency. Authors of this study dont
mentionthetimeofexperiment.Moreover,itcannotbeprovedbytheanalysisthat
most efficient systems have less cost than systems with DC/DC converters of lower
averageefficiency.
22
Furthermore, in a more recent work a model developed for the prediction of the
efficiency of both systems [19]. Modeling results showed high level of accuracy
comparingtoexperimentalmeasurements.Itwaspointedoutthatconnectionthrough
DC/DCconvertermadethesystemcapabletoproducehydrogenduringcloudyperiods
when direct coupled system stop to operate. Hence, a PVelectrolysis system that
allows both direct and indirect coupling depending on the weather conditions by
including adjustable DC/DC converters and switches is suggested as the best option
[19].
Clarkeetal.alsostudiedtheapproachofdirectcouplingofelectrolyserandPVarray,
demonstratingtimeextendedexperimentsunderstronglyvariableconditions[6].The
mainconclusionofthepaper,whichissimultaneouslyanissuethatwasnottakeninto
account by the previous studies, was that the performance of the electrolyser stack
degradedafterthefirstfivedaysoftheveryearlystagesoftheexperiment(Figure8).
The widely variable power inputs as well as the inappropriate stack design were
thoughttobeoneofthepossiblecausesforthisrapiddegradation.
Figure8MeasuredstackVIandefficiencyinitially(left)andover5dataperiod(right)fora13
cellstack[6]
The overall system efficiency found to be around 4.7%, largely differing to similar,
previously discussed, studies [5,18,15]. However, emphasis is given to the fact that
resultsobtainedafteralongtestingperiod(1519h)whenthesystemperformedsafely
andsatisfactorily.
23
Inaddition,Arriagaetalapproachedthedirectcouplingofacommercialelectrolyser
withanappropriateconfigurationofPVpanels[20].Thecomparisonbetweencurrent
voltage characteristics showed that efficient energy transfer can be conducted
between600800W/m2ofradiation,whichistypicalvalueforMexico.Asitisreviewed
in other study [6], the specific research lacks of quantitative results about energy
transferandalsotestingperiodisnotmentioned.
Moreover, a model based on experimental results has been developed for the
predictionoftheperformanceofdirectcoupledsolarhydrogensystem[21].However,
theoverallefficiencyachievedwasabout5.15.4%dependingontheirradiancelevel
as the system configuration was not optimized. Authors, finally, noticed that model
canbealsoappliedwithgoodaccuracyfordifferentsizedsystems.
Asaconclusiontothisgroupofstudies,itcanbesaidthatapartfromthepromising
resulted values of efficiency direct coupled solar hydrogen systems not only havent
proved their durability but also electrolyser performance shows degradation over
variable operation. On the other hand, DC/DC converters provide the capability of
solartohydrogenconversionduringunfavorableconditions.Also,therearedoubtsif
the appropriate configuration of the PV array and the electrolyser stack required for
thedirectcouplingcorrespondtolowercapitalcosts.
3.3. THELIMITATIONSOFCOUPLINGPVANDELECTROLYSER
Besides the recent research efforts to connect directly the PV array and the
electrolyserstack,thereareseveralissuesconcerningtheelectrolyseroperationunder
intermittentpowerinputandlowloads.
Currently,twotypesofwaterelectrolysersareavailableinthemarket,thealkalineand
the proton exchange membrane (PEM) electrolysers [22]. Alkaline electrolysers are
proposedforlargeunits,havinglowerenergyconsumptionandcapitalcostthanPEM
[22,23].Also,anoperationrequirementthatalkalinetypehasisthelimitedacceptable
range of power input which is about 2550% of the rated capacity [7]. On the other
24
hand,commercialPEMelectrolysersaresuitableforsmallinstallationspromotingthe
capability to operate at a very wide range of power loads [8,22]. However, their
durability has not been proved yet [7], while some concerns about the variable
operationhavebeennotedintheliteratureandpresentedbrieflybelow:
Atverylowloadstherateofhydrogenandoxygenproductionmaybelower
thantherateneededtopermeatethroughtheelectrolyteandmixwitheach
other [8]. Such a situation can drive to hazardous conditions inside the
electrolyser. Hence, electrolyser must be set out of order when energy
transferfromPVbecomeslowerthanaspecifiedlimit.
Duetointermittentoperation,electrolysermayoperatemostofthetimeat
a lower temperature than nominal, a fact that results in a lower efficiency
[8]. As a consequence, power input is preferable to be either constant or
variable at a narrow range for a continuous operation time period so
nominaltemperatureandhighestpossiblevaluesofefficiencytobeachieved
byelectrolyser.
Frequentonoffswitchingshouldbeavoidedasthecommercialelectrolysers
arefeaturedbyamaximumofcycles,afterwhichperformancedegrades[9].
Also,afteranelectrolysershutdown,ittakesabout3060minutestobeable
tostartupagain[7].Electrolyserstartupshouldbefollowedbyaseriesof
requirementswhichensureitsoperationtobemaintainedforatimeperiod
andasuddenstopnottobecaused.Duetovariablenatureofsolarenergy,
gridassistanceorashorttermmeansofenergystoragewouldberequired
for powering the electrolyser in case of insufficient energy production by
PV[8]. Minimization of electrolyser cycles should be an issue for solar
hydrogensystems.
InacouplingthroughDC/DCconverter,maximumefficiencyoftheconverter
canbeachievedinaverynarrowrangeofpowernearitsdesignpoint[8].As
a result, the power sent to the electrolyser should be leveled by means of
energy storage at the designed range of DC/DC converter for the power
losseselimination.
25
Overall,itcanbeconcludedthatindependentlytotheelectrolysertypeandthenature
of coupling, power input should vary in narrow limits near the design point of the
electrolyser,somaximumvaluesofefficiencytobeachievableforthelongestperiod
possible and malfunctions to be avoided. From that point of view, standalone solar
hydrogen systems can be reliable only in case that shortterm energy storage, like
batteries, is included to cover the deficit of power available comparing to almost
constant power demand. Finally, electrolyser onoff switching can be eliminated as
power availability would not be relied only on an unpredictable and intermittent
energysource.
3.4. WAYSTOOVERCOMECOUPLINGLIMITATIONS
Asanalyzedintheprevioussection,anasstableaspossibleandneartothenominal
operation is desired for the electrolyser. Such a condition cannot be satisfied by the
exclusivepoweringbyPVduetotheirvariabilityofenergyproduction.Asolutionisto
combine PV with input from the grid [8]. Electrolyser operation at design point,
increase of electrolyser capacity factor, and economics improvement would be the
advantagesofthisapproach.However,whengridisnotavailableandtheobjectiveof
the approach is to produce totally zero emissions hydrogen, a short term energy
storagesystem,asbatteriesare,wouldbeanalternativeoption.
Figure9ElectrolysercyclesvsHysteresisandstateofchargeofbattery[9]
26
TheincorporationofabatterywasapproachedinHarIprojectwheretheperformance
of a system renewablehydrogen system was studied [9]. The postponement of
alkalineelectrolyser(34kW)performancedegradationduetorapidvariationofenergy
production by wind turbines was the reason for batteries installation. Advance
batteries were used to capture peaks and troughs of energy production. In this
application, electrolyser operation is determined by the battery state of charge.
Electrolyser can operate when the charge of battery is between an upper and lower
level (hysteresis). System simulation resulted that the largest the hysteresis the less
electrolyser cycles noticed (Figure 9). Optimization showed that electrolyser cycles
wereminimizedusingstateofchargecontrol.
The effect that power management strategy has on the performance of standalone
systemswhererenewableenergysourcesincorporatewithelectrolyser,fuelcellsand
batteries to cover the demand of a load, has been investigated [10,11]. Ipsakis et al
developed and compared the effect of three different power control strategies. The
reduction in accumulator cycles and the avoidance of frequent electrolyser startups
were objectives of the study [11]. The effect of minimum allowable battery state of
charge on the system performance was one of the parameters. Focusing on the
electrolyser battery cooperation, it was ended up that when the solar radiation is
insufficient, battery should assist the electrolyser powering so to operate at its
minimumanddeteritfromsuddenstops.Strategieswhereelectrolyserwaspowered
solely by renewable energy sources could not guarantee the smooth operation of
electrolyser besides the highest hydrogen production resulted. Authors recommend
thebatteryassistancetotheelectrolyserpoweringforinstallationsinareasofnothigh
solarpotentialasaviableoption.
Ulleberg developed a model for the simulation of a complete PVH2 system which is
verifiedbytheoperationofPHOEBUSplantinGermany[10].Theaimofthepaperisto
investigatethedemonstrationtheofpowercontrolstrategiesinsuchsystems.Tothis
direction, the comparison between electrolyser operation under constant and not
widely variable conditions was conducted. System simulation showed that variable
operation of the electrolyser advances over fixed mode due to the fact that better
27
systemperformance,lesselectrolyserstartups,moreruntimefortheelectrolyserand
reduction in battery use resulted. However, it must be commented that it was not
describedhowsteadyoperationofelectrolyserwasapproached.Also,duetothefact
that system objective is to provide electricity to a load, electrolyser operation is
thoughtto beseasonal.Asaresult,electrolyserundergoesonlyfavorableconditions
during the summer. Author suggests, finally, that a technoeconomic optimization of
thesystemtakingintoaccountbothsystemdesignandcontrolstrategyisnecessary.
Toconcludethat,powermanagementstrategiesfortheelectrolyseroperationbased
on the battery state of charge has been noticed in the literature. In these studies,
systems are much more complex, using hydrogen as longterm storage and not as a
product. The area of solar hydrogen production systems lacks of involving control
strategyinthedesignofthesystem.However,conclusionsandideasaboutthepower
management strategy can be used as reference and fitted in a solar hydrogen
productionscheme.
3.5. SUMMARY
Theanalysisofthestudiesthattookplaceintheprevioussectionscanbesummarized
atthefollowing:
Stronglyvariablepowerinputtotheelectrolyser,evenviaDC/DCconverters,
resultstolowerefficiencyandelectrolyserperformancedegradation.
Asitcanbenoticedlater,literaturereviewconclusionsdirectthemethodologyandthe
objectivesofthepresentdissertation.
29
4. MODELDESCRIPTION
4.1. INTRODUCTION
The present chapter is devoted to the description of the model developed for the
energysimulationofastandaloneelectrolysissystempoweredbyphotovoltaicsand
assistedbybatteries.CodedevelopmentconductedusingMatlabsoftwareaswellas
theoptimizationofitthattookplaceapplyingtheMatlabOptimizationToolbox(MOT).
Ingeneral,modelapproachestheoperationofaspecificsizedsysteminenergyterms
while an economic evaluation of it can also be exported using empirical economic
data.Figure10presentsgeneralflowchartofthecodewhichreflectsthemethodology
themodeldevelopedwith.
Themodelisconsistedby3mainparts.Atthefirstpart,thedatarequiredforthecode
run are stated. Hence, both variables and constant data for the system components
mustbeprovided.Subsystems(PV,batteries,electrolyserandhydrogenstoragetank)
size and other system features such as number of automotives refueled, minimum
daily electrolyser operation are considered as model variables. The values of these
parameters are defined by the user manually. Regarding the constant data, some
typical values were assumed for magnitudes needed for the modeling of each
component.Furthermore,numericalcodealsohasasinputasolardatatimeseriesof
1hourtimestepwhichisusedforthecalculationofhourlyPVenergyproductionby
thespecificsizedpanels.PVenergyproductionandotherhourlyvariablemagnitudes
suchasthebatterystateofchargeandhydrogenstoragetankstateoffulfillmentare
usedforpowermanagementstrategyimplementation.Atthesecondpart,thesystem
operation strategy, the heart of the model, is developed. The power control is of
significant importance for the system operation as it should be able to choose the
most efficient route to exploit the energy produced by the PV panels, confronting
simultaneously with the electrolyser operation limitations that mentioned at the
literature review. Applying the strategy for each time step, respective operation
magnitudesareresultedforeachsystemcomponent.
30
START
MODEL OVERVIEW
FLOW CHART
i=1
YES
HOUR OF
THE DAY=9
NO
CUMULATIVE MAGNITUDES
CALCULATION
YES
i=i+1
i<8760
NO
CALL ECONOMICS FUNCTION
SYSTEM CAPITAL & MAINTENANCE
COSTS ESTIMATION
EVALUATION MAGNITUDES
CALCULATION
END
Figure10Modeloverview
31
Finally, after repetitions that correspond to 1 year period (8760 h), summarized
magnitudesandperformanceindicators,suchastheannualhydrogenproductionand
the annual unexploited energy (energy rejection from now on), are estimated and
can be used for specificdimensioned (according to variable parameters) system
evaluation. In addition, assuming no precise but average specific capital and
operationalcostsforeverydevice,systemcostcanbeestimatedbasedeitheronthe
variable parameters or the summarized performance magnitudes. As a result,
economic evaluation of the system can be also implemented, calculating the specific
hydrogenproductioncost.
However, the detailed analysis of the code is carried out at the following three
sections. At first, technical analysis of the modeling for each system component and
the assumptions made are presented. To notice, that each components model is
developed at a separate programming function (subroutine) and is interconnected
with the others through the operation strategy function. In addition, the empirical
equationsusedforthesystemcostestimationbythemodelthatareincludedinthe
economicsfunctionarestated.Then,adiscussionforthepowermanagementstrategy
developed and used in the model takes place. The flow chart that reflects the
recommended operation strategy is given while the exported timestep magnitudes
are also defined. At the last section, the definitions of cumulative magnitudes and
resulted indicators (technical and economic) which are used for the specific sized
systemevaluationareprovided.
4.2. DATAINPUTSUBSYSTEMSMODELINGDESCRIPTION
4.2.1. TECHNICAL
Priortotheanalysisofthemodelingforeachsystemcomponentandthepresentation
oftheconstantdataincludedintherespectiveMatlabfunctions,modelvariableshave
to be defined. The magnitudes, which values should be entered by the user and the
systemperformanceisreliedon,canbefoundinTable1.Inaddition,modelvariables
areusedasinputinsubsystemsandpowermanagementstrategyfunctions.Theeffect
32
Description
Abbreviation
PVinstalledpeakpower
BatteryCapacity
Electrolyserinstalledpower
TankCapacity
Minimumcontinuouselectrolyseroperation
Numberofcarsrefueledbythestation
Location
Units
kW
kWh
kW
Nm3
Solardatatimeseries
Hours
Cars
Then, the subroutines (functions) developed, for the modeling of each part of the
system,areanalyzed.Also,tonoticethatsystemcomponentsmodelingispurelybased
on data reported in the literature. The interconnection among the subroutines is
accomplishedbythepowermanagementstrategy.Anymagnitudepresentedbelow,is
expressedinunitswhichhavebeenstatedinthelistofabbreviations.Onlyifspecific
referenceexists,unitsofmagnitudesdonotfollowthelistprovided.
4.2.1.1.
PHOTOVOLTAICARRAY
Photovoltaicarrayconstitutestheprimaryenergysourceforthestandalonesystem.
The present subroutine has the capability to estimate the energy produced by
photovoltaicpanels,
,inaspecificiteration, .Theinstantambienttemperature,
,theinstantglobalsolarradiation,
,andthePVinstalledpeakpower,
mustbeprovidedasinputs(Table2).
Concerningtheinstantmagnitudes,
and
,thatarerequired,datatimeseries
of1hourtimestepfordifferentlocationsareused.Inthisproject,weatherdatatime
series were extracted from the METEONORM software. The locations chosen the
installation of such a system to be investigated for Athens (Greece), Frankfurt
(Germany) and Tokyo (Japan) applying the corresponding data time series. Also, the
33
solarradiationseriesisdecidedtobeforpanelinclinationequaltothelatitudeofthe
locationforbestannualsolarenergyreceived[24].
Ontheotherhand,thePVinstalledpeakpower,
,isoneofthemodelvariables
,byeq.1,assumingthat1
andisrelatedwiththeeffectiveareaofthePVpanels,
m2ofareacorrespondsto100WpofPV[25].
10
(1)
Table2PVarraysubroutineSummary&ConstantData
Functionname: PVarray
Input:
AmbientTemperature,
Globalsolarradiation,
PVinstalledpeakpower,
Output:
PVenergyproduction,
PVenergyproduction,
ConstantData
Description
Abbreviation
Nominaloperatingcelltemperature
Value
Units
48
Solarcellefficiencyatnetradiation
14
Temperaturecoefficientofthesolarcell
0.005
C1
Celltemperatureatareferenceambienttemperature
25
DegradationfactorofthePVarray
9599
Efficiencyofthepowerconditioningdevices
95
WiringefficiencyofthePVsystem
95
previously.
Thesolarcelltemperatureaffectssignificantlythecellefficiencyandisgivenbyeq.2:
34
(2)
where,
,isthenominaloperatingcelltemperature[24].
ThevalueassumedinthepresentstudyfortheNominalOperatingCellTemperature,
,aswellasforotherconstantmagnitudesareprovidedinTable2[24,26].
,byeq.3:
where,
(3)
,thesolarcellefficiencyatnetradiation, ,temperaturecoefficientofthe
solarcelland ,thecelltemperatureatareferenceambienttemperature[26].
Hence,thePVenergyproductionforthespecifictimestepiscalculatedbyeq.4,taking
into account also the degradation factor of the PV array,
powerconditioningdevices, ,andthewiringefficiencyofthePVsystem,
[26].
(4)
Due to the fact that solar panel efficiency decreases in low irradiances, degradation
factor,
between95100%forthelowestandhighestradiationrespectively.
Finally,thepowerproduced,
incasethetimestepis ,isgivenbyeq.5.
/ (5)
4.2.1.2.
HYDROGENFUELDEMAND
,thatstationshouldprovidein
dailybasis.Theinputrequiredisthenumberofcarsrefueledbythestation, ,which
issystemvariable.
35
Assumingatypicalvalue(Table3)forthedailyfuelconsumptionbyahydrogenfueled
car,
, close to assumptions made in other studies [12,25], the daily hydrogen fuel
demand,FD ,inunitsofNm3(1kgofH2isequalto11.891Nm3)isgivenbyeq.6.
11.891 (6)
Table3FueldemandsubroutineSummary&ConstantData
Functionname: FuelDemand
Input:
Numberofcarsrefueledbythestation,
Output:
Dailyhydrogenfueldemand,
ConstantData
Description
Abbreviation
Averagedailyhydrogenfuelconsumptionpercar
Value
Units
Kg/daycar
Fortheoperationstrategysimplification,fueldemand(carsrefueling)issupposedto
comeuponlyonceperdayandlateatnight(9pm)whenelectrolyserisswitchedoff.
4.2.1.3.
BATTERIES
BatteriesasshorttermenergystorageeitherabsorbtheexcessofPVenergyoutputor
powertheelectrolyserlevelingtheenergycomingfromPV.Hence,ineverycomputing
timestep, the battery state of charge and the charging/discharging capability should
be estimated. The function for the batteries modeling requires as input the current
stateofcharge,
1 ,thebatterycapacity,
,thedepthofdischarge,
andtheenergyavailable/requiredforcharging/discharging,
.Table4provides
asummaryforthespecificsubroutine.Dueto complexityofthissubroutine,itsflow
chartispresentedinFigure11.Thereafter,thesequenceofcalculationsandstrategyin
thefunctionarealsodescribed.
36
Figure11Flowchartofbatteriessubroutine
Inthisstudy,batteryisapproachedasblackboxinenergyterms,assumingabattery
capacity and the limits that state of charge can vary. The minimum state of charge,
,isdefinedbyeq.7.
(6)
37
Table4BatteriessubroutineSummary&ConstantData
Functionname: Batteries
Inputs:
Currentstateofcharge,
Batterycapacity,
Depthofdischarge,
Energyavailable/requiredforcharging/discharging,
Output:
NewStateofCharge,
Energy/Powerabsorbed/providedbythebattery,
Energy/Powercharged/discharged,
Energy/Powerrejected,
ConstantData
Description
Abbreviation
Value
Units
Chargingenergyefficiency
85
Dischargingenergyefficiency
85
Maximumcharge/dischargerate
/10
Maximumallowablestateofcharge
DepthofDischarge
kW
100
60
Depthofdischargeforleadacidbatteriesisdirectlyrelatedwiththeirlifecycle[27].
However,duetothefactthatdepthofdischarge,
,issetat60%inthissimulation
approach, it was assumed that battery life cycle ranges at about 5 years, which is a
typicaltimeperiodforreplacement[24].
/10(7)
38
Generally, battery capacity is measured in Ah, is not fixed and varies regarding
conditions such as battery end voltage, discharge rate and temperature [24,28].
However, the nature of the study does not require detailed modeling of the battery
and its capacity is considered as fixed in units of kWh. The constant work voltage is
also assumed here. To repeat that total battery capacity,
, is a model variable
(table1).
Ineverycomputingtimestep,thecharginganddischargingcapability,
_
and
correspondingmaximumrateareestimatedapplyingtheeq.8,9.
(8)
(9)
Consequently,themaximumenergythatcanbeprovidedtoelectrolyser,
orcanbestored,
,atthespecifictimesteparegivenbyeq.10,11including
thelossesobservedduringcharginganddischarging.
(10)
(11)
Regardingthecharging,
,andthedischarging,
efficiencywhichareincludedin
eq.10,11, typical values have been assumed (Table 4). To be more precise, it is
accepted that the roundtrip energy efficiency for a leadacid battery is about 70%
[29,30].Hence,acharginganddischargingefficiencyof85%each(0.85 0.85
thatassumedinthisstudyseemsreasonable.
39
0.72)
magnitudeappearsnegative.Otherwise,energyavailableexistsforbatterycharging.
Soif,
0,batterydischargingisrequired.However,theamountofenergythat
batterycanprovideinthistimestephasalreadybeenestimatedineq.10.Thebattery
is discharged only if the stored energy,
demand,
.Asaresult,thenetenergyprovidedbythebattery
is:
0,
_
_
(12)
(eq.13).
(13)
Insuchcase,theenergyrejectedbythebatteryiscalculatedbyeq.14.
(14)
It is obvious that sign separation is maintained also for the energy provided or
absorbed,
lossesisgivenbyeq.15.
40
(15)
Inanycase,thenewstateofcharge,
,isestimatedapplyingeq.16.
(16)
/ (17)
4.2.1.4.
ELECTROLYSER
Asitwasdescribedforbatteries,electrolyserunitappearsalsoasa blockboxinthe
systemmodel.Electrolyserfunctioncalculatesthehydrogenproductioninthespecific
timesteprequiringthepoweravailable,
,
,andtheinstalledpowerofelectrolyser,
asinput(Table5).
Table5ElectrolysersubroutineSummary&ConstantData
Functionname: Electrolyser
Inputs:
Installedpower,
Output:
Poweravailable,
,
Hydrogenproduction,
Power/Energyconsumed,
Power/Energyrejected,
/
_
ConstantData
Description
Abbreviation
SpecificEnergyConsumption
Value
Units
kWh/Nm3
Minimumoperationpoint
95
Maximumoperationpoint
100
The literature review that presented in the first chapter ended up that electrolyser,
independently to the type it is, should operate at a very narrow range close to its
41
,operationpointsdefinethepoweringlimitsofelectrolyser(eq.18).
(18)
,shouldvarybetweentheselimits.Hence,incasethat
Then,thepowerconsumed,
theavailablepower,
,isgreaterthanthemaximumlimit,anamountofpoweris
rejected and the electrolyser operates at its design point. On the other hand,
electrolyserissetoutoforderwhentheavailablepowerisnotsufficient.Inbriefthe
powerconsumedbytheelectrolyser,
,isdefinedineverytimestepas(eq.19):
0,
,
(19)
Moreover,thepowerrejected,
,isestimatedbyeq.20.
(20)
Settingtheminimumoperationpointclosetothedesigned(table5),itcanbeassumed
thatthespecificenergyconsumption,
,(orelectrolyserefficiency)isconstant.Ithas
been shown that operation in low loads results to higher than nominal values of
energy consumption [31]. It is accepted that commercial PEM electrolysers are
featuredbyspecificenergyconsumptionof4.56kWh/Nm3[8].
Inthisstudy,itisassumedthataPEMelectrolyserischosenasithasthecapabilityof
high pressure and purity hydrogen production and eliminates the demand for
hydrogencompressorsandpurificationunitsbeforestoringintanks[1,8].Also,small
42
PEMelectrolyserunitsare,currently,availableinthemarketandseemsuitablefora
smallhydrogenrefuelingstation.
Finally,thehydrogenproductioninthetimestep, ,canbecalculatedbyeq.21.
(21)
4.2.1.5.
HYDROGENSTORAGETANK
Ignoringanyintermediatetreatmentofthehydrogenproducedbeforeitsstorageand
any losses that may appear during the hydrogen input or output, the subroutine for
storage tank estimates the updated state of fulfillment,
for storing,
, and refueling,
1 ,andthe
whichismodelvariable,stateoffulfillmentatprevioustimestep,
hydrogenflow,
,eitherasdemand(negative)orstorage(positive),arerequired
asinput.
Table6StorageTanksubroutineSummary&ConstantData
Functionname: StorageTank
Inputs:
TankCapacity,
HydrogenFlow,
Output:
CurrentStateoffulfillment,
NewStateoffulfillment,
Volumeavailableforstorage,
Hydrogenvolumeavailableforrefueling,
ConstantData
Description
Abbreviation
Value
Units
MinimumStateoffulfillment
10
MaximumStateoffulfillment
100
Ineverytimestep,thenewstateoffulfillment,
43
,isgivenbyeq.22.
(22)
Thereafter,theavailabilityofthestoragetankforfurtherstorage,
_
,andrefueling,
,isestimatedbyeq.23,24respectively.
(23)
(24)
Where,
fulfillment.TheassumptionsmadefortheselimitsareprovidedinTable6.
4.2.2. ECONOMICS
Model also includes a rough estimation of the system total capital and maintenance
costs. A subroutine is developed that takes into account the size of each system
component and applying empirical relationships for their investment cost results to
thetotalsystemcapitalcost(Table7).Regardingthesystemoperationcosts,asmall
percentage of every components capital cost is defined. However, batteries and
electrolyser unit lifetime is limited. The replacement of them is either linked to
operation features (electrolyser) or scheduled over a specific time period (batteries).
Replacementcostsareincludedintheoperationcosts.Then,theequationsappliedfor
each subsystems cost estimation are given. Table 7 provides also the constant data
usedduringsubroutinesimplementation.
Table7EconomicssubroutineSummary&ConstantData
Functionname: Economics
Inputs:
PVinstalledpower,
Batteriescapacity,
Electrolyserpower,
HydrogenStoragetankcapacity,
44
Output:
SystemCapitalcost,
SystemMaintenancecost,
ConstantData
Description
Abbreviation
Value
Units
Years
PVlifetime
20
Years
StorageTanklifetime
20
Years
20
years
3000
Cycles
4000
/kWp
Batterieslifetime
Systemlifetime
ElectrolysermaximumONOFFswitchingcycles
PVspecificcapitalcost
_
_
Batteriesspecificcapitalcost
70
/kWh
Batteriesregularmaintenancecost(%ofcapitalcost)
Electrolysermaintenancecost(%ofcapitalcost)
PVmaintenancecost(%ofcapitalcost)
0.5
Storagetankmaintenancecost(%ofcapitalcost)
4.2.2.1.
PHOTOVOLTAICARRAY
PVcostsshowsignificantdecreaseduringthelasttwodecades,of20%reductionwith
everydoubleofproduction[24,32].Currently,atypicalspecificcost,
powerplantisabout4.7/Wp[32].Inthisstudy,thePVcapitalcost,
,foraPV
isassumed
tobegivenbyeq.25(so4/Wp).
(25)
, is taken as a
,ofitstotalcapitalcost(eq.26).
(26)
replacement costs are not taken into account setting the system lifetime, , at 20
years.
45
4.2.2.2.
BATTERIES
, is
about$100/kWh[30].ConvertingtoBritishpounds,totalcapitalcostforthebatteries
subsystem,
,isestimatedbyeq.27.
(27)
,hasinfluence
onthemaximumcyclespermittedandconsequentlyitslifetime[24,27],inthepresent
study a constant lifetime,
also followed by other design study [33]. The cost for replacement is considered to
split into equal installments through the total system lifetime, . Hence, taking into
account the cost of 3 replacements through the 20 years of system lifetime and the
regular maintenance costs,
costforbatteriesisgivenbyeq.28.
(28)
4.2.2.3.
ELECTROLYSER
Electrolyser capital cost is often assumed to vary linearly with the installed power,
havingaspecificcapitalcostof$1000/kW[34].However,aresearchonthemarketof
Europe showed a correlation between electrolyser specific capital cost and its
production capacity (Figure 12) [35]. The majority of the electrolysers in the sample
taken were alkaline due to immature status at the time of study conduction.
Nevertheless,itisassumedthatPEMelectrolysersspecificcostshowssimilarvariance
regardingtheproductionrate,
,whichisdefinedineq.29inthisstudy.
46
(29)
Figure12Electrolysercomponentcostsobtainedfrom10majorsuppliersinEurope[35]
Hence,anempiricalequationwasexportedthatestimatestheelectrolysercapitalcost,
,basedonthemaximumproductionrate,
(eq.30).
64,547
(30)
Literaturereviewshowedthattheamountofcycles(onoffswitching)thatelectrolyser
undergoes has negative impact on the electrolyser performance. As noted in HARI
project [36], manufacturers often recommend a maximum of cycles after which
performancedegrades.Inthisstudy,takingintoaccountthecyclesperformedthrough
asimulationof1year(8760h),
,andsettinganupperlimit,
, is estimated
(eq.31).
(31)
, is
, (2% of
(32)
4.2.2.4.
HYDROGENSTORAGETANK
Asmentionedinthedescriptionofelectrolysermodeling,itisassumedthathydrogen
isbeingproducedathighpressureingasform.Asaresult,conventionalsteeltanksfor
pressurizedhydrogengasshouldbeusedasmeansofstorage.Asurveyinthemarket
of compressed hydrogen storage tanks resulted to Figure 13 [35]. It was shown that
specific storage price correlates well with the storage capacity. The equation for the
fittingcurveofthedatacollectedwasexportedandusedfortheestimationofstorage
tankcapitalcost,
Figure13Price/capacityrelationsforcommerciallyavailablecompressedstorageoptionsfor
hydrogeninthemediumtolargescalerange[35]
,byeq.33.
222.25
48
(33)
While,themaintenancecosts,
,arealsodefinedasapercentage,
,of
the capital cost (eq.34). A percentage of 0.5% was assumed for the modeling of a
similar system [35]. Finally, replacement costs are ignored as storage tanks lifetime
exceedsthestudiedsystemlifetime.
(34)
4.2.2.5.
SYSTEM
Concluding the subroutine for system economics, the total system capital cost,
,isestimatedasthesumofthesubsystemscapitalcostandaddinga5%for
otherunpredictablecosts(eq.35).
1.05
(35)
Similarly,systemmaintenancecost,
,isgivenbyeq.36.
1.05
(36)
4.3. POWERMANAGEMENTSTRATEGY
49
POWER MANAGEMENT
STRATEGY
FLOW CHART
START
NO
YES
ELECTROLYSER
REQUIREMENTS
EBAT_TH(i)=EPV(i)-t*PEL_MIN
ESUR(i)=0
NO
PPV(i)>PEL_MIN
YES
CALL BATTERY
FUNCTION
(DISCHARGE)
EBAT_TH(i)=EPV(i)
NO
EBAT_NET(i)<0
YES
PAVAIL(i)=min(PEL_MAX, (VAVAIL(i)*CEL)/t, PPV(i))
PAVAIL(i)=PPV(i)+PBAT_NET(i)
ESUR(i)=EPV(i)-PAVAIL(i)*t
CALL BATTERY
FUNCTION
(CHARGE)
CALL ELECTROLYSER
FUNCTION
QTANK(i)=QH2(i)
NO
END
ESUR(i)>0
YES
EBAT_TH(i)=ESUR(i)
Figure14Powermanagementstrategyflowchart
50
operation.Then,theinterconnectionofsubsystemsmodels(functions)underdifferent
operationcasesispresented.Powermanagementstrategydevelopedforthespecific
standalonesystemisreflectedinFigure14.
InputsrequiredforstrategyfunctionimplementationareprovidedinTable8.Most
ofthemarealsoinputsforfunctionsthatarecalledduringstrategyimplementation.In
addition, timestep status of batteries and storage tank is needed to check whether
therequirementsforelectrolyseroperationaresatisfied.Duetothefactthatstrategy
function just defines the way of operation followed in every timestep, its output
coincides with the output of functions called. Equations 6 and 18 that presented in
othersubroutinesarealsoappliedhere.
Table8StrategysubroutineSummary&ConstantData
Functionname: Strategy
Inputs:
PVenergyproduction,
PVpowerproduction,
Volumeavailableforstorage,
BatteryStateofcharge,
Electrolyserpower,
Batterycapacity,
Tankcapacity,
Timestep,
BatteryDepthofDischarge,
Tankstateoffulfillment,
Minimumpermittedelectrolyserhoursofoperationperday,
Output:
Storagetank,Batteries,Electrolyserfunctionsoutput
ConstantData
Description
Abbreviation
SpecificEnergyConsumption
Value
Units
kWh/Nm3
Minimumoperationpoint
95
Maximumoperationpoint
100
85
100
Dischargingenergyefficiency
Maximumallowablestateofcharge
51
4.3.1. ELECTROLYSEROPERATIONCONSTRAINTS
Ithasnotbeenpoweredpreviouslyinthespecificdayand,
0.6
37
,evenifthePVenergyproductionatthefollowingtimestepsis
zero(eq.38).
52
(38)
,itsvalueshouldbesettakingintoaccountthatanelectrolyser
o Potentialhydrogenproductionresultedbyelectrolyseroperationatits
maxima,
availablevolume,
,ensuringitsstorage(eq.39).
(39)
operation hours has not been covered. In such a case, it should continue to
operate independentlyto the amount of PV energy production. Batteries and
storagetankavailabilityhasbeenguaranteedbytheinitial,atthespecificday,
electrolyserpowering.
o PV power production,
electrolyserpower,
0.1
40
dayincreasesthesystemefficiencyasPVenergyproductionisdirectly
exploitedandnotthroughbatteries.
(41)
It has not performed a cycle during the current day. It was decided that
electrolyserwouldnotundertakemorethanonecycleperday,sotoeliminate
the frequent onoff switching which shortens its lifetime and results to
performancedegradation[7].
Toconclude,electrolyserisswitchedononlyduringtheday,atatimewhenPVenergy
issufficientandtheavailabilityofbatteriesandstoragetankguaranteesaminimumof
operationhours.Then,itstopstooperatewhenthePVenergyproductiondropsclose
tozero(atnight).Tonotice,electrolyseralsogetsswitchedoffincaseofinsufficient
stored energy in batteries (mentioned at the following section). In addition, tank
availability is checked in every timestep in order total hydrogen production to be
stored. Otherwise, electrolyser is set out of order. Finally, only one onoff switching
cycleperdayisaccepted.
4.3.2. ENERGYPRODUCTIONMANAGEMENT
54
Two main cases are described regarding whether the electrolyser constraints are
covered.ThedescriptionwhichtakesplacebelowisreflectedinFigure14.
Incasethatelectrolyserisnotallowedtooperateduetothefactthatsomeof
the constraints described in the previous section are not satisfied, all the PV
energyproduction,
,isprovidedforbatteriescharging(eq.42).
(42)
Hence, function used for batteries modeling is called having as input eq.42.
Then, depending on the battery state of charge and total capacity, available
energy is either completely or partial stored, or even totally rejected (see
batteriessubroutine).
On the other hand, if all the requirements are covered, electrolyser has the
potentialtooperate.However,theavailabilityofenergyproductionresultsthe
casesbelow.
o If,
,electrolyserhasthepotentialtobepoweredbyPV
, is
,and
(eq.43).
(43)
.Asitis
verifiedbyelectrolyserconstraints(eq.41)thatevenmaximumpossible
hydrogen production in the specific timestep can be stored in tank,
55
(44)
batteryandusedincaseofenergydeficit.Theamountofenergysurplus
isgivenbyeq.45.
(45)
Thereafter,batteryfunctioniscalledhavingtheinputdefinedbyeq.46.
(46)
o If,
electrolyserasthepowerproducedisnotsufficient,eventooperateat
minimum point. As the hydrogen production is systems first priority,
the energy deficit resulted should be covered by batteries and power
theelectrolyser.Batteriesfunctioniscalledhavingasinputtheamount
ofenergydemanded,
,whichisestimatedbyeq.47.
(47)
Nevertheless,atthispointtwocases,regardingthebatteriescapability
tocoverthedemand,exist.
56
If,
(48)
Asaresult,electrolyseroperateshavingtheminimumhydrogen
production rate. It is chosen batteries to provide the minimum
requiredamountofenergysoeithertostartortomaintainthe
electrolyser operation. Moreover, batteries discharging is
devotedforsatisfyingtheelectrolyserconstraintsandnotspent
formaximizingtheproductionrate.However,tonoticethatthe
electrolyseroperationlimitsatthisstudyarealmostthesameso
analysisabovehasnotagreatimpacthere.
If,
0,batteriesstoredenergyisinsufficienttocover
4.4. RESULTSOUTPUT
At the final stage of the numerical code, a group of magnitudes are calculated and
usedforsystemevaluation.Suchperformanceindicatorsarealsogoingtoconstitute
the objectives for system optimization. The definition of these evaluators is given by
eq.4954.
DuetointermittentnatureofPVandothersystemcomponentsavailability,itisalmost
impossible to exploit the total energy production. Hence, the percentage of energy
rejection,
,definedastheratioofannualenergyrejectionbybatteriestoannual
,isalwaystrue.Thefraction,
,shouldbetheminimumpossiblemeaning
thatsystemhastheappropriatesizingintermsofperformance.
(49)
Inaddition,capacityfactorsareoftenusedfordeterminingsystemcomponentsusage
rate. In this study, capacity factorsfor PV,
, and electrolyser,
, are given by
eq.50,51.
PV capacity factor,
exploited(eitherstoredinbatteriesorconsumedbyelectrolyser)toenergyproduced
if PV worked at peak power for the whole year (eq.50). The resulted value of this
evaluator is indicative of locations level of annual irradiance and PV size adequacy.
DuetoPVnature,valuesof25%andlowerareexpected.
_
,
58
8760(50)
, is equal to the ratio of annual
energyconsumptiontothepotentialconsumptionincaseitwasworkingcontinuously
forayearatnominalpower(eq.51).Amaximizedelectrolysercapacityfactorshould
beanobjectiveduringsystemssizing.
8760(51)
Inaddition,theefficiencyofsolartohydrogenconversion,
,isalsoused.Itis
.Inthis
(52)
Furthermore, the reliability of the system, , is also estimated. Regarding the tank
availability,dailyfueldemand,
,iscoveredeithertotallyorpartially.Hence,system
reliabilityisdefinedastheratiooftheannualsatisfiedfueldemandtotheannualfuel
demand (eq.53). The effect of system demanded reliability on system sizing is
investigated.Itissupposedthatincreasedreliabilityresultstooversizedsystems.
(53)
Where
,isthedailyfueldemandcoveredbystoredhydrogenintank.
attentiongivenduringthestrategydevelopment,electrolysercycleseliminationwillbe
anissuefortheoptimizationstudies.
,isgivenbyeq.55.
(54)
(55)
Where,r,thediscountrateassumedtobe10%forthisstudy.
Finally,thedefinitionofautonomythatbatteriesandstoragetankprovidetakesplace
as both magnitudes are used either for system evaluation or parameters
determination.Firstly,systemenergyautonomy,
activebatterycapacity,
,iscalculatedastheratioofthe
,toelectrolyserinstalledpower(eq.56).System
energyautonomydeterminesthetimethatelectrolysercanbeexclusivelypoweredby
batteries.
(56)
,todailyfueldemand,
(57)
Hydrogen autonomy reflects the max time that tank can guarantee cars refueling if
systemhasbeensetoutoforder.
60
5. RESULTS
Theapplicationofthemodeldescribedtakesplaceatthischapter.Atfirst,acasestudy
forareasonablesizedsystemispresented.Then,anoptimumsystemofminimumcost
andhighreliabilityisresultedusingtheMOT.Inaddition,theeffectthatsomeofthe
systemparametershaveonthesizingofminimumcostsystemsisinvestigatedthrough
various parametric studies. Finally, double objective optimization studies are
conductedapproachingtherelationshipbetweenevaluationindicatorsand/orsystem
parameters.
5.1. CASESTUDY
Thesimulationofasystem,thesizingofwhichisreliedonreasonableassumptions,is
presented here. System sizing and values of other parameters chosen for this case
studyareprovidedinTable9.
Table9CaseStudySystemparameters
No
Description
Abbreviation
Value
Units
PVinstalledpeakpower
48
kW
BatteryCapacity
240
kWh
Electrolyserinstalledpower
24
kW
HydrogenAutonomy
Days
Minimumcontinuouselectrolyseroperation
Hours
Numberofcarsrefueledbythestation
Cars
Location
Athens
Datatimeseries
Hence,itwasdecidedsuchasystem,installedinAthens,tobeabletorefuel2cars,
providing 2 kg of hydrogen gas per day(eq. 6). Assuming an average daily operation
timeforelectrolyserof5hours(minimumoperationat3hours),nominalproduction
rateshouldbeabout4Nm3/handconsequentlyitsnominalpowerat24kW(eq.21).
Concerningtheenergystoragesystem,batteriesaresizedat240kWhoftotalcapacity,
61
Thesimulationofspecificsizedsystemisresultedtofig.1521.Systemperformanceis
reflected in graphs for two indicative weeks of the year. Three different graphs are
givenforeachoftheseweeks.Atfirst,thewaythatPVpowerproductionisdistributed
tothesystemisshowninfig.15,18.Powerproducediseitherconsumedbybatteries
(red)andelectrolyser(blue)orrejected(yellow).Ontheotherhand,fig.16,19reflect
electrolyseroperationwhichiseitherpoweredbyPV(green)orbattery(orange).Also,
thevariationofthestateofchargenotonlyforbatteries(blue)butalsoforstorage
tank(red)ispresentedbyfig.17,20.
50
40
Battery
Electrolyser
Rejection
30
20
10
0
8256
8280
8304
8328
8352
8376
Hour of the year (h)
8400
8424
Figure15ConsumptionsourcesforPVpowerproduction(winterweek)
Electrolyser powering
Power (kW)
25
20
Battery
PV
15
10
5
0
8256
8280
8304
8328
8352
Hour of the year (h)
8376
8400
8424
Figure16Electrolyserpoweringsources(winterweek)
62
0.6
0.4
0.2
8256
8280
8304
8328
8352
8376
Hour of the year (h)
8400
8424
Figure17Batteriesandstoragetankstatusvariation(winterweek)
Regardingthewinterweek,itisshowninFigure15thatPVpowerproductionisalmost
completely exploited either powering the electrolyser or charging the battery.
However,atthebeginningoftheweek,energyproductionisnotsufficienttopower
electrolyser(reachtheinitialrequiredlimitsetbyeq.37)andisrejectedasbatteryis
fullycharged(Figure17).Inaddition,atfirsthoursofeachday,PVenergyproductionis
driven to batteries and electrolyser is turned on at the solar peak of the day (Figure
15).Thereafter,electrolyseroperationismaintainedduringthetimethatPVisactive
and consequently more than the time limit (3 hours). Nevertheless, batteries
assistance to electrolyser powering is significant as shown in Figure 16. As a result,
batteries state of charge reaches the minimum limit at the mid of the week due to
batteries contribution to the system and then, increases exploiting the higher PV
production(Figure17).Finally,hydrogenstoragetankstateoffulfilmentvariesinlow
levels(Figure17),asthesolarconditionsdonotprovidethepotentialtooperatethe
electrolyserlongerindailybasis.
50
40
Battery
Electrolyser
Rejection
30
20
10
0
4320
4344
4368
4392
4416
4440
Hour of the year (h)
4464
4488
Figure18ConsumptionsourcesforPVpowerproduction(summerweek)
63
Electrolyser powering
Power (kW)
25
20
Battery
PV
15
10
5
0
4320
4344
4368
4392
4416
Hour of the year (h)
4440
4464
4488
Figure19Electrolyserpoweringsources(summerweek)
0.6
0.4
0.2
4320
4344
4368
4392
4416
4440
Hour of the year (h)
4464
4488
Figure20Batteriesandstoragetankstatusvariation(summerweek)
Ontheotherhand,PVenergyproductionismostlyresultedgreaterthanelectrolyser
demandduringasummerweek(Figure18).Asaconsequence,electrolyserisdirectly
powered by PV (Figure 19) filling the hydrogen storage tank completely (Figure 20).
However,anenergysurplus,thatcannotbestoredinbatterieswhicharefullycharged
(Figure20),isrejected.Also,electrolyserswitchesoffafteracertaintimeduetothe
factpotentialhydrogenproductioncannotbestoredinthefulltank,showninFigure
20. From Figure 18, electrolyser shut down occurs before the PV energy production
stopresultingtoenergyrejection.
Furthermore, batteries and hydrogen storage tank status over a simulation year is
showninFigure21.Batteriesstateofchargevaries,evenreachingitsminimumlimit,
during winter (02500h, 65008760h), although it remains almost constant at
maximum level during summer period (30006500h). Storage tank undergoes similar
64
variation; however, state of fulfillment is resulted high for longer time period (1000
7000h).
0.6
0.4
0.2
1460
2920
4380
5840
Hour of the year (h)
7300
8760
Figure21Batteriesandstoragetankstatusvariation(annual)
Finally, results for evaluation indicators are given in Table 10. System simulated
showedhighreliability(99.7%)butpoorconversionefficiency(3.43%).Also,capacity
factor for PV and electrolyser was resulted as 12 and 23%, respectively. System
performanceisalsoreflectedatthefractionofenergyrejectedwhichwasestimated
44.5%. It is obvious, that low performance results had impact to specific production
costwhichfoundatabout6.2/Nm3.Inaddition,electrolyseroperatedalmostindaily
basis through the year, performing 353 cycles (only one cycle per day is permitted
accordingtooperationstrategy).
Table10CaseStudyResults
No
Description
Reliability
Specifichydrogenproductioncost
Energyrejection
PVcapacityfactor
Electrolysercapacityfactor
Systemefficiency
Electrolysercycles
Abbreviation
Units
99.7
6.176
/Nm3
44.5
11.9
22.9
3.43
353
Cycles
65
Value
5.2. SINGLEOBJECTIVEOPTIMIZATION
UsinggeneticalgorithmsofMOT,aseriesofoptimizationstudieswasconducted.Here,
theresultsofasingleobjectiveoptimizationarepresented.
The minimization of hydrogen specific production cost was set as the objective.
However, a constraint about system reliability was also included. Hence, system
resulted should be featured by reliability of 99.5% and over. Table 11 provides the
rangeofvaluesforvaryingsystemparameters.Valuesofotherparameterswhichare
assumedasconstantinthisstudyarealsogiven.Thehalfofsystemparameters,such
as location and number of cars refueled remained constant in contrast to the case
studyinorderacomparisonbetweenthemtobefeasible.
Table11SingleobjectiveoptimizationSpecificproductioncostParametersvaluesrange
Parameters
No
Description
Abbreviation
Value
Units
PVinstalledpeakpower
1070
kW
BatteryCapacity
100500
kWh
Electrolyserinstalledpower
1050
kW
HydrogenAutonomy
115
Days
Minimumcontinuouselectrolyseroperation
Hours
Numberofcarsrefueledbythestation
Cars
Location
Athens
>99.5
Datatimeseries
Constraint
1
Reliability
Objective convergence was not attained in a single run. Different limits were
attempted so to achieve the best objective value. Beginning with wide values range
and continuing with more and more decreased range, actual optimum system was
achieved.TheconvergenceforoneoftheseattemptsisshowninFigure22,whilethe
optimumsystemresultedispresentedinTable12.
66
SpecificProductionCost(/Nm3 )
20
40
60
80
100
Generations
Figure22Convergenceinsingleobjectiveoptimization
power.
Table12SingleObjectiveoptimizationSpecificproductioncostOptimumSystemparameters
No
Description
Abbreviation
Value
Units
PVinstalledpeakpower
39.9
kW
BatteryCapacity
210.3
kWh
Electrolyserinstalledpower
19.3
kW
HydrogenAutonomy
10.4
Days
Similarly to case study (Table 10), summary results indicative for the cost optimum
system are given in Table 13. As expected, evaluators are improved, although the
difference is not significant. Hydrogen produced costs about 5.46 /Nm3, a relatively
highvalue.Moreover,33%ofenergyproducedbyPVisrejectedresultingto4%solar
hydrogenconversionefficiency.Ontheotherhand,electrolysercapacityfactorcame
upat28%,whichisacceptableasitisforcedtooperateonlyduringtheday.
67
Table13SingleObjectiveoptimizationSpecificproductioncostOptimumSystemresults
No
Description
Abbreviation
Reliability
Specifichydrogenproductioncost
Energyrejection
PVcapacityfactor
Electrolysercapacityfactor
Systemefficiency
Electrolysercycles
Value
Units
99.5
5.462
/Nm3
33
14.3
28.2
4.1
355
Cycles
At this point, the contribution of each system component to total capital and
maintenance has special interest. As shown in, PV capital cost constitutes about the
half of total capital cost, while electrolyser comes as the second more expensive
component with 35% contribution. On the other hand, batteries cost is resulted less
significant,beingthe5%ofthetotalcost.
Figure23Capital(left)/Maintenance(right)costshareintheoptimumsystem
seems
not
to
have
important
impact
68
on
the
system
costs.
5.3. PARAMETRICSTUDIES
Systemsizing,basedoncostcriteria,isaffectedbyconstraintsandconstantparameter
valuesthathavesetbeforetheoptimization.Hence,itwasdecidedtoinvestigatethe
influence of some parameters on optimal economic systems. Optimization studies
havingasobjectivetheminimizationofthespecificproductioncostbutunderdifferent
conditions,comparingtoprevious(4.2.)section,arepresentedhere.Beforepresenting
the results, Table 14 defines the set of parameters used in following parameter
studies.ChangesinTable14occuronlyafterspecialreferenceineverystudy.
Table14ParametricStudiesParametersandconstraintdefinition
Parameters
No
Description
Abbreviation
Value
Units
PVinstalledpeakpower
Variable
kW
BatteryCapacity
Variable
kWh
Electrolyserinstalledpower
Variable
kW
HydrogenAutonomy
Variable
Days
Minimumcontinuouselectrolyseroperation
Hours
Numberofcarsrefueledbythestation
Cars
Location
Athens
>99.5
Datatimeseries
Constraint
1
Reliability
5.3.1. LOCATION
At the first two studies, weather data time series from Athens were used. However,
solar conditions in Greece are very advantageous for PV installations due to high
annual solar radiation density (about 6.3 GJ/m2). In this study, solar data time series
from Tokyo (4.7 GJ/m2) and Frankfurt (4.2 GJ/m2) are also applied and changes
performedinsystemsizingarenoticed.
69
In order to depict the effect of installation area on system sizing, each location is
featuredbyitsannualsolarenergydensityinGJ/m2.Itwasfoundthatcomponentssize
is largely affected which, however, could not be correlated with the annual solar
energydensityofthearea.Tobemoreprecise,PVelectrolysissystemestablishedin
Tokyo(4.7GJ/m2)orinFrankfurt(4.2GJ/m2)requiresabouttwotimeslargerbattery
capacity (500 kWh) than in Athens (see 5.2). On the other hand, PV and electrolysis
size of a system in Frankfurt resulted 2.5 times bigger than in Athens and Tokyo. As
expected, the specific production cost for optimum systems proved to increase in
areasoflowerannualsolarenergydensity(Figure24).
SpecificProductionCost(/Nm3 )
8
6
4
2
0
3
3.5
4.5
5.5
6.5
AnnualSolarEnergyDensity(GJ/m2 )
Figure24Productioncostvariationregardingtheannualsolarenergydensity
Itmustbepointedoutthattherateofproductioncostreduction,increasesrapidlyfor
areasof4.5GJ/m2andlowerannualsolarenergydensity.Hence,whileTokyosystem
produces hydrogen with 6.7 /Nm3, the optimum system for Frankfurt has a much
higher production cost of 11.3 /Nm3. Such a sudden increase in cost goes with the
demand for more installed PV panels and larger electrolyser unit which mentioned
before.
70
ElectrolyserCycles
y=49.781x1.0705
3.5
4.5
5.5
6.5
AnnualSolarEnergyDensity(GJ/m2 )
Figure25Electrolysercyclesvsannualsolarenergydensityinoptimaleconomicsystems
Theamountofcyclesthatelectrolyserperformsinayearisfoundtobeindicativeof
theareasolarconditionsandthesystemcomponentssize.ItisshowninFigure25that
annualelectrolysercyclesarelinearlyrelatedwiththeannualsolarenergydensity.To
give an example, electrolyser turned on 225 and 268 times in Frankfurt and Tokyo
respectively,incontrasttoAthensthatperformed355cycles.Itisobviousthatwhere
solarpotentialisweaklikeFrankfurtandTokyo,electrolyserunitturnsuplargerand
consequentlyisforcedtooperatefewertimestocoverthefueldemand.
5.3.2. RELIABILTY
Then,aparametricstudyregardingthesystemreliabilityisconducted.Discretevalues
(80,90,95and99.5%)forsystemreliabilityweresetasconstraintsfortheoptimum
economicsystemsexport.
Although the less hydrogen production required for systems of lower reliability,
specifichydrogenproductioncostisfoundtodecreasewithfueldemandsatisfaction
rate. As shown in Figure 26, cost for 80% reliable system reduced about 20 %
comparingwiththesystemof99.5%reliability.Itcanbesaidthathydrogenproduction
costcorrelateswithreliability.
71
SpecificProductionCost(/Nm3 )
y=0.1714x0.7464
3
2
1
0
75
80
85
90
95
100
Systemreliability(%)
Figure26Effectofsystemreliabilityonspecificproductioncostofoptimumeconomic
system
From a general point of view, optimizations resulted that the lower reliability that a
systemshouldhave,thesmallercomponentssizeisrequired.Hence,itwasprovedthe
ratioofPVtoelectrolyserinstalledpowerisrelatedwellwithsystemreliability(Figure
27). More specifically, PV peak power should be more than two times greater than
electrolyser installed power in order to achieve high reliability (99.5%). The almost
same relationship exists also for values of reliability greater than 90%. Then, ratio
dropsto1.75for80%reliablesystem.
PV/Electrolyserinstalledpower
y=0.9291e0.0081x
2
1.75
1.5
1.25
1
75
80
85
90
95
100
Systemreliability(%)
Figure27RatioofPVtoelectrolyserinstalledpowerrequiredforvariouslevelsofreliability
72
In addition, it was pointed at the single objective optimization section that energy
rejectionisquitesignificantforthehighreliablesystemwhichcameup.Aninteresting
resultofthisstudyisthatenergyrejectionincreaseswithreliability.Alinearequation
depicts well the relationship between energy rejection and reliability (Figure 28).
Hence,every5%ofreliability,about57%isaddedtoenergyrejection.Also,itcanbe
noticedthatenergyrejectedintheoptimalsystemof80%reliabilityisonly4%.Itmust
bepointedoutthatthisdropinenergyrejectionisnotowedtosizeincreaseofsystem
storage devices, as both batteries and storage tank are resulted much smaller (25%
and80%maxreductionrespectively)inlowerlevelsofreliability.
Percentage(%)
30
25
EnergyRejection
20
PVcapacityfactor
15
Electrolysercapacityfactor
10
Systemefficiency
5
0
80
85
90
95
100
Systemreliability(%)
Figure28Performanceindicatorsofoptimumeconomicsystemswithdifferentreliability
73
5.3.3. NUMBEROFCARSREFUELED
Until now, optimization studies were conducted for a specific daily fuel demand (2
cars).Here,theimpactofthefueldemandonsystemsizingisinvestigatedbysetting
differentnumberofcars(2,10,50)toberefueledbythestation.
PV/Electrolyserinstalledpower
y=2.1099x0.049
1.8
1.75
1.7
0
10
20
30
40
50
60
Numberofcarsrefueledbystation
Figure29PV/electrolyserinstalledpowerinoptimaleconomicsystemsfordifferentfuel
demand(carsrefueled)
Regarding the objective of the study, it is shown by Figure30 that system size has a
positiveimpactontheoptimalspecifichydrogenproductioncost.Tonotice,thatrate
of cost decrease is resulted higher in the low range (210 cars) of number of cars
refueled. As a consequence, a station that undertakes the refuelling of 10 cars,
74
produces hydrogen at 4.2 /Nm3, having 23% reduction comparing toa station for 2
cars.Ontheotherhand,hydrogenproductioncostfora50carsstationdecreasesto
3.7/Nm3whichmeansthatthedecreaserateintherangeof1050carsislimitedto
12%.
SpecificProductionCost(/Nm3 )
y=5.7883x0.118
4
3
2
1
0
0
10
20
30
40
50
60
Numberofcarsrefueledbystation
Figure30Optimalproductioncostvariationwiththenumberofcarsrefuelledbystation
5.3.4. HYDROGENAUTONOMY
The volume of hydrogen storage tank guarantees a time period of autonomy, when
carsrefuelingcantakeplaceindependentlytosolarconditions.Surely,thebiggerthe
storagetankis,thelessenergyrejectionoccurs.However,volumeoccupiedbystorage
tank is an issue for such stations. Sometimes, storage tank is predefined due to
conditions of installation area. At this section, days of autonomy, that storage tank
provides,aresetasdiscretevalues(3.5,5,7.5,10,12.5,15days)foroptimaleconomic
systems.Theeffectofdaysofautonomyonsystemcharacteristicsisstudied.
Asageneraltrend,itwasprovedthatPVand electrolysersizearenotinfluencedby
thechangeinstoragetankvolume.Hence,PVpeakpowerismaintainedastwice(see
5.2) of the electrolyser installed power for every tank volume. On the other hand,
batteries,astheothermeansofstorageinthesystem,shouldhaveincreasedcapacity
incaseofsmallstoragetank(daysofautonomy).However,asshowninFigure31,
75
thehoursofautonomythatbatteriesprovidetoelectrolyserreachalowlimitfor7.5
daysanduprefuelingautonomy.Itcanbeconcludedthatelectrolysershouldbeable
tobeexclusivelypoweredbybatteriesatleastfor6.5hours.
EnergyAutonomy(hours)
10
12
14
16
HydrogenAutonomy(days)
Figure31Hoursofautonomyrequiredbybatteriesvsdaysofautonomyprovidedbystorage
tankinoptimaleconomicsystems
Theresultsdescribedabovearealsoreflectedatthevariationofhydrogenproduction
costwithhydrogenautonomy(Figure32).Firstly,itisprovedthatanoptimaleconomic
systemshouldincludeatankthatprovidesatleast3.5daysofautonomy.Otherwise,
the required reliability (99.5%) cannot be achieved. Certainly, optimization was also
approachedfor2.5daysofhydrogenautonomy,butdidnthaveanyresults.Then,an
interestingfinding,depictedbyFigure32,isthatalimitexistsforhydrogenautonomy,
afterwhich,thecostreductionisinsufficient.Tobemoreprecise,asystemof3.5days
hydrogen autonomy produces hydrogen at about 6.2 /Nm3, while for 7.5 days the
correspondingcostisat5.55 /Nm3,resultingto10.5%costdecrease.But,afurther
increase in days of hydrogen autonomy drives to inadequate system economics
improvement. It can be noticed that a system which compromises minimum
production cost with minimum hydrogen autonomy is found by the tangent to the
slope of the fitting curve shown in Figure 32. In such a case, this compromise suits
better at the point of 5.65 /Nm3 hydrogen production and 6 days hydrogen
autonomy.
76
SpecificProductionCost(/Nm3 )
10
12
14
16
HydrogenAutonomy(days)
Figure32Optimalproductioncostfordifferentdaysofautonomy(storagetank)required
5.3.5. CAPITALCOSTREDUCTION
Asresultedincasestudy(see5.1),systemcapitalcostismainlyowedtophotovoltaics
and electrolyser unit. Even though that, currently, capital costs of both technologies
arehigh,theprospectsareverypromising.PVcostshavebeendecliningsteadilyover
thelasttwodecades,havinga20%costreductioneverydoublingofproduction[32].
Also, PEM electrolysers seem to dominate the market in the future, due to their
operationcharacteristicsandreducedcost,takingadvantageoftherapidevolutionin
PEMfuelcellsarea[1].Asaconsequence,anoptimizationstudyhavingasparameter
thepotentialcapitalcostreductionforPVandelectrolyserwasconsideredaspointed.
Furthermore, whether capital cost reduction effect is influenced by the fuel demand
(carsrefueled)isinvestigated.
Threedifferentscenariosofpossiblecostreduction(25,50,75%)areimplemented.To
notice that cost reduction applied was the same for both components in every
scenario. What it was shown, is that specific hydrogen production cost decreases
linearly with potential capital cost reduction (Figure 33). Nevertheless, rate of
77
improvement in production cost is greater for small systems (2 vehicles). In the best
scenario of 75 % cost reduction, the hydrogen production cost varies between 1.37
1.88/Nm3forrefueling50and2carsrespectively.
SpeificPRoductionCost(/Nm3 )
10
50
1
0
0
15
30
45
60
75
CapitalCostReduction(%)
Figure33Optimalhydrogenproductioncostfordifferentfueldemand(carsrefuelled)under
threescenariosofPVandelectrolysercapitalcostreduction.
System sizing under the different scenarios applied did not perform significant
changes. The slight increase of PV and electrolyser installed power with capital cost
reductioninoptimalsystemsisnoticeable.Ontheotherhand,meansofstoragetend
to reduce their size. As a result, system performance magnitudes, such energy
rejection, capacity factors and solar to hydrogen efficiency, decline with the capital
costreduction(Figure34).
Percentage(%)
EnergyRejection
PVCapacityFactor
ElectrolyserCapacityFactor
Systemefficiency
20
40
60
80
CapitalCostReduction(%)
Figure34Performanceindicatorsforoptimaleconomicsystemsabletorefuel2carsunder
differentcapitalcostreductionscenarios
78
5.4. DOUBLEOBJECTIVEOPTIMIZATION
At the previous studies, genetic algorithms of MOT were applied for determining
systemparameterswhichresulttheoptimalvalueofoneobjective.However,system
should be featured by more than one optimal objective. System optimization having
two competitive objectives is approached here. Multiobjective optimization using
geneticalgorithmsofMOTwasappliedfortheimplementationofthestudy.Tonotice
that,eithersystemparametersand/orperformanceindicatorsareusedasobjectives.
Then, three pairs of objectives are presented. It must be pointed out that the
minimization of production cost does not constitute an issue from now on, and also
hydrogenstoragetanksizeisassumedtobeconstant(10daysofhydrogenautonomy).
Summary of system parameters for the following studies is given in Table 15. Any
specificchangewillbementionedineachstudy,separately.
Table15DoubleobjectiveoptimizationsParametersandconstraintdefinition
Parameters
No
Description
Abbreviation
Value
Units
PVinstalledpeakpower
Variable
kW
BatteryCapacity
Variable
kWh
Electrolyserinstalledpower
Variable
kW
HydrogenAutonomy
10
Days
Minimumcontinuouselectrolyseroperation
Hours
Numberofcarsrefueledbythestation
Cars
Location
Athens
>99.5
Datatimeseries
Constraint
1
Reliability
79
Anassumptionthatmodeldevelopedtookintoaccountistheminimumofhoursthat
electrolyserrequiredtooperateincaseofswitchingon.Suchaconstraintwassetdue
to the fact that electrolyser cannot achieve instantly its nominal operating
temperature.Moreover,thefrequentonoffswitchingduringadayisavoided.Hence,
battery,asenergystoragedevice,shouldensureelectrolyserautonomousoperationat
least for the minimum required time. As a result, batteries capacity is expected to
increaseasminimumoperationhoursaregettingmore.However,theminimizationof
batteries capacity will contribute to system economics while the maximization of
minimumhoursrequiredforelectrolyseroperationwouldincreaseitscapacityfactor
andeliminatetheonoffswitching(cycles).
Doubleobjectiveoptimization
Minelectrolyseroperation/Energyautonomy
Energyautonomy(h)
12
10
8
y=1.4389x0.9524
6
4
2
0
3
Minimumelectrolyseroperation(h)
Figure35Energyautonomyvsminimumelectrolyseroperationfor99.5%reliablesystem
autonomy more than the required electrolyser minimum operation time. To give an
example,incasethatelectrolysershouldoperateatleast4hoursafteraswitchingon,
battery must be sized so that system has energy autonomy of 5.3 hours. To repeat
that,Figure35includesonlysystemsofhighreliability(over99.5%).
5.4.2. ELECTROLYSERCYCLESENERGYAYTONOMY
Itisalreadymentionedthattheeliminationofelectrolysercyclesisanobjectiveofthe
present project. Besides the special consideration in power management strategy,
according to which onoff switching is limited to 1 per day, further annual reduction
wouldbepossible.Hence,asalsoacontinuityofthepreviousstudy,theminimization
ofelectrolysercyclesisexaminedasacompetitiveobjectivetominimizationofsystem
energyautonomy.However,electrolyserminimumoperationissetnowasconstantat
3hours,whileotherparametersaresimilarlydefined.
DoubleObjectiveOptimization Electrolyser
Cycles/EnergyAutonomy
14
EnergyAutonomy(h)
12
10
8
6
4
2
0
250
270
290
310
330
350
370
ElectrolyserCycles
Figure36EnergyautonomyvsElectrolysercyclesfora99.5%reliablesystem
Therelationshipbetweenenergyautonomyandelectrolysercycles,whichcomeupby
optimizationstudy,isdepictedbyFigure36.Aninterestingfindingisthatelectrolyser
cycles decrease, increasing the energy autonomy provided by batteries. To be more
81
precise,260cyclesperyearcanbeachievedifsystemenergyautonomyisatleast12
h.Lineardecreaseofenergyautonomyismaintaineduntilthepointwhere290cycles
can be obtained with about 7 hours of autonomy. However, the next system that
turned up, satisfying the constraint (more than 99.5% reliability), was featured by 4
hours of energy autonomy and 355 electrolyser cycles. It should be pointed out that
such energy autonomy is the minimum possible. To remind that, in an optimal
economic system (see 5.2), electrolyser performs 355 cycles but battery provides
about6.5hoursofautonomy.
5.4.3. PVELECTROLYSERCAPACITYFACTORS
Capacity factors of PV and electrolyser are indicative of their performance and the
appropriatesizingofthem.Therelationshipbetweenthemhasbeenconsideredasan
issue in a study [8]. Here, it is proved that PV and electrolyser capacity factors are
competitivemagnitudeseachother.Thepresentsectionapproachesthesimultaneous
maximizationofbothcapacityfactors.
Electrolysercapacityfactor(%)
DoubleObjectiveOptimizationPV/Electrolyser
capacityfactor
35
34
33
32
31
30
29
28
8
10
11
12
13
14
15
16
PVcapacityfactor(%)
Figure37RelationshipbetweenoptimalPVandelectrolysercapacityfactors
82
CapacityfactorsthatfeaturetheoptimalsystemsexportedareplottedinFigure37.It
canbeobservedthattheyareinverselyproportional,asthehighestcapacityfactorfor
PV corresponds to the lowest for electrolyser and vice versa. Capacity factor of PV
ranges at 9.515% while the one of electrolyser from 29% to 34%. An intermediate
compromisecouldbefoundbythetangenttothefittingcurve,resultingtoasystemof
12.2% PV and 32% electrolyser capacity factor. To notice that such values are
concerned with at least 99.5% reliable systems. However, that the parametric study
(see5.3.2)regardingthereliabilityofthesystemshowedthatbettercapacityfactors
can be obtained at lower levels of reliability. Finally, it must be noticed that optimal
economicsystemforthesameparameterswasfeaturedby14.3%and28.2%PVand
electrolysercapacityfactor,respectively(see5.2).
83
6. DISCUSSION
6.1. OPERATION
Electrolyser operation features were depicted in case study simulation (fig. 1521).
However,theirpotentialadverseimpactonsystemefficiencyisanissue.Thedemand
ofminimumcontinuousoperationandthelimitationtooneonoffswitchingperday,
results to longer electrolyser operation and so requires lower hydrogen production
capacity (Nm3/h). At this point, adequate battery capacity is necessary to satisfy the
constraint.Furthermore,the,atleast,60%coverofelectrolysernominalpowerbyPV
issetasrequirementforhydrogenproductionstartup,sotomatchthePVproduction
withelectrolyseroperationandextendedbatterydischargingtobeavoided.Suchan
assumption makes anamount of PV energy production either to be used for battery
chargingortoberejection.ThecaseofbatterychargingcanbeobservedinFigure15,
where battery is charged before solar peak is sufficiently reached and turn the
electrolyseron.Nevertheless,duringsummernotonlysolarpotentialisstrongerthan
the system demand (Figure 18) but also battery is fully charged (Figure 21) and so
increasedenergyrejectionoccurs.Hence,itcanbeendedupthatalthoughPVbecome
the major electrolyser supplier, a sufficient amount of energy is rejected during
summer. The variation of solar radiation through the year in conjunction with the
electrolyserconstraintsmayresulttosystemoversizing.Inaddition,electrolyserisset
to operate very close to its design point. As a consequence, batteries should assist
electrolyser powering in case of low PV energy production (Figure 16). But, battery
undergoesdeepdischargingduringextendedperiodsoflowsolarradiation(Figure17).
Finally,electrolyserisdecidednottooperateduringthenightsobatterynaturetobe
maintained as power leveling and not as exclusive powering source. Regarding this
84
feature,casestudysimulationshowedaweaknessoftheapproach.Aperiodofalmost
zerohydrogenvolumestoredintankwaspointedduringwinter(Figure17),afactthat
does not ensure the total satisfaction of refueling demands. Observing Figure 17,
batterystateofchargehasreacheditsmaximumatthesameperiod.Hence,itwould
be preferred to allow the exclusive electrolyser powering by batteries, in case of
emergency(lowtankstateoffulfillment).
Overall,systemsimulationshowedthatoperationconstraintsforareliableelectrolyser
lifecycle have been applied into the power management strategy. However, most of
theelectrolyserrequirementsseemnottomatchwiththebestenergyexploitation.In
this study, the limitation of battery assistance in electrolyser powering was also
approached.Probably,furtherstrategydevelopmentordifferentconceptsareneeded
to compromise electrolyser steady operation and energy rejection elimination,
althoughtheircompetitivenature.
6.2. SIZING
System sizing was shown that is influenced by installation area (see 5.3.1), reliability
demanded(see5.3.2)andthenumberofcarsrefueled(see5.3.3).Ontheotherhand,
parametric study regarding the hydrogen autonomy (see 5.3.4) investigated the
relationship between system means of storage. However, some general trends for
systemsizingcanbehighlightedduetothefactthatmostofstudiesconductedforthe
same parameters base. To notice that, analysis below, deals only with optimal
economicsystems.
OptimalsystemresultedfortheareaofAthensand2carstoberefueledwasfoundto
requirePVpeakpoweralmostthetwiceoftheelectrolyserinstalledpower(Table12).
It can be ended up that each car refueled corresponds to 1.61 Nm3/h (19.39 kW/6
kWh/Nm3 *2 cars) of electrolyser capacity. Optimal design suggests such a small
electrolyser unit (about 7 hours of daily electrolyser operation are required) as the
capital cost for such small units is too high (Figure 12). Such a proportion is mainly
drivenduetothefactthatPVshouldbeabletopowerelectrolyserduringperiodsof
85
lowradiation.Toremindthat,the,atleast,60%poweringbyPVisarequirementfor
electrolyserstartup.However,duringsummerperiod,PVreachesitspeakandshould
be exploited either charging the battery or powering the electrolyser. Regarding
batterycapacity,aminimumofacceptablevaluesisdefinedbearinginmindthatmax
discharge rate (
Needlesstosaythatenergyautonomyisdirectlyrelatedwiththeminimumcontinuous
electrolyseroperationrequired.AsdepictedinFigure35,batteriessizeshouldprovide
at least 1.32 additional hours of autonomy to the minimum electrolyser operation
(2+3 h). It has to be reminded that a requirement for electrolyser startup is the
minimum continuous operation which must be guaranteed by batteries state of
charge.However,suchaminimumdoesnotconstitutetheoptimaleconomicchoice.
To give an example, 6.5 hours of energy autonomy were resulted for 3 hours of
minimumhoursofelectrolyseroperation.
Atthispoint,therelationshipbetweensystemenergyandhydrogenautonomyshould
bealsopointed.Furthermore,asalreadymentionedtheminimumofenergyautonomy
(battery capacity) is defined by the electrolyser power. Such a minimum is reached
whenhydrogenautonomybecomeshigherthan7.5days(Figure31).Anyadditionto
storage tank capacity then has insufficient impact on system sizing. To notice, that
optimal economical system (Table 12) resulted with 10 days of hydrogen autonomy
due to the slightly less specific production cost (Figure 32). However, if volume and
area issues occur during system installation, a minimized storage tank is suitable. A
compromisebetweenspecificproductioncostandstoragetankcapacityispointedat
about6daysofhydrogenautonomy(Figure32).
86
Then,thealmostproportionalrelationshipbetweenthenumberofcars(fueldemand)
and the optimum systems sizing was presented (see 5.3.3). Such a conclusion is
obvious as the fuel demand has been assumed as a daily constant value. However,
somesmallchangeshavebeennoticedinsizingofbiggersystems.Morespecifically,as
showninFigure29,ratioofPVpeakpowertoelectrolyserinstalledpowerdropsfrom
2.1(2cars)to1.75(50cars).Thisdifferenceoccursduetonotlinearrelationbetween
electrolyserspecificcapitalcostandcapacity(Figure12).Thedecreasedspecificcapital
costinhighercapacityrangesmakestheinstallationofabiggerunitacceptable.Onthe
otherhand,meansofstoragearesizedsimilarlytothesingleoptimizationreference
(see5.2).
Regarding the system location, optimal sized systems were resulted for the cities of
TokyoandFrankfurt.Eventhoughtheirsimilarannualsolarenergyreceived(4.7and
4.1GJ/m2respectively),significantdifferencesoccurinsystemsizing.Thedoublingof
requiredbatterycapacityinTokyosystem,incontrasttothatofAthens(Table12),is
owedtothelimitedpeaksolarhoursduringsomedaysofthecorrespondingsolardata
time series. On the other hand, all the system components of optimal system for
Frankfurtturnedupto2.5timesgreaterthanthesamesysteminAthens.Thereason
for such a result is supposed to be an extended period of very low radiation. It is
obviousthatthedemandofhighsystemreliabilityrapidlyincreasescomponentssizein
caseofweaksolarpotential.
87
Asageneraltrendfortheoptimalsystemofhighreliability(99.5%)andestablishedin
areas of high solar potential (Athens), electrolyser should have a specific capacity of
1.61.85Nm3/h(250carstation)percarrefueled.Then,PVsizingisestimatedat2
1.75 times (250 car station) of electrolyser installed power. Regarding the means of
storage, batteries max discharge rate should be equal to electrolyser power. Finally,
storagetankhastoprovideaminimumofabout6daysofhydrogenautonomy(incase
thatvolumelimitationexists).
6.3. PERFORMANCE
Inthisstudy,systemperformanceisevaluatedthroughtheestimationofcomponents
capacityfactors,energyrejectionandconversionefficiency.Also,theamountofcycles
thatelectrolyserperformsisalsotakenintoaccount.Mostofthecommentsstatedfor
systemoperationandsizingarereflectedintheseevaluationmagnitudes.
CapacityfactorsofPVandelectrolyserarethepictureoftheirusagefrequency.Itwas
shownthattheyareadverselyrelatedeachother(Figure37).Thiscanbereasonedas
excess of PV production makes the electrolyser to operate longer but results also to
increased energy rejection. From Figure 37, maximum electrolyser capacity factor
found to be about 34% (for 9.5% PV capacity factor), which is acceptable keeping in
mindtheconstraintssetforitsoperation(see4.3.1).Inaddition,optimumPVcapacity
factorislimitedto15%.Electrolyseroperationconstraintsandsystemnature(stand
alone)arethemaincausesforsuchlowPVcapacityfactor.However,itseemsaclose
to maximum possible PV capacityfactor (14%) to be preferred for optimal economic
systems (Table 13), while electrolyser usage rate is at 28%. Coincidence of PV and
electrolyseroperationandsystemsizingareobviouslythereasonsforsucharesult.In
anycase,electrolysercapacityfactorshouldbehigherthanthecorrespondingofPVin
optimaleconomicsystems.Suchafactagreeswithrelativesizingadvicespresentedin
otherstudy[8].
Also,energyrejectionandsystemefficiencyareindicativeofsizingadequacy.Asnoted
inTable13,optimumsizingresultsto33%energyrejection.Demandforelectrolyser
88
steady operation during the whole year and the onesided exploitation of PV energy
production (electrolyser powering) cause such a high amount of rejection. However,
the increased PV peak power has also impact on system efficiency (eq. 52). As
hydrogen production is constant for determined system reliability, the only way to
improvesystemefficiencyistoeliminatePVeffectivearea(installedpower).Certainly,
PV peak power required is influenced by electrolyser operation mode. In this study,
steady operation was approached resulting to a maximum efficiency of 4.1% (Table
13). Comparing to similar studies [12,13,14] which ended up at about 10% solar to
hydrogen efficiency, such a system efficiency is considered as low. However, this
difference was expected due to variable electrolyser operation followed in these
simulations[12,13,14].
Itshouldbepointedoutthatasreliabilityaffectssystemsizing,impactalsooccursin
performanceevaluators(Figure28).Aninterestingresultisthealmostfullexploitation
of PV energy production (4% energy rejection) in optimal sized system of 80%
reliability. Also, for the same demand satisfaction, system efficiency achieves a
percentage of 5.75%. Considering that maximum efficiency can be achieved when
energy rejection is minimized, such a result is thought to be the greatest possible
undertheoperationconstraintsapplied.Asaconsequence,poweringsourcesstudied
(PV,batteries)cancontributetoelectrolyserpoweringwithoutsystemoversizinguntil
80% reliability. Then, the application of other concepts as the integration of diesel
generatororgridassistanceisrecommendedforcompletefueldemandsatisfaction.
dailyelectrolyseroperation(355cycles).Nevertheless,lesselectrolysercyclescanbe
obtained in reliable systems of greater energy autonomy (Figure 36). In case of high
energy autonomy, electrolyser has lower installed power and is forced to operate
longeronceitisswitchedon.Fromacostview,itwasfoundthatelectrolysershould
operateindailybasisandbatterycapacitytobeeliminated.
6.4. HYDROGENPRODUCTIONCOST
Minimizationofspecifichydrogenproductioncostwastheobjectiveofthemajorityof
the optimizations presented. With no doubt, economic results exported by
optimizationstudiesaretoohigh.Thiscanberealizedcomparingwithcostsofother
simulation studies, different hydrogen production systems and competitive fuels.
However,otherissuesshouldbenoticedpriortoresultscriticism.
Consequenceofchangesperformedinsystemsizingfordifferentparametervaluesis
thedifferentiationinhydrogenproductioncost.Hence,parametersaslocation(Figure
24),numberofcarsrefueled(Figure30)andreliabilitydemanded(Figure26)modify
theresultedproductioncost.Reasoningfortheeffectofthemhasalreadytakenplace
duringthediscussionofsystemsizingandperformance.Theresultedproductioncost
of99.5%reliablesystemsinstalledinAthensisdiscussedbelow.
Tobemoreprecise,optimalhydrogenproductioncostforthepresentsystemranges
at3.735.48/Nm3forlarge(50cars)andsmall(2cars)stationsrespectively(Figure
30). At this point, it should be highlighted that operating conditions set for system
simulationarenotfavorable(steadyelectrolyseroperation,limitedonoffswitching).
90
ThedifferentscenariosofpotentialcostreductionofPVandelectrolyserwereapplied
for the determination of possible production in the future. It was shown that
differenceinproductioncostbetweendifferentsizedsystemsbecomeslessascapital
cost decreases (Figure 33). This fact is owed to the normalization of electrolyser
specificcapitalcostinhighnominalproductionrates(Figure12).Asaresult,production
cost decreases to 1.361.88 /Nm3 (50 2 cars station) for an assumed capital cost
reductionof75%(Figure33).
Certainly, even under such a great reduction, system production cost is far greater
than grid connected electrolysis systems (710 $/kg or equally 0.4 0.6 /Nm3) [38].
Thishappensduetohighenergyrejectionandcomponentscapitalcostswhichsetthe
electricity production more expensive than other sources (grid). Comparing the
consumption of a hydrogen and a gasoline fueled car, it can be concluded that
hydrogenshouldbeproducedwithlessthan0.3/Nm3inordertobecompetitive[38].
Itisunderstoodthatimprovementinefficiencyofhydrogenproductionviaelectrolysis
isneeded.Inthisstudy,anuppervalueforelectrolyserenergyconsumption(6kWh/
Nm3) was assumed and had also impact on system sizing and production cost
estimation.
On the other hand, simulation study for PVelectrolysis systems has also resulted to
low costs (1.06 /Nm3) [14]. Major differences of the present project to more
economicallyattractiveapproaches[12,13,14]istheapplicationofelectrolysersteady
operationandthebatteriesincorporationasshorttermstoragedevice.Furthermore,
systemoperationwassimulatedas100%standalone.
91
To sum up, assumptions made and power control followed by this study was not
favorableforlowproductioncostresults.Itwasprovedthat,evenwithagreatcapital
costreduction,suchaninvestmentisnotattractive.However,differentapproachesof
thesystemsuchasinstallationsubsidies,alternativeloadaddition,gridassistance,use
of diesel generator as a reserve, exploitation of oxygen produced, can improve the
hydrogenproductioncost.
92
7. CONCLUSIONS
The energy simulation and optimal design of a standalone PVelectrolysis system
incorporating electrolyser limitations was achieved with success. Simulation graphs
showedthatrespectofelectrolyserconstraints,bestenergyexploitationanddemand
satisfactioncannotbecompletelycompromisedduetotheircompetitivenature.
Then, optimization studies resulted to general trends for a reliable (99.5%) system
sizing installed in favorable areas (Athens). As a guide, electrolyser installed power
should be defined by the number of cars to be refueled, having a proportion of 1.6
1.85 Nm3/h (502 cars) per car. In addition, due to electrolyser operation
requirements, PV peak power should be almost 1.72 (502 cars) times greater than
electrolysernominalpower.Concerningthebatteriescapacityislargelyaffectedbythe
storage tank capacity. In case of no tank volume limitations, battery should have at
least a maximum charge rate able for the exclusive electrolyser powering. Results
showed that 6.5 hours of energy autonomy is required, while minimum hydrogen
autonomyof6dayscanbeappliedwithoutasignificantcostincrease.
Concerning system performance, although results were not competitive, they are
explained by system nature (standalone) and the power management strategy
followed.PVandelectrolysercapacityfactorsareestimatedatabout14%and28%for
optimal economic systems. On the other hand, demand for high system reliability
drove to high energy rejection (33%) and relatively low solar to hydrogen energy
efficiency (4.1%). Also, the as long as possible daily operation of electrolyser in
conjunctionwiththeeliminationofitsinstalledpower(highcapitalcostatpresent)is
recommended.Moreover,itshouldbenotedthatPVelectrolysissystemissuitablefor
80% demand satisfaction, as optimal systems exported for this reliability level are
featuredbyalmostzeroenergyrejection(4%).Asaconsequence,itcanbeconsidered
that a system under such operation constraints is suggested to be assisted by an
additional powering source (e.g. diesel generator) for the complete cover of fuel
demand.
93
Ontheotherhand,hydrogenproductioncostscalculatedfoundtovaryat3.735.48
/Nm3 regarding the amount of fuel demand (502 cars refueled). Gradual PV and
electrolysercapitalcostreductionwasappliedandshowedthatproductioncostcanbe
limitedto1.361.88/Nm3(502cars)forapotential75%decrease.Evenunderthese
conditions, system investment seems not to be competitive not only for gasoline
replacement but also to grid connected electrolyser units. However, it is true that
operation strategy (electrolyser constraints), system nature (standalone) and
assumptionsmade(lowestpossiblecomponentsefficiency,overestimatedcapitaland
maintenance costs) were the reasons for such high resulted production cost. Finally,
installation subsidies, exploitation of oxygen production, are considered as potential
solutionstoimprovesystemeconomics.
94
APPENDIX
A. PVFUNCTION
function[PV_en_prod,PV_p_prod]=PV(Temperature,Radiation,Area,Time)
%Functionwhichcalculatesphotovoltaicsenergyproductionintimestep
%Inputs:Temperature(C),
%Solarradiationdatatimeseries(kWh/m2),
%Modulesarea(m2)
%Output:Photovoltaicenergyproduction(kWh)
Ta=Temperature;%Ambienttemperature
G=Radiation;%Globalradiation
A=Area;%Modulesarea
t=Time;%Computingtimestep
NOCT=48;%Nominaloperationcelltemperature
Tc=Ta+(NOCT20)*(G/0.8);%Celltemperature
nr=0.14;%Solarcellefficiencyatreferencedconditions
B=0.005;%SolarcellTemperaturecoefficient
Tr=25;%ReferencedTemperature
ne=nr*(1B*(TcTr));%Solarcellefficiencytemperatureeffectincluded
nc=0.95;%MPPTefficiency
wl=0.05;%Wiringlosses
de=0.05;%Degradationeffect
deul=1;%Degradationeffectupperradiationlimit
dell=0.2;%Degradationeffectlowerradiationlimit
%Degradationefficiency
ifG>=deul
nd=1;
elseifG<=dell
nd=1de;
else
nd=(G/deul)*de+(1de);
end
ndc=0.95;%DC/DCconverter/chargeregulatorefficiency
ifG>0%Ifnegativeradiancevalues(night)zeroproductionassumed
Epv=A*G*ne*nd*nc*(1wl)*ndc;%PVenergyproduction
Ppv=Epv/t;%PVpowerproduction
else
Epv=0;
Ppv=0;
end
%Outputstatement
PV_en_prod=Epv;%PVenergyproduction
PV_p_prod=Ppv;%PVpowerproduction
end
95
B. BATTERIESFUNCTION
%Functionthatsimulatesbatteryoperationinenergyterms
%Inputs:DepthofDischargeDoD(%),
%Energyforcharging(+)/discharging()(kWh),
%BatteryCapacity(kWh)
%Time(h)
%CurrentStateofCharge(%)
%Output:NewStateofChargeSOC(%)
%Energytobeprovided(discharge)()/PVEnergyexploited(charge)(+)(kWh)
%Powertobeprovided(discharge)()/PVpowerexploited(charge)(+)(kW)
%Energycharged(+)/disharged()(kWh)
%Powercharged(+)/discharged()(kW)
%Batteryenergylosses(always+nomatterischarge/discharge)(kWh)
%Batterypowerlosses(always+nomatterischarge/discharge)(kW)
%Energyrejected(batterylossesnotincluded)(kWh)
%Powerrejected(batterylossesnotincluded)(kW)
function
[State_of_charge,En_bat_net,P_bat_net,En_bat_in,P_bat_in,En_bat_los,P_bat_los,En_bat_rej,P_bat_rej
]=Battery(Depth_of_Discharge,B_Capacity,Time,Energy_for_dis_charging,Current_SOC);
DoD=Depth_of_Discharge;%DepthofDischarge(%)
t=Time;%ComputingTimestep(h)
C=B_Capacity;%BatteryCapacity(kWh)
Ebatn=Energy_for_dis_charging;%EnergyavailablefromPVforbatterycharging(+)orenergydemand
forelectrolyserpowering()(kWh)
SOC=Current_SOC;%CurrentStateofCharge(%)
SOCmax=1;%MaximumallowedStateofcharge(%)
SOCmin=SOCmaxDoD;%MinimumpermittedStateofCharge(%)
Rmax=C/10;%Maximumcharge/dischargerate(kW)
Sdr=0;%0.02*t/(30*24);%Selfdischargerate(14%montlhy)(kW)AppliedinSystem.mfile
nch=0.85;%Chargingefficiency(%)
ndis=0.85;%Dischargingefficiency(%)
Dis_cap=min(((SOCSdr)SOCmin)*C,Rmax*t);%DischargeCapability(kWh)
Ch_cap=min((SOCmax(SOCSdr))*C,Rmax*t);%ChargeCapability(kWh)
Epch=Ebatn*nch;%Potentialcharge(kwh)
Epdis=Ebatn/ndis;%PotentialDischarge(kWh)
ifEbatn<0%Dischargemode
ifabs(Epdis)>Dis_cap%Batterynotcapabletocoverthedemand
Ebat=0;%Batterydischarge
else%Batterycapabletocoverthedemand
Ebat=Epdis;%Batterydischarge
end
Ebatl=Ebat*ndis;%Energytobeprovidednet()
Erej=0;
Prej=0;
else%Chargemode
96
Ebat=min(Epch,Ch_cap);%Batterycharged
Ebatl=Ebat/nch;%Energyavailableexploitedincludingbatterylosses(+)
Erej=EbatnEbatl;%Energyrejected(batterylossesnotincluded)
Prej=Erej/t;%Powerrejected(batterylossesnotincluded)
end
SOC=(SOCSdr)+Ebat/C;%StateofChargeupdate
Pbat=Ebat/t;%Powercharge/discharge
Pbatl=Ebatl/t;%Powertobeprovided(discharge)()/PVpowerexploited(charge)(+)(kW)
Eblos=abs(EbatlEbat);%Batteryenergylosses(always+nomatterischarge/discharge)(kWh)
Pblos=abs(PbatlPbat);%Batterypowerlosses(always+nomatterischarge/discharge)(kW)
%Outputstatement
State_of_charge=SOC;%NewStateofChargeSOC(%)
En_bat_net=Ebatl;%Energytobeprovided(discharge)()/PVEnergyexploited(charge)(+)(kWh)
P_bat_net=Pbatl;%Powertobeprovided(discharge)()/PVpowerexploited(charge)(+)(kW)
En_bat_in=Ebat;%Energycharged(+)/disharged()(kWh)
P_bat_in=Pbat;%Powercharged(+)/discharged()(kW)
En_bat_los=Eblos;%Batteryenergylosses(always+nomatterischarge/discharge)(kWh)
P_bat_los=Pblos;%Batterypowerlosses(always+nomatterischarge/discharge)(kW)
En_bat_rej=Erej;%Energyrejected(batterylossesnotincluded)(kWh)
P_bat_rej=Prej;%Powerrejected(batterylossesnotincluded)(kW)
end
97
C. ELECTROLYSERFUNCTION
function
[Hydrogen_Production,Power_consumption,Energy_consumption,Power_rejection,Energy_rejection]=El
ectrolyzer(Power_available,Nominal_production_rate,Time,ELmax,ELmin,Eecon)
%Functionthatsimulateselectrolyseroperation
%Input:Power_available(kW),
%Productionrate(Nm3/h),
%Timestep(h)
%MinMaxoperationpoints
%Energyconsumption(kWh/Nm3)
%Output:Hydrogenproduction(Nm3),
%PowerConsumed(kW),
%Energyconsumed(kWh),
%Powerrejection(kW),
%Energyrejection(kWh)
Pav=Power_available;%Availablepowerforelectrolyser(kW)
Prrate=Nominal_production_rate;%NominalProductionrate(Nm3/h)
t=Time;%Computingtimestep(h)
Pelmax=ELmax*Eecon*Prrate;%Nominalelectrolyserpower(kW)
Pelmin=ELmin*Eecon*Prrate;%Minimumelectrolyserpoweracceptable(kW)
ifPavPelmin<0.001%Electrolyseroutofordernotsufficientpower
Pcons=0;%Totalpowerexploited
elseifPavPelmax>0.001%Poweravailablegreaterthanthenominalpower
Pcons=Pelmax;%Electrolyseroperatesatmaxima
else%Powerlevelacceptablebyelectrolyseroperationlimits
Pcons=Pav;
end
%Specificconsumptionestimation
EL_op_point=Pcons/Pelmax;%Electrolysercurrentoperationpoint
%NondimensionedrelationshipobtainedbyUllebergetal
El_sp_cons_nd=2.1478*(EL_op_point)^3+4.2745*(EL_op_point)^22.7977*(EL_op_point)+1.6403;
El_sp_cons=El_sp_cons_nd*Eecon;%Electrolyserspecificconsumption(kWh/Nm3)
Hydpr=Pcons*t/El_sp_cons;%Hydrogenproduction
Econs=Pcons*t;%Energyconsumption
Prej=PavPcons;%Powerrejection
Erej=Prej*t;%Powerrejection
%Outputstatement
Hydrogen_Production=Hydpr;%Hydrogenproduction(Nm3)
Power_consumption=Pcons;%PowerConsumed(kW)
Energy_consumption=Econs;%Energyconsumed(kWh)
Power_rejection=Prej;%Powerrejection(kW)
Energy_rejection=Erej;%Energyrejected(kWh)
end
98
D. STORAGETANKFUNCTION
function
[Volume_available,Stored_H2_available,New_state_of_fulfilment]=Tank(Capacity,Current_state_of_fulfi
lment,Hydrogen_Flow)
%Functionforhydrogenstoragetankfulfilment
%Input:Currentstateoffulfilment(%)
%Capacity(Nm3)
%HydrogenFlowstorage(+)/Demand()
%Output:Volumeavailableforstorage(Nm3)
%Hydrogenvolumeavailableforrefueling(Nm3)
%Newstateoffulfilment(%)
Vt=Capacity;%TankCapacity
SOF=Current_state_of_fulfilment;%CurrentHydrogenlevel
Qhyd=Hydrogen_Flow;%Outlet(),Inlet(+)
SOFmax=1;%Maximumleveloffulfilment
SOFmin=0.1;%Minimumleveloffulfilment
Vrem=(SOFmaxSOF)*Vt;%Volumeavailabilityforstorage
Vav=(SOFSOFmin)*Vt;%Hydrogenavailabilityforrefueling
SOF=SOF+Qhyd/Vt;%Newfulfilmentlevel
%Outputstatement
Volume_available=Vrem;%Volumeavailableforstorage(Nm3)
Stored_H2_available=Vav;%Hydrogenvolumeavailableforrefueling(Nm3)
New_state_of_fulfilment=SOF;%Newstateoffulfilment(%)
End
99
E. POWERMANAGEMENTSTRATEGYFUNCTION
function
[State_of_charge,En_bat_net,P_bat_net,En_bat_in,P_bat_in,En_bat_los,P_bat_los,En_bat_rej,P_bat_rej
,Hydrogen_Production,Power_consumption,Energy_consumption,Power_rejection,Energy_rejection,Vol
ume_available,Stored_H2_available,New_state_of_fulfilment,T_en_rej,T_p_rej]=Strategy(PV_en_prod,P
V_p_prod,Volume_available,Stored_H2_available,Time,Nominal_production_rate,Depth_of_Discharge,
B_Capacity,Capacity,Current_SOC,Current_state_of_fulfilment,Hour_of_the_day,EL_ONOFF,ELmax,ELmi
n,Eecon,Min_hours);
%Functionforpowermanagementstrategyforastandalonesolarhydrogen
%systemhavingasstoragedevice,batteries.
%Input:Photovoltaicpowerproduction(kw)
%Photovoltaicenergyproduction(kWh)
%Availablevolumeforhydrogenstorageintank(Nm3)
%VolumeofstoredHydrogenintank(Nm3)
%Electrolysernominalproductionrate(Nm3/h)
%BatteryDepthofDischarge(%)
%BatteryCapacity(kWh)
%StorageTankCapacity(Nm3)
%SimulationTimestep(h)
%BatteryStateofCharge(%)
%Currentstateoftankfulfilment(%)
%MinMaxoperationpoints
%ElectrolyserEnergyconsumption(kWh/Nm3)
%Output:NewStateofChargeSOC(%)(BATTERY)
%Energytobeprovided(discharge)()/PVEnergyexploited(charge)(+)(kWh)(BATTERY)
%Powertobeprovided(discharge)()/PVpowerexploited(charge)(+)(kW)(BATTERY)
%Energycharged(+)/disharged()(kWh)(BATTERY)
%Powercharged(+)/discharged()(kW)(BATTERY)
%Batteryenergylosses(always+nomatterischarge/discharge)(kWh)(BATTERY)
%Batterypowerlosses(always+nomatterischarge/discharge)(kW)(BATTERY)
%Energyrejected(batterylossesnotincluded)(kWh)(BATTERY)
%Powerrejected(batterylossesnotincluded)(kW)(BATTERY)
%Hydrogenproduction(Nm3)(ELECTROLYZER)
%PowerConsumedbyBattery(kW)(ELECTROLYZER)
%EnergyconsumedbyBattery(kWh)(ELECTROLYZER)
%PowerrejectionbyBattery(kW)(ELECTROLYZER)
%Energyrejection(kWh)(ELECTROLYZER)
%Volumeavailableforstorage(Nm3)(STORAGETANK)
%Hydrogenvolumeavailableforrefueling(Nm3)(STORAGETANK)
%Newstateoffulfilment(%)(STORAGETANK)
%Totalenergyrejection(kWh)(SYSTEM)
%Totalpowerrejection(kW)(SYSTEM)
Prrate=Nominal_production_rate;%NominalProductionrate(Nm3/h)
t=Time;%ComputingTimestep(h)
EL_en_cons=Eecon;%Electrolyserenergyconsumption(kWh/Nm3)
%ElectrolyseroperationData
EL_max_power=ELmax*EL_en_cons*Prrate;%Nominalelectrolyserpower(kW)
EL_min_power=ELmin*EL_en_cons*Prrate;%Minimumelectrolyserpoweracceptable(kW)
100
%Checkwhethertherequirementsforelectrolyseroperationaresatisfied.
%CaseA:Someoftherequirementsabovenotsatisfied.PVpowerproductionusedforbatterycharging
if (EL_ONOFF==0 & (((Volume_available)*EL_en_cons)<Min_hours*EL_max_power | (Current_SOC(1
|
Depth_of_Discharge))*B_Capacity<Min_hours*EL_max_power))
((Volume_available)*EL_en_cons)/Time<EL_max_power
|
(EL_ONOFF==0
&
PV_p_prod<0.6*EL_max_power)|EL_ONOFF==3|EL_ONOFF==4|(PV_p_prod<0.1*EL_max_power&
EL_ONOFF~=10)
Energy_for_dis_charging=PV_en_prod;%TotalPVenergyproductionusedforbatterycharging
%CallBatteryfunction
[State_of_charge,En_bat_net,P_bat_net,En_bat_in,P_bat_in,En_bat_los,P_bat_los,En_bat_rej,P_bat_rej
]=Battery(Depth_of_Discharge,B_Capacity,Time,Energy_for_dis_charging,Current_SOC);
Power_available=0;%Electrolysernotpowered
%CallElectrolyserfunction
[Hydrogen_Production,Power_consumption,Energy_consumption,Power_rejection,Energy_rejection]=El
ectrolyzer(Power_available,Nominal_production_rate,Time,ELmax,ELmin,Eecon);
Hydrogen_Flow=Hydrogen_Production;%Duetopreviouscheckalltheproductionstorageisensured
%CallTankfunction
[Volume_available,Stored_H2_available,New_state_of_fulfilment]=Tank(Capacity,Current_state_of_fulfi
lment,Hydrogen_Flow);
%CaseB:Alltherequirementsaresatisfied.Electrolysercanoperateifsufficientenergyexists.
else
%CaseB1:PVpowerproductionenoughforelectrolyzeroperationatminima
ifPV_p_prod>EL_min_power
In_rej=abs((Power_availablePV_p_prod)*t);%Energyrejectedenergysurplus
%CallElectrolyserfunction
101
[Hydrogen_Production,Power_consumption,Energy_consumption,Power_rejection,Energy_rejection]=El
ectrolyzer(Power_available,Nominal_production_rate,Time,ELmax,ELmin,Eecon);
%CallTankfunction
[Volume_available,Stored_H2_available,New_state_of_fulfilment]=Tank(Capacity,Current_state_of_fulfi
lment,Hydrogen_Flow);
Energy_rejection=Energy_rejection+In_rej;
%CaseB1i:ElectrolyserrejectsanamountofPVenergyproduction
ifEnergy_rejection>0
%CallBatteryfunction
[State_of_charge,En_bat_net,P_bat_net,En_bat_in,P_bat_in,En_bat_los,P_bat_los,En_bat_rej,P_bat_rej
]=Battery(Depth_of_Discharge,B_Capacity,Time,Energy_for_dis_charging,Current_SOC);
%CaseB1ii:ElectrolyserexploitsalltheamountofPVenergyproduction
else%Repeatedonlyforthezerooutput
Energy_for_dis_charging=Energy_rejection; % Energy rejected by electrolyser can be stored in
battery
%CallBatteryfunction
[State_of_charge,En_bat_net,P_bat_net,En_bat_in,P_bat_in,En_bat_los,P_bat_los,En_bat_rej,P_bat_rej
]=Battery(Depth_of_Discharge,B_Capacity,Time,Energy_for_dis_charging,Current_SOC);
end
%CaseB2:PVpowerproductionnotenoughtopowerelectrolyseratminima
else
Energy_for_dis_charging=(EL_min_powerPV_p_prod)*t;%Batteryiscalledtoprovidetheenergy
deficitsoelectrolysertooperateatminima
%CallBatteryfunction
[State_of_charge,En_bat_net,P_bat_net,En_bat_in,P_bat_in,En_bat_los,P_bat_los,En_bat_rej,P_bat_rej
]=Battery(Depth_of_Discharge,B_Capacity,Time,Energy_for_dis_charging,Current_SOC);
%CaseB2i:Batterycanprovidetheenergydeficit
ifabs(P_bat_net)>0
Power_available=PV_p_prod+abs(P_bat_net);%ThesumofPVandbatterypowerproductionis
providedtoelectrolyser
%CallElectrolyserfunction
[Hydrogen_Production,Power_consumption,Energy_consumption,Power_rejection,Energy_rejection]=El
ectrolyzer(Power_available,Nominal_production_rate,Time,ELmax,ELmin,Eecon);
102
%CallTankfunction
[Volume_available,Stored_H2_available,New_state_of_fulfilment]=Tank(Capacity,Current_state_of_fulfi
lment,Hydrogen_Flow);
%CaseB2ii:Batterycannotprovidetheenergydeficit,sotheelectrolysercannotbepowered
else
Power_available=0;%Electrolysernotpowered
Energy_for_dis_charging=PV_en_prod;%PVproductiontobestoredinbattery
%CallBatteryfunction
[State_of_charge,En_bat_net,P_bat_net,En_bat_in,P_bat_in,En_bat_los,P_bat_los,En_bat_rej,P_bat_rej
]=Battery(Depth_of_Discharge,B_Capacity,Time,Energy_for_dis_charging,Current_SOC);
%CallElectrolyserfunction
[Hydrogen_Production,Power_consumption,Energy_consumption,Power_rejection,Energy_rejection]=El
ectrolyzer(Power_available,Nominal_production_rate,Time,ELmax,ELmin,Eecon);
%CallTankfunction
[Volume_available,Stored_H2_available,New_state_of_fulfilment]=Tank(Capacity,Current_state_of_fulfi
lment,Hydrogen_Flow);
end
end
end
%Outputstatement
T_en_rej=En_bat_rej;%SystemEnergyrejection(kWh)
T_p_rej=P_bat_rej;%SystemPowerrejection(kW)
End
103
F. ECONOMICSFUNCTION
function
[
Hydrogen_prod_cost
]
=
Economics(Area,B_Capacity,Nominal_production_rate,Capacity,Total_H2_prod,Bat_cycles,Electrolyser_
cycles,Depth_of_Discharge,EL_Cost_reduction,PV_Cost_reduction);
%Functionthatestimatesthesystemcapital,operationalcostandthe
%hydrogenproductioncost
%INPUT:PVarea(m2)
%BatteryCapacity(kWh)
%Electrolysernominalproductionrate(Nm3/h)
%AnnualHydrogenproduction(Nm3)
%AnnualBatterycycles
%AnnualElectrolysercycles
%BatteryDepthofDisharge(%)
%OUTPUT:HydrogenProductionCost(Pounds/Nm3)
%Dataabouttheeconomicsofthesystem
n=20;%Systemlife(years)
r=0.1;%Discountrate
%PV
PV_power=Area*0.1;%empeiricalrelationship:1m2correspondsto100Wp
PV_cap_cost=4000*PV_power*(1PV_Cost_reduction);%PVcapitalcost
PV_OM_cost=0.02*PV_cap_cost;%PVO&Mcost
%BATTERIES
Bat_cap_cost=70*B_Capacity;%Batteriescapitalcost
Bat_max_cycles=500/Depth_of_Discharge;%MaximumnumberofcyclesforbatterydependingonDoD
Bat_Life_cycle=min((Bat_max_cycles/Bat_cycles)*0.75,5);%Batterylifetime(years)
Bat_replace=round(n/Bat_Life_cycle);%Batteryreplacementsthroughthesystemlifetime
ifBat_replace<1%Incasethatthebatterylifetimeprovedtobeoverthesystemlifetime,noneedfor
replacement
Bat_replace=1;
else
end
Bat_OM_cost=0.01*Bat_cap_cost+((Bat_replace1)*Bat_cap_cost)/n;%BatteriesO&Mcost
%ELECTROLYSER
EL_cap_cost=(1EL_Cost_reduction)*64547*((Nominal_production_rate)^(
0.562))*Nominal_production_rate;%Electrolysercapitalcost
EL_max_cycles=2500;%Maxelectrolyseronoffswitchingbeforedegradation(typicalvalue)
EL_life=(1.5*EL_max_cycles/Electrolyser_cycles); % Electrolyser lifetime accepting a period with
degradedefficiency(years)
EL_replace=floor(n/EL_life);%Electrolyserreplacements
EL_OM_cost=0.02*EL_cap_cost+((EL_replace1)*EL_cap_cost)/n;%ElectrolyserO&Mcost
%STORAGETANKS
104
Tank_cap_cost=222.25*(Capacity^(0.258))*Capacity;%Storagetankcapitalcost
Tank_OM_cost=0.005*Tank_cap_cost;%StoragetankO&Mcost
%OTHER
Other_cap_cost=0.05*(PV_cap_cost+Bat_cap_cost+EL_cap_cost+Tank_cap_cost);
Other_OM_cost=0.05*(PV_OM_cost+Bat_OM_cost+EL_OM_cost+Tank_OM_cost);
%SYSTEM
Capital_Cost=PV_cap_cost+Bat_cap_cost+EL_cap_cost+Tank_cap_cost+Other_cap_cost; % System
capitalcost
OM_Cost=PV_OM_cost+Bat_OM_cost+EL_OM_cost+Tank_OM_cost+Other_OM_cost; % System O&M
cost
%Capitalcostannualdepreciation
Da=Capital_Cost*r/(1(1+r)^(n));%Annualdepreciation
%Hydrogenproductioncost
Hydrogen_prod_cost=(Da+OM_Cost)/(Total_H2_prod);
End
G. FUELDEMANDFUNCTION
function[Fuel_Demand]=Demand(Number_of_Vehicles)
%Functionfordetermingthehydrogendemandof1dayforanumberofvehicles
%Input:NumberofVehicles
%Output:FuelDemand
%DemandStatement
Daily_Fuel_dem=1;%Hydrogendemandof1carfor1day(kg)
Daily_Fuel_demand=Daily_Fuel_dem*11.2;%TransformingtoNm3
Fuel_Demand=Number_of_Vehicles*Daily_Fuel_demand; % Total Daily demand for the number of
vehicles
end
105
H. SYSTEMFUNCTION
function[f]=System(x)
%FunctionCodefortheenergysimulationofastandalonepvelectrolyser
%systemhavingbatteriesasstoragedevice.
%Parameters:Location(Solarradiation(kWh/m2)+temperature(C))
%PVArea(m2)
%BatteryDepthofDischarge(%)
%BatteryCapacity(kWh)
%ElectrolyserNominalProductionRate(Nm3/h)
%Systemautonomy(days)
%Objectives:HydrogenProductionCost(Pounds/Nm3)
%Reliability(%)
%Overallsystemefficiency(%)
%PVcapacityfactor(%)
%ElectrolyserCapacityfactor(%)
%Energyrejection(%)
%Electrolysercycles
%%%%%%%%%%%%%%%%%%
F
U
N
C
T
I
O
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%Parametersstatement
PV_power=x(1);%PVpeakpower(kW)
Bat_Cap=x(2);%Batterycapacity(kWh)
EL_power=x(3);%Electrolysernominalpower(kW)
Days_of_autonomy=x(4);%Daysofautonomy(days)
Vehicles_No=10;%x(6);%Vehiclestoberefueledeveryday
EL_Cost_reduction=0.5;%Futureelectrolysercapitalcostreduction(%)
PV_Cost_reduction=0.5;%Futurephotovoltaicscapitalcostreduction(%)
%CallDemandfunction
[Fuel_Demand]=Demand(Vehicles_No);
Tank_cap=Days_of_autonomy*Fuel_Demand;%Tankcapacityspecifiedbydaysofautonomy
%Electrolyseroperationfeatures
ELmax=1;%Maxoperationpoint(%)
ELmin=0.95;%Minoperationpoint(%)
Eecon=6;%Electrolyserelectricityconsumption(kWh/Nm3)
Min_hours=3;%Electrolyserminimumhours(h)
Bat_DoD=0.6;%x(3);%BatteryDepthofDischarge(%)
%EstimationofPVareaandElectrolysernominalrate
PV_area=PV_power*10;%empeiricalrelationship:1m2correspondsto100WpPV(m2)
EL_Pr_rate=EL_power/Eecon;%Electrolysernominalpower(Nm3/h)
%Parametersrename
Area=PV_area;%PVmodulestotalarea(m2)
106
Depth_of_Discharge=Bat_DoD;%BatteryDepthofDischarge(%)
B_Capacity=Bat_Cap;%BatteryCapacity(kWh)
Nominal_production_rate=EL_Pr_rate;%ElectrolyserH2productionrate(Nm3/h)
Capacity=Tank_cap;%HydrogenStorageTankcapacity(Nm3)
Number_of_Vehicles=Vehicles_No;%NumberofVehicles
%HydrogenHigherHeatingvalue
HHV=3.52226;%(kWh/Nm3)
%Optimization?
opt=1;
%Openfiles
ifopt==0
fid=fopen('Results.txt','w');%Afilethatresultsareexported
fid2=fopen('Summary_table.txt','w');%Asummarytable
else
end
%Fileincludingsolarradiation,temperaturedatatimeseriesloaded
loadAthens.dat;
%Simulationtimestepdeterminedbytheinputfile
Time=length(Athens)/(365*24);
%Cumulativemagnitudesnullifiedbeforethebeginningoftheiterations
Annual_Dem=0;%AnnualHydrogenDemandatthebeginningoftheyear
Annual_Dem_ns=0;%AnnualHydrogennonstatisfieddemandatthebeginningoftheyear
Total_H2_prod=0;%AnnualHydrogenproduction
Total_PV_prod=0;%AnnualPVenergyproduction
Total_rejection=0;%Annualenergyrejection(batteryelectronicdeviceslossesexcluded)
Total_el_consumption=0;%Annualelectrolyserenergyconsumption
Total_radiation=0;%Totalsolarradiationtothepanels
Electrolyser_cycles=0;%AnnualElectrolysercycles
Bat_cycles=0;%AnnualBatterycycles
%Iterationsstartup
fori=1:length(Athens)%Procedurerepeatedequaltimeswiththerowsofinputfile
a=Athens(i,:);%Everyiterationreadsthedatafrom1x7array
Hour_of_the_year=a(1);%Inputdefinition
Hour_of_the_day=a(2);
Day_of_the_month=a(3);
Month_of_the_year=a(4);
Temperature=a(5);
Radiation_Horizontal=a(6)/1000;
Radiation=a(7)/1000;
%Settingthestateofcharge/fulfilmentofbattery/tankamongtheiterations
ifi==1%Initialstates
SOF(i)=0.1;
SOC(i)=1.0;
else%Thishourinitialstateequalstothelastprevioushourstate
SOF(i)=SOF(i1);
SOC(i)=SOC(i1)0.02*Time/(30*24);%Selfdischargerate(14%montlhy)(kW);
end
107
%Initializingthedailyelectrolysercycles
ifHour_of_the_day==1
EL_ON_OFF(i)=0;%Nopreviousonoffswitchduringthepresentday
else
EL_ON_OFF(i)=EL_ON_OFF(i1);
end
%CallPVfunctionforenergyproductionestimation
[PV_en_prod,PV_p_prod]=PV(Temperature,Radiation,Area,Time);
Current_state_of_fulfilment=SOF(i);%Inputfortankfunctionstatament
%CallTankfunctionforcheckingthestoredhydrogenavailability
Hydrogen_Flow=0;%Functioniscalledonlyforthestatusdetermination
[Volume_available,Stored_H2_available,New_state_of_fulfilment]=Tank(Capacity,Current_state_of_fulfi
lment,Hydrogen_Flow);
SOF(i)=New_state_of_fulfilment;
%CallDemandfunction
[Fuel_Demand]=Demand(Number_of_Vehicles);
%Assumption:Vehiclesrefuelingtakesplaceat9pm
ifHour_of_the_day==21
Hydrogen_Flow=min(Fuel_Demand,Stored_H2_available);
else
Fuel_Demand=0;
Hydrogen_Flow=0;
end
Non_cov_dem=Fuel_Demandabs(Hydrogen_Flow);%AmountofH2demandnotavailable
Current_state_of_fulfilment=SOF(i);%Inputfortankfunctionstatement
%CallTankfunction
[Volume_available,Stored_H2_available,New_state_of_fulfilment]=Tank(Capacity,Current_state_of_fulfi
lment,Hydrogen_Flow);
SOF(i)=New_state_of_fulfilment;%UpdateofTankfulfilimentaftertherefueling
Current_state_of_fulfilment=SOF(i);%Inputforstrategyfunction
Current_SOC=SOC(i);%Inputforstrategyfunction
EL_ONOFF=EL_ON_OFF(i);%Inputforstrategyfunction
%Callstrategyfunction
[State_of_charge,En_bat_net,P_bat_net,En_bat_in,P_bat_in,En_bat_los,P_bat_los,En_bat_rej,P_bat_rej
,Hydrogen_Production,Power_consumption,Energy_consumption,Power_rejection,Energy_rejection,Vol
ume_available,Stored_H2_available,New_state_of_fulfilment,T_en_rej,T_p_rej]=Strategy(PV_en_prod,P
V_p_prod,Volume_available,Stored_H2_available,Time,Nominal_production_rate,Depth_of_Discharge,
108
B_Capacity,Capacity,Current_SOC,Current_state_of_fulfilment,Hour_of_the_day,EL_ONOFF,ELmax,ELmi
n,Eecon,Min_hours);
SOF(i)=New_state_of_fulfilment;%Finalstateoftankfulfilmentforthepresenttimestep
SOC(i)=State_of_charge;%Finalbatterystateofchargeforthepresenttimestep
%CountingBatteryCycles
%Thedefinitionofacompletechargeanddischargeisfollowed
SOC_ref=1Depth_of_Discharge;%ThereferenceSOCisassumedtobetheminimumpermitted
Marg=0.42*Depth_of_Discharge;%Themarginusedinbatterycyclesestimation(%)
%Description:Initially,Flag_1=0&Flag_2=0.
%Thefirsttimethatreachestheupperlimit>>Flag_1=1
%Ifthenreachesthelowerlimit>>Flag_2=1
%Ifbothflagsare1andreachestheupperlimitfora2ndtimea
%cycleiscounted(Flag_1=2)
ifi==1%Settingtheinitialstatus
if(abs(SOC(i)1)<Marg)
Bat_high(i)=1;
Bat_low(i)=0;
elseif(abs(SOC(i)SOC_ref)<Marg)
Bat_high(i)=0;
Bat_low(i)=1;
else
Bat_high(i)=0;
Bat_low(i)=0;
end
else
if(abs(SOC(i)1)<Marg)&(Bat_low(i1)==1&Bat_high(i1)==1)
Bat_high(i)=2;
Bat_low(i)=Bat_low(i1);
elseif(abs(SOC(i)1)<Marg)&((Bat_low(i1)==0)&(Bat_high(i1)==1|Bat_high(i1)==0))
Bat_high(i)=1;
Bat_low(i)=Bat_low(i1);
elseif(abs(SOC(i)1)<Marg)&(Bat_low(i1)==1&Bat_high(i1)==0)
Bat_high(i)=1;
Bat_low(i)=Bat_low(i1);
elseif(abs(SOC(i)SOC_ref)<Marg)
Bat_high(i)=Bat_high(i1);
Bat_low(i)=1;
else
Bat_high(i)=Bat_high(i1);
Bat_low(i)=Bat_low(i1);
end
end
%Whena"complete"chargefollowsa"complete"dischargeacycleisadded
if(Bat_high(i)==2&Bat_low(i)==1)
Bat_cycles=Bat_cycles+1;
Bat_high(i)=1;
Bat_low(i)=0;
else
end
109
%Countingelectrolyseronoffswitching
%Statusdefinition(EL_ON_OFF(i)):
%0=Electrolyserisoffuntilthistimestepoftheday
%1=Electrolyserturnedonattheprevioustimestep
%2=Itoperatesmorethatonetimestep
%2.x=Itoperatesforx+1timesteps
%3=Itwasswitchedoffattheprevioustimestep
%4=Ithasbeenswitchedoffmorethanonetimestep
%10=Ithasexceededtheminimumtimestepsofoperation
ifHydrogen_Production>0
Elec_on(i)=1;
ifi>Min_hours
ifElec_on(i1)==0
EL_ON_OFF(i)=1;
elseifElec_on(i1)==1&EL_ON_OFF(i1)==1
EL_ON_OFF(i)=2;
elseifElec_on(i1)==1&EL_ON_OFF(i1)~=1
EL_ON_OFF(i)=EL_ON_OFF(i1)+0.1;
else
end
ifEL_ON_OFF(i)>(Min_hours/10)0.1+2
EL_ON_OFF(i)=10;
else
end
else
end
else
Elec_on(i)=0;
ifi>1
ifElec_on(i1)==1%Whenitisturnedonacycleisadded
Electrolyser_cycles=Electrolyser_cycles+1;
EL_ON_OFF(i)=3;
elseifElec_on(i1)==0&(EL_ON_OFF(i1)==3);
EL_ON_OFF(i)=4;
else
end
else
end
end
%PLOTS
%Plot1PVproduction
plot_1A=Hour_of_the_year;
ifP_bat_net>0
plot_1B=P_bat_net;
plot_1C=plot_1B+Power_consumption;
else
plot_1B=0;
110
plot_1C=Power_consumption+P_bat_net;
end
plot_1D=plot_1C+T_p_rej;
plot_1E=PV_p_prod;
%Plot2Electrolyserconsumption
ifP_bat_net>0
plot_2B=0;
plot_2C=PV_p_prodT_p_rejP_bat_net;
else
plot_2B=abs(P_bat_net);
plot_2C=PV_p_prod+plot_2BT_p_rej;
end
plot_2D=Power_consumption;
%Plot3Tankslevel
plot_3B=SOF(i);
plot_3C=SOC(i);
ifopt==0
fprintf(fid,'%6.0f %12.8f %12.8f %12.8f %12.8f %12.8f %12.8f %12.8f %12.8f %12.8f
\r\n',plot_1A,plot_1B,plot_1C,plot_1D,plot_1E,plot_2B,plot_2C,plot_2D,plot_3B,plot_3C);
else
end
%Balancescheck
%Specificconsumptionestimation
EL_op_point=Power_consumption/(ELmax*EL_power);%Electrolysercurrentoperationpoint
%NondimensionedrelationshipobtainedbyUllebergetal
El_sp_cons_nd=2.1478*(EL_op_point)^3+4.2745*(EL_op_point)^22.7977*(EL_op_point)+1.6403;
El_sp_cons=El_sp_cons_nd*Eecon;%Electrolyserspecificconsumption(kWh/Nm3)
%OVERALLENERGYMANAGEMENT
ifi>1
if abs(PV_en_prodEn_bat_netEnergy_consumptionT_en_rej)>0.00001 | abs(((SOF(i)SOF(i
1))*Capacity+(Fuel_DemandNon_cov_dem))*El_sp_consEnergy_consumption)>0.001
disp(i)
disp(PV_en_prod)
disp(P_bat_net)
disp(Energy_consumption)
disp(T_en_rej)
disp(SOF(i))
disp(PV_en_prodEn_bat_netEnergy_consumptionT_en_rej)
else
end
else
end
%BATTERY
111
ifi>1
ifEn_bat_net>0
ifabs(En_bat_net*0.85(SOC(i)SOC(i1)+0.02*Time/(30*24))*B_Capacity)>0.001
disp(i)
disp(En_bat_net)
disp(En_bat_net*0.85)
disp(SOC(i))
disp(SOC(i1))
else
end
else
ifabs(En_bat_net/0.85(SOC(i)SOC(i1)+0.02*Time/(30*24))*B_Capacity)>0.001
disp(i)
else
end
end
else
end
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%%%%%%%%%%%%%%
%Cumulativemagnitudescalculation
Annual_Dem=Annual_Dem+Fuel_Demand;%Cumulativeannualfueldemand
Annual_Dem_ns=Annual_Dem_ns+Non_cov_dem;%Cumulativenoncovereddemand
Total_H2_prod=Total_H2_prod+Hydrogen_Production;%AnnualHydrogenproduction
Total_PV_prod=Total_PV_prod+PV_en_prod;%AnnualPVenergyproduction
Total_rejection=Total_rejection+T_en_rej; % Annual energy rejection (batteryelectronic devices
lossesexcluded)
Total_el_consumption=Total_el_consumption+Energy_consumption; % Annual electrolyser energy
consumption
Total_radiation=Total_radiation+Radiation*Area;%Totalsolarradiationtothepanels
end
%ResultsOutputEvaluationMagnitudes
%CallEconomicsfunction
[
Hydrogen_prod_cost
]
=
Economics(Area,B_Capacity,Nominal_production_rate,Capacity,Total_H2_prod,Bat_cycles,Electrolyser_
cycles,Depth_of_Discharge,EL_Cost_reduction,PV_Cost_reduction);
En_rej=Total_rejection/Total_PV_prod;%EnergyrejectiontoPVproduction
PV_CF=(Total_PV_prodTotal_rejection)/(PV_power*length(Athens));%PVcapacityfactor
EL_CF=Total_el_consumption/(EL_power*length(Athens));%ELcapacityfactor
Overall_Efficiency=Total_H2_prod*HHV/Total_radiation;%Annualoverallsystemefficiency
EL_cycles=Electrolyser_cycles;%AnnualElectrolysercycles
Battery_cycles=Bat_cycles;%AnnualBatterycycles
Reliability=(1Annual_Dem_ns/Annual_Dem);%Reliabilityofthespecificdimensionedsystem(%)
%Summarytable
ifopt==0
fprintf(fid2,'%6.3f\r\n',abs(Reliability));
fprintf(fid2,'%6.3f\r\n',Hydrogen_prod_cost);
fprintf(fid2,'%6.3f\r\n',En_rej);
fprintf(fid2,'%6.3f\r\n',abs(PV_CF));
112
fprintf(fid2,'%6.3f\r\n',abs(EL_CF));
fprintf(fid2,'%6.4f\r\n',abs(Overall_Efficiency));
fprintf(fid2,'%4.0f\r\n',EL_cycles);
else
end
%Constraintsusedforoptimizationspeedup
%Filesclose
ifopt==0
fclose(fid);
fclose(fid2);
else
end
%Optimizationobjectives
f(1)=Hydrogen_prod_cost;
%f(2)=Days_of_autonomy;
%f(3)=B_Capacity/EL_power;
end
113
REFERENCES
[1] J. Hu, D.L. King, Y. Wang J.D. Holladay, "An overview of hydrogen production
technologies,"CatalysisToday,no.139,pp.244260,2009.
[2] C.J. Winter, "Hydrogen energy Abundant, efficient, clean: A debate over the
energysystemofchange," International Journal of Hydrogen Energy, no. 34, pp.
152,2009.
[3] R. McConnell, S. Licht K. Rajeshwar, "Solar Hydrogen Generation Toward a
RenewableEnergyFuture,"SpringerScienceBusinessMedia,2008.
[4] .Markvart,SolarElectricity,2nded.:JohnWileyandSonsLtd,2000.
[5] T. L. Gibson, D. B. Ouwerkerk N. A. Kelly, "A solarpowered, highefficiency
hydrogen fueling system using highpressure electrolysis of water: Design and
initialresults,"InternationalJournalofhydrogenenergy,no.33,pp.27472764,
2008.
[6] S. Giddey, F.T. Ciacchi, S.P.S. Badwal, B. Paul, J. Andrews R.E. Clarke, "Direct
coupling of an electrolyser to a solar PV system for generating hydrogen,"
InternationalJournalofHydrogenEnergy,no.34,pp.25312542,2009.
[7] T. Nakken, A. Ete . Ulleberg, "The wind/hydrogen demonstration system at
Utsira in Norway: Evaluation of system performance using operational data and
updated hydrogen energy system modeling tools," International Journal of
hydrogenenergy,no.35,pp.18411852,2010.
[8] F. Barbir, "PEM electrolysis for production of hydrogen from renewable energy
sources,"SolarEnergy,no.78,pp.661669,2005.
[9] M.Thomson,D.InfieldM.Little,"Electricalintegrationofrenewableenergyinto
standalone power supplies incorporating hydrogen storage," International
JournalofHydrogenEnergy,no.32,pp.15821588,2007.
[10].Ulleberg,"TheimportanceofcontrolstrategiesinPVhydrogensystems,"Solar
Energy,no.76,pp.323329,2004.
[11]S. Voutetakis, P. Seferlis, F. Stergiopoulos, C. Elmasides D. Ipsakis, "Power
management strategiesfor a standalone power system using renewable energy
114
sourcesandhydrogenstorage,"InternationalJournalofHydrogenEnergy,no.34,
pp.70817095,2009.
[12]E.Bilgen,"Domestichydrogenproductionusingrenewableenergy,"SolarEnergy,
no.77,pp.4755,2004.
[13]E. Bilgen, "Solar hydrogen from photovoltaicselectrolyser systems," Energy
Conversion&Management,no.42,pp.10471057,2001.
[14]A.Contreras,T.N.VezirogluJ.M.Vidueira,"PVautonomousinstallationtoproduce
hydrogen via electrolysis, and its use in FC buses," International Journal of
HydrogenEnergy,no.28,pp.927937,2003.
[15]N. A. Kelly Th. L. Gibson, "Optimization of solar powered hydrogen production
using PV electrolysis devices," International Journal of Hydrogen Energy, no. 33,
pp.59315940,2008.
[16]M.B.Pate,N.K.OlsonJ.R.Bartels,"Aneconomicsurveyofhydrogenproduction
from conventional and alternative energy sources," International Journal of
hydrogenenergy,no.xxx,pp.114,2010.
[17]J.M.Joubert,B.Lachal,K.YvonP.Hollmuller,"Evaluationofa5kWpPVhydrogen
production and storage installation for a residential home in Switzerland,"
InternationalJournalofHydrogenEnergy,no.25,pp.97109,2000.
[18]J.AndrewsB.Paul,"OptimalcouplingofPVarraystoPEMelectrolysersinsolar
hydrogen systems for remote area power supply," International Journal of
HydrogenEnergy,no.33,pp.490498,2008.
[19]N. A. Kelly Th. L. Gibson, "Predicting efficiency of solar powered hydrogen
generation using PVelectrolysis devices," International Journal of Hydrogen
Energy,no.35,pp.900911,2010.
[20]W. Martnez, U. Cano, H. Blud L.G. Arriaga, "Direct coupling of a solarhydrogen
systeminMexico,"InternationalJournalofHydrogenEnergy,no.32,pp.2247
2252,2007.
[21]O. Atlam, "An experimental and modelling study of a photovoltaic / proton
exchange membrane electrolyser system," International Journal of Hydrogen
Energy,no.34,pp.65896595,2009.
[22]J.Ivy,"SummaryofElectrolyticHydrogenProduction,"NationalRenewableEnergy
115
Laboratory,2004.
[23]D. Zhang K. Zeng, "Recent progress in alkaline water electrolysis for hydrogen
productionandapplications,"ProgressinEnergyandCombustionScience,no.36,
pp.307326,2010.
[24]L.CastanerT.Markvart,Ed.,PracticalHandbookofPhotovoltaicsFundamentals
andApplications.:Elsevier,2003.
[25]M. K. Mann, R. M. Margolis, A. Milbrandt J. I. Levene, "An analysis of hydrogen
productionfromrenewableelectricitysources,"SolarEnergy,no.81,pp.773780,
2007.
[26]K.Agbossou,J.Hamelin,T.K.BoseM.Kolhe,"Analyticalmodelforpredictingthe
performance of photovoltaic arraycoupled with a wind turbine in a standalone
renewable energy system based on hydrogen," Renewable Energy, no. 28, pp.
727742,2003.
[27]T. Muneer, P. Clarke A. N. Celik, "Optimal Sizing and Life Cycle Assessment of
Residential Photovoltaic Energy Systems With Battery Storage," Progress in
Photovoltaics:ResearchandApplications,no.16,pp.6985,2008.
[28]S. A. Sharkh D. Doerffel, "A critical review of using the Peukert equation for
determining the remaining capacity of leadacid and lithiumion batteries,"
JournalofPowerSources,no.155,pp.395400,2006.
[29]D. Zafirakis, K. Kavadias J.K. Kaldellis, "Technoeconomic comparison of energy
storage systems for island autonomous electrical networks," Renewable and
SustainableEnergyReviews,no.13,pp.378392,2009.
[30]T.B.ThomasD.Linden,HandbookofBatteries,3rded.NewYork,USA:McGraw
Hill,2002.
[31].UllebergH.Miland,"Testingofasmallscalestandalonepowersystembased
on
solar
energy
and
hydrogen,"
Solar
Energy,
no.
doi:10.1016/j.solener.2008.04.013,2008.
[32]P.FranklM.Raugei,"Lifecycleimpactsandcostsofphotovoltaicsystems:Current
stateoftheart,"Energy,no.34,pp.392399,2009.
[33]Z.Wei,L.ChengzhiY.Hongxing,"Optimaldesignandtechnoeconomicanalysisof
ahybridsolarwindpowergenerationsystem,"AppliedEnergy,no.86,pp.163
116
169,2009.
[34]X.J. Zhu, G.Y. Cao, S. Sui, M.R. Hu C.H. Li, "Dynamic modeling and sizing
optimization of standalone photovoltaic power systems using hybrid energy
storagetechnology,"RenewableEnergy,no.34,pp.815826,2009.
[35]R.Glockner,N.Lymberopoulos,T.Tsoutsos,I.Vosseler,O.Gavalda,H.J.Mydske,
P.TaylorE.I.Zoulias,"Integrationofhydrogenenergytechnologiesinstandalone
powersystemsanalysisofthecurrentpotentialforapplications,"Renewableand
SustainableEnergyReviews,no.10,pp.432462,2006.
[36]Amitava Roy, John Barton, Matthew Little Rupert Gammon, "HYDROGEN AND
RENEWABLES INTEGRATION (HARI)," CREST (Centre for Renewable Energy
SystemsTechnology),LoughboroughUniversity,UK,2006.
[37]John M., Jr. and Robert E. Chapman Watts, "Engineering Economics," in SFPE
HandbookofFireProtectionEngineering.:NFPA,QuincyMA,2002,ch.7.
[38]J. Shayegan M. Qadrdan, "Economic assessment of hydrogen fueling station, a
casestudyforIran,"RenewableEnergy,no.33,pp.25252531,2008.
117