Réjection de perturbation sur un système multi-sources - Application à une propulsion hybride

École doctorale : Sciences et ingénierie pour l'information, mathématiques - S2IM (Poitiers)

Secteur de recherche : Automatique, productique

Présentée par :

Ping Dai

Réjection de perturbation sur un système multi-sources

Application à une propulsion hybride

Directeur(s) de Thèse :

Patrick Coirault, Sebastien Cauet

Soutenue le 19 janvier 2015 devant le jury

Jury :

Président

Xavier Moreau

Professeur, ENSEIRB, Université de Bordeaux

Rapporteur

Ahmed El Hajjaji

Professeur, Université de Picardie, Amiens

Rapporteur

Malek Ghanes

Maître de conférences, ENSEA, Cergy-Pontoise

Membre

Patrick Coirault

Professeur, Université de Poitiers

Membre

Sebastien Cauet

Maître de conférences, LIAS, ENSIP, Université de Poitiers

Pour citer cette thèse :

Ping Dai. Réjection de perturbation sur un système multi-sources - Application à une propulsion hybride [En ligne].

Thèse Automatique, productique. Poitiers : Université de Poitiers, 2015. Disponible sur Internet

Université de Poitiers

THÈSE

présentée à

l'Universitéde Poitiers envue del'obtention du

Diplme de Do torat

Spé ialité:Automatique

Présentéepar

Ping DAI

Réje tion de perturbation sur un système

multi-sour es - Appli ation à une propulsion hybride

Dire teur dethèse: Patri kCOIRAULT

Co-en adrement:SébastienCAUET

Soutenuepubliquementlelundi19janvier2015

Jury

Rapporteurs : AhmedELHAJJAJI Professeurdesuniversités Universitéde Pi ardie,Amiens

Malek GHANES Maîtrede Conféren es HDR

ENSEA,Paris

Examinateurs: XavierMOREAU Professeurdesuniversités

ENSEIRB,Universitéde Bordeaux Patri k COIRAULT Professeurdesuniversités

Universitéde Poitiers

Sébastien CAUET MaîtredeConféren es HDR

Universitéde Poitiers

A knowledgments

This thesisis not only a summaryof thethree-year resear h intheLaboratoire d'Automatique et d'Informatique (LIAS), but also a milestone in almost a de ade of study in the domain of ele tri al engineering ontrol. I would like to thank my advisor,Professor Patri kCOIRAULT, dire torofthelaboratoryforoeringmesu hanopportunitytodotheproje tandforwel oming me inthe lab. Throughout these threeyears,he hasbeen providingme a ademi guidan e and professional supportandled meto afresh advan ed ontroleld.

I would like to extend the gratitude to my o-supervisor Mr. Sébastien CAUET, Maître de onféren e HDRattheUniversityofPoitiers,notonlyforhis ontinueden ouragement,butalso for his help and guidan e in power ele tri al domain and in the experimental test ben h. His professional qualitieshave given mea deep impression,and would stillbevaluable inmyfuture areer.

Furthermore,Iwouldliketoextendmyappre iationtoallmembersofthejuryforevaluatingmy

work :Professor Ahmed EL HAJJAJI, Université de Pi ardie à Amiens, Mr.Malek GHANES,

MaîtredeConféren esHDRàl'ENSEAandProfessorXavierMOREAU,l'universitédeBordeaux I. I would like to thank parti ularly Mr. GHANES for his inspiration whi h has tremendously pushedforward our study.

Inaddition,Iwouldalsoliketoexpress mygratitudetothepresidentofour department Poitou-Charentes for the fundingsupportduring thesethree years.

Moreover, I would like to thank my lab for giving the opportunity of tea hing the experiment pra ti al lessons during thelast twoyears,where Igot a perfe t han eto review, enhan eand pra ti e the professionalknowledge and a umulate tea hing experien e. It was mypleasure to playthe roleof tea her, andI appre iateeverymoment beingwithmystudents.

Mythanksandappre iations alsoextend toea h individualinLIAS.Thanksto every olleague inLIAS for givingmeenthusiasti on ern andhelp. Espe iallyto Baya,Duong, Farah, Fayçal, Amin for the ompanionand helpall along the three years, and to Lila, Mariem for their help during theshort but herish period together.I oerthem all mybest wishesintheir future.

Finally,I wouldlike to thank myparents, my auntsand my wholefamily for their endless love and supportall alongtheroad.Withoutthem,Iwillneverrea hwhere Iam.Also thankyouto my dearfriends inChina, in Fran e and those all over theworld for always being there,for all

Table des matières

General introdu tion 1

1 Thestate of theart of hybridenergy storagesystem forvehi ularappli ation 5

1.1 Introdu tion . . . 6

1.2 Energy storagesystems . . . 6

1.2.1 Fuel ells . . . 7

1.2.2 Batteries . . . 9

1.2.3 Ultra apa itors . . . 10

1.2.4 Comparisons of dierent devi es . . . 12

1.3 Hybridization of energy storagesystems . . . 13

1.3.1 Congurations ofhybrid energy storagesystem . . . 13

1.3.2 Congurations ofDC-DC onverters . . . 17

1.4 Control methodologies . . . 18

1.4.1 Energy management . . . 18

1.4.2 Power onverter ontrol . . . 20

1.5 Con lusions . . . 23

2 Disturban e reje tion theory and appli ation 25 2.1 Introdu tion . . . 26

2.2 Disturban es inele tri alpart ofHEVs . . . 26

2.2.1 Me hani al equations. . . 26

2.2.2 Ele tri al equations. . . 27

2.2.3 Current disturban esinDC bus. . . 28

2.2.4 Control obje tives . . . 29

2.3 Disturban e reje tion theories . . . 30

2.3.1 Linear systems . . . 31

2.3.2 Nonlinear systems . . . 34

2.3.3 Hamiltonian systems . . . 37

2.4 Theory appli ation . . . 39

2.4.1 Battery side onverter . . . 40

2.4.2 Ultra apa itor side onverter . . . 42

3 Hybrid battery/ultra apa itor ontrolstru ture design 51 3.1 Introdu tion . . . 52 3.2 System Modeling . . . 52 3.2.1 Averagemodel . . . 53 3.2.2 Hamiltonian modeling . . . 53 3.3 Control strategies . . . 55

3.3.1 Port- ontrolled Hamiltonian systems . . . 55

3.3.2 Passivity-based ontrol . . . 56

3.3.3 Singularperturbation theories . . . 57

3.4 Controller design . . . 57

3.4.1 Stati and dynami solutions . . . 57

3.4.2 Cas ade ontrol stru ture . . . 59

3.4.3 Internal modeldesign . . . 65

3.5 Simulationresults. . . 68

3.6 Con lusions . . . 71

4 Experiment implement and results analysis 75 4.1 Introdu tion . . . 76 4.2 Experimental equipment . . . 76 4.2.1 Battery system . . . 77 4.2.2 Ultra apa itor . . . 77 4.2.3 Exosystem. . . 80 4.2.4 Power onverters . . . 81

4.3 Control algorithm embedding . . . 82

4.4 Experimental results ofbattery/ultra apa itor hybrid system . . . 83

4.4.1 Sinusoidal disturban e . . . 84 4.4.2 Transient disturban e . . . 87 4.4.3 General disturban e . . . 88 4.5 Con lusions . . . 90 General on lusions 93 Referen es 94

Table des gures

1.1 S hemati of stru ture andoperation me hanism offuel ells. . . 7

1.2 S hemati of stru ture andoperation me hanism ofbatteries . . . 10

1.3 S hemati of stru ture andoperation me hanism ofsuper apa itors . . . 11

1.4 Spe i energy and powerof dierent devi es (Sour e:[Zho06℄) . . . 12

1.5 Parallel onne tionoffuel ells, batteriesand ultra apa itors . . . 14

1.6 Congurations ofbatteries/UC hybridization . . . 15

1.7 Battery and UCsharing one ommon DC-DC onverter . . . 16

1.8 Battery and UChybridizationwith anadditional diode. . . 17

1.9 Integratedmagneti stru ture . . . 17

1.10 Topologies ofDC-DC onverters. . . 18

1.11 Commonly useddriving y les. . . 19

1.12 Cas ade ontrol stru ture for power onverters . . . 21

1.13 RST onverter ontrol stru ture . . . 21

1.14 Slidingsurfa e and the systemmovement . . . 22

2.1 Battery/Ultra apa itor hybrid energy storagesysteminhybrid ele tri vehi les . 27 2.2 Persistent disturban eand harmoni de omposition . . . 29

2.3 Transient disturban e and lteredsignals . . . 30

2.4 System withexogenousdisturban es . . . 30

2.5 Output regulator forlinear systems . . . 34

2.6 Stru tureof the ontrolled systemfornonlinear ase . . . 36

2.7 Considered topology of the hybrid energy storagesystem . . . 40

2.8 Zerodynami orrespondingto theoutputvoltage . . . 41

2.11 Duty y lesof the ontrolsignals of the onverters . . . 49

2.12 Traje toryevolutionforbatteryside onverter (left:fromalarge losed traje tory to a desired equilibrium point) and traje tory evolution for ultra apa itor side onverter (right:from asmall losed traje tory to adesired limit y le) . . . 49

3.1 Topology of ele tri alDC part. . . 53

3.2 Control stru ture of thepower onverters . . . 59

3.3 Simulation results with sinusoidal urrent disturban e (From top to bottom: ex-ternal urrent, urrentthrough

L

1

,voltageintheDCbus, urrentthrough

L

2

and ultra apa itor voltage) . . . 69

3.4 Spe trumanalysis of the urrent through

L

2

with (right) and without (left) the ontrol algorithm . . . 69

3.5 Simulationresults withthreeharmoni urrentdisturban e(Fromtopto bottom: external urrent, urrent through

L

1

, voltage inthe DCbus, urrent through

L

2

and ultra apa itor voltage) . . . 70

3.6 Spe trum analysis of the urrent through

L

2

with (right) and without (left) the ontrol algorithm . . . 70

3.7 S hemati diagram oflters . . . 71

3.8 External urrent and theoutputs ofthe lters . . . 71

3.9 Simulationresults withsinusoidalandstep urrentdisturban e (Fromtopto bot-tom: urrent through

L

1

,voltage intheDCbus, urrent through

L

2

and ultra a-pa itorvoltage) . . . 72

3.10 Desired traje tories of ultra apa itor side onverter (left: when the external dis-turban e

ω

1

is persistent sinusoidal signal with dierent harmoni s; right: when

ω

1

istransient signalplusa sinusoidal signal) . . . 72

4.1 Experimental test ben h . . . 77

4.2 Classi alultra apa itor model . . . 79

4.3 An a uate ultra apa itormodel . . . 79

4.4 Exosystem . . . 80

4.5 Overvoltage prote tion forthe DCbus. . . 81

4.6 Experimental ontrol stru ture . . . 82

4.7 Anti-windup . . . 83

4.8 System responses with and without disturban e reje tion algorithm under sinu-soidaldisturban e. . . 85

4.9 Battery urrentstati spe trumwithandwithoutdisturban ereje tionalgorithm undersinusoidaldisturban e . . . 85

TABLE DESFIGURES ix

4.11 Ultra apa itor urrent andits referen eundersinusoidaldisturban e . . . 86

4.12 Systemresponseswithout disturban ereje tionalgorithm undertransient distur-ban es . . . 87

4.13 Systemresponseswithdisturban e reje tionalgorithmundertransient disturban es 88

4.14 Contrasts of battery and ultra apa itor urrent with and without disturban e reje tion algorithm undertransient disturban es. . . 89

4.15 The errorbetween ultra apa itor urrent and its referen e . . . 89

4.16 System responses with and without disturban e reje tion algorithm under both transient disturban esandsinusoidaldisturban es . . . 90

Liste des tableaux

1.1 Chara teristi s ofdierent devi es (Sour e:[Zho06℄) . . . 12

4.1 Ele tri parameters of the experimental system . . . 78

Notations et abréviations

Notations :

R

: eldofreal numbers

R

m

, R

n

_{, R}

q

: linearspa eof realve tors ofdimension

m, n, q

t

: time,nonnegative real numbers

R

x

: shortfor

R

t

0

x(s)ds

,timeintegration

d

dt

x = ˙x

: time derivative of

x

∂x

∂ξ

: derivative of

x = f (ξ)

I

: identity matrix

A

⊤

: the transposeof matrix

A

A

−

1

: the inverseof matrix

A

sym{A}

:

sym{A} = A + A

⊤

the sum ofmatri es

A

and

A

⊤

L

s

c(ω)

: Lieoperation

L

s

c(ω) = s(ω)

∂

∂ω

c(ω)

L

q

s

c(ω)

: Lieoperation

L

q

s

c(ω) = s(ω)

_∂ω

∂

[L

q−1

s

c(ω)]

J

: the massmoment of inertiaof the vehi le

ω

m

: the rotational speedof thevehi le

T

pmsm

: the torqueofpermanent magnet syn hronous ma hine

T

p

: the torquegenerated bythe pressure inthe ylinder inthe ombustion pro ess

T

i

: the torquegenerated bythe os illating masses and onne tingrod

T

l

: the loadtorque

i

d

,

i

q

: urrent inthe

dq

referen eframe rotatingat thesame speed oftherotor

p

: pairof poles oftheele tri ma hine

φ

m

: the magnetuxof theele tri ma hine

S

1,2,3,4

: ontrolled swit hesofthepower onverters

u, µ

: duty y les of the ontrolled swit hes( ontrol inputs ofthe power onverters)

L

l

, L

2

: indu tan esinthepower onverters

C, C

sc

: apa itan eintheDC bus, apa itan e inthesuper apa itor/ultra apa itor

r

: innerresistan eof battery

R

: resistan eoftotal lossinthe ir uit

R

sc

: innerresistan eof thesuper apa itor/ultra apa itor

V

dc

: voltageon the DC bus

V

sc

: voltageof thesuper apa itor/ultra apa itor

I

L

1

, I

L

2

: urrents throughtheindu tors

L

1

and

L

2

φ

L

l

, φ

L

2

: uxinthe indu tors

L

1

and

L

2

q

C

, q

C

sc

: harge inthe apa itors

C

and

C

sc

I

ex

= ω

1

: external urrent ofthe power onverters

¯

ω

1

: onstant omponent of

ω

1

˜

ω

1

: variable omponent of

ω

1

ω

: exogenerousdisterban e

x, x

∗

: statevariables ofa system,referen e ofthestate variables

u, u

∗

: ontrol input ofa system,referen eof the ontrol input

Π, Γ

: matrixmappings from

{x, u}

to

{x

∗

, u

∗

}

oflinear systems

π(ω), c(ω)

: mappingsfrom

{x, u}

to

{x

∗

, u

∗

}

ofnonlinear systems

k

p

, k

i

: proportional gain, integral gain

V

: magneti energy inan indu tor

T

: ele tri energy ina apa itor

H

: Hamiltonian/energy ofa system

H

d

: desiredHamiltonian/energy ofa system

∇H =

∂H(x)

∂x

: derivative of

H

intermsof

x

J

: inter onne tion matrix

J

d

: desiredinter onne tion stru ture

R

: dampingmatrix

R

d

: desireddamping

R

a

: inje ted damping

R

d

= R + R

a

Abréviations : AC : AlternateCurrent DC : Dire t urrent EV : Ele tri Vehi le

HEV : HybridEle tri Vehi le

HESS : HybridEnergyStorage System

ICE : Internal Combustion Engine

IDA : Inter onne tion and DampingAssignment

LMI : LinearMatrix Inequality

PBC : Passivity BasedControl

PCH : Port-Controlled Hamiltonian

PI : Proportional Integral

PMSM : Permanent MagnetSyn hronous Ma hine

PWM : Pulse-WidthModulation

SMC : SlidingMode Control

SOC : State of Charge

General introdu tion

Context

Inthe19th entury,petrolor gasolinepoweredautomobileshavegraduallyrepla edsteam powe-redautomobileswhi hhaveo upiedthehistoryfora entury.Petrolorgasolinepoweredinternal ombustion engine (ICE), being a dominant means of propulsion nowadays, has improved the speedofvehi les dozensor evenhundredsoftimes. A tually,itisnotonly thevehi lespeedthat in reases, but also the onsumption of fossil fuels and exhaust emissions. While we enjoy the onvenien e broughtbyautomobiles, wearealsowithstandingtheexhaustpollution.Inthe20th entury,withthein reaseoftheawarenessoftheprote tionofnon-renewableenergyandtheair ondition,ele tri alpoweredautomobilesappearedandbegan too upythemarkettodayinthe 21st entury.Hybridele tri vehi les(HEV),beinganoptimal andidate,signi antlyredu ethe toxi and harmful gasemissions withoutsa ri ing thetravellingvelo ity.Itis still freshinour memory thesmog loudhanging over Paris this Mar h.Even thelandmark EielToweralmost disappearedinthe smog.Thegovernment hastakenmeasures in ludingalternating drivingban to limit thepollution,whileHEVisan ex eptionandis notin luded intheban.Theadvantage of HEV is thus evident.In fa t, thepoli yand nan ial dual supportfrom thegovernment has essentiallypromotedand advan edthe resear h and themarketofHEV.

During all these years in the study of HEV, we have been ommitted to reate a omfortable driving environment with lessnoiseand smootherrunning. Thisis a hieved via thesuppression of the torque ripples generated by the ICE. Instead of utilizing the traditional passive ontrol method with a ywheel, Mohamed NJEH [Nje11℄ has ontributed an a tive ontrol method to ontrol the torqueofthe ele tri ma hine, apermanent magnetsyn hronous ma hine(PMSM), to ompensate the ICE torque ripples. The essen e of the study is based on the harmoni de- omposition.The ripple disturban eisde omposedinto dierent harmoni sandthentreated in parallel. The torque ripple ompensation ontroller is thus a parallel onne tion of a series of sub ontrollers, withea h sub ontrollerdealing withonemain harmoni .Thea tive ontroller is for the AC-DC onverter onne ting to the DC bus. Consequently, the torqueripples ompen-sation pro ess results in a transfer of ripple disturban e in the DC bus. This disturban e also

Problem statement

The DCbus voltage issupplied bya battery onne tingwitha DC-DC onverter.The problem leftfor usisthento dealwiththedisturban e inthe DCbusintrodu ed inthepro essoftorque ripple ompensation. Thedire tdamage ausedbythedisturban eisbatterywear. Exposingto su h persistent os illations, batteries, with the property of high energy density but low power density, annotaord su hsubstantial andfrequentpowervariation y les,and thus weardown very qui kly. Consequently, it is quite ne essary to nd a solution to reje t the disturban e in theDC bus.

A tually, talking about batteries, it is easy to think of ultra apa itors being widely used as a omplementary powersour e ina hybrid battery/ultra apa itor energy storagesystem. Hybrid energy storage system has been ommonly studied in ele tri vehi les and other appli ations in ludingrenewableenergy powergeneration. Inmost studies, theappli ation ofultra apa itors invehi ularsystemismainlytodealwiththetransientdisturban e,anotherfa t ausingbattery wear,introdu edbytheloadpowerdemandsudden hangeduringa elerationandde eleration. In ontrast withbatteries, ultra apa itors, withhigher power density, is apt to provide a large amount of energy in a short period and experien e frequent and rapid harge dis harge y les. Therefore,battery/ultra apa itorhybridstru ture providesaperfe tsolutiontoredu ethe bat-tery wearifwe ouldfor e the ultra apa itor to absorbboth thetransient disturban e and the persistent disturban e. In order to a hieve this, it is ne essary to design a high-performan e ontrol algorithm, andthis isthe ore ofour study.

Obje tives

Ourresear h obje tisthenabattery/ultra apa itorhybridenergystoragesystem onne tingto the DC bus where there are exogenous transient and persistent disturban es. Our obje tive is to reje tthedisturban esinthe batteryandabsorbthemintheultra apa itor. Other a essory ontrolobje tivesin ludemaintaininga onstantDCvoltageand ompensatingtheself-dis harge of the ultra apa itor to maintain at thenominalstate of harge.

Fromenergy management point ofview,it requiresproperly managing theenergy allo ation in thebattery/ultra apa itor systemsoastoa hieve the ontrol obje tive.Spe i ally,thetaskof delivering steadyenergy intheDC bus isassignedto thebattery,while thetaskofdisturban e energy absorptionisallo atedtotheultra apa itor. Fromos illatingdisturban e reje tionpoint of view,itrequiresan a tive ontroller to turntheultra ap itor into an a tivedamping.

Thebatteryandthe ultra apa itorare onne tedwithaDC-DC onverterrespe tively.Forthis DC-DC power ele tri al system, the average model of the whole system is a four-dimensional nonlinear system and is not easy to be linearized, and thus it is not feasible to dire tly apply lassi allineartheories.Fa ingthisdi ulty,wehaveexploredvariousnonlinear ontroltheories, fromslidingmode ontroltopassivity-based ontrolofEuler-Lagrangesystems,andtononlinear error output regulation, and to the inter onne tion and damping assignment of Hamiltonian systems.Aftersueringnumerousfrustrations, wenallyunderstand deeplytheprin iple ofthe hybrid energy systemand ontribute a disturban e reje tion ontrol algorithm. The pro ess is

Generalintrodu tion 3

Organization of the dissertation

The dissertation isdividedinto four hapters.

The rst hapter presents a state of the art of hybrid energy storage system. In this hapter, we rst review the working prin iples and hara teristi s of the widely used DC power sour e and storagedevi es :fuel ells, batteries andultra apa itors. Moreover, we summarize dierent ongurations ofhybrid storagesystemanddierent topologies ofDC power onvertersapplied inhybridstoragesystems.Furthermore,wepresentaseriesof ontrolmethodsaimingtoa hieve a properenergy management inhybrid storagesystems.

These ond hapterelaboratestheoriginofthesinusoidalpersistentdisturban e ausing battery wear and explores the output regulation theories of linear and nonlinear system. The related theory is then applied to our system. To a hieve this, we simplify the problem by onsidering the battery side onverter and the ultra apa itor side onverter separately, and apply the out-put regulation ontroller to theultra apa itor side onverter, while a passivity-based ontroller to the battery side onverter. Simulation results have shown theee tiveness of the ontrol al-gorithm. However, separating the battery and the ultra apa itor means that the intera tions between them arealso ignored. Moreover, the ultra apa itor modelused in this hapter is also simplied as an ideal model without onsidering its self-dis harge phenomenon. Consequently, la k of omprehensive onsideration, experiments in real-time are not implemented with this ontrol algorithm.

Thethird hapterpresentsour ontributeddisturban ereje tion ontrolalgorithm.Inthis hap-ter, the battery/ultra apa itor hybrid energy storage system is onsidered as a whole, and the ultra apa itor self-dis harge phenomenon is taken into onsideration by integrating a resistive loss inthe systemmodel. The whole systemis therefore a four-dimensional nonlinear Hamilto-niansystem.However, originalmethodtodesign the ontroller ofnonlinearHamiltoniansystem is not amenable to our system due to the omplexities and di ulties of solving partial die-rential equations. Consequently, we exploitan alternative way.Inorder toa hieve theobje tive of disturban e absorption in the ultra apa itor, we dene a stati solution and a dynami so-lution. The stati solution is the desired equilibriumpoint of thesystem when thedisturban e doesn't exist,while thedynami solution isthedesired dynami traje tories of thestates when theultra apa itorabsorbsthedisturban e.Thus,basedonthesingularperturbationtheory,the system an be onsidered intwo time-s ales, a slow one and a fastone, and ontrolled through a as ade stru ture. The ontrol obje tive an then be a hieved by driving the system to the desiredequilibriumpointthroughtheouterslowloopandimposingthedesireddynami through the inner fast loop. Besides, the outer slow loop ontroller is designed via inter onne tion and damping assignment, while the inner fast loop is regulated via a simple proportional integral ontroller.Simulationresultsaregivenandshowtheee tivenessofthe as ade ontrolmethod.

In the fourth hapter, experiments are arriedout to further verify theee tiveness of the dis-turban e reje tion ontrolalgorithm designed inthethird hapter. The ommuni ationbetween the analogue signals of theexperimental devi es and thedigital signals of the ontrol terminal is based on dSPACE hardware and software. A Lithium-Ion battery and an ultra apa itor are onne tedtotheDCbusviaaboost onverterandabi-dire tionalDC-DC onverterrespe tively. The disturban es ausing battery wearare emulated via an exosystem onsisting of a resistive load and an AC power sour e. Real-time system responses under sinusoidal persistent

distur-kind of disturban e, a ontrast of systemresponses withand without the designed disturban e reje tion algorithm isgiven andveries theee tiveness ofthe ontroller.

Chapitre 1

The state of the art of hybrid energy storage

system for vehi ular appli ation

Sommaire

1.1 Introdu tion . . . 6

1.2 Energy storagesystems . . . 6

1.2.1 Fuel ells . . . 7

1.2.2 Batteries . . . 9

1.2.3 Ultra apa itors . . . 10

1.2.4 Comparisonsofdierentdevi es . . . 12

1.3 Hybridizationof energystorage systems . . . 13

1.3.1 Congurationsofhybridenergystoragesystem . . . 13

1.3.2 CongurationsofDC-DC onverters . . . 17

1.4 Controlmethodologies . . . 18

1.4.1 Energymanagement . . . 18

1.4.2 Power onverter ontrol . . . 20

1.1 Introdu tion

Intoday'sso iety,peoplehavegradually realizedtheimportan eof environment prote tionand sustainabledevelopment.The ombustionoffossilfuels annolonger satisfytheenergy demand due to its pollution to the environment and e ology. As alternative energy sour es, renewable energy su has wind energy,solar energy and wave's energy et . are growing rapidly.Moreover, asopposedtotraditionalenergyprodu tionthatreliesonpetroleumandnaturalgas, ele tro he-mi alenergyprodu tionismoreenvironmentallyfriendlyandmoresustainable.Ele tro hemi al energy storage and onversion devi es in lude fuel ells, batteries, and ele tro hemi al apa i-tors. Redu ing the onsummation offossil-fuel is an issuerequiringthe ooperation of not only s ientists,but also e ologists, e onomistsand politi ians.

In the onguration of vehi les, an energy storage devi e is ne essary. Generally, re hargeable batteriesareutilizedinvehi lestosupplythe urrenttostartupandstoreredundantenergy.Fuel ells arewidelyapplied to pureele tri vehi les due to its highenergy density.Ultra apa itors, or super apa itorsarere ently studied asanauxiliary powerto apture theinstantaneous peak powers. Inthe onguration of hybrid DC power sour es, fuel ells andbatteries are oftenused asprimarypowersour ewhereasultra apa itors oftenserveasauxiliary powersour e.Theyare onne ted in series or inparallel to the DC bus via power onverters and supply power to the load(ele tri motors).RegulationoftheDCbusvoltageandtheenergydistributionmanagement be omesa mainresear h issue.

In this hapter, we will rst present the working prin iple and hara teristi s of dierent DC power sour es : fuel ells, batteries and ulter apa itors. Moreover, we will show some hybrid energy storage system ongurations and various DC onverter topologies applied in ele tri vehi lesandhybridvehi les.Finally,wewillsummarizeaseriesof ontrolte hnologiesaimingto manage the energy distribution for hybrid energy storagesystems in order to redu e the stress to the energy storage systems and thus extend their serve y le. A various ontrol theories to ontrol DC-DCpower onverterswill alsobe presented.

1.2 Energy storage systems

DC power sour es in luding fuel ells, batteries, and super apa itors are all ele tro hemi al energy storageand onversiondevi es. The ommon featuresof them isthat they all onsistof two ele trodes and ele trolyte solution, and the energy generation happens at the interfa e of theele trodesandtheele trolyte where theele tronsand theionsseparateandtransportinthe oppositedire tion [WB04℄.The basi operation me hanism of ea h systemwill be presented in thefollowing se tions.

In automotive appli ation, fuel ells or batteries generally serve as permanent sour e providing therequiredpermanentenergytoguaranteethesystemautonomy.Whereas,duetotheir hemi al stru tures,they annottotallysatisfytheloaddemand,andthus super apa itorsareintrodu ed

1.2 Energystoragesystems 7

H

2

O

2

H O

₂

e

e

hydrogen

anode

electrolyte

cathode

load

ions

−

−

Figure 1.1S hemati ofstru ture and operation me hanismof fuel ells.

1.2.1 Fuel ells

Fuel ells, asa leanand high-e ien y energy sour e,aremainly used inele tri vehi les,and they are intended to repla e traditional ombustion engines in automotive eld. In fa t, these two have a similar operating me hanism, that is they both need external fuels supply. Similar withinternal ombustion engines,fuel ellssupplyenergy aslongasthefuels aresupplied. The dieren e is that the fuels relled in the tank of fuel ells are hydrogen and oxygen (or air) and thatthe exhaustiswater.Whereas,internal ombustionengines onsumea largeamountof non-renewableenergy and generate greenhousegases and harmful gases.

1.2.1.1 Working prin iple and hara teristi s

The hydrogen and hydro arbon fuels used in fuel ells ontain a large amount of energy. This energy is signi antly higher than that found in ommon battery materials [WB04℄. Through hemi al rea tion of hydrogen and oxygen, fuel ells onvert the energy into ele tri al energy. Methanol and natural gas may be used instead of hydrogen gas. In this ase, a fuel reformer is ne essary to onvert the hydro arbon fuel into hydrogen-ri h gas. In this pro ess, quite a little arbondioxideandnitrogenoxidesaregenerated.Ingeneral,thistypeof hemi alrea tion produ esonlywaterinadditiontoele tri ity.Ele tri alvehi leswithfuel ellsaspropulsionare therefore onsideredtobethemostenvironmentallyfriendlyvehi lesand alledgreenvehi les".

Figure 1.1 shows a s hemati of the stru ture and operation me hanism of fuel ells. It an be seen from the gure that fuel ells onsist of an anode,where hydrogen is oxidized, a athode, where oxygen is redu ed by absorption of theele trons to the anions whi h rea t dire tly with the hydrogen ions to form water, andan ele trolyte, where ionsmovement take pla es between thetwo ele trodes.The fuels(hydrogen andoxygen) aresupplied ontinuously fromanexternal sour e, insteadof ontaining within the fuel ell ompartment.This feature allows fuels ellsto serve longer time than other ele tro hemi al ells as longas the fuelsare su ient,and avoids the harge-dis harge y les.

viaareform devi e.The hemi alrea tionsbetween thefuels(hydrogenand oxygen)releasethe energy intheforms of ele tri alenergy and heat.

Theearliestfuel ell anbetra edba kto1838.Germanphysi istChristianFriedri hS hönbein invented therst fuel ell by inserting two platinum wires in hydrogen and oxygen into hydro- hlori a id. Nowadays, in almost 200 years' study, s ientists have developed many dierent stru turesoffuel ells. A ording todierenttypesofele trolyte anddierent operation tempe-rature,fuel ellsare lassiedinto thefollowingtypes:Polymer Ele trolyteMembraneFuelCell (PEMFC),AlkalineFuelCell (AFC),Dire tMethanolFuelCell(DMFC),SolidOxideFuelCell (SOFC), Molten Carbonate Fuel Cell (MCFC), and Phosphori A id Fuel Cell (PAFC) [Nf ℄.

Among them, the lightweight and ompa t PEMFC (also alled Proton Ex hange Membrane

FuelCell) is the most widely used for automotive appli ations. A fuel ell sta kin automotive appli ation is integrated with several individual fuel ells in order to deliver powerful enough propulsion ofvehi les.

1.2.1.2 Advantage and disadvantages

As opposed to internal ombustion engines (ICE), other than lower emissions and lean power generation, fuel ells are more e ient in energy onversion. For example,the energy e ien y of traditional internal ombustion engine is only about 25%, while PEMFC may rea h energy e ien y up to 60%. This e ien y may be in reased up to 85-90% if the produ ed heat an be apturedandproperlyreused[SVP

+

91℄.Moreover,fuel ells ansupply ontinuousenergyas longastherea tantsareavailable.Hen e,itmayserveasenergygeneratorandsupply onstantly thesteady energyto vehi les.

Nowadays,fuel ell powergenerationsystemshave be ome aprimary hoi e forele tri vehi les to supplypowerto ele tri al ma hines. Manymajor manufa turers have invested in thestudies of fuel ell vehi les.January 2011,Mer edes-Benz Class BF-CELLfuel ellvehi les, have om-pleted globaljourney in125 days;GeneralMotors ChevroletEquinox100 fuel ellvehi les have ompleted 1.4millionmilesin2010;DaimlerMotorCompanyhasalreadyannoun edtheir pro-je t ofindustrialization offuel ell arsin 2014;Other manufa turers like Toyota, Honda,GM, Mer edes-Benz,Audi,Hyundaiet .havealsoshowntheir utting-edgete hnologyandambitions infuel ell vehi les[FCv℄.

In spite oftheaforementioned advantages and despitetheadvan edte hnologies, there arestill some barriers in using fuel ells. Firstly, the fuel ell resear h in automotive appli ation is still inthe initial stageandis notaswell developed asinternal ombustion engines,whi hleads to a relatively high ostoffuel ellpowersystem.Se ondly,duetoitsele tro hemi al hara teristi s, it is unable to allow bidire tion urrent ow. Current owing into fuel ells may ause severe damage,andthereforeitisunableto apturebrakingenergyduringvehi lede elerations.Thirdly, fuel ells have relatively highinternal resistan e and show slow dynami response. Thisresults indi ultiesduring oldstart-up.Besides,rapidloadpowerdemandduringvehi lea eleration may ausefuelstarvationphenomenon(dryingoftheFCmembrane)andthusredu eitslifetime.

Consequently, itrequiresa spe ialdesign while using fuel ells. Generally, adiode isutilized to avoidthe urrentowingintothefuel ell.Furthermore,inordertosolvetheforegoingproblems,a hybridizationoffuel ellswithbatteriesand/orsuper apa itorsisoftenapplied.Batteriesand/or

1.2 Energystoragesystems 9

powerduringa elerations.Thus, thehybrid power onguration ompensatestheinadequa ies of fuel ellsand redu es the overallsize and ost aswell.

1.2.2 Batteries

There aremainly three lasses of batteries :non-re hargeable (primary) batteries,re hargeable (se ondary) batteries and spe ialty batteries [WB04℄. As the names suggest, non-re hargeable batteries annotbere hargedandarethereforedis ardedon etheyarefullydis harged. Re har-geable batteriesmayre overtheir originalstate of hargeafterdis harged, andthereforewidely applied in portable ele troni devi es, su h as laptops, mobile phones, ameras, et . Spe ialty batteriesarenot ommonlyusedindailylife.Theyarespe iallydesignedtofulllsomepurposes inmilitary or medi al appli ations.

Anautomotivebatteryisne essaryintraditionalvehi lestosupplyele tri alpowertothestarter motor, the lights, and the ignition system(SLI battery) ofa vehi le's engine.Lead-a id type is the most frequently used SLI battery, and provide a relatively low voltage around

12V

. For heavy vehi les,su hastra torsor highwaytru ks equippedwithdieselengines, theSLI battery applied is about

24V

.Inele tri vehi lesor hybrid vehi les,batteries serving asa powersour e of ele tri alma hinearereferredto astra tionbatteries.Inparallel hybrid ele tri vehi les,the ele tri motor shares the load demand with the engine. A ording dierent operating modes, the per entage shared by the ele tri motor ould vary from 0 to 100%. This requires tra tion batteries to have a relatively high energy storage apa ity, so as to perfe tly satisfy the load demand.

1.2.2.1 Working prin iple and hara teristi s

Batteries onvert the self- ontained hemi al energy into ele tri al energy on demand. A s he-mati ofthestru tureand operationme hanismof batteriesisdepi ted inFigure1.2.Asshown inthegure,the basi elementsofbatteriesare:ananode(positive ele trode),a athode (nega-tive ele trode), andan ele trolyte.The athode isoftenmadeof materialseasily looseele trons su has zin or lithium. While the anode is often made ofthe ones easily a ept ele trons su h as manganese dioxide or lithium obalt oxide. The ele trolyte provides an environment where the ionstransport takes pla e [WB04℄.In thegure, the ionsmovement between the separated ele trolyte is a hieved withthe help ofthebridge.

In1780,theItalianphysi ianLuigiGalvaninoti edthatafrogleg,whi h ameinto onta twith opper and iron, repeatedly shrugged and thought itwas an ele tri al ee t. Therst working galvani element and thus the rstbattery waspresentedintheformof thevoltai pile in1800 by Alessandro Volta. A voltai ell is an ele tro hemi al energy storage means and an energy onverter. During dis harge, the stored hemi al energy is onverted into ele tri energy via ele tro hemi al redoxrea tion.

Re hargeable batteries an be lassied into the following types : lead-a id battery, ni kel- admiumbattery,ni kel-metalhydride(Ni-MH)batteryandlithium-ion(Li-ion)battery.Among them,lead-a idbatterieshavearelativelyhighenergystorage apa ityandlow ost,thuso upy adominantpositioninthere hargeablemarket.Ni kel- admiumbatteriesarerelativelyold,and have relatively small apa ity. They have a memory ee t that need to fully dis harge before

electrolyte

bridge

anode

cathode

load

e

−

e

−

Figure 1.2S hemati ofstru ture and operation me hanismofbatteries

Ni-MH batteries. Ni-MH batteries are a new generation of re hargeable batteries. They have veryhighenergystorage apa ities,lightweight, andalmostnomemoryee ts.Theyhave been applied in hybrid ele tri vehi les.Li-ion batteries have very good appli ation prospe ts. They have a high voltage, high apa ity,and no memory ee t.The self-dis hargeis only 0.5% - 2% per month, while Ni-MH is around 25%. Li-ion batteries have been used in portable ele troni devi es. As the produ tion te hnology be omes in reasingly sophisti ated, Li-ion batteries are expe tedto repla eNi-MH and be usedason-board batteries inthenearfuture[Re ℄.

There areseveral riteria of hoosing batterysystems,in ludingself-dis hargerate, hargetime, energystorage apability, ostand y lelife,et .Basedonanoverall onsideration,inourstudy, a

100V

lithium-ionbatteryis hosentosupplyenergy totheele tri ma hineusedinour hybrid ele tri vehi le.

1.2.2.2 Advantage and disadvantages

Batteryhasrelativelyhighspe i energybutrelativelylowinspe i power.Thepowerresponse is faster than fuel ells, but slowerthan super apa itors. Batteries an be used inboth ele tri vehi les and hybrid vehi les. Being dierent from fuel ells, batteries ontain their hemi al materials inside, and thus they have a limited lifetime. A automotive SLI battery may last around6years.Whenbatteriesrea htheirlifelimit,itisne essarytowelldisposeof.Otherwise, the toxi materials in batteries may ause damage to the environment. Hen e, battery is one form ofele troni waste(e-waste).

Consequently, it is ne essary to ontrol battery systems appropriately so as to maximize their lifetime.Inhybridvehi leappli ations,transientpeakpowerloaddemandisverylikelytoredu e batteries lifetime.Asaresult,itisne essarytolimitthebattery urrentslope.Toa hievethis,a super apa itor anbeaddedtosupplythetransientpeakpowerand apturethebrakingpower.

1.2.3 Ultra apa itors

1.2 Energystoragesystems 11

anode

separator

_cathode

electrolyte

Figure1.3 S hemati ofstru ture andoperation me hanism ofsuper apa itors

energy re overyand burst-mode powerdelivery devi e.

1.2.3.1 Working prin iple and hara teristi s

Super apa itorswererstdevelopedbyGeneralEle tri (GE)engineersin1950s[JH07℄. Super- apa itorisalso alledultra apa itoror ele tri double-layer apa itor(EDLC).Figure1.3gives a stru tureofthe prin iple ofsuper apa itors. Comparing to apa itors, theanode and athode ofsuper ap itorshavemu hhighersurfa earea,andthus ana umulate mu hmore hargesat the surfa es. Moreover, the arbonele trodeshave highenergy storage apa ity.Being dierent fromfuel ellsandbatteries,theenergystoragepro essofsuper apa itorsisnotthrough hemi- al rea tions,and thus isreversible.Asaresult, super apa itors an be repeatedly hargedand dis hargedhundreds ofthousands of times.

Super apa itors exhibit a mu h longer lifetime than batteries. Generally, super apa itors do not rely on hemi al hanges in the ele trodes. Their lifetime depends mainly on the rate of evaporation ofthe liquidele trolyte. Thisevaporation dependsonthetemperature.Asaresult, the higherthetemperatureis, thefasterthe evaporation goes,andthus,theshorterthelifetime will be.

1.2.3.2 Advantage and disadvantages

Super apa itorshavealarge apa itan edensityandverysmalltime onstantsduetotheirvery lowinternalresistan es.Therefore,theyhaveadvantagesinappli ationswherealargeamountof energy inarelativelyshortperiodisdemandedor whereaveryhighnumberof harge/dis harge y les is needed. The energy density of Super apa itors is only about 10% of the same weight batteries, while their power density is about 10 to 100 times greater [KL10℄. Ultra apa itors have therefore mu h faster harge and dis harge y les. They also tolerate many more harge and dis harge y les than batteries. However, ultra apa itors alone annot supply load power requirement due toits self-dis harge property.

Chargetime Dis harge Time Spe i energy (Wh/kg) Spe i po-wer(W/kg) Life Cy le Batteries 1

∼

5 hrs 0.3

∼

3hrs 10

∼

100

< 1000

<1000 Ultra apa itors 0.3

∼

30 s 0.3

∼

30s 1

∼

10

< 10, 000

> 500, 000

Capa itors

10

−

3

_∼10

−

6

s

10

−

3

_∼10

−

6

s

< 0.1

< 100, 000

> 500, 000

Table 1.1 Chara teristi sof dierent devi es (Sour e:[Zho06℄)

Lithium−Ion

batteries

batteries

Lead acid

batteries

NI metal hybrid

Ultrarcapacitors

0.01

0.1

1

10

100

1000

Fuel cell

Specific power (W/kg)

Specific energy (Wh/kg)

10,000

10

100

1000

capacitors

Electrolytic

capacitors

Double−Layer

Figure 1.4Spe i energyand power ofdierent devi es (Sour e :[Zho06℄)

apa itors aresuitable to be asupplement when arapid loadpowerisrequired. Ultra apa itors anbeusedto apturelargebrakingpowerduringde elerationandprovidetransientpeakpower demand during a eleration. Integrating ultra apa itors to energy storage systems of EVs and HEVs will helpredu e theoverall volume andextend theserving life.Withproper ontrol stra-tegy,this integration would also improve fuele ien ies, redu eexhaust emissions,and extend theele tri driving ranges.

1.2.4 Comparisons of dierent devi es

Table1.1givesa omparison ofthe hara teristi s ofbatteries,ultra apa itors and onventional apa itors in terms of harge/dis harge time, spe i energy/power, and life y les. Figure 1.4 gives a omparison of spe i energy and spe i power of the three energy storage devi es. Spe i energy (Wh/kg) and spe i power (W/kg) are two important indi ators of energy storagedevi es.Theydes ribethe energyandpower apa itiesperunitmassof various devi es. The indi ators are sometimes measured perunit volume, and arereferred to asenergy density

(Wh/L)and power density (W/L).

It anbeseenfromthe gurethatthefuel ellshaverelativelyhighspe i energybutrelatively low spe i power. On the ontrary, the apa itors have high spe i power but low spe i energy. The property of batteries is between them. Consequently, fuel ells and batteries are often used to supplythe main power, while super apa itors are used to meet the rapid power requirements. The hybridization of dierent devi es improves the overall performan e of the

1.3 Hybridizationofenergy storagesystems 13

1.3 Hybridization of energy storage systems

Wehave seemfrom the se tionabove thatfuel ells, batteries andultra apa itors all have their ownadvantagesanddisadvantages.Neitheroneofthem anserveallappli ationsbythemselves. Hybridization of two or three of the energy storage systems an provide a better performan e andmeetwiderenergyrequirements.Forhybridvehi les,hybridizationofenergystoragesystems mayalso ontributeto improve fuele ien y indire tly [KL10℄.

Insteadofusingasignalpowersour e,thehybridizationofmultiplepowersour esmaya hievea ompromisebetweenthespe i powerandthespe i energy.Generally,inthe ongurationof twopowersour es:we hooseonesour ewithhighspe i energyandanotherwithhighspe i power.Theonewithhighspe i energysuppliesthemainenergyrequiredbythesystem.While, theotherone withhighspe i power meetstheloadpower demandinashorttime. Thus,this onguration an provide rapidandadditional powerduringa elerationwhenpeakloadpower is required, and a hieve a braking power re overy during de eleration. Besides, it has a longer y le and servi e life. Moreover, the overall size, volume and ost of the energy storage system an beee tively redu ed.

Inahybrid onguration,theroleoffuel ellsandbatteries isgenerallytosupplymainpowerto theload,whereasultra apa itors dealwiththetransientsandnon-stationary u tuatingsignals duringa elerationand braking.Dierent topologiesofhybrid energysour eshavebeenstudied inthe literature. Thiswill bepresentedinthefollowing se tion.

1.3.1 Congurations of hybrid energy storage system

Ahybrid energystoragesystemis omposedoffuel ellssta k,batterypa ksandultra apa itor modulesoranytwoofthem.[WIAT11,TRD09℄and[ABH11℄giverespe tivelytwo ongurations of hybrid energy storagesystems withfuel ell, batteries and ultra apa itors for hybrid ele tri vehi les. The advantage of three energy storage sour es ombination is that thebattery an be harged bythefuel ell when the loadisrelatively light.

[WIAT11,TRD09℄ propose a dire t parallel onne tion of three energy sour es. As shown in Figure 1.5, the fuel ell, battery and ultra apa itor are onne ted to theDC bus via a DC-DC onverter respe tively.Duetoitssymmetryandexibility,itiseasytorealizeanenergyanalysis and ontrol.Whereas,the number oftheele troni powerdevi es inthis topology ismore than in other topologies, and thus, this leads to a potential in rease of ost. In this onguration, the onverters onne ted to batteries and ultra apa itors are bidire tional onverters allowing a urrent ow in both dire tions, whereas, the onverter onne ted to the fuel ells is a single dire tion onverteravoiding the urrent owinto thefuel ells,inorderto prote t thefuel ells.

Other than a ombination of all three energy storage systems, a hybridization of two energy sour es is more ommon. [GFGJ10℄ resear hes a fuel ell and batteries ombination applied in Metro Centro tramway in Spain. [KP07℄ has studied fuel ell/battery hybrid vehi les. Ho-wever, [Gao05℄ has ompared the performan e of a fuel ell-battery hybrid system and a fuel ell-ultra apa itor hybrid systemandhas on luded thatthelatter, withtheexisten eof ultra- apa itors, shows better performan e, e onomizes more fuels, and meet better the load power demand. Similarly, [BK08℄ hasalsodone a omparative study bybuildingan obje tivefun tion

0

1

00

11

00

00

11

11

0

0

1

1

0

1

0

0

1

1

DC

DC

DC

DC

DC

DC

sta k Fuel ell DCbus Ultra apa itor module Battery pa k

Figure 1.5 Parallel onne tion offuel ells, batteries and ultra apa itors

ell-battery,fuel ell-ultra apa itor,andfuel ell-battery-ultra apa itor vehi les,fuel ell-battery hybridization shownthe poorest performan e.

Comparetothe hybridizationoffuel ellandbattery,the ombination offuel ell/ultra apa itor and battery/ultra apa itor show better performan e and are preferred by most resear hers. In these two ongurations, ultra apa itor ee tively omplementsthe ommon drawba ks of fuel ell andbattery,that is, respond for rapid dynami power demandwill ause severedamage to themselves.Thus, inthis omplement stru ture,fuel ellsor batteries,due to theirhighspe i energy,playaroleofsupplying ontinuoussteadyenergy.Whereas,ultra apa itor,duetoitshigh spe i power, serves as energy sour e during a elerations to satisfythe peak power demand, absorbs energy during de elerations andrealizes braking energyre overy.

There areseveraltopologiesforhybridizationoftwoenergystoragesystems[LMT13℄.Seven dif-ferenttopologiesaredepi ted inFigure1.6.Here,we onsideronlythehybridizationofbatteries andUC.Similar topologies anbeextendedtothehybridizationofanyothertwoenergystorage systems.

In Figure 1.6(a), battery pa k, paralleled with UC bank is dire tly linked to the DC bus. The battery harges the UC and supply power to the DC bus. This topology has an advantage of simpli ity; however, it is la k of exibility sin e it is di ult to ontrol the urrent owing through the battery and the UC, and the voltage on the DC bus is also un ontrollable and alwaysvarieswiththe battery voltage.

Toin reasetheexibility,DC-DC onverters anbeapplied,Figure1.6(b, ,d)arethreedierent as ade topologies, alsoknown asseriestopologies. InFigure1.6(b), abidire tional onverter is added to onne t the UCandthe DC bus.In thistopology,theDC bus voltage anbeboosted to a desired value while it is still di ult to regulate thepower inthe UC sin e it operates at thesamevoltageasthebattery.Therefore,the topologyisoftenreferredtoasapassive as ade stru ture [KL10℄. In Figure 1.6( ) and (d), two onverters are applied, and are often referred to as a tive as ade stru ture due to their ontrol exibility. Comparing to (d), ( ) may boost the DC voltage to a higher leveldue to the two onvertersinterfa ing between the batteryand

1.3 Hybridizationofenergy storagesystems 15

0

1

00

11

0000000

1111111

₀₀

₁₁

0

1

0

0

1

1

00

11

0

1

00

11

0

1

00

00

11

11

0

1

00

11

00

11

0

1

DC

DC

DC

DC

DC

DC

DC

DC

DC

DC

DC

DC

DC

DC

DC

DC

DC

DC

DCbus (a) (b) (d) (e) (f) (g) ( )

00

11

0

0

1

1

DC bus

−

+

S

1a

S

1b

S

2a

S

2b

Figure 1.7Battery and UCsharing one ommon DC-DC onverter

requirement,and thus,redu e the ost.

Other than as ade onne tions, Figures 1.6(e, f,g) showthree dierent hybrid energy sour es topologies with parallel onne tions. (e) and (f), withonly one onverter, aremore simple but less exible than (g). In Figure 1.6(e), a DC-DC onverter onne ts the UC to the DC bus, making it possible to ontrol the power inthe UC, and the DC bus voltage an be maintained at a ertain value without regulation. In Figure 1.6(f), the onverter is linked on the battery side, providing a higher and adjustable voltage in the DC bus. However, neither of these two topologies allows a hieving an optimum power sharing between the batteries and the UC. To a hieve thissoasto extend thelife y leofenergy sour es, thetopology 1.6(g) isproposedand is oftenreferredto asan a tive parallel stru ture.The two energy sour esare onne ted to the DC bus by a onverter respe tively. Manyresear hers have studied this topology and proposed various ontrol strategies.

In order to redu e the ost, size and omplexity of ontrol, resear hers have proposed other topologies to transform the foregoing topologies. [DC03℄ hasproposed to onne t both battery and UC to one ommon DC-DC onverter via two parallel swit hes, asshown inFigure 1.7. In this onguration, the onverter is a two-quadrant DC-DC onverter and works inboost mode when energy sour essupply theload and works inbu k mode for re overing braking energy to re hargethebatteriesandtheUC.The ongurationee tivelyredu esthenumberof onverters. Whereas,it in reases thenumberofswit hesand thus ompli ates ontrol strategy.

[CE09℄haveproposedanothertopology,asshowninFigure1.8.Beingdierentfromthetopology of 1.6(f). A diode is added between thebattery and theDC bus. Depending on theload power demand, the diode isreverse biased whenthe loaddemand islow andforward biased when the load demand is high. Therefore, the battery an be swit hed in to supply more power when a highload poweris required.

Other than the topologies aforementioned, resear hers have also introdu ed galvani isolated onverter into hybrid energy storage system. This is a hieved via repla ing the two indu tors by a transformer. As shown in Figure 1.9 [OK12℄, with an integrated magneti stru ture, two indu tors share one ore. This onguration ee tively redu es the weight and volume of the system. Moreover, due to the indu tor oupling stru ture, the ripples of the input urrent and theoutput voltage an be ee tively an elled, and thesystemresponses to theloadtransients

1.3 Hybridizationofenergy storagesystems 17

0

1

00

00

11

11

+

−

UC

S

1

L

DCbus Battery

S

2

D

Figure 1.8 Batteryand UChybridization withan additionaldiode

00

00

11

11

00

11

00

11

0

1

+

−

S

1

DCbus

S

2

S

4

S

3

UC

L

1

L

2

C

2

C

1

Figure 1.9 Integrated magneti stru ture

1.3.2 Congurations of DC-DC onverters

In terms of galvani isolation, DC-DC onverters are divided into isolated onverters and non-isolated onverters.Isolated onverterisolatestheele tri al onne tionbetweeninputandoutput, andthisisa hievedbyintegratingatransformerintheele tri ir uit.This ongurationisbased on the users' se urity onsideration. Moreover, it has strong noise and interferen e blo king apability, thus provides theloadwitha leanerDCsour e whi h isrequired bymany sensitive loads. Isolated onverters have been used in hybrid ele tri vehi les. An isolated half bridge based DC-DCbidire tional onverter isproposed in[HD08℄ for Plug-in Hybrid Ele tri Vehi le appli ations. In the onverter, the leakage indu tan e of thetransformer is the primary energy transfer element.

However, theexisten e of transformer requiresa de oupling of the indu tan e, and thus makes the ontroller design more ompli ated. Furthermore, asopposed to ommon non-isolated DC-DC onverters, isolated onverters are more bulky, heavier and ost more. Therefore, isolated DC-DC onverters arenot onsidered inourstudy.

Fornon-isolated onverter,asdepi tedinFigure1.10,therearethreemaintopologies[SB03℄.As shown inFigure1.10(a), a basi bidire tional onverter, worksinboost mode whenthe urrent owsforwardsand worksinbu kmodewhenthe urrentowsba kwards. Thistopologyisalso alledhalf-bridge onguration. A ombination oftwohalf-bridgeformsafull-bridgebu k-boost onverter as shown in Figure 1.10(b). In this onguration, the number of a tive omponents doubles, but the ele tri al and thermal stresses of the ir uit be ome lower. Another ommon ongurationisCuk onverter,seeFigure1.10( ).AnadvantageofCuk ir uitisthattheinput and output urrent an be rather smooth without ripples. Therefore, itis quitesuitable in fuel ell appli ations. However, Cuk onverter requires more a tive omponents than the other two

+

−

+

−

+

−

+

−

+

−

+

−

V

out

(b) (a) ( )

L

1

C

L

2

C

2

V

in

C

1

S

1

S

2

C

2

V

out

S

2

S

1

V

in

C

1

S

4

S

3

L

L

S

1

S

2

C

2

V

out

V

in

C

1

Figure 1.10 Topologies ofDC-DC onverters

Other DC-DC onverter topologies in lude parallel interleaved stru ture whi h is a parallel onne tion of swit hing onverter witha shared input and output [BDB

+

14℄;hybrid swit hed- apa itor bidire tional onverterwhi h isa ompositeof apa itorsand swit hes[PMC12℄;and Z-Sour e network onverter whi h isa Z-form ombination of indu torsand apa itors [Pen08℄. In our study,a 4-quadrant bidire tional onverter,as showninFigure 1.10(a), isfavourable for battery/UC hybrid energy sour e systeminhybrid ele tri vehi les.

1.4 Control methodologies

In HEVs,inorderto exploittheenergy storagesystemsproperlysoastoextend their lifetimes, in reaseele tri driverange,improve energye ien y,ande onomizer fuel onsumption,proper energymanagementforenergystoragesystemsisquitene essaryandhasattra tedtheattention of many resear hers. Various ontrol strategies are dis ussed in this se tion. Rule-based power split ontrol; ostfun tionoptimization;wavelet-basedloadsharingalgorithm,et .areproposed onsidering powerdistributionbetweendierentenergy storagesystems.Flatness-based ontrol, polynomial(RST) ontrol, PI ontrol,sliding mode ontrol andpassivity-based ontrol arealso applied onsidering the onverter hara teristi s.

1.4.1 Energy management

1.4 Controlmethodologies 19

Speed (mile/h)

Time (s)

Time (s)

Speed (km/h)

120

20

40

100

60

80

1200

1000

800

600

400

200

0

urban

extra−urban

0

400

800

1200

1600

2000

phase

Hot start

Transient

10

20

30

40

50

60

70

Cold start

phase

phase

(b) FTP−75

(a) New European Driving Cycle

Figure 1.11 Commonly useddriving y les

requirement,batteries need to supplysustainable and steadyenergy to theele tri ma hine, so asto support ontinuouspropulsion. On theotherside,UC needsto be responsiblefor vehi les starts/stops,a elerationandde eleration.Spe i ally,UCisin hargeofsupplyingpeakpower during a eleration and absorbing braking power during de eleration. At the same time, it is important to maintain the state of harge (SOC) inbatteries and UC, this isree ted interms of their terminal voltage. Inorderto manage theenergy distribution of dierent energy storage andmaintainthestate of harge,anee tiveandpowerful ontrolstrategyisrequiredtoa hieve ahighperforman e. Overthepastfewde ades,frompowersplittingstrategytopower onverter ontrol, resear hershave madegreat eortsto design su h ahigh-performan esystem.

1.4.1.1 Driving y le

Driving y le is an important standard of automobile operation. A driving y le is a urve des ribing the vehi le velo ity versus time under ertain tra onditions su h as highways or urban roads. There aredierent driving y les proposed by dierent organizations. In general, there are three ountries generating driving y les whi h are Europe (NEDC, ECE15), Japan

(10-15 Mode),and United States (FTP,SC03SFTP, UDDS,US06 andLA92) [TT13℄. Themost

ommonlyused aretheEPA Federal Test Pro edure(FTP-75), and theNewEuropean Driving Cy le (NEDC)des ribingurban driving onditions.Driving y leisusedasa speedreferen ein o-linesimulationtostudythe performan eofpropulsionsystemandtheenergystoragesystem. AsshowninFigure1.11,NEDC istheoreti allyderived,thus its urveisrathersmooth.FTP-75 is measuredfroma real drivingpattern, andthus, itisa non-stationary signal.

1.4.1.2 Controlmethodologies of energy management

Rule-based ontrol isone ofthesupervisory ontrolmethodsoftheoveralllow-level omponents ontrol.Rule-based ontrolisbasedonhumanexpertise,intuitionandheuristi [TT13℄.Therules are designed to share the loaddemanded powerbetween primary and auxiliary energy sour es. The ontroller an be a hieved through fuzzy logi ontrol. Rule-based ontroller is favourable due to its simpli ity and possibility to implement onboard for real-time ontrol. However, it

Another supervisory ontrol method is optimization approa h, whi h is realized through opti-mizing an obje tive fun tion.Optimization approa h,whi h is generally exploited to maximize thee ien y onthe powertrain andminimizethelosses, an beextendedto managetheenergy distribution for hybrid energy system. Optimizationapproa h in ludes linearprogramming, dy-nami programming (DP), sto hasti DP,et . Su han approa h is basedona prioriknowledge ofthepowerdemandsoveradriving y le,andthustheresultsarenotne essarilyoptimalwhen it isimplementedon-line.

[SSL14℄ presents a omparison of a predi tive ontroller, a dynami programming algorithm, and a rule-based strategy for vehi ular batteries/UC hybrid system. Referring to the study, all the three methods ee tively redu e battery wear, and thus extend its lifetime. Moreover, the designed dynami programming algorithm shows the best performan e among these three methods.

Themainideaofenergymanagementforhybridenergysystemisthattheauxiliaryenergysour e UCshould apturethedynami omponentsoftheload urrent whiletheprimaryenergysour e operates at a steady urrent level. Then the obje tive is to get the urrent referen e for UC. Generally, this an be simply obtained by passing the load urrent through a high-pass lter. The lteredhighfrequen y urrent isthen the urrent referen efor UC. However, simple lters mayleadtoadelayandthus theUCmaynotabsorbthe peakpowerpromptly.[AS10℄proposes a low-phase-shiftlter to redu ethe phaseshift introdu ed bythe lter.

[UA08℄proposesawavelet-basepowersharingalgorithm forfuel ell/UChybridvehi ularpower system. Theauthors utilizeADVISOR,an ee tiveanalysistoolfor vehi lesimulation, togeta requiredpowerdemandprole orrespondingtothevehi ledriving y le,andthenapplywavelet transforms to the prole to lter the transient sharp peak powerdemands. The obtained peak powerdemandis then hosen asthe powerreferen efor UC.

[PPMT08,PPMTD11℄ proposea atness-based ontrol for energy management of fuel ell/UC hybridpowersour e.Thepaperprovesthatthenonlineardierentialequationsdes ribinghybrid energy storage system have atness property. Hen e, all states and inputs of the system an be expli itly expressed in terms of the at output and a nite number of its derivatives. The onsidered output of the system is the energy stored in the DC-bus apa itor. The system is ontrolledbyplanningthedesiredreferen etraje tories ontheatoutputspa e,andimplement state feedba k regulators to for ethe statesto followtheir referen es.

1.4.2 Power onverter ontrol

Ifwe ould saythat energy management isatop-level ontrol,thenthepower onverter ontrol ouldbereferredtoasalow-level ontrol.Energymanagementisa hievedthroughpower onver-ter ontrol. Top-level ontrol alone is not enough, it givesonly an general overview. The study of power onverters isrelatively developed. Various ontrol theorymethods havebeen proposed byresear hers.

1.4.2.1 Linear ontrol

1.4 Controlmethodologies 21 Voltage ontroller ontroller Current onverter DC/DC

V

ref

V

I

ref

I

u

Figure1.12 Cas ade ontrol stru ture for power onverters

-V

I

+

R(z

−1

₎

1

S(z

−1

₎

B(z

−1

₎

A(z

−1

₎

I

ref

T

(z

−1

₎

Figure1.13 RST onverter ontrol stru ture

an beregulatedthrougha as adestru turewitha urrentinnerloopandavoltageouterloop. The urrent and voltage ontrollers an be simply realized with a proportional term and an integral term.

[CGG

+

10,CDG12℄ propose a polynomial ontrol strategy,also known as RST digital ontrol, for voltage and urrent management of battery/UC system. The prin ipal ontrol stru ture is presentedinFigure1.13. Thekeypoint ofthe ontrolstrategy isto determinate

R(z), S(z)

and

T (z)

polynomials.Where,

R(z), S(z)

arededu edfromDiophantine equation. Thisalgorithm is robust and easyto implement ina mi ro ontroller.

1.4.2.2 Sliding mode ontrol

Other than Pulse-width modulation (PWM) ontrol and peak urrent mode ontrol, sliding mode ontrolgivesanothermethodto ontrolpowerele tri swit hes.Beingdierentfromother ontrollers, sliding mode ontroller doesn't give a ontinuous duty y le signalfor swit hes but a dis ontinuous ontrol signalwhi h an be dire tlyusedto ontrolswit hes. Consequently,this propertyleadstoastipushingmotiontothesystemandresultsin hatting problem.Thebasi ideaofslidingmode ontrolistondtheslidingsurfa e(seeFigure1.14)andfor ethesystemto movealong thesurfa e until rea hingtheequilibrium point. Thesliding mode alongthesliding surfa e

s = 0

isexist only whenthe ondition

s ˙s < 0

is satised[UGS99℄.

For power onverter ontrol, sliding mode ontrol is often utilized to ontrol an inner urrent loopina as ade ontrol stru tureto onverge the urrent to its referen e[UGS99℄.[ABDM07℄ has proposed to apply sliding mode ontroller to energy management of fuel ell/batteries/UC hybridenergy storagesystem.A as ade ontrol ongurationwithvoltage ontrolasouterloop

0

0

1

1

0

0

1

1

Equilibrum point Slidingsurfa e

s = 0

x

˙

x

Figure 1.14 Slidingsurfa e andthe systemmovement

onverter urrent and the UC onverter urrent. The outer voltage loop is ontrolled by a PI linear ontroller.

In[DCC13b℄,theslidingmode ontrollerisappliedtoasuper apa itor onne tedDC-DC onver-ter to absorb the urrent disturban e in theDC bus. The sliding surfa e is theerror dieren e between the urrent anditsreferen e.The urrentreferen eisnotasimplenominalvalue,but a varyingmanifoldaimingtoabsorbthe urrentdisturban e.Thedisturban eabsorptionobje tive is easily a hieved withsliding mode ontrol. Moredetails an bend inthereferen e.

1.4.2.3 Passivity based ontrol

Passivity-based ontrol,rstproposedbyOrtegaattheendof20th entury,dealingwithsystem total energy,isaquitepowerful nonlinear ontrolmethod.Consider asystemwithstates

x ∈ R

n

input

u ∈ R

m

and output

y ∈ R

m

, ifthere exists a non-negative storage fun tion

H(x)

whi h an bewritten inthe following form, thenthe systemissaid to be passive.

H[x(t)] − H[x(0)] ≤

Z

t

0

u

⊤

(s)y(s)ds

(1.1)

The inequality an berewritten as:

H[x(t)] − H[x(0)]

|

{z

}

storedenergy

=

Z

t

0

u

⊤

(s)y(s)ds

|

{z

}

suppliedenergy

−

d(t)

|{z}

dissipatedenergy (1.2)

Where, theleft sideof theequationrepresentstheenergy storedinthesystem, andtheintegral term ontheright sideistheenergy suppliedto thesystem,and

d(t) ≥ 0

isthedissipatedenergy of the system.

Let the ontrol input

u(x) = β(x(t)) + v(t)

,andmake thefollowing assumption :

−

Z

t

0

1.5 Con lusions 23

Where,

n

0

is a onstant of integration

n

0

= H

a

(x(0))

. Substitute

u

into (1.2) , and after some manipulation, we mayget :

H

d

(x(t)) − H

d

(x(0)) =

Z

t

0

v(s)

⊤

_{(y(s))ds − d(t)}

(1.4)

with energy fun tion

H

d

(x(t)) = H(x(t)) + H

a

(x(t))

being losed-loop energy. It an be seen from(1.4)thatwiththe new ontrolinput

v(t)

,the losed-loopsystemisstillpassive.Moreover, if

H

d

(x)

hasastri t(lo al)minimumat

x

∗

,then

x

∗

isanequilibriumpoint,and

H

d

(x)

de reases while

x

onverges to

x

∗

asymptoti ally.Thisisthe prin iple of PBC.

Passivity based ontrol is designed via inter onne tion and damping assignment, based on an Euler- Lagrange model or a Hamiltonian model [OvdSME02℄. Passivity-based ontrollers have beenstudied to ontrolaboosttypeand abu k-boosttypeDC-DC onverter in[SRO95℄.Ithas been proved thatan outputvoltage dire t ontrol leads to anunstable ontroller, and thus, the ontroller is based on an indire t urrent ontrol. PBC has also been studied in hybrid energy storage systems. [ABH

+

10℄ applies PBC to a DC-DC onverter as ade onguration for fuel ell/UC hybrid energy sour e. The system is written as a port- ontroller Hamiltonian system, and the ontroller isdesigned bypreservingthesystempassivity.

1.5 Con lusions

Inthis hapter,we have summarizedtheprin iplesand hara teristi soffuel ells,batteriesand UC, andanalyzedtheir rolesinEVs andHEVs.Fuel ellsandbatteries,duetotheir highenergy density,areoftenusedtosupplysteadyenergytotheload.Fuel ells,primaryenergysour e,are oftenappliedinele tri alvehi les.Batteriesworkasprimarysour esmainlyinHEVs.UC,dueto itshighpowerdensity,isexploitedto apturethebrakingpowerandsuppliesinstantaneouspeak powerdemand. Hybridization of two or three of them provides a better energy storage system. Moreover, we have shown dierent hybridization stru tures and various topologies of DC-DC onverters in vehi ular appli ations. Among them, a parallel batteries/UC onguration with bidire tionalDC-DC onvertersis hoseninourstudyduetoitsexibility.Furthermore,inorder to ontrol hybrid energy storagesystems,andproperlymanage theenergy distributionbetween energystoragesystems,aseriesof ontrolstrategiesand ontroltheorymethodshavebeenbriey presented.However,alltheseresear hesaimtodealwithinstantaneouspeakpower.Inourstudy, wewillfo usnotonlyoninstantaneouspeakpowerbutalsoonsinusoidaldisturban esintrodu ed byICEtorqueripple ompensation,and this will be elaborated inthefollowing hapter.

Chapitre 2

Disturban e reje tion theory and appli ation

Sommaire

2.1 Introdu tion . . . 26 2.2 Disturban esin ele tri alpart ofHEVs . . . 26 2.2.1 Me hani alequations . . . 26 2.2.2 Ele tri alequations . . . 27 2.2.3 Currentdisturban esin DCbus . . . 28 2.2.4 Controlobje tives . . . 29 2.3 Disturban ereje tion theories . . . 30 2.3.1 Linearsystems . . . 31 2.3.2 Nonlinearsystems . . . 34 2.3.3 Hamiltoniansystems . . . 37 2.4 Theoryappli ation. . . 39 2.4.1 Batteryside onverter . . . 40 2.4.2 Ultra apa itorside onverter . . . 42 2.4.3 Simulationandresults . . . 48 2.5 Con lusions . . . 49