Você está na página 1de 165

Eunice de Ftima Fragoso Ribeiro

Fault Diagnosis in Energy Systems for Telecommunications


Eunice de Ftima Fragoso Ribeiro

Fault Diagnosis in Energy Systems for Telecommunications

Doctor of Philosophy Thesis in Electrical and Computer Engineering supervised by Professor Antnio Joo Marques
Cardoso and presented in the Electrical and Computer Engineering Department of the Faculty of Sciences and
Technology of the University of Coimbra

September 2013
UNIVERSIDADE DE COIMBRA
UNIVERSITY OF COIMBRA

Fault Diagnosis in Energy Systems for Telecommunications

Eunice de Ftima Fragoso Ribeiro

Thesis submitted for the fulfillment of the requirements for the degree
Doctor of Philosophy in Electrical and Computer Engineering

Supervisor: Professor Antnio Joo Marques Cardoso


University of Beira Interior

2013
This work was supported by the Portuguese Foundation for Science and Technology (FCT)
under the grant number SFRH/BD/47741/2008.
Acknowledgments

My PhD was a crazy roller coaster ride with many emotional ups and downs. Through-
out it all, I have been very fortunate for having by my side many people without whom
this work would not have been possible. Therefore, I would like to acknowledge my debt
to all those who took part in this ride for helping me.

First and foremost I would like to express my utmost gratitude to my supervisor, Pro-
fessor Antnio Joo Marques Cardoso, for his guidance, remarkable patience and constant
support throughout my PhD work. I have benefited greatly from his great experience and
background knowledge. Just like Isaac Newton once wrote: if I have seen further it is by
standing on the shoulders of giants. I will always be thankful for the great opportunities
he provided me which helped to shape who I am as a researcher and person.

I would like to extend my heartfelt gratitude to Professor Chiara Boccaletti for her
support, kind availability and suggestions to this work. Her contributions have been of
great value for the success of my research.

My deep appreciation goes to my laboratory partners Sylvie, Jorge and Nuno for their
support, good advice and friendship. Their collaboration was extremely important to sus-
tain my motivation during my PhD experimental work as they always contributed for a
stimulating work environment in the laboratories we have shared.

My gratitude is also extended to my parents and my little sister for their endless
support and unconditional love. Without them, I would not have had the courage to
accept such a challenging task as a PhD. They have always been my driving force in life
and I owe them everything.

A word of gratitude also goes to my friend Ana Raquel Soares for her support in difficult
times and for all the fun we had together.

I would like to address a special word of thanks to my boyfriend Lus Picciochi. His
sense of humor and faith in myself have always made all the stress and problems disappear.
During the latest and most difficult years, his support and affection have granted me the
strength that enabled me to complete this work.

Finally, I would like to thank the Portuguese Foundation for Science and Technology

i
(FCT) for their financial support under the grant number SFRH/BD/47741/2008.

Eunice Ribeiro

ii
Abstract

Telecommunications networks are expanding for remote locations with the increased
dependency on new mobile communication technologies. Therefore, high reliability and
reduced energy consumption costs are mandatory targets for the telecommunications com-
panies. Alternative and grid independent energy solutions based on renewable energies are
very attractive to accomplish those targets. The main goal of this work is to develop fault
diagnostic methods for the power interfaces applied to telecommunications power supplies
based on renewable energies and energy storage devices technologies.

An in-depth research though small and medium wind and photovoltaic (PV) power
conversion systems and the DC telecommunications power supply requirements identified
non-isolated DC-DC converters as key power conversion interfaces, because they optimize
the renewable energies generation. Therefore, four different DC-DC converters have been
selected for interfacing PV-batteries and wind-batteries power systems, namely unidirec-
tional single-ended DC-DC converters, three-phase interleaved DC-DC converters, three-
level boost DC-DC converters and bidirectional half-bridge DC-DC converters.

Photovoltaic arrays, wind generators and batteries models presenting the best accu-
racy and computational effort compromise for electrical circuits simulation were chosen for
simulating power supply systems including generators, storage devices and power inter-
faces. Using these simulation models, theoretical studies were performed presenting a good
agreement with experimental results.

The mentioned DC-DC converters under healthy condition and with introduced power
switch open-circuit faults are comprehensively studied. Both analyses provide the required
data for developing four novel fault diagnostic techniques for the mentioned DC-DC convert-
ers. The presented methods are robust against transients and use only the control variables
without requiring extra sensors. They mainly use basic mathematical calculations which
results in a quick processing time and low computational requirements. Therefore, they
permit to easily improve the DC-DC converters reliability with a small effort.

Finally, as soon as the fault is detected, fault-tolerant strategies must be used to avoid
the DC-DC converters stoppage. The interleaved DC-DC converter is the only topology
with inherent fault tolerance, although its performance is greatly reduced. The unidirec-

iii
tional single-ended DC-DC converters and the bidirectional half-bridge DC-DC converters
have a reduced number of components, which makes them attractive for fault-tolerant
strategies based on redundancy. A novel reconfiguration fault-tolerant strategy for the
three-level boost converter applied to photovoltaic applications is proposed. It requires
only a few components added to the original topology, so that, under an open-circuit power
switch fault, it can be partly reconfigurated into a two-level boost converter ensuring energy
supply.

Keywords

Standalone Power Systems, Renewable Energies, Power Electronics Converters, Fault Di-
agnosis.

iv
v
vi
Resumo

As redes de telecomunicaes esto a expandir-se para locais remotos face ao aumento


do uso das mais recentes tecnologias de comunicao sem fios. Como tal, as empresas de
telecomunicaes tm como principais objectivos reduzir os custos do consumo de energia
dos seus equipamentos bem como melhorar a sua fiabilidade. Novas solues alternativas
de sistemas autnomos de alimentao recorrendo a energias renovveis so actualmente a
opo mais atractiva para atingir estes objectivos. No presente trabalho desenvolveram-se
mtodos de diagnstico de avarias para os conversores de electrnica de potncia aplicados
como interface em sistemas de alimentao que renem energias renovveis e componentes
de armazenamento de energia.

Um estudo aprofundado de sistemas de alimentao elicos e fotovoltaicos, bem como


de sistemas de energia DC para telecomunicaes, permitiu constatar que os conversores
DC-DC no isolados so um componente importante de interface que permite optimizar a
produo das energias renovveis. Por este motivo, quatro conversores DC-DC diferentes
foram escolhidos para estes sistemas, nomeadamente conversores DC-DC no isolados com
uma s sada e unidireccionais, conversores DC-DC entrelaados com trs fases, conver-
sores DC-DC de trs nveis para aumentar a tenso e conversores DC-DC de meia ponte
bidireccionais.

Para concretizar os estudos tericos de simulao dos sistemas de alimentao que


incluem geradores elicos ou fotovoltaicos, armazenamento de energia e conversores de
electrnica de potncia, foram escolhidos modelos de painis fotovoltaicos, geradores elicos
e baterias conjungando da melhor forma o compromisso entre a preciso dos resultados e
o esforo computacional exigido. Os resultados obtidos em simulao foram validados com
sucesso pelos resultados obtidos experimentalmente.

O funcionamento normal e em avaria de circuito aberto de um dos semicondutores dos


conversores DC-DC mencionados foi extensivamente estudado. Ambas as anlises permi-
tiram desenvolver quatro novos mtodos de diagnstico de avarias para estes conversores.
Estes mtodos so robustos face ocorrncia de transitrios e usam apenas as variveis
de controlo, no tendo a necessidade de incluir mais sensores no conversor. Em grande
parte, os clculos utilizados so triviais e, uma vez que so necessrios poucos recursos
computacionais, o tempo de processamento diminuto. Com reduzido esforo, os mtodos

vii
desenvolvidos permitem facilmente melhorar a fiabilidade dos conversores DC-DC.

Por fim, assim que a avaria detectada, estratgias de tolerncia a falhas podem
ser aplicadas aos conversores DC-DC para evitar a interrupo do seu funcionamento. O
conversor DC-DC entrelaado o nico que possui uma intrnseca tolerncia a falhas,
apesar de, aps a ocorrncia de uma avaria, o seu desempenho ser reduzido. Os conversores
DC-DC no isolados com uma s sada e unidireccionais e de meia ponte bidireccionais
so compostos por um nmero reduzido de componentes, o que os torna muito atractivos
para estratgias de tolerncia a falhas baseadas em redundncia. apresentada uma nova
estratgia de reconfigurao do conversor DC-DC de trs nveis elevador de tenso para
aplicaes fotovoltaicas. Esta requer apenas que sejam adicionados alguns componentes
constituio original do conversor DC-DC de trs nveis para que, aps a ocorrncia de
uma avaria, este seja reconfigurado num conversor DC-DC de dois nveis para assegurar a
continuidade do seu funcionamento.

Palavras Chave

Sistemas de Energia Autnomos, Energias Renovveis, Conversores de Electrnica de Potn-


cia, Diagnstico de Avarias.

viii
ix
x
Contents

Acknowledgments i

Abstract iii

Resumo vii

Contents xi

Nomenclature xv

1 Introduction 1
1.1 Outline and Contributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

2 Generators and Batteries Mathematical Models 7


2.1 PV Generator Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.1.1 Basic Features of Solar Cells . . . . . . . . . . . . . . . . . . . . . . 7
2.1.2 An Overview of PV Models . . . . . . . . . . . . . . . . . . . . . . . 10
2.1.3 PV Array Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.1.4 Model Validation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.2 Wind Generator Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.2.1 Basic Features of Wind Energy Conversion Systems . . . . . . . . . 15
2.2.2 An Overview of PMSG Models for Wind Energy Conversion Systems 16
2.2.3 PMSG Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.3 Battery Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.3.1 Basic Features of Batteries . . . . . . . . . . . . . . . . . . . . . . . 19
2.3.2 An Overview of Battery Models . . . . . . . . . . . . . . . . . . . . . 20
2.3.3 Battery Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.3.4 Model Validation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

3 Telecommunications Power Systems 25


3.1 Renewable Energy Based Power Systems . . . . . . . . . . . . . . . . . . . . 26
3.1.1 Small Wind Turbines Power Systems . . . . . . . . . . . . . . . . . . 26
3.1.2 PV Power Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

xi
3.2 Power Conditioning Interfaces . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.2.1 Unidirectional Single-Ended and Bidirectional Half-Bridge DC-DC
Converters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
3.2.1.1 Wind Power System Simulation Model and Experimental
Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
3.2.2 Interleaved Boost DC-DC Converter . . . . . . . . . . . . . . . . . . 32
3.2.2.1 PV Power System Simulation Model and Experimental Setup 33
3.2.3 Three-Level Boost DC-DC Converter . . . . . . . . . . . . . . . . . . 35
3.2.3.1 PV Power System Simulation Model and Experimental Setup 36

4 Fault Analysis of Non-Isolated DC-DC Converters 39


4.1 Single-Ended Unidirectional DC-DC Converters . . . . . . . . . . . . . . . . 40
4.1.1 Healthy and Faulty Operating Principle . . . . . . . . . . . . . . . . 40
4.1.2 Simulation Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
4.1.3 Experimental Results . . . . . . . . . . . . . . . . . . . . . . . . . . 42
4.2 Interleaved DC-DC Converters . . . . . . . . . . . . . . . . . . . . . . . . . 43
4.2.1 Healthy and Faulty Operating Principle . . . . . . . . . . . . . . . . 43
4.2.2 Simulation Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
4.2.3 Experimental Results . . . . . . . . . . . . . . . . . . . . . . . . . . 50
4.3 Three-Level Boost DC-DC Converters for PV Applications . . . . . . . . . 52
4.3.1 Healthy and Faulty Operating Principle . . . . . . . . . . . . . . . . 52
4.3.2 Simulation Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
4.3.3 Experimental Results . . . . . . . . . . . . . . . . . . . . . . . . . . 57
4.4 Bidirectional Half-Bridge DC-DC Converter . . . . . . . . . . . . . . . . . . 58
4.4.1 Healthy and Faulty Operating Principle . . . . . . . . . . . . . . . . 58
4.4.2 Simulation Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
4.4.3 Experimental Results . . . . . . . . . . . . . . . . . . . . . . . . . . 63

5 Fault Diagnostic Methods for Non-Isolated DC-DC Converters 65


5.1 Fault Diagnosis in Single-Ended Unidirectional DC-DC Converters . . . . . 66
5.1.1 Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
5.1.2 Simulation Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
5.1.3 Experimental Results . . . . . . . . . . . . . . . . . . . . . . . . . . 68
5.2 Fault Diagnosis in Interleaved DC-DC Converters . . . . . . . . . . . . . . . 69
5.2.1 Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
5.2.2 Simulation Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
5.2.3 Experimental Results . . . . . . . . . . . . . . . . . . . . . . . . . . 76
5.3 Fault Diagnosis in Three-Level Boost DC-DC Converters for PV Applications 81
5.3.1 Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
5.3.2 Simulation Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
5.3.3 Experimental Results . . . . . . . . . . . . . . . . . . . . . . . . . . 85

xii
5.4 Fault Diagnosis in Bidirectional Half-Bridge Converters . . . . . . . . . . . 85
5.4.1 Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
5.4.2 Simulation Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
5.4.3 Experimental Results . . . . . . . . . . . . . . . . . . . . . . . . . . 91

6 Fault Tolerant Strategies Applied to Non-Isolated DC-DC Converters 95


6.1 Interleaved DC-DC Converter Inherent Fault Tolerance . . . . . . . . . . . 96
6.2 Three-Level Boost DC-DC Converter Reconfiguration Strategy for PV Ap-
plications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
6.2.1 Simulation Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
6.2.2 Experimental Results . . . . . . . . . . . . . . . . . . . . . . . . . . 99
6.3 Energy Storage Systems Modularity and/or Redundancy . . . . . . . . . . . 101
6.3.1 Simulation Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
6.3.2 Experimental Results . . . . . . . . . . . . . . . . . . . . . . . . . . 103

7 Conclusions and Future Work 105


7.1 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
7.2 Recommendations for Future Work . . . . . . . . . . . . . . . . . . . . . . . 107

A Simulation Parameters 109


A.1 PV Module BP 4175B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
A.2 PMSG . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
A.3 Battery . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
A.4 Electrical Components . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110

B Experimental Setup Data 113


B.1 PV Module BP 4175B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
B.2 PMSG . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
B.3 Battery . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
B.4 Electrical Components . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116

Bibliography 119

xiii
xiv
Nomenclature

BattI, BattV Wind-battery system battery packs

CCM Continuous conduction mode

DC Direct current

DCM Discontinuous conduction mode

DSP Digital signal processor

HV bus High voltage side of a bidirectional non-isolated DC-DC


converter

IGBT Insulated gate bipolar transistor

LV bus Low voltage side of a bidirectional half-bridge DC-DC con-


verter

MOSFET Metaloxidesemiconductor field-effect transistor

MPP Maximum power point

MPPT Maximum power point tracking

NOCT Nominal operating cell temperature

P&O Perturb and observe MPPT algorithm

PMSG Permanent magnet synchronous generator

PV Photovoltaic

rpm Revolutions per minute

SOC Battery state-of-charge

STC PV modules standard test conditions

xv
SC Supercapacitor

TRIAC Triode for Alternating Current

ZVS Zero-Voltage-Switching

A Exponential area amplitude (V)

aK , bK , cK , dK Auxiliary fault diagnostic variables

Aw Area (m2 )

a1 , a2 PV model ideality diodes factor


 
B Exponential area time constant inverse (Ah)1

cp Power coefficient

C1, C2 Capacitors

Cond Auxiliary fault diagnostic variable

D Duty-cycle

dA , dA1 , dA2 , dB , dB1 , dB2 Auxiliary fault diagnostic variables

Dp Damping coefficient (Nms/rad)

DM Fraction of period of time for which the current is differ-


ent from zero while the power switch is OFF in a DC-DC
converter operating in DCM

DX Input current Iin derivative sign during interval X

D1, D2 Diode

E Battery open-circuit voltage (V)

EExp Battery voltage at the end of the exponential discharge area


(V)

EF ull Battery fully charged voltage (V)

EG0 Band gap extrapolated to the absolute zero

EN om Battery voltage at the end of the nominal discharge area


(V)

xvi
E0 Battery constant voltage (V)

F1 , F2 , F3 Fault diagnostic variables

FQ , FQ1 , FQ2 Fault diagnostic variables

FI1 , FI2 Fault diagnostic variables

FV 1 , FV 2 Fault diagnostic variables

Gn STC solar irradiation (W/m2 ))

h Fault diagnostic threshold

H, PV temperature indenpendent constants

I Power converter inductor current (A)

Ibatt Battery output current (A)

IBattI , IBattV Bidirectional half-bridge converters inductors currents (A)

id ,iq Stator phase currents in the dq synchronous rotating refer-


ence frame (A)

Idec , Idec1 , Idec2 Current sampled while the gate command signal is decreas-
ing (A)

IEV , IEV 1 , IEV 2 Auxiliary fault diagnostic variables

Ii Inductor current sample i

Iin Interleaved boost converter input current (A)

Iinc , Iinc1 , Iinc2 Current sampled while the gate command signal is rising
(A)

Ik Interleaved DC-DC converter ohase k current (A)

io , iout Converter output currents (A)

iL Load current (A)

ILAM ,ILAM 1 , ILAM 2 Auxiliary fault diagnostic variables

Ilow , Ilow1 , Ilow2 Auxiliary fault diagnostic variables

IM Variable calculated using (4.9)

xvii
Impp PV MPP current (A)

Iph PV light-induced current (A)

Iph,n PV light induced current at STC (A)

Ipv PV output current (A)

iP V 1 , iP V 2 PV modules output currents (A)

IW Buck inductor current (A)

Imppn PV module MPP current at STC (V)

Isc PV module short-circuit current (A)

Iscn PV module short-circuit current at STC (A)

I01 , I02 PV diode saturation currents (A)

I1 Interleaved DC-DC converter phase 1 current (A)

I2 Interleaved DC-DC converter phase 2 current (A)

I3 Interleaved DC-DC converter phase 3 current (A)

J Moment of inertia (kg.m2 )

Boltzmann constant m2 .kg.s2 .K1



k

K Polarization voltage (V)

KI Current temperature coefficient (A/K)

KV Voltage temperature coefficient (V/K)

L, L1 , L2 Power converter inductances (H)

Ld , Lq Stator inductances in the dq synchronous rotating reference


frame (H)

MX Auxiliary fault diagnostic variable

n Interleaved boost converter phases number

N Number of samples

xviii
Nd Total number of samples of the converter inherent switching
delay

Ns Series connected cells in a PV module

N npX Total number of non-positive samples of the input current


Iin derivative during interval X

N psX Total number of samples of the input current Iin derivative


during interval X

N pX Total number of positive samples of the input current Iin


derivative during interval X

p Pole pairs number

P , P 1, P 2, P 3 Switching commands

Pmax,e Maximum experimental power from the PV module manu-


facturer datasheet (W)

Pmax,m Maximum power calculated by the PV model used (W)

Ppv PV output power (W)

P V 1, P V 2 PV modules

P 1I, P 2I,P 1V ,P 2V Gate command signals

P 1P 2P 3 Switching commands states: P1 ON, P2 ON, P3 ON

P 1P 2P 3 Switching commands states: P1 OFF, P2 OFF, P3 ON

P 1P 2P 3 Switching commands states: P1 ON, P2 OFF, P3 ON

P 1P 2P 3 Switching commands states: P1 ON, P2 OFF, P3 OFF

P 1P 2P 3 Switching commands states: P1 OFF, P2 ON, P3 ON

P 1P 2P 3 Switching commands states: P1 OFF, P2 ON, P3 OFF

P 1P 2P 3 Switching commands states: P1 OFF, P2 OFF, P3 ON

P 1P 2P 3 Switching commands states: P1 OFF, P2 OFF, P3 OFF

Q, Q1, Q2, Q3 Power switch

xix
R Rotor radius (m)

Rs PV model series resistance ()

Rsh PV model parallel resistance ()

Rsh,min Initial guess for Rsh for the PV module model implementa-
tion ()

Rst Stator winding per-phase resistance ()

q electronic charge (C)

Qc Battery capacity (Ah)

QN om Battery capacity at the end of the nominal discharge area


(h)

Q1I, Q2I,Q1V ,Q2V Power switches

Rbatt Battery internal resistance ()

S1, S2 Power switch

S1Of f , S2Of f Fault diagnostic variables

T Switching period (s)

Tc Sampling period (s)

Td Converter inherent switching delay time (s)

Te PMSG electromagnetic torque (N.m)

tExp Time at which the battery is at the end of the exponential


discharge area (h)

tF ull Time at which the battery is fully charged (h)

TL Load torque (Nm)

tN om Time at which the battery is at the end of the nominal


discharge area (h)

Tu PV cell temperature (K)

Tun STC temperature (K)

xx
va , vb , vc PMSG phase voltages (V)

Vbatt Battery output voltage (V)

vd , vq Stator phase voltages in the dq synchronous rotating refer-


ence frame (V)

Vdc Wind-battery system DC-bus voltage (V)

VHV Voltage of the high voltage side of a bidirectional half-bridge


DC-DC converter (V)

Vin Converter input voltage (V)

VL Power converter inductor voltage (V)

VLV Voltage of the low voltage side of a bidirectional half-bridge


DC-DC converter (V)

Vmpp PV MPP voltage (V)

Vout Converter output voltage (V)

Vpv PV output voltage (V)

vP V 1 , vP V 2 PV modules output voltages (V)

Vt Thermal voltage (V)

vwind Wind speed (m.s1 )

V C1 Capacitor C1 Voltage (V)

V C2 Capacitor C2 Voltage (V)

V mppn PV module MPP voltage at STC (V)

V oc PV module open-circuit voltage (V)

V ocn PV module open-circuit voltage at STC (V)

X Any interval with the same number of ON switching com-


mands

Z Auxiliary fault diagnostic variable

P max Pmax,m and Pmax,e error

xxi
Battery eficciency

Blade pitch angle ( )

e Electrical position of the PMSG rotor (rad)

m Mechanical position of the PMSG rotor (rad)

d ,q Stator flux linkages in the dq synchronous rotating reference


frame (Wb)

f Rotor magnets flux linkage (Wb)

Air density (kg.m3 )

Rotor angular velocity (rad/s)

m PMSG mechanical rotor speed (rad/s)

Tip speed ratio

Tip speed ratio

xxii
Chapter 1

Introduction

Uninterrupted power supply for remote base stations has been a chal-
lenge since the founding of the wireless industry, but alternative sources
have a chance of succeeding where traditional solutions have failed.

By Zhang Hongguan & Zhang Yufeng from Huawei

Technologies brought new means of telecommunication that have been easily accom-
modated in the daily life of society, such as Internet, communications satellites, radio,
networks, mobile phones, among others. They have globally revolutionised not only the
way people communicate, but also their economy, politics, health, social life, safety and
culture. The main social structures and activities are currently established through infor-
mation networks powered by communication technologies which have caused an explosive
growth of mobile internet.

The growing need for wide availability of mobile technologies in great outdoors de-
termined the telecommunications networks expansion for remote sites. In such locations,
the electrical grid might not be easily available and providing its connection can be highly
expensive. It might also lack the quality required by telecommunications apparatus. Addi-
tionally, expanding these networks involves installing more equipment and the subsequent
rising of energy consumption costs. Therefore, technological progress in telecommunica-
tions, and their subscribers great demand, also brought new challenges for their power
supply systems, such as sustainability issues, economical and reliable off-grid energy solu-
tions and high efficiency on both telecommunications equipment and power interfaces.

Alternative and grid independent generation systems are attractive off-grid energy so-
lutions that can make economic sense. There are many possibilities of energy sources for
such systems, but renewable energy generators are especially advantageous for being pollu-
tion free and naturally replenished while complying with environmental concerns. However,
their main drawbacks are related to their intermittence and unpredictable nature and the

1
2 Chapter 1. Introduction

subsequent need for oversizing, leading to a high initial cost. To overcome this, energy stor-
age devices are combined with different renewable energy generators, in order to achieve
a self-sufficient locally limited generator. Such an attractive arrangement is usually called
standalone power system [1].

A standalone power system is not only a mix of different generators types and backup
equipment. It requires high efficiency technology, individual equipment control that takes
advantage of its maximum performance and a robust energy management. Power convert-
ers allow not only to connect different electric devices together (whether they are loads,
generators or storage equipment), but also to provide suitable control for optimizing and
protecting these devices. They ensure uninterruptible operation, safety and long life, which
are basic requirements in telecommunications power systems [2]. They are also responsible
for accomplishing the strict telecommunications equipment requirements, such as nominal
voltage and operational range, noise and acceptable ripple levels [2].

Growing concerns on environmental issues and fuel depletion claim for a more effective
application of clean and alternative generators based on renewable sources. They can
be used as a complement source to grid connected telecommunications power supplies to
reduce their energy consumption costs. Besides their application in off-grid solutions, they
also increase telecommunications power supplies availability during grid outages which are
conventionally bridge by diesel generators and batteries [3]. Unlike renewable energies
technologies, diesel generators are highly pollutive and noisy. Addittionally, they require
constant maintenance and fuel replenishment [4].

There is a plethora of technologies available for efficient energy generation for telecom-
munications, including generators, storage equipment, power converters, among others.
Each telecommunications power supply project should be individually studied, in order to
conceive the most reliable and economically viable solution. Renewable energies sources
power conversion, among the available local potentialities, should be seriuously considered,
particularly the most technologically advanced ones which are wind and solar.

Both wind and photovoltaic (PV) power systems for off-grid DC applications benefit
from a good reliability when compared with other similar energy systems (systems using
diesel generators, for example) [3], but, as in many other power applications, power con-
verters are the most likely components to fail [514]. These systems have been extensively
studied in literature covering aspects like operating performance, optimal sizing, economical
viability, simulation models, power converters, control, but fault diagnostic methods and
fault-tolerance strategies applied to the power converters have been neglected. The devel-
opment of fault diagnostic methods and fault-tolerance strategies for their power converters
improves their availability and increases their application in telecommunications.

The availability of telecommunications power systems is commonly improved through


redundancy of components, including power converters [2]. Their redundancy is typically
1.1 Outline and Contributions 3

provided by running two independent circuits or two independent full power systems, in
order to avoid any service degradation in case of failure [2].

The development of power converters with self-diagnosing ability and capable of keeping
their normal operation after fault occurrence, in spite of performance degradation, is already
a cost-effective practice in many applications. Industrial manufacturers, for example, are
also conscious of the significant losses in production and profits due to faulty operation of
power converters. Therefore, controlled rectifiers and several topologies of inverters applied
to power drives with quite a few different control methods applied, benefit from various
power switch fault diagnostic methods and fault-tolerant strategies [1517] which are cost-
effective when compared to redundancy inherent high costs of duplicating converters.

An excess of electrical and thermal stresses is the main cause for the failure of power
switches [15], which can be classified as open-circuit and short-circuit faults [15]. Short-
circuit fault protection is already a standard practice integrated in most power switches
drives [18]. However, such protections are not able to detect open-circuit faults, whose
causes can be the lifting of the bonding wires during thermal cycling, a driver failure or a
short-circuit fault induced rupture [15]. Although their reliability is a critical issue on all
power converters, only a few fault diagnostic methods and fault-tolerance strategies for DC-
DC converters have been presented in the literature [1930]. Furthermore, these methods
and strategies face some limitations for their application in renewable energies systems.

All considered, PV and wind DC-DC power converters still lack reliability for integra-
tion in DC telecommunications power supply. The main objective of the present work is
to develop open-circuit fault diagnostic methods for power converters applied to DC stan-
dalone power systems based on wind and PV generators and using batteries as storage
component. This requires an in-depth study of the open-circuit fault converter behaviour,
which is also presented in this work. The post-fault operation of these converters is con-
sidered as well. Due to the importance of availability, some conventional fault-tolerant
strategies are presented and a new fault-tolerant reconfiguration strategy is described.

1.1 Outline and Contributions

The content of this work is organised into seven chapters. The first chapter provides
a brief characterization of the current energy challenges faced by the telecommunications
industry in supplying their equipment. The objectives of this work are contextualized on a
present scenery of a high-tech and energy consumer telecommnunications industry, whose
social, cultural and economic impact is very significant on the living society. The renewable
energies power systems emerge as potential solution for their main supply issues, apart from
their known reliability drawbacks.

The main objective of the second chapter is to fully describe and explain the PV array,
4 Chapter 1. Introduction

wind generator and batteries simulation models used in the theoretical studies presented in
the following chapters. It starts by introducing the most mature techonologies related to PV
and wind generation and batteries. Their basic features are important to understand the
physical and chemical principles underlying their models. Then, a brief literature overview
of these components models is also included in this chapter. It is followed by the full
description of the mathematical models implemented in Matlab/Simulink environment for
the PV array, wind generator and batteries simulation in the studied power systems.

The third chapter investigates DC power systems for telecommunications including


renewable energies. As telecommunications power systems follow the latest technological
trends, other similar power systems available in the literature are also studied. It is conluded
that DC-DC converters are very important power interfaces for the renewable energies op-
timized generation. Therefore, four different DC-DC converters are chosen for PV-battery
and wind-battery systems: single-ended unidirectional DC-DC converters, three-phase in-
terleaved DC-DC converters, three-level boost DC-DC converters (including its application
to PV power systems) and bidirectional half-bridge DC-DC converters. Three PV-battery
and a wind-battery systems to be studied in the upcoming chapters are presented.

The fourth chapter provides an in-depth analysis of both healthy and open-circuit fault
behaviour of the mentioned non-isolated DC-DC converters. Such analysis was used to
develop the fault diagnostic methods presented in the fifth chapter.

The open-circuit fault diagnostic methods for DC-DC converters presented in the lit-
erature are reviewed in the fifth chapter. Their reduced number (when compared with
fault diagnostic methods for other power converters) and limitations highlight the need
for new methods. Therefore, four fault diagnostic methods are proposed in this chapter.
Their principles and algorithms are thoroughly described and validated by simulation and
experimental results.

Once the fault is detected, the converters malfunction and/or stoppage must be avoided
using fault-tolerant strategies, which are discussed in the sixth chapter. A literature sur-
vey of the fault-tolerant strategies applied to DC-DC converters is introduced and a new
fault-tolerant reconfiguration strategy for the three-level boost converter is proposed. Con-
ventional fault-tolerant strategies that might take advantage of the effectiveness of the
previously proposed fault diagnostic methods are presented. A new fault-tolerant reconfig-
uration strategy is also presented.

Finally, the seventh chapter presents the main conclusions that have resulted from this
work as well as related future research proposals.

In summary, the main original contributions of this work are:

Comprehensive study of the faulty operation of the following non-isolated DC-DC


converters: unidirectional single-ended DC-DC converters, three-phase interleaved
1.1 Outline and Contributions 5

DC-DC converters, three-level boost DC-DC converters for PV appplications and


bidirectional half-bridge DC-DC converters;

New fault diagnostic methods for the following non-isolated DC-DC converter topolo-
gies: unidirectional single-ended DC-DC converters, three-phase interleaved DC-DC
converters, three-level boost DC-DC converters for PV applications and bidirectional
half-bridge DC-DC converters;

New faul-tolerant reconfiguration strategy for the three-level boost DC-DC converter
for PV applications.
6 Chapter 1. Introduction
Chapter 2

Generators and Batteries


Mathematical Models

Renewable energy techonlogies for electrical energy conversion have a random electrical
generation highly dependent on climatic conditions. Therefore, many different scenarios of
climatic conditions and load demand can result in different challenges for the power system
to remain reliable. Simulation provides the means to safely and at a relatively low cost
study those situations with reasonable accuracy and computational effort.

This chapter aims to describe the simulation models used in this work for simulating
PV arrays, wind generators and batteries. These models reflect physical and chemical
principles related to PV, wind and battery electrical energy conversion. Therefore, these
basic concepts are introduced and a brief overview of other literature models is also provided.

2.1 PV Generator Model

2.1.1 Basic Features of Solar Cells

Solar cells are semiconductors with a special design and construction to use sunlight
to produce electrical energy efficiently [31]. Above the PN junction that constitutes the
solar cell, there is a metallic grid, which allows the contact between the light and the
semiconductor. The grid lines make electrical contact. On the other side of the electrical
cell, there is a metallic layer, which is the other electrical contact. Between the grid lines,
an antireflective layer improves the efficiency of the solar cell, because it enhances the light
transmission.

PV generators are non-linear power sources as their current-voltage relationship shows


in Figure 2.1. Its non-linear behavior is characterized by a maximum power point (MPP) in
the curves knee, which changes with solar irradiation and temperature. In Figures 2.2 and

7
8 Chapter 2. Generators and Batteries Mathematical Models

2.3, it is graphically represented the influence of irradiation and temperature, respectively,


in the current-voltage curve. For particular irradiation and temperature values, the current-
voltage curve is characterised by the open-circuit voltage (indicated by V oc in Figure 2.1)
and the short-circuit current (indicated by Isc in Figure 2.1).

To design PV generators, the photovoltaic modules manufacturers provide the men-


tioned points at standard test conditions (STC), which are defined as a cell temperature of
25 C and an irradiance level of 1000 W.m2 . Occasionally, they also provide the temper-
ature coefficients of power, short-circuit current and open-circuit voltage and the nominal
operating cell temperature (NOCT), which occurs at 800W.m2 irradiance, 20 C ambient
air temperature and at a wind speed of 1 m/s. These parameters are essential to build
photovoltaic generators models as well.

Figure 2.1: Current-voltage curve of a PV module.

Figure 2.2: Irradiance effect on the current-voltage curve of PV modules.

To obtain the desired values of voltage and current from a PV generator, it is neces-
sary to associate photovoltaic cells together in parallel and/or series. Due to photovoltaic
cells electrical characteristics and other undesirable factors (like shadows, dirt, and power
degradation due to solar cells aging), such an assemblage has consequences on the efficiency
of the photovoltaic generator when compared to each single photovoltaic cell.

Series connections of solar cells are arranged when it is necessary to increase the PV
generator output voltage. The total output voltage provided by their connection is equal
to the sum of the output voltage of each cell. When two series PV cells are similar and
2.1 PV Generator Model 9

Figure 2.3: Temperature effect on the current-voltage curve of PV modules.

exposed to the same irradiance values, the resultant voltage will be twice the voltage of only
one cell. Unfortunately, this does not always happen, because solar cells could be exposed
to different irradiance levels. If one cell of the series connection is dirty or shadowed, it
will affect the performance of the whole set, because the total generated current value is
limited by the one generated by the less illuminated cell. The voltage provided by the cells
will be unequal distributed, but, even worstly, under certain operating conditions, the less
illuminated solar cell could be reverse biased [32] and this will lead to power dissipation on it
and a subsequent temperature rise (hot-spot). This leads to an even worse performance and
possible damages. If the solar cell is totally covered and not exposed to sunlight or if it has
suffered a severe fault and is not working anymore, the whole series string of photovoltaic
cells will behave like this cell, i. e., the whole system will be in open-circuit. To avoid this,
a bypass diode should be placed in all series strings of solar cells. With this simple solution,
PV modules will produce a higher output power and will have lower temperature rise when
a part of them is shaded [33].

To increase the current produced by a PV generator, solar cells are connected in parallel.
The total output current of a parallel association of solar cells will be the sum of each
cell generated current. As solar cells have internal losses, which are due to losses of the
semiconductor material and metallic contacts, the current produced by two similar solar
cells will never be the same even if they are under identical operating conditions.

Performance decreases due to partially shadowing of the PV generator or aging is a


crucial concern for designers and researchers. In order to make the PV generator less
vulnerable as possible, two possible solar cells configurations have been proposed in [34]:
total-cross-tied array and bridge-linked array. Significant improvements to PV modules
configurations in PV arrays have also been presented in [35, 36] to improve PV power
generation under partial shading conditions and reduce mismatch losses (losses due to
differences in the PV modules current-voltage characteristics).
10 Chapter 2. Generators and Batteries Mathematical Models

2.1.2 An Overview of PV Models

The whole process of converting sunlight into electrical energy is determined by spe-
cific properties of the semiconductors that form the solar cell. Among those properties, the
most important ones are related to their capacity to absorb and reflect photons (which are
sunlight particles carrying specific amounts of energy), generation of free carriers charge,
migration of charge and charge separation by an applied electric field [31, 32]. Based on
these properties and on the current-voltage characteristic, different photovoltaic genera-
tor models have been developed with different degrees of complexity concerning different
approximations of the solar cells behaviour. They can be classified according to Figure 2.4.

Figure 2.4: PV models classification.

The first group is called Point by Point [37], where each operating point can be
calculated with two different equations (one for the operating current and the other for
the operating voltage) function of temperature, irradiance and empirical parameters. The
Method of Anderson, the method of Bleasser and IEC-891 procedure are some of the de-
veloped methods to reproduce the photovoltaic generator behaviour [37].

Due to the difficult computations of the complex mathematical photovoltaic generator


models, graphical models based on the current-voltage characteristic were developed. The
non-linear photovoltaic generator current-voltage characteristic can be divided in several
linear parts to easily extract equations that model the photovoltaic behavior. In [38], three
different areas are defined in the photovoltaic generator current-voltage characteristic. The
method seems to diverge on the knee zone of the curve and subsequently is not very accurate.
Dividing the current-voltage characteristic in more than three parts allows to obtain a more
accurate model. Based on this, the Piecewise Model described in [39] allows achieving any
degree of accuracy. Despite of being graphical models, they are based on the single-diode
model.

The third main group for classifying photovoltaic generator models is the most used in
different applications, especially the one-diode model with constant parameters [4044], due
to its good trade-off of accuracy and easy computation. The one-diode equivalent circuit
of a solar cell is represented in Figure 2.5. This circuit is widely used in the literature to
2.1 PV Generator Model 11

explain the physical phenomenon occurring in a solar cell.

Figure 2.5: One-diode equivalent circuit for a PV cell.

The ideal solar cell equivalent circuit is composed only by the light-induced current
source and the diode in anti-parallel with it without the resistances, which represent differ-
ent kinds of operating losses. The light-induced current (Iph ) results from the separation
and drift of the electron-hole pairs generated by photon absorption under an electric field.
This current is directly proportional to the irradiance applied on the solar cell.

The diode in anti-parallel with the current source represents the PN junction that con-
stitutes the solar cell. This diode does not reveal a total ideal current-voltage characteristic
and it is necessary to add a parameter a1 to incorporate these non-idealities.

The current path through the semiconductor, metallic grid and metallic contactors
faces resistive losses, which are represented by the series resistance (Rs ) in the equivalent
circuit of Figure 2.5.

Due to a mass production of solar cells, some shunts at the boards of solar cells can
happen during the production process, which will affect the solar cell performance. Their
effect is represented by the shunt resistance (Rsh ) in the equivalent circuit.

As a result of the formation of electron-hole pairs occurring in the depletion regions of


the solar cell (transition region between the n-type and p-type semiconductors), a second
diode is usually added to this equivalent circuit in order to represent this current [32,
45]. This originates a more accurate but difficult to be implemented model, usually
denominated double diode model. This second diode has a different saturation current and
ideality diode factor.

The equation that combines all the above concepts and describes the double diode solar
cell equivalent circuit is:

q (Vpv + Ipv Rs ) q(Vpv + Ipv Rs Vpv + Ipv Rs


       
Ipv = Iph I01 exp 1 I02 exp 1 ,
a1 kTu a2 kTu Rsh
(2.1)

where q is the electronic charge, k is the Boltzmann constant, Tu is the solar cell tempera-
ture, I01 and I02 are the PV diodes saturation currents, Vpv is the solar cell output voltage,
12 Chapter 2. Generators and Batteries Mathematical Models

a1 and a2 are the diodes ideality constants. The one-diode model equation does not have
the second diode term in equation 2.1.

In the literature, different ways to calculate the parameters of this equation are de-
scribed. Different approaches can be used to estimate the PV model parameters, such as
semiconductors physics [46], application of complex numerical estimation techniques to a
real PV cell current-voltage curve [46, 47] and parametrization based on the PV module
datasheet [48, 49]. For instance, in the constant parameter model, a few parameters like
the series resistance and the ideality diode factor are considered constant [37, 4043, 50].
In contrast, there is also the adaptative parameter model [45], which is more accurate and
where all parameters vary with operating conditions. The effect of the various parameters
on single diode cell performance has been studied in [51], where it is also concluded that
the single diode model results are accurate enough when compared with the double diode
model.

Simplifications to the solar cell equivalent circuits must be done carefully and having in
mind the accuracy needed for the application. In [52, 53], the series resistance is neglected
in the developed models. This affects the accuracy of results, because the series resistance
affects the value of the generated current value. However, not only the electrical steady-
state behaviour and transient effects are studied in literature. References [34] and [54]
describe simulation models to study undesired factors in solar cells, such as the shadowing
effects and the mismatch losses. In [41], electro-thermal and heating/cooling processes of a
solar cell are modeled. The existence of many PV models based on the same assumptions
for specific applications incited the creation of a generalized approach in [55].

2.1.3 PV Array Model

As mentioned in the previous section 2.1.2, the one-diode equivalent circuit of a solar
cell (which is represented in Figure 2.5) is widely used in the literature, due to its good
compromise between simplicity and accuracy for power electronics simulations [49,51]. The
model described in this section has been proposed in [49].

Using the single-diode model, a PV module is described by the following equation:

q (Vpv + Ipv Rs ) Vpv + Ipv Rs


   
Ipv = Iph I01 exp 1 . (2.2)
a1 Vt Rsh

Most of its parameters have already been presented in the previous section when equa-
tion (2.1) for the double diode PV model was introduced. While equation (2.1) is valid
for only one cell, equation (2.2) generalizes this equation for a PV module with Ns cells
connected in series. Therefore, Rsh and Rs are the equivalent parallel and series resistances
of the PV module. A new variable Vt is introduced (usually named as thermal voltage),
which is given by
2.1 PV Generator Model 13

Ns kTu
Vt = . (2.3)
q

Each parameter can be determined from appropriate equations using only some vari-
ables (voltage and current temperature coefficients - KI and KV , respectively) and the
three current-voltage curve points provided by the commercial PV module datasheet: open-
circuit voltage (V ocn ), short-circuit current (Iscn ) and the maximum power voltage and
current (V mppn , Imppn ). Those three PV operating points provided by the manufacturer
datasheet are measured at nominal conditions of temperature (Tun ) and irradiation (Gn ).

As mentioned, the light-generated current Iph depends linearly on the solar irradiation.
To estimate this variable, the temperature influence should also be considered:

G
Iph = [Iph,n + KI (Tu Tun )] , (2.4)
Gn

where Iph,n is the light induced current at nominal conditions (it will be defined later), Tu
is the actual temperature, and G is the actual solar irradiation.

The diode saturation current I01 is strongly dependent on the operating temperature
Tu . It can be calculated using the following equation proposed in [49]:

Iscn + KI (Tu Tun )


I01 = . (2.5)
exp [V ocn + KV (Tu Tun ) /a1 Vt ] 1

The values of Rs and Rsh are determined through an iterative method based on the
fact that there is an only pair (Rs , Rsh ) that verifies the following condition [49]:

Pmax,m = Pmax,e = V mppn Imppn , (2.6)

with Pmax,m as the maximum power calculated by the presented model and Pmax,e as the
maximum experimental power from the manufacturer datasheet at the MPP. The resultant
equation from condition (2.6) is solved to calculate the value of Rsh .

q V mpp+ Rs Imppn V mppn + Rs Imppn


     
Pmax,m = V mppn Iph I01 exp 1
kTu a1 Ns Rsh
(2.7)

V mppn (V mppn + Imppn Rs )


Rsh = n h i o
n Rs ) q
V mppn Iph V mppn I01 exp (V mppn +Impp
Ns a1 kTu + V mpp n I01 P max,e
(2.8)
14 Chapter 2. Generators and Batteries Mathematical Models

The iterative procedure to determine Rsh and Rs is illustrated in the flowchart of Figure
2.6. The initial guess for Rs may be zero and Rsh may be given by

V mppn V ocn V mppn


Rsh,min = . (2.9)
Iscn Imppn Imppn

Each iteration updates both values of Rsh and Rs towards the best model solution and
the following equation for the light-induced current under nominal conditions will be valid:

Rsh + Rs
Iph,n = Iscn . (2.10)
Rsh

Figure 2.6: Algorithm of the method used to adjust the model current-voltage curve [49].

2.1.4 Model Validation

The current-voltage relationship of the BP 4175 (175 W) PV module from the manu-
facturers datasheet [56] (whose data is included in Appendix B.1) is plotted in Figure 2.7a
2.2 Wind Generator Model 15

under a radiation of 1000 W.m2 and different specified cell temperatures. At the same con-
ditions of solar radiation and cell temperature, the current-voltage relationship obtained
from the model simulation is plotted in Figure 2.7b. The simulation parameters used are
included in Appendix A.1. The simulation results are in good agreement with the manu-
facturers data. At STC (green simulation curve in Figure 2.7b), the simulated open-circuit
voltage, short-ciruit current and MPP correspond exactly to the information presented in
manufacturers datasheet. The shape of both curves obtained in simulation and provided
by the manufacturers are very similar, although the simulation slope of the linear parts
of the curve (before and after the curves knee) is slightly larger than the manufacturers
curves which lack accuracy for a better analysis.

(a) (b)

Figure 2.7: Current-voltage curves of BP 4175 PV module: (a) manufacturer datasheet [56]; (b)
curves obtained using the simulation model and Appendix A.1 parameters.

2.2 Wind Generator Model

2.2.1 Basic Features of Wind Energy Conversion Systems

Wind energy conversion systems comprise a wind turbine, an electrical genetor, power
interfaces and control systems. Each of these components has a specific role. The wind
turbine converts the wind kinetic energy into mechanical energy. Its rotating shaft spins
16 Chapter 2. Generators and Batteries Mathematical Models

an electrical generator for producing electrical power, which must be conditioned by power
converters and their control systems.

The kinetic power available in the wind passing through an area Aw with velocity vwind
is [57]:

1
Pwind = Aw vwind
3
, (2.11)
2

where is the air density. Only part of this power (2.11) is extracted by the wind turbine. Its
theoritical maximum limit is 59% and it has been demonstrated by the German physicist
Albert Betz [57]. This portion of useful mechanical power is usually named as power
coefficient (cp ). Commercial wind turbines have lower power coefficients due to inherent
losses [58]. Therefore, the useful wind turbine mechanical power is:

1
Pmec = cp Aw vwind
3
, (2.12)
2

The power coefficient cp is a nonlinear function of the blade pitch angle and the tip
speed ratio [59]. The latter is the ratio between the blade tip speed and the wind speed:

R
= , (2.13)
vwind

where R is the rotor radius and is the angular velocity of the rotor.

A typical power coefficient curve is presented in Figure 2.8, assuming a fixed-pitch-


angle wind turbine. There is a maximum value for the power coefficient at a specific value
of the tip speed ratio which maximizes the mechanical power extracted from the wind.

The mechanical power versus the generator rotor speed for different wind speeds is
shown in Figure 2.9. For each wind speed, there is an optimum rotor speed where the
maximum power is extracted from the wind.

2.2.2 An Overview of PMSG Models for Wind Energy Conversion Sys-


tems

The wind turbine is connected to an electrical generator which converts the mechanical
power into electric one. There are many options of electrical generators to fulfill this role,
but the most attractive and common solution for small and medium power applications is
the permanent magnet synchronous generator (PMSG) [60].

PMSGs models have been very important for studying their design and operating con-
ditions. Analytical models based on the 2-D solution of the Maxwells equations using the
separation of variables methods have been proposed for studying the machine magnetic
2.2 Wind Generator Model 17

Figure 2.8: Power coefficient versus tip speed Figure 2.9: Power speed characteristic.
ratio.

field properties [61], winding faults [62] and design issues [63]. A generalized equivalent
magnetic circuit model is presented in [64] also for design purposes avoiding the use of com-
putationally expensive methods required by the previously mentioned models. Regarding
their electrical generation issues, the most used PMSG model in the literature is the dq
model in the synchronous rotating reference frame. This model is based on the mathe-
matical model of the conventional synchronous generator, without including the damping
winding, because the excitation is constant and provided by permanent magnets instead of
field current [65]. The accuracy of this method can be improved by taking into account both
the machine magnetic parameters variations and its core losses [66, 67]. However, increas-
ing the computational effort might be unattractive for some applications. The compromise
between accuracy and computational effort should be sought for each model application
and a simplified numerical average-value model of the PMSG has been proposed in [68].

Simulation models require specific PMSG parameters. Unfortunately, these parame-


ters are not necessarily included in the PMSG nameplate. A few standstill and loaded
tests together with measurements and identification approaches might be required for their
determination, which are presented in [67, 6972].

2.2.3 PMSG Model

Assuming that there is no saturation, a sinusoidal back emf and negligible eddy current
and hysteresis losses, the PMSG voltage equations in the synchronous rotating reference
frame (dq) [65] are:

dd
vd = Rst id + s q (2.14)
dt

dq
vq = Rst iq + + s d (2.15)
dt
18 Chapter 2. Generators and Batteries Mathematical Models

where Rst is the stator winding per-phase resistance, s is the synchronous angular speed,
s and q are the dq axis stator flux linkages and their equations are:

d = Ld id + f (2.16)

q = Lq iq (2.17)

Ld and Lq are the d, q axis inductances, respectively, and f is the rotor magnets flux
linkage.

The electrical model of the PMSG uses its currents equations expressed in the rotor
reference:

did 1 Rst Lq
= vd id + s iq (2.18)
dt Ld Ld Ld

diq 1 Rst Ld f s
= vq iq + s id (2.19)
dt Lq Lq Lq Lq

The angular velocity of the rotor is given by:

s = pm , (2.20)

with p as the number of pole pairs and m as the rotor speed.

The mechanical and electrical position of the PMSG rotor are, respectively:

Z
m = m dt, (2.21)

e = pm , (2.22)

ds 1
= (Te Dp s TL ). (2.23)
dt J

Finally, the electromagnetic torque equation of the PMSG is [73]:

3
Te = p [f iq + (Ld Lq )) id iq ] (2.24)
2
2.3 Battery Model 19

The following equations are used to perform the Park and inverse Park transformations,
respectively:

   
cos e 2 cos e + 2

vd cose v
2  3   3  a
vq = sin sin e 2 sin e + 2 (2.25)
e 3 vb

3 3
v0 1 1 1 vc
2 2 2

1

va cose
 
sin
 e  vd
v = cos e 2 sin e 2 1 (2.26)

b vq

 3   3 
vc cos e + 2
3 sin e + 2
3 1 v0

In this work, a 2.2 kW PMSG has been modelled using the parameters included
in Appendix section A.2. These parameters were provided by the PMSG manufacturer
(Yaskawa), allowing to closely predict its operating behaviour using mathematical simula-
tions.

2.3 Battery Model

2.3.1 Basic Features of Batteries

Batteries are electrochemical power sources that store electrical energy and deliver it
in case of energy demand. They result from an ensemble of electrochemical power cells,
which can either be connected in series or in parallel, according to the required voltage or
capacity, respectively. Each of these cells includes two conductor electrodes immersed in
a electrolyte which enables the exchange of ions between them. Chemical reactions occur
during charge and discharge processes resulting in electric current.

Commercial batteries are usually identified by their type and their output rated voltage
and capacity (which is expressed in Ampere-hour, Ah). These are the main parameters for
choosing and sizing a battery system for a specific application, besides their technology
characteristics. Lead-acid batteries are the ones incorporating the most mature technology,
although they have a low life cycle and low specific energy and power. Nickel-Cadmium
batteries have a longer life and low maintenance requirements than lead-acid batteries.
However, their success in the battery market has been limited by their high cost and toxic
components. Among the most known and used batteries, lithium-ion batteries are widely
used in small applications. Although their high energy density and specific energy are de-
sirable in many applications, lithium-ion batteries self-discharge more quickly than other
battery types and require protection circuits to ensure that their operation is always un-
der safe voltage and temperature values. The flammability of its components is also a
main worry for the security of these batteries application. Taking all this battery type
20 Chapter 2. Generators and Batteries Mathematical Models

characteristics summary into account, the most suitable battery technology for telecommu-
nications is lead-acid [2] for its maturity, reasonable price, low maintenace possibility and
low self-discharge.

2.3.2 An Overview of Battery Models

Battery models are powerful tools which provide a thorough analysis of their stor-
age capacity during different operating conditions in any power system application. They
permit to study transients, output voltage (which drops during discharge), state-of-charge
(SOC) variations, power losses, chemical reactions, among many other important battery
issues. They can be classified into four different main types [74]: electrochemical models,
analytical models, electrical-circuit models and stochastic models.

Electrochemical models reproduce in detail the chemical processes that occur inside
batterys cells under different operating conditions [75]. Most of these phenomena are
formulated by partial differencial equations which are computationally expensive and time-
consuming to deal with. Their simulation results are very accurate and useful for battery
design.

Analytical models are a simplified version of the electrochemical models focusing only
on some battery properties and their consequences [74,76]. Therefore, they are less complex
and more intuitive than electrochemical models. They also use differencial equations, but
they are not as computationally expensive to deal with as electrochemical models [77].

Stochastic models provide a good qualitative description and pretty accurate results of
the battery behaviour using Markov chains [78, 79]. They are also focused on some battery
properties like the battery discharging and recovery effect [80].

None of the battery model categories briefly presented are practical for electrical circuit
simulations [81]. To study power electronics interfaces and management of power systems
using batteries, their models should be expressed in terms of electrical networks composed
by electrical components (such as resistances, capacitors, inductors, power sources). Two
different approaches can be used for such a demanding modeling task: modelling each
battery part with different electrical circuits or developing an electrical circuit representative
of the battery behaviour [81]. The latter aprroach is the most common one in the literature
and allows intuitive understanding of the battery operation in electric quantities [82, 83].
Most of these models have shown a reasonable compromise between complexity, operation
and precision [84].
2.3 Battery Model 21

2.3.3 Battery Model

An electrical circuit battery model is obviously the most suitable for power converters
simulations studies [85]. Although their results are accurate enough for these applications,
they are based on a few assumptions and have some limitations. The whole chemical
phenomena taking place inside a battery are very complex and it is difficult to include them
all in a simulation model. Furthermore, the mathematical model that represents the battery
electrical behaviour can cause an algebraic loop and consequently simulation instability.
In [86], an electrical-circuit based model is presented which gives accurate enough results
and it is not restricted by the mentioned simulation problems. Its good results and low
simulation efforts have urged its application on many power applications [87, 88].

The presented model assumes the following conditions [86]:

similar battery characteristics during its charge and discharge and consequently only
the discharging curve data is used to deduce its model parameters;

the electrical current amplitude does not influence the battery capacity;

the temperature influence is not taken into account;

the model does not include battery self-discharge characteristic;

the memory effect is not considered.

Besides the aforementioned assumptions, the presented dynamic battery model has a
few limitations:

batterys minimum capacity is 0 Ah;

no maximum capacity limit;

no limitation for the batterys maximum voltage;

batterys minimum no-load voltage is 0 V.

The battery equivalent electrical circuit is shown in Figure 2.10. Its internal resistance
Rbatt is constant and its controlled voltage source is characterized by the following equations:

Qc
 Z 
E = E0 K R + A exp B Ibatt dt (2.27)
Qc Ibatt dt

Vbatt = E Rbatt Ibatt (2.28)

The parameters of these equations are:

E is the battery open-circuit voltage;

E0 is the battery constant voltage;


22 Chapter 2. Generators and Batteries Mathematical Models

Figure 2.10: Battery electrical circuit model.

K is the polarisation voltage;

Vbatt is the battery voltage;

Qc is the battery capacity;

Ibatt is the battery current;

(Ibatt dt) is the actual battery charge;


R

A is the exponential area amplitude;

B is the exponential area time constant inverse;

Rbatt is the internal resistance.

This model requires just a few parameters which are determined from the discharge
curve usually given by the manufacturers datasheet. A typical battery discharge curve is
shown in Figure 2.11, whose area is divided into sections identified as exponential (green
coloured area) and nominal (grey coloured area) areas. Three approximate points are
also indicated which are enough to determine the battery model parameters: (EF ull ,tF ull ),
(EExp , tExp ) and (EN om , tN om ).

Figure 2.11: Typical battery discharge curve.

The batterys discharge curve exponential area parameters (A and B) are determined
2.3 Battery Model 23

from the fully charged voltage (Ef ull ) and the end of the exponential area discharge curve
point (using both voltage EExp and current IExp ), according to the following equations:

A = EF ull EExp ; (2.29)

3
B= . (2.30)
IExp (tExp tF ull )

The fully charged voltage EF ull and the end of the nominal area discharge curve point
(both voltage and current) are applied to equation 2.27 to calculate the polarisation voltage
K:

(EF ull EN om + A (exp (BQN om ) 1)) (Qc QN om )


K= , (2.31)
QN om

where

QN om = IN om (tN om tF ull ) . (2.32)

Finally, the value of the battery constant voltage E0 is also calculated from equation
2.27, but using only the fully charged voltage value EF ull :

E0 = EF ull + K + RBatt IBatt A. (2.33)

The value of the battery internal resistance should be previously determined. Two
different aproaches can be used: equation (2.34) dependent on the battery efficiency [86]
and equation (2.35) which supposes connecting a resistive load and measuring its voltage
ELoad and current iLoad .

1
RBatt = EN om (2.34)
0.2 QN om

EF ull ELoad
RBatt = (2.35)
ILoad

2.3.4 Model Validation

Sealed gelled electrolyte lead-acid batteries were simulated using the previously pre-
sented model. The battery manufacturer datasheet does not include its discharge curve and
24 Chapter 2. Generators and Batteries Mathematical Models

most of the data provided is presented in Appendix B.3. Therefore, it has been measured
using a battery test application provided by a programmable DC electronic load.

Both simulation and experimental results of the battery discharge curve are presented
in Figure 2.12. After several tests which were performed as described in section 2.3.3, the
obtained battery model parameters are included in Appendix A.3. As expected, lead-acid
batteries present a discharge curve with a flat shape. Their output voltage varies very slowly
with their SOC variation. It is clear that both curves have similar shape and approximate
values. There is a good agreement between simulation and experimental results which
validates the battery model.

Figure 2.12: Lead-acid battery discharge curve.


Chapter 3

Telecommunications Power
Systems

All the necessary conditions to make telecommunications equipment work properly are
achieved, not only by doing efforts in this specific area, but also by conceiving a robust
power supply unit in order to satisfy all its requirements. The basic prerequisites imposed
to telecommunications power systems are related to their safety, long life and uninter-
ruptible power [2, 89, 90]. They all depend on a good power system design and predictive
maintenance. Other requirements are associated with the electrical characteristics of the
telecommunications equipment, such as nominal voltage and operational range, noise and
acceptable ripple levels.

The most frequent rated voltage values for telecommunications systems are 24, 48 and
60 V. The choice of the voltages depends on the safety regulations of the country where
they are applied and the most usual one is 48 V [2, 91]. Due to the dynamic behaviour of
energy sources, power grid and power electronics, there is a tolerance range for variations
of this quantity, whose lowest value must not compromise the reliable operation of the
telecommunications equipment and the highest value should be limited according to the
heat generated by the components. There is also a top critical voltage which must not be
exceeded, even for a short time, as components may be destroyed. Some equipment have
even a bottom critical voltage of less or equal to 40 V (for a time less than or equal to 1
ms).

The quality and purity of the supplying voltage is also an important issue. It is impor-
tant to size filters to the power electronics converters in order to minimize ripple as much
as possible. The dynamic operation of the power supply can have temporary deviations of
the steady-state voltage to higher (causing equipment dropouts) and lower values (leading
to the damage of power supplies and electronic equipment).

In this chapter, the introduction of renewable energies in telecommunications power

25
26 Chapter 3. Telecommunications Power Systems

systems is covered together with the technology development of other similar power systems.
In all PV and wind power systems, the DC-DC converter has a key role in stepping up and
down the generators output voltage, optimizing their power generation and charging the
output batteries. Four different DC-DC converters were chosen to be studied in this work.
They are presented in this chapter together with the complete simulation model and the
experimental setup used.

3.1 Renewable Energy Based Power Systems

3.1.1 Small Wind Turbines Power Systems

Grid-connected large wind turbines, which are usually organized in wind farms, have
been extensively studied in the literature regarding technical designs [92], electrical gener-
ators [93, 94], control systems [95], power electronics interfaces [96, 97] and reliability [98].
The challenges for the operation of small wind turbines (defined as having a rotor swept
area less than 200 m2 [99] and rated power range lower than 200 kW [60]) are different as
they are suitable for specific applications related to power supply remote sites [99], water
pumping, electric vehicle charge stations [100] and battery charging [101]. Their technical
feasibility and competitive cost have been long considered for supplying telecommunications
apparatus [102].

Small wind turbines power systems for telecommunications have always followed tech-
nological trends for their most effective implementation. In the past, asynchronous gener-
ators were the most common solution for small wind conversion systems [60] and also for
contemporaneous wind power telecommunications systems like the one presented in [103]
which was implemented by the French manufacturer of wind turbines Vergnet Eolien. More
recently, the PMSG with a high number of pole pairs has dominated the low power wind
market [60]. In fact, this generator presents the best features for these applications [104].
It uses permament magnets for excitation instead of DC supply, which eliminates the need
for an excitation system along with cooling system. Its flexible design allows smaller and
lighter generators. In addition, its maintenance costs are lower when compared with other
generators and the regulation of the generator speed does not need unreliable elements
like gears or gearbox. Besides, it is able to achieve high torques at low wind speed [104].
Therefore, recent wind telecommunications power supplies already include a PMSG [105].
Its drawbacks are related to the cost of the permanent magnets and to their ability to
demagnetize when the generator is exposed to high temperatures or to several overloading
or short-circuit operating problems [106].

PMSG output power must be conditioned for optimizing its generation and to fulfill
telecommunications load power supply requirements. Wind power systems intermittence
and unpredictable power generation can be harmful for fluctuation-sensitive loads like the
3.1 Renewable Energy Based Power Systems 27

telecommunications equipment. Therefore, energy storage devices are useful in providing


power balance and quality. The electrical energy produced by low power wind conver-
sion systems can be converted into electrochemical energy in batteries and supercapacitors
(SCs) [101, 107109] or into chemical energy in fuel cells [110114]. Besides fluctuation
suppression, these devices can have additional applications, such as DC-bus voltage con-
trol [115], backup energy storage and power peaks demand shaving [99, 116].

A typical small wind turbine power system is depicted in Figure 3.1. Focusing on
its power interface, it joins a diode rectifier and a non-isolated DC-DC converter [60].
Using a controlled rectifier allows for a wider range of control and a better performance
[60, 117, 118]. However, in comparison, for low power wind conversion its complexity and
high cost are more significant. Besides, many control strategies for wind power generation
optimizing and apropriate battery charging have been proposed in the literature for the
power converter configuration of Figure 3.1. Such control strategies are algorithms which
change the power generator characteristics for maximum power point tracking (MPPT) and
they are basically divided into two groups: algorithms based on the knowledge of turbine
parameters and alghorithms based on iterative search [99]. The former group requires prior
knowledge of the optimum wind turbine specific variables relationship. The DC power
versus DC voltage characteristic is used for MPPT in [119]. In [120], the MPPT algorithm
is based on the optimum DC voltage-DC current relationship. Both MPPT algorithms
presented in [119,120] do not need mechanical sensors and rely only on voltage and current
sensors. However, many field tests are required to build the lookup tables of the mentioned
relationships. The method presented in [121] uses the power-rotor speed characteristic
for MPPT. This method requires aerodynamic tests using wind tunnel [99]. Besides their
high cost, it is difficult to provide the real environmental operating conditions. Therefore,
the power-rotor speed characteristic will not be accurate enough for MPPT. In [104], the
wind conversion system DC voltage-DC current relationship is formulated by an equation
instead of a lookup table. However, the equation has two coefficients that still require
field tests. A hybrid method of both aforementioned MPPT proposed classifications is
presented in [122]. This method determines a linear relationship between the DC voltage
squared and the DC current through a proposed experimental process. Then, an iterative
method based on the Hill-Climbing search method is used to find the missing coefficient
of the established relationship providing the optimum DC current value. This method is
faster than the conventional Hill-Climbing searching methods presented in [123126]. The
Hill-Climbing search methods consist in regulating the DC-side voltage or current towards
the MPP, according to successive comparisons of the wind generator output measurements.
They only use current and voltage sensors. Those iterative methods have a slow response
and can be inefficient under fast variations of wind speed. Therefore, another improved
solution of the Hill-Climbing search methods has been proposed in [127] which ensures that
changing wind conditions do not lead the iterative search in the wrong direction. Finally,
a novel generator system that automatically tracks the MPP without any power interface
28 Chapter 3. Telecommunications Power Systems

control has been proposed in [128].

Figure 3.1: Small wind turbine power system.

3.1.2 PV Power Systems

PV energy and its related technology differs from others renewable energies in some
important features, such as a wide power range (wind, hydro and geothermal energy are
not so flexible for small power applications), suitability for all climates (as long as there
is sunlight, electric energy is being produced, even if the climatic conditions are not the
best ones), modularity, and low need for maintenance. These are the main reasons why
photovoltaic systems are getting spread as an alternative power solution for a variety of
end uses. Calculators, traffic lights and signalizations, water pumps, warning systems, and
remote DC telecommunications equipment are some of the most common small and medium
size applications, but they are also suitable for high power applications like central power
plants and electrification of remote areas.

Due to industry efforts in module manufacturing technology, in improving the perfor-


mance and in reducing degradation, solar modules benefit from a high reliability [6] and
their warranties last at least 20 years [7]. Obviously more efforts should be made in getting
cheaper, more reliable and efficient modules, but, at this development level, other worries
arise on photovoltaic power systems, which are associated to the other system components
responsible for power conditioning and energy storage. They should operate in an opti-
mized and efficient way. It is important to reduce losses in the whole system as much as
possible and to use the most appropriate procedure to store energy, in order to get a less
expensive and reliable system.

High power central power plants with a large number of photovoltaic modules in a
huge area and householder photovoltaic systems are the most common practices to connect
the electric power produced from the photovoltaic cells to the grid. The main concerns of
these systems are related to their sizing, design concept and power electronic converters
requirements and controls.

The MPPT for PV systems is an optional component, but it highly increases the
system efficiency, because it consists of a DC-DC converter with a robust control algorithm
which tracks the PV modules MPP as it changes with the atmospheric conditions (solar
radiation and temperature). Several algorithms have been developed and improved to
achieve the highest power level over a wide range of inputs with high convergence speed
3.2 Power Conditioning Interfaces 29

and efficiency and low components cost, such as: perturb and observe (P&O), hill climbing,
incremental conductance, fractional open-circuit voltage, fractional short-circuit current,
fuzzy logic control, neural network, current sweep, DC-link capacitor droop control, load
current or load voltage maximization, dP/dV or dP/dI feedback control, slide control.
Many other methods and control strategies have been and will continue to be studied in
the literature [129]. However, two of those thousands of MPPTs available in the literature
are used in most PV applications and studies: P&O and incremental conductance.

PV technologies have been always attractive for telecommunications networks power


supply [130] and many PV systems for this purpose have been successfully installed around
the whole world [131135]. PV arrays energy production is intermittent and highly depen-
dent on climatic conditions, which is inconsistent with the 100% availability requirement
of telecommunications equipment. Therefore, standalone power systems based on the PV
technology and storage devices, which can be a combination of an electrolyzer and fuel
cells [136, 137], batteries or SCs [138, 139], are also alternative solutions for low power re-
mote equipment [140]. To supply high power loads, other kinds of renewable energies (such
as wind) can interact simultaneously with the photovoltaic generator complementing its
operation. Such a power system combining different energy sources and energy storage
devices is called a hybrid power system, which is also quiet popular for telecommunications
power supply [1, 105, 141143].

Several DC-DC converters topologies have been proposed in the literature which differ
in the number of power stages, efficiency, control complexity, among others. PV mod-
ules output voltage is very low and requires a high step-up conversion [144]. Therefore,
new alternative and complex DC-DC converter topologies have been proposed in litera-
ture offering a reasonable step-up voltage conversion [145, 146]. Flexibility for different
applications [147], low input ripple [144] and adequacy for the non-linear behaviour of PV
cells [148, 149] are the main criteria for the develpment of new DC-DC converters for PV
applications. PV cell-level power conversion is a challenging power conditioning interface
that provides high efficiency, reliability and modularity [150].

3.2 Power Conditioning Interfaces

In PV and wind power systems, MPPT algorithms highly optimise their energy produc-
tion. In the considered power range, although there may be several algorithms for MPPT,
they all have in common using a non-isolated DC-DC converter to control the generators
output variables. In addition, telecommunications systems operate from a DC supply volt-
age with a specific voltage range. Therefore, the role of DC-DC converters in standalone
power systems for telecommunications based on renewable energies is very important [151].
The literature presents many topologies and families of these converters. They differ in
many characteristics, such as isolation, number of inputs, number of outputs, number of
30 Chapter 3. Telecommunications Power Systems

levels, operating principles, number of phases, control, efficiency, size, among ohters char-
acteristics.

3.2.1 Unidirectional Single-Ended and Bidirectional Half-Bridge DC-DC


Converters

Single-ended non-isolated unidirectional DC-DC converters are the most basic topolo-
gies and commonly named according to their purpose, such as buck, boost and buck-
boost [152]. Other topologies have been included to this category such as Cuk and SEPIC.
They all include an electronic switch, an inductor to store energy, and a flywheel diode. The
electronic switch commutation is controlled in order to maintain the required load voltage.
The flywheel diode carries the current during that part of the switching cycle when the
electronic switch is off. The buck converter topology is presented in Figure 3.2.

Figure 3.2: Buck converter.

The bidirectional half-bridge DC-DC converter is presented in Figure 3.3 and two differ-
ent control approaches are shown in Figure 3.4. This converter results from the combination
of a boost converter on one direction with a buck converter on the opposite direction (it is
also named bidirectional buck-boost topology for this reason [153]) and has higher efficiency
when compared with other bidirectional non-isolated topologies [154]. During the buck op-
eration the low voltage side source is charged while the boost stage is used to discharge it.
Using the control approach of Figure 3.4a both power switches are synchronously switched,
while the control approach of Figure 3.4b always deactivates one switch, according to the
required current direction.

Figure 3.3: Bidirectional half-bridge converter.


3.2 Power Conditioning Interfaces 31

(a)

(b)

Figure 3.4: Bidirectional half-bridge converter control schemes: (a) synchronously switching con-
trol; (b) non synchronously switching control.

3.2.1.1 Wind Power System Simulation Model and Experimental Setup

The power system used to study single-ended unidirectional and bidirectional half-
bridge DC-DC converters is presented in Figure 3.5. It comprises a wind conversion system
based on PMSG and two sets of batteries. The PMSG three-phase output voltage is recti-
fied by a diode-bridge rectifier which is followed by a buck DC-DC converter. The PMSG
electromagnetic torque is controlled through the mentioned buck converter by controlling
the value of the inductor current IW . A modular design of the power system energy storage
part has been implemented by using two separate battery sets with different control pur-
poses. A bidirectional half-bridge converter interfaces the battery pack BattV controlling
the load input voltage Vdc around 50 V and the other battery pack BattI is also interfaced by
a bidirectional half-bridge converter which regulates the current IBattI for different specified
values. The control approach of Figure 3.4a, where both power switches are synchronously
switched, was used in both bidirectional half-bridge converters of the system.

A dynamic simulation of the wind power sytem presented in Figure 3.5 was performed
in Matlab/Simulink environment using its toolbox SimPowerSystems. A 2.2 kW PMSM
has been modelled using the model already presented in section 2.2.3. Each battery pack
comprises 3 3 12 V 12 Ah lead-acid batteries, whose model has been presented in section
2.3.3. All simulation parameters are included in Appendix sections A.2, A.3 and B.4. This
system is supplying a variable resistive load. The switching frequency imposed through the
PWM techniques was around 1 kHz for all three DC-DC converters.

All parameters used in the simulation correspond to the experimental setup components
32 Chapter 3. Telecommunications Power Systems

Figure 3.5: Wind power system.

parameters. It includes a 2.2 kW PMSG coupled to a motor (PMSM) in a test bench (the
latter emulates the spin of a wind turbine), two battery packs of 3 3 12 V 12 Ah lead-acid
batteries, Semikron IGBTs and diodes, voltage and current sensors, a dSPACE DS 1103
digital controller and a variable resistive load. All relevant parameters of the mentioned
componets are included in Appendix B. The switching frequency imposed through the
PWM techniques was around 1 kHz for all three DC-DC converters. Both system simulation
model and experimental setup share the same equipment.

All control strategies were implemented in the DS1103 digital controller board, using a
sample time of 20 s. The real-time monitoring was provided by the dSPACE ControlDesk
software installed on a personal computer.

3.2.2 Interleaved Boost DC-DC Converter

Interleaved DC-DC converters are composed of n phases operating at the same switch-
ing frequency. The scheme of a 3-phase interleaved boost DC-DC converter is presented in
Figure 3.6.

The inductor currents of each phase are interleaved by delaying the gate command
signals of each phase power switch by T /n (where T is the switching period). Interleaved
operation results in a large degree of ripple cancellation on the input and output converter
3.2 Power Conditioning Interfaces 33

Figure 3.6: Three-phase interleaved converter.

currents. Therefore, this DC-DC converter topologies provides great benefits on efficiency,
size, thermal management and modularity [155158], which are critical concerns in many
power applications including telecommunications power supply [2]. They are widely used
in many modern and demanding applications [159164], including PV power systems [10,
165168].

3.2.2.1 PV Power System Simulation Model and Experimental Setup

This section presents the PV-battery power system of Figure 3.7, which includes an
interleaved DC-DC converter as a power interface. In typical PV applications with MPPT
control, only the PV arrays output voltage and current are monitored for control purposes.
The use of an interleaved DC-DC converter also requires acquiring the DC-link current Iin
to limit the converter input current for safety issues. It controls the PV module output
voltage, so that its MPP can be tracked using the P&O algorithm [169]. The DC bus
voltage is set by the battery pack used.

Figure 3.7: PV system with three-phase interleaved converter.

The literature presents many MPPT algorithms for PV arrays. The main criteria of
choosing it for this system was simplicity, maturity and efficiency. Therefore, the P&O
algorithm was chosen to be implemented [169], which is one of the most common MPPT
in several applications [170, 171]. The diagram of this method is illustrated in Figure
34 Chapter 3. Telecommunications Power Systems

3.8. The operating PV module voltage is modified and the resultant power is compared
with its preceding value. While that modification increases the PV module output power,
similar modifications are performed until the output power decreases (when the modification
follows another direction). The output power will oscillate around its maximum value until
it stabilizes.

Figure 3.8: P&O MPPT algorithm.

A model of this system was simulated in Matlab/Simulink environment using its toolbox
SimPowerSystems. The models presented in sections 2.1.3 and 2.3.3 were used for simulat-
ing the PV module and the batteries, respectively. The model parameters are presented in
Appendix sections A.1, A.3 and B.4.

The experimental setup includes one BP 4175 PV module, a battery pack composed of
12 V 12 Ah lead-acid batteries, three inductances (10 mH), Semikron IGBTs and diodes,
a dSPACE DS1103 digital controller, and current and voltage sensors. It is presented in
Figure 3.9. The switching frequency of each phase current is around 1 kHz (which was also
used for simulation) and two sample times were used for different situations: 50 s and 25
s. The dSPACE ControlDesk software permits to introduce the control strategies and to
perform real-time control and monitoring.
3.2 Power Conditioning Interfaces 35

Figure 3.9: PV system with three-phase interleaved converter experimental prototype.

3.2.3 Three-Level Boost DC-DC Converter

The three-level boost converter is presented in Figure 3.10 and its operating state is
characterized by four operating modes, according to the power switches conduction state,
which can be both conducting, or both turned off or one of them conducting and the other
turned off. These operating modes are controlled to regulate the converter output voltage
and to balance the DC link capacitors voltage V C1 and V C2.

Figure 3.10: Three-level boost converter.

This converter is able to double the voltage gain of conventional two-level non-isolated
topologies, while halving the power switch voltage stress [172174]. Therefore, conduction
and switching losses and electromagnetic interference noise are reduced. Nonetheless, if
not correctly sized, the output-diode reverse recovery can be a serious problem, due to the
hard switching of power devices [144]. Taking into account their advantages, they have an
extensive application [175177]. In the literature, new control approaches and designs have
been proposed for this converter for PV applications [172, 178, 179].
36 Chapter 3. Telecommunications Power Systems

3.2.3.1 PV Power System Simulation Model and Experimental Setup

The PV-batteries system represented in Figure 3.11 (including a scheme of the control
part) consists of a PV array supplying batteries with the proposed open-circuit power switch
fault tolerant converter. It is based on a conventional three-level boost converter, whose
application in PV arrays with high output voltage requirements has proven to be very
efficient [144]. In this specific case, for fault tolerant purposes, a very few extra components
are needed, when compared to the conventional topology, and the photovoltaic array must
have at least two series modules or any other even number of series modules association,
as long as it is possible to locate a middle point dividing the module series to connect it to
the converter. Examples of possible combinations for the application of this fault tolerant
DC-DC converter topology are shown in Figure 3.12. This scheme is convenient for many
applications, because the output voltage of photovoltaic modules is very low. Besides the
above compulsory feature, the topology requires at least two inductances and two capacitors
at the input capacitor bank (so that these components can be placed according to Figure
3.11) whose values can sum up to the original designed one. It also requires two more
sensors than the conventional topology for measuring the input current and voltage of each
division of the series combination of photovoltaic modules. The conventional topology only
needs the total input current and voltage measurement of the whole photovoltaic array,
while both topologies need voltage sensors at the output capacitors for balancing the DC
link voltage, which is imposed by the batteries. The extra devices are inside the dashed
line in Figure 3.11. The mentioned extra sensors, if applied for each parallel connecting
branch of a series-parallel photovoltaic array, can also be used for diagnosing photovoltaic
array faults according to the method presented in [180], in addition to their use for this
fault tolerant topology. Lastly, a TRIAC is also required (also inside a dashed line in Figure
3.11) connecting the midpoint of the input capacitor bank to the midpoint of the IGBTs.
This TRIAC is only triggered in case of a fault occurrence.

During normal state, the circuit behaviour is identical to the conventional topology
(presented in the previous section 3.2.3), although its control has been adapted for a PV
system. Power switch S1 is controlled in order to provide MPPT by means of using the P&O
method, which has been introduced in the section 3.2.2.1, and it uses the total array output
voltage (the sum of vP V 1 and vP V 2 in Figure 3.11) and one of the monitorised currents (iP V 1
or iP V 2 in Figure 3.11). Power switch S2 balances the output DC link capacitors voltages
(V C1 and V C2 in Figure 3.11). These controls, during the normal state, are detailed
in Figure 3.13. The circuit analysis is divided in: normal state (operation of the circuit
without any fault), faulty state (transient after an open-circuit power switch fault) and
rebuilt state (post-fault circuit operation with the proposed fault tolerant strategy which
requires hardware and software reconfiguration), which will be introduced in the following
sections.

This system has been simulated in Matlab/Simulink environment. The models pre-
3.2 Power Conditioning Interfaces 37

Figure 3.11: PV fault-tolerant three-level boost converter.

sented in sections 2.1.3 and 2.3.3 were used for simulating the PV module and the batteries,
respectively. Their parameters are presented in Appendix sections A.1, A.3 and B.4. The
converter model aggregates components of the SimPowerSystems toolbox.

The PV-battery system experimental setup includes two BP 4175B photovoltaic mod-
ules and a battery pack of 10 series 12V 12 Ah lead-acid batteries. The parameters of the
photovoltaic modules and the converter components are presented in Appendix B, respec-
tively. The described converter control was implemented in a dSPACE DS1103 controller
board with a sampling time of 50 s, including MPPT control, DC link capacitors voltage
balance and fault-tolerant strategy control. The switching frequency of the power converter
is around 5 kHz (which was also used for simulation). Using interface and isolation boards,
voltage and current signals are connected to the dSPACE controller and the IGBT gate
commands are sent out.
38 Chapter 3. Telecommunications Power Systems

(a) (b)

Figure 3.12: Possible PV modules configuration for the application of the proposed fault-tolerant
converter: (a) parallel-series array (with middle point connection); (b) series-parallel array.

Figure 3.13: Fault-tolerant converter control during normal state where it operates like a conven-
tional three-level boost converter.
Chapter 4

Fault Analysis of Non-Isolated


DC-DC Converters

Power electronic converters are key elements on any power application [181]. They
provide efficient energy conversion, adequate control of the system variables (voltages and
currents), and power conditioning. Therefore, they are determinative for any power applica-
tion satisfactory operation. High temperature and stress levels, which usually characterise
power converters operating conditions, can compromise their reliability [14, 182]. The most
vulnerable components are capacitors and semiconductors, according to different statistical
surveys [183185]. A power switch fault might not result in the stoppage of equipment
operation, but, when not diagnosed in time, its consequences aggravate until it might be
too late to avoid high repair costs or even complete replacement. Power converters fault
analysis is an essential tool to predict malfunctions and to take measurements to avoid
them and protect the converter operation.

Power switch faults can be classified as open- and short-circuit faults. The diagnostic of
the latter ones is already a standard practice integrated in most drive power switches [18].
However, such protections are not able to detect open-circuit faults, whose causes can be
the lifting of the bonding wires during thermical cycling, a driver failure or a short-circuit
fault induced rupture [15, 18, 21, 186]. Therefore, their study for preventing their potential
destructive consequences is very important.

Healthy and open-circuit switch faulty operation of non-isolated DC-DC converters


topologies are studied in this chapter. The information provided by both analyses is the
basis for the developed fault diagnostic methods to be presented in the following chapter.
The theoretical analysis using circuit equations and waverform schemes is validated with
simulation and experimental results.

39
40 Chapter 4. Fault Analysis of Non-Isolated DC-DC Converters

4.1 Single-Ended Unidirectional DC-DC Converters

4.1.1 Healthy and Faulty Operating Principle

Non-isolated unidirectional DC-DC converters open-circuit fault switch behaviour has


already been introduced in [24] for continuous conduction mode (CCM). It is revised and
generalized for the discontinuous conduction mode (DCM) in Figure 4.1, where the induc-
tor current waveform is presented together with the switch gate command signal. While
the gate command signal is ON, during healthy condition, the inductor current waveform
increases almost linearly. This situation is common for both conduction modes, either CCM
or DCM.

(a) (b)

Figure 4.1: Unidireccional DC-DC converter operation in healthy and faulty conditions: (a) CCM;
(b) DCM.

When the gate command signal is OFF, the inductor current waveform linearly de-
creases. If the inductor current is discontinuous (DCM), it will be zero during part of the
interval in which the gate command signal is OFF.

A detailed explanation is provided for the buck converter presented in Fig. 3.2 which
can be extended easily to other unidirectional non-isolated DC-DC converter topologies,
because their operating principle is similar [24]. The slope of the inductor current I is

dI VL
= , (4.1)
dt L

where L is the inductance and VL is the inductor voltage. The parameter L can be con-
sidered positive and constant, while VL changes according to the switch operation. The
inductor voltage VL determines the current slope value and its sign.

When the switch is ON, the inductor voltage VL is

VL = Vin Vout , (4.2)


4.1 Single-Ended Unidirectional DC-DC Converters 41

where Vin is the buck input voltage and Vout is the buck output voltage.

A buck converter is a step-down converter which produces a lower average output


voltage than its input voltage. Therefore,

Vin > Vout , (4.3)

and the inductor current slope is positive.

While the switch is OFF, the inductor voltage VL is

VL = Vout , (4.4)

and the inductor current slope is negative.

After an open-circuit fault occurrence on the switch, it will be always OFF, which
means that (4.4) is the inductor voltage VL value, independently of the gate command
signal state and conduction mode. The inductor current will decrease until it becomes
zero, as shown in Figure 4.1.

4.1.2 Simulation Results

The prototype previously presented in section 3.2.1.1 was simulated to theoretically


validate the faulty behaviour of a buck DC-DC converter under two different conduction
modes. As previously mentioned, all single-ended non-isolated DC-DC converters have a
similar operation [24], but only the buck DC-DC converter topology is considered here.

To study the buck converter CCM operation, the power system presented in section
3.2.1.1 is supplying a load of 35 with an input voltage of 50 V and the buck converter
inductor current IW reference value is 5 A. The battery pack BattI is charging with a current
of 4 A. The operating conditions of the PMSG are modified to obtain a discontinuous
inductor current on the buck converter. Therefore, the buck converter inductor current IW
reference value is reduced to 0.2 A. In this case, the battery pack BattI is supplying 1 A.
The results were obtained for a PMSG mechanical speed of 800 rpm.

Figure 4.2 shows the simulation results for both inductor current conduction modes.

As expected, during healthy condition, the inductor current IW waveform only increases
while the the gate command P is ON.

For the case of CCM, presented in Figures 4.2a and 4.2c, the open-circuit fault is
introduced at 4 ms while the gate command P is ON (Figure 4.2a). Consequently, after
the fault occurrence, the current stops increasing and decreases until it becomes null, as
shown in Figure 4.2c.
42 Chapter 4. Fault Analysis of Non-Isolated DC-DC Converters

In figures 4.2b and 4.2d, the gate command signal P and the inductor current IW for
the discontinuous mode are presented. A fault is introduced at 6.4 ms. At this instant, the
inductor current IW was imminent to increase and the gate command signal was increasing.
As a consequence of the open-circuit fault, the current does not increase, independently of
the gate command signal state.

(a) (b)

(c) (d)

Figure 4.2: Simulation results of a buck DC-DC converter operating with an open-circuit fault:
(a) gate command P (CCM); (b) gate command P (DCM); (c) inductor current IW (CCM); (d)
inductor current IW (DCM).

4.1.3 Experimental Results

The experimental results were acquired under the same operating conditions previously
mentioned for the simulations and the faults were introduced at nearly the same time. The
experimental setup described in section 3.2.1.1 was used. Figure 4.3 shows the experimental
results for the gate command signal P and the inductor current IW during CCM and DCM,
respectively. Comparing both the simulation and experimental results, their similarity is
obvious. The experimental waveforms of the inductor current IW are very similar to the
simulated ones in both operating conditions.

In addition, a few more experimental results are shown in Figure 4.4. Figures 4.4a and
4.4b show the control loop response to the open-circuit fault. It will try to compensate the
fault consequences on the inductor current IW by increasing the duty-cycle D. The PMSG
rectified output voltage (Figures 4.4d and 4.4c) increases while the inductor current IW
is decreasing. The duty-cycle variation depends on the current controller speed, which is
intentionally different in both cases. The PMSG rectified output voltage variation caused
by the open-circuit fault is harsher in the continuous mode operating conditions, because
the inductor current IW is substantially higher.
4.2 Interleaved DC-DC Converters 43

(a) (b)

(c) (d)

Figure 4.3: Experimental results of a buck DC-DC converter operating with an open-circuit
fault: (a) gate command P (continuous conduction mode); (b) gate command P (discontinuous
conduction mode); (c) inductor current IW (continuous conduction mode); (d) inductor current IW
(discontinuous conduction mode).

(a) (b)

(c) (d)

Figure 4.4: Experimental results of a buck DC-DC converter with an open-circuit fault: (a) duty-
cycle D (CCM); (b) duty-cycle D (DCM); (c) PMSG rectified output voltage VW (CCM) ; (d)
PMSG rectified output voltage VW (DCM).

4.2 Interleaved DC-DC Converters

4.2.1 Healthy and Faulty Operating Principle

The analysis presented here for the non-isolated interleaved DC-DC converters will be
mostly focused on the control monitored variables. However, the monitored variables in
44 Chapter 4. Fault Analysis of Non-Isolated DC-DC Converters

these converters can vary, according to the application. A common control method is to use
a current loop for each parallel phase and a unique voltage loop for controlling the output
voltage, when voltage regulation is required [187]. Such an approach has an implicit high
cost which limits the number of parallel phases (because of the high number of current sen-
sors required) [155, 156]. Therefore, different current estimation and balancing techniques
have been proposed in literature to avoid the use of current loops in each parallel phase
of the converter [188]. Nevertheless, according to [155, 156], if each phase is operating in
DCM, all phase currents will be very close to balance without using any complex technique
or current loops. In the latter case, the DC-link current Iin (which is the sum of the phase
currents) is always monitored for control or circuit safety purposes.

Each phase current of the converter in Fig. 3.6 is controlled with the same switching
frequency and has a phase-shift angle of 2/3 radians. Similarly to what has been explained
in the previous section 4.1 for the single-phase DC-DC converter, each phase current of the
interleaved boost DC-DC converter can operate in different conduction modes, which are
presented in Figure 4.5, including switching pattern, phase current waveform and phase
current derivatives.

Figure 4.5: Switching pattern, inductor phase current and derivative for three conduction modes:
CCM, BCM and DCM.

The input current Iin results from the sum of each phase current, according to the
Kirchhoffs current law:

Iin = I1 + I2 + I3 . (4.5)

The derivative of a sum is the sum of derivatives:


4.2 Interleaved DC-DC Converters 45

dIin dI1 dI2 dI3


= + + . (4.6)
dt dt dt dt

Therefore, the derivative of the DC-link current in multi-phase interleaved converters


can be calculated by analysing each current derivative.

When the phase switch is ON, each inductor phase voltage is positive and equal to Vin .
The rising slope of the inductor phase current is given by (4.1).

When the phase switch is OFF, each inductor phase voltage is negative, which results
from (4.2), where Vin is always lower than Vout . Thus, the falling slope of the inductor
phase current is

dIk Vin Vout


= , (k = 1, 2, 3). (4.7)
dt L

While the phase switch is OFF, in DCM, the derivative of the current is also zero after
DM T when the inductor phase current is zero.

The relationship between the input voltage Vin and the output voltage Vout , according
to [152] is

Vout DM
= , (4.8)
Vin DM D

where D is the duty-cycle and DM corresponds to the fraction of the period of time for
which the current is different from zero. Equation (4.8) is generalized for any operating
mode, because if DM = 1, the relationship between the input voltage Vin and the output
voltage Vout for BCM and CCM will be obtained [152].

Considering the positive variation value of Ik as

Vin D T
IM = , (4.9)
L

the derivative of each phase current can be normalized according to (4.10).


dIk IM for 0 t DT
= DT
(4.10)
dt IM
(DM D)T for DT < t T

Equations (4.6) and (4.10) can be used to calculate the derivative values of the current
Iin . According to (4.10), the duty-cycle determines the end of the rising slope (DT ).
Therefore, the duty-cycle determines when the rising slope of each phase current overlaps,
defining the conditions (4.11), (4.12) and (4.13) [189]. For condition (4.11), there is no
overlap of the rising slope as shown in Figure 4.6a, whereas for condition (4.12) two phase
46 Chapter 4. Fault Analysis of Non-Isolated DC-DC Converters

currents rising slopes overlap (Figure 4.6b) and for condition (4.13) the three phase currents
rising slopes overlap (Figure 4.6c). For each condition, the derivative values of the current
Iin are presented in Table 4.1 divided according to the respective condition, considered
interval and DM values effect.

1
0<D (4.11)
3

1 2
<D< (4.12)
3 3

2
D<1 (4.13)
3

(a) (b) (c)

Figure 4.6: Phase currents and divided switching pattern intervals for the following conditions:
(a) (4.11); (b) (4.12); (c) (4.13).

In Figure 4.6, the different switching patterns of each condition are considered and rep-
resented. For each condition, the phase currents, the DC-link current Iin and its derivative
values are subdivided into six subintervals, according to a constant switching pattern. This
means that when there is a change in one switching command signal (by turning it ON or
OFF), a new subinterval is considered. For example, for condition (4.12), in Figure 4.6b, the
six subintervals are indicated as: P 1P 2P 3, P 1P 2P 3, P 1P 2P 3, P 1P 2P 3, P 1P 2P 3, P 1P 2P 3.
The phase currents ripple frequency (which is equal to the switching frequency) is one third
of DC-link current Iin ripple frequency. Therefore, for calculating its derivative values only
two of the six considered intervals are necessary. The result will also depend on the value
of DM .

The waveforms presented in Figure 4.5 also apply to an interleaved buck converter
phase current whose shape is identical. Although the analysis is focused on an interleaved
boost DC-DC converter, it also applies identically to an interleaved buck DC-DC converter.
4.2 Interleaved DC-DC Converters 47

Interval Iin derivative values DM condition Sign


Condition (4.11)
P 1P 2P 3 IM
DT DM < 1
3

P 1P 2P 3 IM
DT (DMID)T
M
DM 1
3
Any
P 1P 2P 3 IM
DT (DM2ID)T
M
DM 2
3

(DMID)T
M
DM < D + 13
P 1P 2P 3 (DM2ID)T
M
DM D + 13 Negative
(DM3ID)T
M
DM D + 23
Condition (4.12)
P 1P 2P 3 2IM
DT DM < 2
3

P 1P 2P 3 Any
2IM
DT (DMID)T
M
DM 2
3
P 1P 2P 3
P 1P 2P 3 IM
DT DM < 2
3

P 1P 2P 3 IM
DT (DMID)T
M
DM D + 13 Any
P 1P 2P 3 IM
DT (DM2ID)T
M
DM > D + 13
Condition (4.13)
P 1P 2P 3 2IM
DT DM < 2
3

P 1P 2P 3 At least, partly non-positive


2IM
DT (DMID)T
M
Any DM
P 1P 2P 3
P1P2P3 3IM
DT Any DM Positive

Table 4.1: Derivative equations and sign analysis of the DC-link current Iin during healthy con-
ditions.

This analysis is focused on the derivative of Iin for a fault occurrence in Q1 of the
interleaved boost converter in Figure 3.6. Similar results would be obtained for other faults
at Q2 or Q3 considering the symmetry of the circuit. The following analysis also applies to
an interleaved buck converter, whose waveform shapes are similar. After a fault occurrence
in Q1, while P 1 is ON, I1 stops increasing and decreases until it becomes null, as shown
in Figure 4.7a. The negative slope of I1 resulting from that will affect the value of Iin
derivative. Although such transient can have consequences similar to the faulty steady
state conditions (as in Figure 4.7), its occurrence is random and it does not provide enough
information for a general fault diagnosis. Following this transient, the derivative of I1 under
steady state conditions is zero as shown in Figure 4.7b.

After a power switch open-circuit fault, the current Iin ripple frequency is equal to the
phase current ripple frequencies, resulting in a significant ripple increase. Therefore, unlike
48 Chapter 4. Fault Analysis of Non-Isolated DC-DC Converters

the previous analysis, the whole switching period must be analyzed in order to calculate the
derivative of current Iin . The variation of the signs of current Iin derivatives during different
intervals with the same number of ON switching commands already indicates the presence
of a fault in one power switch. This is evident in Figure 4.7, where Iin derivative sign is
positive for intervals P 1P 2P 3 and P 1P 2P 3 while it is negative for interval P 1P 2P 3, unlike
in healthy state where it is positive for all intervals. Therefore, Iin derivative analysis during
particular intervals carries important information for fault detection and localization.

(a) (b)

Figure 4.7: Phase currents and divided switching pattern intervals: (a) transient after open-circuit
fault at Q1 for condition 4.12; (b) steady-state with an open circuit at Q1 for condition 4.12.

Similarly to the previous analysis, equations (4.6) and (4.10) were used to calculate the
derivative values of current Iin to prove the mentioned variation of signs and the result is
presented in Tables 4.2, 4.3 and 4.5, for conditions (4.11), (4.12) and (4.13), respectively.
In Table 4.3, the different signs among Iin derivative values during different intervals with
the same number of ON switching command signals are not so obvious as in 4.2 and 4.5.
Though, a further analysis according to divided subintervals of DM during these intervals,
which is presented in Table 4.4, permits to conclude that the sign of current Iin derivative
actually differs at least in one of the two groups with the same number of ON switching
command signals.

4.2.2 Simulation Results

The prototype previously presented in section 3.2.2 was used to study the three-phase
interleaved DC-DC converter faulty behaviour.The PV module operating conditions are a
temperature of 25 C and a solar irradiation of 850W.m2

For this simulation, whose results are presented in Figure 4.8, the interleaved boost
DC-DC converter has operated in CCM with a D around 0.8 and DM equal to 1. A fault
4.2 Interleaved DC-DC Converters 49

Interval Iin derivative values DM condition Sign


Condition (4.11)
0 DM < 1
3
P 1P 2P 3 (DMID)T
M
DM 1
3
0
(DM2ID)T
M
DM 2
3
IM 2
DT DM < 3
P 1P 2P 3
IM
DT (DMID)T
M
DM 2
3
>0

0 DM < D + 31
P 1P 2P 3 immediately following P 1P 2P 3 (DMID)T
M
DM D + 31 0
(DM2ID)T
M
DM D + 32

Table 4.2: Derivative equations and sign analysis of the DC-link current Iin for condition (4.11)
during a fault in Q1.

Interval Iin derivative values DM condition Sign


Condition (4.12)
P 1P 2P 3 2IM
DT Any DM >0
P 1P 2P 3 2IM
DT DM < 2
3
Any
P 1P 2P 3 IM
DT (DMID)T
M
DM 2
3

0 DM < 2
3
P 1P 2P 3 (DMID)T
M
DM D + 13 0
(DM2ID)T
M
DM > D + 13
IM
DT DM D + 13
P 1P 2P 3 Any
IM
DT (DMID)T
M
DM > D + 13
IM
DT (DMID)T
M
Any DM
P 1P 2P 3 Any
IM 2
DT DM < 3

Table 4.3: Derivative equations and sign analysis of the DC-link current Iin for condition (4.12)
during a fault in Q1.

has been introduced in Q1 at 0.541 s. The current Iin is shown in Figure 4.8a and the phase
current I2 is shown in Figure 4.8c. Its derivative during intervals with all gate command
signals ON is presented in Figure 4.8b and its derivative during intervals with one gate
command signal OFF is presented in Figure 4.8d. During intervals with the same number
of ON switching command signals, the derivative of Iin presents the same sign under healthy
operating conditions, unlike in faulty conditions.
50 Chapter 4. Fault Analysis of Non-Isolated DC-DC Converters

Interval DM < 2
3
2
3 DM D + 31 D + 13 < DM 2D DM > 2D
P 1P 2P 3 >0 >0 >0 >0
P 1P 2P 3 >0 0 0 0
P 1P 2P 3 >0 0 0 0
P 1P 2P 3 0 <0 <0 <0
P 1P 2P 3 >0 >0 0 0
P 1P 2P 3 >0 >0 0 0

Table 4.4: Derivative equations of the DC-link current Iin for condition (4.12) during an open-
circuit fault in Q1, according to the value of DM . The colored cells represent intervals with the
same number of ON switching command signals, but with different Iin derivative signs.

Interval Iin derivative values DM condition Sign


Condition (4.13)
P 1P 2P 3 2IM
DT
Any DM >0
P 1P 2P 3
P 1P 2P 3 2IM
DT (DMID)T
M
Any DM
At least, partly non-positive
P 1P 2P 3 2IM
DT DM < 1

Table 4.5: Derivative equations and sign analysis of the DC-link current Iin for condition (4.13)
during a fault in Q1.

The derivative of Iin during P 1P 2P 3 is positive during both healthy and faulty oper-
ating conditions. This result is expected according to Tables 4.1 and 4.5.

In healthy state, the derivative of Iin during intervals P 1P 2P 3, P 1P 2P 3 and P 1P 2P 3


is partly non-positive as expected from the valid equation in Table 4.5.

During faulty state, the derivative of Iin during P 1P 2P 3 is positive. During intervals
P 1P 2P 3 and P 1P 2P 3, the derivative of Iin is tottally negative. These results are in
agreement with Table 4.5.

Although it does not stop the converter, the DC-link current Iin ripple increases after
the open-circuit fault occurrence, as expected, because of the losing of one interleaved phase
current. Its frequency also changes from 3 kHz to 1 kHz.

4.2.3 Experimental Results

The simulated operating conditions of the interleaved DC-DC converter presented in


section 4.2.2 were also experimentally implemented, although with a slightly lower fre-
quency. Adittionally, the PV module was under random temperature and irradiation con-
4.2 Interleaved DC-DC Converters 51

(a) (b)

(c) (d)

Figure 4.8: Simulation results of the: (a) DC-link current Iin ; (b) Iin derivative sign DX during
intervals with all gate command signals ON; (c) phase current I2 ; (d) Iin derivative sign DX during
intervals with one gate command signal OFF.

ditions available when the results were acquired. The acquired results are presented in
Figure 4.9.

(a) (b)

(c) (d)

Figure 4.9: Experimental results of the: (a) DC-link current Iin ; (b) Iin derivative sign DX during
intervals with all gate command signals ON; (c) phase current I2 ; (d) Iin derivative sign DX during
intervals with one gate command signal OFF.

Both experimental and simulated waveforms of the current Iin presented in Figures
4.9a and 4.8a have very similar shapes. The simulated fault transient is harsher than
the experimental one, because of the imbalanced phases in CCM experimental operation
that might cause different phase current amplitudes. Those amplitudes depend on many
factors [189], including the exactly phase inductance, which is difficult to accurately measure
and reproduce in simulation. Both simulated and experimental phase currents I2 also
present a small amplitude difference.

The current Iin dervative sign DX experimentally acquired is presented in Figure 4.9b
for intervals with all switching command signals ON and in Figure 4.9d for intervals with
only one switching command signal OFF. During the healthy state, the results are very
52 Chapter 4. Fault Analysis of Non-Isolated DC-DC Converters

similar. However, under faulty conditions, the DX values during intervals P 1P 2P 3 and
P 1P 2P 3 are not entirely negative like the ones obtained in simulation. Those initial positive
values presented in Figure 4.9d for intervals P 1P 2P 3 and P 1P 2P 3 are due to the converter
inherent delay related to the non-ideal behavior of the power switches and sensors (which
is not simulated).

4.3 Three-Level Boost DC-DC Converters for PV Applica-


tions

4.3.1 Healthy and Faulty Operating Principle

The normal state of a three-level boost converter is characterized by four operating


modes, according to the power switches conduction state, which can be both conducting,
both turned OFF, or one of them conducting and the other turned OFF. In a PV ap-
plication, they are controlled in order to provide MPPT and DC-link capacitors voltage
balance [172]. However, when a fault occurs both control operations are ceased.

When an open-circuit fault occurs in any of the power switches, the converter stops
working soon. Before that happens, a transient state occurs which is illustrated in Figure
4.10a for an open-circuit fault in S1 and in Figure 4.10b for an open-circuit fault in S2.
After an open-circuit fault occurrence, the remaining healthy IGBT continues working while
it receives impulses to turn it ON (until the control stops working). Only two operating
modes are possible: both power switches OFF and one power switch ON (the healthy one)
and the other OFF (the faulty one). During the period where both S1 and S2 are OFF,
diodes D1 and D2 are forward biased and conducting and both output DC-link capacitors
are charging. Assuming that all elements are ideal and the supply voltage is constant during
one switching period, the equations related to this operation mode are:

diL
vP V 1 + vP V 2 = (L1 + L2 ) + V C1 + V C2 (4.14)
dt

dV C1
C1 + iout = iL (4.15)
dt

dV C2
C2 + iout = iL (4.16)
dt

According to (4.15) and (4.16), the charging current of both capacitors is iL iout .

During the period where S2 is ON and S1 is OFF owing to an open-circuit fault, diode
D2 is reverse biased and diode D1 is conducting. The charging current of capacitor C1 is
4.3 Three-Level Boost DC-DC Converters for PV Applications 53

(a)

(b)

Figure 4.10: Operating modes under faulty state: (a) open-circuit fault in power switch S1; (b)
open-circuit fault in power switch S2.

iL iout and capacitor C2 will be discharging with a current iout . The following equations
are related to this operating mode:

diL
vP V 1 + vP V 2 = (L1 + L2 ) + V C1 (4.17)
dt

dV C1
C1 + iout = iL (4.18)
dt

dV C2
C2 + iout = 0 (4.19)
dt

In case of a fault in switch S1, C2 will charge with a current iL iout while C1 discharges
with a current iout . This will result in an unbalance between both output DC-link capacitors
voltages, according to how much time the control will be sending impulses to the healthy
power switch and the rate of the capacitors charging or discharging. As aforementioned, the
54 Chapter 4. Fault Analysis of Non-Isolated DC-DC Converters

capacitor which is charging or discharging depends on where the open-circuit fault occurs.

The three-level DC-DC converter is proposed for interfacing PV applications. Tem-


perature and radiation changes disturb the MPP of the PV array output voltage, current
and power in different ways which are illustrated in Figure 4.11. If the radiation increases
during the day, the MPP voltage slightly increases and the current MPP has a higher in-
crement, while the MPP power increases (Figures 4.11c and 4.11d). If the solar radiation
decreases, the effect will be the opposite with the MPP voltage slightly decreasing and a
major difference in the reduction of the MPP current, while the MPP power decreases. If
the photovoltaic modules temperature increases, the MPP voltage decreases and the MPP
current value slightly rises, while the MPP power decreases (Figures 4.11a and 4.11b). If the
temperature decreases, the MPP voltage increases and the MPP current hardly decreases,
while the MPP power increases.

(a) (b)

(c) (d)

Figure 4.11: Effects of temperature and irradiation changes on PV modules.

These transients caused by temperature and irradiation at the MPP of PV modules


can be mathematically verified using the Ipv Vpv characteristic of the ideal PV cell [49],
which is
4.3 Three-Level Boost DC-DC Converters for PV Applications 55

qVpv
   
Ipv = Iph I01 exp 1 , (4.20)
a1 kTu

where all variables have already been presented in section 2.1.3. Series and shunt resistive
losses of PV cells were not taken into account, because their influence in the temperature
and irradiation effects on the PV cell behaviour is not significant [32, 190, 191].

Using (4.20), the power of a PV cell is given by

qVpv
    
Ppv = Vpv Ipv = Vpv Iph I01 exp 1 , (4.21)
a1 kTu

When the derivative of power is zero, the MPP current and voltage can be calculated
according to (4.22) and (4.23), which are transcendent equations [32]. Therefore, the lin-
ear dependence of MPP voltage and current on the open-circuit voltage and short-circuit
current, respectively, is used to study the effect of temperature and irradiance on these
electrical parameters [191]. This useful property of PV cells has been proved by numerical
methods [192].

qVmpp qVmppt qVmpp


     
Impp = Iph I01 exp 1 I01 exp , (4.22)
a1 kTu a1 KTu a1 KTu

a1 qVmpp
 
Vmpp = V oc ln 1 + , (4.23)
a1 kTu a1 kTu

Applying the short-circuit condition, Vpv = 0, in equation (4.20) the short-circuit current
Isc is

Isc = Iph , (4.24)

The light induced current I01 depends linearly on the solar irradiation and also increases
with the temperature [49] according to equation (2.4). The same properties are valid for
the short-circuit current Isc and the current at the MPP Impp . Their values increase with
temperature and solar irradiation.

The open-circuit condition is Ipv = 0 and, by applying it to equation (4.20), the open-
circuit voltage can be calculated and related to the short-circuit current according to equa-
tion [32]

a1 kTu Iph a1 kTu Isc


   
V oc = ln ln . (4.25)
q I01 q I01

Equation (4.25) shows clearly that the open-circuit voltage depends logarithmically on
56 Chapter 4. Fault Analysis of Non-Isolated DC-DC Converters

the short-circuit current, which in turn depends linearly on solar irradiation. Therefore, this
results in a logarithmic dependence of the open-circuit voltage on irradiance [32], meaning
that the open-circuit voltage, and subsequently the voltage at the PV cell MPP, increases
with solar irradiation (although its effect on the voltage at the PV cell MPP is lower than
on the current).

To study the effect of the temperature on the open-circuit voltage, the saturation
current equation must be substituted into equation (4.25), because it is dependent on
temperature as shown in the following equation [32]:

EG0
 
I01 = HTu exp , (4.26)
kTu

where H and are temperature independent constants and EG0 is the band gap extrapolated
to the absolute zero.

Considering that the temperature effect on the short-circuit current is very small and
can be neglected [32], the derivative of the open-circuit voltage as a function of temperature
is given by

dV oc
EG0
q V oc + kT
q
u

= . (4.27)
dTu Tu

The parameters of equation (4.27) can be substituted by their values for silicon (which
is the most common material used in PV cells) and the result will be 2.3 mV / C [32],
showing that the open-circuit voltage, and subsequently, the voltage at the PV cell MPP
decrease with the temperature.

According to the mentioned properties of the voltage and current at the PV cell MPP,
the effect of temperature and solar irradiation on maximum power can be easily derived.
As the temperature increases, the current at PV cell MPP increases, but the effect on the
voltage is more significant. Therefore, the effect of temperature on the PV cell maximum
power will be similar to the effect that it has on the voltage at the PV cell maximum point.
The solar irradiation has similar qualitative effects on both electrical parameters at the PV
cell MPP. Consequently, the effect will be qualitatively the same on the maximum power
point. During a fault transient, the power and current suddenly drop, while the voltage
increases (until the photovoltaic array open-circuit value).

4.3.2 Simulation Results

Figure 4.12 shows both healthy and faulty behaviour of the PV power system presented
in section 3.2.3.1, under a constant irradiation value of 315 W.m2 and a temperature value
of 20 C.
4.3 Three-Level Boost DC-DC Converters for PV Applications 57

(a) (b)

(c) (d)

(e) (f)

Figure 4.12: Simulation results for an open-circuit in switch S1 and S2: (a) MPPT control input
voltage for an open-circuit in switch S1; (b) MPPT control input voltage for an open-circuit in
switch S2; (c) MPPT control input current for an open-circuit in switch S1; (d) MPPT control
input current for an open-circuit in switch S2; (e) output DC-link capacitor voltages (V C1 and
V C2) for an open-circuit in switch S1; (e) output DC-link capacitor voltages(V C1 and V C2) for
an open-circuit in switch S2.

Before a fault occurrence, during the normal state, both modules in series are used to
track the maximum power point of the array. Therefore, the input of the MPPT algorithm
control is the sum of both modules output voltages until the instant when a fault occurs.
When a power switch open-circuit fault is introduced, the PV array output voltage increases
(until its open circuit voltage) while its output current and power suddenly decreases. Under
faulty state, the DC-link capacitors voltage is not properly controlled which will result in
their imbalance presented in Figures 4.12e and 4.12f for an open-circuit fault in switches S1
and S2, respectively. If an open-circuit fault occurs in switch S1, the capacitor C1 charges
while the capacitor C2 discharges, as shown in Figure 4.12e. In Figure 4.12f, the opposite
situation occurs due to an open-circuit fault in switch S2.

4.3.3 Experimental Results

Using the PV system setup described in section 3.2.3.1, experimental results for the
healthy and faulty operation of this system were acquired and are presented in Figure 4.13.
58 Chapter 4. Fault Analysis of Non-Isolated DC-DC Converters

Although the climatic conditions of solar irradiance and temperature were not monitored,
an effort was made to tune these parameters on the simulation model to obtain similar
PV modules voltage, current and power values. Although there are still some very small
quantitative differences, the simulation and experimental results are in good agreement.

(a) (b)

(c) (d)

(e) (f)

Figure 4.13: Experimental results for an open-circuit in switch S1 and S2: (a) MPPT control
input voltage for an open-circuit in switch S1; (b) MPPT control input voltage for an open-circuit
in switch S2; (c) MPPT control input current for an open-circuit in switch S1; (d) MPPT control
input current for an open-circuit in switch S2; (e) output DC-link capacitor voltages (V C1 and
V C2) for an open-circuit in switch S1; (e) output DC-link capacitor voltages(V C1 and V C2) for
an open-circuit in switch S2.

4.4 Bidirectional Half-Bridge DC-DC Converter

4.4.1 Healthy and Faulty Operating Principle

For high average inductor current values, the healthy operation of a bidirectional non-
isolated half-bridge DC-DC converter is similar for both control approaches previously
presented in section 3.2.1, although the switching pattern differs, according to Figure 4.14
for control approach of Figure 3.4a and to Figure 4.15 for the control approach presented
in Figure 3.4b.
4.4 Bidirectional Half-Bridge DC-DC Converter 59

(a) (b)

Figure 4.14: Healthy and faulty switching pattern and inductor current waveforms of a bidirec-
tional half-bridge converter with high average inductor current using the control approach of Figure
3.4a when the power is transferred from: (a) the LV bus to the HV bus; (b) the HV bus to the LV
bus.

(a) (b)

Figure 4.15: Healthy and faulty switching pattern and inductor current waveforms of a bidirec-
tional half-bridge converter with high average inductor current using the control approach of Figure
3.4b when the power is transferred from: (a) the LV bus to the HV bus; (b) the HV bus to the LV
bus.

When the power is transferred from the low voltage side (LV bus) to the high voltage
side (HV bus), which is the conventional positive direction, the converter operates as a boost
converter. Independently of the control approach used, for high average inductor current
values, only IGBT Q2 and diode D1 conduct. While Q2 is ON, the inductor voltage is

VL = VLV . (4.28)

Using (4.1) and (4.28), the inductor current slope results positive. When diode D1 is
forward biased (period during which Q1 gate command signal is OFF and Q2 gate command
signal is ON or OFF, according to the control approach used), the inductor voltage is

VL = VLV VHV . (4.29)


60 Chapter 4. Fault Analysis of Non-Isolated DC-DC Converters

As previously mentioned, the converter is operating in boost mode, where

VLV < VHV . (4.30)

Therefore, while D1 is conducting, the inductor current slope is negative.

The converter operates in buck mode when the power is transferred from the HV bus
to the LV bus, which is considered the negative direction. When switch Q1 is ON, the
inductor voltage is given by equation (4.29). Applying condition (4.30) to equation (4.29),
the inductor current slope results negative.

For high average inductor current values, diode D2 is forward biased when Q1 gate
command signal is OFF (independently of the Q2 gate command signal state, according to
the control approach used) and the inductor voltage is given by (4.28). The slope of the
inductor current is positive while diode D1 is conducting.

For low average inductor current values, the two control approaches can have different
inductor current waveforms in healthy conditions. As shown in Figure 4.16, using the con-
trol approach of Figure 3.4a, the inductor current waveform is always continuous, because
both switches are synchronously switched. The control approach of Figure 3.4b always
deactivates one switch, according to the required current direction. Therefore, the inductor
current is blocked from going in the reverse direction and it can be discontinuous, as shown
in Figure 4.17.

(a) (b)

Figure 4.16: Healthy and faulty switching pattern and inductor current waveforms of a bidirec-
tional half-bridge converter with high average inductor current using the control approach of 3.4a
when the power is transferred from: (a) the LV bus to the HV bus; (b) the HV bus to the LV bus.

An open-circuit fault in one of the switches can have different consequences, according
to the operating mode, the faulty switch and the control approach. If an open-circuit fault
occurs in switch Q1 while the converter is operating in boost mode, the consequences will
not be severe for the current or voltage regulation and it might remain unnoticed on all
converter electrical parameters until buck mode operation is required. Oppositely, if an
4.4 Bidirectional Half-Bridge DC-DC Converter 61

(a) (b)

Figure 4.17: Healthy and faulty switching pattern and inductor current waveforms of a bidirec-
tional half-bridge converter with high average inductor current using the control approach of 3.4b
when the power is transferred from: (a) the LV bus to the HV bus; (b) the HV bus to the LV bus.

open-circuit fault occurs in switch Q2, the circuit operation is compromised. The inductor
current instantaneous absolute value will decrease until zero. Using the control approach
of Figure 3.4b, the faulty operation behaviour is similar to the unidirectional non-isolated
topologies. However, for conventional negative power flow (which corresponds to the buck
mode), after an open-circuit fault in switch Q1, the inductor current will rise until zero,
independently of the gate command signal. For the control method of Figure 3.4a, the
inductor current consequences of an open-circuit fault are similar until it reaches the zero
value. After reducing its absolute value to zero, the inductor current sign changes and its
waveform is discontinuous.

4.4.2 Simulation Results

The faulty analysis introduced in the previous section 4.4.1 has been validated in the
bidirectional half-bridge converters interfacing the batteries in the wind power system pre-
sented in section 3.2.1.1. Only the control aprroach of Figure 3.4a is used, because the fault
analysis of the control approach of Figure 3.4b is similar to the one already presented for
single-ended unidirectional DC-DC converters in section 4.1.

An open-circuit fault has been introduced in IGBTs Q1I and Q1V in each half-bridge
converter of the system under two different operating modes. The simulation results ob-
tained for the gate command signals and inductor currents are presented in Figure 4.18.
62 Chapter 4. Fault Analysis of Non-Isolated DC-DC Converters

(a) (b)

(c) (d)

(e) (f)

Figure 4.18: Simulation results for an open-circuit fault in switch Q1I and Q1V : (a) Q1I gate
command signal; (b) Q2I gate command signal; (c) Q1V gate command signal; (d) Q2V gate
command signal; (e) Inductor current IBattI with a high average value; (f) Inductor current IBattV
with a low average value and signal variation during healthy state.

In Figure 4.18e, the battery pack BattI (see Figure 3.5) is charged with a current of
4 A and a fault is introduced in IGBT Q1I at 0.1587 seconds. During healthy conditions,
the current waveform has a positive slope while the gate command signal P 1I is ON, and
a negative one while the gate command signal P 2I is ON. After the fault occurrence, the
current goes up to zero independently of the gate command signals state. Then, the current
will be discontinuous.

Figure 4.18f shows the inductor current IBattV which has a transient caused by a fault
introduced at IGBT Q1V at 0.1569 seconds. During healthy operation, the converter
inductor current waveform is characterized by a sign variation. However, similarly to Figure
4.18e, the current waveform has a positive slope while the gate command signal P 1V is
ON, and a negative one while the the gate command signal P 2V is ON. After the fault
occurrence, both switches are not synchronously switched. Therefore, the current waveform
changes into discontinous.
4.4 Bidirectional Half-Bridge DC-DC Converter 63

4.4.3 Experimental Results

The faulty analysis of bidirectional half-bridge DC-DC converters was also experimen-
tally implemented and studied using the setup presented in section 3.2.1.1. The acquired
results are presented in Figure 4.19. Despite the irregularities on the experimental inductor
current waveform and quantitative differences, both simulation and experimental results
are in relatively good agreement.

(a) (b)

(c) (d)

(e) (f)

Figure 4.19: Experimental results for an open-circuit fault in switch Q1I and Q1V : (a) Q1I
gate command signal; (b) Q2I gate command signal; (c) Q1V gate command signal; (d) Q2V gate
command signal; (e) Inductor current IBattI with a high average value; (f) Inductor current IBattV
with a low average value and signal variation during healthy state.
64 Chapter 4. Fault Analysis of Non-Isolated DC-DC Converters
Chapter 5

Fault Diagnostic Methods for


Non-Isolated DC-DC Converters

Fault diagnostic methods make economic sense in many applications, such as industrial
processes, power supply systems, renewable energy generators power interfaces, telecom-
munications, automotive industry, among others. Their development has been embracing
several topologies of inverters [15, 193] and controlled rectifiers [193]. Recently, fault di-
agnostic methods have been extended to DC-DC converters as well, due to their vital
role in power systems, such as DC power supplies and automotive applications. An open-
circuit power switch diagnostic method for two cascaded buck non-isolated converters is
presented in [19, 20]. This method uses the measured output voltage and current at the
source and load converters. The third statistical moment of the measured signals provides
the required information about the location and type of the fault. Diagnostic methods for
isolated topologies are presented in [2123, 194]. The diagnostic method presented in [21]
only detects the MOSFET fault of a Zero-Voltage-Switching (ZVS) DC-DC converter at
the starting moment, which is a disadvantage as it does not prevent fault propagation
during the operation of the DC-DC converter. It uses the integral and peak values of the
DC-link current patterns. Another diagnostic method limited to ZVS DC-DC converters
is presented in [22]. It detects any open-circuit power switches fault using the transformer
primary voltage of the converter followed by triggering an active phase shifted in the control
system to locate it. The magnetic near field of DC-DC converters is used as a diagnostic
variable by analyzing its Fast Fourier Transform [23]. Despite its wide application (both
isolated and non-isolated DC-DC converters topologies are covered), it requires high pro-
cessing time due to the use of a neural network. Less complex fault diagnostic methods
for non-isolated DC-DC converters have been proposed in [24, 29], but only applicable to
the CCM of operation. However, the operation of non-isolated converters in DCM is also
important for some power applications, where both reduction of power switching losses and
inductance size are critical aspects [195]. The method presented in [29] uses three induc-

65
66 Chapter 5. Fault Diagnostic Methods for Non-Isolated DC-DC Converters

tor current values which are sampled at different gate command signal instants during a
switching period. These values are compared and different conditions are used to detect an
healthy or a faulty operation. One of the conditions used to ensure that the circuit is in a
normal status (i. e., healthy condition) is that two of the sampled points should have almost
the same magnitudes. Such condition is ambiguous and it might lead to the definition of
thresholds, which is not included in [29]. The method presented in [24] is based on two
algorithms to detect faults at any conditions and guarantees a fast detection time. It can
be used to detect both open- and short-circuit faults, but, as mentioned, it does not apply
to all non-isolated DC-DC converters operating modes.

Fault diagnostic methods can be generically assigned into two different categories, ac-
cording to the type of data used for fault diagnostics [196]: methods based on the analysis
of converter variables [19, 23, 24, 29, 197] and methods based on the power switches voltages
and currents [196, 198200]. The latter ones are dependent on the use of power switch
drivers with embedded voltage-sensing and current-sensing circuits or additional hardware,
while the former ones only use common control variables and add a few operations to per-
form the fault diagnostic algorithm. Despite the mentioned drawbacks, the latter category
is very flexible and it can be applied to any power converter.

This literature review highlights that further research is needed in order to develop fault
diagnostic methods that are independent of the converters operating conditions, robust
against transients, and that present a straightforward implementation, without any extra
sensors. The fault diagnostic methods here proposed are based on the the analysis of
converter variables already provided in Chapter 4 and aim to accomplish all mentioned
advantages [201, 202].

5.1 Fault Diagnosis in Single-Ended Unidirectional DC-DC


Converters

5.1.1 Algorithm

For diagnosing an open-circuit fault in unidirectional DC-DC converters, the inductor


current values sampled while the gate command signal rises (Iinc ) and decreases (Idec )
are hold and compared. While the gate command signal is ON, the inductor current
slope is positive and, consequently, the current values sampled while the gate command
signal is rising Iinc are lower than the values sampled at the end of the gate command
signal (while it is decreasing) Idec . Under faulty conditions, the opposite situation occurs,
because the slope is negative or this slope condition does not apply such as in DCM mode.
Therefore, comparison (5.1) is used as a fault diagnostic variable, whether the converter
is operating in CCM or DCM. Delays and dead times related to the circuit operation do
not affect this fault diagnostic variable. However, unpredictable transients might occur,
5.1 Fault Diagnosis in Single-Ended Unidirectional DC-DC Converters 67

such as output or input voltage changes due to load or source variations, turning condition
(5.1) true and consequently leading to false diagnostic alarms. Using the example of a buck
converter presented in Figure 3.2, if condition (4.3) is not verified due to a possible transient
condition (even for a very short period of time), condition (5.1) is verified leading to a false
diagnostic. To avoid this, condition (4.3) is necessary for validating the fault diagnostic
method. Equivalent conditions should be applied to other single-ended unidirectional DC-
DC converters (for example, a boost converter should have an output voltage greater than
its input voltage).

Iinc Idec (5.1)

The described open-circuit fault diagnostic method for unidirectional non-isolated DC-
DC converters is presented in Figure 5.1. While input Cond in Figure 5.1 is ON, the value
of the inductor current I is sampled and hold. This method does not require any extra-
sensors, because it uses only control variables. Even for voltage regulation, the inductor
current I is usually monitored for safety purposes.

Figure 5.1: Fault diagnostic method for unidirectional buck DC-DC converters.

5.1.2 Simulation Results

The fault diagnostic method previously introduced was applied to the buck DC-DC
converter of the prototype presented in section 3.2.1.1 under two different conduction op-
erating modes, although it could be adapted for any single-ended DC-DC converter. The
faulty behaviour of this converter has already been presented in section 4.1. Using the
same operating conditions of the acquired results in sections 4.1.2 and 4.1.3, the simulation
results for the inductor current IW and the fault diagnostic variables Iinc , Idec and FQ are
presented in Figure 5.2.

As in sections 4.1.2 and 4.1.3, under the CCM operating conditions presented in Figure
5.2a, the open-circuit fault occurs at 4 ms. The current decreases during an interval of
approximately 2 ms and condition 5.1 becomes true at 4.5 seconds (which is the fault
68 Chapter 5. Fault Diagnostic Methods for Non-Isolated DC-DC Converters

detection instant). Under DCM operating conditions presented in Figure 5.2b, the open-
circuit fault in switch Q is introduced at 6.4 ms. The fault is detected at 6.8 ms, which is
when condition 5.2b becomes true.

(a) (b)

(c) (d)

(e) (f)

Figure 5.2: Simulation results of a buck DC-DC converter operating with an open-circuit fault:
(a) inductor current IW in CCM; (b) inductor current IW in DCM; (c) auxiliary fault diagnostic
variables in CCM; (d) auxiliary fault diagnostic variables in DCM; (e) fault diagnostic variable FQ
in CCM; (d) fault diagnostic variable FQ in DCM.

5.1.3 Experimental Results

The previously mentioned operating conditions were experimentally reproduced (as in


section 4.1.3). These experimental results are presented in Figure 5.3 and they are very
similar to the simulation ones, except for the current ripple magnitude which presents a
very small variation in the experimental results (this analysis has been already presented
in section 4.1.3).

Furthermore, the behaviour of the fault diagnostic variables during transients is also
presented in Figure 5.4. The reference current of the buck DC-DC converter has been
increased in Figure 5.4a and decreased in Figure 5.4b. Such transients do not cause false
alarms. The auxiliary fault diagnostic variables are presented in Figures 5.4c and 5.4d.
Condition 5.1 only causes false alarms if a harsh transient disturbes the output voltage
5.2 Fault Diagnosis in Interleaved DC-DC Converters 69

(a) (b)

(c) (d)

(e) (f)

Figure 5.3: Experimental results of a buck DC-DC converter operating with an open-circuit fault:
(a) inductor current IW in CCM; (b) inductor current IW in DCM; (c) auxiliary fault diagnostic
variables in CCM; (d) auxiliary fault diagnostic variables in DCM; (e) fault diagnostic variable FQ
in CCM; (d) fault diagnostic variable FQ in DCM.

to be lower than the converter input voltage (influencing the inductor current slope as
explained in section 5.1.1). However, such anomalous situation is already covered by this
method by simply comparing both voltages (Figure 5.1). Therefore, the fault diagnostic
method is immune to transients.

5.2 Fault Diagnosis in Interleaved DC-DC Converters

5.2.1 Algorithm

The derivative of the DC-link current Iin for a switch open-circuit fault was analyzed
in the previous section and it was evident that it reveals important information concerning
power switch fault detection. A carefull study of Iin derivative during each interval with
the same number of ON switching command signals even allows to clearly identify the
faulty power switch. Therefore, the inputs for the fault diagnostic method are the DC-link
current, the switching signal commands for the IGBTs and the duty-cycle. A simplified
block diagram of the proposed diagnostic method for interleaved DC-DC converters is
70 Chapter 5. Fault Diagnostic Methods for Non-Isolated DC-DC Converters

(a) (b)

(c) (d)

Figure 5.4: Experimental results of a buck DC-DC converter during transients: (a) inductor
current IW (increasing reference current); (b) inductor current IW (decreasing reference current); (c)
fault diagnostic variables Iinc and Idec (increasing reference current); (d) fault diagnostic variables
Iinc and Idec (decreasing reference current).

presented in Figure 5.5. As mentioned previously, the DC-link current is always monitored
for control or circuit safety purposes, and thus no additional sensors are needed to implement
this diagnostic method.

The DC-link current Iin must be analyzed during particular intervals with the same
number of ON switching command signals. Those intervals are represented by X which
stands for the following signals: P 1P 2P 3, P 1P 2P 3, P 1P 2P 3, P 1P 2P 3, P 1P 2P 3, P 1P 2P 3.
Each IGBT switching command signals are used to obtain the required intervals to analyse
the Iin derivative. This initial process is illustrated in Figure 5.6 for interval P 1P 2P 3 (a
similar process is used for all other intervals).

Each Iin derivative sign interval DX related to a signal X (any interval with the same
number of ON switching command signals) will be analyzed for counting its positive values
(N pX ), the non-positive (N npX ) and their sum (N psX ), only while signal X is turned ON.
A sample time Tc is used and the process is illustrated in Figure 5.7.

Figure 5.8 illustrates the data processing for determining if the DC-link current deriva-
tive is totally non-positive or positive during the whole considered interval. A constant Nd
is considered to take into consideration the converter delays (such as non-ideal behavior of
power switches, delays and dead-times [24]) in these calculations. This constant is related
to the delay time Td :

Td
Nd = . (5.2)
Tc
5.2 Fault Diagnosis in Interleaved DC-DC Converters 71

Figure 5.5: Open-circuit switch fault diagnosis method for interleaved DC-DC converters.

The non-positive fall of any considered interval X is detected and multiplied by each
counter output, in order to use only the total number of counts. The total N pX is subtracted
to the total N npX . The resulting sign will determine if, during the analyzed interval X, DX
is mostly positive or non-positive. If they are equal, no fault has certainly occurred and the
process will not proceed. If one of the conditions is satisfied, the largest resultant value will
be compared to the total N psX , which should be bigger than 2Nd . Otherwise, the method
cannot be applied or it will lead to dubious results. This condition is crucial for the method
application and also a limitation, because a fault can only be identified if the number of the
total positive counts of the constant switching interval fulfills this requirement. Asymmetry
in the number of samples of the same constant switching pattern group can also lead to
false diagnostic results, as the diagnostic method derives from the similarity in the DC-
link current Iin during the healthy state. Duty-cycle pronounced variations during load
transients and control issues can affect the symmetry of these intervals. To avoid false
diagnostics and improve the robustness of the method against the mentioned asymmetry,
N psX of each constant switching pattern of the same type are summed up and divided by
the total number of phases, which in this particular case is 3. Only the analyzed intervals
that respect the condition of an N psX greater or equal to this calculated value can be
used for diagnosing a fault. Extreme asymmetries can be a limitation for the diagnostic
72 Chapter 5. Fault Diagnostic Methods for Non-Isolated DC-DC Converters

Figure 5.6: Fault diagnostic method input initial process: DC-link current Ii n derivative sign
divided into fault diagnostic method required intervals (exemplified by interval P 1P 2P 3).

Figure 5.7: Fault diagnostic method first analysis of DC-link current Iin derivative sign intervals
DX : counting process of positive values (N pX ), the non-positive (N npX ) and their sum (N psX ),
only while signal X is turned ON.

method. However, they should not be common for the good operation of the converter.
Figure 5.9 illustrates this procedure for intervals P 1P 2P 3, P 1P 2P 3, and P 1P 2P 3, but it
is also applied to P 1P 2P 3, P 1P 2P 3 and P 1P 2P 3.

Finally, Figure 5.10 shows how all data that have been processed can be used to detect
and identify a faulty switch. A variable F1 is defined as a fault diagnostic variable for an
open-circuit fault in Q1. Similar variables, F2 and F3 , can be defined for an open-circuit
fault in Q2 and Q3, respectively. The criteria are different according to the values of the
duty-cycle. For example, if a fault occurs in Q1 at a duty-cycle lower than 31 , Iin current
derivative will be non-positive during interval P 1P 2P 3 and positive during P 1P 2P 3. Those
conditions are presented in Table 5.1 for any power switch fault and any duty-cycle value,
according to Iin derivative sign during a particular interval. Besides fulfilling the conditions
of Table 5.1, MX must also be true, so that the considered intervals are symmetric (avoiding
false alerts during transients). Figure 5.10b shows the process used to take into account all
the variables needed for fault diagnostic. Each counter sums up the input variables during
one switching period. The switching gate signal P 1 is used, in this case, as reset signal
after one switching period. If the positive rise counter satisfies the output conditions of its
sum value and duty cycle, during one switching period, an open-circuit fault is diagnosed.

This method is also robust to current imbalance between phases, as it is based only on
the phase current derivatives sum (which remains the same whether they are imbalanced
or not).

Figure 5.11 shows graphically the fault detection process for an open-circuit fault at
5.2 Fault Diagnosis in Interleaved DC-DC Converters 73

Figure 5.8: Fault diagnostic process for determining if DX is totally positive or non-positive.

Figure 5.9: Fault diagnostic method process for verifying if all the analyzed intervals are symmetric.

IGBT Q1 during condition (4.12). The faulty transient mode resulted in a similar Iin
derivative sign as to the faulty steady state mode. Consequently, the fault is detected in
less than one switching period (the switching command signal P 1 is used as reference for
the mentioned switching period and for the proposed method, as shown in Figure 5.5).
If the faulty transient had different consequences on the Iin derivative sign or if the fault
had occurred at a later stage during the analysed period, another switching period would
be necessary for fault detection. Therefore, after fault occurrence, the minimum detection
time can be less than one phase current switching period and its maximum value can be
less than two switching periods, if the fault occurs at the beginning of the reference gate
command signal (before the conditions for fault detection have been checked) or closer to
its end, respectively.
74 Chapter 5. Fault Diagnostic Methods for Non-Isolated DC-DC Converters

(a) (b)

Figure 5.10: Fault diagnostic method final process for detecting an open-circuit fault in switch
Q1: (a) general process; (b) positive rise counter used in Figure 5.10a.

D Fault Diagnostic Criteria


Q1 Open-Circuit Fault
D< 1
3 (DP 1P 2P 3 0)AN D(DP 1P 2P 3 > 0)
1
3 D 2
3 [(DP 1P 2P 3 0)AN D(DP 1P 2P 3 > 0)]AN D[(DP 1P 2P 3 > 0)AN D(DP 1P 2P 3 0)]
D> 2
3 DP 1P 2P 3 > 0
Q2 Open-Circuit Fault
D< 1
3 (DP 1P 2P 3 0)AN D(DP 1P 2P 3 > 0)
1
3 D 2
3 [(DP 1P 2P 3 0)AN D(DP 1P 2P 3 > 0)]AN D[(DP 1P 2P 3 > 0)OR(DP 1P 2P 3 0)]
D> 2
3 DP 1P 2P 3 > 0
Q3 Open-Circuit Fault
D< 1
3 (DP 1P 2P 3 0)AN D(DP 1P 2P 3 > 0)
1
3 D 2
3 [(DP 1P 2P 3 0)AN D(DP 1P 2P 3 > 0)]AN D[(DP 1P 2P 3 > 0)OR(DP 1P 2P 3 0)]
D> 2
3 DP 1P 2P 3 > 0

Table 5.1: Fault identification conditions based on duty-cycle and Iin derivative sign during defined
intervals.

5.2.2 Simulation Results

The fault diagnostic method previously described in section 5.2.2 was applied to the
3-phase interleaved DC-DC converter included in the simulation model presented in section
5.2 Fault Diagnosis in Interleaved DC-DC Converters 75

Figure 5.11: Fault detection during condition 4.12 for a fault at Q1.

3.2.3 under the operating conditions defined in sections 4.2.2 and 4.2.3.

Figure 5.12 shows the DC-link current Iin , its derivative sign during specific intervals
(which are auxiliar fault diagnostic variables) and the fault diagnostic variables. A fault has
been introduced in Q1 at 0.541 seconds and it is followed by a short fault transient. During
faulty steady state conditions, the Iin derivative sign is different during intervals with only
one switching command signal OFF. It is totally positive during interval P 1P 2P 3 and
negative during intervals P 1P 2P 3 and P 1P 2P 3. Using the mentioned DX values signals
and the fault diagnostic criteria defined in Table 5.1, it can be easily concluded that there
is a fault in switch Q1. This is in agreement with the fault diagnostic variables results
presented in Figure 5.12c. The fault is identified at 0.5424 seconds. All simulated DX
intervals are symmetrical as it can be easily concluded from Figure 5.12b.

(a) (b)

(c)

Figure 5.12: Simulation results of the: (a) DC-link current Iin ; (b) Iin derivative sign DX during
intervals with one gate command signal OFF; (c) fault diagnostic variables.
76 Chapter 5. Fault Diagnostic Methods for Non-Isolated DC-DC Converters

5.2.3 Experimental Results

The developed fault diagnostic algorithm for interleaved DC-DC converters was also
experimentally validated in laboratory under the same operating conditions used for simu-
lation. The acquired experimental results are presented in Figure 5.13.

The experimental results for the DC-link current Iin and its derivative sign DX have
already been detailedly explained and compared with the simulation ones in section 4.2.3.
Despite the differences between the experimental and simulation results concerning the
variable DX used as an auxiliary variable for fault diagnosis, the fault is diagnosed at
the same time (0.5424 seconds), which means that the method is robust against inherent
operating delays in experimental results. Figure 5.14 shows the diagnostic variables for
these operating conditions. A sample frequency of 20 kHz was used. There is a minor
control phase asymmetry that does not affect the fault diagnostic either.

(a) (b)

(c)

Figure 5.13: Experimental results of the: (a) DC-link current Iin ; (b) Iin derivative sign DX
during intervals with one gate command signal OFF; (c) fault diagnostic variables.

Asymmetric and imbalanced operating conditions have been intentionally applied to


prove the robustness of the method. The phase currents will be slightly imbalanced, espe-
cially if the converter is operating in CCM [156, 189].

Figure 5.15 shows the DC-link current when the interleaved DC-DC converter is oper-
ating in CCM and DM has a value of 1. The duty-cycle D is around 0.48. A fault occurs
at 0.1812 seconds in IGBT Q1.

Figure 5.14 show the variables used for fault diagnostic purposes. The sample frequency
is 20 kHz.

The derivative of Iin during intervals with two IGBTs OFF is presented in Figure 5.16a.
The considered intervals are asymmetric. This is particularly evident in Figure 5.16c, which
shows the number of sampled points in the interval, where the considered signal is ON. The
number of sampled points during the interval P 1P 2P 3 is lower than the average of the same
5.2 Fault Diagnosis in Interleaved DC-DC Converters 77

value for the three considered intervals. Therefore, the diagnostic variables calculated for
this interval are not valid for fault diagnostic and they will be ignored during the application
of the method, avoiding false diagnostics (while limiting the diagnostic for this particular
values of duty-cycle). Before the fault occurrence, the derivative of the current Iin during
each considered interval is majorly non-positive (except for the interval P 1P 2P 3, due to
the mentioned asymmetry). Considering a delay of 50 s, which corresponds to the sample
period, the considered intervals are completely non-positive during healthy state. According
to the valid equations in Table 4.1 (having in mind that this particular case is related to
condition 4.12 and DM is equal to 1), this is the expected result. The changes occurring
during the transient fault are not relevant for the fault diagnostic. During the resultant
steady state, the derivative of Iin during the interval P 1P 2P 3 remains non-positive (as
expected from condition 4.12 with DM = 1 and equations in Table 4.3), but during interval
P 1P 2P 3 it is positive (as expected from condition 4.12 with DM = 1 and the valid equation
in Table 4.3). The consequent result of this analysis in variables aK and cK is presented in
Figures 5.16i and 5.16k, respectively.

Figure 5.16 also presents similar results for the derivative of current Iin during different
intervals with only one IGBT OFF. Both steady states during the healthy and faulty op-
erations are majorly positive, as expected from equations in Tables 4.1 and 4.3. There are
changes occurring during the fault transient, but they do not influence the fault diagnos-
tic, as expected. After the fault occurrence, variables aQ1 , bQ1 and cT 1 are simultaneously
ON, as shown in Figure 5.16, and they permit to identify and locate a fault at IGBT Q1.
Finally, Figure 5.17 shows all the final diagnostic variables F1 , F2 and F3 . Due to the fault
transients (which are different than the steady state expected transients for fault detection),
the fault detection takes slightly more than one switching period (the fault was diagnosed
at 0.18258 s, which is 1.58 ms after the fault has occurred).

In Figure 5.18 (showing Iin , the interleaved converter is operating with a D around 0.2
and DM around 0.56 (DCM operating conditions). In this case, a sampling time of 25 s
has been used (for both sampling data and the diagnostic method). A fault is introduced
in Q1 at 0.9745 seconds.

Figure 5.19 shows the variables used for fault diagnostics and Figure 5.20 shows the
final diagnostic variables.

The operating condition corresponds to 4.11 and, using Table 4.1, the derivative of Iin
during intervals with two IGBTs OFF is mostly positive. After the fault, the derivative of
the current Iin during interval P 1P 2P 3 changes for a non-positive value, as expected from
the valid equation in Table 4.2. Using the respective equation in Table 4.2, the derivative
of the current Iin during interval P 1P 2P 3 remains a positive value. This Iin derivative sign
difference in both mentioned intervals is enough to identify and locate a fault in Q1. The
fault transient has similar consequences to the fault steady state and the fault is accordingly
identified in less than a phase current switching period, i. e., at 0.9751 s (in 0.6 ms).
78 Chapter 5. Fault Diagnostic Methods for Non-Isolated DC-DC Converters

(a)

(b)

(c)

(d)

(e) (f)

Figure 5.14: Experimental results of the variables used for fault diagnostic related to Iin derivative
sign DX during interval X, only while it is ON: (a) Iin derivative sign DX during interval X, (b)
number of positive values N pX ; (c) number of non-positive N npX ; (d) sum of N pX and N npX
(N psX ); (e) variable dK means that DX (the correspondent DX has the same color) is always
non-positive during the considered interval and it is used to diagnose IGBT K; (f) variable bK
means that DX (the correspondent DX has the same color) is always positive during the considered
interval and it is used to diagnose IGBT K.

Figure 5.15: DC-link current Iin .


5.2 Fault Diagnosis in Interleaved DC-DC Converters 79

(a) (b)

(c) (d)

(e) (f)

(g) (h)

(i) (j)

(k) (l)

Figure 5.16: Experimental results of the variables used for fault diagnostic related to Iin derivative
sign DX during interval X, only while it is ON: (a) Iin derivative sign DX during interval X with
one IGBT ON; (b) Iin derivative sign DX during interval X with two IGBTs ON; (c) number of
positive values N pX during interval X with one IGBT ON; (d) number of positive values N pX
during interval X with two IGBTs ON; (e) number of non-positive N npX during interval X with
one IGBT ON; (f) number of non-positive N npX during interval X with two IGBTs ON; (g) sum of
N pX and N npX (N psX ) during interval X with one IGBT ON; (h) sum of N pX and N npX (N psX )
during interval X with two IGBTs ON; (i) variable aK means that DX (the correspondent DX has
the same color) is always non-positive during the considered interval and it is used to diagnose IGBT
K; (j) variable dK means that DX (the correspondent DX has the same color) is always non-positive
during the considered interval and it is used to diagnose IGBT K; (k) variable cK means that DX
(the correspondent DX has the same color) is always positive during the considered interval and it
is used to diagnose IGBT K; (l) variable bK means that DX (the correspondent DX has the same
color) is always positive during the considered interval and it is used to diagnose IGBT K.
80 Chapter 5. Fault Diagnostic Methods for Non-Isolated DC-DC Converters

Figure 5.17: Fault diagnostic variables. Figure 5.18: DC-link current Iin .

(a) (b)

(c) (d)

(e) (f)

Figure 5.19: Experimental results of the variables used for fault diagnostic related to Iin derivative
sign DX during interval X, only while it is ON: (a) Iin derivative sign DX during interval X; (b)
number of positive values N pX ; (c) number of non-positive N npX ; (d) sum of N pX and N npX
(N psX ); (e) variable aK means that DX (the correspondent DX has the same color) is always
non-positive during the considered interval and it is used to diagnose IGBT K; (f) variable cK
means that DX (the correspondent DX has the same color) is always positive during the considered
interval and it is used to diagnose IGBT K.

Figure 5.20: Fault diagnostic variables.


5.3 Fault Diagnosis in Three-Level Boost DC-DC Converters for PV
Applications 81

5.3 Fault Diagnosis in Three-Level Boost DC-DC Converters


for PV Applications

5.3.1 Algorithm

The conventional three-level boost converter employs a reduced number of control vari-
ables (only the input current and voltage for MPPT purposes and the output DC link
capacitor voltages for balancing them) and they are also useful for fault detection, which
is desirable for avoiding extra hardware and costs. The fault diagnostic method must be
reliable and robust against transients related to load variations or climatic conditions which
affect the photovoltaic array. Table 5.2 summarizes this information that allows diagnosing
an open-circuit fault occurrence by monitoring the input power, voltage and current. These
variables must be filtered so that sensor noises, small transients related to the MPPT con-
trol and variable ripples do not compromise the diagnostic results. This implies the use of
a small cut-off frequency, which will delay the signal acquisition. The fault detection is not
affected by this fact, because only the signal qualitative variations (an increase or decrease)
are used for fault detection and their value is not important. Concerning the filter design,
a good performance is obtained with first-order low-pass filters with a frequency between
0.5 and 1 Hz.

Voltage Current Power


Temperature increase Decrease Increase Decrease
Temperature decrease Increase Decrease Increase
Irradiation increase Increase Increase Increase
Fault Increase Decrease Decrease

Table 5.2: Effects of temperature, irradiation and open-circuit power switch fault on PV array
electrical characteristics.

After detecting the fault, it is necessary to locate it, because this topology has two
power switches. For fault localization, the unbalance between the capacitors voltage is
used. Both power switches open-circuit fault transients have been previously analysed and
in both cases the DC link capacitors voltage balance is lost, but with different consequences.
Therefore, a fault diagnostic variable can be created using the difference between the output
DC link capacitors voltages:

Z = V C1 V C2. (5.3)

If Z is bigger than a pre-defined positive threshold h, then the faulty switch is S1. If
it is smaller than a negative threshold h, then the faulty switch is S2. This threshold
82 Chapter 5. Fault Diagnostic Methods for Non-Isolated DC-DC Converters

was empirically chosen during tests and 3% value of DC-link voltage was found to be a
good compromise between the diagnosing time and the usual transients caused by climatic
variations (which are slow and cause reduced transients in balancing the output DC-link
capacitors voltage). The whole diagnostic method is summarized in Figure 5.21.

Figure 5.21: Fault diagnostic method.

5.3.2 Simulation Results

Figure 5.22 shows the simulation results of the implementation of the proposed fault
diagnostic method for an open-circuit fault in power switches S1 and S2 in the PV power
system presented in section 3.2.3.1. The simulation operating conditions are the same
already introduced in section 4.3.2.

As demonstrated in section 4.3, when a power switch open-circuit fault is introduced,


the PV array output voltage increases (until its open circuit voltage, if the fault is not
detected) while its output current and power suddenly decreases. These conditions, together
with the output DC-link capacitor voltage imbalance (shown in Figures 5.22g and 5.22h
for an open-circuit fault in power switches S1 and S2, respectively), allow to diagnose
the fault. The fault diagnostic variables turn on when the output DC-link capacitors
voltage unbalance exceeds the threshold value. These fault diagnostic variables are shown
in Figures 5.22i and 5.22j. As soon as the fault is detected (S1Of f and S2Of f change
their output value in Figures 5.22i and 5.22j), the converter is immediately reconfigured for
fault-tolerant operation, which is presented in section 6.2, changing the converter monitored
variables. The fault detection time depends on the charge/discharge rate of the output
DC-link capacitors as shown in Figures 5.22g and 5.22h for an open-circuit fault in power
switches S1 and S2, respectively.
5.3 Fault Diagnosis in Three-Level Boost DC-DC Converters for PV
Applications 83

(a) (b)

(c) (d)

(e) (f)

(g) (h)

(i) (j)

Figure 5.22: Simulation results for an open-circuit in switch S1 and S2: (a) MPPT control input
voltage for an open-circuit in switch S1; (b) MPPT control input voltage for an open-circuit in
switch S2; (c) MPPT control input current for an open-circuit in switch S1; (d) MPPT control
input voltage for an open-circuit in switch S2; (e) input power for an open-circuit in switch S1;
(f) input power for an open-circuit in switch S2; (g) output DC-link capacitor voltages (V C1 and
V C2) for an open-circuit in switch S1; (h) output DC-link capacitor voltages(V C1 and V C2) for
an open-circuit in switch S2 (i) fault diagnostic variables for an open-circuit in switch S1; (j) fault
diagnostic variables for an open-circuit in switch S2.
84 Chapter 5. Fault Diagnostic Methods for Non-Isolated DC-DC Converters

(a) (b)

(c) (d)

(e) (f)

(g) (h)

(i) (j)

Figure 5.23: Experimental results for an open-circuit in switch S1 and S2: (a) MPPT control
input voltage for an open-circuit in switch S1; (b) MPPT control input voltage for an open-circuit
in switch S2; (c) MPPT control input current for an open-circuit in switch S1; (d) MPPT control
input voltage for an open-circuit in switch S2; (e) input power for an open-circuit in switch S1;
(f) input power for an open-circuit in switch S2; (g) output DC-link capacitor voltages (V C1 and
V C2) for an open-circuit in switch S1; (h) output DC-link capacitor voltages(V C1 and V C2) for
an open-circuit in switch S2 (i) fault diagnostic variables for an open-circuit in switch S1; (j) fault
diagnostic variables for an open-circuit in switch S2.
5.4 Fault Diagnosis in Bidirectional Half-Bridge Converters 85

5.3.3 Experimental Results

The fault diagnostic method proposed was validated experimentally using the experi-
mental setup described in section 3.2.3.1 and the same operating conditions of section 4.3.3.
The acquired results are presented in Figure 5.23 and are pretty similar to the simulation
ones as expected after tuning the simulation climatic conditions to obtain similar PV output
voltage, current and power.

5.4 Fault Diagnosis in Bidirectional Half-Bridge Converters

5.4.1 Algorithm

The fault diagnostic method for the single-ended unidirectional converters presented in
section 5.1.1 can be modified to diagnose faults in bidirectional half-bridge converters by
including both gate command signals and changing condition (5.1) according to the inductor
current behaviour already detailedly explained in section 4.4 (new conditions using each
gate command signal given by 5.4 and 5.5). If both switches are not synchronously switched
(using the control approach of Figure 3.4b), the inductor current behaviour is similar to
the unidirectional converters already presented. Therefore, their fault diagnostic algorithm
is presented in Figure 5.24.

Iinc1 Idec1 (5.4)

Iinc2 Idec2 (5.5)

If an open-circuit fault occurs during light load conditions as presented in Figure 4.17
for the control approach of Figure 3.4a, the converter will continue its operation with a
discontinuous inductor current, even if the fault stops the voltage or current regulation.
Therefore, in this case, the fault diagnostic algorithm presented in Figure 5.24 will not
be effective alone. Under severe faults, the discontinuous inductor current and its low
average value after the open-circuit fault occurrence will provide the means for another
fault detection algorithm.

The fault diagnostic method for the bidirectional half-bridge converter with synchronous
switching is presented in Figure 5.25 and it comprises both algorithms. The algorithm inside
the dashed line is added to the previously described one.

In healthy condition, the inductor current is continuous. Consequently, its arithmetic


average value, while the gate command signal is ON, is close to the sum of both interval
extreme values divided by 2. It can be defined as
86 Chapter 5. Fault Diagnostic Methods for Non-Isolated DC-DC Converters

Figure 5.24: Fault diagnostic method for the bidirectional half-bridge converter using the control
approach of 3.4b.

Iinc + Idec
IEV = . (5.6)
2

The inductor current average value can also be defined as:

n
1 X
ILAM = Ii , (5.7)
N i=1

where N is the number of samples of the inductor current I and Ii is the value of the
inductor current sample i. During healthy condition and CCM operation, both variables
(5.6) and (5.7) have similar values.

If an open-circuit fault occurs with severe consequences for the converter regulation, the
inductor current instantaneous absolute value will decrease (until zero and then changes
its sign) and it will become discontinuous. Therefore, under faulty condition and DCM
operation, the arithmetic mean value ILAM will be no longer close to variable IEV and
it will be instead closer to the inductor current values sampled during the gate command
sign rising Ilow (which are the lowest inductor current values). The algorithm presented
in Figure 5.25 uses this property to diagnose any open-circuit faulty switch. The gate
command signal used to compute the diagnostic variables permits to identify which switch
is faulty as shown in Figure 5.25.
5.4 Fault Diagnosis in Bidirectional Half-Bridge Converters 87

Figure 5.25: Fault diagnostic method for the bidirectional half-bridge converter using the control
approach of Figure 3.4a.
88 Chapter 5. Fault Diagnostic Methods for Non-Isolated DC-DC Converters

During one period and while the gate command signal is ON, all inductor current
samples are summed up and divided by their total number to calculate the arithmetic
mean value ILAM . The total number of samples is calculated by sampling the command
gate signal using the same sample time. By sampling and holding the inductor current
values while the gate command signal rises and decreases, these values are summed up and
divided by two to calculate variable IEV . As soon as the command gate signal used for
calculation starts decreasing, the following variables are determined:

dA = |IEV ILAM |, (5.8)

dB = |ILAM Ilow |, (5.9)

where Ilow corresponds to the inductor current values sampled during the gate command
signal rising. Comparing variables dA and dB while one of the gate command signals is
decreasing, if condition (5.10) is true (which means that ILAM is closer to Ilow than IEV ),
the open-circuit switch fault is detected.

dA > dB (5.10)

The fault diagnostic method is only valid while the following condition is true:

VLV < VHV . (5.11)

5.4.2 Simulation Results

The fault diagnostic method presented for the bidirectional half-bridge using the control
approach of Figure 3.4b is very similar the one already used for unidirectional DC-DC
converters in section 5.1. After all, using the control approach of Figure 3.4a, the half-
bridge bidirectional converter is an unidirectional boost converter on one direction and an
unidireccional buck converter on the opposite direction. Therefore, only the fault diagnostic
method developed for the control approach of Figure 3.4a is verified in this section. It was
applied to the bidirectional half-bridge converters interfacing the batteries in the wind
power system presented in section 3.2.1.1. The operating conditions are the same already
used for introducing the converter fault analysis in section 4.4.2.

The battery pack BattI is charged with a current of 4 A and a fault is introduced in
IGBT Q1I at 0.1587 seconds. After the fault occurrence, condition (5.4) becomes true and
the inductor current average value increases. These diagnostic variables are presented in
Figure 5.26. The fault was detected at 0.1592 seconds.
5.4 Fault Diagnosis in Bidirectional Half-Bridge Converters 89

(a)

(b)

(c)

Figure 5.26: Simulation results for an open-circuit fault in switch Q1I: (a) Inductor current IBattI ;
(b) auxiliary fault diagnostic variables Iinc1 and Idec1 ; (c) fault diagnostic variables FI1 and FI2 .

Figure 5.27: Simulated inductor current IBattV .

In Figure 5.27, a fault is introduced in IGBT Q1V at 0.1569 seconds. During healthy
operation, the converter inductor current waveform is characterized by sign variation. The
fault diagnostic variables are shown in Figure 5.28. Only when a fault occurs and causes
discontinuous inductor current with a very low mean value, ILAM is closer to Iinc than IEV ,
and the fault is detected.
90 Chapter 5. Fault Diagnostic Methods for Non-Isolated DC-DC Converters

(a) (b)

(c) (d)

(e) (f)

(g) (h)

(i)

Figure 5.28: Simulation results for an open-circuit fault in Q1V: (a) auxiliary fault diagnostic
variable ilow1 ; (b) auxiliary fault diagnostic variable ilow2 ; (c) auxiliary fault diagnostic variable
iLAM 1 ; (d) auxiliary fault diagnostic variable iLAM 2 ; (e) auxiliary fault diagnostic variable iEV 1 ;
(f) auxiliary fault diagnostic variable iEV 2 ; (g) auxiliary fault diagnostic variables dA1 and dB1 ; (h)
auxiliary fault diagnostic variables dA2 and dB2 (i) fault diagnostic variables FV 1 and FV 2 .
5.4 Fault Diagnosis in Bidirectional Half-Bridge Converters 91

5.4.3 Experimental Results

The fault diagnostic method theoretically validated in the previous section was also
experimentally tested on the setup described in section 3.2.1.1. The operating conditions
are the same used for the simulation and the acquired experimental results are presented
in Figures 5.29, and 5.31.

(a)

(b)

(c)

Figure 5.29: Experimental results for an open-circuit fault in switch Q1I: (a) Inductor current
IBattI ; (b) auxiliary fault diagnostic variables Iinc1 and Idec1 ; (c) fault diagnostic variables FI1 and
FI2 .

Figure 5.30: Inductor current IBattV experimentally obtained.

Both simulation and experimental results are qualitatively similar. The fault is intro-
duced and detected at the same time. However, some irregularities on the inductor current
92 Chapter 5. Fault Diagnostic Methods for Non-Isolated DC-DC Converters

waveform are clearly perceptible on the experimental results, as already mentioned in sec-
tion 4.4.3. The small quantitative differences between the simulation and experimental
results of the inductor current IBattV values led to a very small discrepancy between the
diagnostic variables presented in Figures 5.28 and 5.31.

(a) (b)

(c) (d)

(e) (f)

(g) (h)

(i)

Figure 5.31: Simulation results for an open-circuit fault in Q1V: (a) auxiliary fault diagnostic
variable ilow1 ; (b) auxiliary fault diagnostic variable ilow2 ; (c) auxiliary fault diagnostic variable
iLAM 1 ; (d) auxiliary fault diagnostic variable iLAM 2 ; (e) auxiliary fault diagnostic variable iEV 1 ;
(f) auxiliary fault diagnostic variable iEV 2 ; (g) auxiliary fault diagnostic variables dA1 and dB1 ; (h)
auxiliary fault diagnostic variables dA2 and dB2 (i) fault diagnostic variables FV 1 and FV 2 .
5.4 Fault Diagnosis in Bidirectional Half-Bridge Converters 93

Figure 5.32 presents details of a fault transient in the inductor current IBattI , which
could potentially cause a false diagnostic. Condition (5.4) is true during part of the transient
interval zoomed. However, this is not a consequence of a fault in the bidirectinal half-
bridge converter, but of a temporary transient where DC-bus voltage Vdc is lower than the
battery set BattI output voltage VBattI . Therefore, condition 5.11 is not fulfilled and a false
diagnostic is avoided. This harsh transient was caused by introducing a fault in the buck
converter IGBT Q, which has been presented in section 5.1.3.

(a) (b)

(c) (d)

(e) (f)

Figure 5.32: Experimental results of a fault transient in the inductor current IBattI , which could
potentially cause false diagnostics: (a) inductor current IBattI ; (b) fault diagnostic variables Iinc1
and Iinc2 ; (c) gate command signal P 1I; (d) result of the condition Vdc > VBattI ; (e) battery bank
BattI output voltage VBattI ; (f) load input voltage Vdc .
94 Chapter 5. Fault Diagnostic Methods for Non-Isolated DC-DC Converters
Chapter 6

Fault Tolerant Strategies Applied


to Non-Isolated DC-DC Converters

In telecommunications power supplies, fault-tolerant strategies based on redundancy


are a common practice. For each supplying circuit, two independent ones are provided so
that the continuity of their service is guaranteed in case of failure [2]. Other fault-tolerant
strategies based on reconfiguration and modularity are more cost-effective.

Only a very few fault tolerant designs have been proposed in the literature for DC-
DC converters. In [2527], fault-tolerant strategies for multilevel DC-DC converters are
proposed. They are based on redundancy and remedial control and they are meant for
modern vehicle designs. Modular DC-DC converters topologies as the one presented in [28]
for space applications ensure the continuity of service with a reasonable performance using
the remaining healthy modules. A reconfiguration fault-tolerant strategy is proposed in [29]
for a hybrid electric vehicle power system including a controlled rectifier, a boost DC-DC
converter, a bidirectional half-bridge DC-DC converter and an inverter. It uses only a
redundant switch and circuit control reconfiguration to sustain the functionalities of the
DC-DC converters, in case of a fault in their power switches. Another reconfiguration
fault-tolerant strategy is proposed in [30] by introducing a redundant phase-leg shared by
the three converters of a standard series hybrid electric vehicles, in case any power switch
fails. In [21], an isolated phase-shifted full-bridge DC-DC converter is reconfigured into an
asymmetrical half-bridge converter to operate in post-fault conditions under a reduced load.
It uses a redundant winding on the secondary side of the used high frequency transformer
used.

In this chapter, three fault-tolerant strategies are addressed. Although the interleaved
DC-DC converters have fault-tolerant capability, it is possible to improve it by readapting
the phaseshift angle between the converter phases using the still available healthy ones,
in order to avoid current ripple beyond reasonable limits. A new reconfiguration fault-

95
Chapter 6. Fault Tolerant Strategies Applied to Non-Isolated DC-DC
96 Converters

tolerant scheme for the three-level boost converter in a PV power systems using batteries
as storage devices is proposed and fully described. Finally, a conventional fault-tolerant
strategy based on redundancy and modularity is also introduced for bidirectional half-bridge
DC-DC converters, which can be extended to any other DC-DC converter or application.

6.1 Interleaved DC-DC Converter Inherent Fault Tolerance

Interleaved DC-DC power converters are composed of n phases, whose currents are
phase-shifted by 2/n radians. Such operation has great benefits on efficiency, size, thermal
management, modularity and ripple cancellation output current [155158]. Additionally, if
one IGBT in one phase becomes faulty, the converter does not stop its operation [10], unlike
other DC-DC converter topologies [24]. Therefore, it has inherent fault-tolerance. On the
one hand, this fact shows that the intrinsic reliability of the converter is good, but on the
other hand the benefits related to its low output current ripple are lost and this can have
serious consequences for the load being supplied [203]. Therefore, an open-circuit fault
diagnostic can be useful to avoid ripple beyond reasonable limits and improve the fault
tolerance behavior of interleaved converters by readapting their control for n 1 phases
(using only the healthy ones).

6.2 Three-Level Boost DC-DC Converter Reconfiguration


Strategy for PV Applications

After having detected the fault and identified the faulty switch, the three-level boost
DC-DC converter is reconfigured for post-fault operation. The circuit and the whole control
system are represented with much detail in Figure 6.1. When the three-level boost converter
is in the normal state, the control system uses the PV array total output voltage and current
for the MPPT and the output DC link capacitors voltage for balancing it. Therefore, the
PV array output voltage sensor signals are summed up and only one of the current sensors
is used. During normal operation, power switch S1 is used for MPPT control and power
switch S2 is used for balancing the output DC link capacitors voltage.

After fault detection, the control system used to keep the converter operation is dif-
ferent, depending on whether the fault has occurred in power switch S1 or power switch
S2. However, the concept is the same: when one of the power switches stops working,
the three-level boost converter is reconfigured into a conventional two-level boost converter
with MPPT control for only half of the module array while the other half is still operating
without MPPT control for not overcharging the remaining part of the circuit. As a result,
after a fault occurrence, the converter does not stop working, although providing less power.

If power switch S1 suffers an open-circuit fault, after its detection, the TRIAC is
6.2 Three-Level Boost DC-DC Converter Reconfiguration Strategy for PV
Applications 97

Figure 6.1: Fault-tolerant control strategy.

triggered, providing the hardware reconfiguration for a two-level boost converter. The
control system is also reconfigured according to Figure 6.1. The measured current now is
that from the other current sensor, providing only the second half PV array output current
and also only its output voltage is monitored. Then, power switch S2 stops controlling the
output DC-link capacitor voltages (which are not effective after a fault) and provides MPPT
control for this rearranged two-level boost configuration. Alternatively, if power switch S2
suffers an open-circuit fault, after its detection, the TRIAC is also triggered, providing again
the hardware reconfiguration for a similar two-level boost converter. The control system
is reconfigured with a similar goal, but the changes are different. The monitored current
remains the same, which is the first half PV array output current (see Figure 6.1), and
also only its output voltage is monitored. Subsequently, S1 continues the MPPT control,
but only for a two-level boost converter. The remaining half of the PV array continues
supplying the batteries, but without MPPT control.

6.2.1 Simulation Results

The proposed post-fault reconfguration strategy of a three-level boost DC-DC con-


verter has been simulated in the PV power system presented in section 3.2.3.1 and using
the operating conditions introduced in section 4.3.2. The obtained results are presented in
Chapter 6. Fault Tolerant Strategies Applied to Non-Isolated DC-DC
98 Converters

Figure 6.2, including the normal, the faulty and the reconfiguration states (with simultane-
ous fault detection and fault-tolerant strategy) for an open-circuit fault in power switches
S1 and S2.

(a) (b)

(c) (d)

(e) (f)

(g) (h)

Figure 6.2: Simulation results for an open-circuit in switch S1 and S2: (a) MPPT control input
voltage for an open-circuit in switch S1; (b) MPPT control input voltage for an open-circuit in
switch S2; (c) MPPT control input current for an open-circuit in switch S1; (d) MPPT control
input current for an open-circuit in switch S2; (e) input power for an open-circuit in switch S1;
(f) input power for an open-circuit in switch S2; (g) output DC-link capacitor voltages (V C1 and
V C2) for an open-circuit in switch S1; (h) output DC-link capacitor voltages(V C1 and V C2) for
an open-circuit in switch S2.

Both fault and the simultaneous fault detection and reconfiguration instants are marked
by a blue dashed line in Figure 6.2. Until the fault is not detected, the circuit operates
under faulty state. The duration of this mode depends on the charge/discharge rate of the
output DC-link capacitors as shown in Figures 6.2g and 6.2h for an open-circuit fault in
6.2 Three-Level Boost DC-DC Converter Reconfiguration Strategy for PV
Applications 99

power switches S1 and S2, respectively. As soon as the fault is detected, the converter
is immediately reconfigured into a partly two-level boost converter with MPPT control
only for one of the PV modules, while the other still produces energy, but without MPPT
control.

6.2.2 Experimental Results

The proposed fault-tolerant strategy for three-level boost DC-DC converters for PV
applications has also been experimentally validated in the experimental setup described in
section 3.2.3.1. The operating conditions are the same already used to acquire the results
presented in sections 4.3.3 and 5.3.3. Figure 6.3 presents the converter monitored variables
during healthy, faulty and reconfiguration modes. The results are qualitatively similar.
There are some small quantitative differences and the transients during reconfiguration are
harsher in experimental results than in simulation.

Each PV module output power, voltage and current are shown in Figure 6.4 for an
open-circuit fault in power switches S1 and S2, respectively, and under different climatic
conditions. When the fault occurs in S1, after fault detection, PV module P V 1 (see Figure
3.13) is monitored for MPPT control. After the reconfiguration for post-fault operation, it
produces the same power it was already producing before the fault occurrence and it presents
the same output voltage and current, unless the climatic conditions change. PV module
P V 2 still produces power, but its output power, voltage and current are very different
from the previous ones during the healthy state, because its MPP is not being tracked
anymore, otherwise it would overcharge the converter. Similar results were obtained for a
fault occurring in S2, but, in this case, PV module P V 2 is monitored for MPPT control
while the other is not.

The reconfigured converter produces approximately 30% less power than the original
one in both cases and for different climatic conditions. At the same time, it introduces
higher stresses on the remaining healthy IGBT, due to the output DC-link capacitor voltage
unbalance, making the IGBT voltage 30% higher. Besides, the PV module with MPPT
control has higher output voltage and current ripple. These are the main disadvantages
and limitations of this strategy, which should be expected, because they are related to the
disadvantages of two-level boost power converters over the three-level ones. However, the
converter remains operating until it can be replaced without stop supplying batteries, which
can be useful in critical applications where the power supplying should be uninterrupted.
Chapter 6. Fault Tolerant Strategies Applied to Non-Isolated DC-DC
100 Converters

(a) (b)

(c) (d)

(e) (f)

(g) (h)

Figure 6.3: Experimental results for an open-circuit in switch S1 and S2: (a) MPPT control
input voltage for an open-circuit in switch S1; (b) MPPT control input voltage for an open-circuit
in switch S2; (c) MPPT control input current for an open-circuit in switch S1; (d) MPPT control
input current for an open-circuit in switch S2; (e) input power for an open-circuit in switch S1;
(f) input power for an open-circuit in switch S2; (g) output DC-link capacitor voltages (V C1 and
V C2) for an open-circuit in switch S1; (h) output DC-link capacitor voltages(V C1 and V C2) for
an open-circuit in switch S2.
6.3 Energy Storage Systems Modularity and/or Redundancy 101

(a) (b)

(c) (d)

(e) (f)

Figure 6.4: Experimental results for an open-circuit in switch S1 and S2: (a) PV module output
voltage for an open-circuit in switch S1; (b) PV module output voltage for an open-circuit in switch
S2; (c) PV module output current for an open-circuit in switch S1; (d) PV module output current
for an open-circuit in switch S2; (e) PV module output power for an open-circuit in switch S1; (f)
PV module output power for an open-circuit in switch S2.

6.3 Energy Storage Systems Modularity and/or Redundancy

Single-ended unidirectional and bidirectional half-bridge DC-DC converters are charac-


terized by a reduced number of components, easy control and low cost, which makes them
attractive for redundant and modular power systems. A modular fault-tolerant energy
storage system using bidirectional half-bridge DC-DC converters is presented in this sec-
tion. It can be extended to single-ended unidirectional DC-DC converters in any modular
application, such as PV power generation.

Energy storage devices are mandatory in standalone power systems, due to renewable
sources intermittence, difficult climatic conditions forecast and load demand fluctuations.
As a backup and energy buffer, the energy storage system should be the most adaptable
and reliable part of a renewable energy based power supply. If all else fails, it should re-
main a flexible and robust solution for load supply. Batteries might seem like the preferable
option, due to their well-known mature technology and wide application. Moreover, it is
not very frequent that a commercial single unit of batteries (or of any other energy storage
Chapter 6. Fault Tolerant Strategies Applied to Non-Isolated DC-DC
102 Converters

device, such as SCs) fulfills all the requirements in a typical power application [204]. Many
devices are usually connected in series and/or in parallel to accomplish the desired power
and voltage levels. As a result, instead of using all devices connected in a single full-size
bank, many smaller banks can be formed and matched by modular converters working to-
gether while optimizing the volumetric efficiency of the devices (the volumetric efficiency
of a battery bank, for example, comprising several batteries in series is determined by the
battery with the lowest state-of-charge) [205]. Then, energy storage sizing is an important
matter, usually accomplished using local climatic data during different seasons. However,
an early sized energy storage system might become unsuitable, due to unusual climatic
conditions, generators damages, maintenance periods or even an update of the system ac-
cording to demand changes by increasing or reducing its generation. A modular design of
the energy storage system would provide easy power resizing, backup time improvement,
and redundant modules by the simple addition (or removal) of equipment and power con-
verter modules (instead of acquiring a whole new power converter). Furthermore, modular
converters benefit from improved reliability, smaller components with consequent reduced
converter volume (smaller inductances also improve control performance) [155] and opti-
mized control strategies that permit to reduce output ripple [205], better efficiency, and
active storage devices balance [206].

Different modular bidirectional DC-DC converter topologies and successful accomplish-


ments have been presented for many applications, such as series connection of dual active
bridges [206], a topology based on the cascaded half-bridge modular multilevel converter
with common DC bus [207], full bridge modular topology for UPS (uninterruptible power
supplies) applications [208], interleaved multiphase DC-DC converters for automotive appli-
cations [209], modular architectures in power systems [210] and different modular topologies
for SCs applications [211]. Their compliance with different storage devices led to their suc-
cessful application in storage systems with both SCs and batteries [212, 213]. A modular
converter configuration with dual active bridge converters for this application (and the en-
ergy management scheme) is presented in [205]. Its complexity and high component rating
are its main disadvantages. However, such modular architectures have not been studied for
redundancy fault-tolerant operation.

A small modular converter architecture is proposed in section 3.2.1.1. It features bat-


teries of the same technology divided into two banks. One is controlling the system DC-bus
voltage, while the other is storing the surplus energy generated and supplying energy if
needed. Using the fault diagnostic method described in section 5.4, a fault can be eas-
ily located and the modular converter architecture is subsequently reconfigured to keep
the continuity of its service. If an open-circuit fault occurs in the battery bank BattV
converter, which controls the DC-bus voltage, it will have damaging consequences for the
whole power system. Without a constant DC-bus voltage, the load requirements will no be
accomplished. Therefore, the control of battery bank BattI can be reconfigured to provide
DC-bus voltage regulation.
6.3 Energy Storage Systems Modularity and/or Redundancy 103

6.3.1 Simulation Results

The battery pack BattI is charged with a current of 2 A and the converter connected
to the battery pack BattV is regulating the DC-bus voltage at 50V. A fault is introduced
in IGBT Q1V at 0.1054 seconds as shown in Figure 6.5. After fault occurrence, the fault
diagnostic method presented in section 3.2.1.1 locates the faulty switch at 0.1061 seconds.
As soon as the fault is detected, the control of the converter connected to the battery pack
BattI is reconfigured to provide DC-bus voltage regulation. In Figure 6.5c, after the fault
detection, the DC-bus voltage Vdc stops increasing while the battery pack BattI converter
control is reconfigured for voltage regulation. The fault diagnostic method has a main
role to provide the service uninterruptedly when fault-tolerant strategies are applied to
power systems architectures. In this case, as presented in Figure 6.5, the DC-bus voltage
perturbations due to the fault occurrence were minimum.

(a)

(b)

(c)

Figure 6.5: Simulation results: (a) battery pack BattV converter inductor current; (b) battery
pack BattI converter inductor current; (c) DC-bus voltage.

6.3.2 Experimental Results

The fault diagnostic method theoretically validated in the previous section was also
experimentally tested on the setup described in section 3.2.1.1. The operating conditions
Chapter 6. Fault Tolerant Strategies Applied to Non-Isolated DC-DC
104 Converters

are the same used for the simulation and the acquired experimental results are presented
in Figure 6.6. Apart from quantitative differences between both acquired IBattI fault-
transient and the Vdc ripple, both simulation and experimental results are in relatively
good agreement.

(a)

(b)

(c)

Figure 6.6: Experimental results: (a) battery pack BattV converter inductor current; (b) battery
pack BattI converter inductor current; (c) DC-bus voltage.
Chapter 7

Conclusions and Future Work

An efficient telecommunications network is the foundation upon which


an information society is built.

By Talal Abu-Ghazaleh

7.1 Conclusions

Telecommunication networks equipment are energy dependent and safety critical re-
quiring high efficient technologies and uninterruptible power supply. In this thesis different
online fault diagnostic methods for many non-isolated DC-DC converters topologies meant
for telecommunications wind and PV power systems are proposed. Self-diagnosing power
converters allow the easy introduction of fault-tolerant control strategies which keep the
converter normal operation, although with a reduced performance. Such control approaches
are also presented, including a new reconfiguration fault-tolerant strategy for PV applica-
tions which have inherent modularity structure.

This work starts with a research study of the models for PV arrays, wind generators
and batteries. These technologies basic features were presented, which are essential to
understand their electrical operation and to provide them an optimized control power in-
terface. Literature models were briefly reviewed and the ones presenting the best accuracy
and computational effort compromise for electrical circuits simulation have been choosen
for simulating power systems.

Taking into account the existing technologies for small and medium wind and PV
power conversion systems and the DC telecommunications power supply requirements,
non-isolated DC-DC converters have been identified as key power conversion interfaces.
Therefore, four different topologies have been selected to be studied while interfacing PV-

105
106 Chapter 7. Conclusions and Future Work

batteries and wind-batteries power systems, namely single-ended unidirectional DC-DC


converters, three-phase interleaved DC-DC converters, three-level boost DC-DC converters
and bidirectional half-bridge DC-DC converters. All of them have proven their efficiency
and ability in such systems. Their control and operating principle for these applications is
introduced.

The healthy and faulty behaviour of the mentioned DC-DC converter topologies is thor-
oughly studied. Both analyses are the basis to optimize the converter operation, improve
converter design, develop simulation models, prevent damages, predict malfunctions, build
fault protection devices, develop fault diagnostic methods and apply fault-tolerant strate-
gies. The theoretical analysis using circuit equations and waverform schemes is validated
with simulation and experimental results.

Regarding the provided faulty analysis, it was concluded that an open-circuit switch
fault leads to the converter operation stoppage of the single-ended unidirectional DC-DC
converter and three-level boost DC-DC converters. An open-circuit switch fault on bidi-
rectional half-bridge converts only prevents its operation in one of the directions. Among
the topologies studied, the interleaved DC-DC converter is the only one that has inherent
fault tolerant operation. Apart from increased ripple amplitude and ripple frequency reduc-
tion, an open-circuit switch fault might remain unnoticed in interleaved DC-DC converters.
However, such open-circuit switch fault consequences might not comply with the applica-
tion requirements, particularly when supplying a sensitive load (such as telecommunications
apparatus).

The fault analysis has provided enough data to develop four novel fault diagnostic tech-
niques for the mentioned DC-DC converters topologies. These fault diagnostic techniques
use only control variables and do not require extra sensors. Besides increasing cost and
complexity, extra sensors also decrease the converter reliability. The proposed fault diag-
nostic techniques do not deal with complex calculations, such as Fourier analysis or mean
values, which are related to a harder computational effort and high processing time. They
might have one derivative and many counters, sums and a few multiplications, which are
basic mathematical calculations and do not represent a problem for most DSPs/FPGAs in
the market for PV or wind applications.

Among the existent fault diagnostic methods, only a few are able to diagnose open-
circuit switch faults in single-ended unidirectional DC-DC converters, three-phase inter-
leaved DC-DC converters, three-level boost DC-DC converters and bidirectional half-bridge
DC-DC converters. However, some of them require extra-sensors to be implemented. Oth-
ers are based on the Fast Fourier Transform of magnetic near field or Kalman filters and
surely require higher processing time with more computational effort than the diagnostic
method proposed which uses only basic mathematical operations. Some methods are re-
stricted for only some operating conditions (such as CCM, for example). Therefore, with
the exception of the method developed for the three-level boost DC-DC converter which
7.2 Recommendations for Future Work 107

is meant only for PV aplications, there are no other methods in the literature with the
extension of the ones proposed.

Furthermore, all methods are robust against transients. The interleaved DC-DC con-
verter diagnostic method is even robust against converter inherent delays, phases control
asymmetry and phases current imbalance. However, those situations should be avoided for
the good converter operation.

Finally, fault-tolerant strategies taking advantage of the developed fault diagnostic


methods are proposed, because the fault-tolerant reconfiguration starts as soon as the fault
is detected. Only the three-level boost reconfiguration for fault-tolerant operation in PV
applications is originally provided by this work. The remaining strategies are based on
redundancy and inherent interleaved DC-DC converter fault tolerance.

The single-ended unidirectional DC-DC converters and the bidirectional half-bridge


DC-DC converters are characterized by a reduced number of components, easy control
and low cost, which makes them attractive topologies for applications that might require
modularity or redundancy.

The three-level converter boost proposed is more cost-effective than redundancy or


multiphase DC-DC converter. It rearranges the converter components so that the healthy
part of the three-level boost converter is converted into a two-level one. This converter
remains operating with MPPT control using only one of the PV modules, which is an
advantage over the operation of PV modules without such control. The other PV module
is operating without MPPT control for not overcharging the reconfigured converter. The
fault-tolerant operation of the converter results in less produced power, more PV module
output voltage and current ripple and higher stresses on the power components. However,
the converter rebuilds itself to keep its operation after fault detection and it proves to be an
effective and low cost choice for all those applications in which the uninterrupted supplying
is a critical issue.

7.2 Recommendations for Future Work

This work contributes to the development of self-diagnosing DC-DC converters. How-


ever, not all DC-DC converter topolgies have been covered. Isolated topologies were not
considered although they do not benefit from a wide range of open-circuit switch fault
diagnostic methods. Further research is needed to increase their reliability for DC telecom-
munications power supplies.

The studied systems only deal with PV, wind and battery technologies. There are
other potential sources and technologies whose advantages can bring many benefits for
telecommunications power supplies, such as fuel cells, supercapacitors, flywheels, among
108 Chapter 7. Conclusions and Future Work

others. These components rely upon specific power converters and control strategies, which
also require fault diagnostic methods for an improved reliability.

The experimental results were performed on laboratory setups. A complete base station
supplied by only renewable energies and storage devices is possible and it is a very chal-
lenging task. It would be a lot more interesting to see the developed work implemented in
a real application. However, such ambitious power system requires higly optimized energy
management strategies, individual control of each power generator and correct sizing of the
power converters for an efficient operation. All these subjects require further research.

This work presents a fault diagnostic method for bidirectional half-bridge DC-DC con-
verters. Many applications rely on energy storage. They are mandatory as a backup in
critical applications, for instance telecommunications (as presented in this work), hospi-
tals, industrial processes, among others. Hybrid and electric cars energy storage units are
charged from the engine or the fuel cell (depending on the type of electric car) or from
the electric grid (in the case of a plug-in hybrid electric vehicle) and are used to power
electric motors and recover energy during braking. Distributed generation, electric grid
and standalone power systems also take advantage from energy storage, by reducing peak
demands and allowing greater flexibility for integrating alternative renewable energy gen-
erators (which are known for their intermittency) while improving its cost-effectiveness,
reliability, power quality, efficiency and reducing environment impact. More recently, mod-
ern electric drive applications with high ratio of the peak to average power and with braking
needs, such as lifts, tooling machines, oil pumps, among others, also use storage devices for
efficiency improvement and uninterruptible operation. Maintenance, energy management,
weight, size, lifetime, cost, energy and power density are important characteristics of these
systems, concerning not only the storage technologies, but also their power interfaces which
play an important role in their operating behaviour. Consequently, they represent a very
important research topic.

Using the new storage technologies, such as supercapacitors, it is possible to optimize


already existent old technologies, such as the diesel generator. The diesel generator fuel
emission can be reduced by operating it at the lowest fuel consumption speed and its
slow dynamic can be compensated by supercapacitors. This is a very interesting research
field that has not received the deserved attention. Further research is required for being
implemented in real power supply applications.
Appendix A

Simulation Parameters

This section presents the parameters of the main components simulation models.

A.1 PV Module BP 4175B

Iscn 5.4 A

V ocn 44 V

V mppn 35.7 V

Imppn 4.9 A

KV 160 mV/ C

KI 0.065 %/ C

Ns 72

Gn 1000 W/m2

Tn 25 + 273.15 K

Rs 0.57

Rsh 149.2

Table A.1: BP 1475B PV module model parameters.

109
110 Appendix A

A.2 PMSG

Rst 0.415

Ld , Lq 5.13 mH

f 0.121 Wb

p 5

Table A.2: PMSG model parameters.

A.3 Battery

Rbatt 0.6

E0 15.9341 V

K 0.015715 V

A 0.43591 V

B 59.8802 (Ah)1

Table A.3: Battery model parameters.

A.4 Electrical Components

Wind Power System

LW 10 mH

LBattV 10 mH

LBattI 10 mH

CW 680 F

CBattV 680 F

CBattI 680 F

Table A.4: Power converter model parameters.


A.4 Electrical Components 111

PV Power System with an interleaved boost dc-dc converter

L1 10 mH

L2 10 mH

L3 10 mH

Cin 2350 F

Cout 680 F

Table A.5: Power converter model parameters.

PV Power System with a three-level boost dc-dc converter

L1 10 mH

L2 10 mH

C1 680 F

C2 680 F

Ci1 4700 F

Ci2 4700 F

Table A.6: Power converter model parameters.


112 Appendix A
Appendix B

Experimental Setup Data

This section presents the manufacturers data from the main equipment and components
used to build the experimental setups.

B.1 PV Module BP 4175B

Figure B.1: PV array and its laboratory interface.

Rated power 175 W

Power tolerance 5%

Nominal voltage 24 V

Limited Warranty 25 years

Table B.1: BP 4175B performance data.

113
114 Appendix B

Maximum power 175 W

Voltage at maximum power 35.7 V

Current at maximum power 4.9 A

Short-circuit current 5.4 A

Temperature coefficient of the short-circuit current (0.0650.015)%/ C

Temperature coefficient of the open-circuit voltage ((16010)mV/ C

Temperature coefficient of power (0.50.05)%/ C

NOCT 472 C V

Table B.2: BP 4175B electrical characteristics data.

B.2 PMSG

Figure B.2: PMSG coupled to a motor (PMSM) in a test bench (the latter emulates the spin of
a wind turbine).
B.3 Battery 115

Rated power 2.2 kW

Frequency 145.8 Hz

Speed 1750 rpm

Voltage 200 V

Current 10.4 A

Table B.3: PMSG nameplate data.

B.3 Battery

Figure B.3: Batteries.

Rated voltage 12 V

Capacity 12 Ah

Table B.4: Battery datasheet data.


116 Appendix B

B.4 Electrical Components

LW 10 mH

LBattV 10 mH

LBattI 10 mH

CW 680 F

CBattV 680 F

CBattI 680 F

Table B.5: Wind power system.

L1 10 mH

L2 10 mH

L3 10 mH

Cin 2350 F

Cout 680 F

Table B.6: PV Power System with an interleaved boost dc-dc converter.


B.4 Electrical Components 117

L1 10 mH

L2 10 mH

C1 680 F

C2 680 F

Ci1 4700 F

Ci2 4700 F

Table B.7: PV Power System with a three-level boost dc-dc converter.


118 Appendix B
References

[1] C. Boccaletti, G. Fabbri, and E. Santini, Innovative solutions for stand alone system
powering, in 29th International Telecommunications Energy Conference, pp. 294
301, 2007.

[2] D. W. Reeve, DC Power System Design for Telecommunications. John Wiley & Sons,
2007.

[3] O. Onar, O. Shirazi, and A. Khaligh, Grid interaction operation of a telecommuni-


cations power system with a novel topology for multiple-input buck-boost converter,
IEEE Transactions on Power Delivery, vol. 25, no. 4, pp. 26332645, 2010.

[4] E. Ribeiro, A. J. M. Cardoso, and C. Boccaletti, Standalone diesel-photovoltaic


system with batteries and supercapacitors as energy storage components for telecom-
munications, in International Telecommunications Energy Conference, pp. 18, 2012.

[5] L. Moore, H. Post, H. Hayden, S. Canada, and D. Narang, Photovoltaic power plant
experience at Arizona public service: a 5-year assessment, Progress in Photovoltaics:
Research and Applications, vol. 13, no. 4, pp. 353363, 2005.

[6] N. Dhere, Reliability of PV modules and balance-of-system components, in IEEE


Photovoltaic Specialists Conference, pp. 15701576, 2005.

[7] A. Ristow, M. Begovic, A. Pregelj, and A. Rohatgi, Development of a methodol-


ogy for improving photovoltaic inverter reliability, IEEE Transactions on Industrial
Electronics, vol. 55, no. 7, pp. 25812592, 2008.

[8] J. Wiles, B. Brooks, and B.-O. Schultze, PV installations, a progress report, in


IEEE Photovoltaic Specialists Conference, pp. 14611464, 2002.

[9] P. Zhang, Y. Wang, W. Xiao, and W. Li, Reliability evaluation of grid-connected


photovoltaic power systems, IEEE Transactions on Sustainable Energy, vol. 3, no. 3,
pp. 379389, 2012.

[10] S. Dhople, A. Davoudi, A. Domnguez-Garca, and P. Chapman, A unified approach


to reliability assessment of multiphase DC-DC converters in photovoltaic energy con-

119
120 REFERENCES

version systems, IEEE Transactions on Power Electronics, vol. 27, no. 2, pp. 739
751, 2012.

[11] A. Golnas, PV system reliability: An operators perspective, IEEE Journal of Pho-


tovoltaics, vol. 3, no. 1, pp. 416421, 2013.

[12] A. Rogers, J. Manwell, J. McGowan, and A. Ellis, Design requirements for medium-
sized wind turbines for remote and hybrid power systems, Renewable Energy, vol. 26,
no. 2, pp. 157 168, 2002.

[13] J. Ribrant and L. Bertling, Survey of failures in wind power systems with focus on
swedish wind power plants during 1997-2005, IEEE Transactions on Energy Con-
version, vol. 22, no. 1, pp. 167173, 2007.

[14] G. Petrone, G. Spagnuolo, R. Teodorescu, M. Veerachary, and M. Vitelli, Reliability


issues in photovoltaic power processing systems, IEEE Transactions on Industrial
Electronics, vol. 55, no. 7, pp. 25692580, 2008.

[15] B. Lu and S. Sharma, A literature review of IGBT fault diagnostic and protection
methods for power inverters, IEEE Transactions on Industry Applications, vol. 45,
no. 5, pp. 17701777, 2009.

[16] J. O. Estima and A. J. M. Cardoso, A new algorithm for real-time multiple open-
circuit fault diagnosis in voltage-fed PWM motor drives by the reference current
errors, IEEE Transactions on Industrial Electronics, vol. 60, no. 8, pp. 34963505,
2013.

[17] J. Li, A. Huang, Z. Liang, and S. Bhattacharya, Analysis and design of active NPC
(ANPC) inverters for fault-tolerant operation of high-power electrical drives, IEEE
Transactions on Power Electronics, vol. 27, no. 2, pp. 519533, 2012.

[18] K. Rothenhagen and F. Fuchs, Performance of diagnosis methods for IGBT open cir-
cuit faults in voltage source active rectifiers, in IEEE 35th Annual Power Electronics
Specialists Conference, vol. 6, pp. 43484354, 2004.

[19] R. Jayabalan and B. Fahimi, Monitoring and fault diagnosis of cascaded multicon-
verter systems in hybrid electric vehicles, in IEEE Conference Vehicle Power and
Propulsion, 2005.

[20] R. Jayabalan and B. Fahimi, Monitoring and fault diagnosis of DC-DC multistage
converter for hybrid electric vehicles, in IEEE International Symposium on Diag-
nostics for Electric Machines, Power Electronics and Drives, pp. 16, 2005.

[21] X. Pei, S. Nie, Y. Chen, and Y. Kang, Open-circuit fault diagnosis and fault-tolerant
strategies for full-bridge DC-DC converters, IEEE Transactions on Power Electron-
ics, vol. 27, no. 5, pp. 25502565, 2012.
REFERENCES 121

[22] S. Y. Kim, K. Nam, H. Song, and H. Kim, Fault diagnosis of a ZVS DC-DC converter
based on dc-link current pulse shapes, IEEE Transactions on Industrial Electronics,
vol. 55, no. 3, pp. 14911494, 2008.

[23] Y. Chen, X. Pei, S. Nie, and Y. Kang, Monitoring and diagnosis for the DC-DC
converter using the magnetic near field waveform, IEEE Transactions on Industrial
Electronics, vol. 58, no. 5, pp. 16341647, 2011.

[24] M. Shahbazi, E. Jamshidpour, P. Poure, S. Saadate, and M. Zolghadri, Open- and


short-circuit switch fault diagnosis for nonisolated dc-dc converters using field pro-
grammable gate array, IEEE Transactions on Industrial Electronics, vol. 60, no. 9,
pp. 41364146, 2013.

[25] K. Ambusaidi, V. Pickert, and B. Zahawi, New circuit topology for fault tolerant
H-bridge DC-DC converter, IEEE Transactions on Power Electronics, vol. 25, no. 6,
pp. 15091516, 2010.

[26] K. Ambusaidi, V. Pickert, and B. Zahawi, Computer aided analysis of fault tolerant
multilevel DC/DC converters, in International Conference on Power Electronics,
Drives and Energy Systems, pp. 16, 2006.

[27] K. Ambusaidi, V. Pickert, B. Zahawi, and B. Ji, PSPICE analysis of PWM con-
trolled inductorless fault tolerant multilevel DC/DC converters, in IET Conference
on Power Electronics, Machines and Drives, pp. 490494, 2008.

[28] V. Choudhary, E. Ledezma, R. Ayyanar, and R. Button, Fault tolerant circuit topol-
ogy and control method for input-series and output-parallel modular DC-DC convert-
ers, IEEE Transactions on Power Electronics, vol. 23, no. 1, pp. 402411, 2008.

[29] T. Park and T. Kim, Novel fault tolerant power conversion system for hybrid electric
vehicles, in IEEE Vehicle Power and Propulsion Conference, pp. 16, 2011.

[30] Y. Song and B. Wang, Analysis and experimental verification of a fault-tolerant HEV
powertrain, IEEE Transactions on Power Electronics, vol. 28, no. 12, pp. 58545864,
2013.

[31] A. Luqe and S. Hegedus, Handbook of Photovoltaic Science and Engineering. Wiley,
2003.

[32] L. Castaner and S. Silvestre, Modelling Photovoltaic Systems Using PSpice. New
York: Wiley, 2002.

[33] H. Yoshioka, S. Nishikawa, S. Nakajima, M. Asai, S. Takeoka, T. Matsutani, and


A. Suzuki, Non hot-spot PV module using solar cells with bypass diode function,
in IEEE Photovoltaic Specialists Conference, pp. 12711274, 1996.
122 REFERENCES

[34] N. Kaushika and N. Gautam, Energy yield simulations of interconnected solar PV


arrays, IEEE Transactions on Energy Conversion, vol. 18, no. 1, pp. 127134, 2003.

[35] B. Rani, G. Ilango, and C. Nagamani, Enhanced power generation from PV array
under partial shading conditions by shade dispersion using su do ku configuration,
IEEE Transactions on Sustainable Energy, vol. 4, no. 3, pp. 594601, 2013.

[36] M. El-Dein, M. Kazerani, and M. Salama, An optimal total cross tied intercon-
nection for reducing mismatch losses in photovoltaic arrays, IEEE Transactions on
Sustainable Energy, vol. 4, no. 1, pp. 99107, 2013.

[37] E. Ortiz-Rivera and F. Peng, Analytical model for a photovoltaic module using
the electrical characteristics provided by the manufacturer data sheet, in Power
Electronics Specialists Conference, pp. 20872091, 2005.

[38] V. Autier, V. Molcrette, and H. Roise, Contribution to the determination of the


elements of photovoltaic panel equivalent electric mode. Application in the maximal
power point available analytical expression, in International Conference on Renew-
able Energies and Power Quality, 2003.

[39] R. Campbell, A circuit-based photovoltaic array model for power system studies,
in 39th North American Power Symposium, pp. 97101, 2007.

[40] R. Chenni, M. Makhlouf, T. Kerbache, and A. Bouzid, A detailed modeling method


for photovoltaic cells, Energy, vol. 32, no. 9, pp. 1724 1730, 2007.

[41] S. Liu and R. Dougal, Dynamic multiphysics model for solar array, IEEE Transac-
tions on Energy Conversion, vol. 17, no. 2, pp. 285294, 2002.

[42] G. Yu, Y. Jung, J. Choi, I. Choy, J.-H. Song, and G. Kim, A novel two-mode MPPT
control algorithm based on comparative study of existing algorithms, in Photovoltaic
Specialists Conference, pp. 15311534, 2002.

[43] K. Hussein, I. Muta, T. Hoshino, and M. Osakada, Maximum photovoltaic power


tracking: an algorithm for rapidly changing atmospheric conditions, IEEE Proceed-
ings Generation, Transmission and Distribution, vol. 142, no. 1, pp. 5964, 1995.

[44] J. Gow and C. Manning, Development of a photovoltaic array model for use in
power-electronics simulation studies, IEE Proceedings - Electric Power Applications,
vol. 146, no. 2, pp. 193200, 1999.

[45] W. Xiao, W. Dunford, and A. Capel, A novel modeling method for photovoltaic
cells, in Power Electronics Specialists Conference, vol. 3, pp. 19501956, 2004.

[46] D. S. H. Chan and J. C. H. Phang, Analytical methods for the extraction of solar-cell
single- and double-diode model parameters from I-V characteristics, IEEE Transac-
tions on Electron Devices, vol. 34, no. 2, pp. 286293, 1987.
REFERENCES 123

[47] W. D. Soto, S. Klein, and W. Beckman, Improvement and validation of a model for
photovoltaic array performance, Solar Energy, vol. 80, no. 1, pp. 78 88, 2006.

[48] Y. Mahmoud, W. Xiao, and H. Zeineldin, A parameterization approach for enhancing


PV model accuracy, IEEE Transactions on Industrial Electronics, vol. 60, no. 12,
pp. 57085716, 2013.

[49] M. Villalva, J. Gazoli, and E. Filho, Comprehensive approach to modeling and


simulation of photovoltaic arrays, IEEE Transactions on Power Electronics, vol. 24,
no. 5, pp. 11981208, 2009.

[50] M. Milosevic, G. Andersson, and S. Grabic, Decoupling current control and maxi-
mum power point control in small power network with photovoltaic source, in Power
Systems Conference and Exposition, pp. 10051011, 2006.

[51] M. Suthar, G. Singh, and R. Saini, Comparison of mathematical models of photo-


voltaic (PV) module and effect of various parameters on its performance, in Interna-
tional Conference on Energy Efficient Technologies for Sustainability, pp. 13541359,
2013.

[52] A. Bellini, S. Bifaretti, and V. Iacovone, Resonant DC-DC converters for photovoltaic
energy generation systems, in International Symposium on Power Electronics, Elec-
trical Drives, Automation and Motion, pp. 815820, 2008.

[53] D. Sera, R. Teodorescu, J. Hantschel, and M. Knoll, Optimized maximum power


point tracker for fast changing environmental conditions, in IEEE International
Symposium on Industrial Electronics, pp. 24012407, 2008.

[54] N. Kaushika and A. K. Rai, An investigation of mismatch losses in solar photovoltaic


cell networks, Energy, vol. 32, no. 5, pp. 755 759, 2007.

[55] W. Xiao, F. Edwin, G. Spagnuolo, and J. Jatskevich, Efficient approaches for mod-
eling and simulating photovoltaic power systems, IEEE Journal of Photovoltaics,
vol. 3, no. 1, pp. 500508, 2013.

[56] BP4175 photovoltaic module datasheet.

[57] G. M. Masters, Renewable and Efficient Electric Power Systems. John Wiley & Sons,
2004.

[58] M. Stiebler, Wind Energy Systems for Electric Power Generation. Springer, 2008.

[59] J. Slootweg, S. W. H. De Haan, H. Polinder, and W. Kling, General model for


representing variable speed wind turbines in power system dynamics simulations,
IEEE Transactions on Power Systems, vol. 18, no. 1, pp. 144151, 2003.
124 REFERENCES

[60] N. Orlando, M. Liserre, R. Mastromauro, and A. DellAquila, A survey of control


issues in PMSG-based small wind-turbine systems, IEEE Transactions on Industrial
Informatics, vol. 9, no. 3, pp. 12111221, 2013.

[61] A. Bellara, H. Tiegna, Y. Amara, and G. Barakat, On load analytical modelling of


the magnetic field for axial flux surface-inset permanent magnet machines with semi-
closed slots, in International Conference on Electrical Machines, pp. 28522858,
2012.

[62] H. Tiegna, Y. Amara, G. Barakat, and P. Reghem, 2D analytical solution of mag-


netic field in axial flux permanent magnet machines for the inter-turn short-circuit
stator fault case, in International Conference on Renewable Energies and Vehicular
Technology, pp. 218223, 2012.

[63] S. Vaez-Zadeh and A. Isfahani, Enhanced modeling of linear permanent-magnet


synchronous motors, IEEE Transactions on Magnetics, vol. 43, no. 1, pp. 3339,
2007.

[64] M. Hsieh and Y. Hsu, A generalized magnetic circuit modeling approach for design of
surface permanent-magnet machines, IEEE Transactions on Industrial Electronics,
vol. 59, no. 2, pp. 779792, 2012.

[65] P. Pillay and R. Krishnan, Modeling of permanent magnet motor drives, IEEE
Transactions on Industrial Electronics, vol. 35, no. 4, pp. 537541, 1988.

[66] A. Wijenayake and P. Schmidt, Modeling and analysis of permanent magnet syn-
chronous motor by taking saturation and core loss into account, in International
Conference on Power Electronics and Drive Systems, vol. 2, pp. 530534 vol.2, 1997.

[67] N. Urasaki, T. Senjyu, and K. Uezato, A novel calculation method for iron loss re-
sistance suitable in modeling permanent-magnet synchronous motors, IEEE Trans-
actions on Energy Conversion, vol. 18, no. 1, pp. 4147, 2003.

[68] S. Chiniforoosh, P. Alaeinovin, A. Davoudi, J. Jatskevich, and P. Chapman, Dynamic


modeling of stand-alone micro-wind turbine generator systems for telecommunication
power supply, in IEEE International Telecommunications Energy Conference, pp. 1
6, 2008.

[69] A. Cavagnino, M. Lazzari, F. Profumo, and A. Tenconi, Axial flux interior PM syn-
chronous motor: parameters identification and steady-state performance measure-
ments, IEEE Transactions on Industry Applications, vol. 36, no. 6, pp. 15811588,
2000.

[70] M. Kazerooni and N. Kar, Methods for determining the parameters and characteris-
tics of PMSM, in IEEE International Electric Machines Drives Conference, pp. 955
960, 2011.
REFERENCES 125

[71] B. Stumberger, B. Kreca, and B. Hribernik, Determination of parameters of syn-


chronous motor with permanent magnets from measurement of load conditions,
in IEEE International Electric Machines and Drives Conference, pp. WB2/1.1
WB2/1.3, 1997.

[72] H.-P. Nee, L. Lefevre, P. Thelin, and J. Soulard, Determination of d and q reac-
tances of permanent-magnet synchronous motors without measurements of the rotor
position, IEEE Transactions on Industry Applications, vol. 36, no. 5, pp. 13301335,
2000.

[73] B. Cheng and T. Tesch, Torque feedforward control technique for permanent-magnet
synchronous motors, IEEE Transactions on Industrial Electronics, vol. 57, no. 3,
pp. 969974, 2010.

[74] T. Kim and W. Qiao, A hybrid battery model capable of capturing dynamic cir-
cuit characteristics and nonlinear capacity effects, IEEE Transactions on Energy
Conversion, vol. 26, no. 4, pp. 11721180, 2011.

[75] R. Klein, N. Chaturvedi, J. Christensen, J. Ahmed, R. Findeisen, and A. Kojic, Elec-


trochemical model based observer design for a lithium-ion battery, IEEE Transac-
tions on Control Systems Technology, vol. 21, no. 2, pp. 289301, 2013.

[76] P. Rong and M. Pedram, An analytical model for predicting the remaining battery
capacity of lithium-ion batteries, IEEE Transactions on Very Large Scale Integration
(VLSI) Systems, vol. 14, no. 5, pp. 441451, 2006.

[77] D. Rakhmatov, S. Vrudhula, and D. Wallach, A model for battery lifetime analysis
for organizing applications on a pocket computer, IEEE Transactions on Very Large
Scale Integration (VLSI) Systems, vol. 11, no. 6, pp. 10191030, 2003.

[78] C. Chiasserini and R. Rao, Energy efficient battery management, IEEE Journal on
Selected Areas in Communications, vol. 19, no. 7, pp. 12351245, 2001.

[79] V. Rao, G. Singhal, A. Kumar, and N. Navet, Battery model for embedded systems,
in 18th International Conference on VLSI Design, pp. 105110, 2005.

[80] C. Chiasserini and R. Rao, A model for battery pulsed discharge with recovery
effect, in IEEE Wireless Communications and Networking Conference, pp. 636639
vol.2, 1999.

[81] M. Ceraolo, New dynamical models of lead-acid batteries, IEEE Transactions on


Power Systems, vol. 15, no. 4, pp. 11841190, 2000.

[82] S. Barsali and M. Ceraolo, Dynamical models of lead-acid batteries: implementation


issues, IEEE Transactions on Energy Conversion, vol. 17, no. 1, pp. 1623, 2002.
126 REFERENCES

[83] T. Hu, B. Zanchi, and T. Zhao, Simple analytical method for determining parameters
of discharging batteries, IEEE Transactions on Energy Conversion, vol. 26, no. 3,
pp. 787798, 2011.

[84] M. Einhorn, F. Conte, C. Kral, and J. Fleig, Comparison, selection, and parameter-
ization of electrical battery models for automotive applications, IEEE Transactions
on Power Electronics, vol. 28, no. 3, pp. 14291437, 2013.

[85] Y. Kim and H. Ha, Design of interface circuits with electrical battery models, IEEE
Transactions on Industrial Electronics, vol. 44, no. 1, pp. 8186, 1997.

[86] O. Tremblay, L.-A. Dessaint, and A.-I. Dekkiche, A generic battery model for the dy-
namic simulation of hybrid electric vehicles, in IEEE Vehicle Power and Propulsion
Conference, pp. 284289, 2007.

[87] G. Byeon, T. Yoon, S. Oh, and G. Jang, Energy management strategy of the DC
distribution system in buildings using the EV service model, IEEE Transactions on
Power Electronics, vol. 28, no. 4, pp. 15441554, 2013.

[88] O. Laldin, M. Moshirvaziri, and O. Trescases, Predictive algorithm for optimizing


power flow in hybrid ultracapacitor/battery storage systems for light electric vehicles,
IEEE Transactions on Power Electronics, vol. 28, no. 8, pp. 38823895, 2013.

[89] E. Ribeiro, A. J. M. Cardoso, and C. Boccaletti, New challenges in power con-


ditioning for telecommunications, in 36th Annual Conference on IEEE Industrial
Electronics Society, pp. 582587, 2010.

[90] E. Ribeiro, A. J. M. Cardoso, and C. Boccaletti, Uninterruptible energy production


in standalone power systems for telecommunications, in International Conference on
Renewable Energies and Power Quality, 2009.

[91] H. Gumhalter, Power Supply in Telecommunications. Springer, 1995.

[92] A. Abdullah and A. Fekih, An overview of the current state of wind energy technol-
ogy development in the US, in IEEE Green Technologies Conference, pp. 120126,
2013.

[93] Y. Duan and D. Ionel, A review of recent developments in electrical machine de-
sign optimization methods with a permanent-magnet synchronous motor benchmark
study, IEEE Transactions on Industry Applications, vol. 49, no. 3, pp. 12681275,
2013.

[94] N. Patil and Y. Bhosle, A review on wind turbine generator topologies, in Interna-
tional Conference on Power, Energy and Control, pp. 625629, 2013.
REFERENCES 127

[95] R. Cardenas, R. Pena, S. Alepuz, and G. Asher, Overview of control systems for the
operation of DFIGs in wind energy applications, IEEE Transactions on Industrial
Electronics, vol. 60, no. 7, pp. 27762798, 2013.

[96] Z. Chen, J. Guerrero, and F. Blaabjerg, A review of the state of the art of power
electronics for wind turbines, IEEE Transactions on Power Electronics, vol. 24, no. 8,
pp. 18591875, 2009.

[97] F. Blaabjerg and K. Ma, Future on power electronics for wind turbine systems,
IEEE Journal of Emerging and Selected Topics in Power Electronics, early access
2013.

[98] N. M. A. Freire and A. J. M. Cardoso, A fault-tolerant direct controlled PMSG drive


for wind energy conversion systems, IEEE Transactions on Industrial Electronics,
vol. 61, no. 2, pp. 821834, 2014.

[99] R. Kot, M. Rolak, and M. Malinowski, Comparison of maximum peak power tracking
algorithms for a small wind turbine, Mathematics and Computers in Simulation,
vol. 91, pp. 29 40, 2013.

[100] A. Izadian, H. Yang, and N. Girrens, Wind energy harvesting control for green
cellphone towers with dSPACE implementation, in 37th Annual Conference on IEEE
Industrial Electronics Society, pp. 25112516, 2011.

[101] K. Lo, Y. Chen, and Y. Chang, MPPT battery charger for stand-alone wind power
system, IEEE Transactions on Power Electronics, vol. 26, no. 6, pp. 16311638,
2011.

[102] G. C. Ferrell, W. R. Thorn, M. Agnello, and D. Smith, The application of small


wind energy conversion systems to the telecommunications industry, in International
Telecommunications Energy Conference, pp. 438444, 1979.

[103] W. Hopkins, Wind turbines for professional telecommunication applications: the 15


year running of a wind turbine for a microwave relay station in Southern France and
subsequent developments, in International Telecommunications Energy Conference,
pp. 741746, 1995.

[104] A. Mirecki, X. Roboam, and F. Richardeau, Architecture complexity and energy effi-
ciency of small wind turbines, IEEE Transactions on Industrial Electronics, vol. 54,
no. 1, pp. 660670, 2007.

[105] M. Vilsan and I. Nita, A hybrid wind-photovoltaic power supply for a telecommuni-
cation system, in International Telecommunications Energy Conference, pp. 589591,
1997.
128 REFERENCES

[106] J. A. Baroudi, V. Dinavahi, and A. Knight, A review of power converter topologies


for wind generators, in IEEE International Conference on Electric Machines and
Drives, pp. 458465, 2005.

[107] W. Li, G. Joos, and J. Belanger, Real-time simulation of a wind turbine generator
coupled with a battery supercapacitor energy storage system, IEEE Transactions on
Industrial Electronics, vol. 57, no. 4, pp. 11371145, 2010.

[108] C. Abbey and G. Joos, Supercapacitor energy storage for wind energy applications,
IEEE Transactions on Industry Applications, vol. 43, no. 3, pp. 769776, 2007.

[109] A. Gee, F. Robinson, and R. Dunn, Analysis of battery lifetime extension in a


small-scale wind-energy system using supercapacitors, IEEE Transactions on Energy
Conversion, vol. 28, no. 1, pp. 2433, 2013.

[110] A. Haruni, M. Negnevitsky, M. Haque, and A. Gargoom, A novel operation and


control strategy for a standalone hybrid renewable power system, IEEE Transactions
on Sustainable Energy, vol. 4, no. 2, pp. 402413, 2013.

[111] A. Malekpour, T. Niknam, A. Pahwa, and A. Kavousi Fard, Multi-objective stochas-


tic distribution feeder reconfiguration in systems with wind power generators and fuel
cells using the point estimate method, IEEE Transactions on Power Systems, vol. 28,
no. 2, pp. 14831492, 2013.

[112] E. Ribeiro, A. J. M. Cardoso, and C. Boccaletti, Fuel cell systems for telecommu-
nications, in Emerging Trends in Technological Innovation (L. Camarinha-Matos,
P. Pereira, and L. Ribeiro, eds.), vol. 314 of IFIP Advances in Information and Com-
munication Technology, pp. 469476, Springer Berlin Heidelberg, 2010.

[113] E. Ribeiro, A. J. M. Cardoso, and C. Boccaletti, Fuel cell-supercapacitor system for


telecommunications, in IET International Conference on Power Electronics, Ma-
chines and Drives, pp. 16, 2010.

[114] E. Ribeiro, A. J. M. Cardoso, and C. Boccaletti, Grid interface for a wind turbine-
fuel cell system, in International Conference on Electrical Machines, pp. 16, 2010.

[115] T. Hirose and H. Matsuo, Standalone hybrid wind-solar power generation system
applying dump power control without dump load, IEEE Transactions on Industrial
Electronics, vol. 59, no. 2, pp. 988997, 2012.

[116] N. M. A. Freire, E. Ribeiro, A. J. M. Cardoso, and C. Boccaletti, Experimental


analysis of a standalone renewable energy based hybrid system, in Technological
Innovation for Value Creation (L. Camarinha-Matos, E. Shahamatnia, and G. Nunes,
eds.), vol. 372 of IFIP Advances in Information and Communication Technology,
pp. 337344, Springer Berlin Heidelberg, 2012.
REFERENCES 129

[117] J. O. Estima, J. Fernandes, and A. J. M. Cardoso, Faulty operation analysis of


permanent magnet synchronous generator drives for wind turbine applications, in
5th IET International Conference on Power Electronics, Machines and Drives, pp. 1
6, 2010.

[118] S. Zhang, K. Tseng, D. Vilathgamuwa, T. Nguyen, and X. Wang, Design of a robust


grid interface system for PMSG-based wind turbine generators, IEEE Transactions
on Industrial Electronics, vol. 58, no. 1, pp. 316328, 2011.

[119] K. Tan and S. Islam, Optimum control strategies in energy conversion of PMSG
wind turbine system without mechanical sensors, IEEE Transactions on Energy
Conversion, vol. 19, no. 2, pp. 392399, 2004.

[120] Z. Chen and E. Spooner, Grid power quality with variable speed wind turbines,
IEEE Transactions on Energy Conversion, vol. 16, no. 2, pp. 148154, 2001.

[121] A. De Broe, S. Drouilhet, and V. Gevorgian, A peak power tracker for small wind
turbines in battery charging applications, IEEE Transactions on Energy Conversion,
vol. 14, no. 4, pp. 16301635, 1999.

[122] Y. Xia, K. Ahmed, and B. Williams, Wind turbine power coefficient analysis of a
new maximum power point tracking technique, IEEE Transactions on Industrial
Electronics, vol. 60, no. 3, pp. 11221132, 2013.

[123] R. Datta and V. T. Ranganathan, A method of tracking the peak power points
for a variable speed wind energy conversion system, IEEE Transactions on Energy
Conversion, vol. 18, no. 1, pp. 163168, 2003.

[124] E. Koutroulis and K. Kalaitzakis, Design of a maximum power tracking system for
wind-energy-conversion applications, IEEE Transactions on Industrial Electronics,
vol. 53, no. 2, pp. 486494, 2006.

[125] C. Liu, K. Chau, and X. Zhang, An efficient wind-photovoltaic hybrid generation


system using doubly excited permanent-magnet brushless machine, IEEE Transac-
tions on Industrial Electronics, vol. 57, no. 3, pp. 831839, 2010.

[126] Q. Wang and L. Chang, An intelligent maximum power extraction algorithm for
inverter-based variable speed wind turbine systems, IEEE Transactions on Power
Electronics, vol. 19, no. 5, pp. 12421249, 2004.

[127] S. M. R. Kazmi, H. Goto, H. Guo, and O. Ichinokura, A novel algorithm for fast and
efficient speed-sensorless maximum power point tracking in wind energy conversion
systems, IEEE Transactions on Industrial Electronics, vol. 58, no. 1, pp. 2936,
2011.
130 REFERENCES

[128] B. Kou, Y. Bai, and L. Li, A novel wind power generator system with automatic max-
imum power tracking capability, IEEE Transactions on Energy Conversion, vol. 28,
no. 3, pp. 632642, 2013.

[129] B. Subudhi and R. Pradhan, A comparative study on maximum power point track-
ing techniques for photovoltaic power systems, IEEE Transactions on Sustainable
Energy, vol. 4, no. 1, pp. 8998, 2013.

[130] H. Kelly, Photovoltaic power for telecommunications, in International Telephone


Energy Conference, pp. 104109, 1978.

[131] K. Kurozumi, Y. Kawagoe, T. Yamanaka, and T. Hidaka, Development of photo-


voltaic systems using DC interactive converter for telecommunication systems, in
19th International Telecommunications Energy Conference, pp. 211214, 1997.

[132] A. Steinhuser, W. Schulz, H. Schmidt, and W. Roth, Photovoltaic power supply for
telecommunication network components in remote areas, in International Telecom-
munications Energy Special Conference, pp. 221225, 2000.

[133] N. Darghouth, M. Blazek, and D. Kammen, Alternative power generation technolo-


gies: Use of distributed power for telecommunications facilities and operations, in
IEEE International Symposium on Electronics and the Environment, pp. 6770, 2006.

[134] A. H. Olden, Photovoltaic and other remote-site power generation systems for the
Papua New Guinea telecommunications network, in Fifth International Telecommu-
nications Energy Conference, pp. 166174, 1983.

[135] A. Murray, Photovoltaics for telecommunications in the 1990s, in 10th International


Telecommunications Energy Conference, pp. 350352, 1988.

[136] P. Thounthong, A. Luksanasakul, P. Koseeyaporn, and B. Davat, Intelligent model-


based control of a standalone photovoltaic/fuel cell power plant with supercapacitor
energy storage, IEEE Transactions on Sustainable Energy, vol. 4, no. 1, pp. 240249,
2013.

[137] E. Ribeiro, A. J. M. Cardoso, and C. Boccaletti, Power converters analysis in a


renewable energy based hybrid system with hydrogen storage, in IET Conference on
Renewable Power Generation, pp. 16, 2011.

[138] S. Duryea, S. Islam, and W. Lawrance, A battery management system for stand-
alone photovoltaic energy systems, IEEE Industry Applications Magazine, vol. 7,
no. 3, pp. 6772, 2001.

[139] E. Ribeiro, A. J. M. Cardoso, and C. Boccaletti, Power conditioning of an energy


storage system for telecommunications, in International Symposium on Power Elec-
tronics Electrical Drives Automation and Motion, pp. 11231128, 2010.
REFERENCES 131

[140] E. Kaplani, Design and performance considerations in stand alone PV powered


telecommunication systems, IEEE Latin America Transactions, vol. 10, no. 3,
pp. 17231729, 2012.

[141] K. Kurozumi, T. Tawara, T. Tanaka, Y. Kawagoe, T. Yamanaka, H. Ikebe, K. Shin-


dou, and T. Miyazato, A hybrid system composed of a wind power and a photovoltaic
system at NTT Kume-Jima radio relay station, in International Telecommunications
Energy Conference, pp. 785789, 1998.

[142] D. B. Chia, B. L. Yong, R. Rajkumar, and V. K. Ramachandaramurthy, Design of


a PV/wind hybrid system for telecommunication load in Borneo region, in 9th In-
ternational Conference on Environment and Electrical Engineering, pp. 7376, 2010.

[143] S. Tanezaki, T. Matsushima, and S. Muroyama, Stand-alone hybrid power supply


system composed of wind turbines and photovoltaic modules for powering radio relay
stations, in 25th International Telecommunications Energy Conference, pp. 457462,
2003.

[144] W. Li and X. He, Review of nonisolated high-step-up DC/DC converters in pho-


tovoltaic grid-connected applications, IEEE Transactions on Industrial Electronics,
vol. 58, no. 4, pp. 12391250, 2011.

[145] T. K., H. C., and W. Shih, A high step-up converter with a voltage multiplier module
for a photovoltaic system, IEEE Transactions on Power Electronics, vol. 28, no. 6,
pp. 30473057, 2013.

[146] K.-C. Tseng and C.-C. Huang, High step-up high-efficiency interleaved converter
with voltage multiplier module for renewable energy system, IEEE Transactions on
Industrial Electronics, vol. 61, no. 3, pp. 13111319, 2014.

[147] H. Kim, B. Parkhideh, T. Bongers, and H. Gao, Reconfigurable solar converter:


A single-stage power conversion PV-battery system, IEEE Transactions on Power
Electronics, vol. 28, no. 8, pp. 37883797, 2013.

[148] E. Ribeiro, A. J. M. Cardoso, C. Boccaletti, and A. M. S. Mendes, Photovoltaic DC-


DC converter for telecommunications energy systems, in International Conference
on Clean Electrical Power, pp. 433438, 2009.

[149] T. Suntio, J. Leppaaho, J. Huusari, and L. Nousiainen, Issues on solar-generator


interfacing with current-fed MPPT converters, IEEE Transactions on Power Elec-
tronics, vol. 25, no. 9, pp. 24092419, 2010.

[150] A. Imtiaz, F. Khan, and H. Kamath, All-in-one photovoltaic power system: Fea-
tures and challenges involved in cell-level power conversion in AC solar cells, IEEE
Industry Applications Magazine, vol. 19, no. 4, pp. 1223, 2013.
132 REFERENCES

[151] E. Ribeiro, A. J. M. Cardoso, and C. Boccaletti, Power conditioning and energy


management in a renewable energy based hybrid system for telecommunications, in
International Telecommunications Energy Conference, pp. 19, 2011.

[152] N. Mohan, T. M. Undeland, and W. P. Robbins, Power Electronics: Converters,


Applications, and Design. John Wiley & Sons, 2003.

[153] D. R. Northcott, S. Filizadeh, and A. R. Chevrefils, Design of a bidirectional buck-


boost dc/dc converter for a series hybrid electric vehicle using PSCAD/EMTDC, in
IEEE Vehicle Power and Propulsion Conference, pp. 15611566, 2009.

[154] R. Schupbach and J. Balda, Comparing DC-DC converters for power management
in hybrid electric vehicles, in IEEE International Electric Machines and Drives Con-
ference, vol. 3, pp. 13691374 vol.3, 2003.

[155] L. Ni, D. Patterson, and J. Hudgins, High power current sensorless bidirectional
16-phase interleaved DC-DC converter for hybrid vehicle application, IEEE Trans-
actions on Power Electronics, vol. 27, no. 3, pp. 11411151, 2012.

[156] O. Garcia, P. Zumel, A. De Castro, and J. A. Cobos, Automotive DC-DC bidirec-


tional converter made with many interleaved buck stages, IEEE Transactions on
Power Electronics, vol. 21, no. 3, pp. 578586, 2006.

[157] H. Kosai, J. Scofield, S. McNeal, B. Jordan, and B. Ray, Design and performance
evaluation of a 200 C interleaved boost converter, IEEE Transactions on Power
Electronics, vol. 28, no. 4, pp. 16911699, 2013.

[158] Y. Gu and D. Zhang, Interleaved boost converter with ripple cancellation network,
IEEE Transactions on Power Electronics, vol. 28, no. 8, pp. 38603869, 2013.

[159] M. Pavlovsky, G. Guidi, and A. Kawamura, Assessment of coupled and independent


phase designs of interleaved multiphase buck/boost DC-DC converter for EV power
train, IEEE Transactions on Power Electronics, early access 2013.

[160] S. Moon, G. Koo, and G. Moon, A new control method of interleaved single-stage
flyback AC-DC converter for outdoor led lighting systems, IEEE Transactions on
Power Electronics, vol. 28, no. 8, pp. 40514062, 2013.

[161] S. Zhang and X. Yu, An unified analytical modeling of the interleaved pulse width
modulation (PWM) DC-DC converter and its applications, IEEE Transactions on
Power Electronics, vol. 28, no. 11, pp. 51475158, 2013.

[162] C. de Beer, P. Barendse, and A. Khan, Development of an HT PEM fuel cell emulator
using a multiphase interleaved DC-DC converter topology, IEEE Transactions on
Power Electronics, vol. 28, no. 3, pp. 11201131, 2013.
REFERENCES 133

[163] F. Yang, X. Ruan, Y. Yang, and Z. Ye, Interleaved critical current mode boost PFC
converter with coupled inductor, IEEE Transactions on Power Electronics, vol. 26,
no. 9, pp. 24042413, 2011.

[164] K. Kajangpan and B. Neammanee, High gain double interleave technique with maxi-
mum peak power tracking for wind turbine converter, in International Conference on
Electrical Engineering/Electronics, Computer, Telecommunications and Information
Technology, vol. 01, pp. 292295, 2009.

[165] M. Veerachary, T. Senjyu, and K. Uezato, Neural-network-based maximum-power-


point tracking of coupled-inductor interleaved-boost-converter-supplied pv system us-
ing fuzzy controller, IEEE Transactions on Industrial Electronics, vol. 50, no. 4,
pp. 749758, 2003.

[166] C. Ho, H. Breuninger, S. Pettersson, G. Escobar, and F. Canales, A comparative


performance study of an interleaved boost converter using commercial Si and SiC
diodes for PV applications, IEEE Transactions on Power Electronics, vol. 28, no. 1,
pp. 289299, 2013.

[167] C.-m. Ho, H. Breuninger, S. Pettersson, G. Escobar, L. Serpa, and A. Coccia, Prac-
tical design and implementation procedure of an interleaved boost converter using
SiC diodes for PV applications, IEEE Transactions on Power Electronics, vol. 27,
no. 6, pp. 28352845, 2012.

[168] F. Edwin, W. Xiao, and V. Khadkikar, Dynamic modeling and control of interleaved
flyback module-integrated converter for PV power applications, IEEE Transactions
on Industrial Electronics, vol. 61, no. 3, pp. 13771388, 2014.

[169] C. Hua, J. Lin, and C. Shen, Implementation of a DSP-controlled photovoltaic sys-


tem with peak power tracking, IEEE Transactions on Industrial Electronics, vol. 45,
no. 1, pp. 99107, 1998.

[170] C. Patsios, M. Antonakopoulos, A. Chaniotis, and A. Kladas, Control and analysis


of a hybrid renewable energy-based power system, in XIX International Conference
on Electrical Machines, pp. 16, 2010.

[171] G. Spagnuolo, G. Petrone, S. Araujo, C. Cecati, E. Friis-Madsen, E. Gubia, D. Hissel,


M. Jasinski, W. Knapp, M. Liserre, P. Rodriguez, R. Teodorescu, and P. Zacharias,
Renewable energy operation and conversion schemes: A summary of discussions dur-
ing the seminar on renewable energy systems, IEEE Industrial Electronics Magazine,
vol. 4, no. 1, pp. 3851, 2010.

[172] J. Kwon, B. Kwon, and K. Nam, Three-phase photovoltaic system with three-level
boosting MPPT control, IEEE Transactions on Power Electronics, vol. 23, no. 5,
pp. 23192327, 2008.
134 REFERENCES

[173] M. Zhang, Y. Jiang, F. Lee, and M. Jovanovic, Single-phase three-level boost power
factor correction converter, in Tenth Annual Applied Power Electronics Conference
and Exposition Conference, pp. 434439, 1995.

[174] B. Lin and H. Lu, Single-phase three-level PWM rectifier, in IEEE 1999 Interna-
tional Conference on Power Electronics and Drive Systems, vol. 1, pp. 6368 vol.1,
1999.

[175] V. Yaramasu and B. Wu, Three-level boost converter based medium voltage
megawatt PMSG wind energy conversion systems, in IEEE Energy Conversion
Congress and Exposition, pp. 561567, 2011.

[176] P. Thounthong, Control of a three-level boost converter based on a differential flat-


ness approach for fuel cell vehicle applications, IEEE Transactions on Vehicular
Technology, vol. 61, no. 3, pp. 14671472, 2012.

[177] J. Kolar and T. Friedli, The essence of three-phase PFC rectifier systems - Part I,
IEEE Transactions on Power Electronics, vol. 28, no. 1, pp. 176198, 2013.

[178] H. Chen and W. Lin, MPPT and voltage balancing control with sensing only inductor
current for photovoltaic-fed, three-level, boost-type converters, IEEE Transactions
on Power Electronics, vol. 29, no. 1, pp. 2935, 2014.

[179] Y. Zhang, J. Sun, and Y. Wang, Hybrid boost three-level DC-DC converter with
high voltage gain for photovoltaic generation systems, IEEE Transactions on Power
Electronics, vol. 28, no. 8, pp. 36593664, 2013.

[180] X. Xu, H. Wang, and Y. Zuo, Method for diagnosing photovoltaic array fault in so-
lar photovoltaic system, in Asia-Pacific Power and Energy Engineering Conference,
pp. 15, 2011.

[181] B. Bose, Global energy scenario and impact of power electronics in 21st century,
IEEE Transactions on Industrial Electronics, vol. 60, no. 7, pp. 26382651, 2013.

[182] Y. Song and B. Wang, Survey on reliability of power electronic systems, IEEE
Transactions on Power Electronics, vol. 28, no. 1, pp. 591604, 2013.

[183] S. Yang, A. Bryant, P. Mawby, D. Xiang, L. Ran, and P. Tavner, An industry-based


survey of reliability in power electronic converters, IEEE Transactions on Industry
Applications, vol. 47, no. 3, pp. 14411451, 2011.

[184] S. Yang, D. Xiang, A. Bryant, P. Mawby, L. Ran, and P. Tavner, Condition monitor-
ing for device reliability in power electronic converters: A review, IEEE Transactions
on Power Electronics, vol. 25, no. 11, pp. 27342752, 2010.
REFERENCES 135

[185] A. M. R. Amaral and A. J. M. Cardoso, On-line fault detection of aluminium elec-


trolytic capacitors, in step-down DC-DC converters, using input current and output
voltage ripple, IET Power Electronics, vol. 5, no. 3, pp. 315322, 2012.

[186] P. Duan, K. Xie, L. Zhang, and X. Rong, Open-switch fault diagnosis and sys-
tem reconfiguration of doubly fed wind power converter used in a microgrid, IEEE
Transactions on Power Electronics, vol. 26, no. 3, pp. 816821, 2011.

[187] J. Sun, Y. Qiu, M. Xu, and F. Lee, High-frequency dynamic current sharing analyses
for multiphase buck VRs, IEEE Transactions on Power Electronics, vol. 22, no. 6,
pp. 24242431, 2007.

[188] R. Foley, R. Kavanagh, and M. Egan, Sensorless current estimation and sharing in
multiphase buck converters, IEEE Transactions on Power Electronics, vol. 27, no. 6,
pp. 29362946, 2012.

[189] D. Kim, G. Choe, and B. Lee, DCM analysis and inductance selection method of in-
terleaved boost converter for electric vehicles, in IEEE Vehicle Power and Propulsion
Conference, pp. 17, 2011.

[190] R. Kadri, J.-P. Gaubert, and G. Champenois, Nondissipative string current diverter
for solving the cascaded DC-DC converter connection problem in photovoltaic power
generation system, IEEE Transactions on Power Electronics, vol. 27, no. 3, pp. 1249
1258, 2012.

[191] J. Soon and K. Low, Photovoltaic model identification using particle swarm opti-
mization with inverse barrier constraint, IEEE Transactions on Power Electronics,
vol. 27, no. 9, pp. 39753983, 2012.

[192] M. A. S. Masoum, H. Dehbonei, and E. Fuchs, Theoretical and experimental anal-


yses of photovoltaic systems with voltageand current-based maximum power-point
tracking, IEEE Transactions on Energy Conversion, vol. 17, no. 4, pp. 514522,
2002.

[193] N. M. A. Freire, J. O. Estima, and A. J. M. Cardoso, Open-circuit fault diagnosis


in PMSG drives for wind turbine applications, IEEE Transactions on Industrial
Electronics, vol. 60, no. 9, pp. 39573967, 2013.

[194] E. Ribeiro, A. J. M. Cardoso, and C. Boccaletti, Fault analysis of dual active


bridge converters, in 38th Annual Conference on IEEE Industrial Electronics So-
ciety, pp. 398403, 2012.

[195] Z. Sun, K. W. R. Chew, H. Tang, and L. Siek, Adaptive gate switching control
for discontinuous-conduction mode DC-DC converter, IEEE Transactions on Power
Electronics, early access 2013.
136 REFERENCES

[196] L. Barote, C. Marinescu, and M. Cirstea, Control structure for single-phase stand-
alone wind-based energy sources, IEEE Transactions on Industrial Electronics,
vol. 60, no. 2, pp. 764772, 2013.

[197] A. Izadian and P. Khayyer, Application of Kalman filters in model-based fault diag-
nosis of a DC-DC boost converter, in 36th Annual Conference on IEEE Industrial
Electronics Society, pp. 369372, 2010.

[198] Q. An, L. Sun, L. Sun, and T. Jahns, Low-cost diagnostic method for open-switch
faults in inverters, Electronics Letters, vol. 46, no. 14, pp. 10211022, 2010.

[199] Q.-T. An, L.-Z. Sun, K. Zhao, and L. Sun, Switching function model-based fast-
diagnostic method of open-switch faults in inverters without sensors, IEEE Trans-
actions on Power Electronics, vol. 26, no. 1, pp. 119126, 2011.

[200] B. Ji, V. Pickert, W. Cao, and B. Zahawi, In situ diagnostics and prognostics of wire
bonding faults in IGBT modules for electric vehicle drives, IEEE Transactions on
Power Electronics, vol. 28, no. 12, pp. 55685577, 2013.

[201] E. Ribeiro, A. J. M. Cardoso, and C. Boccaletti, Fault-tolerant strategy for a pho-


tovoltaic DC-DC converter, IEEE Transactions on Power Electronics, vol. 28, no. 6,
pp. 30083018, 2013.

[202] E. Ribeiro, A. J. M. Cardoso, and C. Boccaletti, Open-circuit fault diagnosis in in-


terleaved DC-DC converters, IEEE Transactions on Power Electronics, early access
2013.

[203] D. Guilbert, A. Gaillard, A. NDiaye, and A. Djerdir, Energy efficiency and fault
tolerance comparison of DC/DC converters topologies for fuel cell electric vehicles,
in IEEE Transportation Electrification Conference and Expo, pp. 17, 2013.

[204] X. Tan, Q. Li, and H. Wang, Advances and trends of energy storage technology
in microgrid, International Journal of Electrical Power & Energy Systems, vol. 44,
no. 1, pp. 179 191, 2013.

[205] H. Zhou, T. Bhattacharya, D. Tran, T. Siew, and A. Khambadkone, Composite en-


ergy storage system involving battery and ultracapacitor with dynamic energy man-
agement in microgrid applications, IEEE Transactions on Power Electronics, vol. 26,
no. 3, pp. 923930, 2011.

[206] M. Bragard, N. Soltau, S. Thomas, and R. De Doncker, The balance of renewable


sources and user demands in grids: Power electronics for modular battery energy
storage systems, IEEE Transactions on Power Electronics, vol. 25, no. 12, pp. 3049
3056, 2010.
REFERENCES 137

[207] I. Trintis, S. Munk-Nielsen, and R. Teodorescu, A new modular multilevel converter


with integrated energy storage, in 37th Annual Conference on IEEE Industrial Elec-
tronics Society, pp. 10751080, 2011.

[208] A. Stepanov and I. Galkin, Concept of modular uninterruptible power supply sys-
tem with alternative energy storages and sources, in 7th International Conference-
Workshop Compatibility and Power Electronics, pp. 98102, 2011.

[209] G. Su and L. Tang, A multiphase, modular, bidirectional, triple-voltage dc-dc con-


verter for hybrid and fuel cell vehicle power systems, IEEE Transactions on Power
Electronics, vol. 23, no. 6, pp. 30353046, 2008.

[210] E. Ortjohann, M. Lingemann, O. Omari, A. Schmelter, N. Hmasic, A. Mohd, W. Sin-


sukthavorn, and D. Morton, Modular architecture for decentralized hybrid power
systems, in 13th Power Electronics and Motion Control Conference, pp. 21342141,
2008.

[211] D. Montesinos-Miracle, M. Massot-Campos, J. Bergas-Jane, S. Galceran-Arellano,


and A. Rufer, Design and control of a modular multilevel DC/DC converter for
regenerative applications, IEEE Transactions on Power Electronics, vol. 28, no. 8,
pp. 39703979, 2013.

[212] F. Caponio, A. Abba, P. Baruzzi, G. Ripamonti, and A. Geraci, Modular and bi-
directional energy storage system compliant with accumulators of different chemistry,
in 11th International Conference on Electrical Power Quality and Utilisation, pp. 16,
2011.

[213] E. Ribeiro, A. J. M. Cardoso, and C. Boccaletti, Power conditioning supercapacitors


in combination with batteries for stand-alone power systems, in International Sympo-
sium on Power Electronics, Electrical Drives, Automation and Motion, pp. 914919,
2012.
138 REFERENCES