Design of Wind Dominated Hybrid Remote Area Power Supply Systems Nishad Mendis University of Wollongong

University of Wollongong Research Online

University of Wollongong Thesis Collection University of Wollongong Thesis Collections

2012 Design of Wind Dominated Hybrid Remote Area Power Supply Systems Nishad Mendis University of Wollongong

Recommended Citation Mendis, Nishad, Design of Wind Dominated Hybrid Remote Area Power Supply Systems, Doctor of Philosophy thesis, School of Electrical, Computer and Telecommunications Engineering, University of Wollongong, 2012. http://ro.uow.edu.au/theses/3489

Research Online is the open access institutional repository for the University of Wollongong. For further information contact the UOW Library: [email protected]

Design of Wind Dominated Hybrid Remote Area Power Supply Systems

A thesis submitted in fulﬁlment of the

requirements for the award of the degree

Doctor of Philosophy

from

University of Wollongong

Nishad Mendis, BSc(Eng)

School of Electrical, Computer and Telecommunications

Engineering

April 2012 Dedicated to my parents... Acknowledgements

This thesis would not have become a realisation without the contributions from many people and institutions.

Foremost, I would like to express my sincere gratitude to my main supervisor

A/Prof. Kashem M. Muttaqi for oﬀering a timely and interesting research topic.

Besides my main supervisor, I would like to thank my co-supervisor A/Prof. Sarath

Perera for giving me an opportunity to pursue my doctoral studies at the University of Wollongong (UOW). I appreciate their patience, motivation, enthusiasm, and immense knowledge, moral support and guidance helped me throughout the research and writing of this thesis.

The project was ﬁnancially funded by Australian Research Council (ARC) and

Hydro Tasmania Linkage Grant, LP0669245. I extremely grateful for this generous support, as well as the ﬁnancial assistance provided by the Endeavor Energy Power

Quality and Reliability Centre (EEPQRC).

I would like to thank Dr. Saad S. Sayeef, former post doctoral fellow at EEPQRC in addition for his insight technical contributions and friendly attitude. Also, a special thank goes to Dr Sridhar R. Pulikanti for his guidance provided me during the last year of my PhD research. Thanks to Dr. Ashish Agalgaonkar and Dr. Lasanatha

Meegahapola for the assistance provided. I am indebted to Dr. Vick Smith, Gerrard

Drury, Sean Elphick and Esperanza Gonzalez who are with EEPQRC at UOW, given their immense support for administrative and software related matters. Many thanks also goes to Roslyn Causer-Temby and Sasha Nikolic of the School of Electrical, Com- puter and Telecommunications Engineering (SECTE) at UOW, involved in solving administrative related problems and providing perspectives. The technical assistance provided by the SECTE technical staﬀ is highly appreciated.

I’m extremely thankful for Dilini Kumarasinghe for her support and encourage- iii iv ment provided during the writing of this thesis. A special thanks go to my friends

Dr. Sankika Tennakoon and Dr. Prabodha Paranavithana, previously with the EEP-

QRC, and Radley De Silva, Kalyani Dissanayake, Upuli Jayathuga, Devinda Perera,

Dothinka Ranamuka, Brian Perera, Vidarshika Jayawardene, Kai Zou and Yinchin

Choo for being supportive in many ways especially during the hard times along the way.

Finally and most importantly, my heartiest gratitude goes to my parents, sis- ter and brother-in-law, niece and my wonderful relatives for their encouragement, guidance all the sacriﬁces you all made on behalf of me. Certiﬁcation

I, Nishad Mendis declare that this thesis, submitted in fulﬁlment of the requirements for the award of Doctor of Philosophy, in the School of Electrical, Computer and

Telecommunications Engineering, University of Wollongong, is entirely my own work unless otherwise referenced or acknowledged. This manuscript has not been submitted for qualiﬁcations at any other academic institute.

Nishad Mendis

v Abstract

Hybrid remote area power supply (RAPS) systems can be regarded as an emerging power generation technology for rural and remote communities. These power systems combine the best features of conventional (e.g. diesel based power generation) and non-conventional (e.g. renewable energy) power generation technologies. Hybridi- sation of such energy sources provides superior performance in terms of eﬃciency, lower carbon emission levels, reduced generation cost and improved supply quality and reliability.

Although hybrid RAPS systems seem to oﬀer promising solutions, there are various challenges associated with design and operation of such generating schemes which include: (a) voltage and frequency regulation on customer side, (b) control coordination between the system components (e.g. energy storage, dump load), (c) development of individual control strategies for each system component and (d) maximum power extraction from renewable energy resources. This thesis addresses the above stated issues in relation to wind based RAPS systems where the wind turbine generator performs as the main source of energy. In this regard, two types of popular wind turbine generator technologies, namely: doubly fed induction generator (DFIG) and permanent magnet synchronous generator (PMSG) are considered to form RAPS systems. In addition, auxiliary system components such as an energy storage system, dump load and other types of generating schemes including diesel and hydrogen are combined with wind turbine generator to perform the hybrid operation.

Robust control strategies are developed for the converter systems of the wind turbine generators with a view to regulate the voltage and frequency on the load side.

In addition, a battery storage system and a dump load are integrated to regulate the power balance of the RAPS systems. Moreover, separate conﬁgurations of the dump load are proposed for DFIG based and PMSG based RAPS systems. Individual vi vii controllers are implemented for the battery storage systems and dump loads whose operation is managed through a coordinated control approach. The coordinated control approach is designed to perform as an integrated controller of the RAPS system which manages the power ﬂow between the system components and coordinate responses of individual components in a designated manner. In addition, control strategies are developed to operate the wind turbine generators on their maximum power tracking characteristics to ensure optimum system performance.

Operation of a battery storage system is coordinated with a supercapacitor with a view to improve the battery life by reducing ripple content of battery current. Two diﬀerent power electronic conﬁgurations are proposed to interface the hybrid energy storage (i.e. battery storage and supercapacitor) of DFIG based and PMSG based

RAPS systems. The operation of the hybrid energy storage system is coordinated through the implementation of an energy management algorithm which is developed with a view to reduce the depth of discharge and ripple content of the battery current.

Applicability of a dual mode operation of diesel generating system (i.e. either as a synchronous condenser or as synchronous generator) for wind based RAPS system is examined. The dual mode operating mechanism is controlled via a friction clutch that helps to improve the fuel economy by avoiding the low load factor operation of the diesel generating scheme. Technical viability of such a diesel generating scheme is implemented giving due consideration to its modelling aspects together with respective controllers. Also, reactive power management schemes are implemented between the wind energy conversion system and diesel generating system.

To improve the autonomy of operation of the RAPS systems, hydrogen based generating schemes are introduced. In this regard, the technical feasibility of integrating a hydrogen based generating system consisting of a fuel cell system, an electrolyser and a storage tank to a wind based RAPS system is examined. Individual control viii strategies are implemented for each component of the hydrogen storage system and their functions are coordinated to perform as a self generating unit.

The performance evaluation utilising linearised component based RAPS system is also undertaken with a view to compare the results that are obtained using the corresponding detailed models. Above stated DFIG based and PMSG based RAPS systems are also investigated under changing wind and varying load conditions. Through simulation studies it is revealed that the proposed control strategies developed for the

RAPS systems are capable in regulating the voltage and frequency on the load side while extracting the maximum power from wind. List of Principal Symbols and Abbreviations

ρ Air density in Kgm−3

A Area swept by the rotor blades in m2 ib Battery current

Pb Battery power output vb Battery voltage

Csup Capacitance of the supercapacitor in F

−1 (vw)cut−in Cut-in wind speed in ms

−1 (vw)cut−out Cut-out wind speed in ms D Damping constant vdc DC bus voltage

ωd Diesel engine speed

Pde Diesel power output

Pd Dump load power vds, vqs d and q axes voltage on load side vdr, vqr d and q axes voltage on rotor side of the DFIG

φds, φqs d and q axes stator ﬂux components of DFIG

φdr, φqr d and q axes rotor ﬂux components of DFIG ielz Electrolyser current

Pelz Electrolyser power consumption

Te Electrical torque of wind turbine generator velz Electrolyser voltage F Faraday constant in Ckmol−1

ηF Faraday eﬃciency

Lf Filter inductance ∆f Frequency deviation ix x

Pfc Fuel cell power output fs Frequency on load side ifc Fuel cell current vfc Fuel cell voltage R Gas constant in J(kmolK)−1

M Inertia constant

σ Leakage factor of the DFIG

PL Load demand

Plf ,Phf Low and high frequency power component

Lm Magnetising inductance of the DFIG

Pm Mechanical power output of the wind turbine generator α Minimum loading condition of diesel generator

Qmag No-load reactive power output from DFIG

(Te)opt Optimum torque of the wind turbine generator

(Pw)opt Optimum wind power output β Pitch angle of wind turbine

∆p Power deviation

PDFIG Power output of the DFIG

PPMSG Power output of the PMSG kp,ki Proportional and integral gains of a PI controller

φrated Rated ﬂux of the DFIG

Rdump Resistance of the dump load

Qs Reactive power of the stator of the DFIG

Qde Reactive power output from diesel generator

QDFIG Reactive power output from DFIG

QPMSG Reactive power output from PMSG inverter xi

φr Rotor Flux of the DFIG

Rr Rotor resistance of the DFIG

Lr Rotor inductance of the DFIG

Pr Rotor power of the DFIG

ωr Rotor speed of the wind turbine generator s Slip of the DFIG

φs Stator ﬂux of the DFIG

Ls Stator inductance of the DFIG

Rs Stator resistance of the DFIG

Psc Supercapacitor power output

Ps Stator power of the DFIG

ωs Synchronous speed λ Tip-speed ratio vdcge Unregulated DC voltage output of PMSG vsc Voltage across the supercapacitor vs Voltage on load side

Pw Wind power output

−1 vw Wind speed in ms Publications arising from this Thesis

1. N. Mendis, K. Muttaqi and S. Perera, “Voltage quality behaviour of a wind tur-

bine based Remote Area Power System”, International Conference on Industrial

Electronics (ICIT2008), Gippsland, Australia , 10-13 Feb. 2008, pp. 1-6.

2. N. Mendis, K. Muttaqi, S. Sayeef and S. Perera, “Power generation in iso-

lated and regional communities: Application of a doubly-fed induction gener-

ator based wind turbine”, 19th Australasian Universities Power Engineering

Conference (AUPEC2009), Adelaide, Australia, 27-30 Sept. 2009, pp. 1-7.

3. S. Sayeef, N. Mendis and K. Muttaqi, “Optimisation of Component Sizes for

a Hybrid Remote Area Power Supply System”, 19th Australasian Universities

Power Engineering Conference (AUPEC2009), Adelaide, Australia , 27 30 Sept.

2009, pp. 1 - 6.

4. N. Mendis, K. Muttaqi, S.Sayeef and S. Perera, “Control Coordination of a

Wind Turbine Generator and a Battery Storage Unit in a Remote Area Power

Supply System”, Power Engineering Society General Meeting (PES2010), Min-

neapolis, Minnesota, USA, 25-29 Jul. 2010.

5. N. Mendis, K. Muttaqi, S.Sayeef and S. Perera, “Autonomous Operation of

Wind-Battery Hybrid Power System with Maximum Power Extraction Ca-

pability”, International Conference on Power System Technology (POWER-

CON2010), Hangzhou, China, 24-28 Oct. 2010.

xii xiii

6. N. Mendis, K. Muttaqi, Sayeef and S. Perera, “A Control Approach for Volt-

age and Frequency Regulation of a Wind-Diesel-Battery based Hybrid Remote

Area Power Supply System”, 36th Annual Conference of the IEEE Industrial

Electronics Society (IECON2010), Phoenix, Arizona , 7-10 Nov. 2010.

7. N. Mendis, K. Muttaqi, Sayeef and S. Perera, “Hybrid Operation of Wind-

Diesel-Fuel Cell Remote Area Power Supply System”, International Conference

on Sustainable Energy Technologies (ICSET2010), Kandy, Sri Lanka , 6-9 Dec.

2010.

8. S. Sayeef, N. Mendis and K. Muttaqi, “Enhanced Reactive Power Support of

a PMSG based Wind Turbine for a Remote Area Power System”, 20th Aus-

tralasian Universities Power Engineering Conference (AUPEC2010), Christchurch,

New Zealand , 5-8 Dec. 2010.

9. N. Mendis, K. Muttaqi, S. Sayeef and S. Perera, “Hybrid Operation of a Wind-

Diesel-Battery based Hybrid Remote Area Power Supply System”, International

Conference on Electrical and Computer Engineering (ICECE2010), Dhaka, Bangladesh

, 18-20 Dec. 2010.

10. N. Mendis, K. Muttaqi, S. Sayeef and S. Perera, “Application of a Hybrid En-

ergy Storage in a Remote Area Power Supply System”, International Energy

Conference (EnergyCON2010), Manama, Bahrain , 18-22 Dec. 2010. xiv

11. N. Mendis, S. Sayeef, K. Muttaqi and S. Perera, “Hydrogen Energy Storage

for a Permanent Magnet Wind Turbine Generator Based Autonomous Hybrid

Power System”, Power Engineering Society General Meeting(PES2011), De-

troit, Michigan, USA, 25-29 Jul. 2011.

12. N. Mendis, K. Muttaqi and S. Perera, “A Novel Control Strategy for Stand-

alone Operation of a Wind Dominated RAPS System”, Industry Applications

Society Annual Meeting(IAS2011), Orlando, Florida, USA, 9-13 Oct. 2011. Table of Contents

1 Introduction 1 1.1 Statement of the Problem ...... 1 1.2 Aim, Research Objectives and Methodologies ...... 7 1.2.1 Aim of the Research Work ...... 7 1.2.2 Research Objectives and Methodologies ...... 8 1.3 Outline of the Thesis ...... 10

2 Literature Review on Remote Area Power Supply Systems 12 2.1 Introduction ...... 12 2.2 An Overview of RAPS Systems ...... 13 2.3 Wind Energy Systems ...... 16 2.3.1 Wind Speed Distribution ...... 16 2.3.2 Wind Energy Conversion ...... 16 2.3.3 Maximum Power Extraction From Wind ...... 18 2.3.4 An Overview of General Wind Turbine Concepts ...... 21 2.3.5 Standalone Operation of Wind Turbine Generators ...... 23 2.4 Energy Storage Systems for Wind Power Applications ...... 26 2.4.1 Importance of Energy Storage Systems in Standalone Wind Power Applications ...... 26 2.4.2 Types and Sizing of Energy Storage Systems ...... 27 2.4.3 Energy Storage for Wind based Remote Area Power Supply System ...... 30 2.5 Diesel Generators for Standalone Wind Power Applications ...... 32 2.5.1 Importance of Diesel Generator Systems in Standalone Wind Power Applications ...... 32 2.5.2 Operating Principles of Diesel Generating System ...... 33 2.5.3 Operational Aspects of Wind-Diesel Remote Area Power Sup- ply Systems ...... 35 2.6 Hydrogen Based Storage Systems for Wind Power Applications . . . . 37 2.6.1 Operating Principles of a Fuel Cell System, Electrolyser and Storage Tank ...... 37 2.6.2 Operational Aspects of Wind-Fuel Cell based Remote Area Power Supply Systems ...... 41 2.7 Chapter Summary ...... 44

3 Wind Turbine Generator Technologies for RAPS Applications 45 3.1 Introduction ...... 45 3.2 Doubly-Fed Induction Generator Modelling, Operation and Control . 46 3.2.1 Overview of Operating Principle of the DFIG ...... 46 3.2.2 Mathematical Model of the DFIG ...... 48 3.2.3 Mathematical Model of the Back-to-Back Converter System . 51

xv xvi

3.2.4 Rotor Side Converter Control ...... 57 3.2.5 Line Side Converter Control ...... 59 3.3 Operating and Modelling Aspects of Permanent Magnet Synchronous Generator (PMSG) ...... 60 3.3.1 Overview of Operating Principles of PMSG ...... 60 3.3.2 Inverter Control of PMSG ...... 62 3.3.3 DC/DC Converter Control ...... 63 3.4 Active Power Control Techniques ...... 66 3.4.1 Pitch Angle Control ...... 66 3.4.2 Application of Dump Load for Remote Power Applications . . 67 3.5 Standalone Operating Performance of the Wind Turbine Generators in RAPS Environments ...... 70 3.5.1 Performance of the DFIG based RAPS System ...... 71 3.5.2 Performance of the PMSG based RAPS System ...... 79 3.6 Chapter Summary ...... 85

4 Application of Battery Energy Storage for Wind Energy Based RAPS Systems 87 4.1 Introduction ...... 87 4.2 Linearised Model of Wind Energy and Battery Storage based RAPS Systems ...... 88 4.3 Detailed Model of Wind-Battery Remote Area Power Supply Systems 90 4.3.1 Beneﬁts of Energy Storage System for a Standalone Wind Power Application ...... 90 4.3.2 Coordinated Control Approach of Wind-Battery RAPS system 93 4.3.3 Controller Design ...... 96 4.3.4 DFIG based Wind-Battery Hybrid RAPS System ...... 98 4.3.5 PMSG based Wind-Battery Hybrid RAPS System ...... 104 4.4 Performance Evaluation of the Hybrid Wind-Battery RAPS Systems 108 4.4.1 Performance of the Linearised Model of Wind-Battery RAPS System ...... 109 4.4.2 Performance of the DFIG-Battery RAPS System ...... 110 4.4.3 Performance of the PMSG-Battery RAPS System ...... 118 4.5 Chapter Summary ...... 127

5 Application of Hybrid Energy Storage for Standalone Wind Energy Conversion Systems 129 5.1 Introduction ...... 129 5.2 Comprehensive Model of Wind-Battery-Supercapacitor based Remote Area Power Supply System ...... 130 5.2.1 Importance of Hybrid Energy Storage System in a Standalone Wind Power Application ...... 130 5.2.2 A Coordinated Control Approach for Hybrid Energy Storage based RAPS Systems ...... 132 5.2.3 Energy Management Algorithm (EMA) ...... 133 xvii

5.2.4 Application of Hybrid Energy Storage in DFIG based RAPS System ...... 139 5.2.5 Application of Hybrid Energy Storage in PMSG based RAPS System) ...... 143 5.3 Simulation Results and Discussions ...... 145 5.3.1 Performance of the Hybrid Energy Storage in a DFIG based RAPS System ...... 145 5.3.2 Performance of the PMSG-Hybrid Energy Storage RAPS System151 5.4 Chapter Summary ...... 157

6 Wind-Diesel Hybrid RAPS Systems 159 6.1 Introduction ...... 159 6.2 Linearised Model of the Wind-Diesel and Energy Storage System . . 161 6.3 Modelling Aspects of a Diesel Generating System and its Diﬀerent Operating Modes for a Remote Power Application ...... 163 6.3.1 Importance of Diesel Generator System in a RAPS System . . 163 6.3.2 Dual Mode Operation of a Diesel Generating System . . . . . 164 6.4 Detailed Model of Wind-Diesel-Battery RAPS System ...... 169 6.4.1 Coordinated Control Approach ...... 169 6.4.2 DFIG based Wind-Diesel Hybrid RAPS System ...... 171 6.4.3 PMSG based Wind-Diesel Hybrid RAPS System ...... 175 6.5 Performance of the Hybrid Wind-Diesel-Battery based RAPS System 175 6.5.1 Performance of the linearised model of Wind-Diesel-Battery RAPS System ...... 176 6.5.2 Performance of the Detailed Model of DFIG-Diesel-Battery based RAPS System ...... 179 6.5.3 Performance of the PMSG-Diesel-Battery based RAPS System 184 6.6 Chapter Summary ...... 191

7 Hydrogen as Energy Storage for Wind Dominated RAPS System 193 7.1 Introduction ...... 193 7.2 Hydrogen Storage Systems ...... 194 7.2.1 Fuel cell System ...... 194 7.2.2 Electrolyser and Storage Tank ...... 197 7.3 Application of Hydrogen Storage for a Standalone Wind Energy System199 7.3.1 Hydrogen as Storage for DFIG-Diesel based Hybrid RAPS System200 7.3.2 Coordinated Control Approach for DFIG based RAPS System 202 7.3.3 PMSG based Wind-Hydrogen Hybrid RAPS System ...... 207 7.3.4 Coordinated Control Approach for PMSG based RAPS System 207 7.4 Performance of the Hybrid Wind-diesel-Battery RAPS System with Hydrogen as Energy Storage ...... 210 7.4.1 Performance of the DFIG-Diesel-Hydrogen based RAPS System 210 7.4.2 Performance of the PMSG-Hydrogen based RAPS System . . 215 7.5 Chapter Summary ...... 221 xviii

8 Conclusions and Recommendations for Future Work 223 8.1 Conclusions ...... 223 8.2 Recommendations for Further Work ...... 233

Appendices

A 250 A.1 Co-ordinate Transformation ...... 250 A.2 DFIG based RAPS System ...... 251 A.2.1 Internal Model Control (IMC) Principle for Tuning the PI Con- trollers Associated with the LSC ...... 255 A.2.2 Parameters of DFIG based RAPS System ...... 263 A.3 PMSG Based RAPS System ...... 263 A.4 Power Quality Issues Of DFIG based RAPS-A Case Study ...... 265

B 270 B.0.1 Wind Turbine Power Characteristics - Linearised Model . . . . 270 B.0.2 RAPS System Parameters ...... 271 B.0.3 Wind Turbine Power Characteristic Curves ...... 273 B.0.4 Torque Constant Estimation for the Induction Motor Driven Pump Load ...... 274 B.0.5 Battery Inverter Control for DFIG based Wind-Battery-Supercapacitor RAPS system ...... 275 B.0.6 Estimation of Supercapacitor Current for DFIG based RAPS System ...... 277

C 279 C.0.7 Wind-Diesel RAPS System Performance ...... 279 C.0.8 Modiﬁed RSC Arrangement ...... 284 C.0.9 Diesel Engine Model and Exciter System ...... 285 C.0.10 Parameters of the Wind-Diesel-Battery RAPS System . . . . . 286

D 288 D.0.11 Parameters Associated with Hydrogen Based RAPS Systems . 288 D.0.12 Parameters Associated with Components of the Hydrogen Stor- age System ...... 290 List of Figures

2.1 A typical arrangement of a wind power based standalone power supply system...... 13 2.2 Probability density of the Rayleigh distribution at King Island-Tasmania. 17 2.3 Typical power curve for a variable speed pitch controlled wind turbine. 19 2.4 Wind turbine power characteristics with maximum power extraction. 21 2.5 Typical configurations of wind turbine technologies: (a) Type A, (b) Type B, (c) Type C and (d) Type D...... 22 2.6 Application of an energy storage in a WEC system (a) wind power output, Pw, and load demand, PL, and (b) energy storage charging/discharging status...... 27 2.7 Specific energy versus specific power ranges for various energy storage systems...... 28 2.8 Different configuration of diesel generating systems: (a) fixed speed operation, (b) fixed speed with dual mode operation and (c) variable speed operation...... 35 2.9 Polarisation curve of a fuel cell...... 39 2.10 Polarisation curve of an electrolyser...... 41

3.1 Typical configuration of DFIG...... 47 3.2 Steady-state modified equivalent circuit of the DFIG...... 48 3.3 Space vector equivalent circuit for arbitrary reference frame...... 49 3.4 Stator flux oriented vector representation...... 53 3.5 Stator flux oriented equivalent circuit of a DFIG(superscript S denotes that the space vectors are referred to the stator flux reference frame). 54 3.6 Filter model associated with LSC...... 55 3.7 Voltage vector orientation scheme of the LSC...... 56 3.8 RSC control scheme...... 59 3.9 LSC control scheme...... 60 3.10 Simplified single phase equivalent circuit of round pole PMSG. . . . . 61 3.11 Typical configuration of PMSG wind energy system...... 61 3.12 Inverter control of PMSG based RAPS system...... 63 3.13 Boost converter operation for regulation of the DC bus voltage. . . . 65 3.14 Control strategy of the boost converter of the PMSG based WECS. . 65 3.15 Pitch angle control strategy for a variable speed wind turbine generator. 66 3.16 Dump load control strategy of the DFIG...... 68 3.17 Dump load and its controller for PMSG...... 70 3.18 DFIG based hybrid RAPS system...... 71 3.19 Performance of the DFIG wind turbine system: (a) wind velocity, (b) speed of DFIG and (c) pitch angle...... 72 3.20 Response of the DFIG based RAPS system: (a) voltage on load side, (b) frequency on load side and (c) DC bus voltage...... 74

xix xx

3.21 Power sharing between system components: (a) DFIG power output, (b) dump load power and (c) load demand...... 74 3.22 Reactive power sharing between DFIG and loads...... 75 3.23 Actual and reference d-axis component currents of RSC...... 76 3.24 Actual and reference q-axis component currents of RSC...... 76 3.25 Stator ﬂux components of the DFIG in d-q domain...... 77 3.26 Actual and reference d-axis component currents of LSC...... 78 3.27 Actual and reference q-axis component currents of LSC...... 78 3.28 PMSG based hybrid RAPS system...... 79 3.29 Performance of the PMSG wind turbine system: (a) wind velocity and (b) speed of wind turbine generator...... 80 3.30 Response of the PMSG based RAPS system: (a) voltage on load side, (b) frequency on load side and (c) DC bus voltage...... 81 3.31 Power sharing between system components: (a) PMSG power output, (b) dump load power and (c) load demand...... 81 3.32 Reactive power sharing between inverter and loads...... 82 3.33 Actual and reference d-axis component currents of the inverter. . . . 83 3.34 Actual and reference q-axis component currents of the inverter. . . . 83 3.35 Inverter voltage components d-q domain...... 84

4.1 The linearised block diagram of the wind-battery hybrid RAPS system. 90 4.2 Schematic of the simpliﬁed standalone power supply system...... 91 4.3 Controller for the energy storage system...... 91 4.4 RAPS system performance with battery storage system: (a) supply voltage, (b) DC link voltage and (c) battery current...... 92 4.5 DC link voltage in the absence of the battery storage...... 93 4.6 Control coordination of a wind-battery hybrid power system...... 94 4.7 Power ﬂow directions of the components during (a) over-generation and (b) under-generation...... 95 4.8 Decision making process associated with the state transition of a device. 97 4.9 DFIG based wind-battery RAPS system...... 99 4.10 Bi-directional buck-boost converter arrangement for battery storage system...... 101 4.11 Battery storage control strategy for DFIG...... 102 4.12 PMSG based wind-battery RAPS system...... 105 4.13 Dynamics of DC bus of the PMSG arrangement...... 106 4.14 Maximum power extraction control strategy for DC/DC converter-1. 108 4.15 Power Sharing of the RAPS system under variable wind and load conditions: (a) wind speed, (b) power output of wind turbine generator, (c) battery storage power output, (d) load demand...... 110 4.16 (a) power imbalance and (b) frequency deviation of the RAPS system. 111 4.17 Response of the DFIG based wind-battery RAPS system under wind and load step changes: (a) wind speed, (b) voltage on load side, (c) frequency on load side, and (d) DC link voltage...... 112 xxi

4.18 Power sharing of the DFIG based wind-battery RAPS system under wind and load step changes: (a) wind generator power output, (b) battery storage power, (c) dump load power and (d) load demand. . . . . 113 4.19 Maximum power extraction from wind in the DFIG based wind-battery RAPS system under wind and load step changes...... 114 4.20 Response of the DFIG based wind-battery RAPS system under high wind regimes: (a) wind speed, (b) voltage on load side, (c) frequency on load side, and (d) DC link voltage...... 115 4.21 Power sharing of the DFIG based wind-battery RAPS system under high wind regimes: (a) wind generator power output, (b) battery storage power, (c) dump load power, (d) load demand and (e) pitch angle. 115 4.22 Maximum power extraction from wind in the DFIG based wind-battery RAPS system under high wind regimes...... 116 4.23 Response of the DFIG based wind-battery RAPS system with an induction pump load: (a) wind speed, (b) voltage on load side (i.e. system voltage), (c) frequency on load side, and (d) DC link voltage...... 117 4.24 Power sharing of the DFIG based wind-battery RAPS system with an induction motor driven pump load: (a) wind generator power output, (b) battery storage power and (c) load demand...... 118 4.25 Response of the PMSG based wind-battery RAPS system under wind and load step changes: (a) wind speed, (b) voltage on load side, (c) frequency on load side and (d) DC link voltage...... 120 4.26 Power sharing of the PMSG based wind-battery RAPS system under wind and load step changes: (a) wind generator power output, (b) battery power, (c) dump load power and (d) load demand...... 121 4.27 Behaviour of DC link voltage of PMSG without employing dump load under wind and load step changes...... 122 4.28 Maximum power extraction from wind in PMSG based wind-battery RAPS system under wind and load step changes...... 122 4.29 Response of the PMSG based wind-battery RAPS system under a realistic wind profile: (a) wind speed, (b) voltage on load side, (c) frequency on load side and (d) DC link voltage...... 123 4.30 Power sharing of the PMSG based wind-battery RAPS system under a realistic wind profile: (a) wind power, (b) battery power, (c) dump load power and (d) load demand...... 124 4.31 Maximum power extraction from wind in PMSG based RAPS system under realistic wind profile. (The PwActual is as same as Pw in Fig. 4.30-(a)) ...... 124 4.32 Response of the PMSG based wind-battery RAPS system with an induction pump load: (a) wind speed, (b) voltage on load side, (c) frequency on load side and (d) DC link voltage...... 125 xxii

4.33 Power sharing of the PMSG based wind-battery RAPS system with an induction motor driven pump load: (a) wind power, (b) battery power, (c) dump load power and (d) load demand...... 126

5.1 Simpliﬁed model of a power system with hybrid energy storage. . . . 131 5.2 Current sharing between battery storage and supercapacitor...... 132 5.3 Control coordination of a wind-battery-supercapacitor based hybrid RAPS system...... 134 5.4 Battery voltage under diﬀerent discharge rates...... 136 5.5 Estimation of reference power for battery storage and supercapacitor. 136 5.6 Operating frequency ranges of the energy storage systems: supercapacitor and battery storage system...... 136 5.7 Equivalent circuits of supercapacitor (a) high frequency model and (b) low frequency model...... 138 5.8 Hybrid energy storage in a DFIG based RAPS system...... 140 5.9 Inverter control of the battery storage system for DFIG wind-hybrid energy storage based RAPS system...... 141 5.10 Control strategy for supercapacitor in a hybrid energy storage of a DFIG based RAPS system...... 142 5.11 Hybrid energy storage in a PMSG based RAPS system...... 144 5.12 Energy management scheme for a hybrid energy storage in a PMSG based RAPS system...... 144 5.13 Response of the DFIG based wind-hybrid energy storage RAPS system during variable wind and load conditions: (a) wind speed, (b) voltage on load side, (c) frequency on load side, and (d) DC link voltage. . . 146 5.14 Power sharing of the DFIG based wind-hybrid energy storage RAPS system during variable wind and load conditions: (a) wind power, (b) hybrid energy storage power (i.e. battery power, Pb and supercapacitor Psc power (c) dump load power and (d) load demand...... 148 5.15 MPPT tracking capability of the DFIG in a RAPS system with energy storage integrated...... 148 5.16 Battery current for the case with no supercapacitor of the DFIG based RAPS system with hybrid energy storage integrated...... 149 5.17 Currents of the hybrid energy storage: battery current, ib and supercapacitor current, ic of DFIG based RAPS system...... 150 5.18 Frequency spectrum of the battery storage system of a hybrid energy storage in a DFIG based RAPS system...... 150 5.19 Frequency spectrum of the supercapacitor of hybrid energy in a DFIG based storage RAPS system...... 151 5.20 Response of the PMSG based wind-hybrid energy storage RAPS system under variable wind and load conditions. (a) wind Speed, (b) voltage on load side, (c) frequency on load side, and (d) DC link voltage. 152 xxiii

5.21 Power sharing of the PMSG based wind-hybrid energy storage RAPS system under variable wind and load conditions. (a) wind Power, (b) battery power, (c) supercapacitor power (d) dump load power and (e) load demand...... 154 5.22 Maximum power extraction from wind in the PMSG based RAPS system with hybrid energy storage integrated...... 154 5.23 Battery current for the case with no supercapacitor in the PMSG based RAPS system with hybrid energy storage integrated...... 155 5.24 Currents of the hybrid energy storage: battery current, ib and supercapacitor current, ic of the PMSG based RAPS system with hybrid energy storage integrated...... 156

6.1 The linearised block diagram of the proposed RAPS system...... 162 6.2 Specific fuel consumption of a loaded diesel engine...... 165 6.3 Dynamics associated with the clutch system of the diesel generating system...... 167 6.4 Generation of clutch signal of the diesel engine...... 168 6.5 Diesel engine model...... 169 6.6 Instantaneous power flow control of the wind-diesel-battery RAPS system...... 172 6.7 DFIG based wind-diesel-battery RAPS system...... 173 6.8 Battery storage controller for the DFIG based wind-diesel-battery RAPS system...... 175 6.9 PMSG based wind-diesel-battery RAPS system...... 176 6.10 Power sharing of the linearised model of the wind-diesel-battery RAPS System under variable wind and load conditions: (a) wind speed, (b) wind power, (c) diesel power,(d) battery power, and (e) load demand.178 6.11 (a) Frequency deviation, and (b) power imbalance of the linearised wind-diesel-battery RAPS system. (‘Delta’ in this figure represents ‘∆’.)178 6.12 Response of the DFIG based wind-diesel-battery RAPS system under variable wind and load conditions: (a) wind speed, (b) load side voltage, (c) frequency on load side, and (d) DC link voltage...... 181 6.13 Power sharing of the DFIG based wind-diesel-battery RAPS system under variable wind and load conditions: (a) wind power, (b) diesel power, (c) battery power and (d) load demand...... 182 6.14 Frequency and active power deviation of the DFIG based wind-diesel- battery RAPS system: (a) active power imbalance and (b) frequency deviation...... 183 6.15 Maximum power point tracking characteristics of DFIG based wind turbine generator of the wind-diesel-battery RAPS system...... 183 6.16 Reactive power sharing between the components of the DFIG based wind-diesel-battery RAPS system...... 184 6.17 Diesel generator performance of the DFIG based wind-diesel-battery RAPS system: (a) rotor speed and (b) load angle...... 185 xxiv

6.18 Speeds of the engine, synchronous machine and clutch signal of the DFIG based wind-diesel-battery RAPS system...... 185 6.19 Response of the PMSG based wind-diesel-battery RAPS system under variable wind and load conditions: (a) wind speed, (b) load side voltage, (c) frequency on load side, and (d) DC link voltage...... 187 6.20 Power sharing of the PMSG based wind-diesel-battery RAPS system under variable wind and load conditions: (a) wind power, (b) battery power, (c) diesel power, and (d) load demand...... 188 6.21 Maximum power point tracking characteristics of PMSG based wind turbine generator for wind-diesel-battery RAPS system...... 189 6.22 Active power imbalance of the PMSG based wind-diesel-battery RAPS system. (“Delta” in this ﬁgure represents ∆) ...... 189 6.23 Reactive power sharing between the system components of the PMSG based wind-diesel-battery RAPS system...... 190

7.1 Detail model of the SOFC system...... 196 7.2 Equivalent circuit of a fuel cell...... 197 7.3 Model of Hydrogen storage...... 199 7.4 DFIG based wind-diesel-hydrogen RAPS system...... 201 7.5 Control coordination approach for DFIG based wind-diesel-hydrogen hybrid RAPS system...... 202 7.6 Switching signal of boost converter of the fuel cell, ((Pfc)ref = PL − (Pw)opt)...... 205 7.7 (a)V-I and (b) Power characteristics of the SOFC fuel cell system. . . 205 7.8 Switching signal of buck converter of the electrolyser...... 206 7.9 V-I characteristic of a electrolyser unit...... 207 7.10 PMSG based wind-hydrogen RAPS system...... 208 7.11 Control coordination approach for PMSG based wind-hydrogen hybrid RAPS system...... 209 7.12 Response of the DFIG based wind-diesel-hydrogen RAPS system under variable wind and load conditions: (a) wind speed, (b) voltage on load side (c) frequency on load side, and (d) DC link voltage...... 211 7.13 Power Sharing of the DFIG based wind-diesel-hydrogen RAPS system under variable wind and load conditions: (a) wind power, (b) fuel cell power, (c) electrolyser power, (d) diesel power and (e) load demand. . 213 7.14 Maximum power point tracking characteristic of DFIG based wind- diesel-hydrogen RAPS system...... 214 7.15 Speeds of the engine, synchronous machine and clutch signal of DFIG based wind-diesel-hydrogen RAPS system...... 214 7.16 Performance of the hydrogen based power generation unit in DFIG based wind-diesel-hydrogen RAPS system: (a) molar ﬂow rate of electrolyser, (b) molar ﬂow rate of fuel cell and (c) tank pressure. . . . . 215 7.17 Voltage characteristics of fuel cell and electrolyser of DFIG based wind- diesel-hydrogen RAPS system...... 216 xxv

7.18 Response of the PMSG based wind-hydrogen RAPS system under variable wind and load conditions: (a) wind speed, (b) voltage on load side, (c) frequency on load side, and (d) DC link voltage...... 217 7.19 Power Sharing of the PMSG based wind-hydrogen RAPS system under variable wind and load conditions: (a) wind power, (b) fuel cell power, (c) electrolyser power and (d) load demand...... 218 7.20 Maximum power point tracking from wind in PMSG based wind-hydrogen RAPS system...... 219 7.21 Performance of the hydrogen based power generation unit of PMSG based wind-hydrogen RAPS system: (a) molar ﬂow rate of electrolyser, (c) molar ﬂow rate of fuel cell and (b) tank pressure...... 220 7.22 Voltage characteristics of fuel cell and electrolyser of PMSG based wind-hydrogen RAPS system...... 220

A.1 Control structure (a) IMC control structure and (b) Classical PID control structure...... 256 A.2 IMC based control structure for the LSC ﬁlter. Where (if )ref and if are reference and actual ﬁlter currents respectively...... 258 A.3 LSC arrangement with DC bus...... 260 A.4 IMC based control structures: (a) without active damping and (b) with active damping ...... 261 A.5 PMSG: (a) d-axis circuit (b) q-axis circuit...... 263 A.6 Boost converter of the PMSG...... 264 A.7 DFIG based remote area power supply system...... 265 A.8 Case I - THD at PCC and DFIG busbar with resistive load only. . . . 266 A.9 Case II-THD at PCC and DFIG busbar with resistive load and induction motor load...... 267 A.10 Frequency spectrum of voltage at PCC with resistive load only. . . . 268 A.11 Frequency spectrum of voltage at PCC with resistive and induction motor load...... 268 A.12 Switching harmonics of voltage at PCC with resistive and induction motor load...... 269

B.1 Power characteristics of DFIG based wind turbine...... 273 B.2 Power characteristics of PMSG based wind turbine...... 274 B.3 Torque speed characteristic of an induction motor driven pump load. 275

C.1 Wind-Diesel-Battery hybrid remote area power supply system with the circuit breaker arrangement...... 280 C.2 Response of the DFIG based wind-diesel RAPS system under variable wind and load conditions: (a) wind speed, (b) voltage on load side, (c) frequency on load side, and (d) DC link voltage...... 281 xxvi

C.3 Power sharing of the DFIG based wind-diesel RAPS system under variable wind and load conditions: (a) wind power, (b) diesel power, (c) battery power and (d) load demand...... 282 C.4 Maximum power point tracking characteristic from wind in the DIF based wind-diesel RAPS system...... 283 C.5 Modiﬁed RSC control arrangement for unity power factor operation of the DFIG...... 284 C.6 Model of the diesel engine with its associated control...... 285 C.7 IEEE type I exciter system...... 285 List of Tables

2.1 Types of stanadlone power supply systems ...... 19 2.2 Diﬀerent types of fuel cell systems ...... 39

4.1 Transfer function parameters of wind generator, energy storage system and load demand ...... 89 4.2 Control logic to determine the device status ...... 98

6.1 Transfer function parameters of wind generator, diesel generator energy storage and load demand ...... 162 6.2 Control logic associated with mode transition of the diesel generating system ...... 166 6.3 Active and reactive power support from each system component . . . 174 6.4 “ON” and “OFF” conditions of the system components ...... 179

8.1 Qualitative comparison of diﬀerent types of RAPS systems ...... 229

A.1 Parameters of DFIG ...... 263 A.2 Parameters of PMSG ...... 264

B.1 Parameters of DFIG based RAPS System ...... 271 B.2 Parameters of PMSG-Battery-Dump load RAPS System ...... 272

C.1 Parameters of DFIG-Diesel-Battery-Dump load RAPS System . . . . 286 C.2 Parameters of PMSG-Diesel-Battery-Dump load RAPS System . . . . 286 C.3 Parameters of 350 kW diesel generator used for DFIG based RAPS system ...... 287 C.4 Parameters of 85 kW diesel generator used for DFIG based RAPS system287

D.1 Parameters of DFIG-Diesel-Hydrogen RAPS System ...... 288 D.2 Parameters of PMSG-Hydrogen RAPS System ...... 289 D.3 Parameters of the fuel cell system ...... 290 D.4 Parameters of the electrolyser system ...... 291

xxvii Chapter 1

Introduction

1.1 Statement of the Problem

Electricity is identiﬁed as one key commodity which can be used as a medium for economic growth in rural and regional areas. In these locations, electricity is mainly used for lighting, heating and other household purposes. In addition, it can be utilised for mechanisation of farming operations (e.g. threshing, milking and hoisting grain for storage etc.) which increase the economic productivity and strengthen the social cohesion. Electricity for remote areas that are located close to a main grid1 can be supplied by extending the existing grid relatively cheaply. However, in the newly formed rural areas including islands, the cost of supplying electricity to every new customer has increased. Further, the income levels of dwellers in remote locations are relatively low and tend to purchase less electricity which will lead to reduced

financial returns to the utilities2. These factors do not help promotion of grid-based rural electrification schemes as the first choice to serve rural communities. Instead, locality and decentrality based generation schemes are considered as viable methods in

1This is often used loosely to describe the totality of the network with its stiﬀness. 2In general this is used to explain commercial entities who engage in power generation, transmis- sion and distribution.

1 2 supplying electricity to remote customers. The ﬁrst approach is based on independent power generating schemes which target a single customer (e.g. solar photovoltaic for houses, diesel generators for factories etc.). The power generating schemes that are able to supply electricity to a large community (e.g. mini-grids to supply power to village) is covered by the second approach.

Decentralised power schemes are categorised as remote area power supply (RAPS) systems that supply power to rural communities. The importance of employing RAPS systems for rural electriﬁcation is further justiﬁed by the report, World Energy Out- look 2010, released by the International Energy Agency3. According to this report,

1.2 billion people in the world will continue to seek access to electricity by 2030, 87% of them will live in rural areas. Further, this report states that, in providing universal access to electricity by 2030, only 30 % of rural communities will have access to main grid supply systems while 75% of the remainder will need mini-grids and the other

25% requires standalone power supply systems. Currently, majority of the remote locations are supplied by diesel power based generating systems due to their low installation cost, reliability and simplicity in operation. One of the major drawbacks of this type of dencentralised generating systems is the access to the fuel which may not be possible throughout the year. In addition, suﬃcient mechanical skills must also be available to maintain the equipment in proper operating conditions. Furthermore, poor eﬃciency at low load conditions, environmental concerns associated with the use of diesel generating systems and high diesel prices have been detrimental to their popularity as a viable generating method.

With increasing importance being placed on energy security and sustainable development, role of renewable energy based power generating schemes has become ever more signiﬁcant, especially for rural electriﬁcation. The well known renewable energy

3The International Energy Agency (IEA) is an autonomous organisation which works to ensure reliable, affordable and clean energy for its 28 member countries and beyond. 3 forms include sunlight, wind, rain, tides, bio-mass and geothermal heat which can be transformed into electricity using appropriate technologies. At present, nearly 20% of the world’s energy demand is met through renewable energy based power generating schemes which include both gird-connected and off-grid schemes. However, one of the major challenges associated with renewable based power generating schemes is their intermittency4. To overcome this issue, hybrid remote area power supply schemes are now considered as a new emerging technological solution in supplying electricity to isolated and remote areas. Typically, a hybrid RAPS system is equipped with a primary energy source (e.g. a renewable energy source such as wind, solar), secondary energy source (e.g. diesel or liquefied petroleum gas (LPG) generators) and possibly other auxiliary components (e.g. battery storage, supercapacitor, flywheel, dump load etc.). Hybrid RAPS systems usually trace the best quality features of each of the energy resources available and able to supply grid-quality electricity to the remote customers. Also, such RAPS systems can be designed to operate at improved effi- ciencies while reducing their environmental impacts and ensuring that the generation costs are comparable to those of conventional generating schemes (e.g. diesel generators). Further, hybrid RAPS systems can also be retrofitted into the existing diesel based power systems easily. Moreover, hybrid RAPS systems are able to be upgraded to form grid interactive microgrids through grid connection in the future. Another advantage of hybrid RAPS systems consisting of several other types of energy sources

(e.g. diesel generators, energy storage systems) is avoidance of dependency to a single large capacity renewable energy source (e.g. wind generators, solar PVs).

Although hybrid RAPS systems seem to oﬀer attractive advantages and solutions in rural electriﬁcation, the design, control and operational aspects involved with such power systems are very sophisticated. As stated earlier, due to the intermittency and

4That is not continuously available. 4 uncertainties, the renewable energy resources are not able to supply reliable and quality power to remote customers unless it is appropriately conditioned and managed.

Therefore, renewable energy based power generating schemes are always equipped with the power electronic arrangements along with their respective control techniques.

However, the use of power electronic interfaces in hybrid RAPS systems creates many challenges. Firstly, power electronic interfaces lower the overall system inertia which will adversely affect the voltage and frequency at the customer end. Secondly, the costs associated with power electronic systems are considerably high. Therefore, it is vital to select the best hybrid RAPS configuration without compromising on the supply quality and reliability. In addition to the above, other challenges associated with RAPS systems are: (a) coordination of the functions among all components, (b) harvesting maximum power output from renewable energy sources, (c) maintaining the power quality and supply reliability and (d) optimising the financial returns.

Selection of suitable energy sources to form a hybrid RAPS system depends entirely on the availability of resources within the locality. Among all renewable energy options, wind power has gained the momentum as one of the most widespread and commercially viable renewable energy generation technologies. Wind power generation has gained signiﬁcant levels of deployment in many countries over the world due to its economic potential. According to the World Wind Energy Report5 2010, all wind turbines installed globally by the end of year 2010 contribute potentially 430

Terawatthours to the worldwide electricity supply which supply 2.5 % of the global electricity demand per year. Although wind power generation schemes are seen to oﬀer great opportunities in supplying power, they encompass many challenges in standalone mode of operation. While being a renewable energy source, wind power generation is characterised by its intermittency and hence it should be operated with

5This is prepared by World Wind Energy Association (WWEA); http://www.wwindea.org 5 other components (e.g. diesel generator, energy storage and dump load etc.) to supply reliable power to remote customers.

The challenges and complexities associated with the design of hybrid wind based

RAPS systems include but not limited to: (a) the selection of the technology of the conﬁguration6, (b) design of an energy management scheme, (c) methods of supplying additional reactive power, (d) developing individual controllers for each component,

(e) operating the backup generators7 with minimal fuel consumption and (e) maximum power extraction from wind for optimal operation of the wind turbine generator.

To minimise the generation-demand mismatch, it is vital to establish an energy management mechanism to direct and control the power delivery between the components within the RAPS system. Further, the operation of the wind turbine generator can be maximised by operating at its maximum power tracking point8. To address these challenges, control strategies for each system component should be designed and implemented. Unlike in a grid-connected system, the reactive power supply through a wind generating system is limited or may not be possible. If the reactive power management is not properly implemented and coordinated, RAPS systems can experience severe voltage problems that may lead to voltage collapse. As a solution, a diesel generator can be operated parallel with the wind generator system to provide the required reactive power of the RAPS system. As stated earlier, diesel generators exhibit poor eﬃciency at low load conditions9. Therefore, control mechanisms should be designed and implemented to restrict the diesel generator operation at low load conditions. Similar to the diesel generator system, the design and control of the other auxiliary components (e.g. energy storage, dump load etc.) are imperative to achieve

6The technology of the conﬁgurations can be classiﬁed according to the voltage they are coupled with; this is, using DC, AC and mixed (DC and AC) bus lines 7e.g. diesel generators, LPG generators 8To achieve the greatest possible power harvest, during moment to moment variations of wind 9Typically the diesel generators should be operated 20-30% above its rated capacity. 6 a stable hybrid operation of the RAPS system.

Selection of a specific wind turbine generator technology is also an important design factor in a wind based RAPS system. Usually, variable speed wind turbine generator technologies are preferred in a standalone power system, as they are able to provide better voltage and frequency regulation when compared to constant speed generators such as induction generators. In this regard, doubly fed induction generators (DFIGs) and permanent magnet synchronous generators (PMSGs) are identified as preferred variable speed generator technologies for wind power applications. How- ever, the selection of the size and technology of a wind turbine generator for a specific application is based on several factors such as maximum load demand, financial returns, wind profile data and technical aspects covering their reactive power capability, low voltage ride through capability10 etc. It is well known that DFIG based wind turbine generator systems are preferred for high power applications whereas PMSG based wind turbines are suitable for low or medium power levels. The two types of wind turbine generators: namely DFIG and PMSG can be implemented with different system configurations/layouts with a conventional generator and other auxiliary components to perform their hybrid standalone operation. In this regard, the control strategies needed for each component of the two RAPS systems (i.e. PMSG based and DFIG based) are unique and different from one to another. Thus, design and modelling aspects of such power systems are seen to be a challenging task.

Most of the studies that have been undertaken in relation to wind energy applications have investigated the performance of induction generator based RAPS systems.

Although research attention has been paid to examine the performance of standalone

DFIG and PMSG based RAPS systems, they lacked detailed attention to design, modelling and control aspects. Further, most of the existing work cover speciﬁc be-

10Ability to get connected a wind turbine generator in an event of a fault 7 haviour of the component/s of RAPS systems rather than investigating the overall system behaviour. In addition, majority of the research outcomes have been reported employing simplistic mathematical models (e.g. linearised models) of the components in a RAPS system which are not suitable to examine their dynamics precisely.

1.2 Aim, Research Objectives and Methodologies

1.2.1 Aim of the Research Work

As stated in Section 1.1, the aim of the work presented in this thesis is to investigate the hybrid operation of wind dominated hybrid RAPS systems under changing wind speed and variable load conditions. From a customer perspective, the magnitude of voltage and its frequency at load side are the most important features that need speciﬁc attention in addition to the supply reliability. The aim of the research work presented in this thesis can hence be stated as:

Design and development of wind dominated RAPS systems to regulate the load side voltage and frequency within acceptable limits taking into consideration wind speed variations and load changes.

To achieve the above stated main objective, it is vital to manage the active and reactive power contributions of the components of the RAPS system. In this regard, control coordination strategies are to be developed and implemented amongst other components present within the RAPS system. In addition, individual control algorithms are to be developed based on an appropriate coordinated control approach with a view to regulate the magnitude of the voltage and its frequency on the load side. In this thesis, the following RAPS systems consisting of diﬀerent types of components, including a wind turbine generator as the main component are considered: 8

• wind turbine generator, battery storage with dump load

• wind turbine generator, battery storage, supercapacitor and dump load

• wind turbine generator, battery storage, diesel generator and dump load

• wind turbine generator, fuel cell system, electorlyser and hydrogen storage tank

The suitability of the proposed controllers for each system component and control coordination methodology that is unique for each RAPS system needs to be investigated under variable wind and changing load conditions. Also, emphasis is placed on the operation of the RAPS systems covering the scenarios such as over-generation, under-generation and emergency conditions.

1.2.2 Research Objectives and Methodologies

Several tasks have been undertaken with a view to achieve the following objectives which contribute to fulﬁll the main research aim stated in Section 1.2.1. These objectives include the extension of existing techniques (e.g. wind turbine control techniques and development of new concepts (e.g. coordinated control approaches).

As the work presented in this thesis is based on a wind dominated (i.e. high wind penetrated) hybrid RAPS systems, the control associated with the inverters of wind turbine generators (i.e. DFIG and PMSG) are identified as the key contributors for regulating the voltage and frequency on the load side. In this regard, as the first objective, vector control algorithms have been developed and implemented using the d-q vector control approach. As stated earlier, however, the wind turbine generator control alone cannot be used to regulate its voltage and frequency on the customer side. Therefore, the second objective is to select appropriate RAPS topology and to extend the system configuration with one or more of the other components such as 9 diesel generator, energy storage systems, dump load, hydrogen energy system11. In this regard, the modelling aspects of the system components have been undertaken considering the scope of the study. As an example, various types of modelling aspects

(i.e. electrical, thermal or as a combination of both) are available in the current literature to characterise the behaviour of the fuel cell system alone. To precisely examine its electrical dynamics relevant to the work presented in thesis, a fuel cell system can be used to operate at constant temperature by neglecting its thermal characteristics.

The third objective is to investigate the behaviour of the RAPS systems stated earlier in this section. In this regard, the auxiliary components have been integrated giving due attention to their power electronic interfaces. In addition, individual controllers have been proposed that ultimately contribute to achieve the main objective (i.e. to regulate the load side voltage and frequency). Also, control coordination methodologies have been proposed which are used as the basis to design the controllers associated with each system component. These control coordination approaches are aimed at energy management between all system components, which are considered to be the key to coordinate the diﬀerent generators/sources/auxciliary components.

Due to the fundamental differences associated with the working principles of DFIG and PMSG, there are differences between the modelling aspects of the components, connection topology and controller objectives. The suitability of each RAPS system is investigated in relation to the load side voltage and frequency regulation capability taking into consideration the wind speed changes and load step variations. Also, the behaviour of each RAPS system has been observed under different situations such as over-generation, under-generation and emergency situations12. In addition to the detailed13 simulation of the RAPS systems, corresponding linearised models of the

11This consists of fuel cell system, electrolyser and hydrogen storage tank. 12An example of no-wind condition where wind farm need to be disabled its operation. 13This mainly refers to describe the use of non-linear higher order mathematical models. 10 hybrid RAPS systems have also been considered with the aim of comparing with the results that have obtained using the detailed models of the RAPS systems. Simu- lation studies have been carried out using SimPowerSystems blocksets in MATLAB which is identiﬁed as one of the best simulation platforms to design and investigate the performance of power systems in general.

1.3 Outline of the Thesis

A brief description on the contents of the remaining chapters is given below:

Chapter 2 is a literature review providing an overview on general background on wind turbine generator based RAPS systems. The general working principles of diﬀerent types of wind turbine generator technologies are illustrated. A basic introduction, followed by a review on challenges associated with hybrid RAPS systems is discussed. The key section of this chapter describes the importance of integrating the auxiliary components which provides the background information required to carry out work presented in Chapters 4 - 7.

In Chapter 3, the modelling aspects and an approach for developing control methodologies which will be applied for both the DFIG and PMSG for their standalone operation are illustrated. Also, diﬀerent types of dump load conﬁgurations and their control strategies are explained. The proposed control algorithms relevant to wind turbine generator and dump load are validated in relation to voltage and frequency regulation capability.

Taking the wind turbine generator models presented in Chapter 3 as the basis, the importance of integrating energy storage for wind energy systems is highlighted in Chapter 4. In this regard, the application of battery storage system to ensure power balance is illustrated. Control strategies are developed and explained for the battery storage system that contribute towards regulating the load side voltage and 11 frequency of the hybrid RAPS system.

The necessity for incorporating a supercapacitor that improves the life span of battery storage is explained in Chapter 5. In this regard, the operation of the battery storage system is coordinated with the supercapacitor to perform as a hybrid energy storage. Diﬀerent power electronic conﬁgurations are proposed for the hybrid energy storage systems and their respective energy management algorithms are developed with an aim to achieve lower depth of discharge (DOD) rates and reduced ripple in the battery current.

As a continuation of the work presented in Chapter 4 and 5, Chapter 6 describes the use of diesel generator for wind dominated RAPS applications. The control strategy proposed for the diesel generator is explained. A control coordination methodology is proposed and established to coordinate the response of the components (i.e. wind turbine generator, diesel generator and battery storage) of the RAPS system and also to achieve acceptable voltage and frequency regulation throughout its operation. Simulation results are used to verify the suitability of the proposed RAPS systems.

Chapter 7 presents the importance of integrating hydrogen as a energy storage system for a wind dominated RAPS system. In this regard, the application of fuel cell, electrolyser and storage tank is elaborated as the key components of the proposed

RAPS systems. Mathematical modelling aspects in proceeding chapters are derived using existing component models where relevant. The simulated behaviour of the entire RAPS system is presented.

Finally, Chapter 8 summaries the major outcomes of the work presented in the thesis and makes recommendations and suggestions for future work. Chapter 2

Literature Review on Remote Area

Power Supply Systems

2.1 Introduction

This chapter provides an overview of various types of wind turbine generator based remote area power supply systems, covering the associated challenges when they operate in standalone environments, various system conﬁgurations with diﬀerent system modules and the relevant control strategies. Section 2.2 gives a brief background to

RAPS systems, followed by a review on associated concepts, challenges, current technological developments and trends. Section 2.3 covers the widely used wind turbine generator technologies including DFIG, PMSG and induction generator, giving sig- niﬁcant attention to the ﬁrst two types on which the thesis is primarily based on.

The application of diﬀerent types of energy storage systems for wind based RAPS systems is discussed in Section 2.4. In this regard, emphasis is placed on battery storage systems and supercapacitors which provide the basis for Chapter 4 and Chapter

5 respectively. Section 2.5 describes diﬀerent operating principles of diesel generating systems and their suitability for remote area power applications which are closely 12 13

related to the work presented in Chapter 6. The other key section of this chapter,

Section 2.6, describes the concepts, principles and applicability of a hydrogen based

generating system covering a fuel cell, an electrolyser and a hydrogen storage tank

for a wind based RAPS system forming the background for Chapter 7. The chapter

is summarised in Section 2.7.

2.2 An Overview of RAPS Systems

The selection of suitable energy sources to form a hybrid RAPS system depends

mainly on the availability of resources which are speciﬁc to a geographical location.

Considering wind as the main renewable energy source, an example of a typical layout

of a hybrid standalone power system is shown in Fig. 2.1.

Hydrogen based storage system DC bus ~ AC bus Inverter + _ Dump load Energy storage system Wind turbine generator Domestic load SG

Diesel engine Generator

Figure 2.1: A typical arrangement of a wind power based standalone power supply system. 14

It consists of a wind turbine generator representing the renewable energy segment of the generation mix, a diesel generating system to characterise a conventional generating scheme, an energy storage system and a dump load as indicating the ancillary components. In addition, a hydrogen based generating scheme is also integrated to the RAPS system, to emphasise the importance of hydrogen economy1 which is gain- ing recognition as a viable storage medium for wind power applications [1], [2] and [3].

In real life hybrid RAPS applications, usually the case is to select two or more components depicted in Fig. 2.1 with the wind turbine generator to form diﬀerent types of hybrid RAPS systems (e.g. wind-battery or wind-diesel). Fundamentally, the most critical tasks of such a hybrid remote area power supply system are the regulation of the voltage and frequency within acceptable levels2. In general, to achieve these levels, it is vital to maintain the active and reactive power balance of the RAPS system which can be described using (2.1)3 and (2.2) respectively.

dW 1 d(ΣJω2) ΣP − ΣP = ke = = 0 (2.1) sources sinks dt 2 dt

ΣQsources − ΣQsinks = 0 (2.2)

where, P is active power, Wke is kinetic energy of the system, J is combined moment of inertia of rotating machines (e.g. wind turbine generator, diesel generator),

ω is angular velocity of the rotating machine and Q is reactive power.

In addition to voltage and frequency regulation, as stated in Chapter 1, power

quality issues, coordination of the operation of the diﬀerent components, cost eﬀec-

tiveness and optimal operation are some of the other aspects which need considerable

1A common terminology used to represent a system which delivers energy using hydrogen. 2Acceptable limits deﬁned by the respective power quality standards e.g. IEEE 1574. 3This equation is only applicable for electrical machine (e.g. diesel generator, synchronous generator etc.) based power generating systems. 15 research attention [4], [5], [6], [7] and [8].

The level of renewable energy penetration in a hybrid RAPS system is an important parameter from design and economic perspectives. The quantiﬁcation of the renewable energy sources in a RAPS system can be expressed using average penetration level (APL) and instantaneous penetration level (IPL) given in (2.3) and (2.4) respectively [9]- [10].

P Ere AP L = P (2.3) EL P Pre IPL = P (2.4) PL

where, Ere is energy from renewable based power generation (kWh), EL is total energy delivered to the loads (kWh), Pre is power from renewable based energy sources

(kW) and PL is power delivered to the load (kW). The term, APL is used as an economic measure of a hybrid RAPS system as it determines the cost of energy from the hybrid system while indicating how much of the total generation comes from the renewable energy sources. In contrast, IPL is used as a technical measure which has an impact on the system layout. Hybrid RAPS systems with high IPL levels (e.g. 3) require sophisticated operating mechanisms since renewable energy sources dominate the system dynamics. However, the additional cost and degree of complexity associated with high IPL based RAPS systems can often be justiﬁed by the much greater fuel savings achieved by reducing the use of conventional generating mechanisms such as diesel or gas generators [11], [12], [13],

[14] and [15]. 16 2.3 Wind Energy Systems

2.3.1 Wind Speed Distribution

Wind speed at a given location varies randomly and hence it cannot be characterised

or predicted easily. In this regard, the Weibull distribution function given in (2.5) is

used as widely advocated probabilistic approach [16] to quantify the randomness of

wind speed, v. It is a function of two parameters: k, a shape factor, and c, a scale

factor where both these variables are functions of expected wind speed,v ¯ and Euler’s

gamma function, Γ given in (2.6) and (2.7) respectively. k v v f(v) = ( )k−1e−( )k (2.5) c c c c 1 v¯ = Γ (2.6) k k Z ∞ Γ(x) = t(x−1)e(−t)dt (2.7) 0

If k=2, the Weibull distribution function is known as Rayleigh distribution function and its scaling factor c is given in (2.8)

2 c = √ v¯ (2.8) π

As an example, the wind speed probability density function of the Rayleigh distribution of King Island4 which has an average wind speedv ¯, of 7 m/s is shown in

Fig. 2.2.

2.3.2 Wind Energy Conversion

Wind turbine generators exhibit greater uncertainty and variability in their power

output levels and are not easily dispatchable in the traditional sense. To accommodate

4A remote island of Australia located close to Tasmania. 17

0.12

0.1

0.08

0.06

Probability density 0.04

0.02

0 0 5 10 15 20 25 30 35 Wind speed (m/s)

Figure 2.2: Probability density of the Rayleigh distribution at King Island-Tasmania.

these uncertainties, advanced control methods should be employed [17]- [18].

Power generation using wind turbines consists of two conversion processes which

include the extraction of kinetic energy from wind which can be using (2.9).

1 P = ρAv3 (2.9) w 2

where, A is area swept by the rotor blades, v is wind speed, ρ is air density. Not all kinetic energy available from wind can be extracted by the wind turbine and hence power coeﬃcient Cp, deﬁned as in (2.10) which is a function of tip-speed ratio given by (2.11) and pitch angle, β is employed. During the second conversion process, the mechanical power/torque extracted by the wind turbine as given by

(2.12) is converted to electrical power using a suitable generator [19].

Pm Cp(λ, β) = (2.10) Pw ω R λ = r (2.11) v 1 P = C (λ, β)Aρv3 (2.12) m 2 p 18

where, Pm is mechanical power extracted from wind, Cp(λ, β) is power coeﬃcient of turbine, R is radius of blade, λ is tip-speed ratio, ω is rotational speed of rotor and

β is pitch angle.

Power extraction from wind is not feasible at all wind speeds which can be illustrated using Fig. 2.3. It shows the power extraction characteristics of a typical wind turbine subjected to diﬀerent wind speeds. Below cut-in wind speeds vcut−in, the wind turbine does not generate any power at all due to the limited energy content in the airﬂow. If the wind speed lies between vcut−in and rated wind speed vrated, the power output of the wind turbine is proportional to the cube of wind speed as indicated by

(2.12) where maximum rotor eﬃciency5 can be achieved. However, when the wind speed increases beyond the rated wind speed (i.e. v > vrated), the power output of the wind turbine cannot increase further and hence its aerodynamic eﬃciency is reduced by means of power control mechanisms. These power control mechanisms may include stall or pitch regulation where the applicability of such schemes is based on the wind turbine technologies which are discussed in Section 2.3.4. In the case where wind speed increases to levels above the cut-out speed vcut−out, the wind turbine needs to be shut down to avoid mechanical damage.

Employing the installed wind capacity as the basis, a classiﬁcation of standalone wind based power supply systems is given in Table 2.1 [9]. The work presented in this thesis is based on Type II and Type III RAPS systems which are based on PMSG and DFIG wind turbine generators respectively.

2.3.3 Maximum Power Extraction From Wind

The maximum power from wind can be extracted when a wind turbine is operated with an optimum power coeﬃcient, (Cp)opt. This can be achieved by operating the

5This is associated with maximum power extraction from wind which is explained in Section 2.3.3. 19

v v v cut-in rated cut-out 1.1 No generation Maximised rotor efficiency Rated power, reduced rotor efficiency No generation 1

0.9

0.8

0.7

0.6

0.5 Active power (pu)Active 0.4

0.3

0.2

0.1

0 0 2.5 5 7.5 10 12.5 15 17.5 20 22.5 25 27.5 30 Wind speed (m/s)

Figure 2.3: Typical power curve for a variable speed pitch controlled wind turbine.

Table 2.1: Types of stanadlone power supply systems Type Installed capacity (kW) Terminology Description I <1 Micro systems Single point DC system II 1-100 Village power systems Small power system III 100 -10000 Island power systems Isolated grid systems IV >1000 Large interconnected systems Large remote power system 20

turbine at a desired speed to obtain the optimal tip-speed ratio λopt [20]- [21]. The maximum power point tracking (MPPT) from wind without considering the asso-

ciated system losses (i.e. ideal condition) can be described using (2.13) where the

optimised constant kopt, and optimim tip-speed ratio are given by (2.14) and (2.15) respectively . A typical MPPT characteristic curve of a wind turbine is given in Fig.

2.4. For the sake of illustration of the MPPT principle, assume that wind turbine

generator operates on its maximum power curve, at point A corresponding to a wind

seed of v1. If the wind speed changes from v1 to v2, then the turbine changes its power output from A to B. The wind generator cannot respond to this wind speed

change quickly due to the inertia associated, thus retains the same electrical power

(i.e. power at point A). As a result, the mechanical power input from the turbine to

the generator is greater than its electrical power, causing the system to accelerate.

This acceleration would lead mechanical power to follow the path from B to C (B

→ C) while generator electrical power from A to C (A → C). Finally the system

becomes stable at point C.

3 (Pw)opt = kopt[(ωr)opt] (2.13)

1 R 3 kopt = (Cp)optρA( ) (2.14) 2 λopt (ω ) R λ = r opt (2.15) opt v

where, (Cp)opt is optimum power coeﬃcient of turbine, A is area swept by the rotor blades, v is wind speed, ρ is air density, R is radius of blade, λopt is optimum

tip-speed ratio, β is pitch angle, (Pw)opt is optimal wind power output. 21

Power (Pm)

MPPT characteristics

v3 Turbine characteristics B C

A v2

Wind turbine speed (ωr)

Figure 2.4: Wind turbine power characteristics with maximum power extraction.

2.3.4 An Overview of General Wind Turbine Concepts

In recent years, substantial scaling up of wind turbine generators has taken place taking into consideration both their size and the scale of projects [17]. Various types of wind turbine generators are currently available in the wind power market. The main differences which exist among these configurations depend on the generating system and the way how their aerodynamic efficiency of the rotor is limited during operation.

Applying speed control mechanism as the basis, wind turbine conﬁgurations can be classiﬁed into the following four major types which are illustrated in Fig. 2.5 [22]- [23].

• Type A : Fixed-speed wind turbine system with a squirrel cage induction gen-

erator (SCIG).

• Type B : Limited variable speed wind turbine with a wound rotor induction

generator (WRIG) consisting of variable resistors.

• Type C : Limited variable speed wind turbine with a WRIG consisting of a

partial frequency converter. 22

• Type D : Full variable speed wind turbine with a wound rotor synchronous

generator (WRSG), WRIG or PMSG consisting of a full frequency converter.

Variable resistance

Gear box SCIG Loads Gear box WRIG Loads

Capacitor bank Capacitor bank

(a) (b)

Gear box WRIG Loads Gear box Loads

Frequency converter PMSG/WRSG/WRIG Frequency converter

Figure 2.5: Typical conﬁgurations of wind turbine technologies: (a) Type A, (b) Type B, (c) Type C and (d) Type D.

The ‘Type A’ configuration represents the commonly known Danish concept with no-load compensated induction generator. In principle, the rotor speed of the machine is determined by the gearbox and the pole-pair number of the generator. This concept falls in the fixed speed wind turbine category where the speed variations are limited, which is approximately 1-2% and may or may not use the blade-angle control. A configuration, similar to the Danish concept, is given in ‘Type B’. However, in this case the variable resistors are controlled using an optically controlled converter which is mounted on the rotor shaft allowing variable speed operation. Typically, the allowable speed range of this type of configuration is 0-10% above the synchronous speed. The well known DFIG configuration is explained by ‘Type C’ category. The stator of the generator is directly connected to the load side and the rotor of the machine is connected to the load via a partial load-frequency converter allowing variable speed operation within ± 30% of its rated speed with pitch control capability. The ‘Type 23

D’ conﬁguration corresponds to a full-variable speed machine which connects to the load through a full-scale frequency converter.

Among all the above stated wind turbine technologies, doubly-fed induction and permanent magnet machines are identiﬁed as the more popular and widely used hence dominant wind turbine generator technologies in the current wind power industry [23].

While this is the case, the DFIG based wind turbine generator systems, (Type C), are preferred for high power applications whereas PMSG based wind turbine generator systems, (Type D), are suitable for lower or medium power levels. Fixed speed squirrel cage induction turbine generators6 are still in service and used for oﬀ-grid as well as on-grid applications.

2.3.5 Standalone Operation of Wind Turbine Generators

Variable wind speed with fluctuating load profiles make the operation of wind based power systems challenging, particularly when they operate in standalone mode. The random variations of wind speed lead to fluctuating torque on the wind turbine generator which eventually appear as voltage and frequency excursions in the RAPS system. In contrast to grid connected WEC systems, whose magnitude and frequency of the voltage are supported by the stiffness of the main grid supply system, standalone WEC systems need to be controlled to maintain the required voltage and frequency [15]. The way in which standlaone wind turbine generator contributes to regulate the voltage and frequency depends on the type of wind turbine generator technology and their adopted control strategies. For example, a doubly-fed induction generator needs complex control strategies as it employs a back-to-back converter arrangement compared to an induction generator.

As stated in Section 2.3.4, DFIG based wind power systems are considered as the

6This type is represented by Type A. 24 best option for large wind power applications. They offer numerous advantages over other types of wind generator systems including but not limited to smaller power electronic converter ratings (which is limited to about 20-30% of the total capacity of the wind generator system), maximum power extraction capability in variable wind speed conditions, reactive power capability, reduced mechanical stresses and ability to generate power in sub-synchronous and super-synchronous modes [16]. In contrast, permanent magnet synchronous generators can be utilised for small and medium wind power applications which provide the gearless operation, self excitation capability, improved thermal characteristic due to absence of field excitation losses, high efficiency and maximum power extraction capability [24].

In the existing literature, the standalone operation of DFIGs and PMSGs and their design, modelling and control aspects have gained a little research attention compared to their grid-connected counterparts. Some well established control approaches that have been adopted for DFIGs and PMSGs to operate them in standalone mode with their advantages and disadvantages are described below:

(a) DFIG Control Strategies for wind application

The control and modelling aspects associated with standalone operation of a DFIG are illustrated in [25]- [26]. An approach using vector ﬁeld oriented control in d-q domain7 to operate a DFIG in standalone mode has been investigated in [25]- [27].

The use of magnetisation current of a DFIG to control the load side voltage is reported in [25]. In the results presented, it is seen that frequency regulation is achieved by ensuring indirect stator ﬂux orientation. However, the experimental results presented corresponds to half the rated stator voltage which might be a safety step that has been taken while operating the machine in unsaturated region. Another control approach

7Refer to Appendix A for more information. 25 known as direct voltage control method applied for operation of DFIG in standalone mode is reported in [28]- [29]. This method is based on polar rotating reference frame which is used to control the magnitude of the voltage and phasor angle using the rotor currents. Also, this method eliminates the use of mechanical sensors or rotor position estimators as suggested in ﬁeld oriented control method discussed in [26].

However, the implementation of controllers associated with this method are noted to be complicated as they consist of complex algebraic loops with derivative terms that might cause stability issues in the control system. Further, the results are presented using a DFIG system consisting of a inverter-rectiﬁer8 arrangement.

(b)PMSG Control Strategies for wind application

Recent studies in relation to standlaone operation of PMSGs have been carried out predominantly using the ﬁeld oriented control approach as given in [24], [30]. How- ever, depending on power electronic arrangement used to interface the PMSG to loads, various control strategies have been developed to regulate the voltage and frequency.

In this regard, most commonly advocated power electronic arrangements that have been widely used are: (a) AC/DC controlled converter on the generator side and a

DC/AC controlled converter on the load side with a DC link between two converters [30] and (b) the generator side converter in (a) replaced with a diode rectiﬁer while keeping the same DC/AC controlled converter on the load side, and possibly incorporating an additional DC/DC converter in the DC bus [24]. The voltage and frequency control of a remote non-linear load connected to a similar arrangement given in (a) is discussed in [30]. It particularly investigates the improvement of power quality aspects with regard to harmonics and voltage unbalance. However, the presence of two inverters increase the cost of the installation which may not be justiﬁable for remote

8Inverter is connected to the rotor and rectiﬁer is connected to the load side, allowing only uni-directional power ﬂow. 26 power applications. An arrangement similar to (b) is demonstrated in [24]. Current control mode is adopted for the inverter to regulate the voltage and frequency of the

RAPS system. However, this mode of operation of the inverter is entirely suitable for grid connected operation of inverters and is not suitable for standalone mode. In the absence of other types of reactive power sources, voltage control mode is seen to be the best suited method for standalone operation of the inverters. Also, the results are illustrated for operation with resistive loads only.

2.4 Energy Storage Systems for Wind Power Applications

2.4.1 Importance of Energy Storage Systems in Standalone Wind Power

Applications

Due to the variable nature of wind (i.e. intermittency), a wind turbine generator alone cannot supply power to loads due to its inability to match the load demand. In this regard, a wind farm to be dispatchable similar to other conventional generation units such as a diesel generator, the generated power has to be regulated at a desired level. With rapid developments currently taking place on energy storage devices, their application in wind energy systems is seen to provide a promising opportunity to mitigate the issues associated with wind power ﬂuctuations as shown by Fig. 2.6.

An energy storage system can be categorised in terms of its role in a RAPS power system: either for energy management or for power quality enhancement [31]. These two objectives can be further explained using the following sub-objectives.

• to improve the eﬃciency of the entire system,

• to reduce the primary fuel (e.g. diesel) usage, 27

Pw power PL

(a) Time

Charging

power Time Discharging

(b)

Figure 2.6: Application of an energy storage in a WEC system (a) wind power output, Pw, and load demand, PL, and (b) energy storage charging/discharging status.

• to function as an alternative source or a buﬀer in the absence of other types of

components (e.g. diesel generators, dump loads etc.) and

• to provide better security and enhanced power quality of energy supply .

2.4.2 Types and Sizing of Energy Storage Systems

An ideal energy storage system in a standalone WECS should be able to provide both high energy and power capacities to handle situations such as wind gusts and load step changes which may exist for seconds or minutes or even longer. At present, various types of storage technologies are available to fulﬁl either power or energy requirements of a RAPS system. Widely advocated energy storage technologies that currently employ in wind farms are: batteries, supercapacitors, ﬂywheels, compressed air energy storage, hydro pumped storage, superconducting magnetic energy storages

(SMES), fuel cells etc. [32]- [33]. The power and energy density ranges associated with these energy storage systems are shown in Fig. 2.7 [34]. Energy storage systems 28 with high energy density levels are usually termed as “long term storages” as they are able to operate over a long period of time (e.g. minutes to hours). Similarly, energy storage systems with high power density are termed as “short term storages” as they are capable of handling the transients which occur over a short periods of time (e.g. seconds to minutes).

Figure 2.7: Speciﬁc energy versus speciﬁc power ranges for various energy storage systems.

Among all energy storage systems depicted in Fig. 2.7, fuel cells are seen to have the highest energy density while the supercapacitors seem to have the highest power density. In principle, the energy storage requirement in a RAPS system could be provided either by a fuel cell or a battery system due to their inherent high power and energy capacities over other types of storage systems [35]. However, the economic viability of fuel cell systems is questionable and hence at present, battery storage systems are widely employed in most real life RAPS applications. To further improve the performance of these two types of energy storage systems, a supercapacitor can be incorporated to perform a hybrid operation [36]- [37]. In this way, the combined energy storage system is able to satisfy both power and energy requirements of the 29

RAPS system.

The size estimation of energy storage system for a given wind application is extremely important and site9 speciﬁc. In addition, it also relies on many factors such as total wind turbine inertia, low voltage ride through requirement (LVRT)10 given by

(2.18), short circuit ratio and ﬁnancial returns [38]. There exists substantial research outcomes in relation to optimising an energy storage system for RAPS systems taking the economic aspects as the basis. Nevertheless, only limited work exists in relation to the sizing of energy storage systems considering the technical constraints11. In [39], it is stated that the ratio given in (2.16) is an important design parameter that can be used to characterise the capacity of an energy storage system needed for a small wind/battery hybrid power system. According to a survey based on 25 wind energy systems ranging from 400 W to 50 kW, the ratio given by (2.16) varies between 3 and 10 in majority of the systems. The sizing of a supercapacitor in a standalone power supply system consisting of a wind turbine generator, a fuel cell, an electrolyser and a supercapacitor is explained in [40]. It estimates the capacitance needed for the supercapacitor storage considering the worst case scenario where it is able to support the maximum demand-generation mismatch for a predetermined period of time as given by (2.17). The LVRT capability is used as the basis to estimate the capacitance value of the supercapacitor [38]. Further, a size optimising method of a battery storage which can be used to achieve both technical feasibility and economic proﬁtability are explained in [41] - [42].

9Refers to a geographic place where the wind farm is physically located. 10The ability of a wind turbine generator to remain connected to a main grid supply in the event of a fault. 11This may include allowable voltage and frequency limits, DC bus voltage variations etc. 30

Battery capacity (Ah) A = (2.16) bat Rated current of wind turbine generator 2Esc C = 2 2 (2.17) ((vsc)max) − ((vsc)min) 2ELV RT C = 2 (2.18) ((vdcsc)ref )

where, Esc is energy rating of the supercapacitor, (vsc)max,(vsc)max is maximum and minimum operating voltages of supercapacitor respectively and C is capacitance,

ELV RT is area in the no trip region of the LVRT capability curve and (vdcsc)ref is allowable maximum voltage across the supercapacitor.

2.4.3 Energy Storage for Wind based Remote Area Power Supply Sys-

tem

An energy storage system can be connected to the DC bus or AC side of a wind based RAPS system12. Depending on the connection topology used in the wind energy system, an energy storage system can be designed to provide active and/or reactive power into the RAPS system [43], [44] and [45]. If an energy storage system is connected to the DC bus of the wind energy generator system, it is only able to provide active power support [43]. In contrast, both active and reactive power support can be provided if the energy storage is interfaced via an inverter system to the load side [44]- [45].

Recent studies [43], [46] and [47] in relation to the modelling and control aspects of an energy storage system for a grid connected wind application have received substantial attention. However, such work related to standalone wind-energy storage systems has received only little research attention. The applicability of battery storage

12Refer to wind turbine technologies depicted in Fig. 2.5. 31

as the energy storage for a grid connected variable wind generator is described in [46]-

[47]. As reported in this literature, the core objective behind the use of the energy

storage system (i.e. battery banks) is to mitigate the eﬀect of wind speed ﬂuctuations

and thereby to ensure smooth power output from the wind turbine generator. A

hybrid energy storage system including both battery system and supercapacitor for

a grid connected variable wind turbine generator is given in [36].

In the absence of other generating (e.g. diesel generators) resources, an energy

storage system can be fully utilised for the purpose of load levlling13. In such situations, a long term energy storage system should be selected for the RAPS system as the ﬁrst choice to replenish the demand-generation mismatch. The performance of an energy storage system in a standalone PMSG based WECS is given in [48]- [49]. A

RAPS system consisting a PMSG based wind turbine generator together with a battery storage system and a solar photovoltaic system where the latter two components are connected to the DC bus of the wind energy system is explained in [48]. One of the main drawbacks in the modelling exercise undertaken is the use a of simplistic battery storage system that is directly connected to the DC bus of the wind turbine generator without integrating any power electronic arrangements. The voltage and frequency stabilisation of a battery assisted standalone PMSG based wind turbine generating system is proposed in [50]. However, in the absence of other auxiliary components such as a dump load, the above system requires a sizeable battery storage which is not economically viable. The autonomous operation of a DFIG with energy storage systems are discussed in [49]- [51]. Similar to the case noted in [48], the battery storage system is directly connected to the DC bus of the DFIG system described in [51]. With such an arrangement, the required number of batteries that need to be connected in series to match the DC bus voltage is large and cannot be jus-

13Method that can be used for minimising the demand-generation mismatch. 32

tiﬁed in real life applications. The suitability of a supercapacitor for a remote DFIG

RAPS system is discussed in [49]. In addition to a back-to-back converter system,

two additional DC/DC converters are placed in the DC bus where the supercapaci-

tor is placed in between them. Such an arrangement, however, increases the cost of

the entire system and generates unnecessary switching harmonics. In addition, the

application of supercapacitors for a standalone power system is not eﬃcient due to

the limited energy densities14.

The application of a hybrid energy storage system for grid connected and standalone wind energy application is given in [36] and [52]. This literature particulary focuses on the performance of the hybrid energy storage system rather considering the system level investigations. It is identiﬁed that, the application of hybrid energy storage in relation to a DFIG has not received much research attention.

2.5 Diesel Generators for Standalone Wind Power Applica-

tions

2.5.1 Importance of Diesel Generator Systems in Standalone Wind

Power Applications

Fixed speed diesel generators are still predominantly used in most remote areas for supplying power to regional communities including islands, due to their lower installation cost, reliability and simplicity of operation [9]. However, with the increased attention given to environmental issues, diesel fuel transportation problem, poor eﬃ- ciency at low load factor operation, high operating cost and penetration of renewable energy technologies into the energy industry have made diesel based power generation a less favourable method [53]. Further, standalone power supply systems which con-

14Refer to Fig. 2.7. 33 sist of wind-energy storage systems discussed in Section 2.4 are usually characterised by inherently low inertia, low X/R ratios and poor reactive power capability [54] and [55]. Moreover, due to the inherent capacity limitations associated with energy storage systems, they would not be able to supply the necessary energy to the loads over a long period of time. In addition, such generating schemes are susceptible to any changes in the operating conditions15 which will eventually have an impact on the stability of the RAPS system [39] and [56]. To alleviate above stated issues associated with the two generating schemes, namely: (a) ﬁxed speed diesel generator and

(b) wind-energy storage based hybrid RAPS16, a feasible scheme would be a hybrid system consisting of a wind turbine generator, a diesel generator and possibly with an energy storage system [9], [56] and [57]

2.5.2 Operating Principles of Diesel Generating System

Traditionally, a fixed speed engine employs a conventional synchronous generator to supply power to a load as shown in Fig. 2.8-(a). A diesel engine (DE) refers to an internal combustion engine whose operation is based on diesel cycle17 exhibits non- linear operational characteristics [58]. A diesel engine uses compression ignition to burn the fuel which is injected into the combustion chamber during the final stage of compression. The fuel injection is controlled by the governor which is responsible for controlling the frequency of the generated voltage, whereas the excitation system is used to control the magnitude of terminal voltage [58]. However, the efficiency of a diesel generator is considerably low at light load factor18 operation. Hence, its operation below the minimum loading condition should be avoided as per instructions

15e.g. intermittency of wind, variable load proﬁles etc. 16This is discussed in Section 2.4. 17It is the thermodynamic cycle which approximates the pressure and volume of the combustion chamber of the diesel engine. 18It is the average power divided by the rated power, over a period of time. 34 provided by manufacturers [53] and [59]. At light loads, the fuel economy of ﬁxed speed synchronous generators is poor due to incomplete fuel combustion. One of the possibilities is to combine a dump load to ensure that the diesel generator meets its minimal loading condition. However, dissipating energy through a dump load may lower the plant economy [60]. To alleviate the issues associated with low load factor operation, an improved operating mechanism that can be employed for the diesel generating system shown in Fig. 2.8-(b). In this scheme, the synchronous machine is coupled to the diesel engine to operate as a generator at its higher load factor.

Otherwise, the synchronous machine is disconnected from the diesel engine and made to operate as a synchronous condenser which can be used only to satisfy the reactive power requirement of the loads. Such an arrangement also avoids the necessity for synchronising the diesel generating system with the existing networks [61]. Another solution to overcome the barriers associated with the operation of a diesel generator under low load conditions is to operate in variable speed mode as explained in [62]-

[63]. The variable speed diesel generator is connected via a power electronic interface as shown in Fig. 2.8-(c), which decouples the frequency of the generator and loads.

More importantly, such an arrangement allows the generator to operate at variable speed using the inverter control that drives engine to track its optimum operating curve19. In this regard, ﬁxed speed ﬁeld excited synchronous generator should be replaced with PMSGs or DFIGs due to their ability to operate at variable speeds.

One of the major drawbacks of the variable speed operation of a diesel engine is that its inability to respond to system transients promptly due to lack of inertia and its associated power margin, thus requiring the integration of an energy storage system [53] and [64]. Moreover, the operation of variable speed generators need complex control strategies to utilise them together with renewable energy systems

19For a given output power there is an optimal speed where the fuel consumption is minimum [62]. 35 due to the presence of parallel connected and islanded inverters which is seen to be a challenging task [65].

DE SG Loads DE SG Loads

Clutch system (a) (b)

PMSG/DFIG ~ DE Loads

Frequency converter

(c)

Figure 2.8: Different configuration of diesel generating systems: (a) fixed speed operation, (b) fixed speed with dual mode operation and (c) variable speed operation.

2.5.3 Operational Aspects of Wind-Diesel Remote Area Power Supply

Systems

The operation of wind-diesel systems can be categorised into three types depending on the penetration level of wind power output, namely: (a) low penetrated, (b) medium penetrated, and (c) high penetrated systems. While the ﬁrst and second types require the diesel generator to be in operation at all times along with wind turbine generator, in the third type, the continuous operation of the diesel generator may not be required to allow maximum deployment of the wind energy while minimising the operating costs. The requirement for integrating an energy storage system into a wind-diesel system depends on wind penetration level [66]. Following this criterion, an energy storage system may not be required for low or medium penetration wind-diesel RAPS systems, as the diesel generator dominates the system dynamics as in the case of a conventional generator. In such a situation, the intermittency associated with wind 36 power generation together with variability of load demand have minimal impact on the overall operation when compared to a high penetrated wind-diesel system. Indeed, high penetrated wind-diesel systems require robust sophisticated controllers and other types of auxiliary components20 to ensure stable operation. However, the additional cost and complexities involved with high penetrated wind-diesel system can often be justiﬁed by the much greater fuel savings and reduced diesel generator operating periods [67].

The operation of high penetrated wind-diesel systems can be classiﬁed into the three modes of operation: (a) wind-only (WO), (b) wind-diesel (WD) and (c) diesel

Only (DO). In the WO mode, the wind turbine generator is able to supply the required power along with auxiliary system components to satisfy the load demand. In the WD mode, the power output from the wind turbine generator is less than the load demand and hence the diesel generator is used to supply the demand-generation deﬁcit. In the DO mode, only the diesel generator operates, possibly with an energy storage system to meet the load demand in situations where the wind turbine generator does not generate power under abnormal limitations (e.g. when wind speed goes above cut-out speed or below cut-in-speed).

Recent research outcomes in relation to hybrid operation of wind-diesel RAPS system are given in [64], [68], [69] and [70]. Among all these reported research outcomes, a high priority is given to investigate the operation of induction generator based wind-diesel systems. However, only limited work exists in relation to PMSG/DFIG based wind-diesel systems. Some of this work includes, a RAPS system consisting of a PMSG as a wind turbine generator with a diesel generator system explained in [68].

Nevertheless, the proposed synchronistation mechanism of the diesel generator system involves the measurement of the system impedance including loads which is neither

20This may include energy storages, dump load etc. 37 a robust mechanism nor predictable. The performance analysis of a PMSG-diesel

RAPS system together with hybrid energy storage is discussed in [69]. It focuses on optimising the system performance through a rule-based energy management algorithm. However, rule based energy management is discrete in nature and also it may not able to cover all the operational situations of the RAPS system. To examine the power sharing among a wind turbine generator, a battery storage and a diesel generator using a linearisation model based approach is explained in [70]. In this study, every system component is represented by a ﬁrst order transfer function. However, such an approach does not allow to examine the precise system dynamics due to lack of detailed representation of power electronic interfaces and their corresponding control strategies.

2.6 Hydrogen Based Storage Systems for Wind Power Ap-

plications

2.6.1 Operating Principles of a Fuel Cell System, Electrolyser and

Storage Tank

Hydrogen has become a potential energy carrier and a storage medium for renewable energy systems [1] and [71]. A hydrogen-based standalone power system requires some form of hydrogen production, storage and utilisation, often combined with some short- term energy storage. An electrolyser is used to generate hydrogen by means of an electrochemical reaction. The hydrogen generated by the electrolyser can be utilised by a fuel cell which converts it to electrical energy where any excess hydrogen can be stored in liquid or gaseous form [72], [73] and [74] . 38

(a) Fuel Cell Systems

The integration of a fuel cell into a RAPS system oﬀers many advantages, including high energy density, virtually no carbon emission, absence of moving parts etc. compared to other alternatives such as diesel engines or energy storage systems [75]-

[76]. The operational behaviour of a fuel cell is a complex non-linear process which needs precise modelling to characterise its thermal and electrical aspects. However, for the purpose of illustration, the operation of a fuel cell can be described using four stages: (1) reactant delivery, (2) electrochemical reaction, (3) ionic conduction and

(4) product removal [77]. The first stage includes the supplement of fuel and oxidant which mainly depends on the mechanical structure of the fuel cell such as shape, size and pattern of flow. Once the reactants are supplied to the electrodes, the chemical reactions take place in the second stage. The ions and electrons (i.e. electricity) produced by the second stage needs to be transferred from the location where they are generated to the locations where they are consumed, the process described by the third stage above. The final stage encompasses the procedures associated with the removal of byproducts after generating electricity. For example, a hydrogen-oxygen fuel cell generates water as a by-product which has to be removed to improve the reactant delivery. Depending on the type of electrolyte21 used, fuel cells can be mainly classi-

ﬁed into ﬁve types: namely, proton exchange membrane fuel cell (PEMFC), alkaline

fuel cell (AFC), molten carbonate fuel cell (MCFC), solid oxide fuel cell (SOFC) and

phosphoric fuel cell (PFC) of which some of the characteristics are listed in Table

2.2 [78].

Several research publications have covered the modelling aspects of fuel cell sys-

tems considering either their electrical or thermal characteristics or a combination

of the two [75]- [76]. Some of the reported work includes empirical equations which

21The medium which only permits to pass the appropriate ions. 39

Table 2.2: Different types of fuel cell systems fuel cell type charge carrier operating temperature (C) fuel efficiency (%) + PEMFC H 80 Pure H2 35 + AFC OH 120-150 Pure H2 40-60 −2 MCFC CO3 650 H2,CO, CH4 50-60 −2 SOFC O 800-1000 H2,CO, CH4 50-65 + PFAC H 200 H2,CO 40-60 are based on experimental data obtained from commercially available fuel cell systems [79]. However, to examine the specific behaviour of a fuel cell operation in an electrical application, it is vital to use an appropriate equivalent circuit of a fuel cell which reflects its V-I characteristic or the commonly known polarisation curve as shown in Fig. 2.922 which depicts the three operating regions of a fuel cell: namely; activation, ohmic and concentration. Since a unit cell has a low voltage as shown in

Fig. 2.9, a fuel cell stack can be built by connecting a number of cells in series.

1.2 Activarion voltage Ohmic voltage Concentration voltage 1.1

0.9

0.8

0.7 Voltage (V) 0.6

0.5

0.4

0.3

0.2 0 100 200 300 400 500 600 700 800 900 1000 Current Density (mA/cm2)

Figure 2.9: Polarisation curve of a fuel cell.

22This represents the V-I response of BALLARD MK5-E fuel cell operating at ﬁxed 40 0C which is reproduced according to the data provided in [74].

(b) Electrolyser and Storage System

An electrolyser is an electrochemical device which produces hydrogen and oxygen by decomposing water by passing a DC current passed between two electrodes separated by an electrolyte having good ionic conductivity. The characteristics of an electrolyser is broadly conceived as the reverse process of a hydrogen fuel cell system and its associated electrochemical reaction with water electrolysis as explained by

(2.19) [74] and [80].

− 1 − ) Anode : 2OH (aq) → O2(g) + H2O + 2e 2 (2.19) − Cathode : 2H2O(l) + 2e → H2(g)

Similar to the case of a fuel cell, the modelling aspects of an electolyser consists of

diﬀerent components: electrochemical, electrical, thermal and hydraulic. Depending

on the scope of interest, highly theoretical based [80] to more application oriented

modelling aspects [81] can be incorporated to illustrate the behaviour of an electrol-

yser. In this regard, to examine the performance of an electrolyser in relation to an

electrical application, usually the case is to exclude the thermal aspects and operate it

at ﬁxed temperature [82]. Similar to a fuel cell system, the electrical characteristics of

an electrolyser can be quantiﬁed by means of its V-I characteristics (i.e. polarisation

curve) as shown in Fig. 2.1023.

The ability of hydrogen-based storage systems to stockpile the fuel received from

an electrolyser, and for back-up provision to a fuel cell, provides an added advantage

to the user. The means of storing the hydrogen fuel varies according to the nature

of the application, although some types of storage schemes are more popular than

others. Again, modelling aspects of a hydrogen storage system may include complex

molar ﬂow dynamics associated with hydrogen fuels. However, to examine the storage

23This represents the V-I response of Stuart electrolyser an aggregated model operating at ﬁxed 53.51 0C which is reproduced according to the data provided in [79]. 41

36 Voltage (V) 34

26 0 10 20 30 40 50 60 70 80 90 100 Current (A)

Figure 2.10: Polarisation curve of an electrolyser. characteristics from an electrical point of view, without considering its compression dynamics, simplistic models can be employed as given in [83]. Further, information on the modelling aspects of a hydrogen storage tank is given in Chapter 7.

2.6.2 Operational Aspects of Wind-Fuel Cell based Remote Area Power

Supply Systems

The autonomy of operation permitted by the wind based RAPS systems discussed in

Section 2.4 and Section 2.5 is mainly limited by the dependency on energy storage systems and diesel generator systems respectively. Moreover, the limited capacity of the energy storage systems and uncertainties associated with fuel availability of diesel generators lower the autonomy of RAPS system operation. Incorporating a hydrogen based power generation scheme into such a RAPS system improves its performance and enhances the autonomy of operation. In a situation where a hydrogen based storage system is in operation with a renewable energy based standalone power supply scheme, the electrolyser can be used to generate hydrogen by utilising the excess 42 energy available during over-generation situations. The generated hydrogen is stored in a tank and is used by a fuel cell system to generate power during under-generation situations. However, the electrolysers and fuel cells are DC-current devices and hence an appropriate power electronic arrangement should be employed to interface such systems with a wind energy generating system.

There exist significant number of research publications explaining the grid-connected mode of operation of hydrogen storage based generating systems. In contrast, the application of hyrdrogen storage based generating systems for RAPS system has received a little research attention [73], [84], [85] and [86]. Further, the suitability of hydrogen storage system in relation to PMSG and DFIG based wind power systems has not received much attention compared to other types of wind generating systems. An approach based on energy management of a solar PV, a PMSG and a hydrogen storage system is explained with experimental validation in [73]. However, it does not cover the associated control strategies related to each system component. The standalone operation of PMSG based hydrogen wind energy system is explained in [85]. It uses the battery storage to maintain the DC bus voltage where the fuel cell system enables its operation when the battery storage system is out of its operation. Furthermore, the electrolyser system is used as a dump load of the system where the details of the storage tank are not presented. Also, the paper does not present the control strategies that applied for the PMSG inverter and the coordination between the electrolyser, fuel cell and hydrogen storage tank is not illustrated. The hybrid operation of a DFIG based wind turbine generator with hydrogen generating system connected to a weak grid system is demonstrated in [86]. The control strategy adopted for the DFIG is used to extract the maximum power from wind and not to support the voltage and frequency. The simulated behaviour of the AC voltage at load end is seen to be vulnerable to load and wind speed changes. Also, the behaviuor of the 43 supercapacitor is not included. The active power flow of a RAPS system consisting of a wind turbine generator, fuel cell, electrolyser and a storage tank is demonstrated using linearised system components in [84]. However, such limited models are not sufficient to investigate the precise system dynamics as every system component is represented by a first order transfer function. The concept of wind-diesel-hydrogen generating system is explained in [87]. However, the scope of the work is limited only to the component level modelling and to investigate the specific component behaviour of the RAPS system. 44 2.7 Chapter Summary

This chapter has provided general information in relation to the operational aspects of wind based remote area power supply systems on which this thesis is preliminary based on.

Sections 2.4-2.6 emphasised the importance of integrating energy storage systems, diesel generating systems and hydrogen storage systems for wind energy applications. In addition, the individual performance, existing technologies/developements and characteristics of each component, i.e. energy storage system, diesel generator and hydrogen storage system were explained. Further, existing knowledge and theory presented in current literature in relation to the modelling aspects of system components, control approaches and their relative merits and demerits were examined establishing the background for the proceeding chapters. Based on the literature review, it can be concluded that the wind dominated RAPS systems have received least research attention compared to their grid connected counterparts. Further, the detailed investigation of such RAPS system in relation their power management, control strategies, system response (e.g. voltage and frequency proﬁles) and component modelling have not received much research attention which forms the basis for the following chapters in this thesis. Chapter 3

Wind Turbine Generator

Technologies for RAPS

Applications

3.1 Introduction

As stated in Chapter 2, the operation of wind turbine generators have been mostly studied in relation to grid connected applications compared to their standalone counterparts. Furthermore, the modelling aspects of the variable speed generators, namely:

DFIG and PMSG for their standalone operation have received little research attention. Although these two generator technologies are seen to be similar due to the presence of power electronic converters, in principle there exists several diﬀerences in relation to their functional aspects and control features. Before investigating the behaviour of these two types of generator systems in diﬀerent RAPS environments, it is vital to understand the respective fundamental concepts, operating principles, terminologies and control approaches.

The operating principles, conﬁgurations of wind generators and their proposed 45 46

control strategies to operate them in standalone mode are explained in Section 3.2

and Section 3.3 respectively. The importance of integrating a dump load and its

modelling aspects in relation to a wind turbine generator application as an auxiliary

component is explained in Section 3.4.

The suitability of the control strategies that are applied for wind turbine generat-

ing systems and dump loads are investigated under variable wind and load conditions

and the relevant simulation results are presented in Section 3.5.

3.2 Doubly-Fed Induction Generator Modelling, Operation

and Control

Operating principle, mathematical model, control aspects of DFIG based wind turbine

generator system are discussed in the following sub-sections.

3.2.1 Overview of Operating Principle of the DFIG

A typical conﬁguration of a DFIG based wind turbine generator system is shown

in Fig. 3.1, the operation of which can be categorised into two modes: (a) super-

synchronous and (b) sub-synchronous. The diﬀerence between operation of these two

modes can be determined from the rotor speed ωr, compared to the synchronous

speed ωs, and direction of power ﬂowing through the back-to-back converter. In the super-synchronous mode, the rotor speed of the DFIG is kept above synchronous speed leading to a negative slip s < 0, as evident from (3.1). During the super-synchronous mode, the generated wind power passes to the load through the stator, as well as through the rotor, of the DFIG which is given by (3.2) and (3.3) respectively (i.e. Pr > 0). In contrast, during the sub-synchronous mode of operation, the rotor speed is kept below the synchronous speed. The generated wind power 47

T ,ω m r DFIG Ps , Qs PL, QL

To isolated loads

Pr , Qr AC DC AC

PLSC , QLSC

RSC LSC

Figure 3.1: Typical conﬁguration of DFIG. is supplied to the load by the stator while slip power is absorbed through the rotor

(i.e. Pr < 0). The total mechanical power input to the DFIG from the wind turbine generator is given by (3.4) [88]- [89].

ω − ω s = s r (3.1) ωs P P = m (3.2) s (1 − s)

Pr = − sPs (3.3)

Pm = Ps + Pr (3.4)

The steady state equivalent circuit of the DFIG indicating mechanical power Pm, stator power Ps, and rotor power Pr is shown in Fig. 3.2. It can be seen that, the active power dissipation in the ﬁctitious resistance Rr(1/s − 1), and virtual voltage source vr(1/s − 1), shown within the shaded box in Fig. 3.2 [90] represents the mechanical power Pm which is the input power to DFIG supplied by the wind turbine.

In Fig. 3.2, Rs, Rr represent stator and rotor resistances respectively, is, ir are stator and rotor currents respectively, Lsσ, Lrσ are leakage inductances of stator 48

v (1/s-1) Rr(1/s-1) r

L R Rs Lsσ rσ r ir is v vs Lm r

Ps Pr

Figure 3.2: Steady-state modiﬁed equivalent circuit of the DFIG.

and rotor respectively, Lm is magnetising inductance and vs, vr are stator and rotor voltages respectively.

Operation of the DFIG is mainly determined by control aspects associated with back-to-back converter system, namely: rotor side converter (RSC) and line side converter (LSC) as shown in Fig. 3.1. Based on the speciﬁc functions expected from the RSC and LSC, diﬀerent control strategies should be implemented. As an example, during the grid-connected operation, the RSC can be used for torque/speed control, together with terminal voltage or power factor control. Contrarily, in standalone operation the RSC can only be used to control the load side voltage and frequency in the absence of other types of generating sources (e.g. diesel generators).

3.2.2 Mathematical Model of the DFIG

The space vector equivalent circuit of the DFIG shown in Fig. 3.3 is used to derive the control laws for the RSC. The voltage and current vectors are explained in d-q reference frame1 as follows: 1This is commonly known as park transformation. For more information refer to Appendix A. 49

L R R /s Rs Lsσ rσ r r

v vr/s s Lm

ω ω ω Φ j eΦs j( e-)r r is ir

Figure 3.3: Space vector equivalent circuit for arbitrary reference frame.

Noting vs, vr are stator and rotor voltages, is, ir are stator and rotor currents and

φs, φr are stator and rotor ﬂux components respectively.

vs = vds + jvqs, vr = vdr + jvqr, is = ids + jiqs, ir = idr + jiqr, φs = φds + jφqs and

φr = φds + jφqs. The d-axis and q-axis components of the DFIG stator and rotor voltages are given as in (3.5)-(3.6) and (3.7)-(3.8) respectively [88].

dφ v = R i + ds − ω φ (3.5) ds s ds dt e qs dφ v = R i + qs + ω φ (3.6) qs s qs dt e ds φ v = R i + dr − (ω − ω )φ (3.7) dr r dr dt e r qr φ v = R i + qr − (ω − ω )φ (3.8) qr r qr dt e r dr

where, ωe is angular frequency of arbitrary rotating reference frame, ωr is rotor frequency, Rs is stator resistance, ids, iqs are stator d-axis and q-axis currents respectively, φqs are stator d-axis and q-axis ﬂux respectively, idr, iqr are rotor d-axis and q-axis currents respectively and φrd, φrq are rotor d-axis and q-axis ﬂuxes respectively. 50

Further, the ﬂux components of the stator and rotor of the DFIG in d-q reference frame can be given by (3.9)-(3.10) and (3.11)-(3.12) respectively [88].

φds = Lsids + Lmidr (3.9)

φqs = Lsiqs + Lmiqr (3.10)

φdr = Lridr + Lmids (3.11)

φqr = Lriqr + Lmiqs (3.12)

2 where, Ls, Lr are stator and rotor inductances respectively. Moreover, the active and reactive power output of the DFIG in d-q reference frame can be explained using (3.13) and (3.14) respectively [25].

3 P = (v i + v i ) (3.13) s 2 ds ds qs qs 3 Q = (−v i + v i ) (3.14) s 2 qs ds ds qs

However, the cross-coupling3 that exists in active and reactive power ﬂow can adversely inﬂuence the dynamic performance of the controllers associated with the

RSC. Therefore, it is necessary to decouple these variables in order to provide ﬂexible and robust control. In this regard, an appropriate ﬁeld oriented4 vector scheme should

be employed with regard to the back-to-back converter system.

2 Ls = Lm + Lls and Lr = Lm + Llr where, Lls and Llr are leakage inductance of stator and rotor of the DFIG and Lm is magnetising inductance of the DFIG respectively. 3This refers to a representation of any measurement as a combination of d-axis and q-axis variables. 4Usually, this term is reserved for controllers which maintain a π/2 spatial orientation between critical ﬁeld components. 51 3.2.3 Mathematical Model of the Back-to-Back Converter System

The RSC and LSC are modelled as current controlled voltage source inverters in

which the ﬁeld oriented vector control schemes are employed to develop the respective

control schemes. The control objectives that are related to RSC and LSC can be listed

as follows:

• RSC: voltage and frequency control on the stator side

• LSC: DC bus voltage control of the back-to-back converter system and to pro-

vide any reactive power if necessary for loads.

For the sake of simplicity, the illustration of above stated objectives can be explained using the steady-state operating conditions of the machine. Firstly, the voltage and frequency regulations at the stator of the DFIG are achieved by controlling the air-gap ﬂux of the machine at the rated value. From (3.15)5, it is evident that

the rated ﬂux φrated, of the machine ensures the rated voltage (vs)rated, and rated

frequency (f)rated, on the stator of the DFIG.

(vs)rated φrated = (3.15) (f)rated

Secondly, the DC bus voltage regulation is achieved by the LSC maintaining the

power balance on the DC bus as described by (3.16). However, due to the decoupled

6 operation of the back-to-back converter system, the power from the RSC, PRSC , is regarded as one of the disturbances for the purpose of controlling7 the DC link voltage

which can be explained using (3.16).

5 This relationship can be easily derived from, E = 4.44kwNphfφm. E is air gap voltage, φm is magnetising ﬂux, Nph is number of turns per phase and f is operating frequency. 6The DC link capacitor of the back-to-back converter acts as a temporary energy storage and decouples the control of the RSC from the control of the LSC. 7Refer to Appendix A which describes the PI controller tuning process associated with the DC link voltage using Internal Model Control (IMC) principle [91]. 52

dv P P − P dc = = RSC LSC (3.16) dt Cvdc Cvdc

where, P is the net power ﬂow into the capacitor, PRSC is power from the rotor side

converter, PLSC is power from the line side converter, C is the DC link capacitance and vdc is the DC bus voltage.

(a) Rotor Side Converter (RSC) Model

The selection of the RSC for voltage control purpose is mainly due its ability to inject the reactive power through rotor circuit which scales the reactive power by the

1 8 factor of a slip given by s . Compared to other types of ﬁeld orientated techniques, stator ﬂux orientation (SFO)9 scheme is generally preferred for the RSC in order to

provide ﬂexible decoupled control of active and reactive power. However, due to the

absence of a main grid supply, the corresponding orientation angle of SFO cannot

be determined using stator voltage10. Therefore, in standalone operation the stator

ﬂux orientation is achieved indirectly by setting the q-component of the stator ﬂux

φqs, given in (3.10) to zero. Further, it should be noted from (3.5) that if the stator resistance Rs, is neglected, the voltage reference frame is also principally identical to the stator ﬂux11 reference frame. This relationship can be vectorially represented as shown in Fig. 3.4.

8e.g. Rotor flux orientation and stator voltage orientation: First scheme is sensitive to machine parameters such as Lm while the second scheme cannot be directly applied due to the absence of a stiff grid. 9 In the absence of stator resistance Rs, SFO is also similar to the air-gap flux orientation. 10 R Using Clark transformation, the flux quantities in α − β domain can be given as φαs = (vαs − R Rsiαs)dt and φβs = (vβs − Rsiβs)dt. The corresponding orientation angle for stator flux oriented mode can be estimated as tan−1( φβs ). φαs 11 This can be achieved by substituting φqs = 0, owing to SFO and considering the steady state dφds operation dt = 0 in (3.5). 53

r o s t t o q a axis - to r a q- r q x i f - s l a u x x i s ux ωe stator fl tor sta s axi d- vqs

ϕds -axis ωr rotor d ϑs

ϑr stator -axis d

Figure 3.4: Stator ﬂux oriented vector representation.

When the DFIG is controlled using the stator ﬂux oriented mode, the respective

rotor voltages given earlier in (3.7)-(3.8) can be further simplifed12 into (3.17)-(3.18)

which forms the basis for developing the controllers of RSC.

∗ vdr = vdr − σLriqr(ω − ωr) (3.17) 2 ∗ Lmims vqr = vqr + (ω − ωr)[idrσLr + ] (3.18) Ls

where,

L2 σ = (1 − m ) (3.19) LsLr i v∗ = R i + σL dr (3.20) dr r dr r dt di v∗ = R i + σL qr (3.21) qr r qr r dt

and, φqs is q-axis component of stator ﬂux, Ls is stator inductance, Lm is magnetising inductance, Lr is rotor inductance, idr, iqr are d and q axes components of

12Refer to Appendix A for further detailed derivations. 54

rotor current respectively, vdr,vqr are d and q axes components of rotor voltage re-

spectively, ωr is rotor speed and ω is synchronous speed.

With stator ﬂux oriented scheme applied for the RSC, the active and reactive

power output of the DFIG given by (3.13) and (3.14) can be simpliﬁed to (3.22) and

(3.23) respectively:

3 P = v i (3.22) s 2 qs qs −3 Q = ( )v i (3.23) s 2 qs ds

Further, with SFO scheme, the equivalent circuit of the DFIG shown in Fig. 3.2 can be simpliﬁed to the form shown in Fig. 3.5. It is to be noted that the leakage inductance of the circuit is only referred to the rotor of the DFIG and if the stator resistance is neglected, then the stator voltage appears to be similar to the air gap voltage of the machine. s ϕ dqr Rs Lσ Rr

is s dqs i dqr

Lm s s v dqr v dqs

Figure 3.5: Stator ﬂux oriented equivalent circuit of a DFIG(superscript S denotes that the space vectors are referred to the stator ﬂux reference frame). 55

(b) Line Side Converter (LSC) Model

The LSC is used to control the DC bus voltage of the back-to-back converter system and to supply any reactive power to the loads if needed. In this regard, the

L-R ﬁlter model shown in Fig. 3.6 is used to develop the mathematical model of the controllers for LSC.

Inverter Rf ia,ib,ic Lf

va1 va + Load side vdc - vb1 vb

vc1 vc

Figure 3.6: Filter model associated with LSC.

The voltage balance across the ﬁlter components, Lf and Rf are given by (3.24).

        va ia ia va1             d     v  = Rf i  + Lf i  + v  (3.24)  b  b dt  b  b1         vc ic ic vc1

The vector representation of these balanced three-phase system and their equiva-

lent vectors in d-q rotating reference frame is given by (3.25)-(3.28).

∗ vds1 = vds − vds + Lf ωiqs (3.25)

∗ vqs1 = vqs − vqs − Lf ωids (3.26) di v∗ = R i + L ds (3.27) ds f ds f dt di v∗ = R i + L qs (3.28) qs f qs f dt 56

where, va, vb, vc are voltages on load side, ia, ib, ic are currents through the ﬁlter circuit, Lf , Rf are ﬁlter inductance and resistance respectively, va1, vb1, vc1 are voltages

at the inverter output, vds, vqs are d and q axes components of the load side AC voltage respectively, ids, iqs are d and q axes components of inverter current respectively and vds1, vqs1 are d and q axes components of the inverter output voltage respectively.

As stated earlier, assuming that load side voltage is regulated by RSC and readily available for LSC to use, a voltage orientation scheme is adopted for the LSC. In this regard, the q-axis component of the stator voltage, vqs is set to zero where the corresponding vector representation is shown in Fig. 3.7.

q - axis b - axis ω is

iq d = Vv id d - axis θ a - axis

c - axis Figure 3.7: Voltage vector orientation scheme of the LSC.

With above voltage orientation scheme, the active and reactive power associated

with the LSC are given by (3.29) and (3.30) respectively:

3 P = v i (3.29) LSC 2 ds ds 3 Q = v i (3.30) LSC 2 ds qs 57 3.2.4 Rotor Side Converter Control

As shown in Fig. 3.8, the RSC controller consists of inner-loops which have fast

ﬁeld oriented current control and the slow outer-loops that generate the reference

currents for the inner loops. The modelling work presented for the RSC in Section

3.2.3 is incorporated with this stage to develop a suitable control strategy taking into

consideration the main objectives (i.e. voltage and frequency regulation) as stated in

Chapter 1.

The voltage controller of the DFIG is developed using a reactive power based

control approach. In this regard, the total stator reactive power output Qs, of DFIG given in (3.23) can be further expanded into (3.31)13.

2 3 vs Lm Qs = [− + vs idr] (3.31) 2 ωLs Ls

The rotor d-axis current idr, consists of two components, namely: magnetising current, idrmag, which is mainly used for magnetisation purpose of the DFIG and idrgen which is used to satisfy the reactive power requirements of the loads. The corresponding reactive power components of these two currents, namely: Qmag and

Qgen are given by (3.32) and (3.33) respectively.

2 3 vs Lm Qmag = [− + vs idrmag] (3.32) 2 ωLs Ls 3 Lm Qgen = vsidrgen (3.33) 2 Ls

The no-load reactive power can be compensated by imposing the condition14 given by (3.34). In addition, the reference current of idrgen can be established by considering the voltage error which is compensated through a PI controller as in (3.35). Therefore,

13Refer to Appendix A for detailed derivations. 14 This can be achieved by making Qmag = 0 given by (3.32). 58

the reference d-axis component of the current which is used to satisfy the magnitude

of the stator voltage can be given as in (3.36).

vs idrmag = (3.34) ωLm

(idrgen)ref = (kp + ki)((vs)ref − vs) (3.35)

(idr)ref = (idrgen)ref + idrmag (3.36)

where, kp and ki are proportional and integral gains of the PI controller respectively.

As stated in Section 3.2.3, the stator ﬂux orientation scheme for the machine is

ensured by setting the q-axis component of the stator ﬂux to zero. Mathematically

this condition can be given as in (3.37) and is regarded as a criterion which needs to

be followed by the DFIG in order to regulate the frequency at the stator or load side.

Ls iqr = − iqs (3.37) Lm Therefore, q-axis component of the rotor current given in (3.37) is considered as the reference q-axis component of the rotor current which is used to achieve frequency regulation. In addition, a virtual phase lock loop (PLL) is used to deﬁne the reference frequency for the entire control scheme of the RSC as shown in Fig. 3.8.

RSC control algorithm is implemented by mainly considering the conditions given in (3.36) and (3.37) which are used to deﬁne the d and q axes reference currents

respectively for the inner-loop controllers as shown in Fig. 3.8. These reference

currents are compared with the actual rotor currents, idr and iqr and the error signals are then compensated using the PI controllers15 to generate the switching signals for

15The PI controllers are tuned using internal model control (IMC) principle [91]. An illustrative example is given in Appendix A. 59

the RSC. In addition, feed forward terms which are explained by (3.17)-(3.18) are

integrated into the control loops to avoid the cross coupling terms. The entire control

16 structure associated with RSC is shown in Fig. 3.8 .

Lrωslip irqσ

(v)ref (idr)ref + dq + PI + PI + PI - - + - abc

v i ms P To RSC ϑ 1/Lω idr PLL + W - M

ϑr (iqr)ref dq iqs L /L + PI - s m - + abc

2 iqr ωslip(Lrσ ird+Lmims÷Ls)

Figure 3.8: RSC control scheme.

3.2.5 Line Side Converter Control

The control scheme of LSC consists of a fast inner current control loop which controls

the current through the ﬁlter circuit given in Fig. 3.9. The outer slower control

loops are used to regulate the DC bus voltage of the back-to-back converter and

control reactive power supply through LSC. With reference to (3.29) and (3.30), it

is evident that the d and q axes components of currents17 through ﬁlter can be used

to regulate the DC link voltage and reactive power supply to the loads respectively.

Although there is a possibility of supplying reactive power through LSC similar to

a static synchronous compensator (STATCOM), in the present work, the reactive

16The equations that govern the entire control scheme of RSC is given in Appendix A. 17Active power is related to d-axis component current while reactive power is related to q-axis component current as evident from (3.29) and (3.30) respectively. 60

18 power reference Qref is set at zero . The corresponding control scheme implemented for LSC is shown in Fig. 3.9.

Liqs

(ids)ref + dq (vdc)ref + PI + PI - - - + abc

P v v  To LSC dc ids ds PLL + W + 2 M

(iqs)ref dq (Q)ref + PI + PI - - - + abc

Q L i iqs  qs

Figure 3.9: LSC control scheme.

3.3 Operating and Modelling Aspects of Permanent Magnet

Synchronous Generator (PMSG)

3.3.1 Overview of Operating Principles of PMSG

For the purpose of discerning the operating principle, a simpliﬁed representation of

the steady state equivalent circuit of a non-salient pole PMSG is shown in Fig. 3.10.

The circuit representation neglects the stator resistance. The internal voltage, E

19 behind the inductance Ld, is given as function of rotor speed, ωr and permanent

20 magnet ﬂux φpm. Further, the electromagnetic torque produced by the PMSG can be explained as in (3.38).

18It is assumed that the reactive power is entirely supplied through RSC and therefore avoids the complexities in coordinating the reactive power sharing between RSC and LSC. 19The winding resistance is neglected. 20 In this work, a non-salient pole machine is considered and hence xsd=xsq. The d-q equivalent circuit of a PMSG is given in Appendix A. 61

r r Te = iqsφpm (3.38)

r r where, iqs is the q- axis component of the stator current and φpm is the permanent magnet ﬂux linkage. ω j rLd

is + ω j rΦpm v - s

Figure 3.10: Simpliﬁed single phase equivalent circuit of round pole PMSG.

Considering the simplicity and reduced cost, the arrangement shown in Fig. 3.11

where a PMSG is connected to an uncontrolled three phase diode bridge rectiﬁer-

inverter system is employed in the present work.

Full bridge DC/DC converter Inverter PMSG rectifier idg G PL,QL

vdcge vdc

Wind turbine

Figure 3.11: Typical conﬁguration of PMSG wind energy system.

Unlike in a DFIG based wind generating system, total power generated by the

PMSG turbine passes through the rectiﬁer-inverter arrangement. The unregulated 62

DC bus voltage vdcge, which appears at uncontrolled diode bridge rectiﬁer is propor-

tional to the speed ωgm of PMSG and hence vdcge varies in an unregulated manner.

21 Therefore, a DC/DC converter is connected to regulate the DC bus voltage vdc of the system. The controlled DC link voltage is then converted to AC voltage using an inverter as shown in Fig. 3.11.

3.3.2 Inverter Control of PMSG

Noting that the PMSG generator is not directly connected to the load as shown in

Fig. 3.11, its excitation mechanism is not important for voltage regulation on the load side in comparison to a DFIG. It is therefore not the generator that supplies the reactive power required by the load and controls the load side voltage but the inverter. Noting these facts, the inverter control associated with the PMSG is used to regulate the magnitude of the AC voltage and frequency of the RAPS system. In this regard, a control algorithm has been developed considering the voltage balance across the ﬁlter circuit22 similar to the case of inverter which was explained in Section 3.2.3.

However, in this case the inverter is modelled as a voltage controlled voltage source inverter as its main objective is to control the load side voltage. Moreover, to achieve the decoupled control, the q- axis component of the stator voltage vqs is made equal to zero. In addition, the angular velocity ω, of the rotating axis system is deﬁned using a virtual PLL. The entire control structure adopted for the inverter control is shown in Fig. 3.12. All PI controllers associated with inverter control scheme are tuned using the internal model control principle as described in [91].

21 Usually vdcge < vdc and therefore, a boost converter needs to be placed just after the diode bridge rectiﬁer. 22Refer to Appendix A for more information. 63

Inverter R L va vdc vb Load side vc

abc v , v Control Signals ds qs dq

ϑ dq PI + (vds )ref abc -

P ϑ W PLL vds M dq (v ) PI + qs ref abc -

v qs

Figure 3.12: Inverter control of PMSG based RAPS system.

3.3.3 DC/DC Converter Control

23 As shown from Fig. 3.11, the output voltage vdcge of full bridge rectiﬁer varies in an uncontrolled manner. Therefore, a DC/DC converter is included which transforms

the unregulated DC link voltage to a regulated value. In this regard, a buck or

boost converter can be used to interface with the DC bus. Selection of buck or

boost converter conﬁguration depends on the output voltage vdcge that appears at the output of the uncontrolled diode rectiﬁer. The operating terminal voltage range of the

PMSG can be easily determined considering the voltage constant24 which is usually

provided by the machine manufacturer. Considering the machine parameters25, a √ 23 3 2vpmsg (vdcge) = π ; where vpmsg is line to line voltage which appears at stator of the PMSG. 24 This is given as (Vpeak)LL/krpm. where (Vpeak)LL is the peak line to line voltage and krpm is speed of the PMSG. 25Refer to Appendix A. 64

boost converter26 is selected as the preferred DC-DC converter in the present case.

As stated earlier, the rectiﬁed voltage output vdcge of the uncontrolled rectiﬁer is a

27 function of the speed of the generator ωr as given by (3.39)-(3.41) is shown in Fig. 3.13.

vs = | jωrφpm − jωrLdis | (3.39) √ 3 2v vdcge = (3.40) √π 3 2 q v = ω (φ )2 + (L i )2 (3.41) dcge π r pm d s

where vs is voltage appears across the stator terminal of the PMSG. In real life PMSG applications, the speed of the PMSG generator is kept within the allowable speed limit (ωr)min < ωr < (ωr)max. If the PMSG exceeds the permissible speed range, a dump load or pitch regulation can be activated to control the speed of the generator. Further details on the dump load and pitch angle control aspects are discussed in Section 3.4.

The proposed control scheme for the DC/DC converter is shown in Fig. 3.14.

The outer control loop measures the DC link voltage vdc, which is compared with the reference DC link voltage (vdc)ref , and the error is compensated through a PI controller to generate the reference current through the inductor of the boost converter,

(idc)ref as in (3.42). This current is then compared with the actual battery current ib, and the corresponding error is compensated through the second PI controller to generate the switching signal for the DC-DC converter. The main objective behind this control scheme is to regulate the generator current which is directly proportional to the load torque28 of the generator as given by (3.43). Further, the highest boosting

26The circuit diagram is given in Appendix A. 27This mathematical relationship can be derived using Fig. 3.10. 28The torque is in d-q domain is given in (3.38) 65

vdcge

regulated voltage vdcge2=vdc

safe operating speed range

vdcge1

(ω r)min (ω r)max ω r Figure 3.13: Boost converter operation for regulation of the DC bus voltage.

factor bf , of the boost converter is recorded at lowest generator speed (ωgm)min and can be given as in (3.44).

(idc)ref = ∆vdc(kp + ki/s) (3.42)

TPMSG = KT Idc (3.43)

(bf )max = vdc/vdcge1 (3.44)

where, KT is equivalent linkage ﬂux of the PMSG, idc is the current through the inductor of the boost converter, vdc is the regulated DC bus voltage and vdcge1 is lowest unregulated voltage present at the output of diode bridge rectiﬁer.

v (i ) To DC/DC converter (v ∆ dc dc ref dc)ref + - PI + - PI + - Limiter Comparator

vdc idc Triangular carrier waveform

Figure 3.14: Control strategy of the boost converter of the PMSG based WECS. 66 3.4 Active Power Control Techniques

3.4.1 Pitch Angle Control

A pitch angle regulator can be regarded as a mechanical control scheme that can be utilised to limit: (a) power output and (b) speed of a wind turbine generator. Al- though variable speed wind turbine generators allow operation under diﬀerent speeds, there is a maximum safe operating speed limit for each type of generator (e.g. The maximum speed of a DFIG based wind turbine is limited to 1.2 or 1.3 pu of its rated speed). If the wind turbine generator exceeds the maximum speed limit, pitch regulation can be employed in a manner that the power output of the wind turbine generator is regulated by adjusting the angle of the turbine blades to compensate for wind speed variations. There are various pitch regulation schemes employing for wind turbine generators. The adopted pitch regulation control scheme is shown in

Fig. 3.15. The pitch controller computes pitch angle β by comparing the diﬀerence between the maximum speed (ωr)max, and operating speed ωr.

ωr β k + - Pitch angle Pitch gain Limiter Rate limiter (ωr) max

Figure 3.15: Pitch angle control strategy for a variable speed wind turbine generator.

However, the pitch angle controller is a mechanically controlled mechanism which cannot be eﬀectively utilised to limit the power output of the wind turbine generator quickly due to slower mechanical dynamics. As an alternative solution, a dump load can be employed into a RAPS system which provides fast electrical dynamics compared to the latter option. 67 3.4.2 Application of Dump Load for Remote Power Applications

In general, dump loads are equipped with RAPS systems to absorb the instanta-

neous excess energy available that would otherwise cause unacceptable voltage and

frequency excursions. While providing the dynamic load balancing between the ﬂuc-

tuating wind energy and varying customer power demand, dump load control ensures

governing of the total rotary system29 by minimising the inertia gust, transients and stabilising the ﬂuctuation which arise due to wind and load proﬁles.

In practical RAPS systems, a dump load is a system which is capable of utilising the excess energy, an example of which is a space or water heating system. Moreover, the application of a dump load can be used for ice making, water desalination or water heating. The dump load has to be able to handle variable power input as the nature of the excess energy in a RAPS system is highly variable due to continuously changing wind and load conditions. In most of the cases, dump load consists of resistive elements which can be connected to either DC or AC sides of a wind energy system.

(a) Dump load for DFIG

Due to the limited power capacity associated with the back-to-back converter system30, the DC bus of the back-to-back converter is not identiﬁed as the best location to connect the dump load. Noting this issue, a dump load can be suitably located in AC side of the system where the capacity of the dump load is not restricted by any inverter constraints. In principle, dump load consists of three phase resistive elements which are connected across the switches. The control of the switches is executed at zero crossing points of voltage waveform to ensure minimum impact on the system voltage quality.

Noting the fact that the frequency control strategy of the DFIG suggested in

29The rotary system may include wind and diesel generators. 30Usually the maximum capacity is 20-30% of the rated rating of DFIG. 68

Section 3.2.3 is made independent of the loading condition and speed of generator, the power imbalance associated with the system is selected as the input signal to the controller. This analog input is converted to digital signals using analog to digital conversion unit and are fed into the switches. The maximum power that can be dissipated through a dump load can be expressed as in(3.45):

n (Pd)max = (2 − 1)Pstep (3.45)

where, n is the number of three phase resistive elements, (P )step is power that can be absorbed per resistor step. The condition under which the dump load operation is enabled can be described using (3.46):   Pd PDFIG > PL Pd = (3.46)  0 otherwise

where, PDFIG is power available through DFIG and PL is load demand. A simple schematic of the dump load controller is shown in Fig. 3.16.

PDFIG + A/D Conversion Dump Load -

Limiter PL

Figure 3.16: Dump load control strategy of the DFIG.

(b) Dump load for PMSG

Power generated by the PMSG, shown in Fig. 3.11, passes through the DC bus of the wind generating system. Therefore, any power imbalance that occur due to the generation-demand mismatch in the RAPS system is reﬂected as a DC bus voltage

fluctuation. In over-generation situations, the excessive power available in the RAPS system reflects as an over voltage condition of the DC bus. Contrarily, in under- voltage situations, the DC bus voltage is reflected as an under-voltage condition. 69

Hence, the DC bus voltage vdc, can be selected as an input signal to design a controller for the dump load. However, the DC bus voltage is controlled using the DC/DC

converter as stated in Section 3.14. As depicted in Fig. 3.13, the boost converter is

able to control the DC bus voltage only if the speed of the generator stays within

the allowable speed range given by (ωr)min-(ωr)max. However, during over-generation situations, the speed of the generator could exceed the maximum allowable speed of the generator, (ωr)max. In this case, the DC bus voltage can be regulated by activating the dump load operation which provides fast electrical dynamics compared to the pitch regulation which is relatively slow due to slower mechanical dynamics. The switching function is performed when the DC link voltage exceeds a pre-determined value as given by the condition (3.47):  0  Pd [vdc − (vdc)ref ] > β (vdc)ref Pd = (3.47)  0 otherwise

0 0 where, Pd is dump load power, β is fraction (0 < β < 1) and (vdc)ref is DC link reference voltage.

The rating of the resistor of the dump load can be given by (3.48).

2 (vdc) Rdump = 0 (3.48) α PPMSG

0 0 31 where, α is a fraction (0 < α < 1) and PPMSG is rated capacity of the PMSG. The arrangement of the dump load which is essentially a DC resistor connected via a switch is shown in Fig. 3.17. An additional PI controller and a hysteresis controller32 are integrated with a view to regulate the DC link voltage given by (3.47).

31This is related to the duty ratio of the switch in Fig. 3.17. 32This enables the operation of dump load based on condition given in (3.47). 70

(v ) dc ref + PI + - - vdc Limiter Comparator

vdc -1 Triangular carrier waveform Dump load resistor R

Hysteresis comparator

Figure 3.17: Dump load and its controller for PMSG.

3.5 Standalone Operating Performance of the Wind Turbine

Generators in RAPS Environments

Investigations have been carried out with a view to examine the following aspects of

RAPS systems:

• operation of the hybrid RAPS system (i.e. wind turbine generator and dump

load) and

• performance of wind turbine generators (i.e. DFIG and PMSG) alone33.

Hybrid operation of wind turbine generator is investigated to observe the suit-

ability of the proposed control strategies for each type of RAPS systems (i.e. DFIG

and PMSG). In this regard, the RAPS systems are assessed in terms of their band-

width of the voltage and frequency regulation capability and power sharing between

the components. However, it is not possible to investigate the system performance

during under-generation conditions34 due to the absence of other types of generating

33This is to investigate component level behaviour in d-q domain. 34Where wind power output is not able to satisfy the load demand. 71 source (e.g. diesel generator or energy storage) models which will be presented in the proceeding chapters. Therefore, the simulated results are presented to cover the over-generation situations only. The parameters associated with each RAPS system are given in Appendix A. Apart from the system level behavioural studies, an investigation is also conducted to observe the performance of the inverters (e.g. RSC, LSC and inverter of PMSG) in d-q domain with regard to their current and voltages. The parameters associated with each type of RAPS system is listed in Appendix A.

3.5.1 Performance of the DFIG based RAPS System

(a) Standalone operation of the hybrid DFIG based RAPS system

The hybrid RAPS system consisting of a DFIG as the wind turbine generator with a dump load is shown in Fig. 3.18. The control strategies discussed in relation to the

RSC and LSC in Sections 3.2.4 and 3.2.5 are employed for the DFIG. The methods discussed in Sections 3.4.1 and 3.4.2 are adopted to control the pitch angle and dump load respectively. DFIG

RSC LSC

Dump load

Main loads v,f vdc ∆p

Figure 3.18: DFIG based hybrid RAPS system.

The wind proﬁle under which the DFIG based RAPS system is simulated is shown in 3.19. As shown in Fig. 3.19 -(a), initially the wind speed is set at 12 m/s. At t=4 72 s, the wind velocity drops to 9 m/s, then increased to 11 m/s at t=6 s. The speed variation of the wind turbine generator and pitch angle behaviour are shown in Fig.

3.19-(b) and Fig. 3.19-(c) respectively and are discussed below.

10 V_w (m/s) 5 1 2 3 4 5 6 7 8 9 10 (a) 1.4

1.2 w_r (pu) 1 1 2 3 4 5 6 7 8 9 10 (b) 1

0.5

0 Pitch anglePitch (deg.) 1 2 3 4 5 6 7 8 9 10 (c) Time (s) Figure 3.19: Performance of the DFIG wind turbine system: (a) wind velocity, (b) speed of DFIG and (c) pitch angle.

System response of the DFIG based power system under variable wind and load conditions is shown in Fig. 3.20. The corresponding power sharing that takes place between the system components is shown in Fig. 3.21. As shown in Fig. 3.21-(c), initially the load demand is set to 0.6 pu and after t=5 seconds and t=8 seconds the load is reduced (i.e. step reduction) to 0.4 pu and 0.3 pu respectively. As evident from Fig. 3.20-(a), the load side voltage is maintained within ± 2% during the entire operating period. Also, the load voltage is not seen to be inﬂuenced by the load variations or wind speed changes. The frequency of the RAPS system on load side is also closely regulated at 1 pu. Moreover, the frequency variations are limited to

± 0.05% and are not seen to be aﬀected by wind speed or load variations and hence provides a better agreement with the control strategy adopted for the RSC in Section

3.2.4 (i.e. frequency regulation is achieved in such a way that it is independent of loading conditions and speed variations of the wind turbine generator). The DC 73

bus voltage variation of the RAPS system is shown in Fig. 3.20-(c). Upon close

examination, it can be noted that the variations of the DC bus voltage are only

limited to ± 1% of its rated value.

For simulation purposes, the slip of the DFIG was initially set to s=-0.1 which corresponds to super-synchronous mode of operation. As seen in Fig. 3.21-(a), the dump load starts absorbing the excessive power available in the RAPS system. How- ever, after t=2.5 seconds, the dump load reaches its maximum capacity and hence, the excessive wind power output is limited by activating the pitch control of the wind turbine system as evident in Fig. 3.19 -(c). At t=4 seconds, wind speed drops to 9 m/s causing a reduction in wind power output as shown in Fig. 3.21-(a) resulting the dump load to reduce the absorption rate of excess energy. The load step down occurs at t=5 seconds leading to a situation of excess energy available in the RAPS system and hence the dump load is activated to increase its power consumption. At t=8 seconds, the load demand is further reduced and the excess energy absorbed by the dump load and reaches its maximum capacity. As a result of load step reduction and also reaching of the dump load maximum capacity, wind turbine generator reduces its electrical power output to maintain the power balance of the system as evident from Fig. 3.21-(a). Reduction of electrical power output from wind turbine generator leads to acceleration35 of the wind turbine, which needs to be managed through pitch control as evident from Fig. 3.19-(c). The reactive power supply through DFIG to loads is shown in Fig. 3.22. As seen, the DFIG is able to supply the total reactive power demand of the loads.

The power quality behavour of a DFIG based RAPS system has been assessed through case studies. The voltage quality in terms of harmonic content of the load side voltage of the RAPS system has been investigated and the coressponding results

35 dω This can be easily identiﬁed using Tm −Te = J dt ; where Tm is mechanical torque, Te is electrical torque, J is moment of inertia of the machine and ω is speed of the wind turbine generator. 74

1.05

1 V_L (pu)

0.95 1 2 3 4 5 6 7 8 9 10 (a) 1.005

1 f_L (pu)

0.995 1 2 3 4 5 6 7 8 9 10 (b) 1.04

1.02

1 V_dc (pu)

0.98 1 2 3 4 5 6 7 8 9 10 (c) Time (s)

Figure 3.20: Response of the DFIG based RAPS system: (a) voltage on load side,

(b) frequency on load side and (c) DC bus voltage.

0.8

0.6 P_w (pu)

0.4 1 2 3 4 5 6 7 8 9 10 (a) 0.4

0.2 P_d (pu)

0 1 2 3 4 5 6 7 8 9 10 (b) 1

0.5 P_L (pu)

0 1 2 3 4 5 6 7 8 9 10 (c) Time (s)

Figure 3.21: Power sharing between system components: (a) DFIG power output, (b) dump load power and (c) load demand. 75

0.2 QL

0.15 QDFIG

0.1

0.05

-0.05 Reactive powerReactive (pu)

-0.1

-0.15

-0.2 1 2 3 4 5 6 7 8 9 10 Time (s)

Figure 3.22: Reactive power sharing between DFIG and loads. presented in Appendix A36. The existing distributed generation (DG) power quality standards are taken as the basis to compare the voltage quality of the simulated waveforms. The simulation results indicate the harmonic performance is anticipated to be improved if adequate ﬁltering is employed.

(b) Standalone operation of DFIG based wind turbine generator system

In this section, the performance of the DFIG is investigated in the d-q domain in relation to current, voltage and ﬂux components. The d-q axes rotor currents of the

RSC are shown in Fig. 3.23 and Fig. 3.24 respectively. Upon close examination, it can be realised that the actual rotor currents given by (idr)actual and (iqr)actual follow their corresponding reference currents (idr)ref and (iqr)ref ensuring robust voltage and frequency regulation as evident from Fig. 3.20. Stator ﬂux orientation scheme is employed for the RSC as stated in Section 3.2.3. The ability of the RSC to track the

SFO scheme is investigated during the operation of the hybrid RAPS system which is depicted in Fig. 3.25. As expected, the q-axis component of stator ﬂux is regulated closely at 1 pu while the d-axis component is maintained at zero.

36Refer to Section A.4 for more information. 76

1 (i ) dr ref (i ) dr actual 0.8

0.6

0.4 Current magnitudeCurrent (pu) 0.2

0 1 2 3 4 5 6 7 8 9 10 Time (s)

Figure 3.23: Actual and reference d-axis component currents of RSC.

1 (i ) qr ref 0.9 (i ) qr actual 0.8

0.7

0.6

0.5

0.4

Current magnitudeCurrent (pu) 0.3

0.2

0.1

0 1 2 3 4 5 6 7 8 9 10 Time (s)

Figure 3.24: Actual and reference q-axis component currents of RSC. 77

φφφ ds 1.2 φφφ qs

0.8

0.6

0.4 Flux magnitude (pu) 0.2

-0.2 1 2 3 4 5 6 7 8 9 10 Time (s)

Figure 3.25: Stator ﬂux components of the DFIG in d-q domain.

The performance of the LSC is also observed in d-q domain in relation its current and voltage components. Similar to the behaviour that is observed for RSC, the actual d and q axes currents associated with LSC follows their respective reference currents as evident from Fig. 3.26 and Fig. 3.27 respectively. Further, it should be noticed that, the q-axis component current is maintained at zero thus ensuring zero reactive power supply through LSC. 78

0.2 (i ds )ref

(i ds )actual 0.1

-0.1

-0.2

Current magnitudeCurrent (pu) -0.3

-0.4

-0.5 1 2 3 4 5 6 7 8 9 10 Time (s)

Figure 3.26: Actual and reference d-axis component currents of LSC.

0.25 (i ) qs ref 0.2 (i ) qs actual 0.15

0.1

0.05

-0.05

Current magnitudeCurrent (pu) -0.1

-0.15

-0.2

-0.25 1 2 3 4 5 6 7 8 9 10 Time (s)

Figure 3.27: Actual and reference q-axis component currents of LSC. 79 3.5.2 Performance of the PMSG based RAPS System

(a) Standalone operation of the hybrid PMSG based RAPS system

The hybrid RAPS system consisting of a PMSG as the wind turbine generator with the dump load is shown in Fig. 3.28. The strategies discussed for inverter and

DC/DC converter in Section 3.3 are employed to control the PMSG. The method discussed in Section 3.4 is adopted to control the dump load.

Full bridge Inverter PMSG rectifier DC/DC converter G

Wind turbine vdc v, f Dump load

v dc Main loads

Figure 3.28: PMSG based hybrid RAPS system.

The performance of the PMSG based RAPS system is investigated using similar load and wind conditions that are used in Section 3.5.1. The relevant wind speed proﬁle used and the corresponding generator speed are shown in Fig. 3.29.

The system response of the DFIG based power system under variable wind and load conditions is shown in Fig. 3.30. The respective power sharing that takes place between the system components is shown in Fig. 3.31. Compared to the performance of the DFIG based RAPS response presented in Section 3.5.1, the voltage and frequency regulation of the PMSG based RAPS system shown in Fig. 3.30-

(a) and (b) are seen to provide better performance during steady state operation.

However, the voltage and frequency proﬁles experience slight variations at t=3 and t=8 seconds which are mainly due to the load step changes. Also, it can be noted 80

10 V_w (m/s) 8 0 1 2 3 4 5 6 7 8 9 10 (a)

1.5

1 w_r (pu) 0.5 0 1 2 3 4 5 6 7 8 9 10 (b) Time (s)

Figure 3.29: Performance of the PMSG wind turbine system: (a) wind velocity and (b) speed of wind turbine generator. that the variation of voltage during transient conditions37 is limited to ± 5% of the rated value, while the frequency variation is less than 0.05% of the nominal level.

The DC bus voltage of the system is shown in Fig. 3.30-(c) and is regulated within

2% of its rated value between t=0 to 4 and t=7 to 10 seconds. As stated in Section

3.4.2-(b), this bandwidth regulation (i.e. 2%) of DC bus voltage is achieved through the dump load hysteresis controller where the bandwidth of the hysteresis controller of the dump load determines the lower and upper bounds of the DC bus voltage limits38. Further, the contribution of the dump load in regulating the DC bus voltage can be seen by referring to Fig. 3.30-(c) and Fig. 3.31-(b) where, the DC voltage is regulated within ± 2%. The reactive power supply through the inverter of the wind energy system to loads is shown in Fig. 3.32. As seen, the PMSG is able to satisfy the total reactive power demand of the loads.

37This corresponds to the situations such as load step changes 38Lower limit is zero and upper limit is selected to be 2%. 81

1.1

1 V_L (pu)

0.9 1 2 3 4 5 6 7 8 9 10 (a) 1.005

1 f_L (pu)

0.995 1 2 3 4 5 6 7 8 9 10 (b) 1.1

1 V_dc (pu)

0.9 1 2 3 4 5 6 7 8 9 10 (c) Time (s)

Figure 3.30: Response of the PMSG based RAPS system: (a) voltage on load side, (b) frequency on load side and (c) DC bus voltage.

0.5 P_w (pu) 0 1 2 3 4 5 6 7 8 9 10 (a) 2

1 P_d (pu) 0 1 2 3 4 5 6 7 8 9 10 (b) 1

0.5 P_L (pu) 0 1 2 3 4 5 6 7 8 9 10 (c) Time (s)

Figure 3.31: Power sharing between system components: (a) PMSG power output, (b) dump load power and (c) load demand. 82

0.25 Q L 0.2 Q inv 0.15

0.1

0.05

-0.05

Reactive powerReactive (pu) -0.1

-0.15

-0.2

-0.25 0 1 2 3 4 5 6 7 8 9 10 Time (s)

Figure 3.32: Reactive power sharing between inverter and loads.

(b) Standalone operation of PMSG based wind generator system

Inverter operation of the PMSG is investigated in relation to d-q axes currents

which is illustrated in Fig. 3.33 and Fig. 3.34 respectively. It can be seen that the

actual d and q axes components of the inverter currents follow their respective refer-

ence currents. As anticipated39, d-axis component current follows the same pattern

as active power output as shown in Fig. 3.31-(b). Similarly, the q-axis component of the inverter current follows the same behaviour of the reactive power40 as depicted in Fig. 3.35. As stated in Section 3.3.2, voltage orientation scheme adopted for the inverter and its performance are illustrated in Fig. 3.35. As anticipated, the q-axis

component of the load side voltage is maintained at zero while the d-axis component

of the same is set to 1 pu.

39 3 PLSC = 2 vdsids; where vds is the d-axis component of the voltage and ids is d-axis component of the current. 40 3 The relationship that exists among these variables can be quantiﬁed as; QLSC = 2 vdsiqs. 83

1.4 (i ) ds actual (i ) 1.2 ds ref

0.8

0.6

0.4 Current magnitudeCurrent (pu)

0.2

-0.2 0 1 2 3 4 5 6 7 8 9 10 Time(s)

Figure 3.33: Actual and reference d-axis component currents of the inverter.

0 (i ) qs actual (i ) -0.05 qs ref

-0.1

-0.15

-0.2

-0.25

-0.3 Current Current magnitude (pu) -0.35

-0.4

-0.45

-0.5 1 2 3 4 5 6 7 8 9 10 Time (s)

Figure 3.34: Actual and reference q-axis component currents of the inverter. 84

v ds 1.2 v qs 1

0.8

0.6

0.4

0.2 Magnitude of volatge (pu)

-0.2 2 4 6 8 10 Time (s)

Figure 3.35: Inverter voltage components d-q domain. 85 3.6 Chapter Summary

This chapter has addressed the modelling aspects of DFIG and PMSG for their standalone operation which are considered as the key components of the hybrid RAPS systems. In addition, diﬀerent dump load topologies with their respective control strategies have also been introduced. Moreover, a speed control mechanism has been employed for both RAPS systems.

Simulation exercises have been carried out to observe the suitability of the proposed RAPS systems in relation to their voltage and frequency regulation capability.

In this regard, the operation of two RAPS systems namely: DFIG and PMSG have been investigated under variable wind and load conditions. Based on the results obtained, the following conclusions can be drawn:

• Both RAPS systems are capable in regulating the voltage and frequency within

acceptable limits during the wind and load step changes.

• The transient performance of the DFIG based RAPS system is relatively better

compared to that of the PMSG based RAPS system. This is mainly due to the

control strategies that have been implemented for RSC. Frequency control is

achieved in such a manner that it is made independent of the rotational speed of

the generator and load changes. Also, the inertia provided by the DFIG helps in

mitigating the frequency variations compared to the PMSG counterpart. With

the adopted SFO scheme, the stator voltage appears to be equal to the air gap

voltage which is mainly controlled by the magnetising current41 and hence the

air gap ﬂux. In addition, the way how the voltage regulaion is achieved during

transient conditions can be simply explained by considering the SFO equivalent

circuit of the DFIG42, where the voltage that appears across the magnetising

41Refer to Fig. 3.5. 42Refer to Fig. 3.5. 86

inductance does not change quickly leading to a better voltage regulation.

• The PMSG based RAPS system seems to provide better steady state voltage

and frequency support compared to the DFIG counterpart. The performance

of the inverter used in PMSG based RAPS system is related to its rating and

the controller gain. Further, the reference frequency of the system is deﬁned by

a phase lock loop. Furthermore, inverter provides no inertial support compared

to the DFIG counterpart. The voltage regulation is achieved by operating the

inverter in voltage control mode where the performance are mainly determined

by the PI controller gains. Moreover, the voltage excursions that occur on the

load side are reﬂected in the DC side which primarily arise due to the fact that

DC bus dynamics are controlled by separate power electronic interfaces (e.g.

DC/DC converter and chopper interface of the dump load).

• The dump load control strategies adopted for both DFIG and PMSG are work-

ing well in maintaining the power balance of the system. The dump load control

approach that has been employed for DFIG based RAPS system is seen to pro-

vide a better power balance in the system compared to the dump load used in

PMSG. However, the dump load that has been introduced in the case of the

PMSG based RAPS system is seen to provide fast dynamic load balancing while

helping in maintaining the DC bus voltage within a pre-deﬁned value. Chapter 4

Application of Battery Energy

Storage for Wind Energy Based

RAPS Systems

4.1 Introduction

Chapter 3 introduced the modelling aspects of standalone wind energy conversion

systems (WECS), namely: DFIG and PMSG and their hybrid operation with dump

loads. This chapter continues with further development of such hybrid RAPS systems

incorporating energy storage systems. Battery storage systems are considered to be

the best option compared to other types of energy storage systems for standalone

wind energy applications due to their economic viability, long-term storage capability

and high density energy levels1. This chapter mainly investigates the suitability of such a battery storage system for a DFIG and a PMSG to perform the hybrid operation. In this regard, a linearised model of the RAPS system is undertaken where every component in the system is represented as a ﬁrst order transfer function.

1This can be characterised by long-term storage capability.

87 88

In addition, major emphasis is given to investigate the comprehensive operation2 of a

battery storage assisted RAPS systems (e.g. DFIG and PMSG). However, an energy

storage system cannot be simply added to a WECS without employing appropriate

control techniques. To address these issues, a control coordination methodology which

involves power exchange between sources and loads in the hybrid RAPS system is

proposed. In addition, an individual control strategy for each system component is

employed while giving due emphasis to the battery storage system.

4.2 Linearised Model of Wind Energy and Battery Storage

based RAPS Systems

The dynamic analysis of a RAPS system can be carried out considering higher or-

der mathematical models by incorporating associated nonlinearities of the system

components. While this is the case, acceptable results can also be obtained using

linearised models. In this regard, every system component can be modelled as a ﬁrst

order transfer function [70]. The transfer functions of the wind turbine generator and

energy storage3 are given as in (4.1) and (4.2) respectively.

KWTG ∆PWTG GWTG(s) = = (4.1) 1 + sTWTG ∆PW KESS ∆PESS GESS(s) = = (4.2) 1 + sTESS ∆f

where, PWTG is electrical power output of the wind generator, PESS is energy

storage power output, TWTG, TESS are wind turbine generator and energy storage time constants respectively, ∆f is frequency deviation of the system and KWTG,

KESS are constants. 2This mainly considers high order non-linear models of each system component. 3In this case, a battery storage system is considered. 89

Due to the inherent time delay that exists between system frequency variation and power deviation, the system characteristic equation can be given as in (4.3). Further, the wind turbine generator is considered to be an uncontrolled energy source of which the wind turbine characteristics are explained in Appendix B.

∆f 1 GSYS(s) = = (4.3) ∆pe D + sM

where, M is equivalent inertia constant, D is damping constant of the system and

Pe is demand-generation mismatch. Time constants of the transfer functions are selected by considering practical op-

erating conditions and characteristics of each component. For example, the time

constant of the energy storage system is selected to be small as it consists of a power

electronic interface. The numerical values of the parameters of each transfer function

used is listed in Table 4.1.

Table 4.1: Transfer function parameters of wind generator, energy storage system and load demand

Wind turbine generator KWTG = 1 TWTG = 1.5s

Energy storage system KESS = 1 TESS = 0.01s system characteristics D = 0.012 M = 0.012s

A simpliﬁed block diagram of the entire RAPS system is shown in Fig. 4.1. The

control strategy discussed in [70] is employed in the current study. The input to

the battery controller is taken as the sum of the error in supply demand ∆Pe, and the product of frequency deviation of the system ∆f and frequency characteristic

constant Kb of the battery storage system as shown in Fig. 4.1. 90

PI Kb

Battery Storage KBESS /(1+sT BESS )

vwind ∆Pe ∆ f K WTG /(1+sT WTG ) 1/(Ms+D)

Wind turbine Wind generator Load demand Pref

Figure 4.1: The linearised block diagram of the wind-battery hybrid RAPS system.

4.3 Detailed Model of Wind-Battery Remote Area Power

Supply Systems

In this section, the beneﬁts of energy storage system, coordinated control approach

and controller design for both DFIG and PMSG based RAPS systems are illustrated.

4.3.1 Beneﬁts of Energy Storage System for a Standalone Wind Power

Application

The arrangement shown in Fig. 4.2 can be used to investigate the beneﬁts of having an

integrated energy storage for standalone wind systems. To represent the intermittency

associated with the wind turbine power output, a variable power supply which consists

of a steady component vm of 230 V at fundamental frequency of fm, 50 Hz and a variable supply at frequency fs, 120 Hz with a voltage source of magnitude vs, 50 V is used.

As shown in Fig. 4.2, a battery system is selected to represent the energy storage

system which is incorporated into the DC bus of the wind energy system. The main

objective of the battery storage system is to regulate the DC bus voltage vdc. In this regard, a bi-directional boost converter is used to interface the battery storage system 91

vm , fm vs , fs Diode Rectifier Inverter L Rf Lf

vdc

Filter + Battery storage _

DC-DC converter Load

Figure 4.2: Schematic of the simpliﬁed standalone power supply system. to the DC bus. The conditions under which the battery storage system is operated can be explained by (4.4). Any power imbalance associated with the RAPS system shown in Fig. 4.2 can lead to DC bus voltage variation. Hence, the dc bus voltage variation is used as the input signal of the controller as shown in Fig. 4.3 for the battery storage system.

  chariging mode; ∆v > 0  dc  Battery status = discharging mode; ∆vdc < 0 (4.4)    idling mode; ∆vdc = 0

comparator (I ) (vdc)ref b ref + - PI + - PI + - Limiter I To DC-DC converter vdc b Triangular carrier waveform

Figure 4.3: Controller for the energy storage system.

The suitability of the energy storage system which helps in regulating the DC link voltage is observed under ﬂuctuating wind and load conditions. Initially, the load is set at 25 kW and after t = 3 s, the load demand is increased to 40 kW. The supply side voltage, which is used to simulate the power output of the wind turbine generator 92 is shown in Fig. 4.4-(a). The simulated behaviour of the DC bus and load side voltage with and without the energy storage system are shown in Fig. 4.4-(b) and Fig. 4.5 respectively. The corresponding battery storage current is shown in Fig. 4.4-(c). It can be seen that the DC link voltage is regulated within ± 5% of its rated value (i.e.

600 V) in the presence of the battery storage system. In contrast, the variation of

DC link voltage without the battery storage system is seen to vary within +5% and

-15% of its rated value. This simulation exercise clearly indicates the beneﬁts of the energy storage system for a standlaone wind application.

500

0 V_s (V) -500

1 1.5 2 2.5 3 3.5 4 4.5 5 (a)

640

620

600

580 V_dc (VDC) V_dc 560 1 1.5 2 2.5 3 3.5 4 4.5 5 (b)

0 -20 -40 -60 I_b (A) I_b -80 1 1.5 2 2.5 3 3.5 4 4.5 5 (c) Time (s)

Figure 4.4: RAPS system performance with battery storage system: (a) supply voltage, (b) DC link voltage and (c) battery current. 93

620

600

580

560 V_dc (V) V_dc

540

520

500 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 Time (s)

Figure 4.5: DC link voltage in the absence of the battery storage.

4.3.2 Coordinated Control Approach of Wind-Battery RAPS system

Coordinated control approach for RAPS system consisting of wind energy system, battery storage and dump load is necessary in order to maintain the active and reactive power balance of the RAPS systems. In this regard, the reactive power requirement of the RAPS system is satisﬁed by the inverter control associated with the wind energy conversion system. In contrast, the active power of the system is maintained according to the coordinated control approach depicted in Fig. 4.6.

The proposed coordinated control approach mainly emphasises on the control logic associated with the decision making process of active power sharing between the different system components covering three diﬀerent scenarios, namely: over-generation

4 (Pw > PL), under-generation (Pw < PL) and emergency (Pw = 0) situations. Among these scenarios, emphasis is given to investigate the behaviour of the RAPS system

4e.g. wind turbine generator experiencing a wind speed below its cut-in speed. 94

wind power generation, Pw

vcut-in < v < v cut-out

Yes

No Battery Yes P +(P

P L Discharging w b)max L Shedding

Yes No Battery Charging

No Dump Load No Pitch Pb<(Pb)max P < (P ) "ON" d d max Regulation

Yes Yes

Frequency Regulation

Figure 4.6: Control coordination of a wind-battery hybrid power system. 95 during over-generation and under-generation situations given by (4.5) and (4.6) respectively. The sign convention associated with the power ﬂow direction between the system components is diagrammatically shown in Fig. 4.7.

Pw = PL + Pd + Pb (4.5)

Pw + Pb = PL + Pd (4.6)

where, Pw is power output of wind turbine generator, Pb is power from battery storage system, Pd is dump load power and PL is load demand.

Pw Battery charging (+)

P b Pd PL (a) Pw Battery discharging (-)

P b Pd PL

(b)

Figure 4.7: Power ﬂow directions of the components during (a) over-generation and (b) under-generation.

Power sharing between the system components shown in Fig. 4.6 can be explained as follows: If wind speed, vw stays within safe limits (i.e. vcut−in < vw < vcut−out) and the power output of the wind turbine generator Pw, is greater than the load demand PL, the battery storage is used to absorb the excess power given by (Pw-

PL). However, if the excess generation (Pw-PL), exceeds the maximum capacity of the battery storage system (Pb)max, the dump load starts absorbing the additional 96 power. When the dump load power consumption Pd, reaches its maximum rating

(Pd)max, the wind turbine pitch regulation is activated in order to control the power output of the wind turbine. During under-generation conditions, the power output of the wind turbine generator is lower than the load demand (Pw < PL), and hence the battery storage system moves into the discharge mode of operation. If the combined power output of the wind generator and maximum battery storage capacity is less than the load demand (i.e. Pw + (Pb)max) < PL) a load shedding scheme can be implemented.

The above stated control coordination concept has been realised by applying individual control strategies on each system component of both RAPS systems. In this chapter, it is assumed that Pw and Pb are suﬃcient to supply the load demand at all times except emergency situations5.

4.3.3 Controller Design

In order to avoid the oscillatory transients during state transitions (i.e. from “ON” to

“OFF” and vice versa) between the system components, each decision making block associated with active power given in Fig. 4.6 is further enhanced as shown in Fig.

4.8.

Each decision making block associated with real power compares the active power measurements of two diﬀerent system components (e.g. wind generator supply and load demand) or compare the active power measured with its own maximum/minimum power rating (e.g. Pd with (Pd)max) denoted by, Px and Py. The outcome of such decision making blocks is further compared using a hysteresis controller. The bandwidth of the hysteresis controller, which represents the minimal boundaries of power diﬀerence, ∆Pe(t) is used to determine the logic state (LS) of the component (i.e. ei-

5Behaviour of the RAPS system under emergency situations will be discussed in Chapter 6. 97

Px(t) , Py(t)

LS=logic state

Pe(t ) = Px(t) - Py(t)

of P e(t ) Hysterisis controller

t=t+δ

Time Delay

LS of LS No LS of of Pe ( t+δ ) P e(t ) = control logic = 0 Pe ( t+ δ ) (Device "OFF")

Yes control logic = 1 (Device "ON") Figure 4.8: Decision making process associated with the state transition of a device. 98

ther logic ‘1’ or ‘0’). The bandwidth can be estimated considering the maximum

allowable frequency deviation, (∆f)max as given in (4.7). Once the measurement,

∆Pe(t) is obtained at t, the corresponding LS is processed with a time delay δ. The next LS at t = t + δ is obtained using the next set of the measurements, Px and

Py and then compared with the previous logic state. Table 4.2 describes the logic state and the device (e.g. battery storage system) operating status. In addition, the estimation of the time delay, denoted by t = δ should satisfy the condition given by

(4.8). This condition is to ensure that the devices act to maintain the power balance

6 of the RAPS system to avoid any frequency deviation (i.e. fmin ≤ f ≤ fmax). 1 ∆P = (∆f) [ ] (4.7) e max D + sM 1 tδ ≤ (4.8) (∆f) max

where, M is equivalent inertia constant, D is damping constant of the system and tδ is time delay.

Table 4.2: Control logic to determine the device status

∆Pe(t) ∆Pe(t + δ) Status of the Device 0 0 OFF 0 1 OFF 1 0 OFF 1 1 ON

4.3.4 DFIG based Wind-Battery Hybrid RAPS System

The RAPS system shown in Fig. 4.9 consists of a DFIG as the main energy source

and a battery storage system which is connected to the DC bus of the back-to-back

6Typically the maximum allowable frequency deviation is deﬁned by the respective standards such as IEEE 1547. In this study fmin=0.99 pu and fmax=1.01 pu are considered as the extreme range of the frequencies. 99 converter system. In addition, a dump load is connected to the load side of the RAPS system. Both, battery storage and dump load act as auxiliary system components which help in regulating the active power balance of the RAPS system.

DFIG

RSC LSC

Dump load Main loads DC-DC v,f vdc converter ∆P + _

Battery storage

∆P

Figure 4.9: DFIG based wind-battery RAPS system.

As indicated Sections 3.2.4 and 3.2.5 of Chapter 3, the control strategies associated with the DFIG are used to regulate the AC voltage and frequency. In this regard, the control strategy implemented for the DFIG acts as the primary frequency control. The secondary frequency control is realised by ensuring the active power balance of the RAPS system as discussed in Section 4.3.2. The control approaches developed for the battery storage system and dump load are discussed below. 100

(a) Battery Storage System Control

The battery storage system is connected to the DC bus using a bi-directional

buck-boost converter as shown in Fig. 4.10. During over-generation, power output

of the wind turbine generator PDFIG is greater than the load demand PL and hence the battery storage acts as a load while operating on charging mode (i.e. enabling

the buck operation of the bi-directional buck-boost converter). Contrarily, during

under-generation conditions the battery storage operates on discharging mode (i.e.

boost operation). Under balanced operating conditions where the wind power output

matches with the load demand, the battery storage system stays at the idling7 mode.

All three possible operating conditions of the battery storage system are described in

(4.9)8.   P > P ; chariging mode (buck operation)  DFIG L  Battery status = PDFIG < PL; discharging mode (boost operation) (4.9)    PDFIG = PL; idling mode

The controller for the battery storage system is developed with a view to achieve two objectives. Firstly, it should be able to minimise the demand-generation mismatch. Secondly, the controller helps in extracting the maximum amount of power from wind allowing the wind turbine generator to operate in its optimum mode. The demand-generation mismatch of the RAPS system can be estimated based on (4.9) and is used to obtain the reference battery current, (ib)ref . To achieve the second objective, optimum wind power9 from the DFIG is considered as one of the inputs to

estimate the reference battery current as given in (4.10).

7Battery storage system is neither in charging nor in discharging mode of operation. 8 The same is explained in relation to the DC bus voltage, vdc variation of a WECS is explained in (4.4). 9This refers to MPPT algorithm given in Chapter 2, Section 2.3.3. 101

Q1 DC bus of back-to-back L

converter Q2 + Battery storage _ Switching signals

Figure 4.10: Bi-directional buck-boost converter arrangement for battery storage system.

(PDFIG)opt − PL (ib)ref = (4.10) vb

where, PL is load demand, vb is battery voltage and (PDFIG)opt is optimum power from DFIG.

For grid connected DFIG applications, the q-axis component of rotor current iqr is used to extract the maximum power from wind using RSC converter control [88]-

[89]. Contrarily, in standalone mode of operation, the current iqr, is utilised to operate

10 the RSC in SFO mode as given by (4.11) . Therefore, iqr cannot be directly used to extract the maximum power from wind during standalone operation of the wind turbine generator. Instead, the condition given by (4.10) can be regarded as an indirect technique in harvesting the maximum power from wind compared to grid- connected operation.

Ls iqr = (− )iqs (4.11) Lm 10This relationship is reproduced in Section 3 in (3.37) of Chapter 3. 102

The key idea behind the proposed indirect maximum power extraction strategy is to regulate the power ﬂow of the battery storage in a manner that helps in extracting maximum power from wind by imposing an appropriate toque as in (4.12). When the

DFIG operates at the maximum power extraction mode, the torque generated by the

DFIG given in (4.13) should be equal to the optimum torque as in (4.12). As evident from (4.14), the maximum power extraction from DFIG can be realised by varying iqs, the q-axis component of the stator current. Therefore, by allowing the battery storage current as given in (4.10), it is possible to vary iqs resulting DFIG to operate on maximum power extraction mode.

2 (Te)opt = kopt(ωr) (4.12)

Lm Vs Te = iqr (4.13) Ls + Lm ωs s Lsvs ωr = ( )iqs (4.14) (Ls + Lm)kopt

where, Te is electromagnetic torque of the DFIG, iqr, iqs are rotor and stator q- axis currents respectively, Ls, Lm are stator inductance and magnetising inductance respectively.

The proposed control strategy for the battery storage system is shown in Fig. 4.11.

The reference battery current, (ib)ref is estimated using (4.10) and compared with the actual battery current ib, to generate the switching signal for the bi-directional buck-boost converter.

Comparator To bi-directional (P buck-boost converter DFIG)opt P (ib)ref + - ÷ + - PI + - Limiter

v ib PL b Triangular carrier waveform

Figure 4.11: Battery storage control strategy for DFIG. 103

The capacity estimation of the battery storage system depends on many factors

such as wind proﬁle and load demand. However, in the current study, the size of the

battery storage system is estimated using (4.15). In real life wind-battery applica-

tions, the capacity of the battery storage needs to satisfy the inverter constraint given

by (4.16).

t µ × i × ( ) = (Ah rating) × k0 (4.15) rated 60

(Pb)max ≤ smax × (PDFIG)rated (4.16)

where, µ is fraction of the rated current of the load demand, irated is rated current corresponding to load demand, t is time duration over which the battery provides

power to the system, k0 is average discharge/charge current of the battery storage in

pu, smax is maximum allowable slip of DFIG and (PDFIG)rated is rated capacity of the DFIG.

The dynamics associated with the DC bus voltage vdc, in the presence of the battery storage system can be expressed using (4.17)-(4.20) which also form the basis

to compute the PI compensator control parameters for LSC using the internal model

control (IMC) principle[91].

PLSC = Pdc ± Pb (4.17) 3 v i = v i ± v i (4.18) 2 ds ds dc dc b b dv C dc = i (4.19) dt dc 1 Z 4 v = ( mi ∓ k0i ) (4.20) dc C 3 ds b

where, PLSC is active power through load side converter, Pdc is DC power due to

DFIG operation only, Pb is power from the battery storage, vds is d-axis component of LSC output voltage, ids is d-axis component of LSC output current, vdc is DC link 104

voltage, idc is DC current relevant to DFIG operation only, vb is battery voltage, ib is battery current, C is capacitance of the DC link, m is modulation index of the LSC

and k0 is ratio of battery voltage to DC link voltage.

(b) Dump load controller

An arrangement similar to the one shown in Chapter 3 in Section 3.4 is used to

control the dump load. In this case, however, the dump load operation is enabled

during situations where the battery storage system is not able to handle the excess

power (e.g. when Pb > (Pb)max) during the over-generation conditions. The necessary and suﬃcient conditions under which the dump load enables the operation is given

by (4.21).   Pd (PDFIG)opt + (Pb)max > PL Pd = (4.21)  0 otherwise

Moreover, the dump load will start dissipating the additional power after the battery storage system reaches its rated capacity, (Pb)max. In addition to load balancing, the dump load is also used to extract the maximum power from wind as evident from

(4.21) where the reference dump load power is calculated considering optimal wind

power output (PDFIG)opt, as one of the input. The key concept behind the maximum power extraction from wind utilising the dump load remains similar to the method

discussed in Section 4.3.4-(a).

4.3.5 PMSG based Wind-Battery Hybrid RAPS System

The layout of the PMSG based RAPS system consisting of a wind turbine generator,

a battery storage and a dump load is shown in Fig. 4.12. The coordinated control

approach that is illustrated in Fig. 4.6 is employed for this RAPS system as well. The

battery storage system is connected to the DC bus of the wind energy system using 105

DC/DC converter-211 that is used to regulate the DC bus voltage. Also, as discussed in Chapter 3 in Section 3.5.2 the dump load is used to regulate the DC bus voltage.

In addition, DC/DC converter-112 is used to extract the maximum power from wind.

The control approach that is applied for the inverter remains unchanged as stated in

Chapter 3 in Section 3.3.2. Hence, signiﬁcant attention is given to develop the control strategies for the battery storage system, dump load and DC/DC converter-1.

Full bridge PMSG rectifier DC/DC converter-1 Inverter G

Wind turbine (PPMSG )MPPT v, f DC/DC converter-2 Dump load + Battery storage vdc _ Main loads

v dc Figure 4.12: PMSG based wind-battery RAPS system.

(a) Control Strategy for DC-DC Converter-2

The battery storage is used to regulate the DC bus voltage of the wind energy system. In contrast to the DFIG arrangement shown in Fig. 4.9, all power generated by PMSG passes through the DC bus of the wind energy conversion system. There- fore, any power imbalance that occurs in the RAPS system can introduce DC bus voltage variations as described by (4.22).   P > P v > (v ) battery charging  PMSG L dc dc rated  Battery status = PPMSG < PL vdc < (vdc)rated battery discharging (4.22)    PPMSG = PL vdc = (vdc)rated battery idling

If the wind power output PPMSG, is greater than the load demand PL, the excessive

11This is a bi-directional buck boost converter. 12This is a boost converter. 106

power (∆P = PPMSG − PL) will lead to over-voltage conditions. Contrarily, when the power output of the wind turbine generator is smaller than the load demand, the

power deﬁcit (i.e. (−∆P )) lead to under-voltage conditions. With the integration of the battery storage system into the DC bus, the power balance at DC node can be given by (4.23).

PPMSG ± Pb = PL (4.23)

where, Pb is power output from battery storage.

DC node

PPMSG PL

Generator v Inverter side side dc

Pb + Battery

storage _

Figure 4.13: Dynamics of DC bus of the PMSG arrangement.

The battery storage system is interfaced into the DC bus using a bi-directional

buck-boost converter (i.e. DC/DC converter-2) is shown in Fig. 4.12. The pro-

posed controller for the battery storage system is similar to the scheme shown in Fig.

4.3 where the outer loop controls the DC bus voltage and inner loop generates the

reference current of the battery storage system. 107

(b) Control Strategy for DC-DC Converter-1

As stated earlier in Section 3.3.3 of Chapter 3, the DC-DC converter-1 shown in

Fig. 4.12 represents a boost converter13. In the present context, the controller asso-

ciated with this boost converter is designed to extract maximum power (PPMSG)opt

14 from wind. As evident from (4.24) , the torque of the generator TPMSG, can be

15 controlled using the rectiﬁed current , idc. As stated before, the corresponding generator torque at optimum power can be given by (4.12). If the generator operates

at maximum power extraction mode, the relationship that exists between generator

speed and the current through the boost converter can be given as (4.25). It can be

seen that by varying the rectiﬁed current idc, it is possible to operate the generator at optimum power tracking mode.

TPMSG = KT idc (4.24) s KT ωr = idc (4.25) Kopt

where, KT is equivalent ﬂux linkage of the PMSG and idc is current through the inductor of the boost converter.

The process associated with generation of control signal for the boost converter

is illustrated in Fig. 4.14. It consists of two PI controllers where the generated wind

power of the PMSG is controlled in the outer loop while the inner loop controls the

current idc, through the inductor of the boost converter. The outer loop compares the

actual wind power (PPMSG) with the reference optimum power (PPMSG)opt, which can be obtained using MPPT algorithm. The subsequent error is compensated through

a PI controller to produce the reference current (idc)ref , through the boost converter.

13This is due to unregulated rectiﬁer voltage that is lower than the DC bus voltage. 14This is reproduced from Chapter 3 at Section 3.14. 15The current passes through the inductor of the boost converter. 108

Comparator (P ) (idc )ref w opt + PI + PI + - - - To DC/DC Limiter converter

Pw idc Triangular carrier wave form

Figure 4.14: Maximum power extraction control strategy for DC/DC converter-1.

The dump load is used to regulate the DC bus voltage in situations16 where the

battery storage alone cannot regulate the DC bus voltage. The arrangement of the

dump load remains similar to the construction that was depicted in Chapter 3, Section

3.4. With the integration of the dump load into the DC bus of the wind energy system,

the DC bus voltage dynamics can be expressed using (4.26).

s Z 2 2 (vdc) vdc = (PPMSG − ± Pb − Pinv)dt (4.26) C Rd

where, vdc is DC link voltage, C is capacitance, PPMSG is wind power output from

PMSG, Rd is resistance of dump load, Pb is battery power, Pinv is inverter power output.

4.4 Performance Evaluation of the Hybrid Wind-Battery RAPS

Systems

The hybrid operation of the wind-battery RAPS systems has been studied considering their linearised and detailed models. In this regard, the detailed models of the

RAPS systems have been examined in relation to diﬀerent types of systems described

16Situations where the battery storage reaches its full capacity and it cannot participate in regulating the DC bus voltage. 109 in Sections 4.3.4, and 4.3.5 respectively. Moreover, the suitability of the adopted control strategies for such RAPS systems has also been assessed with regard to system response (e.g. voltage and frequency regulation), power sharing between diﬀerent system components and maximum power extraction capability from wind. Results have been observed under variable wind and load conditions. All the relevant information

(e.g. rating of wind turbine generator, DC bus voltage, battery storage rating etc.) associated with RAPS systems are listed in Appendix B.

4.4.1 Performance of the Linearised Model of Wind-Battery RAPS

System

The simulated behaviour of the hybrid power system is shown in Fig. 4.15. The wind velocity variation is shown in Fig. 4.15-(a) whereas the corresponding wind power output is depicted in Fig. 4.15-(b). Initially, the average wind velocity is set to 12 m/s and the corresponding wind power17 output from the wind generator is

0.73 pu. As evident from Fig. 4.15-(d), initially the load demand is set at 0.6 pu where the excess power (PL − Pw) = 0.13 pu is absorbed by the battery storage as shown in Fig. 4.15-(c). The wind velocity drops to a value of 10 m/s at t = 75 s. The corresponding wind power output is nearly 0.6 pu where the corresponding power imbalance ∆P , is supplied through the battery storage system. At t = 150 s, the load demand increases to 1 pu causing the battery storage to supply nearly

0.4 pu power to maintain the power balance of the system. The power imbalance associated with the system ∆P = Pw ± Pb − PL, is shown in Fig. 4.16-(a) which is closely maintained at zero. The corresponding system frequency deviation ∆f is shown in Fig. 4.16-(b). Upon close examination, it can be seen that the frequency of the system is regulated within ±0.1%. As expected, the highest frequency deviation

17Refer to Appendix B for further information. 110 is seen to occur at t = 150 s during the step load change.

15 10

V_(w)(m/s) 5 50 100 150 200 250 300 (a) 0.8

0.6

P_w (pu) P_w 0.4 50 100 150 200 250 300 (b)

0.5 0

P_b (pu) P_b -0.5 50 100 150 200 250 300 (c)

P_L (pu) P_L 0.5 50 100 150 200 250 300 (d) Time (s)

Figure 4.15: Power Sharing of the RAPS system under variable wind and load conditions: (a) wind speed, (b) power output of wind turbine generator, (c) battery storage power output, (d) load demand.

4.4.2 Performance of the DFIG-Battery RAPS System

(a) Behaviour of DFIG Based RAPS System Below Rated Power

The system response and power sharing between the diﬀerent components of the

DFIG based RAPS system are shown in Fig. 4.17 and Fig. 4.18 respectively. The wind condition under which the system has been simulated is shown in Fig. 4.17-(a) and Fig. 4.18-(d) shows the load demand which is set initially at 0.35 pu. At t = 4 s, the load demand is increased to 0.6 pu as shown in Fig. 4.18-(d). The voltage on the load side is shown in Fig. 4.17-(b) which is not seen to be aﬀected by the wind speed changes. At the time of load addition (i.e. at t = 4 s), the load voltage drops to 0.975 pu but recovers quickly to its rated value. However, the load side voltage stays within 111 0.4 0.2 0

Delta(pu) P -0.2 50 100 150 200 250 300 (a) -3 x 10 5

Deltaf(pu) -5 50 100 150 200 250 300 (b) Time (s)

Figure 4.16: (a) power imbalance and (b) frequency deviation of the RAPS system.

±1% of its rated value during normal operation. Fig. 4.17-(c) shows the operating

frequency which is regulated very close to 1 pu. As indicated in Chapter 3 in Section

3.2.4, the proposed frequency control methodology of the DFIG is independent of

the shaft speed and load condition of the system. Therefore, as anticipated, the

frequency is not seen to be aﬀected by the wind speed and load variations. The DC

link voltage of the DFIG is depicted in Fig. 4.17-(d). The simulated behaviour of the

DC link voltage shows that it is regulated at the rated value throughout the operation

except during the load step changes. When load step change occurs at t = 4 s, the

battery storage system suddenly changes its direction of power ﬂow (i.e. charging to

discharging mode of operation) as evident from Fig. 4.18-(b). This instantaneous

power ﬂow reversal causes the DC link voltage ﬂuctuation which can be described

dvdc using the capacitor voltage equation (i.e. ic = C dt ). At the time of load step increase at t = 4 s, the DC link voltage stays within +10% and −5% of its rated value.

The wind power variation of the system is shown in Fig. 4.18-(a). For simulation purposes, the slip of the wind turbine is initially set to s=-0.1 which corresponds to 112

10 V_w (m/s) 5 1 2 3 4 5 6 7 (a) 1.05

1 V_L (pu) V_L 0.95 1 2 3 4 5 6 7 (b) 1.005

1 f_L (pu) 0.995 1 2 3 4 5 6 7 (c) 1.2

V_DC (pu) 0.8 1 2 3 (d) 4 5 6 7 Time (s)

Figure 4.17: Response of the DFIG based wind-battery RAPS system under wind and load step changes: (a) wind speed, (b) voltage on load side, (c) frequency on load side, and (d) DC link voltage.

super synchronous mode of operation. Referring to the wind turbine characteristics18, the corresponding maximum power output of the wind generator is 0.73 pu at a shaft speed of 1.2 pu for a wind speed of 12 m/s. From Fig. 4.18-(a), the power output of the DFIG is seen to rise to a value of 0.73 pu until the wind speed changes at t = 3 s. At this time, the RAPS system experiences an over-generation scenario where the excess power (Pw − PL), is shared between the battery storage and dump load as shown in Fig. 4.18-(b) and 4.18-(c) respectively. At t = 1.5 s, the battery storage reaches to its maximum capacity of 0.3 pu and hence the excessive power is seen to be absorbed by the dump load. At t = 3 s, the wind velocity drops to 9 m/s causing a reduction in wind power output and as a result, the dump load power consumption is reduced as shown in Fig. 4.18-(c). However, at t = 4 s, the load is increased to 0.6 pu and the RAPS system experiences under-generation, where the

18Refer to Appendix B for further information. 113 battery storage system changes its operation from charging to discharging mode as evident from Fig. 4.18-(b). When the load step reduction occurs at t = 6 s, the

RAPS system experiences an over-generation situation again and the battery storage moves from discharging to charging mode to maintain the power balance of the RAPS system.

Fig. 4.19 shows the actual wind power output (Pw)Actual and the optimal wind power output (i.e. (Pw)MPPT ). It can be seen that throughout the operation, the actual power output of the wind turbine generator follows the optimum power curve, thus ensuring the suitability of the proposed MPPT control strategy of the DFIG based RAPS system.

0.5 P_w (pu) P_w 0 1 2 3 (a) 4 5 6 7

0.4 0.2

P_b (pu) 0 1 2 3 4 5 6 7 (b) 0.1

0 P_d (pu) -0.1 1 2 3 (c) 4 5 6 7 1

0.5 P_L (pu) 0 1 2 3 (d) 4 5 6 7 Time (s)

Figure 4.18: Power sharing of the DFIG based wind-battery RAPS system under wind and load step changes: (a) wind generator power output, (b) battery storage power, (c) dump load power and (d) load demand. 114

1 P Actual w P MPPT 0.9 w

0.8

0.7

Power (pu) 0.6

0.5

0.4

1 2 3 4 5 6 7 Time (s)

Figure 4.19: Maximum power extraction from wind in the DFIG based wind-battery RAPS system under wind and load step changes.

(b) Behaviour of DFIG based RAPS System above Rated Power

The system response under high wind regimes is shown in Fig. 4.20, where the initial wind speed is set at 13.5 m/s. At t = 3 s, the wind speed drops to 9 m/s, then it is increased to 11 m/s at t = 5 s. The power sharing between the system components is shown in Fig. 4.21, where the initial load demand is set at 0.425 pu. At t = 4 s the load demand is increased to 0.725 pu and is reduced back to the original value at t = 6 seconds. As evident from Fig. 4.21-(c), at t = 2.2 s the dump load reaches its maximum capacity at which the pitch angle control is activated in order to limit the wind generator as shown in 4.21-(e). The maximum power extraction from wind is shown in Fig. 4.22. 115

V_w (m/s) 5 1 2 3 4 5 6 7 (a) 1.1

1 V_L (pu) 0.9 1 2 3 4 5 6 7 (b) 1.005

1 f_L (pu) 0.995 1 2 3 4 5 6 7 (c) 1.2

V_DC (pu) 0.8 1 2 3 (d) 4 5 6 7 Time (s)

Figure 4.20: Response of the DFIG based wind-battery RAPS system under high wind regimes: (a) wind speed, (b) voltage on load side, (c) frequency on load side, and (d) DC link voltage.

0.5

P_w (pu) P_w 0 1 2 3 4 5 6 7 (a) 0.5

P_(b) (pu) -0.5 1 2 3 (b) 4 5 6 7 0.4 0.2 0 P_d (pu) 1 2 3 (c) 4 5 6 7 0.8 0.6

P_L (pu) 0.4 1 2 3 (d) 4 5 6 7

1 0.5

Theta (deg) 0 1 2 3 4 5 6 7 (e) Time (s)

Figure 4.21: Power sharing of the DFIG based wind-battery RAPS system under high wind regimes: (a) wind generator power output, (b) battery storage power, (c) dump load power, (d) load demand and (e) pitch angle. 116

Pw Actual P MPPT 0.9 w

0.8

0.7

0.6 Power (pu)

0.5

0.4

1 2 3 4 5 6 7 Time (s)

Figure 4.22: Maximum power extraction from wind in the DFIG based wind-battery RAPS system under high wind regimes.

Driven Pump Load

Performance of the DFIG based RAPS system with an induction motor driven pump load19 is investigated. Response of the RAPS system during starting and running of the pump load is shown in Fig. 4.23.

As shown in Fig. 4.23-(a), the initial wind velocity is set at 12 m/s and at t = 4 s, wind velocity drops to 9 m/s. At the beginning, the load demand of the RAPS system is set at 0.65 pu. An induction motor driven pump load20 having a nominal rating of 0.1 pu is added at t = 4 s and it is disconnected at t = 6 s. As anticipated, the highest voltage and frequency deviation of the RAPS system occurs at t = 4 s which is mainly due to the direct online starting of the induction motor. However, the voltage and frequency of the RAPS system are restored quickly back to their rated values within t = 0.25 s. The corresponding voltage and frequency deviations

19Refer to Appendix B for more information. 20Refer to Appendix B for the torque estimation of the pump load. 117

V_w (m/s) 5 1 2 3 4 5 6 7 (a)

1 V_L (pu) 0.8 1 2 3 4 5 6 7 (b) 1.01

1 f_L (pu) 0.99 1 2 3 4 5 6 7 (c) 1.2

1 (c)

V_DC (pu) 0.8 1 2 3 (d) 4 5 6 7 Time (s)

Figure 4.23: Response of the DFIG based wind-battery RAPS system with an induction pump load: (a) wind speed, (b) voltage on load side (i.e. system voltage), (c) frequency on load side, and (d) DC link voltage. of the RAPS system during the pump load addition are -0.18 pu < ∆v < +0.5 pu and -0.05< ∆f <-0.05 pu respectively. Upon close examination it can be seen that, the step reduction of load (i.e. disconnection of the pump load) at t = 6 s is not seen to be reﬂected adversely on the load side voltage and frequency of the RAPS system.

The DC bus voltage of the RAPS system is shown in Fig. 4.23-(d). The drop in the stator voltage21, causing a reduction in real power delivery to the entire load makes the energy not supplied to the load to return to the DC link and hence the DC link voltage is seen to rise to a high value at t = 4 s. The power sharing between the system components are shown in Fig. 4.24.

21This is due to the high reactive power absorbtion associated with the starting process of the induction motor driven pump load 118

0.5

P_w (pu) 0 1 2 3 4 5 6 7 (a)

0.5

P_b (pu) -0.5 1 2 3 4 5 6 7 (b)

1 0.8

P_L (pu) 0.6 1 2 3 4 5 6 7 (c) Time (s)

Figure 4.24: Power sharing of the DFIG based wind-battery RAPS system with an induction motor driven pump load: (a) wind generator power output, (b) battery storage power and (c) load demand.

4.4.3 Performance of the PMSG-Battery RAPS System

(a) Behaviour of PMSG Based RAPS System Under Wind and Load Step

Changes

The system response of the PMSG based power system under variable wind and load conditions is shown in Fig. 4.25. The corresponding power sharing between the system components is shown in Fig. 4.26. The wind proﬁle under which the system is simulated is shown in Fig. 4.25-(a). It can be seen that the wind velocity is set initially at 12 m/s. At t = 3 s, the wind velocity drops to 9 m/s, then it is increased to 11 m/s at t = 5 s. The load demand is initially set at 0.425 pu. At time t = 4 s, the load is increased to a value of 0.725 pu and reduced back to the original value at t = 6 s as shown in Fig. 4.26-(d). The corresponding simulated behaviour of the load voltage is shown in Fig. 4.25-(b). It can be seen that the voltage of the system is not aﬀected by the wind speed change which occurs at t = 3 s and t = 5 s but 119 a momentary voltage dip (i.e. vL=0.9 pu) occurs at t = 4 s which corresponds to a 0.3 pu load step addition as evident from Fig. 4.26-(d). A sudden voltage rise of

10% is seen to occur at t = 6 s which corresponds to the load step reduction. The

AC voltage is maintained within ±0.5% during normal operation (i.e. during steady state). The behaviour of the operating frequency of the RAPS system is shown in

Fig. 4.25-(c). It can be seen that the frequency stays within 1 ± 0.005 pu during the entire operation regardless of wind or load changes. However, the highest frequency deviations are seen at t = 4 s and t = 6 s which correspond to load step changes. In addition, wind speed changes do not have any effect on the operating frequency as evident from Fig. 4.25-(c). The DC link voltage of the wind energy system is shown in Fig. 4.25-(d) where the DC link voltage magnitude is maintained very close to its rated value. However, small fluctuations of the DC link voltage are visible until t = 3 s. It is seen that the dump load of the system is active during this time period as shown in Fig. 4.26-(c) and the discrete nature of the dump load operation is the cause for the small ripple in the DC link voltage. As expected, DC link voltage variations are seen to occur at t = 3 s, t = 5 s and t = 4 s, t = 6 s which correspond to wind speed and load step changes respectively. At these instances, the active power balance of the DC link cannot be maintained, and this causes slight fluctuations in the DC link voltage. The simulated behaviour of the DC link voltage in the absence of the dump load is shown in Fig. 4.27. It can be seen that until t = 3 s, during which the battery storage stays at its maximum capacity, the excess energy flows into the DC link capacitor causing a rise in the DC link voltage. This shows the importance of having a dump load in the RAPS system in order to regulate the DC link voltage in situations where the excess energy cannot be handled by the battery storage system alone (e.g. situations such as the battery storage system reaches to its maximum capacity). 120

10 V_w (m/s) 5 1 2 3 4 5 6 7 (a)

1.2

V_L (pu) V_L 0.8 1 2 3 4 5 6 7 (b) 1.005

1 f_L (pu) 0.995 1 2 3 4 5 6 7 (c) 1.2

V_DC (pu) V_DC 0.8 1 2 3 4 5 6 7 (d) Time (s)

Figure 4.25: Response of the PMSG based wind-battery RAPS system under wind and load step changes: (a) wind speed, (b) voltage on load side, (c) frequency on load side and (d) DC link voltage.

Initially, the wind speed of 12 m/s generates 0.83 pu of power until the wind velocity changes at t = 3 s as seen in Fig. 4.26-(a). Until t = 4 s, the demand of the system remains at 0.425 pu. Majority of the excess power (Pw − PL), is absorbed by the battery storage up to its maximum capacity of 0.4 pu, and the remaining power is diverted to the dump load to regulate the DC link voltage. The battery storage and the dump load power consumptions are shown in Fig. 4.26-(b) and Fig.

4.26-(c) respectively. However, at t = 3 s, the wind velocity drops to a value of 9 m/s causing a reduction in the generated power to 0.425 pu which is almost equal to the load demand. Therefore the battery storage stays in its idle mode of operation.

However, with an increase in the load that occurs at t = 4 s, a demand-generation mismatch leads to an under-generation scenario, where the deﬁcit power (PL − Pw), is supplied by the battery storage by changing its mode of operation from idling to discharging mode as evident from Fig. 4.26-(b). At t = 5 s, the wind velocity changes 121 from 9 m/s to 11 m/s causing an increment in the wind power output. Again, this leads to a balanced generation-demand condition where the battery storage stays in its idle mode of operation until t = 6 s. When load step reduction occurs at t = 6 s, the demand-generation mismatch shows an over-generation scenario thus changing the mode of operation of the battery storage from idling to the charging mode. The maximum power extraction from the wind in the PMSG based RAPS system is shown in Fig. 4.28. It can be seen that the actual power output from the wind closely follows the reference power (Pw)MPPT for the entire time period except during transients as evident from Fig. 4.28.

0.5 P_w (pu) P_w 0 1 2 3 4 5 6 7 (a) 0.5

0 P_b (pu) -0.5 1 2 3 4 5 6 7 (b) 1

0.5 P_d (pu) 0 1 2 3 4 5 6 7 (c) 0.8

0.6 P_L (pu) 0.4 1 2 3 (d) 4 5 6 7 Time (s)

Figure 4.26: Power sharing of the PMSG based wind-battery RAPS system under wind and load step changes: (a) wind generator power output, (b) battery power, (c) dump load power and (d) load demand.

(b) Behaviour of PMSG Based RAPS System Under Random Varia- tions of Wind and Load Changes

The PMSG based RAPS system is simulated using a random variation in wind 122

2.5

1.5 V_DC (pu) V_DC

0.5

0 1 2 3 4 5 6 7 Time (s)

Figure 4.27: Behaviour of DC link voltage of PMSG without employing dump load under wind and load step changes.

1 Pw Actual

0.9 Pw MPPT

0.8

0.7

0.6

Power (pu) 0.5

0.4

0.3

0.2 1 2 3 4 5 6 7 Time (s)

Figure 4.28: Maximum power extraction from wind in PMSG based wind-battery RAPS system under wind and load step changes. 123 proﬁle as shown in Fig. 4.29. The wind speed varies between 6 m/s and 15 m/s.

The power sharing between the system components is shown in Fig. 4.30 where the initial load demand is set at 0.425 pu. At t = 4 s it is increased to 0.725 pu that is disconnected from the system at t = 6 seconds and the corresponding maximum power extraction from wind is shown in Fig. 4.31.

15 10

V_w (m/s) 5 1 2 3 4 5 6 7 (a) 1.1

V_L (pu) 0.9 1 2 3 (b) 4 5 6 7 1.005

1 f_L (pu) 0.995 1 2 3 (c) 4 5 6 7 1.2

V_DC (pu) 0.8 1 2 3 4 5 6 7 (d) Time (s)

Figure 4.29: Response of the PMSG based wind-battery RAPS system under a realistic wind proﬁle: (a) wind speed, (b) voltage on load side, (c) frequency on load side and (d) DC link voltage.

Driven Pump Load

Behaviour of the PMSG based RAPS system with an induction motor driven pump load is shown in Figs. 4.32 and 4.33. The wind condition under which the system has been simulated is shown in Fig. 4.32-(a) where the wind velocity is set initially at 12 m/s and reduced to 9 m/s at t = 3 s and increases to 11 m/s at t = 4 s. At the beginning, the load demand is set to 0.425 pu and it increases nearly to 0.675 pu (i.e. the capacity of the induction pump load is set to 0.25 pu). The corresponding voltage and frequency on load side is shown in Fig. 4.32-(b) and Fig. 4.32-(c) respectively. 124

1 P_w (pu) P_w 0 1 2 3 4 5 6 7 (a) 0.5

0 P_b (pu) -0.5 1 2 3 4 5 6 7 (b) 1

0.5 P_d (pu) 0 1 2 3 4 5 6 7 (c)

0.6 0.4 P_L (pu) P_L 0.2 1 2 3 (d) 4 5 6 7 Time (s)

Figure 4.30: Power sharing of the PMSG based wind-battery RAPS system under a realistic wind proﬁle: (a) wind power, (b) battery power, (c) dump load power and (d) load demand.

1.1 Pw Actual 1 Pw MPPT

0.9

0.8

0.7

0.6

Power (pu) Power 0.5

0.4

0.3

0.2

0.1 1 2 3 4 5 6 7 Time (s)

Figure 4.31: Maximum power extraction from wind in PMSG based RAPS system under realistic wind proﬁle. (The PwActual is as same as Pw in Fig. 4.30-(a)) 125

As expected, the highest voltage and frequency excursions occur at t = 4 s which correspond to the induction motor driven pump load addition. The corresponding voltage and frequency excursions are within the range of -0.46 pu < ∆v < +0.5 pu and -0.01 pu< ∆f <+0.01 pu respectively. However, the magnitudes of the voltage and frequency are recovered back to their nominal values (i.e. 1 pu) within t = 0.5 s and t = 0.25 s respectively. The second highest voltage and frequency deviation is seen at t = 6 s which corresponds to the disconnection of the pump load from the RAPS system. The DC bus voltage is shown in Fig. 4.32-(d) and recorded its highest variations at t = 3 s and t = 6 s. The DC bus voltage variation of the RAPS system is limited within ± 5% at its rated value. Power sharing between the system components of the RAPS system is depicted in Fig. 4.33.

V_w (ms) 5 1 2 3 4 5 6 7 (a) 1

V_L (pu) 0.5 1 2 3 4 5 6 7 (b) 1.02

1 f_L (pu) 0.98 1 2 3 4 5 6 7 (c) 1.1

V_DC (pu) 0.9 1 2 3 4 5 6 7 (d) Time (s)

Figure 4.32: Response of the PMSG based wind-battery RAPS system with an induction pump load: (a) wind speed, (b) voltage on load side, (c) frequency on load side and (d) DC link voltage. 126

0.5 P_w (pu) 0 1 2 3 4 5 6 7 (a) 0.5

0 P_b (pu) -0.5 1 2 3 4 5 6 7 (b) 1

0.5 P_d (pu) 0 1 2 3 4 5 6 7 (c) 0.8 0.6

P_L (pu) 0.4

1 2 3 (d) 4 5 6 7 Time (s)

Figure 4.33: Power sharing of the PMSG based wind-battery RAPS system with an induction motor driven pump load: (a) wind power, (b) battery power, (c) dump load power and (d) load demand. 127 4.5 Chapter Summary

This chapter has addressed the hybrid operation of diﬀerent types and arrangements of RAPS systems which consist of a wind turbine generator, a battery storage system and a dump load. In the RAPS systems considered, wind turbine generator acts as the main energy source which is responsible for controlling the voltage and frequency.

In order to maintain the power balance of the RAPS systems, a control coordination strategy is formulated which has been taken as the basis to design the controllers for the battery storage system and dump load. In addition, attention has been paid to extract the maximum power from wind thus operating each RAPS system in its optimum power extraction mode. In order to simulate diﬀerent scenarios indicating the operation of wind turbine generator under high and low wind regimes, step changes and highly variable random wind patterns have been considered.

Simulation studies have been carried out to observe the suitability of the proposed RAPS systems together with their respective control strategies (e.g. individual controllers, control coordination etc.) in relation to their voltage and frequency regulation and maximum power extraction capabilities. Based on the results obtained, the following conclusions can be drawn:

• All types of RAPS systems presented in this chapter were able to regulate

the voltage and frequency within acceptable limits during their steady state

operation. However, PMSG based RAPS system exhibited the highest AC side

voltage variation compared to the DFIG counterpart. Wind turbine generator

outputs (i.e. AC voltage and frequency) were seen to be less sensitive to wind

step and load step changes on variations in wind and load demand.

• The power sharing between the system components was implemented in a des-

ignated manner between the battery storage and dump load while capturing 128

optimum wind energy from the wind turbine. The ability to operate the battery

storage system as a load and a source is seen to provide a signiﬁcant ﬂexibility

to the RAPS system in regulating the power balance during over and under

generation scenarios.

• The maximum power extraction method applied for the DFIG based RAPS

system was seen to be an indirect method where the battery storage and dump

load participated in placing optimum torque on the generator shaft. In contrast,

the maximum power extraction from PMSG was seen to be a direct method

where the DC/DC converter (i.e. boost converter) is conﬁgured to extract

the maximum power. The simulated results showed that the MPPT scheme

is working satisfactorily in both RAPS systems while showing some deviations

during transient conditions which were unavoidable. Chapter 5

Application of Hybrid Energy

Storage for Standalone Wind

Energy Conversion Systems

5.1 Introduction

Importance of having an integrated battery storage system for standalone wind energy conversion systems was described in Chapter 4. In general, an energy storage system employed in a standalone wind energy system needs to have both high power and energy density capabilities in order to satisfy both transient conditions (e.g. wind gust) and energy requirements (i.e. demand-generation mismatch) respectively. How- ever, at present, no single energy storage technology meet these requirements. In this regard, hybridisation of two energy storage systems: one with high energy capability and the other with high power density, is seen to provide an improved performance in terms of handling the transients as well as energy requirements of a wind based

RAPS system. As indicated in Chapter 4, battery energy storage systems are well characterised by their high energy density levels. In contrast, supercapacitors are 129 130 recognised by their high power energy density levels. A hybrid energy storage consisting of a supercapacitor and a battery storage system oﬀers competitive advantages

(e.g. lower depth of discharge (DOD) of the battery storage system) compared to the situation where they operate alone and individually in a RAPS system.

This chapter mainly investigates the performance of hybrid energy storage application consisting of a battery energy storage and supercapacitor which can be integrated into standalone wind energy conversion systems. In this regard, two type of wind energy systems: (a) DFIG and (b) PMSG are investigated. Further, two diﬀerent interface arrangements suitable for a hybrid energy storage system are presented. Furthermore, an energy management algorithm (EMA) is proposed for the hybrid operation of the battery storage and supercapacitor. Moreover, such EMA is integrated into the coordinated control approach that govern the entire power ﬂow management of the hybrid RAPS systems.

5.2 Comprehensive Model of Wind-Battery-Supercapacitor

based Remote Area Power Supply System

This section covers the advantages of having a hybrid energy storage system, coordinated control approach and controller design for both DFIG and PMSG based RAPS systems. Energy management algorithm (EMA) for battery-supercapacitor hybrid energy storage is also illustrated.

5.2.1 Importance of Hybrid Energy Storage System in a Standalone

Wind Power Application

A simple arrangement shown in Fig. 5.1 is used to demonstrate the signiﬁcance of integrating hybrid energy storage into a standalone wind power application where the 131

ﬂuctuating current source represents the variable wind power output which consists of two components. The main steady component corresponds to DC current with a magnitude of A=50 A and a ripple component of B=7.5 A at f=20 Hz that is superimposed on the main component to simulate the ﬂuctuating power component of the wind turbine generator. The battery storage system is represented by the series resistor, rb, with a constant DC voltage source vb of 25 V. The supercapacitor is represented by a capacitor of which the voltage is vcap, connected in series with resistor, rcap.

Battery Storage Supercapacitor

r b rcap + Is=A+B sin (2 πft) A Vb Vcap Variable current source _

Figure 5.1: Simpliﬁed model of a power system with hybrid energy storage.

The simulated behaviour of the system is shown in Fig. 5.2. It can be seen that the ﬂuctuating component of the current is absorbed by the supercapacitor whereas the battery storage system absorbs the steady component of 50 A. The ﬂuctuation of the battery current is limited to ± 1A thus ensuring safe operation of the battery storage system. 132

80 I s 70 I b I 60 cap

30 Current (A) Current 20

-10 1 1.5 2 2.5 3 3.5 4 4.5 5 Time (s) Figure 5.2: Current sharing between battery storage and supercapacitor.

5.2.2 A Coordinated Control Approach for Hybrid Energy Storage

based RAPS Systems

The control coordinated approach described in Chapter 4 in Section 4.3.2 is further

modiﬁed with a view to emphasise the contribution made by the supercapacitor.

The decision making process associated with the control coordination algorithm of

wind-battery-supercapacitor based RAPS system is shown in Fig. 5.3. During over-

generation conditions where the power output from the wind turbine generator Pw,

is greater than the load demand PL, the hybrid energy storage (i.e. battery storage

and supercapacitor) should absorb the excess power (Pw − PL), and shared between the battery storage system and supercapacitor according to the energy management

algorithm (EMA) presented in Section 5.2.31. If the capacities of the energy storage

1According to the proposed energy management algorithm, the supercapacitor should be able to absorb the ripple or high frequency power component of demand-generation mismatch leaving the steady component for the battery storage system. 133

systems reach their maximum limits, the dump load needs to absorb the excess power.

However, if the dump load power Pd, reaches its maximum rating (Pd)max, the pitch angle control of the wind turbine generator has to be activated in order to reduce

the power output of the wind turbine generator. During under-generation situations

where the power output of the wind turbine generator is less than the load demand,

i.e. (Pw − PL) < 0, it is assumed that the hybrid energy storage is able to supply the

required power deﬁcit, PL−Pw. During emergency situations, such as no power output from wind turbine generator due to wind speed being below cut-in level or above cut-

out level, a load shedding scheme can be implemented which will be discussed in

Chapter 6. In addition, the controller design criteria elaborated in Chapter 4 in

Section 4.3.3 is considered when designing the controllers of the DFIG and PMSG

based RAPS systems.

5.2.3 Energy Management Algorithm (EMA)

Depending on the objectives to be achieved (e.g. minimisation of the demand-

generation mismatch, ensuring safe operation of the energy storage systems2 etc.), several types of energy management algorithms can be designed and implemented for the hybrid energy storage system. As indicated in Section 5.2.2, in the present context, the energy management algorithm is implemented encompassing the battery storage and supercapacitor with a view to achieve the following objectives: • to help maintain the power balance of the RAPS system,

• to operate the wind turbine generator based on the maximum power point

tracking algorithm and

• to improve the performance of the battery storage system by reducing ripple

current and high rate of depth of discharge (DOD).

2Avoiding the operation of the battery storage system to high depth of discharges. 134

Wind Power, Pw

vcut-in

Yes

No Hybrid Energy Storage No Load Pw-P L>0 System - Battery Pw+Pb+Psc -P L>0 Shedding Discharging Yes Yes

Hybrid Energy Storage System - Battery Charging

Energy Management Algorithm Battery Discharging Battery - Low Frequency Supercapacitor - High Power Component Frequency Power Component

Battery Charging

No Dump Load No P <(P ) P < (P ) Pitch b b max "ON" d d max Regulation

Yes Yes

Frequency Control

Figure 5.3: Control coordination of a wind-battery-supercapacitor based hybrid RAPS system. 135

The ﬁrst objective is achieved by generating the input signal for the controllers of hybrid energy storage system using the demand-generation mismatch (Pw − PL) of the RAPS system. To realise the second objective, the wind power output Pw, of the demand-generation mismatch is estimated using the optimal wind power (Pw)opt. By controlling the power ﬂow into/out of the battery storage system and supercapacitor according to the maximum power tracking algorithm, it is possible to impose an appropriate torque on the wind turbine generator3 to extract the maximum power from wind. Under heavy DOD levels, the battery storage reaches quick charge regulation as shown in Fig. 5.4. Therefore, the third objective is related to managing the DOD levels of the battery storage system which is achieved by separating the demand-generation mismatch into two frequency components through a high-pass

τs ﬁlter, 1+τs as shown in Fig. 5.5. The demand-generation mismatch is explained in two frequency domains which can be given by (5.1). The high frequency power component of the demand-generation mismatch Phf , is used to estimate the reference current of supercapacitor, (ic)ref . Contrarily, the low frequency power component of the demand-generation mismatch Plf , is used to generate the reference current of the battery storage (ib)ref . The operating frequency range of each type of energy storage systems is diagrammatically shown in Fig. 5.64.

∆PwL = Phf + Plf (5.1)

where, Phf is the high frequency component of the demand generation mismatch and Phf is the low frequency component of the demand-generation mismatch. Diﬀerent power electronic conﬁgurations can be employed to interface the hybrid energy storage system into wind based RAPS systems. As an example, both a battery

3In this case wind turbine generators are referred to DFIG and PMSG and the methods of extracting the MPPT have been described in Chapter 4. 4 fc represents the cut-oﬀ frequency of the ﬁlter circuit. Typically fc is 0.5 Hz [36]. 136

Battery voltage, vb

2C 1C 0.1C Charge Regulation 3C

Capacity Ah

Figure 5.4: Battery voltage under diﬀerent discharge rates.

High pass filter (P w)ref P τs + - τs+1 Phf=(Psc )ref

PL + - P (P ) lf = b ref Figure 5.5: Estimation of reference power for battery storage and supercapacitor.

Plf Phf

Battery storage power Supercapacitor power f c Figure 5.6: Operating frequency ranges of the energy storage systems: supercapacitor and battery storage system. 137 storage and a supercapacitor can be connected to the DC bus of the power electronic converter associated with the wind energy system using appropriate DC/DC converter

(e.g. buck converter, boost converter or bi-directional buck-boost converter). Another alternative arrangement is that one of the energy storage systems is connected to the

DC bus of the wind generator system using a DC/DC converter, while the other is located on the AC load side5 by means of an inverter. To examine the suitability of these topologies for two wind generating schemes: (a) DFIG and (b) PMSG are examined with the following conﬁgurations:

• DFIG with battery storage system, supercapacitor and dump load where the

battery storage system is connected to an inverter (i.e. on the AC load side)

and the supercapacitor is connected to the DC bus using a DC/DC converter.

• PMSG with battery storage system, supercapacitor and dump load where both,

the battery storage system and supercapacitor are connected to the DC bus of

the wind energy system using two parallel connected DC/DC converters, one

for each storage system.

Further information regarding the conﬁguration, control strategies and their respective design aspects are discussed below.

Considering the operating frequency range as the criterion, two diﬀerent models of a supercapacitor can be depicted as in Fig. 5.7. The ﬁrst model indicated in Fig. 5.7-

(a)6 is known as detailed model which includes the non-linear Faraday capacitance.

The low frequency domain model which can be used under power system operating frequency range7 presented in Fig. 5.7-(b) is employed in the current work.

5i.e. AC side of RAPS system. 6 R1 represents the equivalent series resistance, Ca is the anode dielectric capacitance, Ra is equivalent parallel resistance of dielectric materials, CF is Cathode Faraday capacitance and Zf is the faraday impedance which includes the resistance of charge movement which is not a pure resistance. 7In this case, the power system operating frequency is 50 Hz.

138

R1 Ca Cc R1 Ca

Ra Zf

(a) (b)

Figure 5.7: Equivalent circuits of supercapacitor (a) high frequency model and (b) low frequency model.

In real life applications, the operation of a supercapacitor needs to satisfy the conditions given in (5.2), (5.3) and (5.4). The ﬁrst condition given by (5.2) emphasis the safe operating voltage of a supercapacitor which is usually indicated in the manufacturer data sheet. The second condition given by (5.3) indicates the maximum possible peak current of a supercapacitor. The third condition presented in (5.4) deﬁnes the maximum allowable power from the supercapacitor during its operation.

(vsc)min < vsc < (vsc)max (5.2)

0.5Csupvsc (Ic)pk = (5.3) C(ESRdc) + 1 dv (P ) = ± C v | sc | (5.4) sc max sup sc dt max

where, Csup is capacitance value of the supercapacitor, (vsc)max,(vsc)min are maximum and minimum operating voltages of supercapacitor respectively, ESRDC is equivalent series resistor of the supercapacitor, Csup is supercapacitance (Psc)max is 139

dvsc maximum power rating of the capacitor and | dt |max is the maximum rate of change of voltage across the supercapacitor.

Similar to the case of the battery storage system, the estimation of size of the supercapacitor is extremely design speciﬁc for a given site. However, in the present case, the value of the capacitance of the supercapacitor is estimated considering the worst case scenario where it is able to supply energy subjected to the wind energy inverter constrains over a certain time period t, as given by (5.5)-(5.6). Furthermore, the size of the battery storage system is estimated using the method discussed in

Section 4.3.4 of Chapter 4.

2Esc Csup = 2 2 (5.5) ((vsc)max) − ((vsc)min)

  (smax(PDFIG)rated)t for DFIG Esc = (5.6)  ((PPMSG)rated − (Pb)rated)t for PMSG

smax is the maximum allowable operating slip of DFIG, (PDFIG)rated is the rated capacity of the DFIG, (PPMSG)rated is the rated capacity of the PMSG and t is the time duration over which the supercapacitor is able to supply power.

5.2.4 Application of Hybrid Energy Storage in DFIG based RAPS

System

The conﬁguration of the DFIG based RAPS system consisting of the hybrid energy storage and dump load with main loads is shown in Fig. 5.8. The battery storage is connected to the load side using an inverter whereas the supercapacitor is interfaced to the DC bus of the back-to-back converter system by means of a bi-directional buck 140 boost converter. The operation of the entire RAPS system is designed according to the coordinated control approach given in Fig. 5.3. As pointed out in Section 5.2.3, the energy management algorithm developed for the hybrid energy storage system based on the operating frequency of battery storage and supercapacitor as is shown in Fig. 5.5 is taken as the basis for the development of the respective controllers associated with the power electronic interfaces. The control strategies that are used to control the back-to-back converter system of the DFIG are the same as which were illustrated in Section 3.2.4 of Chapter 3. The control strategies associated with the battery storage system, supercapacitor and dump load are illustrated in the following sections.

DFIG

RSC LSC +

Dump _ load Battery storage Main loads v,f vdc DC-DC Plf converter ∆P

Supercapacitor

P hf

Figure 5.8: Hybrid energy storage in a DFIG based RAPS system.

(a) Battery Storage Controller

As stated in Section 5.2.3, the battery storage is used to meet the steady component of the demand-generation mismatch thus avoiding higher depths of discharging.

As shown in Fig. 5.8 an inverter is used to interface the battery storage system with the RAPS system. The inverter control associated with the battery storage is developed by following the EMA and depicted in Fig. 5.9. The reference current of the 141

8 battery (ib)ref , is generated considering the low frequency component of the demand-

9 generation mismatch, Plf as given in (5.7). The inverter is operated at unity power factor by setting (iqs)ref equal to zero. The PI controllers associated with control schemes shown in Fig. 5.9 are tuned using the ZieglerNichols method as given in [92].

The derivation of the control algorithm associated with the inverter shown in Fig.

5.9 is given in Appendix B.

(ib)ref = (Plf − Pb)(kp + ki/s) (5.7)

where Pb is actual battery power and kp, ki are proportional and integral gains of the PI controller. ω Li qs

(P ) b ref (i b )ref + dq + PI + PI - - - + abc

(P ) i vds switching b actual ds P ϑ signals to inverter PLL W M

(i qs )ref dq + PI - - + abc

iqs Li ds ω

Figure 5.9: Inverter control of the battery storage system for DFIG wind-hybrid energy storage based RAPS system.

(b) Supercapacitor Controller

As shown in Fig. 5.8, the supercapacitor is connected to the DC bus of the back- to-back converter system using a bi-directional buck-boost converter system. The

10 high frequency component of the demand-generation mismatch Phf is met by the

8 (ib)ref is used to represents the (ids)ref in Fig. 5.9. 9 Plf =(Pb)ref where (Pb)ref is the reference battery power shown in Fig. 5.9. 10 Phf =(Psc)ref where (Psc)ref is the reference supercapacitor power shown in Fig. 5.10. 142

supercapacitor where the corresponding reference current (isc)ref is estimated using (5.8). The adopted control strategy for the supercapacitor is illustrated in Fig. 5.10.

Phf (isc)ref = (5.8) vsc

Comparator To bi-directional (Psc )ref (Isc )ref buck-boost converter ÷ + - PI + - Limiter

vsc Isc Triangular carrier waveform

Figure 5.10: Control strategy for supercapacitor in a hybrid energy storage of a DFIG based RAPS system.

An arrangement similar to the dump load discussed in Chapter 3 in Section 3.4.2-

(a) is employed in the current work. However, in the present context, the operation of

dump load is enabled when the battery storage reaches its maximum capacity. There-

fore, the condition under which the dump load operation is enabled is given by (5.9).

It is to be noted that the contribution of the supercapacior power is not considered in

the process of estimating the dump load power considering its performance in short

term window and high frequency domain.   Pd, (Pw) + (Pb)max − PL > 0 Pd = (5.9)  0, otherwise

where, Pw is generated wind power output from DFIG and (Pb)max is maximum capacity of the battery storage. 143 5.2.5 Application of Hybrid Energy Storage in PMSG based RAPS

System)

A PMSG based RAPS conﬁguration with hybrid energy storage is shown Fig. 5.11. In this case, the hybrid energy storage consists of a battery storage and a supercapacitor connected to the DC bus of the wind energy system using bi-directional buck-boost converters: C-1 and C-2 respectively. The power ﬂow between all system components is designed following the coordinated control approach discussed in Section 5.2.2. In addition, the operation of the hybrid energy storage system is designed with a view to achieve the objectives of the EMA described in Section 5.2.3. By considering the issues11 associated with integrating several components12 into the DC bus, in this case, the dump load is connected to the load side (or AC bus) of the RAPS system. The control schemes employed for the PMSG inverter and boost converter remains same as described in Chapter 3, Section 3.3.213. Furthermore, the dump load arrangement and its operating principle remain the same as given in Section 5.2.4.

Therefore, in this section emphasis is directed towards the controller design of hybrid energy storage system only.

The energy management algorithm related to the battery storage and supercapacitor is depicted in Fig. 5.12. In order to ensure the extraction of maximum power from wind, the demand-generation mismatch of the system given by ∆P is estimated by considering the maximum power output of the wind power, (Pw)ref . In order to operate the safer battery operation, the demand-generation is splitted into two frequency ranges using a high pass ﬁlter. The high frequency power component of the demand-generation mismatch Php, is used to estimate the reference capacitor current

11The dynamic operation of several components will lead to instability in DC bus voltage. 12In this case, it refers to hybrid energy storage and dump load. 13Inverter control is used to regulate the voltage and frequency whereas the boost converter is used to regulate the DC bus voltage. 144 Full bridge PMSG rectifier DC/DC converter-1 Inverter G

Wind turbine (vdc ) C -1 v, f + Battery storage ∆ p _ Dump load Main loads

Plf C -2

Supecapacitor

Phf Figure 5.11: Hybrid energy storage in a PMSG based RAPS system.

(ic)ref , which is subsequently compared with the actual supercapacitor current ic, to generate the switching signal for the converter C-1. Contrarily, the low frequency power component of the demand-generation mismatch Plf , is used to estimate the reference battery current (ib)ref , which is compared with the actual battery current ib, to generate the switching signal for the converter C-2.

Comparator (P P τs (i ) To C-2 w)ref + c ref τs ÷ + PI - +1 - + - Limiter i PL + - vc c Triangular carrier waveform ÷ (ib)ref vc To C-1 + PI + vb - - Limiter Comparator ib

Triangular carrier waveform

Figure 5.12: Energy management scheme for a hybrid energy storage in a PMSG based RAPS system. 145 5.3 Simulation Results and Discussions

Operation of hybrid energy storage systems applied to DFIG and PMSG based RAPS systems have been investigated under variable wind and load conditions and their responses is observed in relation to voltage and frequency regulation bandwidth capabilities. In addition, the behaviour of the battery storage system is observed with and without the supercapacitor as a storage element. Also, the maximum power extraction capability of the wind turbine generators are examined. The parameters associated with each type of RAPS system is listed in Appendix B.

5.3.1 Performance of the Hybrid Energy Storage in a DFIG based

RAPS System

The system response of the DFIG based RAPS system is shown in Fig. 5.13 whereas

Fig. 5.14 illustrates the power sharing between diﬀerent system components. The wind condition under which the system has been simulated is shown in Fig. 5.13-(a).

It can be seen that the wind velocity is set initially at 12 m/s. At t = 3 s, the wind velocity drops to 9 m/s, and it increases to 11 m/s at t = 7 s. As seen in Fig. 5.14-(d), initial load demand is set at 0.425 pu and then it is increased to a value of 0.7 pu at t = 4 s and the added additional load (i.e. 0.275 pu) is disconnected from the system at t = 6 s. The voltage on load side is shown in Fig. 5.13-(b) which is not seen to be aﬀected by the wind speed or resistive load step changes. The load voltage of the system stays within ±2% of its rated value throughout the operation. The operating frequency of the RAPS system is shown in Fig. 5.13-(c). As anticipated, the operating frequency is closely regulated at its rated value of 1.0 pu and is not seen to be inﬂuenced by the wind speed or load step changes. Furthermore, it can be seen that the frequency of the system is maintained within 0.2% of its rated value. The 146

DC link voltage of the DFIG is depicted in Fig. 5.13-(d) which is well regulated at its rated value throughout the operation except during load step changes. However, the highest DC link voltage variations are seen to occur at t = 4 s and t = 6 s which correspond to the load step changes as evident from Fig. 5.13-(d). It can be noted that, the supercapacitor quickly changes its direction of power ﬂow as evident from

Fig. 5.14-(b) which causes the ﬂuctuations in the DC link voltage. However, even during such transient conditions, the DC link voltage variation is limited to within ±

10% of its rated value.

10 V_w (m/s) 5 1 2 3 4 5 6 7 8 9 10 (a) 1.05

1 V_L (pu) 0.95 1 2 3 4 5 6 7 8 9 10 (b)

1.005

1 f_L (pu) 0.995 1 2 3 4 5 6 7 8 9 10 (c) 1.2

1 V_DC (pu) 0.8 1 2 3 4 5 6 7 8 9 10 (d) Time (s)

Figure 5.13: Response of the DFIG based wind-hybrid energy storage RAPS system during variable wind and load conditions: (a) wind speed, (b) voltage on load side, (c) frequency on load side, and (d) DC link voltage.

The DFIG power output is shown in Fig. 5.14-(a). For simulation purposes, initially the slip of the wind turbine is set to s = - 0.1 which corresponds to super 147 synchronous mode of operation. According to the wind turbine characteristics14, the corresponding maximum power output of the wind generator is 0.73 pu at the speed of 1.2 pu for a wind speed of 11 m/s. As shown in Fig. 5.14-(a), the power output of the DFIG is seen to rise to a value of 0.69 pu at t = 3 s and the corresponding load demand is at 0.425 pu. This situation simulates an over-generation condition where the excess power is shared between the battery storage system and supercapacitor as as evident from Fig. 5.14-(b). It can be seen that the supercapacitor responds to the fast varying power variations while the battery absorbs the slow varying power variations of the demand-generation mismatch. Also, the supercapacitor responds quickly to load step changes which occur at t = 4 s and t = 6 s avoiding high rates of

DOD of the battery storage system. Furthermore, the battery storage system reaches its full capacity at t = 2 s leading to the operation of dump load which absorbs the additional power as evident from Fig. 5.14-(c). The battery storage system operates in its charging mode until load step addition which occurs at t = 4 s leading to a change of its mode of operation from charging to discharging. However, the rate of discharge is reduced after the load step reduction that occurs at t = 6 s. The maximum power point tracking behaviour of the DFIG is shown in Fig. 5.15. It can be seen that DFIG closely follows the MPPT curve except during the transient conditions that occur at t = 4 s and t = 6 s.

The battery current for the case with no supercapacitor is shown in Fig. 5.16.

The battery current consists of high frequency ﬂuctuating component and exhibits steep DOD during load step changes. The current levels of the hybrid energy storage with the supercapacitor is shown in Fig. 5.17. It can be seen that the battery storage system has near ripple free current with low value of DOD rate at the time of load step changes.

14Refer to Appendix B for further information. 148 1

0.5 P_w (pu) 0 1 2 3 4 5 6 7 8 9 10 (a)

0.5

0 P P c b P_h (pu) -0.5 1 2 3 4 5 6 7 8 9 10 (b) 0.02

0 P_d(pu) -0.02 1 2 3 4 (c) 5 6 7 8 9 10

0.5 P_L (pu) 0 1 2 3 4 5 6 7 8 9 10 (d) Time (s)

Figure 5.14: Power sharing of the DFIG based wind-hybrid energy storage RAPS system during variable wind and load conditions: (a) wind power, (b) hybrid energy storage power (i.e. battery power, Pb and supercapacitor Psc power (c) dump load power and (d) load demand.

1 (P w)actual 0.9 (P w)MPPT 0.8

0.7

0.6

0.5

Power (pu) 0.4

0.3

0.2

0.1

0 1 2 3 4 5 6 7 8 9 10 Time (s)

Figure 5.15: MPPT tracking capability of the DFIG in a RAPS system with energy storage integrated. 149

400

300

200

100

0 Current (A) Current -100

-200

-300

-400 1 2 3 4 5 6 7 8 9 10 Time (s)

Figure 5.16: Battery current for the case with no supercapacitor of the DFIG based RAPS system with hybrid energy storage integrated.

The supercapacitor current consists mostly of high frequency component and seen

to respond quickly to the transients. However, the safety feature integrated with the

controller of the supercapacitor limits its current to 300 A15.

The frequency spectrum of the battery current and the supercapacitor current are given Fig. 5.18 and Fig. 5.19 respectively. The battery current is seen to be free from ripples where the supercapacitor current consists of considerable amount of high frequencies16 which ensures the safe operation of the battery storage.

15The maximum peak current of the supercapacitor for 1 sec can be given by 0.5CV , where C(ESRDC )+1 ESRDC is the equivalent series resistor [93]. Refer to Appendix B. 16In the present study, the battery storage is used to respond the power ﬂuctuations below 0.5 Hz whereas the high frequency power variations are managed through supercapacitor. 150

Figure 5.17: Currents of the hybrid energy storage: battery current, ib and supercapacitor current, ic of DFIG based RAPS system.

400

350

300

250

200

150 Battery Current (A) Battery Current 100

0 0 20 40 60 80 100 Frequency (Hz)

Figure 5.18: Frequency spectrum of the battery storage system of a hybrid energy storage in a DFIG based RAPS system. 151 100

20 Supercapacitor (A) Current

0 0 20 40 60 80 100 Frequency (Hz)

Figure 5.19: Frequency spectrum of the supercapacitor of hybrid energy in a DFIG based storage RAPS system.

5.3.2 Performance of the PMSG-Hybrid Energy Storage RAPS Sys-

tem

The entire PMSG based RAPS system with hybrid energy storage interagted is simulated under variable wind and load conditions. Fig. 5.20 shows the voltage and frequency behaviour of the RAPS system whereas Fig. 5.21 shows the power sharing between different system components. The wind conditions under which the system has been simulated is shown in Fig. 5.20-(a). It can be seen that the wind velocity is set initially to 12 m/s. At t = 3 s, the wind velocity drops to 9 m/s, which then increases to 10 m/s at t = 5 s. The load demand is initially set at 0.4 pu. At t = 4 s, the load is increased to a value of 0.75 pu and the same load is disconnected from the system at t = 6 s as shown in Fig. 5.21-(d). The AC voltage at the point of common coupling is shown in Fig. 5.20-(b). It can be seen that the load side voltage shows slight fluctuations at t = 4 s and t = 6 s which correspond to load step changes. The highest voltage variation is seen to occur due to load step down at t = 6 s which is limited to within +10% of its rated value. Also, it can be seen that wind changes have 152 minimal influence on the load side voltage. The operating frequency of the system is regulated within 1± 0.25% pu with minor fluctuations due to load step changes as shown in Fig. 5.20-(c). The DC bus voltage is shown in Fig. 5.20-(d) which is regulated well at its rated value.

V_w (m/s) V_w 5 1 2 3 (a) 4 5 6 7

1.1 1

V_L (pu) V_L 0.9 1 2 3 (b) 4 5 6 7 1.005

f_L(pu) 0.995 1 2 3 4 5 6 7 (c) 1.1

V_DC (pu) V_DC 0.9 1 2 3 4 5 6 7 (d) Time (s)

Figure 5.20: Response of the PMSG based wind-hybrid energy storage RAPS system under variable wind and load conditions. (a) wind Speed, (b) voltage on load side, (c) frequency on load side, and (d) DC link voltage.

The wind power variation of the system is shown in Fig. 5.21-(a). According to the wind turbine characteristics, the corresponding maximum power output of the wind generator is 0.83 pu at rated wind speed of 12 m/s17. Until t = 3 s, the power output of the PMSG stays at 0.83 pu and during this time period, the load active power demand is set to 0.4 pu as depicted in Fig. 5.21-(e). This simulates an over- generation situation where the additional power given by (Pw −PL) is shared between the hybrid energy storage and dump load. However, the power sharing between hybrid energy storage units occurs according to the energy management algorithm discussed in Section 5.2.3. The battery storage power is shown in Fig. 5.21-(b) and it is seen at

17Refer to Appendix B for further information. 153 t = 3 s, the battery reaches its full capacity whereas the supercapacitor absorbs the high ﬂuctuating power component of demand-generation mismatch as shown in Fig.

5.21-(c). When the battery storage reaches its full capacity, the excess slow varying power component is absorbed by the dump load as shown in Fig. 5.21-(d). At t = 3 s, the wind speed reduces to 9 m/s thus lowering wind power output to nearly 0.375 pu as depicted in Fig. 5.21-(a). At this time, the RAPS system experiences an under- generation scenario, where the deﬁcit power, (PL-Pw) is supplied through the battery storage. At t = 3 s, the dump load operation is disabled as shown in 5.21-(d). Also, the sudden wind speed changes cause rapid variations of the wind power output which is seen to be absorbed by the supercapacitor as shown in Fig. 5.21-(c). At t = 4 s, the load is increased by 0.3 pu (i.e. resistive component) and RAPS system further experiences an under-generation condition. In this situation, the battery storage increases its discharge rate. After t = 5 s, the wind speed increases to 10 m/s thus increasing the power output from the wind generator. However, the RAPS system still experiences an under-generation condition where the power deﬁcit is supplied through the battery storage. With this load reduction which occurs at t = 6 s, the system experiences an over-generation condition causing the battery storage to move from discharging to charging mode of operation to maintain the power balance of the

RAPS system. Throughout the operation, the supercapacitor absorbs the fast varying power component of demand-generation mismatch during transient conditions that occur due to wind and load step changes as evident from Fig. 5.21-(c).

To examine the eﬀectiveness of integrating an hybrid energy storage into a PMSG based RAPS system, a comparative study has been carried out in relation to the battery storage current. The behaviour of the battery current for the case with no supercapacitor is shown in Fig. 5.23. It can be seen that the battery storage current consists of a high frequency component which is not favourable. In addition, high 154

1 ) u p ( 0.5 w _ P 0 1 2 3 4 5 6 7 (a) 0.5 ) u p ( 0 b _ P -0.5 1 2 3 4 5 6 7 (b) 0.5 ) u p (

c 0 _ P -0.5 1 2 3 4 5 6 7 (c) 0.2 ) u p ( 0.1 d _ P 0 1 2 3 (d) 4 5 6 7

) 1 u p (

L 0.5 _ P 0 1 2 3 (e) 4 5 6 7 Time (s) Figure 5.21: Power sharing of the PMSG based wind-hybrid energy storage RAPS system under variable wind and load conditions. (a) wind Power, (b) battery power, (c) supercapacitor power (d) dump load power and (e) load demand.

Pw Actual

Pw MPPT 0.9

0.8

0.7 ) u p (

r 0.6 e w o P 0.5

0.4

0.3

0.2 1 2 3 4 5 6 7 Time (s)

Figure 5.22: Maximum power extraction from wind in the PMSG based RAPS system with hybrid energy storage integrated. 155 depth of discharge rates which occur during transient conditions including wind and load step changes will further cause damage to the battery storage system.

200

150

100

50 ) A (

t n

e 0 r r u C -50

-100

-150

-200 1 2 3 4 5 6 7 Time (s)

Figure 5.23: Battery current for the case with no supercapacitor in the PMSG based RAPS system with hybrid energy storage integrated.

The battery storage current for the case with supercapacitor is shown in Fig 5.24.

It is clearly visible that the high frequency component (i.e. 0.5 Hz) is absorbed by the supercapacitor and provides a smoother transition from one operational mode to another with lower depth of discharge for the battery storage. The maximum power extracted from wind is shown in Fig. 5.22 from which it can be seen that the PMSG runs on its maximum power extraction mode of operation throughout its entire operation. 156

Figure 5.24: Currents of the hybrid energy storage: battery current, ib and supercapacitor current, ic of the PMSG based RAPS system with hybrid energy storage integrated. 157 5.4 Chapter Summary

As a continuation of Chapter 4, this chapter has addressed the benefits of adding a supercapacitor to a battery storage system in a wind based hybrid RAPS system. In this regard, an energy management algorithm (EMA) has been established between the battery storage system and supercapacitor to operate both energy storage systems in a designated manner. In addition, to coordinate the power flow between the system components, a control coordination strategy was implemented by incorporating the aforementioned EMA. Moreover, two different connection topologies for hybrid energy storages were introduced. In the DFIG based RAPS system, the battery energy storage and supercapacitor were connected to the load side using an inverter and DC bus of the DFIG using a DC/DC converter system respectively. For PMSG based

RAPS system, the battery storage system and supercapacitor were integrated using parallel connected DC/DC conveners.

Simulation studies have been conducted in order to test the suitability of the proposed schemes in relevance to the RAPS applications. Based on the simulated behaviour, following conclusions can be drawn:

• Both types of hybrid RAPS systems (i.e. DFIG based and PMSG based) are

able to regulate the voltage and frequency on the AC side within tight limits. In

this regard, the control coordination strategy discussed in Chapter 4 is further

modiﬁed by integrating the EMA that was developed for hybrid energy storage

system. In addition, individual controllers for the battery storage system and

supercapacitor were developed considering the concept proposed in EMA.

• The algorithm to satisfy demand-generation mismatch has been implemented

using EMA where, the demand-generation mismatch of the RAPS systems was

split in the two frequency operating regions using a high pass frequency ﬁlter. 158

The high frequency power component was managed through the supercapacitor

while leaving the low frequency power component for the battery storage system.

• In both RAPS systems (i.e. DFIG and PMSG), the maximum power extrac-

tion has been achieved by coordinating the operation of hybrid energy storage

system taking the maximum power extraction algorithm as one of their inputs.

• The reduction of the rate of DODs has been achieved by operating the super-

capacitor to respond the high frequency power component associated with the

demand-generation mismatch, thus avoiding high rate of DOD on the battery

energy storage system.

• It was revealed that the supercapacitor responded to the fast varying power

component of demand-generation mismatch while leaving the slow varying or

steady component which was handled via the battery storage system. Chapter 6

Wind-Diesel Hybrid RAPS

Systems

6.1 Introduction

The application of energy storage systems containing battery banks and supercapacitors for wind energy systems were investigated in Chapters 4 and 5 respectively.

However, energy storage systems are not able to provide reactive power support unless they are integrated through inverter systems. Moreover, due to low or no inertia contribution, energy storage systems do not participate in damping the frequency

ﬂuctuations in RAPS systems. In addition, energy storage systems are not able to provide power for long periods due to capacity limitations. While this is the case, a diesel generator system in such a RAPS system is able to provide both reactive and active power support together with inertia contribution towards the entire RAPS system. However, diesel generators need careful consideration with regard to their operational performance such as avoidance of low load factor operation and synchronisation with other generating schemes. The necessity for integrating an energy storage

159 160 system to such a wind-diesel system is mainly influenced by the penetration level1 of the wind power in the RAPS system. In addition, an energy storage system integrated with a wind-diesel RAPS system can provide a means for dealing with, or at least reducing the adverse operational characteristics of diesel generating systems such as unacceptable start/stop operation or minimisation of the transient effects caused by load or wind profiles.

The work presented in this chapter gives emphasis to modelling aspects of a diesel generating system for two diﬀerent wind turbine technologies: DFIG and PMSG, where both types of wind generators are intergated with a battery storage system.

The hybrid operation of such a RAPS system can be classiﬁed into three operating modes, namely: (a) wind only (WO), (b) wind-diesel (WD) and (c) diesel only (DO).

In WO mode, the diesel generator does not participate in supplying any active power to the load whereas the wind turbine generator and battery storage system are used to satisfy the load demand. In WD mode, the wind turbine, diesel generator and possibly together with battery storage system are able to meet the load demand; whereas only the diesel generator and battery storage system are used to supply the load demand during the DO mode. The operation of wind-diesel RAPS systems are investigated by employing their corresponding linearised and detailed models. In this regard, every system component of the linearised model is represented as a ﬁrst order transfer function whereas the detailed model consists of non-linear high order mathematical models of each system component followed by their respective control schemes. 1Typically a high penetrated wind based RAPS system needs an energy storage to regulate the power ﬂuctuations 161 6.2 Linearised Model of the Wind-Diesel and Energy Storage

System

Linearised model of the wind-diesel-energy storage is shown in Fig. 6.1. As indicated in Fig. 6.1, all system components including wind generator, diesel generator, energy storage system and load demand are deﬁned as ﬁrst order transfer functions, as given in (6.1)-(6.4) respectively. Energy storage system and diesel generator system are considered as controlled energy sources which are managed employing PI controllers to determine the power sharing. The control method described in Chapter 4 in Section

4.4.1 is employed in the current work. The time constants associated with each system component is given in Table 6.1.

KWTG ∆PWTG GWTG(s) = = (6.1) 1 + sTWTG ∆PW KDE ∆PDE GDE(s) = = (6.2) 1 + sTDE ∆f KESS ∆PESS GESS(s) = = (6.3) 1 + sTESS ∆f ∆f 1 GSYS(s) = = (6.4) ∆pe D + sM

where, PW is mechanical power input of the WTG, PWTG is electrical power output of the wind generator, PDE is diesel generator output, PESS is the power output of the energy storage system, M is equivalent inertia constant, D is damping constant of the system, f is frequency of system and pe is the demand-generation mismatch. 162

PI KB

Battery Storage System KESS /(1+sT ESS )

∆Pe ∆f vwind K /(1+ sT ) WTG WTG 1/(Ms+D)

Wind Turbine Wind Generator Pref

Diesel Generator KDE /(1+sT DE )

PI KDE

Figure 6.1: The linearised block diagram of the proposed RAPS system.

Table 6.1: Transfer function parameters of wind generator, diesel generator energy storage and load demand

wind turbine generator KWTG = 1 TWTG = 1.5s

diesel generator system KDE = 1 TDE = 2s

battery storage system KESS = 1 TESS = 0.01s load demand D = 0.012 M = 0.012s 163 6.3 Modelling Aspects of a Diesel Generating System and its

Diﬀerent Operating Modes for a Remote Power Appli-

cation

6.3.1 Importance of Diesel Generator System in a RAPS System

In general, wind based RAPS systems inherit low inertia, low damping and poor reactive power capability. To address these issues, a diesel generator system can be incorporated into a RAPS system. However, the operation of a wind-diesel RAPS system may need to pay consideration to following aspects to ensure eﬀective utilisation of wind and diesel systems:

• minimisation of the number of start/stop operations of diesel generator that

can arise due to time varying load and wind proﬁles,

• avoidance of running of diesel generator at low-load,

• active and reactive power coordination between sources and sinks and

• maximum power extraction from wind.

In the present context, the main function of the diesel generator in a high penetrated wind based RAPS system is to satisfy the demand-generation mismatch by performing as a back-up generating system. However, variability of wind and load profiles may lead to frequent start/stop operations of the diesel generator. Further- more, the operation of the diesel generator at low load conditions is not desirable which leads to increased wear and maintenance causing reduced life span. Moreover, the specific fuel consumption of the diesel generator significantly increases with reduced load as evident from Fig. 6.2 [94]. In real life operating conditions, a fixed 164 speed diesel generator should operate above the minimum loading2 level which is specified by the generator manufacturer [59]. To overcome this issue, typically a dump load is integrated with a diesel generator where the operation of dump load is enabled to ensure the commitment of minimum loading condition of the diesel gen- erator3. However, such a scheme is not economically feasible as the generated power from diesel generator is not fully utilised to meet the demand of the main load of the system. Moreover, complex synchronisation techniques may need to be employed during the mode transition of the RAPS system (e.g. WO to WD modes of operation). To address the first two challenges stated above, a different design approach of a diesel generator scheme is proposed in the present work. The proposed diesel generating scheme particulary focuses on improving its performance during low load operation thus ensuring high operating efficiency. In addition, the control features are integrated to avoid unnecessary start/stop operation of the diesel generating system.

Furthermore, in the proposed scheme, diesel generator operates independently but in parallel with other sources thus avoiding the need for a synchronisation mechanism.

The third challenge is achieved by adopting a coordinated control approach which will be discussed in Section 6.4. The fourth aspect is achieved by developing control strategies for each system component while giving signiﬁcant attention to the energy storage system.

6.3.2 Dual Mode Operation of a Diesel Generating System

The diesel generating system is able to operate in dual mode, either as a synchronous condenser or as a generator which is controlled via a friction clutch as shown in

Fig. 6.3. The mode selection of the generating system is mainly determined by two parameters of the RAPS system namely: (a) demand-generation mismatch, ∆PwL =

2Typically this is 20-30% of the rated capacity of the diesel generator. 3Refer to Appendix C which illustrates a case study representing this scenario. 165

Figure 6.2: Speciﬁc fuel consumption of a loaded diesel engine.

Pw − PL and (b) frequency deviation, ∆f. The relationship that exists between the operating mode of the synchronous machine and the ﬁrst criterion ∆PwL, can be described using (6.5).

  (PL − Pw) if (∆PwL) ≥ α(Pde)rated (generting mode) Pde = (6.5)  0 otherwise (synchronous condenser mode)

where, Pde is diesel power output, PL is load demand, Pw is load demand, ∆PwL is demand-generation mismatch and α is minimum loading condition in per unit4.

If the generation-demand mismatch satisﬁes the minimum loading condition, the synchronous machine performs as a generator, otherwise it works as a condenser.

As stated earlier, the second criterion which is given by ∆f can be regarded as an indirect measure of the power imbalance associated with the RAPS system that can be described using (6.4). However, when the synchronous machine needs to operate

4It is a fraction between 0 and 1. 166

in the generating mode under such a frequency behviour5, it needs to satisfy the ﬁrst

criterion given by (6.5). If the condition given in (6.5) is not satisﬁed, the frequency

regulation is achieved using another type of energy source (e.g. energy storage system)

while allowing the synchronous machine to operate as a condenser. Considering all

these facts, the generation of control logic associated with the mode selection of the

synchronous machine can be given as in Table 6.2.

Table 6.2: Control logic associated with mode transition of the diesel generating system

(∆PwL) ≥ α(Pde)rated ∆f < (∆f )min Operating mode of the synchronous machine 0 0 synchronous condenser 0 1 synchronous condenser 1 0 generator 1 1 generator

The dynamics associated with the friction clutch of the diesel generating system

is shown in Fig. 6.3. The friction clutch is used to select the modes of operation of

the diesel generator. When the clutch signal is “ON” state (i.e. control logic ‘1’),

6 the diesel engine allows transfer of the combined torque Tc, given by (6.6) to the synchronous machine. Contrarily, when the clutch is in “OFF” state (i.e. control

logic ‘0’) which makes the torque input to the synchronous machine Tc, to zero thus leading the generating scheme to operate as a condenser.

HsTd + HdTs Tc = (6.6) Hs + Hd

where, Tc is equivalent torque transmitted to synchronous machine when clutch signal = 1, Hs and Hd are inertia constants of synchronous machine and diesel engine

5 i.e. ∆f < (∆f)min 6Combined torque refers to the torque due to diesel engine and synchronous machine. 167

respectively and Ts, Td are torque components of synchronous machine and diesel engine respectively.

Clutch Diesel Engine ωd ωs Synchronous Machine

T , H d d Ts, H s

Figure 6.3: Dynamics associated with the clutch system of the diesel generating system.

The entire process associated with generating the clutch signal is shown in Fig.

6.4. During WO mode of operation, the state of the clutch signal is set to logic zero

‘0’, and Don which is generated based on the logic conditions given in Table 6.2 is also set at the logic zero state. When the conditions are satisﬁed to activate Don to logic ‘1’, the diesel engine starts its cranking process7 allowing the engine speed to

reach the firing speed8. Once the diesel engine reaches its firing speed, the cranking process is stopped and the reference speed of the engine is set to the synchronous speed (i.e. 1 pu). In order to simplify the cranking process of the diesel generating system, instead of reaching standstill situation, the speed of the diesel engine is set to 0.3 pu of the rated synchronous speed. To represent the time taken for completing the cranking process a time delay is integrated as shown in Fig. 6.4. Once the diesel engine reaches its firing speed, the cranking process is switched off while the speed of the engine ωr given by (6.7) increases until it reaches the synchronous speed, ωs. However, at this point, the clutch needs to satisfy two conditions in order to enable its operation which are shown in Fig. 6.4. Firstly, as stated before, the Don signal

7The system controller will switch on the DC supply of the starter motor of the engine. 8The speed at which the diesel generator starts energising the sparks plugs to initiate the process of internal combustion. 168 which is determined by the condition given in Table 6.2 needs to be set to logic ‘1’.

Secondly, the speed diﬀerence of the diesel engine ωd and synchronous machine ωs, should be less than a minimum speed condition ξ, (typically ξ=10−5) as given in (6.8).

These two conditions ensure smooth coupling or synchronisation between the shafts of the diesel engine and synchronous machine thus generating a combined torque Tc, given by (6.6). Once these two conditions are satisﬁed, the logic controller9 which is responsible for generating the clutch signal shown in Fig. 6.4 decides the clutch operation (i.e. logic ‘1’ or ‘0’) accordingly. 1 Z ωd = (Td − Tc) (6.7) 2Hd

|ωd − ωs| ≤ ξ (6.8)

ω To clutch s + U - Logic Comparator AND GATE controller ∃ ωd

P L + - NOT GATE Logic Comparator t=T0 Pw Table 7.2 (f) rated + - Don Comparator f Figure 6.4: Generation of clutch signal of the diesel engine.

The proposed diesel engine model is shown in Fig. 6.5. The transfer function of the diesel engine consists of a speed regulator and an actuator which are represented as a PID controller and second order system10 respectively. In addition to the above

9For simulation purpose, a SR type ﬂip-ﬂop is chosen as the logic controller. 10Refer to Appendix C for further information. 169

suggested control procedures, an additional PI controller is integrated into the gov-

ernor system to ensure the fulﬁlment of required power deﬁcit of the system. The

control signal associated with PI controller is enabled only when the clutch signal is

at the “ON” state. Otherwise the output of that PI controller is set to zero. IEEE

type-1 voltage regulator11 and an exciter system are used with the diesel generating

scheme which are responsible for regulating the load side voltage [95].

Rated speed 1.0 Diesel engine Td + + - - transfer function 0.3 Selector Switch 1 ωd Firing speed + 1/2H d ωd - ∫

Don P t=T0 Tc Clutch PI 0 OR gate Selector Switch 2 Time delay 0 Selector Switch 3 Figure 6.5: Diesel engine model.

6.4 Detailed Model of Wind-Diesel-Battery RAPS System

6.4.1 Coordinated Control Approach

The control coordination approach for the two types of RAPS systems: (a) DFIG

and (b) PMSG is formulated with a view to minimise the active and reactive power

imbalance associated with each system which can be given as in (6.9)-(6.10)12.

Pw + Pde ± Pb = PL + Pd (6.9)

Qde + Qw = QL (6.10)

where, Pw is wind power output, Pde is diesel generator power output, Pb is battery

power output, Pd is dump load power, PL is load demand of the system, Qde is reactive

11Refer to Appendix C for further information. 12 It is assumed that Qde and Qw are leading and QL is lagging. 170

power output from diesel generator, Qw is reactive power supply from wind generator

and QL is reactive power demand of loads. A ﬂow chart represents the control coordination logic associated with the decision

making process of active power sharing between the diﬀerent system components

is depicted in Fig. 6.6. It illustrates the power sharing process among the system

components for three situations: (a) over-generation, (b) under-generation and (c)

emergency situations where the wind turbine generator shuts down its operation.

In over-generation situations, the power output of the wind turbine generator Pw is

greater than the load demand PL, the battery storage system absorbs the additional

power, (∆PwL = Pw − PL). If the excessive generation, is greater than the maximum capacity of the battery storage (Pb)max, then the dump load starts consuming the

additional power. If the dump power Pd is larger than its maximum rating (Pd)max, then the wind turbine pitch regulation activates in order to limit the power output

from the wind turbine generator. In under-generation situations, where the power

output of the wind turbine generator Pw, is less than the load demand PL (i.e. ∆PwL

= Pw-PL < 0), two options are available, which are (a) to operate the battery storage system or (b) to operate diesel generating system. However, the decision for selecting

either option is mainly determined by the magnitude of generation-demand mismatch

given by ∆PwL or logic table given in Fig. 6.2. If ∆PwL is less than the minimum loading condition of the diesel generator, the battery storage system needs to be

operated in order to satisfy the required power deﬁcit. Otherwise, the diesel generator

starts supplying the power to the load together with the wind generator and battery

storage system. These conditions can be given as in (6.11).

  Pw + Pb ∆PwL < β0(PL)rated PL = (6.11)  Pw + Pde ± Pb otherwise 171

During emergency situations, such as no power output from wind turbine generator due to wind speed falling below cut-in or going above cut-out speeds, a load shedding scheme can be implemented if necessary where the existing load is then supplied by the diesel generator and the battery storage system.

Two distinct reactive power regulation schemes are applied on the DFIG and

PMSG based RAPS systems separately. Considering the inherent reactive power capability of diesel generating system, the DFIG is used only to self compensate its no- load reactive power (QDFIG)NL, while the reactive power demand of the loads is met by using the diesel generating system which can be described by (6.12). Contrarily, the reactive power sharing between the PMSG and diesel generating scheme is performed in an uncoordinated manner as given in (6.13) [96].

(QDFIG)NL + Qde = QL (6.12)

QPMSG + Qde = QL (6.13)

where (QDFIG)NL is no load reactive power of DFIG, Qde is reactive power output of DFIG, QL is reactive power demand and QPMSG is reactive power output from PMSG.

6.4.2 DFIG based Wind-Diesel Hybrid RAPS System

The conﬁguration of the hybrid RAPS system consisting of a DFIG as the wind turbine generator, a diesel generating system, a battery storage system and a dump load is shown in Fig. 6.7. In brief, the operation of the system is as follows. When the wind energy is suﬃcient (i.e. in WO) to satisfy the load demand, the diesel generating system behaves as a synchronous condenser which supplies only the reactive power to 172

Wind Power Generation Pw Load Demand PL

No (vw)cut-in >vw>(vw)cut-out Pw=0

Yes

No P -P > (P ) Yes Pw> P L L w de min Diesel "ON"

Yes No No Yes Pw+Pde -P L>0 Pb>(Pb)max Yes P =P -P d w L No

Charge battery

Discharge No Pw-P L>(Pb)max battery

Yes

Pd=Pw-P L-(Pb)max

Pitch Angle Yes Regulation Pd >(P d)max

Frequency Regulation

Figure 6.6: Instantaneous power ﬂow control of the wind-diesel-battery RAPS system. 173 the loads while the active power requirement of the load is then supplied by the DFIG and battery storage system. In WD mode, the diesel generating scheme operates as a generator which supplies active and reactive power to the loads. In addition, the

DFIG and battery storage system participate in regulating the active power balance of the RAPS system. Apart from that, the battery storage system is used to extract the maximum power from wind. The capacity of active and reactive power support provided by each system component are summarised in Table 6.3. The control strategies applied for the diesel generator and dump load remain the same as illustrated in Section 6.3.2 of this chapter and in Section 4.3.4 of Chapter 4 respectively. The control strategies applied for the DFIG, battery storage system and dump load are discussed in the following subsections.

DFIG

RSC LSC Dump load

∆P v,f Main loads DC-DC vdc converter G + _ Engine Clutch Synchronous machine Battery storage ∆P Figure 6.7: DFIG based wind-diesel-battery RAPS system.

(a) DFIG control

As stated in Section 6.4, DFIG is used to self compensate its no load reactive power, Qmag which is given by (6.14) and the corresponding rotor reference d-axis current (idr)ref , can be estimated using (6.15) and (6.16). The other control aspects 174

Table 6.3: Active and reactive power support from each system component System component Active power Reactive power Wind generator X X Synchronous machine X X battery storage X × dump load X ×

of the RSC13 (i.e. to control the frequency) and LSC (i.e. to regulate the DC link voltage) are identical to the control strategies discussed in Section 3.2.4 of Chapter

2 Lmvs vs Qmag = − (6.14) Ls + Lm ωs(Ls + Lm)

Qmag = 0 (6.15)

vs (idr)ref = (6.16) ωsLm

(b) Battery Storage System Controller

There are two main objectives associated with the battery storage where the ﬁrst is to maintain the demand-generation mismatch and the second objective is to extract the maximum power from wind. The ﬁrst objective is achieved by determining the generation-demand mismatch based on the condition given in (6.17). The second ob-

14 jective is realised by generating the battery reference current (ib)ref , by considering the optimum wind power output from the DFIG, (PDFIG)opt as one of the inputs of the battery storage controller.

  (PDFIG)opt − PL; charging (Pb)ref = (6.17)  (PDFIG)opt + Pde − PL; discharging

13Refer to Appendix C for modiﬁed RSC control scheme. 14 (Pb)ref This can be estimated using (6.17) where (ib)ref = ; vb is the battery voltage. vb 175

The battery storage system is connected to the DC bus of back-to-back converter

using a bi-directional buck-boost converter and the adopted battery storage control

strategy is shown in Fig. 6.8.

Comparator To bi-directional (P buck-boost converter DFIG)opt P (ib)ref + - ÷ + - PI + - Limiter

v ib PL b Triangular carrier waveform

Figure 6.8: Battery storage controller for the DFIG based wind-diesel-battery RAPS system.

6.4.3 PMSG based Wind-Diesel Hybrid RAPS System

The conﬁguration of PMSG based RAPS system is shown in Fig. 6.9. The battery

storage system and dump load are connected to the DC bus of the wind energy system

whereas the diesel generating system is connected to the AC side of the RAPS system.

The main objective of the battery storage system and dump load is to regulate the

DC link voltage. The DC/DC converter is also used to extract the maximum power

from wind. The control methodologies that are adopted for the PMSG inverter,

battery storage system, DC/DC converter and dump load are identical to the control

approaches which were illustrated in Section 4.3.5 of Chapter 4.

6.5 Performance of the Hybrid Wind-Diesel-Battery based

RAPS System

The performance of the hybrid RAPS systems is investigated using linearised and

detailed models. In this regard, the suitability of the proposed control strategies

of each system component together with the coordinated control approach are ob- 176

Full bridge PMSG rectifier DC/DC converter-1 Inverter i G dc

vdc

Wind turbine (PPMSG )opt v, f DC/DC converter-2 Dump load + Battery storage vdc _ Main loads G

vdc Clutch Engine Generator

Figure 6.9: PMSG based wind-diesel-battery RAPS system. served with regard to their voltage and frequency bandwidth regulation capabilities and maximum power extraction from wind under changing wind and variable load conditions. In addition, special attention is given to observe the performance of the diesel generating system. The parameters associated with each type of RAPS system is listed in Appendix C.

6.5.1 Performance of the linearised model of Wind-Diesel-Battery RAPS

System

In the linearised model, the cut-in wind speed, rated wind speed and cut-out speed of the wind turbine are selected as 7 m/s, 11 m/s and 20 m/s, respectively. Also, it is assumed that the battery storage system and diesel generator are able to serve

30% and 70% of the rated load demand respectively. The simulated behaviour of the hybrid power system is shown in Fig. 6.10. The wind velocity variation is shown in

Fig. 6.10-(a) and the corresponding wind power output is depicted in Fig. 6.10-(b).

At t = 75 s, the wind velocity is set approximately to 15 m/s and the corresponding power output from the wind generator is 0.8 pu. Initially the load demand is set at

0.7 pu as shown in 6.10-(d) where the excess power given by ∆PwL = (PL − Pw) is 177

at 0.1 pu. This excess power is absorbed by the battery storage system as evident

from Fig. 6.10-(d). At t = 75 s, the wind velocity drops to nearly 10 m/s causing

a reduction in wind power output to 0.55 pu. The corresponding power imbalance

15 ∆PWL of 0.15 pu is then supplied by the diesel generator while keeping the battery storage in the idling mode as shown in Fig. 6.10-(c). At t = 150 s, the load demand

is increased to 0.9 pu which makes the diesel generator to increase its power output

to 0.35 pu. At t = 250 s, the wind velocity is reduces to 5 m/s which is below the

cut-in-speed causing the wind turbine generator to shut down its operation. However,

the maximum rating of the diesel generator is set to 0.7 pu and therefore the load

demand is shared between the battery storage system and diesel generator. It can be

seen that, the diesel power output reaches its maximum capacity of 0.7 pu, while the

remaining load demand (i.e. 0.2 pu) is supplied through the battery storage. At time

t = 300 s, the load demand is reduced to 0.4 pu which is entirely supplied using the

diesel generator while keeping the battery storage power output at zero or kept in

idling mode as evident from Fig.6.10-(c) and (d) respectively. The power imbalance

(∆P = Pw +Pde ±Pb −PL) associated with the RAPS system is shown in Fig. 6.11-(b). It can be seen that the power imbalance ∆P , of the RAPS system is always maintained

at near zero level except during the transient conditions. The corresponding system

frequency deviation ∆f is shown in Fig. 6.11-(a). Upon close examination, it can be seen that the frequency of the RAPS system is regulated within ±0.1%. As expected,

the highest frequency deviation is seen to occur at t = 150 seconds which corresponds

to the step load change. The summary of the results observed in Fig. 6.10 is given

in Table. 6.4. 15α in (6.5) is considered as 0.15. 178

V_w (m/s) V_w 0 50 100 150 200 250 300 350 (a) 1

0.5

P_w (pu) P_w 0 50 100 150 (b) 200 250 300 350 1

0.5

P_DE (pu) P_DE 0 50 100 150 (c) 200 250 300 350 0.5

P_b (pu) P_b -0.5 50 100 150 (d) 200 250 300 350 1

0.5

P_L (pu) P_L 0 50 100 150 (e) 200 250 300 350 Time (s) Figure 6.10: Power sharing of the linearised model of the wind-diesel-battery RAPS System under variable wind and load conditions: (a) wind speed, (b) wind power, (c) diesel power,(d) battery power, and (e) load demand.

-3 x 10 5

-5

Delta (pu) f Delta -10

-15 50 100 150 200 250 300 350 (a) 0.5

Delta PDelta (pu) -0.5

-1 50 100 150 200 250 300 350 (b) Time (s)

Figure 6.11: (a) Frequency deviation, and (b) power imbalance of the linearised wind-diesel-battery RAPS system. (‘Delta’ in this ﬁgure represents ‘∆’.) 179

Table 6.4: “ON” and “OFF” conditions of the system components

T ime(s) Pw PDE Pb PL pu description 0 − 75 ON OFF ON 0.7 75 − 150 ON ON OFF 0.7 wind velocity ↓ 150 − 250 ON ON OFF 0.9 load ↑ 250 − 300 OFF ON ON 0.9 wind velocity ↓ 300 − 350 OFF ON OFF 0.4 load ↓

6.5.2 Performance of the Detailed Model of DFIG-Diesel-Battery based

RAPS System

The system response and power sharing between the system components under changing wind and variable load conditions are shown in Fig. 6.12 and Fig. 6.13 respectively. The wind condition under which the system is simulated is shown in Fig.

6.12-(a). It can be seen that the wind velocity is initially set at 12 m/s. At t = 4 s, the wind velocity drops to 9 m/s, then it increases to 11 m/s at t = 8 s. The load demand is initially set at 0.6 pu which is increased to a value of 0.775 pu by adding a new load of 0.175 pu at time t = 3 s. The new load (i.e. 0.175 pu) is disconnected from the RAPS system at t = 9 s as shown in Fig. 6.13-(d). The corresponding load side voltage is shown in Fig. 6.12-(b) which is not seen to be aﬀected by the wind speed or resistive load step changes. A sudden voltage excursion can be seen at t = 9.25 s which corresponds to the clutch disengagement of the diesel engine (i.e. mode transfer from synchronous generator to synchronous condenser mode) as evident from Fig. 6.13-(b). At t = 9.25 s the synchronous machine enters the motoring mode before changing to the synchronous condenser mode of operation thus causing voltage excursions in the system as evident from Fig. 6.13-(b) and Fig. 6.12-(b). The load side voltage of the system stays within ±1% of its rated value during normal operation. Fig. 6.12-(c) shows the frequency of the system voltage. As anticipated, 180 the frequency is closely regulated at the rated value of 1.0 pu and is not seen to be inﬂuenced by the wind speed change, load step change and mode transition of the synchronous machine (i.e. synchronous condenser to generating mode and vice versa).

The frequency of the system is maintained within 0.2% of its rated value during normal operation which is mainly regulated by the RSC of the DFIG while maintaining the active power balance of the RAPS system. The DC link voltage of the DFIG is depicted in Fig. 6.12-(d). The simulated behaviour of the DC link voltage shows that it is well regulated around its rated value throughout the operation except during load variations, wind step changes and state transitions of synchronous machine

(i.e. synchronous condenser to synchronous generator and vice versa). However, the highest DC link voltage variations are seen to occur at t = 7 s and t = 9.25 s which correspond to the mode transition of the synchronous machine as evident from Fig.

6.13-(b). Also, during the same time interval the battery storage quickly changes its direction of power (i.e. charging to discharging mode of operation) as evident from

Fig. 6.13-(c). Even during the transient conditions, the DC link voltage variation is limited to within ± 5% of its rated value.

The wind power variation of the system is shown in Fig. 6.13-(a). For simulation purposes, initially the slip of the wind turbine is set to s = - 0.1 which corresponds to super synchronous mode of operation. According to the wind turbine characteristics, the corresponding maximum power output of the wind generator is 0.73 pu at a shaft speed of 1.2 pu for a wind speed of 11 m/s. From Fig. 6.13-(a), the power output of the DFIG is seen to rise to a value of 0.73 pu at t = 4 s. At this time the load demand is set to 0.6 pu where the additional power is absorbed by the battery storage system. At t = 3 s, the load demand is increased to 0.775 pu and then the battery storage changes its mode of operation (i.e. charging to discharging) and provides the required power deﬁcit demanded by the loads. At t = 4 s, the wind speed is reduced 181

10 V_w (m/s) 5 1 2 3 4 5 (a) 6 7 8 9 10 1.05

1 V_L (pu) 0.95 1 2 3 4 5 (b) 6 7 8 9 10 1.005

1 f_L (pu) 0.995 1 2 3 4 5 (c) 6 7 8 9 10 1.1

V_DC (pu) 0.9 1 2 3 4 5 (d) 6 7 8 9 10 Ti me (s)

Figure 6.12: Response of the DFIG based wind-diesel-battery RAPS system under variable wind and load conditions: (a) wind speed, (b) load side voltage, (c) frequency on load side, and (d) DC link voltage. to 9 m/s causing a reduction in wind power output. When the diﬀerence between the load demand and wind power (PL − Pw), is greater than 0.3 pu (i.e. β0 in (6.11) is set to 0.3), the condition under which the diesel generator should operate is satisﬁed.

However, the synchronous machine still behaves as a synchronous condenser which needs extra time16 to reach its rated speed (i.e. 1 pu). At t = 7 s, the synchronous machine changes its mode of operation from condenser to generating mode as evident from Fig. 6.13-(b). The diesel generator keeps supplying power until the load step reduction occurs at t = 9 s. During the generating mode of operation (i.e. t = 7 to

9 s) of the synchronous machine, the battery storage system operates as a buﬀer to ensure maximum power extraction from wind whose depth of discharge is at minimal levels as shown in Fig. 6.13-(c). At t = 9 s, the synchronous machine changes its mode of operation from generating to condenser mode where the load demand is then entirely supplied through the DFIG and battery storage system.

16The time taken for the cranking process. 182

0.5 P_w (pu) 0 1 2 3 4 5 6 7 8 9 10 (a) 0.5

0 P_d (pu) -0.5 1 2 3 4 5 6 7 8 9 10 (b) 0.5

0 P_b (pu) -0.5 1 2 3 4 5 6 7 8 9 10 (c) 0.8

0.6 P_L (pu) 0.4 1 2 3 4 5 (d) 6 7 8 9 10 Ti me (s) Figure 6.13: Power sharing of the DFIG based wind-diesel-battery RAPS system under variable wind and load conditions: (a) wind power, (b) diesel power, (c) battery power and (d) load demand.

The corresponding active power imbalance associated with the entire RAPS system and frequency deviation of the system voltage are shown in Fig. 6.14-(a) and (b) respectively. It can be seen that the active power imbalance of the RAPS system is maintained between 0 to 0.5% except during the mode transition of the synchronous machine (i.e. t = 7.25 s and t = 9 s). The frequency deviation of the system stays within ± 0.2% of its rated value even during the transient conditions. The maximum power tracking characteristics of the DFIG is shown in Fig. 6.15. It can be seen that the DFIG runs on its maximum power extraction mode except during transient conditions.

The reactive power sharing between the DFIG, synchronous machine and load is shown in Fig. 6.16. It can be seen that the DFIG operates almost at unity power factor without consuming or supplying reactive power into the system. The entire reactive power requirement of the load is supplied by the diesel generating scheme.

The performance of the diesel generating scheme is also observed in relation to 183

-3 x 10 10

0 Delta P (pu)

-5 1 2 3 4 5 6 7 8 9 10 (a)

-3 x 10 5

0 Delta f (pu)

-5 1 2 3 4 5 6 7 8 9 10 (b) Ti me (s)

Figure 6.14: Frequency and active power deviation of the DFIG based wind-diesel- battery RAPS system: (a) active power imbalance and (b) frequency deviation.

Pw 0.9 PMPE 0.8

0.7

0.6

0.5 P_w (pu) 0.4

0.3

0.2

0.1

0 1 2 3 4 5 6 7 8 9 10 Ti me (s)

Figure 6.15: Maximum power point tracking characteristics of DFIG based wind turbine generator of the wind-diesel-battery RAPS system. 184

Q DFIG 0.6 Q de Q 0.4 L

0.2

-0.2 Reactive Power (pu) -0.4

-0.6

1 2 3 4 5 6 7 8 9 10 Time (s)

Figure 6.16: Reactive power sharing between the components of the DFIG based wind-diesel-battery RAPS system.

its mechanical speed and load angle. The mechanical speed of the diesel generator is

maintained at almost 1 pu. However, considerable speed deviations are seen to occur

during the clutch changes at t = 6.75 s and t = 9 s. The load angle of the diesel

generator stays at nearly zero degrees when it operates as a synchronous condenser

and increases to about 47 degrees when it functions as a generator as evident from

Fig. 6.17. The speed of the synchronous machine ωs, speed of diesel engine ωr, and clutch signal are shown in Fig. 6.18. It can be seen that the clutch signal is enabled or set to logic ‘1’ nearly at t = 7 s when the respective speeds of diesel engine and synchronous machine are equal to each other.

6.5.3 Performance of the PMSG-Diesel-Battery based RAPS System

The PMSG based RAPS system is simulated under variable wind and load conditions where Fig. 6.19 shows the system response and Fig. 6.20 illustrates the power sharing among diﬀerent system components. The wind condition under which the system is 185

1.05

(pu) 1 r w

0.95 1 2 3 4 5 6 7 8 9 10 (a)

100

0 Theata (deg) -50 1 2 3 4 5 6 7 8 9 10 (b) Time (s)

Figure 6.17: Diesel generator performance of the DFIG based wind-diesel-battery RAPS system: (a) rotor speed and (b) load angle.

1.4 ωωω r ωωω 1.2 s Clutch Signal 1

0.8

0.6 Speed (pu)

0.4

0.2

0 1 2 3 4 5 6 7 8 9 10 Time (s)

Figure 6.18: Speeds of the engine, synchronous machine and clutch signal of the DFIG based wind-diesel-battery RAPS system. 186 simulated shown in Fig. 6.19-(a). It can be seen that the wind velocity is set initially at 12 m/s. At t = 3 s, the wind velocity drops to 9 m/s, then it is reduced to 0 m/s at t = 7 s where the RAPS system experiences no-wind condition (i.e. emergency situation). The load demand is initially set at 0.7 pu and at time t = 4 s, the load is increased to a value of 1 pu by adding a new load of 0.3 pu. Then the new load

(i.e. 0.3 pu) is disconnected from the system at t = 6 s as shown in Fig. 6.20-

(d). The AC voltage at load side is shown in Fig. 6.19-(b). It can be seen that the AC load voltage is aﬀected by the wind speed and load step changes (i.e. load addition and reduction) which occur at t = 3 s, t = 4 s and t = 6 s, respectively.

The highest voltage variation is experienced at load addition which occurs at t = 4 s. Also, the synchronous machine changes its mode of operation from synchronous condenser to generating mode by enabling the clutch at t = 4.2 s as seen in Fig. 6.20-

(c). However, the highest load voltage variation is limited to ± 4% of its rated value which is mainly due to the dynamics associated with load step change together with the mode transition of synchronous machine. The load side voltage of the system stays within ±1% of its rated value during normal operation. The frequency of the system voltage is shown in Fig. 6.19-(c). As expected, the frequency excursions are seen to occur during wind speed change at t = 3 s and load step changes at t = 4 s and t = 6 s. However, the frequency of the system is able to maintain within 0.25% of its rated value throughout the operation. The DC bus voltage also experiences slight

ﬂuctuations during wind speed changes at t = 3 s and t = 7 s and load step changes at t = 4 s and t = 6 s. The highest DC bus voltage variation is experienced at t = 4 s which is around ± 15% at its rated value.

The wind power variation of the system is shown in Fig. 6.20-(a). Initially, the power output of the PMSG is at nearly 0.7 pu. After t = 3 s, the RAPS system experiences a under generation scenario where the battery storage is in discharge 187

15 10

5 V_w (m/s) 0 1 2 3 (a) 4 5 6 7 8 1.1

1 V_L (pu) 0.9 1 2 3 (b) 4 5 6 7 8

1.005

1 f_L (pu) 0.995 1 2 3 4 5 6 7 8 (c)

1.2 1

V_DC (pu) 0.8 1 2 3 4 5 6 7 8 (d) Time (s)

Figure 6.19: Response of the PMSG based wind-diesel-battery RAPS system under variable wind and load conditions: (a) wind speed, (b) load side voltage, (c) frequency on load side, and (d) DC link voltage. mode of operation. At t = 3 s, the wind speed drops to 9 m/s causing a reduction of wind power output and hence the battery storage system further increases its discharge rate to satisfy the load demand as depicted in Fig. 6.20-(b). With the load step addition which occurs at t = 4 s, the synchronous machine changes its mode of operation from condenser to generator mode by enabling its clutch signal as shown in Fig. 6.20-(c). During this time period, the battery storage lowers it depth of discharge to almost idling mode (i.e. very low level of discharge rate). At t = 6 s, the load is reduced to 0.7 pu as shown in Fig. 6.20-(d) which causes the battery storage to operate in charging mode. The RAPS system experiences a no-wind condition or an emergency situation at t = 7 s where the battery storage and diesel generator are in operation to satisfy the load demand. The maximum power extracted from the wind is shown in Fig. 6.21. It can be seen that the PMSG operates on its maximum power extraction mode except during transient and emergency conditions 188 which are unavoidable. The active power imbalance or power mismatch associated with the entire RAPS system is shown in Fig. 6.22. Upon close examination, it can be seen that the power imbalance of the system almost stays at zero. This reﬂects the suitability of the proposed coordinated control methodology in managing and stabilising the operation of the RAPS system.

0.5 P_w (pu) 0 1 2 3 (a) 4 5 6 7 8 1

0 P_b (pu) -1 1 2 3 (b) 4 5 6 7 8 1

0 P_de (pu) -1 1 2 3 (c) 4 5 6 7 8 1.5

1 P_L (pu) 0.5 1 2 3 4 5 6 7 8 (d) Time (s)

Figure 6.20: Power sharing of the PMSG based wind-diesel-battery RAPS system under variable wind and load conditions: (a) wind power, (b) battery power, (c) diesel power, and (d) load demand.

The reactive power sharing of synchronous condenser and inverter is shown in

Fig. 6.23. Initially, the load reactive power is set to 0.3 pu. The highest reactive power support is provided through LSC where the rest is supplied by the synchronous condenser. During the load step change that takes place at t = 4 s, the reactive power demand is increased to 0.5 pu where the highest proportion of reactive power is now supplied through synchronous condenser while the rest is provided through LSC. 189

1 P actual w 0.9 P MPPT w

0.8

0.7

0.6

0.5

Power (pu) 0.4

0.3

0.2

0.1

0 1 2 3 4 5 6 7 8 Time (s)

Figure 6.21: Maximum power point tracking characteristics of PMSG based wind turbine generator for wind-diesel-battery RAPS system.

-5 x 10 11

8 Delta Delta P (pu) 7

5 1 2 3 4 5 6 7 8 Time (s)

Figure 6.22: Active power imbalance of the PMSG based wind-diesel-battery RAPS system. (“Delta” in this ﬁgure represents ∆) 190

0.6

Qinv

0.5 QL

Qsyn 0.4

0.3 ) u p

( 0.2

r e w o

p 0.1

e v i t c

a 0 e R

-0.1

-0.2

-0.3

-0.4 1 2 3 4 5 6 7 Time (s)

Figure 6.23: Reactive power sharing between the system components of the PMSG based wind-diesel-battery RAPS system. 191 6.6 Chapter Summary

This chapter considered the case where a diesel generator operates in connection with a highly wind penetrated RAPS system. In this regard, the operation of the diesel generator has been modelled to operate in two modes: (a) synchronous condenser and (b) generating mode. The applicability of the proposed control strategies for each system component together with the proposed control coordinated approach have been investigated under changing wind and variable load conditions. Based on the results obtained, following conclusions can be drawn:

• Both RAPS systems are capable in regulating the voltage and frequency within

acceptable limits during the wind and load step changes. However, the transient

performance of the DFIG based RAPS system was seen to be comparatively bet-

ter than the PMSG counterpart. This can be illustrated by considering the pen-

etration levels of each type of RAPS system. In the case of DFIG based RAPS

system, the ratio given by (PDFIG)rated is high compared to ratio (PPMSG)rated of (Pde)rated (Pde)rated the PMSG based RAPS system. Therefore, during the mode transition of the

diesel generating system, DFIG based RAPS system has minimal impact com-

pared to PMSG counterpart. In addition, the inertia provided by the DFIG

in damping out the frequency ﬂuctuations helps in providing better transient

performance compared to the PMSG. Moreover, the reactive power sharing of

the DFIG based RAPS system is made in a coordinated manner where the

diesel generating system is responsible in establishing the system voltage. In

contrast, where the uncoordinated reactive power control of the PMSG based

RAPS system may lead to voltage variations.

• The power sharing between the diﬀerent RAPS system components was found to

be similar in the investigations carried out using linearised and detailed models 192

of the RAPS systems. In addition, detailed models of the RAPS systems (i.e.

DFIG and PMSG) were capable of following their maximum power tracking

characteristics during the operating period. Furthermore, it has been shown

that the proposed reactive power control of the DFIG helped in operating at

unity power factor whereas the PMSG inverter and diesel generating system

shared their reactive power in an uncoordinated way while satisfying the reactive

power demand of the RAPS system.

• The proposed diesel generating scheme performed well in a manner that its op-

eration was enabled only when it needs to satisfy at least the minimal loading

condition thus ensuring eﬃcient operation. Moreover, the proposed diesel gen-

erating scheme has avoided a synchronisation scheme while operating in parallel

with the wind turbine generator.

• The entire operation of the RAPS systems was enhanced by adopting techniques

for extracting maximum power from wind and by avoiding the operation of diesel

generating system at low load conditions. Chapter 7

Hydrogen as Energy Storage for

Wind Dominated RAPS System

7.1 Introduction

The importance of integrating a diesel power generating scheme to a wind based autonomous power system was illustrated in Chapter 6. However, with the increasing attention placed on environmental issues associated with such diesel power generating scheme, focus is now being given for “carbon free” generating technologies. In addition to carbon emission, increasing diesel fuel prices and the issue associated with fuel transportation to remote locations further justify the selection of such “carbon free” generating schemes. In this regard, hydrogen based power generating schemes can be considered as one of the emerging power generating technologies which are seen to provide feasible solutions for wind based RAPS systems.

Integration of a hydrogen based generating scheme, which consists of a fuel cell system, an electrolyser and a hydrogen storage tank, into a wind based RAPS system can be regarded as the ﬁnal leap of the autonomy of operation in a wind based hybrid

RAPS system. In such a RAPS system, hydrogen functions as an energy carrier 193 194 which is generated and utilised by an electrolyser and a fuel cell system respectively.

In addition, a hydrogen storage tank can be added to store the hydrogen fuel.

This chapter presents the performance of a hydrogen generating scheme in relation to the following aspects of the wind based RAPS systems:

• as a substitute for the battery energy storage system used in the DFIG based

RAPS system examined in Chapter 6 and

• as a complete replacement for the diesel generating system of the PMSG based

RAPS system examined in Chapter 6 by a hydrogen generating system

Behaviour of both RAPS systems are examined giving signiﬁcant consideration to the modelling aspects of the components of the hydrogen generating system. In addition, the suitability of the control strategies which are implemented for each system component is investigated in relation to voltage and frequency regulation together with maximum power extraction capability from wind.

7.2 Hydrogen Storage Systems

7.2.1 Fuel cell System

Fuel cell system is a static electrochemical device which converts the chemical energy of a fuel1 into electricity. The chemical reaction that takes place in the electrodes of the fuel cell system can be given as in (7.1) [86].

2− − ) Anode : H2 + O → H2O + 2e (7.1) 1 − 2− Cathode : 2 O2 + 2e → O

Due to the high power capacity, eﬃciency and thermal stability, solid oxide fuel cell (SOFC) can be eﬀectively employed for high power standalone power application.

1In this case, the fuel represents Hydrogen. 195

There are various models currently available in the literature which describe the

behaviour of SOFC systems in relation to their electrical, electrochemical, hydraulic

and thermal aspects. In the present work, the SOFC model explained in [85] is

taken as the basis where its thermal aspects are excluded2 and therefore constant

temperature mode of operation is assumed.

The SOFC system that is employed in the current work is shown in Fig. 7.1 which

comprises of two main components namely: fuel processor and electrical processor.

in The fuel processor is used to estimate the input molar ﬂow rates of Hydrogen qH2 , and

in Oxygen qO2 whereas the electrical processor is used to estimate the reacted molar ﬂow

r r rates of Hydrogen qH2 , and Oxygen qO2 , during the process of generating the electric current. The molar ﬂow rates of Hydrogen and Oxygen are then used to estimate

the partial pressures of water, Hydrogen and Oxygen indicated by pH2O, pH2 and pO2 respectively. These partial pressures are then used to estimate the reversible voltage

or Nernst’s voltage3 and thereby the voltage that appears across the output terminals

of the fuel cell is given by (7.2).

√ RTfc pH2 pO2 vfc = nfc(E0 + [ln ] − vdrops) (7.2) 2F pH2 O

where, vfc is fuel cell voltage, nfc is number of series fuel cells connected in the

stack, E0 is voltage associated with reaction free energy, Tfc is operating temperature of the fuel cell, R is universal gas constant [J/(KmolK)], F is Faraday constant.

The diﬀerent voltages that drop within the fuel system, vdrops refer to (7.3) include:

2The thermal characteristics of a fuel cell system has minimum or no impact on its electrochemical characteristics. √ 3 RTfc pH2 pO2 Nernst voltage is given by (E0 + [ln ]). 2F pH2 O 196

Input molar flow rate { Losses Umax /2K r

Id Limiter Load demand 1/(1+Ts)

2K r Umin /2K r Electrical processor Kr vact,cell vohm,cell vconc,cell r r q H2 q O2 Ohmic - - + + + Id 2k/U 1/(1+T ) 1/ r + + opt f in HO in Activation q H2 q O2 Diffusion Fuel processor

1/K /(1+ S) 1/K /(1+ S) H2 τ H2 H2O τ H2O 1/KO2 /(1+ τ O2 S)

Partial pressures PH2 PH2O PO2 - 0.5 n (E0+RT/2F[ln(PH2 P O2 /PH2O )] + vfc number of cells Nernst's equation in series

Figure 7.1: Detail model of the SOFC system.

activation4, ohmic5 and concentration6 losses, given by (7.4) - (7.6) respectively, are

incorporated in the fuel cell model [86]. The numerical values associated with their

expressions are listed in Appendix D.

vdrops = nfc(vact,cell − vohm,cell − vconc,cell) (7.3)

vact,cell = η0 + (Tfca + Tfcblnifc) (7.4)

vohm,cell = rohmifc (7.5)

RTfc ifc vconc,cell = − (1 − ) (7.6) zF ilimit

where, ifc is fuel cell current, ilimit is limiting current, r is internal ohmic resistor

4Activation loss is associated with the rate or slowness of the electrochemical reaction taking place at the electrode. 5Ohmic loss is due to resistive elements in the fuel cell system. 6Concentration loss is due to inability of transporting and maintaining adequate concentration of reactants. 197 of the fuel cell, η0, a, b and c are parameters of SOFC and z is compressibility factor. The equivalent circuit of the fuel cell is shown in Fig. 7.2 which consists of voltage

7 8 Ecell , and all the resistive elements representing the losses associated with the fuel cell and a capacitor is used to characterise the double layer charging eﬀect.

Rohmic

Ract vfc

Rconc

+ E - cell

Figure 7.2: Equivalent circuit of a fuel cell.

7.2.2 Electrolyser and Storage Tank

An electrolyser is used to produce hydrogen by utilising the excess energy available in the RAPS system. The excess energy is used to decompose water containing an electrolyte into Hydrogen, a process described by (7.7)9.

7Commonly known as Nernst’s voltage 8 To represents the vdrops 9This chemical reaction was described in Section 2.6 of Chapter 2. 198

( 1 − ) Anode : 2OH −)(aq) → O2(g) + H2O + 2e 2 (7.7) − Cathode : 2H2O(l) + 2e → H2(g)

An empirical current-voltage relationship which describes the kinetics associated with electrolyser given by (7.8) [83] is employed in the current work. Referring to

Faraday’s law, the Hydrogen production rate nH2 , of an electrolyser is directly proportional to the electrical current which passes through the equivalent circuit given by (7.9). Assuming the constant temperature mode of operation and taking the working temperature as 400C, the Faraday eﬃciency of the electrolyser can be given as in (7.10). In this study, the warming up time required by an electrolyser during its real-life operation is not considered.

t2 t3 r + r T t + + 2 v = v + 1 2 elz i + s log( 1 T T i + 1) (7.8) elz rev A elz A elz η n i n = F c elz (7.9) H2 2F ( 0.09 75.5 ηF = 96.5e − 2 ) (7.10) ielz ielz

where, velz is operating voltage of the electrolyser, vrev is reversal voltage, Telz is operating temperature, ielz is electrolyser current, ηF is Faraday eﬃciency, nc is number of electrolyser cells in series and F is the Faraday constant.

The hydrogen storage tank model shown in the Fig. 7.3 where the tank pressure of hydrogen storage is calculated using (7.11)10 and (7.12). A ﬁrst order transfer

1 function, τS+1 , (typically, τ=3) is used to represent the compressor dynamics.

10This relationship is based on Dalton’s law of partial pressure. 199

pt = pelz − pfc (7.11)

NH2 RTt pt − pi = z (7.12) MH2 V

where, pt, pelz, pfc are pressures of tank, electrolyser and fuel cell systems re-

spectively, NH2 is net hydrogen moles per second delivered to the tank, pi is initial tank pressure, R is universal gas constant, V is volume of the tank, Tt is operating

temperature of the tank and MH2 is molar mass of hydrogen.

1 pelz + τ pt - s +1 Compressor Limiter

p fc

Figure 7.3: Model of Hydrogen storage.

7.3 Application of Hydrogen Storage for a Standalone Wind

Energy System

A Hydrogen storage system can be intergraded to a wind based RAPS system to

provide partial or full autonomy of operation11. In this regard, the technical feasibil-

ity and performance of the hydrogen storage system is examined in relation to the

following wind based RAPS systems:

• RAPS system consisting of DFIG, diesel and hydrogen storage system and

11The autonomy of operation permitted by a RAPS system depends on the type of technology and fuel usage such as diesel. 200

• RAPS system consisting of PMSG and hydrogen storage

The main objective behind the ﬁrst RAPS system is to identify the possibilities of replacing the existing auxiliary system components (e.g. energy storage system and dump load) of the RAPS system discussed in Section 6.4.2 of Chapter 6 with a hydrogen storage system. The autonomy of operation permitted by such a hybrid

RAPS system is enhanced by the use of hydrogen storage system. In contrast, the second type of RAPS system (i.e. PMSG with hydrogen storage system) is used to investigate the hybrid operation while allowing the full autonomy of operation, as the generated power is free from the diesel power generation. The control strategies that are implemented for each system component and the respective control coordination methodologies associated with each of the RAPS system are described in the proceeding sub-sections.

7.3.1 Hydrogen as Storage for DFIG-Diesel based Hybrid RAPS Sys-

tem

The conﬁguration of the hybrid RAPS system consists of a DFIG as the wind turbine generator, a diesel generating system as a back-up generator and a hydrogen system consisting of fuel cell system, an electrolyser and a hydrogen storage tank are shown in Fig. 7.4. The electrolyser and fuel cell systems are connected to the DC bus of the back-to-back converter system. The control strategies applied for the back-to-back converter of the DFIG and diesel generating system are identical as illustrated in

Chapter 612. The control strategies which are implemented for the hydrogen based generating system are described in the following sections.

12In this case, RSC of the DFIG is used to compensate its no-load reactive power while regulating the frequency. 201

DFIG

RSC LSC

v,f vdc Main loads Clutch G

Fuel cell system DC-DC converter-1 Engine Generator

+ - Electrolyser DC-DC converter-2

Compressor Storage tank

Figure 7.4: DFIG based wind-diesel-hydrogen RAPS system. 202 7.3.2 Coordinated Control Approach for DFIG based RAPS System

A ﬂow chart which describes the control coordination logic associated with the DFIG based RAPS system is shown in Fig. 7.5.

Wind Power Generation Pw Load Load Shedding Demand No PL No (vw)cut-in >vw>(vw)cut-out Pw=0 Pde = PL

Yes Yes

No No Diesel generator Pw > P L ∆PwL <(P ) de min "ON"

Yes Yes

Electrolyser "ON"

Fuel cell "ON"

Yes Pelz >(Pelz )max

Pitch Angle No Regulation

Frequency Regulation

Figure 7.5: Control coordination approach for DFIG based wind-diesel-hydrogen hybrid RAPS system.

The control coordinated approach given in Fig. 7.5 describes the power sharing between the system components under three situations: (a) over-generation, (b) under-generation and (c) emergency situations. The working philosophy behind the proposed control coordination strategy of the DFIG based hybrid RAPS system is as follows:

During over-generation conditions, the power output of the DFIG, Pw, is greater than the load power demand, PL where the electrolyser absorbs the additional power, 203

Pelz (i.e. ∆PwL = Pw −PL). If this excess power is greater than the maximum capacity of the electrolyser (Pelz)max, the wind turbine pitch regulation needs to be activated in order to control the wind power output. During over-generation conditions, the power balance that exists in the hybrid RAPS system can be given by (7.13).

Pw = Pelz + PL (7.13)

During under-generation conditions, the power output of the DFIG, Pw, is smaller than the load demand, PL, leading to a power deﬁcit (i.e. ∆PwL = (Pw − PL) < 0) in the RAPS system. Two options are available for consideration depending on the magnitude of the demand-generation mismatch ∆PwL. When ∆PwL is less than a minimum loading level β0, where β0 is a fraction of the rated load demand (PL)rated, the fuel cell Pfc, needs to supply the required deﬁcit power together with the DFIG. Otherwise, the synchronous machine needs to perform as a generator to satisfy the power requirement of the load demand. These two operating conditions can be expressed using (7.14):   Pw + Pfc ∆PwL < β0(PL)rated PL = (7.14)  Pw + Pde otherwise

During emergency situations, unavailability of power output from wind turbine generator due to wind speed being below cut-in level or above cut-out level, a load shedding scheme can be implemented where the reduced load is then supplied by the diesel generator as given in (7.15).

Pde = PL (7.15) 204

It is to be noted that, some of the practical issues associated with the hydrogen based generating scheme such as warming up time requirement and safety hazards have not been taken into the consideration. The proposed coordinated control approach is realised by developing individual control strategies for each system component which are explained in the following sub-sections.

(a) Fuel cell system controller

The fuel cell system is interfaced with the DC bus using a boost converter where the adopted control strategy is shown in Fig. 7.6. In addition to satisfying the generation-demand mismatch, the fuel cell system is utilised to extract the maximum power from wind. In this regard, the reference current of the fuel cell system (ifc)ref , is generated by considering the optimum wind power (Pw)opt as one of the inputs to the controller. As stated in Section 7.3.2, the controller operation of the fuel cell system is enabled only when the demand-generation mismatch is less than β (typically in the range of 0.2-0.3) a fraction of the rated load demand which can be given by

(7.16):  (PL−(Pw)opt)  if (∆PwL) < β0(PL)rated vfc (ifc)ref = (7.16)  0 otherwise

where, vfc is the fuel cell voltage The estimation of the size of a fuel cell system is extremely application specific and depends on many factors such as wind profile and wind generator capacity. In this study, the fuel cell system is sized to provide 25% of the rated load demand which also satisfies the constraints associated with the rating of back-to-back converter system given by (7.17). An aggregated fuel cell system which consists of four fuel cell units each rated 0.075 pu is connected in series to provide the required power. The V-I characteristic of one of the fuel cell units is shown in Fig. 7.713.

13The corresponding parameters associated with the fuel cell system is listed in Appendix D. 205

(Pfc)max ≤ smax × (PDFIG)rated (7.17)

where, smax is maximum allowable slip and (PDFIG)rated is rated capacity of the DFIG.

(Pfc)ref (ifc )ref To boost PI ÷ + - + - converter Limiter Comparator

vfc ifc Triangular carrier waveform

Figure 7.6: Switching signal of boost converter of the fuel cell, ((Pfc)ref = PL − (Pw)opt).

800

700

600 (rated current and voltage) Voltage (V) 500

400 (280,400) 0 50 100 150 200 250 (a)

150

100 (112kW) (rated power)

50 Power (kW)

0 0 50 100 150 200 250 (b) Current(A)

Figure 7.7: (a)V-I and (b) Power characteristics of the SOFC fuel cell system.

(b) Electrolyser and Hydrogen storage systems and its controller 206

As stated in Section 7.3.2, the operation of the electrolyser is enabled during over-

generation conditions. When the electrolyser is in operation, the optimal wind power

(Pw)opt, is extracted by estimating the reference current of the electorlyser (ielz)ref as given by (7.18). A buck converter is used to interafce the electrolyser system with the

DC bus and its corresponding control strategy is shown in Fig. 7.8. The capacity of

the electorlyser is selected as 0.25 pu of the rated load in order to satisfy the inverter

constraints associated with the DFIG given by (7.19). In this study, an aggregated

model of an electrolyser which consists of three electrolyser units, each rated 0.083

pu are connected in parallel to generate 0.25 pu power. The V-I characteristic of one

of the electrolyser units is shown in Fig. 7.914.  (Pw)opt−PL  if (∆PwL) > 0 velz (ilez)ref = (7.18)  0 otherwise

where, velz is electrolyser voltage

(Pelz)max ≤ smax × (PDFIG)rated (7.19)

(Pelz )ref (ielz )ref To buck converter ÷ + - PI + - ≥ 0 Limiter

velz ielz Triangular carrier waveform Figure 7.8: Switching signal of buck converter of the electrolyser.

14The parameters associated with the electrolyser are listed in Appendix D. 207

600

500

400

300 Volatge (V) 200

100

0 50 100 150 200 250 300 Current (A)

Figure 7.9: V-I characteristic of a electrolyser unit.

7.3.3 PMSG based Wind-Hydrogen Hybrid RAPS System

The arrangement of the PMSG based RAPS system consisting of a hydrogen generating scheme is shown in Fig. 7.10. The adopted control coordination strategy is depicted in Fig. 7.11. The proposed control strategies for PMSG remain similar to the methodologies which were discussed in Section 3.3.2 of Chapter 3. Furthermore, the control strategies implemented for the hydrogen generating system remain similar to the approaches that were presented in Section 7.3.2.

7.3.4 Coordinated Control Approach for PMSG based RAPS System

Control coordination logic associated with the decision making process of active power sharing between the diﬀerent system components covering three diﬀerent situations:

(a) over-generation, (b) under-generation and (c) emergency situations are depicted in Fig. 7.11. During over-generation conditions where the power output of the PMSG

Pw, is greater than the load power demand PL, the electrolyser is used to absorb the 208

Full bridge PMSG rectifier DC/DC converter-3 Inverter i G dc

vdc

Wind turbine vdc v, f

Main loads Fuel cell system DC-DC converter-1

+ - Electrolyser DC-DC converter-2

Compressor Storage tank Figure 7.10: PMSG based wind-hydrogen RAPS system.

excess power given by (Pw − PL). If the excessive generation ∆PwL = Pw − PL, is greater than the maximum capacity of the electrolyser (Pelz)max, then the wind turbine pitch regulation activates to limit the power output from the wind turbine

generator. During over-generation conditions, the power balance of the system is

given by (7.20).

Pw = Pelz + PL (7.20)

In under-generation situations, where the power output of the PMSG, Pw, is

samller than the load demand, PL (i.e. ∆PwL = Pw- PL)<0), leading the fuel cell system to operate to satisfy the demand-generation mismatch which is given by (7.21).

Pw + Pfc = PL (7.21) 209

During emergency situations, the wind generator does not supply any power to the loads. In such situations, load shedding schemes need to be employed and then the reduced power is supplied by the fuel cell system. Under emergency conditions, the power balance of the system can be given by (7.22).

Pfc = PL (7.22)

It is assumed that the fuel cell and electrolyser systems are available throughout the operation to maintain the power balance of the RAPS system.

Wind Power Generation Load Pw Demand PL No (vw)cut-in >vw>(vw)cut-out Pw= 0

Yes

P > P No Fuel cell w L "ON"

Yes

No Load Electrolyser "ON" Pw+Pfc = PL Shedding Yes

Yes Pelz >(Pelz )max

Pitch Angle No Regulation

Frequency Regulation

Figure 7.11: Control coordination approach for PMSG based wind-hydrogen hybrid RAPS system. 210 7.4 Performance of the Hybrid Wind-diesel-Battery RAPS

System with Hydrogen as Energy Storage

The performance and suitability of the proposed control strategies and their corresponding coordination methodologies of both RAPS systems (i.e. DFIG and PMSG) are investigated under variable wind and load conditions. In this regard, the system response, power sharing between the system components, maximum power extraction capabilities of wind turbine generators and the component level behaviour of the hydrogen generating systems are presented. The parameters associated with each type of RAPS system is listed in Appendix D.

7.4.1 Performance of the DFIG-Diesel-Hydrogen based RAPS System

(a) System Response

The entire hydrogen based RAPS system has been simulated under variable wind and load conditions. Fig. 7.12 shows the system response whereas Fig. 7.13 shows the power sharing between diﬀerent system components. The wind condition under which the system has been simulated is shown in Fig. 7.12-(a). It can be seen that the wind velocity is set initially at 12 m/s and at t = 4 s, wind velocity drops to 9 m/s. Also, at the beginning, the load demand is set at 0.6 pu. At time t = 3 s, the load is increased to a value of 0.775 pu resistive as shown in Fig. 7.13-(e). The load side voltage is shown in Fig. 7.12-(b). It is found that the voltage is not aﬀected by the load step addition at t = 3 s and the wind speed drop at t = 4 s respectively. The load side voltage stays within ±1% except during the transition of operating mode of the synchronous machine which occurs at t = 9 s. The operating frequency of the

RAPS system is shown in Fig. 7.12-(c) which is closely regulated at its rated value of 211

1 pu throughout the operation. The operating frequency of the RAPS system is not seen to be aﬀected by the wind speed or load changes. Upon close examination it can be seen that, with the mode transition of the synchronous machine which occurs at time t = 6.75 s, the frequency excursion of the RAPS system is comparatively high compared to other transient changes (e.g. load step changes) due to wind and load variations. The highest frequency deviation is limited within ±0.25 Hz. The DC link voltage of the DFIG is depicted in Fig. 7.12-(d). The simulated behaviour of the DC link voltage shows that it is well regulated at its rated value throughout the operation except in the event of load step addition which occurs at t = 3 s and mode transitions of the synchronous machine that happen at t = 6.75 s and t = 9 s.

V_w (m/s) 5 1 2 3 4 5 (a) 6 7 8 9 10

1.05

1 V_L (pu) 0.95 1 2 3 4 5 (b) 6 7 8 9 10

1.005

1 f_L (pu) 0.995 1 2 3 4 5 6 7 8 9 10 (c) 1.2

V_DC (pu) 0.8 1 2 3 4 5 (d) 6 7 8 9 10 Time (s) Figure 7.12: Response of the DFIG based wind-diesel-hydrogen RAPS system under variable wind and load conditions: (a) wind speed, (b) voltage on load side (c) frequency on load side, and (d) DC link voltage. 212

(b) Active power sharing between system components

The wind power variation of the system is shown in Fig. 7.13-(a). For simulation purposes, the slip of the wind turbine is set initially to s=-0.1 which corresponds to super synchronous mode of operation. According to the wind turbine characteris- tics15, the corresponding maximum power output of the wind generator is 0.73 pu at a shaft speed of 1.2 pu for the wind speed of 11 m/s. From Fig. 7.13-(a), the wind power of the machine is seen to rise to a value of 0.7 pu. At this time the load demand is set to a value of 0.6 pu. The additional power is consumed by the electrolyser until t = 3 s as shown in Fig. 7.13-(d). When the load step increase occurs at t = 3 s, the generation-demand mismatch, ∆PwL = Pw − PL, leads to an under generation scenario which is smaller than 20% of the rated capacity of the system thus enabling the operation of the fuel cell system as shown in Fig 7.13-(c). At t = 6.75 s, the generation-demand mismatch of the RAPS system, ∆PwL = Pw − PL, becomes greater than 20% of the rated capacity of the load demand which activates the state transition of the synchronous machine from a synchronous condenser to a generator mode of operation as shown in Fig. 7.13-(b). With the load step decrease which occurs at t = 9 s, the operating mode of the synchronous machine changes from generating mode to condenser mode. Also, with the load step reduction, the demand-generation mismatch under an over-generation scenario causes the electrolyser to start absorbing the additional energy associated with the RAPS system.

The maximum power extraction of the wind turbine generator is shown in Fig.

7.14. It can be seen that the wind turbine generator is able to follow its maximum power tracking except during the transient conditions which is unavoidable. The synchronous machine speed ωs, diesel engine speed ωr, and clutch signal are shown in

15Refer to Appendix A. 213

1 0.5

P_w (pu) 0 1 2 3 4 5 (a) 6 7 8 9 10 0.4 0.2

p_fc (pu) 0 1 2 3 4 5 (b) 6 7 8 9 10 0.2

0.1

p_elz p_elz (pu) 0 1 2 3 4 5 (c) 6 7 8 9 10 0.5 0

-0.5P_de (pu) 1 2 3 4 5 (d) 6 7 8 9 10 0.8 0.6

P_L (pu) 0.4 1 2 3 4 5 (e) 6 7 8 9 10 Time (s) Figure 7.13: Power Sharing of the DFIG based wind-diesel-hydrogen RAPS system under variable wind and load conditions: (a) wind power, (b) fuel cell power, (c) electrolyser power, (d) diesel power and (e) load demand.

Fig. 7.15. It can be seen that the clutch signal is enabled by setting it to logic ‘1’ at nearly t = 6.75 s when the speeds of diesel engine and synchronous machine are equal to each other.

(c)Performance of hydrogen generating system

Performance characteristics of the hydrogen based power generating system are shown in Figs. 7.16 and 7.17. The molar ﬂow rates associated with the electrolyser and fuel cell are shown in Fig. 7.16-(a) and 7.16-(b) respectively. As expected, during over-generation conditions (i.e. until t = 3 s), the molar ﬂow rate of the electrolyser rises and thereafter starts decaying due to the compressor dynamics. The fuel intake of the fuel cell is seen to rise from t = 3 s to t = 6.75 s which corresponds to an under- generation condition under which the fuel cell system starts supplying power into the 214

1 P w 0.9 (P ) w opt 0.8

0.7

0.6

0.5

Power (pu) 0.4

0.3

0.2

0.1

0 1 2 3 4 5 6 7 8 9 10 Time (s)

Figure 7.14: Maximum power point tracking characteristic of DFIG based wind- diesel-hydrogen RAPS system.

Figure 7.15: Speeds of the engine, synchronous machine and clutch signal of DFIG based wind-diesel-hydrogen RAPS system. 215 loads. The behaviour of the hydrogen storage tank pressure is shown in Fig.7.16-(c).

It is assumed that the tank pressure is maintained within 5 atm. Initially the tank pressure rises due to the generation of hydrogen by the electrolyser. At t = 3 s the fuel cell starts generating power by consuming hydrogen fuel which experiences a pressure reduction in the storage tank. The DC voltages of the fuel cell and electrolyser are shown in Fig. 7.17. The variation of the fuel cell voltage is maintained within 15% of rated value whereas the electrolyser voltage is maintained within 10% of its rated value.

-3 x 10 2

1.5

0.5

0 1 2 3 4 5 6 7 8 9 10 (nH2)_elz (Kmol/sec) (nH2)_elz

-3 (a) x 10 1.5

0.5

(nH2)_fc (Kmol/sec) 1 2 3 4 5 6 7 8 9 10 (b)

5.1

5.05

5 P_t (atm)P_t

4.95 1 2 3 4 5 6 7 8 9 10 (c) Time (s)

Figure 7.16: Performance of the hydrogen based power generation unit in DFIG based wind-diesel-hydrogen RAPS system: (a) molar ﬂow rate of electrolyser, (b) molar ﬂow rate of fuel cell and (c) tank pressure.

7.4.2 Performance of the PMSG-Hydrogen based RAPS System

Fig. 7.18 shows the system response whereas Fig. 7.19 shows the power sharing between diﬀerent system components. The wind condition under which the system has been simulated is shown in Fig. 7.18-(a). It can be seen that the wind velocity is 216

1.25 v fc 1.2 v elz 1.15

1.1

1.05

0.95 DC Voltage (pu) DC Voltage 0.9

0.85

0.8

0.75 1 2 3 4 5 6 7 8 9 10 Time (s)

Figure 7.17: Voltage characteristics of fuel cell and electrolyser of DFIG based wind- diesel-hydrogen RAPS system.

set initially at 12 m/s. At t = 3 s, the wind velocity drops to 9 m/s, then it increases

to 11 m/s at t = 5 s. Also, the initial demand is set at 0.6 pu. At time t = 4 s,

the load is increased to a value of 0.8 pu as shown in Fig. 7.18-(d). The load side

voltage is shown in Fig. 7.18-(b) which is not seen to be aﬀected by the wind speed

changes except with the load step changes which occur at time t = 4 s and t = 6 s

respectively. The load voltage stays within ±1% during its steady state operation.

The operating frequency of the RAPS system is shown in Fig. 7.18-(c). As expected,

it is regulated at its rated value of 1 pu throughout the operation. The operating

frequency is not seen to be aﬀected by the wind speed or load changes. The frequency

regulation is seen to be well regulated within ±0.05 Hz of its rated value. The DC bus voltage of the system is shown in Fig. 7.18-(d). The simulated behaviour of the

DC link voltage shows that it is regulated at its rated value throughout the operation except during the wind and load changes which occur at t = 3 s, t = 5 s and t = 4 s, t = 6 s respectively. 217

V_w (m/s) 5 1 2 3 4 5 6 7 (a) 1.05

V_L (pu) 0.95 1 2 3 (b) 4 5 6 7 1.001

1 f_L (pu) f_L 0.999 1 2 3 4 5 6 7 (c) 1.05

V_DC (pu) 0.95 1 2 3 (d) 4 5 6 7 Time (s)

Figure 7.18: Response of the PMSG based wind-hydrogen RAPS system under variable wind and load conditions: (a) wind speed, (b) voltage on load side, (c) frequency on load side, and (d) DC link voltage.

The wind power variation of the system is shown in Fig. 7.19-(a). Until t = 3 s, the RAPS system operates in an over-generation scenario where the wind power output is greater than the load demand Pw>PL, and excess power is absorbed by the electrolyser as evident from Fig. 7.19-(c). At t = 3 s the wind velocity drops to 9 m/s from 11 m/s causing a reduction in the wind power output. From t = 3 s to t = 6 s, the RAPS system experiences an under-generation scenario where the wind power output is less than the load demand i.e. Pw

0.25 pu to 0.475 pu. With the wind speed change which occurs at t = 5 s, the fuel cell system reduces its power generation to 0.125 pu. The load step reduction which occurs at t = 6 s leading to an over-generation scenario where the excess energy is absorbed by the electrolyser as shown in Fig. 7.19-(c). The maximum power tracking capability of the PMSG based RAPS system is shown in Fig. 7.20. As noted, the 218

PMSG is able to follow its maximum power point tracking characteristic during the entire operation.

0.5 P_w (pu) P_w 0 1 2 3 (a) 4 5 6 7 0.5

0 P_fc (pu) -0.5 1 2 3 (b) 4 5 6 7 0.4

0.2 P_elz (pu) 0 1 2 3 (c) 4 5 6 7 1

0.8

P_L (pu) 0.6

1 2 3 (d) 4 5 6 7 Time (s)

Figure 7.19: Power Sharing of the PMSG based wind-hydrogen RAPS system under variable wind and load conditions: (a) wind power, (b) fuel cell power, (c) electrolyser power and (d) load demand.

Performance of the hydrogen based power generating system is shown in Figs.

7.21 and 7.22. The molar ﬂow rates associated with the electrolyser and fuel cell are shown in Fig. 7.21-(a) and 7.21-(b) respectively. As expected, during over-generation

(i.e. until t = 3 s), the molar ﬂow rate of the electrolyser rises and afterwards it decays to zero. Again, at t = 6 s, the RAPS system experiences an over-generation condition which enables the hydrogen production through electrolyser as evident from

Fig. 7.21-(a). The fuel intake of the fuel cell system is seen to rise from t = 3 s to t = 5 s which corresponds to an under-generation situation. The behaviour of the hydrogen storage tank pressure is shown in Fig. 7.21-(c). It is assumed that the tank pressure is initially maintained at nearly 5 atm pressure. Upon close examination, it can be seen that, during over-generation situation the electorlyser generates hydrogen 219

1 Pw Actual P MPPT 0.9 w

0.8

0.7

0.6 Power (pu) Power 0.5

0.4

0.3

0.2 1 2 3 4 5 6 7 Time (s) Figure 7.20: Maximum power point tracking from wind in PMSG based wind- hydrogen RAPS system. by utilising excess energy available in the RAPS system leading to an increased tank pressure. However, during t= 3 s and t= 6 s the fuel cell starts generating power thus consuming the hydrogen stored in the tank leading to a reduced pressure. The

DC voltages of the fuel cell and electrolyser are shown in Fig. 7.22. The variation of the fuel cell voltage is maintained within +5% and -15% of rated value whereas the electrolyser voltage is maintained within ±5% of its rated value. 220

-4 x 10 5

0 1 2 3 4 5 6 7

(nH2)_elz (Kmol/sec) (a) -4 x 10 4

0 1 2 3 4 5 6 7

(nH2)_fc (Kmol/sec) (b) 5.02

5.01 P_t (atm)

5 1 2 3 4 5 6 7 (c) Time (s) Figure 7.21: Performance of the hydrogen based power generation unit ofPMSG based wind-hydrogen RAPS system: (a) molar ﬂow rate of electrolyser, (c) molar ﬂow rate of fuel cell and (b) tank pressure.

1.15 V fc V 1.1 elz

1.05

0.95 DC Voltage (pu) Voltage DC 0.9

0.85

0.8 1 2 3 4 5 6 7 Time (s)

Figure 7.22: Voltage characteristics of fuel cell and electrolyser of PMSG based wind- hydrogen RAPS system. 221 7.5 Chapter Summary

This chapter presented the performance analysis of a hydrogen based generating scheme which was incorporated with a standalone wind based RAPS system. In this regard, special attention was given to the component modelling of the hydrogen generating system which includes a fuel cell system, an electorlyser and a hydrogen storage tank. The performance of the hybrid RAPS systems was examined and component level behaviour of the hydrogen generating scheme was also investigated. Based on the results that were obtained through simulation studies, following conclusions can be drawn:

• The power sharing between the system components were seen to be in accor-

dance with the two distinct proposed coordinated control approaches where the

two types of wind turbine generator (i.e. DFIG and PMSG) were able to follow

the maximum power point tracking characteristics throughout their operation.

• The performance of the hydrogen generating system exhibited the anticipated

behaviour during over-generation and under-generation situations demonstrat-

ing robustness of the proposed control strategies for the each component (i.e.

fuel cell system, electrolyser and hydrogen storage tank). In addition, the hy-

brid operation with the coordination between the components (i.e. fuel cell,

electrolyser and storage tank) of the hydrogen generating scheme ensured its

suitability as a self generating unit for remote wind power applications.

• This chapter mainly investigated the technical feasibility of the hydrogen based

RAPS systems. The simulation results showed that voltage, frequency and

power balance were well regulated while satisfying the operating requirements16,

16Permissible voltage and frequency variations 222 thereby validating the applicability of the proposed RAPS systems in real applications. The hydrogen generating scheme has not been assessed with respect to its economic viability which can lead to limitations in real life situations. With the ongoing research activities in the area of hydrogen storage systems, it can be expected that economics associated with the hydrogen generating scheme would be fabvourable in the future. Chapter 8

Conclusions and Recommendations for Future Work

8.1 Conclusions

Through the comprehensive literature review undertaken, it was revealed that the existing research outcomes on renewable energy based hybrid RAPS systems have received limited research attention when compared to grid connected counterparts.

In this thesis, research has been conducted in relation to the following aspects of

RAPS systems: (a) modelling of the system components, (b) development of control strategies for each system component, (c) development of methodologies for robust control coordination of system components and (d) development of methods to capture optimal power from the renewable energy. This thesis has addressed above stated challenges in relation to high penetrated wind based RAPS systems. In this regard,

DFIG and PMSG were selected as the preferred wind turbine generator technologies due to their fast growing popularity over other types of wind turbine generators (e.g. induction generator) for remote wind power applications. Moreover, these two wind turbine generators: DFIG and PMSG were selected with a view to design a structure 223 224 of a hybrid RAPS system to supply in an island or a village or rural and regional areas where main grid supply is not available. The main objective of the work presented in this thesis was to achieve robust voltage and frequency regulation of the RAPS systems while capturing the optimum power from wind.

Several approaches have been proposed in the existing literature for controlling the power electronic converters associated with wind turbine generators. Among which vector control has been widely advocated for most of the industry standard wind power applications. Therefore, ﬁeld oriented vector control has been employed for the DFIG and PMSG based wind turbine generators considered in this thesis. In this regard, the controllers associated with wind turbine generators were developed with an aim to regulate the voltage and frequency on the load side. In order to investigate the suitability of suggested control strategies applied for the wind turbine generators, dump loads were integrated to form hybrid RAPS systems. Two diﬀerent dump load arrangements with their respective control strategies were developed. Dump load operation was enabled only during over-generation situations. The dump load, which was employed for the DFIG was connected to the load side (i.e. AC side), whereas the dump load arrangement that was used with the PMSG, was located on the DC bus of the power electronic arrangement. The performance of the RAPS systems were investigated in Chapter 3 through simulations. It was observed that the proposed hybrid RAPS systems consisting of a wind turbine generator and a dump load were able to maintain the voltage and frequency within acceptable limits. Moreover, the behaviour of the dump loads showed their ability in utilising the excessive power available in the RAPS system during over generation situations, thus verifying the suitability of the proposed control approaches.

As a continuation of the work described in Chapter 3, integration of an energy storage system to a wind based RAPS system was examined in Chapter 4. Typi- 225 cally battery storage systems can be characterised by their high energy density levels making the choice feasible or attractive for remote area wind energy systems. Two diﬀerent control approaches were suggested for the battery storage systems which were incorporated with the DFIG and PMSG. In this regard, direct active power based control approach has been employed for the battery storage system connected to the DFIG based RAPS system whereas the DC bus voltage regulation based control approach was proposed for the PMSG based counterpart. In addition, a control coordination methodology was developed which performed as the central controller of the RAPS systems. The control coordination strategy consisting of the decision making process associated with the active power sharing between the system components were developed. Moreover, two diﬀerent control approaches were implemented for the DFIG and PMSG to extract the maximum power from wind. The maximum power from DFIG was realised through the operation of a battery storage system and a dump load which can be regarded as an indirect mechanism for control. In contrast, the maximum power extraction from wind using the PMSG was achieved using the control strategy that was implemented for its boost converter. Through simulation studies it has been demonstrated that both RAPS systems were capable in maintaining the voltage and frequency regulation within satisfactory levels throughout their operation while extracting the maximum power from wind. Also, it has been noted that the power sharing between the system components were accomplished in accordance with the proposed coordinated control methodology. In addition to the detailed model, corresponding linearised model of the wind-battery storage system was also undertaken with a view to compare the results obtained through the respective detailed models. It was found that linearised and detailed models exhibited similar behaviour in relation to the power sharing between the system components.

In real life situations, the operation of battery storage systems need careful con- 226 siderations such as avoidance of heavy DOD rates and reduced ripple content in the battery current. To mitigate these issues, a supercapcitor was integrated with the battery storage system to form the hybrid energy storage (i.e. battery and supercapacitor) and its performance was investigated in Chapter 5. Diﬀerent power electronic conﬁgurations were employed for the hybrid energy storage systems and their respective energy management algorithms were developed with an aim to achieve lower

DOD rates and reduced ripple in the battery current. Through simulation studies, it has been demonstrated that the battery energy storage systems incorporated with each type of RAPS system were able to operate with reduced ripple and DODs. Fur- thermore, it has also been noted that the supercapacitors integrated to the DFIG and

PMSG were able to handle the transients caused by wind speed and load changes ef- fectively. Also, it was observed that the supercapacitor was capable in absorbing high frequency power component associated with the demand-generation mismatch while leaving the absorption of steady state component for the battery storage. Moreover, the hybrid operation of the RAPS systems were capable in maintaining the voltage and frequency within satisfactory limits while allowing wind turbine generators to operate on their maximum power extraction modes.

RAPS systems are often characterised by their poor reactive power capabilities and uncertainties of power generation due to the intermittency associated with wind.

In order to provide a suitable solution for these issues, a diesel generating system which can be operated in dual mode (i.e. as a synchronous generator and synchronous condenser) was integrated to the wind based RAPS system and its performance was investigated in Chapter 5. In this regard, complete modelling aspects of the diesel generating system together with its respective control strategies were proposed. Per- formance of the diesel generating system was optimised by avoiding its operation at low load conditions. Moreover, a control coordination methodology was developed 227 between the system components with a view to achieve three objectives which were:

(a) to maintain active and reactive power balance of the RAPS systems, (b) to operate wind turbine generator in maximum power point tracking mode and (c) to operate the diesel generator above its minimum loading condition. In addition, diﬀerent reactive power control strategies were implemented on DFIG and PMSG based RAPS systems. In this regard, DFIG control was designed to operate at unity power factor while diesel generating was used to meet all reactive power requirements of the

RAPS system. An alternative scheme for reactive power control was undertaken in relation to PMSG based RAPS system by making the reactive power sharing in an arbitrary manner. Simulation studies were conducted on the DFIG and PMSG based

RAPS systems and all above stated objectives were realised under changing wind and load conditions. Through simulation studies it was demonstrated that the proposed reactive power control approaches were capable in satisfying the reactive power requirement of the RAPS systems. In addition to the detailed model, a linearised model of the wind-battery storage system was also undertaken with a view to compare the results obtained through detailed models. It was shown that both linearised and detailed models exhibited similar behaviour in relation to power sharing between the system components.

With the increasing importance placed on energy security and clean power generation technologies, hydrogen generating systems can be regarded as one of the emerging technologies. There are various challenges which exist in such generating technologies in terms of ﬁnancial and technological perspectives. However, this thesis investigated the technical feasibility of such a hydrogen generating system for a remote wind application by giving due attention to the modelling aspects of the system components and their corresponding control strategies. Hydrogen generating system consisting a fuel cell system, an electrolyser and a hydrogen tank was modelled using 228 the methods available in the literature. The individual models of the fuel cell system, electorlyser and hydrogen storage tank were integrated to perform as a single self generating unit. The role of a hydrogen generating system for the wind energy systems was investigated under following conditions: (a) as a replacement for auxiliary system components (i.e. battery storage and dump load) in the wind-diesel

RAPS system and (b) as a back up generator for wind based RAPS system. The ﬁrst option was examined in relation to the DFIG based RAPS system while the second option was adopted with the PMSG based RAPS system. In this regard, control strategies of hydrogen generating system were designed and implemented based on their respective coordinated control approaches. Through simulation studies it was demonstrated that the hydrogen based generating system can perform as an energy storage system or as a back up generating system in a standlone wind applications.

Also, it has been noted that RAPS systems were capable in regulating the voltage and frequency within satisfactory limits throughout the operation while capturing maximum power from wind. Furthermore, it was observed that the fuel cell system was able to generate power during under-generation conditions by using hydrogen available from a storage tank while the electorlyser produces hydrogen by utilising the excessive power available during over-generation of the RAPS system.

This thesis presented several possible wind based RAPS conﬁgurations while giving a signiﬁcant attention to their component level modelling aspects, control coordination methodologies and individual control strategies for each system component.

In addition, control coordination strategies were developed with a view to maintain the power balance of the RAPS systems. Through simulation studies, it was demonstrated that the robust control strategies proposed for hybrid RAPS systems were able to achieve voltage and frequency regulation under variable wind and load conditions.

A comparative study has also been undertaken to evaluate the diﬀerent types 229 of RAPS systems illustrated in this thesis. Technical aspects together with simulation results have been taken into the consideration to conclude/compare each type of RAPS system. Table 8.1 shows the qualitative comparison of diﬀerent types of

RAPS systems in terms of their voltage regulation, frequency regulation and MPPT capability.

Table 8.1: Qualitative comparison of diﬀerent types of RAPS systems

Performance characteristics Type of RAPS system Voltage regulation Frequency regulation MPPT capability DFIG-battery storage and dump load • Voltage regulation is • Better frequency regu- • Better track- (Chapter 4 - Section satisfactory. lation3 is obtained. ing charac- 4.3.4) • Better transient per- • This is due to the fact teristics are (TYPE I) formance is obtained that inverter controls obtained ex- compared to its (i.e. RSC) and in- pect during PMSG counterpart1. ertial support of the the transient • However, in this case, DFIG help in main- conditions. only the DFIG is avail- taining the frequency able to supply the re- within tight limits. active power demand of the RAPS system. Further, DFIG has its own limitations2 with regard to reactive power generation.

1As an example better performance is observed during the start-up process of an induction motor driven pump load illustrated in Section 4.4.2 of Chapter 4. 2e.g. stator and rotor current, inverter rating etc. 3This includes the steady state and transient responses of the RAPS system. 230

PMSG-battery storage and dump load • Voltage regulation is • Acceptable frequency • Better track- (Chapter 4 - Section satisfactory. regulation is achieved. ing charac- 4.3.5) • Better steady state • However, the tran- teristics are (TYPE II) performance is ob- sient response is less obtained ex- tained. Typically, the favourable compared pect during performance is mainly to TYPE I RAPS the transient based on the PI con- system. This is due to conditions5. troller gains and their no or lack of inertial bandwidths associated support provided by with inverter. the inverter. Also, it • However, in this case is to be noted that inverter associated the PMSG is fully with the PMSG has decoupled through its own reactive power the power electronic limitations4. interfaces. There- fore inertia of the PMSG provides no contribution towards frequency restoration of the RAPS system.

DFIG-battery storage- • Improved voltage reg- • Improved frequency • Similar per- supercapacitor and ulation regulation formance is dump load • Able to offer better • This is because, the observed as (Chapter 5 - Section transient response due fast varying power in the case 5.2.4) to the presence of the fluctuations are able of TYPE (TYPE III) supercapacitor. This to be absorbed by I RAPS is because, the DC the supercapacitor system. voltage variations are and therefore, better absorbed by the su- frequency response is percapaitor and hence anticipated. DC bus voltage variations are not reflected on to the AC side voltage. • However, the DFIG reactive power limitations still exist as indicated in TYPE I RAPS system.

4e.g. inverter rating 5However the tracking characteristics are much faster compared the DFIG based RAPS systems owing the lower inertia associated with the PMSG. 231

PMSG-battery storage- • Improved voltage reg- • Improved frequency • Similar per- supercapacitor and ulation regulation compared formance is dump load • Able to oﬀer even bet- to TYPE II RAPS observed as (Chapter 5 - Section ter transient perfor- system. As similar to in the case 5.2.5) mance compared to the case in TYPE III of TYPE (TYPE IV) TYPE II RAPS sys- RAPS systems, the II RAPS tem due to the pres- supercapacitor is able system. ence of the superca- to absorb the rapid pacitor thus mitigat- power variations. ing the DC bus ﬂuctu- • However, the tran- ations. sient response is still • However, reactive less favourable than power limitations TYPE III due to lit- involved with the tle or no contribution PMSG inverter reac- from the actual PMSG tive power limitations inertia6. still exist as in the case of TYPE II RAPS system.

DFIG-diesel generator-battery • Excellent voltage reg- • Excellent frequency • Similar per- storage and dump ulation regulation formance is load • Diesel generator sup- • This is due to the fact experienced (Chapter 6 - Section plies entire reactive that diesel generating as in TYPE 6.4.2) power demand of the system operates as a I RAPS (TYPE V) RAPS system by op- back-up power gener- system. erating as a condenser ator. Furthermore, in in WO mode and as the presence of diesel a generator in WD generator, the RAPS mode. system is able to • However, the state provide an additional transition that occurs active/reactive power between the compo- and inertial support nents (e.g. WO to compared to TYPE I WD) are seen to cre- or TYPE III RAPS ate voltage variations systems. in the RAPS system. • However, extra care Therefore, extra care needs to be taken to needs to be taken in minimise the tran- the design process to sient eﬀects that ensure better transient arise due to the state response. transitions occur between the system components (e.g. synchronous condenser to generating mode).

6PMSG is decoupled from the load side through the power electronic arrangement. 232

PMSG-diesel generator-battery • Excellent voltage reg- • Excellent frequency • Similar per- storage and dump ulation regulation formance is load (Chapter 6 - • Diesel generating sys- • Diesel generating sys- experienced Section 6.4.3) (TYPE tem and inverter of the tem provides better as in TYPE VI) PMSG supply reactive active/reactive power II RAPS power demand of the and inertial support as system. RAPS system. in the case of TYPE V • A suitable reactive RAPS systems. How- power coordination7 ever, the transient re- approach needs be em- sponse of this type of ployed for improved RAPS system is less performance. improved compared to the TYPE V RAPS system.

DFIG-diesel generator and Hydrogen storage • Similar performance • Similar performance is • Similar per- (Chapter 7 - Section is experienced as in obtained as in the case formance is 7.3.1) the case of TYPE V of TYPE V RAPS sys- observed as (TYPE VII) RAPS system. tem. in TYPE • This is due to the • This is because, the I RAPS fact that diesel gen- active power sharing system. erating system sup- between the system plies the entire re- components is similar active power demand to the case presented of the RAPS system for TYPE V. while the DFIG is controlled to operate at unity power factor.

PMSG and Hydrogen storage • Only the inverter asso- • Similar performance • Similar per- (Chapter 7 - Section ciated with the PMSG as in the case of TYPE formance is 7.3.3) supplies the reactive I RAPS system. observed as (TYPE VIII) power demand of the • This is due to the in the case RAPS system. There- fact that the hydrogen of TYPE fore, similar perfor- storage system substi- II RAPS mance is obtained as tutes the duties per- system. in the case of TYPE I formed by the battery RAPS system. storage and dump load as in the case of TYPE I RAPS system.

7An uncoordinated reactive power control is illustrated in Section 6.4.3 of Chapter 6. 233 8.2 Recommendations for Further Work

As extensions to the work presented in this thesis, following is a description of further activities that can be undertaken in relation to standalone RAPS system:

• Development of the control strategies for each component of RAPS

systems with a view to operate them under unbalanced load condi-

tions

The work presented in this thesis mainly examined the performance of the

RAPS systems under steady state operation and balanced conditions. How-

ever, in real life conditions, RAPS systems may be exposed to unbalanced load

conditions. In such situations, the RAPS system may experience diﬃculties

in regulating the voltage and frequency within acceptable limits. Therefore,

existing control strategies needs to be further ﬁne tuned in order to ensure

robust voltage and frequency regulation. In this regard, the existing positive

vector control approach that has been employed for wind turbine generators

can be further extended in the negative sequence domain. In addition, the con-

trol strategies that were applied for other system components such as battery

storage systems can be modiﬁed to handle the unbalanced situations.

• Design of a protection strategy for hybrid RAPS systems

RAPS systems may experience symmetrical and unsymmetrical fault con-

ditions. In this regard, wind turbine generators need to be protected using

suitable protection schemes such as crow-bar protection systems. In addition,

the control strategies can be further extended for the system components by

considering various contingencies of RAPS operation, such as prioritising the

operation of components in the event of an emergency and minimising the tran-

sient eﬀects. 234

• Integration of other types of renewable energy systems (e.g. solar

photovoltaic) to wind based RAPS systems.

All RAPS system described in this thesis were based on wind as a renewable

energy source. However, some remote areas may have a potential for power

generation by more than one renewable energy sources (e.g. wind and solar).

Therefore, utilisation of a number of renewable energy sources increases the

penetration level of renewable energy proportion in a RAPS system thus leading

to reduced carbon emission. However, the integration of additional renewable

energy sources into the existing wind based RAPS systems requires redesigning

of the RAPS system conﬁgurations with additional power electronic components

and their respective controllers. In addition, the existing control strategies of

wind turbine generators and other system components (e.g. battery storage

and dump load) needs to be modiﬁed by considering the eﬀect of additional

renewable energy sources.

• Further development of the existing control strategies of the RAPS

systems to operate as a grid interactive micro-grid

Wind based RAPS systems examined in this thesis were solely developed

assuming that they operate in standalone environments. However, with the

rapid developments that take place in remote areas would expect grid connected

power supplies in the future. In such situations, existing standalone RAPS

systems can be connected to grid supply systems to operate as grid-interactive

microgrids. In this regard, the existing control strategies need to be investigated

further with a view to operate the RAPS systems in grid-connected as well as in

islanding mode. Furthermore, the power quality aspects of such RAPS systems

needed to be investigated in order to fulﬁl requirements of the standards (e.g. 235

IEEE 1547, AS4777 etc.) associated with the grid-integration of the microgrids.

Moreover, the LVRT capabilities of the wind turbine generators can be assessed

and control strategies should be developed to improve (low voltage ride through

capability) LVRT capability.

• Optimise the size of the system components and apply them to ex-

isting detailed simulation models

RAPS systems presented in this thesis have been given the necessary emphasis

to examine the technical feasibility of high wind penetration. The component

selection was made assuming speciﬁc operating conditions (e.g. battery storage

is sized to satisfy a speciﬁc fraction of the load demand). However, the operation

of such RAPS systems can be optimised by selecting the suitable system com-

ponents by considering the ﬁnancial and technical aspects into account. In this

regard, appropriate algorithms can be developed by taking the technical, ﬁnan-

cial and environmental constraints as inputs (e.g. technical inputs are allowable

voltage and frequency limits and ﬁnancial inputs are number of hours that each

generator is required to operate, operating cost and initial investments) and

environmental aspects such as limitations associated with greenhouse gas emis-

sion levels. The outcomes (i.e. appropriate sizes of the components) from such

algorithms can be used in the simulation models of the RAPS systems with

minor modiﬁcations to their respective control strategies.

• Development of prototype models for experimental validation of the

results

Simulation studies that were illustrated in this thesis can be experimentally 236 validated through laboratory set-up. In this case, a prototype model of a small hybrid RAPS system can be developed. The hybridisation of diﬀerent system components should be tested for eﬀective operation and control. Furthermore, the proposed coordinated control approach that proposed for each type of RAPS system can be tested using available test facilities. References

[1] K. Rajashekara. Hybrid fuel-cell strategies for clean power generation. IEEE

Transactions on Industry Applications, 41(3):682 – 689, May - Jun. 2005.

[2] T. Zhou and B. Francois. Energy management and power control of a hybrid ac-

tive wind generator for distributed power generation and grid integration. IEEE

Transactions on Industrial Electronics, 58(1):95 – 104, Jan. 2011.

[3] G. Weidong and Y. Zhuoyong. Research on non-grid-connected wind

power/water-electrolytic hydrogen production system. International Journal of

Hydrogen Energy, In Press, May 2011.

[4] D. P. Kaundinya, P. Balachandra, and N. H. Ravindranath. Grid-connected

versus stand-alone energy systems for decentralized power-a review of literature.

Journal of Renewable and Sustainable Energy Reviews, 13(8):2041 – 2050, Oct.

2009.

[5] M. Datta, T. Senjyu, A. Yona, T. Funabashi, and C. Kim. A coordinated con-

trol method for leveling PV output power ﬂuctuations of PV and diesel hybrid systems connected to isolated power utility. IEEE Transactions on Energy Con-

version, 24(1):153 – 162, Mar. 2009.

[6] G. Lalor, A. Mullane, and M. O’Malley. Frequency control and wind turbine

technologies. IEEE Transactions on Power Systems, 20(4):1905 – 1913, Nov.

2005.

[7] Advanced management strategies for remote-area power-supply systems. Journal

of Power Sources, 133(1):141 – 146, May 2004.

237 238

[8] M. Liserre, T. Sauter, and J. Y. Hung. Future energy systems: Integrating re-

newable energy sources into the smart power grid through industrial electronics.

IEEE Industrial Electronics Magazine, 4(1):18 – 37, Mar. 2010.

[9] T. Ackermann. Wind Power in Power Systems. Jhon Weily and Sons, Ltd, 2005.

[10] 20% of wind energy by 2030, increasing wind energy’s contribution to us elec-

tricity supply. Technical report, US Department of Energy, July 2008.

[11] T. S. Ustun, C. Ozansoy, and A. Zayegh. Recent developments in microgrids and

example cases around the world–a review. Renewable and Sustainable Energy

Reviews, 15(8):4030 – 4041, Oct. 2011.

[12] K. Malmedal, B. Kroposki, and P. K. Sen. Distributed energy resources and

renewable energy in distribution systems: Protection considerations and pene-

tration levels. In IEEE Industry Applications Society Annual Meeting, 2008 (IAS

’08), Edmonton, Alberta, Canada, pages 1 – 8, Oct. 2008.

[13] T. H. M. El-Fouly, H. H. Zeineldin, E. F. El-Saadany, and M. M. A. Salama.

Impact of wind generation control strategies, penetration level and installation

location on electricity market prices. IET Journal of Renewable Power Genera-

tion, 2(3):162 – 169, Sept. 2008.

[14] E. Rikos, S. Tselepis, C. Hoyer-Klick, and M. Schroedter-Homscheidt. Stability

and power quality issues in microgrids under weather disturbances. IEEE Journal

of Applied Earth Observations and Remote Sensing, 1(3):170 –179, Sept. 2008.

[15] D. Weisser and R. S. Garcia. Instantaneous wind energy penetration in isolated

electricity grids: Concepts and review. Journal of Renewable Energy, 30(8):1299

– 1308, Jul. 2005. 239

[16] Andreas Peterson. Analysis, Modeling and Control of Doubly-Fed-Induction Gen-

erators for Wind Turbines. PhD thesis, Division of Electric Power Engineering,

Department of Energy and Environment, Chalmers University of Technology,

Goteborg, Sweden, 2005.

[17] J. Smith, R. Thresher, R. Zavadil, E. DeMeo, R. Piwko, B. Ernst, and T. Ack-

ermann. A mighty wind. IEEE Power and Energy Magazine, 7(2):41 – 51, Mar.

- Apr. 2009.

[18] E. A. DeMeo, G. A. Jordan, C. Kalich, J. King, M. R. Milligan, C. Murley,

B. Oakleaf, and M. J. Schuerger. Accommodating wind’s natural behavior. IEEE

Power and Energy Magazine, 5(6):59 – 67, Nov. - Dec. 2007.

[19] N. Urasaki T. Funabashi H. Fujita T. Senjyu, R. Sakamoto and H. H. Sekine.

Output power leveling of wind turbine generator for all operating regions by

pitch angle control. IEEE Transaction of Energy Conversion, 21(2):467 – 475,

Jun. 2006.

[20] E. Koutroulis and K. Kalaitzakis. Design of a maximum power tracking sys-

tem for wind-energy-conversion applications. IEEE Transactions on Industrial

Electronics, 53(2):486 – 494, Apr. 2006.

[21] 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, 58(1):29 –36,

Jan. 2011.

[22] J. G. Slootweg and W. L. Kling. Is the answer blowing in the wind? IEEE

Power and Energy Magazine, 1(6):26 – 33, Nov. - Dec. 2003. 240

[23] H. Li and Z. Chen. Overview of diﬀerent wind generator systems and their

comparisons. IET Journal of Renewable Power Generation, 2(2):123 – 138, Jun.

2008.

[24] M. E. Haque, M. Negnevitsky, and K. M. Muttaqi. A novel control strategy for

a variable-speed wind turbine with a permanent-magnet synchronous generator.

Industry Applications, IEEE Transactions on, 46(1):331 – 339, Jan. - Feb. 2010.

[25] R. Pena, J. C. Clare, and G. M. Asher. A doubly fed induction generator using

back-to-back pwm converters supplying an isolated load from a variable speed

wind turbine. IEE Electric Power Applications, 143(5):380 – 387, Sept. 1996.

[26] D. G. Forchetti, G. O. GarcIa, and M. I. Valla. Adaptive observer for sensorless

control of stand-alone doubly fed induction generator. IEEE Transactions on

Industrial Electronics, 56(10):4174 – 4180, Oct. 2009.

[27] M. Shahabi, M. R. Haghifam, M. Mohamadian, and S. A. Nabavi-Niaki. Mi-

crogrid dynamic performance improvement using a doubly fed induction wind

generator. IEEE Transactions on Energy Conversion, 24(1):137 –145, Mar. 2009.

[28] G. Iwanski and W. Koczara. Sensorless direct voltage control method for stand-

alone slip-ring induction generator. In 11th European Conference on Power Elec-

tronics and Applications, Dresden, Germany, Sept. 2005.

[29] G. Iwanski and W. Koczara. Sensorless direct voltage control of the stand-

alone slip-ring induction generator. IEEE Transactions on Industrial Electronics,

54(2):1237 – 1239, Apr. 2007.

[30] M. Fatu, L. Tutelea, I. Boldea, and R. Teodorescu. Novel motion sensorless

control of stand alone permanent magnet synchronous generator (PMSG): Har- monics and negative sequence voltage compensation under nonlinear load. In 241

European Conference on Power Electronics and Applications, 2007, pages 1 –

10, Sept. 2007.

[31] Q. Li, S. S. Choi, Y. Yuan, and D. L. Yao. On the determination of battery

energy storage capacity and short-term power dispatch of a wind farm. IEEE

Transactions on Sustainable Energy, 2(2):148 –158, Aprl. 2011.

[32] P. F. Ribeiro, B. K. Johnson, M. L. Crow, A. Arsoy, and Y. Liu. Energy storage

systems for advanced power applications. Proceedings of the IEEE, 89(12):1744

– 1756, Dec. 2001.

[33] M. Beaudin, H. Zareipour, A. Schellenberglabe, and W. Rosehart. Energy storage

for mitigating the variability of renewable electricity sources: An updated review.

Energy for Sustainable Development, 14(4):302 – 314, 2010.

[34] Quantum Wired. http://www/quantumwired.com/products/.

[35] T. Patrick and Moseley. Energy storage in remote area power supply (RAPS) systems. Journal of Power Sources, 155(1):83 – 87, 2006.

[36] 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, 57(4):1137 –1145, Apr. 2010.

[37] P. Thounthong, S. Rael, and B. Davat. Control strategy of fuel cell and super-

capacitors association for a distributed generation system. IEEE Transactions

on Industrial Electronics, 54(6):3225 –3233, Dec. 2007.

[38] C. Abbey and G. Joos. Short-term energy storage for wind energy applications.

In 40th Industry Applications Conference (IAS2005),Hong Kong, volume 3, pages

2035 – 2042, Oct. 2005. 242

[39] D. Corbus, C. Newcomb, E. I. Baring-Gould, and S. Friedly. Battery voltage

stability eﬀects on small wind turbine energy capture. In AWEA WINDPOWER,

Portland, USA, Oct. 2002.

[40] L. M. Ramirez-Elizondo, G. C. Paap, and N. Woudstra. The application of a

fuel cell-electrolyzer arrangement as a power balancing set-up in autonomous

renewable energy systems. In 40th North American Power Symposium, (NAPS

08), Calgary, Canada, pages 1 – 8, Sept. 2008.

[41] P. Mercier, R. Cherkaoui, , and A. Oudalov. Optimizing a battery energy storage

system for frequency control application in an isolated power system. IEEE

Transactions on Power Systems, 24(3):1469 – 1477, Aug. 2009.

[42] T. K. A Brekken, A. Yokochi, A. von Jouanne, Z. Z. Yen, H. M. Hapke, and D. A.

Alamay. Optimal energy storage sizing and control for wind power applications.

IEEE Transactions on Sustainable Energy, 2(1):69 – 77, Jan. 2011.

[43] L. Qu and W. Qiao. Constant power control of DFIG wind turbines with supercapacitor energy storage. IEEE Transactions on Industry Applications,

47(1):359 –367, Jan.- Feb. 2011.

[44] S. W. Mohod and M. V. Aware. Micro wind power generator with battery energy

storage for critical load. IEEE Systems Journal, 6(1):118.–125, Mar. 2012.

[45] Y. S. Shiao, C. E. Lin, M. T. Tsai, and C. L. Huang. Harmonic and reactive

current compensation using a voltage source inverter in a battery energy storage

system. Journal of Electric Power Systems Research, 25(1):25 – 33, 1992.

[46] S. Teleke, M. E. Baran, A. Q. Huang, S. Bhattacharya, and L. Anderson. Control

strategies for battery energy storage for wind farm dispatching. IEEE Transac-

tions on Energy Conversion, 24(3):725 –732, Sept. 2009. 243

[47] A. Abedini and H. Nikkhajoei. Dynamic model and control of a wind-turbine

generator with energy storage. IET Renewable Power Generation, 5(1):67 –78,

Jan. 2011.

[48] B. S. Borowy and Z. M. Salameh. Dynamic response of a stand-alone wind

energy conversion system with battery energy storage to a wind gust. IEEE

Transactions on Energy Conversion, 12(1):73 –78, Mar. 1997.

[49] M. Rashed, A. E. Ghanem, A. Ei-Sayes, and Mansy. Control stratgey for an

isolated DFIG based micro-grid with integrated super-capacitors. The online Journal on Electronics and Electrical Engineering (OJEEE), 1(2):81–88, Oct.

2009.

[50] A. M. O. Haruni, M. E. Haque, A. Gargoom, and M. Negnevitsky. Control

of a direct drive ipm synchronous generator based variable speed wind turbine

with energy storage. In 36th Annual Conference on IEEE Industrial Electronics

Society (IECON2010), Glendale, AZ, USA, pages 457 –563, Nov. 2010.

[51] F. A. Bhuiyan and A. Yazdani. Multimode control of a DFIG-based wind-power unit for remote applications. IEEE Transactions on Power Delivery, 24(4):2079

–2089, Oct. 2009.

[52] W. Li and G. Joos. A power electronic interface for a battery supercapacitor

hybrid energy storage system for wind applications. In IEEE Power Electronics

Specialists Conference, 2008, (PESC 2008), Rhodes, Greece, pages 1762 –1768,

Jun. 2008.

[53] D. G. Inﬁeld, G. W. Slack, N. H. Lipman, and P. J. Musgrove. Review of

wind/diesel strategies. IEEE Proceedings of Physical Science Measurement and 244

Instrumentation, Management and Education - Reviews, 130(9):613 – 619, Dec.

1983.

[54] L. A. de Souza Ribeiro, O. R. Saavedra, S. L. de Lima, and J. Gomes de Matos.

Isolated micro-grids with renewable hybrid generation: The case of lenis island.

IEEE Transactions on Sustainable Energy, 2(1):1 – 11, Jan. 2011.

[55] K. A. Nigim and W. J. Lee. Micro grid integration opportunities and chal-

lenges. In IEEE Power Engineering Society General Meeting, (PES2007),

Tampa, Florida, USA, pages 1 – 6, Jun. 2007.

[56] S. Roy. Reduction of voltage dynamics in isolated wind-diesel units susceptible

to gusting. IEEE Transactions on Sustainable Energy, 1(2):84 – 91, Jul. 2010.

[57] C. V. Nayar, S. J. Phillips, W. L. James, T. L. Pryor, and D. Remmer. Novel

wind/diesel/battery hybrid energy system. Journal of Solar Energy, 51(1):65 –

78, 1993.

[58] R. J. Best, D. J. Morrow, D. J. McGowan, and P. A. Crossley. Synchronous

islanded operation of a diesel generator. IEEE Transactions on Power Systems,

22(4):2170 –2176, Nov. 2007.

[59] M. A. Kashem and G. Ledwich. Distributed generation as voltage support for sin-

gle wire earth return systems. IEEE Transactions on Power Delivery, 19(3):1002

– 1011, Jul. 2004.

[60] M. Ross, R. Hidalgo, C. Abbey, and G. Joos. Energy storage system scheduling

for an isolated microgrid. IET Renewable Power Generation, 5(2):117 –123, Mar.

2011. 245

[61] R. Sebastian. Modelling and simulation of a high penetration wind diesel sys-

tem with battery energy storage. International Journal of Electrical Power and

Energy Systems, 33(3):767 – 774, Mar. 2011.

[62] R. Pena, R. Cardenas, J. Proboste, J. Clare, and G. Asher. Wind−diesel gen-

eration using doubly fed induction machines. IEEE Transactions on Energy

Conversion, 23(1):202 – 214, Mar. 2008.

[63] J. Lee, S. Lee, and S. Sul. Variable-speed engine generator with supercapacitor:

Isolated power generation system and fuel eﬃciency. IEEE Transactions on

Industry Applications, 45(6):2130 – 2135, Nov. - Dec. 2009.

[64] R. Sebastin and R. Pea Alzola. Eﬀective active power control of a high penetra-

tion wind diesel system with a ni-cd battery energy storage. Renewable Energy,

35(5):952 – 965, May 2010.

[65] J. M. Guerrero. Guest editorial editorial special issue on power electronics for

microgrids - part i. IEEE Transactions on Power Electronics, 25(12):2885 –2

888, Dec. 2010.

[66] E. I. Vrettos and S. A. Papathanassiou. Operating policy and optimal sizing of

a high penetration res-bess system for small isolated grids. IEEE Transactions

on Energy Conversion, 26(3):744 –756, Sept. 2011.

[67] L. A. de Souza Ribeiro, O. R. Saavedra, S. L. de Lima, and J. Gomes de Matos.

Isolated micro-grids with renewable hybrid generation: The case of Lenis island. IEEE Transactions on Sustainable Energy, 2(1):1 – 11, Jan. 2011.

[68] A. M. O. Haruni, A. Gargoom, M. E. Haque, and M. Negnevitsky. Voltage

and frequency stabilisation of wind-diesel hybrid remote area power systems. 246

Australian Journal of Electrical and Electronics Engineering, 8(1):75 – 84, Jan.

2011.

[69] C. Abbey, W. Li, and G. Joos. An online control algorithm for application of a

hybrid ESS to a winddiesel system. IEEE Transactions on Industrial Electronics, 57(12):3896 – 3904, Dec. 2010.

[70] D. Lee and L. Wang. Small-signal stability analysis of an autonomous hybrid

renewable energy power generation/energy storage system part I: Time-domain simulations. IEEE Transactions on Energy Conversion, 23(1):311 – 320, Mar.

2008.

[71] S. Bogosyan. Recent advances in renewable energy employment. IEEE Industrial

Electronics Magazine, 3(3):54 –55, Sept. 2009.

[72] E. A. Gray, C. J. Webb, J. Andrews, B. Shabani, P. J. Tsai, and S. L. I. Chan.

Hydrogen storage for oﬀ-grid power supply. International Journal of Hydrogen

Energy, 36(1):654 – 663, Jan. 2011.

[73] K. Agbossou, M. Kolhe, J. Hamelin, and T. K. Bose. Performance of a stand-

alone renewable energy system based on energy storage as hydrogen. IEEE

Transactions on Energy Conversion, 19(3):633 – 640, Sept. 2004.

[74] M. J. Khan and M. T. Iqbal. Analysis of a small wind-hydrogen stand-alone

hybrid energy system. Applied Energy, 86(11):2429 – 2442, Nov. 2009.

[75] N. Lu, Q. Li, X. Sun, and M.A. Khaleel. The modeling of a standalone solid-

oxide fuel cell auxiliary power unit. Journal of Power Sources, 161(2):938 – 948,

Jun. 2006. 247

[76] J. Jia, Y. Wang, Q. Li, Y. T. Cham, and M. Han. Modeling and dynamic charac-

teristic simulation of a proton exchange membrane fuel cell. IEEE Transactions

on Energy Conversion, 24(1):283 – 291, Mar. 2009.

[77] S. Obara. Fuel Cell Micro-grids. Springer-Verlag London Limited, 2009.

[78] US Department of Energy. http:www.eere.energy.gov/hydrogenandfuelcells/.

[79] M. L. Doumbia, K. Agbossou, and E. Granger. Simulink modelling and simu-

lation of a hydrogen based photovoltaic/wind energy system. In International

Conference on Comupter as Tool, Warsaw , Poland, pages 2067 –2072, Sept.

2007.

[80] F. Marangio, M. Santarelli, and M. Cali. Theoretical model and experimen-

tal analysis of a high pressure pem water electrolyser for hydrogen production.

International Journal of Hydrogen Energy, 34(3):1143 – 1158, Feb. 2009.

[81] O. Atlam and M. Kolhe. Equivalent electrical model for a proton exchange

membrane (pem) electrolyser. Journal of Energy Conversion and Management,

52(8 - 9):2952 – 2957, 2011.

[82] N. Gyawali, Y. Ohsawa, and O. Yamamoto. Dispatchable power from DFIG based wind-power system with integrated energy storage. In IEEE Power and

Energy Society General Meeting, (PES2010), Minneapolis, USA, pages 1 –7, Jul.

2010.

[83] O. C. Onar, M. Uzunoglu, and M. S. Alam. Dynamic modeling, design and simu-

lation of a wind/fuel cell/ultra-capacitor-based hybrid power generation system.

Journal of Power Sources, 161(1):707 – 722, Oct. 2006. 248

[84] T. Senjyu, T. Nakaji, K. Uezato, and T. Funabashi. A hybrid power system using

alternative energy facilities in isolated island. IEEE Transactions on Energy

Conversion, 20(2):406 – 414, Jun. 2005.

[85] C. N. Bhende, S. Mishra, and S. G. Malla. Permanent magnet synchronous

generator-based standalone wind energy supply system. IEEE Transactions on

Sustainable Energy, 2(4):361 –373, Oct. 2011.

[86] N. Gyawali and Y. Ohsawa. Integrating fuel cell/electrolyzer/ultracapacitor sys-

tem into a stand-alone microhydro plant. IEEE Transactions on Energy Con-

version, 25(4):1092 –1101, Dec. 2010.

[87] Y. Xue, L. Chang, and Z. Guo. A load controller for wind/hydrogen/diesel weak

grid. In Canadian Conference on Electrical and Computer Engineering, (CCECE

’09), Newfoundland, Canada, pages 37 –40, May 2009.

[88] Krisztina Leban. Doubly Fed Induction Generator Fault Simulations. PhD thesis,

Institute of Energy Technology, Aalborg University, Alborg, Denmark, 2009.

[89] R. Pena, J. C. Clare, and G. M. Asher. Doubly fed induction generator using

back-to-back pwm converters and its application to variable-speed wind-energy

generation. IEE Proceedings - Electric Power Applications, 143(3):231 – 241,

May 1996.

[90] D. Santos-Martin, S. Arnaltes, and J. L. R. Amenedo. Reactive power capabil-

ity of doubly fed asynchronous generators. Journal of Electric Power Systems

Research, 78(11):1837 – 1840, Nov. 2008.

[91] L. Harnefors and H. P. Nee. Model-based current control of ac machines using

the internal model control method. IEEE Transactions on Industry Applications,

34(1):133 – 141, Jan./Feb. 1998. 249

[92] J. C. Basilio and S. R. Matos. Design of PI and PID controllers with transient performance speciﬁcation. IEEE Transactions on Education, 45(4):364 – 370,

Nov. 2002.

[93] Maxwell Technologies. Ultaracapacitors. http://www.maxwell.com/products//.

[94] University of Texsas. The Austin-Centre for Electromechanics.

http://www.utexas.edu/research/cem/projects/.

[95] X. C. Zhang, G. H. Cheng, and Z. Xu. User deﬁned excitation system models

for power system stability analysis in PSASP. In IEEE Asia and Paciﬁc Trans- mission and Distribution Conference and Exhibition, (IEEE/PES 2005), China,

pages 1 –5, 2005.

[96] M. Kayikci and J. V. Milanovic. Reactive power control strategies for DFIG- based plants. IEEE Transactions on Energy Conversion, 22(2):389 –396, Jun.

2007. Appendix A

A.1 Co-ordinate Transformation

(a) Clark Transform

      1 −1 −1 a  2 2    α 2       =  1 −1 −1   b    3  2 2    β √ √  3 − 3    0 2 2 c

    a 1 0     2   α   =  −1 3     b   2 2      3   β    −1 3  c 2 − 2

      d cos(θ) cos(θ − 2π ) cos(θ + 2π ) a   r  3 3      2      q  =  sin(θ) sin(θ − 2π ) sin(θ + 2π )   b    3  3 3    √ √ √    2 2 2    0 2 2 2 c

250 251

(d) Inverse of Park Transformation

   √    a cos(θ) sin(θ) 2 d   r  2      2  √     b  =  cos(θ − 2π ) sin(θ − 2π ) 2   q    3  3 3 2    √    2π 2π 2    c cos(θ + 3 ) sin(θ + 3 ) 2 0

A.2 DFIG based RAPS System

(a) Mathematical Model of DFIG

The mathematical relationship which describes the stator and rotor voltages of the DFIG given in Section 3.2.2 of Chapter 3 are reinstated as follows:

dφ v = R i + ds − ωφ (A.1) ds s ds dt qs dφ v = R i + qs + ωφ (A.2) qs s qs dt ds φ v = R i + dr − (ω − ω )φ (A.3) dr r dr dt r qr φ v = R i + qr − (ω − ω )φ (A.4) qr r qr dt r dr

φds = Lsids + Lmidr (A.5)

φqs = Lsiqs + Lmiqr (A.6)

φdr = Lridr + Lmids (A.7)

φqr = Lriqr + Lmiqs (A.8)

With ISFO the q-axis component of the stator ﬂux given by (A.6) is equal to zero and the following set of equations can be obtained: 252

φqs = Lsiqs + Lmiqr = 0 (A.9)

−Lm iqs = iqr (A.10) Ls q 2 2 φqs + φds = φds = Lmsims , ims - stator magnetising current.(A.11)

φds = Lmsims = Lsids + Lmidr (A.12)

(ims − idr) ids = Lm (A.13) Ls

The q and d axes components of the rotor ﬂux given by (A.14) and (A.15) can be obtained by substituting (A.10) in (A.8) and (A.13) in (A.7) respectively.

φqr = σLriqr (A.14) 2 Lm φdr = σLridr + ims (A.15) Ls

2 where, σ = (1 − Lm ) is the leakage factor. The stator and rotor voltages can be LsLr further deduced to (A.16)-(A.17) and (A.18)-(A.19) respectively.

dφ v = R i + ds (A.16) ds s ds dt

vqs = Rsiqs + ωφds (A.17) i v = R i + σL dr − σL i (ω − ω ) (A.18) dr r dr r dt r qr r 2 diqr Lmims vqr = Rriqr + σLr + (ω − ωr)[iqrσLr + ] (A.19) dt Ls

dφs Assuming that dt is zero at steady state conditions and the stator resistance Rs is negligible, the d and q axes components of the stator voltages can be given by 253

(A.20) and (A.21) respectively.

vds ' 0 (A.20)

vqs = ωφsd (A.21)

The control algorithm that used to implement the inner current control associated

∗ ∗ with RSC is given by (A.22) and (A.23). The terms vrd and vrq can be deﬁned as (A.24) and (A.25) respectively

∗ vdr = vdr − σLriqr(ω − ωr) (A.22) 2 ∗ Lmims vqr = vqr + (ω − ωr)[iqrσLr + ] (A.23) Ls

where,

i v∗ = R i + σL dr (A.24) dr r dr r dt di v∗ = R i + σL qr (A.25) qr r qr r dt

(b) Reactive Power Control Approach of RSC

As stated in Section 3.2.4 of Chapter 3, the reactive power control approach is used to regulate the magnitude of the voltage. Following explain the detailed derivation of the control algorithm which is implemented for regulating the magnitude of the load voltage.

The total reactive power output of the stator of the machine Qs, can be given by 254

(A.26).

3 Q = (−v i + v i ) (A.26) s 2 qs ds ds qs

With the ISFO, the φqs = 0 and leads to vds ' 0. Therefore, the reactive power given by (A.26) can be further simpliﬁed into (A.27).

3 Q = (−v i ) (A.27) s 2 qs ds

From (A.5), the d-axis component of the current can be given by (A.28). (φds − Lmidr) ids = (A.28) Ls

Substituting (A.28) into (A.27) leads to (A.29). 3 (φds − Lmidr) Qs = − vqs (A.29) 2 Ls

Substituting (A.21) into (A.29) leads to (A.30). vs 3 ω − Lmidr) Qs = − vs( (A.30) 2 Ls

The total reactive power of the stator can be further simpliﬁed to (A.31). The rotor d-axis current consists of two components namely: magnetising component irdmag, and reactive component irdgen, given by (A.32). The ﬁrst current component is used for magnetisation purpose of the DFIG whereas irdgen is used to satisfy the reactive power demand of the loads. The corresponding reactive power components

Qmag and Qgen are given by (A.34) and (A.35) respectively. 255

2 3 vs Lm Qs = [− + vs idr] (A.31) 2 ωLs Ls

idr = irdgen + irdmag (A.32)

Qs = Qmag + Qgen (A.33) 2 3 vs Lm Qmag = [− + vs irdmag] (A.34) 2 ωLs Ls 3 Lm Qgen = vs irdgen (A.35) 2 Ls

A.2.1 Internal Model Control (IMC) Principle for Tuning the PI Con-

trollers Associated with the LSC

Internal model control principle was originally developed for chemical engineering applications and it is considered as one of the robust control method [91]. It is currently being applied to machine controller designs. The design of the controller is straight forward and easy to implement.

As stated in Section 3.2.5 of Chapter 3, the LSC consists of two PI controllers, one for current control and the other for DC voltage control. The following explains the procedure involved with PI controller tuning associated with the LSC of the DFIG.

The classical IMC control structure and classical PID control structures are shown in Fig. A.1.

The relationship that exists between the IMC controller C(s) and classical PID controller F(s) can be given by (A.36) and (A.37) respectively. The term α is a function of the rise time of the step response of the plant and can be given by (A.38). r(t) e(t) u(t) 256 + C(s) G(s) y(t) - Controller Plant

+ G(s)

Plant model

(a)

r(t) e(t) u(t) + F(s) G(s) y(t) - Controller Plant

(b)

Figure A.1: Control structure (a) IMC control structure and (b) Classical PID control structure.

α C(s) = Gˆ−1(s) (A.36) s + α α F (s) = Gˆ−1(s) (A.37) s ln(9) α = (A.38) tr

(a) PI control tuning of inner current loop

The current control for the LSC ﬁlter is found based on IMC principle. The active damping is used to remove the cross coupling of the disturbances (i.e. load voltage

(vg)). The voltage drop across the ﬁler associated with LSC in d-q domain can be

0 given by (A.39). The term indicated by Raf is included in the modiﬁed voltage vfdq given in (A.40) to represent the active active damping term which is used to damp 257

the eﬀects on ﬁlter current due to the variation of load voltage. The relationship that

0 exists between the filter voltage vfdq and modified filter voltage vfdq is given by (A.41).

The term jωLif is included in the inner feedback control loop to decouple the d and q axes current components. However the cross coupling that exists between d and

q variables may not be serious for the IMC due to slow ﬂux dynamics. As a result,

the transients in d-axis current may not be aﬀected by the q- axis current. The IMC

based control structure for the LSC ﬁlter is shown in Fig.A.2. The inner feedback loop

H(s) is given by (A.42). The modiﬁed transfer function of Gp(s) should be derived

0 using modiﬁed voltage (i.e. vfdq) and the ﬁlter current ifdq as given in (A.44). The

0 selection of vfdq is made in a manner that it should not consist of the cross coupling

0 terms. In addition, vfdq should consist of the damping term in order to minimise the eﬀect of disturbances.

di v = L fdq + Ri ± jωi + v (A.39) fdq dt fdq fqd gdq 0 di v = L fdq + Ri + R i (A.40) fdq dt fdq af fdq 0 vfdq = vfdq − (Raf ± jωLif ) (A.41)

H(s) = (Raf − jωL) (A.42)

0 di v − (R ∓ jωLi ) = L fdq + Ri ± jωi + v (A.43) fdq af f dt fdq fqd gdq if (s) 1 Gp(s) = 0 = (A.44) vfdq(s) (sL + R + Raf )

The transfer function between the disturbance (i.e. load voltage, vgdq) and the

output (i.e. ﬁlter current if ) can be determined using the (A.45)- (A.48). The estimation of term α can be given by (A.49) which ensures the damping eﬀects of disturbances (i.e. variations on load voltage). vg

258 (i f)ref v'(s) v(s) + if + GPID (s) + + Gp(s) - -

H(s)

Figure A.2: IMC based control structure for the LSC ﬁlter. Where (if )ref and if are reference and actual ﬁlter currents respectively.

s G (s) = G (s) (A.45) dy s + α p s 1 Gdy(s) = (A.46) (s + α) (sL + R + Raf ) 1 s 1 Gdy(s) = (A.47) L (s + α) R+Raf (s + L ) 1 s 1 G (s) = (A.48) dy L (s + α) (s + α) (r + R ) α = af (A.49) L 1 s G (s) = (A.50) dy L (s + α)2

The relationship that exists between the transfer functions associated with the PI controller1 and process plant can be described by (A.51)-(A.54). The corresponding parameters of PI controller gains can be estimated using (A.55) and (A.57).

1i.e. PI controller associated with the LSC ﬁlter. 259

α G (s) = G (s)−1 (A.51) PID (s) p α G (s) = (sL + R + R ) (A.52) pID s af R + R G (s) = αL + α af (A.53) PID s (αL) G (s) = αL + α (A.54) PID s K G (s) = K + i (A.55) PID p s

Kp = αL (A.56)

2 Ki = α L (A.57)

(a) PI control tuning of outer slower loop

As indicated in Fig. A.3, the power balance between the DC bus and the output of the LSC can be explained by (A.58) -(A.61). The DC bus dynamics can be explained using (A.63) where the current i2 is regarded as a disturbance. The Laplace transform of (A.64) is given by (A.65). The process transfer function , Gp, given by (A.66) can be obtained by substituting (A.65) into (A.62). 260

3 v i = v i (A.58) dc 1 2 ds ds √ vds = 2vs (vs = vLN ) (A.59)

mvdc vs = √ (A.60) 2 2 3 mv v i = ( ) dc i (A.61) dc 1 2 2 ds 3 i = mi (A.62) 1 4 ds dv i − i = C dc (A.63) 1 2 dt dv i = C dc (A.64) 1 dt

i1(s) = Csvdc(s) (A.65) v 3m dc = (A.66) ids 4Cs

where, i1 and i2 are DC currents on RSC and LSC, m is the modulation index and vLN is the line to neutral voltage.

Inverter Rf ia,ib,ic Lf i2 i1 v v a1 a Load side C vdc vb1 vb

vc1 vc

Figure A.3: LSC arrangement with DC bus.

The DC current on the RSC (i.e. i2) in Fig. A.3 is identiﬁed as a disturbance for 261 controlling the DC bus voltage. To minimise this disturbance i2 that can inﬂuence the DC bus voltage, an active damping term is introduced to the control structure.

The IMC based PI control structure is shown in Fig. A.4. The process transfer function with the active damping can be expressed as (A.67). The transfer function between the disturbance i2 and the output vdc is described by (A.68)- (A.71). The term ’k’ should be determined in such a manner that it will damp the disturbance with the same time constant of the control loop. Therefore the value of α and damping constant, k are selected as (A.72) and (A.73) respectively.

(vdc )ref v G (s)=1/Cs dc + Gc(s) p - + Gp(s) -

(a)

(vdc )ref vdc i' ds ids + + GPID (s) + + Gp(s)=3m/4Cs - -

(b) Figure A.4: IMC based control structures: (a) without active damping and (b) with active damping 262

1 Gp(s) = 4Cs (A.67) ( 3m + k) s G (s) = G (s) (A.68) dy (s + α) p s 1 G (s) = (A.69) dy (s + α) 4Cs ( 3m + k) 3m s 1 G (s) = (A.70) dy 4C (s + α) 3mk (s + 4c ) 3m s s G (s) = (A.71) dy 4C (s + α) (s + α) 3mk α = (A.72) 4c 4αC k = (A.73) 3m

The values for the IMC based PI controller gains have to be determined based on the approach given by (A.74)-(A.76). The corresponding values of the controller gains kp and kpi are given by (A.78) and (A.79) respectively.

α G (s) = G (s)−1 (A.74) PID s p α 4Cs G (s) = ( + k) (A.75) PID s 3m α4Cs αk G (s) = + (A.76) PID 3m s K G (s) = K + i (A.77) PID p s 4αC K = (A.78) p 3m 4α2C K = (A.79) i 3m 263 A.2.2 Parameters of DFIG based RAPS System

Table A.1: Parameters of DFIG

Rated power output (PDFIG) 750 kW

Stator resistance (Rs) 0.00706 pu

stator Leakage inductance (Lls) 0.171 pu

Rotor resistance (Rs) 0.005 pu

Rotor leakage inductance (Llr) 0.156 pu

Magnetising inductance (Lm) 2.2 Vs Inertia constant (H) 2.04 S Number of pole pairs (P ) 3

Filter inductance at LSC (Lf ) 0.3 pu

Filter resistance at LSC (Rf ) 0.3/100 pu

DC bus voltage (vdc) 750 V

Stator voltage (vs) 400 V

Operating frequency (fs) 50 Hz

A.3 PMSG Based RAPS System

(a) Equivalent circuit of PMSG

The equivalent circuits of the PMSG in d-q domain are shown in Fig. A.5.

Figure A.5: PMSG: (a) d-axis circuit (b) q-axis circuit. 264

(b) Parameters of PMSG

Trapezoidal model of the permanent magnet machine available in SimpowerSys- tems blockset in MATLAB is employed in this thesis. The model parameters of the

PMSG are as follows:

Table A.2: Parameters of PMSG

Rated power output (PPMSG) 100 kW

stator resistance (Rs) 0.0275 ohms

stator inductance (Ls) 4e-3 H

Flux linkage (Φs) 1.125 Vs

Torque constant (Tk) 9 Nm/Apk

voltage constant (Vk) 535.45 (vpk)LL/Krpm Inertia (J) 0.3 (Jkg2) Number of pole pairs (P ) 4

The boost converter is located between the uncontrolled rectiﬁer and DC bus of the PMSG wind energy conversion system as shown in Fig. A.6.

Switching signal from DC bus of boost controller converter in PMSG

full bridge rectifier

Figure A.6: Boost converter of the PMSG. 265 A.4 Power Quality Issues Of DFIG based RAPS-A Case

Study

The power quality behaviour of a DFIG based RAPS system shown in Fig. A.7 is investigated in relation to the harmonic content of the voltage at PCC. Load bus (PCC) DFIG bus bar DFIG

Distribution line

Dummy Load LSC RSC Load

Figure A.7: DFIG based remote area power supply system.

In super synchronous mode of operation RSC acts as a rectiﬁer and LSC acts as an inverter. At any slip ’s’ of the DFIG, the interaction between the rectiﬁer and inverter harmonics cause unwanted DC current pulsations at the DC link. The corresponding harmonic frequencies of the DC link current is given by (A.80) and (A.81).

R fdc,h = 6k|s|fs k = 1, 2, 3.. (A.80)

I fdc,h = 6fsm m = 1, 2, 3.. (A.81)

R I where, fdc,h- harmonics due to the rotor side controller, fdc,h- harmonics due to the line side controller, fs- stator frequency, |s|- slip of the DFIG

Furthermore, the stator current harmonics of the DFIG can be expressed using

(A.82) which excludes the high frequency switching harmonics of the inverter of the line side converter. 266

fs,h = | 1 ± 6ks ± 6m | fs k, m = 0, 1, 2, 3, ...... (A.82)

In the case, the harmonic content of the voltages at the PCC and the DFIG busbar have been examined for two diﬀerent cases:

• case I : Resistive load (500 kW) and

• case II : Resistive load (350 kW) and induction motor load (150 kW)

The wind speed is set nearly constant at 10.5 m/s.

The variation of the total harmonic distortion (THD) of voltage corresponding to above scenarios are illustrated in Figs. A.8 - A.9. It is seen that THD levels in both cases are quite similar in relation to the two locations of the network. Upon close examination it can be seen that distortion levels are slightly higher in case I compared to case II. The harmonic rich voltage at DFIG busbar together with IM armature reaction would contribute to such harmonic behaviour. 0.16

0.14

DFIG busbar

0.12

0.1

0.08 Total Harmonic Distortion (pu)

0.06

Point of common couplying

0.04 0.5 1 1.5 2 Time (s) Figure A.8: Case I - THD at PCC and DFIG busbar with resistive load only.

The frequency spectra of the PCC bus voltage for case I and case II are shown in Figs. A.10 - A.11 respectively. The interharmonics and non integer harmonics of 267 0.17

0.16

0.15

DFIG busbar 0.14

0.13

0.12

0.11 Point of common coupling TotalHarmonic Distortion (pu) 0.1

0.09

0.08 0.5 1 1.5 2 Time (s)

Figure A.9: Case II-THD at PCC and DFIG busbar with resistive load and induction motor load. small magnitudes are seen to occur around 50 Hz as evident from the above ﬁgures.

Also a relatively high harmonic voltage can be observed at 80 Hz in both frequency spectra. This harmonic frequency can be obtained by setting m=0, k=1 and s=0.1 in

(A.82). The switching harmonics content of the voltage at PCC in case II is shown in Fig. A.12. The switching frequency of the back to back PWM converters are used to operate at 2500 Hz. These harmonics represent the substantial magnitude of the frequency spectra at PCC compared to the harmonic at 80 Hz.

According to IEEE 519-1992, the maximum permissible voltage THD level at the

PCC should be 5% whereas the maximum individual voltage harmonics should be less than 3%In˙ relation to the present study, the THD values in both cases are higher than 5%. Also the individual harmonic limits, especially at high frequency range are not satisfying the above limit thus making harmonic ﬁltering a necessity. 268

0.9

0.8

0.7

0.6

0.5

Voltage (pu) 0.4

0.3

0.2

0.1

0 0 50 100 150 200 250 300 350 400 450 500 Frequency (Hz) Figure A.10: Frequency spectrum of voltage at PCC with resistive load only.

0.9

0.8

0.7

0.6

0.5

Voltage (pu) 0.4

0.3

0.2

0.1

0 0 50 100 150 200 250 300 350 400 450 500 Frequency (Hz)

Figure A.11: Frequency spectrum of voltage at PCC with resistive and induction motor load. 269

0.07

0.06

0.05

0.04

0.03 Voltage (pu)

0.02

0.01

0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 Frequency (Hz)

Figure A.12: Switching harmonics of voltage at PCC with resistive and induction motor load. Appendix B

B.0.1 Wind Turbine Power Characteristics - Linearised Model   0 0 ≤ v(t) < v  ci   2  (A + Bv(t) + Cv (t)) × Pr vci ≤ v(t) < vr PW(t) = (B.1)   P r vr ≤ v(t) < vco    0 vco ≤ v(t) where constants A, B, C can be expressed as:

vci vci + vr 3 A = 2 (vci + vr − 4vr( ) ) (B.2) (vci − vr) 2vr

vci + vr vci + vr 3 3vci − vr B = 2 (4( ) − ) (B.3) (vci − vr) 2vr vci + vr

1 vci + vr 3 C = 2 (2 − 4( ) ) (B.4) (vci − vr) 2vr

270 271 B.0.2 RAPS System Parameters

(a) DFIG-Battery-Dump Load RAPS System

Table B.1: Parameters of DFIG based RAPS System

Rated power output (PDFIG) 750 kW

Stator resistance (Rs) 0.00706 pu

stator Leakage inductance (Lls) 0.171 pu

Rotor resistance (Rs) 0.005 pu

Rotor leakage inductance (Llr) 0.156 pu

Magnetising inductance (Lm) 2.2 Vs Inertia constant (H) 2.04 S Number of pole pairs (P ) 3

Filter inductance at LSC (Lf ) 0.3 pu

Filter resistance at LSC (Rf ) 0.3/100 pu

DC bus voltage (vdc) 750 V

Stator voltage (vs) 400 V

Operating frequency (fs) 50 Hz

Rating of battery storage system (Pb) 420 kWh Allowable SOC of the battery system (SOC) 40%-80%

Rated dump load power ((Pd)max) 200 kW 272

(b) PMSG-Battery-Dump load RAPS System

Table B.2: Parameters of PMSG-Battery-Dump load RAPS System

Rated power output (PPMSG) 100 kW

stator resistance (Rs) 0.0275 ohms

stator inductance (Ls) 4e-3 H

Flux linkage established by the magnets (Φs) 1.125 Vs

Torque constant (Tk) 9 Nm/Apk

voltage constant (Vk) 535.45 (vpk)LL/Krpm Inertia (J) 0.3 (Jkg2) Number of pole pairs (P ) 4

Rating of battery storage system ((Pb)max) 60 kWh

Battery voltage (vb) 250 V average charge/discharge rate of the battery storage (k) 0.4 Allowable SOC of the battery system (SOC) 40%-80%

Resistance of the dump load (Rd) 14 ohms

DC bus voltage vdc 750 V

Stator voltage (vs) 400 V

Operating frequency (fs) 50 Hz 273 B.0.3 Wind Turbine Power Characteristic Curves

(a) DFIG based wind turbine power characteristic curve

1.2 13 m/s

1 12 m/s

0.8 Max. power at base wind speed (11 m/s) and beta = 0 deg 11 m/s

0.6 10 m/s

0.4 9 m/s 8 m/s

Turbine output power (pu ) 0.2 7 m/s 6 m/s 1.2 pu 0

-0.2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 Turbine speed (pu )

Figure B.1: Power characteristics of DFIG based wind turbine. 274

(b) PMSG based wind turbine power characteristic curve 1.2 13.2 m/s 1 Max. power at base wind speed (12 m/s) and beta = 0 deg 0.8 12 m/s

0.6 10.8 m/s

0.4 9.6 m/s 8.4 m/s 0.2 7.2 m/s 6 m/s

Turbine output power (pu) 1 pu 0

-0.2

0 0.2 0.4 0.6 0.8 1 1.2 1.4 Turbine speed (pu)

Figure B.2: Power characteristics of PMSG based wind turbine.

B.0.4 Torque Constant Estimation for the Induction Motor Driven

Pump Load

The induction motor driven pump load is rated at 75 kW which represents one tenth of the rated load capacity of the DFIG based wind-battery RAPS system. The torque which is applied for the induction motor can be estimated using (B.5)-(B.7). The corresponding torque constant k associated with the pump load can be estimated using (B.8)- (B.10). The operating characteristic of an induction motor driven pump load is depicted in Fig. B.3. 275

Pmotor = Tmotorωmotor (B.5) 75 × 103 × 60 T = (B.6) motor 2π × 1500

Tmotor = 477.47Nm (B.7)

2 Tpump = k(ωpump) (B.8) 477.47 k = (B.9) (1500 × 2π)2 k = 0.0194 (B.10)

Where ωmotor, ωpump are speeds of induction motor and pump load respectively,

Tmotor, Tpump are torque applied to motor and pump load respectively and k is torque characteristic constant.

Torque Pump load

Induction motor operating point

ω ω motor = pump Speed

Figure B.3: Torque speed characteristic of an induction motor driven pump load.

B.0.5 Battery Inverter Control for DFIG based Wind-Battery-Supercapacitor

RAPS system

With the adopted voltage orientation scheme (i.e. vqs=0) for the battery inverter, the corresponding active and reactive power output can be given by (B.11) and (B.12) 276

respectively.

Pb = vbids (B.11)

Qb = vbiqs (B.12)

where, vb is battery voltage, idb, iqb are d-q axes components of inverter currents. It can be noted that the active power output of the inverter can be controlled using

the d-axis current of the inverter. In contrast, the reactive power can be controlled

by the q-axis inverter current. A control algorithm similar to LSC converter which

is explained in Section 3.2.3 of Chapter 3 is employed for the inverter of the battery

storage system. Therefore, the control algorithm associated with the inner current

control loop of the battery inverter system1 can be described as follows:

∗ vds1 = vds − vds + Lf ωiqs (B.13)

∗ vqs1 = vqs − vqs − Lf ωids (B.14) di v∗ = R i + L ds (B.15) ds f ds f dt di v∗ = R i + L qs (B.16) qs f qs f dt

where, va, vb, vc are voltages on load side, ia, ib, ic are currents through the ﬁlter cir-

cuit, Lf , Rf are ﬁlter inductance and resistance respectively, va1, vb1, vc1 are voltages

at the inverter output, vds, vqs are d and q axes components of the load side AC volt-

age respectively, ids, iqs are d and q axes components of inverter current respectively and vds1, vqs1 are d and q axes components of the inverter output voltage respectively.

1This is the reinstatement of the control algorithm presented for the LSC of the DFIG in Chapter 3. 277

The d and q axes reference currents of the battery inverter can be generated by deﬁning the reference active and reactive power components as given in (B.17) and

(B.18) respectively.

(ids)ref = ((Pb)ref − Pb)(Kpdb + Kidb/s) (B.17)

(iqs)ref = ((Qb)ref − Qb)(Kpqb + Kiqb/s) (B.18)

where, (Pb)ref and (Qb)ref are reference active and reactive power output of the battery inverter, Kpdb, Kpqb and Kidb, Kiqb are proportional and integral gains of the battery inverter controller.

B.0.6 Estimation of Supercapacitor Current for DFIG based RAPS

System

As stated in Chapter 5, the supercapacitor is used to supply smax times the rated capacity of the DFIG power, PDFIG. The safe operating voltage limits associated with the supercapacitor is selected to be as follows:

250V < vsc < 500V (B.19)

Assuming in the worst case scenario, the supercapacitor is able to provide the maximum slip power of DFIG smaxPDFIG for time t = 10s, the capacitance value of the supercapacitor can be calculated as follows: 278

C = smax(PDFIG)ratedt (B.20) 0.3 × 750 × 1000 × 10 × 2 C = ≈ 20F (B.21) 5002 − 2002

Therefore peak supercpacitor current (Ic)pk for time t = 1s can be given as follows:

0.5CV (Ic)pk = (B.22) C(ESRDC ) + 1 0.5 × 20 × 500 (I ) = ' 370A (B.23) c pk (500 × 25 × 10−3) + 1 Appendix C

C.0.7 Wind-Diesel RAPS System Performance

The operational behaviour of a RAPS system consisting of a Doubly Fed Induction

Generator (DFIG) as the wind turbine generator, diesel generator system, battery storage, dump load and main loads is investigated. The diesel generator is connected to the system via a circuit breaker. It is connected to the main system only when the generation-demand mismatch satisﬁes the minimum loading condition of the diesel generator (i.e. 20% of diesel generator rating). The size of the diesel generator has been selected such that it will meet the minimum loading condition1 by supplying power to the local loads to avoid low load factor operation. The battery storage can be used as a source or load depending on the wind and loading conditions of the system. A control coordination among the system components is established with a view to regulate the system voltage and frequency while extracting maximum power from wind.

The proposed hybrid RAPS system was simulated under changing wind and variable load conditions. Fig. C.2 shows the system response whereas Fig. C.3 shows the power sharing among diﬀerent system components. The maximum power point tracking characteristics of the wind turbine generator is shown in Fig. C.4.

1In this case it is assumed 20% of rated capacity of the diesel generator.

279 280

DFIG

~ ~

RSC LSC D

G Main loads Dump load

+ Circuit

_ Engine Generator breaker Battery storage DC-DC converter

Local load

Figure C.1: Wind-Diesel-Battery hybrid remote area power supply system with the circuit breaker arrangement. 281

14 12

10 V_w (m/s) 8 1 2 3 4 5 6 7 (a)

1.1 1.05 1 V_L (pu) 0.95 1 2 3 (b) 4 5 6 7 1.005

1 f_L (pu) f_L 0.995 1 2 3 4 5 6 7 (c)

1.1 1 0.9 V_DC (pu) 0.8 1 2 3 4 5 6 7 (d) Ti me (s)

Figure C.2: Response of the DFIG based wind-diesel RAPS system under variable wind and load conditions: (a) wind speed, (b) voltage on load side, (c) frequency on load side, and (d) DC link voltage. 282

0.5 P_w (pu) 0 1 2 3 (a) 4 5 6 7 0.5

P_DE (pu) 0 1 2 3 4 5 6 7 (b) 0.5

0 P_b (pu) -0.5 1 2 3 (c) 4 5 6 7 1

0.8

P_L (pu) 0.6 1 2 3 4 5 6 7 (d) Ti me (s)

Figure C.3: Power sharing of the DFIG based wind-diesel RAPS system under variable wind and load conditions: (a) wind power, (b) diesel power, (c) battery power and (d) load demand. 283

1 Pw actual MPPT 0.9

0.8

0.7

0.6

0.5 Power (pu) Power 0.4

0.3

0.2

0.1

0 1 2 3 4 5 6 7 Ti me (s)

Figure C.4: Maximum power point tracking characteristic from wind in the DIF based wind-diesel RAPS system. 284 C.0.8 Modiﬁed RSC Arrangement

The modiﬁed RSC2 controller which only compensates the no-load reactive power is shown in Fig. C.5. The modiﬁed section of the RSC is shown in the circled area.

Lrω slip irq σ

vref + dq 1/Lω + PI - (idr )ref - abc

ϑ P To RSC idr PLL + W - M

ϑr iqs_actual (iqr )ref dq L /L + PI - s m - + abc

2 i ω (L σ ird L i ÷L ) qr slip r + m ms s

Figure C.5: Modiﬁed RSC control arrangement for unity power factor operation of the DFIG.

2This is explained in Section 6.4.2 of Chapter 6 285 C.0.9 Diesel Engine Model and Exciter System

The MATLAB models which represent the diesel engine and exciter are shown in Fig.

C.6 and Fig. C.7 respectively.

Figure C.6: Model of the diesel engine with its associated control.

vq v vstab f vref / v f 0 ka

Figure C.7: IEEE type I exciter system. 286 C.0.10 Parameters of the Wind-Diesel-Battery RAPS System

(a) DFIG Based Wind-Diesel RAPS System

Table C.1: Parameters of DFIG-Diesel-Battery-Dump load RAPS System

Rating of wind turbine generator (PDFIG) 750 kW

Rating of diesel generator (Pde) 350 kW

Rating of battery storage system (Pb) 150 kWh Allowable SOC of the battery system (SOC) 40%-80%

DFIG speed range of operation (ωr) 0.7 pu-1.3 pu

Rated DC link voltage of back-to-back converter (vdc) 750 V

Rated load side voltage (vs) 400 V

Operating frequency (fs) 50 Hz

(b) PMSG Based Wind-Diesel RAPS System

Table C.2: Parameters of PMSG-Diesel-Battery-Dump load RAPS System

Rating of wind turbine generator (PPMSG) 100 kW

Rating of diesel generator (Pde) 85 kW

Rating of battery storage system (Pb) 120 kWh Allowable SOC of the battery system (SOC) 40%-80%

Rated DC link voltage of back-to-back converter (vdc) 750 V

Rated load side voltage (vs) 400 V

Operating frequency (fs) 50 Hz 287

Table C.3: Parameters of 350 kW diesel generator used for DFIG based RAPS system

Direct axis synchronous reactance unsaturated (Xd) 349%

Quadrature axis synchronous reactance unsaturated (Xq) 209% 0 Open circuit time constant (Td0) 1738 ms 0 Direct axis transient reactance saturated (Xd) 20.1% 0 Short circuit transient time constant(Td 100 ms 0 Direct axis subtransient reactance saturated (Xd) 16.1% 00 Subtransient time constant (Td ) 10ms 00 Quadrature axis subtransient reactance saturated (Xd ) 21.8%

(d) 85 kW Diesel Generator Parameters

Table C.4: Parameters of 85 kW diesel generator used for DFIG based RAPS system

Direct axis synchronous reactance unsaturated (Xd) 2.2 pu

Quadrature axis synchronous reactance unsaturated (Xq) 1.1 pu 0 Open circuit time constant (Td0) 2555 ms 0 Direct axis transient reactance saturated (Xd) 20.1% 0 Short circuit transient time constant(Td 30 ms 0 Direct axis subtransient reactance saturated (Xd) 0.17 pu 00 Subtransient time constant (Td ) 8 ms 00 Quadrature axis subtransient reactance saturated (Xd ) 0.12 pu Appendix D

D.0.11 Parameters Associated with Hydrogen Based RAPS Systems

(a) DFIG-Diesel-Hydrogen based RAPS System

Table D.1: Parameters of DFIG-Diesel-Hydrogen RAPS System

Rating of wind turbine generator (PDFIG) 750 kW

Rating of diesel generator (Pde) 350 kW

Rating of fuel cell system (Pfc) 225 kW

Rating of electrolyser (Pelz) 225 kW 3 Volume of hydrogen tank (Vt) 2.5 m

DFIG speed range of operation (ωr) 0.7 pu-1.3 pu

Rated DC link voltage of back-to-back converter (vdc) 750 V

Rated load side voltage (vs) 400 V

Operating frequency (fs) 50 Hz

288 289

(b) PMSG-Hydrogen based RAPS System

Table D.2: Parameters of PMSG-Hydrogen RAPS System

Rating of wind turbine generator (PPMSG) 100 kW

Rating of fuel cell system (Pfc) 50 kW

Rating of electrolyser (Pelz) 50 kW 3 Volume of hydrogen tank (Vt) 2.5 m

Rated DC link voltage (vdc) 750 V

Rated load side voltage (vs) 400 V

Operating frequency (fs) 50 Hz 290 D.0.12 Parameters Associated with Components of the Hydrogen Stor-

age System

(a) Fuel Cell System

Table D.3: Parameters of the fuel cell system Faradays constant (F ) 96484600 [Ckmol−1]

Hydrogen time constant (τH2 ) 3.37 s −5 −1 Hydrogen valve constant (KH2 ) 4.2210 [kmol(satm) ]

Hydrogenoxygen ﬂow ratio (rHO) 1.168

N0 −6 −1 Kr constant (= 4F ) 1.844910 [kmol(sA) ]

No load voltage (E0) 0.8 [V]

Oxygen time constant (τO2 ) 6.74 [s] −1 Oxygen valve constant, (KO2 ) [kmol(satm) ] Absolute temperature (T ) 343 [K] Universal gas constant (R) 8314.47 [J(kmolK)−1]

Water time constant (τH2O) 18.418 [s] −1 Water valve constant (KH2O) 7.71610-6 [kmol(satm) ] Utilisation factor (U) 0.8

Fuel processor time constant (Tf ) 5 s

Electrical processor time constant (Ts) 0.8 s 291

(b) Electrolyser System

Table D.4: Parameters of the electrolyser system Area of Electrode (A) 0.25 m2 reversal voltage (E) 1.22 V −5 2 r1 8.05 × 10 Ωm −7 2 r2 −2.5 × 10 Ωm s 0.185 V −1 2 t1 −1.002A m C −1 2 t2 8.424A m C −1 2 2 t3 247.3A m C 0 Telz 40 C