Development of Control and Operational Strategies for

Wind Power Based Microgrid with Integrated Energy

Storage

Netra Prasad Gyawali

2011
Development of Control and Operational Strategies for Wind
Power Based Microgrid with Integrated Energy Storage

A Dissertation Submitted to the

Graduate School of Engineering, Kyoto University
in Candidacy for the Degree of Doctor of Philosophy

Netra Prasad Gyawali

2011
ii
Abstract
Wind power has emerged as the most dynamically growing energy source all over the world.
There is sig nificant adva ncement in i ts te chnologies with in crease in size and powe r capture
capacity, enabling it cost competitive to conventional power plants. With increased penetration of
wind power into the grid, the industries and the utilities are confronting with many challenges,
such a s frequency regulation, p ower fluctuation, voltage/reactive p ower control a nd s poradic
tripping of the upstream power network. Wind power is further insufficient in islanded operation,
where the frequency and voltage regulations are the main issues beside the load-following problem.
Scrutiny of these challenges reveals that their root cause is the intermittent power output, owing to
its de pendability on va rying wind spe ed. I n o rder to c ope with the u npredictability of the
generation, the best solution, despite its high cost, is to deploy electrical energy storage units. Such
storage units c an b e employed to ma ny applic ations, such as long-term st orage for energy
management, mid-term storage for frequency regulation and peak shaving, and short-term/buffer
applications for instantaneous power balance.
Choice of the particular storage type is guided by a number of factors and specific to the
application. In the c ase of power syste m whe re the penetration level of ren ewable sou rces is
considerable or when the renewable based power system needs to be operated in an islanded mode,
a combination of short and long-term storage is desired for reliable operation. These storage units
can be configured in diverse ways, often requiring power electronic converters to connect with the
regulated dc or ac bus. In t he case of v ariable s peed wind tur bine generator systems, there is
indispensable u se of ac-d c-ac conv erters, so the storag e d evices can be connected to th e
intermediate d c bu s v ia dc-dc converters. Th is dissertation proposes a d edicated sto rage mix
embedded at the int ermediate dc bus of a doubly-fed indu ction generator, whi ch provides the
functionalities of short-term power management and long-term energy management. The storage
mix consists of a fuel cell/electrolyzer (FC/ELZ) pair and an ultracapacitor (UC) as the primary
power so urce and th e secondary power sou rce, respectively; i n w hich the FC/ ELZ pair is
accommodated as a single u nit by a common dc-d c conv erter. To a ccomplish the o bjective o f
transient and long-term p ower m anagement, th e control schemes asso ciated with different
converters are proposed. The control scheme applied to the line-side converter enables the wind
power and storage s ystem (WPSS) to operate in the isl anded m ode or the dispatcha ble

iii
grid-connected mode, by drawing compensating quota of power from th e st orage system. The
controller associated with energy storage system employs the UC to balance the transient part by
consuming the transient fluctuation, while the steady-state imbalance is compensated by operating
in the FC or ELZ mode. The salient point of the scheme is that a unified control scheme is used
with each con troller to op erate in o ne’s re spective mo des in a smoo th ma nner. Further, an
improved power s haring s cheme is als o introduce d in the case w hen t here are m ore tha n one
controllable power sources.
In order to investigate th e performance, the sy stem co nfiguration is in vestigated and the
control scheme has been designed with linearized models of various subsystems. Based on the
dynamic component models, a simulation model for the proposed hybrid power system has been
developed in MATLAB/Simulink environment. The overall operation strategy for coordinating
the po wer flows among t he d ifferent en ergy sou rces has b een presented in d etail. Simulation
studies have been carried out to verify the system performance under different operating scenarios.
The r esults show th at the ov erall power management strategy is e ffective and the power flow
among the different energy sources and the load demand are balanced successfully.

iv
Contents
Abstract iii
List of figures ix
Acknowledgements x iii

Chapter 1 Introduction 1

1.1 Background . ...................................................................................................................1

1.2 Wind Power Issues and Energy Storage .........................................................................2
1.3 Microgrid ........................................................................................................................3
1.4 Literature Review ...........................................................................................................4
1.5 Proposed Study ...............................................................................................................6
1.6 List of Publications .........................................................................................................7
References .............................................................................................................................8

Chapter 2 System Configuration and Sizing 11

2.1 Introduction ....................................................................................................................11

2.2 Wind Turbine Generator System ....................................................................................11
2.3 Fuel Cells....................................................................................................................... 13
2.4 Ultracapacitor ................................................................................................................15
2.5 Electrolyzer . ...................................................................................................................17
2.6 Power Electronic Converters ..........................................................................................18
2.7 Layout of Proposed Scheme ...........................................................................................20
2.8 Unit Sizing ......................................................................................................................23
2.9 Summary .........................................................................................................................24
References ..............................................................................................................................24

Chapter 3 System Modeling 27

3.1 Wind Energy Estimation ................................................................................................28

3.2 Wind Turbine Generator System ...................................................................................30
3.2.1 Wind speed model .................................................................................................30
3.2.2 Wind turbine model ...............................................................................................31

v
 3.2.3 Doubly-fed induction generator model .................................................................36
a) Machine model.....................................................................................................37
b) Rotor-side converter model..................................................................................39
c) Line-side converter, filter and ac bus ..................................................................40
d) dc-link voltage .....................................................................................................41
e) Summary of doubly-fed induction generator model ............................................41
3.2.4 Model analysis .......................................................................................................42
a) Voltage source converter .....................................................................................42
b) Doubly-fed induction generator ...........................................................................44
3.3 Dedicated Energy Storage System..................................................................................44
3.3.1 Fuel cell ..................................................................................................................45
a) Partial pressure dynamics ....................................................................................46
b) Fuel cell output voltage ........................................................................................47
c) Fuel processing unit .............................................................................................52
3.3.2 Model analysis of SOFC .........................................................................................52
3.3.3 Electrolyzer ..............................................................................................................54
3.3.4 Compressor and tank model ....................................................................................56
3.3.5 dc-dc converters.......................................................................................................58
3.3.6 Ultracapacitor ..........................................................................................................60
3.3.7 Summary of dynamics in the DESS ........................................................................62
3.4 Remarks on System Modeling ........................................................................................63
References ..............................................................................................................................63

Chapter 4 Control and Operation Schemes 67

4.1 Internal Model Control Based Design of Controllers . ...................................................67

4.2 Wind Turbine Generator System Controller...................................................................72
4.2.1 Inner current control of rotor-side converter ...........................................................73
4.2.2 Speed control of wind turbine .................................................................................75
4.2.3 Maximum power point tracking scheme .................................................................78
4.2.4 Overall block diagram and test of controller ...........................................................81
4.3 Dedicated Energy Storage System Controller ................................................................82
4.3.1 Control approach for fuel cell side dc-dc converter ................................................83

vi
 4.3.2 Control approach for isolating dc-dc converter .......................................................88
4.4 Line-Side Converter Control ..........................................................................................89
4.4.1 Frequency and active power control........................................................................90
4.4.2 Voltage control ........................................................................................................91
4.4.3 Active power management ......................................................................................95
4.4.4 Reactive power management ...................................................................................97
4.5 Summary .........................................................................................................................99
References ..............................................................................................................................99

Chapter 5 Simulation Results and Discussion 101

5.1 System Startup and Operation in Islanded Mode ...........................................................102

5.2 Power Dispatching with Grid .........................................................................................105
5.3 Regulation of Key Variables ..........................................................................................106
References ..............................................................................................................................112

Chapter 6 Power Sharing Scheme with Multiple VSCs 113

6.1 Introduction ....................................................................................................................113

6.2 Applicability of Droop Regulation .................................................................................114
6.3 Introduction of Virtual Impedance .................................................................................119
6.4 Small Signal Modeling and Control Design Rules .........................................................120
6.5 Simulation Results ..........................................................................................................122
6.6 Summary .........................................................................................................................125
References ..............................................................................................................................125

Chapter 7 Conclusion and the Future Course 127

Appendix 131

List of Symbols 135

List of Acronyms 141

vii
viii
List of Figures
Fig. 2-1 Types of WTGS. ............................................................................................................12
Fig. 2-2 Schematic block diagram of FCs. ..................................................................................15
Fig. 2-3 Schematic diagram of a typical UC................................................................................16
Fig. 2-4 Schematic diagram of alkaline electrolyzer. ..................................................................17
Fig. 2-5 General block diagram of power electronic converters for DERs. ................................19
Fig. 2-6 Schematic diagram of proposed system. ........................................................................22
Fig. 3-1 Probability density of the Rayleigh distribution. ...........................................................28
Fig. 3-2 General structure of a wind turbine generator system....................................................30
Fig. 3-3 Wind turbine characteristic curves. ................................................................................33
Fig. 3-4 Typical variable speed pitch control strategy plotted on the vw-ωr-β space. ..................34
Fig. 3-5 Variable speed pitch controlled wind turbine operation regions....................................35
Fig. 3-6 Simulation model for wind turbine. ...............................................................................35
Fig. 3-7 Schematic diagram of a typical "loss-less" DFIG. .........................................................36
Fig. 3-8 Equivalent circuit for the DFIG in the synchronously rotating reference frame. ..........37
Fig. 3-9 Models for voltage source converter. .............................................................................39
Fig. 3-10 Voltage and current response curves of the VSC.........................................................43
Fig. 3-11 DFIG startup process at no-load...................................................................................44
Fig. 3-12 Schematic diagram of SOFC. .......................................................................................46
Fig. 3-13 Equivalent electrical model of SOFC...........................................................................51
Fig. 3-14 Static V-I and P-I response of the SOFC......................................................................51
Fig. 3-15 Model validation of the SOFC. ....................................................................................53
Fig. 3-16 Static V-I characteristics of an alkaline ELZ. ...............................................................55
Fig. 3-17 Block diagram of hydrogen production and storage model. ........................................57
Fig. 3-18 Electrical diagram of the bi-directional dc-dc converter. .............................................57
Fig. 3-19 Electrical diagram of the full bridge dc-dc converter.. ................................................59
Fig. 3-20 Electrical diagram of UC..............................................................................................60
Fig. 3-21 Overall block diagram of the dedicated energy storage system… ...............................61
Fig. 4-1 Plant control topologies..................................................................................................68
Fig. 4-2 Control block diagram of inner current control loops. ...................................................74

ix
Fig. 4-3 Block diagram of speed control loop. ............................................................................76
Fig. 4-4 Block diagram of rotor pitch angle control. ...................................................................77
Fig. 4-5 Block diagram of overall control scheme for the DFIG.................................................79
Fig. 4-6 Output responses of control block shown in Fig. 4-3. ...................................................81
Fig. 4-7 Bode plot of control block shown in Fig. 4-3. ...............................................................82
Fig. 4-8 Control block diagram of FC/ELZ/UC/DC1 system......................................................85
Fig. 4-9 Pole zero plot of FC/ELZ/DC1 system. .........................................................................86
Fig. 4-10 Step response of the inner current control loop............................................................86
Fig. 4-11 Responses of DC1 controller........................................................................................87
Fig. 4-12 Control scheme for the DC2/dc-link bus system. ........................................................89
Fig. 4-13 Block diagram of P/f control. .......................................................................................91
Fig. 4-14 Control block of the line-side voltage source converter. .............................................93
Fig. 4-15 Responses of the outer voltage control loop. ...............................................................94
Fig. 4-16 Active power management scheme for VDC2.............................................................95
Fig. 4-17 Reactive Power management scheme for VSC2. .........................................................97
Fig. 4-18 Overall control block diagram of line-side converter (VSC2) system. ........................98
Fig. 5-1 Wind speed profile. ........................................................................................................102
Fig. 5-2 Power flow responses of different sources.....................................................................103
Fig. 5-3 Responses of frequency and voltages.............................................................................104
Fig. 5-4 Rotor speed response......................................................................................................107
Fig. 5-5 Responses of pitch angle and power conversion coefficient..........................................107
Fig. 5-6 Response of output voltages...........................................................................................108
Fig. 5-7 Responses of H2 and O2 partial pressures in the SOFC. ................................................109
Fig. 5-8 Molar flow and H2 pressure responses in the storage tank. ..........................................109
Fig. 5-9 LVRT of the proposed system. ......................................................................................110
Fig. 6-1 Single line diagram of microgrid with multiple voltage source converters. ..................113
Fig. 6-2 Voltage source connected to the local bus. ....................................................................114
Fig. 6-3 Conventional droop schemes. ........................................................................................116
Fig. 6-4 Power flow dynamics of VSCs. .....................................................................................117
Fig. 6-5 Voltage response of the load bus....................................................................................118
Fig. 6-6 Block diagram of proposed voltage and reactive power regulation scheme. .................119

x
Fig. 6-7 Power flow dynamics of proposed scheme. ...................................................................123
Fig. 6-8 Voltage dynamics of proposed scheme. .........................................................................124
Fig. 6-9 Active and reactive power sharing. ................................................................................124

xi
xii
Acknowledgements

This thesis is dedicated to my parent whose constant support, encouragement, and love, are
always with me in whatever I pursue.
First and foremost, I would like t o expres s pro found gratitude to m y supervisor, prof.
Ohsawa, for his invaluable support, supervision and useful suggestions throughout this research
work. His en couragement and continuous gu idance, not limited on ly to t he research, bu t also
enabled me to complete my work successfully. As my teacher and mentor, he has taught me more
than I could e ver give h im credit for here . I am ver y t hankful wi th Dr. Yamamoto for h is
continuous support, comment and any kind of assistance in the time whenever I needed.
My si ncere g ratitude goes to prof. Hiki hara and prof. Hagiwara for t heir inval uable
comments and suggestions on my research work as the co-supervisors. Special thank goes to Dr.
Zhou for his supportive and helping hand in all areas since the day I came in Japan. He was always
ready to answer all kinds of questions, and provided meaningful advices in spite of his busy time
schedule.
I am thankful to all the members in the Ohsawa lab and my friends, who provided immense
help directly and indirectly all the time.
Last but not the least , I wish to thank m y loving a nd suppo rtive wife who provided t he
endless support and care which are too numerous to mention. The acknowledgement will, perhaps,
be incomplete without remembering my wonderful children, Sambodhi and Snidh, whose smiling
faces always inspire me to dedicate at my work in all situations.

xiii
xiv
Chapter 1 Introduction

1.1 Background
Rising environmental concerns, increasing fossil fuel depleting rates, and the recent advent
of po wer elect ronics techn ologies, are favo ring th e deployment of re newable ty pe m odular
generators, also called distributed generation (DG) units, around the world. In rural and remote
areas, the DG sources have drawn further attention for the stand-alone operation due to the high
cost a nd c omplexity of th e tr ansmission network [ 1]. Am ong th e renewable energy r esources,
wind p ower has emerged as t he most dyn amically g rowing e nergy source. There is significant
advancement i n i ts te chnologies with i ncrease i n s ize a nd power c apture c apacity, making the
wind po wer cost competitive to conv entional power p lants. Worldwid e n ameplate capacity o f
wind powered generators was 157.9 GW at the end of 2009, with 90% installation between 2000
and 2009 only [2]. Most of the countries have put great emphasis on wind power installation and
it is likely that this trend will continue in coming decades.
Generating electricity from wind takes place in two steps. Firstly, the aerodynamic energy
in the wind passes over the blades exerting a turning force. The rotating blades, then, turn a shaft
inside the nacelle, which goes into a gearbox. The gearbox increases the rotation speed for the
generator. Consequently, th e generator converts the rotational energy into the electrical energy
with the help of applied magnetic fields. The blades on the wind turbines may be rotated in two
axes – horizontal or v ertical axis. Horizontal axis wind turbines are the m ore familiar windmill
type where the blades (normally three) are bolted into a hub on its tower and rotated in a vertical
plane about a h orizontal axis facing to the wind. In contrast, vertical axis turbines rotate around
the r espective vertical axis and do not need a particul ar orient ation t oward the wind. An
advantage of v ertical axi s turbines is th at th e generator a nd th e gearbox ca n be placed on the
ground, so a high tower is not required to mount the machine. Further, a yaw mechanism is not
needed to turn the rotor against the wind. However, at th e ground level, the wind speed is low
and it c ontains h igh t urbulence, s o it is desired to ins tall w ind blades a t a hei ght to get better
wind ca pture. This is ac hieved by using the horizontal a xis turbi ne wi th it s more efficient
propellers. Therefore, most of the modern wind turbines are designed as the horizontal axis type.

1
 Wind power is seldo m op erated in an islanded or a stand -alone mo de, because the wind
power outpu t is un controlled, unlike th e conventional po wer g enerator sy stems. S o, it makes
impossible to s upply power t o the unpredictable load from th e uncontrolled source. The
conventional approach is, therefore, to connect wind power generators with the utility gr id and
supply power as a “negative load”. Direct connection of wind turbines (WTs) to the external ac
point of common c oupling, s o called f ixed speed wind t urbine, is simple. The sim plicity is,
however, in th e exp ense o f e fficient a nd flexible operation. Usually, there are mi smatches in
frequency a nd voltage between the wind tu rbine generator s ystem (WT GS) and t he c ommon
coupling poi nt, wh ich n ecessitates some sorts of po wer electronic conv erters f or flexible
functionalities. Consequently, the WTGS can be operated in a wide range of rotor speed as their
name, variable-speed wind turbines. Compared to the fixed-speed operation, the v ariable-speed
wind turbine provides 20–30% higher energy output, lower mechanical stress, and reduction in
acoustic noise. Further, it als o provides f lexibility to control ac tive a nd reactive power by
utilizing ful l or partial rat ed power electronic conv ersion t echnology [ 3, 4] . Most of th e m ajor
wind turbine manufactures are developing larger wind turbines in the range of 3 to 5 MW range,
based o n th e v ariable-speed operation and th e pitch con trol u sing a direct driven synchronous
generator (without gearbox) or a doubly-fed induction generator (DFIG) [5]. The use of power
electronic converters in the v ariable-speed w ind turbine no t only p rovides th e f lexible
functionality and efficient operation, bu t also avails the int ermediate dc bus f or em bedding
energy storage devices. This, in turn, helps improving the power quality, system stability, better
low voltage ride through capability, among the others [3, 6].

1.2 Wind Power Issues and Energy Storage

The t ransition from fossil fuel b ased po wer generation to power g eneration b ased on
renewable e nergy resources (s uch a s win d power) i ntroduces c hallenging demands on the
operation of electricity systems. At l ow penetration, wind turbines are not expected to take part
in voltage and frequency control. If a disturbance occurs, the wind turbines are disconnected, and
reconnected when normal operation has been resumed. With the high penetration into the power
system, the wind po wer is, no longer, assume d a s th e “n egative load”; me aning th at
uncontrollable power is injected to th e s ystem, un like the “load” that con sumes u ncontrollable
power fro m t he system [7 ]. Th e wi nd energy source must int eract with th e rest of t he power
generation units in a s ystem to m ake it possible f or the system to s ecure a balance between

2
supply and demand. With increased penetration of wind power into the grid, the industries and
the u tilities a re c onfronting wi th many c hallenges, s uch as f requency regulation, power
fluctuation, voltage/reactive power control and sporadic tripping of the upstream power network.
Wind power is f urther ins ufficient i n the i slanded operation, w here t he frequency a nd voltage
regulation are the main issues beside the load-following [8]. Scrutiny of these challenges reveals
that their root cause is the intermittent power output, owing to its dependability on varying wind
speed. In order to cope with the unpredictability of the ge neration, the best solution, despite its
high cost, is to d eploy ele ctrical energy sto rage un its. Such sto rage un its can b e employed to
many a pplications: l ong-term sto rage f or e nergy management, m id-term sto rage f or f requency
regulation and peak shaving, and short-term/buffer applications for instantaneous power balance
[6].
Table A.1 in the appendix illustrates a comparative analysis of the different energy storage
systems a pplicable to the w ind pow er system. For th e short-term energy storag e, vi able
technologies include the flywheel energy storage, battery energy storage systems, ultracapacitors
(UC), a nd the sup erconducting magn etic energy s torage (SMES). For lo ng-term en ergy
management, th e high cap acity techno logies in clude hydrogen sto rage [9 ], th e redox flow
batteries and the compressed air technology [10 ]. These sto rage units are e ither con figured a s
one aggregated unit that serves for the whole wind farm, or the distributed one connected to each
wind turbine generator system. Since no single energy storage is perfect, the choice among the
different types is an o pen iss ue g uided by the cost and technical tra de off am ong the others.
Irrespective of the type of storage systems, it is advisable to integrate local distributed generators,
load and sto rage devices, making a n auto nomous lo cal power network (LPN) to e xploit m ost
effective functionalities of hybrid power system.

1.3 Microgrid
One way o f the efficient control, operation and management of hybrid distributed energy
resources is to form an auto nomous p ower ne twork or a m icrogrid. The m icrogrid may b e
defined a s a cl uster of sm all generating s ources, en ergy storage units and lo ads (h eat and
electricity), which can be operated in parallel with the utility g rid o r i n an i ntentional island
mode to provide a customized level of high reliability and resilience to grid disturbances [1]. An
essential element of a microgrid is the capability to control the balance of the generating capacity
and demand within the c onfines of t he m icrogrid i tself. Regardless t he m icrogrid being

3
connected to the ut ility grid or operating in dependently in a is landed m ode, this c ontrol is
essential to ens ure the stable s upply of ene rgy to th e consumers. With proper ope ration and
management scheme, the microgrid appears as a dispatchable load or a generator from the utility
side, so it behaves as a “good citizen” [11]. From the consumer perspective, it not only provides
the improved reliability of electricity supply (as it can also supply electricity during the failure of
the upstream network), but also opens the opportunity to sell the energy back to the utility from
their g enerators in m ore flexible m anner. By combining the re newable po wer so urces w ith
storage d evices, consumers can respond to the p rice of ele ctricity b y participating in dem and
response. Further, th e consumers c an also integrate th e functionality o f h eat and electricity,
thereby enabling to use th e combined heat and power with the onsite generators, like fuel cells
and micro-turbines.
To achieve de sired o bjectives from a hybrid lo cal p ower network, the i ndividual
components should be c onfigured pr operly w ith c learly d efined control and o perational laws.
Unfortunately, there i s n o predefined t opology a nd control sc heme due t o diverse nature of
distributed energy resources (DERs). The renewable based local power network or microgrid has,
thus two open issues; how to design th e suitable topo logy fo r int erconnecting the individual
components? What sort of control strategies (e.g. hierarchical, coordinated or both) are necessary
for the system, so that the subsystems would fit to fulfill the overall operational objective? In the
view point of f uture “ smarter” p ower network, w here th e co ntribution of wind po wer will b e
significant, the energy storage system is emerging as an indispensable unit for a microgrid. Being
an important subset of the advanced future power network, research work on the microgrid has
great potential to investigate the components’ behaviors and fit them into a system [12].

1.4 Literature Review

Research on wind energy conversion is fairly mature and can be f ound in various studies.
Conceptual design of a wind en ergy c onversion system is pres ented in [ 3, 1 3]. M odeling,
control a nd simulation of the w ind tur bine generator system are illustrated in [ 14-16]. T hese
studies have detailed discussion on overall wind turbine generator system with their embedded
power electronic conversion configurations. Due to the varying nature of wind power input, a
backup st orage unit is needed to r educe t he wind power f luctuations a nd ensure the power
balance. The topo logy of su ch hybrid power sy stems, in which th e wi nd power is assi sted by
secondary backup systems, has been reported in many studies. Wind and diesel combination has

4
been used in stand-alone network since the long time [17]. However, every diesel generator has
several major d isadvantages, such as lo w flexibility to respond the d emand ch anges, h igh
pollution and noise, and low efficiency, thereby making them less attractive for integrating with
clean power generation systems. In the variable-speed wind turbine, there is indispensable use
of a c-dc-ac ty pe power e lectronic co nverters. The ac-dc-ac converters c an be used f or
multipurpose, in which the inte rmediate dc-link bus helps augmenting various types of storage
devices th at re quire dc interconnection. Owing to its mat ure and cost effective sol ution, the
battery system has been extensively used with wind power for various purposes, such as battery
charging [18], suppressing the power fluctuation [19], and power management [20]. However,
its size, limited charging/discharging cycles and the disposal issues are the constraints for wide
scale application [21]. Other advance battery technologies like NiMH, NiCd and Li battery are
superior in terms of efficiency, compactness and higher charging/discharging cycles, and mostly
recommended for low voltage portable and vehicular applications [22].
Introduction of a short-term energy storage device in the intermediate dc-link bus is found
in various studies. Embedding a flywheel energy system in [23], [24], SMES system in [25, 26]
and ultracapacitor in [27], [28] are s ome examples to smooth the wind power fluctuation. For
long-term energy balance, the pump hydro is a matured technology. However, its application is
restricted by the geog raphical constr aints and water a vailability. A fuel-cell (FC) and an
electrolyzer (ELZ ) p air is em erging a s a c andidate f or t he long-term p ower m anagement
utilizing pol lution-free H 2 production a nd consumption technologies [ 29]. T he multitude of
studies has ad opted the s torage a pplication e ither f or the transient power fluc tuation
management, or the l ong-term energy management. There is very limited research exposure in
the area of storage mix combining the short and long-term storage for the application of a wider
time frame. When re newable re sources bec ome dominant i n p ower ge neration, the short a nd
long-term s torage devices are c omplementary for a reliable operation i n the islanded or
dispatchable grid-connected mode. In tegration of s uch mix unit i nto wi nd turbine ge nerator
system is re latively a new topic and the database of t hese studies is limited. In the distributed
power system application, the ultracapacitor is a favorable candidate for the transient mitigation
in term s o f its spe ed, effic iency a nd cost [ 10], th us a co mbination of ult racapacitor a nd fu el
cell/electrolyzer offers a promising technical so lution fo r th e s hort to long-term power
management [30]. A f ew works are reported in this field [30, 31], however, they do not depict

5
the clear picture on the power and energy balance among the energy resources having different
time constants.

1.5 Proposed Study

The aim o f th is rese arch w ork is to a ddress sh ort an d lo ng-term powe r management
requirements of the wind power based microgrid by embedding an energy storage system. The
storage system, termed as t he dedicated energy storage system (DESS), consists of a fuel cell/
electrolyzer and an ultracapacitor as the primary and secondary power source, respectively. We
propose a new configuration of DESS, in which the fuel cell and electrolyzer are accommodated
as a single unit, by a c ommon conv erter and c ontrol action . To ac complish the objective of
transient and long-term p ower m anagement, the c ontrol schemes associated with the l ine-side
converter and DESS are de veloped. The control sc heme ap plied to the l ine-side co nverter
enables the wind power/storage system to operate in either an islanded or a dispatchable grid-
connected mode, by drawing the compensating power fro m the DESS. Th e DESS employs the
ultracapacitor t o sm ooth the transient flu ctuation, whil e th e stea dy-state imbalance is
compensated by operating in either the fuel cell or electrolyzer mode. The salient feature of the
controllers is th at a un ified cont rol sch eme is used by each controller to operate in one’s
respective modes in a seamless manner. Further, an improved power sharing scheme among the
multiple inverters is als o pr oposed for th e a utonomous local power network h aving a high
resistance-inductance ratio (R/X).
The rest o f the thesis proceeds as follows. Beginning with the synoptic v iew of the wind
power/storage and the rela ted p ower electronic c onversion system, chapter two pr esents the
layout and unit sizing of the proposed system. Chapter three discusses on the dynamic modeling
of system components. Chapter four presents the detailed design of the control and operational
schemes of th e o verall win d power a nd the storage system followed by the d iscussion of
simulation results in chapter five. In chapter six, the power sharing scheme among the inverters
in an au tonomous m icrogrid is investigated, and chap ter se ven gives t he con clusion and
prospective direction of the research work.

6
1.6 List of Publications
Some of the results presented in this thesis have been published in the following publications.

International Journal Papers

1. Netra Gyawali and Yasuharu Ohsawa, "Improved power sharing scheme for a microgrid
with multiple distributed generators," IEEJ Trans. EIS, vol. 128(11), pp.1635-1640, 2008.
2. Netra Gy awali a nd Yasuharu Oh sawa, "Integrating fue l cell /electrolyzer/ultracapacitor
system into a stand-alone micro-hydro plant," IEEE Trans. on Energy Conversion System,
vol. 25(4), pp. 1092-1101, Dec. 2010.
3. Netra Gyawali, Osamu Yamamoto and Yasuharu Ohsawa, "Power dispatching from cage
induction g enerator based wind po wer system with in tegrated smar t en ergy sto rage,"
IEEJ Tran s. o n Electrica l and Elec tronics Engineering, vol. 6(2 ), pp. 134–143, Mar.
2011.
4. Netra Gyawali, Osamu Yamamoto and Yasuharu Ohsawa, "Power management of DFIG
based w ind p ower s ystem wit h integrated s mart ene rgy storage having SMES /fuel-
cell/electrolyzer," IET Journal of Renewable Energy (under review).
International Conference Papers
1. Netra Gyawali and Yasuharu Ohsawa, "Power management schemes for a microgrid with
inverter i nterfaced distributed generators," I nternational c onference of powe r s ystem
CPRI-IEEE/PES Bangalore, Dec. 2007.
2. Netra G yawali and Y asuharu Ohsawa, "Effective voltage an d f requency regulation
strategy for a stand-alone sy stem wi th induction generator/fuel cell /ultracapacitor,"
International conference o n I ntegration of Wi de-Scale Renewable Resources in to the
Power Delivery System, Cigre/IEEE Power Society, Calgary, Jul. 2009.
3. Netra Gyawali, Os amu Yam amoto and Yasuharu O hsawa, "Dispatchable po wer f rom
DFIG based wind-power s ystem with integrated energy s torage," IEEE/PES G eneral
Meeting Minnesota, Jul. 2010.

7
References

[1] Abu-Sharkh, S. , R. Y ao, K. Ste emers, J. Koh ler, R. A rnold, A . Wi lson, B. Li, T.
Markvart, and N. Ross, "Micr ogrids: Di stributed on -site generation," Ty ndall research
centre technical report, University of Southampton, 2005.
[2] World win d energy association, A nnual r eport, March 2 009, ava ilable o nline:
http://www.wwindea.org/home/index.php.
[3] T . Ackermann, Wind power in power systems, John Wiley and Sons Ltd, 2006.
[4] T. Burton, D. Sharpe, N. Jenkins and E. Bossanyi, Wind Energy Handbook, John Wiley
and Sons Ltd, 2004.
[5] T. Ackermann and L. Söder, "An overview of wind energy-status 2002," Renewable and
Sustainable Energy Reviews, vol. 6(1-2), pp. 67-127, 2002.
[6] C. A bbey and G. J oos, "E nergy storage a nd management in wind turbine generator
systems," Procee ding of 12 th in ternational Power Electronics and Motion Control
Conference, pp. 2051-2056, 2006.
[7] E. DeMeo, W. Grant, M. Mill igan an d M. Schu erger, "W ind plant integr ation," Power
and Energy Magazine, IEEE, vol. 3, pp. 38-46, 2005.
[8] H. Holttinen, P. Meibom, A. Orths, F. Van Hulle, C. Ensslin, L. Hofmann, J. McCann, J.
Pierik, J. O. Tande, A. Estanqueiro, L. Sö der, G. Strbac, B. Parsons, J. C. Sm ith and B.
Lemström, "Design and operation of power systems with large amounts of wind Power,
first results of IEA collaboration," Global Wind Power Conference, Australia, Sep. 2006.
[9] J. Baker a nd A. Collinson, "E lectrical energy st orage at the turn of the Mil lennium,"
Power Engineering Journal, vol. 13(3), pp. 107-112, 2002.
[10] H. Ibrahim, A. Ili nca an d J. Perron, "E nergy sto rage systems-characteristics and
comparisons," Renewable and Su stainable En ergy Reviews, vol. 12 (5), pp. 1221 -1250,
2008.
[11] R. Lasseter and P. Paigi, "Microgrid: A c onceptual solution," Proceeding on IEEE 35th
Annual Power Electronics Specialists Conference , vol. 6, pp. 4285-4290, Jun. 2004.
[12] F. Katirae i, R. Ira vani, N. Hat ziargyriou a nd A. Dim eas, "Microgrids management,"
IEEE Power and Energy Magazine, vol. 6, pp. 54-65, 2008.

8
[13] R. C. Bansal, T . S. Bhatti a nd D. P . K othari, " On s ome of the design aspects of wind
energy conversion systems," Energy conversion and management, vol. 43(16), pp. 2175-
2187 , 2002.
[14] P. Delarue, A. Bouscayrol, A. Tounzi , X. Guillaud and G. Lancigu, "Modelling, control
and simulation of a n overall w ind e nergy c onversion system," Renewable En ergy, vol.
28(8), pp. 1169-1185, 2003.
[15] T. Petru and T. Thiringer, "Modeling of wind turbines for power system studies," IEEE
Transactions on Power Systems, vol. 17(4), pp. 1132-1139, 2002.
[16] A. Ha nsen, C. Jauch, P. Søre nsen, F. Io v a nd F. Blaabjerg, " Dynamic win d turbine
models i n power s ystem sim ulation tool DIgSILENT," Risø Na tional La boratory f or
Sustainable Energy, 2007.
[17] D. Das, S.K. Aditya and D.P. Kothari, "Dynamics of die sel and wind turbine generators
on an iso lated power system," International Journal of Electrical P ower & Energy
Systems, vol. 21(3), pp. 183-189, 1999.
[18] E. Mul jadi, S. Dr ouilhet and R. H olz "A nalysis of win d power f or battery c harging,”
Proceeding on Wind Energy Symposium, pp. 190-19728, United States, Jan.- Feb. 1996.
[19] K. Yoshimoto, T. Nanahara and G. Koshimizu, "New control method for regulating state-
of-charge of a ba ttery in hybrid wind power/battery energy storage system," Proceeding
on IEEE PES Power Systems Conference and Exposition, pp. 1244-1251, Oct.-Nov. 2006.
[20] A. Yazdani, "Islanded operation of a doubly-fed induction generator (DFIG) wind power
system with integrated energy storage," IEEE Electrical Power Conference, pp. 153-159,
Canada, Oct. 2007.
[21] "Outlook of energy storage technologies, 2007," Europian parliament policy department,
available online: http://www.storiesproject.eu/docs/study_energy_storage_final.pdf.
[22] A. Kh aligh a nd Z. Li, "Battery, ultracapacitor, fuel cel l, and hybrid energy storage
systems for electric, hybrid electric, fuel cell, and plug-in hybrid electric vehicles: state of
the art," IEEE Transactions on Vehicular Technology, vol. 59(6), pp. 2806-2814, 2010.
[23] R. Cardenas, R. Pena, G. A sher a nd J. Clare, " Power sm oothing in wind generation
systems using a s ensorless vector controlled induction m achine driving a flywheel,"
IEEE Transactions on Energy Conversion, vol. 19(1), pp. 206-216, 2004.

9
[24] R. Takahashi, Wu L i, T. Mura ta and J . Tam ura, "An application of f lywheel energy
storage system for wind energy conv ersion," Pro ceeding Inte rnational Conference on
Power Electronics and Drives Systems, pp. 932-937, Kuala Lumpur, Apr. 2006.
[25] S. Nomura, e t al ., "W ind f arms l inked by SMES systems," IEEE Transactions o n,
Applied Superconductivity, vol. 15(2), pp. 1951-1954, 2005.
[26] M. Shei kh, S. Muy een, R. T akahashi a nd J. T amura, " Smoothing control of wi nd
generator ou tput fluctuations by u sing su perconducting Magnetic Energy Storage un it,"
Proceeding International Conference on Electrical Machines and Systems, pp. 1-6, Nov.
2009.
[27] C. Abb ey and G. Joos, "Sup ercapacitor energy sto rage for wind en ergy a pplications,"
IEEE Transactions on Industry Applications, vol. 43(3), pp. 769-776, 2007.
[28] X. Li, C. H u , C. Li u a nd D. X u, "Modeling and control of a ggregated supe r-capacitor
energy storag e system fo r win d po wer g eneration," Proceeding on 34th Annu al
Conference of IEEE Industrial Electronics, pp. 3370-3375, Nov. 2008.
[29] L. Ntziachristos, C. Kouridis, Z. Samaras and K. Pattas, "A wind-power fuel-cell hybrid
system study o n the n on-interconnected Aegean islands g rid," Re newable Energy, vol.
30(10), pp. 1471-1487, 2005.
[30] O.C. On ar, M. Uzunoglu and M.S. Al am, "Mod eling, con trol a nd simulation of a n
autonomous wind turbine/photovoltaic/fuel cell/u ltracapacitor," Journal of Pow er
Sources, vol. 185(2), pp. 1273-1283, Dec. 2008.
[31] W. Cais heng a nd M. H. Nehrir, "Power m anagement of a sta nd-alone
wind/photovoltaic/Fuelcell energy system," IEEE Transaction on Energy Conversion, vol.
23(3), pp. 957-967, Sep. 2008.

10
Chapter 2 System Configuration and Sizing

2.1 Introduction
In this study, a multi sources hybrid distributed generation system‒consisting of a wind
turbine generator system (WTGS), a fuel cell (FC), an electrolyzer (ELZ) and an ultracapacitor
(UC)‒is proposed. Wind is the primary power source of the system to take the full advantage of
renewable energy around us. Since a reliable storage system is necessary with wind power, an
FC/ELZ/UC mix system is employed for the power and energy management. The FC/ELZ
combination is used as a backup and long term storage system, whereas the UC unit is used for
the fast transient and ripple power mitigation. In the following sections, a brief discussion of the
hybrid distributed energy resources is presented first, followed by the detailed system
configuration and unit sizing. The system components modeling and the system control schemes
will be discussed in chapter three and chapter four respectively.

2.2 Wind Turbine Generator System

The wind turbine generator system (WTGS) is a structure that converts kinetic energy of
wind into the electrical energy. The major components of a typical wind energy conversion
system include a wind turbine, a generator, interconnection units, and control systems. The
electromechanical conversion technologies adopted for wind turbines are synchronous and
induction generators. For small to medium power wind turbines, permanent magnet synchronous
generators or squirrel-cage induction generators are often used because of their reliability and
cost effective technologies, whereas doubly-fed induction generators (DFIGs) or permanent
magnet synchronous generators are currently used for the medium to high power wind turbines.
Modern wind turbines can be classified broadly as the fixed-speed and variable-speed wind
turbines. The fixed-speed wind turbines are directly connected to the grid using squirrel-cage
induction generators for electro-mechanical power conversion. The induction generator requires
reactive power to operate, which can be either supplied from the utility grid or by capacitors
connected at the machine terminals (see Fig. 2-1(a)). These generators cannot deliver any
reactive power and generally require a soft-starter to reduce the inrush current during the start-up
[1].

11
 (a)

(b)

(c)

Fig. 2-1 Types of WTGSs. (a) Fixed-speed WTGS, (b) full-scale converter based WTGS, and (c)
partial-scale converter based WTGS. Here, Ps, Pr, PWT, ωr and ωs represent stator active power,
rotor active power, power from wind turbine, rotor angular speed and synchronous speed of
rotating flux, respectively.

12
 The variable-speed wind turbines (WTs) use power electronic converters (PECs) that act as
interfacing devices to achieve power control, soft start, and interconnection functions. These
WTs are further classified into the turbines with full-scale and partial-scale rated PECs systems.
With the full-scale rated converter system, the generators are usually conventional or permanent
magnet type synchronous generators to convert the turbine power into the variable-frequency
variable-voltage electric power that changes with the wind speed (see Fig. 2-1 (b)). A power
electronic based rectifier and an inverter are then used to convert the generator output power to
the power that is compatible with the utility system. This topology causes extra losses in the
power conversion system, but it provides the additional power gain through the variable-speed
operation by capturing more energy from the wind [1], [2].
In this study, partial-scale rated PECs system is used, which is a compromising solution
between fixed-speed operation (in term of cost) and full-scale PEC based variable-speed
operation (in term of operational flexibility). This configuration has been widely adopted
commercially in medium to large scale WTGSs. With the partially-rated converter system, a
DFIG is used as the electro-mechanical power conversion system, in which a voltage source
converter (VSC) connected to the rotor through slip rings controls the rotor current (see Fig. 2-1
(c)). This configuration enables the WTGS to have some degree of variable-speed operation. If
the generator is running super-synchronously, the electrical power (Ptotal) is delivered through
both the rotor (Pr) and the stator (Ps). If the generator is running sub-synchronously, the
electrical power is delivered into the grid through the stator terminal only. The advantage of this
design is that it provides reactive power compensation and increased energy capture from the
wind. This arrangement further allows the rectifier system to undersize about 30% of the full
load value, which equals to the maximum power flow from the rotor circuits [2].

2.3 Fuel Cells

Fuel cells are the static electrochemical devices which convert chemical energy of H2 fuel
into electrical energy. They generate electricity inside a cell through reactions between a fuel
(H2) and an oxidant (O2) in the presence of an electrolyte, in which the reactants flow into the
cell and the products flow out of it. Unlike conventional electrochemical batteries, fuel cells are
the thermodynamically open system, such that the reactant from the external source is consumed
[3]. By contrast, the batteries store electrical energy chemically, and hence represent a

13
thermodynamically closed system.
The difference between the fuel cell and a combustion engine depends on the manner by
which energy is converted. In internal combustion engines, the heat generated from combustion
needs to be converted into mechanical energy and then into electrical energy, causing losses at
each level of transformation. In contrast, fuel cells directly produce electricity from the chemical
reaction. Fuel cells offer numerous advantages over conventional power plants to help them
achieve the goal and widespread adoption, such as [4]:
 high efficiency even at partial load,
 few moving parts resulting in quiet operation, higher reliability, lower maintenance and
longer operating life,
 fuel diversity,
 zero or low emission of greenhouse gases,
 combined heat and power capability, without the need for additional systems (i.e., low
temperature fuel cells can provide district heating, while high temperature fuel cells can
provide high-quality industrial steam),
 flexible and modular structure, and
 increased energy security by reducing the reliance on large central power plants and oil
imports.
In a typical fuel cell, fuel is fed continuously to the anode and oxidant is fed continuously
to the cathode. The electrochemical reactions take place at the electrodes to convert chemical
energy into electricity. The anode is the electrode from which electrons leave and cathode is the
electrode to which the electrons are coming. The most commonly used fuel for fuel cells is
hydrogen, while oxygen or air acts as the oxidant. Among the various types of fuel cells (FCs),
the polymer electrolyte membrane fuel cell (PEMFC) and solid oxide fuel cell (SOFC) have
good prospective for the distributed power generation [4-6]. In a PEMFC, H+ flows across the
electrolyte as a charge carrier producing H2O at the cathode; whereas O2- flows across the
electrolyte as the charge carrier in an SOFC, producing H2O at the anode. The respective
electrochemical phenomena are also illustrated in Fig. 2-2.

14
 (a) (b)
Fig. 2-2 Schematic block diagram of FCs. (a) SOFC (b) PEMFC

In this dissertation, the SOFC has been utilized because of its notable merits: high
efficiency, long-term stability, fuel flexibility, low emissions, and relatively low cost, despite of
their few constraints related to the high operating temperature which results in longer start-up
times and mechanical/chemical compatibility issues. The energy conversion efficiency of the
SOFC stack can reach up to 65% and its overall efficiency with combined heat and power
applications, can reach even above 80% [7], [8]. Further, due to the high operating temperature,
the SOFC allows internal reforming of gaseous fuel inside the fuel cell, providing the multi-fuel
applicability [4].

2.4 Ultracapacitor
Ultracapacitors (supercapacitors, double-layer capacitors or electrochemical capacitors) are
energy storage devices which polarize an electrolytic solution to store energy electrostatically.
Though it is an electrochemical device, no chemical reactions are involved in its energy storage
mechanism. The storage mechanism is highly reversible that allows the UC to be charged and
discharged hundreds of thousands of times. Consequently, it can store more energy than
traditional capacitors and discharge this energy at higher rates than rechargeable batteries.
Ultracapacitors are notable for their ability to store more energy per unit weight and volume than

15
conventional capacitors. Therefore, UCs are attractive in power application for transient
mitigation with renewable power generation systems and plug-in-hybrid-vehicles [9],[10].
Fig. 2-3 depicts the schematic diagram of an ultracapacitor cell, which can be viewed as
two nonreactive porous plates, or collectors, suspended within an electrolyte. The main factors
determining how much electrical energy a capacitor can store are: the surface area of the
electrodes, their distance from each other, and the dielectric constant of the material separating
them. Because of the presence of porous material having pore diameter in the nanometer range,
the charge can be sponged into it, and the effective surface area will be more than 100000 times
than the outer surface area. Consequently, the capacitance would be very high because more
charge can be stored on larger surface area in compare with the conventional capacitor having
the same size [10], [11].

Fig. 2-3 Schematic diagram of a typical UC. With finer dimensions and more uniform distribution,
carbon nanotubes enable greater energy storage in ultracapacitors than activated carbon does. The
photograph is taken from [10].

The competitive technologies of UCs, in terms of the transient mitigation ability

characterized by their high power density, are the flywheel, SMESs and battery systems (see
Table A.1 in appendix). These technologies have own merits and demerits. Despite of the
mature and cost effective solution, the application of lead acid battery is limited due to its limited
charging and discharging cycle and the disposal problem. Although the flywheel demonstrates a
high efficiency, it requires large space for the moving part. In case of the SMES, its wide scale

16
application is limited due to its high cost per kWh, unless some breakthroughs are achieved in its
technologies. Therefore, the UC is adopted in this study as a compromising candidate in terms of
cost, compatibility, efficiency and operating life.

2.5 Electrolyzer
In the context of H2 technology, electrolyzers (ELZs) are the devices to produce hydrogen
and oxygen from the electrolysis of water, which can be widely distributed to meet the hydrogen
and oxygen requirements of different users, such as renewable energy systems, fueling stations
and industrial applications. Recently, there is a rising trend of utilizing ELZs as distributed
energy resources with the renewable power systems like photovoltaic (PV), wind power and
small hydro power generation systems [12], [13].
The electrolysis process can also be considered as the reverse process of hydrogen fueled
fuel cell, in which H2 and O2 undergo electrochemical reaction to generate power. When an
electrical power source is connected to two electrodes (typically made from some inert metal
such as platinum or stainless steel) and placed in the water, hydrogen is produced at the cathode
and oxygen is produced at the anode. The electrolysis cells are characterized by their electrolyte
type. There are two types of low temperature electrolysis: alkaline and proton exchange
membrane. Alkaline electrolysis utilizes a liquid electrolyte consisting of highly concentrated
potassium hydroxide (KOH), whereas the proton exchange membrane electrolysis is based on
the use of a solid conducting polymer that conducts ions when hydrated with water.

Fig. 2-4 Schemaic diagram of alkaline electrolyzer.

17
 Among these electrolyzers, the alkaline electrolyzer (AE) is the well-established and the
dominating technology today with its availability in a wide range of power ratings at a lower cost,
therefore alkaline water electrolysis is adopted in this study. Fig. 2-4 depicts the operating
principle of an AE which employs an alkali solution (usually potassium hydroxide of around 25-
30% concentration at 80oC) as the electrolyte for providing the ion-transfer process. Typical
current densities of the AE are 0.2-0.4 A/cm2 and designed for either atmospheric or operation at
up to 30 bar. The energy conversion efficiencies can range from 60 to 90%. Without auxiliary
purification equipment, purities of 99.8% for H2 can be achieved. The alkaline electrolysis
technology can be implemented at a variety of scales from less than 1 kW to large industrial
electrolyzer plant over 100 MW as long as there is availability of a proper dc electricity supply
[14].

2.6 Power Electronic Converters

Due to uncertain and variable nature of energy input, the voltage and frequency from
renewable power sources vary accordingly. Therefore, unlike the conventional generators, they
cannot be connected to the point of common coupling (PCC) bus directly. Further, most of the
distributed power sources generate power in a dc form, which is not compatible with the ac load
or the power network. A power electronic converter (PEC) system plays a vital role to match the
characteristics of the distributed generation units and the requirements of the interconnection;
including the frequency, voltage, the active and reactive power regulation, minimization of
harmonics, among the others [8]. Therefore, power electronics technology is an indispensable
part of any renewable power systems, and it often shares upto 40% cost of the combined system
[15], [16].

A PEC system accepts power from the distributed energy source and converts it into the
desired voltage and frequency level. There are various types of PEC topologies, namely ac-ac,
ac-dc, dc-ac and dc-dc interfaces. The choice of a particular one is specific to the type of the
energy source or the storage application. An ac-ac conversion topology, such as cycloconverters
or matrix converters, is used when power is generated in ac form with a frequency different from
the load frequency like in microturbines, wind power etc. The ac-dc conversion topology, also
known as the rectifier, is utilized to generate a controlled dc voltage from either an uncontrolled
ac source (i.e., microturbine, wind turbine) or a controlled ac source (i.e., utility supply) [17]. For

18
dc output systems like photo voltaics, fuel cells, batteries and ultracapacitors, a dc-dc converter
is typically needed to change the dc voltage level. The dc-ac inverter is the most generic
converter which converts a dc source to the grid-compatible ac power. Further, a bidirectional
interfacing is required with the storage systems to exchange power with the external system.
Table 2.1 summarizes the type of the power electronic conversion topologies to be used with
distributed energy resources (DERs).

Table 2.1 Power electronic converters for different distributed energy resources

Distributed energy resources type Power conversion Common module name

Wind, micro-turbine, Internal ac-dc Rectifier
combustion engine, flywheel
Battery, SMES dc-ac Inverter
Fuel cell, electrolyzer, dc-dc Boost, buck, buck-boost,
ultracapacitor, SMES chopper
Wind, micro-turbine ac-ac Cycloconverter, matrix
converter
Wind turbine, flywheel ac-dc-ac Back-to-back converter,
rectifier inverter

Fig. 2-5 General block diagram of power electronic converters for DERs.

A general approach to organize the interface into modules, in which each can be designed
with a range of distributed generators (DGs) and storages, is illustrated in Fig. 2-5. It includes
four major modules for a power electronic conversion, namely the source input converter module,

19
an inverter module, the output interface module and the controller module. The first and second
modules are the power electronic converter units, while the third and fourth are their associative
parts. The blue unidirectional arrows depict the power flow path for the distributed energy
resources, whereas the red arrows show the bidirectional power flows for the energy storage
devices [16], [18].
In this study, wind power represents the DG unit, and a UC, an SOFC and an ELZ are
utilized as the energy storage devices. As explained in the earlier section, the wind turbine adopts
the back-to-back ac-dc-ac converters. Unlike in a usual DFIG configuration where the rotor side
converter (ac-dc) and the line-side converter (dc-ac) are equally rated, the configuration in this
study adopts a full-scale line-side converter. The reason of the increased size of the line-side
converter is to handle the storage system to be connected to the intermediate dc bus. For the
SOFC/ELZ and the UC, a two stages dc-dc conversion system is applied; in which the
bidirectional chopper accommodates the SOFC and ELZ, and the full bridge isolated dc-dc
converter regulates the dc-link voltage isolating the energy storage system from the rest of the
network. The detailed configuration, topology and sizing of the hybrid system are discussed in
the following sections.

2.7 Layout of Proposed Scheme

In a hybrid power system, DGs can provide the terminal power in the form of dc or ac, so
their integration can be done through the ac or dc coupling. In a dc coupling configuration,
different alternative energy sources are connected to a dc bus through appropriate power
electronic converters. Then the dc energy is converted into 60 Hz (or 50 Hz) ac form through a
dc-ac converter which can be bi-directional. The dc consumers are also coupled to the dc side,
whereas the ac consumers are connected to the ac terminal of the inverter. The power range for
dc coupled hybrid systems is broad and can be used cost-effectively for various off-grid
applications. In this configuration, any numbers of DGs can be connected to the dc bus with a
proper power management scheme and seen as a “virtual power plant” from the grid side. In an
ac coupled hybrid system, the DGs are linked at the regulated ac bus through proper PEC(s). In
general, when systems increase in size, they are implemented as ac coupled hybrid systems, and
the extent to which dc buses are avoided can vary. In the ac coupled hybrid systems, an island
grid or an isolated microgrid is built and different concepts can be utilized to operate it. One of

20
the merits of ac linked systems is that the inverter can be employed to share active/reactive
power in a relatively easy control structure [21].
In the both concepts mentioned above, the system has to cope with extra losses due to the
inversion and rectification of the current for each energy flow. This two-stages conversion
significantly reduces the overall system efficiency, since the number of PECs is not optimized.
The third hybrid configuration, which is the mix of both ac and dc coupling, can be designed to
exploit the benefit of both configurations by optimizing the number of converters used, so it is
adopted in this study.
The schematic diagram of the proposed wind power and storage system (WPSS) is
illustrated in Fig. 2-6. The system consists of a wind turbine, a doubly-fed induction generator
(DFIG), a dedicated energy storage system (DESS), two bi-directional voltage source converters
(VSC1 and VSC2) and the local load. The configuration can be visualized as a subset of a
consumer scale local network or a microgrid that can be operated autonomously in different
modes. For the medium size power generation, the DFIG is a recommended choice due to its
variable speed operation with a reduced size of the ac-dc-ac converter [2], so the DFIG is
adopted here with a partially rated (30%) ac-dc conversion system (VSC1). In this topology, the
rating of the line-side dc-ac converter (VSC2) is higher, because it should handle the storage
power as well. The stator terminals of the DFIG are directly connected to the ac bus, whereas the
rotor terminals are connected to the intermediate dc bus through VSC1, so that the rotor circuit
would participate to transfer active and reactive power. As shown in the figure, the line-side
converter and the DESS are connected in parallel at the dc link bus. The line-side converter
provides the power electronic conversion between the regulated dc-link bus and the coupling bus
(bus #2). In the conventional operation as the grid-connected non-dispatchable power source,
line-side converters are mainly assigned to control dc-link voltage and ensures the converter
operation at the desired power factor. By introducing the DESS at the dc-link bus, the line-side
converter provides the increased flexibility of operation that evolves the WPSS, making it self-
sustainable and dispatchable entity. The ac buses (bus #1 and bus #2) may either be lumped in a
single bus or connected by a line depending on their distance. The whole system is connected to
the local distribution grid via a breaker (B) whose switching state decides whether the system is
in islanded or grid-connected mode.

21
 Ptie Qtie

Rgd
Lgd
wind

Pr,Qr
control
WTGS

Ll
VSC1

Rl
Energy Storage
Controller
Storage

(a)
Fuel flow Power Signal
O2
Fuel Ii Idc1 Vdc Vuc
Regulation
H2
O2 H2
Controller

+ Ifc
FC
Li Li
Ii Isto
-
Vpwm Idc1
Vdc
Cuc
Vuc
ELZ
+
I elz DC1 DC2

(b)
Fig. 2-6 Schematic diagram of proposed system. (a) Overall system topology, and (b) the dedicated
energy storage subsystem. Here the symbols I,V,P,Q,L and R represent the current, voltage, active power,
reactive power, inductance and resistance, respectively, and the subscripts denote the quantities for the
corresponding part/unit.

22
 The configuration of the DESS consists of an UC, a solid oxide fuel cell (SOFC) and an
ELZ with a H2 storage tank, as illustrated in Fig. 2-6(b). The UC and the SOFC/ELZ are
connected electrically at the dc-link bus through dc-dc interfaces, DC1 and DC2. It is notable
that the proposed scheme accommodates the ELZ and SOFC as a single unit, which can share the
same converter (DC1) for the control of current flow. The DESS is capable of performing
versatile activities; it monitors and controls many variables (currents and voltages of the UC,
SOFC and ELZ, and fuel flow) which are applicable to manage transient and long-term power
balance. The DESS can also assist in the supervisory control and demand response by providing
the information of fuel reserve. In the worst case, when there is no reserve of H2 or the DESS is
out of service, it delivers its status signal to the VSC2 subsystem availing to alter the latter’s
operational scheme without disruption.
Since the power from the DFIG is non-dispatchable in nature and often regarded as a
negative load or disturbance input, the DESS plays a role of the reliable power/energy balancing
source. In particular, it is responsible for the voltage control of dc-link bus (Vdc) by enabling
DC1 to inject the fast transient power and DC2 for the steady state power balance. The main
focus of DC2 control action is on the inherently slow fuel cell dynamics which renders it unable
in tracking fast current, leading to undesired pressure drop and fuel starvation [6]. During the
surplus of power, ELZ/UC is operated to produce the H2, while SOFC/UC comes into effect
during the power deficit by the consumption of H2. As a consequence, VSC2 becomes the fully
controllable unit responsible for the voltage (reactive power) and the frequency (active power)
regulation of bus #2 by adjusting the unbalanced power with the DESS.

2.8 Unit Sizing

Unit sizing in a power system is actually a multi-objective optimization problem that
addresses number of operational criteria; such as minimizing the electricity production cost,
ensuring that the load is served according to certain reliability criteria, and minimizing the loss
[19], [20]. In the case of a generation system that uses renewable sources, the sizing is more
complex in comparison with the conventional systems, due to the randomness of the renewable
resources and the high cost of wind generators and storage modules [20]. Since the investigation
of optimal unit sizing is beyond the scope, this study considers the operating requirement of the
system in sizing the individual components.

23
 While sizing the units, it is roughly estimated that the generation of energy would match
the demand in long term by using H2 storage, with no need of importing it. The maximum load
of the system is kept equal to the rated capacity of the WTGS as 50 kW. Assuming that 20% of
the connected load can be curtailed, an SOFC with 40 kW would be able to supply
uninterruptable load during the worst situation when the DFIG is not generating any power.
Similarly, the size of the ELZ is selected considering the extreme situation that the rated power
from the WTGS would be consumed by the ELZ, with all loads are switched off. That would
make the rating of the ELZ about 50 kW. The ratings of the other components (DC1, DC2 and
VSCs) are selected as the technical demand of the configuration. That would make the ratings of
DC1 and DC2 as 50 kW, the rating of rotor side converter (VSC1) as 20 kVA, and the rating of
line-side converter (VSC2) as 75 kVA.

2.9 Summary
This chapter presented a brief discussion of hybrid distributed energy sources and power
electronic conversion mediums utilized in the study. Distributed energy sources consist of
diverse nature of generators that produce electricity in dc or ac forms. Beginning with brief
discussion of energy resources and power electronic converters, the layout of proposed system
has been illustrated in detail followed by the unit sizing of the system.

References
[1] Z. Chen and F. Blaabjerg, "Wind Energy –The World’s Fastest Growing Energy Source,"
IEEE Power Electronics Society , Third Quarter Newsletter, 2006.
[2] F. Blaabjerg and Z. Chen, "Power electronics for modern wind turbines," Synthesis
Lectures on Power Electronics, vol. 1, pp. 1-68, 2005.
[3] M. Winter and R. Brodd, "What are batteries, fuel cells, and supercapacitors?," Chem.
Rev, vol. 104, pp. 4245-4270, 2004.
[4] A. Dicks and J. Larminie, Fuel cell systems explained: John Wiley & Sons, 2000.
[5] O.C. Onar, M. Uzunoglu and M.S. Alam, "Modeling, control and simulation of an
autonomous wind turbine/photovoltaic/fuel cell/ultracapacitor," Journal of Power
Sources, vol. 185(2), pp. 1273-1283, Dec. 2008.

24
[6] P. Thounthong, S. Raël and B. Davat, "Control strategy of fuel cell and supercapacitors
association for a distributed generation system”, IEEE Transactions on Industrial
Electronics, vol. 54(6), pp. 3225 - 3233, Dec. 2007.
[7] C. Wang and H. Nehrir, "A physically based dynamic model of solid oxide fuel cells,"
IEEE Transaction on Energy Conversion, vol. 22(4), pp. 887 - 897, Dec. 2007.
[8] F. Blaabjerg, Z. Chen and S. Kjaer, "Power electronics as efficient interface in dispersed
power generation systems," IEEE Transactions on Power Electronics, vol. 19(5), pp.
1144-1154, Sep. 2004.
[9] C. Abbey and G. Joos, "Supercapacitor energy storage for wind energy applications,"
IEEE Transactions on Industry Applications, vol. 43(3), pp. 769-776, 2007.
[10] J. Schindall, "The charge of the ultracapacitors," IEEE Spectrum, vol. 44, pp. 42-46, 2007.
[11] A. Burke, "Ultracapacitors: why, how, and where is the technology," Journal of power
sources, vol. 91(1), pp. 37-50, 2000.
[12] T. Zhou and B. Francois, “Modeling and control design of hydrogen production process
for an active hydrogen/wind hybrid power," International Journal of Hydrogen Energy,
vol. 34(1), pp. 21-30, Jan. 2009.
[13] Z. Yumurtaci and E. Bilgen, "Hydrogen production from excess power in small
hydroelectric installations," International Journal of Hydrogen Energy, vol. 29(7), pp.
687-693, Jul. 2004.
[14] A. Smith and M. Newborough, "Low-cost polymer electrolysers and electrolyser
implementation scenarios for carbon abatement," Report to the Carbon Trust and ITM
Power, 2004.
[15] S. Blazewicz, "Distributed energy resources integration research program power
electronics research assessment," Navigant Consulting California Energy Commission
vol. CEC-500-2005-206, Sep. 2005.
[16] W. Kramer, S. Chakraborty, B. Kroposki, and H. Thomas, "Advanced power electronic
interfaces for distributed energy systems part 1: systems and topologies," National
Renewable Energy Laboratory, Golden, Colorado, Tech. Rep. NREL/TP-581-42672,
2008.
[17] Y. Khersonsky and G. Robinson, "PEBB modules in distributed generation applications,"
IEEE Power Engineering Society General Meeting, Jul. 2003.

25
[18] S. Chakraborty, B. Kramer and B. Kroposki, "A review of power electronics interfaces
for distributed energy systems towards achieving low-cost modular design," Renewable
and Sustainable Energy Reviews, vol. 13(9), pp. 2323-2335, 2009.
[19] R. Chedid and S. Rahman, "Unit sizing and control of hybrid wind-solar power systems,"
IEEE Transactions on Energy Conversion, vol. 12(1), pp. 79-85, 1997.
[20] D.B. Nelson, M.H. Nehrir and C. Wang, "Unit sizing and cost analysis of stand-alone
hybrid wind/PV/fuel cell power generation systems," Renewable energy Journal, vol.
31(10), pp. 1641-1656, 2006.
[21] K. Moutawakkil and S. Elster, "RE hybrid systems: Coupling of Renewable Energy
Sources on the AC and DC Side of the Inverter" Refocus, vol.7(5), pp. 46-48, 2007.

26
Chapter 3 System Modeling

The previous chapters described reasons for the interest and application of two time scale
based energy storage system, namely the electro-chemical based long-term energy storage and
the utracapacitor based short term energy storage in a wind power based microgrid, with the
suitable inter-connecting layouts and sizing. This chapter will investigate the appropriate models
of the components applicable to this study, and provide a detailed mathematical treatment to
each of them; thereby presenting a computational platform useful for analyzing the hybrid power
system.
For the effective analysis and testing, it is better to have a real time set up and observations
of the system. However, this approach presents some drawbacks which cannot be overlooked,
such as high costs of the elements under test, the infrastructure and security requirements
(especially for a hydrogen storage and supply system) and the complexity associated to the
performance of the test when a high number of elements are involved. Therefore, the most cost
effective approach at the designing stage is to carry out computational analysis based upon the
mathematical models involved in the system. Although a physical system cannot exactly be
described by a mathematical expression, it is often convenient to utilize an approximated model
for design and analysis purposes in engineering systems. In our case, the system includes a large
number of components with diverse nature of dynamics. Since it is too complex and time
consuming to take account of the dynamics in much greater depth, we will adopt those models
which are adequate and simple enough to observe the system behaviors within the framework of
the research objectives.
On this ground, the following sections will explain the dynamic model of the subsystems,
namely the dedicated energy storage system (DESS), the wind turbine generator system (WTGS)
and the line-side converter system. The DESS includes a fuel cell, an electrolyzer, an
ultracapacitor, two dc-dc converters, and a H2 regulation/storage unit. The WTGS includes a
wind turbine, a doubly-fed induction generator (DFIG) and a voltage source converter (VSC1).
Likewise, the line-side converter system has a voltage source converter unit (VSC2). In the
following sections, the dynamic model for each of the subsystems is presented in detail. These
models will be used in chapter four to design the control and management schemes and in

27
chapter five to innvestigate ooperational modes andd behavior of the oveerall system
m through
numericcal simulatioons.

3.1 W
Wind Enerrgy Estim
mation
Thhe energy off wind is inn its kinetic which is a varying
v quaantity affected by severral factors,
such ass geographhic locationn, climate characterisstics, heighht above ground,
g andd surface
topograpphy. Wind tturbines interact with the
t wind, caapture part oof its kineticc energy, annd convert
it into thhe usable ennergy. The kknowledge of
o the wind distributionn is very cruucial in desiigning and
sizing oof a wind energy connversion syystem. The most comm
monly usedd probabilitty density
functionn to describbe the wind speed and aavailable ennergy is the Weibull funnction, basedd upon the
annual wind
w speedd data [1]. T
The Weibulll distributioon is describbed by the ffollowing pprobability
functionn,
k 1
k v  k
f (vw )   w  e( vw / c ) , (3.1)
c c 
where k is a shape pparameter, c is a scale pparameter annd vw is the wind speedd. The averaage wind
speed (oor the expeccted wind sppeed) can bee calculated as

c 1
vw   vw f (vw )dvw   , (3.2)
0
k k
where  is the Eulaar Gamma ffunction deffined by

( z )   t z 1et dt. (3.3)
0

Fig. 3-1 Probability density of the Rayleigh distribution.

d T
The average w
wind speeds are 5.4 m/s ((solid), 6.8
m/s (dashhed) and 8.2 m
m/s (dotted).

28
 For the application of wind power system, the Weibull distribution function can be reduced
to the single parameter model, in which k = 2, such that a simpler Rayleigh distribution is
constructed. Fig. 3-1 shows the annual Rayleigh distribution of wind speed. The average wind
speeds in the figure are 5.4 m/s, 6.8 m/s and 8.2 m/s. A wind speed of 5.4 m/s correspond to a
medium wind speed site in Sweden [2], while 8–9 m/s are wind speeds available at sites located
outside the Danish west coast [3].
The probability distribution function is utilized in sizing the wind turbine and determining
its rated wind speed. Considering the aerodynamic power in the wind passing through an area
(Aw) as

Pwind  Awvw3 , (3.4)
2
the average power in the wind would be [1]


Pwind  Aw  vw3 f (vw )dvw . (3.5)
2 0

Here, ρ represents the density of air. The function v w3 f ( v w ) starts at zero for vw  0 , reaches a

peak value at some wind speed vwm and finally returns to zero at large values of vw . The yearly

energy production in kWh at the wind speed vw is the power obtained multiplied by the number

of hours in the year. The wind speed vwm is the speed which produces more energy than any

other wind speed. Therefore, the maximum energy obtained from the given wind speed vwm is


Wmax  3
8760 Awvwm f (vwm ). (3.6)
2
The turbine should be designed to maximize the energy in its best operating wind speed
range. Some applications will even require a turbine to be designed with a rated wind speed
equal to vwm . This can be found by multiplying right hand side of (3.1) by vw3 , setting the

derivative equal to zero, and solving for vw as

1/ k
k 2
vwm  c  . (3.7)
 k 
Thus, we see that vwm is greater than c, so it will be greater than the mean speed, vw . If the

mean speed is 6 m/s, then vwm will typically be about 8 or 9 m/s.

29
3.2 Wind Turbine Generator System
A general structure of a WTGS is depicted in Fig. 3-2. It consists of important subsystem
models; a wind speed model, a rotor, a drive train and a generator. In this section, the modeling
of these subsystems, which are of interest in this thesis, are discussed.

Fig. 3-2 General structure of a wind turbine generator system.

3.2.1 Wind speed model

The wind speed model gives the time series data of wind speed. One approach to model a
wind speed sequence is to use data from measurements. If the certain wind speed range or
turbulence intensity, which is not available in the measured data, is to be simulated, a user
defined model is often used. This makes it possible to simulate a wind speed sequence with the
desired characteristics, by setting the corresponding parameters to the appropriate values [4]. In
the simulation of wind power in an electrical power system, it is often assumed that the wind
speed is made up by the sum of the following four components [4], [5]:
a) the average value;
b) ramp component, representing a steady increase in wind speed;
c) gust component, representing a gust; and
d) turbulence component, representing fast variation.
This leads to the following equation,
v w (t )  v wa (t )  v wr ( t )  v wg (t )  v wt ( t ), (3.8)

in which vwa is the average value of the wind; vwr is the ramp component; vwg is the gust
component; and vwt is the turbulence component. The ramp component is characterized by three
parameters: the amplitude of the wind speed ramp (Arm), the starting time of the wind speed
ramp (Tsr), and the end time of the wind speed ramp (Ter) as

30
  0, for t  Tsr ;

 (t  Tsr )
vwr   Arm , for Tsr  t  Ter ; (3.9)
 (Ter  Tsr )
 Arm , for t  Ter .

The gust component is characterized by three parameters: the amplitude of the wind speed gust
(Ag), the starting time of the wind speed gust (Tsg), and the end time of the wind speed gust (Teg).
The wind gust is modeled using the following equation,
 0, for t  Tsg ;

   t  Tsg 
vwg  0.5 Ag 1  cos 2    , for Tsg  t  Teg ; (3.10)
  T T
  eg sg  

 0, for t  Teg .

Finally, the turbulence component of the wind speed is characterized by a power spectral density.
Here, the following power spectral density is used
1 5/3
  h 2   f l
PDt ( f w )  lw ln    1  1.5 w  , (3.11)
      vwa 

which depends upon the tower height (h), turbulence length (l), frequency (fw), and roughness
length (  ) that defines the landscape of the region. A power spectral density can be used to
derive information about the amplitude of a signal’s component at a given frequency. Then, a
large numbers of sinusoidal, with a random initial phase angle and amplitude calculated from the
power spectral density, are added for each time step. Thus, we can generate a time domain signal
with a power spectral density that is the sampled equivalent of the original power spectral
density.

3.2.2 Wind turbine model

Wind turbine is the medium which convert the aerodynamic energy from the wind into the
mechanical energy in the form of blades’ rotation, which is then used to drive a generator that
converts this energy into electricity. In comparison to the vertical axis type, the horizontal axis
type turbine is more favorable due to their higher energy capturing capacity (due to the greater
tower height) with more efficient propellers [6]. Therefore, this dissertation has adopted the
horizontal axis wind turbine in the study and the corresponding modeling is presented here.

31
 The steady-state mechanical power ( PWT ) available at the turbine shaft, can be determined

by [4]
PWT  0.5C P (tip ,  )  Aw vw3 , (3.12)
where
 
C P (tip ,  )  c1 c2 / i  c3   c4  c5  c6 exp(c7 / i ) 

1 1 c9 
  3 . (3.13)
i tip  c8    1 
tip  rT r / vw 

Here, λtip is the tip speed ratio (TSR), β is the pitch angle of turbine blade, ρ is the air mass
density, rT is the radius of turbine rotor, ωr is the rotor angular speed, and ci (i=1,2..7) are the

constant parameters. CP is the power conversion coefficient which varies with λtip and β.

Equation (3.12) shows that PWT is a fraction of the available power at the rotor shaft determined

by CP . The theoretical maximum value of CP is about 59% as derived by the Belz law [1], [4].

For practical cases, it is well below 59%. The plot of CP vs. tip is shown in Fig. 3-3(a). It can be

noted that CP is maximum at the certain value of β (zero degree in this case). In a variable-speed

wind turbine, this curve is utilized to set CP to its maximum value ( CP-max ) by forcing λtip to its

optimum value ( tipopt ), as long as the power or rotor speed is below its rated value. This, in turn,

is made by adjusting the rotor speed to its optimum value (  ropt ) as,

tipopt
 opt
r  vw . (3.14)
rT
Consequently, the mechanical power output from the turbine will be the maximum. The locus of
the maximum power over a wide range of vw is also depicted in PWT vs. ωr plot in Fig. 3-3(b) by
the black-dashed line. This concept is utilized for the maximum power point tracking (MPPT)
scheme.

32
 CP

λtipp

(a)

(b)
Fig. 3-3 Wind tuurbine characcteristic curvees. (a) CP as a function off TSR and (b) power vs. rrotor speed
ccharacteristiccs of proposeed wind turbiine.

33
 D

 max

III

II c
B

rmax I
vwmax
A
vwnom
rmin

Fig. 3-4 Typical variable speed pitch control strategy plotted on the vw-ωr-β space.

For the higher wind speeds, mechanical power output from the turbine becomes greater
than its rated value. There are two major ways of performing the aerodynamic power control, a
stall control (passive or active) and a pitch control. The stall control implies that the blades are
designed to stall in high wind speed and no pitch mechanism is required [7],[8]. The pitch
control is the most common method of controlling the aerodynamic power generated by the
turbine rotor in modern variable-speed wind turbines. In the pitch control, the pitch angle is
regulated to decrease the angle of attack, i.e. the angle between the chord line of the blade and
the relative wind direction so that fraction of aerodynamic power spills away from the turbine
fans. Fig. 3-4 illustrates a typical variable speed pitch control strategy plotted on the vw-ωr-β
space. For the wind speed between 4 m/s to 10 m/s (region I),  ropt is set according to (3.14); for
the speed between 10 m/s to 12 m/s (region II), it is transition of low to high speed region; and
for the speed greater than 12 m/s, the rotor speed is regulated at its rated value by utilizing the
pitch angle regulation. An alternative treatment of power capture vs. wind speed is illustrated in
Fig. 3-5 which indicates the different operating regions. Below 4 m/s, there is no power
generation; for 4  vw  12 m/s, there is maximized generation by keeping CP to CP-max; for

34
12  vw  25 m/s, there is constant generation through the pitch regulation; and above 25 m/s,
wind turbine is shut down to protect it from storm.

Fig. 3-5 Variable speed pitch controlled wind turbine operation regions.

0.5 Aw

(r )

( )

Fig. 3-6 Simulation model for wind turbine.

Fig. 3-6 illustrates the simulation model of the wind turbine developed for this study, in
which the inputs are: wind speed, air density, radius of the wind turbine, mechanical speed of the
rotor referred to the wind turbine side and the pitch angle. The output is the torque (Tm) which
drives the electrical generator. The wind turbine calculates the TSR from the input values and
estimates the value of the power coefficient from the performance curves. Here, inputs r and

 are the controlled inputs, in which r is regulated by speed controller and the pitch angle

35
controller maintains the value of the blade pitch at the optimum value until the power output of
the wind turbine equals the rated power.

3.2.3 Doubly-fed induction generator model

A schematic diagram of a DFIG system with a back-to-back converter is shown in Fig. 3-7.
The back-to-back converter consists of two converters: a rotor-side converter (VSC1) and a line-
side converter (VSC2) connected back-to-back. Between the two converters, a dc-link capacitor
is placed as energy storage to suppress the voltage variations in the dc-link bus. At this bus, other
energy storage devices can be augmented for power and energy management. With the converter
VSC1, it is possible to control the torque or the rotor speed and the power factor of the stator
terminals; while the objectives of the converter VSC2 vary depending upon the operational
modes. It is notable that in a DFIG, the power flow direction through the rotor is negative or
positive depending upon the wind speed. For positive slip (lower wind speeds), the rotor draw
power from the coupling bus; at zero slip, there is no power transfer from the rotor circuit; and at
negative slip (higher wind speeds), the rotor also supplies power to the coupling bus (see Table
3.1). In following subsections, the detailed model of each component of the DFIG is presented.

Fig. 3-7 Schematic diagram of a typical "loss-less" DFIG. It shows the power flow components through
stator (Ps) and rotor circuit (Pr) as a function of slip (sl).

Table 3.1 Example of the power flow for different slips of a DFIG system.

Slip, sl Rotor speed, ωr [pu] Rotor power, Pr Stator power, Ps

0.3 0.7 -0.43 PWT 1.43 PWT
0 1 0 PWT
-0.3 1.3 0.23. PWT 0.77 PWT

36
a) Machine model
The induction machine considered here has a three-phase stator and a wound rotor in which
the terminal of the rotor windings are accessible from the slip rings. During the modeling, we
adopt the classical assumptions; such as linearity of the materials (no saturation), sinusoidal
distribution of the field in the air-gap, core losses are negligible, changes in resistance due to
heating are negligible, and a balanced structure. Further, the wind turbine generator system
assumes the wind turbine and the DFIG as a lumped unit, represented by their equivalent
parameters: the moment of inertia and the damping coefficient. The equivalent circuit of a DFIG
in the reference frame rotating at synchronous angular speed (ωe) is shown in Fig. 3-8 [9],[10].
The expressions of the dynamic equation are given in the space vector form, in which
ψ , i, and v represent space vector for flux, current, and voltage, respectively. The quantities for
the rotor side and stator side are denoted respectively by subscripts ‘r’ and ‘s’; whereas the
quantities transformed to the rotating d-axis and q-axis are represented by subscripts ‘d’ and ‘q’,
respectively. Thus, we have the general expression
x  xd  jxq ,

for x  ψ s , ψ r , i r , i s , v s , v r ; where x   s , r , ir , is , vs , vr and j  1.

ψs
ψr

j (e  r )ψ r
je ψ s

Fig. 3-8 Equivalent circuit for the DFIG in the synchronously rotating reference frame.

The stator and rotor flux state vectors in Fig. 3-8, can be written as
ψ s  Ls i s  Lm i r 
, (3.15)
ψ r  Lm i s  Lr i r 

where
Ls  Lls  Lm, Lr  Llr  Lm . (3.16)

37
Here, L is the inductance; r is the resistance; subscripts s and r are used for denoting the stator or
rotor quantity, respectively. Similarly, Lm represents the mutual inductance of the machine. The
machine has fifth order dynamic equations, in which the first four are derived by applying the
Kirchhoff’s voltage law and the last from equation of motion, as
dψ s / dt  v s  rs i s  je ψ s 

dψ r / dt  v r  rr i r  j (e  r )ψ r  , (3.17)
d r / dt  (Tm  Tem ) / J m  Dm / J m 

where
Tem   0.75 n p Re{ψ s i *s }. (3.18)

Here, Tem is the electromagnetic torque, ωr is the rotor angular speed, Jm is the combined moment
of inertia of the turbine and the generator mass, and np is the number of pole pairs. Similarly, the
expression of active (Pig) and reactive power (Qig) is evaluated as
Pig  3 Re[ v s i *s ] / 2 
. (3.19)
Qig  3 Im[ v s i *s ] / 2 

In the case of cage induction generators, the expressions (3.15) to (3.19) would be the same,
except that the value of rotor voltage vector in (3.17) is zero. For the purpose of analysis, the
equivalent stator magnetizing current vector (imo) is defined so that
ψ s  Ls i s  Lmi r  Lmi mo . (3.20)
That gives,
i mo  Ls i s / Lm  i r 
. (3.21)
i s  Lm (i mo  i r ) / Ls 

Further, from (3.15) and (3.21), we have

ψ s  L2m i mo / Ls   Lr i r (3.22)

where   1  L2m / Ls Lr . Substituting ψ s in (3.20), is in (3.21) and ψ r in (3.15) into (3.17), yields
the stator and rotor voltages in the synchronous d-q reference frame as
v s  rs i s  Lm di mo / dt  j e ψ s 
. (3.23)
v r  rr i r   Lr di r / dt  ( L2m / Ls ) di mo / dt  j ( e   r ) ψ r 

38
 sa sb sc

(a)
vr,an
ira

d1aVdc Isto
I2,dc

vr,bn
irb
d1bVdc Cdc Rdc

vr,cn
irc Ir,dc=d1circ+d1birb+d1aira
d1cVdc

(b)
Fig. 3-9 Models for voltage source converter. (a) Switching model, and (b) averaged model. The notation
‘′’ with the switching function represents the complimentary value, such that si  si  1 (i  a, b, c) . Here,
Ir,dc, Isto and I2,dc represent the dc current injected from VSC1, from the storage system and from VSC2,
respectively.

b) Rotor-side converter model

The model of voltage source converter consists of 3-phase transistor switching circuits
operated through the PWM principle as detailed in Fig. 3-9(a). The relationship between the
switching functions (sa,sb and sc), and the terminal ac and dc voltages can be written as [11],[12]

39
  vr ,an   2 1 1  sa 
  Vdc   
 vr ,bn    1 2 1  sb  ,
(3.24)
  3  1 1 2   sc 
v
 r ,cn 
where Vdc is the dc link voltage, vr,in and si (i=a,b,c) represents the ith phase to neutral point rotor
voltage and switching function of the ith switch, respectively. The complexity of a converter
model increases with the desired level of accuracy. While certain applications demand the most
precise model available, other application are satisfactory with a simplified averaged model. In
case of a large system having many converters, the exact model is not feasible due to complexity
of analysis and constraints of simulation time and data storage. In such cases, the fundamental
frequency model is sufficient assuming that the high-frequency harmonics produced by the
inverter as a result of the PWM control techniques are mostly filtered by a sinusoidal line filter.
Therefore, the averaged modeling approach is utilized in this study to manage the simulation
time and storage requirements. This makes the ac side of VSC1 a sinusoidal voltage source and
dc side a current source, as represented by the following equations [13]

[vr ,an vr ,bn vr ,cn ]t  [d1a d1b d1c ]t Vdc 

 (3.25)
I r ,dc  [ir ,an ir ,bn ir ,cn ][d1a d1b d1c ]t 

where,

Tsw  2 1 1  sa 
1  1 2 1  s  dt ,
d1c ]  
t
[d1a d1b   b
Tsw 0
 1 1 2   sc 

and iri (i=a,b,c) represents the ith phase current from the rotor circuit. Fig. 3-9 (b) shows the
schematic diagram of the averaged model. Noting that d1  d1d  jd1q , the expression of rotor

voltage dynamics can be written as

d 1Vdc  rr i r   Lr d i r / dt  ( L2m / Ls ) d i mo / dt  j ( e   r ) ψ r . (3.26)
This equation, together with (3.15) to (3.22), are used to develop the control scheme of
VSC1 discussed in the next chapter.

c) Line-side converter, filter and ac bus

Applying the same treatment as with VSC1, the time averaged fundamental frequency
model of VSC2, the line filter and the grid voltage can be expressed by the following set of
second order dynamic equations,

40
 e 2  R f i 2  L f di 2 / dt  je i 2 L f  v 2 

v 2  Rth i 2  Lth di 2 / dt  v th 
, (3.27)
e 2  d 2 Vdc 
d 2  d 2 d  jd 2 q 

with e 2  ed  jeq , d 2  d 2 d  jd 2 q , i 2  i2 d  ji2 q , v 2  v2 d  jv2 q and v th  vthd  jvthq .

Here, d2 , e2 , v2 , i 2 ,and vth represents vector quantities for the duty cycle vector of VSC2, the

inverter terminal voltage, the voltage at bus #2, the current injected to bus #2, and the Thevenin’s
voltage of outer power network seen from bus #2, respectively. Similarly, R f , L f , Rth and Lth are

the interfacing resistance and interfacing inductance, the Thevenin resistance of external network
and Thevenin inductance of the external network, respectively (see Fig. 2-6 (a)). This equation
will be used to develop the control scheme for VSC2.

d) dc-link voltage
Considering the power conservation principle in the proposed VSC1 model, we have
1.5(ird vrd  irq vrq )  I r ,dcVdc 
. (3.28)
1.5(i2 d v2 d  i2 q v2 q )  I 2,dcVdc 

Applying, KCL at the dc link node, we get

dVdc V
Cdc  I r , dc  I sto  I 2,dc  dc . (3.29)
dt Rdc
The variables and the parameters in above equations are defined in Fig. 3-9.

e) Summary of doubly-fed induction generator model

From the discussion in section 3.2.3, the dynamic model of DFIG can be summarized by
the following set of equations.

41
 di mo 
Lm  v s  rs i s  je ψ s 
dt

di r Lm 
 Lr  v r  rr i r  ( v s  rs i s  je ψ s )  j (e  r )ψ r 
dt Ls

d r 
Jm  (Tm  Tem )  Dmr L f 
dt  (3.30)
dV V 
Cdc dc  I r ,dc  I sto  I1,dc  dc 
dt rc

di 2 
 e 2  rf i 2  je i 2  v 2 
dt 

where,

ψ s  L2m i mo / Ls   Lr i r 
ψ r  Lm i s  Lr i r 

i s  Lm (i mo  i r ) / Ls 

ir ,dc  1.5(irdVrd  irqVrq ) / Vdc  .
 (3.31)
i1,dc  1.5(i2 dV1d  i2 q v2 q ) / Vdc 
Tem  0.75n p Re{ψ s i*s } 

3 2 
Tm  0.5CP (tip ,  )  ArT r 

3.2.4 Model analysis

a) Voltage source converter
To test the model of the voltage source converter, the ac terminal of the VSC is connected
to the 0.8 pu resistive load through filter, and 750V dc source is connected to the dc side. The
PWM technique is employed for the switching of the transister/diode model built-up in
MATLAB/Simpower toolbox. Likewise, the switching model is replaced by the average model
of VSC with the identical value of filter impedances. Fig. 3-10(a) illustrates the current and the
voltage response against the load with the switching model and Fig. 3-10(b) depicts those of the
averaged model. It is noted that though there are harmonics in voltage and current from the
switched model, its output voltage and current waveform are close to those of the averaged
model when a properly designed filter is applied to the system. Nevertheless, the averaged model
is capable of simulating a VSC with good accuracy for large time scale simulation and analysis
purposes.

42
 (a)

(b)

Fig. 3-110 Voltage annd current ressponse curvees of the VSC

C. (a) Switching model, and (b) averagged model

43
 Fig. 3-11 DFIIG startup proocess at no-looad.

b) Dooubly-fed in
nduction geenerator
Foor the analyysis process,, the inductiion generatoor (IG) is coonfigured with
w parameeters given
in the apppendix; thee rotor is coonnected to dc bus via an ideal invverter; and tthen the volltage build
up proccess is obseerved at thee stator term
minal with no-load conndition. Figg. 3-11 show
ws the ac
voltage and the speed build-uup process. Initially, thhe machine draws large power froom the dc
source aand it starts as a motor. After a shhort transiennt, it reachess to the steaady-state conndition, at
which thhe rotor speeed and the ac voltage settle at theeir rated values becausee the electriical torque
These resultts are simiilar to those obtained from the
equals tto the mecchanical couunterpart. T
dynamicc model of tthe IG reporrted in the standard textts [10].

3.3 Dedicated E
Energy Sttorage Syystem
Ass seen in the previous cchapter, thee storage system consissts of the FC
C/ELZ and UC units;
in whichh not only the electrocchemical phhenomena, bbut also thee other proccess, such as thermal,
chemicaal, fluid dynamic, etc., are also ppresent. If all these phhenomena should
s be ttaken into
accountt at a time, the modelling processs would bee too compllex and tim
me consuminng. When

44
applying the modeling in engineering processes, different approaches can be considered,
depending on the purpose of the study. For example, fluid dynamic, thermal, chemical and
electrochemical approaches may suitable for the development stage of electrochemical system,
especially for the wider time frame analysis. But for the application stage, in which the
electrochemical system has to interact with the load to which it is connected, an electric
modeling of the system could be more desirable. Therefore, in this chapter, we will focus on
those models more suitable to effective integration in simulation tools of electrical network.

3.3.1 Fuel cell

An SOFC cell consists of yttria-stabilized zirconia electrolyte, onto which a porous cathode
(lanthanum manganite) and a porous anode (Ni/ZrO2-cermet) are deposited (see Fig. 3-12). At
the cathode-electrolyte interface, oxygen molecules accept electrons coming from the external
circuit to form oxide ions; the electrolyte layer allows only oxide ions to pass through; and at the
anode-electrolyte interface, hydrogen molecules react with oxide ions to form steam and
electrons get released. These electrons pass through the external circuit and reach the cathode–
electrolyte layer, and thus the circuit is closed [14]. These FCs are characterized by their
electrolyte material, the solid oxide or ceramic electrolyte. There are various models of SOFC
depending upon the scope of interest. Some are the highly theoretical and are based on empirical
equations [15],[16], and other are more application oriented [17]. This study utilizes the detailed
dynamic model which considers the diffusion phenomena of reactant gases and the double layer
charging effect [14],[18]. The assumptions in modeling are as follows.
 One-dimensional treatment of reactants flow.
 H2 is directly available from the tank, so fuel combustion is not included.
 Large stoichiometric quantity of H2 at anode and O2 at cathode.
 The combustion zone is not included in the SOFC thermal model. The fuel and air are
assumed to be pre-heated.
 The gases are ideal.
 Parameters for individual cells can be lumped together to represent a fuel cell stack.
In the following paragraphs modeling of different components of the fuel cell is presented.

45
 O 2 +4e-  2O --
2H 2 +2O--  H 2 O+4e-

Fig. 3-12 Schematic diagram of SOFC.

a) Partial pressure dynamics

Partial pressures of H2, H2O and O2 at anode and cathode channels are deducted in terms of
the fuel cell operating parameters (e.g., fuel cell temperature, fuel and oxidant flow rates) by
using material conservation equations. The resultant partial pressure values are, then, used to
calculate the fuel cell output voltage.
The dynamics of the partial pressures of hydrogen and water vapor in the anode gas flow
channel can be determined through the ideal gas equations as follows [18],[19].

Ua d I 
pH 2  M Hin2  M Hout2  fc 
RT fc dt 2F 
I 
Ua d 
pH 2O  M Hin2O  M Hout2O  fc  , (3.32)
RT fc dt 2F 
Uc d I 
pO2  M Oin2  M Oout2  fc 
RT fc dt 4F 
where the Ua and Uc are the volume of the anode and cathode side, respectively; pi is the partial
pressure of the ith species (i=H2, H2O, O2); Mi is the mass flow rate of ith species; F is the
Faraday constant; Ifc is fuel cell current; and R is the gas constant.

46
 The values of inlet and outlet molar flows of the gases in (3.32) are given by,
pHin2 
M in
H2 M x in
a H2  Ma 
Pa 
pH out 
M Hout2  M a xHout2  M a 2 
Pa 

pHin2O 
M H 2 O  M a xH 2O  M a
in in

Pa 
. (3.33)
pHout2O 
M H 2 O  M a xH 2O  M a
out out

Pa 

pO
in

M Oin2  M c xOin2  M c 2 
Pc 
pO out 
M Oout2  M c xOout2  M c 2 
Pc 

Here, Ma and Mc represent the total molar concentration at the anode and cathode, respectively,
and x represents the fraction of flow quantity as specified by the subscripts. As a result, molar
inflow (outflow) rate becomes proportional to the ratio of inflow (outflow) partial pressure and
net gas pressure inside the electrodes. Assuming that the partial pressure of the species in the
electrodes as an average of input and output pressures, i.e., pH 2  ( pHin2  pHout2 ) / 2 , (3.32) can be

rewritten as

d 2 RT fc in 2M a RT fc RT fc 
pH 2  M H2  pH 2  I fc 
dt Ua PaU a 2 FU a 
d 2 RT fc in 2M a RT fc RT fc 
pH 2 O  M H 2O  pH 2 O  I fc  . (3.34)
dt Ua PU
a a 2 FU a 
d 2 RT fc in 2M c RT fc RT fc 
pO2  M O2  pO2  I fc 
dt Uc PU
c c 4 FU c 
Note that the dynamic equation has effective time constants,  a  U a Pa / 2 M a RT fc and

 c  U c Pc / 2 M c RT fc . They are variable quantities that depend on the molar flow rates of the

species at the anode or cathode.

b) Fuel cell output voltage

At the cathode, oxygen from air is converted to oxygen ions which move through the
electrolyte membrane and reacts with the fuel at the anode-electrolyte interface, this process

47
results the open circuit potential due to change in the Gibbs energy. The phenomenon of
generation of electro-chemical potential of an SOFC is shown in Fig. 3-12. The chemical
reaction taking place at the electrodes is written as,

Anode : H2 +O2-  H2O+2e- 

.
Cathode :1 2 O2 +2e-  O2- 
Based upon the chemical reactions at the cathode and the anode, the voltage generated in a
cell (Ecell) can be written by the Nernst equation as [18], [20]

RT fc  pH pO 
Ecell  E0,cell  k E (T fc  298)  ln  2 2
, (3.35)
2F  p H 2O 

where E0,cell and kE represents the Gibbs potential and empirical constant, respectively. Ecell
calculated from (3.35) is the open-circuit voltage of the fuel cell. However, when the fuel cell is
under a loading condition, the output voltage is less than Ecell due to activation loss, ohmic
voltage drop and concentration over-potential. Consequently, the output voltage of a cell can be
written as
Vcell  E cell  Vact , cell  Vohm , cell  Vcon , cell , (3.36)

where Vact ,cell , Vohm , cell and Vcon , cell denote activation voltage drop, ohmic voltage drop and

concentration voltage drop, respectively. Finally, the net output voltage (Vfc) and power flow
(Pfc) from the fuel cell stack can be obtained as,

V fc  N cellVcell , Pfc  V fc I fc , (3.37)

where Ncell is the total number of cell in a stack. The voltage drop terms in (3.36) is determined
as follows.
i) Activation Voltage Drop
Activation voltage loss represents an activation energy barrier that must be overcome
before the chemical reaction occurs. During the process, a proportion of the voltage generated is
lost in driving the chemical reaction that transfers the electrons to or from the electrode. A
general equation for activation loss, for all working value of current, can be deduced from
Butler–Volmer equation as [20],
2 RT fc
sinh 1  I fc / 2i0  , io  ac e
 ( bc / RT fc )
Vact ,cell  , (3.38)
zF

48
where ac and bc are the empirical parameters, and z is the number of electrons participating in the
reaction. The current density i0 is known as the exchange current density, which depends on the
temperature. According to (3.38), the activation voltage drop will be zero when the load current
is zero. The ohmic and concentration voltage drops (to be discussed in the following
subsections) are also zero when the fuel cell is not loaded (Ifc=0). However, even the open-circuit
voltage of an SOFC is known to be less than the theoretical value given by (3.35) [13]. To
include the effect of no-load activation voltage drop, a constant and a temperature-dependent
term can also be added to (3.38) [17]
2 RT fc
Vact ,cell   0  1T fc  sinh 1  I fc / 2i0  (3.39)
  
zF ,
Vact 0,cell
Vact 1,cell

where 0 is the constant term of the activation voltage drop and 1 is the temperature coefficient of
the activation voltage drop. In the above expression, Vact 0,cell is the part of activation drop affected
only by the fuel cell internal temperature, while Vact1,cell depends both on current and temperature.

ii) Concentration loss

This results from the change in concentration of the reactants at the surface of the
electrodes when the fuel is used. Since the concentration affects the voltage, this type of
irreversibility is sometimes called concentration loss or mass transport loss. The expression for
the concentration loss is derived by [14]
RT fc
Vcon,cell   ln(1  I fc / Ilimit ). (3.40)
zF
Here, Ilimit is the limiting current density at which the fuel is consumed at the maximum supply
rate. The current density cannot rise above this value, because the fuel gas cannot be supplied at
a greater rate. At this current density the pressure would have just reached zero.
iii) Ohmic voltage drop
The ohmic voltage drop results from the resistance to the electrons in the interconnection of
collecting plates and carbon electrodes, and the resistance to the oxygen ions in solid membrane.
The total voltage drop due to resistive effect can be represented as [18], [21]
Vohm , cell  Vohm , elecyt  Vohm ,interc  I fc Rohm , cell , (3.41)

where Vohm,elecyt is the voltage drop in electrolyte medium, Vohm,interc is the voltage drop in
interconnection, and Rohm is the equivalent resistance of the fuel cell. The resistance, Rohm normally

49
decreases as temperature increases. It can be expressed as
 elecyt e
( belecyt / T fc )
 interc e
( binterc / T fc )

Rohm,cell   elecyt   interc , (3.42)

Acell Acell
where αis and bis (i=elecyt, interc) are the constant parameters, δis are the resistivity of the
material, and Acell is the cross-section area of the fuel cell.
iv) Double layer effect
The phenomenon known as charge double layer exists in an SOFC because two electrodes
are separated by the electrolyte (see Fig. 3-12) and two boundary layers are formed (anode-
electrolyte layer and electrolyte-cathode layer). The charge layer acts as the storage of electrical
charge behaving like an electrical capacitor. This causes a first order delay in charging and
discharging. Such delay affects the activation and concentration potentials. However, the ohmic
over-potential is not affected, since it is related linearly to the cell current through Ohm’s law.
The model for the SOFC considering this effect can be described by the equivalent circuit shown
in Fig. 3-13.
The parameters Rohm,cell, Ract,cell and Rconc,cell in the figure are equivalent resistances of
ohmic voltage drop, activation and concentration voltage drops, which can be calculated
according to (3.42), (3.38) and (3.40), respectively. Cfc represents the equivalent capacitance of
the double-layer charging effect. Since the electrodes of the SOFC are porous, the value of Cfc is
large and can be in the order of several farad [14]. The voltage across Cfc can be expressed as

Vdl   I fc  C fc dVdl / dt  ( Ract ,cell  Rcon,cell ) . (3.43)

Based upon the aforementioned discussion, the V-I curve for the SOFC stack is presented in Fig.
3-14 depicting the three regions. Voltage drop across the FC associated with low currents is due
to the activation loss inside the SOFC, voltage drop in the middle of the curve (approximately
linear) is due to the ohmic loss in the FC stack, and voltage drop at the end of the curve is due to
the concentration loss.

50
 Roohm,cell Ifc

Rcon,cell

Vdl
Cfc
Ract,cell Vfc

E0=E0,cell -V
Vact0,cell

Fig. 3-113 Equivalennt electrical m

model of SOF
FC. The term
m Vact0,cell is paart of Vact,cell w
which is affeected only
by thee fuel cell intternal temperrature (not byy current so is
i added withh E0,cell).

Fig. 33-14 Static V-

V-I and P-I ressponse of thee SOFC.

51
c) Fuel processing unit
Considering the pure hydrogen is stored at the storage tank, a simple model of a fuel
regulator operated in constant fuel utilization mode can be written by the first order equation in
the Laplace transformed model as [18]

1  1 
M Hin2   I , (3.44)
 f F  sTd  1  fc
where  f is the utilization factor and Td is the time constant of the fuel regulator. Equation

(3.44) shows that inflow rate of fuel is proportional to the current followed by the system delay.
In case of oxygen, excess oxygen is provided to allow it completely react with hydrogen and
maintain the pressure difference between the anode and the cathode below a certain threshold
value. Oxygen flow rate is, then determined using the hydrogen-oxygen flow ratio, rH O . 2

3.3.2 Model analysis of SOFC

To perform the model verification, a 5 kW SOFC model is developed using parameters
given in the appendix. The dynamic and the static response of the fuel cell given in [17] are
compared with the model as developed in the above section. The static characteristics of [17] are
approximately obtained from the given V-I curve for the 5 kW model. To deduce the voltage
response, the stack current is increased from 0 to 160 A gradually. The response of the output
voltage from the simulink model and the sampled values obtained from [17] (mark by ‘×’) are
plotted in Fig. 3-15 (a). It can be seen that developed model agrees with the model as suggested
in [17]. The voltage drop at the beginning and end of the curves showed in Fig. 3-15(a) are due
to activation and concentration losses, respectively. The voltage drop in the middle of the curves
(approximately linear) is due to the ohmic loss in the fuel cell stack.
Similarly, Fig. 3-15 (b) compares the dynamic responses between these models. To check
the voltage response, the current waveform used in [17] is subjected to the developed model. The
dynamic property the SOFC depends mainly on the double-layer charging effects, delays in fuel
and oxidant flow and fuel regulation dynamics. In this case, the fuel regulation block is not
included for the comparison with the standard model. Although the capacitance (Cfc) due to the
double-layer charging effect is large (in the order of several farads for each cell), the time
constant C fc ( Rcon , cell  Ract , cell ) is normally small (less than 1 s) because of the small value of net

resistance ( Rcon , cell  Ract , cell ) . Therefore, Cfc will only affect the transient response of the SOFC in

52
the very short time range. On the other hand, the fuel and oxidant flows cannot follow the load
current changes instantaneously and the delays can be in the range to several tens of seconds. Fig.
3-15 (b) also illustrates that the developed model agrees with the model of [17]. It should,
however, be noted that the voltage response will delay further when included the fuel regulation
dynamics.

(a)

(b)
Fig. 3-15 Model validation of the SOFC. (a) Static and (b) dynamic model.

53
3.3.3 Electrolyzer
ELZs are the electrochemical devices which produce hydrogen and oxygen by consuming
electricity and water, and broadly characterized as the opposite of the fuel cell system [22]. The
modeling for the alkaline electrolyzer consists of the electro-chemical part, electrical part,
thermal part and hydraulic part. In this study, the thermal modeling is excluded and constant
temperature mode is adopted assuming the large time constant (around several minutes) of the
thermal model [22],[23].
The electrochemical reaction taking place in the alkaline ELZ with the application of
electricity is written as

Anode : 2OH - (aq)  1 2 O2 (g )+H 2O+2e- 


Cathode :2H 2O(l )+2e-  H 2 (g )+2OH - (aq)  .
l  liquid , g  gas, aq  aqueous 

According to the Faraday law, the hydrogen molar production rate can be written as
N elz
M H 2 , pro   (Telz , J elz ) I elz . (3.45)
2F
Here Jelz, Ielz, Nelz, Telz, and η are current density, current, number of cell in the stack, temperature
of the ELZ and the Faraday efficiency, respectively. Equation (3.45) shows that molar
production of H2 is proportional to current, and it is modulated by the Faraday efficiency which
is defined as the ratio between the actual and the theoretical maximum amount of hydrogen
produced in the ELZ. The Faraday efficiency is often called the current efficiency, because it is
caused by parasitic current losses along the gas ducts. The parasitic currents increase with
decreasing current densities due to an increasing share of electrolyte and result a lower electrical
resistance. Further, an increase in temperature leads to the increase parasitic currents losses,
thereby lowering the Faraday efficiencies. An empirical expression for Faraday efficiency as the
function of current density (Jelz) and the temperature is given by [22]
2
J elz 
 f2 
f1  J elz
2
. (3.46)
with, f1  50  2.5Telz , f 2  1  0.00075Telz 

The electrical model of the ELZ represents the cell output voltage as a function of input
current. It is often derived from the empirical parameters deducted from the experiments. The V-I
relationship of the ELZ is written as

54
 r1  r2Telz t t /T t /T2 
ucell  u0  I elz  u1 log  1 2 elz 3 elz I elz  1 . (3.47)
Aelz  Aelz 
Here ucell , u0 , u1 ,and Aelz denote the terminal voltage, thermodynamic cell voltage, overvoltage

parameter and electrode surface area, respectively. The parameters ris and tis represent the
dependency of ohmic resistance and overvoltage coefficient on temperature. The terminal
voltage across the ELZ cell is the sum of three components: thermodynamic cell voltage which is
the function of temperature and pressure, the ELZ overvoltage, and ohmic drop. Based on the
parameters values (given in the appendix), static I-V curve is plotted (see Fig. 3-16). Clearly, the
ELZ can be conceived as a voltage source with an internal resistance; both are nonlinear and
sensitive to the temperature and current flow through it.
Since a unit cell has a low value of output potential, a stack is build embedding a number of
such units in series, giving the net terminal voltage (Velz) as
Velz  Nelz ucell . (3.48)

Fig. 3-16 Static V-I characteristics curve of the alkaline ELZ.

For a given temperature, an increase in hydrogen production (i.e., an increase in current

density) increases the cell voltage (Fig. 3-16), which consequently decreases the energy
efficiency. For a given current density, the energy efficiency increases with increasing cell
temperature. In order to calculate the overall performance of an ELZ system, information about
the number of cells in series and/or parallel per stack and the number of stacks per unit is needed.
The rated voltage of an ELZ stack is found from the number of cells in series, while the number
of cells in parallel yields the rated current (and thus H2 production). The total power is simply the

55
product of the current and voltage. While developing the stack of the rated value, it is considered
that the no-load voltages of the SOFC and ELZ are equal. In other words making NcellEcell=Nelzu0,
there would be a continuous static I-V curve for the SOFC/ELZ unit leading to smooth transition
between the SOFC to ELZ operation modes.
The hydraulic part of an ELZ consists of the dynamics of the resultant gases. If we ignore
the H2 leakage rate, the accumulating rate of the H2 inside the ELZ is difference between the
production and outlet rate. Based upon the ideal gases law, the resultant pressure dynamics can
be written as
U elz d
pH ,elz  M H 2 , pro  M H 2 ,out , (3.49)
RTelz dt 2
where p H 2 , elz
is the partial pressure of hydrogen inside the ELZ, Uelz is the volume of electrolyzer

and M H 2 , out
is the outflow rate of hydrogen from the ELZ. At steady state, the outflow rate is

maintained equal to the production rate to make the pressure constant.

3.3.4 Compressor and tank model

According to the polytropic compression model, the relationship between the hydrogen
molar flow rate and the compressor power is [23]
 com 
M H 2 ,out  Pcomp 
w
 1

   (3.50)
 T  p  
w  elz  tank   1
  1  pH ,elz 
 
   .
2

Equation (3.50) shows outflow of H2 from the ELZ is decided by the compressor power (Pcomp),
the polytropic work (w), polytropic coefficient (  ) and the compression efficiency (αcom).
Polytropic work is derived from the thermodynamic law, by which H2 is supplied to the higher
pressure (ptank). The stored hydrogen rate in the tank depends upon the difference between the
inflow rate and the outflow rate. Thus, the pressure of stored hydrogen in the tank can be written
as
U tank d
ptank  M H 2 ,out  M Hin2 , (3.51)
RTtank dt
where Utank and Ttank represent the volume and temperature of storage tank, respectively. Based

56
upon the above equations, the block diagram of hydrogen production and storage is constructed
and shown in Fig. 3-17.

N elz RTelz M H 2 , out

 1
2F U elz
s

1 RTelz
M Hin2
s U elz
Fig. 3-17 Block diagram of hydrogen production and storage model.

(a)

Idc1

(b)
Fig. 3-18 Electrical diagram of the bi-directional dc-dc chopper. (a) Circuit diagram, and (b) switching
logic. The variable swi are the gate signals for the transistor Ti (i=1,2,3).

57
3.3.5 dc-dc converters
Proposed DESS employs two dc-dc converters, the FC/ELZ side converter (DC1) and
isolating converter (DC2). While the converter DC1 is responsible for the regulation of the
FC/ELZ current and ultracapacitor terminal voltage, the converter DC2 is responsible for the dc-
link bus regulation. In other words, DC1 and DC2 are operated in current and voltage regulation
modes, respectively.
The model for the chopper (DC1) employed to regulate the current through the FC/ELZ
and Vuc is illustrated in Fig. 3-18(a). This model resembles the classical two-quadrant dc-dc
chopper with minor modifications in switching elements to accommodate the FC and ELZ
control action by the same converter. The working principle of DC1 is easily understood from
the switching pattern depicted in Fig. 3-18(b). The direction Idc1 determines the operating modes
of DC1‒ buck or boost mode. During the power deficit (for Idc1>0), the duty cycle (d3) generates
sw3 to turn T3 through the PWM action activating the FC, and then turns T1/T2 OFF (with
sw2=sw1=0) at the same time. Similarly during power surplus (for Idc1<0), the duty cycle
generates sw2/sw1 to turn T2/T1 in complementary way and activates ELZ, while turning T3 OFF
(with sw3=0) at the same time. By forcing the desired value of Idc1 (hence the current flowing
through the FC/ELZ), the control action provides direct control over the UC voltage without
using a separate dc-dc interface. With d2 (d3) as the duty cycle of T2 (T3) and noting that d2 (d3)
is activated when Idc1<0 (Idc1>0), the averaged equations for voltage and current can be written
as
I fc  d 3 I dc1 , I elz   d 2 I dc1 
. (3.52)
V pwm  d3V fc  d 2Velz 
From Kirchhoff’s voltage law, we can write
dI dc1
Li  I dc1 Ri  V pwm  Vuc . (3.53)
dt
For the regulation of Vdc, a full-bridge converter (DC2) is implemented because of its (i)
high power capability, (ii) zero voltage/current switching provision for loss reduction, and (iii)
isolation of FC/UC/ELZ system from the rest of the electrical network. The switching model of
DC2 is shown in Fig. 3-19(a). For the simplicity of analysis and managing the simulation
time/storage, the average model is used here, which can be deducted with following set of
equations [24],[25]

58
 VucVdc  2  
P    , Vuc I i  Vdc I dc  ; (3.54)
2 f s Lt    

Vdc  2  Vuc   2  
Ii      , I dc      ; (3.55)
2 f s Lt    2 f s Lt    

where δ is the phase shift angle, and the other parameters/variables in the above equations are
defined in Fig. 3-19(a). From (3.54) and(3.55), it can be observed that the phase shift variable
determines the magnitude and direction of the current. The positive value of δ indicates the
power deficit (current supplying to dc bus), and the negative value indicates the power surplus
(current drawing from the dc-link bus). The approximated averaged model is depicted in Fig.
3-19(b).

(a)

(b)

Fig. 3-19 Electrical diagram of the full bridge dc-dc converter. (a) Switching model, and (b) averaged
model on the basis of (3.55).

59
3.3.6 Ultracapacitor
This study has used the classical model of a UC (see Fig. 3-20), which consists of an
electrical double layer capacitance (Cuc), an equivalent series resistance (Rsr) and a parallel
resistance (Rpl). Rsr represents the charging and discharging resistance, whereas Rpl models the
leakage effects applicable in the long-term energy storage performance of the UC [26],[27]. The
dynamic equation representing the Kirchhoff’s current law is written as
dVuc Vuc
Cuc   I dc1  I i . (3.56)
dt R pl

Unlike battery system, a UC allows a wide range of voltage variation to leverage the
maximum energy exchange. The amount of energy consumed/drawn from the UC bank (Euc) in
extreme condition is deducted, with the specified values of the initial and final value of the UC
voltages, as
1
Euc  Cuc Vi 2  V f2  . (3.57)
2
When the energy is released from the UC, the magnitude of the voltage is decreased and
vice versa. While sizing the UC, we consider the amount of energy that UC can exchange on the
extreme conditions. This, in turn, is decided by the transient behavior of the FC and the loading
conditions. Since the FC has poor transient response, the UC should have a large value, sufficient
enough, to fulfill the transient demand (by remaining within the allowable range of terminal
voltage), until the FC provide the rated steady-state current. To match the desired voltage level
and the capacitance, numbers of UC units are assembled in the series and parallel fashion, based
upon the data sheet of Maxwell BMOD0110® [28].

Fig. 3-20 Electrical diagram of UC.

60
 Multiplication point
Summing point FC/ELZ/UC/DC1
system

1
Ii 2 f sw Lt
Idstb

-
d2,d3 Ife FC/ Vfe Vpwm 1 I dc1 - R pl Rsr Cuc s  Rsr  R pl Vuc Isto 1
Vdc δ
ELZ + s Li  R i Iuc R pl Cuc s  1 sC dc
+ +
-
1

61
2 f sw Lt

+ 1 M H 2 ,out
Ielz N RTelz DC2/dc-link bus
Demux elz Eq (3.44)
2F s Velz system
-

+
Ifc Fuel M Hin2
-
ptank
1 RTtank
Regulation
s Vtank
Block

H2 production and storage

system

Fig. 3-21 Overall control block diagram of the dedicated energy storage system.
3.3.7 Summary of dynamics in the DESS
The dynamic equations for DESS are summarized by the following set of equations.

d 2 RT fc in 2 M a RT fc RT fc 
pH 2  M H2  pH 2  I fc 
dt Ua PaU a 2 FU a 
d 2 RT fc in 2 M a RT fc RT fc 
p H 2O  M H 2O  p H 2O  I fc 
dt Ua PaU a 2 FU a 
d 2 RT fc in 2 M c RT fc RT fc 
pO2  M O2  pO2  I fc 
dt Uc PU 4 FU c 
c c
 (3.58)
Vdl   I fc  C fc dVdl / dt  ( Ract ,cell  Rcon ,cell ) 

dM Hin2 1 
Td   M Hin2  I fc 
dt f F 
in 
dM O2 rH 2O 
Td   M O2 
in
I fc
dt f F 

dI dc1 
Li   I dc1 Ri  Vuc  d3V fc  d 2Velz 
dt 
dV V 
Cuc uc  I dc1  uc  I i  (3.59)
dt Rp 

U elz d 
pH 2 ,elz  M H 2 , pro  M H 2 ,out 
RTelz dt 
 (3.60)
U tank d 
ptank  M H 2 ,out  M H 2in

RTtank dt 

The overall control block diagram of DESS is depicted in Fig. 3-21. It consists of three
units; FC/ELZ/UC/DC1 system, DC2/dc-link bus system, and H2 production and storage system.
The FC/ELZ/UC/DC1 system has two inputs (duty cycle as command reference and Ii as the
disturbance input) and two outputs (Ifc and Vuc); DC2/dc-link bus system has three input variables
(δ, Vuc and Idstb) and an output variable (Ii); and H2 production and storage system. The first and
second block will be used to design the control schemes for DC1 and DC2, respectively; while
the last one is the feed forward block reflecting the fuel production/consumption.
While observing the set of differential equations, it can be noticed that (3.60) only appears
at the output part and does not include the feedback loop, so it has no role in the system stability.
The remaining state transition matrix is almost diagonal except those related to (3.59) where

62
cross coupling exists. Here, the equilibrium point of the system is at origin as the stable node.
That means with no inputs, all the states finally converse into the origin irrespective of the initial
conditions. Further, the system has time constants of different time scales. Time constant for the
reactants’ partial pressure, fuel processing in (3.58), have larger values of around several seconds
dominating the system dynamics; while the time constants for Idc1 and Vdl is around few
milliseconds. The presence of the sluggish dynamics in the FC, represents the attribute of poor
power density; that means its inability for transient mitigation. Therefore, an energy storage
device with high power density is required, and the UC fulfils this demand. These characteristics,
together with the block diagram in Fig. 3-21, will be in focus while designing the control laws
for DC1 and DC2 to be explained in the next chapter.

3.4 Remarks on System Modeling

In the system modeling explained in this chapter, we observed that dynamic equations are
mainly the linear types under various assumptions, which are selected to investigate the
fundamental aspects of dynamic behavior within the frame work of the research objectives. It
applies to the WTGS, DESS and VSC models. These models may be too optimistic when we try
to study the system responses under the broader operating conditions and system severities. In
such cases, adoption of more comprehensive models with in-depth nonlinear dynamics may be
suggested, which can provide a larger picture on the equilibrium points and the stability regions.
In that context, the component models reported here may not be enough; thereby demanding
more work, especially, on the element level research. Nevertheless, in the integrated system level
study with given operational objectives of the proposed microgrid, the models adopted here are
expected to expose system responses without losing the accuracy level.

References
[1] G. Johnson, Wind energy systems: Prentice-Hall Englewood Cliffs (NJ), 1985.
[2] T. Thiringer and J. Linders, "Control by variable rotor speed of a fixed-pitch wind turbine
operating in a wide speed range," IEEE Transactions on Energy Conversion, vol. 8(3), pp.
520-526, 1993.
[3] Horns rev offshore wind farm, available online at:
http://www.hornsrev.dk/nyheder/brochurer/Horns_Rev_GB.pdf.
[4] T. Ackermann, Wind power in power systems, John Wiley and Sons Ltd, 2006.

63
[5] P. Anderson and A. Bose, "Stability simulation of wind turbine systems," IEEE
Transactions on Power Apparatus and Systems, Vol. PAS-102(12), pp. 3791-3795, 1983.
[6] Danish Wind Energy Association, Guided Tour, available online:
http://guidedtour.windpower.org/en/tour/wtrb/stall.htm
[7] T. Burton, D. Sharpe, N. Jenkins and E. Bossanyi, Wind Energy Handbook, John Wiley
and Sons Ltd, 2004.
[8] F. Bianchi, et al., Wind Turbine Control Systems: Principles, Modelling and Gain-
scheduling Design, Springer, 2006.
[9] J. Hu and Y. He, "Dynamic modelling and robust current control of wind-turbine driven
DFIG during external AC voltage dip," Journal of Zhejiang University Science, vol. 7(10),
pp. 1757-1764, 2006.
[10] C. Ong, Dynamic simulation of electric machinery: using Matlab/simulink, Prentice Hall,
New Jersey, 1998.
[11] M. Rashid, Power electronics handbook: Academic Press, 2001.
[12] M. Tsai and W. Tsai, "Analysis and design of three-phase AC-to-DC converters with
high power factor and near-optimum feedforward," IEEE Transactions on Industrial
Electronics, vol. 46(3), p. 535, 1999.
[13] V. Blasko and V. Kaura, "A new mathematical model and control of a three-phase AC–
DC voltage source converter," IEEE Transactions on Power Electronics, vol. 12(1), 1997.
[14] A. Dicks and J. Larminie, Fuel cell systems explained: John Wiley & Sons, 2000.
[15] R. Bove and P. Lunghi, "SOFC mathematic model for systems simulations-Part 2:
definition of an analytical model," International Journal of Hydrogen Energy, vol. 30(2),
pp. 189-200, 2005.
[16] P. Aguiar, D. Chadwick and L. Kershenbaum, "Modeling of an indirect internal
reforming solid oxide fuel cell," Chemical Engineering Science, vol. 57(10), pp. 1665-
1677, 2002.
[17] J. Padulles, G. Ault and J. McDonald "An integrated SOFC plant dynamic model for
power systems simulation," Journal of Power Sources, vol. 86(1-2), pp. 495-500, 2000.
[18] C. Wang and H. Nehrir, "A physically based dynamic model of solid oxide fuel cells,"
IEEE Transaction on Energy Conversion, vol. 22(4), Dec. 2007.

64
[19] G. Hatsopoulos and J. Keenan, Principles of general thermodynamics: Wiley New York,
1965.
[20] Fuel cell handbook: EG & G Services, 2002.
[21] S. Chan, C. Low and O. Ding, "Energy and exergy analysis of simple solid-oxide fuel-
cell power systems," Journal of Power Sources, vol. 103(2), pp. 188-200, 2002.
[22] Ø. Ulleberg, "Modeling of advanced alkaline electrolyzers: a system simulation
approach," International Journal of Hydrogen Energy, vol. 28(1), pp. 21-33, Jan. 2003.
[23] T. Zhou and B. Francois, “Modeling and control design of hydrogen production process
for an active hydrogen/wind hybrid power," International Journal of Hydrogen Energy,
vol. 34(1), pp. 21-30, Jan. 2009.
[24] S. Inoue and H. Akagi, "Bi-directional dc/dc converter for an energy storage system,"
22nd IEEE Applied Power Electronics conference, Feb.-Mar. 2007.
[25] R. De Doncker, D. Divan and M. Kheraluwala, "A three-phase soft-switched high power
density DC/DC converter for high power applications," IEEE Transaction on Industry
Application, vol. 27(1), pp. 63-73, 1991.
[26] O.C. Onar, M. Uzunoglu and M.S. Alam, "Modeling, control and simulation of an
autonomous wind turbine/photovoltaic/fuel cell/ultracapacitor," Journal of Power
Sources, vol. 185(2), pp. 1273-1283, Dec. 2008.
[27] R. L. Spyker and R. M. Nelms, "Analysis of double-layer capacitors supplying constant
power loads," IEEE Trans. Aerospace Electron, vol. 36(4), pp. 1439-1443, Oct. 2000.
[28] "Electric Double Layer Capacitor: BOOSTCAP_ Ultracapacitor," Available online at:
http://www.maxwell.com/pdf/uc/datasheets/mc_power_series_48_1009365_rev3.pdf.

65
66
Chapter 4 Control and Operation Schemes

The purpose of an autonomous electric power system is to produce electricity from its
generators so that it can supply the local demand and transfer predefined power to the external
network at acceptable quality and reliability. In the conventional power generation system, the
balance of supply and demand is made by controlling the fuel input to the generator’s governing
system, so the load-following approach can be handled easily. In a power system with renewable
generators, the load-following approach cannot be achieved since the power generation is
affected by the intermittent weather conditions. Instead, the generators are operated in the way to
capture maximum input energy available. Consequently, integration of the dedicated energy
storage system (DESS) is indispensable to make renewable based power system an autonomous
one, which serves as the medium of energy and power balance.
The main objectives of the proposed hybrid power system are: to capture as much as wind
energy available, and supply the local load demand and dispatchable power using the dedicated
energy storage system. To fulfill the objectives, there are three subsystems to be controlled. In
particular, the control action for the wind turbine generator system (WTGS) enables in getting
optimum power from wind; the control action for the line side converter (VSC2) regulates the
power, frequency and voltage of the point of common coupling (PCC) bus; and the control action
for the DESS balances the supply and demand of power (in transient condition) and energy (in
steady-state condition), irrespective of fluctuations in load and variations in generation. Each of
these subsystems has many control variables to achieve the desired results, and the control laws
of the variables are based on the dynamic models of each subsystem. There are many approaches
to the controller or the compensator design. Due to simplicity for designing, internal model
control (IMC) scheme is used in this study to select the parameters of the controllers. In the
following sections, the conceptual framework of the IMC based design is described first,
followed by the description of the control/operational schemes for the each subsystem.

4.1 Internal Model Control Based Design of Controllers

The goal of a typical control design is to get fast and accurate set-point tracking against the
effect of external disturbances. Further, the resultant system should be capable to operate over a

67
wide range of operating conditions with enough stability margins. The controller is usually
designed based on an approximate model of the real plant, and it is likely that the parameters of
the plant vary with operating conditions and time. Therefore, it is essential to design a control
system that shows robust performance against mismatches. In this study, proportional-integral-
derivative (PID) type controllers are applied due to their relatively simple structures, which can
be comprehended and implemented easily in practices. Finding a simple design technique for
PID type controllers with the desired performances is an important issue in tuning the controllers.
There are a number of methods for tuning PID controllers, among which the IMC based design
approach is often used in process control systems because it demonstrates a clear tradeoff
between the closed loop performance and robustness, by means of a single tuning parameter
namely the filter time constant [1]. The following paragraphs describe an evolution of a simple
IMC structure to obtain the PID control schemes, in brief.

Gd ( s )

C ( s) G(s)

Gˆ ( s )

(a)

F (s) Gd (s)

C(s) C(s)

Gˆ ( s )

(b)

Fig. 4-1 Plant control topologies. (a) Internal model structure, and (b) equivalent classical model.

68
 A commonly used IMC structure is illustrated in Fig. 4-1(a). The structure uses an internal
model Gˆ ( s ) in parallel with the controlled plant G(s) . To get the desired response, a control
block C(s) is augmented so that the output is related as [2]
 C ( s )G ( s )   1  C ( s )Gˆ ( s ) 
Y (s)    R( s )   Gd ( s)  D( s )
1  [G ( s )  Gˆ ( s)]C ( s )  1  [G ( s )  Gˆ ( s)]C ( s ) (4.1)
    
GCLR GCLD

where “^” denotes the measured value of the parameter; the symbols r, y, u and d denote the
reference input, the controlled output, the plant input and the disturbance input, respectively as
specified in the figure. Similarly, Gd(s) is the disturbance transfer function and the variables with
capital letters represent the mathematical expressions in the Laplace Transformation domain.
This IMC can be considered as a special case of the classical control structure as shown in Fig.
4-1(b), where an equivalent controller block F(s) is related to the internal model in the following
way
F ( s )  [1  C ( s )Gˆ ( s )]1 C ( s ). (4.2)
If we desire to get set-point tracking, it is necessary to make C (0)Gˆ (0)  1  0 . This guides us to

set C (0)Gˆ (0)  1 . For the sake of simplified understanding of the IMC, consider the case when

the internal model is perfect, i.e., G ( s )  Gˆ ( s ) , then there is no feedback in Fig. 4-1(a).
Consequently, the output becomes
Y ( s )  C ( s )G ( s ) R ( s )  [1  C ( s )G ( s )]Gd ( s ) D ( s ). (4.3)

In this case, the close loop system is stable if C (s) and G(s) are both stable, implying that we
have stablizing controllers F (s) , given by (4.2), for the stable plant G(s) . Under this condition,

if we set C ( s )  Gˆ 1 ( s ) , all plant dynamics would be canceled and the output signal would attain
the set point instantaneously. Though the approach seems tempting, it is not applied because of
several reasons, such as: (i) the possibility of G(s) being non-mimimal phase (presence of zeros

in the right-half plane, which makes G 1 (s) unstable), (ii) the possibility of G 1 (s) being
improper, and (iii) the approach is highly sensitive to the model error. This conpect, however,
can still be used with a minor modifications in the control structure. A general approach is to
factorize G( s) = GM (s)GA (s) ; such that GA ( s) is the allpass part of G(s) including all zeros in

69
the right-half complex plane and the time delay, and GM (s) is the remaining part. Then, we

introduce a filter
n
 1 
L( s )    , (4.4)
 1  s 
where n is a positive integer chosen sufficiently large so that C(s) becomes proper, and  is the
parameter to adjust the filtering function that provides the tradeoff between performance and the
robustness. Now, by setting C ( s )  Gˆ M1 ( s ) L ( s ) , we can address all the aforementioned issues

with C ( s )  Gˆ 1 ( s ) [1]. Let us derive the resultant close loop response as

 Hˆ ( s)[1  EM ( s )]   1  Hˆ ( s ) 
Y ( s)    R( s )   Gd ( s )  D( s ) , (4.5)
1  Hˆ ( s ) EM ( s) 
 1  Hˆ ( s) EM ( s)

   
GCLR GCLD

where Hˆ ( s )  L ( s )Gˆ A ( s ) and EM ( s ) is the model error given by EM ( s )  [G ( s )  Gˆ ( s )] / Gˆ ( s ) .

For the special case of a perfect model ( EM ( s)  0 ), (4.5) reduces to

Y ( s )  Hˆ ( s ) R ( s )  [1  Hˆ ( s )]Gd ( s ) D ( s ) , (4.6)
as a result of pole-zero cancellation. Clearly, (4.6) demonstrates that for the model with no
mismatch, the nominal close loop transfer function Hˆ ( s )  L ( s )Gˆ A ( s ) is the designer’s choice

provided Gˆ A ( s ) contains all the right-half plane zeros and delays, and L(s) must be of
sufficiently high order to avoid physically unrealistic control action. This idea leads us to
designing the equivalent classical model of feedback control structure via F (s) as defined in
(4.2) and Fig. 4-1(b).
Substituting the value of C(s) in (4.2), we get
L( s)
F (s)  . (4.7)
GM ( s )[1  L ( s )Gˆ A ( s )]
ˆ

In our study model, there are no zeros on the positive half-plane nor the delay part (i.e
Gˆ A ( s )  1 ), which simplifies the design part substantially. As an illustration, consider a first

order system G ( s )  k M / ( p s  1) where kM and τp are the value of gain and time constant, it is

sufficient to set n = 1, in which case α becomes the desired bandwidth of the closed-loop system.
Consequently, the controller becomes an ordinary proportional-integral (PI) controller expressed
by

70
 1 ˆ 1 s  1  1
F (s)  G (s)  p  p  . (4.8)
s s k M  k M s k M
Then, the output response with the nominal model becomes
1 s
Y (s)  R( s)  G ( s ) D ( s ). (4.9)
 s 1  s 1 d
Thus, the tracking of command input and disturbance rejection is matter of selecting

desired value of α.

Improved IMC

While the conventional filter suggested by (4.4) provides good performance for set point
tracking, it may result poor performance for disturbance rejection. To show this, consider the
closed-loop output response, in which Gd(s) having the same dynamics as that of the plant in Fig.
4-1. Then, the output response of (4.9) can be written as
1 kM  s
Y ( s)  R(s)  D(s) . (4.10)
1  s ( s  1)( s p  1)

If  p is large, the disturbance rejection will be poor as guided by  p , whatever the value of  we

select. That means there will exist a long tail in y(t) due to d(t). To fix this problem, an improved
filter form (for type I with step input) is suggested as [3]
1  s
L( s )  . (4.11)
(1   s)n
The constant parameter  is chosen so that the slow pole of Gd(s) is canceled by a zero (via pole
zero cancellation) in the transfer function, Y (s) / D(s) . Such a pole-zero cancellation is effective
if: (i) the plant dynamics is relatively known, (ii) plant does not have unstable open loop poles,
and (iii) the system does not have internal disturbances associated with the dynamics slower than
the desired closed-loop response. Consequently, by substituting the value of L(s) from (4.11) to
(4.7), the expression for F(s) can be written as
 s 1
F (s)  . (4.12)
Gˆ M ( s )[( s  1) n  ( s  1)]
As an illustration, for the first order system with G ( s )  k M / ( s p  1) as in the earlier example, we

set n = 2, in which case F(s) becomes

71
 ( s  1) ( s  1) s p  1
F (s)  Gˆ M 1 ( s )  . (4.13)
s ( s  2   )
2
s ( s  2   ) k M
2

By setting  p   2 / (2   ) or   2   2 /  p , the output response with ideal model becomes

( s  1) k (2   ) s
Y (s)  R( s )  M D( s ) . (4.14)
( s  1) 2
( s  1)2
Thus, we can select the desired settling time for the command point tracking and disturbance
rejection by specifying the parameters α and  . In case of models having mismatches with the
real plants, the resultant output response will be close to (4.14) as guided by (4.5) and (4.7).
This concept of the controller design as summarized by (4.8) and (4.12) is used in tuning
them and the resultant control transfer function is given in the appendix at the respective places.

4.2 Wind Turbine Generator System Controller

The main objectives of the WTGS controller are to regulate and smooth the generated
power; maximize the energy captured, alleviate the transient loads throughout the wind turbine,
and reduce the machine rotor flux at light load. These objectives are effectively accomplished by
utilizing a variable-speed drive and pitch angle control. While the variable-speed operation is
employed to the rotor-side converter (VSC1) throughout operation, the pitch angle control is
activated only at the wind speed above the rated value (see section 3.2.2).
The control of VSC1 deals with the dynamic characteristics given in (3.17) and (3.29). At
first glance, the design of control scheme seems to be complex, as it has sixth order non-linear
multiple-input multiple-output (MIMO) system. The design part, however, can considerably be
simplified if we note following points.
 Since the stator terminal is connected to the voltage regulated ac bus (either it is regulated
by the VSC2 controller or it is impressed by the grid voltage), the derivative of imo is very
small and the voltage equation becomes algebraic one.
 The dc-link bus voltage regulation is delegated to the DESS, so the control scheme does
not have to deal with it [4].
 In the vector control scheme, the cross-coupling effect is nullified. It means that the
variables of the other axis are termed as disturbance signals. This leads to simplifying the
MIMO system into the separate single-input single-output (SISO) systems.

72
With these conditions, the control law associated with VSC1 can be expressed by two SISO
systems that handle the third order dynamics. The detail description of control scheme is
presented in the following subsections.

4.2.1 Inner current control of rotor-side converter

The main task of the rotor-side converter is to control the machine speed and flux. Due to
the machine inherent dynamics, the speed and the flux control is accompanied with the fast inner
current control loop that controls the rotor current. To achieve the effective control, it is
preferred to decouple the torque and flux quantities by employing the field orientation concept,
such that q-axis component of the rotor current dictates the torque and the d-axis component is
used to regulate the excitation flux or the reactive power.
In the stator field oriented control, synchronously rotating d-axis is oriented along the
stator flux vector position. Assuming the stator terminal is connected to the voltage regulated bus,
the magnetizing current vector (imo) is nearly constant. Further, the influence of stator resistance
is very negligible. Under this orientation, expressions for the quantities of interest (voltages,
torque, rotor speed, stator flux and stator reactive power) can be deducted from (3.15) to (3.22)
as [4, 5]

ψ s   sd  Ls i mo  Ls isd  Lm ird 
, (4.15)
 rd  L2mimo / Ls   Lr ird ,  rq   Lr irq 

vrd  rr ird   Lr (dird / dt )  vdcom 


vrq  rr irq   Lr (dirq / dt )  vqcom  , (4.16)

J m (dr / dt )  (Tm  Tem )  Dmr 
with
vdcom  (e  r ) Lr irq 
 (4.17)
vqcom  (e  r ) Lr irq  ( Lm / Ls ) sd 

Tem  0.75n p sd irq (4.18)

Qs  1.5(e  r ) Lm sd (ird  sd / Lm ) Ls 1 . (4.19)

To design the observer for the stator flux vector position, the stator voltage and the current
quantities are taken as feedback variables. By using stationary Clark’s transformation, the flux
and its phase angle is calculated as [6]

73
  s   (vs  rr is ) dt ,  s    (vs  rr is ) dt 


 f  tan 1 ( s /  s ) , (4.20)

 s   s2   s2 

where the subscripts α and β represents the variables in α and β axes, respectively; and  f

denotes angle of rotating flux. From (4.16), it is possible to include a feed-forward compensating
term in the control law that will compensate the tracking error caused by the variation of vdcom
and vqcom. Consequently, the desired rotor voltage is derived as

vrd  (irdref  ird )GC1 ( s )  vˆdcom 

, (4.21)
vrq  (irqref  irq )GC1 ( s )  vˆqcom 

where the superscript ‘ref’ represents the set value of the respective quantity. Here, the same
controller GC1 ( s) is used for the d and q rotor current control because of similar electrical

dynamics in those axes.

vˆdcom vdcom

irdref 1 ird
GC1 ( s )
 Lr s  rr

(a)

vˆqcom vqcom

irqref 1 irq
GC1 ( s )
 Lr s  rr

(b)
Fig. 4-2 Control block diagram of inner current control loops. (a) d-axis current control, and (b)
q-axis current control.

74
 With help of dynamic equations given in (4.16) and control laws expressed in (4.21), the
overall control diagram is constructed as in Fig. 4-2. The aim of the controller GC1(s) is to
regulate current against the disturbance in the feed forward path. While deriving the control law
of control block GC1(s), we assume that the dc-link voltage (Vdc) is at the rated value (so that the
rotor voltage is written as vrd  Vdcref d1d ) and ωr varies slowly with compare to current (in fact ωr

is regulated by outer q-axis control loops, so for the inner loops it is assumed as constant).
Consequently, the model can be linearized at the nominal values ( Vdc  Vdcrated and  r   rrated )

and the controller is designed using the IMC concept. With the help of the expression in (4.8),
the GC1(s) is reduced to a proportional-integral controller defined as
1
GC1 ( s )  ( Lr s  rr ) , (4.22)
1c s
where 1c is the desired time constant of the close loop system, and its value is selected as 10 ms

to assist the fast tracking of the current. The reference value irqref in (4.21) is set from the outer

speed control loop (to be discussed later), whereas the setting of irdref is guided by (4.19). For the
operation of an induction machine in the generating mode, the reactive power is always drawn by
the machine from the system. If the reactive component of rotor current is zero (ird =0), the stator
reactive power (Qs) assumes a negative value, meaning that stator absorbs the required reactive
power from bus #1, in which case Qs is drawn from VSC2 (keeping the operation of the
microgrid at unit power factor mode). This may not be a desired condition as VSC2 would be
overloaded. The reactive power demand for the generator can be supplied by VSC1 if ird is
imposed such that
irdref   sd / Lm . (4.23)

While setting the reference current by (4.23), the kVA rating of VSC1 is to be increased for
a given value of generating torque. In this study, Qs is nullified through the rotor current (ird) to
reserve the kVA capacity of the grid-side converter (VSC2) for storage management.

4.2.2 Speed control of wind turbine

There are two modes of rotor speed control with a variable speed wind turbine. The first
mode is in the region of lower wind speeds, in which the set-point speed is varied with wind
speed (known as maximum power point tracking) and speed regulation is performed by the rotor-
side converter by varying the power output. On the other hand, the second mode is in the region

75
of higher wind speeds, in which the set-point of rotor speed is kept constant and the rotor speed
is regulated via pitching the rotor blade until the output power reduces to the rated value.
Following paragraphs explain the control approach for each scheme.

(a) Speed control through rotor-side converter

The rotor speed control of is related to the outer control loop of the rotor-side converter
shown in Fig. 4-3, in which the deviation in the rotor speed caused by the disturbance torque (the
mechanical torque) is nullified by employing the compensator GC2(s). The controller action
generates the set-point electromagnetic torque ( Temref ) given by

Temref  ( rref   r )GC 2 ( s ) , (4.24)

where  rref is the command reference of rotor speed set by the maximum power point tracking

(MPPT) technique. Noting that Tem  0.75n p sd irq (from (4.18)) and the stator flux ( sd ) is fairly

constant, the controller output generates q-axis set-point current ( irqref ) consequently.

For the speed control scheme, we set the time constant 1c of the inner current dynamics

sufficiently smaller than the speed dynamics, so that the current control loop can be assumed as
the unit gain, while looking from the outer loop. Considering the mechanical dynamics of the
WTGS given in (4.16), the transfer function from Temref to r becomes the linear function, given by

r ( s ) 1
 . (4.25)
T ref
em ( s) sJ m  Dm

rref 1 irqref 1 irq 1 r

Temref
GC 2 ( s ) 0.75n p ds
0.75n p ds 1c s  1 s J m  Dm

Fig. 4-3 Block diagram of speed control loop.

It should be noted that the IMC method described in (4.8) would be insufficient to set the
parameters for GC2(s), as the time constant of the plant (Jm/Dm) is quite larger than the time
constant of the associated filter L(s). Therefore, the improved IMC is utilized for tuning. From
(4.13) with n=2, the expression for GC2(s) can be derived as

76
 (1  s 2 c ) Dm
GC 2 ( s)  ( sJ m / Dm  1).
s ( 2 c s  1)2  ( 2 c  1) 

Rearranging, we get
(1  s 2 c ) Dm
GC 2 ( s )  ( s tg  1),  tg  J m / Dm . (4.26)
  22c 
s (2 2 c   2 c )  s  1
 2 2 c   2 c 
 
Here,  2c and  2c are the constant parameters to be set for the IMC. To accomplish the fast

disturbance rejection, we set  tg   22c / (2 2 c   2 c ) , which gives

(1  s 2 c ) Dm Dm  2 c Dm
GC 2 ( s )    . (4.27)
s (2 2 c   2 c ) s (2 2 c   2 c ) (2 2 c   2 c )
Finaly, the output response can be written as

( 2 c s  1) ref D (2 2 c   2 c ) s
r ( s )  r ( s )  m Tm ( s ) . (4.28)
( 2 c s  1) 2
( 2 c s  1) 2

Thus, by the choice of suitable value of  2c and  2c , we can eliminate the long settling time of

the rotor angular velocity due to the disturbance torque, irrespective of the disturbance transfer
function, Gd(s). In this case, the time constant of the speed control loop is selected as 0.7 s
considering the slower mechanical dynamics of the rotor, which leads to setting  2 c  1.35s .

vw
Tem
 rated  max
1  1
r
r Tm
 ref s pit  1 sJ m / n p  Dm
0
r

Fig. 4-4 Block diagram of rotor pitch angle control.

(b) Speed regulation through pitch angle control

As discussed earlier, the pitch angle control is operated in high wind speeds. In those
circumstances, the rotor speed can no longer be controlled by increasing the generated power, as
this would lead to overloading the generator and the converter. Once maximum rating of the
generator/converter is reached, the pitch angle is increased; which in turn reduces the power

77
conversion coefficient (CP), and consequently the excess power is spilled away from the turbine.
It should be noted that during this period, the set-point torque for the variable speed drive
(VSC1) reached to its upper limit due to saturation, so it cannot provide the speed regulation.
The control block diagram for the pitch control scheme is shown in Fig. 4-4. The wind
turbine and generator model acts as the plant for the system represented by (3.12) and (3.13)
(chapter three), in which a compensator is employed for the speed regulation to generate the set-
point value of pitch angle (  ref ), given by [7],

 ref  GCP ( s )(rrated  r ) , (4.29)

where GCP ( s) denotes the compensator transfer function. To put the blades into the necessary

position, pitch servos are employed as actuators which may be hydraulic or electrical systems.
Because of large mass of the rotor, the actuator cannot regulate the rotor speed quickly.
Therefore, the actuator is represented by an integral function, and subsequently the transfer
function for the pitch angle control loop becomes the first order delay (with time constant,  pit )

and a pitching rate limiter [7]. As a result, the regulated pitch angle alters the power conversion
coefficient (CP) till the turbine power output equals to the rated value.
To design the compensator GCP ( s) for the rotor speed control, a PID controller is selected

with the parameters as suggested in [7, 8], represented as

 1 s 
GCP ( s )  k pit  1   Td , pit ,
 Ti , pit s 1  0.1Td , pit s 

where k pit , Ti , pit , and Td , pit are the control parameters of GCP ( s) .

4.2.3 Maximum power point tracking scheme

To capture the maximum power from the wind turbine, it is necessary to alter the set-point
of the rotor angular speed according to the variation in wind speed. The conceptual framework
for setting optimum value of rotor speed, referred to maximum power point tracking (MPPT)
technique, can be understood from Fig 3.3(b). It can be observed that the steady-state maximum
power from the turbine is obtained when power conversion coefficient is at its maximum value
(CP-max). This can be achieved by setting the rotor speed proportional to the measured value of
wind speed as

78
 tipopt
ropt  vw , (4.30)
rT

where  ropt is the optimum rotor speed related to the maximum power output from wind turbine.

From (4.30), we note that  ropt is set by direct measuring of wind speed, which may not be

always desirable. To avoid the direct measurement of wind speed, it is modified to set the rotor
speed corresponding to the measured value of the active power from the generator (Pig), with
copper loss included, as [9]
0.5 ArT3cP  max 
Pig  kopt (ropt )3 , kopt 

(tipopt )3 
. (4.31)
P 
ropt  3 ig

kopt 
Since kopt is a constant parameter for the turbine, only measuring the transferred power
allows calculating the reference speed which, in turn, seeks the optimal operation. This speed can
be imposed in the controller scheme, and optimum operation can be achieved after a few
computational iterations following the dynamics, (1  s 2 c ) / (1  s 2 c ) 2 .

rrated  max

r 0
v dcom
Tref ref
i v qcom
ropt em rq
Vrref
 abc
1
isq V dc
Pig
1
irdref
Lm isd

 dt
d / dt

Fig. 4-5 Block diagram of overall control scheme for the DFIG.

79
 (a)

(b)
Fig. 4-6 Output respoonses of conttrol block shhown in Fig. 4-3
4 with  200% variation of plant paraameters: rr,
Lm, Lr , Dm and Jm . (a) Step coommand inpuut, and (b) steep disturbancce input.

80
4.2.4 Overall block diagram and test of controller
Fig. 4-5 shows the overall control block diagram of the system illustrating how the
individual control schemes fit into the plant. For the VSC1 control action, the input is the
electrical power measured from the generator output terminal. Based upon the measured power,
the command speed is calculated using the MPPT technique. The set value of command torque
( Temref ) is calculated either by the outer speed control loop (when  ropt   rrated ) or set as

Temrated (when  ropt   rrated ) , and then the set value of the rotor q-axis current ( irqref ) is deduced.

Likewise, the rotor d-axis current ( irdref ) is calculated from the measured value of stator flux. The

current control loop generates duty cycles for the converter that produces the required rotor
voltage space vector ( v r ) at the terminal of VSC1. In case when  ropt   rrated or Pig  Pigrated , the

pitch angle controller is guided to generate the desired value of pitch angle (  ). This, in turn,
adjusts CP below its optimum value, so that the generated power can be kept at its nominal value.
To check the applicability of controllers GC1(s) and GC2(s) in the block diagram of Fig. 4-2
and 4-3, a number of tests are performed. The tests include the robustness of the designed
controllers under the variation of the plant parameters. Due to lack of real data for the parameters
uncertainties, this dissertation assumes  20% variation of the parameters from their respective
nominal values. The dynamic uncertainty is, however, not included in the analysis. The
performance test includes
 capability of reference point tracking,
 disturbance rejection, and
 wide stability margin.
Fig. 4-6 shows the output response ( r ) under the unit step command input (  ropt ) and

disturbance input ( Tm ). With uncertainty of plant parameters ( J m , Lm , Lr , rr and Dm ) specified

above, there is minor variation in rise time with the step input of  ropt in Fig. 4-6(a). Note that the

desired value of settling time (5 s) is achieved with zero steady state error for all ranges of
parameters’ variation. Similarly, the disturbance rejection is also observed with similar settling
time, as that of reference input tracking, thanks to the modified IMC control (see Fig. 4-6(b)). It
may be noted that the response curve shows the band between the most optimistic and
pessimistic response. At the boundary, the fastest and the slowest rise times correspond to

81
min{J m , Lm , Lr }  max{rr , Dm } and max{J m , Lm , Lr }  min{rr , Dm }, respectively. For the

intermediate responses, the curves are related to the different combinations of the parameters.
Since we are concerned with the values in the boundaries, the determination of each combination
related to the intermediate responses are of less important. Likewise, Fig. 4-7 depicts the Bode
plot of the tuned system. Clearly, the plot has wide stability margin (75o phase margin) even
under the most pessimistic plant model. These figures validate the desired robust performances
of the controllers under the assumed level of parameters’ uncertainty.

Fig. 4-7 Bode plot of control block shown in Fig. 4-3 with  20% variation of plant parameters: rr, Lm, Lr ,
Dm and Jm.

4.3 Dedicated Energy Storage System Controller

The objectives of the DESS controller are to generate the actuating signals for the dc-dc
converters, DC1 and DC2. The converter DC2 is assigned for regulating the dc-link bus voltage
in the expense of drawing (delivering) compensating current, while the converter DC1 forces to
share the compensating current from the UC and FC/ELZ units in the expense of consuming

82
(generating) H2. Fig. 3-21 (chapter three) provides the guideline for selecting control behavior.
The system has nonlinear dynamics with two command inputs: duty cycle for DC1 switches and
phase shift angle for DC2 (d2/d3, δ), a disturbance current input (Idstb), and number of outputs. By
observing the block diagram following information is noted.
 In the DC2/dc-link bus system block, the dynamics of the UC voltage (Vuc) is quite
slow in compare to the dynamics of the phase shift variable (δ).
 The DC2 input current (Ii) appears as the disturbance input for the FC/ELZ/UC/DC1
system.
These properties lead to decoupling the DESS into two SISO blocks, simplifying the
control design. The control schemes associated with these units are discussed in the following
subsections.

4.3.1 Control approach for fuel cell side dc-dc converter

The converter DC1 is an actuator that assists in sharing the desired current in a manner that
transient and steady-state parts are supplied by the UC and FC/ELZ, respectively. This can be
achieved by implementing a control scheme in two loops, in which the inner one is employed for
current regulation and the outer one is designated for the UC voltage regulation. The plant
associated with the DC1 control action is the FC/ELZ/UC/DC1 system shown in Fig. 3-21
(chapter three), in which the controller blocks GC3(s) and GC4(s) are augmented to achieve the
desired closed loop response. The resultant control block diagram for the system is depicted in
Fig. 4-8. Here the blocks GC3(s) and GC4(s) represent the controllers for inner current control
loop and outer voltage control loop, respectively.
While designing the inner controller, GC3(s), the dynamics of system given in the
FC/ELZ/DC1 part has to be dealt with. Since the system has nonlinearity as represented in (3.58),
it is linearized around the stable operating point to investigate the control approach. The dynamic
equations along with other related algebraic equations are linearized at the nominal operating
point of the SOFC (rated value of fuel cell current, I dc1  I dcrated
1 ; rated value of voltage,

V fc  V fcrated ; and nominal value of duty cycle for switches: d 2 , d 3  D2 , D3 ). The entire pole and

zeros of the linearized system lie on the left-half of the complex plane, in which the dominant
pole-zeros are depicted in Fig. 4-9. Clearly, the system is stable, but it has a sluggish dynamics
due to the poles near -0.2 (state represented by fuel regulation dynamics). This leads to selecting

83
a proportional-integral-derivative (PID) controller for tracking the current. Since the response of
the current control loop should be fast, we specify the bandwidth and phase margin as 200 rad/s
and 85 degree respectively. The resultant close loop step response plot with current as the
reference input is depicted in Fig. 4-10. It can be observed that, the control loop has fast current
tracking property thanks to the applied PID controller.
In the fuel cell model, the states related to the partial pressure dynamics do not have the
provision of feedback regulation. The presence of these non-controllable states having sluggish
dynamics puts the constraints of selecting the reference current for the FC/ELZ in a way that the
current dynamics should not have smaller time constant than that of the dominant poles.
Otherwise, it may result fuel starvation in the FC. This demands a high farad rating of the UC
which takes care of the transient part until the steady state current flows through the FC/ELZ part.
From (3.52), the output current from DC1 (Idc1) is proportional to the FC or ELZ current (Ifc or
Ielz), so it would be convenient to control Idc1 alone. For this, a reference output current for DC1
is generated as
Ii
I dcref1   GC 4 ( s)(Vucref  Vuc ) . (4.32)
Td s  1
The first part of (4.32) ensures that only slow varying and the steady state part of current
flows through the FC/ELZ. The current rise time is decided by the consideration of the SOFC
parameters, the maximum value of the fuel cell current ramp rate and time constant of the fuel
regulation block. The second term is associated with the outer control loop that regulates the UC
voltage. The design of the controller GC4(s) is based upon the allowed range of Vuc. This, in turn,
is decided by the minimum operating voltage for the SOFC (Vfc,min), the maximum allowable
voltage for the ELZ (Velz,max) and the capacitance of the UC (Cuc). In this study, the range of Vuc
is taken from 225 to 275 V with its nominal value at 250V. By doing so, it subjects the rated UC
voltage below Vfc,min , ensuring the desired buck/boost operation for the DC1 converter. By these
considerations, GC4(s) is represented by a PID controller to obtain the close loop step response
having settling time of 20 s and the maximum overshoot of about 5% with a satisfactory stability
margin.

84
 1
sTd 1

V ferated

Vucref Idcref1 1 Idc1 R pl Rsr Cuc s  Rsr  R pl

GC 4 (s) GC3 (s) I dcrated
1 D2 , D3
sLi  Ri R pl Cuc s  1

85
D2 , D3

Fig. 4-8 Control block diagram of FC/ELZ/UC/DC1 system.

Fig. 4-99 Pole zero pplot of FC/EL
LZ/DC1 systeem. “zero-maark” and “crooss-mark” arre set for zeroo and pole
respectiively. The nuumber in the radial line reepresents the cosine of anngle between the radial linne and the
neegative real aaxis.

Fig. 4-100 Step responnse of the innner current coontrol loop.

86
 (a)

Phase
margin=75o

(b)
Fig. 4-11 Reesponses of D
DC1 controlleer with  20%
% variation of plant parameeters: Td,Cfc ,Ri and Li

from their nnominal valuues. (a) Unit sstep responsee plot, and (bb) frequency rresponse plot.

87
Test of controller
In order to test the applicability of the proposed control actions, the control block diagram
is subjected to various tests, by considering the uncertainties of the plant parameters
( Td , Li , Cuc , Ri and Rsr ) as  20% with their nominal values. Fig. 4-11(a) represents the step

response plot for the outermost control loop given in Fig 4-8. It can be observed that the
controller provides good tracking of the reference input against the uncertainties of the plant
parameters. The effect of uncertainties is on rise time, but the settling time and steady state
responses are not affected. It is also notable that the response curves show the band between the
most optimistic and pessimistic responses. On the boundary, the fastest rise time and the slowest
rise time correspond to the min{Td , Li , Cuc }  max{Ri , Rsr } and max{Td , Li , Cuc }  min{Ri , Rsr },

respectively. The intermediate response curves, on the other hand, are related to the different
combinations of the parameters.
In order to test the stability margin of the system over the range of parameter uncertainties,
the open loop Bode plot is also presented in Fig. 4-11(b). It is shown that the controller provides
wide stability margin of about 75o even in the assumed pessimistic scenario. These control
parameters are transferred to the original large signal model to test the system under various
operating conditions.

4.3.2 Control approach for the isolating dc-dc converter

The control block for the isolating dc-dc converter is associated with the DC2/dc-link bus
system (illustrated in Fig. 3-21), having a simple first order linear system. Since the time constant
of the UC terminal voltage ( Vuc ) is very large with compared to the dc-link voltage (Vdc), the
former varies very slowly and assumed as a constant quantity while designing the Vdc control
loop. Thus, by linearizing at the rated UC voltage ( Vucref ), the control block diagram is
constructed as in Fig. 4-12(a). The controller block, GC5(s) is employed here for the regulation
of Vdc at the designated value, against the disturbance current (Idstb) imposed from the dc-load,
VSC1 and VSC2. In this case, the controller GC5(s) needs to have good disturbance rejection, as
the command input ( Vdcref ) is the constant quantity. This is achieved by a PID controller, in

which the settling time of the close loop step input is chosen about 0.4 s. Clearly, the desired
disturbance response of the controller is achieved for the given choice of the parameters (see Fig.
4-12 (b)), and consequently the fast voltage regulation is achieved.

88
 I dsttb

-
Vdcref I sto
δ V ref
uc 1
GC 5 ( s ) 2 Lt f sw Vdc
+ Cdc s
+
-

Plant

(a)

(b)

Fig. 4-112 Control scheme for thhe DC2/dc-linnk bus system

m. (a) Blockk diagram , annd (b) step reesponse of
the disturbance input.

4.4 L
Line-Side Converteer Controll
Thhe controlleer employedd with the lline-side coonverter (VS
SC2) has wide
w range oof control
actions. In case of a line convverter operattion when thhe DESS iss not conneccted, main oobjectives
remain tto inject avaailable pow
wer of wind power
p to the grid by reegulating thee dc-link vooltage and
maintainn the desired power facctor, which m
means to opperate in a nnon-dispathaable mode. However,
H
by incorporating thhe DESS, itt can controol the frequuency and vvoltage of tthe PCC buus (during
islandedd mode) orr regulate tthe dispatchhable tie-linne power (in
( grid intterconnected mode).
Therefoore, the conntrol action should be designed too operate inn both moddes like a traditional
t

89
excitation and governor controlled synchronous generator. To accomplish the objective, the
subsystem is designed to mimic the conventional excitation and governor control schemes
system, by utilizing the artificial frequency-active power and voltage-reactive power droop
concepts [10],[11].

4.4.1 Frequency and active power control

Basic idea behind the frequency control is to introduce an artificial droop such that any
increase in the power demand decreases the system frequency, and vice versa. The expression of
frequency-power droop and phase angle of terminal voltage can be written as

eref  0  m ( P2ref  P2 ),  e   eref dt. (4.33)

Here, P2 is the active power output from the inverter, m is the droop coefficient, 0 is the

nominal value of angular frequency of terminal voltage. Likewise,  eref and θe are the reference

frequency and the phase angle of VSC2 terminal voltage, resepctively.

From Fig. 2.6(a) (see chapter two), the impedance between the terminal of inverter and bus
#2 being inductive in nature (2πLf >>Rf), the phase angle and the voltage drop become relatively
linear with the injected active and reactive power, respectively, as

E2V2 E2 
P2  e , Q2   E2  V2  , X f  e L f , (4.34)
Xf Xf 

where Q2 is the reactive power flow from VSC2. The other variables and parameters are defined
in Fig. 2.6(a). From (4.33) and (4.34), we have

d e mE2V2
  e  0  mP2ref . (4.35)
dt Xf

The value of P2ref is derived from the outer active power management loop (to be discussed

in the next section). Equaion (4.35) illustrates that the power injecting to bus #2 with droop of
the inverter frequency resembles the governor and the inertia characteristics of traditional
generators and provides a degree of negative feedback, aiding the smooth frequency regulation.
This phenomenon is also illustrated in Fig. 4-13. The frequency becomes the nominal value only
when the desired and the actual active power flow are equal. Otherwise, it deviates from the

90
nominal value as decided by (4.33). The integral of this frequency becomes the phase shift (θe)
that controls the active power flow. At steady state, the phase angle corresponds to
( mP2ref   0 ) X f / mE2V2 following the dynamics. Consequently, the power injected from VSC2

becomes
P2  P2ref   0 / m, (4.36)

where the droop coeffiecient is defined as

0   fullload
 m. (4.37)
P2rated
Here,  fullload is the full-load frequency of VSC2 terminal voltage (E2). The selection of m is based

upon the allowable range of the frequency deviation from no-load to full loaded condition. In
case of an inverter interfaced system, m is usually set to 2% [12]. Alternately, it decides the
tracking error by 0 / m , which is nullified later by the secondary frequency control.

0

P2
P2ref 1 V2 E2
s Xf

Fig. 4-13 Block diagram of P/f control.

4.4.2 Voltage control

The concept of voltage and reactive power regulation (V/Q control) is similar to the
traditional excitation control with increased flexibility thanks to the converter application. Here,
the voltage control is exercised in the rotating d-q coordinate system, in which the output voltage
magnitude reference aligned with the d-axis of the inverter reference frame as presented as
v2refd  V2 , v2refq  0 ; (4.38)

where, v2d and v2 q are respectively the d and q axis components of v2 . The command value of

voltage ( v2refd ) is derived from the outer V/Q management loop (to be discussed later). It is notable

91
that by forcing v2 q to zero, the angle θe generated from the frequency control action becomes the

phase shift between E2 and V2, and thereby making P2 proportional to θe. There are two cascaded
control loops for both d and q axes. The outer loops regulate the ac bus voltage and provide the
reference signals for the inner current loops. The purpose of the inner current control is to
decouple the d-q axis variables, reject high frequency disturbances and provide sufficient
damping for the output filter.
The design of inner current control is based upon the dynamic given in (3.27) which are
rewritten as
e 2  R f i 2  L f di 2 / dt  je L f i 2  v 2 
. (4.39)
v 2  Rth i 2  Lth di 2 / dt  v th 
We note that the control law for the inner current loop is similar to (4.21), where the cross-
coupling voltage drop terms are compensated. The control law is defined as
e2  GC 6 (s)(i 2ref  i 2 )k pwmi  je L f i 2  vˆ 2 , (4.40)

where GC6(s) is the transfer function of the controller; and k pwmi is the gain of the pulse width

modulation (PWM) generator, which is considered as the constant parameter neglecting the high
frequency dynamics of the PWM logic. It is because the rise time of a PWM generator gain is in
the range of microseconds whereas the rise time for the current control loop is in the range of
milliseconds. The parameters for the controller is determined using the IMC defined in (4.8). In
this case, GC6(s) is designed as

1 Lˆ f Rˆ f
GG 6 ( s )  ( sL f  R f )   . (4.41)
k pwmi 6 c s k pwmi 6 c k pwmi 6 c s

Consequently, the current is tracked quickly following the first order dynamics with bandwidth
( 1/  6c ) as

1
i2  i ref . (4.42)
 6c s  1
Since we desire fast current tracking for the inner loop, we set the value of bandwidth as 100
rad/s.
The design of the outer voltage control loop is relatively slower than the current control
loop, for which we specify its band width ten times smaller than the inner one ( 7c  106c ). This

92
can be achieved by employing GC7(s) with the voltage error signal, producing the current
reference signal as

2  ( v 2  v 2 )GC 7 ( s ).
i ref ref
ˆ (4.43)
For voltage control loop, the inner current loop can be approximated as a simple delay
block given by (4.42) to simplify the analysis, without losing the accuracy [13]. Consequently,
the IMC scheme in (4.7) with the filter L( s)  1/ ( s 7 c  1) , gives the desired controller GC 7 ( s) as

1  s 6 c  1 
GC 7 ( s )  . (4.44)
 7 c s  sLˆ'  Rˆ ' 
The overall control block, including the inner and the outer control loops, is depicted in Fig.
4-14. The plant is constructed from (4.39), (4.40) and (4.43), in which the controllers are defined
by (4.41) and (4.44). With the value of designated parameters, the unit step response and Bode
plot for the outer voltage tracking loops are plotted and depicted in Fig. 4-15. While plotting the
response curves, the robustness of the controller is checked against the variation of parameters
( L f , R f , Lth , Rth ), around their nominal values. It can be seen that utilizing the proposed control

approach, the overall control demonstrates the desired performances (set-point tracking, stability,
robustness), against the assumed level of the parameters’ uncertainty.

v̂2 v2 vth
Plant
v ref i ref
2 + e2
- i2 +
2
+ + 1
GC 7 ( s ) GC 6 ( s ) k pwmi + L' s  R '
Lf s  Rf + v2
+ -
- -
 je L f

 je L f L'  L f  Lth

R '  R f  Rth

Fig. 4-14 Control block of the line-side voltage source converter, including outer voltage control
and inner current control loops. Note that switching function of VSC2 is replaced by the averaged
duty cycle variables, and the d-q axes control actions are accommodated in a single complex
variable domain.

93
 (a)

Phaase
n=85o
margin

(b)

Fig. 44-15 Responsses of the outter voltage coontrol loop. ((a) Unit step response, annd (b) Bode pplot. The
ressponses are obtained by varying
v the pllant parameteers (Rf and Lf) by 20% of
o nominal vaalues.

94
4.4.3 Active power management

Power flow management scheme of VSC2 is the secondary level control utilized to
determine the preset values of active power ( P2ref ) and voltage magnitude ( V2ref ) for the inner
frequency and voltage regulation blocks, respectively. VSC2 can be operated in three different
modes: the islanded, dispatchable grid-connected and non-dispatchable grid-connected modes.
In the first and second modes, the DESS is assigned for the dc-link voltage regulation by
injecting the compensating current ( I sto ) from it. In the third case (when the DESS is not
working), VSC2 can still supply the power in the non-dispatchable mode. Similar objective of
the power management is also discussed in [14]. However, due to the different storage system
and control strategies applied in the inner loops of the VSC2 controller, a modified power
management scheme is introduced here, making it compatible with the proposed configuration.

P2ref

Ptieref

Vdcref

Fig. 4-16 Active power management scheme for VSC2.

The concept of the active power management scheme is illustrated in Fig. 4-16. Each mode
of the operations (islanded, dispatchable grid-connected and non-dispatchable grid connected
mode) is assigned to a separate loop that produces P2ref which is the sum of these three
components associated with proportional-integral action for (i) frequency regulation, (ii) tie-line
regulation and (iii) dc-link voltage regulation, as defined by (4.45)
P Pt Pdc
       
P2ref  GC 8 ( s){k (0  e )  ktie ( Ptie  Ptie )}ss  GC 9 ( s ) kdc (Vdc  Vdc ) ss .
ref ref
(4.45)

95
Here, Ptieref and Ptie are respectively the command and actual value of tie-line power; GC8(s) and
GC9(s) are the transfer functions of the respective controllers; ss is the status signal of the DESS,
having value ‘1’ and ‘0’ for working and non-working conditions, respectively; and ss is the
complement of ss ( ss  1  ss ) . The control parameters kω, ktie and kdc are the coefficients of
frequency, tie-line power and the dc-link voltage regulation, respectively. Likewise, Pω, Pt and
Pdc are the components of Ptieref related to secondary frequency, tie-line and dc-link voltage
regulations, respectively.
When the system is assigned to operate in the dispatchable grid-connected mode, selection
of Ptieref is necessary. It is derived from the higher-level supervisory control of the WPSS,
considering the power exchange agreement and amount of fuel reserve. With preset value of
Ptieref and the system being connected to the grid, the stiff grid impresses the system frequency

( e  0 ), so the component Pω becomes zero. Assuming the DESS is servicing, ss becomes ‘1’.
This implies that the contribution from dc-link voltage regulation part also zero (Pdc=0).
Consequently, the PI control action is effective only to the tie-line regulation. The dispatchable
mode is attractive from the utility point of view because the system is reflected as a controllable
entity that encapsulates all the local fluctuations.
In case of the islanded mode, the DESS should be in operative mode and there is no power
exchange with the utility grid, thereby resulting both Pt and Pdc equal to zero. The PI control
scheme, thus, acts as the secondary frequency regulation, producing Ptieref as the control output. In
extreme case when the DESS is not working, the dc voltage (Vdc) remains unregulated and the
status signal of the DESS becomes ‘0’ (ss  0) . If we desire to keep Vdc at the designated value,
the disturbance at the dc-link bus should be nullified. That means the available power from
VSC1 should be supplied to the utility grid. In this case, the contribution from the secondary
frequency regulation and the tie-line regulation is zero. Instead, the PI action is applied for the
dc-link bus regulation to generate Ptieref . In short, the active power management scheme, with the
aid of the DESS, provides a smooth operation during the islanded and the dispatchable grid-
connected modes. For the faulted DESS case, the power management loop functions as the dc-
link bus regulation, similar to a normal grid connected non-dispatchable wind turbine generator
system.

96
 The selection of the parameters kω and ktie in the control loops is achieved by observing the
trade-off between the speed and the stability of the active power management. As an illustration,
we can represent the control loop in perturbed equation (when ss=’1’) as

k 
P2ref   ( k8 P  8 I )( k  e  ktie Ptie ) 
s 
, (4.46)
  e 
Ptie  Tg
s 
where “~” represents the perturbed variables, Tg is the coefficient of grid power transfer, and k8P

and k8 I are the parameters of GC8(s). Rearranging (4.33) and(4.46), we can derive the expression
for characteristic equation as
s 2  ( k k8 I m  k8 P k8 I Tg m ) s  mTg ktie k8 I  0 . (4.47)

Knowing that the participation of ktie and kω are respectively on the speed and damping, we select
their values for desired dynamic specification.

4.4.4 Reactive power management

The concept of reactive power management is described in Fig. 4-17. The reactive power
management scheme generates V2ref as the sum of droop of reactive power (Qtie) and nominal
value of bus #2 voltage (V0) as
V2ref  V0  k qt (Qtieref  Qtie ) , (4.48)

where kqt is a constant parameter and Qtieref is the set value of tie-line reactive power. In grid-
connected mode, the stiff grid dictates the PCC voltage (V2). Therefore, the scheme is employed
to regulate Qtie (usually at zero), while supplying all the local reactive power demand. If the
WPSS is required to supply reactive power to the grid, Qtieref may be adjusted by a supervisory
control technique.

V2ref

Fig. 4-17 Reactive power management scheme for VSC2.

97
 In case of the islanded operation, Qtie would be zero. So, the control scheme is guided to
regulate V2 ( v2 d ) at the nominal value (V0). That would be achieved when all the reactive
demand inside the system are met. Thus, the same V/Q power management scheme works in all
modes without interruption or switching to different control blocks.
The overall block diagram of the control and operational schemes is represented in Fig. 4-
18. It illustrates how the individual control schemes fit into the whole system. As desired, the
reactive and active power control/management schemes are decoupled to each other. While V/Q
control and P/f control loops are situated in the inner loop, reactive power management and
active power management schemes appear at the outermost loops. The proposed scheme may be
compared to the conventional governor and excitation control approaches applied in synchronous
machine. It also illustrates how the different operating schemes are utilized by the VSC2 system
with the aid of the DESS.

Reactive power management

V/Q Control
0 kqt +
+ V2ref Vd
- + + +
GC7(s) GC6(s)
Qtie V0 +
- -
I2d -
V2d
L
V2
abc-dq abc-dq L dq-abc
I2 ref
E2-abc
V  I2q -
e - q +
GC7(s) + +
0 GC6(s)
e
+ Pω + e
ω0 1
kω + P ref m
- + 2 s
-
GC8(s)
ωe P2
Ptie
- + +
Pt P/f Control
Ptieref ktie

ss(‘1’ or ’0')
- Vdc
V ref + Pdc
dc
kdc GC9(s)
Active power management

Fig. 4-18 Overall control block diagram of the line-side converter (VSC2) system. The figure shows how
the individual control scheme fits into the overall control and management scheme.

98
4.5 Summary
In this chapter, control and operational schemes of proposed hybrid power system has been
developed. The system has three subsystems, namely WTGS, DESS and VSC2. In the WTGS,
control law is associated with the variable speed drive control, in which the stator flux
orientation control is utilized for the inner current and the MPPT is employed for the rotor speed
tracking in the outmost control loop. Further, the pitch angle control is applied to shed the power
at the higher wind speed. The control strategy for the DESS is associated with dc-link voltage
regulation, in which net compensating power is shared by the UC and the FC/H2/ELZ unit
through the coordinated control action. Likewise, the control approaches of the line-side
converter system include the active and reactive power regulation under different operating
conditions, particularly the stand-alone, dispatchable grid-connected and non-dispatchable grid-
connected modes. The control scheme for each subsystem is decoupled in nature because the
variables of the one system appear to be disturbance input to the other one. The parameters in the
controllers are designed by using linearized models and tuned by applying the IMC method.
Simulated results of control schemes have also been demonstrated to validate the operating
performances, such as command point tracking, disturbance rejection and wide stability margin,
under the variation of plant parameters.

References
[1] D. Rivera, M. Morari and S. Skogestad, "Internal model control: PID controller design,"
Ind. Eng. Chem. Process Des. Dev., vol. 25 (1), pp 252–265, Jan. 1986.
[2] L. Harnefors and H. Nee, "Model-based current control of AC machines using the
internal model control method," IEEE Transactions on Industry Applications, vol. 34(1),
p. 133-141, 1998.
[3] I. G. Horn, J. R. Arulandu, J. G. Christopher, J. G. Van Antwerp and R. D. Braatz,
"Improved filter design in internal model control," Ind. Eng. Chem. Res, vol. 35(10), pp.
3437-3441, 1996.
[4] W. Leonhard, Control of electrical drives: Springer Verlag, 2001.
[5] J. Hu and Y. He, "Dynamic modelling and robust current control of wind-turbine driven
DFIG during external AC voltage dip," Journal of Zhejiang University-Science A, vol.
7(10), pp. 1757-1764, 2006.

99
[6] R. Pena, J.C. Clare and G.M. Asher, "Doubly fed induction generator using back-to-back
PWM convertersand its application to variable-speed wind-energy generation," IEE
Proceedings Electric Power Applications, vol. 143(3), pp. 231-241, 1996.
[7] J.G. Slootweg, H. Polinder, W.L. Kling, “Dynamic modelling of a wind turbine with
doubly fed induction generator” IEEE power engineering summer meeting, vol.1, pp.
644– 649, Vancouver Canada, 2001.
[8] N. Horiuchi and T. Kawahito, “Torque and power limitations of variable speed wind
turbines using pitch control and generator power control,” IEEE power engineering
summer meeting, vol.1, pp. 638 – 643, Vancouver Canada, 2001.
[9] A. Mirecki and F. Roboam, "Comparative Study of Maximum Power Strategy in Wind
Turbines," IEEE International Symposium on Industrial Electronics, vol. 2, pp. 993-998,
2004.
[10] N. Pogaku, M. Prodanovic and T. Green, "Modeling, analysis and testing of autonomous
operation of an inverter-based microgrid," IEEE Transactions on Power Electronics, vol.
22(2), pp. 613-625, 2007.
[11] J. A. Peças Lopes, C. L. Moreira and A. G. Madureira, "Defining control strategies for
microgrids islanded operation," IEEE Transactions on Power Systems, vol. 21(2), 2006.
[12] A. Engler and N. Soultanis, "Droop control in LV grids," Proceeding International
Conference on Future Power Systems, pp. 993 - 998, Nov. 2005.
[13] M. Tsai and W. Tsai, "Analysis and design of three-phase AC-to-DC converters with
high power factor and near-optimum feedforward," IEEE Transactions on Industrial
Electronics, vol. 46(3), pp. 535-543, Jun. 1999.
[14] A. Yazdani, "Islanded Operation of A Doubly-Fed Induction Generator (DFIG) Wind-
Power System with Integrated Energy Storage," Proceeding IEEE Electrical Power
Conference, pp. 153-159, Canada, Oct. 2007.

100
Chapter 5 Simulation Results and Discussion

The aim of the numerical simulation is to observe the system dynamics over a designated
period of time that presents the detail of transient response and predicts the steady state response,
with the wide variations of wind speed and load profile. To generate the time series wind data,
the technique proposed in section 3.2.1 is employed, which includes four components of wind
speed (average, ramp, gust and the turbulence). For the simulation purpose, a few assumptions
are made:
 the startup time for fuel cell and electrolyzer are not considered,
 the electrical system is operated in 3-ϕ balanced load with no current flow through the
neutral,
 switching is perfect, i.e. there is no loss during switching and the higher order harmonics
are neglected.
These assumptions are justifiable for the purpose and the selected simulation time.
In order to investigate the performance, the dynamic model of the components in the
proposed wind power and storage system (WPSS) is developed using MATLAB/Simulink [1].
The control schemes have been designed with linearized models, and the overall control and the
large-signal component models are integrated and simulated. To test the effectiveness of the
proposed operational and control schemes, the WPSS is operated in two modes, islanded and
grid-connected. The particular interest is how the embedded energy storage system facilitates the
wind turbine generator system (WTGS) working in an islanded mode and a dispatchable grid-
connected mode, while fulfilling the operational requirements. To validate the control schemes
proposed in the previous chapters, the dynamics of crucial variables from the different
subsystems are observed under the impression of disturbances. Based upon the test of the
controllers against the parameters uncertainty in chapter four, it is expected that the same is
applied for the overall system, so the simulation is performed with nominal values of the
parameters. The detailed simulation results are discussed in the following subsections.

101
5.1 S
System Staartup and
d Operatioon in Islan
nded Mod
de

Beefore the sttartup proceess, the prooposed hybrrid power ssystem (deppicted in Fiig. 2.6) is
islandedd by openinng the switcch B, all loocal loads arre turned off, and the fuel cell/ellectrolyzer
(SOFC//ELZ) unit is The initial conditions of the statee variables
i kept at thhe standby condition. T
related to
t the WTG med as zeroo to observee the generaator’s startupp process. Inn cases of
GS are assum
the DES DC2, the initial states are set to ttheir respecctive nominnal values. Then, the
SS and VD
inductioon machine is magnetizzed by the stored
s chargge in the ulttracapacitor (UC) and tthe startup
process is carried out
o with the prevailing w
wind speed of about 133 m/s. Durinng the startuup process,
the UC supplies thhe necessarry excitatioon power foor the geneerator making DFIG rrun in the
motorinng mode. After speed bbuilds up, the
t new pow
wer balancee mechanism
m shifts DF
FIG to the
generatoor mode, annd then a seequence of events are introduced into the syystem as described in
Table 5.1.
Table 5.1 Sequence oof events in the simulattion processs
Timee E
Event
0s S
Simulation sttarts, generatoor startup
1s E is switchhed ON
ELZ
5s (
(25+j8) kVA load added
10 s G connectiion with Ptie=
Grid =10kW
15 s W
Wind speed rramp down too 8 m/s
20 s (
(10+j4) kVA load added
30 s E of simulaation
End

Fig. 5-1 Wind speeed profile.

102
 (a)

(b)
Fig. 55-2 Power floow responses of different sources.
Thhe nature off wind speedd profile subbjected to thhe wind turbbine is depiccted in Fig. 5-1. Note
that thiss profile is cchosen to oobserve the types of speed componnents illustrrated in secttion 3.2.1.
Until t=
=15 s, the avverage speeed of wind iis 13 m/s, w
with the gust part and tturbulent coomponents
added too it. At t=155 s, wind raamp is subjeected, whichh causes thee reduction in
i wind speeed around
8 m/s. W
With no locaal load and the fuel-celll/electrolyzzer (SOFC/E
ELZ) system
m being discconnected,
the nett generatedd power is consumedd in charging the dcc-link capaacitor (Cdc) and the
Fig. 5-3(a). After t=1 s when the generator
ultracappacitor (UC)), as illustraated in Fig. 5-2 and F
stabilizees at the rateed speed, thhe control acction of the isolating dcc-dc convertter (DC2) iss activated
and the surplus pow
wer is fed too the ELZ. S
Subsequentlyy, H2 is prodduced and pumped
p to thhe storage
tank. Att t= 5 s, a loocal load off (25+j8) kV
VA is introdduced at thee bus #2, w
which causess temporal
decreasee in the dc-link voltagee (Vdc). The deviation iin Vdc is dettected by the DC2 conttroller, the

103
controller generatess phase shifft angle (δ) as
a the contrrolled variabble, the UC supplies thee transient
power, and then thhe fast restoration of Vdc ce in sequences (see Fig.
d takes plac F 5-3(a)). Since the
generateed power is greater thann the load (P t excess power still fllows throughh the ELZ
Pig>Pload), the
towardss the new steeady state.

(a)

(b)

(c)
Fig. 5-3 Responses of frequenncy and voltaages.
Fig. 5-3(b) shhows how thhe frequenccy of V2 is affected
a by the disturbbances in thee islanded
mode. B t step loaad at t= 5 s, the controll action of thhe line-sidee converter (VSC2) is
Because of the
invokedd. Consequeently, the iinner P/f coontrol actioon is activaated as guiided by (4..31). This
phenom
menon is sim
milar to thee droop conntrolled synnchronous ggenerator coonnected too the grid,

104
which avoids the isochronous mode of operation. The droop mechanism also plays the important
role in sharing loads when multiple inverters are operated in parallel. Because of the droop action,
the inverter frequency decreases from the nominal value temporarily with the increase in load. In
the islanded mode, there is neither the grid frequency to impress the system nor the tie-line flow
(i.e. Ptieref  Ptie  0 ). Therefore, the power management block in Fig. 4-15 functions as the
secondary frequency regulation, nullifying the initial deviation of ωe. Similarly, the step load
causes voltage drop in bus #2 voltage (V2), then the feedback V/Q control action is activated to
restore V2 to its reference value of 460 V (see Fig. 5-3 (c)). Thus, in the islanded mode, the
VSC2 controller performs the designated job; the regulation of the frequency and magnitude of
the terminal voltage by supplying the surplus power to the DESS. The DESS system separates
the power into the transient and steady state parts, and passes them through the UC and the
SOFC/ELZ respectively.

5.2 Power Dispatching with Grid

At t=10 s, the synchronization process is carried out and the WPSS is connected to the grid
via the closing of switch B. Then, a tie-line command ( Ptieref ) of 10 kW is subjected to the VSC2

controller. Fig. 5-2(a) shows that the tie-line power (Ptie) settles at the designated value (10 kW)
after a short transient guided by the active power management scheme. The subjection of the tie-
line command causes decrease in Vdc, which is detected by the DC1 controller in the dedicated
energy storage system (DESS). Since the net generated power is still greater than the sum of the
load demand and the tie-line power, i.e. Pig<(Ptie+Pload), a part of the power still flows through
the ELZ and advances towards the lower steady state value. At t=15 s, the wind speed suddenly
decreases to around 8 m/s (Fig. 5-1) and net active power from the generator (Pig) drops to its
one-third value (15 kW). This causes a drop in Vdc, which is responded by the DESS and supplies
the charging current (Isto) to the dc-link bus. Inside the DESS, the UC supplies the transient
power, whereas the SOFC provides the steady state deficit because the generated power is now
less than the sum of tie-line power and load demand, i.e. Pig<(Pload+Ptie). Due to the slow
dynamics of the SOFC, the current gradually rises toward the newer steady state value as guided
by (4.28). Similarly, another step load of (10+j4) kVA is invoked at t=20 s. The transient is
suppressed by the UC and the SOFC supplies additional steady state current to balance the
increased deficit (see Fig. 5-2(b)).

105
 Fig. 5-3(b) and (c) illustrate the responses of line-side converter (VSC2) to the
aforementioned disturbances. After the connection of bus #1 with grid at t= 10 s, the grid
frequency is impressed on the system. So, the power management scheme now functions as the
tie-line regulation. Likewise, the grid voltage dictates bus #2 voltage (V2), and the control action
is directed to regulate Qtie.
Part of the objectives and the results of power sharing scheme explained above may be
compared with [3], in which a battery system applied for the power management. It has been
reported that the battery system provides the dispatchable power or the load following capability
in the islanded mode. However, since the dc-link voltage regulation and charge/discharge control
both need to be handled by the line-side converter alone, the operational scope is quite limited.
Replacing the battery system by the DESS has wider range of application with more complex
configuration and control scheme. Further, the control and operational scheme introduced in this
research are more compatible in hybrid configurations. It is because the control scheme aims to
impress the frequency as in a conventional synchronous generator, unlike in [3] where the
frequency is indirectly regulated by employing the current compensation.
Thus, in the grid connected mode, the VSC2 controller performs the duty of regulating the
tie-line power (Ptie and Qtie) by injecting the unbalance power from the DESS. Consequently, the
fast transient current on the SOFC and ELZ is strategically avoided. The avoidance of transient
power is very crucial to both the SOFC and ELZ operations. In case of the SOFC, it causes the
fuel starvation, while for the ELZ, the fast transient may lower the operational efficiency and
shortens its life [2]. Further, it is also observed that there is smooth transition from the islanded to
grid-connected mode by the same control scheme.

5.3 Regulation of the Key Variables

With the nature of given wind profile in Fig 5-1, the resultant rotor speed response is
illustrated in Fig. 5-4. During the time t  15 s, wind speed exceeds the rated value, so the rotor
speed remains at the rated value generating the rated power with the pitch control action. Fig.
5-5(a) illustrates the action of the pitch controller, in which the pitch angle varies with wind
speed to adjust the wind power conversion coefficient (CP) as shown in Fig. 5-5(b). This causes
the DFIG to generate the active power to its rated value, irrespective of the wind speed. Note that
higher the wind speed, higher will be the pitch angle and lower will be CP. After t  15 s , wind

106
speed beeing less thaan the ratedd value, the rrotor speed is regulatedd at its optim
mum value (illustrated
(
in Fig. 55-4). Duringg this period, the pitchh angle is drropped to zero degree and CP is kkept at the
maximuum value to capture thee maximum
m available power
p emplloying the M
MPPT technnique (see
Fig. 5-55(b)).

Fig. 5-44 Rotor speedd response.

(a)

(b)
Fiig. 5-5 Respoonses of pow
wer conversion coefficientt and pitch anngle.

107
 (a)

(b)
F 5-6 Respponse of outpput voltages. (a) SOFC annd ELZ term
Fig. minal voltagess, and (b) UC
C voltage.

W
With the variaation of the load and generation, ccurrent from
m the SOFC//ELZ also vvaries, and
their terrminal volttages (Velz,V
Vfc ) changee accordinggly (see Figg. 5-6(a)). IIt is notablle that by
selectingg a proper sstack size foor the SOFC
C and ELZ, the open cirrcuit voltagges are madee equal (at
320 V). This leadss to treat thee SOFC andd ELZ as a single unitt having a ccontinuous II-V curve.
Consequuently, a coommon conttrol schemee on DC2 ennables smoooth transitioon between the SOFC
and ELZ
Z modes. W
While regulaating Vdc, theere is inevittable deviatiion in the U
UC state-of-ccharge (or
its term
minal voltage) from its rated value (see Fig. 5-6(b)), w
which shouldd be restoreed for the
proper ooperation. T
The outerm
most controll loop of D
DC2 is guidded to nulliify the deviiation and
maintainn the UC vvoltage arouund 250 V. Due to the large valuee of Cuc, setttling time of
o the UC
voltage control loopp is quite laarge, and heence it is exppected to seettle at the rated value aafter some
time as discussed inn controller design (secction 4.3.1).

108
 Fig. 5-7 R
Responses of H2 and O2 ppartial pressuures in the SO
OFC.

Fig. 5-8 M
Molar flow annd H2 pressurre responses in the storagge tank.

109
 Fig. 5-7 depiccts the dynam
mics of reacctants’ partiial pressure.. Because off the dynam
mics in fuel
regulatoor and delayy in pressuree gain, the tiime constannt for partiall pressure dyynamics is large. The
pressuree is not a coontrolled varriable here, so it is requuired to limit the rise tiime and the ramp rate
of the cuurrent to connform the SOFC
S dynam
mics.
Thhe dynamiccs of molarr productionn/consumptiion rate off H2 in the ELZ/SOFC
C and the
pressuree of storagee tank is deepicted in F
Fig. 5-8. Inittially the laarge surpluss power is utilized
u to
producee H2, so the molar produuction resullting the risee in pressuree of H2. Thee rate of chaange of H2
pressuree is, howeveer, smaller thhan the mollar flows (beeing pressurre as integraation of the fuel flow).
After t=
=18 s, the cuurrent is trannsferred to the SOFC bby consuminng the H2 from
fr the tannk, thereby
reducingg pressure oof the tank.

Fig. 5-9 L
LVRT of the proposed syystem.

Onne of the nootable meritss of the propposed WPSS system liees in its low
w voltage ridde through
(LVRT)) capability,, which is thhe capabilitty to restoree the voltage when the voltage in the grid is
temporaarily reduced due to a ffault or loadd change in the grid. Thhe voltage oof the ac grrid may be
reducedd in one, tw
wo or all thhe three phaases. The seeverity of thhe voltage dip
d is definned by the
voltage level during the dip (m
may go dow
wn to zero) aand the durration of thee dip. In thee new grid
integratiion codes, it is required to stay connected during the fault (low voltage perriods) and
support the grid wiith reactive power to reestore the vvoltage leveel during thee post fault condition
VRT capability of the proposed hhybrid systeem. At t=1 s a 3φ to
[4],[5]. Fig. 5-9 shhows the LV
ground fault is invooked, whichh persists unntil 1.2 s, annd then it iss cleared. A
As shown, thhe voltage
restores at its rated value beccause the storage
s systtem effectivvely injects the desiredd reactive

110
power during the post fault period (t>1.2 s).
Some of the operation scheme of the DESS explained in this section may be compared with
the results reported in [6], [2]. In [6], the FC/ELZ/UC are used as the storage system but has a
different system topology, in which the FC and ELZ are both linked to the regulated dc-bus via a
separate dc-dc converter for each. Moreover, the power sharing scheme among the FC, ELZ and
UC is not compatible with the dynamics of the FC, because the situation of fuel starvation is not
considered clearly. For example, the FC (ELZ) has to provide transient part too, even if the UC
voltage (Vuc) is below (above) the working range. In case of [2], the FC/UC are used as the
storage system to manage the fluctuation, and the FC dynamics is considered in power sharing.
However, this configuration is not applicable if a power generator source (like photovoltaic or
wind power) is connected at the dc-link bus, because it cannot handle the power surplus. Further,
the configuration needs the UC of higher voltage rating (thereby necessitating more cells in the
stack), and a separated converter may be needed if the ELZ is adopted.
These results show that the subsystems have depicted successful performance of meeting
the requirements. As desired, the WPSS provides the solution of short to long-term power
management in the islanded and grid-connected modes, and demonstrates an evolution in terms
of the power and energy management with reference to the previous works (citations [22]-[28] of
chapter one). The WTGS captures optimum power from wind and transfer it to the dc and ac
buses, the DESS serves as the dedicated energy balance for wider time-scale, and the line-side
converter system regulates the frequency, power and voltage seamlessly in the various operating
circumstances, validating the applicability of the scheme. In the proposed scheme, the wind
power is taken as the disturbance input, in which the disturbances are subjected to the ac bus and
the dc bus. The energy storage system and the line-side converter are jointly employed to reject
the disturbances, in the form of load or intermittent power, at the dc and ac buses. Thus, this
concept is equally applicable to the other configurations; such as full-scale power electronics
converter based WTGS, fixed-speed WTGS and photo voltaic based microgrid. Further, the
control and operational approaches can also be utilized in the system having different ratings of
generators/loads because the control and operation laws were deducted in per unit base. The
major limitation, however, to implement the scheme in a real system is the cost-effectiveness of
the proposed storage technologies. With dynamic research and advancement in this field, it is
expected that the economy would be favorable in the future.

111
Reference

[1] Mathworks-MATLAB and Simulink for technical computing (Sep. 2009), available
online: http://www.mathworks.com.
[2] P. Thounthong, S. Raël, and B. Davat, "Control strategy of fuel cell and supercapacitors
association for a distributed generation system”, IEEE Transactions on Industrial
Electronics, vol. 54(6), pp. 3225 - 3233, Dec. 2007.
[3] A. Yazdani, "Islanded operation of a doubly-fed induction generator (DFIG) wind-power
system with integrated energy storage," IEEE Electrical Power Conference, pp. 153-159,
Canada, Oct. 2007.
[4] FERC LVRT standards, Available at: http://www.ferc.gov/industries/electric/indus-
act/gi/wind/appendix-G-lgia.doc
[5] C. Jauch, J. Matevosyan, T. Ackermann and S. Bolik , "International comparison of
requirements for connection of wind turbines to power systems," Wind energy, vol. 8, pp.
295-306, 2005.
[6] O.C. Onar, M. Uzunoglu and M.S. Alam, "Modeling, control and simulation of an
autonomous wind turbine/photovoltaic/fuel cell/ultracapacitor," Journal of Power
Sources, vol. 185(2), pp. 1273-1283, Dec. 2008.

112
Chapter 6 Power Sharing Scheme with Multiple
VSCs

6.1 Introduction

In the previous chapters, the proposed autonomous system had a single inverter system
connected to the grid or point of common coupling (PCC) bus. Therefore, the inverter is solely
responsible for the control of power flow and the terminal voltage. However, in a microgrid,
there is a possibility of having many inverters connected in parallel and linked with the utility
grid. In such cases, it is always desired to share the reactive and/or active power among the
inverters, if possible, without using the control information among them. The real and reactive
power controls should be independent with each other, and the inverters should share common
real and reactive load in proportion to a pre-determined ratio, regardless of plant parameters.

Fig. 6-1 Single line diagram of microgrid with multiple voltage source converters.

Fig. 6-1 shows the configuration of such microgrid, in which an additional controllable
power source is connected to bus#1 via VSC1’. Note that the configuation now possesses two
controllable power sources, unlike in Fig. 2.6(a) where VSC2 was the only one disspatchable
source. In this configuration, the dc voltage regulation is delegated to their respective energy

113
storage system. So, from the inverters’ perspective, the dc power/storage system can be replaced
by a constant dc source for the time period of interest. These voltage source converters (VSCs)
are expected to share the load and tie-line power (Ptie), if desired, according to their rating. There
are various approaches to design the scheme. The important classes of autonomous load-sharing
techniques that have been proposed to date are: (i) the frequency and voltage droop technique
[1],[2], termed as conventional droop scheme, (ii) P-Q sharing using communication between
inverters [3], (iii) the signal injection technique [4], and (iv) the adaptive impedance method [5].
Although the active power sharing is satisfactory with these schemes, neither of them could
satisfactorily address the reactive power sharing in a distribution network with a high R/X value.
Utilizing the insights gained from a detailed study of conventional droop techniques, this section
first investigates the applicability of conventional droop schemes in P-Q sharing with respect to
varying R/X, and proposes a new reactive power sharing scheme by including an idea of virtual
impedance. The new scheme ensures that the inverters on a distributed power network share a
common load regardless of the line impedance. Here, the VSCs are operating in an islanded
mode to share the load at the different buses. If the utility grid is to be connected, the power
exchange between the microgrid and the utility grid can be determined by the appropriate
management schemes; such as the tie-line regulation, as discussed in the section 4.4 of the
previous chapter.
Si  Pi  jQi

Zi
Ei  i Vi 0

Ii

Fig. 6-2 Volltage source connected to the local bus.

6.2 Applicability of Droop Regulation

The idea behind the conventional droop control is that the generators themselves can set
their instantaneous P-Q flow based upon the decoupled relationship between the active power
and the reacitive power, respectively with the frequency and voltage magnitude, so that the
power demand within a system can be shared among the generators in pre-specified manner.
However in low-voltage (LV) and medium-voltage (MV) microgrids, the line connection being
highly resistive, the decoupling is not possible, and hence the wireless droop scheme is quite

114
difficult. To get the clear picture of coupling nature of network variables, consider a system as in
Fig. 6-1, in which the active and reactive power supplied by the ith voltage source to the PCC bus
can be represented by

 EV V2  EV 
Pi   i i cos  i  i  cos i  i i sin i sin  i 
 Zi Zi  Zi 
, (6.1)
 EiVi Vi  EiVi
cos i sin  i 
2
Qi   cos  i   sin i 
 Zi Zi  Zi 
in which the variables and the parameters are defined in Fig. 4-2. Defining the control
variable as vi=Ei= Eicosi - Vi and  i sini we have,
 Pi  Vi  Ei sin i cos i   i 
Q     E cos  sin i  vi 
. (6.2)
 i  Zi  i i

Clearly, (6.2) depicts the fact that Pi-Qi is decoupled with i-vi only in ideal case; namely
pure resistive (i=0) or pure inductive (i =π/2) case. For the high X/R value, traditional droop
sharing scheme can be implemented, where the active power and the reactive power are
decoupled with the frequency and voltage magnitude, respectively. Conversely for a highly
resistive line coupling, the droop sharing is needed to be reversed. In practice, LV lines have R/X
ratio lying between 2 to 6. In such cases, neither of the above techniques fits well.
In order to find a proper control scheme, the applicability of conventional droop scheme
against different values of R/X is checked first; and based upon the limitations, the proper
scheme will be investigated. The frequency and the voltage droop derived from the conventional
scheme for ith VSC is written as [1]

i  0  mi ( Pi ref  Pi ) 

Vi  V0  ni Qi 
0  min 
with, mi  . (6.3)
Pi rated 
V0  Vmin 
ni  
Qmax ,i 

Here, V0 and 0 are the nominal values of voltage and frequency, respectively. Similarly, Pi ref ,

Qmax,i, min and Vmin represent the rated active power, maximum reactive power, frequency at
full load power and minimum load voltage of the system, respectively, of the ith inverter. The
coefficient mi and ni are the droop coeffiecients as defined by the equation. For a microgrid, mi

115
and ni are normally chosen as  2% and  5%, respectively, in per unit system [6].
The control law for the control variables (i, vi) can be written as,
d i 
 0  mi ( Pi ref  Pi ) 
dt . (6.4)

dvi
 (V0  Vi )  ni Qi 
dt 

Here, Vi represents the voltage at the point of ith common coupling. It has been reported that by
controlling vi, rather than Ei or Vi, the impedance mismatch of coupling reactance (Zi) is
minimized [7]. The block diagram of the droop regulation is also shown in Fig. 6-3.

Vi ref

(a)

i ref 1 i
ref
Pi
s

Pi 0

(b)
Fig. 6-3 Conventional droop schemes. (a) V/Q droop, (b) P/f droop
From (6.2) to (6.4) and performing the small signal analysis around the operating point, we
have

d i  Vi   mi Ei sin  i  mi cos  i  i 

    . (6.5)
dt  vi  Z i  ni Ei cos  i  ni sin  i   vi 

Here, the variables with ‘~’ represent the perturbed values, whereas the quantity with capital
letter are the values at the operating point. The characteristic equation of the system represented
by (6.5) can be written as

116
 (a)

(b)
Fig. 6-4 Poweer flow dyanaamics of VSC
Cs. (a) VSC11’ (b) VSC2.

117
 s 2  B1s  C1  00, 

Vi sin  i EiVi 2 . (6.6)
B1  ( Ei mi  ni ) andd C1  2 mi ni 
Zi Zi 
Sinnce B1 , C1  0 , the polees always siituate at the stable planne indicatingg that the sm
mall signal
model oof the system f any valuue of R/X. Thhe choice of mi and ni iis a trade-offf between
m is stable for
the stability and thee voltage reggulation.
Diiscussion : The abovve scheme is tested too the system
m shown inn Fig. 6-1, under the
MATLA
AB/Simulinkk platform [8].
[ The ratiings of VSC
C1’ and VSC2 are keptt 20 kVA annd 30 kVA
respectively. The ggraphs in Fiig. 6-4 show
ws the dynam
mics of thee power flow
w through tthe VSCs,
with diffferent R/X vvalues of linne impedancce. Initially a load of (115+j8) kVA was conneccted to the
bus#1 aand (5+j3) kkVA at bus #2.
# At t=0.22 s another step load of
o (8+j3)kVA
A is introduuced at the
bus #1. It can be seeen that P is apparentlyy regulated against diffferent R/X vvalues, wherreas the Q
is heaviily affected by the valuue. With thee increasingg value of R
R/X, the cirrculating Q increases,
though the voltage regulation is satisfactoory (Fig. 6--5). This phhenomenon of circulatinng current
due to thhe mismatchh of conneccting impedaance and higgh resistive nature of cconnecting impedance
are repoorted in [10] and [6]. The circulaating reactivve current iss not alloweed in the syystem as it
creates aan extra burrden to the ssystem. Hennce, the connventional ddroop schem
me is not appplicable in
the systeem with ressistance dom
minated line.

F 6-5 Voltaage response of the load bbus.

Fig.

118
6.3 Introduction of Virtual Impedance

While observing the power flow dynamics in Fig. 6-4, it can be noticed that the origin of the
circulating reactive power is due to the mismatch of impedance between the respective inverter
and the load bus. Larger the impedance, the smaller is the reactive power flow and vice-versa.
Thus, the the inverters with smaller interconnecting impedance supplies the circulating reactive
power to the inverter with larger interconnecting impedance.
To solve this, a large interface inductors can be included between the inverter and the PCC
bus, but they are heavy and bulky. Resistive compensation can be made by adding its effect in
setting the reference voltage in [6]. However, the amount of resistive part to be compensated is
related to the location of load change, which being the random event, can not be predicted in
multi load bus system. In [5], authors have proposed the concept of adaptive impedance which
varies with the reactive power flow. However, the model is not appealing for the predominately
resistive line as it compensates mainly the reactance part. Moreover, the effective impedance
sought by the VSCs is susceptible with the current transient. This study tries to address these
limitations by introducing the virtual impedance which varies with reactive power as

z Ei  ( z 0  k1i Qi ) , (6.7)
where z0 is the reference impedance and the k1i is the constant which determines the sensitivity
of Qi over zEi. Both parameters are carefully selected depending upon the network and
operational circumstances. This impedance compensates the reactive power difference due to
line impedance unbalance by providing the proper reference voltage given by

Vi  Vo  ni' Pi  ( z0i  k1i Qi ) I i . (6.8)

Vi ref

Fig. 6-6 Block diagram of proposed voltage and reactive power regulation scheme.

119
Here, ni’ is the constant coefficient to decouple the effect of active power on Vi.. It is worthwhile
to note that the voltage regulation in (6.8) has sought two important considerations: (i) regulation
of virtual impedance, (ii) decoupling of the effect of Pi over Vi. Thus, the set of differential
equations to represent the Pi-Qi regulation evolves from (6.4) to (6.9), given by

d i 
 0  mi ( Pi ref  Pi ) 
dt
. (6.9)
dvi
 (V0  Vi )  n Pi  ( z0  k1i Qi ) I i 
,

dt 

Note that, though the active power is also affected by the output impedance, it is effectively
controlled by self-adjusting of the phase angle alone. The block diagram of the proposed scheme
is also shown in Fig. 6-6.

6.4 Small Signal Modeling and Control Design Rules

In order to investigate the stability and the transient response of the system, a small-signal
analysis is performed. The closed-loop system dynamics has been derived, taking into acccount
of well known stiff load bus approximation [9]. The small-signal dynamics of the phase angle
difference and the voltage difference (i,vi) are obtained by linearizing (6.2) and (6.9) at =45o.
This is fairly reasonable for the small-signal anlysis and controller gain estimation of the
considered microgrid with the power network of R/X>1. Although the complete modeling
requires the several low-pass filters with signals Pi, Qi, and Ii, the low-pass filters are excluded
from the small signal analysis, without loosing the accucary of the result. It is because the filter
will not cause any considerable effect on the closed loop dynamics if the corner frequency of
low-pass filter is selected quite higher than the dominant poles. The modified dynamic equations
in pertubed form becomes
d i 
  mi p i 
dt
. (6.10)
dvi 
 ni , p i  ( z0ii  k1qi I i  k1i Qi ii )
dt 

Noting that
Si 2 
Ii   ( Pi 2  Qi2 ) / Vi , i.e. ii   Pi p i  Qi qi  ,
Vi SiVi 
and substituting the values from (6.2) to (6.10) we get,

120
 d i 
  mi p i 
dt 
.
dvi
  p i   qi 
dt 
Simplifying, we get

d i  Vi  mi Ei mi  i 
     ,. (6.11)
dt  vi  Z i 2  (   ) Ei      vi 

where
 2Pi 
  ni ,  ( z0  k1i Qi ) 
 SiVi 
1 2Qi 
 ( z0  k1i Qi )  k1i Si  .
Vi  Si 
Thus, the characteristic equation of the system becomes,
s 2  B2 s  C2  0 , (6.12)
with

Vi k1 Si  P  Qi   
B2  mi Ei  ni   2 i   z0  k1i Qi   
,

Z i 2  Vi  Si    .

Vi mi Ei  2Qi  
C2  ( z0  k1i Qi )  k1i Si  
Z12  Si  

The small signal stability of the system is ensured if B2, C2>0. For a VSC, the value of n’ is
comparable to mcos, implying that miEi>ni’. For the lagging reactive power, Qi is positive
implying that the stability can be achieved for any value of the parameter. But, when the inverter
is supplying the leading reactive power, Qi becomes negative. In that case, the stability is
granted if
z0  k1i Qmaxi  0 
. (6.13)
or , z0  k1i Qmax,i 

In this case, Qmax,i is the maximum leading reactive power supplied by the inverter. That would
make C2>0 because 2Qi  k1i Si2 for any operating conditions. Thus, it can be concluded that the

small signal stability is confirmed when (6.13) is satisfied.

121
6.5 Simulation Results

The proposed scheme is tested on the MATLAB/Simulink model of the microgrid shown in
Fig. 6-1. The loads are assumed to be three phase balanced. The main aim of the scheme is to
regulate the reactive power, against the different values of R/X, as the conventional scheme is
only able to regulate the active power. It should be noted that the rating of inverters are chosen to
be same (20 kVA each), which is not necessary in general (the rating of inverters only need to
alter the values of mi,ni). Fig. 6-7 shows the power flow dynamics through the VSCs when the
line impedance is changed. Initially the load of (20+j8) kVA at the bus #1 and (5+j3) kVA bus #2
are shared by the VSCs. At t=0.2 s, another load of (5+j3) kVA is connected to bus #1, which is
followed by the injection of another load of (5+j3) kVA at bus #2 at t=0.35 s. It can be seen that
the active power and reactive power are shared between the inverter, proportional to their ratings,
in all conditions, irrespective of the different R/X value of the line impedance.
Similarly, the voltage dynamics curve shown in Fig. 6-8. indicates that load voltage
regulation is not affected by the R/X value due to the elimination of the circulating reactive
power. The deviation in the load voltage is regulated to the desired value by the secondary
voltgae control. Another notable merit of the scheme is that the reactive load sharing by the
VSCs can be made flexible. For example, it is not likely that the VSCs which are separated far
apart share the common reactive load equally, irrespective of the load location. It is expected that
the VSC near the load should share more reactive load than that of VSC situated in farther
distance, so that the line loss can be reduced (this concept is, however, not effective for the active
power sharing as there are many constraint in doing so). The proposed scheme can provide this
option by the choice of the of z0 and k1i. Fig. 6-9 shows how the individual VSC responds with
the load change in the nearest load bus. For the ease in analysis purpose, the equal ratings of
VSCs are selected again. Initially, when both buses were equally loaded, the power sharing
between the VSCs was equal. When a load of (5+j3) kVA is applied to bus #1, VSC1’ shares
more reactive power than VSC2. Again when another load of (5+j3) kVA is connected to bus #2,
VSC2 shares more reactive power than VSC1’. Thus, the converter situated nearer to the load
can be commanded to share more reactive load until they are overloaded. Active power sharing
remain, however, unaffected by the choice of k1 and z0, as required.

122
 (a)

(b)
Fig. 6-77 Power flow
w dynamics of
o proposed scheme. (a) VSC1’, and (b)VSC2.

123
 Fig. 6-8 Voltage dynamics off proposed sccheme.

Fig. 6-9 Activve and reactivve power shaaring.

P
Part of the above resuult may be compared with the sttudy reported in [5], where
w the
adaptivee droop conncept is propposed to deecouple the mismatch oof coupling impedancee in power
sharing without communicatioon lines. In the cited reeport, the m
main focus was on the inductive
mismatcch which iss compensatted by the adaptive inductance appplied to a single phasse system.
This stuudy, howeveer, focuses on
o overall iimpedance tto be applieed on three--phase systeem, which

124
demonstrates a more comprehensive reactive power scheme. For example, the converter situated
nearer to the load can be commanded to share more reactive load. Although the study was made
for two inverter system, it can be applied to the microgrid system with any number of inverter
connected to the dedicated energy storage. The scheme also works with inverter connecting the
non-dispatcheble DGs, in which case reactive power can be shared as proposed above. The
active power, in such case, is kept unregulated (in fact captured maximum available power as
expalined in previous chapters) and the grid frequency is imposed to the system by a phase
locked loop . Thus, we can operate both type of inverter connected systems, connected to the
energy storage devices and intermittent renewable DGs, to form an autonomous distribution
network.

6.6 Summary

In this chapter a modified power-sharing scheme applicable for parallel operating VSCs in
a distribution microgrid has been proposed. The scheme was based on the droop regulation,
which did not necessitate the communication channel among the VSCs. In contrast to
conventional droop schemes, which are only applicable to line with X>>R, it can provide proper
power sharing in distribution microgrid (where R/X>2 in general). Based on the fact that line
parameter (R/X) negligibly affects the active power sharing and highly affects the reactive power
sharing, the scheme introduces virtual impedance in the Q-V d roop keeping the P-f d roop
unchanged. This allows the sharing of active and reactive power without sacrificing
frequency/amplitude regulation in the steady-state. Another notable merit of the scheme is that it
provides the flexible reactive power sharing among the VSCs so that the VSCs near the load can
share more var if required.

Reference

[1] R. Lasseter and P. Paigi, "Microgrid: A conceptual solution," IEEE 35th Annual Power
Electronics Specialists Conference, vol. 6, pp. 4285-4290, 2004.
[2] J. A. Peças Lopes, C. L. Moreira, and A. G. Madureira, "Defining control strategies for
microgrids islanded operation," IEE E Tr ansactions on P ower S ystems, vol. 21(2), pp.
916-924, 2006.

125
[3] M.N. Marwali, J. Jung and A. Keyhani, "Control of distributed generation systems-Part
II: Load sharing control," IEEE Transactions on Power Electronics, vol. 19(6), pp. 1551-
1561, 2004.
[4] A. Tuladhar, H. Jin, T. Unger and K. Mauch, "Control of parallel inverters in distributed
AC power systems with consideration of line impedance effect," IEEE Transactions on
Industry Applications, vol. 36(1), pp. 131-138, 2002.
[5] J. Guerrero, J. Matas, L. Vicuña, M. Castilla and J. Miret, "Wireless control strategy for
parallel operation of distributed-generation inverters," IEEE Transactions o n I ndustrial
Electronics, vol. 53(5), pp. 1461-1470, 2006.
[6] A. Engler, "Applicability of droops in low voltage grids," International J ournal of
Distributed Energy Resources, vol. 1(1), pp. 3-15, Sep. 2005.
[7] C. Sao and P. Lehn, "Autonomous load sharing of voltage source converters," IEEE
Transactions on Power Delivery, vol. 20(2), pp. 1009-1016, 2005.
[8] Mathworks-MATLAB and Simulink for technical computing (Sep. 2009), available
online: http://www.mathworks.com.
[9] E. Coelho, P. Cortizo and P. Garcia, "Small-signal stability for parallel-connected
inverters in stand-alone AC supply systems," IEEE Transactions on Ind ustry
Applications, vol. 38(2), pp. 533-542, 2002.
[10] L. Chen, X. Xiao, C. Gong and Y. Yan, "Circulating current’s characteristics analysis and
the control strategy of parallel system based on double close loop controlled VSI, " Proc.
IEEE PESC, pp. 4791–4797, 2004.

126
Chapter 7 Conclusion and the Future Course

This d issertation h as dev eloped the op erational a nd c ontrol schem es fo r the doubly-fed
induction generator based wi nd tu rbine ge neration s ystem inc orporating a dedicated e nergy
storage m ix. T he st orage mix h as consisted of a solid oxide f uel ce ll, a n e lectrolyzer a nd a n
ultracapacitor, in which the fue l cell and the eletrolyzer are confi gured i n a wa y to uti lize
minimum power electronic converters. The m ain focuses of the dissertation were on the control
and operational aspects of the dedicated energy storage system (DESS) and the grid or line-side
converter, so that all the subsystems would fit for overall power management approach.
In order to in vestigate the performance, the system c onfiguration an d the c ontrol scheme
have be en p resented with the linearized models of various subsystems. Based o n th e dyn amic
component models, a simulation model for the proposed wind power and storage system (WPSS)
has been d eveloped using MAT LAB/Simulink. T he detailed o peration stra tegy ha s b een
presented f or c oordinating the power f lows among the different energy s ources. Simulation
studies have bee n c arried out to verify the s ystem perf ormance u nder dif ferent ope rating
scenarios. Th e results show th at th e ov erall power m anagement strategy is ef fective, an d the
power flow among the diff erent energy sources and the l oad d emand is b alanced successfully.
The conclusions of the research work reported in this dissertation are summarized below.
Chapter two presented the layout and the unit sizing of the hybrid WPSS by investigating
the candidate distributed energy resources. We identified the suitable interconnecting topology in
the fo rm o f mix c oupling (ac a nd d c c oupling), so th at less number o f power electronic
converters could be used. An important feature of integrating topology was to accommodate the
fuel cell a nd th e elec trolyzer into a sing le dc-dc converter. Consequently, th e WPSS was
configured as an autonomous local power network or microgrid, which was capable to operate in
the islanded and dispatchable grid-connected modes.
Chapter th ree i nvestigated the prevailing d ynamic models of t he i ndividual components.
We explored the proper mathematical models relevant to the scope of the research. To reduce the
degree of complexity and manage the computational time, few simplifications were made in the
mathematical model–such as lu mping of t he tur bine and th e g enerator bo dy, one dimensional
treatment of th e fu el c ell e xcluding th e th ermal dy namic a nd average modeling of swit ching

127
elements–without jeopardizing the accuracy of the results. Based on the numerical computation,
major components were te sted and compared with simil ar models d eveloped by t he earlier
researchers.
In chapter four, the control and operational schemes of the WPSS were developed based on
the mathematical models developed in chapter three. While designing the control parameters, the
models are linearized around their operating points, wherever applicable. The designated control
schemes for the s ubsystems were: regulating rotor speed to capture optimum power from wind,
employing the DESS to inject two time-scale currents from the storage devices, and enabling the
line-side converter to operate WP SS in different ope rating modes seamlessly. O ne of the
important features of the control approach is that the three controllers (the wind turbine generator
system controller, the DESS controller and the line-side converter controller) were decoupled. It
means th at each sub system u tilizes own termin al signa ls as th e fe edback, treating si gnals fro m
the o ther s ubsystems a s th e disturbance i nputs. T his a voids the inf luence of mismatching and
delay caused from the outside. Further, the test s of controllers against the parameters variations
demonstrated their robustn ess in tracking th e comman d inputs and rejecting disturbance in puts
with a wide stability margin.
Chapter five presented th e numerical simu lation of th e ov erall sy stem. The simulation
results depicted that the subsystems operated in accordance with the theoretical model developed
in c hapter f our. T he de dicated e nergy s torage s ystem demonstrated the c oordinated c ontrol
scheme, in which the c ompensating p ower ( to balance the supply and the demand) h as been
shared between the fuel cell/electrolyzer and the ultracapacitor. The fuel cell/electrolyzer subunit
of the DESS p rovides the steady-state b alance an d th e u ltracapacitor act s a s a bu ffer for th e
transient power. Utilizing H2 a s the sto rage me dium, t he fu el ce ll and the ultracapacitor were
operated to supply the designated power during the wind power deficit, whereas the electrolyzer
and the ultr acapacitor would came into effect during the power surplus. The c rucial part of the
controller design was to consider the slow dynamics of the reactants’ partial pressure and the fuel
regulation parts, so that the fast transients would not pass through the fuel cell/electrolyzer and
cause the fuel starvation. Facilitated by functionalities of the DESS, th e operational scheme of
the line-side converter enabled the WPSS to operate as a dispatchable unit and a self-sustainable
unit in th e grid-connected and th e island ed mo des, re spectively. In th e absence o f t he sto rage
system, the WPSS could also be operated in the non-dispatchable grid-connected mode similar to

128
the conv entional grid-connected wi nd turbine g enerator sy stem. Th e sc heme also provided
seamless transition among the different modes without necessitating an additional controller for
each mode.
Chapter six soug ht the im proved power sh aring schem es in the case wh en the re are
multiple inverters operated in parallel. The proposed sche me en abled t o sh are reac tive po wer
effectively among the inverters in the distribution network having high resistance to inductance
ratio.
In this study, the wind power is taken as the disturbance input, in which the disturbances
are subjected to the ac and dc buses. The energy storage system and the line-side converter were
jointly employed to reject the disturbance, in the form of load or intermittent power, at the dc and
the ac buses. Thus, this concept is e qually applicable to the other topologies, such as fu ll-scale
power electronic converter based wind turbines, fixed-speed wind turbines and the photo voltaic
power based microgrid.
Based upon t he abov e findings, the p rospective research m ay b e sugg ested to ward the
different directions. One important part of future study is in the validation of proposed schemes
to the real system. The analysis in th e di ssertation was th e theo retical one, wh ich wa s tested
through t he numerical model. T herefore, it is ne cessary t o c onfirm the obtained res ults
experimentally so that they could be applied in the practical situations. Another direction of the
future study is on the u nit sizing. Wh ile designing t he pr oposed s cheme, the unit sizing was
based on the technical viewpoint of fulfilling the operating requirements. To decide the proper
size of the components, it is ad visable to take ac count o f seve ral operating constraints a nd
objective functions. So, the resulting size would be the solution of multi-objective optimization.
Likewise, th e m ain operating sc hemes of t he proposed hybrid m icrogrid were f ocused on the
primary and secondary le vels control, such as the isla nded mode or the dispatchable gri d-
connected mode. I n th is system, a tertiary le vel or s upervisory control ca n be im plemented,
which takes care o f the online electricity tariff, weather fo recast and th e state o f the energy
storage to em ploy the demand response phenomena. Su ch a comprehensive a nd autonomous
local n etwork structure ca n pr ovide a substantial contribution in realiz ing th e fu ture “sma rter”
power network.

129
130
Appendix

Table A.1 Comparison of storage technologies [Barton et al., “Energy storage and its use with
intermittent renewable energy,” IEEE Trans. Energy Convers., vol. 19(2), pp. 441–8, Jun. 2004].

The Parameters Used in the Study

Table A.2 Electrical network parameters
Parameter value unit
Vdc 750 V
V2 460 V
ω0 377 rad/s

Table A.3 Wind turbine parameters (Ref. [4], [5] of chapter 3)

Parameter value unit
P rated 40 kW
WT
kopt 0.0075 W/(rad/s)3
vw,cutin 4 m/s
vw,cutout 25 m/s
ρ 1.205 kg/m3
CP-max 0.47

131
 vwrated 12 m/s
ci (i=1,2..8) 0.73, 151, 0.58, 0.002, 2.14, 13.2, 18.4,
-0.02, -0.003
rT 15 m

Table A.4 Induction generator parameters

Parameter value unit
Rated Power (Srated) 50 kVA
Rated line voltage (Vs) 460 V
Lls 0.057 pu
rs 0.0175 pu
Llr 0.057 pu
rr 0.01029 pu
Lm 2.019 pu
Jm 0.2 pu
Dm 0.05 pu
np 2

Table A.5 Parameters for Phoebus alkaline electrolyzer (Ref. [22] of chapter 3)
Parameter value unit
Aelz 0.1 m2
f1 250
f2 0.96
Nelz 270
Pelzrated 50 kW
r1 8.05×10-5 Ωm2
r2 2.5×10-7 Ωm2 °C
u0 1.21 Volt
u1 0.185 V
Telz 80 °C
t1 -1.002 A-1m2
t2 8.424 A-1m2 °C
t3 247.3 A-1m2 °C 2
z 2

Table A.6 Parameters for SOFC (Ref. [18] of chapter 3)

Parameter value unit
Aelecyt 200×10-4 m2
Ainterc 45×10-4 m2
E0,cell 1.18 V
F 96484.6 C/mol
Ilimit 160 A
Ncell 95
Pa 3 atm
Pc 3 atm
Pfcrated 5 kW
R 8.314 J mol-1K-1

132
 μf 0.85
Tfc 1173 K
Va 61.7×10-6 m3
Vc 99.02×10-6 m3
kE 3.78×10-4
z 2
0 0.15 V
ac 53.5041
bc 65000
 elecyt 4.0866×10-4
 elecyt 6900
 elecyt 40×10-6
 interc 0.0175
interc 3126.7
 interc 40×10-6

Table A.7 Parameters for the ultracapacitor (Ref. [28] of chapter 3)

Parameter value unit
C 110 F
Ileakage 1.5 mA
Imax 30 A
Rsr 5.5 mΩ
Vmax 16.2 V
UC stack
Cuc 32 F
Nsr 17
Npl 5
Prated 40 kW

Table A.8 Parameters for WTGS controller

Parameter value unit
 rated 1.5 Weber
s
GC1(s) 0.1582+9.102/s
α1c 0.01
α2c 0.7
γ2c 1.35
GC2(s) 1.3786+1.0204/s
Ti,pit 0.5 s
Td,pit 0.8 s

Table A.9 Parameters for the fuel cell side dc-dc converter (DC1) controller
Parameter value unit
V ref 275 V
fc
D2,D3 0.6

133
 I dcrated 100 A
1
Li 1 mF
Ri 0.1 mΩ
Td 5 s
GC3(s) 0.00256+0.00535/s
GC4(s) 10+0.0084/s+0.036s/(s+0.36)
Vucref 250 V

Table A.10 Parameters for isolating DC2 controller

Parameter value unit
V ref 750 V
dc
GC5(s) 1.98745+14.037/s
Lt 0.1 mH
fsw 2 kHz
Cdc 5 mF

Table A.11 Parameters for line-side converter controller

Parameter value unit
V0 460 V
Lf 0.2 mH
L’ 0.248 mH
Rf 0.0063 Ω
R’ 0.01 Ω
GC6(s) 0.08+2.5264/s
GC7(s) 40/s(0.0025s+1)/( 0.248s +0.01)
kqt 0.1
kω 0.5
kdc 0.11
m 20
GC8(s) 0.1+1.23/s
GC9(s) 0.1+1.23/s

Table A.12 Parameters for multiple voltage source converter system controller
Parameter value unit
k11 10-5
k12 10-5
m1  / 20  10 3
m2  / 30  10 3
n1 0.2  10 3
n2 0.3  10 3
z01 0.02 Ω
z02 0.02 Ω

134
List of Symbols

Wind data
Ag Amplitude of the wind speed gust
Arm Amplitude of the wind speed ramp
Aw Rotor sweep area by wind
c Scale parameter for wind speed
fw Frequency of wind speed
h Tower height
k Shape parameter for wind speed
l Turbulence length
PDt Power spectral density of wind speed
Pwind Power from wind
Teg End time of wind speed gust
Ter End time of wind speed ramp
Tsg Starting time of wind speed gust
Tsr Starting time of wind speed ramp
vw Wind speed
vwa Average wind speed
vwg Gust component of wind speed
vwm Wind speed corresponding to maximum energy capture
vwr Ramp component of wind speed
vwt Turbulence component of wind speed
Wmax Maximum energy produced at vwm
 Roughness length
ρ Density of air

Wind turbine
ci Parameters for the wind turbine (i=1,2..8)
CP Wind power conversion coefficient
CP-max Maximum value of CP
PWT Mechanical power at the turbine shaft
rT Rotor radius
Tm Mechanical torque at the turbine shaft
β Pitch angle
λtip Tip speed ratio
tipopt Optimum tip speed ratio
ωr Rotor speed

135
ropt Optimum rotor speed

Doubly-fed induction generator system (IG, VSC1, VSC2 and dc-link bus)
Cdc Capacitance of the dc-link terminal
Dm Equivalent frictional coefficient
d1 Duty cycle space vector of rotor-side converter (VSC1)
d2 Duty cycle space vector of line-side converter (VSC2)
E2 Terminal rms voltage of line-side converter (VSC2)
e2 Terminal voltage space vector of line-side converter (VSC2)
I2 rms current of bus #2
i2 Current space vector of bus #2
imo Magnetizing current state vector
ir Rotor current space vector
is Stator current space vector
I2,dc Net dc current drawn from the line-side converter
Ir,dc Net dc current injected from the rotor-side converter
Isto Net dc current injected from the storage system
Jm Equivalent moment of inertia
Lf Inductance of coupling inductor
Lls Stator leakage inductance
Llr Rotor leakage inductance
Lm Mutual inductance
Lr Rotor self-inductance
Ls Stator self-inductance
Lth Thevenin inductance
np Number of pole pair
Pig Net active power from the doubly-fed induction generator
Pr Active power drawn from the rotor circuit of the doubly-fed induction generator
Ps Active power drawn from the stator circuit of the doubly-fed induction generator
Qig Net reactive power from the doubly-fed induction generator
Qs Reactive power from drawn from the stator terminal
Rdc Parallel resistance of the dc-link capacitor
Rf Resistance of the coupling resistor
Rth Thevenin resistance
rr Rotor resistance
rs Stator resistance
sl Slip of induction generator
Tem Electromagnetic torque
Tm Mechanical torque at the shaft
V2 Voltage of bus #2 in rms
Vdc dc-link voltage

136
v1 Voltage space vector of bus #1
v2 Voltage space vector of bus #2
vr Rotor voltage space vector
vs Stator voltage space vector
ψs Stator flux space vector
ψr Rotor flux space vector
ωe Angular frequency of stator voltage

Fuel cell
ac,bc Empirical parameters for activation voltage drop
Cfc Double layer capacitance
Ecell Nernst voltage
E0,cell Gibbs potential
F Faraday constant
i0 Exchange current density
Ifc Fuel cell current
Ilimit Limiting current density
kE Emprical constant
Ma Molar value at anode
Mc Molar value at cathode
Mx Molar flow of species x (x = H2, O2 or H2O)
Ncell Number of cell in the stack
Pfc Power from the SOFC
px Partial pressure of species x (x = H2, O2 or H2O)
R Universal gas constant
Rohm,cell equivalent resistances representing ohmic voltage drop
Ract,cell equivalent resistances representing activation voltage drop
Rconc,cell equivalent resistances representing concentration voltage drop
rH 2O hydrogen-oxygen flow ratio
Td Time constant for fuel regulation
Tfc Fuel cell temperature
Ua,Uc Volume of anode and cathode part
μf Fuel utilization ratio
Vact 0,cell Activation voltage drop affected by the temperature
Vact1,cell Activation voltage drop affected by the temperature and current
Vact,cell Fuel cell activation potential
Vcon,cell Fuel cell concentration potential
Vfc Fuel cell terminal voltage
Vohm,cell Ohmic potential drop

137
 αi,bi, δi Parameters for resistive voltage drop in the electrolyte of the solid
oxide fuel cell (i=electrolyte, interconnection)
z Number of electron participating in the chemical reaction
0 Constant term of activation voltage drop
1 Temperature coefficient the activation voltage drop

Electrolyzer and hydrogen production

Aelz Electrodes’ surface area

Ielz Electrolyzer current
Jelz Electrolyzer current density
MH2,out Hydrogen outflow rate
MH2,pro Hydrogen production rate
Nelz Number of cell in a stack
pelz,H2 Hydrogen pressure of at cathode
ri Parameters of ohmic resistance (i=1,2)
Telz Temperature of the electrolyzer
ti Parameter of overvoltage ( i=1,2,3)
u0 Thermodynamic cell voltage
u1 Parameter of overvoltage
ucell Terminal voltage across one electrolyzer
Velz Net terminal voltage across the electrolyzer stack
Uelz Volume of the cathode

Compressor and H2 storage

Pcomp Compressor power
pelz , H 2 Hydrogen pressure of at cathode
ptank Pressure of storage tank
Ttank Temperature of storage tank
Utank Volume of storage tank
w Polytropic work
αcom Compression efficiency
 Polytropic coefficient

Ultracapacitor and dc-dc converters

Cuc Capacitance of the ultracapacitor
di Duty cycle for ith switch in the converter fuel cell side dc-dc converter
( i=1,2,3)
Euc Energy available from the ultracapacitor
fsw Switching frequency of isolating dc-dc converter (DC2)

138
Idc1 Output current from the fuel cell side dc-dc converter (DC1)
Idstb Net disturbance current imposed to the dedicated energy storage system
Ii Input current injected to the isolating dc-dc converter (DC2)
Isto Net current injected from the dedicated energy storage system (DESS)
Li Output inductance of the fuel cell side dc-dc converter (DC1)
Lt Inductance of isolating transformer
Ri Output resistance of the fuel cell side dc-dc converter
Rsr Equivalent series resistance of the ultracapacitor
Rpl Equivalent parallel resistance of the ultracapacitor
Vf Final voltage for the ultracapacitor
Vi Initial voltage for the ultracapacitor
Vuc Voltage across the ultracapacitor terminal
δ Phase shift in the isolating dc-dc converter ( DC2)

Control and operational scheme

C(s) Transfer function of internal model controller
d Disturbance signal
G(s) Plant transfer function
GCi(s) Transfer function of ith PID controller, (i=1,2..9)
GCP(s) Transfer function of pitch control
Gd(s) Disturbance transfer function
F(s) Transfer function equivalent proportional-integral-derivative controller
Idstb Total disturbance current to the dedicated energy storage system
kiP, kiI Coefficient of proportional and integral control of converter GCi ( s ) (i=1,2..8)
kdc Parameter for dc voltage control
kM Parameter for the first order plant
kpwmi Pulse width modulation gain
ktie Parameter for tie-line active power control
kqt Parameter for tie-line reactive power control
L(s) Filter transfer function
m Droop coefficient
n Integer quantity
P2 Active power flow through the line-side voltage source converter (VSC2)
Ptie Tie-line active power flow
Qtie Tie-line reactive power flow
r Command input signal
ss Status signal
Tg Coefficient of grid power transfer
u Manipulating variable
y Output signal
α Filter parameter for the internal model control
α1c Parameters of internal model control for GC1 ( s)
α2c Parameter of the internal model control for GC 2 ( s )

139
 α6c Parameter of the internal model control for GC 6 ( s )
θf Angle of rotating stator flux
τp Time constant of the first order plant
τtg Time constant of turbine generator lump system
τpit Time constant of pitch angle control loop
γ Parameter for the improved internal model control
γ 2c Parameter of the internal model control for GC 2 ( s )
ropt Optimum rotor angular velocity
ω0 Nominal value of ac angular frequency
ωs Slip speed

Microgrid with multiple VSCs

δi Phase difference of phase between the ith inverter terminal and ith PCC voltage
ei Difference of magnitudes between ith inverter terminal and ith PCC voltage
Ei Voltage of ith inverter terminal
Ii rms value of the current injected from the ith inverter
mi P/f droop of ith inverter
ni Q/V droop of ith inverter
Pi Active power flow from ith inverter terminal
Qi Reactive power flow from ith inverter terminal
Qmax,i Maximum reactive power supplied by the ith inverter.
Si Apparent power flow from ith inverter terminal
Vi Voltage of ith PCC bus
ZEi Virtual output impedance of ith inverter
Zi Impedance of ith coupling inductor

Subscripts
r Quantity related to the rotor circuit of the doubly-fed induction generator
s Quantity related to the stator circuit of the doubly-fed induction generator
d d axis quantity of the doubly-fed induction generator
q q axis quantity of the doubly-fed induction generator
dc Quantities related to the dc-link bus
elz Quantities related to the electrolyzer
fc Quantities related to solid oxide fuel cell
uc Quantities related to ultracapacitor
Superscripts
opt Optimum value of the quantity
ref Reference value of the quantity
rated Rated value of the quantity
in Inflow rate of the quantity
out Out flow rate of the quantity

140
List of Acronyms

AE Alkaline electrolyzer
DESS Dedicated energy storage system
DFIG Double fed induction generator
DER Distributed energy resource
DG Distributed generators
ELZ Electrolyzer
ESS Energy storage system
DC1 Electrolyzer and fuel cell and side dc-dc converter
DC2 dc-link bus side isolating dc-dc converter
DESS Dedicated energy storage system
FC Fuel cell
FES Flywheel energy storage system
IG Induction generator
IMC Internal model controller
LVRT Low voltage ride through
LPN Local power network
LV Low voltage
MIMO Multiple input multiple output
MPPT Maximum power point tracking
MV Medium voltage
PCC Point of common coupling
PE Power electronics
PEC Power electronic converter
PI Proportional-integral control
PID Proportional-integral-derivative control
PV Photo voltaic
PWM Pulse width modulation
SISO Single input single output
SMES Superconducting magnetic energy storage
SOFC Solid oxide fuel cell
TSR Tip speed ratio
UC Ultracapacitor
VSC Voltage source converter
VSC1 Rotor-side converter
VSC2 Line-side converter

141
WPSS Wind power and storage system
WT Wind turbine
WTGS Wind turbine generator system

142