Amended Thesis Wang Yu - Fianl

This document is downloaded from DR‑NTU (https://dr.ntu.edu.
sg)
Nanyang Technological University, Singapore.
Application and control of energy storage systems

in smart grid
Wang, Yu
2016
Wang, Y. (2016). Application and control of energy storage systems in smart grid. Doctoral
thesis, Nanyang Technological University, Singapore.
http://hdl.handle.net/10356/69420
https://doi.org/10.32657/10356/69420
Downloaded on 08 Feb 2024 17:20:30 SGT

APPLICATION AND CONTROL OF ENERGY
STORAGE SYSTEMS IN SMART GRID
WANG YU
INTERDISCIPLINARY GRADUATE SCHOOL

ENERGY RESEARCH INSTITUTE @ NTU (ERI@N)
2016
APPLICATION AND CONTROL OF ENERGY
STORAGE SYSTEMS IN SMART GRID
WANG YU
Interdisciplinary Graduate School

Energy Research Institute @ NTU (ERI@N)
A thesis submitted to the Nanyang Technological

University in partial fulfilment of the requirement for
the degree of
Doctor of Philosophy
2016
Statement of Originality
I hereby certify that the work embodied in this thesis is the result of original
research and has not been submitted for a higher degree to any other University
or Institution.
8/8/2016
................. ...........................
Date Wang Yu
Abstract
Abstract
Due to the depletion of traditional fossil fuels and the increase in air pollution and
greenhouse gas emissions, the demand on renewable energy in the world will
continuously grow. Recently, renewable energy sources and energy storage systems
(ESSs) based distributed generation (DG) units have been largely integrated into
distribution networks to deal with the problems caused by fossil fuels. Many power
conversion methods and the corresponding control strategies have been proposed
and investigated for the integration of DG units in conventional power systems.
Therefore, it is necessary to improve the current techniques and develop new control
and management techniques for the integration of various sources based DG units
into utility grid or microgrids for better utilization efficiency. This thesis presents
several control schemes for application of ESSs in existing power systems.
For distribution networks with high penetration of photovoltaics (PVs) and electric
vehicles (EVs), the distribution networks may encounter voltage rise/drop issues
during peak generation/load period. First, distributed ESSs are proposed to solve the
voltage rise/drop issues in distribution networks through a coordinated control
method. The coordinated control method is achieved by both distributed and
localized controls of multiple ESSs. The distributed control using consensus
algorithm regulates the feeder voltages within the required limits, while the localized
control regulates the real-time state of charge (SoC) of each ESS within the desired
SoC range. The entire control structure ensures voltage regulation while effective
utilization of storage capacity during daily operation under different weather
conditions.
Then, the first research work is extended that distributed electric vehicle chargers
(EVCs) are applied to deal with the voltage rise/drop issues in distribution networks.
i
Abstract
With the concept of vehicle-to-grid (V2G), plug-in EVs at EVCs can be served as
energy storage devices and utilized for voltage regulation. Therefore, a two-stage
coordination method of distributed EVCs is proposed to regulate the voltages in
distribution networks. In the stage one, the distributed voltage control is proposed to
coordinate each EVC for voltage regulation. In the stage two, algorithms for
estimating the V2G capacity of the plug-in EVs and determining the real and reactive
power outputs of EVCs are proposed, so that the voltage regulation is achieved
within the required operation limits. The entire control structure ensures that the
voltages along the distribution feeders are within the allowable voltage range through
coordinated control of the power output of each EVC during daily operation.
Finally, the application of distributed ESSs in an islanded ac microgrid is

investigated. In order to achieve power balancing and SoC management of the
microgrid, a hierarchical control scheme is proposed to coordinate all available PVs,
ESSs and loads in the system. The primary control is realized by a novel model
predictive control, which provides fast dynamic response and small steady-state error
for power inverters in the microgrid. The secondary control is performed by
consensus based distributed control, which provides distributed control signals for
primary controlled inverters and loads at each agent. With the proposed control
scheme, the power balancing and SoC management can be achieved by managing
the power output of each available unit in a hierarchical way. The proposed
methodologies and control algorithms throughout the research is verified in
MATLAB/Simulink under various system topologies and different types of PV and
load data.
ii
Acknowledgements
Acknowledgements
This thesis would not have been possible without the guidance and the help of several
individuals who in one way or another contributed and extended their valuable
assistance in the preparation and completion of this study.
First, I wish to express my utmost gratitude to my supervisor, Associate Professor

So Ping Lam, for his continuous insight, enthusiasm and encouragement throughout
the whole research program. His deep knowledge on electric power systems, power
quality and smart grids has been very valuable for my research. He has always been
my inspiration and role model as I hurdle all the obstacles in the completion of this
research program. I would also like to sincerely thank Associate Professor Chan Chi
Chui Julian. His professional attitude and support deserve special acknowledgment.
I am deeply impressed by their serious attitude and enthusiasm towards academic
research which will have positive effect on my future research and work.
Then, I would like to thank Tan Kuan Tak, Peng Xiaoyang, Thomas John and Wang
Benfei, who are my fellows. Tan Kuan Tak helps me a lot during my PhD studies.
As a senior student, he generously shares his experience on research and makes a
good example of a PhD student for me. Moreover, he patiently participates in writing
and revision of my papers. Peng Xiaoyang provides me assistance in learning and
understanding some control theories and operation principles of microgrids. Thomas
John works with me on the model predictive control of power inverters in microgrids.
Wang Benfei gives me many useful suggestions and comments during my PhD study.
I would also like to thank my group members for the in-depth and fruitful discussions
on the research work. They are Ravi Kishore, Nandha Kumar, Sivaneasan
Balakrishnan and Wang Zhe.
iii
Acknowledgements
Furthermore, I am grateful to Nanyang Technological University for providing me

an opportunity to pursue my graduate study with a research scholarship. I am also
thankful for the technical support provided by the laboratory staff Chia-Nge Tak
Heng, Lin Zhi Ren and Foo Mong Keow, Thomas in Clean Energy Research
Laboratory. They generously offer their help during my study in the laboratory. It is
their careful administration that makes the laboratory a comfortable and convenient
place for research.
Last but not least, I want to present my deepest appreciation to my family. Their
endless love and persistent encouragement are the most greatly sources of motivation.
iv
Table of Contents
Table of Contents
Abstract ..................................................................................................................... i
Acknowledgements ................................................................................................ iii
Table of Contents .................................................................................................... v
Table Captions ........................................................................................................ ix
Figure Captions ...................................................................................................... xi
Abbreviations ........................................................................................................ xv
Chapter 1 Introduction ...................................................................................... 1

1.1 Background and Motivations ..................................................................... 2
1.1.1 Background ................................................................................................ 2
1.1.2 Motivations ................................................................................................ 6
1.2 Objectives and Scope ................................................................................. 7
1.3 Contributions of Thesis .............................................................................. 9
1.4 Organization of Thesis ............................................................................. 12
References ............................................................................................................... 14
Chapter 2 Literature Review .......................................................................... 19

2.1 Energy Storage Systems ........................................................................... 20
2.1.1 Comparison of Energy Storage Technologies .......................................... 20
2.1.2 Energy Storage for Utility Grid ............................................................... 24
2.1.3 Energy Storage for Transportation........................................................... 28
v
Table of Contents
2.2 Voltage Regulation Methods in Distribution Networks ........................... 31

2.3 Control Hierarchy of Power Systems ....................................................... 36
2.3.1 Primary Control ....................................................................................... 37
2.3.2 Secondary Control ................................................................................... 40
2.3.3 Tertiary Control ....................................................................................... 44
2.4 Consensus Algorithm ............................................................................... 45
2.5 Summary .................................................................................................. 45
References ............................................................................................................... 50
Chapter 3 Coordinated Control of Distributed Energy Storage Systems for

Voltage Regulation in Distribution Networks .................................................... 63
3.1 Introduction .............................................................................................. 64
3.2 Problem Description and Analysis ........................................................... 64
3.3 Proposed Control Method ........................................................................ 67
3.3.1 Vanadium Redox Battery ......................................................................... 68
3.3.2 Distributed Control Method ..................................................................... 70
3.3.3 Localized Control Method ....................................................................... 73
3.4 Results and Discussions ........................................................................... 76
3.4.1 Case Study: 6-Bus Distribution Feeder.................................................... 76
3.4.2 Case Study: 13-Bus Distribution Network .............................................. 87
3.5 Conclusion ................................................................................................ 91
References ............................................................................................................... 92
Chapter 4 Two-stage Coordination of Distributed Electrical Vehicle Chargers

for Voltage Regulation in Distribution Networks............................................... 95
4.1 Introduction .............................................................................................. 96
4.2 Modeling and Assumptions ...................................................................... 97
4.2.1 Lithium-ion Battery Model ...................................................................... 97
vi
Table of Contents
4.2.2 Assumptions ............................................................................................. 99

4.3 Proposed Voltage Regulation Method ................................................... 100
4.3.1 Stage One: Distributed Voltage Control ................................................ 101
4.3.2 Stage Two: Localized Power Allocator .................................................. 104
4.4 Results and Discussions ......................................................................... 107
4.4.1 Test Case 1: Sunny PV and Light Load Condition ................................ 110
4.4.2 Test Case 2: Cloudy PV and Heavy Load Condition ............................. 114
4.5 Conclusion .............................................................................................. 118
References ............................................................................................................. 119
Chapter 5 Hierarchical Control and Power Management of Islanded

Microgrids with Energy Storage Systems and Renewable Energy Sources .. 121
5.1 Introduction ............................................................................................ 122
5.2 System Description ................................................................................ 122
5.2.1 System Configuration ............................................................................ 122
5.2.2 Structure and Control Scheme of ESS/PV ............................................. 124
5.3 Model Predictive Control of Voltage Source Inverters .......................... 125
5.3.1 Modeling of Voltage Source Inverters ................................................... 125
5.3.2 Proposed Primary Model Predictive Control ......................................... 127
5.4 Consensus Based Distributed Control .................................................... 129
5.4.1 Communication Network and Consensus Algorithm ............................ 129
5.4.2 Proposed Secondary Distributed Control............................................... 131
5.4.3 Distributed Signal Allocation ................................................................. 132
5.5 Results and Discussions ......................................................................... 135
5.5.1 Test Case 1: Short-Term PV Generation and Load Demand Variation .. 136
5.5.2 Test Case 2: Long-Term Power Imbalance and PV Curtailment ........... 139
5.5.3 Test Case 3: Long-Term Power Imbalance and Load Shedding ............ 142
5.6 Conclusion .............................................................................................. 144
vii
Table of Contents
References ............................................................................................................. 145
Chapter 6 Conclusions and Future Work .................................................. 147

6.1 Conclusions ............................................................................................ 148
6.2 Future Work ............................................................................................ 151
List of Publications ............................................................................................. 155
viii
Table Captions
Table Captions
Table 2.1 Characteristics of energy storage technologies [3] ........................... 22

Table 2.2 Charging power levels of EVCs [35] ............................................... 29
Table 2.3 Comparison of typical voltage regulation methods .......................... 33
Table 3.1 Parameters of the VRB based ESS .................................................. 69
Table 3.2 Parameters of the proposed control method..................................... 76
Table 3.3 Parameters of the simulation system................................................ 77
Table 4.1 Data fitting results of the lithium-ion battery………………...…….98
Table 4.2 Parameters of the simulation system .............................................. 108
Table 4.3 Usage data of distributed EVCs..................................................... 109
Table 5.1 Parameters of the simulation system……………..……………….135
Table 5.2 Parameters of the proposed control scheme................................... 136
ix
Table Captions
x
Figure Captions
Figure Captions
Figure 1.1 Renewable generation capacities in World, EU-28, BRICS and top
seven countries at the end of 2015 [9]. ............................................... 3
Figure 1.2 Classification of electrical energy storage technologies according to
energy form [17]. ............................................................................... 4
Figure 2.1 Different energy storage technologies for grid application [15] ....... 24
Figure 2.2 Control hierarchy of power electronic interfaced power system. ..... 37
Figure 2.3 Control architecture of centralized and distributed secondary control.
………………………………………………………………………41
Figure 3.1 A typical radial distribution feeder with multiple buses…………...64
Figure 3.2 Voltage magnitudes along radial distribution feeder. (a) Voltage rise.
(b) Voltage drop. ............................................................................... 66
Figure 3.3 The structure of proposed control method for customer at ith bus. .. 68
Figure 3.4 The OCV/SoC characteristic curve of a single VRB cell. ................ 69
Figure 3.5 The control scheme of the proposed distributed control. .................. 70
Figure 3.6 Communication graph for N-bus radial distribution feeder. ............. 72
Figure 3.7 Single line diagram of the 6-bus test feeder. ..................................... 77
Figure 3.8 Influence on feeder voltages of bus number and power injection. .. 77
Figure 3.9 PV generation, residential load demand and ESS power for test case
1……………………………………………………………………79
Figure 3.10 Power injection with/without ESSs for test case 1. .......................... 79
Figure 3.11 24-h voltage profiles without ESSs for a sunny day. ........................ 80
Figure 3.12 24-h voltage profiles with ESSs and proposed method for a sunny
day…………………………………………………………………80
Figure 3.13 SoC profiles for test case 1. .............................................................. 81
Figure 3.14 PV generation and residential load demand for test case 2. .............. 82
Figure 3.15 24-h voltage profiles without ESSs for a cloudy day. ....................... 83
xi
Figure Captions
Figure 3.16 24-h voltage profiles with ESSs and proposed method for a cloudy
day…………………………………………………………………..83
Figure 3.17 Power outputs of ESSs for test case 2. .............................................. 84
Figure 3.18 SoC profiles of ESSs for test case 2.................................................. 85
Figure 3.19 Voltage profiles with droop based method. ....................................... 86
Figure 3.20 Power curtailed with constant droop based method. ........................ 87
Figure 3.21 Configuration of the 13-bus residential distribution network. .......... 88
Figure 3.22 Communication graph for the 13-bus distribution feeder. ................ 88
Figure 3.23 24-hr voltage profiles without ESSs. ................................................ 89
Figure 3.24 24-hr voltage profiles with ESSs and proposed method. .................. 89
Figure 3.25 Power outputs of ESSs 4, 7, 10 and 12. ............................................ 90
Figure 3.26 SoC profiles of ESSs 4, 7, 10 and 12. ............................................... 90
Figure 4.1 The equivalent circuit model of the lithium-ion battery…………...98
Figure 4.2 The charging characteristic of the lithium-ion battery. ..................... 99
Figure 4.3 The discharging characteristic of the lithium-ion battery. ................ 99
Figure 4.4 Entire control structure of the proposed voltage regulation method.
101
Figure 4.5 Configuration of the 17-bus residential distribution network. ........ 108
Figure 4.6 PV generation and load profiles for simulation study. .................... 109
Figure 4.7 Voltage profiles with uncoordinated EV charging for test case 1. . 111
Figure 4.8 Voltage profiles with the proposed method for test case 1. ............ 111
Figure 4.9 Real and reactive power absorption and SoC profile of each EVC for
test case 1…………………………………………...………..…...112
Figure 4.10 Voltage profiles with uncoordinated EV charging for test case 2. .. 115
Figure 4.11 Voltage profiles with the proposed method for test case 2. ............ 115
Figure 4.12 Real and reactive power absorption and SoC profile of each EVC for
test case 2……………………………………..…………………...116
Figure 5.1 The configuration of an ac microgrid with communication
xii
Figure Captions
agents………………………………………………………..…….123
Figure 5.2 Structure and hierarchical control scheme of each ESS/PV unit. ... 124
Figure 5.3 Proposed model predictive controller for primary control. ............ 128
Figure 5.4 Thresholds of load shedding and load recovering. ......................... 134
Figure 5.5 Events of the simulation for test case 1. ......................................... 137
Figure 5.6 Real power outputs of ESSs for test case 1..................................... 137
Figure 5.7 Real power outputs of PVs for test case 1. ..................................... 137
Figure 5.8 SoC profiles for test case 1. ............................................................ 138
Figure 5.9 Control objective of the leader for test case 1. ................................ 138
Figure 5.13 Power reference changes for test case 2. ........................................ 141
xiii
Figure Captions
xiv
Abbreviations
Abbreviations
AC Alternating Current
CAES Compressed Air Energy Storage
CB Circuit Beaker
CCM Current Control Mode
DC Direct Current
DER Distributed Energy Resource
DG Distributed Generation
EDLC Electrochemical Double Layer Capacitor
EPRI Electric Power Research Institute
ESS Energy Storage System
EU European Union
G2V Grid-to-Vehicle
LV Low-Voltage
MAS Multi-Agent System
MPC Model Predictive Control
MPPT Maximum Power Point Tracking
MV Medium Voltage
NTU Nanyang Technological University
OLTC On-Load Tap Changer
OPF Optimal Power Flow
PHES Pumped Hydro Energy Storage
PI Proportional-Integral
PQ Power Quality
PR Proportional- Resonant
PU Per Unit
PV Photovoltaic
xv
Abbreviations
RES Renewable Energy Source

SM Smart Meter
SOC State of Charge
SOD State of Discharge
TOU Time-of-Use
UPS Uninterruptible Power Supply
US United States
V2G Vehicle-to-Grid
VCM Voltage Control Mode
VRB Vanadium Redox Battery
VSI Voltage Source Inverter
WT Wind Turbine
xvi
Introduction Chapter 1
Chapter 1
Introduction
This chapter presents the background of this research work, which

includes the developments, advantages and challenges of energy storage
for smart grid applications. Motivated by reducing the carbon emissions,
air pollution and the demand of fossil fuels, renewable energy sources
together with energy storage systems have been largely penetrated into
existing power systems. This thesis aims to design effective control
methods for better utilization of energy storage systems in distribution
networks or microgrids with high renewable penetrations. Besides, the
research contributions and organization of the thesis are also presented.
1
1.1 Background and Motivations
1.1.1 Background
Nowadays, the dependence on traditional fossil fuels such as coal and oil has been
and will continue to be a major concern of climate change and air pollution globally.
These environmental problems and diminishing supply of fossil fuels accelerate the
development of renewable energy technologies [1]-[3]. Therefore, more and more
renewable energy sources (RESs) and energy storage systems (ESSs) have been
penetrated into existing distribution networks or microgrids through power
electronics interface. The energy production becomes more and more decentralized
other than conventional centralized generation. This type of power generation is
termed as distributed generation (DG) and is becoming an increasingly popular
solution for the future green energy development. In the meantime, efficient and
reliable communications and advanced control technologies, together with
increasing application of smarter electrical facilities, such as storage batteries,
electric vehicles (EVs) and smart meters, have resulted in a new power grid, also
named smart grid. The smart grid controls the behaviors of electrical devices based
on the information gathered from suppliers to consumers, which will greatly increase
efficiency, reliability, power quality and economics of the system. Consequently, the
entire power system will benefit from these new developments and become more
intelligent, interactive and flexible [4]-[7].
2
Figure 1.1 Renewable generation capacities in World, EU-28, BRICS and top
seven countries at the end of 2015 [9].
In the last decade, renewable energy markets are continuing to grow rapidly and at
least 30 nations all over the world already have renewable energy contributing more
than 20 percent of total energy supply [8]. According to REN21's 2016 report,
renewable energy has reached 23.7% worldwide electricity generation in 2015 [9].
The renewable generation capacities of several large countries and groups at the end
of 2015 are shown in Figure 1.1. Among all the renewable energy generation, solar
panels and wind turbines are the most widely used, since they are clean, infinite,
scalable and cost-effective [10], [11]. However, photovoltaic (PV) generation keeps
fluctuating due to the daily solar irradiance changes, and wind generation is affected
largely by the seasonal variation of wind at the installed location [12], [13]. PV and
wind generation cannot guarantee constant power supply due to their stochastic and
intermittent characteristics, which will influence the stability, reliability and power
quality of power systems [14], [15].
3
Energy storage is a relative new phenomena following in the wake of increased

installation of renewable energy generation in the 21st century. As various problems
caused by the penetration of solar panels and wind turbines generation become more
common, energy storage is considered as one promising solution to deal with them
[16]. For all kinds of energy storage technologies, there are two modes of operation,
which are charging mode and discharging mode. Different kinds of energy storage
technologies transform different forms of energy to electricity and vice versa. A
widely used approach for classification of energy storage technologies is according
to the form of energy used. Electrical energy storage technologies can be categorized
by the form of storing energy into five groups: mechanical, electrochemical,
chemical, electrical and thermal energy storage [17]. The electrical energy storage
devices belonging to each classification are shown in Figure 1.2.
Figure 1.2 Classification of electrical energy storage technologies according to energy

form [17].
4
Large investments have been placed on research and application of energy storage
technologies in countries such as Germany, the United States (U.S.) and China where
the installation of renewable energy generation has progressed fast over the last few
years [18], [19]. Energy storage systems including batteries, ultra-capacitors,
flywheels, etc. have been widely utilized for both grid-connected power systems and
off-grid renewable generation in remote areas. For grid-connected application, ESSs
are of great importance since they can cope with the intermittency and power
variation caused by RESs in power systems, which increase system stability and
reliability significantly [20], [21]. Furthermore, the incorporation of ESSs are also
critical for improving the power quality and voltage profiles of distribution networks,
and performing various demand side management functions such as peak shaving
and load shifting [22]-[24]. While for off-grid application, ESSs play the key role to
balance the renewable generation and load consumption instantaneously in order to
keep the frequency and voltage of the islanded microgrid stable [25].
Another important aspect related to the adoption of energy storage technologies is

transportation electrification, which is an effective way to substantially reduce the
carbon emissions and air pollution [26]-[28]. In the last decade, EVs have been
receiving widespread attention because of their green and economic features.
Therefore, for the sake of environmental protection and reduction of reliance on
fossil fuels, governments of many countries such as U.S., European Union, China
and Japan significantly invest in this field and promulgate corresponding policy for
the development of EVs [29], [30]. Based on moderate expectations, by 2020 up to
35% of the total vehicles in the U.S. will be EVs according to the Electric Power
Research Institute (EPRI) [31]. Plug-in EVs can be connected to power grid through
electric vehicle chargers (EVCs) and operated in discharge mode known as vehicle-
to-grid (V2G) or in charge mode known as grid-to-vehicle (G2V) [32]. With the fact
that private vehicles are parked, on average, 93%–96% of their lifetime, EVs can be
5
viewed as mobile ESSs, which have many similar functions as stationary ESSs for
grid application [33]-[35]. The V2G concept has attracted attention of both grid
operators and vehicle owners. However, effective utilization of storage capacity and
proper recharging strategies of EVs are necessary to realize the benefits of V2G
capabilities.
1.1.2 Motivations
The driving forces, which are from the urgent demand for reduction carbon
emissions, air pollution and needs on fossil fuel, promote the growing research
interest and investment in renewable energy and energy storage technologies. The
integration of RESs into the existing distribution networks will lead to power
variation as well as more complex power flow, which will change the voltage profiles
of distribution networks dramatically. The impacts of increasing penetration of RESs
on the existing distribution grids need to be investigated. In order to facilitate
applications of renewable energy, energy storage has become an emerging topic for
both utility grid and transportation. The research on ESSs and EVs for renewable
energy penetration and grid ancillary services is becoming more and more attractive
and important. In addition, one significant characteristic of modern power systems
is that RESs and ESSs based DG units are interfaced to the system through power
electronics devices and the control of the system is primarily referred to the
coordinated control of the interfacing inverters of DG units. Hence, the major
challenges in this research are to find effective methods to model, control and
supervise of various sources based DG units.
In review of the above situations, the motivation of this thesis is to design effective
control strategies for different system structures consisting of RESs, ESSs and EVs
to solve the problems when integrating of renewable energy into power systems.
6
With the control strategies developed in this research work, ESSs and EVs should
be able to deal with the voltage problems which arise due to the renewable
penetration and load demand in distribution networks. Furthermore, ESSs can be
utilized to solve the power balancing and power management problems in islanded
microgrids. Throughout this research, the state of charge (SoC) of storage batteries
in ESSs should be monitored and controlled properly, to avoid depletion or saturation
of the storage batteries. In general, the proposed control strategies should contribute
to the penetration of renewable energy and realize the benefits of energy storage
techniques in the smart grid.
1.2 Objectives and Scope
With the challenges and motivations to be explored, the objectives of this thesis are
to solve problems related to the application, modeling, control and operation of the
ESSs in the smart grid. In this thesis, the ESSs are mainly applied for voltage
regulation in distribution networks and power balancing in microgrids. In order to
design effective control systems, mathematic models of electrical components such
as power inverters, solar panels and storage batteries are needed to be built. Then
control systems can be developed to control the parameters provided by the
mathematic models to achieve the different control objectives. After that, simulation
platform including system data such as line impedances, PVs, EVs and loads must
be implemented to validate the performance of the proposed control method. At last,
the results under different conditions are to be demonstrated and discussed. The
detailed objectives of this thesis are listed as follows:
1. The first objective is to design a coordinated control method to control the power
outputs of distributed ESSs for voltage regulation in distribution networks with high
PV penetration. Generally, during peak PV generation period, the voltages are
7
mitigated by charging the ESSs, and the stored energy is discharged for voltage
support during peak load period. The impact of storage devices integrated with PV
system on feeder voltages needs to be investigated in detail. Proper battery model is
to be developed to simulate the behavior of ESSs for grid-scale application. The
proposed coordinated control method is aimed to regulate the voltages in the
distribution networks with the required limits and fully utilize the available storage
capacity of ESSs during daily operation. Therefore, both distributed voltage
controller and localized SoC controller will be properly designed to achieve the
control objective.
2. The second objective is to apply distributed EVCs and their charging EVs for
voltage regulation in distribution networks with high penetration of PVs and EVs.
With the concept of V2G, plug-in EVs can also be served as energy storage devices
and contribute to voltage regulation in distribution networks. When the real power
of the plug-in EVs is not available for voltage regulation, the reactive power of EVCs
can be utilized as a backup. First, voltage regulation method of distributed EVCs will
be designed. The coordination of distributed EVCs can be achieved through
communication links between them using consensus algorithm. Then algorithms for
the estimation of V2G capacity of the plug-in EVs and determination of the real and
reactive power outputs of each EVC are to be developed, so that the voltage
regulation objectives can be achieved within the required operation limits.
3. The third objective is to investigate the application of ESSs in an islanded

microgrid. Different from the conditions of grid-connected application, the main
function of ESSs is to balance the PV generation and load demand within the
islanded microgrid in order to maintain the frequency and voltage stability. To
achieve fast dynamic response and small steady-state error of power output, model
predictive control (MPC) of power inverters will be adopted. The discrete time state-
8
space models need to be derived using the equivalent circuit model depending on the
control objectives of the power inverter. A hierarchical control strategy will be
designed to coordinate all available units including ESSs, PVs and loads together for
the power balancing and SoC management in the microgrid. PV curtailment and load
shading will be conducted if the remaining capacities of ESSs are not enough during
the operation of the microgrid.
4. Finally, to verify the performance of the proposed control methods in this thesis,
simulation platforms are to be built using MATLAB/Simulink. The proposed control
methods will be implemented in simulation platforms and tested under various
scenarios. The simulation results will be demonstrated that the entire system is able
to operate stably and the control design is accurate and effective under various
operation conditions. Different types of PV and load profiles will be used to test both
steady-state and dynamic performance of the proposed control methods.
1.3 Contributions of Thesis
The main contributions of this thesis are summarized as follows:
1. In this thesis, different control strategies are proposed for the application of ESSs
in distribution networks and microgrids. The centralized control requires fast and
reliable communication links, and the reliability of the control system is highly
dependent on the performance of the centralized controller. While the decentralized
control is difficult to achieve fully coordination between available units due to lack
of global information. Since the limitation of centralized and decentralized controls,
distributed control is adopted to coordinate available units in this thesis. When
distributed control is applied to control EESs in distribution networks, further
research which considers actual PV and load profiles, different capacities and SoCs
9
of ESSs and various topologies of distribution networks still needs to be done.

Besides, DG units are assumed as ideal power sources and the type and capacity of
DG units are not considered in many research works so far. The SoC and power
limits of energy storage based DG unit will greatly influence the stability and
reliability of the system during operation, especially in distribution networks with
high PV penetration and variable loads. Therefore, characteristics and constraints of
ESSs are taken into account for control system design in this research work.
2. This thesis addresses how to coordinate distributed ESSs to solve the voltage
rise/drop issues in low-voltage (LV) distribution networks. A coordinated control
method which includes both distributed and localized controls is proposed. For the
distributed control design, a consensus algorithm is proposed to coordinate all
available ESSs to regulate the feeder voltages. If the voltages at certain buses violate
voltage limits, the consensus algorithm will determine the power output of each ESS
to eliminate the voltage violation. For localized control design, a SoC control
strategy is proposed to control the SoC of each ESS within an allowable range of the
desired SoC. The SoC control avoids depletion or saturation of ESSs under different
operation conditions. The entire control method ensures voltage regulation while
effectively utilizes all available capacities of ESSs during daily operation. Besides,
the ESS is modeled as vanadium redox battery (VRB) whose SoC is estimated for
the SoC control. Case studies with real PV and load data under different whether
conditions are implemented to verify the performance of this method. In addition,
the effectiveness of the proposed method in distribution networks with multiple
laterals is also validated.
3. Innovated by the previous research work on distributed ESSs, the real and reactive
power outputs of distributed EVCs are applied to solve the voltage regulation
problem in distribution networks with high penetration of PVs and EVs. To delineate
10
the behaviors of battery storage devices on EVs, a lithium-ion battery model is

derived by fitting the experimental data. A novel two-stage coordination method of
distributed EVCs for voltage regulation is proposed. The first control stage obtains
the voltage regulation target by measuring voltages of critical buses and distributes
the control signal to each EVC through communication links. The second control
stage determines the amount of real and reactive power which can be provided by
each EVC, considering the constraints and charging objectives of each EVC and its
charging EV. Different types of PV and load profiles are utilized to evaluate the
performance of the proposed method in a practical distribution network.
4. Finally, a hierarchical control scheme is proposed for islanded ac microgrids

formed by the ESSs, PV systems, and loads. To achieve fast dynamic response and
small steady-state error of voltage source inverters (VSIs), a novel N-step MPC is
developed for the primary control. Depending on different cost functions of MPC
method, the VSI of each unit can operate in current control mode (CCM) or voltage
control mode (VCM). The secondary control is achieved through communication
links between each unit using the consensus algorithm. The objective of secondary
control is to solve large power imbalance and overcharging/overdischarging
problems in the microgrid. This control scheme ensures the power balancing and
energy management can be achieved hierarchically under different PV and load
conditions. Both short-term and long-term simulation studies are conducted to
demonstrate the performance of the proposed hierarchical control method.
11
1.4 Organization of Thesis
The thesis is organized into six main chapters as follows:
Chapter 1: Introduction
This chapter provides background information of energy storage technologies

including their developments, applications, advantages and challenges. The
motivations, objectives and contributions of this thesis, as well as the organization
of this thesis are also discussed.
Chapter 2: Literature Review
In this chapter, a comprehensive review of the related research work on the

application and control of energy storage technologies in utility grid and
transportations is presented. Then, the voltage regulation problem in the distribution
networks, which is one of the main focuses of this thesis, is discussed in detail. After
that, a review on the state-of-art control hierarchy of power systems is provided.
Finally, the consensus algorithm, which is the critical control method adopted in this
thesis, is discussed.
Chapter 3: Coordinated Control of Distributed Energy Storage Systems for Voltage

Regulation in Distribution Networks
In this chapter, distributed ESSs are proposed to solve the voltage rise/drop issues in
LV distribution networks with high penetration of rooftop PVs. The impact of energy
storage device integrated with PV source on feeder voltages is investigated in detail.
A coordinated control method which includes both distributed and localized controls
12
is proposed for distributed ESSs. The entire control structure ensures voltage
regulation while effective utilization of storage capacity under various operation
conditions. The proposed control method is evaluated in two typical LV distribution
networks with real PV and load data, and the simulation results validate the
effectiveness of this method.
Chapter 4: Two-stage Coordination of Distributed Electrical Vehicle Chargers for

Voltage Regulation in Distribution Networks
A two-stage coordination method of distributed EVCs is proposed to regulate the

voltages in distribution networks in this chapter. In the stage one, the distributed
voltage control is proposed to coordinate each EVC for voltage regulation. In the
stage two, the power allocator is developed to determine the active and reactive
power outputs of each EVC, so that the voltage regulation is achieved within the
required operation limits. The proposed two-stage control ensures that the voltages
along the distribution network are within the allowable voltage range by coordinated
control of the power output of each EVC during daily operation. The proposed
voltage control method is evaluated in a LV residential network under various
operation conditions and the simulation results validate the effectiveness of the
proposed method.
Chapter 5: Hierarchical Control and Power Management of Islanded Microgrids

with Energy Storage Systems and Renewable Energy Sources
In this chapter, a hierarchical control scheme is presented for PV systems, ESSs and
loads in an islanded ac microgrid to deal with both the short-term and long-term
power imbalance. The primary control is realized by a novel MPC method, which
provides fast dynamic response and small steady-state error for power inverters in
13
the microgrid. The secondary control is performed by consensus based distributed

control, which provides distributed control signal for primary controlled inverters
and loads. With the proposed control scheme, the power balancing and SoC
management can be achieved by controlling the power output of each available unit
in a hierarchical way. The control scheme is implemented in an ac microgrid and the
simulation results demonstrate the performance of the proposed approach.
Chapter 6: Conclusions and Future Work
An overall conclusion of this research is discussed in this chapter. Moreover, areas

for further research are proposed as well.
References
[1] B. K. Bose, “Global warming: energy, environmental pollution, and the impact
of power electronics,” IEEE Industrial Electronics Magazine, vol. 4, no. 1, pp.
6-17, Mar. 2010.
[2] N. L. Panwar, S. C. Kaushik, and S. Kothari, “Role of renewable energy
sources in environmental protection: a review,” Renewable and Sustainable
Energy Reviews, vol. 15, no. 3, pp. 1513-1524, Apr. 2011.
[3] T. Strasser, F. Andren, J. Kathan, et al., “A review of architectures and
concepts for intelligence in future electric energy systems,” IEEE Trans.
Industrial Electronics , vol. 62, no. 4, pp. 2424-2438, Apr. 2015.
[4] S. Chowdhury, S. P. Chowdhury, and P. Crossley, Microgrids and Active
Distribution Networks. UK: Institution of Engineering and Technology, 2009,
pp. 1-13.
[5] J. M. Guerrero, M. Chandorkar, T. Lee, and P. C. Loh, “Advanced control
architectures for intelligent microgrids-part I: decentralized and hierarchical
14
control,” IEEE Trans. Industrial Electronics, vol. 60, no. 4, pp. 1254-1262,
Apr. 2013.
[6] J. M. Guerrero, J. C. Vasquez, J. Matas, L. G. de Vicuña, and M. Castilla,
“Hierarchical control of droop-controlled AC and DC microgrids-a general
approach toward standardization,” IEEE Trans. Industrial Electronics, vol. 58,
no. 1, pp. 158-172, Jan. 2011.
[7] J. Driesen and F. Katiraei, “Design for distributed energy resources,” IEEE
Power and Energy Magazine, vol. 6, no. 3, pp. 30-40, May/Jun. 2008.
[8] J. C. Wu, K. D. Wu, H. L. Jou, and S. K. Chang, “Small-capacity grid-
connected solar power generation system,” IET Power Electronics, vol. 7, no.
11, pp. 2717-2725, Nov. 2014.
[9] Renewables 2016 Golbal Status Report. [Online]. Available:
http://www.ren21.net/status-of-renewables/global-status-report/.
[10] Y. Duan and R. G. Harley, “Present and future trends in wind turbine generator
designs,” in Proc. Power Electronics and Machines
in Wind Applications, 2009, pp. 1-6.
[11] G. J. Shirek and B. A. Lassiter, “Photovoltaic power generation: modeling
solar plants' load levels and their effects on the distribution system,” IEEE
Industry Applications Magazine, vol. 19, no. 4, pp. 63-72, Aug. 2013.
[12] J. C. Wu and Y. H. Wang, “Power conversion interface for small-capacity
wind power generation system,” IET Generation, Transmission and
Distribution, vol. 8, no. 4, pp. 689-696, Apr. 2014.
[13] A. M. Azmy and I. Erlich, “Impact of distributed generation on the stability of
electrical power systems,” in Proc. IEEE Power Engineering Society General
Meeting, 2005, pp. 1056-1063.
[14] J. A. Martinez and J. Martin-Arnedo, “Impact of distributed generation on
distribution protection and power quality,” in Proc. IEEE Power Engineering
Society General Meeting, 2009, pp. 1-6.
15
[15] S. Vazquez, S. M. Lukic, E. Galvan, L. G. Franquelo, and J. M. Carrasco,

“Energy storage systems for transport and grid applications,” IEEE Trans.
Industial Electronics, vol. 57, no. 12, pp. 3881–3895, Dec. 2010.
[16] B. R. Alamri and A. R. Alamri, “Technical review of energy storage
technologies when integrated with intermittent renewable energy,” in Proc.
International Conference on Sustainable Power Generation and Supply, Apr.
2009, pp.1–5.
[17] Electrical energy storage project team, “Electrical energy storage.”
International Electrotechnical Commission, Dec. 2011.
[18] B. B, Mckeon, J. Furukawa and S. Fenstermacher, “Advanced lead–acid
batteries and the development of grid-scale energy storage systems,”
Proceedings of the IEEE, vol. 102, no. 6, pp. 951-963, Mar. 2014.
[19] “Battery energy storage for smart grid applications,” EUROBAT, 2013.
[20] S. Bacha, D. Picault, B. Burger, I. E. Otadui, et al., “Photovoltaics in
microgrids: an overview of grid integration and energy management aspects,”
IEEE Industrial Electronics Magazine, vol. 9, no. 1, pp. 33-46, Mar. 2015.
[21] M. J. Alam, K. M. Muttaqi, and D. Sutanto, “A novel approach for ramp-rate
control of solar PV using energy storage to mitigate output fluctuations caused
by cloud passing,” IEEE Trans. Energy Conversion, vol. 29, no. 2, pp. 507-
518, Mar. 2014.
[22] I. Serban and C. Marinescu, “Control strategy of three-phase battery energy
storage systems for frequency support in microgrids and with uninterrupted
supply of local loads,” IEEE Trans. Power Electronics, vol. 29, no. 9, pp.
5010-5020, Sep. 2013.
[23] R. Sebastian, “Application of a battery energy storage for frequency regulation
and peak shaving in a wind diesel power system,” IET Generation,
Transmission and Distribution, vol. 10, no. 3, pp. 764-770, Feb. 2016.
16
[24] H. Kakigano, Y. Miura, and T. Ise, “Distribution voltage control for DC

microgrids using fuzzy control and gain-scheduling technique,” IEEE Trans.
Power Electronics, vol. 28, no. 5, pp. 2246-2258, Sep. 2012.
[25] S. J. Lee, J. H. Kim, C. H. Kim, S. K. Kim, et al., “Coordinated control
algorithm for distributed battery energy storage systems for mitigating voltage
and frequency deviations,” IEEE Trans. Smart Grid, vol. 7, no. 3, pp. 1713-
1722, Jun. 2015.
[26] L. Situ, “Electric vehicle development: the past, present and future,” in Proc.
IEEE International Conference on Power Electronics Systems and
Applications, 2009, pp. 20-22.
[27] S. Amjad, S. Neelakrishnan, and R. Rudramoorthy, “Review of design
considerations and technological challenges for successful development and
deployment of plug-in hybrid electric vehicles,” Renewable and Sustainable
Energy Reviews, vol. 14, no. 13, pp. 1104-1110, Apr. 2010.
[28] E. Karden, S. Ploumen, B. Fricke, T. Miller, and K. Snyder, “Energy storage
devices for future hybrid electric vehicles,” Journal of Power Sources, vol.
168, no. 1, pp. 2-11, May 2007.
[29] J. Du and M. Ouyang, “Review of electric vehicle technologies progress and
development prospect in China,” in Proc. Electric Vehicle Symposium and
Exhibition, 2013, pp. 17-20.
[30] K. Rajashekara, “Present status and future trends in electric vehicle propulsion
technologies,” IEEE Journal of Emerging and Selected Topics in Power
Electronics, vol. 1, no. 1, pp. 3-10, Mar. 2013.
[31] M. Duvall and E. Knipping, “Environmental assessment of plug-in hybrid
electric vehicles. Volume 1: nationwide greenhouse gas emissions,”
EPRI/NRDC, Palo Alto, CA, Final Rep. 1015325, 2007, pp. 1–56.
17
[32] W. Kempton and J. Tomic, “Vehicle-to-grid power implementation: From

stabilizing the grid to supporting large-scale renewable energy,” Journal of
Power Sources, vol. 144, no. 1, pp. 268–279, 2005.
[33] H. Turton and F. Moura, “Vehicle-to-grid systems for sustainable
development: An integrated energy analysis,” Journal of Technological
Forecasting and Social Change, vol. 75, no. 8, pp. 1091-1108, Oct. 2008.
[34] U. C. Chukwu and S. M. Mahajan, “V2G parking lot with PV rooftop for
capacity enhancement of a distribution system,” IEEE Trans. Sustainable
Energy, vol. 5, no. 1, pp. 119-127, Sep. 2013.
[35] H. M. Yang, C. Y. Chung, and J. H. Zhao, “Application of plug-in electric
vehicles to frequency regulation based on distributed signal acquisition via
limited communication,” IEEE Trans. Power Systems, vol. 28, no. 2, pp. 1017–
1026, Aug. 2012.
18
Literature Review Chapter 2
Chapter 2
Literature Review
In this chapter, a comprehensive review of the related research works on

energy storage technologies for grid and transportation applications is
presented in detail. Then, as the main problem to be solved in this thesis,
voltage rise/drop issues and corresponding regulation methods are
discussed. After that, a review of state-of-art control hierarchy of modern
power system is conducted. Besides, consensus algorithm which is the
basic control method adopted in this thesis is introduced. Finally, the
literature review has been summarized.
19
2.1 Energy Storage Systems
Energy storage refers to a process of converting electrical energy from power

systems into a form that can be stored for converting back to electrical energy when
needed [1]-[2]. Due to the development of renewable energy technologies, energy
storage systems have been widely applied for utility grid and transportation
applications [3]. For utility grid application, energy storage plays an important role
in integration of renewable energy into power systems and ancillary services such as
peak load shaving, reserve, frequency and voltage regulation. According to the report
of market research company HIS, grid connected energy storage will increase rapidly
and reach 6GW in 2017, and will be over 40GW in 2020 [4]. On the other hand,
hybrid electric vehicles and battery electric vehicles have been developed for
decades and already in service in order to reduce CO2 emissions and fossil fuel
consumption. V2G capacities of plug-in EVs can offer a possible backup for RESs
including solar and wind energy, contributing integration of intermittent renewable
energy generation in power systems. V2G technologies can also provide additional
opportunities for grid operators, such as reactive power compensation, active power
control, load levelling by valley filling/peak load shaving, and harmonic current
filtering [5].
2.1.1 Comparison of Energy Storage Technologies
Different energy storage technologies can be compared from technical and economic
point of view. The investment cost of various energy storage technologies is studied
in [6]-[8], which will not be taken into consideration in this research. The main
technical characteristics of energy storage are energy density, power density, energy
efficiency, life cycle (life time), and self-discharge, which are explained as follows
[9], [10].
20
1) Energy density (Wh/kg) and power density (W/kg) are the energy to weight
ratio and the power to weight ratio respectively. Power density refers to the
ability of energy storage to provide instantaneous power, while energy density
specifies the ability to provide continuous energy over a period of time. Energy
storage with high power density can discharge large amounts of power on
demand, and energy storage with high energy density can discharge energy for
a long period.
2) Energy efficiency (%) is the ratio of released energy to stored energy or the
output and input electricity without concerning of self-discharge.
3) Life cycle or durability (cycles) is the number of times which energy storage
can release the designed energy level after each recharge. One cycle means one
charge and one discharge.
4) Self-discharge is the ratio of initial stored energy to dissipated energy over a
non-use time.
The technical characteristics of various energy storage technologies are summarized

in Table 2.1 [3]. As shown in Table 2.1, electrochemical double layer capacitors
(EDLC) and flywheel have the highest energy efficiency, while compressed air
energy storage (CAES) has very low energy efficiency and Ni-MH has the lowest
energy efficiency among battery energy storage technologies (i.e. Lead-Acid, Ni-Cd,
Ni-MH, Lithium-ion, Li-polymer, NaS and VRB). The mechanical energy storage
technologies which include pumped hydro energy storage (PHES), CAES, and
flywheel have very long life cycle and are very durable. Battery energy storage
usually have short life cycle while lead-acid battery (lead-acid) is the shortest in this
group. Although EDLC and flywheel have long life time, their self-discharge is quite
high. It means EDLC and flywheel are very suitable for short-term storage
application, considering their high self-discharge and high energy efficiency. The
21
self-discharge can be neglected for PHES and VRB, which means they are very
suitable for long-term storage application.
Table 2.1 Characteristics of energy storage technologies [3]
Energy Power Energy

Life Cycle Self-
Type Density Density Efficiency
(cycles) Discharge
(Wh/kg) (W/kg) (%)
Lead-Acid 20-35 25 70-80 200-2000 Low
Ni-Cd 40-60 140-180 60-90 500-2000 Low
Ni-MH 60-80 220 50-80 <3000 High
Li-ion 100-200 360 70-85 500-2000 Medium
Li-polymer 200 250-1000 70 >1200 Medium
NaS 120 120 70 2000 -
VRB 25 80-150 80 >16000 Negligible
EDLC <50 4000 95 >50000 Very High
PHES 0.3 - 65-80 >20 years Negligible
CAES 10-30 - 40-50 >20 years -
Flywheel
5-30 1000 95 >20000 Very High
(steel)
Flywheel
>50 5000 95 >20000 Very High
(composite)
In general, energy storage technologies can be separated into two groups: high
energy density and low power density, and low energy density and high power
density. Energy storage with high energy density and low power density like
batteries is suitable for a long period service (about an hour) but its time response is
large. Energy storage with low energy density and high power density such as
supercapacitor and flywheel provides fast time response but it cannot be in service
for a long period (in several seconds to minutes). Different energy storage
technologies should be chosen according to the applications, considering the
characteristics in Table 2.1.
22
In this thesis, two kinds of battery energy storage technologies are applied, which
are VRB and lithium-ion battery. The energy storage systems refer to the battery
packs with multiple battery cells series and parallel connected, which penetrated into
power systems at the dc part of power inverters. As shown in Table 2.1, VRB has
relatively small energy density and power density compared with lithium-ion battery.
While the advantages of VRB are the long life cycle and low self-discharge rate. For
utility grid application or renewable energy integration, energy storage capacity,
energy efficiency and life cycle are key performance criteria. The VRB exhibits
many advantages such as independent power and energy ratings, fast charge and
discharge response time, high energy efficiency, long life cycle, low self-discharge,
and convenient for SoC estimation, which makes VRBs very suitable for utility
applications [11], [12]. For transportation applications, portability, scalability, and
energy and power density are key performance criteria. Lithium-ion battery has both
very high energy density and power density compared with other batteries. This
characteristic makes lithium-ion battery the mostly used batteries in transportation
application because it needs small weight and volume for high energy and power
[13]-[15].
In addition, storage capacity is a primary limitation of energy storage. Energy storage

cannot absorb or supply energy if they are already full or empty respectively. Since
either energy stored or energy used by energy storage has to be elaborated, the
concept of state of charge and depth of discharge (DoD) are introduced [16], [17].
State of charge defines the level of energy which remains in energy storage
comparing to the total capacity of energy storage in percentage. Depth of discharge
is the contrary of SoC. It defines how much energy is used or discharges in
percentage to the total capacity. In control system, SoC is one important parameter
which has profound impacts on the control strategy of energy storage. For example,
if energy storage cannot supply energy because SoC is very low, control strategy has
23
to recharge storage as soon as possible and other sources have to be responsible for
supplying compensating energy.
2.1.2 Energy Storage for Utility Grid
Energy storage technologies have been widely utilized in modern smart grid for
different applications. According to the function or role of ESSs in smart grid, they
can be classified into three main applications, which are: grid ancillary services for
power quality improvement, uninterrupted power supply (UPS) for off-grid systems,
and energy management for optimal operation. These applications will contribute to
the penetration of renewable energy into the existing power system and improvement
the power quality as well as the efficiency of the entire system. The power/capacity
ranges of different energy storage technologies are shown in Figure 2.1 [15]. Figure
2.1 also shows the suitability of various ESSs for utility grid application. In the
following subsections, some related research works of this field are reviewed.
Figure 2.1 Different energy storage technologies for grid application [15].
24
1) Grid Ancillary Services

The grid ancillary services of ESSs for power quality improvement usually last for a
few seconds to a few hours, which include RESs intermittency mitigation [18]-[20],
frequency and voltage regulation [21]-[23].
The ESS has become one major method for smoothing solar and wind power
generation fluctuations. In [18], a method where the ramp-rate of PV panel output is
measured to control the ramp-rate of inverter output to a desired level by utilizing
the ESS is proposed. During the ramping up/down period, the desired ramp-rate is
controlled by the ESS according to an inverse relationship with the ramp-rate of PV
panel output to improve the performance of PV fluctuation mitigation. For the same
purpose, a power smoothing strategy for a 1-MW grid-connected PV power plant is
proposed in [19]. To smooth the fluctuating power output of the PV plant, a hybrid
ESS composed of a VRB and a supercapacitor bank is applied. The power
management system of the hybrid ESS is designed to reduce the required power
rating of the supercapacitor to one-fifth of the VRB rating and to avoid the operation
of the VRB at low power levels, which increases its overall efficiency. In addition, a
smoothing control method for reducing hybrid PV/wind system power output
fluctuations and regulating the SoC of battery ESS under different operation
conditions is proposed in [20].
The frequency and voltage variation is another issue brought by the penetration of
RESs, which can be solved by utilizing ESSs. The integration impact of battery ESSs
on the short-term frequency control in autonomous microgrids is investigated in [21].
This research improves the short-term frequency stability in the microgrid by a
control method of ESSs and enhances both inertial response and adaptive droop
characteristic. In [22], a power system consisting of a high penetration of wind
generation and ESS is investigated for the regulation of grid frequency. A strategy
25
called SoC feedback control is proposed to improve the performance of the grid
frequency deviation response and control the SoC of the ESS. In [23], a coordinated
control strategy to compensate for inertia of the wind farm is proposed. This strategy
can improve the temporary frequency support ability of the wind turbine generators
based on the coordinated control of the wind turbine generators and the ESSs.
2) UPS Application
The UPS refers to the operation of ESSs in off-grid power systems or islanded
microgrids, which has been studied by many researchers for both ac and dc
microgrids [24]-[27].
In grid-connected condition, power can be imported from the grid to charge the
battery packs of ESSs. While in stand-alone condition, the ESSs can supply local
load demand together with other sources. A line-interactive UPS for ac microgrids
and the corresponding control system are introduced and discussed in [24]. Seamless
transfer between grid-connected and stand-alone operation of the microgrid is
achieved by a frequency and voltage drooping method. The drooping coefficients are
properly chosen to limit the power absorbed by the UPS when connecting back to
the utility grid as well as improve the transient response. In [25], a cooperative
control strategy of distributed energy resources (DERs) and the ESS during islanded
operation is proposed and evaluated by both simulation and experiment. The ESS is
controlled by a primary control to handle the short-term frequency and voltage
regulation, while the secondary control make the DERs respond if the frequency and
voltage deviation last for a long time.
In the last few years, much attention has been paid on dc microgrids. DC microgrids
enable dc output sources such as PV systems, storage batteries, and fuel cells to be
interconnected without ac/dc conversion, which improve the system efficiency. For
26
dc microgrids which have multiple ESSs, a voltage control method which combines
fuzzy control and gain-scheduling approaches to realize both power sharing and
energy management is proposed in [26]. The research results demonstrate that the dc
voltages in the system are regulated within 340V ± 5%, and the SoCs of multiple
ESSs are almost equal. In [27], a SoC-based adaptive droop control method is
presented to balance the SoCs of distributed ESSs in dc microgrids. This control
method is achieved in a decentralized way, in which the droop coefficient is inversely
proportional to the nth order of SoC, so that the load demand is equally shared among
the distributed ESSs in the system.
3) Energy Management
Energy management is another important application area of energy storage, which
includes load leveling, peak shaving, system reverse, economic dispatch, etc. [28]-
[30]. Sizing of ESSs is also an important topic to achieve the energy management
function of ESSs [31], [32].
Energy management or scheduling of distributed ESSs plays an essential role in the

process of improving the efficiency, reliability and economics of power grids. In [28],
a smart power system in which users are equipped with energy storage devices is
considered. By scheduling the energy consumption and storage of users by a central
controller, the energy payment can be minimized. A distributed economic dispatch
strategy for microgrids with multiple ESSs is introduced in [29]. This strategy
overcomes the challenges of dynamic couplings among all decision variables and
stochastic variables in a centralized dispatching formulation. In [30], an algorithm
that uses demand profile information and a minimal set of ESS parameters is
formulated to obtain the ESS operation schedules to achieve peak demand shaving
and load-levelling. In [31], an effective sizing strategy for distributed battery ESSs
in the distribution networks under high PV penetration level is proposed. The main
27
objective of this method is to optimize the size of distributed battery ESSs and derive
the cost-benefit analysis when ESSs are applied for voltage profile improvement and
peak load shaving. In [32], a comprehensive planning framework is introduced for
ascertaining the most cost effective siting and sizing of ESSs that maximizes their
benefits in distribution networks. A probabilistic approach is adopted that includes
the consideration of the stochastic nature of system components. This approach
allows determining the optimal operation of ESSs at each load state. In general, the
time scale of energy management for ESSs can vary from a few hours to months,
which usually has longer time scale than power quality services.
2.1.3 Energy Storage for Transportation
Electric vehicles have a large potential amount of storage capacity, since the energy
stored in the battery is not fully used by most of EVs in regular daily trips. The
utilization of spare battery capacity of EVs for grid support applications with V2G
concept is becoming popular and important. V2G concept describes that EVs can
communicate with the power system to provide demand respond services by either
delivering electricity into the grid or throttling their charging rate. When plugged
into an electricity network, the EV can act either as a controllable load or an energy
storage unit in the system. The stability and reliability of the renewable energy
penetrated power system will be enhanced with the vast storage of EVs connected to
the grid. In order to provide ancillary service from the V2G system, a group of EVs
is necessary to coordinate together as an aggregator. The EV aggregators are
expected to collect EVs into a group through electrical connection and
communication, which create a larger, more desirable system for the utility
application [33]. EVs can be charged in EV charging stations, car parks and home
with charging outlets. Industry has defined three typical rates of EVCs, which can
be classified into three power levels, as shown in Table 2.2 [34]. Normal EV
28
charging can be achieved by Level 1 and Level 2 EVCs, while fast charging of EVs
is implemented by Level 3 EVC.
Table 2.2 Charging power levels of EVCs [35]
Power Level Description Power Level
Opportunity charger (any available 1.4kW (12A)

Level 1
outlet) 1.9kW (20A)
4kW (17A)
Level 2 Primary dedicated charger
19.2kW (80A)
Level 3 Commercial fast charger Up to 100kW
The EV charging pattern is an important factor which will influence the load demand
profiles as well as the voltage profiles in distribution networks. To be summarized,
there are three different EV charging patterns: constant EV charging, price-based EV
charging, and V2G. Constant EV charging represents the situation where no upper
level controller coordinates the EV charging. The EV will charge with a constant
charging power once plugged into the grid and stop charging when either the battery
is fully charged or the EV is plugged out from the grid. In residential area, EV
charging with constant power tends to coincide with evening peak load demand. This
will increase the burden of distribution networks and also cause severe voltage drop
issue. Therefore, this charging pattern is only suitable for low EV penetration
condition where EVs are just treated as passive loads. Price-based EV charging
allocates EV charging activities over time in a cost-minimizing manner while
retaining driving feasibility. The EV charging will consider the time of use tariff and
consequent to off peak charging. Price-based EV charging is an EV management
problem which can be formulated into an optimization problem. With this charging
pattern, the EV charging is directly managed and can be treated as a responsive load.
The last charging pattern ‘V2G’ enables the EV owners to sell electricity back and
29
provides ancillary services for utility grid. Ancillary services are necessary in the
power system for maintaining grid reliability and power quality, balancing supply
and demand, and supporting the transmission of power from generation to
consumption. The ancillary services provided by V2G include: load levelling and
peak power management [36], [37], quick frequency and voltage regulation [38]-
[40], and effective renewable supporting and balancing [41], [42]. As V2G
technology is the main focus of this thesis, the application of V2G is further
discussed below.
1) Load Leveling and Peak Load Shaving

Depending on the stress on the grid during peak load periods, the peak-shaving
support from the EVs can be very useful. The battery packs of EVs have limited
capacity and the capacity usage is also constrained by travel requirements. Therefore,
in [36], a control strategy is proposed for an effective utilization of the available EV
capacity for peak shaving. Dynamic adjustments in EV discharging rates are
developed to fully utilize the limited EV battery capacity when peak shaving is most
valuable based on the demand pattern. In [37], the modeling of a multi charging
station for EVs and its utilization for grid support is investigated. A control
architecture is proposed based on fuzzy control techniques, which maintains the node
voltage within the prescribed limit by valley filling and peak shaving.
2) Frequency and Voltage Regulation

With frequency regulation capabilities of EVs, a V2G control strategy is proposed
for the aggregator to dispatch regulation requirements to EVs and EV charging
stations in [38]. In each EV charging station, the frequency regulation control is
calculated based on the V2G power at present time, and V2G power is presented
considering both frequency regulation and charging demands. In [39], a V2G control
method is proposed for EVs to take part in primary frequency control considering
30
charging demands from EV users in a decentralized way. The battery SOC holder
along with adaptive frequency droop control is proposed to maintain the SoC of the
EV battery around the residual SOC. A V2G reactive power support strategy for
optimal coordinated voltage regulation in distribution networks with high DG
penetration is proposed in [40]. The proposed algorithm employs EVs, DGs, and on-
load tap changers (OLTCs) to satisfy EV charging demands and grid voltage
requirements with relaxed tap operation, and minimum DG active power curtailment.
3) Renewable Supporting and Balancing

The V2G capacities of EVs can provide additional reserve for renewable generation
in power system. When the power injected to the grid from RESs become very high,
either the centralized power plants or the DG units must decrease their power
generation. EVs can help match generation and consumption by charging and
discharging batteries, so that the utility generators do not need to reduce the power
output. A distributed V2G control scheme providing a distributed spinning reserve
for the unexpected intermittency of RESs is proposed in [41]. The EV user’s
requirement is satisfied by performing a scheduled charging for the V2G control.
The authors in [42] aims to examine the potential and technical benefits of using PV
systems as energy suppliers for charging EVs. An urban scale integrated system with
the self-consumption evaluation is presented in order to optimize the interaction
between the PV system and charging station for EVs.
2.2 Voltage Regulation Methods in Distribution Networks
In the last few years, the rapid developments of renewable technology have led to a
significant growth in the amount of DGs in distribution networks. The increasing
penetration of DG units will cause two major voltage problems: voltage rise and drop
issues. With high penetration of rooftop PVs, PV generation may largely exceed
31
customer load demand in the midday. The reverse power flow from customer end to
the utility grid can potentially cause unacceptable voltage rise along the distribution
network [43]-[45]. On the other hand, the massive charging of plug-in electric
vehicles will introduce additional power demand, which may result in unacceptable
voltage drop along the distribution network [46]-[48]. So there is a possibility of
voltage limits violation during both peak load and peak PV generation periods, which
may lead to poor power quality and even equipment failure.
Many techniques have been proposed to deal with the voltage rise/drop issues in
distribution networks. These methods can be employed by both utility operators and
customers. The utility grid facilities have been installed and utilized by operators for
voltage regulation, which include increasing the conductor size [49], installing
voltage regulators (switched capacitors) [50]-[52], and changing the set points of
secondary transformer tap [53], [54]. Voltage regulators and tap-changing
transformers have been widely used for voltage regulation in traditional distributed
networks where the power flow is unidirectional and the load demand is easy to
predict. However, the power flow in distribution networks becomes bidirectional and
the load demand becomes hard to predict due to the penetration of RESs and EVs.
The bidirectional power flow and power variation could cause frequent switching of
the set points of the voltage regulators and transformers. Nowadays, as the fast
development of rooftop PVs and plug-in EVs in residential distribution networks,
power generation and load demand at customer end can also be utilized for voltage
regulation. The methods include curtailing the power of PV generation [55], [56],
reactive power compensation of PV inverters [57], [58], utilizing energy storage
systems [59], [60], and demand response [61], [62]. Table 2.3 summarizes the
features and drawbacks of typical voltage regulation methods as mentioned above.
32
To achieve voltage regulation in modern distribution networks, these voltage

regulation methods can be optimized and controlled according to their features. For
on load tap changing transformers, voltage regulators and switched capacitors which
have the limitations on taps/switches change, the system voltage regulation can be
formulated as an optimization problem. The taps/switches can be optimized by the
grid operators based on the renewable generation and load prediction. The real-time
control can be adopted by DG units to deal with voltage variations caused by the
uncertainty of renewable energy. In this thesis, the research is mainly focused on the
real-time voltage regulation provided by DG units at customer side. Therefore, the
voltage regulation methods from customer side are further discussed as follows.
Table 2.3 Comparison of typical voltage regulation methods
Voltage regulation
Features Drawbacks
methods
Increase the power rating of the Rapid increasing renewable

Increasing the conductor
distribution system, relieve both generation leads to frequently
size [49]
voltage drop/rise upgrading of distribution lines
Installation cost of voltage
Voltage regulators [50]- Mainly used for MV networks,
regulators, not suitable for
[52] limited tap change
frequent tap change.
Frequently changing the
Secondary transformer Change the voltage profiles along
transformer taps reduce the
[53], [54] the distribution feeder directly
transformer lifetime
Utilize reactive power for voltage Less effective in LV
Reactive power
regulation, limited reactive power distribution networks, increase
compensation [57], [58]
capacity power loss
Reduce the production of PV
Power curtailment of Easy to implement, no additional
generation, less effective in MV
DG units [55], [56] cost
distribution networks
Energy storage systems Enhance the production of PV

Installation cost of ESSs
[59], [60] generation, fast response time
Demand response Customers behave as responsive Limited by the types of load at

[61], [62] loads. customer side.
33
A local voltage regulation technique that utilizes 15s PV power forecasts to avoid
upcoming upper voltage limit violation or overvoltage scenario is proposed in [55].
The proposed algorithm employs an active power curtailment based on these PV
power forecasts, when the reactive power estimate given by a droop-based method
cannot provide the desired voltage regulation within predefined power factor limits.
The curtailment threshold values are calculated in such a way that this voltage
regulation technique can reduce possible voltage limit violations. In [56], the use of
droop-based active power curtailment techniques for overvoltage prevention in
radial LV feeders as a means for increasing the installed PV capacity and energy
yield is discussed. Two schemes are proposed and tested in a typical 240V/75kVA
Canadian suburban distribution feeder for 12 houses with roof-top PV systems. The
droop-based curtailment proposed in [55], [56] is decentralized control approach.
Each PV system will curtail the PV output according to only local voltage
measurement. The disadvantage of PV curtailment for voltage regulation is that this
method reduces the production of PV generation.
One of the solutions for voltage rise issue in distribution networks is reactive power
compensation by PV systems. German Grid Codes introduces an active power
dependent standard characteristic curve for inverter-coupled distributed generators.
In the active power dependent method, the reactive power is determined based on
the local feed-in active power of each PV system. [57] utilizes the voltage sensitivity
matrix and quasi-static analysis in order to locally and systematically develop a
coordinated characteristic for every PV system in the distribution network. This
method regulates the feeder voltages under the upper steady-state voltage limit. In
[58], a voltage control method for PV inverters that maintains the feeder voltages
within acceptable ranges by providing and absorbing reactive power is implemented.
This method can be viewed as a form of distributed reactive power based voltage
control, which is conventionally conducted by coordinated control of capacitor based
34
voltage regulators and tap changing transformer. The reactive power compensation
for voltage regulation is originally applied in transmission system where the line
reactance X is much larger than resistance R. However, R/X ratio in LV distribution
networks is much larger than that in transmission networks, which makes reactive
power compensation less effective.
As the cost of adding batteries to residential PV systems has continued dropping in

recent years, ESSs become a promising solution for voltage regulation in distribution
networks. In [59], a coordinated control of distributed ESS with traditional voltage
regulators including the OLTC and step voltage regulators is proposed to solve the
voltage rise problem caused by the high PV penetration in the LV distribution
network. The main objective of this coordinated control is to relieve the tap changer
transformer operation stress, shave the distribution network peak load and decrease
the transmission and distribution resistive power losses under high solar power
penetration. The author in [60] presents utilization of PVs and battery ESSs
coordinately to address voltage rise/drop issues. The power and capacity of PV
inverters together with droop-based ESSs is evaluated for urban and rural
distribution feeders with different R/X ratios. The results demonstrate that for the
urban feeders with low resistance, reactive compensation of PV inverters is sufficient
to maintain acceptable voltage profile. But for the rural feeders with high resistance,
the power outputs of both PV inverters and ESSs are required.
Demand response at customer side has traditionally targeted peak shaving problem
for the optimal allocation of power consumption on a time scale that ranges from
minutes to hours. As the fast development of advanced monitoring and
communication infrastructure, the real-time demand response has potential to
provide ancillary services for distribution networks. In [61], a large-scale ice-thermal
storage has been viewed as a smart load for fast voltage control and demand side
35
management in renewable energy penetrated power systems. The possibility of

modifying a conventional thermal load has been practically demonstrated in a
refrigerator using power electronics technology. The results have demonstrated that
voltage and frequency fluctuations can be mitigated using ice-thermal storage in
urban buildings in weak power grids. In [62], a decentralized demand response
control mechanism has been proposed to provide ancillary services to the utility grid
using a large amount of elastic appliances. The voltage control requirement of the
grid can be continuously supported by broadcasting low-bit rate control signals.
2.3 Control Hierarchy of Power System
With respect to the control architecture of a power system, there are two distinctive
and opposite approaches: centralized and decentralized controls. A fully centralized
control depends much on the performance of the central controller which does the
required calculations and decides the control behaviors for all the controllable units,
which requires extensive communication between the central controller and
controllable units [63]. On the other side, a fully decentralized control controls each
unit only by the local controller, measuring local information and is neither fully
aware of system-wide parameters nor other controllers’ behaviors [64]. As power
systems usually cover large geographic areas, it is impractical to implement a fully
centralized control due to the extensive communication and computation
requirements. Due to the strong connection between the operations of controllable
units in the system, a fully decentralized control is also not possible. The hierarchical
control scheme, consisting of three control levels: primary, secondary and tertiary, is
a compromise approach between fully centralized and fully decentralized control
schemes. The differences between each control level are response speed and time
scale where they need to operate, and infrastructure requirements. Although
distribution networks or microgrids are not necessary as geographically large as
36
conventional power systems, they can also benefit from this control hierarchy
because of the large number of controllable DG units which have power electronics
interface with fast transient performance [65]-[67]. The control hierarchy of power
system with inverter based DG units is shown in Figure 2.2. The rest of this section
presents an overview of each level in the control hierarchy.
Tertiary Control
... ...
Response Time
Secondary Secondary Secondary

Control Control Control
DG-1 DG-n DG-1 DG-1 DG-n

DG-n
DG-2 DG-2
DG-2
Primary Control Primary Control Primary Control
Figure 2.2 Control hierarchy of power electronic interfaced power system.
2.3.1 Primary Control
Primary control is the first level in the control hierarchy which features the fastest
dynamic response compared to the other control levels. The primary controller only
depends on local measurements and has no communication requirements. Controller
speed requirement, islanding detection, grid synchronization, output control and
power sharing control are included in this category [68], [69]. Voltage source
inverters as the interface for DG units or as a part of back-to-back converters, require
a specially designed control to simulate the inertia characteristic of synchronous
generators and provide appropriate power output. For this purpose, primary control
37
of power converters is composed of two stages: inverter output control and power
sharing control.
The first stage of primary control is inverter output control which normally consists
of an outer voltage control loop and an inner current regulation loop. The primary
control of electronically coupled DG units in power systems has become an
important research topic. A general overview of inverter output control is given in
[70], [71], in which power inverter controllers are categorized into three kinds based
on their reference frame: synchronous (dq), stationary (αβ) and natural (abc). The
synchronous reference frame is linked with dc variables and proportional-integral
(PI) based controllers. The stationary reference frame is related to sinusoidal
variables and proportional-resonant (PR) controllers. The natural reference frame
used controllers realized in the form of PI, PR, hysteresis, or deadbeat [70].
Depending on their operation in power systems, the control of inverters can be
classified into voltage control mode and current control mode. The inverters in VCM
are operated as ac voltage sources and controlled to supply the reference voltage
amplitude and frequency. The inverters in CCM are operated as current sources and
controlled to supply the reference real and reactive power. These power inverters are
suitable to operate in parallel with other inverters in power systems. Particularly, the
inverters for PV and wind turbine will operate in CCM to provide their maximum
power at most of their operation time [72].
Recently, MPC as an alternative method for inverter output control has been widely
investigated in literature [73]-[77]. Compared with other control methods, MPC
offers many advantages, for example it can have a variety of control objectives, be
simply implemented to muti-input muti-output systems, have a good dynamic
response, and easily include nonlinearities and constraints in the design of the
controller [73]. MPC applied to power electronics can be classified into two main
38
categories: continuous control set MPC and finite control set MPC. In order to
simplify the implementation of MPC, finite control set MPC is usually applied
because it takes advantage of the inherent discrete nature of power converters. The
finite control set MPC controls select the switching signals of power converters only
from a finite set of switching states. The prediction based on state-space model of
each step is applied to minimize a cost function, and the switching state with
minimum value of cost function is chosen. This control method has already been
widely utilized for power converters and motor drive applications [74]. In [75], the
state-of-art development of finite control set MPC in power electronics has been
reviewed in detail. Some applications of this technology including drives, active
filters, power conditioning, distributed generation and renewable energy are covered.
In [76], the authors show that the predictive control framework allows one to
embellish classical linear controllers (PI, PR) with novel features and design new
control architectures to address various challenges posed by power electronics
applications. In [77], an MPC strategy for a grid connected three-level neutral point
clamped converter is presented. An LCL filter is used to reduce the harmonics
generated by the neutral point clamped inverter.
The second stage of the primary control is known as the power sharing control.
Different from inverter output control, the power sharing control is responsible for
sharing the active and reactive power imbalance properly in the power systems [68],
[70], [78]. The second stage of the primary control can be classified depends on
whether or not the concept of droop is performed. The power sharing control which
is implemented by a centralized control scheme can also be viewed as a part of the
secondary control. The idea of P/f and Q/V droop control method is initially from the
steady-state operation of conventional synchronous generators, and the droop control
method can be artificially applied for electronically interfaced DG units. The external
droop control loop functions to share active and reactive power among DG units.
39
Besides, the droop control can improve the system stability and performance and
adjust both the frequency and amplitude of the output voltage without
communication [79]. The droop control relationship can be expressed as
     m( P  P ) (2.1)
V  V   n(Q  Q ) (2.2)
where ω and V are the measured output frequency and voltage, ω* and V* are their
references. m and n coefficients define the droop slopes are based on steady-state
performance criteria [80]. P and Q are calculated active and reactive power, P* and
Q* are their references. The main limitation of conventional droop control method is
that there will be frequency and voltage deviations in the microgrid. To eliminate the
deviation caused by droop method, communication between each unit is required for
higher level control [81], [82].
2.3.2 Secondary Control
Secondary control, also known as energy management system (EMS), is designed

for the stable and efficient operation of microgrids or distribution networks in both
grid-connected and islanded condition. Secondary control is the highest level in the
control hierarchy for a single system operating in islanded condition. Secondary
control usually operates for a longer time scale as compared to the primary control
which allows enough time to perform complex calculations. The communication
bandwidth requirement can be reduced by using sampled measurements of the
system variables. The low bandwidth communication can be easily achieved in
distribution networks or microgrids with limited geographical area with standard
communication protocols, which only requires slowly changing parameters such as
set point for real and reactive power [83]. Generally, the functions of secondary
40
control include energy management, frequency and voltage restoration, increasing

power sharing accuracy and power quality enhancement.
The secondary control can be realized in a centralized or distributed architecture.

The implementation of a centralized architecture needs all the required control data
and information can be collected simultaneously and optimized at a central
controller. Furthermore, the centralized EMS does not provide the DG units with
plug and play capability. On the contrary, a distributed architecture allows new DG
units to be integrated into distribution grid easily without requiring any changes to
the control settings. However, the employment of a distributed controller requires
proper algorithms to coordinate between the DG units. The control architectures of
centralized and distributed secondary control are shown in Figure 2.3. The
centralized and distributed control architectures to realize secondary control are
described in detail in following subsections.
Power System Power System
DG-1 DG-2 ... DG-i ... DG-n DG-1 DG-2 ... DG-i ... DG-n
Agent-1 Agent-2 ... Agent-i ... Agent-n

Central controller
(a) Centralized control (b) Distributed control
Figure 2.3 Control architectures of centralized and distributed secondary control.
1) Centralized Architecture
In the approach to realize a centralized secondary control, important information
such as the generation capacities and the operating limits of DG units, load
41
measurements from smart meters and energy sensors, transformer tap positions and
circuit breaker status are all transmitted to a central controller for monitoring and
processing either through wireline or wireless communication technologies. Based
on the information gathered, the central controller will compute an appropriate
optimal operating condition for the microgrid and generate the reference power and
voltage set-points for the primary controllers of all the DG units.
The centralized energy management of microgrid is usually formulated into a multi-

objective optimization problem and different techniques have been proposed in
several research works [84], [85]. To address the renewable penetration into
microgrid, the authors in [84] propose the off-line optimization approach to achieve
energy management for a single microgrid system. To minimize the total energy cost
of the conventional energy drawn from the main grid over a finite horizon, this
approach optimizes the energy charged/discharged by ESSs over time subject to
practical load and storage constraints. A centralized energy management method for
a microgrid that includes RESs, diesel generators, battery ESSs and various loads is
proposed in [85]. To address various uncertainties, a risk-constrained scenario-based
stochastic programming framework is proposed using the conditional value at risk
method. The designed model is solved by two levels of stochastic optimization
methods.
There are also many research works focusing on grid ancillary services using the
secondary central controller [86]-[88]. A centralized control system that coordinates
parallel operations of different DG inverters within a microgrid is proposed in [86].
A novel MPC algorithm is employed by the control design for the DG inverters,
which allows faster computational time for large power systems by optimizing the
steady-state and the transient control problems separately. Harmonic compensation,
load sharing, voltage regulation and fault level increase is achieved by coordinate
42
control of different DG units during both grid-connected and islanded operations. In

order to have a reliable communication with minimal end-to-end delay during the
event of node failure, a find reliable link scheme is proposed in [87]. The low
bandwidth communication among the distributed generators and control center can
overcome the system operation challenges posed by network line parameters, failure
of distributed generators and power shortage in the system. In [88], a vehicle-to-grid
reactive power support strategy for optimal coordinated voltage regulation in
distribution networks with high DG penetration is presented. The voltage regulation
problem is formulated as a nonlinear programming in the centralized control and
consists of three consecutive stages, in which successive stages apply the outputs of
their preceding stages as constraints.
2) Distributed Architecture
The decentralized secondary control aims to manage the energy in a microgrid while
allowing all DG units and loads to operate independently. Although the decentralized
control approach still employs a hierarchical architecture consisting of a central
controller and several local controllers for the exchange of data, the key feature that
distinguishes itself from the centralized control approach is that the decisions on the
control variables are made at the local controllers. The central controller is
responsible for managing the overall operation of the microgrid together with the
distribution grid to achieve economical and reliable operation while the task of the
local controllers is to control individual DG units or an aggregation of them to meet
both local and global objectives by communicating with other local controllers and
the central controller [89].
To achieve the distributed secondary control architecture, the multi-agent system
(MAS) framework has been proposed in many research works [90]-[92]. In the
MAS, large complex control tasks are decomposed into simpler and manageable
ones, and are then allocated to various intelligent agents which are provided with
43
local information to achieve multiple local and global objectives, without any
intervention from the central controller. The distributed management of building
energy system with MAS control architecture is proposed in [90]. This method
utilizes a semi-centralized decision-making methodology for battery energy
management system for electrical, heating, and cooling energy zones with combined
heat and power system optimizations aimed at improving energy efficiency and
reducing energy costs. In [91], DG units and price-sensitive loads are proposed to
participate in a microgrid energy market. Each DG unit and price-sensitive load are
controlled by the respective agents which performs various functions such as
scheduling, coordination and market clearing subject to the DG unit and load
constraints. Every agent is assigned to one of objectives of other agents to maximize
DG unit and load surpluses. The authors in [92] illustrate the application of a MAS-
based scheme for the optimization problem in power system. The voltage regulation
problem is solved by dispatching the power outputs of DG units on a distribution
feeder using MAS based control net protocol. Fast communication speed among the
DGs is required to ensure fast response time in emergency conditions.
2.3.3 Tertiary Control
Tertiary control is the highest level of control and to manage the operation of the
microgrid in grid-connected condition. Depending on the requirements of the utility
grid, tertiary control sets long-term and optimal set points for lower level controller.
The tertiary control is responsible for coordinating the operations of several
microgrids interacting with each other in the system, and communicating
requirements like unit commitment, economic dispatch, optimal power flow, voltage
support and frequency regulation from the main power grid. This control level
typically operates in tens of minutes, providing signals to secondary level controls at
microgrids and other subsystems that form the full grid. Tertiary control can be
44
considered as a part of the main utility grid, and not the microgrid. This control level
is not discussed further in this thesis [93].
2.4 Consensus Algorithm
Consensus problems have a long history in computer science and form the
foundation of the field of distributed computing. Formal study of consensus
problems in groups of experts was originated in management science and statistics
in 1960s [94]. In recent years, the consensus problem of MAS has received
increasingly attention. This is partly due to the applications of MAS in many areas
such as cooperative control of automatic vehicles, flocking of birds, distributed
sensor networks, attitude alignment for cluster of satellites and congestion control in
communication networks [95]. When multiple agents agree to the value of a variable
of interest, they are said to have reached consensus. Information consensus
guarantees that agents sharing information over a communication network topology
have a consistent view of information that is critical to the coordination task. To
achieve consensus, there must be a shared variable of interest, called the information
state, as well as appropriate algorithmic methods for negotiating to reach consensus
on the value of that variable, called the consensus algorithms. In general, consensus
algorithms are designed to be distributed, assuming only neighbor-to-neighbor
interaction between each agent. Each agent updates the value of their information
states based on the information states of their neighbors. The goal is to design an
updated control law so that the information states of all the agents in the network
converge to a common value. The graph theory and consensus algorithm are
introduced as follows:
The communication network of a multi-agent system can be described by a graph,

which is defined as G = (V, E) with a set of N nodes V = {v1, v2, …,vN} and a set of
45
edges E = V × V. Each node is assigned with an agent in the system, and edges
represent communication links for data exchange. If communication links are
bidirectional, (vi, vj) ∈ E ⇒ (vj, vi) ∈ E ∀i, j, the graph is said to be undirected,
otherwise it is directed. The set of neighbors of ith node is denoted Ni. Equivalently,
if j ∈ Ni, then vi receives information from vj. A matrix called adjacency matrix A =
[aij] is associated to the edges. Aij represents the weight for information exchanged
between agents i and j, where aij > 0 if agents i and j are connected through an edge
(vi, vj) ∈ E, otherwise, aij=0. The Laplacian matrix L is defined as L= D - A, where
D is called in-degree matrix and is defined as D = diag{di} with di   jN aij . A
directed path form node i to node j is a sequence of edges. A leader node can be
added to the graph and connected to at least one node by unidirectional edges. The
nodes connected to the leader node and the corresponding connecting edge are called
pinned nodes and pinning edge respectively [96], [97].
In the cooperative synchronization problem, also called leaderless consensus, all

agents synchronize to a common value that is not prescribed. Each agent updates the
information state with the most common continuous-time consensus algorithm given
by
xi (t )   a (t )( x (t )  x (t ))
j N i
ij j i (2.3)
where i=1,2,…, n; n is the number of agents in the system; xi and xj are the
information states of agent i and its neighboring agent j.
While in the tracking synchronization problem, also called pinning control, all agents
synchronize to leader nodes that act as a command generator. The consensus
algorithm updates the information state for both pinned and unpinned agents as
46
n
xi (t )   aij (t )( x j (t )  xi (t ))  gi ( xref (t )  xi (t )), i  1,..., n. (2.4)
j 1
where xref and gi are the local set point and pinning gain of the ith agent respectively.
The pinning gain gi is positive if the agent i is pinned by a leader node, and is zero
otherwise.
Consensus based distributed control has been adopted by many research works in
controlling the DG units in distribution networks [98]-[101]. In [98], a fully
distributed multi agent based load restoration algorithm is proposed. According to
the algorithm, each agent makes synchronized load restoration decision according to
discovered information. The author extends his work and apply the consensus based
distributed control for optimal charging rate control of plug-in EVs in [99]. The
proposed strategy is implemented based on a multi-agent system framework, which
is convenient for plug and play operation which allows EVs to join and leave at
arbitrary times. In [100], a distributed control algorithm that controls the power
output of multiple PV generators in a distribution network is developed. The
consensus based cooperative control method is applied to make distributed PV
generators converge and operate at a certain ratio of power outputs, which is
determined by the conditions of the distribution network. In [101], a distributed
control and generation estimation approach is developed to dispatch multiple DGs,
each of which consists of a PV and a controllable load. A distributed weights
adjustment algorithm adaptively makes the adjacency matrix doubly stochastic so
that the aggregated power generation capacity can be estimated. Then, the expected
consensus operational point of the DGs is calculated by those DGs that can obtain
power dispatch command from the supervisory control and data acquisition system
and is propagated to the rest of the DGs with a consensus algorithm.
The consensus algorithm has also been widely applied for secondary and tertiary
47
control in microgrids in many research works [102]-[104]. In [102], a distributed

control approach is proposed for the voltage restoration which enables the voltages
of all the DG units converging to the reference values in finite time. In the meantime,
frequency restoration is realized by a consensus-based distributed frequency control,
considering constraints of certain control input. A distributed hierarchical control
framework to ensure reliable operation of dc microgrid clusters is proposed in [103].
In the secondary level of the hierarchical control, a distributed consensus-based
voltage regulator is introduced to eliminate the average voltage deviation in the
microgrids. Then another distributed policy is employed to regulate the power flow
among the microgrids according to their local SOCs. A distributed tertiary control
method is proposed to handle power sharing among a cluster of dc microgrids in
[104]. The load sharing among the DERs within a dc microgrid is managed by
primary and secondary control. The tertiary level control is required to provide the
higher level load sharing among multiple microgrids within a cluster. The proposed
control method applies a cooperative approach to adjust voltage set points for each
individual microgrid and navigate the power flow among them accordingly. An
updating policy using load mismatch among neighbor microgrids is also proposed to
adjust voltage set points and mitigate such mismatches.
2.5 Summary
The integration of power electronic interfaced DG units into the existing power
systems will continue increasing in the near future due to rapid advancement of
technological and liberalization of the electricity market. However, the high
penetration of RESs and EVs with intermittent and stochastic characteristics will
bring along many stability, reliability and power quality issues in future power
systems. The main challenges that should be taken into consideration include, but
not limited to, the stability problems caused by low inertia of power electronic
interfaced systems, the reliability of DG units and communication based control
48
strategies, and the power quality issues caused by the integration of DG units.
Particularly, the power generation and load demand mismatch caused by the
variability of RESs will lead to severe frequency and voltage control problems.
Therefore, the development of suitable control strategies is essential to ensure
reliable and stable operation of power systems. Furthermore, the coordinated control
of large amount of DG units having different operating characteristics and
constraints will pose a great challenge due to limited communication and possible
conflicting operating requirements between the different types of DG units.
In this chapter, many research works related to increasing the integration of

renewable energy into the existing power systems and the power quality issues
caused by high renewable penetration have been reviewed. However, there are still
limitations and gaps of existing works which require further research attention.
When the penetration rate of RESs is further increasing at different voltage levels of
power systems, there will be various problems to be solved. At transmission voltage
level, the high penetration of RESs will bring new frequency and stability issues. At
distribution level, the load profiles of demand side as well as the voltage profiles of
distribution networks will change dramatically due to the increasing penetration of
DG units. Especially for LV residential networks, the system R/X ratio is high and
bus voltages are very sensitive to real power output, which make the voltage
regulation methods quite different compared with high voltage levels. The research
work of this thesis is focused on voltage regulation issues in residential LV networks
and it is a challenge to promote future renewable penetration. In addition, the battery
characteristics of ESSs and EVs for voltage regulation application have not been
considered by many research works so far. Therefore, proper battery models need to
be built and battery parameters such as SoC, efficiency, self-discharge rate, state-of-
health, lifetime and cost should be further considered for grid application. Specific
control strategies should be designed considering operation constraints and limits of
49
ESSs and EVs. Furthermore, most research works on voltage regulation problems so
far have been achieved in decentralized or centralized control architecture. The main
limitation of decentralized control is the lack of global information such that it is
hard to achieve coordination between each controllable unit. The application of
centralized control architecture for real-time control also faces some drawbacks,
such as communication burden and reliability of central controller. As a compromise
of decentralized and centralized control, distributed control has become a promising
solution for real-time control of DG units which can achieve coordination between
DG units with sparse communication. As a typical method to achieve distributed
control, consensus algorithm for multi-agent systems has been discussed and adopted
in this thesis.
References
[1] J. N. Baker and A. Collinson, “Electrical energy storage at the turn of the
millennium,” Power Engineering Journal, vol. 13, no. 3, pp. 107–112, Jun.
1999.
[2] H. Chen, T. N. Cong, W. Yang, C. Tan, Y. Li, and Y. Ding, “Progress in
electrical energy storage system: a critical review,” Progress in Natural
Science, vol. 19, no. 3, pp. 291–312, Mar. 2009.
[3] S. Vazquez, S. M. Lukic, E. Galvan, L. G. Franquelo, and J. M. Carrasco,
“Energy storage systems for transport and grid applications,” IEEE Trans.
Industrial Electronics, vol. 57, no. 12, pp.3881–3895, Dec. 2010.
[4] E. Maza, “Grid-connected storage market set to explode,” Energy Storage
Association, Jan. 2014.
[5] M. Yilmaz and P. T. Krein, “Review of the impact of vehicle-to-grid
technologies on distribution systems and utility interfaces,” IEEE Trans.
Power Electronics, vol. 28, no. 12, pp.5673–5689, Nov. 2012.
[6] Energy storage opportunities and challenges. Ecofys, 2014.
50
[7] S. M. Schoenung, “Characteristics and technologies for long-term vs. short-

term energy storage: a study by the DOE energy storage systems program,”
Sandia National Laboratories, SAND2001-0765, Mar. 2001.
[8] I. Gyuk, M. Johnson, J. Vetrano, K. Lynn, et al., “Grid energy storage,” U.S
Department of Energy, Dec. 2013.
[9] Grid integration of large-capacity renewable energy sources and use of large-
capacity electrical energy storage. International Electrotechnical Commission,
2012.
[10] H. Ibrahim, A. Ilinca, and J. Perron, “Energy storage systems-characteristics
and comparisons,” Renewable Sustainable Energy Reviews, vol. 12, no. 5, pp.
1221–1250, Jun. 2008.
[11] M. Mohamed, P. Leung, and M. Sulaiman, “Performance characterization of a
vanadium redox flow battery at different operating parameters under a
standardized test-bed system,” Applied Energy, vol. 137, pp. 402–412, Jan.
2015.
[12] T. A. Nguyen, M. L. Crow, and A. C. Elmore, “Optimal sizing of a vanadium
redox battery system for microgrid systems,” IEEE Trans. Sustainable Energy,
vol. 6, no. 3, pp. 729–737, Apr. 2015.
[13] J. Jaguemont, L. Boulon, and Y. Dube, “Characterization and modeling of a
hybrid-electric-vehicle lithium-ion battery pack at low temperatures,” IEEE
Trans. Vehicular Technology, vol. 65, no. 1, pp. 1–14, Jan. 2015.
[14] A. Hoke, A. Brissette, K. Smith, A. Pratt, et al., “Accounting for lithium-ion
battery degradation in electric vehicle charging optimization,” IEEE Journal
of Emerging and Selected Topics in Power Electronics, vol. 2, no. 3, pp. 691–
700, Apr. 2014.
[15] V. A. Boicea, “Energy storage technologies: The past and the present,”
Proceedings of the IEEE, vol. 102, no. 11, pp. 1777-1794, Nov. 2014.
51
[16] Y. Zhang, C. H. Zhang, and X. F. Zhang, “State-of-charge estimation of the

lithium-ion battery system with time-varying parameter for hybrid electric
vehicles,” IET Control Theory & Applications, vol. 8, no. 3, pp. 160–167, Feb.
2014.
[17] Z. Chen, Y. H. Fu, and C. C. Mi, “State of charge estimation of lithium-ion
batteries in electric drive vehicles using extended kalman filtering,” IEEE
Trans. Vehicular Technology , vol. 62, no. 3, pp. 1020–1030, Dec. 2012.
[18] M. J. E. Alam, K. M. Muttaqi, and D. Sutanto, “Mitigation of rooftop solar PV
impacts and evening peak support by managing available capacity of
distributed energy storage systems,” IEEE Trans. Power Systems, vol. 28, no.
4, pp. 3874–3884, Nov. 2013.
[19] G. S. Wang, M. Ciobotaru, and V. G. Agelidis, “Power smoothing of large
solar PV plant using hybrid energy storage,” IEEE Trans. Sustainable Energy,
vol. 5, no. 3, pp. 834–842, Mar. 2014.
[20] X. J. Li, D. Hui, and X. K. Lai, “Battery energy storage station (BESS)-based
smoothing control of photovoltaic (PV) and wind power generation
fluctuations,” IEEE Trans. Sustainable Energy, vol. 4, no. 2, pp. 464–473,
Mar. 2013.
[21] I. Serban, R. Teodorescu, and C. Marinescu, “Energy storage systems impact
on the short-term frequency stability of distributed autonomous microgrids, an
analysis using aggregate models,” IET Renewable Power Generation, vol. 7,
no. 5, pp. 531–539, Sep. 2013.
[22] J. Dang, J. Seuss, L. Suneja, and R. G. Harley, “SoC feedback control for wind
and ESS hybrid power system frequency regulation,” IEEE Journal of
Emerging and Selected Topics in Power Electronics, vol. 2, no. 1, pp. 79–86,
Nov. 2013.
52
[23] L. Miao, J. Y. Wen, C. Y. Yue, et al., “Coordinated control strategy of wind

turbine generator and energy storage equipment for frequency support,” IEEE
Trans. Industry Applications, vol. 51, no. 4, pp. 2732–2742, Jan. 2013.
[24] M. A. Abusara, J. M. Guerrero, and S. M. Sharkh, “Line-interactive UPS for
microgrids,” IEEE Trans. Industrial Electronics, vol. 61, no. 3, pp. 1292–
1300, May 2013.
[25] J. Y. Kim, J. H. Jeon, S. K. Kim, C. H. Cho, et al., “Cooperative control
strategy of energy storage system and microsources for stabilizing the
microgrid during islanded operation,” IEEE Trans. Power Electronics, vol. 25,
no. 12, pp. 3037–3048, Sep. 2010.
[26] H. Kakigano, Y. Miura, and T. Ise, “Distribution voltage control for DC
microgrids using fuzzy control and gain-scheduling technique,” IEEE Trans.
Power Electronics, vol. 28, no. 5, pp. 2246–2258, Sep. 2012.
[27] X. N. Lu, K. Sun, J. M. Guerrero, J. C. Vasquez, et al., “State-of-charge
balance using adaptive droop control for distributed energy storage systems in
DC microgrid applications,” IEEE Trans. Industrial Electronics, vol. 61, no.
6, pp. 2804–2815, Aug. 2013.
[28] H. K. Hguyen, J. B. Song, and Z. Han, “Distributed demand side management
with energy storage in smart grid,” IEEE Trans. Parallel and Distributed
Systems, vol. 26, no. 12, pp. 3346–3357, Nov. 2014.
[29] M. Mahmoodi, P. Shamsi, and B. Fahimi, “Economic dispatch of a hybrid
microgrid with distributed energy storage,” IEEE Trans. Smart Grid, vol. 6,
no. 6, pp. 2607–2614, Feb. 2015.
[30] S. U. Agamah and L. Ekonomou, “Peak demand shaving and load-levelling
using a combination of bin packing and subset sum algorithms for electrical
energy storage system scheduling,” IET Science, Measurement & Technology,
vol. 10, no. 5, pp. 477–484, Jul. 2016.
53
[31] Y. Yang, H. Li, A. Aichhorn, J. P. Zheng, et al., “Sizing strategy of distributed

battery storage system with high penetration of photovoltaic for voltage
regulation and peak load shaving,” IEEE Trans. Smart Grid, vol. 5, no. 2, pp.
982–991, Sep. 2013.
[32] A. S. A. Awad, T. H. M. El-fouly, and M. M. A. Salama, “Optimal ESS
allocation for benefit maximization in distribution networks,” IEEE Trans.
Smart Grid, in press.
[33] D. Wu, D. C. Aliprantis, and L. Ying, “Load scheduling and dispatch for
aggregators of plug-in electric vehicles,” IEEE Trans. Smart Grid, vol. 3, no.
1, pp. 368–376, Mar. 2012.
[34] SAE electric vehicle and plug-in hybrid electric vehicle conductive charge
coupler, SAE Standard J1772, Jan. 2010.
[35] M. J. E. Alam, K. M. Muttaqi, and D. Sutanto, “A controllable local peak-
shaving strategy for effective utilization of PEV battery capacity for
distribution network support,” IEEE Trans. Industry Applications, vol. 51, no.
3, pp. 2030–2037, Nov. 2014.
[36] A. Schuller, et al., “Charging strategies for battery electric vehicles: economic
benchmark and V2G potential,” IEEE Trans. Power System, vol. 29, no. 5, pp.
2014-2022, Sep. 2014.
[37] M. Singh, P. Kumar, and I. Kar, “A multi charging station for electric vehicles
and its utilization for load management and the grid support,” IEEE Trans.
Smart Grid, vol. 4, no. 2, pp. 1026–1037, Jan. 2013.
[38] H. Liu, Z. H. Hu, Y. H. Song, J. H. Wang, et al., “Vehicle-to-grid control for
supplementary frequency regulation considering charging demands,” IEEE
Trans. Power Systems, vol. 30, no. 6, pp. 3110–3119, Dec. 2014.
[39] H. Liu, Z. H. Hu, Y. H. Song, and J. Lin, “Decentralized vehicle-to-grid control
for primary frequency regulation considering charging demands,” IEEE Trans.
Power Systems, vol. 28, no. 3, pp. 3480–3489, Mar. 2013.
54
[40] M. A. Azzouz, M. F. Shaaban, and E. F. El-Saadany, “Real-time optimal

voltage regulation for distribution networks incorporating high penetration of
PEVs,” IEEE Trans. Power Systems, vol. 30, no. 6, pp. 3234–3245, Jan. 2015.
[41] Y. Ota, H. Taniguchi, T. Nakajima, K. M. Liyanage, et al., “Autonomous
distributed V2G (vehicle-to-grid) satisfying scheduled charging,” IEEE Trans.
Smart Grid, vol. 3, no. 1, pp. 559-564, Oct. 2011.
[42] S. Weckx and J. Driesen, “Load balancing with EV chargers and PV inverters
in unbalanced distribution grids,” IEEE Trans. Sustainable Energy, vol. 6, no.
2, pp. 635–643, Mar. 2015.
[43] R. Tonkoski, D. Turcotte, and T. H. M. El-Fouly, “Impact of high PV
penetration on voltage profiles in residential neighborhoods,” IEEE Trans.
Sustainable Energy, vol. 3, no. 3, pp. 518-527, Jul. 2012.
[44] M. Hasheminamin, V. G. Vassilios, V. Salehi, R. Teodorescu, et al., “Index-
based assessment of voltage rise and reverse power flow phenomena in a
distribution feeder under high PV penetration,” IEEE Journal of Photovoltaics,
vol. 5, no. 4, pp. 1158-1168, May 2015.
[45] G. G. Pillai, G. A. Putrus, and N. M. Pearsall, “Impact of distribution network
voltage rise on PV system energy yield,” in Proc. 2013 Annual IEEE India
Conference (INDICON), Dec. 2013, pp. 1-5.
[46] C. Zhang, C. Chen, J. W. Sun, P. W. Zheng, et al., “Impacts of electric vehicles
on the transient voltage stability of distribution network and the study of
improvement measures,” in Proc. 2014 IEEE PES Asia-Pacific Power and
Energy Engineering Conference (APPEEC), pp. 1-6, Dec. 2014.
[47] X. Luo and K. W. Chan, “Real-time scheduling of electric vehicles charging
in low-voltage residential distribution systems to minimise power losses and
improve voltage profile,” IET Generation, Transmission and Distribution, vol.
8, no. 3, pp. 516-529, Mar. 2014.
55
[48] P. Richardson, D. Flynn, and A. Keane, “Optimal charging of electric vehicles

in low-voltage distribution systems,” IEEE Trans. Power Systems, vol. 27, no.
1, pp. 268-279, Jun. 2011.
[49] C. L. Masters, “Voltage rise: The big issue when connecting embedded
generation to long 11kV overhead lines,” Power Engineering Journal, vol. 16,
pp. 5-12, 2002.
[50] M. E. Elkhatib, R. El Shatshat, and M. M. A. Salama, “Optimal control of
voltage regulators for multiple feeders,” IEEE Trans. Power Delivery, vol. 25,
no. 4, pp. 2670-2675, Dec. 2009.
[51] H. Ravindra, M. O. Faruque, K. Schoder, M. Steurer, et al., “Dynamic
interactions between distribution network voltage regulators for large and
distributed PV plants,” in Proc. Transmission and Distribution Conference
and Exposition (T&D), May 2012, pp. 1-8.
[52] S. Kongtripop, T. Kasirawat, C. Pongsriwat, and S. Premrudeepreechacharn,
“Effect of voltage variation from distributed generation on very long
distribution line with multiple voltage regulators,” in Proc. 2009 Asia-Pacific
Power and Energy Engineering Conference, Mar. 2009, pp. 1-4.
[53] M. Todorovski, “Transformer voltage regulation-compact expression
dependent on tap position and primary/secondary voltage,” IEEE Trans.
Power Delivery, vol. 29, no. 3, pp. 1516-1517, Apr. 2014.
[54] B. A. Robbins, H. Zhu, and A. D. Dominguez-Garcia, “Optimal tap setting of
voltage regulation transformers in unbalanced distribution dystems,” IEEE
Trans. Power Systems, vol. 31, no.1, pp. 256-267, Feb. 2015.
[55] S. Ghosh, S. Rahman, and M. Pipattanasomporn, “Distribution voltage
regulation through active power curtailment with PV inverters and solar
generation forecasts,” IEEE Trans. Sustainable Energy, in press.
[56] R. Tonkoski, L. A. C. Lopes, and T. H. M. El-Fouly, “Droop-based active
power curtailment for overvoltage prevention in grid connected PV inverters,”
56
in Proc. 2010 IEEE International Symposium on Industrial Electronics, pp.

2388-2393, Jul. 2010.
[57] P. M. S. Carvalho, P. F. Correia, and L. A. F. Ferreira, “Distributed reactive
power generation control for voltage rise mitigation in distribution networks,”
IEEE Trans. Power Systems, vol. 23, no. 2, pp. 766–772, May 2008.
[58] P. Jahangiri and D. C. Aliprantis, “Distributed volt/var control by PV inverters,”
IEEE Trans. Power Systems, vol. 28, no. 3, pp. 3429-3439, Aug. 2013.
[59] X. H. Liu, A. Aichhorn, L. M. Liu, and H. Liu, “Coordinated control of
distributed energy storage system with tap changer transformers for voltage
rise mitigation under high photovoltaic penetration,” IEEE Trans. Smart Grid,
vol. 3, no. 2, pp. 897-906, Feb. 2012
[60] M. N. Kabir, Y. Mishra, G. Ledwich, Z. Y. Dong, et al., “Coordinated control
of grid-connected photovoltaic reactive power and battery energy storage
systems to improve the voltage profile of a residential distribution feeder,”
IEEE Trans. Industrial Informatics, vol. 10, no. 2, pp. 967-977, Jan 2014.
[61] X. Luo, C. K. Lee, W. M. Ng, et. al, “Use of adaptive thermal storage system
as smart load for voltage control and demand response,” to be published IEEE
Trans. Smart Grid.
[62] K. Christakou, D. C. Tomozei, J. Y. Leboudec, and M. Paolone, “GECN:
primary voltage control for active distribution networks via real-time demand-
response,” IEEE Trans. Smart Grid, vol. 5, no. 2, pp. 622-631, Mar. 2014.
[63] N. Pogaku, M. Prodanovic, and T. C. Green, “Modeling, analysis and testing
of autonomous operation of an inverter-based microgrid,” IEEE Trans.
Industrial Electronics, vol. 22, no. 2, pp. 613-625, Mar. 2007.
[64] M. D. Ilic and S. X. Liu, Hierarchical Power Systems Control: Its Value in a
Charging Industry (Advances in Industrial Control). London: Springer, 1966.
57
[65] Y. A.-R. I. Mohamed and A. A. Radwan, “Hierarchical control system for

robust microgrid operation and seamless mode transfer in active distribution
systems,” IEEE Trans. Smart Grid, vol. 2, no. 2, pp. 352-362, Jan. 2011.
[66] A. Bidram and A. Davoudi, “Hierarchical structure of microgrids control
system,” IEEE Trans. Smart Grid, vol. 3, no. 4, pp. 1963-1976, Dec. 2012.
[67] R. Majumder, G. Bag, and K. H. Kim, “Power sharing and control in
distributed generation with wireless sensor networks,” IEEE Trans. Smart
Grid, vol. 3, no. 2, pp. 618-634, Jun. 2012.
[68] F. Katiraei, M. R. Iravani, and P. W. Lehn, “Micro-grid autonomous operation
during and subsequent to islanding process,” IEEE Trans. Power Delivery, vol.
20, no. 1, pp. 248-257, Jan. 2005.
[69] C. T. Lee, R. P. Jiang, and P. T. Cheng, “A grid synchronization method for
droop-controlled distributed energy resource converters,” IEEE Trans.
Industry Application, vol. 49, no. 2, pp. 954-962, Jan. 2013.
[70] F. Blaabjerg, R. Teodorescu, M. Liserre, and A. V. Timbus, “Overview of
control and grid synchronization for distributed power generation systems,”
IEEE Trans. Industrial Electronics, vol. 53, no. 5, pp. 1398–1409, Oct. 2006.
[71] A. Timbus, M. Liserre, R. Teodorescu, P. Rodriguez, and F. Blaabjerg,
“Evaluation of current controllers for distributed power generation systems,”
IEEE Trans. Power Electronics, vol. 24, no. 3, pp. 654–664, Mar. 2009.
[72] J. T. Bialasiewicz, “Renewable energy systems with photovoltaic power
generators: Operation and modeling,” IEEE Trans. Industrial Electronics, vol.
55, no. 7, pp. 2752-2758, Jul. 2008.
[73] P. Cortes, G. Ortiz, J. I. Yuz, J. Rodriguez, J. Vazquez, and L. G. Franquelo,
“Model predictive control of an inverter with output LC filter for UPS
applications,” IEEE Trans. Industrial Electronics, vol. 56, no. 6, pp. 1875-
1883, Feb. 2009.
58
[74] S. Kouro, P. Cortes, R. Vargas, U. Ammann, and J. Rodriguez, “Model

predictive control-A simple and powerful method to control power converters,”
IEEE Trans. Industrial Electronics, vol. 56, no. 6, pp. 1826-1838, Jun. 2009.
[75] J. Rodriguez, M. P. Kazmierkowski, J. R. Espinoza, P. Zanchetta, et al., “State
of the art of finite control set model predictive control in power electronics,”
IEEE Trans. Industrial Informatics, vol. 9, no. 2, pp. 1003-1016, May 2013.
[76] G. Mirzaeva, G. C. Goodwin, B. P. McGrath, C. Teixeira, et al., “A generalized
MPC framework for the design and comparison of VSI current controllers,”
IEEE Trans. Industrial Electronics, in press.
[77] H. Miranda, R. Teodorescu, P. Rodriguez, and L. Helle, “Model predictive
current control for high-power grid-connected converters with output LCL
filter,” 2009 35th Annual Conference of IEEE Industrial Electronics, Nov.
2009, pp. 633–638.
[78] J. A. P. Lopes, C. L. Moreira, and A. G. Madureira, “Defining control
strategies for microgrids islanded operation,” IEEE Trans. Power Systems, vol.
21, no. 2, pp. 916-924, May 2006.
[79] J. C. Vasquez, J. M. Guerrero, E. Gregorio, P. Rodriguez, R. Teodorescu, and
F. Blaabjerg, “Adaptive droop control applied to distributed generation
inverters connected to the grid,” in Proc. IEEE International Symposium on
Industrial Electronics ISIE’08, 2008, pp. 1–6.
[80] Y. Li, D. M. Vilathgamuwa, and P. C. Loh, “Design, analysis, and real-time
testing of a controller for multibus microgrid system,” IEEE Trans. Power
Electronics, vol. 19, no. 5, pp. 1195–1204, Sep. 2004.
[81] J. C. Vasquez, J. M. Guerrero, M. Savaghebi, J. E. Garcia, et al., “Modeling,
analysis, and design of stationary-reference-frame droop-controlled parallel
three-phase voltage source inverters,” IEEE Trans. Industrial Electronics, vol.
60, no. 4, pp. 1271-1280, Apr. 2012.
59
[82] C. Ahumada, R. Cardenas, D. Saez, and J. M. Guerrero, “Secondary control

strategies for frequency restoration in islanded microgrids with consideration
of communication delays,” IEEE Trans. Smart Grid, vol. 7, no. 3, pp. 1430-
1441, Aug. 2015.
[83] M. Prodanovic and T. C. Green, “High-quality power generation through
distributed control of a power park microgrid,” IEEE Trans. Industrial
Electronics, vol. 53, no. 5, pp. 1471–1482, Oct. 2006.
[84] K. Rahbar, J. Xu, and R. Zhang, “Real-time energy storage management for
renewable integration in microgrid: an off-line optimization approach,” IEEE
Trans. Smart Grid, vol. 6, no. 1, pp. 124-134, Sep. 2014.
[85] J. S. Shen, G. Bag, and K. H. Kim, “A microgrid energy management system
and risk management under an electricity market environment,” IEEE Access,
vol. 4, pp. 2349-2356, Apr. 2016.
[86] K. T. Tan, X. Y. Peng, P. L. So, Y. C. Chu, et al., “Centralized control for
parallel operation of distributed generation inverters in microgrids,” IEEE
Trans. Smart Grid, vol. 3, no. 4, pp. 1977-1987, Aug. 2012.
[87] R. Majumder, C. W. Jiang, Y. Y. Liu, and X. Wang, “Power sharing and
control in distributed generation with wireless sensor networks,” IEEE Trans.
Smart Grid, vol. 3, no. 2, pp. 618-634, Feb. 2012.
[88] M. A. Azzouz, M. F. Shaaban, and E. F. El-Saadany, “Real-time optimal
voltage regulation for distribution networks incorporating high penetration of
PEVs,” IEEE Trans. Power Systems, vol. 30, no. 6, pp. 3234-3245, Jan. 2015.
[89] F. Katiraei, M. R. Iravani, N. Hatziargyriou, and A. Dimeas, “Microgrids
management: Controls and operation aspects of microgrids,” IEEE Power and
Energy Magazine, vol. 6, no. 3, pp. 54-65, 2008.
[90] P. Zhao, S. Suryanarayanan, and M. G. Simoes, “An energy management
system for building structures using a multi-agent decision-making control
60
methodology,” IEEE Trans. Industry Applications, vol. 49, no. 1, pp. 322-330,
Nov. 2012.
[91] Y. S. F. Eddy, H. B. Gooi, and S. X. Chen, “Multi-agent system for distributed
management of microgrids,” IEEE Trans. Power Systems, vol. 30, no. 1, pp.
24-34, May 2014.
[92] M. E. Baran and I. M. El-Markbi, “A multiagent-based dispatching scheme for
distributed generators for voltage support on distribution feeders,” IEEE Trans.
Power Systems, vol. 22, no. 1, pp. 52-59, Feb. 2007.
[93] D. E. Olivares, A. M. Sani, A. H. Etemadi, C. A. Canizares, et al., “Trends in
microgrid control,” IEEE Trans. Smart Grid, vol. 5, no. 4, pp. 1905-1919, May
2014.
[94] R. O. Saber, J. A. Fax, and R. M. Murray, “Consensus and cooperation in
networked multi-agent systems,” Proceedings of the IEEE, vol. 95, no. 1, pp.
215-233, Jan. 2007.
[95] R. O. Saber and R. M. Murray, “Consensus problems in networks of agents
with switching topology and time-delay,” IEEE Trans. Automatic Control, vol.
49, no. 9, pp. 1520-1533, Sep. 2004.
[96] C. Godsil and G. Royle, Algebraic Graph Theory. New York, NY, USA:
Springer, 2001.
[97] Z. Qu, Cooperative Control of Dynamic Systems: Applications to Autonomous
Vehicles. New York, NY, USA: Springer-Verlag, 2009.
[98] Y. L. Xu and W. X. Liu, “Novel multiagent based load restoration algorithm
for microgrids,” IEEE Trans. Smart Grid, vol. 2, no. 1, pp. 152-161, Mar. 2011.
[99] Y. L. Xu, “Optimal distributed charging rate control of plug-in electric vehicles
for demand management,” IEEE Trans. Power Systems, vol. 30, no. 3, pp.
1536-1545, May 2015.
61
[100] H. H. Xin, Z. H. Qu, J. Seuss, and A. Maknouninejad “A self-organizing

strategy for power flow control of photovoltaic generators in a distribution
network,” IEEE Trans. Power Systems, vol. 26, no. 3, pp. 1462-1473, Aug.
2011.
[101] H. H. Xin, Y. Liu, Z. H. Qu, and D. Q. Gan, “Distributed control and
generation estimation method for integrating high-density photovoltaic
systems,” IEEE Trans. Energy Conversion, vol. 29, no. 4, pp. 988-996, Sep.
2014.
[102] F. H. Guo, C. Y. Wen, J. F. Mao, and Y. D. Song, “Distributed secondary
voltage and frequency restoration control of droop-controlled inverter-based
microgrids,” IEEE Trans. Industrial Electronics, vol. 62, no. 7, pp. 4355-4364,
Dec. 2014.
[103] Q. Shafiee, T. Dragicevic, J. C. Vasquez, and J. M. Guerrero, “Hierarchical
control for multiple dc-microgrids clusters,” IEEE Trans. Energy Conversion,
vol. 29, no. 4, pp. 922-933, Oct. 2014.
[104] S. Moayedi and A. Davoudi, “Distributed tertiary control of dc microgrid
clusters,” IEEE Trans. Power Electronics, vol. 31, no. 2, pp. 1717-1733, Feb.
2016.
62
Distributed ESSs for Voltage Regulation Chapter 3
Chapter 3*
Coordinated Control of Distributed Energy Storage

Systems for Voltage Regulation in Distribution Networks
In this chapter, distributed energy storage systems are proposed to solve

the voltage rise/drop issues in LV distribution networks with high
penetration of rooftop PVs. The impact of energy storage integrated with
PV source on feeder voltages is investigated in detail. A coordinated
control method which includes both distributed and localized controls is
proposed to control the operation of distributed ESSs.
*This chapter published substantially as Y. Wang, K. T. Tan, and P. L. So, “Coordinated control of
distributed energy storage systems for voltage regulation in distribution networks,” IEEE Trans.
Power Delivery, vol. 31, no. 3, pp. 1132-1141, Jul. 2015.
63
3.1 Introduction
This chapter addresses how to coordinate distributed ESSs in LV distribution

networks for voltage regulation. The capacity and SoC of each ESS are taken into
account to accommodate daily operation. A coordinated control method which
includes both distributed and localized controls is proposed. For the distributed
control, a consensus algorithm is proposed to regulate voltages along the distribution
feeder. If the voltages at certain buses violate voltage limits, the consensus algorithm
will estimate the power outputs of ESSs to eliminate the voltage violation. For
localized control, a SoC control strategy is proposed to control the SoC of each ESS.
The SoC control avoids depletion or saturation of ESSs for different operation
conditions. The entire control method ensures voltage regulation while effectively
utilizes all available capacities of ESSs in the system. The ESS is modeled as VRB
whose SoC is estimated for the SoC control. Various case studies are implemented
to verify the performance of the proposed method.
3.2 Problem Description and Analysis
Figure 3.1 A typical radial distribution feeder with multiple buses.
64
A typical radial distribution feeder with multiple buses as shown in Figure 3.1 is used
to illustrate the voltage rise/drop issues. In Figure 3.1, each bus comprises a PV
source, an ESS and a local load, which is viewed as a customer. The integration of
ESS into PV source forms a dispatchable DG unit, which is convenient to control for
both grid-connected and islanded operations [1].
Generally, the power imbalance between PV generation and load demand will cause
power flow along the distribution feeder. Depending on the direction and amount of
the power flow, voltages along the feeder will rise/drop to some extent. Especially at
the end of the feeder, the voltage may exceed the limits during peak PV generation
or peak load period. Distribution networks are usually allowed a maximum voltage
deviation of 5%-10% (depend on national standards) from the secondary transformer
to the customer located at the end of the feeder [2]. Therefore, methods that reduce
the power imbalance between PV generation and load demand can mitigate the
voltage rise/drop issues. This can be achieved by utilizing distributed ESSs at the
customer side. The functions of ESSs are to reduce the net power injection at each
bus during the day and reduce the power absorption during the night. Therefore, the
current flow in the distribution feeder will be reduced so that the voltage at the end
of the feeder will be regulated within the voltage limits.
Figure 3.2 shows the voltage magnitudes along radial distribution feeder. The voltage
rise along the feeder during peak PV generation period is shown in Figure 3.2(a).
The voltage drop along the feeder during peak load period is shown in Figure 3.2(b).
The solid line shows the voltage along the feeder with only PV penetration. In this
condition, the voltage at the end of the feeder exceeds the voltage limits. With proper
operation of ESSs, the power imbalance between PVs and loads will be reduced. The
voltages along the feeder will be regulated to the dashed line which is within the
limits. This problem is further illustrated as follows.
65
Voltage rise Voltage drop
Feeder Voltage
Feeder Voltage
Vmax
V0
V0 Vmin
Feeder Length Feeder Length

(a) (b)
Figure 3.2 Voltage magnitudes along radial distribution feeder. (a) Voltage rise. (b)
Voltage drop.
Considering the radial feeder shown in Figure 3.1, from the perspective of power
injection, the complex power injected at ith bus Si is composed of PV power, ESS
power and load power. The voltage relationship and current flow between (i-1)th bus
and ith bus in vector form can be expressed by using power flow analysis as follows:
Si  Vi Iinj
*
,i
(3.1)
N N
Sk*
Ii   Iinj ,k   *
(3.2)
k i k i Vk
Vi  Vi 1  Ii ( Ri  jX i ) (3.3)
where Vi and Vi* are the voltage and its conjugate at ith bus, I inj ,i is the current
injected at ith bus, I i and Ri  jX i are the current and impedance between (i-1)th bus
and ith bus. The voltage deviation between two adjacent buses can be expressed
using the real part in (3.3)
N N
 Pk  Rk  Qk  X k
Vi  Vi  Vi 1  k i k i
(3.4)
Vi 1
Thus the voltage deviation between the transformer bus V0 and the last bus VN can
66
be derived as
N N
N N  Pk  Rk  Qk  X k
VN  V0   Vi   k i k i
(3.5)
i 1 i 1 Vi 1
where Pk and Qk are the real and reactive power injected at ith bus respectively.
According to (3.4) and (3.5), both real and reactive power can be utilized to reduce
the voltage deviation. However, for typical distribution networks with large R/X
ratio, real power is more effective to compensate the voltage deviation compared to
reactive power. Large amount of reactive power requires higher power rating
inverters and leads to higher losses as well as lower power factors along the LV
distribution feeder [3]. In the meantime, compared with PV curtailment method
proposed in [2], the utilization of ESSs can improve the efficiency of PV generation.
In this chapter, the function of ESSs is to reduce the real power injected by customers
during the day and the real power absorbed by customers during the night. Therefore,
the current flow in the distribution feeder will be reduced so that the voltage rise/drop
along the feeder can be mitigated.
3.3 Proposed Control Method
The structure of the proposed control method for customer at ith bus is shown in
Figure 3.3, in which the solid lines show the power flow and the dotted lines show
the control signal flow. For each customer, PV panels and storage devices are fed to
the DC link through the boost converter and the bidirectional converter respectively.
The power transfer between the DC link and the AC grid is achieved by the DC/AC
inverter. Maximum power point tracking (MPPT) control is adopted by the PV
panels to obtain the power reference of the boost converter.
67
Figure 3.3 The structure of proposed control method for customer at ith bus.
As mentioned before, the objective of this research is to coordinate distributed ESSs

for voltage regulation while effectively utilizes the available storage capacity under
various operation conditions. The proposed control method includes both distributed
and localized controls to achieve this goal. As shown in Figure 3.3, the distributed
control signal utilization ratio ui is generated by a consensus algorithm which is used
to regulate the feeder voltages with distributed ESSs. The localized control signal
availability ratio εi is provided by SoC control which is used to regulate the SoC of
each ESS within the desired SoC range. Finally, the power output of the ESS is
determined by multiplying ui and εi with its rated power. A VRB model is
implemented for grid application, and the SoC is estimated for localized SoC control.
The details of the VRB model and proposed control method are explained in the
following subsections.
3.3.1 Vanadium Redox Battery
Due to the high energy efficiency, high scalability, fast response, long time duration
and low maintenance requirements, VRB is well suitable for grid-scale applications
[4]. In this chapter, a residential VRB based ESS rated 5kW/20kWh is implemented
68
for grid energy storage purpose. The VRB is modeled based on the data presented in
[4], [5]. The open-circuit voltage (OCV)/SoC characteristic curve of a single VRB
cell is shown in Figure 3.4. The VRB will be operated in the linear region which
ranges between 15%-85% of SoC. Each ESS consists of 38 VRB cells connected in
series. The parameters for the VRB based ESS are shown in Table 3.1. In addition,
the SoC of the VRB is estimated using the ampere-hour counting method as follows:
SoC (t )  SoC (t  t )  SoC (3.6)
Pbat t Vbat Ibat t

SoC   (3.7)
CB CB
where CB is the storage capacity of the VRB, t is the time interval, Vbat and Ibat are
the output voltage and current of the VRB respectively.
Figure 3.4 The OCV/SoC characteristic curve of a single VRB cell.
Table 3.1 Parameters of the VRB based ESS
Cell Configuration 38 Series Cell Stack

Rated Power 5kW
Rated Capacity 20kWh
Operation Voltage 42-57V
min max
SoC Region (SoC , SoC ) 15%, 85%
69
3.3.2 Distributed Control Method
The first objective of the proposed control method is to coordinate distributed ESSs
for voltage regulation. A distributed control using consensus algorithm is applied for
ESSs to regulate the feeder voltages within the voltage limits. Consensus algorithm
is achieved by sharing variable of interest, called the information state among all
available units. The initial information state is determined by the virtual leader which
measures the information of the critical point [6], [7]. In this chapter, the information
state of each ESS is represented by utilization ratio u.
For radial distribution feeder with multiple buses as shown in Figure 3.1, the last bus
is the critical bus which has the highest/lowest voltage in the system. Thus the last
bus is chosen as the virtual leader to initiate the ESSs coordination. The utilization
ratio of the leader is determined by measuring the voltage of the last bus. The
utilization ratio of the leader is then shared to available ESSs through communication
links to determine the utilization ratio of each ESS to realize the desired voltage
regulation objective. The control scheme of the proposed distributed control is shown
in Figure 3.5.
Figure 3.5 The control scheme of the proposed distributed control.
To avoid overvoltage and undervoltage, the voltage upper and lower references (Vup
and Vlow) are chosen within the voltage limits (Vmax and Vmin) for the consensus
algorithm. During both peak PV generation and peak load periods, for arbitrary bus
70
voltage Vi in distribution feeder, the condition given by (3.8) will be met with the
proposed consensus algorithm:
V low  Vi (t )  V up (3.8)
The utilization ratio of the leader uref is updated at discrete time steps. During peak
PV generation period, if the voltage of the last bus VN goes beyond upper voltage
reference Vup, uref will be increased. During peak load period, if the voltage of the
last customer goes below lower voltage reference Vlow, uref will be reduced. During
other periods, uref will be equal to the former value. In general, the utilization ratio
of the leader will be updated as follows:
 u ref (t  ts )  k1 (VN (t )  V up ) if VN (t )  V up

u ref (t )   u ref (t  ts ) if V low  VN (t )  V up (3.9)
u ref (t  t )  k (V (t )  V low ) if VN (t )  V low
 s 2 N
where VN is the voltage of the last bus, V up and V low are the upper and lower voltage
references respectively. Parameters k1 and k2 are the control gains, which affect the
convergence speed and control accuracy of the distributed control. ts is the sampling
interval. For daily operation, it should pay attention that the initial value of uref is 0.
uref will be back to 0 when the control changes from charging mode to discharge
mode, and vice versa.
The utilization ratio provided by the leader is then communicated to available ESSs
through communication links. The instantaneous communication topology can be
represented by the following matrix:
 s11 (t ) s12 (t ) s1N (t ) 

 s (t ) s (t ) s2 N (t ) 
S (t )   21 22
(3.10)
 
 
 sN 1 (t ) sN 2 (t ) sNN (t ) 
71
where sij denotes the communication link between the ith and jth customers. sii = 1
for all i, sij = 1 if the utilization ratio of the jth bus is received by the ith bus at time
t, and sij = 0 if otherwise. In this thesis, it is assumed that customers can only
communicate with their neighbours, which means all s(i)(i-1) = 1 and s(i)(i+1) = 1 for
radial feeder shown in Figure 3.1.
The utilization ratio for the ith customer is updated as follows:
N 1
ui (t )   dij (t )u j (t  ts )  diN (t )u ref (t  ts ) (3.11)
j 1
where dij(k) is the (i, j) entry of a row stochastic matrix (i.e., row sum of 1) D(t)
which can be found in each discrete time data exchange by
ij sij (t  ts )
dij (t )  N
(3.12)

k 1
s (t  ts )
ik ik
where the weights ωij are set to 1 in this research to share the required real power
equally among distributed ESSs and sij are the entries of communication matrix,
given in (3.10). The ith customer will know the ith row of matrix D(t) to perform the
consensus operation. The communication graph for the N-bus radial feeder can be
presented using state transitions diagram as shown in Figure 3.6.
1/2 1/3 1/3 1/2
1/3 1/3 1/2

1 2 i N
1/2 1/3 1/3
Figure 3.6 Communication graph for N-bus radial distribution feeder.
72
Although the proposed algorithm is derived under radial distribution feeder with one
lateral, it is also applicable for distribution networks with multiple laterals. In such
condition, the last bus at each lateral is the critical bus which has the highest/lowest
voltage in that lateral. So the last bus of each lateral will be chosen as a virtual leader.
If the last bus of this lateral is regulated within the voltage limits, the voltages along
this lateral will be within the limits. The utilization ratios of available ESSs are still
updated by (3.11) according to the communication matrices S(t) and D(t). This
condition will be further illustrated in the second case study of this chapter.
3.3.3 Localized Control Method
The proposed distributed control will regulate the feeder voltages and determine the
utilization ratio of each ESS. However, due to different SoC of ESSs and
unpredictable weather/load conditions, the ESSs may get fully charged/discharged
when they are needed. The second objective is to effectively utilize the available
storage capacity in the network for voltage regulation. Therefore, for each ESS, a
localized control based on local SoC information is implemented to adjust the
charging/discharging speed.
The localized SoC control proposed in this research regulates the SoC of each ESS
within a desired range of its predefined reference SoC. The estimated SoC SoC(t) is
compared with the reference SoC SoCref(t) to determine the availability ratio ε. The
reference SoC is a function of t, which donates expected SoC change of each ESS
during daily operation. For the ESS charging mode of SoC control, ε is defined by
the following equation:
73
 0 if SoC (t )  SoC ref (t )  a


 (t )  1  k3 ( SoC (t )  SoC ref (t )) if b  SoC (t )  SoC ref (t )  a (3.13)
 1 if SoC (t )  SoC ref (t )  b

where parameters a and b are thresholds of SoC control, which define the allowable
SoC deviation range from reference SoC. Parameter k3 is equal to 1/(a-b). If SoC(t)
of the ESS is larger than SoCref(t)+a, the availability ratio becomes 0 to stop the
charging of the ESS. If SoC(t) is within the range [SoCref(t)+b, SoCref(t)+a], the
availability ratio is between 0 and 1 to slow down the charging of the ESS. If SoC(t)
of the ESS is smaller than SoCref(t)+b, the availability ratio becomes 1 to make full
use of the ESS. The condition for the ESS discharging mode of SoC control is similar,
which is not further illustrated here.
Finally, the required power output of ith ESS can be determined by
,i (t )  ui (t )   i (t )  PESS ,i
ref rated
PESS (3.14)
1  ui (t )  1, 0   i (t )  1 (3.15)
rated
where PESS ,i is the rated power output of ESS at ith bus.
It is noted that the operation of the ESS should also take into account power and
energy constraints. The power limits and SoC limits of the ESS during operation can
be described by (3.16) and (3.17) respectively
 PESS
rated
 PESS (t )  PESS
rated
(3.16)
SoC min  SoC (t )  SoC max (3.17)
When the proposed control method is applied for daily operation, there are several
points should be considered. First, based on historical PV and load data, the total
74
storage capacity should be enough to deal with voltage rise/drop issues in the
network. Next, although ESSs are installed for voltage regulation purpose, it is
expected that the storage capacity is fully used for 24-h operation. Besides, the
control method has charging and discharging modes to accommodate daily operation.
There will be mode changes at specific time during the operation. Finally, the
parameters of the proposed control method should be properly chosen. The upper
and lower voltage references define how much power to be charged/discharged by
ESSs with distributed control. The reference SoC defines how fast the ESSs should
be charged/discharged during the operation. In this chapter, the upper and lower
references are set as 1.05 p.u. and 0.96 p.u. respectively, so that the ESSs will get
fully charged/discharged under different weather conditions (sunny and cloudy PV
profiles in this chapter) with only distributed control. For other conditions, if the
capacities of ESSs in the system are large, Vup and Vlow can be set more close to 1p.u..
Whereas if the capacities of ESSs in the system are small, Vup and Vlow can be set
more close to voltage limits. The reference SoC, parameters a and b for localized
SoC control should be chosen according to distributed control, which regulate the
power output of the ESS if the SoC exceeds the desired SoC range during the
operation. This control design ensures fully utilizing the capacity of ESSs while
avoiding overcharge or overdischarge of ESSs. The parameters of the proposed
control method are presented in Table 3.2.
75
Table 3.2 Parameters of the proposed control method
Parameter Value
Voltage Reference (Vlow, Vup) 0.96p.u., 1.05p.u.
Distributed Control (k1, k2) 0.001, 0.001
SoC Control Charging Mode
(a, b, k3) 0.05, 0, 20
SoC Control Discharging Mode
(a’, b’, k3’) -0.05, 0, -20
3.4 Results and Discussions
3.4.1 Case Study: 6-Bus Distribution Feeder
In this section, a distribution feeder consisting of six buses is implemented in

Matlab/Simulink to investigate the performance of the proposed control method. The
overhead residential feeder supplies electricity to five customers by a 100kVA,
22/0.4kV transformer. The distance between each bus is 100m. The single line
diagram of the 6-bus test feeder is shown in Figure 3.7. The PV systems at the
residential households are operated at unity power factor and have the capacity of
9kWp. A VRB rated at 5kW/20kWh is integrated with the rooftop PV system of each
customer which is a typical size for residential rooftop PV applications [8]. The
voltage at the secondary of the transformer is set at 230V (1.0p.u.). This research
allows a maximum ±6% voltage deviation in distribution networks. The voltage
upper reference (1.05p.u.) and lower reference (0.96p.u.) are chosen within this limit.
The parameters of the simulation system are presented in Table 3.3. Figure 3.8 shows
the influence on feeder voltages of bus number and real power injection of the test
feeder. As shown in Figure 3.8, the last bus has the largest voltage deviation in the
feeder for the same power injection at each bus. On the contrary, bus 1 of the test
76
feeder has the smallest voltage deviation. Moreover, the last bus is more sensitive
than other buses to the power injection changes in the test feeder.
Figure 3.7 Single line diagram of the 6-bus test feeder.
Table 3.3 Parameters of the simulation system
Feeder Length 500m

Bus to Bus Distance 100m
Conductor Size 0.549 + j0.072Ω/km
MV/LV Transformer Size 100kVA, 22/0.4kV
PV Size 9kWp
VRB Rating 5kW/20kWh
Maximum Voltage Deviation ±6%
Figure 3.8 Influence on feeder voltages of bus number and power injection.
77
To test the developed methodology in a realistic environment, two solar PV output

profiles and a residential load profile are used. The PV data is measured using GL130
PV module. It was measured in the summer of the year 2014 at Clean Energy
Research Laboratory, Nanyang Technological University, Singapore. 5-min PV data
is used as input data for the PV systems. Customer loads are aggregated into a typical
residential load profile, which is presented in [9]. The PV and load data are scaled
up for this application. The simulation sampling interval is 0.4s. To cope with the
simulation speed, the PV and load data hold the same value for every five minutes.
Besides, this study focuses on distribution networks in a community scaled
residential area. So it is assumed that customers present identical PV and load
profiles in this thesis.
1) Test Case 1: Sunny Weather Condition
The first test case demonstrates the performance of the proposed control method
under sunny weather condition with identical initial SoC of each ESS. The initial
SoC of each ESS is 20%. The PV output, the load demand, the ESS power of the last
bus (bus 5) of the radial test feeder is shown in Figure 3.9. The corresponding power
injection at bus 5 with/without ESSs is shown in Figure 3.10. As PV generation, load
demand and initial ESS condition are identical for each bus, ESSs at buses 1, 2, 3
and 4 have the same power outputs and SoC profiles as the ESS at bus 5, which are
not shown here.
As shown in Figures 3.9 and 3.10, from 8:30h to 17:10h, there is more power
generated by the PVs than consumed by the loads. Thus the feeder will export real
power to the grid which results in voltage rise along the feeder. However, during the
other periods, there is no energy produced by PVs so that the load demand is larger
than PV generation. The feeder will import real power from the grid which results in
78
voltage drop. If ESS is installed at each bus, it will start charging at 9:45h when the
voltage at bus 5 exceeds the upper voltage reference (1.05p.u.). The control changes
to discharging mode at 17: 00h. The ESSs will start discharging at 17:50h when the
voltage at bus 5 below the lower voltage reference (0.96p.u.). In Figure 3.10, due to
the operation of ESS at each bus, the power injection/absorption at bus 5 with ESSs
(solid line) is smaller than the power without ESSs (dotted line). The energy stored
during the daytime is utilized for voltage support during the evening.
Figure 3.9 PV generation, residential load demand and ESS power for test case 1.
Figure 3.10 Power injection with/without ESSs for test case 1.
79
Figure 3.11 24-h voltage profiles without ESSs for a sunny day.
Figure 3.12 24-h voltage profiles with ESSs and proposed method for a sunny day.
Figures 3.11 and 3.12 present the corresponding voltage profiles of each bus without
and with ESSs respectively. In Figure 3.10, the maximum voltage (1.093p.u.) occurs
at noon and the minimum voltage (0.912p.u.) occurs at evening both in bus 5. The
voltage exceeds the upper voltage limit (1.06p.u.) from 10:00h to 15:45h. Buses 3, 4
and 5 experience overvoltage for at least a small period of time during the daytime.
The voltage drops below the lower voltage limit (0.94p.u.) from 18:20h to 23:00h.
80
Buses 3-5 experience undervoltage for at least a small period of time during the
evening. As shown in Figure 3.12, with ESSs installed at each bus and proposed
control method, the voltage is regulated below Vup (1.05p.u.) for overvoltage period,
and the voltage is regulated above Vlow (0.96p.u.) for undervoltage period.
Figure 3.13 SoC profiles for test case 1.
The corresponding SoC profiles of the ESS at bus 5 are shown in Figure 3.13. The
SoC changing speed is corresponding to the power output of the ESS. The SoC
reference curve used in this test case for the SoC control is also shown in Figure 3.13.
As shown in Figure 3.13, the SoC increases to 84.5% at the end of charging process
and reduces to 19.8% at the end of discharging process. At the first part of charging
process, the SoC profile is lower than the SoC reference, which means the ESS is
fully available during this period and the availability ratio ε=1. At the second part of
the charging process, the SoC profile is higher than the SoC reference, which means
that the ESS is partly available and the availability ratio ε is between 0 and 1 to
restrict the overcharging of the ESS. During most time of discharging process, the
ESS is fully available for discharging. It is also observed in Figure 3.13 that the SoC
profile without the SoC control (dotted line) will exceed the SoC operation range of
the VRB at the end of charging process.
81
2) Test Case 2: Cloudy Weather Condition
The second test case further demonstrates the performance of the proposed control
method under cloudy weather condition with different initial SoC of each ESS. In
this test case, the initial SoCs of ESSs at buses 1, 2, 3, 4 and 5 are 50%, 35%, 30%,
25% and 20% respectively. ESS at bus 1 is plugged out at 12:00h, and plugged back
into the feeder at 13:00h. Figure 3.14 shows the PV output and the load demand in
this condition. In Figure 3.14, as there are several large variations of solar irradiation
during the daytime, the total PV generation is smaller than in test case 1. Figures
3.15 and 3.16 present the voltage profiles of each bus without and with ESSs
respectively in test case 2. In Figure 3.15, due to the variations of solar irradiation,
overvoltage time is from 10:50h to 12:10h and 13:05h to 14:45h, which is shorter
than in Figure 3.11. Buses 3, 4 and 5 still experience overvoltage and undervoltage
for some period of time. As shown in Figure 3.16, with ESSs installed at each bus
and proposed control method, voltages along the feeder are regulated within the
voltage reference.
Figure 3.14 PV generation and residential load demand for test case 2.
82
Figure 3.15 24-h voltage profiles without ESSs for a cloudy day.
Figure 3.16 24-h voltage profiles with ESSs and proposed method for a cloudy day.
The corresponding power outputs and SoC profiles of ESSs are shown in Figures
3.17 and 3.18 respectively. The time sequences of the operation are explained as
follows:
1) At 10:50h, the voltage of bus 5 exceeds the upper voltage reference. ESSs 2,
3, 4, and 5 start to charge while the power output of ESS 1 remains 0 until
11:30h. The reason is that the initial SoC of ESS 1 is 50%, which is out of the
83
desired SoC range (larger than (SoCref(t)+0.05)) of SoC control. The

availability ratio ε of ESS1 keeps 0 to prevent the charging of ESS1.
2) From 11:30h to 12:00h, the charging power of ESS 1 is smaller than the
charging power of ESSs 2, 3, 4, and 5. The availability ratio ε of ESS 1 is
between 0 and 1 but still smaller than 1.
3) From 12:00h to 13:00h, ESS 1 is plugged out while ESSs 2, 3, 4 and 5 are still
operated normally in the distribution feeder. The SoC of ESS 1 keeps the same
value for an hour.
4) From 13:00h to 15:10h, all ESSs are fully available (ε=1) for distributed
control.
5) From 15:10h to 17:00h, the charging speed of each ESS starts to slow down.
The availability ratio ε of each ESS drops between 0 and 1 to reduce the
charging power of each ESS.
6) At 17: 00h, the SoC control changes from charging mode to discharging mode.
7) From 17:00h to 24:00h, each ESS starts to discharge when the voltage of bus
5 drops below lower voltage reference at 17:50h. The SoC of each ESS reduces
during the discharging process.
Figure 3.17 Power outputs of ESSs for test case 2.
84
Figure 3.18 SoC profiles of ESSs for test case 2.
3) Test Case 3: Comparison with Droop Based Method
In this part, a comparison with droop based method proposed in [2] and [10] is made
to further illustrate the features of the proposed method. The droop based method is
a typical voltage regulation method which was first applied to curtail PV power in
[2] and then applied in [10] to achieve decentralized control of distributed ESSs. This
method uses local voltage to define how much power should be curtailed from PV
inverters or stored in ESSs. The comparison is implemented with the same condition
as in Test Case 1. Figure 3.19 shows the voltage profiles with droop based method.
Figure 3.20 shows the power curtailed with constant droop based method during the
daytime. As shown in Figure 3.20, the PV energy curtailed at buses 3, 4 and 5 are
about 7kWh, 18.1kWh and 23.3kWh respectively. While the total energy generation
for the sunny PV condition is about 58.6kWh. This means that 11.9%, 30.9% and
39.8% of total PV production are curtailed by PV at buses 3, 4 and 5 respectively.
Even though the curtailed energy can be stored in ESSs as proposed in [10], there
are still several limitations of this method. ESSs are usually randomly deployed in
85
the system other than well planned. However, as shown in Figure 3.19, the
requirement of storage capacity with constant droop method increases along the
feeder. In the meanwhile, variable droop method allows random deployment of ESSs,
but this method requires specific calculation of droop parameters according to
system condition. In practical situation, parameters of the system are always
changing. Real-time estimation of droop parameters makes this method difficult to
realize. Besides, ESSs in distribution networks usually have plug and play ability,
which leads to different SoCs of each ESS. The ESS at certain bus may be fully
charged/discharged when it is needed. In order to regulate voltage at this particular
bus, PV curtailment and reactive compensation is inevitable. Compared with droop
based method, the proposed method allows arbitrary deployment of distributed ESSs
as long as the total storage capacity is enough for voltage regulation. All available
storage capacity in the system can be effectively utilized for voltage regulation.
Figure 3.19 Voltage profiles with droop based method.
86
Figure 3.20 Power curtailed with constant droop based method.
3.4.2 Case Study: 13-Bus Distribution Network
In this section, the performance of the proposed control method is demonstrated in a

residential distribution network with multiple laterals. The configuration of the test
system is a 13-bus distribution network which is a part of IEEE 34-bus system [11]
as shown in Figure 3.21. The overhead residential feeder supplies electricity to
twelve customers by a 22/0.4kV transformer, with seven households installed PV
and ESS integrated DG units. The distance between each bus is 50m. The PV profile
is the same as that in Figure 3.14. The load demand profile is 50% of that in Figure
3.14. The initial SoCs of ESSs at buses 2, 3, 6 and 7 are 20%, and the initial SoCs of
ESSs at buses 4, 10 and 12 are 30%, 40% and 50% respectively.
As shown in Figure 3.21, the distribution network has three laterals. Buses 4, 10 and
12 are at the end of each lateral and have the highest/lowest voltage in each lateral.
So they are chosen as virtual leaders for distributed control. It is also assumed that
each bus can communicate with its adjacent buses. The communication graph for the
87
13-bus distribution network is presented using state transitions diagram as shown in

Figure 3.22.
Figure 3.21 Configuration of the 13-bus residential distribution network.
1/2 1/4 1/3 1/4 1/2
1/4 1/3 1/4 1/2

2 3 6 7 10
1/2 1/4 1/3 1/4
1/4 1/4
1/2 1/2
1/2 4 1/2 12
Figure 3.22 Communication graph for the 13-bus distribution feeder.
Figures 3.23 and 3.24 show the voltage profiles of each bus without and with ESSs
respectively in this case study. In Figure 3.23, the overvoltage and undervoltage are
even severe than in Figure 3.15. The voltages at most buses will violate the voltage
limits during peak generation/load periods. Bus 10 has the longest distance to the
88
secondary transformer, so the voltage at bus 10 is the highest/lowest in the network.

As shown in Figure 3.24, with the proposed control method adopted by ESSs,
voltages in the network are regulated within the voltage references. The
corresponding power outputs and SoC profiles of ESSs are shown in Figure 3.25 and
Figure 3.26 respectively. ESSs 2, 3, 6 and 7 have the same power outputs and SoC
profiles, so only the profile of ESS 7 is shown here.
Figure 3.23 24-hr voltage profiles without ESSs.
Figure 3.24 24-hr voltage profiles with ESSs and proposed method.
89
Figure 3.25 Power outputs of ESSs 4, 7, 10 and 12.
Figure 3.26 SoC profiles of ESSs 4, 7, 10 and 12.
The time sequences of the operation are detailed as follows:

1) At 9:30h, ESSs 2, 3, 6 and 7 start to charge. The power output of ESS 10
remains 0 until 10:30 h and the power output of ESS 12 remains 0 until 11:30
h.
2) At 12:10h, due to the large variation of solar irradiation, there is a sudden drop
of charging power of each ESS.
3) From 13:10h to 15:30h, ESSs 2, 3, 6 and 7 have more charging power than
ESSs 4, 10 and 12.
90
4) From 15:30h to 17:00h, the charging speed of each ESS slows down quickly.
5) At 17:00h, the SoC control changes from charging mode to discharging mode.
6) From 17:00h to 24:00h, each ESS starts to discharge at 17:30h. The SoC of
each ESS reduces during the discharging process.
3.5 Conclusion
In this chapter, distributed ESSs have been utilized to regulate the voltages in LV
distribution networks with high PV penetration. The impact of ESS integrated with
PV source on feeder voltages has been investigated. A coordinated control method
has been proposed for voltage regulation while effectively utilizing storage
capacities of ESSs during daily operation. The power outputs of ESSs are determined
by both distributed control and localized SoC control. Real-time SoC of each ESS is
estimated for the proposed control method. A simulation platform has been
implemented in Matlab/Simulink to validate the performance of the proposed control
method under various operation conditions. The simulation results have
demonstrated that the proposed control method ensures the voltages in the networks
within the voltage limits during the daily operation. Based on the SoC condition of
each ESS, the power sharing among each ESS for voltage regulation is automatically
determined. So the SoC of each ESS is regulated within desired SoC range during
the operation. The proposed method can also be implemented to distribution
networks with multiple laterals as shown by the case study conducted in 13-bus
distribution network.
However, there are several limitations of the research work in this chapter. First, the
control parameters Vup, Vlow and the SoC references of the proposed control method
are chosen according to the author’s experience. For future applications, upper level
controller should be designed to optimize these parameters according to the PV and
91
load prediction. Second, the power conversion efficiency of VRB model and power
inverter model is not considered. Besides, the total capacity of ESSs is considered to
be enough for voltage regulation in the distribution network. However, the ESS
capacity may not be enough in practical situation. Therefore, the coordination
between ESSs and other voltage regulation methods such as OLTCs, voltage
regulators and demand response should be further investigated.
References
[1] M. B. Delghavi and A. Yazdani, “A unified control strategy for electronically

interfaced distributed energy resources,” IEEE Trans. Power Delivery, vol. 27,
no. 2, pp. 803-812, Feb. 2012.
[2] R. Tonkoski, L. A. C. Lopes, and T. H. M. El-Fouly, “Coordinated active
power curtailment of grid connected PV inverters for overvoltage prevention,”
IEEE Trans. Sustainable Energy, vol. 2, no. 2, pp. 139-147, Apr. 2011.
[3] F. Viawan and A. Sannino, “Voltage control with distributed generation and
its impact on losses in LV distribution systems,” IEEE Power Tech, Russia,
2005, pp.1-7.
[4] Q. Xin, T. A. Nguyen, J. D. Guggenberger, M. L. Crow, and A. C. Elmore, “A
field validated model of a vanadium redox flow battery for microgrids” IEEE
Trans. Smart Grid, vol. 5, no. 4, pp. 1592-1601, Jul. 2014.
[5] L. Barote, C. Marinescu, and M. Georgescu, “VRB modeling for storage in
stand-alone wind energy systems,” IEEE PowerTech, Bucharest, 2009, pp. 1-
6.
[6] A. Maknouninejad and Q. Zhihua, “Realizing unified microgrid voltage profile
and loss minimization: a cooperative distributed optimization and control
approach,” IEEE Trans. Smart Grid, vol. 5, no. 4, pp. 1621-1630, Apr. 2014.
92
[7] G. Mokhtari, A. Ghosh, G. Nourbakhsh, and G. Ledwich, “Smart robust

resources control in LV network to deal with voltage rise issue,” IEEE Trans.
Sustainable Energy, vol. 4, no. 4, pp. 1043-1050, Oct. 2013.
[8] Y. Ru, J. Kleissl, and S. Martinez, “Storage size determination for grid-
connected photovoltaic systems,” IEEE Trans. Sustainable Energy, vol. 4, no.
1, pp. 68-81, Jan. 2013.
[9] J. A. Jardini, C. M. V. Tahan, M. R. Gouvea, S. U. Ahn, et al., “Daily load
profiles for residential, commercial and industrial low voltage consumers,”
IEEE Trans. Power Delivery, vol. 15, no. 1, pp. 375-380, Jan. 2000.
[10] M. J. E. Alam, K. M. Muttaqi, and D. Sutanto, “Mitigation of rooftop solar PV
impacts and evening peak support by managing available capacity of
distributed energy storage systems,” IEEE Trans. Sustainable Energy, vol. 28,
no. 4, pp. 3874-3884, May 2013.
[11] IEEE PES, Distribution Test Feeders, Sep. 2010. [Online]. Available:
http://www. ewh.ieee.org/soc/pes/dsacom/testfeeders/.
93
94
Distributed EVCs for Voltage Regulation Chapter 4
Chapter 4
Two-stage Coordination of Distributed Electrical Vehicle

Chargers for Voltage Regulation in Distribution Networks
In this chapter, distributed EVCs are proposed to solve the voltage

rise/drop issues in LV distribution networks with a high penetration of
rooftop PVs and EVs. A two-stage voltage regulation method is proposed
to coordinate the operation of distributed EVCs.
95
4.1 Introduction
Nowadays, increasing amount of EVs are plugged in the home EVCs in distribution
networks, which will cause serve voltage problems with poor coordinated charging.
Due to the concept of V2G, EVs with proper control strategy can be served as energy
storage devices and utilized for voltage regulation. The availability of EVs for V2G
application will vary according to the owner’s preference, which can be obtained
from the travelling preference of different types of EVs. The travelling preference
usually includes the daily travel distance and the parking/charging time of the EV
users. In Singapore, the daily average travel distances of EVs are 55km, which are
within the typical EV travelling ranges between 90km to 160km [1]. The time during
which the EVs are parked is available to be used for V2G application. The majority
of the EVs are generally used for transportation to and from workplace between
08:00 to 10:00 hours and 16:00 to 20:00 hours respectively. These are the time
periods when the availability of EVs for grid ancillary services are low. However,
during other periods (i.e., between 10:00 to 14:00 hours and 22:00 and 05:00 hours)
the EVs are usually not used or immobile as they are typically plugged in for
charging at the workplace and home respectively. These are the time periods when
the availability of EVs for grid ancillary services are high. Besides, the EV charging
manner could be quite different from commercial areas to residential areas. In high
density commercial and residential areas, EVs are aggregated in EV charging
stations and charged together in a centralized way. While in low density residential
areas, EVs are usually charged by home charging outlets in a distributed way. In this
chapter, the voltage regulation problem in LV distribution networks with high PV
penetration and typical residential EV charging is considered.
To delineate the behaviors of storage devices on EVs, a lithium-ion battery model

based on data fitting is derived to estimate SoC for the proposed control method in
96
this chapter. A novel two-stage coordination method of distributed EVCs is proposed.

The first control stage obtains the voltage regulation target by measuring voltage of
critical bus and distributes the control signal to each EVC through communication
links. The second control stage determines the amount of real and reactive power
which can be provided by each EVC, considering the constraints and charging
objectives of the EVC and its charging EV. Different types of PV and load profiles
are applied to evaluate the performance of the proposed method in a practical
distribution network.
4.2 Modeling and Assumptions
4.2.1 Lithium-ion Battery Model
An electrical circuit model consists of a controlled voltage source in series with a

variable resistance is used to characterize the lithium-ion battery, as shown in Figure
4.1 [2]. In Figure 4.1, the open circuit voltage Voc is modeled as a cubic function of
SoC, while the internal resistance Rb is modeled as a linear function of SoC. Through
fitting the experimental data of the lithium-ion battery in [3], the following equations
can be derived:
Voc  a1SoC 3  a2 SoC 2  a3 SoC  a4 (4.1)
 R  b SoC  b2
Rb   ch 1 (4.2)
 Rdis  b3 SoC  b4
where the coefficients a1-a4 and b1-b4 are shown in Table 4.1, Rch and Rdis are the
charging and discharging internal resistance respectively. According to Figure 4.1,
the battery output voltage Vb can be derived as follows:
Vb  Voc  Rb Ib (4.3)
97
where Ib is the charging/discharging current of the battery.
Rdis ( SoC )
Ib
+
Rch ( SoC )
Voc ( SoC ) +
Rb Vb
Figure 4.1 The equivalent circuit model of the lithium-ion battery.
Table 4.1 Data fitting results of the lithium-ion battery
Voc (V)
a1 (V) a2 (V) a3 (V) a4 (V)
1.729 -2.3 1.457 3.24
Rch (mΩ) Rdis (mΩ)
b1 (mΩ) b2 (mΩ) b3 (mΩ) b4 (mΩ)
-0.7357 3.957 -1.782 5.191
Based on (4.1)-(4.3), the charging and discharging characteristics of the proposed

lithium-ion battery can be obtained as shown in Figures 4.2 and 4.3 respectively. It
can be observed that the battery output voltage Vb is a function of SoC at a given
charging/discharging current Ib. Furthermore, for a certain SoC, Vb is proportional to
the charging current, while Vb is inversely proportional to the discharging current.
The SoC of the lithium-ion battery model is estimated using the ampere-hour
counting method as
I b (k )t
SoC(k )  SoC(k  1)  (4.4)
CB
where CB is the capacity of the lithium-ion battery and t is the time interval.
98
Figure 4.2 The charging characteristic of the lithium-ion battery.
Figure 4.3 The discharging characteristic of the lithium-ion battery.
4.2.2 Assumptions
The EV is modeled as Nissan Leaf with storage capacity of 24 kWh, which has a
battery pack consisting of 120 lithium-ion battery cells connected in series. The EVC
is 230V/16A Level II battery charger with 3.6 kVA charging power rating [4]. The
same power rating is assumed to be valid for the discharging operation in V2G mode.
99
All EVs are charged in a distributed way at the home charging outlets (i.e., EVCs)
in the community-scale residential network. Once an EV is plugged in the EVC, the
EV user’s individual information including arrive time, estimated departure time and
SoC is recorded. Then the power allocator judges whether the EV is in V2G mode
or maximum power charging mode. If the EV is in V2G mode, the EV will participate
in the voltage regulation once there is a voltage violation at critical buses. Otherwise,
the EV is charged with the maximum power allowed by the EVC. All EV users will
be kept above a desired SoC of 80% by departure. This is preferable from the EV
users' point of view that the EV will have sufficient energy for the next trip. Besides,
the EV will operate at linear operating range of lithium-ion battery, which is between
10% and 90% of SoC, to avoid depletion or saturation of EV’s battery pack.
4.3 Proposed Voltage Regulation Method
In order to coordinate distributed EVCs for voltage regulation in distribution

networks, this research proposes a new efficient approach to control the real and
reactive power outputs of EVCs. The proposed control method includes two control
stages which are distributed voltage control and localized power allocator. The entire
control structure of the proposed voltage regulation method is shown in Figure 4.4.
In Figure 4.4, the voltages of the critical buses are measured by virtual leader to
initiate the voltage regulation. In the stage one, utilization ratio u of each EVC is
updated with the signals received from adjacent EVCs. Then the power allocator
determines how much real and reactive power will be provided by each EVC in the
stage two. The details of the proposed voltage regulation method are explained in the
following subsections.
100
Figure 4.4 Entire control structure of the proposed voltage regulation method.
4.3.1 Stage One: Distributed Voltage Control
The stage one of the proposed control method is to coordinate EVCs for voltage
regulation using distributed voltage control. The distributed control can be achieved
by the consensus algorithm which shares the information state among all available
units through communication links between them. The information state is initiated
by the virtual leader which measures the information of the critical points [5]. With
the information state provided by the virtual leader, the distributed control will
determine the utilization ratio u for each EVC. The number of EVCs is represented
by i=1, 2, … , N. For radial distribution feeders with only one lateral, the last bus
has the highest/lowest voltage in this feeder, which is defined as a critical bus. For
101
distribution networks with multiple laterals, the last bus of each lateral may have the
highest/lowest voltage in the whole system. Therefore, there will be several critical
buses in this condition, and they will be chosen as virtual leaders to initiate the
distributed control. The information state of the virtual leader uref at a critical bus is
updated as follows:
u ref (t  ts )  k1 (Vc (t )  1.05V0 ) Vc (k )  1.05V0

 ref
u (t  ts ) 1.05V0  Vc (k )  1.03V0

u (t )  0
ref
1.03V0  Vc (k )  0.97V0 (4.5)
 ref
u (t  ts ) 0.97V0  Vc (k )  0.95V0
u ref (t  t )  k (V (t )  0.95V ) Vc (k )  0.95V0
 s 2 c 0
where Vc is the voltage of the critical bus, V0 is the voltage of the transformer bus,
which is set to 1.0p.u. in this thesis. Parameters k1 and k2 are the control gains, which
affect the convergence speed and control accuracy of the distributed control, and ts
is the sampling interval.
The distribution systems are normally allowed a maximum voltage deviation of 5%-
10% (depending on national standards) from the secondary of the transformer to the
customer end. In this chapter, the upper and lower voltage references in (4.5) are
chosen as 1.05V0 and 0.95V0 respectively. So the proposed method can regulate the
voltage of the critical bus within the allowable range if EVCs can provide enough
real and reactive power. According to (4.5), if the voltage of the critical bus Vc goes
beyond 1.05V0, uref will start to increase. If the voltage of the critical bus Vc reduces
below 0.95V0, uref will start to decrease. If Vc is between 1.03 V0 and 0.97V0, uref will
become 0. Otherwise, uref will keep the value of previous step.
The information state uref of the virtual leader is then communicated to each EVC
through communication links. The instantaneous communication graph can be
102
represented by the following matrix:
 s11 (t ) s12 (t ) s1N (t ) 

 s (t ) s (t ) s2 N (t ) 
S (t )   21 22
(4.6)
 
 
 sN 1 (t ) sN 2 (t ) sNN (t ) 
where sij denotes the communication link between the ith and jth EVCs. sii = 1 for all
i, sij = 1 if the information state of the jth bus is received by the ith bus at time t, and
sij = 0 if otherwise. In this thesis, it is assumed that all EVCs only communicate with
their neighbours, which means all s(i)(i-1) = 1 and s(i)(i+1) = 1.
The communication graph can be known by each EVC as a prior knowledge.

However, this restricts the application when there is a communication failure or new
EVCs participate for voltage regulation. Under these conditions, global
communication graph will change due to the addition and removal of certain EVCs.
Hence, a graph discovery algorithm is proposed to determine the matrix S and the
number of EVCs N participated in the voltage regulation in a distributed way.
It is assumed that each EVC i is assigned a unique identifier ID(i), and Gi(k) denotes
the neighbor set obtained by EVC i. The graph discovery algorithm is designed as
follows:
1) At k=0, each EVC i initializes the set Gi(k) as Gi(k)={ID(i) [ID(j)]}, where ID(j)
is identifier of the EVC j which has communication link with EVC i. Gi(0) will
be sent to all its neighbors EVC j.
2) For each step k≥1, EVC i updates its set Gi(k) as Gi (k )  Gi (k  1) G j (k  1) .
3) If Gi (k )  Gi (k  1), then EVC i stops exchanging information with its neighbors,

and this time instant is recorded as K. Otherwise, go to step 2.
4) Finally, the total EVC number N can be extracted by counting the number of
103
elements in set Gi(K), which is also the dimension of matrix S. The identifier of
EVC which participates in the voltage regulation will be known by each EVC.
With information state of virtual leader uref and communication graph S obtained
above, the utilization ratio for the EVC i is updated as follows:
For EVC i at critical bus,

N
ui (t )   dij (t )u j (t  ts )  dii (t )u ref (t ) (4.7)
j i
For EVC i at non-critical bus,

N
ui (t )   dij (t )u j (t  ts ) (4.8)
j 1
where dij(t) is the (i, j) entry of a row stochastic matrix (i.e., row sum of 1) D(t) which
can be found in each discrete time data exchange as follows:
ij sij (t  ts )
dij (t )  N (4.9)
  s (t  t )
j 1
ij ij s
where the weights ωij are set to 1 in this thesis to share the required power equally
among distributed EVCs and sij are the entries of the communication matrix, given
in (4.6).
4.3.2 Stage Two: Localized Power Allocator
The stage two of the proposed voltage regulation method is to determine the active
and reactive power outputs of each EVC by power allocator. There are two
requirements for designing the power allocator of each EVC. Firstly, the voltage
regulation target required by the first control stage must be achieved within the SoC
104
limits of each EV and the power limits of each EVC. As mentioned before, the power
capacity of the EVC Smax is 3.6kVA. So the maximum charging/discharging power
output of the EVC Pmax is 3.6kW. SoCmax and SoCmin are the maximum and minimum
SoC limits of the EV’s battery pack, which is 90% and 10% respectively.
The second requirement is that the EV needs to be charged to a desired SoC level of
80% before leaving. Therefore, the power allocator needs to determine the mode of
operation of the EV plugged in the EVC. In V2G mode, the EV can participate in the
voltage regulation and work as an energy storage device. The real power of the EV
can be utilized dependent on the utilization ratio and real-time SoC. While in
maximum power charging mode, the EV cannot participate in the voltage regulation
and work as a constant power load in the system.
The following algorithm is developed to judge whether the EV plugged in EVC i

will work as V2G mode or maximum power charging mode.
1) At k=1, the EV’s current SoC value SoC(1) and estimated departure time td are
recorded.
2) For each step k≥1, with SoC information SoC(k) and (4.1)-(4.3) of battery model,
the maximum charging current Imax at time instant k is derived as follows:
1
I max (k )  (Voc (k )  Voc (k ) 2  4 R(k ) P max ) (4.10)
2 R( k )
where Voc and R can be obtained by substituting SoC(k) into (1) and (2).
3) With maximum charging current calculated by (4.10), SoC(k+1) of the next step
can be as estimated as follows:
I max (k )t
SoC (k  1)  SoC (k )  (4.11)
CB
4) If SoC(k) ≥ 80%, this time instant is recorded as K, and go to the next step.
Otherwise, go back to step 2. K∆t is the charging time required in which the EV
105
is charged from SoC(1) to 80% with maximum charging power.

5) Compare K∆t with the time left for the EV before leaving (td - t).
If K∆t ≥ (td - t), the EV will work in maximum power charging mode.
If K∆t <(td - t), the EV will work in V2G mode.
In the above algorithm, the minimum charging time of the EV (i.e., the charging time
with the maximum charging power) is estimated through steps 1-4. The EV’s
minimum charging time is compared with the time left before leaving to determine
the mode of operation in step 5. When the EV operates in V2G mode, real power
output of the EVC can be controlled by the utilization ratio u considering the power
limits and SoC limits of the first requirements as follows:
 P max ui (t )  1& SoCi (t )  0.1

ui (t ) P
max
0  ui (t )  1& SoCi (t )  0.1

Pi ref (t )  ui (t ) P max  1  ui (t )  0 & SoCi (t )  0.9 (4.12)
 max
 P ui (t )  1& SoCi (t )  0.9
0 otherwise
When the EV operates in maximum power charging mode, the EV must be charged
with maximum charging power Pmax.
The reactive power output of each EVC is a supplement for real power output when
there is no EV plugged in the EVC or the remaining capacity of the EV is not enough
for the voltage regulation. The reactive power output of each EVC is determined by
the utilization ratio u and reactive power limits as follows:
Q max (t ) ui (t )  2

(ui (t )  1)Q (t ) 1  ui (t )  2
max

Qiref (t )  0  1  ui (t )  1 (4.13)

(ui (t )  1)Q (t )  2  ui (t )  1
max
Q max (t ) ui (t )  2

106
The maximum reactive power which can be provided by each EVC is determined by
the power capacity and real power output of the EVC.
Qmax (t )  (S max )2  Pi ref (t )2 (4.14)
Finally, the power output of each EVC will be provided according to Pi ref and Qiref
at each time instant.
A 17-bus distribution network is implemented in Matlab/Simulink to evaluate the

performance of the proposed voltage regulation method. The configuration of the
test distribution network is shown in Figure 4.5. It is assumed that there are both PV
generation and load demand at each bus, and totally 10 buses installed the EVCs in
the distribution system, as shown in Figure 4.5. The residential distribution network
is connected to the utility gird through a 100kVA, 22/0.4kV transformer. The voltage
at the secondary of the transformer is set as 230 V (1.0 p.u.). The distribution network
has one main feeder and three laterals. The bus to bus distances of the main feeder,
laterals 1, 2 and 3 are 50m, 100m, 40m and 30m respectively. Buses 6, 12 and 16 are
the critical buses which potentially have the highest voltage in the system. The
parameters of the simulation system are given in Table 4.2.
To evaluate the proposed control method is a realistic environment, real PV penal

and residential load profiles are used in the simulation. The PV output data are
collected every 5 minutes using the GL 130 PV module in the summer 2014 at Clean
Energy Research Laboratory, Nanyang Technological University, Singapore. The
load profiles are typical residential load profiles, which is presented in [6]. The PV
and load data are scaled up for this study, which is shown in Figure 4.6. In addtion,
this study focuses a distribution network in a community-scaled area. Therefore,
107
identical PV and load profiles can be applied for each bus. Two test cases are
conducted and discussed in the following subsections. Table 4.3 shows the usage
data of distributed EVCs for two test scenarios.
Figure 4.5 Configuration of the 17-bus residential distribution network.
Parameter Value
System Frequency 50Hz
MV/LV Transformer Size 100kVA, 22/0.4kV
Conductor Size 0.549 + j0.072 Ω/km
EVC Rating 3.6kVA
Nissan Leaf EV capacity 24kWh
Distributed control (k1, k2) 0.01,0.01
108
Table 4.3 Usage data of distributed EVCs
EV Data
EVC ID (i) Initial Arrival Departure
SoC (%) Time (h) Time (h)
EVC 1 50 16:00 06:00
EVC 2 45 17:00 07:00
EVC 3 40 18:00 08:00
EVC 4 35 19:00 08:30
EVC 5 30 20:00 07:30
EVC 6 25 21:00 06:30
EVC 7 20 10:00 18:00
EVC 8 30 12:00 20:00
EVC 9 40 14:00 22:00
EVC 10 40 08:00 07:59
Figure 4.6 PV generation and load profiles for simulation study.
109
4.4.1 Test Case 1: Sunny PV and Light Load Condition
The first test case demonstrates the proposed method under sunny PV generation and
light load condition. The voltage profile of each bus with uncoordinated charging
(i.e., charging with maximum power once plugged in) is shown in Figure 4.7. The
voltage profiles with the proposed voltage regulation method are shown in Figure
4.8. As shown in Figure 4.7, there are voltage rise and drop beyond the allowable
voltage range [0.95p.u., 1.05p.u.] in this distribution network. From 10.2h to 15.1h
in the daytime, buses 8-16 violate the voltage upper reference 1.05p.u. and the
highest voltage is 1.067p.u. at bus 16 at 14h. From 18.5h to 22.5h in the evening,
buses 8-16 violate the voltage lower reference 0.95p.u. and the lowest voltage is
0.941p.u. at bus 16 at 20h. As shown in Figure 4.8, with the proposed method applied
for each EVC, the voltages can be regulated within the allowable voltage range for
both peak PV generations in the daytime and peak load generation in the evening.
The real and reactive power absorption and the SoC profile of each EVC during the
operation period are shown in Figure 4.9, where the operation of each EVC at each
time is clearly demonstrated. In addition, in Figure 4.9, positive value means
absorption of power by the EVC, while negative value means injection of power by
the EVC.
110
Figure 4.7 Voltage profiles with uncoordinated EV charging for test case 1.
Figure 4.8 Voltage profiles with the proposed method for test case 1.
111
Figure 4.9 Real and reactive power absorption and SoC profile of each EVC for test
case 1.
112
The simulation results shown in Figures 4.7-4.9 for test case 1 are further illustrated
in time sequences as follows:
1) 8h≤t<10.3h. Although PV generation is larger than the load demand in this
period, the voltages of all buses are still below 1.05p.u. as shown in Figure 4.8.
Therefore, no voltage regulation is needed during this period. As shown in
Figure 4.9, the real and reactive power outputs of each EVC remain 0.
2) 10.3h≤t<13.2h. The PV generation increases significantly during the midday
and becomes much larger than the load demand. The voltages of critical buses
exceed 1.05p.u., which initiates the distributed voltage control at 10.3 h. The
utilization ratio u is between 0 and 1 in this period and the real power of EVCs
7, 8 and 10 is enough for voltage regulation.
3) 13.2h≤t<15.3h. As PV generation increases further, the real power absorbed
by each EVC is not enough for voltage regulation. The utilization ratio u is
between 1 and 2 and EVCs 1-6 which cannot absorb real power start to absorb
reactive power for voltage regulation. Besides, as shown in Table 4.3, an EV
begins charging at 14h at EVC 9. Therefore, real power instead of reactive
power is absorbed at EVC 9 for voltage regulation from 14h as shown in Figure
4.9.
4) 15.3h≤t<18.7h. Although PV generation decreases significantly, the voltages
of critical buses drop below 1.03p.u. but still above the voltage lower reference
0.95p.u. as shown in Figure 4.8. Therefore, voltage regulation is not needed
during this period. In addition, as the SoC of the EV plugged in EVC 7 is above
80%, mode change is not needed before its leaving at 18h.
5) 18.7h≤t<23.1h. As the load demand increases further in the evening, the
voltages of critical buses drop below 0.95p.u. at 18.7h. The utilization ratio u
decreases between -1 and 0 in this period and the real power outputs of all
available EVCs are enough for voltage regulation. As EVs plugged in EVC 8
and EVC 9 will leave in this period, they need to be charged with maximum
113
charging power until their SoCs reaching 80% before their leaving, as shown
in Figure 4.9.
6) 23.1h≤t≤24h. As the load demand decreases in this period, the voltages of
critical buses increase above 0.97p.u. as shown in Figure 4.8. Therefore, no
voltage regulation is required during this period.
4.4.2 Test Case 2: Cloudy PV and Heavy Load Condition
The second test case demonstrates the proposed method under cloudy PV generation
and heavy load condition. The voltage profile of each bus with uncoordinated
charging for test case 2 is shown in Figure 4.10. The voltage profiles with the
proposed voltage regulation method for test case 2 are shown in Figure 4.11. As
shown in Figure 4.10, there are still bus voltage rise and drop beyond the allowable
voltage range [0.95p.u., 1.05p.u.] in the distribution networks under this condition.
Due to the cloudy weather condition in the daytime, there are several large voltage
variations along the distribution feeder. In the daytime, the voltage rise problem
becomes less severe than test case 1 due to the lower PV generation. While in the
evening, the voltage drop problem becomes more severe than test case 1 due to the
heavy load demand, and the lowest voltage is 0.93p.u. at bus 16 at 20h. As shown in
Figure 4.11, with the proposed method applied for each EVC, the voltages can be
regulated within the allowable voltage range. The real and reactive power absorption
by each EVC and the SoC profile of each EV user during the operation period are
shown in Figure 4.12.
114
Figure 4.10 Voltage profiles with uncoordinated EV charging for test case 2.
Figure 4.11 Voltage profiles with the proposed method for test case 2.
115
Figure 4.12 Real and reactive power absorption and SoC profile of each EVC for test
case 2.
116
The simulation results shown in Figures 4.10-4.12 for test case 2 are further
illustrated in time sequences as follows:
1) 8h≤t<11.2h. Although PV generation is larger than the load demand, the
voltages of all buses are below 1.05p.u. as shown in Figure 4.11. Therefore, no
voltage regulation is needed during this period.
2) 11.2h≤t<14.7h. As PV generation increases in the midday, the voltages of
critical buses are beyond 1.05p.u. at 11.2h. The real power outputs of EVCs 7,
8, 9 and 10 are utilized for voltage regulation if available during this period, as
shown in Figure 4.12. The voltage regulation stops working due to the large
PV variation from 12h to 13.1h.
3) 14.7h≤t<18.4h. The voltages of critical buses drop below 1.03p.u. but still
above the voltage lower reference 0.95p.u. as shown in Figure 4.11. The
voltage regulation is not needed in this period. In addition, the SoC of the EV
plugged in EVC 7 is not fully charged during the daytime, so the EV changes
to maximum power charging mode at 15.4h and is charged to 80% of the SoC
at 18h before leaving as shown in Figure 4.12.
4) 18.4h≤t<19.8h. The voltages of critical buses drop below 0.95p.u. at 18.4h.
EVCs 1, 2, 3, 4, 9 and 10 start to inject real power for voltage regulation as
shown in Figure 4.12. The real power outputs of all available EVCs are enough
for voltage regulation in this period.
5) 19.8h≤t<22.8h. As the load demand increases further, the real power
discharged by each EVC is not enough for voltage regulation. The utilization
ratio u decreases between -1 and -2 during this period. EVCs which cannot
inject real power start to inject reactive power as shown in Figure 4. 12.
6) 22.8h≤t≤24h. As the load demand decreases significantly in this period, the
voltages of critical buses increase above 0.97p.u. as shown in Figure 4.11.
Therefore, voltage regulation is no longer needed in this period.
117
4.5 Conclusion
In this chapter, distributed EVCs have been utilized to regulate the voltages in a LV
distribution network. A lithium-ion battery model has been developed for SoC
estimation. A two-stage voltage control method has been proposed to coordinate the
power outputs of distributed EVCs during daily operation. The power outputs of the
EVCs are determined by both distributed voltage control and localized power
allocator. A 17-bus distribution network has been implemented in Matlab/Simulink
to validate the performance of the proposed method. Based on the utilization ratio
generated by distributed voltage control, the power allocator of each EVC can
determine the real and reactive power outputs to realize voltage regulation. The
simulation results have demonstrated that the proposed control method can ensure
the voltages along the distribution network within allowable voltage range under
various operation conditions.
There are still several limitations of the research work in this chapter. First, the state-
of-health (SoH) is not considered in the lithium-ion battery modelling in this research.
However, due to the limited life cycle of lithium-ion battery, the SoH and capacity
fading are important issues to be considered for long time operation. Second, the EV
user data is considered as a given information in this research. In future research,
probability based model can be built to estimate the SoCs of the EVs and the
parking/charging time of the EV users. Furthermore, the research scope is limited to
the voltage regulation and EV charging problem in a low density residential area.
While for different areas (commercial or residential), the time and amount of EV
charging could be quite different. The real-time voltage regulation through V2G
should also consider EV management strategies.
118
References
[1] Land Transport Authority of Singapore, Singapore Land Transport Statistics in

Brief [Online]. Available: http://www.lta.gov.sg/.
[2] X. J. Li, D. Hui, and X. K. Lai, “Battery energy storage station (BESS)-based
smoothing control of photovoltaic (PV) and wind power generation
fluctuations,” IEEE Trans. Sustainable Energy, vol. 4, no. 2, pp. 464-473, Mar.
2013.
[3] A. Rahmoun and H. Biechl, “Modelling of Li-ion batteries using equivalent
circuit diagrams,” Przeglad. Elektrotechniczny (Electrical Review) 88
(7b/2012) 152-156, 2012.
[4] J. G. Pinto, V. Monteiro, H. Goncalves, and J. L. Afonso, “Onboard
reconfigurable battery charger for electric vehicles with traction-to-auxiliary
mode,” IEEE Trans. Vehicular Technology, vol. 63, no. 3, pp. 1104-1116, Sep.
2013.
[5] A. Maknouninejad and Q. Zhihua, “Realizing unified microgrid voltage profile
and loss minimization: a cooperative distributed optimization and control
approach,” IEEE Trans. Smart Grid, vol. 5, no. 4, pp. 1621-1630, Apr. 2014.
[6] J. A. Jardini, C. M. V. Tahan, M. R. Gouvea, S. U. Ahn, et al., “Daily load
profiles for residential, commercial and industrial low voltage consumers,”
IEEE Trans. Power Delivery, vol. 15, no. 1, pp. 375-380, Jan. 2000.
119
120
Distributed ESSs in Islanded Microgrid Chapter 5
Chapter 5
Hierarchical Control and Power Management of Islanded

Microgrids with Energy Storage Systems and Renewable
Energy Sources
In this chapter, a hierarchical control scheme is presented for an

islanded AC microgrid formed by PVs, ESSs and multiple loads. Both
the short-term and long-term power imbalance in the microgrid is solved
by the proposed control scheme.
121
5.1 Introduction
It is of great importance to ensure efficient utilization of RESs while avoiding

depletion or saturation of the ESSs in the microgrids under various operation
conditions. In this chapter, a hierarchical control scheme is presented for PV systems,
ESSs and loads in an islanded AC microgrid to deal with both the short-term and
long-term power imbalance. ESSs will function as UPS to make the microgrid
continue working at islanded condition. To achieve fast dynamic response and small
steady-state error of VSIs, a novel N-step MPC is developed for the primary control.
Depending on different cost functions of MPC method, the VSIs of each unit can
operate in CCM or VCM. The secondary control is taken effect when the primary
control cannot handle the power imbalance in the microgrid. The secondary control
is achieved through communication links between each agent using the consensus
algorithm. This control scheme ensures the power balancing and power management
can be achieved hierarchically under different operation conditions in the microgrid.
5.2 System Description
5.2.1 System Configuration
The configuration of an ac microgrid with multiple communication agents is shown

in Figure 5.1. The microgrid consists of two EESs, two PV systems and four
distributed loads which is interfaced to the utility grid through a 1MVA, 22/0.4kV
transformer. The ESSs and PVs are connected to each bus through VSIs. It assumes
that there is communication agent at each bus, which can communicate with other
agents through low bandwidth communication links [1]. The distributed control
signal of each agent will be executed by the local ESS/PV and load.
122
Figure 5.1 The configuration of an ac microgrid with communication agents.
In grid-connected condition, the utility grid can support any shortage or surplus of
power in the microgrid. In this condition, the primary control of all VSIs will operate
in CCM. The charging/discharging of the ESSs can be controlled by secondary
control for peak shaving [2]. While in islanded condition, which is the focus of this
research, the circuit breaker (CB) of the microgrid shown in Figure 5.1 will be open.
The ESSs will function as UPS and coordinate with PV systems and loads to supply
power imbalance within the microgrid. For primary control, ESS-1 operates in VCM
to control the voltage and frequency of the islanded microgrid, and ESS-2, PV-1 and
PV-2 operate in CCM to supply the required power. ESS-1 will work as a master
unit while ESS-2 and PVs will work as slave units. For secondary control, if the
master unit is not able to balance the power in the microgrid, the slave units will be
coordinated for power balancing. In this research, ESS-1 has large power rating and
capacity than ESS-2, so it is more suitable to be chosen as a master unit, which will
provide more stable voltage support for other devices in the islanded microgrid. If
all ESSs are not enough to supply the power imbalance, PV curtailment or load
shedding will be conducted. For practical applications, in order to balance the total
load demand for a long time, backup generators can be started up during the
operation of ESSs. After that, the backup generators will play the role of UPS to
123
continue supplying the load demand in the islanded microgrid.
5.2.2 Structure and Control Scheme of ESS/PV
Figure 5.2 shows the structure of each ESS/PV and its control scheme. The
configuration of each ESS/PV consists of two parts: the dc source and power
conversion stage (PCS). The dc source of ESS refers to battery management system
(BMS) of battery bank. The function of the BMS is to balance the SoC and the output
voltage of each cell in the battery bank [3]. The dc source of PV system refers to the
PV panels and the dc/dc boost converter, which is controlled by maximum power
point tracking algorithm (MPPT) to obtain the maximum power output [4]. As this
research is focused on the control of PCS, the dc part will be simplified into dc
voltage source, which provides constant dc link voltage.
Figure 5.2 Structure and hierarchical control scheme of each ESS/PV unit.
124
The PCS of the ESS/PV contains a three-phase dc/ac VSI and a LC filter. The VSI
of each unit is controlled with the proposed hierarchical control scheme to realize
power balancing in islanded microgrid. As shown in Figure 5.2, there are two modes
of operation for primary control, which can be changed according to the MPC cost
function. For the microgrid in Figure 5.1, ESS-1 operates in VCM and ESS-2, PV-1
and PV-2 operate in CCM at primary control. In secondary control, the leader control
is only for ESS-1, which measures the outputs to initiate the consensus algorithm.
Other agents will update their distributed control signals ui with the control signals
received from neighbor agents. Then the distributed control signals will be allocated
to the local controllable units of each agent (ESS, PV and load) to determine how to
operate them. The detailed control scheme will be further illustrated in the following
sections.
5.3 Model Predictive Control of Voltage Source Inverters
5.3.1 Modeling of Voltage Source Inverters
The state-space model of the inverter is used to predict the future behavior of the
state variables, which will be used by the cost function to determine the switching
signals for the inverter. Originated from the fact that the VSI can be modeled as a
system with finite number of switching states, the implementation of MPC can be
further simplified [5]. For the VSI shown in Figure 5.2, the switching states (S1-S6)
of the switches have only binary values “1” and “0” in a closed state and an open
state respectively. The lower switches have the complementary values of their upper
switches. Therefore, the switching signals of each phase are defined by
Sa  S1  S4 , Sb  S2  S5 , Sc  S3  S6 (5.1)
125
The output voltage of the VSI in vectorial form vi ( ) can be expressed related to the
switching signals by
2
vi ( )  ( Sa  Sb e j (2 /3)  Sc e j (2 /3) )Vdc (5.2)
3
Totally eight output voltage vectors can be obtained, considering eight possible
combinations of the switching signals Sa, Sb and Sc, which is further interpreted in
[5]. In order to simplify the modeling of three-phase VSI, all the parameters in abc
frame are transformed to vectors in  frame. The vector u( ) is a complex number,
which can be obtained using the following transformation:
ua 
j (2 /3) 
 ub 
2  j (2 /3)
u( )  u  ju  1 e e  
(5.3)
3
uc 
With the transformation (5.3), the filter current iL ( ) , the capacitor voltage vC ( ) ,
and the output current io ( ) can be expressed. Applying Kirchhoff’s current and
voltage laws to the LC filter of the VSI, the following equations are obtained
dvC ( )
C  iL ( )  io ( ) (5.4)
dt
diL ( )
L  vi ( )  vC ( )  RiL ( ) (5.5)
dt
Compared to the high sampling frequency, the output current is changing very slowly.
It can be assumed that
dio ( )
0 (5.6)
dt
126
Discretize (5.2), (5.4), (5.5) and (5.6) using the forward Euler difference method,
and define state variable as
T
x(k )  vC ( ) (k ) iL ( ) (k ) io( ) (k )  (5.7)
The discrete time state-space model with a sampling time TS can be obtained as
follows:
x(k  1)  Ax(k )  Bvi ( )
y (k )  Cx(k ) (5.8)
where
 Ts T
 1  s
C C 0
  T  1 0 0 
A   s  s 1 0  , B   s  , C  
T RT
 L  .
L L 0 1 0 
  0
 0 0 1   
 
5.3.2 Proposed Primary Model Predictive Control
The block diagram of the proposed MPC for primary control of VSI is shown in
Figure 5.3. Using the state-space model (5.8), the measured state variable x(k) is used
to predict its value at next sampling instant for all the possible output voltage vectors
of the VSI. Here, a N-step prediction is adopted to reduce the steady-state error of
the controller.
127
Figure 5.3 Proposed model predictive controller for primary control.

The reference capacitor voltage vC ( ) (k ) is generated by
 230sin(2 fkTs ) 
j (2 /3) 
 230sin(2 fkTs  2 / 3) 
2  j (2 /3)
v
C ( ) (k )  1 e e  
(5.9)
3
 230sin(2 fkTs  2 / 3) 
With the values of P(k) and Q(k) at sampling instant k obtained from the secondary

control, the reference output current io ( ) (k ) can be obtained as follows:
P(k )  jQ(k )
io( ) (k )  (5.10)
vC ( ) (k )

Then the reference filter current iL ( ) (k ) can be derived from (5.5) as follows:
Cf
iL* ( ) (k  1)  io*( ) (k  1)  (vC* ( ) (k )  vC* ( ) (k  1)) (5.11)
Ts
As mentioned above, there are two modes of operation of VSI, which are VCM and
CCM. The cost function of VCM is defined as the difference between the predicted
capacitor voltage after N-step using the state-space model and its reference value as
follows:
128
JV  vC* ( ) (k )  vC ( ) (k  N ) (5.12)
While the cost function of CCM is defined as the difference between the predicted
inductor current after N-step using the state-space model and its reference value as
follows:
J I  iL* ( ) (k  1)  iL ( ) (k  N ) (5.13)
where JV and JI are the cost functions of the proposed primary MPC in VCM and
CCM respectively. N is number of the prediction steps which is set to 3 in this
research.
To choose the optimal control signals, the voltage vector vi ( ) that minimizes the
cost function is selected, and the switching signals corresponding to this voltage
vector are chosen as control signals of the VSI at next sampling instant.
5.4 Consensus Based Distributed Control
5.4.1 Communication Network and Consensus Algorithm
Consensus based distributed control offers interaction of control units in a dynamic

multi-agent system. Each control agent exchanges data with some other agents on a
communication network and processes the data to control its associated system.
Graph theory is used for graphical representation of the communication network [6].
The communication network of a multi-agent system can be described by a graph,
which is defined as G = (V, E) with a set of N nodes V = {v1, v2, …,vN} and a set of
edges E = V × V. Each node is assigned with a agent in the system, and edges
represent communication links for data exchange. If communication links are
bidirectional, (vi, vj) ∈ E ⇒ (vj, vi) ∈ E ∀i, j, the graph is said to be undirected,
otherwise it is directed. The set of neighbors of ith node is denoted Ni. Equivalently,
129
if j ∈ Ni, then vi receives information from vj. A matrix called adjacent matrix S = [sij]
is associated to the edges. sij represents the weight for information exchanged
between agents i and j, where sij = 1 if agents i and j are connected through an edge
(vi, vj) ∈ E, otherwise, sij = 0.
In this chapter, the distributed control problem can be categorized as cooperative

tracking problem, which is aimed to have consensus for each agent with a control
objective provided by the leader [7]. The control objective is shared to each agent
through the communication network to reach consensus. The node connected to the
leader is called pinned node and the graph contains at least one pinned node.
The consensus algorithm updates the distributed control signals for both pinned and
unpinned agents as
ui (k  1)   d u (k )  g u
jNi
ij j i
ref
(k ) (5.14)
where uj(k) is the local information discovered by agent j at iteration k, ui(k+1) is the
update of ui(k) at iteration k+1, dij is the communication coefficient between agents
i and j, gi is the pinning gain of the agent i which is positive if there is a leader at this
agent, and is zero otherwise, and u ref (k ) is control objective function of the leader.
As discussed in [7], different methods for dij determination provide different
converging speeds. The improved Metropolis method is adaptive to changes of
communication network topology and is able to provide near optimal converging
speed. The dij updating law is represented by
sij
dij  N
, i  1, 2,..., n. (5.15)
s
j 1
ij
where sij are the entries of adjacent matrix S.
130
5.4.2 Proposed Secondary Distributed Control
For islanded microgrid with only proposed primary control scheme, the power is
balanced by VCM-ESS. Any power imbalance in the microgrid will be supplied by
the VCM-ESS. However, when there is large or long-term power imbalance which
leads to rated power or SoC limit violation, the secondary control will be responsible
for coordination of each available unit in the microgrid. It is noted that the operation
of the ESS should also take into account the power and energy constraints. The power
limits and SoC limits of the ESS can be described by (5.16) and (5.17) respectively.
 PESS
rated
 PESS (k )  PESS
rated
(5.16)
SoC min  SoC (k )  SoC max (5.17)
The SoC of the ESS can be simply estimated by the ampere-hour counting method
as
1
CESS 
SoC  SoCt 0  PESS dt (5.18)
where SoCt 0 is the initial value of SoC, CESS is the capacity of the ESS, and PESS is
the power output of the ESS.
As mentioned above, the control objective determined by the leader is shared through
communication network to each available agent. In this research, the leader is located
at VCM-ESS. The objective function of the leader uref is updated at discrete time
steps. If the measured charging power of the VCM-ESS goes beyond upper reference
Pup, uref is increased. If P goes below lower reference Plow, uref is reduced. In between,
uref is equal to the value of previous step. In general, the control objective of the
leader is updated as follows:
131
 u ref (k  1)  k1 ( P(k )  Pup (k )) if P(k )  P up (k )


u ref (k )   u ref (k  1) if P low (k )  P(k )  P up (k ) (5.19)
u ref (k  1)  k ( P(k )  Plow (k )) if P(k )  P low (k )
 2
where parameters k1 and k2 are the control gains, which affect the convergence speed
and control accuracy of the distributed control. Considering the operation limit of
ESS, the upper and lower power reference is designed as follows:
SoC max  SoC1 (k ) rated

P (k ) 
up
PESS1 (5.20)
SoC max  SoC min
SoC1 (k )  SoC min rated

P (k )  
low
PESS1 (5.21)
SoC max  SoC min
With the proposed secondary control, the power output of the VCM-ESS will be
controlled within the upper and lower power references. According to (5.20), (5.21),
the power references are functions of SoC, which will change according to the SoC
of the VCM-ESS for each time instant. Both the upper and lower power references
will increase during the ESS discharging condition. On the contrary, the upper and
lower power references will decrease during the ESS charging condition. Thus this
design restrains the overcharging when the SoC becomes high and the
overdischarging when the SoC becomes low.
5.4.3 Distributed Signal Allocation
The distributed secondary control will initiate once there is power reference violation
of (5.19) and each agent will update its distributed signal according to (5.14). Finally,
each agent will have the same value which is equal to the control objective of the
leader uref when they reach consensus (i.e., steady-state). The allocation of
distributed signal is designed to determine how to utilize this control signal for power
132
balancing and SoC management of the entire microgrid. The detail is explained as
follows:
If ui=0, the VCM-ESS is enough to balance the power in the microgrid with only
primary control, so secondary control will not work in this condition.
If 0<ui<1, the power output of the VCM-ESS is larger than the upper power reference.
The CCM-ESS is charged to balance the power with the distributed signal of
secondary control. Similarly, if -1<ui<0, the power output of the VCM-ESS is
smaller than the upper the power reference and the CCM-ESS is discharged. The
charging and discharging of CCM-ESS are determined by the following equations:
  (k ) PESS
rated
2 ui (k )  1

u (k ) (k ) PESS 2 0  ui (k )  1
rated
PESS 2 (k )   i (5.22)
ui (k )  (k ) PESS 2  1  ui (k )  0
rated
   (k ) P rated ui (k )  1
 ESS 2
 SoC max  SoC2 (k )

 SoC2 (k )  SoC up
 (k )   SoC max  SoC up (5.23)
1 SoC2 (k )  SoC up

1 SoC2 (k )  SoC low


 (k )   SoC (k )  SoC min (5.24)

2
SoC2 (k )  SoC low
 SoC low  SoC min
where SoCup and SoClow are the upper and lower SoC references to reduce the power
output of the CCM-ESSs respectively. α and β are the utilization ratios of ESS-2 for
the charging and discharging process respectively.
133
If ui>1, the CCM-ESS is already charging with the maximum power. PV generation
curtailment will be conducted by the CCM-PV to reduce the PV generation. The
curtailed power is defined by
max
PPV (k )  (ui  1) P MPPT (k ) (5.25)
where PMPPT is the maximum power determined by MPPT.

If ui<-1, the CCM-ESS is discharging with the maximum power. Load shedding will
be conducted to shed the non-critical loads to reduce the total load demand. In order
to avoid repetitive ON-OFF switch of loads, a relay control is used for load shedding
and load recovering, as shown in Figure 5.4. The priority for load shedding from
high to low is Load-3, Load-2 and Load-1. uL-ON and uL-OFF represent the thresholds
to switch on and off each load respectively.
ui Load recovering
uL1ON
uL 2ON
uL 3ON Load-1
uL1OFF Load-2
uL 2OFF Load-3
Load shedding
uL 3OFF
PL
Figure 5.4 Thresholds of load shedding and load recovering.
134
The ac microgrid shown in Figure 5.1 is implemented in Matlab/Simulink as the

platform of proposed hierarchical control scheme. The parameters of the simulation
system are shown in Table 5.1 and the parameters of the proposed control scheme
are shown in Table 5.2. To validate the performance of the proposed control scheme,
both short-term and long-term test cases are demonstrated in this section. It is noted
that small capacity of the ESS is chosen to make SoC change faster in this research.
However, the proposed control scheme can also be applied to microgrids with larger
ESS capacity. ESS-1 is controlled in VCM while ESS-2 and PVs are controlled in
CCM at primary control. It is assumed that each agent can only communicate with
its neighbors. The initial SoCs of ESS-1 and ESS-2 are set to 50% for all test cases.
Parameter Value
Nominal Bus Frequency 50Hz
Nominal Bus Voltage 230V
Filter Capacitance 20μF
Filter Inductance 1.3mH
Load Ratings: Real/ Reactive Power
Load-1 3kW/0.7kVAr
Load-2 4.2kW/0.9kVAr
Load-3 4kW/ 1.2kVAr
Critical Load-4 6.8kW/ 1.5kVAr
PV Ratings: Maximum Power

PV-1 7.8kW
PV-2 9kW
ESS Ratings: Capacity/Rated Power

ESS-1 0.14kWh/5kW
ESS-2 0.07kWh/4kW
135
Table 5.2 Parameters of the proposed control scheme
Parameter Value
Leader Control (k1, k2) 0.001, 0.001
Communication Speed 10ms
min
SoC Safety Range (SoC ,
0.1, 0.9
SoCmax)
SoC Reference (SoClow, SoCup) 0.3, 0.7
Load Shedding Thresholds
(uL1-OFF, uL2-OFF, uL3-OFF) -1.2, -1.5, -2
Load Recovering Thresholds
(uL1-ON, uL2-ON, uL3-ON) 0, -0.25, -0.5
5.5.1 Test Case 1: Short-Term PV Generation and Load Demand Variation
The first test case demonstrates the performance of the proposed method under short-
term large power variation condition. The microgrid with the proposed control
method will be tested under large PV/load variation to evaluate the instant power
balancing ability. Figure 5.5 shows the events of the simulation in test case 1. The
initial state and switch on/off behaviors of PVs and loads during the simulation
period are shown on the timeline. Figure 5.6 shows the real power output of each
ESS in test case 1. The values are lined out at the end of each period. Figure 5.7
shows the corresponding real power output of each PV in this test case. The
corresponding profiles of SoC and control objective of the leader during this period
are shown in Figures 5.8 and 5.9 respectively. As shown in Figure 5.8, the SoC will
change according to the real power outputs shown in Figure 5.6. Figure 5.9 shows
that control objective of the leader uref is between -1 and 1 from 0 to 5s, and increases
to 1.2 from 5s to 6s. According to the distributed signal allocation defined in this
chapter, each agent will take measures properly.
136
Figure 5.5 Events of the simulation for test case 1.
Figure 5.6 Real power outputs of ESSs for test case 1.
Figure 5.7 Real power outputs of PVs for test case 1.
137
Figure 5.9 Control objective of the leader for test case 1.
The results in Figures 5.6-5.9 are further explained in time sequences as follows:
1) 0≤t<1s. Initially, the power generation of PV-2 is larger than power
consumption of Load-4 for 2.2kW. The surplus power is charged by ESS-1 as
shown in Figure 5.6. The power output of ESS-1 is within the power reference
calculated by (20) and (21), so only primary control is operated.
2) 1≤t<2s. As Load-1 is switched into the microgrid at t=1s, ESS-1 is discharged
with 0.8 kW to match the power imbalance as shown in Figure 5.6. The power
output of ESS-1 is still within the power reference, so the primary control
responds fast to deal with the sudden load change.
138
3) 2≤t<3s. As Load-2 is switched into the microgrid at t=2s, the load demand is
larger than PV generation for 5kW. The power output of ESS-1 increases larger
than the upper power reference, so uref drops to -0.6 and the secondary control
starts to work. Both ESS-1 and ESS-2 are discharged with 2.5kW to supply the
power imbalance as shown in Figure 5.6.
4) 3≤t<4s. As PV-1 is switched into the microgrid at t=3s, the PV generation is
larger than load demand for 2.8kW for this period. The power output of ESS-
1 decreases smaller than the lower power reference, so ESS-1 and ESS-2 are
charged with 2.5kW and 0.3kW respectively as shown in Figure 5.6.
5) 4≤t<5s. As Load-1 is switched off the microgrid at t=4s, the PV generation is
large than load demand for 5.8kW for this period. ESS-1 and ESS-2 are
charging with 2.5kW and 3.3kW respectively to match the power imbalance as
6) 5≤t≤6s. As Load-2 is switched off the microgrid at t=5s, the power imbalance
increases further to 10 kW. ESS-1 and ESS-2 are charging with 2.5kW and
4kW respectively but still not enough to balance the power in the microgrid.
uref increases further to 1.2 as shown in Figure 5.9. Therefore, the power output
of PV-1 and PV-2 are curtailed to 6.4kW and 5.9kW as shown in Figure 5.7.
5.5.2 Test Case 2: Long-Term Power Imbalance and PV Curtailment
The second test case validates the performance of the proposed method under long-
term power imbalance. In this condition, the PV generation is larger than the load
demand and the power imbalance lasts long enough to test how of SoC changes
influence the proposed control scheme. If the remaining capacity of the ESSs is not
enough to absorb the surplus PV generation, PV curtailment will be conducted
proportional to the power rating of the PV system. All the PVs and loads except
Load-1 are connected into the microgrid in this test case. The real power outputs of
139
ESSs and PVs are shown in Figures 5.10 and 5.11 respectively. Figures 5.12 and 5.13
show the corresponding SoC profiles and power reference changes in test case 2
respectively.
140
Figure 5.13 Power reference changes for test case 2.
As shown in Figure 5.12, the SoC will change according to the real power outputs
of ESS-1 and ESS-2 shown in Figure 5.10. The power references shown in Figure
5.13 will change according to SoC of ESS-1, as determined by (5.20) and (5.21). The
results in Figures 5.10-5.13 are further explained in time sequences as follows:
1) 0≤t<30s. Initially, the surplus power generation is 1.8kW, which is within the
power reference as shown in Figure 5.13. Therefore, only ESS-1 is operated
with primary control to balance the power as shown in Figure 5.10.
2) 30≤t<165s. At t=30s, the charging power of ESS-1 (negative value of power
output in Figure 5.10) reaches the upper power reference. The secondary
141
control starts to work and ESS-2 absorbs the remaining surplus power as
3) 165≤t≤300s. The charging power of ESS-2 starts to reduce due to the notably
decease of utilization ratio α in (5.23). As the total power of ESS-1 and ESS-
2 is not enough to balance the power in the microgrid, PV curtailment will be
conducted based on the distributed control signals. The power of PV-1 and PV-
2 will be gradually reduced to 8kW and 7kW as shown in Figure 5.11.
5.5.3 Test Case 3: Long-Term Power Imbalance and Load Shedding
Similar to the second test case, the third test case validates the performance of the
proposed method for long-term load demand larger than PV generation. Load
shedding will be conducted when the remaining capacity of ESSs is not enough to
supply the load demand in this condition. Initially, all the PVs and loads are
connected into the microgrid in this test case. The real power outputs of ESSs and
PVs are shown in Figures 5.14 and 5.15 respectively. It is assumed that all PV
generation (PV-1 and PV-2) reduces to 80% of maximum power output at 400s as
shown in Figure 5.15. Figure 5.16 shows the corresponding SoC profiles in test case
3. As shown in Figure 5.16, the SoC will change according to the real power outputs
of ESS-1 and ESS-2 shown in Figure 5.14 for this test case.
142
143
The results in Figures 5.14-5.16 are further explained in time sequences as follows:
1) 0≤t<95s. As the power shortage is 1.2kW during this period, only ESS-1 is
discharged for power balancing as shown in Figure 5.14.
2) 95≤t<295s. At t=95s, the discharging power of ESS-1 reaches the lower power
reference. To balance the power in this condition, secondary control starts to
work and ESS-2 will provide the remaining shortage power as shown in Figure
5.14.
3) 295≤t≤400s. Due to the total discharging power of ESS-1 and ESS-2 are not
able to support the power shortage, distributed control signals drop below -
1.2 at 295s. Therefore, Load-1 will be shed due to the violation of load
shedding threshold uL1-OFF. After shedding of Load-1, the total PV generation
becomes larger than the load demand for 1.8kW. ESS-1 changes back to
charging mode during this period.
4) 400s≤t<470s. As PV generation drops to 80% of maximum power output in
Figure 5.15, the total power shortage becomes 1.56kW. ESS-1 changes to
discharging mode to balance the power shortage as shown in Figure 5.14.
5) 470s≤t<500s. As the distributed control signals drop to -1.5 at t=470s, which
is load shedding threshold uL2-OFF, Load-2 will be shed to deal with the power
shortage in the microgrid. After the load shedding, ESS-1 is charged with
2.64kW for power balancing as shown in Figure 5.14
5.6 Conclusion
In this chapter, a hierarchical control scheme has been designed to achieve power
balancing and power management in islanded microgrids. The primary control is
performed by a novel MPC, providing fast dynamic response and small steady-state
errors. The secondary control is initiated when the primary control is not enough to
deal with the power imbalance in the microgrid. Then the distributed control signal
144
of each agent is updated autonomously through the signals received from neighbors.
The allocation of the distributed control signal has been properly designed to
effectively coordinate each available unit in the microgrid. A simulation platform has
been implemented in Matlab/Simulink to evaluate the performance of the
hierarchical control scheme under different test scenarios. The simulation results
have validated the effectiveness of the proposed control scheme to deal with both
short-term and long-term power imbalances in the microgrid. Proper actions such as
PV curtailment and load shedding are conducted to ensure the safety operation of the
ESSs. With the proposed control scheme, the power balancing and power
management of the microgrid can be achieved hierarchically.
The limitations of the research work in this chapter are discussed as follows. First,
the proposed hierarchical control method is only applied for a small scale islanded
microgrid. The effectiveness of primary model predictive control and secondary
distributed control need to be validated through hardware experiment. Furthermore,
real-time control and operation of larger microgrid with more complex system
structure is also needed. In addition, the reliability issues of the communication
network are not considered in this research. In practical situation, there may be
communication topology changes, communication failures and communication
delays in the communication network. These problems related to communication
layer are still needed to be investigated in future research.
References
[1] H. Q. Li, G. Chen, T. W. Huang, and Z. Y. Dong, “High-performance consensus

control in networked systems with limited bandwidth communication and
time-varying directed topologies,” IEEE Trans. Neural Networks and
Learning Systems, in Press.
145
[2] Y. Z. Wang, X. Lin, and M. Pedram, “Adaptive control for energy storage
systems in households with photovoltaic modules,” IEEE Trans. Smart Grid,
vol. 5, no. 2, pp. 992-1001, Feb. 2014.
[3] C. H. Kim, M. Y. Kim, H. S. Park, and G. W. Moon, “A modularized two-stage
charge equalizer with cell selection switches for series-connected lithium-ion
battery string in an HEV,” IEEE Trans. Power Electronics, vol. 27, no. 8, pp.
3764-3774, Jan. 2012.
[4] A. Chikh and A. Chandra, “An optimal maximum power point tracking
algorithm for PV systems with climatic parameters estimation,” IEEE Trans.
Sustainable Energy, vol. 6, no. 2, pp. 644-652, Mar. 2015.
[5] J. Rodriguez, J. Pontt, C. A. Silva, P. Correa, et al., “Predictive current control
of a voltage source inverter,” IEEE Trans. Industrial Electronics, vol. 54, no.
1, pp. 495-503, Feb. 2007.
[6] Z. Qu, Cooperative Control of Dynamic Systems: Applications to Autonomous
Vehicles. New York, NY, USA: Springer-Verlag, 2009.
[7] L. Xiao, S. Boyd, and S. J. Kim, “Distributed average consensus with least-
mean-square deviation,” in Proc. 17th International Symposium Mathematical
Theory of Networks and Systems, Kyoto, Jul. 2006.
146
Conclusions and Future Work Chapter 6
Chapter 6
Conclusions and Future Work
In this chapter, the summary and overall conclusions of this thesis are
discussed. The performance of the proposed control methods has been
verified through various test cases in this thesis. The research results
have demonstrated that the proposed methods are effective to control
energy storage systems for different applications in power systems.
Moreover, areas for further research are presented as well, such as
application of energy storage systems for voltage regulation in medium-
voltage distribution networks, renewable intermittency mitigation and
frequency regulation.
147
6.1 Conclusions
In the last few years, electric power systems have undergone many major
transformations from centralized generation to distributed generation. To meet the
world’s rising energy demand, RESs together with ESSs and EVs have been largely
penetrated into modern power systems, which can provide many economic and
environmental benefits. The development of smart grid technologies provides
approaches to realize the benefits of RESs as well as ESSs, and improve the power
quality and system reliability. As the key elements in future smart grid, ESSs are of
great importance for both grid-connected and islanded operations. Many research
and development efforts which aim to provide efficient solutions have been
conducted in this area.
The large penetration of RESs and EVs into the distribution networks or microgrids
changes the voltage profiles dramatically and results in various power quality
problems. Among them, voltage rise issue caused by large PV generation in the
midday and voltage drop issue caused by the uncoordinated charging EVs require
significant research attention. In this thesis, ESSs and EVCs have been utilized to
overcome these voltage problems. Coordinated control methods have been proposed
to control the operation of distributed ESSs and EVCs in the system, considering
storage capacity and power limits of each device. Both upper level control and lower
level control have been well designed to achieve the voltage regulation and
effectively utilize the storage capacity. Besides, the utilization and control of ESSs
in islanded microgrid has also been investigated.
In Chapter 3, distributed ESSs have been utilized to regulate the voltages in LV

distribution networks with high PV penetration. The impact of ESS integrated with
PV source on feeder voltages has been investigated in detail. After that, the VRB
148
model has been applied to simulate the characteristics of battery packs in ESS and
real-time SoC of each ESS is estimated for the control system. A coordinated control
method which consists of distributed control and localized control has been proposed
for voltage regulation while effectively utilizing storage capacities of ESSs during
daily operation. With consensus algorithm and communication links between each
ESS, all available ESSs have been coordinated together to deal with the voltage
rise/drop issues in the system. Based on the utilization ratio and availability ratio for
each sampling interval, power outputs of each ESS have been determined.
The simulation platform in Matlab/Simulink has been built to study the performance
of the proposed coordinated control method for distributed ESSs. The proposed
control method has been tested in a 6-bus distribution feeder with different types of
PV and load profiles. The simulation results have demonstrated that the proposed
control method ensures the voltages in the distribution networks within the voltage
limits during the daily operation. Based on the SoC condition of each ESS, the power
sharing among each ESS for voltage regulation is automatically determined.
Therefore, the SoC of each ESS is regulated within the desired SoC range during the
daily operation. The advantage of the proposed method has been interpreted through
comparison with droop based curtailment method. In addition, the proposed method
has also been implemented in a 13-bus distribution network with random located
ESSs. The results have further demonstrated the effectiveness of this method to solve
voltage regulation problems in LV distribution networks.
To extend of the research work in Chapter 3, distributed EVCs have been proposed
to regulate the voltages in LV distribution networks with high penetration of PVs and
plug-in EVs in Chapter 4. To characterize the behaviors of storage devices on EVs,
a lithium-ion battery model based on data fitting has been developed to estimate SoC
for the proposed control method. Proper assumptions have been made to clarify the
149
control objective and constraints of the distributed EVCs and their plug-in EVs. A
two-stage voltage control method has been proposed to coordinate the power outputs
of distributed EVCs during daily operation. Based on the utilization ratio generated
by distributed voltage control, the power allocator of each EVC determines the real
and reactive power outputs to realize voltage regulation. Consensus algorithm has
been used to design the distributed voltage control for both critical and non-critical
nodes. The localized control determines the mode of operation of the plug-in EV as
well as the real and reactive power outputs of each EVC.
A 17-bus distribution network has been implemented in Matlab/Simulink to validate

the performance of the proposed method. Simulations of two test scenarios have been
conducted under two different types of PV and load profiles. The voltage profiles
with no voltage regulation and uncoordinated charging have shown that there are
obvious voltage rise in the midday and voltage drop in the evening, while the voltage
profiles with the proposed control method have demonstrated that the voltages along
the distribution network are regulated within allowable voltage ranges. Besides, the
simulation results have also shown that the proposed voltage control method can
determine the real and reactive power outputs of each EVC based on utilization ratio
of the EVC as well as the SoC and departure time of the plug-in EV automatically.
In Chapter 5, a hierarchical control scheme has been designed to achieve power

balancing and power management in islanded microgrids. The primary control is
performed by a novel MPC, providing fast dynamic response and small steady-state
errors. Depending on different cost function, the primary MPC controller can operate
in VCM to control the output voltage and in CCM to control the output current. The
secondary control is initiated when the primary control is not able to deal with the
power imbalance in the microgrid. Then the distributed control signal of each agent
is updated autonomously through the signals received from its neighbors. In addition,
150
the allocation of the distributed control signal has been properly designed to
effectively coordinate each available unit in the microgrid. Proper actions such as
PV curtailment and load shedding should be conducted to ensure the safety operation
of the microgrid.
The simulation platform has been implemented in Matlab/Simulink to validate the

performance of the hierarchical control scheme under different test scenarios. The
results of first test case have demonstrated that the proposed control scheme can deal
with short-term power variation properly. The results of secondary test case have
shown that PV generation is curtailed when the PV generation is larger than the load
demand for a long time and the SoCs of ESSs are very high. Similarly, the loads in
the microgrid have been shed according to their priority when the load demand is
larger than the PV generation for a long time and the SoCs of ESSs are very low. The
simulation results have validated the effectiveness of the proposed control scheme
to deal with both short-term and long-term power imbalance in the microgrid. With
the proposed control scheme, both power balancing and SoC management of the
microgrid can be achieved hierarchically.
In summary, the previous proposed control methods and specific findings have been
verified by favorable research results in this thesis. The simulation results have
demonstrated that these control methods are effective and robust to coordinate ESSs
for various applications in smart grids.
6.2 Future Work
With the rapid development of advanced control and communication technologies,

the future smart grid will process different characteristics. There are many problems
and challenges to be explored in this research field. The design of advanced control
151
systems requires extensive research on various aspects of the stability, reliability and
power quality issues associated with the integration of various DERs. In this thesis,
several control techniques for application of ESSs in smart grid have been proposed.
But there are many assumptions and limitations of the modelling and control
methods in this thesis, which still need to be considered and investigated in future
research. The following are some recommendations for future research and
development of control systems for ESSs in smart grid.
1. Although actual PV and load data have been used to validate the performance of
the proposed control methods in simulation platform. The proposed control methods
for ESSs still need to be further validated through experimental studies. The results
obtained and the analysis carried out in this thesis serve as a preliminary step towards
the design of control systems for future hardware implementation. The power and
capacity levels of the distributed networks in this research can be scaled down and
implemented in laboratory test bed. After that, real-time test of the proposed methods
for voltage regulation could be conducted in existing power distribution networks.
2. This thesis mainly focuses on LV distribution networks with tens of buses.

However, a real distribution network may contain thousands of buses with various
electrical devices and different topologies. To deal with this problem, the network
partition methods are to be investigated in order to realize zonal voltage control in a
short control response time with the minimum amount of active and reactive power
compensation. The voltage levels of distribution networks can also be classified into
medium-voltage (MV) and low-voltage (e.g., 22kV/6.6kV for MV, 400V for LV in
Singapore). In LV distribution networks, real power has larger impact than reactive
power on feeder voltages. But in MV distribution networks, reactive power
compensation is more effective for voltage regulation. The coordination of ESSs
together with PVs, EVs and other voltage regulation devices (e.g., transformers and
152
switched capacitors) for different voltage levels is another important problem to be

explored. In addition, the PVs, loads, EVs and ESSs in one residential feeder should
be aggregated together for analysis and control in MV distribution networks. With
aggregated PVs and loads for each LV feeder, the voltage profiles and compensation
power required in MV distribution networks can be calculated and analyzed. Then
the aggregated ESSs and EVs with the control methods proposed in this research can
be applied for voltage regulation in MV distribution networks.
3. Although voltage regulation for LV distribution networks has been successfully

achieved in this research, the voltage fluctuation issues still exist within the voltage
limits. These voltage fluctuations are mainly caused by the intermittency of
renewable energy generation. When a large number of PV panels are connected to
the distribution networks, sudden changes in solar irradiance and temperature will
cause the PV power outputs to vary which consequently result in voltage level
fluctuations at the connecting buses. Therefore, the PV intermittency problem needs
to be further studied for improving the power quality and voltage profiles in the
distribution networks as well as increasing the efficiency of PV generation. The
ramp-rate control method is suggested for reducing the ramp-rate of PV systems to
a desired level, which is realized by controlling the power outputs of distributed
ESSs. The ramp rates are typically specified by the utility operators in kilowatts per
second (kW/s), and are common feature of new solar PV installation agreements
between utility operators and independent power producers. Another issue which
should be taken into consideration is the response time of ESSs for this application.
The ramp-rate control requires ESSs provide large amount of power in few seconds.
Hence, hybrid ESSs which contain batteries and ultracapacitors is very suitable for
this application, where ultracapacitors are responsible for short-term PV
intermittency with high ramp-rate while batteries are responsible for long-term PV
153
intermittency with low ramp-rate. The control algorithms for power allocation in the
hybrid ESSs should be investigated in the future research as well.
4. The main focus of this thesis is on the voltage regulation issue, while frequency
regulation using ESSs is another important issue to be explored. In power systems,
frequency regulation reserves are usually categorized in three types: primary reserve
(frequency-response reserve), secondary reserve (spinning reserve) and tertiary
reserve (starting up backup generators). The primary reserve needs to be online
within 8 seconds for at least 30 seconds. The secondary reserve should be online
within 30 seconds for at least 10 minutes. The tertiary reserve should be online within
10 minutes for at least 30 minutes. Because of the fast response time of ESSs, they
can be participated in primary reserve. Transfer functions based frequency regulation
model needs to be built to analyze the impacts of ESSs to system frequency. The
control algorithms to coordinate the operation of distributed ESSs are to be
developed for frequency regulation purpose. In addition, because of the limited
capacity of ESSs, the SoCs of ESSs should also be taken into account for control
system design.
5. Last but not the least, various energy management functions such as load leveling,
economic dispatch and optimal power flow can be incorporated into the control
design considering the time-of-use (TOU) of electricity. This thesis mainly focuses
on the power quality improvement of distribution networks. The future research
could consider more on energy management and economic operation of ESSs in
power systems. The cost functions to real-time optimize the operation of ESSs based
on renewable generation forecasting, load demand forecasting, TOU of electricity
and other factors should be developed to maximize the economic potential of ESSs.
154
List of Publications
Journal Publications
[1] Y. Wang, K. T. Tan, and P. L. So, “Coordinated control of distributed energy

storage systems for voltage regulation in distribution networks,” IEEE Trans.
Power Delivery, vol. 31, no. 3, pp. 1132-1141, Jun. 2016.
[2] Y. Wang and P. L. So, “Two-stage coordination of distributed electrical

vehicle chargers for voltage regulation in distribution networks,” submitted to
IEEE Trans. Smart Grid. (Under Review)
[3] Y. Wang, T. John, and P. L. So, “Hierarchical control and power management
of islanded microgrids with energy storage systems and renewable energy
sources,” submitted to IEEE Trans. Sustainable Energy. (Under Review)
[4] T. John, Y. Wang, P. L. So, and K. T. Tan, “A series-shunt network device for
an interconnected microgrid system,” submitted to IEEE Trans. Energy
Conversion. (Under Review)
Conference Publications
[5] Y. Wang, K. T. Tan, and P. L. So, “Coordinated control of battery energy

storage system in a microgrid,” in Proc. IEEE PES Asia-Pacific Power and
Energy Engineering Conference, Hong Kong, Dec. 2013.
[6] Y. Wang, B. F. Wang, and P. L. So, “A voltage regulation method using

distributed energy storage systems in LV distribution networks,” in Proc.
IEEE International Energy Conference, Leuven, Apr. 2016.
[7] T. John, Y. Wang, K. T. Tan, and P. L. So, “Model predictive control of

distributed generation inverter in a microgrid,” in Proc. IEEE Asia Innovative
Smart Grid Technology Conference, Kuala Lumpur, May 2014, pp. 657-662.
155
[8] T. John, Y. Wang, P. L. So, and K. T. Tan, “Coordinated operation of a

microgrid with a distribution network device,” in Proc. IEEE Asia Innovative
Smart Grid Technology Conference, Bangkok, Nov. 2015.
156

Amended Thesis Wang Yu - Fianl

Uploaded by

Copyright:

Available Formats

Amended Thesis Wang Yu - Fianl

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Amended Thesis Wang Yu - Fianl

Uploaded by

Copyright:

Available Formats

This document is downloaded from DR‑NTU (https://dr.ntu.edu.

Application and control of energy storage systems

Downloaded on 08 Feb 2024 17:20:30 SGT

INTERDISCIPLINARY GRADUATE SCHOOL

Interdisciplinary Graduate School

A thesis submitted to the Nanyang Technological

Finally, the application of distributed ESSs in an islanded ac microgrid is

First, I wish to express my utmost gratitude to my supervisor, Associate Professor

Furthermore, I am grateful to Nanyang Technological University for providing me

Acknowledgements ................................................................................................ iii

Table of Contents .................................................................................................... v

Table Captions ........................................................................................................ ix

Figure Captions ...................................................................................................... xi

Chapter 1 Introduction ...................................................................................... 1

Chapter 2 Literature Review .......................................................................... 19

2.2 Voltage Regulation Methods in Distribution Networks ........................... 31

Chapter 3 Coordinated Control of Distributed Energy Storage Systems for

Chapter 4 Two-stage Coordination of Distributed Electrical Vehicle Chargers

4.2.2 Assumptions ............................................................................................. 99

Chapter 5 Hierarchical Control and Power Management of Islanded

References ............................................................................................................. 145

Chapter 6 Conclusions and Future Work .................................................. 147

List of Publications ............................................................................................. 155

Table 2.1 Characteristics of energy storage technologies [3] ........................... 22

RES Renewable Energy Source

This chapter presents the background of this research work, which

1.1 Background and Motivations

Energy storage is a relative new phenomena following in the wake of increased

Figure 1.2 Classification of electrical energy storage technologies according to energy

Another important aspect related to the adoption of energy storage technologies is

1.2 Objectives and Scope

3. The third objective is to investigate the application of ESSs in an islanded

1.3 Contributions of Thesis

The main contributions of this thesis are summarized as follows:

of ESSs and various topologies of distribution networks still needs to be done.

the behaviors of battery storage devices on EVs, a lithium-ion battery model is

4. Finally, a hierarchical control scheme is proposed for islanded ac microgrids

1.4 Organization of Thesis

The thesis is organized into six main chapters as follows:

This chapter provides background information of energy storage technologies

Chapter 2: Literature Review

In this chapter, a comprehensive review of the related research work on the

Chapter 3: Coordinated Control of Distributed Energy Storage Systems for Voltage

Chapter 4: Two-stage Coordination of Distributed Electrical Vehicle Chargers for

A two-stage coordination method of distributed EVCs is proposed to regulate the

Chapter 5: Hierarchical Control and Power Management of Islanded Microgrids

the microgrid. The secondary control is performed by consensus based distributed

Chapter 6: Conclusions and Future Work

An overall conclusion of this research is discussed in this chapter. Moreover, areas

[15] S. Vazquez, S. M. Lukic, E. Galvan, L. G. Franquelo, and J. M. Carrasco,

[24] H. Kakigano, Y. Miura, and T. Ise, “Distribution voltage control for DC

[32] W. Kempton and J. Tomic, “Vehicle-to-grid power implementation: From

In this chapter, a comprehensive review of the related research works on

2.1 Energy Storage Systems

Energy storage refers to a process of converting electrical energy from power

2.1.1 Comparison of Energy Storage Technologies

The technical characteristics of various energy storage technologies are summarized

Table 2.1 Characteristics of energy storage technologies [3]

Energy Power Energy