Power System Operation Optimization Cooperate Wind Power … · Power System Operation Optimization...
Transcript of Power System Operation Optimization Cooperate Wind Power … · Power System Operation Optimization...
Power System Operation Optimization Cooperate Wind
Power with Energy Storage System Considering Emission Problem
By
Yang Zhang
This thesis is presented for degree of Doctor of Philosophy of
The University of Western Australia
School of Electrical, Electronic and Computer Engineering
2015
1
Declaration
I hereby declare that this thesis is composed of my original work and that, to
the best of my knowledge and belief, it reproduces no material previously
published or written nor material which has been accepted for the award of
any other degree or diploma, except where due acknowledgment has been
made in the text.
I have clearly stated the contribution of others to my thesis as a whole,
including statistical assistance, survey design, data analysis, significant
technical procedures, professional editorial advice, and any other original
research work used or reported in my thesis. I have clearly stated which
parts of my thesis, if any, have been submitted to qualify for another award.
I acknowledge that copyright of all material contained in my thesis resides
with the respective copyright holder(s).
_______________ (Signed)
Yang Zhang (Please Print)
2
This page is intentionally left blank
3
Acknowledgements
I would like to express my sincere gratitude to my supervisors, Prof Herbert
Ho-Ching Iu and Prof Tyrone Fernando, for giving me an opportunity to
reach this goal, for their direction, support, and advice over the course of my
candidature. It would have been impossible for me to finish this thesis
without their enthusiasm, inspiration, and guidance. I also would like to give
my great appreciation to Prof Kitpo Wong for providing his excellent
guidance and constant support.
I would like to thank all my friends and fellow students at the School of
Electrical, Electronics and Computer Engineering who have helped me in
one way or another.
I would especially like to express my love and thanks my wife, Lin Miao.
This could not have happened without her.
This thesis is dedicated to my parents, Shaoxuan Zhang and Biyan Wan, and
my families. I cannot thank them enough for all they have done for me.
4
This page is intentionally left blank
5
Abstract
This thesis is devoted to the development of a new operation model and
computational frameworks for the power system economic dispatch (ED)
and unit commitment (UC) with wind power, energy storage system (ESS)
and environmental problem.
Electric power systems, which are controlled by electrical and mechanical
systems, play a fundamental role in modern society. Economic dispatch is a
crucial process in the power system operation, which aims to allocate power
generation to match load demand at minimal possible cost while satisfying
all generators and system constraints. UC is an optimization problem of
determining operational schedules for generating units in a power system
with a number of constraints. The main objective of UC is to decide the
on/off statuses of generators over the scheduling period to meet the system
load demand and reserve requirements at lowest cost. Basically, the UC
outputs are on/off statuses on an hourly basis for a given timeframe, such as
24 hours.
In today’s society global warming is becoming a matter of concern for more
and more people, especially for governments and electric power experts. As
a result, carbon tax is applied in many countries to reduce carbon emissions.
The carbon tax aims to limit the emission and then minimize greenhouse
gases (GHGs) emission cost. Here the GHGs include carbon dioxide (CO2)
and nitrous oxides (N2O). In order to reduce generating emission, a dispatch
model including wind farms and carbon tax is developed.
6
As one of the forms of renewable energy, wind energy is being widely used
throughout the entire world. However, the most serious problem many
power industry enterprisers talk about centres on the intermittency and
uncertainty of wind power. These problems make it difficult to integrate
wind power into power systems. To reduce fluctuation of the wind energy
output, a battery energy storage system (BESS) in a renewable energy
generation system is adopted. Wind farms combined with battery energy
storage can enhance system reliability, power availability and quality, and
operational efficiency.
The rapid deployment of energy storage and the increasingly significant
amount of variability introduced by new wind generation systems will be
two of the most important and interesting changes to electricity grids over
the next few decades. Energy storage is frequently described as the solution
for the variability of wind power (as well as a host of other issues), but the
relationship between storage and variable generation is subtle and complex.
Variability is not new to the electricity grid. Well established methods for
mitigating the effects of variability already exist, and energy storage
technologies are both novel and costly. Thus, while energy storage is able to
eliminate the negative effects of wind variability, it should not be assumed
that storage is the ideal complement to increasing quantities of wind
generation. Only through prudent comparison of potential wind integration
techniques will we identify the best options and determine how energy
storage contributes to the solution.
In this research work, a computational framework for ED and UC
cooperating wind power with battery storage considering carbon tax is
presented. Given the complexity of the model, a solution approach based on
hybrid particle swarm optimization (PSO) algorithms such as sequential
quadratic programming (SQP), combine particle swarm optimization (PSO)
and quantum-inspired particle swarm optimization (QPSO) are also
7
proposed. The hybrid PSO algorithms have very strong search ability and
high convergence speed. The dispatch model is tested on a standard power
system involving thermal units and wind farms integrating BESS using the
real wind speed data obtained from two meteorological stations in Tasmania,
Australia.
In summary, the research reported in this thesis provides a complex model
for power system operation combined with renewable energy and carbon
emissions, which are validated effectively for proposed power test systems.
It also covers advanced power system random and probabilistic data
analysis techniques that can provide more accurate simulation results. The
novel method gives a better emission and operation solution efficiently and
economically.
Key Words:
Power System,Wind Power, Energy Storage System, Emission, Economic
Dispatch, Unit Commitment, Optimization
8
This page is intentionally left blank
9
Publications
Journal Papers
Y. Zhang, F. Yao, H. C. Iu, T. Fernando, and K. Wong, "Sequential
quadratic programming particle swarm optimization for wind power
system operations considering emissions," Journal of Modern Power
Systems and Clean Energy, vol. 1, pp. 231-240, 2013/12/01.
Y. Zhang, F. Yao, H. H. C. Iu, T. Fernando, and H. Trinh, "Wind–
thermal systems operation optimization considering emission
problem," International Journal of Electrical Power & Energy
Systems, vol. 65, pp. 238-245, 2. 2015.
Y. Zhang, H. H. C. Iu, T. Fernando, “Operation Scheduling
Integrating Wind-Thermal Generation with Energy Storage System
Considering Carbon Emission," IEEE Systems Journal. (second
revision)
Y. Zhang, H. H. C. Iu, T. Fernando, "Cooperative Dispatch of BESS
and Wind Power Generation Considering Carbon Emission
Limitation In Australia," IEEE Transactions on Industrial
Informatics. (second revision)
10
Y. Zhang, J. K. Zhao, K. Emami, T. Fernando “Environmentally
Constraint Economic Dispatch Integrates Wind Power Using Hybrid
SQP-PSO,” Journal of Renewable and Sustainable Energy.
(submitted)
G. Y. Zhang, G. Y. Wu, Y. Zhang, X. L. Dai “A Simple Model for
Probabilistic Interval Forecasts of Wind Power Chaotic Time Series”
Acta Phys. Sin. Vol. 63, No. 13, 2014.
G. Y. Zhang, G. Y. Wu, Y. Zhang “A Wind Speed Forecasting
Method Based on EEMD and Quantum Bacterial Foraging
Optimization” Acta Energiae Solaris Sinica (accepted)
Kianoush Emami, Tyrone Fernando, Brett Nener, Hieu Trinh, Yang
Zhang, “A Functional Observer Based Fault Detection Technique
for Dynamical Systems” Journal of the Franklin Institute. (accepted)
Conference Papers
Y. Zhang, F. Yao, H. H. C. Iu, T. Fernando, and H. Trinh,
“Operation Optimization of Wind-Thermal Systems Considering
Emission Problem," IEEE Industrial Electronics Society conference,
2014.
11
Contents
Declaration ..................................................................................................... 1
Acknowledgements ........................................................................................ 3
Abstract .......................................................................................................... 5
Publications .................................................................................................... 9
Contents ........................................................................................................ 11
List of Figures .............................................................................................. 19
List of Tables ................................................................................................ 21
Chapter 1 23
Introduction .................................................................................................. 23
1.1. Background .................................................................................... 23
1.2. Motivation ...................................................................................... 25
1.3. Objectives ...................................................................................... 27
1.4. Contributions ................................................................................. 28
1.5. Thesis Outline ................................................................................ 28
Chapter 2 31
Wind Power and Energy Storage System ..................................................... 31
2.1. Introduction of Wind Energy ......................................................... 31
2.2. Wind Power Production ................................................................. 33
2.2.1. Wind Power Curve ................................................................. 33
12
2.2.1.1. Cut-in Wind Speed ......................................................... 33
2.2.1.2. Rated Wind Speed .......................................................... 34
2.2.1.3. Cut-out Wind Speed ....................................................... 35
2.2.2. Optimizing Rotor Diameter and Generator Rated Power ...... 35
2.2.3. Variable Slip Induction Generators ....................................... 37
2.2.4. Pole-Changing Induction Generators ..................................... 37
2.2.5. Rotor Efficiency ..................................................................... 37
2.2.6. Wind Power Calculation ........................................................ 37
2.2.7. Wind Power Impacting Factors ............................................. 39
2.2.7.1. Load Factor ..................................................................... 39
2.2.7.2. Seasonal and Diurnal Variation of Wind Power ............ 39
2.2.7.3. Wind Statistics ................................................................ 39
2.3. Connection Between Wind Farm and Power Grid ........................ 40
2.3.1. Wind Farms ............................................................................ 40
2.3.2. Wind Farm Distribution ......................................................... 40
2.3.3. Challenges with Wind Connection ........................................ 41
2.3.3.1. Local Impacts ................................................................. 42
2.3.3.2. Low Frequency Operation .............................................. 42
2.3.3.3. Low Power Factor .......................................................... 42
2.3.3.4. Reduce Grid Security and Reliability ............................. 43
2.4. Energy Storage System ................................................................. 43
2.4.1. Energy Storage Technologies ................................................ 46
2.4.2. Battery Energy Storage System ............................................. 51
2.4.3. Battery Types ......................................................................... 51
13
2.5. Conclusion ..................................................................................... 56
Chapter 3 59
Hybrid Power System & Data Analysis Methodologies .............................. 59
3.1. Introduction .................................................................................... 59
3.2. Hybrid Wind Power Generation Systems ...................................... 60
3.2.1. Wind Power Integration ......................................................... 60
3.2.2. Embedded Energy Storage Systems ....................................... 64
3.3. Data Analysis in Wind Power System ........................................... 65
3.3.1. Power System Operation with Wind Power ........................... 65
3.3.2. Optimization Approach .......................................................... 68
3.3.2.1. Genetic Algorithm ........................................................... 68
3.3.2.2. Immune Algorithm .......................................................... 71
3.3.2.3. Particle Swarm Optimization .......................................... 74
3.3.2.4. Comparison ..................................................................... 76
3.3.3. Advanced Techniques ............................................................ 76
3.4. Conclusion ..................................................................................... 77
Chapter 4 79
Power System Dispatch Considering Wind Energy and Emission .............. 79
4.1. Introduction .................................................................................... 79
4.2. Power System Dispatch Integrating Wind Energy with Emission
…………………………………………………………………80
4.3. Probabilistic Modelling of Wind Power for ED Modelling .......... 83
4.4. Mathematical Formulation of CEED Problem with Wind Power
………………………………………………………………... 86
14
4.4.1. Objective Function ................................................................. 86
4.4.2. System Constraints ................................................................ 88
4.5. Hybrid Optimization Algorithm .................................................... 89
4.5.1. Sequential Quadratic Programming (SQP) ............................ 89
4.5.2. Particle Swarm Optimization ................................................. 90
4.5.3. Composite Computation Approach ....................................... 91
4.6. Case Studies .................................................................................. 92
4.6.1. Case-I. ELD Model with and without Wind Farm ................ 95
4.6.2. Case-II. CEED Model with and without Wind Farm ............ 97
4.6.3. Case-III. Comparisons with Other Approaches ..................... 99
4.7. Conclusion ................................................................................... 100
Chapter 5 103
Unit Commitment with Wind Power Generation and Carbon Tax
Considered ................................................................................................. 103
5.1. Introduction ................................................................................. 103
5.2. Unit Commitment Considering Wind Power and Carbon Tax ... 103
5.3. Probabilistic Modeling of Wind Power ....................................... 106
5.4. Mathematical Formulation of CUCE Problem with Wind Power
…………………………………………………………………..107
5.4.1. Objective Function ............................................................... 107
5.4.2. System Constraints .............................................................. 110
5.5. Hybrid Optimization Algorithm .................................................. 111
5.6. Numerical Simulation ................................................................. 111
5.6.1. Parameter Sensitivity Analysis ............................................ 112
15
5.6.2. Case Studies ......................................................................... 115
5.6.3. Comparisons ......................................................................... 118
5.7. Conclusion ................................................................................... 119
Chapter 6 121
Wind-Thermal Generation Scheduling Optimization Integrating ESS with
Carbon Emission ........................................................................................ 121
6.1. Nomenclature ............................................................................... 121
6.2. Introduction .................................................................................. 122
6.3. Wind Power Forecasting and BESS ............................................ 125
6.3.1. Wind Power Prediction ........................................................ 126
6.3.2. Battery Energy Storage System (BESS) .............................. 127
6.3.2.1. Selection ........................................................................ 127
6.3.2.2. Operation ....................................................................... 128
6.4. Problem Formulation ................................................................... 129
6.4.1. Objective Function ............................................................... 129
6.5.1. System Constraints ............................................................... 131
6.5. Hybrid Optimization Algorithm .................................................. 132
6.6. Simulation Result and Discussion ............................................... 133
6.6.1. Parameter Analysis ............................................................... 133
6.6.2. Case Studies ......................................................................... 133
6.6.2.1. CUCE Result without Wind Farms and BESS ............. 133
6.6.2.2. CUCE Result with Wind Farms and no BESS ............. 134
6.6.2.3. CUCE Result with Wind Farms and BESS .................. 137
6.6.3. Comparisons ......................................................................... 141
16
6.7. Conclusion ................................................................................... 142
Chapter 7 145
BESS and Wind Power Cooperative Dispatch with Emission Limitation in
Australia ..................................................................................................... 145
7.1. Introduction ................................................................................. 145
7.2. Probability Analysis of Wind Power and Battery Energy Storage
System (BESS) ....................................................................................... 148
7.2.1. Probability of Wind ............................................................. 148
7.2.2. Battery Energy Storage System (BESS) .............................. 148
7.3. The Proposed Economic Dispatch Model ................................... 150
7.4. Quantum-Inspired Particle Swarm Optimization ........................ 153
7.4.1. Particle Swarm Optimization ............................................... 154
7.4.2. Quantum-Inspired Particle Swarm Optimization ................ 154
7.4.3. Procedure of QPSO .............................................................. 158
7.5. Case Studies and Discussion ....................................................... 159
7.5.1. Benefits of Carbon Tax ........................................................ 161
7.5.2. Integration of Wind Power .................................................. 164
7.5.3. Combined with BESS .......................................................... 165
7.6. Conclusion ................................................................................... 166
Chapter 8 167
Conclusion and Future Work ..................................................................... 167
8.1. Summary of Contributions .......................................................... 167
8.2. Suggestions for Future Work ...................................................... 171
17
8.2.1. Wind Power + Solar Power + Storage Dispatch/Unit
Commitment Considering Emission Problem ..................................... 171
8.2.2. Large-Scale Renewables and Energy Demand .................... 172
Bibliography ............................................................................................... 173
18
This page is intentionally left blank
19
List of Figures
Fig. 2.1 Wind turbine power curve ............................................................... 33
Fig. 2.2 (a) Increasing rotor diameter gives rate power at lower wind speed
...................................................................................................................... 36
Fig. 2.2 (b) Increasing the generator size increases rates power .................. 36
Fig. 2.3 Energy storage technology comparison .......................................... 50
Fig. 3.1 Schematic diagram of general isolated wind-diesel hybrid power
system ........................................................................................................... 61
Fig. 3.2 The aggregated (left) and distributed (right) ESS configurations in
wind farms .................................................................................................... 65
Fig. 3.3 Flowchart of a typical GA ............................................................... 70
Fig. 3.4 Flowchart of a typical IA ................................................................ 73
Fig. 3.5 Flowchart of a typical PSO ............................................................. 75
Fig. 4.1 Computational framework considering wind power uncertainties . 82
Fig. 4.2 Simplified wind turbine power curve .............................................. 83
Fig. 4.3 Wind speed distribution and Weibull fitting ................................... 93
Fig. 4.4 Solutions of ED models with and without wind farm ..................... 96
Fig. 4.5 Solutions of CEED models with and without wind farm ................ 98
Fig. 5.1 Wind power uncertainties for computational framework ............. 114
Fig. 5.2 Wind speed distribution and Weibull fitting ................................. 114
Fig. 5.3 Forecasted system demand ............................................................ 116
Fig. 5.4 Forecasted wind power vs scheduled wind power ........................ 117
Fig. 6.1 Structure of wind power generation system integrating BESS ..... 124
Fig. 6.2 Algorithm for producing probabilistic wind power forecast ........ 127
Fig. 6.3 Predicted and real wind power ...................................................... 136
20
Fig. 6.4 BESS operation with forecasted and actual wind power .............. 139
Fig. 6.5 Thermal units output with/without BESS operation .................... 139
Fig. 6.6 State of charge of the battery ........................................................ 140
Fig. 6.7 Wind penetration CUCE with/without BESS ............................... 141
Fig. 7.1 Structure of wind power generation system integrating BESS .... 146
Fig. 7.2 The quantum rotation gate ............................................................ 156
Fig. 7.3 Flow chart of quantum-inspired particle swarm optimization ..... 158
Fig. 7.4 Simplified 14-generator, 50 Hz system. ....................................... 159
Fig. 7.5 Carbon emission of ED models with/without wind farm (WF) & (a)
with carbon tax; (b) without carbon tax ..................................................... 162
Fig. 7.6 Carbon emission without/with different level of BESS in three
system demand ........................................................................................... 163
Fig. 7.7 Wind penetration without/with different level of BESS in three
system demand ........................................................................................... 164
21
List of Tables
Table 2.1 Offshore Wind Farms in Europe. ................................................. 41
Table 2.2 Attributes of Different Battery Types .......................................... 56
Table 3.1 Comparisons of the Algorithms ................................................... 76
Table 4.1 Wind Power Factor ....................................................................... 93
Table 4.2 Fuel Cost Coefficients .................................................................. 94
Table 4.3 Fuel Cost Coefficients .................................................................. 94
Table 4.4 Emission Price .............................................................................. 94
Table 4.5 Emission Factor of Units .............................................................. 95
Table 4.6 Forecast System Demand and Wind Farm Output ....................... 95
Table 4.7 Solution of ELD without Wind Farm ........................................... 95
Table 4.8 Solution of ELD with Wind Farm ................................................ 96
Table 4.9 Solution of CEED without Wind Farm ........................................ 97
Table 4.10 Solution of CEED with Wind Farm ........................................... 98
Table 4.11 Comparison of Different Approaches ...................................... 100
Table 5.1 Wind Power Factors ................................................................... 113
Table 5.2 Generator Parameters ................................................................. 113
Table 5.3 Generator Constraint .................................................................. 113
Table 5.4 Emission Factors of Units .......................................................... 115
Table 5.5 Emission Price ............................................................................ 115
Table 5.6 Forecasted Wind Farm Power and System Demand .................. 117
Table 5.7 Generator Schedules ................................................................... 117
Table 5.8 Solution of CUCE without Wind Farm ...................................... 118
Table 5.9 Solution of CUCE with Wind Farm ........................................... 118
Table 5.10 Comparison of Different Approaches ...................................... 119
22
Table 6.1 BESS Parameters ....................................................................... 134
Table 6.2 Load Demand and Forecasted Wind Power .............................. 134
Table 6.3 Generators Schedule of CUCE .................................................. 136
Table 6.4 Generators Schedule of CUCE with Wind Farm ....................... 137
Table 6.5 Generators Schedule of CUCE with Wind Farm and BESS ..... 140
Table 6.6 Comparison of Different Approaches ........................................ 142
Table 7.1 Generator Parameters ................................................................. 160
Table 7.2 Emission Factors ........................................................................ 160
Table 7.3 Wind Turbine Parameter ............................................................ 160
Table 7.4 Predict System and Wind Farm Output ..................................... 161
Table 7.5 BESS Parameter ......................................................................... 161
Table.7.6 Solutions of Different ED Models in Three System Demand ... 162
Table 7.7 Generation Output and Cost with Different Level BESS .......... 163
23
Chapter 1
Introduction
1.1. Background
Many countries around the world are introducing programs aimed at
reducing the emissions produced by the power plants and increasing the
utilization of renewable generation. Among different types of renewable
energy technologies, wind power is expected to be one of the most popular
types of renewable in the near future [1]. Wind energy has a number of
advantages such as no pollution, relatively low capital cost and a short
gestation period. As mentioned above wind power has many advantages,
however it also causes intermittent and volatile characteristics which may
impact on power system security and stability. High penetration of wind
power can introduce new challenges and reduce the power system
dispatchability and economic efficiency [2-7]. As a result, the decreasing
power system stability margins will lead to unacceptable operating
conditions and power system collapses. In addition, the uncontrollable
nature of wind power will lead to an additional cost of managing the
intermittency as the intermittency of wind can cause strain on the electricity
grid.
24
One way to overcome this inherent variability with wind is by using energy
storage. If there existed an efficient and cheap method at a scale of hundreds
of MWh to store the energy generated from these renewable sources, then
this source could be used on a dispatchable basis much like natural gas
plants, allowing generation to match demand. Incorporating such forms of
energy storage with wind and solar technologies would enable the large-
scale integration of renewable and non-emitting generation to the electricity
grid.
Recently, combining an energy storage system (ESS) together with a wind
power has been proposed in order to provide economic and technical
benefits to power systems [8]. Energy storage systems have been shown to
be quite suitable in mitigating the negative impacts resulting from the
integration of wind generation. In order to reduce fluctuation of the wind
energy output, a battery energy storage system (BESS) is integrated into the
renewable energy generation system [9]. BESS can also act as a means of
mitigating the intermittency of wind power.
With the increasing development of wind energy (and some development of
solar energy), the intermittent nature of wind becomes a significant problem.
Utilities have already struggled to meet fluctuating demand when they
control the output of the generators; adding generators that they cannot
control will add significant complexity to this problem. This problem is
magnified as the penetration level of wind energy on the utility system
increases. For example, it is not nearly as great of a problem as it will be
once the utility reaches 20% wind penetration. There is not an exact break
point where a solution must be found, rather, the cost per MWh to integrate
wind energy into the system increases as the penetration level increases [10]
Carbon tax policy has been implemented in many countries and will affect
the generation dispatch to some extent. The power system economic
25
dispatch pursues not only economic benefit but also many others such as
low carbon emissions and environmental protection. In order to reduce the
GHGs emissions, the combined economic emission dispatch (CEED) was
proposed, which can take into account fuel cost and emission tax together.
The aim of this research is to develop an advanced dispatch and a
mathematical model of wind power cooperating with BESS systems for
resolving the operation problems and minimizing the power system
operation and emission costs. Moreover, novel advanced and effective data
analysis techniques will be developed.
1.2. Motivation
In order to solve the problems of today’s power system operation, it is
necessary to use new techniques of numerical analysis, control
methodologies and equipment modelling to improve the operation
efficiency, and minimize the wind power and BESS operation and emission
cost.
Recently, it is a trend to use alternative energy resources to thermal energy
power generation. Wind and solar energy are the most popular alternative
choices. One of the major benefits of the renewable resources is that there is
no extra cost in the production of power after the initial land and capital cost
and maintenance cost. In this research work, the wind speed will be
assumed to follow the Weibull distribution. Wind power is stochastic in
nature and errors will always exist in wind power forecasts. Therefore
analysis of the uncertainty of wind power by probabilistic methods is a key
part in addressing the problem.
26
From a fundamental perspective, the main disadvantage of wind energy is
the intermittent nature of wind. Wind is extremely unreliable, equally likely
to blow as to not blow during peak hours. Although daily and seasonal wind
patterns exist, wind is still highly unpredictable when it comes to timing and
strength. The intermittency and uncertainty of wind makes the dispatch of
wind energy a difficult task [11]. High wind power penetration could impact
the system security and reliability of power grid [12]. To reduce fluctuation
of the wind energy output, a battery energy storage system (BESS) is
integrated into a renewable energy generation system [9]. Wind farms
combined with battery energy storage can enhance system reliability, power
availability and quality, and operational efficiency.
Other techniques used to improve the wind power system operation are
economic dispatch (ED) and unit commitment (UC). Economic dispatch
deals with the minimum cost of power production in electrical power system
analysis [13]. The main task of ED is to try to find the optimal allocation of
the electrical power output from various available generators. Normally, the
ED problem includes two or three power generators, and only one is an
exhausted resource such as fuels. UC is an optimization problem of
determining operational schedules for generating units in a power system
with a number of constraints [14]. The main objective of UC is to decide the
on/off statuses of generators over the scheduling period to meet the system
load demand and reserve requirements at lowest cost.
Furthermore, previous research works did not consider the emission issue
and wind power prediction overestimation/underestimation situations, which
are critical in the implementation of a wind power system. Here, complete
optimization-based economic dispatch models with wind power BESS and
emission problems are presented. Both wind turbines and conventional
generators are taken into account in the power generation.
27
There are many research works about optimizing the power system
operation. An approach to evaluate the contribution that wind power can
make to the load carrying capability of a power generating system in an
operating scenario was studied in [15]. A novel UC formulation for a power
system with significant levels of wind generation was proposed in [16]. In
[17], the authors proposed an approach to evaluate the uncertainties of the
balancing capacity, ramping capability, and ramp duration requirements.
Furthermore, various numerical optimization methods, i.e., genetic
algorithm (GA) [18], [19], evolutionary programming (EP) [20], the
quantum-inspired evolutionary algorithms (QEA) [21], simulated annealing
(SA) [22], artificial neural networks (ANN) [23-25] and particle swarm
optimization (PSO) [26], have been employed to solve the UC problems.
Due to the complex power dispatch model, coordinately dispatched
traditional generation sources with wind power and BESS while satisfying
all the determined and probabilistic constraints becomes more complicated.
One of the consequences is that more advanced and reliable computation
approaches are required.
1.3. Objectives
The main objectives of the research that reported in this thesis are following:
Develop a novel framework for power system economic dispatch
operation optimization with renewable energy and emission.
Develop a novel framework for power system unit commitment
operation optimization with renewable energy and emission.
Design a novel approach to optimizing power system operation
combining renewable energy and energy storage system.
28
Implement a hybrid operation method to renewable energy and
energy storage system with carbon tax.
Develop hybrid algorithms for the power system operation.
1.4. Contributions
The main contributions of the research reported in this thesis are set out
below:
A novel framework for power system economic dispatch operation
optimization with renewable energy and emission.
A novel framework for power system unit commitment operation
optimization with renewable energy and emission.
A hybrid power system economic dispatch model for renewable
energy and energy storage system with carbon tax.
A hybrid power system unit commitment model for renewable
energy and energy storage system with carbon tax.
A new Sequential Quadratic Programming combined with Particle
Swarm Optimization that is to be used for solving economic
dispatch/unit commitment problems.
A new Quantum Particle Swarm Optimization that is to be used for
solving economic dispatch problems.
1.5. Thesis Outline
This thesis is organised as follows:
29
Chapter 1 presents the objectives and background of this thesis along with
an introduction to power system operation. This includes an introduction to
wind energy and storage systems.
Chapter 2 focuses on the wind power and storage energy system
fundamental knowledge. Wind energy source theory and wind power
conversion is introduced. Storage system conception and application also
have been discussed.
Chapter 3 describes the hybrid power system which includes renewable
generation. A comprehensive review of power system data analysis
methodologies for wind power integration and operation is provided.
Chapter 4 looks at power system dispatch considering wind energy and
emission.
Chapter 5 discusses the power system unit commitment problem with wind
power generation and carbon tax considered.
Chapter 6 examines a novel wind-thermal scheduling model integrating
energy storage system.
Chapter 7 is concerned with the power dispatch problem cooperate wind
power and battery storage in Australia.
Chapter 8 concludes the thesis and suggests the future research direction.
30
This page is intentionally left blank
31
Chapter 2
Wind Power and Energy Storage System
2.1. Introduction of Wind Energy
There has been a rapid growth of renewable energy sources in power
systems due to environmental concerns in energy generation from
conventional sources. Growing demand for electrical energy and concerns
associated with limited reserves of fossil fuels such as coal, oil, and natural
gas are also responsible for the development and increase in renewable
energy utilization.
Wind energy is one of the fastest growing renewable energy sources. A total
of 120,791 MW of wind capacity has been installed throughout the world
[27] by the year 2008. The cost of energy from wind has dropped to the
point that in some sites it is nearly competitive with conventional sources.
The current total installed wind capacity in Canada is 2,577 MW, which is
about 1 % of Canada’s total electricity demand [28]. The city of
Saskatchewan currently has 171.2 MW of installed wind capacity, with the
completion of the 150 MW centennial wind project in 2006 [28]. The World
Energy Council has estimated that wind energy capacity worldwide may
total as high as 474,000 MW by the year 2020 [29]. In Canada; Ontario,
Nova Scotia and Prince Edward Island have committed to generate 10%, 5%
32
and 15% respectively of the total electricity production from renewable
energy sources by the year 2010 [28]. Many countries around the world are
implementing different policies to promote the growth of renewable energy.
The Renewable Portfolio Standard (RPS) is a policy that requires those who
sell electricity to have a certain percentage of renewable power in their mix
[30]. In the USA, 13 states have written the RPS into state law to increase
the percentage of renewable power to 10%-20% before the year 2010.
Renewable energy policies, such as the Fixed Feed-in-Tariffs in Germany,
Denmark, and Spain [31], and Renewable Obligation in the UK [32], have
driven the development of wind power in these countries.
The first wind turbine for electricity generation was developed at the end of
the 19th century. From 1940 to 1950, two important technologies, three
blades structure of wind turbine and the AC generator which replaced DC
generator were developed [33]. Wind power is one of the renewable energy
sources which have been widely developed in recent years. Wind energy has
many advantages such as no emission pollution, relatively low capital cost
and a short gestation period. According to the Global Wind Energy Council
(GWEC), global wind power capacity has increased from 5400 MW at the
end of 1995 to 223 GW by 2012. However wind power still accounts for
less than 1.5% of the world’s electrical demand. It is inferred that wind
energy will develop to about 15% of the world’s electrical supply by 2025
[34]. During the period of 1973 to 1979, research into wind generation
increased as a result of the oil crisis. By the end of 2000, wind power has an
important role in sustainable energy development. At the same time, wind
turbine technologies were developed throughout the world, especially in
Denmark, Germany, and Spain. Today, wind energy is the fastest growing
energy source.
A lot of developments have been taken place in the design of wind energy
conversion systems (WECS). Modern wind turbines are highly sophisticated
machin
industry
to deliv
propert
2.2.
2.2.1.
The po
power
electric
2.2.1.1.
When t
the win
sufficie
generat
nes built on
y, incorpora
ver energy
ty will be di
Wind P
Wind Po
ower curve i
curve show
cal output.
. Cu
the wind sp
nd turbines
ent to overc
tor is not ab
the aerodyn
ating advan
across a w
iscussed in t
ower Pr
ower Curv
is an import
ws the rela
Fig. 2.1
t-in Wind S
peed is belo
cannot star
ome friction
ble to genera
33
namic princ
nced materia
wide-range
the followin
roduction
ve
tant item fo
ationship b
1 Wind turbin
Speed
ow cut-in w
rt [35, 36].
n in the driv
ate any usef
ciples devel
als and elec
of wind sp
ng section.
n
or a specific
between win
ne power curve
wind speed
Power in th
ve train of t
ful power b
loped from
ctronics and
peeds. The
c wind turbi
nd speed a
e.
(VC) show
he low spee
he turbine.
elow cut in
the aerospa
d are design
e wind pow
ine. The wi
and generat
wn in Fig. 2
ed wind is n
Therefore t
speed.
ace
ned
wer
nd
tor
.1,
not
the
34
2.2.1.2. Rated Wind Speed
It can be seen from Fig. 2.1 that as the wind speed increases, the power
delivered by the generator will increase as the cube of wind speed. When
the wind speed reached the rated wind speed (VR), the generator can deliver
the rated power. If the wind speed exceeds the rated wind speed, there must
be some methods to control the wind power supply to the generator or else
the generator may be damaged. Basically, there are three control approaches
for large wind power machines: active pitch-control, passive stall-control,
and the combination of the two.
In pitch-control system, an electronic system monitors the generator output
power. If the power exceeds the rated power, the pitch of the turbine blades
will adjust to shed some wind. The electronic system will control a
hydraulic system to slowly rotate the blades about the axes, and turn them a
few degrees to reduce the wind power. In conclusion, this strategy is to
reduce the blade’s angle of attack when the wind speed exceeds the rated
wind speed.
For the stall-controlled machines, the turbine blades can reduce the
efficiency automatically when the winds exceed the rated speed. In this
control method, it has moving parts to increase the angle but it happens
automatically with wind speed. The majority of modern, large wind turbines
use this passive, stall-controlled approach.
For the large (above 1.0 MW) size wind turbines, when the wind speed
exceed the rated wind speed, the turbine machine will not reduce the angle
of attack but increase it to induce stall.
For the small size wind turbines, there are a variety of techniques to spill
wind. The common way is the passive yaw control that can cause the axis of
the turbine to move more and more out of the wind. Another way relies on a
35
wind vane mounted parallel to the plane of the blades. As winds get stronger,
wind pressure on the vane rotates the machine away from the wind.
From Fig. 2.1 we can see that there is no power generated at wind speeds
below VC, the output is equal to the rated power of the generator at wind
speeds between VR and VF, above VF the turbine is shut down [35, 36].
2.2.1.3. Cut-out Wind Speed
Sometimes, the wind is too strong to damage the wind turbine. In Fig. 2.1
this wind speed is called cut-out or the furling wind speed (VF). Above VF,
the output power is zero. In terms of active pitch-controlled and passive
stall-controlled machines, the rotor can be stopped by rotating the blades
about their longitudinal axis to create a stall. However, for the stall-
controlled machines, there will be the spring-loaded on the large turbine and
rotating tips on the ends of the blades. When it is necessary, the hydraulic
system will trip the spring and blade tips rotate 90◦ out of the wind and stop
the turbine.
2.2.2. Optimizing Rotor Diameter and Generator Rated
Power
Fig. 2.2 [74] shows the trade-offs between rotor diameter and generator size
as methods to increase the energy delivered by a wind turbine. As shown in
Fig. 2.2(a), increasing the rotor diameter and maintaining the same
generator will shift the power curve upward. In this situation, the turbine
generator can get the rated power at a lower wind speed. The result for
keeping the same rotor but increasing the generator size is presented below
in Fig. 2.2 (b).
F
Fig. 2.2
Fig. 2.2 (b) In
2 (a) Increasin
ncreasing the g
ng rotor diame
generator size
36
eter gives rate
increases rate
power at low
es power (win
er wind speed
nd turbine size
d.
e same).
37
2.2.3. Variable Slip Induction Generators
It is known that the speed of a normal induction generator is around 1% of
the synchronous speed. The slip in the generator is a function of the dc
resistance in the rotor conductors [37]. If we add a variable resistance to the
rotor, then the slip can range up to about 10% [37].
2.2.4. Pole-Changing Induction Generators
In terms of the induction generator, the rotor spins at a frequency which is
largely controlled by the number of poles. If it is possible for us to change
the number of poles, we can make the wind turbine spin at different
operating speeds. The stator can have external connections that switch the
number of poles from one value to another without change in the rotor.
2.2.5. Rotor Efficiency
For a given wind speed, the rotor efficiency is a function of rotor turning
rate. If the rotor turns too slowly, the efficiency drops off because the blades
are letting too much wind pass by unaffected. However, if the rotor turns too
fast, efficiency will reduce as the turbulence caused by one blade
increasingly affects the blade that follows. The tip-speed ratio (TSR) is a
function which can illustrate the rotor efficiency. The definition of the tip-
speed-ratio is:
TSR = rotor tip speed/wind speed = (πdN)/60v (2.1)
where N is rotor speed in rpm, d is the rotor diameter (m); and v is the wind
speed (m/s) upwind of the turbine.
2.2.6. Wind Power Calculation
The total power available in wind is equal to the product of mass flow rate
of wind mw, and V2/2. Assuming constant area or ducted flow, the continuity
38
equation states that mw=ρAV, where ρ is the density of air in kg/m3, A is the
blades area in m2, and V is velocity in m/s.
Thus, the total wind power,
2 3( ) / 2 ( ) / 2w wm V AVP (2.2)
Here, the ρ is a function of pressure, temperature and relative humidity. Let
us assume the inlet wind velocity is Vi and the output velocity is Vo, then the
average velocity is (Vi +Vo)/2.
The wind power recovered from the wind is given as
2 2 2 2
2 3
/ 2 / 4
/
( )
2 1
wout i o i o i o
w
P m V V A V V V V
P x x x
(2.3)
where x= Vo/ Vi. Differentiating Eq. (2.3) with respect to x and setting it to
zero gives the optimum value of x for maximum power output
2) /( 0 1 2 3out dxd P x x (2.4)
and then we can get .xmax p=1/3.
Substituting the value of xmax p in Eq. (2.3), the maximum power recovered
is
max 16 / 27 0.593w wout P PP (2.5)
It can be found that the maximum power from a wind system is 59.3% of
the total wind power.
The electrical power output is,
e p m g wC PP (2.6)
39
where Cp is the efficiency coefficient of performance when the wind is
converted to mechanical power. ηm is mechanical transmission efficiency
and ηg is the electrical transmission efficiency [38]. The optimistic values
for these coefficients are Cp=0.45, ηm=0.95 and ηg=0.9, which give an
overall efficiency of 38% [38]. For a given system, Pw and Pe will vary with
wind speed.
2.2.7. Wind Power Impacting Factors
2.2.7.1. Load Factor
There are two main objectives in wind turbine design. The first is to
maximize the average power output. The second one is to meet the
necessary load factor requirement of the load. Load factor is very important
when the generator is pumping irrigation water in asynchronous mode [39].
Commonly assumed long-term average load factors may be anywhere from
25% to 30%.
2.2.7.2. Seasonal and Diurnal Variation of Wind Power
It is clear that the seasonal and diurnal variations have significant effects on
wind. The diurnal variation can be reduced by increasing the height of wind
power generator tower. In the early morning, the average power is about 80 %
of the long term annual average power. On the other hand, in early
afternoon hours, the average power can be 120% of the long term average
power [75].
2.2.7.3. Wind Statistics
Wind resource is a highly variable power source, and there are several
methods of characterizing this variability. The most common method is the
power duration curve [40]. Another method is to use a statistical
representation, particularly a Weibull distribution function [41]. Long term
40
wind records are used to select the rated wind speed for wind electric
generators. The wind is characterized by a Weibull density function.
2.3. Connection between Wind Farm and Power
Grid
2.3.1. Wind Farms
Nowadays, a single wind turbine is just used for a particular site, such as an
off-grid home in rural or off-shore areas. On a good windy site, normally
there will be lots of wind turbines which are often called wind farms or
wind parks. The advantages of wind farms are reduced site development
costs, simplified connections to transmission lines, and more centralized
access for operation and maintenance.
The numbers of wind turbines can be installed at a wind site. If the wind
turbines are located too close, it will result in upwind turbine interfering
with the wind received by those located downwind. However, if the wind
turbines are located too far, it means site space is not properly utilized.
When the wind passes the turbine rotor, the energy will be extracted by the
rotor and the power which is available to the downwind machines will be
reduced. Recent studies show that the wind turbines performance will
degrade when the wind turbines are too close to each other [37].
2.3.2. Wind Farm Distribution
In Europe, offshore projects are now springing up off the coasts of Belgium,
Denmark, France, Germany, Irelands, Netherlands, UK, Sweden and
Scotland. The total offshore wind farm installed capacity in 2009 has
41
reached 2055 MW. Table 2 shows offshore wind farms in Europe that have
capacities over 100 MW up to the year 2009 [42].
Table 2.1
Offshore Wind Farms in Europe
Country Project name Capacity
(MW)
Number of
turbines
Wind turbine
manufacturer
Netherlands Egmond Aan zee 108 38 Vestas
Denmark Nysted 165.6 72 Siemens
Sweden Lillgund 110.4 48 Siemens
Denmark Horns Rev 1 160 80 Vestas
Netherlands Prinses Amaila 120 60 Vestas
Gunfleet Clacton-on Sear 104.4 29 Siemens
Denmark Horns Rev 2 209 91 Siemens
2.3.3. Challenges with Wind Connection
The variability and limited predictability of wind power have raised
concerns about the impacts on power system reliability and cost. The
impacts of wind power on power systems can roughly be divided into local
impacts and system-wide aspects [43], taking into account both the
electrical aspects of wind turbines and the characteristics of the wind.
Furthermore, the connection of wind power challenges the planning and
operation of the grid. Another aspect is the formulation of grid-code
requirements especially for wind power. Last, the design of electricity
markets also has consequences for the system integration of wind power.
All of these aspects are discussed below.
42
2.3.3.1. Local Impacts
The integration of small-scale wind power mostly involves the connection
of individual wind turbines to distribution grids. The local impacts of wind
power therefore mainly depend on local grid conditions, connected wind-
turbine type. The effects become less noticeable with the (electrical)
distance from the source. The observed phenomena include changed branch
flows, altered voltage levels, increased fault currents and the risk of
electrical islanding, which complicate system protection, and possibly
power quality problems, such as harmonics and flicker [44]. Modern wind
turbines are equipped with versatile power electronics and can be designed
to mitigate some of these problems. The rest must be captured by strict grid
requirements and new designs for the distribution networks.
2.3.3.2. Low Frequency Operation
There is no doubt that the low frequency operation of the wind generation
will affect the output power. Normally, when the frequency is less than 48
Hz, many wind power generations do not integrate with grid. The power
output loss could be around 5–10% on account of low frequency operation
[45].
2.3.3.3. Low Power Factor
A synchronous generator can supply both active and reactive power.
However, reactive power is needed by the wind power generation with
induction generator for the magnetization. With respect to wind power
generation with induction generators, instead of supplying reactive power to
the grid, they will absorb reactive power from grid. As a result, suitable
reactive power compensation devices are required to supply the reactive
power to the wind generator/grid [46, 47].
43
2.3.3.4. Reduce Grid Security and Reliability
The poor grid stability may cause 10–20% power loss [45], and this
deficiency may be the main reason for low energy output of wind power
generation.
In China, many wind farms are not connected to the power grid because of
the stability issues and difficulties in dispatching by the system operators.
Major wind power research is being conducted in the aspects of dispatch
issues and long distance transmission issues.
In the Australian National Electricity Market (NEM), before the connection
of a wind farm to a power grid, the (wind) generation service provider must
conduct connectivity studies by itself and/or with the transmission network
service provider for which the wind farm is to be connected [45]. The
connectivity study needs to check if the proposed wind generator can be
hosted by the existing power grid in view of stability as well as reliability
aspects. Depending on the study results conducted by the transmission
network service provider, the cost associated and the suitability of the
connection point of the proposed wind farm will be given for the generation
company to make further decisions regarding its investment.
2.4. Energy Storage System
There are many methods that have been proposed to mitigate the variability
of these stochastic sources, and one of the most promising is that of energy
storage. Energy storage of various kinds has been around for numerous
years. Since the advent of electrical power, energy has typically been stored
in the form of fuel, and then converted to electricity as needed. By far the
most common form of this today is pumped hydro, in which water is
pumped up an elevation to a reservoir during times of low power demand,
44
and allowed to flow back downhill through hydroelectric generators during
time of high power demand. This method, while relatively efficient,
affordable, and scalable, has the same fundamental limiting factor as
hydroelectric generation at large which is that it is not viable in places
without both an abundant supply of water and a high topographical gradient.
Among the other forms of energy storage, including flywheels,
supercapacitors, and compressed air storage, electrochemical energy storage
in the form of batteries has the most promise for mitigating short to long-
term variability in wind and solar power generation. The reason batteries
show the most promise is because of their high level of efficiency, low
response time, and dispatchability. The cost and limited service life of
batteries continues to restrict their mainstream implementation, but these
factors are changing because of the lowering costs of battery technology
from new chemical compositions, as well as evolutionary improvements in
existing compositions. As numerous civil bodies establish goals for the
percentage of their energy generated from renewables, the need grows to
alleviate concerns of the reliability of these sources. Batteries offer the
speed, reliability, and dispatchability to directly address these issues.
Energy storage technology has great potential to improve electric power
grids, to enable growth in renewable electricity generation, and to provide
alternatives to oil-derived fuels in the nation’s transportation sector. In the
electric power system, the promise of this technology lies in its potential to
increase grid efficiency and reliability—optimizing power flows and
supporting variable power supplies from wind and solar generation. In
transportation, vehicles powered by batteries or other electric technologies
have the potential to displace vehicles burning gasoline and diesel fuel,
reducing associated emissions and demand for oil.
45
Federal policy makers have become increasingly interested in promoting
energy storage technology as a key enabler of broad electric power and
transportation sector objectives. The Storage Technology for Renewable
and Green Energy Act of 2011 (S. 1845), introduced on November 10,
2011, and the Federal Energy Regulatory Commission’s Order 755,
Frequency Regulation Compensation in the Organized Wholesale Power
Markets, are just two recent initiatives intended to promote energy storage
deployment in the United States. Numerous private companies and national
laboratories, many with federal support, are engaged in storage research and
development efforts across a very wide range of technologies and
applications.
Energy storage technologies for electric applications have achieved various
levels of technical and economic maturity in the marketplace. For grid
storage, challenges include roundtrip efficiencies that range from under 30%
to over 90%. Efficiency losses represent a trade-off between the increased
cost of electricity cycled through storage, and the increased value of greater
dispatchability and other services to the grid. The capital cost of many grid
storage technologies is also very high relative to conventional alternatives,
such as gas-fired power plants, which can be constructed quickly and are
perceived as a low risk investment by both regulated utilities and
independent power producers. The existing market structures in the electric
sector may also undervalue the many services that electricity storage can
provide. For transportation storage, the current primary challenges are the
limited availability and high costs of both battery-electric and hydrogen-
fueled vehicles. Additional challenges are new infrastructure requirements,
particularly for hydrogen, which requires new distribution and fueling
infrastructure, while battery electric vehicles are limited by range and
charging times, especially when compared to conventional gasoline vehicles.
46
Substantial research and development activities are underway in the United
States and elsewhere to improve the economic and technical performance of
electricity storage options. Changes to market structures and policies may
also be critical components of achieving competitiveness for electricity
storage devices. Removing non-technical barriers may be as important as
technology improvements in increasing adoption of energy storage to
improve grid performance.
2.4.1. Energy Storage Technologies
This research attempts to summarize the current state of knowledge
regarding energy storage technologies for both electric power grid and
electric applications. It is intended to serve as a reference for policymakers
interested in understanding the range of technologies and applications
associated with energy storage, comparing them when possible, in a
structured way to highlight key characteristics relevant to widespread use.
While the emphasis is on technology (including key performance metrics
such as cost and efficiency), this report also addresses the significant policy,
market, and other non-technical factors that may impede storage adoption. It
considers eight major categories of storage technology: pumped hydro,
compressed air, batteries, capacitors, superconducting magnetic energy
storage, flywheels, thermal storage, and hydrogen.
There are several energy storage technologies available to utilities, and
many are under development. These storage technologies all have different
purposes; some are high power, low energy, and some can provide various
amounts of energy for much longer durations. Some have no startup time
required and can truly provide grid stabilization; others have startup times in
the seconds to minutes timeframe and are much better suited to load
following and peak shaving. For any energy storage application there are
usually one or two technologies that are feasible solutions.
47
Batteries are probably the first thing someone thinks of when they think of
electricity storage. There are many different types of batteries, including
lead acid, nickel-cadmium, lithium-ion, sodium-sulphur, zinc-bromine, and
vanadium redox batteries [48]. Lead-acid batteries are the most common as
they have been around since the 19th century, but have a low energy density
and power density compared to newer technologies. Nickel-cadmium and
nickel-metal hydride batteries are an option for consumer electronics as well
as medium-scale energy storage. A 40 MW, 10 MWh energy storage system
using nickel-cadmium batteries for grid stabilization and backup opened in
Fairbanks, Alaska in September, 2003 [49]. Lithium-ion batteries are also a
newer battery technology with a higher energy density than lead-acid, but
lithium-ion batteries are being developed primarily for electronics and
electric or hybrid-electric automobiles.
There are some new battery technologies that are more favorable for utility-
scale energy storage. NGK Insulators of Japan is now developing a 34 MW,
245 MWh sodium-sulphur batteries for medium-scale load following and
peak shaving applications with wind integration [50]. Certain flow batteries
(vanadium redox and zinc bromine, for example) can produce those levels
of power for longer durations than sodium-sulphur batteries. There has been
significant development of these technologies in the past few years and
utilities are beginning to install these batteries in several locations
throughout the United States.
Flywheels are another energy storage technology that can provide power
quality and voltage regulation, as well as a large amount of power for a
short duration when needed. They are a mechanical energy storage system
which stores energy in the rotating inertia of a mass. This concept has been
around for over 100 years, but commercial development has been slow.
Beacon Power in Massachusetts makes both high power and low power
flywheels [51]. The high power flywheels have a short duration (minutes),
48
but can provide a significant amount of power. This is the most popular
application of flywheels, providing the ancillary services of grid
stabilization and spinning reserves.
Superconducting Magnetic Energy Storage (SMES) and ultracapacitors are
energy storage technologies which provide grid stabilization and voltage
and frequency regulation. SMES stores energy as a continuously circulating
current through a superconducting magnetic coil. An ultracapacitor is a very
large electrochemical capacitor that acts just like a conventional capacitor in
a circuit, storing energy due to an applied voltage. These technologies are
unique from other energy storage technologies discussed here because they
actually store electricity rather than converting the electricity to another
form of energy and then storing the energy in that form. They are only
capable of providing power for a few seconds, so these technologies are
only used for power quality applications.
Pumped hydro energy storage (PHES) is the most developed and most
widely used energy storage technology. According to a study done by
Electricity Storage Association, there are about 90 GW of electric energy
storage worldwide, almost all of which is pumped hydro storage. There are
approximately 280 pumped hydro energy storage facilities worldwide [52].
Traditional hydro generating stations are not storage facilities in the strictest
sense but their output can be controlled to provide regulation support.
Unlike a hydro generating station, a pumped hydro storage system allows
two–way water flow. During off–peak hours, the generator acts as a motor
to store water in an elevated reservoir. During peak hours, the water is
released to produce electricity.
Although large hydro stations have high generation efficiency, losses due to
evaporation and leakage water reduces the overall efficiency for pumped
hydro storage units. The overall efficiency of this technology is roughly 75%
49
[53], [54]. Very high energy and power capacities are two main features of
pumped hydro storage. Large pumped hydro storage units can hold up to
1000s of MWh energy. For example, a recently constructed unit in the Alps
can store up to 8.5 GWh of energy and supply 1.06 GW of power
[55].Pumped hydro plants are capable of providing maximum ramp rate,
approximately equivalent to their rated capacity, in less than 1 minute
response time [56]. Life time of pumped hydro storages is very long – some
have been in operation for over 50 years [54].
Of the major difficulties with pumped hydro storage is its low energy
density. Siting of these storage facilities require large areas preferably with
different elevation levels. If geographical restrictions do not allow upper and
lower reservoirs, underground reservoirs can be used as well. This option
leads to higher cost of construction and longer lead time. Lack of suitable
locations and impact on the environment are major drawbacks of pumped
hydro storage.
Compressed air energy storage (CAES) devices store electrical energy by
converting it into mechanical form. Air is compressed into a large container
and, when energy needs to be discharged, the air is expanded to release the
mechanical energy. Usually large salt caverns, abandoned mines and
aquifers are used as the air container. CAES devices have been part of grid
operation since 1970s [56].
The energy and power capacity of CAES devices depend primarily on the
size of the container. Given a large mine can be located in the area, these
devices can feature very high energy and power capacities. The compression
process develops heat and, unless this heat is conserved, the cycle efficiency
of CAES devices becomes low. In addition to the slow response rate due to
dependence on a mechanical system, CAES devices’ specific siting
requirements make it an unviable option for most power systems. If aquifers
ar
ne
ca
A
Po
ap
ne
tw
hy
re used as th
ear its vicin
atastrophic r
A compariso
ower is sho
pplications
eeded for sy
wo technolo
ydroelectric
he air conta
nity. The c
raptures.
on of variou
own on the
can produc
ystem wide
ogies that a
c energy sto
Fig. 2.3
ainer, CAES
cavern or m
us energy st
horizontal
ce high pow
e wind inte
are truly hig
rage and co
3 Energy stora
50
S systems m
mine based
torage tech
axis, and t
wer for sev
egration. In
gh-energy s
ompressed a
age technolog
may interfer
containers
hnologies is
ime on the
veral hours,
fact, there
storage tech
air energy st
gy comparison
re with the
are suscep
shown in F
vertical ax
, which is
are current
hnologies: p
torage
n.
eco-life
ptible to
Fig. 2.3
xis. Few
what is
tly only
pumped
51
2.4.2. Battery Energy Storage System
There are many different concerns that may arise as batteries are integrated
into power generation systems. This thesis addressed some of the most basic
and ambiguous of these in a comprehensive and holistic manner. In doing so,
I outline the probable benefits of several different configuration choices for
these storage systems, so that utilities and project designers have a clearer
indication of the capabilities offered by the inclusion of such a system. I
have taken a system approach to problems that are somewhat beyond the
scope of electrical engineering in the strictest sense, in the hope of bringing
perspective to the higher-level design choices. In particular, I have
attempted to define a battery storage system as the completion of a complete
energy system.
2.4.3. Battery Types
The battery type is of the utmost importance in terms of the possible range
of the parameters of the BESS. Some of the technical specifications of the
battery types were not used explicitly in the model, but rather implicitly.
The technical specifications for some battery types commonly used in grid-
level storage are outlined in this section, as well as some less-commonly
used ones to give the reader a sense of perspective and why some qualities
are preferred over others when choosing a battery configuration for a
particular system. For each of these types, the relevant model parameter
range is given and related to the chemical or physical makeup of the battery
type. These are then summarized in Table 2.2.
Lead-Acid (Pb-Acid) batteries are known for their high power to weight
ratio from high maximum current throughput, but are hampered by their low
overall energy to weight ratios. It is for this reason that their primary
application has been in traditional internal combustion in automobiles as a
52
method to supply power to the lighting and ignition systems [57]. They are
not typically found in consumer electronics or electric vehicle applications
because of this low energy to weight ratio. In applications where weight and
size are unimportant, such as grid-level energy storage, they are more
common. Still, they are often overshadowed in grid storage applications by
other battery types such as NiCd and NiMH that offer substantially greater
capabilities in terms of energy and power density, and average cycle life —
albeit at a higher cost.
Nickel-cadmium (NiCd) batteries have similar applications to those of
NiMH batteries. They have the disadvantage of being produced using the
highly toxic metal cadmium. However until recently they were cheaper to
produce than other similar batteries [57]. The lowering cost of production of
NiMH batteries has led to stricter regulation on NiCd batteries for consumer
use, though their use in specialized application remains popular because of
their ability to tolerate high discharge rates with no loss of capacity or
damage to the battery cells. They can also be discharged much deeper and
for more prolonged periods of time than other batteries.
NiCd batteries have a charge cycle energy efficiency of between 60% and
90%— the upper extent of which places them ahead of almost every other
battery type [57]. They have an energy density of 40—60 Wh/kg, and a
power density of 140-180 W/kg. The self-discharge rate is low, at about 1%
per day, though this would not likely be relevant in a BESS system
connected at a common bus with a PV array [58]. In this setup, the battery
would likely cycle once every 1-2 days. One of the more relevant aspects of
NiCd batteries, however, is their cycle life —which averages approximately
3000 charge cycles. This would be a major component of the lifetime cost
of a BESS that selected this particular battery chemistry.
53
NiCd batteries are some of the most commonly used battery types in grid-
scale energy storage. They are the battery type used in the world’s current
largest battery array — the Golden Valley Electric Association BESS in
Fairbanks, Alaska. There, four strings of 344 battery modules connected in a
series configuration to create a tested maximum output power of 46 MW
[59], though it is designed to deliver 27 MW of power for a period of 15
minutes [60]. The project includes a total of 13,760 battery cells, and each
battery has a stated anticipated life of between 20 and 30 years. The overall
cost of the project is stated at $35 million [60].
Nickel-metal hydride (NiMH) batteries have two to three times the energy
density of nickel-cadmium batteries – though less than that of lithium-ion.
Their main disadvantage is a higher rate of self-discharge relative to other
battery chemistries. They can discharge safely from 1.4V/cell at full charge
to a maximum discharge of just over 1V/cell, with an average 1.25V/cell
during discharge, (under a current load of 0.5 A). This delivers a more
constant voltage over the entire charge cycle than other battery types,
though over-discharging can damage the cells by polarity reversal [61, 62].
NiMH batteries are well suited for high current drain applications because
of their low internal resistance. They are often used in digital cameras, as
well as electric automobiles.
NiMH batteries are well suited for many applications where high power-to-
weight ratio is a priority. They are typically not chosen for grid-level storage
because there is rarely an impetus to find a smaller, lighter battery — only
one that stores enormous amounts of energy efficiently. Unfortunately,
overall charge cycle efficiency is not a strong suit of a typical NiMH battery,
with documented rates of between 50-80%. Furthermore, their self-
discharge rate without separator devices is considered relatively high. They
have a working life of between 500 and 2000 cycles, depending on
54
application [61]. Because of these reasons, they have not been included in
the model.
Molten salt batteries, (alternatively referred to as a liquid metal batteries or
thermal batteries), are not reactive at ambient temperatures, but when heated
to a certain threshold, they achieve a measure of power and energy density
unmatched by many other batteries. Furthermore, because of the lack of
necessity for thermal control devices required of most typical batteries, they
are generally much cheaper to build and deploy. The three primary varieties
of thermal batteries are sodium ion, sodium-sulphur, and magnesium-
antimony [63].
Sodium-sulphur (NaS) batteries are a type of molten salt battery that
consists of positive and negative electrodes of liquid sulphur and liquid
sodium, respectively, separated from one another by a beta alumina ceramic
electrolyte in a solid state [57]. While a phenomena referred to as thermal
runaway is considered a potential problem that must be avoided in most
batteries, part of the simplicity of NaS batteries and molten salt batteries in
general is that their operating temperature is extremely high (572-680◦ for
NaS batteries), and what would be considered thermal runaway with other
battery types is in fact the normal operating temperature for these types. In
fact, heating is sometimes required when the battery is not in active use.
Although the requirement of external heat requires additional energy, it is
usually less than the cooling required of traditional battery types, and the
systems required for this aspect of the battery bank are less complex. In
most configurations, this heating is only required initially and during times
of standby [64]. Finally, many of the required materials for the construction
of these batteries is inexpensive and can be sourced locally — further
decreasing the resultant cost and overall environmental footprint of any
project utilizing these batteries. The electrochemical specifications of NaS
make them ideal for grid storage applications. They have an excellent stated
55
cycle life that is dependent on the manner in which they are used. Their
operating life is stated at 15 years, and the cycle is determined by the depth
of discharge (DOD) at which they are operated. When operating at a DOD
of 100% (i.e. the battery is discharged each charge cycle until no voltage
difference remains), the cycle life will consist of 2500 cycles. When
operating at a DOD of 85% (i.e. the battery is discharged each charge cycle
until it arrives at a 15% charged state), the cycle life will consist of 4500
cycles [64]. Furthermore, their charge and discharge efficiency lies between
89% and 92%, placing them well within the “high-efficiency” classification
of batteries. These characteristics are greater than that in almost every other
battery type. While recent developments of molten salt batteries have been
promising, such as those by Ambri and their magnesium antimony batteries
[53], this battery type is still considered experimental, and real-world
deployments of BESSs utilizing this battery chemistry are not yet common.
However, one such example of this technology in action is the American
Electric Power (AEP) Charleston Energy Storage project, in Charleston, VA.
This project consists of a NaS battery array rated at 1.2 MW of
instantaneous power, and an energy storage capacity of 7.2 MWh [64]. This
is the first MW-scale NaS project outside Japan, and it was installed to
provide peak-shaving and grid support services at the local substation [67].
Lithium-ion (Li-ion) batteries have good energy efficiency rates of 85-95%,
and have excellent energy density and power density ratings of 100-200
Wh/kg and 360 W/kg, respectively. It is these characteristics that enable
them to be the dominant form of energy storage in consumer electronics,
where small size and weight are paramount. They have a relatively low self-
discharge rate between 5% and 10% per month, depending on the specific
type of Li-ion battery chemistry, and their working lifetime consists of
approximately 3000 charge cycles — in line with previously mentioned
battery types. Their cost is prohibitively high for most applications in grid-
56
level energy storage, where larger, cheaper alternatives abound. The cost of
lithium based batteries is expected to continue to rise as they are used more
frequently in consumer electronics and electric vehicle applications, while
the world’s known supply of lithium is being depleted faster than new stores
are discovered [57]. However, at the AES Laurel Mountain project in West
Virginia, 32 MW of Li-Ion batteries provide reserve capacity for a 97 MW
wind farm. The installation costs of the BESS are said to have been $900
per kW [68], bringing the total installation cost to $28.8 million.
Table 2.2
Attributes of Different Battery Types
Type Min. Charge (%) Efficiency (%) Cycles Cost ($/kWh)
Pb-Acid 30 75 1500 135
Ni-Cd 0 75 3000 540
Na-S 0/15 89 2500/4500 500
Li-Ion 20 70 10000 915
2.5. Conclusion
In this chapter, a number of wind power issues, such as wind power
conversion, impacts of wind power, maximum rotor efficiency, speed
control for maximum power, some of the design considerations in wind
turbine design, wind farms, latest trend of wind power generation, problems
related with grid connections and promotion of wind power generation have
been discussed. This chapter also provides a summary of the effect energy
storage and those currently developed under research for energy storage
systems. Multiple storage techniques are introduced and different kinds of
battery for storage system are presented. Technical and financial challenges
of renewable energy and storages are also included. In the next chapter, the
57
methodologies of data analysis and hybrid generating system will be
discussed.
58
This page is intentionally left blank
59
Chapter 3
Hybrid Power System & Data Analysis
Methodologies
3.1. Introduction
A survey of state-of-the-art research techniques that facilitate hybrid wind
power generation and operation is provided in this chapter. The relevant
literature review comprises broadly of the two sections outlined below. In
the first section, research on the power generation system combined with the
wind power and energy storage system is discussed. The basic concepts of
wind-thermal generation and wind-storage systems are first reviewed and
this is followed by comprehensive discussions of existing techniques. In the
second section, the availability of new computational intelligence based
methods for wind power system operation is studied and discussed. This
chapter reviews a series of popular evolutionary algorithms. The advantages
and disadvantages of each algorithm are discussed in detail. This is followed
by comprehensive comparisons of these approaches.
60
3.2. Hybrid Wind Power Generation Systems
It is no doubt that the purpose of all types of energy generation ultimately
depends on the economics. The wind power generation costs have been
falling over recent years. It is estimated that wind power in many countries
is already competitive with fossil fuel and nuclear power if
social/environmental costs are considered [69].
The installation cost of a wind system is the capital cost of a wind turbine,
land, tower, and its accessories, and it accounts for less than any state or
federal tax credits. The maintenance cost of a wind system is normally very
small and annual maintenance cost is about 2% of total system cost [70].
The cost of financing to purchase the wind system is significant with respect
to the overall cost of wind system. Furthermore there are extra costs such as
property tax, insurance of wind system and accidents caused from the wind
system. One of the main advantages of generating electricity from the wind
system is that wind is free. The cost of wind system just occurs once (the
maintenance cost included in install cost). On the other hand, the cost of
non-renewable energies are getting more and more expensive, which makes
renewable energies such as wind power more and more economically viable.
Recently, research and development make the wind power generation
competitive with other non-renewable fuels such as fossil fuel and nuclear
power. Lots of efforts have been done to reduce the cost of wind power by
design improvement, better manufacturing technology, finding new sites for
wind systems, development of better control strategies (output and power
quality control), development of policy and instruments, human resource
development, etc [71].
3.2.1. Wind Power Integration
There a
electric
and sup
autonom
electric
reliable
system
are very
The co
wind p
size of
reduce
or mor
biomas
autonom
hybrid
various
micro h
Fig
are still man
cal connecti
pply power
mous, isola
city to these
e and has b
are that th
y high.
ost of electr
ower gener
diesel engin
pollution. S
e renewable
ss, etc.) to
mous hybri
system wi
s options l
hydro, etc.
g. 3.1 Schemat
ny locations
ion to grid
to such area
ated power
e loads by d
een proved
he cost of fu
ricity can b
ration. This
ne and batte
Such system
e energy ba
meet the e
id power sy
ith main c
like wind-d
tic diagram of
61
s in differen
supply. A
as is called
system, et
diesel power
d for many y
fuel, transpo
be reduced
system has
ery storage
ms having p
ased source
electric dem
ystems. Fig
omponents
diesel, win
f general isola
nt parts of th
power sys
a remote, d
tc. It is a c
r plants. Th
years. The
ortation, op
by integrat
s another ad
system, wh
parallel ope
s (wind, ph
mand of an
g. 3.1 show
[72]. A h
nd-diesel-ph
ated wind-dies
he world tha
tem which
decentralized
common w
he diesel sys
main probl
eration and
ting diesel
dvantage of
hich can sav
ration of di
hotovoltaic,
isolated ar
ws a typica
hybrid syste
hotovoltaic,
sel hybrid pow
at do not ha
can genera
d, standalon
way to supp
stem is high
lems of dies
d maintenan
systems w
f reductions
ve the fuel a
iesel with o
micro hydr
rea are call
al wind-dies
em can ha
wind-dies
wer system.
ave
ate
ne,
ply
hly
sel
nce
ith
in
and
one
ro,
led
sel
ave
el-
62
The operation system of a diesel engine is very important. Normally there
are two main modes of system operation which are running the diesel
engine either continuously or intermittently. Continuous diesel system
operation has the advantage of technical simplicity and reliability. The main
disadvantage of this approach is low utilization of renewable energy sources
(wind) and not very considerable fuel savings. Basically, the minimum
diesel loading should be 40% of the rated output, and then minimum fuel
consumption will be around 60% of that at full load [73]. In order to get
large fuel savings, it is expected that diesel engine runs only when wind
energy is lower than the demand. Nevertheless unless the load is
significantly less than the energy supplied by the wind turbine, the diesel
generator will not be able to stay off for long time. The start-stop can be
reduced by using the energy storage methods. To make the supply under
these circumstances continuous, it is required to add complexity in the
architecture or control strategy.
As wind is highly fluctuating in nature and it will affect the quality of
supply considerably, the system may be damaged in the absence of proper
control mechanism. Main parameters to be controlled are the system
frequency and voltage, which determine the stability and quality of the
supply. In a power system, frequency deviations are mainly due to real
power mismatch between generation and demand, whereas voltage
mismatch is the primary indicator of reactive power unbalance in the system.
In the power system active power balance can be achieved by controlling
the generation, i.e., by controlling the fuel input to the diesel electric unit
and this method is called automatic generation control (AGC) or load
frequency control (LFC) or by scheduling or management of the output
power. The function of load frequency controller is to generate, raise or
lower command, depending upon the disturbance, to the speed-gear changer
of the diesel engine which in turn changes the generation to match the load.
63
Different methods of controlling the output power of autonomous hybrid
power systems are dump load control, priority load control, battery storage,
flywheel storage, pump storage, hydraulic/pneumatic accumulators, super
magnetic energy storage, etc [74].
It is equally important to maintain the voltage within specified limits, which
is mainly related with the reactive power control of the system [75, 76]. In
general, in any hybrid system there will be an induction generator for
wind/hydro system. An induction generator offers many advantages over a
synchronous generator in an autonomous hybrid power system. Reduced
unit cost, ruggedness, brushless (in squirrel cage construction), absence of
separate DC source for excitation, ease of maintenance, self-protection
against severe overloads and short circuits, etc., are the main advantages
[77].
However the major disadvantage of the induction generator is that it
requires reactive power for its operation. In the case of grid-connected
system, an induction generator can get the reactive power from
grid/capacitor banks, whereas in the case of isolated/autonomous system
reactive power can only be supplied by capacitor banks. In addition, most of
the loads are also inductive in nature, therefore, the mismatch in generation
and consumption of reactive power can cause serious problems of large
voltage fluctuations at generator terminals, especially in an isolated system.
The terminal voltage of the system will sag if sufficient reactive power is
not provided, whereas surplus reactive power can cause high voltage in the
system, which can damage the consumer’s equipment and affect the quality
of supply. To take care for reactive power/voltage control an appropriate
reactive power compensating device is required [46, 72, 74]. Another
approach available from ENERCON27 consists of a wind turbine based on
an annular generator connected to a diesel generator with energy storage to
form a stand-alone power system [72].
64
3.2.2. Embedded Energy Storage Systems
Due to the intermittent nature of renewable energy sources, hybrid
combinations of two or more energy sources along with energy storage can
improve reliability and ensure a continuous and cost-effective power supply.
In renewable energy-based hybrid power system applications, energy
storage is considered as an integral part of the system [77]-[84]. Energy
storage can improve transient stability of the system when wind and load
variation occurs [79]-80]. Most importantly, they are used for load levelling
and peak shaving applications [81], [82]. However, proper technology
selection, operation and control strategies, structure of the hybrid power
system, and generation unit sizing are also vital to construct a robust
renewable energy based hybrid power supply system [68]-[82]
There are various ways to integrate different energy sources and storage to
form a hybrid power system. Among them, dc-coupled, ac-coupled and
hybrid-coupled are the most popular options [83]-[84].
The energy storage system (ESS) in a wind farm can be configured as either
one aggregated unit that serves the whole wind farm, or distributed ones,
installed in each wind turbine generator (WTG), Fig. 3.2 [85]. Other
configurations involve a number of partially aggregated units, each of which
serves a group of WTGs.
Few papers look into the structure and performance differences of the ESSs
in these different configurations. Due to the smoothing effect from the
spatial distribution of WTGs, the total wind farm power is less fluctuant
than the individual WTG powers. For this same reason, the aggregated ESS
is often assumed to have a superior performance to the distributed ESS of
the same total capacity. This is partially verified in some simplified
conditions in [86].
Fig. 3.
3.3.
In this
discuss
compar
intellig
emphas
system
3.3.1.
Power
generat
schedu
uncerta
wind p
generat
scales.
conven
2 The Aggreg
Data An
section, th
sed, with th
risons. The
gence based
sised. Final
s operation
Power Sy
systems no
tors, which
les or vary
ainty and in
ower into a
tion should
Under the
ntional gene
gated (left) an
nalysis in
he importa
he focus be
e importan
d methods f
lly, the fut
in a deregu
ystem Op
ormally incl
can control
the output i
ntermittency
a power sys
d be conside
e normal ci
erator will
65
nd distributed
n Wind P
ance of win
eing on exis
nce of intr
for wind p
ture needs
ulated marke
peration w
lude an abu
l their outp
in accordan
y of wind re
stem. At the
ered in the
ircumstance
be schedu
(right) ESS co
Power S
nd power
sting appro
roducing a
power syste
of data an
et environm
with Wind
undance of
ut power fo
nce with sys
esource, it is
e same time
e power sys
e of therm
uled to mee
onfigurations
System
system dat
oaches and
advanced c
em data ana
nalysis for
ment are disc
d Power
convention
ollowing the
stem deman
s a challeng
e, the variab
stem over d
al power s
et the syst
in wind farms
ta analysis
their detail
computation
alysis is th
wind pow
cussed.
nal fossil-fir
e system lo
nd. Due to t
ge to integra
bility of wi
different tim
system, eve
tem load a
s.
is
led
nal
hen
wer
red
oad
the
ate
ind
me
ery
and
66
reserve requirements at minimum operating cost. It is necessary for the units
to vary output power to match the system load changes over the scheduling
time, which requires the generators to have the capability to meet the load
fluctuation as well as sudden, unexpected changes in the system demand
[87].
Along with the integration of wind power come more complicated control,
requirements and reserves should be considered. If the wind generation is
involved into the thermal power system, the system operators may be forced
to alter the generator loading levels, ramping requirements, spinning reserve
and other relevant issues. Having recognized the wind power integration
problem, it follows that the power system operators must develop a plan of
action.
Here, the emphasis is on to two important problems of operational planning
for power systems with wind power generation, economic load dispatch and
unit commitment.
Economic load dispatch (ELD) [88] is an important topic in the operation of
thermal power plants which can help to build up effective generating
management plans. It aims to allocate power generation to match load
demand at minimal possible cost while satisfying all the units and system
constraints. In previous research, different approaches have been suggested,
including linear programming and non-linear programming [89]-[91].
Linear programming methods are fast and reliable, but the main drawback is
that it is associated with the piecewise linear cost approximation [92]. The
non-linear programming methods have a problem of algorithm convergence
and complexity [93].
Recently, different heuristic approaches have been proven to be effective,
such as evolutionary programming (EP) [88]-[90], SA [91], tabu search (TS)
[92], pattern search (PS) [93], GAs [94],[95], DE [96], and PSO [97]. EP
67
can be a quite powerful evolutionary method; however, it is rather slow
converging to a near optimum for some function optimization problems [98].
Both SA and TS can be quite useful in solving complex reliability
optimization problems, but SA is very time consuming, and cannot easily be
utilised to tune the control parameters of the annealing schedule. TS is
difficult in defining effective memory structures and strategies which are
problem dependent. Although GAs can ensure the colony evolves and the
solutions change continually, they often lack a strong capacity of producing
the best offspring individuals and thus cause the slow convergence near
global optimum and sometimes may be trapped into local optima. DE is no
doubt a very powerful method, but the greedy updating method and intrinsic
differential property usually leads the computing process to be trapped by
local optima. The PSO converges quickly, but has a slow fine-tuning ability
of the solution. Once it gets stuck into the local optima, it is very hard to
jump out of it.
In today’s society, the power system generation scheduling problem can be
divided into two relevant optimization sub-problems: unit commitment (UC)
and economic dispatch (ED). The main objective of the unit commitment is
to decide the ON/OFF statuses of generators over the scheduling period to
meet the system load demand and reserve requirements at the lowest cost.
Basically, the unit commitment outputs are ON/OFF statuses on an hourly
basis for a given time scales, such as 24 hours. In addition, a unit
commitment is an optimization problem that determines which and when a
generator is to be working and for how long. Unit commitment schedule is
approached by satisfying the system constraints such as ramp rate limits,
spinning reserve as well as minimum up and down time limits.
In the literatures, many researchers have shown the interests to unit
commitment methods and various numerical optimization techniques have
been employed to solve the unit commitment problems. In the traditional
68
UC problem, many mathematical methods have been proposed such as
priority list (PL) [99 100] approaches, dynamic programming (DP) [101],
branch-and-bound (BB) [102] methods, mixed-integer programming (MIP)
[103] and Lagrangian Relaxation (LR) [104, 105] methods. Recently,
optimization solvers based on heuristics techniques have been proved to be
effective with promising performance, including genetic algorithm (GA)
[106-109], evolutionary programming (EP) [110], fuzzy logic (FL) [111],
artificial neural network (ANN) [112], simulated annealing (SA) [113],
particle swarm optimization (PSO) [114] as well as hybrid techniques [115-
117]. Many researchers are attracted by heuristic optimization methods.
Apart from providing local optimal solutions, those approaches provide
global optimal solution and easily dealing with various difficult nonlinear
constraints.
3.3.2. Optimization Approach
In this section, a group of Evolutionary Algorithms (EAs) will be reviewed,
which take inspirations from evolutionary or adaptive systems in the
biological and physical world, using to solving optimization problems. In
the EAs, normally a population of candidates is generated randomly within
search spaces first, and then evolves according to kinds of distinguished
implementations, such as selection, crossover, mutation, or recombination.
With fitness function evaluation, the population evolves towards global
optimum in the search space. Four kinds of popular EAs are introduced as
follows, including the GA, IA, and PSO.
3.3.2.1. Genetic Algorithm
GAs [118] are one of the most famous families of EAs. It is implemented as
a computer simulation of gene evolution in which a population of gene
representations of candidate solutions to a specific problem evolves toward
69
better solutions. Originally, these solutions are represented in binary as
strings “0” and “1”. GAs usually begin with a randomly generated
population of individuals within the search space. In each generation, the
fitness of every individual is evaluated, and then undergoes selection,
crossover, and mutation to form a new population. Commonly, GA
terminates when either a maximum number of generations or satisfactory
fitness value has been reached. In this section, the procedures of the
classical binary-coded GA are represented.
Step-1. Initialization. Each unit is a value decoded from a gene which can be
represented as a binary string. For a five-digit binary string and unit range is
[-10,10], the gene {0,0,0,0,0}can be decoded to -10, and gene {1,1,1,1,1}
can be decoded to 10.
Step-2. Selection. From the theory of natural evolution selection, the
individuals with higher fitness values are more likely to produce better
offspring. Normally, the roulette selection is used in the selection procedure.
A roulette wheel on which each member of the population is given a sector
whose size is proportional to the fitness of individual is constructed [118].
Then the wheel is spun and the selected individual becomes parent.
Step-3. Crossover. Crossover is a random implementation of recombination
in which each parent contributes part of its genetic structure to offspring.
Here the single-point crossover is employed. Based on the crossover
possibility, individual exchange of characters between two strings is
performed.
1 1 0|0102 1 1|011
SS
(3.1)
Suppose in choosing a random integer in [1,4], if in case of 2, the crossover
occurs after the second number can be seen below
St
th
bi
of
as
A
tep-4. Muta
he value. Wi
it to a diffe
ffspring, an
s follows.
A flow chart
ation. Mutat
ith the bina
erent represe
d a new po
of a basic G
F
1' 1 0|02 ' 1 1|0
SS
tion is the i
ary string re
entation. Th
pulation wi
3 1 0 01S
GA is given
Fig. 3.3 Flowc
70
011011
implementa
epresentation
hen the par
ill be gener
11 3'S
n in Fig. 3.3
chart of a typi
ation of occ
n, this simp
rents will be
ated. An ex
1 0 111
, [118].
ical GA.
casional tun
ply means ch
e replaced b
xample can
(3.2)
nning of
change a
by their
be seen
(3.3)
71
3.3.2.2. Immune Algorithm
With the development of immunology, the mechanism of biologic immune
system has been gradually discovered by researchers. Because of the
powerful capability of information processing and special characteristics
such as diversity, adaptive trait, biologic immune system has become a hot
spot of artificial intelligence research. Immune algorithm (IA) [119]-[121]
imitates the principle of the defence system annihilates foreign disease-
causing bacteria or viruses through self-learning and self-adjusting
mechanism. Although IA is very similar to GA, there are essential
differences in the production theory for population. Compared to GA and
other kinds of EAs, IA enhances searching ability through the mechanism of
memory pool. Meanwhile, it achieves self-adjusting by introducing two
distinguished discriminators, affinity and concentration. To some extent, it
can avoid premature convergence. It should be noted that similar techniques,
such as sharing function method, have been used with other EAs to discount
the fitness values of closely located individuals in the search domain, in
order to achieve higher diversity in the search process. The evolutionary
procedures of IA are represented as follows.
Step-1. The antigens and antibodies in IA represent the objective functions
and feasible solutions, respectively. The affinity and concentration are used
as discriminators of the quality of solutions, which are calculated by
1( ) (1 )iiAs t r r (3.4)
where,
r random number in (0,1);
72
i location index of antibodies in current population which are
rearranged in terms of the values in ascending sequence, i ϵ [1, p], where p
is population size.
1
1( )
p
i mni
t Ksp
Cs
(3.5)
1, || ( ) ( ) ||
0,m n
mn
Ab t Ab t lKs
otherwise
(3.6)
where,
||◦|| Euclidean distance;
l distance threshold;
Step-2. Then, a roulette selection is implemented based on the selection
probabilities for the antibodies. This allocates each antibody a probability of
being selected proportionally according to affinity and concentration. The
selection rates can be calculated by
i
ii p
i
i=1 i
As (t)Cs (t)
(t) =As (t)Cs (t)
Ps
(3.7)
Step-3. After that, crossover and mutation are implemented. Crossover is
one of the primary IA operators that promote the new region exploration
ability in the space. Generally, crossover rate should be chosen
comparatively large, between 0.7 and 1.0. Mutation is another operator
which guarantees the population diversity. And the mutation rate should be
chosen between thousandths and hundredths.
An arithmetic crossover operator is described as follows
( ) ( ) (1 ) ( )i m nt b t b tAb Ab Ab (3.8)
And mu
followi
Step-4.
next ge
antibod
values.
A flow
utation ope
ing formula
( )i tAb
Finally, an
eneration a
dies will be
chart of a b
rator can be
e
( ) (i tAb
ntibodies wh
and be adde
inserted int
basic IA is g
Fig. 3.
73
e selected in
1
1) 1t b
which have h
ed into me
to populatio
given in Fig
.4 Flowchart o
n the algori
1r
m
t
T Ab
high affinity
mory pool.
on, replacin
g. 3.4, [121]
of a typical IA
ithm are des
( ) ( )m nt tAb
y values wi
. Given nu
g those with
].
A.
scribed as t
(3
ill evolve in
umber of ne
th low affin
the
.9)
nto
ew
ity
74
3.3.2.3. Particle Swarm Optimization
PSO is a global search technique originally introduced by Kennedy and
Eberhart [121]. It simulates the social evolvement knowledge, probing the
optimum by evolving the population which may include candidate solutions.
Compared with other EAs, PSO shows incomparable advantages in
computational speed and precision [123]. In short, the PSO is characterized
as a simple heuristic of well-balanced mechanism with flexibility to enhance
and adapt to both global and local exploration abilities, which gains lots of
attention in power system applications [124],[125]. In order to improve the
global search ability, avoiding being trapped into local optima in solving
multimodal problems; many revised versions of PSO appeared, mainly
concentrating in improving the evolution implementations and exploring the
best parameters combinations.
The origins of PSO are best described as sociologically inspired, since the
algorithm was based on the sociological behaviour associated with bird
flocking [123]. In the conventional PSO, each individual is treated as a
particle in the space, with position and velocity vectors. The algorithm
maintains a swarm of particles, where each particle represents a potential
solution to the objective problem. For a given n-dimensional problem, the
position and velocity vectors of a particle in the PSO can be represented as
,,1 ,2
,,1 ,2
( ) [ ( ), ( )......, ( )]
( ) [ ( ), ( )......, ( )]j j nj j
j j nj j
x t x t x t x t
v t v t v t v t
(3.10)
The core idea of classical PSO is the exchange of information among the
global best, population best, and current particles, which can be done as
follows
(
(j
j
x
v
where,
φ, η
r1, r2
ω
Ppb
Pgb
vj
The flo
( 1) (
( 1)jt x
t v
parameter
random nu
inertia wei
local best p
global best
velocity ve
ow chart of a
1
) ( 1)
( )j
j
t v t
v t r
s;
mber in (0,
ght;
particle;
t particle;
ectors;
a typical PS
Fig. 3.5
75
)
[ ( )pbp t x
1);
SO is given
5 Flowchart of
2( )]jx t r
in Fig. 3.5,
f a typical PSO
[ ( )pbp t x
[122]
O.
( )]jx t (3.
11)
76
3.3.2.4. Comparison
Although the heuristic methods do not always guarantee discovering
globally optimal solutions in finite time, they often provide a fast and
reasonable solution. Generally speaking, all these algorithms are the same,
only with different theory background and evolutionary implementations.
Each method has its own merits and drawbacks, and the problem of local
optima is unavoidable. Consequently, the research emphasis may focus on
how to improve search capability and computing efficiency. Many attempts
try to merge some of their individual implementations together into a new
algorithm, so that it can overcome individual disadvantages as well as
benefit from each other’s’ advantages. Based on previous algorithms
research experience, compared with other alternatives, PSO is
computationally inexpensive in terms of memory and speed. The most
attractive features of PSO can be summarized as: simple concept, easy
implementation, fast computation, and robust search ability [126].
Table 3.1
Comparisons of the Algorithms
Approaches` Theory Speed Accuracy Variations
GA Gene evolution ♦♦ ♦♦ ♦♦♦
IA Immunology ♦ ♦♦♦ ♦
PSO Social evolvement ♦♦♦ ♦ ♦♦♦
♦ represents the degree or the score of each class.
3.3.3. Advanced Techniques
Along with the introduction of wind power forecasting and wind power
system operation, the amount of data associated from power system
considering wind power has been increasing sharply. This has introduced
difficulties for wind power system data analysis with the traditional
approaches. As a result, it is necessary to introduce advanced approaches
77
into wind power system data analysis, such as artificial neural networks and
time series models.
3.4. Conclusion
This chapter provides a summary of available approaches and those
currently under research for optimal design of hybrid renewable energy
systems. Different approaches for the configuration and energy management
of hybrid systems are presented. Detailed reviews of a wind energy
generation system embedding energy storage systems are presented.
Technical and financial challenges of renewable energy combined energy
storage systems are also included.
This chapter has described the importance of data analysis for wind power
operation. It is clear that despite the many hundreds of approaches that have
been developed for these problems, however each method has its own
advantages and disadvantages, the comprehensive comparisons have
provided after detailed discussion of these techniques.
Due to the deregulation and growth of power system and market, the
existing approaches cannot provide satisfactory performance any longer.
More advanced data analysis techniques should be introduced into solving
power system problems. Meanwhile, the computational power of modern
computers enables the employment of new data analysis techniques to be
practical and effective. The possibility and availability of employing new
computational intelligence based methods for wind power system operation
has been studied and discussed.
In the next chapter, power system economic dispatch considering wind energy and emission problem will be discussed.
78
This page is intentionally left blank
79
Chapter 4
Power System Dispatch Considering Wind
Energy and Emission
4.1. Introduction
In this chapter, a computation framework for addressing combined
economic and emission dispatch (CEED) problem with valve-point effects
as well as stochastic wind power considering unit commitment (UC) using a
hybrid approach connecting sequential quadratic programming (SQP) and
particle swarm optimization (PSO) is proposed. The CEED problem aims to
minimize the scheduling cost and greenhouse gases (GHGs) emission cost.
Here the GHGs include carbon dioxide (CO2), nitrogen dioxide (NO2), and
sulphur oxides (SOx). A dispatch model including both thermal generators
and wind farms is developed. The probability of stochastic wind power
based on the Weibull distribution is included in the CEED model. The
model is tested on a standard system involving six thermal units and two
wind farms. A set of numerical case studies are reported. The performance
of the hybrid computational method is validated by comparing with other
solvers on the test system.
80
4.2. Power System Dispatch Integrating Wind
Energy with Emission
Power system generation scheduling problem can be divided into two sub-
problems, unit commitment (UC) and economic dispatch (ED). ED is an
important task in the power system operation, which aims to allocate power
generation match load demand at minimal possible cost while satisfying all
the units and system constraints [127-129]. Suitable improvements in the
unit outputs scheduling can contribute to significant cost savings.
With the awareness of environmental pollution contributed by the
combustion of fossil fuels, building a low-carbon world has attracted
widespread attention. Many countries are trying to utilise clean energy in
order to mitigate the greenhouse effects. The primary source of greenhouse
gases (GHGs) is the combustion of fossil fuels. Coal, oil, and gas are the
three major types of regular fuels, which produce emissions represented by
GHGs, such as CO2, NO2, and SOx. In order to reduce the GHGs emissions,
the combined economic emission dispatch (CEED) considering UC was
proposed, which can take account of fuel cost and emission tax together. As
the amount of emissions from fossil-based thermal generators depends on
the amount of generated power, therefore the emission cost increase leads to
reduced overall power generated by thermal units, which in turn lowers
emissions. Moreover, the natural economic forces will also help to catalyse
the move to greater energy efficiency and use of renewable sources. Wind
energy is among the major contributors to an overall reduction in GHGs
emissions. Dispatch strategies normally can provide quick solutions to
improve the current situation of system operation and reduce carbon
emissions dramatically. On the other hand, utilising renewable energy is
another effective way to mitigate energy source deficiency, control GHGs
emissions, and achieve the smart grid vision [130–132]. Wind power being
81
one of the most appealing renewable energy resources has gained
widespread concerns during the last few decades. Along with the
introduction of various emission reduction schemes, increasing the number
of wind turbines that have been installed around the world [133]. However,
due to the intermittent and stochastic characteristics of wind resource, wind
power brings great challenges to power system economic dispatch problems.
One of the major challenges is how to effectively accommodate the wind
forecasting errors. Because variations of wind speed directly influence the
power output of wind farms, which then causes difficulties in estimating
suitable system reserve margins to ensure secure and reliable system
operations. As a result, high penetration of wind power also causes high
potential risks and more difficulties in power system operation. Moreover,
many publications have indicated that wind speed approximately follows a
Weibull distribution [134]. In order to assist with management of the
uncertainties of wind forecasts, extensive research has been conducted to
develop kinds of probabilistic optimization strategies [135, 136]. In this
chapter, a schematic representation of computational framework contains
wind power forecast and stochastic unit commitment/economic dispatch,
which is adopted from [137], is shown in Fig. 4.1.
In order to accommodate the revised dispatch strategy, more efficient
solvers are needed. Different heuristic techniques have been developed to
solve the classical ED problems with constraints, to namely simulated
annealing (SA) [138], genetic algorithm (GA) [139], evolutionary
programming (EP) [140, 141], tabu search (TS) [142], pattern search (PS)
[143], particle swarm optimization (PSO) [144, 145], as well as differential
evolution (DE) [146, 147]. Based on our experience, when compared with
other approaches, the PSO is computationally inexpensive in terms of
memory and speed. However, these heuristic methods do not always
guarantee discovering globally optimal solutions in finite time, especially
w
so
op
w
se
no
so
ap
w
hy
SQ
op
m
ef
when being a
ophisticated
ptimization
widespread a
equential qu
onsmooth f
olve the ED
pproach com
was develope
ybrid optim
QP is one
ptimization.
method over
fficiency an
Fig. 4.
applied into
d computa
techniques
attention. I
uadratic pro
fuel cost fu
D problem w
mbining DE
ed to addre
mization met
e of best
. SQP–PSO
r a large
nd percentag
.1 Computatio
o large-scale
ational too
which com
In [148], th
ogramming
unction. A
with kinds o
E with biog
ss both con
thods were
nonlinear-p
O technique
number of
ge of succes
onal framewor
82
e optimizati
ols are r
mbine differ
he authors
(SQP) for
hybrid self
of constrain
geography-b
nvex and no
found to be
programmin
e is an effe
f test prob
ssful solutio
rk considering
ion problem
required.
rent approac
presented
solving the
f-tuning DE
nts in [149].
based optim
on-convex E
e more effe
ng method
ective nonli
blems in te
ons.
g wind power
ms. Therefor
Recently,
ches have r
a hybrid E
e ED proble
E was prop
In [150], a
mization (DE
ED problem
ctive and a
ds for cons
inear-progra
erms of ac
uncertainties.
re, more
hybrid
received
EP and
em with
posed to
a hybrid
E/BBO)
m. These
accurate.
strained
amming
ccuracy,
.
83
4.3. Probabilistic Modelling of Wind Power for ED
Modelling
Wind power, one of the most appealing renewable energy sources, has been
widely developed in the recent years. Wind power provides many
advantages over alternative sources such as no pollution, relatively low
capital cost, and a short gestation period. However, the wind resource
changes with locations and climates resulting in high uncertainties in the
produced energy. The total power available from a wind turbine is equal to
the product of the mass flow rate of the wind mW, and V2/2. Assuming
constant area or ducted flow, the continuity equation states that mW=ρAV,
where ρ is the density of the air in kg/m3, A is the blades area in m2, and V is
the velocity in m/s. Thus, the total wind power becomes
PW=(mWV2)/2=(ρAV3)/2 (MW). In this equation, the wind speed V is a
random variable. Ignoring minor nonlinearities, the function relation
between a given wind speed and power output can be described in Fig. 4.2.
inv rv outv
Wind Speed m s
Win
dP
ower
MW
rw
Fig. 4.2 Simplified wind turbine power curve.
In the above figure, w (MW) is the wind energy conversion systems (WECS)
output power; wr (MW) is the WECS output rated power; vin (m/s), vr (m/s),
vout (m/s) is the WECS cut-in speed, rated speed, and cut-out speed,
respectively. From Fig. 4.2, we can see that there is no power generated at
84
wind speeds below vin or above vout; at wind speeds between vr and vout, the
output is equal to the rated power of the generator; at wind speeds between
cut-in wind speed and rated wind speed, the output is a linear function
power.
Therefore, the wind power output can be described as,
0,,
,
in out
in r
r r out
W V v or V vW aV b v V vW w v V v
(4.1)
where r
r in
wa
v v
, in r
r in
v wb
v v
.
Weibull distribution is the most popular density function that can be used to
describe wind speed frequency curve [151]–[153]. An extensive review of
various probability density functions of wind speed was provided in [153],
and comparisons were made. The results indicated that the two-parameter
Weibull distribution is the widely accepted model. Using two-parameter
Weibull distribution, cumulative distribution function (CDF) and probability
density function (PDF) of wind speed are
1 exp , 0k
VvF v vc
(4.2)
1
expk k
Vk v vf vc c c
(4.3)
where k>0 is the shape parameter, c>0 is the scale parameter.
According to Eq. (4.1), three portions of WECS power output can be
analyzed and the corresponding probabilities (CDF or PDF) can be
calculated.
(i) For inV v or outV v ,
85
0
1
1 exp exp
in out
in outV V
k kin out
P W P V v P V v
F v F v
v vc c
(4.4)
(ii) For in rv V v , in r
r in
V v wW aV b
v v
, depending on the definition of
cumulative distribution function (CDF), the CDF of WECS output power
can be described as,
in rW
r in
r in r inin inV
r r
V v wF w P W w P W w
v v
v v w v v wP V v F v
w w
(4.5)
We can obtain the PDF of W by differentiating with respect to w. The chain
rule for derivatives can be used, dF dF du
dw du dw , where u is the argument of F,
r inin
r
v v wu v
w
, and we then obtain
1
exp
k
r inin
r in rW
r
k
r inin
r
v v wv
k v v wf w
cw c
v v wv
wc
(4.6)
(iii) For r outv V v ,
86
exp exp
r r out
out rV V
kkoutr
P W w P v V v
F v F v
vvc c
(4.7)
4.4. Mathematical Formulation of CEED Problem
with Wind Power
This section describes the problem formulation of the proposed CEED
considering UC model including wind power. The model aims at
minimizing the total operation costs (including fuel cost, wind farm cost)
and emission cost while satisfying the given constraints. In [145], an
economic dispatch (ED) model incorporating wind power is developed. In
order to accurately characterize the uncertainty in the availability of wind
energy, penalty costs functions for both underestimation and overestimation
cases were added. Inspired by the practical application, a similar CEED
model was developed with an additional term incorporated to account for
government wind farm subsidy. To address the uncertainties in wind power
production, wind speed distribution probability functions are applied in
formulating the optimization model.
4.4.1. Objective Function
The objective function is formulated to minimize the total system operation
costs and greenhouse gases (CO2 and NO2) emission costs. A cost function
is obtained based on the ripple curve for more accurate modelling which
contains higher order nonlinearity and discontinuity due to the valve point
effect [143], and should be refined by a sine function [153]. The overall
objective function can be expressed as the sum of these two terms,
87
1 2.Min Cost Cost (4.8)
1.) Total system scheduling costs
, ,11 1
, , , ,1 1
, ,1
M N
i i w j j avi j
N N
u j j av j o j j j avj j
N
s j j avj
Cost C p C w
C W w C w W
C w
(4.9)
2,minsini i i i i i i i i iiC p a b p c p d e p p (4.10)
where i iC p is the fuel cost function of thermal generator i. , ,w j j avC w is
the wind power cost of the wind farm. If the wind farm is owned by the
system operator, this term may not exist. In this chapter, the wind farm is
assumed to be owned by the operator, so this cost is equal to zero. The
underestimation cost , ,u j j av jC W w occurs if the generated wind power is
more than the predicted, thus the system operator should compensate for the
surplus wind power cost. On the other hand, if the actual wind power is less
than the predicted scheduling power, the operator needs to purchase from an
alternate source and pay the overestimation cost , ,o j j j avC w W . The last
term in the Eq. (4.9) is the wind power subsidy cost , ,s j j avC w . As one of the
renewable energy subsidy projects, wind farm in many countries receive a
largely covert subsidy. An excellent example is the Renewable Obligation
(RO) in UK. The RO is designed to encourage generation of electricity from
eligible renewable sources in the UK [154]. In this chapter, the wind farm
was assumed to receive a fix cost subsidy for generating every MW wind
power.
88
According to [155], the cost of underestimation will be assumed as follow,
,
, ,
, , ,
,
r j
j
r j r j
j j
w
u j j av j u j j Ww
w w
u j jW Ww w
C W w C w w f w dw
C wf w dw w f w dw
(4.11)
In terms of overestimation case, the cost equation will be in the similar
manner,
, , , 0
, 0 0
j
j j
w
o j j j av o j j W
w w
o j j W W
C w W C w w f w dw
C w f w dw wf w dw
(4.12)
Eq. (4.11) and (4.12) can be solved through the wind power probability Eqs.
(4.4)-(4.7).
2.) Greenhouse gases (GHGs) emission costs
2 ,1
M
iGHG ii
Cost F p
(4.13)
where
, i i iGHG iF p h EM p (4.14)
2( )i i i i i i i iEM p ef d e p f p (4.15)
Eq. (4.13) represents the GHGs emission cost function. In Eq. (4.14), h is
the given GHGs emissions price which is determined by regulations and
markets. ( )i iEM p is the GHGs emissions of thermal generator i and is
calculated by the Eq. (4.15). efi is the fuel emission factor of GHGs for
thermal generator i, while di, ei, and fi are fuel consumption coefficients.
The GHGs are CO2 and NO2 in this chapter.
4.4.2. System Constraints
89
,max,min i iip p p (4.16)
,0 j r jw w (4.17)
1 1
M N
i j d lossi j
p w p p
(4.18)
Inequality constraint Eq. (4.16) defines the limitations of thermal units
output from the lower to the upper bound, and constraint Eq. (4.17) shows
the wind power output limitations. Eq. (4.18) gives the power balance
between generations and loads including the transmission losses. pi and wj
are thermal generator and wind generator output, pd is system load demand,
ploss is transmission losses.
4.5. Hybrid Optimization Algorithm
In this section, a hybrid optimization algorithm is presented, which
combines SQP and PSO together.
4.5.1. Sequential Quadratic Programming (SQP)
Since its popularization in the late 1970s, SQP has arguably become the
most successful approach for solving nonlinearly constrained optimization
problems [156]. Backed by a mature and solid theoretical background, SQP
has been developed and used to solve a remarkably large number of
practical problems. The basic principle of sequential approximations is to
replace the given problem by a sequence of quadratic sub-problems that are
easier to solve [157], [158].Consider the application of the SQP
methodology to nonlinear optimization problems,
Min.f(x)
90
( ) 0.
( ) 0h x
s tg x
(4.19)
The Lagrangian of this problem can be written as,
( , . ) ( ) ( ) ( )TL x f x h x g x (4.20)
where and are vectors of multipliers. SQP is an iterative procedure
which models the problem for a given iterate kx by a quadratic
programming sub-problem, solves that quadratic programming sub-problem,
and then uses the solution to construct a new iterate 1kx
The sub-problem can be constructed by linearizing the constraints around kx ,
and it can be written as,
Min. 1( )( ) ( ) ( )( )
2k k k T k kf x x x x x Hf x x x
( ) ( )( ) 0.
( ) ( )( ) 0
k k k
k k k
h x h x x xs t
g x g x x x
(4.21)
We need to update the estimates of the multipliers, and define the
corresponding search directions, and then choose a step size and define the
next iteration.
4.5.2. Particle Swarm Optimization
PSO is a global search technique originally introduced by Kennedy and
Eberhart [159]. It simulates the social evolvement knowledge, probing the
optimum by evolving the population which may include candidate solutions.
In the classical PSO, each individual is treated as a particle in the space,
with position and velocity vectors. The algorithm maintains a swarm of
particles, where each particle represents a potential solution to the objective
91
problem. For a given n-dimensional problem, the position and velocity
vectors of a particle in the PSO can be represented as
,,1 ,2
,,1 ,2
( ) [ ( ), ( )......, ( )]
( ) [ ( ), ( )......, ( )]j j nj j
j j nj j
x t x t x t x t
v t v t v t v t
(4.22)
The core idea of the classical PSO is the exchange of information among the
global best, population best, and current particles, which can be done as
follows
1 2( 1) ( ) [ ( ) ( )] [ ( ) ( )]
( 1) ( ) ( 1)j j j jpb gb
j j j
v t v t r p t x t r p t x t
x t x t v t
(4.23)
where vj is the velocity vectors, is inertia weight, ppb is the local best
particle and pgb is the global best particle, =1.65, =1.81
4.5.3. Composite Computation Approach
The procedures of the proposed hybrid algorithm are summarized as the
follows,
Step-1. Load history wind data, generators and wind turbines settings,
emission parameters, and forecast wind power output.
Step-2. Solve the ED and CEED problem without considering valve-point
effects incorporating wind power using SQP.
Step-3. Calculate the updated constraints using Eq. (4.24) [160], and
randomly generate initial population around the solution obtained from SQP
for PSO.
,min ,min
,max ,max
max([ ], )
min([ ], )
/ (1 )
i ii i
i i i i
i i
p p p
p p p
e
(4.24)
92
Step-4. Solve the ED and CEED problem with valve-point effects
incorporating wind power using PSO.
Step-5. Save and output final solution. Application of this approach in ED
and CEED problem incorporating wind power are presented in the
following section.
4.6. Case Studies
The QPSO is implemented on a modified IEEE 30-bus system. The
benchmark system consists of 6 thermal generators, 1 wind farm, 41
branches, and 21 loads. These thermal generators include 3 coal-fired units,
2 gas-fired units and 1 oil-fired unit.
In the case study, the CEED model with wind power was evaluated using
the historical wind speed dataset from a wind observation station in
Tasmania, Australia. The data was provided by the Australian Bureau of
Meteorology [162]. Here we assume that the wind speed data from a large
wind farm and use the data to estimate the generated wind power. The wind
speed distribution frequency and the corresponding Weibull distribution
parameters are presented in Fig. 4.3.
The Vestas V90 3.0 MW wind turbine is selected for the case studies. It is a
pitch regulated upwind wind turbine with active yawing and a three-blade
rotor. It has a rotor diameter of 90 m with a generator rated at 3.0 MW. The
Vestas V90 3.0 MW is widely used in the wind plants in Australia and has a
proven high efficiency. The parameters of the associated Weibull
distribution factor and wind farm parameters can be calculated from the
wind speed data and are given in Table 4.1.
The pro
generat
units, a
of 100
area. T
is 15%
generat
case or
system
power.
coeffic
Fi
c k
5.5 1.89
oposed algo
tors and 1 l
and 1 oil-fir
Vestas V90
he predicte
% of the rat
ted wind po
r underestim
under inve
The fuel c
ients are sh
g. 4.3 Wind sp
ө vin
9 0 4
orithm is im
large wind
red unit in t
0 3.0MW w
d power ou
ed power,
ower, the ex
mation case
estigation is
cost coeffic
own in Tab
93
peed distribut
Table 4
Wind Power
vout vr
25 16
mplemented
farm. Ther
this test sys
wind turbin
utput for eac
which is 0
xtra cost wi
e. Accordin
s 2030 MW
cients, gene
ble 4.2 and T
tion and Weib
4.1
Factors
wr Cw,j
3 0
on a test sy
re are 3 coa
stem. The w
es located i
ch wind turb
0.45 MW. D
ill be determ
ngly the ma
W, and 2330
erator limits
Table 4.3 [1
bull fitting.
Cu,j Co,j
60 20
stem includ
al-fired unit
wind farm to
in a coheren
bine is deno
Depending
mined by o
aximum ca
MW incorp
s, and fuel
61].
Cs,j
10
ding 6 therm
ts, 2 gas-fir
otally consi
nt geograph
oted as wj a
on the actu
overestimati
apacity of t
porating wi
consumpti
mal
red
sts
hic
and
ual
ion
the
ind
ion
94
Table 4.2
Fuel Cost Coefficients
Unit Fuel Cost Coefficients
ai bi ci di ei
G1 (Coal) 2000 10 0.002 200 0.084
G2 (Coal) 2500 15 0.0025 300 0.035
G3 (Coal) 6000 9 0.0018 400 0.042
G4 (Gas) 923.4 18 0.00315 150 0.063
G5 (Gas) 950 20 0.0032 100 0.084
G6 (Oil) 124.8 23.4 0.003432 80 0.098
Note: The coefficients of ai, bi, ci and ei are in $, $/MW and $/MW2, and $/MW.
Table 4.3
Fuel consumption Coefficients
Note: the coefficients of fi, gi, and hi are in t, t/MW and t/MW2 for coal/oil units. The coefficients of fi, gi, and hi are in m3, m3/MW and m3/MW2for gas unit.
In this chapter, two of most concerned GHGs emissions, CO2 and NO2 are
considered in the model. The emission characteristics of the units and
emission allowance price are shown in the Table 4.4 and 4.5.
Table 4.4
Emission Price
Greenhouse gas CO2 ($/t) NO2 ($/kg)
Price 1.5 5.0
Unit Fuel Consumption Coefficients
Pmin Pmax fi gi hi
G1 (Coal) 40 0.2 0.00004 20 110
G2 (Coal) 50 0.3 0.00005 20 100
G3 (Coal) 80 0.12 0.000024 120 600
G4 (Gas) 2462.4 48 0.0084 110 520
G5 (Gas) 2500 50 0.009 110 500
G6 (Oil) 1.248 0.234 3.43e-05 40 200
G7 (Wind) 0 0 0 0 300
95
Table 4.5
Emission Factor of Units
Emission
Factor Coal(kg/kg) Gas (kg/m3) Oil (kg/kg)
efco2 3.1604 1.84 2.8523
efno2 1.29e-03 3.4e-04 3.3e-04
Table 4.6
Forecast System Demand and Wind Farm Output
Case Index Case I Case II
Demand(MW)
G7
1200
45
1600
45
4.6.1. Case-I. ELD Model with and without Wind Farm
In this case study, the system load is 1200 MW and the system loss power is
assumed to be zero. The predict system demand and wind farm output are
listed in Table 4.6. The basic ELD model with and without wind farm are
tested on the system and the simulation results are shown in Table 4.7 and
Table 4.8.
Table 4.7
Solution of ELD without Wind Farm
Unit Power (MW) Operation Cost ($)
G1 (Coal) 96.9286 2969.04
G2 (Coal) 99.4079 4122.16
G3 (Coal) 593.5730 12359.45
G4 (Gas) 259.1281 5808.68
G5 (Gas) 110.6357 3207.22
G6 (Oil) 40.3266 1076.58
Total 1200.0000 29538.56
Overall Cost 29538.56
It
ge
H
in
m
O
F
can be show
enerated po
However, th
ncreased slig
many advant
Unit
G1 (Coal)
G2 (Coal)
G3 (Coal)
G4 (Gas)
G5 (Gas)
G6 (Oil)
G7 (Wind)
Total
Overall Cost
Fig. 4.4 Solut
wn that the
ower and op
e outputs a
ghtly. The r
tages, the o
T
Solution of E
Power (
94.928
99.97
592.02
258.99
110.00
40.54
3.86
1200.00
ions of ED mo
solution of
peration cos
and schedu
reason is th
peration co
96
Table 4.8
ELD with Win
(MW)
86
710
273
938
097
473
21
000
31115
models with and
f ED with w
sts of some
uling costs
hat although
osts caused
nd Farm
Operation C
2967.33
4125.33
12290.3
5802.31
3189.00
1083.54
1657.76
31115.6
5.63
d without win
wind farm su
fuel units
of generato
h wind pow
by wind pr
Cost($)
3
3
5
1
0
4
6
63
nd farm.
ucceeds in re
(G1, G3, G
ors (G2, G6
wer generato
rediction err
reducing
G4, G5).
6) were
ors have
rrors are
97
expensive. For wind power generators, part of the load of high cost units
(G1, G3, G4, G5) is shifted to comparative low cost units (G2, G6). The
operational cost of the solution of ED with wind farms are highly increased
in comparison to the solution of ED without wind farms. In addition, the
wind power government subsidy is insignificant due to the low wind power
output.
The generated wind power in this case is 3.8621 MW which is far less than
the predicted wind power (45 MW), therefore the cost incurred by
overestimation will be applied. As a result the operator needs to purchase
more power from another source. Furthermore, the common ED model does
not take in account the emission issue. The incorporation of wind power in
simple ED problem is not an economic solution due to the wrong estimation
cost of wind power.
4.6.2. Case-II. CEED Model with and without Wind Farm
In this case study, the system load is 1600 MW and the system loss power is
assumed to be zero. The CEED model with and without wind farm are
performed on the test system and the simulation results are shown in Table
4.9, Table 4.10 and Fig. 4.5.
Table 4.9
Solution of CEED without Wind Farm
Unit Power (MW) Operation Cost($) Emission Cost($)
G1 (Coal) 95.5408 2986.10 3202.99
G2 (Coal) 20.7747 2820.83 3029.61
G3 (Coal) 598.7496 12414.63 8641.41
G4 (Gas) 509.7226 10924.30 852.97
G5 (Gas) 333.1363 7978.45 590.56
G6 (Oil) 42.0759 1131.62 495.65
Total 1600.0000 38255.93 16813.19
Overall Cost 55069.121
Th
le
w
op
cl
sh
Unit
G1 (Co
G2 (Co
G3 (Co
G4 (Ga
G5 (Ga
G6 (O
G7 (Wi
Tota
Overall
Fi
he system l
ess than the
wind power.
peration cos
lear that pa
hifted to the
S
t Pow
oal) 9
oal) 2
oal) 5
as) 5
as) 2
Oil) 4
ind) 7
al 16
Cost
ig. 4.5 Solutio
load is incre
e maximum
. The obje
sts and gre
art of the lo
e zero emiss
Ta
Solution of CE
wer (MW)
95.3455
21.3548
569.0520
507.6528
296.0316
40.0636
70.4998
600.0000
ons of CEED m
eased to 16
m capacity f
ective of C
enhouse ga
oad of high
sion wind p
98
able 4.10
EED with Win
Operation C
2980.
2835.
11708
10885
7159.
1068.
3230.
39868
56367
models with a
600 MW in
for both th
CEED is to
ases (CO2 a
hly polluted
power gener
nd Farm
Cost ($) Em
80
68
.60
.02
18
30
97
.55
7.12
and without w
this case. B
hermal units
o minimize
and NO2) em
d fuel fired
rator (G7).
mission Cost (
3200.81
3039.05
8404.66
849.54
530.05
474.47
0.00
16498.57
ind farm.
But the load
s and syste
e the total
mission cos
d units (G1~
Although th
($)
d is still
em with
system
sts. It is
~G6) is
he wind
99
power cost is expensive, emission costs were decreased in the solution of
CEED with wind farm. The reason is that the government wind power
subsidy is directly proportional to the output wind power. In this case, the
real generated wind power is 70.4998 MW which is larger than the
predicted wind power (45 MW). The underestimation situation will be
considered and the cost for not using all wind power available from wind
turbine should be applied. From Table 4.9 and Table 4.10, we can find that
the CEED model with wind farm reduces the emission cost dramatically in
comparison with CEED solution without wind power because of the zero
emission characteristic of wind energy. In Eq. (4.9), the government wind
power subsidy is directly proportional to the output of wind power. Thus,
the overall cost is acceptable from the standpoint of the wind farm operator.
Therefore, the results have shown that the proposed CEED with wind
energy gives a better emission solution efficiently and economically.
4.6.3. Case-III. Comparisons with Other Approaches
In order to evaluate the performance of the proposed method, GA, Immune
Algorithm (IA) [163], and PSO are employed in the case studies. For
comparison purposes, these algorithms are used directly to solve the CEED
problem with wind power. For the proposed SQP-PSO algorithm, the
population size is 100 and the maximum number of iterations is 3 for PSO.
Meanwhile, in order to make a fair comparison of the other approaches, we
fixed the same population size as 100 and tested them to reach maximum
iteration 100. The initial crossover and mutation rates for GA and IA were
all set as 80% and 5%, respectively. All the programs were run on a 2.66
GHz, Intel Core 2, with 4G RAM desktop. Table 4.11 shows the results out
of 50 runs with each method.
A comparison with other approaches is made to evaluate the proposed
algorithm which is shown in Table 4.11. As shown, we can conclude that
100
the proposed hybrid approach can greatly enhances the searching ability,
ensures quality of average solutions, saves computation time and also
efficiently manages the system constraints.
Table 4.11
Comparison of Different Approaches
Algorithm Best Solution ($) Average Solution ($) Average Time (s)
GA 57369.97 57916.20 13.28
IA 57180.98 57669.57 12.57
PSO 56714.06 57417.04 8.01
SQP+PSO 56367.12 56538.19 1.29
4.7. Conclusion
This chapter developed a hybrid method combining the SQP and PSO to
achieve faster and better optimization performance. The method was
successfully applied to solve the power system ED problem considering
GHGs emissions and wind power in an integrated CEED model, where the
valve-point effect is also taken into account. In the present work, the wind
speed distribution probability functions are applied in formulating the
optimization model to address the uncertainties involved. The proposed
hybrid method was applied to solve the CEED problem of a test system
involving 6 thermal units and 1 wind farm. The comparisons were made
between the classical ED and the proposed CEED model with and without
wind farms. The proposed CEED model with wind farms shows a better
performance in terms of less emission cost. In addition, the resultant overall
dispatching cost is also optimized considering the government subsidy.
Furthermore, the proposed hybrid optimization method was compared with
other optimization approaches for the studied cases. The simulation results
show that the hybrid method is better in terms of the speed and accuracy.
101
Compared to the classical PSO and other methods, it can be concluded that
the hybrid method greatly enhances the searching ability and efficiently
manages the system constraints, therefore providing a new and efficient tool
for the CEED problem.
102
This page is intentionally left blank
103
Chapter 5
Unit Commitment with Wind Power
Generation and Carbon Tax Considered
5.1. Introduction
In this chapter, we propose a Combined Unit Commitment and Emission
(CUCE) model integrating with wind energy, and carbon tax. This model
differs from existing works as it pays special attention to the wind-thermal
cooperation dispatch considering carbon tax. Wind generation, as a
renewable resource gradually becomes an integral part of smart grid
infrastructure. The introduction of a carbon tax can optimize carbon
emissions. In order to address the advanced dispatch strategy a hybrid
computational framework based on Sequential Quadratic Programming
(SQP) and Particle Swarm Optimization (PSO) is adopted.
5.2. Unit Commitment Considering Wind Power and
Carbon Tax
Power system generation scheduling problem can be divided into two sub-
problems, unit commitment (UC) and economic dispatch (ED). UC is an
104
optimization problem of determining operational schedules for generating
units in a power system with a number of constraints [127]–[129]. The main
objective of UC is to decide the on/off statuses of generators over the
scheduling period to meet the system load demand and reserve requirements
at the lowest cost. Basically, the UC outputs are on/off statuses on an hourly
basis for a given timeframe (e.g. 24 hours). In addition, UC schedule is
approached by meeting the system constraints such as ramp rate limits,
spinning reserve, as well as minimum up and down time limits. Suitable
improvements in the unit outputs scheduling can contribute to significant
cost savings.
In the past few decades, wind power has been one of the most important
renewable energy resources and has gained widespread attention. Wind
energy plays a major role in reducing the global greenhouse emissions,
easing the energy shortage throughout the entire world. An increasing
number of wind turbines have been built around the world with the
introduction of various emission reduction schemes [133], [164]. Significant
challenges to power system unit commitment and economic dispatch
problems exist because of the intermittent and stochastic characteristics of
wind resource [131], [165]. Effectively accommodating the wind
forecasting error is one of the major challenges. Variations of wind speed
directly impacts wind farms output, which influences reliable system
operations and difficulties in evaluating appropriate system reserve margins
to guarantee secure operation. As a result, high potential risks and additional
problems in power system operation and forecasting exist as a result of the
high penetration of wind power. A single predictor is not effective in
forecasting wind speed. Comprehensive studies have developed some kinds
of probabilistic optimization strategies [135], [146] for management of the
uncertainties of wind forecasts.
105
With the awareness of environmental pollution contributed by the
combustion of fossil fuels, building a low-carbon world has attracted
widespread attention. Many countries are encouraging the production of
clean energy in order to mitigate the greenhouse effects by introducing an
emission tax. The primary source of greenhouse gases (GHGs) is the
combustion of fossil fuels. Oil, gas and coal are the three major types of
regular fuels, which produce emissions represented by GHGs, such as NO2,
CO2 and SOx. In order to reduce the GHG emissions, a Combined Unit
Commitment and Emission (CUCE) framework is proposed in this chapter,
which can take into account fuel cost and emission tax altogether.
Obviously, increasing the tax on emissions will reduce power generated by
thermal units, which in turn lowers emissions. Moreover, the natural
economic forces will also help to catalyse the movement to greater energy
efficiency and use of renewable sources. Unit scheduling strategies
normally can offer quick solutions to improve the current situation of
system operation. At the same time, developing renewable energy is another
effective way to alleviate the depletion of energy sources, achieve a smart
grid vision and control GHG emissions [166-169].
In the literature, many researchers have shown great interest in
incorporating wind power in the analysis of UC. For instance, an approach
to evaluate the contribution that wind power can make to the load carrying
capability of a power generating system in an operating scenario was
studied in [170]. A novel UC formulation for a power system with
significant levels of wind generation was proposed in [171]. In [172], the
authors proposed an approach to evaluate the uncertainties of the balancing
capacity, ramping capability and ramp duration requirements. Furthermore,
various numerical optimization methods such as genetic algorithm (GA)
[173], [174], evolutionary programming (EP) [175], the quantum-inspired
evolutionary algorithms (QEA) [176], simulated annealing (SA) [176],
106
artificial neural networks (ANN) [177-179] and particle swarm optimization
(PSO) [180], have been employed to solve the UC problems. These heuristic
optimization methods are attractive because they can provide a fast and
reasonable solution, and they can deal with the constraints easily.
Nevertheless, when addressing large-scale optimization problems in finite
time, these heuristic algorithms do not always guarantee globally best
solutions. Therefore, more sophisticated computational tools are required. In
this chapter, we present an effective hybrid technique, which combines the
Sequential Quadratic Programming (SQP) and the Particle Swarm
Optimization (PSO) algorithms. The SQP seems to be one of the best
nonlinear-programming methods for constrained optimization problems. It
outperforms every other nonlinear-programming method in terms of
efficiency, accuracy and percentage of successful solutions, over a large
number of test problems. It guarantees a local optimum for nonconvex
optimization problems [156], [181]. Likewise, PSO [159] is one of the
modern heuristic algorithms and has gained lot of attention in various power
system applications. In this work, we exploit the advantages of both SQP
and PSO algorithms: SQP is used to obtain an initial solution and boundary
conditions which is then used in the PSO algorithm to obtain a final solution.
5.3. Probabilistic Modeling of Wind Power
The introduction of wind energy probabilistic has been discussed in chapter
4.
107
5.4. Mathematical Formulation of CUCE Problem
with Wind Power
The emission formulation of the UC model includes the wind power
generation described in this section. The aim of this UC model is to
minimize emission cost (carbon tax) and the operation costs (including wind
power cost, fuel cost) while satisfying the given constraints. The wind speed
distribution probability functions are applied in formulating the optimization
model to solve the wind uncertainties.
5.4.1. Objective Function
The objective function [182] is programmed to make the greenhouse gases
(CO2 and NO2) emission costs and total system operation costs minimum.
The overall objective function can be expressed as the sum of these two
terms
cos 1 2tf Cost Cost
(5.1)
Total system scheduling cost is the first term, i.e.,
, , ,
1
, , , ,
p e susdi t i t i t
i
w u o st j t j t j t j t
j
C C C
CostC C C C
(5.2)
where
, , ,pi t iC I i t p i t (5.3)
, , ( , )ei t tax iC I i t C EM p i t (5.4)
, , ,susdi tC SU i t SD i t (5.5)
108
, ,, ,wj t w j avC Q j t W j t (5.6)
, ,, , ,uj t u j avC Q j t E W j t w j t (5.7)
, ,, , ,oj t o j avC Q j t E w j t W j t (5.8)
, ,, ,sj t s j avC Q j t W j t (5.9)
where ,p
i tC is production cost of thermal unit i at time t . ,ei tC is emission cost
of thermal unit i at time t . ,susdi tC is start up and shut down cost of thermal unit
i at time t . ,wj tC is production cost of wind unit j at time t . ,
oj tC and ,
uj tC are
overestimation and underestimation cost of wind unit i at time t . ,sj tC is
government subsidy of wind unit j at time t . ,I i t is the on/off status of
thermal unit i at time t . ,Q j t is the on/off status of wind unit j at time t .
,SU i t and ,SD i t are startup and shunt down cost of thermal unit i at time
t . ,u j and ,o j are coefficient for not using all generated wind power due to
the underestimation and overestimation case. ,s j is government subsidy
coefficient of power generated by wind unit j . ,avW j t is power generated
by wind unit j at time t . The ,i p i t is the fuel cost function of thermal
generator i at time t, ,p i t is power generated by thermal unit i at time t
2, , ,, , ,i i t i t i tp i t a b p i t c p i t (5.10)
where , ,i i ia b c are production cost coefficients of thermal unit i in Table 5.2
, ,ws j tC is the wind power cost of the wind farm. If the wind farm is owned by
the system operator, this cost is zero which is considered in the case studies
of this chapter later on. , ,gs j tC is the wind power subsidy cost. As one of the
renewable energy subsidy projects, wind farms in many countries receive a
109
largely covert subsidy. When the predicted wind power is less than the
generated power the underestimation cost ,uj tC occurs, therefore the surplus
wind power cost compensated from the system operator. However, if
scheduled power is more than the actual wind power, the operator should
pay the overestimation cost ,oj tC and buy other alternate source power. When
we have determined the operation status and the time t, the on/off status and
the subscript t can be dropped. According to [135], the overestimation cost
will be assumed as follow,
, , 0
, 0
, ,f
j j
w
o j av o j j W
w w
o j j W Wo
E w j t W j t w w f w dw
w f w dw w f w dw
(5.11)
The cost of underestimation will take a similar form,
,
, ,
, ,
,
, ,r j
j
r j r j
j j
w
u j av u j j Ww
w w
u j jW Ww w
E W j t w j t w w f w dw
w f w dw w f w dw
(5.12)
,w j t is the predicted wind power generated by wind unit j at time t .The
wind power probability equations (4.4)-(4.7) address the equations (5.11)
and (5.12).
The emission function is second item, which can be written as,
,21
( )M
ei t tax i i
i
Cost C C EM p
(5.13)
2( ) ( )i i i i i i i iEM p ef f g p h P (5.14)
Carbon tax is an environmental tax that is levied on the carbon content of
fuels. As Australia is one of the world’s biggest greenhouse gas polluters,
due to its heavy reliance on coal for electricity, the Australian government
110
has proposed detailed carbon tax policies. In 2012, the Gillard government
has announced publicly that the 500 largest polluters in Australia will be
imposed a carbon tax at AUD 23/t of carbon emission, effective from 01
July, 2013. Through this carbon tax policy, the government encourages the
power industry to invest cleaner forms of power, like wind and solar energy.
Although it imposes great impacts on the traditional coal industry, for the
renewable energy sector this tax is a positive kick start. Equation (5.13) can
calculate the EMi(pi), which is the carbon emissions of thermal unit i, efi is
the fuel emission factors of CO2 for thermal generator i, coefficients of fuel
consumption are fi, gi and hi. All coefficients are listed in Table 5.2.
Equation (5.14) expresses the carbon emission function, Ctax is the given
carbon tax price which is determined by Australian regulations and markets.
5.4.2. System Constraints
System real power balance
1 1
, , , ,M N
avd lossi j
p t p t I i t p i t Q j t W j t
(5.15)
dp t is total system demand, lossp t is total transmission losses.
Unit generator limits
,max,min , iip p i t p (5.16)
,minip and ,maxip are minimum and maximum out power of thermal generators.
Wind power unit limits
,0 j r jw w (5.17)
jw is out power of wind generators.
111
System spinning reserve requirements
, ,s sI i t r i t R t (5.18)
,sr i t is the spinning reserve of thermal unit i at time t . sR t is the
spinning reserve requirement at time t .
Thermal unit minimum starting up/down times
1 1 0
1 1 0
on oni i i i
off offi i i i
X t T I t I t
X t T I t I t
(5.19)
oniX and off
iX are duration for which thermal unit i has remained on and off
time at time t . oniT and off
iT are minimum up/down time of thermal unit i .
Ramp rate limits
, , 1
, 1 ,
p i t p i t UR i
p i t p i t DR i
(5.20)
UR i and DR i are power output ramp-down/ramp up rate of thermal unit
i .
5.5. Hybrid Optimization Algorithm
The hybrid Algorithm has been introduced in section 4.5.
5.6. Numerical Simulation
In this chapter, a schematic representation of computational framework
which contains wind power forecast and stochastic unit commitment/
112
economic dispatch is shown in Fig. 5.1. The SQP-PSO is tested on a
modified IEEE 30-bus system. The benchmark system consists of 2 wind
farms, 41 branches, 21 loads and 6 thermal generators (including 1 oil-fired
unit, 2 gas-fired units, and 3 coal-fired units). Depending on the generated
wind power, the extra cost will be determined by the overestimation case or
the underestimation case.
5.6.1. Parameter Sensitivity Analysis
In the case study, the CUCE model with wind power was evaluated using
the historical wind speed dataset from a wind observation station in
Tasmania, Australia. The data was provided by the Australian Bureau of
Meteorology [162]. Here we assume that the wind speed data from a large
wind farm and use the data to estimate the generated wind power. The
corresponding Weibull distribution parameters and the wind speed
distribution frequency are presented in Fig. 5.2.
In total the wind farms have 20 Sinovel SL3000 3.0 MW wind turbines and
30 Vestas V90 3.0 MW wind turbines, located in two coherent geographic
areas. There are two kinds of wind turbines: three-blade rotor and pitch
regulated upwind wind turbines with active yaw. The wind turbine power
curve is linearized in the computation. The parameters of the associated
Weibull distribution factor and wind farm parameters are given in Table 5.1.
Locations of generators, generator limits, emission factors, fuel
consumption coefficients, and the fuel cost coefficients are shown in Table
5.2 and Table 5.3.
In this chapter, two of the most concerning GHG emissions, CO2 and NO2
are considered in the model. The emission characteristics of the units and
emission allowance prices are shown in the Tables 5.4 and Table 5.5.
113
Table5.1
Wind Power Factors
c k θ vin vout vr wr αw,j αu,j αo,j αs,j
5.5 1.89 0 4 25 16 3 0 60 20 10
Table 5.2
Generator Parameters
Note: (1) The coefficients of ai, bi, and ci, are in $, $/MW, $/MW2, and $/MW. (2) The coefficients of di, ei, and fi are in t, t/MW, and t/MW2 for coal/oil units, are in m3, m3/MW, and m3/MW2 for gas unit.
Table 5.3
Generator Constraint
Unit Production Cost Coefficients Fuel Consumption Coefficients
ai bi ci di ei fi G1 (Coal) 2200 12 0.003 45 0.3 0.00005 G2 (Coal) 2400 15 0.0020 50 0.25 0.00004 G3 (Coal) 6500 11 0.0022 90 0.14 0.00003 G4 (Gas) 930.5 20 0.00320 2430.5 55 0.009 G5 (Gas) 900 15 0.002 2000 0.212 0.007 G6 (Oil) 130.2 20.5 0.004125 1.248 0.334 0.0000342 G7 (Wind) 0 0 0 0 0 0 G8 (Wind) 0 0 0 0 0 0
Unit pmin
(MW) pmax
(MW)
Ramp Up Rate
(MW/h)
Ramp Down Rate
(MW/h)
Tup (h)
Tdn (h)
Start Up Cost
(Cold) ($)
Start Up Cost (Hot)
($)
Shut Down Cost
($)
Initial Status
G1 (Coal) 30 120 40 65 5 5 900 500 3200 -5
G2 (Coal) 20 110 30 50 4 3 780 360 3200 -6
G3 (Coal) 130 700 80 110 6 4 4800 2250 3200 1
G4 (Gas) 100 500 100 130 4 3 7000 3600 3200 1
G5 (Gas) 120 550 100 120 4 3 6600 3300 3200 -1
G6 (Oil) 45 210 55 62 3 4 4200 2230 3200 -1
G7 (Wind)
G8(Wind)
0
0
90
60
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
Fiig. 5.1 Wind p
Fig. 5.2 W
power uncerta
Wind speed di
114
ainties for com
istribution and
mputational fra
d Weibull fitti
amework.
ng.
115
Table 5.4
Emission Factors of Units
Emission Factor Coal (kg/kg) Gas (kg/m3) Oil (kg/kg)
efco2 3.1604 1.84 2.8523
efno2 1.29e-03 3.4e-04 3.3e-04
Table 5.5
Emission Price
Fuel CO2 ($/t) NO2 ($/kg)
Price 2.0 4.5
5.6.2. Case Studies
In the following case studies, the system loss power is assumed to be zero.
The CUCE model with and without wind farm are performed on the test
system and the simulation results are shown. The load curve during 24 hours
is shown in Fig.5.3. The predicted wind power and system demand for 24
hours are given in Table 5.6. Table 5.7 shows the schedules of 7 units for 24
hours by minimizing the cost. In this table, the value of “1” represents an
on-line state of each unit, and the value of “0” represents an off-line state of
each unit. From this table, it is clear that wind turbines were not scheduled
for 20 hours. The reason is that the wind power generating cost is
comparatively high. Fig. 5.4 presents the scheduled wind power generation.
The blue line presents the forecasted wind power for 24 hours and the red
line presents the scheduled wind power.
The objective of CUCE is to minimize the total system operation costs and
greenhouse gas (CO2 and NO2) emission costs. It is clear that part of the load
of highly polluted fuel fired units (G1-G6) is shifted to the zero emission
wind power generator (G7-G8). Although the wind power cost is expensive,
emission cost was decreased in the solution of CUCE with wind farms. The
re
th
pr
th
G
an
fin
dr
be
th
th
st
th
ef
eason is that
he output wi
roposed UC
he generator
G8) replace t
nd less pollu
nd that the
ramatically
ecause of th
he governme
he wind po
andpoint of
he proposed
fficiently an
t the govern
ind power. I
C model, the
rs will be im
the highly p
uted gas gen
e CUCE m
in compa
he zero emis
ent wind po
ower outpu
f the wind f
d CUCE w
nd economic
F
nment wind
It can be see
e carbon em
mpacted. Zer
polluted coa
nerators (G4
model with
arison with
ssion charac
ower subsid
ut. Thus, th
farm operato
with wind e
cally.
Fig. 5.3 Foreca
116
power subs
en that if th
mission is r
ro emission
al fired unit
4). From Ta
wind farm
CUCE so
cteristic of w
dy is shown
he overall
or. Therefor
energy give
asted system d
sidy is direc
he carbon ta
educed, and
polluted wi
ts (G1-G3),
able 5.8 and
ms reduces
olution wit
wind energy
n to be direc
cost is ac
re, the resul
es a better
demand.
ctly proport
x is include
d the output
ind generato
oil generat
d Table 5.9,
the emissi
thout wind
y. In equatio
ctly proport
ceptable fr
lts have sho
emission s
tional to
ed in the
uts of all
ors (G7,
tor (G6)
, we can
ion cost
d power
on (5.2),
tional to
rom the
own that
solution
Unit G1(Coal) G2 (Coal) G3 (Coal) G4 (Gas) G5 (Gas) G6 (Oil) G7 (Wind)
TimeWind
DemanTimeWind
Deman
01 02 030 0 0 0 0 0 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0
e 01 d 44 nd 609 e 13 d 59 nd 1600
Fig. 5
3 04 05 00 0 00 0 01 1 01 1 10 0 10 0 10 0 1
02 0373 69507 4314 1558 43
1633 155
5.4 Forecasted
Forecasted W
06 07 08 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 0 0 0
3 04 9 76 3 397
5 16 .2 27 59 1478
117
d wind power
Table 5
Wind Farm Pow
Table 5
Generator S
09 10 11 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 0 0 1
05 0691 84388 41717 183 6
1503 151
vs scheduled
5.6
wer and Syste
5.7
Schedules
12 13 14 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0
6 07 4 92 7 569
8 19 7
9 1532
wind power.
m Demand
15 16 17 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0
08 09 86 13 741 927 20 21 11 7
1463 1293
18 19 200 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0
10 44
1109 22 43
3 1081
0 21 22 20 0 00 0 01 1 11 1 11 1 10 0 00 0 0
11 12 65 62.9
1359 151023 24 54 61 888 712
3 24 0 0 1 1 1 0 0
118
Table 5.8
Solution of CUCE without Wind Farm
Unit Operation Cost ($) Emission Cost($)
G1 (Coal) 2986.10 3203.99
G2 (Coal) 2820.83 3029.61
G3 (Coal) 12414.63 8641.41
G4 (Gas) 10924.30 852.97
G5 (Gas) 7978.45 590.56
G6 (Oil) 1131.62 495.65
Total 38255.93 16814.19
Table 5.9
Solution of CUCE with Wind Farm
Unit Operation Cost ($) Emission Cost($)
G1 (Coal) 2980.80 3200.81
G2 (Coal) 2835.68 3001.05
G3 (Coal) 11317.27 8404.66
G4 (Gas) 10479.25 849.54
G5 (Gas) 6214.13 530.05
G6 (Oil) 940.74 474.47
G7,G8 (Wind) 3230.97 0
Total 37998.84 16498.58
5.6.3. Comparisons
For evaluating the performance of the proposed method, GA, QEA, PSO
and SQP-PSO are employed in the case study, which are shown in table
5.10. For comparison purposes, these algorithms are used directly to solve
the CUCE problem with wind power. For the proposed SQP-PSO algorithm,
the population size is 100 and the maximum number of iterations is 3 for
PSO. Meanwhile, for making a fair comparison, fixing the same population
size as 100 and tested them to reach maximum iteration count of 100. The
initial crossover and mutation rates for GA were all set as 80 (%). All the
programs were run on a 3.0 GHz, Intel Core 5, with 4G RAM desktop. The
119
CPU time to obtain the solution for the best, the worst and the average
results for different text algorithms are shown in Table 5.10. Accordingly,
for the comparison result, the SQP-PSO has shown the superiority to the
existing methods.
Table 5.10
Comparison of Different Approaches
Method Execution Time (s) Best Cost ($) Mean Cost ($) Worst Cost ($)
GA 13.28 57369.97 57916.20 58562.40
PSO 12.57 57180.98 57669.57 58231.16
QEA 8.01 56714.06 57417.04 58067.03
SQP-PSO 7.29 56367.12 56538.19 56721.31
5.7. Conclusion
This chapter developed a hybrid method combining the SQP and PSO to
achieve faster and better performance optimization. The method has been
successfully applied to solve the power system UC problem considering
GHG emissions and wind power in an integrated CUCE model. Wind power
has shown to have impacted the smart grid, especially when most wind
energy injects into the transmission grid. Carbon tax is an important factor
that affects the operation of smart grid and hence the cost, therefore it also
can effectively reduce the GHG emissions. To address the uncertainties in
wind power production, the wind speed probability functions are applied in
formulating the optimization model. The proposed hybrid method has been
applied to address the CUCE problem of 2 wind farms and 6 thermal units
system. Comparisons have been made by the proposed CUCE model with
and without wind farm. The proposed CUCE model with wind farm shows a
better performance in terms of less emission cost. Due to the uncertainties of
wind energy, it is crucial to improve power system forecasting accuracy. In
addition, the resultant overall unit scheduling cost is optimized considering
120
the government subsidy. Furthermore, the proposed hybrid optimization
algorithm has been compared with other methods for the studied cases. The
simulation results show that the hybrid method is better in terms of speed and
accuracy. According to the comparison, the hybrid approach efficiently
manages the system constraints and greatly enhances the searching ability,
which offers a convenient and efficient tool for the UC problem with carbon
tax and wind power generation.
121
Chapter 6
Wind-Thermal Generation Scheduling
Optimization Integrating ESS with Carbon
Emission
6.1. Nomenclature
, ,i i ia b c Production cost coefficients of thermal unit i .
, ,i i id e f Coefficients of fuel consumption for thermal unit i .
,i j tB Susceptance between Bus i and Bus j.
,p
i tC Production cost of thermal unit i at time t .
,wj tC Production cost of wind unit j at time t .
,sj tC Government subsidy of wind unit j at time t .
,susdi tC Start up and shunt down cost of thermal unit i at time t .
maxC Maximum energy capacity of the BESS.
minC Minimum energy capacity of the BESS.
i Cost function of thermal unit i .
,s j Government subsidy coefficient of power generated by wind unit.
,w j Production cost coefficient of wind unit j .
,I i t The on/off status of thermal unit i at time t .
M Number of thermal units.
N Number of wind units.
122
,p i t Actual power generated by thermal unit i at time t .
BESSp t BESS charging/discharging power.
dp t Total system demand.
( )gp t Total power generated by thermal units.
,i j tp Real Power on a transmission line.
,maxi jp Maximum transmission capacity of a line.
,mini jp Minimum transmission capacity of a line.
lossp t Total transmission losses.
( )wp t Total power generated by wind farm.
,Q j t The on/off status of wind unit j at time t .
,sr i t The spinning reserve of thermal unit i at time t .
sR t The spinning reserve requirement at time t .
( )SOC t State of charge of BESS at time t .
,SU i t The startup cost of thermal unit i at time t .
,SD i t The shunt down cost of thermal unit i at time t .
,r jw Rated wind power from wind turbine unit j .
,w j t Predicted wind power generated by wind unit j at time t .
,avW j t Actual power generated by wind unit j at time t.
ch Storage battery charge efficiency.
dis Storage battery discharge efficiency.
, ,,i t j t Bus angle.
6.2. Introduction
In recent years, fossil fuel combustion has caused significant environmental
pollution, creating a need to establish a low-carbon world, which has
attracted extensive attention. Many countries are implementing policies to
produce clean energy in order to mitigate the greenhouse effects by
123
introducing an emission tax. The combustion of fossil fuels are the main
source of greenhouse gases (GHGs). The three major kinds of conventional
fuels are oil, gas and coal, which produce emissions such as, NO2, CO2 and
SF6. For reducing air pollution and building a clean energy environment,
carbon emission limitation and high efficiency power scheduling strategies
are two important factors that should be considered.
Wind power is one of the most important renewable energy resources
playing a major role in reducing the global greenhouse emissions, easing the
energy shortage throughout the entire world. An increasing number of wind
turbines have been built around the world with the introduction of various
emission reduction schemes [133], [164]. The intermittency and uncertainty
of wind makes the dispatch of wind energy a difficult task [183]. High wind
power penetration could impact the system security and reliability of the
power grid [184]. To reduce fluctuation of the wind energy output, a battery
energy storage system (BESS) is integrated into the renewable energy
generation system [185]. Wind farms combined with battery energy storage
can enhance system reliability, power availability and quality, and
operational efficiency [186]. Fig. 6.1 illustrates grid-connected wind farm-
BESS schematic diagram. The hybrid model is an integration system
including thermal generators, wind farms and storage systems. The BESS is
connected to the system at the point of common coupling and is
charged/discharged through a power converter to smooth the net power
injected to system. The energy storage facility in this model is located at the
wind farm and connected to the system through a transmission line. In this
case, only electric power generated by wind unit is stored for future use, and
the wind turbine generators have the priority to serve the system load. Wind
energy can be stored in the device when the system load is low. On the other
hand, during peak load interval, the stored electric energy can be transferred
into the power system [168]. Wind power combined with BESS can be
124
utilised to shave the peak load and smooth out the intermittent power of the
wind [188, 189].
Fig. 6.1 Structure of wind power generation system integrating BESS.
Unit commitment (UC) is an optimization problem of determining
operational schedules for generating units in a power system with a number
of constraints [127, 128]. The main objective of UC is to decide the on/off
statuses of generators over the scheduling period to meet the system load
demand and reserve requirements at the lowest cost. Basically, the UC
outputs are on/off statuses on an hourly basis for a given timeframe (e.g. 24
hours). In addition, UC schedule is approached by meeting the system
constraints such as ramp rate limits, spinning reserve, as well as minimum
up and down time limits. Suitable improvements in the unit outputs
scheduling can contribute to significant cost savings and reduction of GHG
emissions. In the literature, many researchers have shown great interest in
incorporating wind power in the analysis of scheduling strategies. In [189-
192] authors introduce wind power into scheduling operation without
energy storage system (ESS). In [193, 194] the authors combine wind power
with ESS without considering the carbon emission limit.
chBESSp
disBESSp
125
In this chapter, a combined unit commitment and emission (CUCE)
framework is proposed to reduce the GHG emissions, which can take into
account the fuel cost and emission tax altogether. Obviously, increasing the
tax on emissions will reduce power generated by thermal units, which in
turn lowers emissions. Wind power penetration of power system can reduce
the environmental pollution and minimize the gas emissions. However, the
intermittent and uncertainty characteristics of wind could impact the
stability and security of power system when the wind power integrates into
the power grid. To mitigate the negative impacts of wind energy for the
power system, a BESS is incorporated with the CUCE model. In this
chapter, a hybrid UC model is used to precisely calculate the operating cost
of generation. The objective function of the proposed model is designed to
accomplish all of these tasks while taking into account the non-idealities of
a real system. In many instances these non-idealities are high order and
nonlinear, the fuel cost is represented by a quadratic function. When the
total limits of CO2 emission over the scheduling horizon are imposed, the
CUCE problem becomes much more difficult to solve because the region of
feasible solution becomes much smaller. We developed a hybrid algorithm
which combines the sequential quadratic programming (SQP) and the
particle swarm optimization (PSO) to address the economic dispatch
emission problem, and the hybrid algorithm has been shown to have
excellent performance in non-idealities are high order and nonlinear
problem. In this chapter, we present the effective hybrid technique to
address the CUCE combined with BESS problem.
6.3. Wind Power Forecasting and BESS
In this section, the model of wind power forecasting and battery are
introduced. In renewable power management, obtaining a good forecast is a
126
significant procedure for carrying out the relevant optimization. The BESS
plays an important role of mitigating the variability of wind in the hybrid
power system.
6.3.1. Wind Power Prediction
One of the major challenges of wind generation is how to effectively
accommodate the wind forecasting errors. Because variations of wind speed
directly influences the power output of wind farms, which then causes
difficulties in estimating suitable system reserve margins to ensure secure
and reliable system operations. As a consequence, high penetration of wind
power also causes high potential risks and more difficulties in power system
operation.
In this chapter, a scenario reduction method calculates the singularly
forecasted wind power. Fig. 6.2 illustrates the process of the predicting
method [195]. The single wind power forecast was calculated using the
singular spectrum analysis (SSA) technique. Given the forecast errors of the
aggregated outputs of the wind farms, lower and upper limits were
determined for each time step where the effective wind power output is
most likely to fall within this range. Then a large number of random
forecasts were generated within the lower and upper bounds using Monte
Carlo simulation. It is worth noting that using the wider forecast error
bounds improves the effectiveness of the stochastic programming methods.
Limited information about the possible wind power output is provided by a
small amount of wind power forecast scenarios. It is very difficult to
numerically obtain a solution for a stochastic optimization problem using
the large number of wind power forecast scenarios. On the other hand, a
small number of wind power forecast scenarios provide less information
about the possible wind power outturns. For addressing the above issues
127
many wind power forecast scenarios are generated, followed by a scenario
reduction algorithm to merge the forecast scenarios that are very close
together.
Fig. 6.2 Algorithm for producing probabilistic wind power forecast.
6.3.2. Battery Energy Storage System (BESS)
6.3.2.1. Selection
Recently, combining an energy storage system (ESS) together with a wind
power has been proposed in order to provide economic and technical
benefits to power systems [196, 197]. Different energy storage systems
(ESS) technologies such as pumped hydroelectric, compressed air, super-
capacitors, magnetic storage, electrical batteries, and flywheels are proposed.
128
For large-scale electricity storage, large fuel cells seem to be appropriated
[198].
Technical factor, operating process, and economics should be considered for
the selection of a suitable battery type. The important battery parameters,
which may affect power system operation [200], are minimum and
maximum storage capacity (MWh), charging and discharging rate
(MWh/hour), state of charge (SOC) and depth of discharge (DOD). Battery
storage can be controlled to charge or discharge in constant or variable rate
based on system operation requirements. The merit of BESS is the fast rate
of charging/discharging [199].
6.3.2.2. Operation
Practically, storage size equivalent to 15%–25% of the wind farm capacity
is suggested to realize an effective hourly dispatch [200]. In day-ahead
scheduling, wind farm operators predict wind power for the next hour.
When the forecast value is smaller than actual power output (underestimate),
the excess energy can be stored in energy storage. If the actual value is less
than the predicted value (overestimate), energy from the storage can supply
to meet the system load demand [201]. In systems with large wind
penetration, the system operator may impose a limit on the wind power and
BESS. The charge and discharge of the BESS is subject to stored energy
limits. The minimum and maximum energy stored in the battery bank are
specified. The SOC lower limit is 20% and upper limit is 80% of battery full
capacity respectively. For instance, we assume the initial state of the BESS
is SOC lower limit, and then the ESS will continue in the charging state to
store energy until it reaches the upper limit state, where the charging process
is finished. In this case, the capacity of the ESS is around 80% of its full
capacity. Once at 80% capacity the BESS discharges until it reaches the
lower limit of SOC [202].
129
With regard to the BESS, an important design consideration is the finite
number of charge–discharge cycles the BESS can undertake over its useful
lifetime [168]. The deep charge/discharge cycles have been minimized in
order to extend the lifetime of the battery. The number of such cycles, called
the cycle life, depends on the depth of discharge, the BESS has to undergo.
In this aspect, deep-cycle batteries have been suggested as suitable level, too
deep a discharge (SOC> 80 % or <20 %) should be avoided as it leads to
permanent physical damage to the BESS and an exceedingly low cycle life
During the charging and discharging procedure only one state is in process,
such that charging and discharging cannot happen at the same time.
6.4. Problem Formulation
The emission formulation of UC model including wind power and battery
energy storage system are described in this section. The aim of this UC
model is to minimize emission cost (carbon tax) and the operation costs
(including wind power cost, fuel cost, battery cost) while satisfying the
given constraints. The BESS is embedded into the model to shave the load
peak and address the wind intermittency.
6.4.1. Objective Function
cos , , ,1 1 1
, , ,1 1 1
.M M M
p e susdt i t i t i t
i i i
N N Nw s BESSj t j t i t
j j i
Minimize F C C C
C C C
(6.1)
where
, , ,pi t iC I i t p i t (6.2)
130
, , ,susdi tC SU i t SD i t (6.3)
, ,, ,wavj t w jC Q j t W j t (6.4)
, ,, ,savj t s jC Q j t W j t (6.5)
The ,i p i t is the fuel cost function of thermal generator i at time t , which
contains higher order nonlinearity and discontinuity due to valve point effects
[16]. It can be defined as
2, , , ,min, , , sin ( )i i t i t i t i i iip i t a b p i t c p i t d e p p (6.6)
,wj tC is the wind power cost of the wind farm. This term is zero when the wind
farm is owned by the system operator.
The ,1
Mei t
i
C item is the emission function and it can be represented as
,1
( )M
ei t i iT a x
iC C E M p
(6.7)
2( ) ( )i i i i i i i iE M p e f f g p h P (6.8)
equation (6.8) can calculate the ( )i iEM p , which is the carbon emissions of
thermal unit i, ief is the fuel emission factors of CO2 for thermal generator i.
if , ig and ih are coefficients of fuel consumption. TaxC is the market carbon
tax price.
The last component ,1
NBESSi t
i
C is the operation cost of BESS and can be
indicated as
,1
( ) ( )N
BESS dis chi t BESS BESS BESS
i
C p t p t
(6.9)
131
where BESS is coefficient of BESS consumption, in this work the BESS is
$ 0.1 / kW h .
6.4.2. System Constraints
System real power balance is given by
( ) ( ) ( ) ( ) ( )d loss g w BESSp t p t p t p t p t . (6.10)
Network Constraints
, , , ,i j t i j t i t j tp B (6.11)
, ,max,mini j i j t i jp p p . (6.12)
Unit generator limits are
,max,min , iip p i t p . (6.13)
Wind power unit limits are
,0 j r jw w . (6.14)
System spinning reserve requirements are given by
( , ) ( , ) ( )s sI i t r i t R t . (6.15)
Thermal unit minimum starting up/down times satisfy
1 1 0
1 1 0
on oni i i i
off offi i i i
X t T I t I t
X t T I t I t
. (6.16)
Ramp rate limits are presented as
, , 1
, 1 ,
p i t p i t UR i
p i t p i t DR i
. (6.17)
132
BESS charge/discharge power limits are
,max
,max
0 ( )
0 ( )
ch chBESS BESS
dis disBESS BESS
p t p
p t p
. (6.18)
where , maxchBESSp and ,maxdis
BESSp are the maximum charging and discharging rate of
battery. We assume the battery cannot charge and discharge at the same
time.
( ) ( )=0dis chBESS BESSp t p t . (6.19)
BESS storage constraints are
( )L USOC SOC t SOC (6.20)
max( ) ( ) /BESSSOC t C t C . (6.21)
where LSOC and USOC are the lower and upper SOC of the battery
( ) ( ) ( )ch disiniBESS BESS BESSch disC t C p t p t (6.22)
and iniC is initial value of capacity of the BESS.
6.5. Hybrid Optimization Algorithm
The hybrid optimization algorithm which combines SQP and PSO together
is presented in chapter 4. This advanced algorithm has been applied to solve
the combined economic and emission dispatch problem [182]. Due to the
excellent performance of the hybrid technique for nonlinear problem, the
composited PSO and SQP algorithm will be used to solve the CUCE
combined with BESS problem.
133
6.6. Simulation Result and Discussion
6.6.1 Parameter Analysis
In this chapter, The SQP-PSO is tested on a modified IEEE 30-bus system.
The test system and two wind farms are represented in [137]. [160] lists the
generator limits, emission factors, fuel cost coefficients, and fuel
consumption coefficients. The carbon tax is fixed as AUD 21/t. The wind
farm and storage system are owned by the system operator. We select
battery storage size at 20% of the wind farm capacity. The BESS model
composed of NaS batteries [36], the characteristics are listed in Table 6.1.
The operation cost of BESS includes storage system installation cost and
battery degradation fee. All storage costs can calculate by coefficient cost of
BESS BESS . A wind observation station provides the historical wind speed
in Tasmania, Australia [161]. The 24-hour system load and forecasted wind
power are presented in Table 6.2. In this simulation test the transmission
loss is assumed to zero.
6.6.2. Case Studies
6.6.2.1. CUCE Result without Wind Farms and BESS
In this case study, the proposed CUCE problem is solved to determine the
commitment of the units. Results from Table 6.3 show that after the carbon
tax embeds in to UC model, the generating outputs units will be impacted,
and the relevant cost is changed. A portion of the highly polluted coal fired
units (G1–G3) for load is shifted to less polluted gas generators (G4, G5),
and oil generator (G6). The daily generation cost is $555,935.5.
134
Table 6.1
BESS Parameter
Quantity Value Quantity Value
maxC (MWh) 50 , maxchBESSp
(MWh/hour)
10
minC (MWh) 0 ,maxdisBESSp
(MWh/hour)
10
iniC (MWh) 10 ch 0.83
LSOC 20% dis 0.83
USOC 80% BESS 0.1/ kW h
Table 6.2
Load Demand and Forecasted Wind Power
Time Wind(MW) Load(MW) Time Wind(MW) Load(MW)
01 55 609 13 106 1600
02 85 507 14 97 1633
03 94 433 15 82 1559
04 101 397 16 34 1478
05 100 388 17 17 1503
06 105 417 18 20 1519
07 125 569 19 21 1532
08 128 741 20 38 1463
09 100 927 21 60 1293
10 88 1109 22 70 1018
11 102 1359 23 68 888
12 108 1510 24 62 712
6.6.2.2. CUCE Result with Wind Farms and no BESS
The CUCE model combines with the wind farm to minimize the greenhouse
135
gas emissions and total system operation costs. From Table 6.4, it can be
seen that parts of load shift to clean power wind generators (wind farm 1
and 2). The power generation from highly polluting fuel fired units (G1-G6)
is shown to be decreased. Zero emission wind generators replace the highly
polluting coal fired units (G1-G3), oil generator (G6) and less polluted gas
generators (G4). Due to the government subsidy of wind energy is directly
proportional to the wind farm generation power. Thus, the overall cost is
acceptable from a standpoint of wind farm operator. Even though the wind
power cost is expensive, emission cost is decreased in the solution of CUCE
with wind farms. From Table 6.3 and Table 6.4, it can be seen that the
CUCE model with wind farms reduces the emission cost remarkably in
comparison with CUCE solution without wind energy as a result of the wind
energy zero emission characteristic. Therefore, according to the results the
proposed CUCE with wind energy gives an efficient emission solution.
However, the wind power intermittency and uncertainty sometimes make
the dispatch of wind energy become difficult. As shown in Fig. 6.3, the
forecasted wind power is compared with the real wind power output. In
parts of schedule time intervals, the wind speed exceeds the cut-out speed or
below the cut-in speed of the wind generator, so there is no wind power
injecting into the load system. For other case, during the peak load period
while the wind speed in low (not below the cut-in speed), the system needs a
large amount of input power, but the wind farms cannot supply enough wind
power. On the other hand, the wind speed is high (not exceeding cut-out
speed) while the load demand is low. High wind power penetration could
impact the system security and reliability of the power grid.
136
Fig. 6.3 Predicted and real wind power.
Table 6.3
Generators Schedule of CUCE
Time G1 G2 G3 G4 G5 G6 Power (MW) Emission ($) Production ($) Cost ($)
1 0 0 355 254 0 0 609 3949.365 9506.49 13455.86
2 0 0 355 152 0 0 507 3287.895 7914.27 11202.17
3 0 0 300 133 0 0 433 2808.005 6759.13 9567.135
4 0 0 277 120 0 0 397 2574.545 6197.17 8771.715
5 0 0 268 120 0 0 388 2516.18 6056.68 8572.86
6 0 0 287 130 0 0 417 2704.245 6509.37 9213.615
7 0 0 355 214 0 0 569 3689.965 8882.09 12572.06
8 0 0 355 250 136 0 741 4805.385 11567.01 16372.4
9 0 0 355 320 252 0 927 6011.595 14470.47 20482.07
`10 0 134 355 320 300 0 1109 7191.865 17311.49 24503.36
11 84 300 355 320 300 0 1359 8813.115 21213.99 30027.11
12 125 300 355 320 300 110 1510 9792.35 23571.1 33363.45
13 185 300 355 320 300 140 1600 10376 24976 35352
14 203 300 355 320 300 155 1633 10590.01 25491.13 36081.14
15 154 300 355 320 300 130 1559 10110.12 24335.99 34446.11
16 103 300 355 320 300 100 1478 9584.83 23071.58 32656.41
17 128 300 355 320 300 100 1503 9746.955 23461.83 33208.79
18 144 300 355 320 300 100 1519 9850.715 23711.59 33562.31
19 147 300 355 320 300 110 1532 9935.02 23914.52 33849.54
20 100 300 355 320 300 88 1463 9487.555 22837.43 32324.99
21 68 300 355 320 250 0 1293 8385.105 20183.73 28568.84
22 0 143 355 320 200 0 1018 6601.73 15890.98 22492.71
23 0 83 355 250 150 0 888 5758.68 13861.68 19620.36
24 0 30 355 200 127 0 712 4617.32 11114.32 15731.64
Total 163,125.5 392810 555935.5
0
20
40
60
80
100
120
140
1 2 3 4 5 6 7 8 9 101112131415161718192021222324
Win
d P
ow
er G
ener
ated
(M
W)
Time (Hour)
forecasted
actual
137
Table 6.4
Generators Schedule of CUCE with Wind Farm
Time G1 G2 G3 G4 G5 G6 Wind Power (MW) Emission ($) Production ($) Cost ($)
1 0 0 337 217 0 0 55 609 3507.84 10072.86 13580.7
2 0 0 291 130 0 0 86 507 2920.32 8385.78 11306.1
3 0 0 219 120 0 0 94 433 2494.08 7161.82 9655.9
4 0 0 176 120 0 0 101 397 2286.72 6566.38 8853.1
5 0 0 168 120 0 0 100 388 2234.88 6417.52 8652.4
6 0 0 192 120 0 0 105 417 2401.92 6897.18 9299.1
7 0 0 264 180 0 0 125 569 3277.44 9411.26 12688.7
8 0 0 288 215 110 0 128 741 4268.16 12256.14 16524.3
9 0 0 345 282 200 0 100 927 5339.52 15332.58 20672.1
10 0 169 330 272 250 0 88 1109 6387.84 18342.86 24730.7
11 87 300 310 260 300 0 102 1359 7827.84 22477.86 30305.7
12 135 300 300 259 300 110 108 1510 8697.6 24975.4 33673
13 190 300 292 262 300 150 106 1600 9216 26464 35680
14 205 300 300 276 300 155 97 1633 9406.08 27009.82 36415.9
15 170 300 300 277 300 130 82 1559 8979.84 25785.86 34765.7
16 128 300 310 306 300 100 34 1478 8513.28 24446.12 32959.4
17 170 300 310 306 300 100 17 1503 8657.28 24859.62 33516.9
18 165 300 320 314 300 100 20 1519 8749.44 25124.26 33873.7
19 152 300 335 315 300 110 20 1532 8824.32 25339.28 34163.6
20 100 300 325 312 300 88 38 1463 8426.88 24198.02 32624.9
21 70 300 310 303 250 0 60 1293 7447.68 21386.22 28833.9
22 0 145 303 300 200 0 70 1018 5863.68 16837.72 22701.4
23 0 85 301 254 180 0 68 888 5114.88 14687.52 19802.4
24 0 30 300 195 125 0 62 712 4101.12 11776.48 15877.6
Total 144,944.6 416,212.5 561,157.1
6.6.2.3. CUCE Result with Wind Farms and BESS
In this case, the CUCE model combines with the wind farms and BESS. We
set the BESS initial value of capacity at 20% of full capacity. BESS can
store wind at off-peak hours when demand is low but the actual wind power
out is greater than predicted. The upper limit of BESS capacity is 80% of
the full capacity [201]. When the BESS stores the wind energy, the battery
is charging until it reaches its upper limit SOC. During the load peak
interval, the BESS injects the stored energy into the load until it reaches the
low limit of SOC. The forecasted wind power, actual output power from
wind farm and BESS charging/discharging operation are shown in Fig. 6.4.
138
From this figure, since the BESS stores the wind energy when the actual
value is greater than forecasted value, the wind curtailment is lower than
that in Case 2 (CUCE with wind farm no BESS). However, the BESS
operation can reduce wind curtailment. From Table 6.5, it can be seen the
output of expensive generators G1 and G2 are less than that in Case 2. The
advanced model reduces operation and emission costs dramatically. Fig. 6.3
and Fig. 6.4 indicate the CUCE model combined wind farm without/with
BESS charging/discharging operation respectively. Those two profiles
illustrate that BBES increases the value of electricity generated from wind
resource by making it available regardless of when it was generated. BESS
can provide peak-shaving capability or reduce peak demand by storing
energy during off-peak hours from the grid or wind resource and release
during peak hours. Lower generation operation emission cost is obtained in
this case. These results show the lower cost and emissions of using BESS
for supplying the load in the system.
After the BESS is adopted in the CUCE with wind farm model, the thermal
generators output will be affected as well. As discussed previously, the
battery stored the wind energy in the time of off-peak period while wind
generation is high, releasing at high load demand interval. At the same time,
the commitment of thermal units is altered. From Fig. 6.5, it is clear that the
output of all thermal units is reduced at the peak load time intervals. On the
contrary, during the low load demand the thermal generators output is
increased. The generated power from G1 to G6 changes more gradually than
CUCE wind farm model without BESS. It demonstrates that the BESS can
shave the system peak load and improve the security and stability of the
power system.
Fig. 6.6 illustrates that the BESS has undergone two entire charge/discharge
cycles, which cost $ 12,000. Note that by limiting the SOC to be between 20%
and 80%, the deep charge/discharge cycles have been minimized in order to
139
extend the lifetime of the battery. The cost of BESS is entirely covered by
the economic benefits obtainable from the dispatch ability, i.e. the reduction
of the production cost. With the BESS operation SOC limits, power output
has a smooth trend.
The battery system can store the overestimated wind power in off-peak, and
supply it during peak hours. The wind energy curtailment will be reduced,
however, high wind penetration would affect the power system security and
reliability. As shown in Fig. 6.7, in CUCE combined wind power with
BESS model, wind power penetration and curtailment will be improved,
which mitigates the volatility of the wind.
Fig. 6.4 BESS operation with forecasted and actual wind power.
Fig. 6.5 Thermal units output with/without BESS operation.
-20
0
20
40
60
80
100
120
140
1 2 3 4 5 6 7 8 9 101112131415161718192021222324
Win
d Po
wer
Gen
erat
ed (M
W)
Time (Hour)
BESS
forecasted
actual
0
200
400
600
800
1000
1200
1400
1600
1800
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
The
rmal
uni
ts G
ener
ated
(MW
)
Time (Hour)
without BESS with BESS
140
Table 6.5
Generators Schedule of CUCE with Wind Farm and BESS
Time G1 G2 G3 G4 G5 G6 Wind BESS Power (MW) Emission ($) Production ($) Cost ($) 1 0 0 337 214 0 0 58 -3 609 3124.17 9963.24 13087.41 2 0 0 291 120 0 0 96 -10 507 2600.91 8294.52 10895.43 3 0 0 212 120 0 0 101 -7 433 2221.29 7083.88 9305.17 4 0 0 169 120 0 0 108 -7 397 2036.61 6494.92 8531.53 5 0 0 166 120 0 0 102 -2 388 1990.44 6347.68 8338.12 6 0 0 191 120 0 0 106 -1 417 2139.21 6822.12 8961.33 7 0 0 265 180 0 0 124 1 569 2918.97 9308.84 12227.81 8 0 0 295 215 110 0 121 7 741 3801.33 12122.76 15924.09 9 0 0 345 284 200 0 98 2 927 4755.51 15165.72 19921.23 10 0 139 340 294 250 0 86 2 1109 5689.17 18143.24 23832.41 11 85 300 315 270 300 0 89 10 1359 6971.67 22233.24 29204.91 12 135 300 315 270 300 110 100 8 1510 7746.3 24703.6 32449.9 13 190 300 292 270 300 150 98 0 1600 8208 26176 34384 14 205 300 300 278 300 155 95 0 1633 8377.29 26715.88 35093.17 15 170 300 295 272 300 130 92 -10 1559 7997.67 25505.24 33502.91 16 105 300 333 300 300 100 40 -6 1478 7582.14 24180.08 31762.22 17 130 300 335 316 300 100 22 -5 1503 7710.39 24589.08 32299.47 18 145 300 335 314 300 100 25 -5 1519 7792.47 24850.84 32643.31 19 145 300 335 318 300 110 24 -4 1532 7859.16 25063.52 32922.68 20 100 300 335 314 300 88 26 10 1463 7505.19 23934.68 31439.87 21 70 300 315 303 250 0 55 5 1293 6633.09 21153.48 27786.57 22 0 145 310 301 200 0 62 10 1018 5222.34 16654.48 21876.82 23 0 85 305 254 180 0 64 4 888 4555.44 14527.68 19083.12 24 0 30 300 195 127 0 60 1 712 3652.56 11648.32 15300.88
129091.32 411683.04 540774.36 BESS Cost =12,000 Total 552,774.36
Fig. 6.6 State of charge of the battery.
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time (Hour)
SOC
141
Fig. 6.7 Wind penetration CUCE with/without BESS.
6.6.3. Comparisons
For evaluating the performance of the proposed method, GA, QEA, PSO
and SQP-PSO are employed in the case study, which are shown in Table 6.6.
For comparison purposes, these algorithms are used directly to solve the
CUCE problem with wind power. For the proposed SQP-PSO algorithm, the
population size is 100 and the maximum number of iterations is 3 for PSO.
To make a fair comparison, the population size is fixed at 100 and tested for
a maximum iteration count of 100. The initial crossover and mutation rates
for GA were all set as 80 (%). All the programs were run on a 3.0 GHz,
Intel Core 5, with 4G RAM desktop. The CPU time to obtain the solution of
the best, the worst and the average results for different test algorithms are
shown in Table 6.6. According to the comparison results, the SQP-PSO has
shown the superiority to the existing methods.
0%
5%
10%
15%
20%
25%
30%
1 2 3 4 5 6 7 8 9 101112131415161718192021222324
Win
dpe
netr
atio
n
Time (Hour)
with BESS
without BESS
142
Table 6.6
Comparison of Different Approaches
CUCE with Wind Farm and BESS
Method Execution
Time (s)
Best Cost
($)
Mean Cost
($)
Worst Cost
($)
GA
MIP
13.72
8.29
559,162
561,699
560,369
563,162
561,162
568,640
PSO 12.26 559,189 556,957 556,311
QEA 9.01 554,170 557,171 560,673
SQP-PSO 7.33 550,123 552,744 557,213
6.7. Conclusion
This chapter develops a hybrid method combining the SQP and PSO to
achieve faster and better performance optimization. The method has been
successfully applied to solve the power system UC problem considering
GHG emissions and wind power in an integrated CUCE model. The
proposed hybrid method has been applied to solve the carbon tax with UC
problem and tested on 2 wind farms and 6 thermal units system. To address
the uncertainties in wind power production, a reduction algorithm for
forecasting is applied in formulating the optimization model. Comparisons
have been made for the proposed CUCE model with and without wind
farms. The wind farm combined model shows a better performance in terms
of less emission cost. In addition, the resultant overall unit scheduling cost
is also optimized considering the government subsidy.
Due to the intermittency and volatility of wind energy, the dispatch of wind
energy is a difficult task. A high level of wind power penetration could
impact the system security and reliability of power grid. In order to reduce
fluctuation and smooths the output curve, a battery energy storage system
(BESS) is introduced in the previous method. A novel coordinated unit
commitment operation combined wind-thermal with BESS is proposed. The
143
proposed hybrid model has been compared with the model without BESS in
the studied cases. The simulation results show that the novel method for
CUCE with wind energy gives a better emission and operation solution,
both efficiently and economically. At same time, BESS will be charged
during off-peak periods and discharged during peak periods for economic
operation. BESS reduces wind curtailment as wind penetration increases in
a system, thereby reducing the operating costs. According to the simulation
results, we observe that CUCE integrating wind and BESS can improve the
peak load reduction, system operating cost, GHG emissions and
commitment of the units, which offers a convenient, economic and efficient
tool for the UC problem.
144
This page is intentionally left blank
145
Chapter 7
BESS and Wind Power Cooperative Dispatch
with Emission Limitation in Australia
7.1. Introduction
Due to concerns regarding air pollution and global warming, establishing a
low-carbon world has attracted extensive attention, and the activity of
utilizing clean energy has been accelerated. The combustion of fossil fuels
produces greenhouse gas (GHG) emissions such as NO2, CO2 and SF6.
Many countries are implementing policies to produce clean energy in order
to mitigate the greenhouse effects by introducing an emission tax. A variety
of electric smart grid operators are paying substantial attention towards
reducing the operation and emission costs. Integrating renewable energy
sources is a key factor for reducing air pollution and building a clean energy
environment. At the same time, highly efficient economic dispatch
strategies should also be considered.
As wind generation does not create any harmful emissions, government
often promote it as a means to reduce the national emissions levels [182].
However, the intermittent and uncertainty of wind makes planning and
power dispatch difficult. If a high level of wind power penetration is
integrated into the power system, there needs to be sufficient thermal
146
generation reserves to meet the power demand and to cope with the
intermittency of wind [165]. The high overestimation and underestimation
compensation cost of wind turbines make it difficult for the power
generation enterprises to introduce wind energy into the grid. As a result, it
is crucial to improve the wind power forecasting system which can greatly
help the integration process, since system operators rely on accurate wind
power forecasts to design operational plans and assess system economy
[203]. Energy storage systems have been shown to be quite suitable in
mitigating the negative impacts resulting from the integration of wind
generation. In order to reduce fluctuations of the wind energy output, a
battery energy storage system (BESS) is integrated into renewable energy
generation systems [168], [187], [199]. Fig. 7.1 shows a grid-connected
wind farm with BESS.
Fig. 7.1 Structure of wind power generation system integrating BESS.
In some existing works, researchers have shown great interest in
incorporating wind power in the analysis of scheduling strategies. In [204-
206], the authors introduce wind power into scheduling operation without
chBESSp
disBESSp
147
energy storage system (ESS). In [207,208], the authors combine wind power
with ESS without considering the carbon emission limit. In this work, we
consider a combined economic emission dispatch (CEED) which takes into
account of fuel cost and emission tax together. In [137], author developed a
hybrid algorithm which combines the quantum-inspired evolutionary
algorithms QEAs and PSO, i.e., a quantum-inspired particle swarm
optimization (QPSO) to address the economic dispatch emission problem,
and the hybrid algorithm has been shown to have excellent performance. In
this chapter we use the same QPSO algorithm as in [137], and build an
advanced economic dispatch model for smart gird system implementation
which is different from existing works and pays special attention to
emission constraint, wind power, and energy storage system integration. A
large penetration of unpredictable and variable generation introduces
additional constraints on the system. Any imposed constraints on system
operation can make the dispatch more difficult and increase the operational
cost. A battery energy storage system (BESS) embedded into the dispatch
model can improve the utilization of wind energy. Carbon tax which can
achieve large scale CO2 emission reduction is also introduced into the model.
The objective function of proposed model is designed to accomplish all of
these tasks while taking into account the non-idealities of a real system. In
many instances these non-idealities are high order and nonlinear. When the
total limits of CO2 emission are imposed over the scheduling horizon, the
dispatch model becomes more difficult to solve because the region of
feasible solutions becomes much smaller.
The QEAs can explore the target space with a smaller number of individuals
and exploit global solution within a short span of time [209]. Hence, the
QEAs can strike the right balance between exploration and exploitation
more easily for constrained optimization problems such as the one
considered in this chapter. Quantum bit is used as probabilistic
148
representation of particles, defined as the smallest information unit. A string
of quantum bits consist of many quantum bit individuals. Quantum rotation
gate is defined as an implementation to drive the individuals moving
towards better solutions, and eventually find the global optimum. Likewise,
PSO is one of the modern heuristic algorithms and has gained a lot of
attention in various power system applications. We take advantage of both
QEA and PSO algorithms: QEA is used to obtain an initial solution and
boundary condition which is then used in the PSO algorithm to obtain a
final solution. In this chapter, we present the hybrid technique to address the
emission economic dispatch problem. To realize the benefits of the control
of storage, a sensitivity analysis is performed based on different levels of
BESS.
7.2. Probability Analysis of Wind Power and Battery
Energy Storage System (BESS)
7.2.1. Probability of Wind
The probability of wind has been discussed in chapter 4.
7.2.2. Battery Energy Storage System (BESS)
Combining an energy storage system (ESS) together with wind power has
been proposed in order to provide economic and technical benefits to power
systems [199]. Different energy storage systems (ESS) technologies such as
pumped hydroelectric, compressed air, super-capacitors, magnetic storage,
electrical batteries, and flywheels are proposed. For large-scale electricity
storage, the large fuel cells seem to be appropriate [198]. Technical factors,
operating process, and economy should be considered in the selection of a
suitable battery type. The important battery parameters, which may affect
149
power system operation, are the minimum and maximum storage capacity
(MWh), the charging and discharging rate (MWh/hour), and the state of
charge (SOC). Battery storage can be controlled to charge or discharge in
constant or variable rate based on system operation requirements. The merit
of BESS is the fast rate of charging/discharging. Three different levels of
BESS will be embedded into the dispatch model in this work, 15%, 20%,
and 25% respectively. Under the varying levels of BESS, the different
operating characteristics, such as wind penetration, carbon emission,
generating cost would be changed. Sensitivity analysis could find the
appropriate level of BESS for this optimization dispatch model. In the
emission economic dispatch problem, when the forecast value is smaller
than actual power output (underestimate), the excess energy can be stored in
energy storage. If the actual value is less than the predicted value
(overestimate), energy from the storage can supply to meet the system load
demand [209]. The minimum and maximum energy stored in the battery
bank are specified, the SOC lower limit is 20% and upper limit is 80% of
battery full capacity respectively. For instance, we assume the initial state of
the BESS is SOC lower limit and the BESS will be in charging state to store
energy. When the battery bank reaches the upper limit state, the charging
process is finished. In this case, the capacity of the BESS is around 80% of
its full capacity, and then BESS will discharge until it reaches the lower
limit of SOC [202]. It should be noted that by limiting the SOC to be
between 20% and 80%, the deep charge/discharge cycles have been
minimized in order to extend the lifetime of the battery. During the charging
and discharging procedure only one state is in process, that is charging and
discharging cannot happen at the same time.
150
7.3. The Proposed Economic Dispatch Model
The formulation of ED model including wind power, battery energy storage
system and carbon tax are described in this section. The aim of this ED
model is to minimize emission cost and the operation costs (including wind
power cost, fuel cost and battery cost) while satisfying a set of given
constraints. The BESS is adopted into the model to shave the load peak and
address the wind intermittency. The objective function to be minimized is
given below:
cos , , ,1 1 1
, ,1 1
, ,1 1
.
( )
( ) +
M N Mp w e
t i t j t i ti j i
N Ns
j t o t oej j
N NBESS
u t ue i tj i
Minimize F C C C
C E C W
E C W C
(7.1)
where M and N are the number of thermal generators and wind turbine units.
The thermal generator cost is a quadratic function, which contains higher
order nonlinearities and discontinuities due to valve point effects:
2
, , , , ,min, , sin ( )
p
t i i t i t i t i i i iC a b p i t c p i t d e p p (7.2)
where p(i,t) is the actual power generated by thermal unit i at time t, and
here ai, bi, ci, are the production cost of thermal unit i at time t, and di, ei, fi
are the coefficients of fuel consumption for thermal unit i. We denote ,wj tC
as the wind power cost of the wind farm,
, ,, ,wj t w j a vC Q j t W j t (7.3)
where αw, j is the production cost coefficient of wind unit j, Q(j,t) is the
on/off status of wind unit j at time t and Wav (j,t) is the actual power
151
generated by wind unit j at time t. The term Wav (j,t) should be zero if the
system operator owns the wind farm. The third item is the emission function
and it can be represented as
,1
( )M
ei t T a x i i
i
C C E M p
(7.4)
2( ) ( )i i i i i i i iE M p e f f g p h P (7.5)
The term EMi (pi) in (7.5) is the carbon emissions of thermal unit i, efi is the
fuel emission factor of CO2 for thermal generator i. The terms fi, gi and hi
are the coefficients of fuel consumption whereas CTax is the market carbon
tax price and ,wj tC is the government subsidy of wind unit j at time t,
, ,, ,sj t s j avC Q j t W j t (7.6)
where αs ,j is the government subsidy coefficient of wind power. From the
PDF functions given in (4.4)-(4.7), the mean of the overestimation penalty
cost of the wind power is assumed as:
, , 0
, 0 0
( )j
j j
w
o t os o j j W
w w
o j j W W
E C W C w w f w dw
C w f w dw wf w dw
(7.7)
where wj is the predicted power from wind turbine unit j. Co,j is the cost
coefficient for purchasing reserve power from other source due to
overestimation. Similarly, the underestimation penalty cost is given as
,
, ,
, ,
,
( )r j
j
r j r j
j j
w
u j ue u j j Ww
w w
u j W j Ww w
E C W C w w f w dw
C wf w dw w f w dw
(7.8)
152
where wr,j is the rated power of wind turbine j. The term Cu,j is the cost
coefficient for not using all generated wind power due to underestimation.
The last component is the operation cost of BESS, it can be indicated as
,1
( ) ( )N
B E S S d is c hi t B E S S B E S S B E S S
i
C p t p t
(7.9)
where πbess is the coefficient of BESS consumption, in this work the πbess is
$ 0.1 / kW h , pBESS is the battery charging/ discharging power. System real
power balance is given by
, ,1 1 1
M N N
i av j BESS j d lossi j j
p w p p p
(7.10)
where pd is the total system demand, ploss is the total transmission losses.
The transmission loss is taken as approximately 5% of the total demand of
the system. Unit generator limits are
,min ,max,i ip p i t p (7.11)
Wind power unit limits are
,0 j r jw w (7.12)
BESS charge/discharge power limits are
,max
,max
0 ( )
0 ( )
ch chBESS BESS
dis disBESS BESS
p t p
p t p
. (7.13)
where ,maxchBESSp and ,maxdis
BESSp are the maximum charging and discharging rate of
battery. We assume the battery cannot charge and discharge at the same
time, ie.,
153
( ) ( )=0dis chBESS BESSp t p t . (7.14)
BESS storage constraints are
( )L USOC SOC t SOC (7.15)
max( ) ( ) / .BESS BESSSOC t C t C (7.16)
where SOC is the state of charge of BESS at time t. SOCL and SOCU are the
lower and upper SOC of the battery. Denote CBESS as the energy capacity of
the BESS,
( ) ( ) ( )ch disBESS ini BESS ch BESS disC t C p t p t (7.17)
where Cini is the initial value of the capacity of BESS, ηch and ηdis are
respectively the storage battery charge and discharge efficiency.
7.4. Quantum-Inspired Particle Swarm Optimization
The proposed ED model is a complicated optimization problem with highly
nonlinear and non-convex models. Traditional methods can easily be
trapped by local optima if applied directly. To solve this problem, we
propose a quantum inspired particle swarm optimization algorithm for the
economic dispatch problem. The quantum inspired particle swarm
optimization is a probabilistic search algorithm that implies quantum
behaviour in the PSO algorithm. We combine the advantages of both PSO
and the quantum inspired evolution algorithms (QEAs): the initial solution
and boundary conditions obtain from QEAs, and then use the PSO
algorithm to obtain a final solution.
154
7.4.1. Particle Swarm Optimization
The concept of Particle Swarm Optimization has been introduced in Chapter
4.
7.4.2. Quantum-Inspired Particle Swarm Optimization
QPSO has stronger search ability and quicker convergence speed since it not
only introduces the concepts of quantum bit and rotation gate but also the
implementation of self-adaptive probability selection and chaotic sequences
mutation. In the QPSO, the state of a particle is depicted by quantum bit and
angle, instead of particle position and velocity in classical PSO.
Quantum bit, the smallest unit in the QPSO, is defined as a pair of numbers,
1,2 ,,
1,2 ,ji
ji
j mi n
(7.18)
The modulus 2( ) || ji t and 2( ) || ji t give the probabilities that the quantum
bit exists in states “0” and “1”, respectively, which must satisfy,
2 2( ) | ( ) | 1| |ji jit t (7.19)
A string of quantum bits consists of a quantum bit individual, which can be
defined as,
1
1
1
( ), , ( ), , ( )
( ), , ( ), , ( )( )
( ), ( ), ( ), ,
ji jnjj
ji jnj
ji jnj
t t t
t t tq t
q t q t q t
(7.20)
A quantum bit is able to represent a linear superposition of all possible
solutions due to its probabilistic representation [210]. In total 2n kinds of
individuals can be represented by combinations of different quantum bit
155
states. This quantum bit representation has better characteristics of
generating diversity in population than other representations.
Because of the normalization condition, the quantum angle can be represented as,
1| ( ) =cos ( ) | 0 sin ( ) |1
( ) arctan
ji jij
jiji
ji
q t t t
t
(7.21)
0>, and 1> give the probabilities that the quantum bit exists in states “0” and
“1”
The quantum bit individual can be represented in the form of quantum
angles,
1
1
( )
( )
( ), ( ), ( )
( ), ( ), ( )
, ,
, ,
j
j
ji jnj
ji jnj
q t
t
q t q t q t
t t t
(7.22)
The fundamental update mechanism of QPSO is evolving quantum bits and
angles, by which the updated quantum bits should still satisfy the
normalization condition. The quantum rotation gate update equation could
be calculated by,
1
2
( 1) ( ) ( )
( )
j j pb j
gb j
t t r t
r t
(7.23)
where, j is angle change, j is current angles, pb is local best angles, and
gb is global best angles.
( 1) ( )cos ( 1) sin ( 1)
sin ( 1) cos ( 1)( 1) ( )
ji ji
ji ji
ji ji
ji ji
t tt t
t tt t
(7.24)
And quantum rotation gate can be illustrated in Fig. 7.2, [211].
A
ch
co
se
ad
Th
fit
po
de
w
de
ra
Th
Although th
haracteristic
ould still ap
elf-adaptive
dopted.
he individu
tness value
opulation in
esigned by u
where, r is ra
efinition is t
ather than re
he individu
he quantum
cs of popula
ppear. In ord
probabilit
F
ual affinity
of every in
n terms of f
using locati
As q
andom numb
that the affi
eal fitness v
al concentra
Cs q
m bit and
ation divers
der to addre
ty selection
Fig. 7.2 The q
value can b
dividual in
fitness valu
ion index of
( ) (1jq t r
ber in (0,1).
nity value i
alue.
ation can be
1( )
m
aj
Kq t
156
rotation ga
sity, the pre
ess this pro
n and chao
quantum rotati
be defined
the current
ue in ascend
f quantum b
1) jr
. The most a
is only relev
e defined as
( ), (j aKs q t q t
m
ate represen
emature con
oblem, the i
otic sequen
on gate.
as follows
population
ding sequen
bit individua
attractive fe
vant to the l
s,
)t
ntation has
nvergence p
implementa
nces mutati
. We calcu
and rearran
nce. The aff
al.
eature of thi
ocation inde
s better
problem
ations of
tion are
ulate the
nged the
ffinity is
(7.25)
is
dex
(7.26)
157
1, || ( ), ( ) ||
0,( ), ( ) j a
j aq t q t l
otherwiseKs q t q t
(7.27)
Roulette selection is implemented based on the computed selection
probabilities. This allocates every quantum bit individual a probability
of being selected proportionally according to selection probabilities.
The selection probabilities are,
( )
( )
( )
( )1
( )
j
j
j
j
As q t
Cs q t
j m As q t
Cs q tj
Ps q t
(7.28)
Therefore, the quantum bit individuals can be selected according to
individuals selection probabilities, guaranteeing that individuals having high
affinity values are selected; and the one that has high concentration value
could be rejected, which ultimately helps the algorithm converge at optimal
solutions.
Chaotic sequences mutation is implemented next. A widely used system
evidencing chaotic behaviour is the logistic map, which can be expressed as
follows
( 1) ( )[1 ( )], [0,4]g t g t g t (7.29)
The behaviour of the above chaotic system is greatly influenced by the
parameter, which determines whether it stabilizes at a constant size,
oscillates between limited sequences of sizes, or behaves chaotically in an
unpredictable pattern [211]. A small difference in the initial value causes
substantial differences in long time behaviour. Here we select μ=4, and the
mutation implementation can be defined as,
1( ) 4 ( )[1 ( )]( ) (0,1), 1,2, ,
i i i
i
g t g t g tg t i n
(7.30)
158
And
1 1 ( )( ) ( )j jt
s g tT
q t q t
(7.31)
Notice that there is a user-defined control variable s, which is the mutation
control constant. Selection of this value depends on practical problem. In
general, with little knowledge about global optimum, it is difficult to
constrain the mutation size to a sufficiently small region. Initial solutions
are usually far from the global optimum; hence larger mutation may prove
to be beneficial. But as the evolution progresses, later solutions may be
nearer to the global optimum and the mutation size should be reduced
gradually to help quick convergence. According to our experience, the range
[0.1, 0.5] is suitable.
7.4.3. Procedure of QPSO
Fig. 7.3 Flow chart of quantum-inspired particle swarm optimization.
7.5.
In this
generato
and tabl
coeffici
table 7.3
the histo
demand
of them
turbines
characte
Case St
section, the
or model of
le 7.2 list th
ents and win
3. The carbo
orical wind s
d and wind p
m consists of
s (3.0 MW
eristics are li
udies an
novel dispa
the Australia
he generator
nd turbine c
n tax is fixed
speed in Tas
ower output
25 wind turb
W). The BE
isted in table
Fig. 7.4 Simp
159
nd Discu
atch model
a power syst
parameters
characteristic
d as AUD 21
smania, Aust
are presente
rbines (3.0 M
SS model
7.5.
plified 14-gen
ssion
is implemen
tem shown i
and emission
cs of the two
1/t. A wind o
tralia [162]. T
ed in table 7.
MW), and the
is compose
erator, 50 Hz
nted on the
n Fig. 7.4 [2
n factors. Th
o wind farm
observation st
The forecaste
4. For the w
e other consi
ed of Na-S
system.
simplified 1
212]. Table 7
he penalty c
ms are listed
station provid
ted system lo
wind plants, o
ists of 40 wi
S batteries,
14-
7.1
ost
in
des
oad
one
ind
its
160
Table 7.1
Generator Parameter
Note: The coefficients of fi, gi, and hi are in t, t/MW and t/MW2 for coal/oil units.
The coefficients of fi, gi, and hi are in m3, m3/MW and m3/MW2 for gas unit.
Table 7.2
Emission Factors
Emission Factor Coal(kg/kg) Gas (kg/m3) Oil (kg/kg)
efco2 3.21 1.70 2.631
Table 7.3
Wind Turbine Parameter
Plant Model No C k ө vin vout vr wr Cw,j Cu,j Co,j Cs,j
G13 Sinvoel 25 4.7 1.89 0 3 25 16 3 0 50 20 10
G14 Vestas 40 4.3 1.71 0 4 25 16 3 0 40 30 10
Unit Fuel Cost Coefficient Fuel Cost Coefficient
Pmin (MW) Pmax (MW) ai bi ci di ei fi gi h
G1 (Coal)
G2 (Coal)
2000
2000
10
10
0.002
0.0025
200
200
0.084
0.080
45
40
0.30
0.20
0.00004
0.00005
20
20
120
120
G3 (Coal) 2500 15 0.0030 300 0.035 50 0.30 0.00004 20 110
G4 (Coal) 2500 13 0.0025 280 0.040 44 0.25 0.00005 20 110
G5 (Coal) 6000 10 0.0018 400 0.042 80 0.12 0.00003 120 600
G6 (Coal) 5800 9 0.0020 380 0.040 75 0.15 0.00025 110 600
G7 (Gas) 925 18 0.0031 150 0.063 2462 48 0.0085 120 520
G8 (Gas) 930 20 0.0032 145 0.060 2430 45 0.0084 110 510
G9 (Gas) 950 20 0.0035 100 0.080 2500 50 0.0090 100 500
G10 (Gas) 960 23 0.0035 100 0.085 2550 50 0.0095 100 510
G11 (Oil) 124 23 0.0038 80 0.010 1.250 0.25 0,00003 30 200
G12 (Oil) 130 25 0.0040 85 0.098 1.100 0.23 0.00003 35 190
G13(Wind) 0 0 0 0 0 0 0 0 0 90
G14(Wind) 0 0 0 0 0 0 0 0 0 120
161
Table 7.4
Predict System and Wind Farm Output
Index Case A Case B Case C)
Demand(MW)
G13
G14
2400
44
60
2800
10
40
3200
30
50
Table 7.5
BESS Parameter
Quantity Value Quantity Value
maxC (MWh) 50 , maxchBESSp (MWh/hour) 10
minC (MWh) 0 ,maxdisBESSp (MWh/hour) 10
iniC (MWh) 10 ch 0.83
LSOC 20% dis 0.83
USOC 80% BESS 0.1 / kW h
7.5.1. Benefits of Carbon Tax
In this case study, the simulation results are shown in table 7.6. In Cases A,
B and C, “a” group represent the ED model without carbon tax while “b”
group represent the carbon tax model. Comparing groups “a” with “b”, it
can be seen that the generating outputs units will be impacted: part of the
power supplied from the highly polluted coal fired units (G1–G6) is shifted
to the less polluted gas generators (G7-G10) and oil generators (G11, G12).
The carbon tax model’s power generation cost is much higher than that
without a carbon tax, the reason is that less polluted generators operation
cost is higher. In Fig. 7.5, under three different cases of system demand, as
expected, the emission is less with the ED model with carbon tax than that
of the ED model without carbon tax.
162
Table.7.6
Solutions of Different ED Models in Three System Demand
Index Case A Case B Case C
ED
Units
Without wind farm With Wind farm Without wind farm With Wind farm Without wind farm With Wind farm
a b a b a b a b a b a b
G1(Coal) 106.52 22.89 103.82 54.9 107.86 29.84 94.82 62.95 110.06 45.36 96.33 30.12
G2 (Coal) 108.94 23.13 103.3 56.1 109.34 29.04 93.96 64.33 109.82 43.66 94.21 27.84
G3 (Coal) 101.73 21.47 99.2 21.06 100.24 23.03 96.95 20.49 99.86 22.63 96.58 20.59
G4 (Coal) 98.11 22.01 98.98 20.78 99.02 22.87 96.11 21.93 98.82 20.37 99.32 23.01
G5 (Coal) 581.93 568.65 567.84 570.18 588.21 589.34 596.25 578.05 579.37 569.36 566.51 569.33
G6 (Coal) 583.15 570.23 567.48 556.54 589.25 588.12 592.23 557.69 578.19 560.36 571.23 574.97
G7 (Gas) 260.18 409.37 209.32 355.26 417.02 502.71 320.12 477.81 511.39 550.39 456.25 471.26
G8 (Gas) 257.88 404.79 213.96 354.98 415.3 501.85 318.8 476.35 507.29 561.85 448.01 465.06
G9 (Gas) 111.86 139.15 147.08 120.33 146.02 216.04 177.96 137.28 258.22 365.42 241.29 364.36
G10 (Gas) 108.98 136.47 149.02 120.55 147.72 215.88 176.36 135.68 261.22 373.58 274.17 385.32
G11 (Oil) 40.99 40.26 39.96 48.5 40.13 41.32 54.02 59.09 42.45 42.25 50.22 50.23
G12 (Oil) 39.73 41.58 40.54 48.98 39.89 39.96 53.88 58.67 43.31 44.77 48.82 47.09
G13(Wind) 0 0 16.66 18.5 0 0 41.6 57.66 0 0 67.82 69.22
G14(Wind) 0 0 42.84 53.34 0 0 80.94 92.02 0 0 89.24 101.6
Total(MW) 2400 2400 2400 2400 2800 2800 2800 2800 3200 3200 3200 3200
Cost ($) 59231.4 97338.7 59026.2 97054.4 67373.6 106693.8 66519.6 105379.4 75683.6 116071.8 75203.2 115398.4
Note: “a” represents without carbon tax, “b” means with carbon tax.
Fig. 7.5 Carbon emission of ED models with/without wind farm (WF) & (a) with carbon tax; (b) without carbon tax.
1750
1800
1850
1900
1950
2000
2050
2100
2150
Em
isso
n (
ton
)
Case A
With out WF a
With out WF b
With WF a
With WF b
2050
2100
2150
2200
2250
2300
2350
2400
2450
2500
Em
isso
n (
ton
)
Case B
Without WF a
Without WF b
With WF a
With WF b
235 0
240 0
245 0
250 0
255 0
260 0
265 0
270 0
275 0
280 0
285 0
Em
isso
n (
ton
)
Case C
Without WF a
Without WF b
With WF a
With WF b
163
Table 7.7
Generation Output and Cost with Different Level BESS Index Case A ED with carbon tax & wind farm Case B ED with carbon tax & wind farm Case C ED with carbon tax & wind farm
BESS
Unit 15% 20% 25% 15% 20% 25% 15% 20% 25%
BESS 29.25 39 48.75 27.14 34.32 43.21 28.28 36.21 45.89
G1(Coal) 53.85 52.7 51.9 57.68 58.33 52.77 26.34 25.37 24.49
G2 (Coal) 55.2 53.3 52.1 56.35 58.79 55.28 24.38 23.51 23.08
G3 (Coal) 22.03 22.06 21.08 21.49 21.28 22.49 21.36 21.39 20.31
G4 (Coal) 21.91 21.78 20.56 21.46 22.39 21.27 22.53 22.28 21.53
G5 (Coal) 559.17 553.09 549.37 569.25 568.41 566.63 567.47 558.27 558.19
G6 (Coal) 542.03 541.65 539.54 554.73 553.27 549.46 572.44 569.14 568.26
G7 (Gas) 350.23 348.46 343.32 471.32 468.01 469.15 465.18 466.32 461.07
G8 (Gas) 349.17 347.98 355.43 474.88 466.03 466.73 461.76 459.29 451.91
G9 (Gas) 119.34 118.21 116.12 131.97 129.28 134.23 359.18 358.03 361.43
G10 (Gas) 121.48 119.67 118.23 132.69 128.68 133.35 383.46 376.61 381.69
G11 (Oil) 49.2 50.5 52.43 62.02 59.09 60.29 46.23 49.86 51.83
G12 (Oil) 53.23 51.96 54.26 63.32 65.54 61.86 43.52 48.17 49.27
G13(Wind) 19.3 22.3 21.4 59.33 61.36 60.16 71.57 74.09 72.02
G14(Wind) 54.61 57.34 55.51 96.28 105.22 103.12 106.3 111.46 109.03
Total(MW) 2400 2400 2400 2800 2800 2800 3200 3200 3200
Cost ($) 96833.6 96143.5 96710.8 104853.2 103984.9 104438.9 114892.1 114017.9 114426.2
Fig. 7.6 Carbon Emission with different level of BESS in three system demand.
1760
1780
1800
1820
1840
1860
1880
1900
1920
Em
isso
n (to
n)
Case A
without BESS
BE SS 15%
BE SS 20%
BE SS 25%
2060
2080
2100
2120
2140
2160
2180
2200
2220
Em
isso
n (to
n)
Case B
without BESS
BE SS 15%
BE SS 20%
BE SS 25%
2360
2380
2400
2420
2440
2460
2480
2500
2520
2540
Em
isso
n (
ton
)
Case C
without BESS
BESS 15%
BESS 20%
BESS 25%
2.70%
2.80%
2.90%
3.00%
3.10%
3.20%
3.30%
3.40%
Win
d P
enet
rati
on
F
7.
Th
em
pa
G
de
it
re
of
th
In
po
va
co
w
un
ge
Cas
Without BESS
BE SS 15%
BE SS 20%
BE SS 25%
Fig. 7.7 Wind
.5.2. Integ
he ED mod
missions an
art of the po
G14). Even
ecreased wi
can be show
emarkably in
f the wind e
he proposed
n Case A, th
ower. In Ca
alue. Clearl
onsidered a
wind powe
nderestimat
eneration m
se A
d penetration w
gration of
del incorpor
nd total syst
ower supply
though the
ith the prop
wn that the
n compariso
energy havi
ED with w
he generated
ase C, the
ly, the over
and the cost
er should
ion compen
more expensi
without/with d
f Wind Po
rates the wi
tem operati
y is shifted t
e wind pow
osed solutio
ED model w
on with the
ng no emis
wind energy
d wind pow
generated w
restimation
t for compe
be appl
nsation cost
ive and was
164
different level
ower
ind farm to
ion costs. F
to the clean
wer cost is
on of ED w
with wind f
e solution w
ssions. Ther
gives an ef
wer of G13 i
wind powe
and undere
ensating fo
lied. The
t of wind tu
ste energy s
l of BESS in th
minimize t
From table
n power win
expensive
with wind fa
farm reduce
without wind
refore, acco
fficient emis
is less than
er is larger
estimation s
r the surplu
high ov
urbines mak
ource.
5.0
5.
5.2
5.3
5.4
5.5
5.6
5.7
5.8
5.9
Win
d P
enet
rati
on
hree system d
the greenho
7.6, it is cl
nd generator
, emission
arm. From F
es the emissi
d energy as
rding to the
ssion solutio
the predicte
than the pr
situation sh
us and insu
verestimatio
ke the wind
00%
10%
20%
30%
40%
50%
60%
70%
80%
90%With
BES
BES
BES
demand.
ouse gas
lear that
rs (G13-
cost is
Fig. 7.5,
ion cost
a result
e results
on.
ed wind
redicted
hould be
ufficient
on and
d power
Case C
hout BESS
SS 15%
SS 20%
SS 25%
165
7.5.3. Combined with BESS
The wind power intermittency and uncertainty sometimes make it difficult
to perform the dispatch of wind energy. High wind power penetration and
curtailment could reduce the power grid dispatchability. In our research, to
reduce the variability of wind energy and increase its dispatchability, BESS
is combined with wind generators with the consideration of carbon tax.
Comparing all groups bolded “b” in table 7.6 with table 7.7, BESS adopted
in wind-thermal generation system can decrease the generation cost. At the
same time, according to Fig. 7.6, after introducing BESS the power
generation emission decreases significantly.
The profiles in Fig. 7.7 show that BESS cooperative dispatching with wind
ensures grid dispatchability and economy to improve the penetration of
wind power.
In order to determine the appropriate storage size, we test three different
levels of BESS for the ED problem with wind farm and carbon tax. Table
7.7 shows that when the BESS is 20% of wind farm capacity, the operation
cost is lowest in those three groups. The cost of BESS with 25% wind farm
capacity cost is lower than with 15%. Similarly, the carbon emission in the
20% group is the lowest, while the emission of the 25% group is less than
the 15 % group in Fig 7.6. A suitable level of BESS could enhance the wind
power penetration in the generating system. It can be shown when the BESS
is 20% of wind farm capacity, the wind penetration is highest. At same time,
the dispatch model still maintains its dispatchability and economy.
166
7.6. Conclusion
In this chapter, a cooperative dispatch model is proposed, which
incorporates wind generation, emission limitation, and energy storage
system. This model was applied successfully in a simplified 14-generator
model of the Australian power system. This work focuses on BESS and
wind energy generation coordination and optimization with an ED model.
The benefits of coordination were verified using case studies. The power
generation cost and emission are shown to reduce dramatically by wind-
BESS coordination. It was also shown that a higher wind energy penetration
is possible by a coordinative operation. Different levels of BESS were
evaluated for the ED problem considering carbon tax and wind energy
integration. The proposed solution framework can determine the most
suitable and economical storage capacity and dispatch for the Australia
power grid. The proposed novel method for coordinating BESS with wind
energy gives a better emission and operation solution for the ED problem of
the Australian power system.
167
Chapter 8
Conclusion and Future Work
8.1. Summary of Contributions
In the past decade, wind power has become a generation technology of
significance in a number of countries, and its growth is foreseen to continue.
However, the integration of wind power is a challenge due to the
intermittency and uncertainty wind resource. When integrating significant
amounts of wind power in power systems, technical challenges arise due to
the uncontrollability of the primary energy source, the wind. While power
system operation requires a continuous power balance between generation
and load, the variability and limited predictability of wind power introduces
additional uncertainty into power system operation. The question arises, to
what extent can the power system accommodate wind power while
maintaining a reliable electricity supply?
This research has been focused on using new techniques of numerical
analysis, control methodologies and equipment modelling to improve the
power system operation efficiency, and minimize the wind power and BESS
operation and emission cost. A hybrid optimization method connecting
sequential quadratic programming (SQP) and particle swarm optimization
(PSO) was developed to solve the combined economic and emission
168
dispatch (CEED) and unit commitment (CUCE) problem with stochastic
wind power. A quantum-inspired particle swarm optimization (QPSO) has
been proposed so as to overcome many drawbacks that affect the original
PSO and solve the economic dispatch (ED) problem considering
probabilistic wind power and carbon tax.
Moreover, in order to reduce fluctuation of the wind energy output, a battery
energy storage system (BESS) in renewable energy generation system is
adopted. Wind farm combining with battery energy storage can enhance
system reliability, power availability and quality, and operational efficiency.
Storage technique was used for the economic dispatch (ED) unit
commitment (UC) with wind power and emission issue.
Chapter 2 described the background of wind power, such as wind power
conversion, impacts of wind power, maximum rotor efficiency, speed
control for maximum power, some of the design considerations in wind
turbine design, wind farms, latest trends of wind power generation,
problems related with grid connections and the promotion of wind power
generation have been discussed. This chapter also provides a summary of
the effect energy storage and those currently developed under research for
energy storage systems. Multiple storage techniques are introduced and
different kinds of battery for storage system are presented. Technical and
financial challenges of renewable energy and storages are also included.
Chapter 3 discussed ongoing research for optimal design of hybrid
renewable energy systems. Different approaches for the power generation
system combined the wind power and energy storage system. The basic
concepts of wind-thermal generation and wind-storage systems are first
reviewed and this is followed by comprehensive discussions of existing
techniques. Furthermore, existing techniques relevant to solving the wind
power system problem is studied and discussed. Then the field of
169
evolutionary algorithms was discussed, focusing in particular on
comparisons between these algorithms. It has been shown that all these
algorithms are effectively the same except for their different background
theories and evolutionary implementations. Each method has its own merits
and drawbacks, which are discussed in detail.
Chapter 4 considered the combined economic and emission dispatch (CEED)
with wind power. Nowadays, hybrid optimization methods combining
different techniques have received widespread concerns. The previous
research works have proved that the result from a composite of optimization
methods is often superior to those produced by any individual approaches.
The reason is that the combination of optimization techniques can overcome
individual disadvantage and benefit from each other’s advantages. In this
chapter, we developed a novel hybrid optimization algorithm connecting
sequential quadratic programming (SQP) and particle swarm optimization
(PSO) for solving combined economic and emission dispatch (CEED)
problem with valve point effects as well as stochastic wind power. The
probability of stochastic wind power based on the linear wind power output
curve is involved in the proposed CEED model. The test system is
composed of six thermal units and one wind farm. A set of numerical
experiments have proved the effectiveness of the hybrid computational
method.
Chapter 5 proposed a unit commitment (UC) considering probabilistic wind
power and emission problem. Special attention has been paid to the Wind-
Thermal cooperation dispatch considering carbon tax. Wind generation, as a
renewable source gradually becomes an integral part of smart grid
infrastructure. The introduction of a carbon tax can optimize carbon
emissions. In order to address the advanced dispatch strategy a hybrid
computational framework based on Sequential Quadratic Programming
(SQP) and Particle Swarm Optimization (PSO) is adopted. Based on the
170
analytic probability of stochastic wind power, the final scheduled outputs of
the wind farm have been calculated.
Chapter 6 provided a hybrid renewable generation model to address the
combined unit commitment and emission (CUCE) problem. By considering
a model which includes thermal generators, wind farms and energy storage
system (ESS), the proposed novel scheduling scheme can minimize the
operation cost and greenhouse gases emission cost. The intermittency and
uncertainty of wind make the dispatch of wind energy a difficult task. High
wind power penetration could impact the system security and reliability of
the power grid. To reduce fluctuation of the wind energy output, a battery
energy storage system (BESS) in renewable energy generation system is
adopted. In order to solve the schedule problem, a hybrid computational
framework based on sequential quadratic programming (SQP) and particle
swarm optimization (PSO) is introduced. The viability of the novel
scheduling scheme is demonstrated using a set of numerical case studies.
The result shown that CUCE integrating wind and BESS can improve the
peak load reduction, system operating cost, GHGs emission and
commitment of the units, which offers a convenient, economic and efficient
tool for the UC problem.
Chapter 7 focused on the power system operations with wind power and
carbon tax embedding storage in Australia. The newly proposed quantum-
inspired particle swarm optimization (QPSO) was researched. QPSO has
stronger search ability and quicker convergence speed since it not only
introduces the concepts of quantum bit and rotation gate, but also involves
the implementation of self-adaptive probability selection and chaotic
sequences mutation. It was shown here that the QPSO has superior search
capability and speed. Battery energy storage system (BESS) is incorporated
with wind generation in order to reduce fluctuation of wind energy output.
To realize the benefits of storage control, a sensitivity analysis is performed
171
based on different levels of BESS. In this work, BESS and wind energy
generation coordination and optimization with an ED model. The benefits of
coordination were verified using case studies. The power generation cost
and emission are shown to reduce dramatically by wind-BESS coordination.
It was also shown that a higher wind energy penetration is possible by a
coordinative operation. Different levels of BESS were evaluated for the ED
problem considering carbon tax and wind energy integration. The proposed
solution framework can determine the most suitable and economical storage
capacity and dispatch for the Australian power grid.
8.2. Suggestions for Future Work
8.2.1. Wind Power + Solar Power + Storage Dispatch/Unit
Commitment Considering Emission Problem
In this work, the stochastic nature of wind power and solar power will be
solved by the derived cumulative distribution function and Monte Carlo
sampling technique. The results of two simulation methods will be
compared. For the Monte Carlo sampling, Variance Reduction such as
Importance Sampling (IS) and Latin Hypercube Sampling (LHS) will be
applied. Unlike simple random sampling, IS and LHS ensure a full
convergence onto the range of variables by maximally satisfying marginal
distribution.
In the case study part, the model with wind power was evaluated using the
historical wind speed dataset. We assume that the wind speed data from a
large wind farm and use the data to estimate the generated wind power. The
solar power distribution will be assumed as normal distribution and we will
use the data from the Newcastle Solar Power Station. Before supplying the
172
electric power, the current went through a full bridge inverter, filter inductor,
and transformer. The solar power generation can work with off-grid or on-
grid forms, so it can be very flexible and convenient. The output power of
photovoltaic cells is affected by the intensity of sunshine, and the battery
junction temperature and other factors. In terms of the optimization method,
a new hybrid approach such as Fuzzy GA combined improved PSO will be
applied.
8.2.2. Large-Scale Renewables and Energy Demand
At present, renewables such as wind power, geothermal and solar altogether
supply less than 1% of energy demand worldwide. With a substantially
larger penetration, the variability and periodic unavailability of renewables
presents a formidable challenge for existing energy systems in general and
for power systems in particular. The methodologies developed here for the
power system integration of wind power are in principle applicable for a
wide range of renewables, since most renewables have a limited
controllability of their respective energy sources. Future research in the
integration of renewables in existing energy systems should revolve around
three questions: 1) How large are the differences in time and in size between
the energy demand and the energy supply? 2) How do increased amounts of
renewables influence these differences? 3) How can the differences between
demand and supply be narrowed in the best possible way? The electricity
system must be regarded as an integral part of a much larger energy supply
system. The development of integrated solutions will require substantial
research efforts.
173
Bibliography
[1] N. Navid and G. Rosenwald, "Market Solutions for Managing Ramp Flexibility With High Penetration of Renewable Resource," IEEE Transactions on Sustainable Energy, vol. 3, pp. 784-790, 2012.
[2] M. Milligan, E. Ela, D. Lew, D. Corbus, W. Yih-huei, B. Hodge, and B. Kirby, "Operational Analysis and Methods for Wind Integration Studies," IEEE Transactions on Sustainable Energy, vol. 3, pp. 612-619, 2012.
[3] G. Yang, J. D. McCalley, and N. Ming, "Coordinating Large-Scale Wind Integration and Transmission Planning," IEEE Transactions on Sustainable Energy, vol. 3, pp. 652-659, 2012.
[4] T. Aigner, S. Jaehnert, G. L. Doorman, and T. Gjengedal, "The Effect of Large Scale Wind Power on System Balancing in Northern Europe," IEEE Transactions on Sustainable Energy, vol. 3, pp. 751-759, 2012
[5] S. Faias, J. de Sousa, F. S. Reis, and R. Castro, "Assessment and Optimization of Wind Energy Integration Into the Power Systems: Application to the Portuguese System," IEEE Transactions on Sustainable Energy, vol. 3, pp. 627-635, 2012.
[6] L. Jiaqi, S. Grijalva, and R. G. Harley, "Increased Wind Revenue and System Security by Trading Wind Power in Energy and Regulation Reserve Markets,"IEEE Transactions on Sustainable Energy, vol. 2, pp. 340-347, 2011.
[7] N. Aparicio, I. MacGill, J. Rivier Abbad, and H. Beltran, "Comparison of Wind Energy Support Policy and Electricity Market Design in Europe, the United States, and Australia," IEEE Transactions on Sustainable Energy, vol. 3, pp. 809-818, 2012.
174
[8] A. K. Srivastava, A. A. Kumar, and N. N. Schulz, "Impact of Distributed Generations With Energy Storage Devices on the Electric Grid," IEEE Systems Journal, vol. 6, pp. 110-117, 2012.
[9] EnerNex Corporation. Wind Integration Study for the Public Service Company of Colorado. May 2006.
[10] J. P. S. Catalao, H. M. I. Pousinho, and V. M. F. Mendes, "Optimal offering strategies for wind power producers considering uncertainty and risk," IEEE Systems Journal, vol. 6, pp. 270-277, 2012
[11] A. Borghetti, M. Bosetti, S. Grillo, S. Massucco, C. A. Nucci and M. Paolone., "Short-Term Scheduling and Control of Active Distribution Systems With High Penetration of Renewable Resources," IEEE Systems Journal, vol. 4, pp. 313-322, 2010.
[12] J. Kabouris and C.C. Vournas. “Application of interruptible contracts to increase wind-power penetration in congested areas”, IEEE Trans. Power Syst, 2004, 19, (3), pp. 1642-1649.
[13] H.Y. Yamin, Q. El-Dwairi, and S.M. Shahidehpour, “A new approach for GENCOs profit based unit commitment in day-ahead competitive electricity markets considering reserve uncertainty”, Int. J. Elec. Power, vol. 29, no. 8, pp. 609-616, Oct. 2007.
[14] R. Billinton, B. Karki, R. Karki, and G. Ramakrishna, “Unit commitment risk analysis of wind integrated power systems,” IEEE Transactions on Power Systems, vol. 24, no. 2, pp. 930–939, 2009.
[15] J. Restrepo and F. Galiana, “Assessing the yearly impact of wind power through a new hybrid deterministic/stochastic unit commitment,” IEEE Transactions on Power Systems, vol. 26, no. 1, pp. 401–410, 2011.
[16] Y. Makarov, P. Etingov, J. Ma, Z. Huang, and K. Subbarao, “Incorporating uncertainty of wind power generation forecast into power system operation, dispatch, and unit commitment procedures,” IEEE Transactions on Sustainable Energy, vol. 2, no. 4, pp. 433–442, 2011.
175
[17] S. Kazarlis, A. Bakirtzis, and V. Petridis, “A genetic algorithm solution to the unit commitment problem,” IEEE Transactions on Power Systems, vol. 11, no. 1, pp. 83–92, 1996.
[18] D. Dasgupta and D. McGregor, “Thermal unit commitment using genetic algorithms,” IEE Proceedings-Generation, Transmission and Distribution, vol. 141, no. 5, pp. 459–465, 1994
[19] G. Sheble and T. Maifeld, “Unit commitment by genetic algorithm and expert system,” Electric Power Systems Research, vol. 30, pp. 115 –121, 1994.
[20] T. W. Lau, C. Chung, K. P. Wong, T. S. Chung, and S. Ho, “Quantuminspired evolutionary algorithm approach for unit commitment,” IEEE Transactions on Power Systems, vol. 24, no. 3, pp. 1503–1512, 2009
[21] A. H. Mantawy, Y. Abdel-Magid, and S. Selim, “A simulated annealing algorithm for unit commitment,” IEEE Transactions on Power Systems, vol. 13, pp. 197–204, 1998.
[22] M. Rashidi, M. Ali, N. Freidoonimehr, and F. Nazari, "Parametric analysis and optimization of entropy generation in unsteady MHD flow over a stretching rotating disk using artificial neural network and particle swarm optimization algorithm," Energy, vol. 55, pp. 497-510, 2013
[23] M. Rashidi, O. A. Bég, A. B. Parsa, and F. Nazari, "Analysis and optimization of a transcritical power cycle with regenerator using artificial neural networks and genetic algorithms," Proceedings of the Institution of Mechanical Engineers, Part A: Journal of Power and Energy, p. 0957650911407700, 2011.
[24] S. Shamshirband, D. Petković, R. Hashim, S. Motamedi, and N. B. Anuar, "An appraisal of wind turbine wake models by adaptive neuro-fuzzy methodology," International Journal of Electrical Power & Energy Systems, vol. 63, pp. 618-624, 2014.
[25] J. Lang, L. Tang, and Z. Zhang, “An improved binary particle swarm optimization for unit commitment problem,” Asia-Pacific Power and Energy Engineering Conference (APPEEC), pp. 1–4, 2010.
176
[26] Global Wind Energy Council, http:// www.gwec.net/, “Wind is a Global Power Source”.
[27] Canadian Wind Energy Association Website, http://www.canwea.ca/, “Wind Version for Canada”.
[28] Windustry Website, http://www.windustry.com/, “Introduction to Wind Energy”.
[29] American Wind Energy Association Website, http://www.awea.org/, “Renewable Portfolio Standard”.
[30] T. Ackermann, Historical Development and Current Status of Wind Power, Chapter 2 of book: “Wind Power in Power Systems”, Edited by Ackermann T., John Wiley & Sons Ltd, 2005.
[31] M. P. Connor, The UK Renewable Obligation, Chapter 7 of book: Switching of Renewable Power, Edited by V. Lauber, Earthscan, 2005.
[32] R.C. Bansal, T.S. Bhatti and D.P. Kothari, “On some of the design ascpects of wind energy conversion systems,” Energy Conversion and Management (2002) 2175-2187.
[33] “Global wind scenario,” Power Line 7 (2003) 49-53.
[34] G.L. Johnson, Wind Energy Systems (Prentice-Hall, Englewood Cliffs, New Jersey, 1985).
[35] J.F. Manwell, J.G. McGowan and A.L. Rogers, Wind Energy Explained Theory, Design and Application (John Wiley & Sons, Inc., Ho, New Jersey, 2002).
[36] M.R. Patel, Wind and Solar Power Systems (CRC Press LLC, Boca Raton, Florida, 1999).
[37] B. Singh, “Induction Generator-A Prospective,” Electric Machines and Power Systems (1995) 163–177.
[38] G.D. Rai, Non Conventional Energy Sources, 4th edition (Khanna Publishers, New Delhi, India, 2000).
177
[39] P.K. Sandhu Khan and J.K. Chatterjee, “Three-phase induction generators: A discussion on performance,” Electric Machines and Power Systems (1999) 813–832.
[40] P. Gipe, Wind Power, Chelsea Green Publishing Company, Post Mills, Vermount, 1995.
[41] European Wind Energy Association, http://www.ewea.org.
[42] S. Boyd and L. Vandenberghe, “Convex Optimization”, Cambridge, MA Cambridge Univ. Press, 2004, p.307.
[43] J. Soens. Impact of Wind Energy in a Future Power Grid. PhD thesis, Katholieke Universiteit Leuven, Belgium, 2005.
[44] R.C. Bansal, T.S. Bhatti and D.P. Kothari, “Some aspects of grid connected wind electric energy conversion systems,” Interdisciplinary J. Institution on Engineers (India) 82 (2001) 25–28.
[45] Z. Saad-Saund, M.L. Lisboa, J.B. Ekanayka, N. Jenkins and G. Strbac, “Application of STATCOMs to Wind Farms,” IEE-Proc. Generation, Transmission and Distribution 145 (1998) 511–516.
[46] J. Bonefeld and J.N. Jensen, “Horns rev-160 MW offshore wind,”Renewable Energy World 5 (2002) 77–87.
[47] EPRI and U.S. Department of Energy. 2003. EPRI-DOE Handbook of Energy Storage for Transmission and Distribution Applications. 1001834.
[48] J. McDowell “Golden Valley Electric Association-Battery Energy Storage System Update.” ESA Conference. 2006.
[49] T. Tamakoshi and K. Asano. “NaS Battery Applications for Prompt Reserves and Suppression of Power Fluctuations.” ESA Conference. 2006.
[50] M. Lazarewicz and C. Lyons. “Flywheel Based Frequency Regulation Demonstration Projects.” ESA Conference 2006.
[51] EIA, Energy Information Administration, Oct. 2007, Available: http://www.eia.doe.gov/
178
[52] J. Enslin, and G. Thijssen, “Cost Comparison for a 20 MW Flywheel–based Frequency Regulation Power Plant”, Beacon Power Corporation, Sep. 2007
[53] G. Balzer, et. al., “Wind Energy Storages-Possibilities”, Proc. IEEE Lausanne POWERTECH, 2007, pp. 615-620.
[54] M. Kamibayashi and K. Tanaka, "Recent Sodium Sulphur Battery Applications", IEEE/PES Transmission and Distribution Conf. and Expo, Nov. 2001, vol. 2, pp. 1169 – 1173
[55] J. Desteese, et. al., “On the use of energy storage technologies for regulation services in electric power systems with significant penetration of wind energy”, 5th Int. Conf. European Electricity Market, May 2008, pp. 1-6.
[56] Sergio Vazquez, Srdjan M. Lukic, Eduardo Galvan, Leopoldo G. Franquelo, and Juan M. Carrasco. 2010. “Energy storage systems for transport and gridapplications.” IEEE Transactions on Industrial Electronics, 57, no. 12: 3881-3895. 2010.
[57] G. Torres. 2006. “The Truth About NiCd Batteries.” Last modified March6, 2006.http://www.hardwaresecrets.com/article/The-Truth-About-NiCd.
[58] ABB, Inc. 2013. “World’s Largest Battery Energy Storage System.” Accessed September16.http://www05.abb.com/global/scot/scot232.nsf/veritydisplay/3c4e15816e4a7bf1c12578d100500565/$file/case note bess gvea fairbanks-web.pdf.
[59] Golden Valley Electric Association. 2013. “Battery System.” Accessed September 16. http://www.gvea.com/energy/bess.
[60] W. B. Gu, C. Y. Wang, S. M. Li, M. M. Geng, and B. Y. Liaw. 1999.“Modeling discharge and charge characteristics of nickel metal hydride batteries.” Electrochemical Acta 44, no. 25: 4525-4541.
[61] Energizer. 2013. “Energizer NH15-2300.” Form No. EBC-7102WB. Accessed June 5. http://data.energizer.com/PDFs/nh15-2300.pdf\
179
[62] David J. Bradwell, Hojong Kim, Aislinn HC Sirk, and Donald R. Sadoway.2012. “Magnesium Antimony Liquid Metal Battery for Stationary Energy Storage.” Journal of the American Chemical Society 134, no. 4.
[63] Ken Mattern, A. Ellis, S. E. Williams, C. Edwards, A. Nourai, and Dave Porter. 2008. “Application of inverter-based systems for peak shaving and reactive power management.” In Transmission and Distribution Conference and Exposition, 2008. T&D. IEEE/PES, pp. 1-4.
[64] Ambri 2013. Ambri Brochure Accessed April 25. http:www.ambri.com/storage/documents/ ambri-brochure.pdf
[65] Sandia National Laboratories. 2013. “Charleston Energy Storage Project. “DOE International Energy Storage Database. Last modified April 1.http://www.energystorageexchange.org/projects/182.
[66] William Tahil 2007. “The trouble with lithium.” Implications of Future PHEV Production for Lithium Demand.” January. Meridian International Research. http ://www.inference. phy.cam.ac.uk /sustainable /refs/ nuclear/Trouble Lithium. Pdf
[67] Prachi Patel and Brian Perusse. 2012. “Grid battery storage gets big in the States.” MRS bulletin 37, no. 11.
[68] J. Beurskens and P.H. Jensen, “Economics of Wind Energy Prospects and Directions,” Renewable Energy World 4 (2001) 103–121.
[69] Bonefeld, J. and Jensen, J.N. (2002) Horns Rev-160 MW offshore wind’, Renewable Energy World, Vol. 5, No. 3,pp. 77-87.
[70] Enercon Wind Diesel Electric System, http://www.enercon.de.
[71] R.C. Bansal and T.S. Bhatti, “Small Signal Analysis of Isolated Hybrid Power Systems: Reactive Power and Frequency Control Analysis” Alpha Science International U.K. & Narosa Publishers, New Delhi, 2008.
180
[72] R. Hunter and G. Elliot, Wind-Diesel Systems, “A Guide to the Technology and Its Implementation” Cambridge University Press, 1994.
[73] R.C. Bansal, “Automatic reactive power control of isolated wind-diesel hybrid power systems,” IEEE Transactions on Industrial Electronics, 2006.p.1116–1126
[74] G.M. Masters, Renewable and Efficient Electrical Power Systems (John Wiley & Sons, Inc., Ho, New Jersey, 2004).
[75] T.S. Jayadev, “Windmills Stage a Comeback,” IEEE Spectrum (1976) 45–49.
[76] A.A.F. Al-Ademi, “Load-frequency control of stand-alone hybrid power systems based on renewable energy sources,” Ph.D Thesis, Centre for Energy Studies, Indian Institute of Tech- nology, Delhi (1996).
[77] R. Karki, et al., “A simplified wind power generation model for reliability evaluation”, IEEE Transactions on Energy Conversion, 2006. 21(2): p. 533-540.
[78] R. Karki, et al., “Reliability Evaluation Considering Wind and Hydro Power Coordination”, IEEE Transactions on Power Systems, 2010. 25(2): p. 685-693.
[79] Phoon Hee Yau, “Generation System Reliability Evaluations with Intermittent Renewables”.2006, University of Strathclyde.
[80] P. Bresesti, et al. “Probabilistic model for adequacy evaluation of electric networks withsizeable wind power generation”, in Power Systems Conference and Exposition, 2004. IEEE PES.2004.
[81] M.R Milligan, “A Sliding Window Technique for Calculating System LOLP Contributions of Wind Power Plants”, in AWEA Wind power 2001. 2001: Washington DC.
[82] William Carr, “Adequacy assessment of the Irish generation system including wind capacity”.2005, Dublin Institute of Technology.
181
[83] R. Billinton and Gao Yi, “Multistate Wind Energy Conversion System Models for Adequacy Assessment of Generating Systems Incorporating Wind Energy”. IEEE Transactions on Energy Conversion, 2008. 23(1): p. 163-170.
[84] S. Kyung, K.J McKenzie, Y. Liu, and S.Atcitty, “A study on applications of energy storage for the wind power operation in power systems,” Power Eng. Soc. General Meeting, 2006.IEEE, pp. 5
[85] W. Li, and G. Joós, “Comparison of energy storage system technologies and configurations in a wind farm,” IEEE Power Electronics Specialists Conf., PESC 2007, June 2007 pp. 1280 – 1285
[86] Brendan Fox, Damian Flynn, Leslie Bryans, Nick Jenkins, David Miborrow, Mark O’Malley, Richard Watson and Olimpo Anaya-Lara, “Wind power integration connection and system operational aspects, London,” The Institution of Engineering and Technology, 2007.
[87] H.T. Yang, P.C. Yang, and C.L. Huang, “Evolutionary programming based economic dispatch for units with non-smooth fuel cost functions,” IEEE Transactions on Power Systems, vol. 11, no. 1, Feb. 1996, pp. 112-118.
[88] N. Sinha, R. Chakrabarti, and P.K. Chattopadhyay, “Evolutionary programming techniques for economic load dispatch,” IEEE Transactions on Evolutionary Computation, vol. 7, no. 1, Feb. 2003, pp. 83-94.
[89] K.P. Wong, and J. Yuryevich, “Evolutionary-programming-based algorithm for environmentally-constrained economic dispatch,” IEEE Transactions on Power Systems, vol. 13, no. 2, May 1998, pp. 301-306.
[90] K.P. Wong, and C.C. Fung, “Simulated annealing based economic dispatch algorithm,” IEE Proceedings of Generation, Transmission and Distribution, vol. 140, no. 6, Nov. 1993, pp. 509-515.
182
[91] W.M. Lin, F.S. Cheng, and M.T. Tsay, “An improved Tabu search for economic dispatch with multiple minima,” IEEE Transactions on Power Systems, vol. 17, no. 1, Feb. 2002, pp. 108-112. 121
[92] J.S. Al-Sumait, A.K. AL-Othman, and J.K. Sykulski, “Application of pattern search method to power system valve-point economic load dispatch,” International Journal of Electrical Power and Energy Systems, vol. 29, no. 10, 2007, pp. 720-730.
[93] D.C. Walter, and G.B. Sheble, “Genetic algorithm solution of economic dispatch with valve point loading,” IEEE Transactions on Power Systems, vol. 8, no. 3, Aug. 1993, pp. 1325-1332.
[94] K.S. Swarup, and S. Yamashiro, “Unit commitment solution methodologies using genetic algorithm,” IEEE Transactions on Power Systems, vol. 17, no .1, Feb. 2002, pp. 87-91.
[95] L.S. Coelho, and V.C. Mariani, “Combining of chaotic differential evolution and quadratic programming for economic dispatch optimization with valve-point effect,” IEEE Transactions on Power Systems, vol. 21, no. 2, May 2006, pp. 989-996.
[96] J.G. Vlachogiannis, and K.Y. Lee, “Economic load dispatch-A comparative study on heuristic optimization techniques with an improved coordinated aggregation-based PSO,” IEEE Transactions on Power Systems, vol. 24, no. 2, May 2009, pp. 991-1001.
[97] K.P. Wong, and Y.W. Wong, “Combined Genetic Algorithm/Simulated Annealing/Fuzzy Set approach to short-term generation scheduling with take-or-pay fuel contract,” IEEE Transactions on Power Systems, vol. 11, no. 1, Feb. 1996, pp. 128-136.
[98] R. M. Burns and C. A. Gibson, “Optimization of priority lists for a unit commitment program,” IEEE PES Summer Meeting, 1975.
[99] T. Senjyu, T. Miyagi and A. Saber, “Emerging solution of large-scale unit commitment problem by stochastic priority list,” Electric Power Systems Research, pp. 283-292, 2006.
183
[100] Z. Ouyang and S. M. Shahidehpour, “An intelligent dynamic programming for unit 122commitment application,” IEEE Transaction on Power System, pp. 1203-1209, 1991.
[101] A. I. Cohen and M. Yoshimura, “A branch-and-bound algorithm for unit commitment,” IEEE Trans. Power App. Syst, pp.444-451, 1983.
[102] H. Daneshi, A. L. Choobbari, S. M. Shahidehpour and Z. Li, “Mixed integer programming methods to solve security constrained unit commitment with restricted operating zone limits,” IEEE Conference on EIT, pp. 187-192, 2008.
[103] S. J. Wang, S. M. Shahidehpour, D. S. Kirschen, S. Mokhtari and G. D. Irisarri, “Short-term generation scheduling with transmission and environmental constraints using an augmented Lagrangian relaxation,” IEEE Trans. Power Systems, pp. 1294-1301, 1995.
[104] W. Ongsakul and N. Petcharaks, “Unit commitment by enhanced adaptive Lagrangian relaxation,” IEEE Trans. Power System, pp.620-628, 2004.
[105] S. A. Kazarlis, A. G. Bakirtzis and V. Petridis, “A genetic algorithm solution to the unit commitment problem,” IEEE Trans. Power Systems, pp83-92, 1996.
[106] D. Dasgupta and D. R. McGregor, “Thermal unit commitment using genetic algorithms,” IEE Proc. C Generation, Transmission and Distribution, pp. 459-465, 1994.
[107] G. B Sheble and T. T. Maifeld, “Unit commitment by genetic algorithm and expert system,” Electric power system research, pp. 115-121, 1994.
[108] D. Srinivasan and J. Chazelas, “A priority list-based evolutionary algorithm to solve large scale unit commitment problem,” IEEE PowerCon, pp. 1746-1751, 2004.
[109] K. A. Juste, H. Kita, E. Tanaka and J. Hasegawa, “An evolutionary programming solution to the unit commitment problem,” IEEE Trans. Power System,pp.1452-1459, 1999.
184
[110] L. G. Zwe, “Discrete particle swarm optimization algorithm for unit commitment,” 123 IEEE Power Engineering Society General Meeting. Pp. 418-424, 2003.
[111] X. Yuan, H. Nie, A. Su, L. Wang and Y. Yuan, “AN improved binary particle swarm optimization for unit commitment problem,” Expert System with Applications, pp. 8049-8055, 2009.
[112] A. Y. Saber, T. Senjyu, N. Urasaki and T. Funabashi, “Unit commitment computation-a novel fuzzy adaptive particle swarm optimization approach power systems conference and exposition,” IEEE PSCE, pp. 1820-1828, 2006.
[113] T. Ting, M. Rao and C. Loo, “Solving unit commitment problem using hybrid particle swarm optimization,” Journal of heuristics, pp. 507-520, 2006.
[114] C. P. Cheng, C. W. Liu and C. C. Liu, “Unit commitment by Lagrangian relation and genetic algorithms,” IEEE Trans. Power Systems, pp. 707-714, 2000.
[115] H. Balci and J. Valenzuela, “Scheduling electric power generations using particle swarm optimization combined with Lagrangian relaxation method,” Journal of appl. Math. Comput, Sci., pp. 411-421. 2004.
[116] T. Logenthiran and D. Srinivasan, “Short term generation scheduling of a microgrid,” IEEE TENCON, pp. 1-6, 2009.
[117] J.H. Holland, “Adaptation in Natural and Artificial Systems”, The University of Michigan Press, 1975.
[118] H. Bersini, and F. Varela, “Hints for adaptive problem solving gleaned from immune networks,” Parallel Problem Solving from Nature, vol. 496, 1991, pp. 343-354.
[119] D. Dasgupta and S. Forrest, “An anomaly detection algorithm inspired by the immune system,” Artificial Immune System and Their Applications, Berlin: Springer-Verlag. 1998, pp. 262-277.
[120] K. Meng, Z.Y. Dong, X. Yin, and K.P. Wong, “Electricity market clearing price forecasting,” invited book chapter in L. B. Shi edit.
185
Computational Intelligence in Power Systems, Springer, 2009, pp. 89-106.
[121] J. Kennedy and R. Eberhart, “Particle swarm optimization,” IEEE International Conference on Neural Networks, vol. 4, 1995, pp. 1942-1948.
[122] J. Kennedy and R. Eberhart, “Swarm Intelligence”, Morgan Kaufmann Publishers, 2001.
[123] H. Yoshida, K. Kawata, Y. Fukuyama, S. Takayama, and Y. Nakanishi, “A particle swarm optimization for reactive power and voltage control considering voltage security assessment,” IEEE Transactions on Power Systems, vol. 15, no. 4, Nov. 2000, pp. 1232-1239.
[124] M.A. Abido, “Optimal design of power system stabilizers using particle swarm optimization,” IEEE Transactions on Energy Conversion, vol. 17, no. 3, Sep. 2002, pp. 406-413.
[125] J.B. Park, K.S. Lee, J.R. Shin, and K.Y. Lee, “A particle swarm optimization for economic dispatch with nonsmooth cost functions,” IEEE Transactions on Power Systems, vol. 20, no. 1, Feb. 2005, pp. 34-42.
[126] J. Z. Zhu, “Optimization of power system operation”. Piscataway, N.J.:Wiley, 2009.
[127] K. P. Wong and Y. W. Wong “Thermal generator scheduling using hybrid genetic/simulated-annealingapproach,” IEE proceeding-Generation, Transmission and Distribution, vol. 142, pp. 372-380, 2005.
[128] K. P. Wong and S.Y. Wong "Combined genetic algorithm/simulated annealing/fuzzy set approach to short-term generation scheduling with take-or-pay fuel contract." IEEE Trans. Power Syst, Vol.11, no. 1, 1996:pp. 112-118.
[129] J. H. Zhao, Z. Y. Dong, X. Li, and K. P. Wong, "A framework for electricity price spike analysis with advanced data mining methods," IEEE Trans. Power Syst., vol. 22, pp. 376-385, 2007.
186
[130] J. H. Zhao, F. S. Wen, Z. Y. Dong, Y. S. Xue, and K. P. Wong, "Optimal dispatch of electric vehicles and wind power using enhanced particle swarm optimization," IEEE Trans. Ind. Informat., vol. 8, pp. 889-899, 2012.
[131] V. C. Gungor, D. Sahin, T. Kocak, S. Ergut, C. Buccella, C. Cecati, andG. P. Hancke, “Smart grid technologies: communication technologies and standards,” IEEE Trans. Ind. Informat., vol. 7, no. 4, pp. 529–539,Nov. 2011.
[132] Z. Y. Dong, K. P. Wong, K. Meng, F. J. Luo, F. Yao, and J. H. Zhao,“Wind power impact on system operations and planning,” in IEEE PESGen. Meeting, Minneapolis, MN, Jul. 2010.
[133] M. R. Patel, “Wind and solar power systems”. Boca Raton, FL: CRCPress, 2006.
[134] J. Hetzer, D. C. Yu, and K. Bhattrarai, “An economic dispatch model incorporating wind power,” IEEE Trans. Energy Convers., vol. 23, no.2, pp. 603–611, Jun. 2008.
[135] X. Liu and W. Xu, “Minimum emission dispatch constrained by stochastic wind power availability and cost,” IEEE Trans. Power Syst., vol.25, no. 3, pp. 1705–1713, Aug. 2010.
[136] F. Yao, Z. Y Dong, K. Meng, Z. Xu, H. Iu, and K. P. Wong, "Quantum-inspired particle swarm optimization for power system operations considering wind power uncertainty and carbon tax in Australia", IEEE Trans on Industrial Informatics, vol.8, pp. 880-888, 2012.
[137] K. P. Wong and C. C. Fung, "Simulated annealing based economic dispatch algorithm," IEE Proc.Gener.Transm.Distrib., vol. 140, pp. 509-515, 1993.
[138] K. P. Wong and Y. W. Wong, "Genetic and genetic/simulated-annealing approaches to economic dispatch", IEE Proc.Gener.Transm.Distrib, vol. 141, pp. 507-513, 1994.
[139] J. G. Vlachogiannis and K. Y. Lee, “Quantum-inspired evolutionary algorithm for real and reactive power dispatch,” IEEE Trans. PowerSyst., vol. 23, no. 4, pp. 1627–1636, Nov. 2008.
187
[140] M. S. Tsai and F. Y. Hsu, “Application of grey correlation analysis in evolutionary programming for distribution system feeder reconfiguration,”IEEE Trans. Power.Syst., vol. 25, no. 2, pp. 1126–1133, May2010.
[141] X. N. Lin, S. H. Ke, Z. T. Li, H. L.Weng, and X. H. Han, “A fault diagnosis method of power systems based on improved objective function and genetic algorithm-tabu search,” IEEE Trans. Power Delivery, vol.25, no. 3, pp. 1268–1274, Jul. 2010.
[142] J. S. Al-Sumait, A. K. AL-Othman, and J. K. Sykulski, “Application of pattern search method to power system valve-point economic load dispatch,” Electr. Power Energy Syst., vol. 29, no. 10, pp. 720–730,Dec. 2007.
[143] S. H. Ling, H. H. C. Iu, F. H. F. Leung, and K. Y. Chan, “Improved hybrid particle swarm optimized wavelet neural network for modeling the development of fluid dispensing for electronic packaging,” IEEE Transaction on Industrial Electron., vol. 55, no. 9, pp. 34–42, Sep. 2008.
[144] K. Meng, H.G. Wang, Z.Y. Dong, and K.P.Wong, “Quantum-inspired particle swarm optimization for valve-point economic load dispatch,”IEEE Trans. Power Syst., vol. 25, no. 1, pp. 215–222, Feb. 2010.
[145] F. Yao, K. Meng, X. Zhao, Z. Y. Dong, H. H. C. Iu, and K.P. Wong, "Differential evolution algorithm for multi-objective economic load dispatch considering minimum emission costs," in Power and Energy Society General Meeting, 2011 IEEE, 2011, pp. 1-5.
[146] N. Duvvuru and K. S. Swarup, “A hybrid interior point assisted differential evolution algorithm for economic dispatch,” IEEE Trans. Power Syst., vol. 26, no. 1, pp. 541–549, Feb. 2011.
[147] P. Attaviriyanupap, H. Kita, E. Tanaka, and J. Hasegawa, “A hybrid EP and SQP for dynamic economic dispatch with non-smooth fuel cost function,” IEEE Trans. Power Syst., vol. 17, no. 2, pp. 411-416, May. 2002.
188
[148] S.K. Wang, J.P. Chiou, and C.W. Liu, “Non-smooth/non-convex economic dispatch by a novel hybrid differential evolution algorithm,” IETGen. Trans. Distrib., vol. 1, no. 5, pp. 793-803, Sep. 2007.
[149] A. Bhattacharya and P. K. Cattopadhyay, “Hybrid differential evolution with biogeography based optimization for solution of economic load dispatch,” IEEE Trans. Pwr. Syst., vol.25, no.4, pp.1955-1964, Nov. 2010.
[150] G. M. Masters, “Renewable and Efficient Electric Power Systems”. Hoboken, NJ: Wiley, 2005.
[151] J. A. Carta, P. Ramirez, and S. Velazquez, “A review of wind speed probability distributions used in wind energy analysis: Case studies in the Canary Islands,” Renew. Sust.Energ. Rev., vol. 13, no. 5, pp.933–955, Jun. 2009.
[152] X. Liu, “Economic load dispatch constrained by wind power availability: A wait-and-see approach,” IEEE Trans. Smart Grid, vol. 1, no.4, pp. 347–355, Dec. 2010.
[153] A. E. Olsson, “Particle Swarm Optimization”: Theory, Techniques and Applications. New York: Nova Science, 2011.
[154] Office of the Gas and Electricity Markets, last accessed 2012. [Online]. Available:http://www.ofgem.gov.uk/Sustainability/Environment/RenewablObl/Pages/RenewablObl.aspx
[155] P.T. Boggs and J.W. Tolle, “Sequential quadratic programming,” ActaNum., vol. 4, pp. 1-51, 1995.
[156] S.P. Han, “A globally convergent method for nonlinear programming,” J. Optim. Theory Appl., vol. 22, no. 3, pp. 297-309, Aug. 1975.
[157] M.J.D. Powell, “A fast algorithm for nonlinearly constrained optimization calculations,” Numer.Anal., G.A. Watson ed., Lecture Notes in Mathematics, Springer Verlag, vol. 630, 1978.
189
[158] J. Kennedy and R. Eberhart, “Particle swarm optimization,” Proc. IEEEInt. Conf. Neural Netw., pp. 1942-1948, 1995.
[159] N. Duvvuru and K.S. Swarup, “A hybrid interior point assisted differential evolution algorithm for economic dispatch,” IEEE Trans. Power Syst.,vol. 26, no. 1, Feb. 2011
[160] P. Venkatesh and K. Y. Lee, “Multi-objective evolutionary programming for economic emission dispatch problem,” in Proc. IEEE PES Gen. Meeting, Pittsburgh, PA, Jul. 2008
[161] Australian Bureau of Meteorology, last accessed 2012. [Online].
[162] J.D. Farmer, N. Packard and A. Perelson, “The immune system, adaptation and machine learning,” Phys. D, vol. 2, no. 1-3, pp. 187-204, 1986.
[163] X. Liu and W. Xu, “Economic load dispatch constrained by wind power availability: A here-and-now approach,” IEEE Transactions on Sustainable Energy, vol. 1, pp. 2–9, 2010.
[164] Y. Xu, Z. Y. Dong, Z. Xu, K. Meng, and K. P. Wong, “An intelligent dynamic security assessment framework for power systems with wind power,” IEEE Transactions on Industrial Informatics, vol. 8, pp. 995–1003, 2012.
[165] J. H. Zhao, Z. Y. Dong, X. Li, and K. P. Wong, “A framework for electricity price spike analysis with advanced data mining methods,” IEEE Transactions on Power Systems, vol. 22, no. 1, pp. 376–385, 2007.
[166] V. Gungor, D. Sahin, T. Kocak, S. Ergut, C. Buccella, C. Cecati, and G. Hancke, “Smart grid technologies: Communication technologies and standards,” IEEE Transactions on Industrial Informatics, vol. 7, no. 4, pp. 529–539, 2011.
[167] Q. Li, S. Choi, Y. Yuan, and D. Yao, “On the determination of battery energy storage capacity and short-term power dispatch of a wind farm,” IEEE Transactions on Sustainable Energy, vol. 2, pp. 148–158, 2011.
190
[168] W. Su, H. Eichi, W. Zeng, and M.-Y. Chow, “A survey on the electrification of transportation in a smart grid environment,” IEEE Transactions on Industrial Informatics, vol. 8, pp. 1–10, 2012.
[169] R. Billinton, B. Karki, R. Karki, and G. Ramakrishna, “Unit commitment risk analysis of wind integrated power systems,” IEEE Transactions on Power Systems, vol. 24, no. 2, pp. 930–939, 2009.
[170] J. Restrepo and F. Galiana, “Assessing the yearly impact of wind power through a new hybrid deterministic/stochastic unit commitment,” IEEE Transactions on Power Systems, vol. 26, no. 1, pp. 401–410, 2011.
[171] Y. Makarov, P. Etingov, J. Ma, Z. Huang, and K. Subbarao, “Incorporating uncertainty of wind power generation forecast into power system operation, dispatch, and unit commitment procedures,” IEEE Transactions on Sustainable Energy, vol. 2, no. 4, pp. 433–442, 2011.
[172] S. Kazarlis, A. Bakirtzis, and V. Petridis, “A genetic algorithm solution to the unit commitment problem,” IEEE Transactions on Power Systems, vol. 11, no. 1, pp. 83–92, 1996.
[173] D. Dasgupta and D. McGregor, “Thermal unit commitment using genetic algorithms,” IEE Proceedings-Generation, Transmission and Distribution, vol. 141, no. 5, pp. 459–465, 1994
[174] G. Sheble and T. Maifeld, “Unit commitment by genetic algorithm and expert system,” Electric Power Systems Research, vol. 30, pp. 115 –121, 1994.
[175] T. W. Lau, C. Chung, K. P. Wong, T. S. Chung, and S. Ho, “Quantum inspired evolutionary algorithm approach for unit commitment,” IEEE Transactions on Power Systems, vol. 24, no. 3, pp. 1503–1512, 2009
[176] A. H. Mantawy, Y. Abdel-Magid, and S. Selim, “A simulated annealing algorithm for unit commitment,” IEEE Transactions on Power Systems, vol. 13, pp. 197–204, 1998.
[177] M. Rashidi, M. Ali, N. Freidoonimehr, and F. Nazari, "Parametric analysis and optimization of entropy generation in unsteady MHD
191
flow over a stretching rotating disk using artificial neural network and particle swarm optimization algorithm," Energy, vol. 55, pp. 497-510, 2013
[178] M. Rashidi, O. A. Bég, A. B. Parsa, and F. Nazari, "Analysis and optimization of a transcritical power cycle with regenerator using artificial neural networks and genetic algorithms," Proceedings of the Institution of Mechanical Engineers, Part A: Journal of Power and Energy, p. 0957650911407700, 2011.
[179] S. Shamshirband, D. Petković, R. Hashim, S. Motamedi, and N. B. Anuar, "An appraisal of wind turbine wake models by adaptive neuro-fuzzy methodology," International Journal of Electrical Power & Energy Systems, vol. 63, pp. 618-624, 2014.
[180] A. E. Jeyakumar, “Hybrid pso-sqp for economic dispatch with valvepoint effect,” Electric Power Systems Research, vol. 71, pp. 51 – 59, 2004
[181] Y. Zhang, F. Yao, H. H. C. Iu, T. Fernando, and K. P. Wong, "Sequential quadratic programming particle swarm optimization for wind power system operations considering emissions," Journal of Modern Power Systems and Clean Energy, vol. 1, pp. 231-240, 2013.
[182] J. P. S. Catalao, H. M. I. Pousinho, and V. M. F. Mendes, "Optimal offering strategies for wind power producers considering uncertainty and risk," IEEE Systems Journal, vol. 6, pp. 270-277, 2012.
[183] A. Borghetti, M. Bosetti, S. Grillo, S. Massucco, C. A. Nucci, M. Paolone, et al., "Short-term scheduling and control of active distribution systems with high penetration of renewable resources," IEEE Systems Journal, vol. 4, pp. 313-322, 2010.
[184] A. K. Srivastava, A. A. Kumar, and N. N. Schulz, "Impact of distributed generations with energy storage devices on the electric grid," IEEE Systems Journal, vol. 6, pp. 110-117, 2012.
[185] S. W. Mohod and M. V. Aware, "micro wind power generator with battery energy storage for critical load," IEEE Systems Journal, vol. 6, pp. 118-125, 2012.
192
[186] S. Teleke, M. E. Baran, A. Q. Huang, S. Bhattacharya, and L. Anderson, "Control strategies for battery energy storage for wind farm dispatching," IEEE Transactions on Energy Conversion, vol. 24, pp. 725-732, 2009.
[187] K. Heussen, S. Koch, A. Ulbig, and G. Andersson, "Unified System-Level Modeling of Intermittent Renewable Energy Sources and Energy Storage for Power System Operation," IEEE Systems Journal, vol. 6, pp. 140-151, 2012.
[188] S. Surender Reddy, P. R. Bijwe, and A. R. Abhyankar, "Real-time economic dispatch considering renewable power generation variability and uncertainty over scheduling period," IEEE Systems Journal, vol. PP, pp. 1-12, 2014.
[189] T. Niknam and F. Golestaneh, "Enhanced Bee Swarm Optimization algorithm for dynamic economic dispatch," IEEE Systems Journal, vol. 7, pp. 754-762, 2013.
[190] T. Niknam, R. Azizipanah-Abarghooee, M. Zare, and B. Bahmani-Firouzi, "Reserve constrained dynamic environmental/economic dispatch: a new multiobjective self-adaptive learning bat algorithm," IEEE Systems Journal, vol. 7, pp. 763-776, 2013.
[191] B. Mohammadi-Ivatloo, A. Rabiee, and A. Soroudi, "Nonconvex dynamic economic power dispatch problems solution using hybrid immune-genetic algorithm," IEEE Systems Journal, vol. 7, pp. 777-785, 2013.
[192] A. Y. Saber and G. K. Venayagamoorthy, "Resource scheduling under uncertainty in a smart grid with renewables and plug-in vehicles," IEEE Systems Journal, vol. 6, pp. 103-109, 2012.
[193] T. Niknam, F. Golestaneh, and M. S. Sadeghi, "θ-multiobjective-teaching-learning-based optimization for dynamic economic emission dispatch,," IEEE Systems Journal, vol. 6, pp. 341-352, 2012.
[194] M. Qadrdan, W. Jianzhong, N. Jenkins, and J. Ekanayake, "Operating strategies for a GB integrated gas and electricity network
193
considering the uncertainty in wind power forecasts," IEEE Transactions on Sustainable Energy, vol. 5, pp. 128-138, 2014.
[195] M. S. Lu, C. L. Chang, W. J. Lee, and L. Wang, "Combining the wind power generation system with energy storage Equipment," IEEE Transactions on Industry Applications, vol. 45, pp. 2109-2115, 2009.
[196] S. Teleke, M. E. Baran, S. Bhattacharya, and A. Q. Huang, "Rule-based control of battery energy storage for dispatching intermittent renewable sources," IEEE Transactions on Sustainable Energy, vol. 1, pp. 117-124, 2010.
[197] H. Ibrahim, A. Ilinca, and J. Perron, "Energy storage systems—characteristics and comparisons," Renewable and Sustainable Energy Reviews, vol. 12, pp. 1221-1250, 6 2008.
[198] D. L. Yao, S. S. Choi, K. J. Tseng, and T. T. Lie, "A statistical approach to the design of a dispatchable wind power-battery energy storage system," IEEE Transactions on Energy Conversion, vol. 24, pp. 916-925, 2009.
[199] S. X. Chen, H. B. Gooi, and M. Q. Wang, "Sizing of energy storage for microgrids," IEEE Transactions on Smart Grid, vol. 3, pp. 142-151, 2012.
[200] B. Hartmann and A. Dan, "Cooperation of a grid-connected wind farm and an energy storage-demonstration of a simulation tool," IEEE Transactions on Sustainable Energy, vol. 3, pp. 49-56, 2012.
[201] K. Yoshimoto, T. Nanahara, and G. Koshimizu, "New control method for regulating state-of-charge of a battery in hybrid wind power/battery energy storage system," in IEEE Power Systems Conference and Exposition 2006, pp. 1244-1251.
[202] M. N. Kabir, Y. Mishra, G. Ledwich, Z. Y. Dong, and K. P. Wong, "Coordinated control of grid-connected photovoltaic reactive power and battery energy storage systems to improve the voltage profile of a residential distribution feeder," IEEE Transactions on Industrial Informatics, vol. 10, pp. 967-977, 2014.
194
[203] C. H. Yang, G. Zhabelova, C. W. Yang, and V. Vyatkin, "Cosimulation environment for event-driven distributed controls of smart grid," IEEE Transactions on Industrial Informatics, vol. 9, pp. 1423-1435, 2013.
[204] S. Jun, V. Palade, X. Wu, W. Fang, and Z. Wang, "Solving the power economic dispatch problem with generator constraints by random drift particle swarm optimization," IEEE Transactions on Industrial Informatics, vol. 10, pp. 222-232, 2014
[205] G. Binetti, A. Davoudi, D. Naso, B. Turchiano, and F. L. Lewis, "A distributed auction-based algorithm for the nonconvex economic dispatch problem," IEEE Transactions on Industrial Informatics, vol. 10, pp. 1124-1132, 2014.
[206] P. Garcia, C. A. Garcia, L. M. Fernandez, F. Llorens, and F. Jurado, "ANFIS-based control of a grid-connected hybrid system integrating renewable energies, hydrogen and batteries," IEEE Transactions on Industrial Informatics, vol. 10, pp. 1107-1117, 2014.
[207] M. Jongerden, A. Mereacre, H. Bohnenkamp, B. Haverkort, and J. Katoen, "Computing optimal schedules of battery usage in embedded systems," IEEE Transactions on Industrial Informatics, vol. 6, pp. 276-286, 2010.
[208] B. Hartmann and A. Dan, "Cooperation of a grid-connected wind farm and an energy storage-demonstration of a simulation tool," IEEE Transactions on Sustainable Energy, vol. 3, pp. 49-56, 2012.
[209] J.G. Vlachogiannis and K.Y. Lee, “Quantum-inspired evolutionary algorithm for real and reactive power dispatch,” IEEE Trans. Power Syst., vol. 23, no.4, Nov. 2008, pp. 1627-1636.
[210] K.H. Han and J.H. Kim, “Quantum-inspired evolutionary algorithm for a class of combinatorial optimization,” IEEE Trans. Evol. Comput, vol. 6, no.6, Dec. 2002, pp. 580-593.
[211] X.F. Yan, D.Z. Chen, and S.X. Hu, “Chaos-genetic algorithm for optimizing the operating conditions based on RBF-PLS model,” Comput. Chem. Eng., vol. 27, Oct. 2003, pp. 1393-1404
195
[212] M. Gibbard and D. Vowles, "Simplified 14-generator model of the SE Australian power system," The University of Adelaide, South Australia, 2008.