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


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


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


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


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


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


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


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.


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.


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 .


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


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.


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 .


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.


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


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


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.

Power System Operation Optimization Cooperate Wind Power … · Power System Operation Optimization...

Documents

Transcript of Power System Operation Optimization Cooperate Wind Power … · Power System Operation Optimization...