Eleanor Denny - A Cost-Benefit Analysis of Wind Power PDF

A Cost Benefit Analysis
of Wind Power
by
Eleanor Denny B.A., M.B.S.
Thesis submitted for the degree of

Philosophiae Doctor
from the
School of Electrical, Electronic and Mechanical Engineering
National University of Ireland
University College Dublin, Ireland
Supervisor of Research: Prof. M. O’Malley

Co-Supervisor of Research: Prof. J. Fitz Gerald
Nominating Professor: Prof. M. O’Malley
January 19, 2007

Abstract
Concern over global warming has led policy makers worldwide to accept the im-
portance of reducing greenhouse gas emissions. As wind generation does not itself
create any harmful emissions, policy makers often promote it as a means of reduc-
ing national emission levels while also meeting renewable obligations. In fact, wind
generation has been the fastest growing form of renewable energy in Europe and the
United States in the past decade. Like other forms of generation reliant on underly-
ing weather conditions, wind generation output is variable, i.e. the output of these
units depends upon weather conditions that cannot be controlled by the operator
of the generator. As well as being variable, wind generation also faces a challenge
of being relatively unpredictable. Since the underlying energy source cannot be di-
rectly controlled, the renewable generation is high when conditions are favourable
and low when unfavourable.
The function of power system operators is to supply electricity to customers
in a reliable manner at a sustainable cost. This involves ensuring that the gener-
ation meets the demand at all times and that any short term gaps between load
and generation are bridged quickly and precisely to maintain the integrity of the
power system. Since the output of some renewable generators, in particular wind
generation, cannot be actively controlled and is difficult to predict accurately, this
balancing of generation and load can become more challenging as renewable pene-
i
ii
trations increase. Thus, the cost of operating the power system can increase. The
work presented here discusses in detail the costs imposed on the power system by
wind generation and also the benefits than can accrue from added wind generation.
The costs included are the capital and operating costs of wind generation, the deep
network reinforcement costs, the additional reserve costs and the costs associated
with additional cycling of conventional units. The benefits are the capacity benefit
of wind, the emissions savings and the savings in fuel costs.
In order to investigate the costs and benefits of wind generation, it is necessary
to analyse wind in the context of the power system within which it is installed.
Thus, a case study of a real electricity system is conducted and a thorough model
of the power system is developed. This model determines the operating levels of
conventional generators as penetrations of wind generation increase. The resulting
operating schedules are then analysed to determine wind’s impact on the operation
of conventional generators and their resulting cycling costs, emissions and fuel costs.
The costs and benefits are combined to generate net benefit curves which show
the wind penetrations where the costs exceed the benefits, beyond which no further
investment should be made in wind generation. Three test years are examined and
the sensitivity of the net benefits to a large range of assumptions is tested. It is
found that increased interconnection, high CO2 prices and a flexible plant mix are
particularly beneficial for wind generation, and that there are positive net benefits
for wind energy penetrations of 17% and higher under the chosen set of assumptions
in the case study.
A number of applications relating to the case study were conducted in this anal-
ysis and these are also discussed in detail. The impact of carbon prices on the merit
order of generators and their resulting cycling costs is examined. The results show
that by introducing a carbon price, the cycling costs can actually exceed the value
of the saved CO2 , and the effect on cycling costs is exacerbated by the addition of
wind generation. An alternative method of operating a power system with wind
generation known as the fuelsaver approach is investigated as are a range of connec-
tion policy options. The impacts of tidal generation on power system operation are
also investigated.
Acknowledgements
I would like to thank everyone who helped and supported me throughout my PhD,
in particular, special thanks goes to the following people.
Professor Mark O’ Malley, whose guidance, support and encouragement has been
invaluable throughout this project. Mark’s commitment to my PhD has been an
inspiration from the very outset and despite having the most demanding of work
schedules, he always makes time for his PhD students. I am extremely grateful for
all he has done for me, without him I would not have completed this PhD project.
I am greatly indebted to him. Thank you.
Professor John Fitz Gerald of the Economic and Social Research Institute whose
advice and guidance on all things economic was invaluable in completing this thesis.
John devoted a lot of time and effort to my project and for this I am truly grateful.
Dr. Laura Malaguzzi Valeri of the Economic and Social Research Institute for her
many useful discussions, interactions and suggestions.
Colleagues in industry, for the provision of vital data, advice and observations,
particularly Michael O’Mahony from ESB Power Generation, Donal Phelan in ESB
Networks, Michael Kelly in EirGrid, Morgan Bazillian in SEI and Dermot O’Kane
from Airtricity.
iii
iv
Ronan Doherty and Garth Bryans with whom I collaborated on some of the work
presented in this thesis.
All my new postgrad friends especially to Andrew Keane for his help, advice, en-
couragement, and friendship. Also particular thanks goes to Aidan Tuohy for proof
reading this thesis. Thanks all round to the occupants of room 157 over the last
four years, Ciara, Gill, Shane, Rebecca, Ronan D, Alan, Hugh, Garth, Tim, Ronan,
Emma and Daniel for all the moral support, the tea breaks, the stories and for
wholeheartedly welcoming an economist into the world of engineering!
Rose Mary Logue who works tirelessly to look after us postgraduates. Thanks for
all the encouragement over the last four years.
All my friends for putting up with me and supporting my social life for the last
four years! You’ve been great, and at last I might have an answer to the dreaded
question ‘so when will you be finished?’ !
My fantastic family, Michael, Frances, Conor and Clı́odhna. Thank you for your
constant support and encouragement, not just over the last four years but in ev-
erything I do. Thanks also to my extended family, in particular Granddad Paddy
who is an inspiration to everyone he meets and Sarah with whom I lived for a most
enjoyable year at the start of my PhD and who can throw a cracking party! I love
you all. Thank you.
And finally Eoin, for the endless encouragement, support, humour, patience and love
which have guided me through the last number of years. Thank you for everything.
Publications Arising from Thesis
Journal Papers:
1. E. Denny, L. Malaguzzi Valeri, J. Fitz Gerald & M. O’Malley, “Carbon Prices

and Asset Degradation - A Costly Combination for Electric Power Systems”,
The Energy Journal (in review), January 2007. Also to be presented at IAEE
International Energy Conference, Wellington, New Zealand, Feb. 2007.
2. E. Denny & M. O’Malley, “Quantifying the Total Net Benefits of Grid Inte-
grated Wind”, IEEE Transactions on Power Systems (in press), Nov 2006.
3. A. Keane, E. Denny, and M. OMalley, Quantifying the Impact of Connection

Policy on Distributed Generation, IEEE Transactions on Energy Conversion,
vol. 22, no. 1, Mar. 2007. Also to be presented at IEEE PES General Meeting,
Tampa, Florida, June 2007.
4. E. Denny & M. O’Malley, “Wind Generation, Power System Operation and

Emissions Reduction”, IEEE Transactions on Power Systems, vol. 21, no. 1,
pp. 341-347, 2006.
5. M. Bazilian, E. Denny, & M. O’Malley, “Challenges of Increased Wind Energy

Penetration in Ireland”, Wind Engineering Journal, vol. 28, no. 1, pp. 43-56,
2003.
6. A. Keane, Q. Zhou, E. Denny, J. Bialek, and M. O’Malley, ”A Novel Curtail-

ment Method for Distributed Generation Voltage Management”, IEEE Trans-
actions on Power Systems, in preparation, 2007.
v
vi
Conference Papers:
7. E. Denny, & M. O’Malley, “A Cost Benefit Analysis of Tidal Generation”, in

9th IAEE European Energy Conference (in review), Florence, Italy, June 2007.
8. E. Denny, G. Bryans, J. Fitz Gerald & M. O’Malley, “A Quantitative Analysis

of the Net Benefits of Grid Integrated Wind”, in IEEE Power Engineering
Society General Meeting (Panel Session), Montreal, Quebec, June 2006.
9. E. Denny, & M. O’Malley, “Impact of Wind Generation on Emissions under

Alternative Power System Operation Approaches”, in 40th Universities Power
Engineering Conference, Cork, Ireland, 2005.
10. E. Denny & M. O’Malley, “Impact of Increasing levels of Wind Generation in

Electricity Markets on Emissions Reduction”, in 7th IAEE European Energy
Conference, Bergen, Norway, Aug, 2005.
11. G. Bryans, E. Denny, B. Fox, P. Crossley, & M. O’Malley, “Study of the Effect
of Tidal Generation on the Irish Grid System: Resource and Emissions”, in
CIGRE Symposium on Power Systems with Dispersed Generation, Athens,
Greece, April 2005.
12. R. Doherty, E. Denny, & M. O’Malley, “System Operation with a Significant

Wind Power Penetration”, in IEEE Power Engineering Society General Meet-
ing, Denver, Colorado, USA, June 2004.
Contents
Abstract i
Acknowledgements iii
Publications Arising from Thesis v
List of Figures xii
List of Tables xv
1 Introduction 1
1.1 Wind Generation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.2 The Impacts of Wind Generation . . . . . . . . . . . . . . . . . . . . 6
1.3 Quantifying the Costs & Benefits of Wind . . . . . . . . . . . . . . . 12
1.4 The Scope of this Thesis . . . . . . . . . . . . . . . . . . . . . . . . . 18
1.5 An Outline of the Thesis . . . . . . . . . . . . . . . . . . . . . . . . . 21
2 The Case Study 23

2.1 Wind Generation in Ireland . . . . . . . . . . . . . . . . . . . . . . . 25
2.2 Ireland’s Electricity Demand . . . . . . . . . . . . . . . . . . . . . . 28
2.3 Ireland’s Electricity Market . . . . . . . . . . . . . . . . . . . . . . . 30
vii
CONTENTS viii
2.3.1 Dominance in the Irish Electricity Market . . . . . . . . . . . 31

2.4 Ireland’s Current Generation Plant Mix . . . . . . . . . . . . . . . . 33
2.5 The Future Plant Mix of the Irish System . . . . . . . . . . . . . . . 37
2.6 Summary of the Key Features of the Case Study . . . . . . . . . . . 38
3 The Dispatch Model 41

3.1 Optimisation in the Dispatch Model . . . . . . . . . . . . . . . . . . 42
3.1.1 Validating the Dispatch Model . . . . . . . . . . . . . . . . . 44
3.2 Wind in the Dispatch Model . . . . . . . . . . . . . . . . . . . . . . 45
3.2.1 Wind Curtailment . . . . . . . . . . . . . . . . . . . . . . . . 47
3.2.2 Wind Shortfall . . . . . . . . . . . . . . . . . . . . . . . . . . 48
3.2.3 Load Factors for the Wind Generation . . . . . . . . . . . . . 48
3.3 The Dispatch Model Representing an Electricity Market . . . . . . . 49
3.4 Specifics of the Dispatch Model for Case Study . . . . . . . . . . . . 51
3.4.1 Reserve Capacity . . . . . . . . . . . . . . . . . . . . . . . . . 51
3.4.2 Hydro and Pumped Storage . . . . . . . . . . . . . . . . . . . 52
3.4.3 Outage Schedules . . . . . . . . . . . . . . . . . . . . . . . . . 52
3.4.4 Demand . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
3.4.5 Fuel Prices . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
3.4.6 Peat Generation . . . . . . . . . . . . . . . . . . . . . . . . . 53
3.4.7 Wind Capacity . . . . . . . . . . . . . . . . . . . . . . . . . . 54
3.4.8 Network Constraints . . . . . . . . . . . . . . . . . . . . . . . 54
3.5 Summary of Key Features of the Dispatch Model . . . . . . . . . . . 55
4 The Cost of Wind Generation 56

4.1 Wind Development Costs . . . . . . . . . . . . . . . . . . . . . . . . 57
4.2 Network Upgrade Costs . . . . . . . . . . . . . . . . . . . . . . . . . 59
4.3 Additional Reserve Requirement . . . . . . . . . . . . . . . . . . . . 60
4.4 Cycling Conventional Units . . . . . . . . . . . . . . . . . . . . . . . 62
4.4.1 The Cost of Cycling . . . . . . . . . . . . . . . . . . . . . . . 66
4.5 Summary of Costs Associated with Wind Generation . . . . . . . . . 70
CONTENTS ix
5 The Benefits of Wind Generation 72

5.1 Capacity Benefit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
5.2 Emissions in Conventional Generators . . . . . . . . . . . . . . . . . 75
5.2.1 Carbon Dioxide Formation . . . . . . . . . . . . . . . . . . . 75
5.2.2 Sulphur Dioxide Formation . . . . . . . . . . . . . . . . . . . 76
5.2.3 Nitrogen Oxides Formation . . . . . . . . . . . . . . . . . . . 76
5.2.4 Coal, Peat and Heavy Fuel Oil Generators . . . . . . . . . . . 77
5.2.5 Gas Fired Generators . . . . . . . . . . . . . . . . . . . . . . 78
5.2.6 Emissions During Cycling . . . . . . . . . . . . . . . . . . . . 81
5.3 Emissions Benefits of Wind Generation . . . . . . . . . . . . . . . . . 81
5.4 Fuel Savings Benefit . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
5.5 Summary of Benefits of Wind Generation . . . . . . . . . . . . . . . 86
6 The Net Benefits of Wind Generation 88

6.1 The Net Benefits of Wind Generation in the Base Case . . . . . . . . 88
6.2 Sensitivity Analysis of Net Benefits . . . . . . . . . . . . . . . . . . . 91
6.2.1 Wind Load Factor Scenario . . . . . . . . . . . . . . . . . . . 94
6.2.2 Wind Power Forecast Accuracy Scenario . . . . . . . . . . . . 95
6.2.3 Demand Growth Scenario . . . . . . . . . . . . . . . . . . . . 96
6.2.4 Fuel Price Scenario . . . . . . . . . . . . . . . . . . . . . . . . 97
6.2.5 Capital Cost Scenario . . . . . . . . . . . . . . . . . . . . . . 97
6.2.6 Cycling Cost Scenario . . . . . . . . . . . . . . . . . . . . . . 98
6.2.7 Emissions Price Scenario . . . . . . . . . . . . . . . . . . . . . 99
6.2.8 Discount Rate Scenario . . . . . . . . . . . . . . . . . . . . . 99
6.2.9 Worst and Best Case Scenarios . . . . . . . . . . . . . . . . . 100
6.3 Conclusions of Net Benefits . . . . . . . . . . . . . . . . . . . . . . . 101
7 Applications of the Net Benefits Methodology 103

7.1 Carbon Prices and Cycling Costs . . . . . . . . . . . . . . . . . . . . 104
7.1.1 Emissions Prices in the Cycling Analysis . . . . . . . . . . . . 107
7.1.2 Results & Discussion . . . . . . . . . . . . . . . . . . . . . . . 108
CONTENTS x
7.1.3 Summary of Carbon Prices and Cycling Costs . . . . . . . . . 114

7.2 The Fuelsaver Approach to Wind Generation . . . . . . . . . . . . . 115
7.2.1 Analysing the Fuelsaver Approach . . . . . . . . . . . . . . . 116
7.2.2 Fuelsaver Cycling Costs . . . . . . . . . . . . . . . . . . . . . 120
7.2.3 Fuelsaver Emissions Benefits . . . . . . . . . . . . . . . . . . 120
7.2.4 Fuelsaver Fuel Benefits . . . . . . . . . . . . . . . . . . . . . . 122
7.2.5 Net Benefits of Wind under the Fuelsaver Approach . . . . . 123
7.2.6 Conclusions of Fuelsaver Approach . . . . . . . . . . . . . . . 124
7.3 The Impact of Connection Policy on Distributed Generation . . . . . 125
7.3.1 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
7.3.2 Ireland’s Distribution Network . . . . . . . . . . . . . . . . . 128
7.3.3 The Energy Resources . . . . . . . . . . . . . . . . . . . . . . 130
7.3.4 Optimal Allocations of Distributed Generation . . . . . . . . 131
7.3.5 Load Flow Simulation . . . . . . . . . . . . . . . . . . . . . . 132
7.3.6 Scaling Outputs . . . . . . . . . . . . . . . . . . . . . . . . . 133
7.3.7 Distributed Generation in the Dispatch Model . . . . . . . . 134
7.3.8 The Costs and Benefits of Distributed Generation . . . . . . 135
7.3.9 Net Benefits . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
7.3.10 Conclusion of Connection Policy . . . . . . . . . . . . . . . . 138
7.4 Tidal Generation on the Irish System . . . . . . . . . . . . . . . . . . 139
7.4.1 Modelling the Tidal Generation . . . . . . . . . . . . . . . . . 141
7.4.2 Tidal Cycling Costs . . . . . . . . . . . . . . . . . . . . . . . 142
7.4.3 Tidal Emissions Benefits . . . . . . . . . . . . . . . . . . . . . 143
7.4.4 Fuel Savings with Tidal Generation . . . . . . . . . . . . . . 144
7.4.5 The Net Benefits of Tidal Generation . . . . . . . . . . . . . 144
7.4.6 Conclusions of Tidal Generation in Ireland . . . . . . . . . . 147
8 Discussion and Conclusions 149

8.1 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
8.2 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
8.3 Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158
CONTENTS xi
References 160
A The Characteristics of Generators on the Irish System 181
B The Optimisation of the Distribution Network 187

List of Figures
1.1 Net benefits of wind generation . . . . . . . . . . . . . . . . . . . . . 19
2.1 Existing wind farms in RoI in January 2007 . . . . . . . . . . . . . . 26

2.2 Comparison of electricity consumption in the EU15 . . . . . . . . . . 29
2.3 Typical daily demand patterns in RoI in 2006 . . . . . . . . . . . . . 29
2.4 Installed generation capacity by fuel type in Ireland . . . . . . . . . 33
2.5 The age of the generating units on the Irish system . . . . . . . . . . 34
2.6 Load duration curve with merit order of conventional units . . . . . 34
2.7 Conventional generation availability in Ireland . . . . . . . . . . . . 36
3.1 Generator reserve characteristics . . . . . . . . . . . . . . . . . . . . 44

3.2 Wind power data series . . . . . . . . . . . . . . . . . . . . . . . . . 47
3.3 Assumed load factors with increasing installed wind generation . . . 49
3.4 Illustration of market model . . . . . . . . . . . . . . . . . . . . . . . 50
4.1 Assumed capital cost per MW of installed wind generation . . . . . 58

4.2 Network reinforcement costs with increased wind capacity . . . . . . 60
4.3 System reserve level with different wind power forecast errors . . . . 61
4.4 Damage associated with different cycling activities . . . . . . . . . . 64
4.5 Creep fatigue interaction resulting in premature component failure . 65
xii
LIST OF FIGURES xiii
4.6 Operating profiles of two Irish units with different wind penetrations 66
4.7 Cycling costs with increasing penetrations of wind generation . . . . 70
5.1 The capacity credit of wind generation in Ireland . . . . . . . . . . . 74

5.2 Typical NOX emissions from a CCGT and an OCGT . . . . . . . . . 80
5.3 Emissions savings with increasing wind generation . . . . . . . . . . 82
5.4 Value of emissions savings with increasing wind generation . . . . . . 84
5.5 Fuel consumption with increasing wind generation . . . . . . . . . . 85
5.6 Value of fuel savings with increasing wind generation . . . . . . . . . 86
6.1 The base case net benefits of wind generation . . . . . . . . . . . . . 90

6.2 The Critical Values of Wind Generation Part I . . . . . . . . . . . . 92
6.3 The Critical Values of Wind Generation Part II . . . . . . . . . . . . 93
6.4 The net benefits of wind with changes in load factor . . . . . . . . . 94
7.1 Merit order with carbon price of e30/ton . . . . . . . . . . . . . . . 108

7.2 Cycling costs and CO2 savings at different carbon prices . . . . . . . 109
7.3 Cycling costs and CO2 savings with wind and e30/ton CO2 . . . . . 111
7.4 Cycling costs with wind generation and carbon prices . . . . . . . . 112
7.5 Carbon savings with wind generation and carbon prices . . . . . . . 113
7.6 Net savings with wind generation and carbon prices . . . . . . . . . 113
7.7 Wind curtailment in 2010 under fuelsaver . . . . . . . . . . . . . . . 118
7.8 Average number of committed units under fuelsaver . . . . . . . . . 118
7.9 Average operating levels of committed units under fuelsaver . . . . . 119
7.10 Cycling costs under the fuelsaver approach . . . . . . . . . . . . . . . 121
7.11 Emissions savings under the fuelsaver approach . . . . . . . . . . . . 121
7.12 Fuel savings under the fuelsaver approach . . . . . . . . . . . . . . . 123
7.13 The net benefits of wind under the fuelsaver approach . . . . . . . . 124
7.14 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
7.15 Future distribution of installed wind power capacity in percent . . . 130
7.16 Output of DG generators on a sample day . . . . . . . . . . . . . . . 133
7.17 The marine current turbines design . . . . . . . . . . . . . . . . . . . 141
LIST OF FIGURES xiv
7.18 The power output during a spring and neap tide . . . . . . . . . . . 142
7.19 Cycling costs with increases in tidal generation . . . . . . . . . . . . 143
7.20 Tidal generation emissions savings . . . . . . . . . . . . . . . . . . . 144
7.21 Tidal generation fuel savings . . . . . . . . . . . . . . . . . . . . . . 145
7.22 The total benefits and cycling costs of tidal generation . . . . . . . . 146
A.1 Generator Information for 2007 . . . . . . . . . . . . . . . . . . . . . 182

A.5 Generator Operating Characteristics . . . . . . . . . . . . . . . . . . 186
List of Tables
2.1 The Irish electricity system (at year end 2005) . . . . . . . . . . . . 24

2.2 Price support caps for wind generation under REFIT . . . . . . . . . 28
3.1 Fuel costs in e 2006/GJ . . . . . . . . . . . . . . . . . . . . . . . . . 53
4.1 Costs associated with increased cycling of units . . . . . . . . . . . . 67

4.2 Cost per single on-off cycle . . . . . . . . . . . . . . . . . . . . . . . 67
4.3 Cycling activity expressed as an equivalent hot start . . . . . . . . . 68
5.1 Typical components of various fuel types . . . . . . . . . . . . . . . . 77

5.2 Typical components of natural gas . . . . . . . . . . . . . . . . . . . 78
5.3 Emissions prices in e per ton . . . . . . . . . . . . . . . . . . . . . . 83
6.1 Base case assumptions summarised . . . . . . . . . . . . . . . . . . . 90

6.2 Worst case and best case assumptions summarised . . . . . . . . . . 100
7.1 Assumed Energy Resource for the Five Network Sections (MW) . . . 131
7.2 Allocations for the five network sections . . . . . . . . . . . . . . . . 131
7.3 Scaled allocations for the whole system . . . . . . . . . . . . . . . . . 134
7.4 Connection costs for wind generation . . . . . . . . . . . . . . . . . . 135
7.5 Additional reserve costs for distributed wind generation . . . . . . . 136
xv
LIST OF TABLES xvi
7.6 Additional cycling costs with wind generation . . . . . . . . . . . . . 136

7.7 Emissions savings (in tons) with wind generation . . . . . . . . . . . 137
7.8 Fuel savings in petajoules with wind generation . . . . . . . . . . . . 137
7.9 Net benefits of wind generation (in em) . . . . . . . . . . . . . . . . 138
7.10 Boundary annual costs . . . . . . . . . . . . . . . . . . . . . . . . . . 147
CHAPTER 1
Introduction
H
UMAN activities are changing the composition of the Earth’s atmosphere
through the buildup of greenhouse gases. This has resulted in an increase in
the temperature of the earth’s surface by about 1◦ celsius in the past century, with
the 10 warmest years of the 20th century occurring in the 15 years between 1985 and
2000 (US EPA, 2006a). The snow cover in the Northern Hemisphere and floating
ice in the Arctic Ocean have decreased, worldwide rainfall has increased by about
one percent and global sea levels have risen by 4-8 inches over the past century.
The consequences of global warming range from manageable to catastrophic
(Goodstein, 2002). According to the World Health Organisation (2002) approxi-
mately 150,000 people currently die needlessly each year as a direct result of global
warming. Agricultural output will suffer significantly as the land becomes drier and
outputs fall. This situation will be particularly severe in many Third World coun-
tries which lack resources for irrigation. The Hadley Centre (2006) predicts that
by 2010, 300 million people in sub-Saharan Africa will be suffering from chronic
malnutrition and a further 30 million people worldwide will be at risk of hunger
due to climate change by 2050. Massive disruption of natural ecosystems is likely
1
Chapter 1. Introduction 2
to occur as forests die and up to 50 million species could be driven to extinction by

2050 (Roach, 2004). Diseases and pests are also likely to thrive in a warmer climate
(Goodstein, 2002). An increase in sea-level would flood many areas of the world and
as many as 1 billion people could be directly impacted. Low lying countries such as
Bangladesh and the Netherlands and many coastal areas along the eastern coast of
the United States and the Gulf Coast would become uninhabitable (Lovgren, 2004).
From an economic perspective, it has been shown that ignoring climate change will
eventually lead to major disruption to economic and social activity later this cen-
tury, on a scale similar to those associated with World War II and the economic
depression of the first half of the 20th century, and it will be difficult or impossible
to reverse these effects (Stern, 2006).
From being a research topic, global warming has now moved to the political
stage and policy makers worldwide have accepted the importance of reducing green-
house gas emissions (Helm, 2005). As a result, there has been an international
movement in the promotion of policy mechanisms for the reduction of greenhouse
gas emissions. The Kyoto Protocol is an agreement made under the United Nations
Framework Convention on Climate Change (Kyoto, 1992) and countries that ratify
this protocol commit to reducing their emissions of carbon dioxide. Emissions trad-
ing regimes which came into force EU-wide in 2005 (EU ETS, 2003), and globally
in 2008, will aid countries struggling to meet their greenhouse gas obligations un-
der the Kyoto Protocol. In order to reduce European emissions of sulphur dioxide,
nitrogen oxides (NOX ), volatile organic compounds and ammonia, the Gothenburg
Protocol indicates national emissions ceilings for each of these compounds for all EU
Member States for 2010. Once the Protocol is fully implemented, it is projected that
Europe’s NOX emissions should be cut by at least 41% and its sulphur emissions by
63% compared to 1990 levels (The Gothenburg Protocol, 1999).
In addition to national emission limits, there are also directives limiting emissions
from individual installations. One such directive is the Large Combustion Plant
Directive (LCPD) which limits NOX , SO2 , and dust emissions from combustion
plants over 50MW in size. Under the LCPD, any combustion plant which will not
meet its emissions targets must cease operation in 2010 or run for a limited number
of hours between 2010 and 2015 (EU Directive 2001/80/EC, 2001).
As well as being a major contributor of greenhouse gas and other emissions, the
electricity industry is also highly fossil fuel dependent. This reliance on fossil fuels
leaves the industry highly exposed to price increases as global fossil fuel resources
become more depleted (EIA, 2004). The increasing concentration of the remaining
resources in certain geographical areas may also make fossil fuel prices more volatile.
In a bid to reduce their reliance on fossil fuels, while simultaneously reducing power
system emissions, there has also been a move towards the promotion of clean renew-
able technologies for electricity generation. In an aim to achieve a European wide
target of 22.1% of gross electricity consumption produced from renewable energy
sources by 2010, the EU Directive 2001/77/EC obliges all member states to achieve
a given percentage of their electricity consumption from renewable energy sources.
These targets are based on the current installed capacity in each country, the po-
tential for future renewable generation and their Kyoto Protocol emission targets:-
the renewable energy targets range from between 5.7% for Luxembourg to 78.1% for
Austria (EU Directive 2001/77/EC, 2001).
As wind generation does not itself create any harmful emissions, policy makers
often promote it as a means of reducing a country’s national emission levels while
meeting its renewable obligations. This has ensured that wind energy has become
the new renewable energy of choice for most northern European countries. This is
reflected in figures showing wind energy as the fastest growing form of renewable
energy in Europe (EWEA, 2005). In addition, the United States is currently ex-
periencing the strongest wind capacity growth period ever witnessed (Smith et al.,
2004). As the market for wind energy has grown, the costs per kilowatt hour have
reduced dramatically. Ambitious renewable energy targets together with successes
to date and reducing costs, should ensure that the increased penetration of wind
energy continues in European electricity networks.
1.1 Wind Generation
The majority of renewable energy ultimately comes from the sun. The sun radiates
174,423,000,000,000 kilowatts to the Earth per hour (DWEA, 2006). According to
the BWEA (2006) this equates to more energy per second than all the electricity
used in the UK in a month! About 1 to 2 per cent of the energy coming from the sun
is converted into wind energy (DWEA, 2006). Wind is formed by the movement of
air over the surface of the Earth, from areas of high pressure to low pressure. These
areas of high and low pressure are caused by the heating of the Earth’s surface by
the sun and the rotation of the earth (BWEA, 2006).
Globally air rises from the equator and moves north and south in the upper
atmosphere until it reaches an area of low pressure where it will fall. As the air rises
from the equator there will be a low pressure area close to ground level attracting air
back from the north and south. This determines the prevailing wind in a country,
however, local climatic conditions also have an influence on the most common wind
directions. Land masses are heated by the sun more quickly than the sea in the
daytime. The air rises, flows out to the sea, and creates a low pressure at ground
level which attracts the cool air from the sea. At night, the temperature of the
land drops much faster than the sea and the wind blows in the opposite direction.
In addition, close to the Earth’s surface, wind can be affected by friction from
geographical features such as mountains, hills, forests, cities etc. (DWEA, 2006;
BWEA, 2006).
A wind turbine converts the energy of the wind into a torque acting on the
rotor blades which then drives a generator. The amount of energy which the wind
transfers to the rotor depends on the density of the air, the rotor area, and the wind
speed (Miller et al., 1997). Thus, the kinetic energy in the wind is converted to
mechanical energy which is then converted to electrical energy (Ahmed and Cohen,
1994).
Wind generation, like solar, tidal and wave generation, exhibits ‘variable’ output.
The output of these units depend upon weather conditions that cannot be controlled
by the operator of the generator. This is known as being ‘non-dispatchable’. For
example, the amount of electricity generated by a wind turbine fluctuates as wind

speed changes and that of a photovoltaic array with the intensity of sunlight. Thus,
the control of their output is limited as operators can only reduce the potential
output of such generators (Gross et al., 2006). When significant penetrations of
these forms of generation are connected to an electricity network, it can result in a
requirement to alter the operation of the system to accommodate the variability of
these generators.
As well as being variable, wind generation also faces a challenge of being rela-
tively unpredictable. Since the underlying resource cannot be directly controlled,
the renewable generation is high when conditions are favourable and low when un-
favourable. Thus, forecasts of weather conditions are crucial when examining renew-
able generation sources. Tidal generation is governed by the gravitational forces of
the moon and the sun on the Earth’s oceans and can be predicted almost perfectly
over long time horizons (Bryans, 2006). Wind generation on the other hand, requires
complex forecasting techniques which account for wind speed, wind direction, hub
height, geographical surroundings, wind farm size, turbine dispersion, etc. Given the
large number of factors which must be taken into account when forecasting wind
generation, the margin for error can be significant and increases as the time horizon
lengthens (Giebel et al., 2003).
The function of power system operators is to supply electricity to customers in
a reliable manner at a sustainable cost. Reliability of electricity supply is defined
as “the ability to supply adequate electric service on a nearly continuous basis with
few interruptions over an extended period of time” (IEEE/CIGRE, 2004). This
involves ensuring that the generation meets the load at all times and that any
short term gaps between load and generation are bridged quickly and precisely to
maintain the integrity of the power system (Kirschen and Strbac, 2004). Generators
are scheduled to meet the forecasted load and must alter their operating levels
to follow the load as it fluctuates throughout the day. Since the output of some
renewable generators, in particular wind generation, cannot be actively controlled
and is difficult to predict accurately, this balancing of the generation and the load
can become more challenging (Gross et al., 2006).

Given the optimistic targets for renewable energy in Europe and the specific
challenges associated with increased variable wind generation, there has been much
debate on the advantages and disadvantages associated with wind generation. Some
argue that wind generation imposes high costs on the power system operation and
increased wind power threatens the stability of the system and could result in
greatly increased prices (Etherington, 2006). Others argue that wind generation
is the panacea for the challenge of depleting fossil fuel resources and that it should
be widely supported and if necessary subsidised to ensure its further development
(Brown, 2006).
Policy makers are continually bombarded with groups on both sides of the wind
debate and as a result, it is often the case that vague objectives with regard to wind
generation are made. It was thus recognised that research was required to identify
and quantify the costs and benefits associated with increased wind penetration. The
work presented here discusses in detail these costs and benefits, quantifies the im-
pacts that wind generation has on an electricity system and determines the optimal
penetrations of wind generation beyond which no further investment should be made
in wind generation.
1.2 The Impacts of Wind Generation
The interaction of wind energy and the electricity system is complex and there
are significant challenges posed to system operators when large amounts of vari-
able generation are introduced (Bazilian et al., 2004). Under the European Union
EU Directive 96/92/EC (1996) the Transmission System Operator (TSO) is respon-
sible for “ensuring a secure, reliable and efficient electricity system and, in that con-
text, for ensuring the availability of all necessary ancillary services”. These ancillary
services include the provision of sufficient reserve capacity to meet the load under
unexpected system operating conditions (Wang and Billington, 2004). As shown in
Zhu et al. (2000) these unexpected system conditions can include the loss of a unit,
a transmission line trip and unexpected load fluctuations. Söder (1993) showed that
an increase in the capacity of wind generation on an electricity system increases the

uncertainty in the system, as wind generation is relatively unpredictable and non-
dispatchable, which results in a requirement to carry additional reserve capacity in
order to maintain system security.
Depending on the electricity system, there are a number of reserve categories
which a system operator must carry - from fast acting reserve, with a reaction time
in the order of seconds, to reserve which responds more slowly (Rashidi-Nejad et al.,
2002; Kundur, 2004). It has been shown in Watson and Landberg (2003), Marti et al.
(2006) and Kariniotakis and Pinson (2003), that the standard deviation of wind
power forecast errors over short time frames are small. As such Doherty and O’Malley
(2005), Dany (2001) and Gouveia and Matos (2004), found that increasing wind
power capacity has little effect on the reserve categories that operate over a short
time frame (seconds to minutes), however, wind capacity causes a greater increase
in the need for categories of reserve that act over longer periods of time.
In addition to the added reserve requirement with increases in wind generation,
an increase in variable generation on an electricity system may require the system
operator to alter how conventional generation is dispatched (Gardner et al., 2003;
Hatziargyriou et al., 2000). Conventional generation may be obliged to operate at
lower operating levels in order to be available to ramp up to accommodate the
inherent variability of the wind generation (Gross et al., 2006). There may also be an
increase in the number of start-ups and shut-downs of conventional units as system
operators attempt to coordinate the following of the fluctuating load throughout
the day and the variable output of the wind generation (Holttinen, 2004; EirGrid,
2004).
Thermal units are designed to be at their most efficient when online and running
at a stable load (Flynn, 2003). It has been shown by the IEEE Subcommittee (1990),
that in general, units are optimised for continuous rather than cyclical operation and
when operating in their normal range can operate for relatively long periods with
relatively low risk of failure and loss of equipment life. An increase in the ramping
and starting of conventional units, as a result of increased wind penetration, can
result in increased wear and tear on the machine and result in a shortening of the life
span of the unit (Lefton et al., 1997). It is estimated that one on-off cycle for a single
unit can cost as much as e500,000 (Grimsrud and Lefton, 1995), thus an increase
in cycling associated with increased wind generation could result in significant cost
increases.
Currently many wind turbines have protection equipment installed which discon-
nects the turbine following a fault (Mullane et al., 2005). This sympathetic tripping
can cause serious problems for system security, increasingly so with larger numbers
of wind generators connecting to the system (Xiang et al., 2006). Essentially, this
sympathetic tripping aggravates a fault event resulting in increasing pressure on
reserves and reliability standards. In periods of low demand and high wind speeds
(e.g. during windy nights) a significant contribution of wind power can be reached
even when the overall share of wind power in the electricity supply is still modest
(Hurley and Watson, 2002). When the wind power penetration level is considerable
it will no longer be feasible for wind turbines to disconnect during voltage or fre-
quency disturbances, as this would lead to a large generation deficit and could lead
to a reduction in the stability of the system (McArdle, 2004; Lalor et al., 2005). This
issue has led to a review of many grid codes across Europe to include a necessity
for wind turbines to have fault ride through capabilities (Causebrook and Fox, 2004;
Johnson and Urdal, 2004).
The most ideal sites for wind generation often coincide with remote locations
which are frequently in weak areas of the electricity grid. This can result in con-
nection challenges for wind generation with issues such as voltage control, harmonic
emissions, short circuit levels and losses playing a significant role (Keane and O’Malley,
2005b). This has led to the possibility of distribution networks switching to ‘active’
rather than ‘passive’ systems through the installation of complex control systems
and by equipping transformers with tap changers that can control voltage locally
(IEE, 2003). Rural locations for wind farms also cause problems for developers. For
cost minimisation the location of plant is very important and there is often a trade-
off between the windiest sites and easy access to the distribution network. There are
large economies of scale associated with wind farm development which often cannot
be realised in remote areas due to limitations on access (DWEA, 2006). In addition,
wind generation developers often come up against significant planning opposition
from local residents in remote areas which in conjunction with lengthy connection
application procedures can considerably delay the development process (SEI, 2003;
IWEA, 2004).
Wind generation is variable and its output at any particular time in the future
is significantly less-assured than conventional fuel-based technologies. However, in
some respects all power plants are variable in that their reliability is never perfectly
guaranteed, and system operators have simply learned how to deal with unplanned
outages (Ahmed and Cohen, 1994). It is a common myth in the media that since
the wind generation cannot be relied upon 100% of the time, that every MW of
installed wind generation must be matched by a MW of additional conventional
generation. For example, as stated by the well known environmental commenta-
tor David Bellamy “wind turbines are completely effete because they need backup
all the time and help to produce CO2 , not reduce it” (Gross et al., 2006). This is
entirely incorrect as wind generation does in fact add to the capacity of a system,
and the extent to which it does is generally given by its capacity credit. According
to Castro and Ferreira (2001), the capacity credit of a unit can be defined as “the
amount of conventional resources (mainly thermal) that could be ‘replaced’ by the
renewable production without making the system any less reliable”. It is generally
considered that wind generation has a capacity factor of between 20 and 40% de-
pending on location, which is a function of the existing installed wind, meaning
that 100 MW of installed wind generation can avoid the building of between 20 and
40MW of conventional generation.
For system security reasons, it is highly beneficial for a country to generate its
electricity from a diverse range of sources so as not to be reliant on the supply
of an individual fuel source. Wind generation adds to the plant portfolio and the
underlying energy source is free and will never be depleted. Since it reduces a
country’s reliance on fossil fuel it reduces its exposure to fluctuations in supply as a
result of international politics (Ahmed and Cohen, 1994).

In addition to the capacity benefit of installing wind generation, the opera-
tion of wind generation can provide significant savings to the economy. Under
EU Directive 2001/77/EC (2001), wind generation must receive priority dispatch.
This means that if the wind is blowing, the system operator must accept the elec-
tricity produced by the wind generator. Since generation must always equal de-
mand, this can result in the displacement of the output of a conventional unit
(Malik and Awsanjli, 2004). A reduction in output generally leads to a reduction in
fuel consumption and accordingly the fuel bill. Given the rapid economic growth in
Eastern countries such as China and India and conflicts in fossil fuel rich countries,
fossil fuel prices have been growing worldwide. This rise in prices is likely to continue
as fossil fuel resources become more depleted (EIA, 2004). Thus, the fuel savings
from wind generation are likely to grow in value into the future as fossil fuel prices
rise.
In addition to saving on the fuel costs by displacing conventional generation,
wind can also reduce the harmful emissions from the burning of these fuels. Harmful
emissions are created in combustion plants through the burning of fuels at elevated
temperatures. Carbon Dioxide (CO2 ) is a greenhouse gas and is one of the main
emissions responsible for global warming (US EPA, 2006a). The amount of carbon
dioxide released is in direct proportion with the amount of carbon in the fuel and
the quantity of fuel burnt. Thus, if wind generation results in a reduction in the
operation of a generation plant which burns a carbon intensive fuel, it will result
in a reduction in carbon dioxide emissions. Sulphur Dioxide (SO2 ) is a colourless
gas which damages trees and crops and can lead to chronic breathing difficulties in
humans. It is also a contributor to acid rain (Alberta Environment, 2003). Like
carbon dioxide, sulphur dioxide emissions are directly related to the sulphur content
of the fuel and the calorific value of the fuel, and reduced operation will result in
reduced SO2 emissions. Nitrogen Oxide (NOx ) is an atmospheric pollutant which
contributes to ozone-depletion and reacts with hydrocarbons, ozone and light to
produce smog. NOx also reacts with water to produce acid, which causes corrosion
problems and acid rain. Electric power plants are among the highest contributors of
NOx emissions along with motor vehicles and industrial boilers (CEPAARB, 2004).
Nitrogen Oxides formation is more complex than CO2 and SO2 formation. NOx
does not depend solely on the nitrogen content of the fuel but is also affected by the
flame temperature, the oxygen concentration and the residence time (Kesgin, 2003).
It is not always the case that a reduction in output will lead to a reduction in NOX
emissions (Mansour et al., 2001).
It has been shown by Sims et al. (2003) that wind generation is among the
cheapest ways of reducing carbon dioxide emissions and is far cheaper than solar
and carbon sequestration. With current political pressure to reduce CO2 emissions
under the Kyoto Protocol and the cost to an economy of purchasing carbon permits
(Conniffe et al., 1997), emissions reduction from wind generation is highly signifi-
cant.
Wind generation development can also greatly aid local communities. The con-
struction and operation of wind turbines creates local jobs, and Mazza (2005) esti-
mates that the construction of a 75MW wind farm could create employment for up
to 200 labourers. Additionally, a study for the United States Department of Energy
(US DOE, 2005) found that wind energy produces 27% more jobs per kilowatt hour
than coal plants and 66% more jobs than gas fired plants. Private landowners can
earn significant annual rents from developers for the citing of wind turbines on their
land (Ritsema et al., 2003). The land is not made redundant by the turbines and
can continue to be used as agricultural land for the duration of the turbine life. The
combination of these benefits multiplies within communities as the wages earned by
employees of the wind projects may be spent in local businesses, the supplies for
construction and maintenance may be purchased locally and an increase in tourism
can result (Galluzzo, 2005). In order to access wind generation sites, developers
often need to enhance the quality of the local road network. The cost of these im-
provements to the local infrastructure are covered by the developers and improve
the standard of living for all those living in the vicinity.
Wind generation is a publicly visible example of an environmentally friendly
technology and a country’s commitment to renewable generation. Wind generation

can act as a reminder to those who see it that global warming is a serious issue, and
this can have a knock-on effect of encouraging people to be more energy efficient in
their daily lives. Wind generation is accessible engineering for young children and
is often used by environmental groups to promote environmental awareness from a
young age (DWEA Kids, 2006).
This is not an exhaustive summary of the impacts of wind generation but rather
an illustration of some of the more direct impacts. Depending on the scope of
an analysis, the costs and benefits of wind generation investigated could be much
further reaching, such as the reduction in future health costs as a result of a reduction
in carbon dioxide emissions, or the environmental impact costs of construction in
elevated areas etc. However, the costs and benefits investigated in this thesis are
limited to the most direct costs and benefits, and are summarised in Section 1.4.
1.3 Quantifying the Costs & Benefits of Wind
A number of studies in the past have attempted to estimate the value of these costs
and benefits of wind generation and a selection of these are reviewed here.
Kennedy (2005) give a generic approach to calculating the long term costs and
benefits of wind generation. Here the average variable, fixed and environmental costs
for conventional generators over a fixed period of time are calculated without wind
in the plant portfolio. A fixed capacity of offshore wind generation is then added
and the generator’s costs are recalculated. The difference in the costs is taken to be
the social benefit of wind generation. However, this analysis is limited to looking at
the impact on two distinct units rather than the impact on the electricity system
as a whole. These units are highly efficient advanced gas fired units and relatively
low emitters of CO2 and NOX with no SO2 emissions, which underestimates the
potential emissions benefits of wind generation. In addition, load duration curves
are used to evaluate the operation of the units rather than a time sequential model
of the system as a whole. A perfect wind forecast is assumed and no account is
made of additional costs associated with increased cycling of the conventional units
and any additional reserve that may be required.

Munksgaard and Larsen (1998) compare the socio-economic impacts of wind gen-
eration in Denmark to central gas and coal fired plants. This paper presents an
interesting study on both the internal and external costs and benefits of wind gener-
ation when compared to conventional coal and gas units. However, the wind turbine
sizes analysed are small in comparison to today’s standards, in addition, the authors
tend to underestimate the capacity credit of wind generation at between just 10 -
20%. However, the main drawback of this work is admitted by the authors when
they state that to accurately calculate the costs and benefits of wind generation “re-
quires a simulation model which can simulate the operation of the electricity system.
A model like this has not been used in this study”.
Bergmann et al. (2006) claim to “estimate the magnitude of these external costs
and benefits for the case of renewable technologies in Scotland”. The approach
adopted uses a ‘choice experiment’ technique to determine the social welfare gain
of increased wind generation. This involved members of the public ranking the
advantages and disadvantages of wind generation in terms of importance as they
see them and their willingness to pay a subsidy on their electricity bill for these
benefits. This approach is highly limited in that no attempt was made to quantify
the external cost of wind generation on the electricity system and the use of eco-
nomic utility curves to determine the value of wind generation is highly subjective.
Alvarez-Farizo and Hanley (2002) use a similar methodology to estimate the value
of the environmental benefits of wind generation through public preferences.
Pepermans et al. (2005) provide a comprehensive summary of the costs and ben-
efits that can accrue from small scale wind generation installed on the distribution
network. Although this paper provides insightful analysis of the costs and benefits
of wind generation, it does not attempt to place a monetary value on these costs
and benefits.
A number of more comprehensive studies have been conducted on different elec-
tricity systems in order to quantify the real cost of integrating wind generation onto
a network. The results from these different papers are difficult to compare due to
different methodologies, data, tools, as well as terminology and metrics used in the
results (Holttinen et al., 2006). In addition, many of the studies concentrate on just
a single cost such as the added reserve cost. Also, many of the studies provide only
a very limited estimate of the benefits associated with wind generation, if quanti-
fied at all. Comparisons across work done in different countries must be done with
caution as the costs and benefits are relevant only in the context of the particular
system in which they are embedded (Gross et al., 2006). However, a brief summary
of the methodologies used in a selection of these studies for different countries and
the salient results are presented here.
The European Union currently contributes approximately 75% of the installed
wind generation worldwide (Montes et al., 2007). According to EWEA (2005), an
increase in the existing contribution of wind power for electricity production in Eu-
rope is technically and economically feasible. Large penetrations of wind generation
require changes in approaches to operating power systems to maintain a stable and
reliable supply. EWEA (2005) found that the impact of wind power depends on the
wind power penetration level, the power system size, generation capacity mix, the
degree of interconnection and load variations. Across Europe, it is estimated that
the cost of providing additional reserve for wind generation ranges from e1-4/MWh
(EWEA, 2005).
For the UK, Dale et al. (2004) estimate the costs that underlie the price paid by
electricity end users, i.e. the generation costs, the conventional units’ fuel costs and
network costs. This analysis is conducted for two scenarios, a ‘conventional’ scenario
which sees all electricity produced by mainly thermal units in 2020 and a ‘wind’ sce-
nario which has 20% of electricity produced by wind generation in 2020. The results
are based on the work presented in the comprehensive report by ILEX and Strbac
(2002) which uses a system model to determine the likely operating levels of conven-
tional units with increased wind penetration. The results show that the predominant
cost associated with increased wind generation is related to the non-dispatchability
of the wind generation resulting in additional reserve requirement and lower con-
ventional unit efficiency. Gross et al. (2006) provide a systematic review of a large
number of studies on the costs and impacts of intermittent generation with an em-
phasis on the UK electricity system. In this study, intermittency refers to generation
which is both variable and relatively unpredictable. They found that of over 200
studies analysed, the major issue for wind generation integration was its variability
and its unpredictability. This variability and unpredictability results in a need for
additional balancing of system load and generation and a need to maintain capac-
ity to ensure reliability of supplies. For the UK, they found that the cost of this
intermittency ranged from £2-£3/MWh (£1 = e1.48 in Jan 2007). If the costs
associated with capacity developments required to ensure future reliability of sup-
ply are included, then the cost of additional intermittent generation ranges from
£5-£8/MWh.
In Germany, renewable energy sources currently account for 10% of electricity
generation consumption. The goal of the DENA (2005) study undertaken by the
German Energy Agency was to enable fundamental and long term energy-economy
planning with increased wind generation. The main focus of this work was on the
extension of the electricity network, however, the additional reserve requirement was
also calculated. It was found that on average wind generation required an increase
in reserve capacity by an amount equal to 9% of the installed wind generation.
The study also found that the development of wind power requires a shift in the
conventional plant mix towards more flexible units.
The Norwegian electricity system is almost entirely made up of hydro genera-
tion (99% electricity consumption) and currently has a shortage of capacity which
has led to a reliance on imports to help meet its electricity demand. As a result,
the Norwegian government has committed to installing 1000MW of wind generation
by 2010 (Hagstrom et al., 2005). A comprehensive analysis of the impacts of wind
generation on the entire Nordic system is detailed in Holttinen (2004). The costs
analysed include reserves and balancing costs and the benefits covered include re-
placed energy, reduced fossil fuel consumption and CO2 emissions savings. It was
found that wind generation resulted in a requirement for an increase in the provision
of reserve in the 15min - 1hour time frame. It was estimated that wind generation
would result in a saving of 620-700gCO2 /kWh by displacing mainly coal units in

Finland and Denmark. Although the results shown by Holttinen (2004) are very
interesting, the availability of a large amount of hydro generation which can be used
to smooth the variability of the wind generation may tend to underestimate the
costs of wind generation which would be incurred in a system which was less flexi-
ble. A large amount of interconnection is also apparent in the Nordic system which
is beneficial to wind generation. Emissions of SO2 and NOx are not considered in
this study.
Ireland presents a very interesting case study for wind generation as it is a
relatively small island system with limited interconnection and a large and growing
wind energy penetration. A report by Gardner et al. (2003) analysed the impact of
wind generation on the island of Ireland and estimated that 1000MW of installed
wind could result in fuel savings of e75m per annum. No estimate was made in
this study on the extent of the system costs of wind generation such as additional
reserve requirement. In addition, this study is rather simplistic and assumes that
“the conventional generation that would run if there was no wind generation on the
system will also run when wind generation is connected. As the output of wind
generation is increased, the output of the conventional generators will be reduced
accordingly, but no conventional generation will be shut down”. This method of
power system operation is known as the ‘fuelsaver’ approach. This study assumes
that wind generation has no capacity benefit and results in an underestimation of the
potential benefits of wind generation. A study undertaken by EirGrid (2004), the
system operator in Ireland, attempted to estimate the future benefits and limitations
of wind generation in the Republic of Ireland. This study utilised the Siemens
PROMOD software to dispatch the system with wind generation. This study utilised
conservative estimates of the plant mix and the findings of the study are critical of
wind generation. EirGrid (2004) conclude that the cost of CO2 abatement with wind
generation would exceed e120/tonne and 1500MW of installed wind generation
would add e196m to the cost of electricity generation per annum. This study
omitted the potential of SO2 and NOx emissions reductions.
The Production Tax Credit in the United States has been extended through the
end of 2007 and is resulting in the strongest wind capacity growth period ever wit-
nessed in the country (Smith, 2006). A large number of studies have been conducted
on wind generation in the United States. Dragoon and Milligan (2003) estimate the
relative value of wind generation compared to other conventional resources using
dispatch models for the American utility PacifiCorp in Wyoming. The costs of
wind generation analysed were the increased reserve requirements and imbalance
costs associated with increased wind penetration. The system analysed has a large
penetration of hydro generation (14%) which could be used to balance some of the
variation in the wind generation. However, an ‘off the shelf’ dispatch model was
used in this analysis which was not detailed enough to deal with the operation of
the hydro units so optimising their operation with respect to the wind generation
was not considered in this study. Thus, the variability costs of wind to this particular
system may have been overestimated. By using a commercially available dispatch
model, the authors were limited in the assumptions that could be altered and since
the model was not designed to incorporate wind generation, perfect wind forecasts
were assumed. No attempt is made to quantify the benefits accruing to additional
wind generation.
A study on the XCEL-NORTH system by Brooks et al. (2003) used traditional
utility simulation scheduling to estimate the system costs of increased wind gen-
eration. This study concentrated solely on the additional reserve cost of 280MW
installed wind generation in Minnesota and found that the integration costs were
$1.85/MWh ($1 = e0.77 in Jan 2007). However, the models used in this analysis
are rather crude by today’s standards and the assumptions regarding forecasts were
conservative (Smith et al., 2004). Increased wind penetrations of 15% installed ca-
pacity on this system were investigated in Zavadil et al. (2004) and it was found
that the integration costs rose to $4.60/MWh.
GE Energy (2005) investigated the ability of the New York electricity grid to
accommodate 10% wind by capacity by 2008. This study used locational marginal
pricing to conduct a market based approach to determine the market prices with
and without wind generation and it was found that net revenue for wind generation
amounted to $35/MWh, the same price as the current power purchase agreement for
wind in New York. This study highlighted the benefits accruing to improvements in
wind power forecasting with savings of up to $10/MWh.
Work by Hirst (2001, 2002) for the Bonneville Power Administration concen-
trated on additional ancillary service provision for wind generation as did other
studies by We Energies (2003) for Wisconsin, GRE Study (2003) for Minnesota and
Kirby et al. (2003) for California (Smith et al., 2004). The general results from these
studies found that the costs associated with wind integration ranged from $1.47 -
4.53/MWh depending on the wind penetration and the underlying plant mix.
While many of the wind integration studies discussed here are interesting, the
results are difficult to compare across different systems and many have concentrated
solely on quantifying the costs and not the benefits. In addition, none of the cost
benefit studies have attempted to quantify the optimal level of wind generation for
a particular system. In much of the previous work conducted, the system modelling
techniques are questionable and some studies omit a system model altogether. The
work presented in this thesis aims to overcome some of these limitations.
1.4 The Scope of this Thesis
This thesis quantifies the costs and benefits of wind generation. The approach
adopted attempts to maximise social welfare and thus incorporates both direct and
indirect costs and benefits into the analysis. Thus, it is assumed that society pays
the capital cost for every MW installed of wind generation and reaps the benefits of
every MWh of electricity produced from it. Any savings in the fuel bill are societal
savings and any premature depreciation in the lifespan of a conventional generator
from increased stresses is considered a societal cost. A comprehensive model of the
system is developed and is used to analyse the impact of wind generation on the
operating profiles of the conventional units.
The costs considered are the development costs of wind generation, such as the
capital, operation and maintenance costs. The system costs analysed are the cost
of deep reinforcement of the network, the additional reserve cost and the cost of
any additional cycling of the conventional units. Wind generation is systematically
added to the plant mix in the case study and the generating units are dispatched
accordingly. The resulting dispatches are then used to calculate any changes in
costs. The benefits investigated include the capacity benefit of wind generation, the
emissions savings benefits and the fuel savings as increasing penetrations of wind
generation are added.
The costs and benefits are then brought together for each penetration level of
wind generation, and the net benefits are determined. It is hypothesised that the
shape of the net benefits curve for wind generation would be similar to the curve
in Figure 1.1. This hypothesis is tested with regard to the case study. The optimal
level of wind generation penetration is illustrated by the marker and is the point
beyond which no further investment should be made in wind generation.
Net Benefits
Optimal
0 Installed Wind
Figure 1.1: Net benefits of wind generation
A comprehensive model of the case study system is designed and simulations are
carried out for a year on a variety of plant mixes for a range of wind generation
penetrations. This thesis provides a wide ranging sensitivity analysis to determine
the impact of key assumptions on the costs and benefits of wind generation. However,
in order to limit the scope of this study, a number of assumptions were required:
Markets: This study represents a near perfectly competitive gross pool electricity
market (Katz and Rosen, 1998). Thus, the generators are assumed to be profit
maximisers and price takers and gaming of the electricity market by individual
generators is not taken into account (Guan et al., 2001). However, it is likely that if
a generating company had a large degree of market power (and was not being actively
controlled by the electricity system regulator) that strategic bidding behaviour could
alter the market outcome with generators withholding output in order to achieve
higher prices (Guan et al., 2001). This results in suboptimal solutions and since this
study aims to maximise social welfare, a perfectly competitive market is assumed. As
discussed in Correia (2006) this approach to electricity market modelling is entirely
valid.
Incentives: This thesis excludes areas such as direct investments and renewable
generation support incentives that may be needed for large scale wind power, how-
ever, carbon prices are included in the analysis and in reality these are likely to aid
in the promotion of wind generation.
Network: Issues relating to voltage control, short circuit levels etc. are not explic-
itly covered in this work. However, the dispatch model developed is used as part of
a cost-benefit study on distribution connection policies (Keane et al., 2006), and is
discussed in Chapter 7. The cost-benefit analysis of connection policies forms part of
a larger study on distribution connected generation by Keane and O’Malley (2005b,
2006) which incorporates detailed network analysis. Initial estimates of network
reinforcement costs with wind generation are however included in the analysis.
Network Congestion and System Dynamics: These are important issues for
wind integration, (Palsson et al., 2003; Akhmatov et al., 2000; CER, 2004a), but are
highly system specific and require a large scale engineering system model beyond
the scope of this study and as such have been omitted in this analysis. However, the
likely impacts of these issues on the net benefits of wind generation are discussed
briefly in Chapter 8.
Other: In addition, in an attempt to limit the number of assumptions required, it
was necessary to omit ‘softer’ factors such as the visual impact of wind generation,
the creation of jobs, improvements in infrastructure, etc. The likely impact of these
issues on the net benefits of wind generation are discussed briefly in Chapter 8.
1.5 An Outline of the Thesis
This thesis consists of eight chapters covering different aspects of the cost benefit
analysis and detailed results, discussions and conclusions are given in each chapter.
Two appendices are also included. The first appendix details the characteristics of
all the generators on the Irish electricity system. The second appendix presents the
network study conduced by Keane and O’Malley (2005b, 2006) which is used as the
basis for some of the results shown in Chapter 7.
Chapter 2 gives a detailed description of the underlying system in the case study.
Since many of the results shown throughout this thesis are dependent on the un-
derlying plant mix, this chapter details the unit types and characteristics as well as
the proposed market arrangements and operational details of the power system in
the case study. Appendix A is provided to detail further the characteristics of the
generators on this system. The feasibility of wind generation and the likely resources
are shown. This study analyses three test years: 2010; 2015 and 2020. The assumed
plant mixes for each of these years is also described in detail.
Chapter 3 details the economic dispatch model used in this analysis and how it
is validated against a full unit commitment model of the case study system. The
intricacies of both models are described in this chapter along with details of the bid
prices for the generators. The wind power data and the inclusion of wind forecasts
in the dispatch model is discussed in detail in this chapter.
Chapter 4 discusses the costs of wind generation. These costs include the de-
velopment and operation costs of the wind turbines, the cost of deep network rein-
forcement, the additional reserve requirement with increasing penetrations of wind
generation and the cost of cycling of generators on the system. The dispatches from
the model are used to calculate the cycling costs and the costs are shown for each
of the three years for the case study.
Chapter 5 describes the benefits of wind generation. The value of the added
capacity of wind generation is quantified in addition to the emissions and fuel saving
benefits.
Chapter 6 combines the results from chapters 4 and 5. The net benefits of in-
creasing penetrations of wind generation are calculated by subtracting the costs from
the benefits. The penetration level where the costs equal the benefits is the critical
point beyond which the costs will exceed the benefits and no further investment
should be made in wind generation. This critical level is tested under a wide range
of scenarios and is discussed in detail in this chapter.
Chapter 7 details four applications of the dispatch model and the cost benefit
methodology employed in previous chapters. The first application investigates the
impact of carbon prices on the cycling costs and carbon dioxide emissions of gen-
erators. The second application investigates an alternative method for including
wind generation in system operation, and is known as the fuelsaver approach. The
third application analyses the costs and benefits of different distribution connection
policies to determine the access policy which will yield the maximum net benefits.
The final application investigates the costs and benefits of tidal generation for the
case study.
Chapter 8 presents a discussion of the salient issues raised in this thesis and
considers the impact of including additional costs and benefits. The main conclu-
sions of the thesis are presented and some directions for possible future research are
provided.
CHAPTER 2
The Case Study
I
RELAND is the case study chosen for this analysis. It is a small, open and
trade dependent economy with a population of approximately 4.24 million in the
Republic and 1.7 million in Northern Ireland (CSO, 2006). Over the last decade the
level of GDP in the Republic has almost doubled and the economy outperformed all
other European economies in the 1990s recording a growth rate three times the EU
average, with an average growth of 8.1%, coining the term The Celtic Tiger. In 2004,
the IMD World Competitiveness Year Book (IMD, 2004) ranked the Republic of
Ireland (RoI) third for GDP per capita in the world, ahead of the United States (5th)
and Switzerland (11th). The Republic is currently operating at full employment,
aided by a favourable corporate tax rate and a knowledge based workforce, which
has led to widespread foreign direct investment, particularly in the pharmaceutical
and information technology industries (IDA, 2006). Economic growth in Northern
Ireland (NI) has also been strong in the last decade, however, growth rates did not
reach the same heights as those in the Republic and have averaged at approximately
3% per annum over the last four years (NICS, 2005). Economic growth is the single
most important driver of electricity demand and as such both the Republic of Ireland
23
Chapter 2. The Case Study 24
and Northern Ireland have seen a consistent growth in electricity demand over the
last 15 years.
The Irish electricity system is made up of two separately operated but AC inter-
connected systems, one in the Republic and one in Northern Ireland. In August 2004,
the Commission for Energy Regulation in the Republic of Ireland and the Northern
Ireland Authority for Energy Regulation made a commitment to develop a single
wholesale electricity market on the island of Ireland. As such, this thesis studies an
‘all island’ electricity system, covering the Republic of Ireland and Northern Ireland
(referred to jointly as ‘Ireland’). Some characteristics of the electricity systems in
RoI and NI are summarised in Table 2.1 below (AIP, 2005; EirGrid, 2005b; SONI,
2005). Given the size of the electricity system in the Republic compared to Northern
Ireland, the information on the case system provided in this chapter is driven by the
characteristics of the electricity system in the Republic1 .
Table 2.1: The Irish electricity system (at year end 2005)
RoI NI All Island

Peak Demand (MW) 4,566 1,663 6,229
Total Generation (MW) 6,403 2,013 8,416
Ireland has a large and growing installed wind capacity which currently repre-
sents 8% of total installed capacity. In addition, Ireland has limited interconnection
making the effects of wind generation on system operation easier to identify as they
are not influenced by interconnection to other systems. There is a relative generation
capacity shortage in the Republic and temporary diesel generators are currently em-
ployed during periods of peak demand during the winter months (EirGrid, 2005b).
This illustrates the need for additional capacity on the system but, in conjunction
with limited interconnection, also highlights the limitations of the system in re-
sponding to large fluctuations in wind generation production. Ireland is thus an
ideal case study to investigate the system impacts of wind generation as the effects
1
The total peak demand in Table 2.1 has been calculated as the sum of the ROI and NI system
peaks. In reality, the times of peak demand may not actually occur at the same time.
are relatively easier to identify.

Section 2.1 discusses the current and future wind situation in Ireland. Section
2.2 details the load growth and daily load profile for the Irish system. The proposed
electricity market design is discussed in Section 2.3. The current generation plant
portfolio is described in Section 2.4 and Section 2.5 describes the assumptions made
about the plant mix for the different years of the undertaken study. A summary of
the key characteristics of Ireland as a case study are given in Section 2.6.
2.1 Wind Generation in Ireland
Wind generation currently represents approximately 8% of the total installed gen-

erating capacity in Ireland and is expected to grow to 12% by 2010 and 20% by
2020 (EirGrid, 2005b). A study undertaken by SEI (2004a) estimates that the ac-
cessible, feasible wind generation penetration in Ireland for 2020 could be as high
as 10,000MW installed.
Wind generation has grown rapidly in Ireland from 70MW installed at the be-
ginning of 2000 to 740 MW at the end of 2006. There are currently an additional
675MW of wind generation with signed connection agreements and 3,286MW in
the connection process awaiting decisions (EirGrid, 2006c). Of the 740MW of wind
generation currently installed, approximately 44% is connected to the transmission
system with the remainder embedded on the distribution network. The rate of
growth has been so rapid that in March 2004 the system operator was forced to
impose a wind moratorium which prohibited any new offers for connections to the
transmission or distribution systems being issued to wind farms. This was largely
due to the lack of a comprehensive grid code for wind generation leading to concerns
regarding security of supply. This cessation of connection was lifted in July 2004
following the publication of a new grid code for wind. Figure 2.1 illustrates the
locations around the Republic of Ireland where wind generation is connected and
committed, as of December 2006 (EirGrid, 2006d). It can be seen that the majority
of wind generation is connected in the west of the country and there is currently
just one offshore development, located off the east coast.
Figure 2.1: Existing wind farms in RoI in January 2007
In order to deal with the large number of interacting applications for wind gener-
ation in Ireland, the Transmission and Distribution System Operators (TSO/DSO)
in Ireland, have introduced a new method through which wind generators will be
granted connection offers. The previous method, on a first come first served basis,
while effective for large conventional generating plant and for a small number of in-
teracting applications, is not effective for large scale management of interacting wind
generation applications. Under the new scheme, applications are grouped according
to their location on the network and their interaction. The TSO/DSO then issues
a connection offer to the individual applications within the group. This connection
offer will remain valid irrespective of whether other applicants in the same group
accept their connection offers. In the case of a major change in the shared connec-
tion design, the connection offer may vary from the initial offer. This is to avoid
over-development of the network and the creation of stranded network assets (ESB,
2004). This new grouping process should greatly reduce the processing time with
which applications are dealt. Wind generation connection on the distribution net-
work is discussed further in Chapter 7 and proposals are made to optimally utilise
the existing distribution network to allow the connection of increased renewable
generation.
To date, wind generation in Ireland has been supported by the Irish Government
through the Alternative Energy Requirement (AER) scheme. The AER scheme has
been funded by a levy imposed on all electricity customers, known as the Public
Service Obligation (DCMNR, 2002). Under the AER programme, developers of
wind generation were invited to tender for the development and operation of wind
projects with a 15 year power purchase agreement with the Electricity Supply Board
(ESB). The tender prices bid by developers were intended to reflect the prices at
which they would be willing to sell their electricity to the ESB. Those developments
with the lowest bid prices were then offered an AER contract. Other technologies
covered by the AER scheme included small-scale hydro, combined heat and power
and biomass generation. There were six rounds of auctions which resulted in the
supporting of 500MW of renewable generation (DCMNR, 2006).
In 2006, a new support mechanism for renewable generation was announced,
called the Renewable Energy Feed In Tariff (REFIT) programme. One of the main
problems with the AER scheme was the awarding of support to renewable projects
which were never actually developed. In order to alleviate this problem, under
the REFIT support scheme, developers of renewable generation must have a signed
network connection agreement prior to applying for support, this should ensure that
projects are developed within a reasonable time frame. Under REFIT, renewable
generators can negotiate their purchase price with any supplier (not just ESB) in
the liberalised market. The REFIT programme will then reimburse the supplier
up to the REFIT price cap. Like the AER scheme, the funding for this support
mechanism will come from the public service obligation levy (DCMNR, 2002). The
REFIT scheme is similar to other fixed feed in tariffs which have proven successful
in many EU countries (REFIT, 2006). The fixed price caps were announced in
September 2006 and are shown for wind generation in Table 2.2. These are the
prices up to which the Government will reimburse suppliers for contracting with
wind generators.
Table 2.2: Price support caps for wind generation under REFIT
Technology Size Tariff (per MWh)

Large Scale Wind > 5MW e 57
Small Scale Wind < 5MW e 59
Putting these figures into perspective, in Germany the fixed feed in tariff ranges
from e60 - 89/MWh, in France the feed in tariff is e30.50 - 83.80/MWh and in
Spain it is just e27/MWh (RISØ and EWEA, 2005).
2.2 Ireland’s Electricity Demand
Ireland’s electricity demand is relatively small with a peak demand of just 6,229MW
in 2006. Figure 2.2 illustrates Ireland’s electricity consumption relative to the other
countries within the EU15 (Eurostat, 2006; IAEA, 2006). Although Ireland’s elec-
tricity demand is small in relative terms (with Luxembourg being the only EU15
country with less), recent annual growth rates in electricity demand in Ireland and
projected into the future are uniquely positive for a developed country (Deloitte,
2005). Economic growth is the predominant driver for electricity demand and in the
period between 1990 and 2000, total annual demand in Ireland increased by 69%,
corresponding to a 5.4% increase per year, compared to an IEA average of 2.8% per
annum (Deloitte, 2005).
In Ireland, forecasts of electricity demand are calculated by the System Opera-
tor and are given in the Generation Adequacy Report (EirGrid, 2005b). Demand
forecasts are calculated based on predictions of growth in GDP and in the Personal
600
1990
2004
500
400
TWh
300
200
100
Luxembourg
Sweden
Germany
Italy
UK
Belgium
Denmark
Netherlands
Austria
Finland
France
Greece
Ireland
Portugal
Spain
Figure 2.2: Comparison of electricity consumption in the EU15
5000
4500
4000
3500
Demand MW
3000
2500
2000
Winter Peak
1500 Typical Winter
Typical Summer
1000 Summer Minimum
500
0
0 6 12 18 24
Hour
Figure 2.3: Typical daily demand patterns in RoI in 2006

Consumption of Goods and Services Index (PCGS). It is envisaged that although

demand will continue to grow, the rate of growth is likely to be less than the record
growth rates experienced in the 1990s. Three scenarios are presented as represen-
tative of the likely levels of electricity demand growth into the future. The ‘high’
demand scenario envisages growth rates of 4.3%, the ‘medium’ scenario with growth
rates of 3.7% and the ‘low’ scenario with 2.5% growth. These growth rates are used
as inputs in the model of the Irish system, discussed further in Chapter 3.
In Ireland, peak demand occurs during winter weekday evenings and minimum
demand occurs during summer weekend night-time hours. Figure 2.3 shows the
typical daily demand profile for the Republic of Ireland in 2006 (EirGrid, 2006a).
It can be seen that the magnitude of the demand is much lower in summer than in
winter, and in addition, the peak demand in the summer occurs much earlier in the
day than in the winter (EirGrid, 2005b). Electricity demand in Northern Ireland
follows the same general pattern (SONI, 2005).
2.3 Ireland’s Electricity Market
Ireland’s electricity market is currently based on highly regulated bilateral trading

with a top-up and spill balancing mechanism. The current arrangements are ex-
pected to be phased out in 2007 to make way for a Single Electricity Market (SEM)
which will commence trading on 1st November 2007. The SEM is considered to be
the first stage in creating an all island electricity system and is expected to result in
reduced electricity costs to consumers and greater security of supply (AIP, 2006b).
The SEM will be a mandatory gross pool market, all trading must take place
through the pool and no physical bilateral trading will be permitted outside the
pool. The market will have a single market operator (MO) and will commence
with two system operators with graduation to a single system operator (SO) over
time. All generators will submit energy bids to the MO/SO in price quantity pairs.
The market will have centralised commitment with a single market clearing price.
All dispatched generators will receive the single marginal price regardless of their
initial bid. Financial contracts, such as contracts for differences (CfDs), can be
made between participants outside of the pool as a hedging instrument to minimise

exposure to variable prices (SEM, 2005).
The SEM will also contain a capacity payments mechanism which will reward
generators for having available plant. This capacity mechanism is intended to aid
generators in recouping their capital costs and should act as an incentive to new
entrants in the market (SEM, 2005). In addition, the presence of a capacity mech-
anism should ensure the energy bids from generators do not incorporate a capital
costs element. The model described in detail in Chapter 3, aims to reflect the salient
features of the proposed SEM market design.
2.3.1 Dominance in the Irish Electricity Market
Given its small size, Ireland’s electricity supply North and South was traditionally
governed by two separate vertically integrated utilities. In the Republic there was
the Electricity Supply Board (ESB) and in the North, Northern Ireland Electricity
(NIE). These utilities were responsible for the generation, supply and transmission
of electricity.
The ESB was founded in 1927 with the development of a hydro power station
at Ardnacrusha. ESB is 95% owned by the Irish Government with the remaining
shares held by employees. ESB has a number of divisions which are ringfenced
and operate separately in the electricity market. These entities are ESB Power
Generation, ESB Customer Supply and ESB Networks. System operations were
traditionally controlled by ESB National Grid, however, this responsibility has now
been passed to an independent system operator, EirGrid (ESB, 2006).
NIE traditionally controlled all the generation, supply, transmission and distri-
bution in Northern Ireland. In 1992, NIE sold four generating facilities in private
transactions and in 1993 the remainder of NIE generation was sold on the London
Stock Exchange. Following this privatisation, the generation, supply and distri-
bution functions of NIE were divided up and are now fully owned subsidiaries of
Viridian Group PLC. These subsidiaries operate independently and are made up of:
Viridian Power and Energy responsible for generation; Huntstown power generation
which is a power station owned by Viridian in the Republic of Ireland; NIE customer
supply; Energia customer supply; and NIE infrastructure division which controls the
transmission and distribution network in Northern Ireland. System operations are
controlled by the independent System Operator Northern Ireland (SONI) (Viridian,
2006).
Liberalisation of the electricity market is required under EU Directive 96/92/EC
(1996). Generation and supply are generally considered as two segments of the elec-
tricity industry where competition is possible (transmission and distribution net-
works are considered natural monopolies). Even with an all island electricity market,
it is likely that the ESB will maintain a highly dominant position in the electricity
market with ownership of over 60% of the generating capacity (AIP, 2005). With a
pool type market using a single marginal price, there exists scope for gaming of the
market by a dominant participant. The marginal unit will be the price setter, and
all other dispatched units will be price takers. Currently ESB owns the majority
of units which are likely to be operating on the margin. This places ESB in a po-
sition which would allow it to bid high prices on the margin, ensuring a high price
for its portfolio of generating assets. In addition, strategic manipulation of outage
schedules could ensure higher prices by withholding the output of baseloaded units
(Deloitte, 2005).
The regulator has decided that it will heavily regulate the operation of ESB in
the SEM through firm market control, transparency and bidding rules rather than
attempt to disaggregate its generating and supply assets to reduce it’s market share
(CER, 2004, 2006a; AIP, 2006a). Given the complex nature of gaming behaviour,
it has been assumed in this thesis that gaming of the Irish electricity market will
be comprehensively controlled by the regulator and will not occur. Thus the model
described in Chapter 3 represents a near perfectly competitive market, which would
provide the most optimal solutions for society.
2.4 Ireland’s Current Generation Plant Mix
Ireland’s generation plant mix was traditionally based on large coal and oil fired
generation plant with a small number old thermal gas generators and peat plants.
Peat is made up of partially decomposed plant debris and is considered an early
stage in the development of coal (EIA, 2004). These peat units use fluidised bed
technology and their operation is similar to that of coal fired units. Peat fired
generation accounts for approximately 350MW of Ireland’s installed capacity. Since
1990, the share of high carbon content fuels such as coal has fallen in Ireland with
a large increase in the use of natural gas CCGT plants. Gas fired generation now
accounts for over 45% of the installed capacity in Ireland (AIP, 2005). Figure 2.4
illustrates the current installed generating plant mix on the island of Ireland.
Peat: 4%
Wind & Other RE: 9%
Coal: 14%
Gas: 45%
Hydro: 6%
Oil: 23%
Figure 2.4: Installed generation capacity by fuel type in Ireland
A profile of the age, size and unit types of generators on the Irish system is
shown in Figure 2.5. In 2004, a number of peat fired units were decommissioned
and were replaced with more efficient fluidised bed units. Thus, as can be seen from
Figure 2.5, with the exception of the replacement of the peat units, all investment
in new generation in Ireland has been in gas fired units, predominantly CCGTs. It
is anticipated that this trend is likely to continue into the future (EirGrid, 2005b;
Doherty et al., 2006; CER, 2006b).
80
70
Hydro
Peat
60 Oil
Gas
Coal
50
Age in Years
40
30
20
10
0
0 100 200 300 400 500 600
Size of Plant (MW)
Figure 2.5: The age of the generating units on the Irish system
7500
6000 Distillate
Gas
Demand MW
4500 Oil
Gas
3000
2450
1500
Coal
Hydro & Peat
0 2000 4000 6000 8000
Hours
Figure 2.6: Load duration curve with merit order of conventional units
Figure 2.6 illustrates a load duration curve for the island of Ireland. This shows
the number of hours in the year when the demand exceeds a given MW level. Also
illustrated is the typical dispatch order for the Irish generating units in 2006, based
on fuel costs (AIP, 2005). The hydro generation is dependent on the available
resource and therefore does not operate at all hours during the year. However, for
ease of illustration, in Figure 2.6 the hydro units are shown at the base as they are
the cheapest units on the system.
Figure 2.6 shows that the load on the island of Ireland is above 2450MW for
all 8760 hours of the year. Thus, the units that operate below this demand level,
operate all 8760 hours of the year and are said to operate at the base. From Figure
2.6 this is seen to be the hydro, peat, coal and some gas fired units. This dispatch
order is based on efficiency and fuel costs with the exception of the peat fired units.
Despite being more expensive and inefficient the Irish Government made a policy
decision to adopt a ‘must run’ approach for all peat fired generators for security of
supply reasons, thus making them baseloaded. The peat is then subsidised through
the public service obligation levy on all electricity bills (DCMNR, 2002). For ease
of illustration, the wind generation has been omitted from Figure 2.6 as it is not
considered dispatchable and the output varies throughout the day and depends on
the time of year.
As the demand increases above 2450MW more gas fired units are dispatched,
followed by oil fired units, then peaking units fired by gas and distillate. Thus, the
marginal unit will be gas, oil or distillate depending on the level of demand. For
example, gas fired units will operate at the margin for approximately 3,510 hours
of the year (8760-5250), oil fired units for 4,450 hours (5250-800), followed by gas
and distillate units during the small number of hours of peak demand in the year.
These marginal units will be required to alter their output in order to follow the
load variations throughout the day.
Ireland has one interconnector, known as the Moyle Interconnector, which con-
nects Northern Ireland to Scotland through submarine cables. The capacity of this
interconnector is 500MW with an operating capacity of 400MW and it entered into
full commercial operation in April 2002. The 400MW capacity of this interconnector
is not always available and it is estimated that only 155MW of its capacity will be
available in the market due to existing contracts (NIAER, 2003).
Apart from Ireland, only one other country in the EU15, the Netherlands, has a
similar reliance on imported gas and oil, and only three countries in the EU15 have
levels above 40% (Deloitte, 2005). In addition, those countries that have material
penetrations of oil and gas capacity, almost always have compensating capacity of
nuclear and hydro generation which have lower operating costs which reduces their
exposure to fluctuations in imported fuel prices and supplies. Ireland has no nuclear
generation and a small penetration of hydro generation and all feasible sites for hydro
generation are considered to be exhausted. The issue of fuel import dependence is
important and leaves Ireland highly exposed to international fuel shortages, which
could have a dramatic effect on the reliability of the Irish system.
Ireland’s generating units already suffer from availability levels well below the
EU average. Figure 2.7 shows the average availability of the conventional units on
the Irish system from 1994 to 2005 (EirGrid, 2005b). It can be seen that there is no
consistent improvement in generator availability apparent in recent years and that
the availability of units in 2003 was particularly bad.
88
System Availability (%)
86
84
82
80
78
76
1995 1997 1999 2001 2003 2005
Figure 2.7: Conventional generation availability in Ireland

The poor availability of the generating units on the Irish system adds to the
inflexibility of the system in responding to large variations in wind generation. In
addition, the availability of certain units could temporarily reduce further in the
coming years as they undergo extensive overhauls in line with EU emissions reduc-
tion directives (e.g. EU Directive 2001/80/EC (2001)). It is expected that the coal
units will install fuel gas desulphurisation (FGD) technology to reduce sulphur diox-
ide emissions, and Selective Catalytic Reduction (SCR) technology, also known as
scrubbers, to reduce nitrogen oxides (NOX ) emissions. Two of the oil units will also
install SCRs to reduce NOX emissions (ESB, 2003). Emissions from individual units
will be described in more detail in Chapter 5.
Further details of the operating characteristics, including fuel type, maximum
capacity and year of commission of the units on the Irish system are shown in
Appendix A.
2.5 The Future Plant Mix of the Irish System
In this thesis, the costs and benefits are calculated for Ireland for the years 2010,
2015 and 2020. The assumed plant mixes for these years were based on the pro-
jected generation plant mix considering increased wind penetration for Ireland from
EirGrid (2005b) and the optimal future generation portfolio for Ireland to accommo-
date large penetrations of wind generation from Doherty et al. (2006). Preliminary
results from the Government sponsored ‘All Island Renewable Grid Study’ (AIRGS,
2006b) specifically aimed at determining the optimal plant mix for 2020 were also
used to inform the assumptions on the generation portfolios. Increases in hydro
generation and pumped storage were considered in AIRGS (2006b) and were found
not to be viable for the Irish system, and are thus not considered in this thesis.
The plant portfolio for 2010 was based on generation projections given in EirGrid
(2005b). Three new gas fired units are expected to be in operation by 2010:
Huntstown II, Aughinish Aluminum and Tynagh (see Appendix A). In addition,
the FGD and SCR technologies in the coal and oil units are assumed to be installed
by 2010.
For 2015 it is envisaged that new additions to the conventional plant mix will be
gas fired (EirGrid, 2005b; CER, 2006b; Doherty et al., 2006). As such, three new
gas fired units are expected to come online by 2015 (EirGrid, 2005b). In addition
it is assumed that two of the oil fired units that were in operation in 2010 will be
retired (Doherty et al., 2006; CER, 2006a).
In 2004, the Irish Government approved the development of an electrical inter-
connector between Ireland and Wales for security of supply and competition pur-
poses. According to EirGrid (2005b), this interconnector is likely to be completed
in 2012. Its installed capacity will be 500MW but its maximum operating capacity
is expected to be 400MW. This interconnector is included in the plant mix of 2015
and 2020 but not 2010. The Moyle interconnector with a capacity of 400MW is
included in the plant mix for each year.
By 2020, another two oil fired units are retired due to age and are replaced with
an additional four gas fired units (Doherty et al., 2006). The details of the plant
portfolios in each of the three test years are given in Appendix A.
In two of the additional applications investigated in Chapter 7 the assumed
test year is 2007. The reasons for the change in test year in these applications are
discussed in Chapter 7 and a summary of the plant mix in 2007 is given in Appendix
A.
2.6 Summary of the Key Features of the Case Study
Ireland is an ideal case study to investigate the costs and benefits of wind generation
as it has limited interconnection allowing for a controlled study on the impacts of
its large and growing wind penetration. The costs and benefits presented in this
work are, in the main, dependent on the underlying electricity system. Thus, the
most significant characteristics of the Irish system with regard to wind generation
are summarised here:
• Ireland has a large wind resource and there is currently 740MW installed,
675MW under development and 3286MW in the planning process, leading to
a potential penetration of over 4000MW. If all 4000MW were developed, it is

projected it could represent over 40% of the installed generating capacity in
2010.
• Ireland has a high electricity demand growth requiring the development

of new capacity on the system. It is envisaged that any new conventional
development will be gas fired. In addition, a 500MW interconnector to Wales
is expected to be completed by 2012 (EirGrid, 2005b).
• Ireland’s underlying plant mix is currently made up of predominantly gas, oil

and coal units. The plant mix is relatively inflexible with only a small number
of fast acting open cycle gas turbines.
• Ireland has a small penetration of hydro units and it is assumed that all
feasible sites for hydro generation have been exhausted. Thus increased pen-
etrations of hydro generation are considered not viable for Ireland (AIRGS,
2006b).
• Ireland has one pumped storage unit. In other countries, pumped storage
can be used to help reduce the variability in the wind generation, however, this
is not possible in Ireland as this unit is already being used for system security
purposes. Increases in pumped storage were considered in AIRGS (2006b) but
were found to be not viable for the Irish system and as such are not considered
in this thesis.
• Ireland’s electricity market will be an all island market, covering the Repub-
lic and Northern Ireland. This will improve the security of the system with
generation capacity increased by a third over the Republic alone.
• The proposed electricity market is a centralised gross pool electricity market

with a separate capacity mechanism. This should ensure that bids in the
energy market are marginal cost based. It is assumed that the dominance of
the ESB will be actively controlled by the electricity regulator
Further details on particular aspects of the Irish system will be described where
relevant in Chapters 3, 4 and 5, and detailed descriptions of the units on the Irish
system are given in Appendix A. In Chapter 6 the costs and benefits are brought
together to determine the optimal penetrations of wind generation. This is in con-
trast to work conducted by Doherty et al. (2006) where an optimal conventional
plant portfolio was determined for given installations of wind generation. Here, the
optimal penetrations of wind generation are determined for a given conventional
plant portfolio.
CHAPTER 3
The Dispatch Model
I
N order to quantify the costs and benefits of wind generation it is necessary to
analyse it in the context of the power system in which it is installed and to
investigate wind’s impacts on the operation of the underlying plant mix. An electric
power system based predominantly on hydro power will result in very different costs
and benefits for wind generation than a power system made up of coal units. This
chapter details how the operation of the case study power system was modelled to
determine how wind generation impacted on the operation of the conventional units.
The model produces operating schedules for all the conventional generators on the
system with increasing penetrations of wind generation. These dispatches are then
used to quantify the costs and benefits of wind generation in subsequent chapters.
The plant mix in the case system is described in Chapter 2 and details on the
characteristics of the individual units are shown in Appendix A. As discussed in
Chapter 2, the proposed all Ireland single electricity market (SEM) is a gross pool
market with centralised commitment (SEM, 2005). The model used in this paper
aims to reflect this market and is validated against a full unit commitment model
of the system. Although many electricity markets are moving away from centrally
41
Chapter 3. The Dispatch Model 42
controlled dispatch, it has been shown in Xu and Christie (1999) that a model which
allows the generators to bid incrementally and self commit produces solutions al-
most identical to those from a centralised unit commitment (Xu and Christie, 1999;
Ede et al., 2000). In fact, centralised unit commitment can be used for the predic-
tion of operating decisions in decentralised markets (Xu and Christie, 1999). Thus,
although the model used here reflects a centralised gross pool electricity market, it
could also reflect the operating decisions in alternative electricity market designs.
Section 3.1 describes the optimisation algorithm used to dispatch the generators.
Section 3.2 details how wind generation was included in the model. A practical
description of how the model represents an electricity market is given in Section 3.3.
The assumptions in the model to tune it to the Irish system are given in Section
3.4. A summary of the key features of the model are given in Section 3.5.
3.1 Optimisation in the Dispatch Model
The dispatch model aims to represent the salient features of the proposed market
design in Ireland. Thus, it is a centralised gross pool electricity market and it is
assumed that generators bid their marginal energy costs. Generators are assumed
to have linear marginal cost reflective bids for energy and since there will be a
separate capacity payment mechanism, these bids should reflect their fuel costs and
not their capital costs. In addition, the dispatch model also dispatches for operating
reserves. Thus, each generator also bids into the pool to provide reserve capacity.
The dispatch model uses a linear programming market clearing formulation to co-
optimise unit operating points and reserve levels to find a least cost solution on an
hourly basis. The aim is given by the following objective function:
N N
!
X X
min cpi Pi + cri Ri (3.1)
i=1 i=1
where Pi is the power from unit i and Ri is the operating reserve from unit i. The
energy and reserve bids of generator i are given by cpi and cri respectively and N
is the number of generators. If losses are neglected then the minimisation is subject
to a load balancing constraint and an operating reserve target.
N
X
Pi = Load (3.2)
i=1
N
X
Ri ≥ Reserve (3.3)
i=1
The characteristics for each unit i are given by equations (3.4) to (3.7).
0 ≤ Pi ≤ P maxi (3.4)
0 ≤ Ri ≤ Rmaxi (3.5)
1
Pi − Ri ≤ P maxi (3.6)
Rslopei
−R maxi
Pi + Ri ≤ 0 (3.7)
P mini
Equation (3.4) ensures that no generator can operate above its maximum rated
capacity and Equation (3.5) ensures that no generator can be dispatched for more
than its maximum reserve capacity. Equations (3.6) and (3.7) describe the reserve
characteristics of the generators. Equation (3.6) ensures that a generator cannot be
dispatched in such a way as its power output plus its reserve exceed its maximum
capacity. Equation (3.7) ensures that a generator cannot be dispatched to provide
reserve unless it is dispatched for power and is operating above its minimum operat-
ing point. The nature of these reserve constraints is illustrated graphically in Figure
3.1.
Discrete decisions must be made about the on/off status of a generator to ensure
it is not dispatched below its minimum rated capacity. Since the dispatch model
does not explicitly contain discrete decision variables, the following approach was
adopted to mimic these on/off decisions. The model is first run with each generator’s
Reserve
Rmax i
slope = -1
Pmin i Pmax i Power

Figure 3.1: Generator reserve characteristics
power output permitted to range between 0 and maximum generation. This results
in optimality, however some units may be dispatched below their minimum operating
point. Since it is assumed that generators have linear bids for energy and reserve it
is assumed that those units which have been dispatched below their minimum must
be utilised to ensure load balance. These units are thus required to be turned on,
and as such, this first run of the dispatch model essentially determines which units
should be on and which off. The algorithm is then run a second time, this time only
including those units which have been turned on in the first stage and forcing them
to operate between their minimum and maximum rated capacities. The algorithm
now returns a feasible dispatch. The dispatch model was run for each hour for an
entire month for each of the test years. The results were then scaled up by a weighted
factor for each month to represent the entire year.
3.1.1 Validating the Dispatch Model
A full unit commitment model of the case system was designed in the PLEXOS
(2006) environment by Bryans (2006). This model used mixed integer programming
and accounted for temporal constraints including minimum stable generation, max-
imum generation, maximum ramp-up rates, maximum ramp-down rates, minimum
down-times, start times (from cold, warm and hot) and scheduled outages. Although
this model was extremely rigourous, the length of time for each monthly model run
(over 72 hours) made it too cumbersome to use for sensitivity analysis and economic
evaluation. In addition, the PLEXOS model was unable to handle updated wind
forecasts and as such a perfect wind forecast was required.

The dispatch model was benchmarked against the PLEXOS model at a number
of different installed wind capacities to ensure that the dispatches from the dispatch
model were reasonable. It was found that although the dispatch model does not
explicitly include temporal constraints, the dispatch model returned realistic dis-
patches when compared with the PLEXOS dispatches. In addition, the resulting
operating schedules for each unit from the dispatch model were examined to deter-
mine their compliance with the temporal constraints. It was found that the dispatch
model produced operating schedules which were over 95% compliant with the ramp
rate constraints and had compliance levels in excess of 90% for the minimum up and
down time and start time constraints.
3.2 Wind in the Dispatch Model
It is the two stage approach of the dispatch model which is one of its key advantages
when looking at wind generation. The first stage of the model essentially returns
a ‘commitment’ decision, by deciding which units should be turned on and which
should be off. The second stage of the model does not turn on or off any units but
determines their operating levels (economic dispatch). This approach can account
for the unpredictability of wind generation forecasts by using a long range wind
forecast in the first stage and an updated wind generation forecast in the second
stage, representing actual system operation. As such, the commitment decisions
are based on a longer range wind forecast and the operating levels are based on an
updated forecast as would be expected in reality. In order to reflect wind forecasting
characteristics, the time between the two model runs was assumed to be 6 hours
(Marti et al., 2006; Meibom et al., 2006).
This two stage approach, with commitment and dispatch taking account of
wind power forecasts, has also been adopted in the Siemens PROMOD software,
used in EirGrid (2004), and the two stage methodology is also employed in the
MORE CARE control software, which is described in detail for an island system in
Hatziargyriou et al. (2002) and Hatziargyriou et al. (2000).
In the process of establishing a transparent market for the island of Ireland, the
Irish system regulators gathered a comprehensive data set of operating characteris-
tics of all generators on the system, including a detailed wind generation data series
(AIP, 2005). This real wind power data is assumed to represent the updated wind
power forecast in the second run of the dispatch model.
The wind power forecast used in the first run of the model is generated based
on this real wind power output and the standard deviations of wind power forecast
errors (Doherty and O’Malley, 2005). An ‘average’ wind power forecast error with
a standard deviation of 9% of the installed capacity was used in the base case
(Doherty and O’Malley, 2005). This implies that 68% of the time, the forecast of
the wind power output is within 1 standard deviation (9% installed capacity) of the
actual wind power output. A ‘best’ case and a ‘worst’ case wind power forecast with
a 7% and 14% standard deviation respectively are also tested, representing the most
accurate and the least accurate the total wind power forecast error is likely to be, for
a forecast horizon of 6 hours ahead (Doherty and O’Malley, 2005)1 . The sensitivity
of the net benefits of wind generation to changes in the wind power forecast error is
investigated in Chapter 6.
Figure 3.2 represents a two day sample of the wind series data used in the
dispatch model for an assumed wind penetration of 1000MW. The long range forecast
is based on a standard deviation of wind power forecasts equal to 9% of the installed
capacity of the wind generation. This forecast is used in the first run of the dispatch
model. The updated wind power forecast, used in the second run of the model, is
given by the real wind power output from AIP (2005).
Under EU Directive 2001/77/EC (2001) of the European Union, when it is avail-
able, wind generation must receive priority dispatch. In order to comply with this
directive, the wind generation is assumed to bid into the dispatch model with a
marginal cost of zero. This ensured it was always dispatched when available.
1
Assuming the ‘worst’ case wind power forecast scenario could also represent a situation where
commitment and dispatch were made further than 6 hours apart i.e. if commitment decisions were
made more than 6 hours ahead, the wind power forecasts would be less accurate and the errors
larger.
900
800
Long Range Wind Forecast
(commitment)
700
Real Wind Output (dispatch)
Wind Power Output MW
600
500
400
300
200
100
0
0 6 12 18 24 30 36 42 48
Hour
Figure 3.2: Wind power data series
3.2.1 Wind Curtailment
If the long range wind forecast for a given hour grossly underestimated the actual
wind output, surplus conventional generation could be committed in the first stage
of the dispatch model, i.e. generation could exceed the load even when all committed
conventional generators are running at their minimum. If this situation occurs then
no conventional unit is switched off and the wind generation is curtailed instead.
Wind generation is curtailed if Equation (3.8) holds.
N
X
(δi,j Pmin,i ) + Windj > Loadj ∀j (3.8)
i=1
where N is the number of generators on the system, Pmin,i is the minimum

generation for unit i, δ is either 1 or 0 depending on whether unit i has been
committed in hour j. Windj is the updated wind power forecast for hour j and
Loadj is the load in hour j.
This situation results in the curtailment of wind generation by an amount equal
to the load minus the sum of the minimum generation of the committed generators.
This new wind output is then used as the updated wind forecast for that hour in
the second run of the dispatch model. This situation occurred rarely but increased
as the installed wind capacity increased.
3.2.2 Wind Shortfall
If the long range wind power forecast overestimates the actual wind output too few
plant may be committed to meet the load in the first run of the dispatch model.
This would happen if Equation 3.9 were to hold.
N
X
(δi,j (Pmax,i − Ri )) + Windj < Loadj ∀j (3.9)
i=1
where Pmax,i is the maximum capacity of generator i and Ri is the dispatched

reserve for unit i. When this happens, the dispatch model forces on fast acting plant,
such as Open Cycle Gas Turbines (OCGTs), to meet the shortfall in generation in the
second stage of the model. The breach of constraint (3.9) occurred more frequently
than (3.8) and increased as the installed wind capacity increased.
3.2.3 Load Factors for the Wind Generation
The electricity output of a wind generator is based on its load factor. The load
factors for wind generation in Ireland typically range from 25% - 40% depending on
a large range of factors including location, turbine size, season etc. (BWEA, 2005).
A range of different load factors are tested in this thesis and are summarised in
Figure 3.3.
It is assumed that the most favourable sites for wind turbines will be developed
first and these are likely to have the highest load factor. For this reason, it is
assumed that the load factor per MW installed decreases with increasing installed
capacity in the ‘low’ and ‘mid’ load factor cases as the number of feasible favourable
sites reduces. For comparative purposes, a ‘constant’ load factor has been included
and the ‘high’ load factor case represents the situation where new development in
50
45
40
35
Load Factor %
30
25
20
High Load Factor
Constant Load Factor
15
Mid Load Factor
Low Load Factor
10
0
0 500 1000 1500 2000 2500 3000 3500 4000
Wind Generation MW
Figure 3.3: Assumed load factors with increasing installed wind generation
wind penetration is in offshore sites with higher load factors and in onshore sites
with advances in wind turbine efficiency. These load factors are used when the wind
generation output from AIP (2005) is being scaled up for increasing penetrations of
wind generation. The sensitivity of the net benefits of wind generation to the chosen
load factor is investigated in Chapter 6.
3.3 The Dispatch Model Representing an Electricity

Market
The dispatch model aims to represent a realistic electricity market. The first stage
of the dispatch model can be considered to represent the forward market where
generators bid for energy and reserve and the MO/SO determines commitment. The
second stage of the dispatch model mimics gate closure where the final operating
and reserve levels are determined. Running the model in this manner is a realistic
way of capturing the operation of the market and also the error in the wind forecast
between ahead markets and real time. Figure 3.4 illustrates this methodology. Here,
the system is being dispatched to generate electricity and reserve for hour j.
MW
Hour j
Load
b
aaaaaaaaaaaaaaaaaa
aaaaaaaaaaaaaaaaaa
a aaaaaaaaaaaaaaaaaa
caaaaaaaaaaaaaaaaaa
aaaaaaaaaaaaaaaaaa
aaaaaaaaaaaaaaaaaa
aaaaaaaaaaaaaaaaaa
Forward Gate Real Time

Market Closure Time
Reserve
aaaaaConventional Generation
aaaaa
aaaaaWind or Conventional Generation
aaaaa
aaaaa
Wind
Figure 3.4: Illustration of market model
In Figure 3.4, the x-axis represents the time-line approaching real time. The
long range wind power forecast is bid into the forward market (point a in Figure
3.4) and the conventional generation is dispatched up to the forecasted load plus the
reserve target (b). Thus, in this forward market those generators to be turned on
to be available for generation and reserve at real time are determined, based on the
long range wind forecast. It is assumed that due to the large levels of installed wind
generation being investigated, the wind generation forecast errors will dominate the
load forecast errors (Doherty and O’Malley, 2005) thus a perfect load forecast is
assumed. This assumption could be relaxed by altering the load in the second run
of the dispatch model in a similar manner to how the updated wind forecasts are
included.
Between the forward market and gate closure, the wind forecasts are readjusted
as forecasting becomes more accurate as the forecast horizon decreases. Thus, at
gate closure the wind profile has been adjusted to an ‘updated’ profile (c) and the
operating levels of those generators switched on in the forward market are set. It
may be necessary at gate closure to curtail the wind generation or turn on fast acting
units depending on the magnitude of the error in the wind forecasts (Sections 3.2.1
and 3.2.2). In this thesis, all of the costs and benefits are calculated on the dispatched
levels of the generators at gate closure. In reality, between gate closure and real time
the wind power output may change further. This balancing is represented by the
white shaded area in Figure 3.4. This real time error is not considered in this thesis
but will be discussed briefly in Chapter 8.
3.4 Specifics of the Dispatch Model for Case Study
The specific input assumptions for the dispatch model are described here in order
to tune the model for the Irish system.
3.4.1 Reserve Capacity
In many systems, the amount of operating reserve carried is equal to the size of the
largest infeed, which for Ireland is 480MW (EirGrid, 2006b). An increase in the
capacity of wind generation on an electricity system increases the uncertainty in the
system which results in a requirement to carry additional reserve capacity in order to
maintain system security (Söder, 1993). Thus, in this analysis, the assumed reserve
requirement is equal to the size of the largest infeed plus the additional reserve based
on the installed wind generation. The levels and cost of this additional reserve
requirement with wind generation is discussed in detail in Chapter 4. Generator
reserve capacities and bid prices (based on Doherty and O’Malley (2005)) are shown
in Appendix A.
3.4.2 Hydro and Pumped Storage
Ireland has one pumped storage station and a small concentration of hydro power
plants. Given the complex operating nature of these units, the hydro and pumped
storage plant were optimised in the full Unit Commitment model described in Section
3.1.1, and the operating levels were then used as an input in the dispatch model.
3.4.3 Outage Schedules
No unit is available for all hours of the year due to maintenance and outages. In
order to account for this, the calendar for generators’ scheduled outages from the
data set in AIP (2005) was used to incorporate the outage rates. A generator’s
maximum capacity was then set to zero for all 24 hours of the days it was scheduled
to be offline.
3.4.4 Demand
Hourly demand data for the Republic of Ireland is taken from the Irish transmission
system operator (EirGrid, 2006a). The Northern Ireland electricity system is about
one third of the size of the system in the Republic of Ireland so the load values for
the Republic were scaled up by a factor of 1.33 to represent the two systems. The
demand levels for 2010, 2015 and 2020 were based on the projected load growth
levels in EirGrid (2005b). Three demand growths are investigated: 2.5%, 3.7% and
4.3%. In order to calculate the demand for each year the following formula was used.
Lj,y = Lj,2005 × (1 + g)y−2005 ∀j (3.10)
where Lj,y is the load in hour j for year y and y is either 2010, 2015 or 2020.
The demand in each of these years is grown from the base year of 2005. Thus, the
load in hour j in 2005 is given by Lj,2005 . The assumed growth rate is g. Thus a
flat growth rate for each hour of each day is assumed. This may not necessarily be
the case in reality, with the demand in some hours growing more than in others,
however, based on EirGrid (2005b), it is assumed that the average growth over the
year will be one of the three growth rates given.
3.4.5 Fuel Prices
There are a range of fuel prices investigated in the analysis, based on CER (2006b),
UK DTI (2004) and Doherty et al. (2006), and shown in Table 3.1. The same growth
rate is not used for each fuel price as some fuel prices are expected to grow by more
than others. For example, it is likely that gas prices will grow more than coal due
to an increase in the development of cleaner more efficient gas fired plant (CER,
2006b).
Table 3.1: Fuel costs in e 2006/GJ
Year Scenario Gas Coal Oil Peat Distillate

Low 4.16 2.16 4.13 3.14 7.83
2010 Mid 4.37 2.17 4.27 3.23 8.08
High 4.58 2.18 4.40 3.33 8.34
Low 4.41 2.17 4.30 3.25 8.14
2015 Mid 5.04 2.22 4.71 3.53 8.92
High 5.66 2.26 5.12 3.82 9.70
Low 4.91 2.21 4.63 3.48 8.76
2020 Mid 6.41 2.31 5.62 4.15 10.65
High 7.91 2.41 6.61 4.83 12.52
The bid price for each generator is obtained by multiplying the appropriate fuel
price by the generator’s consumption of energy per MWh (given for each generator
in Appendix A). The sensitivity of the net benefits of wind generation to the chosen
fuel price is investigated in Chapter 6.
3.4.6 Peat Generation
Ireland has a large indigenous natural peat resource and peat fired generation ac-
counts for approximately 350MW of Ireland’s installed capacity. The Irish Gov-
ernment have adopted a ‘must-run’ attitude towards peat generation and in 2001
the European Commission agreed to allow the Irish Government to continue this
policy. Under this scheme, electricity customers subsidise the use of indigenous
peat as a fuel for power through a Public Service Obligation on all electricity bills
(DCMNR, 2002). This is part of the national policy of fuel diversity, which is part
of Ireland’s strategy to ensure security of electricity supply. This ‘must-run’ policy
however, distorts the optimal scheduling of the generators. Since this is a social wel-
fare maximising study, it is assumed for the purposes of this analysis, that the peat
generators must bid their marginal cost into the gross pool like all other generators.
The assumed prices are shown in Table 3.1 previously.
3.4.7 Wind Capacity
In this thesis installed wind capacities up to 4000MW are investigated. While this
thesis does not explicitly conduct a resource evaluation of wind generation on the
case study system, other studies suggest that the feasible resource could be larger
than 4000MW (SEI, 2004a). However, it is deemed that the dispatch model may
become less reliable at very high levels of installed wind generation. Thus, it is
suggested that a more complex engineering model which can include the temporal
characteristics of the generators and issues of system dynamics would be required
for higher assumed installed wind capacities.
3.4.8 Network Constraints
Network constraints are not explicitly included in the dispatch model. However
initial estimates from the Irish transmission system operator of the network rein-
forcement necessary with increases in wind generation are discussed in Chapter 4.
These network reinforcements will help ameliorate potential congestion. Network
congestion is discussed briefly in Chapter 8. In addition, an application of the dis-
patch model has been conducted in Chapter 7 which utilises network analysis for
the distribution system.
3.5 Summary of Key Features of the Dispatch Model
This chapter detailed the model used to simulate the operating schedules of gener-
ators on the Irish system with increasing penetrations of wind generation. The key
features of this model are summarised here.
• The model used is called the dispatch model and it is a linear programming
formulation which cooptimises generators operating and reserve levels on an
hourly basis.
• The model represents a gross pool electricity market. The generators en-
ergy bids consist of their marginal fuel cost (prices given in Table 3.1 and
energy consumption per MWh in Appendix A). Their reserve bids are given
in Appendix A.
• The model is run in two stages with the commitment decisions made in the
first stage and the operating decisions made in the second. The model is
validated against a full unit commitment model of the system.
• The wind generation bids into the pool with a marginal cost of zero. A long
range forecast is used in the commitment stage of the model and an updated
forecast is used to determine the generators’ operating levels.
• The operation of the hydro and pumped storage units is not optimised
and is based on the schedules from the unit commitment model.
• A perfect demand forecast is assumed. A range of demand growths are in-

vestigated for the years 2010, 2015 and 2020.
• The network is simplified to a single bus-bar system and the optimisation

does not account for losses.
The sensitivity of the costs and benefits of wind generation to many of the
assumptions discussed in this Chapter are investigated in Chapter 6.
CHAPTER 4
The Cost of Wind Generation
T
HIS chapter presents the costs associated with wind generation. As this study
represents a social welfare maximising analysis, all the costs associated with
wind generation are assumed to accrue to society in general rather than any par-
ticipant in particular. For example, it has been debated whether wind generation
should be charged for the additional reserve costs it may impose on a system. This
work does not answer this question but rather attempts to estimate the magnitude
of the costs.
The costs of developing and operating wind generation is a cost of increased wind
generation and is included here. A number of scenarios are investigated with respect
to wind capital costs and are discussed in Section 4.1 (CER, 2001). Work done by
AIRGS (2006b) included initial estimates of the additional network upgrade costs
associated with increasing penetrations of wind generation in Ireland. These initial
estimates of deep network reinforcement costs are discussed in Section 4.2.
An increase in variable and relatively unpredictable wind generation on the sys-
tem can increase the uncertainty for system operators. As such they may be required
to carry additional reserve capacity to maintain standards of security. The level, cost
56
Chapter 4. The Cost of Wind Generation 57
and implications of this additional reserve capacity is discussed in Section 4.3 of this
Chapter.
An increase in variable wind generation may result in an increase in the cycling
of conventional units on the system as operators attempt to coordinate the following
of the fluctuating load throughout the day and the variations in the wind generation
(EirGrid, 2004). The dispatch model described in Chapter 3 is used to investigate
the changes in the operating patterns of conventional generators as the installed
wind penetration increases. These costs are discussed in Section 4.4. A summary of
the costs of wind generation is provided in Section 4.5.
4.1 Wind Development Costs
Wind generation has zero fuel costs, thus, its costs are based mainly on its capital
cost, which represents about 75-80% of its total cost. Average capital costs for
wind generation currently range from e900 to e1,100 per kW installed including
the cost of shallow connection to the grid (EWEA, 2003). Shallow connection costs
are the costs of the local assets required to connect a generator to the electricity
network. Deep connection costs are the cost of the transmission system assets that
may be required with an increase in generation. These deep costs are discussed in
the following Section 4.2. Four different capital cost curves for wind generation in
Ireland are investigated in this thesis and are shown in Figure 4.1.
In each of the four scenarios, the capital costs are initially assumed to be e1
million per MW installed. In ‘High Capital’, the wind development costs grow as
the installed wind generation increases. This is due to increases in the costs of
production of wind turbines associated with a growth in fossil fuel prices worldwide,
an increase in the development of more expensive offshore wind farms and a reduction
in the economies of scale for smaller onshore wind farms. As a control scenario,
‘Constant Capital’ assumes that the cost per MW remains constant as the installed
wind generation increases. In ‘Mid Capital’ and ‘Low Capital’ the cost reduces
incrementally, most likely as a result of advances in technology and turbine size.
Under ‘Mid Capital’ it has been assumed that the most accessible and efficient sites
1.1
Capital Cost per MW in em 1.05
0.95
0.9
High Capital
Constant Capital
0.85 Mid Capital
Low Capital
0.8
0 500 1000 1500 2000 2500 3000 3500 4000
Installed Wind Capacity MW
Figure 4.1: Assumed capital cost per MW of installed wind generation
are developed first. As more and more wind generation is connected, the number of
feasible wind farm sites reduces and the size of wind farms is likely to decline. It is
typically the case that there is a reduction in the economies of scale for smaller wind
farms resulting in higher costs per MW installed. Thus, these reduced economies
of scale combined with a likely increase in off-shore wind development result in the
assumption that the cost per MW will begin to increase beyond a certain point in
‘Mid Capital’. This point, where the capital cost begins to increase, was based on
a forecast of the current feasible installed wind capacity on the island of Ireland,
given in SEI (2004b). Capital costs continue to decrease under ‘Low Capital’ under
the assumption of technology advances.
The total capital cost is given by the total installed wind capacity multiplied
by the cost per MW from Figure 4.1. However, throughout this thesis the costs
are analysed on an annual basis, thus, in order to annualise the capital costs, they
are assumed to be equivalent to an annuity (CER, 2006b; RAE, 2004). The annual
capital cost is therefore given by multiplying the total capital cost by A in Equation
4.1.
i
A= (4.1)
−T
1 − (1 + i)
where i is the assumed interest rate and T is the term over which the capital
costs are discounted. The interest rates of 6%, 8% and 10% are tested in Chapter
6. The term is assumed to be 20 years (CER, 2006b; RAE, 2004).
Operating and maintenance costs for wind energy include repair and insurance
and represent about 20-25% of the total production costs per kWh for wind energy
(EWEA, 2003). Operation and maintenance costs are assumed to be e35,000 per
MW installed per annum (Doherty et al., 2006).
4.2 Network Upgrade Costs
As the installed penetration of wind generation increases, it is likely that the trans-
mission and distribution networks will require upgrading. Workstream 3 of the
Government Sponsored All-Ireland Renewable Grid Study (AIRGS, 2006a) will in-
vestigate the network development options and costs associated with increases in
wind generation penetration in Ireland. As yet, no results are available from this
study, however, Workstream 2A includes initial estimates from the transmission
system operator of the likely deep network reinforcement costs with increased wind
generation. These estimates are illustrated in Figure 4.2.
As with the capital costs of wind generation, these network costs must be ex-
pressed as an annual cost for the purposes of this study. This is done by expressing
them as an annuity using Equation (4.1) with an interest rate of 8% and a term of
20 years.
Workstream 3 will include a load flow analysis of the Irish system to determine
potential constraints and bottlenecks (AIRGS, 2006a). This requires a large scale
engineering model of the system which is considered to be beyond the scope of this
study. Thus, the model described in Chapter 3 is based on a single bus bar system
and transmission system losses are ignored.
800
700
600
Network Costs em
500
Deep Network Costs
400
300
200
100
0
0 500 1000 1500 2000 2500 3000 3500 4000
Figure 4.2: Network reinforcement costs with increased wind capacity
4.3 Additional Reserve Requirement
In many systems, the amount of operating reserve carried is equal to the size of
the largest infeed (EirGrid, 2006b). An increase in the capacity of wind generation
on an electricity system increases the uncertainty in the system as wind generation
is relatively unpredictable and non-dispatchable. This results in a requirement to
carry additional reserve capacity in order to maintain system security Söder (1993).
The level of this additional reserve requirement for Ireland was based on the results
shown in Doherty and O’Malley (2005).
Doherty and O’Malley (2005) relate the reserve level on the system in each hour
to the reliability of the system over the year. The reserve is thus allocated in such
a way as to keep the average risk of a load shedding incident in each hour the
same for all hours. Doherty and O’Malley (2005) incorporate both load and wind
power forecast errors when calculating the reserve requirement for each hour. Load
forecast errors are not particularly sensitive to the forecast horizon and are usually
proportional to the load at any given hour. The wind power forecast errors generally
increase as the forecast horizon increases. In Doherty and O’Malley (2005), the load
and wind forecast errors for hour h are modelled as Gaussian stochastic variables
with a mean of zero and a standard deviation of σload,h and σwind,h respectively and
the total system forecast error is given by Equation (4.2).
q
σtotal,h = 2
σload,h 2
+ σwind,h (4.2)
Based on results given in Pinson and Kariniotakis (2003), Doherty and O’Malley
(2005) determined the most accurate and the least accurate wind power forecasts
errors are likely to be. It was found that the average wind power forecast error
had a standard deviation of approximately 9% of the installed wind capacity. The
best and the worst case forecast errors had standard deviations of 7 and 14% respec-
tively (See Section 3.2 in Chapter 3). Figure 4.3 below indicates the reserve capacity
required to maintain reliability on the Irish system associated with these three stan-
dard deviations of wind power forecast errors. These graphs are extrapolated from
Doherty and O’Malley (2005).
1600
1400 Worst Wind Forecasts

Average Wind Forecasts
Best Wind Forecasts
1200
Reserve (MW)
1000
800
600
400
200
0
0 500 1000 1500 2000 2500 3000 3500 4000
Installed Wind Capacity (MW)
Figure 4.3: System reserve level with different wind power forecast errors
The sensitivity of the net benefits to changes in the forecast error of the wind
power output is tested in Chapter 6. Thus, when a wind power forecast error
with a standard deviation of 7% is tested in Chapter 6, the ‘best case’ reserve
requirement shown in Figure 4.3 is also used. Thus, in this analysis, the assumed
reserve requirement is equal to the size of the largest infeed (480MW) plus the
additional reserve based on the installed wind generation.
Total wind power variations over short time frames are small, thus increasing
wind power capacity has little effect on the reserve categories that operate over a
short time frame (seconds to minutes) (Doherty and O’Malley, 2005). Increasing
amounts of wind capacity causes a greater increase in the need for categories of
reserve that act over longer periods of time (20min - 1 hour), known as replace-
ment reserve. This is the reserve category illustrated in Figure 4.3. The Irish TSO
pays generators e1.15 per MW per hour for the replacement reserve they provide
(EirGrid, 2006b). Thus, the additional reserve cost associated with increases in wind
generation is assumed to be equal to e1.15 multiplied by the additional reserve ca-
pacity from Figure 4.3.
Since the dispatch model described in Chapter 3 cooptimises energy and reserve,
there are occasions when units are switched on in order to be dispatched to provide
reserve. As the reserve requirement increases, the committment of units to provide
reserve increases. In addition, the burden of having to provide an increased level
of reserve results in the operation of more plants at lower levels of operation and
thus lower efficiencies. These indirect costs of additional reserve are captured by a
greater cycling cost which is discussed in Section 4.4 plus lower emissions and fuel
benefits, discussed in Chapter 5.
4.4 Cycling Conventional Units
In order to meet the fluctuating electricity demand throughout the day, generating
units are often required to start up and shut down and vary their output in line
with the demand changes. Certain units, generally the cheapest ones, are on all the
time and are referred to as baseloaded units. A mid merit unit would turn on in
the morning and turn off again at night when the electricity demand is low and a
shoulder unit turns on once a day when the electricity is at its highest. A peaking
unit is a flexible unit which is required less than once a day when the demand reaches
unusually high levels or when other units are unavailable (EirGrid, 2006a).
In general, most conventional units are designed for continuous rather than
variable operation and when operating in their normal range can operate for rel-
atively long periods with relatively low risk of failure and loss of equipment life
(Lefton et al., 1997). When a generating unit is required to vary its output to meet
the demand, the components in the unit are subject to stresses and strains. This
is known as cycling and includes ramping up and down during load following and
switching on and off. When a unit is turned off and on, the boiler, steam lines,
turbine and auxiliary components undergo large temperature and pressure stresses
which result in damage. This damage accumulates over time and eventually leads
to accelerated component failures and forced outages (Lefton et al., 1997).
As illustrated in Figure 4.4, once a unit is on, changing the output is the least
damaging of the cycling activities. Starting a unit is much more damaging. There
are three types of starts, a hot start, a warm start and a cold start. A hot start
refers to a unit starting up within a few hours of shutting down, while the metals in
the unit are still hot. A warm start would be when the unit has been off for slightly
longer, and the metals have begun to cool but are not yet cold. A cold start would
occur if the unit had been off for a long period and the metals were completely
cooled. This is the most damaging of all cycling activities (Lefton and Besuner,
2001).
This wear and tear on the components of the generating units is exacerbated by
a phenomenon known as creep-fatigue interaction. Creep is the change in the size or
shape of a material due to constant stress on the material over time. This is likely
in units which are operated at continuous output over long periods of time, such as
baseloaded units. Creep stems from continuous exposure to elevated temperatures
and high pressures. (Grimsrud and Lefton, 1995). Fatigue occurs when a material
is exposed to fluctuating stresses resulting in fractures and failures. Fatigue is likely
Load Hot Warm Cold

Start
Operating Level
Following Start Start
Pmax
Pmin
Damage
Figure 4.4: Damage associated with different cycling activities
during cyclical operation when the materials experience large transients in both
pressures and temperatures (Lefton et al., 1998).
Older units which were designed and used for baseloaded operation over a number
of years and are then forced to cycle on a regular basis are very susceptible to com-
ponent failure through creep and fatigue damage interaction (Grimsrud and Lefton,
1995). A brand new component can withstand a lot of fatigue damage before it
fails. However, a component that has already undergone 50% of its life creep damage
can fail with only 10% fatigue damage (American Society of Mechanical Engineers,
1990). This phenomenon is illustrated in Figure 4.5. The unit illustrated has
been operating at a relatively stable output over a long period of time (such as
a baseloaded unit) and is 50% through its expected component lifespan at point A
on the curve. If it were to continue to operate in this manner, it would achieve its
design life expectancy and would fail at point B with 100% of its creep damage.
However, if at point A, the unit is required to ramp up and down and turn on
and off more frequently, the components will experience fatigue damage. The unit
will then fail prematurely with only 10% of allowable fatigue damage at point C
(American Society of Mechanical Engineers, 1990).
Component Life Remaining

100%
Creep
50% Creep Damage
A
50%
Fatigue
C B
0
10% Damage
Figure 4.5: Creep fatigue interaction resulting in premature component failure
This depletion of the life expectancy of conventional units is a serious issue when
analysing wind generation penetration as it is likely that some units will be moved
down the merit order (as wind generation is assumed to have priority dispatch
EU Directive 2001/77/EC (2001)) and will switch from being baseloaded to being
required to be more flexible. In order to illustrate this point, Figure 4.6 shows the
operating profiles of two units on the Irish system in 2010 with 0MW and 3000MW
of installed wind.
Figure 4.6 illustrates the operating levels for a CCGT unit and a coal unit.
Both of these units operate at baseload almost all the time at low levels of installed
wind generation. It can be seen however, that when the installed wind generation
increases to 3000MW, these units become much more cyclical in operation. These
units are likely to have already undergone significant creep damage and thus the
issue of creep-fatigue interaction will be paramount if these units are required to
cycle more frequently with increased wind generation. In addition, switching from
baseloaded to variable operation will have an impact on the revenues earned by
generators in the market and may even result in the stranding of assets. This issue
is discussed in Chapter 8.
400
350
300
250
Output MW
200
150
CCGT 0
100 CCGT 3000
Coal 0
50
Coal 3000
0
0 12 24 36 48 60 72 84 96
Hour
Figure 4.6: Operating profiles of two Irish units with different wind penetrations
The actual cost of cycling is very difficult to estimate and must be conducted
on a plant by plant basis (Grimsrud and Lefton, 1995). The following approach has
been adopted to estimate the cost of cycling for conventional units.
4.4.1 The Cost of Cycling
Cycling results in an increased fuel cost for the unit as well as other more complex
costs. The costs associated with cycling are given in Table 4.1 (Lefton et al., 1997).
It is estimated that these costs can range from e200 to e500,000 per single on-off
cycle depending on the type of unit (Lefton and Besuner, 2001). Grimsrud and Lefton
(1995) calculated each of the other cycling costs as shown in Table 4.1 above and
these are shown in the ‘True Cost’ column in Table 4.2. From Table 4.2 it is clear
that the true cost of a single on-off cycle greatly exceeds the fuel cost. The cycling
costs depend mainly on the type of boiler in the unit rather than the fuel burnt,
however, for ease of reference, Table 4.2 illustrates the typical fuel types used in
Table 4.1: Costs associated with increased cycling of units
1. Replacement energy & capacity due to changes in forced outage rates

2. Additional O&M spending associated with overhaul
3. Higher heat rates due to low load and variable load operation
4. Efficiency changes due to component degradation
5. Auxiliary power, fuel and chemicals during start up
6. Unit life shortening
7. Increased operator error as a result of greater hands-on operation
Table 4.2: Cost per single on-off cycle
Boiler Type Common Fuel Used Typical Fuel Cost e True Cost e
HRSG Gas 500 1,500 - 25,000
Subcritical Drum Coal or Oil 5,000 15,000 - 100,000
Large Supercritical Coal 10,000 30,000 - 500,000
each boiler as well as their typical fuel costs for an on-off cycle.
From Table 4.2, combined and open cycle gas turbines often use a Heat Recovery
Steam Generator (HRSG) boiler, also known as a waste heat boiler. The majority
of the gas plants on the Irish system use HRSG boilers. The terms subcritical and
supercritical come from the definition of the temperature and pressure at which
water vapour and liquid water are indistinguishable, known as the critical point.
This occurs at a temperature of 374◦ C and a pressure of 22MPa. Thus, subcritical
boilers operate below this critical point and are often found in coal and oil plants. All
of the Irish coal and oil plants use this boiler design (EirGrid, 2005b). Supercritical
units operate at pressures and temperatures above the critical point to improve
efficiency and can be found in some large coal units (Cregan and Flynn, 2003). There
are currently none of these units installed on the Irish system, however they can be
found in some European countries, Asia and the United States (Dalton, 2004). It is
apparent from Table 4.2 that the true cost of cycling is many times greater than the
fuel cost. The fuel cost typically represents only a fraction, between 2% and 30%,
of the real cost of cycling a generating unit.
In order to estimate the cycling costs of the generators on the Irish system with
increases in wind generation, the following approach was adopted (Lefton et al.,
1997). Once the dispatch of the generators had been determined, as described in
Chapter 3, the number of cold, warm and hot starts for each generator were counted.
In addition, when a generator’s output varied, the change in output was expressed
as a percentage of the maximum rated capacity of the generator. In order to compile
the total cycling cost, these cycling activities were expressed in terms of the damage
from a single hot start (Combined Cycle Journal, 2004). The rate of temperature
and pressure change under each of the cycling activities was determined and this was
considered to be equivalent to the damage incurred for each activity (Lefton et al.,
1997). Each of the cycling activities could then be expressed in terms of the damage
incurred during a hot start, referred to from here as an ‘equivalent hot start’ (EHS)
as shown in Table 4.3 (Lefton et al., 1997).
Table 4.3: Cycling activity expressed as an equivalent hot start
Activity EHS
Cold Start 2.28
Warm Start 1.77
Hot Start 1.00
10% Load Swing 0.05
Thus, from Table 4.3, the damage from a cold start is equivalent to 2.28 times the
damage from a hot start. A load swing equal to 10% of the rated capacity of the unit
is the equivalent of 0.05 times a hot start. Thus, each category of cycling activity
was multiplied by the figures in Table 4.3 to give the total number of equivalent hot
starts for each generator over the course of the year.
Once the number of EHS for each generator had been determined, the costs
could be calculated. As shown in Table 4.2, the fuel cost for a cycling activity
represents between 2 and 30% of the true cost of cycling depending on the unit. Thus,
using specific generator information (Appendix A), the age and previous operating
patterns of the generators (in an attempt to estimate creep damage and flexibility)
and unit cycling descriptions from Lefton et al. (1998), the fuel cost for each unit
was assumed to represent a percentage of its total cycling cost.
For example, Ireland has a number of large coal fired units that have been in
operation for over 20 years. These units are baseloaded throughout the year, in
fact, Lefton et al. (1998) found that these units are more baseloaded than any in
the United States. The energy consumed in one hot start for one of these units is
4,360GJ (Appendix A). Using the ’mid’ fuel price for coal of e2.17/GJ (see Table 3.1
in Chapter 3) this implies that the fuel cost of a single hot start is e9,461. These
units use large subcritical drum boilers and have been heavily baseloaded over a
long period of time, and are likely to have incurred a large amount of creep damage
(Lefton et al., 1998). The fuel cost is therefore estimated to represent 10% of the
total cycling cost for these units. Thus, the true cost of a single hot start for one of
these coal units is e9,461 ÷ 0.10 = e94,610.
The dispatch model described in Chapter 3 was used to determine the operating
schedules of conventional generators with increasing penetrations of wind generation.
In order to calculate the cycling costs associated with increased wind generation, the
total cycling cost with zero installed wind generation was determined as described
above. This was then subtracted from the cycling costs at each level of installed
wind generation. This was conducted for each of the test years and the resulting
additional cycling costs are shown in Figure 4.7.
As Figure 4.7 shows, wind generation causes an increase in the cost of cycling
across each of the three test years. By analysing the operating schedules of the
generators, it is seen that wind generation causes an increase in the cycling of the
marginal units. In 2010 and 2015 there is a reduction in the added cycling costs
between 500 and 2000MW of installed wind generation. This is due to the wind
generation pushing the oil units down the merit order so they are used less frequently
and are replaced on the margin by more flexible gas fired generation. As the installed
wind generation increases beyond 2000MW, more units are displaced and plants
200
180
160 2010
2015
140 2020
Cycling Cost (em)
120
100
80
60
40
20
0
0 500 1000 1500 2000 2500 3000 3500 4000
Figure 4.7: Cycling costs with increasing penetrations of wind generation
which had been previously baseloaded are moved down to the margin. These are
units with very high cycling costs and thus, at high penetrations of wind generation,
the cycling costs are large.
4.5 Summary of Costs Associated with Wind Genera-

tion
This chapter outlines the costs associated with increasing penetrations of wind gen-
eration. The costs included were the capital costs of wind generation, the operation
and maintenance costs, the deep network reinforcement costs and the cost of the
additional cycling of the conventional units. The salient results associated with each
of these costs is summarised here:
• Four different capital cost profiles will be examined in this thesis. These
capital costs include the assumed shallow connection cost to the network and
are represented as an annuity with a term of 20 years.
• The operation and maintenance costs are assumed to be equal to e35,000

per annum per MW installed.
• The deep network reinforcement costs are taken from AIRGS (2006b)
and come into force above 1500MW installed. These costs are included as an
annuity with a term of 20 years.
• The additional reserve costs are based on the reserve capacities with in-
creased wind generation in Doherty and O’Malley (2005). Three different re-
serve scenarios will be examined in Chapter 6 based on the accuracy of the
assumed wind power forecasts. The reserve cost is assumed to be e1.15/MWh
(EirGrid, 2006b).
• The additional cycling costs are the only costs calculated based on the dis-
patches from the model described in Chapter 3. The cycling activities are
expressed in terms of equivalent hot starts and the cost is calculated based
on their fuel cost during a hot start. Increasing penetrations of wind genera-
tion cause an increase in cycling of marginal units across all three test years
resulting in increased costs.
The sensitivity of the net benefits to many of the assumptions discussed here will
be investigated in Chapter 6. In addition, the costs of wind generation described
in this chapter are not an exhaustive list of the costs of wind generation, thus the
implications of including other costs in the analysis is discussed in Chapter 8.
CHAPTER 5
The Benefits of Wind Generation
T
HIS chapter discusses the benefits associated with increased wind generation.
As with the costs, the benefits are not enjoyed by any entity in particular but
are assumed to be attributable to society in general.
The capacity credit of wind generation is a measure of the amount of conventional
generation that could be displaced by the renewable production without making the
system any less reliable (Castro and Ferreira, 2001). For low levels of installed wind
generation, the capacity credit tends to approximate the average wind output. How-
ever, as wind penetrations increase, the capacity credit tends to reduce as the corre-
lation of the wind generation output on the system increases (Castro and Ferreira,
2001). The value of the capacity benefit of wind generation is discussed in Section
5.1.
As the installed capacity of wind generation increases it displaces conventional
generation which has an impact on the emissions from the conventional units. Emis-
sions of carbon dioxide (CO2 ) and sulphur dioxide (SO2 ) depend on the quantity of
carbon and sulphur in the fuel respectively and the quantity of fuel burnt. Thus,
a reduction in the operation of a thermal unit will result in a reduction in CO2
72
Chapter 5. The Benefits of Wind Generation 73
and SO2 emissions. Nitrogen Oxides (NOX ) formation is more complex and does
not depend solely on the nitrogen content of the fuel. The formation of emissions
in combustion plants is discussed in Section 5.2. The dispatch model described in
Chapter 3 is used to quantify the emissions savings from wind generation and these
benefits are shown in Section 5.3.
When wind generation displaces electricity produced from thermal units, the
quantity of fuel burnt by the thermal units reduces and wind generation provides
a fuel saving (Malik and Awsanjli, 2004). The value of this saving depends on the
price of the fuel saved and is impacted by a number of factors. The load factor of
the wind generation gives the MWh produced per MW installed and as this reduces
the incremental fuel saving reduces. An increase in the cycling of conventional units
will reduce the fuel savings from wind generation as start ups are fuel expensive.
The fuel savings associated with increases in wind generation are determined using
the dispatch model and are discussed here in Section 5.4.
A summary of the benefits of wind generation and the main assumptions dis-
cussed in this Chapter are given in Section 5.5.
5.1 Capacity Benefit
One of the key benefits associated with increased wind generation is the additional
capacity it adds to the system. The extent to which wind generation can substitute
for conventional generation without reducing the reliability of the system is given
by the capacity credit of wind (Castro and Ferreira, 2001). Variable sources of
generation, such as wind, make a different contribution to the capacity on the system
than dispatchable generation. Although wind generation can serve a large proportion
of the load, it may not necessarily be the case that the times of high wind generation
coincide with times of high demand (Doherty et al., 2006).
Doherty et al. (2006) use a monte-carlo approach to determine the capacity
credit of wind generation in Ireland. The reliability criteria to be maintained with
increasing wind generation was a loss of load expectation (LOLE) of 8 hours per
year. Wind generation was added to the system, and the wind power profiles were
subtracted from the load profile. This resulted in a reduction in the LOLE as the
installed generation on the system was relatively larger compared to the load. The
load was then systematically increased until the LOLE returned to 8 hours per year.
The capacity credit of wind generation was calculated by dividing the installed wind
generation by the increase in load. Figure 5.1 illustrates the capacity credit of wind
generation from Doherty et al. (2006) which is used in this analysis.
0.5 100
Capacity Credit
0.4 80
Capacity Credit
Benefit in em
0.3 60
0.2 40
0.1 20
Value of Capacity Benefit
0
0 500 1000 1500 2000 2500 3000 3500 4000
Figure 5.1: The capacity credit of wind generation in Ireland
As illustrated by Figure 5.1, as wind penetrations increase, the capacity credit

reduces. This is due to an increase in the correlation of the wind generation output
on the system (Castro and Ferreira, 2001). As the capacity credit is defined as
the extent to which wind generation can substitute for conventional generation, the
capacity benefit of wind generation can therefore be deemed as the saved cost of
having to build and maintain additional conventional generation in its place. In
other words, from Figure 5.1, 1000MW of installed wind generation has a capacity
credit of 34%. Thus, the development of 1000MW of wind generation would save
building and maintaining a 378MW conventional plant with an availability of 90%

(378MW × 0.9 = 340MW).
For the purposes of this thesis, it is assumed that new conventional generation
built in Ireland will be gas fired, with a capital cost of e600,000 per MW installed
and operation and maintenance costs of e50,000 per MW per year (CER, 2006b;
Doherty et al., 2006). The capital cost is converted to an annuity in order to be
expressed as an annual cost. The broken line in Figure 5.1, represents the annual
capacity benefit of wind generation with the monetary values on the secondary y-
axis.
5.2 Emissions in Conventional Generators
Harmful emissions are created in combustion plants through the burning of fuels
at elevated temperatures. Wind generation can result in a reduction in the output
of conventional units on the system, however, it is not necessarily the case that a
reduction in output will result in a reduction in emissions. This thesis investigates
the impact of wind generation on emissions of carbon dioxide (CO2 ), sulphur dioxide
(SO2 ) and nitrogen oxide (NOX ). This Section describes how these emissions are
created in combustion units.
5.2.1 Carbon Dioxide Formation
Carbon dioxide is generated by the combustion of fuels containing carbon. The

amount of carbon dioxide released is in direct relationship with the amount of carbon
in the fuel and the quantity of fuel burnt (EIA, 2004). Thus a generation plant which
burns a carbon intensive fuel will generate more carbon dioxide at increased levels
of operation. The following expression was used to calculate the CO2 emissions of
a combustion plant.

E
CO2 /M W h = (CO2 /kgfuel) (5.1)
CV
where E represents the generator’s energy consumed in megajoules (MJ) per

MWh and CV is the calorific value of the fuel in MJ per kilogram. CO2 /kg fuel
represents the amount of CO2 emissions released per kilogram of fuel burnt.
5.2.2 Sulphur Dioxide Formation
Sulphur (S) is found in hydrocarbon fuels and is mostly in pure form. During com-
bustion sulphur combines with carbon, hydrogen and oxygen to form SO2 , SO3 , SO,
CS, CH, COS, H2 S, S and S2 (Hunter, 1982). Given the high temperatures and
oxygen concentrations during combustion, sulphur dioxide (SO2 ) is the principal
sulphur compound formed in combustion. Even with 20% air deficiency, 90% of the
sulphur is in the form of SO2 and as little as 0.1% is as SO3 ; SO accounts for the
remainder of the sulphur (Harris, 1990). As a result, the analysis that follows will
concentrate on emissions of SO2 rather than the alternative sulphur compounds.
Also the NECD and the Large Combustion Plant Directive indicate limits on allow-
able emissions of SO2 (EU Directive 2001/80/EC, 2001). The expression used to
calculate sulphur dioxide emissions is the same as that given in (5.1) substituting
SO2 for CO2 .
5.2.3 Nitrogen Oxides Formation
Oxides of Nitrogen (NOX ) are formed by the combination of nitric oxide (NO) and
nitrogen dioxide (NO2 ). NO and NO2 are formed during combustion by the reaction
of nitrogen present in the combustion system, either in the fuel or in the combustion
environment. Normally NO is formed in much larger amounts than NO2 , and NO2
is formed by further reaction of NO (Cunningham, 1978). Thus, NO formation
determines the total amount of NOX emitted.
Unlike CO2 and SO2 , NOX formation does not depend solely on the nitrogen con-
tent of the fuel. It is also significantly affected by the flame temperature, the oxygen
concentration and the residence time. The formation of NOX can be attributed
to four distinct chemical kinetic processes: thermal NOX formation, prompt NOX
formation, fuel NOX formation and reburning. Thermal NOX is formed by the
oxidation of atmospheric nitrogen present in the combustion air. Prompt NOX is
produced by high-speed reactions at the flame front, and fuel NOX is produced by
oxidation of nitrogen contained in the fuel. The reburning mechanism reduces the to-
tal NOX formation by accounting for the reaction of NO with hydrocarbons (Kesgin,
2003). Although these processes are different they may operate concurrently.
5.2.4 Coal, Peat and Heavy Fuel Oil Generators
As described above, the carbon dioxide and sulphur dioxide emissions from a gen-
eration plant depend on the chemical content and the calorific value of the fuel. A
summary of the characteristics of coal, peat and heavy fuel oil are given in Table
5.1 (EIA, 2004; Molero, 1997). Gas fired units are dealt with separately in Section
5.2.5.
Table 5.1: Typical components of various fuel types
Carbon Sulphur Nitrogen Calorific Value CO2 /kg fuel

Coal 65% 0.8% 1.3% 28 MJ/kg 2.49
Peat 25% 0.15% 0.65% 7.75 MJ/kg 1.25
Heavy Fuel Oil 87% 2% 0.2% 40 MJ/kg 0.79
The composition of a fuel depends on the composition of the source fuel and is
manufacturer dependent. Since this is unique and varies from source to source and
geographical origin, Table 5.1 give typical values only. It is clear from Table 5.1 that
although coal may contain more carbon than peat and as a result will produce more
CO2 per kilogram of fuel used, the quantity of peat required to produce one MWh is
significantly more than that required of coal. Heavy fuel oils usually contain higher
amounts of sulphur than other petroleum products as sulphur tends to concentrate in
the residue during the refining processes (Lightman and Street, 1981). Low sulphur
heavy fuel oil would have a lower sulphur content, about 0.5 - 1.0%
Fuel NOX is the major source of NOX emissions from the combustion of nitrogen
bearing fuels such as heavy oils, coal and peat (Li et al., 2003). During combustion
the fuel bound nitrogen is converted to fixed nitrogen species such as HCN (a combi-
nation of hydrogen, carbon and nitrogen) and ammonia (NH3 ). These, in turn, are
readily oxidised to form NO when they reach the lean zone of the flame. Between
20 and 80 percent of the bound nitrogen is typically converted to NOX , depending
on the design of the combustion equipment (Li et al., 2003).
5.2.5 Gas Fired Generators
Gas fired generation, differs from coal, oil and peat generation and is dealt with
separately here. The typical characteristics of natural gas are shown in Table 5.2.
Table 5.2: Typical components of natural gas
Carbon Sulphur Nitrogen Calorific Value CO2 /kg

Gas 70% 0% 1.3% 48 MJ/Nm3 2.68
Carbon dioxide formation in a gas fired generator is formed as described in

Equation (5.1). As shown in Table 5.2, natural gas contains a negligible amount
of sulphur, thus emissions of SO2 are not significant for gas turbines (EIA, 2000).
Thermal NOX is the predominant source of NOX emissions from a gas turbine. Due
to the increasing number of gas fired generation plants on electricity systems, and
their unique NOX characteristics, the formation of NOX in a gas turbine will be
outlined in more detail below.
Thermal NOX is formed by the following reactions, known as the Zeldovich
mechanism (Delabroy et al., 1998):
O + N2 ⇔ N O + N (5.2)
N + O2 ⇔ N O + O (5.3)
N + OH ⇔ N O + H (5.4)
The reaction (5.2) determines the rate of thermal NOX production, and shows
that NO can be minimised by reducing the concentration of [O] or [N2 ]. This rate
of thermal NOX production becomes significantly faster at high temperatures, thus,
reducing the temperature results in a reduction in NO formation (Delabroy et al.,
1998). However, reducing the flame temperature reduces the efficiency of the plant.
Combined cycle gas turbine developers are constantly striving to improve the effi-
ciency of the turbines and this is often done through increased flame temperature.
Combined cycle gas turbine (CCGT) systems generate around 40% more elec-
tricity than conventional generating systems from the same amount of gas. In order
to achieve reduced emissions, gas turbine manufacturers have adopted lean premixed
combustion as a standard technique. This premix (of fuel and air) achieves low levels
of pollutant emissions without the need for additional hardware for steam injection
or selective catalytic reduction. By premixing the fuel and air prior to firing, lo-
calised regions of near stoichiometric fuel-air mixtures are avoided and a subsequent
reduction in thermal NOX can be realised (Mansour et al., 2001). Lean premixed
combustion is limited by the presence of combustion instabilities, which induce high
pressure fluctuations, which can produce turbine damage, flame instability and even
flame extinction (Cabot et al., 2004). For this reason the fuel and air premix is not
possible during startup and at reduced load levels (below about 65-70% of maxi-
mum capacity). As a result, NOX emissions in a CCGT increase significantly at
lower loads. Figure 5.2 shows the NOX characteristics of a CCGT and an open cycle
gas turbine (OCGT), also included for illustrative purposes are the NOX emissions
from a coal unit which has installed a Selective Catalytic Reducer (SCR or scrubber)
to reduce its NOX emissions (O’Mahony, 2004).
From Figure 5.2 it is clear that if a CCGT is forced to operate below approxi-
mately 70% of its maximum rated capacity, its NOX emissions will increase threefold.
For an OCGT, operation below 60%, will result in NOX emissions increasing by up
to eight times. The NOX characteristics of these units are critical when examining
wind generation. As discussed in Chapter 4, wind generation may result in the
system operator requiring more units to operate at lower efficiencies due to the in-
creased reserve requirement. In addition, wind generation causes an increase in the
4.5
4
CCGT
OCGT
3.5
NOX Emissions (kg/MWh)
Coal SCR
2.5
1.5
0.5
0
0 10 20 30 40 50 60 70 80 90 100
Percentage of Maximum Capacity
Figure 5.2: Typical NOX emissions from a CCGT and an OCGT
cycling of marginal units. Thus, an increase in wind generation may result in more
CCGT units operating at lower loads, increasing their NOX emissions. In other
words, although wind generation itself does not produce any harmful emissions, it
may in fact result in an increase in NOX emissions through the reduced operation
of gas fired units.
Figure 5.2 also shows the NOX emissions from a coal fired unit with a scrubber in-
stalled. Under the large combustion plant directive, all the coal units on the Irish sys-
tem are expected to have this technology installed by 2010 (EU Directive 2001/80/EC,
2001; ESB, 2003). As such, a coal fired unit could actually produce less NOX emis-
sions than a CCGT, if the gas unit was operating at less than 70% of maximum
capacity.
5.2.6 Emissions During Cycling
According to the IPPC (1996), CO2 emissions during start up must be included in
the total CO2 emissions of a unit. These can be very large. For example, a 285MW
coal fired unit on the Irish system uses 14,620GJ of energy when started from cold
(see Appendix A). From Table 5.1, the calorific value of coal is 28MJ/kg and the
CO2 emissions per kg of fuel burnt are 2.49kg. Thus, using Equation (5.2.1), the
CO2 emissions for a single cold start are given by:

1
CO2 = (14, 620) (2.49) = 1, 300tons (5.5)
0.028
From Table 3.1, if the price for coal was assumed to be e 2.17/GJ, then the fuel
cost of a cold start would be e 31,725. If the price of carbon was e 30/ton, then
the carbon cost of a cold start would be e 39,004. In other words, with a carbon
price of e 30/ton, the carbon cost of starting this unit would exceed its fuel cost.
CO2 and SO2 emissions do not change significantly during ramping. However,
due to the increased oxygen levels present during ramping, NOx emissions increase
by about 10% over steady state conditions during periods of significant ramping
(O’Mahony, 2004).
5.3 Emissions Benefits of Wind Generation
Once the operating levels of the conventional units had been attained using the
model described in Chapter 3, the resulting CO2 , SO2 and NOx emissions from
the conventional units were calculated for each hour by using specific emissions
information for each individual generator (EirGrid, 2006a; O’Mahony, 2004; AIP,
2005). Figure 5.3 illustrates the emissions benefits from increasing levels of wind
generation for CO2 , SO2 and NOx for the 2010 plant mix and load.
The magnitude of CO2 emissions is much larger than for the other two emissions,
however, for ease of illustration all three emissions have been plotted on the same axis
in Figure 5.3. It can be seen that as wind generation increases, the system emissions
of CO2 , SO2 and NOx are reduced. However, the relationship is non-linear. It
30000
25000
20000
Kilotons
30
25
20
15
CO2
10 SO2
NOx
5
0
0 500 1000 1500 2000 2500 3000 3500 4000
Figure 5.3: Emissions savings with increasing wind generation
can be seen that across all three emissions the incremental savings reduce as the
wind increases and that for high levels of installed wind generation, the emissions
actually begin to increase. This is due to a number of reasons. Firstly, as the wind
generation increases, its load factor reduces and thus the conventional generation it
displaces reduces. Secondly, increasing penetrations of wind generation require an
increased reserve capacity to be carried on the system, this results in more units
operating at reduced levels. Operating at a reduced output, the generating units
are less efficient and thus the emissions savings are reduced. Thirdly, as shown in
Chapter 4, increasing wind generation causes an increase in the cycling of units.
Thus, the potential CO2 and NOX emissions savings are reduced due to an increase
in startups and ramping respectively.
At low levels of wind capacity, the marginal units are mainly oil and peat fired,
thus displacing these units results in reduced CO2 , SO2 and NOX emissions. How-
ever, as the installed wind goes beyond 2500MW, the marginal units tend to be
made up of gas fired units which produce less CO2 emissions per MWh and no SO2
emissions, hence the negligible SO2 reductions. Above about 1000MW of installed
wind, the reduction in NOX emissions from the displacement of oil units is coun-
teracted by an increase in emissions from CCGTs running at lower loads. Beyond
3000MW there is an increased use of peaking units which results in an increase in
SO2 emissions from the distillate generators and in NOX emissions from the OCGTs.
For the purposes of this analysis it is necessary to express these emissions savings
in monetary terms. Under the EU ETS (2003), there is currently an EU wide CO2
emissions market where generators buy and sell allowances for CO2 . Thus, the
CO2 emissions will be valued at representative market prices of CO2 . There is not
currently an emissions market for SO2 and NOx in Europe, however, there is a
market for these emission in the United States. Thus, the assumed value of these
emissions are based on the prices in these emissions markets in the United States
(US EPA, 2006b,c). Four different emissions prices are considered in this thesis and
are summarised in Table 5.3.
Table 5.3: Emissions prices in e per ton
Emission Low Mid High Highest

CO2 10 30 50 70
SO2 50 150 250 350
NOx 1000 3000 5000 7000
In the dispatch model, an emissions factor is included in the bid price of each
generator to represent the price of CO2 , SO2 and NOx . This is done by calculating
the average emissions level per MWh of production for each generator on the Irish
system. These levels were then multiplied by the assumed emissions prices in Table
5.3 and added to the fuel costs to give the bid prices of generators. At the margin,
all emissions must be included, thus generators are assumed to pay the emissions
price for the full extent of their emissions1 .
1
The inclusion of emissions prices in the bids of generators can alter the merit order, with heavy
polluters becoming more expensive relative to light polluters. This can have a significant impact on
the cycling costs. The interaction between emissions prices and cycling costs is developed in detail
in Chapter 7.
Figure 5.4 illustrates the monetary value of the emissions savings from wind in
2010. For each curve, the emissions savings are valued at the emissions prices, and
these prices are also used in the dispatch of the generators.
500
Highest Emission Prices
450 High Emission Prices
Mid Emission Prices
400 Low Emission Prices
Emissions Benefit (em)
350
300
250
200
150
100
50
0
0 500 1000 1500 2000 2500 3000 3500 4000
Figure 5.4: Value of emissions savings with increasing wind generation
It can be seen that the emissions savings peak at 3000MW installed and then
reduce in line with the actual emissions savings. The sensitivity of the net benefits
to changes in the emissions prices are discussed in detail in Chapter 6.
5.4 Fuel Savings Benefit
As wind generation displaces electricity produced from thermal units the quantity
of fuel burnt by the thermal units changes. However, like the emissions savings, a
reduction in output does not mean a linear reduction in fuel burnt. When operating
at low loads, generators are less efficient and thus their fuel consumption per MWh
is higher. In addition, an increase in the cycling of generators causes increased fuel
consumption.
The model described in Chapter 3 determined the operating schedules for the
conventional generators with increasing penetrations of wind generation. Once these
dispatches had been determined, the consumption of fuel was calculated by analysing
the GJ of energy consumed per MWh for each generator (see Appendix A). Figure
5.5 illustrates the fuel consumption by conventional generators on the Irish system
for the year 2010 as the installed wind capacity increases.
200
180
Energy Consumed in Petajoules
160 Gas
Coal
140 Peat
Oil
120
100
80
60
40
20
0
0 500 1000 1500 2000 2500 3000 3500 4000
Figure 5.5: Fuel consumption with increasing wind generation
Figure 5.5 illustrates the annual fuel consumption in petajoules for each fuel type
on the Irish system for 2010, and shows how the consumption changes with increasing
wind generation. The majority of generation on the Irish system is gas fired so wind
generation has a significant impact on the consumption of gas. Although much
less generation is provided by oil and peat units, their consumption is significantly
reduced by the wind generation. The coal units tend to be baseloaded and thus
their fuel consumption is really only affected at high wind penetrations.
For the purposes of this cost benefit analysis, these fuel savings are valued at the
assumed fuel prices for each fuel type. The assumed fuel prices are shown in Table
3.1 in Chapter 3. Figure 5.6 shows the value of the fuel savings in 2010 for each of
the three fuel price scenarios.
250
200
Fuel Savings in em
150
100 High Fuel Prices

Mid Fuel Prices
Low Fuel Prices
50
0
0 500 1000 1500 2000 2500 3000 3500 4000
Figure 5.6: Value of fuel savings with increasing wind generation
It can be seen from Figure 5.6, that although changes in the assumed fuel prices
do alter the value of the fuel savings, the impact is not hugely significant. As was
seen with the emissions savings, the value of the fuel savings increase up to 3000MW
installed and beyond this point they reduce again.
5.5 Summary of Benefits of Wind Generation
This chapter outlined the benefits associated with increased wind capacity on the
Irish system. The capacity benefit was determined based on the capacity credit of
wind generation while the emissions and fuel savings were based on the operating
schedules from the dispatch model. A summary of the key points from this chapter
are given here.
• Although it is variable in nature, wind generation adds to the capacity on the

system and the level of this benefit is given by the capacity credit. The capacity
credit decreases as the installed wind generation increases and is valued as the
saved cost of having to build and maintain an equivalent conventional unit.
• The creation of harmful emissions in conventional units was discussed and

it was shown that reduced operation reduces CO2 and SO2 emissions but not
necessarily NOX emissions. In addition, reduced operating levels and increased
cycling may ameliorate the potential emissions savings from wind generation.
• It was shown that wind generation provides emissions benefits at all penetra-
tions, however the incremental savings reduce as the installed wind generation
increases. The emissions savings were based on the operating schedules in the
dispatch model.
• Wind generation provides fuel savings for conventional units, however, the
incremental savings reduce as the installed wind generation increases due to
low load and cyclical operation by conventional units.
The sensitivity of the net benefits to changes in the assumptions surrounding

the benefits of wind generation will be investigated in Chapter 6. In addition, this
chapter touched on the impact of emissions prices on the dispatch of the generators,
this will be discussed in detail in Chapter 7. The softer benefits of wind generation,
such as the creation of jobs and improvements in local infrastructure are not explic-
itly dealt with in this thesis, however, the implications of including these benefits is
discussed briefly in Chapter 8.
CHAPTER 6
The Net Benefits of Wind Generation
C
HAPTERS 4 and 5 set out the costs and benefits associated with wind gen-
eration for the case study. In this chapter, the quantified costs and benefits
are brought together to determine the optimal levels of installed wind generation for
the case study.
Section 6.1 shows the net benefits curve and discusses the optimal penetration
levels of wind generation in the base case for each of the test years. The assumptions
underlying the base case are also summarised. Section 6.2 presents a sensitivity
analysis and discusses the implications of changing the assumptions from the base
case and determines the optimal penetrations of wind generation for each sensitivity.
6.1 The Net Benefits of Wind Generation in the Base

Case
The annual costs of wind generation were described in detail in Chapter 4 and in-
cluded the capital cost of wind generation, the operation and maintenance costs, the
deep network reinforcement costs, the reserve costs and the cycling costs. In order
88
Chapter 6. The Net Benefits of Wind Generation 89
to calculate the net benefits of wind generation, these annual costs are combined to
give a single total cost (TC) for each level of installed wind generation w, shown by
Equation (6.1):
T
X 1
T Cw = (Cw + OMw + N Ww + Rw + CYw ) (6.1)
t
(1 + r)t
where r represents the assumed discount rate and T the term over which the costs
and benefits are discounted. The capital costs for wind penetration w are expressed
as an annuity and are denoted by C w . The annual operation and maintenance costs
are given by OM w , NW w is the network reinforcement costs expressed as an annuity,
R w is the additional annual reserve cost and CY w is the added annual cycling costs.
The investigated benefits of wind generation were discussed in Chapter 5 and
are summed to give the total benefits TB w , shown in Equation (6.2).
T
X 1
T Bw = (CPw + Ew + F Sw ) (6.2)
t
(1 + r)t
where CP w is the capacity benefit of wind generation, derived from the saved
cost of building and maintaining equivalent conventional generation, and is expressed
as an annuity. The annual emissions savings are given by E w and the annual fuel
savings by FS w .
The net benefits for each level of installed wind generation are then determined
by subtracting the total costs from the total benefits for each level of installed
wind capacity. This Section discusses the net benefits of wind generation under the
assumptions chosen in the base case. These assumptions were discussed in Chapters
3, 4 and 5 and are summarised in Table 6.1. Figure 6.1 illustrates the net benefits
curve for wind generation for each of the test years 2010, 2015 and 2020 under these
base case assumptions.
The net benefit curves illustrate where the incremental net benefits begin to
decline and more importantly, the points where the costs equal the benefits (where
net benefits equal 0), representing the critical points beyond which no further in-
vestment should be made in wind generation. Beyond the critical point, the costs
Table 6.1: Base case assumptions summarised
Assumption Scenario
Load Factor ‘Mid’ scenario in Figure 3.3
Wind Power Forecast Error Standard Deviation of 9% (Section 3.2)
Demand Growth 3.7% per year
Fuel Price Growth ‘Mid’ prices in Table 3.1
Wind Capital Costs ‘Mid Capital’ in Figure 4.1
Wind O&M Costs e35,000 p.a. per MW installed
Network Upgrade Costs See Figure 4.2
Reserve Costs ‘Average’ curve in Figure 4.3
Cycling Cost See Figure 4.7
Capacity Benefit See Figure 5.1
Emissions Costs ‘Mid’ emissions prices in Table 5.3
Discount rate 8%
Term 20 years
4000
3000
2000
Optimal
Net Benefit em
1000
0
2020
2015
−1000
2010
−2000
−3000
0 500 1000 1500 2000 2500 3000 3500 4000
Figure 6.1: The base case net benefits of wind generation

will exceed the benefits and the net benefits will be negative. The critical point
for 2010 occurs at approximately 2,790MW of installed wind capacity, representing
approximately 21% of total electricity generation. The net benefits for 2010 are
less than in 2015 and 2020 due to the installed plant mix being more inflexible and
having less interconnection. In addition, the presence of an east west interconnector
with a capacity of 400MW makes a highly significant impact on the net benefits of
wind generation in 2015 and 2020. The replacement of the oil fired units with gas
fired units is also highly beneficial to wind in 2015 and 2020. From 2010 to 2020 the
underlying plant mix evolves to represent more closely the optimal plant portfolio
for the test system with large penetrations of wind generation (Doherty et al., 2006).
The optimal plant portfolio for the case system should have less baseloaded units
and more open cycle gas turbines, as is the case for the 2020 plant mix. The critical
point for wind generation in 2015 is 2,994MW and in 2020 is 3,705MW, representing
approximately 21.5% and 22.2% of electricity generated respectively.
Given the large number of assumptions required in this study, perhaps more
interesting than the absolute values of the critical points of wind generation are the
sensitivity of these critical points to changes in the underlying assumptions. Thus,
Section 6.2 investigates the impact on the net benefits of wind generation of altering
the assumptions in base case.
6.2 Sensitivity Analysis of Net Benefits
A range of scenarios are examined in this Section and for ease of comparison across
the three test years, the critical values will be expressed in terms of percentage of
electricity generated from wind rather than the installed MW. For each scenario, the
net benefits curve is recalculated and the critical value of installed wind generation,
where the net benefits equal zero, is determined. Unless otherwise stated, the only
value altered in each scenario is the one under investigation, and all other values
are as they are in the base case. Figures 6.2 and 6.3 represent a graphical summary
of the critical values for wind generation across each of the investigated scenarios.
Each scenario is then explained and discussed in turn in the following subsections.
Chapter 6. The Net Benefits of Wind Generation
BC = Base Case
LLF = Low Load Factor
CLF = Constant Load Factor
HLF = High Load Factor
WFC =Worst Forecasts
BFC = Best Forecasts
LD = Low Demand
HD = High Demand
LFP = Low Fuel Prices
HFP = High Fuel Prices
LCp = Low Capital
CCp = Constant Capital
HCp = High Capital
92
Figure 6.2: The Critical Values of Wind Generation Part I
Chapter 6. The Net Benefits of Wind Generation
BC = Base Case
50Cy = Low Cycling Costs
80Cy = Very Low Cycling Costs
LEp = Low Emissions Prices
HEp = High Emissions Prices
HHEP = Highest Emissions Prices
6% = Discount Rate of 6%
10% = Discount Rate of 10%
Bad = Worst Case Scenario
Good = Best Case Scenario
93
Figure 6.3: The Critical Values of Wind Generation Part II
Although this thesis analysed only installed wind capacities up to 4000MW (rep-
resenting 27.6% of electricity generated in 2010) Figures 6.2 and 6.3 show percentages
up to 30%. This is due to the fact that in some scenarios, the critical values were
above 4000MW and in these cases, the critical values are found by linear extrapola-
tion from the net benefits curve at 4000MW.
6.2.1 Wind Load Factor Scenario
In Chapter 3 the assumptions regarding the load factors of wind generation were
discussed. Four scenarios were illustrated with low, mid, constant and high load
factors respectively. Represented in Figure 6.2 by the columns ‘LLF’ meaning low
load factor, ‘CLF’ meaning constant load factor and ‘HLF’ for high load factor in
Figure. The mid load factor was assumed in the base case scenario (BC). Figure 6.4
illustrates the impact assuming the other load factors would have on the net benefits
of wind generation in 2010.
3000
2000
1000
Net Benefits em
High Load Factor (HLF)

−1000 Constant Load Factor (CLF)
Base Case (BC)
Low Load Factor (LLF)
−2000
−3000
0 500 1000 1500 2000 2500 3000 3500 4000
Installed Wind Generation
Figure 6.4: The net benefits of wind with changes in load factor
It can be seen from Figure 6.4 that assuming a high load factor shifts the net
benefits curve upwards from the base case. This is because for the same installed
capacity, more energy is reaped from the wind generation resulting in higher fuel
and emissions savings but for the same capital cost. However, the upward shift is
not the same across all penetrations of wind generation and it can be seen that the
critical value for the ‘constant’ and ‘high’ cases are similar. This is because although
there are higher benefits to a high load factor, at increased penetrations of wind,
there will also be an impact on the cycling costs. A low load factor shifts the net
benefits curve significantly downwards resulting in a reduced critical value. It is also
interesting to note that at a low load factor, the net benefits of wind generation at
low levels of penetration actually dip below zero. This is due to the lower emissions
and fuel benefits but a relatively high capital cost.
The net benefits curves for the load factor sensitivities for 2015 and 2020 were
also calculated and the curves followed the same general patterns. These net benefits
curves were calculated for each test year for all the sensitivities to determine the
optimal penetrations in each, however, for ease of illustration, only the net benefits
curves for the sensitivity to changes in the load factor are shown. A summary of the
actual critical points for all the sensitivities are given in Figures 6.2 and 6.3.
6.2.2 Wind Power Forecast Accuracy Scenario
Wind power forecasts were included in the dispatch model by assuming a long range
forecast in the commitment stage of the model and an updated forecast in the
dispatch stage. The base case assumption for these wind forecasts was that the
standard deviation of the wind power forecast errors was 9% of the installed wind
capacity. The impact on the net benefits of wind generation if a worst case wind
forecast, with 14% standard deviation is used and if a best case wind power forecast
with 7% standard deviation was examined for each of the test years. These scenarios
are represented in Figure 6.2 by ‘WFC’ and ‘BFC’ respectively.
It was found that assuming the worst case wind power forecasts has a dramatic
impact on the critical values of wind generation, dropping to 16% in 2010. Wind
power forecasts impact on the reserve capacity being carried on the system, as
discussed in Section 4.3. Thus, assuming the worst case wind power forecasts results
in more reserve capacity being carried on the system. This increases the reserve costs.
Also, more units are operating at lower efficiencies resulting in reduced emissions
and fuel saving benefits. In addition, there is an increase in the switching on of
peaking units to deal with the unexpected shortfalls in generation when wind power
forecasts overestimate the actual production. This has a knock-on effect on cycling
costs. Each of these factors combine to increase the costs of wind generation and
reduce the benefits, thereby reducing the overall net benefits of wind generation,
and shifting the net benefits curve downwards. As the generation plant mix in
2020 is more flexible, the cost of poor wind forecast accuracy is lower. On the
other side, assuming the best case wind power forecasts, significantly improves the
critical values of wind generation by requiring less reserve capacity, more efficient
conventional unit operation and lower cycling costs.
6.2.3 Demand Growth Scenario
Chapter 3 discussed the assumptions regarding the assumed demand growth for the
years 2010, 2015 and 2020. A ‘mid’ level demand growth of 3.7% from EirGrid
(2005b) was assumed in the base case. Here a ‘low’ demand growth of 2.5% and a
‘high’ demand growth of 4.3% are investigated and are given in Figure 6.2 as ‘LD’
and ‘HD’ respectively.
This scenario shows an interesting result whereby high demand growth is pre-
ferred to low demand growth. With low demand growth, wind generation represents
a higher proportion of the dispatched plant. This results in the marginal plants be-
ing higher up the merit order, and in Ireland, this means older, more inefficient
coal fired units. These units are then required to cycle more frequently resulting in
large cycling costs. On the other hand, if the demand growth was high, the wind
generation represents a smaller proportion of the dispatched plant and the marginal
units are more likely to be gas fired units which are more flexible and have lower
cycling costs.
6.2.4 Fuel Price Scenario
Chapter 3 set out the fuel price assumptions used in this analysis. A low, mid and
high range of fuel prices was provided for each of the test years and the mid prices
were assumed in the base case. The impact on the critical values of wind generation
of assuming the other fuel prices are shown in Figure 6.2 and are given by ‘LFP’,
‘MFP’ and ‘HFP’ representing low, mid and high fuel prices respectively.
It is seen that high fuel prices are beneficial for wind generation since the value
of the savings in the fuel bill from wind generation are increased. This is partic-
ularly evident under the high fuel prices option in 2020 where the gas prices are
particularly high and the critical point is 24.5%. The impact of fuel prices on the
net benefits of wind generation is relatively intuitive. Not illustrated in Figure 6.2
is an alternative scenario which assumes high fuel prices as well as high capital costs
for wind generation. This could be due to an increase in manufacturing costs for
wind turbines associated with the high fuel prices. It is found assuming the high
capital cost scenario outweighed the benefits of the increased fuel savings and the
critical values were slightly lower than in the base case.
6.2.5 Capital Cost Scenario
Chapter 4 discussed four scenarios relating to the capital costs of wind generation.
The ‘High Capital’ scenario showed capital costs starting at e1m/MW and increas-
ing to e1.075m/MW at 4000MW installed. The ‘Constant Capital’ scenario held
the capital cost constant at e1m/MW. The ‘Mid Capital’ case showed costs initially
decreasing and then increasing again beyond 2750MW installed. The ‘Low Capi-
tal’ scenario showed capital costs decreasing to e0.855m/MW at 4000MW. These
scenarios are represented in Figure 6.2 by ‘LCP’, ‘CCP’ and ‘HCP’ indicating low,
constant and high capital costs respectively. The ‘mid’ scenario was used in the base
case assumptions, and is thus incorporated in the ‘BC’ results.
From Figure 6.2, it is clear that higher capital costs lead to a lower critical
value. In addition, it is interesting to note that although wind generation is highly
capital intensive, assuming a constant capital cost still maintains the shape of the
net benefits curve and the critical value is less than the base case. A number of
fairly intuitive additional scenarios with respect to the cost of wind generation were
also run but are not illustrated in Figure 6.2. The first is an alteration in the
operating and maintenance costs of wind generation. The base case assumed O&M
costs of e35,000/MW per annum. Increasing these costs to e50,000/MW, which is
the average cost for an offshore wind turbine reduces the critical values to 20% in
2010 and 21% and 21.2% in 2015 and 2020 respectively. The second investigates the
impact of changing the interest rate faced by the developer of the wind turbine from
8% to 6%. This increases the critical value of wind by approximately 1% across each
of the years. Decreasing the term of the loan faced by the developer from 20 to 15
years has a slightly larger impact on the critical values which decrease by between
1.5% and 2%.
6.2.6 Cycling Cost Scenario
When the cycling costs were considered in Chapter 4, they were calculated based on
the fuel cost of each unit, their age, previous operating profiles etc. The fuel cost was
assumed to represent between 2 and 30% of the true cost of cycling depending on
the unit. Here the sensitivity of the net benefits of wind to a reduction in these costs
is investigated. A more flexible plant mix would have a lower cost of cycling, thus,
two scenarios relating to lower cycling costs were investigated. The first assumed
that the fuel costs represented 50% of the true cycling costs for all units, and the
second that the fuel costs represented 80% of the true costs illustrated by ‘50Cy’
and ‘80Cy’ respectively in Figure 6.3.
As would be expected, it is seen that as the cost of cycling the generating units
reduces, the net benefits of wind generation increase. Wind generation causes an
increase in the cycling of units on the margin, thus with more flexible units on the
margin, with lower cycling costs, the cost imposed by wind generation is reduced.
This is also seen by comparing the critical values across the three test years, with
2020 having the most flexible plant mix.
6.2.7 Emissions Price Scenario
Chapter 5 discussed in detail the creation of emissions in combustion units and

calculated the emissions savings with increases in wind capacity. The value of these
saved emissions is based on the assumed emissions prices and four emissions price
scenarios were set out in Chapter 5: a ‘low’ emissions price; a ‘mid’ price (which was
used in the base case); a ‘high’ price; and a ‘highest’ price. Figure 6.3 represents
these scenarios as ‘LEp’, ‘HEp’ and ‘HHEp’ for the low, high and highest cases.
It was found when calculating the results that given the magnitude of the CO2
emissions, the emissions benefits were mainly dependent on the price of CO2 rather
than the other emissions. At the ‘low’ emissions price, the merit order of the con-
ventional units is virtually unchanged by the inclusion of the emissions price in the
marginal costs of generators and the wind generation displaces some oil units and
many gas units on the margin. The gas units have the lowest CO2 emissions per
MWh of all the conventional units on the Irish system. This explains the dramatic
reduction in emissions savings at the low emissions price. As the emissions price
increases to the ‘mid’ scenario (the base case), the coal units are shifted down the
merit order where they operate on the margin more often, resulting in much greater
emissions savings. However, this shift of coal units to the margin also increases
the cycling costs, thus as the emissions price increase further, the critical value is
restrained at a level not much greater than the base case. This interaction between
emissions prices, wind generation and cycling costs is an interesting issue which is
developed further in Chapter 7.
6.2.8 Discount Rate Scenario
When the net benefits of wind generation are being calculated, each of the costs and
benefits are assumed to accrue annually for a certain number of years and these are
then discounted by the assumed interest rate. In the base case, this discount rate
was 8%. Figure 6.3 illustrates the impact of reducing the discount rate for all costs
and benefits to 6% and increasing them to 10%. As would be expected, a lower
discount rate increases the net benefits of wind generation and a higher discount
rate reduces them.
6.2.9 Worst and Best Case Scenarios
The previous sections illustrated the sensitivity of the net benefits of wind generation
to changes in the underlying assumptions. These results are now brought together
to determine the two extremes for wind generation. The ‘worst case scenario’ for
wind generation embodies all of the assumptions that brought about the lowest net
benefits for wind generation, and the ‘best case scenario’ captures all the assumptions
which resulted in the highest critical values. Table 6.2 summarises the assumptions
which have been altered to generate the two extreme scenarios. The critical values
of the worst case and best case scenarios for wind generation are shown in Figure
6.3 as ‘bad’ and ‘good’ respectively.
Table 6.2: Worst case and best case assumptions summarised
Assumption Worst Case Best Case

Load Factor ‘Low’ ‘High’
Wind Power Forecast Error Worst (std dev 14%) Best (std dev 7%)
Demand Growth ‘Low’ ‘High’
Fuel Price Growth ‘Low’ ‘High’
Wind Capital Costs ‘High Capital’ ‘Low Capital’
Cycling Cost As in Base Case 80%
Emissions Costs ‘Low’ ‘Highest’
Discount rate 10% 6%
Term 15 years 20 years
By assuming the worst case scenario for wind generation, the critical values of
wind generation are very low, representing 549MW, 629MW and 1275MW installed
in 2010, 2015 and 2020 respectively. These penetrations represent 5%, 6% and 9%
of electricity generated by wind in 2010, 2015 and 2020. If the best case is assumed,
the critical values are above 30% of electricity generated in each of the three test
years.
6.3 Conclusions of Net Benefits
This Chapter presented the net benefits associated with wind generation and the
optimal penetrations of wind generation were determined under a wide range of
scenarios. The key results are summarised here.
• The net benefits curves were presented for the base case and the optimal
penetrations were found where the net benefits equal zero. This was found
to be 2790MW in 2010, 2994MW in 2015 and 3705MW in 2020, representing
21%, 21.5% and 22.2% of electricity generated respectively.
• The underlying plant mix is also very important and assuming a flexible
plant mix, as in 2020, significantly increases the net benefits of wind generation.
• The optimal points were tested under a large range of scenarios and it was
found that they were highly sensitive to the assumed load factor. A high
load factor results in much increased net benefits due to greater emissions and
fuel savings with the same capital costs as under low load factor scenarios.
• The capital cost of wind generation has an impact on the net benefits and
low capital costs increase the net benefits of wind generation.
• Accurate wind power forecasting increases the net benefits of wind gener-
ation by reducing the reserve and cycling costs.
• A low demand growth reduces the net benefits of wind generation signifi-
cantly.
• High fuel prices are beneficial for wind generation, increasing the value of
the saved fuel consumption with increasing penetrations of wind generation.
• A more flexible plant mix would have lower cycling costs thus two scenarios
with lower cycling costs were investigated. It was found that the net benefits
of wind generation were increased when the cycling costs were reduced.
• A reduced discount rate increased the net benefits of wind generation.

• Low emissions prices have a significant detrimental impact on the net ben-
efits of wind generation for two reasons. Firstly the saved emissions are worth
less at lower emissions prices. And secondly, with a low emissions price, the
wind generation displaces mainly gas fired units which have relatively lower
emissions per MWh. As the emissions price increases, the merit order shifts
such that heavier polluters are displaced by the wind generation. Thus a higher
emissions price results in more emissions saved and these emissions are valued
at a higher price.
• A worst case scenario was developed which incorporated all of the assump-
tions which would be detrimental to wind generation. Under this assumption,
it was found that just 5%, 6%, and 9% of wind generation should be developed
in 2010, 2015 and 2020 respectively.
• A best case scenario was also developed embodying all of the assumptions
which increased the net benefits of wind generation. It was found that in each
of the three test years, the optimal penetration of wind generation exceeded
30% of electricity generated in this scenario. These worst case and best case
scenarios can be thought of as the extremes for the potential wind generation
deployment in Ireland.
This chapter illustrated the net benefits results for wind generation for the case
study. The costs and benefits were discussed in Chapters 4 and 5 and were based on
the output of the dispatch model. A number of further applications of the dispatch
model and the methodology employed in calculating the net benefits are developed
in the following Chapter 7.
CHAPTER 7
Applications of the Net Benefits Methodology
T
HIS chapter discusses the results of some interesting applications of the dis-
patch model presented in Chapter 3 and related to the issues of renewable
generation and system impacts. As discussed in Chapter 6 emissions prices alter
the merit order in which generators are dispatched, with heavy polluters becoming
more expensive relative to light polluters. This can result in units which were pre-
viously baseloaded, being required to operate on the margin more often which can
have a knock on impact on cycling costs. The impact of emissions prices on cycling
costs is investigated in detail in Section 7.1. While the previous chapters investi-
gated many of the costs and benefits of wind generation, this application on carbon
prices isolates just the cycling costs and the emissions benefits and investigates the
relationship between the two.
In the work presented so far, the power system was operated by incorporat-
ing wind generation forecasts into the dispatch model. However, in a study by
Gardner et al. (2003) for the same case system, an alternative approach to power
system operation is assumed. Here, the units are committed regardless of the wind
power forecasts and then at real time, if the wind generation is available, these con-
103
Chapter 7. Applications of the Net Benefits Methodology 104
ventional units are deloaded to accommodate the wind generation. This method is
known as the fuelsaver approach and is discussed in detail in Section 7.2.
Section 7.3 discusses the impact that different connection policies have on the
development of renewable generation on the distribution network. The resulting
renewable energy penetration are then analysed in terms of their net benefits to
determine the connection policy which yields the maximum net benefits. This work
was conducted in collaboration with Keane and O’Malley (2005b).
Wind generation is variable and relatively unpredictable, whereas, tidal genera-
tion on the other hand is variable and almost perfectly predictable with in the time
frames of interest to power system operators. Thus, Section 7.4 discusses the fea-
sible tidal generation penetrations on the Irish system. The capital costs necessary
to ensure positive net benefits for tidal generation are also shown. Bryans (2006)
investigated the tidal generation resources in Ireland and his findings feed into the
results shown in this Section.
7.1 Carbon Prices and Cycling Costs
Due to international concern over climate change, policy makers worldwide are in-
troducing numerous instruments to help curb global emissions, such as carbon taxes,
emissions trading, allocation credits, emissions caps etc. These instruments cover
a wide range of industries including electricity generation, and are intended to in-
ternalise the cost of emissions for polluters with the aim of altering their behaviour
patterns to reduce their greenhouse gas emissions. In electricity markets, genera-
tors generally consider their operating costs to be made up of their internal costs.
However, these policy instruments have made the additional emission costs appli-
cable to generators based on the amount of environmental pollution they cause
(Baumol and Oates, 1975).
In January 2005, the European Union Greenhouse Gas Emission Trading Scheme
(EU ETS) commenced operation as the largest multi-country, multi-sector green-
house gas emission trading scheme worldwide. Phase I of the EU ETS covers the
period 2005-2007 and Phase II will run from 2008-2012 to coincide with the first
Kyoto Protocol commitment period. Under the EU ETS, each generator must sub-
mit a number of allowances equal to their total emissions for the preceding calender
year. These allowances for the year are subsequently cancelled. In Phase I any
installation with annual emissions exceeding their submitted number of allowances,
will be penalised e40 for each additional ton of carbon dioxide1 . In addition to the
payment of this excess emissions penalty, the installation is also required to submit
allowances for those excess emissions when submitting allowances in relation to the
following calendar year (EU ETS, 2003).
An emissions trading scheme requires electricity generators to internalise their
emissions costs. This in turn alters the marginal costs of generators such that
heavy polluters become relatively more expensive compared to lighter polluters
(Borchiellini et al., 2000). As shown by Doherty et al. (2005), placing a cost on
carbon emissions results in a shift towards fuels with a lower carbon content, and
in particular a shift from coal fired production to gas. This result is supported
by Klaassen and Riahi (2006) who show that the use of coal for electricity pro-
duction declines steadily into the future with the introduction of a carbon tax.
Voorspools and D’haeseleer (2006) studied the impact of a carbon tax on a num-
ber of EU countries and their results illustrate the significant impact that a carbon
tax has on the operation of the coal plants, which shift from being baseloaded or
mid merit to operating on the margin much more frequently with a carbon tax
of just e10 per ton. Putting this value in context, the price of carbon in Phase
I of the EU ETS has averaged at e18.52/ton with a peak price of e32.25/ton
(European Climate Exchange, 2006; Point Carbon, 2006).
Palmer et al. (2006) conducted a cost benefit analysis of emissions reduction in
the electricity sector, encompassing the electricity generation costs and the social
benefits. This study found that the benefits of emissions reduction far outweigh the
costs. However, this analysis did not account for the cost associated with the change
in operating patterns of units which had been previously baseloaded and were now
operating on the margin.
1
This will rise to e100 from January 2008.
As discussed in Chapter 4, generating units are usually designed to be at their

most efficient when turned on and running at a stable output (Le et al., 1990). If
a unit is operating on the margin, it will have to cycle more frequently to meet the
varying load. The cost of cycling significantly increases if a unit is operated at a
continuous output over a long period of time and is then required to cycle. Thus, if
a carbon price is introduced which causes a coal fired unit to become relatively more
expensive and shift from operating continuously as a baseload plant to operating on
the margin, the resulting cycling costs could increase significantly and could dampen
the benefits associated with its reduced emissions.
The introduction of a carbon price can act as an incentive for wind generation
which becomes relatively cheaper compared to the conventional carbon emitting
units. Under the EU Directive 2001/77/EC (2001), the output from a wind gen-
erator must be accepted when it is available. As a result, wind generation can
displace the output of a conventional unit and thereby reduce carbon dioxide (CO2 )
emissions, as shown in Chapter 5. However, as discussed in Chapter 4, wind gen-
eration also increases the cycling costs of the marginal units. Given that a carbon
price could change the type of unit operating at the margin, the effect of increasing
penetrations of wind generation could further exacerbate the cycling costs resulting
from a carbon price. Thus, in this Section, three aspects of the issues surrounding
cycling costs and carbon emissions are investigated. Firstly the impact of carbon
prices on cycling costs and carbon emissions is determined. Secondly the impact
of wind generation on cycling costs and carbon emissions is shown, and thirdly the
combined impact of a carbon price plus increased wind generation is investigated. In
an attempt to reduce the number of assumptions which could impact on the results,
a number of changes were made to the assumptions used in net benefits study.
• The assumed plant mix and load for this analysis are for 2007. The charac-
teristics of the system in 2007 are known with much more certainty than for
2010 and beyond and are discussed in detail in EirGrid (2005b). A summary
of the assumed plant mix for 2007 is given in Appendix A.
• The dispatch model described in Chapter 3 was used to determine the dis-
patches with increasing emissions prices and wind generation. However, in an

attempt to reduce the number of factors which may influence the results, the
model was simplified to dispatch for energy only i.e. the reserve constraint was
omitted. Including the reserve constraints introduces technical complications
without changing the fundamental insights of the results of this analysis. In
fact, adding reserves increases the cycling costs, thus the magnitude of the
cycling costs may even be underestimated in this Section.
• In the net benefits analysis, the peat fired generation was assumed to bid into
the dispatch model like all other units. However, for 2007, it is expected that
the Government will continue to impose its must-run policy towards peat fired
generation, thus, in this Section, the peat fired generation bids a low price into
the dispatch model to ensure baseloaded operation.
Section 7.1.1 summarises how the emissions prices were included in the bid prices
of generators and the impact that this has on the merit order. Section 7.1.2 illus-
trates the results showing the cycling costs and emissions savings firstly with the
introduction of a carbon price, secondly with an increase in wind generation and
thirdly with a combination of a carbon price and wind generation.
7.1.1 Emissions Prices in the Cycling Analysis
As discussed in Chapter 5, to include the carbon price in the bid price of generators,
their CO2 emissions per MWh were multiplied by the carbon price and this was
added to the fuel price per MWh to give their bid price. When the generating units
internalise their carbon costs it significantly altered the merit order of the dispatch.
The merit order in the absence of carbon costs was illustrated graphically in Figure
2.6 and is shown in Appendix A. Figure 7.1 here, illustrates what would happen to
the merit order if the price of carbon was e30 per ton of CO2 , and the units were
dispatched based on their marginal cost plus their CO2 emissions.
The baseloaded plant is now almost entirely made up of gas fired units and the
coal units have been pushed up to the margin. As discussed in Section 4.4, units
7500
6000 Distillate
Gas
Demand MW
4500 Oil
Gas
3000 Coal
2450
1500 Gas
Hydro & Peat

0 2000 4000 6000 8000
Hours
Figure 7.1: Merit order with carbon price of e30/ton
that were previously operated at baseload and are then required to cycle are most
prone to creep-fatigue interaction and will have very high cycling costs. If the coal
units are operating on the margin their operating profiles will change dramatically
which is likely to result in a significant increase in cycling costs.
7.1.2 Results & Discussion
The dispatch model was run for each hour for an entire year for a large range of
scenarios and the associated cycling and carbon dioxide emissions were calculated.
Figure 7.2 illustrates the annual additional cycling cost in e millions on the primary
y-axis, and the saved CO2 emissions in millions of tons on the secondary axis (shown
by the dotted line). The value of the CO2 emissions savings are calculated by
multiplying the emissions savings (in millions of tons) by the carbon price at each
point (illustrated by the dashed line).
250 10
Added Cycling Cost

Value of Saved CO2
200 8
Million Tons CO2 saved

150 6
e million
100 4
50 2
Saved CO2
0 0
0 10 20 30 40 50 60 70
CO2 price in e
Figure 7.2: Cycling costs and CO2 savings at different carbon prices
Figure 7.2 illustrates an interesting result, that at a carbon price of less than
e50 per ton, the cycling costs would far exceed the value of the saved CO2 . As can
be seen at a carbon price of e20/ton, the added cycling costs amount to e140m
while the amount of saved carbon is just 1.8 million tons, which at a value of e20
per ton amounts to a benefit of e36m. It is also clear that between a carbon price
of e10 and e20 per ton, the cycling costs escalate rapidly. This is because above
e10/ton, the marginal cost of coal exceeds that of gas. This results is supported
by Voorspools and D’haeseleer (2006). Increasing the price of carbon above e20
per ton does not have such a dramatic impact on the merit order and coal and oil
continue to operate on the margin. However, coal does get relatively more expensive
compared to the oil units and as a result gets utilised less and less as the carbon
price increases. This explains the stabilising of the cycling costs beyond e20 per
ton, and the slight reduction beyond e40 per ton.
In reality it is likely that given their large cycling costs, these coal units may bid
into the market in such a way as to minimise their cycling activities. They may alter
their bidding strategies in order to ensure they are baseloaded or to ensure they are
adequately compensated if they are required to operate on the margin. Alterations
in bidding strategies such as these may lead to the stranding of assets in the long
run which can cause significant security of supply issues. In addition, without a
more complex model it is unclear if the electricity system would in fact be able to
physically operate with such a significant alteration in the merit order. These issues
of strategic bidding behaviour and feasible operation are beyond the scope of this
study but raise interesting issues which could be covered in future research.
The introduction of a carbon price may result in an increase in wind generation
penetrations. As shown in Chapter 5, wind generation can aid in the reduction of
CO2 emissions by displacing generation on the margin. However, wind generation
can also causes an increase in the cycling of generating units on the margin (EirGrid,
2004). Figure 7.3 illustrates CO2 savings with increasing penetrations of wind as
well as the cycling costs in 2007. In order to clearly see the impact of wind gener-
ation alone on the cycling costs and emissions savings, Figure 7.3 assumes that the
conventional generators on the system have not internalised any of their CO2 costs.
For ease of comparison with Figure 7.2, Figure 7.3 values the carbon dioxide savings
at a price of e30/ton.
Figure 7.3 shows that although wind generation does cause an increase in cycling
costs, at a carbon price of e30/ton the carbon emissions savings exceed the cycling
costs at all penetrations. Thus, even in the absence of generators internalising their
carbon costs, wind generation results in carbon dioxide emissions savings by dis-
placing generation on the margin. By comparing Figures 7.2 and 7.3, it is apparent
that both a carbon price and wind generation can reduce CO2 emissions, however,
cycling costs are much higher when a carbon price is used to reduce emissions than
when wind generation is employed2 .
As discussed in Section 7.1.1, the introduction of a price on carbon results in a
2
As stated previously, this Section isolates the issues of cycling costs and emissions prices. These
results were embedded in the cost benefit analysis conducted in previous Chapters but are investi-
gated in isolation in this Chapter.
100 10
80 Added Cycling Cost 8

Value of Saved CO2
Million Tons CO2 saved

60 6
e million
40 4
20 2
Saved CO2
0 0
0 500 1000 1500 2000 2500 3000
Figure 7.3: Cycling costs and CO2 savings with wind and e30/ton CO2
shift in the merit order, with the baseloaded units operating on the margin more
often. Since wind generation causes an increase in the cycling of the marginal units,
a carbon price combined with increased wind generation could further exacerbate
the cycling costs. Figure 7.4 illustrates the cycling costs when generators internalise
their carbon costs combined with an increase in wind generation. It can be seen
that both wind generation and a price on carbon increase cycling costs and the
combination of both result in even higher cycling costs. This is due to the carbon
price resulting in previously baseloaded coal units with high cycling costs operating
on the margin, combined with the wind generation increasing the cycling of these
units. As shown in Voorspools and D’haeseleer (2006), carbon prices pushed coal to
the margin and Holttinen (2004) showed that wind generation increases the cycling
of marginal units. It is clear from Figures 7.4 that the main driver of the cycling
costs is the switching of the coal units from baseloaded to variable operation.
Figure 7.5 shows how wind generation and carbon prices interact to result in
200
180
160
Cycling Cost (em)
140
120
100
3000MW Wind
80 2000MW Wind
1000MW Wind
60 0MW Wind
40
20
0
0 10 20 30 40 50 60 70
Carbon Price (e)
Figure 7.4: Cycling costs with wind generation and carbon prices
CO2 savings. With a carbon price of zero, 3000MW of wind generation results in
the saving of almost 2.7 million tons of CO2 . With zero installed wind generation,
a carbon price of e70/ton generates a saving of 3 million tons. However, when
wind generation and a carbon price are combined, the savings amount to more than
simply the sum of the two parts. A carbon price of e70/ton combined with 3000MW
of wind generation results in a CO2 emissions saving of almost 7.5 million tons. This
is due to the wind generation displacing the marginal units which, because of the
carbon price, are the heavy polluting coal units.
In order to illustrate the combined results of the cycling costs and the carbon
dioxide emissions savings, the cycling costs were subtracted from the value of the
saved emissions to give a ‘net saving’. These net savings are illustrated in Figure 7.6.
The CO2 emissions savings are valued at the price of carbon. Thus with 1000MW
of wind generation and a carbon price of zero, the net savings are made up of just
the cycling costs from wind generation as the emissions savings equal zero.
6
CO2 Savings (Mton)
2
3000MW Wind
2000MW Wind
1 1000MW Wind
0MW Wind
0
0 10 20 30 40 50 60 70
Carbon Price (e)
Figure 7.5: Carbon savings with wind generation and carbon prices
350
300 3000MW Wind

2000MW Wind
250 1000MW Wind
0MW Wind
200
Net Saving (em)
150
100
50
−50
−100
−150
0 10 20 30 40 50 60 70
Carbon Price (e)
Figure 7.6: Net savings with wind generation and carbon prices
The point where the net savings are at a minimum occurs at a carbon price
of e20/ton and installed wind generation of zero. At this point, the cycling costs
exceed the CO2 emissions savings by over e100m. As wind generation is added to
the system, the CO2 emissions savings it produces are greater than the cycling costs
it adds so the deficit in net savings reduces. However, even with 3000MW of wind
installed, net savings are negative at e20/ton. At the peak, with a carbon price of
e70/ton and wind generation of 3000MW, the value of the savings in carbon dioxide
exceed the cycling costs by almost e340m for the year. This represents CO2 savings
of over 2.8 times the cycling costs. At this point, the coal and oil units are on the
margin, and although this results in high cycling costs (Figure 7.4) it also means
that the wind is displacing the highest emitting units on the system resulting in
much greater CO2 savings.
7.1.3 Summary of Carbon Prices and Cycling Costs
It was shown here that carbon dioxide reduction mechanisms can have a dramatic
impact on system cycling costs and may end up costing more to the system than
the value of the saved CO2 emissions. This situation is particularly severe in Ire-
land where there is a large proportion of the plant mix made up of large relatively
inflexible units with very large cycling costs. By increasing the flexibility of the
plant mix, the cycling costs could reduce significantly. For example, an increase in
open cycle gas turbines, responsive demand and hydro generation, could reduce the
system cycling costs by carrying some of the variability on the margin. It is likely
that an increase in interconnection could help reduce the need to cycle the marginal
units by helping to smooth out the variations on the margin.
Carbon mechanisms result in a shift in the merit order with high carbon emitters
becoming relatively more expensive compared to low emitters. This can result in
units which were previously baseloaded, operating on the margin much more fre-
quently. This can have a significantly large impact on cycling costs. In this Section
the impact that carbon prices can have on the cycling costs of generators and on
their carbon dioxide emissions savings was investigated. It was found that the car-
bon dioxide reduction benefits can be completely outweighed by the added cycling
costs. For the case study it was found that the cycling costs exceeded the carbon
savings when carbon prices were less than e50/ton. In addition, it was shown that
the most significant contributor to the cycling costs is the switching of coal units to
the margin. The coal units in the case system become more expensive than gas at
a carbon price of between e10 and e20/ton.
A carbon mechanism can promote increased penetrations of wind generation and
thus the impacts of wind generation on the system were also investigated. Wind
generation is variable and relatively unpredictable which causes an increases in the
cycling of generating units on the margin. However, the wind generation also dis-
places production by the marginal units and thus results in carbon dioxide emissions
savings. It was found that the carbon dioxide emissions savings produced by wind
generation outweigh the cycling costs it imposes.
The impact on cycling costs and carbon dioxide emissions when a carbon price is
combined with increased wind generation was then analysed. A carbon price shifts
high emitting baseloaded units to the margin and the wind generation increases the
cycling of the marginal units. Thus when a carbon price and wind generation are
combined, it was found that the cycling costs are increased further as the marginal
units have higher cycling costs. However, the combination of a carbon price and
wind generation have a much greater impact on carbon dioxide reductions. With a
carbon price the marginal units are the high carbon emitting units, thus when wind
generation is introduced it displaces some of the output of the heaviest polluting
units on the system. This results in carbon dioxide savings much greater than
the sum of the savings from the carbon price and the wind generation separately.
However, it was found that even with 3000MW of installed wind generation, the
cycling costs can still exceed the value of the saved carbon dioxide emissions.
7.2 The Fuelsaver Approach to Wind Generation
For the cost benefit analysis discussed in previous chapters, the dispatch model was
run with long range wind power forecasts in the first model run and updated wind
power forecasts in the second model run (Chapter 3). In other words, commitment
decisions of conventional units were made based on long range wind forecasts and
operating levels were decided based on an updated wind power forecast. Operating
a power system in this manner is highly efficient with regard to wind generation,
however, it may require an alteration in the mentality of the power system operator
who traditionally dealt with units which were almost entirely dispatchable.
A simpler approach to operating a power system with wind generation, known
as the fuelsaver approach, has been presented in Gardiner et al. (2003) and DETI
(2003). Under this approach, wind generation is not considered in the scheduling
of the plants and the unit commitment decisions are made ignoring any installed
wind capacity. Once the commitment decision has been made, the wind generation
is considered. If wind generation is available it is used and marginal conventional
plants which were dispatched are deloaded to accommodate the wind generation.
A conventional plant could be deloaded as far as its minimum but no plants are
switched off. If wind production reaches a level such that no more conventional
generation can be deloaded, then any further wind production is curtailed. This
operational strategy considers that the only benefit of wind generation is a fuel-
saving one and it assumes that wind generation has a capacity value of zero. This
is a simplistic approach and it allows issues of forecasting and reliability of wind
production to be ignored. However, this method of power system operation with
wind generation results in an over commitment of conventional units, and these
units will be running at much lower efficiencies than under the approach adopting
wind power forecasts. The impact of operating the power system using a fuelsaver
approach is investigated here and it is shown that it is a highly inefficient way to
operate the system with significant wind power penetration.
7.2.1 Analysing the Fuelsaver Approach
The impact of operating the power system under a fuelsaver approach is investigated
here and the net benefits of wind generation are determined for the year 2010. In
order to simulate the fuelsaver approach, the dispatch model described in Chapter
3 was altered slightly. In the commitment stage of the model, where conventional
generators’ output can range between 0 and maximum generation, wind power out-
put is assumed to equal zero. This implies that the commitment decisions are made
regardless of wind generation. In the second stage of the dispatch model, the wind
power is introduced and the operating levels of the committed conventional units
are determined. If the sum of the minimum stable generation of the committed
units plus the wind power output exceeds the load for a given hour, then the wind
generation is curtailed.
By analysing the dispatches resulting from fuelsaver model, it is seen that with
very low installed capacities of wind generation, the dispatches are quite similar to
those from the model using wind power forecasts. However, as the wind capacity
increases, the fuelsaver method becomes less and less efficient with much greater
wind power curtailment, more units turned on, and more units running at lower
efficiencies. Figure 7.7 illustrates the amount of wind curtailment for the year 2010
under the fuelsaver approach when compared to the original approach, referred to
from here as the ‘forecasted’ approach. The wind curtailment is shown in GWh
curtailed over the year.
As can be see from Figure 7.7, under the fuelsaver approach, wind generation is
curtailed from a lower installed capacity and at a much higher magnitude. Under
the fuelsaver approach, with 4000MW of installed wind generation, there would
be 3450GWh of curtailed wind, representing over 30% of the annual wind power
production (assuming a ‘mid’ load factor). In the forecasted approach curtailment
results when the long range wind power forecast grossly underestimates the actual
wind output. At 4000MW installed, the likely annual curtailment is approximately
11% of annual production. Thus, with the forecasted approach, significantly more
energy is reaped from the same installed capacity of wind generation operated under
a fuelsaver approach (2.2TWh over the year at 4000MW installed).
Figure 7.8 illustrates the average number of units dispatched to meet the load
under both the fuelsaver and forecasted approaches. Under the forecasted approach,
less and less units are dispatched to meet the load since wind generation is considered
3500
Fuelsaver
Forecasted
GWh of Wind Curtailment 3000
2500
2000
1500
1000
500
0
0 500 1000 1500 2000 2500 3000 3500 4000
Figure 7.7: Wind curtailment in 2010 under fuelsaver
22
Fuelsaver
Average Number of Committed Units
21 Forecasted
20
19
18
17
16
15
14
13
12
0 500 1000 1500 2000 2500 3000 3500 4000
Figure 7.8: Average number of committed units under fuelsaver

in the scheduling of the units. The reduction saturates at approximately 2000MW

installed as the reserve requirement with wind ensures that a certain number of units
must be committed. With the fuelsaver approach however, the units are scheduled
independently of the wind generation. Since the load is unchanged across each of the
wind penetrations in the fuelsaver scenario, the number of units committed across
each of the wind penetrations remains unchanged.
In addition to more units being on under the fuelsaver approach, those units are
also running at lower operating levels than under the forecasted approach. Figure
7.9 shows the average operating level, as a percentage of maximum capacity, under
the two approaches.
65
Average Operation as % of Maximum Capacity
Fuelsaver
Forecasted
60
55
50
45
40
0 500 1000 1500 2000 2500 3000 3500 4000
Figure 7.9: Average operating levels of committed units under fuelsaver
It is clear from Figure 7.9 that the operating levels of the conventional units are
dramatically reduced under the fuelsaver approach when compared to the forecasted
approach, this has a knock on effect on the efficiency of these units. Operating at
reduced levels results in increased fuel consumption and emissions per MWh.
The costs and benefits which depend on the schedule of generators are the cycling
costs and the emissions and fuel savings. Under the fuelsaver approach, the wind
power forecasts are disregarded and the system is dispatch as though there were
no wind generation. Thus, it is assumed in the fuelsaver approach that there is no
increased reserve requirement with wind generation as the wind does not add to the
uncertainty of the system. The fuelsaver approach alters the dispatch of generators
from the forecasted approach, thus, the costs and benefits which are discussed here
are the cycling costs, the emissions benefits and the fuel savings.
7.2.2 Fuelsaver Cycling Costs
Under the forecasted approach, the main cause of cycling costs was the ramping
of units on the margin to deal with changes in the hourly wind power output, in
addition to the switching on of some peaking units if the forecasted wind output
was greater than actual. Under the fuelsaver approach however, there are no cycling
costs attributable to on/off cycling as it is assumed under fuelsaver that no units
are switched on or off. As shown in Figure 7.7, the fuelsaver approach results in the
curtailment of a large proportion of the wind generation. This actually smooths out
the wind power production and thus, under the fuelsaver approach, cycling costs are
actually less than under the forecasted approach as illustrated in Figure 7.10.
However, although the fuelsaver approach results in lower cycling costs, the fact
that the units are operating at much lower efficiencies has a dramatic impact on the
emissions and fuel savings as shown here in Section 7.2.3 and 7.2.4
7.2.3 Fuelsaver Emissions Benefits
The dispatches from the fuelsaver approach were analysed with regard to the emis-
sions savings with increases in wind capacity. Since the conventional units operate
at much lower outputs under the fuelsaver approach, they are less efficient. In addi-
tion, as discussed in Chapter 5, NOX emissions increase significantly in CCGTs and
OCGTs if they are operated at reduced outputs. Figure 7.11 illustrates the absolute
emissions reduction with increases in wind generation under the fuelsaver approach.
120
Fuelsaver
Forecasted
100
80
Cycling Cost em
60
40
20
0
0 500 1000 1500 2000 2500 3000 3500 4000
Figure 7.10: Cycling costs under the fuelsaver approach
30000 200
25000
20000 Value of Emissions Savings em

Emissions in kilotons
CO2
SO2
NOx
30
25 100
20
15
’Forecast’ Savings
’Fuelsaver’ Savings
10
0 0
0 500 1000 1500 2000 2500 3000 3500 4000
Figure 7.11: Emissions savings under the fuelsaver approach

The emissions in kilotons are illustrated on the primary y-axis and it is seen that the
CO2 emissions greatly exceed the SO2 and NOX emissions. The secondary y-axis
illustrates the value of these saved emissions at the ‘mid’ emissions price scenario
(Table 5.3) in Chapter 5 and also shows the value of the saved emissions under the
forecasted approach for ease of comparison
It is clear from Figure 7.11 that the forecasted approach results in much improved
emissions savings than the fuelsaver approach. In addition, it is very interesting to
note that if a fuelsaver approach is adopted, wind generation can actually cause an
increase in NOX emissions over a case with no wind. This is due to the CCGT units
operating at low loads. Thus, although wind generation does not itself produce
harmful emissions, it can actually cause an increase in system NOX emissions if the
power system is operated under a fuelsaver approach.
7.2.4 Fuelsaver Fuel Benefits
As with the emissions savings, the fuelsaver approach was also analysed with respect
to the reductions it produced in fuel consumption. Figure 7.12 illustrates the re-
duction for each of the fuel types in petajoules over the year on the primary y-axis,
and the monetary savings assuming the ‘mid’ fuel prices on the secondary y-axis.
Again, the fuel savings under the forecasted approach are included for illustrative
purposes.
Since no units are switched off under the fuelsaver approach, no units are entirely
displaced by the wind generation. Thus the fuel savings stem from a reduction in
output of the committed generators. As the wind generation increases, more and
more units are required to operate at lower levels of operation. Thus, a saving in
each of the fuel types is seen, however this is not a linear reduction as the units are
less efficient at lower loads. It is seen from the secondary y-axis that the value of
the savings are significantly less than in the forecasted approach.
200 300
Gas
Coal
Peat
Value of Fuel Savings em

150
Oil
Fuel Consumed (PJ)
200
100
100
Forecasted Savings
50 Fuelsaver Savings
0 0
0 500 1000 1500 2000 2500 3000 3500 4000
Figure 7.12: Fuel savings under the fuelsaver approach
7.2.5 Net Benefits of Wind under the Fuelsaver Approach
The cycling costs and the emissions and fuel saving benefits of wind generation under
a fuelsaver approach have been calculated in the previous Sections. It is assumed
that the development, operation and maintenance costs of wind generation, remain
unchanged regardless of the method of system operation with wind generation as do
the network costs since these costs depend on the level of installed wind generation.
It is considered that operating the system under a fuelsaver approach requires no
additional reserve capacity. The fuelsaver approach assumes that wind generation
has a capacity value of zero thus the only benefits of wind generation under the fuel-
saver methodology are the emissions and the fuel savings. Figure 7.13 illustrates the
total cost function, the total benefit function and the net benefits of wind generation
operated under a fuelsaver approach.
Figure 7.13 shows that under the fuelsaver approach the costs exceed the benefits
at all levels of installed wind generation. Thus, at all levels of wind generation, wind
6000
Costs
Benefits
4000 Net Benefits
2000
em
−2000
−4000
−6000
0 500 1000 1500 2000 2500 3000 3500 4000
Figure 7.13: The net benefits of wind under the fuelsaver approach
power forecasts should be included in the dispatch decisions of generators as if they

are omitted, the costs will exceed the benefits at all levels.
7.2.6 Conclusions of Fuelsaver Approach
This Section analysed an alternative power system operation approach with wind
generation, known as the fuelsaver approach, which was employed in Gardner et al.
(2003) and DETI (2003). The fuelsaver approach ignores wind power forecasts in the
scheduling of generation and wind power is only considered if it is available at real
time. The fuelsaver approach results in large amounts of wind energy curtailment,
of up to 30% of the annual output at high levels of installed wind power. In addition,
it results in the over commitment of conventional generation and those committed
generators are required to operate at low operating levels. This has a knock on
impact on the emissions and fuel savings of wind generation. In fact, under the
fuelsaver approach, NOx emissions can actually be greater than in the absence of any
wind generation. By adopting the fuelsaver approach the costs of wind generation
exceed the benefits at all levels of installed wind generation, which implies that to
employ this method of power system operation would be entirely futile.
7.3 The Impact of Connection Policy on Distributed

Generation
Much of the wind generation development to date has been in low capital, small
scale developments connected to local distribution networks (Collinson et al., 1999).
These local distribution networks are low voltage networks which were traditionally
designed for the delivery of electricity to end use customers rather than the gener-
ation of electricity. However, with the connection of small scale generators, such as
wind turbines, the characteristics of the distribution network are changing leading
to changes in the operation of these networks (OFGEM, 2003).
This small-scale generation, which is not centrally dispatched and is connected
to local distribution networks is referred to as distributed generation (DG). Wind
generation is the fastest growing form of DG, with significant penetrations connected
already in many countries (Keane et al., 2006). Traditionally distribution system
operators have only allowed DG to connect to the distribution network on a ‘firm’
basis. This means that the generator can connect up to a capacity which will
ensure that the generator will never violate the technical constraints of the network,
even when operating at its limits. Thus the output from a firm generator will be
accepted onto the network at all times. The impact of a firm connection policy
on the potential for DG has been demonstrated in Keane and O’Malley (2005a) and
Wallace and Harrison (2003). Adopting a firm connection policy can lead to network
sterilisation, which limits the potential for further generation capacity connecting to
that section of network3 (Keane and O’Malley, 2005b; Vovos et al., 2005).
Non-firm access has been proposed as a method to increase the penetration
3
Network sterilisation results when capacity is allocated to the bus/buses whose voltage and/or
short circuit levels are most sensitive to power injections. Thus no more generation can be connected
as the buses are constrained.
of distributed generation (Collinson et al., 1999). Non firm access refers to output
greater than the firm amount, at which generators may be allowed operate depending
on the load and generation levels throughout the year. Active controllers are then
used to manage any constraint breaches, and if necessary to curtail some of the
output of the distributed generators. Keane and O’Malley (2006) have shown the
scope for DG if non firm access is employed.
Collaboration work with Keane and O’Malley (2006) form the basis for the re-
sults shown in this Section. Keane and O’Malley (2005b, 2006) determined the
optimal allocation of energy resources on the distribution network under three con-
nection policy scenarios. It is found that by employing non-firm connection policies,
significantly more renewable generation can connect to the distribution network
without the need for additional investment in the network, thus ensuring efficient
use of the existing infrastructure.
Using the methodologies set out in Chapters 3 to 6 a cost benefit analysis is
conducted on the optimal penetrations to determine the overall monetary benefit
of using one connection policy over another. It is shown that a significant increase
in the net benefits of distributed generation is gained if the appropriate connection
policy is utilised from the outset and conversely that significant costs are incurred
if ad hoc policies are employed. Furthermore, it is shown that non firm access has
the scope to facilitate a significant extra amount of distributed generation capacity.
The methodology employed to determine the optimal penetrations of generation
on the distribution network is outlined in Section 7.3.1. The distribution network
for the case study is discussed in Section 7.3.2 and the potential energy resources
are shown in 7.3.3. Section 7.3.4 illustrates the results of the optimisation of the
distribution network. The load flow analysis, system scaling and the dispatch model
are discussed in Sections 7.3.5 to 7.3.7. The costs and benefits of the optimal pene-
trations are discussed in Section 7.3.8 and the net benefits are given in Section 7.3.9.
The conclusions are given in Section 7.3.10.
7.3.1 Methodology
The optimisation of generation on the distribution network is discussed in detail in

Keane and O’Malley (2005b, 2006) and Keane et al. (2006). The objective of the
methodology is to maximise the amount of energy output per euro of investment
in distributed generation by making best use of the existing network assets and
available energy resources. This is done subject to the technical constraints on
the network. Appendix B summarises the methodology which was employed by
Keane and O’Malley (2005b) to optimise the allocation of distributed generation on
the distribution network. The resulting allocations are then used as the basis for
the cost benefit analysis which is the focus of the work presented here.
A number of representative samples of the distribution network were analysed.
These sections were chosen based on their geographical location and also on their
characteristics (i.e. rural or semi urban). Likely energy resource portfolios were asso-
ciated with each section. Using the methodology shown in Appendix B, each section
was analysed and the optimal allocation of the energy resources was determined for
3 cases:
• Firm
• Non Firm
• Firm + Non Firm
The first case is the base case of the maximum firm allocation optimised on a
MWh/e basis, i.e. the amount of energy per euro of connection costs is maximised.
The objective function and constraints utilised by Keane and O’Malley (2005b) for
the ‘firm’ allocation are shown in Appendix B. The second case refers to the max-
imum non firm access permitted. Here voltage sensitivities are utilised in the ob-
jective function to reduce the instances of overvoltage (see Appendix B). The third
case examined is the case where firm access is initially only permitted and then non
firm access is permitted at a later stage. These two non firm cases are optimised on
a MWh/ekV basis.
A summary of the implementation and stages of the methodology are shown in

Figure 7.14. Portion ‘A’ of the Figure highlights the ‘network’ part of the analysis.
Based on the technical constraints, the allocations under the firm and non firm
objectives are calculated. The firm allocation (MWh/e) is fed as a constraint into
the non firm allocation, which yields the ‘firm + non firm’ allocation. This process
of determining the optimal penetrations of distributed generation was conducted by
Keane and O’Malley (2005b) and is summarised in Appendix B.
Portion ‘B’ of Figure 7.14 translates the installed DG capacities into annual
energy outputs. Each of the three allocations are simulated on the distribution
network over a year using load flow analysis (Keane et al., 2006). This load flow
analysis determines the instances of DG output curtailment in the non firm cases.
The analysis for the network section is then scaled up to represent the system as a
whole.
Portion ‘C’ of the analysis inputs the annual energy profiles of the DG allocations
into the dispatch model to determine their impact on the operating levels of the other
conventional units on the system. The resulting dispatches are then used to calculate
the costs and benefits of each of the allocations.
7.3.2 Ireland’s Distribution Network
In Ireland, distributed generation is typically connected at 38kV or by direct feed

to the 38kV/MV station on the MV network (10kV or 20kV). Voltage rise tends to
be the dominant constraint due to the network in Ireland, which outside the main
cities is typically a weak rural network with a large amount of conductor. A weak
network means a network with a low short circuit level or fault level.
Five network sections are chosen as a representative sample of the types of net-
work encountered on the Irish distribution network where generation will seek to
connect. Figure 7.15 shows the likely future distribution of installed wind power per
county, reproduced from Doherty and O’Malley (2005). The five network sections
were selected from along the western coastline and also in the southeast of the coun-
try. It is evident from Figure 7.15 that it is in these locations that connections of
SAMPLE
NETWORK A
SECTIONS
TECHNICAL
CONSTRAINTS
FIRM ALLOCATION
(MWh/Euro)
NON FIRM
ALLOCATION
(MWh/Euro kV)
ANNUAL LOAD FLOW SIMULATION & CURTAILMENT

B
SCALING
ANNUAL ECONOMIC DISPATCH

(Chapter 3)
C
COST BENEFIT ANALYSIS

(Chapters 4 – 6)
CAPITAL, O&M & CONNECTION COSTS,

RESERVE & CYCLING COSTS,
CAPACITY BENEFIT,
EMISSIONS BENEFIT,
FUEL SAVING.
Figure 7.14: Methodology

DG, and in particular wind generation, will and are being sought.
Figure 7.15: Future distribution of installed wind power capacity in percent
The network sections were also selected based on their characteristics and struc-
ture from network data available from ESB Networks (2005). They represent the
types of network encountered on the Irish distribution network, from voltage con-
strained rural networks to more urban networks where voltage is less of a problem
and short circuit level can become significant.
7.3.3 The Energy Resources
A representative energy resource portfolio was drawn up for each network section
based on its geographical location and with reference to ESBI and ETSU (1997)
and Figure 7.15. Using typical values for the load factor of each type of gener-
ation (BWEA, 2006), these resources are incorporated into the objective function
(Equation (B.1) in Appendix B). Each network section has its own distinct assumed
energy resource. Table 7.1 shows the total assumed energy resource and load factors
across the five representative network sections.
Table 7.1: Assumed Energy Resource for the Five Network Sections (MW)
Section Capacity (MW)

Energy Source 1 2 3 4 5 Total Load Factor
Wind 20 31.5 39 23.5 24 138 0.35
Biomass 0 0 10 10 15 35 0.85
LFG 12 0 0 0 0 12 0.76
Hydro 2 2 1 0.8 0 5.8 0.32
Total 34 33.5 50 34.3 39 190.8 -
7.3.4 Optimal Allocations of Distributed Generation
The allocations for the five network sections were determined under the three con-
nection policies using the methodology described in Appendix B and the total allo-
cations are shown in Table 7.2.
Table 7.2: Allocations for the five network sections
Firm +
Firm Non Firm
Non Firm
Wind 48.6 94 94
Biomass 34.4 35 35
Installed MW
LFG 12 12 12
Hydro 5.0 5.8 5.8
Wind 153 291 289
Biomass 259 264 264
Energy GWh
LFG 80 80 80
Hydro 14 16.5 16.5
Given their significantly higher load factor, any available biomass and landfill
gas (LFG) are allocated the available firm capacity first. As a result, there is very
little difference, between their firm and non firm capacities, leading to the conclusion
that when a maximisation of energy strategy is followed, forms of generation with
higher load factors are largely independent of the connection policy as they will
always be connected first. The small scale hydro generation also has very little
difference between its firm and non firm allocation, but this is due to the limited
resource available. When these factors are combined with the high wind resource,
it is evident that it is the costs and benefits of wind generation that are affected
by the connection policy and it is wind generation therefore that is the focus of the
results in this Section (Keane et al., 2006).
7.3.5 Load Flow Simulation
Once the allocations of DG capacity on the network sections are determined, the an-
nual energy output for each DG resource is ascertained. For the firm case, the annual
energy output is determined using actual historical data series from ESB Networks
(2005). For the non firm cases, over the year, constraint breaches will occur at times
of high generation and low load. Annual simulations are carried out for the network
sections to determine the time and magnitude of energy curtailment. These simu-
lations consist of load flow calculations carried out for half hourly data. The data
included in the simulations includes active and reactive load and generation profiles
for each of the energy resources, along with data on the frequency of N-1 outages
and the sending voltage at the transmission station (Keane et al., 2006).
If there is a constraint breach in the non firm case, due to over voltage, the output
of the distributed generation is curtailed. The level of curtailment is calculated using
Equation (7.1).
Vi − VM ax
PCurtail i = i ∀ N. (7.1)
2νi
Where PCurtail i (MWh) is the amount of energy that is curtailed over a half
hour period at the ith bus. Vi and VM ax (kV) are the voltage at the ith bus and the
maximum permissable voltage respectively. νi is the voltage sensitivity (kV/MW)
of the ith bus and N is the number of distribution buses. νi is calculated for each
bus drawing on the analysis used in Keane and O’Malley (2005b). This curtailment
method results in the least amount of energy being curtailed, by curtailing the
generator who has the highest voltage sensitivity at that bus. The annual load flow
simulation yields the amount of energy curtailed over the year, which in turn gives
modified generation profiles, which can then be used in the dispatch model. Figure
7.16 shows the output profile for each of the four allocated DG types over a single
day in March on a single section of network. This day is illustrated as it is an
example of a day of particularly high wind generation when the output had to be
curtailed in the non-firm case.
100
90
80
70
% Generation
60
50
40
30
Biomass
20 LandFill Gas
Wind
Non Firm Wind
10
Hydro
0:00 4:00 8:00 12:00 16:00 20:00 0:00

Time
Figure 7.16: Output of DG generators on a sample day
7.3.6 Scaling Outputs
From Table 7.2 it is evident that given a large wind resource, non firm access has
the potential to facilitate much higher penetrations than firm access (94MW com-
pared to 48.6MW). The allocations shown in Table 7.2 are only for five network
sections, however, to understand fully the effect of the connection policy of DG,
it was necessary to scale the allocations up to represent the whole system. This
scaling incorporates the characteristics of the distribution network, and the likely
penetrations and locations of renewable energy in Ireland (ESBI and ETSU, 1997;
AIRGS, 2006b; Doherty et al., 2006; ESB Networks, 2005). The installed capacities
and the annual energy outputs for the whole system, based on actual output data
series (AIP, 2005) and the total curtailed energy, are given in Table 7.3.
Table 7.3: Scaled allocations for the whole system
Firm +
Firm Non Firm
Non Firm
Wind 655 1392 1392
Biomass 401 410 410
Installed MW
LFG 96 96 96
Hydro 64 77 77
Wind 2065 4333 4302
Biomass 3018 3089 3089
Energy GWh
LFG 639 639 639
Hydro 184 221 221
Table 7.3 illustrates the scaled installed capacities and also the annual energy
outputs for each technology. It is seen that for the same installed capacity, the non
firm scenario produces more energy output than the firm + non firm case. This is
because there is more freedom in the allocating of capacity on the network in the
non firm case resulting in reduced capacity allocation at the most voltage sensitive
buses. This results in less curtailment under the non firm case when compared to
the ‘firm + non firm’ case.
It should be noted that the capacities given here are representative of the dis-
tribution connected generation and not transmissions connected. Thus they are less
than the capacities investigated in previous Chapters.
7.3.7 Distributed Generation in the Dispatch Model
The annual output from the distributed generation is determined as described in

Section 7.3.6. In order to quantify the net benefits of distributed generation, it is
necessary to quantify the impact that it has on the operation of the power system
as a whole. Thus, the dispatch model described in Chapter 3 is used. In order
to include the output profiles of the DG, the dispatch model was altered slightly.
The assumed plant mix and load were for the year 2007 and the model was run at
half hour intervals for the entire year (rather than hourly intervals as in previous
Chapters). The test year 2007 was used in the dispatch model for consistency as
the network analysis was based on local load profiles which were provided by the
Irish distribution network operator for 2007 (ESB Networks, 2005). The DG profiles
were bid into the commitment stage of the dispatch model with a bid prices of zero,
ensuring they were always on when available. The resulting dispatches were then
used to calculate the cycling costs and emissions and fuel savings.
7.3.8 The Costs and Benefits of Distributed Generation
The total connection costs for wind under the three connection policies across the
whole system were calculated by Keane et al. (2006) and were assumed to be variable
per km of line and are taken to be 50,000e/km. The connection costs are shown
in Table 7.4. A considerable saving in connection costs is made when a non firm
connection policy is pursued from the outset. In some cases, the extra non firm
capacity can share the existing connection line, but it is seen from the results that
on other occasions new connection lines must be built.
Table 7.4: Connection costs for wind generation
Connection Policy Installed MW Connection Cost

Firm 655 e61,300,000
Non Firm 1392 e128,000,000
Firm + Non Firm 1392 e156,800,000
As discussed in Section 7.3.7, the DG output levels for each technology under the
three connection policy scenarios are bid into the dispatch model with a bid price
of zero, ensuring priority dispatch. The resulting dispatches are then analysed to
determine the costs and benefits of each of the three connection policy approaches.
The additional reserve requirement for the system with increased wind penetra-
tion was discussed in Chapter 4 and is valued at e1.15/MWh. The reserve costs
for the distributed wind generation allocations are given in Table 7.5. The installed
wind capacity is the same in both of the non firm cases thus the additional annual
reserve costs are the same.
Table 7.5: Additional reserve costs for distributed wind generation
Connection Policy Installed MW Additional Reserve Cost

Firm 655 e101,090
Non Firm 1392 e290,990
Firm + Non Firm 1392 e290,990
The additional annual cycling costs over an installed wind penetration of zero
were calculated as described in Chapter 4 and are given in Table 7.6.
Table 7.6: Additional cycling costs with wind generation
Connection Policy Additional Cycling Cost

Firm e52,396,400
Non Firm e66,437,200
Firm + Non Firm e64,858,700
The reason for the slight difference in the cycling costs for the two non firm cases
is the energy output. The non firm case has less curtailment than the ‘firm + non
firm’ case for the same installed capacity resulting in higher annual output. This
means that over the year the non firm capacity displaces more generation. However,
at times of low load this displaced generation can be units which have higher cycling
costs. This non firm case causes a slight increase in the cycling of these units over
the ‘firm + non firm’ case.
The dispatch model was also used to calculate the CO2 , SO2 and NOx emissions
for each of the conventional generators. Table 7.7 shows the emission savings across
the year compared to a no wind case. The emissions savings are valued at the ‘mid’
emissions prices.
Table 7.7: Emissions savings (in tons) with wind generation
Connection CO2 SO2 NOx Benefit e

Firm 1,086,116 2,685 1,147 36,830,000
Non Firm 2,189,700 3,659 1,965 72,136,000
Firm + Non Firm 2,181,300 3,659 1,959 71,463,000
The additional wind generation on the system also provides a saving in fuel
consumption as thermal units are displaced. Table 7.8 shows the fuel savings with
increased wind generation. The fuel prices given in Table 7.8 were also used in the
dispatch of the generators.
Table 7.8: Fuel savings in petajoules with wind generation
Fuel Type Gas Coal Peat Oil

Benefit e
Price e/GJ 4.16 2.15 3.14 4.13
Firm 9.97 0.05 0.98 4.75 64,273,000
Non Firm 19.99 0.22 2.25 7.63 122,240,000
Firm + Non Firm 19.87 0.22 2.25 7.63 121,740,000
7.3.9 Net Benefits
The costs and benefits above were combined and the net benefits of each connec-
tion policy were determined. The capital cost of wind generation is assumed to be
e900,000 per MW here. This is less than that assumed in Chapter 5 as the shallow
connection costs were included in the capital cost in Chapter 5. However, in this
analysis they are dealt with separately and are illustrated in Table 7.4. The capacity
benefit is taken to be 37% at 655MW and 31% at 1392MW, as discussed in Chapter
5. It was found that although the benefits outweighed the costs under all three
connection policies there was a significant difference in the net benefits. Table 7.9
shows the net benefits of each of the connection policies.
Table 7.9: Net benefits of wind generation (in em)
Cap. (MW) Total Cost Total Benefit Net Benefit

Firm 655 1,346 1,379 33
Non Firm 1392 2,461 2,684 223
Firm+NonFirm 1392 2,476 2,673 197
As is evident from Table 7.9, while the two non firm policies both result in the
same installed capacity of generation, the costs and benefits of wind generation show
significant differences. There is a significant increase of 12% in net benefits under
non firm connection than under a firm and then non firm scenario for the same level
of installed wind. The increases in net benefits results from the different pattern on
connected generation across the buses, which leads to altered connection costs and
curtailed energy, which in turn impacts upon a number of the costs and benefits as
detailed above.
It is clear that the connection policy affects the energy output and connection
costs and furthermore that this has a considerable knock on effect on system costs
and benefits such as cycling costs, emissions and fuel use. It has also been shown
that as some costs, such as the additional reserve cost and the capital cost, are
dependent on the capacity of the plant installed, they are unaffected by the non
firm connection policy employed.
7.3.10 Conclusion of Connection Policy
Much of the recent development in renewable generation has been in low capital,
small scale developments connected to local distribution networks. Distribution
networks were traditionally designed for the delivery of electricity to end users rather
than the generation of electricity, leading to changes in the characteristics of the
distribution network and changes in the operation of these networks. To ensure

a sustainable future electricity supply, renewable generation should be facilitated
in such a way as to maximise connection while making best use of the existing
distribution network infrastructure.
This Section compared the optimal allocation of energy resources on the distribu-
tion network under three connection policy scenarios. It is found that by employing
non-firm connection policies, significantly more renewable generation can connect to
the distribution network without the need for additional investment in the network,
thus ensuring efficient use of the existing infrastructure.
The optimal penetrations of distributed generation under a number of planning
policies are determined and these are then used in the dispatch model to determine
their impact on the system as a whole. A complete cost benefit analysis is then
conducted on these penetrations to determine the overall monetary benefit of using
one connection policy over another.
It is shown that a significant increase in the net benefits of distributed gener-
ation is gained if the appropriate connection policy is utilised from the outset and
conversely that significant costs are incurred if ad hoc policies are employed. Fur-
thermore, it is shown that non firm access has the scope to facilitate a significant
extra amount of distributed generation capacity.
7.4 Tidal Generation on the Irish System
Wind generation is variable and relatively unpredictable. As discussed in Chapter 3,

wind power forecasts can be incorporated into the scheduling of generators, however,
there still remains a degree of uncertainty and a need for real time balancing with
wind curtailment and the switching on of peaking units. Here tidal generation is
investigated as an alternative to wind generation as although it is variable, it is
almost perfectly forecastable within the time frames of interest to power system
operators.
The majority of the energy contained within the tides is generated from the
gravitational forces of the sun and moon on the deep oceans. The rotation of the
earth relative to both the sun and the moon produces a 12.4 hour cycle resulting in
two high waters and two low waters per day. The size of the high water is dependent
on the position of the moon relative to the sun. When they are in line the forces
are constructive and there is a spring tide. When they are at 90◦ the forces are
destructive and there is a smaller neap tide (Bryans et al., 2005).
Traditionally tidal energy has been harnessed using a barrage system to establish
a head of water, which can in turn power a turbine, much as in a hydroelectric
dam. An example of such a scheme can be seen at the La Rance tidal barrage,
Brittany, France. Recent developments in tidal energy devices (TEDs) have focused
on harnessing the kinetic energy present whilst the tide is flowing into the shelf
sea, rather than to harness the potential energy established by the rise in sea level
(Bryans et al., 2005; Bryans, 2006).
The progress of TED development has been slow, with only 15 projects in de-
velopment around the world. One tidal device is almost market ready, developed by
Marine Current Turbines, and two other devices are still in the development stage:-
the Engineering Business’s ‘Stingray’ project and the Hammerfest Strøm project.
The design developed by Marine Current Turbines (MCT) is illustrated in Figure
7.17. It utilises two turbines on a wing either side of a support beam driven into the
seabed. The wing holding the turbines can be jacked up the beam enabling the tur-
bines to be raised out of the water, removed and serviced on land. The MCT device
has been designed to take advantage of the best tidal resources and is commercially
viable in areas of 20 to 40m water where the peak spring tidal current velocity is
greater than 2.25 m/s (Bryans et al., 2005; Whittaker et al., 2003). MCT installed a
prototype with a single 750 kW turbine off Lynmouth in the Bristol Channel during
2003.
Bryans et al. (2004) determined the resource for tidal energy around Ireland us-
ing a 2 dimensional tidal model to simulate the tidal flows for the waters surrounding
the entire island with a 405m by 405m grid. They found that the resource currently
accessible to the tidal device shown in Figure 7.17 is 374 MW around Ireland. How-
ever, Bryans et al. (2004) predict that into the future, TED development will lead
Figure 7.17: The marine current turbines design
to larger turbines which will be financially viable at greater depths and lower spring
current velocities. Based on the predictions by Bryans (2006), a tidal resource of up
to 560MW is investigated here.
7.4.1 Modelling the Tidal Generation
The power output from the potential tidal generation in Ireland was determined by
Bryans et al. (2005) and is shown for a spring tide and a neap tide in Figure 7.18
with 565MW installed.
The power output from a turbine or group of turbines will only reach its max-
imum output during a spring tide, which occurs for a short time twice a month.
Therefore it is not envisaged that developers would consider it economically viable
to rate the electrical equipment to harness all of the energy available at a spring
tide. Instead, the maximum power from the turbine would be down rated by alter-
ing the pitch of the blades. This has been termed Electrical Down Rating (EDR).
It is anticipated that in Ireland, the installed TEDs will undergo 40% downrating
of the maximum rated capacity of the turbines. Thus, the maximum power output
realised is 336MW although there are 560MW installed, as seen in Figure 7.18. It is
500
Spring
Neap
400
Power Output (MW)
300
200
100
00:00 06:00 12:00 18:00 00:00

Time of Day
Figure 7.18: The power output during a spring and neap tide
envisaged that developers will balance the cost of the grid connection and the rating
of electrical components against the capacity saving from spilling energy at higher
tidal flow rates (Bryans, 2006; Bryans et al., 2006).
In order to model the tidal generation, a net load curve was developed by sub-
tracting the tidal generation from the load. In addition, as the tidal generation is
considered to be perfectly forecastable, no additional reserve was required with in-
creasing capacities of tidal generation. The model was run as described in Chapter
3 with the installed wind generation assumed to equal zero. The following figures
illustrate the cycling costs, and emissions and fuelsaving benefits resulting from tidal
generation.
7.4.2 Tidal Cycling Costs
As illustrated in Figure 7.18, the tidal generation has four peaks and troughs per
day representing the tidal current coming in and out twice a day. This fluctuation is
particularly apparent during a spring tide when the variations are at their maximum.
The conventional generation on the system will be required to ramp up and down
inline with these variations in the tidal generation which results in cycling costs. The
magnitude of these variations increase as the installed tidal generation increases,
thus, the cycling costs increase as illustrated in Figure 7.19.
40
35
30
Cycling Costs em
25
20
15
10 Tidal Cycling Costs
0
0 80 160 240 320 400 480 560
Installed Tidal Capacity (MW)
Figure 7.19: Cycling costs with increases in tidal generation
7.4.3 Tidal Emissions Benefits
Figure 7.20 illustrates the emissions benefits from increasing installed penetrations
of tidal generation on the Irish system for the 2010 plant mix and load. It is seen
that the emission savings are quite modest as the installed levels of tidal generation
increase. This is because the installed penetrations are relatively small compared to
the size of the system (representing less than 2% of electricity generation (Bryans,
2006)). The tidal generation output peaks four times a day with a maximum output
of 60% of the installed capacity. This results in the tidal generation having a rela-
tively low load factor when compared to the wind generation - approximately 22%
for tidal generation compared to over 40% for low levels of installed wind generation
(Bryans, 2006).
30000 50
25000
Value of Emissions Savings in em

CO2
20000 SO2 40
NOx
30 30
Kilotons
25
20 20
15
10 10
5 Value of Emissions Savings
0 0
0 80 160 240 320 400 480 560
Installed Capacity (MW)
Figure 7.20: Tidal generation emissions savings
7.4.4 Fuel Savings with Tidal Generation
The fuel savings with increases in tidal generation are shown in Figure 7.21.
As with the emissions savings, because of the relative size of the installed tidal
with respect to the size of the system, the fuel savings are modest. The value of the
saved fuel is shown on the secondary y-axis.
7.4.5 The Net Benefits of Tidal Generation
The previous sections discussed the cycling costs associated with increases in tidal
generation and the emissions and fuel savings. Because tidal generation is per-
fectly forecastable it does not add to the reserve requirement of the system. Bryans
(2006) calculated the capacity benefit of tidal generation to be 22% using the MCT
200 60
Gas
Coal
Value of Fuel Savings in em

Oil
150
Energy Consumed (PJ)
Peat
40
100
20
50
Value of Fuel Savings
0 0
0 80 160 240 320 400 480 560
Installed Capacity (MW)
Figure 7.21: Tidal generation fuel savings
technology with 40% EDR. Without any EDR, this drops to 18%. Thus, the capac-
ity benefit of tidal generation can be thought of as the saved cost of building and
maintaining a conventional generator with a capacity of 22% of the installed tidal
generation.
Since tidal generation is still in its infancy clearly defined capital costs have
not yet been established and forecasting the likely capital costs for 2010 could be
erroneous. In addition, there have been no comprehensive network reinforcement
studies completed for Ireland with respect to tidal generation. Thus, rather than
illustrating the total net benefits of tidal generation, Figure 7.22 illustrates the net
present value of the cycling costs, and the net present value of the capacity, emissions
and fuel saving benefits with an assumed discount rate of 8% and a term of 20 years.
From Figure 7.22 the total benefits of tidal generation are seen to exceed the
cycling costs at all penetrations of tidal generation, however, the capital, operation
and network costs have still to be included. Table 7.10 illustrates the maximum that
800
700
600
500
em
400
300
200
100
NPV of Total Benefits
NPV of Cycling Costs
0
0 80 160 240 320 400 480 560
Installed Tidal Capacity (MW)
Figure 7.22: The total benefits and cycling costs of tidal generation
these other costs could be each year to ensure that the benefits of tidal generation
were greater than the total costs.
The amounts in Table 7.10 represents the maximum that the combined capital,
O&M and network costs can be each year to ensure positive net benefits for tidal
generation. In other words, if the annual capital, O&M and network reinforcement
costs exceeded the amounts shown in Table 7.10 then the costs of tidal generation
will exceed the benefits and the resource should not be developed.
Putting these figures into perspective, if it is assumed that the operation and
maintenance costs of tidal generation were equal to e50,000 per MW per annum
(less than that of an offshore wind farm (Doherty et al., 2006)), annual O&M cost
for 160MW of tidal generation would be e8m. If it were assumed that no network
reinforcement was required with 160MW of tidal, then the capital costs would have
to be less than e2,257,807 per annum to ensure positive net benefits (e 10,257,807
- e8m). If this is the annual cost of capital, then the total capital cost of 160MW
Table 7.10: Boundary annual costs
Installed Tidal Generation Annual Cost em

0 0
80 4,059,204
160 10,257,807
240 14,602,503
320 17,973,800
400 22,023,430
480 28,582,690
560 33,693,617
of tidal would be e22.2m (assuming an interest rate of 8% and a term of 20 years).

This represents an unrealistic capital cost of approximately e140k per MW installed
of tidal generation. In other words, to ensure tidal generation produced positive net
benefits the capital cost would have to be equal to or less than e140,000 per MW
installed. Considering that the best new entrant on the Irish system has a capital cost
of e600,000/MW installed, it is not unreasonable to conclude that tidal generation
will produce negative net benefits at all penetrations (CER, 2006b).
7.4.6 Conclusions of Tidal Generation in Ireland
This section has discussed the potential for tidal generation in Ireland. The feasible
tidal resource and output profiles were based on Bryans et al. (2005) and Bryans
(2006) and these were then subtracted from the load to give a net load curve. The
dispatch model was then run with this net load curve to determine the costs and
benefits of tidal generation. It was found that tidal generation resulted in increased
cycling costs as tidal penetrations increased. The nature of the tidal generation,
with four daily peaks and troughs in output, results in a low load factor for tidal
generation. This leads to relatively small emissions and fuel saving benefits for tidal
generation. To calculate the net benefits of tidal generation it was assumed that
there were no deep network reinforcements necessary with increased tidal generation
and the operation and maintenance costs were assumed to be slightly less than those
for an offshore wind turbine. However, even with these assumptions, in order to
produce positive net benefits for tidal generation, the capital costs would have to
be less than approximately e140,000 per MW installed. This is considered to be an
unrealistic level of capital cost, thus, it is concluded that tidal generation is currently
not a feasible option for Ireland.
CHAPTER 8
Discussion and Conclusions
This thesis presented a cost benefit analysis of wind generation with specific reference
to Ireland as a case study. A range of costs and benefits were included and a large
number of assumptions were tested. A number of additional applications of the
methodology were presented analysing cycling costs, fuelsaver operation, connection
policy and tidal generation. In order to limit the scope of this thesis, it was necessary
to concentrate on the most direct costs and benefits.
Section 8.1 presents a discussion on a number of costs and benefits which could
be included in a wider cost benefit study and some additional issues which were
raised in this thesis are discussed. Section 8.2 reviews the main conclusions of the
work presented throughout this thesis. Section 8.3 outlines some areas for future
research arising from the work presented in this thesis.
8.1 Discussion
In Chapter 1 some costs and benefits of wind generation were introduced which
were not examined in detail in the body of this thesis. The possible implications
of including these costs and benefits are investigated here. In addition, some wider
149
Chapter 8. Discussion and Conclusions 150
issues regarding electricity prices, wind power forecasting, gaming and conventional
plant portfolios are also discussed briefly here.
The dispatch model described in Chapter 3 is a single bus bar system which
neglects any load flow analysis. Thus, issues of transmission congestion are not
included. Congestion limits supplier’s ability to make the most economic decisions
in planning, sourcing and supplying electricity to customers (ESBIE, 2005) resulting
in increased costs. However, as stated by Milborrow (2006), ‘no general rules apply
to transmission congestion economics, which can only be studied on a case-by-case
basis’. This is because congestion is highly system specific and depends on the
location of the generators and loads on the network. Thus, without modelling the
exact location of generators and the network characteristics, it is not possible to
quantify congestion costs. An increase in wind generation, like any conventional
generation, will increase the flows on the lines which may have a knock on effect
on congestion (SEI, 2004c). Workstream 3 of the All Island Renewable Grid study
(AIRGS, 2006a) will conduct load flow analysis for the case study and will identify
potential bottlenecks with increased penetrations of renewable generation. However,
although the cost benefit analysis presented in this thesis did not explicitly include
congestion costs, it did include a cost for network reinforcement. Increased network
capacity ameliorates potential bottlenecks and congestion, thus, the costs of any
additional congestion with increases in wind generation, which were not included
here, may not in fact be significant.
Another cost of wind generation which was not included in this analysis, is the
potential visual impact turbines have on the landscape. Wind turbines tend to
be situated in elevated areas, and as such are considered by some to be a blight
on the landscape (Etherington, 2006). Hurtado et al. (2004) and Möller (2006)
present methodologies to determine a ‘visibility index’ criteria which establishes the
visibility of a wind farm from neighbouring dwellings, however, neither of these
studies attempt to place a value on the cost of this visibility . The cost of the
visual impact of wind generation is very difficult to determine and previous studies
tend to rely on highly subjective public preference models which use questionnaires
to determine the public’s perception of wind turbines (Gamboa and Munda, 2007;
Bishop and Miller, 2007). A more concrete approach to estimating the cost of the
visual impact of wind generation may be to investigate its impact on house prices in
the local area. A report by the Royal Institute of Chartered Surveyors (RICS, 2004)
found that a planning application for a wind farm affects local house prices, however,
the negative impact diminishes as time goes by. In other words, the biggest impact
on the value of local property is at the time of the application for the wind farm,
however, house prices tend to recover within two years of wind farm operation. This
survey also included an assessment for agricultural land, and in some cases, it was
found that the presence of wind turbines on the land actually increased the value of
the agricultural land. This leads to the conclusion that the perception of the visual
impact of wind farms is in fact much more than the actual impact.
This thesis did not investigate issues of dynamic security with increasing pen-
etrations of wind generation. As discussed in Lalor et al. (2005), rates of change
of system frequency will increase as wind generation displaces conventional units.
If doubly fed induction generator (DFIG) wind turbines are employed, additional
static reserve may have to be available to maintain system frequency above a given
threshold. This thesis included additional reserves for increasing penetrations of
wind generation, however, it did not break these down into different types of re-
serves i.e. it did not distinguish between static and dynamic reserve. Thus, a future
cost benefit analysis could include the costs and implications of providing additional
static reserve.
In this thesis, additional operating reserve capacity was dispatched with increas-
ing penetrations of wind generation. This assumes that the existing conventional
generation portfolio has the capability of providing this additional reserve capacity.
It may however, be the case that high penetrations of installed wind generation re-
quire investment in conventional generation to continue to maintain the necessary
levels of reserve capacity for system security. If this were to be the case, it could be
considered a cost attributable to wind generation (although it would be improving
the security of the system as a whole). This long term reserve capacity was not
considered in this thesis. In addition, capacity issues relating to the operating of the
system were not included. It may be the case that by including large penetrations
of wind generation in the plant mix some conventional units will be pushed out
of the merit order entirely. The cost of these stranded assets and any premature
retirement of conventional units was not considered here. Also not considered here
was the physical ability of the system to operate at the extremes e.g. the feasibil-
ity of running the system with a carbon price of e70 per ton resulting in minimal
electricity generation from coal. This may diminish the quality of the results from
the dispatch model. These issues could be included in a wider analysis of wind
generation on a power system.
In addition, as was mentioned in Chapter 3, the costs and benefits of wind gener-
ation calculated in this thesis were based on the scheduled outputs of the dispatched
generators at gate closure. In reality, there will be some real time balancing required
to accurately follow the load and wind variations. This may involve the utilisation
of some of the dispatched reserve, the deloading of conventional units or the curtail-
ment of wind generation. This real time balancing of the generation and load with
wind is an interesting issue which could be included in future research.
This thesis omitted some of the ‘softer’ benefits of wind generation development
such as the creation of local jobs, improvements in local infrastructure leading to
improvements in the standard of living in rural areas. These benefits are very difficult
to estimate and will vary depending on the location of the wind farm. Recent
research by Murphy and Walsh (2002) and Forfás et al. (2003) suggest that very
little benefit should be attributed to creating additional jobs in Ireland because there
is effectively full employment. In fact they suggest that at most you should include a
value of between just 10 - 20% of a job. In addition, wind generation also reduces the
reliance on imported fuels and as such can act as a hedge against international fuel
price and supply variations. This benefit was not included in this thesis. The Irish
Government currently supports the operation of peat fired generation in Ireland by
a levy on all electricity bills, known as the public service obligation (DCMNR, 2002).
The reasoning behind the support of peat is for fuel diversity purposes for security
of supply (ESB, 2001). For the years 2004, 2005 and 2006, the average income
from the PSO levy for the 350MW of installed peat generation has been e57.88m
per annum (CER, 2004b; EirGrid, 2005a). Thus, the Irish public pay on average
e57.88m per year for the security of supply benefits and the local economy benefits
of the peat fired generation. If wind generation was assumed to create these same
benefits, given it’s capacity factor, the value of 350MW of installed wind generation
could be assumed to equal 42% of e57.88m (e24.3m). This equates to a benefit of
e69,450 per MW installed which is approximately e19/MWh. Alternatively, the
feed-in tariff price proposed by the Irish Government could be seen to represent the
value of the fuel diversity benefits of wind generation. As discussed in Chapter 2,
the feed-in tariff for wind generation is approximately e58/MWh.
While this thesis presented a cost benefit analysis of wind generation, it did not
investigate the impact of wind generation on wholesale consumer electricity prices.
Holttinen (2004) found that in the Nordic market, the inclusion of wind generation
in the plant mix results in a reduction in the average spot price and this result
is supported by GE Energy (2005) who showed that spot prices could reduce by
up to 10% in the New York area with the addition of wind generation. Reduced
electricity prices are beneficial for the economy with knock on effects on the standard
of living, industry, trade, employment etc. These potential indirect benefits of wind
generation have not been included here but if they were they would shift the net
benefits curve shown in Chapter 6 to the right resulting in higher critical values for
wind generation. As a caveat, low electricity prices are not a goal in themselves.
Economies should strive for ‘sustainable’ electricity prices as if prices are too low,
there will be issues associated with investment in the electricity industry which could
lead to serious security of supply issues.
Wind forecasting is based on weather forecasting and historical behaviour. The
historical aspect refers to the specific behaviour of the site in response to meteoro-
logical conditions in the past. The accuracy of the wind power forecast is a func-
tion of the method used and the comprehensiveness of the historical data employed
(GE Energy, 2005). This thesis examined three scenarios for wind power forecast
accuracy and it was found that high wind power forecast accuracy increased the
net benefits for wind generation by requiring less reserve capacity, more efficient
conventional unit operation and lower cycling costs. However, it has been shown
by GE Energy (2005) that the benefits to improvements in wind power forecasting
are limited. This study analysed the economic benefits of including 3,300MW of
wind on the New York electricity system under a range of operating approaches. It
found that including wind power forecasts in the operating decisions of conventional
generators resulted in much greater savings than simply assuming a ‘fuelsaver’ type
approach (this result has also been supported in this thesis in Chapter 7). Wind fore-
casts based on the current ‘best practice’ were used in this analysis and the benefits
were compared to a perfect wind forecast. Using the wind forecasts in the commit-
ment decisions resulted in a reduction in electricity system variable costs by $430m
per annum compared to $335 when a ‘fuelsaver’ type approach was used. However,
assuming perfect wind power forecasts provided only a further $25m saving. This
shows that additional investment in improving wind power forecast accuracy above
current levels may in fact result in only relatively small benefits in electricity sys-
tem operating costs. In fact, if the cost of research, development and equipment to
improve wind power forecasts were included, the cost of improved accuracy could
be greater than the benefits it provides.
This thesis assumed a perfectly competitive electricity market, however, in reality
it is possible that generators may behave strategically and distort the optimal results.
There has been a large body of work examining gaming behaviour in electricity
markets, such as Guan et al. (2001), Gountis and Bakirtzis (2004) and Hobbs et al.
(2000). An interesting study conducted by Wen and David (2001) compares the
social welfare benefits of a perfectly competitive electricity market to one where
generators strategically bid into the market to maximise their own revenue based
on rival bids. This study simulated a numerical example and showed that strategic
bidding resulted in higher market clearing prices and thus revenues to the majority
of generators. With reference to the Irish system, a report on potential gaming in the
Irish electricity market prepared by ILEX energy consulting for the Irish regulator
found that strategic behaviour would result in increased prices throughout the day,
but in particular, during periods of peak demand (ILEX, 2003). With reference to
the net benefits of wind generation discussed in Chapter 6, strategic bidding could
result in alterations to the schedule of conventional generators on the system. This
could have a knock-on impact on the cycling costs and the emissions and fuel saving
benefits of wind generation. Analysing the impact of electricity market gaming on
the benefits of renewable generation would provide an interesting study and it is
proposed in Section 8.3 that this is an area of potential future work.
The results shown in this thesis were given for a case study on the Irish system.
In general, the costs and benefits of wind generation will be highly dependent on
the underlying plant mix, however, the methodology described in this thesis could
easily be applied to other systems and other forms of renewable generation. This
thesis examined the optimal penetrations of wind generation given a number of
specified conventional plant mixes on the Irish system. However, looking into the
future, if it is envisaged that wind generation will play a major role in the plant
portfolio, then the conventional plant mix should be optimised to accommodate this
wind generation. A study conducted by Doherty et al. (2006) examines the optimal
future conventional plant portfolios with high levels of installed wind generation.
Their analysis shows that with increasing penetrations of wind generation, there is
a reduction in the necessity for baseloaded generation and an increase in peaking
capacity. In particular, the results point towards a reduction in coal fired generation
and an increase in OCGTs with increasing wind generation penetrations.
8.2 Conclusions
This thesis presented a methodology for determining the optimal penetrations of

wind generation on a power system given specified conventional plant mixes. The
methodology was applied to a case study and a cost benefit analysis of wind gener-
ation on the case system was conducted. In addition, a number of applications of
the net benefits methodology were also presented. The main conclusions arising out
of the cost benefits analysis and the applications are presented here.
• It was found that the additional reserve requirement for wind generation causes
a knock on effect on the cycling costs, and the emissions and fuel saving ben-
efits.
• The cycling cost of conventional units on the system were shown to increase as
the installed wind generation increased, particularly at high levels of installed
wind generation when previously baseloaded units began to be displaced by
the wind output.
• It was found that wind generation results in a reduction in carbon dioxide,

sulphur dioxide and nitrogen oxide emissions. The carbon dioxide emissions
are of a much greater magnitude than the other two emissions, thus the price
of carbon was the main determinant in the value of the emissions savings. The
emissions reductions are dependent on the particular units being displaced and
are affected by low load operation and cyclical operation.
• The most significant fuel savings with wind generation for the case study were
in gas. The fuel savings are also affected by low load and cyclical operation. In
addition, the load factors of wind generation have an impact on the potential
emissions and fuel savings.
• It was shown that wind generation has positive net benefits up to 21% of
electricity generated in 2010, and 21.5% in 2015 and 22.2% for the base case
assumptions for the case study.
• A sensitivity analysis of the net benefits of wind showed that the net benefits of
wind were significantly affected by the underlying plant mix. It was also shown
that high load factors, low cycling costs, low capital costs and accurate wind
power forecasts significantly increased the net benefits of wind generation. Low
emissions prices significantly reduced the net benefits as did increased discount
rates and low demand growth.
• A worst and best case scenario were developed which illustrate the extreme
critical values for wind generation. In the worst case, wind generation ex-
pressed negative net benefits with penetrations of just 5% of electricity gener-

ation. Positive net benefits were experienced up to 30% of electricity generated
through wind in the best case scenario.
• The carbon dioxide reduction benefits of a carbon price can be completely

outweighed by the added cycling costs. For the case study it was found that
the cycling costs exceeded the carbon savings when carbon prices were less
than e50/ton. In addition, it was shown that the most significant contributor
to the cycling costs is the switching of coal units to the margin.
• When a carbon price and wind generation are combined, the cycling costs
are increased further, however, the combination of a carbon price and wind
generation have a much greater impact on carbon dioxide reductions. It was
found that even with 3000MW of installed wind generation, the cycling costs
can still exceed the value of the saved carbon dioxide emissions.
• Operating a power system under a fuelsaver approach is entirely futile as it

produces negative net benefits at all levels of installed wind generation.
• Employing non-firm connection policies significantly increases the renewable

generation capacity that can connect to the distribution network without the
need for additional investment. Also a significant increase in the net benefits of
distributed generation is gained if the appropriate connection policy is utilised
from the outset and conversely that significant costs are incurred if ad hoc
policies are employed.
• The net benefits of tidal generation were investigated and it was shown that
the capital costs of tidal generation would have to be less than e140k per MW
installed for tidal generation to have positive net benefits for the case study
in 2010. This is an infeasibly low capital cost resulting in the conclusion that
tidal generation is not a feasible option for the Irish system at the given time.
8.3 Future Work
The costs and benefits calculated in this thesis were based on the output of the
dispatch model. In this model, generators bids were based on their fuel costs. Given
the significant nature of the cycling costs as illustrated in Chapters 4 and 6, it is
likely that in reality generators bids would be more complex to ensure they were
compensated for any starts, shutdowns or excessive cycling. As the dispatch model
did not account for temporal constraints the impact of complex bids could not be
accounted for. A full unit commitment type model would be required for this type of
analysis and a cost benefit analysis of wind generation based on a unit commitment
model would provide an interesting area for future research.
This thesis also assumed a perfectly competitive electricity market, which in
reality may not always be the case. Strategic bidding by generators could alter the
merit order in which they are dispatched which would have a knock on effect on
the costs and benefits of wind generation. The impact of gaming in the electricity
market on the net benefits of wind generation is a complex issue and could provide
insightful results on the realistic operation of electricity markets.
This thesis concentrated on the costs and benefits of wind and tidal generation in
isolation, however the methodology could easily be adapted to incorporate the costs
and benefits of other forms of renewable generation. A broader study comprising
other forms of renewable generation would provide very useful results indicating
the forms of renewable generation which are most suited to the underlying plant
mix and also, the forms of renewable generation which should be promoted and if
necessary supported by Government. Also, other forms of generation such as nuclear
or advanced coal units could be included.
In addition, this thesis did not investigate support mechanisms for renewables.
An interesting area for future research would be on the analysing different support
mechanisms for wind generation and determining which one is most appropriate to
achieve the optimal levels of penetration at least cost. In addition, if the net benefits
of other forms of renewable energy were included, those support mechanisms could
be tested to determine if they supported the appropriate technologies, to the optimal
penetrations and at the correct time.

Issues relating to network congestion and system dynamics were not included
in this study. It is likely that wind generation will have an impact on these issues
and future work to incorporate these factors in a cost benefit analysis would be very
interesting.
References
Ahmed, S., Cohen, J., 1994. Renewable energy sources: technology and economics.
In: IEEE International Conference on Systemts, Man. and Cybernetics, Humans,
Information and Technology. pp. 2355–2360.
AIP, 2005. All Island Project (AIP). All-island modelling project. Available:
http://www.allislandproject.org.
AIP, dec 2006a. All Island Project (AIP). Market power mitigation in the SEM,
market monitoring in the SEM decision paper - AIP/SEM/217/06. Available:
http://www.allislandproject.org.
AIP, nov 2006b. All Island Project (AIP). Transition from the existing bilateral mar-
kets to the single electricity market. Available: http://www.allislandproject.org.
AIRGS, 2006a. All-Island Renewable Grid study (AIGS). work-stream 3: Network

study. Available: http://www.dcmnr.gov.ie.
AIRGS, 2006b. All-Island Renewable Grid study (AIRGS). High level assessment of
suitable generation portfolios for the all-island system in 2020. Work-Stream 2A.
Akhmatov, V., Knudsen, H., Bruntt, M., Nielsen, A., Pedersen, J., Poulsen, N.,
2000. A dynamic stability limit of grid connected induction generators. In: Proc.
160
REFERENCES REFERENCES
of IASTED International Conference on Power and Energy Systems. Marabella,

Spain, pp. 235–244.
Alberta Environment, 2003. Sulphur dioxide: environmental effects, fate and be-
haviour - prepared by WBK and Associates Inc. Available: http://www.gov.ab.ca/.
Alvarez-Farizo, B., Hanley, N., 2002. Using conjoint analysis to quantify public
preferences over the environmental impacts of wind farms. an example from spain.
Energy Policy 30, 107–116.
American Society of Mechanical Engineers, Dec 1990. Class 1 components in elevated

service, section III division 1. Cases of ASME Boiler and Pressure Vessel Code (N-
47, Rev. 29).
Baumol, W., Oates, W., 1975. The Theory of Environmental Policy, 1st Edition.
Cambridge University Press, Cambridge, UK.
Bazilian, M., Denny, E., O’Malley, M., 2004. Challenges of increased wind energy
penetration in ireland. Wind Engineering 28 (1), 43–56.
Bergmann, A., Hanley, N., Wright, R., 2006. Valuing the attributes of renewable
energy investments. Energy Policy 34, 1004–1014.
Bishop, I. D., Miller, D. R., 2007. Visual impact of off-shore wind turbines: the
influence of distance, contrast, movement and social variables. Renewable Energy
32, 814–831.
Borchiellini, R., Massardo, A., Santarelli, M., 2000. Analytical procedure for carbon
tax evaluation. Energy Conversion and Management 41, 1509–1531.
Brooks, D., Lo, E., Zavadil, R., Santoso, S., Smith, J., may 2003. Characterizing the
impact of significant wind generation facilities on bulk power system operations
planning. Utility Wind Integration Group (UWIG): http://www.uwig.com.
Brown, L., 2006. Plan B 2.0: Rescuing a planet under stress and a civilisation in
trouble. Chapter 10 - Stabilizing Climate. Available: www.energy-policy.org.
161
Bryans, A., 2006. Impacts of Tidal Stream Devices on Electrical Power Systems.
PhD Thesis, The Queen’s University Belfast.
Bryans, A. G., Denny, E., Fox, B., Crossley, P., O’Malley, M., 2005. Study of the
effect of tidal generation on the irish grid system. In: CIGRE Symposium on
Power Systems with dispersed generation. Athens, Greece.
Bryans, A. G., Fox, B., Crossley, P., O’Malley, M. J., 2006. The predictability and
variability effects of renewable generation. IEEE Transactions on Power Systems
(in review).
Bryans, A. G., Fox, B., Crossley, P., Whittaker, T. J. T., O’Malley, M. J., 2004. Tidal
stream resource and impact assessment for Ireland. In: International Conference
on Advanced Power System Automation and Protection. Juju Island, Korea.
BWEA, 2005. British Wind Energy Association (BWEA). Can we rely on the wind
- online. Available: http://www.bwea.com/energy/rely.html.
BWEA, 2006. British wind energy association - wind energy. Available:

http://www.bwea.com/.
Cabot, G., Vauchelles, D., Taupin, B., Boukhalfa, A., 2004. Experimental study of
lean premixed turbulent combustion in a scale gas turbine chamber. Experimental
Thermal and Fluid Science 28, 683–690.
Castro, R., Ferreira, L., 2001. A comparison between chronological and probabilistic
methods to estimate wind power capacity credit. IEEE Transactions on Power
Systems 16, 904–909.
Causebrook, A., Fox, B., 2004. Decoding grid codes to accommodate diverse gener-
ation technologies.
CEPAARB, 2004. California Environmental Protection Agency Air Resources Board

(CEPAARB) Report to the Legislator - gas-fired power plant NOx emission con-
trols and related environmental impacts. Available: www.arb.ca.gov.
162
CER, 2001. Commission for Energy Regulation (CER). Charges for connection to
the ESB’s transmission system. Available: www.cer.ie.
CER, 2004. Commission for Energy Regulation (CER). A regulatory approach to

esb dominance. Available: http://www.cer.ie.
CER, 2004a. Commission for Energy Regulation (CER). Options for operational
rules to curtail wind generation. Available: www.cer.ie.
CER, 2004b. Commission for Energy Regulation (CER). Public service obligation
levy 2004. CER/03/186. Available: www.cer.ie.
CER, 2006a. Commission for Energy Regulation (CER). CER submission to the de-
partment of communications, marine and natural resources on the green paper on
energy: towards a sustainable energy future for ireland - CER/06/259. Available:
http://www.cer.ie.
CER, 2006b. Commission for Energy Regulation Ireland. Best new entrant price
2007 - CER/06/148. Available: http://www.cer.ie.
Collinson, A., Earp, G. K., Howson, D., Owen, R. D., Wright, A. J., 1999. The effect
of increasing levels of embedded generation on the distribution network - ETSU
K/EL/00184/REP. Available: www.dti.gov.uk.
Combined Cycle Journal, 2004. Editorial. Profitable operation requires knowing how
much it costs to cycle your unit. Available: http://www.psimedia.info/ccj.htm
Second Quarter.
Conniffe, D., FitzGerald, J., Scott, S., Shortall, F., 1997. The cost to Ireland of
greenhouse gas abatement. ESRI Policy Research Series 32.
Correia, P., 2006. Decentralised unit commitment in a market structure: problem

formulation and solution advancement. IEE Proceedings on Generation, Trans-
mission and Distribution 153 (1).
Cregan, M., Flynn, D., 2003. Thermal Power Plant Simulation and Control - Editor:
D. Flynn. IEE Power and Energy Series, London, UK, Ch. 1.
163
CSO, 2006. Central Statistics Office Ireland (CSO). Census 2006 - preliminary re-
port. Available: http://www.cso.ie.
Cunningham, A., 1978. The reduction of atmospheric pollutants during the burning
of heavy fuel oil in large boilers. Journal of the Institute of Fuel 51, 20–29.
Dale, L., Milborrow, D., Slark, R., Strbac, G., 2004. Total cost estimates for large-
scale wind scenarios in UK. Energy Policy 32, 1949–1956.
Dalton, S., 2004. Cost comparison IGCC and advanced control.Roundtable on de-
ploying advanced clean coal plants. Available: http://www.climatevision.gov.
Dany, G., Sep. 2001. Power reserve in interconnected systems with high wind power
production. In: IEEE Power Tech Conference. Porto, Portugal.
DCMNR, 2002. Department of Communications Marine and Natural Resources

(DCMNR). Statutory instrument no 217 of 2002 - electricity regulation act 1999:
(public service obligations) order 2002. Available: www.dcmnr.gov.ie.
DCMNR, 2006. Department of Communications, Marine and Natural Resources

(DCMNR). AER programme and profile of AER I - VI competitions. Available:
http://www.dcmnr.gov.ie.
Delabroy, O., Haile, E., Lacas, F., Candel, S., Pollard, A., Sobiesiak, A., Becker,
H., 1998. Passive and active control of NOx in industrial burners. Experimental
Thermal and Fluid Science 16, 64–75.
Deloitte, 2005. Review of the electricity sector in Ireland - final report. Available:
DENA, 2005. Deutsche Energie-Agentur GmbH (DENA).Planning of the grid in-

tegration of wind energy in germany onshore and offshore up to the year 2020.
[Online] http://www.dena.de.
DETI, 2003. Department of the Enterprise Trade and Investment (DETI). A study
into the economic renewable energy resource in northern ireland and the abil-
164
ity of the electricity network to accommodate renewable generation up to 2010.

Available: www.energy.detini.gov.uk.
Doherty, R., O’Malley, M., 2005. New approach to quantify reserve demand in sys-
tems with significant installed wind capacity. IEEE Transactions on Power Sys-
tems 20 (2), 587–595.
Doherty, R., Outhred, H., O’Malley, M., Nov. 2005. Generation portfolio analysis
for a carbon constrained and uncertain future. In: Future Power Systems. Ams-
terdam.
Doherty, R., Outhred, H., O’Malley, M., 2006. Establishing the role that wind may
have in future generation portfolios. IEEE Transactions on Power Systems 21 (3),
1415–1422.
Dragoon, K., Milligan, M., May 2003. Assessing wind integration costs with dispatch
models: a case study of pacificorp. In: Windpower. Austin, Texas.
DWEA, 2006. Danish Wind Energy Association (DWEA). [Online]

http://www.windpower.org/en/tour.htm.
DWEA Kids, 2006. Danish Wind Energy Association (DWEA) in association with
the Danish Ministry of Education and the Danish Ministry of the Environment
and Energy. [Online] http://www.windpower.org/en/kids/index.htm.
Ede, S., Zimmerman, R., Mount, T., Thomas, R., Schulze, W., 2000. An economic
analysis of the self commitment of thermal units. In: Proc. 33rd Hawaii Interna-
tional Conference on System Sciences.
EIA, 2000. Energy Information Administration (EIA). Electric power annual 2000.
Available: www.eia.doe.gov II.
EIA, 2004. Energy Information Administration (EIA) - International energy outlook.

Available: http://www.eia.doe.gov.
EIA, 2004. Energy Information Administration (EIA) energy glossary.

www.eia.doe.gov.
165
EirGrid, 2004. Impact of wind power generation in ireland on the operation of con-
ventional plant and the economic implications. Available: http://www.eirgrid.com.
EirGrid, 2005a. Quarterly review of the irish electricity market - issue 12. Available
http://www.eirgrid.com.
EirGrid, 2005b. Transmission System Operator Ireland. Generation adequacy report:

2006 - 2012. Available: www.eirgrid.com.
EirGrid, 2006a. Market and system data (download section). Available:

EirGrid, 2006b. Statement of charges and payments for ancillary services providers
2006. Available: http://www.eirgrid.ie.
EirGrid, 2006c. Transmission and distribution connections. Available

EirGrid, 2006d. Transmission forecast statement 2006-2012. Available:

www.eirgrid.com.
ESB, 2001. Electricity Supply Board (ESB). ESB welcomes EU backing for peat
generation - press release. Available: http://www.esb.ie/.
ESB, 2003. Electricity Supply Board (ESB). Strategy to reduce emissions

of transboundary air pollution by 2010 - submissions by ESB. Available:
http://www.environ.ie/.
ESB, 2004. Networks and National Grid. Group processing approach for renewable
generator connection applications - joint TSO/DSO proposal to CER. Available:
www.cer.ie CER/04/317.
ESB, 2006. Electricity Supply Board (ESB). ESB at a glance. Available:

http://www.esb.ie/.
ESB Networks, 2002. Irish distribution code.
166
ESB Networks, 2005. ESB Networks infrastructure.

URL http://www.esb.ie/esbnetworks/
ESBI, ETSU, 1997. Total renewable energy resource in ireland. Available:

ESBIE, 2005. Electricity Supply Board Independent Energy (ESBIE). Independent

energy news winter 2005. Available: www.esbie.ie.
Etherington, J. R., 2006. The case against windfarms. Available.

http://www.countryguardian.com.
EU Directive 2001/77/EC, 2001. of the European Parliament and of the Council

on the promotion of electricity produced from renewable energy sources in the
internal electricity market. Available: www.europa.eu.int.
EU Directive 2001/80/EC, 2001. of the European Parliament and of the Council.

The limitation of emissions of certain pollutants into the air from large combustion
plants, october 2001. Available: www.europa.eu.int.
EU Directive 96/92/EC, 1996. of the European Parliment and of the Council of

19 december 1996 concerning common rules for the internal market in electricity.
Available: http://www.europa.eu.
EU ETS, 2003. European union greenhouse gas emissions trading scheme under
directive 2003/87/EC. Available: http://www.europa.eu.
European Climate Exchange, 2006. Carbon financial instruments. Available:

http://www.europeanclimateexchange.com.
Eurostat, 2006. Gas and electricity market statistics - data 1990 to 2006. Available:
http://www.europa.eu.
EWEA, 2003. European Wind Energy Association (EWEA). Wind energy - the
facts. An analysis of wind energy in the EU-25. Available: http://www.ewea.org.
167
EWEA, dec 2005. European wind energy agency (EWEA) - large scale integration of
wind energy in the european power supply : analysis, issues and recommendations.
Available: http://www.ewea.org.
Flynn, D., 2003. Thermal Power Plant simulatoin and control, 1st Edition. IEE
Power and Energy Series 43, IEE, London.
Forfás, Murphy, A., Barry, F., 2003. Forfas: The National Policy and Advisory
Board for Enterprise, Trade, Science, Technology, and Innovation. The economic
appraisal system for projects seeking support from the industrial development
agencies dublin.
Galluzzo, T., 2005. Small packages, big benefits: economic advantages of local wind
projects. Available: www.iowapolicyproject.org.
Gamboa, G., Munda, G., 2007. The problem of wind farm location: a social multi-
criteria evaluation framework. Energy Policy 35, 1564–1583.
Gardiner, P., Snodin, H., Higgins, A., Goldrick, S. M., Feb. 2003. The impacts of
increased levels of wind penetration on the electricity systems of the republic of
ireland and northern ireland (3096/GR/04).
Gardner, P., Snodin, H., Higgins, A., Goldrick, S. M., 2003. The impacts of increased
levels of wind penetration on the electricity systems of the Republic of Ireland and
Northern Ireland. Final Report to the Commission for Energy Regulation/OFREG
by Garrad Hassan 3096/GR/04.
GE Energy, 2005. The effects of integrating wind power on transmission system

planning, reliability, and operations: Report on phase 2. Prepared for The New
York State Energy Research and Development Authority.
Giebel, G., Landberg, L., Kariniotakis, G., Brownsword, R., 2003. State-of-the-
art on methods and software tools for short-term prediction of wind energy pro-
duction. In: European Wind Energy Conference and Exhibition EWEC. Madrid,
Spain.
168
Goodstein, E., 2002. Economics and the environment, 3rd Edition. John Wiley and
Sons Inc., New York, USA.
Gountis, V. P., Bakirtzis, A. G., 2004. Efficient determination of cournot equilibria

in electricity markets. IEEE Transactions on Power Systems 19 (4), 1837–1844.
Gouveia, E., Matos, M., Sep. 2004. Operation reserve in a power system with a large
amount of wind power. In: International Conference on Probabilistic Methods
Applied to Power Systems. Ames, Iowa.
GRE Study, 2003. Great River Energy (GRE) wind integration study. In: UWIG
Technical Workshop. Seattle, WA.
Grimsrud, P., Lefton, S., 1995. Economics of cycling 101: what do you need to know
about cycling costs and why? APTECH Engineering Technical Paper TP098,
http://www.aptecheng.com.
Gross, R., Hemptonstall, P., Anderson, D., Green, T., Leach, M., Skea, J., 2006.
The costs and impacts of intermittency: An assessment of the evidence on the
costs and impacts of intermittent generation on the british electricity network.
A report of the Technology and Policy Assessment Function of the UK Energy
Research Centre.
Guan, X., Ho, Y., Pepyne, D., 2001. Gaming and price spikes in electric power
markets. IEEE Transactions on Power Systems 16.
Hadley Centre, 2006. Climate Prediction and Research. The UK Met Office - the
impact of climate change on the food supply. Available: http://www.metoffice.com.
Hagstrom, E., Norheim, I., Uhlen, K., july 2005. Large scale wind power integration
in norway and impact on damping in the nordic grid. Wind Energy 8, 375–384.
Harris, B., 1990. Conversion of sulphur dioxide to sulphur trioxide in gas turbine
exhaust. Journal of Engineering for Gas Turbine and Power 112, 585–589.
Hatziargyriou, N., Contaxis, G., Papadopoulos, M., Papadias, B., Matos, M., Lopes,
J. P., Nogaret, E., Kariniotakis, G., Halliday, J., Dutton, G., Dokopoulos, P.,
169
Bakirtzis, A., Androutsos, A., Stefanakis, J., Gigantidou, A., 2000. Operation and
control of island systems - the crete case. In: IEEE Power Engineering Society
Winter Meeting.
Hatziargyriou et al., 2002. Energy management and control of island power sys-
tems with increased penetration from renewable energy sources. In: IEEE Power
Engineering Society Winter Meeting.
Helm, D., 2005. Climate-change policy, 1st Edition. Oxford University Press, Oxford,
UK.
Hill, J., Thornley, V., Barbier, C., Maloyd, A., Oliver, A., 2005. Innovative voltage
control techniques for optimum connection of renewable and intermittent genera-
tion to distribution networks. In: Proc. of 40th International Universities Power
Engineering Conference. Cork.
Hird, C., Leite, H., Jenkins, N., Li, H., Mar. 2004. Network voltage controller for
distributed generation. IEE Proc.-Gener. Transm. Distrib. 151 (2), 150–156.
Hirst, E., 2001. Interactions of wind farms with bulk-power operations and markets.
Prepared for the project for Sustainable FERC Energy Policy.
Hirst, E., 2002. Integrating wind energy with the bpa power system: Prelimary
study. Prepared for Power Business Line Bonneville Power Administration.
Hobbs, B. F., Metzler, C. B., Pang, J. S., 2000. Strategic gaming analysis for electric
power systems: an mpec approach. IEEE Transactions on Power Systems 15 (2),
638–645.
Holttinen, H., 2004. The impact of large scale wind power production on the Nordic
electricity system - VTT publications 554. Julkaisija, VTT technical research cen-
tre of finland.
Holttinen, H., Meibom, P., Ensslin, C., Hofmann, L., McCann, J., Pierik, J., Tande,
J., Hagstrom, E., Estanqueiro, A., Amaris, H., Söder, L., Strbac, G., Parsons, B.,
170
may 2006. Design and operation of power systems with large amounts of wind
power, IEA collaboration. In: Nordic Wind Power Conference. Espoo, Finland.
Hunter, S., 1982. Formation of SO3 in gas turbines. Journal of Engineering for
Power 104, 44–51.
Hurley, T., Watson, R., 2002. An assessment of the expected variability and load fol-
lowing of a large wind penetration in ireland. In: Global Wind Power Conference.
Paris, France.
Hurtado, J. P., Fernádez, J., Parrondo, J. P., Blanco, E., 2004. Spanish method
of visual impact evaluation in wind farms. Renewable and Sustainable Energy
Reviews 8, 483–491.
IAEA, 2006. International Atomic Energy Agency (IAEA). Energy and environment
data reference bank. Available: http://www.iaea.org.
IDA, 2006. Industrial Development Agency (IDA) Ireland. URL:

http://www.idaireland.com.
IEE, 2003. Embedded generation issues - a briefing note. An Energy and Environ-
ment Fact Sheet provided by the IEE.
IEEE Subcommittee, 1990. Current Operating Problems Subcommittee. A report

on the operational aspects of generation cycling. IEEE Transactions on Power
Systems 5 (4), 1194–1203.
IEEE/CIGRE, 2004. The Institute of Electrical and Electronics Engineers

(IEEE)/International Council on Large Electric Systems (CIGRE). Joint task
force on stability terms and definitions. Definition and classification of power sys-
tem stability. IEEE Transactions on Power Systems 19 (2), 1387–1401.
ILEX, 2003. ILEX Energy Consulting, in conjuction with UMIST and University
College Dublin.The price and dispatch impact of a centralised wholesale electricity
market in ireland. Available: http://www.eirgrid.com.
171
ILEX, Strbac, G., 2002. Quantifying the system cost of additional renewable gener-
ation in 2020. Available: http://www.dti.gov.uk.
IMD, 2004. International Institute for Management Development (IMD). World

competitiveness yearbook 2004. Available: http://www.imd.ch.
IPPC, 1996. Integrated Pollution Prevention and Control (IPPC) Directive.Council

directive 96/61/EC of 24 september 1996 concerning integrated pollution and
control. Available: www.europa.eu.int.
IWEA, 2004. Irish Wind Energy Association (IWEA) submission on CER proposed
direction on resuming connection offers for wind generators (CER/04/354). Avail-
able: www.cer.ie.
Johnson, A., Urdal, H., 2004. Technical connection requirements for wind farms.
Kariniotakis, G., Pinson, P., 2003. Evaluation of the MORE-CARE wind power
prediction platform. performance of the fuzzy logic based models. In: European
Wind Energy Association Conference. Madrid, Spain.
Katz, M., Rosen, H., 1998. Microeconomics, 3rd Edition. McGraw-Hill Companies.
Keane, A., Denny, E., O’Malley, M., 2006. Quantifying the impact of connection
policy on distributed generation. IEEE Transactions on Energy Conversion (in
press).
Keane, A., O’Malley, M., 2005a. Impact of distribution network constraints on dis-
tributed generation capacity. In: Proc. of 40th International Universities Power
Engineering Conference. Cork.
Keane, A., O’Malley, M., 2005b. Optimal allocation of embedded generation on

distribution networks. IEEE Transactions on Power Systems 20 (3), 1640–1646.
Keane, A., O’Malley, M., Jun. 2006. Optimal distribution generation plant mix with
novel loss adjustment factors. In: IEEE PES General Meeting. Montreal, Canada.
172
KEMA, May 2005. The contribution to distribution network fault levels from the
connection of distributed generation. Tech. Rep. DG/CG/00027/00/00, Depart-
ment of Trade & Industry UK.
Kennedy, S., 2005. Wind power planning: assessing long term costs and benefits.
Energy Policy 33, 1661–1675.
Kesgin, U., 2003. Study on prediction of the effects of design and operating param-
eters on NOx emissions from a leanburn natural gas engine. Energy Conversion
and Management 44, 907–921.
Kirby, B., Milligan, M., Hawkins, D., Makarov, Y., Jackson, K., Shu, H., 2003. Cali-
fornia renewable portfolio standard renewable generation integration cost analysis.
phase 1: One year analysis of existing resources, results and recommendations;.
Kirschen, D., Strbac, G., 2004. Fundamentals of Power system economics. John
Wiley and Sons Ltd., West Sussex, England.
Klaassen, G., Riahi, K., 2006. Internalising externalities of electricity generation:

An analysis with message-macro. Energy Policy (in press).
Kundur, P., 2004. Power system stability and control. McGraw Hill Inc.
Kyoto, 1992. United Nations Framework Convention on Climate Change. Kyoto

protocol. Available: www.unfccc.int.
Lalor, G., Mullane, A., O’Malley, M., 2005. Frequency control and wind turbin
technologies. IEEE Transactions on Power Systems 20 (4), 1905–1913.
Le, K., Jackups, R., Feinstein, J., Thompson, H., Wolf, H., Stein, E., Gorski, A.,
Griffith, J., 1990. A report prepared by the IEEE current operating problems
subcommittee: Operational aspects of generation cycling. IEEE Transactions on
Power Systems 5 (4), 1194–1203.
Lefton, S., Besuner, P., 2001. Power plant cycling operations and unbundling
their effect on plant heat rate. APTECH Technical Paper TP134, Available:
http://www.aptecheng.com.
173
Lefton, S., Besuner, P., Grimsrud, G., 1997. Understand what it really costs to cycle
fossil-fired units. Power 141 (2), 41–42.
Lefton, S., Besuner, P., Grimsrud, G., Bissel, A., Norman, G., 1998. Optimizing
power plant cycling operations while reducing generating plant damage and costs
at the irish electric supply board. In: International conference on plant mainte-
nance for managing life and performance. Baltica IV, Helsinki, Stockholm.
Li, K., Thompson, S., Peng, J., 2003. Modelling and prediction of NOx emission in
a coal-fired power generation plant. Control Engineering Practice 12, 707–723.
Lightman, P., Street, P., 1981. The chemistry and physics of heavy oil combustion.
In: The Heavy End of the Barrel-Future Trends in Oil Firing, Institute of Energy
Symposium. The Institute of Energy, Portsmouth, England, pp. 44 – 55.
Lovgren, S., 2004. Warming to cause catastrophic rise in sea level? Available:
http://www.nationalgeographic.com.
Malik, A., Awsanjli, A., May 2004. Energy fuel saving benefit of a wind turbine. In:
IEEE MELECON. Dubrovnik, Croatia, pp. 1041–1044.
Mansour, A., Benjamin, M., Straub, D., Richards, G., 2001. Application of macro-
lamination technology to lean, premix combustion. AMSE Journal of Engineering
for Gas Turbines and Power 123 (4), 796–802.
Marti, I., Kariniotakis, G., Pinson, P., Sanchez, I., Nielsen, T., Madsen, H., Giebel,
G., Usaola5, J., Palomares, A., Brownsword, R., Tambke, J., Focken, U., Lange,
M., Sideratos, G., Descombes, G., 2006. Evaluation of advanced wind power fore-
casting models - results of the anemos project. Available: http://anemos.cma.fr.
Mazza, P., 2005. Harvesting clean energy for rural development. Part one. Climate
solutions special report. Available: http://www.harvestcleanenergy.org.
McArdle, J., 2004. Dynamic modelling of wind turbine generators and the impact
on small lightly interconnected grids. Wind Engineering 28 (1).
174
Meibom, P., Barth, R., Brand, H., Weber, C., 2006. Deliverable D6.2 (b) -
Documentation Methodology of the Scenario Tree Tool. Wind power integra-
tion studies using a multi-stage stochastic electricity system model. Available:
http://www.wilmar.risoe.dk.
Milborrow, D., 2006. No limits to high wind penetration. Wind Power Monthly
22 (9).
Miller, A., Muljadi, E., Zinger, D., 1997. A variable speed wind turbine power
control. IEEE Transactions on Energy Conversion 12 (2), 181–186.
Molero, L., 1997. Pollutant formation and interaction in the combus-

tion of heavy liquid fuels - PhD thesis, university of london. Available:
www.chemeng.ucl.ac.uk/research/combustion.
Möller, B., 2006. Changing wind-power landscapes: regional assessment of visual

impact on land use and population in northern jutland, denmark. Applied Energy
83, 477–494.
Montes, G., Martin, E., Garcia, J., 2007. The current situation of wind energy in
spain. Renewable and Sustainable Energy Reviews 11, 467–481.
Mullane, A., Lightbody, G., Yacamini, R., 2005. Wind-turbine fault ride-through
enhancement. IEEE Transactions on Power Systems 20 (4), 1929–1937.
Munksgaard, J., Larsen, A., 1998. Socio-economic assessment of wind power - lessons
learned from denmark. Energy Policy 26 (2), 85–93.
Murphy, A., Walsh, B., autumn 2002. The rationale for subsidising jobs in a fully-
employed economy. Irish Banking Review.
NIAER, 2003. Northern Ireland Authority for Energy Regulation (NIAER). Inter-
connector trading from 1 april 2004 - a decision paper. Available: www.nics.gov.uk.
NICS, 2005. Northern Ireland Civil Service (NICS). Northern ireland executive eco-
nomic overview. Available: www.nics.gov.uk.
175
OFGEM, 2003. Office of Gas and Electricity Markets (OFGEM).Structure of elec-

tricity distribution charges - initial conclusions. Available: www.ofgem.gov.uk.
O’Mahony, M., 2004. ESB Power Generation. Personal communication with the
electricity research centre. Electricity Supply Board, Dublin, Ireland.
Palmer, K., Burtraw, D., Shih, J., 2006. The benefits and costs of reducing emisions
from the electricity sector. Journal of environmental management (in press).
Palsson, M., Toftevaag, T., Uhlen, K., Tande, J., 2003. Control concepts to en-
able increased wind power penetration. In: IEEE PES General Meeting. Toronto,
Canada, pp. 1984–1990.
Pepermans, G., Driesen, J., Haeseldonckx, D., Belmans, R., D’haeseleer, W., 2005.
Distributed generation: definition, benefits and issues. Energy Policy 33, 787–798.
Pinson, P., Kariniotakis, G., 2003. Wind power forecasting using fuzzy neural net-
works enhanced with on-line prediction risk assessment. In: Proc. 2003 IEEE
Bologna Power Tech Conference: ref. BPT03-387. Bologna.
PLEXOS, 2006. PLEXOS for power systems - electricity market simulation.

www.draytonanalytics.com.
Point Carbon, 2006. Historic prices. Available: http://www.pointcarbon.com.
RAE, 2004. Royal Academy of Engineering (RAE). The costs of generating electric-
ity. Available: www.raeng.org.uk.
Rashidi-Nejad, M., Song, Y., Javidi-Dasht-Bayaz, M., Jan. 2002. Operating reserve
provision in deregulated power markets. In: IEEE PES Winter Meeting. New
York, pp. 1305–1310.
REFIT, 2006. Renewable Energy Feed In Tariff (REFIT). A competi-

tion for electricity generation from biomass, hydro and wind. Available:
176
RICS, 2004. Royal Institute of Chartered Surveyors (RICS). Impact of wind

farms on the value of residential property and agricultural land. Available:
http://www.rics.org.
RISØ, EWEA, 2005. RISØ National Laboratory, Systems Analysis Department

Denmark and European Wind Energy Association (EWEA). Re-Xpansion work
package 2 - evaluation of renewable support schemes. Final report. Available:
http://www.ewea.org.
Ritsema, M., Edelman, M., Otto, D., 2003. Renewable power and energy efficiency:
Policies in Iowa and other states. a report for the Iowa policy project. Available:
http://www.iowapolicyproject.org.
Roach, J., 2004. By 2020 warming to doom million species. Available:

http://www.nationalgeographic.com.
SEI, 2003. Energy Policy Statistical Support Unit of Sustainable Energy Ireland
(SEI). Energy in ireland 1990 - 2002: Trends, issues and indicators. Available:
http://www.sei.ie.
SEI, 2004a. Renewable energy resources in ireland for 2010 and 2020 - a methodology.
Available: http://www.sei.ie.
SEI, 2004b. Sustainable Energy Ireland (SEI). Final report- economic analysis of
RES-E support mechanisms. Available: www.sei.ie.
SEI, 2004c. Sustainable Energy Ireland (SEI).A study on renewable energy in the
new irish electricity market. a report prepared for SEI by the Brattle Group Ltd.
Available: http://www.sei.ie.
SEM, 2005. The Single Electricity Market (SEM). High level design decision paper.
Available: www.allislandproject.org.
Sims, R., Rogner, H., Gregory, K., 2003. Carbon emissions and mitigation cost com-
parisons between fossil fuel, nuclear and renewable energy resources for electricity
generation. Energy Policy 31, 1315–1326.
177
Smith, J., 2006. Personal communication with Mark O’Malley regarding the north
american literature and project review on wind penetration studies.
Smith, J., DeMeo, E., Parsons, B., Milligan, M., 2004. Wind power impact on
electric power system costs: summary and perspective on work to date. In: Global
WINDPOWER Conference. Chicago, Illinois.
Söder, L., 1993. Reserve margin planning in a wind-hydro-thermal power system.

IEEE Transactions on Power Systems 8, 564–571.
SONI, 2005. System Operator Northern Ireland (SONI). Seven year generation ca-
pacity statement: 2005/06-2011/12. Available: http://www.nie.co.uk.
Stern, N., 2006. Stern review on the economics of climate change. Available:
http://www.hm-treasury.gov.uk.
The Gothenburg Protocol, 1999. to the 1979 convention on long range transbound-
ary air pollution. to abate acidification, eutrophication and ground-level ozone.
Available: www.unece.org.
UK DTI, 2004. UK Department of Trade and Industry. UK coal production prospects

for 2008-2016. Available: http://www.dti.gov.uk.
US DOE, 2005. United States Department of Energy (US DOE). Wind energy for
rural economic development. Available: http://www.nrel.gov.
US EPA, 2006a. United states environmental protection agency - global warming.

Available: http://www.epa.gov/.
US EPA, 2006b. United States Environmental Protection Agency - NOx State Imple-
mentation Plan. Evaluating ozone control programs in the eastern united states.
Available: http://www.epa.gov.
US EPA, 2006c. United States Environmental Protection Agency (USEPA).SO2

allowance market analysis. Available: http://www.epa.gov.
178
Viridian, 2006. Viridian Group PLC Website. Available:

http://www.viridiangroup.co.uk.
Voorspools, K., D’haeseleer, W., 2006. Modelling of electricity generation of large in-
terconnected power systems: How can a co2 tax influence the european generation
mix. Energy Conversion and Management 47, 1338–1358.
Vovos, P. N., Harrison, G., Wallace, A., Bialek, J., May 2005. Optimal power flow as
a tool for fault level-constrained network capacity analysis. IEEE Trans. Power
Syst. 20 (2), 734–741.
Wallace, A. R., Harrison, G., May 2003. Planning for optimal accommodation of dis-
persed generation in distribution networks. In: CIRED 17th International Con-
ference on Electricity Distribution. Barcelona.
Wang, P., Billington, R., May 2004. Reliability assessment of a restructured power
system considering reserve agreements. IEEE Transactions on Power Systems
19 (2), 972–978.
Watson, R., Landberg, L., 2003. Evaluation of prediktor wind power forecasting
system in ireland. In: European Wind Energy Association Conference. Madrid,
Spain.
We Energies, 2003. System operation impacts of wind generation integration -

etk2044. Available: http://www.uwig.org.
Wen, F., David, A. K., 2001. Optimal bidding strategies for competitive generators
and large consumers. Electrical power and energy systems 23, 37–43.
Whittaker, T. J. T., Fraenkel, P. L., Bell, A., Lugg, L., 2003. The potential for
the use of marine current energy in Northern Ireland. Department of Trade and
Industry, Department of Enterprise, Trade and Investment and Northern Ireland
Electricity study.
World Health Organisation, 2002. World Health Report. Available: www.who.int.
179
Xiang, D., Ran, L., Tavner, P. J., Yang, S., 2006. Control of doubly fed induction
generator in a wind turbine during grid fault ride-through. IEEE Transactions on
Energy Conversion 21 (3), 652–662.
Xu, J., Christie, R., 1999. Decentralised unit commitment in competitive energy
markets. in DIMAC Conference on Unit Commitment, Rutgers, New Jersey. Also
published as a chapter in Hobbs, Rothkopf, O’Neill and Chao, eds., The Next
Generation of Electric Power Unit Commitment Models, Kluwer.
Zavadil, R., King, J., Xiadong, L., Ahlstrom, M., Lee, B., Moon, D., Finley, C.,
Alnes, L., Jones, L., Hudry, F., Monstream, M., Lai, S., Smith, J., Sep. 2004.
Xcel energy and the minnesota department of commerce, wind integration study
- final report. EnerNex Corporation and Wind Logics, Inc.
Zhu, J., Jordan, G., Ihara, S., 2000. The market for spinning reserve and its impact
on energy prices. In: IEEE PES Winter Meeting. Singapore, pp. 1202–1207.
180
APPENDIX A
The Characteristics of Generators on the Irish System
This Appendix sets out the characteristics of the generators on the Irish system and
is based on the data set available at AIP (2005). Figure A.1 gives the generators
and their characteristics for the year 2007. The order of the generators in column
1 is the merit order in which they were dispatched. The energy prices in e/MWh
are based on the ‘mid’ fuel prices shown in Table 3.1. Figures A.2,A.3 and A.4 set
out the assumed plant mixes and generator characteristics for 2010, 2015 and 2020
respectively. Figure A.5 gives the no load and start up energy for each generator.
The following abbreviations are used to specify boiler type:
• SC - D refers to a ‘subcritical drum’ boiler
• HRSG refers to a ‘heat recovery steam generator’ boiler, also known as a

waste heat boiler
• FB - D refers to a ‘fluidised bed drum’ boiler
• OT - SC refers to a ‘once through subcritical’ boiler
181
Appendix A. Generator Information 182
Year of
Fuel Reserve Boiler
Unit Min Max GJ/MWh Euro/MWh Reserve Comm-
Type Mid Fuel Prices Euro/MWh Type
ission
Ardnacrusha 1 hydro 12 21 0 0.00 4 0.5 1929 NA
Erne 1 hydro 4 10 0 0.00 3 0.5 1950 NA
Erne 2 hydro 4 10 0 0.00 3 0.5 1950 NA
Erne 3 hydro 5 23 0 0.00 5 0.5 1950 NA
Erne 4 hydro 5 23 0 0.00 5 0.5 1950 NA
Lee 1 hydro 3 15 0 0.00 0 0.5 1957 NA
Lee 2 hydro 1 4 0 0.00 0 0.5 1957 NA
Lee 3 hydro 3 8 0 0.00 0 0.5 1957 NA
Liffey 1 hydro 3 15 0 0.00 1 0.5 1944 NA
Liffey 2 hydro 3 15 0 0.00 1 0.5 1944 NA
Liffey 3 hydro 0 4 0 0.00 0 0.5 1944 NA
Liffey 4 hydro 0 4 0 0.00 0 0.5 1944 NA
Moneypoint 3 coal 115 285 11.03 23.78 45 1.6 1985 SC-D
Kilroot coal 64 201 11.10 23.92 30 1.6 1981 SC-D
Kilroot coal 64 201 11.10 23.92 30 1.6 1981 SC-D
Dublin Bay Power gas 200 396 6.55 27.26 42 2.6 2002 HRSG
Coolkeeragh CCGT gas 260 404 6.60 27.47 40 2.6 2005 HRSG
Huntstown gas 223 343 6.65 27.69 25 2.6 2002 HRSG
Ballylumford CCGT gas 116 240 6.83 28.41 37 2.6 2002 HRSG
Poolbeg CCGT gas 280 480 6.99 29.10 150 2.6 2000 HRSG
West Offaly Power peat 46 137 9.86 30.94 9 2 2004 FB-D
Lough Ree Power peat 40 91 10.07 31.61 12 2 2004 FB-D
North Wall 4 gas 112 163 9.60 39.96 32 2.6 1982 HRSG
Marina gas 77 112 9.71 40.41 35 2.6 1979 HRSG
Edenderry peat 40 118 13.07 41.01 9 2 2000 FB
Aghada gas 35 258 10.00 41.62 20 1.8 1981 OT-SC
Tarbert 4 oil 35 241 10.70 44.19 25 1.8 1977 OT-SC
Tarbert 3 oil 35 241 10.90 45.02 25 1.8 1977 OT-SC
Poolbeg 3 gas 56 242 11.00 45.78 20 1.7 1981 OT-SC
Poolbeg 1 gas 55 110 11.15 46.41 8 1.7 1971 OT-SC
Ballylumford 5 gas 56 170 11.20 46.61 35 1.7 1968 OT-SC
Poolbeg 2 gas 55 110 11.30 47.03 8 1.7 1971 OT-SC
Gt. Island 3 oil 30 108 11.70 48.32 20 1.7 1971 SC-D
Aghada OCGT gas 10 88 12.15 50.57 20 1 1982 NA
Tarbert 1 oil 25 54 12.84 53.02 3 1.7 1969 SC-D
Tarbert 2 oil 25 54 12.86 53.11 3 1.7 1969 SC-D
Gt. Island 1 oil 25 54 12.87 53.15 3 1.7 1967 SC-D
Gt. Island 2 oil 25 54 12.91 53.32 3 1.7 1967 SC-D
Aghada distillate 5 52 14.12 59.63 14 1 1981 NA
Rhode 1 distillate 5 52 10.63 83.92 20 1 unknown NA
Asahi distillate 5 52 10.63 84.90 0 1 unknown NA
Aghada CT distillate 10 90 12.14 97.16 20 1 1982 NA
North Wall 5 distillate 5 108 12.90 101.41 72 1 1983 NA
Coolkeeragh distillate 8 53 14.20 112.72 16 1 1959 NA
Kilroot distillate 5 29 14.35 113.96 7 1 1982 NA
Ballylumford 2 distillate 8 53 14.45 115.73 15 1 1968 NA
Figure A.1: Generator Information for 2007

Year of
Fuel Reserve Boiler
ission
Erne 1 hydro 4 10 0 0.00 3 0.5 1950 NA
Erne 2 hydro 4 10 0 0.00 3 0.5 1950 NA
Erne 3 hydro 5 23 0 0.00 5 0.5 1950 NA
Erne 4 hydro 5 23 0 0.00 5 0.5 1950 NA
Lee 1 hydro 3 15 0 0.00 0 0.5 1957 NA
Lee 2 hydro 1 4 0 0.00 0 0.5 1957 NA
Lee 3 hydro 3 8 0 0.00 0 0.5 1957 NA
Liffey 1 hydro 3 15 0 0.00 1 0.5 1944 NA
Liffey 2 hydro 3 15 0 0.00 1 0.5 1944 NA
Liffey 3 hydro 0 4 0 0.00 0 0.5 1944 NA
Liffey 4 hydro 0 4 0 0.00 0 0.5 1944 NA
Kilroot coal 64 201 11.10 24.09 30 1.6 1981 SC-D
Kilroot coal 64 201 11.10 24.09 30 1.6 1981 SC-D
Tynagh CCGT gas 202 404 6.57 28.71 32 2.6 2006 HRSG
Huntstown 2 gas 200 401 7.00 30.59 50 2.6 2007 HRSG
Aughinish CCGT gas 40 150 8.72 38.12 10 2.6 2006 HRSG
Edenderry peat 40 118 13.07 42.22 9 2 2000 FB
Marina gas 77 112 9.71 42.43 35 2.6 1979 HRSG
Aghada gas 35 258 10.00 43.70 20 1.8 1981 OT-SC
Tarbert 4 oil 35 241 10.70 45.69 25 1.8 1977 OT-SC
Tarbert 3 oil 35 241 10.90 46.54 25 1.8 1977 OT-SC
Poolbeg 3 gas 56 242 11.00 46.97 20 1.7 1981 OT-SC
Poolbeg 1 gas 55 110 11.15 47.61 8 1.7 1971 OT-SC
Poolbeg 2 gas 55 110 11.30 48.25 8 1.7 1971 OT-SC
Gt. Island 3 oil 30 108 11.70 49.96 20 1.7 1971 SC-D
Aghada OCGT gas 10 88 12.15 53.10 20 1 1982 NA
Tarbert 1 oil 25 54 12.84 54.82 3 1.7 1969 SC-D
Tarbert 2 oil 25 54 12.86 54.91 3 1.7 1969 SC-D
Gt. Island 1 oil 25 54 12.87 54.95 3 1.7 1967 SC-D
Gt. Island 2 oil 25 54 12.91 55.12 3 1.7 1967 SC-D
Aghada distillate 5 52 14.12 61.70 14 1 1981 NA
Asahi distillate 5 52 10.63 85.93 0 1 unknown NA
Coolkeeragh distillate 8 53 14.20 114.72 16 1 1959 NA

Year of
Fuel Reserve Boiler
ission
Erne 1 hydro 4 10 0 0.00 3 0.5 1950 NA
Erne 2 hydro 4 10 0 0.00 3 0.5 1950 NA
Erne 3 hydro 5 23 0 0.00 5 0.5 1950 NA
Erne 4 hydro 5 23 0 0.00 5 0.5 1950 NA
Lee 1 hydro 3 15 0 0.00 0 0.5 1957 NA
Lee 2 hydro 1 4 0 0.00 0 0.5 1957 NA
Lee 3 hydro 3 8 0 0.00 0 0.5 1957 NA
Liffey 1 hydro 3 15 0 0.00 1 0.5 1944 NA
Liffey 2 hydro 3 15 0 0.00 1 0.5 1944 NA
Liffey 3 hydro 0 4 0 0.00 0 0.5 1944 NA
Liffey 4 hydro 0 4 0 0.00 0 0.5 1944 NA
Kilroot coal 64 201 11.10 24.64 30 1.6 1981 SC-D
Kilroot coal 64 201 11.10 24.64 30 1.6 1981 SC-D
New CCGT (1) gas 200 450 6.71 33.82 20 1 2012 HRSG
New CCGT (2) gas 200 420 6.72 33.87 20 1 2012 HRSG
Ballylumford CCGT gas 116 240 6.83 34.40 37.1 2.6 2002 HRSG
Edenderry peat 40 118 13.07 46.14 9.4 2 2000 FB
Marina gas 77 112 9.71 48.94 35 2.6 1979 HRSG
Tarbert 4 oil 35 241 10.70 50.40 25 1.8 1977 OT-SC
Aghada gas 35 258 10.00 50.40 20 1.8 1981 OT-SC
Tarbert 3 oil 35 241 10.90 51.34 25 1.8 1977 OT-SC
Gt. Island 3 oil 30 108 11.70 55.11 20 1.7 1971 SC-D
Poolbeg 3 gas 56 242 11.00 55.44 20 1.7 1981 OT-SC
Poolbeg 1 gas 55 110 11.15 56.20 8 1.7 1971 OT-SC
Poolbeg 2 gas 55 110 11.30 56.95 8 1.7 1971 OT-SC
Tarbert 1 oil 25 54 12.84 60.47 3 1.7 1969 SC-D
Tarbert 2 oil 25 54 12.86 60.57 3 1.7 1969 SC-D
Aghada gas 5 52 14.12 71.16 14 1 1982 NA
New OCGT gas 10 160 10.63 94.86 15.5 1 2012 NA
Kilroot distillate 5 29 14.35 128.01 7.25 1 1982 NA
Ballylumford 2 distillate 8 53 14.45 128.86 14.5 1 1968 NA

Year of
Fuel Reserve Boiler
ission
Ardnacrusha 1 hydro 11.9 21 0 0.00 4 0.5 1929 NA
Erne 1 hydro 4 10 0 0.00 3 0.5 1950 NA
Erne 2 hydro 4 10 0 0.00 3 0.5 1950 NA
Erne 3 hydro 5 22.5 0 0.00 5 0.5 1950 NA
Erne 4 hydro 5 22.5 0 0.00 5 0.5 1950 NA
Lee 1 hydro 3 15 0 0.00 0 0.5 1957 NA
Lee 2 hydro 1 4 0 0.00 0 0.5 1957 NA
Lee 3 hydro 3 8 0 0.00 0 0.5 1957 NA
Liffey 1 hydro 3 15 0 0.00 1 0.5 1944 NA
Liffey 2 hydro 3 15 0 0.00 1 0.5 1944 NA
Liffey 3 hydro 0.4 4 0 0.00 0 0.5 1944 NA
Liffey 4 hydro 0.2 4 0 0.00 0 0.5 1944 NA
Kilroot coal 64 201 11.10 25.64 30 1.6 1981 SC-D
Kilroot coal 64 201 11.10 25.64 30 1.6 1981 SC-D
Huntstown gas 222.8 342.7 6.65 42.64 25 2.6 2002 HRSG
CCGT (1) gas 200 450 6.71 43.01 20 1 2012 HRSG
CCGT (2) gas 200 420 6.72 43.08 20 1 2012 HRSG
New CCGT or similar gas 220 400 6.79 43.52 3 1.7 2018 HRSG
New CCGT or similar gas 230 400 6.79 43.52 3 1.7 2018 HRSG
Edenderry peat 40 117.6 13.07 54.24 9.4 2 2000 FB
Tarbert 4 oil 35 241 10.70 60.13 25 1.8 1977 OT-SC
Tarbert 3 oil 35 241 10.90 61.26 25 1.8 1981 OT-SC
New CCGT gas 63 103 9.70 62.18 8 2.6 2013 HRSG
Marina gas 77 112 9.71 62.24 35 2.6 1979 HRSG
Aghada gas 35 258 10.00 64.10 20 1.8 1981 OT-SC
Gt. Island 3 oil 30 108 11.70 65.75 20 1.7 1971 SC-D
OCGT gas 10 160 10.63 94.86 15.5 1 2012 NA
New OCGT gas 10 80 10.63 68.17 15.5 1 2018 NA
Poolbeg 3 gas 56 242 11.00 70.51 20 1.7 1981 OT-SC
Poolbeg 1 gas 55 110 11.15 71.47 8 1.7 1971 OT-SC
Poolbeg 2 gas 55 110 11.30 72.43 8 1.7 1971 OT-SC
New OCGT gas 15 52 14.12 90.51 0 0 2018 NA
Aghada gas 5 52 14.12 90.51 14 1 1982 NA

No Load Cold Start Warm Start Cold Start

Unit
(GJ/hour) GJ GJ GJ
Moneypoint 3 148 14620 6920 4360
Moneypoint 1 148 14620 6920 4360
Moneypoint 2 148 14620 6920 4360
Kilroot 213 2247 1645 973
Kilroot 213 2247 1645 973
Dublin Bay Power 533 7700 2600 2600
Tynagh CCGT 467 2811 1633 1144
Coolkeeragh CCGT 496 5000 2000 1000
Huntstown 350 1000 650 500
Ballylumford CCGT 496 50 50 50
Ballylumford CCGT 496 50 50 50
Poolbeg CCGT 716 100 100 100
West Offaly Power 124 500 500 500
Lough Ree Power 90 320 320 320
Aughinish CCGT 506 2900 1700 1150
North Wall 4 347 80 80 80
Marina 251 50 50 50
Edenderry 498 2010 1084 436
Aghada 187 4302 2185 1273
Tarbert 4 258 3180 1934 1072
Tarbert 3 258 3180 1934 1072
Poolbeg 3 274 4302 2185 1273
Poolbeg 1 84 1025 625 353
Ballylumford 5 179 2124 1527 847
Poolbeg 2 88 1025 625 353
Ballylumford 6 179 2124 1527 847
Gt. Island 3 105 743 600 293
Ballylumford 1 98 50 50 50
Aghada OCGT 87 50 50 50
Tarbert 1 50 562 449 218
Tarbert 2 50 562 449 218
Gt. Island 1 51 562 449 218
Gt. Island 2 51 562 449 218
Aghada 187 4302 2185 1273
Rhode 1 85 50 50 50
Rhode 2 85 50 50 50
Asahi 85 50 50 50
Aghada CT 268 63 63 63
Aghada CT 268 63 63 63
North Wall 5 317 50 50 50
Coolkeeragh 177 16 16 16
Kilroot 101 8 8 8
Kilroot 101 8 8 8
Figure A.5: Generator Operating Characteristics

APPENDIX B
The Optimisation of the Distribution Network
This Appendix sets out the methodology employed in Keane and O’Malley (2005b,
2006) and Keane et al. (2006) to optimise the generation capacity on the distribution
network.
Objective Function
The optimisation problem is formulated as a linear program, with the amount of

constraint breaches that arise taken into account. Load factors express the energy
output of a generator as a fraction of the maximum possible energy output that is
produced by a generator in a year and these are included in the objective function.
This ensures that the available capacity is allocated based on the amount of energy
that is delivered. The objective function is given in Equation (B.1).
M X
N
X PAvail j P lantij LFj
J= (B.1)
ConnCostij νi
j=1 i=1
Where PAvail j is the jth available energy resource. P lantij are the control variables
representing the fraction of PAvail j allocated to the ith bus. LFj is the load factor
187
Appendix B. Optimisation of the Distribution Network 188
of the jth energy resource. νi gives the total voltage sensitivity of the ith bus to
power injections at all other buses. ConnCostij is the connection costs of the jth
energy resource at the ith bus. M & N are the number of energy resources and
buses respectively. The geographical dispersion of DG impacts the connection costs.
The connection costs of each resource to each bus are determined and are minimised
in the objective function. The objective function is maximised with respect to
the technical constraints described below. The thermal constraint, the transformer
rating, the short circuit level and the short circuit ratio constraints are the same in
both the firm and non firm cases and are shown in Equations (B.2), (B.3), (B.5)
and (B.6) (Keane et al., 2006). The only difference between the firm and non firm
case is the voltage constraint (B.9), which is not included in the non firm scenario.
Thermal Constraint
This is a stand alone constraint, which ensures that the rated current of the lines
must not be exceeded. It is given by Equation (B.2).
Ii < IiRated i ∀ N. (B.2)
Where Ii is the current flowing from generator i to bus i and IiRated is the
maximum rated current for the line between each generator and its corresponding
bus. Under standard voltage and power factor conditions the rated current of the
line can be translated directly into a rated active power for that line.
Transformer Capacity
The amount of generation connected minus the summer valley load must not ex-
ceed the rating of the transformer at the higher voltage. If there is some existing
generation then this must be subtracted from the total. The result is the remaining
capacity available below that station. In the case of two parallel transformers, the
capacity is taken as the rating of the smaller transformer plus the summer valley
load. The constraint is expressed formally as in Equation (B.3).
PT x ≤ PT raf oCap . (B.3)
Where PT x refers to power flow through the transmission substation transformer

and PT raf oCap refers to the rating of that transformer.
Short Circuit Level
A maximum short circuit rating for all equipment is laid down in the distribution
code (ESB Networks, 2002). A short circuit calculation is carried out to ensure that
this constraint is not exceeded as the level of installed capacity increases. The short
circuit level (SCL) is highest at the transmission system bus. Buses close to this
bus may find their capacity limited as a result. The constraint is given by Equation
(B.4).
SCLT x < SCLRated . (B.4)
Where SCLT x is the short circuit level at the transmission substation busbar and
SCLRated is the highest current that switchgear can safely break under fault condi-
tions. The contribution of increasing levels of generation at each bus to SCLT x is
determined by short circuit analysis. The SCL contributions of generation at each
bus are combined and formalised into an algebraic equation as shown in Equation
(B.5).
N
X
δjT x PDG j + αT x ≤ SCLRated . (B.5)
j=1
Where δjT x is the dependency of the SCL at the transmission station to power
injections at bus j, i.e. the slope of the SCL vs. power injection characteristic of
the jth bus. PDG j is the power injection at the jth bus, αT x is the initial SCL at
the transmission bus with no generation present.
Short Circuit Ratio
The short circuit ratio (SCR) is the ratio of generator power PDG (MW) at each bus
to the short circuit level at each bus SCLBus (MVA). It gives an indication of the
voltage dip experienced near the generation in the event of a feeder outage and is
shown in Equation (B.6).
N
X
PDG i − 0.1 cos(φ) δji PDG j ≤ 0.1 cos(φ)αi . (B.6)
j=1
Where cos(φ) is the power factor at the generator. For this constraint the short
circuit level used at the transmission station is the summer night valley level.
Voltage Rise Effect
The voltage at the generator is given by Equation (B.7).
RPL + XQL XPL − RQL

⇒ VG = VL + +j (B.7)
VL VL
Where PDG and QDG are the active and reactive power of the generator respec-
tively, Z=R+jX is the impedance of the line, PL and QL are active and reactive
power at the bus and VG and VL are the voltages at the generator and bus respec-
tively. Thus it can be seen that the generator voltage will be the load/bus voltage
plus some value related to the impedance of the line and the power flows along that
line. It is evident that the larger the impedance and power flow the larger the volt-
age rise. The increased active power flows on the distribution network have a large
impact on the voltage level because the resistive element of the lines on distribution
networks are higher than other lines. This leads to an X/R ratio of approximately
1 rather than a more typical value of 5 on transmission networks. The voltage must
be kept within standard limits at each bus as given by Equation (B.8).
Vmin i < Vi < Vmax i i ∀ N. (B.8)
Where Vmin i & Vmax i refer to the minimum and maximum voltage limits at the
ith bus. The relationship between voltage and power injections at each bus is de-
termined. As MWs are added at each bus the voltage rises. Increasing levels of
generation are added incrementally at each bus in turn and load flow analysis is
carried out to determine a voltage vs. active power characteristic for each bus. Next
the interdependence of the bus voltage levels is examined. Once again increasing
levels of generation are added incrementally at each bus, but now the voltage level
at every other bus is examined. Thus characteristics are determined for voltage
levels at each bus due to generation at all other buses. By combination of these
characteristics the voltage constraint may be formalised into algebraic equations for
each bus as shown in Equation (B.9).
N
X
µi PDG i + βi + µji PDG j ≤ Vmax i i ∀ N, i 6= j. (B.9)
j=1
Where µi is the dependency of the voltage level at bus i on power injections at bus
i, i.e. the slope of the voltage vs. power injection characteristic of the ith bus. βi
refers to the initial voltage level at the ith bus with no generation, µji refers to the
dependency of the voltage level at bus i on power injections at bus j. This analysis
is carried out under minimum load conditions as this is the worst case scenario for
voltage rise. Both the standby and normal forward feed conditions are considered.
There is usually more than one possible standby feeding arrangement, but the most
severe feeding condition is usually readily identifiable.
Non Firm Constraint Management
Voltage rise is one of the dominant technical constraints on DG. The assessment of
the voltage constraint at a N-1 peaking condition can present a significant barrier
to further penetration of DG. While infrequent, it is important for the operation
of the system that the voltage stays within its limits. A number of voltage control
techniques have been proposed to mitigate the voltage rise effect (Hird et al., 2004;
Hill et al., 2005). These techniques along with others facilitate non firm management
of the voltage constraint. The short circuit level may be dominant in more urban
areas. However, active management of fault levels is some way off and is likely to
be very expensive (KEMA, 2005). Hence, it is non firm management of the voltage
constraint which is considered here.

Eleanor Denny - A Cost-Benefit Analysis of Wind Power PDF

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Eleanor Denny - A Cost-Benefit Analysis of Wind Power PDF

Uploaded by

Copyright:

Available Formats

A Cost Benefit Analysis

Thesis submitted for the degree of

Supervisor of Research: Prof. M. O’Malley

January 19, 2007

1. E. Denny, L. Malaguzzi Valeri, J. Fitz Gerald & M. O’Malley, “Carbon Prices

3. A. Keane, E. Denny, and M. OMalley, Quantifying the Impact of Connection

4. E. Denny & M. O’Malley, “Wind Generation, Power System Operation and

5. M. Bazilian, E. Denny, & M. O’Malley, “Challenges of Increased Wind Energy

6. A. Keane, Q. Zhou, E. Denny, J. Bialek, and M. O’Malley, ”A Novel Curtail-

7. E. Denny, & M. O’Malley, “A Cost Benefit Analysis of Tidal Generation”, in

8. E. Denny, G. Bryans, J. Fitz Gerald & M. O’Malley, “A Quantitative Analysis

9. E. Denny, & M. O’Malley, “Impact of Wind Generation on Emissions under

10. E. Denny & M. O’Malley, “Impact of Increasing levels of Wind Generation in

12. R. Doherty, E. Denny, & M. O’Malley, “System Operation with a Significant

Publications Arising from Thesis v

List of Figures xii

2 The Case Study 23

2.3.1 Dominance in the Irish Electricity Market . . . . . . . . . . . 31

3 The Dispatch Model 41

4 The Cost of Wind Generation 56

5 The Benefits of Wind Generation 72

6 The Net Benefits of Wind Generation 88

7 Applications of the Net Benefits Methodology 103

7.1.3 Summary of Carbon Prices and Cycling Costs . . . . . . . . . 114

8 Discussion and Conclusions 149

A The Characteristics of Generators on the Irish System 181

B The Optimisation of the Distribution Network 187

1.1 Net benefits of wind generation . . . . . . . . . . . . . . . . . . . . . 19

2.1 Existing wind farms in RoI in January 2007 . . . . . . . . . . . . . . 26

3.1 Generator reserve characteristics . . . . . . . . . . . . . . . . . . . . 44

4.1 Assumed capital cost per MW of installed wind generation . . . . . 58

5.1 The capacity credit of wind generation in Ireland . . . . . . . . . . . 74

6.1 The base case net benefits of wind generation . . . . . . . . . . . . . 90

7.1 Merit order with carbon price of e30/ton . . . . . . . . . . . . . . . 108

A.1 Generator Information for 2007 . . . . . . . . . . . . . . . . . . . . . 182

2.1 The Irish electricity system (at year end 2005) . . . . . . . . . . . . 24

3.1 Fuel costs in e 2006/GJ . . . . . . . . . . . . . . . . . . . . . . . . . 53

4.1 Costs associated with increased cycling of units . . . . . . . . . . . . 67

5.1 Typical components of various fuel types . . . . . . . . . . . . . . . . 77

6.1 Base case assumptions summarised . . . . . . . . . . . . . . . . . . . 90

7.6 Additional cycling costs with wind generation . . . . . . . . . . . . . 136

to occur as forests die and up to 50 million species could be driven to extinction by

1.1 Wind Generation

example, the amount of electricity generated by a wind turbine fluctuates as wind

can become more challenging (Gross et al., 2006).

1.2 The Impacts of Wind Generation

an increase in the capacity of wind generation on an electricity system increases the

result of international politics (Ahmed and Cohen, 1994).

technology and a country’s commitment to renewable generation. Wind generation

1.3 Quantifying the Costs & Benefits of Wind

and any additional reserve that may be required.

would result in a saving of 620-700gCO2 /kWh by displacing mainly coal units in

1.4 The Scope of this Thesis

Figure 1.1: Net benefits of wind generation

1.5 An Outline of the Thesis

The Case Study

RoI NI All Island

are relatively easier to identify.

2.1 Wind Generation in Ireland

Wind generation currently represents approximately 8% of the total installed gen-

Figure 2.1: Existing wind farms in RoI in January 2007

Technology Size Tariff (per MWh)