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. OMalley Co-Supervisor of Research: Prof. J. Fitz Gerald Nominating Professor: Prof. M. OMalley

January 19, 2007

Abstract

Concern over global warming has led policy makers worldwide to accept the importance of reducing greenhouse gas emissions. As wind generation does not itself create any harmful emissions, policy makers often promote it as a means of reducing 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 underlying 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 directly 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 generation 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 dicult to predict accurately, this balancing of generation and load can become more challenging as renewable pene-

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 benets 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 benets are the capacity benet of wind, the emissions savings and the savings in fuel costs. In order to investigate the costs and benets 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 winds impact on the operation of conventional generators and their resulting cycling costs, emissions and fuel costs. The costs and benets are combined to generate net benet curves which show the wind penetrations where the costs exceed the benets, beyond which no further investment should be made in wind generation. Three test years are examined and the sensitivity of the net benets to a large range of assumptions is tested. It is found that increased interconnection, high CO2 prices and a exible plant mix are particularly benecial for wind generation, and that there are positive net benets 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 analysis 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 eect 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 connection 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. Marks 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 eort 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 OMahony from ESB Power Generation, Donal Phelan in ESB Networks, Michael Kelly in EirGrid, Morgan Bazillian in SEI and Dermot OKane 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, encouragement, 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! Youve been great, and at last I might have an answer to the dreaded question so when will you be nished? ! My fantastic family, Michael, Frances, Conor and Cl odhna. Thank you for your constant support and encouragement, not just over the last four years but in everything 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 nally 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. OMalley, 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. OMalley, Quantifying the Total Net Benets of Grid Integrated 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. OMalley, 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. OMalley, 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. OMalley, A Novel Curtailment Method for Distributed Generation Voltage Management, IEEE Transactions on Power Systems, in preparation, 2007. v

vi Conference Papers: 7. E. Denny, & M. OMalley, A Cost Benet 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. OMalley, A Quantitative Analysis of the Net Benets of Grid Integrated Wind, in IEEE Power Engineering Society General Meeting (Panel Session), Montreal, Quebec, June 2006. 9. E. Denny, & M. OMalley, 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. OMalley, 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. OMalley, Study of the Eect 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. OMalley, System Operation with a Signicant Wind Power Penetration, in IEEE Power Engineering Society General Meeting, Denver, Colorado, USA, June 2004.

Contents

Abstract Acknowledgements Publications Arising from Thesis List of Figures List of Tables 1 Introduction 1.1 1.2 1.3 1.4 1.5 Wind Generation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The Impacts of Wind Generation . . . . . . . . . . . . . . . . . . . . Quantifying the Costs & Benets of Wind . . . . . . . . . . . . . . . The Scope of this Thesis . . . . . . . . . . . . . . . . . . . . . . . . . An Outline of the Thesis . . . . . . . . . . . . . . . . . . . . . . . . .

i iii v xii xv 1 4 6 12 18 21 23 25 28 30

2 The Case Study 2.1 2.2 2.3 Wind Generation in Ireland . . . . . . . . . . . . . . . . . . . . . . . Irelands Electricity Demand . . . . . . . . . . . . . . . . . . . . . . Irelands Electricity Market . . . . . . . . . . . . . . . . . . . . . . . vii

CONTENTS 2.3.1 2.4 2.5 2.6 Dominance in the Irish Electricity Market . . . . . . . . . . .

viii 31 33 37 38 41 42 44 45 47 48 48 49 51 51 52 52 52 53 53 54 54 55 56 57 59 60 62 66 70

Irelands Current Generation Plant Mix . . . . . . . . . . . . . . . . The Future Plant Mix of the Irish System . . . . . . . . . . . . . . . Summary of the Key Features of the Case Study . . . . . . . . . . .

3 The Dispatch Model 3.1 Optimisation in the Dispatch Model . . . . . . . . . . . . . . . . . . 3.1.1 3.2 Validating the Dispatch Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Wind in the Dispatch Model 3.2.1 3.2.2 3.2.3

Wind Curtailment . . . . . . . . . . . . . . . . . . . . . . . . Wind Shortfall . . . . . . . . . . . . . . . . . . . . . . . . . . Load Factors for the Wind Generation . . . . . . . . . . . . .

3.3 3.4

The Dispatch Model Representing an Electricity Market . . . . . . . Specics of the Dispatch Model for Case Study . . . . . . . . . . . . 3.4.1 3.4.2 3.4.3 3.4.4 3.4.5 3.4.6 3.4.7 3.4.8 Reserve Capacity . . . . . . . . . . . . . . . . . . . . . . . . . Hydro and Pumped Storage . . . . . . . . . . . . . . . . . . . Outage Schedules . . . . . . . . . . . . . . . . . . . . . . . . . Demand . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Fuel Prices . . . . . . . . . . . . . . . . . . . . . . . . . . . . Peat Generation . . . . . . . . . . . . . . . . . . . . . . . . . Wind Capacity . . . . . . . . . . . . . . . . . . . . . . . . . . Network Constraints . . . . . . . . . . . . . . . . . . . . . . .

3.5

Summary of Key Features of the Dispatch Model . . . . . . . . . . .

4 The Cost of Wind Generation 4.1 4.2 4.3 4.4 Wind Development Costs . . . . . . . . . . . . . . . . . . . . . . . . Network Upgrade Costs . . . . . . . . . . . . . . . . . . . . . . . . . Additional Reserve Requirement . . . . . . . . . . . . . . . . . . . . Cycling Conventional Units . . . . . . . . . . . . . . . . . . . . . . . 4.4.1 4.5 The Cost of Cycling . . . . . . . . . . . . . . . . . . . . . . .

Summary of Costs Associated with Wind Generation . . . . . . . . .

CONTENTS 5 The Benets of Wind Generation 5.1 5.2 Capacity Benet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Emissions in Conventional Generators . . . . . . . . . . . . . . . . . 5.2.1 5.2.2 5.2.3 5.2.4 5.2.5 5.2.6 5.3 5.4 5.5 Carbon Dioxide Formation . . . . . . . . . . . . . . . . . . . Sulphur Dioxide Formation . . . . . . . . . . . . . . . . . . . Nitrogen Oxides Formation . . . . . . . . . . . . . . . . . . . Coal, Peat and Heavy Fuel Oil Generators . . . . . . . . . . . Gas Fired Generators . . . . . . . . . . . . . . . . . . . . . . Emissions During Cycling . . . . . . . . . . . . . . . . . . . .

ix 72 73 75 75 76 76 77 78 81 81 84 86 88 88 91 94 95 96 97 97 98 99 99

Emissions Benets of Wind Generation . . . . . . . . . . . . . . . . . Fuel Savings Benet . . . . . . . . . . . . . . . . . . . . . . . . . . . Summary of Benets of Wind Generation . . . . . . . . . . . . . . .

6 The Net Benets of Wind Generation 6.1 6.2 The Net Benets of Wind Generation in the Base Case . . . . . . . . Sensitivity Analysis of Net Benets . . . . . . . . . . . . . . . . . . . 6.2.1 6.2.2 6.2.3 6.2.4 6.2.5 6.2.6 6.2.7 6.2.8 6.2.9 6.3 Wind Load Factor Scenario . . . . . . . . . . . . . . . . . . . Wind Power Forecast Accuracy Scenario . . . . . . . . . . . . Demand Growth Scenario . . . . . . . . . . . . . . . . . . . . Fuel Price Scenario . . . . . . . . . . . . . . . . . . . . . . . . Capital Cost Scenario . . . . . . . . . . . . . . . . . . . . . . Cycling Cost Scenario . . . . . . . . . . . . . . . . . . . . . . Emissions Price Scenario . . . . . . . . . . . . . . . . . . . . . Discount Rate Scenario . . . . . . . . . . . . . . . . . . . . .

Worst and Best Case Scenarios . . . . . . . . . . . . . . . . . 100

Conclusions of Net Benets . . . . . . . . . . . . . . . . . . . . . . . 101 103

7 Applications of the Net Benets Methodology 7.1

Carbon Prices and Cycling Costs . . . . . . . . . . . . . . . . . . . . 104 7.1.1 7.1.2 Emissions Prices in the Cycling Analysis . . . . . . . . . . . . 107 Results & Discussion . . . . . . . . . . . . . . . . . . . . . . . 108

CONTENTS 7.1.3 7.2

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

The Fuelsaver Approach to Wind Generation . . . . . . . . . . . . . 115 7.2.1 7.2.2 7.2.3 7.2.4 7.2.5 7.2.6 Analysing the Fuelsaver Approach . . . . . . . . . . . . . . . 116 Fuelsaver Cycling Costs . . . . . . . . . . . . . . . . . . . . . 120 Fuelsaver Emissions Benets . . . . . . . . . . . . . . . . . . 120 Fuelsaver Fuel Benets . . . . . . . . . . . . . . . . . . . . . . 122 Net Benets of Wind under the Fuelsaver Approach . . . . . 123 Conclusions of Fuelsaver Approach . . . . . . . . . . . . . . . 124

7.3

The Impact of Connection Policy on Distributed Generation . . . . . 125 7.3.1 7.3.2 7.3.3 7.3.4 7.3.5 7.3.6 7.3.7 7.3.8 7.3.9 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . 127 Irelands Distribution Network . . . . . . . . . . . . . . . . . 128 The Energy Resources . . . . . . . . . . . . . . . . . . . . . . 130 Optimal Allocations of Distributed Generation . . . . . . . . 131 Load Flow Simulation . . . . . . . . . . . . . . . . . . . . . . 132 Scaling Outputs . . . . . . . . . . . . . . . . . . . . . . . . . 133 Distributed Generation in the Dispatch Model . . . . . . . . 134 The Costs and Benets of Distributed Generation . . . . . . 135 Net Benets . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137

7.3.10 Conclusion of Connection Policy . . . . . . . . . . . . . . . . 138 7.4 Tidal Generation on the Irish System . . . . . . . . . . . . . . . . . . 139 7.4.1 7.4.2 7.4.3 7.4.4 7.4.5 7.4.6 Modelling the Tidal Generation . . . . . . . . . . . . . . . . . 141 Tidal Cycling Costs . . . . . . . . . . . . . . . . . . . . . . . 142 Tidal Emissions Benets . . . . . . . . . . . . . . . . . . . . . 143 Fuel Savings with Tidal Generation . . . . . . . . . . . . . . 144 The Net Benets of Tidal Generation . . . . . . . . . . . . . 144 Conclusions of Tidal Generation in Ireland . . . . . . . . . . 147 149

8 Discussion and Conclusions 8.1 8.2 8.3

Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155 Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158

CONTENTS References A The Characteristics of Generators on the Irish System B The Optimisation of the Distribution Network

xi 160 181 187

List of Figures

1.1 2.1 2.2 2.3 2.4 2.5 2.6 2.7 3.1 3.2 3.3 3.4 4.1 4.2 4.3 4.4 4.5

Net benets of wind generation . . . . . . . . . . . . . . . . . . . . . Existing wind farms in RoI in January 2007 . . . . . . . . . . . . . . Comparison of electricity consumption in the EU15 . . . . . . . . . . Typical daily demand patterns in RoI in 2006 . . . . . . . . . . . . . Installed generation capacity by fuel type in Ireland . . . . . . . . . The age of the generating units on the Irish system . . . . . . . . . . Load duration curve with merit order of conventional units . . . . . Conventional generation availability in Ireland . . . . . . . . . . . .

19 26 29 29 33 34 34 36 44 47 49 50 58 60 61 64 65

Generator reserve characteristics . . . . . . . . . . . . . . . . . . . . Wind power data series . . . . . . . . . . . . . . . . . . . . . . . . . Assumed load factors with increasing installed wind generation . . . Illustration of market model . . . . . . . . . . . . . . . . . . . . . . . Assumed capital cost per MW of installed wind generation . . . . .

Network reinforcement costs with increased wind capacity . . . . . . System reserve level with dierent wind power forecast errors . . . . Damage associated with dierent cycling activities . . . . . . . . . . Creep fatigue interaction resulting in premature component failure . xii

LIST OF FIGURES 4.6 4.7 5.1 5.2 5.3 5.4 5.5 5.6 6.1 6.2 6.3 6.4 7.1 7.2 7.3 7.4 7.5 7.6 7.7 7.8 7.9 Operating proles of two Irish units with dierent wind penetrations Cycling costs with increasing penetrations of wind generation . . . . The capacity credit of wind generation in Ireland . . . . . . . . . . . Typical NOX emissions from a CCGT and an OCGT . . . . . . . . . Emissions savings with increasing wind generation . . . . . . . . . . Value of emissions savings with increasing wind generation . . . . . . Fuel consumption with increasing wind generation . . . . . . . . . . Value of fuel savings with increasing wind generation . . . . . . . . . The base case net benets of wind generation . . . . . . . . . . . . . The Critical Values of Wind Generation Part I . . . . . . . . . . . . The Critical Values of Wind Generation Part II . . . . . . . . . . . . The net benets of wind with changes in load factor . . . . . . . . .

xiii 66 70 74 80 82 84 85 86 90 92 93 94

Merit order with carbon price of e30/ton . . . . . . . . . . . . . . . 108 Cycling costs and CO2 savings at dierent carbon prices . . . . . . . 109 Cycling costs and CO2 savings with wind and e30/ton CO2 . . . . . 111 Cycling costs with wind generation and carbon prices . . . . . . . . 112 Carbon savings with wind generation and carbon prices . . . . . . . 113 Net savings with wind generation and carbon prices . . . . . . . . . 113 Wind curtailment in 2010 under fuelsaver . . . . . . . . . . . . . . . 118 Average number of committed units under fuelsaver . . . . . . . . . 118 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 benets 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 benets and cycling costs of tidal generation . . . . . . . . 146 A.1 Generator Information for 2007 . . . . . . . . . . . . . . . . . . . . . 182 A.2 Generator Information for 2010 . . . . . . . . . . . . . . . . . . . . . 183 A.3 Generator Information for 2015 . . . . . . . . . . . . . . . . . . . . . 184 A.4 Generator Information for 2020 . . . . . . . . . . . . . . . . . . . . . 185 A.5 Generator Operating Characteristics . . . . . . . . . . . . . . . . . . 186

List of Tables

2.1 2.2 3.1 4.1 4.2 4.3 5.1 5.2 5.3 6.1 6.2 7.1 7.2 7.3 7.4 7.5

The Irish electricity system (at year end 2005) . . . . . . . . . . . . Price support caps for wind generation under REFIT . . . . . . . . . Fuel costs in e 2006/GJ . . . . . . . . . . . . . . . . . . . . . . . . . Costs associated with increased cycling of units . . . . . . . . . . . . Cost per single on-o cycle . . . . . . . . . . . . . . . . . . . . . . . Cycling activity expressed as an equivalent hot start . . . . . . . . . Typical components of various fuel types . . . . . . . . . . . . . . . . Typical components of natural gas . . . . . . . . . . . . . . . . . . . Emissions prices in e per ton . . . . . . . . . . . . . . . . . . . . . . Base case assumptions summarised . . . . . . . . . . . . . . . . . . .

24 28 53 67 67 68 77 78 83 90

Worst case and best case assumptions summarised . . . . . . . . . . 100 Assumed Energy Resource for the Five Network Sections (MW) . . . 131 Allocations for the ve network sections . . . . . . . . . . . . . . . . 131 Scaled allocations for the whole system . . . . . . . . . . . . . . . . . 134 Connection costs for wind generation . . . . . . . . . . . . . . . . . . 135 Additional reserve costs for distributed wind generation . . . . . . . 136 xv

LIST OF TABLES 7.6 7.7 7.8 7.9

xvi

Additional cycling costs with wind generation . . . . . . . . . . . . . 136 Emissions savings (in tons) with wind generation . . . . . . . . . . . 137 Fuel savings in petajoules with wind generation . . . . . . . . . . . . 137 Net benets of wind generation (in em) . . . . . . . . . . . . . . . . 138

7.10 Boundary annual costs . . . . . . . . . . . . . . . . . . . . . . . . . . 147

CHAPTER

Introduction

UMAN activities are changing the composition of the Earths atmosphere through the buildup of greenhouse gases. This has resulted in an increase in

the temperature of the earths 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 oating 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) approximately 150,000 people currently die needlessly each year as a direct result of global warming. Agricultural output will suer signicantly as the land becomes drier and outputs fall. This situation will be particularly severe in many Third World countries which lack resources for irrigation. The Hadley Centre (2006) predicts that by 2010, 300 million people in sub-Saharan Africa will be suering 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

Chapter 1. Introduction

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 ood 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 century, on a scale similar to those associated with World War II and the economic depression of the rst half of the 20th century, and it will be dicult or impossible to reverse these eects (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 greenhouse 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 trading 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 under 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 Europes 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

Chapter 1. Introduction

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 renewable 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 potential 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 countrys 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 reected in gures showing wind energy as the fastest growing form of renewable energy in Europe (EWEA, 2005). In addition, the United States is currently experiencing 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.

Chapter 1. Introduction

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 Earths 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 inuence on the most common wind directions. Land masses are heated by the sun more quickly than the sea in the daytime. The air rises, ows 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 Earths surface, wind can be aected 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

Chapter 1. Introduction

example, the amount of electricity generated by a wind turbine uctuates 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 signicant 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 relatively unpredictable. Since the underlying resource cannot be directly controlled, the renewable generation is high when conditions are favourable and low when unfavourable. Thus, forecasts of weather conditions are crucial when examining renewable generation sources. Tidal generation is governed by the gravitational forces of the moon and the sun on the Earths 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 signicant 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 dened 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 uctuates throughout the day. Since the output of some renewable generators, in particular wind generation, cannot be actively controlled and is dicult to predict accurately, this balancing of the generation and the load

Chapter 1. Introduction can become more challenging (Gross et al., 2006).

Given the optimistic targets for renewable energy in Europe and the specic 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 benets associated with increased wind penetration. The work presented here discusses in detail these costs and benets, quanties the impacts 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 signicant challenges posed to system operators when large amounts of variable generation are introduced (Bazilian et al., 2004). Under the European Union EU Directive 96/92/EC (1996) the Transmission System Operator (TSO) is responsible for ensuring a secure, reliable and ecient electricity system and, in that context, for ensuring the availability of all necessary ancillary services. These ancillary services include the provision of sucient 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, o a transmission line trip and unexpected load uctuations. Sder (1993) showed that

Chapter 1. Introduction

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 nondispatchable, 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 OMalley (2005), Dany (2001) and Gouveia and Matos (2004), found that increasing wind power capacity has little eect 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 uctuating 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 ecient 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

Chapter 1. Introduction

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-o 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 signicant cost increases. Currently many wind turbines have protection equipment installed which disconnects 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 signicant 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 frequency disturbances, as this would lead to a large generation decit 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 connection challenges for wind generation with issues such as voltage control, harmonic emissions, short circuit levels and losses playing a signicant role (Keane and OMalley, 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 tradeo between the windiest sites and easy access to the distribution network. There are

Chapter 1. Introduction

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 signicant 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 signicantly 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 commentator David Bellamy wind turbines are completely eete 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 dened 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% depending 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 benecial 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 countrys reliance on fossil fuel it reduces its exposure to uctuations in supply as a

Chapter 1. Introduction result of international politics (Ahmed and Cohen, 1994).

10

In addition to the capacity benet of installing wind generation, the operation of wind generation can provide signicant 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 electricity produced by the wind generator. Since generation must always equal demand, 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 conicts 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 diculties 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 caloric 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

Chapter 1. Introduction

11

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 aected by the ame 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 (Connie et al., 1997), emissions reduction from wind generation is highly signicant. Wind generation development can also greatly aid local communities. The construction and operation of wind turbines creates local jobs, and Mazza (2005) estimates 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 red plants. Private landowners can earn signicant 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 benets 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 improvements 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

Chapter 1. Introduction

12

technology and a countrys 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 eect of encouraging people to be more energy ecient 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 benets 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 benets investigated in this thesis are limited to the most direct costs and benets, and are summarised in Section 1.4.

1.3

Quantifying the Costs & Benets of Wind

A number of studies in the past have attempted to estimate the value of these costs and benets of wind generation and a selection of these are reviewed here. Kennedy (2005) give a generic approach to calculating the long term costs and benets of wind generation. Here the average variable, xed and environmental costs for conventional generators over a xed period of time are calculated without wind in the plant portfolio. A xed capacity of oshore wind generation is then added and the generators costs are recalculated. The dierence in the costs is taken to be the social benet 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 ecient advanced gas red units and relatively low emitters of CO2 and NOX with no SO2 emissions, which underestimates the potential emissions benets 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

Chapter 1. Introduction and any additional reserve that may be required.

13

Munksgaard and Larsen (1998) compare the socio-economic impacts of wind generation in Denmark to central gas and coal red plants. This paper presents an interesting study on both the internal and external costs and benets of wind generation when compared to conventional coal and gas units. However, the wind turbine sizes analysed are small in comparison to todays 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 benets of wind generation requires 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 benets 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 benets. 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 economic 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 benets of wind generation through public preferences. Pepermans et al. (2005) provide a comprehensive summary of the costs and benets that can accrue from small scale wind generation installed on the distribution network. Although this paper provides insightful analysis of the costs and benets of wind generation, it does not attempt to place a monetary value on these costs and benets. A number of more comprehensive studies have been conducted on dierent electricity systems in order to quantify the real cost of integrating wind generation onto a network. The results from these dierent papers are dicult to compare due to

Chapter 1. Introduction

14

dierent 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 benets associated with wind generation, if quantied at all. Comparisons across work done in dierent countries must be done with caution as the costs and benets 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 dierent 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 Europe 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 scenario 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 conventional 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 conventional unit eciency. Gross et al. (2006) provide a systematic review of a large

Chapter 1. Introduction

15

number of studies on the costs and impacts of intermittent generation with an emphasis 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 capacity 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 supply 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 exible units. The Norwegian electricity system is almost entirely made up of hydro generation (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 benets covered include replaced 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

Chapter 1. Introduction

16

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 exible. A large amount of interconnection is also apparent in the Nordic system which is benecial 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 benet and results in an underestimation of the potential benets of wind generation. A study undertaken by EirGrid (2004), the system operator in Ireland, attempted to estimate the future benets 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 ndings 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.

Chapter 1. Introduction

17

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 witnessed 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 PaciCorp 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 o 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 benets 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 generation. 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 todays standards and the assumptions regarding forecasts were conservative (Smith et al., 2004). Increased wind penetrations of 15% installed capacity 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

Chapter 1. Introduction

18

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 benets accruing to improvements in wind power forecasting with savings of up to $10/MWh. Work by Hirst (2001, 2002) for the Bonneville Power Administration concentrated 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 dicult to compare across dierent systems and many have concentrated solely on quantifying the costs and not the benets. In addition, none of the cost benet 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 quanties the costs and benets of wind generation. The approach adopted attempts to maximise social welfare and thus incorporates both direct and indirect costs and benets into the analysis. Thus, it is assumed that society pays the capital cost for every MW installed of wind generation and reaps the benets 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 proles 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

Chapter 1. Introduction

19

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 benets investigated include the capacity benet of wind generation, the emissions savings benets and the fuel savings as increasing penetrations of wind generation are added. The costs and benets are then brought together for each penetration level of wind generation, and the net benets are determined. It is hypothesised that the shape of the net benets 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

Installed Wind
Figure 1.1: Net benets 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 benets 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

Chapter 1. Introduction

20

market (Katz and Rosen, 1998). Thus, the generators are assumed to be prot 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, however, 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 explicitly covered in this work. However, the dispatch model developed is used as part of a cost-benet study on distribution connection policies (Keane et al., 2006), and is discussed in Chapter 7. The cost-benet analysis of connection policies forms part of a larger study on distribution connected generation by Keane and OMalley (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 specic 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 benets of wind generation are discussed briey 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

Chapter 1. Introduction issues on the net benets of wind generation are discussed briey in Chapter 8.

21

1.5

An Outline of the Thesis

This thesis consists of eight chapters covering dierent aspects of the cost benet analysis and detailed results, discussions and conclusions are given in each chapter. Two appendices are also included. The rst appendix details the characteristics of all the generators on the Irish electricity system. The second appendix presents the network study conduced by Keane and OMalley (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 underlying 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 development and operation costs of the wind turbines, the cost of deep network reinforcement, 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 benets of wind generation. The value of the added capacity of wind generation is quantied in addition to the emissions and fuel saving

Chapter 1. Introduction benets.

22

Chapter 6 combines the results from chapters 4 and 5. The net benets of increasing penetrations of wind generation are calculated by subtracting the costs from the benets. The penetration level where the costs equal the benets is the critical point beyond which the costs will exceed the benets 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 benet methodology employed in previous chapters. The rst application investigates the impact of carbon prices on the cycling costs and carbon dioxide emissions of generators. 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 benets of dierent distribution connection policies to determine the access policy which will yield the maximum net benets. The nal application investigates the costs and benets 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 benets. The main conclusions of the thesis are presented and some directions for possible future research are provided.

CHAPTER

The Case Study

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 interconnected 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 Peak Demand (MW) Total Generation (MW) 4,566 6,403 NI 1,663 2,013 All Island 6,229 8,416

Ireland has a large and growing installed wind capacity which currently represents 8% of total installed capacity. In addition, Ireland has limited interconnection making the eects of wind generation on system operation easier to identify as they are not inuenced by interconnection to other systems. There is a relative generation capacity shortage in the Republic and temporary diesel generators are currently employed 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 responding to large uctuations in wind generation production. Ireland is thus an ideal case study to investigate the system impacts of wind generation as the eects
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.
1

Chapter 2. The Case Study are relatively easier to identify.

25

Section 2.1 discusses the current and future wind situation in Ireland. Section 2.2 details the load growth and daily load prole 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 dierent 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 generating 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 accessible, 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 beginning 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 oers 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 oshore development, located o the east coast.

Chapter 2. The Case Study

26

Figure 2.1: Existing wind farms in RoI in January 2007 In order to deal with the large number of interacting applications for wind generation 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 oers. The previous method, on a rst come rst served basis, while eective for large conventional generating plant and for a small number of interacting applications, is not eective 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 oer to the individual applications within the group. This connection oer will remain valid irrespective of whether other applicants in the same group

Chapter 2. The Case Study

27

accept their connection oers. In the case of a major change in the shared connection design, the connection oer may vary from the initial oer. 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 network 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 reect the prices at which they would be willing to sell their electricity to the ESB. Those developments with the lowest bid prices were then oered 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 Tari (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

Chapter 2. The Case Study

28

mechanism will come from the public service obligation levy (DCMNR, 2002). The REFIT scheme is similar to other xed feed in taris which have proven successful in many EU countries (REFIT, 2006). The xed 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 Large Scale Wind Small Scale Wind Size > 5MW < 5MW Tari (per MWh) e 57 e 59

Putting these gures into perspective, in Germany the xed feed in tari ranges from e60 - 89/MWh, in France the feed in tari is e30.50 - 83.80/MWh and in Spain it is just e27/MWh (RIS and EWEA, 2005).

2.2

Irelands Electricity Demand

Irelands electricity demand is relatively small with a peak demand of just 6,229MW in 2006. Figure 2.2 illustrates Irelands electricity consumption relative to the other countries within the EU15 (Eurostat, 2006; IAEA, 2006). Although Irelands electricity 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 Operator 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

Chapter 2. The Case Study

29

600 1990 2004 500

400

TWh

300

200

100

Luxembourg

Germany

Sweden

Italy

Netherlands

Denmark

Belgium

France

Greece

Finland

Austria

Ireland

Figure 2.2: Comparison of electricity consumption in the EU15

5000 4500 4000 3500

Demand MW

3000 2500 2000 1500 1000 500 0 0 6 12 18 24 Winter Peak Typical Winter Typical Summer Summer Minimum

Hour

Figure 2.3: Typical daily demand patterns in RoI in 2006

Portugal

Spain

UK

Chapter 2. The Case Study

30

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 representative 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 prole 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

Irelands Electricity Market

Irelands electricity market is currently based on highly regulated bilateral trading with a top-up and spill balancing mechanism. The current arrangements are expected 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 rst 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 dierences (CfDs), can be

Chapter 2. The Case Study

31

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 mechanism should ensure the energy bids from generators do not incorporate a capital costs element. The model described in detail in Chapter 3, aims to reect the salient features of the proposed SEM market design.

2.3.1

Dominance in the Irish Electricity Market

Given its small size, Irelands 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 distribution 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 distribution 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

Chapter 2. The Case Study

32

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 electricity industry where competition is possible (transmission and distribution networks 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 position 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 rm market control, transparency and bidding rules rather than attempt to disaggregate its generating and supply assets to reduce its 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.

Chapter 2. The Case Study

33

2.4

Irelands Current Generation Plant Mix

Irelands generation plant mix was traditionally based on large coal and oil red 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 uidised bed technology and their operation is similar to that of coal red units. Peat red generation accounts for approximately 350MW of Irelands 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 red 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 prole 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 red units were decommissioned and were replaced with more ecient uidised 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 red units, predominantly CCGTs. It is anticipated that this trend is likely to continue into the future (EirGrid, 2005b; Doherty et al., 2006; CER, 2006b).

Chapter 2. The Case Study

34

80 70 60 Hydro Peat Oil Gas Coal

Age in Years

50 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 Oil

Demand MW

4500

3000

Gas

2450 1500
Coal Hydro & Peat

2000

4000

6000

8000

Hours
Figure 2.6: Load duration curve with merit order of conventional units

Chapter 2. The Case Study

35

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 red units. This dispatch order is based on eciency and fuel costs with the exception of the peat red units. Despite being more expensive and inecient the Irish Government made a policy decision to adopt a must run approach for all peat red 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 red units are dispatched, followed by oil red units, then peaking units red by gas and distillate. Thus, the marginal unit will be gas, oil or distillate depending on the level of demand. For example, gas red units will operate at the margin for approximately 3,510 hours of the year (8760-5250), oil red 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 connects Northern Ireland to Scotland through submarine cables. The capacity of this interconnector is 500MW with an operating capacity of 400MW and it entered into

Chapter 2. The Case Study

36

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 uctuations 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 eect on the reliability of the Irish system. Irelands generating units already suer 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

Chapter 2. The Case Study

37

The poor availability of the generating units on the Irish system adds to the inexibility 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 reduction 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 dioxide 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 benets are calculated for Ireland for the years 2010, 2015 and 2020. The assumed plant mixes for these years were based on the projected generation plant mix considering increased wind penetration for Ireland from EirGrid (2005b) and the optimal future generation portfolio for Ireland to accommodate large penetrations of wind generation from Doherty et al. (2006). Preliminary results from the Government sponsored All Island Renewable Grid Study (AIRGS, 2006b) specically 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 red 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.

Chapter 2. The Case Study

38

For 2015 it is envisaged that new additions to the conventional plant mix will be gas red (EirGrid, 2005b; CER, 2006b; Doherty et al., 2006). As such, three new gas red units are expected to come online by 2015 (EirGrid, 2005b). In addition it is assumed that two of the oil red 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 interconnector between Ireland and Wales for security of supply and competition purposes. 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 red units are retired due to age and are replaced with an additional four gas red 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 benets 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 benets presented in this work are, in the main, dependent on the underlying electricity system. Thus, the most signicant 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

Chapter 2. The Case Study

39

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 red. In addition, a 500MW interconnector to Wales is expected to be completed by 2012 (EirGrid, 2005b). Irelands underlying plant mix is currently made up of predominantly gas, oil and coal units. The plant mix is relatively inexible 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 penetrations 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. Irelands electricity market will be an all island market, covering the Republic 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

Chapter 2. The Case Study

40

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 benets are brought together to determine the optimal penetrations of wind generation. This is in contrast 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

The Dispatch Model

N order to quantify the costs and benets of wind generation it is necessary to analyse it in the context of the power system in which it is installed and to

investigate winds impacts on the operation of the underlying plant mix. An electric power system based predominantly on hydro power will result in very dierent costs and benets 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 benets 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 reect 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 almost 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 prediction of operating decisions in decentralised markets (Xu and Christie, 1999). Thus, although the model used here reects a centralised gross pool electricity market, it could also reect 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 reective bids for energy and since there will be a separate capacity payment mechanism, these bids should reect 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 cooptimise unit operating points and reserve levels to nd a least cost solution on an hourly basis. The aim is given by the following objective function:
N N

min
i=1

cpi Pi +
i=1

cri Ri

(3.1)

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

Chapter 3. The Dispatch Model to a load balancing constraint and an operating reserve target.
N

43

Pi = Load
i=1 N

(3.2)

Ri Reserve
i=1

(3.3)

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)

Pi

1 Ri P maxi Rslopei

(3.6)

R maxi Pi + Ri 0 P mini

(3.7)

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 operating point. The nature of these reserve constraints is illustrated graphically in Figure 3.1. Discrete decisions must be made about the on/o 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/o decisions. The model is rst run with each generators

Chapter 3. The Dispatch Model

44

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 rst run of the dispatch model essentially determines which units should be on and which o. The algorithm is then run a second time, this time only including those units which have been turned on in the rst 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, maximum 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

Chapter 3. The Dispatch Model forecasts and as such a perfect wind forecast was required.

45

The dispatch model was benchmarked against the PLEXOS model at a number of dierent 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 dispatches when compared with the PLEXOS dispatches. In addition, the resulting operating schedules for each unit from the dispatch model were examined to determine 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 rst stage of the model essentially returns a commitment decision, by deciding which units should be turned on and which should be o. The second stage of the model does not turn on or o 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 rst 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 reect 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).

Chapter 3. The Dispatch Model

46

In the process of establishing a transparent market for the island of Ireland, the Irish system regulators gathered a comprehensive data set of operating characteristics 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 rst run of the model is generated based on this real wind power output and the standard deviations of wind power forecast errors (Doherty and OMalley, 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 OMalley, 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 OMalley, 2005)1 . The sensitivity of the net benets 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 rst 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 available, 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.

Chapter 3. The Dispatch Model

47

900

800

700

Wind Power Output MW

Long Range Wind Forecast (commitment) Real Wind Output (dispatch)

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 rst 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 o and the wind generation is curtailed instead. Wind generation is curtailed if Equation (3.8) holds.
N

(i,j Pmin,i ) + Windj > Loadj

i=1

(3.8)

where N is the number of generators on the system, Pmin,i is the minimum generation for unit i, 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

Chapter 3. The Dispatch Model

48

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 rst run of the dispatch model. This would happen if Equation 3.9 were to hold.
N

(i,j (Pmax,i Ri )) + Windj < Loadj

i=1

(3.9)

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 dierent 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 rst 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

Chapter 3. The Dispatch Model

50 45 40

49

Load Factor %

35 30 25 20 15 10 5 0 0 500 1000 1500 2000 2500 3000 3500 4000 High Load Factor Constant Load Factor Mid Load Factor Low Load Factor

Wind Generation MW Figure 3.3: Assumed load factors with increasing installed wind generation wind penetration is in oshore sites with higher load factors and in onshore sites with advances in wind turbine eciency. 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 benets 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 rst 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 nal 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,

Chapter 3. The Dispatch Model the system is being dispatched to generate electricity and reserve for hour j.

50

MW
Hour j

Load

a
Forward Market Reserve

aaaaaaaaaaaaaaaaaa aaaaaaaaaaaaaaaaaa aaaaaaaaaaaaaaaaaa caaaaaaaaaaaaaaaaaa aaaaaaaaaaaaaaaaaa aaaaaaaaaaaaaaaaaa aaaaaaaaaaaaaaaaaa

Gate Real Closure Time

Time

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 OMalley, 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

Chapter 3. The Dispatch Model included.

51

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 prole has been adjusted to an updated prole (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 benets 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 briey in Chapter 8.

3.4

Specics of the Dispatch Model for Case Study

The specic 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 (Sder, 1993). Thus, in this analysis, the assumed reserve o 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 OMalley (2005)) are shown in Appendix A.

Chapter 3. The Dispatch Model

52

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 generators maximum capacity was then set to zero for all 24 hours of the days it was scheduled to be oine.

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)y2005

(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 at 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

Chapter 3. The Dispatch Model year will be one of the three growth rates given.

53

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 ecient gas red plant (CER, 2006b). Table 3.1: Fuel costs in e 2006/GJ Year 2010 Scenario Low Mid High Low Mid High Low Mid High Gas 4.16 4.37 4.58 4.41 5.04 5.66 4.91 6.41 7.91 Coal 2.16 2.17 2.18 2.17 2.22 2.26 2.21 2.31 2.41 Oil 4.13 4.27 4.40 4.30 4.71 5.12 4.63 5.62 6.61 Peat 3.14 3.23 3.33 3.25 3.53 3.82 3.48 4.15 4.83 Distillate 7.83 8.08 8.34 8.14 8.92 9.70 8.76 10.65 12.52

2015

2020

The bid price for each generator is obtained by multiplying the appropriate fuel price by the generators consumption of energy per MWh (given for each generator in Appendix A). The sensitivity of the net benets 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 red generation accounts for approximately 350MW of Irelands installed capacity. The Irish Government have adopted a must-run attitude towards peat generation and in 2001 the European Commission agreed to allow the Irish Government to continue this

Chapter 3. The Dispatch Model

54

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 Irelands strategy to ensure security of electricity supply. This must-run policy however, distorts the optimal scheduling of the generators. Since this is a social welfare 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 reinforcement necessary with increases in wind generation are discussed in Chapter 4. These network reinforcements will help ameliorate potential congestion. Network congestion is discussed briey in Chapter 8. In addition, an application of the dispatch model has been conducted in Chapter 7 which utilises network analysis for the distribution system.

Chapter 3. The Dispatch Model

55

3.5

Summary of Key Features of the Dispatch Model

This chapter detailed the model used to simulate the operating schedules of generators 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 energy 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 rst 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 investigated for the years 2010, 2015 and 2020. The network is simplied to a single bus-bar system and the optimisation does not account for losses. The sensitivity of the costs and benets of wind generation to many of the assumptions discussed in this Chapter are investigated in Chapter 6.

CHAPTER

The Cost of Wind Generation

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 participant 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 system 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 uctuating 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 dierent 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 oshore 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 ecient sites

Chapter 4. The Cost of Wind Generation

1.1

58

Capital Cost per MW in em

1.05

0.95

0.9 High Capital Constant Capital Mid Capital Low Capital

0.85

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 rst. 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 o-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.

Chapter 4. The Cost of Wind Generation

59

A=

i 1 (1 + i)T

(4.1)

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 transmission and distribution networks will require upgrading. Workstream 3 of the Government Sponsored All-Ireland Renewable Grid Study (AIRGS, 2006a) will investigate 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 expressed 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 ow 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.

Chapter 4. The Cost of Wind Generation

800 700 600

60

Network Costs em

500 Deep Network Costs 400 300 200 100 0 0

500

1000

1500

2000

2500

3000

3500

4000

Installed Wind Capacity MW 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 o carry additional reserve capacity in order to maintain system security Sder (1993). The level of this additional reserve requirement for Ireland was based on the results shown in Doherty and OMalley (2005). Doherty and OMalley (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 OMalley (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

Chapter 4. The Cost of Wind Generation

61

proportional to the load at any given hour. The wind power forecast errors generally increase as the forecast horizon increases. In Doherty and OMalley (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 load,h and wind,h respectively and the total system forecast error is given by Equation (4.2).

total,h =

2 2 load,h + wind,h

(4.2)

Based on results given in Pinson and Kariniotakis (2003), Doherty and OMalley (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% respectively (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 standard deviations of wind power forecast errors. These graphs are extrapolated from Doherty and OMalley (2005).
1600 1400 1200 Worst Wind Forecasts Average Wind Forecasts Best Wind Forecasts

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 dierent wind power forecast errors

Chapter 4. The Cost of Wind Generation

62

The sensitivity of the net benets 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 eect on the reserve categories that operate over a short time frame (seconds to minutes) (Doherty and OMalley, 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 replacement 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 capacity 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 eciencies. 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 benets, discussed in Chapter 5.

4.4

Cycling Conventional Units

In order to meet the uctuating 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

Chapter 4. The Cost of Wind Generation

63

the morning and turn o 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 exible 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 relatively 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 o. When a unit is turned o 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 o 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 o 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 uctuating stresses resulting in fractures and failures. Fatigue is likely

Chapter 4. The Cost of Wind Generation

64

Operating Level

Load Following

Hot Start

Warm Start

Cold Start

Pmax

Pmin

Damage
Figure 4.4: Damage associated with dierent 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 component 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 o 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).

Chapter 4. The Cost of Wind Generation

65

Component Life Remaining

100%

Creep
A

50% Creep Damage

50%

Fatigue
0 C B

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 exible. In order to illustrate this point, Figure 4.6 shows the operating proles 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 signicant 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.

Chapter 4. The Cost of Wind Generation

66

400 350 300

Output MW

250 200 150

CCGT 0

100 50 0 0 12 24 36 48 60 72

CCGT 3000 Coal 0 Coal 3000

84

96

Hour

Figure 4.6: Operating proles of two Irish units with dierent wind penetrations The actual cost of cycling is very dicult 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-o 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-o 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

Chapter 4. The Cost of Wind Generation 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. Eciency 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

67

Table 4.2: Cost per single on-o cycle

Boiler Type HRSG Subcritical Drum Large Supercritical

Common Fuel Used Gas Coal or Oil Coal

Typical Fuel Cost e 500 5,000 10,000

True Cost e 1,500 - 25,000 15,000 - 100,000 30,000 - 500,000

each boiler as well as their typical fuel costs for an on-o 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 denition 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 eciency 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

Chapter 4. The Cost of Wind Generation

68

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 generators 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 Cold Start Warm Start Hot Start 10% Load Swing EHS 2.28 1.77 1.00 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 gures 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,

Chapter 4. The Cost of Wind Generation

69

using specic generator information (Appendix A), the age and previous operating patterns of the generators (in an attempt to estimate creep damage and exibility) 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 red 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 exible gas red generation. As the installed wind generation increases beyond 2000MW, more units are displaced and plants

Chapter 4. The Cost of Wind Generation

200 180 160 2010 2015 2020

70

Cycling Cost (em)

140 120 100 80 60 40 20 0 0 500

1000

1500

2000

2500

3000

3500

4000

Installed Wind Capacity (MW) 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 Generation

This chapter outlines the costs associated with increasing penetrations of wind generation. 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 dierent capital cost proles will be examined in this thesis. These capital costs include the assumed shallow connection cost to the network and

Chapter 4. The Cost of Wind Generation are represented as an annuity with a term of 20 years.

71

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 increased wind generation in Doherty and OMalley (2005). Three dierent reserve 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 dispatches 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 generation cause an increase in cycling of marginal units across all three test years resulting in increased costs. The sensitivity of the net benets 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

The Benets of Wind Generation

HIS chapter discusses the benets associated with increased wind generation. As with the costs, the benets 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. However, as wind penetrations increase, the capacity credit tends to reduce as the correlation of the wind generation output on the system increases (Castro and Ferreira, 2001). The value of the capacity benet 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. Emissions 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 Benets 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 benets 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 benets of wind generation and the main assumptions discussed in this Chapter are given in Section 5.5.

5.1

Capacity Benet

One of the key benets 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 dierent 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 proles were

Chapter 5. The Benets of Wind Generation

74

subtracted from the load prole. 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

0.4

Capacity Credit

80

Capacity Credit

0.3

60

0.2

40

0.1 Value of Capacity Benefit

20

0 0

500

1000

1500

2000

2500

3000

3500

4000

Installed Wind Capacity MW

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 dened as the extent to which wind generation can substitute for conventional generation, the capacity benet 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

Benet in em

Chapter 5. The Benets of Wind Generation

75

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 red, 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 benet of wind generation with the monetary values on the secondary yaxis.

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 CV

CO2 /M W h =

(CO2 /kgfuel)

(5.1)

where E represents the generators energy consumed in megajoules (MJ) per

Chapter 5. The Benets of Wind Generation

76

MWh and CV is the caloric 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 combustion 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 deciency, 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 allowable 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 content of the fuel. It is also signicantly aected by the ame 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

Chapter 5. The Benets of Wind Generation

77

produced by high-speed reactions at the ame front, and fuel NOX is produced by oxidation of nitrogen contained in the fuel. The reburning mechanism reduces the total NOX formation by accounting for the reaction of NO with hydrocarbons (Kesgin, 2003). Although these processes are dierent 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 generation plant depend on the chemical content and the caloric 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 red units are dealt with separately in Section 5.2.5. Table 5.1: Typical components of various fuel types Carbon Coal Peat Heavy Fuel Oil 65% 25% 87% Sulphur 0.8% 0.15% 2% Nitrogen 1.3% 0.65% 0.2% Caloric Value 28 MJ/kg 7.75 MJ/kg 40 MJ/kg CO2 /kg fuel 2.49 1.25 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 signicantly 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 rening 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 xed nitrogen species such as HCN (a combi-

Chapter 5. The Benets of Wind Generation

78

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 ame. 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 red generation, diers 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 Gas 70% Sulphur 0% Nitrogen 1.3% Caloric Value 48 MJ/Nm3 CO2 /kg 2.68

Carbon dioxide formation in a gas red 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 signicant 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 red 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

Chapter 5. The Benets of Wind Generation

79

that NO can be minimised by reducing the concentration of [O] or [N2 ]. This rate of thermal NOX production becomes signicantly faster at high temperatures, thus, reducing the temperature results in a reduction in NO formation (Delabroy et al., 1998). However, reducing the ame temperature reduces the eciency of the plant. Combined cycle gas turbine developers are constantly striving to improve the eciency of the turbines and this is often done through increased ame temperature. Combined cycle gas turbine (CCGT) systems generate around 40% more electricity 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 ring, localised 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 uctuations, which can produce turbine damage, ame instability and even ame 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 maximum capacity). As a result, NOX emissions in a CCGT increase signicantly 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 (OMahony, 2004). From Figure 5.2 it is clear that if a CCGT is forced to operate below approximately 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 eciencies due to the increased reserve requirement. In addition, wind generation causes an increase in the

Chapter 5. The Benets of Wind Generation

4.5 4 CCGT OCGT Coal SCR

80

NOX Emissions (kg/MWh)

3.5 3 2.5 2 1.5 1 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 red units. Figure 5.2 also shows the NOX emissions from a coal red unit with a scrubber installed. Under the large combustion plant directive, all the coal units on the Irish system are expected to have this technology installed by 2010 (EU Directive 2001/80/EC, 2001; ESB, 2003). As such, a coal red unit could actually produce less NOX emissions than a CCGT, if the gas unit was operating at less than 70% of maximum capacity.

Chapter 5. The Benets of Wind Generation

81

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 red unit on the Irish system uses 14,620GJ of energy when started from cold (see Appendix A). From Table 5.1, the caloric 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 0.028

CO2 = (14, 620)

(2.49) = 1, 300tons

(5.5)

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 signicantly 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 signicant ramping (OMahony, 2004).

5.3

Emissions Benets 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 specic emissions information for each individual generator (EirGrid, 2006a; OMahony, 2004; AIP, 2005). Figure 5.3 illustrates the emissions benets 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

Chapter 5. The Benets of Wind Generation

30000 25000 20000

82

Kilotons

30 25 20 15 10 5 0 0 500 1000 1500 2000 2500 3000 Installed Wind Capacity MW 3500 4000 CO2 SO2 NOx

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 ecient 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 red, thus displacing these units results in reduced CO2 , SO2 and NOX emissions. However, as the installed wind goes beyond 2500MW, the marginal units tend to be made up of gas red units which produce less CO2 emissions per MWh and no SO2

Chapter 5. The Benets of Wind Generation

83

emissions, hence the negligible SO2 reductions. Above about 1000MW of installed wind, the reduction in NOX emissions from the displacement of oil units is counteracted 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 dierent emissions prices are considered in this thesis and are summarised in Table 5.3. Table 5.3: Emissions prices in e per ton Emission CO2 SO2 NOx Low 10 50 1000 Mid 30 150 3000 High 50 250 5000 Highest 70 350 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 signicant impact on the cycling costs. The interaction between emissions prices and cycling costs is developed in detail in Chapter 7.

Chapter 5. The Benets of Wind Generation

84

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.
500 450 Highest Emission Prices High Emission Prices Mid Emission Prices Low Emission Prices

Emissions Benet (em)

400 350 300 250 200 150 100 50 0 0

500

1000 1500 2000 2500 3000 Installed Wind Capacity MW

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 benets to changes in the emissions prices are discussed in detail in Chapter 6.

5.4

Fuel Savings Benet

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 ecient and thus their fuel consumption per MWh is higher. In addition, an increase in the cycling of generators causes increased fuel consumption.

Chapter 5. The Benets of Wind Generation

85

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

Energy Consumed in Petajoules

180 160 140 120 100 80 60 40 20 0 0 500 1000 1500 2000 2500 3000 Installed Wind Capacity MW 3500 4000 Gas Coal Peat Oil

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 red so wind generation has a signicant impact on the consumption of gas. Although much less generation is provided by oil and peat units, their consumption is signicantly reduced by the wind generation. The coal units tend to be baseloaded and thus their fuel consumption is really only aected at high wind penetrations. For the purposes of this cost benet analysis, these fuel savings are valued at the assumed fuel prices for each fuel type. The assumed fuel prices are shown in Table

Chapter 5. The Benets of Wind Generation

86

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 Installed Wind Capacity MW

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 signicant. 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 Benets of Wind Generation

This chapter outlined the benets associated with increased wind capacity on the Irish system. The capacity benet 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

Chapter 5. The Benets of Wind Generation

87

system and the level of this benet 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 benets at all penetrations, 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 benets to changes in the assumptions surrounding the benets 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 benets of wind generation, such as the creation of jobs and improvements in local infrastructure are not explicitly dealt with in this thesis, however, the implications of including these benets is discussed briey in Chapter 8.

CHAPTER

The Net Benets of Wind Generation

HAPTERS 4 and 5 set out the costs and benets associated with wind generation for the case study. In this chapter, the quantied costs and benets

are brought together to determine the optimal levels of installed wind generation for the case study. Section 6.1 shows the net benets 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 Benets of Wind Generation in the Base Case

The annual costs of wind generation were described in detail in Chapter 4 and included 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 Benets of Wind Generation

89

to calculate the net benets 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

T Cw =
t

1 (Cw + OMw + N Ww + Rw + CYw ) (1 + r)t

(6.1)

where r represents the assumed discount rate and T the term over which the costs and benets 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 benets of wind generation were discussed in Chapter 5 and are summed to give the total benets TB w , shown in Equation (6.2).
T

T Bw =
t

1 (CPw + Ew + F Sw ) (1 + r)t

(6.2)

where CP w is the capacity benet 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 benets for each level of installed wind generation are then determined by subtracting the total costs from the total benets for each level of installed wind capacity. This Section discusses the net benets 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 benets curve for wind generation for each of the test years 2010, 2015 and 2020 under these base case assumptions. The net benet curves illustrate where the incremental net benets begin to decline and more importantly, the points where the costs equal the benets (where net benets equal 0), representing the critical points beyond which no further investment should be made in wind generation. Beyond the critical point, the costs

Chapter 6. The Net Benets of Wind Generation

90

Table 6.1: Base case assumptions summarised Assumption Load Factor Wind Power Forecast Error Demand Growth Fuel Price Growth Wind Capital Costs Wind O&M Costs Network Upgrade Costs Reserve Costs Cycling Cost Capacity Benet Emissions Costs Discount rate Term Scenario Mid scenario in Figure 3.3 Standard Deviation of 9% (Section 3.2) 3.7% per year Mid prices in Table 3.1 Mid Capital in Figure 4.1 e35,000 p.a. per MW installed See Figure 4.2 Average curve in Figure 4.3 See Figure 4.7 See Figure 5.1 Mid emissions prices in Table 5.3 8% 20 years

4000

3000

2000

Net Benet em

Optimal

1000

0 2020 2015 2010

1000

2000

3000 0

500

1000

1500

2000

2500

3000

3500

4000

Installed Wind Capacity (MW)

Figure 6.1: The base case net benets of wind generation

Chapter 6. The Net Benets of Wind Generation

91

will exceed the benets and the net benets 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 benets for 2010 are less than in 2015 and 2020 due to the installed plant mix being more inexible and having less interconnection. In addition, the presence of an east west interconnector with a capacity of 400MW makes a highly signicant impact on the net benets of wind generation in 2015 and 2020. The replacement of the oil red units with gas red units is also highly benecial 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 benets of wind generation of altering the assumptions in base case.

6.2

Sensitivity Analysis of Net Benets

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 benets curve is recalculated and the critical value of installed wind generation, where the net benets 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 Benets 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 Benets 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

Chapter 6. The Net Benets of Wind Generation

94

Although this thesis analysed only installed wind capacities up to 4000MW (representing 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 extrapolation from the net benets 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 benets of wind generation in 2010.
3000

2000

Net Benets em

1000

0 High Load Factor (HLF) Constant Load Factor (CLF) Base Case (BC) Low Load Factor (LLF)

1000

2000

3000 0

500

1000

1500

2000

2500

3000

3500

4000

Installed Wind Generation Figure 6.4: The net benets of wind with changes in load factor

Chapter 6. The Net Benets of Wind Generation

95

It can be seen from Figure 6.4 that assuming a high load factor shifts the net benets 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 benets 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 benets curve signicantly downwards resulting in a reduced critical value. It is also interesting to note that at a low load factor, the net benets of wind generation at low levels of penetration actually dip below zero. This is due to the lower emissions and fuel benets but a relatively high capital cost. The net benets curves for the load factor sensitivities for 2015 and 2020 were also calculated and the curves followed the same general patterns. These net benets 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 benets 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 benets 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

Chapter 6. The Net Benets of Wind Generation

96

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 eciencies resulting in reduced emissions and fuel saving benets. 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 eect on cycling costs. Each of these factors combine to increase the costs of wind generation and reduce the benets, thereby reducing the overall net benets of wind generation, and shifting the net benets curve downwards. As the generation plant mix in 2020 is more exible, the cost of poor wind forecast accuracy is lower. On the other side, assuming the best case wind power forecasts, signicantly improves the critical values of wind generation by requiring less reserve capacity, more ecient 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 preferred to low demand growth. With low demand growth, wind generation represents a higher proportion of the dispatched plant. This results in the marginal plants being higher up the merit order, and in Ireland, this means older, more inecient coal red 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 red units which are more exible and have lower cycling costs.

Chapter 6. The Net Benets of Wind Generation

97

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 benecial for wind generation since the value of the savings in the fuel bill from wind generation are increased. This is particularly 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 benets 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 benets 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 increasing 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 Capital 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

Chapter 6. The Net Benets of Wind Generation

98

net benets 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 rst 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 oshore 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 proles 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 benets of wind to a reduction in these costs is investigated. A more exible plant mix would have a lower cost of cycling, thus, two scenarios relating to lower cycling costs were investigated. The rst 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 benets of wind generation increase. Wind generation causes an increase in the cycling of units on the margin, thus with more exible 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 exible plant mix.

Chapter 6. The Net Benets of Wind Generation

99

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 benets were mainly dependent on the price of CO2 rather than the other emissions. At the low emissions price, the merit order of the conventional 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 benets of wind generation are being calculated, each of the costs and benets 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 benets to 6% and increasing them to 10%. As would be expected, a lower discount rate increases the net benets of wind generation and a higher discount

Chapter 6. The Net Benets of Wind Generation rate reduces them.

100

6.2.9

Worst and Best Case Scenarios

The previous sections illustrated the sensitivity of the net benets 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 benets 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 Load Factor Wind Power Forecast Error Demand Growth Fuel Price Growth Wind Capital Costs Cycling Cost Emissions Costs Discount rate Term Worst Case Low Worst (std dev 14%) Low Low High Capital As in Base Case Low 10% 15 years Best Case High Best (std dev 7%) High High Low Capital 80% Highest 6% 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.

Chapter 6. The Net Benets of Wind Generation

101

6.3

Conclusions of Net Benets

This Chapter presented the net benets 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 benets curves were presented for the base case and the optimal penetrations were found where the net benets 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 exible plant mix, as in 2020, signicantly increases the net benets 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 benets 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 benets and low capital costs increase the net benets of wind generation. Accurate wind power forecasting increases the net benets of wind generation by reducing the reserve and cycling costs. A low demand growth reduces the net benets of wind generation signicantly. High fuel prices are benecial for wind generation, increasing the value of the saved fuel consumption with increasing penetrations of wind generation. A more exible plant mix would have lower cycling costs thus two scenarios with lower cycling costs were investigated. It was found that the net benets of wind generation were increased when the cycling costs were reduced. A reduced discount rate increased the net benets of wind generation.

Chapter 6. The Net Benets of Wind Generation

102

Low emissions prices have a signicant detrimental impact on the net benets 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 red 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 assumptions 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 benets 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 benets results for wind generation for the case study. The costs and benets 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 benets are developed in the following Chapter 7.

CHAPTER

Applications of the Net Benets Methodology

HIS chapter discusses the results of some interesting applications of the dispatch 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 previously 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 investigated many of the costs and benets of wind generation, this application on carbon prices isolates just the cycling costs and the emissions benets and investigates the relationship between the two. In the work presented so far, the power system was operated by incorporating 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 Benets 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 dierent 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 benets to determine the connection policy which yields the maximum net benets. This work was conducted in collaboration with Keane and OMalley (2005b). Wind generation is variable and relatively unpredictable, whereas, tidal generation 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 feasible tidal generation penetrations on the Irish system. The capital costs necessary to ensure positive net benets for tidal generation are also shown. Bryans (2006) investigated the tidal generation resources in Ireland and his ndings 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 introducing 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 internalise the cost of emissions for polluters with the aim of altering their behaviour patterns to reduce their greenhouse gas emissions. In electricity markets, generators generally consider their operating costs to be made up of their internal costs. However, these policy instruments have made the additional emission costs applicable 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 greenhouse 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 rst

Chapter 7. Applications of the Net Benets Methodology

105

Kyoto Protocol commitment period. Under the EU ETS, each generator must submit 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 red production to gas. This result is supported by Klaassen and Riahi (2006) who show that the use of coal for electricity production declines steadily into the future with the introduction of a carbon tax. Voorspools and Dhaeseleer (2006) studied the impact of a carbon tax on a number of EU countries and their results illustrate the signicant 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 benet analysis of emissions reduction in the electricity sector, encompassing the electricity generation costs and the social benets. This study found that the benets 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.

Chapter 7. Applications of the Net Benets Methodology

106

As discussed in Chapter 4, generating units are usually designed to be at their most ecient 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 signicantly 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 red 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 signicantly and could dampen the benets 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 generator 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 generation 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 eect 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 benets study. The assumed plant mix and load for this analysis are for 2007. The characteristics 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-

Chapter 7. Applications of the Net Benets Methodology

107

patches with increasing emissions prices and wind generation. However, in an attempt to reduce the number of factors which may inuence the results, the model was simplied 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 benets analysis, the peat red 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 red generation, thus, in this Section, the peat red 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 illustrates the results showing the cycling costs and emissions savings rstly 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 signicantly 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 red units and the coal units have been pushed up to the margin. As discussed in Section 4.4, units

Chapter 7. Applications of the Net Benets Methodology

108

7500

6000

Distillate Gas Oil Gas

Demand MW

4500

3000

Coal

2450 1500
Gas

Hydro & Peat

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 proles will change dramatically which is likely to result in a signicant 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).

Chapter 7. Applications of the Net Benets Methodology

250 Added Cycling Cost Value of Saved CO2 200 8 10

109

e million

150

100

50 Saved CO2

0 0

10

20

30

40

50

60

0 70

CO2 price in e
Figure 7.2: Cycling costs and CO2 savings at dierent 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 benet 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 Dhaeseleer (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

Million Tons CO2 saved

Chapter 7. Applications of the Net Benets Methodology

110

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 signicant 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 signicant 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 generation 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 displacing 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 benet analysis conducted in previous Chapters but are investigated in isolation in this Chapter.

Chapter 7. Applications of the Net Benets Methodology

100 10

111

80

e million

60

40

20 Saved CO2

0 0

500

1000

1500

2000

2500

0 3000

Installed Wind Capacity MW

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 Dhaeseleer (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

Million Tons CO2 saved

Added Cycling Cost Value of Saved CO2

Chapter 7. Applications of the Net Benets Methodology

200 180 160

112

Cycling Cost (em)

140 120 100 80 60 40 20 0 0 10 20 30 40 50 60 70 3000MW Wind 2000MW Wind 1000MW Wind 0MW Wind

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.

Chapter 7. Applications of the Net Benets Methodology

113

8 7

CO2 Savings (Mton)

6 5 4 3 2 1 0 0

3000MW Wind 2000MW Wind 1000MW Wind 0MW Wind 10 20 30 40 50 60 70

Carbon Price (e)

Figure 7.5: Carbon savings with wind generation and carbon prices
350 300 250 3000MW Wind 2000MW Wind 1000MW Wind 0MW Wind

Net Saving (em)

200 150 100 50 0 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

Chapter 7. Applications of the Net Benets Methodology

114

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 decit 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 Ireland where there is a large proportion of the plant mix made up of large relatively inexible units with very large cycling costs. By increasing the exibility of the plant mix, the cycling costs could reduce signicantly. 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 frequently. This can have a signicantly 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-

Chapter 7. Applications of the Net Benets Methodology

115

bon dioxide reduction benets 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 signicant 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 displaces 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 benet analysis discussed in previous chapters, the dispatch model was run with long range wind power forecasts in the rst model run and updated wind

Chapter 7. Applications of the Net Benets Methodology

116

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 ecient 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 o. 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 benet of wind generation is a fuelsaving 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 eciencies 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 inecient way to operate the system with signicant 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 benets of wind generation are determined for the year 2010. In order to simulate the fuelsaver approach, the dispatch model described in Chapter

Chapter 7. Applications of the Net Benets Methodology

117

3 was altered slightly. In the commitment stage of the model, where conventional generators output can range between 0 and maximum generation, wind power output 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 ecient with much greater wind power curtailment, more units turned on, and more units running at lower eciencies. 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, signicantly 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

Chapter 7. Applications of the Net Benets Methodology

118

3500 Fuelsaver Forecasted 3000

GWh of Wind Curtailment

2500

2000

1500

1000

500

0 0

500

1000

1500

2000

2500

3000

3500

4000

Installed Wind Capacity (MW) Figure 7.7: Wind curtailment in 2010 under fuelsaver

22

Average Number of Committed Units

21 20 19 18 17 16 15 14 13 12 0 500 1000 1500 2000 2500 3000

Fuelsaver Forecasted

3500

4000

Installed Wind Capacity (MW) Figure 7.8: Average number of committed units under fuelsaver

Chapter 7. Applications of the Net Benets Methodology

119

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. Average Operation as % of Maximum Capacity
65 Fuelsaver Forecasted 60

55

50

45

40 0

500

1000

1500

2000

2500

3000

3500

4000

Installed Wind Capacity (MW) 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 eect on the eciency of these units. Operating at reduced levels results in increased fuel consumption and emissions per MWh.

Chapter 7. Applications of the Net Benets Methodology

120

The costs and benets 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 benets which are discussed here are the cycling costs, the emissions benets 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/o cycling as it is assumed under fuelsaver that no units are switched on or o. 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 eciencies 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 Benets

The dispatches from the fuelsaver approach were analysed with regard to the emissions savings with increases in wind capacity. Since the conventional units operate at much lower outputs under the fuelsaver approach, they are less ecient. In addition, as discussed in Chapter 5, NOX emissions increase signicantly 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.

Chapter 7. Applications of the Net Benets Methodology

120 Fuelsaver Forecasted 100

121

Cycling Cost em

80

60

40

20

0 0

500

1000

1500

2000

2500

3000

3500

4000

Installed Wind Capacity (MW) Figure 7.10: Cycling costs under the fuelsaver approach
30000 25000 200

Emissions in kilotons

20000 CO2 SO2 NOx 30 25 20 15 10 5 0 0 500 1000 1500 2000 2500 3000 3500 0 4000 Forecast Savings Fuelsaver Savings 100

Installed Wind Capacity (MW)

Figure 7.11: Emissions savings under the fuelsaver approach

Value of Emissions Savings em

Chapter 7. Applications of the Net Benets Methodology

122

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 Benets

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 reduction 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 o 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 ecient at lower loads. It is seen from the secondary y-axis that the value of the savings are signicantly less than in the forecasted approach.

Chapter 7. Applications of the Net Benets Methodology

200 300

123

150

200

100

50

Forecasted Savings Fuelsaver Savings

100

0 0

500

1000

1500

2000

2500

3000

3500

0 4000

Installed Wind Capacity (MW) Figure 7.12: Fuel savings under the fuelsaver approach

7.2.5

Net Benets of Wind under the Fuelsaver Approach

The cycling costs and the emissions and fuel saving benets 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 benets of wind generation under the fuelsaver methodology are the emissions and the fuel savings. Figure 7.13 illustrates the total cost function, the total benet function and the net benets of wind generation operated under a fuelsaver approach. Figure 7.13 shows that under the fuelsaver approach the costs exceed the benets at all levels of installed wind generation. Thus, at all levels of wind generation, wind

Value of Fuel Savings em

Fuel Consumed (PJ)

Gas Coal Peat Oil

Chapter 7. Applications of the Net Benets Methodology

6000 Costs Benefits Net Benefits

124

4000

2000

em

2000

4000

6000 0

500

1000

1500

2000

2500

3000

3500

4000

Installed Wind Capacity (MW) Figure 7.13: The net benets 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 benets 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

Chapter 7. Applications of the Net Benets Methodology

125

wind generation. By adopting the fuelsaver approach the costs of wind generation exceed the benets 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 generation 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 signicant 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 rm 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 rm generator will be accepted onto the network at all times. The impact of a rm connection policy on the potential for DG has been demonstrated in Keane and OMalley (2005a) and Wallace and Harrison (2003). Adopting a rm connection policy can lead to network sterilisation, which limits the potential for further generation capacity connecting to that section of network3 (Keane and OMalley, 2005b; Vovos et al., 2005). Non-rm 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.

Chapter 7. Applications of the Net Benets Methodology

126

of distributed generation (Collinson et al., 1999). Non rm access refers to output greater than the rm 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 OMalley (2006) have shown the scope for DG if non rm access is employed. Collaboration work with Keane and OMalley (2006) form the basis for the results shown in this Section. Keane and OMalley (2005b, 2006) determined the optimal allocation of energy resources on the distribution network under three connection policy scenarios. It is found that by employing non-rm connection policies, signicantly more renewable generation can connect to the distribution network without the need for additional investment in the network, thus ensuring ecient use of the existing infrastructure. Using the methodologies set out in Chapters 3 to 6 a cost benet analysis is conducted on the optimal penetrations to determine the overall monetary benet of using one connection policy over another. It is shown that a signicant increase in the net benets of distributed generation is gained if the appropriate connection policy is utilised from the outset and conversely that signicant costs are incurred if ad hoc policies are employed. Furthermore, it is shown that non rm access has the scope to facilitate a signicant 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 ow analysis, system scaling and the dispatch model are discussed in Sections 7.3.5 to 7.3.7. The costs and benets of the optimal penetrations are discussed in Section 7.3.8 and the net benets are given in Section 7.3.9. The conclusions are given in Section 7.3.10.

Chapter 7. Applications of the Net Benets Methodology

127

7.3.1

Methodology

The optimisation of generation on the distribution network is discussed in detail in Keane and OMalley (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 OMalley (2005b) to optimise the allocation of distributed generation on the distribution network. The resulting allocations are then used as the basis for the cost benet 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 associated 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 rst case is the base case of the maximum rm 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 OMalley (2005b) for the rm allocation are shown in Appendix B. The second case refers to the maximum non rm access permitted. Here voltage sensitivities are utilised in the objective function to reduce the instances of overvoltage (see Appendix B). The third case examined is the case where rm access is initially only permitted and then non rm access is permitted at a later stage. These two non rm cases are optimised on a MWh/ekV basis.

Chapter 7. Applications of the Net Benets Methodology

128

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 rm and non rm objectives are calculated. The rm allocation (MWh/e) is fed as a constraint into the non rm allocation, which yields the rm + non rm allocation. This process of determining the optimal penetrations of distributed generation was conducted by Keane and OMalley (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 ow analysis (Keane et al., 2006). This load ow analysis determines the instances of DG output curtailment in the non rm 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 proles 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 benets of each of the allocations.

7.3.2

Irelands 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 network 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 OMalley (2005). The ve network sections were selected from along the western coastline and also in the southeast of the country. It is evident from Figure 7.15 that it is in these locations that connections of

Chapter 7. Applications of the Net Benets Methodology

129

SAMPLE NETWORK SECTIONS

TECHNICAL CONSTRAINTS

FIRM ALLOCATION (MWh/Euro)

NON FIRM ALLOCATION (MWh/Euro kV)

ANNUAL LOAD FLOW SIMULATION & CURTAILMENT

SCALING

ANNUAL ECONOMIC DISPATCH

(Chapter 3)

COST BENEFIT ANALYSIS

(Chapters 4 6) CAPITAL, O&M & CONNECTION COSTS, RESERVE & CYCLING COSTS, CAPACITY BENEFIT, EMISSIONS BENEFIT, FUEL SAVING.

Figure 7.14: Methodology

Chapter 7. Applications of the Net Benets Methodology DG, and in particular wind generation, will and are being sought.

130

Figure 7.15: Future distribution of installed wind power capacity in percent The network sections were also selected based on their characteristics and structure from network data available from ESB Networks (2005). They represent the types of network encountered on the Irish distribution network, from voltage constrained rural networks to more urban networks where voltage is less of a problem and short circuit level can become signicant.

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 generation (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 ve representative network sections.

Chapter 7. Applications of the Net Benets Methodology

131

Table 7.1: Assumed Energy Resource for the Five Network Sections (MW) Section Capacity (MW) Energy Source Wind Biomass LFG Hydro Total 1 20 0 12 2 34 2 31.5 0 0 2 33.5 3 39 10 0 1 50 4 23.5 10 0 0.8 34.3 5 24 15 0 0 39 Total 138 35 12 5.8 190.8 Load Factor 0.35 0.85 0.76 0.32 -

7.3.4

Optimal Allocations of Distributed Generation

The allocations for the ve network sections were determined under the three connection policies using the methodology described in Appendix B and the total allocations are shown in Table 7.2. Table 7.2: Allocations for the ve network sections Firm Wind Biomass LFG Hydro Wind Biomass LFG Hydro 48.6 34.4 12 5.0 153 259 80 14 Non Firm 94 35 12 5.8 291 264 80 16.5 Firm + Non Firm 94 35 12 5.8 289 264 80 16.5

Installed MW

Energy GWh

Given their signicantly higher load factor, any available biomass and landll gas (LFG) are allocated the available rm capacity rst. As a result, there is very little dierence, between their rm and non rm 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

Chapter 7. Applications of the Net Benets Methodology

132

always be connected rst. The small scale hydro generation also has very little dierence between its rm and non rm 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 benets of wind generation that are aected 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 annual energy output for each DG resource is ascertained. For the rm case, the annual energy output is determined using actual historical data series from ESB Networks (2005). For the non rm 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 simulations consist of load ow calculations carried out for half hourly data. The data included in the simulations includes active and reactive load and generation proles 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 rm 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 i N. 2i

PCurtail i =

(7.1)

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. i is the voltage sensitivity (kV/MW) of the ith bus and N is the number of distribution buses. i is calculated for each bus drawing on the analysis used in Keane and OMalley (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 ow

Chapter 7. Applications of the Net Benets Methodology

133

simulation yields the amount of energy curtailed over the year, which in turn gives modied generation proles, which can then be used in the dispatch model. Figure 7.16 shows the output prole 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-rm case.
100 90 80 70

% Generation

60 50 40 30 20 10 0:00 4:00 8:00 12:00 Biomass LandFill Gas Wind Non Firm Wind Hydro 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 rm access has the potential to facilitate much higher penetrations than rm access (94MW compared to 48.6MW). The allocations shown in Table 7.2 are only for ve network sections, however, to understand fully the eect 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

Chapter 7. Applications of the Net Benets Methodology

134

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 Wind Installed MW Biomass LFG Hydro Wind Energy GWh Biomass LFG Hydro 655 401 96 64 2065 3018 639 184 Non Firm 1392 410 96 77 4333 3089 639 221 Firm + Non Firm 1392 410 96 77 4302 3089 639 221

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 rm scenario produces more energy output than the rm + non rm case. This is because there is more freedom in the allocating of capacity on the network in the non rm case resulting in reduced capacity allocation at the most voltage sensitive buses. This results in less curtailment under the non rm case when compared to the rm + non rm case. It should be noted that the capacities given here are representative of the distribution 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 benets of distributed generation, it is

Chapter 7. Applications of the Net Benets Methodology

135

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 proles 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 proles which were provided by the Irish distribution network operator for 2007 (ESB Networks, 2005). The DG proles 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 Benets 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 rm connection policy is pursued from the outset. In some cases, the extra non rm 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 Firm Non Firm Firm + Non Firm Installed MW 655 1392 1392 Connection Cost e61,300,000 e128,000,000 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

Chapter 7. Applications of the Net Benets Methodology

136

determine the costs and benets of each of the three connection policy approaches. The additional reserve requirement for the system with increased wind penetration 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 rm cases thus the additional annual reserve costs are the same. Table 7.5: Additional reserve costs for distributed wind generation Connection Policy Firm Non Firm Firm + Non Firm Installed MW 655 1392 1392 Additional Reserve Cost e101,090 e290,990 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 Firm Non Firm Firm + Non Firm Additional Cycling Cost e52,396,400 e66,437,200 e64,858,700

The reason for the slight dierence in the cycling costs for the two non rm cases is the energy output. The non rm case has less curtailment than the rm + non rm case for the same installed capacity resulting in higher annual output. This means that over the year the non rm capacity displaces more generation. However, at times of low load this displaced generation can be units which have higher cycling costs. This non rm case causes a slight increase in the cycling of these units over the rm + non rm case. The dispatch model was also used to calculate the CO2 , SO2 and NOx emissions

Chapter 7. Applications of the Net Benets Methodology

137

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 Firm Non Firm Firm + Non Firm CO2 1,086,116 2,189,700 2,181,300 SO2 2,685 3,659 3,659 NOx 1,147 1,965 1,959 Benet e 36,830,000 72,136,000 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 Price e/GJ Firm Non Firm Firm + Non Firm Gas 4.16 9.97 19.99 19.87 Coal 2.15 0.05 0.22 0.22 Peat 3.14 0.98 2.25 2.25 Oil 4.13 4.75 7.63 7.63 Benet e 64,273,000 122,240,000 121,740,000

7.3.9

Net Benets

The costs and benets above were combined and the net benets of each connection 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 benet is taken to be 37% at 655MW and 31% at 1392MW, as discussed in Chapter

Chapter 7. Applications of the Net Benets Methodology

138

5. It was found that although the benets outweighed the costs under all three connection policies there was a signicant dierence in the net benets. Table 7.9 shows the net benets of each of the connection policies. Table 7.9: Net benets of wind generation (in em) Cap. (MW) Firm Non Firm Firm+NonFirm 655 1392 1392 Total Cost 1,346 2,461 2,476 Total Benet 1,379 2,684 2,673 Net Benet 33 223 197

As is evident from Table 7.9, while the two non rm policies both result in the same installed capacity of generation, the costs and benets of wind generation show signicant dierences. There is a signicant increase of 12% in net benets under non rm connection than under a rm and then non rm scenario for the same level of installed wind. The increases in net benets results from the dierent 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 benets as detailed above. It is clear that the connection policy aects the energy output and connection costs and furthermore that this has a considerable knock on eect on system costs and benets 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 unaected by the non rm 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

Chapter 7. Applications of the Net Benets Methodology

139

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 distribution network under three connection policy scenarios. It is found that by employing non-rm connection policies, signicantly more renewable generation can connect to the distribution network without the need for additional investment in the network, thus ensuring ecient 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 benet analysis is then conducted on these penetrations to determine the overall monetary benet of using one connection policy over another. It is shown that a signicant increase in the net benets of distributed generation is gained if the appropriate connection policy is utilised from the outset and conversely that signicant costs are incurred if ad hoc policies are employed. Furthermore, it is shown that non rm access has the scope to facilitate a signicant 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

Chapter 7. Applications of the Net Benets Methodology

140

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 owing 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 development 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 Businesss Stingray project and the Hammerfest Strm 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 turbines 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 o Lynmouth in the Bristol Channel during 2003. Bryans et al. (2004) determined the resource for tidal energy around Ireland using a 2 dimensional tidal model to simulate the tidal ows 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. However, Bryans et al. (2004) predict that into the future, TED development will lead

Chapter 7. Applications of the Net Benets Methodology

141

Figure 7.17: The marine current turbines design

to larger turbines which will be nancially 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 maximum 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 altering 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

Chapter 7. Applications of the Net Benets Methodology

500 Spring Neap 400

142

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 ow rates (Bryans, 2006; Bryans et al., 2006). In order to model the tidal generation, a net load curve was developed by subtracting the tidal generation from the load. In addition, as the tidal generation is considered to be perfectly forecastable, no additional reserve was required with increasing capacities of tidal generation. The model was run as described in Chapter 3 with the installed wind generation assumed to equal zero. The following gures illustrate the cycling costs, and emissions and fuelsaving benets 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 uctuation is

Chapter 7. Applications of the Net Benets Methodology

143

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 5 0 0

Tidal Cycling Costs

80

Installed Tidal Capacity (MW)

160

240

320

400

480

560

Figure 7.19: Cycling costs with increases in tidal generation

7.4.3

Tidal Emissions Benets

Figure 7.20 illustrates the emissions benets 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 relatively low load factor when compared to the wind generation - approximately 22%

Chapter 7. Applications of the Net Benets Methodology

144

for tidal generation compared to over 40% for low levels of installed wind generation (Bryans, 2006).
30000 25000 20000 CO2 SO2 NOx 40 50

Kilotons

30 25 20 15 10 5 0 0 80 160 240 Value of Emissions Savings 320 400 480

30

20

10

0 560

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 Benets 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 perfectly forecastable it does not add to the reserve requirement of the system. Bryans (2006) calculated the capacity benet of tidal generation to be 22% using the MCT

Value of Emissions Savings in em

Chapter 7. Applications of the Net Benets Methodology

200 Gas Coal Oil Peat 40 60

145

Energy Consumed (PJ)

150

100

20 50 Value of Fuel Savings

0 0

80

160

240

320

400

480

0 560

Installed Capacity (MW) Figure 7.21: Tidal generation fuel savings

technology with 40% EDR. Without any EDR, this drops to 18%. Thus, the capacity benet 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 dened 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 benets 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 benets with an assumed discount rate of 8% and a term of 20 years. From Figure 7.22 the total benets 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

Value of Fuel Savings in em

Chapter 7. Applications of the Net Benets Methodology

800 700 600 500

146

em

400 300 200 100 0 0

NPV of Total Benefits NPV of Cycling Costs 80 160 240 320 400 480 560

Installed Tidal Capacity (MW) Figure 7.22: The total benets and cycling costs of tidal generation

these other costs could be each year to ensure that the benets 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 benets 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 benets and the resource should not be developed. Putting these gures 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 oshore 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 benets (e 10,257,807 - e8m). If this is the annual cost of capital, then the total capital cost of 160MW

Chapter 7. Applications of the Net Benets Methodology Table 7.10: Boundary annual costs Installed Tidal Generation 0 80 160 240 320 400 480 560 Annual Cost em 0 4,059,204 10,257,807 14,602,503 17,973,800 22,023,430 28,582,690 33,693,617

147

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 benets 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 benets 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 proles 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 benets 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 benets for tidal generation. To calculate the net benets of tidal generation it was assumed that there were no deep network reinforcements necessary with increased tidal generation

Chapter 7. Applications of the Net Benets Methodology

148

and the operation and maintenance costs were assumed to be slightly less than those for an oshore wind turbine. However, even with these assumptions, in order to produce positive net benets 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

Discussion and Conclusions

This thesis presented a cost benet analysis of wind generation with specic reference to Ireland as a case study. A range of costs and benets 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 benets. Section 8.1 presents a discussion on a number of costs and benets which could be included in a wider cost benet 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 benets 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 benets 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 briey here. The dispatch model described in Chapter 3 is a single bus bar system which neglects any load ow analysis. Thus, issues of transmission congestion are not included. Congestion limits suppliers 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 specic 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 ows on the lines which may have a knock on eect on congestion (SEI, 2004c). Workstream 3 of the All Island Renewable Grid study (AIRGS, 2006a) will conduct load ow analysis for the case study and will identify potential bottlenecks with increased penetrations of renewable generation. However, although the cost benet 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 signicant. 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 o on the landscape (Etherington, 2006). Hurtado et al. (2004) and Mller (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 dicult to determine and previous studies tend to rely on highly subjective public preference models which use questionnaires

Chapter 8. Discussion and Conclusions

151

to determine the publics 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 aects 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 penetrations 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 dierent types of reserves i.e. it did not distinguish between static and dynamic reserve. Thus, a future cost benet analysis could include the costs and implications of providing additional static reserve. In this thesis, additional operating reserve capacity was dispatched with increasing 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 require 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

Chapter 8. Discussion and Conclusions

152

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 feasibility 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 benets of wind generation 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 curtailment 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 benets 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 benets are very dicult to estimate and will vary depending on the location of the wind farm. Recent a research by Murphy and Walsh (2002) and Forfs et al. (2003) suggest that very little benet should be attributed to creating additional jobs in Ireland because there is eectively 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 benet was not included in this thesis. The Irish Government currently supports the operation of peat red 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

Chapter 8. Discussion and Conclusions

153

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 benets and the local economy benets of the peat red generation. If wind generation was assumed to create these same benets, given its 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 benet of e69,450 per MW installed which is approximately e19/MWh. Alternatively, the feed-in tari price proposed by the Irish Government could be seen to represent the value of the fuel diversity benets of wind generation. As discussed in Chapter 2, the feed-in tari for wind generation is approximately e58/MWh. While this thesis presented a cost benet 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 benecial for the economy with knock on eects on the standard of living, industry, trade, employment etc. These potential indirect benets of wind generation have not been included here but if they were they would shift the net benets 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 specic behaviour of the site in response to meteorological conditions in the past. The accuracy of the wind power forecast is a function of the method used and the comprehensiveness of the historical data employed (GE Energy, 2005). This thesis examined three scenarios for wind power forecast

Chapter 8. Discussion and Conclusions

154

accuracy and it was found that high wind power forecast accuracy increased the net benets for wind generation by requiring less reserve capacity, more ecient conventional unit operation and lower cycling costs. However, it has been shown by GE Energy (2005) that the benets to improvements in wind power forecasting are limited. This study analysed the economic benets 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 forecasts based on the current best practice were used in this analysis and the benets were compared to a perfect wind forecast. Using the wind forecasts in the commitment 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 benets in electricity system 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 benets 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 benets 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

Chapter 8. Discussion and Conclusions

155

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 benets 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 benets of wind generation. Analysing the impact of electricity market gaming on the benets 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 benets 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 specied 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 red 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 specied conventional plant mixes. The methodology was applied to a case study and a cost benet analysis of wind generation on the case system was conducted. In addition, a number of applications of the net benets methodology were also presented. The main conclusions arising out of the cost benets analysis and the applications are presented here.

Chapter 8. Discussion and Conclusions

156

It was found that the additional reserve requirement for wind generation causes a knock on eect on the cycling costs, and the emissions and fuel saving benets. 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 aected by low load operation and cyclical operation. The most signicant fuel savings with wind generation for the case study were in gas. The fuel savings are also aected 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 benets 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 benets of wind showed that the net benets of wind were signicantly aected by the underlying plant mix. It was also shown that high load factors, low cycling costs, low capital costs and accurate wind power forecasts signicantly increased the net benets of wind generation. Low emissions prices signicantly reduced the net benets 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-

Chapter 8. Discussion and Conclusions

157

pressed negative net benets with penetrations of just 5% of electricity generation. Positive net benets were experienced up to 30% of electricity generated through wind in the best case scenario. The carbon dioxide reduction benets 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 signicant 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 benets at all levels of installed wind generation. Employing non-rm connection policies signicantly increases the renewable generation capacity that can connect to the distribution network without the need for additional investment. Also a signicant increase in the net benets of distributed generation is gained if the appropriate connection policy is utilised from the outset and conversely that signicant costs are incurred if ad hoc policies are employed. The net benets 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 benets 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.

Chapter 8. Discussion and Conclusions

158

8.3

Future Work

The costs and benets 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 signicant 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 benet 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 eect on the costs and benets of wind generation. The impact of gaming in the electricity market on the net benets 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 benets of wind and tidal generation in isolation, however the methodology could easily be adapted to incorporate the costs and benets 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 dierent 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 benets 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

Chapter 8. Discussion and Conclusions penetrations and at the correct time.

159

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 benet 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. 23552360. 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 markets 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. 235244. Alberta Environment, 2003. Sulphur dioxide: environmental eects, fate and behaviour - 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, 107116. 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 (N47, Rev. 29). Baumol, W., Oates, W., 1975. The Theory of Environmental Policy, 1st Edition. Cambridge University Press, Cambridge, UK. Bazilian, M., Denny, E., OMalley, M., 2004. Challenges of increased wind energy penetration in ireland. Wind Engineering 28 (1), 4356. Bergmann, A., Hanley, N., Wright, R., 2006. Valuing the attributes of renewable energy investments. Energy Policy 34, 10041014. Bishop, I. D., Miller, D. R., 2007. Visual impact of o-shore wind turbines: the inuence of distance, contrast, movement and social variables. Renewable Energy 32, 814831. Borchiellini, R., Massardo, A., Santarelli, M., 2000. Analytical procedure for carbon tax evaluation. Energy Conversion and Management 41, 15091531. Brooks, D., Lo, E., Zavadil, R., Santoso, S., Smith, J., may 2003. Characterizing the impact of signicant 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

REFERENCES

REFERENCES

Bryans, A., 2006. Impacts of Tidal Stream Devices on Electrical Power Systems. PhD Thesis, The Queens University Belfast. Bryans, A. G., Denny, E., Fox, B., Crossley, P., OMalley, M., 2005. Study of the eect 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., OMalley, M. J., 2006. The predictability and variability eects of renewable generation. IEEE Transactions on Power Systems (in review). Bryans, A. G., Fox, B., Crossley, P., Whittaker, T. J. T., OMalley, 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, 683690. Castro, R., Ferreira, L., 2001. A comparison between chronological and probabilistic methods to estimate wind power capacity credit. IEEE Transactions on Power Systems 16, 904909. Causebrook, A., Fox, B., 2004. Decoding grid codes to accommodate diverse generation technologies. CEPAARB, 2004. California Environmental Protection Agency Air Resources Board (CEPAARB) Report to the Legislator - gas-red power plant NOx emission controls and related environmental impacts. Available: www.arb.ca.gov.

162

REFERENCES

REFERENCES

CER, 2001. Commission for Energy Regulation (CER). Charges for connection to the ESBs 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 department 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 eect 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. Protable operation requires knowing how much it costs to cycle your unit. Available: http://www.psimedia.info/ccj.htm Second Quarter. Connie, 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, Transmission 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

REFERENCES

REFERENCES

CSO, 2006. Central Statistics Oce Ireland (CSO). Census 2006 - preliminary report. 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, 2029. Dale, L., Milborrow, D., Slark, R., Strbac, G., 2004. Total cost estimates for largescale wind scenarios in UK. Energy Policy 32, 19491956. Dalton, S., 2004. Cost comparison IGCC and advanced control.Roundtable on deploying 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 prole 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, 6475. Deloitte, 2005. Review of the electricity sector in Ireland - nal report. Available: http://www.dcmnr.gov.ie. DENA, 2005. Deutsche Energie-Agentur GmbH (DENA).Planning of the grid integration of wind energy in germany onshore and oshore 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 abil164

REFERENCES

REFERENCES

ity of the electricity network to accommodate renewable generation up to 2010. Available: www.energy.detini.gov.uk. Doherty, R., OMalley, M., 2005. New approach to quantify reserve demand in systems with signicant installed wind capacity. IEEE Transactions on Power Systems 20 (2), 587595. Doherty, R., Outhred, H., OMalley, M., Nov. 2005. Generation portfolio analysis for a carbon constrained and uncertain future. In: Future Power Systems. Amsterdam. Doherty, R., Outhred, H., OMalley, M., 2006. Establishing the role that wind may have in future generation portfolios. IEEE Transactions on Power Systems 21 (3), 14151422. Dragoon, K., Milligan, M., May 2003. Assessing wind integration costs with dispatch models: a case study of pacicorp. 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 International 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

REFERENCES

REFERENCES

EirGrid, 2004. Impact of wind power generation in ireland on the operation of conventional 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:

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

http://www.eirgrid.com. 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

REFERENCES ESB Networks, 2005. ESB Networks infrastructure. URL http://www.esb.ie/esbnetworks/

REFERENCES

ESBI, ETSU, 1997. Total renewable energy resource in ireland. Available: http://www.dcmnr.gov.ie. 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 nancial 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

REFERENCES

REFERENCES

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. Forfs, Murphy, A., Barry, F., 2003. Forfas: The National Policy and Advisory a 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 benets: economic advantages of local wind projects. Available: www.iowapolicyproject.org. Gamboa, G., Munda, G., 2007. The problem of wind farm location: a social multicriteria evaluation framework. Energy Policy 35, 15641583. 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 eects 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-theart on methods and software tools for short-term prediction of wind energy production. In: European Wind Energy Conference and Exhibition EWEC. Madrid, Spain. 168

REFERENCES

REFERENCES

Goodstein, E., 2002. Economics and the environment, 3rd Edition. John Wiley and Sons Inc., New York, USA. Gountis, V. P., Bakirtzis, A. G., 2004. Ecient determination of cournot equilibria in electricity markets. IEEE Transactions on Power Systems 19 (4), 18371844. 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 Oce - the impact of climate change on the food supply. Available: http://www.metoce.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, 375384. Harris, B., 1990. Conversion of sulphur dioxide to sulphur trioxide in gas turbine exhaust. Journal of Engineering for Gas Turbine and Power 112, 585589. Hatziargyriou, N., Contaxis, G., Papadopoulos, M., Papadias, B., Matos, M., Lopes, J. P., Nogaret, E., Kariniotakis, G., Halliday, J., Dutton, G., Dokopoulos, P., 169

REFERENCES

REFERENCES

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 systems 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 generation 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), 150156. 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), 638645. Holttinen, H., 2004. The impact of large scale wind power production on the Nordic electricity system - VTT publications 554. Julkaisija, VTT technical research centre of nland. Holttinen, H., Meibom, P., Ensslin, C., Hofmann, L., McCann, J., Pierik, J., Tande, J., Hagstrom, E., Estanqueiro, A., Amaris, H., Sder, L., Strbac, G., Parsons, B., o

170

REFERENCES

REFERENCES

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, 4451. Hurley, T., Watson, R., 2002. An assessment of the expected variability and load following of a large wind penetration in ireland. In: Global Wind Power Conference. Paris, France. Hurtado, J. P., Ferndez, J., Parrondo, J. P., Blanco, E., 2004. Spanish method a of visual impact evaluation in wind farms. Renewable and Sustainable Energy Reviews 8, 483491. 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 brieng note. An Energy and Environment 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), 11941203. 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 denitions. Denition and classication of power system stability. IEEE Transactions on Power Systems 19 (2), 13871401. 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

REFERENCES

REFERENCES

ILEX, Strbac, G., 2002. Quantifying the system cost of additional renewable generation 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 oers for wind generators (CER/04/354). Available: 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., OMalley, M., 2006. Quantifying the impact of connection policy on distributed generation. IEEE Transactions on Energy Conversion (in press). Keane, A., OMalley, M., 2005a. Impact of distribution network constraints on distributed generation capacity. In: Proc. of 40th International Universities Power Engineering Conference. Cork. Keane, A., OMalley, M., 2005b. Optimal allocation of embedded generation on distribution networks. IEEE Transactions on Power Systems 20 (3), 16401646. Keane, A., OMalley, M., Jun. 2006. Optimal distribution generation plant mix with novel loss adjustment factors. In: IEEE PES General Meeting. Montreal, Canada.

172

REFERENCES

REFERENCES

KEMA, May 2005. The contribution to distribution network fault levels from the connection of distributed generation. Tech. Rep. DG/CG/00027/00/00, Department of Trade & Industry UK. Kennedy, S., 2005. Wind power planning: assessing long term costs and benets. Energy Policy 33, 16611675. Kesgin, U., 2003. Study on prediction of the eects of design and operating parameters on NOx emissions from a leanburn natural gas engine. Energy Conversion and Management 44, 907921. Kirby, B., Milligan, M., Hawkins, D., Makarov, Y., Jackson, K., Shu, H., 2003. California 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., OMalley, M., 2005. Frequency control and wind turbin technologies. IEEE Transactions on Power Systems 20 (4), 19051913. Le, K., Jackups, R., Feinstein, J., Thompson, H., Wolf, H., Stein, E., Gorski, A., Grith, J., 1990. A report prepared by the IEEE current operating problems subcommittee: Operational aspects of generation cycling. IEEE Transactions on Power Systems 5 (4), 11941203. Lefton, S., Besuner, P., 2001. Power plant cycling operations and unbundling their eect on plant heat rate. APTECH Technical Paper TP134, Available: http://www.aptecheng.com. 173

REFERENCES

REFERENCES

Lefton, S., Besuner, P., Grimsrud, G., 1997. Understand what it really costs to cycle fossil-red units. Power 141 (2), 4142. 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 maintenance 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-red power generation plant. Control Engineering Practice 12, 707723. 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? http://www.nationalgeographic.com. Malik, A., Awsanjli, A., May 2004. Energy fuel saving benet of a wind turbine. In: IEEE MELECON. Dubrovnik, Croatia, pp. 10411044. Mansour, A., Benjamin, M., Straub, D., Richards, G., 2001. Application of macrolamination technology to lean, premix combustion. AMSE Journal of Engineering for Gas Turbines and Power 123 (4), 796802. 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 forecasting 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). Available:

174

REFERENCES

REFERENCES

Meibom, P., Barth, R., Brand, H., Weber, C., 2006. Deliverable D6.2 (b) Documentation Methodology of the Scenario Tree Tool. Wind power integration 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), 181186. 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. Mller, B., 2006. Changing wind-power landscapes: regional assessment of visual o impact on land use and population in northern jutland, denmark. Applied Energy 83, 477494. Montes, G., Martin, E., Garcia, J., 2007. The current situation of wind energy in spain. Renewable and Sustainable Energy Reviews 11, 467481. Mullane, A., Lightbody, G., Yacamini, R., 2005. Wind-turbine fault ride-through enhancement. IEEE Transactions on Power Systems 20 (4), 19291937. Munksgaard, J., Larsen, A., 1998. Socio-economic assessment of wind power - lessons learned from denmark. Energy Policy 26 (2), 8593. Murphy, A., Walsh, B., autumn 2002. The rationale for subsidising jobs in a fullyemployed economy. Irish Banking Review. NIAER, 2003. Northern Ireland Authority for Energy Regulation (NIAER). Interconnector trading from 1 april 2004 - a decision paper. Available: www.nics.gov.uk. NICS, 2005. Northern Ireland Civil Service (NICS). Northern ireland executive economic overview. Available: www.nics.gov.uk. 175

REFERENCES

REFERENCES

OFGEM, 2003. Oce of Gas and Electricity Markets (OFGEM).Structure of electricity distribution charges - initial conclusions. Available: www.ofgem.gov.uk. OMahony, 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 benets 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 enable increased wind power penetration. In: IEEE PES General Meeting. Toronto, Canada, pp. 19841990. Pepermans, G., Driesen, J., Haeseldonckx, D., Belmans, R., Dhaeseleer, W., 2005. Distributed generation: denition, benets and issues. Energy Policy 33, 787798. Pinson, P., Kariniotakis, G., 2003. Wind power forecasting using fuzzy neural networks 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 electricity. 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. 13051310. REFIT, 2006. Renewable Energy Feed In Tari (REFIT). A competition for electricity generation from biomass, hydro and wind. Available: http://www.dcmnr.gov.ie.

176

REFERENCES

REFERENCES

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. RIS, EWEA, 2005. RIS 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 eciency: 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 comparisons between fossil fuel, nuclear and renewable energy resources for electricity generation. Energy Policy 31, 13151326. 177

REFERENCES

REFERENCES

Smith, J., 2006. Personal communication with Mark OMalley 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. Sder, L., 1993. Reserve margin planning in a wind-hydro-thermal power system. o IEEE Transactions on Power Systems 8, 564571. SONI, 2005. System Operator Northern Ireland (SONI). Seven year generation capacity 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 transboundary air pollution. to abate acidication, 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 Implementation 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

REFERENCES Viridian, 2006. Viridian Group PLC

REFERENCES Website. Available:

http://www.viridiangroup.co.uk. Voorspools, K., Dhaeseleer, W., 2006. Modelling of electricity generation of large interconnected power systems: How can a co2 tax inuence the european generation mix. Energy Conversion and Management 47, 13381358. Vovos, P. N., Harrison, G., Wallace, A., Bialek, J., May 2005. Optimal power ow as a tool for fault level-constrained network capacity analysis. IEEE Trans. Power Syst. 20 (2), 734741. Wallace, A. R., Harrison, G., May 2003. Planning for optimal accommodation of dispersed generation in distribution networks. In: CIRED 17th International Conference 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), 972978. 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, 3743. 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

REFERENCES

REFERENCES

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), 652662. 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, ONeill 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 - nal 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. 12021207.

180

APPENDIX

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 uidised bed drum boiler OT - SC refers to a once through subcritical boiler

181

Appendix A. Generator Information

Fuel Type
hydro hydro hydro hydro hydro hydro hydro hydro hydro hydro hydro hydro hydro hydro hydro coal coal coal coal coal gas gas gas gas gas gas peat peat gas gas peat gas oil oil gas gas gas gas gas oil gas gas oil oil oil oil distillate distillate distillate distillate distillate distillate distillate distillate distillate distillate distillate distillate

182

Unit
Ardnacrusha 1 Ardnacrusha 2 Ardnacrusha 3 Ardnacrusha 4 Erne 1 Erne 2 Erne 3 Erne 4 Lee 1 Lee 2 Lee 3 Liffey 1 Liffey 2 Liffey 3 Liffey 4 Moneypoint 3 Moneypoint 1 Moneypoint 2 Kilroot Kilroot Dublin Bay Power Coolkeeragh CCGT Huntstown Ballylumford CCGT Ballylumford CCGT Poolbeg CCGT West Offaly Power Lough Ree Power North Wall 4 Marina Edenderry Aghada Tarbert 4 Tarbert 3 Poolbeg 3 Poolbeg 1 Ballylumford 5 Poolbeg 2 Ballylumford 6 Gt. Island 3 Ballylumford 1 Aghada OCGT Tarbert 1 Tarbert 2 Gt. Island 1 Gt. Island 2 Aghada Rhode 1 Rhode 2 Asahi Aghada CT Aghada CT North Wall 5 Coolkeeragh Kilroot Kilroot Ballylumford 2 Ballylumford 3

Min
12 12 12 12 4 4 5 5 3 1 3 3 3 0 0 115 115 115 64 64 200 260 223 116 116 280 46 40 112 77 40 35 35 35 56 55 56 55 63 30 56 10 25 25 25 25 5 5 5 5 10 10 5 8 5 5 8 8

Year of Boiler Reserve CommMax GJ/MWh Euro/MWh Reserve Mid Fuel Prices Euro/MWh Type ission
21 22 19 24 10 10 23 23 15 4 8 15 15 4 4 285 285 285 201 201 396 404 343 240 240 480 137 91 163 112 118 258 241 241 242 110 170 110 103 108 170 88 54 54 54 54 52 52 52 52 90 88 108 53 29 29 53 53 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 11.03 11.05 11.07 11.10 11.10 6.55 6.60 6.65 6.83 6.83 6.99 9.86 10.07 9.60 9.71 13.07 10.00 10.70 10.90 11.00 11.15 11.20 11.30 11.35 11.70 11.62 12.15 12.84 12.86 12.87 12.91 14.12 10.63 10.63 10.63 12.14 12.15 12.90 14.20 14.35 14.35 14.45 14.45 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 23.78 23.82 23.86 23.92 23.92 27.26 27.47 27.69 28.41 28.41 29.10 30.94 31.61 39.96 40.41 41.01 41.62 44.19 45.02 45.78 46.41 46.61 47.03 47.24 48.32 48.36 50.57 53.02 53.11 53.15 53.32 59.63 83.92 83.92 84.90 97.16 97.37 101.41 112.72 113.96 113.96 115.73 115.73 4 4 4 4 3 3 5 5 0 0 0 1 1 0 0 45 45 45 30 30 42 40 25 37 37 150 9 12 32 35 9 20 25 25 20 8 35 8 8 20 35 20 3 3 3 3 14 20 20 0 20 20 72 16 7 7 15 15 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 1.6 1.6 1.6 1.6 1.6 2.6 2.6 2.6 2.6 2.6 2.6 2 2 2.6 2.6 2 1.8 1.8 1.8 1.7 1.7 1.7 1.7 2.6 1.7 1.7 1 1.7 1.7 1.7 1.7 1 1 1 1 1 1 1 1 1 1 1 1 1929 1929 1929 1929 1950 1950 1950 1950 1957 1957 1957 1944 1944 1944 1944 1985 1985 1985 1981 1981 2002 2005 2002 2002 2002 2000 2004 2004 1982 1979 2000 1981 1977 1977 1981 1971 1968 1971 1968 1971 1968 1982 1969 1969 1967 1967 1981 unknown unknown unknown 1982 1982 1983 1959 1982 1982 1968 1968 NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA SC-D SC-D SC-D SC-D SC-D HRSG HRSG HRSG HRSG HRSG HRSG FB-D FB-D HRSG HRSG FB OT-SC OT-SC OT-SC OT-SC OT-SC OT-SC OT-SC OT-SC SC-D OT-SC NA SC-D SC-D SC-D SC-D NA NA NA NA NA NA NA NA NA NA NA NA

Figure A.1: Generator Information for 2007

Appendix A. Generator Information

183
Year of Boiler CommEuro/MWh Type ission
0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 1.6 1.6 1.6 1.6 1.6 2.6 2.6 2.6 2.6 2.6 2.6 2.6 2.6 2 2 2.6 2.6 2 2.6 1.8 1.8 1.8 1.7 1.7 1.7 1.7 2.6 1.7 1.7 1 1.7 1.7 1.7 1.7 1 1 1 1 1 1 1 1 1 1 1 1 1929 1929 1929 1929 1950 1950 1950 1950 1957 1957 1957 1944 1944 1944 1944 1985 1985 1985 1981 1981 2002 2006 2005 2002 2002 2002 2000 2007 2004 2004 2006 1982 2000 1979 1981 1977 1977 1981 1971 1971 1968 1968 1971 1968 1982 1969 1969 1967 1967 1981 unknown unknown unknown 1982 1982 1983 1959 1982 1982 1968 1968 NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA SC-D SC-D SC-D SC-D SC-D HRSG HRSG HRSG HRSG HRSG HRSG HRSG HRSG FB-D FB-D HRSG HRSG FB HRSG OT-SC OT-SC OT-SC OT-SC OT-SC OT-SC OT-SC OT-SC SC-D OT-SC NA SC-D SC-D SC-D SC-D NA NA NA NA NA NA NA NA NA NA NA NA

Unit
Ardnacrusha 1 Ardnacrusha 2 Ardnacrusha 3 Ardnacrusha 4 Erne 1 Erne 2 Erne 3 Erne 4 Lee 1 Lee 2 Lee 3 Liffey 1 Liffey 2 Liffey 3 Liffey 4 Moneypoint 3 Moneypoint 1 Moneypoint 2 Kilroot Kilroot Dublin Bay Power Tynagh CCGT Coolkeeragh CCGT Huntstown Ballylumford CCGT Ballylumford CCGT Poolbeg CCGT Huntstown 2 West Offaly Power Lough Ree Power Aughinish CCGT North Wall 4 Edenderry Marina Aghada Tarbert 4 Tarbert 3 Poolbeg 3 Poolbeg 1 Poolbeg 2 Ballylumford 5 Ballylumford 6 Gt. Island 3 Ballylumford 1 Aghada OCGT Tarbert 1 Tarbert 2 Gt. Island 1 Gt. Island 2 Aghada Rhode 1 Rhode 2 Asahi Aghada CT Aghada CT North Wall 5 Coolkeeragh Kilroot Kilroot Ballylumford 2 Ballylumford 3

Fuel Type
hydro hydro hydro hydro hydro hydro hydro hydro hydro hydro hydro hydro hydro hydro hydro coal coal coal coal coal gas gas gas gas gas gas gas gas peat peat gas gas peat gas gas oil oil gas gas gas gas gas oil gas gas oil oil oil oil distillate distillate distillate distillate distillate distillate distillate distillate distillate distillate distillate distillate

Reserve Min Max GJ/MWh Euro/MWh Reserve

Mid Fuel Prices

12 12 12 12 4 4 5 5 3 1 3 3 3 0 0 115 115 115 64 64 200 202 260 223 116 116 280 200 46 40 40 112 40 77 35 35 35 56 55 55 56 63 30 56 10 25 25 25 25 5 5 5 5 10 10 5 8 5 5 8 8

21 22 19 24 10 10 23 23 15 4 8 15 15 4 4 285 285 285 201 201 396 404 404 343 240 240 480 401 137 91 150 163 118 112 258 241 241 242 110 110 170 103 108 170 88 54 54 54 54 52 52 52 52 90 88 108 53 29 29 53 53

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 11.03 11.05 11.07 11.10 11.10 6.55 6.57 6.60 6.65 6.83 6.83 6.99 7.00 9.86 10.07 8.72 9.60 13.07 9.71 10.00 10.70 10.90 11.00 11.15 11.30 11.20 11.35 11.70 11.62 12.15 12.84 12.86 12.87 12.91 14.12 10.63 10.63 10.63 12.14 12.15 12.90 14.20 14.35 14.35 14.45 14.45

0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 23.94 23.99 24.03 24.09 24.09 28.62 28.71 28.84 29.07 29.83 29.83 30.56 30.59 31.84 32.54 38.12 41.95 42.22 42.43 43.70 45.69 46.54 46.97 47.61 48.25 48.94 49.60 49.96 50.78 53.10 54.82 54.91 54.95 55.12 61.70 85.93 85.93 85.93 98.06 98.17 104.21 114.72 115.96 115.96 116.73 116.73

4 4 4 4 3 3 5 5 0 0 0 1 1 0 0 45 45 45 30 30 42 32 40 25 37 37 150 50 9 12 10 32 9 35 20 25 25 20 8 8 35 8 20 35 20 3 3 3 3 14 20 20 0 20 20 72 16 7 7 15 15

Figure A.2: Generator Information for 2010

Appendix A. Generator Information

Fuel Type
hydro hydro hydro hydro hydro hydro hydro hydro hydro hydro hydro hydro hydro hydro hydro coal coal coal coal coal gas gas gas gas gas gas gas gas peat gas gas peat gas peat gas gas oil gas oil oil gas gas gas gas gas oil oil gas gas distillate distillate distillate distillate distillate distillate distillate distillate

184

Unit
Ardnacrusha 1 Ardnacrusha 2 Ardnacrusha 3 Ardnacrusha 4 Erne 1 Erne 2 Erne 3 Erne 4 Lee 1 Lee 2 Lee 3 Liffey 1 Liffey 2 Liffey 3 Liffey 4 Moneypoint 3 Moneypoint 1 Moneypoint 2 Kilroot Kilroot Dublin Bay Power Tynagh CCGT Coolkeeragh CCGT Huntstown New CCGT (1) New CCGT (2) Ballylumford CCGT Ballylumford CCGT West Offaly Power Poolbeg CCGT Huntstown 2 Lough Ree Power Aughinish CCGT Edenderry North Wall 4 Marina Tarbert 4 Aghada Tarbert 3 Gt. Island 3 Poolbeg 3 Poolbeg 1 Ballylumford 5 Poolbeg 2 Ballylumford 1 Tarbert 1 Tarbert 2 Aghada New OCGT Aghada CT Aghada CT Aghada CT North Wall 5 Kilroot Kilroot Ballylumford 2 Ballylumford 3

Year of Boiler Reserve CommMin Max GJ/MWh Euro/MWh Reserve Mid Fuel Prices Euro/MWh Type ission
12 12 12 12 4 4 5 5 3 1 3 3 3 0 0 115 115 115 64 64 200 202 260 223 200 200 116 116 46 280 200 40 40 40 112 77 35 35 35 30 56 55 56 55 56 25 25 5 10 10 10 10 5 5 5 8 8 21 22 19 24 10 10 23 23 15 4 8 15 15 4 4 285 285 285 201 201 396 404 404 343 450 420 240 240 137 480 401 91 150 118 163 112 241 258 241 108 242 110 170 110 170 54 54 52 160 90 88 88 108 29 29 53 53 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 11.03 11.05 11.07 11.10 11.10 6.55 6.57 6.60 6.65 6.71 6.72 6.83 6.83 9.86 6.99 7.00 10.07 8.72 13.07 9.60 9.71 10.70 10.00 10.90 11.70 11.00 11.15 11.20 11.30 11.62 12.84 12.86 14.12 10.63 12.14 12.15 12.15 12.90 14.35 14.35 14.45 14.45 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 24.49 24.54 24.58 24.64 24.64 33.01 33.11 33.26 33.53 33.82 33.87 34.40 34.40 34.80 35.24 35.28 35.56 43.96 46.14 48.38 48.94 50.40 50.40 51.34 55.11 55.44 56.20 56.45 56.95 58.56 60.47 60.57 71.16 94.86 108.25 108.38 108.38 115.04 128.01 128.01 128.86 128.86 4 4 4 4 3 3 5 5 0 0 0 1 1 0 0 45 45 45 30 30 42 32 40 25 20 20 37.1 37.1 9 150 50 12 10 9.4 32 35 25 20 25 20 20 8 35 8 35 3 3 14 15.5 20 20 20 72 7.25 7.25 14.5 14.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 1.6 1.6 1.6 1.6 1.6 2.6 2.6 2.6 2.6 1 1 2.6 2.6 2 2.6 2.6 2 2.6 2 2.6 2.6 1.8 1.8 1.8 1.7 1.7 1.7 1.7 1.7 1.7 1.7 1.7 1 1 1 1 1 1 1 1 1 1 1929 1929 1929 1929 1950 1950 1950 1950 1957 1957 1957 1944 1944 1944 1944 1985 1985 1985 1981 1981 2002 2006 2005 2002 2012 2012 2002 2002 2004 2000 2007 2004 2006 2000 1982 1979 1977 1981 1977 1971 1981 1971 1968 1971 1968 1969 1969 1982 2012 1982 1982 1982 1983 1982 1982 1968 1968 NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA SC-D SC-D SC-D SC-D SC-D HRSG HRSG HRSG HRSG HRSG HRSG HRSG HRSG FB-D HRSG HRSG FB-D HRSG FB HRSG HRSG OT-SC OT-SC OT-SC SC-D OT-SC OT-SC OT-SC OT-SC OT-SC SC-D SC-D NA NA NA NA NA NA NA NA NA NA

Figure A.3: Generator Information for 2015

Appendix A. Generator Information

Fuel Type
hydro hydro hydro hydro hydro hydro hydro hydro hydro hydro hydro hydro hydro hydro hydro coal coal coal coal coal peat peat gas gas gas gas gas gas gas gas gas gas gas gas peat gas oil oil gas gas gas gas oil gas gas gas gas gas gas gas gas gas distillate distillate distillate distillate distillate distillate distillate

185
Year of Boiler CommEuro/MWh Type ission
0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 1.6 1.6 1.6 1.6 1.6 2 2 2.6 2.6 2.6 2.6 1 1 1.7 1.7 2.6 2.6 2.6 2.6 2 2.6 1.8 1.8 2.6 2.6 2.6 1.8 1.7 1 1 1.7 1.7 1.7 1.7 1.7 0 1 1 1 1 1 1 1 1 1929 1929 1929 1929 1950 1950 1950 1950 1957 1957 1957 1944 1944 1944 1944 1985 1985 1985 1981 1981 2004 2004 2002 2006 2005 2002 2012 2012 2018 2018 2002 2002 2000 2007 2000 2006 1977 1981 1982 2013 1979 1981 1971 2012 2018 1981 1971 1968 1971 1968 2018 1982 1968 1968 1982 1982 1982 1983 1982 NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA SC-D SC-D SC-D SC-D SC-D FB-D FB-D HRSG HRSG HRSG HRSG HRSG HRSG HRSG HRSG HRSG HRSG HRSG HRSG FB HRSG OT-SC OT-SC HRSG HRSG HRSG OT-SC SC-D NA NA OT-SC OT-SC OT-SC OT-SC OT-SC NA NA NA NA NA NA NA NA NA

Unit
Ardnacrusha 1 Ardnacrusha 2 Ardnacrusha 3 Ardnacrusha 4 Erne 1 Erne 2 Erne 3 Erne 4 Lee 1 Lee 2 Lee 3 Liffey 1 Liffey 2 Liffey 3 Liffey 4 Moneypoint 3 Moneypoint 1 Moneypoint 2 Kilroot Kilroot West Offaly Power Lough Ree Power Dublin Bay Power Tynagh CCGT Coolkeeragh CCGT Huntstown CCGT (1) CCGT (2) New CCGT or similar New CCGT or similar Ballylumford CCGT Ballylumford CCGT Poolbeg CCGT Huntstown 2 Edenderry Aughinish CCGT Tarbert 4 Tarbert 3 North Wall 4 New CCGT Marina Aghada Gt. Island 3 OCGT New OCGT Poolbeg 3 Poolbeg 1 Ballylumford 5 Poolbeg 2 Ballylumford 1 New OCGT Aghada Ballylumford 2 Ballylumford 3 Aghada CT Aghada CT Aghada CT North Wall 5 Kilroot

Reserve Min Max GJ/MWh Euro/MWh Reserve

Mid Fuel Prices

11.9 11.9 11.9 11.9 4 4 5 5 3 1 3 3 3 0.4 0.2 115 115 115 64 64 46 40 200 202 260 222.8 200 200 220 230 116 116 280 200 40 40 35 35 112 63 77 35 30 10 10 56 55 56 55 56 15 5 8 8 10 10 10 5 5

21 22 19 24 10 10 22.5 22.5 15 4 8 15 15 4 4 285 285 285 201 201 137 91 396 404 404 342.7 450 420 400 400 240 240 480 401 117.6 150 241 241 163 103 112 258 108 160 80 242 110 170 110 170 52 52 53 53 90 88 88 108 29

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 11.03 11.05 11.07 11.10 11.10 9.86 10.07 6.55 6.57 6.60 6.65 6.71 6.72 6.79 6.79 6.83 6.83 6.99 7.00 13.07 8.72 10.70 10.90 9.60 9.70 9.71 10.00 11.70 10.63 10.63 11.00 11.15 11.20 11.30 11.62 14.12 14.12 14.45 14.45 12.14 12.15 12.15 12.90 14.35

0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 25.49 25.53 25.58 25.64 25.64 40.91 41.81 41.99 42.11 42.31 42.64 43.01 43.08 43.52 43.52 43.75 43.75 44.82 44.87 54.24 55.91 60.13 61.26 61.54 62.18 62.24 64.10 65.75 94.86 68.17 70.51 71.47 71.79 72.43 74.48 90.51 90.51 92.60 92.60 129.25 129.40 129.40 137.35 152.84

4 4 4 4 3 3 5 5 0 0 0 1 1 0 0 45 45 45 30 30 9 12 42 32 40 25 20 20 3 3 37.1 37.1 150 50 9.4 10 25 25 32 8 35 20 20 15.5 15.5 20 8 35 8 35 0 14 14.5 14.5 20 20 20 72 7.25

Figure A.4: Generator Information for 2020

Appendix A. Generator Information

Unit Moneypoint 3 Moneypoint 1 Moneypoint 2 Kilroot Kilroot Dublin Bay Power Tynagh CCGT Coolkeeragh CCGT Huntstown Ballylumford CCGT Ballylumford CCGT Poolbeg CCGT West Offaly Power Lough Ree Power Aughinish CCGT North Wall 4 Marina Edenderry Aghada Tarbert 4 Tarbert 3 Poolbeg 3 Poolbeg 1 Ballylumford 5 Poolbeg 2 Ballylumford 6 Gt. Island 3 Ballylumford 1 Aghada OCGT Tarbert 1 Tarbert 2 Gt. Island 1 Gt. Island 2 Aghada Rhode 1 Rhode 2 Asahi Aghada CT Aghada CT North Wall 5 Coolkeeragh Kilroot Kilroot Ballylumford 2 Ballylumford 3 No Load (GJ/hour) 148 148 148 213 213 533 467 496 350 496 496 716 124 90 506 347 251 498 187 258 258 274 84 179 88 179 105 98 87 50 50 51 51 187 85 85 85 268 268 317 177 101 101 180 180 Cold Start GJ 14620 14620 14620 2247 2247 7700 2811 5000 1000 50 50 100 500 320 2900 80 50 2010 4302 3180 3180 4302 1025 2124 1025 2124 743 50 50 562 562 562 562 4302 50 50 50 63 63 50 16 8 8 16 16 Warm Start GJ 6920 6920 6920 1645 1645 2600 1633 2000 650 50 50 100 500 320 1700 80 50 1084 2185 1934 1934 2185 625 1527 625 1527 600 50 50 449 449 449 449 2185 50 50 50 63 63 50 16 8 8 16 16

186
Cold Start GJ 4360 4360 4360 973 973 2600 1144 1000 500 50 50 100 500 320 1150 80 50 436 1273 1072 1072 1273 353 847 353 847 293 50 50 218 218 218 218 1273 50 50 50 63 63 50 16 8 8 16 16

Figure A.5: Generator Operating Characteristics

APPENDIX

The Optimisation of the Distribution Network

This Appendix sets out the methodology employed in Keane and OMalley (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 N

J=
j=1 i=1

PAvail j P lantij LFj ConnCostij i

(B.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. i 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 rm and non rm cases and are shown in Equations (B.2), (B.3), (B.5) and (B.6) (Keane et al., 2006). The only dierence between the rm and non rm case is the voltage constraint (B.9), which is not included in the non rm 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 owing 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 exceed 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).

Appendix B. Optimisation of the Distribution Network

189

PT x PT raf oCap .

(B.3)

Where PT x refers to power ow 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 nd 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 conditions. 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

jT x PDG j + T x SCLRated .
j=1

(B.5)

Where jT 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, T x is the initial SCL at the transmission bus with no generation present.

Appendix B. Optimisation of the Distribution Network

190

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

PDG i 0.1 cos()

j=1

ji PDG j 0.1 cos()i .

(B.6)

Where cos() 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 Eect

The voltage at the generator is given by Equation (B.7). XPL RQL RPL + XQL +j VL VL

VG = VL +

(B.7)

Where PDG and QDG are the active and reactive power of the generator respectively, 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 respectively. 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 ows along that line. It is evident that the larger the impedance and power ow the larger the voltage rise. The increased active power ows 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

Appendix B. Optimisation of the Distribution Network

191

ith bus. The relationship between voltage and power injections at each bus is determined. As MWs are added at each bus the voltage rises. Increasing levels of generation are added incrementally at each bus in turn and load ow 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

i PDG i + i +
j=1

ji PDG j Vmax i i N, i = j.

(B.9)

Where i 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. i refers to the initial voltage level at the ith bus with no generation, ji 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 identiable.

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 signicant 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 eect (Hird et al., 2004; Hill et al., 2005). These techniques along with others facilitate non rm management of the voltage constraint. The short circuit level may be dominant in more urban

Appendix B. Optimisation of the Distribution Network

192

areas. However, active management of fault levels is some way o and is likely to be very expensive (KEMA, 2005). Hence, it is non rm management of the voltage constraint which is considered here.