Loading...
HomeMy WebLinkAboutThe Costs & Impacts of Intermittency UK 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 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, with financial support from the Carbon Trust Robert Gross Philip Heptonstall Dennis Anderson Tim Green Matthew Leach Jim Skea March 2006 Preface This report has been produced by the UK Energy Research Centre's Technology and Policy Assessment (TPA) function. The TPA was set up to inform decision making processes and address key controversies in the energy field. It aims to provide authoritative and accessible reports that set very high standards for rigour and transparency. The subject of the report was chosen after extensive consultation with energy sector stakeholders. It addresses the following question: What is the evidence on the costs and impacts of intermittent generation on the UK electricity network, and how are these costs assigned? This UKERC report was part funded by the Carbon Trust and was undertaken by a team of experts from Imperial College London and the Supergen Future Network Technologies Consortium. The work was overseen by a panel of experts, and provides a systematic review of more than 200 reports and studies from around the world. The report provides a detailed review of the current state of understanding of the engineering and economic impacts of intermittent, or renewable energy sources, such as wind and solar power. It seeks to provide a review of this complex topic that is accessible to the non-specialist. This report is the first output of the UKERC’s TPA function, which was established to produce a wide variety of policy relevant reports on the energy sector to stimulate and inform debate between policymakers, researchers and the wider energy community. About UKERC It is the UK Energy Research Centre's (UKERC) mission to be the UK's pre-eminent centre of research, and source of authoritative information and leadership, on sustainable energy systems. UKERC undertakes world-class research addressing the whole-systems aspects of energy supply and use while developing and maintaining the means to enable cohesive research in energy. To achieve this we are establishing a comprehensive database of energy research, development and demonstration competences in the UK. We will also act as the portal for the UK energy research community to and from both UK stakeholders and the international energy research community. Executive Summary Overview The output of many types of renewable electricity generation, such as wind, wave and solar, is intermittent in nature. Output varies with environmental conditions, such as wind strength, over which the operator has no control. Assimilating these fluctuations has the potential to affect the operation and economics of electricity networks, markets and the output of other forms of generation. It can affect the reliability of electricity supplies and the actions needed to ensure demand meets supply every instant. This report aims to understand and quantify these impacts, and therefore addresses the question ‘What is the evidence on the impacts and costs of intermittent generation on the British electricity network, and how are these costs assigned?’ It is based on a review of over 200 international studies. The studies have been categorised and assessed. The review process has been overseen by an expert group and the final report has been peer-reviewed by international experts. Stakeholders were consulted through a workshop, and materials produced throughout the assessment process were posted on the UKERC website. This study focuses only on the electricity system implications of the uncontrollable variability of some renewable energy sources, often referred to as intermittency’. It therefore does not address: the basic costs of renewable generation relative to conventional generation; the environmental impacts of renewable generation; or the direct costs of extending the transmission system to accommodate new generation. The report focuses on incremental developments to the existing electricity system, with a timeframe approximately twenty years into the future. It does not consider the long term potential to reconfigure electricity networks in order to maximise the use of sustainable energy technologies, nor the costs or options for doing so. The benefits of renewable generation 5. Renewable electricity generation helps to reduce the need to operate power stations burning fossil fuels such as coal and gas. This means that carbon dioxide emissions are reduced. It is sometimes said that wind energy, for example, does not reduce carbon dioxide emissions because the intermittent nature of its output means it needs to be backed up by fossil fuel plant. Wind turbines do not displace fossil generating capacity on a one-for-one basis. But it is unambiguously the case that wind energy can displace fossil fuel-based generation, reducing both fuel use and carbon dioxide emissions. Wind generation does mean that the output of fossil fuel-plant needs to be adjusted more frequently, to cope with fluctuations in output. Some power stations will be operated below their maximum output to facilitate this, and extra system balancing reserves will be needed. Efficiency may be reduced as a result. At high penetrations (above 20%) energy may need to be ‘spilled’ because the electricity system cannot always make use of it. But overall these effects are much smaller than the savings in fuel and emissions that renewables can deliver at the levels of penetration examined in this report. ‘Terminology is controversial. many lean towards the term ‘variable’ others toward ‘intermittent’. Neither term is perfect: the Impacts on reliability of electricity systems None of the 200+ studies reviewed suggest that introducing significant levels of intermittent renewable energy generation on to the British electricity system must lead to reduced reliability of electricity supply’. Many of the studies consider intermittent generation of up to 20% of electricity demand, some considerably more. It is clear that intermittent generation need not compromise electricity system reliability at any level of penetration foreseeable in Britain over the next 20 years, although it may increase costs. In the longer term much larger penetrations may also be feasible given appropriate changes to electricity networks, but this report does not explore the evidence on this topic. The introduction of significant amounts of intermittent generation will affect the way the electricity system operates. There are two main categories of impact and associated cost. The first, so called system balancing impacts, relates to the relatively rapid short term adjustments needed to manage fluctuations over the time period from minutes to hours. The second, which is termed here ‘reliability impacts’, relates to the extent to which we can be confident that sufficient generation will be available to meet peak demands. No electricity system can be 100% reliable, since there will always be a small chance of major failures in power stations or transmission lines when demands are high. Intermittent generation introduces additional uncertainties, and the effect of these can be quantified. System balancing impacts 10. The vast bulk of electricity in Britain is supplied through market arrangements comprising bilateral contracts of varying durations between generators and suppliers (wholesalers of electricity). However relatively small, but crucial, adjustments are needed to ensure demand and supply balance each instant. These are made by the system operator, the company with a statutory duty to ensure that electricity supply continuously meets demand. The system operator balances the system by purchasing services from generators or adjustable loads. To ensure these services are available in the timescales required, the system operator enters into contracts for system balancing reserves. System balancing entails costs which are passed on to electricity consumers. Intermittent generation adds to these costs. For penetrations of intermittent renewables up to 20% of electricity supply, additional system balancing reserves due to short term (hourly) fluctuations in wind generation amount to about 5-10% of installed wind capacity. Globally, most studies estimate that the associated costs are less than £5/MWh of intermittent output, in some cases substantially less. The range in UK relevant studies is £2 - £3/MWh. System reliability impacts and additional system capacity requirements 12. 13. To maintain reliability of supplies in an electricity system, peak demand must not exceed the production capability of the installed generation at that moment. Historically central planners sought to ensure that installed generation capacity could meet forecast peak demand within a planning horizon. In liberalised markets, individual market participants are responsible only for ensuring adequate generation capacity is available to meet their own contracts to supply electricity. In either case,a system margin can be measured which is the amount by which the total installed capacity of all the generating plant on the system exceeds the anticipated peak demand. Unless there is a large amount of responsive or controllable demand, a system margin is needed to cope with unavailability of installed generation and fluctuations in electricity requirements (e.g. due to the weather). Conventional plant — coal, gas, nuclear — cannot be completely relied upon to generate electricity at times of peak demand as there is, very approximately, a one-in-ten chance that unexpected failures (or “forced outages”) in power plant or electricity transmission networks will cause any individual conventional generating unit not to be available to generate power. Even with a system margin. there is 14. 15. 16. 17, 19. The risk of demand being unmet can be characterised statistically, and the measure commonly used to quantify this risk is called Loss of Load Probability (LOLP). This measures the likelihood that any load (demand) is not met, and it is usually a requirement of electricity systems that LOLP is kept small’. Intermittent generation increases the size of the system margin required to maintain a given level of reliability. This is because the variability in output of intermittent generators means they are less likely to be generating at full power at times of peak demand. The system margin needed to achieve a desired level of reliability depends on many complex factors but may be explored by statistical calculations or simplified models. Intermittent generation introduces new factors into the calculations and changes some of the numbers, but it does not change the fundamental principles on which such calculations are based. Intermittent generators can make a contribution to system reliability, provided there is some probability of output during peak periods. They may be generating power when conventional stations experience forced outages and their output may be independent of fluctuations in energy demand. These factors can be taken into account when the relationship between system margin and reliability is calculated using statistical principles. There is some debate over the extent to which existing measures of reliability, particularly LOLP, fully capture the changes that arise when intermittent sources are added to the network. This is because intermittent generation changes the nature of the unreliability that may arise (for example, increasing the number of occasions in which relatively small curtailments of demand may be required). These aspects may be represented by using different statistics to calculate risk, in addition to a simple LOLP. Capacity credit is a measure of the contribution that intermittent generation can make to reliability. It is usually expressed as a percentage of the installed capacity of the intermittent generators. There is a range of estimates for capacity credits in the literature and the reasons for there being a range are well understood. The range of findings relevant to British conditions is approximately 20 — 30% of installed capacity when up to 20% of electricity is sourced from intermittent supplies (usually assumed to be wind power). Capacity credit as a percentage of installed intermittent capacity declines as the share of electricity supplied by intermittent sources increases. The capacity credit for intermittent generation, the additional conventional capacity required to maintain a given level of reliability and thus the overall system margin are all related to each other. The smaller the capacity credit, the more capacity needed to maintain reliability, hence the larger the system margin. The amount by which the system margin must rise in order to maintain reliability has been described in some studies as “standby capacity”, “back-up capacity” or the “system reserves”. But there is no need to provide dedicated “back-up” capacity to support individual generators. These terms have meaning only at the system level. Costs of maintaining reliability 20. The additional capacity to maintain reliability entails costs over and above the direct cost of generating electricity from intermittent sources. There has been some controversy over how to estimate the costs associated with the additional thermal capacity required to maintain reliability. In part this reflects the fact that under current market arrangements there is no single body with responsibility to purchase system margin. This is one reason why costs are less transparent than they are for system balancing services. Some studies have assessed the costs of the capacity required to maintain reliability based on assumptions about the nature of plant providing ‘system reserves’. Others have assessed only the change in the total costs of the electricity system as a whole’. There is broad agreement between both approaches on the total change to system costs. ‘e.g. the LOLP of the pre-privatised electricity system in Great Britain was planned not to exceed 9% - nine winters per 21. 22. 23. 24. We have identified the need for an agreed definition for reporting the ‘system reliability costs of intermittency’. We suggest that this be based on the difference between the contribution to reliability made by intermittent generation plant and the contribution to reliability made by conventional generation plant. This comparison should be drawn between plants that provide the same amount of energy when operated at maximum utilisation. This provides a measure of the cost of maintaining system reliability and is in addition to the direct costs of intermittent plant. In the main text and Annex 2, we explore this relationship in depth and show that it can be expressed as follows: System reliability cost = fixed cost of energy-equivalent thermal plant (e.g. CCGT) minus avoided fixed cost of thermal plant (e.g. CCGT) displaced by the capacity credit of intermittent plant (e.g. wind). It should be noted that all forms of generation have the potential to impact on system costs, and this is an important topic for ongoing and future research’. The comparison with conventional generating plant at maximum utilisation (i.e. on ‘baseload’) is crucial to this calculation. Policymakers and others often seek to compare the average costs of different types of generating plant on a ‘like with like’ basis. For example, they may wish to compare the cost of wind power with the cost of coal power. This comparison uses levelised costs (£/MWh) that assume that plants are operating at maximum utilisation. If intermittency costs are calculated in any other way there is a danger that comparisons of this nature will not be meaningful. Using the definition set out in paragraph 21, the cost to maintain system reliability lies within the range £3 - £5/MWh under British conditions. Again, relative to a comparitor plant operated at maximum utilisation. Impacts can also be expressed in MW terms; additional conventional capacity to maintain system reliability during demand peaks amounts to around 15% to 22% of installed intermittent capacity. This assumes around 20% of electricity is supplied by well dispersed wind power. Current costs are much lower; indeed there is little or no impact on reliability at existing levels of wind power penetration. The cost of maintaining reliability will increase as the market share of intermittent generation rises. Comparing different electricity systems 25. It is tempting to read across the results of studies on intermittency costs from one country to another, or from one system to another. This can be another source of controversy. The greatest care must be taken in trying to make such comparisons. The impacts and costs of intermittent generation can be assessed only in the context of the particular type of system in which they are embedded. The impacts depend on: — The quality of the environmental resource on which renewable generation depends, for example the strength of the wind and the degree to which it fluctuates. — The robustness of the electricity grid and the capacity to transfer power from generators to consumers. — Regulatory and operating practices, in particular how far ahead the use of system balancing reserve is planned (known as ‘gate closure’). The closer to real time reserves are committed, the more reliable will be forecasts of intermittent generation, which can reduce the need for more expensive fast-acting reserve. — Accuracy of forecasting of intermittent output. Better forecasting can improve the efficiency with which intermittency is managed, both by the system operator after gate closure and by markets over longer timescales. Weather patterns in some regions are more predictable than in others. — The extent to which intermittent generators are geographically dispersed or are located in a particular area. If wind generators are located close together their output will tend to fluctuate up and down at the same time, increasing variability of the total output and increasing the costs of both system halancina and mainrsining ralishilitv 26. Some conditions in Britain (quality of wind resource, robustness of the grid, relatively late gate closure) will tend to mitigate the impacts of intermittency and keep associated costs relatively low. Others (notably the relative lack of interconnection and relatively small geographical area over which resources are dispersed) will tend to increase the costs of managing the system relative to other regions. Comparisons between Britain and other countries must be treated with the greatest of caution. Intermittency costs in Britain 27. 28. The aggregate ‘costs of intermittency’ are made up of additional short-run balancing costs and the additional longer term costs associated with maintaining reliability via an adequate system margin. Intermittency costs in Britain are of the order of £5 to £8/MWh, made up of £2 to £3/MWh from short- run balancing costs and £3 to £5/MWh from the cost of maintaining a higher system margin. For comparison, the direct costs of wind generation would typically be approximately £30 to £55/MWh. If shared between all consumers the impact of intermittency on electricity prices would be of the order 0.1 to 0.15 p/kWh. These estimates assume that intermittent generation is primarily wind, that it is geographically widespread, and that it accounts for no more than about 20% of electricity supply. At current penetration levels costs are much lower, since the costs of intermittency rise as penetrations increase. If intermittent generation were clustered geographically, or if the market share were to rise above 20%, intermittency costs would rise above these estimates, and/or more radical changes would be needed in order to accommodate renewables. Recommendations for reporting the costs of intermittency 29. When reporting the costs associated with intermittent electricity generation, we recommend that: a) there be a clear statement of which costs are included and those which are excluded, i.e. short-run balancing costs versus long-run capacity requirements; b) there be a clear statement of the methodological basis for calculating intermittency impacts; ¢) when comparing the costs of intermittent sources versus baseload conventional generation the method described in paragraph 21 be used; d) that the context of the system into which intermittent generation is being embedded be clearly described. Recommendations for UK-relevant research and policy 30. 31. We recommend that additional steps are put in place to continuously monitor the effect of intermittent generation on system margin and existing measures of reliability. The effectiveness of market mechanisms in delivering adequate system margin also needs to be kept under review. Intermittent generation can make a valuable contribution to energy supplies, but to ensure reliability of supply, additional investment in thermal capacity is also required. In the short run older plant is likely to provide system margin but, in the long run, investment in new capacity will be needed. Flexible and reliable generation is an ideal complement to intermittent renewables. Policy should encourage and not impede investment in plant that is well suited to complement renewable energy sources and contribute to both reliable operation and efficient system balancing. 32. We recommend that more research be undertaken on the following topics: — Renewable energy deployment scenarios in which intermittent generation is clustered in particular regions of the UK, including the system impacts of very large offshore wind farms. — Measures of reliability appropriate to intermittent sources. In particular the merits of, and options for, going beyond ‘loss of load probability’, (LOLP) in characterising the reliability of an electricity system at high levels of intermittent generation. LOLP measures the likelihood of a capacity shortfall rather than its severity. — Using these improved measures of reliability, there is a need for on-going monitoring of the British market to assess how actual market response (i.e. decisions to invest in new generation or maintain existing generation in-service) compare to those that would be consistent with the improved reliability measures. — The definition of an agreed convention for reporting the costs associated with maintaining system reliability. — Further work on the development of methodologies for assessing the system cost implications of new generating technologies (intermittent or otherwise), in terms of the impacts on the utilisation of incumbent generation. — The extent to which intervention may be needed to ensure that adequate investment in appropriate thermal plant to maintain reliability is delivered, and the policy options available to do so. — The implications of different combinations of thermal plant on the costs and impacts of integrating renewable energy in the short to medium term. In particular, the relative impacts of different sizes and types of thermal generation, and of inflexible versus flexible plant, on efficiency of system operation and integration of wind and other renewables. — Options for managing the additional power fluctuations on the system due to intermittency — including new supply technologies, the role of load management, energy storage etc. Opportunities and challenges for re-optimisation of the electricity system in the long term to cope with intermittent generation, including research on much higher penetrations of renewable sources than the relatively modest levels considered in this report. Contents | Introduction 1.1 What is this report about? 1.2 Why is this report needed? 1.3 How is this report different? 1.4 The structure of this report 2 Understanding the impacts of intermittent generation 2.1 Introduction 2.2 Context: managing fluctuations in demand and supply 2.3 Introducing intermittent supplies - what changes? 2.4 Calculating costs 2.5 Summary 3 Evidence on the costs and impacts of intermittency 3.1 Introduction 3.2 Historical development of research on intermittency 3.3 Quantitative findings 3.4 Discussion of key issues from the quantitative evidence 3.5 Summary of key findings 3.6 Summary of all findings and data used in Ch. 3 4 Conclusions 4.1 The impacts of integrating intermittent generation 4.2 The costs of integrating intermittent generation 4.3 Factors that affect the costs of integrating intermittent generation 4.4 Confusion and controversy 4.5 Recommendations for policy 4.6 Issues for further research References Annex |: Project team, expert group and contributors Annex 2: Costs of maintaining system reliability Annex 3: Full list of included documents Annex 4: Full list of excluded documents Annex 5:Technical annex to Ch.|: search terms and databases used Annex 6: Technical annex to Ch. 2: terminology a wa . ‘ . oe oe NO wWwnen- = an RauunrwWunun bh wwwwWwn — SR<SISSERESSSSEVCGESZSESANA[-S aR Glossary’ BETTA BM BNFL BPA BWEA CCGT CEGB CHP CO: DENA DNUoS DTI EC EPSRC EST EU GRE Grid Code GW GWh JESS kW kWh ICEPT IEA LOLE LOLP British Electricity Trading and Transmission Arrangements Balancing Mechanism British Nuclear Fuels Ltd. Bonneville Power Administration British Wind Energy Association Combined Cycle Gas Turbine Central Electricity Generating Board Combined Heat and Power Carbon Dioxide Deutsche Energie-Agentur - German Energy Agency Distribution Network Use of System charges Department of Trade and Industry Commission of the European Union Engineering and Physical Sciences Research Council Energy Saving Trust European Union Great River Energy A document that defines obligatory features of a power generator that is connected to the electricity system Gigawatt - a measure of power, one thousand MW Gigawatt Hour - unit of electrical energy, one thousand MW of power provided for one hour Joint Energy Security of Supply Committee (comprised of representatives from National Grid, DTI and Ofgem) Kilowatt - a measure of power, one thousand watts Kilowatt Hour - unit of electrical energy, one thousand watts of power provided for one hour Imperial (College) Centre for Energy Policy and Technology International Energy Agency Loss of Load Expectation Loss of Load Probability MW MWh NETA NGC NGET NISM OCGT Ofgem OPEC OU PIU PV p/kWh RIIA R&D sD Thermal Plant TPA TNUoS TSO TW TWh UCD UKERC Watt (VW) WECS Megawatt - a measure of power, one thousand kW Megawatt Hour - unit of electrical energy, one thousand kW of power provided for one hour New Electricity Trading Arrangements National Grid Company National Grid Electricity Transmission (formerly NGC‘) Notification of Inadequate System Margin Open Cycle Gas Turbine Office of Gas and Electricity Markets Organisation of the Petroleum Exporting Countries Open University Performance and Innovation Unit - now the Prime Minister's Strategy Unit, UK Government Cabinet Office Photovoltaic - apparatus that transforms sunlight into electricity Pence per kWh, the common unit of pricing energy sold to consumers Royal Institute of International Affairs (Chatham House) Research and Development Standard Deviation Conventional steam-raising electricity generators - this includes coal, gas, oil and nuclear plant Technology and Policy Assessment Function (of the UKERC) Transmission Network Use of System charges Transmission System Operator - in the UK, this is National Grid Terawatt - a measure of power, one thousand GW. Terawatt Hour - unit of electrical energy, one thousand GW of power provided for one hour University College Dublin UK Energy Research Centre The standard (SI) unit to measure the rate of flow of energy Wind Energy Conversion Systems - in practice, a wind turbine List of tables and figures Table 3.1 Table 3.2 Table 3.3 Table 3.4 Table 3.5 Table 3.6 Table 3.7 Table 3.8 Table 3.9 Table 3.10 Table A2.1 Figure 2.1 Figure 2.2 Figure 3.1 Figure 3.2 Figure 3.3 Figure 3.4 Figure A7.1 Figure A7.2 Figure A7.3 Figure A7.4 Overview of the evidence base Example studies 1978 - 1989 Example studies 1990 - 1999 Example studies from 2003 & 2004 Relationship between capacity credit and cost, GB relevant capacity credits and system characteristics Summary of additional reserve requirements with intermittent generation Summary of findings relating to reserve costs Range of findings for fuel and carbon dioxide savings metrics Range of findings for energy spilling metrics Range of findings for capacity credit The sensitivity of reliability cost to capacity factor and capacity credit Seasonal variation in daily demand patterns Winter 24 hour load profile on National Grid system Range of findings related to additional reserves with increasing penetration of intermittent supplies Range of findings on the cost of additional reserve requirements Range of findings on capacity credit of intermittent generation Frequency distribution of findings for capacity credit where intermittent generation provides 10% of energy Frequency Distribution of System Margin When Conventional Generation Supplies 100% of the Energy. Loss-of-Load Probability = 2.5 % Frequency Distribution of System Margin When Conventional Generation Supplies 80% of Energy and Intermittent Generation 20%, but with no Additional Investment in Capacity to Maintain LOLP (LOLP rises to 30%) Frequency Distribution of System Margin When Conventional Generation Supplies 80% of the Energy, Intermittent Generation 20%, and Backup Capacity is Installed to Maintain Loss-of-Load probability to =~ 2.5% Available conventional capacity (including backup) corresponding to Figure A7.3: 80% of energy is supplied by thermal and 20% by intermittent generation; backup capacity 20% (19.2% capacity credit) 1.1. What is this report about? Some forms of renewable electricity generation exhibit what is often referred to as ‘intermittent’ output. The output of these kinds of electricity generators depends upon environmental conditions that may be predictable to some degree but are outside the control of plant owners or system operators. For example, the amount of electricity generated by a wind turbine fluctuates as wind speed changes and that of a photovoltaic array with the intensity of sunlight. Their output is controllable only insofar as operators can curtail or reduce the potential output of such generators. When such devices are connected to electricity networks in significant numbers this will affect the operation of the network, actions within electricity markets, and the need for and output of other forms of generation. The impacts of intermittent generation on system operation and reliability and the extent of any new costs (relative to costs and impacts imposed on the system by other generating options) are the subjects of this report. The report focuses on the UK", but draws upon experience and analysis from several countries. The focus is also largely upon issues raised by existing renewable energy targets or goals, and changes that might therefore be required in the period to around 2020. This time period was considered to be most relevant to policymakers and other industry stakeholders''. The report therefore draws on literature that is largely concerned with incremental change to existing electricity systems, rather than with the design or conceptualisation of radically different systems — such as those that might be conceived for the more distant future”. The report addresses this question: What is the evidence on the costs and impacts of intermittent generation on the UK electricity network, and how are these costs assigned? ‘This report refers to such generators as ‘intermittent’. Other terms such as ‘variable’ have been proposed, and are arguably more accurate (Grubb 1991). Buc ‘intermittent’ is common parlance, despite its limitations and the fact that alternative descriptors were first mooted more than a decade ago. All terms have limitations and ‘intermittent’, though perhaps unsatisfactory, is widely utilised. "More specifically the main focus of the report is the Great Britain (GB) electricity network, since Northern Ireland has rather different regulatory arrangements. See Ch. 3 for more details. "'See the UKERC User Needs Assessment http://www.ukerc.ac.uk/content/view/|24/105/ Some oarticioants in the stakeholder workshop on this renort nranaunded a more ‘visinnary’ annrnach and an idaalicad 1.2 Why is this report needed? A substantial fraction of the UK’s 2010 and 2015 targets for renewable energy will be met by intermittent generation, particularly wind power. The Government's aspiration is that renewables will meet 20% of electricity demand by 2020. This, and any subsequent renewable energy targets, is likely to increase the role of intermittent sources still further (DTI 2003). Managing intermittency may have important implications for the costs of meeting these targets, and/or affect security of electricity supply. It is important to note that intermittent generators are not alone in imposing system costs, nor are other technologies without security of supply implications. The report therefore focuses upon what is different about intermittency, and what changes are needed to deal with the integration of intermittent renewables. Although there have been numerous engineering and economic studies of intermittency in many countries”, the topic remains contentious with a wide range of cost estimates in the literature'*. As a result, stakeholders interviewed by UKERC prior to the initiation of this assessment suggested that intermittency is a key controversy in the energy field’. Intermittency is also a high profile topic, often linked to a wider debate over the future of wind power that has caught the media's attention'®. This report attempts to show where the balance of evidence lies and explain the reasons for the disparities in the literature. It is important to note that there are a range of issues where evidence is limited or research is ongoing, it is also impossible for this report to cover each and every aspect of integrating renewables in to electricity networks. Where possible we highlight evidence gaps and topics for further research. 1.3 Howis this report different? The object of this report is not to undertake new research on intermittency. Rather, it is to provide a thorough review of the current state of knowledge on the subject, guided by experts and in consultation with a range of stakeholders. It also aims to explain its findings in a way that is accessible to non-technical readers and is useful to policymakers. A key goal is to explain controversies, where they arise. To do this the UKERC undertook a systematic search for every report and paper related to the costs and impacts of intermittent generation. This highly specified search revealed over two hundred reports and papers on the subject, each of which was categorised and assessed in terms of the issues covered and the methodology of the analysis. Experts from all sides of the debate and a wide range of stakeholders were invited to comment and contribute through an expert group and stakeholder workshop. Each stage of the process has been documented so that readers and reviewers can identify the origins of our findings and how the literature we consider and discuss was revealed. We describe this in a review protocol, published on UKERC's website. Relevant materials were also posted on the website as work progressed, including the project scoping note, discussion paper and workshop proceedings”. The complexity of the subject matter and confusion surrounding the debate were highlighted in our interactions with stakeholders. Our research also revealed a relatively limited attention to accessible exposition of principles in the literature. This report therefore provides an introduction to some key principles of electricity network operation and explains the factors affected when intermittent generation is added to it. Finally, the review team undertook its own analysis using statistical first principles to inform the exposition and assist in assessment of the findings revealed in the literature’. PUK research dates back to the late 1970s. Recent work was undertaken for the UK Energy Review (Milborrow 2001), Energy White Paper (Ilex and Strbac 2002) and Carbon Trus/DTI Network Impacts Study (Mott MacDonald 2003). Similar research has been conducted in most countries with renewables programmes. “See Ch 3 for the range of estimates. 'SUKERC User Needs Assessment htto://www.ukerc.ac.uk/content/view/|24/105/ The project team was drawn from the Supergen Future Network Technologies Consortium. The Expert Group was chosen to provide economic, policy and engineering expertise and a diversity of perspectives. It provided advice and scrutiny at a series of meetings throughout the project. Peer review was provided in January 2006 by international experts”. Databases, bibliographies, catalogues, and other sources, together with key words, were agreed with the expert group, refined in collaboration with stakeholders and published in the Scoping Note and Protocol. The approach aims to provide a comprehensive, transparent and replicable assessment of the balance of evidence on the intermittency debate. As a result this assessment is able to draw firm conclusions about what is known, what remains uncertain and where more research is needed. We hope that this serves to overcome some of the controversy and facilitates a better informed debate. yn Expert | Pee 1.4 The structure of this report Ch. 2 provides an introduction to the principles of network operation relevant to integration of intermittent generators, what changes when intermittent sources are added to the network, and the techniques used to assess their impacts. It also provides an overview of key controversies related to each impact. These issues are explored further in Ch. 3 and Ch. 4. Ch. 3 provides analysis of the evidence on each impact. In each case it provides the reasons that findings differ and discusses the implications of the quantitative evidence. Ch. 4 draws out the principal findings, conclusions, implications for policy and discusses areas where further work is needed. ey comp Understanding the impacts of intermittent generation 2.1 Introduction This chapter provides an introduction to the operation of technically mature electricity supply systems, such as that of the UK. It explains how a combination of market mechanisms and actions by the body responsible for the technical operation of the transmission system — the transmission system operator (TSO) — ensure that supply and demand are kept in balance. This includes the provisions that are made in case conventional generators fail or demand is higher than expected. All forms of generation have the potential to increase or decrease system costs. The chapter therefore explores the additional issues introduced when intermittent supplies are introduced and their implications for system composition, operation and cost. The chapter deals with a range of concepts for understanding the impacts of intermittency, and the generic tools used to assess and maintain standards for reliable and secure operation of electricity networks. Some of these techniques and tools are themselves the subject of ongoing research and development, as we explain further in Section 2.3 and elsewhere. There is also ongoing research on the characteristics of intermittent resources and the best techniques for managing them. This chapter (and-Ch. 3) provides in some respects a ‘snapshot’ of current understanding. What follows is intended to provide an overview accessible to a non-technical reader. It attempts to ensure consistency with the processes and practices in place within the current UK market and regulatory arrangements”. Terminology can give rise to misunderstanding as terms are used differently in different countries, or have changed over time. We discuss potential for misunderstanding due to terminology below and a review of the technical terms we use in this chapter is provided in Box 2.1. Full details are in Annex 6. *For the most bart the chapter uses the terminology of the British Electricitv Trading and Transmission Arrangements example in moving from ‘engineering to. regulatory or comme result, confusion can arise. Annex 6 ‘provides : a comprehensi review here a few key terms, unlikely to be familiar to the lay | reader, but essential to ‘aad chapter and those that follow. Where terms differ between sources the specific ic defi inition | report is provided: ‘These follow the GB Grid Code. Term _ Definition Balancing Set of arrangements in place after gate c closure (see below) i in which the ‘system oper mechanism can take bids and offers to balance the system. The prices of bids and offers are de mined by market participants and, once accepted, are fi irm contracts, paid at the bid price. hese |. bilateral contracts are between market participants and the system operator. ees Balancing services Services purchased from balancing service providers by the system ‘Operator. Includes Balancing I Mechanism bids - offers, other energy trades, Response, Reserve, and other ke system services. Re A : é BETTA British Electricity Trading and Transmission ‘Arrangements The area alee qe Binich generators, ‘suppliers and the system operator operate. These include the GB Grid Code, Balancing & Settlement Code and Connection & Use of ‘System Code which, contains detailed definitions of contracts and rules relevant to this section. Capacity credit plant with no increase in A measure of the amount of load that can be served on an ‘electricity system by i intermittent, e loss-of- load probability (LoLp), which is often, expressed i ine terms of conventional thermal capacity that an intermittent generator can replace. - aes Capacity factor Energy produced bya generator as a percentage of that which would be achieved i 20% = 40%.* Gate closure ‘| for contracts to balan demand and supply. mee see “balancing mechanism’. The point in time (one hour before real time und ie Ramping rates _ A measure of how quickly any Pere on. the sysremn< can increase or. decrease its ouipis ae “ A normally measured in MW/h. : ieshe : Z pee Response and reserve services Reserve and response services are purchased iy he system operator in order to ensure ~ there is sufficient capability in the short-term to undertake system balancing actions and frequency control. Response (frequency response) may be utilised in seconds through ; automatic controls on generators or loads. Steam generators may be held below maximum output to facilitate this, Reserve is a capability to change output to meet system operator requests within a few minutes. Utilisation of this capability may ‘be subject to payment in the Balancing Mechanism or through other balancing service agreements. There are various categories of reserve depending on speed of delivery and the nature of its Provision: Fast Reserve can be provided by demand reduction, pump storage or part | loaded steam” Se plant connected to the system. Standing Reserve is ready for action within twenty minutes. As well as demand reductions it might consist of fast: starting gas turbines, c or u c tration. Residual Reserve - This is the capability provided i inthe . Balancing Mechanism (ie. reserves that at be erect in ESPs to market Prices s rather than contracted by the TSO). ee, Contingency Reserve - This is the capacity that pou be established i in the 4 hour ahead period by the market. It is not usually purchased by the TSO buti is monitored to ensure adequate short-term reserves will be available. System margin The difference between installed capacity, including imports and exports, and peak demand. — Operating margin is the difference between available generation and actual demand, Suctam anerator far [Tha eAmisand Ar hAdWretaAAciila (apiehalsachntal AnaeaticW AP tha alaeticiedieranemiecian 2.2 Context: managing fluctuations in demand and supply 2.2.1 Introduction Electricity demand changes continuously. It fluctuates from second to second and also goes through very large swings over a few hours. Most consumers are not under the control of system Operators nor are they direct participants in wholesale electricity markets”',so the system operator and market mechanisms must ensure increased electricity generation as demand increases, and reductions as demand falls. To prevent serious problems this adjustment must be continuous, and almost instantaneous. Figure 2.1 illustrates the extent of diurnal variations in electricity demand, and how these vary by time of year, reflecting seasonal effects. Some forms of generation can vary their output rapidly, others only over a longer time period. Some are largely inflexible and must operate at a fixed and constant level. Figure 2.2 in section 2.2.3 shows an example 24 hour ‘load profile’ and the operation of different types of plant to meet demand. However no plant is able to operate 100% of the time; all types of generator require periodic maintenance, and every power station will suffer occasional unplanned outages due to a breakdown or fault. As a result, power systems are engineered to cope with both demand fluctuations and periods when several power stations are unavailable due to planned maintenance or unexpected breakdowns. We explore the processes through which demand fluctuations and supply side failures are managed in the following sections. Figure 2.1. Seasonal variation in daily demand patterns” 55000 *'Some very large consumers (such as steel works or chemical plants) can reduce demands in resnanse ta hich nrices ar enter 2.2.2. Basic principles - meeting demand fluctuations and ensuring reliability Section 2.2.3 provides an overview of the way that the UK's electricity market helps keep electricity demand and supply in balance, as well as current provisions for reserves and contingencies to deal with unpredictable events. Market mechanisms have replaced the centrally planned electricity system that used to exist in the UK. However, the way the market relates to the technical operation of the electricity system is complex. By way of background, we first revisit the basic principles of electricity system operation in an historical context. In the very early days of electricity, single generating stations provided individual local networks with electricity. These generators had to be inherently flexible in output, and/or variations in demand had to be restricted. They needed 100% ‘back up’ to ensure reliability in case of a fault. As interconnection between local networks expanded — first on a municipal scale, then nationally — two changes occurred. First, risks of breakdown could be shared across many plants, so the amount of ‘back up’ could be reduced without compromising reliability. Second, demands were aggregated which tended to smooth fluctuations and also meant that a range of types of generation could be used — some more flexible than others. In very simple terms, the following principles of operation apply to all large, advanced networks: — Arrange of plants are used to meet different portions of the daily demand curves seen in Figures 2.1 and 2.2 - from very flexible plant designed to meet rapid swings in demand to inflexible (but cheap to operate) plant that runs all the time. A process for ‘dispatching’ plant to meet demand is needed, historically this was under the direct control of the system operator/owner, and used least cost criteria. System balancing reserves are needed to deal with unexpected short term fluctuations (minutes to hours) caused by either demand changes or faults at power stations or power lines. These reserves are sized on a statistical basis according to the range of unpredicted variation in demand, reliability of conventional generators and the scale of potential faults. The aim is to meet specific criteria for operational reliability — that is to ensure that the risk of demand being unmet is small. — In addition to the short term reserves made available each day a larger ‘system margin” of maximum possible supply over peak demand is provided for when planning the development of the system. The size of this margin can be determined using statistical principles to do with the number and reliability of generators and the variability of demands. Previous UK practice was to ensure installed capacity should be approximately 20% larger than expected peak demand. Current practice is to monitor and report on this margin. Again, such criteria are aimed at ensuring a specific measure of reliability is sustained, and the risk of demand being unmet is small (e.g. the LOLP of the pre-privatised electricity system in Great Britain was planned not to exceed 9% - nine winters per century). Any new generating plant can contribute to system margin (to varying extent) and may increase or decrease reserve requirements. In very broad terms, very large, less reliable, and unpredictable or inflexible generators tend to increase system costs™. Smaller, reliable, predictable and flexible generators tend to reduce system costs. Nowadays the central planner is gone. Many of its functions are taken care of by markets, and key technical duties now rest with the transmission system operator (TSO). Nevertheless the same basic actions must be undertaken and requirements met. We explain how this is achieved in the section that follows, and then consider what changes when intermittent generators are added to the system. *System margin is the current UK Grid Code term.The concept has been referred to historically as variously ‘capacity margin’ “system reserves’ and ‘plant margin’. *A recent example of a new non-renewable generator affecting reserve reauirements is that of the oressurised water reactor 2.2.3. Short term balancing and long term capacity provision Short term balancing Short term balancing is achieved in part through actions of the system operator, but also through decisions taken in markets. Balancing through the market Currently, in the UK and many other countries, most of the variation illustrated in section 2.2.1 is handled by markets. Demand variation is reflected in market prices and/or supply contracts that ensure more generation when demand is high and less when it is low. Markets operate under different rules in different countries. The UK market arrangements ‘BETTA’ (see Box 2.1) are reflected in the following forms of contract and market activity: — Firstly, generators and wholesalers/suppliers enter into bilateral contracts — more than 90% of UK electricity is traded in this way. Such contracts can be long term — months or even years ahead of real time. Contracts incorporate time of day variations. — Second, small amounts of electricity trade through a number of spot markets (known as ‘power exchanges’) that allow market players to buy and sell electricity for rolling half hourly time periods. These markets operate from a couple of days until one hour ahead of real time. At the one hour point in time bilateral trading between generators and consumers is suspended and the energy volume of bilateral trades between generators and suppliers is notified to the settlement system. This is known as ‘gate closure’. — After gate closure, a balancing mechanism operates in the period from one hour ahead of real time and allows anticipated shortfalls or excesses — perhaps as the result of a known fault at a power station — to be accommodated through direct trades between the system operator and large consumers or generators of electricity’. — Finally (also after gate closure), the system operator can instruct plant with which it has contracts for balancing services to increase or decrease output and frequency response will occur automatically due to the action of a range of automatic controls. These reserve and response provisions for system are quite complex, and highly important to the intermittency debate. They are discussed below. Balancing by the system operator The market activities described reflect anticipated demand and supply. In addition, over short timescales, relatively small (but crucial) adjustments are made by the system operator. These allow residual market errors and events occurring post gate-closure, such as demand prediction errors or sudden failures at power stations, to be managed. Adjustments are made through automatic controls on power stations and by the system operator calling upon fast responding reserve plants. It does this through the balancing mechanism and directly with operators with which it has entered into reserve service contracts. These actions are described below, using terminology defined in terms of the range of services that the system operator contracts for, and the grid code issued by the electricity market operator”. Output from plant contracted to provide response may be delivered within seconds and its utilisation is controlled by automatic control systems sensitive to system frequency. Contracted fast reserve can be brought into operation within seconds to minutes. Contracted standing reserve can be brought into operation within 20 minutes and must be able to sustain its output for some hours. Some reserve capacity (called residual reserve) may be provided by part-loaded generation that participates in the market but is not contracted by the system operator. The sum of contracted and residual reserves are called operating reserves. Reserves in excess of operating reserves that appear available prior to gate-closure are referred to as contingency reserves. The system operator will monitor such reserves as real-time is approached to ensure the required operating and contingency reserves are maintained. In some cases, the reserve provided by the market, together with that contracted in advance by the system operator may be insufficient and so additional reserve must be established by ’warming’ unsynchronised generation, which requires several hours to achieve (a ‘warming’ payment may be made to generators by the TSO). It is important to note that the strict definition of reserves described here may differ from international or historic norms. In particular, these definitions encompass all system balancing reserves, but do not include what have sometimes been referred to as ‘capacity’ or ‘system’ reserves which relate to system margin (see below). System balancing reserves contracted by the British TSO (National Grid) currently stand at around 2.5 GW”. Reserves are sized in relation to three factors: — The largest single credible generation in-feed loss on the system” — The expected availability of all conventional plant on the system - A given amount of demand prediction errors The effect of the latter two factors is determined statistically. We discuss the way this is done in Section 2.3. In the sections that follow we explain how intermittent supplies impact on the activity of electricity markets and the need for reserve and response services. Figure 2.2. Winter 24 hour load profile on National Grid system” 60 $s: s88 88 ao oOo rt Ho oF OD g g|f om GNuclear @Gas Olmports CLarge Coal O Coal Oi GPumped Storage Other *Winter Outlook Report, 2005/6 Published by National Grid Plc and available at hetp://www.ofgem.gov.uk/temp/ofgem/cache/cmsattach/|2493_214_05.pdflwtfrom=/ofgem/index.jsp “Reserves are sized to cover the sudden loss of the single largest generating unit (a criteria known as n-I, where n is the number of plants and | che single largest plant, currently either Sizewell B nuclear power station at |!260MW, or one bipole of the interconnector with France at [OOOMW. or generators subiect to instantaneous trinning for system ournoses of un to f response to immediate events. Timeframe (period ahead of real time) Years _ Actions by ae Be Planning and construction of new plant Long term bilateral contracts Purchase of transmission entry capacity “Actions of system operator Seven year statement. Transmission system pricing New transmission capacity — Months Generation maintenance schedules “Transmission maintenance i : schedules Return of mothballed plant Winter outlook report * Monitoring and ref orting of Purchase of short-t term transmission. Sry capacity ‘ ee “Bilateral contracts j Weeks ‘Bilateral contracts Days” Short term bilateral contracts Commitment of inflexible iveneratines: units : Notification of Inadequate : 3 System Margin (NISM) ‘Hours Participation in balancing mechanism Establishment of required Commitment of less flexible, generating response and reserve capacity units a 4 Variation in output from flexible/load Operation ot balancing following plant to maintain contracted . echanism ti Ay positions : : Registration ofe energy volumes at gate closure et contracts Notification of intended physical - : i positions to System Operator i aa eae Minutes Variation in output from flexiblelload Utilisation of fast reserve ‘following plant to maintain contracted positions. Adjustments to reflect acceoted balancine mechanism bids & Operation of balancing” mechanism Ensuring reliability through capacity provision — system margin In addition to the short term operational requirements of the system, reliable supply of electricity also requires that electricity markets deliver enough capacity to meet expected peak demands. To be confident of reliable supplies a system margin by which installed capacity exceeds peak demand is desirable. This is because there are bound to be some plants that have to be taken out of service for maintenance, some that break down, or times when peak demand is higher than anticipated. The relationship between system margin and reliability can be quantified. It is a function of potential errors in demand prediction, outages in generation and is normally estimated using concepts such as the Loss of Load Probability (LOLP, see box 2.3). Historically these calculations led directly to the planning of generation capacity by system owners and operators. This is no longer the case in the UK where the system margin emerges from decentralised market-based investment decisions. At present, in the UK, the system operator monitors, but does not contract for, system margin. Since it takes time to build new plant or repair plant that has been taken out of service (‘mothballed’), system margin is monitored from a time horizon of several months to years before real time. The British system operator compares market notified margin with an ‘indicative’ level of desired system margin, and publishes market participant estimates of expected system margin at winter peak periods”. These statements highlight periods when system margin is expected to be smaller than desired’. They provide the market with detailed information about the level of system margin, and are used by market participants to determine expected future prices. This has proved effective as a mechanism to incentivise generation plant owners to bring mothballed plant back into operation™. It is important to note that system margin is not the same as the reserve described above. Dedicated reserves are purchased in order to react to unexpected events quickly, whereas system margin provides a more general contingency. System margin is much larger than dedicated reserve: in the UK, around 2.5GW” of dedicated reserve is kept available, whilst National Grid's indicative level of adequate system margin is around 20% above peak demand or 12 - 14GW*™. Under current arrangements system margin is estimated net of the response and reserve services contracted by National Grid. Close to real time the amount of system margin over peaks will normally become smaller, as breakdowns, maintenance and decisions to remove generation for commercial reasons become manifest. Expected and actual system margins are monitored by the system operator which can take a range of actions if the margin is smaller than it believes it should be to ensure reliability. The primary mechanism is Notification of Inadequate System Margin (NISM)*. In the sections that follow we explain how intermittent generation can contribute to reliability, and how we assess the extent of this contribution. We also discuss the controversy related to the costs of maintaining reliable supplies when intermittent generation is added to the system. See Seven Year Statement and Winter Outlook Report, 2005/6 Published by National Grid Plc *'See Ibid. Note that system margin is actually monitored by the system operator (National Grid) on behalf of a team of experts drawn from the TSO, the DTI and Ofgem (che JESS committee). *The TSO provides information on a range of factors. For example, it shows scenarios for demand in both ‘normal’ and exceptionally cold winters. The information that the system operator provides also takes into consideration the likely availability of plant. For 2006/6 National Grid oravide scenarios that also show the available margin assuming 91% é supplied. ‘Some interruptions of sh arise from equi “transmission and distribution ne when they < are exceeded by the available capacity, but have to be shed of course wh ¢ the area when the available margin is negative provides a measure ‘of what i is called the loss of load probability. : The ‘spread’ of the distribution shown i is measured by its standard deviation*, denoted here by. G, and there i is a simple statistical relationship between this quanti and the standard deviatior variations in demand (relative to expected demand) and plant availability (failure rate), denoted by ao and o, respectively (the square of such quantities is called the variance): 0,7 = a4 + a fees The greater the supply uncertainties the greater the system margin needs to be (whether dete mined by market forces or otherwise) if the reliability of supply is to be maintained. Intermittent plants chai ‘the spread and the shape of this distribution, as we explain in Box 2.7. Akos a ‘Distribution en OF Available Plant ‘ Margin 03 02 10 5 0 5 10 15 20 25 30 | : % ; Available Plant Margin as % Peak Demand It is important to note the LOLP provides for a simplified ¢ comparison of the reliability of prospective Bi] generation systems as it does not provide any indication of the : frequency, duration. and the potential shortages. These factors have been identified as an important area of further res to faire pectic system Sea ere in the UK (Ernst. Seung 2005).: i sdeandard’ deviation is a measure that tells you how tightly nes a ‘set oy values are around the mean “value of a set the standard deviation is small che ‘bell curve’ depicted above is steep and narrow. When ic is large the curve broadens out. 2.3 Introducing intermittent supplies — what changes? 2.3.1 Introduction As we have seen, regardless of any intermittent generation that might be installed on an electricity network, provision must be in place to respond to changes in demand and failures in supply. Large demand fluctuations are normal, and no form of generation is 100% reliable. These are handled through market mechanisms and actions taken by the system operator. Quantifying the implications of adding intermittent generation to a network is therefore primarily a matter of understanding the extent to which a number of factors change when such generation is installed. 2.3.2 What is different about intermittent plant? Intermittent renewable generation has a range of characteristics that distinguish it from conventional generation plant. Intermittent generators can provide energy, have zero fuel costs and can reduce emissions. It would usually make sense to operate such plant whenever it is available. The energy supplied by an intermittent generator is a function of the resource available to it, and the amount of generation capacity installed. Taking wind energy as an example, in aggregate, the average capacity factor of all British wind farms is in the region of 27% - 30%*. Capacity factor is a measure of the average power output relative to the installed capacity. Generally speaking the capacity factor that can be achieved by intermittent generators is low relative to that of conventional generators. This means that a larger amount of intermittent capacity is required to replace the energy from a given capacity of conventional stations. This has implications for important concepts related to keeping electricity supply reliable (such as system margin) and for the maximum amount of ‘back up’ that might be required as a result of adding intermittent generators to the system. These are discussed later in this chapter. Intermittent renewable plants also show a wide variation of output, indeed for much of the time the output of a wind farm or other installation might be less than half of its maximum potential output. The nature of the outputs of intermittent generators varies markedly, depending on the nature of the It might be largely predictable (solar power in sunny regions), entirely predictable (tidal power) or much more stochastic (wind power in some regions, solar in UK). But all forms of intermittent renewable energy contrast with a conventional generator which (if required) would be expected to operate close to its maximum output for most of the time, with a relatively narrow range of output variation — even allowing for unplanned outages. Whilst all plant is intermittent, insofar as it will suffer occasional outages, intermittent renewables fluctuate to a much greater degree. Depending upon technology, location and timing of demand peaks, their output may or may not be available during peak demand periods. In many cases, the contribution to reliability is lower than for conventional stations, because there is more uncertainty surrounding the contribution of intermittent stations to meeting peak demands than there is for conventional generators contributing a similar amount of energy. These factors can be quantified and give rise to changes to provisions for system balancing and reliability, which we explore and explain later in this chapter. 2.3.3 Limitations There are a range of important issues that relate to the integration of renewables, and may be affected by intermittency, which are beyond the scope of this report or only dealt with relatively briefly. Examples include: the impacts of renewables on transmission infrastructure, particularly if generation is concentrated in limited geographical areas; the costs and impacts of supply interruptions; the role of demand side management and bulk storage systems in accommodating intermittent energy; the impact of generation system flexibility on the ability of electricity systems to absorb wind energy. Moreover, the sections that follow and the evidence presented in Ch. 3 tend to focus on a particular set of issues, and a particular approach to ensuring reliability (see Boxes 2.3 and 2.7). These approaches have limitations, some of which we discuss (see section 2.3.5), and are still under development through ongoing research. It is also important to note that there are areas where empirical evidence can improve understanding, examples include: a more detailed understanding of a range of intermittent resources; impact of extreme weather conditions; the effects of geographical clustering; understanding and quantifying the impacts of different nature of off- Box 2.4 Popular misconceptions Two related assertions that receive regular airings in the mainstream media are paraphrased below: ‘Wind turbines only operate 3 30% of the | time, therefore we must provide 70% backup’ ‘Wind turbines need back up so they don’t save any CO:* Both these assertions are incorrect. In both, the use of the term ‘back up’ may in itself give rise to misunderstanding. Irrespective of terminological issues, the assertions are in error for ‘the followi reasons: { ; t — The former assertion confuses the Sapaeny factor (see above and Annex 6) that might be achieved | by a ane typical UK wind farm (which would indeed be around 30% in a location with good wind coni with the amount of time it is operational. In fact, most wind turbines will be operational fe ) Rae 80% of the time — but usually operate at less than their rated capacity. This i ‘is because the rat lc pacity” i of a wind turbine is its maximum output, v which i is typically associated with wind speeds in excess of © 11-15 m/s (40-54 km/h). Yet most wind turbines operate ina range of wind speeds from around 4 mis to around 25 m/s. ‘ — The capacity factor of renewable energy abes not tell us anything about ‘back up’ requirements. The. capacity factor simply provides an indication of the amount of energy, on average, a given capacity of | renewable plant would be expected to ‘provide. The actions needed to manage intermittency are Rie derived statistically, as this chapter explains. ‘ ij — However, capacity factor does indicate the size of the comparator plant against whi h ‘intermittent generators should be assessed when determining what is required to maintain rel vil ty. A 1000oMW. wind farm with a 30% load factor delivers the same energy as a 350MW modern gas power station, “allowing for the 15% outage rate of such generators. Hence, even if the intermittent station cannot ‘ : contribute anything to reliability i its ‘back up’ in this example ‘won "tC exceed 35% of i inst e This is why claims that renewable generators need ‘100% (or even n 60% or Beate) ‘back iu per MW ij installed are muddled and incorrect. ‘ i ; : — The ‘latter assertion conflates energy and power. “Intermittent sources are Unlikely. to be abl provide the same level of reliable power output d iring ¢ demand | peaks as a conventio! g nerator. This al wey, give rise toa need for additional capacity to maintain reliability (see Ch. ane full details), ’ ‘than i its contribution to reliability In fact, even ‘if the “contribution of an intermittent source 2 at peak periods is ‘expected to be zero (as would be the case for PV power in the UK for example), its contribution to CO: savings are still a direct function of its energy output. Actual CO, savings are dependant on what fossil fuel plant is displaced, reduced by efficiency losses in thermal plant affected by intermittency and additional use of reserve and response. As we show i in Ch. 3, these losses are a small proportion of the energy provided. Links to other grids can mean that CO: savings are ‘exported’ so might not be realised in the country of origin. But the CO: savings are, within a few percentage points, directly linked to the energy that renewable stations generate. 2.3.4 Principal impacts of intermittent generation on electricity networks The principal impacts that we discuss in this section fall into two broad categories”: 1. System balancing impacts The primary benefit (indeed purpose) of adding intermittent generation to an electricity system is to save fuel and hence reduce emissions as fossil fuel stations are used less. Direct economic benefits might be made available to renewable energy developers through polices such as (for example) the Renewables Obligation. These savings will be reduced to the extent that intermittent plant gives rise to an increase in: — Response and reserve requirements contracted for by the system operator, to manage unpredicted short term fluctuations and referred to here as response and reserve impacts. — Effects on the utilisation of other plants in the electricity market which are termed here system efficiency impacts. Examples include losses due to increased variation in the output of thermal plant and wasted energy if intermittent output exceeds the ability of the system to use it. 2. Effects on capacity requirements to ensure reliability If intermittent generators can make a contribution to reliability — that is if there is some probability of them generating during peak periods — they may be able to displace (or avoid future investment in) thermal plant without reducing system reliability. This is a benefit of intermittent generation over and above their role as a ‘fuel saver’. However, the contribution of intermittent generators to reliability is often lower than a conventional generator that can deliver the same amount of energy. Hence there are two, counterpoised, impacts, the allocation of which can give rise to confusion and controversy, as we explain later in this chapter: — Capital cost savings from any conventional plant that can be replaced or retired without compromising reliability - Capital costs of conventional plant retained or constructed to maintain reliability at peak demand. We now consider how each of these categories of impact is measured and assessed. For each category we consider: — The range of impacts — The information needed to quantify each impact — The techniques used to assess the scale of each impact — The implications of each impact It is important to note that any new generator has the potential to increase or decrease reserve requirements or reliability. It is therefore important to assess impacts, and in particular costs through a comparison with a given alternative form of generation that can provide the same amount of energy. ”Porentiallv important imoacts that we do not discuss in ereat derail here include: 2.3.5 System balancing 1: Response and reserve services Range of impacts Additional short run fluctuations in output can increase the utilisation of automatic controls on the output of conventional power stations. It is also necessary to have more part loaded plant running that can rapidly ramp up or ramp down i.e. increase or reduce its output as intermittent stations pick up or drop off. Fluctuations over minutes to several hours can require increased fast and standing reserve. It is important to note that predictable variations will have implications for the utilisation of non-reserve plant, and are discussed in the section on system efficiency impacts, below. This section is restricted to unpredicted variations. Information needed Three interrelated factors determine the amount of extra reserve required when intermittent generation is added to the system. |. The extent of rapid and unpredicted variations in intermittent plant output, which has two aspects: Firstly, how rapidly the outputs of different penetrations of different types of intermittent plant will fluctuate. This is sometimes called the ramping rate. Secondly, the possible scale of total, system-wide, changes in a given period. The system may cover a large area. Hence, this requires a representation of the aggregated behaviour of individual intermittent plants, based on weather data, size of units, inertia, the scope for ‘smoothing’ of outputs — for example by geographical dispersion — and a range of other 2. How accurately fluctuations over the minutes-to- hours timescale can be forecast. This is important because the more accurate the forecasting the greater the opportunity to use (lower cost) planned changes as opposed to holding reserve plant in readiness — in particular reserves comprised of extra part loaded plant, which can be costly and less efficient. In market terms, the effects of predicted fluctuations can be contractually committed prior to gate closure, which should permit the market to reveal the most cost effective means to manage these variations. Again, it is the prediction accuracy of total aggregated intermittent generation that is relevant, forecasting for a large amount of distributed resources reduces forecast errors. 3. How existing variations of demand or load compare with that of intermittent output and the reserve capabilities that already exist on the system. These existing reserve capabilities are a function of the variability of demand, the reliability of existing plant, the number of plants on the system, and the size of the largest single unit. Box 2.5 Reserve and response services for short term system balancing | : The amount of reserve : needed to handle ‘unpredicted short term vari tions - ell hel er due to dem: d analytic techniques viskerted here ae approximate results but im ns. are need deal. with more complex situations, for example where correlations between. variables Xist. We Present t the : analytical approach in order to provide an Jeraiatet of principles that, cor into 2 ply. : Historically, reserves have been sized t to cover approx. 3 Sona deviations of the potential uncertain — Mae fluctuations that arise from this combined demand prediction error -and generation plant failure, plus : provision for the sudden loss of the largest single unit (n-I criteria, or disturbance reserve). The +3 criteria ensure 99% of unpredicted demand or supply fluctuations are covered by reserves: Reserves = #3) ee + io (Plus disturbance reserve) where Tae, represent the standard deviati tuatior must be combined ‘statistically with the | ‘variance of ‘demand and conve: Suggest that reserve impacts from, intermittence wil be ‘relatively mode SD of conventional generation and demand a4 ie a2) i is s around 340 M : Therefore the SD with wind would be (340 - + 140°) or 368 MW - a minor addition Response and fast aR reserve requirements would be +3368 or | 143 MW - compared | to | i : cases total GB reserves would also require | 1.1 GW ‘sizing’ reserve — hence i in this example 10 GW wit de accounts for around 130 MW reserve needs out of approx 2.2. GW total reserves needs. A related. example, in this case from Denmark, is provided Brephicaly, below. z Wind fluctuation and time horizons (from Milborrow, 2001) Number of changes, % ~ j so ° Hours ahead Fe * . 4 10 °° ¥ e z “ee 2 = ee e 6. of 1 /_o* “er . ¢ *o"s 0 . wee 03 ow ee, ee * * . at oo zs © © © se How do we work it out? The relationship of interest to analysts is that between intermittent fluctuation and load variability. Statistical principles (and simulation models based upon them) can be used to assess the total unpredicted fluctuation that the system operator might have to manage over different timeframes. System operators are concerned with the total amount by which the system might be out of balance. As a result, reserve requirements are a function of both the unpredicted load variation and unpredicted intermittency. Fluctuations in load and intermittent output might amplify each other, be completely unrelated to each other, or they may cancel each other out”. Hourly variations in wind farm output are also a function of geographical dispersion. Even extreme fluctuations fall from +30% of installed capacity when the area is in the order of 40,000 km? (about the size of Denmark) to about +20% for an area of 160,000 km? (e.g. Germany or the state of lowa) and then to about +10% in larger areas covering several countries e.g. the Nordic states (Holttinen 2005). Normal fluctuations are much more modest. For this reason we need to know the degree to which variations might correlate. If demand increases tend to occur at the same time as decreases in intermittent output then the amount of reserves needed tend to increase. The obverse might also be the case, or loads and intermittent output might show no correlation. In all cases analysis requires a statistical treatment of both demand and intermittent generation, since we are dealing with probabilities rather than determinate functions. At its simplest this might take the form of a statistical ‘rule’ — such as a sum of squares rule (see Box 2.5). Statistical algorithms and computer models can also capture more complex inter- relationships and correlations. Implications Reserve requirements tend to represent relatively small proportions of the intermittent generation capacity installed; the evidence from many studies bears this out — see Ch. 3. This is because the short run fluctuations and prediction errors associated with wind capacity are comparable to other variations in the supply-demand balance and so little increase in reserve provision is called for. The reasons for this are explained in more detail in Box 2.5. A statistical derivation of system balancing requirements is also provided in the working paper that accompanies this report”. Another reason reserve additions tend to be modest is that reserves are determined by two aspects: unpredicted fluctuations described above and a ‘dimensioning’ factor that allows for the failure of the largest single generating unit. This can be a major determinant of reserve margins. Even large wind farms are much smaller than large conventional stations. Hence, there may already be more than sufficient reserve capacity on the system to deal with intermittency — particularly if the amount of intermittent generation is a small proportion of total supply (this varies according to system characteristics but might be defined as below 10% of energy — see Ch. 3). Box 2.6 Confusion arising from use of terminology _ Reserves, back up, stand by have all been ‘used to denote conventional plants t held i in readiness to respond to fluctuations from intermittent stations. Each term pies! rise to some problems. There are two important areas of misunderstanding. — Actions to manage ener term fluctuations and maintain relia bility is; electrical networks should be. assessed on the basis of plants interconnected and operated asa syst . Dedicated ‘back up’ is not required. Rather, intermittent plants may increase the amount of reserves and response needed for balancing the system, may impact on the effi iciency of other plants, and may increase the amount of capacity on the system required to maintain reliability. — The additional actions needed to manage fluctuations from i intermittent plants are also affected by the nature of fluctuations resulting from demand and conventional stations on the system. This i is because the fluctuations from intermittent plants can be ‘expected to diversify with these other fluctuations to some extent, depending on their relative magnitudes as well as correlations. Failure to assess these requirements in a systemic fashion would only be consistent if it were applied ¢ to all generation, since all experience unplanned outages. In this case the benefits of interconnected networks, which share reserve and reliability across all pian, would be lost, The second problem is associated ‘with the use ef the term ‘reserves’. The term is tised for quite different types of reson on different timescales. Two pigad Gatecoties of usage can be found i in the literature: |. Reserves has a strict and narrow sense, restricted to the requirem nts ice ‘fast responding r reserves for ; "short term system balancing ‘that are contracted for by the system perator. These a are the only ane oi As and for which the system reserves for which the system operator has a responsibility for establis operator may directly purchase i in the UK (see Box 2.1). 2. A broader defi nition ace icone: the addie nal capacity that r may ibe required to ensure reliability. when viewed from a long term, or planning, time horizon. System | margin i is the current terminology used to refer to this capacity, and in Britain there is no. mechanism for direct procurement c of system “margin. Yet historically, and in other regions, capacity over and above peak demand has also been : referred to as capacity reserves (see Annex 6). This gives rise to confusion, and may mean that comparisons are drawn between studies that are using the term differently. For example: ; — Some studies of the ‘cost of intermittency’ in fact only quantify the cost of additional system balancing = the capacity to maintain reliability may be neglected, or not directly addressed. ais may, give rise toa. ‘reserve cost’ estimate that understates the full cost of i intermittency. - " , — However, where the term ‘reserves’ is used to refer. to both capacity provision to. m intain ‘reliability i ~~ and short term reserves, this too can create confusion — since it leads to cost estimates ‘considerably larger than those directly attributable to the only. reserve services actually purchased by | the system operator. 2.3.6 System balancing 2: Other system efficiency impacts Range of impacts As mentioned previously, the principal impact of intermittent generation on the operation of other plant on the system is to replace the output of fossil fuel stations and hence secure fuel and emissions savings”. However fuel saving may be partially offset by a range of efficiency impacts: — More frequent changes in the output of load- following plant and/or greater use of flexible plant to manage predicted variations. This may decrease the efficiency of thermal plant and cause more fuel to be burnt. Frequent start up and shut down of certain types of plant can use a lot of fuel to ‘warm’ plant, without generating any electricity. The way such changes are provided for is also affected by the accuracy with which fluctuations can be forecast. In general terms better forecasting results in fewer losses, since the most efficient changes can be planned. However improved forecasting does not eliminate these costs, since the need to manage predicted fluctuations will still lead to the effects described above. — If maximum output of intermittent plants exceeds the ability of the system to absorb its energy (normally determined by the minimum output, for either technical or economic reasons, of conventional plant at periods of low demand) it may be necessary, particularly at large penetrations of intermittent generation, to curtail output or ‘spill’ energy. — Depending on the location and size of intermittent plant it may increase or decrease transmission investment and operating costs. Information needed Estimating these impacts requires quantification of four factors: |. Average energy provision (e.g. per year) by intermittent plant. This is the maximum prospective fuel saving, neglecting all losses, efficiency impacts and curtailment. 2. A time of day representation of the typical (or actual) output of the intermittent generators, since different plants can be displaced at different times of the day and/or year. Different plants have different fuels, efficiencies and emission levels. 3. Assessment of the nature of the plants used to manage variability (the primary load-following plant) and what changes in the operation of all plants result from the addition of intermittent generation. 4. Assessment of the difference between minimum demand and the minimum output of inflexible plant. How do we work it out? Analysts need to identify the impact of intermittent output on the commitment of other plant. The main technique used to assess impacts of this nature is a time series assessment of the behaviour of all the plant on the system, and demand, at each hour of the day, throughout the year. Averages may also be used to quantify some impacts, depending on the degree of accuracy needed. For example, energy spilling may be calculated using data that indicates the average overall amount of wind output each year that is coincident with periods of low demand. More complex assessments would normally make use of time series simulation models, which represent the commitment of all the plant on the system. Getting the unit commitment ‘right’ is an economic issue, and other factors such as robust markets can also play into the unit commitment decision. Some models look ahead with perfect foresight, both in regards to load and weather forecasting, and may need some modification to take the effect of intermittency into account. Implications We explore the scale of these impacts in Ch. 3. The extent to which overall generation efficiency is reduced due to the need for other generators to vary their outputs more, or because energy is ‘spilled’ will depend on both the nature and penetration of intermittent sources, and on the nature of conventional plant on the system. In general terms, smaller and more flexible generators can assist the accommodation of intermittent sources, whereas larger and less flexible generators make efficient integration of renewables more challenging. These impacts are mediated through market signals, and it is therefore important that the benefits of flexibility and high efficiency at a range of outputs are captured in market rewards. 2.3.7 Capacity requirements to ensure system reliability Range of impacts Intermittent generation may be able to replace some conventional plant. The extent to which intermittent generation can replace thermal plant without compromising system reliability is referred to as its ‘capacity credit’. It is important to note that capacity credit is a derived term because it can only be calculated in the context of a more general assessment of reliability across peaks". See Boxes 2.3 and 2.7. It might be thought that intermittent plant cannot contribute to reliability at all since in most cases we cannot be certain that it will be available at times of peak demand (there are exceptions, see later in this section). However, there is a possibility that any plant on the system will fail unexpectedly, so reliability is always calculated using probabilities. Intermittent plant can contribute to reliability provided there is some probability that it will be operational during peak periods. Put another way, it is possible that intermittent plant will be running when a conventional plant breaks down and demand is high, so it can contribute to reliability. However, intermittent plant is usually less predictable than conventional generation, so the capacity credit of intermittent plant is usually lower, per unit of energy delivered, than it is for conventional generation. This means that there must be more installed capacity on the system than there would be without intermittent generators. Information needed How much capacity can be replaced by intermittent plant without compromising reliability is determined by the probability of intermittent generation providing electricity at periods when demands are high. Quantifying this depends upon the behaviour of demand, conventional stations and intermittent generators during the times of the year when demand is at its highest level. We need information about: |. The timing and duration of demand peaks. 2. The variability of demand during peak periods (expected demand and range of possible demand levels). 3. The expected output and possible range of outputs from conventional stations during peak periods. 4. The range of possible outputs from intermittent stations during peak periods. In principle the output of aggregated intermittent stations can fluctuate from near zero to almost 100% of installed capacity. We need to know both the expected (most likely) output at peak periods and the probabilities of the range of potential output at peak periods. Note that wider geographical dispersion will tend to reduce, possibly eliminate, the probabilities of either near zero or maximum output, and the evidence from several countries indicates that aggregated fluctuations lie within a well defined range that reaches neither zero nor maximum output”. How do we work it out? The complex relationship between the range and average output of intermittent and conventional plants, and the range and expected level of demand at peak times can be assessed using statistical algorithms or models based on statistical principles. The key determinants of capacity credit are as follows: |. The degree of correlation between demand peaks and intermittent output. — Positive correlation between high output and high demand will tend to increase the capacity credit of intermittent stations; the obverse will have the opposite effect. — For this reason, capacity credit varies considerably according to the interplay between demand and renewable resource. For example, in the UK photovoltaic panels are unable to provide any contribution to peak demands, because these peaks occur in winter evenings, when it is dark. This does not detract from the prospective benefits of PV as an energy provider. However, in some 2. The range of intermittent outputs. Where demand and intermittent output are largely uncorrelated, for example in the case of wind energy in Britain, a decrease in the range of intermittent outputs will tend to increase capacity credit. In statistical terms this is because the variance decreases. Taking wind as the example again, more consistent wind regimes decrease variance and increase capacity credit. Variance can be reduced through geographical dispersion of plants. This has the effect of smoothing outputs such that overall variation decreases as geographical dispersion increases. We explore this relationship empirically in Ch. 3. Having different types of intermittent plant on a system can also decrease variance and increase overall capacity credit. This is because different types of renewable resource fluctuate over different timescales, which also has the effect of smoothing outputs such that overall variation decreases. regions demand peaks are driven by air conditioning loads that are highest on hot sunny days, in which case there is a very high probability of significant PV output that is highly correlated with demand. PV has zero capacity credit in the UK, but can have a high capacity credit in warmer regions. Correlation occurs where both diurnal and seasonal fluctuations in demand and output show a strong coincidence, or indeed have the same cause (solar radiation in the PV example above). However, a partial relationship between demand and renewable output does not necessarily imply a meaningful correlation. Wind energy in Northern Europe tends to have higher availability and higher average output in winter, when peak demand also occurs. However, wind does not exhibit any meaningful diurnal pattern in winter months (being driven largely be weather fronts), and demand and wind output are therefore assumed in most studies to be uncorrelated on a day to day basis. 3. The average level of output. A higher level of average output over peak periods will tend to increase capacity credit. Again, taking UK wind as an example, there is little correlation between wind output and demand. However, wind farm outputs are generally higher in winter than they are in summer. For this reason analysts use winter quarter wind output to calculate capacity credit. maintain LOLR Box 2.3 een how LOLP « can be calculated statistically fro Beale the va intermittent sour LOLP as desc that the LOLP Implications The primary implication of the above is that more plant will be needed to ensure reliability than would be the case without intermittent stations. The need to retain or construct plant alongside renewable generators, in order to ensure reliability, will give rise to a cost. In Ch. 3 we review the range of findings on capacity credit, and hence the scale of its contribution to reliability, requirements for ‘back up’ and costs. Redefining system margin The second implication is that ‘system margin’ as defined in Section 2.3.2, becomes less meaningful when intermittent generation is introduced onto a system. The difference between installed capacity and expected peak demand is no longer a good indicator of how reliable supplies are likely to be. Intermittent generators will be generating at full capacity for only a small percentage of the time, and only at 30% or less of their capacity (assuming a 30% capacity factor) for half the time, and at 15% for one quarter of the time (see Annex 7). Yet they can contribute to meeting peak loads and sometimes do this when ‘conventional’ generation is down; i.e. they have a ‘capacity credit’. From the definitions given earlier, the following relationships can be derived: (i) For a system comprised entirely of ‘conventional’ plant the system margin as a percentage of peak demand is defined as: System margin = Capacity on the system — Peak demand” (ii) When intermittent generation is added to the system, it is: System margin = ‘Conventional’ capacity on the system + Capacity credit of intermittent generation — Peak demand If the capacity credit is estimated such that the loss- of-load probability with intermittent generation is the same as that on a thermal only system then system margin is the same in the two cases. This provides a familiar yardstick by which the adequacy of system margins may be assessed. Limitations to LOLP It is important to note that the LOLP function is one tool by which reliability may be measured. There are others (see Annex 6). LOLP may not capture the full range of impacts associated with intermittent generation — for example the chronology and duration of lost loads. There are a range of metrics (see Annex 6) and there has been extensive research into measures of reliability, but this is an important area of ongoing and future research. As illustrated in the figures in Box 2.7 and Annex 7, the shape of the area where load is not served may change in a system with intermittent generation. The impact of intermittency on reliability is determined by the distribution of intermittent outputs. Impacts range from low probability events with significant impact, to more frequent, but less severe, fluctuations. For example, output from a very large number of intermittent stations may be either zero or low. This ‘high impact’ event typically has very low probability, but can have a significant impact on capacity credit at large penetrations. ‘Low impact’ events, where there is a small capacity shortage, have a much higher probability but little effect of security of supply. ; Box. 2.8 The effect of the ‘low wind cold snap’ scenario Some commentators ie Meron the walidey er allocating any capacity credit to ind _generatio _ in particular, This i is ‘because of concerns about weather events that affect much of mainland Britain and result in low wind output coincident with high demand. The following quotes were provided by Graham Sinden for his presentation to the workshop « on this Plolect ‘on Sth July oor (See ire [hw ukere: ac. e-uldcontendview! 124/105/): , ‘ : “There are several periods during qa year when the UK i is covered by an anti-<cylone and there i is no a wind and al no waves.” a (rel a) i 2 we = must not lose sight of the fact that wind only blows a third of, the time.” (Foulkes 2003) It is iaporeant: to note that intermittent output need not fall to Zero, but need only to be very ioe in most areas of the country for. this concern to have signifi cance. Two important observations can be, s) _ made about the ‘cold snap low wind’ issue; the first relates to the ‘implications of such | events for Ge capacity credit, the second to. the empirical evidence that such events occur. Capacty credit is a function of probabilities, and no plant i is 100% Pelabla Even ‘if low wind events occur with regularity, capacity credit need not be zero, provided there is no direct correlation between high demand and low output. The reason is that no plant can be 100% guaranteed available >. that's why probabilities are used. If this argument is applied to conventional plant | but with ‘cold clear spell’ substituted by ‘unplanned outage’ which has the same result of a plant not generating \ when it’s > wanted, then it would follow that capacity credit of any plant i is zero and would require 100% ‘back up’. The studies reviewed i in Ch.3 provide a wide range of capacity credit estimates, but most UK | studies suggest that. capacity ¢ credit at penetrations in excess of 10% of energy | from ind are 15% - per ee 25% of the installed capacity of wind energy. Several studies have ‘observed that capacity credit would. increase if resources were more’ diverse. It is important to note however, ‘that existing estimates of “capacity credit generally use LOLP as a Measure of reliability. As we note in the text in Section 2: 2 this measure may not capture all the impacts from intermittent generation. This i is an im of ongoing research. : ae Some existing studies. use relatively short term 1 weather daa’ sets. ‘Capacity ¢ credit i is estimated most accurately’ using multiple years of data. If such ‘cold snap low wind’ events occur fich greater regularity than has been allowed for i in existing studies of the capacity credit of the UK system, it ; does have a capacity credit. In Germany the relatively poor wind regime ‘and more limited — : geographical dispersion r result i in capacity credits around half ‘that estimated for the UK (DENA Bro} Steering Group 2005; E. ON Netz 2005). This illustrates that weather is an important determinant of” capacity credit, but even in this case, capacity credit is not zero and 100% ‘back | up’ is not required. | The final point to note with regard to capacity credit and weather data i is ‘that even nif capacity credit i is zero intermittent stations can still save fossil fuel, contribute to diversity and security of ‘supply and ~ reduce emissions from fossil fuel generators. The cost implications of low and zero capacity credit | have been considered by several actcls e. eB (Dale et al 2003; llex and ‘Strbac 2002) Bnd. are discussed Sin Ch 3% ey i 3 ; seg Sa eeu additions, and all 2 the lower capacity credit of intermittent stations, can be more complex t 2.4.1 Introduction As described in section 2.3 intermittent generation brings a range of changes; these can also be differentiated in terms of benefits and costs relative to conventional technologies: Prospective benefits: — Fuel and other variable cost savings and emissions reduction as fossil fuel stations are used less — Capital and other fixed cost savings from any conventional plant that can be displaced Prospective costs: — Capital and operating costs of intermittent generation plant itself — Additional response and reserve to manage unpredicted short term fluctuations — Additional fuel burn due to increased variation in the output of load following plants — Conventional plant retained or new plant constructed to maintain reliability at peak demand — Wasted energy if intermittent output exceeds demand* We have now shown that the scale of each of these impacts can be quantified, which provides the basis for a cost (and benefit) assessment. This is highly context specific, for example, fuel costs (and hence the value of fuel savings), plant mix, plant margin and cost of reserve provision vary markedly between regions, countries or systems. In addition, short run marginal costs will differ from long run marginal costs. There may be opportunities to reoptimise the system in the longer run, which may reduce long run costs. On the other hand, short run costs may be held down through the use of older plant (for example to provide system margin) that will eventually need to be retired and replaced. This means that these costs can only be assessed from a systemic perspective. Quantification of the costs of intermittency requires a comparison of the capital, operating and fuel costs of a system with new intermittent generation against a credible counterfactual scenario without intermittent plant. Both scenarios must provide the same level of energy, power quality and reliability. Some costs (for example, additions to short term reserves) are sufficiently self-contained to permit a relatively straightforward assessment of the additional costs associated with intermittent plant. Whilst the costs in question are system specific, and accrue at a system (as opposed to individual plant) level, it is relatively easy to determine the cost in question and ‘attach’ these costs to intermittent plant. Other system costs — such as the implications of a lower capacity credit — appear to be more difficult to account for. We explain this below. A system wide approach may also be thought to militate against a traditional ‘like with like’ cost comparison between new generating options — usually based upon a ‘factory gate’ average cost figure (£L/MWh). However, there is a clear balance of evidence for this approach — see Ch. 3 for details. As we discuss in the following section, most of the problems associated with system-wide analysis are tractable. In what follows we provide a brief description of the approaches taken to costing system balancing impacts, and consider the issues surrounding the costs of maintaining reliability. “Transmission losses also will either be reduced or increased depending on how wind power is sited related to load centres. For higher penetrations these losses will probably increase. We do not deal with transmission losses in this report. *Strictly speaking lost intermittent output is not itself a cost. since marginal cost of production is zero. However. the need 2.4.2 System balancing costs 1. Response and Reserves Operating and capital costs for reserve plant used for system balancing can be calculated in a relatively straightforward fashion once the additional reserves associated with intermittent generation have been assessed as described in section 2.3.3. The usual approach is to determine the least cost option for provision of such reserves. An alternative approach is to use market prices for reserve services. Both the need for and cost of provision will vary from system to system - for example depending on the size and nature of existing reserves. The principal problem with estimating costs of reserve and response services arises from terminological, operational and regulatory differences between countries. Reserve costs vary according to which actions fall to system operators and which are dealt with by markets. We provide a range of estimates from the literature in Ch. 3. 2. System efficiency impacts The time series approach described previously allows fuel savings and efficiency losses to be accounted for. There is no other means by which these overarching aspects can have their costs quantified. It is important to note that in those electricity networks that operate through market processes, there is no single body with responsibility for optimising efficiency but rather each market participant optimises their own position such that, in an efficient market design, overall efficiency is achieved. Total system efficiency impacts and the costs thereof, are therefore something of an abstract concept for individual market participants but, nonetheless, can be monitored in terms of total fuel burn. It is important to consider the potential difference between central optimisation of fuel and emission savings and those delivered by the decentralised market. This provides a comparator against which market solutions can be judged, both in terms of costs and in terms of other impacts. 2.4.3, Costs related to capacity required to maintain system reliability In many cases adding intermittent generators to an electricity network will tend to increase the amount of plant required to provide a given measure of reliability if compared to delivering the same energy, to reliability of a thermal generation that delivers the same energy output. The total change in costs can be assessed by comparing a system that contains intermittent generators with one that meets the same reliability criteria without those intermittent generators — assuming that both systems have the same energy output. It is important to note, however, that in the UK at present there is no explicit payment for ‘reliability services’. Unlike the additional reserve and response services that intermittency might give rise to, the system operator does not contract for plant in order to maintain system margin or to act as ‘back up’ to intermittent generators. Two distinct strands of thought can be found in the literature on how to conceptualise the costs associated with any additional capacity required to maintain reliability when intermittent generators are added to an electricity network. The first does not explicitly define a ‘capacity cost’ rather it assesses the overall change in system costs that arises from additional capacity. More plant is required than would be the case in the absence of intermittent stations. The approach depends upon an estimate of the additional capacity needed to maintain reliability in order to derive capacity credit. However, this approach does not attempt to directly attribute a cost of ‘capacity reserves’ or ‘stand by’ due to intermittent stations (Dale et al 2003; Milborrow 2001). The reason for this is that there is no explicit market for, or central procurer of, such services. Some commentators note that it is possible to derive the cost of maintaining reliability using the above approach by assessing the impact on system load factors (Dale et al 2003). This is because one effect of adding intermittent generators is that the load factor of the remaining conventional stations on the system will fall, since additional capacity is needed to provide a given energy output. All new generators have the potential to affect system load factors. Quantification of these impacts is an important topic of ongoing research. However it is unlikely that intermittency will affect each type of generator equally. In fact, it is possible that particular categories of generating plant might be used to maintain reliability. These include plants used for peak demand such as oil fired stations and open-cycle gas turbines and/or older plant retained and maintained only for peaking duty. This has led The second line of thought directly costs the additional ‘capacity reserve’ put in place to ensure reliability. Using this approach, costs are assessed by costing the provision of ‘back up’ or ‘capacity reserve’ sufficient to close any gap between the capacity credit of intermittent stations and that of conventional generation that would provide the same amount of energy. Costs will vary depending upon what form of generation is assumed to provide ‘back up’. This can give rise to a degree of uncertainty, since there is no market for this ‘back up’ and the nature and cost of available ‘back up’ may vary according to system circumstance and technology. It is also not clear that we can know the /ong run marginal cost of such capacity, as this will be a product of future system optimisation (market based or otherwise), which will be affected by new technologies or practices. In a working paper that accompanies this report a simple algebraic exposition is developed which allows both techniques to be reconciled”. In principle both approaches should arrive at the same change in total system costs. Therefore, a simple identity can be derived that can be rearranged to allow the derivation of the capacity credit related cost of intermittency. Algebraic derivation of this term is provided in a working paper that accompanies this report. We provide a short description in Annex 2. This shows that the system reliability cost of intermittency = fixed cost of energy-equivalent thermal plant (e.g. CCGT) minus avoided fixed cost of thermal plant (e.g. CCGT) displaced by capacity credit of wind®*. The benefit of this approach is that it allows the capacity credit related costs associated with adding intermittent plant to the system to be made explicit in a way that is consistent with systemic principles, making no judgement about the nature of the plant that actually provides capacity to maintain reliability. Instead, all that is required is determination of the least cost energy equivalent comparator, i.e. the thermal plant that would supply the same energy in the absence of intermittent generation (normally assumed to be CCGT). This approach is used in Ch. 3 to consider the range of costs implied by the range of capacity credit estimates we found in the literature that are relevant to the British electricity network. 2.5 Summary This section has explored the principles of electricity supply system operation, and the provisions that are made through regulation and market actions to ensure that electricity demand is met by supply. We have seen how reliability is measured and maintained, and requirements estimated for a variety of reserve and response services. Demand fluctuations are substantial and not entirely predictable, whilst all forms of generation suffer occasional unplanned outages. In all cases, the effects on system reliability and efficiency can only be quantified using a system wide, and essentially statistical, approach. The principal impacts of intermittency, and their implications, are as follows: — System balancing impacts. These include both the additional response and reserve requirements that must be purchased by the system operator and the effects on market participants. They reflect the need to manage and accommodate fluctuations over the period from seconds to hours. — Capacity to ensure system reliability. This relates to the capacity that must be built or retained on the system with intermittent generation to ensure that a defined measure of reliability of supply during peak demand is maintained. Ch. 3 reviews the empirical evidence on each of these issues and the history and nature of the studies that have been undertaken into intermittency. It seeks evidence on the following questions: — What is the scale, and range of estimates, of additional reserves that are required to accommodate intermittent generation? — Can we quantify other impacts, such as efficiency losses? — How much does this cost? — What is the scale, and range of estimates, of the capacity credit of intermittent renewables? — What are the reasons for this range? — What are the implications for the UK? There are important issues relevant to the integration of renewables that this report deals with only briefly. Others lie entirely outside its scope. The impacts of supply interruptions and of various scenarios of renewables development are both examples. There are also limitations to existing approaches to estimating the impacts of intermittent generation, for example, the range of impacts captured in the reliability measure LOLP. In many areas research is ongoing, both empirical and analvtical. 3 Evidence on the costs and impacts of intermittency 3.1 Introduction The remainder of this chapter provides the following information: This section provides an overview of the findings from the in-depth review of the literature on * Overview of historical developments in research intermittency undertaken for this assessment. It on intermittency seeks to identify where the weight of evidence lies, * Quantitative findings: and understand the origins of contention. Ch. | and — Additional reserve and response services for Annex 5 describe the protocol, search terms and system balancing; databases used to gather data. As noted, these are — Other system balancing impacts derived from best practice in systematic review and — The capacity requirements to ensure informed by the stakeholder workshop and expert reliability group. — Implications for costs * Discussion of key issues A total of 212 documents were reviewed, of which * Conclusions 58 were excluded because they were irrelevant or duplicative. The remaining 154 documents were categorised according to the major issue which each document addresses and the approach adopted by the authors. The categories and numbers of documents falling into each category are summarised in Table 3.1. Table 3.1: Overview of the evidence base Primary aspect covered Method/approach. = | Number of documents Reliability, reserves and balancing _| Statistical and/or time series simulation 64 Review 57 Other I Sub-total 122 Connection, transmission and N/A 19 network issues Resource characteristics N/A 13 Total included documents 154 Total excluded documents 58 EEE ee | Total all documents 212 3.2 Historical development of research on intermittency The systematic search undertaken for this assessment revealed a rich and technically detailed literature spanning more than 25 years. The focus of work has changed over time, reflecting the evolution of understanding, the development of wind power in several countries and changing market and regulatory context. This section provides a short review and lists some key studies by way of example. It is important to note that the most recent period revealed by far the largest number of reports, and we include only a short excerpt here. Early studies: 1978 - 1989 The literature uncovered in our review dates from 1978, with an initial cluster of reports dating from this time until 1987. Many studies were carried out by, or for, what were then state owned utilities and in response to the OPEC induced oil price shocks. Many studies focus on the basic principles of how to represent intermittent generators on an integrated network. Most are concerned with transmission system level reliability, reserve and balancing issues, with a particular focus on the role of wind and other renewables as ‘fuel savers’. In all cases the ‘context’ is very different, in that centralised operation of electricity networks was still extant in all countries. As a result ‘optimisation’ of networks Table 3.2: Example studies 1978 - 1989 with intermittent sources is conceptualised in rather different terms than it is currently. However, the technical issues are largely unchanged. Many reports are concerned with development of methodological principles and apply these only to relatively simple — and obviously at that time hypothetical — scenarios. Date | Author ; Title ; betes 1978 | Johanson E, Goldenblatt M An economic model to establish the value of WECS to a utility system. 1979 | General Electric; W D Marsh Requirement assessment of wind power plants in utility systems 1979 | Rockingham A A probabilistic simulation model for the calculation of the value of wind energy to electric utilities 1980 | Farmer ED, Newman VG, Economic and operational implications of a complex of wind-driven Ashmole PH power generators on a power system 1980 | Rockingham AP System economic theory for WECS 1980 | Zaininger Engineering Co. Wind power generation dynamic impacts on electric utility systems 1981 | Whittle G The effects of wind power and pumped storage in an electricity generating system 1982 | Gardner GE, Thorpe A System integration of wind power generation in Great Britain 1982 | Meier RC, Macklis SL Interfacing wind energy conversion equipment with utility systems 1982 | Moretti PM, Jones BW. Analysis method for non-schedulable generation in electric systems 1983 | Danish Energy Ministry Vindkraft | Elsystemet 1983 | Brian Martin, John Carlin Wind-load correlation and estimates of the capacity credit of wind power:An empirical investigation 1983 | Brian Martin, Mark Diesendorf The economics of large-scale wind power in the UK, a model of an optimally mixed CEGB electricity grid 1983 | Halliday JA, Lipman NH, Studies of wind energy integration for the UK national electricity grid Bossanyi EA, Musgrove PJ 1983 | Yamayee ZA, Ma FS Effect of size and location of conventional and intermittent generation ‘on system reliability 1984 | Halliday JA Analysis of wind speed data recorded at 14 widely dispersed U.K meteorological stations 1987 | Swift-Hook DT Firm power from the wind 3 Methodological development: 1990 - 1999 During the 1990s some differences of emphasis emerged relative to the earlier analyses. There was a marked decrease in the number of utility studies compared to the early 1980s, though academic work continued in the US, UK and Nordic countries. One notable addition to the body of knowledge in this period was a series of ten country studies sponsored by the European Commission. The break up of national monopolies is possibly reflected in a Table 3.3: Example studies 1990 - 1999 marked reduction in emphasis on the benefits of wind and other renewables (fuel saving and system optimisation). Instead, work in this period has a noticeable focus on detailed methodological issues and in particular costs of system balancing and calculation of capacity credit. Several studies pay attention to methodological refinement and development, for example, through incorporation of stochastic variables into simulation models. Date | Author i Tithe es eee ie aaa i 1990 | Holt, Milborrow, Thorpe Assessment of the impact of wind energy on the CEGB system 1991 | Grubb The integration of renewable electricity sources 1991 | Grubb Value of variable sources on power systems 1992 | EC Commission Wind Power Penetration Study, The Case of Denmark 1992 | EC Commission Wind Power Penetration Study, The Case of Germany 1992 | EC Commission Wind Power Penetration Study, The Case of Greece 1992 | EC Commission Wind Power Penetration Study, The Case of Italy 1992 | EC Commission Wind Power Penetration Study, The Case of Portugal 1992 | EC Commission Wind Power Penetration Study, The Case of Spain 1992 | EC Commission Wind Power Penetration Study, The Case of the Netherlands 1993 | Billinton & Gan Wind power modelling and application in generating adequacy assessment 1993 | Bouzguenda, Rahman Value analysis of intermittent generation sources from the system operations perspective 1993 | Soder Reserve margin planning in a wind-hydro-thermal power system 1993 | Watson SJ, Landberg L, Halliday | Wind speed forecasting and its application to wind power JA integration 1993 | Yih-huei Wan, Brian K.Parsons | Factors Relevant to Utility Integration of Intermittent Renewable Technologies 1994 | South Western Electricity plc Interaction of Delabole wind farm and South Western Electricity‘s Distribution system 1994 | Watson SJ, Landberg L, Halliday | Application of Wind speed forecasting to the integration of wind JA energy into a large scale power system 1995 | Michael R.Milligan, Alan Miller, Estimating the Economic Value of Wind Forecasting to Utilities Francis Chapman 1996 | Reconnect Ltd Wind turbines and load management on weak networks 1996 | Milborrow D Capacity credits - Clarifying the issues 1996 | Wind energy weekly How Difficult is it to Integrate Wind Turbines With Utilities? 1997 | Michael Milligan, Brian Parsons | A Comparison and Case Study of Capacity Credit Algorithms for Intermittent Generators 1999 | RJ Fairborn Electricity network limitations on large scale deployment of wind energy Renaissance: 21st century research The beginning of the 2!st century saw a very significant increase in research activity on intermittency. References for the last five years outnumber those from both the previous decades by more than two to one. Whilst most utilities have been privatised, system operators, regulators and governments have funded a significant number of studies. Attention to methodological issues has been sustained and extended. In addition, an increasing amount of empirical data has been combined with increasingly sophisticated scenarios of wind power and other intermittent generation installation. Table 3.4: Example studies from 2003 & 2004” ABB Inc. Date | Author Title es : 2003 | Dale, Milborrow, Slark, Strbac A shift to wind is not unfeasible (Total Cost Estimates for Large scale Wind Scenarios in UK) 2003 | Doherty R, O'Malley M Quantifying reserve demands due to increasing wind power penetration 2003 | Dragoon K (PacifiCorp), Assessing Wind Integration Costs with Dispatch Models:A Case Milligan M (NREL) Study of PacifiCorp 2003 | Environmental Change Institute | The Practicalities of Developing Renewable Energy Stand-by University of Oxford Capacity and Intermictency Submission to The Science and Technology Select Committee of the House of Lords 2003 | Milligan M Wind Power Plants and System Operation in the Hourly Time Domain 2003 | Mott MacDonald The Carbon Trust & DTI Renewables Network Impact Study Annex 4: Intermittency Literature Survey & Roadmap 2003 | SeckT GRE wind integration study 2003 | Sveca J, Soder L Wind power integration in power systems with bottleneck problems 2003 | Xcel Energy Characterizing the impacts of significant wind generation facilities on bulk power systems operations planning 2004 | Auer H Modelling system operation cost and grid extension cost for different wind penetrations based on GreenNet 2004 | Bach P Costs of wind power Integration into Electricity Grids: Integration of Wind Power into Electricity Grids Economic and Reliability Impacts 2004 | Brooks D L,Anthony J, Lo E, Quantifying System Operation Impacts of Integrating Higgins B Bulk Wind Generation at We Energies 2004 | Doherty R, Denny E, O'Malley M| System operation with a significant wind power penetration 2004 | E.ON- Net Z Wind report 2004 2004 | Electric Systems Consulting Integration of Wind Energy into the Alberta Electric System Stage 4: Operations Impact 2004 | EnerNex Corporation, Wind Xcel Energy and the Minnesota Department of Commerce Wind Logics Integration Study - Final Report 2004 | Holttinen H The impact of large scale wind power production on the Nordic electricity system 2004 | Ilex, The Electricity Research Operating Reserve Requirements as Wind Power Penetration Centre (ERC), The Electric Increases in the Irish Electricity System Power and Energy Systems Research Group (EPESRG) 2004 | KEMA-XENERGY for Intermittent wind generation: Summary report of impacts on grid California energy commission system operations 2004 | Milborrow D Assimilation of wind energy into the Irish electricity network ~ 2004 | Royal Academy of Engineering, | The Costs of Generating Electricity PB Power 2004 | Soder L Simulation of wind speed forecast errors for operation planning of 3.3. Quantitative findings 3.3.1 Introduction This section provides an overview and discussion of the principal quantitative findings from the literature, through meta-analysis of the included studies and reports. It discusses the main ranges and the reasons for differences between studies. It also provides a review of the different measures, or metrics, utilised in different studies and considers the potential for confusion that might arise from this and other factors. The main issues identified in Ch. 2 provide the basis for the following categorisation of findings, described in the principal sub-sections below: — System balancing part |: Response and reserve services (impacts and costs) — System balancing part 2: Other system efficiency impacts — Capacity requirements to ensure reliability We begin with a general overview of the characteristics of the literature and limitations to the search. 3.3.2 Overview of the evidence base Sixty-seven documents were found to contain quantitative data under one or more of the headings above®. It should be noted that this number does not include those documents which presented results of other studies (provided that these other studies had been captured by the search process). The number does, however, include those reports which presented data from other studies and introduced additional new data (care was required to ensure that in these cases findings were not double counted). Relatively few studies attempted to measure the cost attributable to the (usually) Of the sixty-seven documents, twenty-four are academic research papers, six are collaborative studies undertaken by academic and industry representatives, two commissioned by a learned society, sixteen are reports by industry participants (generating/supply companies, network operators or trade bodies), and nineteen are from/by government or government bodies such as regulators and executive agencies. Nearly all of the studies reviewed focused exclusively on wind generation rather than intermittent renewable generation as a whole, which reflects the relatively advanced penetration of wind power relative to other emerging renewables. Limitations The search may have a number of limitations: — The principal focus was on English language references and those commonly translated into English. This (along with the success of wind energy in these countries) may explain the predominance of Nordic, US, UK and Irish studies. There are relatively few studies from Spain. — The search engines utilised may not have revealed the full range of government and industry reports in all countries. — Whilst citation trails were followed, notably from key references highlighted by the expert group, it was not possible to follow each and every citation and reference in the time available. Hence, whilst every attempt has been made to be extensive, the review is by no means exhaustive. Nevertheless, this review provides by far the most extensive assessment of this nature that has been undertaken to date in the mainstream literature. General observations One striking characteristic of the data is the range of different metrics used to assess the impacts. This means that for each of the categories of impact identified above, the numbers are presented in several different formats. This creates the potential for confusion and the risk that comparisons between results is not on a genuinely like for like basis. Attempts to normalise data from a range of studies to facilitate comparison run the risk of losing important detail or, at worst, suggesting that figures are comparable when they are not. The differing methods of presentation of data within each heading are identified in tabular form for each of our categories of impact, described below — see Tables 3.6 to 3.10. We present the principal/most common metrics in graphical form. Even where studies have used ostensibly the same metric it is not always possible to compare the results because a study has focussed on a particular element’ of a metric, or other system dimensions are not declared. Examples include studies which do not identify the extent to which intermittent generation displaces existing plant. Other studies are not explicit regarding total intermittent generation levels, total system capacity, or total system demand, all of which hamper the derivation of the penetration level. These issues do not imply a criticism of the studies reviewed — they are used to illustrate that it is prudent to exercise caution when drawing comparisons between results. These risks not withstanding, the remainder of this section presents the quantitative findings through a combination of charts and tables. Where there is a particular issue of comparability, this is identified. It is also notable that attention to capacity credit tends to focus on the scale of the impact (i.e. calculation of capacity credit), with limited attention to costs thereof. By contrast more studies that consider system balancing provide cost estimates than provide an indication of the scale of the impact. Finally, a relatively small number of studies provide quantitative evidence on system efficiency effects such as fuel saving. We discuss each of these points in more detail below. “For example, some studies include only the reserve requirement for frequency regulation and at the other extreme some appear to include an element of system margin for reliability requirements. “If a paper exoresses Denetration level using a metric based on system capacity and the capacity credit is not specified. what 3 Out of a total of 18 studies that provide quantitative evidence on the additional reserve services associated with increasing penetrations of intermittent generation, ten use these two metrics, although two of the studies do so in a way which means that they cannot be represented on the chart. 3.3.3 System balancing part 1: response and reserve services Impacts (MW and % additions to reserve requirements) The additional reserve and response requirements attributable to intermittency are presented in Figure 3.1. In this figure we present findings which estimate additions to reserves in two ways: Two other types of presentation were found in the literature. These findings are presented in Table 3.6 at the end of this chapter: — Asa percentage of installed intermittent generation capacity at given levels of intermittent generation penetration, and where penetration level is expressed as the percentage of total system energy provided from intermittent generation. These appear as a point or series of points in Figure 3.1. — Asa percentage of installed intermittent generation capacity, but no penetration level given. These appear as a horizontal line in Figure 3.1. Reserves expressed as a percentage of installed intermittent generation capacity at given levels of intermittent generation penetration, where penetration level is expressed as the percentage of total system installed capacity provided from intermittent generation (four studies use this formulation). Expressed as a percentage of installed intermittent generation capacity at given levels of intermittent generation penetration, where penetration level is expressed as the installed intermittent generation capacity as a percentage of peak system load (four studies use this formulation). Figure 3.1 Range of findings related to additional reserves with increasing penetration of intermittent supplies Additional reserve requirement 80 7 = ee é 70 E 4 : - - - 3 50 | s 368 } a 6 s | #€ 30: 35 | * 2 j 7 3 20 | 7b earn” 186 10 7 8 : a eEeEEeEeEeEeEeEeEeEeEeEeEeE—EeEEEEEEE———e t oe ste 229 oo Ot 7 7 0 5 10 15 20 25 30 35 40 Intermittent aeneration penetration level (% of total system enerav) Comments on the range of values for reserve requirements” 5% penetration level There are only four data points at this penetration level, representing data from just two studies. The striking characteristic is the very high outliers from a German study (E.ON Netz 2004) (reference number 57) which are two orders of magnitude greater than the other values (Holttinen 2004) (ref.67). The wording of the Eon Netz report is such that it is not clear whether the ‘reserve’ costs that they cite refer to balancing services only, or also include an element of capacity provision that reflects the relatively low capacity credit of German wind farms™. Moreover, particular difficulties are faced within the Eon Netz region, which has extensive wind energy developments: — Factors which tend to exacerbate the scale of swings in output: Low average wind speeds and low capacity factor for wind output (see Ch. 2 for an explanation of the relationship between these factors and capacity credit); and substantial ‘clustering’ of wind farms in the North West of the control area which limits potential smoothing of short to medium term fluctuations. — Limited interconnection with regions to the East and West. — ‘Gate closure’ 24 hours ahead of real time, which means that the forecasting error that must be managed by reserve plants is much larger than it would be in the UK and other countries with a much shorter period between scheduled unit commitment and real time. E.ON ‘firms’ up the wind based on day ahead wind forecasts, independently of the load forecast errors. This increases the reserve requirement, and associated costs. By contrast Holttinen covers a large area (4 countries) and only considers the sub-hour variations of wind power. 10% penetration level There are eight data points at this penetration level, which lie in the range between approximately 2% and 8%. The high value is from the Dena Grid Study (DENA Project Steering Group 2005) (ref.74) - another report based on the German electricity 20% penetration level At this level of intermittent generation, six of the seven data points are in the range between approximately 3% and 9%. There are no low outliers, but one higher value of 19% (DENA Project Steering Group 2005) (ref.74). Penetration level not specified These values are represented in Figure 3.1! as horizontal lines, since it is unclear what penetration level they represent. Two studies, (Milligan 2001) and (Hudson et al 2001) (refs.186 and 229) lie within the normal range of the 10% and 20% penetration levels described above. A third (Doherty et al 2004a) (ref.178), at 25%, is above the trend. This finding reflects lreland’s small system size and limited interconnection. Other comments Different analysts use different definitions of ‘reserves’, which means that a range of impacts are captured. For example, some studies look exclusively at ‘spinning’ reserve (part loaded plant), and so have not included the impact of intermittent generation on other system balancing services such as the level of standing reserve. Others identify figures for frequency control and load following reserve but do not analyse the impact on generating unit commitment (the requirement to instruct plant in advance of when it is required to allow sufficient time for it to be brought into operation). Within the data on reserve impacts and costs we have included (but not shown on figure 3.1) a notable outlier (Royal Academy of Engineering and PB Power 2004) (ref.239). This report is difficult to categorise. This is because the report does not use the systemic approach to estimating system costs common to other studies, but works on the premise that wind generation requires dedicated back up. Since this back up would be expected to provide both balancing and reliability, the data in this study are therefore a combination of system balancing reserves and capacity installed to maintain reliability. This highlights the scale of the implications of methodological differences and the importance of terminology to estimates of the impacts of intermittency. 5 Response and reserve services costs Twenty-three studies provide quantitative evidence on costs associated with additional reserve and response requirements attributable to the addition of intermittency. The main findings are represented in Figure 3.2. In this figure we present findings which used the following approach: — Cost per MWh of electricity from intermittent generation at given levels of intermittent generation penetration, where penetration level is expressed as the percentage of total system energy provided from intermittent generation (as in figure 3.1). Fifteen studies use this approach. The database contains a further eight studies, summarised in Table 3.7 and discussed below, which used the following metrics: Cost per MWh of electricity from intermittent generation at given levels of intermittent generation penetration, where penetration level is expressed as the percentage of total system installed capacity provided from intermittent generation. Four studies use this approach. Cost per MWh of electricity from intermittent generation at given levels of intermittent generation penetration, where penetration level is expressed as the installed intermittent generation capacity as a percentage of peak system load (three studies use this formulation). Cost per MWh of electricity from intermittent generation, but no penetration level given. One study use this approach. Figure 3.2 Range of findings on the cost of additional reserve requirements* Additional reserve costs 9.00 095 8.00 3 ee 235 a ale —_ = cpleeie nasal aatenerai sas =e — 7.00 j aia aan ee eeEEeeEeeee as = 6.00 95 5.00 +2065 = = = = 1 5 Intermittent generation level penetration level (% of total system energy) 10 15 20 T 1 25 30 35 40 45 Key: 51 (Mott MacDonald 2003), 67(Holttinen 2004), 79(Dale et al 2003), 83(Ilex and Strbac 2002), 89(Milborrow 2004), 95(Bach 2004), 125(Ilex et al 2004), 129(Pedersen 193(Hirst 2002), 199(Hirst 200), 206(Fabbri et al 2005), et al 2002), 132(Milborrow 2001), 187(Seck 2003), 232(Dale 2002), 235(Milborrow 2005) Shaded area represents the range of values for UK studies Comments on the range of values for costs of additional reserve requirements with intermittent generation® 5% penetration level” Four of the seven data points lie in the range between £0.6 and £1.7/MWh. There is one very high outlier, which presents interpretation and analysis of data from the Eon Netz 2004 report —a value of £8.1/MWh (Milborrow 2005) (ref.235). It is worth noting that this report also highlighted some of the particular difficulties faced within the Eon Netz region (discussed previously in section 3.3.3). The other relatively high figure of £4.3-£4.8/MWh, from (Fabbri et al 2005) (ref. 206), reflects the price of procuring electricity in the Spanish market to cover the difference between predicted and actual generation from wind plant. 10% penetration level At this level of intermittent generation, there are eleven data points with values ranging from £0.2 to £2.9/MWh. There are no clear outlying values at either the upper or lower end of this (wide) range. 15% penetration level Values range from £0.5 to £2.8/MWh, with seven data points, again with no clear outliers. 20% penetration level There are eight data points, six of which lie in the range £1.3 to £3/MWh, with two high outliers of £5.6 and £8.4/MWh from one of the studies relating to the Danish Eltra system (Bach 2004) (ref.95). We note that there is a disparity between the numbers presented in this study and the other study based on the Eltra system (Pedersen et al 2002) (ref.129). Other comments Studies which refer to the share of installed capacity, rather than energy, have costs in the range £0.4 and £4.81/MWh. Twenty studies, out of a total of twenty-three, conclude that additional costs are less than £5.0/MWh of intermittent output at penetration levels of up to 20%, with estimates ranging between £0.2 and £4.81/MWh. One study suggests costs remain in this range at much higher penetrations. All studies find that reserve costs tend to rise as penetration level increases, but the range of costs across studies is broadly similar at each penetration level, that is, there is no appreciable convergence or divergence as penetration rises. The difference between individual studies is typically larger than the increase in costs within each study resulting from increasing penetration levels. This suggests that the reserve cost is particularly sensitive to assumptions about system characteristics, existing reserves, and what is included within the definition of reserve requirements (see Ch. 3). The study that does not show a penetration level (Royal Academy of Engineering and PB Power 2004) (ref.239) is an extremely high outlier at a cost of £17/MWh. This report has the unusual characteristics noted previously, and appears to be an amalgamation of balancing and reliability costs. 3.3.4 System balancing part 2: other system efficiency impacts Intermittent stations will affect the operation of generating plant other than, and in addition to, operating reserves (see Ch. 2 for further details). Load following plant may be required to respond to variations in intermittent generation, which may affect efficiency. In addition, if the output of intermittent plants cannot be absorbed by the system energy may need to be discarded. Both factors will serve to decrease the potential value of intermittent generation in terms of its ability to deliver fuel savings and emissions reductions. This sub-section considers these two factors, and discusses the quantitative evidence available on the scale of their impacts. There is limited evidence on these impacts, in part reflecting the influence of market rules, available transmission to neighbouring countries and flexible demand on the results. 3 Fuel and carbon dioxide emissions savings In the UK's electricity system, output from renewable generators would normally be expected to displace electricity generated in conventional plants burning coal and natural gas (usually coal plants). The theoretical maximum fuel and emissions savings would be realised where each MWh of renewable electricity displaces a MWh of fossil fuel electricity and where the conventional plant fuel burn is reduced accordingly (i.e. where there is no loss of operating efficiency resulting from the conventional plant's reduced output). In practice, the theoretical savings are reduced because the efficiency of conventional plant can be affected by intermittent renewable generation in two ways: — Through an increase in the variability of generation (increasing losses as plant is more frequently shutdown and restarted, or output is ramped up or down). — Asa result of a lower overall load factor, even if output is relatively stable at the lower load factor (because generating plant fuel efficiency is typically maximised at close to a plant's designed output). This section is concerned with the extent to which the theoretical fuel and carbon dioxide emissions savings are reduced as a result of these potential efficiency losses. There are a number of factors which can influence the net savings: The operational mode of the conventional generating plant on the system is key. At one extreme is the ‘fuel saver’ mode in which all of the conventional thermal plant that would be operated if there had been no intermittent renewable generation is left running but output (and therefore fuel burn) is reduced when intermittent generation is able to supply electricity. The result is that there will tend to be more part-loaded conventional plant running than is really needed, which will exacerbate efficiency losses. A closer to optimal approach is the ‘forecast’ mode in which renewable electricity generation is predicted (for example, based on forecast wind speeds) and surplus conventional plant (ref, 26), (Ilex et al 2004) (ref.125), (Holt et al 1990) (ref.160), (Denny and O’Malley 2005) (ref.181) conclude that better forecasting of intermittent resource availability will maximise fuel and carbon dioxide emission savings. Some commentators have argued that the design and operation of the electricity market can affect the potential fuel and carbon dioxide savings. If the market penalties for intermittent generation significantly exceed the true system cost, as may be the case in the UK (BWEA 2005) (ref.50), (Milborrow 2001) (ref.132), then generators with a mix of conventional plant and intermittent renewables may keep more conventional generation on-line than is theoretically required (from a whole- system perspective). They do this in order to avoid the market penalties which they would otherwise be exposed to as a result of their intermittent generation. The consequence may be a system that as a whole is operated sub-optimally (with the attendant efficiency losses). The type of conventional thermal plant that is displaced by intermittent generation has a major effect on carbon dioxide savings. This is a consequence of the much higher carbon content of coal (per kWh of energy) compared to natural gas, and the greater thermal efficiency of combined cycle gas turbines. Strictly, this is not a pure intermittency issue, but is related to it because the operating characteristics of different conventional plant may make it more or less likely to be displaced by intermittent renewable generation. If coal fired plant is displaced then the carbon dioxide savings will be greater than if gas-fired plant is displaced. The issue of displacement can be assessed analytically based on the operating characteristics of existing (or planned) generating units. However, there appears to be a continuing debate as to what type of conventional plant is displaced by wind generation in the UK, e.g. (BWEA 2005) (ref.50), (Milborrow 2004) (ref.89). A small number of studies have explicitly addressed the efficiency losses of thermal plant resulting from intermittent renewable generation (those that use the C2 and C5 metrics described in table 3.8 at the end of this chapter). There is no evidence to suggest that efficiency is reduced to such a degree as to significantly undermine fuel and carbon dioxide emissions savings: — The fuel savings not realised because of the reduced efficiency tend to increase as intermittent generation penetration level rises but the actual losses are generally small - up to the 20% penetration level, the studies present efficiency losses ranging between a negligible level and 7% (as a percentage of theoretical maximum fuel savings). Energy Spilling Energy spilling will occur when the available renewable generation from installed plant at a particular point in time exceeds the ability of the system to absorb it, or when it is not economic to continue operating the intermittent generation. The circumstances under which energy may have to be spilt are dependant on a range of system characteristics (Denny and O'Malley 2005) (ref.181) (Sveca and Soder 2003) (ref.18), (Holttinen 2004) (ref.67), (Bach 2004) (ref.95): — The point at which intermittent renewable generation capability is not fully utilised will be lower on systems with a high proportion of inflexible plant. This inflexibility may be a result of technical constraints (such as in the case of nuclear generation, or Combined Heat and Power plants bound to the heat demand), or policy constraints (such as the ‘must-run’ peat fuel plants in Ireland). — Itis important to note, however, that all systems will require some minimum quantity of plant that can provide the full range of frequency response, reactive power and other essential system services and that some types of intermittent renewable plant do not supply such services. — The degree of correlation between the renewable resource availability and demand will have a major impact on the threshold at which energy spilling will occur. The threshold will be lower for those systems where high resource availability is positively correlated with periods of low demand. — The level of spare capacity on the transmission lines between areas of high resource availability and areas of high demand will also influence energy spilling. If the resource is remote from demand and the transmission system has little spare capacity, the likelihood of transmission bottlenecks will be higher. Such bottlenecks could require that generation capacity is constrained off the system, and potential electricity generation would be lost. The studies identified in table 3.9 show that the proportion of energy spilt tends to increase as the intermittent generation penetration level rises. The conclusion is that: — At penetration levels up to approximately 20%, the spilt energy ranges between zero and less than 7% for five out of the six studies. The remaining study (which relates to the transmission network-constrained Swedish system) concludes that energy spill levels would reach 16.7% at an 11% penetration level, assuming all the new wind capacity was located in the north of the country and there were no grid reinforcements to the south. 3.3.5 Capacity requirements to ensure reliability: the capacity credit of intermittent generation Twenty-nine studies provide quantitative evidence on the capacity credit of intermittent generators. All use a statistical or simulation approach based upon a measure of reliability such as LOLP*. The main findings are represented in Figures 3.3 and 3.4. In this figure we present findings which use the following metric: — Capacity credit expressed as a percentage of installed intermittent generation capacity at given levels of penetration, where penetration level is expressed as the percentage of total system energy provided from intermittent generation. Nineteen studies use this approach. 3 The database contains a further ten studies, summarised in Table 3.10 and discussed below, which use the following metrics: — Percentage of installed intermittent generation capacity, where penetration level is expressed as intermittent generation capacity as a percentage of peak system load. Three studies use this approach. — Percentage of installed intermittent generation capacity, where penetration level is expressed as the percentage of total system installed capacity provided by intermittent generation. Four studies use this approach. — Percentage of installed intermittent generation capacity, but no penetration level given. Three studies use this approach. Comments on the range of values for capacity credit® 5% penetration level At this level of intermittent generation the capacity credit values lie in a range between 17% and 35% for all but one of the ten studies which provided data points at this penetration level. There are no particularly marked outlying values at the upper end of the range. The one clear low outlier (E.ON Netz 2005) (ref.246) has a value at 8% capacity credit that is less than half that of the next lowest value. 10% penetration level At this penetration level, eleven of the twelve data points lie in the range 15% to 30%. The values follow a similar pattern to those at the 5% penetration level with no clear upper outliers, but one low outlier (DENA Project Steering Group 2005) (ref.74), with a value that is half that of the next lowest. Figure 3.3: Range of findings on capacity credit of intermittent generation & £ & N a _ a = o Capacity Credit (% of installed intermittent generation capacity) 8 o | i : | ° | | °o 5 10 15 Intermittent generation penetration level (% of total system energy) — + 20 25 30 35 40 Key for studies used in figure 3.3: | 7(Watson 2001), 51(Mott MacDonald 2003), 74(DENA Project Steering Group 2005), 79(Dale et al 2003), 83(Ilex and Strbac 2002), 121 (Giebel 2000), 160(Holt et al 1990), 204(Grubb 1991), 238(Martin and Carlin 1983), 240(Commission of the European Union 1992b), 241 (Danish Energy Ministry 1983), 242(Commission of the European Union 1992d), 243(Commission of the European Union 1992a), 244(Commission of the European Union !992g) 246(E.ON Netz 2005), 247(Sinden 2005), 248(Commission of the European Union 15% penetration level At this penetration level, seven of the eight data points lie in the range 11% to 20%. The one low outlier (ELON Netz 2005) (ref.246), has a value that is well under half that of the next lowest. There are no upper outliers. 20% penetration level At this penetration level, six of the seven data points lie in the range 15% to 20%. The lower boundary of this range, at 15% capacity credit, runs counter to the general trend that capacity credit falls as penetration level rises (the corresponding value at 15% penetration was |1% capacity credit). This is because the studies which tended to provide the lower boundary numbers at the 5%-15% penetration levels do not extend as far as the 20% penetration level. The trend for one low outlying value continues in line with other penetration levels, with the low outlier (DENA Project Steering Group 2005) (ref.74) value being considerably less than half the next lowest. Other comments The capacity credit data in Figure 3.3 have two clear messages - firstly that all these studies conclude that intermittent generation does have a capacity credit value greater than zero, and secondly that capacity credit expressed as a percentage of intermittent capacity declines as the penetration of intermittent generation rises®. Findings from other capacity credit metrics (see Table 3.10) show the same trends. There is a moderate amount of convergence within the data available - that is to say that the range of findings narrows slightly as renewables penetration increases. One study (Commission of the European Union 1992g) (ref.244) does not follow the progressive downward trend of all others. This is thought to be caused by the methodology adopted by this particular study, which displaced specific conventional installations of varying size as more wind generation was modelled on the system. The findings also demonstrate the sensitivity of the capacity credit to resource availability and the degree of correlation between resource availability and periods of high demand. Capacity credit values are adversely affected where there is a low degree of correlation between resource availability and peak loads. This is particularly well illustrated in an early US study (General Electric and Marsh 1979) (ref.217), which used data from four separate sites in the US. The lowest capacity credit value in this study was from a site with negative correlation between resource availability and load and a mismatch between actual wind speeds and wind turbine design speeds. Studies relevant to British conditions, all of which focus on wind power, indicate that output and demand are largely uncorrelated. The relationship between resource and capacity credit is also demonstrated by studies using data from operating wind farms in a region with low average wind speeds. At each of the penetration levels described above, the low outlying values are from the German studies (DENA Project Steering Group 2005) (ref.74) and (EON Netz 2005) (ref.246). The results show the effect that the relatively weak wind resource in Germany has on the capacity credit value (see Ch. 3 for an overview of the relationship between average output and capacity credit). It is not clear why two other German studies - (Auer 2004) (ref.84) and (Commission of the European Union 1992c) (ref.250), produced results that lie within the normal range. Figure 3.4 presents the distribution of findings at the 10% penetration level. It shows a cluster of findings in the 20 - 25% range, which accords well with recent UK studies reviewed in detail in the UKERC working paper that accompanies this assessment (see Box 3.1). It also indicates that 80% of the findings reviewed here provide estimates of capacity credit in the range 15% - 30%. It is important to note that there is a direct relationship between the capacity credit and the amount of additional thermal capacity required to maintain reliability. This is because capacity credit is calculated by adding thermal plant in order to maintain a defined standard for reliability such as LOLP (see Ch. 2, box 2.7). The amount of capacity required is also a function of the capacity factor of conventional plant and of intermittent generators". We explore the range of capacities, and associated costs, under a set of assumptions relevant to British conditions in the following section. 3.3.6 The costs and capacities of maintaining reliability with intermittent generation Defining a convention for cost allocation There is some controversy over the means by which the cost implications of the relatively low capacity credit of intermittent stations should be calculated. As we discuss in Ch. 2, some analysts consider such costs as manifesting themselves through a reduction in system load factor, whilst others have assessed the cost of various forms of ‘back up’ capacity. In contrast to the system balancing costs discussed in Section 3.3.3, there does not appear to be a generally accepted approach to calculating ‘reliability costs’. In fact, many assessments note capacity credit, but do not attempt to derive an associated cost term at all, which is why we are unable to report a cost range in the analysis above. In order to overcome this difficulty, UKERC have developed a simple formula that makes explicit the additional costs of maintaining reliability’. This can be added to the balancing costs discussed in Section 3.3.3 in order to provide a total ‘cost of intermittency’. The formula can be expressed in words as follows: Reliability cost of intermittent generation = Fixed costs of energy equivalent thermal plant minus the avoided fix costs of thermal plant that is displaced by the capacity credit of wind. In what follows we use this formulation to provide an indication of the range of capacity costs that are associated with a sub-set of the range of capacity credits reported in Section 3.3.5. Reliability costs under UK conditions The table below takes a range of capacity credits for 10% and 20% penetration of wind energy from the data assembled in Table 3.10. The range is chosen to represent UK relevant findings, and is also close to the centre of the range of the findings in Figure 3.3. We combined this range of capacity credits with fixed data for total system size, thermal equivalent capacity costs, thermal equivalent capacity factor, and wind capacity factor. These data represent a future least cost thermal comparator, GB electricity system and wind output*. In each illustration, the only figures changed are the capacity credit and total wind®. Figure 3.4: Frequency distribution of findings for capacity credit where intermittent generation provides 10% of energy Capacity credit values 30% = 25% = 20% | 15% - % of studies 10% 0% 15 Capel ence en et tet ila onsets “Further work is needed on the means by which these impacts can be quantified in order to allow a transparent comparison of the effects on incumbent generation of adding various types of new generation plant. “See Annex 2 and Ch 2 for the derivation of this formulation. “The data are derived from a recent and widely cited UK study (Dale et al 2003). “This is a simolification. since canacitv factor and capacity credit are related variables. Nevertheless the range of canacitv 3 This analysis is predicated on the principle that the reliability component of the costs of intermittency can be determined only through a comparison between the contribution of an intermittent generator to reliability and that of a thermal generator which provides the same amount of energy. The actual cost of providing system reliability will always be system and context specific. Dedicated peaking plant, maintaining older power stations that can be made available for a small number of peak hours each year, storage, and demand management may all offer the most cost effective means to provide system margin. Implications for additions to thermal capacity There is a MW corollary of these cost ranges. The formulation noted in Section 3.3.5% would provide the following ranges of additional thermal capacity, expressed as a percent of installed intermittent capacity, assuming the capacity factors and capacity credits as per Table 3.5. Additional thermal capacity (10% energy from intermittents): | 1.2% - 21.8% Additional thermal capacity (20% energy from intermittents): 15.2% - 22.1% Zero and low capacity credits There is widespread consensus about the range of capacity credit relevant to UK conditions. However some British studies explore the possibility of very low or zero capacity credit in recognition of the concerns highlighted in Box 2.8 (Dale et al 2003; Ilex and Strbac 2002). If capacity credit were zero and all other characteristics held as per Table 3.5, costs of maintaining reliability would rise to £9/MWh of wind energy. It is also important to note that capacity credits and capacity factors are linked, reflecting the fact that a lower capacity factor is usually associated with a lower capacity credit. Lower capacity factor results in higher costs per unit of output. Non-UK studies, particularly those from Germany noted above, exhibit low capacity factors relative to Britain, and commensurately lower capacity credits. Such conditions result in modest increases in reliability costs - however this is because much larger capacities are needed to supply an equivalent amount of energy, hence generating costs and total costs rise considerably (we explore a range of capacity factors in annex 2). Table 3.5: Relationship between capacity credit and reliability cost, GB relevant capacity credits and system characteristics System characteristics Wind energy penetration level Capacity | Reliability cost : = ne creditrange =| (LIMWhofwind) 10% (40 TWh of wind energy, 13 GW of wind 19.4% £4.76 installed) 30% £2.44 20% (80 TWh of wind energy, 26.1 GW of wind 19.1% £4.82 installed) 26% £3:32 400 TWhiyr Total system energy Wind capacity factor 35% Thermal equivalent capacity factor” 85% Thermal equivalent cost £67,000/MW/year “Assumptions taken from Dale et al 2003, and seeking to represent a future GB electricity system with demand of 400 TWhiyr, a mix of on and off-shore wind, and where CCGT continues to provide the least cost form of new electricity generation plant. “A key principle of chis approach is that comparator plant is assumed to be lowest cost new generation. Such plant would be operated at maximum capacity factor (CF), and is assumed here to be CCGT. We use 85% CF as an approximation; in fact some new plant exceeds this. Availability at peak demand is probably higher (above 90%, see National Grid winter outlook revort). whilst system load factor (typically around 58%) or thar of rhe enrire fleet of CCGT as onerated at present (tvnicallv 3 Implications for the cost of intermittency This analysis suggests that adding intermittent generation to the British electricity network will impose a capacity/reliability cost of less than £5/MWh with a 20% penetration of intermittent generation, with a range that starts a little above £3/MWh*. Section 3.3.2 indicates that the majority of estimates of system balancing costs are also less than £5/MWh, in many cases substantially less, and the range for UK studies is £2 - £3/MWh. Section 3.3.4 also indicates that there may be an efficiency reduction mediated through the electricity market of the order of 1% of the electrical output of intermittent generators. The impact of this on overall costs is likely to be negligible. Hence, the total cost of intermittency at a 20% penetration on the British electricity network is likely to be in the range of a little under £5/MWh up to around £8/MWh. This range accords well with a range of UK studies reported in Box 3.1. Box 3.1 Case study comparison of Capacity and Balancing Costs: Estimates of four well known studies compared to estimates from first principles A comparative assessment between four key UK studies and the work from first principles undertaken in support of this assessment*. This work provides a review of the statistical approach to estimating provisions for both system balancing and long term reliability. This box is taken from the working Paper. It compares the key estimates in the case where there is a high level of wind energy on the system (in the range 15% to 20%). SCAR Carbon R.A. UKERC Dale et report Trust Eng. working al paper Capacity Factor for Wind, %* 35 35 35 35 35 Capacity creidt for wind, MW/MW. wind capacity, % 19.2 22.9 20.0 Not estd 22.1 Capacity required to ensure reliability MW thermal/MW wind, %¢ 18.9 18.3 21.2 65.0 19.1 Cost of this capacity 0.39 0.26 0.45 1.86 0.44 Energy costs of increased variability 0.08 Not estd | Notestd | Not estd 0.05< Balancing costs 0.27 0.22 0.20 Not estd 0.25° Total costs, p/kWh 0.74 0.48° 0.65 1.86 0.74 Source data used in the UKERC working paper is from: (Dale et al 2003), (Ilex and Strbac 2002), (Mott Macdonald 2003), (Royal Academy of Engineering and PB Power 2004). In order to provide a consistent data set adjustments haye been made to some variables — for example all figures are presented at a 10% discount rate, which may vary from figures in the originals. *See Anderson, 2005, Power System Reserves and Costs with Inermitcent Generation http://www.ukerc.ac.uk/content/view/124/105/ Notes: The estimates in each report have been converted to a 10% discount rate: the tables attached to the working paper provide the original estimates at the discount rates thay have used plus further footnotes on assumptions. All costs are in p/kWh. a Capacity factors for wind are based upon the assumption that roughly half the capacity will be offshore. b Not an independent estimate, as discussed above, but based on the estimates of the first three eports (the upper estimate in the case of the Carbon Trust). c The report by Dale et al is based on 20% energy market share for wind, that shown here for the present study is 15%, which partly accounts for the higher estimate in the former. d- Estimated directly by the R.A. Eng. and the present study, and for the other studies inferred from the identity between capacity credit and capacity costs of intermittency presented above. The working paper refers to capacity reserves, terminology has been adapted to ensure consistency with the rest of this report. f — Represents the incremental cost of an increase in from 10% to 20%. There is reasonable agreement among four of the above five studies. Estimates of the capacity credit range from 19% to 23% of wind capacity. The range of estimates of the overall costs of providing for capacity margin and balancing is from 0.65 to 0.74p/kWh of electricity generated by wind after allowing for transmission losses. 3 3.4 Discussion of key issues from the quantitative evidence General comments Unless the assumptions and characteristics of the system being analysed are very clearly understood there is a danger that the results are misinterpreted, or that invalid comparisons are drawn. It is apparent from analysis of each study that the results of any individual work are sensitive to a set of system characteristics: — The existing generation mix (in particular the degree of flexibility of existing plant and suitability for part loading, and the rate at which existing plant can increase or decrease output). — Existing requirements for reserve services for system balancing. — The spatial distribution of intermittent generation plant. — The mix of intermittent generation technologies. — Transmission network constraints and size of links to other networks. — The absolute level of renewable resource available and the degree of correlation of resource availability with demand peaks and troughs. — Generating unit commitment time horizon and accuracy of renewable resource forecasting. — The overall system reliability/security target level. It is important to note that data limitations, methodological details and scope of impacts/costs may differ between studies. It is only possible for this report to highlight significant outliers and general trends. Relevance of simulation and empirical studies The majority of the studies reviewed use simulated data, real data extrapolated or real data run through a range of models. The main exception is experience from Germany's Eon Netz, which tends to show relatively high costs for reserves. Moreover, it has been contended that experience in Denmark and Germany suggests that simulation studies in the UK may have failed to capture the extent of prospective fluctuations”. However, it is also important to note that experience cannot supersede simulation if the experience is not directly relevant. We would not conclude (for example) that PV should have a significant capacity credit in the UK because of experience with solar plants in California. It is also important to note that there are important differences between Denmark, Germany and the UK: — Denmark is a small country and the scope for geographical dispersion is limited. The system must also integrate output from heat demand- constrained CHP plant, and has very high penetration level of wind energy. (Bach 2004; Holttinen 2004; Pedersen et al 2002) - Denmark is heavily interconnected to both the Nordel and German electricity systems and hence able to manage intermittency in ways unavailable to the UK. — We have discussed some of the differences between Britain and Germany (most notably the lower capacity factor of German wind farms) and the specific issues that relate to the geography and operating practices of the Eon Netz region. It is also clear that the DENA Grid Study, which looks at a wider geographical area, takes a more optimistic view than Eon Netz. It is important that key problems are not ‘assumed away’. Some existing studies explicitly explore key effects, such as regional concentration of some renewables (Ilex and Strbac 2002). However, others have assumed that wind energy will be geographically dispersed and hence may have failed to identify an important prospective cost. It has been suggested that wind developments tend to cluster in areas with good wind resources, and that in future large individual offshore developments may present problems for system operators”. These impacts must be explored in analytic research and monitored as empirical evidence increases. 3.5 Summary of key findings Exactly where in each range of values a particular study falls depends on the penetration level of intermittent generation, the characteristics of the system being modelled, and the methodology adopted by the study. Summary of impacts on system balancing reserves The majority of the studies which are applicable to the UK find that up to an intermittent generation penetration level of 20%, the additional reserve requirements imposed on the system are generally less than 10% of the installed capacity of the intermittent generators. The studies which present higher reserve requirements either represent systems which are not directly comparable to the UK or use a methodological approach which is not consistent with widely accepted practice. Different system operation principles, such as determining reserves from day-ahead prediction errors as in Germany, are also relevant. All the studies that present reserve requirements over a range of intermittent generation penetration levels show that the reserve requirement, expressed as a percentage of intermittent generation capacity, will rise as the penetration level increases. Summary of impacts on system efficiency Only a small number of studies explicitly address the efficiency losses of thermal plant resulting from intermittent renewable generation. Losses tend to increase as the intermittent generation penetration level rises but the actual losses are small. At the 20% penetration level, the studies present efficiency losses ranging between a negligible level and 7% of intermittent output. The studies which address energy spilling show that the proportion of energy spilt tends to increase as intermittent generation penetration level rises, but the proportion of energy spilt is relatively small - at penetration levels up to approximately 20%, the spilt energy ranges between zero and less than 7% for all but one of the studies. The remaining study relates to a transmission network-constrained system and concludes that it may be more economic under some circumstances to spill energy rather than dimension the transmission network to cove with Summary of capacity requirements to ensure reliability: capacity credit All the studies show that intermittent generation does contribute to system reliability through a positive capacity credit, and that capacity credit expressed as a percentage of intermittent output declines as intermittent generation penetration level rises. The capacity credit value is, however, particularly sensitive to the degree of correlation between resource availability and peak demand periods and to geographical dispersion. This is reflected in the relatively wide range of results at each penetration level. Nevertheless, 80% of the studies concluded that, at the 10% penetration level, the capacity credit lies in the range between 15% and 30%. A significant proportion of the studies do not extend to the 20% penetration range, but most of the studies that do present a capacity credit range between 15% and 20%. Those studies that present lower capacity credit values relate to systems with relatively low resource availability (compared to UK conditions), poor correlation between peak demand and intermittent output, or both. Summary findings on costs Over 80% of the studies concluded that the cost of providing additional reserves would be less (and in many cases substantially less) than £5 per MWh of intermittent generation at intermittent generation penetration levels up to, and in some cases exceeding, 20%. British relevant studies fall into the range £2 - £3/MWh. Those studies which present higher costs relate either to systems with much higher penetration levels, or where resource availability is not comparable with Britain, or are based on methodology that is inconsistent with UK regulatory and system operation practices. Costs of maintaining reliability fall into the range £3 - £5/MWh for penetrations up to 20% and under British electricity system and weather conditions. 3 3.6 Summary of all findings and data used in Ch. 3 Table 3.6 Summary of additional reserve requirements with intermittent generation Power, 2004, The cost of generating electricity system plant margin requirements Document reference, author, date and title Metric type and notes — Penetration Reserve range ae ; R Sie Bee Rs Ge heey level ranges f2u| 5: =) ae 26,Watson et al, 1994, Application of wind RI, spinning reserve only 9.9-37.9% 6.6-24.5% speed forecasting to the integration of wind so numbers not in figure energy into a large scale power system 3.1 Report also has values for higher penetration levels but excluded due to the influence of discarded energy 51, Mott MacDonald, 2003, Carbon Trust RI 10-20% 3.3-7.6% and DTI intermittency survey & roadmap 57, E.ON Netz, 2004,Wind report 2004 RI, appears to include 4% 50-80% system plant margin requirements 67, Holttinen, 2004, The impact of large scale RI 5-20% 0.8-4.2% wind power production on the Nordic electricity system 74, Dena, 2005, Dena grid study RI 10-20% 8.3-19.4% 79, Dale et al, 2003,A shift to wind is RI 20% 5% not unfeasible 6, Doherty, 2005, A new approach to R2 quantifyreserve demand in systems with significant installed wind capacity 14, Doherty & O'Malley, 2003, Quantifying R2 13-31% 3-7% reserve demands due to increasing wind power penetration 42, Dragoon and Milligan, 2003, Assessing R2 (figure is for % 3-23.8% 2-103% wind integration costs with dispatch increase in reserve) models:A case study with PacifiCorp 117, Kema-xenergy, 2004, Intermittent wind R2 10% 0.6% generation: summary of report of impacts ‘on grid system operations 178, Doherty, 2004, Wind penetration studies R3 [NIA 25% on the island of Ireland 186, Milligan, 2001,A chronological reliability R3 NIA 11-20% model to assess operating reserve allocation to wind power plants 219, Farmer at al, 1980, Economic and R3, values based on N/A ar 7-16% operational implications of a complex of 5-10GW of wind capacity. wind-driven generators on a power system UK cotal capacity in 1980 was approximately 63GW | 239, Royal Academy of Engineering & PB R3, appears to include N/A 65% 173, Electric Systems Consulting ABB Inc, R4, report described as 13.3% 10-40% 2004, Integration of wind energy into the preliminary analysis and Alberta electric system - stage 4: operations does not use the ‘sum of impact squares’ rule to combine demand and wind variance 191, Milligan, 2003, Wind power plants and R4, values are for the 5.7-22.7% 3.4-12.4% system operation in the hourly time domain additional load following requirement imposed by having wind generators on the system 229, Hudson at al, 2001, The impact of wind R4, value is for ‘regulation’ 4.5% 0.2% generation on system regulation only (frequency response) requirements _ Number of Metric description : RI: Percentage of installed intermittent generation capacity, where intermittent penetration ‘studies — 6 level is expressed as the percentage of total system energy from intermittent generation. R2: Percentage of installed intermittent generation capacity, where intermittent penetration 4 level is expressed as the percentage of total system installed capacity from intermittent generation. R3: Percentage of installed intermittent generation capacity, but no penetration level given. 4 R4: Percentage of installed intermittent generation capacity, where intermittent penetration 4 level is expressed as the installed intermittent generation capacity as a percentage of eak system load. peak sys lee Total 18 Table 3.7 Summary of findings relating to reserve costs Document reference, author, date and title. Metric type and notes "| Penetration © Reserve ae Pee | level range Fange 51, Mott MacDonald, 2003, Carbon Trust and RCI 10-20% £1.7-£2.9 DTI intermittency survey & roadmap 67, Holttinen, 2004, The impact of large RCI 10-20% £0.6-£1.3 scale wind power production on the Nordic electricity system 79, Dale et al, 2003,A shift to wind is RCI 10-20% £2.5-£3.0 not unfeasible 83, Ilex & Serbac, 2002, Quantifying the RCI, incremental cost of 17-27% £2.3-£2.7 system costs of additional renewables in moving from 7% to 17% 2020 & 27%, figures are for reserve costs but this report does have figures for capacity costs as well. Figures are for the ‘north wind, high demand’ scenario. 89, Milborrow, 2004, Assimilation of wind RCI 5-20% £0.6-£1.6 energy into the Irish electricity network 95, Bach, 2004, Costs of wind power RCI 21% £5.6-£8.5 Integration into Electricity Grids: Integration of Wind Power into Electricity Grids Economic and Reliability Impacts 125, llex at al, 2004, Operating Reserve RCI, figures derived from 8-17% £0.15-£0.6 Requirements as Wind Power Penetration modelling individual days, Increases in the Irish Electricity System not whole year - these are the highest cost days (for some scenarios the cost is negative) 129, Pedersen, 2002, Present and future RCI 16.3% £1.8 integration of large scale wind power into Eltra's power system 132, Milborrow, 2001, Penalties for RCI 10-45% £1.5-£3.3 intermittent generation sources 187, Seck, 2003, GRE wind integration study RCI 2.4-9.5% £2.1-£2.9 193, Hirst, 2002, Integrating wind energy RCI 6% £1.1-£1.7 with the BPA power system: preliminary study 199, Hirst, 2001, Interactions of wind farms RCI, numbers are for 0.1% £0.6-£2.3 with bulk-power operations and markets ‘regulation’ (frequency response) and load following only 206, Fabbri et al, 2005, Assessment of the RCI, figures are the 4% £4,3-£4.8 Cost Associated With Wind Generation market costs of Prediction Errors in a Liberalized Electricity procuring the difference Market between predicted and actual generation 232, Dale, 2002, NETA and wind RCI 2-10% £0.1-£2.4 235, Milborrow, 2005, Windstats newsletter RCI, figure derived from 6% £8.1 analysis of the EON ALA. ee 42, Dragoon and Milligan, 2003, Assessing RC2, numbers are for 20% £3.5 wind integration costs with dispatch load following and unit models:A case study with PacifiCorp commitment only 45, EnerNex and Wind Logics, 2004, Xcel RC2 13.1% £2.6 Energy and the Minnesota Department of Commerce Wind Integration Study - Final Report Wind integration study - final report 46, Xcel Energy, 2003, Characterizing the RC2 3.5% £1.2 impacts of significant wind generation facilities on bulk power systems operations planning 239, Royal Academy of Engineering & PB RC3, appears to include N/A £17.5 Power, 2004, The cost of generating system plant margin electricity requirements 84, Auer, 2004, Modelling system operation RC4, figures are for 5.1-30.4% £0.04-£1.0 cost and grid extension cost for different reserve costs but this wind penetrations based on GreenNet report does have figures for capacity costs as well 151, Brooks et al, 2004, Quantifying System RC4 4-289% £1.1-£1.6 Operation Impacts of Integrating Bulk Wind Generation at We Energies 229, Hudson at al, 2001, The impact of wind RC4, value is for 4.5% £0.04 generation on system regulation ‘regulation’ only requirements (frequency response) Metric description RCI: Cost per MWh of electricity from intermittent generation, where penetration level is expressed as the percentage of total system energy provided from intermittent generation. RC2: Cost per MWh of electricity from intermittent generation, where penetration level 4 is expressed as the percentage of total system installed capacity from intermittent generation. RC3: Cost per MWh of electricity from intermittent generation, but no penetration level given. | RC4: Cost per MWh of electricity from intermittent generation, where intermittent penetration 3 level is expressed as the installed intermittent capacity as a percentage of peak system load. Total 23 Table 3.8 range of findings for fuel and carbon dioxide savings metrics Document reference, author, date and title Metric type and notes Penetration i : Beit | level range . 22, Doherty et al, 2004, System operation with Cl, lower value is for fuel 17.6% Power System Operation and Emissions Reduction speed forecasting to the integration of wind energy into a large scale power system a significant wind power penetration saver mode, higher value is for forecast mode 181, Denny and O'Malley, 2005,Wind Generation, | Cl 5.4-10.5% 3.5-9% 26, Watson et al, 1994, Application of wind SACD aad es 9.9-48.3% 0-48% 79, Dale et al, 2003,A shift to wind is not unfeasible | C2 20% 221,Whittle, 1981, Effects of wind power and C2 2.5% pumped storage in an electricity generating system 50, BWEA, 2005, Blowing Away the Myths c3 125, llex at al, 2004, Operating Reserve | C4, figures derived from 8-17% Requirements as Wind Power Penetration modelling individual days, Increases in the Irish Electricity System not whole year - these are the peak demand days L 222, Halliday et al, 1983, Studies of wind energy cs 15-42% integration for the UK national electricity grid 223, Gardener and Thorpe, 1983, System integration] Cé 20-60% level is expressed as the percentage of total system energy provided from intermittent generation. of wind power generation in Great Britain 67, Holttinen, 2004, The impact of large scale C7, lower figure is based 4-12% 300-700g/kWh wind power production on the Nordic on wind displacing mainly electricity system CCGT plant, higher figure based on displacing mainly coal plant Metric description Number ofa 2 : : y studies Cl: Total CO2 savings in percentage terms, where penetration level is expressed as the percentage 2 of total system installed capacity provided from intermittent generation. C2: Reduction in CO2 savings (when compared to theoretical maximum savings), where penetration 3 C3; CO2 savings per kWh of electricity from intermittent generation, but no penetration level given. C4:Total CO2 savings in percentage terms, where penetration level is expressed as the percentage of total system energy provided from intermittent generation. CS: Reduction in CO2 savings (when compared to theoretical maximum savings), where penetration level is expressed as the percentage of total system installed capacity provided from intermittent generation. Cé6:Total CO2 savings in percentage terms, where penetration level is expressed as the percentage is expressed as the percentage of total system energy provided from intermittent generation. 18, Sveca and Soder, 2003, Wind Power ESI 3-11% 1.9-16.7% Integration in Power Systems with Bottleneck Problems 178, Doherty, 2004, Wind penetration studies ESI 13-38% 0-40% on the island of Ireland 222, Halliday et al, 1983, Studies of wind Es 15-42% 2-45% energy integration for the UK national electricity grid ‘ oe : 26, Watson et al, 1994, Application of wind ES2 9.9-48.3% 0-39.2% speed forecasting to the integration of wind energy into a large scale power system 132, Milborrow, 2001, Penalties for Ese 10-15% 0.1-0.7% intermittent generation sources 223, Gardener and Thorpe, 1983, System Es3 20-60% 2-56% integration of wind power generation in Great Britain Metric description " ‘Number of ES!: Percentage of intermittent generation output which must be spilled, where penetration level 3 is expressed as the percentage of total system installed capacity provided from intermittent generation. ES2: Percentage of intermittent generation output which must be spilled, where penetration level 2 ES3: Percentage of intermittent generation output which must be spilled, where penetration level is expressed as the percentage of peak system energy demand provided from intermittent generation Total Table 3.10 Range of findings for capacity credit ‘Document reference, author, date and title ; Metric type and notes © : Penetrat ae fy pat pate eet! | Tevel range 17,Watson, 2001, Large scale integration of ccl 20.5% wind power in an island utility - an assessment of the likely variability of wind power production in Ireland 51, Mott MacDonald, 2003, Carbon Trust and ccl 10-20% 27.5-20% DTI intermittency survey & roadmap 74, Dena, 2005, Dena grid study ccl 10-20% 7.7-6% 79, Dale et al, 2003,A shift to wind is not ccl 20% 19.1% unfeasible 83, llex & Strbac, 2002, Quantifying the system CCI, Figures are for the 17-27% 22.9-18.4% costs of additional renewables in 2020 ‘north wind, high demand’ scenario. 121, Giebel, 2000, The capacity credit of wind ccl 20% 19.3% energy in Europe, estimated from reanalysis data 160, Holt et al, 1990, CEC Wind energy ccl 2-15% 31-15.6% penetration study 204, Grubb, 1991, The integration of renewable ccl 5-38% 35-19.3% electricity sources 238, Martin & Carlin, 1983,Wind-load ccl 5-20% 35.2-15.1% correlation and estimates of the capacity credit of wind power:An empirical investigation. 240, EC, 1992,Wind power penetration study, ccl 5-15% 30-20% the case of Denmark? 241, Danish Energy Ministry, 1983,Vindkraft | ccl 5-15% 23-11% Elsystemet 242, EC, 1992,Wind power penetration study, ccl 2.5-15% 38-17% the case of Greece 243, EC, 1992,Wind power penetration study, ccl 5-15% 20-13% the case of The Netherlands 244, EC, 1992, Wind power penetration study, ccl 1.5-15% 10-15.6% the case of Spain 246, EON Netz, 2005, Wind report 2005 ccl 4.7-12.5% 8-4.8% 247, Sinden, 2005, Wind power and the cCcl 10% 23.1% resource 248, EC, 1992,Wind power penetration study, ccl 7.8-30.6% 36.5-22.9%the case of Portugal 249, EC, 1992, Wind power penetration study, ccl 2.5% 22.6% the case of Italy 250, EC, 1992,Wind power penetration study, ccl 10% 15% the case of Germany 84, Auer, 2004, Modelling system operation cc2 5.1-30.4% 35.2-22.9% cost and grid extension cost for different wind penetrations based on GreenNet 104, GE Energy Consulting, 2005, The effects CC2, the report has a 10% 10% of integrating wind power on transmission much higher capacity system planning, reliability and operations credit value of 36% for a single offshore site 203,Wan & Parsons, 1993, Factors relevant cc2 1-10% 41-15% to utility integration of intermittent 45, EnerNex and Wind Logics, 2004, Xcel Ccc3 3.5-13.1% 33.8-26.7% Energy and the Minnesota Department of Commerce Wind Integration Study - Final Report Wind integration study - final report 59, Royal Academy of Engineering, 2003, Ccc3 11.6-31.3% 26.7-16% Response to the House of Lords Science and Technology committee inquiry into the practicalities of developing renewable energy 117, Kema-xenergy, 2004, Intermittent wind cc3 4.8% 25.9-0% generation: summary of report of impacts on grid system operations 217,GE & Marsh, 1979, Requirements CC3, covers 4 different 5-20% 5-2% assessment of wind power plants in electric sites 22-6% utility systems 37-17% 47-28% 133, Garrad Hassan and Partners, 2003, CC4, value based on up N/A 20% The impacts of increased levels of wind to 800MW of wind Penetration on the electricity systems of the capacity. Island of Ireland Republic of Ireland and Northern Ireland systems total capacity is approximately 7.5GW 137, Milligan, 2001, Factors relevant to CcCc4 N/A 21-51% incorporating wind power plants into the generating mix in restructured electricity markets 212, Milborrow, 1996, Capacity credits - CC4, results from several N/A 58-7% clarifying the issues earlier studies Metric description | Number of : a pasties te : Fa Meaae tay pees cleustudies cay CCI: Percentage of installed intermittent generation capacity, where penetration level is expressed 19 as the percentage of total system energy provided from intermittent generation. CC2: Percentage of installed intermittent generation capacity, where penetration level is expressed 3 as intermittent generation capacity as a percentage of peak system load. CC3: Percentage of installed intermittent generation capacity, where penetration level is expressed 4 as the percentage of total system installed capacity provided from intermittent generation. CC4: Percentage of installed intermittent generation capacity, but no penetration level given. 3 Total 29 Conclusion This report is the product of a systematic review of the literature on the costs and impacts of intermittent generation. It seeks to provide an overview of the main results of the review, together with a non-technical exposition of the key principles of electricity network operation. It is international in scope but draws out the key findings relevant to the British electricity network. It assesses the integration of intermittent renewables in the immediate future and on the basis of incremental change to electricity network design and operation. Its principal conclusions are summarised below. 4.1 The impacts of integrating intermittent generation None of the studies reviewed in our assessment suggest that intermittency is a major obstacle to the integration of renewable sources of electricity supply. Almost all of the literature deals with the impacts of intermittency using a statistical representation of the main factors, or through simulation models based upon statistical principles. At the levels of penetration foreseeable in the next 20 years, it is neither necessary nor appropriate to allocate dedicated ‘back up’ or reserve plant to individual renewable generators when these are integrated into modern electricity networks. Nevertheless additional capacity is likely to be needed, and operational changes will need to be made. The primary impacts and costs introduced through connecting increasing amounts of intermittent supply arise from additional system balancing actions and the need to install or maintain capacity to ensure reliability of supplies. Such costs cannot be assessed without a counterfactual that permits the costs of intermittent sources to be compared to those imposed by conventional generation making an equivalent contribution to energy and reliability. 4.2 The costs of integrating intermittent generation System balancing costs For intermittent penetrations of up to 20% of electricity supply most studies estimate that costs are less than £5/MWh of intermittent output, in some cases very substantially less. The range in studies relevant to Britain is £2 - £3/MWh. These costs arise from the need to schedule additional response and reserve plant to manage unpredicted fluctuations on the timescale from minutes to hours. Additional system balancing reserves represent no more than 5-10% of installed wind capacity in the vast majority of cases. System balancing services are purchased directly by the system operator, and additions can be calculated directly using statistical techniques. They are not controversial, and although there is a range of estimates in the literature the reasons for the range are well understood. System balancing will also be undertaken by market participants as prices change in response to predicted fluctuations in intermittent output. This, together with the additional system balancing actions under the control of the system operator, may affect the efficiency with which thermal generators operate and hence give rise to costs. These costs may be revealed through markets or calculated using system simulations. Most studies find that efficiency losses are a small fraction of the energy output of intermittent generators; typically no more than a few percent. The costs of maintaining reliability Our analysis suggests that adding intermittent generation to the British electricity network will impose a capacity/reliability cost of less than £5/MWh with a 20% penetration of intermittent generation, with a range that starts a little above £3/MWh. This range is based upon results relevant to Britain revealed in our review of the literature, and uses the convention for costing the impact of intermittency on reliability chat we describe in section 4.4 below. These costs arise because the amount of capacity required to meet a given measure of reliability will increase when intermittent generation is added to an electricity network. Intermittent generators are, generally speaking, less certain to be generating power at times of peak demand than conventional generators. Capacity credit is a measure of the contribution that intermittent generation can make to available capacity at times of peak demand. It is expressed as a percentage of the maximum instantaneous output of the generators. There is a range of estimates for capacity credits in the literature and the reasons for there being a range are well understood. The range of findings relevant to British conditions is approximately 20 - 30% of installed capacity when up to 20% of electricity is sourced from intermittent supplies. In percentage terms, capacity credit falls as the intermittent generation penetration level rises. Capacity credit and additional conventional capacity required to maintain a given level of reliability are corollaries. The smaller the capacity credit, the more capacity will be needed to maintain reliability. This in turn determines the reliability costs highlighted above. In addition, capacity credit expressed as a percentage of installed intermittent capacity declines as the share of electricity supplied by intermittent sources increases. For this reason costs also increase as penetration of intermittent generation rises. The total costs Total costs of intermittency comprise system balancing costs plus the costs of maintaining reliability. In Britain these are likely to lie in the range £5 - £8/MWh (0.5p - 0.8p/kWh) of intermittent output. This range is sensitive to a number of factors, as we discuss below. 4.3 Factors that affect the costs of integrating intermittent generation System balancing Smoothing through aggregation and better forecasting decreases costs: System balancing costs will tend to be higher if the output of intermittent generators fluctuates more rapidly or more substantially over short time periods, if fluctuations are less predictable or if decreases in renewable output and increases in consumer demand correlate strongly. System operators are concerned primarily with aggregate fluctuation, potentially from large numbers of generators. Decreasing the correlation between the output of individual generators decreases aggregate fluctuation, effectively smoothing outputs. This means that wide geographical dispersion and a diversity of renewable sources tends to decrease system balancing costs. Interconnection between regions can further decrease costs. Conversely, geographical concentration will increase cost, and it may be that wind developments tend to cluster in regions with the best resource. Much larger individual wind farms could be developed, particularly offshore, increasing the fluctuation seen at an individual connection point. Both factors need further research. The nature of conventional plant and regulatory practice affect costs: The characteristics of renewable sources are not the only determinant of system balancing impacts. The nature of thermal plants operating on the system and regulatory practices are also relevant. — If system balancing actions are determined close to real time (known as ‘gate closure’, which occurs one hour before time in Britain) system balancing costs are minimised, since intermittent output can be forecast with a high degree of accuracy at such timescales. In countries where balancing decisions are made a long period ahead of time (gate closure is up to a day ahead in some regions) forecasting, and indeed demand fluctuations and failures of conventional plants, is much less accurate. Reserve costs rise as a result. — In general terms, relatively small and flexible plants assist the integration of intermittent renewables and reduce balancing costs. A large penetration of inflexible thermal generating units would make it more difficult to absorb large amounts of renewable output and increase the likelihood of intermittent output being curtailed. Large single generating units can also have a significant impact on reserve requirements, since the system needs to be able to cope with the sudden failure of the largest generating unit. This requirement will usually have a much larger impact on system balancing reserves than the fluctuations introduced by renewable generators. Capacity to maintain reliability Output over peak periods is the principal determinant of the cost to maintain reliability: The costs for maintaining reliability at times of peak demand are determined by the capacity credit of intermittent generators. This depends upon average output during peak periods, the geographical dispersion of generators and the relationship between fluctuations in electricity demand and intermittent output. Correlations between peak output and peak demand can either increase or decrease capacity credit: Strong positive correlations can lead to high capacity credit. At the other extreme, if peak demand always correlates with low or zero output, capacity credit would be very low or zero. Where demand and intermittent output are uncorrelated, average output and the distribution of output over peak periods determines capacity credit: Other things being equal, higher average output will lead to higher capacity credit. The wider the variance of output the lower the capacity credit. Variance is reduced through geographical dispersion and diversifying the range of intermittent sources utilised. In all cases, capacity credit is a derived term and cannot be calculated independently of a wider assessment of system reliability. It is context and system specific. It is also important to note that all types of generating plant have the potential to affect the utilisation of incumbent generation plant. More work is needed to provide a transparent methodology for assessing these impacts and how they differ between technology types. 4.4 Confusion and controversy A number of factors give rise to confusion in the literature, and may be one reason for ongoing debate on the subject of intermittency. There is a widespread tendency for terminology to be used in different ways: Words can be given multiple meanings. A good example is the use of ‘reserve’. In some studies ‘reserve’ is specifically operational reserve, used for short term balancing. In others it also denotes the ‘back up’ capacity required to maintain reliability because intermittent generators have capacity credits lower than their capacity factors. We contend that this confusion over language gives rise to widespread misunderstanding. It can result in inappropriate cost comparisons across studies and give rise to ongoing confusion and disagreement. The literature also exhibits a wide range of metrics through which the costs and impacts of reserve and balancing issues are expressed. This makes cross comparison hazardous, which also serves to perpetuate conflict and debate. There has been some controversy over how to estimate the costs associated with the additional thermal capacity required to maintain reliability: Some studies have assessed the costs of the capacity required to maintain reliability based on assumptions about the nature of plant providing ‘system reserves’. Others have assessed only the change in the total costs of the electricity system as a whole. There is broad agreement between both approaches on the total change to system costs”. We recommend that the ‘reliability cost of intermittency’ be defined as follows: The additional cost of adding a unit of intermittent generation to an electricity system, over and above the direct costs of investing in and operating the intermittent generator, compared with the cost of building and operating conventional generating plant at base load. This can also be expressed as: Reliability cost equals the fixed cost of energy-equivalent thermal plant (e.g. CCGT) minus the avoided fixed cost of thermal plant (e.g. CCGT) displaced by capacity credit of wind”. "There is a range of costs associated with ‘back up’ and the range arises from differing assumptions on the nature of the plant The comparison with conventional generating plant at baseload is crucial to the calculation. Policymakers and others often seek to compare the average costs of different types of generating plant on a ‘like with like’ basis — for example the cost of wind power compared to the cost of coal power. This usually uses levelised costs (£/MWh). If intermittency costs are calculated any other way meaningful comparisons of this nature are impossible. 4.5 Recommendations for policy We recommend that additional steps are put in place to continuously monitor the effect of intermittent generation on system margin and existing measures of reliability. The effectiveness of market mechanisms in delivering adequate system margin also need to be kept under review. Policies need to encourage widespread geographical distribution of intermittent generators if the costs of intermittency are to be minimised. A judgement is needed on the relative costs of intermittency and transmission upgrading. This cannot be done without the development of detailed scenarios recommended below. Intermittent generation can make a valuable contribution to energy supplies but, to ensure reliability of supply, investment in thermal capacity is also required. In the short run older plant may provide system margin but, in the long run, investment in new capacity will be needed. Flexible and reliable generation is an ideal complement to intermittent renewables. Policy should encourage and not impede investment in plant that is well suited to complement renewable energy sources and contribute to both reliable operation and efficient system balancing. 4.6 Issues for further research In some countries wind development has clustered in specific geographical regions, and problems have been highlighted recently by some system operators. Some of the literature assumes wide geographical dispersion. The impacts of geographical clustering, its likelihood and interaction with transmission cost issues needs to be better understood. Related to this, much larger individual wind farms are envisaged, particularly offshore. The implications of their fluctuations need to be better understood. The risk of demand being unmet is characterised statistically, and the measure commonly used to quantify this risk is called Loss of Load Probability (LOLP). This measure defines the likelihood that some load is not served, and the normal convention in advanced electricity networks is that LOLP is kept very small. This is done by ensuring that the generation capacity on the system exceeds peak demand by some amount, known as the system margin. There is some debate over the extent to which existing measures of reliability, particularly LOLP, fully capture the changes that arise when intermittent sources are added to the network. This is because intermittent generation changes the nature of the statistics used to calculate risk, and not all of these changes are represented within existing measures of reliability. Most of the studies reviewed in this report take an incremental approach and assess the impacts of intermittent generation on existing electricity networks. Optimisation of operating practices, development of electricity systems and new technologies designed to facilitate the integration of intermittent sources could radically reduce the costs of integrating intermittent generation. Conversely, some technologies and practices are not well suited to the efficient integration of intermittent generation. Analysis through modelling and scenarios could assist our understanding of the prospects for this. We recommend that more research is therefore undertaken on the following topics: — Renewable energy deployment scenarios in which intermittent generation is clustered in particular regions of the UK and analysis of the impacts on electricity networks of very large individual wind farms. — Measures of reliability appropriate to intermittent sources. In particular the merits of, and options for, going beyond ‘loss of load probability’ (LOLP) in characterising the reliability of an electricity system at high levels of intermittent generation. LOLP measures the likelihood of a capacity shortfall rather than its severity. — Using these improved measures of reliability, there is a need for on-going monitoring of the British electricity market to assess how actual market response (i.e. decisions to invest in new generation or maintain existing generation in- service) compare to those that would be consistent with the improved reliability measures. — The definition of an agreed convention for reporting the costs associated with maintaining system reliability. Further work on the development of methodologies for assessing the system cost implications of new generating technologies (intermittent or otherwise), in terms of the impacts on the utilisation of incumbent generation. The extent to which intervention may be needed to ensure that adequate investment in appropriate thermal plant to maintain reliability is delivered, and the policy options available to do so. The implications of different combinations of thermal plant on the costs and impacts of integrating renewable energy in the short to medium term. In particular, the relative impacts of different sizes and types of thermal generation, and of inflexible versus flexible plant, on efficiency of system operation and integration of wind and other renewables. Options for managing the additional power fluctuations on the system due to intermittency — including new supply technologies, the role of load management, energy storage etc. Opportunities and challenges for re-optimisation of the electricity system in the long term to cope with intermittent generation, including research on much higher penetrations of renewable sources than the relatively modest levels considered in this report. References Auer, 2004, Modelling system operation cost and grid extension cost for different wind penetrations based on GreenNet, IEA Workshop on Wind Integration, Paris, Available at http://www.iea.org/textbase/work/2004/nea/auer.pdf Bach, 2004, Costs of wind power Integration into Electricity Grids: Integration of Wind Power into Electricity Grids Economic and Reliability Impacts, IEA Workshop on Wind Integration, Paris, Available at http://www.iea.org/textbase/work/2004/nea/bach.pdf BWEA, 2005, Blowing Away the Myths,A critique of the Renewable Energy Foundation’s report: Reduction in carbon dioxide emissions: estimating the potential contribution from wind power, Published by the British Wind Energy Association, London, Available at http://www.bwea.com/pdf/ref_three.pdf Commission of the European Union, 1992a,Wind Power Penetration Study-The case of the Netherlands, CEC, Brussels, Luxembourg, CEC ref. EUR 14246 EN, pp 46 Commission of the European Union, 1992b, Wind Power Penetration Study-The case of Denmark, CEC, Brussels, Luxembourg, CEC ref. EUR 14248 EN, pp 124 Commission of the European Union, 1992c,Wind Power Penetration Study-The case of Germany, CEC, Brussels, Luxembourg, CEC ref. EUR 14249 EN, pp 68 Commission of the European Union, 1992d,Wind Power Penetration Study-The case of Greece, CEC, Brussels, Luxembourg, CEC ref. EUR 14252 EN, pp 71 Commission of the European Union, 1992e,Wind Power Penetration Study-The case of Italy, CEC, Brussels, Luxembourg, CEC ref. EUR 14244 EN, pp 73 Commission of the European Union, 1992f,Wind Power Penetration Study-The case of Portugal, CEC, Brussels, Luxembourg, CEC ref. EUR 14247 EN, pp 104 Commission of the European Union, 1992g,Wind Power Penetration Study-The case of Spain, CEC, Brussels, Luxembourg, CEC ref. EUR 14251 EN, pp 81 Dale, 2002, Neta and wind, EPSRC “Blowing” workshop, UMIST, Manchester 2002. Available at http://www.ee.qub.ac/blowing/activity/UMIST/WS3_Lewis_Dale.pdf Dale et al, 2003,A shift to wind is not unfeasible (Total Cost Estimates for Large-scale Wind Scenarios in UK), Power UK, Issue 109, pp 17-25 Danish Energy Ministry, 1983,Vindkraft i Elsystemet (Denmark wind study), Published by the Danish Energy Ministry, Copenhagen EEV 83-02 DENA Project Steering Group, 2005, Planning of the Grid Integration of Wind Energy in Germany Onshore and Offshore up to the Year 2020 (DENA Grid study), Deutsche Energie-Agentur, Berlin, Available at hetp://www.wind-energie.de/fileadmin/dokumente/Themen_A-Z/Netzausbau/stud_summary-dena_grid.pdf Denny and O'Malley, 2005, Wind Generation, Power System Operation and Emissions Reduction, IEEE Transactions on Power Systems,Volume 21, Issue |, pp 341-347 Doherty et al, 2004a,Wind Penetration Studies on the Island of Ireland, Wind engineering, Volume 28, Issue I, Doherty and O'Malley, 2005,A New Approach to Quantify Reserve Demand in Systems With Significant Installed Wind Capacity, IEEE Transactions on Power Systems 2005,Volume 20, Issue 2, pp 587-595 DTI, 2003, Energy White Paper, DTI, London Published by the Stationary Office, London, ISBN 010157612-9 , Available at http://www.dti.gov.uk/energy/whitepaper/ourenergyfuture.pdf E.ON Netz, 2004,Wind report 2004, Published by EON Netz GMBH, Bayreuth, Germany E.ON Netz, 2005,Wind Report 2005, Published by EON Netz GMBH, Bayreuth, Germany Ernst & Young, 2005, UK's Future Generation Mix: The Coincidence of Corporate and Government Objectives?, Available at http://www.ey.com/global/download.nsf/UK/uk_generation_mix/$file/uks_future_generation_mix.pdf Fabbri et al, 2005,Assessment of the Cost Associated With Wind Generation Prediction Errors in a Liberalized Electricity Market, IEEE Transactions on Power Systems, Volume 20, Issue 3, pp 1440-1446 Fells, 2004, Submission to House of Lords Committee on Practicalities of Developing Renewable Energy. Evidence, summary available at http://www.parliament.the-stationery- office.co.uk/pa/ld200304/Idselect/Idsctech/ | 26/126we25.htm., Full HoL report: HoL Science and Technology Committee, 4th Report: Renewable Energy: Practicalities: Report Volume |; Evidence Volume II HL Paper | 26, I- I 15.7.04. Foulkes, 2003, Press Release - ‘ICE delight at wind investment, but not in isolation’. Press release from the Institute of Civil Engineers, Available at http://www.ice.org.uk/news_events/press_archive.asp General Electric and Marsh, 1979, Requirement assessment of wind power plants in electric utility systems, EPRI ER-978-SY, Volumes |-3, January 1979, Electric Power Research Institute, California, USA Giebel, 2000, The Capacity Credit of Wind Energy in Europe, Estimated from Reanalysis Data, Risoe National Laboratory working paper, 200la. Giebel, 2005,Wind power has a capacity credit, Gregor Giebel, Risoe National Laboratory, Denmark, Available at hetp://ejournal.windeng.net/3/01/GGiebel-CapCredLit_WindEngEJournal_2005_right_links.pdf Grubb, 1991, The integration of renewable electricity sources, Energy Policy Sept 1991,Volume 19, Issue 7, pp 670-688 Hirst, 2001, Interactions of wind farms with bulk-power operations and markets, Project for Sustainable FERC Energy Policy ,Virginia, USA. Available at http://cwec.ucdavis.edu/rpsintegration/library/Wind%20farms%20and%20bulkpower%20interactions%20Sep0 | % 20Hirst.pdf Hirst, 2002, Integrating wind energy with the BPA power system: Preliminary study, Bonneville Power Administration, Oregon, USA. Prepared for Power Business Line, Bonneville Power Administration by Consulting in Electric-Industry Restructuring , Oak Ridge, Tennessee, USA . Available at http://www.bpa.gov/power/pgc/wind/Wind_Integration_Study_09-2002.pdf Holt et al, 1990, Assessment of the impact of wind energy on the CEGB system, CEC, Brussels, Luxembourg, CEC contract ref. EN3W. 0058. UK (H) Holttinen, 2004, The impact of large scale wind power production on the Nordic electricity system, Licentiate Hudson et al, 2001, The impact of wind generation on system regulation requirements, American Wind Energy Association Windpower 200! conference proceedings, Washington, DC llex and Strbac, 2002, Quantifying the system costs of additional renewables in 2020, Ilex for DTI, London, Available at http://www.dti.gov.uk/energy/developep/080scar_report_v2_0.pdf llex et al, 2004, Operating Reserve Requirements as Wind Power Penetration Increases in the Irish Electricity System, Sustainable energy Ireland, Available at http://www.sei.ie/uploadedfiles/InfoCentre/llex Martin and Carlin, 1983,Wind-load correlation and estimates of the capacity credit of wind power:An empirical investigation, Wind engineering, Volume 7, Issue 2, pp 79-84 Milborrow, 2001, Penalties for intermittent sources of energy ,Working Paper for PIU Energy Review, Available at http://www.pm.gov.uk/files/pdf/Milborrow.pdf Milborrow, 2004, Perspectives from Abroad: Assimilation of wind energy into the Irish electricity network, Sustainable Energy Ireland, Dublin Milborrow, 2005, German report skews picture of wind on the grid, Windstats Newsletter, Volume 18, Issue I, pp |-2 Milligan, 2001,A Chronological Reliability Model to Assess Operating Reserve Allocation to Wind Power Plants, National Renewable Energy Laboratory, The 200! European Wind Energy Conference, Available at http://www.nrel.gov/docs/fy0 I losti/30490.pdf Mott MacDonald, 2003, The Carbon Trust & DTI Renewables Network Impact Study Annex 4: Intermittency Literature Survey & Roadmap, The Carbon Trust & DTI, London, Available at http://www.thecarbontrust.co.uk/carbontrust/about/publications/Annex4.pdf Pedersen et al, 2002, Present and future integration of large-scale wind power into Eltra’s power system, Eltra, Denmark, Available at http://www.eltra.dk/media/showMedium.asp?14556_LCID 1033 Royal Academy of Engineering and PB Power, 2004, The Cost of Generating Electricity, The Royal Academy of Engineering, London, ISBN |-903496-11-X Seck, 2003, GRE wind integration study, Great River Energy, UWIG technical wind workshop proceedings, Available at http://www.uwig.org/seattlefiles/seck.pdf Sinden, 2005,Wind Power and the UK Resource, Environmental Change Institute, University of Oxford, for DTI, London. Available at hetp://www.eci.ox.ac.uk/renewables/UKWind-Report.pdf Sveca and Soder, 2003,Wind power integration in power systems with bottleneck problems, Volume 2, pp 7, 2003 IEEE Power Tech Conference Proceedings, Bologna Watson, 2001, Large scale integration of wind power in an island utility-an assessment of the likely variability of wind power production in lreland,Volume 4, pp 6, 2001 IEEE Power Tech Conference Proceedings, Porto, Portugal Watson et al, 1994,Application of Wind speed forecasting to the integration of wind energy into a large scale power system, Generation, Transmission and Distribution, IEE Proceedings, Volume 141, Issue 4, pp 357-362 Annex I: Project team, expert group and contributors Project team The team was drawn from the EPSRC-funded Supergen Future Network Technologies Consortium, operating under the management of ICEPT at Imperial College, who run the TPA function of the UKERC. The team is as follows: Director: Jim Skea, UKERC Research Director Project Manager: Robert Gross, TPA Manager Principal Contributors: Dennis Anderson, Tim Green, Matt Leach Researchers: Philip Heptonstall, Sree Payyala Expert group The expert group was chosen for its combination of economic, energy policy and engineering expertise and to provide a diverse perspective. It met three times during the course of the project, providing input to the initial framing of the issues, literature search, synthesis and drafting. Several members of the group made additional contributions in the form of reference provision, invited submissions on particular issues and through bilateral interviews on specific areas of expertise. Nick Hartley, Oxera - Expert Group Chair Lewis Dale, National Grid Plc Michael Laughton, QMW College & The Royal Academy of Engineering David Milborrow, independent consultant Mark O'Malley, University College Dublin Phil Ruffles, The Royal Academy of Engineering Goran Strbac, University of Manchester (since joined Imperial College London) Peer reviewers Hannele Holttinen, VTT Finland Michael Milligan, National Renewable Energy Laboratory, USA Stakeholder workshop attendees Adrian Bull BNFL Richard Ford BWEA Michael Grubb Cambridge/Carbon Trust/Imperial College Karsten Neuhoff Cambridge University David Vincent Carbon Trust Andrew Henderson Ceasa, Spain Simon Watson CREST, Loughborough University Fred Starr DG-JRC Institute of Energy, NL Janusz Bialek Edinburgh University Nick Eyre EST Doug Parr Greenpeace Tim Green Imperial College Robert Gross Hugh Sharman’ David Milborrow Antony Price Ronnie Belmans (sent sub.) Lewis Dale (sent sub.) Ben Willis Steven Argent Bob Everett Nick Hartley Graham Sinden’* Nick Mabey Michael Laughton Richard Plozchek Gaynor Hartnell Walt Patterson Alan Laird Chris Bronsdon Mark Barrett Jim McDonald (sent sub.) Gordon MacKerron Steve Sorrell Mark O’Malley Jim Skea Goran Strbac David Andrews Imperial College/UKERC Incoteco Independent Institute of Civil Engineers Katholieke Universiteit Leuven, Belgium National Grid Npower Ofgem Ou Oxera Oxford University Prime Minister's Strategy Unit QMW College London RAEng Renewable Power Association RIIA Scottish Power Scottish, Executive, Estrata Ltd SENCO Strathclyde University Sussex Energy Group Sussex Energy Group/UKERC UCD, Ireland UKERC University of Manchester Wessex Water Annex 2: Costs of maintaining system reliability Two distinct strands of thought can be found in the literature on how to conceptualise the costs associated with any additional capacity required to maintain reliability when intermittent generators are added to an electricity network. The first does not explicitly define a ‘capacity cost’ rather it assesses the overall change in system costs that arises from additional capacity. The second includes an explicit ‘capacity cost’, which can be estimated provided we know or make an assumption about the nature of the plant that provides ‘back up’. In a working paper that accompanies this report a simple algebraic exposition is developed of both configurations which allows both techniques to be reconciled”. We provide a short description here: |. Total system cost approach. This approach compares a system with intermittent stations with an equivalent (same energy output, same reliability) system without such generation in place. On this view the cost of accommodating the lower capacity credit of intermittent stations is manifest through a depression of the load factors of the conventional plant on the system. Whilst more plant (intermittent plus conventional) is required than would be the case in the absence of intermittent stations, this approach does not attempt to directly attribute ‘capacity reserves’ due to intermittent stations (Dale et al 2003; Milborrow 2001). This approach is fully consistent with the systematic approach explained earlier in Ch. 2,and provides an estimate of the total cost of intermittent generators without being drawn into any controversy about the nature or need for any ‘reliability back up’ plant and the attribution of costs to particular aspects of intermittency. The approach derives the total change in system costs which result from replacing a proportion of thermal generating plant (e.g. CCGT) with intermittent generation (e.g. wind). It can be expressed in the simplest terms as follows: nge i m = f buildin: ing i itten - fuel sav wind - avoided fixed cost of thermal plant displaced by ~ fit of | , ; The procedure is: i. Start with the fixed and variable costs of the intermittent generating plant ii. Add system balancing costs, and any efficiency losses caused by intermittency ii. Subtract the thermal generation variable costs avoided (primarily fuel cost savings) iv. Subtract the fixed costs avoided due to being able to retire” some of the thermal plant (this is the benefit of the capacity credit of the wind) v. The remainder is the change in system cost The main limitation of the approach is that it produces a figure for the change in total system costs that includes but does not specifically identify the costs attributable to the lower capacity credit of intermittent compared to conventional stations. In other words, it does not explicitly identify the ‘capacity deficit’ cost. An alternative approach does attempt to derive this cost: 2. Capacity reserve approach. This approach conceptualises the impact of the lower capacity credit in the form of additional ‘capacity reserve’ put in place to ensure reliability. Using this approach, costs are assessed by costing the provision of ‘back up’ or ‘capacity reserve’ sufficient to close any gap between the capacity credit of intermittent stations and that of conventional generation that would provide the same amount of energy. This approach may be expressed in the most simple terms as follows: hange ii = ilding ani and variable cost of energy-equivalent CCGT” i. Start with the fixed and variable costs of the wind generating plant ii. Add system balancing costs, and any efficiency losses caused by intermittency iii, Add the capacity cost (this is the cost that will arise if the capacity credit of wind is lower than its capacity factor) iv. Subtract the fixed and variable costs of energy- equivalent CCGT generation v. = Change in system cost 7UKERC Working paper available at http://www.ukerc.ac.uk/content/view/124/105/ To use precise economic terms; the long run marginal costs saved by non-replacement of existing capital stock. "This is the thermal plant that would orovide the same amount of enerev as the wind plant at minimum cast. As an This approach may give rise to controversy because line (iii) may be derived using a range of methods and assumptions about the nature and amount of ‘back up’ that is needed. This is because cost estimates provided will vary according to assumptions about the nature of the plant that provides ‘back up’. Also, in the absence of a central planner, it is not clear by what means such plant is provided. Different assumptions are found in the literature, ranging from, for example, the capital and operating costs of new gas-fired peaking plant, projected future costs of storage devices, or the maintenance and operating costs of retaining old power stations that would otherwise be retired. (Ilex and Strbac 2002; Milborrow 2001; Royal Academy of Engineering and PB Power 2004). 3. Reconciliation. In principle both approaches should arrive at the same change in system costs. Therefore, a simple identity can be derived that can be rearranged in order to allow the derivation of the capacity credit related cost of intermittency. Algebraic derivation of this term is provided in a working paper that accompanies this report, in which it is shown that the change in variable costs cancel. Simplified, this term is as follows: i = fi f energy-equivalent thermal plant (e.g. CCGT) - avoided fixed cost of thermal plant (e.g. CCGT) ispl: i it of wi The benefit of this approach is that it allows the capacity credit related costs associated with adding intermittent plant to the system to be made explicit in a way that is consistent with systemic principles, without making any judgement about the nature of any ‘back up’. Instead, all that is required is determination of the least cost energy equivalent comparator, i.e. the thermal plant that would supply the same energy in the absence of intermittent generation (normally assumed to be CCGT). In section 3.3.6 we used this simplified term to demonstrate the effect of different capacity credit values on the capacity cost (i.e. the cost of maintaining reliability), whilst keeping all other system characteristics unchanged. These characteristics were chosen to be representative of a future British electricity network and expected capacity credit for wind power. However, capacity credit and capacity factor are related variables and relatively low capacity credit values tend to be associated with relatively low capacity factors. We therefore also explored the sensitivity of the cost of maintaining reliability to a range of capacity credit and capacity factor values, the results of which are shown in table A2.1. As in section 3.3.6, the system characteristics are derived from (Dale et al 2003). The only changes in each calculation are the wind capacity factor and wind capacity credit. The lower wind capacity factors require proportionately more installed wind capacity to deliver the same amount of energy. In isolation, this has no impact on the reliability cost. This is illustrated by the reliability cost being the same for each capacity factor where the capacity credit is the same fraction of capacity factor (e.g. capacity factor/capacity credit combinations of 20%/10%, 30%/15%, 40%/20%). For any given capacity factor the reliability cost reduces as the wind capacity credit increases. It is the size of capacity credit relative to capacity factor which determines the cost of maintaining reliability - low capacity credit relative to capacity factor gives rise to higher reliability costs. Table A2.1 The sensitivity of reliability cost to capacity factor and capacity credit Annex 3: Full list of included documents on weak networks DeCarolis J F The Economics and Environmental 2004 | 169} Carnegie Mellon Impacts of Large-Scale Wind Power in a University, Pittsburgh, Carbon Constrained World Pennsylvania DeCarolis J F The Costs of Wind's Variability: ls There a | 2005 | 197 | http://www.ucalgary.ca/ Keith DW Threshold? ~keith/papers/72. Decarolis.2005.Threshold. e.pdf Dena Project Planning of the Grid Integration of Wind 2005 | 74 '| Deutsche Energie-Agentur, Steering Group Energy in Germany Onshore and Offshore Berlin TGEAD up to the Year 2020(Dena Grid study) Denny E, O'Malley M_| Wind Generation, Power System Operation} 2005 | 181 | IEEE TRANSACTIONS and Emissions Reduction ON POWER SYSTEMS Deutsches Wind Turbine Grid Connection and 2001 | 63 | http://europa.eu.incomm/ Windenergie-Institut | Interaction energy/res/sectors/doc/ GmbH Germany, wind_energy/maxi Tech-wise A/S brochure_final_version.pdf Denmark, DM Energy United Kingdom Doherty R, Bryans L, | Wind Penetration Studies on the Island 2004 | 178} Wind Engineering Gardner P, of Ireland : O'Malley M ; Doherty R, Denny E, | System operation with a significant wind 2004 | 22 | Power Engineering Society O'Malley M power penetration General Meeting, 2004. IEEE Doherty R, Quantifying reserve demands due to 2003 | 14 | http://ieeexplore.ieee. O'Malley M increasing wind power penetration org/iel5/9135/28975/0130 4288. pdf?tp=&arnumber=| 304288&isnumber=28975 Doherty R, A New Approach to Quantify Reserve 2005| 6 | IEEE Transactions on O'Malley M Demand in Systems With Significant Power Systems Installed Wind Capacity : Dragoon K Assessing Wind Integration Costs with 2003 | 42 | WINDPOWER 2003, (PacifiCorp), Dispatch Models: A Case Study of Austin, Texas PacifiCorp E.ON- Net Z Wind Report 2005 2005 | 246 | EON Netz, Germany E.ON- Net Z Wind report 2004 2004 | 57 | EON Netz, Germany EC Commission Wind Power Penetration Study, 1992 | 249| CEC, Brussels, The Case of Italy Luxembourg - see ref. 245 EC Commission Wind Power Penetration Study, 1992 | 243 CEC, Brussels, ; The Case of the Netherlands Luxembourg - see ref. 245 EC Commission Wind Power Penetration Study, 1992 | 242| CEC, Brussels, The Case of Greece Luxembourg - see ref. 245 EC Commission Wind Power Penetration Study, 1992 | 248} CEC, Brussels, The Case of Portugal Luxembourg - see ref. 245 EC Commission Wind Power Penetration Study, 1992 | 244} CEC, Brussels, The Case of Spain Luxembourg - see ref. 245 EC Commission Wind Power Penetration Study, 1992 | 240} CEC, Brussels, A The Case of Denmark Luxembourg - see ref. 245 EC Commission Wind Power Penetration Study, 1992 | 250} CEC, Brussels © The Case of Germany Luxembourg - see ref. 245 Econnect Ltd Wind turbines and load management 1996 | 226} Document sourced from British Library sor nent ES ERRMSG cA TRC 8 eC aie Lecco ca sacroammtaci iE mOREARLT mC ELTRA Transmission | Specifications for connecting wind farms 2000 | 120] hetp://www.eltra.dk/media/ System Planning to the transmission network showMedium.asp?12321_ LCID 1033 Energy Link and Wind Energy Integration in New Zealand | 2005 | 109] Ministry of Economic MWH NZ for Development Energy Ministry of Economic Efficiency and Development Energy Conservation Authority Efficiency and Conservation Authority EnerNex Xcel Energy and the Minnesota 2004 | 45 | http://www.state.mn.us/mn Corporation, Wind Department of Commerce Wind /externalDocs/Commerce/ Logics Integration Study - Final Report Wind_Integration_Study _092804022437_Wind IntegrationStudyFinal.pdf Environmental Variability of UK marine resources 2005 | 205 | Carbon Trust Change Institute (commissioned by Carbon Trust) ; University of Oxford ‘ Environmental The Practicalities of Developing Renewable | 2003 | 91 | Environmental Change Change Institute Energy Stand-by Capacity and Institute , Oxford University of Oxford | Intermittency Submission to The Science University and Technology Select Committee of the House of Lords Ernst B, Wan Y, Short term power fluctuation of wind 2003 | 106] California Wind Energy Kirby B turbines: Looking at data from the German __| Collaborative 250MW measurement program from the j ancillary services viewpoint ESD limited Maximising the commercial value of wind | 2000} 201 | http://test.netgates.co.uk/ energy through wind forecasting nre/pdf/W1100555.pdf — Fabbri A, Assessment of the Cost Associated With 2005 | 206} IEEE Transactions on GomezSanRomanT, | Wind Generation Prediction Errors in a Power Systems RivierAbbad J, Liberalized Electricity Market MendezQuezada VH Fairbairn R J Electricity network limitations on large 1999 | 200 | http://test.netgates.co.uk/ scale deployment of wind energy nre/pdf/rep529.pdf Farmer E D, Newman | Economic and operational implications of a | 1980 | 219] IEE ProcA V G,Ashmole PH complex of wind-driven power generators on a power system Fells Submission to House of Lords Committee | 2004} 252] Fells Associates on Practicalities of Developing Renewable ; Energy Ford R, Milborrow D_| Integrating Renewables 2005 | 185] Supplied by author Foulkes Press Release - ‘ICE delight at wind 2003 | 254} Institution of Civil investment, but not in isolation’ Engineers Furong Li, Haibin Wan | Dead calm (requirement for wind power 2005} 8 | Power Engineer [see also generation storage devices) Power Engineering Journal] Future Energy The value of energy storage within the UK | 2004 | 131 | http://www.dti.gov.uk/ Solutions electricity network renewables/publications/ pdfs/kel00246. pdf Gardner G E, System integration of wind power 1982 | 223 | EC Contractors’ meeting, Thorpe A generation in Great Britain Brussels Garrad Hassan and The impacts of increased levels of wind 2003 | 133} Commission for Energy Partners Limited penetration on the electricity systems of Regulation the Republic of Ireland and Northern treland: Final report General Electric, Requirement assessment of wind power 1979 | 217 Electric Power Research Marsh W D plants in utility systems Institute, California, USA Giebel G Wind power has a capacity credit 2005 | 245 | hetp://www.drgiebel.de/ thesis.htm? Giebel G The Capacity Credit of Wind Energy in 2000} 121] EXPO 2000, Hannover Europe, Estimated from Reanalysis Data Gilman B, Cheng M, | The value of wind forecasting to Southern | 2001 | 231] Proc AWEA Conference Isaac J, Zack J, Bailey B,| California Edison : Brower M Gjengedal T Integration of wind power and the impact | 2003} 31 | Power Engineering, 2003 on power system operation Large Engineering Systems Conference on Gouveia EM, Operational reserve of a power system 2004 | 19 | 2004 International Matos MA with a large amount of wind Conference on power Probabilistic Methods Applied to Power Systems Grubb M J The integration of renewable electricity 1991 | 204 | Energy Policy sources Grubb M J Value of variable sources on power systems} 1991 | 68 | Generation, Transmission and Distribution [see also IEE Proceedings- Generation, Transmission and Distribution], |EE Proceedings C Grubb M J On capacity credits and wind - load 1988 | 213} Proceedings of 10th correlations in Britain British wind energy association conference, London Halliday J A Analysis of wind speed data recorded at 14| 1984 | 208] Wind Engineering widely dispersed U.K meteorological stations : Halliday J A, Lipman Studies of wind energy integration for the | 1983 | 222 | Wind workshop VI, N H, Bossanyi EA, UK national electricity grid American Wind Energy Musgrove P J Association, Minneapolis Hansen AD, Becho J, | Dynamic modelling of wind farm grid 2002 | 64 | Wind Engineering Blaabjerg F, interaction Sorensen P Hirst E Integrating wind energy with the BPA 2002 | 193} Bonneville Power power system: Preliminary study Administration, Oregon, USA Hirst E Interactions of wind farms with bulk 2001 | 199} Project for Sustainable power operations and markets FERC Energy Policy, Virginia, USA Holt J S, Milborrow Assessment of the impact of wind energy | 1990} 160} Brussels, Luxembourg D,Thorpe A on the CEGB system Holttinen H The impact of large scale wind power 2004} 67 | Helsinki University of production on the Nordic electricity system Technology, Dept. of Engineering, Physics and Mathematics Holttinen H, The effect of large scale wind power ona | 2003} 71 | Proceedings of the 4th thermal system operation Pedersen J International Workshop on Large-Scale Integration of Wind Power’ and Transmission Networks far Offshore Makarov Y, Hawkins D | Quantifying the impact of wind energy on | 2004 | 234} Proc Global Wind power power system operating reserves Conf, Chicago Martin B, Carlin J Wind-load correlation and estimates of the | 1983 | 238 | Wind Engineering capacity credit of wind power:An empirical investigation Martin B, The economics of large-scale wind power | 1983 | 237) Energy Policy Diesendorf M in the UK,a model of an optimally mixed CEGB electricity grid Meier RC, Macklis SL | Interfacing wind energy conversion 1982} 12 | Energy equipment with utility systems Milborrow D Capacity credits - Clarifying the issues 1996 | 212} Proceedings of |8th annual conference, British wind energy association, London Milborrow D Assimilation of wind energy into the Irish | 2004} 89 | Sustainable energy Ireland electricity network Milborrow D Wind and gas = the gap narrows 2005 | 236} Power UK Milborrow D Impacts of wind on electricity systems with | 2004 | 172| Canada Wind Energy particular reference to Alberta Association, CANWEA Milborrow D The real costs and problems of integrating | 2001 | 73 | Blowing work shop, Belfast| wind Milborrow D Penalties for intermittent sources of energy) 2001 | 132} Working Paper for PIU Energy Review Milborrow D False alarm 2005 | 207 | Power Engineer Milborrow D Integrating wind: Ofgem discussion day 2004 | 66 | www.ofgem.gov.ul/temp/ ofgem/cache/cmsattach/77 49_DavidMilborrow_mayO 4.pdf Milborrow D German report skews picture of wind on | 2005 | 235 | Windstats Newsletter the grid Milligan M Modelling Utility-Scale Wind Power Plants | 2002 | 141 | http://www.nrel.gov/docs/ Part 2: Capacity Credit fy02osti/29701I pdf Milligan M Measuring Wind Plant Capacity Value 2002 | 139} http://www.dinet.vt.edu/ repository/previewReposit ory/EE000000/EE008000/E £00800 1/DISK I/DLNET- 12-30-2002-01 1 1/ resources/20493 pdf Milligan M A Chronological Reliability Model to Assess} 2001 | 186 | National Renewable Operating Reserve Allocation to Wind Energy Laboratory, The Power Plants 2001 European Wind Energy Conference Milligan M Factors Relevant to Incorporating Wind 2001 | 137| Proceedings of the 24th Power Plants into the Generating Mix in International Conference Restructured Electricity Markets of the International Association for Energy Economics, Houston, TX Milligan M Wind Power Plants and System Operation | 2003 | 191 | WINDPOWER 2003 in the Hourly Time Domain Austin, Texas Milligan M, Miller A, Estimating the Economic Value of Wind 1995 | 138} Proceedings of Wind Chapman F Forecasting to Utilities power’95, Washington, D.C. i Milligan M, Parsons B_| A Comparison and Case Study of Capacity | 1997} 162] Solar ‘97,Washington, DC Credit Algorithms for Intermittent Generators Moretti PM. lones BW! Analysis method for non-schedulable 1982113. | Solar Energy a nena nnenemnl 5 Sis SSSR a Sac Raa TOA bs SSC MG Pa EAA ad RN A RR AN NEE Mott MacDonald The Carbon Trust & DTI Renewables 2003} 52 | The Carbon Trust & DTI Network Impact Study Annex 5: Gri Code Compliance Nabe CA Capacity credits for wind energy in 2000 | I11 | http://www.energiewirts deregulated electricity markets: chaft.tuberlin.de/mitarbeit Limitations and extensions er/wind2|-paper-V7-5- nabe.pdf Namoyvicz C Updates to Wind Intermittency Algorithms | 2004 | 123} Renewable Energy for AEO2004 Modeller’s Summit NEMMCO Intermittent Generation in the National 2003 | 96 | National Electricity Market! Electricity Market Prepared by: Market Management Company Development Version: 1.0 Limited National Grid Submission to Energy Policy Review, 200! | 230] Supplied by Lewis Dale Company Appendix 2 Pantaleo A, Pellerano | Technical issues for wind energy integration} 2003 | 147} Wind Engineering A,Trovato M in power systems: Projections in Italy Pedersen J,Mortensen | Present and future integration of large scale} 2002 | 129] Eltra, Denmark P Eriksen P wind power into Eltra’s power system : Peters L, Noteboom R| Integrating Wind Resources into the NW | 2003} 44 | Discussion Paper presented! Power System at the TBL Business | Practices Forum Pierik J, Morren J, de | Dynamic models of wind farms for grid- 2004} 65 | Nordic Wind Power Haan S, van EngelenT,} integration studies: Case study results Conference, 1-2 March Wiggelinkhuizen E, 2004, Chalmers University Bozelie J, Energy re- Of Technology search Centre of the Netherlands (ECN), Delft University of Technology (TUD) Resch G, Swider D, Pushing a least cost integration of green 2003} 88 | European Communities, Twidell J, Nielsen L, electricity into the European grid. Dynamics the Sth framework Huber C, Stadler M, | and basic interactions of RES-E with the programme of the Auer H grid, switchable loads and storage European Commission supported by DG TREN Rockingham A A probabilistic simulation model for the 1979 | 211} Proceedings of Ist BWEA calculation of the value of wind energy to wind energy workshop, electric utilities Cranfield Rockingham A P System economic theory for WECS 1980 | 220] 2nd British Wind Energy Association Workshop, Cranfield Royal Academy of Response to the House of Lords Science and | 2003 | 59 | The Royal Academy of Engineering Technology select committee inquiry into the Engineering, London practicalities of developing renewable energy Royal Academy of The Costs of Generating Electricity 2004 | 239] The Royal Academy of Engineering, PB Power Engineering, London Salman SK,Teo AL} Windmill modelling consideration and factors | 2003 | 15 | IEEE Transactions on influencing the stability of a grid-connected Power Systems wind power based embedded generator Seck T GRE wind integration study 2003 | 187| Great River Energy, UWIG | technical wind workshop Sinden G Wind Power and the UK Resource 2005 | 247] Environmental Change Institute, University of Oxford System Operating Costs Summary and Perspective on Work Done to Date Smith ] C (Utility Wind) Wind power impacts on electric power 2004} 48 | American Wind Energy Interest Group), system operating costs: summary and Association Global DeMeoE A (Renewable| perspective on work to date WindPower Conference, Energy Consulting), Chicago, Illinois Parsons B(NREL), Milligan M(NREL) Soder L Reserve margin planning in a wind-hydro- | 1993) 2 | IEEE Transactions on thermal power system Power Systems Soder L Imbalance management and reserve 2002 } 155} Wind power and the requirements impact on power systems IEEE-Cigre workshop, Oslo Soder L Simulation of wind speed forecast errors | 2004} 23 | 2004 International for operation planning of multiarea power Conference on Probabilistic systems Methods Applied to Power| Systems South Western Interaction of Delabole wind farm and 1994 | 225 | Document sourced from Electricity ple SouthWestern Electricity’s Distribution British Library system Strbac G, Jenkins N Network security of the future UK 2001 | 196} MANCHESTER CENTRE electricity system (Report to PIU) FOR ELECTRICAL ENERGY Department of Electrical Engineering &Electronics PO Box 88, Manchester, M60 1QD Sustainable Wind Power in the UK:A guide to the key | 2005] 92 | Sustainable development Development issues surrounding onshore wind power commission Commission development in the UK Sveca J, Soder L Wind power integration in power systems | 2003} 18 | 2003 IEEE Power Tech with bottleneck problems Conference Proceedings, Bologna Swift-Hook DT Firm power from the wind 1987 | 224 | Proceedings of British Wind Energy Association Conference, Edinburgh The Large-Scale Wind | Integration of Large-Scale Wind Generation| 2004 | 98 | http://systemcontroller.trans Integration Working end.com.au/public.asp Group to the NEM Entry Coordination Group Thorpe A A computer model for the evaluation of 1987 | 209} Wind Engineering plant and system operating regimes Union for the Co- Integrating wind power in the European 2004 | 145 | http://www.ucte.org/pdf/ ordination of power systems - prerequisites for Publications/2004/UCTE- Transmission of successful and organic growth position-on-wind- Electricity (UCTE) power. pdf#search="Integ rating%20wind%20power %20in%20the%20European %20power%20systems Usaola J, Ravelo O, Benefits for Wind Energy in Electricity 2004 | 184} Wind Engineering Gonzalez G, Soto F Markets from Using Short Term Wind Davila MC, Diaz- Power Prediction Tools; a Simulation Study Guerra B UWIG Wind Power Impacts on Electric-Power- 2003 | 149} UWIG Annex 4: Full list of excluded documents Author Title Year | Ref.| Source Akhmatov V Analysis of dynamic behaviour of electric | 2003 | 130} http://server.oersted.dtu. power systems with large amount of dk/eltek/res/phd/00-05/ wind power va-thesis.pdf Ancona DF, Krau S, Operational Constraints and Economic 2003 | 101) European Wind Energy Lafrance G, Benefits of Wind-Hydro Hybrid Systems Conference, Madrid, Spain Bezrukikh P Analysis of Systems in the U.S./Canada and Russia Bazilian M, Denny E, | Challenges of Increased Wind Energy 2004 | 183 | Wind engineering O'Malley M Penetration in Ireland Billinton R, Karki R Reliability/cost implications of utilizing 2003 | 40 | Reliability Engineering & photovoltaics in small isolated power System Safety, Univ systems Saskatchewan, Dept Elect Engn, Power Syst Res Grp, Saskatoon, SK S7N 5A9, Canada Blair N, Short W, Reduced Form of Detailed Modelling of 2005 | 171 | WindPower 2005, Denver, Heimiller D Wind Transmission and Intermittency for Colorado Use in Other Models Brobak B, Jones L Real time data acquisition from wind farms | 2002 | 29 | Power Engineering Society in power systems Summer Meeting, 2002 IEEE Brocklehurst F A Review of the UK Onshore Wind Energy] 1997 | 210} ETSU, Harwell, Resource Oxfordshire Brown A, Ellison C, | Transmitting Wind Energy Issues and 1999 | 176 | National Wind Porter K Options in Competitive Electric Markets Coordinating Committee: NREL Burges K Dynamic modelling of wind farms in 2004 | 118] hetp://www.irish-energy. transmission networks ie/uploads/documents/ upload/publications/tech nical_paper_modelling_ wind _KBu_mar_04.pdf Castronuovo ED, Bounding active power generation of a 2004 | 25 | 2004 International Lopes JAP wind-hydro power plant Conference on Probabilistic Methods Applied to Power Systems Catunda SYC, Pessanha| Uncertainty analysis for defining a wind 2004 | 32 | Instrumentation and JEO, FonsecaNeto JV, | power density measurement system Measurement Technology structure Camelo Nj, Silva PRM Conference, 2004. IMTC 04. Proceedings of the 21st IEEE Christiansen P, Grid Connection and Remote Control for | 2002 | 180} http://www.owen.eru.rl. Jorgensen K, the Horns Rev 150 MW Offshore Wind ac.uk/workshop_4/pdfs/ SorensenA Farm in Denmark owen_Christiansen.pdf Danish wind turbine | The energy balance of modern wind 1997 | 81 | Vindmilleindustrien Vester manufacturers turbines Voldgade 106DK [552 association Copenhagen K Dragoon K, Milligan M} Assessing Wind Integration Costs with 2003 | 108] NREL http://anem« download/publicati 200 http Kariniotakis GN, Uncertainty of short-term wind power 2004 | 24 | 2004 International Pinson P forecasts a methodology for on-line Conference on Probabilistic assessment Methods Applied to Power| Systems King D Environment: Climate Change Science: Adapt, Mitigate, or Ignore? 2004} 35 | Science Kjaer C Comments on the Preliminary Draft Vision | 2003 | 190 | http://www.ewea.org/ Document of the High Level Group on documents/| 1_EWEA Hydrogen and Fuel Cells _hydrogen_response.pdf Kosoric KR, Wind changes influence on control of power | 2003 | 30 | Power Engineering Society Katancevic AR systems with high percentage of wind power General Meeting, 2003, ‘ IEEE Milborrow D Hydrogen myths and renewables reality 2003 | 233 | Windpower Monthly Milborrow D Revolutionary potential 2000 | 228 | Windpower Monthly Minnesota Wind resource analysis program 2002 2002 |° 47 | http://www.state.mn.us/ Department of mn/externalDocs/WRAP__ Commerce Report_! 10702040352 : _WRAP2002.pdf Morgan CA Offshore wind economies of scale, 2003} 61 | hetp://www.dti.gov.uk/rene (HMSNCSfDCT) engineering resource and load factors wables/policy/garrad hassanoffshorewind.pdf Mott MacDonald, Renewable Energy Industry Gap Analysis: | 2005} 114} Department of Trade and Bourton Group Summary Report Industry National Grid Evidence to House of Lords’ Select 1999 | 227| The Stationery Office Company; Committee on ‘Electricity from Renewables’. NECA Code change panel: Intermittent generation} 2004 | 97 | http://www.neca.com.au/ forecasting obligations Files%5CC_CCP_ intermittent_generation_ forecasting_obligations _Sep2004.pdf Obersteiner C, Modelling additional system operation cost | 2004 | 86 | GreenNet Dissemination Auer H due to large-scale wind generation : Workshop Prague Oxera What is the potential for commercially 2005 | 60 | http://www.oxera.com/ cmsDocuments/Reports/ DT1%20The%20potential% 20for%20commercially%20 viable%20renewable%20 generation%20technologie s%20February%202005.pdf| Park SJ, Kang BB, Yoon| A study on the stand-alone operating or 2004 | 36 | http://ieeexplore.iece JP, Cha IS, Lim JY photovoltaic/wind power hybrid generation -org/ielS/937 1/29761/ system 01355441 pdfftp= &ar number=135544 | &is number=29761 Renewable energy Wind power resource assessment: wind 2005 | 54 | RERL- MTC Community research lab UoM power on the community scale wind power fact sheet 5 Renewable energy Wind power: Interpreting your wind 2005} 55 | RERL - MTC community research lab UoM resource data wind power fact sheet 6 Royal Academy of Inquiry into the practicalities of developing | 2003 | 153} The Royal Academy of Engineering renewable energy: Memorandum submitted by The Royal Academy of Engineering Engineering, London Annex 5: Technical annex to Ch. 1: search terms and databases used Literature search Following consultations with the expert group the following databases, bibliographies, catalogues, and other sources were utilised. A preliminary list was published in the Scoping Note and Protocol. A number of relevant papers were accessed through expert recommendations, in particular older studies and important pieces of ‘grey’ literature, as well as international sources. Manual searching of key recent documents’ bibliographies — the DTI ‘SCAR’ Report and Network Impacts Study and PIU Working Papers Recommendations from the expert group and stakeholders — In particular international technical reports/consultations/case studies produced by transmission system operators and regional electricity companies. Database searches, using key words and search terms (see below). Databases included: — ‘ESTAR’, the British Library's Electronic Storage and Retrieval System — ‘SIGLE’, the system for Information on Grey Literature in Europe - citations to reports and non-conventional literature published across EU member states since 1980 — Elsevier's ‘Science Direct’ — Academic Working Paper Series available online — PhD theses available online — Engineering databases - IEEE Explore and IEE Inspec * Specific journal archives not covered in the above, in particular for older papers not available in on-line databases — Electrical engineering journals, IEE conference proceedings + Website searches using the keyword combinations (as below). Example sites: - DTI — National Grid and Ofgem — Google — IEA — Wind energy associations — US DoE — NREL Search terms Key words were determined and refined in collaboration with the expert group and stakeholders. They are listed below: Inclusion and categorisation criteria The literature was included and categorised according to relevance. It is not uncommon for systematic reviews to exclude the majority of studies found during the search period on ‘quality’ grounds — for example the exclusion of all non-empirical work”. The approach taken in this report is to include studies and comment on their quality rather than exclude large numbers of reports a priori. A categorisation matrix was developed, which captured key data from the 154 included references. The range of data captured on each reference is summarised in the table below: Annex 6: Technical annex to Ch. 2: terminology To make an accurate assessment of the costs of using intermittent sources of generation in an electricity network one must be careful to properly describe the problem and to use terms consistently. The issues are under discussion in many parts of the world and amongst many groups of people (the general public, economists, engineers and others). Here we define the terms we use, and where relevant how they relate to terms used elsewhere. Energy, Power, Average Power and Rated Power Energy is the ability to do work or is work done. It is, for instance, the work done by an electric motor or a heater. The scientific unit of energy is the joule (J), or its multiples such the megajoule (1,000,000 joules) and the gigajoule (1,000,000,000 joules). It is more normal when discussing electricity systems to measure energy in kilowatt-hours, kWh (domestic electricity tariffs are quoted in p/kWh) or megawatt- hours (wholesale electricity prices are quoted in £/MWh). Power is the rate at which energy is delivered and its scientific unit is watts (VV), and is equivalent to joules per second, J/s. When a power quantity is multiplied by a time it gives an energy quantity. So, a kilowatt-hour is 1000 watts for 3600 seconds and is therefore 3,600,000 joules or 3.6M]. It is also common to discuss the energy per annum. This is actually a power because it is an energy transfer rate (energy per unit time). This is an example of an average power using a year as the period over which the average is taken. For instance, the predicted UK electricity consumption in 2020 is 400,000 GWh per annum. If this is divided by the number of hours in a year (8760) it gives the average power (the average rate of energy delivery) which is 45.6 GW. Most items of generating plant or electrical network equipment have a maximum power capability known as their rated power. This will be determined by its maximum voltage and maximum current or perhaps by a mechanical limitation. Sometimes these limitations depend on air temperature so there can be different rated powers for summer and winter conditions. In alternating current (AC) systems we must also account for reactive power which expresses how much energy per second is stored in but it does need to be present for the proper functioning of the system. Intermittency and Variability Intermittency has become a short hand term for power sources that do not produce a constant output. In every day language the term intermittent would be interpreted as something that turns on and off. All types of power generation are intermittent in this sense. Coal or nuclear power generation plants that are designed to run at full power continuously are still subject to planned shut- downs for maintenance and unplanned shut-downs because of equipment failures. Variability is an alternative term to describe power sources such as the wind whose output is not constant and varies between zero and full power. That variation might be on any or all of the timescales of seconds, minutes, hours, days, seasons and years. The variation may be in part regular (such as tides or patterns of evening on shore winds), it may be predictable, subject to forecast errors (and dependent on weather patterns, sea and air temperatures, or other factors) or it may be random. The variability can be characterised in terms of the changes in the amount of power generation, the frequency of the changes and the rapidity with which the changes occur. System Balancing Gate Closure, Balancinh Mechanism, Balancing Systems Charges and System Frequency The supply of electricity is unlike the supply of other goods. Electricity cannot be readily stored in large amounts and so the supply system relies on exact second-by-second matching of the power generation to the power consumption. Some demand falls into a special category and can be manipulated by being reduced or moved in time. Most demand, and virtually all domestic demand, expects to be met at all times. It is the supply that is adjusted to maintain the balance between supply and demand in a process known as system balancing. There are several aspects of system balancing. In the UK system, contracts will be placed between suppliers and customers (with the electricity wholesalers buying for small customers on the basis of predicted notified to the system operator which in Great Britain is National Grid Electricity Transmission Limited. This hour-ahead point (some countries use as much as twenty-four hour ahead) is known as gate closure. At gate closure the two-sided market of suppliers and consumers ceases. (National Grid becomes the only purchaser of generation capability after gate closure and its purpose in doing so is to ensure secure operation of the system.) What actually happens when the time comes to supply the contracted power will be somewhat different to the contracted positions declared at gate closure. Generators that over or under supply will be obliged to make good the difference at the end of the half hour period by selling or buying at the system sell price or system buy price. Similar rules apply to customers who under or over consume. This is known as the balancing mechanism and the charges as balancing system charges. This resolves the contractual issues of being out-of- balance but not the technical problems. If more power is consumed than generated then all of the generators (which are synchronised such that they all spin at the same speed) will begin to slow down. Similarly, if the generated power exceeds consumption then the speed will increase. The generator speeds are related to the system frequency. Although the system is described as operating at 50 Hz, in reality it operates in a narrow range of frequency centred on 50 Hz. It is National Grid’s responsibility to maintain this frequency using “primary response" plant (defined below). This plant will increase or decrease its power output so that supply follows demand and the frequency remains in its allowed band. The cost of running the primary response plant can be recovered from the balancing charges levied on those demand or supply customers who did not exactly meet their contracted positions. It is possible that a generator or load meets its contract position by consuming the right amount of energy over the half hour period but within that period its power varied about the correct average value. Thus the contract is satisfied but the technical issue of second-by-second system balancing remains. Back-up and Reserve The term back-up power is sometimes used to describe the need for additional power to be available to cover for when intermittent or variable sources are not available. It is important that back- up is matched to the problem it is intended to cover and therefore a classification system is of flexible generation plant that can produce power at short notice is not the type of plant that it is desirable (economically or otherwise) to run for long term energy supply. There is therefore a hierarchy of measures on different timescales. Because this is a study of the UK system, we will follow the classification used by the British transmission system operator, National Grid. The preferred terminology is for reserve generation and this is split into several categories as defined in the following sections. Operating Reserve, Primary Response and Secondary Response Operating reserve is generation capability that is put in place following gate closure to ensure that differences in generation and consumption can be corrected. The task falls first to primary response. This is largely made up of generating plant that is able to run at much less than its rated power and is able to very quickly increase or decrease its power generation in response to changes in system frequency. Small differences between predicted and actual demand are presently the main factor that requires the provision of primary response. There can also be very large but infrequent factors that need primary response such as a fault at a large power station suddenly removing some generation or an unpredicted event on TV changing domestic consumption patterns. The primary response plant will respond to these large events but will not then be in a position to respond to another event unless the secondary response plant comes in to deal with the first problem and allow the primary response plant to resume its normal condition of readiness. Primary response is a mixture of measures. Some generating plant can be configured to automatically respond to changes in frequency. In addition some loads naturally respond to frequency and other loads can be disconnected (shed) according to prior agreement with the customers concerned in response to frequency changes. Secondary response is normally instructed in what actions to take by the system operator and will have been contracted ahead by the system operator. The secondary reserve might be formed of open-cycle gas-turbine power stations that can start and synchronise to the system in minutes. In the past in the UK and presently in other parts of the world, the term spinning reserve has been used to describe a generator that is spinning and ready at very short notice to contribute power to the system. Spinning reserve is one example of Standing Reserve, Contingency Reserve and Capacity Reserve To provide cover for unavailable generating plant over a period of hours requires standing reserve. This might be in the form of thermal plant (such as coal fired power stations) that are kept at operating temperature but without their steam turbines and generators running. In the past this has been known as thermal reserve but the term used here will be standing reserve. It is necessary to keep them warm because they can take several hours to heat up from cold to operating temperature. This type of reserve has to be contracted 24 hours ahead by the system operator. Such notice is termed ‘warming’ and payments are made once warming commences. Contingency reserve consists of the margin of generation over forecast demand which is required in the period from 24 hours ahead down to real time to cover against uncertainties in large power station availability and against both weather forecast and demand forecast errors. It includes generation that the system operator has contracted for but not issued a notice to warm. If a generator has to be taken out of service for a prolonged period then it is expected that there will be a reaction in the pre-gate-closure market. If the out-of-service generator would have been offered in the spot market then there will be a shortage there. If it was part of a long term contract then its owner will seek to cover that contract position by purchasing output from other generators. Other plant owners might now offer generators that would not otherwise have been offered. These plants can be described as capacity reserve. Provision of this capacity reserve is left to the market but because National Grid makes regular statements on system adequacy, the market has signals about when moth- balled plant might become needed to form capacity reserve. Reliability, LOLP, LOLE, LOEE Electricity supply systems operate with high reliability but are not perfectly reliable in that occasionally some customers are not supplied. Some interruptions of supply arise from equipment failures or storm damage in the transmission and distribution networks. Some can arise from inadequate generation capacity. There are many ways of measuring and estimating reliability and the measure used depends on circumstance. The to be shed (forced to disconnect from the system) because insufficient generation is present. This is expressed as a percentage that is the number of years per century in which load shedding will occur. LOLP does not inform us of how much load will be shed or for how long. Loss-of-load expectation, LOLE is slightly different and accounts for how much time would be spent without the load being served. Loss-of-energy expectation, LOEE accounts for how large a collection of load, in terms of its power, is not served and over what time period by measuring how much energy is not supplied (energy being the product of power and time). Capacity, Installed Capacity, Availability, Technical Availability, System Margin, Reserve Capacity and Capacity Credit The capacity of a system is the amount of the generation plant connected to the system. The installed capacity would be all of the connected generation accounted for at its rated power. However, we know that plant is not always available to generate because of planned maintenance, unplanned maintenance or unavailability of the energy source. Technical availability accounts for maintenance only and availability will include the energy source availability too. Statistical methods are needed to assess how much of the plant connected to a system is likely to be available at any time and from this the LOLP can be calculated for a given combination of plant and peak demand. A simple measure of the safety margin in a system is the system margin which is the difference between the installed capacity and the peak demand but one needs to know the type of plant (or the mix of types of plant) in question before this can be interpreted in terms of a system reliability. Capacity margins in the region of 20% peak demand have been common in the UK and because the generation mix has been dominated by thermal (gas turbine and coal) plant of similar probability of availability it has been possible to use this as a simple indication of whether the system was adequate to meet peak demand. When a new generation technology with a quite different availability probability is introduced (or substituted for existing plant) one can reassess the LOLP using statistical methods. A simple representation of the outcome of this assessment is to assign the new generation plant a capacity If the new generation is less likely to be available at peak demand than the incumbent generation then its capacity credit will be less than its installed capacity. Cut-in Speed and Cut-out Speed At low wind speeds, a wind turbine will barely turn and does not produce enough energy to cover its own internal needs for electrical control. Therefore the turbine is not used (and is held stationary) at wind speeds below the cut-in speed (typically 4 m/s). For wind speeds above the cut-in speed the power output rises with wind speed until eventually the maximum power rating of the electrical generator is reached. If the wind speed rises further (to above about |5 m/s), measures are taken to limit the generated power to the rated power. At very high wind speeds, above the cut-out speed (typically 25 m/s), this is no longer possible without endangering the wind turbine structure and the turbine is stopped. In principle the turbine could be designed for a higher cut-out speed but the expense of doing so is judged to not be justified in terms of the additional energy generated during the infrequent periods of very high wind speeds. Efficiency, Load Factor and Capacity Factor Generators take an original energy source and convert it into electrical power through one or more transformations. Even where the original energy is essentially free (such as sunlight or wind) it is important that the plant is used to maximum advantage and a high proportion of the available energy is converted to electrical form. For a wind turbine, the available energy is the kinetic energy in the air mass that passes through the swept area of the turbine blades. The efficiency is defined as the electrical energy output divided by the available kinetic energy. It has been established (and this is known as the Betz limit) that not all of the energy in the air mass can be captured (since this would require bringing that portion of air to a standstill). The theoretical limit on wind turbine efficiency is 59% and practical wind turbines achieve somewhat less than this because of aerodynamic, mechanical and electrical inefficiencies. It must then be recognised that since the wind speed is variable, the turbine produces less than rated power for some of the time and that the average power is less than the rated power. The ratio of average generated power to GB system has fallen in recent years, as a result of changes to the regulatory regime, electricity prices and gas prices. The maximum ratio of generated power to rated power is known as capacity factor. This represents the maximum number of load hours per year net of both planned and unplanned outages, independent of actual utilisation. Use of System Charges The requirement on generators (at least those above a certain size) to pay charges to cover the cost of the balancing mechanism has already been discussed. There are other charges levied on generators including intermittent generators. In using the electrical network to convey power, the generator will be charged transmission network use-of-system charges, TNU0oS for use of the high voltage network and be charged distribution network use-of-system, DNUoS for use of the medium and low voltage networks. There is a further charge levied for making the connection to the system known as the connection charge. Ancillary Services Some generators are contracted by the system operator to provide ancillary services to the gird. The provision of reserve has already been discussed. Providing control of the grid voltage (through provision of reactive power) is also required and some plant will be contracted to do this. A small number of plants are also contracted to supply black-start capability such that should all of the system collapse following a very serious problem, these black-start generators can restart without the assistance of an external electricity supply. Grid Code A grid code is a document that defines obligatory features of a power generator that is to be connected to the electricity transmission or distribution system. An item recently added to the UK grid code is a requirement for fault ride-through from large wind farms. Fault ride-through is the ability of a generator to stay connected to the grid even when the grid is experiencing a fault condition so that once the fault is cleared (and normally the faulted item can be disconnected in less than a second) the wind farm will be available to resume delivering power. Annex 7: Comparing the system margin and loss of load probabilities with and without intermittent generation: an illustrative example The following figures compare the system margins for three systems. In the first, all the energy is generated by conventional plant, and the system margin is such that the loss-of-load probability (LOLP) is 2.5%. The LOLP is indicated by the area shaded red, where demand is greater than available capacity. The maximum available capacity on this system would be about 20% of the peak demand - slightly higher than the sum of the rated capacities of the plant on the system, since for short periods operating the plant above rated capacity is possible. In the second example 80% of the energy output is provided by conventional and 20% by intermittent generation. In this example the mean capacity of the system is the same as that for the conventional system of Figure A7.1. There is no extra investment in thermal capacity to maintain reliability (sometimes termed ‘back up’ or ‘capacity reserves’).The effects are a marked increase in the loss of load probability, from roughly 2.5% to nearly 30% - and also a marked increase in the variance of the margin: In the third example, the increase of LOLP is neutralised by investment in extra capacity (‘backup’), which shifts the distribution to the right. As shown, an increase in the mean capacity on the system is such that the mean available margin rises from 9.0% (see Fig A7.2) to 20.9%, and is sufficient to restore the LOLP to the same level as that for the conventional system. The increased investment is approximately 12% of peak demand or 20% of the capacity of the intermittent plant. (See Figure A7.4 below, which shows the frequency distribution of the available capacity of conventional generation.) The capacity credit is 19.2%. These estimates are similar to those estimated using statistical formula (and also those of several other studies), though for slightly different parameters. Figure A7.1: Frequency Distribution of System Margin When Conventional Generation Supplies 100% of the Energy. Loss-of-Load Probability ~ 2.5 % Figure A7.2: Frequency Distribution of System Margin When Conventional Generation Supplies 80% of Energy and Intermittent Generation 20%, but with no Additional Investment in Capacity to Maintain LOLP. (LOLP rises to 30%) Figure A7.3: Frequency Distribution of System Margin When Conventional Generation Supplies 80% of the Energy, Intermittent Generation 20%, and Backup Capacity is Installed to Maintain Loss-of-Load probability to = 2.5%. Figure A7.4. Available conventional capacity (including backup) corresponding to Figure A7.3: 80% of energy is supplied by thermal and 20% by intermittent generation; backup capacity 20% (19.2% capacity credit). Further points: |. The spread in the margin increases significantly (note the difference in the scales in the axes of Figures A7.1,A7.2 and A7.3) when intermittent generation is added, which of course is a reflection of the greater volatility of output. 2. Although the loss-of-load probability is similar in both cases, in extreme situations (<0.5%) the cuts in supply would be deeper with intermittent generation (compare the red areas in Figures A7.1 and A7.3). The nature and depth of the outages is an important aspect of the problem. 3. In extreme cases the loss-of-load levels due to capacity shortages would be within the compass of demand management practices. As illustrated in Figure A7.4, the capacity of conventional plant on the system would be 108% of peak demand, the average available capacity 101% and the lower probability limit of available capacity 94%. 4. There are significant periods (during times of peak demand) when the output from the intermittent generators raises the available capacity to very high levels; these are periods when the fuel savings over the peak will be large. Assumptions for Preceding Results Monte Carlo simulations using Crystal Ball. No. of trials: 20,000. Calculations compare 20% energy addition from conventional capacity with 20% from intermittent capacity. Demand: Mean value normalised to 100%; Standard deviation, 3.0% of expected value. Thermal capacity: Normal distributions with means 2 standard deviations below installed capacity and standard deviations of 4.0% of mean capacity. Backup capacity: means 2 s.ds below capacity with s.ds = 5.0% of mean capacity. Intermittent capacity: Weibull distribution with mean of 20.0% of expected peak demand, s.d. 12.8% of mean; max. capacity 60% of expected peak demand, min. capacity 0%. Annex 8: Background documents and working papers The following documents were produced as part of the process that contributed to this assessment report. Hard copies are available from the UKERC HQ on request: 58 Prince’s Gate Exhibition Road London SW7 2PG Tel: +44 (0) 207 594 1574 Fax: +44 (0) 207 594 1576 Email: — admin@ukerc.ac.uk They are also available from the intermittency project pages of the UKERC website (hetp://www.ukerc.ac.uk/content/view/77/60). * Scoping note and assessment protocol * Discussion paper on key questions * Stakeholder workshop report + Workshop presentations The following Working Papers are also relevant to this assessment report, and/or the wider work of the TPA. They are available from the TPA pages of the UKERC website (http://www.ukerc.ac.uk/contenvview/55/67). * TNA User Needs Assessment + Working paper on energy and evidence based policy and practice * Power System Reserves and Costs with Intermittent generation, Anderson 2005 + Allocating costs arising from the capacity credit of intermittent options, UKERC 2005 UKERC UK ENERGY RESEARCH CENTRE 58 Prince’s Gate Exhibition Road London SW7 2PG tel: +44 (0)20 7594 1574 email: admin@ukerc.ac.uk www.ukerc.ac.uk UKERC is funded by the UK Research Councils