Options for large-scale wind power deployment in Sweden

Public ISBN nr. 978-82-93150-74-9 Options for large-scale wind power deployment in Sweden Commissioned by Skellefteå Kraft THEMA Report 2015-15

About the project About the report Project number: SKR-15-01 Report name: Options for large-scale wind power deployment in Sweden Project name: Vindkraftutredning Report number: 2015-15 Client: Skellefteå Kraft ISBN-number: 978-82-93150-74-9 Project leader: Berit Tennbakk Availability: Public Project participants: Anders Lund Eriksrud Åsmund Jenssen Completed: May 6, 2015 About Øvre Vollgate 6 0158 Oslo, Norway Company no: NO 895 144 932 offers advice and analysis related to the green energy transition. Our services are based on in-depth knowledge of energy markets and infrastructure, broad understanding of energy and climate policies, long-standing experience, and solid professional expertise within economics, finance, technology and law. Standard disclaimer: AS (THEMA) does not accept any responsibility for any omission or misstatement in this Report. The findings, analysis, and recommendations are based on publicly available information and commercial reports. Certain statements may be statements of future expectations that are based on THEMAs current view, modelling and assumptions and involve known and unknown risks and uncertainties that could cause actual results, performance or events to differ materially from those expressed or implied in such statements. THEMA expressly disclaims any liability whatsoever to any third party. Page ii

CONTENT SAMMANDRAG PÅ SVENSKA... 2 Slutsatser... 2 Beräknade projektkostnader... 2 Skatteberäkningar... 3 Nätinvesteringar och systemkostnader... 4 Investeringskostnader... 4 Systemkostnader... 5 Andra samhällsekonomiska effekter... 5 1 INTRODUCTION... 7 2 METHODOLOGY... 8 3 TOTAL INVESTMENT AND OPERATION COSTS... 9 3.1 Investment and operation costs for wind power generation... 9 3.1.1 Investment and operating costs... 9 3.1.2 Future cost development... 11 3.1.3 Grid tariffs and connection charges... 11 3.1.4 Cost estimations... 12 3.1.5 Discussion and conclusions... 14 3.2 Grid investment costs... 15 3.3 System costs... 18 4 TAX ESTIMATES... 22 4.1 Income tax... 22 4.2 Property tax... 23 4.3 Summary... 24 5 OTHER WELFARE ECONOMIC COSTS AND BENEFITS... 26 5.1 Environmental effects... 26 5.2 Employment effects... 27 REFERENCES... 29

SAMMANDRAG PÅ SVENSKA Denna rapport innehåller en analys av kostnaderna och nyttoeffekterna av två alternativa förslag till storskalig utbyggnad av vindkraft i Sverige. I det första alternativet är vindkraften landbaserad i norra Sverige (prisområde SE1 och SE2) och i det andra alternativet är vindkraften havsbaserad i södra Sverige (prisområde SE3 och SE4). I båda alternativen antar vi att den totala vindkraftproduktionen i Sverige är 55 TWh år 2030, vilket innebär att det måste byggas cirka 37,5 TWh ny vindkraft mellan 2020 och 2030. I båda fallen kopplas all vindkraft till stamnätet. Bakgrunden till analysen är den pågåande diskussionen i Sverige om att införa ett särskilt styrmedel för att stimulera havsbaserad vindkraft. Elforsk publicerade nyligen en översikt över kostnaderna för utveckling och drift av olika typer av elproduktionsanläggningar i Sverige. Rapporten visar att det är betydligt dyrare att bygga vindkraftskapacitet till havs än på land. Förespråkare för havsbaserad vindkraft menar emellertid att det är bättre med produktion i södra Sverige eftersom merparten av konsumtionen sker där, vilket innebär att en storskalig utbyggnad i norr kommer kräva betydande nätinvesteringar. I denna rapport har vi uppskattat storleken på den inkrementella kostnaden för att bygga ut havsbaserad vindkraft. I tillägg kan andra samhällsekonomiska effekter skilja sig mellan de två alternativen. Frågan är dock om de stora skillnaderna i investerings- och driftskostnader mellan de två alternativen uppvägs av andra kostnader. Slutsatser Investerings- och driftskostnader för storskalig havsbaserad vindkraft i södra Sverige är över 40 procent högre än för att bygga ut motsvarande landbaserad vindkraft i norra Sverige. Kraftproduktion i norr betalar en betydligt högre effekttariff än kraftproduktion i söder. Om man bygger ut 37,5 TWh vindkraft i norra Sverige motsvarar betalningen av effekttariff kostnaden för att bygga nio linjer med kapacitet på 500 MW genom snitt 2 (SvKs kostnadsbedömning). Förstärkning av stamnätet krävs vid storskalig vindkraftsutbyggnad både i norr och i syd. Extrakostnaden för att bygga havsbaserad vindkraft i södra Sverige jämfört med landbaserad i norra Sverige motsvarar det dubbla av nuanskaffningsvärdet för det nuvarande svenska stamnätet. Skatteinkomsterna från landbaserat vindkraft er något högre än från havsbaserad vindkraft, huvudsakligen på grund av at avskrivningsreglerna gynnar mer kapitalintensiva anläggningar. Vi kan inte se att skillnader i miljökostnader eller andra ekonomiska effekter försvarar den extra kostnad som utbyggnad av havsbaserad vindkraft leder till, även om utbyggnad av havsbaserad vindkraft ger fler arbetstilfällen än landbaserad vind. Beräknade projektkostnader Havsbaserad vindkraft är över 40 procent dyrare än på land De beräknade totala investeringskostnaderna för de båda alternativen presenteras i nedanstående tabell. Tabellen visar både kostnaden per kwh, totalkostnad i nuvärde och årlig kostnad. Page 2

Beräknade investerings- och driftkostnader för storskalig vindkraft i Sverige (37,5 TWh), i 2015 SEK. Landbaserad vindkraft i norr Havsbaserad vindkraft i söder Skillnad Kostnad [öre/kwh] 62.9 90.5 27.6 44 % Årlig kostnad [mdr. SEK] 23.6 33.9 10.4 44 % Nuvärde 1 [mdr. SEK] 270.5 389.3 118.7 44 % Skillnad (procent av landbaserad) Beräkningarna är baserade på följande antaganden: Det fins en stor potential för land- och havsbaserad vindkraft i Sverige De billigaste projekten realiseras först Lärokurvor innebär att kostnaderna minskar med ökande volymer för båda alternativen, men läroeffekterna är större för havsbaserad vind Skillnader i fullasttimmar mellan land- och havsbaserad vindkraft innebär att det behövs mindre kapacitet för att producera 37,5 TWh till havs än på land Antagandena är baserade på offentliga källor. De innehåller dock en del osäkerheter. Beräkningarna visar att det är stor skillnad i kostnader mellan att bygga landbaserad vindkraft i norr versa havsbaserad vindkraft i söder. Estimaten inkluderar nätkostnader som projekten måste betala, vilka är baserade på dagens nättariffer. Tariffstrukturen innebär att landbaserad vindkraft i norr betalar en betydligt högre nättariff än projekt i söder. I genomsnitt uppskattar vi att havsbaserad vindkraft i söder kostar ca 25 öre/kwh mer än landbaserad vindkraft i norr. Den totala kostnadsskillnaden är nästan 120 miljarder kronor i nuvärde. På årsbasis är skillnaden 10,4 miljarder kronor. Skatteberäkningar Vi har också beräknat storleken på skatteintäkterna till staten för de två alternativen. Projekten betalar fastighetsskatt och inkomstskatt till staten. Nettonuvärde av skatteintäkterna från samtliga projekt (37.5 TWh) under 20 år. Realvärde 2015. Landbaserad vind Havsbaserad vind Skillnad öre/kwh Miljarder SEK öre/kwh Miljarder SEK öre/kwh Miljarder SEK Inkomstskatt 6.91 29.72 2.19 9.42 4.72 20.30 Fastighetsskatt 0.67 2.88 0.67 2.88 0.00 0.00 Total 7.58 32.60 2.86 12.30 4.72 20.30 Skatteberäkningarna visar att landbaserad vindkraft totalt kommer att betala mer skatt än havsbaserad vindkraft. Det beror främst på att investeringskostnaderna är högre för havsbaserad vindkraft vilket innebär särskilt gynnsamma avskrivningsregler. I tillägg påverkar antagandet om att landbaserad vindkraft fortfarande får stöd genom elcertifikat, medan havsbaserad vindkraft får stöd genom anbud. Detta innebär att havsbaserad vindkraft i genomsnitt genererar en något lägre vinst. 1 Netto nuvärdet är beräknat som den årliga kostnaden (2030) upprepat under 20 år med en diskonteringsränta på sex procent. Page 3

Nätinvesteringar och systemkostnader Investeringskostnader Ett argument för att bygga ut havsbaserat vindkraft i söder istället för landbaserad i norr är att en ökad utbyggnad i norr kommer att kräva en ökad kapacitet i stamnätet för att transportera strömmen till södra Sverige. Detta är också argument bakom de högre effekttarifferna för kraftproduktion i norra Sverige. Av flera skäl återspeglar effekttariffen inte nödvändigtvis skillnaderna i nätkostnader för de två alternativen: Storskalig vindkraftsutveckling kommer oavsett alternativ att förändra flödet i näten, och därmed behovet (eller lönsamheten) av att öka kapaciteten i stamnätet De marginella förlustkostnaderna är beroende av hur vindkraftsproduktionen fördelas mellan olika noder i näten Dagens effektavgift i nättariffen återspeglar inte de verkliga ekonomiska kostnaderna som är kopplade til ökad produktionskapacitet i olika delar av nätet Med ett antagande om nuvarande nivå och differentiering av effekttariffer så betyder det att vindkraftsparker i norr kommer att betala i genomsnitt 1,48 öre/kwh i effekttariffer. Det utgör 555 miljoner kronor per år för 37,5 TWh produktion. I Perspektivplan 2025 uppger Svenska Kraftnät att det kostar 930 miljoner kronor att bygga en ny linje på 500 MW genom snitt 2 (eller en ny linje på 800 MW genom snitt 1). Det totala värdet av effekttariffer från en utbyggnad av 37,5 TWh vindkraft i norr motsvarar i givet fall kostnaden för att bygga nio nya linjer genom snitt 2, vilket i sig motsvarar en exportkapacitet på ca 40 TWh. Ökad landbaserad vindkraftsproduktion i norra Sverige En ökad vindkraftsproduktion i norra Sverige leder till ökad export söderut och därmed lägre priser även i södra Sverige. När produktionen ökar, ökar kraftöverskottet i norr, flaskhalsarna mot söder blir fler och prisskillnaden mellan norr och söder ökar. Vid tillräcklig stor prisskillnad, blir det lönsamt att öka överföringskapaciteten genom snitt 1 och 2. Ökad överföring ger prisutjämning, dvs. högre priser i norr och lägre priser i söder. När det totala överskottet i Sverige är tillräckligt stort, och priserna tillräckligt låga, blir det lönsamt att bygga nya exportlinjer till kontinenten. Hur mycket och när överföringskapaciteten måste stärkas, beror också på utbyggnaden av vindkraft i norra och centrala Norge och i norra Finland, nätkapaciteten mellan norr och söder i Norge och Finland, samt utveckling av kärnkraft i norra Finland. Ökad havsbaserad vindkraftsproduktion i södra Sverige Ökad vindkraftsproduktion i södra Sverige leder till lägre priser i både norra och södra Sverige. Priserna är lägre i norr eftersom överskottsenergi i norr fortfarande måste exporteras, medan importefterfrågan i södra Sverige minskas. Detta leder till att överskottet i norr transporteras genom södra Sverige till marknader längre söderut. När det totala överskottet i Sverige är tillräckligt stort, blir det lönsamt att bygga nya exportlinjer till kontinenten. Samtidigt är det nödvändigt att stärka elnätet norrut i Sverige för att nyttogöra sig flexibilitet i vattenkraften i norr som är nödvändig för att balansera variationerna i vindkraftsproduktionen i södra Sverige. Hur mycket och när det blir lönsamt att stärka överföringskapaciteten beror bl.a. på kärnkraftsutvecklingen, vindkraftproduktionen på kontinenten, och flaskhalsar i det tyska och polska nätet. Slutsatser om nätkostnaderna Vi har inte tillräklig faktaunderlag för att kvantifiera nätkostnaderna för de två alternativen i detalj. Kostnaderna beror dessutom på antaganden om utvecklingen av marknaderna i och omkring Sverige i övrigt. Vi noterar dock att det sannolikt kommer att finnas behov av att bygga ut nätet i båda alternativen, dock förmodligen i något högre grad om vindkraften baseras i norr. Frågan är således om kostnadsfördelen för vindkraft i norr försvinner om vi inkluderar högre nätkostnader. Page 4

Det kommer att finnas behov av att bygga ut kapaciteten i snitt 2, och kanske även i snitt 1. Vi har beräknat extrakostnaden för vindkraftsutbyggnad i söder till 120 miljarder SEK. För dessa pengar kan man bygga ca 170 nya linjer genom snitt 1 och snitt 2 (se ovan). Den samlade extrakostnaden motsvarar två gånger nuanskaffningsvärdet av den nuvarande svenska stamnäten. Med andra ord finns det inte fog för argumentet att det ökade behovet av nätinvesteringar betyder att vindkraften bör placeras i söder. Systemkostnader Storskalig utveckling av vindkraft innebär att det svenska kraftsystemet får en hög andel variabel kraftproduktion. Produktion från både landbaserad och havsbaserad vindkraft fluktuerar kraftigt och ofta. Under perioder om flera dagar kan vinden utebli och produktionen vara mycket låg. I ett sådant kraftsystemet finns behov av andra källor som kan balansera variationerna i vindkraft med kort varsel och som kan producera energi när det är vindstilla och efterfrågan är hög. Med mer vindkraft i systemet, måste Svenska Kraftnät förmodligen köpa mer reservkraft för att säkerställa att strömförsörjningen kan upprätthållas. Vad detta kommer att kosta beror på resten av elsystemet: Hur mycket och hur vindkraftsproduktionen varierar Potentialen och kostnaden för flexibilitet i annan produktionskapacitet och i konsumtion Möjligheten att kunna importera flexibilitet från andra marknadsområden Både land- och havsbaserad vindkraft varierar kraftigt. Det krävs således balansresurser för både landbaserad och havsbaserad vindkraft. Medan vindkraftproduktionen i södra Sverige er högt korrelerat med vindkraftproduktionen i Danmark och Tyskland, är vindkraftproduktionen i norra Sverige lite korrelerad med vindkraftproduktionen i söder. Samtidigt är det betydligt större tillgång på flexibilitet i norra Sverige på grund av den stora andelen vattenkraft både i norra Sverige och norra Norge, vilket innebär att det är billigare att balansera vindkraft i norr. Med en ökad utveckling av vindkraft i söder ökar efterfrågan på reglerresurser. Eftersom vindkraftproduktionen i söder är starkt korrelerad med vindkraftsproduktion i Danmark och norra Tyskland, kommer Sverige ofta att tvingas konkurrera om reglerresurser med dessa marknadsområden. Andra samhällsekonomiska effekter Vi har också analyserat skillnader i miljökostnader och sysselsättningseffekter mellan de två alternativen. Miljökostnader De huvudsakliga miljökostnaderna i samband med vindkraft är inngrepp i landskapet som har konsekvenser för naturvård och friluftsliv, samt ljud och effekter på djurlivet. Vindkraft innebär i allmänhet relativt små miljökostnader, även om det finns begränsad kunskap om några av effekterna, särskilt vad gäller påverkan på djur och fåglar. Det finns också individuella skillnader mellan olika projekt, både vad gäller land- och havsbaserad vindkraft. Konsekvenserna kan enligt Naturvårdsverket minimeras genom god planering. Totalt sett är det förmodligen något lägre miljökostnader i samband med havsbaserad vindkraft, men skillnaderna är i allmänhet små. Arbetstillfällen Havsbaserad vindkraft har högre kostnader i samband med både investering, drift och underhåll än landbaserad vindkraft. Det innebär också att det behövs fler årsarbeten både i utvecklingsfasen och produktionsfasen. Studier visar dock att en stor del av sysselsättningen relaterad till havsbaserade vindkraftsprojekt inte kommer ske i Sverige eftersom en del utrustning och processer är mycket Page 5

specialiserade. Detta kan också vara fallet för landbaserad vindkraft, men förmodligen i mindre utsträckning. Baserat på olika studier, uppskattar vi att antalet direkta arbetstillfällen i anknytning till utbyggnad av 37,5 TWh landbaserad vindkraft kan uppgå till ca 40 000 årsarbeten i byggskedet, dvs i genomsnitt kommer 4 000 personer att vara heltidsanställda under den 10-årsperioden när vindkraften byggs ut. I driftsfasen, från 2030 och framåt, uppskattar vi antalet heltidsanställda per år till omkring 1 000. Antalet arbetstillfällen för 37,5 TWh havsbaserad vindkraft är högre, och uppgår til 146.000 årsarbeten i byggskedet. Studier av Storgrundet och Kriegers Flak visar emellertid att bara 25 procent av dessa er årsarbeten som skapas i Sverige. I driftsfasen uppskattas antalet heltidsanställda till drygt 2000, dvs. ungefär dubbelt så högt som för landbaserad vind. Page 6

1 INTRODUCTION Wind power generation in Sweden has increased in recent years due to the ambitious renewables targets for 2020. The ambitions level and support scheme design after 2020 is currently under discussion. For example, the current government have said that they want to introduce a new support scheme for offshore wind in Sweden. Sweden has vast wind resources both onshore and offshore. As onshore wind is cheaper than offshore wind, the technology-neutral Elcertificate market has not provided sufficient support to make offshore wind generation profitable. Therefore, some advocate a separate support scheme to stimulate offshore wind, as offshore wind is mainly located in the south of Sweden, where most of the electricity consumption is located. The south of Sweden is a deficit electricity area, does not have potentials for hydropower generation, and is the area in which nuclear power stations are located as well. The Swedish nuclear plant are nearing their economic lifetime. Others argue that instead, the cheaper onshore wind resources in the north should be deployed large-scale. Their location may not be ideal as the north of Sweden is a surplus area due to large hydro generation already, but on the other hand, the hydro generation offers cheap regulating power, which is necessary to balance wind power generation. The main issue for the analysis presented in this report is: What are the welfare economic costs of large-scale deployment of onshore wind power in northern Sweden, versus the welfare economic costs of offshore wind power in southern Sweden? The underlying question is to what extent the benefits of wind deployment in southern Sweden outweighs the additional costs compared to wind deployment in northern Sweden. The report is commissioned by Skellefteå Kraft. Skellefteå Kraft is a regional energy utility in northern Sweden and a major wind power generator in the region. Page 7

2 METHODOLOGY We analyse a situation where Sweden decides to increase total wind generation to 55 TWh in 2030. This is in accordance with an ambition level suggested by Swedish wind energy. 2 The case is quite extreme, as we assume that the target is for wind power only, and that all new wind generation commissioned after 2020 is either only onshore wind in northern Sweden or only offshore wind in southern Sweden. Although the assumptions are clearly not completely realistic, they provide a clear-cut illustration of the differences. By northern Sweden we refer to bidding area SE1 and SE2. By southern Sweden we refer to bidding area SE3 and SE4. In order for Swedish wind generation to reach 55 TWh by 2030, we assume that a total of 37,5 TWh of wind generation must be commissioned between 2020 and 2030. According to the Swedish Wind Association, annual wind power generation in Sweden was 13,3 TWh at the end of 2014. Projects providing an additional 3 TWh are under construction or has been decided. Until the end of 2020, we expect the Elcertificate market to provide some additional capacity. Comprehensive cost-benefit analyses should include all costs and benefits of the studied activity. The main costs and benefits of wind power generation are Total investment and operation costs Grid investment costs System costs External effects such as environmental costs and security of supply effects In this analysis we compare two alternatives, and focus on the difference in economic consequences between the two. We do not discuss the optimal level of wind generation in Sweden. In addition, we present the distribution of costs and benefits among generators, the Swedish TSO 3 (Svenska kraftnät, hereafter abbreviated SvK), and state tax revenues. All effects cannot be or are not easy to quantify. We discuss non-quantifiable effects qualitatively. Moreover, we base the analysis on existing numbers and public sources. It has not been possible to make new detailed analyses within the scope of this study. We have not analysed the general market effects for the power or the Elcertificate market, nor what support systems should be used in the alternatives. We have assumed that onshore wind will still be supported by the Elcert system, whereas offshore wind will be supported through feed-in auctions. When we base calculations on assumptions about market values, we make it explicit in the text. The report is organised as follows: Chapter 3: Total investment and operation costs Chapter 4: Tax estimates Chapter 5: Other welfare economic costs and benefits 2 In February 2015, the Swedish Wind Association suggested a 2030 renewables target of 55 TWh for Sweden, see http://www.vindkraftsbranschen.se/blog/pressmeddelanden/sverige-topp-tre-i-vindkraftsutbyggnad/ 3 Transmission System Operator Page 8

3 TOTAL INVESTMENT AND OPERATION COSTS 3.1 Investment and operation costs for wind power generation This section presents estimates of the costs seen by the investors. These costs include investment and operating costs, and grid tariffs imposed by the TSO. The grid tariff need not reflect the true welfare economic grid costs, to be discussed in Chapter 4. Taxes are not included in the cost estimates, but treated separately in Chapter 0. This chapter first presents a discussion of the investment and operating costs, the potential for wind power in Sweden, and the future cost of wind power. Further, we discuss the Swedish grid tariffs that are assumed. Finally, we show cost estimates for 37.5 TWh of new wind power in the two alternatives in Chapter 3.1.4. 3.1.1 Investment and operating costs Wind power is a capital-intensive technology. There are no fuel costs, and investment costs make up the majority of the total cost. The local wind conditions, typically expressed through the number of full load hours per year, are very important for the resulting Levelised Cost of Energy (LCoE). The LCoE is a measure of the total investment and operating costs of a technology over the lifespan, and can be interpreted as the price needed (constant for all years) to cover the costs over the economic lifetime of a wind farm. The LCoE is therefore expressed in terms of costs per energy, that is, Swedish öre/kwh in this study. There is a large potential for onshore wind power in Sweden. Energimyndigheten (2014) recently estimated the technical potential in Sweden to about 160 TWh, of which the majority is available at an LCoE between 50 and 60 öre/kwh, see Figure 3.1. The estimate is based on Vindbrukskollen, covering about 90 percent of all wind power projects in Sweden. The study does however separate the potential geographically, and does not give how much of the potential that is found in Northern Sweden. Figure 3.1: The technical potential for onshore wind power in Sweden, estimated by Energimyndigheten (2014). Source: Energimyndigheten (2014) The LCoE of offshore wind power is higher than that of onshore wind power. Yet, the offshore wind conditions are often good, which helps reduce the LCoE. Distance to shore, water depth and local seabed conditions are important cost drivers for offshore wind power. The Baltic Sea offers good conditions for offshore wind power. The waters are shallow, and a less corrosive environment favour the Baltic Sea to other offshore wind locations (THEMA, 2013). Moreover, the seabed is suitable for gravitational foundations, due to the shallow water and solid grounds (Malmberg, 2012). Page 9

Energimyndigheten (2014) provides a numerical example of the cost of Swedish offshore wind power. The example shows an LCoE at around 100 öre/kwh, using a discount rate of 10 percent. However, Energimyndigheten does not give any estimate of the potential at this cost level. Elforsk (2014) estimates an LCoE of offshore wind power to 75 öre/kwh for a 600 MW offshore wind farm. This estimate is low compared to other studies. An operational cost of 18 öre/kwh is used, which is significantly below other studies, and with no detailed explanation. 4 There is great uncertainty about the technical potential for offshore wind power, and how the cost will develop if a large amount of offshore wind power is built. The known potential for offshore wind power in Sweden amounts to approximately 30 TWh, based on projects that hold a licence or has applied for one (Energimyndigheten, 2014). The projects holding a licence comprise approximately 8 TWh. To the best of our knowledge, there exists no study of the technical potential for offshore wind power in Sweden, beyond the identified projects. The cost estimates for onshore wind power used in this study are based on THEMA s in-house developed database of wind power projects. The database models the cost and potential for each identified project bottom-up, e.g., accounting for the number of full load hours and estimates of costs based on number of turbines, turbine size, etc. Moreover, the full cost of wind power is estimated in the database, including for instance cost of grid connection, administration and decommissioning. The database is frequently used in THEMA s analysis of the current elcertificate market. Figure 3.2 (left panel) shows the estimated cost curve for onshore wind power in Northern Sweden. The figure shows all costs, including grid tariffs, and the projects that we expect to be built as part of the current elcertificate target for 2020 are removed from the cost estimate. Figure 3.2: Estimated cost curves on identified projects onshore (Northern Sweden) and offshore. 5 The cost estimates for offshore wind power are based on the analysis in THEMA (2013). We have updated the estimates and used a discount rate of 6 percent and an economic lifetime of 20 years. The resulting cost curve is shown in Figure 3.2 (right panel). The investment cost is estimated as a function of the distance to the coastline and the sea depth, and both the operational costs and the cost of grid connection are estimated based on the distance from the coastline. 4 The operations and maintenance cost was estimated to be in the region 17-35 öre/kwh for global offshore wind power by IRENA (2012) and the analysis from THEMA (2013) estimated average Swedish offshore operations and maintenance costs to 28 öre/kwh. 5 The figure shows cost estimates for 2015 for onshore wind and cost estimates for 2020 for offshore wind. However, in the estimations in section 3.1.4, we account for learning effects, such that onshore wind power obtain five years of additional learning effects, relative to the estimates shown in Figure 3.2. Page 10

3.1.2 Future cost development There future cost development of wind power is uncertain. IEA (2012) presents a survey of studies of the future cost of onshore wind power, and concludes that the LCoE may be reduced between 0 and 40 percent from 2011 to 2030. These estimates are based on various sources, typically utilising learning curves in combination with expert elicitation, engineering models, and near-term market analysis. The cost of onshore wind power in Sweden has fallen significantly in recent years. According to Energimyndigheten (2014), the cost of onshore wind power fell 30 percent in Swedish kronor from 2009 to 2013. The cost reduction was due to turbine manufacturer over-capacity, sinking commodity prices, and increased competition. Because of the recent cost development, the future onshore cost reduction is expected to be modest. Offshore wind power technology is far less mature than onshore technology. Offshore wind power constituted only two percent of the total global installed wind power capacity in 2013 (Energimyndigheten, 2014). Moreover, the cost of offshore wind power increased by 17 percent between 2009 and 2013, because new projects were built further from the coastline. Due to the current high cost level and immaturity of offshore wind power, the future cost reduction of offshore wind power is expected to be significantly larger than that of onshore wind power. However, EC (2013) states that assumptions about the cost development of remote offshore wind power has been very positive, but that capital cost of remote offshore wind power is now expected to decrease more slowly than earlier. In this analysis, we use the estimated cost reduction from Fraunhofer (2013) for onshore and offshore wind power. The average LCoE of onshore wind power reduces by a factor of 0.4 percent per year, whereas the average LCoE of offshore wind power reduces by a factor of 1.4 percent per year. These estimates are close to the reduction factors for investment costs used by EC (2013). 3.1.3 Grid tariffs and connection charges We assume that the new wind power is fed directly into the central grid. The Swedish grid tariffs are divided into two components, the generator (power) tariff and the energy tariff. The energy tariff is based on a marginal loss percentage for the given node and the power price (fixed for one year at the time) and multiplied by a correction factor of 0.8. 6.Hence, the energy tariff covers the cost of grid losses. By including the energy tariff, the cost estimates in this report also take network losses into account, assuming that the tariff reflects the actual costs of losses. We base the energy term on the current marginal loss percentages provided in the grid tariffs for 2015, and an assumption about the future power price development in the Nordics. 7 A power plant located in Southern Sweden receives a negative energy tariff, because this region is a deficit area. Northern Sweden is a surplus area, and the energy tariff is therefore positive. We base the generator (power) tariff on the SvK s grid tariffs for the central grid for 2015. 8 The generator tariff is differentiated between nodes, and is increasing from south to north. There is uncertainty about the future generator tariff. In THEMA (2014), three future scenarios were discussed. We base our calculations on the first scenario, namely an extension of today s generator tariffs, e.g., corresponding to a generator tariff of 1.6 öre/kwh for a wind power plant in Northern Sweden (SE1) with 3000 full load hours. The second scenario in THEMA (2014) results in significantly higher generator tariffs, and the increase is largest in Northern Sweden. The same wind power plant faces a generator tariff of 5.1 öre/kwh in this scenario. The last scenario implies a harmonisation of the Swedish generator tariff with European levels, i.e., below 1 öre/kwh. Thus, the 6 Note that because the energy tariff is corrected by a factor of 0.8 and does not vary over the day or the year, the tariff does not reflect the correct welfare economic cost of losses. 7 We assume that the power price will stay around current levels until 2020, and gradually increase towards 40 EUR/MWh in 2030. 8 http://www.svk.se/contentassets/53e8b6dcdf9e4e12811773a738f02f78/prislista2015.pdf Page 11

uncertainty implies that the generator tariff may be significantly higher than our base case assumption, and that the impact will be greater in Northern Sweden than in Southern Sweden. Table 3.1 shows the resulting grid tariffs for an average project for each technology. Table 3.1: Average grid tariffs in öre/kwh (real 2015). Tariffs Onshore Offshore Generator tariff 1.48 0.93 Energy term (marginal losses) 0.60-0.87 Total 2.07 0.06 3.1.4 Cost estimations We now compare the total costs for the two technologies, as seen from the investors. Neither the onshore nor the offshore cost curve provide sufficient potential to cover 37.5 TWh of wind energy. As our best guess, we therefore extend the two cost curves as described below. We assume that the potential from 0 to 20 TWh consists of mature projects, and represent the best locations in Northern Sweden. In order to meet a demand of 37.5 TWh of onshore wind power, we expand the cost curve from 20 TWh and up (where we observe a cost increase), so that the total potential sums up to 37.5 TWh. Taking into account the abundant onshore potential, at a reasonable cost, identified in the study by Energimyndigheten (2014), we assume that there is a higher potential in this region of the cost curve. The cost and potential for offshore wind power is more uncertain, and identified projects do not meet a demand of 37.5 TWh. In this study, we scale the entire cost curve for offshore wind up to 37.5 TWh. We regard this as an optimistic approach to the cost of offshore wind power, because the potential beyond 30 TWh is unknown, and one may need to use less favourable locations, resulting in higher costs for such projects. However, Malmberg (2012) states that there is a large potential for offshore wind power in the Baltic Sea, without any further quantification. We use a discount rate of 6 percent (real, before tax), following Elforsk (2014). A discount rate of 6 percent serves as an estimate of the welfare economic cost of capital, which need not reflect the Weighted Average Cost of Capital (WACC) of the market participants. Offshore wind power is often treated with a higher WACC than onshore wind power, due to higher risks involved. Typically, a WACC around 10 percent is used for offshore wind power in the literature (e.g., THEMA (2013), Fraunhofer (2013), and Energimyndigheten (2014)). We use an economic lifetime of 20 years, as this is most common in the literature (e.g., done by Fraunhofer (2013), Energimyndigheten (2014), Elforsk (2014) and NVE (2015)). All numbers are given in real 2015 SEK. We further use an exchange rate of 9.3 SEK/EUR. Note that not all projects in the database hold a license or may actually be built. We assume that canceled projects have similar effects to the onshore and the offshore cost curves, so that the estimated differences are not significantly affected. The cost curves used in this study are proxies of the costs, based on available data. A linear development is assumed, i.e., 3.75 TWh of new wind power is built each year between 2021 and 2030. The cheapest projects are built first, so that the most expensive projects experience the greatest cost reduction due to learning effects. Page 12

Figure 3.3: Estimated cost curves after expansions and cost reduction effects. The resulting expanded cost curves after cost reduction effects are shown in Figure 3.3. The learning effects yields flatter cost curves. Results The results from the cost calculations are shown in Table 3.2. Offshore projects are significantly more expensive than onshore projects. We estimate the difference to 44 per cent. Moreover, the estimated net present value (NPV) of the difference amounts to 118.7 billion SEK over a lifetime of 20 years. Table 3.2: Cost estimates. All numbers in real 2015 SEK. Onshore Offshore Difference Difference (percentage of onshore) Cost [öre/kwh] 62.9 90.5 27.6 44 % Annual cost [billion] 23.6 33.9 10.4 44 % NPV [billion] 270.5 389.3 118.7 44 % The costs are split by category in Figure 3.4. The figure shows that both investment costs and operational costs are significantly higher for offshore wind power. The grid tariff is however higher for onshore wind power in Northern Sweden, reflecting that this region is currently a surplus area, whereas Southern Sweden is a deficit area. Page 13

Figure 3.4: Average costs split by category. Table 3.3.4 shows that the needed installed capacity for offshore wind power is smaller than that of onshore wind power in order to obtain an annual generation of 37.5 TWh. The reason is that the number of full load hours are higher for offshore wind power. However, there is a large difference in the LCoE for the two alternatives. Table 3.3: Resulting installed capacities and number of full load hours. Onshore Offshore Installed capacity 13 533 MW 10 766 MW Average full load hours 2 771 3 483 3.1.5 Discussion and conclusions The estimated costs yield a large cost difference between the two alternatives. The uncertainty in the estimates is significant, but the estimates are our best guess based on publicly available information. The largest uncertainty pertains to the cost estimates for offshore wind power, which may turn out substantially higher than assumed. Thus, the uncertainty rather points in the direction of a larger cost difference between offshore and onshore wind power, than a lower difference, compared to the numbers estimated in this study. This does not mean that the cost difference cannot turn out to be smaller, e.g., as a result of an unexpected cost revolution within offshore wind power. Yet, with the current knowledge, the cost difference is more likely to be larger than smaller than our best guess estimate. The cost estimate for offshore wind power is lower than that of the most recent study of Energimyndigheten (2014). The two major reasons are that we use a lower discount rate, and that we account for future cost reductions. We use the current generator tariffs and marginal loss percentages. A large-scale wind power development may result in changed grid tariffs. This will have effects on both alternatives. However, the current grid tariffs comprise a small share of the cost difference between the alternatives, so changed tariffs would not alter the conclusion. Note that we estimate the average cost of a large scale development, not only the cost of single wind power projects. A large scale wind power development implies that not only the very best locations Page 14

are utilised. This drives the cost up, and is an important reason why the estimated cost of onshore wind power is higher than the cost observed today. 3.2 Grid investment costs Grid tariffs and grid costs As we have seen in the calculations presented above, there is a substantial difference in the grid payments for generation capacity in the north and generation capacity in the south of Sweden. According to SvK these differences are cost-based. The correspondence between the geographic differentiation of the capacity charge in the grid tariff and grid investment costs is however unclear. A recent study, commissioned by the Swedish Energy Association, concludes that the current generator tariff model is not cost reflective (THEMA, 2014). In any case, if 37,5 TWh of additional wind power generation is introduced in the Swedish system the structure of the grid tariff would probably need to change. Hence, we cannot derive the welfare economic grid costs associated with large-scale wind power deployment from the difference in grid payments based on the current tariff model. Drivers for grid investment Increasing wind power generation in Sweden to 55 TWh represents a profound transition of the electricity system whatever the geographical distribution of the generation. Hence, both alternatives are likely to require massive grid investments. The question is to what extent the need for investments is different in the two cases. The analysis and discussions in this section are mainly based on the cases analyzed in the grid development plan presented by SvKs network long-term development plan from April 2013 (Perspektivplan 2025), including appendices (hereafter denoted Perspektivplan). A comprehensive analysis of grid investments would require access to a grid model, which we do not have. We expect grid expansion to be based on welfare economic analysis, i.e. that grid investments will be carried out if they are deemed welfare economically profitable: Grid investments are welfare economically profitable if price differences due to congestion between areas are expected to be high Welfare economic benefits include congestion revenues after the grid investment, and changes in generators and consumers surplus in the interconnected system Grid expansion between a deficit and a surplus area typically leads to price convergence, i.e. increasing prices in the surplus area and reducing prices in the deficit area Clearly, then, it matters whether the new wind power generation is located in the north or south of Sweden. However, as noted by SvK in the Perspektivplan (our translation from Swedish): Wind power expansion in the north, requires a strengthening of the transmission grid. But even wind power expansion in the south affect the transmission grid, as hydro power resources in Sweden, Norway and Finland to an increasing degree will be need as regulating resources. Finally, the configuration of grid expansion in Sweden depends on how and where new wind power in Norway and Sweden is located. We note that the internal grid within Sweden, particularly between SE2 and SE3 (snitt 2) must probably be strengthened in both alternatives, i.e. moving wind power investments to the south is not sufficient to avoid grid investments. It is however, difficult to determine exact or proximate thresholds for grid expansions. To what extent will it be necessary to strengthen the grid at lower levels of wind power capacity in the north than in the south? A complicating matter is that the grid consequences do not only depend on the location of wind power in Sweden, but even the development of the surrounding systems, as the Swedish electricity grid is highly integrated with surrounding market areas. Some of the main factors for differences in grid effects between the two alternatives are: Page 15

Deployment in the north o Wind power generation in Northern Norway and Northern Finland o Nuclear capacity expansion in Northern Finland o Grid capacity in Norway and Finland Deployment in the south o Interconnectors from Southern Sweden o Wind power deployment in Denmark and Northern Germany o Internal grid congestion in Germany and Poland SvKs analysis of wind power scenarios in the Perspektivplan In the Perspektivplan, SvK has analyzed different wind power scenarios. The scenarios imply much lower wind deployment (in 2025) than our alternatives (for 2030). However, the results shed some light on the impact on the profitability of, or need for, transmission grid investments in the alternatives. Table 3.5 shows the wind generation assumptions in SvKs scenarios. In VIND1 the wind power generation is quite evenly distributed between the north and the south. In VIND2, two thirds of the wind generation is located in the north, while the situation is the opposite in VIND3. Moreover, a larger share of the wind generation is offshore in VIND3 (almost 8 TWh as compared to only 0,5 TWh in VIND1 and VIND2). Table 3.4: Wind generation in SvKs wind power scenarios, TWh/year, 2025 Scenario Total wind generation in Sweden Wind generation SE1 and SE2 PP2025_VIND1 17,2 7,3 9,9 PP2025_VIND2 19,1 12,7 6,4 PP2025_VIND3 19,3 6,9 12,4 Source: Svenska kraftnät Perspektivplan 2025 Wind generation SE3 and SE4 The Perspektivplan does not report hourly prices or wind generation patterns. From the general price impacts (average annual price levels), we may however, derive some insights from the results. Table 3.6 shows the reported price levels for SE1 and SE4. In all scenarios, the price levels in SE2 and SE3 are between the price levels in SE1 and SE4. Hence, if the price difference between SE1 and SE4 is small, the other area price differences (SE1-SE2, SE2-SE3, SE3-SE4) are even smaller. It should be noted that the grid capacity between SE3 and SE4 (snitt 4) is smaller in VIND3 than in VIND2. However, the difference in (average) prices between SE3 and SE4 is small in both scenarios, so this assumption does not seem to affect the results substantially. Table 3.5 Average annual power prices per area depending on wind power deployment, EUR/MWh Price area PP2025_VIND1 PP2025_VIND2 PP2025_VIND3 VIND2-VIND1 VIND3-VIND1 SE1 59,1 58,2 57,4-0,9-1,7 SE2 55,9 58,9 57,9-0,6-1,6 Difference 0,4 0,7 0,5 Source: Svenska kraftnät Perspektivplan 2025 The changes in average prices and price differences are difficult to use as indications of the impact on grid investments. Generally, two areas may have the same average price level, but differences in the hourly price structure may still imply a high value of exchange. Figure 3.5 shows an example of the hourly (average) price structure during a week for the Swedish price areas. (The graph is equal to the left panel in Figure 3.8 in section 3.3, and is used for illustration here.) The figure reveals that Page 16

higher average prices in the south of Sweden are mainly explained by higher prices during workday daytime. During nights and weekends, congestions seldom occur. Hence, we may assume that increased average price differences stem from relatively higher price differences during the day. In that case, higher average price differences also imply increased value of exchange capacity from north to south. Figure 3.5 Example of average weekly price profiles in the Swedish price areas Source: THEMA model simulations In order to determine the profitability of grid reinforcements, more detailed market and grid analyses are needed. That is beyond the scope of this report. Observations and deductions: 1. The price differences between price areas are small in all scenarios. SvK also concludes that the value of increased grid capacity in interface 1 and 2 in the Swedish grid is low in all scenarios. 2. The price difference between SE1 and SE4 increases with increased wind deployment, i.e. both scenarios indicate an increased need for grid enforcements, regardless of location. 3. The price difference increases more in the scenario where the increase in wind power generation is located in the north (VIND2). Hence, increased wind deployment in the north increases the need for grid enforcements more than increased wind deployment in the south. 4. The price level decreases somewhat more in the scenario where more of the wind generation is located in the south (VIND3), in all price areas. This indicates that the (spot) market value of all power generation decreases more when the wind generation is increasingly located in the south. All the scenarios are similar apart from the wind generation assumptions. Observation number 4 implies that the relative profitability of increased interconnector capacity to continental markets (Germany, Zealand, and/or Poland) increases as well. This increases the probability that it will be profitable to increase the internal Swedish grid capacity (north-south) as well. As noted in the Perspektivplan: With two new interconnectors (to Zealand and Germany), the net present value of increased enforcements in interface 1 and 2 are precisely zero (p. 113). In other words, increased wind generation in the north definitely increases the need for grid capacity expansion, but increased wind generation in the south is very likely to do so as well. According to the Perspektivplan, the cost of a new 500 MW line in Snitt 2 is 100 million Euro, or just under 1 billion SEK. The cost of a new 800 MW line in Snitt 1 is the same. The Perspektivplan states that increased capacity between SE1-SE2-SE3 (i.e. over snitt 1 and snit 2) should be given priority. Page 17

The timing of the investment is indicated to 2020-2025, but (d)et återstår att närmare utreda anslutningspunktera (p. 70). We suspect that the price reduction in the south in VIND3 is partly explained by the (relative) lack of flexible power generation and the correlation with wind generation in Denmark and Germany. Prices in the north are lower in VIND3 because the surplus generation still has to be exported to the south, where the average import price is now lower. We return to this in the discussion about system effects in the next section. The need for grid investments in the two alternatives also depends on the development in nuclear power in Sweden. In 2030, a larger share of nuclear capacity may be phased out. The nuclear capacity is all located in the south (SE3). Hence, all else equal, the deficit in south Sweden may be substantially larger than what is assumed in 2025 in the Perspektivplan. This is likely to affect the need for investments in the Swedish grid and the economics of location of wind capacity. It is however not obvious that such a development will reduce the need for grid investments substantially, due to the intermittency of wind generation and the location of relatively cheap regulating resources in the north. 3.3 System costs A power system with a large share of intermittent generation is generally characterized by: Increased (hourly and diurnal) price variation Larger and faster changes in the residual load (load minus intermittent generation) Increased need for reserves to handle structural imbalances and even unexpected imbalances Prolonged periods with low generation from intermittent sources, i.e. increased need for energy back-up Increased value of flexible resources, i.e. for fast changes, peak load, energy generation and use However, the challenges also depend on The composition and characteristics of the intermittent generation The composition and characteristics of the rest of the capacity (and load) in the system Exchange capacity with other markets Characteristics of onshore and offshore wind power Onshore and offshore wind power are both intermittent energy sources with large and frequent changes in the generation level from hour to hour. In relative terms, offshore wind has more full load hours. We do not know however, to what extent offshore wind power is less intermittent than onshore wind. Both technologies are highly variable, and the minimum production is likely to be very close to zero. Hence, we cannot conclude that offshore wind will need less back-up and balancing resources than onshore wind generation. Characteristics of the rest of the capacity and load The rest of the system is very different in the north and south of Sweden, if regarded separately. Table 3.6 Overview of maximum generation from other sources and industry demand, TWh/year (2025) Area Hydro power CHP Nuclear power Condensing/GT Total generation cap. Industry consumption SE1+SE2 55 5 60 16 SE3+SE4 12 19 72 8 111 40 Source: Svenska kraftnät Perspektivplan 2025, THEMA power market model Page 18