Renewable energy sources in European energy supply and interactions with emission trading

ARTICLE IN PRESS Energy Policy 38 (2010) 2898–2910

Contents lists available at ScienceDirect

Energy Policy journal homepage: www.elsevier.com/locate/enpol

Renewable energy sources in European energy supply and interactions with emission trading ¨ n, Wolf Fichtner Dominik Most Institute for Industrial Production (IIP), Chair for Energy Economics, Universit¨ at Karlsruhe (TH)/Karlsruhe Institute of Technology (KIT), Hertzstraße 16, 76187 Karlsruhe, Germany

a r t i c l e in f o

a b s t r a c t

Article history: Received 26 September 2009 Accepted 13 January 2010 Available online 1 February 2010

This paper presents a model-based approach, which allows to determine the optimised structure and operation of the EU-15 electricity supply under different political and economic framework conditions, with a focus on the integration of renewable energy sources for electricity generation (RES-E) in the EU15 countries. The approach is designed to take into account the characteristics of power production from both renewable and conventional sources, including the technological and economic characteristics of existing plants as well as those of future capacity expansion options. Beyond that, fuel supply structures are modelled, as well as the international markets for power and CO2-certificates with their restrictions. Thus, a profound evaluation of the exploitation of mid-term renewable potentials and an assessment of the market penetration of the various renewable power generation technologies under the (normative) premise of a cost-optimised evolution of the power system becomes possible. Results show that a promotion of renewable energies reduces the scarcity of CO2-emission allowances and thus lowers marginal costs of CO2 reduction up to 30% in 2030. Despite the higher overall costs, a diversification of the energy resource base by RES-E use is observed, as primarily natural gas and nuclear fuels are replaced. & 2010 Elsevier Ltd. All rights reserved.

Keywords: Electricity market modelling Renewable energy Emission trading

1. Introduction Along with efforts to achieve CO2 mitigation and a better energy efficiency, one major contemporary strategic challenge for the European electricity supply system is the integration of substantial amounts of renewable energy sources. In addition to the specific goals for each EU Member State already set in Directive 2001/77/EC for 2010 (Commission of the European Communities, 2001), more ambitious targets have been set for the share of renewable sources in final energy consumption (not only electricity) for 2020 in the Directive 2009/28/EC (Commission of the European Communities, 2009).1 While the performance parameters of renewable energy technologies are continuously being improved, the politically

n

Corresponding author. Tel.: + 49 721 608 4689. ¨ E-mail address: [email protected] (D. Most). 1 These targets have been set for the share of energy from renewable sources in gross final consumption of energy, but no specific targets have been set for the share of electricity from renewables. However, member states are obliged to set up national allocation plans for renewable energies, where the amount of energy coming from renewable sources has to be distinguished for the transport, electricity, heat and cooling sector (see Commission of the European Communities, 2009). 0301-4215/$ - see front matter & 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.enpol.2010.01.023

and environmentally motivated introduction of significant amounts of renewable electricity generation is likely to depend on incentive schemes in the short- to mid-term future. Besides the inhomogenous geographical distribution of renewable energy resources, the temporal evolution of renewable electricity market penetration in the EU Member States will thus be influenced by the different design options for national promotion schemes and their possible future harmonisation. Likewise, the future cost structure of conventional electricity generation also has an influence on the economy of renewable electricity generation and the necessary support. Moreover, physical interdependencies between renewable and conventional power generation exist. In order to develop adequate policies and strategies, policy makers as well as decision makers in utilities must be able to consider the above mentioned interdependencies in order to get an idea of the future consequences of their decisions. In this paper a modelling approach will be presented which enables a quantitative assessment of the long-term role of renewable electricity production under varying framework conditions within the liberalised European electricity market. At first, methodological aspects of RES-E integration will be briefly introduced in Section 2. Subsequently, the chosen modelling approach will be shortly described in Section 3. The focus of this

ARTICLE IN PRESS D. M¨ ost, W. Fichtner / Energy Policy 38 (2010) 2898–2910

paper is set on the evolution of renewable energy production and its cost effects in European electricity supply, which will be presented for selected scenarios in Section 4. The paper ends with a conclusion and outlook.

2. Methodological aspects of RES-E integration In most existing energy system models,2 renewable power production and the associated energy potentials are commonly not part of the optimisation, i.e. the use of available renewable resources is not handled as an endogenous variable in the models, but introduced exogenously as a predetermined expansion path. One of the reasons for this practice is that the representation of individual RES-E potentials and their costs requires comparatively extensive efforts in terms of modelling and data gathering. Due to the hitherto low penetration and the minor relevance for the electricity sector as a whole, these efforts were worth-while only to a limited extent, as for low expansion rates the resulting bias in the model are small in relation to the whole power sector. However, the currently observed and politically fostered increase of the penetration of renewable energy carriers with growing shares of fluctuating generation makes a more appropriate representation of renewable power production in energy sector planning models desirable. Further, in order to assess the effects of renewable electricity promotion instruments adequately, these also need to be integrated into the existing modelling approaches. While some approaches exist, which focus on the composition of the renewable part of the mix and take into account the available dynamic potentials and incentive mechanisms, these are mainly based on static merit order curves for conventional power production. Mostly, the approaches rely on market simulations, which are based on the balancing of supply and demand curves. While the supply curves on the one hand can be derived from the available potentials of individual renewable electricity generation options and their costs, the demand for renewable electricity on the other hand is determined from the electricity price and the price incentives given by the incentive schemes. Partly, the approaches are dynamic and take into account inter-temporal relations, as e.g. the evolution of incentive schemes or cost decreases as a result of learning effects. In the following, some of these dedicated approaches will shortly be introduced and outlined. One example for a dynamic simulation of the renewable electricity market in the EU is the Admire Rebus model (cf. Uyterlinde et al., 2003). Its approach is based on the static simulation methodologies of the REBUS3 (Voogt et al., 2001) and ElGreen (Haas et al., 2001) models. Both REBUS and ElGreen use national static marginal supply cost curves to simulate an ideal TGC market. These curves indicate the correlation between the price of electricity and the amount of electricity produced from a given source and are derived from estimates of different RES-E 2 There exist several well-known models of the long term development of energy markets, such as MARKAL (Market Allocation Model, Fishbone and Abilock, 1981), PRIMES (Capros et al., 1998; Antoniou and Capros, 1999), EFOM (Energy Flow Optimisation Model, Finon, 1974; Van der Voort et al., 1984), MESSAGE (Model for Energy Supply System Alternatives and their General Environmental Impact, Agnew et al., 1979; Messner and Schrattenholzer, 2000), TIMES (The Integrated MARKAL EFOM System, Remme, 2006) and NEMS (National Energy Modelling System, Department of Energy (DOE), 2008). However, these models are in general able to include renewable energy potentials (see e.g. Remme, 2006), but detailed analyses dealing with renewable energies covering EU15 are unknown to the authors. Besides that, it has to be pointed out that most of the above mentioned models are energy system models (covering different energy sectors) and in opposite to that PERSEUS-RES-E focus only on the electricity sector. A more detailed overview about models analysing European energy markets can be found ¨ and Perlwitz, 2009). in (Most 3 Renewable Electricity BUrden Sharing.

2899

potentials, their costs and expected performance. A similar approach as in ADMIRE REBUS is used in the Green-X model (Huber et al., 2004) and the GreenNet model (Obersteiner et al., 2006). As the ADMIRE REBUS model, these models take into account fully dynamic cost-resource curves, i.e. the potential and the cost of each renewable energy technology are determined endogenously in the model, depending on the one hand on the static cost-resource curves, and on the other hand on the outcome of the previously simulated year as well as the policy framework conditions set for the simulation year. In the case of the GreenNet model, special regard is given to the costs for the grid integration of renewable energy sources and various scenarios for their allocation. All these models have in common that the electricity commodity price development is an exogenous model input,4 which can be varied in scenarios. Seasonal load profiles, the international electricity exchange and its interrelation with the ETS are usually not taken into account. Similar limitations apply to the modelling of interactions of large shares of fluctuating RESE use with conventional electricity generation. Modelling approaches explicitly focussing on these interactions usually comprise a shorter time horizon with a high temporal resolution, but without regarding inter-temporal and inter-regional aspects in the optimisation of the capacity and production mix. Furthermore, many of the described model-based simulation or optimisation approaches, even those explicitly developed for a detailed analysis of renewable electricity evolution, neglect the interactions with conventional power production. While this is acceptable for low shares of fluctuating renewable electricity generation, these interactions become more and more significant at higher penetration rates. For those approaches that explicitly investigate the interaction of fluctuating wind power feed-in with conventional electricity generation5 it can be argued6 that they often merely consider the absorption of wind power in today’s base load dominated electricity systems. As the system structure will not be static in the medium to long term, this results in a biased view of the effects in the conventional power system. Consequently, the developed hybrid modelling approach introduced in the following aims at combining the relevant long-term and short-term aspects of conventional and renewable power generation as well as their interactions on different time scales. This hybrid modelling approach consists of the optimising long-term energy system model PERSEUS-RES-E (Rosen, 2007) and the heuristic model Aeolius (Rosen et al., 2007) for the temporally highly resolved simulation of the scheduling of conventional power plants with growing fluctuating wind energy feed-in.

3. Long-term energy system model 3.1. Outline of the developed modelling approach The general objective of the energy system model PERSEUS-RES-E is to provide an analysis tool for the quantification of the economic and technological impacts that the policy framework of the required utilisation of renewable sources for 4 In the case of the GreenNet model, the exogenously given price development results from a stochastic fundamental electricity market model. 5 There exist several approaches which investigate the interaction of fluctuating wind power feed-in with conventional electricity generation. At this point, we refrain from giving a detailed overview about such models and refer to ¨ the following literature dealing with such models: Leonhard and Muller (2002), ¨ Fischedick (1996), Lux (1999), Kramer (2003), Lux (1999), Sontow (2000), Dany (2000), Dena (2005), Holttinen (2004) and Brand et al. (2005). 6 ¨ This argumentation is e.g. also brought forward by Kramer (2003).

ARTICLE IN PRESS 2900

D. M¨ ost, W. Fichtner / Energy Policy 38 (2010) 2898–2910

electricity generation in combination with the CO2 reduction commitments may have on conventional and renewable technology choices, certificate (allowance) prices and inter-regional power exchanges as well as on electricity prices under varying exogenous energy sector framework conditions. In the following the general approach will only be shortly introduced and the interested reader is referred to a detailed description of the model, which can be found in Rosen (2007). The developed model PERSEUS-RES-E is a multi-regional energy and material flow model, representing the electricity sector of the EU-15 as well as that of six further neighbouring European countries (NO, CH, PL, CZ, SK, HU). Methodologically, the model is based on a multi-periodic linear optimisation approach. The target function requires a minimisation of all decisionrelevant costs within the entire energy supply system modelled. This basically comprises fuel supply and transport costs, transmission fees, fixed and variable costs of the physical assets (operation, maintenance, load variation costs, etc.) as well as investments for new plants. The relevant techno-economic characteristics of the real supply system have been considered by implementing further equations covering technical, ecological and political restrictions. The most important technical restrictions, common also to the other models of the PERSEUS family, are:

 Physical energy and material balances: match of demand and  

supply taking into account storage options and time structures of electricity and heat demand (load profiles). Capacity restrictions: transmission capacities, availability of installed capacities, (de)commissioning restrictions and technical lifetime of physical assets. Power plant operation: maximum/minimum hours of full load operation, fuel options, cogeneration options and load variation restrictions.

Ecological and political restrictions in the model comprehend inter alia emission reduction targets or minimum restrictions for the use of renewable energies. The consideration of transmission capacities and losses as well as transmission fees ensures a realistic representation of the real power exchange characteristics within the model. Further, given the obviously strong interdependencies between electricity and CO2 markets, the linkage needs to be adequately reflected by the chosen modelling approach. In order to analyse the impact of the emission trading scheme on the physical electricity market, a second market layer, i.e. the CO2 allowance market, is an integrated part of the model.

deployment for the conventional technologies involved (i.e. transient conditions such as plant start-up, load change, fuel consumption at different output levels, etc.). Along with additional information (e.g. from the German Energy Agency, Dena, 2005) three relevant fluctuation-induced effects of RES-E technologies have been integrated into the long-term energy system model:

 Secured capacity.  Reserve capacity requirements.  Fuel- and CO2-inefficiencies due to balancing power, partial load operation, additional plant start-ups and shut-downs. These values have been calculated for different power plant portfolios for future years with AEOLIUS and the above mentioned effects have then been considered in the long-term optimisation with additional constraints guaranteeing the corresponding secured capacities, reserve requirements as well as the inefficiencies depending on the amount of renewables used.8 3.3. Data basis and main model assumptions The model calculations rely on an elaborate database of conventional as well as renewable technology data. About 1500 individual units represent existing conventional power plant technology classes (where the data is derived on the basis of the World electric power plant database, Platts, 2005) and expansion options (see Rosen, 2007, page 139). Technological developments are included in the way that future expansion options depend on the year of construction. In Table 1 the input data for a pulverized subcritical steam generator with carbon capture and storage on the basis of hard coal is illustrated and shows inter alia different efficiencies for future years. This means that technological progress is based on exogenous parameters.9 In addition, approximately 1500 further units have been integrated into the model to represent the potentials of 15 different renewable energy sources for electricity generation in the member states of the EU-15. This mid-term potential as well as the costs for the utilisation of these potentials in each member state is based on Klobasa and Ragwitz (2004) and Held et al. (2009). These available RES-E potentials are (quasi-static) economic potentials taking social and political barriers into account. Technological progress of renewable energies is included in the way that unused potential of renewables is (in general) larger due to a higher efficiency and cheaper in later years.10 Import potentials (such as Desertec11) of renewable energies outside

3.2. Effects of fluctuating RES-E production Fluctuations are mainly caused by the intermittent availability of wind power. For an adequate representation of their effects the PERSEUS-RES-E model was amended in order to account for the interdependencies of these short-term load-changes with the scheduling of conventional capacities as well as with long-term capacity expansion planning. The challenges for plant scheduling (occurring on a small time scale of 1 h down to 10 min) have been analysed with the power plant dispatch model AEOLIUS7 (see Rosen et al., 2007). The current or future power system structure as determined by the PERSEUS-RES-E optimisation model can be transferred into AEOLIUS and extended by further parameters necessary for the description of short-term power station 7 Fluctuations by all types of RES-E are taken into account. However, fluctuations induced by wind energy have by far the highest impact in Europe and are thus of special importance.

8 The models are coupled with a hard link meaning that each model generates its output files. Selected output files are input files for the other model. After each data exchange and model calculation, differences in the results files are generated and then manually checked. In general only two iterations steps were necessary and the iteration procedure was manually stopped. 9 For the example according to Table 1 this means that CCS technology improves in efficiency between 2020 and 2030 independent of whether and how much this technology is applied in the scenarios. Learning curves are one possibility to include indigenous technology improvements (see e.g. Remme, 2006). However, technological progress has been treated as exogenous parameter in this model as the complexity (and the solvability) of such a large model significantly increases with learning curves and additional findings with learning curves instead of exogenous parameters are limited. 10 Fig. 1 shows that for electricity generation from renewable sources specific electricity generation costs actually increase as capacity or output increases. This is explained by the fact that renewable potential is more and more exhausted. However, the effect of technological improvement, which works in the opposite direction, is considered in the way that quasi-static cost potential curves are used. This means that for later years unused potential of renewables is cheaper and larger due to a better efficiency. 11 http://www.desertec.org/

ARTICLE IN PRESS D. M¨ ost, W. Fichtner / Energy Policy 38 (2010) 2898–2910

2901

Table 1 CCS expansion option. Technology

Year Net efficiency (%)

Specific investment costs (h/kW)_

Fixed costs (h/kW/a)

Additional variable costs (cent/kWhel)

Lifetime Emission factor (kt CO2/TWhel)

Pulverized supercritical steam generator with CCS (hard coal)

2020 39 2030 43

1825 1825

40 40

1.2 1.2

20 20

12

4

AT BE DK FI FR DE GR IE IT LU NL

3

PT

2

ES SE UK

11 Electricity generation costs [ Cent/kWh]

85 77

DE

PT

10

ES

SE

FR

9 IE

8 7

UK

6 5

DK

1 0 100

0

200

300

400

Electricity generation potential [TWh/a] Fig. 1. Electricity generation potential for onshore wind energy in the EU15 (cp. Held et al., 2009).

Table 2 Assumed world market price trends for fossil energy carriers (based on IEA, 2008). World market price (cent2005/kWhtherm)

(2005) Data

2010

2015

2020

2025

2030

Fuel oil Natural gas Hard coal (world market) Natural gas

(3.05) (1.89) (0.72) (1.89)

3.44 2.08 0.63 2.08

3.34 2.09 0.64 2.88

3.43 2.17 0.65 3.79

3.52 2.24 0.67 4.76

3.62 2.31 0.68 5.78

Reference trend Reference trend Reference trend High gas scenario

Europe are not considered in the model. The electricity generation potential for onshore wind energy is exemplarily illustrated in Fig. 1. Each step in Fig. 1 is represented by a single plant with a given capacity and cost structure so that the model can optimize the utilised potential. Furthermore, the data basis of the model includes information on the existing transmission system infrastructure, electricity demand profiles as well as the expected demand increase and other energy-economic framework assumptions. The latter include e.g. fuel supply options and energy carrier price developments. The assumed energy carrier price developments on world markets are depicted in Table 2. Prices of the energy carrier lignite are assumed to be between 3.9 h/MWh (e.g. in Germany) and 4.6 h/MWh. Both taxation of energy carriers and transporting the fuels to the point of use in the respective power stations cause an additional fuel cost component, which is accounted for based on the location of the power plants. The values in the model differ from 0.0 to 0.2 cent/kWhtherm for hard coal, with the lower value applying to Poland and the higher one to Austria and Switzerland, respectively. For natural gas, the range is between 0.0 cent/ kWhtherm, as e.g. for the Netherlands, up to 0.3 cent/kWhtherm, as e.g. for Italy. On the demand side, the total electricity demand per country is exogenously given based on IEA (2008), European Commission (1999) and European Commission (2006). However, losses on transmission line capacities are considered. Depending on the level of electricity exchange between countries, a higher/ lower production is necessary to fulfil the exogenously given demand.

The whole model accounts for approximately 1 million variables in about 0.9 million equations with 4.8 million nonzero elements. The model is implemented in GAMS (General Algebraic Modelling System—see Brooke et al. (1998)) and is solved with CPLEX. Depending on the specifications for the scenarios, calculation times on a PC with 3.0 GHz processor and 4 GB RAM range from 30 min to several hours.

3.4. Analysis options PERSEUS-RES-E is thus a profound tool that allows to derive technically feasible and economically efficient long-term strategies for electricity sector development, taking into account many possible combinations of alternative framework conditions. The detailed representation of available resources for renewable electricity generation and their interactions with the conventional power sector allow a cost-based, normative assessment to be carried out for different design options of quantity- and pricebased incentive mechanisms. It is the aim of these analyses to derive spatially and temporally optimised profiles for the utilisation of conventional and especially of renewable technologies. Among others, the total and specific additional costs for the use of RES-E under the chosen schemes can be determined as a result of these analyses. Moreover, they allow evaluating the relative importance of these costs in comparison to other influences on system expenditures, such as fuel price changes or the stringency of CO2 allowances. Based on the specified green

ARTICLE IN PRESS 2902

D. M¨ ost, W. Fichtner / Energy Policy 38 (2010) 2898–2910

electricity quotas for the considered countries, an inter-regional trade with green certificates is represented in the model. In this case, the marginal costs of compliance with renewable electricity targets are an indicator for an equilibrium price for green certificates. In the case of Europe-wide technology or producer specific renewable quota assignments, e.g. under a possible future Community Support Framework, a green certificate trade directly among renewable electricity producers can easily be implemented into the existing model after assigning the corresponding renewable quotas to the obliged electricity producers. The above features make the modelling approach a useful techno-economic tool to test different design options for future renewable electricity support in advance. Offering a technologically feasible and economically efficient solution for each design option, it can help to derive reasonable, regionally and technologically diversified renewable electricity quotas or feed-in tariff levels, which would allow achieving future renewable electricity targets.12 It also allows determining the most cost-efficient renewable energy burden sharing solution for a breakdown of European targets on a national level and could thus help by the design of the national renewable energy action plans (see Commission of the European Communities, 2009). Thus, the developed models and the results that can be derived make the approach suitable for a variety of applications. For policy advice and also in research applications it can be used as an instrument to judge the effectiveness and efficiency of existing or planned policy measures as well as their interaction with other policy measures, as e.g. RES-E targets and emission restrictions, or nuclear energy policies. For manufacturers of conventional and renewable electricity generation technologies, the model allows to assess the future market potential based on the technology choices made. Further, utilities and independent power producers can benefit from the model results to support their expansion and contingency planning. Regional and technology-specific information on future market shares and the expected prices for power and CO2 certificates can be derived.

it can in principle be assumed that the value of renewable electricity generation is underestimated. The chosen sectorspecific model type implies further limitations regarding the economic analysis of electricity production from renewable sources, as the interrelations with other economic sectors are not accounted for. With the exception of CO2 emissions, also other external effects of energy supply are not accounted for. Furthermore, it has to be pointed out that the strengths of the model can be seen in the integrated assessment of renewable targets, its interaction with the conventional system and its impact on CO2 abatement costs and electricity prices both derived from system marginal costs (for CO2 reduction and electricity demand satisfaction). However, it has to be mentioned that input data is of special importance. The input data has been gathered with care and discussed with several experts. Sensitivity analyses have been performed to analyse the impact of the parameters. This means that main findings also depend on the quasi-static economic cost potential curves of renewable resources. Technological changes have been incorporated as far as it is possible from a today’s view.13 The model can in general be compared with the situation of the following example: when crossing a busy road, every person sets up a picture of the future, which should answer the question of whether the road could, or should, be crossed. In condition that the appraisal of the moving cars on the road reaches the conclusion that there is enough time to cross the road without being run over, the decision is taken to cross the road. It is in this context that such model should also be seen. In contrast to the illustrative example, however, this model gives decision support to stakeholders facing different developments in the energy sector. The scenarios should thus not be seen as a forecast or picture of the future, but should help to support decision making.

3.5. Critical reflection

4.1. Definition of the scenario calculations

However, besides these possible fields of application, the limitations of this type of technology-focussed energy model also shall be discussed. As a consequence of the chosen system boundaries, the model only represents a selected part of the problem areas relevant for the utilisation of renewable electricity generation technologies. Assuming a perfect foresight represents a simplification of real world decision processes, leading to capacity- or production-related decisions being taken instantly when required. The chosen optimisation premise, based on a minimisation of system expenditures, implies a cost-based competition of renewable electricity generation with existing conventional power plants and future expansion options. With investment decisions primarily based on system restrictions, strategic behaviour of market participants, social and other nontechnical barriers and drivers are neglected in the chosen approach. Thus, the model is not suited for the analysis of existing and possible future market imperfections. As the value of renewable electricity in the model is determined by the saved costs of conventional electricity generation and without taking into account market imperfections,

The future use of renewable energy technologies in power generation does not only depend on the technical and economic characteristics of the technologies themselves as compared to conventional power generation technologies, but also on a number of other general framework conditions. Often these framework conditions are of a political nature, as e.g. in the case of CO2 restrictions, renewable targets, or the German nuclear phase-out. They can also be of an economic nature, as e.g. in the case of the fuel price development, or of a technological nature, when technological barriers exist that may either be overcome or remain limiting in the modelled time horizon. Any change in the development of one of these framework conditions or their combination has effects on the relative competitiveness of energy technologies in general, and that of renewable electricity generation in particular. It is thus the aim of this model-based analysis to characterise the economically optimised future evolution of the power system under a range of possible developments of those framework conditions. The modelling approach allows to assess the market penetration of renewable electricity generation under different

12 The yearly fixing of feed-in tariffs is a challenging task and has to be done on a more detailed basis for smaller regions. However, the model can help to give an overview about the whole system and its interdependencies. It is in this context that the model can help to analyse feed-in tariffs for different technologies, their interdependencies with electricity prices and CO2 price developments.

13 Learning curves could be one way to directly incorporate technology changes. Nevertheless, it has to be mentioned that learning curves can also only consider today’s expected developments.

4. Model-based analysis of the role of renewable energy sources in electricity supply in EU 15

ARTICLE IN PRESS D. M¨ ost, W. Fichtner / Energy Policy 38 (2010) 2898–2910

2903

1400

Renewable_nondisp Renewable_disp Gas Oil CCS Coal Lignite Nuclear Hydro

3500 3000 2500 2000 1500 1000 500 0

Electricity generation [TWh/a]

Electricity Generation [TWh/a]

4000

Tide and wave

1200

PV Solar thermal

1000

Geothermal Wind offshore

800

Wind onshore Biogas Biowaste

600

Solid biomass Small hydro

400

Large hydro

200 0

2000

2005

2010

2015

2020

2025

2030

2000

2005

2010

2015

2020

2025

2030

Fig. 2. Structure of installed capacities and power production in total (left side) and only for renewables (right side) in the EU15 in the reference scenario.

framework conditions. In the following selected scenarios will be presented:

 Share of RES-E production (EU-15 or per country), e.g. to assess the effectiveness of incentive systems.

 Technology-specific exploitation of available mid-term RES-E  Reference scenario ‘‘INCENTIVES’’: a continuation of current





promotion schemes for renewable energy sources (e.g. feed-in tariffs, quota systems, etc.) is assumed as well as a continuation of CO2-emission trading with a reduction target of 30% until 2030 in comparison to the values from national allocation plans in 2005. Scenario ‘‘TARGETS 2030’’: stipulation of compliance with individual (national) RES-E targets in 2010 (Directive EC/77/ 2001) and stipulation of compliance with an assumed overall EU RES-E target of 1166 TWh for 2020 and of 1600 TWh for 2030. Scenario ‘‘WITHOUT SUBSIDY’’: a support scheme for renewable energy does not exist in this scenario. This scenario serves as a calculation basis of a totally cost-optimised energy system to derive the additional costs of renewable energy sources in the other scenarios.

Furthermore, the influence of other important (but not RES-E related) framework conditions can be studied, as for example:

 Evolution of fuel prices (especially natural gas).  Options for a (dis-)continued use/construction of nuclear capacities.

 Strictness of post-Kyoto CO2 constraints. In this paper, only a variation of the gas price development will be presented. This selection is motivated by the fact that the development of the gas price is quite uncertain, especially in comparison to other fuel prices (coal, lignite, uranium). Additionally, the CO2 price derived from marginal abatement costs is significantly influenced by the fuel switch-costs (gas to coal) and thus the gas price has a significant impact on model results. Two additional scenarios based on the scenarios WITHOUT SUBSIDY and TARGETS 2030 have been calculated with significant higher gas prices (WITHOUT SUBSIDY-HIGH GAS and TARGETS 2030HIGH GAS). The scenarios and parameter studies can subsequently be evaluated and compared to each other for the EU-15 as a whole or on member state level. The following results are of special interest:

 Evolution of technology mix (renewable and conventional capacities/production).

potentials.

 Cost considerations (e.g. total costs of RES-E use, average costs of tapped potentials, specific additional costs of power generation due to RES-E use). 4.2. Evolution of the European electricity system Within the reference scenario, current promotion and incentive schemes for the support of renewable energies are assumed. This scenario represents a business-as-usual situation in which renewables are fostered according to the national legislation. Additionally, a relatively stringent emission reduction path is introduced. With the Emission Trading Scheme beginning from 2005, the utility sectors in the modelled regions are equipped with emission allowances derived from the national allocation plans (NAP) for the period 2005–2007. From there onwards, emissions have to be reduced by 30% until the year 2030 in comparison to 2005. In the following, the developments in the power sector of the EU-15 and of Germany as well as France (only for the reference scenario) will be described in more detail. In the reference scenario INCENTIVES, model results indicates an increased total installed capacity in 2010 and a significant growth of capacity until 2030, which results from increased renewable capacities as well as from an increase in natural gas-fired capacities. Both types of plants are characterised by a lower capacity utilisation in comparison to typical base load power plants. The capacity utilisation of renewables depends further on the renewable technology. For example, an onshore wind power plant in general has between 1500 and 3000 full load hours depending on the wind quality at the site, while biomass plants have significantly higher full load hours but still less than nuclear plants with about 8000 full load hours.14 The forecasted mix of capacities and power generation from the different energy carriers in the EU15 is depicted in Fig. 2 for the modelled time horizon. The introduced emission restrictions are achieved with renewables and with the less emission-intensive energy carrier natural 14 Full load hours of conventional technologies are endogenously modelled in the short and long-term model. Thus, full load hours could be different within the two model results. However, we observed that both models calculated approximately the same full load hours for the different technologies. In opposite to that, full load hours of renewable technologies strongly depend on the natural availability of the resource and are thus per se in the model calculations the same.

ARTICLE IN PRESS 2904

D. M¨ ost, W. Fichtner / Energy Policy 38 (2010) 2898–2910

4000 Electricity Generation [TWh/a]

1200

800 600 400 200

3500 Renewable nondisp.

3000

Renewable disp.

2500

Gas Oil

2000

CCS

1500

Coal Lignite

1000

Nuclear Hydro

500

2020

2030

2010

2020

Without subsidy

Targets_2030

Incentives

Without subsidy

Targets_2030

Incentives

Without subsidy

Without subsidy

Targets_2030

Incentives

Without subsidy

Targets_2030

Incentives

Without subsidy

Targets_2030

Incentives

2010

Targets_2030

0

0

Incentives

Capacities [GW]

1000

2030

Fig. 3. Structure of installed capacities and power production in the EU15 in the reference scenario (INCENTIVES), the TARGETS2030 scenario and the WITHOUT SUBSIDY scenario.

gas. The capacities as well as the production of the existing CO2intensive technologies lignite are to a very large extent replaced by natural gas and renewables. Compared to the base year 2000, power production from lignite is reduced by 105 TWh in 2030, equivalent to a decrease of 57%. The corresponding values for fueloil are 127 TWh or 98% less. The production from hard coal fired power plants remains on a relative constant level at about 370 TWh/a until 2030, whereof 4 TWh/a are produced by CCS plants, while the production from nuclear power increases by 150 TWh up to 1050 TWh/a until 2030 (about 18% more). Gasfired power plants supply about 900 TWh/a in 2030, which is nearly a doubling of the production in comparison to the period of 2000. The production from hydro power slightly increases by 30 TWh up to 370 TWh/a in 2030 mainly caused by new small hydro power plants. About 320 TWh/a in 2030 are produced by dispatchable renewable energy carriers (biowaste, solid biomass, biogas, geothermal). Nondispatchable renewable energy carriers (wind onshore, offshore, photovoltaic, solar thermal, tide and wave) supply about 710 TWh/a in 2030. The amount from renewable energies is slightly more than the required target for the year 2010. Some countries (e.g. Germany) significantly overcomply their targets from the Directive 2001, while others do not. Nevertheless it has to be kept in mind that this result of the INCENTIVE scenario is calculated under the premise that the quota-based systems in the countries using them will be completely successful in reaching the respective targets in these countries. But this is not always the case as the quota-based instruments are not as successful as they should be (see Ragwitz et al., 2006). However, only 80% of the aimed renewable energy target for 2020 is achieved.15 The evolution of renewable power production is accomplished mainly by the use of wind power resources, both onshore and offshore, as well as renewable biogenic electricity sources (cf. Fig. 2 right side). Available wind potentials are realised up to 70% (330 TWh) offshore, and to more than 85% (335 TWh) onshore. Already in 2015 80 TWh or 20% of the available offshore wind resources are utilised, and 170 TWh or 45% of the available onshore wind energy resources. Concerning biogas, the corresponding utilisation rates are 57% (57 TWh) in 2030 and 27 TWh

15 This figure refers to the assumed target of 33% (corresponding to approx. 1166 TWh) in electricity production in the scenario TARGETS2030.

in 2010. Solid biomass potentials are realised to a maximum of almost 65% (191 TWh) in 2030 and 85 TWh in 2010. The biodegradable fraction of municipal solid waste is exploited to more than 75% (24 TWh) in 2030 and 15 TWh in 2010. Reaching a total electricity production of 67 TWh in 2030, nearly the entire remaining small hydro potential is also developed continuously throughout the modelled time horizon. On the other hand, the assumed available geothermal resources are realised to a large extent (40 TWh), primarily after 2020. Solar potentials are used, with only 49 TWh of PV and 9 TWh of solar thermal electricity generated in 2030. Furthermore, no additional large hydro capacities are built (Fig. 2). For Germany (see Fig. 4 right side) the optimisation results indicate that there are only minor changes in the total generated electricity. Hard coal and lignite capacities are replaced by natural gas fired units and by renewables. The same is true for the nuclear capacities with their CO2-free power production, which are decommissioned partly in 2015 and fully in 2020. Although the extensive construction of natural gas fired capacities partly compensates for the decommissioned fossil and nuclear capacities and leads to 126 TWh of electricity produced from natural gas in 2030, this type of power plant does not reach the same amount of full-load hours as the base load nuclear technology. This fact is reflected in the evolution of the generation mix, where especially the share of nondispatchable renewable energies significantly increases up to a production of 206 TWh/a in 2030. Imports reach a maximum value of almost 90 TWh in 2030, while CO2-intensive coal production is cut to 30 TWh, i.e. 77% less than the initial production of 130 TWh in 2000. Almost identically, power production from lignite decreases by 70% from 131 TWh in the year 2000 to 40 TWh in the year 2030. The situation in France (see Fig. 4 left side) is different to the situation in Germany, especially as nuclear capacities are significantly increased. The production from nuclear power plants augments from about 410 TWh/a in 2000 up to 580 TWh/a in 2030. Beside nuclear, the production from hydro power slightly augments until the year 2030, especially due to new built small hydro power plants, while the production from other renewables achieve a level of nearly 150 TWh/a in 2030. The electricity generation from renewables is mainly based on wind power and biomass. Different from the physically restricted power flows among the regions, the exchange of CO2 certificates has no technical limits.

ARTICLE IN PRESS D. M¨ ost, W. Fichtner / Energy Policy 38 (2010) 2898–2910

France

Germany

140

2010

2020

CCS Coal

200

Lignite Nuclear

100

2030

2010

2020

Without subsidy

Targets_2030

Incentives

0

Hydro

Without subsidy

Without subsidy

Targets_2030

Incentives

Without subsidy

Targets_2030

Incentives

Without subsidy

Targets_2030

Incentives

20

Gas

300

Targets_2030

40

Renewable_disp

Incentives

60

Renewable_nondisp

400

Without subsidy

80

500

Targets_2030

Capacities [GW]

100

Incentives

Electricity Geneeration [TWh]

600

120

0

2905

2030

Fig. 4. Structure of power production in Germany and France in the reference scenario (INCENTIVES), the TARGETS2030 scenario and the WITHOUT SUBSIDY scenario.

As no regulations by trade limits are introduced in the scenarios the price can reach an equilibrium all over Europe. Price information are derived from the marginal costs of the emission restriction. Along with a growing relative shortage of emission allowances due to the rising electricity demand in the European countries, marginal reduction costs increase from 5h in 2005 up to 42h in 2030. The marginal reduction costs for CO2, which can be interpreted as an indicator for the future certificate prices, are described for all scenarios in Section 4.4 in more detail.

4.3. Influence of promotion schemes and renewable targets for renewable energy utilisation Beside the business-as-usual scenario INCENTIVES, two further scenarios will now be introduced: TARGETS2030 and WITHOUT SUBSIDY. Both scenario calculations are based on the business-asusual scenario and only the parameters concerning the promotion of renewable energies are varied. In the so-called TARGETS2030 scenario, the evolution of the capacity and production mix clearly reflects the targeted amount of renewable energies for the years 2010, 2020 and 2030. The targets for the year 2010 are fixed on national level corresponding to the Directive 2001/77/EC. For the year 2020 the target is fixed at 1166 TWh for EU15 and was derived on the basis of the Directive 2009/28/EC and European Renewable Energy Council (EREC) (2005).16 For the year 2030 the target is fixed at 1600 TWh, which is based on discussions concerning the renewable target in this year (see Fraunhofer Institute Systems and Innovation Research (Fraunhofer-ISI), 2009). In opposite to the scenario TARGETS2030, in the scenario WITHOUT SUBSIDY it is assumed that no promotion and incentive scheme for renewable energies exists. This situation is a hypothetic one, but it is used as comparison where all technologies compete on a completely competitive market when any externalities as well as past (and also non-renewable present) subsidies are not taken into account. Scenario TARGETS2030: the framework conditions enable the model to choose the cost-optimised solution of reaching the cumulative European RES-E target by utilizing the most cost efficient potentials on a European scale, instead of having to resort to probably more costly potentials in individual countries. Apart 16 An overall target of a 20% share of RES in primary energy demand, translating into 1166 TWh, or more than one third of electricity generation from RES in 2020, has been proposed by the European Renewable Energy Council (European Renewable Energy Council (EREC), 2005), Uyterlinde et al. (2005).

from predicting the situation in a possible future green certificate trading scheme, the results of this scenario can also give an indication whether the national RES-E targets for 2020 have been chosen in a way that prevents economic inefficiencies, which can occur if more expensive potentials have to be chosen in individual countries, while less expensive potentials in other countries remain unused. The scenario is characterized by a notable increase of renewable electricity capacities, totaling 380 GW in 2020 and 510 GW in 2030. The production from dispatchable renewables increases up to 790 TWh in 2030 (including hydro) and from nondispatchable renewables up to 830 TWh (see Fig. 2). The higher production from renewables mainly substitutes the production from nuclear (about 180 TWh less than in the reference scenario) and from gas-fired plants ( 25 TWh in comparison to the reference scenario). On the other side, the generated electricity from coal-fired plants increases by 20 TWh in comparison to the reference scenario. This is due to the fact, that the higher amount of CO2 free electricity from renewables enables higher emissions in the conventional sector. The evolution of renewable power production is accomplished mainly by the use of wind power resources, both onshore and offshore, as well as renewable biogenic electricity sources (cf. Fig. 5). Available wind potentials are realised to more than 90% (440 TWh) offshore and (366 TWh) onshore in 2030. Already in 2010 110 TWh of the available offshore wind resources are utilised, and 137 TWh of the available onshore wind energy resources. Concerning biogas, the corresponding utilisation rates are 71% (71 TWh) in 2030. Solid biomass potentials are realised to a maximum of almost 92% (270 TWh) in 2030. The biodegradable fraction of municipal solid waste is exploited to more than 90% (30 TWh) in 2030. Reaching a total electricity production of 70 TWh in 2030, nearly the entire remaining small hydro potential is also developed continuously throughout the modelled time horizon. The assumed available geothermal resources are realised with up to 70% (41 TWh/a), primarily after 2020. Solar potentials remain virtually untapped, with 5 TWh of PV and 10 TWh of solar thermal electricity generated in 2030. Furthermore, no additional large hydro capacities are built. On the right side of Fig. 5, the used renewable potential is depicted in comparison to the reference scenario and in Table 3 the share of renewable electricity production for the different European countries is compared to the targets in the European Directives. The scenario calculation TARGETS2030 can also be interpreted as a situation with perfect green certificate trading scheme. Therefore, the marginal costs of RES-E target compliance are

ARTICLE IN PRESS 2906

D. M¨ ost, W. Fichtner / Energy Policy 38 (2010) 2898–2910

500 1800

2030

PV Solar thermal

1200

[TWh/a]

Tide and wave

1400

Geothermal [TWh/a]

unrealised potential

400

1600

1000

300

2020 2010

200

2000

Wind offshore 100

Wind onshore

800

Biogas

0 2000

2005

2010

2015

2020

2025

INCENTIVES

INCENTIVES

TARGETS_2030

INCENTIVES

Geo

TARGETS_2030

INCENTIVES

Biogas Solid Bio- Large Small Tide & Solar PV bio- waste hydro hydro wave thermal mass

2030

TARGETS_2030

INCENTIVES

TARGETS_2030

INCENTIVES

TARGETS_2030

INCENTIVES

TARGETS_2030

INCENTIVES

TARGETS_2030

INCENTIVES

TARGETS_2030

Large hydro

INCENTIVES

Small hydro 200

TARGETS_2030

TARGETS_2030

Solid biomass

400

INCENTIVES

0

Biowaste

TARGETS_2030

600

Wind Wind onoffshore shore

Fig. 5. Evolution of renewable electricity generation in the EU15 by energy carrier in the TARGET_2030 scenario (left side) and comparison of the used renewable energy potential (TARGETS2030 versus INCENTIVE).

Table 3 Targets and model results for the share of energy from renewable sources. Directive 2001/77/EC

Austria Belgium Denmark Finland France Germany Great Britain Greece Ireland Italy Luxemburg Netherlands Norway Portugal Spain Sweden Switzerland

Directive 2009/28/EC

Model results: share of electricity production related to the national electricity demand

National indicative targets for the Share of energy from contribution of electricity produced from renewable resources in gross RES-E to gross electricity consumption (%) final consumption of energy (%) 2010 2005

Target for share of energy from Scenario TARGETS renewable resources in gross 2030 (%) final consumption of energy (%)

Scenario INCENTIVES (%)

2020

2010

2020 2030

2030

Scenario WITHOUT SUBSIDY (%) 2030

78.1 6.0 29.0 31.5 21.0 12.5 10.0

23 2.2 17.0 28.5 10.3 5.8 1.3

34.0 13.0 30.0 38.0 23.0 18.0 15.0

83.1 6.9 38.3 32.9 22.6 19.9 14.6

90.0 13.4 72.9 44.9 33.1 42.4 27.7

89.5 18.2 104.1 50.5 42.0 55.8 39.3

97.2 13.4 75.5 43.1 31.8 58.2 21.0

63.8 2.0 12.7 19.9 11.2 10.4 1.9

20.1 13.2 25.0 5.7 9.0 n.a. 39.0 29.4 60.0 n.a.

6.9 3.1 5.2 0.9 2.4 n.a. 20.5 8.7 39.8 n.a.

18.0 16.0 17.0 11.0 14.0 n.a. 31.0 20.0 49.0 n.a.

22.0 34.3 26.7 8.5 13.0 102.2 48.9 32.0 65.0 64.2

29.0 64.3 25.8 14.0 20.7 92.9 60.1 35.1 77.5 57.5

36.1 93.7 26.2 17.0 30.5 84.0 87.4 44.0 79.2 52.1

35.7 82.9 26.3 17.4 29.9 84.0 60.9 50.0 56.4 52.1

11.1 17.5 13.4 4.4 2.7 84.0 52.3 17.8 49.6 52.1

determined. These can be seen as an indicator for the certificate price under an ideal, Europe-wide green certificate trading scheme. In the EU-Target scenario, with cumulative targets specified for the EU15 in 2020 and 2030, the marginal target compliance costs amount to 68.8 cent/kWh in 2020, and to 37.8 cent/kWh in 2030.17 In Germany the situation in the TARGETS2030 is nearly the same as in the reference scenario (see Fig. 4), while in France the production from renewable energies is significantly higher than in

17 In the model, extension limitations for different renewable technologies are taken into account as it is assumed that a technology has a limited growth rate (an example: it is not possible to construct 10 GW wind power in Germany in one year as the production capacities for wind power plants are limited). To achieve the exogenously given target, the limited growth rate has an impact as maybe a more expensive technology has to be chosen to achieve it. Thus, marginal costs of target compliance could increase. For later years (after 2020), learning effects are assumed which decrease the quasi-static economic potential and thus also marginal costs of compliance are decreasing.

the reference scenario. Renewable electricity production increases up to 280 TWh in 2030, which is about 70 TWh more than in the reference scenario. Thereby, especially wind, solid biomass and biogas are used to a larger extent. The renewable production exclusively replaces electricity from nuclear plants, while the export remains on the same high level. Scenario WITHOUT SUBSIDY: while the demand in the EU15 increases continuously throughout the time horizon, the model results indicate an only marginally increased total installed capacity in 2030 in the scenario WITHOUT SUBSIDY, which results from a more intensive utilisation of existing capacities, before new capacities are constructed. When comparing the beginning and the end of the modelled time horizon, the entire additional demand encountered until 2030 is covered by new natural gas-fired and coal-fired capacities with carbon capture and storage (CCS) as well as nuclear capacities. The increasingly stringent emission restrictions introduced by the ETS and afterwards are triggering this structural change to the more expensive, but less emission-intensive energy carrier natural gas and

ARTICLE IN PRESS D. M¨ ost, W. Fichtner / Energy Policy 38 (2010) 2898–2910

2907

60 WITHOUT

50

SUBSIDY

Certificate price [ /t]

WITHOUT SUBSIDY - High

40

Gas price INCENTIVES

30

TARGETS_2030 TARGETS_2030 High Gas price

20

10

0 2005

2010

2020

2030

Fig. 6. Influence of renewable electricity support on evolution of marginal costs of CO2 reduction.

uranium as well as to the new technology CCS in later years. Compared to the base year 2000, power production from coal and lignite (without CCS) is reduced by 340 TWh in 2030, equivalent to a decrease of 54%, while the production from coal plants with CCS increases after the year 2020 up to 510 TWh in 2030. Beside this significant increase, also the production from nuclear power plants augments by 50% up to 1280 TWh in 2030. Nearly no renewables are constructed without support schemes. Nevertheless, the CO2 reduction target of 30% until the year 2030 can be achieved with the above scrutinized power plant portfolio. In Fig. 3, the structure of installed capacities and power production is compared with the two other scenarios. However, the post 2030 development should not be neglected. Even if this is not modelled, it should be pointed out that it is unclear whether such a development of the electricity sector as described by the WITHOUT SUBSIDY scenario until 2030 can make further, much more ambitious CO2-abatement targets after 2030 (e.g. near to zero net CO2 emissions) possible or economically feasible. The evolution on a European level can also be seen on the national level. In Germany, especially gas-fired plants and coalfired plants with CCS substitute the production from renewables in comparison to the reference scenario. Beside, the electricity production in Germany in the scenario WITHOUT SUBSIDY is significantly lower in the year 2020 in comparison to the other scenarios. This can be explained by the fact that Germany imports larger amounts of nearly CO2-free nuclear power production from neighbouring countries in this scenario. In France, exclusively nuclear power plants compensate the production from renewable energies (see Fig. 4).

4.4. Cost effects of renewable energy sources Beside the development of the renewable electricity generation, the cost effects of the increased renewable electricity shares due to the applied promotion instruments shall be shortly assessed. Fig. 6 shows the evolution of the marginal costs of CO2 reductions. It can be noticed that the marginal costs of CO2 reduction with the additional renewable targets in the TARGETS scenario and the incentive schemes in the INCENTIVE scenario are significantly lower than in the WITHOUT SUBSIDY scenario. This is explained by the fact that with renewables less and cheaper conventional reduction options like the fuel switch from coal to

natural gas are sufficient to achieve the required CO2 reductions. Thus, although the mandatory use of renewable electricity causes higher total system costs, the increased exploitation of RES-E potentials has positive effects on the CO2 certificate prices. In the periods up to 2010 of the INCENTIVES scenario, more RES-E potentials are used than in the scenarios with targets. This is also expressed in lower marginal CO2 reduction costs until then. With a comparatively lower amount of RES-E potentials realised in 2020 and 2030 in the INCENTIVES scenario, the certificate price is higher again, as more expensive conventional reduction measures need to be realised instead. In the scenario WITHOUT SUBSIDY, marginal costs of CO2 reduction increase from 8 h/t in 2005 up to 55 h/t in 2030. Beside these three scenarios, two additional scenarios have been calculated taking higher gas prices into account. In the scenario TARGETS2030-HIGH GAS, marginal costs of CO2 reduction are significantly higher than in the normal TARGETS2030 scenario and increase up to 47 h/t. In this scenario also the generation from coal fired plants with carbon capture and storage is significant higher than in the gas base price scenario. In the WITHOUT SUBSIDY HIGH GAS scenario marginal costs of CO2 reduction are the highest in 2020 but are then lower in 2030 in comparison to the scenario WITHOUT SUBSIDY. This can be explained by the fact, that CCS plays a crucial role when the gas price is that high and with it, the certificate price is capped at a level of approximately 50 h/t. Looking at the development of the marginal costs of power production,18 it can be noted that the mandatory introduction of RES-E mitigates the increase of marginal generation costs in all EU15 countries. This can be traced back to the so-called meritorder effect (see Sensfuß et al., 2008). It was sometimes argued that the merit-order effect only occurs in a static power plant model (without investment planning), nevertheless this effect can also be seen in the long-term energy system model, amongst others when existing capacities are replaced. However, it only occurs, when the amount of renewable energies has to be increased by a stipulated target or incentive scheme. If renewables remain on a constant level, the conventional power system adapts to the situation and marginal costs of power production 18 Marginal costs are derived from the demand restrictions for each time interval of each year. The yearly value is calculated by averaging the marginal costs of each time interval (on a volume weighted basis). Marginal costs of power production can be used to derive a fundamental estimation of electricity prices.

ARTICLE IN PRESS 2908

D. M¨ ost, W. Fichtner / Energy Policy 38 (2010) 2898–2910

70 Italy WITHOUT SUBSIDY Italy INCENTIVES

Marginal costs [EUR/MWh]

60

Italy TARGETS_2030 Germany WITHOUT SUBSIDY

50

Germany INCENTIVES Germany TARGETS_2030

40

France WITHOUT SUBSIDY

30

France INCENTIVES France TARGETS_2030

20

Av. EU15 INCENTIVES

10 0 2005

2010

2020

2030

Fig. 7. Influence of renewable electricity support on evolution of marginal costs of power production in Germany, Italy and France. (Results maybe different for the base year 2005. Main reasons for this deviation are the missing renewable incentives or targets in 2005, which have also an impact on system marginal costs. Beside that interperiodical restrictions can also have an impact. A power plant extension in 2010 has an impact on future years, but sometimes it could also have an impact on former years.)

converge. Marginal costs of power production are derived for the different time segments and countries in the model.19 In Fig. 7 the influence of renewable electricity support on the evolution of marginal costs of power production is exemplified depicted for the countries Germany, France and Italy. The average marginal costs of power production in the EU15 for the different scenarios are shown in Fig. 6. The EU15 average in 2020 is 48 h/MWh in the reference scenario versus 53 h/MWh in the WITHOUT SUBSIDY scenario, while in 2030 marginal costs are quite similar (56 h/ MWh in the WITHOUT SUBSIDY and 53 h/MWh in the INCENTIVE scenario). With the ambitious renewable targets, marginal costs of electricity production are in some countries significant lower than in the scenario WITHOUT SUBSIDY, as e.g. Germany and Spain. This can be explained by the fact that the mandatory use of renewable electricity substitutes conventional technologies and shifts the merit order curve in a way that allows the remaining demand to be covered by technologies with lower marginal generation costs (see Fig. 7). Next, the costs caused in the renewable electricity sector by the promotion of RES-E in the EU15 member states are analysed. For each scenario with promotion of renewable electricity use, these are derived from the difference of renewable power production costs in the respective scenario and the renewable power production costs in the WITHOUT SUBSIDY scenario. With almost 72.4 billion Euro2005 the highest costs are induced in the INCENTIVES scenario. In the TARGETS2030 scenario these costs are somewhat lower, amounting to 63.2 billion Euro2005, although the production from renewable energies is about 20% higher than in the INCENTIVE scenario. The reduced additional costs in the TARGETS2030 scenario are due to the relocation of renewable electricity use from regions where the compliance with the national target is especially expensive (e.g. Italy, Spain and Sweden) to other European regions with less expensive potentials available (mostly Germany and the UK, to a lesser extent also Denmark, Finland and the Netherlands).

19 Typical days are used to describe the demand curve. The demand curve is constructed on 48 time steps per year (4 seasons with working and weekend-days with 6 respectively 3 time slots per day). Marginal costs are derived for these 48 load levels and are averaged to yearly values.

The average generation costs of the used renewable energies are between 68 and 100 h/MWh (depending on the year and the country) in the INCENTIVE scenario and thus slightly higher than in the scenario TARGETS2030 (58–90 h/MWh). Finally, by relating the difference of total system expenditures in the scenarios with RES-E promotion and those in the TARGETS2030 case to the total amount of electricity produced in each region, the net additional costs per kWh of total electricity production are determined. In Fig. 8 the costs caused in the renewable sector by the promotion of RES-E in the EU15 are added to the marginal costs in the corresponding scenario. The merit-order effect is overcompensated by the these costs for renewable energies. Until the year 2020, the costs including the reallocation of the RES-E support costs (red line in Fig. 8) are only slightly higher than the marginal costs in the scenario WITHOUT SUBSIDY. After 2020, especially in 2030 costs including the redistribution of the RES-E costs are significantly higher than the marginal costs in the base case. This means, that until the year 2030 the costs caused in the renewable sector by the promotion of RES-E per MWh increase to 19 h/MWh. In comparison to marginal costs of 55 h/MWh, this is an increase of approximately 34%.20

5. Conclusion and outlook A model-based approach for the assessment of the potential contribution of RES-E to the European power supply and its interactions with existing power generation has been introduced. It makes use of two complementary models, the long-term optimising energy system model PERSEUS-RES-E and the AEOLIUS dynamic simulation model for power plant scheduling at increasing levels of fluctuating (wind) power production. The approach allows to quantitatively determine the influence of RESE policy options and framework conditions, indicating in which 20 However, it has to be pointed out that the differences of yearly system expenditures are added on top to system marginal costs. As system marginal costs can be used to derive electricity prices, this comparison should help to get a better impression of the height of the additional costs for promoting RES-E.

ARTICLE IN PRESS D. M¨ ost, W. Fichtner / Energy Policy 38 (2010) 2898–2910

70

2909

EU WITHOUT SUBSIDY EU INCENTIVES

Costs [EUR/MWh]

60

EU TARGETS_2030 EU INCENTIVES with

50

reallocation of RES-E support EU TARGETS_2030 with

40

reallocation of RES-E support

30 20 10 0 2005

2010

2020

2030

Fig. 8. Influence of the reallocation of renewable electricity support on marginal costs of power production in average in EU15.

EU Member State what kind and what capacity of RES-E facilities are most economical, and when they should be installed. Results show that without targets or financial incentives the current and expected future cost situation in the electricity sector (without taking into account external costs, however) would only allow to a very small amount of renewable sources to become a favourable economic solution, even if CO2 restrictions are introduced. In this case, CO2 reductions in conventional power production (e.g. fuel-switch from coal to gas and introduction of CCS), or increased imports of less CO2-intensive electricity are cheaper options than the development of renewable electricity production. On the other hand, a substantial increase of renewable electricity use with high growth rates throughout the time horizon of the model can be observed when RES-E targets or incentive mechanisms are introduced. Although more expensive in total, a politically induced introduction of renewable electricity reduces the scarcity of CO2 emission allowances and lowers the marginal costs of CO2 reduction up to 30% in 2030, when ambitious targets are specified for this year. Despite the higher overall costs, a diversification of the energy resource base by RESE use is observed, as primarily natural gas and nuclear fuels are replaced. Although not economically valued in the electricity market, this is a bonus in terms of a weakened increase of the energy import dependency. More generally it can be concluded that the approach allows to assess electricity sector market developments under the interrelated effects of design options for various policy instruments and other framework conditions. This does not only concern the support of renewable electricity utilisation, but also e.g. the field of emission reduction policies. Further, the developed model instrument can easily be adapted to take into account other environmental policy measures, such as energy conservation and efficiency measures. The results of the analysis illustrate that the implemented modelling approach is a sophisticated and versatile tool, both for utilities to analyse future market developments, and also for policy planners in order to develop effective and efficient design options, especially for renewable electricity support. The developed model in the current version only represents the electricity sector. As the new Directive 2009/28/EC refer to the share of renewables in gross final consumption and not only for electricity, we aim to extend the model to other sectors, this means especially to include the heat and transport sector. Particularly, the heat sector offers a significant and not expensive potential for the use of renewable energies. Also in the transport

sector a potential for the use of renewables arises with the use of biomass and with the use of new technologies, such as electric mobility. Especially electric mobility will increase the interactions between the transport and electricity sector and thus necessitates an integrated model. The discussed extension of the model would allow using the model for the design and evaluation of the national renewable energy action plans. With this extension, the discussed modelling approach would be a sophisticated tool, especially for policy planners in order to develop effective and efficient design options for renewable electricity support.

Acknowledgements The paper results from the work of the young investigator group ‘‘New methods for energy market modeling’’ of Dr. Dominik ¨ Most. The young investigator group received financial support by the Concept for the future of Karlsruhe Institute of Technology within the framework of the German Excellence Initiative. We also thank Dr. Johannes Rosen, who developed a basis version of the model. References Agnew, M., Schrattenholzer, L., Voß, A., 1979. A Model for Energy Supply System Alternatives and their General Environmental Impact Laxenburg, IIASA. Antoniou, Y., Capros, P., 1999. Decision support system framework of the PRIMES energy model of the European Commission. International Journal of Global Energy 12 (1–6), 92–119. Brand, H., Barth, R., Weber, C., Meibom, P., Swider, D.J., 2005.Extension of wind power: effects on markets and costs of integration. In: Proceedings of the Fourth International Energy Economics Conference. Vienna, Austria. Brooke, A., Kendrick, D., Meeraus, A., Raman, R., 1998. GAMS—A User’s Guide (Edition December 1998). GAMS Development Corporation, Washington. Capros, P., Mantzos, L., Kolokatsas, D., Ioannou, N., Georgakopoulos, T., Filippopoulitis, A., Antoniou, Y., 1998. The PRIMES Energy System Model Reference Manual. National Technical University of Athen. Commission of the European Communities, 1999. European Union Energy Outlook to 2020. Commission of the European Communities, 2001. Directive 2001/77/EC of the European Parliament and of the Council of 27 September 2001 on the promotion of electricity produced from renewable energy sources in the internal electricity market. Brussels, Commission of the European Communities. Commission of the European Communities, 2006. World Energy Technology Outlook 2050 Office for Official Publications of the European Communities, Bruxelles. Commission of the European Communities, 2009. Directive 2009/28/EC of the European Parliament and of the Council of 23 April 2009 on the promotion of the use of energy from renewable sources and amending and subsequently

ARTICLE IN PRESS 2910

D. M¨ ost, W. Fichtner / Energy Policy 38 (2010) 2898–2910

repealing Directives 2001/77/EC and 2003/30/EC. Brussels, Commission of the European Communities. Dany, G., 2000. Kraftwerksreserve in elektrischen Verbundsystemen mit hohem Windenergieanteil. Aachen, Klinkenberg Verlag. ¨ die Netzintegration von WindDena, 2005. Energiewirtschaftliche Planung fur energie in Deutschland an Land und Offshore bis zum Jahr 2020 – Endbericht – Deutsche Energie-Agentur GmbH (dena). Department of Energy (DOE), 2008. Integrating Module of the National Energy Modeling System: Model Documentation 2008, Office of Integrated Analysis and Forecasting, Energy Information Administration, US Departement of Energy. European Renewable Energy Council (EREC), 2005. Renewable Energy Target for Europe—20% by 2020. Renewable Energy House, Brussels. Finon, D., 1974. Optimization Model for the Energy Sector. Energy Policy 6, 136–151. Fischedick, M., 1996. Erneuerbare Energien und Blockheizkraftwerke im Kraftwerksverbund. Technische Effekte, Kosten, Emissionen—Dissertation. Stutt¨ Energiewirtschaft und Rationelle Energieanwendung. gart, Institut fur Fishbone, L.G., Abilock, H., 1981. MARKAL: a linear programming model for energy systems analysis: technical description of the BNL version. International Journal of Energy Research 5/4, 353–375. Fraunhofer Institute Systems and Innovation Research (Fraunhofer-ISI), 2009. ADAM Adaptation and Mitigation Strategies: Supporting European Climate Policy, ADAM 2-degree scenario for Europe—policies and impacts. Karlsruhe, Fraunhofer Institute Systems and Innovation Research (Fraunhofer-ISI). Haas, R., Faber, T., Huber, C., Resch, G., 2001. Final Report of the project ElGreen—Organising a Joint European Green Electricity Market. Institute of Energy Economics, Vienna University of Technology, Vienna, Austria. ¨ OnshoreHeld, A., Krause, H., Ragwitz, M., 2009. Kostenpotenzialkurven fur Windstromerzeugung in der EU basierend auf einem geografischen Informa¨ tionssystem. In: Most, D., Fichtner, W., Grunwald, A. (Eds.), Energiesystemana¨ lyse Karlsruhe. Universitatsverlag Karlsruhe (TH). Holttinen, H., 2004. The impact of large scale wind power production on the Nordic electircity system. VTT Publications, Espoo 554. Huber, C., Green, J., Cleijne, H., Morthorst, P., Gual, M., Ragwitz, M., Orasch, W., Bokemann, M., Lins, C., 2004.Green-X Final Report—Deriving optimal strategies for increasing the share of RES-E in a dynamic European electricity market. Vienna, Austria. IEA, 2008. World energy outlook 2008. International Energy Agency. Klobasa, M., Ragwitz, M., 2004. Final Report: Large Scale Integration of Renewable Energy Resources into the EU Energy System. Karlsruhe, Fraunhofer ISI. ¨ Kramer, M., 2003. Modellanalyse zur Optimierung der Stromerzeugung bei hoher Einspeisung von Windenergie. Fortschritt-Berichte VDI. ¨ Leonhard, W., Muller, K., 2002. Balancing fluctuating wind energy with fossil power stations—where are the limits? Electra 204, 12–18. Lux, Rainer, 1999. Auswirkungen fluktuierender Einspeisung auf die Stromerzeu¨ Stuttgart, gung konventioneller Kraftwerkssysteme. 1. Stuttgart, Universitat Dissertation.

Messner, S., Schrattenholzer, L., 2000. MESSAGE-MACRO: lining an energy supply model with a macroeconomic module and solving it iteratively. Energy 25, 267–282. ¨ Most, D., Perlwitz, H., 2009. Prospects of gas supply until 2020 in Europe and its relevance for the power sector in the context of emission trading. Energy 34 (10), 1510–1522. ¨ Obersteiner, C., Faber, T., Resch, G., Auer, H., Pruggler, W., 2006. Modelling LEastCost RES-E Grid Integration under different Regulatoroy Conditions based on the Simulation Software GreenNet. Deliveralbe of the project GreenNet-EU27 ‘‘Guiding a least cost grid integration of RES-Electricity in an extended Europe’’. Vienna University of Technology, Vienna, Energy Economics Group (EEG). Platts, 2005. World Electric Power Plants Data Base (WEPP) Platts, Division of The McGraw-Hill Companies. Ragwitz, Mario, Held, Anne, Resch, Gustav, Faber, Thomas, Haas, Reinhard, Huber, Claus, Coenraads, R., Voogt, M., Reece, G., Morthorst, P.E., Jensen, S.G., Konstantinaviciute, I., Heyder, B., 2006. Assessment and optimisation of renewable energy support schemes in the European electricity market. Intelligent Energy, Europe. ¨ Remme, U., 2006. Zukunftige Rolle erneuerbarer Energien in Deutschland: ¨ Sensitivitatsanalysen mit einem linearen Optimierungsmodell, Dissertation, ¨ Stuttgart. Universitat Rosen, J., 2007. The future role of renewable energy sources in European electricity ¨ supply: a model-based analysis for the EU-15. Karlsruhe, Universitatsverlag. ¨ Rosen, J., Tietze-Stockinger, I., Rentz, O., 2007. Model-based analysis of effects from large-scale wind power production. Energy 32 (4), 575–583. Sensfuß, F., Ragwitz, M., Genoese, M., 2008. The merit-order effect: a detailed analysis of the price effect of renewable electricity generation on spot market prices in Germany. Energy Policy 36 (8), 3086–3094. Sontow, Jette, 2000. Energiewirtschaftliche Analyse großtechnischer Windstro¨ Energiewirtschaft und merzeugung (Dissertation). Stuttgart, IER Institut fur Rationelle Energieanwendung. Uyterlinde, M.A., Daniels, B.W., de Noord, M., de Vries, H.J., de Zoeten, C., Skytte, K., Meibom, P., Lescot, D., Hoffmann, H., Stronzig, M., Gual, M., del Rio, P., Herna´ndez, F., 2003. Renewable electricity market developments in the European Union—final report of the ADMIRE REBUS project. ¨ Uyterlinde, M.A., Martinus, G.H., Rosler, H., Kouvaritakis, N., Panos, V., Mantzos, L., Zeka-Paschou, M., Kypreos, S., Rafaj, P., Blesl, M., Ellersdorfer, I., Fahl, U., Keppo, ¨ ¨ I., Riahi, K., Bohringer, C., Loschel, A., Sano, F., Akimoto, K., Homma, T., Tomoda, T., Pratlong, F., Le Mouel, P., Szabo, L., Russ, P., Kydes, A., 2005. The contribution of renewable energy to a sustainable energy system - Volume 2 in the CASCADE MINTS project - ECN-C-05-034. Van der Voort, E., Donni, E., Thonet, C., Bois D’Enghien, E., Dechamps, C., Guilmot, J.F., 1984. Energy Supply Modelling Package, EFOM-12C Mark I, Mathematical Description Louvain-La-Neuve. Voogt, M.H., Uyterlinde, M.A., de Noord, M., Skytte, K., Nielsen, L.H., Leonardi, M., Whiteley, M., Chapman, M., 2001. Renewable energy burden sharing REBUS (project report ECN-C-01-030), Petten, ECN.