The basis of the method applied in this work lies firstly in a comprehensive database, the Chalmers Power Plant Database (Chalmers PP Db) describing all existing power plants across the European Union, Norway, Iceland and Switzerland; secondly, in a solar model, DCS-CHP (Distributed Concentrating Solar Combined Heat and Power), which creates simulated output for many types of solar systems based on typical weather data over a grid of Europe; thirdly, in an electricity-supply system model, ELIN (ELectricity INvestment), which links the existing capacity found in the Chalmers PP Db to investments in new generation capacity to the year 2050. The ELIN model calculations include EU-27, Norway and Switzerland.
2.3. The Electricity Investment Model
The ELIN model is a long-term dynamic optimization model (originally formulated by Odenberger et al. [12
] and further developed by Goransson et al. [13
], where these references include full model formulation) that includes the present generation system, as derived from the Chalmers PP Db, together with an extensive array of new and existing technologies that are to be used to meet the changes in future demand as existing capacity comes of age or becomes unprofitable. Both conventional fossil fuel and CCS technologies are available for investment, as well as a portfolio of renewable technologies.
The integration with the DCS-CHP model allows ELIN to have a variety of solar technology investments to choose from. The available solar technology models in DCS-CHP are based on bottom up thermodynamic and electrical modeling of five different technology classes (CSP, tracking High Concentration Photovoltaics (HCPV), non-tracking distributed PV, tracking utility PV and non-tracking utility PV) all with different costs and inherently different production profiles.
Wind technologies, on the other hand, are divided into two technology classes (onshore and offshore) with different costs and further divided into capacity factors based on the wind resources of the locations (the highest being a 35% capacity factor). However, all solar technologies in the same technology class and all wind technologies in the same technology class, respectively, are considered to have the same investment and O&M costs (see Figure A1
a,b). Therefore, the differences in the investments chosen depend entirely on the favorability of each class of technology in each location.
For solar and wind, the investment decisions are based on the available resource at specific locations in addition to the other variables affecting all generation investments such as the transmission capacity and electrical load in the respective countries. Solar production is also to a lesser extent affected by temperature, as the DCS-CHP model accounts for ambient temperature when calculating production profiles.
The ELIN model calculates the sizes and locations of investments, as well as the dispatch of new and existing capacity that minimize the total system cost over the entire modeled period. The time horizon of the ELIN model is from 2010–2050 with each discrete year separately described. The intra-annual time resolution of the ELIN model is 16 time steps, including two daily load segments (night load and day load), weekdays and weekends, which are allocated over four different seasons: winter, summer, spring and autumn. Typical model outputs from the ELIN model include capacity and production levels of electricity by fuel and country until 2050, aggregated investment costs, electricity trade between regions (or countries) and marginal costs of electricity. Generally in the model runs, a CO-emission cap, which is gradually reduced up to the year 2050, is imposed on emissions from the electricity production. Thus, the marginal cost of CO-emission reductions is also part of the model output.
A capacity constraint forces investment in peaking natural gas plants when variable renewables reach a high level in the model. The reason for this is that large capacities of wind and solar can have large short-term variations and hence are balanced with additional plants, which can ramp up and down quickly when needed (peaking natural gas is the most cost-effective of such plants). Although the magnitude of resulting investments in peaking gas are only indicative, this capacity constraint effectively puts a penalty on investments of wind and solar in the optimization process to compensate for the insufficient time resolution of the model.
In addition, distributed solar PV installations can be modeled to compete with retail prices or wholesale market prices (i.e., the relation of small-scale prosumers benefits from having a higher value of locally consumed power can be evaluated). The small-scale producer added value is implemented for PV electricity based on current national differences between retail and wholesale prices from Eurostat consumer price reports [14
], yet excludes the value added tax since this is not part of the investment cost of the installation. When running with this option, the model applies a net metering scheme for small-scale solar PV producers, meaning that all electricity generated by the PVs gains this added value.
Another model output in ELIN is the investments in electrical transmission capacities between countries. We assume that significant transmission investments across Europe are optional, i.e., the model decides endogenously whether they are profitable or not (based on exogenous assumptions of investments costs), from 2020 and onwards. The profitability of a new interconnector depends on whether the wholesale electricity-price difference between two countries is large enough to motivate such an investment.
The first transmission investments are allowed in the model year 2020, which gives the needed time for the construction of new interconnections. Note that these transmission investments are purely between countries (international) and exclude all interconnections within the same country (intranational). The model assumes furthermore the ability to transmit free of cost within a single country.
Highly detailed wind-power availability data across Europe are also included in the ELIN model. The data have primarily been taken from the ERA Interim dataset [16
] made available through the European Centre for Medium-Range Weather Forecasts [17
]. Although the data were originally defined for single spatial cells of 200–700 km2
and covering the entire EU-27, it has been aggregated to fit the ELIN regional model structure (53 intra-national regions defined by key electricity-transmission bottlenecks). Both the annual availability (full-load hours) and the production profiles for wind power have been implemented on a regional level. The estimated potential land availability for wind power, which is also an important model input parameter, is based on a detailed assessment of areas across Europe not suitable for wind power, i.e., densely-populated areas, or transportation infrastructure, waterways, seas or areas under environmental protection [18
]. Based on the remaining available land surface suitable for wind-power installations and on wind availability and investment costs that develop over time, cost-supply curves are generated for new onshore wind power that are used as input to the ELIN model.
Likewise, the land potential for solar is calculated similarly to wind (excluding nature protection areas, water, etc.), but does not exclude densely-populated areas (these areas are assumed suitable for distributed PV only), nor transportation infrastructure.
2.4. The Scenario
The climate market scenario, inspired by the European Commissions Roadmap scenario “Diversified supply technologies” [19
] and “Power choices reloaded” by Eurelectric [20
], investigates a future with focus on stringent CO
emission reduction after the year 2020. Thus, the modeling presented in this work includes current EU targets to 2020 and thereafter only annual CO
emission constraints up to 2050 reaching 50% reduction by 2030 and 93% by 2050 compared to 1990 levels. Up to the year 2020, the constraint development follows national projections reported in the National Renewable Energy Action Plans [21
]. The demand growth for electricity is implemented on a national basis resulting in a 0.91% increase per year on an average European level. This strong growth could indicate, for example, an increasing electrification of other sectors such as transport.
The technology costs (Appendix A.1
, Table A1
, Table A2
, Table A3
and Table A4
) in the base case climate market scenario are taken from the IEA’s World Energy Investment Outlook (WEIO) “New Policy Scenario” [22
]. Both battery and thermal storage technologies are included in the ELIN model with cost curves shown in Figure A1
a produced from extrapolations of [23
]. Other storage technologies such as pumped hydro have not been included due to various constraints including the model’s time steps and are therefore outside the scope of this analysis. Cost curves over time for wind and solar technologies are curve fitted based on the same WEIO data assumptions (Figure A1
a,b). The assumptions about the cost curves (exogenous) used have been compared with expected learning curves for solar PV, and we see fairly good agreement with the magnitude of investments seen here and the expected investments that would be needed to stay on the predicted cost/learning curves. Some of the important initial conditions given for the base case model run in ELIN are outlined in Figure A1
in Appendix A.1
The combustion technologies (including biomass) and nuclear, in contrast to wind and solar, are assumed to have constant investment costs throughout the model runs (i.e., the 2012 costs in WEIO), but increasing efficiency over time, as shown in Figure A1
c–e. The cost of fuel for these technologies is based on a cost-supply fuel curve that remains constant throughout the model where prices at every time step are based on the amount of consumed resources (as shown in Appendix A.2
). The difference in costs for the same technologies shown in Figure A1
c–e reflects the type of power plant (condensing, Combined Heat and Power (CHP) or a back-pressure plant coupled to industrial waste heat). The potential for investments in the comparatively efficient back-pressure plants is limited to current industrial levels in the ELIN model as they are available to be installed only in very specific industrial locations. A sensitivity run eliminating the possibility to invest in these back pressure plants shows little change in overall results (see S1
for all sensitivity results).