The REX model [34] is a greenfield optimization model for capacity investment and dispatch with respect to electricity generation, transmission, storage and demand response. It seeks at minimum cost portfolio for the future electricity system with an overnight investment approach. The objective of the model is to minimize the total annual electricity system costs, given the constraints of meeting electricity demand, renewable energy resource potentials and a CO₂ emission cap. The investment options for generation technologies, transmission and storage capacity are shown in Figure 1. For more details regarding the model, please refer to reference [34].

In this study, we model a future interconnected European electricity system for the year 2050 using an hourly time resolution with a CO₂ emission cap of 10 g/kWh, roughly equivalent to a 98% reduction in CO₂ emissions for the European electricity sector, as compared with the emissions in 1990. Since the study focuses on the northern part of Europe, we also provide a more detailed spatial resolution for this region. We assume all the regions are interconnected with high-voltage direct current (HVDC) transmission grids. The approximation is made based on the fact that many of the international connections are already controllable point-to-point HVDC connections and this trend is likely to continue in the future [35]. In addition, this assumption is consistent with the fact that the cross-border connections are treated with Net Transfer Capacities (NTCs) in market clearing process for many countries [36]. For more detailed assumptions see reference [34]. For the costs of different technologies, see reference [37].

Figure 1

Methods overview. CHP: Combined heat and power. OCGT: Open cycle gas turbine. CCGT: Combined cycle gas turbine. NG: Natural gas

In this study, a MATLAB-based linear programming model (Enerallt) is used to model the power market with an hourly resolution. Modelled region in this study consists of Sweden (SE), Finland (FI), Estonia (EE), Latvia (LV) and Lithuania (LT). Moreover, Sweden is modelled as one bidding area. Regions (bidding areas) that are interconnected with modelled regions are considered as external regions that are able to exchange electricity with the modelled regions depending on the marginal price of electricity and net transfer capacity. External regions include Norway (NO1, NO2, NO3, NO4, NO5), Denmark (DK1, DK2), Germany (DE) and Poland (PL). Due to data availability, electricity exchange with Russia and Belarus is assumed to be fixed at 2018 level. Generation mix in modelled regions is modelled as technology- specific. Included technologies are presented in the Appendix Table A. 1-Table A. 3. Economic data for different technologies are based on costs presented in [34], [35], [38]-[41] (Table A. 7). The objective function and constraints are presented in eqs. (1)–(8). The objective function (1) is to minimize short-term operation costs in the power sector (and in the district heating sector):

$\underset{p_{ij,t},p_{ijk,t},h_{ijk,t}}{\min }\left({\displaystyle \sum }_i{\displaystyle \sum }_j{\displaystyle \sum }_t{c}_{ij,t}{p}_{ij,t}+{\displaystyle \sum }_i{\displaystyle \sum }_j{\displaystyle \sum }_{k\ne j}{\displaystyle \sum }_t{c}_{ij,t}{p}_{ijk,t}+{\displaystyle \sum }_i{\displaystyle \sum }_j{\displaystyle \sum }_n{\displaystyle \sum }_t{c}_{ij,t}{h}_{ijn,t}\right)$ (1)

${\displaystyle \sum }_i{p}_{ik,t}+{\displaystyle \sum }_i{\displaystyle \sum }_j{p}_{ijk,t}-{\displaystyle \sum }_i{\displaystyle \sum }_n\frac{h_{ikn,t}}{{\eta }_i}=d_{k,t},\forall k\ne j,n,t$ (2)

${\displaystyle \sum }_i{\displaystyle \sum }_{k\ne j}{p}_{ijk,t}\le NT{C}_{jk,t},\forall j,t$ (3)

$0\le {\displaystyle \sum }_i{p}_{ik,t}+{\displaystyle \sum }_i{\displaystyle \sum }_j{p}_{ijk,t}\le P_{ij,t},\forall i,j,t$ (4)

$p_{ij,t}-p_{ij,t-1}\le r_{ij}{P}_{ij,t}$ (5)

${\displaystyle \sum }_i{h}_{ikn,t}+{\displaystyle \sum }_i{p}_{ik,t}{\theta }_{i,t}+{\displaystyle \sum }_i{\displaystyle \sum }_j{p}_{ijk,t}{\theta }_{i,t}=q_{kn,t},\forall j,k,t,{\theta }_{i,t}=\frac{H_{i,t}}{P_{i,t}}$ (6)

$0\le {\displaystyle \sum }_i{h}_{ikn,t}\le H_{ikn,t},\forall j,k,t$ (7)

$h_{ikn,t}-h_{ikn,t-1}\le r_{ikn}{H}_{i,t}$ (8)

In (1) p_ij,t represents the power supply of technology i in bidding area j, p_ijk,t is the power supply of technology i in bidding area j that is exported to bidding area k and c_ij,t is the short-term marginal cost of production for technology i in bidding area j. Moreover, in eq. (1) h_ijn,t represents the heat supply of technology i in bidding area j in district heating network n, c_ij,t, is the short-term marginal cost of heat conversion of technology i in bidding area j. In eq. (2) d_k,t is the electricity demand in bidding area k. Moreover, in eq. (2) h_ikn,t/η_i is the electricity demand of heat conversion technology i that is consumed in heat conversion process (e.g. heat pump). In eq. (3) NTC_jk,t is the net transmission capacity between bidding areas j and k. In eq. (4) P_ij,t is the available electricity generation capacity for the technology i in bidding area j. In eq. (5) r_ij is the ramping factor. In eq. (6) q_kn,t is the heat demand in the district heating network n in bidding area k. Moreover, in eq. (6) heat supply of CHP technology i in bidding area k is coupled to the power supply p_i,t by heat-to-power ratio θ_i,t. In eq. (7) H_ikn,t is the available heat conversion capacity for the technology i in district heating network n in bidding area k. In the equations, t is the hour index. A more detailed description of the model (including description of hydropower simulation) is presented in references [31], [42]. Moreover, validation of the model is presented in [42].

In this study, the REX and Enerallt models are combined by utilising the optimal capacity scenarios of REX as input to Enerallt. The REX scenarios are based on the conceptual framework proposed by Kan et al. [34]. Four different scenarios are analysed (Table 2). Nuclear power has been the subject of a long-running debate in Sweden [43]-[45]. Moreover, transmission expansion by 2050 may impact the results. Thus, these scenarios cover two hypotheses. In the first scenario (S1), nuclear power is removed from the electricity mix in Sweden, while transmission lines are fixed. In second scenario (S2), Sweden maintains the present amount of nuclear power and transmission lines are fixed. In the third scenario (S3), transmission lines are expanded and there is no nuclear power in Sweden. In the last scenario (S4), transmission lines are expanded and Sweden maintains nuclear power. Fixed transmission is the actual transmission capacity in 2014 and transmission expansion for 2050 comes from the REX model. The transmission capacities in the two variants are given in Appendix Table A. 4- Table A. 6.

The Enerallt model input, which comes from the REX model and includes capacity mix per country /region, is presented in the appendix (see Table A. 1-Table A. 3). We selected Finland, Sweden and the Baltic countries, for modelling at a country level. The Baltic countries are the target countries for this study, while Finland was included owing to its strong connection with Estonia and being a large producer and consumer, and Sweden was modelled because of the scenarios assessing the nuclear power decisions being made in the country.

Each of the above scenarios in Table 2 provides different electricity system costs. The cost components for the i country are provided in the following equation:

$C_i=I_i+F_i+V_i+E_i-R_i$ (9)

where C_i is the total annual electricity system cost, I_i is the annualized investment cost. The annualized investment cost is computed by considering a 5% discount rate. Since the transmission line capacities are symmetric it is assumed that transmission line investment costs are shared equally between bordering countries. F_i is the fixed operation and maintenance costs (O&M cost). V_i is the variable cost which includes the variable O&M cost and fuel cost. With regard to CHP production, fuel costs are allocated to heat and electricity using energy method and only fuel use in CHP electricity production is included in electricity system costs. In equation (9), E_i is the exchange cost that is the cost of importing electricity minus the revenue from exporting electricity. R_i is the congestion income:

$R_i={\displaystyle \sum }_t{\displaystyle \sum }_j\raisebox{1ex}{$\left|(P_{i,t}-P_{j,t})\times (Q_{ij,t}-Q_{ji,t})\right|$}\!\left/ \!\raisebox{-1ex}{$2$}\right.i\ne j$ (10)

where i and j are the countries trading electricity through transmission lines in hour t. P_i,t − P_j,t describes the area price difference between country i and country j. Q_ij,t − Q_ji,t is the commercial flow between two countries. Eq. (10) comes from the agreement between the Nordic TSOs that congestion income is shared in equal proportions [46]. Moreover, since transmission line capacities are symmetric between country i and country j [34], it is also assumed that investment costs for transmission line capacities are shared equally between country i and country j.

In order to compare the electricity system cost between countries, average cost is computed as follows:

$C_i{}^{\text{ave}}=\raisebox{1ex}{$C_i$}\!\left/ \!\raisebox{-1ex}{$D_i$}\right.$ (11)

where D_i is the demand for the country i.

The economic input data is provided by Kan et al. [34]. Total investment and fixed O&M costs are estimated based on the generation capacity and transmission line capacity data from REX model [34] (Table A. 1-Table A. 6) and technology-based investment and fixed O&M costs presented in [35], [38]-[41] (Table A. 7). Variable costs are estimated based on the electricity production data from the Enerallt model and fuel price and technology-based variable O&M cost presented in [34]. For CHP, the energy method is used to allocate costs for heat and electricity. However, only the electricity part is included. Exchange cost and congestion income come from the Enerallt model. In the scenarios of the Enerallt model, carbon dioxide emission trading price was assumed 30 €/ton CO₂. However, there was practically no fossil electricity generation.

Figure 2 shows electricity production in different scenarios. Electricity production data from 2017 is also shown to provide a comparison with the current situation [12]. Data for current situation of Estonia comes from [47].

Figure 2 shows that transmission expansion would affect electricity production significantly, causing the region to become more dependent on wind power. This level of dependency is less significant in Finland, as nuclear power capacity is considered constant. It is notable that in S3 and S4, no investment is made in solar PV, natural gas and biogas generators. Wind power production especially in the Baltic countries increases income from the electricity trade. Other policies in neighboring countries may affect the optimal capacity mix, which derives from the REX model (see Appendix Table A. 1-Table A. 3).

Electricity production in Lithuania increases significantly in all four scenarios compared to 2017. The electricity generation in Lithuania was 4.2 TWh in 2017. This amount is 12 TWh in S1 and S2, 27 TWh in S3 and 18 TWh in S4.

Figure 2

Selected countries' electricity production in the scenarios studied. The situation in 2017 is presented for comparison. Electricity demand does not change between scenarios

Figure 3 presents detailed itemization of costs in different scenarios for the studied countries. Total system costs include exchange costs and congestion income. Average electricity system cost for the studied five countries is in S1 417 €/MWh, in S2 415 €/MWh, in S3 297 €/MWh and in S4 341 €/MWh. Thus, transmission expansion and phasing out nuclear power in Sweden cause the lowest total system costs for the whole region considered here. For the Baltic countries and Finland, the transmission expansion scenarios entail less costs. For Sweden, cost decreases through transmission expansion while nuclear power is phased out. However, for Sweden the highest cost occurs in S4, where a significant investment cost is needed to provide a huge amount of nuclear and wind capacity.

Figure 3

Electricity system cost components in selected countries. (ATESC is average total electricity system cost)^†

The impacts of these different scenarios on electricity prices are presented in Figure 4. As shown in Figure 4, in S1 and S2, where transmission is fixed, no significant increase in prices was detected. By expanding the transmission line connection in the Nordic-Baltic electricity market and Poland, the scenario with nuclear power in Sweden (S4) has lower electricity prices than does the case where nuclear power in Sweden is decommissioned (S3). This decrease is 6%, 7% and 8% in Estonia, Latvia and Lithuania respectively. In transmission expansion scenarios (S3 and S4), Estonia has the lowest electricity prices (67 €/MWh in S3 and 63 €/MWh in S4), while Lithuania has the highest (74 €/MWh in S3 and 68 €/MWh in S4). In these scenarios, Lithuanian prices are even higher than Sweden (72 €/MWh in S3 and 66 €/MWh in S4) and Finland (72 €/MWh in S3 and 64 €/MWh in S4).

Unlike the Baltic countries, Sweden and Finland have the lowest prices in S4. This can be due to the large number of wind power installations in all countries and nuclear power in Sweden, which affect electricity price decreases in Finland and Sweden compared to the scenarios where transmission is fixed. In S4, maintaining the existing amount of nuclear power brings an opportunity for Sweden to obtain more income from exporting electricity.

Figure 4

Demand-weighted average prices according to the Enerallt model. Situation in 2017 is presented for comparison

Figure 5 provides an overview of import and export between the selected countries and the neighbouring countries. From the Figure 5, we can see that trade appears to be unaffected by Sweden's nuclear power policy when the transmission is fixed (S1, S2). In the transmission capacity expansion scenarios (S3, S4), all countries become net exporters. From S3 to S4, where transmission lines are developed, nuclear power in Sweden decreases the amount of trade in the other countries, especially in Lithuania (8.84 TWh). However, Sweden then exports more electricity. What is interesting about the last scenario is the trade between Lithuania and Poland. Lithuania mainly exports electricity to Poland in the capacity expansion scenarios (S3, S4). However, this trade decreases in the last scenario. Sweden is also exporting electricity to Poland. In S4, Sweden's net export to Poland is 51 TWh. It seems that in S4, where Sweden has nuclear power and transmission lines are expanded, it is less profitable to invest in wind power capacity in the Baltic countries. Thus, Sweden provides more electricity to Poland and the trade between Lithuania and Poland decreases. Through transmission expansion, Finland trades more with its neighbours. Finland imports more electricity from Estonia and Norway, while exporting electricity mostly to Sweden.

Figure 5

Electricity trade between the selected countries according to the Enerallt modelling (Finland (FI), Sweden (SE), Estonia (EE), Latvia (LV) and Lithuania (LT)) and the neighbouring countries (Norway (NO), Denmark (DK), Germany (DE), and Poland (PL)). Positive values mean that electricity is imported from aforementioned countries, and vice versa, negative values mean electricity is exported

In order to analyse countries dependency on importing electricity, hourly time series of trade have been studied. Hours can be categorized into peak, off peak1 and off peak2. Peak refers to the period from 8 am to 8 pm. Off peak1 refers to the time period from midnight to 8 am. Off peak2 refers to the time period from 8 pm to midnight. This classification is based on Nord pool spot [48]. In transmission expansion scenarios (S3 and S4), all countries have less net import. In these scenarios, during peak hours and off peak 1 hours, maintaining nuclear power in Sweden causes more importing for the Baltic countries. Maintaining nuclear power in Sweden in S4 reduces dramatically the hours that Sweden is net importer. In S4, Sweden is net importer during 5% of hours, while this amount is 47% for Estonia, 30% for Latvia and 35% for Lithuania.

As shown in Figure 6, in S1 and S2, where transmission is fixed, nuclear power in Sweden cannot affect the amount and expected income from the electricity trade. In these scenarios, in all countries except Latvia, the electricity they import from other countries is generally more expensive than their income from export.

Transmission line expansion has a significant impact on revenue in the Baltic countries, especially in Lithuania, due to the large number of wind energy installation with large electricity generation capacity, which provides countries with an opportunity to trade electricity. Closer inspection of the Figure 6 shows that by expanding transmission lines, nuclear power removal in Sweden could increase revenue from export in all countries. Revenue from export is decreased in the Baltic countries when Sweden retains its nuclear power and new transmission lines are built between countries. This decrease is 38%, 26% and 46% for Estonia, Latvia and Lithuania, respectively. This can be due to a lower in wind power capacity in the Baltic region in S4 compared to S3, which results in less income from exporting electricity to neighbouring countries.

Figure 6

Revenue from exporting and cost from importing electricity. Positive values mean that electricity is exported to aforementioned countries and the revenue is earned, and vice versa, negative values mean electricity is imported. Situation in 2017 is shown for comparison

Figure 7 illustrates CO₂ intensity rates in the studied countries. For CO₂ intensity, CO₂ emission is divided by the total power production per fuel source [49]. No significant differences were found in the CO₂ intensity rates of countries in S1 to S2 or S3 to S4, where only nuclear power was changed. Transmission expansion affects the Lithuanian CO₂ intensity significantly, as natural gas would not be needed. In S3 and S4, Finland, Estonia and Lithuania all generate electricity mostly with renewables, while Latvia switches 100% to renewables, resulting in further decrease in CO₂ intensity rates in the Baltic countries. The reader should note that in the REX model, where the generation capacity scenarios stem from, at European level the CO₂ emissions were the same in scenarios S1-S4.

Figure 7

CO₂ intensity in the studied countries and scenarios Finland (FI), Sweden (SE), Estonia (EE), Latvia (LV), and Lithuania (LT)

The results for different scenarios are summarized in Table 3. Although the nuclear power policy is a national policy for Sweden, it affects the Baltic countries and Finland. If transmission capacity is fixed, allowing nuclear power in the future Swedish electricity system has little impact on the average electricity system cost for the Baltic countries. However, the inclusion of nuclear power in Sweden might have large impacts on the trade revenues for the neighbouring countries. Specifically, Estonia, Latvia, Lithuania and Finland face a 35%, 11%, 2%, 2%, and 35% drop in income from exporting electricity respectively, as compared to the case where nuclear power is not allowed in Sweden. Thus, Sweden's nuclear power policy on its neighbours' income is considerable even with the current amount of transmission connections.

In the case of optimal transmission expansion, the availability of nuclear power in Sweden enables a significant increase in revenue from electricity trade for Sweden as compared to no nuclear in Sweden. Meanwhile, the trade income for Finland and the Baltic countries is reduced. The average electricity system cost for Baltic countries is higher in S1 than in S3, while the electricity price shows an opposite pattern. In S4, the Baltic countries have less installed generation capacity compared to S3. However, the exchange cost and congestion income have affected the average electricity system cost in these countries. Thus, there could be a possible trade-off for the Baltic countries. In S3 when there is no nuclear power in Sweden, Baltic countries may experience higher prices. However, these higher prices are provided by lower cost due the electricity trade opportunity. In S4, they experience lower prices while they have less opportunity to trade electricity and this can bring about more costs. The main reason is that, in S3, nuclear power is not installed in Sweden, the Nordic and Baltic countries rely more on each other for electricity trade to share variable renewable energy and dispatchable resources, as compared with S4 where nuclear power is installed for Sweden. The relatively high income from electricity trade for the Baltic countries offsets the investment cost in S3. In addition, the exclusion of flexible nuclear power in the Swedish electricity system results in less flexibility for the highly renewable regional electricity, which drives up the electricity price.