The electricity production of hydropower plants connected to reservoirs can be controlled rather well and they are able to provide flexibility. The total installed capacity of these types of power stations adds up to about 35 GW. Furthermore, the available hydro storage sums up to about 125 TWh, whereas the discharge time for some of the biggest reservoirs is several years.

The actual flexibility is determined by the concurrent demand level in the Nordic area. In addition the operation of the hydropower system is limited by discharge constraints as well as maximum and minimum reservoirs levels due to environmental regulations. Furthermore, these flexibility estimations can be limited by the bottlenecks in the transmission system within the Nordic region or between the Nordic region and other countries.

The increase in flexibility is primarily expected to be provided by the increase in generation capacity. NVE² estimates a possible increase of 16.5 GW in western and southern Norway (NVE, 2011). The Centre for Environmental Design of Renewable Energy (CEDREN)³ estimates a potential increase of up to 18.2 GW in southern Norway, without violating current environmental constraints (Solvang et al., 2012). These increases solely include the construction of new power plants, pumps as well as water ways, but no new reservoirs.

The model is a fundamental optimisation model for the mid- and long-term simulation of hydrothermal power systems. The market model has its strength in the strategic exploitation of hydro energy stored in reservoirs. Stochastic dynamic programming (SDP) is employed to calculate water values. The optimisation is stochastic due to the natural variation in climate variables such as temperature, wind speed and inflow, and dynamic since the utilisation of hydro reservoirs couples dispatch decisions in time. The water values reflect the expected marginal value of the water stored in the reservoirs, when substituting hydro power with other production sources (Wolfgang et al., 2009). To represent the seasonal as well as annual variation of renewable energy sources (RES), several climatic years (75 years in this case) are taken into account. At the same time a high temporal resolution is used in the model in order to account for the short-term variation of RES. Fig. 16.3 shows the aggregated inflow of all 75 climatic years to the Norwegian hydropower system.

In addition to the renewable energy sources, thermal power production is modelled by individual power plants. The inputs to this model include the generation capacity and marginal cost for thermal production, as well as wind production, solar production, electricity consumption, transmission capacity, and information about historical climate variables such as inflow, wind speed, solar radiation and temperature.

The model is divided into several market areas, based on a transport model. The exchange between areas is modelled as transport corridors with a capacity equal to the net transfer capacities (NTCs) between areas. A more detailed description of the model can be found in Wolfgang et al. (2009), while the underlying dataset is described in Jaehnert and Doorman (2014).

The tool used in this simulation step is SINTEF's Power System Simulation Tool (PSST) (Korpås et al., 2007; Farahmand, 2012). PSST considers a detailed grid model and computes the optimal generation dispatch and flow along transmission lines for each hour of the simulation year. The optimal solution for each hour is found by minimising the operating cost that essentially expresses the cost of generation based on different marginal generation costs of available power stations. The scenarios and data for generation portfolio and demand in PSST are consistent with those of the EMPS model.

The distinguishing feature that makes the methodology different from the existing DCOPF algorithm is the modelling of hydro generation and reservoirs. Despite the constant marginal cost of thermal generators, the marginal cost of hydro units (water values) is dependent upon the reservoir level and thus on the amount of water that is available for energy production. The hydro generation is coupled in time by the use of water values and reservoir hydro levels, and PSST is run for the entire year (8760 consecutive hours). This enables the simulation to capture the effect of inflow variation and reservoir hydro level variation related to the hydro generators. At each simulation step, the hydro levels of reservoirs are updated according to the generation and pumping level in the last period and inflow scenario in that time period. The water values are determined based upon the reservoir level and the time of the year. As explained in EMPS model, water values are imported as exogenous inputs from the EMPS model simulation, which is targeted to strategic utilisation of available hydro resources.

The model encompasses many countries in Continental Europe listed in Table 16.3.

In order to achieve representative simulation results, a benchmark scenario (2010) is calibrated to represent the generation mix reported from ENTSO-E.⁵ A comparison of the generation mix from the calibrated model and the reported data is shown in Fig. 16.5. The calibration is performed by adjusting the availability factors for the different technologies of thermal power plants. See Jaehnert and Doorman (2014) for further explanations of the calibration process. Apart from Germany, the figures are rather consistent. The difference in Germany is due to the utilisation of the power production on all grid levels in the market model, while ENTSO-E solely reports that for the transmission grid level.

The vast majority of the future offshore wind power will be located in the North and Baltic Seas, hence concentrated in a relatively small geographical area. An illustration is given in Fig. 16.6, including the wind farms planned by 2030 (blue).

The installed capacities, per country, are given in Table 16.4 (Cutululis et al., 2012a)

In the CEDREN study, it is assumed that no new reservoir is built, whereas the new power stations are constructed with new tunnels between existing reservoirs. Scenario 1 (11 GW) includes 12 new power stations, where five are “pump storage power stations” with a combined installed capacity of 5.2 GW and the remainder are “hydro storage power stations.” The pump storage power stations have reversible pump turbines, pumping water between two reservoirs, while hydro storage power stations are not equipped with such pump turbines.

One of these potential expansions is located in Tonstad in Norway, where a new pump storage power station is expected. Two cases were analysed in connection with Tonstad as follows:

The new tunnels in these two cases and associated reservoirs are shown in Fig. 16.7. The detailed results for the water level regulation in the upstream and downstream reservoirs as well as all other cases are explained in Solvang et al. (2012).

For the analysed power system scenario of 2030, scenario 1, with an increase of 11.2 GW of hydro generation capacity, is chosen. It is distinguished between new power generation and pumping capacity.

The expansion in the Norwegian hydropower system is solely an expansion of generation and pumping capacity, with no additional inflow. However, the hydro expansion in Sweden is expected to be in the form of small-scale hydropower facilities. The hydro generation capacity in Sweden is expected to increase by 1 GW. No expansion of hydropower production is assumed for Finland or Great Britain.

The PSST model likewise distinguishes between the expansion of installed production capacity and the expansion of pumping capacity. When the pumping capacity is increased, the hydro production capacity is increased by the identical capacity. The expanded capacity is proportionally divided among generators in one water course based on their capacity. The overview of capacity expansion is presented in Table 16.5 (Solvang et al., 2012).

The water values are used as the marginal production cost for hydro power. Given these production costs as the optimal strategy, the hydropower system is used to balance fluctuating energy inflow, in the short term as well as in the long term.

Figs 16.10 and 16.11 show the accumulated reservoir handling (reservoir trajectories) and hydropower production in Norway for all hours in all 75 climatic years. Plotted are the percentiles for all 75 climatic years in order to display the annual variability.

In the 2010 scenario, the reservoir handling (Fig. 16.10a) is characteristic for the Nordic countries, with depletion during the winter and early spring as well as filling during the rest of the year. Assessing the reservoir handling for the 2030 scenario (Fig. 16.11a) shows minor differences. It can be observed that the reservoir levels become higher in general, while the seasonal reservoir storage capability is utilised less. This means that percentiles of the reservoir handling become more spreadout and flatter. It also indicates a change of the reservoir utilisation from a longer-term perspective as compared to shorter term.

The aggregated hydropower production for Norway illustrates that significant changes in the hydro production pattern can be expected. In 2010 (Fig. 16.10b) there is a rather stable seasonal production trend, with higher production during winter and lower production during the summer, according to the demand in the Nordic area. In addition there is a diurnal pattern, resulting from the differences in demand during day and night as well as the weekend.

The stable seasonal pattern vanishes and more volatile hydropower production occurs in the 2030 scenario (Fig. 16.11b). These changes are due to the significant integration of wind power production in the future power system.

In 2010 (Fig. 16.12), low prices in the north and high prices in the south are observed, with two minor exceptions, southern Sweden and eastern Germany. The congestion rent is highest on the interconnectors between the Nordic area and Continental Europe, especially on the NorNed cable (about 60 €/kW per annum). Rather low area prices can be observed in the UK, which is due to a high generation capacity of base load power plants.

In the 2030 scenario (Fig. 16.13), prices increase significantly in Continental Europe, while there is a certain price decrease in the Nordic area and the UK. This price development is due to the decommissioning of a high share of thermal generation capacity in Continental Europe. On the other side large capacities of wind power productions are expected to be commissioned in the Nordic area (Sweden) as well as the UK (Scotland).

The connections between the Nordic area and Continental Europe are still most congested, whereas the marginal profit is more than doubled compared to 2010. In addition, increasing congestions can be observed on the corridor between England and Scotland. These large marginal profits on the transmission corridors show the need as well as the potential for a further expansion of those transmission corridors.

The development of the average electricity prices in the future scenarios is mainly influenced by two factors, the additional power production from RES and the expected increase of production costs of thermal power plants.

In order to study the optimal offshore grid topology, the optimal offshore grid topology introduced in IEE-EU OffshoreGrid project (Woyte et al., 2011) is considered as a basis. The increased capacity of hydro power in southern Norway is considered in our simulation for 2030 (Solvang et al., 2012). In addition to the assumptions in the IEE-EU OffshoreGrid project (Woyte et al., 2011), it is important to consider further alternatives for offshore grid topologies in order to exploit the potential additional hydropower production in an optimal way. In this respect, three different offshore grid structures are considered, including:

Fig. 16.14 shows different offshore grid alternatives in this study. “Case A” is the design proposed in IEE-EU OffshoreGrid project. A rather limited hydro capacity expansion was considered in Norway by 2030 in Woyte et al. (2011). However, the scenarios for hydro capacity expansion considered in Solvang et al. (2012), motivate the introduction of new offshore grid alternatives “Case B” and “Case C”. In “Case B”, there is a direct connection between the Norwegian offshore node and the onshore grid in the Netherlands, which allows transfer of hydropower flexibility and offshore wind power directly to the onshore grid. In “Case C” the offshore grid is closed in a loop between Norway, Great Britain and Germany offshore nodes. In this offshore grid configuration, power flows circulate among the countries around the North Sea. This configuration facilitates circulation of power, use of hydropower flexibility from the Norwegian system and penetration of installed offshore wind power to the onshore grid.

In order to study the effect of onshore constraints, we superimpose scenarios for onshore grid constraints on the proposed offshore grid alternatives. The scenarios for onshore grid constraints are clustered in three categories including:

The “NC” scenario represents the case without internal constraints in the EU grid. In this case, the internal grid in each country is considered as a copper plate, whereas cross-border grid constraints are considered by limiting the transmission capacity for each individual cross-border transmission line and NTC values for the corridors linking countries and areas inside Germany. The NTCs used between areas inside Germany are taken from the dena Grid study-II (dena, 2010).

The “IC” scenario considers the onshore grid with today's internal grid limitations in the German, the Dutch, the British and the Scandinavian systems.

The “ICE” scenario represents a situation, where the grid in “IC” scenario is expanded. In Norway, the expansion includes the grid reinforcements according to Statnett's Network Development Plan (Statnett, 2013), which includes a transmission corridor connecting the proposed hydropower stations located in southern Norway from the CEDREN study (Solvang et al., 2012). In central Europe the grid is expanded according to the Ten-Year Network Development Plan (ENTSO-E, 2012).

Table 16.6 presents the annual operating cost taking into account the proposed onshore constraints (“NC”, “IC” and “ICE”), and proposed offshore grid cases. The onshore grid constraints in the German and the Dutch systems limit the transmission of wind and hydro production to the load centres inland, hence increasing the operating cost. The comparison indicates that “NC” is always the solution with the lowest operating cost since the highest penetration of offshore wind is possible into the system. In this scenario, “Case B” of the offshore grid (a direct connection between the Norwegian offshore node and the onshore grid in the Netherlands) results in the lowest operating cost.

For “IC” and “ICE” scenarios, the results show a different picture with respect to offshore grid topologies. The optimal offshore grid topology appears to be “Case C”, where the offshore grid is closed in a loop between Norway, Great Britain and Germany offshore nodes.

According to investment estimation in L'Abbate and Migliavacca (2011), the investment cost for offshore HVDC grid is expected to be 1700 €/(MW km). The distance between the Norwegian and the German offshore wind farms is roughly assumed to be 350 km. The distance between the Norwegian offshore wind farm and Eemshaven in the Netherlands is assumed to be approximately 500 km. The connection capacity is expected to be 1000 MW. In this respect, the investment costs for the extra tie-line in “Case B” and “Case C” with respect to “Case A” are approximated to be 850 MEUR and 600 MEUR, respectively. Taking into account 100 MEUR for the costs of DC/AC converters, switchgears and transformers (L'Abbate and Migliavacca, 2011), the overall cost of the interconnections are roughly expected to be 950 MEUR and 700 MEUR for “Case B” and “Case C”, respectively. The life time factor (LFT) for the lifetime of 30 years and 5% discount rate is ∑n=130(1(1+5%)n)=15.3725. This factor allows a comparison of the operational savings accumulated throughout the life time of grid project, which is annual savings × 15.3725. The accumulated savings for “NC-Case B”, “IC-Case C” and “ICE-Case C” are ∼1225 MEUR, ∼936 MEUR and ∼1005 MEUR, respectively. The reference offshore case for this saving is “Case A” (original offshore grid according to the IEE-EU OffshoreGrid project). Comparing these numbers with investment for all expanded corridors indicates that all accumulated savings are greater than investment cost for each corridor. Therefore, it turns out that making investments in those corridors is profitable in comparison with the accumulated savings throughout the lifetime of the corridors.

Fig. 16.15 illustrates the correlation between the Tonstad pumping pattern and the German offshore wind production at wind facilities connected to NordLink HVDC cable. The figure shows that pumping at the Tonstad hydropower station is highly correlated with the variation in wind production at DanTysk and NordseeOst. These results illustrate that in the future power system with a large penetration of wind energy and high level of interconnections, the pumping strategy in the Nordic region will not only be influenced by seasonal inflows but also by the variability of wind production around the North Sea in case of offshore interconnectors in the form of point-to-point HVDC links between countries and a meshed offshore grid.