IV. ESTRATEGIA METODOLÓGICA
4.4 La investigación cualitativa como abordaje metodológico
The assumed wind power capacity for the simulations is displayed in Table 9.1. The modelling routine of the WPP model was described in Chapter 3. The assumed 2020 wind power capacity is based on the expert knowledge of the associated regional wind energy associations polled within the TradeWind EU project [16]. Theemployed 2020 scenario is in accordance with the 2020 high scenario in the TradeWind report, containing nearly a triplication of the currently installed WPP capacity.
Table 9.1: Installed WPP capacity [MW] 2010 & 2020
Areas 2010 2020 Norway 545 6600 Sweden 1250 10 000 Finland 350 3000 Denmark 3700 6000 Germany 24 900 57 300 Netherlands 1000 2950 Belgium 2800 10 400 Sum 34245 96 250
9.2.2
Forecast Error
In a system with a high penetration of WPP, production forecasts are necessary to schedule the production of conventional production units in order to assure a stable operation at all times.
Table 9.2 displays the MAE and NMAE for the 2010 and 2020 scenarios including different forecast horizons. Due to the highly sophisticated NWP models, the system-wide 24 hour NMAE only amounts up to about 3.7% for the 2020 scenario. Correspondingly, a further increase in accuracy is noticeable for the 3 hour forecast, reducing the NMAE to about 0.7%. Even though the values for MAE and NMAE provide a rather optimistic view on the wind forecast error and the required regulating resources in the system, Figure 9.2 displays that there are considerably high wind forecast errors, which especially occur before and after storm fronts.
9. Integration of regulating power markets
Table 9.2: Forecast error
2010 2020 MAE 3 h [MW] 218 883
NMAE 3 h [%] 0.6 0.9 MAE 24 h [MW] 915 3596
NMAE 24 h [%] 2.6 3.7
Figure 9.2 displays the hourly forecast errors for 3 and 24 hours ahead, indicating the improved forecast over a descending time period and thus the significantly reduced need for balancing WPP. The largest deviations in the 2020 high wind scenario reach an absolute value of about 40 GW. This illustrates the challenges the system is confronted with upon the addition of large amounts of intermittent wind power. Even though, the 3 hour forecast MAE is relatively small, the hour to hour variations may rise up to 10 GW (see Figure 9.2 and Figure 9.3). -40 -20 0 20 3h
Wind forecast error 2020 [GW]
0 2000 4000 6000 8000 -40 -20 0 20 Time [Hours] 24h
Figure 9.2: Forecast error of WPP in GW - 3 and 24 hours ahead Figure 9.3 shows the WPP 3 and 24 hour forecast error duration curves for the 2010 and 2020 scenarios. It shows the overall increase of the forecast error as well as a significant increase in the maximum forecast error, as seen at both ends of the curves.
Based on WPP forecasts, producers are able to identify their approximate wind power production and reschedule the preliminary production portfolio con- sidering the technical constraints of thermal power plants. Nevertheless, this requires a functioning intra-day market which gives power producers the possi-
9.3. Market Model 0 2000 4000 6000 8000 -30 -15 0 15 30 Time [Hours] WPP forecast error [GW] 2010 3h 2010 24h 2020 3h 2020 4h
Figure 9.3: WPP forecast error duration curve for the simulated area
bility to balance their production portfolio based on updated WPP forecasts. The simulations using the 24 hour WPP forecasts therefore represent a sce- nario without an intra-day market, while in the simulations using the three hour forecasts it is assumed that wind power producers are able to balance their production portfolio either by re-dispatch or in the intra-day market up to 3 hours before real-time. This reduction of the forecast horizon assumes higher flexibility of the system and will lead to more trading in the intra-day market.
9.3
Market Model
The applied EMPS market model has beend described in Chapter 4.
9.3.1
Market data
For the 2020 scenarios, all the system parameters such as the power plant port- folio, the inter-area transmission capacity and the WPP are updated to incor- porate the supposed system expansion in the upcoming years.
Due to reinforcements and the commissioning of new connections between the Nordic area and Continental Europe, particularly the extension of the Sk- agerrak cable, the Nordlink and as well as the NorNed connection, the estimated transmission capacity will be expanded from the existing 3700 MW up to 6800 MW [118].
Furthermore, the expected increase in hydro production capacity, especially in southern Norway, provides additional balancing resources. The increase of production capacity is done for single power plants and sums up to about 6 GW
9. Integration of regulating power markets
in total.
Due to the variability of WPP, it is necessary to adapt the reserve require- ments to the increased WPP system penetration to be able to provide suffi- cient balancing power. The adjustment of the reserve requirements done in this chapter is based on the 3 sigma approach as discussed in [106] and [119]. The estimation of the reserve requirements is based on the 3 hour WPP forecast error.
Table 9.3: Reserve Requirements [MW] 2010 & 2020
2010 2020
Areas pos. neg. pos. neg. Norway 1200 1200 1485 1485 Sweden 1220 1220 1950 1950 Denmark 1200 1200 1510 1510 Germany 3010 2045 6720 5755 Netherlands 300 300 1330 1330 Belgium 150 150 465 465 Sum 7080 6115 13460 12495
For the 2010 scenario, the reserve requirements for the countries modelled are based on the current requirements set [64]. These are adjusted based on the previously mentioned 3 sigma approach for the 2020 scenario. The overall reserve requirements used in the analysis are shown in Table 9.3. The total requirements are nearly doubled in 2020. The main increase takes place in Germany and the Netherlands, which is primarily due to high increase in offshore WPP.
9.4
Case Studies
The influence of WPP on system operation is studied based on four scenarios. First, 2010 is simulated using the actual installed wind power capacity and the corresponding imbalances as a reference. Secondly, two cases for the 2020 scenario are simulated. Two different cases for the reserve procurement as well as the system balancing are defined. These cases are no market integration and full market integration.
The case no market integration represents the current state. Regulating reserves have to be procured in the respective countries. In countries split into different control areas, e.g., Norway and Germany (see Figure 4.1), reserve
9.5. Results
requirements are defined by control area; however, the procurement can be done country-wide with the consideration of available transfer capacities. Exchange of balancing power with neighbouring countries is not possible.
Full market integration describes a future state in which regulating power
markets in Northern Europe are fully integrated. Besides the procurement of reserves in their respective countries, reserves now can also be procured in the whole simulated area. However, as suggested by ENSTO-E [120], 50% of the required reserves must be procured in the respective country.
Given the available transmission capacity, exchange of regulating energy is enabled in the fully integrated market. This exchange results in the activation of the system-wide most economical reserves.