Scholarly article on topic 'Determining the Economic Value of Offshore Wind Power Plants in the Changing Energy System'

Determining the Economic Value of Offshore Wind Power Plants in the Changing Energy System Academic research paper on "Economics and business"

CC BY-NC-ND
0
0
Share paper
Academic journal
Energy Procedia
OECD Field of science
Keywords
{Offshore / "system costs" / "flexibility costs" / "reserve power" / "optimal renewable energy mix"}

Abstract of research paper on Economics and business, author of scientific article — Christoph Richts, Malte Jansen, Malte Siefert

Abstract This paper analyses the economic effect of integrating large offshore wind energy capacities in the German future energy system. For this purpose three scenarios are compared at a completed transformation status in which 80% of Germany's final energy consumption (electricity, heat and transportation) is provided by renewable energy systems (RES). Emphasis is put towards an overall system perspective accounting for generation costs of renewable energy sources and balancing requirements of renewables (flexibility costs). A second focal point is the analysis of relevant power plant properties of offshore wind power, especially the provision of control reserve power.

Academic research paper on topic "Determining the Economic Value of Offshore Wind Power Plants in the Changing Energy System"

CrossMark

Available online at www.sciencedirect.com

ScienceDirect

Energy Procedia 80 (2015) 422 - 432

12th Deep Sea Offshore Wind R&D Conference, EERA DeepWind'2015

Determining the economic value of offshore wind power plants in

the changing energy system

Christoph Richtsa, Malte Jansena, Malte Sieferta

aFraunhofer Institute for Wind Energy and Energy System Technology (IWES), Königstor 59, 34110 Kassel, Germany

Abstract

This paper analyses the economic effect of integrating large offshore wind energy capacities in the German future energy system. For this purpose three scenarios are compared at a completed transformation status in which 80% of Germany's final energy consumption (electricity, heat and transportation) is provided by renewable energy systems (RES). Emphasis is put towards an overall system perspective accounting for generation costs of renewable energy sources and balancing requirements of renewables (flexibility costs). A second focal point is the analysis of relevant power plant properties of offshore wind power, especially the provision of control reserve power.

©2015Published byElsevier Ltd. Thisisan openaccess article under the CC BY-NC-ND license

(http://creativecommons.Org/licenses/by-nc-nd/4.0/).

Peer-review under responsibility of SINTEF Energi AS

Keywords: Offshore; system costs; flexibility costs; reserve power; optimal renewable energy mix

1. Introduction

Germany is aiming at reducing the GHG emissions until 2050 about 80-95% compared to 1990 level. Given the recent pace of energy transition, the legal base and actual scenarios for future capacity addition, it seems unlikely that this overall goal will be achieved. Energy balance and system analysis shows that about 850 TWh of green electricity is required to empower the electricity sector and to provide emission-free energy for heat and transportation [1]. Today's contribution of RES to electricity supply is about 150 TWh. The government aims at producing 80% of gross electricity consumption from RES in 2050. Still, it is unclear to which extent new consumers for successful sector-coupling are included within this figure and what strategies are applied to enable these technologies (i.e. electric vehicles, heat pumps, power-to-heat, power-to-gas).

However, if the dramatic paradigm shift of a complete energy transition providing 80% of final energy demand from RES is to be achieved, offshore wind power plants need to play a crucial role. In the first part of this paper (chapter 2) it is shown, that contribution can be achieved cost-competitively by demonstrating the value of offshore

1876-6102 © 2015 Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license

(http://creativecommons.org/licenses/by-nc-nd/4.0/).

Peer-review under responsibility of SINTEF Energi AS

doi: 10. 1016/j .egypro .2015.11.446

wind power from a system cost perspective. To do so, in a first step (section 2.1) this paper numbers the renewable capacity and sector-coupling technologies which are needed for a complete energy transition. Thus, the main pillars of the underlying energy concept are outlined. In a second step (section 2.2) a suitable mix of renewable energies for supplying the electricity system is derived by varying the share of wind onshore, offshore and PV minimizing standard deviation (SD) of the residual load (RL) in Germany's electricity sector. By this we can give information on the optimal mix between those three technologies. Wind offshore plays a crucial role in that mix. Two more scenarios are designed which have limits on the installed capacity of wind offshore. In a third step (section 2.3) the economic implications concerning generation and flexibility costs for balancing RES are analyzed. The results between the scenario where offshore wind was not capped and SD of RL have been minimized ("minimized SD scenario") are compared to the two scenarios without significant contribution from offshore wind power plants ("onshore scenario" and "PV scenario").

The second part of this paper (chapter 3) considers an additional benefit of offshore wind power which has not been accounted for in the previous analysis. However, the findings in chapter 2 imply that the characteristics wind onshore are different to wind offshore with wind offshore providing additional benefit to the system. Therefore it is assessed how offshore wind power plants perform in terms of reliability and forecast accuracy compared against onshore wind power plants. This approach shall back the previous findings (chapter 2) by assessing the wind power plant characteristics in detail. The characteristic of the different wind power technologies are quantified by using probabilistic forecasting. To reflect the scenario findings we look at required performance in the power system of today, since we cannot analyze system design in 2050, which will be dependent on many contributing factor. This performance can be expressed as deviations from the schedule and the capability (i.e. potentials p.u.) of balancing reserve provision to the system in the near future. These benefits are referred to as power plant characteristics.

Nomenclature

RES Renewable Energy Systems PtH Power-to-heat

GHG Greenhouse Gases PV Photovoltaic

GW Gigawatt RL Residual load

kW Kilowatt SD Standard deviation

LCOE Levelized costs of electricity TWh Terrawatt hours

PtG Power-to-gas

2. Benefits of offshore wind power from a system perspective

2.1. General scenario design

The upcoming scenario analysis for 2050 is based on the development of a general future energy vision. Therefore an energy balance has been developed and its underlying basic assumption and implications are summarized in the following (more details can be found in Rohrig et al. [3]).

• Energy efficiency: Final energy consumption in 2050 decreases about 38% compared to today's level. This is due to better energy efficiency on the demand side and the usage of electricity-based highly efficient technologies. Efficiency gains are assumed to come from a significantly lower heat demand in all sectors, especially in industrial applications and space heating by better thermal insulation standards. A second important factor to decrease final energy consumption is switching to electricity-based highly efficient technologies. Namely e-mobility in transportation and direct usage of (renewable) electricity for heating (especially heat pumps). Conventional electricity consumption is assumed to decrease as well.

• Sector-coupling: The coupling of the sectors electricity, heat and transportation is vital for effective reduction of GHG-emissions. The required new technologies are largely electricity-based and powered by renewable

sources, adding in total roughly 380 TWh on top of the "conventional" electricity demand (see table 1). Power-to-gas technology ensures seasonal storage and RE gas supply - also for the heat and transportation sector.

• Fluctuating RES: Weather dependent fluctuating RES become the main pillar of the energy supply. Due to the best resource availability they need to provide roughly 800 TWh. Electricity supply from biomass and hydro power is limited due to low potential in Germany. It is assumed that they provide 65 TWh. Remaining electricity demand is covered by gas-fired conventional plants.

An overview of the energy balance for the year 2050 is given in table 1 and compared to today's situation.

Table 1. Facts of Germany's energy situation today and scenario definition for 2050.

Today 2050

RES share - final energy 12.5% 80%

Electricity demand ~600 TWh ~900 TWh

conventional 600 TWh 522 TWh

non-conventional - 378 TWh

e-mobility - 60 TWh

heat pumps - 68 TWh

power-to-gas - 150 TWh

power-to-heat - 100 TWh

Electricity supply ~600 TWh ~900 TWh

Biomass+Hydro ~65 TWh ~65 TWh

Fluctuating RES (PV, wind) ~85 TWh ~800 TWh

Fossil fuels ~450 TWh ~ 35 TWh

2.2. Deriving an optimized RES-mix

The following analysis identifies a mix of fluctuating renewable energies for the general scenario designed in the previous section where SD of RL is minimized. The working hypothesis is that minimizing SD yields lowest system costs at the end. The scenario time-series for power feed-in from on- and offshore wind power and PV are modelled using the simulation environment developed at Fraunhofer IWES (for the details of methodology see Arbach et al. [4]). The time-series is based on historical meteorological data of the year 2011 (wind speed and solar irradiation, COSMO DE reanalysis data, spatial resolution 3km x 3km). The result is power data in hourly resolution for a specific installed capacity so that annual energy supply totals 800 TWh being supplied by offshore wind (30%), onshore wind (50%) and photovoltaic (20%). The power time series are depicted normalized to the installed capacity in chapter 3, figure 4.

For identifying an optimal mix of renewable energy technologies the magnitude of fluctuations induced on the energy system are measured by calculating the SD of RL. The residual load is here defined as the electric historical load of Germany in 2011 minus the power feed-in from fluctuating RES. The electric load is linearly downscaled to the sum of energy of the projected conventional electricity demand in 2050 (see table 1). The same procedure is applied to every possible combination of RES-mixes varying the percentage of each technology in 1%-steps. Accordingly the RE feed-in time series have been linearly down- or up-scaled. The total energy provided by all

Further research is envisaged where the RE feed-in time series are simulated for every 1%-step with the same methodology as the sample time-series instead of simple up- or down-scaling. This would better account for smoothing effects due to more or less total installed capacity of each technology (onshore, offshore, PV) and it's different geographical placement within Germany.

RES is always 800 TWh. The resulting change of SD is drawn in figure 1a indicating absolute values in GW by the color of the plot. Higher values indicate more severe fluctuations. On the y-axis the share of total wind energy is varied. A share of 50% means that there is a 50% contribution from solar energy. On the y-axis the share of offshore wind energy in percent of the total wind energy is shown. 100% means that there is no contribution from onshore wind energy.

According to the findings from this assessment the lowest fluctuation is given for RES-mixes with a high share of offshore wind energy ranging between 40% - 60% of all wind power while PV is contributing 15-25% of the 800 TWh (see figure 1a). Within these possible scenarios changes of the magnitude of fluctuations are almost negligible (SD is between 44 and 42 GW). In contrary scenarios tending towards the employment of only one or two technologies show a stronger reaction in terms of increasing SD of residual load (up to 68 GW given 50% solar contribution, best case only PV and offshore is 48 GW, PV and onshore is 51 GW). However, if potentials for RES in Germany are included in the analysis offshore wind energy and PV contribution is unlikely to be higher than 250 TWh for each technology. Limits to potentials are due to lack of available area (for PV the focus is on roof-top installments). For onshore wind energy the technical potential is very large and limits are depending rather on the socio-economic boundaries (i.e. land usage). In figure 1b it is assumed that up to 2% of Germany's total area can be used for wind onshore power plants. In the figure only the colored area indicates viable RES-mixes within these potential limits which are shown by dashed lines. Thus, a large number of scenarios - especially with very large contribution from offshore and PV, but also onshore - are unrealistic. However, potential limits are indicative here and were derived from different sources [3,5,6,7].

Fig. 1. (a) Standard deviation for different RES-mixes; (b) Standard deviation for realistic scenarios given assumed limits of potential.

In the next section three scenarios are selected and compared to each other in order to analyze the economic impact of offshore wind energy. This is on the one hand the scenario with minimum SD within the potential limits ("minimum SD scenario"). For purpose of comparison two scenarios beyond the potential limits are chosen in which the energy produced by offshore wind energy is substituted by onshore wind energy ("onshore scenario") or PV ("PV scenario").

2.3. Analysis of key parameters and system costs in three scenarios with different share of offshore wind power

In this section a comparative residual load analysis shows the impact of the different RES-mixes on the key parameters of the energy system. To account for the available flexible consumption in 2050 a load management simulation is applied on the residual load in each of the three scenarios. The used algorithm integrates the electricity consumption and flexible load of new consumers such as heat pumps and e-mobility as well as air conditioning and use of household devices minimizing the variance of the standard deviation of RL. This reduces

peak loads and fills load valleys smoothing the availability of excess power from RES so that applied storage technologies work on higher load factors. However, the following residual load analysis is done before any storage commitment. The results are indicators of how much storage capacity is needed to make use of excess power.

The following analysis opposes generation costs which is levelized costs of electricity (LCOE) of RES and flexibility costs which occur for balancing the energy system. These are costs for back-up-capacity covering electricity demand which cannot be satisfied by RES or load management, fuel to power this capacity, storage capacity for power-to-gas (PtG) and power-to-heat (PtH) and curtailment of RES. The sum of generation and flexibility costs is referred to as overall system costs. ^

The generation costs are depicted in table 2 and are scenario-independent cost assumptions for each technology derived from Kost et al. [8] and Hobohm et al. [9]. Depending on the technology share in the scenario and given the 800 TWh of renewable electricity a total sum of generation costs is calculated.

Table 2. Generation costs in the selected three scenarios

scenario- "minimum SD "onshore scenario" "PV scenario"

independent scenario"

LCOE (ct/kWh)

Photovoltaic 7.1 19.0% 19.0% 49.6%

Onshore wind energy 6.1 48.7% 79.3% 48.7%

Offshore wind energy 6.9 32.3% 1.7% 1.7%

Weighted, average LCOE (RES-mix, ct/kWh) 6.548 6.304 6.610

Total generation costs for 800 TWh (bn EUR) 52.4 50.4 52.9

The calculating of the flexibility cost is also based on simple approaches. The relevant parameters for total calculation of flexibility costs are based on the RL analysis. The RL is shown as a power duration curve after applying the load management simulation but before storage use (PtG+PtG) in figure 2a and 2b. Hours with positive RL are more frequent than hours with negative RL.

-minimum SD scenario

-onshore scenario

70 60 _ 50

S 40 g 30 fi 20 10 0

Hours per year

1000 1500 2000 2500

-20,0 -50,0 -80,0 -110,0 -140,0 -170,0 -200,0

Hours per year

2000 3000 4000

5000 6000

PV scenario

Fig. 2. Residual load in the three selected scenarios (a) positive values of RL (b) negative values of RE

f System costs do not reflect the costs of the entire energy economy. For example grid costs, costs for other energy resources (biomass, hydro power, non-electric fuel in heat and transportation) and energy infrastructure are not included. The system costs represent the bulk of primary energy costs and for balancing assuming that other costs are identical in all three different scenarios.

The need for back-up-capacity results from the remaining peak electricity demand of the RL. Annuity costs are calculated based on investment costs of 500 EUR/kW (interest rate 5%) representing the expenditure for gas turbines, CHP-units or gas engine back-up-capacity which cost-effectively and flexibly run on low operating hours. The amount of fuel needed depends on the remaining positive residual energy. Fuel costs are (conservatively) estimated to reach 3.6 ct/kWh^ in 2050 and are derived from future natural gas price projections [10]. However, the fuel can be natural gas or renewable methane. Costs for renewable methane are likely to be higher, but renewable methane production is assumed to be equal in all three scenarios so that the cost difference is not relevant for comparison. It is assumed that the fuel is converted into electricity with an efficiency of 40%.^ Thus total cost for residual energy is a function of remaining electricity demand, specific fuel costs and efficiency. Specific cost of demand not met by fluctuating RES is between 12.4-12.7 ct/kWh including costs for fuel and back-up-capacity. It is thus, higher than LCOE for RES. Storage costs are estimated based on the amount of capacity needed to store 150 TWh of electricity by PtG and to convert 100 TWh by PtH.

Depending on the availability of excess power in each scenario, different full load hours of storage technologies can be reached resulting in different installed capacities. It is assumed that excess power is used with priority for PtG due to significantly higher investment costs. Investment costs are assumed to be 200 EUR/kW for PtH and 1000 EUR/kW for PtG (electrolysis and methanization). Specific costs of curtailment is equal to LCOE as LCOE have been calculated based on total energy production regardless weather is can be integrated in the electricity system or not. Due to different quantity of curtailment given different RES-mixes the total costs is scenario-dependent. The logic behind this approach is that also curtailed or "potential" renewable energy production needs to be refinanced.

The system costs (sum of flexibility and generation costs) are finally summarized in table 3. They allow for comparison of the cost effectiveness of the three analyzed scenarios. The "minimum SD scenario" offers the lowest system costs (63.5 bn EUR/a) followed by the "onshore scenario" (64.5 bn EUR/a) and the PV scenario (69.7 bn EUR/a). Thus, the annual net effect of total savings in the "minimum SD scenario" is 1 billion Euro compared to the "onshore scenario" and 6.2 billion compared to the "PV scenario". The savings are due to the lower flexibility costs due to lower storage and back-up capacity required, lower costs for coverage of residual electricity demand and lower costs for curtailment.

Table 3. Facts of Germany's energy situation today and scenario definition for 2050.

"minimum SD "onshore scenario" "PV scenario"

scenario"

Back-up capacity (maximum peak load, GW) 54.4 62.0 62.6

Annuity costs (bn EUR/a) 1.8 2.0 2.0

Electricity demand of residual load (TWh/a) 53.4 68.9 81.8

Fossil fuel costs (bn EUR/a) 4.8 6.2 7.4

Storage capacity (GW) 67.9 74.3 83.9

Annuity costs (bn EUR/a) 3.2 3.6 4.0

Curtailment of energy from RES (TWh/a) 20.3 35.9 51.2

Costs of curtailment (bn EUR/a) 1.3 2.3 3.4

Flexibility costs per year (bn EUR/a) 11.1 14.0 16.8

Generation costs per year (bn EUR/a) 52.4 50.4 52.9

Electric efficiency is assumed to be 40% due to the fact that most back-up plants only operate on very low fulll load hours (see figure 2a). Therefore it is is more cost-efficient to use technologies with low investment costs despite lower electric efficiciency. Especially small CHP units nonetheless can reach high overall efficiency supplying electricity and heat.

System costs per year (bn EUR/a) 63.5 64.5 69.7

However, sensitivity analysis shows that the results strongly depend on the assumed generation cost of each technology. The net savings are turning into net losses between the "minimum SD scenario" and the "onshore scenario" if offshore LCOE are about 1 ct/kWh higher than assumed and LCOE for other technologies remain untouched. The PV scenario is even then, more expensive. This is also valid if offshore costs remain as initially assumed and PV costs are reduced about 2 ct/kWh. Other crucial assumptions are the costs of fuel for back-up capacity and interest rate of capital. However, they have been chosen conservatively with 3.6 ct/kWh for the year 2050 and 5%. Higher values increase the net savings in the "minimum SD scenario" as the unmet demand and the infrastructure requirements (back-up-capacity and storage) are higher in other scenarios. For the results of the sensitivity analysis see figure 3.

Weather the full load hours (FLH) of wind energy, resulting from the technical assumptions of the turbine development (especially the rotor-generator-ratio) and the resource data, have a severe impact on the results has to be investigated in further research. Simulated FLH reach 4800 hours per year for offshore wind energy and 2700 hours for onshore wind energy in the year 2050. However, it is likely that not the absolute value of FLH but the ratio between offshore- and onshore-FLH is relevant. Different weather years could also have an influence on the results (resource data is based on 2011). Furthermore no grid simulation or congestion analysis has been conducted and no cost assessment for possibly different grid expansion needs within the proposed scenarios has been undertaken. Additional research work can address these issues.

Sensitivity

■ Net effect (compared to "onshore scenario") ■ Net effect (compared to "PV scenario")

Fig. 3. Change of net saving effect in the minimum SD scenario for different assumptions 3. Power plant properties and balancing reserve provision by offshore power plants

In chapter 2 we have assessed whether offshore wind power plants have characteristics that decrease fluctuations of the residual load in the system. We have shown that including large share of offshore wind power plants is important in that aspect and thus facilitate system integration. To further investigate these desirable characteristics we use the results from chapter 2 and present day challenges for wind power plants to explain the differences. One possible way to measure the quality of supply is to look at generated forecast deviations. In the energy system today this would mean the fluctuations on average over a period of 15 minutes, as discusses in section 3.1. Fewer deviations will cause less demand for balancing. The characteristics of the wind power plants also have to address an increasing demand for system services, such as balancing reserves. The potential to deliver this service can be seen in section 3.2. System services are currently provided mostly by conventional power plants. To ensure a secure system operation with very high levels of RES penetration it is paramount that wind turbines, amongst others, deliver all types of system services. These services will have to be provided by variable RES if we want to avoid the creation large amounts of must-run capacities from conventional generation.

Due to this reason we assess the power plant properties of wind turbines and the characteristics of reserve provision from wind turbines. For the following assessment exemplary data from an offshore wind power plant and an onshore wind power plant has been used and combined with information based on weather data as described in chapter 2. Both data sets were created using the same methodology.

Power plant properties can be categorized as schedule accuracy and forecasting error. All electricity production and consumption is forecasted with schedules prior to operation. Any deviation from the schedules will cause a disruption in the balance of the power system and hence cause the dispatch of reserves. Deviations by wind turbines are caused by errors between the forecast and production. It is important to reduce deviations as much as possible. Therefore it will be assessed how offshore wind power plants perform compared to onshore wind power plants in term of schedule accuracy.

The dispatch of balancing reserve is required if deviation of grid frequency from its target value indicates an imbalance between consumption and production. Power plants have to change their performance until the frequency is nominal again. A dispatch of positive reserve means an increase in power production whereas a dispatch of negative reserves would mean lowering the power output.

In order to assess the quality of a wind power forecast one could look at the forecast error. For this paper probabilistic wind power forecasts errors have been created and analyzed. Due to lack of real forecasting and power measurement data for an offshore wind power park the power forecasting error is modelled based on wind speed forecasting errors. Representative frequency distributions of wind speed errors are known and applied to wind speed distributions for a representative onshore and a representative offshore site. Power output is then calculated using an evaluated physical wind power plant model, resulting in the real power output for both sites and a frequency distribution of forecasting errors for the power output. The following comparison uses data for offshore and onshore wind power plants that are modeled in the same way.

3.1. Forecast accuracy

The accuracy of the schedule can be assessed by the analysis of production volatility over time. Compared to onshore wind power plants offshore wind farms have a more constant production. Based on the analysis of the simulated scenario-data in 2050 (see part two of this paper), it was found that offshore wind farms have a production pattern that is generally better predictable. For example, it is far more likely that the wind power plants either produce at full capacity or nothing at all than it is the case for the onshore wind power plants. This makes forecasting more reliable since the behavior creates less hours with partial loading on the wind turbine. Partial loads are in the steep area of the power curve. Small changes in wind speed could cause large changes in power output. The opposite is true when the wind turbine is producing at rated power. Figure 4 shows the annual production pattern of the simulated data in 2050 of offshore wind power plants, onshore wind power plants and photovoltaic systems in form of a power duration curve.

Results of the analysis of the two simulated wind power plants show that the probabilistic forecast errors of offshore wind power plants are smaller and less severe for onshore wind power plants. The average forecast error, defined by the averaged difference between the 25%- to 75%-percentile of the probabilistic forecast, is 25% for offshore wind power plants related to the average feed-in, the one of onshore wind power plants is 60 %. Offshore wind power plants often have very small forecasting errors, and the maximum deviation from the forecast is significantly smaller than with onshore wind power plants (see figure 5), hence offshore wind power plants produce energy more reliable and are better capable to fulfil the announced schedule. Further forecast improvements can be expected since forecasts for offshore wind power plants are at the beginning of their learning curve. This leads to a smaller balancing demand and fewer costs for flexibilities to balance the system.

0 1000 2000 3000 4000 5000 6000 7000 8000 8760

Hours of the Year

Onshore wind Offshore wind Photovoltaic

power plant power plant systems

Fig. 4. Power duration curve of onshore wind power plants, offshore wind power plants and photovoltaic systems in 2050 based on weather

year data 2011

From the system point of view forecast errors of individual wind power plants cancel each other out if wind farms are spread over a large area, especially if the weather conditions differ significantly. Therefore, onshore wind power plants and offshore wind power plants complement each other very well, since their combined forecast error is smaller than the individual one. We conclude that offshore wind energy contributes to the balancing of the

power systeQHObRQmgntiCojoTecagta lerrongfrequencydist errors for

the wind power plants described earlier.

25% -75% percentile of probabilistic forecast

Average real Power

Offshore wind power plant Onshore wind power plant

50% 100% 150%

Forecast error (in % of average feed-in)

Fig. 5. (a) Sketch of the difference of the 25%- and the 75%-percentile in different time steps of the probabilistic forecast (b) Distribution of

forecast errors in % of average feed-in of wind power plants

3.2. Balancing reserve provision

The provision of balancing reserves by wind power plants is another way to assess the characteristics of wind power plants. The provision of balancing reserve requires the wind powerplant to be forecasted most accurately, besides technical aspects. Figure 6 shows the differences in potential reseavo proviso between the modeled offshore and onshore wind power plants. Forecasts are created using a statistical method based on the data modeled for section 3.1. Derived from the deterministic forecast and the schedule error, probabilistic forecasts are calculated using a kernel-density estimator.

These probabilistic forecasts were used to calculate the potential for delivering balancing reserve (compare also [10] and [11]). Probabilistic forecasts combine a power output with a probability. By choosing a power forecast related to a very high reliability, balancing reserve can be delivered to the market as reliable as from thermal generation. This is ensured if the reliability of a forecast is equal or higher than 99.994%. This number would guarantee that the reliability of an offer is no lower than offers from thermal generation. However looking at different level of reliability is important since the required reliability level might be subject to change. Figure 6 shows how much energy can be offered with different levels of reliability, in reference to the total energy output from the wind power plant. An offer potential of 0.2 in the means that 20% of the annual energy provided can be offered as reserve power with a certainty of e.g. 99.994%. Figure 6a shows the potentials for the reserve provision created with a day-ahead forecast whereas figure 6b shows the potentials for a reserve provision using a one hour-ahead forecast.

Offshore wind power plant Onshore wind power plant

99.99 99.994

Offshore wind power plant Onshore wind power plant

Level of reliability in %

Level of reliability in %

Fig. 6. (a) Potential for balancing reserve of offshore and onshore wind power plants for different levels of security for a day-ahead forecast in % over average annual feed-in (b) Potential for balancing reserve of offshore and onshore wind power plants for different levels of security for

one hour-ahead short term forecast in % of average feed-in

Tertiary control (ENTSO-E name: manual FRR) in Germany is procured on a daily basis. This means that the day-ahead potential is important if wind power plants want to offer balancing reserve under current regulations. With the help of a probabilistic forecast offshore wind power plants can offer 18% of their annual power feed-in as balancing reserve. With the same methodology onshore wind would only be able to offer less than 2% of their annual feed-in. Additionally for the near future intraday reserves will become important since the Network Code in Electricity Balancing is aiming at implementing a European balancing reserve market with one hour lead time [12]. As it can be seen in Fig. 6b the difference in potentials between offshore and onshore wind power plants are smaller compared to the day-ahead forecast. The offshore wind farm can offer more than 50% of its annual production as balancing reserve whereas the onshore wind farm could provide about 30% of its annual production. The fact that offshore wind energy can be predicted with a smaller error than onshore wind enables it to deliver more system services.

4. Conclusion

From a system perspective integrating large shares of offshore wind energy into the RES-mix can lead to significant cost reduction in flexibility costs. This is due to the less volatile feed-in from offshore wind energy especially resulting in better load coverage and thus, lower costs for supplying the remaining electricity demand which cannot be covered by fluctuating RES. The second most important aspect is that curtailment of RES is lower

than without offshore contribution, thus saving costs. Additional but minor cost reductions occur due to the fact that less storage and back-up capacity is needed. According to our findings, total savings in costs lie between 1 and 6 billion Euro per year compared to strongly onshore or strongly PV-orientated scenarios. However, overall system savings heavily depend on the development of generation costs for RES and transmission grids which have not been subject to the study, where further research is needed. Thus, if generation costs become more cost competitive then today, offshore energy will play a decisive role for a cost effective and feasible emission-free energy future.

Apart from that it has been shown that offshore wind power can be predicted more accurately than onshore wind power. Thus, fewer reserves for the balancing of forecast errors are needed. Due to the reliable forecast large parts of the production can be used to deliver system services. In fact more than 50% of the power output of an offshore wind power plant can be predicted with a reliability of 99.994% one hour in advance. Around 18% can be predicted day-ahead with the same reliability.

References

[1] Gerhardt et al. Geschäftsmodell Energiewende - Eine Antwort auf das "Die-Kosten-der-Energiewende"-Argument. Kassel: Fraunhofer Institute for Wind Energy and System Technology; 2014.

[2] Rohrig et al. Energiewirtschaftliche Bedeutung der Offshore-Windenergie für die Energiewende - Langfassung. Varel: Foundation Offshore Wind Energy; 2013.

[3] Arbach et al. Virtuelles Stromversorgungssystem - Komplettsimulation zukünftiger Stromversorgungssysteme. Kassel: Fraunhofer Institute for Wind Energy and System Technology; 2013.

[4] Bofinger, Spiekermann. Potenzial der Windenergie an Land - Studie zur Ermittlung des bundesweiten Flächen- und Leistungspotenzials der Windenergienutzung an Land. Berlin: Federal Environment Agency; 2013.

[5] Klaus et al. Energieziel 2050 - 100% Strom aus erneuerbaren Quellen. Berlin: Federal Environment Agency; 2010.

[6] Bofinger et al. Potenzial der Windenergienutzung an Land. Berlin: German Wind Energy Association; 2011.

[7] Kost et al. Stromgestehungskosten Erneuerbare Energien. Freiburg: Fraunhofer Institute for Solar Energy Systems; 2013.

[8] Hobohm et al. Kostensenkungspotenziale der Offshore-Windenergie in Deutschland. Varel: Foundation Offshore Wind Energy; 2013.

[9] Nitsch et al. Langfristszenarien und Strategien für den Ausbau der erneuerbaren Energien in Deutschland bei Berücksichtung der Entwicklung in Europa und global. Köln: German Aerospace center; 2012.

[10] Jansen et al. Pool of Photovoltaic Systems delivering Control Reserve, Proceedings Solar Integration Workshop. London; 2013

[11] Jansen et al. Impact of control reserve provision of wind farms on regulating power costs and balancing energy prices, Proceedings Wind Integration Workshop, Lisbon; 2012

[12] ENTSO-E Network Code on Electricity Balancing (EB). Brussels; 2014; available online on: https://www.entsoe.eu/major-projects/network-code-development/electricity-balancing/Pages/default.aspx (last accessed: 22.01.2015)