Investigation of balancing effects in long term renewable energy feed-in with respect to the transmission grid

A European power system mainly based on renewable sources will have dominant contributions from wind and solar power. However, wind and solar generation facilities have, due to the weather dependent nature of their resources, highly fluctuating feed-in profiles. To overcome the mismatch between power consumption and generation it is important to study and understand the generation patterns and balancing potentials. High temporally and spatially resolved long term weather data was used to simulate the feed-in from wind and photovoltaics for European countries for the years 2003 to 2012. We investigate storage energy and capacity needs in Europe in dependency of the generation mix from wind onshore, wind offshore and photovoltaics and the share of renewables. Furthermore we compute the storage energy and capacity needs for different transmission scenarios. We show that for unlimited transmission storage needs are reduced mostly by high wind offshore shares. We also show that higher shares above 100 % of renewables can decrease the required storage capacity to a higher extent than the required storage energy.


Introduction
The European energy transition is Europe's path towards non-fossil sustainable energy generation.In 2012 the share of renewables in the power sector was already 25 % and it is expected to increase tremendously in the upcoming decades.In such a power system wind and solar power will have most likely the highest shares.To ensure stable electricity transmission grids, supply and demand need to match reliably.However, the feed-in profiles of wind and solar generation facilities depend on the weather.This makes their integration into the power system difficult.Several solutions have been proposed and investigated to solve or reduce the problem of the resulting generation-load mismatch.These are for example excess generation (Heide et al., 2010) or transmission grid extensions (Becker et al., 2014;Rodriguez et al., 2014).Other relevant papers investigating European power systems based on renewable energies are for example (Huber et al., 2014;Rasmussen et al., 2012;Schaber et al., 2012;Steinke et al., 2013).In the following, balancing is investigated by the calculation of storage needs from feed-in for wind and solar and load data.We study storage energy and capacity needs in dependency of the generation mix and for different transmission scenarios.

Data
We investigate a renewable European power system for 34 countries (EU-28, Norway, Switzerland and the Balkan states).The background scenario of this work is Pfluger et al. (2011).Generation capacities of the single countries have been adopted from this scenario.Within the single countries the capacities were distributed according to an empirically derived approach: the cumulative distribution function of the capacities dependent on the average resource (wind speed for wind and irradiance for pv) was derived for Germany and adopted for all countries (Fig. 1).No grid points of the onand offshore regions were filtered out.

Weather data
A large weather database to model feed-in from renewable sources has been set up.This database has a spatial resolution Published by Copernicus Publications. of 7 km × 7 km covering most of Europe at hourly temporal resolution for the years 2003 to 2012.It consists of wind speeds in 140 m and temperatures in 2 m height, that have been derived by statistical downscaling from MERRA reanalysis data (Rienicker et al., 2011).Since MERRA reanalysis only provides hourly wind speeds in 10 and 50 m height with a spatial resolution of 1/2 • × 2/3 • , the MERRA wind speeds were statistically downscaled to the desired resolution of 7 km × 7 km.The grid points of this resolution were chosen to be the grid points of the COSMO-EU analysis (Schulz and Schättler, 2009).The statistical downscaling consists of three steps: in the first step the MERRA wind speeds were spatially interpolated to the 7 km × 7 km grid.In the second step they were logarithmically extrapolated to 140 m under the assumption of a logarithmic wind profile with surface roughness lengths provided by COSMO-EU.The logarithmic wind profile is given implicitly by where s(x) is the wind speed at height x, z is the desired height to be extrapolated to, z 0 is the height for which the wind speed is available and z r is the surface roughness length.
The surface roughness length is the parameter to model the logarithmic wind profile.The final step of the downscaling process consisted of the calculation of linear regression coefficients between MERRA reanalysis and COSMO-EU analysis wind speeds for the year 2012.It means, that the relationship between COSMO-EU windspeeds s c and MERRA wind speeds s m is assumed to be where and b are the calculated parameters of the linear regression for a single grid point.After calculation of these coefficients this regression was applied on all investigated years.
For the pv power calculations images from Meteosat First Generation (MFG) and Meteosat Second Generation (MSG) satellites were used.The Heliosat method (Cano et al., 1986;Hammer et al., 1998) was then applied to derive surface irradiance which is used for pv power calculations.

Calculation of feed-in time series
The powercurve of an Enercon E-126 at 140 m hub height with additional 5 % plain losses was used to convert wind speeds into produced power for every grid cell.Photovoltaics power has been calculated by using the global horizontal irradiation and configurations of pv modules (tilt angles, azimuthal orientations, types, etc.) adopted from the background scenario.The conversion of global horizontal irradiance to the tilted module is based on the Klucher model (Klucher, 1979).Module temperatures were computed as where T a is the ambient temperature, T m the module temperature and I t the irradiance on the inclined surface.σ is a factor that was chosen to be 0.036.The module efficiency η(T m ) is then calculated as a function of the module temperature for a given irradiance where a is a device-specific parameter.Then DC power can be calculated by using the installed capacity for any grid point.DC power was converted into AC power using the parameter of a Sunny Mini Central 8000TL converter.Feed-in of the single grid cells aggregated to the country levels for wind and pv.

Load data and transmission capacities
Load data was retrieved from ENTSO-E for the countries investigated and slightly modified to account for the background scenario.We investigate the storage needs in this work for several scenarios: one with unlimited (copperplate model), one without any inter-country transmission in Europe and the other ones with multiples of net transfer capacities.These net transfer capacities (NTC) are calculated and provided by ENTSO-E.The NTC values for summer 2010 have been used to limit transmission between countries.

Balancing by storages and transmission
This section explains how storage capacity and energy needs, which are the central measures of this investigation, are defined and computed from this data.For every country we have computed a generation time series where P n is the generation from photovoltaics of country n, W n is the generation from wind onshore and U n is the generation from wind offshore.Together with the load data the time series of the mismatch can then for every node n directly be computed as where G n (t) is the produced power, L n (t) is the load and F m is the transfered power via a link m between two different countries.We assume that storages are charged by excess energy and discharged when the mismatch is negative.The overall mismatch at time t is given by The storage filling level can then be described as where S 0 is the initial storage filling level at time t 0 .This is not a detailed storage model, but a measure to quantify the generation-load mismatch problem.It is equivalent to one large storage, that is connected to all countries of the investigated power system.The measures to quantify the results of our investigations in this work are the required storage energy S E and the required storage capacity S C .The required storage energy is the energy that is needed to eliminate the negative mismatch (residual load after transmission) and is defined by where is the Heaviside function, i.e.
It is assumed that no storage losses occur.If the required storage energy reduces to zero all loads can be covered by generation at any time.No additional balancing options are therefore necessary.The required storage capacity is the amount of energy the storage needs to be able to store (reservoir capacity).It is defined for a time interval [t 0 , t 1 ] by It can be directly seen from the definitions, that S C = 0 if and only if S E = 0. We calculate storage energies in units of the overall load, while storage capacities are measured in units of the average yearly load.The flow has been formulated as an optimization problem to minimize the required energy from storages at every timestep: minimize The solution of this optimization problem yields the required energy from storage after transmission.

Results
In this section the storage needs are calculated in dependency of the installation factor (share of renewables) α.It is defined by so it describes the ratio between average generation and average load.For α = 1 load and generation are on average the same.

Optimal mix
Storage energy needs in this section are computed for the mix between wind onshore and offshore in Europe with unlimited transmission.This mix is described by the parameters β and γ in a recursive way as follows: The storage energy needs for a Europe with unlimited intercountry transmission in dependency of the tupel (α, β, γ ) can be seen in Fig. 2 and the storage capacity needs in Fig. 3.Both are strongly depending on the generation mix.Storage energy needs have a minimum at around 12 % of the overall consumption for a mix of β = 0.2 and γ = 0.0.They reduce even further for the same β, γ to 9 % for 20 % of over-installation (α = 1.2).For storage capacity needs the minimum is at the same shares.It is around 12% of the average yearly consumption and reduces to less than 1 % at these shares for over-installation of 20 %.In general, overinstallation reduces storage capacity needs much stronger than storage energy needs.This can be seen in Fig. 4. It shows the relative reductions between storage energy and capacity needs in dependency of β and γ from α = 1.0 to α = 1.1.

Storage needs in dependency of the transmission grid
We calculate storage energy needs for installation factors α from 1.0 to 1.2 for four different transmission scenarios (Fig. 5).The generation mix was kept for these calculations from the background scenario.For α = 1.0 today's transmission grid has the potential to reduce the storage energy needs by ca. 12 %.Extension of every line capacity by the factor 1.5 would reduce storage needs by additional 5 %.This is still comparably low to the reduction by around 45 % that could be achieved by unlimited transmission.At α = 1 .2storage energy needs reduce for unlimited transmission to less than 8 % of the mean load.

Conclusions
Based on a highly spatially and temporally resolved weather data set, we investigated balancing in terms of storage capacity and storage energy needs for 34 European countries in dependency of the generation mix between wind onshore, wind offshore and photovoltaics.For a scenario with unlimited transmission in Europe we concluded, that in 100 % renewable scenarios storage energy and capacity needs are considerably reduced by high shares of wind offshore compared to wind onshore and photovoltaics.Furthermore we showed, that storage capacity needs can be reduced by over-installation of 10 % (α = 1.1) by 70 to 80 %.For the same over-installation of 10 % the relative reduction of storage energy needs is around 25 %.However, transmission is not unlimited in Europe today.We have computed storage needs for different transmission grid scenarios and shown, that today's inter-country transmission grid can reduce storage energy needs by ca. 12 % for a 100 % renewable share.This is still highly extendable because unlimited transmission would reduce storage energy needs in the 100 % renewables scenario by ca.45 %.

Figure 1 .
Figure 1.Cumulative distribution function of wind and photovoltaic capacities in order by decreasing resource of wind speed and irradiance on a 7 km × 7 km grid for Germany (2013).

Figure 2 .
Figure2.Storage energy needs in units of the consumption in Europe.They are computed with unlimited transmission in dependency of the solar share β and the wind onshore share γ for installation factor α = 1.0, 1.1 and 1.2.The dashed lines indicate the optimal solar share in dependency of the wind onshore share.

Figure 3 .
Figure 3. Same as Fig. 2 but for storage capacity needs.

Figure 4 .
Figure 4. Reduction of relative storage energy (left) and storage capacity (right) needs for an increase of the installation factor α = 1.0 to 1.1 for a Europe with unlimited transmission in dependency of the solar β and wind onshore share γ .

Figure 5 .
Figure 5. Storage energy needs in Europe in dependency of the installation factor α for different transmission scenarios in units of the overall load.