Effects of Climate Change on Wind Power Generation: A Case Study for the German Bight

Abstract

Driven by the demands of climate change mitigation, many countries have begun large-scale electricity production from variable renewables, such as solar PV and wind power. Electricity production from wind turbines, in particular, strongly depends on local weather conditions and their changes caused by climate change. Thus, for many countries with a high share of wind power generation, such as Germany, two essential questions arise: how will climate change affect electricity production, and how strong will be this impact for different RCPs? To better assess the impact on existing onshore wind turbines, spatially and temporally resolved data on their power generation are required. In order to create such disaggregated data, this study uses a physical simulation model and climate data modified for the RCP 2.6, RCP 4.5, and RCP 8.5 scenarios. To investigate the effects on a significant region with very high wind power generation in Germany, the numerical simulations were carried out on an ensemble of 22 onshore wind turbines with an installed capacity of 65.5 MW in the German Bight. After model validation, the power generation from this turbine ensemble was simulated for the high-wind year 2008 and the low-wind year 2010. The simulation results are presented with a high temporal resolution, and the observed changes are discussed for the applied RCPs. In summary, the resulting wind power generation of the entire plant ensemble decreases with increasing RCP to values of up to nearly 3 GWh for both years.

1. Introduction

In order to mitigate climate change, one of the world’s most urgent challenges is to reach climate neutrality in all sectors as quickly as possible. Many countries and communities of states have already launched various programs to achieve this goal, such as the European Union’s Green Deal. A key strategy of the Green Deal is to increase the share of solar PV and wind power in the electricity supply and shift as much energy demand as possible to electricity in order to reach climate neutrality by 2050. This large-scale electrification in all sectors, from household, transport, trade and commerce, to industry, would lead to a drastic increase in Europe’s electricity consumption.

Despite its intermittent nature, renewable energy from onshore wind turbines has become an essential pillar for the electricity supply in many European countries, including Germany. Over the past decade, onshore wind power has grown rapidly due to significant advances in grid integration and cost reduction. For instance, the onshore wind capacity awarded in Europe in 2024 was 17 GW, a 24% increase from the previous year. This remarkable increase was primarily driven by the onshore wind capacity awarded in Germany reaching 11 GW [1]. The ongoing growth in onshore wind power is also reflected in the newly installed capacity of 2.6 GW in Germany, which brought the total volume to 63.6 GW in 2024 [2]. Hence, the already high share of variable renewables in the power system, coupled with the planned increase in electricity demand due to electrification, makes the electricity supply increasingly sensitive to weather conditions [3,4,5]. Therefore, it is crucial to better quantify the impact of climate change on wind power generation.

Current climate studies on renewable energies often refer to the frequency and duration of low wind speed periods and investigate their meteorological conditions and the effects on the power system when the insolation is low as well [6,7], i.e., during so-called dark wind lulls or in German “Dunkelflauten” [8,9,10]. Other climate studies on electricity production analyze seasonal differences in extreme weather events for solar PV and wind power generation in Germany, considering both existing and future installed capacities [11]. Many of these studies also predict that, in the long term, climate change may impair electricity production from wind energy more than from other renewable energy sources [12]. Further climate studies on variable renewables, such as the study of [13], assess the impact of climate change and climate variability in Europe. This study found out that, although climate change will significantly impact future power generation from solar PV and wind power, the effects are outweighed by the high and variable impact of North Atlantic Oscillation phases, especially for wind power. However, none of these climate studies or models consider disaggregated power generation from existing onshore wind turbines, taking into account their individual locations and specific plant data. In order to create such disaggregated data of onshore wind turbines available for detailed analyses of the German power system [14,15,16], this study applies a physical simulation model which uses wind turbine data and climate data modified for different RCPs [17,18].

Using this model approach, this study investigates the impact of climate change on power generation from 22 onshore wind turbines in the German Bight, a significant region where the annual wind power generation is among the highest in Germany [14,15]. In order to determine the wind power generation with a high spatial and temporal resolution, this study shows an enhanced wind power model, which is a further development of the wind power model described in [19]. The simulation model presented also belongs to the ReSTEP model collection [20,21,22,23,24].

The remainder of this paper is structured as follows: Section 2 presents the enhanced wind power model and the necessary data used for the performed simulations. After model validation in Section 3, this physical model is applied to an ensemble of 22 onshore wind turbines in the German Bight in order to simulate their electricity production for the high-wind year 2008 and the low-wind year 2010. For this, both an REF scenario providing past weather data and the RCP2.6, RCP4.5, and RCP8.5 scenarios providing climate data are used in the numerical simulations. The simulation results are shown with a high temporal resolution, and the observed changes are discussed for the applied RCPs. This study ends with brief conclusions about the wind power model and the performed simulation results in Section 4.

2. Data and Methods

From an economic and technical point of view, climate change and the accompanying extreme weather events may be associated with strong impacts on electricity production from wind turbines [25,26]. This would also have a high influence on the spatiotemporal availability and the costs of this renewable energy [27,28]. Thus, for Germany with a high share of wind power generation, two essential questions arise: how will climate change affect electricity production, and how strong will be this impact for different RCPs? In order to better investigate these questions, an enhanced wind power model is used with wind turbine data and climate data modified for different RCPs to calculate the power generation with a high spatial and temporal resolution.

2.1. Wind Turbine and Climate Data

The wind power model presented in this paper requires, for the numerical simulations, specific plant data about the investigated onshore wind turbines in the German Bight. This region was chosen because in the German Bight the annual wind power generation is among the highest in Germany and, furthermore, the climate models used for the climate change scenarios work more precise for such coastal-near regions. For each onshore wind turbine, the plant dataset contains the following information: plant location in geographical coordinates, hub height, installed capacity, and time of operation, as listed in

Table 1. Wind turbine and climate data required for the wind power model.

Wind Turbine Data	Climate Data
Location	Location
Hub height	Wind speed
Installed capacity	Air temperature
Operation time	Ground elevation

.

The specific plant data about the selected onshore wind turbines were taken from the internet platform CEMDR [29,30]. After cross-checking and selecting the wind turbines for the investigated years of 2008 and 2010, the compiled dataset comprises 22 wind turbines in the German Bight. This plant dataset includes different types and power classes of the onshore wind turbines, where their installed capacities range from 1.0 MW to a maximum of 7.5 MW. The total installed capacity of the investigated plant ensemble is 65.5 MW. Figure 1 shows the sites of the selected onshore wind turbines in the German Bight.

In addition to the specific plant data of the selected onshore wind turbines, weather or climate data at their locations are also required for realistic simulations of wind power generation. Furthermore, high-resolution weather or climate data with a temporal resolution of at least one hour are a prerequisite for accurate simulation results from physical models [31]. Using climate data in these numerical simulations enables physical models, such as the enhanced wind power model, to estimate the impact of different climate change scenarios on wind power generation.

Such climate data, which were created using past weather information modified by climate change signals for the applied RCP scenarios in climate models, were provided by GERICS. The simulations were performed for typical high-wind and low-wind years from a meteorological perspective, where the climate change impact was imposed on climate data by climate change signals for the RCP2.6, RCP4.5, and RCP8.5 scenarios. These RCPs describes climate change scenarios for the intensities of 2.6 W/m², 4.5 W/m², and 8.5 W/m² as radiative forcing at the end of the 21st century, corresponding to increasing global warming scenarios [17]. This RCP approach emerged from an innovative collaboration between climate modelers, emission inventory experts, integrated assessment modelers, and terrestrial ecosystem modelers [18].

GERICS simulates for each year, i.e., the high-wind year 2008 and the low-wind year 2010, an REF scenario providing past weather data and the RCP2.6, RCP4.5, and RCP8.5 scenarios providing climate data modified by climate change signals. The following meteorological variables were provided by GERICS for each location of the 22 onshore wind turbines: wind speed at a height of 140 m, near-surface (2 m) air temperature, and ground elevation, as listed in Table 1. These variables are required as input data for the wind power model presented in this paper. Further information on the methods and climate models used by GERICS to generate these climate data can be found in Appendix A and in [32].

2.2. Model Description and Calibration

This subsection describes the enhanced wind power model, which is a further development of the ReSTEP model presented in [19]. Wind power generation data can generally be calculated using statistical [33] or physical simulation models [34,35,36]. Unlike statistical approaches, physical models usually base their simulation results on weather or climate data obtained from meteorological measurements or climate models. Thus, an important advantage of physical models, such as the presented wind power model, is their ability to create wind power generation data with a high spatial and temporal resolution.

Figure 2 shows the calculation steps of the enhanced wind power model, in which the wind turbine and weather or climate data serve as input data for the calibrated wind power model. The calculation steps are depicted as grey rectangles, and the input and output data are shown by yellow and blue containers, respectively. The black arrows in Figure 2 indicate the flow of information in the simulation model.

The wind power model presented in this study includes modified calculations that further improve the accuracy of wind speeds at the required turbine hub heights. For the necessary extrapolation of the wind speeds to these hub heights, Hellmann’s exponential law is applied in the enhanced wind power model. This Hellmann’s exponential law [37,38] can be described by the following relationship:

$$v_{\mathrm{h}}\approx {\left({\displaystyle \frac{H_{\mathrm{h}}}{H_0}}\right)}^{\alpha }·v_0$$

(1)

In this relationship, v_h stands for the needed wind speed at the turbine hub height H_h and v₀ is the wind speed at height H₀ provided by the used input data. The exponent α represents the so-called Hellmann exponent. Unlike the wind power model described in [19], which uses wind speed data from 10 m above ground level, the enhanced wind power model uses wind speeds from 140 m, as provided by GERICS. This significantly improves the accuracy of wind speed extrapolation due to smaller differences to the required hub heights.

Similar to the ReSTEP models in [19,20], the presented wind power model also uses sixth-order polynomial approximations for the required power curves. Once a functional representation is derived for a normalized power curve, the output power can be easily determined using the installed capacity, i.e., the rated power, of an onshore wind turbine, along with the extrapolated wind speed and air temperature at its hub height. Figure 3a depicts the specific power curve of a typical wind turbine with its technical parameters. Figure 3b shows the approximated nonlinear part of the normalized power curve from Figure 3a with its coefficient of determination R². This statistical coefficient, better than 0.99, also demonstrates the high accuracy of the chosen mathematical approach. Another advantage of such normalized power curves is that similar wind turbines, for which no information about their power curves is publicly available, can be simulated in the wind power model as well.

If the wind speed becomes too high during extreme weather events, e.g., during a heavy storm, the wind turbine will shut off automatically to prevent damage to the entire system. For most onshore wind turbines, the concerning cut-out speed is 25 m/s at the hub height. However, this value may be higher for modern wind turbines depending on their specific storm control system. The individual features and settings of such storm control systems are not publicly available and, therefore, the common value of 25 m/s is used in this study. In order to more realistically investigate the influence of extreme weather events, i.e., heavy storms with high wind speed periods, the enhanced wind power model applies an additional storm control functionality. This simple storm control shuts down the wind turbine when the wind speed exceeds the cut-out speed of 25 m/s and turns it on again when the wind speed drops below 23 m/s. In other words, this storm control functionality was calibrated with a hysteresis of 2 m/s for the numerical simulations. In Figure 3a this storm control is illustrated in the power curve, i.e., its hysteresis is depicted by red arrows. All of the other calculation steps and their underlying physical laws of the presented wind power model are already described in [19]. Additional information on ReSTEP wind power models can also be found in [20].

In order to obtain realistic simulation results from the enhanced wind power model, additional information is needed to reasonably calibrate it, which is not contained in the plant dataset. For instance, the presented physical simulation model requires data on the specific power curve of a wind turbine, including its cut-in, rated, and cut-out speeds. Such power curves typically show the relationship between the wind speed at the hub height and the output power at an air temperature of 15 °C and an atmospheric pressure of 1.013 bar. Such power curves are publicly available on various internet platforms, such as The Wind Power [39].

Reasonable values of the turbine ensemble are also necessary for the Hellmann exponent (α), which is needed to extrapolate the wind speeds of the weather or climate data to the required hub heights [37,38]. This Hellmann exponent is a function of the topography at a certain geographical location and takes into account its surface roughness. The simulations performed for the plant ensemble use the common value of 0.143 for open land, which fits very well with the selected wind turbine sites in the German Bight, as shown in Figure 1. For other regions with a more heterogeneous topography, such as regions with large forest or mountain areas, the investigated wind turbines can be assigned to different groups using different Hellmann exponents to ensure a high accuracy in the simulations.

In addition, the hysteresis of the embedded storm control functionality in the enhanced wind power model has to be calibrated to a reasonable value. Since such information is not publicly available for onshore wind turbines, a value of 2 m/s was used for the entire plant ensemble in the numerical simulations.

3. Results and Discussion

3.1. Model Validation

In order to verify the presented wind power model and assess its accuracy, the results obtained from numerical simulations must be compared with measurements of existing wind turbines. For model validation, measurements of output power and wind speed are required from a wind turbine with known technical parameters. The measurements originate from a wind turbine General Electric GE 1.5sl with 1.5 MW and 100 m hub height located near the town of Görlitz in Eastern Germany and were performed between 11 and 13 December 2015. For this simulation, the Hellmann exponent was set to zero since the wind speed was measured directly at the turbine hub height. The air temperature at 2 m and the ground elevation, required to correct the output power from the power curve in the simulation model [19], were retrieved via the online tool PVGIS using its CMSAF weather product for this wind turbine site [40,41]. Figure 4 shows the simulated and measured output power as well as the measured wind speed with a temporal resolution of 10 min.

As depicted in Figure 4, the presented wind power model accurately reproduces the measured pattern over the entire period of 54 h. The frequently small underestimation of the simulation results is primarily due to the use of the so-called guaranteed power curve of the General Electric GE 1.5sl, which is often exceeded under real conditions. The RMSE of the 10 min data over the entire period, a common statistical measure of such deviations, reach a low value of 0.088 MW. Despite the existing small deviations, the enhanced wind power model yields very realistic results.

In addition, the embedded storm control functionality must also be checked, to determine whether it works according to the settings in the model calibration. This simulation was performed for a wind turbine Enercon E-126 with 4.2 MW and 135 m hub height. Figure 5 shows the operation of this storm control using the RCP 4.5 scenario of the wind storm “Emma” near the town of Aurich in the German Bight between the 29 February and 2 March 2008. For this extreme weather event, the climate data were provided by GERICS with a temporal resolution of 10 min.

Figure 5 clearly indicates that the storm control also works according to the settings of the model calibration when the wind speed exceeds the cut-out speed of 25 m/s. At this point, the wind turbine interrupts the power generation and starts again when the wind speed drops below a value of 23 m/s. This simple storm control functionality simulates a common behavior of wind turbines, i.e., without any further storm control features and settings, during heavy storms. In addition, Figure 6 shows the simulated output power of the same extreme weather event using the REF scenario and the same onshore wind turbine.

It can be clearly seen in Figure 6 that, in contrast to the RCP4.5 scenario of Figure 5, the cut-out speed is not reached and, therefore, no shutdown of the wind turbine occurs. Furthermore, despite the differences in both scenarios, the total electricity production for this 63 h period is only slightly lower for the RCP4.5 scenario with 249.8 MWh compared to 252.9 MWh for the REF scenario. This is mainly due to the power curve of the applied wind turbine having a wide range of constant output power from the rated to the cut-out speed, as depicted in Figure 3a. In summary, this model validation demonstrates that the presented wind power model can be used to determine the output power and the corresponding power generation with a high temporal resolution. Figure 5; Figure 6 also show that the impact of extreme weather events on wind power generation can be investigated with this simulation model.

3.2. Wind Power Generation

After model validation, the numerical simulations were carried out for the 22 onshore wind turbines with a total installed capacity of 65.5 MW in the German Bight. The wind power generation was simulated for the REF scenario using past weather data and for the RCP2.6, RCP4.5, and RCP8.5 scenarios using climate data. Figure 7 shows the determined power generation from the turbine ensemble for the REF and RCP8.5 scenarios both for the high-wind year 2008 and the low-wind year 2010.

Figure 7 shows that the differences between the REF and RCP8.5 scenarios are not very high for both years and significantly lower than the differences of the monthly power generation between these years. In addition, it is visible both for 2008 and 2010 that most of the electricity is produced during the winter months, which is typical for onshore wind turbines located in Germany. It can also be seen from Figure 7 that for both years the relative differences between the REF and RCP8.5 scenarios are slightly higher in summer, i.e., in July, August, and September, than in the other months. Figure 8 depicts the cumulated differences of the power generation from the turbine ensemble in the German Bight for the RCP2.6, RCP4.5, and RCP8.5 scenarios compared to the REF scenario for the investigated years.

It can be seen that for both years the cumulated differences are in the same order of magnitude, although the total electricity production for the year 2008 is much higher than for the year 2010. Furthermore, it is depicted in Figure 8 that there is a clear trend of decreasing power generation with increasing RCP, and this trend is similar for both years. It also becomes obvious that there are only small differences in the power generation between the RCP2.6, RCP4.5, and RCP8.5 scenarios and the REF scenario for the investigated years. One important reason for this behavior is that a typical wind turbine produces the same amount of electricity over a wide range of higher wind speeds, i.e., from the rated to the cut-out speed according to its power curve. A further reason is that the climate data provided by GERICS for the applied RCPs do not differ significantly from each other. This might be caused by the inability of the applied climate models to create climate data with the spatiotemporal sensitivity required for the enhanced wind power model to fully indicate the local impacts of climate change.

4. Conclusions

It could be clearly shown in this study that the impacts of climate change and extreme weather events on wind power generation can be investigated with the presented physical simulation model using wind turbine data and climate data modified for different RCPs. In this context, it was shown that the wind power generation of the investigated plant ensemble with an installed capacity of 65.5 MW decreases with increasing RCP to values of up to nearly 3 GWh both for the high-wind year 2008 and the low-wind year 2010. Caused by the power curves of wind turbines and the limited sensitivity of the provided climate data, the effects on power generation from onshore wind turbines are not as strong as one might assume. According to GERICS, the used methods and climate models follow the fifth IPCC cycle, which could be a reason for the limited sensitivity. Nevertheless, the model approach presented in this paper clearly demonstrates that realistic disaggregated power generation data can be created using the enhanced wind power model. In addition, it could also be shown that the simulation results agree well with the measured power generation data of an existing wind turbine, as depicted in Figure 4. In summary:

• Realistic wind power simulations can be performed with the presented physical model;
• A clear trend of decreasing wind power generation with increasing RCP could be shown for the investigated years of 2008 and 2010;
• The local effects of climate change on wind power generation are not as strong as one might assume.

Furthermore, this presented approach allows the interested community to develop its own wind power models based on the ideas and information provided in this study. The enhanced wind power model can also be applied to other regions if the necessary input data are available. Many other energy studies can benefit from such highly resolved wind power generation data created by this physical model, including those focusing on energy system modeling [42,43] and also spatially resolved investigations on existing and future power systems [14,16].