Locational (In)Efficiency of Renewable Energy Feed-In Into the Electricity Grid: A Spatial Regression Analysis.

• Author: Hofer, Tim

1. INTRODUCTION

   In order to mitigate climate change, governments all over the world have been transforming their electricity generation systems from conventional power plants to renewable energy sources (RES). This shift in the electricity generation structure is having severe impacts on the entire electricity system, especially on the grid infrastructure. The challenges arise from central power plants with a steady electricity production being replaced by a multitude of small and dispersed renewables with variable electricity generation. (1) Renewables are primarily connected to the distribution grid, whereas coal-fired or nuclear power plants are connected to the transmission grid. (2) At times of high renewable energy output, this can lead to bidirectional power flows and grid overstress. To prevent the failure of electric network components, the responsible system operator (SO) can decrease the electricity output of power plants. If doing so, the SO is obliged to first reduce the output of conventional power plants--the so-called redispatch. (3) Only if the conducted redispatch measures are insufficient to relieve the bottleneck in the electricity grid is the SO allowed to curtail the output of renewables--the so-called RES curtailment. (4) If the SO does reduce the output of a renewable generator, it has to refund the operator of the generator in question for the restrained electricity. The costs for the curtailment comprise the curtailment compensation for the respective renewable energy technology and the difference between the potential and the realized electricity supply. The SO is authorized to pass on the costs resulting from this RES curtailment to the consumers in the region concerned.

   The need for RES curtailment in Germany has increased tremendously over the past years. In 2017, the reduced output of renewables reached 5,518 GWh, whereas it amounted to 555 GWh in 2013 (BNetzA, 2018a). The associated costs for RES curtailment totaled [euro]43.7 million and [euro]610 million in 2013 and 2017, respectively (BNetzA, 2018b). Onshore wind turbines are by far the most frequently and heavily curtailed renewable energy technology in Germany, accounting for 80.8% of the overall quantity of RES curtailment (BNetzA, 2018a). In 2017, 89% of the implemented RES curtailment measures were due to a bottleneck in the transmission grid (BNetzA, 2018a). Furthermore, 89% of all RES curtailment measures were instructed by transmission system operators (TSOs) and conducted by distribution system operators (DSOs) (BNetzA, 2018a). In our analysis, we consider RES curtailment conducted by TSOs and DSOs alike. Most of the implemented RES curtailment measures occurred in northern and eastern Germany, where a high amount of installed renewable energy capacity meets a relatively low electricity load. Figure 1 depicts the amount of RES curtailment compared with the installed capacity of wind energy systems (BNetzA, 2017a, 2018a).

   This study aims at identifying the main drivers for RES curtailment measures and at explaining the regional variability of RES curtailment costs. For this purpose, we analyze the RES curtailment costs of four DSOs in Germany. (5) Since the DSOs publish only general information on their curtailment measures--such as the time and duration of a measure and the type of the curtailed renewable energy technology--but not on the curtailment costs, we have calculated the curtailment costs ourselves. To explain the locational differences of the RES curtailment within the DSO regions, we partition the DSO region into smaller subregions based on the high-to-medium voltage substations (Egger, 2017; Hulk et al., 2017). We then allocate all variables to these subregions and apply a two-step Heckit sample selection model (Heckman, 1976). The first part is a probit model that corrects bias from non-randomly selected samples. The second part is a linear model that captures cross-sectional dependence via spatial lags of the explanatory variables (SLX) and correlated common effects (CCE). Whereas the first part explains the impact of the explanatory variables on the probability of occurrence of RES curtailment, the second part illustrates the effect of the variables on RES curtailment costs. More specifically, we quantify the increase in RES curtailment costs due to an additional megawatt (MW) of installed renewable energy capacity and an additionally generated gigawatt-hour (GWh) of electricity.

   The regional RES curtailment costs are integration costs of renewables in the present, rather inflexible, electricity system. Hirth et al. (2015) classify such integration costs into three categories: grid-related, balancing and profile costs. According to this definition, grid-related costs reflect the marginal value of electricity in different regions and refer to opportunity costs of transporting electricity from the place of generation to the place of consumption. Balancing costs are defined as costs that arise due to forecasting errors of future weather conditions. Profile costs reflect the costs resulting from matching electricity supply and demand and arise due to the variability of the output of renewables. A wide variety of literature exists that investigates the system integration costs of renewables in general and grid-related costs in particular. We refer the interested reader to the publications by Hirth (2015), Hirth et al. (2015), Holttinen et al. (2013), Holttinen et al. (2011), Milligan et al. (2011), and Smith et al. (2007), which also provide reviews of studies on integration costs of renewables.

   In contrast to the large number of studies on grid-related integration costs, the number of studies that explicitly incorporate curtailment costs into their analysis is still limited. The following studies, which do incorporate curtailment costs, focus on calculating the costs instead of quantifying and explaining the regional correlation between installed capacity of renewables and RES curtailment costs. Ueckerdt et al. (2013) developed a mathematical definition of integration costs of renewables, comprising profile, balancing, and grid costs. As an example, the authors parametrize these cost components for increasing wind shares in Germany. A simple power system model is used to quantify the profile costs. The balancing and grid costs are parametrized based on a literature review. Grid-related costs incorporate investment costs in the electricity infrastructure as well as congestion management costs. However, the authors do not quantify or explain the reasons for the varying grid costs in different regions in Germany. Strbac et al. (2007) investigated the costs and benefits of wind energy in the United Kingdom using a generic model of the power system. Their simulation model explicitly incorporates the annual curtailed electricity output of wind turbines for different levels of wind capacities. The overall additional costs of integrating wind power consider balancing costs and grid costs, among others. The authors do not distinguish between different wind turbine locations and different grid costs at these locations. Denny and O'Malley (2007) conducted a cost-benefit analysis of wind energy systems, considering different wind turbine capacities, varying power plant mixes, and distinct electricity demand levels for the electricity system in Ireland. The model calculates the net benefits of wind power, including the curtailment and network reinforcement costs. No differentiation between different locations is made. Ecofys and Fraunhofer IWES (2017) qualitatively investigated the reasons for the occurrence of RES curtailment in all of the federal states of Germany. They conclude that the different proportions between installed wind and solar capacity and the load are the decisive factors for the varying amount of curtailed renewable electricity generation. Rural regions with high installed wind or solar capacities and low load experienced the highest amount of RES curtailment. In contrast, suburban and urban areas with little installed wind or solar capacities and a high load exhibit almost no RES curtailment. This aligns with the results of the studies by Agricola et al. (2012) and Buchner et al. (2014). The studies find that wind and solar power are the main drivers for the reinforcement of the distribution grid and that mainly low-load regions are affected by an increased overstress of the grid. The latter three studies find a positive qualitative correlation between the installed capacity of renewables and the overstress of the electricity grid as well as a negative correlation between the load and RES curtailment. However, the studies do not quantify the effect of renewables on curtailment costs. To the best of our knowledge, no comparable published study has so far quantified the impact of different renewable energy technologies on RES curtailment costs.

   In summary, the main merit of our paper is to quantify and explain the regionally diverging RES curtailment costs by means of an econometric model. The first part of our model elucidates why RES curtailment occurs only in some regions of Germany and not in others. The second part of our model analyzes the correlation of installed capacity and generated output, respectively, of renewables and RES curtailment costs. As part of our analysis, we also calculate the regionally disaggregated amount and costs of RES curtailment in Germany in a higher spatial resolution than available in official publications. These results could, for example, be used to introduce price signals that incentivize a welfare-enhancing deployment of renewables (Haucap and Pagel, 2013). Such price signals would, among other things, incorporate the regionally varying RES curtailment costs. Alternatively, the results of this study could be used to incentivize the reinforcement of the electricity grid or the further implementation of flexibility options in regions with high RES...