Optimal Allocation of Variable Renewable Energy Considering Contributions to Security of Supply.

• Author: Peter, Jakob

1. INTRODUCTION

   The Paris Agreement aims at holding global warming to well below 2 degrees Celsius (United Nations, 2015), which requires deep decarbonization of the global electricity sector. Recent cost reductions suggest that the optimal pathway will be substantially based on variable renewable energy sources (VRE). As a consequence, global electricity markets will increasingly be influenced by generation technologies based on VRE such as wind and solar energy. Electricity generation from VRE differs from dispatchable power generation in its pronounced dependency on weather conditions. These weather-induced variations show spatial dependencies and are not perfectly predictable. Accordingly, there arise important implications for reliability of supply in power systems as electricity is only storable at comparatively high cost and the supply-demand balance has to be maintained at all times in order to prevent outages.

   Reliability of supply has always been a major concern in electricity systems as outages incur high economic losses. With increasing shares of VRE, reliability issues gain further importance due to the variability, spatial dependency and imperfect predictability of electricity generation based on VRE and the resulting risk of unavailability during times of stress (e.g., Cramton et al., 2013). VRE resources are typically less correlated on a wider geographical scope, which reduces fluctuations because of imperfectly correlated generation patterns at different locations. Subsequently, this will be referred to as balancing effects. Markets can benefit from these balancing effects via interconnections and cross-border cooperation. Envisaged reliability levels can thereby be reached at lower costs compared to reliability measures restricted to national borders (e.g., Cepeda et al., 2009; Hagspiel, 2017). Against this background, the following research question arises: What is the optimal mix and allocation of VRE capacity in order to benefit from balancing effects both in generation and contribution to security of supply to reach an envisaged reliability target?

   Assessing the contribution of VRE to security of supply is complex, because of the stochasticity of electricity generation based on weather-dependent resources. The ability of an additional VRE generation unit to provide secure capacity depends on the correlation of its electricity generation with electricity demand and with electricity generation from other units. As an example, consider a simplified case for wind energy: An electricity system has an off-peak demand of one and a peak demand of two, with off-peak periods being more frequent compared to peak demand situations. Additionally, there are two possible sites A and B for investment into wind capacities. Wind generation at site A is perfectly correlated with off-peak demand and wind generation at site B is perfectly correlated with peak demand hours. In this setting, wind capacities at site A supply more electrical energy because off-peak situations are more frequent. Nevertheless, wind investments at site B can be preferable because wind generation capacities at site B generate electricity in the critical peak demand situations. Thus, one unit of wind capacity at site B reduces the need for one unit of dispatchable capacity and therefore contributes to security of supply. Now consider the situation where there is already one unit of wind capacity in place at site B, which generates one unit of electricity in peak demand hours. The remaining residual demand, which must be supplied by dispatchable generation capacity, is one in off-peak and one in peak demand periods. As a result, installing one additional unit of wind capacity at site B cannot contribute to security of supply because firm capacity is still required in the off-peak demand period and thus cannot be substituted. However, if there were wind capacities of one unit installed at both sites, investing in one additional unit of wind capacity at site B would indeed contribute to security of supply.

   The highly stylized example clarifies that the marginal contribution to security of supply from additional generation capacities based on VRE depends on all existing installed capacities within the system as these capacities and their weather-dependent generation determine the critical residual demand situations. Typically, generation patterns of wind and solar power plants at different locations are positively correlated. Therefore, the ability of one unit of VRE generation capacity to substitute firm capacity, which is referred to as its capacity value (or capacity credit), (1) declines as the share of VRE in total generation increases. (2) Nevertheless, economic long-term simulation models for electricity markets, which are widely used in scientific and political practice, often assign fixed exogenous capacity values to wind and solar generation and neglect cross-border effects for reasons of simplification and computational tractability. Similarly, adequacy studies and capacity mechanisms often do not or only crudely allow for participation of VRE and are often confined to national borders. (3)

   The paper at hand develops a new methodology to endogenously determine the contribution of VRE to security of supply in a long-term partial equilibrium model for electricity markets. The proposed methodology builds on an iterative approach that captures the non-linear dependency of the capacity value of VRE on installed capacity and its spatial distribution considering cross-border cooperation via interconnectors. The methodology therefore determines cost-minimal investment into power plants, taking into account electricity generation as well as provision of security of supply of VRE, while keeping the computational tractability in a large-scale application. After introducing our methodology, we apply the model framework to a simple two-country example. Next we apply it to the European electricity system to determine an optimal decarbonization pathway up to the year 2050, starting from today's existing power plant fleet. Our analysis focuses on wind power; however the presented approach can be applied to all VRE technologies. We perform the analysis on a newly developed dataset based on meteorological reanalysis data featuring a high spatial and temporal resolution. The data is therefore well suited to optimally capture the stochastic properties of wind generation and the resulting contribution to security of supply.

   We show that the proposed methodology is capable of endogenously determining the capacity value of wind power in large-scale investment and dispatch models for electricity markets. The results of the large-scale application imply that wind power can substantially contribute to security of supply in a decarbonized European electricity system cooperating with respect to reliability, with an average wind power capacity value of 13% in 2050. Additionally the results show that the capacity value of wind power is heterogeneous across different regions and years, which is a result of varying wind conditions as well as increasing total installed capacities and technological innovation over time. Existing modeling approaches, which typically assign constant exogenous capacity values for wind power, therefore result in inefficient levels of dispatchable capacities required to guarantee security of supply in electricity systems with high shares of VRE. In our application for the European electricity system, the additional yearly costs for firm capacity provision (4) when applying exogenous fixed wind power capacity values of 5% compared to endogenous capacity values amount to 1.5 and 3.8 bn EUR in 2030 and 2050, respectively, which represents additional costs of 3% and 7%. Finally, our results suggest that European market integration can substantially improve the contribution of wind power to security of supply due to cross-border balancing effects.

   Our paper is mainly related to two streams of literature. The first relevant stream examines system adequacy and reliability of supply in electricity systems. Reliability of supply in electricity systems has been subject to extensive scientific research efforts, both from a technical as well as an economic point of view. (5) In particular, the contribution of individual technologies to system adequacy, i.e., the capacity value, has been a topic of interest. The probability theory of the capacity value of additional generation for the cases of statistical independence and dependence is presented in Zachary and Dent (2012). Based on these theories, various contributions investigate empirical methods to evaluate the capacity value of wind power in electricity systems. (6) Cepeda et al. (2009) investigate the positive implications of connecting different electricity systems on reliability and ways to internalize cross-border effects in a two-zone model. Hagspiel et al. (2018) introduce a comprehensive framework to investigate reliability in power systems consisting of multiple technologies and interconnected regions. All the mentioned studies focus on static analyses for given power systems. Consequently, the capacity value is not evaluated within a dynamic model, which determines the optimal future structure of an electricity system. This research gap is filled in this work by developing a dynamic framework, which builds on the framework introduced by Hagspiel et al. (2018) and combines it with an investment planning model through an iterative process.

   The second relevant literature stream focuses on the analysis of electricity systems with high shares of VRE based on long-term dynamic partial equilibrium models. Typical research questions within this literature are optimal decarbonization pathways for electricity systems or optimal allocation of renewable generation capacites. However, the contribution of VRE to security of supply is often only crudely accounted for by assigning...