Modeling variable renewable energy systems

This dissertation addresses challenges arising in the modeling of variable renewable energy systems. A particular emphasis is placed on the role of weather variability and uncertainty, long-duration storage and the heating sector. Chapter 1 provides broader context and motivates the questions covered in subsequent chapters. Chapter 2 tests various configurations of a stabilized and parallelized Benders decomposition algorithm for a two-stage stochastic formulation of an energy system model subject to interannual weather variability, represented by potentially many historical weather years. The refined algorithm outperforms a standard Benders decomposition implementation by a factor of 100 and is faster than the deterministic equivalent when including six weather years or more. Chapter 3 explores the effect of limited foresight in the face of weather uncertainty in the operational decisions of a wind- and solar-based renewable energy system with long-duration storage. The latter hedges the system against extreme weather realizations. Solar generation gains in relative system value for its higher predictability and installed photovoltaic capacities increase by up to 25%, most significantly so in wind-dominated systems. Long-duration storage bids reflect the probabilityadjusted cost of an adverse system state. This probabilistic bidding behavior stabilizes system prices and alleviates concerns about extreme price disparity in renewable electricity systems. Chapter 4 investigates the impact of a large-scale electrification of the heating sector on long-duration storage capacities. Using a large dataset of 78 weather years, the analysis shows a significant average increase in long-duration storage requirements for a renewable European energy system. 75% of the increase is driven by a leverage effect, in which additional load amplifies periods of renewable scarcity. The remainder is due to a compound effect, where cold spells coincide with periods of low variable renewable energy availability. A mix of hydrogenbased long-duration electricity storage and long-duration thermal storage in district heating networks is found to be cost-optimal and policymakers should expedite the creation of regulatory frameworks for both storage technologies to safeguard and support a wide-spread heating electrification. Finally, an empirical analysis of the heating sector during the 2022 energy crisis complements the techno-economic lens of the preceding chapters. Chapter 5 uses a causal forest approach to estimate the behavioral natural gas savings of German households and businesses during the crisis. The analysis finds significant savings, despite incomplete price signals, that contributed to closing a looming supply gap.