Abstract:
Erneuerbare Energien spielen eine wesentliche Rolle bei der Bekämpfung des Klimawandels, da sie das Potenzial haben, die Kohlendioxid Emissionen im Energiesektor deutlich zu reduzieren. Mit dem steigenden Anteil erneuerbarer Energien und der zunehmenden Elektrifizierung werden Energieangebot und nachfrage jedoch stärker von Wetterbedingungen abhängig. Wetterabhängige erneuerbare Energiequellen wie Wind und Solar unterliegen natürlichen Schwankungen, was Unsicherheiten im Energiesystem mit sich bringt. Gleichzeitig wird auch die Energienachfrage vom Klima beeinflusst, da Temperaturveränderungen den Bedarf an Heizung und Kühlung verändern. ... mehrDa extreme Wetterereignisse infolge des Klimawandels häufiger werden, ist das Verständnis des Einflusses klimatischer Variabilität auf sowohl die erneuerbare Stromerzeugung als auch die Stromnachfrage entscheidend für die Analyse von Energiesystemen.
Um diese Einschränkungen zu überwinden und die Lücke zwischen Wetterdaten und Energiesystemanalyse zu schließen, konzentriert sich diese Dissertation auf zwei wesentliche Eingangsdaten für Energiesystemmodelle: Windgeschwindigkeit und Stromnachfrage, und untersucht deren Einfluss auf die Systemresilienz. Die Dissertation bringt methodische Beiträge zur Bewältigung großer Unsicherheiten in der Energiesystemmodellierung. Sie führt einen neuen maschinellen Lernansatz für hochauflösendes Windgeschwindigkeits-Downscaling ein und erweitert die Anwendung von TRFs zur Projektion der Stromnachfrage auf große geografische Skalen. Zudem werden erstmals die dynamischen Eigenschaften von TRFs unter sozioökonomischen Faktoren untersucht. Schließlich konzentriert sich die Studie auf die Erstellung synthetischer Wetterjahre für extreme Residual-Last-Ereignisse und bietet Energiemodellierern eine Orientierung zur Auswahl oder Synthese von Wetterjahren für die Optimierung und Simulation robuster Systeme.
Gemeinsam verbessern diese drei Beiträge die Zuverlässigkeit der Energiesystemmodellierung, indem sie Unsicherheiten in den Eingangsdaten verringern, die Nachfrageprojektion verbessern und die Robustheit des Systems gegenüber extremen Wetterereignissen bewerten. Diese Arbeit bietet wertvolle Einblicke, wie klimatische Variabilität und extreme Wetterereignisse zukünftige Energiesysteme gestalten und bietet eine Grundlage für eine robustere Energieplanung und -politik.
Abstract (englisch):
Renewable energy plays a crucial role in mitigating climate change by significantly reducing carbon emissions in the energy sector. Over the past decades, many regions have transitioned toward renewable energy, increasing the share of renewables in global power generation and driving electrification in modern energy systems. To understand, analyze, and optimize these evolving energy systems, energy systems modeling is essential, as it can provide insights and guidelines for policy decisions. However, the reliability of energy systems modeling is often challenged by the uncertainties, particularly concerning weather variability, future electricity demand, and extreme weather events. ... mehrThese uncertainties in input data pose significant challenges for energy systems modeling and can greatly impact the robustness of system designs. Therefore, this dissertation aims to improve the accuracy and reliability of energy systems modeling by addressing these uncertainties.
One of these uncertainties is the spatial resolution of meteorological input data. This is particularly critical for wind speed, which is highly sensitive to spatial resolution due to its strong local variability influenced by topography. Therefore, this dissertation focuses first on improving the spatial resolution of wind speed data, which is a core input data in energy systems modeling. By applying a machine learning–based statistical downscaling technique, coarse-resolution reanalysis data are downscaled into high-resolution, hourly wind speed time series. The improved data increases the accuracy of renewable energy generation modeling, enables local-scale energy system analysis, improves the assessment of extreme weather events at regional scales, and builds the foundation for more accurate and detailed energy system models.
The main climate change induced uncertainty on the demand side is considered next. Electricity demand is highly sensitive to temperature variations and is influenced by socio-economic factors such as the electrification of demand sectors and energy efficiency improvements in buildings. The evolving patterns of these factors introduce significant uncertainty in accurate demand projections. To better capture the dynamic nature of electricity demand under different climate scenarios and policy implementations, this dissertation investigates future demand trends using Temperature Response Functions (TRFs), which characterize the nonlinear relationship between temperature and electricity demand. Therewith, the role of the building sector is accounted for through a novel piecewise regression model that considers the influence of space cooling, passive cooling, electrification, and thermal insulation. This improved approach enables a more realistic and robust projection of future electricity demand under varying climate and policy conditions.
The third uncertainty addressed in this dissertation concerns extreme weather events. As the frequency and intensity of such events increase, it becomes increasingly important to integrate them into energy systems modeling. By exploring how extreme weather events impact energy system reliability, this dissertation aims to identify the key factors contributing to robust energy system design. To achieve this, methods for generating synthetic extreme weather years that capture critical extreme weather conditions are proposed. These synthetic extreme weather years are then used as input data for a sector-coupled energy system optimization model to investigate their impact on energy systems. This comprehensive approach significantly improves the current literature. The findings reveal that dispatchable generation technologies play a crucial role in ensuring system robustness and that short-term extreme events significantly influence the required dispatchable generation capacities.
The novelty of this dissertation lies in its methodological contributions to addressing these three major uncertainties in energy systems modeling. For high-resolution wind speed data, this dissertation presents a novel machine learning–based statistical downscaling approach. In contrast to commonly used methods such as computationally intensive dynamic downscaling or distribution-based methods, the proposed method enables efficient, high-resolution wind speed downscaling over large geographical areas while requiring minimal computational resources. For electricity demand projection, this dissertation expands the application of TRFs, which are commonly used in the literature but rarely explored at large geographical scales. More importantly, while most studies assume that TRFs remain static over time, this work is the first to investigate the dynamic characteristics of TRFs under various socioeconomic factors. Finally, for robust energy system design, this study focuses on generating a comprehensive volume of synthetic weather years specifically tailored for extreme residual load events, a topic rarely discussed in the literature. By examining the close relationship between dispatchable generation, short-term extreme events, and system robustness, this dissertation provides guidance for energy modelers on selecting or synthesizing weather years for optimization and simulation models.
Together, these three contributions enhance the reliability of energy systems modeling by reducing uncertainties in meteorological input data, improving demand projection, and evaluating system robustness against extreme weather events. This work provides valuable insights into how climate variability and extreme weather shape future energy systems, offering a foundation for more robust energy planning and policy development.
