Comparison of different approaches for spatial complexity reduction in linear energy system optimization model using eTraGo

Kurzfassung

The provision of energy is fundamental to economic growth, societal development, and well-being. In this sense, electrical energy emerges as a versatile, safe, and scalable solution to tackle the global environmental issues like climate change and CO2 emissions. Germany is aiming to achieve net-zero greenhouse gas emissions by 2045 and is targeting sector coupling strategies to integrate electricity, gas, heat and transport into national power and gas grids. Policymakers and politicians are critical in creating and implementing regulations, policies, and laws based on the research carried out by specialized institutes and predictions of grid operators, and here the energy system models play a crucial role. Energy system models are mathematical representations that often use simplified linear equations to simulate the physical processes and interactions of a real-world energy system, e.g. national power and gas grids. Such models can be used to support the stakeholders in investment decisions or decision or decisions on the implementation of new regulations. A typical approach to address is to optimize the system's dispatch and its capacity expansion in different energy system scenarios. For large-scale energy system models this is a complex to solve model, demanding significant computational resources and lengthy solution times. To address this, techniques to reduce temporal and spatial complexity are employed, such as finding typical periods in the dataset or reducing the spatial complexity of the network components. The eTraGo framework is an example for an open-source linear optimization power flow (LOPF) tool that couples the electricity, gas, heat, and e-mobility sectors in Germany and is used to investigate the effects of coupling nonelectrical sectors on the power network. This research focuses on tackling spatial complexity reduction in eTraGo applying the k-means and k-medoids Dijkstra clustering algorithms at different levels of complexity. In this sense, the following research questions are posed. 1. In terms of spatial resolution, which algorithm cluster and preserves more critical features from the original network? 2. What is the effect when increasing the spatial resolution of the eTraGo network on the optimized parameters, and if there are significant changes, why does this occur? The present study comprises the following chapters: Chapter 2: Theoretical background. Address the theory behind k-means and k-medoids Dijkstra algorithms and their relationship to spatial complexity reduction in LOPF models, discussion of the advantages and disadvantages of the chosen algorithms, description of the eGon project and eTraGo tool along with its characteristics, previous work using this model and internal approaches to evaluate clusters. Chapter 3: Methodology. This chapter describes the steps followed to carry out the simulations. We first define the levels of spatial complexity to be simulated and settings for clustering the gas and electricity networks, then we define the four energy sectors with their respective variables based on the eGon2035 dataset and assumptions. Furthermore, we show the optimization results for the baseline model, and finally, the metrics to evaluate the results in terms of spatial resolution and optimized parameters are explained. Chapter 4: Results. This chapter reports the results of all the simulations performed with k-means and k-medoids Dijkstra algorithms for the four levels of complexity (LOCs) defined. Results for simulation times, clustering of the gas and electricity networks, the magnitudes of all parameters optimized by sector, and the results of the evaluation criteria, are presented. Chapter 5: Discussion of Results. This chapter analyzes the differences observed in the optimization of eTraGo with the two clustering methodologies. Which parameters are most affected by the increase in spatial resolution and why? Are there unexpected results? What do the metrics or evaluation indices tell us? Chapter 6: Conclusion and future Work. This chapter summarizes the main findings of the study and answers the research questions. Possible solutions or recommendations to overcome the limitations of the present study are presented for future research.