Conclusion and Outlook
In this thesis, an agent-based electricity market simulation model has been developed and applied to several research questions. The model comprises a day-ahead market for hourly contracts of electricity delivery, a balancing power market at which positive minute reserve capacities are procured by transmission system operators, and an exchange for CO2 emission allowances. Market participants are modeled as software agents who have learning capabilities, represented through reinforcement learning algorithms. The model has been run with data input from the German electricity sector (UCTE system's total load data and power plant portfolios that roughly correspond to those of the main players in the German power markets), and the prices resulting from the dynamic interaction of agents in the modeled markets are compared to prices observed at the corresponding real-world markets in Germany. The developed agent-based simulation model delivers realistic daily and seasonal courses of prices on the day-ahead electricity market and on the balancing power market. It can therefore be used for methodologically supporting questions of how to best engineer markets in the electricity sector. Some examples for this procedure have been calculated in this thesis, and conclusions for policy advice have been drawn.
The main contributions that have been achieved through the present work are summarized in Sect. 7.1. Finally, some suggestions for future work in the field analyzed with the agent-based simulation model are formulated in Sect. 7.2.
KeywordsElectricity Market Electricity Price Electricity Sector Emission Allowance Reinforcement Learning Algorithm
Unable to display preview. Download preview PDF.