Long-term Agent-based Electric Power System Simulation Models

Many electricity generation systems are undergoing drastic changes through rapidly increasing shares of intermittent renewable energy sources (RES) like wind and solar photovoltaics, as well as the prevalence of transport electrification. Models are often used to address the corresponding challenges (technical, economic, policy), by analyzing system operation and planning. The energy system is characterized by multiple actors (generators, system operators, regulators, suppliers, and consumers). These actors interact with one another, in different ways (e.g., through regulation and policies or in a market), at various levels in the system. Each actor has its own specific objectives, can make certain decisions, is facing boundary conditions, and derives expectations towards the future. Furthermore, actors will adjust their behavior based on experiences and interactions with other players. This setting allows for setting up and deploying so-called agent-based models. Such models depart from pure optimization and/or equilibrium models as they try to capture the actual behavior of actors based on the elements mentioned above. The overall aim of this PhD is to develop an agent-based modeling framework for improving investment decision-making algorithms in the energy system. The advantages of agent-based models will be leveraged to have more realistic and detailed descriptions of the behavior of different agents, bounded rationality, and market design than is possible in equilibrium models. First, an extensive literature survey will be conducted, and a core agent-based model framework will be implemented. Next, existing investment decision-making algorithms and their metrics are reviewed and evaluated. Third, prospect theory is deployed to capture bounded rationality in power plant investment decision-making. Fourth, the power plant investment decision-making algorithm is extended to synthesize three major market segments. To summarize, in short, the envisaged research will contribute to building energy system modeling toolboxes, which will then be used to support energy policy decision-making.