Agent Based Modeling of Germanys Electricity System Transition Pathway

Kurzfassung

The accelerating energy transition demands the rapid deployment of sustainable generation, storage, and flexibility technologies. In practice, the decision to invest in new capacity is taken by a heterogeneous set of market participants each operating under distinct assumptions and risk preferences. Most conventional capacity expansion frameworks model these actors as a single, perfectly rational planner that maximizes a system wide objective function. Such formulations, however, fail to capture the bounded rationality and path dependent investment behavior that shape real world outcomes. To address this gap, we extend the electricity market dispatch model AMIRIS with an investment module. The investment module explicitly represents the decision logic of diverse investors with deviating assumptions by using an agent-based modeling (ABM) approach. Furthermore, the granular dispatch simulation of AMIRIS enables the detailed modeling of all relevant system technologies. The model is calibrated with data on the German energy system drawn from recent energy scenario studies, mainly ARIADNE. Technology cost data is taken from the Danish Energy Agency. With this, we ensure that technology cost trajectories and demand forecasts are consistent with the most widely used reference pathways. We complete the parameterization with assumptions about investor-related parameters, such as return requirements. We apply the ABM to simulate the transition pathway of the German electricity system until 2045. A systematic sensitivity analysis explores the impact of key uncertainties, most notably technology cost trajectories for renewable generation and storage technologies, on investment timing and technology mix. Results reveal that investment decisions are highly sensitive to cost assumptions: modest variations in projected turbine or PV module prices shift the build out sequence and alter the relative importance of flexibility assets. Comparative evaluation against the original studies, which employed optimization based capacity expansion models, highlights systematic divergences. For example, stranded assets occur and lead to a different build-out timeline. In conclusion, the agent based approach generates transition pathways that differ from those derived from traditional optimization models. This underscores the importance of capturing market heterogeneity and realistic revenue dynamics when planning the expansion of sustainable energy technology. The findings suggest that policy design should account for these behavioral nuances to avoid misguided investment in critical energy resources.