Increasing the realism of electricity market modeling through market interrelations

Being able to model electricity market prices realistically is strongly beneficial for both market participants and decision-makers who shape the design and regulation of power markets. This is not an easy task, given the complexities of today’s interrelated power markets. Here, we present and evaluate the open-source model flexABLE that models electricity markets using agent-based simulation. We show how the agents’ bidding strategies and the modeled market interrelations increase the realism of simulated prices. Agents can submit multiple bids at the energy-only market, representing their inflexible and flexible parts of the generation capacity. We show that the introduction of additional markets, such as a control reserve market and a district heating market, further increases result accuracy. Besides, negative market prices similar to those observed in real-world markets are seen in the simulation outputs when these markets are added. The full feature model achieves an RMSE of 8.43 €/MWh, and an MAE of 5.36 €/MWh for 2020, when compared to EPEX Spot day-ahead historic prices. The Pearson’s correlation coefficient of modeled and actual prices is 0.94. The simulated prices are much closer to real prices than those from economic dispatch or unit commitment models, both in terms of error metrics and in the observed minimum and maximum prices for 2020.