Modelling of bidding strategies on the day-ahead electricity market

Abstract: Electricity generation coming from variable renewable energy sources is and has beenincreasing significantly. Because of the unpredictability attached to their behavior, fossil fuelpower plants and other renewables have to offer a higher degree of flexibility compared to thepast. In Germany, one of the countries with a greater involvement in renewable energygeneration, the share of renewable energy sources has grown from 14.6% in 2012 to a 29.2%in 2018. Consequently, fossil fuel power plants are forced into more start up and downoperations, making cycling costs higher.This thesis, under the project of improving the agent-based simulation of the DLR’s Instituteof Engineering Thermodynamics of Stuttgart, AMIRIS, contributes to the assessment ofcycling costs in the bidding of power point operators by creating new bidding strategies forAMIRIS that cover the aforementioned cycling cost. These strategies are designed with theoverall objective of maximizing the power plant operator’s profit.The new designed bidding strategies are first implemented in a Python simulation beforebeing introduced in the agent-based environment. Consequently, results of both simulationsare presented in this report, always paying close attention to the comparison between biddingwith these new strategies and bidding under short term marginal cost.In AMIRIS, when applying these new strategies to a very small part of the power plant park,the day-ahead electricity market clearing is almost identical, and these new bidding strategiesyield higher profit to the power plant operator than bidding under short term marginal cost.On the contrary, when applying these new designed bidding strategies to the whole powerplant park, the marginal clearing price suffers greatly from non-accounted interaction betweenthe operators’ bidding and its impact on the clearing of the market.