Price dynamics and market power in an agent-based power exchange

摘要

This paper presents an agent-based model of a power exchange. Supply of electric power is provided by competing generating companies (GENCOs), whereas demand is assumed to be inelastic with respect to price and is constant over time. The transmission network topology is assumed to be a fully connected graph and no transmission constraints are taken into account. The price formation process follows a common scheme for real PEs: a clearing house mechanism with uniform price, i.e., with price set equal across all matched buyer-seller pairs. A single class of GENCOs is considered, characterized by linear cost function for each technology. GENCOs determine their supply functions according to production costs. However, an individual reinforcement learning algorithm characterizes GENCOs behaviors in order to attain the expected maximum possible profit in each auction round. The paper investigates how the market competitive equilibrium is affected by market microstructure and production costs.