Chapter 68 - An Agent-Based Approach to Modeling Electricity Spot Markets

Abstract

This chapter proposes an agent-based approach characterizing auction-style electricity spot market dynamics. This method captures the learning and decision making of rational agents which cannot be modeled using a supply–demand equilibrium approach. It is found that because the auctions are repeated, an agent improves its decisions through time or through adaptive learning. It is observed that instead of elaborately searching for an optimal bidding strategy for each agent, some sensible and practical bidding strategies are assigned to the agents. The proposed two-step decision process simplifies the decision to determine bidding quantity and price separately. One can consider the capacity withholding strategy as a strategy used in a Cournot game, in which generators compete for their best market shares, and also the price setting strategy is considered as a strategy used in a Bertrand game. This strategic behavior depends highly on information available to the agents before a bidding decision is made. The model mainly focuses on understanding the role of learning by the generators and, accordingly, on their strategic bidding behavior in the market. The simulations show that the generators are able to execute strategic behavior in several market conditions, especially when total supply is reduced.