A dynamic model of the electricity generation market

Abstract

This thesis proposes that the bidding process that occurs daily in the competitive short-run power market can be modeled as a dynamic system, or a dynamic game played by electricity generators. Such a game is a finitely repeated one of complete but imperfect information. In this thesis, a dynamic model representing a small shortrun power market is formulated as a repeated game. Daily price competition provides sufficient information for the generators to estimate their bids. The next bids of each generator are proposed as functions of the previous and current bids. Results from the dynamic model show that the generators' bidding strategy affects the dynamics of the modeled power market. Different strategies yield different market clearing price patterns. Moreover, a step-supply function bid, in which an offer price relates to an offer quantity by a marginal-cost function, and the maximum available capacity of each generator, can cause inefficient prices in some scheduling periods and/or might result in inefficient dispatches in each scheduling day. In addition, depending on the bidding strategy that is uniformly applied to the model, although there is certainty of inelastic anticipated demand in the model, the repeated bidding processes tends to allow the generators to "learn" from the market so that they can tacitly collude to create demand deficiency. It is further suggested here that when demand deficiency unexpectedly occurs in peak-load periods, a real-time market (i.e., an hour-ahead market) might be needed so that supply always meets demand.