The effect of pricing rules on a constrained wholesale electricity market: an agent based approach

[摘要] This thesis presents a comparative computational study of the performance of two different pricing mechanisms in a day-ahead wholesale electricity market, where performance is measured as the average level of payments made by the system operator to the generators. It focusses on two key pricing mechanisms: a uniform price based buy-back pricing model, defined as a short run approximation to the market design in Great Britain, and a nodal pricing model based on Locational Marginal Pricing rules. The research uses a game theory based approach for modelling themarket, allowing multiple rounds of the game to be played and statistically reliable results to be obtained. The research develops an agent based simulation of the day ahead markets for both of the pricing mechanisms, and is simulated on a constrained electricity grid. The agents developed for the simulation each represent a generator and are designed to be profit maximising with respect to a parent generation company. Agents employ an evolutionary algorithm in order to create optimised bids for the generation of electricity based on the current market state. Simulations of the market are performed using a styalised 29-Node transmission grid. A series of experiments are performed comparing the performance of the nodal and buy-back pricing mechanisms, under a series of different operating conditions. It is seen in all of the observed cases that the nodal market design averages a higher level of payments to its participants, and the indication is that the agents in a nodal market are able to explore the higher risk strategies more profitably than their buy-back counterparts. This work also highlights the value of creating evolutionary agents that are robust and flexible in analysing market designs. This research demonstrates that the greater the level of competition in a market the more efficiently market participants act. In addition to this the agents competing with a uniform based Buy Back pricing system appears far more restricted by higher levels of competition than their Nodal counterparts.

[发布机构] University:University of Birmingham;Department:School of Computer Science

[关键词] H Social Sciences;HB Economic Theory