Strategic bidding in a day-ahead market by coevolutionary genetic algorithms

In the present work, the problem of energy market price clearing and generation company (Genco) strategic bidding is considered in the framework of existing day-ahead markets with system marginal price auction. The situation of imperfect competition arising when one of the Gencos is large enough to exert market power is considered in detail, showing what bidding behaviors are to be expected when such a market arrangement occurs. The impact that inter-area transmission system congestions may have on the mechanism of system pricing is also addressed. The bidding problem faced by each Genco is formulated as a strategic multi-player game in which the choice between different bidding levels and energy amounts to be sold at the market has to be made. The large size of the problem due to the number of competitors and to the presence of transmission constraints makes the application of classical game theory troublesome. Therefore, an agent based method belonging to the category of coevolutionary genetic algorithm was selected for the solution of this problem. Test cases illustrate the different strategies that the Gencos may implement to optimize their performance at the day-ahead market. Beside some small didactical examples, the situation of the Italian day-ahead market is considered in detail.