Electricity spot market : game-theoretical approach

Abstract

We study models where a finite number of profit-maximising generators compete in the electricity spot market. Each generator chooses as its strategy a supply function, which indicates how much a generator is willing to produce for any positive price. Generators simultaneously submit these supply functions to a system operator, who solves an allocation problem and clears the market. We start with a study of a particular parameterized supply function bidding game which was proved to possess certain good properties in the case of a single market. Our first model is an extension of this game for the the case of several, in particular two, markets. First, for any given topology, Nash equilibrium existence and uniqueness are characterised. We then focus on a case with symmetric players and establish the price of anarchy bounds. Further, we compare resulting social welfare across models with the same players but under different topological structures. A Braess paradox is detected and investigated for the symmetric case. The second model is a proposal of a novel payment rule, which leads to the non-uniform price auction. We prove that truthful bidding is a dominant strategy and show the close relation of the proposed clearing rule in the context of supply function bidding with the Vickrey-Clarke-Groves (VCG) auction. Some further analytic results are obtained for the symmetric case. We conclude with a case study with actual data for the UK electricity spot market in 2014/2015. Under some simplifications, we simulate the supply function game under different mechanism designs, and analyse and compare outcomes across models.