Supply function equilibrium analysis for electricity markets

The research presented in this Thesis investigates the strategic behaviour of generating firms in bid-based electricity pool markets and the effects of control methods and network features on the electricity market outcome by utilising the AC network model to represent the electric grid. A market equilibrium algorithm has been implemented to represent the bi-level market problem for social welfare maximization from the system operator and utility assets optimisation from the strategic market participants, based on the primal-dual interior point method. The strategic interactions in the market are modelled using supply function equilibrium theory and the optimum strategies are determined by parameterization of the marginal cost functions of the generating units. The AC power network model explicitly represents the active and reactive power flows and various network components and control functions. The market analysis examines the relation between market power and AC networks, while the different parameterization methods for the supply function bids are also investigated. The first part of the market analysis focuses on the effects of particular characteristics of the AC network on the interactions between the strategic generating firms, which directly affect the electricity market outcome. In particular, the examined topics include the impact of transformer tap-ratio control, reactive power control, different locations for a new entry’s generating unit in the system, and introduction of photovoltaic solar power production in the pool market by considering its dependencyon the applied solar irradiance. The observations on the numerical results have shown that their impact on the market is significant and the employment of AC network representation is required for reliable market outcome predictions and for a better understanding of the strategic behaviour as it depends on the topology of the system. The analysis that examines the supply function parameterizations has shown that the resulting market solutions from the different parameterization methods can be very similar or differ substantially, depending on the presence and level of network congestion and on the size and complexity of the examined system. Furthermore, the convergence performance of the implemented market algorithm has been examined and proven to exhibit superior computational efficiency, being able to provide market solutions for large complex AC systems with multiple asymmetric firms, providing the opportunity for applications on practical electricity markets