Electricity market analysis with co-evolutionary computation

Abstract

The unique feature of the electricity power industry makes the power market more akin to an oligopoly. A great deal of work has been performed on analyzing power markets using the oligopoly models in the literatures. The Supply Function Equilibrium (SFE) and Cournot models have also been widely employed to model the electricity market. Many techniques, such as empirical analysis, agents-based simulation, iterative Nash Equilibrium (NE) search algorithms and complementarity program methods, have been utilized to determine the power market equilibrium. Despite of these previous efforts, no method is widely recognized as being an effective method for market equilibrium determination. Hence there is an urgent need to develop an effective and powerful approach for power market analysis and equilibrium determination. Co-evolutionary computation is developed from traditional Evolutionary Algorithms (EAs), which simulates the co-evolutionary mechanism in nature and adopts the notion of ecosystem. It is a new methodology used to simulate the bidding behavior of the market players and to determine market equilibrium. The thesis applies co-evolutionary computation algorithms (CCAs) to solve the power market equilibrium and to study several important issues in power market analysis. The first issue is that when transmission constraints are considered, the profit function of Generation Companies (GenCos) may be nonconcave and discontinuous with many local optima. Determination of market equilibrium becomes a more challenging task. A two-level evaluation process is developed in the thesis for the determination of the equilibrium. The Linear Supply Function Equilibrium (LSFE) and the Cournot market models are employed and the market equilibrium is determined by CCA. The existence or non-existence of the NE due to the transmission and generation capacity constraints are illustrated using a 2-bus test system and the IEEE 30-bus system. The effects of different parameters settings of the LSFE model on the equilibria are also studied and compared with those found based on the Cournot model. The second issue examined is to determine the market equilibrium in a multiple pricing period. When compared to actual markets, earlier research works study only a single pricing period market. A multiple pricing period market with inter-temporal constraints should also be studied to make the simulation results comparable to actual market settings. The CCA method used to study single pricing period market is then extended to multiple pricing period market analysis. It is found that the market outcome of a GenCo using a constant supply function across multiple pricing period is contrasted with the case that GenCos using a specified supply function in each pricing period. It is important to observe outcomes of different markets in which obligations for consistent bidding are set up or not. Owing to the recent California electricity crisis, more and more researchers are convinced that forward market plays an important role for market power mitigated in electricity markets. In this thesis, the issue of whether rational GenCos would voluntarily enter forward markets or not is examined and the factors which could affect the bidding behavior are studied. The thesis formulates a two-settlement electricity market as a two-stage game. The LSFE model and Cournot model are used to model strategic bidding for the spot market, while the forward market is modeled by the Cournot model. GenCos' bidding behaviors are analyzed by CCA and two numerical examples are used to verify the theoretical analysis. From the works undertaken in this thesis, it is found that the CCA method is robust and flexible and has the potential to be used to solve the complicated equilibrium problems in real-world electricity markets.