A Multi-Agent Based Approach to Modeling and Analyzing Spot Electricity Markets

Wednesday, November 5, 2003 - 11:45am - 12:20pm
Keller 3-180
Marija Ilic (Carnegie-Mellon University)
This talk is based on the joint work with Dr. Poonsaeng Visudhiphan and my current graduate student Zhyiong Wu. In the first part of this talk a model derived in the PhD thesis of Dr Visudhiphan is briefly reviewed, and simulations are presented to illustrate electricity prices resulting from this model under several bidding strategies. The results are analyzed and compared with the results obtained using two well-known AI learning techniques. An important conjecture concerning equilibrium results as a function of system demand level is stated and illustrated. The second part of the talk highlights the importance of physical constraints on bidding strategies and market power. In particular, it is shown how the unit-commitment constraints, reflected in start-up and shut down costs and rates of various generation technologies, are likely to make the bidding logic much more difficult, and almost impossible. Finally, a generalized definition of market power in electricity markets as a measure of market inefficiencies with the physical constraints accounted for is suggested. Physical and financial risks are assessed using this generalized notion.