Payment cost minimization auction for deregulated electricity markets

Date of Completion

January 2008


Engineering, Electronics and Electrical|Engineering, System Science




Deregulated electricity markets in the US currently use an auction that minimizes the total bid cost to select bids and their levels. Payments are then settled based on market clearing prices. Under this setup, consumer payments could be significantly higher than the minimized bid cost from auction This gives rise to "payment cost minimization," an alternative auction mechanism that minimizes consumer payments. We have previously presented an augmented Lagrangian and surrogate optimization framework to solve the payment cost minimization problem with uniform market clearing prices and without considering transmission. This approach is extended in the first part of this thesis to incorporate current market practices of locational marginal prices (LMPs) with the consideration of transmission constraints. DC power flow equations are used to describe transmitted power, and the LMP definition is characterized by Karush-Kuhn-Tucker conditions of economic dispatch to be embedded as constraints. A regularization scheme is adopted to satisfy constraint qualifications, and surrogate optimization is used to overcome difficulties caused by problem inseparability. Specific methods to satisfy the "surrogate optimization condition" in the presence of transmission constraints are highlighted. Numerical testing results demonstrate the effectiveness of the approach. ^ The second part of this thesis investigates market behaviors under the two auctions without transmission by using a game theoretic framework. Suppliers' bid strategies are discretized to form matrix games with Nash equilibrium as the solution concept. To reduce side effects caused by discretization, the "approximate equilibrium" is introduced to recover lost equilibria, and additional strategy samples are evaluated to eliminate artificially created solutions. Matrix games are solved by examining suppliers' payoffs obtained from running auction algorithms. Auction characteristics are explored to reduce the number of times for running auction algorithms. Numerical testing results show that under payment cost minimization, reduction in payments is achieved at a relatively small loss of production efficiency as compared to bid cost minimization. Also, the "hockey-stick" bidding behavior is found more likely to occur under bid cost minimization. ^