Campuses:

Models for the Electric Power Grid

Sunday, March 7, 2004 - 9:15am - 10:15am
Keller 3-180
Christopher DeMarco (University of Wisconsin, Madison)
While sharing a number of broad qualitative features with problems in control and resource allocation for other large scale networks such as the internet, electric power grids present a range of unique challenges. Three major technological characteristics distinguish control and pricing problems in electric power: (i) the commodity being delivered is inefficient to store, so production tracks consumption across the entire interconnected grid, nearly instantaneously; (ii) the majority of branches in the delivery network are passive elements, with branch flows dictated by nonlinear relations to nodal boundary conditions, rather than by direct control; (iii) many constraints on operation represent physical limits whose violation can yield costly equipment damage and threats to human safety. Adding to the complexity of analysis is the U.S. electric power system's uneven regulatory policy transition, in which certain physical elements contributing to grid control operate in competitive markets (generators), while the others (e.g., switched capacitor banks, adjustable tap transformers) operate under the authority of regulated regional transmission monopolies.

This tutorial will give an overview of the mathematical models used to predict both dynamic and steady state performance of physical quantities in the electric power grid. Starting from the nonlinear constraints on network power flow, and the nature of financial offers and bids for electric power production and consumption, the relation of so-called locational marginal prices to an underlying optimization formulation will be reviewed. Issues in developing effective offer and bid strategies from these locational prices, and related issues for setting regulatory structures to govern these, will be highlighted. The tutorial will conclude with an overview of techniques for predicting cascading failure phenomena. Research to improve these techniques could play a key role in balancing operational strategies that favor efficiency under normal conditions, versus strategies that favor mitigation of risk of extremely high cost, low probability failure events such as the eastern U.S. blackout of August 2003.