Lab Home | Phone | Search
Center for Nonlinear Studies  Center for Nonlinear Studies
 Colloquia Archive 
 Postdoc Seminars Archive 
 Quantum Lunch 
 CMS Colloquia 
 Q-Mat Seminars 
 Q-Mat Seminars Archive 
 Kac Lectures 
 Dist. Quant. Lecture 
 Ulam Scholar 
 Summer Research 
 Past Visitors 
 History of CNLS 
 Maps, Directions 
 CNLS Office 
Wednesday, April 15, 2009
1:30 PM - 3:00 PM
CNLS Conference Room (TA-3, Bldg 1690)


Can we quantify the risk of cascading failure blackouts with branching processes?

Ian Dobson
University of Wisconsin-Madison

Blackouts of the electric power transmission infrastructure are complicated cascading events in a huge network with diverse, interacting failures. In these cascading failures, a series of dependent failures successively weaken the system and making further failures more likely. The cascading causes power law and criticality phenomena in blackout statistics. One contention is that we should study not arbitrary networks, but engineered networks, and we outline a complex systems simulation approach to generate "engineered" data when this data is not otherwise available. We model cascading in a bulk statistical fashion as initial failures propagating probabilistically according to a branching process. We estimate branching process parameters from data and hence estimate the probability of cascading failures of various sizes. Initial testing of these methods on real and simulated data open the possibility that the probabilities of large blackouts could be practically estimated from power system observations or non-exhaustive simulation runs.

Host: Misha Chertkov, T-4