UVM Theses and Dissertations
Format:
Print
Author:
Cotilla Sánchez, José Eduardo
Dept./Program:
Electrical Engineering
Year:
2012
Degree:
Ph. D.
Abstract:
Recent catastrophic events resulting in the disruption of large portions of electricity infrastructure around the globe suggest that today's power systems consistently function near high-risk operating points. The emergence of power-laws in the sizes of power grid disturbances constitutes further evidence for this criticality. With the rapid integration of synchronized phasor measurements and smart metering systems into electric power systems, researchers have a unique opportunity not only to obtain new insights into the criticality of electrical infrastructure but also to advance the state of the art in power engineering by building on the "big data revolution." The transition from a data-scarce to a data-rich electricity system mirrors transitions in other complex systems, such as airline networks and healthcare systems, and could substantially improve the reliability and capacity of power grids.
This dissertation illustrates three different approaches that leverage large data sets to develop computational methods that improve the tractability of dynamic analysis for very large power systems. The first one is the clustering of power networks into electrically cohesive zones as a method to facilitate zone-based planning analyses. The second approach uses a measure of critical slowing down for power systems to estimate proximity to unstable operating points. Lastly, the third component of this dissertation aims to identify the optimal combination of numerical algorithms and computer architectures to solve the hybrid discrete/continuous differential-algebraic equations that make up a cascading failure simulation applicable to large power systems.
This dissertation illustrates three different approaches that leverage large data sets to develop computational methods that improve the tractability of dynamic analysis for very large power systems. The first one is the clustering of power networks into electrically cohesive zones as a method to facilitate zone-based planning analyses. The second approach uses a measure of critical slowing down for power systems to estimate proximity to unstable operating points. Lastly, the third component of this dissertation aims to identify the optimal combination of numerical algorithms and computer architectures to solve the hybrid discrete/continuous differential-algebraic equations that make up a cascading failure simulation applicable to large power systems.