Localization & Mitigation of Cascading Failures in Power Systems, Part I: Spectral Representation & Tree Partition

Cascading failures in power systems propagate non-locally, making the control of outages extremely difficult. In this work, we propose a new framework that offers strong analytical guarantees on both the localization and mitigation of cascading failures in power systems. The key component of this framework leverages the concept of tree partition, which characterizes regions of a power network inside which line failures are automatically localized. In Part I of this paper we establish a mathematical theory that underlies all the performance guarantees of tree partition, as well as its failure localization properties. This theory consists of a set of tools developed using the Laplacian matrix of the transmission network and reveals a novel perspective that precisely captures the Kirchhoff's Law in terms of topological structures. Our results show that the distribution of different families of subtrees of the transmission network plays a critical role on the patterns of power redistribution, and motivates tree partitioning of the network as a strategy to eliminate long-distance propagation of disturbances. These results are used in Parts II and III of this paper to design strategies to localize and mitigate line failures.