Vulnerability analysis of electric power delivery networks

University dissertation from Stockholm : Mark och vatten

Abstract: Disturbances in the services provided by the infrastructuresystems—e.g. electric power supplies and communications—can have serious implications for everyday life,economic prosperity and national security. The disturbances canbe initiated by natural disasters, adverse weather, technicalfailures, human errors, sabotage, terrorism or acts of war. Theaim of this thesis is to study methods for proactivevulnerability analysis of electric power delivery networks(i.e. to analyze their sensitivity to threats and hazards), andto formalize vulnerability as a theoretical concept.The thesis consists of three papers. In the first paper, wediscuss concepts and perspectives for developing a methodologyfor vulnerability studies with the help of the followingthemes: The properties of the infrastructure systems, threatsand hazards, vulnerability and consequence analysis, andmeasures for creating robust and resilient systems.In the second paper we discuss how to assess vulnerabilityof power delivery systems with the help of standard powersystem performance indices. In two case studies, Swedish powerdelivery disturbance data is analyzed with statistical methods.We demonstrate that the disturbance size of large disturbancesfollows a power law distribution, and that the time betweendisturbances is exponentially distributed.In third paper, we model electrical power networks asgraphs, and conduct empirical studies of two power transmissiongrids. We calculate values of topological characteristics ofthe networks and compare their error and attack tolerance, i.e.their performance when vertices are disabled, with twofrequently used model networks. Further, we perform a graphinfluenced vulnerability analysis of a fictitious powernetwork, and evaluate di.erent strategies to decrease thevulnerability of the system.

  This dissertation MIGHT be available in PDF-format. Check this page to see if it is available for download.