Date of Award

4-24-2019

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Systems Engineering

First Advisor

Kamran Iqbal

Abstract

Intentional controlled islanding (ICI), also referred as controlled system separation or system splitting, has been proposed as an effective corrective control action to help detect, prevent and mitigate the risk and spread of the cascading failures. A practical controlled islanding separation scheme needs to address two critical problems: where to island i.e., the splitting points to form sustainable islands and when to island i.e., Islanding time. This dissertation examines the causes and mechanism of power system cascading outages and develops new approaches to help detect, prevent and mitigate the risk and spread of these cascading failures. Four effective solutions: weighted complete graph-based detection of coherent groups of generators scheme, modularity clustering-based detection of coherent groups of generators scheme, a two-step multi-layer spectral clustering separation scheme, and a unified controlled separation scheme i.e., where and when to island for ICI, have been proposed. In this dissertation, a weighted complete graph-based detection of coherent groups of generators scheme i.e., reveals the relationship between the synchronization coefficient and the integrity of the connection grid lines, which is appropriate for real-time analysis, using the synchronization coefficients between generators is proposed. Further, a modularity clustering-based detection of coherent groups of generators scheme is also proposed. This method identifies coherent groups of generators based on the correlation coefficient (CC) between rotor angle/speed oscillations of generators and uses real-time power integrity indices, which discloses the overall system separation status. Further, in this dissertation, to address the question “where to island”, a real-time ICI algorithm based on multi-layer graphs, subspace analysis, and constrained spectral clustering is proposed. The proposed scheme progresses by seeking realistic islanding cutsets, which would cause the minimum power flow disruption, while maintaining generator coherencies, for any given island. Additional, to address the question “when to island”, a unified predictor scheme to predict the time of islanding by evaluating generator's synchronism in different groups using power splitting indices scheme predictor is proposed. Comprehensive simulation studies have been implemented using the IEEE 39 Bus and IEEE 118-bus systems and promising results show the ability of the proposed solutions to help detect, prevent and mitigate cascading outages.

Download

Share

COinS