Improving Power Network Resilience Against Threats

The work reported in this dissertation aims at developing a practical tool for utility transmission planners to improve the power network resilience against high-impact, low-frequency events (HILF). Such events rarely happen, but when occurred are catastrophic. In this dissertation, we studied the impact of HILF from the static and dynamic point of view and proposed the planning strategy to lower the devastating effect of the HILF. In the static analysis, we consider the steady-state situation of the power grids and plan to protect the grid’s components, i.e., generators, buses, and transmission lines, to maximize preserving demand in a disastrous situation. To this end, the mathematical programming and optimization approaches are utilized to improve the power network resilience against threats, namely deliberate attack, and natural hazards. At the first step, the vulnerability of the power system’s components against deliberate attacks is evaluated through a tri-level Mixed-Integer Linear Programming (MILP) optimization problem, which is formulated based on the leader-follower game-theoretic approach. In this study, the system planner can observe how vulnerable are the component should an attack succeed. In addition to the deliberate attack that is an extreme view of the natural hazards, we proposed a game theoretic approach for linearization of the budget allocation problem in power networks reliability improvement. While the static analysis gives an overall insight about the network situation, the real-time behavior of the power networks in response to the HILF need to be pursued through the dynamic analysis. The main players in the dynamic fluctuations of the power network parameters following an event are the synchronous generators that form the coherent groups of generators. Accordingly, we proposed two methods for detecting the coherency among generators, namely algorithm-based and Modularity clustering-based approaches. The trace of the generator coherency paws the ways for forming the stable islands to mitigate the catastrophic blackouts. To this end, we proposed the multi-layer spectral clustering approach to detect the islands’ boundaries in a real-time fashion to avoid the large-scale outage when the system experience failure of its critical components. The efficiency of our proposed approaches is tested by applying our methodology on the standard IEEE test-bed systems