Identification of chains of events leading to catastrophic failures of power systems

Catastrophic failures have been traditionally treated as rare events. However, in view of the manner in which power systems have been evolving over the past decade, and the frequency with which catastrophic failures have occurred in this period, it is imperative that systematic approaches be devised to identify the likely chains of events that lead to these failures, so that appropriate countermeasures - strategic and tactical - may be designed and implemented. The paper discusses the recent body of work on structural issues, such as hidden failures and failure sequences, and identifies the need to explore further how chains conspire to form, and how the transition rates of the dependent events might change as the chain evolves. The paper then proceeds to develop a genetic algorithm based method that searches through the state space to identify likely event trees that connect healthy system states to catastrophic failure states. The method is illustrated via a small test system. The simplicity of the system enables an intuitive understanding of the mechanics of the method.