Utilização de índices de estabilidade transitória para a avaliação de segurança dinâmica

Abstract

Indices that report how much a contingency is stable or unstable in an electrical power system have been the object of several studies in the last decades. In some approaches, indices are obtained from time-domain simulation; others explore the calculation of the stability margin from the so-called direct methods, or even by neural networks.The goal is always to obtain a fast and reliable way of analysing large disturbance that might occur on the power systems. A fast classification in stable and unstable, as a function of transient stability is crucial for a dynamic security analysis. All good propositions as how to analyse contingencies must present some important features: classification of contingencies; precision and reliability; and efficiency computation. Indices obtained from time-domain simulations have been used to classify the contingencies as stable or unstable. These indices are based on the concepts of coherence, transient energy conversion between kinetic energy and potential energy, and three dot products of state variable. The classification of the contingencies using the indices individually is not reliable, since the performance of these indices varies with each simulated condition. However, collapsing these indices into a single one can improve the analysis significantly. In this paper, it is presented the results of an approach to filter the contingencies, by a simple classification of them into stable, unstable or marginal. This classification is performed from the composite indices obtained from step by step simulation with a time period of the clearing time plus 0.5 second. The contingencies originally classified as stable or unstable do not require this extra simulation. The methodology requires an initial effort to obtain the values of the intervals for classification, and the weights. This is performed once for each power system and can be used in different operating conditions and for different contingencies. No misplaced classification o- - ccurred in any of the tests, i.e., we detected no stable case classified as unstable or otherwise. The methodology is thus well fitted for it allows for a rapid conclusion about the stability of th system, for the majority of the contingencies (Stable or Unstable Cases). The tests, results and discussions are presented using two power systems: (1) the IEEE17 system, composed of 17 generators, 162 buses and 284 transmission lines; and (2) a South Brazilian system configuration, with 10 generators, 45 buses and 71 lines.