Margin-Based Framework for Online Contingency Selection in Unbalanced Networks

Abstract

A framework for performing fast online contingency selection in unbalanced power systems is presented in this paper. Selection methods are typically based on the principle of identifying the effect of contingencies on multiple normalized performance indices and ranking them using the results. Presently used performance indices are highly nonlinear and they are known to mask the effects of single contingencies leading to misclassifications. In this paper, we propose two new methods, one relying on margin-based performance indices and another based on state sensitivity. New performance indices are proposed based on margins of 1) circuit loading, 2) bus voltages, and 3) reactive power. In addition, a state sensitivity method is proposed which estimates a system's post contingency operating state via a single iteration of the quadratized power flow model and provides estimates of post contingency line loading, bus voltages, and reactive power levels. Numerical experiments on a three phase version of the IEEE Reliability Test System show that the proposed performance indices yield more accurate results, at a computational cost comparable to a single power flow iteration. The state sensitivity method is more accurate in identifying critical contingencies but its computational cost is higher. The method has been also demonstrated in the larger PEGASE systems.