Evaluation of hurricane impact on composite power system reliability considering common-cause failures

Abstract

Extreme adverse weather such as a hurricane can have a significant impact on composite power system reliability. Since hurricanes can cause the simultaneous failures of multiple system components, common-cause failures (CCF) should be investigated in the reliability evaluation of composite power systems when the effects of hurricanes are considered. A few techniques have been proposed to evaluate the effects of CCF, but they are not suitable for composite power systems. This paper proposes a method based on Bayesian networks (BN) to solve this problem. Basically, the proposed method uses the noisy OR-gate model to reduce the dimensional dilemma of the conditional probability method. This model also considers the independent failures of transmission lines and generating units during hurricanes. The functionality of a BN is determined by its configuration and the conditional probability distributions (CPD) associated with the nodes. In this paper, the CPD of BN is obtained by using random sampling. Since hurricanes usually last for only a limited period of time, a pseudo-repetitive temporal model is used to calculate the time-specific system reliability indices. The proposed method is applied to the modified IEEE Reliability Test System (RTS). The implementation demonstrates that the proposed method is effective and flexible in its applications.