A robust data driven approach to quantifying common-cause failure in power networks.

Abstract

The standard alpha-factor model for common cause failure assumes symmetry, in that all components must have identical failure rates. In this paper, we generalise the alpha-factor model to deal with asymmetry, in order to apply the model to power networks, which are typically asymmetric. For parameter estimation, we propose a set of conjugate Dirichlet-Gamma priors, and we discuss how posterior bounds can be obtained. Finally, we demonstrate our methodology on a simple yet realistic example.