Abstract
A large body of reliability engineering research assumes that components fail in a statistically independent manner: This axiom renders the mathematics tractable, but rarely holds in practice. Several studies address this shortcoming with models that possess additional parameters to describe the correlated failure of components. However, these latter techniques restrict correlation parameters to a subinterval of their valid range. This paper presents a reliability analysis technique that encompasses systems with both negative and positive component correlations. Unlike previous research, the proposed approach places no arbitrary restrictions on a system's correlation parameters. The approach simulates correlated binary variates with the aid of a multi-variate normal covariate representation. A series of examples illustrates the flexibility of the approach. The results quantitatively confirm the intuition that negative component correlation augments the reliability of fault-tolerant systems. This discovery indicates that the proposed approach offers an assessment methodology to measure the utility of negative component correlations.