Modeling common cause failures in diverse components with fault tree applications

Abstract

A common cause failure (CCF) is a single failure event that affects multiple components or functions of a system. Common cause failures are an important part of any reliability or hazard model, since they work to negate the improvements offered by redundancy. They are often the biggest contributors to risk levels, and should thus always be carefully considered. There are many system analysis methods that offer ways of taking common cause failures into account. However, these methods tend to be simple, such as taking a percentage of component failures, and attributing them to common causes. Certain assumptions are often made, such as that components will all share the same common causes, or have the same failure rate and distribution. For instance, the beta-factor model is a very commonly-used method, found in standards such as IEC 61508. To calculate the failure rate due to common causes, the beta factor is simply multiplied by the component failure rate. In essence, the beta factor simply represents the percentage of component failures that are due to common causes. Other papers have been written to explore general analysis methods for common causes, where the failure probability of the common cause is known [1]. In this paper, we shall focus more specifically at methods of expanding upon the beta-factor model so that it can be accurately used to calculate the probability of the CCF event, even where component rates, distributions, and common cause group probabilities differ. In addition to discussing the mathematical models used in these scenarios, we will also consider Fault Tree modeling solutions.