Analysis of fault trees with secondary failures

The Fault Tree methodology is appropriate when the component level failures (basic events) occur independently. One situation where the conditions of independence are not met occurs when secondary failure events appear in the fault tree structure. Guidelines for fault tree construction, which have been utilised for many years, encourage the inclusion of secondary failures along with primary failures and command faults in the representation of the failure logic. The resulting fault tree is an accurate representation of the logic but may produce inaccurate quantitative results for the probability and frequency of system failure if methodologies are used which reply on independence. This paper illustrates how inaccurate these quantitative results can be. Alternative approaches are developed by which fault trees of this type of structure can be analysed.