Use of fault tree analysis for evaluation of system-reliability improvements in design phase

Abstract

Traditional failure mode and effects analysis is applied as a bottom-up analytical technique to identify component failure modes and their causes and effects on the system performance, estimate their likelihood, severity and criticality or priority for mitigation. Failure modes and their causes, other than those associated with hardware, primarily electronic, remained poorly addressed or not addressed at all. Likelihood of occurrence was determined on the basis of component failure rates or by applying engineering judgement in their estimation. Resultant prioritization is consequently difficult so that only the apparent safety-related or highly critical issues were addressed. When thoroughly done, traditional FMEA or FMECA were too involved to be used as a effective tool for reliability improvement of the product design. Fault tree analysis applied to the product as a top down in view of its functionality, failure definition, architecture and stress and operational profiles provides a methodical way of following products functional flow down to the low level assemblies, components, failure modes and respective causes and their combination. Flexibility of modeling of various functional conditions and interaction such as enabling events, events with specific priority of occurrence, etc., using FTA, provides for accurate representation of their functionality interdependence. In addition to being capable of accounting for mixed reliability attributes (failure rates mixed with failure probabilities), fault trees are easy to construct and change for quick tradeoffs as roll up of unreliability values is automatic for instant evaluation of the final quantitative reliability results. Failure mode analysis using fault tree technique that is described in this paper allows for real, in-depth engineering evaluation of each individual cause of a failure mode regarding software and hardware components, their functions, stresses, operability and interactions.