Synthetic tree model: a formal methodology for fault tree construction

Abstract

Fault tree analysis is a recently developed method of reliability analysis and is generally applicable to complex, dynamic systems which include nuclear reactor systems.Fault tree analysis offers a tool by which systems may be optimized in design to achieve, within the limits of engineering capabilities, the dual requirements for maximum safety and plant availability and minimum cost and complexity.The influence, of the application of formal reliability analysis to all systems will result in higher probability of the systems functioning properly when they are called upon t o operate.. This report provides a formal methodology,.Synthetic Tree Model, for constructing fault trees for electrical systems to the point where identifiable primary component failures will directly produce the required fault events.While the model is developed herein only for electrical systems, its implications extend to all fault tree constructions.Existing fault tree terminology'is used in Synthetic Tree Model.The resultant fault trees are in a conventional format and are, consequently, immediately compatible with presently used fault tree solution techniques.Actually, they differ from a conventionally constructed fault tree in few ways.A difference is that, should any number of analysts construct fault trees independently for a given system and main failure event, using Synthetic Tree Model, they will all obtain identical fault trees.This is not a characteristic of conventional fault tree construction.This report puts forth a rr~odel that affords the opportunity to reduce the cost of and time required for a fault tree analysis as well as provides potential for a standard by which fault trees can be constructed or checked.Synthetic Tree Model is a synthesis technique for piecing together, with proper editing, a fault tree from small segments called component failure transfer functions.The component failure transfer functions are obtained from a system-independent failure mode analysis of individual components.This piecing together is an uncomplicated process but does involve "bookkeeping" such that the appropriate editing of the component failure transfer functions can be carried out.The component failure transfer functions are a limiting factor on the resolution of the fault trees resulting from Synthetic Tree Model.While automation of fault tree construction is possible in the framework of Synthetic Tree Model, an experimental computer program, DRAFT, has been written t o accomplish this for certain 'lectrical systems; this automation formulates yet ailother clislinct type of analysis.The automated corislruction has potential as an overall, summary-type analysis that can be routinely done in 3 relatively small amount of time.AutolllaLivrl vl Synthetic Tree Model provides the fault tree analyst a valuable tool to complement his present skills while Synthetic Tree Model itself is immediately applicable to manual fault tree construction with the advantages of this manual analysis.The model is purposely left "open ended" to allow for its extension.Synthetic Tree Modcl shows potential for becorr~irlg a standard for fault tree construction as it is a formal approach t o fault tree construction.The technique is of a general enough nature to allow fault tree construction for systems both in the nuclezir industry and elsewhere.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 32 citations exist upstream.