ABSTRACT
In nuclear PSA complexity of event tree sequences requires application of cut-off techniques to reduce the computational complexity. However this implies the estimation of the Truncation Error (TE) of the sequence unavailability. The lack of suitable methods for the estimation of TE requires the analyst to re-analyse the fault tree two or more times using lower truncation threshold values until the top-event probability shows a constant behaviour. However, if this condition cannot be reached due to insufficient working memory there is no way to correctly perform the analysis: the top-event probability is underestimated. Hence new advanced analysis methods are needed. A new method based on functional decomposition and truncation has been recently developed by the authors to analyse complex coherent fault trees. The fault tree is decomposed into a set of mutually exclusive simpler fault trees up to their dimension is compatible with the available working memory size. The results from the independent analysis of all generated simpler trees can easily be re-combined to obtain the results for the original un-decomposed complex fault tree. This paper extends the application of functional decomposition to the analysis of event tree sequences containing fault trees in normal form and fault trees in negated form. The decomposition is applied to the conjunction of complex coherent trees in normal form; then, each of the generated simpler fault trees are combined with the trees in negated form. The analysis of some complex event tree sequences of a real PSA show the pros and cons of the proposed decomposition approach.
