Analytical Solutions of Large Fault Tree Models using BDD: New Techniques and Applications

Most tools available for quantifying large linked Fault Tree models as used in Probabilistic Safety Assessment (PSA) are unable to produce analytically exact results. The algorithms of such quantifiers are designed to neglect sequences when their likelihood decreases below a predefined truncation limit. In addition, the rare event approximation is typically implemented to the first order, ignoring success paths. In the last decade, new quantification algorithms using the mathematical concept of Binary Decision Diagram (BDD) have been proposed to overcome these deficiencies. Since a BDD analytically encodes Boolean expressions, exact failure probabilities can be deduced without approximation or truncation. However, extended effort is required when converting a given Fault Tree to its BDD form; this turns out to be an optimization problem of NP-complete complexity. Several innovative optimization techniques are developed and investigated as a case study on the fullscope PSA model of the Leibstadt Nuclear Power Plant. We succeeded in converting the Leibstadt PSA model into a BDD with more than 1'500'000 nodes, for a total of 3650 basic events. The BDD covers a complete Event Tree sequence that includes reactor shutdown and cooling with all Emergency Core Cooling Systems and support systems, enabling objective comparisons between quantification tools.