CONDITIONAL MODELING FOR VARIABLE SUCCESS CRITERIA

摘要

Fault tree modeling has traditionally been performed on a functional level -where the system success criteria are largely static; e.g. power distribution system/train is functionally modeled to provide power to all components associated with its system/train. However, under certain specific circumstances, conditions may exist where a function will be limited or expanded variably. The precise limitations differ but are generally associated with procedural restraints and equipment restrictions under explicitly defined conditions (specific initiators/accident sequences, timing, component failure combinations, condition specific design limitation, etc.). Existing methods to address variable success criteria include post-processing of quantification results or developing additional event tree logic. These methods can result in inaccurate component/basic event risk importance measures (importance values can be over/under estimated) or large and redundant fault tree top (accident sequence) logic. An alternate method for treatment of variable success criteria can be addressed by modeling system fault tree logic conditionally through the use of mutually exclusive logic models. The conditional modeling uses modified system fault trees that contain multiple functions in tandem with mutually exclusive logic acting to dynamically control the accident scenarios where each of the functions is applied. The resulting conditional modeling method produces accurate component/basic event risk importance values and can significantly simplify the amount of modeling associated with event tree top logic expansion, creation of postprocessing files and corresponding mutually exclusive logic updates. It should be noted this method was developed for use with CAFTA based fault tree models.