A smart component methodology for reliability analysis of dynamic systems

摘要

The new generation of nuclear power plants are being designed with passive safety systems and advanced digital Instrumentation & Control (I&C) systems for achieving required operational performance and improved reliability. Digital I&C systems often perform complex tasks while interacting with time dependent process dynamics. Similarly, in passive decay heat removal systems, the time dependence arising from the variation of heat load and interaction between redundant loops are important. Since static reliability methods cannot model these systems without significant approximations, dynamic reliability methods are preferred. Dynamic methods improve the correctness of modeling and shift the burden of proof for correctness from the modeling expert to the methodology. Though a number of dynamic reliability methods have been developed, they have not been adopted widely, like that of the traditional event tree/fault tree methods. A significant reason is due to the lack of an intuitive simulation framework to translate and represent the system structure and functional aspects by a faithful system reliability model. Therefore, we propose in this paper a methodology known as Smart Component Methodology (SCM) for reliability modeling of dynamic safety systems. The methodology is based on an intuitive object oriented framework for the representation of the components, system structure, behavior and reliability data of a safety system. A suitable Monte Carlo simulation algorithm is embedded into this framework to quantify the system reliability. The SCM architecture for dynamic system representation and the general Monte Carlo simulation algorithm used for driving the object oriented framework are presented. The method is applied to example I&C systems, passive heat removal systems involving human actions. The results are compared with traditional methods to validate the results and to demonstrate the improved accuracy and ease and generality of modeling. (C) 2019 Elsevier Ltd. All rights reserved.