Probabilistic Risk/Reliability Analysis (PRA)Using a Simulation Approach

摘要

Probabilistic risk/reliability (PRA) analyses for engineered systems arernconventionally based on fault-tree methods. These methods are mature and efficient, and arernwell suited to systems consisting of interacting components with known, low probabilities ofrnfailure. Even complex systems, such as nuclear power plants or aircraft, are modeled by therncareful application of these approaches.rnHowever, for systems that may evolve in complex and nonlinear ways, and where thernperformance of components may be a sensitive function of the history of their workingrnenvironments, fault-tree methods can be very demanding. This paper proposes an alternativernmethod of evaluating such systems, based on probabilistic time-based simulation usingrnintelligent software objects to represent the components of such systems. Using a Monte Carlornapproach, simulation models can be constructed from relatively simple interacting objects thatrncapture the essential behavior of the components that they represent. Such models are capable ofrnreflecting the complex behaviors of the systems that they represent in a natural and realistic way.