Research on Time-Dependent Failure Modeling Method of Integrating Discrete Dynamic Event Tree With Fault Tree

Classical PRA methods such as Fault Tree Analysis (FTA) and Event Tree Analysis (ETA) are characterized as static methods due to predetermined event sequences and success criteria of frontline systems. Unlike classical PRA, Dynamic PRA (DPRA) couples the stochastic random process such as random failure of system with deterministic analysis（by simulation）to determine the risk level of complex systems. It considers the safety significance of the timing and order of events on accident progression and consequences. However, it is time-consuming to establish a complicated system simulation model with all safety systems, equipment in an accident. Meanwhile, thousands of accident scenarios are generated due to various state transitions and parameter uncertainty. In fact, most of accident scenarios in nuclear power plants have been modeled for risk analysis using classical PRA models (such as ET). Some of the components to be analyzed could be modeled by a ETs and FTs, instead of establishing a computationally expensive simulation model. However, it is still challenging to combine the classical PRA method with the dynamic PRA method. For dynamic PRA, timing of events is explicitly modeled while the classical PRA model defines Boolean logic variables (such as True, False, Normal). Therefore, the "integration" method requires that these models must be able to handle Boolean logic variables and the timing of specific events. So in this paper, an integration model is illustrated about how to couple the time-dependent simulation model with the probabilistic analysis. Then possible dependencies and configuration consistency issue accounted for Discrete Event Tree (DET) branch probabilities are discussed. After that, a detailed method of integration FT into DET is introduced which emphasizes method of computing the conditional branch probability with FTs online, as well as developing the DET model in case of temporal relations of failure. Finally, a simple case of Low Pressure Injection System in Large Break Loss-of-Coolant Accident (LBLOCA) scenario is provided as a demonstration.