Application of dynamic fault tree in reliability assessment of reactor protection system

[Background] Reactor protection system (RPS) has complex structure, numerous equipment, and continuous failure behaviors. The traditional static fault tree cannot accurately simulate the sequential failure behaviors of the RPS. [Purpose] This study aims to obtain the failure probability of related system of RPS and the sensitivity of each component by simulating the sequential failure behaviors of RPS, and provide suggestions for the optimal design and optimization of RPS in the future. [Methods] First of all, the dynamic fault tree (DFT) modeling tools was employed to model the reliability of the RPS, a dynamic binary tree method was applied to quantitative analysis of the DFT model. Then, the Latin hypercube sampling method was adopted to quantitatively analyze the uncertainty of the RPS failure probability, and compared with the traditional static fault tree method under the same conditions. [Results] Simulation results show that the failure probability with 0.95 confidence interval of RPS is 0.015 306~0.015 721. The two most sensitive components are the undervoltage drive card and the undervoltage relay. [Conclusions] DFT can accurately simulate the reliability of RPS, and provide more accurate results, which is beneficial for optimization design in the future.【背景】反应堆保护系统（Reactor Protection System, RPS）结构复杂、设备数量繁多且存在持续失效行为，传统静态故障树无法精准模拟该系统的时序失效特性。 【目的】本研究通过模拟反应堆保护系统的时序失效行为，获取其关联子系统的失效概率与各组件的敏感性特征，为后续反应堆保护系统的优化设计与改进提供理论依据。 【方法】首先采用动态故障树（Dynamic Fault Tree, DFT）建模工具开展反应堆保护系统的可靠性建模，并运用动态二元树法对该动态故障树模型进行定量分析；随后引入拉丁超立方抽样法（Latin hypercube sampling）对反应堆保护系统失效概率开展不确定性定量分析，并与同等条件下的传统静态故障树方法进行对比分析。 【结果】仿真结果表明，反应堆保护系统失效概率的95%置信区间为0.015306~0.015721；敏感性排名前两位的组件为欠压驱动卡与欠压继电器。 【结论】动态故障树可精准模拟反应堆保护系统的可靠性特性，输出更为准确的分析结果，可为后续优化设计提供有效支撑。