核电厂厂用水系统概率安全分析方法研究

目的 厂用水系统作为核电厂的最终热阱对核电厂的运行及事故缓解具有重要意义。 方法 采用故障树方法对正常运行工况与停堆工况在寒冷与非寒冷天气条件下的四种情况进行模化，并分析了厂用水系统作为事故缓解手段的可靠性；蒙特卡罗方法被用于分析该系统作为始发事件贡献因素的可运行性；同时，由敏感性分析得到了设备失效率、定期维修周期与年平均不可用时间的关系并就此给出了相关优化建议。 结果 定量化结果给出厂用水系统对于不同任务时间的不可靠度，其中重点关注的正常运行工况在非寒冷天气下的不可靠度为1.47E-01（任务时间1年），事故缓解过程的不可靠度为4.10E-04（任务时间72 h）；停堆工况非寒冷天气下的不可靠度为6.45E-03（任务时间96 h）。丧失厂用水作为始发事件的发生频率约2.0E-03/年，系统的年平均不可用时间约3.4 h。 结论 结果表明:海水预处理系统作为厂用水冷却塔的补水水源，对厂用水系统不可靠度贡献较大，并且该系统中贡献最为显著的主给水管道上电动隔离阀的运行失效，因没有冗余设计而成为单点故障。此外，相比于降低设备的失效率，缩短定期维修周期对于降低年平均不可用时间更具有效性和可操作性。

Study on Probabilistic Safety Analysis for Service Water System of Nuclear Power Plant

Introduction The service water system as the ultimate heat sink is of great significance for the operation and accident mitigation of nuclear power plants.    Method   In this paper, the fault tree was built to model the full power and shutdown modes of the investigated plant under cold and non-cold weather conditions respectively, and then to assess the reliability of service water system as an accident mitigation measure. The Monte Carlo method was used to analyse the operability of the system as an initiating event. Relationships between equipment failure rate, maintenance cycle and the annual average unavailability were also attained via sensitivity analysis.    Result   Quantitative results show the failure rates of service water system for different mission time, particularly under the conditions of non-cold weather at full power the failure rates are 1.47E-01 per year and 4.10E-04 per 72 hours (mission time in accident mitigation process), and at shutdown mode the corresponding failure rate is 6.45E-03 per 96 hours. The frequency of the loss of service water is about 2.0E-03 per year and the annual average unavailability time of the system is about 3.4 hours.    Conclusion   The results suggest that seawater pretreatment system has more significant influence on the unreliability of service water system, of which the operation failure of electric isolation valve at the main pipeline is the dominant contributor due to having no redundant design. In addition, shortening regular maintenance cycle will be a more effective and operable way to reduce the annual average unavailability time compared with reducing the failure rates of components.