【作者】 张敏；

【导师】 张志俭；

【作者基本信息】 哈尔滨工程大学 ， 核能科学与工程， 2019， 博士

【摘要】 风险监测是指在核电站寿期内,对核电机组不同配置下的风险进行评估的过程,其结果可应用于风险指引的核电站运行与维修管理,在保障核安全的前提下,平衡核电站安全性和经济性。Living PSA(Living Probabilistic Safety Analysis)方法是目前国际国内工程界广泛认可的、用以实现风险监测的建模分析方法,风险监测器(Risk Monitor,RM)则是在其方法基础上开发出的软件工具。现有的RM更多的用于离线场景,如核电站的维修计划优化以及系统改造后的初始风险评估,其对风险的评价仍然是离线的、静态的和初始的。得益于近年来核电站信息系统及数字化仪控的发展,可以较为方便地在核电站运行过程中连续获取设备状态信息(包括运行、备用、故障、维修等),为在线风险监测的实施提供了技术基础,即:基于核电站安全相关设备的连续状态监测与反馈,在其运行过程中,连续及时地评估机组风险,该风险是一组对应于系统真实状态的、随时间不断变化的瞬时值。风险监测器从离线应用到在线应用的发展,对Living PSA方法模型的准确性提出了更高的要求。事件树和故障树是PSA的两大基本要素,本课题首先对事件树更新和故障树更新需求进行了研究,确定了以故障树更新实现Living PSA更新的总体技术路线,并确定了Living PSA方法的技术框架与需要解决的关键技术问题。针对核电站可能发生的配置变化情况,对现有的Living PSA方法进行了如下几点改进型研究:1)对故障树中底事件概率模型改进为状态相关、时间相关的概率模型,使其能够考虑设备当前状态及运行历史对其任务可靠性与重要度的影响;并建立了底事件模型与设备状态的对应关系,利用该对应关系可实现底事件的自动更新;2)对工业界广泛使用的共因失效参数模型进行了研究,得出了共因组阶数降低时共因失效模型参数更新的理论计算公式,并结合工程实际,给出了共因失效模型更新流程与简化的参数更新公式;3)对冗余系统存在的顺序失效问题,引入“顺序条件失效”和“顺序依赖失效”的概念,分别采用“条件与”门和“优先与”门,构造了一种顺序故障树模型,给出了其定性分析布尔代数运算规则和最小割集定量化计算公式,并提出了基于穷举式建树的和基于计算机辅助自动建树的两种顺序故障树建模与更新方案;基于以上研究,提出了 Living PSA建模于更新软件的功能需求,并进行了软件设计,包括数据结构、工作流程以及人机界面设计,按照设计方案所开发的软件能够满足基于本课题建议方法的Living PSA建模与更新,并经过了第三方测试验证与业内专家的评审。最后,选择某核电厂设备冷却水系统为对象,并预设一组运行配置变化情况,给出了在线风险监测的Living PSA建模、更新与分析的综合案例。研究表明,与采用现行的LivingPSA建模与更新方法相比,本论文所提出的方法具有以下特点:1)能够指示系统风险的连续变化,以及对应时刻的风险重要贡献项,从而更有针对性地进行运行与维修管理。2)在底事件概率模型中考虑设备运行历史的影响,补偿了现行方法的不保守性;3)对共因失效更新的考虑,相较于当前工程方案更准确;4)对存在顺序失效特性的系统的失效路径描述更加准确,消除了现有模型中部分不必要的保守性,并能够发现现阶段故障树方法未发现的系统失效路径。更多还原

【Abstract】 Risk monitoring is the process of evaluating the risk of a Nuclear Power Plant(NPP)unit in different configurations during its lifetime.The results can be applied to risk-informed safety management in NPPs,to balance the safety and economy under the precondition of ensuring nuclear safety.The Living Probabilistic Safety Analysis(Living PSA)method is an approved and widely used modeling and analyzing method to implement risk monitoring in industry,and many software tools(named as Risk Monitor,RM)have been developed based on Living PSA.Existing RMs are mainly used for off-line applications,such as of maintenance plan optimization and risk assessment after system modification.The assessment is offline,static,and initial(usually not conservative).As the technique development of information system and digital instrumentation&control system in NPPs in recent years,it is implementable to continuously collect equipment status information(operation,standby,failed,in maintenance,etc.)during the NPPs operation.It provides the foundation of online risk monitoring,which is to evaluate the risk of an NPP unit continuously and timely at any time point during plant operating according to the current plant configurations(component status in operation systems and safety-related systems).Instantaneous risk that continuously change with time and the real configurations of units can be evaluated by online risk monitoring.The development of risk monitors from off-line application to online application has led to higher requirments on the accuracy of Living PSA.Event Tree(ET)and Fault Tree(FT)are the two basic elements of PSA.Therefore,the update strategies of event tree and fault tree are studied separately first.And it is found that Event Tree updating can be implemented through Fault Tree updating.Then,research on Living PSA is focused on Fault Tree modeling and updating.And also the technical framework and key issues to be studied for Living PSA methodology are determined.According to the possible changes of Nuclear Power Plants during operation,the following improvements have been made to the existing Living PSA method:1)Time-dependent and status-related Basic Event(BE)probability models are proposed to involve possible factors which may affect the reliability and importance of a component in consideration,including the current status,demand status and operating history.Then,the correspondence relationship among the BE models and equipment states is established,by which the BE models can be updated automatically.2)Research is carried out on widely used Common Cause Failure(CCF)parameter models to find out how to implement the parameter updating when CCF group order is reduced.The theoretical parameter updating models are obtained,and then the update process and simplified parameter update formulas are suggested according to actual engineering processes.3)To evalute the sequential failure issues in redundant system,the concepts of "Sequence-Conditional Failure" and "Sequence-Dependent Failure" are introduced,and"Conditional-AND" gate and "Priority-AND" gate are constructed respectively,forming the sequential fault tree(SFT)model.The Boolean logic laws for qualitative assessment are extended for SFT and also the Minimum Cut Sets(MCSs)quantification formulas.Two SFT modeling and updating schemes are proposed:One is to enumerate all possible situations in a FT model,and the other is based on computer-aided automatic FT construction.Based on the above research,the functional requirements of Living PSA software for modeling and updating are established,and the software design is carried out,including data structure,workflow and human-machine interface design.By the software developed according to the design,modeling and updating can be emplemented based on the proposed Living PSA methodology.The softwared has been verified by independent third-party and industry experts.Finally,a case study is carried out to demonstrate the comprehensive Living PSA modeling and updating process on NPP Component Cooling Water System,with a set of configuration changes presetted.Research tells that,compared to the existing Living PSA methodology,the proposed method has several advantages as follows:1)System risk and corresponding important risk contributors obtained from the proposed Living PSA are related to not only system configuration but also system operating time.Thus,the risk imformed operation and maintenance management can be carried out more respectively.2)The over optimistic in existing Living PSA is compensated by considering the influences of equipment operation history in BE probability models;3)The suggested methodology for Common Cause Failure Update is more accurate,compared to currently engineering adopted measure in commercial RM tools;and4)The failure paths of sequential failure system obtained from the proposed Living PSA model are more accurate,and also the quantification,eliminating unnecessary conservation in existing Living PSA.And there is evidence that the proposed Living PSA method may find new failure paths which are missed by the existing method.更多还原