Risk Informed Safety Margin Characterization for Nuclear Reactors Using Dynamic Risk Assesment Methods

本提案考虑与CANDU业主集团合作开发和应用先进的概率风险评估（PRA）方法。虽然该应用针对的是核电厂，但这些方法是通用的，可能有益于进行风险分析的其他经济部门（例如，石油化工和航空航天）。PRA量化了与各种复杂工程系统运行相关的风险，并构成了核电厂监管许可的重要组成部分。风险的准确量化可以对工厂设计的安全性产生重要的见解，并可以识别工厂设计以及操作和应急响应程序中的漏洞。近年来，PRA的传统方法已被扩展到包括更现实的预测系统响应的风险估计。这种改进的分析形式-称为动态PRA（D-PRA）-包括事故情景进展期间物理系统行为的时间依赖性的显式建模，以及将人类交互纳入模型。这项工作旨在解决目前在D-PRA框架内存在的以下方法缺陷：a）缺乏对动态PRA方法和传统PRA方法的结果进行比较的综合研究，B）在将部分故障和延迟设备和/或操作员响应纳入时变模型方面几乎没有发展，c）需要进一步开发更准确的人为因素模型，特别是在时间敏感的事故场景的背景下，例如福岛型电站停电（SBO），d）希望量化工厂安全裕度和风险估计之间的关系。此外，在学术文献中没有将此类方法应用于加拿大核电厂的应用程序，因此McMaster与CANDU业主集团合作，将使用这些方法进行首次分析。

This proposal considers the development and application of advanced probabilistic risk assessment (PRA) methodologies in collaboration with CANDU Owners Group. While the application is aimed at nuclear power plants, the methods are generic and may have benefit for other economic sectors where risk analyses are performed (e.g., petrochemical and aerospace). PRA quantifies the risk associated with the operation of a broad range of complex engineering systems and forms an important part of the regulatory licensing of nuclear power plants. The accurate quantification of risk can yield important insights into the safety of a plant design and can identify vulnerabilities in plant design and operational and emergency response procedures. In recent years, the traditional methods for PRA have been extended to include more realistic predictions of system response in the risk estimates. This improved form of analysis - called dynamic PRA (D-PRA) - includes explicit modeling of the time-dependence of the physical system behaviour during accident scenario progression, and the inclusion of human interactions into the model. This work has been designed to address the following methodological shortcomings currently present within the D-PRA framework: a) there is a lack of comprehensive study comparing the results of the dynamic and traditional PRA methods, b) there has been little development on incorporating partial failures and delayed equipment and/or operator response into the time-dependent models, c) there is a need for further development of more accurate human factors models, especially in the context of time-sensitive accident scenarios such as a Fukushima-type station blackout (SBO), d) there is a desire to quantify the relationship between plant safety margins and risk estimates. In addition there are no applications in the scholarly literature of such methods being applied to Canadian nuclear power plants so McMaster in collaboration with CANDU Owners Group will perform the first-of-a-kind analysis using these methods.