This paper develops a best estimate plus uncertainty (BEPU) framework for research reactor transient safety analysis. The BEPU framework is developed based on the system level reactor safety analysis code RELAP5-3D and the data analysis platform RAVEN developed by Idaho National Laboratory. Within the framework, a sensitivity analysis procedure is first conducted to identify the contributions and ranks of individual input parameters to the user-defined figures-of-merit (FOMs) associated with specific transient phenomena. An uncertainty analysis procedure is then performed to quantify the uncertainties of the FOMs resulting from the uncertainties of the input parameters. Many useful outcomes can be realized through the BEPU analysis. Specifically, the sensitivity information obtained from the sensitivity analysis will provide insights about the influence of each different input parameter on FOMs. The uncertainty information obtained from the uncertainty analysis will imply the range of response deviations caused by the propagation of errors existing in various input components.

As a case study for research reactors, the developed BEPU framework was employed to perform design-basis accident (DBA) analysis for one conceptual research reactor design proposed at the National Institute of Standards and Technology (NIST). Two hypothetical DBA scenarios, namely the reactivity insertion accident (RIA) and the loss of flow accident (LOFA), were modeled and analyzed through the BEPU framework. To demonstrate the value of the BEPU framework, the BEPU analysis results were compared to that obtained from the conventional transient safety analysis procedure, which was conducted by using commonly used transient safety analysis codes including RELAP5-3D and PARET. The comparison shows that the BEPU analysis is capable of providing additional sensitivity and uncertainty information that help confirm safety margins of the NIST conceptual research reactor during both RIA and LOFA situations, which justifies the advantages and benefits of the BEPU safety analysis framework developed in this work.

本文为研究堆瞬态安全性分析开发了最佳估计加不确定度（BEPU）框架。BEPU框架是基于系统级反应堆安全性分析代码RELAP5-3D和爱达荷州国家实验室开发的数据分析平台RAVEN开发的。在框架内，首先进行灵敏度分析程序，以识别各个输入参数对与特定瞬态现象相关的用户定义的品质因数（FOM）的贡献和等级。然后执行不确定性分析程序以量化由输入参数的不确定性引起的FOM的不确定性。通过BEPU分析可以实现许多有用的结果。特别，从灵敏度分析中获得的灵敏度信息将提供有关每个不同输入参数对FOM的影响的见解。从不确定性分析中获得的不确定性信息将暗示由各种输入组件中存在的误差的传播引起的响应偏差的范围。

作为研究堆的案例研究，已开发的BEPU框架用于对美国国家标准技术研究院（NIST）提出的一种概念研究堆设计进行设计基准事故（DBA）分析。通过BEPU框架对两种假设的DBA场景进行了建模和分析，分别是反应性插入事故（RIA）和流失事故（LOFA）。为了证明BEPU框架的价值，将BEPU分析结果与通过使用包括RELAP5-3D和PARET在内的常用瞬态安全分析代码进行的常规瞬态安全分析程序获得的结果进行了比较。