主催: The Japan Society of Mechanical Engineers
会議名: 第30回 原子力工学国際会議(ICONE30)
開催日: 2023/05/21 - 2023/05/26
Passive safety systems, which are independent of any external input or energy to operate, are considered the most significant candidate to increase the inherent safety of nuclear power plants. Due to the lower driving forces, passive safety systems are more susceptible to interference from external conditions compared to active safety systems, which can affect system performance. Functional failure is an important factor contributing to the operational failure of passive safety systems and needs to be considered in their reliability analysis. However, the assessment of functional failure depends on numerous T-H simulations, which in reality takes a lot of time. Two approaches have been proposed to solve this problem: one is a low-fidelity model that ignores the details of the original model and retains only the main framework of the original model; the other is to use a data-driven model to approximate the model response, which can also be called a metamodel. However, its effectiveness is severely constrained by the quantities of input parameters. The increase in the number of parameters generates a geometric increase in uncertainty. And in order to avoid "dimensional disasters", it is crucial to identify sensitive parameters, which helps to reduce the model complexity and improve the analysis efficiency. In this paper, the T-H model of the Integral-type Pressurized Water Reactor was developed, and based on this model, the elementary effects method and the variance-based method were used to analyze the parameter sensitivity. An efficient and accurate sensitivity analysis framework was proposed in which the number of factors was reduced using the elementary effects method, and then the ANOVA was executed on initially screened parameters to identify key parameters of the PRHRS in IPWR200.