Uncertainty study of LOCA accidents for developing FLEX strategy in nuclear power plants

Abstract

As an aftermath of the Fukushima nuclear accident and its implications for public acceptance of nuclear energy, the nuclear industry has begun to seriously rethink and improve the safety strategies of nuclear power plants. The beyond-design-basis external events, such as the earthquake and tsunami in Fukushima nuclear accident, may lead to an extreme accident which may lead to the failure of the traditional strategies. In the recent decade, the concept of FLEX strategies has been accepted in the nuclear engineering community and attracted a large number of researchers to investigate it for nuclear accident mitigation. The development of FLEX strategies is divided into two steps. One is an uncertainty study of system response, especially the safety injection system. The other is training and testing of machine learning (ML)-based FLEX strategies and to verify the effectiveness of ML algorithms in the development of FLEX strategies for nuclear power plants. This paper focuses on first step uncertainty study of the nuclear accidents by using statistical algorithm to generate a database of incidents and system responses. The conclusion of this paper is useful for analyzing the characteristic of typical nuclear accidents and benefits to the FLEX strategy development, which will be focused on in the near future.