Correlation analysis for screening key parameters for passive system reliability analysis. (March 2015)

• Record Type: Journal Article
• Authors: Yu, Yu; Ma, Guohang; Hao, Zulong; Wang, Shengfei; Niu, Fenglei; Zio, Enrico
• Is Part Of: Annals of nuclear energy. Volume 77(2015:Mar.)
• Page Start: 23
• Page End: 29
• Publication Date: 2015-03
• Subjects: Passive system reliability analysis -- Correlation analysis -- Natural circulation -- Thermal–hydraulic model -- AP1000; Nuclear energy -- Periodicals; Nuclear engineering -- Periodicals; 621.4805
• DOI: 10.1016/j.anucene.2014.10.035
• Languages: English
• ISSNs: 0306-4549

Abstract

Highlights: A method for screening key parameters used in passive system reliability analysis. T–H model is a relationship between output and each input. Correlation coefficient between output and input is influenced by T–H performance. Correlation coefficient is also affected by uncertainty of input. Abstract: Passive systems are widely used in new generation nuclear power plants to enhance their safety. Reliability of passive system operating based on natural circulation must be assessed in terms of functional failure. The functional failure probability evaluation requires repeatedly running the thermal–hydraulic (T–H) code which simulates the system responses under different values of the input parameters. In practice, repeated running of the code is quite costly in terms of running time and artificial neural network (ANN) has been proposed to replace the T–H model. However, the number of input parameters can be too large to satisfy the requirement of the ANN. In this paper we illustrate a systematic methodology to screen the key parameters for passive system operation based on correlation analysis for reducing the number of inputs. Correlation analysis is a well-known statistical method to assess the relationships among parameters. In the case of interest for passive system reliability, we consider the T–H model as a relationship between model inputs and outputs, which can be used in correlation analysis. With this method, key parameters can be screened with limited numbers of samples. The passive containment cooling system in AP1000 is analyzed and 4 parameters are identified as important ones from 47 inputs.