Pedestrian: Probabilistic fracture mechanics analysis code based on direct sampling with replacement. (November 2018)

• Record Type: Journal Article
• Authors: Nagai, Masaki; Miura, Naoki; Yamamoto, Masato
• Is Part Of: International journal of pressure vessels and piping. Volume 167(2018)
• Page Start: 52
• Page End: 58
• Publication Date: 2018-11
• Subjects: Probabilistic fracture mechanics -- Structural integrity evaluation -- Monte Carlo method -- Sampling with replacement; Pressure vessels -- Periodicals; Pipe -- Periodicals; Récipients sous pression -- Périodiques; Tuyaux -- Périodiques; Pipe; Pressure vessels; Periodicals; 681.76041
• DOI: 10.1016/j.ijpvp.2018.10.008
• Languages: English
• ISSNs: 0308-0161

Abstract

Abstract: Probabilistic fracture mechanics (PFM) is a rational methodology in structural integrity evaluation for nuclear power plants, and several PFM analysis codes have been developed. In general, the statistical distributions of random variables such as initial crack sizes and material properties are expressed as probability density functions, and a set of random variables are sampled from those functions for each repeated failure analysis to obtain the cumulative failure probability. Since most random variables are considered to have physical upper and lower limits in their distribution, it is essential to set reasonable cut-offs in the functions to avoid improper samplings. If a sufficient quantity of measured data is available, their population is supposed to intrinsically include the information from such limits. In this study, we developed a PFM analysis code, that can sample the variables based on the direct sampling of measured data by sampling with replacement in order to rationally evaluate the failure probability. The difference in failure probability was discussed between the general sampling procedure, which uses probability density functions, and the direct sampling procedure. Consequently, the failure probability obtained by the direct sampling procedure was approximately an order of magnitude lower than that obtained by the general sampling procedure.