Fragility analysis and probabilistic performance evaluation of nuclear containment structure subjected to internal pressure

Abstract Based on the detailed three-dimensional finite element model of the nuclear containment structure, this study presents reliability and probabilistic performance evaluation of the nuclear containment structure subjected to internal pressure. To realize the automatic running of nonlinear finite element analysis of nuclear containment structure, Python and Matlab scripts are developed. Confidence intervals of fragility parameters are estimated by the statistical inference method and bootstrap method. An analytical method for constructing the confidence interval of the fragility curve is proposed in this study, and confidence interval of the fragility curve predicted by the proposed method is compared with that of bootstrap method. Moreover, statistical characteristics of the cumulative failure probability of nuclear containment structure are estimated by bootstrap method and the proposed analytical method. Finally, probabilistic safety margin of the nuclear containment structure are evaluated by the median value method and confidence interval method. Results indicated that statistical uncertainty has almost no effect on the mean values of the fragility parameters. However, statistical uncertainty has some effects on the variability of the fragility parameter β S . In general, the influence of statistical uncertainty on fragility parameter β S is greater than that of fragility parameter P m . Confidence intervals of P m estimated by the statistical inference method and bootstrap method are almost the same, and statistical inference method overestimates the confidence interval of β S . The proposed method for constructing confidence interval and bootstrap method provide almost the same prediction for the confidence interval of the fragility curve. In general, statistical characteristics for the cumulative failure probability of the nuclear containment structure calculated by the bootstrap method and the proposed analytical method are almost the same. The difference between the safety margin calculated by the median value method and the safety margin with 95% confidence level calculated by confidence interval method is negligible.