Integration of functional reliability analysis and system hardware reliability through Monte Carlo simulation

• Published in: Annals of nuclear energy 2016-09, Vol.95, p.54-63
• Main Author: Ramakrishnan, M.
• Format: Article
• Language: English
• Subjects: Balanced failure biasing; Computer simulation; Decay heat; Failure; Failure probability; Functional reliability; Hardware; Monte Carlo simulation; Nuclear engineering; Nuclear reactor components; Passive system; Pools; Reliability analysis; Uncertainty
• ISSN: 0306-4549; 1873-2100
• DOI: 10.1016/j.anucene.2016.04.032

•Integration of functional reliability analysis and system hardware reliability through Monte Carlo simulation.•Failure probability estimation for crossing the various design safety limits on temperature.•Linear behaviour of failure probability with various standard deviation values of response probability density.•Results compare well with two other methods when process uncertainty contribution to total failure probability is dominant.•Significant difference in results when the process uncertainty contribution to total failure probability is not dominant. Passive safety systems are being increasingly used in innovative reactor designs due to its inherent safety characteristics. The passive safety system failure can occur due to process uncertainties and changes in system hardware state with time. The failure probability contribution due to process uncertainties of a passive decay heat removal system of a fast breeder reactor are quantified through functional reliability analysis. The probability density of the peak hot pool temperature obtained from functional reliability analysis is combined with system hardware reliability through Monte Carlo simulation. The probabilities of crossing the various design safety limits on hot pool temperature are evaluated. The failure probabilities are estimated for different standard deviation values of the peak hot pool temperature probability density. This leads to a linear relationship between failure probability and standard deviation having two distinct regions. The results from the present analysis are compared with the results from two other methods. The results from the three methods converge when the process uncertainty contribution to total failure probability is dominant. The results are significantly different from the three methods when the process uncertainty contribution to total failure probability is not dominant. The difference or the convergence of results by different methods is governed by the extent of process uncertainty contribution to the total failure probability. The proposed method reduces the conservatism as compared to the functional reliability analysis combined with fault tree approach.