Network Reliability with Degenerate Gaussian Copula Description of Common-cause Failures
We consider the application of the degenerate multivariate normal distribution — known as the Gaussian copula — to the description of two-terminal reliability for an arbitrarily large, meshed network undergoing common-cause failures. When all the systems components are identical, with the same availability p and the correlation term p, we can express the exact result using a simple integral. The first correction to the independent-failures case is proportional to the second derivative of the reliability polynomial; corrections of higher order are easily obtained, too. In the case of correlated subgroups of components, we have developed a numerical expression allowing the quick calculation of the correlated probabilities through Pade approximants. A case study illustrates our results.
network reliability common-cause failures multivariate normal distribution
CHRISTIAN TANGUY
Orange Labs, Networks and Carriers, CORE/TPN/TRM, 3&-40 rue du general Leclerc, Issy-les-Moulineaux, 92794,Prance
国际会议
北京
英文
571-577
2011-06-20（万方平台首次上网日期，不代表论文的发表时间）