Methods of reliability model updating through additional events

Abstract Reliability theory is now applied in various areas. A first-order or second-order reliability method gives good approximations to both component and system reliability computations. With these available techniques, reliability models conditioned on additional events can be updated. Existing approaches to this important aspect of the reliability analysis are found numerically involved. New analytical methods, which are both theoretically reasonable and numerically efficient, are thus developed for computing the updated reliability as well as the updated distributions of the basic variables affecting the reliability. Especially, it is found that the distribution of an independent normal or lognormal variable, or the joint distribution of a group of normal or lognormal variables, is invariant through updating. Only its characteristic values need updating. Numerical examples for proof-loading and non-destructive inspection are also illustrated.