Parameter sensitivity of failure probabilities

From MaRDI portal





In many reliability problems the solution depends on various parameters, whose values are not known exactly. Often for these parameters some reasonable estimates are calculated and then these estimates are treated as if they were the true parameter values. The essential lack of these methods is that they neglect of the gap between the given data and the mathematical model. With the data the structure of the model and its parameters can be estimated only with some uncertainty. The problem of calculating the structural safety in such cases is considered in [\textit{A. Der Kiureghian}, J. Eng. Mech. Div. ASCE 115, No. 5, 1119-1140 (1989)]. It is important to have information about the influence of changes in the parameters on the result, which is in general the failure probability. First results for this problem can be found in the Ph.D. thesis of \textit{M. Hohenbichler} (TU Munich, 1984). A drawback of this work is that for non-normal random variables no analytic results are obtained. Further only the influence on the beta value and not on the failure probability is considered. The basic idea for constructing simple approximations for these sensitivity factors is the same as for deriving estimates for the failure probability. The Laplace method is used for obtaining asymptotic approximations of these quantities. For the entire collection see [Zbl 0753.00026].











This page was built for publication: Parameter sensitivity of failure probabilities

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4036679)