The iterative advanced mean value algorithm (AMB+), introduced nearly ten years ago, is now widely used as a cost-effective probabilistic structural analysis tool when the use of sampling methods is cost prohibitive (Wu et al., 1990). The need to establish confidence bounds on calculated probabilities arises because of the presence of uncertainties in measured means and variances of input random variables. In this paper an algorithm is proposed that makes use of the AMV+ procedure and analytically derived probability sensitivities to determine confidence bounds on calculated probabilities.
展开▼