Assessment of statistical parameter uncertainty in the reliability analysis of jacket platforms

Abstract

Statistical uncertainty is undoubtedly introduced in the estimation of parameters usually involved in the physical or probabilistic models employed for reliability analysis. Provided that such models parameters are uncertain, the probability of failure and the reliability index become uncertain variables as well. In this work, a general formulation is introduced to account for parameter uncertainties in the reliability analysis of jacket platforms under storm conditions. A point estimates method based on the Rosenblatt transformation is used to compute the statistics of the failure probability and the reliability index. It has the advantage that the estimating points and weights are readily defined. The effect of considering parameter uncertainties in the probability distribution of maximum wave height and the environmental load model was assessed. The relative influence of the different sources of parameter uncertainty and the implications of neglecting it were examined. Credible bounds of failure probability and coefficients of variation of the reliability index were evaluated as means of charactering uncertainty in their estimation. A decision problem regarding optimal probabilities of failure for structural design using a risk model is also analyzed, showing that depending on the coefficient of variation of the failure probability greater safety margins may be required.