To evaluate the required resistance factor for geotechnical foundation design, an approach was developed to evaluate the annual probability of failure and to calibrate the required deterministic safety factors. The paper describes the approach and presents the calibrated load and resistance factors corresponding to an annual probability of failure of 10-4 for three jacket structures offshore Norway and the U.K. The results of generic calculations of load and resistance factors are also presented. These compared well with the calibrated load and resistance factors obtained for the three “real-life” jackets. The calibration of the load and resistance factors requires the evaluation of the model uncertainty for each pile design method and the statistics of the uncertain parameters in the analyses. The results show that the annual probability of failure varies significantly with pile design method. The annual probability of failure depended strongly on the uncertainty in the geotechnical calculation model. Suggestions for improvements to the methodology and to the codes are made, including how to define and improve the evaluation of the model uncertainty.
The reliability analyses of one rockfill embankment dam in Norway are presented. The event tree analysis and the Bayesian network approaches, the latter combined with Monte Carlo simulations, were used to obtain the annual probability of dam breach. The analyses were run partly in a workshop format, as a process to reach consensus among several experts on the parameters to consider in the calculation of probabilities. The analyses illustrate that the reliability analyses enabled the identification of an unexpected mode of failure and later quantified the effect of rehabilitation measures on the estimated probability of dam breach.