Reliability based design of marine risers
The harsh environment in which offshore structures must operate, their intended service life and the uncertainties inherent to the load processes, have been the impulse for investigation of their reliability. The method most extensively applied for this purpose during the last two decades was the Structural Systems Reliability, which can not be coupled with the finite element method. Therefore the objectives of the present work are to investigate the applicability of a technique which allows the utilization of the reliability analysis methods with a marine riser modelled by the finite element method, FEM, and revision of the reliability levels associated with this riser, including the fatigue life. For these purposes the response surface methodology was selected, among a number of methods. A response surface approach which requires a low number of experiments with the FEM model was elected, calculations for construction of the response surface are further simplified by the assumption of statistical independence among the basic variables. It is demonstrated in the present study that the response surface is capable of producing an equivalent and explicit limit state function which is used at a second stage with the First Order Reliability Method and the Adaptive Importance Sampling simulation technique. However, it was found that the assumption of independence is not always valid. In this case, a method is proposed in which the correlated variables are implicitly considered at the level of the mechanical model. The reliability of the marine riser was reviewed with the proposed algorithms, finding that the validity of the reliability levels depend on the number of basic variables considered and their statistical properties. The significant reduction in required computing time achieved with the response surface methodology allowed parametric studies to be carried out, in order to investigate the impact of different statistical properties of the basic variables. The fatigue reliability case was also investigated with the S-N approach. The introduction of uncertainty in the fatigue life estimation proved that acceptable levels of deterministic fatigue life may render unacceptablelevels of reliability. The uncertainty associated with the stress range is the most significant variable, though the present fatigue reliability formats consider it in a very simplified manner, therefore an approach is suggested with which the stress uncertainty can be considered in a more detailed fashion. However, the algorithm used here for construction of the response surface was unable to produce the required surface. Therefore it is concluded that though the response surface is capable of handling a large number of structural reliability cases, there are instances in which more research efforts are needed.