Stochastic Simulation Technique Of Multivariate Distribution And Its Application In Ocean Engineering

For an economic and safe design of marine structures, the joint probability of multivariate extreme environmental loads should be considered. This work studies two aspects of the simulation technique in the joint probability analysis method and its application to offshore engineering.The first part concentrates on the Importance Sampling Procedure which can reduce statistical variance and enhance efficiency in calculating the failure probability. In this work the procedure is used in the design of jacket platform and jack-up platform. Through the comparison with traditional design methods, the present approach leads to some suggestions about determining the design standard for platform.The second part focuses on the multivariate compound extreme value distribution. Compound extreme value distribution is extended to three dimensions (three variables) in this study. The approach is more flexible and reasonable because the statistical characteristics of the extreme values and the influence of storm occurring frequency are both considered. Also it has great advantages in choosing samples. Furthermore, a particular model-Poisson-Nested Logistic model is developed.As the number of dimensions is increased, the joint probability for a failure criterion is difficult to be computed analytically. Consequently the stochastic simulation method of compound extreme value distribution is brought forward. The present study shows that the simulation method is reliable and can give accurate solution. As an example, Poisson-Nested Logistic model is applied to the design of sea wall. Using the simulation method with optimization, a set of combined values of wave height, water level and wave period with joint return period of 100 years are determined as the design reference for sea wall.