Reliability analysis of tension leg platformsfor qave loadings

Abstract

In this dissertation, two important subjects for the safety assessment of tension leg platforms (TLP``s) are studied. One is the development of a method for the evaluation of hydrodynamic forces, and the other is the development of a TLP reliability analysis model for wave loadings. In order to evaluate the hydrodynamic forces on the TLP hull structure, a wave force evaluation method is developed, which is based on the finite element techniques. To model the fluid domain efficiently, two types of elements are developed. They are the infinite elements for modeling the radiation condition at infinity, and the fictitious bottom boundary elements for avoiding extensive fluid domain discretizations in the case of deep water waves. To illustrate the validity and versatility of those elements, the example analyses are carried out for vertical axisymmetric structures and general three dimensional structures. Comparisons with the results obtained by other available solution methods show that the present method incorporating those elements gives good results. Numerical experiments are also carried out to determine the criteria for proper distance to the infinite and fictitious bottom boundary elements from the body surface. A reliability analysis procedure of a rectangular shaped TLP for wave loadings is formulated systematically by using the domain crossing concept of the random tensile stress in the tendons. In the present reliability model, two limit conditions are considered; i.e., the exceedance of the ultimate tensile capacity and the occurrence of negative tension. Using those conditions, a system limit state for a rectangular shaped TLP is developed in terms of the platform motions in the vertical plane; i.e., heave, roll and pitch. The build-up and decay process of a severe storm event is modeled by multiple segments of stationary seastates. The strength of tendon material is also treated as a random variable with a Weibull distribution. Using the statisti...