You are hereA study of the nonlinearities, dynamics, and reliability of a drag-dominated marine structure

A study of the nonlinearities, dynamics, and reliability of a drag-dominated marine structure

Lancelot Manuel
Principal Advisor: 
C. Allin Cornell
Year Published: 
Fri, 01/01/1993 (All day)
Manuel_Lance.pdf6.48 MB

Dynamically sensitive, drag-dominated marine structures such as deep water jack-up rigs present a broad range of challenging nonlinear, non-Gaussian phenomena. Nonlinearities arise from numerous areas: kinematics (nonlinear wave theories or stretching corrections to Airy's linear wave theory), hydrodynamics (Morison drag forces), soil and/or structural nonlinearities (e.g., hysteresis and permanent deformations), large displacements (P-delta effects), etc.

Analytical results for harmonic input are presented. Simulations of response to irregular seas are studied in depth. For example, this study shows that large differential vertical settlement under the legs of a jack-up might result at soft clay sites. Stiffness degradation effects are modelled in rotational foundation springs for a jack-up resting on clay. The effect of fixity is studied by using several models with different fixity assumptions. In stochastic analyses, short-term (within seastate) studies reveal widely different response moments (first through fourth) and also very different extremes depending on modelling assumptions. Moreover, the interdependence of different aspects significantly alters extreme response estimates. These aspects are treated extensively in parametric sensitivity studies on nonlinear and linearized foundation models.

The randomness in the environmental parameters that characterize a seastate is incorporated in a reliability analysis. With long-term models for the environmental parameters and short-term stochastic simulations for the response parameters, the total reliability of a jack-up rig is computed.

The solution of the reliability problem is carried out by the first-order reliability method (FORM) which requires repeated short-term simulations in the time domain for different values of long-term parameters. There is a conflict between this requirement and accurate solutions of the short-term problem that require multiple, long simulations of input and nonlinear output. It is proposed here to use a scheme of successively more accurate analyses. In linear structure-foundation system models of a jack-up, "approximate" gradients computed from information at the start of each FORM iteration, and in nonlinear system models with load-to-response functions previously obtained, the reliability of these systems is approximately assessed first. With the benefit of these simplified/approximate analyses, the subsequent search using FORM with the exact formulation is considerably shorter. Hence, a strategy wherein less expensive, approximate methods are followed by more expensive, accurate methods can prove to be an economical one.