Computational and modelling aspects of marine riser analysis

Phaal, Robert

Computational and modelling aspects of marine riser analysis

Thesis / Dissertation

1987

Abstract

[page 119, 132 missing] In recent years much interest has been shown iri the design of offshore structures. Flexible members such as marine risers and mooring cables are critical components in such structures. These systems are characterised by nonlinear geometry and loading. In the formulation of an algorithm to model such systems, simplifying assumptions must often be made. This thesis attempts firstly to give an overview of the literature available in this field. The testing of a riser model is important to ensure that it may be used with confidence. There are several sources of error inherent in a model of this nature. Modelling errors are those caused by the simplifying assumptions made in developing the mathematical statement of the model. Numerical errors include those due to the approximation method and the finite accuracy of the computing machine. Finally, there is a level of randomness in the design parameters, which results in a lack of confidence and an uncertainty in the response obtained from a deterministic model. This thesis attempts to qualify, if not quantify, the sources of error innate to the modelling of flexible offshore structures; in particular, marine risers. Extensive use has been made of a sophisticated, commercially available finite element program, ABAQUS. Three commonly occuring riser configurations have been modelled successfully with ABAQUS. These are the standard, the catenary, and the hanging riser configurations (see figure 1.1). A geometrically linearised finite element riser model has also been developed and tested. The linearised model is shown to be applicable to problems where the maximum model deflection is less than approximately 10% of the riser length. A Probabilistic Finite Element Method (PFEM) has been implemented in order to investigate one source of uncertainty in the problem: that of the hydrodynamic loading. The method is shown to have limitations expressed by coefficient of variation bounds for the random parameters. The C data available in the literature is within these bounds, d while the C data is not. These bounds are shown to be a problem dependent, and the PFEM to be more applicable to the modeiling of the uncertainties associated with the analysis of drag dominated problems such as mooring cables.

Keywords

Science

Reference:

Collections

Masters

Full item page