Thesis: Deqiong Kong

Offshore on-bottom pipelines are vulnerable to lateral buckling caused by thermal- and pressure-induced axial expansion. A cost-effective solution is to allow for the buckling to occur in a controlled manner rather than to prevent it. This design approach requires an accurate assessment of the soil resistance experienced by the pipe during large cyclic lateral displacements.

The sequential limit analysis (SLA) method has been developed to study soil-structure interaction problems involving large displacements and large plastic strains in purely cohesive materials such as undrained clay. This approach was chosen because of its high computing efficiency as well as the robustness of limit analysis in solving plasticity problems. New techniques were developed to implement this approach, including model geometry updating routines, treatment of external model boundaries, periodic remeshing and interpolation methods as well as a constitutive model accounting for strain softening and strain rate effects. The SLA method was validated by benchmarking against known analytical solutions and physical model tests, as well as against output from complementary analyses performed using the coupled Eulerian-Lagrangian (CEL) approach in-built in Abaqus. The same constitutive model used in SLA was implemented into the CEL model via an Abaqus VUMAT subroutine to ensure comparability of results. Apart from these comparisons, the computing discrepancy between the lower and upper bound limit analysis solutions and the incompressibility of the deforming material were carefully examined to demonstrate the validity of the SLA approach.

Using the SLA method the vertical penetration behaviour of a pipe was investigated to< an embedment of three diameters. The transition of soil failure mechanisms from shallow to deep embedment was examined carefully via a parametric study, and the strain softening effect was recognised to be extremely significant in determining the penetration resistance. A new simplified vertical penetration resistance calculation was developed, taking account of influences of interface roughness, soil strength gradient, soil unit weight and other parameters related to strain softening and strain rate.

Lateral pipe-soil interaction behaviour (monotonic and cyclic) was explored using the SLA method for lateral displacements up to twenty pipe diameters, with a focus on the soil resistance at the initial ‘breakout’ stage as well as the steady-state residual stage. In addition to the routine output of pipe load-displacement data (e.g. invert trajectory and lateral resistance), yield envelopes during the loading were also derived to provide more comprehensive understanding of the pipe-soil interaction. For monotonic loading, lower and upper bound estimates for the critical pipe weight that differentiates light pipe rising behaviour from heavy pipe diving behavior, were derived. Empirical equations were proposed to predict the residual resistance of a light pipe, accounting for the influences of pipe weight, initial pipe embedment, soil strength gradient, soil unit weight and finally strain softening effects. For cyclic loading, the numerical results for a number of different loading cases compared convincingly to corresponding centrifuge model test data, providing confidence in the numerical modelling approach. Parametric studies were completed exploring the influences of initial pipe embedment, pipe weight, soil strength gradient, soil unit weight, as well as the strain softening effect.