Doctor of Philosophy / Suction caissons have been used in the offshore industry in the last two decades as both temporary mooring anchorages and permanent foundation systems. Although there have been more than 500 suction caissons installed in various locations around the world,understanding of this concept is still limited. This thesis investigates the installation aspect of suction caissons, focusing on the installation in dense sand and layered soils, where sand is inter-bedded by silt and weakly cemented layers. The research was mainly experimental, at both normal gravity and elevated acceleration levels in a geotechnical centrifuge, with some numerical simulations to complement the experimental observations. This study firstly explored the suction caisson installation response in the laboratory at 1g. The influence and effect of different design parameters, which include caisson size and wall thickness, and operational parameters including pumping rate and the use of surcharge were investigated in dense silica sand. The sand heave inside the caisson formed during these installations was also recorded and compared between tests. The 1g study also investigated the possibility of installing suction caissons in layered sand-silt soil, where caissons were installed by both slow and rapid pumping. The heave formation in this case is also discussed. The mechanism of heave formation in dense sand and deformation of the silt layer was further investigated using a half-caisson model and the particle image velocimetry (PIV) technique. The installation response at prototype soil stress conditions was then investigated in a geotechnical centrifuge. The effects of caisson size, wall thickness, as well as surcharge were investigated in various types of sand, including silica sand, calcareous sand dredged from the North Rankin site in the North West Shelf (Australia), and mixed soil where silica sand was mixed with different contents of silica flour. Comparison with the 1g results was also made. The general trend for the suction pressure during installation in homogenous sand was identified. The installation in layered soil was also investigated in the centrifuge. The installation tests were performed in various sand-silt profiles, where the silt layers were on the surface and embedded within the sand. Comparison with the results in homogenous sand was made to explore the influence of the silt layer. Installations in calcareous sand with cemented layers were also conducted. The penetration mechanism through the cemented layer is discussed, and also compared with the penetration mechanism through the silt layer. Finite element modelling was performed to simulate key installation behaviour. In particular, it was applied to simulate the sand deformation observed in the PIV tests. The likely loosening range of the internal sand plug during suction installation in silica sand was estimated. By investigating the development of hydraulic gradient along the inner wall, the principle underlying the suction response for different combinations of selfweight and wall thickness was identified. FE modelling was also performed to explore the influence of the hydraulic blockage by the silt layer. This study found that the caissons could penetrate into all soils by suction installation. Among the key findings are the observations that the suction pressure increases with depth following a distinct pressure slope, corresponding to a critical hydraulic condition along the inner wall; and the installation was possible in both layered sand-silt and uncemented-cemented soils if sufficient pumping was available. While the caisson could penetrate the weakly cemented layers well with no notable adverse effects, problems were observed in the installation in layered sand-silt soil. These include piping failure in slow pumping rate installation at 1g, and the formation of extremely unstable soil heave during installation.

Doctor of Philosophy / Suction caissons have been used in the offshore industry in the last two decades as both temporary mooring anchorages and permanent foundation systems. Although there have been more than 500 suction caissons installed in various locations around the world,understanding of this concept is still limited. This thesis investigates the installation aspect of suction caissons, focusing on the installation in dense sand and layered soils, where sand is inter-bedded by silt and weakly cemented layers. The research was mainly experimental, at both normal gravity and elevated acceleration levels in a geotechnical centrifuge, with some numerical simulations to complement the experimental observations. This study firstly explored the suction caisson installation response in the laboratory at 1g. The influence and effect of different design parameters, which include caisson size and wall thickness, and operational parameters including pumping rate and the use of surcharge were investigated in dense silica sand. The sand heave inside the caisson formed during these installations was also recorded and compared between tests. The 1g study also investigated the possibility of installing suction caissons in layered sand-silt soil, where caissons were installed by both slow and rapid pumping. The heave formation in this case is also discussed. The mechanism of heave formation in dense sand and deformation of the silt layer was further investigated using a half-caisson model and the particle image velocimetry (PIV) technique. The installation response at prototype soil stress conditions was then investigated in a geotechnical centrifuge. The effects of caisson size, wall thickness, as well as surcharge were investigated in various types of sand, including silica sand, calcareous sand dredged from the North Rankin site in the North West Shelf (Australia), and mixed soil where silica sand was mixed with different contents of silica flour. Comparison with the 1g results was also made. The general trend for the suction pressure during installation in homogenous sand was identified. The installation in layered soil was also investigated in the centrifuge. The installation tests were performed in various sand-silt profiles, where the silt layers were on the surface and embedded within the sand. Comparison with the results in homogenous sand was made to explore the influence of the silt layer. Installations in calcareous sand with cemented layers were also conducted. The penetration mechanism through the cemented layer is discussed, and also compared with the penetration mechanism through the silt layer. Finite element modelling was performed to simulate key installation behaviour. In particular, it was applied to simulate the sand deformation observed in the PIV tests. The likely loosening range of the internal sand plug during suction installation in silica sand was estimated. By investigating the development of hydraulic gradient along the inner wall, the principle underlying the suction response for different combinations of selfweight and wall thickness was identified. FE modelling was also performed to explore the influence of the hydraulic blockage by the silt layer. This study found that the caissons could penetrate into all soils by suction installation. Among the key findings are the observations that the suction pressure increases with depth following a distinct pressure slope, corresponding to a critical hydraulic condition along the inner wall; and the installation was possible in both layered sand-silt and uncemented-cemented soils if sufficient pumping was available. While the caisson could penetrate the weakly cemented layers well with no notable adverse effects, problems were observed in the installation in layered sand-silt soil. These include piping failure in slow pumping rate installation at 1g, and the formation of extremely unstable soil heave during installation.

The need for experimental and analytical modeling in the field of deep water offshore anchoring technologies is high. Suction caisson and drag embedment anchors (DEA) are common anchors used for mooring structures in deep water. The installation process of drag embedment anchors has been highly empirical, employing a trial and error methodology. In the past decade analytical methods have been derived for modeling DEA installation trajectories. However, obtaining calibration data for these models has not been economical. The development of a small scale experimental apparatus, known as the Laponite Tank, was developed for this thesis. The Laponite Tank provides a quick and economical means of measuring DEA trajectories, visually. The experimental data can then be used for calibrating models. The installation process of suctions caissons has benefited from from a more rational approach. Nevertheless, these methods require refinement and removal methodology requires development. In this thesis, an algorithm for modeling suction caisson installation in clay has been presented. An analytical method and modeling algorithm for removal processes of suction caissons in clay was also developed. The installation and removal models were calibrated to field data. These analytical and experimental studies can provide a better understanding of installation of drag embedment anchors and the installation and removal of suction caissons.

Suction caissons are a foundation system for offshore structures which offer a number of advantages over traditional piled foundations. In particular, due to the method of installation used, they are well suited for deep-water applications. The suction caisson consists of an open ended cylindrical shell, which is installed below the seabed in a sequence which consists of two loading phases. The caisson is first installed part way under self weight, with the installation being completed by lowering the pressure within the cylinder and thus allowing the ambient water pressure to force the caisson into the ground. This thesis examines a number of structural issues which result from the form of the caisson — essentially a thin walled cylinder — and the interaction of the caisson with the surrounding soil during installation. To do this, variational analysis and nonlinear finite element analysis are employed to examine the buckling and collapse behaviour of these cylinders. In particular, two issues are considered; the influence of the open end, and the interaction between the cylinder and soil on the buckling and collapse loads. First, the behaviour of open ended cylinders is considered, where the boundary condition at the open end is allowed to vary continuously from completely free to pinned, by the use of a variable lateral spring. This lateral spring restraint may be considered to represent the intermediate restraint provided by a ring stiffener which is not fully effective. The effect of various combinations of boundary conditions is accounted for by the use of a multiplier on the lower bound to the buckling load of a cylinder with classical supports. The variable spring at the open end may also be considered to be an initial, simple representation of the effect of soil restraint on the buckling load. More complex representations of the soil restraint are also considered. A nondimensional factor is proposed to account for the influence of this spring on the buckling load. One combination of boundary conditions, where the upper end of the caisson is pinned, and the lower end free (referred to as a PF boundary condition), is found to have buckling and collapse behaviour which is unusual for cylindrical shells. Buckling loads for such shells are much lower than would be found for cylinders with more typical boundary conditions, and of similar dimensions. More unusually however, PF cylinders are shown to have positive postbuckling strength. The behaviour is found to be a result of the large flexibility which results from the low restraint provided by the PF boundary conditions. This is shown by continuously decreasing the flexibility of the cylinder, by increasing the axial restraint at the pinned end. It is shown that this results in a large increase in buckling load, and a return to more usual levels of imperfection sensitivity. In particular, with an intermediate level of axial restraint, buckling loads and imperfection sensitivity are intermediate between those of PF shells with no, and with full, axial restraint. Overall however, collapse loads for PF cylinders with no additional restraint are well below those of cylinders with stiffer boundary conditions, for equal geometries. Eigenvalue buckling of cylinders fully and partially embedded in an elastic material are examined, and two analytical solutions are proposed. One of these is an extension of a method previously proposed by Seide (1962), for core filled cylinders, to pin ended cylinders which have support from both a core and a surrounding material. The second method represents the elastic support as a two parameter foundation. While more approximate than the first method, this method allows for the examination of a wider range of boundary conditions, and of partial embedment. It is found that the buckling load of the shell/soil system decreases as the embedment ratio decreases. Collapse of fully and partially embedded cylinders is also examined, using nonlinear finite element analysis. The influence of plasticity in the soil is also considered. For cylinders with small imperfections, it is found that the collapse load shows a large increase over that of the same cylinder with no soil support. However, as the size of initial geometric imperfections increases, it is found that the collapse load rapidly approaches that of the unsupported cylinder. In particular, in weak soils the gain in strength over the unsupported shell may be minimal. The exception to this is again PF cylinders. As these have relatively low collapse loads, even very weak soils are able to offer an increase in collapse load over the unsupported case. Finally, a summary of these results is provided in the form of guidance for design of such structures.