Abstract
This thesis investigates the interaction between soil and pipeline in sand subjected to lateral ground loading. The purpose of this study is to improve the structural modelling of buried pipelines; and also aim to produce design guidelines and construct normalised charts which will be of direct benefit to practising pipeline engineers. The research was performed entirely using the finite element (FE) method, and utilised the user subroutine of an advanced constitutive soil model that was implemented in this thesis. The problems examined in this research can be categorised into four main topics.
First, 2-D FE analyses were conducted concerning with the effects of loading rate on a laterally-loaded pipeline buried in saturated sand. Depending on the ground movement rate, the behaviour of pipelines can be different in fully-drained, partially-drained or undrained soil conditions. Based on the FE analyses results, fast-loading (undrained condition) analysis gave higher ultimate pipe force (Nq) when compared with slow-loading (drained condition) analysis, especially in dense sand. And, as partial consolidation occurs, Nq will decrease gradually from undrained value towards a fully-drained value. Boundaries for the partially-drained condition also have been derived in between Vn (dimensionless relative loading rate) of 0.02 and 50, where Vn < 0.02 represents fully-drained condition and Vn > 50 describes completely undrained condition. All indications support the conclusion that both the sand dilatancy and hydraulic conductivity of the soil in relation to the loading rate are important factors for mobilisation of the lateral resistance of a pipeline in a saturated soil medium.
Second, soil loading on a pipeline under global soil shearing conditions was investigated by performing different types of relative ground and pipeline movement modes, with the aim of generating both passive and active failure states. It was found that the load-transfer curve varies in initial stiffness; it will soften in the active loading case, but it will stiffen in the passive loading case. Shear banding is the result of localisation of deformation in soil mass, and the shear band mechanism will affect the pipe force prediction. It was found that Nq will increase when a pipeline was placed inside the soil collapsed/failure affected region, but the slip surface (or shear band) will not affect Nq if the pipeline was placed outside the failure zone. Overall, it can be concluded that the effect of global soil shearing on the interaction of soil and pipeline is relatively small in terms of Nq, implying that local soil deformation and soil dilation characteristics are more important and influential factors contributing to the magnitude of the lateral pressure on pipelines.
Third, investigations of the behaviour of an elbow-bend pipe, under lateral soil loading were performed using a 3-D FE modelling method. It was found that deeper burial pipeline, denser soil and elbow-bend pipe with larger bending angle accounted higher Nq. Also Nq at the elbow-bend pipe was about 2.7 times higher than a straight pipe. Based on these predictions, normalised charts that account for pipeline burial depths, initial elbow-bend angles, pipeline deformation modes, sand densities and effect of the soil models were proposed. 3-D multi-linear trend soil-spring models were also constructed for the use in both closing and opening mode elbow-bend pipes. In order to validate the proposed normalised charts and 3-D soil-spring models, several 3-D spring model analyses were performed. The results from the spring model analyses confirmed that the ‘3-D elbow effect’ can be ignored in the closing mode case, but in the opening mode case, the effect was computed at about 17% when compared to a 2-D bilinear soil-spring model case. Additionally, a larger effective plastic strain region was observed when 3-D soil-spring models were adopted in the design.
Fourth, in order to achieve a reliable design procedure against permanent ground deformation (PGD), a full-scale 3-D FE numerical analysis and a full-scale 3-D spring model analysis were both carried out on a 90º elbow pipeline. Encouraging and good results were achieved from both of the numerical models when compared with the data from experiments carried out at Cornell University. Thus, it is shown that the adopted 3-D FE method was able to simulate the observed pipeline performance under PGD ground failure in a reliable way.