Modelling the vertical lifting deformation for a deep-water pipeline laid on a sleeper

Pipelines are widely used for transporting oil resources in offshore oil exploitation. As burying an entire deep-water pipeline is not possible owing to the large water depth, an on-bottom pipeline inevitably exhibits lateral global buckling deformation under high temperature and pressure in practice. Triggering several controllable mitigatory global buckling deformations by installing sleepers under pipelines is a more effective alternative to prevent cross-sectional failure caused by excessive buckling. While sleepers may trigger vertical global buckling and this risk must be verified before they are installed in practice. This study analysed the feature of lifting deformation for a pipeline laid on a sleeper. Nine influential factors of the variation in the lifting displacement were analysed. Based on the nonlinear relationship between lifting displacement and temperature difference, three key points and four relevant key parameters for describing the lifting displacement curve were proposed and calculated for the conditions of a pipeline with different combinations of influential factors. A back propagation neural network was trained to model the relationship between the locations of the three key points and the values of the nine influential factors. The error analysis indicated that the trained network can effectively predict the vertical lifting deformation and is suitable for a pipeline that experiences no lifting deformation. Based on the feature points predicted by the trained network, the approximate profile of vertical lifting displacement during the heating process can be described, and the assessment of whether the lifting displacement with the loading conditions in practice is allowable can be conducted.