Theoretical and data-driven methods to predict the mechanical response of flexible pipe carcass under radial compression

The carcass layer is the innermost structure of flexible pipes and is primarily designed to resist radial loads. Flexible pipes are subjected to compression by tensioners during installation, and excessive radial loads can lead to plastic deformation and premature material yielding, consequently diminishing the structural load-bearing capacity. Despite the critical role of the carcass layer, limited research has been performed on the plastic analysis of its complex cross-section under symmetric radial compression. This study applied the plastic-hinge theory to a planar circular ring, considering factors such as ellipticity and material hardening, and introduced the concept of equivalent radial stiffness for the carcass layer to predict the load-displacement curve during the plastic phase of the carcass layer under radial compression. Simultaneously, owing to the challenge of uneven stress distribution caused by a complex cross-section, this study adopts a combined approach of the attention mechanism and Long Short-Term Memory (LSTM) neural network. By inputting the structural ellipticity, inner diameter, and load–displacement response, the method aims to accurately predict the stress in the carcass layer. This conclusion indicates that the theoretical model exhibits a higher prediction accuracy when material hardening is considered. Errors arise when material hardening is ignored, as the theoretical model fails to account for the deformation of nonuniform cross-sections. By contrast, the data-driven models demonstrated high precision in predicting both radial and circumferential stresses in the carcass layer under radial compression.