Finite Element Analysis of a Flexible Pipe Interlocked Carcass Under Tension Loads

Abstract

To correctly model the structural behavior of a flexible pipe, the contribution of all the layers must be completely understood, among them the interlocked carcass. That carcass is a metallic layer designed to provide radial stiffness to a flexible pipe, mainly supporting pressure differentials and thus preventing failure modes such as collapse and crushing, but its behavior under other loads is worth of investigation. This paper contributes to understanding the carcass behavior under tension. Given its complex helical and interlocked geometry, modelling the carcass through the Finite Element Method is a challenging task, not only due to the large size of the models, but also due to the nonlinearities and convergence difficulties that arise from the self-contacts at the interlocking. For these reasons, most works developed over the past decades have adopted an equivalent layer approach, in which the carcass is replaced by an orthotropic cylindrical layer with equivalent mechanical properties. Although practical, this approach disregards the effects from the interlocking, such as stiffness variations and stress concentrations. Therefore, aiming a more realistic representation and a better understanding of the mechanical behavior of the interlocked carcass, this work presents four different carcass finite element models to analyze this layer under tension loads. The first one is a complete three-dimensional finite element model of an interlocked carcass discretized with second order isoparametric solid elements and surface-to-surface contact elements. The second model consists of a version of the first one with the addition of an inner polymeric sheath. As for the third and fourth models, it was adopted the simplifying ring hypothesis, that is, a carcass with 90 degree lay angle, thus allowing the axisymmetric modelling of the two previous configurations, representing a substantial computational gain by using two-dimensional meshes. The results of those models are then presented and compared, and the validity of the adopted simplifying hypothesis is verified.