Data-driven multiobjective optimization of a flexible riser carcass layer

Abstract

The carcass layer is the innermost structure of a flexible riser and is mainly used to resist external hydrostatic pressure in deep water. Its complex cross-section shape and numerous parameters pose a challenge to the accurate evaluation and optimal design of critical collapse pressure. In this work, a finite element model for the nonlinear buckling analysis of the carcass layer is established. The simulation results of finite element analysis are compared with the experimental results, and the error between these results is found to be 6.93%. On the basis of the simulation results, a data-driven high-precision surrogate model is established to predict the critical collapse pressure of the carcass layer and analyze the sensitivity of key parameters of the carcass layer's cross-section. Thickness is found to play a decisive role in critical collapse pressure, accounting for 64.55%. By taking the maximum critical collapse pressure and minimum mass of the carcass layer as optimization objectives, the key parameters of the carcass layer section are optimized by a multiobjective particle swarm optimization algorithm to obtain the Pareto optimal solution. Three groups of optimal section parameters are obtained in accordance with three groups of application conditions. The errors of critical collapse pressure between the prediction and finite element models for the three groups of optimized cross-sections are less than 3.00%, thus verifying the accuracy of the predicted results.