Flexible Riser Configuration Design for Extremely Shallow Water With Surrogate Model Based Optimization

Abstract

The aim of this paper is to study the optimization design of a steep wave configuration based on a surrogate model for an extremely shallow water application of a flexible riser. As the traditional technique of riser configuration design is rather timeconsuming and exhaustive due to the nonlinear time domain analysis and large quantities of load cases, it will be challenging when engineers address an extreme design, such as the configuration design in the case of extremely shallow water. To avoid expensive simulations, surrogate models are constructed in this paper with the Kriging model and radial basis function (RBF) networks by using the samples obtained by optimal Latin hypercubic sampling (LHS) and time domain analysis in a specified design space. The RBF model is found to be easier to construct and to show better accuracy compared with the Kriging model according to the numerical simulations in this work. On the basis of the RBF model, a hybrid optimization is performed to find the minimum curvature design with corresponding engineering constraints. In addition, an optimized design is found to meet all of the design criteria with high accuracy and efficiency, even though all of the samples associated with construction of the surrogate model fail to meet the curvature criterion. Thus, the technique developed in this paper provides a novel method for riser configuration design under extreme conditions.