Configuration design of a steel double lazy wave riser based on metamodel-assisted metaheuristic algorithms

Abstract

This study presents an approach to the configuration design of a steel double lazy wave riser. Considering the length of each segment and arrangement parameters of buoyancy blocks as design variables, the objectives are to minimize the maximum static stress, horizontal motion scope, and maximum standard deviation of dynamic stress, respectively. A self-developed program was established to calculate the riser dynamics in a fully coupled platform/mooring/riser system. An improved Particle Swarm Optimization (PSO) that can handle both continuous and discrete design variables was proposed to automatically find the optimal solution. Feedforward Neural Networks (FNNs) corresponding were embedded into the optimization algorithm to replace the time-consuming numerical calculation. The results show that the FNN provides high accuracy with an error of no more than 6.34 % and all Determination Coefficients of about 0.95. The static-stress optimization tends to reduce the arch height in the riser; while the dynamic optimizations tend to increase the arch heights. Three optimizations all reach feasible solutions with noticeable objective reductions.