Parallel Dynamic Optimization of Steel Risers

The design of an ultradeepwater steel riser is known to be a challenging task, considering the several environmental load conditions to be simulated and the long time took by each simulation. To aid in this task, optimization algorithms can be used to automate the search for the best design. The problem, however, is very expensive computationally, motivating the implementation and use of a parallel optimization software coupled with a frequency domain dynamic model. This paper deals with the use of two algorithms to optimize a steel riser of given inner diameter and material for both pipe and floater: the genetic algorithm and the simulated annealing. The optimizer is left free to vary the length of the segments and the floater diameter, inside user-defined ranges. Each obtained configuration is tested in a set of load cases with different currents, offsets and waves. Optimization times and performance gain with parallelization is addressed for both algorithms.