Steel Lazy Wave Riser Optimization Using Artificial Neural Networks and Genetic Algorithm

Abstract Over the past few years, a number of deepwater projects that use steel lazy wave risers have been commissioned or are under development. Steel lazy wave risers have an advantage over steel catenary risers as they offer flexibility of use with a floater having severe motion such as FPSO. They also impart lower loads at the interface with the floater compared to a traditional steel catenary riser, and hence can be used in deeper waters. Therefore, design of steel lazy wave risers has gained importance over the years as exploration of oil happens in ever deeper waters. In this paper, artificial neural networks and genetic algorithm are used to automatically generate a steel lazy wave riser design. A dataset of optimized designs of steel lazy wave risers for various inputs such as water depth, pipe OD, wall thickness etc. are generated using genetic algorithm. This dataset is used to train a neural network to automatically output a steel lazy wave riser design. The SLWR configuration that is automatically generated can be used as a starting point for conceptual and pre-FEED studies and help engineers come up with an initial SLWR design capturing the basic requirements without going through rigorous analyses. It has potential for cost savings and meeting schedule demands of fast paced projects as it will speed up the steel lazy wave risers’ design.