Artificial Neural Networks for Optimization procedures of Mooring System of Floating Platforms

Resumo

Floating production systems (FPS) are widely used for oil exploitation by offshore industry. Installation of these systems have been advanced to deep and ultra-deep water subject to both extreme and operational environmental conditions. Under these conditions, mooring systems assume a fundamental role of keeping the FPS on the location and thus ensuring the integrity of other systems. Numerical model of these systems requires rigorous nonlinear static and dynamic analysis in the time domain using Finite Element Method (FEM) which have high computational costs. Therefore, an optimization process that may requires hundreds or thousands of analyses of the candidate solutions can take a long time. Thus, this work aims to optimize a mooring system of a floating production system, changing the evaluation of the objective function and the associated constraints from Finite Element Procedure by an Artificial Neural Network, in order to reduce the computational costs. Such reduction of time consuming may favor the accomplishment of several studies of the system in question. Case study presents a real-world scenario and the optimization tool employs the Particle Swarm Optimization (PSO) method. From the results we can see that the replacement of the FEM analyses by ANN meta-model has a high level of accuracy and presents low computational cost.