Otimização multiobjetivo de risers compósitos utilizando algoritmos bioinspirados

Abstract

Features such as high strength/weight and stiffness/weight ratios and high corrosion and fatigue resistance make composite materials appealing to the oil and gas industry, particularly for the manufacture of risers - pipes that connect the floating oil and gas production facility to the pipelines and equipment at the seabed and that are usually made of steel. The use of fiber-reinforced laminated composites, in particular, enable the riser to be tailored to meet manufacturing, safety and stability constraints under different loading conditions. This work proposes the implementation of a nature-inspired algorithm, namely Nondominated Sorting Genetic Algorithm - II (NSGA-II), and its use in a formulation capable of optimizing composite risers in the free-hanging catenary configuration considering multiple load cases and multiple objective functions. An inextensible cable model that considers vertical static loads, floater offset and maritime current loads in a fast and efficient way, especially when compared to the Finite Element Method, performs the structural analysis of the riser. This formulation is implemented in a program written in C ++ called Bio-Inspired Optimization System (BIOS) developed by the Laboratório de Mecânica Computacional e Visualização of the Universidade Federal do Ceará. A two- step validation was employed in this work: in the first stage, benchmarks problems involving mathematical functions, constrained and unconstrained, were selected and, in the second stage, benchmarks problems involving laminated plates, subjected to buckling and first natural frequency constraints, were studied. Both stages showed results consistent to the literature. The addition of specific operators for the laminated plate problems seemed to be very effective at reducing the number of generations until convergence. To the riser problem, two combinations of objective functions were optimized and different design parameters were studied, such as water depth and top angle, showing the ability of the model to adapt and to simulate several scenarios and the capacity of the algorithm of providing multiple solutions.