Dependency effect on the reliability-based design optimization of complex offshore structure

A tradeoff between cost and safety is essential in the reliability-based design of offshore support structures operating in uncertain harsh environments. This study proposes a dependence-based double-loop optimization framework for complex structural systems under such environmental conditions. It considers the dependency of the environmental variables using a Drawable Vine (D-vine) copula. The reliability (inner loop of the design cycle) is modeled using the adaptive Polynomial Chaos Kriging (PCK) as a metamodel. The study employs a hybrid optimization approach that combines Genetic Algorithm (GA) and Sequential Quadratic Programming (SQP) in the outer loop optimization phase. The dependency effect is demonstrated on a steel column and a deep water segmented Steel Catenary Riser (SCR) at various hang angles to the offshore structure. The study shows the importance of multivariate dependence modeling in Reliability-Based Design Optimization (RBDO). It also highlights the significance of: (i) optimal copula selection, (ii) the impact of variable order in the D-vine copula's dependence tree, and (iii) the efficiency provided by the PCK metamodel. The method described in this paper provides a road map for a dependency-based optimal design of complex ocean structures. Also, it allows for strategic design decision-making under uncertainty, considering cost and safety.