Reliability-based design optimization of offshore wind turbine support structures using RBF surrogate model

An efficient reliability-based design optimization method for the support structures of monopile offshore wind turbines is proposed herein.First,parametric finite element analysis(FEA)models of the support structure are established by considering stochastic variables.Subsequently,a surrogate model is constructed using a radial basis function(RBF)neural network to replace the time-consuming FEA.The uncertainties of loads,material properties,key sizes of structural components,and soil properties are considered.The uncertainty of soil properties is characterized by the variabilities of the unit weight,friction angle,and elastic modulus of soil.Structure reliability is determined via Monte Carlo simulation,and five limit states are considered,i.e.,structural stresses,tower top displacements,mudline rotation,buckling,and natural frequency.Based on the RBF surrogate model and particle swarm optimization algorithm,an optimal design is established to minimize the volume.Results show that the proposed method can yield an optimal design that satisfies the target reliability and that the constructed RBF surrogate model significantly improves the optimization efficiency.Furthermore,the uncertainty of soil parameters significantly affects the optimization results,and increasing the monopile diameter is a cost-effective approach to cope with the uncertainty of soil parameters.