A metamodel-based optimization approach to reduce the weight of composite laminated wind turbine blades

Abstract

In wind turbine blades, the complex resultant geometry due to the aerodynamic design cannot be modified in the successive mechanical design stage. Hence, the reduction of the weight and manufacturing costs of the blades while assuring appropriate levels of structural stiffness, integrity and reliability, require a composite material layout that must be optimally defined. The aim of this work is to present a metamodel-based method to optimize the composite laminate of wind turbine blades. This methodology combines a genetic algorithm (GA) with an artificial neural network (ANN) in order to reduce the computational cost of the optimization procedure. Therefore, at first, representative samples were built to train and validate the ANN model, and then, the ANN model is coupled with GA to find the optimal structural blade design. As an actual case study, the method was applied to redesign a medium-power 40-kW wind turbine blade to reduce its mass while structural and manufacturing constrained are fulfilled. The results indicated that is possible to save of up to 20% of laminated mass compared to a reference design. Furthermore, a 40% reduction of the computational cost was achieved in contrast with the typical simulation-based optimization approach.