A Surrogate-Based Approach for Bend-Twist Coupling Optimization of a Horizontal Axis Wind Turbine Composite Blade

A design study on the effect of bend-twist coupling in a composite blade of horizontal axis wind turbine (HAWT) is conducted. Using specific lamination sequences with unidirectional plies, a laminated composite is obtained whose mechanical stresses are coupled in its bending and twisting. Given the aerodynamic forces on the blade during its operation, mechanical stresses emerge in the material, which, due to its bend-twist coupled behavior, conforms elastically, adapting its geometry to the wind flow. The tailoring of aeroelastic effects can be applied to wind turbine blades to improve the performance of wind turbines. Furthermore, the manufacturing costs, either from material or process perspectives, are potentially reduced with well-designed layup sequences. A blade is first designed according to specific parameters using the Blade Element Momentum Theory (BEMT) under the assumption of total rigidity. These include operational and constructive parameters (e.g., desired output power, airfoil profile characteristics, nominal wind velocity, number of blades, etc.). The effect of bend-twist coupling in the laminated composite is analyzed by fluid-structure interaction (FSI). The layup sequence is parametrized within ANSYS Composites PrepPost (ACP) with different laminations for upper and lower face of the blade, given both geometry and stress distributions differ for each face. The CFD and the structural behavior are coupled in a two-way FSI system, along with material properties evaluated from ACP. Thus, deformations are computed using the pressure distribution from CFD and material properties from ACP. The power coefficient is calculated once the FSI simulation has converged, yielding the resultant torque and rotation, which are then used to evaluate the extracted power.. An optimization routine is implemented to maximize the power extraction as a function of the lamination sequence for the upper and lower faces of the blade. The computational cost of the optimization procedure is reduced by employing radial basis function neural network, acting as a surrogate model. It is predicted an increase of up to 10% in the power coefficient for the generator. The final proposed laminated composite for upper and lower faces of the blade extends the power curve of the generator, allowing it to sustain operation even in overload conditions, delaying stall of the blades and avoiding tower collapse by excessive blade tip axial deflection.