Optimization of the Taper/Twist Stacking Axis Location of NREL VI Wind Turbine Rotor Blade Using Neural Networks Based on Computational Fluid Dynamics Analyses

Abstract

The stacking axis locations for twist and taper distributions along the span of a wind turbine blade are optimized to maximize the rotor torque and/or to minimize the thrust. A neural networks (NN)-based model is trained for the torque and thrust values calculated using a computational fluid dynamics (CFD) solver. Once the model is obtained, constrained and unconstrained optimization is conducted. The constraints are the torque or the thrust values of the baseline turbine blade. The baseline blade is selected as the wind turbine blade used in the National Renewable Energy Laboratory (NREL) Phase VI rotor model. The Reynolds averaged Navier–Stokes (RANS) computations are done using the FINE/turbo flow solver developed by NUMECA International. The k-epsilon turbulence model is used to calculate the eddy viscosity. It is observed that achieving the same torque value as the baseline value is possible with about 5% less thrust. Similarly, the torque is increased by about 4.5% while maintaining the baseline thrust value.