Aeroacoustic airfoil shape optimization enhanced by autoencoders

Aeroacoustic noise is a major concern in wind turbine design that can be minimized by optimizing the airfoils that shape the rotating blades. To this end, we present a framework for airfoil shape optimization to reduce the trailing edge noise for the design of wind turbine blades. Far-field noise is evaluated using Amiet's theory coupled with the TNO-Blake model to calculate the wall pressure spectrum and fast turn-around XFOIL simulations to evaluate the boundary layer parameters. The computational framework is first validated using a NACA0012 airfoil at 0 degrees angle of attack. Particle swarm optimization is used to find the optimized airfoil configuration. The multi-objective optimization minimizes the A-weighted overall sound pressure level at various angles of attack, while ensuring enough lift and minimum drag. Furthermore, traditional shape optimization techniques show difficulties to converge to the optimum, when too many shape parameters are included in the optimization. For this reason, we compare classic parameterization methods to define the airfoil geometry (i.e., CST) to a machine learning method (i.e., a variational autoencoder). We observe that variational autoencoders can represent a wide variety of geometries, with only four encoded variables, leading to efficient optimizations, which result in improved optimal shapes. When compared to the baseline geometry, a NACA0012, the autoencoder-based optimized airfoil reduces by 3% (1.75 dBA) the overall sound pressure level (with decreased noise across the entire frequency range), while maintaining favorable aerodynamic properties in terms of lift and drag.

Automated summary

This study presents a framework for optimizing the airfoil shape of wind turbine blades to reduce trailing edge noise. The framework uses Amiet's theory, the TNO-Blake model, and XFOIL simulations to evaluate noise and boundary layer parameters. Particle swarm optimization is employed to find the optimized airfoil configuration, while traditional shape optimization techniques are compared to machine learning methods using a variational autoencoder. The autoencoder-based optimized airfoil reduces overall sound pressure level by 3% (1.75 dBA) and improves aerodynamic properties compared to the baseline NACA0012 airfoil.