Adjoint-Based Tonal and Broadband Aeroacoustic Optimization of Propeller Blades With Amiet Model

This paper introduces an optimization framework for reducing both tonal and broadband trailing edge noise emissions from small isolated propellers by coupling the semi-empirical Amiet model with a steady-state Reynolds-averaged Navier-Stokes solver. The wall pressure spectra are approximated with Goody’s model. A novel aspect is the estimation of boundary layer thickness for different blade sections near the trailing edge, utilizing an extension to the rotating reference frame of the general method tailored for non-equilibrium flows. This enables the estimation of high-frequency broadband noise, traditionally challenging also for fully unsteady-RANS simulations. The steady-adjoint solver performs the sensitivity analysis, combined with automatic differentiation. The computational cost is independent by the number of design variables, enabling complex modifications of the blade shape. The sensitivities are used by the sequential least squares programming gradient-based algorithm for the research of local minimum. The framework is tested on the C24ND blade geometry and effectively reduce both the energies at the blade pass frequencies and the intensity of noise at higher frequencies.