Low-order methodology for the design of propellers with serrated trailing edges

Abstract

The exponential increase in applications where UAVs or drones are used raises concerns about potential noise pollution in large cities. Trailing edge noise is a significant broadband noise source of hovering UAV propellers, which can be reduced by employing bio-inspired trailing edge serrations. The high computational cost of high-fidelity CFD simulations put them at the end of the design cycle rather than at the beginning. Therefore, propeller designers need a fast method for predicting noise reductions. Analytical models for straight and serrated edges need as essential input the single-point wall-pressure fluctuations spectrum. This can be modeled using low-cost Reynolds-Averaged Navier-Stokes simulations. A low-order methodology is proposed to estimate the potential noise reductions resulting from trailing edge serrations using RANS simulations for a representative drone propeller based on a NACA0012 airfoil with constant pitch and constant chord. Ayton’s theoretical model provides predictions for serrated trailing edge noise generated by a fully turbulent flow over an infinitesimally thin plane. The extension of Ayton’s model, proposed by Li and Lee, provides a heuristic three-dimensional model for a finite span applicable to rotor blades. This model reveals the potential benefits of using a square wave serration compared to the traditional sawtooth serration. This thesis addresses the limitations of Li and Lee’s model by deriving a new model for the square wave using Ayton’s model and Curle’s analogy. Measurements of a NACA0012 airfoil at low-Reynolds numbers, typical of small drones, were performed in an anechoic chamber at the Université de Sherbrooke, for straight, sawtooth, and square wave edges. Noise reductions of up to 5 dB are measured, with the square wave outperforming the sawtooth serration. Theoretical predictions are in reasonable agreement with the experimental results. Li and Lee’s model is then extended to rotating blades using Schlinker and Amiet’s model. The model is verified in the limit of zero serration amplitude finding good agreement at high frequencies and high observer angles. Single-blade passage RANS simulations of the NACA0012 propeller are performed, and aerodynamic validation is made with experimental data. The results highlight the importance of adequately refining the mesh around the propeller tip vortices and using transitional turbulence modeling. The wall-pressure fluctuations spectrum was modeled based on the RANS results, and the propeller far-field acoustics were calculated using the in-house code PyFanNoise. The acoustic predictions agree fairly well with experimental measurements, especially at high rotational speeds, where secondary flows are weaker and the onset of turbulence matches more favorably with the fully turbulent k- SST model used in the RANS. Li and Lee’s model is then used to study the sensitivity of noise reductions to different shapes. The square wave serration is shown to outperform the sawtooth and sinusoidal shapes for all frequencies and observer angles, particularly for small propeller blades typically used in drones. However, for larger chord blades typically used for ducted fans, combinations of sawtooth and sinusoidal serrations provide better noise reductions. The methodology is validated by considering the effects of serration installation and manufac turing. Several propellers were 3D printed and tested in an anechoic chamber, where far-field noise and aerodynamic performance were measured. The baseline configuration exhibits clear evidence of laminar boundary-layer instability noise. Cut-in and add-on serrations alleviate this noise mechanism. Similarly, to overcome the influence of the laminar-to-turbulent transition over the blade surface, some propellers also include additional surface roughness to trigger turbulence. Cut-in serrations experience additional vortex-shedding noise characterized by a Strouhal number based on the serration root thickness. The results show that serrations are a viable method for controlling trailing edge noise at low RPM, where broadband noise dominates over tonal noise, and that add-on serrations with a trip are in better agreement with the theoretical results, thus highlighting the importance of the manufacturing method during the design phase.

Date	17 Dec 2025
Original language	American English
Awarding Institution	École de technologie supérieure
Supervisor	Marlène Sanjosé (Supervisor) & Stéphane Moreau (Co-supervisor)