Trailing Edge Noise Prediction Using the Non-Linear Disturbance Equations

Open Access
Jain, Abhishek
Graduate Program:
Aerospace Engineering
Master of Science
Document Type:
Master Thesis
Date of Defense:
July 15, 2015
Committee Members:
  • Kenneth Steven Brentner, Thesis Advisor
  • Philip John Morris, Thesis Advisor
  • CAA
  • CFD
  • Trailing Edge
  • Turbulence
  • Synthetic Eddy Method
  • Inflow Boundary Conditions
AIRFOIL self-noise consists of five major sources. One of these identified sources is turbulent boundary layer – trailing edge (TBL-TE) noise, which is an important source of rotor and wind turbine broadband noise, and the focus of this thesis. Trailing edge noise is the result of unsteady flow interacting with the trailing edge of an airfoil or other sharp edged flow surface. The presence of the sharp trailing edge scatters the sound generated by the turbulent eddies very efficiently, especially for sources in the immediate vicinity of the edge. There is a need for accurate and computationally efficient methods to calculate the turbulent boundary layer trailing-edge (TBL-TE) noise that are not reliant on empirical data. The majority of the current semi-empirical techniques are based on measurements from symmetric NACA airfoil sections (i.e. NACA 0012). These techniques are generally not coupled with CFD solvers to obtain turbulent boundary layer data that provides pertinent parameters used in the acoustic calculations. Some methods exist that incorporate CFD solutions like Large Eddy simulations (LES) into their noise prediction algorithms. But these are prohibitively expensive and impractical for routine use. The method described in this paper is a first principles approach that aims to predict the TBL-TE noise using computational aeroacoustic (CAA) techniques without resorting to empiricism. The prediction of trailing edge noise requires an accurate calculation of the boundary layer fluctuations in the vicinity of the trailing edge. Scales in the computational domain ranging from the small turbulent boundary layer scales to those of the long-range noise propagation need to be resolved. These data can be obtained using simulation techniques like Direct Numerical Simulation (DNS) or Large Eddy Simulation (LES). However such simulations for complete helicopters or wind turbine rotors are impractical given today’s computational resources. Also, DNS becomes unrealistic for the propagation of the acoustic signal to distant observers. The method described here overcomes these limitations by using a hybrid CAA approach coupled with a flow solver based on the non-linear disturbance equations (NLDE). The overall problem is separated into component problems with the NLDE equations applied over a relatively small noise generating region i.e. approximately the last 10% of the chord or less. This makes the solution more computationally efficient than LES for the full airfoil or rotor and enables the use of the most computationally efficient methods in the required regions. The proposed method is advantageous to helicopter and wind turbine manufacturers as it provides a tool for the prediction of rotor broadband noise at the design stage. This can also be used as a tool to reduce noise through the analysis of appropriate noise reduction devices.