James Gill

Symmetric airfoil geometry effects on leading edge noise.

Computational aeroacoustic methods are applied to the modeling of noise due to interactions between gusts and the leading edge of real symmetric airfoils. Single frequency harmonic gusts are interacted with various airfoil geometries at zero angle of attack. The effects of airfoil thickness and leading edge radius on noise are investigated systematically and independently for the first time, at higher frequencies than previously used in computational methods. Increases in both leading edge radius and thickness are found to reduce the predicted noise. This noise reduction effect becomes greater with increasing frequency and Mach number. The dominant noise reduction mechanism for airfoils with real geometry is found to be related to the leading edge stagnation region. It is shown that accurate leading edge noise predictions can be made when assuming an inviscid meanflow, but that it is not valid to assume a uniform meanflow. Analytic flat plate predictions are found to over-predict the noise due to a NACA 0002 airfoil by up to 3 dB at high frequencies. The accuracy of analytic flat plate solutions can be expected to decrease with increasing airfoil thickness, leading edge radius, gust frequency, and Mach number.

Synthetic turbulence methods for leading edge noise predictions

An advanced digital filter method to generate synthetic turbulence is presented for efficient two- and three-dimensional leading edge noise predictions. The technique, which is based on the Random Particle-Mesh method, produces a turbulent inflow that matches a target isotropic energy spectrum. The discretized equations for the synthetic eddies, and the input parameters needed to recover the desired turbulence statistics, are presented. Moreover, a simple and fast implementation strategy, which does not require an additional boundary condition, is presented under the frozen turbulence assumption. The method is used in a linearized Euler solver to predict turbulence-airfoil interaction noise from a number of configurations, including variations in airfoil thickness, angle of attack and Mach number. For the first time, noise predictions from a digital filter method are directly compared to those provided by synthetic turbulence based on a summation of Fourier modes. The comparison indicates that the advanced digital filter method gives enhanced performance in terms of computational cost and simulation accuracy. In addition, initial tests show that this method is capable of reproducing experimental noise measurements within 3 dB accuracy.

Single velocity-component modeling of leading edge turbulence interaction noise

A computational aeroacoustics approach is used to predict leading edge turbulence interaction noise for real airfoils. One-component (transverse), two-component (transverse and streamwise), and three-component (transverse, streamwise, and spanwise) synthesized turbulence disturbances are modeled instead of harmonic transverse gusts, to which previous computational studies of leading edge noise have often been confined. The effects of the inclusion of streamwise and spanwise disturbances on the noise are assessed. It is shown that accurate noise predictions can be made by modeling only transverse disturbances which reduces the computational expense of simulations. The accuracy of using only transverse disturbances is assessed for symmetric and cambered airfoils, and also for airfoils at non-zero angle of attack.

Evaluation and development of non-reflective boundary conditions for aeroacoustic simulations

The performance of several non-reflective acoustic boundary conditions is quantitatively compared using a plane wave test configuration. The study uses a high-order linearized Euler equation solver to find a non-reflective boundary condition which gives good performance in a variety of cases. The performance for acoustic waves with varying frequency and incident angle is compared in flows with varying Mach number. The performance of buffer zone, far-field, and characteristic non-reflective methods is compared. Some non-reflective boundary methods, such as buffer zones, contain tuneable parameters that are optimized in the current work. This provides a more comprehensive evaluation than previous studies which used constant values of the tuneable parameters. A new generic non-reflective zonal characteristic boundary condition is proposed and is shown to give improved performance in comparison to other tested methods. The performance of the proposed boundary condition is also demonstrated in a two-dimensional airfoil turbulence-interaction case that includes vortical waves leaving the domain, and in a three-dimensional duct mode case.

Prediction of Contra-Rotating Open Rotor broadband noise in isolated and installed configurations

Broadband noise is a significant part of the noise emitted by contra-rotating open rotors. Several noise sources can contribute to the total broadband sound field, with the most dominant ones probably being trailing edge noise, rotor-wake interaction noise and pylon-wake interaction noise. This paper addresses the prediction of these noise sources using analytical models based on Amiet’s flat plate airfoil theory and also to empirical turbulence models, fed by input data extracted from steady and unsteady CFD RANS simulations. The models are assessed against wind tunnel tests of Rolls-Royce’s rig 145 (build 1) conducted at the DNW anechoic open jet test facility using Rolls-Royce blades and Airbus pylons. The study showed promising results in terms of the ability of the models to predict acoustic power spectrum shapes, peak frequencies and absolute levels. The effects of changes in thrust on broadband wake-interaction noise are well reproduced. However, the models significantly underestimate the effect of thrust on trailing edge noise and the effect of rotational velocity on pylon interaction noise.

Reduced dimension modeling of leading edge turbulent interaction noise

A computational aeroacoustics approach is used to model the effects of real airfoil geometry on leading edge turbulent interaction noise for symmetric airfoils at zero angle of attack. For the first time, one-component (transverse), two-component (transverse and streamwise), and three-component (transverse, streamwise, and spanwise) synthesized turbulent disturbances are modeled instead of single frequency transverse gusts, which previous computational studies of leading edge noise have been confined to. The effects of the inclusion of streamwise and spanwise disturbances on the noise are assessed, and it is shown that accurate noise predictions for symmetric airfoils can be made by modeling only the transverse disturbances, which reduces the computational expense of simulations. Additionally, the two-component turbulent synthesis method is used to model the effects of airfoil thickness on the noise for thicknesses ranging from 2% to 12%. By using sufficient airfoil thicknesses to show trends, it is found that airfoil thickness will reduce the noise at high frequency, and that the sound power P will reduce linearly with increasing airfoil thickness.

Aerofoil geometry effects on turbulence interaction noise

Fan broadband is one of the dominant noise sources on an aircraft engine, particularly at approach. The dominant noise generation mechanism is due to turbulent- aerofoil interaction noise (TAI). This thesis investigates the effect of changes in 2D aerofoil geometry on TAI noise. The main focus of this thesis is to attempt to reduce it through the development of innovative leading edge geometries. The first two chapters of the thesis deals with an experimental and numerical investigation into the effect of aerofoil geometry on interaction noise on single aerofoils and on cascades. Consistent with previous work, they show that variations in aerofoil parameters, such as aerofoil thickness, leading edge nose radius and camber, produce only a small changes in broadband interaction noise at approach conditions. Subsequent chapters deal with the development of innovative leading edge serration profiles aimed at reducing interaction noise. Chapter 4 is a detailed study into the limitations of single-wavelength serrations in reducing interaction noise. The optimum profile is identified. Chapters 5, 6 and 7 all deal with the development of innovative profiles that can provide up to 10dB of additional noise reductions compared to single-wavelength serrations. For each of the profiles investigated a simple model is developed to aid the understanding of their interaction mechanism.

Effects of real airfoil geometry on leading edge gust interaction noise

High-order computational aeroacoustic methods are applied to the modeling of noise due to interactions between gusts and the leading edge of real symmetric airfoils. The effects of airfoil thickness and leading edge radius on noise are investigated systematically and in-dependently for the first time, at higher frequencies than previously used in computational methods. Single frequency harmonic gusts are interacted with airfoils of varying geometry at zero angle of attack. Increases in both leading edge radius and thickness are found to reduce the predicted noise. This noise reduction effect becomes greater with increasing frequency and Mach number. The dominant noise reduction mechanism for airfoils with real geometry is found to be related to the leading edge stagnation region. The assumption of uniform meanflow is shown to be invalid when modeling the leading edge noise of real airfoils. However, accurate results are still obtained when an inviscid meanflow is assumed. The accuracy of analytic flat plate solutions can be expected to decrease with increasing airfoil thickness, leading edge radius, gust frequency and Mach number.

Leading edge noise predictions using anisotropic synthetic turbulence

An advanced digital filter method is presented to generate divergence-free synthetic turbulence with homogeneous anisotropic velocity spectra. The resulting fluctuating velocity field is obtained through a superposition of anisotropic Gaussian eddies. This method is used to generate a two-dimensional turbulent flow with the key statistics of homogeneous axisymmetric turbulence. This type of turbulence has been reported in aero-engine intakes, fan wakes and open-jet wind tunnel experiments. The advanced digital filter method is implemented in a linearized Euler solver in order to investigate potential effects of anisotropic turbulence on leading edge noise. Computational aeroacoustic simulations are performed for anisotropic turbulence with streamwise-to-transverse length scale ratios ranging from 0.33 to 3 on a number of isolated airfoil configurations, including variations in mean flow Mach number, airfoil thickness and angle of attack. Noise reduction due to airfoil thickness is assessed on a NACA 0012 airfoil at zero angle of attack, showing similar trends for both isotropic and moderately anisotropic turbulent flows. Effects of anisotropic turbulence on noise become evident for airfoil configurations at non-zero angle of attack.

Gradient Term Filtering for Stable Sound Propagation with Linearized Euler Equations

A new, stable gradient term filtering (GTF) method is applied to time-domain linearised Euler equations (LEE) to compute sound propagation problems. The method employs a Laplace operator as a filter to obtain acoustic wave components. Through the filtering process, both the Kelvin-Helmholtz and the Rayleigh–Taylor instabilities can be removed from the solution process. Stability analysis confirms the stable behaviour of the solution in the presence of a sheared background mean flow, as against the conditional stability of LEE and gradient term suppression (GTS) methods. In accounting for vortical wave propagation, a curl operator can be conventionally utilised to obtain two-dimensional vortical wave components. Several benchmark test cases are studied to validate the proposed methods. Tests show that the proposed method can obtain stable solutions for acoustic wave propagation and is capable of modeling vortical interactions.