Stochastic methods for computational aero-acoustics

Lead Research Organisation: University of Southampton
Department Name: Faculty of Engineering & the Environment

Abstract

The aim of the research programme is to develop a new class of numericalmethods for the solution of stochastic aeroacoustic problems. The random natureof many aeroacoustic problems stems from the presence of turbulent flows whichact as noise sources and also as an uncertain medium through which soundpropagates. The challenge is to devise numerical methods which preserve in somesense the uncertainty of the sound field and still represent accurately thecomplex physics of sound propagation through non-uniform flows. The projectwill focus on solutions of two aeroacoustic problems of urgent practicalinterest: (1) broadband fan noise where fan blades interact with their boundarylayers or turbulent wakes, (2) turbulent scattering by jet mixing layers wherethe unsteady random velocity field of a turbulent flow distribute the originalfrequency content of the source of sound. There is currently no generalsimulation methods available for these two applications and stochastic methodsrepresent a novel approach to the prediction of these sources of sound.Stochastic methods will be developed specifically for aeroacoustic applicationsand will draw on existing methods from other fields of engineering. After areview of the literature a selection of stochastic numerical methods will beimplemented and tested against benchmark problems. The accuracy andcomputational costs of the numerical schemes will be compared. Their ability tosolve large-scale realistic problems will be assess by considering anindustrial application. The proposed project will increase the range ofapplications that can be treated by computational aeroacoustics and willcontribute to the reduction of transport noise and especially aircraft noise.
 
Description The aim of the research programme was to develop a new class of numerical methods for the solution of stochastic aeroacoustic problems. The random nature of many aeroacoustic problems stems from the presence of turbulent flows which act as noise sources and also as an uncertain medium through which sound propagates. The challenge is to devise numerical methods which preserve in some sense the uncertainty of the sound field and still represent accurately the complex physics of sound propagation through non-uniform flows.

The project focused on predicting broadband fan noise which is generated when fan blades or stator vanes interact with their boundary layers or with turbulent wakes from other blades located upstream. After some preliminary tests the project concentrated on a variant of the random-vortex-particle method, where the turbulence impinging on the aerofoil is represented by a set of point vortices with random strengths. This method was preferred against other methods, especially stochastic methods based on Fourier, for its ability to represent efficiently inhomogeneous turbulent fields.

The following is a list of the main achievements of the project:
- The stochastic method was extended to describe non-Gaussian spectrum, by modifying the nature of the filter. This technique has proved to be quite accurate (provided some care is taken with the numerical implementation) and allows for more realistic turbulence spectra to be modelled.
- The problem of representing evolving, rather than frozen, turbulence has been tackled. Again this allowed a more realistic description of the targeted applications. It was demonstrated that using standard Langevin equations to represent the evolution of the vortex strengths is not well suited for numerical simulation of the acoustic field. A more complex, but more robust, stochastic model was devised and validated for that purpose.
- Inhomogeneous turbulence was also modelled using the random-vortex-particle method to provide an accurate description of the train of turbulent wakes seen by the stator vanes in a turbofan engine.

The numerical tools developed in this project were validated against analytical models for a flat plate interacting with a uniform turbulent stream. And numerical predictions have also been compared against experimental data for an isolated aerofoil in a turbulent jet, and very good agreement was observed. The project was able to increase the range of applications that can be treated by computational aero-acoustics by demonstrating that stochastic methods can be used successfully to predict broadband fan noise.
Exploitation Route The prediction methods devised during the project will be further developed during a project funded by TSB and Rolls-Royce, with a particular focus on broadband fan noise.

Also another project, funded by EPSRC, will involve the prediction methods developed in this project to study the spectrum broadening caused by the interaction of sound with turbulence.
Sectors Aerospace, Defence and Marine,Energy,Transport
URL http://www.southampton.ac.uk/engineering/research/projects/synthetic_turbulence_for_broadband_fan_noise_simulations.page?
 
Description Reducing noise emissions from aircraft engines is crucial to the well-being of residents around airports and to the competitiveness of the UK aerospace sector. For modern and future aircraft engines the fan stage is a major contributor to the radiated noise. This project delivered more efficient computational methods to support the design of low-noise fan stages. These methods are currently used and further developed in a project with Rolls-Royce plc. In addition these methods have been adopted by other research institutes in France (ONERA) and in Germany (DLR).
First Year Of Impact 2012
Sector Aerospace, Defence and Marine,Energy,Transport
 
Description HARMONY
Amount £900,000 (GBP)
Funding ID 101367 
Organisation Innovate UK 
Sector Public
Country United Kingdom
Start 02/2013 
End 01/2016
 
Description Collaboration with Rolls-Royce on Stochastic Methods 
Organisation Rolls Royce Group Plc
Department Noise Department
Country United Kingdom 
Sector Private 
PI Contribution We contributed our expertise in computational aero-acoustics, and conducted all the development, validation and assessment of the computational methods developed in this project.
Collaborator Contribution Staff at Rolls-Royce provided feedback and advice on the benchmark problems and the assessment of the computational methods. They helped ensure that the academic research conducted in this project remained relevant from an industrial perspective.
Impact - Assessment of novel computational methods for industrial applications. - Continued collaboration funded by Rolls-Royce and TSB to further develop these methods.
Start Year 2007