Aeroacoustics of Low Mach Number Flows: Fundamentals, Analysis, and Measurement

Aeroacoustics of Low Mach Number Flows

Fundamentals, Analysis and Measurement

First Edition

Stewart Glegg

William Devenport

Table of Contents

Cover image

Title page

Copyright

Dedication

Preface

Part 1: Fundamentals

1: Introduction

Abstract

1.1 Aeroacoustics of low Mach number flows

1.2 Sound waves and turbulence

1.3 Quantifying sound levels and annoyance

1.4 Symbol and analysis conventions used in this book

2: The equations of fluid motion

Abstract

2.1 Tensor notation

2.2 The equation of continuity

2.3 The momentum equation

2.4 Thermodynamic quantities

2.5 The role of vorticity

2.6 Energy and acoustic intensity

2.7 Some relevant fluid dynamic concepts and methods

3: Linear acoustics

Abstract

3.1 The acoustic wave equation

3.2 Plane waves and spherical waves

3.3 Harmonic time dependence

3.4 Sound generation by a small sphere

3.5 Sound scattering by a small sphere

3.6 Superposition and far field approximations

3.7 Monopole, dipole, and quadrupole sources

3.8 Acoustic intensity and sound power output

3.9 Solution to the wave equation using Green's functions

3.10 Frequency domain solutions and Fourier transforms

4: Lighthill's acoustic analogy

Abstract

4.1 Lighthill's analogy

4.2 Limitations of the acoustic analogy

4.3 Curle's theorem

4.4 Monopole, dipole, and quadrupole sources

4.5 Tailored Green's functions

4.6 Integral formulas for tailored Green's functions

4.7 Wavenumber and Fourier transforms

5: The Ffowcs Williams and Hawkings equation

Abstract

5.1 Generalized derivatives

5.2 The Ffowcs Williams and Hawkings equation

5.3 Moving sources

5.4 Sources in a free stream

5.5 Ffowcs Williams and Hawkings surfaces

5.6 Incompressible flow estimates of acoustic source terms

6: The linearized Euler equations

Abstract

6.1 Goldstein's equation

6.2 Drift coordinates

6.3 Rapid distortion theory

6.4 Acoustically compact thin airfoils and the Kutta condition

6.5 The Prantl–Glauert transformation

7: Vortex sound

Abstract

7.1 Theory of vortex sound

7.2 Sound from two line vortices in free space

7.3 Surface forces in incompressible flow

7.4 Aeolian tones

7.5 Blade vortex interactions in incompressible flow

7.6 The effect of angle of attack and blade thickness on unsteady loads

8: Turbulence and stochastic processes

Abstract

8.1 The nature of turbulence

8.2 Averaging and the expected value

8.3 Averaging of the governing equations and computational approaches

8.4 Descriptions of turbulence for aeroacoustic analysis

9: Turbulent flows

Abstract

9.1 Homogeneous isotropic turbulence

9.2 Inhomogeneous turbulent flows

Part 2: Experimental approaches

10: Aeroacoustic testing and instrumentation

Abstract

10.1 Aeroacoustic wind tunnels

10.2 Wind tunnel acoustic corrections

10.3 Sound measurement

10.4 The measurement of turbulent pressure fluctuations

10.5 Velocity measurement

11: Measurement, signal processing, and uncertainty

Abstract

11.1 Limitations of measured data

11.2 Uncertainty

11.3 Averaging and convergence

11.4 Numerically estimating fourier transforms

11.5 Measurement as seen from the frequency domain

11.6 Calculating time spectra and correlations

11.7 Wavenumber spectra and spatial correlations

12: Phased arrays

Abstract

12.1 Basic delay and sum processing

12.2 General approach to array processing

12.3 Deconvolution methods

12.4 Correlated sources and directionality

Part 3: Edge and boundary layer noise

13: The theory of edge scattering

Abstract

13.1 The importance of edge scattering

13.2 The Schwartzschild problem and its solution based on the Weiner Hopf method

13.3 The effect of uniform flow

13.4 The leading edge scattering problem

14: Leading edge noise

Abstract

14.1 The compressible flow blade response function

14.2 The acoustic far field

14.3 An airfoil in a turbulent stream

14.4 Blade vortex interactions in compressible flow

15: Trailing edge and roughness noise

Abstract

15.1 The origin and scaling of trailing edge noise

15.2 Amiet's trailing edge noise theory

15.3 The method of Brooks, Pope, and Marcolini [8]

15.4 Roughness noise

Part 4: Rotating blades and duct acoustics

16: Open rotor noise

Abstract

16.1 Tone and broadband noise

16.2 Time domain prediction methods for tone noise

16.3 Frequency domain prediction methods for tone noise

16.4 Broadband noise from open rotors

16.5 Haystacking of broadband noise

16.6 Blade vortex interactions

17: Duct acoustics

Abstract

17.1 Introduction

17.2 The sound in a cylindrical duct

17.3 Duct liners

17.4 The Green's function for a source in a cylindrical duct

17.5 Sound power in ducts

17.6 Nonuniform mean flow

17.7 The radiation from duct inlets and exits

18: Fan noise

Abstract

18.1 Sources of sound in ducted fans

18.2 Duct mode amplitudes

18.3 The cascade blade response function

18.4 The rectilinear model of a rotor or stator in a cylindrical duct

18.5 Wake evolution in swirling flows

18.6 Fan tone noise

18.7 Broadband fan noise

Appendix A: Nomenclature

A.1 Symbol conventions, symbol modifiers, and Fourier transforms

A.2 Symbols used

Appendix B: Branch cuts

Appendix C: The cascade blade response function

Index

Copyright

Academic Press is an imprint of Elsevier

125 London Wall, London EC2Y 5AS, United Kingdom

525 B Street, Suite 1800, San Diego, CA 92101-4495, United States

50 Hampshire Street, 5th Floor, Cambridge, MA 02139, United States

The Boulevard, Langford Lane, Kidlington, Oxford OX5 1GB, United Kingdom

Copyright © 2017 Elsevier Inc. All rights reserved.

No part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage and retrieval system, without permission in writing from the publisher. Details on how to seek permission, further information about the Publisher's permissions policies and our arrangements with organizations such as the Copyright Clearance Center and the Copyright Licensing Agency, can be found at our website: www.elsevier.com/permissions.

This book and the individual contributions contained in it are protected under copyright by the Publisher (other than as may be noted herein).

Notices

Knowledge and best practice in this field are constantly changing. As new research and experience broaden our understanding, changes in research methods, professional practices, or medical treatment may become necessary.

Practitioners and researchers must always rely on their own experience and knowledge in evaluating and using any information, methods, compounds, or experiments described herein. In using such information or methods they should be mindful of their own safety and the safety of others, including parties for whom they have a professional responsibility.

To the fullest extent of the law, neither the Publisher nor the authors, contributors, or editors, assume any liability for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions, or ideas contained in the material herein.

Library of Congress Cataloging-in-Publication Data

A catalog record for this book is available from the Library of Congress

British Library Cataloguing-in-Publication Data

A catalogue record for this book is available from the British Library

ISBN 978-0-12-809651-2

For information on all Academic Press publications visit our website at https://www.elsevier.com/books-and-journals

Publisher: Joe Hayton

Acquisition Editor: Brian Guerin

Editorial Project Manager: Edward Payne

Production Project Manager: Anusha Sambamoorthy

Cover Designer: Greg Harris

Front cover image credit: Photo courtesy of Virginia Tech/John McCormick.

Typeset by SPi Global, India

Dedication

To our wives, Lisa and Anne

Preface

Stewart Glegg, Florida Atlantic University, Boca Raton, FL, United States

William Devenport, Virginia Tech, Blacksburg, VA, United States

As its name suggests, aeroacoustics is a subject that connects the fields of aerodynamics and acoustics. It is relatively young for a mechanical science and of an age where many of us were taught by or have worked with the pioneers in this field. Part of the appeal of aeroacoustics is in the creativity and generous spirit of cooperation unleashed by this cross-disciplinary collaboration of experts, where extraordinary accomplishments have been made possible by fusion of fields and expertise. Aeroacoustics is a field where applied mathematics, science, and engineering play a pivotal role; a field where experiments in their classical role of testing hypotheses are very much part of current progress. Consider, for example, the sound radiated by an airfoil in turbulence. The science of turbulence is hard enough, and putting an airfoil into it feels likely to be an unmanageable complication. At low Mach number the sound radiated by the interaction is perhaps one millionth part of the pressure field generated. It is hard to imagine a problem that appears more badly posed. Remarkably, it has been rendered not only tractable but, with some simplification, solvable by entirely analytical methods with results that pass near, if not through, experimental measurements. It is very satisfying to be working in a field that regularly produces extraordinary achievements of this type. A number of these achievements are detailed in this book.

Stewart Glegg obtained his BSc degree in engineering science from Southampton University and went on to also obtain his Masters and PhD at Southampton studying under the direction of Mike Fisher at the ISVR. At Southampton he was fortunate to be introduced to the fields of fluid dynamics and acoustics by some of the pioneers in aeroacoustics, notably Geoff Lilley, Chris Morfey, Peter Davies, Phil Doak, and of course Mike. After graduating with his PhD he worked at Westland Helicopters under the direction of Dave Hawkings who introduced him to the field of rotor noise. In 1979 he returned to the ISVR as a faculty member, and in 1985 he moved to Florida Atlantic University in search of a warmer climate and new challenges. It was not long before he was conscripted by Feri Farassat to work on helicopter rotor noise under NASA sponsorship, and since that time he has been actively involved in aeroacoustics in the United States.

William Devenport also graduated in engineering science in the United Kingdom, at the University of Exeter, and went on to do a PhD at the University of Cambridge. His PhD research was in the experimental and computational study of turbulent separated flows under the capable guidance of Peter Sutton. While it was not his focus then, he was fortunate enough to be able to attend lectures on aeroacoustics given by Shôn Ffowcs Williams, to interact with Ann Dowling, and to share an office with some of their graduate students. His experimental work continued when he joined Virginia Tech as a postdoc, under the guidance of Roger Simpson. His interest in aeroacoustics truly began just after he was appointed as assistant professor in 1989. Ever supportive, Roger Simpson introduced him to Tom Brooks at NASA Langley. Tom had already been sponsoring Stewart to do predictions of broadband noise generated by blade wake interactions in helicopter rotors, and had decided that some experiments were needed to define the turbulence structure. Stewart visited Virginia Tech in the Spring of 1990 initiating a collaboration that continues today. Perhaps some of the longevity of our collaboration and interest derives from our different technical backgrounds and areas of expertise.

This book began its life on paper in 2005 when the authors jointly organized a course on aeroacoustics and hydroacoustics taught from Florida Atlantic University to graduate students there and at Virginia Tech. The initial core of the text (written by S.G.) was drafted as notes for that course with the intent of bringing together the fundamentals of aeroacoustics in a form and at a level that would be appropriate to graduate students, and that would give them the background needed to understand most modern papers and developments in the field. Over 11 years later this remains the goal of this book.

We have many to thank for their help in making this book possible. In terms of those who have enabled the long-term study of aeroacoustics of which this book is a part, we first thank the Department of Ocean and Mechanical Engineering at Florida Atlantic University and the Kevin T. Crofton Department of Aerospace and Ocean Engineering at Virginia Tech for providing us with secure and supportive research and teaching environments, over many years. Without their support it would not have been possible to develop the material for this text or to write it up. We also acknowledge the invaluable support and encouragement of our colleagues at these institutions including the members of the Center for Acoustics and Vibration at FAU, and Eric Paterson, Aurélien Borgoltz, and the members of ATFRG and CREATe at VT. We owe a debt of gratitude to the many graduate students we have had the privilege to work with. Some (mentioned below) directly contributed to this book, but all have contributed indirectly through the hard work and inspiration they dedicated to the mutual advance of understanding that underpins the relationship between student and advisor. Our research sponsors have in many ways enabled this book, and we are grateful to all. Among them are Tom Brooks, who along with his colleagues at NASA Mike Marcolini and Casey Burley, found a way to sponsor us continuously for the first 10 years of our collaboration; sponsorship that was fundamental in initiating, building, and linking our research programs. We have been fortunate enough also to receive long-term support from the Office of Naval Research, in particular through the programs in hydroacoustics and turbulence and wakes managed by Ki-Han Kim and Ron Joslin. We are very grateful not only for the sustained funding, but also for the motivation they have provided to move into new and exciting technical areas, and to take on challenges outside our comfort zone. We have also benefited greatly from collaborations and support associated with the Virginia Tech Stability Tunnel and its evolution as an aeroacoustic facility. In particular, we owe debts here to Ricardo Burdisso who played a central role in its transformation and who, along with his company AVEC Inc., have been inspirational partners in its further development and advancement. We also thank Wing Ng, and his company Techsburg Inc., without whom this upgrade would never have been possible. We owe a debt of gratitude to ONR, and particularly Ron Joslin, for his courage to be the first to support this venture when it must have appeared both unconventional and risky, and to ONR and General Electric for their encouragement and support in maturing many of its state of the art capabilities.

In terms of those whose efforts have directly contributed to this book we would first like to thank those who encouraged or inspired the book, in particular, Phil Joseph, Chris Morfey, Bill Blake, and Nigel Peake. We are particularly grateful to Chris who also carefully reviewed and commented on a number of chapters, and whose comments often challenged us to maintain the technical rigor of the book. A number of people provided material for or read and commented on specific sections or components of the book and we thank them all. They include Nathan Alexander, Jason Anderson, Manuj Awasthi, Neehar Balantrapu, Andreas Bergmann, Ken Brown, Ricardo Burdisso, Lou Cattafesta, Ian Clark, Dan Cadel, Alexandra Devenport, Mitchell Devenport, Mike Doty, Marty Gerold, Christopher Hickling, Florence Hutcheson, Remy Johnson, Liselle Joseph, Phil Joseph, Emilia Kawashima, Jon Larssen, Todd Lowe, Lin Ma, Henry Murray, Mike Marcolini, Patricio Ravetta, Michel Roger, David Stephens, Ian Smith, and Hiroki Ura.

Last but not least, above all we thank our spouses Inger Hansen (known to family and friends as Lisa) and Anne Devenport for their constant support of us in our careers, in this endeavor, and for their dedicated efforts in reviewing and correcting large fractions of the book.

Part 1

Fundamentals

1

Introduction

Abstract

In this chapter the subject of aeroacoustics is introduced and placed in the context of Lighthill's acoustics analogy. The important measures of subjective acoustics are discussed so that the reader can relate the sound levels discussed in the rest of the text to the levels of annoyance used to assess sound levels in noise control applications. A section on notation used in the book is also included in the chapter.

Keywords

Aeroacoustics; Lighthill; Sound waves; Turbulence; Sound levels; Annoyance; Symbol conventions

1.1 Aeroacoustics of low Mach number flows

Sound is a fundamental part of human life. We use it for social interaction, for art, for communication and education, and for understanding each other. Sound is also a byproduct of human activities. Unwanted sound, or noise, can adversely affect our productivity, health, security, and quality of life. Flow noise generated by fans, vehicles, wind turbines, and propulsion systems are major contributors to this unwanted sound. Aeroacoustics is the study of noise generation by air flows, and the way in which aerodynamic systems can be designed to minimize noise.

Much of our motivation for understanding flow generated noise originates from the aircraft industry. Regulations require noise certification for all new aircraft and so new products must meet increasingly high standards for noise emissions. Historically, most key developments can be traced to the advent of the jet engine at the end of World War II and its role in ushering in a new era in commercial air transportation. The original engines were unacceptably loud and it was clear that for the jet engine to be viable the noise had to be reduced. At that time there was no understanding of how sound waves could be generated by a turbulent flow, although theories did exist for sound radiation from vibrating surfaces, and even propellers. In 1952 Sir James Lighthill [1] published his theory of aerodynamic sound and the subject of aeroacoustics was born. This theory, which is known as Lighthill's Acoustic Analogy, provides the basis for our understanding of sound generation by flow. It is an exact re-arrangement of the equations of fluid motion, but has certain limitations, which must be carefully understood if it is to be applied correctly. Lighthill's theory has been frequently challenged, but remains the most important and effective analytical tool for the understanding and reduction of flow noise.

The purpose of this book is to provide an introduction to the basic concepts that describe the sound radiation from low Mach number flows, in particular flows over moving surfaces, which are the most widespread cause of flow noise in engineering systems. This includes fan noise, rotor noise, wind turbine noise, boundary layer noise, airframe noise, and aircraft noise, with the exception of jet and shock associated noise. The basic principles are also applicable to hydroacoustics, the study of flow generated noise in underwater applications. The primary difference between aero and hydroacoustics is that water is an almost incompressible fluid and thus usually involves Mach numbers that are an order of magnitude smaller than in aerodynamic applications.

This book is intended to be an introductory text on low Mach number aeroacoustics for graduate students who do not have a background in acoustics but who are familiar with the mathematical foundations of fluid dynamics and thermodynamics. It is written in four parts. Part 1 introduces the fundamentals, including the basic equations of unsteady fluid flow, a review of fluid dynamics concepts, and an introduction to linear acoustics. Lighthill's acoustic analogy is then introduced, leading to Curle's theorem, the Ffowcs Williams and Hawkings equation, the linearized Euler equations, vortex sound, and a discussion of turbulence and turbulent flows. This is followed by Part 2 that details experimental measurement techniques, including the use of microphones and flow measurement devices, signal processing and phased arrays, and wind tunnel testing methods. Part 3 of the text deals with edge and boundary layer noise including leading edge noise, trailing edge noise, and the flow over rough surfaces. Finally, Part 4 considers rotating sources such as fans, propellers, and rotors. It includes chapters on open rotor noise, duct acoustics, and fan noise.

1.2 Sound waves and turbulence

Sound waves cause small perturbations in pressure that propagate through a fluid medium. The speed of propagation, co, depends on the local properties of the fluid, but is typically about 343 m/s in air and 1500 m/s in water. The simplest form of a sound wave is a harmonic wave that has a sinusoidal variation in space with a wavelength λ and a frequency f=co/λ, Fig. 1.1. The human ear responds to frequencies between 20 Hz and 20 kHz, so typical wavelengths of interest are from 17 m to 17 mm in air.

Fig. 1.1 A turbulent flow incident on a structure that radiates sound waves.

In contrast, turbulence is caused by instabilities in viscous shear flows breaking down into random chaotic motion. Common examples are boundary layers on wings and aircraft fuselages, and the wakes behind moving vehicles. At higher speeds (high Reynolds number) the loss of energy due to turbulent mixing far exceeds the loss that occurs directly from viscous action. Turbulence is sometimes conceptualized as if it consisted of eddies convected at constant velocity through the medium without evolving in the convected frame of reference. This simplification is known as Taylor's hypothesis and the average convection speed Uc is typically taken to be about 60–80% of the free stream velocity. The size of the eddies L is usually the same order of magnitude as the smallest dimension of the mean flow, for example the boundary layer, shear layer, or wake thickness.

When convected turbulence encounters a solid body, it generates rapid changes of pressure on the surface of the body that radiate as sound waves through the fluid. The frequencies of the fluctuations which result from this type of interaction are determined by the eddy size and its convection velocity, and can be estimated as f≈Uc/L. The sound waves generated at this frequency will therefore have a wavelength λ≈Lco/Uc. In low Mach number applications, the convection speed is always much less than the speed of sound (Uc≪co), so the acoustic wavelength is always much larger than the size of the eddies generating the sound. The disparity in scales between the turbulent eddy dimensions and the acoustic wavelength is one of the most important features of aeroacoustics and hydroacoustics and one of the reasons that the subject is so challenging.

1.3 Quantifying sound levels and annoyance

The ear responds to the pressure fluctuations of sound waves to cause the sensation of hearing. When carrying out calculations of noise levels and designing for quiet systems it is always important to keep in mind the end goal. This is most often the level of annoyance experienced by a human listener.

The human ear has a remarkable dynamic range and can hear sound waves with amplitudes as low as 20 μPa and as high as 200 Pa before encountering the threshold of pain. The ear's sensitivity is logarithmic and so sound is measured using a decibel scale, referred to as the sound pressure level (SPL). This is given in terms of the root mean square of the fluctuating pressure time history prms and a reference pressure pref as

   (1.3.1)

where the units are stated as dB(re pref). It is important to specify the reference sound pressure pref when quoting a result in decibels because different units are used in different applications. For almost all airborne applications the standard is pref=20 μPa, but results in underwater applications are not as well standardized. The most common reference pressure used in underwater acoustics is pref=1 μPa, but historical data is also given in other units such as pref=1 μbar.

The root mean square pressure prms is the time average of the square of the fluctuating pressure. At any instant the pressure at a point is given as the sum of the mean background pressure po and a time varying perturbation p′(t). If p(t. The root mean square or "rms" pressure is then defined as

   (1.3.2)

where the averaging time must be large enough to include many cycles of the lowest frequency contained in the signal.

The human ear also discriminates the frequency of sound in a logarithmic manner. The frequency content of noise is therefore often characterized by summing up the sound into one-third octave bands that appear as equal intervals on a logarithmic scale. In acoustics, an octave is a doubling of the frequency and thus a third-octave band integrates the sound over a band for which the upper limit is 2¹/³ times its lower limit. The mid-band frequencies in Hertz are given by the relation

   (1.3.3)

where n is known as the band number [2]. Thus the band extends from fn×2−1/6 to fn×2¹/⁶.

The typical response of the ear at different frequencies is plotted in Fig. 1.2 and we see that low and high frequencies are significantly less important than the mid-frequency range around 1 kHz. Note especially how the ear does not respond well to frequencies below 100 Hz, although sometimes low-frequency sounds are identified as being most irritating. To obtain a measure of the perceived loudness of a sound the most commonly used metric is the dB(A) level. This applies a weighting to each one-third octave band level and then each band level is summed on an energy basis. Since the dB(A) level corrects for the sensitivity of the ear, and is easy to measure directly using a sound level meter, it is used extensively in noise control applications. It has also been found to be a good measure of annoyance and so most noise ordinances are defined using this unit with corrections for duration, pure tones, and day/night levels. Other measures of annoyance such as perceived noise level (PNL) and effective perceived noise level (EPNL) are used in assessing aircraft noise.

Fig. 1.2 The dB(A) weighting scale which represents the typical sensitivity of the human ear.

It is important to appreciate that simple physical scaling of sound levels from model scale tests to full scale applications must also incorporate a correction that accounts for annoyance. For example, slowing the rpm of a propeller will not only reduce its source level, but will generate lower frequencies which receive a larger negative A-weighting correction. The dB(A) level will therefore scale quite differently from the SPL, and significant additional noise benefits can be achieved by changing the frequency content of a sound.

In this text we will not attempt to rate noise source levels in subjective units such as dB(A), PNL, or EPNL, but rather the focus will be on the fundamental physics behind the sound generation process. However, it is important to keep in mind that based on the sensitivity of the ear (Fig. 1.2), the frequencies of most concern to humans usually lie between about 350 and 10,000 Hz. This roughly corresponds to wavelengths between 1 m and 3 cm in air.

1.4 Symbol and analysis conventions used in this book

While there is no universally agreed upon nomenclature for aeroacoustics, we use symbol conventions that will be familiar to many aeroacousticians. In this way analysis methods or results described in this book will be, as far as possible, recognizable when seen elsewhere in the aeroacoustics literature. We have also strived to be as consistent as possible and avoid using the same symbol for quantities that may be confused. Recognizing that these desires often conflict with that of familiarity, we have included in Appendix A both a list of symbols for basic quantities that appear repeatedly throughout the book, as well as a chapter-by-chapter nomenclature for more specialized symbols.

Our conventions for symbol modifiers are also listed in Appendix A and the most important ones are summarized here. Vector quantities are represented using bold symbols, such as x for position, or using subscript notation to refer to their Cartesian components, e.g., x1, x2, and x3. Except in special cases, the mean part of a variable will be denoted by subscript "o" or an overbar, the latter being used to indicate an averaging or expected value operation. The fluctuating part (such as p′(tdenotes the complex amplitude of a signal with harmonic time dependence.

Fourier transforms are perhaps the most important mathematical tool of aeroacoustic analysis. Except where otherwise noted we define the Fourier transform of a time history as

   (1.4.1)

where T tends to infinity, and the inverse Fourier transform as

   (1.4.2)

where ω is angular frequency and we are using the symbol i to represent the square root of −1. We define the one-dimensional Fourier transform of a variation over distance as

   (1.4.3)

where R∞ tends to infinity, and the inverse transform as

   (1.4.4)

with two- and three-dimensional forms that are the result of repeated application of Eqs. (1.4.3) or (1.4.4). Here k1 is the wavenumber. Note that in the forward time transform the exponent is positive, and the transformed variable is identified with a tilde, whereas it is negative in the forward spatial transform, and the transformed variable is identified with a double tilde.

It is important to realize that the above Fourier transform definitions are not universal to aeroacoustics analysis since there is no common standard. Our definitions are consistent with the definitions used in the previous aeroacoustics texts by Goldstein [3], Howe [4], and Blake [5]. Other texts on aeroacoustics, such as Dowling and Ffowcs Williams [6], and books focused on related topics, such as signal analysis [7], use different conventions, and it is obviously important to be aware of these distinctions when applying or comparing results.

References

[1] Lighthill M.J. On sound generated aerodynamically. I. General theory. Proc. R. Soc. A Math. Phys. Sci. 1952;211:564–587.

[2] Acoustical Society of America. Specification for Octave-Band and Fractional-Octave-Band Analog and Digital Filters. ANSI S1.11-2004 In: Melville, NY: Acoustical Society of America; 2004.

[3] Goldstein M.E. Aeroacoustics. New York, NY: McGraw-Hill International Book Company; 1976.

[4] Howe M.S. Acoustics of Fluid–Structure Interactions. Cambridge: Cambridge University Press; 1998.

[5] Blake W.K. Mechanics of Flow Induced Sound and Vibration. Orlando, FL: Academic Press; 1986.

[6] Dowling A.P., Ffowcs Williams J.E. Sound and Sources of Sound. Chichester: Ellis Horwood; 1983.

[7] Bendat J.S., Piersol A.G. Random Data: Analysis and Measurement Procedures. fourth ed. New York, NY: Wiley; 2011.

2

The equations of fluid motion

Abstract

This chapter is a review of the basic equations and concepts of fluid dynamics. These also form the foundations of aeroacoustics. We will start by considering the equations of fluid motion, the thermodynamics of small perturbations, and the role of vorticity. We will then evaluate the propagation of energy in the fluid, including the energy associated with sound waves in a moving medium, sound power, and acoustic intensity. Finally, we will review some basic concepts of fluid dynamics and summarize some results and methods of ideal flow theory that are most relevant to low Mach number aeroacoustics.

Keywords

Equations of fluid motion; Momentum; Continuity; Speed of sound; Energy equation; Sound power; Ideal flow; Conformal mapping

This chapter is a review of the basic equations and concepts of fluid dynamics. These also form the foundations of aeroacoustics. We will start by considering the equations of fluid motion, the thermodynamics of small perturbations, and the role of vorticity. We will then evaluate the rate of change of energy in the fluid, including the energy associated with sound waves in a moving medium. Finally, we will review some basic concepts of fluid dynamics and summarize some results and methods of ideal flow theory that are most relevant to low Mach number aeroacoustics.

2.1 Tensor notation

Cartesian tensor notation is useful in aeroacoustics because it provides relatively simple expressions for tensor products. In this section we will give a brief overview of the notation to be used in the following sections.

We are typically concerned with position vectors such as x and y which describe the locations of observers and sources, and flow variables, such as the velocity vector v, which defines the velocity of a fluid particle at a fixed location. In Cartesian coordinates these vectors have three components and if we use tensor notation, each component of the vector is defined by a subscript, say i, which has the values 1, 2, or 3. Hence we define x=(x1, x2, x3) giving the three components of the position vector x. To simplify the notation, we replace x by xi where the subscript i=1, 2, 3 defines each component. Using this approach, the definition of a dot product between the vectors q and v is

(2.1.1)

In general, the summation signs in the above definition are found to be cumbersome and so we introduce the convention that whenever repeated indices occur in a product there is an implied summation over all the components. Hence we have

   (2.1.2)

For example, we can define the magnitude squared of a vector as

(2.1.3)

This notation is particularly useful in the definition of the gradient of a scalar ϕ

   (2.1.4)

where i, j, and k are unit vectors in the i=1, 2, and 3 directions. Similarly, the divergence of a velocity v is

   (2.1.5)

This approach is most valuable when dealing with tensors. For example, the product Sij=vivj is a tensor which represents a matrix with nine elements corresponding to i=1, 2, 3 and j=1, 2, 3.

   (2.1.6)

A common expression found in the equations of motion is the velocity gradient tensor ∂vi/∂xj, which expands as

   (2.1.7)

Care needs to be exercised when we consider the expression Sii since this has repeated indices and so, by the rules defined above

   (2.1.8)

Hence if we want to isolate only one of the diagonal terms of the tensor we write Sii (no summation implied).

Example

Consider the tensor defined by the Kronecker delta function δij (which is defined as zero when i≠j and one when i=j) and evaluate (a) pδij, (b) δkjδik, and (c) Sijδij.

Using the summation rule we obtain

(2.1.9)

Throughout this text the velocity vector is denoted as v or vi and is often considered as the sum of the mean, time invariant, velocity U and the velocity fluctuation about the mean u, so v=U+u. Coordinates are defined using the vectors x, y, or z with the coordinates x and y used to denote observer and source location, respectively, where relevant. A volume is denoted as V and a surface area by S.

The thermodynamic variables of pressure, density, temperature are given their usual symbols p, ρ, Te, and the internal energy, enthalpy, and entropy are expressed, per unit mass, using the variables e, h, and s, respectively. Stagnation enthalpy and specific total energy are H and eT, respectively. A subscript "o" is used to denote the mean, time invariant, values of the thermodynamic variables at a fixed location, whereas a prime is used to indicate the fluctuating part. The mean speed of sound is given by the symbol co, and if this is constant throughout the fluid the symbol c∞ is used.

2.2 The equation of continuity

The concept of conservation of mass requires that mass is neither created nor destroyed in any fluid element. In Fig. 2.1 we show a region of the fluid, which is of volume V, and is bounded by the surface S. We will refer to this as a control volume. The outward pointing normal to the surface is n(o) and the velocity of the fluid is v. If the volume is fixed in space, and the density of the fluid is written as a function of space and time t as ρ(x,t), then the mass of the fluid contained in V is

Fig. 2.1 A control volume of size V bounded by the surface S with fluid moving through the volume at velocity v .

   (2.2.1)

The flow transports fluid in and out of the control volume and the mass flux out of V, across the surface element dS, per unit time, is given by

   (2.2.2)

The rate at which the control volume loses mass must equal the net outward flux of mass, which is given by the integral of Eq. (2.2.2) over the bounding surface S. Hence, for a fixed stationary surface

(2.2.3)

We can simplify this equation by using the divergence theorem to turn the surface integral into a volume integral. The divergence theorem states that the integral of the divergence of a vector over a volume is related to the component of the vector normal to the surface enclosing the volume by

Using this relationship gives

   (2.2.4)

This result is independent of the volume V and so the integral can only be identically zero if the integrand is zero. Hence we obtain the continuity equation in differential form as

   (2.2.5)

In tensor notation this is given as

   (2.2.6)

This is one of the most important equations in fluid dynamics and defines how mass is conserved in a fixed volume. It shows that the rate of change of density with time added to the divergence of the mass flux is zero at a fixed point.

It is also important to consider how mass is conserved in a frame of reference that moves with a differentially small piece of the fluid material, defined as a fluid particle. To consider this, we introduce the substantial or material derivative Df/Dt, which defines the rate of change of the function f in a frame of reference that moves with the particles rather than in the fixed frame which was used above. To define the moving frame of reference, we specify the position of the fluid particles at time t1 by the vector ηi. Since the particles move with velocity vi their location at the time t>t1 will be

   (2.2.7)

The rate of change in the convected frame of reference is the rate of change of f(zi(t),t) when ηi is fixed. Hence

   (2.2.8)

Then by using Eq. (2.2.7) and evaluating the partial derivatives we find the substantial derivative for the location xi=zi(t) in fixed coordinates is

(2.2.9)

, we expand Eq. (2.2.5) as,

(2.2.10)

This shows that if the fluid density is constant in the frame of reference moving with the fluid particles, so Dρ/Dt. However acoustic waves are, by definition, compressible and so this approximation cannot be used in the analysis of sound.

In the following sections we will consider thermodynamic quantities such as entropy or enthalpy, which are defined per unit mass rather than per unit volume. We will therefore need to consider the volume per unit mass, or specific volume, equivalent to the inverse of the density. We then have

   (2.2.11)

and so Eq. (2.2.10) gives

   (2.2.12)

This is an alternative form of the continuity equation, which is useful in applications that involve compressible flow.

2.3 The momentum equation

2.3.1 General considerations

To determine the momentum balance in a fluid we recall that the time rate of change of momentum of a body of fluid is equal to the net force exerted on it. We now apply this principle to the control volume illustrated in Fig. 2.1. The net momentum of the fluid in the control volume is given by

   (2.3.1)

According to the conservation of momentum, the rate of change of this quantity equals the force F applied to the control volume less the net rate at which momentum leaves the volume due to the movement of fluid across its surface. We saw above that the mass flow rate of fluid across a single element of the control surface dS . The rate at which momentum is lost due to this motion is therefore

   (2.3.2)

Hence the rate of change of momentum in the fixed stationary control volume is

(2.3.3)

The forces which are applied to the control volume are of three different types: (i) body forces, such as gravity which are almost never important in aeroacoustic applications and so will be ignored, (ii) pressure forces which apply a net force −pn(o)dS to each surface element shown in Fig. 2.1, and (iii) viscous shear stresses that introduce a net shear force on the surface. Viscous forces are rarely important in sound waves but are often important to the flows that produce them. They are most conveniently expressed using a viscous stress tensor σij which gives the force per unit area in the j direction applied to a surface whose outward normal lies in the i direction. The stress tensor is symmetric so σij=σji and the indices are interchanged without consequence. We can define the viscous shear stress applied to the surface of the control volume as

   (2.3.4)

where nj(o) is the tensor notation for the outward pointing normal to the surface. It is convenient to combine the viscous stress and pressure force into a single tensor pij defined as

   (2.3.5)

This tensor is called the compressive stress tensor and is often used in aeroacoustics to replace the tensor τij=−pij which is more commonly used in texts on fluid dynamics.

These conventions allow the force on the fluid F in the control volume to be written as

   (2.3.6)

Combining this with Eq. (2.3.3) yields, in tensor notation,

(2.3.7)

and applying the divergence theorem then gives

(2.3.8)

As with the continuity equation this integral can only be zero if the integrand is zero so we obtain the momentum equation in the absence of body forces as

   (2.3.9)

This is the conservation form of the momentum equation, which shows that the rate of change of momentum of a fixed volume of fluid is balanced by the flux of momentum out of the volume and the stresses applied to its surface.

We can also write the momentum equation in a nonconservation form which relates the forces applied to a fluid particle to its acceleration Dvi/Dt. Newton's Second Law of motion then requires that the mass per unit volume ρ times the acceleration equals the force applied to the particle per unit volume, which from Eq. (2.3.6) and the divergence theorem is −∂pij/∂xj, so

   (2.3.10)

It is a relatively simple exercise to show that Eqs. (2.3.9), (2.3.10) are the same by expanding the derivatives in Eq. (2.3.10) and using the continuity equation (2.2.6). Eq. (2.3.10) is the well-known Navier Stokes equation.

2.3.2 Viscous stresses

Viscous stresses are caused by molecular diffusion across the boundary enclosing the control volume. If the molecular diffusion causes fluid molecules to move into a region of fluid with a different velocity, then momentum is transferred and a viscous stress exists. The viscous stress causes the velocity parallel to the boundary, say vs, to be sheared in the direction normal to the boundary, so nj(∂vs/∂xj)≠0 or the velocity normal to the boundary, say vn, to be sheared in the direction parallel to the boundary, so (∂vn/∂xj)≠0. Detailed consideration of the viscous shear stress for a compressible fluid is discussed in texts on fluid dynamics [1,2], and requires the definition of a coefficient of bulk viscosity in addition to the shear viscosity μ. The bulk viscosity is usually assumed to be zero (Stokes hypothesis) in which case the viscous stress term is

   (2.3.11)

where μ is the coefficient of viscosity (for a detailed derivation of this equation see Batchelor [1] or Kundu [2]).

The viscous stress term in the momentum equation can be simplified for an incompressible flow. From Eq. (2.3.10) we note that the contribution of the viscous stresses will depend on ∂σij/∂xj and that for an incompressible flow ∂vj/∂xj=0 so we obtain from Eq. (2.3.11)

   (2.3.12)

assuming constant viscosity.

, t#=Ut/L, ρ#=ρ/ρ∞, and p#=p/ρ∞U² where U, L, and ρ∞ are constant reference values with L representing the overall scale of the flow. Substituting these definitions into Eq. (2.3.10), expanded using Eqs. (2.3.5), (2.3.11), gives

Dividing throughout by ρ∞U²/L we obtain

We see that in normalized form the viscous term is divided by the ratio of the scale of the inertial forces ρ∞U²/L to the scale of the viscous forces μU/L². This ratio is defined as the Reynolds number Re. In high speed and/or large scale flows the Reynolds number is high and the viscous term small, indicating that the effects of viscosity can often be ignored.

2.4 Thermodynamic quantities

Acoustic waves are a result of compressible effects in the fluid that cause small perturbations in the local pressure. It is therefore important that we understand how pressure changes are associated with changes in density and temperature. We make the assumption that an acoustic wave is a thermodynamic process which does not involve any exchange of heat or dissipative processes. In this case, the pressure perturbation p′ about the mean pressure po is directly proportional to the density perturbation ρ′ about the mean density ρo with the constant of proportionality given by the isentropic bulk modulus (dp/dρ)s.

   (2.4.1)

We will show later that co is the speed at which sound waves propagate through the medium.

We cannot ignore the role of dissipative processes or heating on the generation of sound or on the flows that generate it. For this reason, we need to discuss the thermodynamic properties of gases in some detail and define the role of quantities such as the enthalpy and entropy, which along with pressure, density, internal energy, kinetic energy, and temperature, define the state of the gas.

The First Law of Thermodynamics requires that the energy of a system can only be changed by the addition of heat or by work done on the system. In this case our system is a fluid particle that (by definition) moves with the flow and has constant mass. The change of internal energy per unit mass of the particle de is given by the sum of the heat added per unit mass δq and the work done on the system per unit mass, δw, so

   (2.4.2)

Note here that the internal energy represents the state of the particle and is independent of how the energy got there, whereas the heat and the work represent path functions and are dependent on the process taking place.

First consider what happens when the molecules in a particle expand to fill a larger volume. When the particle expands from a volume V to a volume V+dV, the work done on the surrounding fluid is given by the force exerted on the surrounding medium times the distance it moves during the expansion. Writing this as a surface integral, the force on each surface element is pdS, and the distance it moves is Δx, giving the total work done by the particle as

where the surface S encloses the volume V. Since the particle is very small the pressure may be considered as constant over the surface and, since

the work done by the system on its surroundings is pdV. This represents a loss of energy to the system and so Eq. (2.4.2) becomes

   (2.4.3)

In this equation dυ represents a change in volume per unit mass and can be related to the density of the fluid in the particle using υ=1/ρ. This example considers a volumetric change which occurs at constant pressure. In many cases we need to consider changes in pressure which occur without a change in volume. In this case the particle increases its capacity to do work. We can define this capacity using a state variable called the enthalpy which is related to the internal energy as h=e+pυ. We note that dh=de+υdp+pdυ which allows the first law (2.4.3) to be written for a change at constant volume as

   (2.4.4)

If heating takes place at either constant volume or constant pressure, then the temperature Te of the system is increased. The sensitivity coefficients are defined as specific heats for constant volume and constant pressure as

   (2.4.5)

Hence if heating takes place at constant volume then the change in internal energy can be related to the heat input by Eq. (2.4.3). Specifically,

In contrast if heating takes place at constant pressure then we can use Eq. (2.4.4) to give the change in enthalpy as,

If e and h are only functions of temperature (the assumption of a perfect gas), then these expressions reduce to de=cvdTe and dh=cpdTe. Subtracting Eq. (2.4.3) from Eq. (2.4.4) and using these relationships give

or

which integrates to the perfect gas law p=ρRTe with R=(cp−cv).

For an ideal gas the specific heats are constant and allow us to relate the internal energy and the enthalpy to the temperature as

   (2.4.6)

To proceed further we introduce the concept of the entropy of a system, which describes its state of disorder. Entropy is a variable like temperature or pressure that gives the state of a gas. It