MODAL ACOUSTIC RADIATION CHARACTERISTICS OF A THICK ANNULAR DISK DISSERTATION Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy in the Graduate School of The Ohio State University By Hyeongill Lee, M.S., B.S. ***** The Ohio State University 2003 Dissertation Committee: Approved by Dr. Rajendra Singh, Adviser Dr. Donald R. Houser _____________________________ Dr. Ahmet Selamet Dr. Robert G. Parker Adviser Department of Mechanical Engineering ABSTRACT New analytical and semi-analytical solution procedures for modal sound radiation from a thick annular disk are proposed. Classically, thin annular or circular plate theory has been used to describe sound radiation from normal surfaces while ignoring contributions from the radial surfaces. But, in many practical cases, the disk thickness is often beyond the thin plate theory limit and consequently a thick plate structural and acoustic formulation must be employed, as illustrated in this study. Also, radiation from in-plane vibration must be considered along with that from out-of-plane vibration to properly estimate the total sound radiation. First, we consider purely modal radiations from a disk with free-free and fixed-free boundaries. A new analytical formulation, based on the thick plate theory, is proposed for radiation from out-of-plane flexural modes. Further, the far-field sound pressures from in-plane radial vibration modes are obtained by using two alternate analytical methods based on the Rayleigh integral technique and a cylindrical radiator model. Analytical predictions are confirmed with measured data (with free-free boundaries only) as well as computational results (with both sets of boundaries) from finite element and boundary element codes in terms of structural eigensolutions, accelerance, acoustic response function spectra, modal sound pressures in far-field and modal directivity patterns. Selected parametric studies investigate the effects of disk ii geometry and vibrating frequencies on the radiation properties. Second, vibro-acoustic response for a multi-modal case, given a multi-directional harmonic force, is formulated based on the modal expansion technique. The analytical method employs the structural eigensolutions (from an analytical or numerical method), measured damping ratios and new analytical modal radiation solutions. This method is confirmed by comparing predictions of acoustic frequency response function, sound power and radiation efficiency spectra with those obtained using purely computational methods. The effects of coupling between structural modes (gap between their natural frequencies) and circumferential separation between two force excitation locations are investigated. Finally, as an example, modal and multi-modal sound radiations from a simplified brake rotor are expressed in terms of the characteristics of a generic thick annular disk having identical geometric dimensions. Coupling between in-plane and out-of-plane vibration modes that is introduced by the hat structure and boundaries of rotor is also investigated. Accuracy of our semi-analytical method is confirmed by purely numerical analyses based on finite element and boundary element models. iii DEDICATION Dedicated to my parents for their love and support throughout my life iv ACKNOWLEDGMENTS I would first like to thank my advisor, Dr. Rajendra Singh for his continuous support, patience, and guidance throughout this work. I would also like to thank Dr. Donald Houser, Dr. Ahmet Selamet and Dr. Robert Parker for their careful examination on this dissertation and kind suggestions and comments throughout this study. I thank my fellow graduate students in Acoustics and Dynamics Laboratory for all their assistance during this research. I also thank all Korean students in Mechanical Engineering department at the Ohio State University for their advice and help. Finally, I would like to thank my family especially my wife, Ji-Young, for her continuous love and support. Without her support, this study would not have been possible. v VITA October 27, 1964 ............................... Born – Korea 1986 ................................................... B.S. Mechanical Engineering, Seoul National University, Seoul, Korea 1993 ................................................... M.S. Mechanical Engineering, The University of Toledo 1986 - 1998........................................ Hyundai Motor Company, 1999 - present .................................... Graduate Research Associate, The Ohio State University PUBLICATIONS Journal Publications 1. H. Lee and R. Singh, “Acoustic radiation from radial modes of a thick annular disk”, submitted to the Journal of Sound and Vibration, 2002. 2. H. Lee and R. Singh, “Acoustic radiation from out-of-plane modes of an annular disk based on thick plate theory”, submitted to the Journal of Sound and Vibration, 2002. Conference Presentations 1. H. Lee and R. Singh, “Vibro-acoustic analysis of a thick annular disk,” The 6th U.S. National Congress on Computational Mechanics, 1 - 3 August 2001 Dearborn, Michigan. vi 2. H. Lee and R. Singh, “New calculation strategy for acoustic radiation from a thick annular disk” The 143rd Meeting of the Acoustical Society of America, 3 - 7 June 2002 Pittsburgh, PA. Abstract published in J. Acoust. Soc. Am.,111(5-2), p. 2474, May 2002. 3. H. Lee and R. Singh, “Vibro-Acoustics of a Brake Rotor with Focus on Squeal Noise” The 2002 International Congress and Exposition on Noise Control Engineering, paper # N613, Aug. 19 – 21, 2002 Dearborn, Michigan. 4. H. Lee and R. Singh, “Sound Radiation from a Disk Brake Rotor Using a SemiAnalytical Method” The 2003 SAE N&V Conference, 03NVC-229, May 5-8, 2003 Traverse City, Michigan. FIELDS OF STUDY Major Fields: Mechanical Engineering Structural Dynamics and Vibration. Vibro-Acoustics and Noise Control. vii TABLE OF CONTENTS ABSTRACT ........................................................................................................................ii DEDICATION ...................................................................................................................iv ACKNOWLEDGMENTS................................................................................................... v VITA ................................................................................................................................vi LIST OF TABLES .............................................................................................................xi LIST OF FIGURES..........................................................................................................xiii LIST OF SYMBOLS ........................................................................................................xx Chapter .................................................................................................................................. 1 Introduction ............................................................................................................. 1 1.1 Motivation ......................................................................................................... 1 1.2 Literature Review .............................................................................................. 1 1.3 Problem Formulation......................................................................................... 5 1.4 Organization ...................................................................................................... 9 REFERENCES FOR CHAPTER 1....................................................................... 12 2 Acoustic radiation from out-of-plane modes using thin and thick plate theories .................................................................................................................. 16 2.1 Introduction ..................................................................................................... 16 2.2 Structural Analysis Based on Thick and Thin Plate Theories......................... 21 2.3 Acoustic Radiation Calculation....................................................................... 39 2.4. Computational and experimental Investigations of Sound Radiation............ 46 2.5 Effect of key Parameters on the Modal Sound Radiation ............................... 52 viii 2.6 Conclusion....................................................................................................... 61 REFERENCES FOR CHAPTER 2....................................................................... 62 3 Sound radiation in-plane vibration........................................................................ 64 3.1 Introduction ..................................................................................................... 64 3.2 Problem Formulation....................................................................................... 66 3.3 Structural Analysis .......................................................................................... 67 3.4 Acoustic Radiation Model............................................................................... 76 3.5 Method I: The Rayleigh Integral Approach .................................................... 78 3.6 Method II: Cylindrical Radiator ...................................................................... 81 3.7 Modal Radiation Results ................................................................................. 87 3.8 Parametric Studies........................................................................................... 96 3.9 Conclusion..................................................................................................... 103 REFERENCES TO CHAPTER 3 ....................................................................... 105 4 Multi-modal vibro-acoustic response.................................................................. 107 4.1 Introduction ................................................................................................... 107 4.2 Assumptions and Objectives ......................................................................... 109 4.3 Vibro-Acoustic Responses to a Harmonic Excitation................................... 109 4.4 Effects of Structural and Acoustic Modal Coupling on the Acoustic Radiation ....................................................................................................... 127 4.5 Conclusion..................................................................................................... 133 REFERENCES FOR CHAPTER 4..................................................................... 134 5 Application to a brake rotor ................................................................................ 135 5.1 Introduction ................................................................................................... 135 5.2 Objectives and Assumptions ......................................................................... 137 5.3 Structural Modal Analysis............................................................................. 137 5.4 Sound Radiation from Structural Modes of Brake Rotor.............................. 140 ix 5.5 Vibro-Acoustic Response to a Multi-Directional Harmonic Force .............. 146 5.6 Conclusion..................................................................................................... 151 REFERENCES FOR CHAPTER 5..................................................................... 152 6 Conclusion........................................................................................................... 154 6.1 Summary ....................................................................................................... 154 6.2 Contributions................................................................................................. 157 6.3 Future Research............................................................................................. 158 BIBLIOGRAPHY ........................................................................................................... 160 x LIST OF TABLES Table Page 1.1 Geometric dimensions and material properties of the example disks..................... 7 2.1 Disk examples with free-free or fixed (r = b) - free boundaries ........................... 18 2.2 List of models developed and methods employed for Disk I................................ 20 2.3 Eigensolutions for Disk I, II and III with free-free boundaries............................. 30 2.4 Measured modal damping ratios for Disk I........................................................... 31 2.5 Eigensolutions of Disk I with fixed-free boundary conditions ............................. 34 2.6 Comparison of directivity patterns for selected modes of Disk I.......................... 50 2.7 Comparison of modal acoustic power and radiation efficiency levels for selected out-of-plane modes of Disk I................................................................... 51 2.8 Modal acoustic powers and radiation efficiencies for first four out-ofplane modes with fixed-free or free-free boundaries. ........................................... 59 3.1 Disk dimensions and material properties .............................................................. 69 3.2 Comparison of disk eigen-solutions for radial modes........................................... 74 3.3 Modal displacement amplitudes used for acoustic analysis.................................. 74 3.4 Comparison of directivity patterns for selected modes ......................................... 92 xi 3.5 Comparison of modal acoustic power and radiation efficiency levels.................. 94 4.1 Self and mutual radiation terms of sound power between elastic modes of the sample disk .................................................................................................... 128 5.1 Geometric dimensions and material properties of the brake rotor ...................... 137 5.2 Selected structural modes of the brake rotor ....................................................... 139 5.3 Sound powers and radiation efficiencies for selected modes.............................. 145 xii LIST OF FIGURES Figure 1.1 Page Thick annular disk example cases a) with free – free boundaries b) with fixed – free boundaries c) disk with hat structure ................................................... 6 2.1 A thick annular disk with an excitation in the normal direction ........................... 17 2.2 Comparison of Disk I mode shapes given free – free boundaries. (a) (1, 0) mode; (b) (0, 2) mode. Key: plate theory (Model C); 2.3 WKLFNSODWHWKHRU\0RGHO$ WKLQ ILQLWHHOHPHQWPHWKRG0RGHO' ...................... 28 /f(ω), acoustic Vibro-acoustic experiment used to measure structural w frequency response functions P/f(ω) and out-of-plane modal sound radiations. .............................................................................................................. 29 2.4 /f(ω) at r = 151.5 mm and ϕ = 180° for Disk I Structural accelerances w with free-free boundaries. Key: - - -, computed using FEM (Model D); ––– –, measured (E). .................................................................................................... 32 2.5 Effect of geometry on the (0, 2) and (0, 3) modes of Disk I with free-free boundaries. (a) Effect of radii ratio (β); (b) Effect of thickness ratio ( h ). Key: WKLFN SODWH WKHRU\ 0RGHO $ WKLFN SODWH WKHRU\ ZLWKRXW rotary inertia effect (Model B); - - -, thin plate theory (Model C); measured (E). ........................................................................................................ 35 xiii 2.6 Effect of geometry on the (1, 0) and (1, 1) modes of Disk I with free-free boundaries. (a) Effect of radii ratio (β); (b) Effect of thickness ratio ( h ). Key: WKLFN SODWH WKHRU\ 0RGHO $ WKLFN SODWH WKHRU\ ZLWKRXW rotary inertia effect (Model B); - - -, thin plate theory (Model C); measured (E). ........................................................................................................ 36 2.7 Effect of geometry on the (0, 2) and (0, 3) modes of Disk I with fixed-free boundaries. (a) Effect of radii ratio (β); (b) Effect of thickness ratio ( h ). Key: WKLFN SODWH WKHRU\ 0RGHO ) WKLFN SODWH WKHRU\ ZLWKRXW rotary inertia effect (Model G); - - -, thin plate theory (Model H)........................ 37 2.8 Effect of geometry on the (1, 0) and (1, 1) modes of Disk I with fixed-free boundaries. (a) radii ratio (β); (b) thickness ratio ( h ). Key: mode with thick plate theory (Model F); ∆∆∆, (0, 0) mode with thick plate theory without rotary inertia effect (Model G); plate theory (Model H); PRGHZLWKWKLQ PRGHZLWKWKLFNSODWHWKHRU\0RGHO F); ◊◊◊, (0, 1) mode with thick plate theory without rotary inertia effect (Model G); - - -, (0, 1) mode with thin plate theory (Model H). ........................... 38 2.9 Sound Radiation from the out-of-plane vibration modes in the sherical coordinate system. (a) a thin disk with an infinite baffle; (b) a thick disk without baffle. ....................................................................................................... 41 2.10 Acoustic frequency response function P/f(ω) given unit impulsive force excitation f(t) in the z direction at r =151.5 mm. (a) θ = π/2 and φ = 0; (b) θ = 0 and φ = 0. Key: –––, analytical calculation (Model L); - - -, PHDVXUHG+HUHT LVWKHUDGLDO computed using BEM (Model M); mode. 2.11 48 Directivity pattern Dmn(θ) given φ = 0 and R = 303 mm. (a) m = 0, n = 2 DQDO\WLFDOPHWKRGEDVHGRQWKLFN mode; (b) m = 0, n = 3 mode. Key: plate theory (Model J); ∆∆∆, analytical method based on thin plate theory xiv (Model K); - - -, semi-analytical method (Model L); , computed using BEM (Model M); 2.12 PHDVXUHG ........................................................................ 49 Effect of radii ratio on the modal sound radiation based on alternate plate 2 theories. (a) Spatially averaged mean-square velocity < w > t , s ; (b) Radiation efficiency σ. Key: (Model J); PRGH ZLWK WKLFN SODWH WKHRU\ PRGHZLWKWKLQSODWHWKHRU\0RGHO. mode with thick plate theory (Model J); - - -, (0, 3) mode with thin plate theory (Model K)................................................................................................... 53 2.13 Effect of radii ratio on the modal sound radiation based on alternate plate 2 theories. (a) Spatially averaged mean-square velocity < w > t , s ; (b) Radiation efficiency σ. Key: (Model J); PRGH ZLWK WKLFN SODWH WKHRU\ PRGHZLWKWKLQSODWHWKHRU\0RGHO. mode with thick plate theory (Model J); - - -, (1, 1) mode with thin plate theory (Model K)................................................................................................... 54 2.14 Effect of thickness ratio on the modal sound radiation based on alternate 2 plate theories. (a) Spatially averaged mean-square velocity < w > t , s ; (b) Radiation efficiency σ. Key: (Model J); PRGH WKLFN ZLWK SODWH WKHory PRGHZLWKWKLQSODWHWKHRU\0RGHO. mode with thick plate theory (Model J); - - -, (0, 3) mode with thin plate theory (Model K)................................................................................................... 56 2.15 Effect of thickness ratio on the modal sound radiation based on alternate 2 plate theories. (a) Spatially averaged mean-square velocity < w > t , s ; (b) Radiation efficiency σ. Key: (Model J); PRGH ZLWK WKLFN SODWH WKHRU\ PRGHZLWKWKLQSODWe theory (Model K); mode with thick plate theory (Model J); - - -, (1, 1) mode with thin plate theory (Model K)................................................................................................... 57 xv 2.16 Modal directivity patterns of Disk I with alternate boundary conditions. (a) n = 2 modes; (b) n = 0 modes. Key: –––, fixed-free; - - -, free-free boundary condition................................................................................................ 60 3.1 A thick annular disk with a radial force ................................................................ 68 3.2 L(λ) for q = 2 and q = 3 modes. Key: –––, q = 2 mode; - - -, q = 3 mode. ........... 73 3.3 Comparison of radial mode shapes Key: solid line, analytical solution given by equation (3.13); discrete point, finite element analysis.......................... 73 3.4 Structural frequency response functions ü/f(ω) at j = 0. Key: ––––, measured; - - - -, computed using BEM. ............................................................... 75 3.5 Spherical sound radiation from a vibrating disk. .................................................. 77 3.6 Sound Radiation from the radial vibration of a thick annular disk in spherical coordinate system. ................................................................................. 79 3.7 Cylindrical radiator of length h using cylindrical coordinate system. .................. 82 3.8 Vibro-acoustic experiment used to measure structural u f (ω) and acoustic P/f(ω) frequency response functions and in-plane modal sound radiation. 88 3.9 Acoustic frequency response functions P/f(ω). (a) θ = π/2 and φ = 0; (b) θ = 0 and φ = 0. Key: –––, measured; - - -, computed using BEM. ......................... 90 3.10 Acoustic frequency response functions P u (ω) . (a) θ = π/2 and φ = 0; (b) θ = 0 and φ = 0. Key: –––, measured; - - -, computed using BEM. ...................... 91 3.11 Directivity pattern Dq(θ) given φ = 0 and R = 303 mm. (a) q = 2 mode; (b) q = 3 mode. Key: - - -, Rayleigh integral method; , cylindrical radiator method; – - –, computed using BEM; •, measured.................................. 93 xvi 3.12 Directivity pattern Dq(φ) given θ = π/2 and R = 303 mm. (a) q = 2 mode; (b) q = 3 mode. Key: –––, Analytical and numerical methods; • • •, measured................................................................................................................ 95 3.13 Effect of radii ratio on the amplitude ratio and phase difference between PqI and PqO. (a) q = 2 mode; (b) q = 3 mode. Key: ο ο ο, Phase difference; ∆ ∆ ∆, Amplitude Ratio......................................................................................... 98 3.14 Effect of radii ratio on Πq and σq. (a) q = 2 mode; (b) q = 3 mode. Key: , Acoustic power; ο ο ο, Radiation efficiency. ............................................. 99 3.15 Effect of ω on Πq and σq. (a) q = 2 mode; (b) q = 3 mode. Key: , Acoustic power; ο ο ο, Radiation efficiency. ..................................................... 101 3.16 Effect of thickness ratio on Πq and σq. (a) q = 2 mode; (b) q = 3 mode. Key: , Acoustic power; ο ο ο, Radiation efficiency.................................... 102 4.1 Sound radiation from a vibrating thick annular disk........................................... 110 4.2 Vibro-acoustic experimental setup used to measure structural frequency /f(ω) or u /f(ω) and acoustic P/f(ω) frequency response functions w response functions. .............................................................................................. 113 4.3 Structural frequency response functions with free-free boundaries /f(ω) at r = 0.1515 and ϕ = 180°; (b) ü/f(ω) at ϕ = 0. Key: , (a) w measured; - - -, computed using FEM; –––, analytical calculation..................... 115 4.4 Acoustic frequency response functions P/f(ω) due to radial excitation. (a) θ = π/2 and φ = 0; (b) θ = 0 and φ = 0 Key: , measured; - - -, computed using BEM; –––, analytical calculation. ............................................................. 117 4.5 Acoustic frequency response function P/f(ω) given impulsive force excitation f(t) at r = a in the z direction. (a) θ = π/2 and φ = 0; (b) θ = 0 xvii and φ = 0. Key: , measured; - - -, computed using BEM; –––, analytical calculation........................................................................................... 118 4.6 Far-field sound pressure spectra P(ω) due to multi-modal excitation. (a) θ = π/4 and φ = 0; (b) θ = 0 and φ = 0. Key: ––––, analytical calculation; - - , computed using BEM........................................................................................ 120 4.7 Acoustic radiation functions due to combined harmonic excitation. (a) acoustic power spectra Π(ω); (b) radiation efficiency spectra σ(ω). Key: – ––, analytical calculation; - - -, computed using BEM........................................ 121 4.8 Example cases for calculation of vibro-acoustic responses due to multipoint excitations .................................................................................................. 124 4.9 Far-filed sound pressure spectra due to two identical harmonic forces with specific circumferential distances. (a) P(ω) at R = 303 mm, φ = 0, θ = π/4; (b) P(ω) at R = 303 mm, φ = 0, θ = π/2. Key: - - -, ∆ϕ = π/12; , ∆ϕ = π/6;. 4.10 , ∆ϕ = π/4; –––, ∆ϕ = π/3. ................................................................... 125 Acoustic radiation functions due to two identical harmonic forces with specific circumferential distances. (a) acoustic power spectra Π(ω); (b) radiation efficiency spectra σ(ω). Key: - - -, ∆ϕ = π/12; , ∆ϕ = π/6; , ∆ϕ = π/4; –––, ∆ϕ = π/3................................................................................... 126 4.11 Effect of natural frequency separation on P/f(ω). (a) θ = π/2 and φ = 0; (b) θ = 0 and φ = 0. Key: , modified; - - -, original case. .................................. 131 4.12 Effects of natural frequency separation on acoustic radiation functions. (a) acoustic power frequency response functions Π/f(ω); (b) radiation efficiency function σ(ω). Key: , modified; - - -, original case...................... 132 5.1 A thick annular disk with a hat structure simulates the brake rotor. Disk is clamped at the inner bolts and free at outer edge. ............................................... 136 xviii 5.2 Finite element model of the brake rotor with 2010 solid elements..................... 138 5.3 Directivity patterns for selected modes ............................................................... 143 5.4 Finite element model used for forced vibration analysis given a multidirectional harmonic force................................................................................... 148 5.5 Far-field sound pressure spectra p/f(ω) for selected receiver positions rp1 and rp2. Key: Analytical 5.6 &RPSXWHGXVLQJ%(0----.................................. 149 Acoustic power spectra Π(w) and radiation efficiency spectra σ(w) of the brake rotor. Key: Analytical &RPSXWHGXVLQJ%(0----. .......................... 150 xix LIST OF SYMBOLS LIST OF SYMBOLS FOR CHAPTER 2 a outer radius of annular disk (mm) b inner radius of annular disk (mm) B Hankel transform c0 speed of sound in the acoustic medium (m/s) Db flexural rigidity of disk (Nm) E Young’s modulus of disk (N/m2) f(t) dynamic force on disk (N) F amplitude of applied force (N) i −1 k acoustic wave number (rad/m) h disk thickness (mm) h disk thickness ratio (h/a) m number of nodal circles in the disk Mr, Mϕ, Mrϕ ending moment in the disk (Nm) n number of nodal diameters in the disk p far field sound pressure (Pa) P spatially-dependent far field sound pressure amplitude(Pa) Pmn far field sound pressure amplitude due to the (m, n)th out-of-plane mode (Pa) R radius of sphere at the far-field location (m) Qr, Qϕ shear forces in the disk (N) r, φ, z cylindrical coordinates R, θ, φ spherical coordinates xx rf position vector of the excitation force f(t) on the disk rp position vector of a receiver position sound pressure rs position vector of a sound source position Ss surface of the sound source Sv boundary surface of the acoustic control volume [TTHICK] characteristic matrix for the thick plate theory [TTHIN] characteristic matrix for the thin plate theory V acoustic control volume w transverse velocity in the disk (m/s) w transverse acceleration in the disk (m/s2) W spatial dependent transverse displacement in the disk (m) W spatial dependent transverse velocity in the disk (m/s) β radii ratio of the annular disk γ angle between the surface normal vector and the vector from source position to receiver position (rad) ∆θ increment in the cone angle θ (rad) λmn dimensionless structural eigenvalue for the (m, n)th flexural mode ν Poisson’s ratio of disk Π acoustic power from the disk vibration (W) Πmn acoustic power from the modal vibrations of the disk (W) ρ0 mass density of the acoustic medium (kg/m3) ρd mass density of the disk (kg/m3) σmn sound radiation efficiency of normal modes of the disk ϕ azimuthal angle of the disk (rad) ∆φ increment in azimuthal angle φ (rad) ℜ radiation resistance of disk Φmn flexural mode shape of the disk Ψr, Ψϕ bending rotations of the disk (rad) ψr, ψϕ spatially-dependent bending rotations of the disk (rad) ω angular frequency (rad/s) ωmn natural frequency of the (m, n)th out-of-plane mode (kHz) xxi ζmn modal damping ratio of the (m, n)th out-of-plane mode (%) Subscripts d disk m, n out-of-plane mode indices 0 acoustic medium p observation point in a far-field location r radial direction of the disk s source (radiator) ϕ circumferential direction of the disk Abbreviations BEM boundary element method FEM finite element method LIST OF SYMBOLS FOR CHAPTER 3 a outer radius of annular disk (mm) b inner radius of annular disk (mm) c0 speed of sound in the acoustic medium (m/s) D extensional rigidity of disk (N/m) Db flexural rigidity of disk (Nm) E Young’s modulus of disk (N/m2) f(t) dynamic force on disk (N) F amplitude of applied force (N) g free space Green’s function (m-1) i −1 k acoustic wave number (rad/m) kz structural wave number in z direction (rad/m) xxii h disk thickness (mm) h disk thickness ratio (h/a) Hn Hankel function of order n m number of nodal circles in the disk n number of nodal diameters in the disk Nr normal force in the radial direction (N) Nϕ normal force in the tangential direction (N) Nrϕ, Nrϕ shear forces in the radial direction (N) N r , N rϕ normalized forces in the radial and shear direction p far field sound pressure (Pa) P spatially-dependent far field sound pressure (Pa) pq far field sound pressure amplitude due to qth radial mode (Pa) q radial mode index R radius of sphere at far-field location (mm) r, φ, z cylindrical coordinates R, θ, φ spherical coordinates rf position vector of force f(t) on the disk rp position vector of sound pressure rs position vector of source Ss surface of sound source Sv boundary surface of acoustic control volume [T(ξ)] transfer matrix u radial displacement (mm) uq radial displacement for the qth radial mode (mm) u radial velocity of disk (m/s) u radial acceleration of disk (m/s2) U spatially-dependent radial displacement of disk (m/s) U spatially-dependent radial velocity of disk (m/s) U spatially-dependent radial acceleration of disk (m/s2) [U(ξ)] utility matrix u amplitude of radial velocity (m/s) xxiii u amplitude of radial acceleration (m/s2) v circumferential displacement (mm) V acoustic control volume Z(z) variation of surface acceleration in z direction β radii ratio of annular disk γ angle between surface normal vector and vector from source point to receiver position ∆θ increment in cone angle θ (rad) λq dimensionless structural eigenvalue for the qth radial mode λ2mn dimensionless structural eigenvalue for the (m,n)th flexural mode ν Poisson’s ratio of disk ξ dimensionless radial coordinate for disk Πq acoustic power from the qth radial mode (W) ρ0 mass density of acoustic medium (kg/m3) ρd mass density of disk (kg/m3) σq sound radiation efficiency of the qth radial mode σ rr normal stress of disk in radial direction (Pa) σ ϕϕ normal stress of disk in tangential direction (Pa) σ rϕ shear stress of disk in tangential direction (Pa) ϕ azimuthal angle of disk (rad) ∆φ increment in azimuthal angle φ (rad) ℜ radiation resistance of disk Φq mode shape of the qth radial mode ω angular frequency (rad/s) ωq angular frequency (rad/s) ω natural frequency ratio of annular disk ζq modal damping ratio Subscripts d disk xxiv I inner radial edge o acoustic medium O outer radial edge p observation point in far field q radial mode index r radial direction of disk s source (radiator) ϕ circumferential direction of disk Superscripts 1 first kind 2 second kind ~ Fourier transform − complex valued Abbreviations BEM boundary element method FEM finite element method LIST OF SYMBOLS FOR CHAPTER 4 a outer radius of annular disk (mm) b inner radius of annular disk (mm) c0 speed of sound in the acoustic medium (m/s) fn(t), fr(t) dynamic force on disk in normal and radial directions (N) F amplitude of applied force (N) i −1 k acoustic wave number (rad/m) h disk thickness (mm) xxv m number of nodal circles in the disk n number of nodal diameters in the disk p far field sound pressure (Pa) P spatial dependent far field sound pressure (Pa) pq far field sound pressure amplitude due to qth radial mode (Pa) q radial mode index R radius of sphere at far-field location (mm) R, θ, φ spherical coordinates rf position vector of force f(t) on the disk rp position vector of sound pressure So surface of sound source Sv boundary surface of acoustic control volume u r radial velocity of disk (m/s) u r radial acceleration of disk (m/s2) U r spatially-dependent radial velocity of disk (m/s) U r spatially-dependent radial acceleration of disk (m/s2) V acoustic control volume w transverse displacement in the disk w transverse acceleration in the disk W spatial dependent transverse displacement in the disk Z(z) variation of radial surface acceleration in z direction η vector of structural modal participation factors Π acoustic power from the disk vibration (W) Πm,n,q acoustic power from the modal vibrations of the disk (W) ρ0 mass density of acoustic medium (kg/m3) σ sound radiation efficiency of the disk ϕ azimuthal angle of the disk (rad) ∆ϕ circumferential separation of two excitation on the disk (rad) Φm,n,q mode shape for mode (m, n, q) Φ vm, n, q velocity modal vector for mode (m, n, q) xxvi ω angular frequency (rad/s) ωm,n,q natural frequencies of the disk (Hz) ζm,n,q modal damping ratios Subscripts m, n, q combined mode indices 0 acoustic medium p observation point in far field r radial direction of disk s source (radiator) ϕ circumferential direction of disk Superscripts V velocity Abbreviations BEM boundary element method FEM finite element method LIST OF SYMBOLS FOR CHAPTER 5 a outer radius of annular disk b inner radius of annular disk c0 speed of sound in the acoustic medium E Young’s modulus of the sample rotor fn(t), fr(t) dynamic force on rotor F amplitude of applied force h disk thickness xxvii H i Hat height −1 I acoustic intensity at a field point Jn Bessel’s function of order n k acoustic wave number kmn structural wave number of the (m, n)th out-of-plane mode kq structural wave number in z direction l tangential mode index m number of nodal circle n number of modal diameters R, θ, φ spherical coordinates for receiver positions q radial mode index Sv boundary surface of acoustic control volume th thickness of hat structure uqI , uqO acceleration on inner and outer radial surfaces of the qth radial mode U spatially-dependent radial velocity U q spatially-dependent radial velocity of the qth radial mode V acoustic control volume w transverse displacement of rotor W spatial dependent transverse displacement Wmn mode spatial dependent transverse displacement of the (m, n)th out-of-plane β radii ratio of annular disk η structural modal participation factor vector Γ modal sound pressure of rotor Π acoustic power Πmn, Πq acoustic power from the single mode of the sample rotor Πmnq, acoustic power from the combined mode of the sample rotor ρ0 mass density of acoustic medium ρd mass density of the rotor σmn, σq sound radiation efficiency of the single mode of the sample rotor xxviii σmn,q sound radiation efficiency of the combined mode of the sample rotor ϕ azimuthal angle of rotor Φ velocity modal vector of rotor ωj natural frequencies of rotor ζj modal damping ratios of rotor Subscripts d disk j mode number I inner radial edge l tangential mode index m,n out-of-plane mode indices m,n,q combined mode indices 0 acoustic medium O outer radial edge p observation point in far field q radial mode index s source (radiator) ϕ circumferential direction of rotor Superscripts 1 first kind 2 second kind Abbreviations BEM boundary element method FEM finite element method xxix CHAPTER 1 INTRODUCTION 1.1 Motivation Annular and circular disk models have been widely used to describe many practical sound radiators such as gears, brake rotors, clutches, flywheels, railway wheels, circular saws, electrical machinery stators, and electro-acoustic transducers. Thickness of such components are often beyond the thin plate theory limit and one must examine both out-of-plane and in-plane vibrations to appropriately control sound radiation from these thick bodies. Traditionally, thin annular disk radiations with infinite baffles have been employed to describe the sound radiation from flexural modes. Consequently, the effect of coupling between in-plane and out-of-plane modes has not been included in the calculation of sound radiation. This research intends to study such unresolved issues and it proposes new analytical and semi-analytical approaches for sound radiation from thick annular disks. 1.2 Literature Review 1.2.1 Structural Dynamics of Annular Disks There is a substantial body of literature on the out-of-pane (flexural) vibration of thin and thick plates [1.1-1.12]. Leissa [1.1-1.4] has summarized natural frequencies and 1 modes of plates with various geometric configurations and boundary conditions. In particular, Vogel and Skinner [1.5] investigated natural frequencies of an uniform annular disk using the thin plate theory. Wang and Thevendran [1.6] used the Rayleigh-Ritz method, based on thin plate theory, to analyze annular plates. Mindlin [1.7] and Mindlin and Deresiewicz [1.8] proposed a sixth-order thick plate theory to describe flexural vibration of thick circular disks. McGee et al. [1.9] used the same theory to solve free vibrations of thick annular plates. Irie et al. [1.10] analyzed the vibration of annular Mindlin plates with nine combinations of inner and outer radial edge conditions. For the admissible functions of the displacement fields, they employed Bessel’s functions in the radial direction and trigonometric function the circumferential direction. Out-of-plane vibration of annular Mindlin plates of varying thickness has been studied using similar approach [1.11]. Liew et al. [1.12] investigated the vibration of circular and annular Mindlin plates of different boundary conditions with multiple internal ring supports. Conversely, published literature on the in-plane vibrations of annular or circular disks is relatively sparse [1.13-1.16]. For instance, Bhuta and Jones [1.13] studied coupled symmetric and torsional vibrations of a thin, rotating circular disk. Burdess et al. [1.14] generalized the analysis to consider asymmetric in-plane vibrations and the properties of forward and backward traveling waves. Chen and Jhu [1.15] examined the in-plane stability of a spinning annular disk and determined the effect of rotational speed on natural frequencies. Irie et al. [1.16] calculated natural frequencies of annular disks using the transfer matrix method. We will use this method for the structural analysis of thick annular disks. 2 1.2.2 Acoustic Radiation from Annular Disks Sound radiation from thin circular and annular disks has been examined by several investigators [1.17-1.21]. For instance, Thompson [1.17] computed self and mutual radiation impedances of a uniformly vibrating annular or circular piston by integrating of the far-field directivity function. Lee and Singh [1.18] proposed a polynomial approximation for modal acoustic power radiation from a thin annular disk but this method was restricted to only out-of-plane modes. Levine and Leppington [1.19] developed an analytical solution for active and reactive powers from a planar annular membrane given axisymmetric motions. Rdzanek and Engel [1.20] suggested asymptotic formulas for power from a thin annular disk with clamped edges. Finally, Wodtke and Lamancusa [1.21] investigated a circular plate using finite element analysis and then calculated the sound radiation via the Rayleigh integral formula. Sound radiation from in-plane modes of thick annular disks has not been investigated thus far. Conversely, many researches on the radiation from cylindrical radiators have been executed using various approaches [1.22-1.27]. Williams et al. [1.22] used semi-analytical method with finite series of eigenfunctions for boundary condition to calculate the acoustic radiation from a finite cylinder. Sandman [1.23] investigated sound radiation from finite cylindrical shells and found that cylindrical baffle has very little influence on sound radiation and concluded that the baffled cylindrical geometry may be assumed to be a reasonable approximation for this problem. Stepanishen [1.24] combined Green’s function and Fourier integral technique to develop integral expressions for the generalized radiation impedance and radiated power and applied it to an infinite cylinder. Junger and Feit developed expressions for far-field sound pressures for finite 3 and infinite cylindrical radiators given arbitrary surface velocity distribution [1.25]. Williams solved the same problem using a 2-dimensional Fourier transform [1.26]. Finally, Wang and Lai calculated the modal-averaged radiation efficiency of a finite length circular cylindrical shell [1.27]. Several studies on the multi-modal sound radiation have been executed so far. For instance, Kelti and Peng [1.28] investigated the effects of modal coupling on the acoustic power from simply supported or clamped unit width plate and concluded that contributions due to the modal coupling may be important for off-resonant excitation cases. Cunefare [1.29] developed a technique for deriving the optimal surface velocity of a beam that minimizes the radiation efficiency of the beam. The author [1.30] also developed analytical and computational tool to assess the contribution of individual modes and interaction effects to the total sound power from planar structures. Cunefare and Currey [1.31] investigated orthogonal acoustic modes of finite baffled beam using singular value decomposition of the radiation operator. Here, acoustic modes are particular velocity patterns on the radiator surface. The authors [1.32] obtained the radiation modes of baffled finite plates using the same approach. Several researchers applied the acoustic radiation mode concept to the active structural acoustic control or the estimation of radiated acoustic power [1.33-1.34]. Multi-modal sound radiation and modal coupling effects on the sound radiation from a thin annular disk have also been investigated using the modal expansion technique [1.18]. 1.2.3 Brake Rotor Dynamics and Squeal Noise Several structural dynamic models have been used to explain the brake squeal generation mechanism. Two most common approaches deal with either the self-excited 4 vibration generated by the “stick-slip” phenomenon that is related to the velocity dependency of friction coefficient [1.35-1.36] or the modal coupling phenomena that involves two system modes coupled together because of the friction interface [1.37-1.39]. Recently, non-linear transient analysis [1.40-1.41] and complex eigen-value problem [1.42-1.46] have been implemented using the finite element method. Dihua and Dongying [1.43] extracted component modal data from finite element models and calculated squeal propensity using a modal synthesis method. They investigated the contribution of each component mode and the sensitivity of each mode to the instability of the system mode. Hamabe et al. [1.44] calculated the participation of an unstable system mode for a drum brake assembly. Squeal was eliminated after they separated the two highest participating drum modes. Kung et al. [1.45] applied the modal participation factor concept to a disk brake and identified a source of a low frequency squeal. Dunlap et al. [1.47] investigated brake squeal using various approaches including numerical and experimental approaches and concluded that natural frequency separation between inplane and out-of-plane modes is critical to high frequency squeal. McDaniel et al. [1.48] investigated coupling between in-plane and out-of-plane modes and concluded that the coupling creates vibrational instability that is characterized by power flow through the transverse motion of the rotor. 1.3 Problem Formulation Almost all of the above mentioned studies have considered sound radiation only from either flexural vibration modes or rigid body piston motions as evident from the review of available literature as presented in Section 1.2. 5 (a) (b) Fixed boundary r r ϕ b z ϕ b a z a h h (c) disk hat r z b a th h H Figure 1.1: Thick annular disk example cases a) with free – free boundaries b) with fixed – free boundaries c) disk with hat structure 6 Outer radius (a) 151.5 mm Inner radius (b) 87.5 mm Radii ratio (β = b/a) 0.54 Thickness (h) 31.5 mm Hat height (H) 24 mm Density (ρd) 7905.9 Kg/m3 Young’s modulus (E) 218 GPa Poisson’s ratio (ν) 0.305 Table 1.1: Geometric dimensions and material properties of the example disks. In such studies, sound radiation from the in-plane modes of a disk has been assumed to be negligible compared to that from the out-of-plane modes. But, if the thickness of a disk is beyond the range of thin plate (shell) theory, in-plane vibration could generate sufficient sound given proper excitation. Therefore, the chief goal of this research is to develop analytical solutions for sound radiation from the normal modes of a thick annular disk. Both analytical and semi-analytical procedures are proposed to calculate structural velocities and radiated acoustic pressure due to uni-directional and arbitrary harmonic excitation forces. Figure 1.1 illustrates the example cases for this study. Also, geometric dimensions and material properties of these example disks are given in the Table 1.1. 7 The primary assumptions for this study are as follows: 1. Structural and acoustic systems are linear time-invariant systems. Non-linear effects arising from the coupling of motion variables are ignored. 2. Structural velocities in the normal direction (z) of flexural modes and in the radial direction (r) of radial modes vary sinusoidally in the ϕ direction. 3. Far field sound pressure due to radial mode is generated by the structural motions of the inner or outer radial surfaces exclusively. 4. Coupling between in-plane and out-of-plane modes is assumed to be negligible in the calculation of sound radiation from normal modes of the disk. But the same effect is assumed to be critical in the sound radiated in the combined excitation problem. Chief objectives of this study are as follows. 1. Develop analytical and semi-analytical solutions for sound radiation from modal vibrations and validate analytical solutions using computational and/or experimental vibro-acoustic methods. 2. Critically examine thick and thin plate theories and investigate the effect of rotary inertia and shear deformations on the structural eigen-solutions and acoustic sound radiation from flexural (out-of-plane) modes. Also, study the effects of the disk geometry and boundary conditions on sound radiation using the proposed analytical solutions. 8 3. Examine the effect of couplings between in-plane and out-of-plane modes as well as couplings within the same type of modes on the total sound radiation considering natural frequency separations. Develop a semi-analytical procedure for calculating sound radiation from the disk (including a simplified brake rotor) given multi-modal harmonic excitations. 1.4 Organization Each chapter of this dissertation is self contained with its own objectives, problem formulation, methodology, results and list of references. Given below is a brief discussion of each chapter. Chapter 2 Structural eigen-solutions for the out-of-plane modes of a thick annular disk with free-free boundaries are calculated using both thick and thin plate theories. New analytical formulation is then proposed for the sound radiation problem. In addition, the same problem is solved by a semianalytical procedure in which the disk surface velocity is numerically defined by a finite element model and sound radiation is then analytically obtained using a modified circular radiator model. Also, the effects of radii and thickness ratios on the structural dynamic and acoustic radiation characteristics are investigated using the analytical procedure. Finally, the effect of boundary conditions is briefly examined. Chapter 3 In-plane modal vibration of a thick annular disk is analytically investigated using the transfer matrix method and confirmed using the results of a finite element model. Sound radiation from structural modes of 9 this type is obtained using two different analytical approaches: (i) the Rayleigh integral formula, (ii) modified cylindrical radiators of finite length along with Fourier series and Sinc function approaches. The problem is also analyzed using a boundary element code. Relevant computational and analytical results are successfully compared with vibroacoustic measurements. Acoustically efficient radial structural modes may be determined based on modal radiation efficiency and acoustic power calculations. Strategies for minimization of sound radiation are also investigated by parametrically varying disk geometry and natural frequencies. Chapter 4 A new semi-analytical procedure for the calculation of sound radiation from a thick annular disk, when it is excited by a multi-modal and multidirectional harmonic force, is introduced. Vibro-acoustic responses for a specific force has been calculated by the modal expansion technique from the structural modal dataset from analytical or numerical methods, and the normal radiation solutions representing the far-field sound pressure due to structural modal vibrations. This method is confirmed by comparing predicted results for a single frequency excitation with numerical calculations. Based on this procedure, acoustic power and radiation efficiency spectra of the sample disk corresponding to a specific force location and direction are obtained. Now, excitation can be applied at multiple locations, with different frequencies. The effects of coupling between structural modes and natural frequency separation on the acoustic 10 radiation are also investigated through the proposed procedure. This study could lead to strategies that would minimize sound radiation as excited by arbitrary harmonic forces. Chapter 5 Modal and multi-modal sound radiation mechanisms from a brake rotor are studied using analytical and semi-analytical procedure proposed earlier in Chapters II – IV. In this study, structural eigen-solutions and modal sound radiation of a brake rotor are expressed in terms of the vibroacoustic characteristics of a generic thick annular disk having similar geometric dimensions. Accuracy of the proposed method is confirmed by purely numerical analyses based on finite element and boundary element models. Vibro-acoustic responses such as surface velocities and radiated sound pressures due to multi-modal excitations are calculated from synthesized structural modes and modal acoustic radiation of the rotor using the modal expansion technique. In addition, acoustic power and radiation efficiency spectra corresponding to a specific force excitation are obtained from the sound pressure data. Predictions are also confirmed by comparisons with analogous numerical analyses. Chapter 6 Contributions of this study are highlighted and several areas of future research are suggested. 11 REFERENCES FOR CHAPTER 1 1.1 A. W. LEISSA 1969 NASA SP-160 Vibration of Plates. 1.2 A. W. LEISSA 1987 The Shock and Vibration Digest 19(3), 10-24. Recent Research and plate vibration, 1981-1985. Part 1: Classical theory. 1.3 A. W. LEISSA 1987 The Shock and Vibration Digest 19(3), 10-24. Recent research and plate vibration, 1981-1985. Part 2: Complicating effects. 1.4 A. W. LEISSA 1993 Vibrations of Plates, New York: Acoustical Society of America. 1.5 S. M. Vogel and D. W. Skinner 1965 Journal of Applied Mechanics December, 926-931 Natural frequencies of transversely vibrating uniform annular disk. 1.6 C. M. WANG and V. THEVENDRAN 1993 Journal of Sound and Vibration 163(1), 137-149 Vibration analysis of annular plates with concentric support using a variant of Rayleigh-Ritz method. 1.7 R. D. MINDLIN 1951 ASME Journal of Applied Mechanics 18, 31-38 Influence of rotatory inertia and shear on the flexural motion of isotropic, elastic plate. 1.8 R. D. MINDLIN and H. DERESIEWICZ 1954 Journal of Applied Physics 25(10), 1329-1332 Thickness-shear and flexural vibration of a circular disk. 1.9 O. G. MCGEE, C. S. HUANG and A. W. LEISSA 1995 International Journal of Mechanical Science 37(5), 537-566 Comprehensive exact solutions for free vibrations of thick annular sectorial plates with simply supported radial edges. 1.10 T. IRIE, G. YAMADA and K. TAKAGI 1982 Transactions of the American Society of Mechanical Engineers, Journal of Applied Mechanics 49, 633-638 Natural frequencies of thick annular plates. 1.11 IRIE, G. YAMADA and S. AOMURA 1979 Journal of Sound and Vibration 66 (1), 187-197 Free vibration of a Mindlin annular plate of varying thickness. 1.12 K. M. LIEW, Y. XIANG, C. M. WANG AND S. KITIPORNCHAI 1993 Computer Methods in Applied Mechanics and Engineering 110, 301-315 Flexural vibration of shear deformable circular and annular plates on ring support. 1.13 G. BHUTA and J. P. JONES 1971 Journal of the Acoustical Society of America 35 (7), 982-989. Symmetric planar vibrations of a rotating disk. 1.14 J. S. BURDESS, T. WREN and J. N. FAWCETT 1987 Proceeding of the Institution of Mechanical Engineers 201, 37-44. Plane stress vibration in rotating discs. 12 1.15 S. CHEN and J. L. JHU 1996 Journal of Sound and Vibration 195 (4), 585-593. On the in-plane vibration and stability of a spinning annular disk. 1.16 T. IRIE, G. YAMADA and Y. MURAMOTO 1984 Journal of Sound and Vibration 97 (1), 171-175 Natural frequencies of in-plane vibration of annular plates. 1.17 W. THOMPSON, JR. 1971 Journal of Sound and Vibration 17 (2), 221-233. The computation of self- and mutual-radiation impedances for annular and elliptical pistons using Bouwkamp integral. 1.18 M. R. LEE and R. SINGH 1994 Journal of the Acoustical Society of America 95 (6) 3311-3323. Analytical formulations for annular disk sound radiation using structural modes. 1.19 H. LEVINE and F. G. LEPPINGTON 1988 Journal of Sound and Vibration 121 (5), 269-275. A note on the acoustic power output of a circular plate. 1.20 W. P. RDZANEK Jr. and Z. ENGEL 2000 Applied Acoustics 60 (5), 29-43. Asymptotic formula for the acoustic power output of a clamped annular plate. 1.21 H. W. WODTKE and J. S. LAMANCUSA 1998 Journal of Sound and Vibration 215 (5), 1145-1163. Sound power minimization of circular plates through damping layer placement. 1.22 W. WILLIAMS, N. G. PARKE, D. A. MORAN AND C. H. SHERMAN 1964 Journal of the Acoustical Society of America 36 (12) 2316-2322. Acoustic Radiation from a Finite Cylinder. 1.23 B. E. SANDMAN 1976 Journal of the Acoustical Society of America 60 (6), 12561264. Fluid loading influence coefficients for a finite cylindrical shell. 1.24 P. R. STEPANISHEN 1978 Journal of the Acoustical Society of America 63 (2) 328338. Radiation power and radiation loading of cylindrical surfaces with nonuniform velocity distribution. 1.25 M. C. JUNGER and D. FEIT 1985 Sound, Structures, and Their Interactions. New York: MIT Press. 1.26 E. G. WILLIAMS 1999 Fourier Acoustics. San Diego: Academic Press. 1.27 C. WANG and J. C. S LAI 2000 Journal of Sound and Vibration 232 (2), 431-447. The sound radiation efficiency of finite length acoustically thick circular cylindrical shell under mechanical excitation I: Theoretical analysis. 1.28 R. F. KELTIE and H. PENG 1987 ASME Trans. J. Vib. Acoust. Stress Reliabil. Des. 109, 48-53. The effect of modal coupling on the acoustic radiation from panels. 13 1.29 K. A. CUNEFARE 1991 Journal of the Acoustical Society of America 90(5), 25212529. The minimum multimodal radiation efficiency of baffled finite beams. 1.30 K. A. CUNEFARE 1992 AIAA J. 30, 2819-2828. Effect of modal interaction on sound radiation from vibrating structure. 1.31 K. A. CUNEFARE and M. N. Currey 1994 Journal of the Acoustical Society of America 96(4), 2302-2312. On the exterior acoustic radiation modes of structures. 1.32 M. N. CURREY and K. A. CUNEFARE 1995 Journal of the Acoustical Society of America 98(3), 1570-1580. The radiation modes of baffled finite plates. 1.33 G. P. Gibbs, R. L. Clark, D. E. Cox and J. S. Vipperman 2000 Journal of the Acoustical Society of America 107(1), 332-339. Radiation modal expansion: Application to active structural acoustic control. 1.34 M. R. Bai and M. Tsao 2002 Journal of the Acoustical Society of America 112(3), 876-883 Estimation of sound power of baffled planar sources using radiation matrices. 1.35 H. MURAKAMI, N. TSUNADA AND T. KITAMURA, “A Study Concerned with a Mechanism of Disc-Brake Squeal,” SAE Paper # 841233. 1.36 H. MATSUI, H. MURAKAMI, H. NAKANISHI and Y. TSUDA, “Analysis of DiscBrake Squeal,” SAE Paper # 920553. 1.37 W. V. NACK AND A. M, JOSHI, “Friction Induced Vibration: Brake Moan,” SAE Paper # 951095. 1.38 J. FLINT AND J. HULTÈN 2002 Journal of Sound and Vibration 254 (1), 1-21. Lining-Deformation–Induced Modal Coupling as Squeal Generator in a Distributed Parameter Disc Brake Model. 1.39 D. N. HERTING, MSC/NASTRAN Advanced Dynamic Analysis User’s Guide, pp. 157-173, 1997. 1.40 Y. K. HU, AND L. I. NAGY, “Brake Squeal Analysis by Using Nonlinear Transient Finite Element Method,” SAE Paper # 971510. 1.41 O. N. HAMZEH, W. W. TWORZYDLO, H. J. CHANG AND S. T. FRYSKA, “Analysis of Friction-Induced Instabilities in a Simplified Aircraft Brake, SAE Paper # 1999-01-3404. 1.42 G. D. LILES, “Analysis of Disc Brake Squeal Using Finite Element Methods,” SAE Paper # 891150. 14 1.43 G. DIHUA AND J. DONGYING, “A Study on Disc Brake Squeal using Finite Element Methods,” SAE Paper # 980597. 1.44 T. HAMABE, I. YAMAZAKI, K, YAMADA, H. MATSUI, S. NAKAGAWA AND M. KAWAMURA, “Study of a Method for Reducing Drum Brake Squeal,” SAE Paper # 1999-01-0144. 1.45 S. W. KUNG, K. B. DUNLAP AND R. S. BALLINGER, “Complex Eigenvalue Analysis for Reducing Low Frequency Squeal,” SAE Paper # 2000-01-0444. 1.46 T. S. SHI, O. DESSOUKI, T. WARZECHA, W. K. CHANG, AND A. JAYASUNDERA, “Advances in Complex Eigenvalue Analysis for Brake Noise” SAE Paper # 200101-1603. 1.47 K. B. DUNLAP, M. A. RIEHLE AND R. E. LONGHOUSE, “An Investigative Overview of Automotive Disc Brake noise” SAE Paper # 1999-01-0142. 1.48 J. G. MCDANIEL AND X. LI, “Analysis of Instabilities and power flow in Brake Systems with Coupled Modes” SAE Paper # 2001-01-1602. 15 CHAPTER 2 ACOUSTIC RADIATION FROM OUT-OF-PLANE MODES USING THIN AND THICK PLATE THEORIES 2.1 Introduction Acoustic radiation from thick plates or disks has not been adequately examined though there is a substantial body of literature on the structural dynamics of thin and thick plates [2.1-2.12]. Limited acoustic studies have considered either flexural vibration modes or rigid body piston motions of thin disks [2.14-2.19]. For instance, Thompson [2.14] computed self and mutual radiation impedances of a uniformly vibrating annular or circular piston by integrating of the far-field directivity function. Lee and Singh [2.15] proposed a polynomial approximation for modal acoustic power radiation from a thin annular disk but this method was restricted to only out-of-plane modes. Levine and Leppington [2.16] developed an analytical solution for active and reactive powers from a planar annular membrane given axisymmetric motions. Rdzanek and Engel [2.17] suggested asymptotic formulas for power from a thin annular disk with clamped edges. Finally, Wodtke and Lamancusa [2.18] investigated a circular plate using finite element analysis and then calculated the sound radiation via the Rayleigh integral formula. In this 16 chapter, new analytical and semi-analytical methods for sound radiation from a thick annular disk will be proposed. In particular, we comparatively evaluate the merits of thin vs. thick plate theories on the calculation of radiation from out-of-plane flexural modes. Vibro-acoustic experiments and large scale finite and boundary elements codes are used to validate the analytical formulation. y fn(t) r ϕ b z x a h Figure 2.1: A thick annular disk with an excitation in the normal direction 17 Outer radius a (mm) Inner radius b (mm) Radii ratio β (= b/a) Thickness h (mm) Thickness Ratio h (=h/a) Density ρd (Kg/m3) Young’s modulus E (GPa) Poisson’s ratio ν Disk I 151.5 82.5 0.54 31.5 0.21 7905.9 218 0.305 Disk II 139.0 82.5 0.59 31.5 0.23 7905.9 218 0.305 Disk III 151.5 82.5 0.54 16.3 0.11 7905.9 218 0.305 Table 2.1: Disk examples with free-free or fixed (r = b) - free boundaries Annular disk idealization can be used to analyze many real-life mechanical components such as gears, brake rotors, clutches, flywheels, railway wheels, circular saws, and electric motor. In many cases, thickness (h) is not negligible relative to other dimensions of the component, and thus one must consider the thickness effects in structural dynamic and acoustic radiation characteristics. Figure 2.1 illustrates the example case that is assumed to be non-rotating and without the complicating effects of inner hub. Disk is assumed to be of uniform h and made of an undamped, isotropic material. First, free-free boundaries at the inner and outer edges are assumed. Then, the inner edge is assumed to be ideally fixed but the outer edge is still free. Table 2.1 provides typical values of 3 disks. Disk I is used for all analytical, numerical, and experimental studies. Additionally, Disk II and III are used for structural modal analysis to examine plate theories. For a 18 complete investigation of the vibro-acoustic characteristics of a thick annular disk, it is necessary to simultaneously consider both in-plane and out-of-plane vibrations. But, the current analysis focuses only on the out-of-plane modal vibration and the resulting sound. Primary assumptions are as follows: (1) Structural and acoustic systems are linear timeinvariant systems and complicating effects such as fluid loading and acoustic scattering from the disk edges are negligible. (2) Structural velocities in the normal direction (z) vary sinusoidally in the ϕ direction. (3) Free and far field sound pressure at the observation point (rp) is generated only by the structural motions of two normal surfaces and the inner or outer radial surfaces at edges does not contribute to the far-field sound. (4) Coupling between in-plane and out-of-plane modes is negligible. Chief objectives of this chapter are as follows. (1) Critically examine thick and thin plate theories and investigate the effect of rotary inertia and shear deformations on the structural eigensolutions and acoustic sound radiation. (2) Develop analytical and semi-analytical solutions for sound radiation from modal vibrations. (3) Validate analytical solutions using computational and/or experimental vibro-acoustic methods. (4) Study the effects of the disk geometry and boundary conditions on sound radiation using the proposed analytical solutions. Only single mode excitations are considered here as the multi-modal excitations and coupling issues will be considered in a future chapter. Table 2.2 summarizes various models or methods that will be employed in this study. For the analytical method, procedure includes analytical determination of structural eigensolutions and resulting sound field. Conversely, the finite element method (FEM) is used for structural analysis for the semi-analytical formulation though the sound field is still computed using a modified circular disk radiator model. 19 Medium Structural Dynamics Structural Dynamics + Acoustic Radiation Model or Method Designation Structural Dynamics Formulation Acoustic Radiation Formulation Disk Boundaries Method Type A Thick plate - Free-Free Analytical B Thick plate without rotary inertia effect - Free-Free Analytical C Thin plate - Free-Free Analytical D Finite elements - Free-Free Computational E Experiment - Free-Free Experimental F Thick plate - Fixed-Free Analytical G Thick plate without rotary inertia effect - Fixed-Free Analytical H Thin plate - Fixed-Free Analytical I Finite elements - Fixed-Free Computational J Thick plate Thick plate Free-Free Analytical K Thin plate Thin plate Free-Free Analytical Thick plate Free-Free SemiAnalytical Boundary elements Free-Free Computational L M Finite elements Finite elements N Experiment Experiment Free-Free Experimental O Finite elements Boundary elements Fixed-Free Computational Table 2.2: List of models developed and methods employed for Disk I 20 2.2 Structural Analysis Based on Thick and Thin Plate Theories 2.2.1 Thick Plate Theory According to the procedure proposed by Mindlin and Deresiewicz [2.7] or Mcgee et al. [2.8], the vibratory displacements of a thick annular disk are assumed as follows, while recognizing the effects of shear and rotating inertia. u = zΨr (r , ϕ, t ) v = zΨϕ (r , ϕ, t ) (2.1) w = w(r , ϕ, t ) Here, u, v and w are components in the radial (r), circumferential (ϕ), and transverse directions (z), Ψr and Ψϕ are the bending rotations of normal to the mid-plane in radial and circumferential directions, respectively. Refer to the list of symbols given at the end of this chapter for a complete list of symbols. The equations of motion in terms of the stress resultants in polar coordinates (r, ϕ) are ρ h 3 ∂ 2 Ψr ∂M r 1 ∂M ϕ 1 + (M r − M ϕ )− Q r = d + r ∂ϕ r 12 ∂t 2 ∂r ∂M rϕ 2 1 ∂M ϕ 2 ρd h3 ∂ Ψϕ + + M rϕ − Qϕ = ∂r r ∂ϕ r 12 ∂t 2 ∂Q r 1 ∂Q ϕ 1 ∂2w + Q r − Qϕ = ρ d h 2 + r ∂ϕ r ∂r ∂t (2.2) (2.3) (2.4) where ρd is the mass density of the annular disk. The stress resultants in terms of moments Mr, Mϕ, and Mrϕ, along with shear forces Qr and Qϕ can be related to the transverse displacements and bending rotations as: 21 ∂Ψϕ ∂Ψ ν M r = Db r + Ψr + ∂ϕ r ∂r ∂Ψϕ 1 M ϕ = Db Ψr + ∂ϕ r M rϕ = M ϕr = , ∂Ψ + ν r , ∂r (1 − ν) Db ∂Ψϕ 1 ∂Ψr + − Ψϕ r ∂ϕ 2 ∂r (2.5) (2.6) (2.7) ∂w Q r = κ 2 Gh Ψr + ∂r (2.8) 1 ∂w Q ϕ = κ 2 Gh Ψϕ + r ∂ϕ (2.9) where Db = Eh3 / 12(1 − ν2 ) is the flexural rigidity, E is the modulus of elasticity, ν is the Poisson’s ratio, κ2=π2/12 is the shear correction factor, and G is the shear modulus of the disk. Assume a harmonic variation with time, Ψr (r , ϕ, t ) = ψ r (r , ϕ) cos ωt Ψϕ (r , ϕ, t ) = ψ ϕ (r , ϕ) cos ωt (2.10) w(r , ϕ, t ) = W (r , ϕ) cos ωt to reduce equations (2.2– 2.4) to ω2 ρ d h 3 ∂M r 1 ∂M ϕ 1 + (M r + M ϕ )− Q r + ψr = 0 + r ∂ϕ r 12 ∂r ∂M rϕ ∂r + ω2 ρ d h 3 1 ∂M ϕ 2 + M rϕ − Q ϕ + ψϕ = 0 r ∂ϕ r 12 ∂Q r 1 ∂Q ϕ 1 + Q r + ω 2 ρ d hW = 0 + r ∂ϕ r ∂r 22 (2.11) (2.12) (2.13) The transverse deflection amplitude (W) and associated angular rotations (ψr and ψϕ) are defined in terms of three potential functions (φ1, φ2, and φ3) as: ψ r = (σ1 − 1) ψϕ = ∂φ1 ∂φ 1 ∂φ 3 + (σ 2 − 1) 2 + ∂r ∂r r ∂ϕ (σ1 − 1) ∂φ1 (σ 2 − 1) ∂φ 2 + ∂ϕ r r ∂ϕ + ∂φ 3 ∂r W = φ1 + φ 2 (2.14) (2.15) (2.16) while introducing the following parameters: σ 1 ,σ 2 = (δ , δ 2 2 2 1 { ) 4 1 Rλ − S [ −1 λ4 2 δ ,δ = R + S ± (R − S ) + 4λ4 2 2 2 2 1 R= (2.17a-b) ] −1 / 2 ρ ω2 h D h2 , S = 2 b , λ4 = d 12 Db κ Gh } (2.18a-b) (2.19a-c) Substitution of equations (2.3–2.9) and (2.14–2.16) in (2.11–2.13) along with a series of subsequent manipulations yields (∇ (∇ (∇ 2 2 2 ) + δ )φ + δ )φ + δ12 φ1 = 0 2 2 2 =0 2 3 3 =0 (2.20) where ∇2 is the harmonic differential operator, and another parameter is introduced as follows: 23 1 δ 23 = 2 Rλ4 − (1 − ν ) S (2.21) The solutions to equations (2.11–2.13) require the determination of the potential functions φ1, φ2, and φ3 that must satisfy equation (2.20). φ1 (r , ϕ) = Rn1 sin(nϕ) φ2 (r , ϕ) = Rn 2 sin(nϕ) (2.22 a-c) φ3 (r , ϕ) = Rn3 cos(nϕ) Introducing equation (2.22) into (2.20) yields r2 ( ) d 2 Rni dR + r ni + δ i2 r 2 − n 2 = 0; i = 1,2,3 2 dr dr (2.23) where n is typically a positive integer. The general solutions to equations (2.23) involve ordinary and modified Bessel functions of the first and second kinds and the six constants of integration that are determined from the boundary conditions. 2.2.2 Thin Plate Theory The thin plate theory essentially neglects the effects of rotary inertia and additional deflections caused by shear forces [2.5]. Consequently, the governing differential equation for transverse displacement w(r, ϕ, t) in the mid-plane of the plate is 2 ∂2 1 ∂ 1 ∂2 ∂2w − ρ =0 Db 2 + + 2 w h d r ∂r r ∂ϕ 2 ∂t 2 ∂r (2.24) Solution to this equation is assumed as follows: w(r , ϕ, t ) = W (r ) cos(nϕ)e− iωt 24 (2.25) Using equations (2.24-2.25), the following Bessel’s equation is obtained. d 2 1 d n2 2 + + r dr r 2 dr 2 W − λ4W = 0 (2.26) General solution to this equation can be written as W ( r ) = C1 J n (λ mn r ) + C 2Y n (λ mn r ) + C 3 I n (λ mn r ) + C 4 K n (λ mn r ) (2.27) where Jn and Yn are the Bessel functions of first and second kinds and In and Kn are modified Bessel functions of first and second kinds. Here n is the order of the Bessel function representing the number of nodal diameters and m is the order of eigenvalues representing the number of nodal circles [2.5]. 2.2.3 Eigensolutions for Free-Free Boundaries For the thick plate theory, free-free boundary conditions at the inner and outer radial edges can be expressed as follows: M r (a, ϕ) = M rϕ (a, ϕ) = Qr (a, ϕ) = 0, M r (b, ϕ) = M rϕ (b, ϕ) = Qr (b, ϕ) = 0 (2.28a-b) From the expressions of Mr, Mrϕ, and Qr as defined in (2.13-2.23) along with boundary conditions defined by (2.28), one could formulate the following equation in matrix form. [TTHICK ]{C}= {0} (2.29) Here, [TTHICK] is a 6 ×6 characteristic matrix with elements of various Bessel functions, {C} is an arbitrary coefficients vector and {0} is a null vector. For the thin plate theory, boundary conditions of (2.28) are simplified as follows: 25 M r (a, ϕ) = M rϕ (a, ϕ) = 0, M r (b, ϕ) = M rϕ (b, ϕ) = 0 (2.30a-b) Given the relations between w and bending moments (Mr, Mrϕ), the following equations can be derived for the boundaries satisfying (2.30). 1 ∂w 1 ∂ 2 w ∂ 2w + ν + 2 2 = 0, ∂r 2 r ∂ r r ∂ϕ (2.31a-b) ∂ ∂ w 1 ∂w 1 ∂ w 1 − ν ∂ ∂w w + + =0 + + ∂r ∂r 2 r ∂r r 2 ∂ϕ2 r 2 ∂ϕ2 ∂r r 2 2 2 The characteristic matrix equation corresponding to thin plate theory, similar to equation (2.29), is as follows. [TTHIN ]{C}= {0} (2.32) The characteristic or frequency equations are obtained from equations (2.29) or (2.32). 2.2.4 Validation Studies Analytical solutions for the free-free boundaries, as obtained by both thin and thick plate theories, are compared in Table 2.3. Results of finite element analyses and structural modal experiments are also provided for Disks I, II, and III. Only the first four modes are listed in Table 2.3 since the relevant upper frequency for acoustic radiation study for Disk I is 8 kHz. In the finite element method (FEM), 11 out-of-plane modes have been obtained in the frequency range from 0 to 16 kHz with a model that includes 4,400 solid brick elements and 6,600 nodes [2.13]. In addition, mode shapes of Disk I from alternate analytical approaches are compared with numerical analysis in Figure 2.2. As shown in this figure, the mode shapes from alternate plate theories are very similar in 26 spite of differences in natural frequencies. In modal experiments, the excitation force f(t) is applied in the z direction by an impulse hammer (PCB GK291C) at ϕ = 0° at the outer edge of the disk. The set up for structural modal experiment is explained in Figure 2.3. The frequency range and resolution (∆f) of this experiment are set as 16 kHz and 8 Hz respectively. Natural frequencies (ωmn) and modal damping ratios (ςmn) are extracted /f(ω) where w is the acceleration and f is the applied force. from accelerance spectra w As shown in Table 2.3, the finite element predictions match well with measurements. Analytical solutions based on the thick plate theory produce more accurate answers than the ones that based on the thin plate theory. Yet, even the thick plate theory predictions show significant errors over the higher frequency range. Also, it is obvious that differences between eigenvalues based on two alternate plate theories are proportional to the thickness ratio ( h = h/a). Modal damping ratios are estimated from measured accelerance spectra using the half-power bandwidth method for every resonant peak and ( r , ϕ) is these results are summarized in Table 2.4. In addition, accelerance spectrum w calculated based on the numerical modal dataset using the forced vibration analysis in FEM. These results are subsequently used as excitation to numerical or analytical methods for the calculation of far-field sound radiation. Figure 2.4 compares computed and measured accelerance spectra and a good agreement over the given frequency range is observed. Dominant peaks in this figure correspond to the out-of-plane modes whose frequencies are listed in Table 2.3. 27 0.4 Displacement (µm) Displacement (µm) 0.1 0 -0.1 0.5 0.75 0.3 0.2 0.1 0.5 1 r/a 0.75 1 r/a Figure 2.2: Comparison of Disk I mode shapes given free – free boundaries. (a) (1, 0) mode; (b) (0, 2) mode. Key: WKLFNSODWHWKHRU\0RGHO$ WKLQSODWHWKHRU\ (Model C); , finite element method (Model D). 28 Microphone p i φ = φi, rp = 303 mm) Anechoic Chamber φ Accelerometer w ° °, ru= 151.5 mm) Impulse Hammer f ° ° rf = 151.5 mm) Rotor Signal Conditioning Unit Signal Conditioning Unit FFT Analyzer FFT Analyzer /f(ω), acoustic Figure 2.3: Vibro-acoustic experiment used to measure structural w frequency response functions P/f(ω) and out-of-plane modal sound radiations. 29 Disk I β=0.54 h =0.21 II β=0.59 h =0.23 III β=0.54 h =0.11 ωmn (kHz) Non-dimensional Eigenvalues λmn2 = ωmn a2(ρd h/Db)1/2 Mode Indices m n Thick Plate Theory (Model A) Thin Plate Theory (Model C) Finite Element (Model D) 0 2 3.82 4.02 3.86 3.92 1.331 1 0 8.85 9.80 8.69 9.02 3.063 0 3 10.59 11.11 10.04 10.25 3.481 1 1 15.42 17.55 13.57 14.00 4.756 0 2 3.62 3.90 3.72 3.72 1.500 1 0 9.14 10.53 8.74 9.33 3.756 0 3 10.04 10.73 9.75 9.78 3.938 1 1 15.36 18.32 13.22 13.81 5.563 0 2 4.06 4.12 4.03 4.03 0.706 1 0 9.55 9.82 8.99 9.53 1.669 0 3 11.11 11.13 10.71 10.78 1.888 1 1 16.86 17.59 15.27 15.99 2.800 Experiment Experiment (E) (E) Table 2.3: Eigensolutions for Disk I, II and III with free-free boundaries. 30 Mode Indices m n Damping Ratio (%) 0 2 0.62 1 0 0.34 0 3 0.26 1 1 0.26 Table 2.4. Measured modal damping ratios for Disk I. 31 150 (0, 4) (0, 3) (0, 2) (1, 1) 2 dB re 20 µm/s -N (1, 0) 100 50 0 2 4 6 8 Frequency (kHz) /f(ω) at r = 151.5 mm and ϕ = 180° for Disk I with Figure 2.4: Structural accelerances w free-free boundaries. Key: - - -, computed using FEM (Model D); ––––, measured (E). 32 2.2.5 Effect of Fixed-Free Boundaries Eigensolutions for a disk with fixed-free boundaries can be easily calculated from the analytical solutions of Chapter 2.2.1. For the thick plate case, boundary conditions at r = b (fixed) and r = a (free) edges are expressed as follows. M r (a, ϕ) = M rϕ (a, ϕ) = Qr (a, ϕ) = 0, W (b, ϕ) = ψ r (b, ϕ) = ψ ϕ (b, ϕ) = 0 (2.33a-b) Likewise, for the thin plate case, these conditions are specified as M r (a, ϕ) = M rϕ (a, ϕ) = 0, Wr (b, ϕ) = ∂Wr (b, ϕ) = 0 ∂r (2.34a-b) From these boundary conditions, matrix equations similar to equations (2.29) and (2.32) can be obtained and eigenvalues can be determined using the same procedure. In addition, natural frequencies are also calculated using FEM and compared with alternate plate theories in Table 2.5. As in the free-free boundary case, eigensolutions based on the thick plate theory are much more accurate than those given by the thin plate theory. Note that modal experiments are not attempted since it is difficult to experimentally simulate the perfect fixed boundary condition at r = b. 2.2.6 Effect of Disk Geometry on Eigensolutions As one can see from equations (2.20-2.23), structural eigensolution of a thick annular disk are affected by the radii ratio (β = b/a), thickness ratio ( h = h/a), as well as by material properties. In this Chapter, effects of β and h on the non-dimensional 33 eigenvalues are examined for selected modes of Disk I with free-free or fixed-free boundaries. In our investigation, such non-dimensional parameters are controlled by adjusting h and b for a fixed a. Eigenvalues are calculated using three alternate analytical methods based on models A, B, and C as described in Table 2.2. Results are shown in Figures 2.5 and 2.6 where differences in the alternate formulations are evident. For the free-free disks, experimental and FEM results are also included in these figures for the sake of comparison. Non-dimensional Eigenvalue λmn2 = ωmn a2(ρd h/Db)1/2 Mode Indices m n Thick Plate Theory Thin Plate Theory 0 0 0 0 0 1 2 3 11.96 13.43 15.28 18.75 15.72 16.05 17.52 21.17 ωmn (kHz) FEM FEM 13.61 13.63 14.28 16.81 4.623 4.628 4.849 5.709 Table 2.5: Eigensolutions of Disk I with fixed-free boundary conditions. 34 (a) (b) 15 15 (0, 3) (0, 3) 10 λ2 λ2 10 5 0 0.3 (0, 2) 0.5 5 0 0.7 β (0, 2) 0 0.2 h Figure 2.5: Effect of geometry on the (0, 2) and (0, 3) modes of Disk I with free-free boundaries. (a) Effect of radii ratio (β); (b) Effect of thickness ratio ( h ). Key: WKLFN plate theory (Model A); WKLFNSODWHWKHRU\ZLWKRXWURWDU\LQHUWLDHIIHFW0RGHO%- -, thin plate theory (Model C); PHDVXUHG( 35 30 (a) (b) 20 (1, 1) λ2 (1, 1) λ2 20 10 (1, 0) 10 (1, 0) 0 0.3 0.5 0 0.7 β 0 0.2 h Figure 2.6: Effect of geometry on the (1, 0) and (1, 1) modes of Disk I with free-free WKick boundaries. (a) Effect of radii ratio (β); (b) Effect of thickness ratio ( h ). Key: plate theory (Model A); WKLFNSODWHWKHRU\ZLWKRXWURWDU\LQHUWLDHIIHFW0RGHO%- -, thin plate theory (Model C); PHDVXUHG( 36 (a) (b) 40 25 (0, 3) 20 30 20 (0, 2) λ2 λ2 15 (0, 3) 10 10 0 0.2 (0, 2) 5 0.5 β 0 0.8 0 0.1 0.2 h 0.3 Figure 2.7: Effect of geometry on the (0, 2) and (0, 3) modes of Disk I with fixed-free boundaries. (a) Effect of radii ratio (β); (b) Effect of thickness ratio ( h ). Key: WKLFN plate theory (Model F); WKLFNSODWHWKHRU\ZLWKRXWURWDU\LQHUWLDHIIHFW0RGHO*- -, thin plate theory (Model H). 37 40 20 λ2 λ2 30 20 10 10 0 0.2 0.5 β 0 0 0.8 0.2 Figure 2.8: Effect of geometry on the (1, 0) and (1, 1) modes of Disk I with fixed-free boundaries. (a) radii ratio (β); (b) thickness ratio ( h ). Key: PRGHZLWKWKLFN plate theory (Model F); ∆∆∆, (0, 0) mode with thick plate theory without rotary inertia effect (Model G); PRGHZLWKWKLQSODWHWKHRU\0RGHO+ PRGH with thick plate theory (Model F); ◊◊◊, (0, 1) mode with thick plate theory without rotary inertia effect (Model G); - - -, (0, 1) mode with thin plate theory (Model H). 38 Differences between the λ2mn values, based on alternate approaches, are proportional to h for all modes. But as shown in Figures 2.5a and 2.6a, the effects of β are mode dependent. For example, differences are proportional to β for (1, 0) and (1, 1) modes. Conversely, for (0, 2) and (0, 3) modes, such differences are inversely proportional to β. The comparison of natural frequencies based on models A, B, and C suggests that differences between eigensolutions are mainly caused by the rotary inertia effect in (1, 0) and (1, 1) modes and by the shear deformations in (0, 2) and (0, 3) modes. Similar investigation has been executed for the fixed-free disk and results are given in Figures 2.7 and 2.8. Differences between eigenvalues based on thick and thin plate theories appear to be caused by the shear deformations regardless of the mode type. Also, in this case, these differences are proportional to thickness and radii ratios for all modes. 2.3 Acoustic Radiation Calculation 2.3.1 Formulation Sound radiation from the flexural vibrations of circular plates or annular disks has been examined by several investigators [2.14-2.18]. For instance, Thompson [2.14] computed self and mutual radiation impedances of a uniformly vibrating annular or circular piston by integrating of the far-field directivity function. Lee and Singh [2.15] proposed a polynomial approximation for modal acoustic power radiation from a thin annular disk using the far-field and radiation impedance approaches. Levine and Leppington [2.16] developed an analytical solution for active and reactive powers from a planar annular membrane given axisymmetric motions. Rdzanek and Engel [2.17] 39 suggested asymptotic formulas for power from a thin annular disk with clamped edges. Finally, Wodtke and Lamancusa [2.18] investigated a circular plate using finite element analysis and then calculated the radiation via the Rayleigh integral formula. However, none of these studies have examined radiation from a thick annular disk. If acoustic scattering from the edges of a vibrating structure is neglected, the far and free field sound radiation from a planar radiator in an infinite baffle can be typically calculated by the Rayleigh integral method as described below. With reference to Figure 2.9a, the sound pressure amplitude (P) is as follows where ρ0 is the mass density of air, c0 is the speed of sound, k is the acoustic wave number, rp and rs are the position vectors of receiver and source positions, and W is the amplitude of vibratory velocity in the z direction at rs. ρ c k P(r p ) = 0 0 2π ∫ e ik r p − rs W (rs ) r p − rs Ss dS (rs ) (2.35) For an axially symmetric radiator, (m, n)th modal sound pressure can be expressed as follows by simplifying equation (2.35) using the Hankel transform [2.19 - 2.20]. Here, Jn is the Bessel function of order n, R = |rp| is the radius of sphere on which the observation positions are defined, and θ and φ are the cone and azimuthal angles of the observation positions. ρ0c0 keik mn Rd cos nφ(−i ) n +1 Β n [w (r )]; Pmn ( R, θ, φ) = Rd Β n [w (r )] = ∫ ∞ 0 w (r ) J n (kr r )rdr ; kr = k sin θ; Rd = rp − rs . 40 (2.36a-d) z (a) Pmn(R, θ, φ) Sv rprs W mn rp R rs Ss (b) Pmn(R, θ, φ) z Sv rprs W mn γ rp rs Ss R Figure 2.9: Sound Radiation from the out-of-plane vibration modes in the sherical coordinate system. (a) a thin disk with an infinite baffle; (b) a thick disk without baffle. 41 For a non-planar source, the far and free field sound pressure can be expressed as equation (2.37) based on the plane-wave approximation within the short-wavelength limits [2..20] along with reference to Figure 2.9b: ik r − r ρ0c0 k e p s W (rs ) P ( rp ) = (1 + cosγ ) dS 4π ∫S s rp − rs (2.37) In our study, the observation positions are defined by a group of points having equal angular increments (∆ϕ, ∆θ) on a sphere (SV) that is centered at the disk center and sound pressures at all of the observation positions are calculated using equations (2.35), (2.36), or (2.37). The modal directivity function Dmn(θ, φ) at frequency ωmn can be defined from the modal pressure Pmn(rp) expression as follows. Dmn (θ, φ) = RPmn ( R, θ, φ)eik mn R (2.38) From the far-field approximation, modal sound power Πmn of the (m, n)th mode is calculated using the following equation. 2 Π mn = I mn SV s 1 2π π P = ∫ ∫ mn R 2 sin θ dθ dφ 2 0 0 ρ0c0 (2.39) Here, Imn is the acoustic intensity, and Sv is area of the control surface. The modal radiation efficiency σmn of an annular disk is determined from Πmn as follows where < w mn 2 > t , s is the spatially averaged mean-square velocity on the two normal surfaces of the annular disk. 42 σ mn = Π mn 2 < w mn >t , s 2 < w mn >t , s = ; (2.40a-b) 1 a 2π(a 2 − b 2 ) ∫b ∫ 2π 0 W dϕ dr 2 mn 2.3.2 Thin Plate Approach The classical method based on the thin plate theory does not consider the effect of h in sound radiation calculation. In this case, h is assumed to be negligible relative to other disk dimensions and sound radiation from the (m, n)th mode of a thick annular disk is generally calculated by assuming the baffled condition. In addition, Rd in equations (2.36) is approximated with R and equation (2.36) is simplified to yield the following equation. ρ 0 c 0 k mn e ik mn R Pmn ( R, θ, φ) = cos nφ( −i) n +1 Β n [w ( r )] R (2.41) As an alternative to equation (2.36), the results of structural analysis of Chapter 2.2 may be used. Define approximate mode shape Ψmn and modal surface velocity w mn as follows where polynomial functions are used to describe the flexural vibrations. N ψ mn (r , ϕ) = cos( nϕ)∑ Cmn , s r s s =0 (2.42a-b) w mn (r , ϕ, t ) = W ( r , ϕ)eiω mn t = −ωmnCmn , s r s cos( nϕ)eiω mn t Here, Cmn,s is an arbitrary constant. Substituting equation (2.37) into equation (2.36), the far-field modal sound pressure Pmn is: 43 Pmn ( R, θ, φ) = N a ρ 0 ck mn e ik mn Rd cos nφ(−i ) n +1 ∑ C mn, s ∫ r s +1 J (k mn r sin θ) dr b Rd s =0 (2.43) By expressing the Bessel function of the first kind with corresponding power series, modal sound pressure can be expressed as: Pmn ( R, θ, φ) = ρ0ckmneik mn Rd cos nφ(−i ) n +1 × Rd N ∞ ∑∑ s =0 l =0 Cmn , s (−1)l (kmn sin θ) 2 m +1 b n + s + 2l + 2 l!(n + l )!2 2l + n (n + s + 2l + 2) (2.44) (1 − β n + s + 2l + 2 ) The analytical method based on the thin plate theory (Model K) uses equation (2.44) for sound pressure calculations. 2.3.3 Thick Plate Approach For the thick plate theory, we do not ignore the thickness (h) effect. Therefore, sound radiations from two normal surfaces should be simultaneously considered. With reference to equation (2.37) along with the far-field assumption, angle γ is approximated by θ for the normal surface facing the field point and by -θ on the opposite side. Consequently, sound pressure at rp is given by the sum as shown below: ik r p − rs' ik r − r e W (rs' ) ρ0ck e p s W (rs ) P ( rp ) = (1 + cosθ) + ∫ (1 − cosθ) dS ∫S s ' Ss 4π rp − rs rp − rs (2.45) Here, rs’ is the position vector of a source point on the normal surface that is away from the field point. In addition, |rp - rs| and |rp - rs’| are expressed as follows: 44 Rd = rp − rs = h hz h x 2 + y 2 + ( z − ) 2 ≈ R(1 − 2 ) ≈ R − cos θ; 2 2 R Rd' = rp − rs' = h hz h x 2 + y 2 + ( z + ) 2 ≈ R(1 + 2 ) ≈ R + cos θ; 2 R 2 (2.46a-f) R = rp ; x = R sin θ cos φ; y = R sin θ sin φ; z = Rcosθ; According to the far-field condition, Rd and Rd’ in the denominator of equation (2.46) can be approximated by R, but the numerator terms cannot be replaced with R especially over the higher frequency range. Substituting (2.46) into (2.45) and employing the Hankel transform used in equation (2.36), we find the following equations with a simplified R expression in the denominator. ρ0ckmn eik mn R − ik mn 2 cos θ P ( R, θ, φ) = e cos nφ( −i ) n +1 Β n [w ( r )] 2R h s mn o Pmn ( R, θ, φ) = ρ0ckmn e 2R ik mn R e ik mn h cos θ 2 (2.47a-b) cos n(φ + π)(−i) n +1 Β n [w (r )] As one can see from these equations, the disk thickness introduces a phase difference that is equal to –kmn(h/2)cosθ for the surface facing the field point and is kmn(h/2)cosθ for the surface away from the field point. The total far-field modal sound pressure is expressed by a sum of sound radiations from two normal surfaces as follows: o Pmn ( R,θ , φ ) = (1 + cos θ ) Pmns ( R,θ , φ ) + (1 − cos θ ) Pmn ( R, θ , φ ) (2.48) The semi-analytical (Model L) and analytical (Model J) methods based on the thick plate theory consider the effect of h and use equation (2.48) for the calculation of sound pressure. 45 2.4. Computational and experimental Investigations of Sound Radiation Modal acoustic radiation properties such as acoustic frequency response functions P/f(ω), modal acoustic power (Πmn), and modal radiation efficiency (σmn) of Disk I are obtained using the analytical methods of Chapter 2.3. Furthermore, the same radiation properties are calculated with uncoupled, direct, exterior, and unbaffled boundary element analyses [2.21]. In the computational study (Model M), 6,146 acoustic field points and 6,144 elements are defined on the sphere (Sv) surrounding the disk that is represented by the finite element model for structural dynamics. The center of this sphere coincides with the disk center. Excitations to this boundary element analysis (BEM) are the normal velocity distribution W ( r , ϕ) on both normal surfaces that are obtained from the forced vibration analysis using the finite element code (Model D). Analytical predictions and numerical analyses are verified by comparing results with measured data obtained from vibro-acoustic experiments conducted in an anechoic chamber as shown in Figure 2.3. Far-field sound pressures are measured with a 6 mm microphone (MTS L130C10 combined with pre-amplifier MTS 130P10) at predetermined field points on a circle of R = 303 mm radius from the disk center in the plane of ϕ = 0° and θ = 90°. Considering the symmetries of the pressure distributions, P is measured in the range of 0° ≤ θ ≤ 90° in the ϕ = 0° plane with an increment of ∆θ = 2.5° and over 0° ≤ ϕ ≤ 90° in the θ = 90° plane with an increment of ∆φ = 5°. The same radius (303 mm) is used in computational and analytical studies. Force and pressure signals are conditioned and analyzed via a 2-channel dynamic signal analyzer (HP 35670A) to obtain p/f(ω) spectra such as the one shown in Figures 2.10. The experimental directivity pattern D(θ, φ) on 46 the sphere SV is synthesized from measured P(θ, φ) data. Our analytical methods accurately predict the far-field sound pressure distributions. This is illustrated in Figure 2.11 where P(θ, φ) results from alternate analytical procedures are compared with measured and computed values for (0, 2) and (0, 3) modes. Further, Table 2.6 compares directivity patterns in a pictorial form. Analytical predictions of Πmn and σmn for two modes are compared with BEM code (Model C) and measured results in Table 2.7.In the experimental case, results at the discrete points over SV have been synthesized using the measured Pmn(θ, φ) data to yield Πmn along with σmn. In this process, the measured Pmn(θ, φ) profile is assumed to have a perfect sinusoidal variation in the φ direction. As shown in Tables 2.6 and 2.7, acoustic radiation properties obtained using analytical solutions (Models J, K, and L) match well with computational predictions (Model M) and measurements. The discrepancies between measured data and analytical or numerical results can be explained by uncertainties in the acoustic experiment. There are two important uncertainties in the acoustic experiment. The first uncertainty is generated by frequency resolution in the experiment. Since this study covers very wide frequency range (0 – 16kHz), relatively coarse frequency resolution (8 – 16 Hz) is used in the experiment. Consequently, measured sound pressure for a given receiver position could lower than the actual sound pressure up to 6 dB. On the other hand, analytical and numerical sound pressures are calculated at exact natural frequencies. The other uncertainty is due to spatial resolution. Angular resolutions of 2.5 and 5.0° are used in the measurement that induces maximum 3.0 dB error in the sound pressure. 2.5° angular resolution is used in the calculations and maximum error is 1.5 dB. 47 P/f (dB re 20µPa/N) 100 80 60 40 20 0 1 2 3 1 2 3 4 5 6 7 8 4 5 6 7 8 P/f (dB re 20µPa/N) 100 80 60 40 20 0 Frequency (kHz) Figure 2.10: Acoustic frequency response function P/f(ω) given unit impulsive force excitation f(t) in the z direction at r =151.5 mm. (a) θ = π/2 and φ = 0; (b) θ = 0 and φ = 0. Key: –––, analytical calculation (Model L); - - -, computed using BEM (Model M); measured. Here q = 0 is the radial mode. 48 Normalized Pressure 1 0.5 0 Normalized Pressure 0 30 60 90 30 60 90 1 0.5 0 0 deg Figure 2.11: Directivity pattern Dmn(θ) given φ = 0 and R = 303 mm. (a) m = 0, n = 2 mode; (b) m = 0, n = 3 mode. Key: DQDO\WLFDOPHWKRGEDVHGRQWKLFNSODWHWKHRU\ (Model J); ∆∆∆, analytical method based on thin plate theory (Model K); - - -, semianalytical method (Model L); , computed using BEM (Model M); PHDVXUHG 49 Structural Mode (0, 2) Mode + - - (0, 3) Mode + + + - + Analytical Method Measured Computed using BEM Table 2.6: Comparison of directivity patterns for selected modes of Disk I. 50 Πmn, dB re 1 pW Analytical Thick Plate Thin Plate (Model J) (Model K) Measured BEM (Model O) SemiAnalytical (Model L) (1, 1) 70.1 70.7 71.6 70.1 70.9 (0, 3) 75.3 76.3 76.2 75.1 74.8 Mode (m, n) Measured BEM (Model O) SemiAnalytical (Model L) (1, 1) 0.77 0.75 1.08 1.23 1.06 (0, 3) 0.81 1.01 1.01 1.28 1.14 Mode (m, n) σmn Analytical Thick Plate Thin Plate (Model J) (Model K) Table 2.7: Comparison of modal acoustic power and radiation efficiency levels for selected out-of-plane modes of Disk I. 51 2.5 Effect of key Parameters on the Modal Sound Radiation As shown in Chapter 2.3, natural frequencies and modes of a thick annular disk are affected by its geometry and boundaries. Furthermore, equations (2.44-2.45) illustrate that Pmn depends on disk geometry, vibrating mode and frequency. In this Chapter, effects of the radii ratio (β = b/a), thickness ratio ( h = h / a ), and boundary conditions on modal sound radiation are studied through variations in < w mn 2 > t , s , Πmn and σmn as introduced by changes in key parameters. As a first step, natural frequencies and mode shapes corresponding to a specific geometric configuration are calculated using thin or thick plate theory and then the modal surface velocities are defined from the corresponding structural eigensolutions. Then, the modal far-field sound pressures are calculated using equation (2.43) or (2.45). Finally, Πmn and σmn are obtained from the sound pressure data using equations (2.39-2.40). In this particular study, the amplitudes of modal vibrations are intentionally adjusted to get the same modal velocity amplitudes regardless of variations in the natural frequencies for a given geometric configuration. 2.5.1 Effects of Radii Ratio First, the effect of β is investigated using Disk I. The results of this investigation are summarized in Figures 2.12 and 2.13 where < w mn 2 > t , s and σmn are significantly affected by β. For a limiting case of β → 0 when the annular disk turns into a circular disk, sound radiation can be solved using the same solution. And, for the other limiting case when β → 1, the annular disk can be considered as a thin cylinder that cannot generate sound with its out-of-plane vibration. 52 (b) 2 50 1 0 0.3 σ 2 2 (pm /s ) (a) 100 0.5 β 0.7 0 0.3 0.9 0.5 β 0.7 0.9 Figure 2.12: Effect of radii ratio on the modal sound radiation based on alternate plate 2 theories. (a) Spatially averaged mean-square velocity < w > t , s ; (b) Radiation efficiency σ. Key: PRGHZLWKWKLFNSODWHWKHRU\0RGHO- PRGHZLWKWKLQ plate theory (Model K); PRGHZLWKWKLFNSODWHWKHRU\0RGHO-- - -, (0, 3) mode with thin plate theory (Model K). 53 (a) (b) 3 2 2 2 (pm /s ) 100 σ 50 1 0 0.3 0.5 β 0.7 0 0.3 0.9 0.5 0.7 0.9 β Figure 2.13: Effect of radii ratio on the modal sound radiation based on alternate plate 2 theories. (a) Spatially averaged mean-square velocity < w > t , s ; (b) Radiation efficiency σ. Key: PRGHZith thick plate theory (Model J); PRGHZLWKWKLQ plate theory (Model K); PRGHZLWKWKLFNSODWHWKHRU\0RGHO-- - -, (1, 1) mode with thin plate theory (Model K). 54 As shown in Figure 2.12, < w mn 2 > t , s and σmn for (0, 2) and (0, 3) modes converge to 0 as β → 1 irrespective of plate theories. But, as shown in Figure 2.13, σmn values for (1, 0) and (1, 1) modes based on thin plate theory significantly fluctuate with β even in the case of β → 1 though the corresponding < w mn 2 > t , s values monotonically decrease. Conversely, σmn values for (1, 0) and (1, 1) modes based on the thick plate theory do not show much fluctuation with β. It is conceivable that Πmn values based on the thin plate theory fluctuate with β but the same values based on the thick plate theory are stable. 2.5.2 Effects of Thickness Next, h is selected as an independent variable and it is varied from 0.025 to 0.35 with a nominal value of h0 = 0.21. Results are summarized in Figures 2.14 and 2.15 where one can observe considerable variations in σmn. The < w mn 2 > t , s values based on the thin plate theory are constant irrespective of the mode type. Conversely, the same data based on the thick plate theory are mode dependent. For instance, < w mn 2 > t , s is proportional to h for (0, 2), (0, 3) or (1, 0) modes, but is inversely proportional to h for (1, 1) mode. As shown in Chapter 2.2, natural frequencies of the out-of-plane modes are proportional to h. If the disk thickness is small enough such that the natural frequency for a specific mode is below the critical frequency, the modal sound radiation is very low [2.20]. For this reason, σmn values are very low in the region of small h regardless of the mode type as shown in Figures 2.14 and 2.15. 55 (a) (b) 1.5 1 2 2 (pm /s ) 150 σ 125 0.5 100 0 0 0 0.2 Á 0.2 Á Figure 2.14: Effect of thickness ratio on the modal sound radiation based on alternate 2 plate theories. (a) Spatially averaged mean-square velocity < w > t , s ; (b) Radiation efficiency σ. Key: PRGHWKLFNZLWKSODWHWKHRU\0RGHO- PRGH with thin plate theory (Model K); PRGHZLWKWKLFNSODWHWKHRU\0RGHO-- -, (0, 3) mode with thin plate theory (Model K). 56 (a) (b) 1.5 1 2 2 (pm /s ) 50 σ 25 0.5 0 0 0 0 0.2 h 0.2 h Figure 2.15: Effect of thickness ratio on the modal sound radiation based on alternate 2 plate theories. (a) Spatially averaged mean-square velocity < w > t , s ; (b) Radiation PRGHZLWKWKLFNSODWHWKHRU\0RGHO- PRGH efficiency σ. Key: with thin plate theory (Model K); PRGHZLWKWKLFNSODWHWKHRU\0RGHO-- -, (1, 1) mode with thin plate theory (Model K). 57 Furthermore, for the thick plate theory, phase difference between sound pressures radiated from two normal surfaces is proportional to h and the effect of this phase difference should be considered in addition to the effect of natural frequency change. For the thin plate theory that considers sound pressure from only one normal surface, radiation is affected only by the natural frequency variation. 2.5.3 Effects of Boundary Conditions The effect of fixed–free boundary conditions on sound radiation is finally studied. Typical Πmn and σmn of two out-of-plane modes for Disk I with either free-free or fixedfree boundary conditions are listed in Table 2.8. The modal acoustic powers and radiation efficiencies for modes with the same number of nodal diameters (n) significantly change when the inner edge is clamped. For example, Π02 and σ02 increase with fixed-free boundaries due to the increases in the corresponding natural frequencies. For instance, ω02 goes up from 1.31 kHz (below the critical frequency that is around 2.0 kHz) to 4.85 kHz (above the critical frequency). Also, Πmn and σmn for n = 0 and n = 1 modes, significantly increase due to the elimination of a nodal circle in the corresponding mode shapes. In addition, directivity patterns of two sample modes with fixed-free boundaries are numerically calculated and compared in Figure 2.16 with those with free-free boundaries. The application of fixed-free boundary conditions increases the number of ripples in the θ direction for both modes. 58 Boundaries Free - Free Fixed - Free Mode Indices ωmn (kHz) Πmn (dB re 1 pW) σmn (0, 2) 1.31 70.9 0.263 (1, 0) 2.95 67.2 0.379 (0, 3) 3.41 76.3 1.030 (1, 1) 4.61 70.7 0.808 (0, 0) 4.62 69.7 0.900 (0, 1) 4.63 73.4 0.884 (0, 2) 4.85 73.2 0.878 (0, 3) 5.71 73.7 0.883 Table 2.8: Modal acoustic powers and radiation efficiencies for first four out-of-plane modes with fixed-free or free-free boundaries. 59 Normalized Pressure 1 0.5 0 Normalized Pressure 0 30 60 90 30 60 90 1 0.5 0 0 θ (deg) Figure 2.16: Modal directivity patterns of Disk I with alternate boundary conditions. (a) n = 2 modes; (b) n = 0 modes. Key: –––, fixed-free; - - -, free-free boundary condition. 60 2.6 Conclusion This chapter has proposed a new analytical solution that explicitly considers the disk thickness effect on sound radiation from out-of-plane modes. In addition, our semianalytical procedure combines the computationally obtained disk surface velocities with analytical solutions for sound radiation. A comparative evaluation of thin and thick plate theories shows that the thick plate theory is more accurate when predictions are compared with computational codes (such as FEM and BEM) and vibro-acoustic experiments. Our procedure can be efficiently used to conduct parametric studies such as the ones reported in this chapter by varying the radii or thickness ratio. In particular, one can easily analyze the limiting cases of a circular plate and a thin cylinder considering only the out-of-plane flexural modes. In a future chapter, we will simultaneously consider both out-of-plane and in-plane components of the disk vibration. Modal interaction effects and sound radiation from coupled modes will also be studied. 61 REFERENCES FOR CHAPTER 2 2.1 A. W. LEISSA 1969 NASA SP-160 Vibration of Plates. 2.2 A. W. LEISSA 1987 The Shock and Vibration Digest 19(3), 10-24. Recent Research and plate vibration, 1981-1985. Part 1: Classical theory. 2.3 A. W. LEISSA 1987 The Shock and Vibration Digest 19(3), 10-24. Recent research and plate vibration, 1981-1985. Part 2: Complicating effects. 2.4 A. W. LEISSA America. 1993 Vibrations of Plates, New York: Acoustical Society of 2.5 S. M. Vogel and D. W. Skinner 1965 Journal of Applied Mechanics December, 926-931 Natural frequencies of transversely vibrating uniform annular disk. 2.6 R. D. MINDLIN 1951 ASME Journal of Applied Mechanics 18, 31-38 Influence of rotatory inertia and shear on the flexural motion of isotropic, elastic plate. 2.7 R. D. MINDLIN and H. DERESIEWICZ 1954 Journal of Applied Physics 25(10), 13291332 Thickness-shear and flexural vibration of a circular disk. 2.8 O. G. MCGEE, C. S. HUANG and A. W. LEISSA 1995 Journal of Sound and Vibration 163(1), 137-149 Comprehensive exact solutions for free vibrations of thick annular sectorial plates with simply supported radial edges. 2.9 T. IRIE, G. YAMADA and K. TAKAGI 1982 Transactions of the American Society of Mechanical Engineers, Journal of Applied Mechanics 49, 633-638 Natural frequencies of thick annular plates 2.10 IRIE, G. YAMADA and S. AOMURA 1979 Journal of Sound and Vibration 66 (1), 187-197 Free vibration of a Mindlin annular plate of varying thickness. 2.11 C. M. WANG and V. THEVENDRAN 1993 International Journal of Mechanical Science 37(5), 537-566 Vibration analysis of annular plates with concentric support using a variant of Rayleigh-Ritz method. 2.12 K. M. LIEW, Y. XIANG, C. M. WANG AND S. KITIPORNCHAI 1993 Computer Methods in Applied Mechanics and Engineering 110, 301-315 Flexural vibration of shear deformable circular and annular plates on ring support. 2.13 I-DEAS User’s manual version 8.2. 2000 SDRC, USA. 2.14 W. THOMPSON, JR. 1971 Journal of Sound and Vibration 17(2), 221-233. The computation of self- and mutual-radiation impedances for annular and elliptical pistons using Bouwkamp integral. 62 2.15 M. R. LEE and R. SINGH 1994 Journal of the Acoustical Society of America 95(6), 3311-3323. Analytical formulations for annular disk sound radiation using structural modes. 2.16 H. LEVINE and F. G. LEPPINGTON 1988 Journal of Sound and Vibration 121(5), 269-275. A note on the acoustic power output of a circular plate. 2.17 W. P. RDZANEK Jr. and Z. ENGEL 2000 Applied Acoustics 60(5), 29-43. Asymptotic formula for the acoustic power output of a clamped annular plate. 2.18 H. W. WODTKE and J. S. LAMANCUSA 1998 Journal of Sound and Vibration 215(5), 1145-1163. Sound power minimization of circular plates through damping layer placement. 2.19 H. LEE and R. SINGH 2002 submitted to Journal of Sound and Vibration Acoustic radiation from out-of-plane modes of an annular disk using thin and thick plate theories, submitted to Journal of Sound and Vibration. 2.20 M. C. JUNGER and D. FEIT 1985 Sound, Structures, and Their Interactions. New York: MIT Press. 2.21 SYSNOISE User’s manual Version 5.4. 1999 NIT, Belgium. 63 CHAPTER 3 SOUND RADIATION FROM IN-PLANE VIBRATION 3.1 Introduction Many mechanical or structural components such as gears, brake rotors, clutches, flywheels, railway wheels, circular saws, electrical machinery stators, and electroacoustic transducers can be idealized as annular disks. Often thick bodies are encountered and one must examine radiation from both out-of-plane vibrations and in-plane vibrations to appropriately control sound radiation from such components. Even though many researchers have studied structural dynamics of annular disks, published literature on the in-plane oscillations of an annular disk is relatively sparse [3.1 – 3.4]. Bhuta and Jones [3.1] studied coupled symmetric and torsional vibrations of a thin, rotating circular disk. Burdess et al. [3.2] generalized the analysis to consider asymmetric in-plane vibrations and the properties of forward and backward traveling waves. Chen and Jhu [3.3] examined in-plane vibration and stability of a spinning annular disk and determined the effect of rotational speed on the natural frequencies. Irie et al. [3.4] calculated natural frequencies of annular disks using the transfer matrix method. We will use this method for the structural analysis of thick annular disks. 64 Sound radiation from thin circular and annular disks has been examined by several investigators [3.5-3.9]. For instance, Thompson [3.5] computed self and mutual radiation impedances of a uniformly vibrating annular or circular piston by integrating of the farfield directivity function. Lee and Singh [3.6] proposed a polynomial approximation for modal acoustic power radiation from a thin annular disk but this method was restricted to only out-of-plane modes. Levine and Leppington [3.7] developed an analytical solution for active and reactive powers from a planar annular membrane given axisymmetric motions. Rdzanek and Engel [3.8] suggested asymptotic formulas for power from a thin annular disk with clamped edges. Finally, Wodtke and Lamancusa [3.9] investigated a circular plate using finite element analysis and then calculated the sound radiation via the Rayleigh integral formula. However, almost all of the above mentioned studies have considered sound radiation only from either flexural vibration modes or rigid body piston motions. In such studies, sound radiation from the in-plane modes of a disk has been assumed to be negligible compared to that from the out-of-plane modes. But, if the thickness of a disk is beyond the range of thin plate (shell) theory, in-plane vibration could generate sufficient sound given proper excitation. This chapter proposes analytical solutions for sound radiation from radial vibration modes of a thick annular disk. Structural eigensolutions are calculated using the transfer matrix method and then the surface velocities at outer and inner radial edges are determined using the modal expansion. The far-field sound pressure distribution is obtained using two alternate methods. In the first method, pressure is calculated using the Rayleigh integral technique. The second method treats sound radiating radial surfaces as cylindrical radiators of finite length. The Fourier series and Sinc function approaches are 65 employed for calculations. Acoustic powers and radiation efficiencies of radial modes are also determined from the far-field sound pressure calculations. Analytical predictions match well with measured data as well as computational results from a finite element code in terms of structural eigensolutions and from a boundary element code in terms of sound pressure, directivity etc. Selected parametric studies include the effects of disk geometry and natural frequencies on the radiation properties. Limiting cases are also examined. 3.2 Problem Formulation Chief objective of this chapter is to develop analytical solutions for acoustic radiation from the radial structural modes of a thick annular disk. Figure 1 illustrates the example case that is assumed to be stationary and free at the inner and outer edges. Table 1 provides sample values used for analytical and experimental studies. Primary assumptions are as follows: 1. Structural and acoustic systems are linear time-invariant systems and the scattering effect is negligible. 2. Free field sound pressure at the observation point is generated only by the radial velocities on inner (r = b) and outer (r = a) edges and other surfaces do not contribute to the far-field sound pressure. 3. Vibration amplitude of the radial surfaces is uniform in the z direction. Structural eigensolutions are analytically obtained using the transfer matrix method. Then modal expansion is used to determine radial velocities for a given harmonic force fr(t). In the first acoustic radiation analysis method, the far-field sound pressure is calculated from the Rayleigh integral formula. In the second method, pressures are obtained by treating sound radiating surfaces as two cylindrical radiators of 66 finite lengths. Then Fourier series and Sinc function approaches are applied. The problem is also analyzed using numerical (finite and boundary element) codes. Relevant computational and analytical results are compared with vibro-acoustic measurements. Acoustically efficient radial structural modes may be determined based on modal radiation efficiency and acoustic power calculations. Strategies for minimization of sound radiation are also investigated by parametrically varying disk geometry and natural frequencies. 3.3 Structural Analysis For in-plane vibration analysis of a thick annular disk, the cylindrical coordinate system (r, ϕ, z) is employed where z is normal to the disk as shown in Figure 3.1. Normal and shear stresses in the z direction (σzz = σrz = σϕz = 0) are ignored. The equations of motion corresponding to the plane stress condition can be obtained from the following: ∂σ rr 1 ∂σrϕ σrr − σϕϕ + + = ρ d u r ∂ϕ r ∂r (3.1) 1 ∂σϕϕ ∂σ rϕ 2 + + σ rϕ = ρd v r ∂ϕ r ∂r (3.2) u 1 ∂v E ∂u + + ν 1 − ν 2 ∂r r r ∂ϕ (3.3) σ ϕϕ = E ∂u u 1 ∂v + + ν 1 − ν 2 ∂r r r ∂ϕ (3.4) σ rϕ = E 1 ∂u ∂v v + − 2(1 + ν ) r ∂ϕ ∂r r (3.5) σ rr = 67 fr(t) u r ϕ b z a v h Figure 3.1: A thick annular disk with a radial force 68 Outer radius (a) 151.5 mm Inner radius (b) 87.5 mm Radii ratio (β = b/a) 0.54 Thickness (h) 31.5 mm Density (ρd) 7905.9 Kg/m3 Young’s Modulus (E) 218 GPa Poisson’s Ratio (ν) 0.305 Table 3.1: Disk dimensions and material properties where u and v are radial and circumferential displacements, σrr and σϕϕ are the normal stresses and σrϕ is the shear stress. The components of in-plane force can be defined from u and v as ∂u ν ∂v , N r = D + u + ∂ϕ ∂r r 1 ∂u ∂v ∂u + ν , N ϕ = D u + ∂r ∂ϕ ∂r r N rϕ = N ϕr = (3.6a-c) (1 − ν) D ∂v 1 ∂u − v , + 2 ∂r r ∂ϕ where Nr and Nϕ are the normal forces in radial and tangential direction and Nrϕ is the shear force. Also refer to the list of symbols given at the end of this chapter for the identification of symbols. In case of the qth radial mode at natural frequency ωq (rad/s), 69 equations (3.1) and (3.2) may be rearranged to yield following equations in dimensionless { form, where {z(ξ)} = u , v, Nrϕ , Nr } T and [U(ξ)] are the state vector and the coefficient matrix respectively. d {z (ξ)}= [U (ξ)]{z (ξ)} dξ ν − ξ q ξ [U (ξ)]= q(1 − ν 2 ) ξ2 (1 − ν 2 ) 2 − λq 2 ξ qν ξ 1 ξ 2 q (1 − ν 2 ) 2 − λq 2 ξ q (1 − ν 2 ) ξ2 − (3.7) 0 2 1− ν 2 − ξ q − ξ 0 qν ξ 1− ν − ξ 1 (3.8) u = au cos qϕ, v = av sin qϕ, Nr = DNr cos qϕ, Nϕ = DNϕ cos qϕ, Nrϕ = DNrϕ sin qϕ, Nϕr = DNϕr sin qϕ, (3.9a-i) ξ = r / a, λq = (2ρd La2ω2q / D)1 / 2 , D = Eh /(1 − ν2 ) In equation (3.8), the non-dimensional frequency parameter λq is introduced. One may define the transfer matrix [T(ξ)] in the following form, where ξ and β = b / a represent normalized radial coordinates at an arbitrary radial location and at the inner edge respectively. {z (ξ)}= [T (ξ)]{z(β)} (3.10) Using equation (3.7) and (3.10), transfer matrix at any given frequency can be obtained from the following relation. 70 ξ ξ ξ [T (ξ)]= exp ∫β [U (ξ′)]dξ′ = [I ]+ 1 ∫β [U (ξ′)]dξ′ + 1 ∫β [U (ξ′)]dξ′ 1! 2! 2 + ⋅⋅ (11) By applying the boundary conditions at ξ = β and ξ = 1, [T(ξ)] of equation (3.10) is reduced to a 2×2 sub-matrix. In our study, we employ the procedure of reference [3.4] to find the eigensolutions. Given the free boundary conditions at both radial edges, N r = N rϕ = 0 , equation (3.9) can be simplified as follows: T31 T41 T32 u 0 = T42 v ξ =β 0 (3.12) where Tij in the sub-matrix is the (i, j)th element of the original [T(ξ)] matrix. The characteristic equation L(λ) is obtained from equation (3.12) and eigenvalues λq for inplane modes may be determined from the relation L(λ) = 0. Typical L(λ) curves for q = 2 and q = 3 modes are plotted in Figure 3.2. The analytical solutions obtained by the transfer matrix method are compared in Table 3.2 with the results of numerical (finite element) analysis and structural modal experiment. In the finite element analysis, 6 inplane modes (including 2 circumferential modes) are obtained with a model that includes 4,400 solid elements and 6,600 nodes [3.10]. In modal experiments, the radial excitation force f(t) is applied by an impulse hammer at ϕ = 0° in the mid-plane of the disk. Natural frequencies (ωq) and modal damping ratios (ςq) are extracted from structural frequency response functions such as u /f(ω) where u is the acceleration and f is the applied force. The upper frequency limit of finite element analysis and modal experiment is 16 kHz. Excellent agreement is evident between analytical, numerical, and experimental 71 eigensolutions. According to the assumptions given earlier, radial displacement of two edges for the qth mode can be expressed as follows where |uq| is the displacement amplitude and ωq is the corresponding angular frequency. uq (ϕ, t ) = U q (ϕ)e iω q t = uq cos(qϕ)e iω q t (3.13) The above expression is consistent with the assumption of radial motion only and further analyses of sound radiation are executed based on this equation. Radial component of the computed mode shapes, from the finite element analysis, are compared in Figure 3.3 with the assumed analytical solution given above. Considering the symmetry of modes, only half of the vibration patterns are plotted in this figure from φ = 0° to φ = 180°. To calculate the far-field sound pressure from the disk vibration, Uq for a specific excitation should be defined. The harmonic response of the disk for a point force f r (t ) = Feiωt at rf = (rf, ϕf) can be expressed as u (r , ϕ; t ) = U (r , ϕ; ω)e −iωt . In frequency domain, U (r , ϕ; ω) can be theoretically calculated via the modal expansion as follows: F ∞ Φ q (rf , ϕ f )Φ q (r , ϕ) U (r , ϕ; ω) = − ∑ ρ d h q =1 (ω2 − ω2q ) + i 2ς q (ω / ωq ) T (3.14) where Φq is the eigenfunction and ςq is the modal damping ratio. In our study, the displacement amplitudes are obtained from a numerical synthesis of forced vibration response based on finite element results. Computed natural frequencies and mode shapes are combined with experimentally obtained ςq values for modal expansion. To find the frequency response function, unit harmonic concentrated force f r (t ) = e iω q t is applied to the mid-plane in the radial direction at ϕ = 0° (see Figure 3.1). 72 at a given ωq Figure 3.2: L(λ) for q = 2 and q = 3 modes. Key: –––, q = 2 mode; - - -, q = 3 mode. Figure 3.3: Comparison of radial mode shapes Key: solid line, analytical solution given by equation (3.13); discrete point, finite element analysis. 73 Non-dimensional Frequency λq = ωq(ρdha2 /D)1/2 Mode q Transfer Matrix Finite Element Experiment 2 0.493 0.498 0.489 3 1.193 1.207 1.185 0 1.296 1.271 1.254 Table 3.2: Comparison of disk eigen-solutions for radial modes Mode q Frequency (kHz) |uq| (µm) 2 2.86 0.063 3 7.00 0.028 0 7.26 0.015 Table 3.3: Modal displacement amplitudes used for acoustic analysis 74 The result of forced vibration analysis is given in Table 3.3. The structural frequency response functions u /f(ω) for specific locations are also obtained via numerical analysis. Figure 3.4 compares computed and measured u /f(ω) spectra and a good agreement over the given frequency range is observed. Each dominant peak in this figure corresponds to radial mode whose frequencies are given in Table 3.2. (dB re 20 µm/s2 N) 150 125 100 75 1.6 3.2 4.8 Frequency (kHz) 6.4 8 Figure 3.4: Structural frequency response functions ü/f(ω) at j = 0. Key: ––––, measured; - - - -, computed using BEM. 75 3.4 Acoustic Radiation Model The sound pressure at the observation location p(rp;t) can be expressed as P(rp) e-iωt where P(rp) is the spatially-dependent pressure at ω. With reference to Figure 3.5, P(rp) in the far and free fields due to a vibrating structure with normal surface acceleration U (rs ) at source point rs can be expressed by the Helmholtz integral equation [3.11]. ∂g P(rp ) = − ∫ P + ρ0U (rs ) g dS (rs ) ss ∂γ (3.15) In this equation, g is the free space Green’s function, ρ0 is the medium density and Ss is source surface (see Figure 3.5). The first and second terms in equation (3.15) represent the sound pressure generated at rp by the surface pressure at rs and surface acceleration ü at rs respectively. If the field point is sufficiently far from the source (k|rp| >> 1), one can express the amplitude of acoustic particle velocity of as P/ρ0c0 where k = ω/c0 is the acoustic wave number and c0 is the sound speed of the medium. Furthermore, since particle velocity is in-phase with the sound pressure in the far-field, the sound intensity at the same location can be uniquely defined as I = P2/2ρ0c0. The sound power Π(ω) from a structure vibrating with ω can be found by integrating the far-field sound intensity over the surface Sv that surrounds the source. Acoustic radiation resistance ℜ(ω) can then be obtained from Π(ω) and spatially-averaged mean square radial velocity < u 2 > t , s as follows [3.12]: ℜ(ω) = Π (ω) = σ(ω)ρ 0 c0 As < u 2 > t , s 76 (3.16) Sv ü γ V rp rs rp rs z Ss Figure 3.5: Spherical sound radiation from a vibrating disk. 77 where σ(ω) is the acoustic radiation efficiency, As is the area of the radiator, and < >t,s implies temporal and spatial averages. 3.5 Method I: The Rayleigh Integral Approach Without restrictions on the source configuration and frequency range, the surface pressure distribution must be obtained through a numerical calculation. If the frequency range is restricted to short-wavelength limit, the solution of original Helmholtz integral equation can be circumvented. With this assumption, equation (3.15) can be simplified as the following equation that can be solved without using a numerical method [3.11]. ik r − r ρ0ck e p s Ur (rs ) P ( rp ) = (1 + cosγ )dS (rs ) 4π ∫S s rp − rs (3.17) In our study, sound radiation from the qth radial mode of a thick annular disk is calculated by assuming the unbaffled condition. Using the mode shape of equation (3.13), normal acceleration on the two radial surfaces is expressed as follows: − iω t − iω t − iω t uq (ϕ, t ) = U (ϕ)e q = uq cos( qϕ)e q = −ω q2 u q cos(qϕ) exp e q (3.18) If the sound generating surfaces are discretized into small elements dS of constant üq, P(rp) from the structure can be easily calculated using equation (3.17). For the annular disk case, dS(rs) in equation (3.17) can be expressed by dS(rs) = 2πa dϕ dz and dS(rs) = 2πb dϕ dz for the outer (r = a) and inner (r = b) radial surfaces respectively. The total sound pressure P at rp can be calculated by integrating sound pressure generated by each element over the entire source surface. In this study, numerical integration is used to solve for the sound pressure distribution. 78 p(R, θ, φ) z rp θ R y h ϕ x Figure 3.6: Sound Radiation from the radial vibration of a thick annular disk in spherical coordinate system. 79 The size of dS element should be selected according to the frequency of vibration. If characteristic dimension of the element is larger than 0.5λ, where λ = 2π/k is the acoustic wavelength, the far-field sound pressure will have some errors and consequently acoustic radiation properties including the directivity patterns will be distorted. In our study, observation positions are defined by a group of points having equal angular increments (∆ϕ, ∆θ) on a sphere that is centered at the disk center. Sound pressures at all of the observation positions are calculated using equation (3.17). With computed far-field sound pressure data, acoustic directivity patterns for all structural modes are obtained. For spherical radiation, the modal directivity function Dq(θ, φ) at ωq can be defined from modal pressure Pq(rp) as follows where R = |rp| is the radius of sphere on which observation positions are defined. Pq ( R, θ, φ) = e ik q R R D q ( θ, φ) (3.19) From the far-field approximation, modal power Πq for the qth radial mode is also calculated from modal pressures on a sphere surrounding the disk by using the following equation where θ and φ are the cone and azimuthal angles of the observation positions. 2 Π q = I sq S s = 1 2 π π Pq R 2 sin θ dθ dφ ∫ ∫ 0 0 2 ρ0c0 (3.20) Modal acoustic radiation resistance ℜq is calculated using equation (3.16) where < u q 2 > t ,s = h / 2 2π( a+b) 1 2 U q dl dz ∫ ∫ h / 2 0 − 4πh(a + b) (3.21) Based on equation (3.16), the modal radiation efficiency σq of an annular disk is determined as follows where 2πh(a + b) is the total area of radiating surfaces. 80 σq = ℜq 2ρ 0 c0 πh ( a + b) (3.22) 3.6 Method II: Cylindrical Radiator 3.6.1 Formulation Outer and inner radial surfaces of the annular disk are treated as two separate cylindrical radiators of finite lengths that have uniform surface velocity amplitudes in the thickness direction (z in Figure 3.7). Sound radiation from a cylindrical radiator has been analytically studied by several researchers [3.11, 3.13-3.14]. Junger and Feit developed expressions for far-field sound pressures for finite and infinite cylindrical radiators given arbitrary surface velocity distribution [3.11]. Williams et al. [3.13] used semi-analytical method with finite series of eigenfunctions for boundary condition to calculate the acoustic radiation from a finite cylinder. Sandman [3.14] investigated sound radiation from finite cylindrical shells and found that cylindrical baffle has very little influence on sound radiation and concluded that the baffled cylindrical geometry may be assumed to be a reasonable approximation for this problem. Stepanishen [3.15] combined Green’s function and Fourier integral technique to develop integral expressions for the generalized radiation impedance and radiated power and applied it to an infinite cylinder. Williams solved the same problem using a 2-dimensional Fourier transform [3.16]. Finally, Wang and Lai calculated the modal-averaged radiation efficiency of a finite length circular cylindrical shell [3.17]. In our approach, the far-field sound pressure has been calculated based on the procedure proposed by Junger and Feit [3.11] along with the approximation suggested by Sandman [3.14]. 81 p(rp, φp, zp) zp rp φp z h Figure 3.7: Cylindrical radiator of length h using cylindrical coordinate system. 82 Therefore, we briefly summarize this procedure in order to clarify the proposed method. Junger and Feit [3.11] analyzed a cylindrical radiator of length h that has arbitrary acceleration distribution Z(z) in the z direction and a sinusoidal distribution (cosqϕ) in the circumferential direction ϕ. Modal surface accelerations can be expressed in the cylindrical coordinate system excluding time dependency as Ur ( z, ϕ) = ur Z ( z ) cos nϕ (3.23) If the Fourier transform [ℑ] in z direction is applied to the Helmholtz equation that is expressed in cylindrical coordinates, the partial differential equation governing the ~ Fourier transformed pressure P ( r, φ; k z ) = ℑ[P( r, φ, z )] can be obtained as follows: ∂2 1 ∂ 1 ∂2 ~ 2 2 2 + + k − k z + 2 2 P (r , φ; k z ) = 0 r r r r ∂φ ∂ ∂ (3.24) where kz is structural wave number in the z direction. The solution to this equation can be assumed as ~ P (r , φ ; k z ) = AH n1 [(k 2 − k z2 ) 1 / 2 r ] cos nφ (3.25) where H n1 is the Hankel function of order n. The coefficient A in this equation can be obtained from boundary conditions at r = a and r = b. ~ ∂P (r , φ; k z ) ~ = −ρ 0 u Z (k z ) cos nφ, at r = a and r = b, ∂r h/2 ~ where Z(k z ) = ∫ Z ( z )e −ik z z dz (3.26a-b) −h / 2 Consequently, A can be expressed as follows where r̂ is the radius of given boundary surface. 83 ~ − ρ0 u Z (k z ) A= ′ 2 2 (k 2 − k z )1 / 2 H n1 [(k 2 − k z )1 / 2 rˆ] (3.26c) ( [ ]) ~ Next, P(r, φ, z) can be calculated by taking the inverse Fourier transform P = ℑ −1 P of equation (3.25) and the final equation is given as ~ ik z 1 2 2 1/ 2 ∞ e z H [(k − k ) ρ0 r ]Z (k z )dk z n z P(r , φ, z ) = − u cos nφ∫ −∞ ′ 2 2 2π (k 2 − k )1 / 2 H 1 [(k 2 − k )1 / 2 rˆ] z n (3.27) z After transforming this equation into spherical coordinates by setting r = Rsinθ and z = Rcosθ, the following expression is obtained using the stationary phase approximation. ~ ρ 0 e ikR Z (k z )(−i) n +1 cos nφ P ( R, θ, φ) = u πkR sin θ H 1′ (krˆ sin θ) n (3.28) As a special case, if Z(z) is a simple sinusoidal function as equation (3.29), Fourier ~ transform Z (k z ) and P ( R, θ, φ) can be obtained as equations (3.30) and (3.31) where km = (1+2m)π / h for m =0, 1, 2,⋅ ⋅ ⋅. Z ( z ) = cos km z, z <h/2 = 0, z > h/2 2k ( −1) m cos(k z h / 2) ~ Z (k z ) = m 2 2 km − kz P ( R, θ, φ) = n +1 2ρ0eikR km ( −1) m cos(kh cos θ / 2) u (−i ) cos nφ ′ πkR sin θ(k m2 − k 2 cos 2 θ) H n1 ( krˆ sin θ) 84 (3.29) (3.30) (3.31) 3.6.2 Sinc Function Approach Next consider a thick annular disk case in which the modal surface acceleration of two radial surfaces is given as Uq (ϕ) = uq cos(qϕ) . In this case, variation in the z direction can be expressed via a square pulse (rectangular) function. Accordingly, Z(z) can be expressed as equation (3.32) and the Fourier transform of this equation is obtained as equation (3.33). Z ( z ) = 1, z < h/2 = 0, z > h/2 sin(k z h / 2) sin(k z h / 2) ~ Z ( k z ) = ℑ[ Z ( z )] = 2 =h = h Sinc( k z h / 2) kz kzh / 2 (3.32) (3.33) where the Sinc function is defined as Sinc(x) = sin(x)/x. Sound pressure PqO from the outer radial surface and PqI from the inner radial surface are generated by diverging and converging waves respectively. These are expressed by the Hankel functions of the first and the second kind respectively [3.16]. Therefore, using equations (3.28) and (3.33), we get ik R Sinc ( k q sin θ h / 2)( −i ) q +1 ρ0e q PqO ( R, θ, φ) = uq h cos qφ 1′ πk q R sin θ H (k a sin θ) (3.34a) ik R Sinc ( k q sin θ h / 2)( −i ) q +1 ρ0e q PqI ( R, θ, φ) = uq h cos qφ 2′ πk q R sin θ H q ( k q b sin θ) (3.34b) q q And, the total modal sound pressure is the sum of sound pressures from two radial edges as follows. 85 Pq ( R, θ, φ) = PqI ( R, θ, φ) + PqO ( R, θ, φ) (3.35) Other modal radiation properties such as Πq, ℜq and σq are calculated based on equations (3.16), (3.20) and (3.22). 3.6.3 Fourier Series Approach The far-field sound pressure can also be calculated via yet another procedure. In the Fourier series approach the square pulse function Z(z) is approximated by a Fourier series of 15 components. 4 14 1 Z ( z ) = ( ) ∑ (−1) m ( ) cos km z , π m=0 2m + 1 = 0, z < h/2 (3.36) z > h/2 Using equation (3.31), the sound pressures generated by the mth component of equation (3.36) are calculated as PqOm ( R, θ, φ) = Pq Im ( R, θ, φ) = q +1 k m ( −1) m cos( k q h cos θ / 2) uq ( −i ) cos qφ 2 ′ πk q R sin θ( k m2 − k q cos 2 θ) H q1 ( k q a sin θ) (3.37a) q +1 k m (−1) m cos( k q h / 2 cos θ) uq ( −i) cos qφ 2 ′ πk q R sin θ( k m2 − k q cos 2 θ) H q2 ( k q b sin θ) (3.37b) 2ρ 0 e 2ρ 0 e ik q R ik q R where PqOm and PqIm are sound pressure from the outer and inner edges due to the mth component. The total PqO (and likewise PqI) at a given rp is obtained from a summation of PqOms and PqIms as 86 PqO ( R, θ, φ) = 14 ∑P qOm ( R, θ, φ), m =0 PqI ( R, θ, φ) = (3.38a-b) 14 ∑P q Im ( R, θ, φ) m =0 As in the Sinc function approach, Pq, Πq, ℜq, and σq are calculated using equations (3.35), (3.16), (3.20), and (3.22). 3.7 Modal Radiation Results Modal radiation properties such as acoustic frequency response functions P/f(ω) and P / u(ω) , directivity function Dq(θ,φ), Πq, and σq of the sample annular disk (Table 1) are obtained by using analytical methods of Chapters 3.5 and 3.6. Further, the same radiation properties are calculated with an uncoupled, direct, exterior, and unbaffled boundary element analysis [3.18]. In the computational model, 4,400 nodes and 6,600 elements are used to describe the source. With this model, the number of elements per acoustic wavelength exceeds 6, below 8 kHz. And 6,146 acoustic field points and 6,144 elements are defined on the sphere surrounding the disk. The center of this sphere coincides with the disk center. Excitation to this boundary element analysis model is the normal velocity on two radial surfaces that is calculated from the forced vibration analysis as explained in Chapter 3.3. Results of analytical solutions and numerical model are verified by measurement data obtained from a vibro-acoustic experiment that is conducted in an anechoic chamber as shown in Figure 3.8. In this experiment, the disk is excited by an impact hammer (PCB GK291C) in the radial direction at mid-plane of the disk. 87 Microphone p i, i, r p = 303 mm) Anechoic Chamber Impulse Hammer f ° °, r f = 151.5 mm) Accelerometer ür ° °, r u= 151.5 mm) Rotor Signal Conditioning Unit Signal Conditioning Unit FFT Analyzer FFT Analyzer Figure 3.8: Vibro-acoustic experiment used to measure structural u f (ω) and acoustic P/f(ω) frequency response functions and in-plane modal sound radiation. 88 Far-field sound pressures are measured with a 6 mm microphone that is used for out-ofplane modes at the predetermined field points on a circle of 303 mm radius that is centered at the center of the disk in the plane of ϕ = 0° and θ = 90°. Field point mesh for out-of-plane modes is used in this experiment since directivity patterns of in-plane modes have similar symmetries to those of out-of-plane. Force and pressure signals are conditioned and analyzed via a 2-channel dynamic signal analyzer (HP 35670A) to obtain the P/f(ω) spectrum as shown in Figure 3.9. Sample FRFs at two rp locations (R = 303 mm, φ = 0, θ = 0 and π/2) are also calculated using BEM and then compared with measured data from Figure 3.8. Excellent agreement between experimental and numerical results is seen especially in the vicinity of the three peaks at 2.86, 7.00 and 7.26 kHz. These three peaks correspond to the first three modes (q =2, 3, and 0) of Table 3.3. Spectral contents of P/f(ω) of Figure 3.9 depend on the receiver positions rp suggesting that the sound source is highly directive. The relation between surface vibration and Pq is obtained in the form of P u (ω) FRF that is based on the u f (ω) and P/f(ω) data of Figures 3.4 and 3.9. These FRFs for two rp locations are given in the Figure 10. Note that P u (ω) spectra for two locations show considerable differences since P/f(ω) is affected by a highly directive sound field but u f (ω) is controlled only by disk modes. The modal directivity pattern Dq(θ, φ) on the sphere Sv is synthesized from measured Pq(θ, φ) data. All of the analytical methods accurately predict far-field sound pressure distributions. This is illustrated in Table 3.4 along with Figures 3.11 and 3.12 where Pq(θ, φ) results are compared with measured and computed values for q = 2 and q = 3 modes. As shown in Figure 3.12, Pq shows a sinusoidal variation in the φ direction. 89 P/f(dB re 20 µPa/N) 100 75 50 25 P/f(dB re 20 µPa/N) 0 1.6 3.2 4.8 6.4 8 4.8 6.4 8 100 75 50 25 0 1.6 3.2 Frequency (kHz) Figure 3.9: Acoustic frequency response functions P/f(ω). (a) θ = π/2 and φ = 0; (b) θ = 0 and φ = 0. Key: –––, measured; - - -, computed using BEM. 90 (dB re 20 µPas2/m) 100 75 50 25 (dB re 20 µPas2/m) 0 1.6 3.2 4.8 6.4 8 4.8 6.4 8 100 75 50 25 0 1.6 3.2 Frequency (kHz) Figure 3.10: Acoustic frequency response functions P u (ω) . (a) θ = π/2 and φ = 0; (b) θ = 0 and φ = 0. Key: –––, measured; - - -, computed using BEM. 91 q = 2 Mode q = 3 Mode Structural Mode Analytical Method Measured Computed using BEM Table 3.4: Comparison of directivity patterns for selected modes. 92 Normalized Pressure Normalized Pressure 1 0.5 0 0 10 20 30 40 50 60 70 80 90 0 10 20 30 40 50 60 70 80 90 1 0.5 0 θ (Deg) Figure 3.11: Directivity pattern Dq(θ) given φ = 0 and R = 303 mm. (a) q = 2 mode; (b) q = 3 mode. Key: - - -, Rayleigh integral method; , cylindrical radiator method; – - –, computed using BEM; •, measured. 93 Πq, dB re 1 pW Analytical Methods Mode q Measured Computed using BEM Rayleigh Integral Cylindrical Radiator Model Sinc Fourier 2 66.5 66.5 66.8 66.0 66.0 3 68.0 67.5 67.2 67.5 67.5 Measured Computed using BEM σq, dB re 1 Analytical Methods Mode q Rayleigh Integral Cylindrical Radiator Model Sinc Fourier 2 -3.5 -4.0 -2.3 -4.0 -4.0 3 -1.5 -1.0 -2.2 -2.0 -2.0 Table 3.5: Comparison of modal acoustic power and radiation efficiency levels 94 Figure 3.12: Directivity pattern Dq(φ) given θ = π/2 and R = 303 mm. (a) q = 2 mode; (b) q = 3 mode. Key: –––, Analytical and numerical methods; • • •, measured. 95 Also, one can observe from Figure 3.11 that the cylindrical radiator model is more accurate than the Rayleigh integral method. The Πq and σq predictions for the first two radial modes are compared with BEM code and measured results in Table 3.5. In the experimental case, Pq over Sv has been synthesized using the Pq(θ, φ) data to yield Πq and σq. In this process, the Pq(θ, φ) profile is assumed to have a perfect sinusoidal variation in the φ direction based on the results of Figure 3.12. As shown in Table 3.5, acoustic radiation properties obtained using analytical solutions exhibit good agreements with computed results and measurements. Since predictions are within 6 dB for Πq and 13 dB for σq, one may conclude that analytical solutions are sufficiently accurate in calculating Πq and σq. These discrepancies can be explained by the uncertainties in the measurement introduced by frequency and spatial resolutions as explained in Chapter 2.4. 3.8 Parametric Studies Acoustic radiation properties from a vibrating disk depend on its geometry, vibrating mode and frequency, and the ratio of structural to acoustic wave numbers. In our study, effects of the vibration frequency (ωq), radii ratio (β = b/a), and thickness ratio ( h = h / a ) on modal sound radiation are investigated by calculating Πq and σq. In addition, the relative contribution of sound radiated from the inner and outer radial edges is investigated using the same procedure. Note that, in general, natural frequencies and geometric dimensions of any structure are closely related and cannot be independently controlled. Nonetheless, in our study, ωq and h are independently varied to explicitly 96 determine the effect of each on Πq and σq. For all parametric studies, the cylindrical radiator method (Sinc function approach) is used for calculation. First, we investigate the effect of β on the relative contribution of sound from inner and outer radial edges to the total sound pressure. Therefore, we define the amplitude ratio ( p qI p qO ) and phase difference (∠PqI − ∠PqO) as a function of β, using the formulation of Chapter 3.7. In this study, the maximum modal sound pressure locations are chosen as the sample receiver positions (rp). Realistically, β cannot be altered without changing the in-plane natural frequencies. However, in this investigation, the vibration frequency and its amplitude are fixed and then β is altered. The radii ratio is controlled by changing the inner radius (b) while the outer radius (a) is fixed, and the range is 0.01 ≤ β ≤ 1.0 where the nominal value of β is 0.54. Results for two radial modes are summarized in Figure 3.13. In the limiting case of β → 0, sound pressure is not generated from the inner radial edge (PqI = 0). Consequently, p qI p qO = 0 and ∠PqI − ∠PqO = −∠PqO. In the other limiting case where β → 1, one can see from equation (3.34) that p qI = p qO , and ∠PqI = −∠PqO. Next, the effect of β on Πq and σq is investigated. The results of this investigation are summarized in Figure 14 where Πq and σq significantly fluctuate with β. For each mode, Πq and σq show similar patterns with β. But the pattern of q = 3 mode fluctuates more rapidly than that of q = 3 mode with an increase in β. The acoustic wavelength of q = 3 mode is smaller than that of q = 2 mode and consequently interference between sound waves from two surfaces is more sensitive to a change in b. 97 (a) 1.00 p qO 0.75 90 0.50 0 0.25 -90 0.00 0.01 -180 0.25 0.50 ∠pqI − ∠pqO p qI 180 0.75 Radii Ratio (β) p qI p qO 180 0.75 90 0.50 0 0.25 -90 0.00 0.01 -180 0.30 0.54 ∠pqI − ∠pqO (b) 1.00 0.80 Radii Ratio (β) Figure 3.13: Effect of radii ratio on the amplitude ratio and phase difference between PqI and PqO. (a) q = 2 mode; (b) q = 3 mode. Key: ο ο ο, Phase difference; ∆ ∆ ∆, Amplitude Ratio. 98 (a) 5.0 0.8 4.0 0.6 3.0 0.4 2.0 0.2 1.0 0.0 0.01 Acoustic Power(µW) Radiation Efficiency 1.0 0.0 0.25 0.50 0.75 1.00 Radii Ratio(β) (b) 10.0 1.5 7.5 1.0 5.0 0.5 2.5 0.0 0.01 0.0 0.25 0.50 0.75 Acoustic Power (µW) Radiation Efficiency 2.0 1.00 Radii Ratio (β) Figure 3.14: Effect of radii ratio on Πq and σq. (a) q = 2 mode; (b) q = 3 mode. Key: , Acoustic power; ο ο ο, Radiation efficiency. 99 As a limiting case in which β → 0, sound pressure is generated only by the outer edge and the far-field sound pressure can be determined by the PqO term of equation (32a). For the other limiting case where β → 1, the annular disk can be considered as a thin cylinder that can be solved using the method of reference [3.11]. The effect of ωq on radiation is also investigated for q = 2 and 3 modes. In this study, the forced vibration amplitude and geometric dimensions are fixed while ωq is varied by ±1,000 Hz about its nominal value ωq0. Typical variations in Πq and σq are expressed as a function of the frequency ratio ω = ωq/ωq0, as illustrated in Figure 3.15. Though Πq and σq show significant variations with ω , their shapes are mode dependent. Since geometric dimensions (h, a and b) of the disk are fixed in this investigation, the amplitude ratio PqI PqO is constant at any frequency but the relative phase ∠PqI − ∠PqO varies linearly with ω at a given receiver position. Consequently, Πq and σq show sinusoidal variations with the frequency ratio. Finally, the effect of disk thickness (h) is investigated. The thickness ratio ( h = h / a ) is selected as an independent variable and it is varied from 0.02 to 0.6 with a nominal value of h0 = 0.21. Recall that ωq of the disk is independent of h. Results are summarized in Figure 3.16 where one can observe an increase in Πq because of the larger radiation area as h is increased. As one can see from equation (3.34), in the limiting case of h → 0, both PqI and PqO vanish for all modes. Consequently, Πq = 0 and σq cannot be defined for any q. 100 (a) 12.0 0.8 9.0 0.5 6.0 0.3 3.0 0.0 0.65 0.83 1.00 1.17 Acoustic Power(µW) Radiation Efficiency 1.0 0.0 1.35 Frequency Ratio 12.0 1.5 9.0 1.0 6.0 0.5 3.0 0.0 0.86 0.93 1.00 1.07 Acoustic Power (µW) Radiation Efficiency (b) 2.0 0.0 1.14 Frequency Ratio ( ω) Figure 3.15: Effect of ω on Πq and σq. (a) q = 2 mode; (b) q = 3 mode. Key: , Acoustic power; ο ο ο, Radiation efficiency. 101 (a) 20.0 0.9 15.0 0.6 10.0 0.3 5.0 0.0 0.0 0.02 0.10 0.18 0.26 0.34 0.42 0.50 Acoustic power (µW) Radiation Efficiency 1.2 0.58 Thickness Ratio (h ) 8.0 0.8 6.0 0.5 4.0 0.3 2.0 0.0 0.0 0.02 0.10 0.18 0.26 0.34 0.42 0.50 Acoustic Power (µW) Radiation Efficiency (b) 1.0 0.58 Thickness Ratio (h ) Figure 3.16: Effect of thickness ratio on Πq and σq. (a) q = 2 mode; (b) q = 3 mode. Key: , Acoustic power; ο ο ο, Radiation efficiency. 102 In the other limiting case where h → ∞, the annular disk can be treated as an infinite cylinder where PqI does not contribute to the total sound pressure Pq that has uniform and sinusoidal (cosqφ) distributions in the z and circumferential directions respectively. In ~ this case, Z (k z ) in equation (3.27) can be expressed as 2πδ(kz) where δ is the Dirac delta function. The integral can be performed without the stationary phase approximation to yield: Pq (r , φ , z ) = − ρ 0 urq cos qφ H q1 (kr ) ′ kH q1 (ka) (39) This equation is identical to the far-field sound pressure expression from an infinite cylinder with standing wave configuration that is reported by Junger et al. [3.11]. 3.9 Conclusion This Chapter has introduced new analytical solutions for sound radiation from the radial modes of a thick annular disk. Neglecting the effect of scattering, two simplified calculation procedures have been proposed. Based on comparisons with computed (using FEM and BEM) and measured vibro-acoustic data, both analytical solutions are deemed to be sufficiently accurate though the cylindrical radiator method (using either the Sinc function or the Fourier series approach) appears to be better in predicting the modal radiation. Limiting cases of disk radiator have also been evaluated, including the infinite cylinder expression that is identical to one given in the literature. Therefore, the radiation problem, given an arbitrary radial force excitation, can be analytically solved using the 103 structural mode synthesis procedure and the Rayleigh integral or cylindrical radiator method. When the proposed theory for radial modes is combined with the analytical radiation solutions for in-plane circumferential and out of plane flexural modes as well as their interactions, the total sound radiation from a thick annular disk having a multidimensional force excitation can be formulated in an efficient manner. This work is in progress. Also, future research should incorporate other boundary conditions (including the fixed inner edge) as well as structural imperfections within the disk body. Finally, real-life noise control problems such as the brake squeal may be addressed using analytically rigorous procedures. 104 REFERENCES TO CHAPTER 3 3.1 G. BHUTA and J. P. JONES 1971 Journal of the Acoustical Society of America 35 (7), 982-989. Symmetric planar vibrations of a rotating disk. 3.2 J. S. BURDESS, T. WREN and J. N. FAWCETT 1987 Proceeding of the Institution of Mechanical Engineers 201, 37-44. Plane stress vibration in rotating discs. 3.3 S. CHEN and J. L. JHU 1996 Journal of Sound and Vibration 195 (4), 585-593. On the in-plane vibration and stability of a spinning annular disk. 3.4 IRIE, G. YAMADA and Y. MURAMOTO 1984 Journal of Sound and Vibration 97 (1), 171-175 Natural frequencies of in-plane vibration of annular plates. 3.5 W. THOMPSON, JR. 1971 Journal of Sound and Vibration 17 (2), 221-233. The computation of self- and mutual-radiation impedances for annular and elliptical pistons using Bouwkamp integral. 3.6 M. R. LEE and R. SINGH 1994 Journal of the Acoustical Society of America 95 (6) 3311-3323. Analytical formulations for annular disk sound radiation using structural modes. 3.7 H. LEVINE and F. G. LEPPINGTON 1988 Journal of Sound and Vibration 121 (5), 269-275. A note on the acoustic power output of a circular plate. 3.8 W. P. RDZANEK Jr. and Z. ENGEL 2000 Applied Acoustics 60 (5), 29-43. Asymptotic formula for the acoustic power output of a clamped annular plate. 3.9 H. W. WODTKE and J. S. LAMANCUSA 1998 Journal of Sound and Vibration 215 (5), 1145-1163. Sound power minimization of circular plates through damping layer placement. 3.10 I-DEAS User’s manual version 8.2. 2000 SDRC, USA. 3.11 M. C. JUNGER and D. FEIT 1985 Sound, Structures, and Their Interactions. New York: MIT Press. 3.12 C. E. WALLACE 1970 Journal of the Acoustical Society of America 51 (3), 946-952. Radiation resistance of a rectangular panel. 3.13 W. WILLIAMS, N. G. PARKE, D. A. MORAN AND C. H. SHERMAN 1964 Journal of the Acoustical Society of America 36 (12) 2316-2322. Acoustic Radiation from a Finite Cylinder. 3.14 B. E. SANDMAN 1976 Journal of the Acoustical Society of America 60 (6), 12561264. Fluid loading influence coefficients for a finite cylindrical shell. 105 3.15 P. R. STEPANISHEN 1978 Journal of the Acoustical Society of America 63 (2) 328338. Radiation power and radiation loading of cylindrical surfaces with nonuniform velocity distribution. 3.16 E. G. WILLIAMS 1999 Fourier Acoustics. San Diego: Academic Press. 3.17 C. WANG and J. C. S LAI 2000 Journal of Sound and Vibration 232 (2), 431-447. The sound radiation efficiency of finite length acoustically thick circular cylindrical shell under mechanical excitation I: Theoretical analysis. 3.18 SYSNOISE User’s manual Version 5.4. 1999 NIT, Belgium 106 CHAPTER 4 MULTI-MODAL VIBRO-ACOUSTIC RESPONSE 4.1 Introduction In many practical conditions, mechanical excitation is not limited to single direction and frequency does not always coincide with the natural frequency of the component. In these cases, several structural modes are simultaneously excited and sound pressures from individual modes contribute to total sound pressure. Furthermore, these modal sound radiation solutions interact with each other and final acoustic field is affected by these interactions. In addition, total acoustic power should include power from coupling effects between the acoustic field generated by different modes as well as sound powers from individual modes. Consequently, a procedure for the calculation of structural and acoustic responses for multi-modal excitations should be introduced to properly control sound radiation from these practical components. Several studies on this subject have been executed so far. For instance, sound radiation from the multi-modal vibration and the coupling effect of the structural modes has been investigated with analytical modal sound pressure of simple structures such as beams or rectangular plates [4.1-4.3]. Also, exterior acoustic radiation modes of simple beams and baffled finite plates are defined from expressions for far-field acoustic power 107 [4.4-4.5]. Several researchers applied the acoustic radiation mode concept to the active structural acoustic control or the estimation of radiated acoustic power [4.6-4.7]. Multimodal sound radiation and modal coupling effects on the sound radiation from a thin annular disk have also been investigated using the modal expansion technique [4.8]. Contrary to the thin plate cases where total sound radiation can be represented by the radiation from out-of-plane vibrations without a considerable error, sound radiation from in-plane and out-of-plane vibrations should be considered simultaneously to appropriately control the sound radiation from a thick annular disk. In addition, effect of coupling between in-plane and out-of-plane modes should be included along with the effect of coupling within out-of-plane modes. But, neither the multi-modal sound radiation nor modal coupling effect of complex structures such as a thick annular disk has been clearly defined thus far. In this study, structural and acoustic responses of a thick annular disk to an arbitrary multi-modal or multi directional excitation are investigated using modal expansion technique based on structural modal participation factors along with modal sound radiation data defined by analytical solutions that have been introduced in Chapter 2 and 3. In addition, vibro-acoustic response of the sample disk to a multi-points excitation is studied using the matrix of modal participation factors corresponding overall excitations. The effect of relative circumferential distance between two identical harmonic forces are investigated in this study. Finally, the inter-modal coupling effect on the total sound radiation from the disk is investigated along with the effect of coupling within the same type modes using the same procedure. The disk for modal sound radiations described in Figure 2.1 and Table 2.1 is used as an example case for this study. 108 4.2 Assumptions and Objectives Sound radiation from a thick annular disk is conceptually shown in the Figure 4.1 in the context of source and receiver positions. In this study, in-plane and out-of-plane vibrations of a disk are considered simultaneously to completely investigate the vibroacoustic properties of the disk along with their interactions. The scope of this study is strictly limited to the frequency domain analysis of a linear time-invariant (LTI) system with free-free boundaries. Complicating effects such as fluid loading and scattering at the disk edges are not considered. Chief objectives of this study are: (1) Develop analytical and semi-analytical procedures for calculating sound radiation from the disk given multi-modal harmonic excitations. (2) Examine the effect of couplings between in-plane and out-of-plane modes as well as couplings within the same type of modes on the total sound radiation. (3) Investigate the effect of natural frequency separation on the sound radiation. (4) Study sound radiation due to multi-point excitations and investigate effect of circumferential distance between two identical harmonic forces on the sound radiation. 4.3 Vibro-Acoustic Responses to a Harmonic Excitation 4.3.1 Modal Formulation for Structural and Acoustic Responses If a disk is excited by a harmonic excitation with an arbitrary frequency and direction, several modes, including both in-plane and out-of-plane modes, are excited 109 P(R, θ, φ) z Sv W γ Ssn Ssr R U ϕ Figure 4.1: Sound radiation from a vibrating thick annular disk. 110 simultaneously. Based on the modal expansion technique, velocity distribution (v) on the disk surfaces can be expressed in terms of elastic velocity mode shapes of the disk. {v}= {η}T {ΦV } {Φ }= {Φ {η}= {η V V 0 , 2 , −1 0 , 2 , −1 , Φ1V, 0, −1 , ΦV0,3, −1 ,, ΦVm, n , −1 , ΦV−1, −1, 2 , ΦV−1, −1,0 ,, ΦV−1, −1, q , η1, 0, −1 , η0,3, −1 ,, ηm, n , −1 , η−1, −1, 2 , η0, 2, 0 ,, η−1, −1, q } } (4.1) where ΦVm, n, q is a velocity modal vector of the disk that can be expressed as mode shapes multiplied by corresponding natural frequencies and ηm,n,q is corresponding modal participation factor. In this expression, new modal index (m, n, q) has been introduced which is the combination of the out-of-plane mode index (m, n) and radial mode index q. The value –1 in the indices is used to represent null index that means the corresponding velocity mode shape corresponding to that index does not contribute to the newly defined modal velocity vector. For instance, ΦV0, 2, −1 is the velocity mode shape of the (0,2) out-ofplane mode and ΦV−1, −1, 2 is the velocity mode shape of the q = 2 radial mode. Structural modal participation factors due to a harmonic excitation with frequency ω on a given structure can be obtained from the modal data set (natural frequencies, mode shapes, and damping ratios) as follows: ηm, n, q = ∑ Φ m, n, q (rf , ϕ f )Φ m, n, q (r , ϕ) T (1 − ω2 / ωm2 , n , q ) + i 2ς m, n, q (ω / ωm, n , q ) (4.2) Lee and Singh expressed far-field sound pressure from a thin annular disk due to a multimodal excitation using structural modal participation factors and modal sound pressures [4.8]. Applying the same procedure to an acoustic problem, the far-field pressure on a 111 sphere (SV) surrounding the disk due to surface velocity of equation (4.1) can be expressed as follows: P = {η} {Γ} T {Γ}= {Γ0, 2, −1, Γ1,0,−1, Γ0,3, −1,, Γm, n, −1 , Γ−1,−1, 2 , Γ−1, −1,0 ,, Γ−1, −1, q } (4.3) where, Γm,n,q is the modal sound pressure obtained from equation (2.48) and (3.35). Acoustic power (Π) from the disk and corresponding radiation efficiency (σ) due to an arbitrary harmonic excitation f(t) is also calculated from far-field sound pressures on a sphere surrounding the disk as follows: Π = IsS σ= where, < v >t , s = 2 s = 1 2π π P H P 2 R sin θ dθ dφ 2 ∫0 ∫0 ρ0c0 (4.4 a b) Π < v >t , s 2 1 4πh(a + b) + 2π(a 2 − b 2 ) {∫ h/2 ∫ 2π(a +b) −h / 2 0 a 2π U 2 dl dz + ∫ ∫ W 2 dϕ dr b 0 } 4.3.2 Structural Response Utilizing the procedure given in the Section 4.3.1, structural response to a specific multi-modal harmonic excitation has been calculated based on the analytical modal datasets given in Chapter 2 and 3 or numerical eigen-solutions along with experimental modal damping ratios. Analytical responses of the sample disk with free-free boundaries for the normal and radial excitation are validated with numerical analyses and structural experiments. 112 Microphone p i φ = φi, rp = 303 mm) Anechoic Chamber , u Accelerometer w ° °, ru= 151.5 mm) φ Impulse Hammer f ° ° rf = 151.5 mm) Rotor Signal Conditioning Unit FFT Analyzer Signal Conditioning Unit FFT Analyzer Figure 4.2: Vibro-acoustic experimental setup used to measure structural frequency /f(ω) or u /f(ω) and acoustic P/f(ω) frequency response functions. response functions w 113 In the structural experiments, excitation force f(t) is applied on a normal surface in the z direction at the outer edge of the disk (out-of-plane) or on the mid-plane of the outer radial edge in the radial direction (in-plane) by an impulse hammer (PCB GK291C) at ϕ = 0°. The set up for the structural experiment is explained in Figure 4.2. The upper frequency limit and resolution (∆f) of this experiment are set as 16 kHz and 8 Hz respectively. Natural frequencies (ωmn and ωq) and modal damping ratios (ςmn and ζq) are /fn(ω) or u /fr(ω) using extracted from measured structural frequency response function w and u are the accelerations and f is applied force the half-power point method where w in the corresponding directions. In the finite element method (FEM), structural frequency /fn(ω) or u /fr(ω) have been obtained in the frequency range from 0 response function w to 8 kHz with a model that was used in the Chapter 2 and 3 [4.9]. In addition, harmonic ( r , ϕ) and u( r , ϕ) given a unit harmonic force in single direction with acceleration w arbitrary frequency is calculated based on the numerical modal dataset using a forced /fn(ω) spectra for a normal force vibration analysis. Figure 4.3 compares uni-directional w and u /fr(ω) spectra for a radial force via analytical methods with computed and measured spectra for given excitation and accelerometer positions. Excellent agreements among the results from three approaches can be observed for both cases over the given frequency range. Dominant peaks in these figures correspond to natural frequencies for out-of-plane and in-plane modes of the sample disk respectively. 114 (a) 150 (0, 4) /f (dB re 20µm/s2N) (0, 3) (0, 2) (1, 1) (1, 0) 100 50 1 2 3 4 5 6 7 8 Frequency (kHz) (b) 150 ü/f (dB re 20µm/s2N) q=0 q=3 q=2 100 50 2 3 4 5 6 7 8 Frequency (kHz) /f(ω) Figure 4.3: Structural frequency response functions with free-free boundaries (a) w at r = 0.1515 and ϕ = 180°; (b) ü/f(ω) at ϕ = 0. Key: , measured; - - -, computed using FEM; –––, analytical calculation. 115 4.3.3 Acoustic Response Elementary radiation properties such as acoustic frequency response function P/f(ω), acoustic power Π(ω), and radiation efficiency σ(ω) spectra of the sample annular disk (Figure 2.1 and Table 2.1) are obtained by the analytical method given in Section 4.3.1. The same properties are calculated with uncoupled, direct, exterior, and unbaffled boundary element analyses [4.10]. The field point model for modal sound radiation solutions is used for this calculation along with the finite element model that has been used in the numerical structural analyses. Excitations to this boundary element model are the normal velocity distribution on the external surfaces obtained from the forced vibration analysis as explained in Section 4.3.2. Analytical and computational P/f(ω) at two rp locations (R = 303 mm, φ = 0, θ = 0 and π/2) due to single directional force (normal or radial) are verified by the vibro-acoustic experiments conducted in an anechoic chamber as shown in Figures 4.4 and 4.5. The excitation for structural experiment has been used in the acoustic experiment. Far-field sound pressures are measured with a 6 mm microphone (MTS L130C10 with pre-amplifier MTS 130P10) at the predetermined field points on the sphere Sv. P/f(ω) given unit harmonic force in single direction with arbitrary frequency is calculated based on the numerical modal dataset using numerical forced vibration analysis in FEM. Figures 4.4 compares unidirectional acoustic response function P/f(ω) due to a radial force from analytical methods with purely numerical analysis and measured data for two rp locations. Figures 4.5 compares the same data due to a normal force for the same rp locations. Dominant peaks in these figures correspond to natural frequencies of the corresponding modes of the sample disk. 116 (a) P/f (dB re 20µPa/N) 100 50 0 2 3 4 5 6 7 8 5 6 7 8 (b) P/f (dB re 20µPa/N) 100 50 0 2 3 4 Frequency (kHz) Figure 4.4: Acoustic frequency response functions P/f(ω) due to radial excitation. (a) θ = π/2 and φ = 0; (b) θ = 0 and φ = 0 Key: , measured; - - -, computed using BEM; –––, analytical calculation. 117 (a) P/f (dB re 20µPa/N) 100 (0, 3) (0, 2) (1, 1) (1, 0) 50 0 1 2 3 4 5 6 7 8 (b) 100 P/f (dB re 20µPa/N) (0, 4) (1, 0) q=0 50 0 1 2 3 4 5 6 7 8 Frequency (kHz) Figure 4.5: Acoustic frequency response function P/f(ω) given impulsive force excitation f(t) at r = a in the z direction. (a) θ = π/2 and φ = 0; (b) θ = 0 and φ = 0. Key: , measured; - - -, computed using BEM; –––, analytical calculation 118 Excellent agreements among analytical, experimental and numerical results can be found over the given frequency range for both receiver positions, especially in the vicinities natural frequencies of the disk. In addition to acoustic responses for the single directional forces, P(ω), Π(ω) and σ(ω) for the multi-directional harmonic force of f n = f r = 1N at a single location are calculated using the proposed analytical procedure and compared with the corresponding numerical investigation results. The combined force is applied at ϕ = 0 location on the mid-plane of the disk. Measured data have not been included in this study because of practical difficulties. Analytical P(ω) data for the receiver locations that have been used in the cases of uni-directional force are compared with numerical results in Figure 4.6. These spectra have dominant peaks at the every frequency where uni-directional p/f(ω) have peaks as shown in Figures 4.4 and 4.5. Also acoustic power and radiation efficiency spectra from two approaches are given in Figure 4.7. Acoustic responses from analytical calculation match relatively well experimental and numerical data for all cases. 119 (a) P(ω) (dB re 20µPa) 100 50 0 1 2 3 4 6 7 8 5 6 7 8 (b) 100 P(ω) (dB re 20µPa) 5 50 0 1 2 3 4 Frequency (kHz) Figure 4.6: Far-field sound pressure spectra P(ω) due to multi-modal excitation. (a) θ = π/4 and φ = 0; (b) θ = 0 and φ = 0. Key: ––––, analytical calculation; - - -, computed using BEM 120 (a) Π (dB re 1pW) 100 50 0 1 2 3 4 5 6 7 8 5 6 7 8 (b) 1.5 σ 1 0.5 0 1 2 3 4 Frequency (kHz) Figure 4.7: Acoustic radiation functions due to combined harmonic excitation. (a) acoustic power spectra Π(ω); (b) radiation efficiency spectra σ(ω). Key: –––, analytical calculation; - - -, computed using BEM 121 4.3.4 Responses for Multiple Excitations Many practical mechanical components are simultaneously excited by several forces on several different locations with different frequencies. Analytical approach developed in the previous sections can be expanded to this kind of practical problems using additional modal participation factors. If a number of harmonic excitations are applied on the disk simultaneously, surface velocity and far-field sound pressure can be expressed as combination of those for individual excitation as follows: {v}= ∑ {ηi }T {ΦV } i {P}= ∑ {ηi }T {Γ} (4.5) i {η }= {η i i 0 , 2 , −1 , η1i ,0, −1 , η0i ,3, −1 , , ηim , n, −1 , ηi−1, −1, 2 , ηi0, 2, 0 , , ηi−1, −1, q } where {ηi } is the vector of modal participation factors corresponding to the ith excitation. As in the cases of single excitation, acoustic power radiation efficiency can be calculated 2 from far-field sound pressure using equation (4.4). In this case < v >t , s should be calculated from {v} that is obtained using equation (4.5) As an example case, sound radiation from the sample disk due to two identical harmonic forces in normal direction applied at different locations having specific circumferential separation (∆ϕ) on the disk (rad). Figure 4.8 explains sample cases used in this study. As shown in the figure first force is applied at r = 0.1515 m, ϕ = 0 and second force is applied at r = 0.1515 m, ϕ = π/12, π/6, π/4, or π/3 according to the specific case. Far-field sound pressure spectra P(ω) at two receiver positions (R = 303 mm, φ = 0, θ = π/4 and π/2), due to identical harmonic forces at two circumferential 122 locations are calculated using the analytical procedure. This result is summarized in Figure 4.9. In addition, acoustic power spectra Π(ω) for above four cases are calculated from previously obtained far-field sound pressure distributions and compared each other in Figure 4.10 (a). Finally, radiation efficiency spectra σ(ω) corresponding to above four 2 2 cases are calculated from Π(ω) and < v >t , s using equation (4.5). Here, < v >t , s is calculated using equation (4.6) from surface velocity obtained by modal dataset and modal participation factors corresponding to two excitations. As shown in the Figure 4.9 and 4.10, far-field sound radiation is significantly affected by the circumferential separation between two excitations. Also, as one can see from the comparison of this result with sound radiation due to a single excitation that has been introduced in Figures 4.4 and 4.5, sound radiation at a specific frequency can be controlled using additional forces at proper location. For instance, sound radiation at 6.15 kHz or 3.4 kHz can be significantly reduced by adding an identical force with ∆ϕ = π/4 ∆ϕ = π/3 respectively. 123 f2 ∆ϕ f1 Z Disk Figure 4.8: Example cases for calculation of vibro-acoustic responses due to multi-point excitations 124 (a) P(ω) (dB re 20µPa) 100 50 0 1 2 3 4 6 7 8 5 6 7 8 (b) 100 P(ω) (dB re 20µPa) 5 50 0 1 2 3 4 Frequency (kHz) Figure 4.9: Far-filed sound pressure spectra due to two identical harmonic forces with specific circumferential distances. (a) P(ω) at R = 303 mm, φ = 0, θ = π/4; (b) P(ω) at R = 303 mm, φ = 0, θ = π/2. Key: - - -, ∆ϕ = π/12; , ∆ϕ = π/6;. , ∆ϕ = π/4; –––, ∆ϕ = π/3. 125 (a) 90 Π (dB re 1pW) 80 70 60 50 40 1 2 3 4 5 6 7 8 5 6 7 8 (b) 1.5 σ 1 0.5 0 1 2 3 4 Frequency (kHz) Figure 4.10: Acoustic radiation functions due to two identical harmonic forces with specific circumferential distances. (a) acoustic power spectra Π(ω); (b) radiation efficiency spectra σ(ω). Key: - - -, ∆ϕ = π/12; , ∆ϕ = π/6; , ∆ϕ = π/4; –––, ∆ϕ = π/3. 126 4.4 Effects of Structural and Acoustic Modal Coupling on the Acoustic Radiation 4.4.1 Effects of Structural Modal Coupling on the Acoustic Radiation If several structural modes are excited simultaneously, total sound power should include power from coupling effects between the acoustic field generated by different modes as well as sound powers from individual modes. For instance, Keltie and Peng [4.1] found that the coupling between two structural modes is as important as the individual mode when both natural frequencies are much lower than the excitation frequency. Cunefare [4.2] developed a quadratic expression for the radiation efficiency of a beam under multi-modal excitation using the far-field intensity integration technique. Lee and Singh [4.7] investigated modal coupling effects on the sound radiation from a thin annular disk and found that nonzero sound power due to the coupled modes exists only when the coupled modes have the same number of nodal diameters. As explained in Chapter 4.3, overall sound power from a thick annular disk excited by a harmonic force can be expressed as equation (4.4). Consequently, total sound power can be decomposed into two groups: 1) Sound power from self-radiations of individual modes, 2) Sound power from mutual radiation between arbitrary couple of structural modes. In a thick annular disk case, being evident from the equation (4.4), modal coupling effects exist between out-of-plane modes and in-plane modes as well as between any two out-of-plane modes and any two in-plane modes. Sound power generated by the interaction between mode (mi, ni, qi) and mode (mj, nj, qj) can be obtained with following equation. Π m j n j rj mi ni ri R2 = 2 0 c0 2π π/2 0 0 ∫ ∫ Pmi ni ri P * m j n j rj sin θdθdφ 127 (4.6) Here, Π mijni rj i j is sound power from a self-radiation when mi = mj, ni = nj and qi = qj, m n r otherwise Π mijni rj i j is that from a mutual radiation between (mi, ni, qi) and (mj, nj, qj) modes. m n r Acoustic powers due to modal coupling between any two structural modes are calculated using equation (4.6). By repeating this procedure, radiated powers associated with individual modes of the disk can be obtained along with sound powers due to the coupling effects between two structural modes of different types. The results are summarized in Table 4.1. 0,2,-1 1,0,-1 0,3,-1 1,1,-1 0,4,-1 1,2,-1 -1,-1,2 -1,-1,3 -1,-1,0 0,2,-1 2.1E+7 0 0 0 0 1.4E+6 2.6E+7 0 0 1,0,-1 0 9.0E+7 0 0 0 0 0 0 1.0E+7 0,3,-1 0 0 2.0E+8 0 0 0 0 5.3E+7 0 1,1,-1 0 0 0 1.7E+8 0 0 0 0 0 0,4,-1 0 0 0 0 4.7E+8 0 0 0 0 1,2,-1 1.4E+6 0 0 0 0 3.7E+8 9.1E+6 0 0 -1,-1,2 2.6E+7 0 0 0 0 9.1E+6 7.8E+7 0 0 -1,-1,3 0 0 5.3E+7 0 0 0 0 6.0E+8 0 -1,-1,0 0 1.0E+7 0 0 0 0 0 0 1.2E+9 Table 4.1: Self and mutual radiation terms of sound powers between elastic modes of the sample disk. 128 As shown in the table, acoustic power due to the coupling between out-of-plane modes exists only when two modes have the same nodal diameters (ni = nj). Similarly, sound powers from the coupling between two modes of in-plane type with different q numbers are negligible. Finally, sound powers from coupling between in-plane and outof-plane modes exist only when n = q otherwise, coupling powers are negligible. For example, (0, 2) out-of-plane mode has non-zero coupling effects with (1, 2) out-of-plane mode and q = 2 in-plane mode. Also, total acoustic power due to multi-modal excitation can be obtained as the linear combination of the sound powers from self and mutual radiations as follows: Π = ∑∑ ηmi ni qi ηm j n j q j Π mijni qj i j m n q (4.7) 4.4.2 Effects of Structural Natural Frequencies Separation on the Acoustic Radiation In the thick annular disk case, any two normal modes of the same type (in-plane or out-of-plane) have enough natural frequencies separation and generally two modes are not excited simultaneously by a single frequency harmonic excitation. But the natural frequencies separation between in-plane and out-of-plane modes can be very small according to the geometric configuration and material properties of the disk. In this section, the effect of natural frequency separation between any combination of an inplane mode and an out-of-plane mode is investigated using the proposed analytical solutions and the results are compared with the results of boundary element analysis. 129 As shown in equation (4.2), modal participation factors for individual structural mode due to a harmonic excitation are functions of frequency ratio (ω ⁄ ωr) between excitation frequency ω and natural frequency of the given mode ωr. Consequently, acoustic power from a thick annular disk, including powers from self-radiation and mutual radiation, is affected by the structural natural frequencies separation. In the analytical investigation, natural frequency of one of the neighboring modes is changed to adjust natural frequency separation between the neighboring modes while other structural natural frequencies, mode shapes, and modal damping ratios of all the disk modes remain constant. Based on these data, structural and acoustic response to a harmonic excitation with the frequency between the two natural frequencies of the neighboring modes. As a sample case, the natural frequency of the (0, 2) out-of-plane mode is increased to 2.85kHz so that the natural frequency separation between (0, 2) out-of-plane mode and q = 2 in-plane mode is 10 Hz. P/f(ω) at two rp locations (R = 303 mm, φ = 0, θ = 0 and π/2), Π/f(ω), and σ(ω) for the modified case are calculated using the proposed analytical procedure and compared with the results of original case. As shown in Figure 4.11 and 4.12 considerable increases in P/f(ω)s and Π(ω), at the vicinity of modified natural frequency (2.85 kHz). But, radiation spectra σ(ω) at the same frequency range remains almost constant in spite of the modification in natural frequency. 130 P(ω) (dB re 20µPa) 100 50 0 1 2 3 2 3 4 5 6 7 8 4 5 6 7 8 P(ω) (dB re 20µPa) 100 50 0 1 Frequency (kHz) Figure 4.11: Effect of natural frequency separation on P/f(ω). (a) θ = π/2 and φ = 0; (b) θ = 0 and φ = 0. Key: , modified; - - -, original case. 131 Figure 4.12: Effects of natural frequency separation on acoustic radiation functions. (a) acoustic power frequency response functions Π/f(ω); (b) radiation efficiency function σ(ω). Key: , modified; - - -, original case. 132 4.5 Conclusion This study has introduced analytical and semi-analytical procedures for the sound radiation from multi-modal vibration of a thick annular disk. Sound radiation modes from out-of-plane and radial modes of the disk using analytical and numerical investigations. Sound radiation from the disk excited by an arbitrary harmonic force is obtained from modal sound radiation and structural modal participation factors using the modal expansion technique. Based on the comparison with numerical analysis results, it is evident that the proposed procedure has sufficient accuracy in predicting sound radiation from a thick annular disk excited by arbitrary harmonic forces. Sound powers from the modal coupling between two structural modes of the same type (in-plane or out-of-plane mode) as well as between one in-plane mode and one out-of-plane mode have been studied using proposed analytical solutions for sound radiation modes. According to the results of analytical and numerical investigations, sound powers from modal coupling between two out-of-plane modes exist only when two modes have same number of nodal diameters n. In case of radial modes, sound power due to modal coupling doesn’t exist between two modes with different radial mode index q. Furthermore non-zero sound power due to modal coupling between in-plane mode and out-of-plane mode exists only when n = q. If natural frequency separation between two neighboring modes is small enough to excite two modes with significant modal participating factors, both modes significantly contribute to the total sound power when the excitation frequency is close enough to the natural frequencies of the modes. 133 REFERENCES FOR CHAPTER 4 4.1 R. F. KELTIE and H. PENG 1987 ASME Trans. J. Vib. Acoust. Stress Reliabil. Des. 109, 48-53. The effect of modal coupling on the acoustic radiation from panels. 4.2 K. A. CUNEFARE 1991 Journal of the Acoustical Society of America 90(5), 25212529. The minimum multimodal radiation efficiency of baffled finite beams. 4.3 K. A. CUNEFARE 1992 AIAA J. 30, 2819-2828. Effect of modal interaction on sound radiation from vibrating structure. 4.4 K. A. CUNEFARE and M. N. Currey 1994 Journal of the Acoustical Society of America 96(4), 2302-2312. On the exterior acoustic radiation modes of structures. 4.5 M. N. CURREY and K. A. CUNEFARE 1995 Journal of the Acoustical Society of America 98(3), 1570-1580. The radiation modes of a baffled finite plates. 4.6 G. P. Gibbs, R. L. Clark, D. E. Cox and J. S. Vipperman 2000 Journal of the Acoustical Society of America 107(1), 332-339. Radiation modal expansion: Application to active structural acoustic control. 4.7 M. R. Bai and M. Tsao 2002 Journal of the Acoustical Society of America 112(3), 876-883 Estimation of sound power of baffled planar sources using radiation matrices. 4.8 M. R. LEE and R. SINGH 1994 Journal of the Acoustical Society of America 95(6), 3311-3323. Analytical formulations for annular disk sound radiation using structural modes. 4.9 I-DEAS User’s manual version 8.2. 2000 SDRC, USA. 4.10 SYSNOISE User’s manual Version 5.4, NIT, Belgium, 1999 134 CHAPTER 5 APPLICATION TO A BRAKE ROTOR 5.1 Introduction Several structural dynamic models have been used to explain the brake squeal generation mechanism based on the self-excited vibration [5.1-5.2] or modal coupling phenomena [5.3-5.5]. Recently, non-linear transient analysis [5.6, 5.7] and complex eigen-value method [5.8-5.12] have been implemented using commercial finite element software. Also, Dunlap et al. [5.13] investigated brake squeal noises in various frequency ranges using appropriate approaches and concluded that natural frequency separation between coupled flexural and tangential modes is critical in generating high frequency squeal noise. McDaniel and Li [5.14] investigated coupling between in-plane and out-ofplane modes and concluded that the coupling creates vibrational instability that is characterized by power flow through the transverse motion of the rotor. But, most prior studies have focused on the structural dynamics of brake rotors and related components. The acoustic radiation mechanism has not been adequately examined. To fill this void, we investigate sound radiation from a simplified brake rotor. A semi-analytical procedure is proposed that is based on structural eigen-solutions from finite element analysis and analytical modal sound radiation solutions developed for a thick annular disk. 135 hat disk r z b a th h H Figure 5.1: A thick annular disk with a hat structure simulates the brake rotor. Disk is clamped at the inner bolts and free at outer edge. Outer radius (a) Inner radius (b) Radii ratio (β = b/a) Disk thickness (h) Hat thickness (th) Hat height (H) Density (ρd) Young’s modulus (E) Poisson’s ratio (ν) 151.5 mm 87.5 mm 0.54 31.5 mm 6 mm 24 mm 7905.9 Kg/m3 218 GPa 0.305 Table 5.1: Geometric dimensions and material properties of the brake rotor. 136 Figure 5.1 describes the geometric configurations of the rotor example used in this study. In addition, geometric dimensions and material properties are given in Table 5.1. 5.2 Objectives and Assumptions Chief objectives of this chapter are as follows. (1) Develop semi-analytical solutions for sound radiation from modal vibrations of a brake rotor. (2) Employ a modal synthesis procedure to calculate the vibro-acoustic response to an arbitrary harmonic excitation. (3) Validate proposed analytical procedures using computational vibro-acoustic methods. As evident from Figure 5.1 and Table 5.1, disk thickness (h) is not negligible compared to other dimensions of the disk. Consequently, for a complete investigation of the vibroacoustic characteristics of a brake rotor, it is necessary to simultaneously consider both in-plane and out-of-plane vibrations. Primary assumptions are as follows: (1) Structural and acoustic systems are linear time-invariant systems and complicating effects such as fluid loading and acoustic scattering from the disk edges are negligible. (2) Sound is radiated by only the thick annular disk area, and the hat structure of Figure 5.1 does not contribute to the radiated sound pressure. (3) Boundary conditions for the mounting bolts can be accurately simulated by the idea clamped boundaries at the same locations. 5.3 Structural Modal Analysis The structural dynamics of the rotor of Figure 5.1 have been investigated using a finite element model with 2010 solid elements and 3960 nodes [5.15]. 137 Figure 5.2: Finite element model of the brake rotor with 2010 solid elements. 138 a. Out-of-Plane Modes (m, n) Mode no. Freq. (Hz) 1, 2 900 (0, 1) Mode Shape 4, 5 1700 (0, 2) 9,10 3500 (0, 3) 6 1920 l=0 20 7240 q=0 13 4000 (1, 0) - b. In-Plane Modes (l, q) Mode no. Freq. (Hz) Mode Shape c. Combined Modes (m, n and q) Mode no. Freq. (Hz) 7, 8 2500 m = 0, n = 1 q=1 11, 12 3800 m = 1, n = 2 q=2 18, 19 7120 m = 1, n = 3 q=3 + + + Mode Shapes - Table 5.2: Selected structural modes of the brake rotor 139 To simulate realistic boundary conditions, clamped nodal restraints have been applied at the locations of mounting bolts. A schematic of this finite element model is given in Figure 5.2. As many as 44 structural modes up to 16 kHz have been defined in our analysis. Selected natural frequencies and mode shapes are listed in Table 5.2. In our study, structural modes are described using four modal indices such as m (number of nodal circles), n (number of nodal diameters), q (radial mode index) and l (tangential mode index). Furthermore, structural modes of a brake rotor are categorized into three types: out-of-plane modes described by (m, n), in-plane modes given by (l, q), and combined modes given by (m, n and q). As shown in the table, the qth radial modes are always coupled with out-of-plane modes having the same number of nodal diameters (n) as q, due to the hat structure and clamped boundary conditions. 5.4 Sound Radiation from Structural Modes of Brake Rotor As shown in Table 5.2, mode shapes of pure out-of-plane (flexural) modes of a brake rotor can be expressed with the same modes of a generic annular disk with identical geometric dimensions. Consequently, sound radiation from these modes can be expressed in terms of the modal sound radiation for the corresponding annular disk. In this study, far-field sound pressure due to the (m, n)th out-of-plane mode of sample rotor is calculated using equation (5.1) that is introduced in Chapter 2. Refer to Chapter 2 for the identification of symbols. 140 o ( R, θ, φ) Pmn ( R, θ, φ) = (1 + cos θ) Pmns ( R, θ, φ) + (1 − cos θ) Pmn ρ 0 ck mn e ik mn R −ik mn h2 cos θ e cos nφ( −i ) n +1 Β n [w (r )] P ( R, θ, φ) = 2R ρ 0 ck mn e ik mn R ik mn 2h cos θ o Pmn ( R, θ, φ) = − e cos nφ( −i ) n +1 Β n [w ( r )] 2R s mn (5.1) ∞ where Β n [w ( r ) ] = ∫ w ( r ) J n (k r r ) rdr ; k r = k sin θ 0 Sound power (Π) for mode (m, n) is calculated from the far-field sound pressure Pmn using the following equation. 2 Π mn = I mn SV 1 2π π P = ∫ ∫ mn R 2 sin θ dθ dφ 2 0 0 ρ0c0 s (5.2) Here, Imn is the acoustic intensity on a sphere Sv where Sv is the control surface. The modal radiation efficiency σmn of an annular disk is determined from Πmn as follows 2 where < w mn >t , s is the spatially averaged mean-square velocity on the two normal surfaces of an annular disk. σ mn = Π mn 2 < w mn >t , s 2 < w mn >t , s = ; (5.3) 1 a 2π(a 2 − b 2 ) ∫b ∫ 2π 0 W dϕ dr 2 mn Next, we consider combined modes. As one can see in the Table 5.2, the qth radial modes are always coupled with (m = 0, n = q) or (m = 1, n = q) out-of-plane modes except for the q = 0 radial mode. Since the thickness (h) of disk is beyond the thin plate theory limit, sound radiation from in-plane (radial) modes should be included in the calculation of total sound radiation from such modes. Modal velocity distributions on the 141 normal and radial surfaces have been defined from numerically estimated natural frequencies and modes shapes. Consequently, the total far-field sound pressure at a given receiver position is expressed as sum of sound pressure from normal surfaces due to the out-of-plane vibration as given by equation (5.1) and that from radial surfaces due to the radial vibration. Sound pressure from radial surfaces can be calculated by following equation that is introduced in Chapter 3. Pq ( R, θ, φ) = PqI ( R, θ, φ) + PqO ( R, θ, φ) ik R Sinc (k q sin θ h / 2)( −i) q +1 ρ0 e q PqO ( R, θ, φ) = uqO h × cos qφ 1′ πkq R sin θ H q (k q a sin θ) Sinc ( k q sin θ h / 2)(−i ) ρ0e q PqI ( R, θ, φ) = uqI h ′ πkq R sin θ H 2 (k b sin θ) ik R q (5.4) q +1 × cos qφ q Modal acoustic power for combined mode (m, n, q) can be obtained using an equation similar to equation (5.2) from the total sound pressure that is sum of Pmn and Pq. In addition, modal radiation efficiencies for the combined mode are obtained using following equation. σmnq = Π mnq (5.5) 2 < vmnq >t , s 2 In this case, < vmnq >t , s should be calculated and averaged over the entire radiating surface. It can be obtained using the following equation. vmnq 2 = t,s 1 4πh(a + b) + 2π(a 2 − b 2 ) {∫ h/2 ∫ 2 π( a + b) −h / 2 0 142 a 2π 2 U q2 dl dz + ∫ ∫ Wmn dϕ dr b 0 } (5.6) Mode BEM Analytical m =0 n=2 m=1 n=2 q=2 q=0 Figure 5.3: Directivity patterns for selected modes 143 For pure in-plane modes, the far-field sound pressure can be calculated by using only equation (5.4). Acoustic power and radiation efficiency can be obtained by the same expressions as equations (5.2) and (5.5). Since normal surfaces do not contribute to the 2 modal sound pressure, < vq >t , s should be calculated over radial surfaces only using the following equation. vq 2 = t,s h / 2 2 π( a + b) 1 U q2 dl dz ∫ ∫ h / 2 0 − 4πh(a + b) (5.7) The accuracy of this procedure has been confirmed through a comparison with a purely numerical analysis in terms of the directivity patterns, sound powers and radiation efficiencies. In the numerical analysis, velocities on the rotor surface are calculated from predicted natural frequencies and mode shapes. And, sound radiation data have been calculated using uncoupled, direct, exterior, and unbaffled boundary element analyses [5.16]. The analytical directivity patterns for selected modes are compared with numerical prediction in Figure 5.3. As one can see, analytical and numerical directivity patterns for 3 types of modes are consistent with each other. In addition to the directivity patterns, modal sound powers and radiation efficiencies are listed in Table 5.3. Like the directivity patterns, modal sound radiation data from analytical solutions match numerical data quite well. Therefore, analytical modal radiation basis can be used to calculate sound radiation from a brake rotor when it is excited by a multi-directional harmonic force. This procedure is explained in the next section. 144 Mode Efficiency (σ) Power (dB) Indices Type Out-ofPlane Analytical BEM Analytical BEM - 74 72 0.15 0.05 - - 83 86 0.37 0.44 3 - - 92 92 0.77 0.56 - - - 0 0 0 - - - - 0 - 99 99 0.52 0.69 0 1 1 - 87 88 0.40 0.31 1 2 2 - 91 91 0.43 0.34 1 3 3 - 96 95 0.71 0.44 m n q l 0 1 - 0 2 0 In-Plane Combined Table 5.3: Sound powers and radiation efficiencies for selected modes. 145 5.5 Vibro-Acoustic Response to a Multi-Directional Harmonic Force If a rotor is excited by a multi-directional harmonic force of arbitrary frequency, several in-plane and out-of-plane modes are simultaneously excited. Based on the procedure introduced in Chapter 4, velocity distribution (v) on the disk area of a rotor surfaces can be expressed in terms of the structural modes of the disk. {v} = {η}T {Φ V } {Φ }= {Φ V V 1 , Φ V2 , Φ V3 , , Φ V44 {η} = {η1 , η 2 , η3 , , η 44 } } (5.8) where Φj is the jth velocity modal vector of the disk and ηi is the corresponding modal participation factor. Also, the far-field pressure (P) on the sphere (SV) surrounding the rotor due to surface velocity of equation (5.8) is expressed as follows: P = {η} {Γ} T {Γ}= {Γ1 , Γ2 , Γ2 , Γ3 ,, Γ44 } (5.9) where, Γj is the modal sound pressure for the jth mode obtained from equation (5.1) or (5.3). Sound power (Π) from the rotor due to an arbitrary harmonic force (f) is also calculated from far-field sound pressures, on a sphere surrounding the disk, as follows: Π = ISV s = 1 2π π P 2 2 R sin θ dθ dφ 2 ∫0 ∫0 ρ0c (5.10) Corresponding radiation efficiency (σ) is calculated using Π from equation (5.10) and 2 < v > t , s that is obtained from velocity on the total radiating surfaces using an equation similar to equation (5.7). As in the case of modal sound radiation, vibro-acoustic 146 responses to a multi-directional force are obtained using the proposed analytical procedure and compared with prediction of a numerical analysis. In the purely computational analysis, structural responses for the excitation are obtained from the forced vibration analysis with the finite element model and acoustic responses are obtained by boundary element analysis given structural velocities. In this example, unit harmonic forces in the normal and tangential directions are applied in the mid-plane of rotor at ϕ = π/2 as shown in Figure 5.4. Far-field sound pressure spectra p/f(ω) are obtained using above procedure from 0.8 to 8 kHz. Results for two receiver positions rp1 (R = 1m, θ = 0, φ = 0) and rp2 (R = 1m, θ = π/4, φ = 0) are compared with purely numerical results in Figure 5.5. In addition, sound power and radiation efficiency spectra have been determined. A comparison of analytically obtained spectra and numerical results is given in Figure 5.6. As shown in Figure 5.5, p/f(ω) from the analytical approach match purely computational predictions quite well for both cases. Furthermore, these results depend on the receiver positions indicating that the source is highly directive. Also, analytical acoustic power and radiation efficiency spectra show relatively good agreement especially below 5 kHz. As seen from Figures 5.5 and 5.6, the proposed analytical procedure has sufficient accuracy in calculating the sound radiation due to a harmonic force. Sound pressure at a given receiver position, sound power or radiation efficiency for a given harmonic excitation can be easily calculated using this process. Furthermore, this process may be easily extended to multi-location or multi-frequency excitation cases by considering the modal participation factors. Even though numerical structural modal data have been used, experimental data could be utilized instead of numerical data. 147 fn fr Figure 5.4: Finite element model used for forced vibration analysis given a multidirectional harmonic force. 148 100 dB re 20µpa/N 80 60 40 20 0 1 2 3 1 2 3 4 5 6 7 8 6 7 8 100 dB re 20µpa/N 80 60 40 20 0 4 5 Frequency (kHz) Figure 5.5: Far-field sound pressure spectra p/f(ω) for selected receiver positions rp1 and rp2. Key: Analytical &RPSXWHGXVLQJ%(0----. 149 Π (dB re 1pW) 100 80 60 40 20 1 2 3 1 2 3 4 5 6 7 8 6 7 8 1 0.8 σ 0.6 0.4 0.2 0 4 5 Frequency (kHz) Figure 5.6: Acoustic power spectra Π(w) and radiation efficiency spectra σ(w) of the brake rotor. Key: Analytical &RPSXWHGXVLQJ%(0----. 150 5.6 Conclusion Modal sound radiation from a brake rotor is successfully calculated by using a semianalytical procedure based on structural modal response yielded by a numerical code and analytical solution for sound radiation from a generic, thick annular disk. Vibro-acoustic response for a harmonic force is synthesized quite reliably from modal sound radiation data and the structural modal participation factors using the modal expansion technique. This procedure could utilize either computational or experimental modal data such as natural frequencies, mode shapes and modal damping ratios. The modal expansion technique used to calculate sound radiation due to a harmonic excitation could be easily generalized to multi-locations or multi-frequencies excitations. One would, however, need to define additional modal participation factor vectors. Effects of geometric modifications on vibro-acoustic characteristics can be easily investigated using this procedure. Also, this procedure could be combined with pre-developed methods using numerical and experimental approaches. Finally, it is possible to develop vibro-acoustic design guidelines using this procedure. 151 REFERENCES FOR CHAPTER 5 5.1 H. Murakami, N. Tsunada and T. Kitamura, “A Study Concerned with a Mechanism of Disc-Brake Squeal,” SAE Paper # 841233. 5.2 H. Matsui, H. MURAKAMI, H. NAKANISHI and Y. TSUDA, “Analysis of Disc-Brake Squeal,” SAE Paper # 920553. 5.3 W. V. Nack and A. M, Joshi, “Friction Induced Vibration: Brake Moan,” SAE Paper # 951095. 5.4 D. N. Herting, MSC/NASTRAN Advanced Dynamic Analysis User’s Guide, pp. 157-173, 1997 5.5 J. FLINT AND J. HULTÈN 2002 Journal of Sound and Vibration 254 (1), 1-21. Lining-Deformation–Induced Modal Coupling as Squeal Generator in a Distributed Parameter Disc Brake Model. 5.6 Y. K. Hu, and L. I. Nagy, “Brake Squeal Analysis by Using Nonlinear Transient Finite Element Method,” SAE Paper # 971510. 5.7 O. N. Hamzeh, W. W. Tworzydlo, H. J. Chang and S. T. Fryska, “Analysis of Friction-Induced Instabilities in a Simplified Aircraft Brake, SAE Paper # 1999-013404. 5.8 G. D. Liles, “Analysis of Disc Brake Squeal Using Finite Element Methods,” SAE Paper # 891150. 5.9 G. Dihua and J. Dongying, “A Study on Disc Brake Squeal using Finite Element Methods,” SAE Paper # 980597. 5.10 T. Hamabe, I. Yamazaki, K, Yamada, H. Matsui, S. Nakagawa and M. Kawamura, “Study of a Method for Reducing Drum Brake Squeal,” SAE Paper # 1999-010144. 5.11 S. W. Kung, K. B. Dunlap and R. S. Ballinger, “Complex Eigenvalue Analysis for Reducing Low Frequency Squeal,” SAE Paper # 2000-01-0444. 5.12 T. S. Shi, O. Dessouki, T. Warzecha, W. K. Chang, and A. Jaya sundera, “Advances in Complex Eigenvalue Analysis for Brake Noise” SAE Paper # 200101-1603. 5.13 K. B. Dunlap, M. A. Riehle and R. E. Longhouse, “An Investigative Overview of Automotive Disc Brake noise” SAE Paper # 1999-01-0142. 5.14 J. G. McDaniel and X. Li, “Analysis of Instabilities and Power Flow in Brake Systems with Coupled Modes” SAE Paper # 2001-01-1602. 152 5.15 I-DEAS User’s manual version 8.2, SDRC, USA, 2000. 5.16 SYSNOISE User’s manual Version 5.4, NIT, Belgium, 1999. 5.17 M. R. Lee and R. Singh, “Analytical formulations for annular disk sound radiation using structural modes”, Journal of the Acoustical Society of America 95(6), pp. 3311-3323, 1994. 153 CHAPTER 6 CONCLUSION 6.1 Summary This study has resulted in the development of new analytical and semi-analytical procedures for the calculation of modal and multi-modal sound radiation from thick annular disk. In-plane and out-of-plane modal sound radiations of the disk have been obtained using analytical solutions based on cylindrical and modified annular plate models respectively. Multi-modal sound radiation has been calculated from modal sound radiation and structural modal participation factors using modal expansion technique. Accuracy of these analytical solutions has been confirmed by numerical analyses and vibro-acoustic experiments. As a practical example, whole procedure applied to a brake rotor and successfully predicted sound radiation from a brake rotor. In Chapter II, structural eigen-solutions for the out-of-plane modes of a thick annular disk with free-free boundaries have been calculated using both thick and thin plate theories. The differences between two approaches are clarified in terms of natural frequency and mode shapes. New analytical formulation for the modal sound radiation from a thick annular disk has been proposed with the consideration of the disk thickness. 154 Two approaches, analytical and semi-analytical procedures have been developed combining this analytical formulation for modal sound radiation with structural modal response yield by an analytical calculation or a numerical prediction. In addition, the same problem has been solved by a purely numerical procedure in which the disk surface velocity is numerically defined by a finite element model and sound radiation is then obtained using a boundary element model. Also, the effects of radii and thickness ratios on the structural and acoustic radiation characteristics have been investigated using the analytical procedure. Finally, the effect of boundary conditions has been examined using the semi-analytical method. In Chapter III, in-plane vibration of a thick annular disk has been investigated using Transfer Matrix Method. Structural modal characteristics from this method have been confirmed by numerical and experimental results with excellent agreements. Three analytical approaches based on modified cylindrical model – Rayleigh integral method, Fourier series method, Sinc function method – have been employed to calculate sound radiation from the vibration of this type. The results also have been confirmed numerical analysis using finite and boundary element models and vibro-acoustic experiment in an anechoic chamber. Excellent agreements among these results could be obtained. The effects of vibrating frequency and geometric configuration on the modal sound radiation have been investigated through parametric studies. Chapter IV proposes a new semi-analytical procedure for the calculation of sound radiation from a thick annular disk when it is excited by a multi-modal and multidirectional harmonic force. Structural response for a specific force has been calculated by 155 the modal expansion technique from the structural modal dataset for both in-plane (radial) and out-of-plane (flexural) modes that have been obtained using analytical or numerical methods. In addition, acoustic responses for various harmonic excitations have been estimated by the same technique utilizing the structural modal participation factors and the normal radiation solutions representing the far-field sound pressures due to the structural modal vibrations. These methods are confirmed by comparing predicted results for a single frequency excitation with numerical calculations. Based on this procedure, acoustic power and radiation efficiency spectra of the sample disk corresponding to a specific force location and direction are obtained. This study has been extended to multipoints and multi-frequencies excitation cases for greater generality of this procedure. The effects of coupling between structural modes and natural frequency separation on the acoustic radiation are also investigated through the proposed procedure. This study could lead to strategies that would minimize sound radiation from a thick annular disk excited by arbitrary harmonic forces. The sound radiation from a brake rotor has been calculated using the whole procedure proposed in previous chapters. Structural modes of the rotor have been expressed by structural modes of a generic annular disk having similar geometric configuration. Modal sound radiations of a brake rotor have been synthesized using those of corresponding generic annular disk. Vibro-acoustic response of the rotor to a multidirectional harmonic excitation has been calculated using the procedure introduced in Chapter IV. Modal and broadband vibro-acoustic characteristics of a brake rotor could be calculated very efficiently using this procedure. 156 6.2 Contributions This study has significantly advanced the literature (based on the thin plate model) on the vibro-acoustic characteristics of a thick annular disk by including sound radiation from radial edges and the effect of disk thickness. In particular, the following major contributions emerge. 1. Analytical and semi-analytical solutions for sound radiation from modal vibrations of a thick annular disk have been introduced and validated using computational and/or vibro-acoustic experimental methods. In the radial mode case, total sound radiation is expressed as a sum of radiation from two radial edges that have been obtained using the modified cylindrical radiator method. Proposed solution has been successfully compared with the classical approach for several limiting cases. Far-field sound pressures from two normal surfaces are combined by considering the disk thickness effects to obtain the total sound radiation from out-of-plane modes. Proposed theory yields more accurate results than the classical solution based on a thin plate model. Modal sound radiation solution for each mode over the given frequency range has been calculated by using these two analytical solutions. 2. This study critically examined thick and thin plate theories and investigated the effect of rotary inertia and shear deformations on the structural eigensolutions and acoustic sound radiation from flexural (out-of-plane) modes. The effects of geometric configuration on the modal sound radiations have been clarified by studying key normalized parameters such as radii ratio and thickness ratio, based on the proposed analytical solutions. Appropriate geometric modification to control sound radiation at 157 any specific frequency can be proposed based on the results of these parametric studies. 3. Analytical and semi-analytical procedures for calculating sound radiation given multi-modal harmonic excitations have been developed. Broadband vibro-acoustic characteristics and responses to an arbitrary harmonic force excitation vector can be easily calculated using analytical or numerical modal sound radiation solution and structural modal participation factors. These procedures can be applied to a thick annular disk to define sound radiation from couplings between in-plane and out-ofplane modes as well as couplings within the same type of modes. Also, acoustic power from self and mutual radiation could be conveniently calculated using these procedures with a reasonable accuracy. 6.3 Future Research Since analytical and semi-analytical solutions for sound radiation from a thick annular disk have been successfully introduced in this study, application of such solutions to practical components such as gears, brake rotors, or clutches could be the main task of future research work. Some specific issues have been identified and proposed for future studies: 1. Investigate the vibro-acoustic characteristics of annular disks with annular or radial slots. Analytical and semi-analytical solutions and procedures proposed in this work could be used in this investigation. 158 2. Investigate the effect of a hat structure on the vibro-characteristics of a brake rotor through a parametric study using the procedures proposed in our study. This study could suggest design guidelines for a hat structure. 3. Examine the role of material damping and sliding friction on the radiation properties of rotors. 4. Conduct a complete investigation on the brake squeal noise by combining analytical solutions within a large scale computational code. 159 BIBLIOGRAPHY A. W. LEISSA 1969 NASA SP-160 Vibration of Plates. A. W. LEISSA 1987 The Shock and Vibration Digest 19(3), 10-24. Recent Research and plate vibration, 1981-1985. Part 1: Classical theory. A. W. LEISSA 1987 The Shock and Vibration Digest 19(3), 10-24. Recent research and plate vibration, 1981-1985. Part 2: Complicating effects. A. W. LEISSA 1993 Vibrations of Plates, New York: Acoustical Society of America. S. M. Vogel and D. W. Skinner 1965 Journal of Applied Mechanics December, 926-931 Natural frequencies of transversely vibrating uniform annular disk. C. M. WANG and V. THEVENDRAN 1993 Journal of Sound and Vibration 163(1), 137-149 Vibration analysis of annular plates with concentric support using a variant of RayleighRitz method. R. D. MINDLIN 1951 ASME Journal of Applied Mechanics 18, 31-38 Influence of rotatory inertia and shear on the flexural motion of isotropic, elastic plate. R. D. MINDLIN and H. DERESIEWICZ 1954 Journal of Applied Physics 25(10), 1329-1332 Thickness-shear and flexural vibration of a circular disk. O. G. MCGEE, C. S. HUANG and A. W. LEISSA 1995 International Journal of Mechanical Science 37(5), 537-566 Comprehensive exact solutions for free vibrations of thick annular sectorial plates with simply supported radial edges. T. IRIE, G. YAMADA and K. TAKAGI 1982 Transactions of the American Society of Mechanical Engineers, Journal of Applied Mechanics 49, 633-638 Natural frequencies of thick annular plates. K. M. LIEW, Y. XIANG, C. M. WANG AND S. KITIPORNCHAI 1993 Computer Methods in Applied Mechanics and Engineering 110, 301-315 Flexural vibration of shear deformable circular and annular plates on ring support. IRIE, G. YAMADA and Y. MURAMOTO 1984 Journal of Sound and Vibration 97 (1), 171175 Natural frequencies of in-plane vibration of annular plates. 160 G. BHUTA and J. P. JONES 1971 Journal of the Acoustical Society of America 35 (7), 982989. Symmetric planar vibrations of a rotating disk. J. S. BURDESS, T. WREN and J. N. FAWCETT 1987 Proceeding of the Institution of Mechanical Engineers 201, 37-44. Plane stress vibration in rotating discs. S. CHEN and J. L. JHU 1996 Journal of Sound and Vibration 195 (4), 585-593. On the inplane vibration and stability of a spinning annular disk. T. IRIE, G. YAMADA and Y. MURAMOTO 1984 Journal of Sound and Vibration 97 (1), 171-175 Natural frequencies of in-plane vibration of annular plates. W. THOMPSON, JR. 1971 Journal of Sound and Vibration 17 (2), 221-233. The computation of self- and mutual-radiation impedances for annular and elliptical pistons using Bouwkamp integral. M. R. LEE and R. SINGH 1994 Journal of the Acoustical Society of America 95 (6) 33113323. Analytical formulations for annular disk sound radiation using structural modes. H. LEVINE and F. G. LEPPINGTON 1988 Journal of Sound and Vibration 121 (5), 269-275. A note on the acoustic power output of a circular plate. W. P. RDZANEK Jr. and Z. ENGEL 2000 Applied Acoustics 60 (5), 29-43. Asymptotic formula for the acoustic power output of a clamped annular plate. H. W. WODTKE and J. S. LAMANCUSA 1998 Journal of Sound and Vibration 215 (5), 1145-1163. Sound power minimization of circular plates through damping layer placement. W. WILLIAMS, N. G. PARKE, D. A. MORAN AND C. H. SHERMAN 1964 Journal of the Acoustical Society of America 36 (12) 2316-2322. Acoustic Radiation from a Finite Cylinder. B. E. SANDMAN 1976 Journal of the Acoustical Society of America 60 (6), 1256-1264. Fluid loading influence coefficients for a finite cylindrical shell. P. R. STEPANISHEN 1978 Journal of the Acoustical Society of America 63 (2) 328-338. Radiation power and radiation loading of cylindrical surfaces with nonuniform velocity distribution. M. C. JUNGER and D. FEIT 1985 Sound, Structures, and Their Interactions. New York: MIT Press. E. G. WILLIAMS 1999 Fourier Acoustics. San Diego: Academic Press. 161 C. WANG and J. C. S LAI 2000 Journal of Sound and Vibration 232 (2), 431-447. The sound radiation efficiency of finite length acoustically thick circular cylindrical shell under mechanical excitation I: Theoretical analysis. R. F. KELTIE and H. PENG 1987 ASME Trans. J. Vib. Acoust. Stress Reliabil. Des. 109, 48-53. The effect of modal coupling on the acoustic radiation from panels. K. A. CUNEFARE 1991 Journal of the Acoustical Society of America 90(5), 2521-2529. The minimum multimodal radiation efficiency of baffled finite beams. K. A. CUNEFARE 1992 AIAA J. 30, 2819-2828. Effect of modal interaction on sound radiation from vibrating structure. K. A. CUNEFARE and M. N. Currey 1994 Journal of the Acoustical Society of America 96(4), 2302-2312. On the exterior acoustic radiation modes of structures. M. N. CURREY and K. A. CUNEFARE 1995 Journal of the Acoustical Society of America 98(3), 1570-1580. The radiation modes of baffled finite plates. G. P. Gibbs, R. L. Clark, D. E. Cox and J. S. Vipperman 2000 Journal of the Acoustical Society of America 107(1), 332-339. Radiation modal expansion: Application to active structural acoustic control. M. R. Bai and M. Tsao 2002 Journal of the Acoustical Society of America 112(3), 876883 Estimation of sound power of baffled planar sources using radiation matrices. H. MURAKAMI, N. TSUNADA AND T. KITAMURA, “A Study Concerned with a Mechanism of Disc-Brake Squeal,” SAE Paper # 841233. H. MATSUI, H. MURAKAMI, H. NAKANISHI and Y. TSUDA, “Analysis of DiscBrake Squeal,” SAE Paper # 920553. W. V. NACK AND A. M, JOSHI, “Friction Induced Vibration: Brake Moan,” SAE Paper # 951095. J. FLINT AND J. HULTÈN 2002 Journal of Sound and Vibration 254 (1), 1-21. LiningDeformation–Induced Modal Coupling as Squeal Generator in a Distributed Parameter Disc Brake Model. D. N. HERTING, MSC/NASTRAN Advanced Dynamic Analysis User’s Guide, pp. 157173, 1997. Y. K. HU, AND L. I. NAGY, “Brake Squeal Analysis by Using Nonlinear Transient Finite Element Method,” SAE Paper # 971510. 162 O. N. HAMZEH, W. W. TWORZYDLO, H. J. CHANG AND S. T. FRYSKA, “Analysis of Friction-Induced Instabilities in a Simplified Aircraft Brake, SAE Paper # 1999-01-3404. G. D. LILES, “Analysis of Disc Brake Squeal Using Finite Element Methods,” SAE Paper # 891150. G. DIHUA AND J. DONGYING, “A Study on Disc Brake Squeal using Finite Element Methods,” SAE Paper # 980597. T. HAMABE, I. YAMAZAKI, K, YAMADA, H. MATSUI, S. NAKAGAWA AND M. KAWAMURA, “Study of a Method for Reducing Drum Brake Squeal,” SAE Paper # 1999-01-0144. S. W. KUNG, K. B. DUNLAP AND R. S. BALLINGER, “Complex Eigenvalue Analysis for Reducing Low Frequency Squeal,” SAE Paper # 2000-01-0444. T. S. SHI, O. DESSOUKI, T. WARZECHA, W. K. CHANG, AND A. JAYASUNDERA, “Advances in Complex Eigenvalue Analysis for Brake Noise” SAE Paper # 2001-011603. K. B. DUNLAP, M. A. RIEHLE AND R. E. LONGHOUSE, “An Investigative Overview of Automotive Disc Brake noise” SAE Paper # 1999-01-0142. J. G. MCDANIEL AND X. LI, “Analysis of Instabilities and power flow in Brake Systems with Coupled Modes” SAE Paper # 2001-01-1602. 163