A Study of the Air-Structure Interaction in a Concert Flute by Jennifer Froling A Project Submitted to the Graduate Faculty of Rensselaer Polytechnic Institute in Partial Fulfillment of the Requirements for the degree of MASTER OF ENGINEERING IN MECHANICAL ENGINEERING Major Subject: SOLID MECHANICS Approved: _________________________________________ Ernesto Gutierrez-Miravete, Project Adviser Rensselaer Polytechnic Institute Hartford, CT May 2012 i CONTENTS LIST OF TABLES ............................................................................................................ iii LIST OF FIGURES .......................................................................................................... iv LIST OF SYMBOLS ......................................................................................................... v GLOSSARY ..................................................................................................................... vi ABSTRACT .................................................................................................................... vii 1. Introduction…………………………………………………………………………….1 2. Methods………………………………………………………………………………..3 2.1 Modeling Validation Methods…………………………………………………..3 2.2 Material and Thickness Study Methods…………………………………………7 3. Results………………………………………………………………………………...11 3.1 Modeling Validation Results…………………………………………………...11 3.2 Material and Thickness Study Results………………………………………….13 4. Conclusions…………………………………………………………………………...19 5. References…………………………………………………………………………….20 ii LIST OF TABLES Table 2.1 Properties of Pipe Model ................................................................................... 3 Table 2.2 Properties of Air Column Model ....................................................................... 6 Table 2.3 Pipe Material Study Properties ........................................................................ 10 Table 3.1 Pipe Model Mode Results................................................................................ 11 Table 3.2 Air Column Mode Results ............................................................................... 12 Table 3.3 Pipe Material Study Results: Harmonic Pressure Amplitude .......................... 16 Table 3.4 Pipe Thickness Study Results: Harmonic Pressure Amplitude ....................... 17 Table 3.5 Pipe Material Study Results: Harmonic Frequency ........................................ 18 Table 3.6 Pipe Thickness Study Results: Harmonic Frequency ...................................... 18 iii LIST OF FIGURES Figure 1.1 Modern Concert Flute ...................................................................................... 2 Figure 2.1 Material and Stiffness Studies Models ............................................................. 9 Figure 3.1 Input Characteristic Impedance of Air Column ............................................. 12 Figure 3.2 Nodal Pressure Response For Baseline Model .............................................. 14 Figure 3.3 Pressure Response in Exterior Acoustic Mesh at 273 Hz .............................. 14 Figure 3.4 Pressure Response in Air Column Acoustic Mesh at 273 Hz ........................ 15 Figure 3.5 Displacement Magnitude in Pipe Model at 273 Hz ....................................... 15 iv LIST OF SYMBOLS a radius [m] A modal coefficient c wave speed [m/s] E Young’s modulus [Pa] f frequency [Hz] h thickness [m] I area moment of inertia [m3] L length [m] k wave number [1/m] m mode number P pressure [Pa] r volumetric drag [kg/(m3s)] Z impedance [kg/s] Ι³ shear viscosity [Paβs] Ι³B bulk viscosity [Paβs] Κ bulk modulus [Pa] λ wavelength [m] µ mass per unit length [kg/m] ν Poisson ratio ρ density [kg/m3] v GLOSSARY Acoustic A general term referring to the behavior of vibrational energy in a system. Decibel (dB) The unit of measure based on a logarithmic scale used to measure sound pressure level. Eigenfrequency The frequency at which a resonance occurs in a system, or when a small excitation in the system causes a large response. Harmonics A series of steady-state resonances occurring at multiples of fundamental resonance frequency of the system. Impedance The resistance of a system to a harmonic excitation measured in units of force per velocity. For example, when a system is in a state of resonance, its impedance is at a minimum. Mode A term referring to the deformed state of a system at a resonance. Natural Frequency See Eigenfrequency Sound Pressure Level The acoustic pressure in a fluid measured in decibels. Tone Refers to the sound produced by a system, such as a musical instrument, that is being excited at a resonance or the frequency at which this resonance occurs. Wavelength The length of an oscillatory disturbance in a solid or fluid medium. Wavespeed The speed at which an oscillatory disturbance travels in a solid or fluid medium. vi ABSTRACT The objective of this project is to estimate the effect that the tubing material and thickness have on the sound spectra produced by a modern concert flute. Since the sound-producing mechanisms in a flute are quite complex and the number of variables affecting the sound produced are numerous, it was necessary to focus on only these two variables using a simplified approach. This simplified approach involved building a Abaqus finite-element model of a cylindrical pipe with approximately the same geometry as a concert flute and coupling it to acoustic elements, both on the inside and outside of the pipe model. After running several analyses with five different materials and three different pipe thicknesses, the results showed that these parameters have very little effect on the sound produced the pipe. vii 1. Introduction Much attention has been paid to examining the physics of the modern concert flute, shown in Figure 1.1, particularly the effects of the air jet produced by a flutists at the embouchure of the instrument and the shape of tone holes, or the key holes, from which the sound chiefly radiates from the instrument for all notes higher than the lowest note of instrument’s range. A thorough review of the available literature on the subject, however, shows that few attempts have been made to quantify the effect of the material of the flute on the sound quality it produces. Even fewer published studies exist for which finite-elements models were used to study this subject. Several experiments were performed by John W. Coltman in which both musically skilled and unskilled listeners were instructed to identify flutes constructed of silver, copper, and wood played by experienced flutists. Coltman found that the listeners were able to correctly identify the type of flute being played only about one-third of the time in 36 trials [1]. Many flutists, however, claim that there is an appreciable different in the quality of sound produced by flutes of different materials. The most famous flute player in modern history, JeanPierre Rampal, always played a solid gold flute and is reported to have said of the tone of gold flutes (in contrast to silver flutes): “a little darker; the colour is a little warmer, I like it” [2]. Whether or not this reported difference is due to the response of the edge tone produced at the embouchure of the instrument, is due to the vibrations of the body of the instrument, is purely psychological, or has some other explanation is still unknown. This project examines only the possibility of an effect of the structural modes of vibration of the instrument on the sound spectra produced, focusing on the lowest tone that can be played by the instrument, approximately 261 Hz or, known to musicians C4 [3]. This tone corresponds to the fingering configuration in which all of the holes or keys are covered. More than any other fingering configuration, this one causes the flute to behavior most like an open-ended pipe, which can be modeled using the finiteelement method. 1 Modern Concert Flute Figure 1.1 2 2. Methods 2.1 Modeling Validation Methods The first part of this project required proving the ability of an Abaqus model to predict the structural modes of a thin metal pipe with geometry and properties similar to that of a modern concert flute. Most flutes of good quality are made of silver with a wall thickness of 0.00036 meters or approximately 0.014 inches. Concert flutes are also typically 0.612 meters or 24 inches long, measured from the center of the embouchure to the opposite end of the instrument. The material and geometrical parameters used in the calculation of the bending, longitudinal, and torsion modes of the pipe are given in Table 2.1 [4]. Table 2.1 Properties of Pipe Model Parameter Symbol Value L Length of Pipe 0.612 a Radius of Pipe 0.009 h Thickness of Pipe 0.00036 E Young’s Modulus of Silver 8.30E+10 ρpipe Density of Silver 10490 ν Poisson Ratio of Silver 0.37 Units m m m Pa kg/m^3 NA The first five bending modes or natural frequencies of the pipe, fn, were calculated using equation 2.1, where I is the area moment of inertia of the pipe cross-section calculated using equation 2.2, µ is the mass per unit length of the pipe calculated using equation 2.3, and A is the modal coefficient given for a simply-supported beam as 9.87, 39.5, 88.9, 158, and 247 for the first five bending modes [5]. ππ,πππππππ = 3 π΄ πΈπΌ √ 2π ππΏ4 (2.1) πΌ = π π3 β (2.2) π = 2ππβπππππ (2.3) The first three torsion and the first two longitudinal modes or natural frequencies of a simply-supported pipe were calculated using Equations 2.4 and 2.6 where cp is the compression wave speed in the pipe, given in Equation 2.5 and m is the mode number [6]. ππ,π‘πππ πππ ππ π (1 − π) √ = 2πΏ 2 π¬ ππ = √ (π − ππ )πππππ ππ,ππππππππ ππππ = ππ π √(π − ππ ) ππ³ (2.4) (2.5) (2.6) A finite-element model was built of a simply-supported pipe with the same material and geometric properties as given in Table 2.1 using Abaqus CAE/Standard software, version 6.10. The pipe model was constructed using S4R elements, or four-noded quadrilateral shell elements, with lateral boundary constraints applied at both ends. The required mesh refinement was calculated using the flexural wave speed of a plate given in equation 2.7 where the maximum analysis frequency is 4000 Hz. The maximum flexural wave speed was then calculated using Equation 2.8 to determine the minimum wavelength expected during the analysis. According to the Abaqus manuals, it is recommended to use an element size that is approximately one sixth the size of the smallest wavelength expected during the analysis [7]. 4 ππ = √2πππππ₯ βππ (2.7) 4 √12 ππππ = ππ ππππ₯ (2.8) The acoustic modes of a cylindrical waveguide with one end driven by a piston and the opposite end open to the air can be found by calculating the impedance at the input end of the pipe. The modes of the air column inside the pipe were calculated using air properties at 20 degrees Celsius given in Table 2.2 [8]. The wave number, k, was calculated using equation 2.9 for a range of frequencies from 1 to 2000 Hz in 1 Hz increments. The acoustic modes were determined by first calculating the radiation impedance at the end of the pipe, ZmL, using equation 2.10. The impedance at the input end of the pipe, Zm0, was calculated using equation 2.11, where S is the cross-sectional area of the pipe [8]. The input impedance was then plotted as a function of frequency. The minima of the impedance as function of frequency correspond to the acoustic modes of the air column. 5 Table 2.2 Properties of Air Column Model Parameter Symbol Length of Air Column Leff a K ρair cair Radius of Air Column Bulk Modulus of Air Density of Air Velocity of Sound in Air π= Value Units 0.6174 m 0.009 m 1.42E+05 Pa 1.21 kg/m^3 343 m/s 2ππ ππππ 1 πππΏ = (ππππ ππππ π)[ (ππ)2 + π0.6ππ] 4 ππ0 = (ππππ ππππ π)[ πππΏ )+ππ‘ππ(ππΏ) ππππ ππππ π ] πππΏ 1+π( )tan(ππΏ) ππππ ππππ π (2.9) (2.10) ( (2.11) A finite-element model was built in Abaqus using AC3D8, or 8-noded brick acoustic elements, with the same approximate element size as the pipe model, which was more than enough refinement to satisfy the six elements per wavelength guideline. The radiation impedance boundary condition at the end of the pipe was modeled as a small addition to the length of the air column. This effective length of the air column was calculated using equation 2.12. Non-reflecting acoustic impedance boundary conditions were applied at both ends of the air column model. πΏπππ = πΏ + 0.6π 6 (2.12) 2.2 Material and Thickness Study Methods To determine the effect of pipe material and thickness on the acoustic modes of an air column inside the pipe, it was necessary to build a finite-element model with coupled structural and acoustic nodal degrees of freedom and an exterior acoustic element mesh to capture the sound radiation from the pipe. The pipe and air column were modeled using the same geometry, element types, and element refinement as the two models described in section 2.1. The exterior acoustic elements were modeled by creating a larger cylinder to enclose the smaller pipe model and filling the volume between the cylinder and the pipe model with AC3D4, or 4-noded tetrahedral acoustic elements. The minimum required size of the elements was determined in a similar method as the pipe model, using equation 2.13 and the six elements per wavelength guideline. ππππ = ππππ ππππ₯ (2.13) The extent of the exterior cylinder was dependent on the maximum expected wavelength calculated using equation 2.14 and the node and element number analysis maximum of 20,000 imposed by the Abaqus CAE/Standard education/teaching license. Due to these constraints, the analysis range was limited to 250-2000 Hz and four separate models were necessary to calculate the response of the model over this frequency range. The four models are shown in Figure 2.1. The first model was used to analyze the response in the frequency range 250-580 Hz, the second model was used for the frequency range 580-1100 Hz, the third model for 1100-1600 Hz, and the fourth model for 1600-2000 Hz. All analyses were run in 1 Hz increments. Each of the four models contains the same pipe and air column models, but is different in the exterior mesh extent and refinement. The first model, for example, has an extent in the longitudinal direction of 0.423 meters and an approximate maximum element size of 0.16 meters since the wavelengths at the lower frequencies are quite long. The fourth model, however, has an extent in the longitudinal direction of 0.076 and an approximate maximum element size of 0.031 meters since the wavelengths at the higher frequencies are much shorter. 7 The geometry of the cylinder containing the exterior acoustic mesh was based on onefourth the maximum expected wavelength for the radius of the cylinder and one-third the maximum expected wavelength plus the length of the pipe for the length of the cylinder. This is based on guidelines also found in the Abaqus manuals indicating that the exterior acoustic mesh be approximately one-third the maximum wavelength in the direction of the acoustic wave where a more accurate solution is necessary, which in this case is the longitudinal direction [7]. The four models are shown in Figure 2.1. In each of the models, the structural nodes of the pipe were coupled to the air column nodes using tie constraints. The nodes at the ends of the air column were also coupled to the nodes of exterior mesh using tie constraints. Non-reflecting acoustic impedance boundary conditions were applied the outside element faces of the exterior acoustic mesh. ππππ₯ = 8 ππππ ππππ (2.14) Figure 2.1 9 Although the actual damping effects of the pipe walls and air are difficult to determine, it was necessary to introduce a small amount of damping into the model to avoid the pressure response from going to infinity at the acoustic modes of the air column inside the pipe. A volumetric drag coefficient was calculated for each analysis frequency using Equation 2.15 from the Abaqus manual to account for viscous damping in the air [7]. The shear viscosity, Ι³, of air at 20 degrees Celsius is 1.85E-5 Pa*s and the bulk viscosity, Ι³B, is given by equation 2.16 [8],[9]. A damping ratio of 0.005 was added to the pipe material [5]. This damping ratio was assumed to be similar for all the pipe materials analyzed. ππππ 4 π(π) = (2ππ)2 ( )(ππ΅ + π) π«πππ 3 (2.15) ππ΅ = 0.6π (2.16) To perform the material and thickness studies, each of the four models was analyzed seven times. The first analysis, or the baseline analysis, was run using the pipe material and pipe thickness parameters given in Table 2.1. For second set of analyses, the material studies, the pipe material parameters were then modified to reflect that of pure gold, platinum, aluminum, and lead as given in Table 2.3 [10],[11],[12],[13]. For the third set of analyses, the thickness parameters, the pipe material properties were returned to that of silver and the thickness was modified twice, first to that of approximately 0.010 inches and then to 0.017 inches. Table 2.3 Property 3 Pipe Material Study Properties Silver Gold Platinum Aluminum Density, kg/m Young’s Modulus, GPa Poisson Ratio 10490 19300 83 79 0.37 0.44 10 21450 168 0.38 Lead 2700 11340 70 16 0.35 0.44 3. Results 3.1 Modeling Validation Results To determine the structural modes of the pipe finite-element model described in section 2.1, the Lanczos eigensolver in Abaqus CAE/Standard was used. The eigenvalue frequencies, or natural frequencies, from the analysis were then compared to the natural frequencies calculated using analytical methods also described in section 2.1. The comparison between the two sets of results is shown in Table 3.1. The longitudinal and torsion mode frequencies compare very well—the finite-element natural frequencies are within 1% of the calculated frequencies. The bending modes did not compare as well— within 10% of the calculated frequencies, but this was an expected result as the analytical results were only approximate. The finite-element bending natural frequencies are likely more accurate than those calculated using the analytical method. Table 3.1 Pipe Model Mode Results Mode Calculated Natural Frequency (Hz) Abaqus Natural Frequency (Hz) 72.7 Percent Error 1st Bending 75.1 2nd Bending 300.5 288.4 4.0 3rd Bending 676.2 640.1 5.3 4th Bending 1201.9 1117.5 7.0 5th Bending 1878.9 1707.6 9.1 1st Longitudinal 2298.1 2297.9 0.0 2nd Longitudinal 4596.2 4594.4 0.0 1st Torsion 1388.3 1375.2 0.9 2nd Torsion 2776.7 2750.8 0.9 3rd Torsion 4165.0 4127.0 0.9 11 3.2 The impedance of the air column as calculated using the analytical methods described in section 2.1 is shown as a function of frequency in Figure 3.1. The Lanczos eigensolver in Abaqus CAE/Standard was used to determine the natural frequencies of the air column finite-element model also described in section 2.1. The comparison of the natural frequencies calculated using the two methods is shown in Table 3.2. The two methods compare very well—all of the natural frequencies compare within 1%. Input Mechanical Impedance of Air Column 40 20 Zm0, kg/s 0 -20 -40 -60 558 -80 837 1116 1395 1674 1953 279 -100 0 500 1000 Frequency, Hz 1500 2000 Figure 3.1 Table 3.2 Air Column Mode Results Harmonic Calculated Natural Frequency (Hz) Abaqus Natural Frequency (Hz) Percent Error Fundamental Tone st 1 Harmonic 279 277 0.56 558 555 0.57 2nd Harmonic 837 832 0.57 3rd Harmonic 1116 1109 0.58 4th Harmonic 1395 1387 0.60 5th Harmonic 1674 1664 0.61 6th Harmonic 1953 1941 0.63 12 3.2 Material and Thickness Study Results A linear perturbation, steady-state dynamic baseline analysis was performed on the each of the four coupled structural-acoustic models shown in Figure 2.1. The baseline analysis used material properties of silver found in Table 2.1 and a 0.014 inch pipe thickness. A harmonic pressure load of 100 Pa was applied at a node in the center of the cross-section of one end of the air column. This level is an approximation of the typical pressure levels produced by a flute player at the embouchure of the instrument [14]. The output data measured was the pressure, P, at a single node located on the longitudinal axis of the model and on the end of the exterior acoustic mesh boundary closest to the output end of the pipe. This pressure was then plotted as a function of frequency for each of the four models in the frequency range appropriate for the exterior mesh of the model. This pressure response was converted to sound pressure level in decibel (dB) units using equation 3.1, where Pref is equal to 20µPa [8]. SPL = 20log10 (P⁄Pref ) (3.1) The four responses from the four models shown in Figure 2.1 were combined in the same plot to show a continuous pressure response as a function of frequency over the entire frequency range analyzed (see Figure 3.2). A contour plot of nodal pressure for an elevation cut down the centerline of the exterior acoustic mesh at the fundamental harmonic of 273 Hz is shown in Figure 3.3. For clarity, the models of the pipe and air column inside the pipe have been removed in Figure 3.3. At the same frequency a side view of the nodal pressure of air column inside the pipe is shown in Figure 3.4. A contour plot of the displacement magnitude at each node at the same frequency is shown as a side view of the pipe model in Figure 3.5. The scale on the displacement contour plot indicates that very small displacements on the order of 10-8 meters are induced in the pipe at this frequency. 13 Nodal Pressure Response for Baseline Model (Pa) 120 dB, re:20E-6µPa 100 80 Model 1 60 Model 2 40 Model 3 20 Model 4 0 0 500 1000 Frequency, Hz 1500 2000 Figure 3.2 Pressure in Exterior Acoustic Element Mesh at 273 Hz Figure 3.3 14 Pressure in Air Column Acoustic Mesh at 273 Hz Figure 3.4 Displacement Magnitude in Pipe Model at 273 Hz Figure 3.5 15 As described in section 2.2, the models were analyzed six more times after the baseline analysis was complete in order to evaluate the effect of different pipe materials and thicknesses. Differences between the models were evaluated by measuring the frequency that each harmonic occurred, or the frequency of each peak in the pressure response, and the amplitude at each peak in sound pressure level units. These results, as well as the percent differences in the peak amplitudes from the baseline response, are shown in Tables 3.3-3.6. The pressure amplitude at the fundamental tone or resonance and the first six harmonics for the pipes of different materials are shown in Table 3.3. The smallest percent deviation, no more than 0.035% from the baseline analysis, was found using gold material properties. This is an expected result since the properties of gold are very similar to that of silver. The greatest percent deviation, 2.796% from the baseline, was found using lead material properties. This result is likely due to the very low Young’s modulus of lead, 16 GPa, compared to that of silver, 83 GPa. The greatest difference in sound pressure level in all of the material studies, however, is less than 3 decibels. Table 3.3 Pipe Material Study Results: Harmonic Pressure Amplitude (dB,ref: 20µPa) and Percent Difference From Baseline (Silver) Harmonic Fundamental Tone Silver Gold %Δ Platinum %Δ Aluminum %Δ Lead %Δ 90.553 90.581 -0.030 90.302 0.278 90.650 0.106 92.909 -2.601 1st Harmonic 89.687 89.719 -0.035 89.403 0.317 89.792 0.117 92.195 -2.796 2nd Harmonic 96.838 96.824 0.014 96.964 -0.130 96.792 -0.048 95.085 1.810 3rd Harmonic 101.972 101.979 -0.007 101.902 0.069 101.996 0.024 102.384 -0.404 4th Harmonic 99.858 99.855 0.002 99.876 -0.018 99.850 -0.008 99.510 0.349 5th Harmonic 102.118 102.117 0.001 102.130 -0.011 102.113 -0.005 101.868 0.245 6th Harmonic 100.282 100.283 -0.001 100.276 0.006 100.284 0.002 100.250 0.032 16 The pressure amplitude at the fundamental tone and the first six harmonics for the pipes of different thickness are shown in Table 3.4. The greatest percent deviation, 0.23% from the baseline, was found using a thickness of 0.010 inches. A greater difference was found by decreasing the thickness of the pipe than by increasing it. This is similar to the result found in the material studies in that decreasing the stiffness of the pipe causes the most change in the pressure response, although it does not change it by any significant amount. Table 3.4 Pipe Thickness Study Results: Harmonic Pressure Amplitude (dB,ref: 20µPa) and Percent Difference From Baseline (0.015'') Harmonic 0.014'' 0.010'' %Δ 0.017'' Fundamental Tone 90.553 90.738 -0.203 90.470 1st Harmonic 89.687 89.894 -0.230 89.593 2nd Harmonic 96.838 96.738 0.104 96.881 3rd Harmonic 101.972 102.020 -0.047 101.950 4th Harmonic 99.858 99.842 0.016 99.864 5th Harmonic 102.118 102.108 0.010 102.122 6th Harmonic 100.282 100.285 -0.003 100.280 %Δ 0.092 0.105 -0.045 0.022 -0.007 -0.004 0.002 The frequencies at which the fundamental tone and the first six harmonics occur for the pipes of different materials and thickness are shown in Tables 3.5 and 3.6. The only observed change in the fundamental and harmonic frequencies was found in the analysis of the lead pipe at the second harmonic when the resonance frequency decreased from 819 Hz to 818 Hz. It is likely that the resonance frequencies in all of the material and thickness studies shifted a negligible amount less than 1 Hz, which was the smallest frequency increment used during all of the analyses. 17 Table 3.5 Pipe Material Study Results: Harmonic Frequency (Hz) and Percent Difference From Baseline (Silver) Harmonic Silver Gold Platinum Aluminum Lead Fundamental Tone 273 273 273 273 273 1st Harmonic 546 546 546 546 546 2nd Harmonic 819 819 819 819 818 3rd Harmonic 1092 1092 1092 1092 1092 4th Harmonic 1365 1365 1365 1365 1365 5th Harmonic 1638 1638 1638 1638 1638 6th Harmonic 1910 1910 1910 1910 1910 Table 3.6 Pipe Thickness Study Results: Harmonic Frequency (Hz) and Percent Difference From Baseline (0.014’’) 0.014'' 0.010'' 0.017'' Harmonic Fundamental Tone 1st Harmonic 2nd Harmonic 3rd Harmonic 4th Harmonic 5th Harmonic 6th Harmonic 273 546 819 1092 1365 1638 1910 18 273 546 819 1092 1365 1638 1910 273 546 819 1092 1365 1638 1910 4. Conclusions The results of the material and thickness studies show that the properties of the pipe finite-element model have a slight effect on the pressure response produced. The largest difference in the response—almost 3% difference in amplitude—was found when the pipe material was changed to lead. Much smaller differences were found with the other pipe materials and pipe thicknesses that were evaluated. This result indicates that the pipe must be extremely flexible or have very little stiffness for it to have any effect on the modes of the air column inside the pipe. Even the differences found in the response for the lead pipe are negligible and suggest that vibrations in the tubing of a flute do not have any appreciable effect on the tone quality that the flute produces. These studies do not prove that the material and tubing thickness of a flute have no effect on the tone quality of a flute, but that the difference may be found in other mechanisms of the sound production such as the air jet vibration at a flute’s embouchure or the sound radiation from open key holes when notes higher than the fundamental tone are being played. 19 REFERENCES [1] Coltman, John W. Effect of Material on Flute Tone Quality. The Journal of the Acoustical Society of America, Vol 49, Part 2. July 27, 1970. [2] “Jean-Pierre Rampal” http://en.wikipedia.org/wiki/JeanPierre_Rampal. December 16, 2011. [3] “ Frequencies for Equal-Tempered Scale” http://www.phy.mtu.edu/~suits/notefreqs.html. December 16, 2011. [4] “Silver.” http://en.wikipedia.org/wiki/Silver. December 16, 2011. [5] Harris, Cyril M. and Charles E. Crede. Shock and Vibration Handbook Vols. 1-3. McGraw-Hill Book Company, New York: 1961. [6] Junger, Miguel and David Feit. Sound, Structures, and their Interaction 2nd Ed. MIT Press, Cambridge, MA: November, 1986. [7] “Abaqus 6.10 Online Documentation” © Dassault Systemes, 2010. [8] Kinsler, Lawrence E. and Austin R. Frey. Fundamentals of Acoustics 4th Ed. John Wiley and Sons, Hoboken, NJ: December, 1999. [9] Pierce, Allan D. Acoustics: An Introduction to its Physical Principles and Applications. Acoustical Society of America, Woodbury, NY: 1989. [10] “Gold.” http://en.wikipedia.org/wiki/Gold. March 10, 2012. [11] “Platinum.” http://en.wikipedia.org/wiki/Platinum. March 10, 2012. [12] “Aluminum.” http://en.wikipedia.org/wiki/Aluminum. March 10, 2012. [13] “Lead.” http://en.wikipedia.org/wiki/Lead. March 10, 2012. [14] Fletcher, N.H. Air Flow and Sound Generation in Musical Wind Instruments. Annual Review of Fluid Mechanics, Vol 11, 1979. 20