ME. CHAPTER 12 Structural Damping ERIC E, UNGAR Bolt Beranek and Newman, Inc. Cambridge, Massachusetts 12.1 THE EFFECTS OF DAMPING ‘The dynamic responses and sound transmission characteristics of structures are determined by essentially three parameters: mass, stiffness, and damping. Mass and stiffness are associated with storage of kinetic and strain energy, respectively. whereas damping relates 10 the dissipation of energy, or more precisely, to the conversion of the mechanical energy associated with a vibration to a form (usually heat) that is unavailable to the vibration. Damping in essence affects only those vibrational motions that are controlled by a balance of energy in a vibrating structure; vibrational motions that depend on a balance of forces are virtually unaffected by damping. For example, consider the response of a classical mass-spring-dashpot system to a steady sinusoidal force. Ifthis force acts at a frequency that is considerably lower than the system's natural frequency, the response is controlled by a quasi-static balance between the applied force and the spring force. If the applied force acts at a frequency that is consid- erably above the system's natural frequency. the response is controlled by a bal- ance between the applied force and the mass’s inertia. In both of these cases, damping has practically no effect on the responses. However, at resonance, where the excitation frequency matches the natural frequency, the spring and inertia ef- Noise and Vibration Control Engineering: Principles and Applications, Edited by Lea L., Beranck and Istvan L. Ver, ISBN 0-471-61751-2 © 1992 John Wiley & Sons, Ine, 451452 STRUCTURAL DAMPING feets cancel each other and the applied force supplies some energy to the system during each cycle; as a result, the system's energy (and amplitude) increases until steady state is reached, at which time the energy input per cycle is equal to the energy lost per cycle due to damping. In the light of energy considerations like the foregoing, one finds that increased damping results in (1) more rapid decay of unforced vibrations, (2) faster decay of freely propagating structure-bome waves, (3) reduced amplitudes at resonances of structures subject to steady periodic or random excitation with attendant redue- tions in stresses and increases in fatigue life, (4) reduced response to sound and increased sound transmission loss (reduced sound transmission) above the coinci- dence frequency (where the disturbing pressure wave moves along the structure in concert with the structural displacement), (5) reduced rate of buildup of vibrations at resonances, and (6) reduced amplitudes of “*self-excited"’ vibrations, in which the vibrating structure accepts energy from an external source (e.g., wind) as the result of its vibratory motion. 12.2 MEASURES AND MEASUREMENT OF DAMPING ‘Most measures of damping are based on the dynamic responses of simple systems with idealized damping behaviors. Damping measurements typically involve ob- servation of some characteristics of these responses. f Decay of Unforced Vibrations Many aspecis of the behaviors of vibrating systems can be understood in terms of the simple ideal linear mass~spring-dashpot system shown in Fig. 12.1. If this system is displaced by an amount x from its equilibrium position, the massless spring produces a force of magnitude kx tending to restore the mass m toward its ‘equilibrium position, and the massless dashpot produces a retarding force of mag- nitude ci. Here & and c are constants of proportionality: k is known as the spring constant and c as the viscous damping coefficient. If this system is displaced from its equilibrium position by an amount Xq and then released, the resulting displacement varies with time ras! X= Xe" cos(ayt + 9) (12.1) te) ro) F F Fig. 12.1 System with single degree of free x dom: (a) schematic representation of mass- spring-dashpot system or of vibrational mode of structure; (b) Free-body diagram of mass, Spring produces restoring force kx: dashpot produces re- tarding force cx. ES12.2 MEASURES ANO MEASUREMENT OF DAMPING © 453 provided that { < 1, Here @ represents a phase angle, which depends on the velocity with which the mass is released, and w, and cy represent the undamped and damped radian natural frequencies of the system. These obey ik a= fon wy = with f, representing the cyclic (undamped) natural frequency. The constant { is called the damping ratio or fraction of critical damping; it is defined as (12.2) £ c= aim = 2me, (12.3) c where ¢, is known as the critical damping coefficient. For the small values of one usually encounters in practice, «wy is sufficiently close to «,, so that one rarely needs to distinguish between the damped and the undamped natural frequencies. Furthermore, the foregoing expression for «, applies only for viscous damping: other relations hold for other damping models. ‘The right-hand side of Eq, (12.1) represents a cosine function with an amplitude ye" that decreases as time 1 increases (see Fig. 12.2); its rate of decrease is fw, and thus is proportional to ¢. However, Eq, (12.1) no longer applies for values of ¢ that equal or exceed unity (or for values of c that equal or exceed ¢,). For such large values of for ¢ one obtains a monoscillatory decay represented by pure exponential expressions instead of the decaying oscillation represented by Eq. (12.1). The eritical damping coefficient c. constitutes the boundary between os- cillatory and nonoscillatory decays. TUR Pye Fig. 12.2. Time variation of displacement of mass-spring-dashpot system released with zero velocity from initial displacement Xp. Light curve: Undamped system (¢ = 5 =O): amplitude remains constant at Xo. Heavy curve: Damped system (0 < ¢ < 630 < $< 1); amplitude decreases according to x = Xpe ““, which is represented by upper dashed curve. Lower dashed curve corresponds tox = —X,e~*"'. Amplitudes X,and X, , illustrate values that may be used to calculate logarithmic decrement from Eq. (12.4) for N = 24 454 STRUCTURAL DAMPING The logarithmic decrement & is a convenient, time-honored representation of how rapidly a free oscillation decays, It is defined by! x xX (12.4) aw tn dn where X, represents the value of x at any selected peak and X,. represents the ‘alu atthe Peak at \ cycles from the aforementioned one. For a viscously damped system, it follows from Eq. (12.1) that 6 = 2x¢. The utility of logarithmic measures of oscillatory quantities has Tong been rec- ognized in acousties, and definitions analogous to acoustical levels have come ince use in the field of vibrations, particularly in regard to measurement, For example, Tne TAY define the displacement level L,, in decibels, corresponding to an oscil, latory displacement x(r) in analogy to sound pressure level, as t= 10 Weg (12.5) ret Where rer denotes a (constant) reference value of displacement. One may then obtain a decay rate A, in decibels per second, and relate it to the damping ratio for a viscously damped system?: ’ 8.69fu, = 54.62 f, (12.6) Also in analogy to acousties, one may define the reverberation time Tuy as the time ittakes for the displacement level to decrease by 60 decibels: thus, 60 1.10 =—=— 12,7) oa ‘ Because velocity and acceleration levels may be defined in full analogy to the definition of displacement level in Eq. (12.5), the decay rate and reverberation time expressions of Eqs, (12.6) and (12.7) also apply ta these other vibration levels. All of the foregoing measures of damping deal with quantities that characterize the decay of free vibrations, and not with energy as such. However, these quan- tities are related to energy. The total energy W of a mass~spring-dashpot system like that of Fig. 12.1 consists of the kinetic energy Wyiy of the mass aad the po- tential energy Wa, stored in the spring. For a system vibrating with amplitude X at its natural frequency w,, one finds? We Wag + Wag = JkX? mi- 122 MEASURES AND MEASUREMENT OF DAMPING © 455 ‘The energy dissipated by the system of Fig. 12.1 corresponds to the work that is done on the dashpot; the energy D dissipated per cycle of a vibration at frequency w, and amplitude X may readily be found to obey D = os,cX*, The ratio of the energy dissipated per cyc/e to the energy present in the system is called the damping capacity J. For a viscously damped system it obeys y = D/W = 4x. The ratio of the average energy dissipated per radian to the energy in the system is called the loss factor. The loss factor 7 is equal to 1/2+ times the damping capacity and, for a viscously damped system, is related to ¢ as D == 2. "= SW 25 (12.9) The expressions relating ¥ and y to the damping ratio ¢ are based on Eq. (12.8) and on the expression for D as well as on the assumption that the amplitude changes little during a cycle, that is, that the decay is slow and the damping is small. However, the definitions of damping capacity and loss factor are not limited to yiscous damping or to small damping values, If any extended structure that is not too highly damped vibrates in absence of external forces at one of its natural frequencies, all points on that structure move cither in phase or in opposite phase with cach other, and the structure is said to vibrate in one of its modes. In addition to the modal natural frequency, there: cor responds to each mode a modal mass, a modal stiffness, and a modal jdamping value. With the aid of these parameters the behavior of a modal vibration may be described in terms of that of an equivalent simple mass-spring-dashpot system! “5, ‘Thus, all of the foregoing discussion concerning this simple system also applies to stmuctural modes. Of course, extended structures.also can exhibit unforced motions at a given frequency in which all points are not in of out of phase with each other. Such motions, which can be described in terms of propagating waves, also decrease due to damping. For flexural waves on a beam or for nonspreading (straight-crested) flexural waves on a plate, the sparial decay raze A,, defined as the reduction in vibration level per wavelength, obeys’ A, = 13.6m in decibels per wavelength. Steady Forced Vibrations If the system of Fig. 12.1 is subject to a sinusoidal force F = Fycos wr, then the displacement of the mass in the steady state is also sinusoidal and obeys x = Xcos(we — 6) = Re{ Xe*"}, where X = |X| denotes the amplitude of the motion and X_= Ne~* is the corresponding phaser or complex amplitude, with Jj = V=1. From the equation of motion for the system, it follows! that ayo Bl=|(\-) 2)] ° Dw fey Fi/k = X, = 1 wf + te tan dé Tw fot (12.10) —ea | 456 STRUCTURAL DAMPING ‘The deflection X,, = Fy/k, which was introduced in order to obtain a nondimen- sional expression, is the quasi-static or zero-frequency deflection that the system would experience due to a statically applied force of magnitude Fy. The ratio X/X... which indicates by what factor the amplitude under dynamic excitation exceeds the quasi-static deflection, is often called the amplification. ‘The foregoing response expressions (and related ones that involve velocity oracceleration A instead of displacement) depend on damping; thus, damping data may be extracted from corresponding measurements. Two widely used measures of damping may be derived readily from the amplification expressions of Eq. (12.10), a plot of which is shown in Fig. 12.3. One is the amplification at reso- nance, conventionally represented by the letter @ and often simply called “the Q"* of the system.’ This corresponds to the value of X/X,, that results if the excitation frequency w is equal to the natural frequency w, and is related to the viscous damp- ing ratio by Q = 1/2g. The second commonly used measure is the relative band- width b = Aw/o, ~ 1/Q, where Aw represents the difference between the two frequencies* (one below and one above w,; see Fig. 12.3) at which the amplifi- cation is equal to @/ V2. ‘ In view of Eq. (12.10), the phase lag also provides a measure of damping. It is particularly convenient to use phase information in “Nyquist plots," that is, in plots of the real and imaginary parts of responses at a number of frequencies, as illustrated in Fig. 12.4. These plots are circles or nearly circles, with a diameter that is equal to Q if the plots are appropriately nondimensionalized.®” t “These frequencies often are called the half power poids because al these the enerey stored in the system (and that dissipated by it), which is proportional 10 the square of the amplitude [see Eq, (12.6)]. is half of the maxicmam value, o_o AMPLIFICATION X/%yq 0 7 = FREQUENCY RATIO 44/4 Fig. 12.3 Steady-state response of mass-spring-dashpot system to sinusoidal force. See Eg. (12.10): @ = value of amplification X/X,, at resonance. Relative bandwidth b = Au /ay is determined from “‘half-power points,"" i.c., from frequencies at which response ampli- tude is 1 /V2 times the maximum.122 MEASURES AND MEASUREMENT OF DAMPING 457 e) “VISCOUS DAMPING STRUCTURAL DAMPING. Fig. 12.4 Nyquist plots of nondimensional responses of viscously and structurally damped mass~spring-damper systems. (a) Real and imaginary parts of amplification X/X,. (b) Real and imaginary pans of mobility Vk/Fy2, = Ve,/2Fo. Amplification plot for structural damping and mobility plot for viscous damping are exact circles: others are approximate eireles that become more nearly circular with decreased damping. Diameter is exactly or approximately equal to Q, Figure plots comespond to f = 0.2, = 0.4. If the system of Fig. 12.1 (or a structural mode modeled by it) is subject to a broadband force (or broadband modal force) rather than a single-frequency sinu- soidal force, then the mean-square displacement ¥° of the mass is given by® (2.11)458 © STRUCTURAL DAMPING where S,{a) denotes the spectral density of the force in terms of radian frequency (i.c., the value of excitation force squared per unit radian frequency interval). Note that the spectral density in cyclic frequency obeys $4 f) = 2xS,(us). The foregoing equation is exact for excitations with spectral densities that are constant for all frequencies. It is a good approximation for excitations with spectral densities that vary only slowly in the vicinity of the system's natural frequency w,: the spectral density value to be used in the equation then is that corresponding to «, Interrelation among Measures of Damping ‘The previously defined measures are related to each other as follows™”: v _ 2.20 _ ca ae 8 rahe a= = (12.12) The foregoing interrelations have been derived for viscous damping, that is, for damping that may be characterized by a retarding force proportional to velocity. However, they apply approximately also for any type of damping as long as the damping is not too great, say ¢ < 0.1 or q < 0.2, which encompasses most practical structures. Measurement of Damping Most approaches to measurement of the damping of structures are based on the previously discussed responses of simple systems, which, as has been mentioned, also correspond to those of structural modes. However, unlike mass-spring-dash- pot systems, structures have a multiplicity of modes and corresponding natural frequencies. Therefore, many of the approaches applicable to simple systems can be applied only to structural modes whose responses can be separated adequately from those of all others because of differences in their natural frequencies or mode shapes, Measurement of logarithmic decrement & typically is applicable only to the fun- damental modes of structures for which a clear record of the amplitude-versus- time trace can be obtained. If more than one mode is present, their decaying re Sponses are superposed and the record becomes difficult to interpret The counting of peaks that is required for determination of the logarithmic dec rement from Eq. (12.4) is not needed if one focuses on the decaying signal's en velope. For purposes of evaluating this envelope it is particularly convenient (0 use a display of the logarithm of the rectified amplitude versus time. Rectification is needed because the logarithm of negative numbers is undefined. In such a log- arithmic display the envelope becomes a straight line whose slope is proportional to fw,, and thus to the decay rate. Not only does measurement of the slope enable ‘one to evaluate the damping, but also observation of deviations of the envelope from a straight line permit one to judge whether the structure's damping is indeed- 12.2 MEASURES AND MEASUREMENT OF DAMPING 459 amplitude independent and whether a superposition of responses with different decay rates is present, Determination of decay rates is useful also in frequency bands in which a mul- titude of modes are excited. A typical measurement here involves excitation of a structure by a broadband force in a given frequency band, cutting off the excitation, then observing the envelope of the logarithm of the rectified signal obtained by passing the output of a transducer (usually an accelerometer) through a bandpass filter® tuned to the excitation band. The center frequency of this passband may be taken to represent «, for all modes in the band, Same judgment in interpreting the resulting envelopes and averaging of results from repeated measurements is gen- erally required because different modes in the band may exhibit somewhat different decays, A conceptually straightforward approach to measuring the damping of a struc ture involves the application of Eq. (12.9) to observed values of the energy dis- sipation and total vibrational energy W [Eq. (12.8)] present in a structure in the steady state. The structure is excited via an impedance head ora similar transducer arrangement that measures the force and motion at the excitation point. The in- stantaneous force and velocity values are multiplied and the product is time aver~ aged to yield the average energy input per unit time, which is equal to the energy dissipated per unit time under steady-state conditions. For a given excitation fre- quency f, the energy D dissipated per cycle is equal to 1 /f times the-energy dis- sipated per unit time. 1 ‘The energy W stored in the structure may be determined from its kinetic energy, which may be calculated from information on its mass distribution and from ve~ locity values measured by a suitable array of accelerometers or other motion trans- ducers, This measurement approach requires particular care in instrumentation se- lection and calibration but has a significant advantage: because it involves direct measurement of the dissipated energy. it does not rely on any particular model of dissipation: It also permits one, for example, to investigate how the loss factor varies with amplitude. Foree-and-motion transducer combinations may also be employed for the direct measurement of complex impedance or mobility (or of other force-motion ratios), from which damping information may be extracted, typically on the basis of Ny- quist plots. Much corresponding specialized modal testing or modal parameter extraction instrumentation and software has recently become available. Simple steady sinusoidal response measurements may also be made to measure Q and the half-power point bandwidth & directly on the basis of their definitions. ‘These measurements require particular care to ensure that the near-resonance Te- sponse of the mode of interest is not affected significantly by the responses of other modes resonance frequencies near that of the mode of interest. “The filter's response must be fast enough so that it can follow the decaying signal; otherwise one observes the decay of the filler response instead of that of the structural vibration, Filters with wider pessbands need to be used to abserve the more rapid decays associated with greater damping,480 STRUCTURAL DAMPING 12.3. DAMPING MODELS Analytical Models All of the foregoing discussion has dealt with “'viscous"’ damping, where energy dissipation results from’a force that is proportional to the velocity of a vibrating system and that acts opposite to the velocity. This viscous mode! of damping action has been used most widely because it results in relatively simple linear differential equations of system motion and because it yields a reasonable approximation to the action of some teal systems, particularly at small amplitudes. Among the many other models that have received considerable attention, most also involve a motion-opposing force that is a function of velocity. In dry friction or Coulomb damping, the force is constant in magnitude (but changes its algebraic sign when the velocity does), and in square law and power law damping the force magnitude is proportional to the square or to some other power of velocity. Of course, moder numerical methods also permit one readily to analyze models in- volving other velocity dependences, such as one might obtain from corresponding experiments = Viscoelastic Damping In contrast to the previously mentioned models, so-called structural or viscoelastic damping involves a force that opposes the velocity but that is proportional to dis- placement. This model is widely held to represent the behavior of structures better than does viscous damping, and it has the advantage of analytical simplicity. par- ticularly in complex (phasor) notation. Ina system undergoing sinusoidal motion described by Xe, the phasor cor- responding to a viscous damping force is given by cV = jwscX, whereas the phasor corresponding to a viscoelastic damping force may be written as jnkX in terms of the loss factor 7 and spring stiffness &. Thus, the sinusoidal response equation for a simple system may be written in phasor form as iF ‘x [-ia® + jac + kot for viscous damping [-ma? + jnk + kM [-ma* + k]-' for viscoelastic damping 2.13) __ In the equation corresponding to viscoelastic damping the complex stiffness ACL + jn) has been introduced. This is convenient because the real part of the stiffness indicates the system's strain energy storage capability and the imaginary part indicates its energy dissipation capability. The above expression pertaining to viscous damping leads directly to the dimensionless response relations of Eq (12.10). One may obtain the corresponding dimensionless response equations for a viscoelastically damped system simply by replacing 2fw /w, by 9 in this equa- tion, It should be noted that the different frequency dependences of the two damping forces imply different system behaviors, except near the natural frequency, Fur-| 12.4 DAMPING MECHANISMS AND MAGNITUDES 461 thermore, the viscoelastic model is often extended to loss factors that are functions of frequency*—particularly in order to represent measured frequency dependences of damping—and a loss factor with appropriate frequency variation may also rep- resent viscous damping. With frequency-dependent loss factors, the viscoelastic damping model can represent the sinusoidal response of any system whose param- eters are independent of amplitude. Selection of Models If one desires o determine the precise response of a system to a prescribed exci- tation, one generally needs to have a complete description of all forces, including the damping forces; that is, one needs a damping model that corresponds to the actual system. This is true, for example, if one wants to study the “*wave shape"” of the motion of a screeching brake or a chattering tool or if one needs to determine the response of a system to a transient, such as a shock In many practical instances, however, the details of the system's motion are of no interest and only the amplitudes are of concern. As has been mentioned, under steady resonant or free decay conditions (and in a few other situations), the am- plitudes are established essentially by the energy in the system. For such condi- tions, the details of the damping model are unimportant as long as the model gives the correct energy dissipation per cyele. It is for this reason that measures of damp- ing that involve only energy considerations have found wide acceptance ,+ One can use them for many practical purposes without investigating the time variation of the damping forces. 12.4 DAMPING MECHANISMS AND MAGNITUDES Since damping involves the conversion of energy associated with a vibration to ‘other forms that are unavailable to the vibration, there are as many damping mech- anisms as there are Ways to remove energy from a vibrating system. These include mechanisms that convert mechanical energy into heat as well as others that trans- port energy away from the vibrating system of concer Energy Dissipation and Transport Material damping, mechanical hysteresis, o internal friction refer to the conver- sion of mechanical energy into heat that occurs within materials due to deforma~ tions that are imposed on them. This conversion may result from a variety of ‘offects on the molecular, crystal lattice, or metal grain level, including magnetic, *Caution is indicated here if one wants to ealeulate-eertain responses by transformation from the fre~ ‘quency to the time domain. Some frequency variations of the loss factor may lead to physically un- ‘realizable results, ¢.g., to system motions thal begin before a force is applied." ‘+Equivalent viscous damping, defined as viscous damping that results in the sume energy dissipation as the damping actually present in the system, is often used in analyses, This damping made! obviously should not be used where details of the system motion are of eoncem.@ 462 © STRUCTURAL DAMPING thermal, metallurgical, and atomic phenomena.'' Figure 12.5 indicates the ranges of the loss factors reported for some common materials. Damping of a vibrating structure may also result from friction associated with relative motion between the structure and solids or fluids that are in contact with it, Also, an electrically conductive structure moving in a magnetic field is subject to damping due to eddy currents that result from the motion and that are converted into heat. =o 88 b8 7 F JUILDING MATERIALS — ab a 4 % ¢ F , IB .. 2 3 Oe METALS is : 8 : i ». tye |S wees Ee e +8 | eb = 2 1° le |e Br’. BE gee le IS K 5 = af 6 a eo l 8 US 2 ere 1 jpa¢/wOQIG=s9s/ Ut Ol x 2=0% wo gg = us OM uogZ= ug =% bunds s109 %yve wagz0'0= ui ze/1 =%y 1SaN|DN oua!ayay ainssaid juaiquio wio | (uty = 4) say0id jaays 10 wn -}WN|D “ul ZEvt UO SUDA jaa1s 10 wnumuanyo 405 S}u1od yop suinju0> u01b a4 paysog. 1Ul [= at 404 S1uIod O}P Sujoju09 uo/bas pasojoug — pa a WAUJLUNI |) eesh (uoyownsa 40)) aBoano zest ETA ot olo|mle ajpunxouddy (ue ssauyaiys 21D ° 3 Zo 9 o y4= %K jua1914j809 uodiosqo peanpay MM. m 465486 © STRUCTURAL DAMPING Damping Due to Acoustical Radiation Sound radiated by a vibrating structure transports energy from the structure and thus contributes damping. For a homogeneous pane! of thickness J+ and material density p,, one may calculate the panel's loss factor ng at frequency f due to sound radiation from one side of the panel from its radiation efficiency 9 by use of the relation'* pic n= 5 Tah” (12.15) where p and c denote the density of the ambient medium and the speed of sound in it, respectively. If the panel can radiate from both of its sides, ny is twice as great as indicated by Eq. (12.15). The magnitude of the radiation efficiency o depends on the vibratory velocity distribution on the panel as well as on frequency and thus generally is different for different excitation distributions, For a plate that has little inherent damping and that is excited at a single point, the radiation efficiency # obeys” (Usm AREF. fort << f. 0 = 40.45 VUE /e forf =f. (12.16) 10 for f >> f. f Where ‘4 denotes the panel's surface area (one side), Y its circumference, and f. the coincidence frequency. This frequency, which is defined as that at which the plate flexural wavelength is equal to the acoustical wavelength in the ambient me- dium, is given by f. = c° /1.8he,, where c, represents the longitudinal wave speed in the plate material, More detailed information on radiation efficiency is provided in Chapter 9. Equation (12.16) may be used for the general estimation of radiation efficiency values for plates that are not too highly damped. This equation alsa provides a reasonable estimate of the radiation efficiency of rib-stiffened plates if twice the total rib length is included in the circumference U. 12.5 VISCOELASTIC DAMPING TREATMENTS Viscoelastic Materials and Material Combinations Materials that have both damping (energy dissipation) and structural (strain energy storage) capability are called “‘viscoelastic.”” Although virtually all materials fall into this category, the term is generally applied only to materials, such as plastics and elastomers, that have relatively high ratios of energy dissipation to energy storage capability. Structural materials with high strength-to-weight ratios typically have little in- herent damping, as is evident from Fig, 12.5, whereas plastics and rubbers that are highly damped tend to have relatively low strength, This circumstance has led to the consideration of combinations of high-strength materials and high-damping a125 VISCOELASTIC DAMPING TREATMENTS 467 viscoelastic materials for applications where both strength and damping are 1e- quired, Additions of viscoelastic materials to structural elements have come to be known as viscoelastic damping treatments, If a composite structure is deflected, it stores energy via a variety of deforma tions (such as shear, tension and compression, flexure) in each structural element If 9; denotes the loss factor corresponding to the ith element deformation and H, represents the energy stored in that deformation, then the loss factor 9 of the entire structure abeys* Wi, r= Bae (12,17) where W;, = E W; denotes the total energy stored in the structure. The foregoing expression indicates that the loss factor » of the composite structure is equal to a weighted average of the loss factors comesponding to all of the clement deforma- tions, with the energy storages serving as the weighting factors. This expression also leads to an important conclusion: An element deformation can make a signif icant contribution to the total loss factor only if (1) the loss factor associated with itis significant and (2) the energy storage associated with it is a significant fraction of the total energy storage. Mechanical Properties of Viscoelastic Materials } Because viscoelastic materials have both energy storage and energy dissipation capability, it is convenient to describe their behavior in terms of elastic and shear moduli that are complex quantities, in analogy to the definition of the complex stiffness introduced in Eq. (12.13). The complex Young's modulus E of a material, defined as the ratio of the stress’ phasor to the strain phasor, may be written as!’ E = Eg + jE; = En(\ + jne), where the real part Ey is called the storage modulus, the imaginary part E, is called the loss. modulus, and the loss factor ay associated with Young's modulus is equal to £)/Eg. A completely analogous definition ap- plies to the complex shear modulus. For plastics and elastomers, the viscoelastic materials of greatest practical terest, the real and imaginary moduli as well as the loss factors vary considerably with frequency and temperature. However, these parameters usually vary rela~ tively little with strain amplitude, preload, and aging."" The loss factor associated with the shear modulus typically is equal to that associated with the Young's mad- ulus for all practical purposes, so that one generally need not distinguish between the two, Also, since most of the viscoelastic materials of practical imerest are virtually incompressible, the shear modulus value is éssentially equal to one-third of the corresponding Young's modulus value. One may also note that often nk << 1, so that |E| = Ep. Figure 12.7 shows how the (real) shear modulus and loss factor of a typical viscoelastic material varies with frequency and temperature. At low frequencies and/or high temperatures, the material is soft and mobile enough for the strain to follow an applied stress without appreciable phase shift so that the damping is ‘small; the material is said to be in its ‘trubbery"” state, At high frequencies and/468 © STRUCTURAL DAMPING. 10 5 - avec “ ieee : a5°t 2 os = rete § ose ms 4 I eC 201 t oe 50 TL] atc 358 2 See 5 8 iz + nee S were 3 © g os + e sec Zz on S—- f 0.08 + + + 0 we Fig? Pot (2) Frequency FH Fig. 12.7 Dependence of shear modulus and loss factor of a polyester plastic on frequency and teshperature,"": (a) functions of frequency at constant temperatare; (6) isometric plots on temperature-frequeney plane: or low temperatures, the material is stiff, immobile, may tend to be brittle, is relatively undamped, and behaves somewhat like glass; it is said to be in its “glassy” state, At intermediate frequencies and temperatures, the modulus takes on intermediate values and the loss factor is highest; the material is said to be in its “‘transition™ state. This material behavior may be explained on the basis of the interactions of the long-chain molecules that constitute polymeric materials. At low temperatures, the molecules are relatively inactive; they remain “locked together,"* resulting in high stiffness, and because they move little relative to each other, there is little imter- molecular “friction”* to produce damping. At high temperatures, the molecules become active; they move easily relative to each other, resulting in low stiffness, and because they interact little, there is again little energy dissipation due to in- termolecular friction, At intermediate temperatures, where the molecules have in- termediate relative motion and interaction, the stiffness also takes on an interme- diate: value and the loss factor is greatest. A similar discussion applies to the effect of frequency on the material properties, with the inertia of the molecules leading to their decreasing mobility and interaction with increasing frequency. ‘The observation that there exists a remperature-frequency equivalence, name! that an appropriate temperature decrease produces the same effect as a given fre~ quency increase, has led to the development of convenient plots in which data for the frequency and temperature variations of each material modulus collapse onto125 VISCOELASTIC DAMPING TREATMENTS 469. single curves.''-'? This collapse is achieved by plotting the data against a reduced frequency fe = fa(T'), where a(7') is an appropriately selected function of tem- perature T. In presentations of data in this form the function a(T’) may be given analytically, in a separate plot or, as has recently come into vogue, in the form of a nomogram that is superposed on the data plot. Figure 12.8 is an illustration of such a plot and nomogram; its use is explained in the figure’s legend, Data on the properties of damping materials are available from knowledgeable suppliers of these materials. Compilations of data appear in references 11 and 21. Key information on some of these materials appears in Table 12.1 in a form that is useful for preliminary material comparison and selection for specific applications in keeping with the concepts discussed in the later portions of this chapter. For each listed material the table shows the greatest loss factor value 7,,., exhibited by the material and the temperatures at which this value is obtained at three frequen- cies. The table also lists three values of the modulus of elasticity: Emax, the greatest value of Young's modulus, applies at low temperatures (i.e., temperatures consid- erably below those corresponding 10 tax); Eins the smallest value of E. applies at high temperatures; the transition value £,,,., applies im the gp, Tange; and Exmax ~ TmarEianse the Maximum value of the loss modulus, applies in the tran- sition range. TEMPERATURE ' ad _ WPS BOFC FOC S0°C BO"C 10°C OT -109 20% =O oy Jarl soe sor wre lee 7 /f i for 3 a x7 5 e = fe * é 2 FF o| / —h |f_|-—— § who: a | fe LA i es ~ ; sn a nl ae EDUGED FREQUENCY ty ~ fy (H2} Fig. 12.8 Reduced frequency plot of élastic modulus E and loss factor m of "*Sylgard 188” silicone potting compound. (After ref. 20.) Points indicate measured data to which curves were fitted. Nomograph superposed on data plot facilitates determination of reduced fre- quency fy corresponding to frequency f and temperature 7. Use of nomogram is illustrated by dashed lines: For f= 15 Hz.and T = 20°C, one finds fy = 5 < 10" Hz and £ = 3.8 x 10° N/m? y = 0.36,@ 470 STRUCTURAL DAMPING TABLE 12.1 Properties of Some Commercial Damping Materiats* Teaperature Cryer for gay & Elastie Modul (PSE)*#* 100 wz 1000 Hz Frax Emin Strang "Imax Antiphon=13 75 720 ES 1.283 1.988 3.83 Blachford squaplas 20125 1,686 SER 2.205 1.185 Barry Controls #-326 “25-10 ES 383 zou 3. he Dow Comming Sylgard 188 #0 0 e.zeu 3222.63 1.563 EAR c-1002 55 9035 2E2 7.783 1.5e8 EAR c-2003 70 1008S GED 2.2K 228K Lord LD-Ho9 Bo 125386 3.3831 Soundooat YAD 601 50-75 BEB 1SE2 6.763. 6.783 Soundosat DYaD 606 yoo 130305 1.2e2 ES BES Soundooat DYKD 609 150 1852S ERY 118 Soundaoat 30070385. TET 4.683 6.983 3M 180-110 17 BG 195 ets0 EN es teg EB 3M ISD-112 H2 10 Ho 1.385 ET 3.283 3.983 3H 1s00113 1.1 N20, SBS 32 BED 2.32 3H B68 0.8 15 505 TMES3E1 ES. 1.683 3H 130-830 1.000 --75 50-20 EB HSER 5.583 5.583 ce SHED 69 50-8128 E35 583 3.9K 3.508 “Approximate values taken from curves in Ref. 11 "To convert to *C, use the formila *C = (5/9) (*F-32) or the approximate table below: °F 80 60 -to 29 © 20 40 60 8 10 120 to 160 180 200 °C 62 -51 -l0 -29 -18 -7 4 16 27 38 49 60 71 82 93 ‘""unbers shown correspond ta storage (real) values of Young's modulus, except that Etay Pepresente the maximun values of the losa (imaginary) modulus.” F, abBtles for low tenperatures and/or high Frequencies: fry), sppiles for" Btn temperatures and/or’ low Trequanctes.. “Ey_cvc"and free, applies in*the range of tana’ Bivlde by 3 40 obtain the cort69BSoaing HB8E- coattan wales” To convert to N/m, multiply tabulated values by 7E3. The nunbér following £ represents the power of 10 by which the number preceding £ 13 to be maltiplied; e.g., 1.263 represents 1.2 £ 103, It is important to keep in mind that the mechanical properties of polymeric materials, including plastics and elastomers, tend to be more variable than those of metals and other classical structural materials. Some of this variability results from a polymer’s molecular structure and molecular weight distribution, which depend not only on the material's chemical composition but also on its processing. Additional variability results from the various types and amounts of plasticizers and fillers that are added to most commercial materials for a number of practical purposes, Thus, it is quite common for nominally identical polymeric materials to exhibit considerably different mechanical behaviors. It may also occur that even material samples from the same production run have considerably different loss factors and moduli at frequencies. and temperatures at which they are intended to12.5 VISCOELASTIC DAMPING TREATMENTS 471 he used, pointing toward the need for careful quality control and performance verification for critical applications. Structures with Viscoelastic Layers One may calculate the loss factor of a structure vibrating in a given mode by use of Eq. (12.17) if one knows the energies HW, stored in the vatious deformations of all of the component elements. Indeed, moder finite-element analysis meth- ods”? proceed by calculating the modal deflections, applying these to evaluate all the energy storage components and then using Eq. (12.17) to find the loss factor. ‘Analytical results have been developed for flexure of uniform beam and plate structures under conditions that are often approximated in practice. These results, which are extremely useful for design guidance and for development of an under- standing of the important parameters, apply to structures whose deflection distri- butions are sinusoidal* and to structural configurations in which an insert or layer of viscoelastic material is subjected predominantly either to tension and compres- sion or to shear. Two-Component Beams In flexure of a uniform beam with an insert or added layer or viscoelastic material, as illustrated by Fig. 12.9, the energy storage (and dissipation) associated with shear and torsional deformations may generally be neglected. If contact between the components is maintained without slippage at all surfaces and if the loss factor of the basic structural (nonviscoelastic) component is negligible, then the loss fac- “The deflection distribution of any beam (or plate) vibrating in one of its natural modes is at least approximately: sinusdidal (in one dimension for beams and in two dimensions for plates) at locations that are one wavelength oF more from the boundaries, regardless of the boundary conditions. Therefore, the assumption of 2 sinusoidal deflection distribution is correct for a larger fraction of the structure as the frequency increases. Fig. 12.9 End views of beams with viscoelastic insemts oradded layers. Structural material is unshaded, viseoelastic material is shown shaded; H\) represents distance between neutral axis of structural component and that of viscoelastic component. Beam deflectian is vertical, with wave propagation along the beam length, perpendicular to the plane of the paper.Gg 472 © STRUCTURAL DAMPING tor 7 of the composite beam is related to the loss factor of the material of the viscoelastic component by* 2 2 [1-90 Re ei (2.18) e Wl + (%/f2Fel | where ce = (1 + £)° + (8k)* and Hy, denotes the distance between the neutral axes of the two components. With subscript 1 referring to the structural (undamped) component and subscript 2 referring to the viscoelastic component, k = Ky/'K,, where K, = EA; denotes the extensional stiffness of component i, expressed in terms of its Young's modulus (real part) £, and cross-sectional area A). Further- more, 7; = vi;/A; represents the radius of gyration of A,, where J, denotes the centroidal moment of inertia of 4), For the often-encountered case where the structural component's extensional stiffness is much greater than that of the viscoelastic component, k << 1 and Eq. (12.18) reduces to BE gly BE, Tp = = 12.19) El, + Edy Ey hy G2) ” where Jy = I, + Hi. the neutral axis of A, ‘The last expression in Eq, (12.19) applies for Eyl << E,/,, which is generally true in practical structures where the area and elastic modulus of the viscoelastic component are small compared to those of the structural component. In this case the composite structure's neutral axis coincides very nearly with that of the struc- tural component and the dominant energy storage is associated with flexure of the structural component (whose flexural stiffness is £\/,). The dominant energy dis- sipation is associated with extension and compression of the viscoelastic compo- nent, with the average extension (equal to the extension at the viscoelastic com- ponent’s neutral axis) given by the flexural curvature and the distance H,, between the neutral axes of the viscoelastic and the structural components.* The flexural curvature is greatest at the antinodes of the vibrating structure: most of the damp- ing action thus occurs at these locations, with little damping resulting from the ‘material near the nodes. The second form of Eq. (12.19) contains two ratios: the first involves only material properties, and the second only geometric parameters, It indicates that the most important dynamic mechanical property of the viscoelastic material is its extensional loss modulus E, = BE. In keeping with the conclusions based on the general energy expression [Eq. (12.17)], good damping of the composite structure (3 + Hj) denotes the moment of inertia of A; about "A “spacer,”" a layer that is ofiff in shear and soft in extension (¢.g., like honeycomb), between the structural and the viscoelastic camponent can increase Hs and thus the damping obwuined with a given amount of viscoelastic material.12.5 VISCOELASTIC DAMPING TREATMENTS 473, can be obtained only from a viscoelastic material that has not only a high loss factor but also a considerable energy storage capability. Plates with Viscoelastic Coatings A strip of a plate (see insert of Fig. 12.10) may be considered as a special case of a two-component beam, where the two components have rectangular cross sections. Thus Eqs. (12.18) and (12.19) apply, with r, = H,/V12 and Hy: = iH, +H), where H, denotes the component thickness, The energy storage and dissipation considerations that were discussed in the foregoing section as well as the foregoing remarks concerning the dominant damping material properties apply here also. Figure 12.10 is a plot based on Eq. (12.18) for 6? << 1. It shows that for small relative thicknesses fy A /'H,, the loss factor ratio 9/8 is proportional to the viscoelastic layer thickness, whereas for very large relative thicknesses the loss factor ratio approaches uni Lots factor ratio még 1073 oF that is, the loss factor of the coated plate approaches J] ee 8a57 Relative thichness of domping layer, he=Me/ Fig. 12.10 Dependence of loss factor 17 of plate strip with added viscoelastic layer on relative thickness and relative modulus of layer.” Curves apply strictly only if loss factor 8 of viscoelastic material is small compared to unity.G 474 STRUCTURAL DAMPING that of the viscoelastic coating, as one would expect.* As also is evident from the figure, at small relative thicknesses the loss factor ratio is proportional to the mod- ulus ratio ¢ = £),/E), For small relative thicknesses, that is, in the regions where the curves of Fig. 12.10 are nearly straight, the loss factor of a coated plate may be estimated from®” = SB hy G+ 6h, + 4h3) (12.20) 1 where hy = H;/H,. If two viscoelastic layers are applied to a plate, one layer on each side, then the loss factor of the coated plate may be taken as the sum of the loss factors contrib- uted by the individual layers, with each contribution calculated as if the other layer were absent, provided that each viscoelastic layer has low relative extensional rigidity, that is, that E,#, << E,Hj. If this inequality is not satisfied, a more complex analysis is required. Three-Component Beams with Viscoelastic Interlayers Figure 12.11 illustrates uniform beams consisting of two structural (nonviscoelas- tic) components interconnected via a relatively thin viscoelastic component. Such three-component beams may be preferable to two-component beams for practical reasons because the viscoelastic material is exposed only at its edges; however, such beams can also be designed to have higher damping than two-component beams of similar weight.+ “For very thick viscoelastic coatings, deformation in the thickness direction (which are nat considered In this simplified analysis) also may play a significant role, particularly at frequencies at which stand- ing-wave resonances may occur in the viscoelastic material. ** It should be noted that design changes to obtain increased damping generally also result in mass and stffiess changes, which tend to affect a siructure's vibratary response and shouild be considered in the design prozess,"°™* a 3 Fh, Fig. 12.111 End views of composite beams made up of two structural components (un shaded) joined via a viscoslastic component (shaded); £4; i8 distance between neutral axes of structural components. Beam deflection is vertical, with wave propagation along the beam length, perpendicular to the plane of the paper.125 VISCOELASTIC DAMPING TREATMENTS 475, In flexure of a three-component beam with a viscoelastic Inyer whose exten- sional and flexural stiffnesses are small compared to those of the structural com- ponents, the dominant energy dissipation is associated with shear in the viscoelas- tic component and the most significant energy storage occurs in connection with extension/compression and flexure of the two structural components. The shear in the viscoelastic component is greatest at the vibrating structure's nodes. Thus, most of the energy dissipation occurs in the viscoelastic material near the nodes, with relatively little resulting in that near the antinodes. For efficient damping, it is important that the shearing action in the viscoelastic material not be restrained (particularly at and near nodes) by structural imterconnections, such as bolts. The loss factor 7 corresponding to a spatially sinusoidal deflection shape of such a three-component beam is related to the loss factor 8 of the viscoelastic material BYX Terre cha ae (221) where a E\A, EyAy i (12.22) pH, Here subscripts | and 3 refer to the structural components and 2 to the viscoelastic component; £), 4,, and J; represent, respectively, the Young's modulus, cross- sectional area, and moment of inertia of component #; Hy; denotes the distance between the neutral axes of the two structural components; and G; represents the shear modulus (real part) of the viscoelastic material, Hy the average thickness of the viscoelastic layer, and b its length as measured on a cross section through the beam. The wavenumber p of the spatially sinusoidal beam deflection obeys ‘ ft (%) am (12.23) Pr \2 aN where A represents the bending wavelength and B denotes the flexural rigidity and the mass per unit length of the composite beam, The structural parameter Y of three-component structures depends only on the geometry and Young's moduli of the two structural components, whereas the shear parameter X depends also on the properties of the viscoelastic layer and on the wavelength of the beam deflection. The shear parameter X is proportional to the Square of the ratio of the beam flexural wavelength to the decay distance,*' that is, the distance within which a local shear disturbance decays by a factor of ¢, where e ~ 2.72 denotes the base of natural logarithms; thus, X also is a measure ‘of how well the viscoelastic layer couples the flexural motions of the two structural components.@ 476 STRUCTURAL CAMPING The (complex) flexural rigidity of « three-component beam is given by xy 1+ x, B= (Eh + Esb) (: + ) X* = XQ — 38) (12.24) Its magnitude is 8 = |B], Thus, for small X, the flexural rigidity B of the com- posite beam is equal to the sum of the flexural rigidities of the structural compo- nents, that is, to the total flexural rigidity that the two components exhibit if they are not interconnected. For ¥ >> 1, however, B approaches 1 + ¥ times the foregoing value, which is equal to the flexural rigidity of a beam with rigidly interconnected structural components 1 and 3. Fora given value of and ¥, the loss factor 9 of the composite beam takes on its greatest value, bY Ye "575 Tie (12.25) at the optimum value of X, which is given by Xoge = 1 + FC + Py? (12.26) With the aid of these definitions, one may rewrite Eq, (12.21) in terms of the ratio R = X/Xu in the following form, which provides a convenient view of the damp- ing behavior of three-component beams: 2 + NOR sie N= (14 bY )Xpe (12.27) Figure 12.12, which is based on Eqs. (12.25) and (12.26), shows how ta,./8 increases monotonically with ¥, indicating the importance of selecting a configu- ration with a large yalue of ¥ in the design of highly damped composite struc- tures.* Figure 12.13 gives approximate values of ¥ for some often encountered configurations. Figure 12.14 shows how 7 /‘Yays Varies with X/X,p.. indicating the importance of making the operating value of ¥ of a given design match Xyy as closely as possible in onder to obtain a loss factor that approaches tan. If one knows the wavenumber p = 2n/2 and the frequency associated with a given beam vibration, one may calculate the loss factor 7 of a composite beam simply by substituting the beam parameters and the material properties at any fre- quency (and temperature) of interest into Eqs. (12.21) and (12.22) or (12.25}- (12.27). The latter set of equations is particularly useful for judging how far from the optimum a given configuration may be operating. ‘7A shear-stiff, extensionally soft “spacer” (e.g., honeycomb) inserted between the viseoelastic and fone ar both structural components can serve 10 increase Ais and, therefore, the value of ¥. See Eq. (12.22). Fora given deflection of the composite structure, a spacer inereases the damping by increasing the shear strain—and thus the energy storage and dissipation—in the viscoelastic component.425 VISCOELASTIC DAMPING TREATMENTS © 477. Pct Bt om Bae max B a 0.08| 4 oF OT 1 © STRUCTURAL PARAMETER, Y Fig. 12.12 Dependence of maxinum loss factor n,.. of three-component beams or plates ‘on structural parameter ¥ and loss factor 8 of viscoelastic material. 3 GENERAL PLATE Sa] GENERAL o€AM sysrem NERAL PLA Kia ¥. ce 3 y_Srit+hy tee SECS 3 Ait Ag " = Hy 2 ‘SMALL ADDED BEAM Hy " E #4 ae cM fe + ram apbeD “TARE” fia tateg TE fl oH WITH = 2:1 HHGRNESS Nao Ym2 aS ‘icing i 3 I—BEAM | 3 eee t rurcmorircaTe MS ‘eater IDM ve ett te a 1 ar a Hy=Hy Fig. 12.13 Values of structural parameter ¥ for three-component beam and plate config- ‘urations with thin viscoelastic components and with structural components of the same material; viscoelastic component is shown cross-hatched. /, moment of inertia: A, cross- sectional area: r, radius of gyration; Hx, distance between neutral axes.478 © STRUCTURAL DAMPING 10 ‘ | ‘ as : © 4 4 x. ‘ { @ ~ — | | Pai) a 4 3 eg 20304080 BO HG tas Bg Fig. 12.14 Dependence of loss factor 9 of three-component beam or plate on shear param eter X [from Eq. (12.27). If one knows only the frequency and not the wavenumber p corresponding to @ beam vibration, one needs to use Eq. (12.23) to determine p. Substitution of B as calculated from Eq. (12.24) into Bq, (12.23), followed by substitution of the result into the first of Eqs. (12.22), leads to a cubic equation in X. Although one may solve this numerically, it is often more convenient to determine X by use of an iteration procedure like that indicated in Fig. 12.15. In contrast to the previously discussed two-component beam, the loss factor 7 of a three-component beam does not depend primarily on the loss modulus (i.e, on the product of the loss factor 6 and storage modulus E; or G,) of the viscoelastic material. The loss factor of a three-component beam depends on 8 and G, sepa- rately, and the separate dependences must be taken into account in the design of such a beam. Design of a highly damped three-component beam structure requires (1) choice of a configuration with a large structural parameter ¥, (2) selection of a damping material with a large loss factor 8 in the frequency and temperature range of interest, and (3) adjustment of the damping material thickness H, and length b so as to make X [as calculated from Eq. (12.22) for the value of Gs applicable to the frequency and temperature of concer] approximately equal to12.5 VISCOELASTIC DAMPING TREATMENTS 479 CALCULATE Y from Eq. 12.22 DETERMINE ond Gz at FREQUENCY and TEMPERATURE of INTEREST from MATERIAL DATA FIND Kop from Eq. 12.26 CALCULATE B from Eq. 12.24 FIND We? from Ea, 12.23 CALCULATE NEW X trom Eq. 12.22 Fig. 12.15 Iteration procedure for determi- nation of loss factor of three-component beams or plates for which wavelength is not known initially. Yes [FIND 7 trom Eq, 12.21 oF 12.25 12.27] Xow [given by Eq. (12.26)]. The resulting design then will have a loss factor ep- proximately equal to n,,,, as given by Eq. (12.25). The expected performance of any design should be checked for the frequency and temperature ranges of interest by means of the procedure described in the previous paragraph. Note that a given design may be expected to perform opti mally—that is, to have X under operating conditions approximately equal to Xiq— only in a limited range of frequencies and temperatures, with reduced performance outside this range. Plates with Viscoelastic Interlayers A strip of a plate consisting of a viscoelastic layer between two structural layers may be considered as a special case of a three-component beam. In a plate strip, all components have rectangular crass sections of the same width; Eqs. (12.22) accordingly become (12.28)480 © STRUCTURAL DAMPING If, furthermore, Eyl, + £y/; in Eq, (12.24) is replaced by yy(£,H} + Esf}) and if 4 in Eq, (12.23) is interpreted as the mass per unit surface area of the plate, then all of the foregoing discussion pertaining to beams also applies to pl REFERENCES 1. W. T. Thomson, Theory of Vibration with Applications, 2nd ed., Prentice-Hall, En- glewood Cliffs, NJ, 1981, 2. R. Plunkett, “Measurement of Damping,” in J. F. Ruzicka (ed.), Structeral Damping, ‘American Socicty of Mechanical Engineers, New York, 1959. 3. E. E. Ungar and E. M. Kerwin, J1 ‘Loss Factors of Viscoelastic Systems in Terms of Energy Concepts.” J. Acoust. Soc, Am. 34(7), 954-957 (1962). 4. K.N. Tong, Theory of Mechanical Vibration, Wiley. New York, 1960. 5. E. E. Ungar, “Mechanical Vibrations,”" in H. A, Rothbart (ed.), Mechanical Design and Systems Handbook. 2nd ed., McGraw-Hill, New York, 1985, Chapter 5. 6. D.J, Ewins, Modal Testing: Theory and Practice, Research Studies Press Lid... Letch- worth, Hertforishire, England, 1986, 7. ¥. H, Neubert, Mechanical Impedance: Modelting/Analysis of Structures, Josens Printing and Publishing Co., State College, PA, distribuied by Naval Sea Systems Command, Code NSEA-SSN, 1987. 8. B. L, Clarkson and J. K. Hammond, ““Randam Vibration,” in R. G, White and J, G. Walker (eds.), Noise and Vibration, by Wiley, New York, 1982, Chapter 5. 9. L. Cremer, M, Heckl, and E. E. Ungar, Structurebarne Sound, 2nd ed., Springer- Verlag, New York, 1988. 10. S. H. Crandall, ‘“The Role of Damping in Vibration Theory,"" J. Sound Vib. 11(1), 3-18 (1970). il. A.D. Nashif, D. I. G. Jones, and J. P. Henderson, Vibration Damping, Wiley, New York, 1985. 12. G. Kurtze, ‘*Korperschalldimpfung durch Kémige Medien" [Damping of Structure- bome Sound by Granular Media}, Acustica 6(Beihett 1), 154-159 (1956), 13. B. E, Ungar and L. G. Kureweil, “Structural Damping Potential of Waveguide Ab- sarbers,"" Trans. iniemoise 84, pp. 571-574, December 1985, 14. M.A. Heckl, “*Measurements of Absorption Coefficients on Plate: Am, 34, 308-808 (1962), 15. G. Maidanik, ‘Energy Dissipation Associated with Gas-Pumping at Structural Joints,”* J. Acoust. Soc. Am, 34, 1064-1072 (1966). 16. E, E, Ungar, “Damping of Panels Due to Ambient Air," in P. J. Torvik, (ed.). Damp- ing Applications in Vibration Control, American Society of Mechanical Engineers. AMD-Vol 38, New York, 1980, pp. 73-81. 17. E.E. Ungarand J, Carbonell, “On Panel Vibration. Damping Due to Structural Joints," AIAA J. 4, 1385-1390 (1966). 18, B. B. Ungar, “Damping of Panels," in L. L. Beranck, (ed.), Noise and Vibration Control, McGraw-Hill, New York, 1971, Chapter 14. 19, D. 1, G, Jones, A Reduced Temperature Nomogram for Characterization of Damping Material Behavior,” Shock Vib. Bull. 48(Pi. 2), 13-22 (1978) J. Acoust, See:20. 21 2. 23. REFERENCES © 481 D. 1. G. Jones and J. P. Henderson, "Fundamentals of Damping Materials,"’ Section 2 of Vibration Damping Short Course Notes, University of Dayton, M. L. Drake (ed.), 1988. J. Soovere and M. L. Drake, “‘Acrospace Structures Technology Damping Design Guide,"" AFWAL-TR-84-3089, Flight Dynamics Laboratory, Wright-Patterson Air Foree Base, OH, December 1985, M. L. Soni and F. K. Bogner, ““Finite Element Vibration Analysis of Damped Struc tures."” ALAA J. 20(5), 700-707 (1982). C. D, Johnson, D, A. Kienholz, and L. C. Rogers, “'Finite Element Prediction of Damping in Beams with Constrained Viscoelastic Layers,"" Shock Vib. Bull. S1(Pt. 1). 71-82 (1981) sner and M. L. Drake, ‘“Damped Structure Design Using Finite Element Shock Vib. Bull, $2(Pt, 5), 1-12 (1982), . EE, Ungar, “Loss Factors of Viscoelastically Damped Beam Structures,’' J. Acoust. Soc. Am. 34(8), 1082-1089 (1962) E. M, Keewin, Jr., “Damping of Flexural Waves in Plates by Spaced Damping Treat- ‘ments Having Spacers of Finite Stiffness,"" Proc. 3nd Int, Congr. Acoust. 1959, Stutt- ‘gan, Elsevier, 1961, pp. 412-415 D. Ross, E. E. Ungar, and E. M, Kerwin, Jr., “Damping of Plate Flexural Vibrations by Means of Viscoelastic Laminae,” in J. Ruzicka (ed.), Structurat Damping, Amer- ican Society of Mechanical Enginers, New York, 1959, Section 3. E. EB. Ungar and E. M. Kerwin, Jr., “Plate Damping Due to Thickness Deformations in Attached Viscoelastic Layers,"* J. Acaust. Sac, Am. 36, 386-392 (1964), D. J. Mead, “Criteria for Comparing the Effectiveness of Damping Materials,"" Noise Controt 1, 27-38 (1961). i bration Control (1),"" in R. G. White and J. G. Walker (eds.), Noise and Vibration, Ellis Horwood Limited, Chichester, West Sussex, England, 1982. E, M. Kerwin, Jr., “Damping of Flexural Waves by a Constrained Viscoelastic Layer,"* J. Acoust. Soe. Am. 31, 952-962 (1959),