Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
net/publication/242439704
CITATIONS READS
14 258
2 authors, including:
Samir Nayfeh
Massachusetts Institute of Technology
58 PUBLICATIONS 1,542 CITATIONS
SEE PROFILE
All content following this page was uploaded by Samir Nayfeh on 14 March 2014.
DETC2003/VIB-48534
an active research area for many years [8]. Nearly 175 years ago, tively, 106 1 0 3 j kPa and 38 9 1 0 4 j kPa. Based on these
Faraday studied the interaction between a vibrating body and measurements, we calculate the complex wave speeds in dilata-
a granular material and found that heavy particles move to the tion and shear to be 32 3 1 0 15 j m/s and 19 7 1 0 20 j m/s,
nodes of vibration, but light particles move to the antinodes [9]. respectively.
Recent experiments by Fricke [10] and Nayfeh et.al. [11] indi- In all of the experiments, the beams are suspended by light
cate that such low-density particles can provide high damping of elastic strings to simulate free-free boundary conditions. An
structural vibration if the speed of sound in the fill is sufficiently electromagnetic shaker provides an excitation at one end of the
low. beam via a force transducer (PCB 208B [17]). The response is
In this paper, we study the flexural dynamics of beams cou- measured by an accelerometer (PCB 333A30 [17]) located at the
pled to media such as foam and granular material in which the same end of the beam as the shaker. A chirp excitation is gener-
speed of sound propagation is much lower than that in the beam. ated by a dynamic signal analyzer and supplied to the shaker via
Foam materials are commonly employed for sound absorption an amplifier.
and noise control (e.g., Kinsler et. al. [12]; Beranek and Ver [13]) In Figure 1, we plot the force-to-acceleration frequency re-
in aircraft and automobiles. They are also widely used as absorp- sponses obtained for the aluminum beam with and without the
tive liners in anechoic chambers, air conditioning systems, intake foam layer. As expected, the beam without foam exhibits very
and exhaust ducts of turbines, and large industrial fans among little damping, with ζ 0 002 for each of the first ten modes.
other applications. In Section 2.1, we report on experiments in When a layer of foam of thickness 12.7 mm is glued to the alu-
which coupling an aluminum beam to foam provides significant minum beam, the increase in damping in the first five modes is
damping over a broad band of frequencies for a relatively low in- very small, but there is a significant increase in the damping of
crease in mass. Next in Section 2.2, we develop a simple, linear the sixth and higher modes. Based on the speeds of sound com-
model in which an Euler-Bernoulli beam is coupled to a foam puted in the foregoing, we find that the sixth mode occurs in a
material in which dilatation and shear waves can propagate. Us- frequency range where quarter-wavelength dilatation and half-
ing a modal expansion, we obtain the accelerance of such a beam wavelength shear waves can be set up through the thickness of
under various boundary conditions. The results of the model are the foam. This suggests that the damping in the frequency range
found to be in close agreement with the responses measured in of 600 to 1600 Hz is the result of strong coupling between the
the experiments. beam and foam arising from the establishment of standing waves
In Section 3.1, we report on experiments in which aluminum through the thickness of the foam.
beams filled with low-density granular material exhibit signifi-
cant damping (as high as 6%) in the first two modes of the beam.
Next, in Section 3.2, we develop a simple, linear model similar 2.2 Model
to that of the foam, idealizing the granular material as a com- In this section, we develop a model by which the responses
pressible fluid. Using the model developed in Section 3.2 and measured in Section 2.1 can be predicted. We consider a beam
the properties of the powder (such as its speed of sound and loss of length L, flexural stiffness EI, and mass per unit length m.
factor) documented in the Appendix, we obtain the accelerance A layer of foam is coupled to the beam as shown in Figure 2.
of the beam under various boundary conditions. The results of The beam is excited by a point-harmonic force at a frequency ω
the model are found to be in close agreement with the measured and distance x f from one end of the beam. We employ a simple
responses. Euler-Bernoulli model for the beam and consider the foam to be
a lossy and isotropic continuum in which waves of dilatation and
distortion can propagate.
2 Damping using Foams The foam material used in the experiments described in Sec-
tion 2.1 is an elastic-porous medium. It is known that three types
2.1 Experiments of waves can propagate in such media: two types of dilatation
We report on experiments conducted on an aluminum beam waves and a shear wave (e.g., Biot [18]; Bolton et.al. [19]). One
(1447.8 38.1 12.7 mm) coupled to a 12.7 mm layer of EAR C- of the dilatation waves depends largely on the bulk elastic prop-
3201 (EAR Corporation [14]) energy-absorbing foam using 3M erties of the material of the foam and the other on the acoustical
Contact 80 neoprene adhesive (3M Corporation [15]). The den- properties of the fluid in the pores, the porosity, and the flow re-
sity of the foam is 104.1 kg/m3 , and its stiffness and loss factor sistivity. Hence the former and latter are respectively referred to
are available from the manufacturer’s data sheets [14], but the as frame and airborne waves. The stress induced on the beam
effective complex stiffness of the foam varies somewhat from lot by the fluidic component of the foam scales as the product of the
to lot and with frequency. In [16], a set of experiments were bulk modulus of the air in the pores times the volumetric strain
carried out to measure the complex extensional and shear stiff- in air. Hence although this type of wave can have a significant
40
20
−20
−40
0 200 400 600 800 1000 1200 1400 1600 1800
Frequency (Hz)
Figure 1. Measured force-to-acceleration frequency responses under chirp excitation: without foam (dotted) and with foam (solid)
σy x 0
τx0
M V V dV M dM
y
dx
L
x
z
h
hf Figure 3.
dx
Free-body diagram of an infinitesimal beam element of length
λ λ 2
νE f 1 ν 1
2ν , and G are respectively, the density, com-
2.2.1 Equations of Motion Consider steady vibration plex Lame constant, and complex shear modulus of the mate-
of the beam-foam system of Figure 2 under harmonic excitation rial of the foam. To obtain the equation governing the vibratory
at a frequency ω by a point force Re Fδ x
x f e jωt . We de-
x and y directions by Re(u x y ω e jωt ) and Re(v x y ω e jωt ), re- a force in y direction whereas the shear stress τxy x 0 ω results
in a moment about the neutral axis. Making use of the above in-
are governed by the following second-order wave equations of teraction between the foam and beam, we write the equation for
d 4V bh ∂τxy x 0 ω
αn x
bσy x 0 ω
Fδ x
x f
EI 4 mω V 2
∂x
dx 2 βn x
sinh kn x
pn cosh kn x
(10)
(3)
where b and h are, respectively, the width and height of the beam.
where kn is the wave number of the nth mode of the beam and pn
2.2.2 Boundary Conditions for the Foam At the is a constant given by
interface between the foam and beam (at y 0), we require that
sin kn L
sinh kn L
pn
the deformations in the foam match those of the surface of the
cosh kn L
cos kn L
(11)
v x 0 ω
V x ω
h2 ∂V (5)
form
∂x
∞
∑
At the free surface of the foam (at y h f ), the normal and shear v x y ω χn y ω α n ξn y ω βn x
x (12)
n 1
stresses must vanish. Hence we have ∞
u x y ω
∑ φn y ω αn x kn
ψn y ω βn x kn (13)
λux x h f ω λ 2G vy x h f ω
n 1
0 (6)
uy x h f ω v x x h f ω
0 (7)
where the primes denote the first derivative and χ, ξ, φ, and ψ are
T
(15)
the system.
and the matrix An is given in Table 1. Next, we solve (14) subject
2.2.3 Approximate Solution Because closed-form to the boundary conditions (4)–(7) to obtain
solutions for (1)–(3) along with the boundary conditions (4)–(7)
are generally difficult to obtain, we seek approximate solutions
Xn Bn eΛn y cnVn (16)
to the coupled beam-foam problem. To this end, we expand the
deflection of the beam in terms of the eigenfunctions of an un-
damped beam as where Bn and Λn are, respectively, the matrices of eigenvectors
and eigenvalues obtained by diagonalizing An , and cn is a con-
∞
V x ω
∑ Vn ω
αn x
βn x
(8)
stant vector whose elements are determined by enforcing the
boundary conditions (4)–(7). Thus, having obtained the func-
tions χn , ξn , φn , and ψn , we compute the stresses σy x 0 and
n 1
τxy x 0 at the interface of the foam and the beam and substitute
0 1 0 0 0 0 0 0
ρω2 kn2 λ 2G λ G
G
0 0 0 0
G kn
0 0
0 0 0 1 0 0 0 0
An
ρω2 kn2 λ 2G λ G
0 0 0 0 0 0 G kn
G
0 0 0 0 0 1 0 0
λ G ρω2 kn2 G
0 k
λ 2G n
0 0 0
λ 2G
0 0
0 0 0 0 0 0 0 1
λ G ρω2 kn2 G
0 0 0
k
λ 2G n
0 0
λ 2G
Because the speed of sound and loss factor of the powder are
not precisely known, we detail in the Appendix a set of exper-
0.063 in iments to measure these properties. From these measurements,
0.125 in 0.8 in we find that the speed of sound does not vary very much with
the frequency and is approximately 58 m/s. Likewise, we find
that the loss factor also does not exhibit a strong dependence on
2 in
1.125 in
frequency and is approximately 0.20.
(a) (b) In all of the experiments, the beam is suspended by soft
elastic strings to simulate free-free boundary conditions. An
Figure 5. Cross section of aluminum beams: (a) box beam with an over impulsive excitation is provided by an impact hammer (PCB
all length of 28.25 inches, and (b) U-Channel beam with an overall length 333A30 [17]) at one end of the beam, and the response is mea-
of 23.25 inches sured by an accelerometer (PCB 333A30 [17]) located at the
same end the beam.
The first set of flexural measurements were conducted on an
of this linear system is equal to the number of terms in the ex- aluminum box beam of length 717.55 mm and cross section as
pansion (8). Next, after computing the values of Vn , we write the sketched in Figure 6(a). The box beam is filled completely with
non-dimensional accelerance R at x x f as 3M Microbubbles and, due to the compressibility of the granu-
lar material, the beam is readily filled to an extent that no free
∞
R ∑
mω2V L mω2Vn L surface remains between the granular fill and the interior of the
αn x f βn x f
(17) beam wall. The total mass of the particle fill is 16 per cent of that
F
n 0 F
of the beam.
Finally, we plot in Figure 4 the force-to-acceleration frequency In Figure 6, we plot the force-to-acceleration frequency re-
response and compare it with the measured response. We find sponse for the completely filled box beam along with that of
that there is good agreement between the measured and predicted the empty (unfilled) beam. As expected, the unfilled beam ex-
frequency responses. hibits very little damping, with ζ 0 002 for each of the first
40
20
−20
−40
0 100 200 300 400 500 600 700 800 900
Frequency (Hz)
Figure 4. Comparison of measured and predicted force-to-acceleration frequency responses: measured without foam (dotted), predicted without foam
(dash-dot), measured with foam (dashed), and predicted with foam (solid)
90 100
80
80
70
60
60
Magnitude (dB)
Magnitude (dB)
50
40 40
30
20
20
10 0
0
−10 −20
0 200 400 600 800 1000 1200 1400 1600 1800 2000 0 200 400 600 800 1000 1200 1400
Frequency (Hz) Frequency (Hz)
Figure 6. Measured force-to-acceleration frequency responses of the Figure 7. Measured force-to-acceleration frequency responses of the U-
rectangular box beam under hammer excitation: unfilled (dotted), filled channel beam under hammer excitation: unfilled (dotted), filled (solid)
(solid)
with a layer of powder 17 mm thick. The mass of granular fill consider a beam of length L, flexural stiffness EI, and mass per
is 9.6 per cent of that of the beam. As in the measurements on unit length m. A layer of granular material of thickness h and
the box beam, we find that the filled U-channel beam exhibits width b is coupled to the beam as shown in Figure 8. The beam
significant damping in each of its first two modes. In this case, is excited by a point-harmonic force at a frequency ω and dis-
the tuned-mass damper effect appears close to the second reso- tance x f from one end of the beam. We employ a simple Euler-
nance of the beam. The critical damping ratios for the first three Bernoulli model for the beam and consider the granular fill to be
modes of the filled beam are found to be 0.005, 0.05, and 0.06, a compressible inviscid fluid in which two-dimensional pressure
respectively. waves can propagate. While the mechanism of dissipation within
the granular material is not completely understood, we character-
ize it by a complex speed of sound c defined by
3.2 Model
In this section, we develop a model to predict the dynamics
of the powder-filled beams described in the previous section. We c c0 1 jηω ω (18)
∑ Vn ω
αn x
βn x
(22)
n 1
y
h
where αn x and βn x are respectively, the propagating and
z
∞ ∞
Figure 8. Diagram showing the parameters used in the model: The
p x y ω
∑ φn y ω α n
x ∑ ψn y ω β n
x (23)
beam has length L and is partially filled with granular material over a
n 1
n 1
width b and to a height h.
ρω2Vn
y
(24)
force Re Fδ x
x f e jωt . We denote the vibratory deflection of λn1 1 e2λn1h
eλn2 y
eλn2 2h
ψn y ω
ρω2Vn
y
(25)
λn2
sure p x y ω in the granular material is governed by the follow-
1 e2λn2h
p 0 (19) wave number of the nth mode of the undamped beam. Next,
∂x2 ∂y2 c2 we use the above expressions for φn and ψn to compute the pres-
At the free surface (at y h), the pressure must vanish. At the
proportional to the acceleration of the beam. Hence we have the and L to obtain a system of linear equations in Vn . The order of
boundary conditions this linear system is equal to the number of terms in the expan-
sion (22). After determining the values of the Vn , we compute
∂p the non-dimensional accelerance R at x x f according to
p x h ω x 0 ω ρω2V x ω
0 and (20)
∂y
∞
R ∑
mω2V L mω2Vn L
where ρ is the density of the powder. As in the case of foam αn x f βn x f
(26)
(Section 2.2.1), we consider the force and moment balance on an F F n 0
infinitesimal element of the beam to obtain the following equa-
tion governing the vibratory motion of the beam: Finally, we compare the predicted and the measured force-
to-acceleration frequency responses for the U-channel beam. As
EI
d 4V
mω2V
bp x 0 ω
Fδ x
x f
(21)
shown in Figure 9, the second resonance in the unfilled beam
dx4
7
"!"! "!"! "!"! "!"!
120
accelerometer
"! "! #$ "! #$ "! #$ #$
acrylic tube
100
"!"! "!"! $#$# "!"! $#$# "!"! $#$# $#$#
plunger powder
$# $# $# $# (' ('
%&% %&% $# $# $# $# + ,+ ,+ ,+ ,+ ('(' ,+ ('('
80
&& +) +*), +*), +*), +, (' +, ('
Magnitude (dB)
60
)+ *),+ *),+ *),+ ,+ ,+
)+)*),+) *),+) *),+) ,+ ,+
40
8
−10 Damping and Isolation, G. Agnes, Ed., vol. 4697, SPIE,
−20
pp. 158–167.
−30
[12] Kinsler, L., Frey, A., Coppens, A., and Sanders, J., 2000.
Magnitude (dB)
−40
Fundamentals of Acoustics. John-Wiley.
−50
−60
[13] Beranek, L., and Ver, L., 1992. Noise and Vibration Con-
−70
trol Engineering: Principles and Applications. Wiley, New
−80
York.
50 100 200 300 400 500
[14] E-A-R S PECIALITY C OMPOSITES. EAR Technical Data
Sheet. 7911 Zionsville Road, Indianapolis, IN 46268.
0
[15] 3M A DHESIVES. Adhesive Data Sheet. St. Paul, MN
−100
55144.
−200
[16] Varanasi, K., and Nayfeh, S., 2003. “Vibration damping by
Phase (deg)