Williamsburg, Virginia

ACTIVE 04
2004 September 20-22

A New Approach to Identify Optimal Properties of Shunting

Elements for Maximum Damping of Structural Vibration
Using Piezoelectric Patches
Junhong Park1, Daniel L. Palumbo2,
National Research Council1, Structural Acoustics Branch2, NASA Langley Research Center,
Hampton, VA 23681-2199,
junhongpark@hotmail.com

ABSTRACT
The use of shunted piezoelectric patches in reducing vibration and sound radiation of structures has several
advantages over passive viscoelastic elements, e.g., lower weight with increased controllability. The performance of
the piezoelectric patches depends on the shunting electronics that are designed to dissipate vibration energy through
a resistive element. In past efforts most of the proposed tuning methods were based on modal properties of the
structure. In these cases, the tuning applies only to one mode of interest and maximum tuning is limited to invariant
points when based on den Hartog’s invariant points concept. In this study, a design method based on the wave
propagation approach is proposed. Optimal tuning is investigated depending on the dynamic and geometric
properties that include effects from boundary conditions and position of the shunted piezoelectric patch relative to
the structure. Active filters are proposed as shunting electronics to implement the tuning criteria. The developed
tuning methods resulted in superior capabilities in minimizing structural vibration and noise radiation compared to
other tuning methods. The tuned circuits are relatively insensitive to changes in modal properties and boundary
conditions, and can applied to frequency ranges in which multiple modes have effects.

1. INTRODUCTION
As advanced structural elements such as honeycomb panels are increasingly used in vehicles, total mass of the
structure continues to decrease, especially in aerospace applications [1]. The decrease in airframe mass can result in
increased cabin noise. To maintain interior noise levels, either existing treatment must be supplemented (adding
mass) or more efficient treatment must be developed. Structural vibration damping contributes to the minimization
of noise-related problems. To increase structural damping and to minimize noise generation, it is common practice
to apply viscoelastic materials to the structures, or to add acoustic treatments [2,3]. Recently, more active methods
utilizing smart components and electro-mechanical couplings have been proposed [4] and have been used in the
control of aircraft interior noise. Control actuators often proposed in such systems are piezoelectric patches, shape
memory alloys, and fluidic actuators. Among them the piezoelectric patches has been most widely used in structural
acoustic control. Active control systems using feedback or feed forward control techniques require supporting
electronics that typically consist of a digital signal processor system and components for sensing and actuating
imbedded elements. By utilizing passive shunting techniques as shown in Figure 1 the required hardware can be
significantly reduced with increased robustness.
The use of shunted piezoelectric patches to control structural vibration was first proposed by Forward [5] to
stabilize optical systems. Hagood and Flotow [6] presented a more complete analysis of the shunted damping of
structures. Effects from the electronic network were analyzed and incorporated in the variation of the elastic
properties of the patch. The performance of the piezoelectric patches in reducing the vibration response was
significantly related to the characteristics of shunting electrical circuits. The tuning of the values of the electronic
components was performed based on the shunt circuit’s effect on the structure’s modal response. Resonant shunt
circuits were optimized using den Hartog’s invariant point concept and the pole placement technique. Many
approaches have been proposed to optimally tune the shunting electrical networks [7-10]. In addition to RL resonant
circuits, various RLC circuits have been proposed [11-15]. One review of these methods can be found in the
reference [16]. However, most of the tuning methods are based on the modal properties of the structure when
treated with piezoelectric patches. Consequently, the tuning applies only to a limited number of modes at natural
frequencies. Also, the tuning of individual circuits, each attached to one of multiple piezoelectric patches, can be
complicated due to interaction between patches through the structure. Difficulty also arises when the circuit’s
bandwidth covers a frequency range in which intermodal couplings are not negligibly small due to high modal
density within the band.
In this study, the optimization of the shunting electrical networks is performed based on the wave propagation
characteristics of the structure and wave transmission characteristics through the piezoelectric patches. The shunting
electrical networks influences the dynamic mechanical properties of the patch. Although many approaches have
been proposed to separate the physical degrees of freedom between the modal properties and electrical charge,
combining the effects of the electrical circuit to the dynamic mechanical properties and deleting the physical degrees
of freedom related to electrical charge yield the same results since the physical quantities related to mechanical and
electrical variables are uncoupled in the system equations when external excitation is not applied to the electrical
circuit. The energy dissipated as the propagating structural waves transmit and are reflected from the piezoelectric
patch is calculated. The optimal tuning is obtained to maximize this structural energy dissipation. The effects of
location of piezoelectric patches on and boundary conditions of the structure are investigated. The performance of
the tuning criteria is confirmed through calculating the forced vibration of the beam and is compared to a
conventional method based on den Hartog’s invariant point concept. An active filter technique to implement the
optimal shunting criteria is proposed.

2. EFFECTS OF SHUNTING NETWORKS

When the piezoelectric patch is shunted as shown in Figure 1, the patch acts as a vibration absorber. In this
case, the effects of shunts on the piezoelectric materials can be analyzed using linear constitutive equation [17] as
⎡ D3 ⎤ ⎡ ε 3
T
d 31 ⎤ ⎡ E3 ⎤ (1)
⎢ =
⎥ ⎢ E ⎥⎢ ⎥
S d
⎣ 11 ⎦ ⎣ 31 s T
11 ⎦ ⎣ 11 ⎦

where D3 is the electrical displacement, S11 is the strain, E3 is the electrical field, T11 is the stress, d31 is the
piezoelectric constant, s11 is compliance at constant electrical field, and ε 3 is the electrical permeability at constant
E T

stress. The dynamic displacement of the patch is coupled to electrical charges through the piezoelectric effects
t
represented by the coefficients, d31. The voltage (V) and current (I) are related to electrical variables as V = E 3dz p

∫0
and I = D3 dA where tp is the thickness and A is the area of the patch. By taking the Laplace transform of the
∫ A
above relationships after assuming uniform spatial distribution of the electrical field and electrical displacement of
piezoelectric patch, equation (1) is converted to a more convenient definition to derive admittance of shunting
networks as [6]
T
⎡ I ⎤ ⎡ sC p sAd 31 ⎤ ⎡V ⎤ (2)
⎢ =
⎥ ⎢ E ⎥⎢ ⎥
S d t
⎣ 11 ⎦ ⎣ 31 p s T
11 ⎦ ⎣ 11 ⎦

where C Tp is the inherent capacitance of the patch and s is the Laplace parameter. In this case, shunting of the patch
has primary impacts on the dynamic stiffness of the patch. When the external electrical source to the piezoelectric
patch is not present (I=0), the effects of electrical network on the elastic moduli of the shunted patch can be
simplified as
E pE (sC Tp + Y SU ) , (3)
E pSU =
(sC Tp + Y SU ) − k312 sC Tp
where E pE (=1/ s11E ) is the elastic moduli of the patch at short circuit, Y SU is the electrical admittance of the shunting
networks, and k31 is the electromechanical coupling coefficient for transverse operation (= d 31 s11E ε 3T ) [6]. From
equation (3) variation of the complex moduli of the patch with the admittance of electrical circuit is obtained after
substituting s=iω where i = − 1 . When the usual complex notation is used, w( x, t ) = Re{wˆ ( x )e iωt } for w the
displacement, complex moduli are defined as,
[ ]
Eˆ pSU = E pSU,d (ω ) + iE pSU,l (ω ) = E pSU,d (ω ) 1 + iη PSU (ω ) , (4)
where E is the dynamic moduli, E is the loss moduli, and η is the loss factor. Arbitrary electrical circuits
SU
p ,d
SU
p ,l
SU
P

composed of multiple resistors, capacitances, and inductances in series or in parallel can be considered. The
bending stiffness of the beam treated by the patch is given as
Dˆ x1 = Eˆ b J b + Eˆ pSU J p = Dx1 (1 + iη DSU ), (5)

2
where Êb is the complex moduli of the beam material, Jb and Jp are the moments of inertia of the beam and the
patch, respectively. The bending stiffness of the beam itself is given as Dˆ x 2 = Eˆ b J b .
The variation of the complex modulus, Ê SU and the bending stiffness, Dˆ x1 , with respect to the admittance, YSU,
p

is most significant near the values which make the imaginary part of the denominator in equation (3) close to zero,
the resonance condition of the RLC circuit. Consequently, its variation is limited when the shunt is consisted only of
resistive elements. Only when the shunting circuit is composed both of reactive and resistive elements, wide
variation of the resulting complex modulus is possible. The dynamic modulus of the shunted patch can be negative
or positive. From the wide variation of the dynamic modulus and loss factor, an arbitrary value of the dynamic
bending stiffness, Dˆ x1 , is obtained. The condition of maximum loss factor or maximum loss moduli of the bending
stiffness does not result in minimum structural vibration response. Instead, the values that minimize the vibration
and sound generation of structures and maximize the vibration energy dissipation at the patch are desired in noise
control applications. To achieve this goal, the forced vibration and resulting sound generation of the beam may be
considered, which require estimation of variation of the modal properties depending on the shunting networks. In
this study, wave propagation characteristic of structural waves through piezoelectric patch is investigated to obtain
optimal shunting properties without requiring calculation of the modal properties.

3. WAVE PROPAGATION NEAR THE PIEZOELECTRIC PATCHES

To analyze reflection, transmission, and dissipation of the wave propagating near the piezoelectric patch, wave
propagations as shown in Figure 2 for different boundary conditions were considered. The edge of the beam in
Figures 2(a) and (b) at x=-L and x=-(L+Le), respectively, can be supported by geometric or general boundary
conditions. The beam was assumed to be infinitely long in the x-direction in order to allow for the calculation of the
reflection ratio of the normally incident bending waves as assumed in calculating reflection ratios at viscoelastically
supported edges [18]. The equation of motion for the free transverse vibrations (when effects of shear deformation
and rotary inertia are neglected) is,
∂ 4w ∂2w
Dx +M 2 =0, (6)
∂x 4
∂t
where Dx is the bending stiffness, w is the transverse displacement, and M is the mass per unit length. To calculate
the wave propagation characteristics at the edge, normally incident harmonic bending waves with complex
amplitude Ĉ22 were assumed to propagate toward the patch from x = ∞. The reflection, transmission and dissipation
of the incident vibration energy are calculated by obtaining the resultant wave propagation induced by this
disturbance near the patch. The calculation depends on various boundary conditions, for example the different cases
as shown in Figure 2.
When the piezoelectric patch is attached to the edge of the beam shown in Figure 2(a) the beam transverse
displacement is given as
wˆ (x ) = (Cˆ 11e − ik ( x + L ) + Cˆ 12 e ik x + Cˆ 13 e − k ( x + L ) + Cˆ 14 e k x )(1 − H ( x )) + (Cˆ 21 e − ik x + Cˆ 22 e ik x + Cˆ 23 e − k x )H ( x ) = w1 ( x )(1 − H ( x )) + w2 ( x )H (x ) (7)
1 1 1 1 2 2 2

where H is the Heavyside step function, k1 and k2 are the wavenumbers related to the circular frequency as
k1, 2 = (ω 2 M 1, 2 Dx1, 2 ) . A continuous neutral surface of zero bending stiffness through the discontinuities at x=0 is
14

assumed. By attaching the patch symmetrically with respect to the neutral surface of the beam, this assumption can
be satisfied. By applying the boundary conditions of the beam, the complex coefficients ( Ĉ11 - Ĉ14 , Ĉ21 , Ĉ 23 ) in
equation (7) are determined. Note that there are six unknowns with Ĉ22 an input variable. The boundary conditions
are
• at x=0
∂wˆ 1
wˆ 1 (0) − wˆ 2 (0) = 0 , (0) − ∂wˆ 2 (0) = 0 , (8a,b)
∂x ∂x
∂ 3 wˆ 1
(0 ) − Dˆ x 2 ∂ wˆ32 (0) = 0 , Dˆ x1 ∂ wˆ21 (0) − Dˆ x 2 ∂ wˆ22 (0) = 0 ,
3 2 2
Dˆ x1 (8c,d)
∂x 3
∂x ∂x ∂x
• at x=-L (for different geometric boundary conditions)
wˆ 1 (− L ) = 0 ,
∂wˆ 1
(− L ) = 0 , (for clamped edge) (8e,f)
∂x
∂ 2 wˆ 1 ( − L) ∂ 3wˆ 1 ( − L)
=0, = 0 , (for free edge) (8g,h)
∂x 2
∂x 3

3
 wˆ 1 ( −L) = 0 , ∂ w1 (2− L) = 0 . (for simply-supported edge)
2
ˆ (8i,j)
∂x
After replacing equation (7) into six boundary conditions in equation (8), the unknown coefficients are
obtained. Consequently, the response of the beam to external excitation of the incident bending waves of complex
magnitude Ĉ22 is calculated. The same numerical procedures are repeated for the piezoelectric patches attached to
different location of the beam. The differences are on the number of boundary conditions and the assumed beam
displacement functions. The beam displacements are given as
• when separated from edge (Figure 2(b))
( ) ( )
wˆ (x ) = Cˆ11e − ik1 ( x + L ) + Cˆ 12 e ik1 x + Cˆ13 e − k1 ( x + L ) + Cˆ14 e k1 x (H ( x + L ) − H ( x )) + Cˆ 21e − ik2 x + Cˆ 22 e ik2 x + Cˆ 23 e − k2 x H ( x ) , (9a)
( 31 32 33 34 )
+ Cˆ e −ik1 ( x + L + Le ) + Cˆ e ik1 ( x + L ) + Cˆ e − k1 ( x + L + Le ) + Cˆ e k1 ( x + L ) (1 − H ( x + L ))
• when far from edges (Figure 2(c))
( ) ( )
wˆ (x ) = Cˆ 11e − ik1 ( x + L ) + Cˆ12 e ik1 x + Cˆ13 e − k1 ( x + L ) + Cˆ 14 e k1x (H ( x + L ) − H ( x )) + Cˆ 21e − ik2 x + Cˆ 22 e ik2 x + Cˆ 23 e − k2 x H ( x ) . (9b)
( )
+ Cˆ e ik1 ( x + L ) + Cˆ e k1 ( x + L ) (1 − H ( x + L ))
32 34
Boundary conditions are defined in the exactly same way as shown in equation (8). The calculation can be
performed with and without taking into account material damping of the structure. When the patch is applied to
conventional metal structures, the damping in the structure itself is negligibly small in general. In such cases, the
dissipation of the vibrational energy through the piezoelectric patches is calculated as
⎧⎛
Cˆ 22 ⎞⎟ × 100
2 2

⎪⎪⎜⎝
1 − Cˆ 21 (for the patch attached near an edge, Figure 2 (a) and (b))
⎠ . (10)
Disspation(%) = ⎨
⎛ 2 2 2
⎞
⎪⎜1 − ⎛⎜ Cˆ 21 + Cˆ 32 ⎞⎟ Cˆ 22 ⎟ × 100 (for the patch far from the edges, Figure 2 (c))
⎪⎩⎝ ⎝ ⎠ ⎠
Maximum dissipation of the vibrational energy is directly related to minimization of the forced vibration and
sound radiation as illustrated for the case of a plate supported by viscoelastic elements [18]. Wave propagation
characteristics are derived by calculating the effects of the shunting electrical networks on the bending stiffness of
the beam. From these results, the optimal shunting network that maximizes the dissipation in equation (10) is
obtained numerically without requiring calculation of the modal properties.

4. BEAM VIBRATION CALCULATION

The performance of the piezoelectric patches in reducing the vibration of structures is estimated through
calculating the forced vibration of a beam of finite size from which tuning efficiency is predicted. Figure 3 shows
the beam supported by general boundary conditions and controlled by piezoelectric patch. By modifying the
stiffness of the springs supporting the beam, the vibration characteristic of the beam of arbitrary edge conditions is
analyzed [19, 20]. The kinetic, TK, and potential energy, VP, for transverse vibration of the beam is calculated using:
2 x2 2
⎛ ∂w ⎞ ⎛ ∂w ⎞
a

∫ (M p − M b )⎜
1 1 , (11a)
TK = ∫ Mb⎜ ⎟ dydx + ⎟ dydx
−a
2 ⎝ ∂t ⎠ x1
2 ⎝ ∂t ⎠

1 ⎛ ∂w(− a ) ⎞ 1 ⎛ ∂w(a ) ⎞ , (11b)

2 x 2 2 2 2
⎛∂ w⎞ ⎛∂ w⎞
a
⎟ dx + ∫ (D p − D b )⎜
2 2 2
1 1 1 1 ⎛a⎞
VP = ⎟ dx + s t1 w(− a ) + s r1 ⎜ ⎟ + s t 2 w⎜ ⎟ + s r 1 ⎜
2
∫ D b ⎜⎜ 2 ⎟ ⎜ 2 ⎟
⎟
−a
2 ⎝ ∂x ⎠ x1
2 ⎝ ∂x ⎠ 2 2 ⎝ ∂x ⎠ 2 ⎝ 2 ⎠ 2 ⎝ ∂x ⎠
a
1
W F = ∫ f ( x , t )w( x , t )dx . (11c)
−a
2
In the Rayleigh-Ritz method, the transverse displacement of the beam is approximated as

∑φ (x)α (t ),
N
w( x, t ) = (12)
m m
m =1

where φm are the trial functions chosen from a complete set, and αm are the generalized coordinates. After
substituting equation (12) into equation (11), Lagrange’s equations of motion,
d ⎛ ∂LS ⎞ ∂LS
⎜ ⎟− = 0, m = 1,2,..., N , (13)
dt ⎜⎝ ∂α& m ⎟⎠ ∂α m
are applied, where LS = TK-VP+WF is the system Lagrangian. This yields a set of equations of motion
[M ]{α&&}+ [K e + K V ]{α } = { f }, (14)

4
where [M], [Ke], and [KV] are the mass and stiffness (from springs at edges and beam strain energies) matrices ,
respectively. When the harmonic excitation of the beam is considered with time dependence of e-iωt, the equations
of motion are
− ω 2 [M ]{α&&}+ [K e + K V ]{α } = { f }. (15)
The polynomial functions are used as trial functions as
φ m ( x ) = x m , m = 0, N . (16)
Consequently, the mass and stiffness matrices are calculated as
[M ] = m +Mp + 1 (a
mp
b m + p +1
− (− a )
m + p +1
) + (Mm + −p M+ 1 ) (x
p b m + p +1
2 − x1m + p +1 ), (17a)

⎧0, m < 2 or p < 2

[K ]
V
mp
⎪
= ⎨ Db m (m − 1) p( p − 1) m + p −3
(
a (
− −a )m + p −3
+ ) ( )
D p − Db m(m − 1) p( p − 1) m + p −3
x2 m + p −3
− x1 ( )
otherwise
, (17b)
⎪ m+ p−3 m+ p−3
⎩
⎧ (− a )m (− a ) p + s t 2 a m a p , m = 0 or p = 0
[K ] = ⎪⎨s
e t1 , (17c)
⎩ s t1 (− a ) (− a ) + s t 2 a a + s r1 mp (− a ) (− a ) + s r 2 mpa a , otherwise
mp m −1 p −1 m −1 p −1
⎪
m p m p

a
[ f m ] = ∫ f (x, t )φ m ( x )dx . (17d)
−a

5. RESULTS AND DISCUSSIONS

If the shunting elements are not present in Figure 1, there is no energy loss in the system when the material
damping is small. Appreciable damping occurs only when electrical energy produced through piezoelectric effects
is dissipated into the electrical network. The patch with resistive shunting element acts as a control device of the
system to absorb and dissipate vibration energy and to minimize undesirable vibration of the structure and noise
generation especially when the external excitation has a broadband characteristic. Two different shunting networks
(resistive and resonant) are considered. To compare results with previously published data, a beam structure and a
piezoelectric patch of similar dimensions and material properties to those in the previous publication [6] are used in
this study. (L = 6.2 cm, a = 14.65 cm, t = 3.17 mm, EbE =73 GPa, E pE = 63 GPa, C pS =0.156 µf, ε eT =1700 ε°, k31 =
0.35, d31 = 180e-12 m/v).

A. Resistive shunting
When the shunting network consists only of a resistor, the resistor increases the material damping of the
piezoelectric patches by dissipating converted electrical energy into heat. Using the wave propagation approach, the
amount of energy dissipation is calculated as presented in section 2. Figure 4 shows the energy dissipation on
patches attached to different locations of the beam at the fundamental frequency. Figure 4 (a) shows the variation
with separation distance from the clamped edge of the beam. The maximum dissipation occurs when the patch is
attached to the clamped edge of the beam. With increasing separation distance, the energy dissipation is decreased.
The maximum dissipation occurred at approximately the same resistance values regardless of the position of the
patch relative to the edge. Obviously, the optimal position of the patch to maximize its performance to induce
maximum energy dissipation is near the clamped edge. Figure 4 (b) shows the comparison for the patch attached to
edge of the beam with different boundary conditions – clamped, free, simply-supported (Figure 2 (a)), and far from
both edges (Figure 2 (c)). The maximum amount of energy dissipation occurred when the edge is clamped. For
other boundary conditions, the energy dissipation was smaller. The energy dissipation was negligibly small when
the patch is attached to the free edge.
Figure 5 shows the variation of the energy dissipation with frequency and resistance when the patch is attached
to the beam’s edge (Figure 1(a)). For each frequency, there is one maximum in the amount of the energy
dissipation. When the patch is attached to the free edge (Figure 5(b)), non-negligible energy dissipation occurs only
at high frequencies where the length of the patch is comparable to the excitation wavelength (λ=10 cm at f=2 kHz).
Figure 6 shows optimal resistance values taken from Figure 5 at the point of maximum energy dissipation. The
optimal values are relatively insensitive to boundary conditions, and are very similar to the values that induce
maximum material damping of the patch and the maximum modal damping of the beam [6].
For the structure and patch considered in this study, the amount of energy dissipation obtained using purely
resistive shunting is generally less than 5 %. This amount of damping does not justify the added cost and

5
complexity when compared to passive techniques [18]. To increase the dissipation, the current flowing through the
resistor must be amplified. This is accomplished with resonant shunting circuits.

B. Resonant shunting
When a resonant shunting element is used, the condition of the maximum dissipation is very sensitive to circuit
parameters. Optimization of the resonant network is not as straight forward as the resistive network. Until now
many optimization techniques have been proposed. When the resonant shunting is used, the resonance due to the
capacitance of the patch and the inductance of the shunting network significantly increases the current flowing
through the resistance and consequently increases the energy dissipation and damping capabilities of the patch. To
investigate these effects, the RLC circuit shown in Figure 1 is considered. Other configurations, such as RLC in
parallel as discussed in several other publications, are also possible.
Figure 7 shows the variation of the vibration energy dissipation with R and L when C=0 at fundamental
frequency of the beam (f=33.5 Hz). Lres is the resonant inductance value, (ω 2C pS )−1 . The calculation was repeated
for different boundary conditions of the beam. In the simulation results, the performance of the patch in dissipating
the vibration energy when the boundary condition is clamped is better than that of any other boundary conditions.
When the boundary condition is clamped or infinite, there are values of R and L that induces completely dissipation
of the incident vibrational energy near R≈300 Ω and L=0.97 Lres at fundamental frequency. This condition
corresponds to the values that lead the dynamic stiffness, Dˆ x1 , to be negative and the loss factor to be positive. As
the resistance and inductance values deviate from the optimal value, the amount of vibrational energy dissipation
starts to decrease very rapidly. Dissipation is more sensitive to the inductance value. The optimal inductance value
is smaller than the resonant inductance by 1-2 %.
In Figure 7(a) and (d), there are another values of R and L than induces complete dissipation of the incident
vibrational energy near R≈5 Ω and L=0.96 Lres. This condition results in positive dynamic moduli and loss factor of
Dˆ x1 . However, the occurrence of this optimal value is extremely sensitive to the inductance values, L, and requires
very small value of R which may not easy to implement using conventional RLC circuit in actual situations
especially when the required inductance value is large. In the following studies, this condition was not considered
and only the optimal values that induce complete vibration energy dissipation with negative dynamic moduli of Dˆ x1
was selected as the optimal values.
Figure 8 shows the variation of these optimal inductance and resistance with frequency. The values obtained
from the den Hartog’s invariant point concept are plotted also. The optimal values are similar to each other, but
considerable difference exists in the values of the resistance. To implement the condition calculated in Figure 8,
frequency dependent variation of the resistance and inductance is required. To satisfy this condition, it is more
straightforward to consider frequency-dependent admittance values. Figure 9 shows the admittance of the electrical
circuit that results in complete absorption of the incident vibrational energy. The variation of the admittances when
the shunting elements are composed of constant values of R and L is also plotted. The optimal value of R and L was
taken as the value that completely absorbs the vibration energy at the first resonance of the beam under
consideration. Note that the desired value of admittance increases with increasing frequency. However, the
admittance of the RL-circuit decreases with increasing frequency. Consequently, the performance of the shunting
network decays very rapidly as the excitation frequency deviates from the design frequency. This is an inherent
disadvantage of the shunting circuit composed only of a resistor and an inductor even when its performance is
improved by adopting the tuning method to minimize the modal response peaks, as in den Hartog’s invariant point
concept. To resolve this problem, an RLC circuit was considered to generate the increasing admittance with
increasing frequency.
The values of R, L, and C were determined by matching the optimal admittance values (real and imaginary
parts) and the first derivative of the imaginary part with respect to frequency at the tuning frequency. Matching the
derivatives of the imaginary parts was necessary due to the strong dependence of performance on the reactance
value. For the beam and the patch considered in this study, the values were determined as R=299 Ω, L=0.4460 H,
and C=-0.1615 µF. A negative capacitance was required to achieve the desired variation of the admittance with
frequency. The results are shown in Figure 10. The admittance follows the desired optimal values very closely near
the tuning frequency. A negative capacitance could be implemented with an active filter such as a negative
impedance converter. However, such a design would require an external power source.

6
 C. Forced vibration of beam
To compare the capabilities of the various tuning criteria, the forced vibration response of a finite beam
controlled by the shunts located at the clamped edge was considered. Figure 11 shows the forced vibration response
calculated using the Rayleigh-Ritz method presented in section 4 for different values of elements in the RLC shunt
circuit. When the admittance matches the optimal values over the entire frequency range, the modal response of the
beam completely disappears and the beam response closely resembles that of an infinite beam. This suggests that
the incident vibrational energy to the piezoelectric patch is completely absorbed and is dissipated into heat. When
the RLC circuit employing negative capacitance is used to approximate the frequency dependent admittance, as in
Figure 10, the performance is limited to a finite frequency range. The first (fundamental) modal response of the
beam completely disappears. The shunting is also effective in reducing the response of the second and third modal
response. At higher frequencies the damping capability rapidly decreases since the admittance deviates from the
optimal values calculated here when the frequency increases from the design frequency. Compared to tuning
methods based on the modal properties and den Hartog’s invariant point concept, the current method based on the
wave propagation approach is superior in regulating the forced vibration response of the beam.

6. CONCLUSIONS AND FUTURE WORK

The absorption and subsequent dissipation of structural vibration energy by piezoelectric patches was
investigated. Dissipating converted electrical energy through piezoelectric effects into shunting electrical networks
enhanced the damping of structural vibrations. Performance was strongly dependent on the parameters of the
shunting elements. Optimization of these parameters was performed based on wave propagation characteristics, i.e.,
absorption, transmission, and reflection of bending waves propagating toward the piezoelectric patches attached to
various locations of the beam. Following this approach, the modal properties of the beam were not considered in
obtaining the optimal properties. The method applies to a broad range of frequencies and to frequency ranges
affected by a large number of modes. The coupling between patches through modal response interaction of the
beam can be neglected in the presented tuning method.
When a resistive shunting element was used, the vibration energy dissipation was maximum when the patch
was attached to a clamped edge as opposed to the simply supported or free edge. When the patch was attached to
the free edge, energy dissipation occurred only at high frequencies where the dimension of the patch was
comparable to the wavelength of the propagating bending wave. The complete absorption of the incident vibration
energy occurred when a resonant shunting network was used. The inherent disadvantage of using RL networks with
frequency-independent resistance and inductance values was identified –the admittance of constant RL circuit
decreases with increasing frequency while the calculated optimal admittance in this study increases with increasing
frequency. Complete absorption was obtained at the design frequency when an RLC circuit with negative
capacitance was employed. Implementation of a negative capacitance requires active filters such as a negative
impedance converter. The RLC circuit designed according to wave propagation characteristics resulted in much
better structural vibration control compared to conventional methods based on modal properties of the beam.
Without requiring complicated control systems, such as feedback and feedforward controllers, the piezoelectric
patches can achieve significantly improved performance by using a simple RLC circuit with appropriate tuning.
Although, this technique was demonstrated here on a beam shunted with piezoelectric patch using RL or RLC
electrical networks, the proposed numerical procedures can be applied to different electrical networks with minor
modifications. Similar optimization procedures can be applied to a shunt patch attached to a plate by neglecting
Poisson effects as for the optimization of the boundary stiffness of a viscoelastically-supported plate [18]. The
experimental confirmation of the developed optimization method is in progress.

7. ACKNOWLEDGEMENTS
The first author gratefully acknowledges the financial support from the NASA Langley Research Center while
in residence under a National Research Council Post-doctoral Research Associateship Award.

References
1. D. Zenkert An Introduction to Sandwich Construction (Chameleon Press Ltd., London, 1997).
2. A. Cummings, H.J. Rice, R. Wilson, “Radiation damping in plates, induced by porous media,” J. Sound Vib.
221, 143-167 (1999).
3. J.S. Bolton, N.M. Shiau, Y.J. Kang, “Sound transmission through multi-panel structures lined with elastic
porous materials,” J. Sound Vib. 191, 317-347 (1996).
4. G.P. Gibbs, R.H. Cabell, D.L. Palumbo, R.J. Silcox, and T.L. Turner “Recent advances in active noise and
vibration control at NASA Langley Research Center,” J. Acoust. Soc. Am. 112, 2346 (2002).

7
5. R.L. Forward, “Electronic damping of vibrations in optimal structures,” Appl. Optics. 18, 690-697 (1979).
6. N.W. Hagood and A.von Flotow, “Damping of structural vibrations with piezoelectric materials and passive
electrical networks,” J. Sound Vib. 146, 243-268 (1991).
7. G.S. Agnes, “Development of a modal model for simultaneous active and passive piezoelectric vibration
suppression,” J. Intel. Mat. Syst. Str. 191, 317-347 (1996).
8. J.J. Hollkamp, “Multimodal passive vibration suppression with piezoelectric materials and resonant shunts,” J.
Intel. Mat. Syst. Str. 191, 317-347 (1996).
9. J. Tang and K.W. Wang, “Active-passive hybrid piezoelectric networks for vibration control: comparisons and
improvement,” Smart Mater. Struct. 10, 794-806 (2001).
10. M.B. Ozer and T.J. Royston, “Passively minimizing structural sound radiation using shunted piezoelectric
materials,” J. Acous. Soc. Am. 114, 1934-1946 (2003).
11. C.H. Park and D.J. Inman, “Enhanced piezoelectric shunt design,” Shock and Vibration. 10, 127-133 (2003).
12. G. Caruso, “A critical analysis of electric shunt circuits employed in piezoelectric passive vibration damping,”
Smart Mater. Struct. 10, 1059-1068 (2001).
13. A.J. Fleming and S.O.R. Moheimani, “Adaptive piezoelectric shunt damping,” Smart Mater. Struct. 12, 26-48
(2003).
14. C.H. Park, “Dynamics modeling of beams with shunted piezoelectric elements,” J. Sound Vib. 268, 115-129
(2003).
15. C.L. Davis and G.A. Lesieutre, “Ac actively tuned solid state vibration absorber using capacitive shunting of
piezoelectric stiffness,” J. Sound Vib. 232, 601-617 (2000).
16. G.A. Lesieutre, “Vibration damping and control using shunted piezoelectric materials,” Shock Vib. Dig. 30,
187-195 (1998).
17. B. Jaffe, R. Cook, and H. Jaffe, Piezoelectric ceramics, (Academic Press, New York, 1971).
18. J. Park, T. Siegmund, and L. Mongeau, “Influence of support properties on the forced vibrations of rectangular
plates,” J. Sound Vib. 264, 775-794 (2003).
19. A. Berry, J.-L. Guyader and J. Nicolas, “A general formulation for the sound radiation from rectangular, baffled
plates with arbitrary boundary conditions,” J. Acous. Soc. Am. 88, 2792-2802 (1990).
20. J. Park, T. Siegmund, and L. Mongeau, “Analysis of the flow-induced vibrations of viscoelastically supported
rectangular plates,” J. Sound Vib. 261, 225-245 (2003).

8
 R L C
F

Piezoelectric patches

Figure 1. Shunted damping of flexural vibration using a

piezoelectric patch.
(a) (b)
Cˆ11 , Cˆ12 , Cˆ13 , Cˆ14 Cˆ11 , Cˆ12 , Cˆ13

L
x
(a)
Cˆ 31 , Cˆ 32 , Cˆ 33 , Cˆ 34 Cˆ 11, Cˆ12 , Cˆ 13 , Cˆ14 Cˆ11, Cˆ12 , Cˆ13

Le L
x (c) (d)
(b)
Cˆ 32 , Cˆ 34 Cˆ11 , Cˆ12 , Cˆ 13 , Cˆ14 Cˆ11 , Cˆ 12 , Cˆ13
Figure 5. Dissipation of incident vibration energy vs.
frequency and resistance for resistive shunt and different
boundary conditions at edges of beam: (a) clamped, (b) free,
L and (c) simply supported; and (d) far from both edges.
x
(c)
Figure 2. Wave propagation near the piezoelectric patch
attached to different locations: (a) at edge, (b) near an edge,
and (c) far from edges.

2a
x2
Optimal resistance

s r1 x1 x sr 2

clamped
free
st1 st 2 simply-supported
infinite
Figure 3. Flexural vibration of the beam under generally maximum loss factor
supported boundary condition. maximum modal damping
3
10
10
f

1.2 1.2
Le: , 0; Clamped Figure 6. Optimal resistance calculated from two different
1.0 1.0 Free
0.8
, 0.2 L;
, 0.4 L;
Simply-supported tuning criteria: maximum dissipation of incident vibrational
0.8 Infinite
, 0.6 L; energy and maximum loss factor.
Dissipation (%)

Dissipation (%)

0.6 , 0.8 L; 0.6

,L
0.4 0.4

0.2 0.2

0.0 0.0

-0.2 -0.2
1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8
10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10
R (Ω) R (Ω )

(a) (b)
Figure 4. Vibration energy dissipation of piezoelectric patches
shunted by resistive networks. (a) Effect of separation
distance from clamped edge, and (b) effects of several
different boundary conditions.

9
 1E-3

1E-4

1E-5

Admitance (1/Ω)
1E-6

1E-7

(a) (b) 1E-8 optimal: Yr

-Yi
1E-9 from RL circuit: Yr
-Yr
1E-10
10 100 1000
f (Hz)

Figure 9. Optimal variation of the admittance to induce

complete absorption of incident bending waves, and the
admittance of the RL-circuit.
(c) (d)
Figure 7. Dissipation of incident vibration energy vs.
inductance and resistance for RL shunt and different boundary 1E-3

conditions at edges of beam: (a) clamped, (b) free, and (c) 1E-4
simply supported; and (d) far from both edges. f=33.5 Hz.
1E-5

Admitance (1/Ω)
1E-6
10000

1E-7
1000
1E-8 optimal: Yr
-Yi
R (Ω), L(H)

100 1E-9 from RLC circuit: Yr

-Yr
1E-10
10 10 100 1000
f (Hz)

Non-reflecting conditions : , R; L Figure 10. Optimal variation of the admittance to induce

0.1
Den-Hartog's invariant concept : , R; ,L complete absorption of incident bending waves, and their
10 100 1000
approximation using RLC circuit (with negative capacitance).
f (Hz)

Figure 8. Optimal values of resistance and inductance to 10

-1

short cirbuited
induce complete absorption of incident bending waves at -2
10
piezoelectric patches attached to the clamped edge of the beam den Hartog's invariant points concept
and its comparison to values calculated to minimize the modal 10
-3

response using den Hartog’s invariant point concept [6].

w(x=L)/F

-4
10

-5
10

-6
10
optimal shunting
-7
10
shunting using negative capacitance
-8
10
10 100 1000
f (Hz)

Figure 11. Damping of beam vibration using piezoelectric

patches shunted by resonant shunting networks and
comparison of different optimization criteria.

10