Email alerting service Receive free email alerts when new articles cite this article - sign up in
the box at the top right-hand corner of the article or click here
An adaptive tuned vibration absorber (ATVA) with a smart variable stiffness element is
capable of retuning itself in response to a time-varying excitation frequency, enabling
effective vibration control over a range of frequencies. This paper discusses novel methods
of achieving variable stiffness in an ATVA by changing shape, as inspired by biological
paradigms. It is shown that considerable variation in the tuned frequency can be achieved
by actuating a shape change, provided that this is within the limits of the actuator.
A feasible design for such an ATVA is one in which the device offers low resistance to the
required shape change actuation while not being restricted to low values of the effective
stiffness of the vibration absorber. Three such original designs are identified: (i) A pinned–
pinned arch beam with fixed profile of slight curvature and variable preload through an
adjustable natural curvature; (ii) a vibration absorber with a stiffness element formed from
parallel curved beams of adjustable curvature vibrating longitudinally; (iii) a vibration
absorber with a variable geometry linkage as stiffness element. The experimental results
from demonstrators based on two of these designs show good correlation with the theory.
Keywords: vibration absorber; vibration control; smart structures; biomimetics
1. Introduction
The tuned vibration absorber (TVA) has been used for many years since its
inception by Ormondroyd and Den Hartog (1928). The device is an auxiliary
spring–mass–damper system whose natural frequency (‘tuned frequency’) is tuned
to suppress the vibration at its point of attachment to a host structure. The TVA
can be used in two distinct ways, depending on the application (von Flotow et al.
1994). When used as a ‘mass damper’, the TVA is tuned so as to dampen the
contribution of a particular mode of the host structure to the vibration over a wide
range of excitation frequencies. In this application, an optimal damping level of
(a) (b)
me me
w = wt= K me
host
structure
f (t) = Re(Fe jw t)
Adaptive tuned vibration absorber with variable shape stiffness element 3957
Various designs for variable stiffness elements in ATVAs have been proposed
(e.g. Hong & Ryu 1985; Longbottom et al. 1990; Franchek et al. 1995; Carneal et al.
2004). Rustighi et al. (2005) recently achieved variable stiffness by using a stiffness
element made of a shape memory alloy, whose Young modulus changes
with temperature, and so can be controlled by adjusting the electrical current
through it. Such a device is easy to manufacture, requiring no actuating mechanism
for stiffness adjustment. However, the maximum variability of the tuned frequency
of such a device is typically found to be around 20% and, for the device in Rustighi
et al. (2005), it took 2 min with a 9 A current to achieve this change. Hence, such a
device is only suitable for applications where any potential mistuning is gradual.
Another strategy for stiffness adjustment involves controlling the shape of the
stiffness element. One such device, first proposed by Walsh & Lamancusa (1992),
had a stiffness element composed of a pair of beams with adjustable gap between
them. This design was modified by Brennan (2000) and developed by Kidner and
Brennan (2002), who used a servomotor to adjust the gap automatically.
A maximum adjustment of 35% in tuned frequency was achieved and the
response was much more rapid than that in Rustighi et al. (2005). However, the
actuating mechanism was still relatively slow to operate and added considerably
to the redundant mass of the device, thereby degrading the vibration
attenuation.
This paper explores the concept of shape control for achieving variable
stiffness in an ATVA, presenting novel ideas inspired by biological paradigms:
changing curvature to adjust stiffness, as observed in plant stems and leaves
(Vincent & Jeronimidis 1991), and controlling stiffness by adjusting geometry, as
in musculo-skeletal structures like the flight motor of an insect (Brennan et al.
2003). Variable curvature is considered first, followed by variable geometry.
Effective designs based on these principles are then identified and results from
two experimental demonstrators, based on variable curvature and variable
geometry, respectively, are presented.
The aim of this section is to consider some designs for ATVAs using the curvature
control concept and to discuss how this could be realized.
Figure 2. Dynamics of pinned shallow arch with end longitudinal restraints of finite stiffness.
and PEL Z p2 EI =L2 is the Euler critical load. Equations (2.1) and (2.2) and the
subsequent analysis assume the arch is shallow, i.e. ðwi0 Þ2 ; ðwP0 0 Þ2 / 1; where
ð Þ 0 Z dð Þ=dx. For small vibrations vZ wKwP0 about the preloaded configuration
wP0 ðxÞ (figure 2c), the neutral surface of the beam is strained by an amount
depending on the stiffness KS of the longitudinal end restraints. In reference
Bonello et al. (2004) it is shown that, for vibrations symmetrical about the mid-
section, the equation of motion is given by
ÐL 0 0
iv 00 00 0 wP0 v dx
EIv C v P0 KwP0 C gv€ Z 0; ð2:3Þ
2f1=2KB Þ C 1=KS g
where E, I, A, g are the Young modulus, second moment of area of cross-section
about neutral axis, cross-sectional area and mass per unit length, respectively, of
the curved beam and KBZEA/L, the axial stiffness of the straightened beam.
Equation (2.3) neglects the effects of inertia in the longitudinal direction. This
equation is a generalization of the case considered in (Reissner 1954) of a pinned–
pinned shallow arch with full end longitudinal restraint and no preload. In that
work, it was shown that the longitudinal inertia was negligible. Equation (2.3)
also applies for vibration that is not symmetric about the mid-section, provided
that KSZN. If vibration in the first symmetric mode is considered, the response
is given approximately by
v Z A sin ðpx=LÞ sin u1 t: ð2:4Þ
Substituting equations (2.4) and (2.1b) in equation (2.3) and using equation (2.2)
gives the equation governing the tuned frequency of a potential TVA consisting
Adaptive tuned vibration absorber with variable shape stiffness element 3959
350
0 1 2 3 4
H h
Figure 3. Variation of natural frequency with curvature of a shallow pinned arch of rectangular
section 2h with no preload and KSZN.
It is assumed that the vibration of the TVA is symmetrical about the mid-section
of the arch. However, in practice, due to inherent slight asymmetry in conditions,
an antisymmetric mode of vibration will also be excited, becoming significant
when the frequency given by equation (2.7) becomes close to the frequency
u2 rad sK1 of the first antisymmetric mode of the pinned–pinned beam. In this
case the TVA will not reduce to the simple model of figure 1b and will cease to
function effectively. Substituting vZ A sin ð2px=LÞ sin u2 t in equation (2.3) with
KSZN, along with the expression for the deformed profile about which vibration
occurs (equation (2.1b)), gives, after some algebra:
u22 Z u2s ð16K4P0 =PEL Þ Z u2s ð12 C 4H =HP0 Þ: ð2:8Þ
(by substituting for P0/PEL from equation (2.2)). The condition u2t ! u22 limits
the variation of H and HP0 in the ATVA since, from equations (2.7) and (2.8):
ðHP0 =rÞ3 K24ðHP0 =rÞK6ðH =rÞ! 0 for u2t ! u22 : ð2:9Þ
The methods (i) and (ii) for varying ut are now considered in turn.
Adaptive tuned vibration absorber with variable shape stiffness element 3961
(a) (b)
slight adjustment in distance fixed distance
HP HP
0 0
0
150 H/h=2 –1
H / h = 2.5 –2
100
–3
P0 H/h=1
–4
50
PE L –5 H/h=2
0 –6
H / h = 2.5
–7
– 50 –8
–9
–100 – 10
0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
HP h HP h
0 0
(c) 5
HP 3
0
h
2
0
0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
H h
Figure 5. Performance curves for shallow arch ATVA with fixed H, variable HP0 (figure 4a).
over a range of HP0 =h (equation (2.9)) is given by the solid line in figure 6c, where
the dashed line represents
pffiffiffiffiffithe undeformed condition and intersects with the solid
one at HP0 =hZ H =hZ 10 as before. It is seen from figure 6a that this technique
works very well for slight curvatures HP0 =h. Moreover, for such slight curvatures,
there are no limits on the variation of H/h, as far as antisymmetric mode
interference is concerned, since the lower limit on H/h is negative in figure 6c.
Unlike the technique in the previous method §2a(i), the preload P0 is provided by
the support reaction, i.e. the actuator does not act against the high longitudinal
stiffness of the beam. Moreover, the actuation would be incorporated into the
beam making the design simple, cheap and efficient.
(c) 50
40 lower limit on H / h (equation (2.9))
HP = H
30 0
H
h 20
10
–10
0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
HP h
0
Figure 6. Performance curves for shallow arch ATVA with fixed HP0 , variable H (figure 4b).
It is worth pointing out that either device depicted in figure 4 is limited by its
small effective mass. However, this limitation can be overcome to some extent by
the addition of a dense mass at the apex.
Adaptive tuned vibration absorber with variable shape stiffness element 3963
(a) (b)
2L
E1
A a/2
R h
neutral axis h
N N
two layer materials. Upon locating axis NN using elementary composite beam
theory, analytical expressions for (INN)1, (INN)2 are determined from the
geometry of figure 7b (Bonello et al. 2004):
ðINN Þ1 Z ðSh3 =8Þ U 3 F1 C UF2 C U K1 F3 C U K1 ð1KDÞ2 F4 ; ð2:12aÞ
3
3 K1 K1 2
ðINN Þ2 Z ðSh =8Þ V F1 C VF2 C V F3 C V ð1 C DÞ F4 ; ð2:12bÞ
C Z E2 =E1 ; ð2:13aÞ
D Z ð1KC Þ=ð1 C C Þ; ð2:13bÞ
2
F1 Z ð1=2Þðh=SÞ½a C sin aKð8=aÞsin ða=2Þ; ð2:14aÞ
2
F2 Z ð1=2Þðh=SÞ½a C sin aKð16=3Þð1=a sin ða=2Þ; ð2:14bÞ
F3 ZKð4=9Þðh=SÞð1=aÞsin2 ða=2Þ; ð2:14cÞ
F4 Z ð128=9Þðh=SÞ
4 K2
9 S 6 S 2 2 S
!ð1=aÞ 4 K 2 C1 KD sin2 ða=2Þ; ð2:14dÞ
a h a h a h
U Z ð2=aÞðS=hÞK1; ð2:14eÞ
V Z ð2=aÞðS=hÞ C 1: ð2:14f Þ
For a flat cross-section, the expression in (2.11) reduces to
I aZ0 Z S aZ0 h 3 =24 2ð1 C C Þ C 6ð1KDÞ2 C 6Cð1 C DÞ2 : ð2:15Þ
Three methods of changing the cross-section curvature that were considered are
as follows.
(a) x (b)
My Ny
y z
Mx
Mx
Nx
Nx
My Ny
laminate exploded view of laminate
(b) Using the stretch-bend effect of cross-ply laminates (Jones 1975). With
reference to figure 8, if the cross-section is a cross-ply laminate consisting of
two orthotropic layers of the same composite material that are oriented such
that their fibres are at 908 to each other, then a tensile force Nx per unit
length along an edge will produce curvature Kx in a plane normal to the edge
according to the equations
" # " # " #" #
Nx A11 A12 30x B11 0 Kx
Z C ; ð2:16aÞ
Ny A12 A22 30y 0 B22 Ky
" # " # " #" #
Mx B11 0 30x D11 D12 Kx
Z C ; ð2:16bÞ
My 0 B22 30y D12 D22 Ky
where the axes x and y are aligned with the axes of symmetry of material
properties, 30x, 30y are the longitudinal strains in the mid-surface and Kx, Ky
the curvatures and the matrix terms are defined in Jones (1975). It is shown
in Jones (1975) that the coupling coefficients between stretching and
bending are maximum for given material properties when the two layers are
of equal thickness h.
(c) Using piezo-actuation. This is discussed later, in §2d (ii).
Adaptive tuned vibration absorber with variable shape stiffness element 3965
100 1.0
50 0.5
0 0 5 10 15 20 25 30 35 40 45 50
5 10 15 20 25 30 35 40 45
a (deg.) a (deg.)
Figure 9. (a) Tuning characteristic for cantilever type TVA in figure 7 estimated from equations
(2.10)–(2.15) for a 2-layer cross-ply cross-section with E1, E2Z130, 10 GPa, y12Z0.3, 2hZ2!10K3 m,
2LZ0.24 mm; (b) Edge force required to achieve cross-section curvature in figure 9(a).
in tuned frequency are possible, provided that the actuator can provide the
necessary changes to the subtended angle a. However, the slope of the graph in
figure 9b is seen to be enormous, indicating that stretching the composite cross-
section is not a practical way of varying the cross-section.
The main drawback with the design in figure 7 is that significant variations
in the subtended angle a are required over a short length S. Since the curvature
KZa/S, this means that the required curvature changes are too high for method
(b). It is also shown in §2d(ii) that piezo-actuation cannot yield the curvature
changes required for this application. Indeed, of methods (a)–(c) listed above,
(a) appears to be the only viable option for this application.
Hence, for the above reasons the concept of anticlastic curvature control for
variable stiffness in figure 7 was not pursued any further.
attach to structure
vibration absorber beam geometry
Figure 10. Vibration absorber with stiffness element formed from parallel curved beams.
(i) Composites
One way of implementing the stretch-bend effect in composites is to use
hydraulic pressure to generate the required tension force along the edge.
Figure 11 shows an initially straight clamped-free beam in which the upper and
lower surfaces are orthotropic laminae of the same composite material and are
oriented such that their fibres are at 908 to each other. The two laminae are
Adaptive tuned vibration absorber with variable shape stiffness element 3967
0.01 m
0.12 m
0.01 m
Figure 11. Finite element model of clamped-free beam with a pressurised fluid-filled cavity
separating a pair of laminae of orthotropic material with material axes at right angles to each other
(thickness of each lamina is 0.25 mm).
separated by a fluid filled cavity. Fibre reinforcements joining the upper and
lower surfaces prevent them from bulging. Figure 11 shows the FE model of the
deformation resulting from the application of a 1 bar pressure. For the
dimensions shown, and a Poisson ratio (y12Z0.3, the tip deflections are
0.13 mm, 0.145 mm, 0.274 mm, respectively, for major and minor Young moduli
(E1, E2) of (130 GPa, 10 GPa), (10 GPa, 5 GPa), (100 GPa, 5 GPa). If such a
beam were incorporated into the devices of figure 4 and figure 10 then these
deflection values would be the changes in the crown height H for a 1 bar change
in pressure. Hence, it is expected that a pressure increase of, say, 20 bar would
result in significant variation to the tuned frequency.
The beam in figure 11 is somewhat analogous to the stem of a plant
(Vincent & Jeronimidis 1991). As in the present case, the turgor pressure in a
stem (typically 20 bar; Vincent & Jeronimidis 1991) is uniform. However, the
bending of the stem is due to the variation of cell wall thickness across the cross-
section, resulting in differential cell wall expansion, and the curvature changes
measured in (Vincent & Jeronimidis 1991) were considerably higher than those
obtained with the composite model of figure 11.
(ii) Piezo-actuation
A piezo-actuator bends a beam to which it is bonded by applying a moment.
The piezo-patch(es) can be placed on both parallel surfaces of the beam with
opposing polarities or one surface only. With one-sided piezo-actuation there is a
tensile force on the beam mid-surface as well. Hence, it is worthwhile
investigating whether the use of a composite beam with a piezo-actuator on
one surface would have any advantage with respect to curvature generation.
If 3p is the longitudinal expansion coefficient (‘free strain’) of the actuator
under a voltage V:
3p Z d31 V =ta ; ð2:18Þ
where d31 is the piezo-electric constant (Srinivasan & Michael McFarland 2001).
With reference to figure 12, following an analysis broadly similar to that in
Srinivasan & Michael McFarland (2001), it can be shown that the piezo will
cause the initially straight beam to bend to a radius R (measured from the beam
mid-surface), with the ratio r of non-dimensional curvature to actuator free
S
ta piezo-actuator
h E1
beam
h E2
beam mid-surface
T Z ta =h; ð2:20cÞ
j Z bb h=ðba ta Þ; ð2:20dÞ
and bb, bs are the beam and actuator widths normal to plane of paper. The
formula applies for a composite beam with two layers of equal thickness h. This
was done to simplify the analysis. Moreover, it was thought that the use of equal
layer thickness might maximise the stretch-bend effect induced by the piezo, as
in a cross-ply laminate with two equal thickness layers under tension. figure 13
shows the variation of jrj with T/2 for baZb b and various C for PZT material
with EaZ59.5 GPa and d31Z212!10K12 V mK1 (Brennan et al. 1994). From
figure 13 it appears that there is some advantage in using a composite beam with
the piezo-actuator. The curves in figure 13 show that there is normally optimum
value of T, Topt, which minimizes the electric field requirement for a given
curvature in a beam of given geometry. Hence, there is the possibility of choosing
the materials of a composite beam to reduce this value of Topt. Substituting
rZr(Topt), taZTopth and 1/RZa/S in equations (2.18) and (2.19) and
eliminating 3p gives the minimum voltage required to generate a subtended
angle of a by piezo-actuation in a beam of given thickness 2h and length S :
Topt h 2 a
Vmin Z : ð2:21Þ
rðTopt Þ d31 S
Figure 14 shows Vmin as a function of a for SZ0.05 m and 2hZ0.5!10K3m and
various material properties. It is evident that high voltages are necessary to
actuate significant changes to the subtended angle a over a short length S,
Adaptive tuned vibration absorber with variable shape stiffness element 3969
0.7
0.6
0.5
h/R 0.4
ep 0.3
E 1= E 2 = 10 GPa E 1= E 2 =130 GPa
0.2
E 1 = 10 GPa, E 2 = 5 GPa E 1 =130 GPa, E 2 = 13 GPa
0.1 E 1 = 5 GPa, E 2 = 10 GPa E 1 =13 GPa, E 2 =130 GPa
0
0.2 0.4 0.6 0.8 1.0 0 0.2 0.4 0.6 0.8 1.0
ta /(2h) ta /(2h)
Figure 13. Variation of curvature to free strain ratio with actuator to beam thickness ratio for baZbb
and EaZ59.5 GPa, d31ZK212!10K12 m VK1.
4000
3500
3000
2500
voltage (V)
2000
1500
E 1= E 2 =10 GPa
1000
E 1 =10 GPa, E 2 = 5 GPa
500
E 1 =5 GPa, E 2 = 10 GPa
0 5 10 15 20 25 30 35 40 45 50
a (deg.)
Figure 14. Minimum voltage versus subtended angle a for b aZb b and 2hZ0.5!10K3 m, SZ0.05 m,
EaZ59.5 GPa, d31ZK212!10K12 m VK1.
Consider the linkage in figure 15a where the rigid links AB, BC are pivoted at
A, B and C and the cantilevers are of flexural stiffness k. This configuration is
based on the simplified model of the flight motor of an insect used in the analysis
of Brennan et al. (2003). The nonlinear expression relating the applied static
force F to y (figure 15b) was derived in Brennan et al. (2003):
( )
F b
y 2 K1=2 y
Z 1K K1 : ð3:22Þ
2kL L L L
(a) (b) F
B
y0 L L
y
A q0 C q
rigid links
cantilevers
stiffness k
2b
undeformed deformed
From this expression, the non-dimensional linear stiffness for small deflections
DyZyKy0 about the stable equilibrium position of figure 15a can be derived:
F=ð2kLÞ 1
K~ z Z K1: ð3:23Þ
Dy=L ðb=LÞ2
Hence, variable stiffness can be achieved by varying the distance between the
cantilevers. Figure 16 shows a schematic diagram of a self-tuning vibration
absorber that uses this concept, its tuned frequency ft being proportional to the
square root of the stiffness given by equation (3.23). Note that in the device
shown in figure 16, the stiffness element is inverted relative to that of the insect
model in Brennan et al. (2003) so that the actuator forms part of the effective
mass of the device, thereby enhancing vibration attenuation (Brennan 1997).
The force required from the actuator to control the stiffness is minimal but the
actuator needs to have a large stroke. A further requirement for the actuator is
that its longitudinal stiffness should be sufficiently high as to prevent variation in
the distance between the cantilevers as a result of vibration-induced deformation.
Hence, a suitable actuating mechanism for the device would be a screw driven by
a servomotor, like that used in Kidner & Brennan (2002). Alternatively, it might
be possible to use the gel actuator developed in associated research (Santulli
et al. 2002) forming part of the same multi-disciplinary project as the research
presented in this paper. This would be in keeping with the musculo-skeletal
analogy behind the device in figure 16, since the gel actuator can be considered as
the engineering analogue of a muscle.
The effectiveness of the device is limited to frequencies for which inertia effects
of the components forming the linkage are negligible. For low values of q0 the
device is prone to ‘snap through’ to its lower stable equilibrium position at high
amplitudes of vibration (Brennan et al. 2003). Moreover, for a given level of
vibration, nonlinear effects are expected to be more prominent at low values of
q0, as shown in figure 17, which compares the nonlinear expression in equation
(3.22) with the linear (tangential) relationship of equation (3.23) for three
different values of q0. It is also seen from figure 17 that, for deflections from the
stable equilibrium position, the spring hardens in tension (FO0) and softens in
compression (F!0).
Adaptive tuned vibration absorber with variable shape stiffness element 3971
adjust distance
mechanism and
absorber mass
cantilevers
b b
q0
rigid link L rigid link
L
structure
In the previous sections it was shown that, while considerable variation in tuned
frequency can be achieved by actuating a shape change, this will be limited by
the maximum force that can be provided by the actuator. A feasible design for an
ATVA is one in which the device offers low resistance to the required shape
change actuation. Of course, this should not restrict the device to low values of
ATVA effective stiffness (i.e. the stiffness of the equivalent model in figure 1).
Hence, the best designs are those in which the actuator force is uncoupled as
much as possible from the effective stiffness of the TVA. For example, the devices
in figures 4b, 10a and 16 are effective since, although their effective stiffness can
be considerably high, it is controlled with relatively little effort. For the device in
figure 4b, this is achieved by merely altering the natural (undeformed) shape,
which is equivalent to the piezo-actuator acting against the relatively low
flexural stiffness of a free–free beam. In figure 10a the effective stiffness of the
device is controlled by changing the curvature of the beams. While the controlled
stiffness can be quite high, the curvature of each beam is relatively easy to adjust
by flexure since it is pivoted at the ends and one end is free to move. In figure 16
the effective stiffness of the device is controlled by adjusting the distance between
the cantilevers and the (hinged) linkage stiffness element offers no resistance to
this adjustment.
5. Experimental demonstrators
Demonstrators were built for two of the three effective designs identified in the
previous section. This section presents some experimental data obtained from
these devices. The aims are twofold: to prove that they perform as predicted
under different levels of excitation and to illustrate the variability in tuned
frequency obtainable. For each setting of the ATVA, the tuned frequency ft
(in Hz) and the damping loss factor h were measured by applying random
excitation at the base of the device and measuring the transmissibility frequency
response relating the acceleration at the base, a1(t), to that of the absorber mass
b / L = 0.7
0.15
b / L = 0.8
0.10 b / L = 0.6
0.05
F
0
2kL
– 0.05
-0.10
linearized, equation (3.23)
– 0.15
– 0.5 0 0.5
y/L
Figure 17. Nonlinear relationship between F and y for various b/L values.
a1(t). The measurements were taken for two levels of excitation, the higher one
corresponding to peak values of a2(t) in excess of 10 m sK2. Assuming a linear
relationship between these two signals and a single degree of freedom (SDOF)
model for the effective part of the demonstrator, as in figure 1b:
A2 1 C jh
TZ Z ; ð4:24Þ
A1 1Kðf =ft Þ2 C jh
where h is the damping loss factor of the stiffness element, f is the frequency and
A1 and A2 are the complex
amplitudes
of a1(t) and a2(t) for harmonic vibration,
i.e. when ar ðtÞZ Re Ar e j2pft , rZ1, 2. jT j is maximum at fZft and ft and h
were accurately located by considering that, at this point, Re{T}Z1 and
Im{T}ZK1/h.
Figure 18 shows a picture of a demonstrator for the ATVA of figure 10a. Each
curved beam is a piezo-actuator (THUNDER TH 7-R; Face International 1998)
consisting of a piezo–ceramic layer 0.25 mm thick sandwiched between a
0.15 mm thick aluminium layer and a 0.2 mm thick steel plate. The beams are
supported on miniature ball bearings at their ends, allowing freedom of rotation.
The absorber mass is guided in the direction of vibration by a linear bearing
running on the central column that is fixed to the base. The beam curvature was
adjusted by varying the piezo-actuator voltage. The piezo-actuators can take
between K300 and C600 V DC (Face International 1998), however, the
experiments results were restricted to a voltage range of K280 to C480 V DC.
Since there is negligible leakage through the high resistance ceramic, the current
required to maintain a given curvature setting is miniscule.
At low vibration amplitudes, the lowest measured value of ft (at K280 V) was
36 Hz and the highest 56 Hz (at C480 V), with the value of ft at 0 V being 42 Hz.
The corresponding frequencies at seven times the vibration amplitude were only
slightly changed (36, 55 and 41 Hz, respectively). The experimental work on this
demonstrator is described in greater detail in a separate paper (Bonello et al.
2005). Figure 19 shows the percentage variation in ft with applied voltage.
Adaptive tuned vibration absorber with variable shape stiffness element 3973
13 cm
Figure 18. Prototype ATVA with variable curvature beams as stiffness element.
50
percentage increase in tuned frequency
20
10
– 10
– 20
– 30
– 300 – 200 – 100 0 100 200 300 400 500
voltage (V, DC)
(a) (b)
10 2
measured
measured
from equation (4.24)
from equation (4.24)
10 1 with measured
with measured
ft and h
ft and h
magnitude
10 0
10 –1
10 – 2
1.0
measured measured
0.5 from equation (4.24) from equation (4.24)
with measured with measured
cos f
0 ft and h ft and h
–0.5
–1.0
0 50 100 150 200 250 300 350 400 0 50 100 150 200 250 300 350 400
frequency (Hz) frequency (Hz)
low amplitude high amplitude
Figure 21. Transmissibility plots at low amplitude (Db=LZ b=LKbnom =L; bnom =LZ 0:8).
magnitude of the transmissibility and the cosine of its phase f for Db/LZ0.1 at
low amplitude. It is seen that linear SDOF theory holds and cos f z0 when fZft.
Figure 21b show that, at 10 times the amplitude, the linear theory still holds in
the region of the tuned frequency and this frequency changes only slightly. The
strong harmonics that appear, indicating non-linearity, will have no effect on the
operation of the ATVA.
Adaptive tuned vibration absorber with variable shape stiffness element 3975
50
predicted
Figure 22. Tuning characteristic (Db=LZ b=LKbnom =L; bnom =LZ 0:8).
Figure 22 compares the predicted and measured tuning characteristics for the
nominal configuration bnom/LZ0.8, where the correlation between simulation and
measurement is seen to remain reasonably good at high amplitude. The maximum
variation in ft is 70% (K40 to C30%) for a total actuator movement of 1.6 cm.
6. Conclusions
This paper has explored the concept of shape control for achieving variable
stiffness in an ATVA, presenting novel ideas inspired by biological paradigms
where a change in curvature or geometry is used to control stiffness. Several
approaches were considered: using a shallow arch in transverse vibration, curved
beams vibrating longitudinally, changing anticlastic curvature, and changing the
geometry of a linkage system. A study on the use of composites and piezo-
actuators to achieve the required shape change was also performed. It was
concluded that considerable variation in the tuned frequency can be achieved by
actuating a shape change, provided that this is within the limits of the actuator.
A feasible design for such an ATVA is one in which the device offers low resistance
to the required shape change actuation while not being restricted to low values of
the effective stiffness of the vibration absorber. Hence, the best designs are those in
which the actuator force is uncoupled as much as possible from the effective
stiffness of the absorber. Three such designs were identified: (i) a pinned–pinned
arch beam with fixed profile of slight curvature and variable preload through an
adjustable natural curvature; (ii) a vibration absorber with a stiffness element
formed from curved beams of adjustable curvature in parallel vibrating
longitudinally; (iii) a vibration absorber with a variable geometry linkage as the
stiffness element. The experimental results from demonstrators based on the latter
two designs showed good correlation with the theory.
The authors would like to acknowledge the Engineering and Science Research Funding (EPSRC) of
the UK for funding this work.
References
Brennan, M. J. 1997 Vibration control using a tunable vibration neutraliser. J. Mech. Eng. Sci.
211, 91–108.
Brennan, M. J. 2000 Actuators for active control—tunable resonant devices. J. Appl. Mech. Eng. 5,
63–74.
Brennan, M. J., Day, M. J., Elliott, S. J. & Pinnington, R. J. 1994 Piezoelectric actuators and
sensors. In Proc. IUTAM Symp. on the Active Control of Vibration, pp. 263–274. Eindhoven:
International Union of Theoretical and Applied Mechanics.
Brennan, M. J., Elliott, S. J., Bonello, P. & Vincent, J. F. V. 2003 The ‘click’ mechanism in
dipteran flight: if it exists, then what effect does it have? J. Theor. Biol. 224, 205–213. (doi:10.
1016/S0022-5193(03)00158-9.)
Bonello, P., Brennan, M. J. & Elliott, S. J. 2004 The design of an adaptive tuned vibration
absorber with a variable shape stiffness element. ISVR Memorandum No. 941, ISVR, University
of Southampton, UK.
Bonello, P., Brennan, M. J. & Elliott, S. J. 2005 Vibration control using an adaptive tuned vibration
absorber with a variable curvature stiffness element. Smart Mater. Struct. 14, 1055–1065.
Carneal, J. P., Charette, F. & Fuller, C. R. 2004 Minimization of sound radiation from a plate
using adaptive tuned vibration absorbers. J. Sound Vib. 270, 781–792. (doi:10.1016/S0022-
460X(03)00257-8.)
Face International 1998 THUNDER TH-7R data sheet. Face International Corporation, Norfolk,
VA, USA.
Franchek, M. A., Ryan, M. W. & Bernhard, R. J. 1995 Adaptive-passive vibration control.
J. Sound Vib. 189, 565–585. (doi:10.1006/jsvi.1996.0037.)
Hong, D. P., Ryu, Y. S. 1985 Automatically controlled vibration absorber. US Patent No. 4935651.
Jones, R. M. 1975 Mechanics of composite materials. New York: Hemisphere.
Kidner, M. R. F. & Brennan, M. J. 1999 Improving the performance of a vibration neutraliser by
actively removing damping. J. Sound Vib. 221, 587–606. (doi:10.1006/jsvi.1998.2027.)
Kidner, M. R. F. & Brennan, M. J. 2002 Variable stiffness of a beam-like neutraliser under fuzzy
logic control. Trans. ASME J. Vib. Acoust. 124, 90–99. (doi:10.1115/1.1423634.)
Longbottom, C. J., Day, M. J., Rider, E. 1990 A self tuning vibration absorber. UK Patent No. GB
218957B.
Ormondroyd, J. & den Hartog, J. P. 1928 Theory of the dynamic absorber. Trans. ASME 50, 9–22.
Reissner, E. 1954 Note on the problem of vibrations of slightly curved bars. Trans. ASME J. Appl.
Mech. June, pp. 195–196.
Rustighi, E., Brennan, M. J. & Mace, B. R. 2005 A shape memory alloy tuned vibration absorber:
design and implementation. Smart Mater. Struct. 14, 19–28. (doi:10.1088/0964-1726/14/1/002.)
Ryder, G. H. 1986 Strength of materials 3rd edn. Basingstoke, UK: Macmillan.
Santulli, C., Patel, S. I., Jeronimidis, G., Davis, F. J. & Mitchell, G. R. 2002 Development of smart
actuators using N-isopropylacrylamide gels. In First World Congress on Biomimetics &
Artificial Muscles, pp. 1–9.
Srinivasan, A. V. & Michael McFarland, D. 2001 Smart structures: analysis and design.
Cambridge: Cambridge University Press.
Vincent, J. F. V. & Jeronimidis, G. 1991 The mechanical design of fossil plants. In Biomechanics
and evolution (ed. J. M. V. Rayner & R. J. Wootton), pp. 21–36. Cambridge: Cambridge
University Press.
von Flotow, A. H., Beard, A. H. & Bailey, D. 1994 Adaptive tuned vibration absorbers: tuning
laws, tracking agility, sizing and physical implementation. In Proc. Noise-Conf. 94, Florida,
pp. 81–101.
Walsh, P. L. & Lamancusa, J. S. 1992 A variable stiffness vibration absorber for the minimization
of transient vibrations. J. Sound Vib. 158, 195–211. (doi:10.1016/0022-460X(92)90045-Y.)