Академический Документы
Профессиональный Документы
Культура Документы
www.elsevier.com/locate/engstruct
Abstract
Passive tuned-mass dampers (TMDs) are a very efficient solution for the control of vibrations in structures subjected to long-duration, narrowband excitations. In this study, a Bidirectional and Homogeneous Tuned Mass Damper (BH-TMD) is proposed. The pendular mass is supported
by cables and linked to a unidirectional friction damper with its axis perpendicular to the direction of motion. Some advantages of the proposed
BH-TMD are: (1) its bidirectional nature that allows control of vibrations in both principal directions; (2) the capacity to tune the device in each
principal direction independently; (3) its energy dissipation capacity that is proportional to the square of the displacement amplitude, (4) its low
maintenance cost. Numerical results show that, under either unidirectional or bidirectional seismic excitations, the level of response reduction
achieved by the proposed BH-TMD is similar to that obtained from an ideal linear viscous device. Moreover, experimental shaking table tests
performed using a scaled BH-TMD model confirm that the proposed device is homogeneous, and, hence, its equivalent oscillation period and
damping ratio are independent of the motion amplitude.
c 2006 Elsevier Ltd. All rights reserved.
Keywords: Tuned mass damper; Passive control; Structural dynamics; Bi-directional control; Homogeneous device; Frictional damping; Low-cost TMD
implementation
1. Introduction
Passive Tuned Mass Dampers (TMDs) are used in vibration
reduction of flexible structures subjected to long-duration
narrow-band excitations [13]. While a TMD does not
necessarily reduce the peak deformation demand in an
inelastic structure subjected to ground motion, it reduces the
corresponding level of damage [5,6].
In the TMD literature, there are publications that deal
with the bidirectional behavior of a structure. Most of this
research aims to control the lateraltorsional response of the
bare structure by means of multiple unidirectional TMDs [7,
8]. In order to use the total weight of the supplemental mass,
a typical design would consider one or multiple bidirectional
TMDs, with frequencies tuned independently in each principal
1549
Fig. 1. Schematic representation of a BH-TMD: (a) 3D-view of the device in the undeformed position; (b) x z-plane motion; and (c) yz-plane motion.
u
1L + L y cos sin
L y sin
r(, ) = v =
(1)
w
L x 1L + L y cos cos
where 1L = L x L y is the difference in TMD lengths;
is the angle (measured in the x z plane) between QC and the
vertical direction; and is the angle (measured in the ABC
plane) between the height of triangle ABC, QC, and cable
C D. For convenience, displacement components u and v are
set as the independent coordinates, and grouped in a degree-of
T
freedom (DOF) vector q = u v . The relationship between
the dependent coordinate w and q can be found from Eq. (1).
An example of contour lines of w(u, v) can be seen in Fig. 2(b),
along with the direction and magnitude of the gradient of
w(u, v), which is related to the restoring force acting on m d
due to the gravitational field.
The engineering axial deformation s along the direction of
the friction damper is given by:
q
s(u, v) = ld (u, v) lo = u 2 + v 2 + (w + lo )2 lo
(2)
where ld (u, v) and lo are the deformed and undeformed lengths
of the device, respectively. The corresponding axial force in the
friction damper is approximated by a rigid-plastic model:
!
s T
q
(3)
f = po sign (s ) = po sign
q
1550
(6)
where
fd (q, q)
=
Fig. 2. (a) Deformed state of the Y-shape cable system; (b) contour lines of
vertical displacement w = w(u, v) indicating direction and magnitude of the
gradient of w for an orthotropic BH-TMD (L x = 100 cm, L y = 60 cm).
T
s
where po is the slip force; s = q
q is the rate of the damper
axial deformation; and sign represents the signum function.
Finally, a restrainer was incorporated into the friction damper
in order to limit the lateral displacement of the pendular mass.
The magnitude of the corresponding force is:
0
if s < slim
fr =
(4)
kr (s slim ) + cr s if s > slim
(q)
d (q, q)
= J m d a Q
q + Cq q + f d (q, q)
T
(5)
r
J = q
second-order term that couples q and q;
is the Jacobian
and
matrix of the kinematic transformation (i.e., r = Jq);
T
a = aTh az is the vector of support accelerations, where ah =
T
ax a y and az are the horizontal and vertical components,
respectively.
Eq. (5) takes into account the actual kinematics of the
BH-TMD and is highly nonlinear. A first-order approximation
az
1+
g
K p + K f (s ) q
m g
d
0
0
p = k px
K
= Lx
md g
0 k py
0
Ly
"
#
2
px
0
= md
0
2py
0
f (s ) = sign(s ) k f x
K
0 k f y
1
1
+
0
lo
= sign(s ) po L x
1
1
0
+
Ly
lo
(7a)
(7b)
(7c)
(8)
which indicates that for small deformations the proposed BHTMD behaves as a first-order nonlinear but homogeneous
system. Because of their simplicity, Eqs. (6) and (7) will be
used later for the design of the proposed BH-TMD.
2.3. Experimental validation
In order to experimentally validate the first-order approximation of the constitutive relationship of the proposed BHTMD (Eqs. (7a)(7c)), a scaled model of an isotropic BHTMD was constructed and tested on a shaking table at the
Structural Engineering Laboratory of the Pontificia Universidad Catolica de Chile. As shown in Fig. 3, the model has a
cylindrical pendular mass of weight Wd = 100 N supported
by
three vertical cables of length L = 10 cm (d x = dy =
g/L = 2.6 rad/s). The initial, undeformed length of the
Fig. 3. Experimental setup of a shaking table test of a 1:4 scaled isotropic BHTMD model.
1551
(9)
Fig. 4. Experimental results for a scaled BH-TMD model: a displacement of the pendular mass with respect to the base (top); and measured base acceleration
(bottom).
1552
(10)
(11a)
(11b)
=
Rs u g (t)
fd (q, q)
Qd (q, q)
JT m L
d
(12)
d
1 /2
(14)
op =
=
s
1+
s
(1 /4)
op =
(15)
4 (1 + ) (1 /2)
where is the ratio between the mass of the TMD and that
of the primary structure; and d and s are the fundamental
nominal frequencies of the TMD and that of the structure in the
direction considered, respectively. Based on these equations (or
on any of the equivalent equations proposed in the literature
[1113]), valid for linear behavior, simple design equations for
the BH-TMD can be easily derived. Because of its orthotropic
properties, the BH-TMD can be tuned in each principal
direction independently, and the pendular lengths are given by
(Eq. (7b)):
g
g
= 2 2
2px
op sx
g
g
Ly = 2 = 2 2 .
py
op sy
(16)
Lx =
(17)
1 Ed
4 E s
Ks
0
y
az
f (s )
q
0
1+
Kp + K
g
"
#
Ms + LTh m d Lh Rs
=
u g (t)
(13)
m d L h Rs
(18)
1553
(r + r )controlled
(r + r )uncontrolled
(20)
Fig. 8. Schematic plan view of the Bidirectional Linear Viscous Tuned Mass
Damper (BLV-TMD) used as benchmark device.
1554
Table 1
Dynamical parameters of the models considered in this study
Parameters
M1 model
M2 model
1
2
3
1.37
8.45
24
1.05
3.66
7.21
1
2
3
1.37
8.45
24
1.47
5.12
10.1
Damping ratio
0.02
0.05
d x (rad/s)
dy (rad/s)
d x
dy
1.32
1.32
0.086
0.086
0.03
1.01
1.42
0.086
0.086
0.03
L x (cm)
L y (cm)
po /Wd
lo /L x
o
57
57
0.12
1.0
0.02
0.03
97
49
0.10
1.0
0.02
0.03
Primary structure
TMD
Fig. 9. 5%-damped pseudo-acceleration response spectra of the earthquake records considered in this study.
and
max (kr (t)k)controlled
(21)
max (kr (t)k)uncontrolled
PN
r (t), N being the number of time
where r (t) = N1 t=1
PN
discretization points of r (t); and r = ( (N 11) t=1
(r (t)
r (t))2 )1/2 .
Since the efficiency of TMD devices is sensitive to the intensity, duration and frequency content of the excitation [15],
ground acceleration histories from different events and soil
were selected: (1) El Centro (Imperial Valley, USA,1930); (2)
Newhall (Northridge, USA, 1994); (3) Melipilla (Chile, 1985);
and (4) SCT (Michoacan, Mexico, 1985). The corresponding
5%-damped pseudo-acceleration response spectra are shown in
Fig. 9. In order to get more insight into the dynamics of TMD2 = 1
1555
Fig. 10. Response to harmonic excitations (PGA = 0.01g) of model M1, with and without TMDs: (a) Frequency Response Functions (FRFs) for the displacement
at the top of the chimney; (b) response histories under resonance condition of normalized interaction force (left) and normalized displacement at the top of the
chimney (right).
1556
Fig. 11. Unidirectional response of model M2, with and without BH-TMD subjected to the E-W component of the SCT record scaled down to (a) 25% and (b)
50%: normalized hysteresis loops of the BH-TMD (top), and response history of normalized base shear Vx (t)/Ws (bottom).
1557
(a) Melipilla.
(b) El Centro.
Fig. 12. Bidirectional response of chimney (model M1) with and without TMDs, to two earthquake records: (a) Melipilla; and (b) El Centro. Displacement paths
are shown at left and response history of total displacement d(t) at the top of the chimney at right.
Table 2
Maximum uncontrolled response (in % of total height H ), and reduction factors 1 and 2 (in brackets) for chimney (model M1)
-X Y
Record
Melipilla
-X
-X Y
SCT
-X
-X Y
Mean
-X
-X Y
0.98
1.05
0.44
0.45
0.48
0.49
0.53
0.57
0.32
(0.12)
0.50
(0.32)
0.50
(0.30)
0.63
(0.58)
0.57
(0.44)
0.21
(0.20)
0.21
(0.23)
0.28
(0.16)
0.29
(0.16)
0.29
(0.13)
0.53
(0.28)
0.41
(0.27)
0.60
(0.53)
0.50
(0.45)
0.21
(0.26)
0.26
(0.24)
0.28
(0.12)
0.23
(0.15)
-X Y
Newhall
-X
0.22
0.30
BLV-TMD
0.22
(0.04)
BH-TMD
0.22
(0.07)
Reduction factors
El Centro
-X
Acknowledgements
1558
(b) El Centro.
Fig. 13. Bidirectional response of 25-story building (model M2) with and without TMDs, to two earthquake records: (a) Newhall scaled to 50%; and (b) El Centro.
Displacement paths are shown at left and response history of total roof displacement d(t) at right.
Table 3
Maximum uncontrolled response (in % of total height H ), and reduction factors 1 and 2 (in brackets) for 25-story building (model M2)
El Centro
-X
-X Y
Newhall
-X
-X Y
Record
Melipilla
-X
-X Y
SCT
-X
-X Y
Mean
-X
-X Y
0.33
0.33
0.82
0.82
0.19
0.26
1.42
1.43
0.69
0.71
0.42
(0.05)
0.40
(0.04)
0.53
(0.13)
0.51
(0.13)
0.24
(0.00)
0.26
(0.09)
0.45
(0.37)
0.43
(0.37)
0.41
(0.14)
0.40
(0.16)
scaled to 50%
0.41
(0.03)
0.37
(0.05)
0.54
(0.13)
0.50
(0.13)
0.22
(0.02)
0.23
(0.10)
0.44
(0.39)
0.42
(0.38)
0.40
(0.13)
0.38
(0.17)
scaled to 100%
0.40
(0.03)
0.36
(0.03)
0.52
(0.12)
0.35
(0.13)
0.22
(0.03)
0.22
(0.10)
0.17
(0.08)
0.15
(0.06)
0.33
(0.05)
0.27
(0.08)
= 12 m d u 2 + v 2 + w 2 = 12 r T M(r ) r is the
where T (q, q)
kinetic energy of the pendular mass, M(r ) = m d I the local
mass matrix (I is a 3 3 identity matrix); Vg (q) = m d gw is
the gravitational potential energy, g the acceleration of gravity;
Vg
d T
T
+
+ Qi + Qe = 0
dt q
q
q
(A.1)
1559
d T
d (q)
(q) q
=
M q = M(q) q + M
dt q
dt
r T T
T
=
=
q j
q j r
= q T
(A.2)
(A.3)
(A.4)
XX
(q)
m i j
qk
d T
(q) q;
M
dt q
T = 1/2
T
q
q j qk .
(B.4)
(B.5)
1
(q)
M = md
0
0
1
(B.6)
XX
i
(A.5)
(q)
m i j qi q j
(B.7)
j
(q)
X X m i j
T
=
qi q j .
qk
qk
i
j
(B.8)
X (q)
d T
(q)
=
m i j q j + m i j q j
dt qi
j
(B.3)
m i j q j
0
dt qi
j
(q)
d T
dt q
qk .
X X m i(q)
X (q)
d T
j
2
m i j q j +
=
qk + (q) q j
0
dt qi
q
k
j
k
j
where M
(ii) Term
(i) Term
qk
or
s
r
w
az .
= m d ah + m d
q
X m i(q)
j
(q)
m i j =
Vg
q
Vg
w
=
m d g.
q
q
T
J
q M(r ) r
q j
!
J T (r )
M J q = q T H j q.
q j
(q)
(B.1)
T
0.
q
(B.9)
(iii) Term w
q
The ith component of vector w
q can be expressed by a
Taylor series as:
X 2 w
w
w
q j + 2 (q)
+
(B.10)
=
qi
qi 0
q
q
i
j
0
j
w
where q
= 0. Hence, the linear approximation of w
q can be
i 0
expressed by:
1
0
w
L
w q;
w =
H
H
(B.11)
x
1
q
0
Ly
1560
w (i, j) = 2 w is the Hessian matrix of w(q)
where H
qi q j 0
evaluated at q = 0.
s
(iv) Term q
In this case, the procedure followed to linearize terms (i) and
(iii) leads to:
1
1
+
0
s
L
lo
s q;
s =
(B.12)
H
H
x
1
1
q
0
+
Ly
lo
2s
s (i, j) =
where H
qi q j 0 is the Hessian matrix of s(q)
evaluated at q = 0.
Substituting Eqs. (B.6), (B.9), (B.11) and (B.12) into Eq. (5)
gives Eq. (6).
[4]
[5]
[6]
[7]
[8]
[9]
[10]
References
[1] Villaverde R, Koyama LA. Damped resonants appendages to increase
inherent damping in buildings. Earthq Eng Struct Dyn 1993;22:491508.
[2] Bernal J. Influence of ground motion characteristic on the effectiveness of
tuned mass dampers. In: Proc. XI world conf. on earthq. engng. 1996.
[3] Ruiz SE, Esteva L. About the effectiveness of tuned mass dampers
on nonlinear systems subjected to earthquakes. In: Manolis GD,
Beskos DE, Brebbia CA, editors. Earthquake resistant engineering
[11]
[12]
[13]