Académique Documents
Professionnel Documents
Culture Documents
www.elsevier.com/locate/engstruct
a,*
Department of Civil Engineering, National University of Singapore, 10 Kent Ridge Crescent, Singapore 119260
b
79 Anson Road, Unit 20-01, Singapore 079906
Received 12 May 2000; received in revised form 25 January 2001; accepted 31 January 2001
Abstract
An active control system involving a tuned liquid column damper (TLCD) is developed for the vibration control of a singledegree-of-freedom tower subjected to wind excitation. This new control system will be referred to as the active tuned liquid column
damper (ATLCD) system. It is found that the wind-induced response of the tower can be reduced substantially by the ATLCD
system. The most efficient type of sensor is found to be the acceleration sensor when compared with displacement- and velocitytype sensors. To assess the effectiveness of such a control system, a comparison study is made between the ATLCD and the active
passive tuned mass damper (APTMD). In seeking the optimal control parameters for both systems, the norm of the root-meansquare of the system output under actual wind excitation, Hrms, is adopted as the objective function instead of the commonly used
H2 (i.e., the norm of the system output under a unit white noise). The results show that the proposed ATLCD system is a good
alternative to the APTMD system. Considering the potential advantages of the liquid column damper, the ATLCD system may be
a better choice for vibration control of wind-induced towers. 2001 Elsevier Science Ltd. All rights reserved.
Keywords: Active tuned liquid column damper; Wind-excited vibration; Towers; Optimal control technique
1. Introduction
Modern tall buildings and towers are relatively light
and slender, and possess low natural damping characteristics. Thus, they are prone to excessive wind-induced
oscillations in severe wind environments. The oscillations may cause failure from the viewpoint of serviceability. It is therefore important to design an effective and economical means of suppressing structural
vibrations induced by wind loading.
To date, many auxiliary damping control systems have
been proposed to suppress excessive structural
vibrations. The most commonly used control systems are
the passive control system and the active control system.
In the passive control system, tuned mass dampers
(TMDs) have been found to be effective in reducing the
wind-induced response of tall buildings and towers (see,
e.g., [1,2]). A newly proposed passive device is the tuned
liquid damper (TLD), which relies on the motion of shal-
* Corresponding author.
0141-0296/01/$ - see front matter 2001 Elsevier Science Ltd. All rights reserved.
PII: S 0 1 4 1 - 0 2 9 6 ( 0 1 ) 0 0 0 1 5 - 3
Nomenclature
A
parameter affecting energy distribution on lower frequency range
frontal area of tower
A0
APTMD activepassive tuned mass damper
ATLCD active tuned liquid column damper
ATMD active tuned mass damper
B
width of tube
breadth of tower
B0
drag coefficient, which is taken to be equal to 1.2
C0
CD, CL damping of platform and liquid in the ATLCD system
CP, CA damping of passive TMD and active TMD, respectively
damping of tower
CS
D
depth of tower
f(t)
along-wind force
frequency of vortex shedding
fS
H
transfer function matrix
norm of root-mean-square of the system output under unit white noise
H2
Hrms norm of root-mean-square of the system output under actual wind load
turbulence intensity
Iu
stiffness of the platform in the ATLCD system
KD
KS, KP, KA stiffness of tower, passive TMD and active TMD, respectively
proportional constant between the control force and the movement of the hydraulic piston
Kt
ground surface drag coefficient, which is taken to be equal to 0.03
k0
wavelength, which is taken to be equal to 1200 m
lx
MD, ML mass of platform and liquid in the ATLCD system
MS, MP, MA mass of tower, passive TMD and active TMD, respectively
frequency of tower
nS
feedback gain of the transducer in the servomechanism
1/R0
collective loop gain of the electrohydraulic servomechanism
R1
S
Strouhal number
proportional constant between the sensed structure response and the output voltage from the sensor
S0
wind power spectral density matrix
SF
power spectral density of wind force
SFF
SxS, SxD, Sy power spectral density of tower, platform and liquid of the ATLCD system, respectively
SxP, SxA power spectral density of passive TMD and active TMD of the APTMD system, respectively
TLCD tuned liquid column damper
TMD tuned mass damper
displacement response of tower in frequency domain
XS
XP, XA relative displacement response of passive TMD and active TMD in frequency domain, respectively
XD, XL relative displacement response of platform and displacement of liquid in frequency domain,
respectively
displacement response of tower in time domain
xS
xP, xA relative displacement response of passive TMD and active TMD in time domain, respectively
xD, xL relative displacement response of platform and displacement of liquid in time domain, respectively
mean wind speed at the reference height, 10 m
U10
mean wind speed at height z
Uz
U(w) control force in frequency domain
u(t)
control force in time domain
z
height of tower
a
widthlength ratio in the ATLCD system
b
normalized frequency ratio
d
head loss
e
normalized loop gain
1055
1056
zS, zP, zA damping factor of tower, passive TMD and active TMD, respectively
zD, zL damping factor of platform and liquid, respectively
mP, mA mass ratio of passive TMD and active TMD over tower, respectively
mD, mL mass ratio of platform and liquid over tower, respectively
density of air, which is taken to be equal to 1.2 kg/m3
r0
sF
standard deviation of along-wind force
standard deviation of lift coefficient
sL
sx S
standard deviation of acceleration of tower in the ATLCD and APTMD system, respectively
sxP, sxA standard deviation of displacement of the passive TMD and active TMD, respectively
standard deviation of displacement response of platform in the ATLCD system
sxD
standard deviation response of control force
su
standard deviation response of structural acceleration with damper control system normalized by that
s x rS
without any control system
standard deviation response of dampers displacements with control system normalized by the
s xrD
maximum permissible stroke
standard deviation response of liquids displacement with control system normalized by the maximum
s yr
permissible motion of liquid
standard deviation response of control force normalized by the wind loading
s ur
t
normalized feedback gain
c
damper mass ratio in the APTMD system
P, A frequency ratio of passive TMD and active TMD over tower, respectively
D, L frequency ratio of platform and liquid over tower, respectively
w
circular frequency of along-wind excitation
wS, wP, wA circular frequency of tower, passive TMD and active TMD, respectively
wD, wL circular frequency of platform and liquid, respectively
not present in a TLCD. Due to the simple physical concepts on which the restoring force is provided in a
TLCD, no activation mechanism is required. Therefore,
maintenance cost is minimized. A TLCD is a flexible
system in meeting architectural demands since it can be
fitted easily in a variety of ways. Due to simplicity of
installation, it may be used in existing structures that
often have severe space constraints, even for temporary
use. Furthermore, a TLCD provides an excellent reservoir of water high up in the building for fire suppression.
Therefore, to make use of several potential advantages
of a TLCD, in this paper, an active tuned liquid column
damper system (ATLCD) is proposed. In the ATLCD
device shown in Fig. 1(a), the TLCD is fixed on a movable platform at the top of the tower. The movement of
the platform is controlled by a spring and a dashpot
when it is driven by a control force. A servo-actuator is
used to generate the control force based on the feedback
from the sensor attached to the top of the tower, which is
modelled as a single-degree-of-freedom system. Before
assessing the effectiveness of ATLCD, the most beneficial sensor type is first selected among the available
displacement-, velocity- and acceleration-type sensors.
Because the liquid in the tube oscillates in sympathy
with the rigid-body motion of the tube under a control
force, the proposed system may be viewed as a hybrid
control system similar to the APTMD. Therefore, the
1057
Fig. 1. Damper control system fitted on a single-degree-of-freedom tower: (a) active tuned liquid damper (ATLCD) system, (b) activepassive
tuned mass damper (APTMD) system.
optimum performance of the ATLCD system is compared with that of the APTMD system by using a frequency-based control method. The optimum control
parameters for both ATLCD and APTMD systems are
achieved by minimizing the root-mean-square (rms)
norm of the system output when the wind spectral densities corresponding to along-wind and across-wind excitation have been adopted as the input excitation.
1
rALy rAdyy2rAgyrAB(xSxD),
2
and
MSx SCSxSKSxSCDxDKDxDu(t)f(t)
aML(xSxD)MLyCLyMLw2Ly0,
(1)
and
MD(xSxD)rAByCDxDKDxDu(t)0,
(3)
(4)
(5)
where
(2)
MLrAL,
(6a)
B
a ,
L
(6b)
1058
wL2g/L
(6c)
and
CL
p L s ,
2MLd
(6d)
xS(t)
xD(t)
y(t)
XS(w)
f(t)
XD(w)
Y(w)
eiwt,
(7)
where XS(w), XD(w), Y(w) and F0 are the Fourier transforms of xS(t), xD(t), y(t) and f(t), respectively, and w is
the frequency of the wind excitation.
For the control force u(t), Roorda [11] has established
the following relationship between the control force
U(w) and the controlled tip displacement of the tower
XS(w) in the frequency domain
te
X (w),
U(w)Kt(ib)
e+ib S
r
(8)
where
XD(w) 0 ,
Y(w)
(10)
h111b2g1(2zSg2)(ib),
(11a)
h12mD2D2mDDzD(ib),
(11b)
h130,
(11c)
h21g1b2mDg2(ib),
(11d)
h22(2Db2)mD2mDDzD(ib),
(11e)
h23ab2mL,
(11f)
h31ab2mL,
(11g)
h32ab2mL,
(11h)
h33(2Lb2)mL2mLLzL(ib)
(11i)
(9a)
MD
mD ,
MS
(12a)
ML
,
MS
(12b)
wD
,
wS
(12c)
wL
L ,
wS
(12d)
CS
,
zS
2MSwS
(12e)
CD
,
zD
2MDwD
(12f)
mL
n
b ,
nS
(9b)
R1
e
2pnS
(9c)
and
(2pnS)rS0
.
R0
and
XS(w)
where
F0
servomechanism; R1 the collective loop gain of the electrohydraulic servomechanism; and S0 the proportional
constant between the sensed structural response and the
output voltage from the sensor.
In view of Eqs. (7)(9), the governing equations (1)
and (4), (5) can be expressed as
(9d)
In Eqs. (8) and (9), the parameter Kt represents a proportionality constant between the control force and the
movement of the hydraulic piston; the parameters e and
t are the normalized loop gain and feedback gain,
respectively; b is the normalized excitation frequency; n
the along-wind excitation frequency; nS the frequency of
tower; 1/R0 the feedback gain of the transducer in the
CL
zL
2MLwL
2p s
1 ad
9
(12g)
and
F0
.
F
MSw2S
(12h)
(13)
XD(w) 1 h11c1
HD(ib)
F
h12 h12c2
and
(14)
(15)
where
(21a)
U
r 10,
U
Uz
(21b)
UzU10
z
10
0.3
(21c)
wSlx
,
2p4U10
(21d)
F0p4 F0p42L
F
MSw2SB 2gaMS
(21e)
and
h23h32h22h33
c 1
h12
(16a)
and
c2h21h33h23h31
h11(h23h32h22h33)
.
h12
(16b)
The displacement power spectra of the tower, the platform and the liquid can be expressed, respectively, as
SxSHS(ib)2SFF(b),
(17)
SxDHD(ib)2SFF(b)
(18)
and
SyHL(ib)2SFF(b),
(19)
where SFF(b) is the non-dimensional power spectral density of the along-wind excitation or the across-wind excitation.
For the along-wind force, the two-sided Harris spectrum is adopted. This non-dimensional power spectral
density S lFF is given by [5]
S lFF(b)
2r p4yb
U
f
,
[2+(p3yb)2]5/6
h22
Y
h11h22 h21 c1
HL(ib)
,
F
h12h23 h12h23 h23 c2
1059
4k0F 2
f,
b
(20)
in which
F0r0A0C0U2z .
k0 is the ground surface drag coefficient, lx the wavelength, z the height of the tower, wS the circular frequency of the tower, U10 the mean wind speed at the
reference height of 10 m, r0 the density of air, A0 the
frontal area of the tower, C0 the drag coefficient and Uz
the mean wind speed at height z.
For the across-wind excitation, a new empirical formula for the across-wind spectra proposed by Choi and
Kanda [12] has been adopted for the present study. The
spectral form can be decomposed into two parts, namely
a narrow spectral peak due to the regular vortex shedding
and a wideband spectral distribution caused by vorticity
in the separated shear layer and the shear layertrailing
edge direct interaction. The non-dimensional spectra
density S cFF of the across-wind force can be written as
follows:
1 ln f+0.5d20
2
d0
g1
g2
Displacement
sensor
Kt te2
KS e2+b2
Kt te
KS e2+b2
Velocity sensor
Kt teb
KS e2+b2
Kt te
KS e2+b2
Acceleration sensor
Kt te2b2
KS e2+b2
Kt teb2
KS e2+b2
(22)
where
B1
Type of sensor
F 2y s2L
(1B1)
(f/k)
B1AC
exp
S cFF(b)
A 5/A
b
[1+(f/k) ]
2pd0
AC
Table 1
Expressions of g1 and g2 for different sensors
(21f)
A(5.0/A)
,
(4.0/A)(1.0/A)
Iu
k1.58
d I
0
(23a)
z
3D
1
3
5B0
z
3
d01.58
for D/B03.0,
B0
D
k1.58
6 0.8
(23c)
B I 3
D
(23b)
(23d)
(23e)
for D/B01.0,
(23f)
1060
F0
F y
,
MSw2Sz
(23g)
1
F0 r0A0U2z ,
2
(23h)
fS n
fS
(23i)
(23j)
and
SUz
fS .
B0
In Eqs. (22) and (23), sL represents the standard deviation of the lift coefficient; () the gamma function; fS
the frequency of vortex shedding; S the Strouhal number;
Iu the turbulence intensity; and B0 and D are the breadth
and depth of the tower block, respectively. The parameter A affects the energy distribution in the lower frequency range. Here, A is assumed to be 2.0 and thus the
value of AC becomes 1.26.
Similarly, the power spectrum of the control force is
described as [8]
te
S (b).
Su Kt(ib)
e+ib xS
r
(24)
s x S
sxSp4
Bw2S
1 4
b SxS db
p
1/2
(25)
s xD
sxDp4
B
1/2
S db
p xD
(26)
s y
syp
1
S db
B
p y
1/2
(27)
(30)
(31)
and
MA(x Sx Px A)CAx AKxAu(t)0,
(33a)
h12mP2P2mPPzP(ib),
(33b)
h130,
(33c)
h21g1b2mPg2(ib),
(33d)
h22(2Pb2)mP2mPPzP(ib),
(33e)
h23 m 2mAAzA(ib),
(33f)
h31g1b2mAg2(ib),
(33g)
h32b mA,
(33h)
h33(2Ab2)mA2mAAzA(ib),
(33i)
2
A A
and
1/2
syp4
1 2
s y
b Sy db
BwS
p
(28)
1/2
Su db
su
sF
SFF db
(29)
(32)
and
MP
mP ,
MS
(34a)
MA
mA ,
MS
(34b)
wP
,
wS
(34c)
wA
,
wS
(34d)
1061
zS
CS
,
2MSwS
(34e)
zP
CP
,
2MPwP
(34f)
Sx(b)H(b)SF(b)H(b),
(34g)
CA
zA
2MAwA
(41)
and
F
F0
.
MSw2S
(34h)
(35)
SxPHP(ib)2SFF(b)
(36)
and
(42)
SFF 0 0
SxAHA(ib)2SFF(b).
(37)
SF 0
0
sx Sp4
s x S
Bw
2
S
sxPp
B
1
S db
p xP
(38)
1/2
(39)
and
s xA
sxAp4
B
S db
p xA
0 0
1
Hrms
2p
1/2
trace[H(b)SF(b)H (b)] db
s xP
(43)
1/2
1 4
b SxS db
p
0 0 ,
(40)
(44)
1/2
1
H2
2p
1/2
trace[H(b)H (b)] db
(45)
1062
4. Numerical examples
We first seek the most appropriate type of sensor
among the displacement, velocity and acceleration sensors. Next, the optimum performance of the ATLCD is
compared with that of the APTMD to show the effectiveness of the former system.
4.1. Tower and damper parameters
The towers studied by Xu [8] and Holmes [19] are
considered here. Their characteristics are listed in
Table 2.
The towers considered here are limited to square section with a fixed aspect ratio D:B0=1:1. The behavior of
the towers subjected to wind loads has been studied by
the authors when they are installed: (1) without any
damper, (2) with optimum TMD, (3) with optimum
ATMD, and (4) with optimum APTMD. The results can
be found in the papers by Wang et al. [17] and Yan et
al. [13].
Based on the sensitivity study done by Yan et al. [13],
for active control, the structural responses have been
found to be rather insensitive to the total damperstructure mass ratios, the damperstructure frequency ratios
and the damping ratios of dampers for both along-wind
and across-wind excitation. In view of this, typical
properties of the dampers have been chosen as shown in
Table 3.
Table 2
Properties of towers
Height (m)
Mass (kg)
Fundamental Damping
frequency
factor
(Hz)
Reference
81
186
278
1.0106
1.52107
3.7107
0.49
0.20
0.16
Holmes [19]
Xu [8]
Holmes [19]
1%
1%
1%
1063
Table 3
Properties of damper systems
Damper system
Remarks
ATLCD
APTMD
The mass of the platform is ignored, i.e., the total mass of ATLCD is
equal to the mass of liquid
c is the damper mass ratio, which is mass ratio of the active mass
damper over the total mass dampers
Fig. 2. Effect of different types of sensor on structural acceleration normalized by corresponding response without any damper: (a) acceleration
sensor, (b) velocity sensor, (c) displacement sensor.
Fig. 3. Effect of different types of sensor on displacement of liquid normalized by corresponding response without any damper: (a) acceleration
sensor, (b) velocity sensor, (c) displacement sensor.
and for other cases of a=0.6, 0.7 and 0.8, but the results
have not been presented as they are similar in nature to
those shown in Figs. 25.
4.3. Comparison of optimum performance between
ATLCD and APTMD
It is to be noted that the liquid in the tube has the
freedom to oscillate in respect to the motion of the tower,
besides undergoing a rigid-body motion with the platform, under a control force. In this regard the ATLCD
system is a hybrid system, like the APTMD system,
1064
Fig. 4. Effect of different types of sensor on displacement of platform normalized by corresponding response without any damper: (a) acceleration
sensor, (b) velocity sensor, (c) displacement sensor.
Fig. 5. Effect of different types of sensor on control force normalized by total wind force: (a) acceleration sensor, (b) velocity sensor, (c)
displacement sensor.
which has been shown to be better than the ATMD system [9,13]. Thus, to investigate the performance of the
ATLCD further, the ATLCD system will be compared
with the APTMD system. It is noted that the most beneficial sensor for the APTMD system was found to be
the acceleration-type sensor. Therefore, the accelerationtype sensor is used in the following study.
To investigate the effects of the control parameters e
and t on the performance of the ATLCD system, the
variations in structural acceleration, structural displacement, displacement of the platform, displacement of the
liquid, control force and Hrms with respect to the control parameters are plotted in Fig. 6(a)(f). Fig. 6(a)(f)
are associated with along-wind excitation. It is seen that
the structural acceleration and displacement responses
can be reduced to an acceptable range with the ATLCD.
In view of the results in Fig. 6(a)(f), it is also found that
the dynamic responses of the system are not sensitive to
the loop gain e but to the feedback gain t. The same
insensitivity of the APTMD system to the loop gain e
is also observed. Therefore, in the following study, the
loop gain is set to a fixed value (say, e=5) while the
1065
Fig. 6. Variations of (a) structural acceleration, (b) structural displacements, (c) platform displacement, (d) liquid displacement, (e) control force
and (f) Hrms with ATLCD against control parameters under along-wind excitation.
1066
Table 4
Optimal feedback gains and corresponding responses for along-wind excitation (e=5)
Height
(m)
Control
system
topt
sx S (cm/s2)
sxS (cm)
s xD or s xA (cm) s y or s xP (cm)
su (kN)
Min. Hrms
81
ATLCD
APTMD
ATLCD
APTMD
ATLCD
APTMD
8.65
8.16
9.10
8.67
9.84
9.35
2.67
2.92
2.48
2.70
2.18
2.45
2.59
2.67
2.38
2.58
2.21
2.35
17.70
19.15
17.74
19.61
10.05
10.93
43.06
43.50
43.07
43.49
42.97
43.85
4.27
4.68
4.24
4.57
2.39
3.27
186
278
27.67
39.51
27.36
36.89
15.22
21.13
Table 5
Optimal feedback gains and corresponding responses for across-wind excitation (e=5)
Height
(m)
Control
system
topt
sx S (cm/s2)
sxS (cm)
s xD or s xA (cm) s y or s xP (cm)
su (kN)
Min. Hrms
81
ATLCD
APTMD
ATLCD
APTMD
ATLCD
APTMD
9.23
8.72
9.40
8.94
10.7
9.78
2.68
2.93
2.51
2.75
2.18
2.44
2.36
2.68
2.24
2.43
2.16
2.25
19.75
20.84
19.63
21.30
10.02
10.90
43.03
42.90
43.18
43.00
42.98
42.81
4.32
4.68
5.14
5.39
2.58
3.35
186
278
23.53
33.14
37.50
49.20
16.91
23.16
5. Conclusions
An active tuned liquid column damper system has
been proposed for the vibration control of towers under
wind excitation. The effectiveness of the ATLCD system
in suppressing wind-induced acceleration response has
been demonstrated for a single-degree-of-freedom tower
in the frequency domain. The results obtained from the
current research have shown that the proposed ATLCD
system can be used to significantly reduce the windinduced acceleration response of towers. The acceleration type of sensor has been found to be most effective
compared with displacement and velocity sensors. Since
the proposed ATLCD is a hybrid control system, its performance is compared with the APTMD, which has previously been shown to be better than the well-known
ATMD. Through the comparison of the optimum performance of the ATLCD and the APTMD, it is found
Table 6
Response ratios under along-wind and across-wind excitations when using optimum feedback gains
Excitation
Height (m)
sATLCD
x S
sATLCD
xS
sATLCD
xD
APTMD
x S
APTMD
xS
APTMD
xA
s
Along-wind
Across-wind
81
186
278
81
186
278
0.91
0.92
0.90
0.91
0.91
0.89
0.97
0.92
0.94
0.88
0.92
0.96
0.92
0.90
0.92
0.95
0.92
0.92
sATLCD
y
sAPTMD
xP
sATLCD
u
sAPTMD
u
0.70
0.74
0.72
0.71
0.76
0.73
0.99
0.99
0.98
1.003
1.004
1.004
References
[1] Kawaguchi A, Terammura A, Omote Y. Time history response
of a tall building with a tuned mass damper. J Wind Eng Ind
Aerodyn 1992;41-44:194960.
[2] Xu YL, Kwok KCS, Samali B. Control of wind-induced tall
building vibration by tuned mass dampers. J Wind Eng Ind Aerodyn 1992;40:132.
[3] Fujino Y, Pacheco BM, Chaiseri P, Sun LM. Parametric studies
on tuned liquid damper (TLD) using circular containers by freeoscillation experiment. J Struct Engrs/Earthquake Eng
1988;5(1):38191.
[4] Sakai F, Takaeda S, Tamaki T. Tuned liquid column damper
new type device for suppression of building vibrations. In: Proceedings of International Conference on Highrise Buildings,
Nanjing, China, 1989:92631.
1067