You are on page 1of 22

ASE 10-3-6 Tse 12/6/07 8:20 am Page 283

Vibration Control of a Wind-Excited Benchmark


Tall Building with Complex Lateral-Torsional
Modes of Vibration

K. T. Tse1,*, K. C. S. Kwok2, P. A. Hitchcock1, B. Samali3 and M. F. Huang2


1CLP Power Wind/Wave Tunnel Facility, The Hong Kong University of Science and Technology, Hong Kong
2Department of Civil Engineering, The Hong Kong University of Science and Technology, Hong Kong
3Faculty of Engineering, The University of Technology, Sydney, Australia

(Received: 28 August 2006; Received revised form: 29 January 2007; Accepted: 27 February 2007)

Abstract: This paper describes a proposed wind-excited benchmark tall building


incorporating three-dimensional lateral-torsional modes of vibration, which is typical
of a significant number of modern tall buildings. A series of wind tunnel pressure tests
were conducted on a 1:400 scale model to determine the translational and torsional
wind forces acting on the benchmark building. A finite element model was also
constructed and mass, damping, and stiffness matrices were subsequently formulated
as an evaluation model for numerical analysis. The evaluation model was further
simplified to a state reduced-order system (ROS) using the state order reduction
method. A numerical vibration control example was conducted to demonstrate the
suppression of the wind-induced three-dimensional lateral-torsional motions using
a bi-directional tuned mass damper (TMD) incorporating two magnetorheological
(MR) dampers, one in each orthogonal direction, to act as a semi-active control
system, referred to as a smart tuned mass damper (STMD). The optimal control forces
generated by the MR dampers were obtained through the linear quadratic regulator
(LQR) to minimize the storey accelerations. The formulation details, methodology and
numerical simulation results are outlined in this paper.

Key words: wind-excited benchmark building, evaluation model, reduced-order system, performance criteria,
Smart Tuned Mass Damper (STMD), Linear Quadratic Regulator (LQR).

1. INTRODUCTION The first generation wind-excited benchmark building


Over the years, there have been significant achievements is a 76-storey, 306 meter tall concrete office tower with
in developing structural control algorithms and devices square cross-section, and it is chamfered at two diagonal
(e.g. Samali et al. 1985; Ankireddi and Yang 1996; Cao corners. Its axes of shear centre and mass centre coincide,
et al. 1997; Yang et al. 1997; Spencer 2002; Kobori et therefore essentially avoiding coupled lateral-torsional
al. 2003). Each algorithm and device has its own merits motion. For further simplicity, all rotational degrees of
and was developed for particular applications. To make freedom were eliminated by static condensation,
direct comparisons between these algorithms and resulting in only the lateral degrees of freedom being
devices, structural control benchmark problems have retained (Yang et al. 2001). Hence, the model is
been developed for earthquake and wind excitations, amenable to research aimed at suppressing sway
respectively (Spencer et al. 1998; Yang et al. 1998). modes of vibration, such as the studies of Samali et al.

*Corresponding author. Email address: timtse@ust.hk; Fax: +852-22430040; Tel: +852-23580170.


Associate Editor: YL Xu

Advances in Structural Engineering Vol. 10 No. 3 2007 283


ASE 10-3-6 Tse 12/6/07 8:20 am Page 284

Vibration Control of a Wind-Excited Benchmark Tall Building with Complex Lateral-Torsional Modes of Vibration

(2004a, 2004b) for liquid column vibration absorbers and 72 m 24 m

active control using a fuzzy controller. The sway motions 8m 236 m

17 Floors @ 4 m
226 m
of the benchmark building were also shown to be Link beam
214 m
mitigated successfully using sliding mode control 202 m
(Kim et al. 2004; Wu and Yang 2004). Recognising the 186 m

limitations of the first generation benchmark building, 8m


170 m
154 m
in which only sway modes of vibration were

17 Floors @ 4 m
138 m
considered, a wind-excited benchmark tall building is 122 m
Steel out
proposed that deliberately incorporates three- rigge truss
106 m
dimensional (3D) lateral-torsional motions, similar to a 90 m
8m
large number of modern tall buildings. 70 m

17 Floors @ 4 m
A series of wind tunnel pressure tests were conducted
at the CLP Power Wind/Wave Tunnel Facility (WWTF) Core
46 m

at The Hong Kong University of Science and Technology h h


22 m
(HKUST) to determine the translational and torsional 12 m
0m
wind loads experienced by the building in a suburban
terrain. A finite element model was also constructed and x y

was used for the determination of an evaluation model,


Figure 1. The proposed wind-excited benchmark building.
including mass, damping and stiffness matrices. To save
computational time, the evaluation model was
subsequently condensed into a model with less degrees of columns on each storey, with 7 columns on each side of
freedom while retaining the key dynamic properties, the wide face and 3 columns on each side of the narrow
referred to as a reduced-order system (ROS), from which face of the building. The two identical reinforced concrete
damping devices may be designed and control algorithms cores are connected by lintel beams and were designed to
evaluated. A bi-directional tuned mass damper (TMD) resist lateral loading such as wind. The two additional
integrated with a magnetorheological (MR) damper in steel out-rigger trusses, connecting the core to the outer
two orthogonal directions, acting as a semi-active columns, are employed as a stabilization system at refuge
system and referred to as a smart tuned mass damper floors located at approximately one-third and two-thirds
(STMD), was employed to illustrate the mitigation of of building height. The out-rigger trusses, which mobilize
wind-induced 3D lateral-torsional motions of the the exterior columns to provide further resistance to
benchmark building. The linear quadratic regulator lateral loads, increase the effective structural depth of the
(LQR) was utilized to establish the target control forces, building in the narrow direction (i.e. along the y-axis as
aiming to minimize the storey accelerations, while the shown in Figure 1). Core setbacks at the two refuge floors
MR dampers were commanded with an inverse dynamic cause a significant shift of mass without significantly
model to generate accurately the control forces. The altering stiffness, resulting in eccentricities between
details of the formulation of the ROS, control algorithm the level-by-level shear centre, mass centre and the
and inverse dynamic model, and the numerical geometrical centre over the building height. The
simulation results are presented in this paper. asymmetrical structural configuration along the x-axis and
the associated eccentricities cause the building to
2. PROPOSED BENCHMARK BUILDING experience 3D modes of vibration.
The proposed benchmark building is a 60-storey, 240 A finite element model (FEM) was constructed using
metre tall reinforced concrete structure with a uniform SAP2000 to model the structural system and mass
rectangular cross-section of 72 m by 24 m throughout its distribution. The peripheral concrete frame members
height, hence a corresponding aspect ratio (H:W:D) of and lintel beams connecting the cores were modelled as
10:3:1, as shown in Figure 1. The total mass of the space frame elements with rectangular cross-sections,
building is almost 130,000 tonnes, resulting in a mass while the outrigger steel members and the concrete core
density of slightly above 300 kg/m3 with a building walls were modelled as truss and shell elements,
volume of more than 410,000 m3. respectively. For ease of construction, the initial sizes of
The reinforced concrete building comprises two concrete members were grouped into four zones as high-
identical reinforced concrete cores, reinforced concrete rise level, mid-rise level, low-rise level and near ground
frames, and two steel out-rigger trusses. The reinforced level, for which the typical dimensions of the outer colu-
concrete perimeter frames are composed of columns at mns were 1500 mm × 1000 mm, 1200 mm × 1000 mm,
regular spacing of 12 m, resulting in 16 perimeter 1000 mm × 1000 mm, and 1000 mm × 800 mm

284 Advances in Structural Engineering Vol. 10 No. 3 2007


ASE 10-3-6 Tse 12/6/07 8:20 am Page 285

K. T. Tse, K. C. S. Kwok, P. A. Hitchcock, B. Samali and M. F. Huang

1st mode 2nd mode 3rd mode 1.8


250
1.5
Gust wind speed
1.2 Turbuence intensity
200

z/h
0.9
AS/NZS 1170.2:2002 Cat. 3
0.6

150 0.3
Height (m)

0.0
0.0 0.2 0.4 0.6 0.8 1.0 1.2
100 a) Normalized wind characteristics

0.0
λL = 250 m
50

log (n.S(n)/var u)
x –1.0
y
z
0
–1 0 1 –1 0 1 –1 0 1 Measured spectrum
–2.0
Figure 2. Mode shapes of the benchmark building. Harris - von Karman spectrum

–3.0
respectively. The range of thickness of the core walls –2 –1 0 1 2
b) Log (n/u)
varied from 200 mm to 750 mm, while the cross-
sectional areas of steel out-rigger members were 0.5 m2. Figure 3. Wind characteristics and longitudinal wind spectrum.
The compressive strength of concrete was assumed to be
60 MPa and the elasticity modulus was 34 GPa.
Dynamic properties such as natural frequencies of 3. EXPERIMENTAL SETUP AND WIND LOADS
vibration and mode shapes were determined by In order to use wind tunnel results to predict wind forces
executing a classical eigenvalue analysis and were acting on a prototype-scale structure, the characteristics
subsequently used to develop a numerical evaluation of the natural wind should be satisfactorily modelled in
model which can be employed to design damping the wind tunnel. The following variables are of
devices and to compare control algorithms. The particular importance: (1) mean longitudinal wind
development and simplification of the numerical velocity at height z, U(z); (2) turbulence intensity at
evaluation model are discussed comprehensively later in height z, Iu(z); (3) power spectral density of the velocity
this paper. The mode shapes corresponding to the first fluctuations, Su(n); and (4) length scale associated with
three modes of vibration, associated with storey mass the modelled building and natural wind, λL.
centre along the height, are displayed in Figure 2. It For this study, a boundary layer wind model
should be noted that the torsional mode shapes are corresponding to an open terrain (Category 3) in the
multiplied by the overall radius of gyration of the AS/NZS 1170.2:2002 (Australian/New Zealand Standard
building to maintain consistent dimensions for the three 2002), was simulated using a combination of solid wooden
(x, y, z) components. fences and roughness elements over a 21 m fetch length at
As intended, the first mode is a 3D lateral-torsional WWTF. The fluctuating wind velocity was measured at
mode with dominance along the y-axis, due to the various heights at the centre of the working section using a
eccentricity of shear centre from mass centre along the hot-wire anemometer. Measured and target gust wind
x-axis. The dominant mode shape was found to have a speed profiles were normalized with respect to the value at
double-curvature with contraflexure noted at mid- building height and are presented in Figure 3(a) together
height, which is a typical deflected shape of an with the measured and target turbulence intensity profiles.
outrigger-core structural system. The second mode is The wind tunnel results are reasonably consistent with the
essentially a translational mode along the x-axis, profiles suggested by the AS/NZS 1170.2:2002, with the
attributable to the symmetry of structural system in that difference generally not exceeding 5%. It can also be seen
direction. The third mode is predominantly torsional from Figure 3(b) that the corresponding spectrum of
with a modest contribution of translation along the longitudinal wind speed at building height is comparable to
y-axis. The corresponding natural frequencies are 0.116, a Harris-von Karman spectrum with a longitudinal
0.213, and 0.229 Hz, respectively. turbulence length scale of approximately 250 m at
prototype scale.

Advances in Structural Engineering Vol. 10 No. 3 2007 285


ASE 10-3-6 Tse 12/6/07 8:20 am Page 286

Vibration Control of a Wind-Excited Benchmark Tall Building with Complex Lateral-Torsional Modes of Vibration

increments from 0° to 90°, where 0° corresponds to wind


normal to the wide face of the building.
External pressures were measured and normalized
with respect to the mean wind speed at building height.
Instantaneous pressure coefficients were determined at
each pressure-tap location and subsequently analyzed
to determine maximum, minimum, mean, and standard
deviation Cp’s. Figure 5 displays the mean pressure
coefficients measured on the surfaces of the
benchmark building model due to flow normal to the
wide face (i.e. 0°) and narrow face (i.e. 90°). Three-
dimensional flow is noted at the building top for the 0°
incident wind angle, resulting in a stagnation point at
about 80% of the building height and the pressure
distribution is symmetrical about the centreline of the
windward face. As expected for a symmetrical
building, the pressure distribution is nearly identical
for the side faces.
Eqn 1 was used subsequently to convert the measured
pressure coefficients into 14 layers of prototype scale
external wind forces,
Figure 4. Pressure-tapped benchmark building model inside
1
Wi (t) = ρU 2 ∑ Cpij (t)∆A ij
wind tunnel.
(1)
A 1:400 scale rigid model of the benchmark building, 2 j

as shown in Figure 4, was constructed and tested at


WWTF to measure building surface pressure, from where ρ is the density of air in kgm–3; U is the prototype
which the alongwind, crosswind, and torsional wind scale mean wind speed at building height in ms–1; Cpij(t)
loads were determined and building responses were is the pressure coefficient time history measured in the
estimated numerically in conjunction with the wind tunnel at pressure-tap j in layer i; and ∆A ij is the
evaluation model. The rigid model was installed with 14 corresponding tributary area in m2. In this study, a 10 year
layers of pressure-taps over its height with 32 pressure- return period serviceability mean wind speed of 36.6 ms–1
taps in each layer. The elevations of the 14 layers are at building height was used for a building located in a
presented in Figure 1. Each pressure-tap was connected coastal cyclone region, in accordance with AS/NZS
to one of 16 ports of an ESP-16HD pressure scanner with 1170.2:2002. For alternative locations, the external
a 750 mm single lumen PVC tube of 1.5 mm internal- wind forces may be adjusted by multiplying with
diameter, without any restrictor. The amplitude and
U 2d / 36.62 , where U d is the desired prototype mean
phase distortion due to the tubing system were
compensated by numerical post-processing. wind speed at building height. Consequently, the 36 s of
In order to satisfy the requirement on the minimum recorded data were converted into equivalent 1 hour
Reynolds number of 5 × 104 for a building model with duration, prototype scale wind forces with appropriate
sharp edges, as recommended by AWES-QAM-1-2001 time scaling, and those external wind forces were used
(Australasian Wind Engineering Society 2001), the tests in the evaluation of building responses, damper design,
were carried out at a mean wind speed of approximately and control algorithms.
13 m/s at the model height, resulting in a velocity scale of In view of the mode shapes computed by the FEM,
approximately 1:4 and a time scale of approximately the development of the numerical evaluation model and
1:100. Surface pressures were measured at a sampling formulation of the equations of motion, the translational
frequency of 400 Hz, which was sufficient to measure and torsional wind forces were determined at the mass
pressure fluctuations with frequencies of up to 2 Hz at centre of the corresponding floor. The distributed mean
prototype scale, which is significantly higher than the wind forces along the building height are presented in
natural frequencies of the first 6 modes of vibration. The Figure 6 for incident wind angles of 0° and 90°. It is
pressure data were recorded for 36 s, which is equivalent noted that, due to the symmetrical cross-section of the
to approximately 1 hour at prototype scale. Measurements building and the simulated wind field, the mean
were taken for five different incident wind angles at 22.5° crosswind suction forces on the side faces are very

286 Advances in Structural Engineering Vol. 10 No. 3 2007


ASE 10-3-6 Tse 12/6/07 8:20 am Page 287

K. T. Tse, K. C. S. Kwok, P. A. Hitchcock, B. Samali and M. F. Huang

0.7 0.6

–0

8.0
.6

– 0 0.2
220 220

6.0

– 0.– 0.2
–0.5
–0.6
0.8

–0


–0.6

.22

6
– 0.5
–0

3
– 0..4
– 0.6

.8
0.06

0.5

–0.7
–0

–0.7

.2

–0
– .58
0.9

–0

– 0.8
200 200

–0
.62

– 0 0.5

–0.4
.62

–0.3
.5 6

– 0.64
180 4 180
– 0.66

–0.2
0.80.7
160 160

0.66

.5

–0.52
Height (m)

Height (m)
–0
140 140
–0.64

–0.4

0.7
0.7
–0.6

6
–0.4
8
0.60.7

120 120
– 0 .58
–0
0.8

.6

0.62

–0
100 100

.18
0.6

– 0.2
0.4 0.5 0.3

80 80

–0.3
–0.5
–0.4
–0.6
0.3 0.50.4
–0.62

– 0.4 – 0.65
.5
–0.

– 0.

–0.7
–0
0.64

– 0.7
0.6
58

60 60

– 0.2
– 0.3
0.5
4
.5

40 40
–0

Side Front Side Back Front Side Back Side


(a) (b)

Figure 5. Mean pressure coefficients on the benchmark building surface for (a) incident wind angle of 0°, and (b) incident
wind angle of 90°.

Alongwind Cross-wind Torque Alongwind Cross-wind Torque

front
back resultant
220 right 220
side resultant
left
side
150 150
Height (m)

Height (m)

front

back
100 100

left
50 side 50 right
side

0 0
–50 0 50 100 – 10 0 0 100 – 50 0 50 –50 0 –50 0 50
kN/m kN/m kN-m/m kN/m kN/m kN-m/m
(a) (b)

Figure 6. Distributed mean wind forces about mass centres of the benchmark building for (a) incident wind angle of 0°, and (b) incident
wind angle of 90˚.

similar over the building height, resulting in a mean 4. MODIFICATION OF THE BENCHMARK
crosswind force essentially equal to zero. The BUILDING
alongwind force increases gradually with the increasing The wind pressure measured from the wind tunnel test
mean wind speed over the building height, reflecting was applied to the benchmark building and the
the approach wind profile. The magnitudes of the alongwind deflection at the roof level was found to
predicted torques decrease significantly at approximately significantly exceed the allowable overall drift
one-third and two-thirds of building height, suggested in BS8110 (1997), and the standard deviation
corresponding to the core set-back at those two levels. It of acceleration of the highest habitable floor exceeded
can be readily shown that the torsional wind forces values suggested in ISO6897 (1984). Although the
decrease as the mass centre offsets towards the benchmark building was deliberately designed to have
centreline while pressure distributions are symmetrical low lateral stiffness and large wind-induced responses
about the centreline as well. so that vibration control is essential, the building

Advances in Structural Engineering Vol. 10 No. 3 2007 287


ASE 10-3-6 Tse 12/6/07 8:20 am Page 288

Vibration Control of a Wind-Excited Benchmark Tall Building with Complex Lateral-Torsional Modes of Vibration

responses were so large that only a special/super control rotations about the jth storey mass centre.
device could suppress the motions to an acceptable The mass matrix M180×180 in kg or kg⋅m2 was
level. Therefore, the structure of the benchmark determined by solving the characteristic equation:
building was modified using appropriate structural MIΩ2 = KI (3)
optimization techniques.
Strength checking was performed at the outset with where I180×180 = [I1 I2 … I180] is the modal matrix with
the sizes of core walls ranging from 300 mm to 600 mm. Ii being the ith mode shape output from SAP2000. The
The natural frequencies of the first three modes of diagonal matrix Ω2 is the so-called spectral matrix
vibration were consequently increased from 0.116, containing eigenvalue elements which are the square of
0.213, and 0.229 Hz to 0.140, 0.224, and 0.262 Hz, the building’s natural frequencies in radians per second.
respectively. However, the overall stiffness of the The proportional damping matrix C180×180 in N⋅s/m or
building system was still relatively low as only the N⋅s⋅m/rad was calculated by assuming a damping ratio
strength requirements were considered. Steel “belt”, of 1.0% of critical for the first six modes of vibration
trusses were hence employed at the outriggers levels to using Rayleigh’s approach. It should be emphasized that
also mobilize other peripheral columns and improve the only the principal equations and procedures are outlined,
overall lateral stiffness as well as torsional rigidity. A as the actual computer programming is complicated with
size optimized design, which aims to satisfy the overall a large range of values (e.g. 10–5 for mode shape, and
drift criteria of H/400, was then carried out by 1012 for stiffness quantities) being handled, sometimes
employing a drift optimization method (Chan, 1997 and resulting in significant round-off error.
2001). Correspondingly, the natural frequencies were
increased to 0.231, 0.429 and 0.536 Hz, respectively. The 6. STATE REDUCED-ORDER SYSTEM
deflection at roof level was successfully reduced to a The equation of motion of the evaluation model of the
reasonable level while vibration control was still benchmark building under wind excitation can be
required to suppress accelerations to satisfy occupant expressed as:
comfort criteria. The first six mode shapes of the
modified benchmark building are displayed in Figure 7. && + Cx& + Kx = GW
Mx (4)
The first mode is still a 3D lateral-torsional mode while in which x = [x1, x2,…, x59, x60, y1, y2,…, y59, y60,
the second mode is no longer purely translation as the θ1,θ2,…, θ59, θ60]’ is the displacement vector in metres
structural system is not symmetrical after the sizing or radians, W is the wind excitation vector in N or N⋅m
optimization. and G is a dimensionless excitation influence matrix
indicating the locations of wind loads. In state space
5. EVALUATION MODEL form, Eqn 4 becomes:
The evaluation model employed to design damping
Z& = AZ + BW (5)
devices and to compare control algorithms in this study
is a mathematical representation of the benchmark '
where Z360×360 = x' , x& '  is the state vector, A is a
building with the same structural dynamic properties of
system matrix and B is a location vector.
the FEM, such as mass and stiffness distributions,
Eqn 5 is referred to as the full-order system (FOS) and
natural frequencies of vibration, and mode shapes. The
usually involves a large number of degrees of freedom. The
evaluation model was determined based on quantities
numerical simulation of a FOS may be computationally
output from the finite element software, SAP2000. The
expensive and time-consuming, and hence a reduced-order
mass and stiffness matrices were resolved in an indirect
system (ROS) was derived through a state reduced-order
way as SAP2000 does not provide a function to output
scheme (SROS) such that the eigen properties of the
those matrices. The stiffness matrix K180×180 in N/m or
selected states and selected modes of the FOS are retained
N⋅m/rad was firstly determined by computing the
(Wu et al. 1998). The state reduced-order scheme was
inverse of a flexibility matrix Flex as following:
selected above other available schemes because the
K = Flex–1 = [f1, f2, …, f179, f180]–1 (2)
retained states in the ROS maintain the physical meaning
in which fi = [x1, x2,…, x59, x60, y1, y2,…, y59, y60, θ1, of the FOS (e.g. storey displacement responses: xi, yi, and
θ2,…, θ59, θ60]’ is the output displacement vector in θi) and hence can be measured directly by the sensors.
metres or radians due to the application of a unit Eqn 5 can be rearranged and partitioned as:
translational load or unit torque acting at the ith degree
&   A A  Z  B 
Z
of freedom, where xj, yj, are the jth storey displacements
& =    +  W
r rr re r r
(6)
along the x-, y-axes, respectively, and θj are the Ze   A er A ee  Ze  Be 

288 Advances in Structural Engineering Vol. 10 No. 3 2007


ASE 10-3-6 Tse 12/6/07 8:20 am Page 289

K. T. Tse, K. C. S. Kwok, P. A. Hitchcock, B. Samali and M. F. Huang

1st mode 2nd mode 3rd mode


250 250 250
y z x x y z y x z

200 200 200


Height (m)

Height (m)

Height (m)
150 150 150

100 100 100

50 50 50

0 0 0
–1 –0.5 0 0.5 –1 –0.5 0 0.5 –0.5 0 0.5 1

4th mode 5th mode 6th mode


250 250 250
y z x x y z y x z

200 200 200


Height (m)

Height (m)

Height (m)
150 150 150

100 100 100

50 50 50
FOS
O ROS
0 0 0
–1 0 1 –1 0 1 –1 0 1

Figure 7. Mode shapes of FOS and ROS models of the benchmark building.

in which Zr is the state vector being retained in the implies the modal coordinates Yi of the ROS differ from
ROS. If it is assumed that only the first n modes those of the FOS. However, identical modal coordinates
contribute significantly to the response, the state vector can be preserved by modifying Eqn 9 as:
can be approximated as:
& = A* Z + B* W
Z (10)
 Zr   I rr I re   Yr   I rr  r r r r

 =    ≈   Yr (7)
Ze  I er I ee  Ye  I er 
(
A*r = Ar + Are I er I rr−1 ; )
where Irr and Ier are sub-matrices of the first n mode
shape vectors re-arranged in accordance with the
retained states and Yr is the modal coordinate vector of B*r = I rr S1' S'2 K S'n  B (11,12)
the first n modes. The modal coordinate vector in Eqn 7
was eliminated, resulting in:
where Si is the ith row vector of I–1.
The ROS for the benchmark building has been derived
Z e = I er I rr−1Z r (8) '
with Zr = x 'r , x& 'r  being the 84-dimensional retained
Substituting Eqn 8 into Eqn 6, the state equation for the state vector and x 'r = [xj, yj, θj] for values of j corres-
ROS was evaluated as: ponding to the 5th, 10th, 15th, 20th, 25th, 30th, 35th, 39th,

( )
43rd, 47th, 51st, 54th, 57th, and 60th storeys. The
& = A + A I I −1 Z + B W
Z (9) refuge floors at the 20th and 39th storeys and the roof
r r re er rr r r
were intentionally retained because damping devices are
It is noteworthy that the approximation made in Eqn 7 generally installed at those locations to avoid occupying

Advances in Structural Engineering Vol. 10 No. 3 2007 289


ASE 10-3-6 Tse 12/6/07 8:20 am Page 290

Vibration Control of a Wind-Excited Benchmark Tall Building with Complex Lateral-Torsional Modes of Vibration

rentable floor space. Additional storeys were selected in direction are considerably smaller because of the
order to achieve smooth mode shapes that were building orientation. The large torsional responses at the
comparable to those obtained from the FOS. Fourteen corner are due to the significant 3D lateral-torsional
storeys were selected so that they were distributed over motion that was introduced deliberately to the building.
the building height and hence the first 42 modes (i.e. It is more meaningful to consider the resultant
14  3DOF) were retained in the ROS. responses of these lateral-torsional dynamic motions,
To validate the accuracy and possibility of with contributions due to translational motion in x- and
substitution of ROS for FOS, one can start by y-directions and torsional motion, such as the roof
comparing the natural frequencies of vibration and level resultant response traces at a corner, as shown in
mode shapes between the models. Further comparison Figure 11. The maximum displacement and acceleration
of the frequency response function and statistical values of the uncontrolled building, represented by the
of building responses, such as peak and standard magnitude of the vectors in Figure 11, are 295 mm and
deviation displacements and accelerations, can be made 62 milli-g, respectively.
through numerical simulations. The first six natural
frequencies of vibration of FOS and ROS were found to 7. DESIGN OF CONTROL DEVICE
be nearly identical, as expected, and the mode shapes of 7.1. Smart Tuned Mass Damper (STMD)
both models are presented in Figure 7. Passive control devices, such as tuned mass damper
To further examine the accuracy of the ROS model (TMD), tuned liquid damper (TLD), and tuned liquid
for the wind-excited benchmark building, an impulse column damper (TLCD), have been shown by many
(delta-function input) load was applied to both models researchers as being capable of mitigating the dynamic
to obtain the unit-impulse response function, from motions of civil structures and having the advantages of
which the frequency response function was subsequently reliability and comparatively low operating and
determined through Fourier Transformation. The maintenance costs (Spencer and Sain 1997). For example:
frequency response function, which represents the Kwok and Samali (1995) have studied the performance of
dynamic properties of a physical system in the frequency both passive and active TMDs under wind loads through
domain, is generally a complex quantity with the parametric studies and full-scale measurements; two
magnitude referred to as the system gain factor. The gain identical TMDs were designed to be installed on a 492 m
factors of ROS and FOS are virtually identical for both high super tall building using wind tunnel high-
models and are presented in Figure 8. frequency-force-balance (HFFB) data and the wind-
Finally, numerical simulations were conducted to induced responses were shown to be considerably
evaluate the dynamic responses of both models under mitigated (Tang and Gu 2006); Tamura et al. (1996) have
the same wind loads obtained from the wind tunnel confirmed the efficiency of TLD by measuring the wind-
pressure test, with proper conversion to prototype scale. induced responses of the 77.6 m high Tokyo International
The displacement and acceleration responses at the roof Airport Tower with and without TLDs; the optimal head-
level of the building are presented in Figures 9 and 10 loss coefficient and damping ratio for TLCD were also
for incident wind angles of 0° and 90°, respectively. It investigated and can be computed in a single step for a
can be seen from Figures 9 and 10 that the responses of given level of wind (Yalla and Kareem 2000). However,
the ROS and FOS models are nearly identical, validating the performance of a passive device is very difficult to
the ROS model for use in the design of damping devices optimize due to uncertainties in the structural dynamic
and the comparison of control algorithms while reducing properties and the excitation, as they are generally
the computational workload. designed for a specific condition (Housner et al. 1997).
Peak and standard deviation displacements and In contrast, semi-active control devices have the
accelerations of selected storeys of the FOS model are potential to address this challenge as they accommodate
tabulated in Table 1 for an incident wind angle of 0°. It the reliability of passive devices and the adaptability
should be noted that the twist responses in radians were of a fully active system, but with lower input energy
multiplied by the distance from the furthest corner to the demands than a fully active system (Spencer and Sain
storey mass centre to represent the torsional responses at 1997). One example of a semi-active control device is to
the corner in length units. The percentage errors of the equip a TMD with a magnetorheological (MR) damper
wind-induced responses of ROS to those of FOS are to provide variable stiffness and damping, referred to as
also tabulated in Table 1 for further comparison. The smart tuned mass damper (STMD). Ni et al. (2004) and
maximum displacement response of 262 mm and Wang et al. (2005) have devised a semi-active tuned
acceleration response of 53 milli-g were observed in the liquid column damper using magnetorheological
alongwind direction. The responses in the crosswind fluid (MR-TLCD) for wind-induced vibration mitigation

290 Advances in Structural Engineering Vol. 10 No. 3 2007


ASE 10-3-6 Tse 12/6/07 8:20 am Page 291

K. T. Tse, K. C. S. Kwok, P. A. Hitchcock, B. Samali and M. F. Huang

Frequency response function


100
ROS
Spectrum magnitude 10–2 FOS
10–4
10–6
10–8
10–10
10–12
10–14
10–16
0.1 1 2
Frequency (Hz)

Figure 8. Frequency response functions of FOS and ROS models of the benchmark building.

Tip translational response in X-direction Tip translational response in X-direction


30 15
FOS
20 10
Acceleration (milli-g)

ROS
Displacement (mm)

10 5

0 0

–10 –5

–20 –10

–30 –15
3000 3100 3200 3300 3400 3500 3600 3000 3100 3200 3300 3400 3500 3600
Time (sec) Time (sec)

Tip translational response in Y-direction Tip translational response in Y-direction


300 60

200 40
Acceleration (milli-g)
Displacement (mm)

100 20

0 0

–100 –20

–200 –40

–300 –60
3000 3100 3200 3300 3400 3500 3600 3000 3100 3200 3300 3400 3500 3600
Time (sec) Time (sec)

Tip twist response at corner Tip twist response at corner


40 20
Acceleration (milli-g)
Displacement (mm)

20 10

0
0

–20 –10

–40 –20
3000 3100 3200 3300 3400 3500 3600 3000 3100 3200 3300 3400 3500 3600
Time (sec) Time (sec)

Figure 9. Tip displacement and acceleration responses of FOS and ROS of the wind-excited benchmark building for an incident wind
direction of 0°.

Advances in Structural Engineering Vol. 10 No. 3 2007 291


ASE 10-3-6 Tse 12/6/07 8:21 am Page 292

Vibration Control of a Wind-Excited Benchmark Tall Building with Complex Lateral-Torsional Modes of Vibration

Tip translational response in X-direction Tip translational response in X-direction


20 10
FOS

Acceleration (milli-g)
Displacement (mm)

10 ROS 5

0 0

–10 –5

–20 –10
3000 3100 3200 3300 3400 3500 3600 3000 3100 3200 3300 3400 3500 3600
Time (sec) Time (sec)

Tip translational response in Y-direction Tip translational response in Y-direction


300 40

200
Acceleration (milli-g)
Displacement (mm)

20
100

0 0

–100
–20
–200

–300 –40
3000 3100 3200 3300 3400 3500 3600 3000 3100 3200 3300 3400 3500 3600
Time (sec) Time (sec)

Tip twist response at corner Tip twist response at corner


50 40
Acceleration (milli-g)
Displacement (mm)

20 20

0 0

–25 –20

–50 –40
3000 3100 3200 3300 3400 3500 3600 3000 3100 3200 3300 3400 3500 3600
Time (sec) Time (sec)

Figure 10. Tip displacement and acceleration responses of FOS and ROS of the wind-excited benchmark building for an incident wind
direction of 90°.

300 60

200 40
Y-acceleration (milli-g)
Y-displacement (mm)

100 20

0 0

–100 –20

–200 –40

–300 –60
–300 –200 –100 0 100 200 300 –60 –40 –20 0 20 40 60
X-displacement (mm) X-acceleration (milli-g)

Figure 11. Tip displacement and acceleration response traces of FOS and ROS of the wind-excited benchmark building.

292 Advances in Structural Engineering Vol. 10 No. 3 2007


ASE 10-3-6 Tse 12/6/07 8:21 am Page 293

K. T. Tse, K. C. S. Kwok, P. A. Hitchcock, B. Samali and M. F. Huang

Table 1. Numerical simulation responses at selected aeroelastic effects are known to produce aerodynamic
storeys of FOS of wind-excited benchmark stiffness and damping, causing variations of building
building and the percentage error of ROS dynamic properties. Evidently, the STMD has the
advantage of real-time tuning of its frequency and hence
Displacement (mm)
it is more robust to the uncertainties of building stiffness
Standard and damping due to inaccuracies in the analytical
deviation Peak estimation or wind-induced aeroelastic effects, and
changes in excitation characteristics.
Storey No. FOS error (%) FOS error (%)
MR damper, a newly developed variable stiffness
X (crosswind) 5 0.2 0.3 0.7 0.5 device, has shown great promise for civil engineering
20 1.2 0.0 5.5 0.1 applications. This type of damper utilizes special
characteristics of a MR fluid that possesses
39 2.9 0.0 13.8 0.0
magnetically controllable yield strength and can
60 5.0 0.0 23.7 0.0 reversibly change from free-flow to semi-solid in milli-s.
Y (alongwind) 5 1.2 0.1 4.3 0.1 Recognizing the significant potential of MR-based
20 13.6 0.0 48.3 0.0
control devices, a number of researchers have
undertaken studies to investigate vibration mitigation
39 39.8 0.0 141.1 0.0 using such devices (e.g. Dyke et al. 1996, 1998; Hansen
60 73.9 0.0 261.9 0.0 et al. 1997; Spencer et al. 1997; Yang 2001; Chang and
Z (torsion) 5 0.7 0.1 2.7 0.3 Zhou 2002; Tse 2002; Ni et al. 2004; Christenson et al.
2006). Furthermore, Yoshida and Dyke (2003, 2005)
20 4.0 0.0 16.7 0.0
have investigated the performance of MR dampers to
39 7.0 0.0 29.9 0.0 mitigate the coupled lateral and torsional motions in
60 8.3 0.0 36.4 0.0 model scale and subsequently extended that research to
experiments on full-scale 9-storey building models.
A 20 tonne MR damper, which was built at the
Acceleration (milli-g)
University of Notre Dame to study its possible
Standard application to real structures (Yang 2001, 2002), was
deviation Peak mounted in each orthogonal direction of a bi-directional
TMD as a semi-active system to numerically illustrate
Storey No. FOS error (%) FOS error (%)
the vibration control of the benchmark building.
X (crosswind) 5 0.1 5.7 3.6 1.4 The schematic diagram of the STMD to be installed on the
20 0.6 0.4 6.5 0.7 benchmark building roof is shown in Figure 12, and
comprises a bi-directional TMD with an inertial mass of
39 1.7 0.0 8.1 0.0
520 tonnes in parallel with a MR damper in each
60 3.0 0.0 11.0 0.0 orthogonal direction of operation to provide additional
Y (alongwind) 5 0.4 0.1 2.5 1.7 controllable damping (forces). The effective mass of the
20 2.6 0.6 9.9 2.3
STMD is about 0.4% of the total mass of the building,
and its undamped natural frequency in the y-direction
39 7.5 0.0 28.3 0.0 and x-direction are set at 0.22 Hz and 0.42 Hz,
60 14.0 0.0 53.4 0.0 respectively. The damping ratios of the STMD for both
Z (torsion) 5 0.3 7.8 2.3 –0.9 directions are 5% of critical. The STMD’s undamped
natural frequencies are slightly less than the first two
20 1.7 0.3 7.3 1.9
fundamental natural frequencies of vibration of the
39 3.6 0.0 16.8 0.0 benchmark building, which are 0.231 Hz and 0.429 Hz,
60 4.9 0.0 28.5 0.0 so that the MR dampers provide additional controllable
stiffness and damping forces to optimize the natural
frequencies of the STMD when operating in its passive
and analysis results showed that MR-TLCD with mode. Because the benchmark building undergoes 3D
optimal parameters can achieve much better vibration lateral-torsional motions, the STMD was offset from the
mitigation capability than a conventional TLCD system. geometrical centre of the building at a distance of 32 m
For the wind-induced responses of a wind-sensitive in the x-direction and 8 m in y-direction, as shown in
structure such as the proposed benchmark building, Figure 12, to suppress both translational and torsional

Advances in Structural Engineering Vol. 10 No. 3 2007 293


ASE 10-3-6 Tse 12/6/07 8:21 am Page 294

Vibration Control of a Wind-Excited Benchmark Tall Building with Complex Lateral-Torsional Modes of Vibration

72 m

4m

Building
roof

24 m 4m

MR damper

spring
M
dashpot

Building perimeter

Figure 12. Schematic diagram of the plan view of a STMD installed on the benchmark building roof.

motions. It should be noted that it was assumed that the through a reduced-order observer (Shahian and
STMD did not induce any additional wind forces on the Hassul 1993) for computing the control forces. The
building although it was installed on the roof. state-space realization for the dynamic system and
the measurement of the building model installed with
7.2. Control Algorithm the control device was expressed as,
In this study, the linear-quadratic regulator (LQR) was & = AZ + BW + Ef
Z c (14)
used for the semi-active control. The control force fc was
determined by minimizing the following quadratic cost
function over a period of time (t), Y = CZ + DW + Ffc (15)

J = ∫0 ( Z' NZ + fc ' Rfc )dt (13) where Z and Y are the state vector and the measured
output, respectively. W is the wind excitation vector in
where N and R are the weighting matrices and Z is the state
N or N-m. The matrices A, B, E, C and D, F are,
vector. By adjusting these weighting matrices, one can
reduce the structural responses with different objectives,  0 I   0   0 
such as the tip displacement, the tip acceleration, or the A= −1 
; B =  −1  ; E =  M −1Λ ;
−1
− M K − M C  M Γ   
inter-storey drifts, with regards to the maximum available
control force generated by the damping device. This (16a-c)
means R should be chosen appropriately so that the control
force is reasonable, or even minimized.
−1
In principle, accelerations and velocities of all C = C − M −1K − M −1C  ; D = C  M Γ  ;
floors can be measured although this is likely to be
impractical and unnecessary in a real application.
Therefore, in the current study, it was assumed that F = C  M −1Λ  (16d-f)
only the accelerations in x- and y-directions at the
corner of the refuge floors and the roof were available where C is the measurement (i.e. accelerometers)
for computing the feedback control forces. The res- location matrix; M and K are the mass and stiffness
ponse state vector of the building model, including matrices, respectively; Γ and Λ represent the wind load
displacement and velocity vectors, was first estimated and control device location matrices, respectively.

294 Advances in Structural Engineering Vol. 10 No. 3 2007


ASE 10-3-6 Tse 12/6/07 8:21 am Page 295

K. T. Tse, K. C. S. Kwok, P. A. Hitchcock, B. Samali and M. F. Huang

Under the condition that the dimension of Y is smaller The coupled Eqns 22 and 23 are the final form of the
than Z, an arbitrary matrix T can be assumed such that reduced-order observer where the state estimate Ẑ can
C  is nonsingular and be obtained by inputting the measured output Y, the
T  control forces fc and the wind loads W. This state
 
estimate can then be used to calculate the control forces,
Y − Ffc  C  fc = −kẐ (25)
  =  Z (17)
 x  T 
where k is the feedback gain matrix and is manipulated
where x is the observer vector. Estimates of Z, Ẑ can from the cost function of the LQR in Eqn 13 with the
then be formed as, weighting matrix N a diagonal matrix with
components [1, 1, 1, …, 300, …] and R also a diagonal
ˆ = P ( Y − Ff ) + Mxˆ
Z c (18) matrix with components [9 × 10–2, 9 × 10–2] for this study.

7.3. An Inverse Model for Control Application


−1
C  One of the challenges in vibration control, particularly
P M  =   (19) semi-active and active control, is to assure the control
T  device generates the target control force determined
from the control algorithm. The failure of control force
where x̂ are the estimates of x and can be designed as,
generation could result in unfavourable building
x&ˆ = A 22 xˆ +  A 21 ( Y − Ffc ) + E 2 fc + B2 W 
responses. It has been suggested (Jung et al. 2004) that
the force generated by a MR damper cannot be directly
commanded. Only the voltage (v) supplied to the current
& − Ff − A ( Y − Ff ) − A xˆ − E f − B W  (20)
+ L Y driver for the MR damper can be arbitrarily changed.
c 11 c 12 1 c 1  Tse and Chang (2004) proposed an inverse dynamic
model, in which the damper force f is simplified as,
with
(c1a + c1b u)
A11 = CAP ; A12 = CAM ; f≅
(c0a + c1a ) + (c0b + c1b )u
(26)
(α a + α b u) zu + (c0 + c0 u) x& + k0 x 
A 21 = TAP ; A 22 = TAM ; (21a-d)  a b 
+ k1 ( x − x 0 )

E1 = CE ; E 2 = TE ; B1 = CG ; B2 = TG (21e-h)
1n 1n
The matrix L in Eqn 20 should be properly chosen so as  A   A 
z ≅ z u = sgn( x& − y& )  ≅ sgn( x& )  (27)
to place the observer eigenvalues arbitrarily in the left-  γ + β   γ + β 
hand side of the complex plane. Let Q = xˆ − L(Y − Ffc ) ,
where c0, c1, k0, k1, α, A, γ, and β are the damper
then Eqns 18 and 20 can be re-written as,
parameters (Spencer et al. 1997). Eqn 26 can be
rearranged to a quadratic function in u:
Zˆ = MQ + (P + ML)( Y − Ffc ) (22)

(α c zu ) u 2 + (α b zu + c0 x& ) u + (α a zu + c0 x& + k0 x − f ) = 0
b a
& = VQ + H( Y − Ff ) + If + J&&s
Q (23)
c c g
(28)
where
Evidently, u at any time instance can be solved rather
V = A 22 − LA12 ; H = FL + A 21 − LA11 ; (24a,b) easily in terms of f, which is the desired control force
from Eqn 25 based on the feedback LQR algorithms.
Consequently, the corresponding command voltage (v)
I = E 2 − LE1 ; J = B2 − LB1 (24c,d) can be obtained numerically from

Advances in Structural Engineering Vol. 10 No. 3 2007 295


ASE 10-3-6 Tse 12/6/07 8:21 am Page 296

Vibration Control of a Wind-Excited Benchmark Tall Building with Complex Lateral-Torsional Modes of Vibration

W
uncertainties in the dynamic properties of the building,
by considering variations to the stiffness and damping
TMD fD Benchmark yi matrices of ±20%. The response quantities of all cases
+ Building without control are tabulated in Tables 2 and 3 for wind
MR
y1, y1 directions of 0° and 90°, respectively.
It can be seen from Tables 2 and 3 that the wind-
induced responses, such as standard deviation and
fD peak values of displacement and acceleration,
V Inverse fC Control
increase for the cases in which the stiffness matrix or
Observer damping matrix was reduced by 20%. This is because
model algorithm
the structure becomes slightly more flexible due to
Figure 13. Block diagram of the control methodology of the wind- the reduction of stiffness, resulting in lower natural
excited benchmark building. frequencies. Therefore, the displacement responses at
the building roof increase by nearly 20% while the
corresponding accelerations increase by only 5%,
. which is attributed to the combined effect of the
v = u + u /η (29) simultaneously lower natural frequency and higher
displacement response (i.e. x&& = ω 2 x ). Furthermore,
Figure 13 shows a block diagram illustrating how the
the reduction of damping increases the displacement
inverse model was used in the control application. The
and acceleration responses by around 10% because
measured outputs, comprising the accelerations in x-
the natural frequency remains unchanged. In contrast,
and y-directions at the refuge floors and the roof, were
the building’s displacement and acceleration
used to estimate the building responses of other floors
responses both decrease as the stiffness matrix or
with an observer. The desired control force based on the
damping matrix is increased by 20%, owing to the
LQR algorithm was then established with the knowledge
rise in natural frequency and energy dissipation
of the building response estimates and the instant-
capacity.
aneous actual force generated by the control device. A
The building responses with control were compared
voltage was subsequently calculated with the inverse
with the uncontrolled responses, as shown in Figures
dynamic model, with which the MR damper can
14 and 15, and the ratios of the controlled response to
generate forces considerably closer to the desired
that of uncontrolled response are tabulated in Tables 4
control force.
and 5. In general, the STMD shows its capability to
mitigate the wind-induced orthogonal translation
7.4. Numerical Results responses as well as the torsion responses. The reduction
The state-space realization for the dynamic system and of acceleration responses are more pronounced than the
the measurement of the building model installed with displacement responses as the LQR was designed to
the control device was numerically simulated. The compute control forces to minimize the acceleration
structural responses (i.e. accelerations at refuge floors responses in this study. In other words, the STMD is
and roof), the observer to estimate other response state also capable of reducing the displacement responses,
vectors of the building model, and the controller if the LQR is designed for that purpose, by changing
comprising a control algorithm and an inverse model the values of the weighting matrix, N. It is noteworthy
were digitally implemented with a time step of 0.001 s. that the magnitudes of the response ratios
To make the simulation realistic, measurement noises corresponding to the cases in which the stiffness or
modelled as Gaussian rectangular pulse processes were damping were varied are of similar magnitudes to the
introduced to the building responses and a computational building with original stiffness and damping matrices,
time-delay of 1 milli-s was considered for the building indicating the robustness of the STMD and the
responses simulation. necessity of MR dampers to provide variable stiffness
For the wind-induced responses of a wind-sensitive and damping.
structure such as the proposed benchmark building, To quantify the control effectiveness, the following
aerodynamic stiffness and aerodynamic damping are performance indices are established:
motion-induced effects that have the potential to cause
variations of building dynamic properties. Such J1 = max(σ xi&& )/σ xTip
&& ; J 2 = max(σ xi )/σ xTip ;
aeroelastic effects were taken into account in this study
to investigate the robustness of the STMD to

296 Advances in Structural Engineering Vol. 10 No. 3 2007


Table 2. Numerical simulation responses at selected storeys of the benchmark building for an incident wind angle of 0ο

Displacement (mm)
ASE 10-3-6 Tse

Standard deviation Peak

Storey No. Original k + 20% k – 20% c + 20% c – 20% Original k + 20% k – 20% c + 20% c – 20%
X (crosswind) 5 0.2 0.1 0.2 0.1 0.2 0.7 0.6 1.0 0.7 0.8
12/6/07

20 1.2 0.9 1.5 1.1 1.2 5.5 4.2 8.0 5.4 5.7
39 2.9 2.3 3.8 2.7 3.1 13.8 10.1 19.9 13.4 14.3
60 5.0 4.0 6.6 4.7 5.3 23.7 17.2 34.2 23.0 24.7
8:21 am

Y (alongwind) 5 1.2 0.9 1.4 1.1 1.3 4.3 4.1 6.9 4.0 4.7
20 13.6 10.4 16.1 12.7 14.9 48.3 48.0 74.8 44.7 52.7
39 39.8 30.2 47.1 37.0 43.6 141.1 140.2 212.4 130.4 154.0
Page 297

60 73.9 56.1 87.5 68.6 81.2 261.9 256.6 388.5 241.9 286.2

Advances in Structural Engineering Vol. 10 No. 3 2007


Z (torsion) 5 0.7 0.5 0.8 0.6 0.7 2.7 2.7 4.3 2.6 2.9
20 4.0 3.1 4.8 3.8 4.3 16.7 16.0 26.7 16.1 18.3
39 7.0 5.4 8.6 6.6 7.6 29.9 28.6 46.8 28.8 33.0
60 8.3 6.4 10.2 7.8 9.1 36.4 32.1 56.2 34.6 40.1
K. T. Tse, K. C. S. Kwok, P. A. Hitchcock, B. Samali and M. F. Huang

Acceleration (milli-g)

Standard deviation Peak

Storey No. Original k + 20% k – 20% c + 20% c – 20% Original k + 20% k – 20% c + 20% c – 20%
X (crosswind) 5 0.1 0.1 0.1 0.1 0.1 3.6 3.3 3.6 3.4 3.6
20 0.6 0.6 0.7 0.6 0.7 6.5 6.2 6.5 6.3 6.5
39 1.7 1.6 1.8 1.5 1.8 8.1 5.8 8.5 5.8 8.5
60 3.0 2.8 3.2 2.7 3.3 11.0 10.1 11.6 10.5 11.8
Y (alongwind) 5 0.4 0.4 0.4 0.4 0.5 2.5 2.1 2.6 2.4 2.5
20 2.6 2.3 2.7 2.4 2.9 9.9 8.4 10.5 9.4 10.2
39 7.5 6.6 7.8 6.8 8.4 28.3 25.3 23.0 26.1 31.0
60 14.0 12.4 14.9 12.7 15.8 53.4 48.1 60.5 48.8 58.9
Z (torsion) 5 0.3 0.3 0.3 0.3 0.3 2.3 2.2 3.5 2.4 4.1
20 1.7 1.6 1.8 1.6 1.9 7.3 6.1 8.5 7.0 10.5
39 3.6 3.3 3.8 3.2 4.0 16.8 12.4 17.5 15.7 22.4

297
60 4.9 4.5 5.2 4.5 5.5 28.5 17.4 29.9 26.7 30.5
Table 3. Numerical simulation responses at selected storeys of the benchmark building for an incident wind angle of 90ο

Displacement (mm)

298
Standard deviation Peak
ASE 10-3-6 Tse

Storey No. Original k + 20% k – 20% c + 20% c – 20% Original k + 20% k – 20% c + 20% c – 20%
X (alongwind) 5 0.1 0.1 0.2 0.1 0.1 0.5 0.4 0.7 0.5 0.5
12/6/07

20 1.0 0.7 1.3 0.9 1.0 4.2 2.6 5.9 4.0 4.4
39 2.5 1.9 3.3 2.3 2.6 10.9 6.6 15.7 10.5 11.5
60 4.3 3.3 5.8 4.1 4.6 19.3 11.5 27.3 18.4 20.3
8:21 am

Y (crosswind) 5 1.3 1.0 1.5 1.2 1.4 5.1 4.1 5.8 4.8 5.3
20 14.2 11.0 17.3 13.2 15.6 55.8 47.6 66.5 52.8 59.2
39 41.5 32.3 50.6 38.5 45.7 160.8 140.7 197.1 152.0 170.8
Page 298

60 77.2 60.1 94.1 71.5 85.2 296.0 262.6 368.9 279.4 314.7
Z (torsion) 5 0.7 0.6 0.9 0.7 0.8 2.9 2.5 3.7 2.8 3.1
20 4.7 3.6 6.1 4.4 5.1 19.1 16.1 26.0 18.0 20.9
39 8.9 6.8 11.7 8.4 9.6 37.8 29.0 54.1 35.6 41.1
60 11.1 8.4 14.9 10.4 12.0 48.7 35.9 71.4 45.8 53.1

Acceleration (milli-g)

Standard deviation Peak

Storey No. Original k + 20% k – 20% c + 20% c – 20% Original k + 20% k – 20% c + 20% c – 20%
X (alongwind) 5 0.1 0.1 0.1 0.1 0.1 2.9 2.7 2.9 1.1 2.9
20 0.5 0.5 0.6 0.5 0.6 3.7 3.4 3.7 2.0 3.9
39 1.4 1.2 1.6 1.3 1.6 5.3 4.1 6.8 5.0 5.5
60 2.5 2.1 2.8 2.3 2.8 9.3 7.4 12.1 8.9 9.7
Y (crosswind) 5 0.3 0.3 0.3 0.3 0.4 2.3 2.2 2.3 2.2 2.4
20 2.8 2.6 2.8 2.5 3.1 10.5 9.8 13.2 10.0 11.4
39 8.0 7.4 8.3 7.3 9.0 29.5 26.9 29.7 26.2 30.1
60 15.1 13.9 15.9 13.7 16.9 54.2 51.7 57.2 51.4 57.5
Z (torsion) 5 0.4 0.4 0.5 0.4 0.4 2.5 2.4 2.7 2.4 3.9
20 2.8 2.4 3.3 2.5 3.1 9.5 9.3 14.3 9.3 15.9
39 5.8 5.1 7.0 5.3 6.4 30.7 21.7 42.6 29.4 32.1

Advances in Structural Engineering Vol. 10 No. 3 2007


Vibration Control of a Wind-Excited Benchmark Tall Building with Complex Lateral-Torsional Modes of Vibration

60 8.0 7.1 9.7 7.3 9.0 43.0 30.4 63.0 41.3 44.9
ASE 10-3-6 Tse 12/6/07 8:21 am Page 299

K. T. Tse, K. C. S. Kwok, P. A. Hitchcock, B. Samali and M. F. Huang

Tip translational response in X-direction Tip translational response in X-direction


30 15
Uncontrol

Acceleration (milli-g)
10
Displacement (mm)

20 STMD
10 5

0 0
–10 –5
–20 –10
–30 –15
3000 3100 3200 3300 3400 3500 3600 3000 3100 3200 3300 3400 3500 3600
Time (sec) Time (sec)

Tip translational response in Y-direction Tip translational response in Y-direction


300 60

Acceleration (milli-g)
200 40
Displacement (mm)

100 20

0 0

–100 –20

–200 –40

–300 –60
3000 3100 3200 3300 3400 3500 3600 3000 3100 3200 3300 3400 3500 3600
Time (sec) Time (sec)

Tip twist response at corner Tip twist response at corner


40 20
Acceleration (milli-g)
Displacement (mm)

20 10

0 0

–20 –10

–40 –20
3000 3100 3200 3300 3400 3500 3600 3000 3100 3200 3300 3400 3500 3600
Time (sec) Time (sec)

Figure 14. The uncontrolled and controlled tip displacement and acceleration responses of the wind-excited benchmark building for an
incident wind angle of 0°.

J3 = σ xTMD /σ xTip (30-32) with the storey acceleration as the control device was
installed to reduce the acceleration, and hence improve
occupant comfort. The second (J2) and third (J3) indices
represent the capability of the control device to reduce
J 4 = max( x&&ˆi )/ x&&ˆTip ; J5 = max( xˆi )/ xˆTip ;
the displacement response and the control effort in terms
of device stroke. The fourth (J4), fifth (J5) and sixth (J6)
J6 = xˆTMD /xˆTip indices are the peak response counterparts of J1, J2 and
(33-35)
J3. The performance indices shown in Table 6 confirm
the ability of the control device to mitigate wind-
where σẍi and σẍTip are the standard deviation induced vibrations. The standard deviation accelerations
accelerations of the ith storey at the corner having the and displacements with control were reduced by over
largest responses and the roof level without control, 50% and the peak responses were also reduced by about
respectively. σ xi , σ xTip and σ xTMD are the standard 25%. The maximum control force generated was 476
deviation displacements of the ith storey at the most kN and the maximum stroke was 526 mm, which are
severe corner, roof level without control and tuned mass within the typical range that can be achieved
damper, respectively. The first index (J1) is associated practically.

Advances in Structural Engineering Vol. 10 No. 3 2007 299


ASE 10-3-6 Tse 12/6/07 8:21 am Page 300

Vibration Control of a Wind-Excited Benchmark Tall Building with Complex Lateral-Torsional Modes of Vibration

Tip translational response in X-direction Tip translational response in X-direction


20 10
Uncontrol

Acceleration (milli-g)
Displacement (mm)

STMD
10 5

0 0

–10 –5

–20 –10
3000 3100 3200 3300 3400 3500 3600 3000 3100 3200 3300 3400 3500 3600
Time (sec) Time (sec)

Tip translational response in Y-direction Tip translational response in Y-direction


300 40

200
Acceleration (milli-g)
Displacement (mm)

20
100

0 0

–100
–20
–200

–300 –40
3000 3100 3200 3300 3400 3500 3600 3000 3100 3200 3300 3400 3500 3600
Time (sec) Time (sec)

Tip twist response at corner Tip twist response at corner


50 40
Acceleration (milli-g)
Displacement (mm)

25 20

0 0

–25 –20

–50 –40
3000 3100 3200 3300 3400 3500 3600 3000 3100 3200 3300 3400 3500 3600
Time (sec) Time (sec)

Figure 15. The uncontrolled and controlled tip displacement and acceleration responses of the wind-excited benchmark building for an
incident wind angle of 90°.

8. CONCLUSIONS A numerical vibration control example was conducted


A wind-excited benchmark tall building with complex to illustrate the suppression of the complex motions of the
lateral-torsional motions is proposed in this paper. A series wind-excited benchmark building using a bi-directional
of wind tunnel pressure tests were conducted to determine TMD integrated with two MR dampers, one in each
the wind loads experienced by the benchmark building. A orthogonal operation direction. The bi-directional TMD
finite element model of the benchmark building was also was designed with natural frequencies slightly less
constructed and an evaluation model was determined. The than the building’s first two natural frequencies and
evaluation model was further simplified to a state ROS equipped with MR dampers to provide additional
using the state order reduction method. Under wind loads controllable damping forces to significantly mitigate
obtained from the wind tunnel pressure test, the building both the acceleration and displacement responses in the
responses obtained using the ROS were found to be translation directions and slightly smaller reductions in
identical to those determined using the FOS. Therefore, the torsion. The maximum generated control forces were
ROS developed in this study is a useful tool that can be 35 kN with a maximum stroke of 42 mm in x-direction
used to design damping devices and to compare and and 476 kN with a maximum stroke of 526 mm in
evaluate control algorithms and devices. y-direction. Correspondingly, the standard deviation

300 Advances in Structural Engineering Vol. 10 No. 3 2007


Table 4. Controlled to uncontrolled response ratios at selected storeys of the benchmark building for an incident wind angle of 0ο

Displacement (%)

Standard deviation Peak


ASE 10-3-6 Tse

Storey No. Original k + 20% k – 20% c + 20% c – 20% Original k + 20% k – 20% c + 20% c – 20%
X (crosswind) 5 81.0 85.2 50.4 84.0 77.1 92.9 85.4 43.9 95.9 89.3
12/6/07

20 76.4 82.6 47.5 79.9 72.1 94.3 91.0 42.7 97.1 91.0
39 73.4 81.0 45.6 77.1 68.8 92.7 94.3 41.8 95.6 89.3
60 71.8 80.2 44.4 75.7 67.1 92.4 95.8 41.7 95.4 88.9
8:21 am

Y (alongwind) 5 64.2 76.4 44.3 68.7 59.0 88.8 75.4 45.7 95.6 81.6
20 61.5 74.5 42.8 66.1 56.2 82.9 72.8 43.7 89.6 76.0
39 60.2 73.7 42.0 64.8 54.9 78.9 72.2 42.6 85.4 72.3
Page 301

60 59.6 73.6 41.5 64.2 54.3 76.9 73.4 42.0 83.3 70.4
Z (torsion) 5 77.2 86.9 50.8 81.6 71.7 92.8 72.4 44.7 90.7 85.5

Advances in Structural Engineering Vol. 10 No. 3 2007


20 76.6 87.5 50.1 81.5 70.8 95.0 76.1 45.6 91.6 85.8
39 78.7 90.2 50.6 84.0 72.4 89.5 82.4 47.2 93.3 86.6
60 84.7 97.0 52.4 90.5 77.8 92.3 92.1 48.9 97.4 89.3
K. T. Tse, K. C. S. Kwok, P. A. Hitchcock, B. Samali and M. F. Huang

Acceleration (%)

Standard deviation Peak

Storey No. Original k + 20% k – 20% c + 20% c – 20% Original k + 20% k – 20% c + 20% c – 20%
X (crosswind) 5 65.6 78.9 67.4 70.7 59.8 98.2 98.8 98.6 97.9 95.5
20 49.5 68.3 52.5 53.8 44.7 97.1 98.0 97.9 96.6 92.3
39 47.5 67.2 50.4 51.7 42.8 61.9 76.0 66.6 64.5 58.5
60 47.4 67.6 50.1 51.6 42.8 58.3 77.7 64.7 61.3 54.5
Y (alongwind) 5 88.8 93.9 90.7 91.1 85.7 97.0 99.1 99.2 97.0 94.0
20 44.6 65.9 53.4 49.0 39.8 96.0 97.9 98.0 96.1 91.2
39 40.0 62.6 49.7 44.1 35.6 59.7 66.8 73.6 64.8 54.5
60 40.2 63.1 49.5 44.3 35.8 47.8 60.7 60.1 52.2 43.3
Z (torsion) 5 85.5 90.3 86.0 92.4 77.6 94.3 95.7 96.1 96.0 94.1
20 82.8 89.0 83.0 90.9 74.0 94.4 95.9 96.4 96.1 91.3
39 83.8 89.4 83.8 92.1 74.8 82.2 91.5 73.7 88.0 76.3

301
60 85.1 91.0 84.3 93.4 75.9 83.4 91.4 72.6 88.9 77.9
Table 5. Controlled to uncontrolled response ratios at selected storeys of the benchmark building for an incident wind angle of 90°

Displacement (%)

302
Standard deviation Peak
ASE 10-3-6 Tse

Storey No. Original k + 20% k – 20% c + 20% c – 20% Original k + 20% k – 20% c + 20% c – 20%
X (alongwind) 5 78.9 86.8 60.6 82.3 74.6 68.1 73.8 56.5 71.2 64.7
12/6/07

20 76.0 85.5 58.2 79.7 71.4 65.6 84.2 49.9 68.5 62.5
39 74.2 84.7 56.6 78.1 69.4 65.4 87.1 48.7 68.3 62.3
60 73.4 84.5 55.8 77.3 68.5 64.1 87.3 49.4 66.9 61.0
8:21 am

Y (crosswind) 5 62.1 75.1 51.8 66.7 56.6 76.9 64.9 68.4 81.0 72.8
20 60.1 73.7 50.3 64.7 54.6 76.1 60.9 65.6 80.4 71.8
39 58.9 73.0 49.4 63.5 53.5 75.7 59.1 63.4 80.1 71.2
Page 302

60 58.3 72.9 48.8 63.0 52.9 75.8 58.5 62.2 80.3 71.3
Z (torsion) 5 77.5 87.8 60.4 82.5 71.5 92.6 69.5 72.3 96.5 85.2
20 79.9 90.0 61.3 85.0 73.7 92.0 75.0 67.3 97.5 84.3
39 83.5 93.3 62.7 88.8 77.1 90.8 89.0 62.4 96.4 83.5
60 88.4 98.7 64.2 94.1 81.5 94.1 98.8 62.6 99.5 86.5

Acceleration (%)

Standard deviation Peak

Storey No. Original k + 20% k – 20% c + 20% c – 20% Original k + 20% k – 20% c + 20% c – 20%
X (alongwind) 5 76.4 90.0 73.6 81.0 70.8 91.7 99.3 76.4 77.7 80.1
20 49.6 73.2 49.7 54.2 44.4 95.5 99.1 75.5 76.1 85.4
39 47.8 72.6 48.3 52.4 42.8 56.8 98.3 57.6 59.4 54.6
60 47.6 72.9 47.8 52.1 42.6 56.1 94.2 54.8 58.6 53.9
Y (crosswind) 5 76.4 85.4 77.0 80.8 71.0 95.7 89.4 68.4 68.9 67.1
20 47.7 67.4 52.8 52.4 42.5 93.2 86.6 64.3 65.2 61.8
39 42.6 64.3 49.0 46.8 37.9 62.5 72.2 65.4 66.7 58.0
60 42.9 64.8 48.9 47.2 38.2 66.1 77.2 66.5 69.7 62.3
Z (torsion) 5 88.2 95.1 78.7 95.9 79.6 87.6 81.8 63.1 90.1 90.6
20 87.2 95.4 76.1 95.4 78.2 95.3 86.4 65.1 93.3 94.0
39 87.3 94.9 75.8 95.6 78.2 90.8 68.8 67.3 94.9 86.8

Advances in Structural Engineering Vol. 10 No. 3 2007


Vibration Control of a Wind-Excited Benchmark Tall Building with Complex Lateral-Torsional Modes of Vibration

60 88.0 96.5 76.3 96.3 78.8 89.4 81.9 62.5 93.1 85.6
ASE 10-3-6 Tse 12/6/07 8:21 am Page 303

K. T. Tse, K. C. S. Kwok, P. A. Hitchcock, B. Samali and M. F. Huang

Table 6. Performance indices seismic response reduction”, Smart Materials and Structures,
Vol. 5, pp. 565–575.
J1 J2 J3 J4 J5 J6
Dyke, S.J., Spencer, Jr. B.F., Sain, M.K. and Carlson, J.D. (1998).
0.402 0.596 1.726 0.479 0.769 2.011 “Experimental study of MR dampers for seismic protection”,
Smart Materials and Structures: Special Issue on Large Civil
Structures, Vol. 7, No. 5, pp. 693–703.
and peak responses were reduced by over 50% and
Guidelines for the Evaluation of the Response of Occupants of Fixed
about 25%, respectively.
Structures, Especially Buildings and Off-shore Structures, to Low-
The results of this study clearly demonstrate that the
frequency Horizontal Motion (0.063 to 1Hz), ISO pp. 1897–1984.
three-dimensional motions of the benchmark building
Hansen, B.C., Gordaninejad, F., Saiidi, M. and Chang, F.K. (1997).
can be effectively mitigated with a control device
“Control of bridges using magneto-rheological fluid dampers”,
operating within a practical operational range. The
Structural Health Monitoring, Fu-Kuo Chang, ed., Technomic
proposed STMD system serves as a practical application
Publishing Co., pp. 81–90.
while acknowledging that it was more effective in
Housner, G.W., Bergman, L.A., Caughey, T.K., Chassiakos, A.G.,
controlling translational motions. Future works by the
Claus, R.O., Masri, S.F., Skelton, R.E., Soong, T.T., Spencer,
authors and other researchers will investigate the
B.F. and Yao, J.T.P. (1997). “Structural control: pass, present,
possibility of using other or additional control systems
and future”, Journal of Engineering Mechanics, ASCE, Vol. 123,
and algorithms with a view to reducing the torsional
No. 9, pp. 897–971.
response in addition to translational responses.
Jung, H.J., Spencer Jr., B.F., Ni, Y.Q. and Lee, I.W. (2004). “State-
of-the-art of semiactive control systems using MR fluid dampers
ACKNOWLEDGEMENT in civil engineering applications”, Structural Engineering and
This research project is funded by a Research Grants Mechanics, ASCE, Vol. 17, pp. 493–526.
Council of Hong Kong Central Allocation Grant Kim, S.B., Yun, C.B. and Spencer Jr., B.F. (2004). “Vibration
(Project PolyU1/02C-CA02/03.EG03) and a HKUST control of wind-excited tall buildings using sliding mode fuzzy
High Impact Areas Grant (HIA02/03.EG01). control”, Journal of Engineering Mechanics, ASCE, Vol. 111,
pp. 505–510.
REFERENCES Kobori, T. (2003). “Past, present and future in seismic response
American Institute of Steel Construction, Inc., (1986). Manual of control in civil engineering structures”, Proceedings of 3rd World
Steel Construction-load and Resistance Factor Design, Chicage, Conference on Structural Control, Wiley, New York, pp. 9–14.
IL, First Edition. 1 Kwok, K.C.S. and Samali, B. (1995). “Performance of tuned mass
Ankireddi, S. and Yang, H.T.Y. (1996). “Simple ATMD control dampers under wind loads”, Journal of Engineering Structures,
methodology for tall buildings subject to wind loads”, Journal of ASCE, Vol. 17, pp. 655–667.
Structural Engineering, ASCE, Vol. 122, pp. 83–91. Ni, Y.Q., Ying, Z.G., Wang, J.Y., Ko, J.M. and Spencer Jr., B.F.
Australian/New Zealand Standard, Structural Design Actions Part 2: (2004). “Stochastic optimal control of wind-excited tall buildings
Wind Actions, AS/NZS 1170.2:2002. using semi-active MR-TLCDs”, Probabilistic Engineering
Cao, H., Reinhorn, A.M. and Soong, T.T. (1997). “Design of an Mechanics, Vol. 19, No. 3, pp. 269–277.
active mass damper for wind response of nanjing TV tower”, Samali, B., Yang, J.N. and Yeh, C.T. (1985). “Control of lateral-
Engineering Structure, Vol. 20, pp. 134–143. torsional motion of wind-excited buildings”, Journal of
Chan, C.M. (1997). “How to optimize tall steel building Engineering Mechanics, ASCE, Vol. 111, pp. 777–796.
frameworks”, Guide to Structural Optimization, ASCE Manuals Samali, B., Al-Dawod, M., Kwok, K.C.S. and Naghdy, F. (2004a).
and Reports on Engineering Practices No.90, American Society “Active control of cross wind response of 76-storey tall building
of Civil Engineers, pp.165–195. using a fuzzy controller”, Journal of Engineering Mechanics,
Chan, C.M. (2001). “Optimal lateral stiffness design of tall buildings ASCE, Vol. 130, No. 4, pp. 492–498.
of mixed steel and concrete construction”, Journal of Structural Samali, B., Mayol, E., Kwok, K.C.S., Mack, A. and Hitchcock, P.
Design of Tall Buildings, Vol. 10, No. 3, pp. 155–177. (2004b). “Vibration control of the wind-excited 76-storey
Chang, C.C. and Zhou, L. (2002). “Neural network emulation of benchmark building by liquid column vibration absorbers”, Journal
inverse dynamics for a MR damper”, Journal of Structural of Engineering Mechanics, ASCE,Vol. 130, No. 4, pp. 478–485.
Engineering, ASCE, Vol. 128, No. 2, pp. 231–239. Shahian, B and Hassul, M. (1993). Control System Design Using
Christenson, R.E., Spencer Jr., B.F. and Johnson, E.A. (2006). Matlab, Prentice-Hall, Inc., Englewood Cliffs, NJ.
“Experimental verification of smart cable damping”, Journal of Spencer, Jr. B.F., Christenson, R.E. and Dyke, S.J. (1998). “Next
Engineering Mechanics, ASCE, Vol. 132, No. 3, pp. 268–278. generation benchmark control problem for seismically excited
Dyke, S.J., Spencer, Jr. B.F., Sain, M.K. and Carlson, J.D. (1996). buildings”, Proceeding of 2nd World Conference on Structural
“Modeling and control of magnetorheological dampers for Control, Vol. 2, pp.1351–1360, John Wiley & Sons, NY.

Advances in Structural Engineering Vol. 10 No. 3 2007 303


ASE 10-3-6 Tse 12/6/07 8:21 am Page 304

Vibration Control of a Wind-Excited Benchmark Tall Building with Complex Lateral-Torsional Modes of Vibration

Spencer, Jr., B.F., Dyke, S.J., Sain, M.K. and Carlson, J.D. (1997). Thesis, The University of Notre Dame, Indiana, US.
“Phenomenological model for magnetorheological dampers”, Yang, G., Spencer Jr., B.F., Carlson, J.D. and Sain, M.K. (2002).
Journal of Structural Engineering, ASCE, Vol. 123, No.3, pp. “Large-scale MR fluid dampers: modelling and dynamic
230–238. performance considerations”, Engineering Structures, Vol. 24,
Spencer, Jr., B.F. and Sain, M.K. (1997). “Controlling buildings: a new No. 3, pp. 309–323.
frontier in feedback”, Special Issue of the IEEE Control Systems Yang, J.N., Agrawal, A.K., Samali, B. and Wu, J.C. (2001). “A
Magazine on Emerging Technology, Vol. 17, No. 6, pp. 19–35. benchmark problem for response control of wind-excited tall
Spencer, Jr., B.F., (2002). “Civil engineering applications of smart buildings”, Proceeding of the International Modal Analysis
damping technology”, Proceeding of 5th International Conference Conference - IMAC., Vol. 1, pp. 151–157.
on Vibration Engineering, Nanjing, China, pp. 771–782. Yang, J.N., Wu, J.C., Agrawal A.K. and Hsu, S.Y. (1997). “Sliding
Structural Use of Concrete: Part 2. Code of Practice for Design and mode control with compensators for wind and seismic response
Construction, BS8110–1997. control”, Journal of Earthquake Engineering and Structural
Tang, Y. and Gu, M. (2006). “Analysis on control of wind induced Dynamics, Vol. 26, pp. 1137–1156.
vibration of a super-tall building with TMD”, Journal of Yang, J.N., Wu, J.C. Samali, B. and Agrawal, A.K. (1998). “A
Vibration and Shock, Vol. 25, No. 2, pp. 16–19. benchmark problem for response control of wind-excited tall
Tse, T. (2002). Design and modelling of magnetorheological buildings”, Proceedings of the 2nd World Conference on
damper, MPhil Thesis, The Hong Kong University of Science and Structural Control, John Wiley & Sons, NY, pp. 1407–1416.
Technology, HK. Yoshida, O. and Dyke, S.J. (2003). “Experimental verification of
Tse, T. and Chang, C.C. (2004). “Shear-mode rotary magneto- torsional response control of asymmetric buildings using MR
rheological damper for small-scale structural control experiments”, dampers”, Journal of Earthquake Engineering and Structural
Journal of Structural Engineering, ASCE, Vol. 130, pp. 904–911. Dynamics, Vol. 32, No. 13, pp. 2085–2105.
Wu, J.C. and Yang, J.N. (2004). “Modified sliding mode control for Yoshida, O. and Dyke, S.J. (2005). “Response control of full-scale
wind-excited benchmark problem”, Journal of Engineering irregular buildings using magnetorheological dampers”,
Mechanics, ASCE, Vol. 130, No. 4, pp. 499–504. Journal of Structural Engineering, ASCE, Vol.131, No. 5, pp.
Wu, J.C., Yang, J.N. and Schmitendorf, W.E. (1998). “Reduced- 734–742.
order H and LQR control for wind-excited tall buildings”, Wang, J.Y., Ni, Y.Q., Ko, J.M. and Spendcer Jr., B.F. (2005).
Engineering Structures, Vol. 20, No. 3, pp. 222–236. “Magneto-rheological tuned liquid column dampers (MR-
Yalla, S.K and Kareem, A. (2000). “Optimum absorber parameters TLCDs) for vibration mitigation of tall buildings: modelling and
for tuned liquid column dampers”, Journal of Structural analysis of open-loop control”, Computers and Structures, Vol. 83,
Engineering, ASCE, Vol. 126, No. 8, pp. 906–915. No. 25–26, pp. 2023–2034.
Yang, G.Q. (2001). Large-scale Magnetorheological Fluid Damper
for Vibration Mitigation: Modelling, Testing and Control, PhD

304 Advances in Structural Engineering Vol. 10 No. 3 2007

You might also like