Professional Documents
Culture Documents
Control Force Characteristics of Different Control Strategies For The Wind-Excited 76-Story Benchmark Building Structure
Control Force Characteristics of Different Control Strategies For The Wind-Excited 76-Story Benchmark Building Structure
Control Force Characteristics of Different Control Strategies For The Wind-Excited 76-Story Benchmark Building Structure
Chunwei Zhang*
Institute for Infrastructure Engineering, University of Western Sydney, NSW 2751, Australia
Key words: benchmark, wind-induced vibration control, energy index, phenomena behavior of control force,
intrinsic mechanism of control force, structural adjacent reaction wall control strategy, performance evaluation.
infrastructure, where the controller is composed of a practical structures. Here the LCVA adopted is
compensation and a convergent part. The compensation composed of four identical columns of water. Different
part uses the structural response measurement and the configurations in the LCVA, e.g. without additional
disturbance measurement resulting in a feedback-feed damping enhancing mechanisms or with orifice plates
forward control loop which is efficient for reducing the have been considered for a direct comparison.
wind-induced vibrations. Lus et al. (2004) presented an Additionally, the robustness issue, the sensitivity of
approach outlining the system identification and LCVAs due to mistuning has also been investigated,
damage detection algorithm for the linear Benchmark and the performance of LCVA compared against the
76-floor structure, where a state-space model using the tuned mass damper (TMD) control strategy. The
Observer/Kalman filter identification algorithm and comparable results show that the LCVA is more
the second-order dynamic model parameters from the attractive due to lower cost in terms of effectiveness in
realized state space model were both investigated. It the suppression of wind-induced motion of the 76-story
attributed the system model from a view point of health benchmark building as well as its robustness to
monitoring approach which is critical when engaging mistuning of traditional TMD.
precise state-space based system model for carrying out Varadarajan et al. (2004) proposed the novel semi-
active control analysis. Mei et al. (2004) proposed active variable stiffness-tuned mass damper (SAIVS-
the Model Predictive Control (MPC) strategy for TMD) and continuously retuning its frequency due to
the Benchmark problem where the MPC is based on the real time control thus it is robust to changes in building
minimization of the difference between the predicted stiffness and damping. The control strategy incorporates
and desired response trajectories which is also subjected a Hilbert transform instantaneous frequency algorithm,
to input/output hard constraints prescribed constraints, and the results show that the effectiveness is comparable
and the results demonstrated the effectiveness and and robust, and is achieved at an order of magnitude less
robustness where the building with minus/positive 15% power consumption than pure active control. Wu et al.
stiffness uncertainty was considered. Peng et al. (2004) (2004) also proposed the modified sliding mode control
introduced a sinusoidal reference control strategy which (MSMC) strategy using dynamic output feedback which
is able to realize adaptive feed-forward vibration incorporates a pre-filter to modulate the control force
control, where the recursive-least-squares algorithm is together with a Kalman–Bucy filter based observer
used and a higher frequency sinusoidal signal is adopted using limited acceleration measurements taken only at
as the reference signal. Both the numerical and strategic locations. The simulation results demonstrate
experimental results showed that the proposed strategy the control effectiveness of structural vibration as well
can reduce structural vibration and achieve adaptability as its robustness. Yang et al. (2004a) proposed two
in real-time with regard to dynamic uncertainties and multi-objective control strategies for the Benchmark
modeling errors. Pham et al. (2004) investigated problem, the energy-to-peak based controller and the
traditional Linear Quadratic Gaussian (LQG) control peak-to-peak based controller, which were used to
approach where a reduced order of system and a minimize the sum of weighted peak responses with the
balanced controller was designed to achieve minimizing constraints or penalties on the peak values of control
control cost on the order of one-third less than the resources. Both the state feedback and dynamic output
standard LQG solution together with guaranteed feedback controllers are compared, and the simulation
stability performance. However, a higher target results illustrate that the proposed control strategies are
performance may require a significantly larger control advantageous compared with the linear quadratic
effort to be exerted. Gaussian controller.
Samali et al. (2004b) established the framework of In addition to the aforementioned researches
the active tuned mass damper (ATMD) incorporated regarding the wind induced vibration control Benchmark
with the fuzzy logic controller, where the inherent problem, Active Mass Driver/Damper (AMD) control
robustness and ability to handle nonlinear behavior of has also received intensive investigations by researchers
structure without a mathematical model has been within structural control area. Since Yao (1972)
presented, and the results showed that the performance proposed the concept of structural active control, the
of such a control strategy dealing with uncertainty in AMD control, recognized by superior control
stiffness which is similar and somewhat advantages effectiveness against control cost, has taken the lead in
over LQG controller. Samali et al. (2004c) also various structural control aspects and become one of the
investigated using liquid column vibration absorbers most extensively researched and applied technique in
(LCVAs) for the Benchmark problem based on previous practical applications (e.g. Soong 1990; Mita et al. 1992;
successful implementations of LCVAs in similar Housner et al. 1997; Spencer
et al. 1997, 2003; Ou 2003; Zhang et al. 2010b). Several be essentially damping force, suggesting that the active
important journals in civil engineering field, such as actuators can be replaced by semi-active devices or even
Journal of Engineering Mechanics ASCE (Issue 4, passive viscous damping devices based on a proper
2004), Journal of Structural Engineering ASCE (Issue 7 design. However, the phenomena behavior of active
2003), Earthquake Engineering and Structural Dynamics force of AMD system was found to be case dependent,
(Issue 11 in 1998; Issue 11 in 2001), reported the-state- which means that the actuator of an AMD system cannot
of-the-art of research and engineering applications of be simply replaced by semi-active or passive device
semi-active control and active control, particularly AMD (Zhang et al. 2010a). The intrinsic mechanism of AMD
control, including this IASCM Benchmark control control force need to be investigated thoroughly,
problem. Spencer and Nagarajaiah have made a particularly on the basis of a representative/typical civil
systematical overview of the applications of active structure background, the IASCM Benchmark model is
control in civil engineering (Spencer and Nagarajaiah shown to be most appropriate.
2003). To date, more than 50 high-rising buildings, Based on the IASCM Benchmark control problem,
including television towers and nearly 15 large-scale the relative advantages and restrictions within each
bridge towers have been equipped with AMD control control strategy will be discussed and quantitatively
systems for reducing wind-induced vibrations or evaluated. To facilitate such comparisons, appropriate
earthquake-induced vibrations of the structure. modifications have been incorporated into the standard
Although AMD control has achieved success, there Benchmark solution to make the problem compatible as
are still unresolved problems and issues concerning a test-bed for different control strategies, which will be
incorporating semi-active devices into active systems referred to as AMD control hereafter. Furthermore, this
(Horvat et al. 1983; Pinkaew et al. 2001; Ou 2003; paper explores phenomena behavior as well as intrinsic
Ricciardelli et al. 2003; Zhang et al. 2010b). On one mechanism of active control forces of selected control
hand, understanding the differences existing in control strategies, including AMD/AMD-2, STI and STA
strategies is essential for design and to achieve best control strategies.
performance in terms of by optimizing control
effectiveness/efficiency and control cost. On the other 2. THE REFERENCE BENCHMARK
hand, understanding the phenomena behavior as well as SOLUTIONS
intrinsic mechanism of active control force Yang et al. (2004b) proposed the reference control
corresponding to each control strategy realized by strategy, where the inertia mass of Hybrid Mass Damper
different control systems will contribute to identifying HMD is 500 ton, i.e. nearly 0.33% of the total structural
the best suitable systems for practical applications. Ou weight. A standard MATLAB program has also been
and Li (2010) showed that linear quadratic regulator developed for running simulations. Under the input of
(LQR) based active control force can be decomposed first 900 seconds of wind load acquired through wind
into an elastic restoring force component and a damping tunnel tests (Samali et al. 2004a; Yang et al. 2004b), the
force component. According to the analysis of the standard solution of control facts, cost and
proportions of the elastic restoring force and the damping effectiveness, for the Benchmark problem can be
force that make up the total active control force, two sets acquired based on the simulation program. The standard
of indices were developed to quantify the damping analysis is conducted on a reduced order of the
characteristics and the negative stiffness characteristics, structural model. However, in the current paper, the
respectively. The latter issue has also been intensively non-reduced original structural model with 76 DOFs is
discussed by Iemura et al. (2005). These indices can be found to be more appropriate for the purposes of control
used to quantify capability of a semi-active damping strategies comparison, which is developed based on a
system and a passive damping system to achieve the thorough comparison on the impact of order reduction
performance of a full active control system. Numerical of the original Benchmark problem (Zhang 2010a). In
results also indicate that negative stiffness characteristics order to exclude any potential impact attributable to
of an active control force exist in an active control control algorithms and weight parameters, the AMD
system, which can be successfully realized by semi- control strategy is also examined based on three
active magnetorheological (MR) damping systems independent algorithms: LQR, output feedback optimal
demonstrated through a field test of a stay cable of the control (named as LQRY), and LQG control. As a
Binzhou Yellow River Highway Bridge. Concurrently, result, figure 1 shows the peak response at the top
the structural interbedded active control strategy for the structural floor under different control algorithms,
IASCM Benchmark problem has also been studied, where all weighting parameters within the Q matrix are
where the behavior of active control force were found to set to be unity, except parameters corresponding to the
32 65
LQR LQR
LQRY LQRY
Displacement reduction (%)
28 55
26 50
24 45
22 40
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
1.2 400
LQR LQR
LQRY LQRY
LQG LQG
1 300
0.8 200
0.6 100
0.4 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
Figure 1. Parametric impact of AMD control based on the non-reduced structural model
state of inertia mass are assigned to be non-zero small AMD control strategy based on the non-reduced order
numbers. The acquired parametric impact results agree structural model are summarized in Table 1, which are
well with the standard Benchmark solution as well as found to have covered the known Benchmark solutions.
other proposed Benchmark solutions based on Although this reference Benchmark solution is straight
stochastic controllers (Yang et al. 2004b; Mei et al. forward, it establishes the link for further comparison
2004; Pham et al. 2004; Samali et al. 2004b). with reference to the known solutions.
Selected quantitative results of the above reference
Table 1. Results of two AMD control strategy for the wind-excited Benchmark problem
x ẍ x ẍ Control Mass
Items force (kN) stroke (cm)
Solution based on the non-reduced structural model 24~30 42~58 36~44 47~60 50~270 55~90
Benchmark standard solution (Yang et al., 2004b) 28.3 49 42.2 49.7 118 73.3
MPC solution (Mei et al., 2004) 28.4~28.8 49~70.3 – – 118 73.7~77.5
(RMS: Root Mean Square value. x is the displacement and ẍ is the acceleration.)
Table 2. Parameters of LQR control algorithm for AMD and STI control
(Note: Φ1 is a vector corresponding to the first column of the structural flexibility matrix, which has been introduced to adjust control objective to achieve
uniform performance throughout such a high-rise building structure; [I]n×n is a unit matrix with dimensions of n×n.)
80 80
60 60
No. of Floors
No. of Floors
40 40
20 Uncontrolled 20 Uncontrolled
AMD control AMD control
STI control STI control
0 0
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35
2
Peak displacement (m) Peak acceleration (m/s )
80 80
70
60 60
No. of Floors
No. of Floors
50
40 40
30
20 Uncontrolled
20 Uncontrolled
AMD control AMD control
10
STI control STI control
0 0
0 0.02 0.04 0.06 0.08 0.1 0.12 0 0.02 0.04 0.06 0.08 0.1
200 80
Peak
100 RMS
70
Force (kN)
−100
60
−200
AMD
−300 50
No. of Floors
200
30
100
Force (kN)
0 20
−100
10
−200
STI-60th floor
−300 0
0 200 400 600 800 1000 100 200 300 400 500 600
Time (s) STI control force (kN)
Figure 3. Comparison of active control force by AMD and STI control strategies
by STI control is 150 times bigger than that of AMD quits effect between each adjacently placed actuator. In
control. Based on the phenomena observation, the terms of the entire structure, inter-story installed
primary reason can be attributed to the reaction and actuator will exert active control force to both ends of
Table 3. Comparison of control results between AMD and STI control strategy
Control results Peak value RMS value Peak value RMS value Peak force (kN) Quantity
AMD control 30.0 43.3 58.7 59.9 265 1
STI control (this paper) 33.2 46.5 61.3 62.7 372~527 76
STI control (Ou, 2003) 33.5 – 46.8 – 110~1500 20
the actuator and transfer to the structure with two forces Where i is the number of floor where the actuator is
with the same magnitude but different directions. For installed, ui is the control force of the corresponding
example, if the actuator is installed between the i-th and actuator, x·i is the interbedded velocity, T is the time
the (i+1)-th floor of the structure and assuming the duration, i.e. the total length of time for analysis; sgn (·) and
active force at the bottom end being pointing to the left H(·) are sign function and unit step function, respectively.
direction, then the active force at the top end will be to
the right direction. Therefore, it is not cost-effective to 1 x ≥ 0 1 x≥0
engage pure active actuators into inter-story for sgn( x ) = ; H (x) = (3)
vibration control of tall building structures, although the −1 x < 0 0 x<0
actuator stroke is relatively small compared with AMD
Eqn 1 defines a virtual time history of the active
actuator stroke. The following section will develop a
control force, where positive values of u*i (tj) stand for
further analysis on the intrinsic mechanism of active
the force being opposite to the actuator velocity x·i (tj) at
force corresponding to STI control.
time step tj, and vice versa. Index γ 1 defines the
proportion of active force being opposite to the relative
3.2.1. Evaluation indices
velocity. Figure 4 shows the time history of virtual force
Ou and Li (2010) proposed an index denoting the
u*i (tj) at 60th floor corresponding to STI control
direction relation between control force and velocity for
strategy.
cable damper. Furthermore, another three sets of indices
It can be seen from Figure 4 that a dominant
are defined in this paper to facilitate further analysis of
percentage of active control force is opposite to its
the relation between active control force and the
velocity by STI control. In comparison, Figure 5 gives
actuator state variables.
the time history of AMD control force based on a
(i.) Index of direction relation between active force
similar process, whereby the direction feature is no
and relative velocity
longer obvious. As a result, although achieving similar
control effectiveness, the characteristics of control
u*i (t ) = − sgn( x&i ui ) ui (t ) (1) forces by the two control strategies are completely
different from each other.
(ii.) Index of direction relation between active force
T and velocity-displacement
∫ H (− x&iui )u i (t )dt
*
γ 1,i = 0 T
(2)
u**i (t ) = −H ( xi x&i )sgn( x&i ui ) ui (t ) (4)
∫0 ui (t ) dt
500
Virtual STI control - 60th floor
400
Force (kN)
300
200
100
0
−100
0 100 200 300 400 500 600 700 800 900
Time (s)
Figure 4. Time history of virtual control force at 60th floor by STI control strategy
200
Virtual AMD control
100
Force (kN) 0
−100
−200
−300
0 100 200 300 400 500 600 700 800 900
Time (s)
T 80
∫ H (− x&iui )u i (t )dt
**
γ1
γ 2 ,i = 0 T
(5) γ2
60
∫0 ui (t ) dt γopp
No. of Floors
40
where xi is the interbedded displacement of i-th
structural floor, and the other symbols have the similar
meanings with Eqn 1. 20
Eqn 4 defines a virtual time history of active control
force, where positive values of u**i (tj) stand for the force 0
being opposite to the actuator velocity x·i (tj) during 0 20 40 60 80 100
Indices (%)
which the actuator moving against its equilibrium
position. Therefore, index γ 2 defines the proportion of Figure 6. Characteristic indices of active control force by STI
active force being opposite to its relative velocity during control strategy
a particular velocity-displacement period.
(iii.) Index of vibration characteristics of actuator
80
Damping
T
∫ H (− xi x&i )dt
γ opp,i = 0 T
60
Stiffness
(6)
∫0 1dt
Floor
40
Figure 8 for STI control and in Figure 9 for AMD where Eopposite stands for the energy of active control
control, respectively. Again, significant difference force when it is opposite to its velocity, and Etotal stands
can be seen between the two control strategies, where for the total energy consumed by active control.
the STI control force is shown to be damping force Figure 10 shows the energy indices distributing
behavior, but the AMD control force is irregular in throughout the entire structure for STI control strategy,
terms of velocity or displacement relevant behavior. where the maximum value of γ Eopp was found to be
99.9977% with a minimum value of 99.9492%, and the
3.2.3. Energy characteristics of active control average value is 99.987%. Therefore, it can be
force concluded that nearly 100% control energy is allocated
In section 3.2.1, index γ 1 gives the directional where control force is opposite to its velocity. Based on
characteristic of active control force, which also equals the above analysis, the active control force of STI
to the area ratio of positive virtual control forces as control strategy is shown to be essential damping force.
shown in Figure 4. However, this index only considers
the force magnitudes, which does not take the 4. STRUCTURAL ADJACENT REACTION
corresponding velocity magnitudes into consideration. WALL CONTROL STRATEGY
Since energy/power is a multiplication product of force According to the previous phenomena behavior
and displacement/velocity, i.e. W = Fx or P = Fx·, the analysis, the reason for high control cost required by
index describing energy characteristics of control force STI control strategy is attributable to the active reaction
can be defined as and quits effect. It is impossible to physically eliminate
this disadvantage. However, for the purpose of
T numerical analysis, an imaginary adjacent wall with
∫ H[− x&i (t ) ⋅ ui (t )] ⋅ x&i (t ) ⋅ ui (t ) dt = Eopposite
γ Eopp = 0 infinite stiffness is introduced and placed next to the
T
(7)
Etotal structure to provide the support for reaction force for
∫0 x&i (t ) ⋅ ui (t ) dt each actuator, thus the quits effect can be numerically
400 400
200 200
STI control force (kN)
0 0
−200 −200
−400 −400
−600 −600
−0.4 −0.3 −0.2 −0.1 0 0.1 0.2 0.3 −0.4 −0.3 −0.2 −0.1 0 0.1 0.2 0.3
Inter-drift (cm) Inter-velocity (cm/s)
200 200
AMD control force (kN)
100 100
0 0
−100 −100
−200 −200
−300 −300
−100 −50 0 50 100 −80 −60 −40 −20 0 20 40 60 80
Mass stroke (cm) Mass velocity (cm/s)
100
60
40
20
γEopp
0
0 10 20 30 40 50 60 70 80
No. of Floors
Figure 10. Energy indexes of active control force by STI control strategy
excluded/ignored. Such a strategy is called structural engaged in order to exclude any potential impact caused
adjacent reaction wall (STA) control strategy. Figure 11 by control algorithms and weight parameters. Table 4
illustrates the concept sketch of STA control. Although summarizes the weight parameters corresponding to each
the name sounds like adjacent-structures or coupled- control algorithm.
structures vibration control, they are completely Structural responses including peak and RMS
different from each other, and the imaginary wall here is displacement and acceleration are given in Figure 12,
stiff allowing no deformation. This STA strategy is where the control effectiveness is comparable with
designed to facilitate further investigation to reveal AMD control as given in Figure 2.
interactions within AMD subsystem components, e.g. Figure 13 shows the control force of each actuator
function of the inertia mass. corresponding to three control algorithms, where LQR
control is shown to be the lowest in terms of control
4.1. Comparison between STA Control and force requirement.
AMD Control Quantitative structural response and control force
Similar approach used in the analysis of STI control was requirement by three control strategies: AMD, STI and
utilized to compare STA control and AMD control. Three STA, are summarized in Table 5. Based on the results,
control algorithms: LQR, LQRY and LQG were again AMD control is shown to possess the highest efficiency
in terms of lower control force cost, i.e. magnitudes and
numbers of actuators. Whereas STA control requires the
Actuators
minimum magnitudes of control forces with respect to
each single actuator, which is in the order of one-tenth
of AMD control in terms of force magnitude
requirement.
Control LQRY
algorithms LQR and LQG
Q matrix [I ]152×152 [I ]228×228
Figure 11. Sketch of structural adjacent reaction wall (STA)
R matrix 3×10–10×[I]76×76×Φ1
control strategy
80 80
60 60
No. of Floors
No. of Floors
40 40
Uncontrolled Uncontrolled
20 STA - LQR 20 STA - LQR
STA - LQRY STA - LQRY
STA - LQG STA - LQG
0 0
0 0.1 0.2 0.3 0.4 0 0.1 0.2 0.3 0.4
Peak displacement (m) Peak acceleration (m/s2)
80 80
60 60
No. of Floors
No. of Floors
40 40
Uncontrolled Uncontrolled
20 STA - LQR 20 STA - LQR
STA - LQRY STA - LQRY
STA - LQG STA - LQG
0 0
0 0.05 0.1 0.15 0.2 0 0.02 0.04 0.06 0.08 0.1
Figure 12. Structural response of STA control under different control algorithms
80 80
60 60
No. of Floors
Floor
40 40
0 0
0 10 20 30 0 2 4 6 8
Peak control force (kN) RMS control force (kN)
Figure 13. Comparison of active control force under three algorithms by STA control strategy
Table 5. Comparison of control forces and structural response by three control strategies
Note: * This is the summation of forces by 76 actuators, where minimum and maximum force by a single actuator ranges from 372 kN to 527 kN. ** This is
the summation of forces by 76 actuators, where minimum and maximum force by a single actuator ranges from 4.6 kN to 21.5 kN.
expressed as
γ g2
60
γ opp
T
∫0 H(− ui y&i )sgn(ui y&i ) ui (t ) dt
No. of Floors
γ gEopp
γ g1,i = − T
(8)
40
∫0 ui (t ) dt
20
T
γ g 2 ,i = −
∫0 H(−ui y&i ) H( yi y&i )sgn(ui y&i ) ui (t ) dt (9)
T 0
∫0 ui (t ) dt 0 20 40 60 80 100
Indices (%)
γ opp,i =
∫0 H (− yi y&i )dt control strategy
T
(10)
∫0 1dt 80
previously.
In addition, the corresponding energy characteristic 40
T 20
∫0 H[− y&i (t ) ⋅ ui (t )] ⋅ y&i (t ) ⋅ ui (t ) dt Damping
g
γ Eopp = T
(11) Stiffness
∫0 y&i (t ) ⋅ ui (t ) dt 0
0 5 10 15 20 25
Figure 14 presents the STA indices corresponding to Peak control force (kN)
Eqns 8 to 11. Based on the results of γ 1g and γ Eoppg, the Figure 15. Decomposition of active control force by STA control
control force is nearly 100% opposite to its velocity. In strategy
addition, Figure 15 shows the results of decomposing
active control force into damping force and stiffness
force, where the damping force is again shown to be the velocity for STA control strategy. Based on the results,
dominant component. Figure 16 gives the hysteresis the force is obviously shown to be essential damping
loops between control force and displacement as well as force.
20 20
10 10
Control force (kN)
−10 −10
−20 −20
−20 −10 0 10 20 −10 −5 0 5 10
Displacement (cm) Velocity (cm/s)
Figure 16. Hysteresis loops of active control force by STA control strategy
5. NEW CONTROL STRATEGY FOR THE comparison, where only one actuator is assigned to the
BENCHMARK PROBLEM top floor of structure. In this case, STA control looks
The STA control is evaluated to require the minimum like AMD control, but still requires an imaginary fixed
magnitudes of control forces with damping force wall for supporting reaction force of the actuator. The
behavior. Whereas, AMD control requires pure active comparative case is the actuator only with inertia mass
control force which can either pull/drag or push the based AMD system, i.e. without damper and spring. To
inertia mass as determined by control algorithm. Both facilitate numerical comparison, only LQR control
AMD control and STA control require additional algorithm is considered with corresponding weight
components, i.e. a moving inertia mass or an imaginary parameters given in Table 6.
fixed mass, which do not belong to the original Figure 17 shows the similar phenomena behavior of
structural components. The similarities as well as AMD control force and STA control force, which
differences between these two control strategies need results in similar controlled structural response, but the
further discussion. motion of the reaction force supporting point is
As shown in Figures 12 and 13, significantly larger different. The AMD control force is generated by the
control forces are required by several lower structural actuator with the reaction force being supported by
floors corresponding to LQRY/LQG control cases. The the inertia mass, which results in the cyclic oscillation of
results indicate that control forces within those lower the mass subsystem on the top floor of the structure.
structural floors are less effective as compared with that Whereas, for STA control, the imaginary wall is
of upper floors in terms of suppressing global structural assumed to be infinity in stiffness, therefore the stroke
vibrations. Therefore, it is more efficient to assign of the actuator is equal to the relative displacement of
control devices into higher level floors as well as to the top structural floor, which is smaller than the mass
reduce actuator resources. Intensive optimal assignment stroke of AMD control system.
of actuator will not be discussed in the current paper, In order to further study the inertia mass ratio impact
however, one extreme case can be designed for further on characteristics of control force, the following
Table 6. Parameters of LQR control algorithm by AMD control and STA control strategies
300
STA control
200 AMD control
Control force (kN)
100
−100
−200
−300
0 100 200 300 400 500 600 700 800 900
Time (s)
Figure 17. Time history comparison of active control force between AMD control and STA control
analysis is conducted with the mass ratio ranging from simply replaced by any other non-active device, but the
0.1% to 10%, which covers the ranges of reasonable STA control strategy can be realized with AMD-2
mass ratio for passive/active/hybrid control. As shown configuration where an actuator can be replaced by a
in Figure 18, AMD-1 stands for the Benchmark active semi-active damper provided the inertia mass achieves
control system with damper and spring, whereas AMD-2 significant ratios with respect to the entire structural
stands for an actuator-mass system which excludes mass. For comparison purposes, the energy index of
damper and spring from the standard AMD control force corresponding to STA strategy is
configuration. The purpose of introducing terms of calculated to be 99.9993% according to Eqn 11, as
AMD-1 and AMD-2 is purely for the simplicity of shown in Figure 18 denoted with blue circle markers.
expression in the following figures and context. Based on the results, when the inertia mass ratio
According to the Benchmark problem definition, AMD-1 increases up to 2% and beyond, energy index indicates
is the Active Tuned Mass Damper (ATMD), AMD-2 is that the AMD-2 force is nearly 100% damping force
defined as a comparable control strategy which is which is exactly the same as STA control force, in
composed of an inertia mass and a controllable actuating terms of both wave forms as shown in Figure 17 and
device, such as an actuator or a semi-active damper. the intrinsic mechanism. Therefore, AMD-2 can be
Based on the quantitative energy index, the behavior of viewed as the simplest equivalent realization of STA
active control force corresponding to the AMD-1 (or control. Besides, based on the force mechanism, the
namely ATMD) strategy as given in Figure 9, is shown active actuator can also be completely replaced by a
to be changing toward the damping force characteristics semi-active controlled damping device, e.g. a
with respect to the increasing of ratio of inertia mass. magnetorheological fluid (MR) damper where the
However, significant difference exists between AMD-1 damping force can be regulated according to
and AMD-2 especially within lower mass ratio region. optimization procedure to trace active control force so
This concludes actuator of ATMD system can’t be as to fully realize active control performance (Ou et al.
2010). The AMD-2 can be viewed as a new control
strategy for the Benchmark problem derived from the
in-depth analysis of STA control strategy. It requires
100 only a mass and a semi-active damper configuration,
where the inertia mass ratio should be significant
enough. In the current example, the 2% inertia mass
80
ratio criteria are acquired based on the Benchmark
γE
40 70
Reduction (%) 35 60
Reduction (%)
30 50
25 40
20 30
15 AMD-1 20 AMD-1
AMD-2 AMD-2
10 10
0 2 4 6 8 10 0 2 4 6 8 10
Mass ratio (%) Mass ratio (%)
(a) Peak displacement at the top floor (b) Peak displacement at the top floor
0.6 250
AMD-1 AMD-1
AMD-2 AMD-2
0.5 200
0.4 150
0.3 100
0.2 50
0.1 0
0 2 4 6 8 10 0 2 4 6 8 10
Mass ratio (%) Mass ratio (%)
(c) Inertia mass strokes (d) Peak values of active control force
Figure 19. Impact of inertia mass ratio on control performance by AMD control strategies
As a further discussion, the inertia mass ratio impact Figure 18, the AMD-1 control force is shown to be
on active control performance, i.e. effectiveness against intrinsically different from that of AMD-2 control
control effort, for both the new control strategy, referred strategy. All these differences also reveal the impact of
to as AMD-2, and the reference Benchmark control damper and spring element within the standard
strategy, referred to as AMD-1, will be evaluated and configuration of AMD control system, where the
compared. ultimate control performance is determined by the
Based on the results given in Figure 19, there is an interactions among actuator, damper, spring and inertia
upper limit in control effect corresponding to AMD-2 mass.
strategy, which is 10%~30% less than AMD-1 strategy.
The control forces and mass strokes required by AMD- 6. CONCLUSIONS
2 are also higher than that of AMD-1 strategy. However, This research has been based on the rigorous model of
the most attractive aspect is that only one semi-active the IASCM Benchmark problem of wind-induced
controllable damper device is needed by AMD-2 vibration control of the 76-storey Melbourne building.
control, which makes the control system simple, low Two new control strategies: STA and Mass-semi-active-
cost, power-fail safe, reliable and energy dissipative damper (AMD-2), have been proposed and compared
rather than consumptive as required by pure active with Benchmark extended AMD control as well as STI
control. With regard to AMD-1 control, increasing control strategies. Phenomena behavior as well as
inertia mass ratio may contribute to achieving higher intrinsic mechanism of control forces has been
control performance. For example, there are different investigated. The main conclusions are summarized as
optimal mass ratios range according to different follows:
categories of criteria, which is 2~6% in terms of (1) A set of innovative evaluation indices, denoting
controlling structural displacement and acceleration the direction and energy relations between
response, 3~8% in terms of mass strokes and 4~10% in control force and its velocity and/or
terms of peak control forces, respectively. As given in displacement, have been proposed to study the
intrinsic mechanism of active force of each and future”, Journal of Engineering Mechanics, ASCE, Vol. 123,
control strategy. Based on the analysis results, No. 9, pp. 897–971.
the active control force of STI and STA and Iemura, H. and Pradono, M.H. (2005). “Simple algorithm for semi-
AMD-2 strategies were all shown to be essential active seismic response control of cable-stayed bridges”,
damping force. However, the mechanism of Earthquake Engineering and Structural Dynamics, Vol. 34, No.
AMD control force was shown to be versatile, 4–5, pp. 409–423.
which can either push or pull the inertia mass to Kim, S.B., Yun, C. and Spencer, B.F. (2004). “Vibration control of
exert active control energy into structures. wind-excited tall buildings using sliding mode fuzzy control”,
(2) Due to the quits effect, STI control was shown Journal of Engineering Mechanics, ASCE, Vol. 130, No. 4,
to be ineffective in terms of active energy pp. 505–510.
exertion. STA control requires minimum Lus, H., Betti, R., Yu, J. and De Angelis, M. (2004). “Investigation
magnitudes of control forces with respect to of a system identification methodology in the context of the
each single actuator. Further investigation also ASCE benchmark problem”, Journal of Engineering Mechanics,
revealed the homogenous property between ASCE, Vol. 130, No. 1, pp. 71–84.
STA and actuator-mass based AMD control Mei, G., Kareem, A. and Kantor, J.C. (2004). “Model predictive
strategies. The impact of inertia mass ratio has control of wind-excited building: Benchmark study”, Journal of
also been analyzed and evaluated with an Engineering Mechanics, ASCE, Vol. 130, No. 4, pp. 459–465.
emphasis on the intrinsic mechanism of active Mita, A. and Kaneko, M. (1992). “Hybrid versus tuned or active
force corresponding to standard-extended AMD mass dampers for response control of tall buildings”, The 1st
control strategy. International Conference on Motion and Vibration Control,
(3) A new control solution for the Benchmark Yokohama, Japan, pp. 304–309.
problem has been developed based on Ou, J.P. (2003). Structural Vibration Control: Active, Semi-Active
comprehensive analysis of STA control and Smart Control, Science Press, Beijing, China.
strategy, which is to engage a bigger inertia Ou, J.P. and Li, H. (2010). “Analysis of capability for semi-active or
mass with a semi-active controllable damper passive damping systems to achieve the performance of active
device. Based on the force mechanism and control systems”, Structural Control and Health Monitoring,
inertia mass ratio impact analysis, the results Vol. 17, No. 7, pp. 778–794.
indicated that the Mass-semi-active-damper Peng, F., Gu, M. and Niemann, H. (2004). “Study on sinusoidal
control system is capable of achieving reference strategy-based adaptive feedforward control applied to
comparative performance at lower cost with benchmark wind-excited building”, Journal of Engineering
other attractive advantages, compared with Mechanics, ASCE, Vol. 130, No. 4, pp. 518–523.
Benchmark solutions. Pham, K.D., Jin, G., Sain, M.K., Spencer Jr, B.F. and Liberty, S.R.
(2004). “Generalized linear quadratic Gaussian techniques for the
ACKNOWLEDGEMENTS wind benchmark problem”, Journal of Engineering Mechanics,
This research is partly supported by the University of ASCE, Vol. 130, No. 4, pp. 466–470.
Western Sydney Research Partnerships Program and the Pinkaew, T. and Fujino, Y. (2001). “Effectiveness of semi-active
Research Grant Scheme. tuned mass dampers under harmonic excitation”, Engineering
Structures, Vol. 23, No. 7, pp. 850–856.
REFERENCES Ricciardelli, F., Pizzimenti, A.D. and Mattei, M. (2003). “Passive
Ahlawat, A.S. and Ramaswamy, A. (2004). “Multiobjective optimal and active mass damper control of the response of tall buildings
fuzzy logic control system for response control of wind-excited to wind gustiness”, Engineering structures, Vol. 25, No. 9, pp.
tall buildings”, Journal of Engineering Mechanics, ASCE, Vol. 1199–1209.
130, No. 4, pp. 524–530. Samali, B., Kwok, K.C.S., Wood, G.S., and Yang, J.N. (2004a).
Battaini, M., Casciati, F. and Faravelli, L. (2004). “Controlling wind “Wind tunnel tests for wind-excited benchmark building”,
response through a fuzzy controller”, Journal of Engineering Journal of Engineering Mechanics, ASCE, Vol. 130, No. 4,
Mechanics, ASCE, Vol. 130, No. 4, pp. 486–491. pp. 447–450.
Horvat D., Barak P. and Rabin, M. (1983). “Semi-active versus Samali, B., Al-Dawod, M., Kwok, K.C.S. and Naghdy, F. (2004b).
passive or active tuned mass dampers for structural control”, “Active control of cross wind response of 76-story tall building
Journal of the Engineering Mechanics, ASCE, Vol. 109, No. 3, using a fuzzy controller”, Journal of Engineering Mechanics,
pp. 691–705. ASCE, Vol. 130, No. 4, pp. 492–498.
Housner, G.W., Bergman, L.A., Caughey, T.K., Chassiakos, A.G., Samali, B., Mayol, E., Kwok, K.C.S., Mack, A. and Hitchcock, P.
Claus, R.O., Masri, S.F., Skelton, R.E., Soong, T.T., Spencer, (2004c). “Vibration control of the wind-excited 76-story
B.F. and Yao, J.T.P. (1997). “Structural control: past, present, benchmark building by liquid column vibration absorbers”,
Journal of Engineering Mechanics, ASCE, Vol. 130, No. 4, Zhang, C.W. and Ou, J.P. (2010a). “Control force characteristics of
pp. 478–485. structural active mass driver systems”, Journal of Vibration
Soong, T.T. (1990). Active Structure Control Theory and Practice, Engineering, Vol. 23, No. 1, pp. 1–6.
Longman Scientific & Technical, New York, USA. Zhang, C.W. and Ou, J.P. (2010b). Mass Inertia Effect Based
Spencer Jr., B.F. and Sain, M.K. (1997). “Controlling buildings: a Vibration Control Systems for Civil Engineering Structure and
new frontier in feedback”, IEEE Control Systems Magazine, Infrastructure, Vibration Control (Mickael Lallart Ed.), Sciyo,
Vol. 17, No. 6, pp. 19–35. Rijeka, Croatia, pp. 105–158.
Spencer Jr., B.F. and Nagarajaiah, S. (2003). “State of the art of
structural control”, Journal of Structural Engineering, ASCE, NOTATION
Vol. 129, No. 7, pp. 845–856. u active control force by actuator
Varadarajan, N. and Nagarajaiah, S. (2004). “Wind response control xi, x·i inter-floor/storey or interbedded relative
of building with variable stiffness tuned mass damper using displacement, velocity
empirical mode decomposition/Hilbert transform”, Journal of T, tj time duration, time step
Engineering Mechanics, ASCE, Vol. 130, No. 4, pp. 451–458. u*i(tj),u**i(tj) virtual control force based on direction
Wu, J. and Yang, J.N. (2004). “Modified sliding mode control for modifications
wind-excited benchmark problem”, Journal of Engineering γ 1,i,γ 2,i,γ opp,i direction index of force with inter-
Mechanics, ASCE, Vol. 130, No. 4, pp. 499–504. floor/storey velocity and/or displacement
Yang, J.N., Lin, S. and Jabbar, F. (2004a). “Linear multiobjective W, P work/energy, power
control strategies for wind-excited buildings”, Journal of Eopposite energy of active control when it is
Engineering Mechanics, ASCE, Vol. 130, No. 4, pp. 471–476. opposite to its velocity
Yang, J.N., Wu, J.C., Samali, B. and Agrawal, A.K. (1998). “A Etotal total energy consumed by active control
g
benchmark problem for response control of wind-excited tall γ Eopp, γ Eopp energy characteristic index of control
buildings”, Proceedings of 2nd World Conference on Structural force
Control, San Francisco, USA, pp. 1408–1416. yi, y·i, displacement and velocity of the i-th
Yang, J.N., Agrawal, A.K., Samali, B. and Wu, J. (2004b). floor relative to the earth ground
g g g
“Benchmark problem for response control of wind-excited tall γ 1,i, γ 2,i, γ opp,i direction index of force with relative
buildings”, Journal of Engineering Mechanics, ASCE, Vol. 130, velocity and/or displacement
No. 4, pp. 437–446. R, Q control algorithm weight parameter or
Zhang, C.W. and Ou, J.P. (2008). “Control structure interaction of matrix
electromagnetic mass damper system for structural vibration Φ1 the first column vector of the structural
control”, Journal of Engineering Mechanics, ASCE, Vol. 134, flexibility matrix
No. 5, pp. 428–437. [ I ]n×n unit matrix with the dimension of n × n