스마트 수동제어 시스템
Simple Control Algorithms for MR Dampers and
Smart Passive Control System
조 상 원 (趙 常 元 SangWon Cho)
건설 및 환경공학과
Department of Civil and Environmental Engineering
한 국 과 학 기 술 원
Korea Advanced Institute of Science and Technology
2 0 0 4
박 사 학 위 논 문
Doctoral Thesis
MR 댐퍼를 위한 간편한 알고리즘과
스마트 수종제어 시스템
Simple Control Algorithms for MR Dampers and
Smart Passive Control System
Simple Control Algorithms for MR Dampers and
Smart Passive Control System
Advisor : Professor InWon Lee
by
SangWon Cho
Department of Civil and Environmental Engineering
Korea Advanced Institute of Science and Technology
A thesis submitted to the faculty of the Korea Advanced Institute of
Science and Technology in partial fulfillment of the requirements for the
degree of Doctor of Philosophy in the Department of Civil and
Environmental Engineering.
Daejeon, Korea
2003. 11. 27
Approved by
Professor InWon Lee
Major Advisor
MR 댐퍼를 위한 간편한 알고리즘과
스마트 수동제어 시스템
조 상 원
위 논문은 한국과학기술원 박사학위논문으로 학위논문
심사위원회에서 심사 통과하였음.
2003 년 11 월 27 일
심사위원장 이 인 원 (인)
심 사 위 원 윤 정 방 (인)
심 사 위 원 김 진 근 (인)
심 사 위 원 곽 효 경 (인)
심 사 위 원 장 평 훈 (인)
i
DCE
995353
조 상 원. SangWon Cho. Simple Control Algorithms for MR Dampers
and Smart Passive Control System. MR댐퍼를 위한 간편한 제어 알고
리즘과 스마트 수동제어 시스템. Department of Civil and
Environmental Engineering. 2004. 101p. Advisor: Professor InWon Lee.
Text in English.
ABSTRACT
This dissertation proposes simple and efficient control algorithms for seismically
excited structures using MR dampers and a smart passive system based on MR dampers.
Magnetorheological (MR) dampers are one of the most promising control devices for
civil engineering applications to earthquake hazard mitigation, because they have many
advantages such as small power requirement, reliability, and low price to manufacture.
A number of control algorithms have been adopted for semiactive systems
including the MR damper. In spite of good features of previous studies, some algorithms
have drawbacks such as poor performances or difficulties in designing the weighting
matrix of the controller. Thus, the control algorithm is required, which is simple to use
and efficient to give comparable or better performance over the previous algorithms.
As a simple and efficient control algorithm, a modal control scheme and a
maximum energy dissipation algorithm (MEDA) are implemented for the MR damper
based control system.
Modal control reshapes the motion of a structure by merely controlling a few
selected vibration modes. Hence, a modal control scheme is more convenient to design
the controller than other control algorithms. Although modal control has been
investigated for the several decades, its potential for a semiactive control, especially for
the MR damper, has not been exploited. Thus, in order to study the effectiveness for the
MR damper system, a modal control scheme is implemented to seismically excited
ii
structures. A Kalman filter is included in a control scheme to estimate modal states from
physical measurements by sensors. Three cases of the structural measurement are
considered as a feedback to verify the effect of each measurement; displacement,
velocity, and acceleration, respectively. Moreover, a lowpass filter is applied to eliminate
the spillover problem. In a numerical example, a sixstory building model with the MR
dampers on the bottom two floors is used to verify the proposed modal control scheme.
The El Centro earthquake is used to excite the system, and the reduction in the drifts,
accelerations, and relative displacements throughout the structure is examined. The
performance of the proposed modal control scheme is compared with that of other control
algorithms that were previously suggested.
The maximum energy dissipation algorithm represents one control class which
employs the Lyapunov’s direct approach to stability analysis in the design of a feedback
controller. However, their potential for civil engineering applications using semiactive
control, especially with MR dampers, has not yet been fully exploited. This paper
investigates the performance and the robustness of the maximum energy dissipation
algorithm for civil engineering structures using MR dampers. The numerical examples
contain the cablestayed bridge and the nonlinear building. Various earthquakes are used
to excite the systems. Through the series of numerical simulations, the performance is
compared with that of other control algorithms that are previously proposed: The
reduction in the drifts, accelerations, and relative displacements throughout the structure
are examined according to the evaluation criteria.
Meanwhile, to reduce the responses of the controlled structure by using MR
dampers, a control system including a power supply, controller, and sensors is needed.
However, it is not easy to apply the MR damperbased control system to largescale civil
structures, such as cablestayed bridges and highrise buildings, because of the difficulties
of building up and maintaining the control system.
iii
Thus, this dissertation proposes a ‘smart’ passive damper system. The smart
passive damper system is based on MR dampers. Of course, the MR damper is a
semiactive device that needs an external power source to change the damping
characteristics of the MR fluids. However, the smart passive damper system based on MR
dampers is not using an external power source, but selfpowered by an electromagnetic
induction (EMI) system that is attached to the MR damper. The EMI system consists of a
permanent magnet and a coil. According to the Faraday’s law of induction, the EMI
system changes the kinetic energy of the MR damper to the electric energy and then the
electric energy is used to vary the damping characteristics of the MR damper. Therefore,
it is easy to build up and maintain the proposed smart damper system that consists of the
MR damper and the EMI system, because it does not require any control system such as a
power supply, controller, and sensors. To verify the effectiveness of the proposed EMI
system, the performances are compared with those of the semiactive MR damper.
iv
TABLE OF CONTENTS
ABSTRACT .......................................................................................................................... i
TABLE OF CONTENTS...................................................................................................... iv
LIST OF TABLES ............................................................................................................... vi
LIST OF FIGURES............................................................................................................. vii
CHAPTER 1 INTRODUCTION........................................................................................ 1
1.1 Background......................................................................................................... 1
1.2 Literature Review ............................................................................................... 4
1.2.1 Magnetorheological (MR) Dampers......................................................... 4
1.2.2 Control Algorithms for MR Dampers....................................................... 7
1.3 MR Fluids and Dampers ................................................................................... 10
1.3.1 MR Fluids............................................................................................... 10
1.3.2 MR Fluids Dampers................................................................................ 12
1.4 Objectives and Scopes ...................................................................................... 15
1.5 Organization ..................................................................................................... 17
CHAPTER 2 MODAL CONTROL SCHEME................................................................ 18
2.1 Modal Control Scheme for MR Dampers......................................................... 18
2.1.1 Modal Control......................................................................................... 18
2.1.2 Design of Optimal Controller ................................................................. 21
2.1.3 Modal State Estimation .......................................................................... 23
2.1.4 Elimination of Observable Spillover ...................................................... 27
2.2 Numerical Example .......................................................................................... 29
2.3 Summary of Results.......................................................................................... 43
v
CHAPTER 3 MAXIMUM ENERGY DISSIPATION ALGORITHM............................ 44
3.1 Control System................................................................................................. 44
3.1.1 Control Devices ...................................................................................... 45
3.1.2 Maximum Energy Dissipation Algorithm for MR Damper.................... 47
3.2 Benchmark Problems........................................................................................ 49
3.2.1 Benchmark CableStayed Bridge ........................................................... 49
3.2.2 Nonlinear Benchmark Building.............................................................. 55
3.3 Numerical Examples......................................................................................... 59
3.3.1 Control Performance............................................................................... 59
3.3.2 Controller Robustness............................................................................. 64
3.4 Summary of Results.......................................................................................... 67
CHAPTER 4 SMART PASSIVE CONTROL SYSTEM................................................. 68
4.1 Electromagnetic Induction System for MR Damper ........................................ 68
4.2 Analytical Model and Design ........................................................................... 73
4.2.1 Analytical Model .................................................................................... 73
4.2.2 Design of the EMI System...................................................................... 76
4.3 Numerical Simulation Results .......................................................................... 80
4.4 Summary of Results.......................................................................................... 87
CHAPTER 5 CONCLUSIONS ........................................................................................ 88
SUMMARY (IN KOREAN) 90
REFERENCES 93
ACKNOWLEDGEMENTS
CURRICULUM VITAE
vi
LIST OF TABLES
1.1 Properties of MR and ER fluids.............................................................................. 11
2.1 Normalized controlled maximum responses
due to the scaled El Centro earthquake................................................................... 34
2.2 Normalized controlled maximum responses of the acceleration feedback
due to the scaled El Centro earthquake................................................................... 40
2.3 Normalized controlled maximum responses of the displacement feedback
due to the scaled El Centro earthquake................................................................... 41
2.4 Normalized controlled maximum responses of the velocity feedback
due to the scaled El Centro earthquake................................................................... 42
3.1 Parameters for MR damper model.......................................................................... 46
3.2 Comparisons of the evaluation criteria for benchmark cablestayed bridge........... 61
3.3 Comparisons of the evaluation criteria for the nonlinear benchmark building....... 62
3.4 Evaluation criteria of modified location and number of MR dampers ................... 63
3.5 Evaluation criteria for ±7% stiffness perturbed system
under El Centro earthquake .................................................................................... 65
3.6 Evaluation criteria for ±30% stiffness perturbed system........................................ 66
4.1 Normalized peak absolute accelerations and interstory drifts ............................... 85
4.2 Percent increment compared to the better clippedoptimal controller case ............ 86
vii
LIST OF FIGURES
1.1 Behavior of MR fluid in magnetic field.................................................................. 10
1.2 Schematic of the prototype 20ton largescale MR fluid damper ........................... 13
1.3 Smallscale SD1000 MR fluid damper.................................................................. 14
1.4 Bypass type 20ton MR fluid.................................................................................. 14
2.1 Schematic diagram of the MR damper implementation ......................................... 29
2.2 Frequency responses of the first floor for the uncontrolled structures
under the scaled El Centro earthquake.................................................................... 32
2.3 Frequency responses of the sixth floor for the uncontrolled structures
under the scaled El Centro earthquake.................................................................... 33
2.4 Variations of evaluation criteria with weighting parameters
for the acceleration feedback .................................................................................. 36
2.5 Variations of evaluation criteria with weighting parameters
for the displacement feedback ................................................................................ 37
2.6 Variations of evaluation criteria with weighting parameters
for the velocity feedback......................................................................................... 38
3.1 Mechanical model of the MR damper .................................................................... 45
3.2 Drawing of the Cape Girardeau Bridge .................................................................. 51
3.3 Cross section of bridge deck................................................................................... 51
3.4 Finite element model ............................................................................................. 54
3.5 Finite element model of the towers......................................................................... 54
3.6 Schematic of the 20story benchmark building ...................................................... 56
4.1 Schematic of a MR damperbased control system.................................................. 68
4.2 Schematic of a MR damper with the EMI system.................................................. 70
4.3 Schematic of a MR damper with the EMI system implementation ........................ 70
4.4 Schematic of a MR damper implementation .......................................................... 73
4.5 Simple mechanical model of the normal MR damper ............................................ 75
viii
4.6 Design of EMI system with S
a
under three earthquakes......................................... 77
4.7 Design of EMI system with S
i
under three earthquakes ......................................... 77
4.8 Design of the clippedoptimal controller with S
a
under three earthquakes............. 79
4.9 Design of the clippedoptimal controller with S
i
under three earthquakes ............. 79
4.10 Velocities and induced voltages under various earthquakes................................... 81
4.11 Normalized peak acceleration and interstory drift ................................................ 83
Chapter 1 Introduction
1
CHAPTER 1
INTRODUCTION
1.1 Background
The tragic consequences of the recent earthquakes have underscored, in terms of
both human and economic factors, the tremendous importance of the way in which
buildings and bridges respond to earthquakes. In recent years, considerable attention has
been paid to research and development of structural control systems. Supplemental
passive, active, hybrid, and semiactive damping strategies offer attractive means to
protect structures against natural hazards. Passive supplemental damping strategies,
including base isolation systems, viscoelastic dampers, and tuned mass dampers, are
widely accepted by the engineering community as a means for mitigating the effects of
dynamic loading on structures. However, these passivedevice methods are unable to
adapt to structural changes, varying usage patterns, and loading conditions.
For more than two decades, researchers have investigated the possibility of using
active, hybrid, and semiactive control methods to improve upon passive approaches to
reduce structural responses (Soong 1990; Soong and Reinhorn 1993; Spencer and Sain
1997; Housner et al. 1997; Kobori et al. 1998, 2003; Soong and Spencer 2002; Spencer
2002). The first fullscale application of active control to a building was accomplished by
the Kajima Corporation on 1989 (Kobori et al. 1991). The Kyobashi Center building is an
11story (33.1m) building in Tokyo, having a total floor area of 423m
2
. A control system
was installed, consisting of two AMDs – the primary AMD is used for transverse motion
and has a mass of 4 t, while the secondary AMD has a mass of 1 t and is employed to
reduce torsional motion. The role of the active system is to reduce building vibration
under strong winds and moderate earthquake excitations and consequently to increase
comfort of occupants of the building.
Chapter 1 Introduction
2
Hybridcontrol strategies have been investigated by many researchers to exploit
their potential to increase the overall reliability and efficiency of the controlled structure
(Housner et al. 1994; Kareem et al. 1999; Nishitani and Inoue 2001; Yang and Dyke
2003; Faravelli and Spencer 2003). A hybrid control system is typically defined as one
that employs a combination of passive and active devices. Because multiple control
devices are operation, hybrid control systems can alleviate some of the restrictions and
limitations that exist when each system is acting alone, Thus, higher levels of
performance may be achievable. Additionally, the resulting hybrid control system can be
more reliable than a fully active system, although it is also often somewhat more
complicated. To date, there have been over 40 buildings and about 10 bridges (during
erection) that have employed feedback control strategies in fullscale implementations
(Spencer and Nagarajaiah 2003).
Although nearly a decade has passed since construction of the Kobashi Seiwa
building, a number of serious challenges remain to be resolved before feedback control
technology can gain general acceptance by the engineering and construction professions
at large. These challenges include: (i) reduction of capital cost and maintenance, (ii)
eliminating reliance on external power, (iii) increasing system reliability and robustness,
and (iv) gaining acceptance of nontraditional technology by the profession. Semiactive
control strategies appear to be particularly promising in addressing a number of these
challenges (Spencer 1996).
Control strategies based on semiactive control devices appear to combine the best
features of both passive and active control systems and to offer the greatest likelihood for
near term acceptance of control technology as a viable means of protecting civil
engineering structural systems against earthquake and wind loading. The attention
received in recent years can be attributed to the fact that semiactive control devices offer
the adaptability of active control devices without requiring the associated large power
sources. In fact, many can operate on battery power, which is critical during seismic
events when the main power source to the structure may fail. According to presently
Chapter 1 Introduction
3
accepted definitions, a semiactive control device is one that can not inject mechanical
energy into the controlled structural system (i.e., including the structure and the control
device), but has properties which can be controlled to optimally reduce the responses of
the system. Therefore, in contrast to active control devices, semiactive control devices do
not have the potential to destabilize (in the bounded input/bounded output sense) the
structural system. Previous studies indicate that appropriately implemented semiactive
systems perform significantly better than passive devices and have the potential to
achieve the majority of the performance of fully active systems, thus allowing for the
possibility of effective response reduction during a wide array of dynamic loading
conditions (Spencer and Sain 1997; Symans and Constantinou 1999; Spencer 2002).
Most of the semiactive control devices have employed some electrically controlled
valves or mechanisms. Such mechanical components can be problematic in terms of
reliability and maintenance. Another class of semiactive devices uses controllable fluids.
The advantage of controllable fluid devices is that they contain no moving parts other
than the piston, which makes them very reliable.
Two fluids that are viable contenders for development of controllable dampers are:
(1) electrorheological (ER) fluids; and (2) magnetorheological (MR) fluids. However,
recently developed MR fluids appear to be an attractive alternative to ER fluids for use in
controllable fluid dampers (Carlson 1994; Carlson and Weiss 1994; Carlson et al. 1995).
MR fluids are magnetic analogs of electrorheological fluids and typically consist of
microsized, magnetically polarizable particles dispersed in a carrier medium such as
mineral or silicone oil. When a magnetic field is applied to the fluid, particle chains form,
and the fluid becomes a semisolid and exhibits viscoplastic behavior similar to that of an
ER fluid. Carlson and Weiss (1994) indicated that the achievable yield stress of an MR
fluid is an order of magnitude greater than its ER counterpart. Moreover, MR fluids are
not sensitive to impurities such as are commonly encountered during manufacturing and
usage. Therefore MR dampers have, over the last several years, been recognized having a
number of attractive characteristics for use in structural vibration control applications.
Chapter 1 Introduction
4
1.2 Literature Review
1.2.1 Magnetorheological (MR) Dampers
Controllable fluid dampers generally utilize either electrorheological (ER) fluids or
magnetorheological (MR) fluids. These fluids are unique in their ability to reversibly
change from freeflowing, linear viscous fluids to semisolids with a controllableyield
strength in only a few milliseconds when exposed to an electric (ER fluids) or magnetic
field (MR fluids). These fluids can be modeled as Newtonian fluids in the absence of a
magnetic field. When a field is applied, the viscoplasticity model (Phillips 1969) may be
used to describe the fluid behavior.
Although the discovery of ER and MR fluids dates back to the 1940s, only recently
have they been applied to civil engineering applications. To date, a number of ER fluid
dampers have been investigated (Burton et al. 1996; Gavin et al. 1996a, 1996b; Kamath
et al. 1996; Makris et al. 1996) for structural vibration control applications in civil
engineering. Gavin et al. (1996a, 1996b) designed and tested an ER damper that consisted
of a rectangular container and a moving plunger comprised of nine rigidly connected flat
plates. Makris et al. (1996) developed an ER damper consisting of an outer cylinder and a
double20 ended piston rod that pushes the ER fluids through an annular duct.
Despite these advances in the development of ER fluid dampers, the development
of commercially feasible damping devices using these fluids is limited by several factors.
First, the fluids have a very limited yield stress; even the best ER fluids currently
available may only achieve stresses of 3.0 to 3.5 KPa. Also, common impurities that
might be introduced during manufacturing significantly reduce the capacity of the fluids.
Additionally, safety, availability and the cost of highvoltage (e.g. ~4000 volts) power
supplies required to control the ER fluids are further considerations. MR fluids, on the
other hand, have a 50 to 100 KPa maximum yield stress, are not affected by most
impurities, and are not sensitive to temperature. Moreover, MR fluids can be controlled
Chapter 1 Introduction
5
with a lowpower (e.g., less than 50 watts), lowvoltage (e.g., ~1224 volts), current
driven power supply with ~12 amps output. Therefore, MR fluids are particularly
promising for natural hazard mitigation and cost sensitive applications (Carlson and
Spencer 1996a, 1996b; Spencer and Sain 1997).
Different techniques have been developed to model the behavior of the controllable
fluid dampers. Basically, two types of models have been investigated: nonparametric and
parametric models. Ehrgott and Masri (1992) presented a nonparametric approach to
model a small ER damper that operates under shear mode by assuming that the damper
force could be written in terms of Chebychev polynomials. Gavin et al. (1996b) extended
this approach to model the ER damper. Chang and Roschke (1998) developed a neural
network model to emulate the dynamic behavior of MR dampers. However, the non
parametric damper models are quite complicated. Stanway et al. (1987) proposed a
simple mechanical model, the Bingham model, in which a Coulomb friction element is
placed in parallel with a dashpot. Gamoto and Filisko (1991) extended the Bingham
model and developed a visoelasticplastic model. The model consists of a Bingham
model in series with a standard model of a linear solid model. Kamath and Wereley
(1997), Makris et al. (1996), and Wereley et al. (1998) developed parametric models to
characterize ER and MR dampers. Dyke et al. (1996a,b), Spencer et al. (1997a) and Yang
et al. (2001a,b) presented the BoucWen model whose versatility was utilized to describe
a wide variety of hysteretic behavior.
A number of experimental studies have been conducted to evaluate the usefulness
of MR dampers for vibration reduction under wind and earthquakes. Dyke et al. (1996a,b,
1998), Jansen and Dyke (2000), Spencer et al. (1996b), and Yi and Dyke (2000) used MR
dampers to reduce the seismic vibration of building structure model. Spencer et al.
(2000), Ramallo et al. (2001) and Yoshioka et al. (2001) incorporated an MR damper
with a base isolation system such that the isolation system would be effective under both
strong and moderate earthquakes. Johnson et al. (2001a,b) employed the MR damper to
Chapter 1 Introduction
6
reduce windinduced stay cable vibration. The experimental results indicate that the MR
damper is quite effective for a wide class of applications.
Moreover, the technology has been demonstrated to be scalable to devices
sufficiently large for implementation in civil engineering structures. Carlson and Spencer
(1996b), Spencer et al. (1999), and Yang et al. (2002) have developed and tested a 20t
MR damper. Recently, Sodeyama et al. (2003) have also presented impressive results
regarding design and construction of largescale MR dampers. In 2001, the first fullscale
implementation of MR dampers for civil engineering application was achieved. The
NihonKagakuMiraikan, the Tokyo National Museum of Emerging Science and
Innovation has two 30tonMR fluid dampers installed between the third and fifth floors.
The dampers were built by Sanwa Tekki using the Lord Corporation MR fluid.
Retrofitted with staycable dampers, the Dongting Lake Bridge in Hunan, China
constitutes the first fullscale implementation of MR dampers for bridge structures. Long
steel cables, such as are used in cablestayed bridges and other structures, are prone to
vibration induced by the structure to which they are connected and by weather conditions,
particularly wind combined with rain, that may cause cable galloping. The extremely low
damping inherent in such cables, typically on the order of a fraction of a percent, is
insufficient to eliminate this vibration, causing reduced cable and connection life due to
fatigue and/or breakdown of corrosion protection. Two Lord SD1005 MR dampers are
mounted on each cable to mitigate cable vibration. A total of 312 MR dampers are
installed on 156 stayed cables. Recently, MR dampers have been chosen for
implementation on the Binzhou Yellow River Bridge in China to reduce cable vibration.
The installation is expected to be completed in October 2003 (Spencer and Nagarajaiah
2003).
Chapter 1 Introduction
7
1.2.2 Control Algorithms for MR Dampers
One challenge in the use of semiactive technology is in developing nonlinear
control algorithms that are appropriate for implementation in fullscale structures.
Numerous control algorithms have been adopted for semiactive systems. In one of the
first examinations of semiactive control, Karnopp et al. (1974) proposed a ‘‘skyhook’’
damper control algorithm for a vehicle suspension system and demonstrated that this
system offers improved performance over a passive system when applied to a single
degreeoffreedom system. Feng and Shinozukah (1990) developed a bangbang
controller for a hybrid controller on a bridge. More recently, a control strategy based on
Lyapunov stability theory has been proposed for electrorheological dampers (Brogan
1991; Leitmann 1994). The goal of this algorithm is to reduce the responses by
minimizing the rate of change of a Lyapunov function. McClamroch and Gavin (1995)
used a similar approach to develop a decentralized bangbang controller. This control
algorithm acts to minimize the total energy in the structure. A modulated homogeneous
friction algorithm (Inaudi 1997) was developed for a variable friction device. Clipped
optimal controllers have also been proposed and implemented for semiactive systems
(Sack et al. 1994; Sack and Patten 1994; Dyke, 1996a,b,c). The effective utilization of
multiple control devices is an important step in the examination of semiactive control
algorithms. A typical control system for a fullscale structure is expected to have control
devices distributed throughout a number of floors. Because of the inherent nonlinear
nature of these devices, one of the challenging aspects of utilizing this technology to
achieve high levels of performance is in the development of appropriate control
algorithms.
As previously mentioned, a number of control algorithms have been adopted for
semiactive control systems using MR dampers (Jansen and Dyke 2000). Among many
control algorithms, modal control represents one control class, in which the motion of a
structure is reshaped by merely controlling some selected vibration modes. Modal control
Chapter 1 Introduction
8
is especially desirable for the vibration control of civil engineering structure, which is
usually a large structural system, may involve hundred or even thousand degrees of
freedom, its vibration is usually dominated by the first few modes. Therefore, the motion
of the structure can be effectively suppressed by merely controlling these few modes
(Yang 1982). To date, numerous procedures and algorithms concerning modal control or
pole assignment have been proposed in literature. A modal control method using full state
feedback may not be practical for a structural system involving a large number of DOFs,
since the control implementation may requires a large amount of sensors. Thus a modal
control scheme, which uses modal state estimation, is desirable. To estimate the modal
states from the sensor output, Luenberger observer (Meirovitch 1990; Luenberger 1971)
and a KalmanBucy filter (Meirovitch, 1967) can be used for the case of low noiseto
signal ratios and for high noisetosignal ratios, respectively. The troublesome of
estimating the modal states for feedback in modal control is the problem of spillover.
Note, however, that a small amount of damping inherent in the structure is often
sufficient to overcome the observation spillover effect (Meirovitch and Baruh 1983). At
any rate, observation spillover can be eliminated if the sensor signals are prefiltered so as
to screen out the contribution of the uncontrolled modes.
On the other side, the maximum energy dissipation algorithm (MEDA) represents
one control class which employs the Lyapunov’s direct approach to stability analysis in
the design of a feedback controller (Brogan 1991). The approach requires the use of a
Lyapunov function that must be a positive definite function of the states of the system.
According to Lyapunov stability theory, if the rate of change of the Lyapunov function is
negative semidefinite, the origin is stable in the sense of Lyapunov. Thus, in developing
the control law based on Lyapunov stability theory, the goal is to choose control inputs
for each deice that will result in making the rate of change of the Lyapunov function as
negative as possible. Jansen and Dyke (2000) suggested MEDA as a variation of the
decentralized bangbang approach proposed by McClamroch and Gavin (1995). It is
Chapter 1 Introduction
9
noticeable that this control law requires only local measurements, which means MEDA is
simply implemented without any design process.
Chapter 1 Introduction
10
1.3 MR Fluids and Dampers
1.3.1 MR Fluids
The initial discovery and development of MR fluids can be credited to Jacob
Rabinow (1948, 1951) at the US National Bureau of Standards in the late 1940s. These
fluids are suspensions of micronsized, magnetizable particles in an appropriate carrier
liquid. Normally, MR fluids are free flowing liquids having a consistency similar to that
of motor oil. However, in the presence of an applied magnetic field, the iron particles
acquire a dipole moment aligned with the external field that causes particles to form
linear chains parallel to the field, as shown in Fig. 1.1. This phenomenon can solidify the
suspended iron particles and restrict the fluid movement. Consequently, yield strength is
developed within the fluid. The degree of change is related to the magnitude of the
applied magnetic field, and can occur only in a few milliseconds.
Figure 1.1 Behavior of MR fluid in magnetic field
There are basically two types of controllable fluids – MR fluids and ER fluids. The
primary advantage of MR fluids stems from their high dynamic yield strength due to the
high magnetic energy density that can be established in the fluid. Energy density in MR
fluids is limited by the magnetic saturation of iron particles. From a practical
implementation perspective, although the total energy requirements for the ER and MR
Chapter 1 Introduction
11
devices are almost equal, only MR devices can be easily driven by common lowvoltage
power sources (Carlson and Spencer 1996a). MR devices can be controlled with a low
voltage, currentdriven power supply outputting only ~12 amps. ER devices, on the other
hand, require a highvoltage power source (~20005000 volts) that may not be readily
available, especially during strong earthquake events. Moreover, such a high voltage may
pose a safety hazard. The properties of both MR and ER fluids are given in Table 1.1.
Table 1.1 Properties of MR and ER fluids (Spencer and Sain 1997)
Property MR Fluids ER Fluids
Max. yield Stress 50100 kPa 25kPa
Maximum field ~250kA/m ~4kV/mm
Plastic viscosity,
p
η
0.11.0Pas 0.11.0Pas
Operable temp. range 40 to 150°C +10 to 90°C
Stability
unaffected by most
impurities
cannot tolerate
impurities
Response time milliseconds milliseconds
Density 3 to 4 g/cm3 1 to 2 g/cm3
2
) (
/
field y p
τ η
10101011 s/Pa 107108 s/Pa
Maxi. energy density 0.1 Joules/cm3 0.001 Joules/cm3
Power supply (typical)
225V
12A
20005000V
110 mA
Chapter 1 Introduction
12
1.3.2 MR Fluid Dampers
The maximum force that an MR damper can deliver depends on the properties of
MR fluids, their flow pattern, and the size of the damper. Virtually all devices that use
MR fluids can be classified as operating in: (a) a valve mode, (b) a direct shear mode, (c)
a squeeze mode, or a combination of these modes (Carlson and Spencer 1996a). To date,
several MR fluid devices have been developed for commercial use by the LORD
Corporation (Carlson et al. 1996; Jolly et al. 1998). Linear MR fluid dampers have been
designed for use as secondary suspension elements in vehicles. MR fluid rotary brakes
are smoothacting, proportional brakes which are more compact and require substantially
less power than competing systems. MR fluid vibration dampers for realtime, active
control of damping have been used in numerous industrial applications.
In civil engineering applications, the expected damping forces and displacements
are rather large in magnitude. Therefore, MR dampers primarily operating under direct
shear mode or squeeze mode might be impractical. Usually valve mode or its
combination with direct shear mode is employed. Some examples of recently developed
MR dampers are given below. These dampers are capable of meeting realworld
requirements and are presently either in commercial production or in production
prototype trials.
A 20ton prototype largescale seismic MR fluid damper was developed under
cooperation between the LORD Corporation and the Structural Dynamics and
Control/Earthquake Engineering Laboratory (SDC/EEL) at the University of Notre Dame
(Carlson and Spencer 1996a; Spencer et al. 1997b,1998; Yang et al. 2000a,b). The MR
fluid damper schematic is given in Fig. 1.2. For the nominal design, a maximum damping
force of 200,000 N (20 tons) were chosen. The damper has an inside diameter of 20.3 cm
and a stroke of ±8 cm. The completed damper is approximately 1 m long, has a mass of
250 kg, and contains approximately 6 liters of MR fluid. However, the amount of fluid
energized by the magnetic field at any given instant is approximately 90 cm
3
.
Chapter 1 Introduction
13
Fig. 1.3 shows a smallscale SD1000 MR fluid damper manufactured by the
LORD Corporation (Carlson and Spencer 1996a; Dyke 1996a,b; Jolly et al. 1998;
Spencer 1997a). In this damper, MR fluids flow from a highpressure chamber to a low
pressure chamber through an orifice in the piston head. The damper is 21.5 cm long in its
extended position, and the main cylinder is 3.8 cm in diameter. Forces of up to 3,000 N
can be generated with this device.
Fig. 1.4 shows a bypasstype 20ton MR fluid damper designed by the Sanwa
Tekki Corporation (Fujitani et al. 2000; Sunakoda et al. 2000). Unlike dampers
mentioned previously, MR fluids in this damper flow from a highpressure chamber to a
lowpressure chamber in valve mode through a bypass outside the main cylinder. The
bypass has an annular gap between the outside of the magnetic pole and the inside of the
bypass cylinder. The magnetic field is generated by a 10stage electromagnet and is
perpendicular to the fluid flow.
Figure 1.2 Schematic of the prototype 20ton largescale MR fluid damper
Chapter 1 Introduction
14
Figure 1.3 Smallscale SD1000 MR fluid damper
Figure 1.4 Bypass type 20ton MR fluid
Bearing & Seal MR Fluid
Coil Diaphragm
Accumulator
Wires to
Electromagnet
Chapter 1 Introduction
15
1.4 Objectives and Scopes
The purpose of this study is to implement simple and efficient control algorithms
for seismically excited structures using MR dampers and to develop a smart passive
system based on the MR damper. The objectives and scopes of this study can be
summarized as follows.
First, the objectives and scopes of the study on implementations of simple and
efficient control algorithms can be summarized as follows:
(1) Implementation of modal control for seismically excited structures using MR
dampers:
In order to study the effectiveness for the MR damperbased semiactive, a modal
control scheme is implemented to seismically excited structures. A Kalman filter is
included in a control scheme to estimate modal states from measurements by
sensors. A lowpass filter is applied to eliminate the spillover problem. In a
numerical example, a sixstory building model with the MR dampers on the bottom
two floors is used to verify the implemented modal control scheme. The
performance of the proposed modal control scheme is compared with that of other
control algorithms previously studied.
(2) Implementation of maximum energy dissipation algorithm for seismic response
reduction of largescale structures using MR dampers:
The performance and the robustness of the maximum energy dissipation algorithm
for civil engineering structures using MR dampers are investigated. The numerical
examples contain the cablestayed bridge and the nonlinear building. Various
earthquakes are used to excite the system. Through the series of numerical
simulation, the performance and the robustness are compared with that of other
Chapter 1 Introduction
16
control algorithms that are previously proposed: The reduction in the drifts,
accelerations, and relative displacements throughout the structure are examined
according to the evaluation criteria.
Next, the objectives and scopes of the study on development of a smart passive
system based on the MR damper can be summarized as follows:
(1) Development of a smart passive system based on the MR damper to reduce
structural responses:
The smart passive damper system is based on MR dampers. The MR damper is a
semiactive device that needs an external power source to change the damping
characteristics of MR fluids. However, the smart passive damper system based on
MR dampers is not using an external power source, but selfpowered by an
electromagnetic induction system (EMI) that is attached to the MR damper. The
EMI system for MR dampers consists of a permanent magnet and a coil. According
to the Faraday’s law of induction, the EMI system changes the kinetic energy of
the MR damper to the electric energy and then the electric energy is used to vary
the damping characteristics of the MR damper. The theoretical backgrounds and
the designing process are presented. To verify the effectiveness of the proposed
smart passive control system, the performances are compared with those of the
semiactive MR damper using clippedoptimal controller.
Chapter 1 Introduction
17
1.5 Organization
This dissertation consists of four chapters. Chapter 1 discusses the background, the
literature review, the characteristics of MR fluids and dampers, and the objectives and
scopes of this study.
In Chapter 2, a modal control scheme is implemented for the MR damperbased
control system. A lowpass filter and the Kalman filter as a modal state estimator are
reviewed and included in the modal control scheme for the MR damperbased control
system in Section 2.1. Reduced design procedure is presented, also, in this section. To
evaluate the proposed modal control scheme for usage with the MR damper, a numerical
example is considered, in which a model of a sixstory building is controlled with four
MR dampers in Section 2.2. The results are summarized in Section 2.3.
In Chapter 3, the maximum energy dissipation algorithm (MEDA) is implemented
for the MR damperbased control system. The control system including the MR device
and MEDA is reviewed in Section 3.1. In Section 3.2, the cablestayed bridge and the 20
story nonlinear building are shown as representative structures of civil engineering and
numerical examples. In Section 3.3, the applicability of the MEDAbased semiactive
control system is examined from the viewpoint of the performance and the robustness
through the numerical examples. The results are summarized in Section 3.4.
In Chapter 4, a smart passive control system is proposed. In Section 4.1, an
electromagnetic induction (EMI) system is proposed for the MR damper. An analytical
model and a design procedure of the proposed EMI system are described in Section 4.2.
To show the effectiveness of the proposed smart passive control system, a set of
numerical simulations are performed for the four historical earthquakes in Section 4.3.
Section 4.4 summarizes the results.
Finally, the conclusions of this dissertation are summarized in Chapter 5.
Chapter 2 Modal Control Scheme
18
CHAPTER 2
MODAL CONTROL SCHEME
2.1 Modal Control Scheme for MR Dampers
In this section, a modal control scheme with a Kalman filter and a lowpass filter is
implemented to a seismically excited structure. A Kalman filter is included in a control
scheme to estimate modal states from various measurements. Moreover, a lowpass filter
is applied to eliminate the spillover problem. After the implementation of the modal
control scheme, numerical simulations are presented in subsequent sections for
comparisons between control algorithms.
2.1.1 Modal Control
Consider a seismically excited structure controlled with m MR dampers. Assuming
that the forces provided by the control devices are adequate to keep the response of the
primary structure from exiting the linear region, the equations of motion can be written
g
MΓ Λf Kx x C x M x t t t t & & & & & − = + + ) ( ) ( ) ( ) ( (2.1)
where M, C and K are the n×n mass, damping, and stiffness matrices, respectively; x is
the ndimensional vector of the relative displacements of the floors of the structure; f =
[ f
1
, f
2
,…, f
m
]
T
is the vector of measured control forces generated by m MR dampers;
g
x& & is
ground acceleration; Γ is the column vector of ones; and Λ is the matrix determined by
the placement of MR dampers in the structure. This equation can be written in the state
space form as
Chapter 2 Modal Control Scheme
19
g
x& & & N Gf Fz z + + = (2.2a)
v Mf Hz y + + = (2.2b)
where z is a state vector; y is a vector of measured outputs; and v is a measurement noise
vector. The displacement can be expressed as the linear combination
Φη x = =
∑
=
n
r
r r
t η t
1
) ( ) ( φ , r = 1, 2,…, n (2.3)
where ) (t η
r
is a r th modal displacement; φ
r
is a r th eigenvector; Φ is a eigenvector set;
and η is a modal displacement vector. The eigenvectors are orthogonal and can be
normalized so as to satisfy the orthonormality conditions
rs r
T
s
δ = φ φ M ,
rs r r
T
s
δ
2
ω φ φ = K , r = 1, 2,…, n (2.4)
where δ
rs
is the Kronecker delta and ω
r
is a natural frequency. Thus inserting (2.3) into
(2.1), multiplying by
T
r
φ and considering orthogonal condition between eigenvectors, we
obtain
g
T
r
T
r r
2
r r r r r
x φ φ η ω η ω 2ζ η & & & & & MΓ Λf − = + + , r = 1, 2,…, n (2.5)
where
r
ζ are modal damping ratios. (2.5) can be written in the matrix form as
g
x t t t t & & & & & ' ) ( ) ( ) ( ) ( E f B' η Ω η ∆ η
2
+ = + + (2.6)
where ∆ is the diagonal matrix listing 2ω
r
ζ
r
; Ω
2
is the diagonal matrix listing
2
1
ω ,…,
2
n
ω ; B′= Λ Φ
T
; and E′= MΓ Φ
T
− . (2.6) can be written in the modal spacestate form as
Chapter 2 Modal Control Scheme
20
g
x t t t & & & E Bf Aw w + + = ) ( ) ( ) ( (2.7a)
) ( ) ( t Cw t y = (2.7b)
where w(t) = [ η
T
η&
T
] is the modal state vector and
(
¸
(
¸
− −
=
∆ Ω
I 0
A
2
,
(
¸
(
¸
=
B'
0
B ,
(
¸
(
¸
=
E'
0
E (2.8)
In modal control, only a limited number of lower modes are controlled. Hence, l
controlled modes can be selected with l < n and the displacement may be partitioned into
controlled and uncontrolled parts as
) ( ) ( ) ( t t t
R C
x x x + = (2.9)
where x
C
and x
R
represent the controlled and uncontrolled displacement vector,
respectively. We refer to the uncontrolled modes as residual. Then, (2.7) can be rewritten
g C C C C C
x t t t & & & E f B w A w + + = ) ( ) ( ) ( (2.10a)
) ( ) ( t t
C C C
w C y = (2.10b)
where w
C
is a 2ldimensional modal state vector by the controlled modes and
(
¸
(
¸
− −
=
C
2
C
C
C
∆ Ω
I 0
A ,
(
¸
(
¸
=
C
C
B'
0
B ,
(
¸
(
¸
=
C
C
E'
0
E (2.11)
are the 2l×2l, 2l×m matrixes and a 2l×1 vector, respectively.
Chapter 2 Modal Control Scheme
21
For a feedback control, the control vector is related to the modal state vector according to
f (t) = −K
C
w
C
(t) (2.12)
where K
C
is an m×2l control gain matrix. Note that, in using the control law given by
(2.12), the closedloop modal equations are not independent.
Because the force generated in the i th MR damper depends on the responses of the
structural system, the MR damper cannot always produce the desired optimal control
force f
Ci
. Only the control voltage v
i
can be directly controlled. Thus, the strategy of the
clippedoptimal control (Dyke et al. 1996a) is used, in which a force feedback loop is
incorporated to induce the force in the MR damper f
i
to generate approximately the
desired optimal control force f
Ci
. To this end, the i th command signal v
i
is selected
according to the control law
] [(
max i i
H )f f f V v
i i C
− = (2.13)
where V
max
is the voltage to the current driver associated with saturation of the MR effect
in the physical device, and H(w) is the Heaviside step function.
2.1.2 Design of Optimal Controller
Referring to the discussions in above section, control gain matrix K
C
should be
decided. Although a variety of approaches may be used to design the optimal controller,
H
2
/LQG (Linear Quadratic Gaussian) methods are advocated because of their successful
application in previous studies (Dyke et al. 1996a,b,c).
For the controller design,
g
x& & is taken to be a stationary white noise, and an infinite
horizon performance index is chosen that weights the modal states by controlled modes
such as
Chapter 2 Modal Control Scheme
22
(
¸
(
¸
+ =
∫
∞ →
)dt ( E
τ
1
lim J
τ
0
T
C
T
C
τ
Ru u w Q w (2.14)
where R is a 2 × 2 identity matrix because the numerical example has two MR dampers,
and Q is a 2l × 2l diagonal matrix. It should be noted that the size of Q is reduced from 2n
× 2n to 2l × 2l because the limited lower modes are controlled. Therefore, it can be said
that it is more convenient to design the smaller weighting matrix of modal control. For
example, when the lowest one mode is selected for calculating the modal control action,
Q is a 2 × 2 diagonal matrix such as
(
¸
(
¸
=
mv
md
q
q
0
0
Q (2.15)
where q
md
is a weighting element for a modal displacement and q
mv
is for a modal
velocity. When the lowest two modes are controlled, Q is the 4 × 4 diagonal matrix.
(
(
(
(
¸
(
¸
=
mv2
mv1
md2
md1
q 0
q
q
0 q
Q (2.16)
The measurement noise is assumed to be identically distributed, statistically independent
Gaussian white noise processes, and 100 / = = γ
i i g g
v v x x
S S
& & & &
. Then, the controller is
C C C C C
s s B LC A I K G
ˆ
)] ( [ ) (
1 −
− − = (2.17)
where ] [
ˆ
LD B L B − = . Here, K
C
is the state feedback gain matrix for the deterministic
regulator problem given by
Chapter 2 Modal Control Scheme
23
P B K
C C
' = (2.18)
where P is the solution of the algebraic Ricatti equation given by
0 ' ' ' = + − +
C C C C C C
QC C P B PB P A PA (2.19)
and
)' ( S C L
C
= (2.20)
where S is the solution of the algebraic Ricatti equation given by
0 = + − +
C C C C C C
E' E γ S C SC' S A SA' (2.21)
2.1.3 Modal State Estimation
An observer for modal state estimation should be provided, since real sensors may
not estimate the full modal states directly or the system may be expensive to prepare the
sensors for the full states. To estimate the modal state vector w
C
(t) from the measured
output y(t), we consider an observer. Luenberger observers are used for low noiseto
signal ratios and KalmanBucy filters for high noisetosignal ratios (Meirovitch, 1990).
A modal control method using the full state feedback may not be practical for a
structural system involving a large number of DOFs, since the control implementation
may requires a large amount of sensors. Thus a modal control scheme that uses a modal
state estimation, is desirable. Moreover, accurate measurements of displacements and
velocities are difficult to achieve directly in fullscale applications, particularly during
seismic activity, since the foundation of the structure is moving with the ground. Hence,
it is ideal to use the acceleration feedback because accelerometers can readily provide
reliable and inexpensive measurements of accelerations at arbitrary points on the structure
Chapter 2 Modal Control Scheme
24
(Dyke et al. 1996a, b). Not only, the acceleration feedback is considered, but also the
state feedback including velocities and displacements, is implemented for the modal state
estimation using a KalmanBucy filter. In any case, we can write a modal observer in the
form
)] ( ) ( ˆ ) ( [ ) ( ) ( ˆ ) ( ˆ t t t x t t t
C C C g C C C C C
f D w C y L E f Β w A w − − + + + = & &
&
(2.22)
where ) ( ˆ t
C
w is the estimated controlled modal state and L is the optimally chosen
observer gain matrix by solving a matrix Riccati equation, which assumes that the noise
intensities associated with earthquake and sensors are known. C
C
is changeable according
to the signals that are used for the feedback and D
C
is generally zero except the
acceleration feedback. For modal state estimation from the displacements, C
C
in (2.22) is
as follows;
C
C
= ] 0 [
C
Φ (2.23)
For control with the velocity feedback,
C
C
= ] 0 [
C
Φ (2.24)
For control with the acceleration feedback,
C
C
=
(
¸
(
¸
− −
C
C
C M K M
Φ
Φ
0
0
] [
1 1
and D
C
= Λ
1 −
M (2.25)
Upon obtaining the estimated controlled modal state from (2.22), we compute the
feedback control forces
f (t) = −K
C
) ( ˆ t
C
w (2.26)
Chapter 2 Modal Control Scheme
25
Until now, the uncontrolled modes are ignored. In reality, however, the sensor signals
will include contributions from all the modes, so that the output vector is corrected to
) ( ) ( ) ( t t t
R R C C
w C w C w C y + = = (2.27)
To examine the effect of the control forces on the uncontrolled modes, residual modes
can be written
g R R R R R
x t t t & & & E f B w A w + + = ) ( ) ( ) ( (2.28)
where w
R
is a residual state vector by uncontrolled modes. Substituting (2.26) into (2.10a)
and considering (2.28), we obtain
g C C C C C C C
x t t t & & & E w K B w A w + − = ) ( ˆ ) ( ) ( (2.29a)
g R C C R R R R
x t t t & & & E w K B w A w + − = ) ( ˆ ) ( ) ( (2.29b)
Moreover, substituting (2.26) and (2.27) into (2.22), we can write the observer equation
in the form
g C R R C C C C C C C C
x t t t t t & &
&
E w LC w w LC w K B A w + + − + − = ) ( ] ) ( ˆ ) ( [ ) ( ˆ ) ( ) ( ˆ (2.30)
Then the error vector is defined
) ( ) ( ˆ ) ( t t t
C C C
w w e − = (2.31)
so that (2.29) and (2.30) can be rearranged
Chapter 2 Modal Control Scheme
26
g C C C C C C C C C
x t t t & & & E e K B w K B A w + − − = ) ( ) ( ) ( ) (
g R C C R R R C C R R
x t t t t & & & E e K B w A w K B w + − + − = ) ( ) ( ) ( ) ( (2.32)
) ( ) ( ) ( ) ( t t t
R R C C C C
w LC e LC A e + − = &
(2.32) can be written in the matrix form
g R
R
C
R
C
C C R
C R R C R
C C C C C
C
R
C
x
t
t
t
t
t
t
& &
&
&
&
(
(
(
¸
(
¸
+
(
(
(
¸
(
¸
(
(
(
¸
(
¸
−
− −
− −
=
(
(
(
¸
(
¸
0
E
E
e
w
w
LC A LC 0
K B A K B
K B 0 K B A
e
w
w
) (
) (
) (
) (
) (
) (
(2.33)
Note that the term
C R
K B − in (2.33) is responsible for the excitation of the residual
modes by the control forces and is known as control spillover (Balas, 1978). If
R
C is
zeros, which means the sensor signal only include controlled modes, the term
C R
K B −
has no effect on the eigenvalues of the closedloop system. Hence, we conclude that
control spillover cannot destabilize the system, although it can cause some degradation in
the system performance. Normally, however, the above system cannot satisfy the separate
principle because the term LC
R
affects eigenvalues of the controlled system by the
observer. This effect is known as observation spillover and can produce instability in the
residual modes. However, a small amount of damping inherent in the structure is often
sufficient to overcome the observation spillover effect.(Meirovitch and Baruh, 1983). At
any rate, observation spillover can be eliminated if the sensor signals are prefiltered so as
to screen out the contribution of the uncontrolled modes (Meirovitch, 1990)
Chapter 2 Modal Control Scheme
27
2.1.4 Elimination of Observable Spillover
(2.33) in above section should be further improved for eliminating the observable
spillover. A lowpass filter is introduced to measure the filtered response vector y
f
defined
as
) ( ) (
) ( ) ( ) (
t t
t t t
y z f
y z
y M z H y
y G z F z
+ =
+ = &
(2.34)
or in the frequency domain
) ( ) ( ) ( jω y jω H jω y
y f
= (2.35)
where ] ) ( [ ) (
1
y y z z y
jω jω M G F I H H + − =
−
. Substituting (2.27) into (2.35), the new
sensor dynamics becomes
)] ( ) ( )[ ( ) ( jω jω jω jω
R R C C y f
w C w C H y + = (2.36)
If the lowpass filter dynamics H
y
(jω) can be selected as a diagonal matrix, (2.36)
becomes
)] ( ) ( [ )] ( ) ( [ ) ( jω jω jω jω jω
R y R C y C f
w H C w H C y + = (2.37)
The pole of the lowpass filter dynamics can be placed by proper selection of the
parameters, H
z
, F
z
, G
y
, M
y
, then the rolloff can be occurred forth the lowest modal
frequency of the residual dynamics. The second term of righthand side of (2.37), which
represents the residual modal state, may have the following characteristics.
 ) (   ) ( ) ( 
1
jω jω jω
R R y
w w H ε ≅ for ∞ ≤ < ω 0 (2.38)
Chapter 2 Modal Control Scheme
28
where 0
1
≅ ε . Otherwise, the first term of righthand side of (2.37), which represents the
controlled modal state, may also have the following characteristics.
 ) (   ) ( ) (  jω jω jω
C C y
w w H ≅ for ∞ ≤ < ω 0 (2.39)
From (2.38) and (2.39), the new sensor dynamics y
f
can be rewritten as
) ( ) ( ) ( ) (
1
ε ε O w C w C y + + ≅ jω jω jω
R R C C f
(2.40)
or in time domain
) ( ) ( ) ( ) (
1
ε ε O w C w C y + + ≅ t t t
R R C C f
(2.41)
Substituting (2.41) into (2.34), the controlled system matrix in (2.33) becomes
g R
C
C
R
C
C C
C R R C R
C C C C C
C
R
C
x
t
t
t
t
t
t
& &
&
&
&
(
(
(
¸
(
¸
+
(
(
(
¸
(
¸
(
(
(
¸
(
¸
−
− −
− −
=
(
(
(
¸
(
¸
0
E
E
e
w
w
LC A ε 0
K B A K B
K B 0 K B A
e
w
w
2
) (
) (
) (
) (
) (
) (
(2.42)
where 0
1 2
≅ =
R
LC ε ε . Thus, the separate principle can be applied in the design of
observer gain since the term L in ε
2
no longer contributes to the characteristics of the
system. In other words, the observable spillover does not occur in this controlled system.
Hence, the controlled modal states in (2.22) may be suppressed by a welldesigned
control input, and the residual modal states may be also attenuated by their natural
damping.
Chapter 2 Modal Control Scheme
29
2.2 Numerical Example
To evaluate the proposed modal control scheme for use with the MR damper, a
numerical example is considered in which a model of a sixstory building is controlled
with four MR dampers (Fig. 2.1). This numerical example is the same with that of Jansen
and Dyke (2000) and is adopted for direct comparisons between the proposed modal
control scheme and other control algorithms. Two MR dampers are rigidly connected
between the ground and the first floor, and two MR dampers are rigidly connected
between the first and second floors.
Figure 2.1 Schematic diagram of the MR damper implementation
(Jansen and Dyke 2000)
Chapter 2 Modal Control Scheme
30
Each MR damper is capable of producing a force equal to 1.8% the weight of the
entire structure, and the maximum voltage input to MR devices is V
max
= 5V. The
governing equations can be written in the form of (2.7) by defining the mass of each
floor, m
i
, as 0.227 N/(cm/sec
2
), the stiffness of each floor, k
i
, as 297 N/cm, and a damping
ratio for each mode of 0.5%. MR damper parameters used in this study are c
0a
= 0.0064
Nsec/cm, c
0b
= 0.0052 Nsec/cmV, α
a
= 8.66 N/cm, α
b
= 8.86 N/cmV, g = 300 cm
2
, b =
300 cm
2
, A = 120, and n = 2. In simulation, the model of the structure is subjected to the
NS component of the 1940 El Centro earthquake. Because the building system considered
is a scaled model, the amplitude of the earthquake was scaled to ten percent of the full
scale earthquake.
Figs. 2.2 and 2.3 show the uncontrolled responses of the first and sixth floors,
respectively, in frequency domain. From Fig. 2.2, it can be seen that the first mode is
dominant in relative displacement and velocity of the first floor, whereas the lowest three
modes are dominant in the absolute acceleration. In Fig. 2.3, however, we can find that
the first mode is dominant in all responses of the sixth floor. Thus, it will be possible to
reduce the responses through modal control that control using the lowest one or two
modes.
The various control algorithms were evaluated using a set of evaluation criteria
based on those used in the second generation linear control problem for buildings
(Spencer et al., 1997a). The first evaluation criterion is a measure of the normalized
maximum floor displacement relative to the ground, given as

.

\

=
max
i
i t,
1
x
 t x
J
) (
max (2.43)
where x
i
(t) is the relative displacement of the i th floor over the entire response, and x
max
denotes the uncontrolled maximum displacement. The second evaluation criterion is a
measure of the reduction in the interstory drift. The maximum of the normalized
interstory drift is
Chapter 2 Modal Control Scheme
31


.

\

=
max
n
i i
i t,
2
d
 /h t d
J
) (
max (2.44)
where h
i
is the height of each floor (30cm), d
i
(t) is the interstory drift of the above ground
floors over the response history, and
max
n
d denotes the normalized peak interstory drift in
the uncontrolled response. The third evaluation criterion is a measure of the normalized
peak floor accelerations, given by


.

\

=
max
a
ai
i t,
3
x
 t x 
J
& &
& & ) (
max (2.45)
where the absolute accelerations of the ith floor, ) (t x
ai
& & , are normalized by the peak
uncontrolled floor acceleration, denoted ) (t x
max
a
& & . The final evaluation criteria considered
in this study is a measure of the maximum control force per device, normalized by the
weight of the structure, given by

.

\

=
W
(t) f
J
i
i t,
4
max (2.46)
where W is the total weight of the structure (1335 N). The corresponding uncontrolled
responses are as follows: x
max
= 1.313 cm,
max
a
d = 0.00981 cm,
max
a
x& & = 146.95 cm/sec
2
.
The resulting evaluation criteria are presented in Table 1 for the control algorithms
previously studied (Jansen and Dyke, 2000). The numbers in parentheses indicate the
percent reduction as compared to the best passive case. To compare the performance of
the semiactive system to that of comparable passive systems, two cases are considered in
which MR dampers are used in a passive mode by maintaining a constant voltage to the
devices. The results of passiveoff (0V) and passiveon (5V) configurations are included.
Chapter 2 Modal Control Scheme
32
0 1 2 3 4 5 6 7 8 9 10
0
0.2
0.4
0.6
0.8
1
1.2
1.4
P
S
D
frequency, Hz
°10
4
P
o
w
e
r
S
p
e
c
t
r
u
m
o
f
V
e
l
o
c
i
t
y
0 1 2 3 4 5 6 7 8 9 10
0
0.5
1
1.5
2
x 10
6
PS
D
frequency, Hz
P
o
w
e
r
S
p
e
c
t
r
u
m
o
f
A
c
c
e
l
e
r
a
t
i
o
n
Frequency, Hz
Figure 2.2 Frequency responses of the first floor for the uncontrolled structures
under the scaled El Centro earthquake
0 1 2 3 4 5 6 7 8 9 10
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
P
S
D
frequency, Hz
P
o
w
e
r
S
p
e
c
t
r
u
m
o
f
R
e
l
a
t
i
v
e
D
i
s
p
l
a
c
e
m
e
n
t
°10
2
x 10
5
Chapter 2 Modal Control Scheme
33
Figure 2.3 Frequency responses of the sixth floor for the uncontrolled structures
under the scaled El Centro earthquake
0 1 2 3 4 5 6 7 8 9 10
0
0.5
1
1.5
2.0
2.5
3.0
P
S
D
frequency, Hz
°10
2
P
o
w
e
r
S
p
e
c
t
r
u
m
o
f
R
e
l
a
t
i
v
e
D
i
s
p
l
a
c
e
m
e
n
t
0 1 2 3 4 5 6 7 8 9 10
0
0.5
1
1.5
2
x 10
5
P
S
D
frequency, Hz
P
o
w
e
r
S
p
e
c
t
r
u
m
o
f
V
e
l
o
c
i
t
y
0 1 2 3 4 5 6 7 8 9 10
0
2
4
6
8
10
12
14
16
x 10
6
P
S
D
frequency, Hz
P
o
w
e
r
S
p
e
c
t
r
u
m
o
f
A
c
c
e
l
e
r
a
t
i
o
n
Frequency, Hz
Chapter 2 Modal Control Scheme
34
Table 2.1 Normalized controlled maximum responses
due to the scaled El Centro earthquake*
Control strategy J
1
J
2
J
3
J
4
Passiveoff 0.862 0.801 0.904 0.00292
Passiveon 0.506 0.696 1.41 0.0178
Lyapunov controller A 0.686(+35) 0.788(+13) 0.756(−16) 0.0178
Lyapunov controller A 0.326(−35) 0.548(−21) 1.39(+53) 0.0178
Decentralized bangbang 0.449(−11) 0.791(+13) 1.00(+11) 0.0178
Maximum energy dissipation 0.548(+8) 0.620(−11) 1.06(+17) 0.0121
Clippedoptimal A 0.631(+24) 0.640(−8) 0.636(−29) 0.01095
Clippedoptimal B 0.405(−20) 0.547(−21) 1.25(+38) 0.0178
Modified homogeneous friction 0.421(−17) 0.599(−20) 1.06(+17) 0.0178
(* Jansen and Dyke 2000)
Chapter 2 Modal Control Scheme
35
For modal control, three cases of the structural measurements are considered;
displacements, velocities and accelerations. Using each structural measurement, a
Kalman filter estimates the modal states. Fig. 2.4 represents the results of the stochastic
response analysis for the acceleration feedback case. The variations of each evaluation
criteria for increasing weighting parameters are shown in a 3dimensional plot.
Previously mentioned, J
1
is evaluation criteria for the maximum displacement, J
2
is for
the maximum interstory drift and J
3
is for the maximum acceleration. In Fig. 4, J
T
is the
summation of evaluation criteria, J
1
, J
2
and J
3
. From the variations of J
T
, we can find the
weighting for reduction of overall structural responses whereas from J
1
, J
2
and J
3
, we can
find the weighting for reduction of related responses. In Fig. 2.4, it can be seen that J
1
is
minimum at q
md
= 400 and q
mv
= 1500, J
2
is at q
md
= 1 and q
mv
= 500, J
3
is at q
md
= 2200
and q
mv
= 100 and J
4
is at q
md
= 500 and q
mv
= 600. Designer can decide which to use
according to control objectives. By using the controller (H2/LQG) with designed
weighting matrices from Fig. 2.4, we can get the results in Table 2.2.
Figs. 2.5 and 2.6 represent the results for the displacement and velocity feedback
cases, respectively. Tables 2.5 and 2.6 summarize the results for each minimum
evaluation criteria of the designed weighting matrices from Figs. 2.5 and 2.6.
For each feedback case, in Tables 2.2 to 2.4, four modal control designs with
different capabilities are considered. In Table 2.2, the modal controller AJ
1
, AJ
2
, AJ
3
and
AJ
T
with acceleration feedback use a weighting that minimize the evaluation criteria J
1
,
J
2
, J
3
and J
T
, respectively. In Tables 2.3 to 2.4, the modal controller DJ
1
, DJ
2
, DJ
3
, and
DJ
T
with displacement feedback and VJ
1
, VJ
2
,VJ
3
, and VJ
T
with velocity feedback use a
weighting which minimize the evaluation criteria J
1
, J
2
, J
3
and J
T
, respectively. For each
weighting, the lowest one and two modes cases are given in Tables 2.2 to 2.4. In the
lowest two modes case, we place identical weighting on the each mode; q
md1
= q
md2
= q
md
and q
mv1
= q
mv2
= q
mv
.
Chapter 2 Modal Control Scheme
36
Figure 2.4 Variations of evaluation criteria with weighting parameters
for the acceleration feedback
J
1
J
2
J
3
J
T
=J
1
+ J
2
+ J
3
q
md
q
mv
q
md
q
mv
q
md
q
mv
q
md
q
mv
Chapter 2 Modal Control Scheme
37
Figure 2.5 Variations of evaluation criteria with weighting parameters
for the displacement feedback
J
1
J
2
J
3
J
T
=J
1
+ J
2
+ J
3
q
md
q
mv
q
md
q
mv
q
md
q
mv
q
md
q
mv
Chapter 2 Modal Control Scheme
38
Figure 2.6 Variations of evaluation criteria with weighting parameters
for the velocity feedback
J
1
J
2
J
3
J
T
=J
1
+ J
2
+ J
3
q
md
q
mv
q
md
q
mv
q
md
q
mv
q
md
q
mv
Chapter 2 Modal Control Scheme
39
The calculated evaluation criteria for various control strategies are compared in
Tables 2.1 to 2.4. The performance of the proposed modal control scheme is generally
better than that of other control strategies. The results show that the modal controller A
and V appear to be quite effective in achieving significant reductions in both the
maximum displacement and interstory drift over the passive case. In fact, the modal
controller AJ
1
achieves a 39% reduction in the relative displacement as compared to the
better passive case. If further reductions in interstory drift and acceleration are desired in
the controller, modal controller AJ
2
and AJ
3
can achieve the reductions in the interstory
drift and absolute acceleration of 30% and 23%, respectively, over the best passive cases,
although the maximum displacement increased. The reduction by modal controller AJ
2
is
resulting in the lowest interstory drift of all cases considered here. In Table 2.4, modal
controller VJ
1
using the lowest two modes and VJ
3
achieve reductions in relative
displacement and absolute acceleration of 41% and 30%, respectively, resulting in the
lowest values of all cases considered here. The modal controller AJ
T
and VJ
T
do not
achieve any lowest value of evaluation criteria, but have competitive performance in all
evaluation criteria. Notice that the designer has some versatility depending on the control
objectives for the particular structure under consideration.
The modal controller D compared with the modal controller A and V appears to be
worse in achieving reductions, which agrees with the fact that the variations of evaluation
criteria are more sensitive to weighting parameter q
mv
than q
md
from Figs. 2.4 to 2.6.
Comparing the lowest one mode case with twomode case, every lowest value of
evaluation criteria occurs at the lowest one mode case, except the modal controller VJ
1
that achieves further reductions by 6% from one mode case (reductions of 41% over the
best passive case) in the relative displacement.
Chapter 2 Modal Control Scheme
40
Table 2.2 Normalized controlled maximum responses of the acceleration feedback
due to the scaled El Centro earthquake
Control strategy J
1
J
2
J
3
J
4
1 mode 0.310(39) 0.529(24) 1.07(+18) 0.0178
Modal control A
J1
(q
md
=400, q
mv
=1500)
2 modes 0.392(23) 0.543(22) 1.05(+16) 0.0178
1 mode 0.398(21) 0.485(30) 0.870(4) 0.0178
Modal control A
J2
(q
md
=1, q
mv
=500)
2 modes 0.413(18) 0.510(27) 0.781(14) 0.0178
1 mode 0.549(+8) 0.618(11) 0.697(23) 0.0178
Modal control A
J3
(q
md
=2200, q
mv
=100)
2 modes 0.548(+8) 0.585(16) 0.741(18) 0.0178
1 mode 0.380(25) 0.488(30) 0.823(9) 0.0178 Modal control A
JT
(q
md
=500, q
mv
=600)
2 modes 0.423(16) 0.533(23) 0.876(3) 0.0178
Chapter 2 Modal Control Scheme
41
Table 2.3 Normalized controlled maximum responses of the displacement feedback
due to the scaled El Centro earthquake
Control strategy J
1
J
2
J
3
J
4
1 mode 0.403(20) 0.560(20) 0.765(15) 0.0178
Modal control D
J1
(q
md
=100, q
mv
=4900)
2 modes 0.325(36) 0.504(28) 1.06(+17) 0.0178
1 mode 0.403(20) 0.560(20) 0.769(15) 0.0178
Modal control D
J2
(q
md
=100, q
mv
=4900)
2 modes 0.325(36) 0.504(28) 1.06(+17) 0.0178
1 mode 0.702(+39) 0.728(+5) 0.671(26) 0.0178
Modal control D
J3
(q
md
=200, q
mv
=4900)
2 modes 0.678(+34) 0.689(1) 0.796(12) 0.0178
1 mode 0.408(19) 0.566(19) 0.721(20) 0.0178 Modal control D
JT
(q
md
=3300,q
mv
=4700)
2 modes 0.329(35) 0.510(27) 1.04(+15) 0.0178
Chapter 2 Modal Control Scheme
42
Table 2.4 Normalized controlled maximum responses of the velocity feedback
due to the scaled El Centro earthquake
Control strategy J
1
J
2
J
3
J
4
1 mode 0.327(35) 0.554(20) 1.06(+17) 0.0178
Modal control V
J1
(q
md
=700, q
mv
=800)
2 modes 0.301(41) 0.530(24) 1.07(+18) 0.0178
1 mode 0.383(24) 0.487(30) 0.874(3) 0.0178
Modal control V
J2
(q
md
=1, q
mv
=400)
2 modes 0.351(31) 0.510(27) 0.941(+4) 0.0178
1 mode 0.541(+7) 0.611(12) 0.632(30) 0.0178
Modal control V
J3
(q
md
=1300, q
mv
=100)
2 modes 0.522(+3) 0.583(16) 0.553(39) 0.0178
1 mode 0.354(30) 0.502(28) 0.825(9) 0.0178
Modal control V
JT
(q
md
=600,q
mv
=500)
2 modes 0.323(36) 0.510(27) 0.827(9) 0.0178
Chapter 2 Modal Control Scheme
43
2.3 Summary of Results
In this study, modal control was implemented to seismically excited structures
using MR dampers. To this end, a modal control scheme was applied together with a
Kalman filter and a lowpass filter. A Kalman filter considered three cases of the
structural measurement to estimate modal states: displacement, velocity, and acceleration,
respectively. Moreover, a lowpass filter was used to eliminate spillover problem. In a
numerical example, a sixstory structure was controlled using MR dampers on the lower
two floors. The responses of the system to a scaled El Centro earthquake excitation were
found for each controller through a simulation of the system.
Modal control reshapes the motion of a structure by merely controlling a few
selected vibration modes. Hence, in designing phase of controller, the size of weighting
matrix Q was reduced because the lowest one or two modes were controlled. Therefore, it
is more convenient to design the smaller weighting matrix of modal control. This is one
of the important benefits of the proposed modal control scheme.
The numerical results show that the motion of the structure was effectively
suppressed by merely controlling a few lowest modes, although resulting responses
varied greatly depending on the choice of measurements available and weightings. The
modal controller A and V achieved significant reductions in the responses. The modal
controller AJ
2
, VJ
1
and VJ
3
achieve reductions (30%, 41%, 30%) in evaluation criteria J
1
,
J
2
and J
3
, respectively, resulting in the lowest values of all cases considered here. The
modal controller AJ
T
and VJ
T
fail to achieve any lowest value of evaluation criteria, but
have competitive performance in all evaluation criteria. Based on these results, the
proposed modal control scheme is found to be suited for use with MR dampers in a multi
input control system. Further studies are underway to examine the influence of the
number of controlled modes on the control performance.
Chapter 3 Maximum Energy Dissipation Algorithm
44
CHAPTER 3
MAXIMUM ENEGRY DISSIPATION ALGORITHM
3.1 Control System
Consider a seismically excited structure controlled with n MR dampers. The
equation of motion can be written.
g
x& & & & & MΓ Λf Kx x C x M − = + + (3.1)
where x is vector of the relative displacements of the floors of the structure;
g
x& & is one
dimensional ground acceleration; f = [ f
1
, f
2
,…, f
n
]
T
is the vector of measured control
forces generated by n MR dampers; Γ is the column vector of ones; and Λ is the matrix
determined by the placement of MR dampers in the structure. This equation can be
written in statespace form as
g
x& & & E Bf Az z + + = (3.2)
v Df Cz y + + = (3.3)
where z is a state vector; y is the vector of measured outputs; and v is a measurement
noise vector. More details of system matrices can be found in Dyke et al (2003) and
Ohtori et al (2000, 2002).
Chapter 3 Maximum Energy Dissipation Algorithm
45
3.1.1 Control Devices
The MR damper with capacity of 1000KN is considered as control devices. To
accurately predict the behavior of controlled structure, an appropriate modeling of MR
dampers is essential. Several types of controloriented dynamic models have been
investigated for modeling MR dampers. Herein, the BoucWen model is considered. The
BoucWen model (Spencer et al, 1997a), which is numerically tractable and has been
used extensively for modeling hysteretic system, is considered for describing the behavior
of the MR damper (Figure 3.1).
Figure 3.1 Mechanical model of the MR damper
The force generated by the damper is given by
x c z f &
0
+ =α (3.4)
where the evolutionary variable z is governed by
x A z x z z x z
n n
& & & + − − =
−
     
1
β γ (3.5)
By adjusting the parameters of the model γ, β, n, and A, the degree of linearity in the
unloading and the smoothness of the transition from the preyield to the postyield region
Chapter 3 Maximum Energy Dissipation Algorithm
46
can be controlled. Some of the model parameters depend on the command voltage u to
the current driver as follows.
u
b a
α α α + = and u c c c
b a 0 0 0
+ = (3.6)
Parameters for both benchmark problems are listed in Table 3.1. Each parameter is
adopted from Yoshida and Dyke (2002) for the nonlinear benchmark building and from
Moon et al. (2003) for the cablestayed bridge.
Table 3.1 Parameters for MR damper model
Value
Parameter
For nonlinear building For cablestayed bridge
α
a
1.087e5 N/cm 500 N/m
α
b
4.962e5 N/(cm⋅V) 671.41 N/(m⋅V)
c
0a
4.40 N s/cm 0.15 N s/m
c
0b
44.0 N s/(cm⋅V) 1.43 N s/(cm⋅V)
η 50 s
1
300 s
1
γ 3 cm
2
300 m
2
β 3 cm
2
300 m
2
A 1.2 120
n 1 1
Chapter 3 Maximum Energy Dissipation Algorithm
47
3.1.2 Maximum Energy Dissipation Algorithm for MR Damper
This control algorithm is presented as a variation of the decentralized bangbang
approach proposed by McClamroch and Gavin (1995). Lyapunov’s direct approach
requires the use of a Lyapunov function, denoted V(x), which must be a positive definite
function of the states of the system x. In the decentralized bangbang approach, the
Lyapunov function was chosen to represent total vibratory energy in the system. Jansen
and Dyke (2000) instead consider a Lyapunov function that represents the relative
vibratory energy in the structure as in
Mx x Kx x V
T T
2
1
2
1
+ = (3.7)
According to Lyapunov stability theory, if the rate of change of the Lyapunov function
) ( x V
&
is negative semidefinite, the origin is stable in the sense of Lyapunov. Using
(3.7), the rate of change of the Lyapunov function is then
Λf) MΓ Kx x C M( x x K x V
T T
+ − − − + =
g
x& & & & &
&
(3.8)
In this expression, the only way to directly effect V
&
is through the last term containing
the force vector f. To control this term and make V
&
as large and negative as possible, the
following control law is obtained:
) (
max i i i
f x H V v Λ & − = (3.9)
where Λ
i
is ith column of the Λ matrix; f
i
is i th column of the f matrix.
Note that MEDA is very simple because only local measurements (i.e., the velocity
and control force) are required to implement this control law. In (3.9), there is no design
parameter to decide, which is essential part in other control laws. In other words, complex
Chapter 3 Maximum Energy Dissipation Algorithm
48
design process can be skipped. This is the important benefit of using MEDA. Otherwise,
the more structures are complex, the more design parameters are considered. Therefore, it
can be said that it is more convenient to use MEDA for structural control, especially for
the largesize civil structures.
Chapter 3 Maximum Energy Dissipation Algorithm
49
3.2 Benchmark Problems
In this study, we consider two kinds of benchmark problem: a cablestayed bridge
and a 20story nonlinear building. The cablestayed bridge and the highrise nonlinear
building model are representative structures of civil engineering. Using both benchmark
problems, we exploit MEDA for civil engineering applications. For the completeness, this
section briefly summarizes both benchmark problems, respectively. More details can be
found in Dyke et al (2003) and Ohtori et al (2000, 2002).
3.2.1 Benchmark CableStayed Bridge
At the Second International Workshop on Structural Control (Dec. 1820, 1996,
Hong Kong), the Working Group on Bridge Control developed plans for a "first
generation" benchmark study on bridges. The cablestayed bridge used for this
benchmark study is the Missouri 74–Illinois 146 bridge spanning the Mississippi River
near Cape Girardeau, Missouri, designed by the HNTB Corporation (Hague, 1997). The
bridge is currently under construction and is to be completed in 2003. Seismic
considerations were strongly considered in the design of this bridge due to the location of
the bridge (in the New Madrid seismic zone) and its critical role as a principal crossing of
the Mississippi River. In early stages of the design process, the loading case governing
the design was determined to be due to seismic effects. Earthquake load combinations in
accordance with American Association of State Highway and Transportation Officials
(AASHTO) division IA specifications were used in the design. Various designs were
considered, including full longitudinal restraint at the tower piers, no longitudinal
restraint, and passive isolation. When temperature effects were considered, it was found
that fully restraining the deck in the longitudinal direction would result in unacceptably
large stresses. Based on examination of the various designs, it was determined that
incorporating force transfer devices would provide the most efficient solution.
Chapter 3 Maximum Energy Dissipation Algorithm
50
Sixteen 6.67 MN shock transmission devices are employed in the connection
between the tower and the deck. These devices are installed in the longitudinal direction
to allow for expansion of the deck due to temperature changes. Under dynamic loads
these devices are extremely stiff and are assumed to behave as rigid links. Additionally,
in the transverse direction earthquake restrainers are employed at the connection between
the tower and the deck and the deck is constrained in the vertical direction at the towers.
The bearings at bent 1 and pier 4 are designed to permit longitudinal displacement and
rotation about the transverse and vertical axis. Soilstructure interaction is not expected to
be an issue with this bridge as the foundations of the cablestayed portion is attached to
bedrock.
As shown in Fig. 3.2, the bridge is composed of two towers, 128 cables, and 12
additional piers in the approach bridge from the Illinois side. It has a total length of
1205.8 m. The main span is 350.6 m in length, the side spans are 142.7 m in length, and
the approach on the Illinois side is 570 m. A cross section of the deck is shown in Fig.
3.3. The bridge has four lanes plus two narrower bicycle lanes, for a total width of 29.3
m. The deck is composed of steel beams and prestressed concrete slabs. Steel ASTM
A709 grade 50W is used, with an f
y
of 344 MPa. The concrete slabs are made of
prestressed concrete with a f
c
' of 41.36 MPa. Additionally, a concrete barrier is located in
the center of the bridge, and a railing is located along the edges of the deck.
The 128 cables are made of high–strength, low–relaxation steel (ASTM A882
grade 270). The smallest cable area is 28.5 cm
2
and the largest cable area is 76.3 cm
2
. The
cables are covered with a polyethylene piping to resist corrosion. The Hshaped towers
have a height of 102.4 m at pier 2 and 108.5 m at pier 3. Each tower supports a total 64
cables. The towers are constructed of reinforced concrete with a resistance, f
c
', of 37.92
MPa. The approach bridge from the Illinois side is supported by 11 piers and bent 15
which are made of concrete. The deck consists of a rigid diaphragm made of steel with a
slab of concrete at the top.
Chapter 3 Maximum Energy Dissipation Algorithm
51
Figure 3.2 Drawing of the Cape Girardeau Bridge
Figure 3.3 Cross section of bridge deck.
Chapter 3 Maximum Energy Dissipation Algorithm
52
Based on the description of the Cape Girardeau bridge provided in the previous
section, a threedimensional finite element model of the bridge was developed in
MATLAB
®
(1997). A linear evaluation model is used in this benchmark study. However,
the stiffness matrices used in this linear model are those of the structure determined
through a nonlinear static analysis corresponding to the deformed state of the bridge with
dead loads (Wilson and Gravelle, 1991). Additionally, the bridge is assumed to be
attached to bedrock, and the effects of soil–structure interaction are neglected. A one
dimensional ground acceleration is applied in the longitudinal direction. This direction is
considered to be the most destructive in cablestayed bridges.
The finite element model employs beam elements, cable elements and rigid links.
The nonlinear static analysis is performed in ABAQUS
®
(1998), and the element mass
and stiffness matrices are output to MATLAB® for assembly. Subsequently, the
constraints are applied, and a reduction is performed to reduce the size of the model to
something more manageable. The first ten frequencies of the evaluation model are
0.2899, 0.3699, 0.4683, 0.5158, 0.5812, 0.6490, 0.6687, 0.6970, 0.7102, and 0.7203 Hz.
To make it possible for designers/researchers to place devices acting longitudinally
between the deck and the tower, a modified evaluation model is formed in which the
connections between the tower and the deck are disconnected. If a designers/researcher
specifies devices at these nodes, the second model will be formed as the evaluation
model, and the control devices should connect the deck to the tower. As one would
expect, the frequencies of this model are much lower than those of the nominal bridge
model. The first ten frequencies of this second model are 0.1618, 0.2666, 0.3723, 0.4545,
0.5015, 0.5650, 0.6187, 0.6486, 0.6965, and 0.7094 Hz. Note that the uncontrolled
structure used as a basis of comparison for the controlled system, corresponds to the
former model in which the decktower connections are fixed (the dynamically stiff shock
transmission devices are present).
The finite element model, shown in Fig. 3.4, has a total of 579 nodes, 420 rigid
links, 162 beam elements, 134 nodal masses and 128 cable elements. The towers are
Chapter 3 Maximum Energy Dissipation Algorithm
53
modeled using 50 nodes, 43 beam elements and 74 rigid links. Constraints are applied to
restrict the deck from moving in the lateral direction at piers 2, 3 and 4. Boundary
conditions restrict the motion at pier 1 to allow only longitudinal displacement (X) and
rotations about the Y and Z axes. Because the attachment points of the cables to the deck
are above the neutral axis of the deck, and the attachment points of the cables to the tower
are outside the neutral axis of the tower, rigid links are used to connect the cables to the
tower and to the deck (see Fig. 3.5). The use of the rigid links ensures that the length and
inclination angle of the cables in the model agree with the drawings. Additionally, the
moment induced in the towers by the movement of the cables is taken into consideration
in this approach. In the case of variable sections, the average of the section is used for the
finite element. The cables are modeled with truss elements. In the finite element model
the nominal tension is assigned to each cable.
The FEM model described above is used directly in cases when the control devices
are employed in the longitudinal direction between the deck and tower. If the
designer/researcher employs no control device at these locations (in which case the shock
transmission devices are included), the model is modified by including four
longitudinally directed, axially stiff beam elements that force the deck to move with the
tower in the longitudinal direction. The uncontrolled structure used as a basis of
comparison corresponds to this second case. Note that the program included with the
benchmark files determines if the designer/researcher has placed devices in this location
and builds the appropriate FEM model.
Note that the Illinois approach is not included in this model (see Fig. 3.2). Because
the bearing at pier 4 does not restrict longitudinal motion and rotation about the X axis of
the bridge, the Illinois approach was found to have a negligible effect on the dynamics of
the cablestayed portion of the bridge. The system matrices are provided at the
benchmark web site: http://wusceel.cive.wustl.edu/quake
Chapter 3 Maximum Energy Dissipation Algorithm
54
Figure 3.4 Finite element model
Figure 3.5 Finite element model of the towers
Chapter 3 Maximum Energy Dissipation Algorithm
55
3.2.2 Nonlinear Benchmark Building
During the 2nd World Conference on Structural Control (June 28 – July 1, 1998,
Tokyo), as a result of the success of the linear benchmark's presented, it was decided to
pursue the nonlinear analysis for the seismically excited buildings. The nonlinear
benchmark building considered here is the 20story benchmark building specified in the
benchmark problem statement (http://wusceel.cive.wustl.edu/ quake).
The 20story benchmark structure is 30.48 m by 36.58 m in plan, and 80.77 m in
elevation. The bays are 6.10 m on center, in both directions, with five bays in the north
south (NS) direction and six bays in the eastwest (EW) direction. The building’s lateral
loadresisting system is comprised of steel perimeter momentresisting frames (MRFs).
The interior bays of the structure contain simple framing with composite floors.
The columns are 345 MPa steel. The interior columns of the MRF are wideflange.
The corner columns are box columns. The levels of the 20story building are numbered
with respect to the ground level (see Fig. 3.6). The building has two basement levels. The
level directly below the ground level is the second basement (B1). The level below B1
is the second basement (B2). Typical floortofloor heights (for analysis purposes
measured from centerofbeam to centerofbeam) are 3.96 m. The floortofloor heights
for the two basement levels are 3.65 m and for the ground level is 5.49 m.
The column lines employ threetier construction, i.e. monolithic column pieces are
connected every three levels beginning with the first level. Column splices, which are
seismic (tension) splices to carry bending and uplift forces, are located on the first, fourth,
seventh, tenth, thirteenth, sixteenth and eighteenth levels at 1.83 m above the centerline
of the beam to column joint. The column bases are modeled as pinned and secured to the
ground (at the B2 level). Concrete foundation walls and surrounding soil are assumed to
restrain the structure at the ground level from horizontal displacement. The floor system
is comprised of 248 MPa steel wideflange beams acting compositely with the floor slab.
Each frame resists one half of the seismic mass associated with the entire structure.
Chapter 3 Maximum Energy Dissipation Algorithm
56
Figure 3.6 Schematic of the 20story benchmark building
Chapter 3 Maximum Energy Dissipation Algorithm
57
The seismic mass of the structure is due to various components of the structure,
including the steel framing, floor slabs, ceiling/flooring, mechanical/electrical, partitions,
roofing and a penthouse located on the roof. The seismic mass, including both NS
MRFs, of the ground level is 5.32´105 kg, for the first level is 5.63´105 kg, for the second
level to 19th level is 5.52´105 kg, and for the 20th level is 5.84´105 kg. The seismic mass
of the above ground levels of the entire structure is 1.11´107 kg. The 20story NS MRF
is depicted in Fig. 3.6.
This benchmark study focuses on an inplane (2D) analysis of the benchmark
structures. The frames considered in the development of the evaluation models are the N
S MRFs (the short, or weak, direction of the buildings) for the structures described in the
previous section. Passive, active and/or semiactive control devices can be implemented
throughout these NS frames of the 20story structure and their performance assessed
using the evaluation models in this section and the evaluation criteria identified in the
Control Design section.
Based on the physical description of the 20story structures, inplane finite element
models of the NS MRFs have been developed. Structural member nonlinearities are
included to capture the inelastic behavior of buildings during strong earthquakes. The
beams and columns of the structures are modeled as planeframe elements, and mass and
stiffness matrices for each of the structures are determined. A bilinear hysteresis model is
used to characterize the nonlinear bending stiffness of the structural members. The
damping matrix is determined based on an assumption of Rayleigh damping. This process
is described in further detail in the following paragraphs.
Nodes are located at beamtocolumn joints. Elements are created between nodes to
represent the beams and columns in the structure. The beam members extend from the
centerline of column to centerline of column, thus ignoring the column panel zone.
Inertial loads, accounting for the seismic mass of the floor slabs, ceiling/flooring,
mechanical/electrical, partitions, roofing and penthouse are uniformly distributed at the
nodes of each respective level assuming a lumped mass formulation.
Chapter 3 Maximum Energy Dissipation Algorithm
58
The 20story building frames contain column splices. The column joint of the
splice story is located 1.83 m above the centerline of the beam. For simplicity the spliced
columns are modeled as having uniform properties over the story height equal to the
weighted average of the upper and lower column properties of that story. There is no
node modeled at the splice. Each node has three degreesoffreedom (DOFs): horizontal,
vertical and rotational. The 20story structures have 414 DOFs prior to application of
boundary conditions/constraints, respectively.
Each element, modeled as a plane frame element, contains two nodes and six
DOFs. The length, area, moment of inertia, modulus of elasticity and mass density are
predefined for each element. The elemental lumped mass and stiffness matrices are
determined as functions of these properties (Sack 1989; Cook, et al. 1989). Global mass
and stiffness matrices are assembled from the elemental mass and stiffness matrices by
summing the mass and stiffness associated with each DOF for each element of the entire
structure. Rotational inertia is ignored; thus, rotational mass is assigned a small value.
The DOFs corresponding to fixed boundary conditions are then constrained by
eliminating the rows and columns associated with these DOFs from the global mass and
stiffness matrices. The simulation of the benchmark buildings is developed to represent
control of the entire structure, including both NS MRFs and the entire mass of each
structure. Researchers/designers should recognize that the control strategies applied
within this study represent the structural control of the entire benchmark building being
considered.
The first 10 natural frequencies of the 20story benchmark evaluation model are:
0.261, 0.753, 1.30, 1.83, 2.40, 2.44, 2.92, 3.01, 3.63 and 3.68 Hz. These results are
consistent with those found by others who have modeled this structure (Barrosa 1999;
Spencer, et al. 1998a,b). Assuming the structures respond in the elastic range, transfer
functions for the displacement and absolute acceleration at the top of each building from
ground acceleration can be determined.
Chapter 3 Maximum Energy Dissipation Algorithm
59
3.3 Numerical Examples
In the previous section, we found that MEDA is very simple to implement. In this
section, we will examine the applicability of the MEDAbased semiactive control system
from the viewpoint of the performance and the robustness. Through a series of numerical
simulations of benchmark problems, the results are compared with those of other control
algorithms: Clippedoptimal controller (CO; Yoshida and Dyke, 2002) and sliding mode
controller (SMC; Moon et. al, 2003). Clippedoptimal controller was suggested and
experimentally examined by Dyke et al. (1996a,b,c). The clippedoptimal control
approach is to design a linear optimal controller that calculates desired control forces
based on the measured structural responses and the measured control force applied to the
structure. A force feedback loop is incorporated to induce the MR damper to generate
approximately the desired optimal control force. Sliding mode controller was developed
specifically for robust control of uncertain nonlinear systems. The fundamental idea of
SMC is to design a controller to drive the state trajectory onto a sliding surface, where the
motion is stable. Thus, SMC is known as a robust controller.
3.3.1 Control Performance
Table 3.2 shows the values of the evaluation criteria for the benchmark cable
stayed bridge under various earthquakes. The definition of each evaluation criteria can be
found in Dyke et al. (2003) and Ohtori et al. (2000, 2002). Each controller employs 24
MR dampers between the deck and abutment and the deck and tower of the bridge, all
oriented to apply forces longitudinally. Four devices are located between each of the
following pairs of nodes on bent 1 and pier 3; and, two devices are located between each
of the following pairs of nodes on piers 2 and 4.
Note that each controller is able to achieve a significant reduction in the base shear
force when compared with the uncontrolled system; the base shears in MEDA are
Chapter 3 Maximum Energy Dissipation Algorithm
60
reduced to 33 ~ 59% levels in the peak value (J
1
) and to 23 ~ 46% levels in the normed
values (J
7
) for the three earthquake excitations. Overturning moments are reduced to 26 ~
56% levels in the peak values (J
3
) and to 23 ~ 45% levels in the normed value (J
9
). It is
clear that most of structural responses generated by three earthquakes are controlled well.
Further, the numerical results show that MEDA performs slightly better than two other
controllers.
Table 3.3 shows the evaluation criteria for the nonlinear benchmark building. Total
65 MR dampers are used. Four devices are located on the first eight stories, three devices
are located on the next nine stories, and two devices are located on the top three stories.
The building was subjected to the four earthquakes specified in the benchmark paper with
various intensities.
The interstory drifts in MEDA are reduced to 50 ~ 80% levels in the peak value
(J
1
) and to 27 ~ 99% levels in the normed values (J
4
), which are better performances than
those of clippedoptimal controller. MEDA, also, prevent the formation of plastic
connections (J
9
) for halfscale of Northridge and Kobe earthquakes. However, note that
MEDA fails to reduce the peak floor acceleration and the peak base shear, giving the
magnification of 241% and 144%, respectively, in maximum value over the uncontrolled
case, whereas clippedoptimal controller success. To improve the performance of MEDA
for the building structure, the location and the number of MR dampers can be changed.
An infinite number of configurations of MR dampers may be possible, which may result
in choosing an optimal one for purpose. In this study, we chose that one device is
installed on every second floor from the first floor, and then a total of 10 MR dampers are
used. The building was subjected to the four earthquakes specified in the benchmark
paper with various intensities. Table 3.4 shows the evaluation criteria for the nonlinear
benchmark building with the modified configuration of MR dampers. Note that the
magnifications of the peak floor acceleration and the peak base shear are reduced to
128% and 118% respectively in maximum value, but are still over the uncontrolled case.
Furthermore, the number and the location of MR dampers could not be chosen at one
Chapter 3 Maximum Energy Dissipation Algorithm
61
time. Various numbers and locations are tried and then one case is selected, which is the
most suitable for the purpose. This process of MEDA for building structure corresponds
to designing phase of other controller, which means MEDA for the building structure lose
its advantage of simple method to use. However, MEDA for the cablestayed bridge is
still simple and efficient controller as listed in Table 3.2.
Table 3.2. Comparisons of the evaluation criteria for benchmark cablestayed bridge
J
1
, Peak base shear J
2
, Peak shear at deck level
Controller El Centro Mexico Gebze El Centro Mexico Gebze
CO 0.391 0.469 0.415 1.084 1.179 1.376
SMC 0.397 0.453 0.392 1.090 1.068 1.146
MEDA 0.331 0.593 0.453 1.108 1.315 1.447
J
3
, Peak overturning moment J
4
, Peak moment at deck level
Controller El Centro Mexico Gebze El Centro Mexico Gebze
CO 0.267 0.466 0.395 0.537 0.472 0.953
SMC 0.300 0.488 0.382 0.557 0.408 1.053
MEDA 0.255 0.558 0.355 0.464 0.381 0.779
J
5
, Peak deviation of cable tension J
6
, Peak deck displacement
Controller El Centro Mexico Gebze El Centro Mexico Gebze
CO 0.189 0.060 0.142 0.933 1.282 2.519
SMC 0.205 0.056 0.159 0.880 1.578 2.941
MEDA 0.185 0.079 0.143 0.709 0.694 1.266
J
7
, Normed base shear J
8
, Normed shear at deck level
Controller El Centro Mexico Gebze El Centro Mexico Gebze
CO 0.234 0.440 0.328 0.975 1.147 1.331
SMC 0.217 0.372 0.286 0.903 0.902 1.271
MEDA 0.234 0.464 0.318 0.883 1.064 1.128
J
9
, Normed overturning moment J
10
, Normed moment at deck level
Controller El Centro Mexico Gebze El Centro Mexico Gebze
CO 0.300 0.393 0.391 0.624 0.656 1.194
SMC 0.193 0.315 0.380 0.577 0.720 1.487
MEDA 0.233 0.453 0.348 0.552 0.552 1.123
J
11
, Normed deviation of cable tension
Controller El Centro Mexico Gebze
CO 0.020 0.007 0.012
SMC 0.018 0.006 0.012
MEDA 0.020 0.011 0.010
Chapter 3 Maximum Energy Dissipation Algorithm
62
Table 3.3 Comparisons of the evaluation criteria for the nonlinear benchmark building
Clippedoptimal controller
(Yoshida and Dyke, 2002)
Maximum energy dissipation algorithm
Earthquake
Intensity
El Centro
0.5/1.0/1.5
Hach.
0.5/1.0/1.5
North.
0.5/1.0
Kobe
0.5/1.0
El Centro
0.5/1.0/1.5
Hach.
0.5/1.0/1.5
North.
0.5/1.0
Kobe
0.5/1.0
J
1
Interstory
drift ratio
0.747
0.748
0.748
0.883
0.887
0.907
0.859
0.942
0.816
0.728
0.639
0.642
0.658
0.713
0.683
0.745
0.666
0.796
0.497
0.636
J
2
Floor
acceleration
0.648
0.646
0.664
0.746
0.743
0.833
0.807
0.904
0.702
0.839
1.653
1.049
0.828
2.410
1.550
1.257
0.965
0.911
0.796
0.849
J
3
Base shear
0.780
0.782
0.909
0.977
0.982
1.010
0.885
0.969
0.925
1.070
1.267
0.996
1.025
1.440
1.223
1.172
0.988
1.122
0.789
1.258
J
4
Normed
interstory
drift ratio
0.662
0.663
0.670
0.885
0.884
0.903
0.724
0.929
0.648
0.230
0.455
0.468
0.486
0.724
0.735
0.768
0.442
0.986
0.273
0.271
J
5
Normed
floor
acceleration
0.563
0.560
0.578
0.658
0.652
0.661
0.592
0.637
0.579
0.713
1.328
0.961
0.812
2.081
1.370
1.091
0.712
0.761
0.624
0.691
J
6
Normed
base shear
0.724
0.723
0.729
0.849
0.848
0.858
0.776
0.841
0.689
0.840
0.710
0.622
0.600
0.933
0.818
0.797
0.474
0.724
0.401
0.638
J
7
Ductility
0.772
0.722
0.722
0.955
0.959
0.943
0.728
0.978
0.688
0.688
0.695
0.648
0.613
0.796
0.807
0.752
0.559
0.809
0.334
0.603
J
8
Dissipated
energy


0.078


0.714
0.220
0.548
0.144
0.323


0


0.008
0
0.328

0.058
J
9
Plastic
connections


0.372


0.791
0.542
0.906
0.308
0.810


0


0.233
0
0.760

0.595
J
10
Normed
ductility
0.733
0.733
0.656
0.847
0.847
0.890
0.632
0.944
0.777
0.227
0.480
0.492
0.449
0.700
0.711
0.730
0.325
0.984
0.248
0.309
J
11
Control
force
0.002
0.003
0.501
0.002
0.004
0.005
0.007
0.008
0.005
0.009
0.009
0.009
0.009
0.009
0.009
0.009
0.009
0.009
0.009
0.009
Chapter 3 Maximum Energy Dissipation Algorithm
63
Table 3.4 Evaluation criteria of modified location and number of MR dampers
Maximum energy dissipation algorithm
(modified location and number of MR dampers)
Earthquake
Intensity
El Centro
0.5/1.0/1.5
Hachinohe
0.5/1.0/1.5
Northridge
0.5/1.0
Kobe
0.5/1.0
J
1
Interstory drift
ratio
0.744
0.743
0.740
0.859
0.833
0.873
0.829
0.895
0.746
0.756
J
2
Floor
acceleration
0.960
0.794
0.755
1.281
0.977
0.972
0.865
0.960
0.732
0.867
J
3
Base shear
0.894
0.823
0.921
1.183
1.072
1.055
0.890
0.981
0.957
1.071
J
4
Normed
Interstory drift
ratio
0.540
0.549
0.576
0.823
0.818
0.839
0.571
1.050
0.487
0.221
J
5
Normed floor
acceleration
0.734
0.540
0.544
1.025
0.732
0.648
0.539
0.632
0.542
0.720
J
6
Normed base
shear
0.662
0.641
0.655
0.873
0.818
0.818
0.610
0.762
0.568
0.737
J
7
Ductility
0.764
0.740
0.686
0.977
0.967
0.961
0.660
0.929
0.663
0.757
J
8
Dissipated
energy


0.001


0.568
0.044
0.493
0.181
0.301
J
9
Plastic
connections


0.163


0.372
0.292
0.917
0.180
0.869
J
10
Normed
ductility
0.573
0.606
0.566
0.800
0.796
0.854
0.463
1.092
0.490
0.263
J
11
Control force
0.009
0.009
0.009
0.009
0.009
0.009
0.009
0.009
0.009
0.009
Chapter 3 Maximum Energy Dissipation Algorithm
64
3.3.2 Controller Robustness
The dynamic characteristics of the real structure may not be identical to those of
the evaluation model and can be changed after construction. Even if the designed
controller were confirmed to have good performance in the evaluation model, it dose not
necessarily guarantee good performance in the actual system. Therefore, the controller
robustness of the MEDAbased semiactive control system using MR dampers was
examined with respect to uncertainties in stiffness for the benchmark cablestayed bridge.
Since the performance of MEDA is deteriorated for the building structures, the robustness
check of MEDA is carried out only for the cablestayed bridge. The stiffness matrix is
perturbed by some factor ε, and the resulting bridge model was simulated using the
controller for the nominal system. The perturbed stiffness was calculated as
) 1 ( ε + = K K
pert
(3.10)
where K = nominal stiffness of the bridge, ε = perturbation parameter, and K
pert
=
perturbed stiffness. Perturbations of 7% and 30% were considered. The configuration of
MR dampers are followings; Four devices are located between each of the following pairs
of nodes on bent 1 and pier 4; and, two devices are located between each of the following
pairs of nodes on piers 2 and 3. Table 3.5 shows the evaluation criteria for ±7% stiffness
perturbed system under El Centro earthquake. SMC is known as robust controller (Moon
et al, 2003). So, the robustness of MEDA is compared to that of SMC and the nominal
performance is listed with the perturbed performance. In Table 3.4, MEDA is stable and
performs well for ±7% perturbed system. MEDA shows more robust performance than
SMC except the peak shear J
2
, the normed overturning moment J
9
, and the normed
deviation of cable tension J
11
. Table 3.6 shows the evaluation criteria for ±30% stiffness
perturbed system under the three earthquakes. MEDA is stable and performs well for
±30% perturbed system showing comparable performances with SMC.
Chapter 3 Maximum Energy Dissipation Algorithm
65
Table 3.5 Evaluation criteria for ±7% stiffness perturbed system
under El Centro earthquake
SMC(Moon et al. 2003) MEDA
Evaluation
Criteria*
ε = 0 ε = ±7% ε = 0 ε = ±7%
J
1
0.394 0.432 0.331 0.395
J
2
1.130 1.323 1.108 1.347
J
3
0.296 0.335 0.255 0.278
J
4
0.560 0.54 0.464 0.443
J
5
0.213 0.224 0.185 0.219
J
6
0.870 0.862 0.709 0.692
J
7
0.218 0.235 0.234 0.233
J
8
0.887 0.901 0.883 0.891
J
9
0.189 0.198 0.233 0.215
J
10
0.551 0.556 0.552 0.547
J
11
0.016 0.017 0.020 0.02
* Evaluation criteria is same with Table 2
Chapter 3 Maximum Energy Dissipation Algorithm
66
Table 3.6 Evaluation criteria for ±30% stiffness perturbed system
El Centro Earthquake Mexico City Earthquake Gebze Earthquake
Evaluation
Criteria*
SMC
**
MEDA SMC
**
MEDA SMC
**
MEDA
J
1
0.422 0.431 0.56 0.761 0.453 0.577
J
2
1.563 1.494 1.384 1.59 1.737 1.744
J
3
0.334 0.36 0.559 0.557 0.481 0.478
J
4
0.626 0.489 0.531 0.359 1.266 1.0967
J
5
0.227 0.232 0.056 0.0879 0.176 0.191
J
6
0.945 0.788 1.69 0.63 2.827 1.478
J
7
0.311 0.31 0.497 0.544 0.387 0.393
J
8
1.045 1.061 1.177 1.334 1.78 1.797
J
9
0.242 0.264 0.392 0.513 0.506 0.498
J
10
0.577 0.469 0.76 0.564 1.944 2.131
J
11
0.019 0.0238 0.007 0.0127 0.021 0.0169
*
Evaluation criteria is same with Table 2
**Data from Moon et al. 2003
Chapter 3 Maximum Energy Dissipation Algorithm
67
3.4 Summary of Results
The applicability of the MEDAbased semiactive control system using MR
dampers in reducing structural responses for seismic loading conditions has been
demonstrated through a series of numerical studies of the benchmark problems. MEDA is
adopted to improve the design efficiency of controller without the deterioration of the
performance and the robustness.
For the benchmark cablestayed bridge, the numerical results show that MEDA can
reduce the vibration of the seismically excited cablestayed bridge structures effectively.
A comparison of results with two other controllers indicates that MEDA performs slightly
better than other two controllers.
For the nonlinear benchmark building, MEDA fails to reduce the peak floor
acceleration and the peak base shear. If the number and the location of MR dampers are
changed, the performance may be improved at the cost of losing an advantage of simple
method to use.
The robustness of MEDA is investigated with respect to the uncertainties in
stiffness for the benchmark cablestayed bridge. For the ±7% perturbed system, MEDA
shows more robust performance than SMC in the most evaluation criteria. Even for the
±30% stiffness perturbed, MEDA is stable and performs well showing comparable
performances with SMC.
Finally, we confirm that the MEDAbased semiactive control system using MR
dampers is efficient and robust for the benchmark cablestayed bridge, but is not
appropriate for the nonlinear benchmark building structure.
Chapter 4 Electromagnetic Induction System
68
CHAPTER 4
SMART PASSIVE CONTROL SYSTEM
4.1 Electromagnetic Induction System for MR damper
A prototype MR damper has been considered to show the schematic of MR
dampers, which was obtained for evaluation from the Lord Corporation and was used by
Dyke et al.(1996a). The damper is 21.5 cm long in its extended position, and the main
cylinder is 3.8 cm diameter. The main cylinder houses the piston, the electromagnet, an
accumulator and 50 ml of the MR fluid, and the damper has a ±2.5 cm stroke. As shown
in Fig. 1.3, the magnetic field produced in the device is generated by a small
electromagnet in the piston head. The current for the electromagnet is supplied by a
power supply such as a battery and regulated by a controller which determines control
commands, resulting in changes of damping characteristics of the MR fluid. Thus, to
reduce the structural responses, The MR damper needs a control system that consists of a
power supply, a controller, and sensors as shown in Fig. 4.1. Although the MR damper
based control system in Fig. 4.1 is simple, many MR dampers are used for civil
engineering structures such as cablestayed bridges and highrise buildings. In that case,
the MR damperbased control system becomes more complicated to build up and
maintain.
Figure 4.1 Schematic of a MR damperbased control system
Chapter 4 Electromagnetic Induction System
69
Thus, this dissertation proposes a ‘smart’ passive control system. The smart passive
control system is based on MR dampers. Of course, the MR damper is a semiactive
device that needs an external power source to change the damping characteristics of MR
fluids. However, the smart passive control system based on MR dampers is not using
external power source, but selfpowered by electromagnetic induction (EMI) system that
is attached to the MR damper. In this study, an EMI system is newly proposed for MR
dampers to replace a control system. Fig. 4.2 shows the MR damper with the EMI system
that consists of a permanent magnet and a coil. The EMI system changes kinetic energy
of reciprocation motion of the MR damper to the electric energy according to the
Faraday’s law of induction (Reitz et. al. 1993; Marshall and Skitek 1990; Miner 1996)
and then the electric energy is used to change the damping characteristics of the MR
damper. The characteristics of the MR fluid is affected by magnetic field. The magnetic
fields at coil 1 solidify the MR fluid resulting increase of damping capacity of the
damper. The magnetic field is arisen by induced current of the EMI system (consists of
permanent magnet and coil 2). Fast relative motions, between the permanent magnet and
coil 2, make high current at coil 1. Slow relative motions, between the permanent magnet
and coil 2, make low current at coil 1. Thus, the MR damper with the EMI system is able
to reduce the vibrations of structures by itself without any power supply and controller as
in Fig. 4.3. From the above explanation, we can call the MR damper with EMI system as
a ‘smart passive control system’ although the MR damper is semiactive device.
Chapter 4 Electromagnetic Induction System
70
Figure 4.2 Schematic of a MR damper with the EMI system
Figure 4.3 Schematic of a MR damper with the EMI system implementation
Bearing & Seal
MR Fluid
Coil 1 Diaphragm
Accumulator
Permanent
Magnet
Coil 2
EMI system
Chapter 4 Electromagnetic Induction System
71
Faraday’s law of induction is
dt
d
N
B
Φ
ε − = (4.1)
where ε is induced electromotive force (emf) that has unit of volt(V), N is number of
turns of coil, and Φ
B
is magnetic flux. Negative sign in (4.1) is the direction of induced
current. In (4.1), magnet flux can be defined
φ Φ cos ⋅ = ⋅ = BdA A d B d
B
r r
(4.2)
where B
r
is magnetic field, A
r
is area of cross section, and φ is the angle between B
r
and
A d
r
. Using (4.2), Faraday’s law can be rewritten
dt
dB
A N
dt
d
N
B
− = − =
Φ
ε (4.3)
Faraday’s law of induction states that the induced emf in a closed loop equals the
negative of the time rate of change of magnetic flux through the loop. External loads such
as earthquakes and winds cause the reciprocal motion of the MR damper. In consequence,
the coil in the EMI system at the end of the pistonaxle moves back and forth inducing
the emf. Thus, the faster MR damper moves, the higher emf is induced and the more
slowly MR damper moves, the lower emf is induced. This induced emf is carried to an
electromagnet in the piston head and generates magnetic field around the electromagnet
that changes the damping characteristics of the MR fluid.
To verify the feasibility of the proposed EMI system, we estimates the maximum
induced emf using Faraday’s law for the prototype MR damper in Fig. 1.3. Provided that
the permanent magnet of 1.2 Tesla is used in the EMI system, the maximum velocity of
reciprocal motion of the MR damper is 9cm/sec (that is about the maximum velocity of
uncontrolled case of following numerical example), and the turns of coil are 900, then the
Chapter 4 Electromagnetic Induction System
72
time rate of change of magnetic field during the full stroke of 5cm is 2.16 Tesla/sec. The
resulting emf induced in the coil is about 2.54V. Considering that saturation of the MR
effect begins in the prototype device when the applied voltage is 2.25V, the maximum
induced emf, 2,54V, is enough to change the damping characteristics of the MR fluid.
Besides, the amount of induced emf can be regulated by the turns of coil or the intensity
of permanent magnet.
The proposed smart passive control system does not need sensors that measure
structural responses for a controller, because the damping characteristics of the MR
damper is automatically regulated in proportion to the time rate of change of magnetic
flux. Also, the power for electromagnet in the piston head is supplied by induced emf of
the EMI system, which means there is no need of a power supply. Therefore, the
proposed smart passive control system has potential to replace the conventional MR
damperbased semiactive control system. This is the important benefit of using the smart
passive control system.
Chapter 4 Electromagnetic Induction System
73
4.2 Analytical Model and Design
4.2.1. Analytical Model
The performances of the smart passive control system are now evaluated through
the simulations. A model of a threestory building configured with a single MR damper is
considered here for direct comparisons with the MR damperbased semiactive system
controlled by the clippedoptimal controller, which is the exact one used by Dyke et al.
(1996a). The MR damper is rigidly connected between the ground and the first floor of
the structure. A schematic of the MR damper implementation is shown in Fig. 4.4.
Figure 4.4 Schematic of a MR damper implementation (Dyke et al. 1996a)
Chapter 4 Electromagnetic Induction System
74
The governing equations of the structure are given by
g
x& & & E f B Az z + + = (4.4)
where
g
x& & is a onedimensional ground acceleration, f is the measured force generated
between the structure and the MR damper, z is the state vector, and
(
¸
(
¸
− −
=
− −
C M K M
I
A
1 1
0
,
(
¸
(
¸
=
−
Γ M
B
1
0
,
(
¸
(
¸
− =
Λ
E
0
,
3 . 98 0 0
0 3 . 98 0
0 0 3 . 98
kg
(
(
(
¸
(
¸
= M
m
N sec
50 50 0
50 100 50
0 50 175
⋅
(
(
(
¸
(
¸
−
− −
−
= C (4.5)
,
84 . 6 84 . 6 0
84 . 6 7 . 13 84 . 6
0 84 . 6 0 . 12
10
5
m
N
(
(
(
¸
(
¸
−
− −
−
= K ,
0
0
1
(
(
(
¸
(
¸
−
= Γ
(
(
(
¸
(
¸
=
1
1
1
Λ
The simple mechanical model of the MR damper was shown in Fig. 4.5. The
equations governing the force f predicted by this model were given by Spencer et al.
(1997a) as follows:
) (
0 1 1
x x k y c f − + = &
) ( ) ( y x A  z  y x β  z  z  y x  γ z
n 1 n
& & & & & & & − + − − − − =
−
(4.6)
)} ( {
) (
y x k x c αz
c c
1
y
0 0
1 0
− + +
+
= & &
Chapter 4 Electromagnetic Induction System
75
Figure 4.5 Simple mechanical model of the normal MR damper
(Spencer et al. 1997a)
To account for the dependence of the force on the voltage applied and the resulting
magnetic current, Spencer, et al.(1997a) have suggested
u u
b a
α α α + = ) (
u c c (u) c
1b 1a 1
+ = (4.7)
u c c (u) c
0b 0a 0
+ =
where u is given as the output of a firstorder filter given by
v) η(u u − − = & (4.8)
and v is the commanded voltage sent to the current driver, which is emf induced by the
EMI system. (4.8) is necessary to model the dynamics.
c
0
x
F
k
1
BoucWen
c
0 k
0
Chapter 4 Electromagnetic Induction System
76
4.2.2 Design of the EMI System
According to the Faraday’s law of induction, the induced emf is proportional to the
turns of the coil and the time rate of change of magnetic flux. Thus, the amount of emf
can be regulated by the turns of the coil with a fixed capacity of the permanent magnet.
Appropriate number of coil turns needs to be determined in the design for better
performance of the smart passive control system. In the design of the EMI system for this
study, the influence of two parameters is considered: S
a
, the summation of peak
accelerations and S
i
, the summation of peak interdrift displacements at each floor, which
are normalized by uncontrolled responses, respectively. To determine the coil turns, the
maximum response approach (Park et al. 2003) is used for parameters S
a
and S
i
with three
earthquakes, El Centro, Hachinohe and Kobe.
Fig. 4.6(a) shows the variations of S
a
for each earthquake and Fig. 4.6(b) is the
envelope of the maximum responses of Fig. 4.6(a). From Fig. 4.6(b), we can determine
the optimal coil turns, which is the minimum point of the envelope denoted by the arrow.
Fig. 4.7 is similar to Fig. 4.7 except that it is for S
i
. Two appropriate coil turns, 2.16×10
4
and 2.6×10
4
(turns/m), are determined from the Figs. 4.6 and 4.7. Finally, an EMI system,
designed for S
a
, is designated EMI
ac
and the other EMI system, designed for S
i
, is called
EMI
dr
.
For the comparison of the performance, a normal MR damper system using the
clippedoptimal controller (Dyke et al. 1996a,b) is considered. This strategy is as follows.
First, an “ideal” active control device is assumed, and an appropriate primary controller
for this active device is designed. Then a secondary bangbang type controller causes the
MR fluid damper to generate the desired active control force, so long as this force is
dissipative.
Chapter 4 Electromagnetic Induction System
77
1 2 3 4 5 6 7 8 9 10
x 10
4
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
1 2 3 4 5 6 7 8 9 10
x 10
4
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
(a) Variations of S
a
(b) Envelope of max. responses
Figure 4.6 Design of EMI system with S
a
under three earthquakes
1 2 3 4 5 6 7 8 9 10
x 10
4
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
1 2 3 4 5 6 7 8 9 10
x 10
4
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
(a) Variations of Si (b) Envelope of max. responses
Figure 4.7 Design of EMI system with S
i
under three earthquakes
Coil turns/m Coil turns/m
S
a
Hachinohe
Kobe
El Centro
Coil turns/m Coil turns/m
S
i
Hachinohe
Kobe
El Centro
Chapter 4 Electromagnetic Induction System
78
In this study, an H
2
/LQG control design is adopted as the primary controller. The
ground excitation is taken to be a stationary white noise, and an infinite horizon
performance index is chosen that weights appropriate parameters of the structure, i.e.,
(
¸
(
¸
+ =
∫
∞ →
τ
τ
τ
0
} {
1
lim dt E J Rf f Qz z
T T
(4.9)
where R is an identity matrix, and Q is the response weighting matrix. The weighting
parameters are as in (4.10).
(
¸
(
¸
=
×
×
3 3
3 3
0
0
I
I
Q
i
a
q
q
(4.10)
where q
a
and q
i
weight the accelerations and interstory drifts at each floor, respectively.
To determine the weighting parameters q
a
and q
i
, the maximum response approach is
used as in the EMI system for three earthquakes.
Fig. 4.8(a) shows the variations of S
a
for increasing weighting parameters in 3
dimensional plot. Each surface corresponds to the variation of S
a
for each earthquake.
Fig. 4.8(b) is the envelope of the maximum response of Fig. 4.8(a). From Fig. 4.8(b), we
can determine the appropriate weighting parameters, which is the minimum point of the
envelope denoted by the arrow. Fig. 4.9 is similar to Fig. 4.9 except that it is for S
i
. Two
sets of the appropriate weighting parameter, q
a
= 5.0×10
13
, q
i
= 1.0×10
5
for S
a
and
q
a
=5.0×10
15
, q
i
=5.0×10
6
for S
i
, are determined from the Figs. 4.8 and 4.9. The clipped
optimal controller, designed for S
a
, with weighting parameters q
a
and q
i
, is designated
CO
ac
and the other, designed for S
i
, is called CO
dr
.
Chapter 4 Electromagnetic Induction System
79
(a) Variations of S
a
(b) Envelope of max. responses
Figure 4.8 Design of the clippedoptimal controller with S
a
under three earthquakes
(a) Variations of Si (b) Envelope of max. responses
Figure 4.9 Design of the clippedoptimal controller with S
i
under three earthquakes
q
a
S
a
q
i
q
a
q
i
q
a
S
i
q
i
q
a
q
i
Chapter 4 Electromagnetic Induction System
80
4.3 Numerical Simulation Results
To verify the effectiveness of the proposed smart passive control system, a set of
simulations is performed for the four historical earthquakes: El Centro, Hachinohe, Kobe,
and Northridge earthquakes. The Northridge earthquake is not considered in the
designing phase, but is included here to check the validation of the design of the EMI
system and the clippedoptimal controller. Simulation results of the proposed smart
passive control system are compared to those of the MR damperbased semiactive control
system using the clippedoptimal controller by the evaluation criteria based on those used
in the second generation linear control problem for buildings (Spencer et al., 1997b). The
first evaluation criterion is a measure of the normalized peak floor accelerations, given by


.

\

=
max
,
1
 ) (
max
a
ai
i t
x
t x
J
& &
& & 
(4.11)
where the absolute accelerations of the i th floor, ) (t x
ai
& & , are normalized by the peak
uncontrolled floor acceleration, denoted ) (
max
t
a
x& & .
The second evaluation criterion is a measure of the reduction in the interstory drift.
The maximum of the normalized interstory drift is


.

\

=
max
n
i
 
d
t d
J
i t
) (
max
,
2
(4.12)
where d
i
(t) is the interstory drift of the above ground floors over the response history, and
max
n
d denotes the normalized peak interstory drift in the uncontrolled response.
Chapter 4 Electromagnetic Induction System
81
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
20
15
10
5
0
5
10
15
20
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
1
0.5
0
0.5
1
1.5
2
2.5
3
(a) El Centro earthquake
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
20
15
10
5
0
5
10
15
20
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
1
0.5
0
0.5
1
1.5
2
2.5
3
(b) Hachinohe earthquake
Figure 4.10 Velocities and induced voltages under various earthquakes
Time(sec)
V
e
l
o
c
i
t
y
(
c
m
/
s
e
c
)
Time(sec)
V
o
l
t
a
g
e
(
V
)
Time(sec)
V
e
l
o
c
i
t
y
(
c
m
/
s
e
c
)
Time(sec)
V
o
l
t
a
g
e
(
V
)
Chapter 4 Electromagnetic Induction System
82
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
20
15
10
5
0
5
10
15
20
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
1
0.5
0
0.5
1
1.5
2
2.5
3
(c) Kobe earthquake
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
20
15
10
5
0
5
10
15
20
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
1
0.5
0
0.5
1
1.5
2
2.5
3
(d) Northridge earthquake
Figure 4.10 Velocities and induced voltages under various earthquakes (continued)
Time(sec)
V
e
l
o
c
i
t
y
(
c
m
/
s
e
c
)
Time(sec)
V
o
l
t
a
g
e
(
V
)
Time(sec)
V
e
l
o
c
i
t
y
(
c
m
/
s
e
c
)
Time(sec)
V
o
l
t
a
g
e
(
V
)
Chapter 4 Electromagnetic Induction System
83
Representative responses of the EMI system to four earthquakes are shown in
graphs. Fig. 4.10 shows the velocities at the first floor where the MR damper is attached
and the induced voltages by the EMI system for each earthquake. For moderate
earthquakes (El Centro and Hachinohe), the velocity of the first floor is smaller than that
of severe earthquakes (Kobe and Northridge) with the consequence that the induced
voltage by the EMI system is lower according to the Faraday’s law of induction. Also, it
can be seen that the higher voltage is induced for severe earthquakes. The maximum
induced voltage is 1.6V, 0.9V, 2.25V, and 2.25V for El Centro, Hachinohe, Kobe, and
Northridge earthquakes, respectively, which is the voltage enough to operate the MR
damper. Besides, it should be noted that the induced voltage is restricted within 2.25V for
the capacity of the MR damper, which is identical condition with the clippedoptimal
controller. However, the induced voltage is continuously varying whereas the command
voltage of the clippedoptimal controller takes on values of either zero or the maximum
value.
0 0.5 1 0 0.5 1
Figure 4.11 Normalized peak acceleration and interstory drift
El Centro
Hachinohe
Kobe
Northridge
Peak acceleration Peak drift
CO
dr
CO
ac
EMI
dr
EMI
ac
Chapter 4 Electromagnetic Induction System
84
Fig. 4.11 shows the values for the peak acceleration and peak interstory drift
normalized by uncontrolled responses, where it can be seen that the EMI system performs
well over the entire suite of earthquakes considered. The reductions in peak acceleration
and drift are comparable to those of the clippedoptimal controller – both giving as much
as nearly 50% decreased compared to the uncontrolled responses, except the Northridge
earthquake. Although the EMI system and the clippedoptimal controller are designed
under the three earthquakes except the Northridge earthquake, both achieve reductions in
the peak acceleration and interstory drift for the Northridge earthquake.
Table 4.1 shows the accelerations and the interstory drifts at each floor for four
cases of two categories (i.e., EMI
ac
, EMI
dr
, CO
ac
, and CO
dr
) normalized by each
uncontrolled response, respectively. Table 4.2 reports the percent response reduction (−)
or increase (+) compared to the better clippedoptimal controller. In Tables 4.1 and 4.2,
the colored cells are the minimum value among four cases at each floor. In the Table 4.2,
the clippedoptimal controllers achieve more reductions over the EMI systems for the
moderate earthquakes such as El Centro and Hachinohe, except that the EMI systems
give minimum value at the first floor. For the severe earthquakes such as Kobe and
Northridge, however, the performances of the EMI system are better than those of the
clippedoptimal controller giving up to 35.5% and 24.1% additional maximum decreases
in the peak acceleration and interstory drift, respectively, compared to the better clipped
optimal controller. Though the EMI system fails to achieve more reductions over the
clippedoptimal controller for the moderate earthquakes, it has comparable performance
to the clippedoptimal controller without the power source, controller, and sensors. This
is the important benefit of using the smart passive control system.
Chapter 4 Electromagnetic Induction System
85
Table 4.1 Normalized peak absolute accelerations and interstory drifts
Accelerations
El Centro (0.3495)* Hachinohe (0.2294) Story
CO
ac
CO
dr
EMI
ac
EMI
dr
CO
ac
CO
dr
EMI
ac
EMI
dr
1
st
0.499 0.551 0.355 0.340 0.492 0.515 0.372 0.377
2
nd
0.354 0.433 0.436 0.396 0.431 0.520 0.526 0.530
3
rd
0.441 0.473 0.512 0.492 0.384 0.465 0.404 0.423
Kobe (0.8337) Northridge (0.8428)
Story
CO
ac
CO
dr
EMI
ac
EMI
dr
CO
ac
CO
dr
EMI
ac
EMI
dr
1
st
0.370 0.429 0.367 0.345 0.897 0.881 0.568 0.568
2
nd
0.494 0.493 0.484 0.485 0.587 0.554 0.612 0.586
3
rd
0.410 0.384 0.387 0.393 0.815 0.800 0.725 0.738
Interstory drifts
El Centro (0.3495) Hachinohe (0.2294) Story
CO
ac
CO
dr
EMI
ac
EMI
dr
CO
ac
CO
dr
EMI
ac
EMI
dr
1
st
0.228 0.212 0.168 0.180 0.295 0.243 0.178 0.194
2
nd
0.423 0.448 0.476 0.457 0.289 0.319 0.357 0.355
3
rd
0.441 0.473 0.512 0.492 0.384 0.465 0.404 0.423
Kobe (0.8337) Northridge (0.8428)
Story
CO
ac
CO
dr
EMI
ac
EMI
dr
CO
ac
CO
dr
EMI
ac
EMI
dr
1
st
0.348 0.308 0.293 0.301 0.563 0.473 0.359 0.382
2
nd
0.456 0.442 0.428 0.435 0.859 0.846 0.827 0.835
3
rd
0.410 0.384 0.387 0.393 0.815 0.800 0.725 0.738
* ( ) is peak ground acceleration (g)
Chapter 4 Electromagnetic Induction System
86
Table 4.2 Percent increment compared to the better clippedoptimal controller case
Accelerations
El Centro (0.3495) Hachinohe (0.2294) Story
CO
ac
CO
dr
EMI
ac
EMI
dr
CO
ac
CO
dr
EMI
ac
EMI
dr
1
st
0 10.4
28.8*
31.9 0 4.7 24.4 23.4
2
nd
0 22.1 23.1 11.7 0 20.5 22.0 22.9
3
rd
0 7.3 16.0 11.5 0 21.4 5.4 10.4
Kobe (0.8337) Northridge (0.8428)
Story
CO
ac
CO
dr
EMI
ac
EMI
dr
CO
ac
CO
dr
EMI
ac
EMI
dr
1
st
0 14.2 0.7 6.6 1.8 0 35.5 35.5
2
nd
0.16 0 1.8 1.5 6.0 0 10.5 5.8
3
rd
6.8 0 0.9 2.4 1.9 0 9.4 7.8
Interstory drifts
El Centro (0.3495) Hachinohe (0.2294) Story
CO
ac
CO
dr
EMI
ac
EMI
dr
CO
ac
CO
dr
EMI
ac
EMI
dr
1
st
7.2 0 20.6 15.3 21.5 0 26.7 20.4
2
nd
0 5.9 12.4 8.1 0 10.4 23.5 22.8
3
rd
0 7.3 16.0 11.5 0 21.4 5.4 10.4
Kobe (0.8337) Northridge (0.8428)
Story
CO
ac
CO
dr
EMI
ac
EMI
dr
CO
ac
CO
dr
EMI
ac
EMI
dr
1
st
12.8 0 5.0 2.5 19.0 0 24.1 19.2
2
nd
3.2 0 3.2 1.6 1.5 0 2.2 1.3
3
rd
6.8 0 0.9 2.4 1.9 0 9.4 7.8
* minus sign means reduction
Chapter 4 Electromagnetic Induction System
87
4.4 Summary of Results
This study has proposed a smart passive control system for a civil engineering
application. The smart passive control system is based on the MR damper with EMI
system. The EMI system consists of a permanent magnet and a coil. According to the
Faraday’s law of induction, the EMI system generates induced voltages that can supply
electricity and control commands to the MR damper, replacing a normal control system
such as a power supply, a controller, and sensors.
To investigate the achievable capabilities of the smart passive control system, two
EMI systems were designed. Then, the effectiveness of performances are evaluated, and
compared with those of the semiactive MR damper system using clippedoptimal
controller. In comparing both systems, it was observed that for the moderate earthquake
such as El Centro and Hachinohe, the smart passive control system showed the
comparable performance to the MR damperbased system controlled by the clipped
optimal controller. For the severe earthquakes such as Kobe and Northridge, the smart
passive control system shows the better performance giving up to 35.5% and 24.1%
additional maximum decreases in the peak acceleration and interstory drift, respectively.
In addition to the comparable performance, the proposed smart passive control
system has the simple structure without any power supply, controller, and sensors.
Therefore, the proposed smart passive control system has potential to be implemented in
real civil structures.
Chapter 5 Conclusions
88
CHAPTER 5
CONCLUSIONS
The study proposes the implementation of simple and efficient control algorithms
for seismically excited structures using MR dampers and a smart passive control system
based on MR dampers
First, the characteristics of the implemented simple and effective control
algorithms are summarized as follows by the analytical and numerical examples:
(1) Modal control scheme with a lowpass filter and modal estimator of Kalman filter,
was implemented to seismically excited structures using MR dampers, resulting in
reducing the size of weighting matrix Q.
The motion of the structure was effectively suppressed by merely controlling a few
lowest modes, although resulting responses varied greatly depending on the choice
of measurements available and weightings.
(2) Maximum energy dissipation algorithm (MEDA) was adopted to improve the
design efficiency of controller without the deterioration of the performance and
robustness.
For the benchmark cablestayed bridge, MEDA can reduce the vibration of the
seismically excited cablestayed bridge structures effectively. For the nonlinear
benchmark building, MEDA fails to reduce the peak floor acceleration and the peak
base shear.
The robustness of MEDA was investigated with respect to the uncertainties in
stiffness for the benchmark cablestayed bridge. For the ±7% and 30% perturbed
Chapter 5 Conclusions
89
system, MEDA shows comparable performances to SMC in the most evaluation
criteria.
And, the characteristics of the proposed smart passive control system are
summarized as follows by the analytical and numerical examples:
(1) The EMI system that consists of a permanent magnet and a coil, generates induced
voltages to supply electricity and control commands for MR dampers, replacing a
normal control system such as a power supply, a controller, and sensors.
(2) In comparing with MR damperbased semiactive control system using the clipped
optimal controller, the smart passive control system shows comparable
performances for the moderate earthquake such as El Centro and Hachinohe. For
the severe earthquakes such as Kobe and Northridge, the smart passive control
system shows better performances giving 35.5% and 24.1% additional maximum
decreases in the peak acceleration and interstory drift, respectively.
요약문
90
요 약 문
MR댐퍼를 위한 간편한 알고리즘과 스마트 수동제어 시스템
본 논문은 MR댐퍼를 기반으로 하는 제어시스템을 대형 구조물등에 설치할
수 있도록 간편한 알고리즘을 구현하였다. 아울러, MR댐퍼를 기반으로 하는 스마
트 수동제어 시스템을 제안하였다.
MR댐퍼는 최근에 각광 받는 반능동 제어장치로써, 배터리와 같은 소규모 전
력을 사용하고, 이에 따른 비상시의 신뢰성, 그리고 상대적으로 저렴한 제작단가 등
의 장점들이 있다. 이러한 MR댐퍼를 운용하기 위한 많은 알고리즘들이 제안되었으
나, 각각의 성능이 비슷함에도 불구하고, 제어기 설계가 복잡한 문제 등이 발생했
다. 따라서 본 논문에서는 MR댐퍼의 반능동 제어장치로써의 비선형성을 고려하는
동시에, 제어기 설계가 간단하고, 그 성능 역시 기존 알고리즘에 뒤지지 않는 제어
알고리즘으로써, 모드제어기법과 Maximum Energy Dissipation Algorithm (MEDA)
을 연구하였다.
모드제어는 선택된 소수의 저차 모드만을 제어한다. 따라서 제어기의 크기가
축소되어 설계가 용이하다. 모드제어가 수년간 연구되었음에도 불구하고 반능동
제어장치, 특히 MR댐퍼에 대한 가능성은 아직 검증되지 않았다. 따라서 본 논문에
서는 지진하중을 받는 구조물의 MR댐퍼 제어를 위해서, 모드제어를 구현하고 그
성능을 평가하였다. 또한 현실상황과 적합하도록, Kalman필터를 상태관측기로 사
용하여 모드 좌표계를 관측하였으며, Lowpass필터를 사용하여 Spillover문제를 해
결하였다. 또한 Kalman필터를 상태관측기로 사용하는데 있어서, 변위, 속도, 그리
고 가속도 피드백을 각각 사용하여, 그 성능을 비교하였다. 이렇게 구현된 모드제어
를 6자유도 구조물에 적용하여, 기존 제어 알고리즘의 성능과 비교하였다.
요약문
91
또한 제어기의 설계에 Lyapunov직접법을 사용하는 MEDA 역시, MR댐퍼를
사용하는 규모가 큰 토목구조물에 대한 그 적용성 및 가능성이 연구되지 않았다.
Lyapunov직접법에 의해 유도된 MEDA는 제어장치가 설치된 곳의 국부변위 등에
의해 제어력을 결정하므로, 특별한 제어기 설계가 필요하지 않다. 따라서 제어기 설
계가 어려운 대형 토목구조물에 적합하다. 따라서 본 논문에서는 MEDA의 대규모
토목구조물에 대한 적용성을 성능과 강인성 면에서 평가하였다. 이를 위해서 수치
예제로써 사장교 및 20층 짜리 비선형 빌딩의 벤치마크문제를 이용하였으며, 성능
및 강인성을 기존 연구의 결과들과 비교하였다.
한편, 기존 MR댐퍼의 효율적인 운용을 위한 연구와 아울러, MR댐퍼를 기반
으로 하는 스마트 제어시스템을 제안하였다. 기존의 MR댐퍼를 사용하여 구조물을
제어하기 위해서는, MR댐퍼를 위한 외부전원, 진동의 정도를 측정할 수 있는 센서,
그리고 MR댐퍼의 거동을 결정하는 제어기가 기본적으로 필요하다. 이러한 MR댐
퍼가 토목구조물과 같은 경우에 대량으로 상용되면, 그 제어 시스템은 설치 및 유지
관리가 어렵다.
따라서 본 논문에서는 MR댐퍼를 기반으로 하는 스마트 수동제어 시스템을
제안하였다. 스마트 수동제어 시스템은 MR댐퍼와 EMI (Electromagnetic Induction)
시스템으로 구성한다. MR댐퍼에 필요한 전원은, 외부전원 대신에 MR댐퍼에 부착
된 EMI 시스템에서 유도되는 유도전류를 사용한다. 즉, MR댐퍼의 왕복운동에너지
를, EMI 시스템에서 Faraday의 전자기 유도법칙에 의해 전기에너지로 전환하고, 이
를 MR댐퍼의 점성특성을 변화시키는데 사용한다. 또한 유도되는 전류의 양이 MR
댐퍼의 왕복운동 강도에 비례하므로, 센서 및 제어기가 필요하지 않다. 따라서 반능
동 제어장치인 MR댐퍼를 기반으로 하지만, 외부전력을 사용하지 않고, 장치 스스
로 외부하중의 강도에 맞추어 점성을 변화시키는 ‘스마트’ 수동제어 시스템이다.
요약문
92
제안된 스마트 수동제어 시스템의 성능평가를 위해서, Clippedoptimal 제어기로 제
어되는 MR댐퍼의 성능을 수치예제를 통해 비교하였다.
References
93
REFERENCES
1. ABAQUS (1998). Hibbitt, Karlsson & Sorensen Inc. Pawtucket, RI.
2. Balas, M.J. (1978). “Feedback control of flexible system,” IEEE Transactions on
Automatic Control, 23(4), 674679.
3. Barrosa, L. R., (1999). “Performance evaluation of vibration controlled steel
structures under seismic loading”, Ph.D. thesis of Stanford University.
4. Brogan, W.L. (1991). Modern Control Theory. Prentice Hall, Englewood Cliffs,
New Jersey.
5. Burton, S.A., Makris, N., Konstantopoulos, I., and Antsaklis, P.J. (1996).
“Modeling the response of ER damper: phenomenology and emulation.” J. Engrg.
Mech., 122, 897–906.
6. Carlson, J.D. (1994). “The Promise of Controllable Fluids.” Proc. of Actuator 94
(H. Borgmann and K. Lenz, Eds.), AXON Technologie Consult GmbH, 266–270.
7. Carlson, J.D. and Weiss, K.D. (1994). “A Growing Attraction to Magnetic Fluids.”
Machine Design, August, 61–64.
8. Carlson, J.D., Catanzarite, D.M. and St. Clair, K.A. (1995). “Commercial
MagnetoRheologicalFluid Devices.” Proc. of the 5th International Conference on
ER Fluids, MR Fluids and Associated Technology, U. Sheffield, UK.
9. Carlson, J.D., Catanzarite, D.N. and St Clair, K.A. (1996). “Commercial Magneto
Rheological Fluid Devices.” Proc. 5th Int. Conf. on ER Fluids, MR Suspensions
and Associated Tech., W. Bullough, Ed., World Scientific, Singapore, 20–28.
10. Carlson, J.D., and Spencer Jr., B.F. (1996a). “Magnetorheological fluid dampers:
scalability and design issues for application to dynamic hazard mitigation.” Proc.
2nd Workshop on Structural Control: Next Generation of Intelligent Structures,
Hong Kong, China, 99–109.
References
94
11. Carlson, J.D., and Spencer Jr., B.F. (1996b). “Magnetorheological fluid dampers
for semiactive seismic control.” Proc. 3rd International Conference on Motion
and Vibration Control, Chiba, Japan, 3, 35–40.
12. Chang, C.C., and Roschke, P. (1998). “Neural network modeling of a
magnetorheological damper.” J. Intelligent Material Systems and Structures, 9,
755–764.
13. Cook, R.D., Malkus, D.S., Plesha, M.E. (1989). Concepts and Applications of
Finite Element Analysis, John Wiley & Sons, New York.
14. Dyke, S.J., Spencer Jr., B.F., Jr.,Sain, M.K., and Carlson, J.D. (1996a). “Modeling
and control of magnetorheological dampers for seismic response reduction.” Smart
Mat. and Struct., 5, 565–575.
15. Dyke, S.J., Spencer Jr., B.F., Sain, M.K., and Carlson, J.D. (1996b). “Seismic
response reduction using magnetorheological dampers.” Proc. IFAC World Cong.,
Int. Fed. of Automatic Control, L. 145–150.
16. Dyke, S.J., Spencer Jr., B.F., Quast, P., Sain, M.K., Kaspari Jr., D.C. and Soong,
T.T. (1996c). “Acceleration Feedback Control of MDOF Structures.” J. of Engrg.
Mech., ASCE, 122(9), 907–918.
17. Dyke, S.J., Spencer Jr., B.F., Sain, M.K., and Carlson, J.D. (1998). “An
experimental study of MR dampers for seismic protection.” Smart. Mat. and
Struct., 5, 693–703.
18. Dyke, S.J., Caicedo, J.M., Turan, G., Bergman, L.A., and Hague, S. (2003). “Phase
1 Benchmark control problem for seismic response of cablestayed bridges.” J. of
Structural engineering, 129(7), 857–872.
19. Ehrgott, R., and Masri, S.F. (1992). “Modeling the oscillatory dynamic behavior of
electrorheological materials in shear.” Smart Mat. and Struct., 1, 275–285.
20. Faravelli, L., and Spencer, B. F., Jr., eds. (2003). Proc., Sensors and Smart
Structures Technology, Wiley, New York.
References
95
21. Feng, Q., and Shinozuka, M. (1990). “Use of a variable Damper for Hybrid Control
of Bridge Respnse Under Earthquake.” Proc. , U.S. Nat. Workshop on Struct.
Control Res., USC Publ. No. CE9013.
22. Fujitani, H., Sodeyama, H., Hata, K., Iwata, N., Komatsu, Y., Sunakoda, K., and
Soda, S. (2000). “Dynamic performance evaluation of magnetorheological
damper.” Proc. Int. Conf. on Adv. in Struc. Dyn., v1, Hong Kong, China, 319–326.
23. Gamoto, D.R., and Filisko, F.E. (1991). “Dynamic mechanical studies of
electrorheological materials: moderate frequencies.” J. Rheology, 35(3), 399425.
24. Gavin, H.P., Hanson, R.D., and Filisko, F.E. (1996a). “Electrorheological dampers,
part 1: analysis and design.” J. Applied Mech., ASME, 63, 669–675.
25. Gavin, H.P., Hanson, R.D., and Filisko, F.E. (1996b). “Electrorheological dampers,
part 2: testing and modeling.” J. Applied Mech., ASME, 63, 676–682.
26. Hague, S. (1997). “Composite Design for Long Span Bridges.” Proceedings of the
XV ASCE Structures Congress, Portland, Oregon.
27. Housner, G. W., Masri, S. F., and Chassiakos, A. G., eds. (1994). Proc., 1st World
Conf. On Structural Control.
28. Housner, G. W., et al. (1997). “Structural control: Past, present and future.” J. Eng.
Mech., 123(9), 897971.
29. Inaudi, J.A. (1997). “Modulated Homogeneous Friction: A Semiactive Damping
Strategy.” Earthquake Engineering and Structural Dynamics, 26(3), 361.
30. Jansen, L.M., and Dyke, S.J. (2000). “Semiactive control strategies for MR
dampers: comparative study.” J. Engrg. Mech., ASCE, 126(8), 795–803.
31. Johnson, E.A., Baker, G.A., Spencer Jr., B.F., and Fujino, Y. (2001a). “Semiactive
Damping of Stay Cables.” J. Engrg. Mech., ASCE, accepted.
32. Johnson, E.A., Christenson R.E., and Spencer Jr., B.F. (2001b). “Semiactive
damping of cables with sag.” ComputerAided Civil Infrastructure Engineering,
accepted.
References
96
33. Jolly, M.R., Bender, J.W., and Carlson, J.D. (1998). “Properties and Applications
of Commercial Magnetorheological Fluids.” Proc. SPIE 5th Annual Int.
Symposium on Smart Structures and Materials, San Diego, CA.
34. Kamath, G.M., Hurt, M.K., and Wereley, N.M. (1996). “Analysis and testing of
Bingham plastic behavior in semiactive electrorheological fluid dampers.” Smart
Mat. and Struct., 5, 576–590.
35. Kamath, G.M., and Wereley, N.M. (1997). “Nonlinear viscoelasticplastic
mechanismsbased model of an electrorheological damper.” J. Guidance, Control,
and Dynamics, 20(6), 1125–1132.
36. Kareem, A., Kijewski, T., and Tamura, Y. (1999). “Mitigation of motions of tall
buildings with specific examples of recent application.” Wind Struct., 2(3), 201
251.
37. Karnopp, D.C., Crosby, M.J., and Harwood, R.A. (1974). “Vibration Control Using
SemiActive Force Generators.” ASME Journal of Engineering for Industry, 96(2),
May.
38. Kobori, T., Koshika, N., Yamada, N., and Ikeda, Y. (1991). “Seismic response
controlled structure with active mass driver system. Part 1: Design.” Earthquake
Eng. Struct. Dyn., 20, 133139.
39. Kobori, T., Inou, Y., Seto, K., Iemura, H., and Nishitani, A., eds. (1998). Proc.,
2nd World Conf. On Structural Control, Wiley, New York.
40. Kobori, T. (2003). “Past, present and future in seismic response control in civil
engineering structures.” Proc., 3rd World Conf. On Structural Control, Wiley,
New York, 914
41. Leitmann, G. (1994). “Semiactive Control for Vibration Attenuation.” J. of
Intelligent Material Systems and Structures, September, 5, 841–846.
42. Luenberger, D.G. (1971). “An Introduction to observers,” IEEE Transactions on
Automatic Control, AC16(6), 596602.
References
97
43. Markis, N., Burton, S.A., Hill, D., and Jordan, M. (1996). “Analysis and design of
ER damper for seismic protection of structures.” J. Engrg. Mech., 122, 1003–1011.
44. Marshall, S.V. and Skitek, G. G. (1990). Electromagnetic concepts and
applications, PrenticeHall.
45. MATLAB (1997). The Math Works, Inc. Natick, Massachusetts.
46. McClamroch, N.H. and Gavin, H.P. (1995). “Closed Loop Structural Control
Using Electrorheological Dampers.” Proc. of the Amer. Ctrl. Conf., Seattle,
Washington, 4173–4177.
47. Meirovitch, L. (1967). Analytical Methods in Vibrations. Macmillan Publishing.
48. Meirovitch, L. and Baruh, H. (1983) “On the Problem of Observation Spillover in
SelfAdjoint DistributedParameter Systems.” Journal of Optimization Theory and
applications, 39(2), 269291.
49. Meirovitch, L. (1990). Dynamics and Control of Structures. John Wiley and Sons,
Inc., New York, N. Y.
50. Miner, G.F. (1996). Lines and electromagnetic fields for engineers, Oxford
University Press.
51. Moon, S. J., Bergman, L. A., and Voulgaris, P. G. (2003). “Sliding Mode Control
of CableStayed Bridge Subjected to Seismic Excitation.” Journal of Engineering
Mechanics, ASCE, 129(1), 7178.
52. Nishitani, A., and Inoue, Y. (2001). “Overview of the application of active/
semiactive control to building structures in Japan.” Earthquake Eng. Struct. Dyn.,
30, 15651574.
53. Ohtori, Y., Christenson, R.E., Spencer, Jr. B.F., and Dyke, S.J. (2000). “Summary
of benchmark control problems for seismically excited nonlinear buildings.” Proc.
Of the 2000 Engineering Mechanics Conf., ASCE, Austin, Texas, May 2124.
54. Ohtori, Y., Christenson, R.E., Spencer, Jr. B.F., and Dyke, S.J. (2002).
“Benchmark control problems for seismically excited nonlinear buildings.” Journal
References
98
of Engineering Mechanics: Special Issue on Structural Control Benchmark
Problems, ASCE(submitted).
55. Park, K.S., Jung, H.J and Lee, I.W. (2003). "Hybrid Control Strategy for Seismic
Protection of a Benchmark CableStayed Bridge." Engineering Structures, 25(4),
405417.
56. Phillips, R.W. (1969). “Engineering applications of fluids with a variable yield
stress.” Ph.D thesis, University of California, Berkeley, California.
57. Rabinow, J. (1948). “The magnetic fluid clutch.” AIEE Transactions, 67, 1308–
1315.
58. Rabinow, J. (1951). “Magnetic fluid torque and force transmitting device.” U.S.
Patent 2,575,360.
59. Ramallo, J.C., Johnson, E.A., and Spencer Jr., B.F. (2001). “‘Smart’ base isolation
systems.” J. Engrg. Mech., ASCE, submitted.
60. Reitz, J.R. Milford, F.J., and Christy R.W. (1993). Foundations of electromagnetic
theory, AddisonWesley Pub. Co.
61. Sack, R.L., (1989). Matrix Structural Analysis, PWSKent Pub. Co., Boston.
62. Sack, R.L., Kuo, C.C., Wu, H.C., Liu, L. and Patten, W.N. (1994). “Seismic
Motion Control via Semiactive Hydraulic Actuators.” Proc. of the U.S. Fifth
National Conf. on Earth. Eng., Chicago, Illinois, 2, 311–320.
63. Sack, R.L. and Patten, W. (1994). “Semiactive Hydraulic Structural Control.”
Proc. of the Int. Workshop on Structural Control, USC Publication Number CE
9311,. 417–431.
64. Sodeyama, H., Sunakoda, K., Fujitani, H., Soda, S., Iwata, N., and Hata, H. (2003).
“Dynamic tests and simulation of magnetorheological dampers.” Comput. Aided
Civ. Infrastruct. Eng., 18, 4557.
65. Soong, T. T. (1990). Active structural control: Theory and practice, Longman
Scientific, Essex, U.K.
References
99
66. Soong, T. T. and Reinhorn, A. M. (1993). “An overview of active and hybrid
structural control research in the U.S.” Struct. Des. Tall Build., 2, 192209.
67. Soong, T. T. and Spencer, B. F., Jr.. (2002). “Supplementary energy dissipation:
Stateoftheart and stateofthepractice.” Eng. Struct., 24, 243259.
68. Spencer Jr., B.F., Jr. (1996). “Recent Trends in Vibration Control in the U.S.A.”
Proc. 3rd Int. Conf. On Motion and Vibr. Control, Chiba, Japan.,Vol. II, K1K6.
69. Spencer Jr., B.F., Timlin, T.L., Sain, M.K., and Dyke, S.J. (1996a). “Series
solution of a class of nonlinear optimal regulators.” J. Optimization Theory and
Applications, 91, 321–345.
70. Spencer Jr., B.F., Dyke, S.J., and Sain, M.K. (1996b). “Magnetorheological
dampers: a new approach to seismic protection of structures.” Proc. Conf. on
Decision and Control, 676–681.
71. Spencer, B. F., Jr., and Sain, M. K. (1997). “Controlling buildings: A newfrontier
in feedback.” IEEE Control Syst. Mag., 17(6), 1935.
72. Spencer Jr., B.F., Dyke, S.J., Sain, M.K., and Carlson, J.D. (1997a).
“Phenomenological model of a magnetorheological damper.” J. Engrg. Mech.,
ASCE, 123, 230–238.
73. Spencer Jr., B.F., Carlson, J.D., Sain, M.K., and Yang, G. (1997b). “On the current
status of magnetorheological dampers: seismic protection of fullscale structures.”
Proc. American Control Conf., 458–462.
74. Spencer Jr., B.F., Yang, G., Carlson, J.D., and Sain, M.K. (1998). “‘Smart’
dampers for seismic protection of structures: a fullscale study.” Proc. 2nd World
Conf. on Struct. Control, Kyoto, Japan, 417–426.
75. Spencer, B.F., Jr., Dyke, S.J. and Deoskar, H.S. (1998a,b). “Benchmark Problems
in Structural Control – Part I: Active Mass Driver System; Part II: Active Tendon
System.” Earthquake Engneering and Structural Dynamics, 27(11), 1127–2247
References
100
76. Spencer, B. F., Jr., Yang, G., Carlson, J. D., and Sain, M. K. (1999). “ ‘Smart’
dampers for seismic protection of structures: A fullscale study.” Proc., 2
nd
World
Conf. Structural Control, Wiley, New York, 1, 417426.
77. Spencer Jr., B.F., Johnson, E.A., and Ramallo, J.C. (2000). “'Smart' isolation for
seismic control.” JSME International Journal: Special Issue on Frontiers of
Motion and Vibration Control, Series C, 43(3), 704–711.
78. Spencer, B. F., Jr., (2002). “Civil engineering applications of smart damping
technology.” Proc., 5th Int. Conf. On Vibration Engineering, Nanjing, China, 771
782.
79. Spencer, B. F., Jr., and Nagarajaiah, S. (2003). “State of the art of structural
control.” J. Struct. Eng., 129(7), 845856.
80. Stanway, R., Sproston, J.L., and Stevens, N.G. (1987). “Nonlinear modeling of an
electrorheological vibration damper.” J. Electrostatics, 20, 167–184.
81. Sunakoda, K., Sodeyama, H., Iwata, N., Fujitani, H., and Soda, S. (2000).
“Dynamic characteristics of magnetorheological fluid damper.” Proc. SPIE Smart
Struc. and Mat. Conf., Newport Beach, California, 3989, 194–203.
82. Symans, M. D., and Constantinou, M. C. (1999). “Semiactive control systems for
seismic protection of structures: A stateoftheart review.” Eng. Struct., 21(6),
469487.
83. Wereley, N.M., Pang, L., and Kamath, G.M. (1998). “Idealized hysteresis
modeling of electrorheological and magnetorheological dampers.” J. Intelligent
Mat., Systems and Struct., 9, 642–649.
84. Wilson, J., and Gravelle W. (1991). “Modelling of a cableStayed bridge for
dynamic analysis,” Earth. Eng. and Struct. Dyn., 20, 707721.
85. Yang, J.N. (1982). “Control of Tall Buildings Under Earthquake Excitations.”
ASCE Journal of Enginering Mechanics Division, 108, No. EM5, 833849.
References
101
86. Yang, G., Ramallo, J.C., Spencer Jr., B.F., Carlson, J.D., and Sain, M.K. (2000a).
“Dynamic Performance of largescale MR fluid dampers.” Proc. 14th ASCE
Engineering Mechanics Conference, CDROM, Austin, Texas.
87. Yang, G., Ramallo J.C., Spencer Jr., B.F., Carlson, J.D., and Sain, M.K. (2000b).
“Largescale MR fluid dampers: dynamic performance considerations.” Proc. Int.
Conf. on Advances in Struc. Dyn. Hong Kong, China, 1, 341–348.
88. Yang, G., Spencer Jr., B.F., Carlson, J.D., and Sain, M.K. (2001a). “Dynamic
modeling and performance considerations on fullscale MR fluid dampers.” Proc.
8th Int. Conf. on Structural Safety and Reliability, Newport Beach, CA.
89. Yang, G., Jung, H.J., and Spencer Jr., B.F. (2001b). “Dynamic model of fullscale
MR dampers for civil engineering applications.” Proc. USJapan Workshop on
Smart Structures for Improved Seismic Performance in Urban Region, Seattle,
WA.
90. Yang, J. N., and Agrawal, A. K. (2002). “Semiactive hybrid control systems for
nonlinear buildings against nearfield earthquakes.” Eng. Struct., 24(3), 271280.
91. Yang, J. N., and Dyke, S. J. (2003). “Kobori Panel Discussion: Future perspectives
on structural control.” Proc., 3rd World Conf. On Structural Control, Wiley, New
York, 279286.
92. Yi, F., and Dyke, S.J. (2000). “Structural control systems: performance
assessment.” Proc. of American Control Conf., Chicago, IL.
93. Yoshida, O., and Dyke, S.J. (2002). “Seismic control of a nonlinear benchmark
building using smart dampers.” J. of Engineering Mechanics: Special Issue on
Structural Control Benchmark Problems, ASCE(submitted).
94. Yoshioka, H., Ramallo, J.C., and Spencer Jr., B.F. (2002). “‘Smart’ base isolation
strategies employing magnetorheological dampers.” J. Struct. Eng., 128(5), 540
551.
감사의 글
사랑하시는 하나님께 감사 드립니다.
또 다른 삶의 한 장을 마무리하며, 함께 했던 여러분들께 감사 드리고자 합니
다. 무엇보다 지난 7년 동안 세심한 지도와 올바른 삶의 태도를 깨우쳐 주신 이인원
교수님께 깊이 감사 드립니다. 세심한 심사와 조언을 해주신 윤정방 교수님, 김진근
교수님, 곽효경 교수님, 기계공학과 장평훈 교수님께 감사 드립니다. 그리고 한국과
학기술원 건설환경공학과 교수님들께도 감사 드립니다. 세미나에 참석하셔서 많은
것을 가르쳐주신 한남대학교 오주원 교수님, 경일대학교 이종헌 교수님, 공주대학
교 고만기 교수님, 성균관대학교 박선규 교수님, 한경대학교 김운학 교수님, 중부대
학교 김춘호 교수님, 세종대학교 정형조 교수님께 감사 드립니다. 아울러 복합재료
를 가르쳐 주신 Korea Composites 김덕현 박사님께도 감사 드립니다.
당당하게 살아가는 멋진 정길호, 김동옥, 김만철, 박정규, 권호철, 김동현, 조
지성, 정환진, 김현택, 이승우 선배님들께 감사 드립니다. 또한 후배지만 어엿한 사
회인으로 살아가는 김기영, 지현우, 조홍기, 신혜린, 장동두, 하준식에게도 고마운
마음을 전합니다. 그리고 여전히 실험실에서 함께 하는 동기 김병완 박사, 박규식,
탄, 문영종, 최강민, 이헌재, 이성진, 장종우, 임현우, 장지은에게도 감사 드립니다.
각자의 길을 성실히 가고 있는 석사 동기들에게 감사의 뜻을 전합니다. 같은
뜻을 품고 캠퍼스를 누볐던 목원대학교와 대전IVFer들, 93동기들, 그리고 항상 따
스함으로 다가오는 늘사랑 교회와 구역 식구들의 애정어린 관심과 사랑을 기억합
니다.
손주 사위를 많이 귀여워 해주신 할머님과 맏아들과 맏사위로서 격려해 주신
아버지, 어머니, 장인, 장모님께 감사 드립니다. 결혼해서 훌쩍 어른이 되어 버린, 여
동생 범주와 남편 준용이 그리고 처제 경훈과 남편 성구형, 그리고 군 복무중인 처
남 진우에게도 고맙다는 말을 전합니다.
마지막으로, 나보다 나를 더 사랑하는 젊고 아름다운 아내, 경진에게 고마운
마음과 사랑을 전합니다.
CURRICULUM VITAE
Name : SangWon Cho
Date of Birth : August 15, 1974
Place of Birth : Seoul, Korea
EDUCATION
Ph.D. Civil and Environmental Engineering
Korea Advanced Institute of Science and Technology (KAIST), Korea
March 1999Present
M.S. Civil and Environmental Engineering
Korea Advanced Institute of Science and Technology (KAIST), Korea
March 1997February 1999
B.S. Civil and Environmental Engineering
Korea Advanced Institute of Science and Technology (KAIST), Korea
March 1993February 1997
EXPERIENCE
March 1999 Graduate Research Assistant,
to Present Department of Civil and Environmental Engineering, KAIST, Korea
LIST OF PUBLICATIONS
Dissertations:
1. “Efficient Mode Superposition Methods for Nonclassically Damped Systems,”
M.S. Dissertation, Department of Civil and Environmental Engineering, Korea
Advanced Institute of Science and Technology, Daejeon, Korea, February 1999.
2. “Simple Control Algorithms for MR Dampers and Smart Passive Control System,”
Ph.D. Dissertation, Department of Civil and Environmental Engineering, Korea
Advanced Institute of Science and Technology, Daejeon, Korea, February 2004.
Journal Papers:
1. SangWon Cho, ByoungWan Kim, HyungJo Jung & InWon Lee, "The
Implementation of Modal Control for Seismic Structures using MR Damper,"
ASCE Journal of Engineering Mechanics, (Accepted for Publication)
2. SangWon Cho, HyungJo Jung & InWon Lee, "Design and Control of
Magnetorheological Dampers with the Electromagnetic Induction System for
Seismic Response Reduction," Smart Material and Structures, (Accepted for
Publication)
3. KangMin Choi, SangWon Cho, ManGi Ko & InWon Lee, "Algebraic Method
for Sensitivity Analysis of Eigensystems with Repeated Eigenvalues," Computers
and Structures, Vol. 82, No. 1, pp. 6369, Dec. 2003.
4. KangMin Choi, SangWon Cho, HyungJo Jung & InWon Lee, "Semiactive
Fuzzy Cotrol for Seismic Response Reduction Using MR Damper," Earthquake
Engineering and Structural Dynamics, (Accepted for Publication)
5. KangMin Choi, SangWon Cho, ManGi Ko & InWon Lee, "Algebraic Method
for Sensitivity Analysis of Eigensystems with Repeated Eigenvalues," KSCE
Journal of Civil Enginieering, Vol. 7, No. 1, pp. 1723, 2003. 1.
6. SangWon Cho, JiSeong Jo, InWon Lee, "Efficient Mode Superposition Methods
for NonClassically Damped Systems," Journal of the Earthquake Engineering
Society of Korea, Vol. 4, No. 1, pp. 8998, 2000. 3. (in Korean)
7. SangWon Cho, HyungJo Jung , DongOk Kim, InWon Lee, "Efficient Mode
Superposition Methods for NonClassically Damped Systems," Journal of the
Korean Society of Civil Engineers, Vol. 19, No. 16, pp. 859868, 1999. (in
Korean)
Conference Papers:
1. SangWon Cho, HyungJo Jung, ChunHo Kim, InWon Lee, "Simple and
Efficient Control Algorithm for Seismic Response Reduction of Large Scale
Structures using MR Dampers," The Sixteenth KKCNN Symposium on Civil
Engineering, Kyeongju, Korea, Dec. 810, 2003. pp. 243248.
2. SangWon Cho, KyuSik Park, ManGi Ko, InWon Lee, "Modal Control for
Seismic Structures using MR Dampers," The Second International Conference on
Structural and Construction Engineering, Rome, Italy, Sep. 2326, 2003. pp. 2063
2066.
3. KangMin Choi, SangWon Cho, InWon Lee, JongHeon Lee, "Algebraic Method
for Sensitivities Analysis of Eigensystem with Repeated Eigenvalues," The Second
International Conference on Structural and Construction Engineering, Rome,
Italy, Sep. 2326, 2003. pp. 597602.
4. SangWon Cho, ByoungWan Kim, KyuSik Park, InWon Lee, "Modal Control
using MR Damper," The Fifth KKCNN Symposium on Civil Engineering, Kent
Ridge, Singapore, Dec. 1920, 2002. pp. 7984.
5. SangWon Cho, KyuSik Park, WoonHak Kim, InWon Lee, "Modal Control for
Seismically Excited Structures using MR Damper," The Second ChinaJapan
Korea Symposium on Optimization of Structural and Mechanical Systems (CJK
OSM 2), Busan, Korea, Nov. 48, 2002.
6. HeonJae Lee, SangWon Cho, HyungJo Jung, JuWon Oh, InWon Lee, "Neuro
Control for Seismic Response Reduction using a Semiactive MR Fluid Damper,"
The Second ChinaJapanKorea Symposium on Optimization of Structural and
Mechanical Systems (CJKOSM 2), Busan, Korea, Nov. 48, 2002.
7. SangWon Cho, ByoungWan Kim, HyungJo Jung & InWon Lee,
"Implementation of Modal Control for Seismically Excited Structures using MR
Damper," International Conference on Advances and New Challenges in
Earthquake Engineering Research, Hong Kong, Aug. 1920, 2002.
8. KyuSik Park, SangWon Cho, InWon Lee, "A Comparative Study on Aseismic
Performances of Base Isolation Systems for MultiSpan Bridge," ASCE's First
Virtual World Congress for Civil Engineering (www.ceworld.org), Online
Conference, July 1, 2002.
9. Sangwon Cho, JiSeong Jo & InWon Lee, "Modified BangBang Control of
Seismically Excited Structures Using MR Damper," KAISTKyoto Univ. Joint
Seminar on Earthquake Engineering, KAIST, Korea, February 25, 2002, pp. 125
132
10. JiSeong Jo, SangWon Cho & InWon Lee, "Modified Sturm Sequence Property
for Damped Systems," KAISTKyoto Univ. Joint Seminar on Earthquake
Engineering, KAIST, Daejeon, Korea, February 25, 2002, pp. 6570
11. DongHyawn Kim, SangWon Cho & InWon Lee, "Intelligent Control of
Structural Vibration using CMAC," The Eighth East AsiaPacific Conference on
Structural Engineering & Construction, Singapore, December 57, 2001, Paper No.
1401.
12. SangWon Cho, DongHyawn Kim & InWon Lee, "NeuroControl of Structures
using CMAC," The First AsianPacific Congress on Computational Mechanics,
Sydney, Australia, November 2023, 2001, pp. 12771282.
13. KyuSik Park, SangWon Cho & InWon Lee, "A Comparative Study on Aseismic
Performances of Base Isolation Systems for Multispan Continuous Bridge," The
Fourteenth KKNN Symposium on Civil Engineering, Kyoto, Japan, November 57,
2001, pp. 3540.
14. SangWon Cho, InWon Lee & JuWon Oh, "Efficient Mode Superposition
Methods for NonClassically Damped System," The 12th KKNN Seminar/
Workshop on Civil Engineering, Daejeon, Korea, August 2022, 1999, pp. 8186
15. JongWoo Jang, SangWon Cho, InWon Lee, & WooHyun Yoon, "Optimal
Placement of MR Dampers for 20story Nonlinear Benchmark Building"
Conference on Korean Society of Civil Engineers, Daegu, Oct. 2425, 2003. (In
Korean)
16. SangWon Cho, HyungJo Jung, SunKyu Park, In Won Lee, "Maximum Energy
Dissipation Algorithm for Seismic Response Reduction of LargeScale Structures
using MR Dampers," Conference on Korean Society of Civil Engineers, Daegu,
Oct. 2425, 2003. (In Korean)
17. JongWoo Jang, SangWon Cho, InWon Lee, & WooHyun Yoon, "Optimal
Placement of MR Dampers for 20story Nonlinear Benchmark Building,"
Conference on Computational Structural Engineering Institute of Korea, Daejeon,
Oct. 11, 2003, pp.153160. (In Korean)
18. JongWoo Jang, SangWon Cho, InWon Lee, & WooHyun Yoon "Optimal
Placement of MR Dampers for 20story Nonlinear Benchmark Building,"
Conference on Earthquake Engineering Society of Korea, Gunsan, Sep. 19, 2003,
pp. 467472. (In Korean)
19. SangWon Cho, HyungJo Jung, JongHeon Lee, & InWon Lee, "Vibration
Control for a Benchmark CableStayed Bridge using Maximum Energy Dissipation
Algorithm," Conference on Earthquake Engineering Society of Korea, Gunsan,
Sep. 19, 2003, pp. 435441. (In Korean)
20. KyuSik Park, HyungJo Jung, SangWon Cho & InWon Lee, "Hybrid Control
with a BangBang Type Controller," Conference on Computational Structural
Engineering Institute of Korea, Ansan, Apr. 12, 2003, pp. 193200. (In Korean)
21. SangWon Cho, HyungJo Jung, JuWon Oh & InWon Lee, "Implementation of
Method Control for Seismically Excited Structures Using MR Damper,"
Conference on Korean Society of Civil Engineers, Busan, Nov. 89, 2002. (In
Korean)
22. SangWon Cho, JuWon Oh & InWon Lee, "Modal Control of Vibration using
MR Damper," Conference on Earthquake Engineering Society of Korea, Asan,
Sept. 28, 2002, pp. 357363. (In Korean)
23. KangMin Choi, SangWon Cho, WoonHak Kim & InWon Lee, "Algebraic
Method for Computation of Eigenpair Sensitivities of Damped Systems with
Repeated Eigenvalues," Conference on Korean Society of Steel Construction,
Seoul, June 8, 2002, pp. 141147. (In Korean)
24. SangWon Cho, ByoungWan Kim, WoonHak Kim & InWon Lee, "Modified
Decentralized BangBang Control Seismically Excited Structures Using MR
Dampers," Conference on Earthquake Engineering Society of Korea, Seoul, Mar.
23, 2002. (In Korean)
25. Sangwon Cho, JiSeong Cho, SunKyu Park & InWon Lee, "Efficient Mode
Superposition Method for NonClassically Damped Systems," Conference on
Korean Society of Noise and Vibration Engineering, Jeju, Jun. 23, 2000, pp. 549
555. (In Korean)
26. SangWon Cho, HyungJo Jung, DongOk Kim & InWon Lee, "Efficient Mode
superposition Method for NonClassically Damped Systems,” Conference on
Korean Society of Steel Construction, Seoul, Jun. 12, 1999, pp. 293300. (In
Korean)
27. SangWon Cho, ManCheol Kim, SunKyu Park & InWon Lee, "Development of
an Efficient Mode Superposition Method for Nonclassically Damped System,"
Conference on Korean Society of Civil Engineers, Seoul, Oct. 2324, 1998, pp.
471474. (In Korean)
Research Reports:
1. “Development of Integrated Aseismic Control Systems for Structures by Advanced
Vibration Control Techniques,” National Research Laboratory (NRL) Program,
(Principal Investigator: InWon Lee), KAIST, Ministry of Science & Technology,
(in Korean), September 2002.
2. “Research Report for Brain Korea 21 (BK21) Program,” Advanced Structural
Engineering Technology Research Team, (Principal Investigator: ChungBang
Yun), KAIST, Ministry of Education, (in Korean), September 2002.
3. “Technology for Improving WindResistant Performance of LargeScaled
Bridges,” (Principal Investigator: InWon Lee), KAIST, Korea Institute of
Construction Technology in Ministry of Construction & Transportation, (in
Korean), February 2000.
4. “Development of PreDesign Program for LRB,” (Principal Investigator: InWon
Lee), KAIST, Unison Industrial, (in Korean), March 2000.
5. “Development of Efficient Dynamic Analysis Method for NonClassically Damped
Systems,” (Principal Investigator: InWon Lee), Hyundai E&C Co. Ltd., December
1999.
Patent:
1. MR damper with electromagnetic induction system to replace power source and
sensor, Korean Patent, Application no: 200261823, October 2002.
MR 댐퍼를 위한 간편한 알고리즘과 스마트 수종제어 시스템
Simple Control Algorithms for MR Dampers and Smart Passive Control System
Simple Control Algorithms for MR Dampers and Smart Passive Control System
Advisor : Professor InWon Lee by SangWon Cho Department of Civil and Environmental Engineering Korea Advanced Institute of Science and Technology A thesis submitted to the faculty of the Korea Advanced Institute of Science and Technology in partial fulfillment of the requirements for the degree of Doctor of Philosophy in the Department of Civil and Environmental Engineering.
Daejeon, Korea 2003. 11. 27 Approved by
Professor InWon Lee Major Advisor
2003 년 11 월 27 일 심사위원장 심 사 위 원 심 사 위 원 심 사 위 원 심 사 위 원 이 인 윤 김 곽 장 원 (인) (인) (인) (인) (인) 정 방 진 근 효 경 평 훈 .MR 댐퍼를 위한 간편한 알고리즘과 스마트 수동제어 시스템 조 상 원 위 논문은 한국과학기술원 박사학위논문으로 학위논문 심사위원회에서 심사 통과하였음.
A number of control algorithms have been adopted for semiactive systems including the MR damper. Thus. As a simple and efficient control algorithm. Although modal control has been investigated for the several decades. a modal control scheme is more convenient to design the controller than other control algorithms. in order to study the effectiveness for the MR damper system. a modal control scheme and a maximum energy dissipation algorithm (MEDA) are implemented for the MR damperbased control system. In spite of good features of previous studies. ABSTRACT This dissertation proposes simple and efficient control algorithms for seismically excited structures using MR dampers and a smart passive system based on MR dampers. because they have many advantages such as small power requirement. Thus. reliability. some algorithms have drawbacks such as poor performances or difficulties in designing the weighting matrix of the controller. Advisor: Professor InWon Lee. Hence. MR댐퍼를 위한 간편한 제어 알고 리즘과 스마트 수동제어 시스템.조 상 원. its potential for a semiactive control. has not been exploited. Text in English. SangWon Cho. Modal control reshapes the motion of a structure by merely controlling a few selected vibration modes. 2004. especially for the MR damper. which is simple to use and efficient to give comparable or better performance over the previous algorithms. 101p. Magnetorheological (MR) dampers are one of the most promising control devices for civil engineering applications to earthquake hazard mitigation. Simple Control Algorithms for MR Dampers DCE 995353 and Smart Passive Control System. Department of Civil and Environmental Engineering. a modal control scheme is implemented to seismically excited i . the control algorithm is required. and low price to manufacture.
controller. displacement. The maximum energy dissipation algorithm represents one control class which employs the Lyapunov’s direct approach to stability analysis in the design of a feedback controller. This paper investigates the performance and the robustness of the maximum energy dissipation algorithm for civil engineering structures using MR dampers. and relative displacements throughout the structure is examined. a lowpass filter is applied to eliminate the spillover problem. a sixstory building model with the MR dampers on the bottom two floors is used to verify the proposed modal control scheme. velocity. such as cablestayed bridges and highrise buildings. Moreover. to reduce the responses of the controlled structure by using MR dampers. A Kalman filter is included in a control scheme to estimate modal states from physical measurements by sensors. However. especially with MR dampers. Various earthquakes are used to excite the systems. Meanwhile. their potential for civil engineering applications using semiactive control. respectively. a control system including a power supply. accelerations. and relative displacements throughout the structure are examined according to the evaluation criteria. and acceleration. ii . However. it is not easy to apply the MR damperbased control system to largescale civil structures. because of the difficulties of building up and maintaining the control system. Three cases of the structural measurement are considered as a feedback to verify the effect of each measurement. accelerations. has not yet been fully exploited. The numerical examples contain the cablestayed bridge and the nonlinear building. and sensors is needed. In a numerical example.structures. The El Centro earthquake is used to excite the system. the performance is compared with that of other control algorithms that are previously proposed: The reduction in the drifts. and the reduction in the drifts. The performance of the proposed modal control scheme is compared with that of other control algorithms that were previously suggested. Through the series of numerical simulations.
Therefore. this dissertation proposes a ‘smart’ passive damper system. Of course. the MR damper is a semiactive device that needs an external power source to change the damping characteristics of the MR fluids. it is easy to build up and maintain the proposed smart damper system that consists of the MR damper and the EMI system. because it does not require any control system such as a power supply. The smart passive damper system is based on MR dampers. the performances are compared with those of the semiactive MR damper.Thus. To verify the effectiveness of the proposed EMI system. but selfpowered by an electromagnetic induction (EMI) system that is attached to the MR damper. However. and sensors. controller. The EMI system consists of a permanent magnet and a coil. the EMI system changes the kinetic energy of the MR damper to the electric energy and then the electric energy is used to vary the damping characteristics of the MR damper. According to the Faraday’s law of induction. iii . the smart passive damper system based on MR dampers is not using an external power source.
......................................................................... vii CHAPTER 1 INTRODUCTION.1........................... 7 1.......................................................................................2 Numerical Example .....3 MR Fluids and Dampers .......................................................... 27 2.................................................3 Modal State Estimation ..........4 Elimination of Observable Spillover .....1 Background.......................................................... 12 1...............................................................................................2........... 1 1..... 17 CHAPTER 2 MODAL CONTROL SCHEME .................. iv LIST OF TABLES .................................................... 43 iv .................................................................................................................1 Modal Control Scheme for MR Dampers.2 Literature Review ......................................................... 15 1..................1 Magnetorheological (MR) Dampers........ 1 1................................................3 Summary of Results....................................1 MR Fluids ............................................................................ 18 2........3.... 21 2.. 18 2......................................................................................................................TABLE OF CONTENTS ABSTRACT ..................5 Organization ............. 10 1.......................................................... 4 1.......2............................................................................................................................................. vi LIST OF FIGURES .1.........................................................1.. i TABLE OF CONTENTS .............................................3.......................................................................................... 18 2.................................................1 Modal Control.... 4 1.......2 Control Algorithms for MR Dampers..........................................................2 MR Fluids Dampers...............................................................2 Design of Optimal Controller .......1........................................ 23 2................................................4 Objectives and Scopes .......... 29 2........................................... 10 1.............................................................................................
.....1...................................... 45 3............... 49 3....1 Control Performance.4 Summary of Results............................................................................1................................................... 76 4....................................2..................................... 55 3....................... 88 SUMMARY (IN KOREAN) REFERENCES ACKNOWLEDGEMENTS CURRICULUM VITAE 90 93 v ........................................................2.... 67 CHAPTER 4 SMART PASSIVE CONTROL SYSTEM ....1 Control Devices ................................................................................. 44 3....2 Analytical Model and Design .................CHAPTER 3 MAXIMUM ENERGY DISSIPATION ALGORITHM ....2 Controller Robustness.................................................................2...........................................................1 Control System ....3 Numerical Examples........................... 68 4.............. 44 3....2....................................1 Analytical Model .............. 73 4..................... 80 4.........3.................................. 59 3..............4 Summary of Results........3 Numerical Simulation Results .....2 Nonlinear Benchmark Building... 87 CHAPTER 5 CONCLUSIONS ................................................. 59 3................................................................................................................................................................................................................. 68 4.......... 64 3................................................ 49 3........1 Benchmark CableStayed Bridge ..........................2 Design of the EMI System..........................................1 Electromagnetic Induction System for MR Damper ..................................3.....................2 Maximum Energy Dissipation Algorithm for MR Damper........................................................................ 73 4............. 47 3.....................2 Benchmark Problems............................................
.2 2...........................................3 3.............................. 86 vi ....................................................2 3.............. 42 Parameters for MR damper model................... 63 Evaluation criteria for ±7% stiffness perturbed system under El Centro earthquake ....LIST OF TABLES 1.............6 4........1 2..... 62 Evaluation criteria of modified location and number of MR dampers .......................................................................................... 46 Comparisons of the evaluation criteria for benchmark cablestayed bridge......3 2.................................................... 40 Normalized controlled maximum responses of the displacement feedback due to the scaled El Centro earthquake......... 65 Evaluation criteria for ±30% stiffness perturbed system .............. 61 Comparisons of the evaluation criteria for the nonlinear benchmark building................. 66 Normalized peak absolute accelerations and interstory drifts ................ 34 Normalized controlled maximum responses of the acceleration feedback due to the scaled El Centro earthquake.......4 3............................4 3...... 11 Normalized controlled maximum responses due to the scaled El Centro earthquake....................1 2......... 41 Normalized controlled maximum responses of the velocity feedback due to the scaled El Centro earthquake........................................1 4..................... 85 Percent increment compared to the better clippedoptimal controller case .......................1 3.................................2 Properties of MR and ER fluids.........5 3.................................................
.................................................................... 37 Variations of evaluation criteria with weighting parameters for the velocity feedback......................................................................................................... 29 Frequency responses of the first floor for the uncontrolled structures under the scaled El Centro earthquake.... 36 Variations of evaluation criteria with weighting parameters for the displacement feedback ...................................2 1.................... 70 Schematic of a MR damper implementation ....5 2................................................... 54 Finite element model of the towers...............................1 2...... 54 Schematic of the 20story benchmark building ...4 2....3 2.......................................... 51 Cross section of bridge deck................................ 45 Drawing of the Cape Girardeau Bridge ................................1 3........................................ 14 Bypass type 20ton MR fluid................ 51 Finite element model .. 32 Frequency responses of the sixth floor for the uncontrolled structures under the scaled El Centro earthquake..................................................... 33 Variations of evaluation criteria with weighting parameters for the acceleration feedback ......................5 Behavior of MR fluid in magnetic field.......................... 10 Schematic of the prototype 20ton largescale MR fluid damper ............................................... 68 Schematic of a MR damper with the EMI system ......... 70 Schematic of a MR damper with the EMI system implementation ..1 4...................... 56 Schematic of a MR damperbased control system..............................................................2 2...................................................................4 2................. 38 Mechanical model of the MR damper ......................................................................2 4........................6 4.........................3 4.........5 3............ 75 vii ...................1 1...........................................LIST OF FIGURES 1............6 3.............4 3.....................................................3 3.................................................................. 73 Simple mechanical model of the normal MR damper ....2 3............................................................................................. 13 Smallscale SD1000 MR fluid damper. 14 Schematic diagram of the MR damper implementation ..................................................................4 4..........3 1............
4.6 4.7 4.8 4.9
Design of EMI system with Sa under three earthquakes......................................... 77 Design of EMI system with Si under three earthquakes ......................................... 77 Design of the clippedoptimal controller with Sa under three earthquakes............. 79 Design of the clippedoptimal controller with Si under three earthquakes ............. 79
4.10 Velocities and induced voltages under various earthquakes................................... 81 4.11 Normalized peak acceleration and interstory drift ................................................ 83
viii
Chapter 1 Introduction
1
CHAPTER 1 INTRODUCTION
1.1 Background
The tragic consequences of the recent earthquakes have underscored, in terms of both human and economic factors, the tremendous importance of the way in which buildings and bridges respond to earthquakes. In recent years, considerable attention has been paid to research and development of structural control systems. Supplemental passive, active, hybrid, and semiactive damping strategies offer attractive means to protect structures against natural hazards. Passive supplemental damping strategies, including base isolation systems, viscoelastic dampers, and tuned mass dampers, are widely accepted by the engineering community as a means for mitigating the effects of dynamic loading on structures. However, these passivedevice methods are unable to adapt to structural changes, varying usage patterns, and loading conditions. For more than two decades, researchers have investigated the possibility of using active, hybrid, and semiactive control methods to improve upon passive approaches to reduce structural responses (Soong 1990; Soong and Reinhorn 1993; Spencer and Sain 1997; Housner et al. 1997; Kobori et al. 1998, 2003; Soong and Spencer 2002; Spencer 2002). The first fullscale application of active control to a building was accomplished by the Kajima Corporation on 1989 (Kobori et al. 1991). The Kyobashi Center building is an 11story (33.1m) building in Tokyo, having a total floor area of 423m2. A control system was installed, consisting of two AMDs – the primary AMD is used for transverse motion and has a mass of 4 t, while the secondary AMD has a mass of 1 t and is employed to reduce torsional motion. The role of the active system is to reduce building vibration under strong winds and moderate earthquake excitations and consequently to increase comfort of occupants of the building.
Chapter 1 Introduction
2
Hybridcontrol strategies have been investigated by many researchers to exploit their potential to increase the overall reliability and efficiency of the controlled structure (Housner et al. 1994; Kareem et al. 1999; Nishitani and Inoue 2001; Yang and Dyke 2003; Faravelli and Spencer 2003). A hybrid control system is typically defined as one that employs a combination of passive and active devices. Because multiple control devices are operation, hybrid control systems can alleviate some of the restrictions and limitations that exist when each system is acting alone, Thus, higher levels of performance may be achievable. Additionally, the resulting hybrid control system can be more reliable than a fully active system, although it is also often somewhat more complicated. To date, there have been over 40 buildings and about 10 bridges (during erection) that have employed feedback control strategies in fullscale implementations (Spencer and Nagarajaiah 2003). Although nearly a decade has passed since construction of the Kobashi Seiwa building, a number of serious challenges remain to be resolved before feedback control technology can gain general acceptance by the engineering and construction professions at large. These challenges include: (i) reduction of capital cost and maintenance, (ii) eliminating reliance on external power, (iii) increasing system reliability and robustness, and (iv) gaining acceptance of nontraditional technology by the profession. Semiactive control strategies appear to be particularly promising in addressing a number of these challenges (Spencer 1996). Control strategies based on semiactive control devices appear to combine the best features of both passive and active control systems and to offer the greatest likelihood for near term acceptance of control technology as a viable means of protecting civil engineering structural systems against earthquake and wind loading. The attention received in recent years can be attributed to the fact that semiactive control devices offer the adaptability of active control devices without requiring the associated large power sources. In fact, many can operate on battery power, which is critical during seismic events when the main power source to the structure may fail. According to presently
e. Moreover. MR fluids are magnetic analogs of electrorheological fluids and typically consist of microsized.Chapter 1 Introduction 3 accepted definitions. been recognized having a number of attractive characteristics for use in structural vibration control applications. Therefore. including the structure and the control device). particle chains form. Most of the semiactive control devices have employed some electrically controlled valves or mechanisms. When a magnetic field is applied to the fluid. and (2) magnetorheological (MR) fluids. The advantage of controllable fluid devices is that they contain no moving parts other than the piston. recently developed MR fluids appear to be an attractive alternative to ER fluids for use in controllable fluid dampers (Carlson 1994. 1995). Therefore MR dampers have. semiactive control devices do not have the potential to destabilize (in the bounded input/bounded output sense) the structural system. Another class of semiactive devices uses controllable fluids. MR fluids are not sensitive to impurities such as are commonly encountered during manufacturing and usage. Spencer 2002).. and the fluid becomes a semisolid and exhibits viscoplastic behavior similar to that of an ER fluid. in contrast to active control devices. Such mechanical components can be problematic in terms of reliability and maintenance. . Carlson et al. However. but has properties which can be controlled to optimally reduce the responses of the system. a semiactive control device is one that can not inject mechanical energy into the controlled structural system (i. Carlson and Weiss (1994) indicated that the achievable yield stress of an MR fluid is an order of magnitude greater than its ER counterpart. Two fluids that are viable contenders for development of controllable dampers are: (1) electrorheological (ER) fluids. magnetically polarizable particles dispersed in a carrier medium such as mineral or silicone oil. which makes them very reliable. thus allowing for the possibility of effective response reduction during a wide array of dynamic loading conditions (Spencer and Sain 1997. Previous studies indicate that appropriately implemented semiactive systems perform significantly better than passive devices and have the potential to achieve the majority of the performance of fully active systems. over the last several years. Symans and Constantinou 1999. Carlson and Weiss 1994.
1996a. Despite these advances in the development of ER fluid dampers. are not affected by most impurities. Additionally. These fluids can be modeled as Newtonian fluids in the absence of a magnetic field. have a 50 to 100 KPa maximum yield stress. Gavin et al.2.2 Literature Review 1. on the other hand. Gavin et al. 1996b. 1996. First. Although the discovery of ER and MR fluids dates back to the 1940s. common impurities that might be introduced during manufacturing significantly reduce the capacity of the fluids. To date. 1996) for structural vibration control applications in civil engineering. availability and the cost of highvoltage (e. ~4000 volts) power supplies required to control the ER fluids are further considerations. Kamath et al. 1996b) designed and tested an ER damper that consisted of a rectangular container and a moving plunger comprised of nine rigidly connected flat plates.5 KPa. Moreover. even the best ER fluids currently available may only achieve stresses of 3. a number of ER fluid dampers have been investigated (Burton et al. only recently have they been applied to civil engineering applications. 1996. Also.Chapter 1 Introduction 4 1. MR fluids. and are not sensitive to temperature. linear viscous fluids to semisolids with a controllableyield strength in only a few milliseconds when exposed to an electric (ER fluids) or magnetic field (MR fluids). the viscoplasticity model (Phillips 1969) may be used to describe the fluid behavior. (1996) developed an ER damper consisting of an outer cylinder and a double20 ended piston rod that pushes the ER fluids through an annular duct.g. (1996a. Makris et al. These fluids are unique in their ability to reversibly change from freeflowing. Makris et al. the fluids have a very limited yield stress.1 Magnetorheological (MR) Dampers Controllable fluid dampers generally utilize either electrorheological (ER) fluids or magnetorheological (MR) fluids. MR fluids can be controlled . the development of commercially feasible damping devices using these fluids is limited by several factors. safety. When a field is applied.0 to 3.
Different techniques have been developed to model the behavior of the controllable fluid dampers. 1998). (1997a) and Yang et al. (1996b). Basically. (1996b) extended this approach to model the ER damper. The model consists of a Bingham model in series with a standard model of a linear solid model. MR fluids are particularly promising for natural hazard mitigation and cost sensitive applications (Carlson and Spencer 1996a.b) employed the MR damper to . currentdriven power supply with ~12 amps output. 1996b.b). and Wereley et al. (1996). Jansen and Dyke (2000). Dyke et al. (1987) proposed a simple mechanical model. (2001a. two types of models have been investigated: nonparametric and parametric models. Spencer et al. Spencer et al. (2001a. Gavin et al. Chang and Roschke (1998) developed a neural network model to emulate the dynamic behavior of MR dampers.. the Bingham model. Makris et al. Spencer and Sain 1997). However.b. Stanway et al. ~1224 volts). Ehrgott and Masri (1992) presented a nonparametric approach to model a small ER damper that operates under shear mode by assuming that the damper force could be written in terms of Chebychev polynomials. and Yi and Dyke (2000) used MR dampers to reduce the seismic vibration of building structure model. (1998) developed parametric models to characterize ER and MR dampers. (1996a.Chapter 1 Introduction 5 with a lowpower (e. less than 50 watts). Kamath and Wereley (1997). (2000). A number of experimental studies have been conducted to evaluate the usefulness of MR dampers for vibration reduction under wind and earthquakes. Therefore.b) presented the BoucWen model whose versatility was utilized to describe a wide variety of hysteretic behavior. lowvoltage (e. (2001) and Yoshioka et al. Spencer et al. (1996a. (2001) incorporated an MR damper with a base isolation system such that the isolation system would be effective under both strong and moderate earthquakes. in which a Coulomb friction element is placed in parallel with a dashpot.g. Dyke et al. the nonparametric damper models are quite complicated. Ramallo et al..g. Johnson et al. Gamoto and Filisko (1991) extended the Bingham model and developed a visoelasticplastic model.
. The installation is expected to be completed in October 2003 (Spencer and Nagarajaiah 2003). Recently. Retrofitted with staycable dampers. The experimental results indicate that the MR damper is quite effective for a wide class of applications. Carlson and Spencer (1996b). typically on the order of a fraction of a percent. the technology has been demonstrated to be scalable to devices sufficiently large for implementation in civil engineering structures. Spencer et al. are prone to vibration induced by the structure to which they are connected and by weather conditions. that may cause cable galloping. such as are used in cablestayed bridges and other structures. MR dampers have been chosen for implementation on the Binzhou Yellow River Bridge in China to reduce cable vibration. The extremely low damping inherent in such cables. Moreover. the Dongting Lake Bridge in Hunan. The NihonKagakuMiraikan. the Tokyo National Museum of Emerging Science and Innovation has two 30tonMR fluid dampers installed between the third and fifth floors. Two Lord SD1005 MR dampers are mounted on each cable to mitigate cable vibration.Chapter 1 Introduction 6 reduce windinduced stay cable vibration. A total of 312 MR dampers are installed on 156 stayed cables. (1999). (2003) have also presented impressive results regarding design and construction of largescale MR dampers. In 2001. causing reduced cable and connection life due to fatigue and/or breakdown of corrosion protection. (2002) have developed and tested a 20t MR damper. Recently. is insufficient to eliminate this vibration. the first fullscale implementation of MR dampers for civil engineering application was achieved. Sodeyama et al. Long steel cables. China constitutes the first fullscale implementation of MR dampers for bridge structures. and Yang et al. particularly wind combined with rain. The dampers were built by Sanwa Tekki using the Lord Corporation MR fluid.
As previously mentioned. The effective utilization of multiple control devices is an important step in the examination of semiactive control algorithms. In one of the first examinations of semiactive control. (1974) proposed a ‘‘skyhook’’ damper control algorithm for a vehicle suspension system and demonstrated that this system offers improved performance over a passive system when applied to a singledegreeoffreedom system. one of the challenging aspects of utilizing this technology to achieve high levels of performance is in the development of appropriate control algorithms. Leitmann 1994). Modal control . a control strategy based on Lyapunov stability theory has been proposed for electrorheological dampers (Brogan 1991. Feng and Shinozukah (1990) developed a bangbang controller for a hybrid controller on a bridge. in which the motion of a structure is reshaped by merely controlling some selected vibration modes. The goal of this algorithm is to reduce the responses by minimizing the rate of change of a Lyapunov function. A modulated homogeneous friction algorithm (Inaudi 1997) was developed for a variable friction device. Among many control algorithms.2 Control Algorithms for MR Dampers One challenge in the use of semiactive technology is in developing nonlinear control algorithms that are appropriate for implementation in fullscale structures. modal control represents one control class. a number of control algorithms have been adopted for semiactive control systems using MR dampers (Jansen and Dyke 2000). McClamroch and Gavin (1995) used a similar approach to develop a decentralized bangbang controller. Clippedoptimal controllers have also been proposed and implemented for semiactive systems (Sack et al.2. Karnopp et al.b. A typical control system for a fullscale structure is expected to have control devices distributed throughout a number of floors.c). More recently. 1994. Because of the inherent nonlinear nature of these devices. Sack and Patten 1994. This control algorithm acts to minimize the total energy in the structure. Numerous control algorithms have been adopted for semiactive systems.Chapter 1 Introduction 7 1. 1996a. Dyke.
It is . the motion of the structure can be effectively suppressed by merely controlling these few modes (Yang 1982). may involve hundred or even thousand degrees of freedom. To estimate the modal states from the sensor output. in developing the control law based on Lyapunov stability theory. The troublesome of estimating the modal states for feedback in modal control is the problem of spillover. is desirable. the origin is stable in the sense of Lyapunov. The approach requires the use of a Lyapunov function that must be a positive definite function of the states of the system. 1967) can be used for the case of low noisetosignal ratios and for high noisetosignal ratios. its vibration is usually dominated by the first few modes. since the control implementation may requires a large amount of sensors. Thus a modal control scheme. which uses modal state estimation. that a small amount of damping inherent in the structure is often sufficient to overcome the observation spillover effect (Meirovitch and Baruh 1983). Thus.Chapter 1 Introduction 8 is especially desirable for the vibration control of civil engineering structure. the goal is to choose control inputs for each deice that will result in making the rate of change of the Lyapunov function as negative as possible. Therefore. To date. if the rate of change of the Lyapunov function is negative semidefinite. Luenberger observer (Meirovitch 1990. numerous procedures and algorithms concerning modal control or pole assignment have been proposed in literature. Note. however. Luenberger 1971) and a KalmanBucy filter (Meirovitch. respectively. which is usually a large structural system. According to Lyapunov stability theory. observation spillover can be eliminated if the sensor signals are prefiltered so as to screen out the contribution of the uncontrolled modes. A modal control method using full state feedback may not be practical for a structural system involving a large number of DOFs. Jansen and Dyke (2000) suggested MEDA as a variation of the decentralized bangbang approach proposed by McClamroch and Gavin (1995). On the other side. At any rate. the maximum energy dissipation algorithm (MEDA) represents one control class which employs the Lyapunov’s direct approach to stability analysis in the design of a feedback controller (Brogan 1991).
. which means MEDA is simply implemented without any design process.Chapter 1 Introduction 9 noticeable that this control law requires only local measurements.
Figure 1. Normally.3 MR Fluids and Dampers 1.1. 1951) at the US National Bureau of Standards in the late 1940s. This phenomenon can solidify the suspended iron particles and restrict the fluid movement. The degree of change is related to the magnitude of the applied magnetic field.Chapter 1 Introduction 10 1. magnetizable particles in an appropriate carrier liquid. and can occur only in a few milliseconds. Consequently. MR fluids are free flowing liquids having a consistency similar to that of motor oil. the iron particles acquire a dipole moment aligned with the external field that causes particles to form linear chains parallel to the field. Energy density in MR fluids is limited by the magnetic saturation of iron particles. From a practical implementation perspective. although the total energy requirements for the ER and MR . as shown in Fig. These fluids are suspensions of micronsized.1 Behavior of MR fluid in magnetic field There are basically two types of controllable fluids – MR fluids and ER fluids.3. yield strength is developed within the fluid. The primary advantage of MR fluids stems from their high dynamic yield strength due to the high magnetic energy density that can be established in the fluid.1 MR Fluids The initial discovery and development of MR fluids can be credited to Jacob Rabinow (1948. However. 1. in the presence of an applied magnetic field.
MR devices can be controlled with a lowvoltage. energy density Power supply (typical) . require a highvoltage power source (~20005000 volts) that may not be readily available. range Stability Response time Density MR Fluids 50100 kPa ~250kA/m 0.1. η p Operable temp. such a high voltage may pose a safety hazard.1 Joules/cm3 225V 12A ER Fluids 25kPa ~4kV/mm 0.001 Joules/cm3 20005000V 110 mA η p /τ 2 y ( field ) Maxi. currentdriven power supply outputting only ~12 amps. yield Stress Maximum field Plastic viscosity. especially during strong earthquake events. Moreover.11. Table 1.Chapter 1 Introduction 11 devices are almost equal.0Pas +10 to 90°C cannot tolerate impurities milliseconds 1 to 2 g/cm3 107108 s/Pa 0. only MR devices can be easily driven by common lowvoltage power sources (Carlson and Spencer 1996a).11. The properties of both MR and ER fluids are given in Table 1. on the other hand.1 Properties of MR and ER fluids (Spencer and Sain 1997) Property Max. ER devices.0Pas 40 to 150°C unaffected by most impurities milliseconds 3 to 4 g/cm3 10101011 s/Pa 0.
1998. 1998). Linear MR fluid dampers have been designed for use as secondary suspension elements in vehicles. 1.3. The damper has an inside diameter of 20. In civil engineering applications. active control of damping have been used in numerous industrial applications. and contains approximately 6 liters of MR fluid. 2000a. For the nominal design. has a mass of 250 kg. .2 MR Fluid Dampers The maximum force that an MR damper can deliver depends on the properties of MR fluids. several MR fluid devices have been developed for commercial use by the LORD Corporation (Carlson et al. MR fluid rotary brakes are smoothacting. However. MR dampers primarily operating under direct shear mode or squeeze mode might be impractical. Some examples of recently developed MR dampers are given below. To date. Virtually all devices that use MR fluids can be classified as operating in: (a) a valve mode. 1996. 1997b. The MR fluid damper schematic is given in Fig. Yang et al. Jolly et al. Usually valve mode or its combination with direct shear mode is employed. a maximum damping force of 200. proportional brakes which are more compact and require substantially less power than competing systems. and the size of the damper. (c) a squeeze mode.3 cm and a stroke of ±8 cm.000 N (20 tons) were chosen. the amount of fluid energized by the magnetic field at any given instant is approximately 90 cm3.Chapter 1 Introduction 12 1. These dampers are capable of meeting realworld requirements and are presently either in commercial production or in production prototype trials. The completed damper is approximately 1 m long. Therefore. A 20ton prototype largescale seismic MR fluid damper was developed under cooperation between the LORD Corporation and the Structural Dynamics and Control/Earthquake Engineering Laboratory (SDC/EEL) at the University of Notre Dame (Carlson and Spencer 1996a.b). (b) a direct shear mode.2. MR fluid vibration dampers for realtime. or a combination of these modes (Carlson and Spencer 1996a). their flow pattern. Spencer et al. the expected damping forces and displacements are rather large in magnitude.
Sunakoda et al. Forces of up to 3. Spencer 1997a). Dyke 1996a. Figure 1. Fig.Chapter 1 Introduction 13 Fig. The magnetic field is generated by a 10stage electromagnet and is perpendicular to the fluid flow. MR fluids in this damper flow from a highpressure chamber to a lowpressure chamber in valve mode through a bypass outside the main cylinder. The damper is 21. In this damper. and the main cylinder is 3. 1. MR fluids flow from a highpressure chamber to a lowpressure chamber through an orifice in the piston head. 2000.2 Schematic of the prototype 20ton largescale MR fluid damper . Jolly et al. Unlike dampers mentioned previously.3 shows a smallscale SD1000 MR fluid damper manufactured by the LORD Corporation (Carlson and Spencer 1996a. 1998.5 cm long in its extended position. The bypass has an annular gap between the outside of the magnetic pole and the inside of the bypass cylinder.4 shows a bypasstype 20ton MR fluid damper designed by the Sanwa Tekki Corporation (Fujitani et al.000 N can be generated with this device.8 cm in diameter. 1. 2000).b.
Chapter 1 Introduction 14 Diaphragm MR Fluid Coil Bearing & Seal Wires to Electromagnet Accumulator Figure 1.3 Smallscale SD1000 MR fluid damper Figure 1.4 Bypass type 20ton MR fluid .
In a numerical example. a modal control scheme is implemented to seismically excited structures. A lowpass filter is applied to eliminate the spillover problem. The objectives and scopes of this study can be summarized as follows. A Kalman filter is included in a control scheme to estimate modal states from measurements by sensors.4 Objectives and Scopes The purpose of this study is to implement simple and efficient control algorithms for seismically excited structures using MR dampers and to develop a smart passive system based on the MR damper. (2) Implementation of maximum energy dissipation algorithm for seismic response reduction of largescale structures using MR dampers: The performance and the robustness of the maximum energy dissipation algorithm for civil engineering structures using MR dampers are investigated. the objectives and scopes of the study on implementations of simple and efficient control algorithms can be summarized as follows: (1) Implementation of modal control for seismically excited structures using MR dampers: In order to study the effectiveness for the MR damperbased semiactive. a sixstory building model with the MR dampers on the bottom two floors is used to verify the implemented modal control scheme. First.Chapter 1 Introduction 15 1. The performance of the proposed modal control scheme is compared with that of other control algorithms previously studied. the performance and the robustness are compared with that of other . Various earthquakes are used to excite the system. The numerical examples contain the cablestayed bridge and the nonlinear building. Through the series of numerical simulation.
The EMI system for MR dampers consists of a permanent magnet and a coil. the objectives and scopes of the study on development of a smart passive system based on the MR damper can be summarized as follows: (1) Development of a smart passive system based on the MR damper to reduce structural responses: The smart passive damper system is based on MR dampers. but selfpowered by an electromagnetic induction system (EMI) that is attached to the MR damper. The MR damper is a semiactive device that needs an external power source to change the damping characteristics of MR fluids. and relative displacements throughout the structure are examined according to the evaluation criteria. According to the Faraday’s law of induction. the smart passive damper system based on MR dampers is not using an external power source. To verify the effectiveness of the proposed smart passive control system. Next. However. accelerations. the EMI system changes the kinetic energy of the MR damper to the electric energy and then the electric energy is used to vary the damping characteristics of the MR damper. the performances are compared with those of the semiactive MR damper using clippedoptimal controller. . The theoretical backgrounds and the designing process are presented.Chapter 1 Introduction 16 control algorithms that are previously proposed: The reduction in the drifts.
a smart passive control system is proposed. A lowpass filter and the Kalman filter as a modal state estimator are reviewed and included in the modal control scheme for the MR damperbased control system in Section 2. the characteristics of MR fluids and dampers. a numerical example is considered. An analytical model and a design procedure of the proposed EMI system are described in Section 4. The results are summarized in Section 3. and the objectives and scopes of this study. To evaluate the proposed modal control scheme for usage with the MR damper.Chapter 1 Introduction 17 1.3. a set of numerical simulations are performed for the four historical earthquakes in Section 4.4 summarizes the results. . an electromagnetic induction (EMI) system is proposed for the MR damper. In Chapter 2. In Section 4.2. the cablestayed bridge and the 20story nonlinear building are shown as representative structures of civil engineering and numerical examples. Finally.2. In Chapter 4. In Chapter 3. The control system including the MR device and MEDA is reviewed in Section 3. In Section 3. Section 4. In Section 3. The results are summarized in Section 2. the conclusions of this dissertation are summarized in Chapter 5.5 Organization This dissertation consists of four chapters. also.3. Reduced design procedure is presented.4. the applicability of the MEDAbased semiactive control system is examined from the viewpoint of the performance and the robustness through the numerical examples.3. To show the effectiveness of the proposed smart passive control system.1.1. Chapter 1 discusses the background. the literature review. a modal control scheme is implemented for the MR damperbased control system. the maximum energy dissipation algorithm (MEDA) is implemented for the MR damperbased control system.2.1. in this section. in which a model of a sixstory building is controlled with four MR dampers in Section 2.
a lowpass filter is applied to eliminate the spillover problem. C and K are the n×n mass. damping. a modal control scheme with a Kalman filter and a lowpass filter is implemented to a seismically excited structure.Chapter 2 Modal Control Scheme 18 CHAPTER 2 MODAL CONTROL SCHEME 2. fm ]T is the vector of measured control forces generated by m MR dampers. the equations of motion can be written & M&&(t ) + Cx (t ) + Kx (t ) = Λf (t ) − MΓ &&g x x (2. respectively.1.1) where M. A Kalman filter is included in a control scheme to estimate modal states from various measurements. numerical simulations are presented in subsequent sections for comparisons between control algorithms. x is the ndimensional vector of the relative displacements of the floors of the structure.1 Modal Control Consider a seismically excited structure controlled with m MR dampers. and Λ is the matrix determined by the placement of MR dampers in the structure. f = [ f1. Assuming that the forces provided by the control devices are adequate to keep the response of the primary structure from exiting the linear region. and stiffness matrices. Moreover. Γ is the column vector of ones. 2.…. After the implementation of the modal control scheme. &&g is x ground acceleration. f2. This equation can be written in the statespace form as .1 Modal Control Scheme for MR Dampers In this section.
2 ωn . Ω 2 is the diagonal matrix listing ω1 . The displacement can be expressed as the linear combination x (t ) = ∑ φr ηr (t ) = Φη . 2. we obtain T T && & ηr + 2ζ r ωr ηr + ωr2 ηr = φr Λf − φr MΓ&&g .1). φr is a r th eigenvector.5) where ζ r are modal damping ratios. and E′= − Φ T MΓ . r =1 n r = 1. Φ is a eigenvector set.3) where ηr (t ) is a r th modal displacement. The eigenvectors are orthogonal and can be normalized so as to satisfy the orthonormality conditions φsT Mφr = δrs . x r = 1.…. and v is a measurement noise vector.2b) where z is a state vector.….4) where δrs is the Kronecker delta and ωr is a natural frequency. n (2. (2. n (2. r = 1.Chapter 2 Modal Control Scheme 19 & z = Fz + Gf + N&&g x (2.….5) can be written in the matrix form as && & x η(t ) + ∆η(t ) + Ω 2 η(t ) = B' f (t ) + E ' &&g (2. 2.6) can be written in the modal spacestate form as . φsT Kφr = ωr2 δrs . 2.2a) y = Hz + Mf + v (2.…. multiplying by φrT and considering orthogonal condition between eigenvectors. Thus inserting (2.3) into (2. B′= Φ T Λ . (2. n (2. and η is a modal displacement vector. y is a vector of measured outputs.6) 2 where ∆ is the diagonal matrix listing 2ωr ζr .
11) are the 2l×2l. only a limited number of lower modes are controlled. respectively. 2l×m matrixes and a 2l×1 vector.7a) y (t ) = Cw (t ) (2. l controlled modes can be selected with l < n and the displacement may be partitioned into controlled and uncontrolled parts as x (t ) = xC (t ) + x R (t ) (2.7) can be rewritten & wC (t ) = AC wC (t ) + BC f (t ) + EC &&g x (2. (2.10b) 0 AC = 2 − ΩC IC 0 0 . Then. respectively. B = B' .Chapter 2 Modal Control Scheme 20 & w (t ) = Aw (t ) + Bf (t ) + E&&g x (2. BC = B' . E = E' − ∆ (2.7b) & where w(t) = [ η T η T ] is the modal state vector and 0 A= 2 − Ω I 0 0 .10a) yC (t ) = C C wC (t ) where wC is a 2ldimensional modal state vector by the controlled modes and (2. Hence. We refer to the uncontrolled modes as residual.9) where xC and xR represent the controlled and uncontrolled displacement vector. .8) In modal control. EC = E' − ∆C C C (2.
1996a. Although a variety of approaches may be used to design the optimal controller.Chapter 2 Modal Control Scheme 21 For a feedback control.c).13) where Vmax is the voltage to the current driver associated with saturation of the MR effect in the physical device. in using the control law given by (2. Thus. control gain matrix KC should be decided. in which a force feedback loop is incorporated to induce the force in the MR damper fi to generate approximately the desired optimal control force fCi. 1996a) is used. 2. Only the control voltage vi can be directly controlled.12) where KC is an m×2l control gain matrix. Because the force generated in the i th MR damper depends on the responses of the structural system. the control vector is related to the modal state vector according to f (t) = −KC wC(t) (2.1. To this end. Note that. the closedloop modal equations are not independent. and H( ) is the Heaviside step function.2 Design of Optimal Controller Referring to the discussions in above section.b. x For the controller design. the strategy of the clippedoptimal control (Dyke et al. the i th command signal vi is selected according to the control law vi = Vmax H [( f C i − f i )fi ] (2. H2/LQG (Linear Quadratic Gaussian) methods are advocated because of their successful application in previous studies (Dyke et al.12). and an infinite horizon performance index is chosen that weights the modal states by controlled modes such as . &&g is taken to be a stationary white noise. the MR damper cannot always produce the desired optimal control force fCi.
Q is the 4 × 4 diagonal matrix.Chapter 2 Modal Control Scheme 22 τ 1 T J = lim E ∫ (wC Q wC + uT Ru )dt 0 τ →∞ τ (2. qmd1 Q= 0 qmd2 qmv1 0 qmv2 (2.17) ˆ where B = [ L B − LD] . and S &x&g &x&g / S vivi = γ = 100 . Here.14) where R is a 2 × 2 identity matrix because the numerical example has two MR dampers. For example. the controller is ˆ GC ( s ) = K C [ sI − ( AC − LC C )]−1 BC (2. when the lowest one mode is selected for calculating the modal control action. Therefore. KC is the state feedback gain matrix for the deterministic regulator problem given by . It should be noted that the size of Q is reduced from 2n × 2n to 2l × 2l because the limited lower modes are controlled. and Q is a 2l × 2l diagonal matrix. statistically independent Gaussian white noise processes. Then. When the lowest two modes are controlled.15) where qmd is a weighting element for a modal displacement and qmv is for a modal velocity. Q is a 2 × 2 diagonal matrix such as q Q = md 0 0 qmv (2.16) The measurement noise is assumed to be identically distributed. it can be said that it is more convenient to design the smaller weighting matrix of modal control.
1. 1990). Moreover.20) SA'C + AC S − SC'C C C S + γ EC E'C = 0 (2. Hence. since the foundation of the structure is moving with the ground. we consider an observer.3 Modal State Estimation An observer for modal state estimation should be provided. A modal control method using the full state feedback may not be practical for a structural system involving a large number of DOFs. it is ideal to use the acceleration feedback because accelerometers can readily provide reliable and inexpensive measurements of accelerations at arbitrary points on the structure .18) PAC + A'C P − PB 'C BC P + C 'C QCC = 0 and (2. since the control implementation may requires a large amount of sensors.19) L = (C C S )' where S is the solution of the algebraic Ricatti equation given by (2.Chapter 2 Modal Control Scheme 23 K C = B 'C P where P is the solution of the algebraic Ricatti equation given by (2. is desirable. accurate measurements of displacements and velocities are difficult to achieve directly in fullscale applications. To estimate the modal state vector wC (t) from the measured output y(t).21) 2. particularly during seismic activity. Thus a modal control scheme that uses a modal state estimation. since real sensors may not estimate the full modal states directly or the system may be expensive to prepare the sensors for the full states. Luenberger observers are used for low noisetosignal ratios and KalmanBucy filters for high noisetosignal ratios (Meirovitch.
26) .24) (2.22).25) Upon obtaining the estimated controlled modal state from (2. we can write a modal observer in the form & ˆ ˆ ˆ wC (t ) = AC wC (t ) + ΒC f (t ) + EC &&g + L [ y (t ) − C C wC (t ) − DC f (t )] x (2. b).23) CC = [ M −1 K M −1C ] ΦC 0 −1 and DC = M Λ 0 ΦC (2. but also the state feedback including velocities and displacements. CC in (2. For modal state estimation from the displacements. 1996a. is implemented for the modal state estimation using a KalmanBucy filter.Chapter 2 Modal Control Scheme 24 (Dyke et al. the acceleration feedback is considered. we compute the feedback control forces ˆ f (t) = −KC wC (t ) (2.22) is as follows.22) ˆ where wC (t ) is the estimated controlled modal state and L is the optimally chosen observer gain matrix by solving a matrix Riccati equation. which assumes that the noise intensities associated with earthquake and sensors are known. CC = [ 0 Φ C ] For control with the acceleration feedback. In any case. CC is changeable according to the signals that are used for the feedback and DC is generally zero except the acceleration feedback. Not only. (2. CC = [Φ C 0 ] For control with the velocity feedback.
the uncontrolled modes are ignored. Substituting (2.30) can be rearranged (2. we obtain & ˆ wC (t ) = AC wC (t ) − BC K C wC (t ) + EC &&g x (2. residual modes can be written & w R (t ) = AR w R (t ) + BR f (t ) + E R &&g x (2.10a) and considering (2. In reality.31) .29) and (2.22).28).29b) Moreover.28) where wR is a residual state vector by uncontrolled modes.27) To examine the effect of the control forces on the uncontrolled modes. however.27) into (2.Chapter 2 Modal Control Scheme 25 Until now. the sensor signals will include contributions from all the modes.26) into (2. so that the output vector is corrected to y = C w (t ) = C C wC (t ) + C R w R (t ) (2.30) x Then the error vector is defined ˆ eC (t ) = wC (t ) − wC (t ) so that (2.26) and (2. we can write the observer equation in the form & ˆ ˆ ˆ wC (t ) = ( AC − BC K C ) wC (t ) + LC C [ wC (t ) − wC (t ) ] + LC R w R (t ) + EC &&g (2. substituting (2.29a) & ˆ w R (t ) = AR w R (t ) − BR K C wC (t ) + E R &&g x (2.
a small amount of damping inherent in the structure is often sufficient to overcome the observation spillover effect. However. If C R is zeros.33) Note that the term − BR K C in (2.Chapter 2 Modal Control Scheme 26 & wC (t ) = ( AC − BC K C ) wC (t ) − BC K C eC (t ) + EC &&g x & w R (t ) = − BR K C wC (t ) + AR w R (t ) − BR K C eC (t ) + E R &&g x (2. we conclude that control spillover cannot destabilize the system. 1978). however. 1983). Normally. the above system cannot satisfy the separate principle because the term LCR affects eigenvalues of the controlled system by the observer.33) is responsible for the excitation of the residual modes by the control forces and is known as control spillover (Balas. This effect is known as observation spillover and can produce instability in the residual modes. the term − BR K C has no effect on the eigenvalues of the closedloop system.32) & eC (t ) = ( AC − LC C )eC (t ) + LC R w R (t ) (2. At any rate. observation spillover can be eliminated if the sensor signals are prefiltered so as to screen out the contribution of the uncontrolled modes (Meirovitch.(Meirovitch and Baruh.32) can be written in the matrix form & wC (t ) AC − BC K C w (t ) = − B K & R C R eC (t ) 0 & 0 AR LC R − BC K C wC (t ) E R − BR K C w R (t ) + E R &&g x AC − LC C eC (t ) 0 (2. 1990) . although it can cause some degradation in the system performance. which means the sensor signal only include controlled modes. Hence.
(2. Hz. Substituting (2.33) in above section should be further improved for eliminating the observable spillover.Chapter 2 Modal Control Scheme 27 2.37) The pole of the lowpass filter dynamics can be placed by proper selection of the parameters.37). Gy.34) y f ( jω) = H y ( jω) y ( jω) (2. My.38) . A lowpass filter is introduced to measure the filtered response vector yf defined as & z (t ) = Fz z (t ) + G y y (t ) y f = H z z (t ) + M y y (t ) or in the frequency domain (2. which represents the residual modal state.36) If the lowpass filter dynamics Hy(jω) can be selected as a diagonal matrix.35). Fz. the new sensor dynamics becomes y f ( jω) = H y ( jω)[C C wC ( jω) + C R w R ( jω)] (2.35) where H y ( jω) = [ H z ( jωI − Fz ) −1 G y + M y ] . may have the following characteristics. The second term of righthand side of (2.1.27) into (2.4 Elimination of Observable Spillover (2. then the rolloff can be occurred forth the lowest modal frequency of the residual dynamics.36) becomes y f ( jω) = C C [ H y ( jω) wC ( jω)] + C R [ H y ( jω) w R ( jω)] (2.  H y ( jω) w R ( jω)  ≅ ε1  w R ( jω)  for 0 < ω ≤ ∞ (2.
Hence. may also have the following characteristics. the controlled system matrix in (2. In other words.37).41) into (2. the controlled modal states in (2. Thus.33) becomes (2.41) & wC (t ) AC − BC K C w (t ) = − B K & R C R eC (t ) 0 & 0 AR ε2 − BC K C wC (t ) EC − BR K C w R (t ) + E R &&g x AC − LC C eC (t ) 0 (2.39). which represents the controlled modal state.  H y ( jω) wC ( jω)  ≅  wC ( jω)  for 0 < ω ≤ ∞ (2. . the observable spillover does not occur in this controlled system.38) and (2.34).39) From (2. and the residual modal states may be also attenuated by their natural damping.40) y f (t ) ≅ C C wC (t ) + ε1C R w R (t ) + O (ε ) Substituting (2. the new sensor dynamics yf can be rewritten as y f ( jω) ≅ C C wC ( jω) + ε1C R w R ( jω) + O (ε ) or in time domain (2.22) may be suppressed by a welldesigned control input.42) where ε2 = ε1 LC R ≅ 0 . the first term of righthand side of (2.Chapter 2 Modal Control Scheme 28 where ε 1 ≅ 0 . the separate principle can be applied in the design of observer gain since the term L in ε2 no longer contributes to the characteristics of the system. Otherwise.
2.Chapter 2 Modal Control Scheme 29 2. a numerical example is considered in which a model of a sixstory building is controlled with four MR dampers (Fig. Two MR dampers are rigidly connected between the ground and the first floor.2 Numerical Example To evaluate the proposed modal control scheme for use with the MR damper.1). and two MR dampers are rigidly connected between the first and second floors.1 Schematic diagram of the MR damper implementation (Jansen and Dyke 2000) . This numerical example is the same with that of Jansen and Dyke (2000) and is adopted for direct comparisons between the proposed modal control scheme and other control algorithms. Figure 2.
the model of the structure is subjected to the NS component of the 1940 El Centro earthquake.2 and 2. mi.227 N/(cm/sec2). the stiffness of each floor. and the maximum voltage input to MR devices is Vmax = 5V. and n = 2. in frequency domain. 1997a).0064 Nsec/cm. c0b = 0. The maximum of the normalized interstory drift is .7) by defining the mass of each floor.86 N/cmV.2. αa = 8. it can be seen that the first mode is dominant in relative displacement and velocity of the first floor. as 0.0052 Nsec/cmV. Because the building system considered is a scaled model. however. From Fig. In simulation. and a damping ratio for each mode of 0. as 297 N/cm. The second evaluation criterion is a measure of the reduction in the interstory drift. Figs. The first evaluation criterion is a measure of the normalized maximum floor displacement relative to the ground. whereas the lowest three modes are dominant in the absolute acceleration. b = 300 cm2. A = 120. 2. αb = 8. given as x (t ) J 1 = max imax t. ki. i x (2.8% the weight of the entire structure. 2. 2. In Fig.Chapter 2 Modal Control Scheme 30 Each MR damper is capable of producing a force equal to 1. The various control algorithms were evaluated using a set of evaluation criteria based on those used in the second generation linear control problem for buildings (Spencer et al.43) where xi(t) is the relative displacement of the i th floor over the entire response. it will be possible to reduce the responses through modal control that control using the lowest one or two modes. The governing equations can be written in the form of (2. the amplitude of the earthquake was scaled to ten percent of the fullscale earthquake.66 N/cm. respectively. Thus.5%. and xmax denotes the uncontrolled maximum displacement.3 show the uncontrolled responses of the first and sixth floors. g = 300 cm2.. MR damper parameters used in this study are c0a = 0.3. we can find that the first mode is dominant in all responses of the sixth floor.
The corresponding uncontrolled max x max responses are as follows: xmax = 1. i dn (2.00981 cm. 2000). The final evaluation criteria considered x max in this study is a measure of the maximum control force per device. given by f (t) J 4 = max i t. &&a = 146. The resulting evaluation criteria are presented in Table 1 for the control algorithms previously studied (Jansen and Dyke. To compare the performance of the semiactive system to that of comparable passive systems. denoted &&a (t ) .Chapter 2 Modal Control Scheme 31 d (t ) /h  J 2 = max i max i t.95 cm/sec2. &&ai (t ) .45) where the absolute accelerations of the ith floor. i xa (2. are normalized by the peak x uncontrolled floor acceleration. The third evaluation criterion is a measure of the normalized peak floor accelerations.46) where W is the total weight of the structure (1335 N). normalized by the weight of the structure.313 cm. The results of passiveoff (0V) and passiveon (5V) configurations are included. i W (2.44) where hi is the height of each floor (30cm). di(t) is the interstory drift of the above ground max floors over the response history. d a = 0. given by && (t ) x J 3 = max aimax && t. . two cases are considered in which MR dampers are used in a passive mode by maintaining a constant voltage to the devices. The numbers in parentheses indicate the percent reduction as compared to the best passive case. and d n denotes the normalized peak interstory drift in the uncontrolled response.
6 0.8 PSD 0.8 0.2 0 0 1 2 3 4 10 x 105 1. Hz 7 8 9 10 1.2 Frequency responses of the first floor for the uncontrolled structures under the scaled El Centro earthquake .Chapter 2 Modal Control Scheme 32 102 1.6 Power Spectrum of Relative Displacement 1.2 1 PSD 0. Hz 7 8 9 10 Frequency.4 0.4 0.6 0.2 0 0 x 10 6 1 2 3 4 5 6 frequency. Hz Figure 2.4 1.2 Power Spectrum of Velocity 1 0. Hz 7 8 9 10 2 Power Spectrum of Acceleration 1.5 PS D 1 0.4 4 5 6 frequency.5 0 0 1 2 3 4 5 6 frequency.
5 0 0 x 10 6 1 2 3 4 5 6 frequency.0 PSD 1. Hz 7 8 9 10 x 10 Power Spectrum of Velocity 2 1.0 Power Spectrum of Relative Displacement 2.5 PSD 1 0.5 2.5 1 0. Hz Figure 2. Hz 7 8 9 10 16 Power Spectrum of Acceleration 14 12 10 PSD 8 6 4 2 0 0 1 2 3 4 5 6 frequency.5 0 0 1 5 2 3 4 5 6 frequency.3 Frequency responses of the sixth floor for the uncontrolled structures under the scaled El Centro earthquake .Chapter 2 Modal Control Scheme 33 102 3. Hz 7 8 9 10 Frequency.
788(+13) 0.756(−16) 1.01095 0.696 0.862 0.631(+24) 0.599(−20) 1.25(+38) J4 0.0178 0.791(+13) 0.39(+53) 1.00292 0.06(+17) 0.620(−11) 0.0178 0.421(−17) J2 0.0121 0.00(+11) 1.548(+8) 0.0178 1.Chapter 2 Modal Control Scheme 34 Table 2.801 0.449(−11) 0.0178 0.636(−29) 1.06(+17) 0.41 J3 0.548(−21) 0.0178 (* Jansen and Dyke 2000) .686(+35) 0.640(−8) 0.904 0.0178 0.326(−35) 0.1 Normalized controlled maximum responses due to the scaled El Centro earthquake* Control strategy Passiveoff Passiveon Lyapunov controller A Lyapunov controller A Decentralized bangbang Maximum energy dissipation Clippedoptimal A Clippedoptimal B Modified homogeneous friction J1 0.506 0.405(−20) 0.547(−21) 0.
four modal control designs with different capabilities are considered.4. J1.3 to 2. JT is the summation of evaluation criteria. 4.6.2. 2. J3 and JT. respectively. displacements. respectively.2 to 2.4 represents the results of the stochastic response analysis for the acceleration feedback case. J2. Figs. 2. and DJT with displacement feedback and VJ1. In Fig. For each weighting. By using the controller (H2/LQG) with designed weighting matrices from Fig. we can get the results in Table 2. J2 and J3. From the variations of JT.5 and 2.2.6 summarize the results for each minimum evaluation criteria of the designed weighting matrices from Figs. J2 is at qmd = 1 and qmv = 500.2 to 2. DJ2. Using each structural measurement. in Tables 2. AJ3 and AJT with acceleration feedback use a weighting that minimize the evaluation criteria J1. 2. In the lowest two modes case. Tables 2. AJ2. qmd1 = qmd2 = qmd and qmv1 = qmv2 = qmv. the lowest one and two modes cases are given in Tables 2. we place identical weighting on the each mode. In Fig. J2 and J3. DJ3.Chapter 2 Modal Control Scheme 35 For modal control. respectively. 2. the modal controller AJ1. the modal controller DJ1. we can find the weighting for reduction of related responses.4.4. J2.4. In Tables 2. a Kalman filter estimates the modal states. Fig. J3 is at qmd = 2200 and qmv = 100 and J4 is at qmd = 500 and qmv = 600. Designer can decide which to use according to control objectives. VJ2. 2.4. three cases of the structural measurements are considered. The variations of each evaluation criteria for increasing weighting parameters are shown in a 3dimensional plot. velocities and accelerations. J3 and JT. J1 is evaluation criteria for the maximum displacement. In Table 2. Previously mentioned. it can be seen that J1 is minimum at qmd = 400 and qmv = 1500.5 and 2. . we can find the weighting for reduction of overall structural responses whereas from J1. J2 is for the maximum interstory drift and J3 is for the maximum acceleration.5 and 2. For each feedback case.VJ3. and VJT with velocity feedback use a weighting which minimize the evaluation criteria J1.6 represent the results for the displacement and velocity feedback cases.
Chapter 2 Modal Control Scheme 36 J1 J2 qmd qmv qmd qmv J3 JT =J1+ J2 + J3 qmv qmd qmd qmv Figure 2.4 Variations of evaluation criteria with weighting parameters for the acceleration feedback .
5 Variations of evaluation criteria with weighting parameters for the displacement feedback .Chapter 2 Modal Control Scheme 37 J1 J2 qmd qmv qmd qmv J3 JT =J1+ J2 + J3 qmd qmv qmd qmv Figure 2.
Chapter 2 Modal Control Scheme 38 J1 J2 qmd qmv qmd qmv J3 JT =J1+ J2 + J3 qmv qmd qmv qmd Figure 2.6 Variations of evaluation criteria with weighting parameters for the velocity feedback .
2. although the maximum displacement increased.4 to 2. If further reductions in interstory drift and acceleration are desired in the controller.4. respectively. resulting in the lowest values of all cases considered here. Notice that the designer has some versatility depending on the control objectives for the particular structure under consideration. The modal controller AJT and VJT do not achieve any lowest value of evaluation criteria. but have competitive performance in all evaluation criteria. In fact. modal controller AJ2 and AJ3 can achieve the reductions in the interstory drift and absolute acceleration of 30% and 23%. every lowest value of evaluation criteria occurs at the lowest one mode case. The performance of the proposed modal control scheme is generally better than that of other control strategies.1 to 2. . modal controller VJ1 using the lowest two modes and VJ3 achieve reductions in relative displacement and absolute acceleration of 41% and 30%. except the modal controller VJ1 that achieves further reductions by 6% from one mode case (reductions of 41% over the best passive case) in the relative displacement. The modal controller D compared with the modal controller A and V appears to be worse in achieving reductions. The reduction by modal controller AJ2 is resulting in the lowest interstory drift of all cases considered here. the modal controller AJ1 achieves a 39% reduction in the relative displacement as compared to the better passive case. Comparing the lowest one mode case with twomode case. which agrees with the fact that the variations of evaluation criteria are more sensitive to weighting parameter qmv than qmd from Figs.4.Chapter 2 Modal Control Scheme 39 The calculated evaluation criteria for various control strategies are compared in Tables 2. respectively. The results show that the modal controller A and V appear to be quite effective in achieving significant reductions in both the maximum displacement and interstory drift over the passive case. In Table 2. over the best passive cases.6.
413(18) 0.Chapter 2 Modal Control Scheme 40 Table 2.543(22) 0.0178 0.310(39) 0.0178 Modal control AJ1 2 modes 1 mode 2 modes 1 mode 2 modes 1 mode 2 modes Modal control AJ2 (qmd=1.488(30) 0.585(16) 0.398(21) 0.697(23) 0.0178 0.07(+18) 1. qmv=500) (qmd=2200.549(+8) 0.05(+16) 0.510(27) 0.876(3) J4 0.2 Normalized controlled maximum responses of the acceleration feedback due to the scaled El Centro earthquake Control strategy (qmd=400.618(11) 0.741(18) 0.0178 0.485(30) 0. qmv=600) .392(23) 0.533(23) J3 1.0178 0.548(+8) 0.529(24) 0.823(9) 0.781(14) 0.380(25) 0.0178 0. qmv=100) Modal control AJ3 Modal control AJT (qmd=500.423(16) J2 0.0178 0.0178 0.870(4) 0. qmv=1500) J1 1 mode 0.
06(+17) 0.702(+39) 0.408(19) 0. qmv=4900) J1 0.04(+15) J4 0.0178 0.06(+17) 0.796(12) 0.0178 0.0178 0.769(15) 1.721(20) 1.0178 0.504(28) 0.504(28) 0.Chapter 2 Modal Control Scheme 41 Table 2.560(20) 0.0178 Modal control DJ1 1 mode 2 modes 1 mode 2 modes 1 mode 2 modes 1 mode 2 modes (qmd=100.728(+5) 0.678(+34) 0.329(35) J2 0.403(20) 0.qmv=4700) Modal control DJT .510(27) J3 0.325(36) 0. qmv=4900) Modal control DJ2 (qmd=200.3 Normalized controlled maximum responses of the displacement feedback due to the scaled El Centro earthquake Control strategy (qmd=100.560(20) 0.0178 0.566(19) 0.0178 0.689(1) 0.765(15) 1.325(36) 0. qmv=4900) Modal control DJ3 (qmd=3300.403(20) 0.0178 0.671(26) 0.
qmv=100) Modal control VJ3 (qmd=600.510(27) 0.487(30) 0.354(30) 0.874(3) 0.351(31) 0.301(41) 0.383(24) 0.qmv=500) Modal control VJT .0178 0.541(+7) 0.632(30) 0.583(16) 0.827(9) J4 0.510(27) J3 1.327(35) 0.611(12) 0.554(20) 0.530(24) 0.553(39) 0.06(+17) 1.0178 0.323(36) J2 0.502(28) 0.522(+3) 0.0178 0.0178 2 modes 1 mode 2 modes 1 mode 2 modes 1 mode 2 modes Modal control VJ2 (qmd=1.4 Normalized controlled maximum responses of the velocity feedback due to the scaled El Centro earthquake Control strategy Modal control VJ1 (qmd=700.Chapter 2 Modal Control Scheme 42 Table 2.0178 0.07(+18) 0. qmv=800) J1 1 mode 0.0178 0. qmv=400) (qmd=1300.0178 0.825(9) 0.0178 0.941(+4) 0.
it is more convenient to design the smaller weighting matrix of modal control. Based on these results. although resulting responses varied greatly depending on the choice of measurements available and weightings.Chapter 2 Modal Control Scheme 43 2. respectively. . Modal control reshapes the motion of a structure by merely controlling a few selected vibration modes. Moreover. a lowpass filter was used to eliminate spillover problem. Further studies are underway to examine the influence of the number of controlled modes on the control performance. a modal control scheme was applied together with a Kalman filter and a lowpass filter. Therefore. respectively.3 Summary of Results In this study. in designing phase of controller. The modal controller A and V achieved significant reductions in the responses. The responses of the system to a scaled El Centro earthquake excitation were found for each controller through a simulation of the system. Hence. 41%. A Kalman filter considered three cases of the structural measurement to estimate modal states: displacement. To this end. The modal controller AJT and VJT fail to achieve any lowest value of evaluation criteria. VJ1 and VJ3 achieve reductions (30%. In a numerical example. 30%) in evaluation criteria J1. This is one of the important benefits of the proposed modal control scheme. the size of weighting matrix Q was reduced because the lowest one or two modes were controlled. velocity. modal control was implemented to seismically excited structures using MR dampers. resulting in the lowest values of all cases considered here. but have competitive performance in all evaluation criteria. J2 and J3. a sixstory structure was controlled using MR dampers on the lower two floors. The modal controller AJ2. The numerical results show that the motion of the structure was effectively suppressed by merely controlling a few lowest modes. and acceleration. the proposed modal control scheme is found to be suited for use with MR dampers in a multiinput control system.
More details of system matrices can be found in Dyke et al (2003) and Ohtori et al (2000. and Λ is the matrix determined by the placement of MR dampers in the structure. & M&& + Cx + Kx = Λf − MΓ &&g x x (3.2) y = Cz + Df + v (3. &&g is one x dimensional ground acceleration. fn ]T is the vector of measured control forces generated by n MR dampers. . 2002).1 Control System Consider a seismically excited structure controlled with n MR dampers. f2. Γ is the column vector of ones.….3) where z is a state vector. f = [ f1. This equation can be written in statespace form as & z = Az + Bf + E&&g x (3. The equation of motion can be written.Chapter 3 Maximum Energy Dissipation Algorithm 44 CHAPTER 3 MAXIMUM ENEGRY DISSIPATION ALGORITHM 3.1) where x is vector of the relative displacements of the floors of the structure. y is the vector of measured outputs. and v is a measurement noise vector.
Several types of controloriented dynamic models have been investigated for modeling MR dampers. 1997a). Herein. is considered for describing the behavior of the MR damper (Figure 3.1 Control Devices The MR damper with capacity of 1000KN is considered as control devices.4) & & & z = −γ  x  z  z n−1 − β x  z n + Ax (3. which is numerically tractable and has been used extensively for modeling hysteretic system. Figure 3. To accurately predict the behavior of controlled structure. the BoucWen model is considered. the degree of linearity in the unloading and the smoothness of the transition from the preyield to the postyield region .Chapter 3 Maximum Energy Dissipation Algorithm 45 3.5) By adjusting the parameters of the model γ.1 Mechanical model of the MR damper The force generated by the damper is given by & f = αz + c0 x where the evolutionary variable z is governed by (3. and A. n. β.1. an appropriate modeling of MR dampers is essential.1). The BoucWen model (Spencer et al.
43 N s/(cm⋅V) 300 s1 300 m2 300 m2 120 1 η γ β A n .15 N s/m 1.962e5 N/(cm⋅V) 4. (2003) for the cablestayed bridge. α = α a + α bu and c0 = c0 a + c0bu (3.1. Some of the model parameters depend on the command voltage u to the current driver as follows. Each parameter is adopted from Yoshida and Dyke (2002) for the nonlinear benchmark building and from Moon et al.1 Parameters for MR damper model Parameter αa αb c0a c0b 1.6) Parameters for both benchmark problems are listed in Table 3.087e5 N/cm 4.41 N/(m⋅V) 0.2 1 Value For nonlinear building For cablestayed bridge 500 N/m 671.Chapter 3 Maximum Energy Dissipation Algorithm 46 can be controlled.0 N s/(cm⋅V) 50 s1 3 cm2 3 cm2 1.40 N s/cm 44. Table 3.
9) Note that MEDA is very simple because only local measurements (i. the velocity and control force) are required to implement this control law. the origin is stable in the sense of Lyapunov..2 Maximum Energy Dissipation Algorithm for MR Damper This control algorithm is presented as a variation of the decentralized bangbang approach proposed by McClamroch and Gavin (1995). Jansen and Dyke (2000) instead consider a Lyapunov function that represents the relative vibratory energy in the structure as in V= 1 T 1 x Kx + x T Mx 2 2 (3.e. Using (3. the only way to directly effect V is through the last term containing & the force vector f.7). if the rate of change of the Lyapunov function & V ( x ) is negative semidefinite.1. In (3.Chapter 3 Maximum Energy Dissipation Algorithm 47 3.9). Lyapunov’s direct approach requires the use of a Lyapunov function. which must be a positive definite function of the states of the system x. denoted V(x). the following control law is obtained: & v i = Vmax H (− xΛi f i ) where Λi is ith column of the Λ matrix.7) According to Lyapunov stability theory. complex . there is no design parameter to decide.8) & In this expression. (3. which is essential part in other control laws. the Lyapunov function was chosen to represent total vibratory energy in the system. the rate of change of the Lyapunov function is then & & & & V = x T Kx + x T M( −Cx − Kx − MΓ &&g + Λf) x (3. In other words. fi is i th column of the f matrix. To control this term and make V as large and negative as possible. In the decentralized bangbang approach.
Otherwise. . especially for the largesize civil structures. it can be said that it is more convenient to use MEDA for structural control. the more design parameters are considered. Therefore.Chapter 3 Maximum Energy Dissipation Algorithm 48 design process can be skipped. the more structures are complex. This is the important benefit of using MEDA.
it was found that fully restraining the deck in the longitudinal direction would result in unacceptably large stresses. Earthquake load combinations in accordance with American Association of State Highway and Transportation Officials (AASHTO) division IA specifications were used in the design. Various designs were considered. 1996. The bridge is currently under construction and is to be completed in 2003.Chapter 3 Maximum Energy Dissipation Algorithm 49 3. respectively.2 Benchmark Problems In this study. The cablestayed bridge and the highrise nonlinear building model are representative structures of civil engineering. we exploit MEDA for civil engineering applications. 1820. In early stages of the design process.1 Benchmark CableStayed Bridge At the Second International Workshop on Structural Control (Dec. Missouri. including full longitudinal restraint at the tower piers. 2002). For the completeness. Hong Kong). the loading case governing the design was determined to be due to seismic effects. 3. we consider two kinds of benchmark problem: a cablestayed bridge and a 20story nonlinear building. The cablestayed bridge used for this benchmark study is the Missouri 74–Illinois 146 bridge spanning the Mississippi River near Cape Girardeau. Seismic considerations were strongly considered in the design of this bridge due to the location of the bridge (in the New Madrid seismic zone) and its critical role as a principal crossing of the Mississippi River. it was determined that incorporating force transfer devices would provide the most efficient solution. this section briefly summarizes both benchmark problems. the Working Group on Bridge Control developed plans for a "first generation" benchmark study on bridges. Based on examination of the various designs. . More details can be found in Dyke et al (2003) and Ohtori et al (2000. and passive isolation. no longitudinal restraint. 1997). designed by the HNTB Corporation (Hague. Using both benchmark problems.2. When temperature effects were considered.
4 m at pier 2 and 108. Additionally.3. Each tower supports a total 64 cables. and 12 additional piers in the approach bridge from the Illinois side.92 MPa. The 128 cables are made of high–strength. A cross section of the deck is shown in Fig.6 m in length.Chapter 3 Maximum Energy Dissipation Algorithm 50 Sixteen 6. Steel ASTM A709 grade 50W is used. and the approach on the Illinois side is 570 m. in the transverse direction earthquake restrainers are employed at the connection between the tower and the deck and the deck is constrained in the vertical direction at the towers. of 37. The bearings at bent 1 and pier 4 are designed to permit longitudinal displacement and rotation about the transverse and vertical axis. It has a total length of 1205.3 cm2. The approach bridge from the Illinois side is supported by 11 piers and bent 15 which are made of concrete. 128 cables. a concrete barrier is located in the center of the bridge. The deck is composed of steel beams and prestressed concrete slabs.67 MN shock transmission devices are employed in the connection between the tower and the deck.7 m in length. 3. These devices are installed in the longitudinal direction to allow for expansion of the deck due to temperature changes. the side spans are 142.5 m at pier 3. The Hshaped towers have a height of 102. low–relaxation steel (ASTM A882 grade 270). The bridge has four lanes plus two narrower bicycle lanes. fc'. Additionally. the bridge is composed of two towers.2. The smallest cable area is 28. and a railing is located along the edges of the deck. The cables are covered with a polyethylene piping to resist corrosion. 3. for a total width of 29. As shown in Fig. The main span is 350. with an fy of 344 MPa. The deck consists of a rigid diaphragm made of steel with a slab of concrete at the top.36 MPa. The concrete slabs are made of prestressed concrete with a fc' of 41.5 cm2 and the largest cable area is 76. Under dynamic loads these devices are extremely stiff and are assumed to behave as rigid links. Soilstructure interaction is not expected to be an issue with this bridge as the foundations of the cablestayed portion is attached to bedrock. .8 m. The towers are constructed of reinforced concrete with a resistance.3 m.
3 Cross section of bridge deck.2 Drawing of the Cape Girardeau Bridge Figure 3. .Chapter 3 Maximum Energy Dissipation Algorithm 51 Figure 3.
and a reduction is performed to reduce the size of the model to something more manageable.5812.5650. 134 nodal masses and 128 cable elements. However. and 0. A onedimensional ground acceleration is applied in the longitudinal direction.3723.4545.6965. 0. shown in Fig. 0. the constraints are applied.7094 Hz. 0. The finite element model employs beam elements.2666.6187. the second model will be formed as the evaluation model. 0. has a total of 579 nodes.7102. 0. 0. 0.6970.4683. and the control devices should connect the deck to the tower. the frequencies of this model are much lower than those of the nominal bridge model. 0. 3. To make it possible for designers/researchers to place devices acting longitudinally between the deck and the tower.5015.6490. a modified evaluation model is formed in which the connections between the tower and the deck are disconnected. 0. 0. A linear evaluation model is used in this benchmark study. cable elements and rigid links.5158. 0. Additionally.Chapter 3 Maximum Energy Dissipation Algorithm 52 Based on the description of the Cape Girardeau bridge provided in the previous section. 0. 0. The nonlinear static analysis is performed in ABAQUS® (1998). Subsequently. The towers are .6486. This direction is considered to be the most destructive in cablestayed bridges. 0. the bridge is assumed to be attached to bedrock.6687. As one would expect. If a designers/researcher specifies devices at these nodes. 420 rigid links. 1991).7203 Hz. The first ten frequencies of the evaluation model are 0. 162 beam elements. Note that the uncontrolled structure used as a basis of comparison for the controlled system. the stiffness matrices used in this linear model are those of the structure determined through a nonlinear static analysis corresponding to the deformed state of the bridge with dead loads (Wilson and Gravelle.3699. 0. and the effects of soil–structure interaction are neglected. 0. and 0.4. The finite element model.2899. The first ten frequencies of this second model are 0. a threedimensional finite element model of the bridge was developed in MATLAB® (1997). and the element mass and stiffness matrices are output to MATLAB® for assembly.1618. corresponds to the former model in which the decktower connections are fixed (the dynamically stiff shock transmission devices are present).
The cables are modeled with truss elements. In the case of variable sections. The FEM model described above is used directly in cases when the control devices are employed in the longitudinal direction between the deck and tower.cive. the average of the section is used for the finite element. 3. the model is modified by including four longitudinally directed. The system matrices are provided at the benchmark web site: http://wusceel. Boundary conditions restrict the motion at pier 1 to allow only longitudinal displacement (X) and rotations about the Y and Z axes. the Illinois approach was found to have a negligible effect on the dynamics of the cablestayed portion of the bridge. The use of the rigid links ensures that the length and inclination angle of the cables in the model agree with the drawings. Note that the program included with the benchmark files determines if the designer/researcher has placed devices in this location and builds the appropriate FEM model. In the finite element model the nominal tension is assigned to each cable. Additionally. 43 beam elements and 74 rigid links. the moment induced in the towers by the movement of the cables is taken into consideration in this approach. Constraints are applied to restrict the deck from moving in the lateral direction at piers 2. The uncontrolled structure used as a basis of comparison corresponds to this second case. If the designer/researcher employs no control device at these locations (in which case the shock transmission devices are included). 3. Note that the Illinois approach is not included in this model (see Fig.5). rigid links are used to connect the cables to the tower and to the deck (see Fig. 3 and 4.wustl.edu/quake . Because the attachment points of the cables to the deck are above the neutral axis of the deck. axially stiff beam elements that force the deck to move with the tower in the longitudinal direction.Chapter 3 Maximum Energy Dissipation Algorithm 53 modeled using 50 nodes. Because the bearing at pier 4 does not restrict longitudinal motion and rotation about the X axis of the bridge. and the attachment points of the cables to the tower are outside the neutral axis of the tower.2).
5 Finite element model of the towers .Chapter 3 Maximum Energy Dissipation Algorithm 54 Figure 3.4 Finite element model Figure 3.
sixteenth and eighteenth levels at 1.77 m in elevation.96 m. it was decided to pursue the nonlinear analysis for the seismically excited buildings. Each frame resists one half of the seismic mass associated with the entire structure. as a result of the success of the linear benchmark's presented. fourth. The floor system is comprised of 248 MPa steel wideflange beams acting compositely with the floor slab.wustl. Typical floortofloor heights (for analysis purposes measured from centerofbeam to centerofbeam) are 3. are located on the first. The building’s lateral loadresisting system is comprised of steel perimeter momentresisting frames (MRFs). The level below B1 is the second basement (B2).Chapter 3 Maximum Energy Dissipation Algorithm 55 3.e. 3. The floortofloor heights for the two basement levels are 3. Column splices. The nonlinear benchmark building considered here is the 20story benchmark building specified in the benchmark problem statement (http://wusceel. The column bases are modeled as pinned and secured to the ground (at the B2 level). The columns are 345 MPa steel. The corner columns are box columns. The building has two basement levels. Tokyo). and 80.2. The level directly below the ground level is the second basement (B1).48 m by 36. The levels of the 20story building are numbered with respect to the ground level (see Fig.49 m. which are seismic (tension) splices to carry bending and uplift forces.10 m on center. in both directions.edu/ quake).6). .2 Nonlinear Benchmark Building During the 2nd World Conference on Structural Control (June 28 – July 1.58 m in plan. with five bays in the northsouth (NS) direction and six bays in the eastwest (EW) direction. i. tenth.cive. The column lines employ threetier construction. Concrete foundation walls and surrounding soil are assumed to restrain the structure at the ground level from horizontal displacement. The interior bays of the structure contain simple framing with composite floors. 1998. thirteenth. The 20story benchmark structure is 30. seventh.65 m and for the ground level is 5.83 m above the centerline of the beam to column joint. monolithic column pieces are connected every three levels beginning with the first level. The interior columns of the MRF are wideflange. The bays are 6.
Chapter 3 Maximum Energy Dissipation Algorithm 56 Figure 3.6 Schematic of the 20story benchmark building .
Structural member nonlinearities are included to capture the inelastic behavior of buildings during strong earthquakes. for the second level to 19th level is 5.63´105 kg. thus ignoring the column panel zone. A bilinear hysteresis model is used to characterize the nonlinear bending stiffness of the structural members. accounting for the seismic mass of the floor slabs.11´107 kg. ceiling/flooring. of the ground level is 5. Nodes are located at beamtocolumn joints. . The seismic mass. 3. roofing and a penthouse located on the roof. for the first level is 5. and for the 20th level is 5. ceiling/flooring. The beam members extend from the centerline of column to centerline of column.32´105 kg. The 20story NS MRF is depicted in Fig. or weak. direction of the buildings) for the structures described in the previous section. partitions. inplane finite element models of the NS MRFs have been developed. roofing and penthouse are uniformly distributed at the nodes of each respective level assuming a lumped mass formulation.52´105 kg. Inertial loads.84´105 kg. mechanical/electrical. active and/or semiactive control devices can be implemented throughout these NS frames of the 20story structure and their performance assessed using the evaluation models in this section and the evaluation criteria identified in the Control Design section. The seismic mass of the above ground levels of the entire structure is 1.6. This benchmark study focuses on an inplane (2D) analysis of the benchmark structures. including both NS MRFs. including the steel framing. partitions.Chapter 3 Maximum Energy Dissipation Algorithm 57 The seismic mass of the structure is due to various components of the structure. The damping matrix is determined based on an assumption of Rayleigh damping. Passive. Based on the physical description of the 20story structures. and mass and stiffness matrices for each of the structures are determined. floor slabs. mechanical/electrical. The frames considered in the development of the evaluation models are the NS MRFs (the short. Elements are created between nodes to represent the beams and columns in the structure. This process is described in further detail in the following paragraphs. The beams and columns of the structures are modeled as planeframe elements.
Chapter 3 Maximum Energy Dissipation Algorithm
58
The 20story building frames contain column splices. The column joint of the splice story is located 1.83 m above the centerline of the beam. For simplicity the spliced columns are modeled as having uniform properties over the story height equal to the weighted average of the upper and lower column properties of that story. There is no node modeled at the splice. Each node has three degreesoffreedom (DOFs): horizontal, vertical and rotational. The 20story structures have 414 DOFs prior to application of boundary conditions/constraints, respectively. Each element, modeled as a plane frame element, contains two nodes and six DOFs. The length, area, moment of inertia, modulus of elasticity and mass density are predefined for each element. The elemental lumped mass and stiffness matrices are determined as functions of these properties (Sack 1989; Cook, et al. 1989). Global mass and stiffness matrices are assembled from the elemental mass and stiffness matrices by summing the mass and stiffness associated with each DOF for each element of the entire structure. Rotational inertia is ignored; thus, rotational mass is assigned a small value. The DOFs corresponding to fixed boundary conditions are then constrained by eliminating the rows and columns associated with these DOFs from the global mass and stiffness matrices. The simulation of the benchmark buildings is developed to represent control of the entire structure, including both NS MRFs and the entire mass of each structure. Researchers/designers should recognize that the control strategies applied within this study represent the structural control of the entire benchmark building being considered. The first 10 natural frequencies of the 20story benchmark evaluation model are: 0.261, 0.753, 1.30, 1.83, 2.40, 2.44, 2.92, 3.01, 3.63 and 3.68 Hz. These results are consistent with those found by others who have modeled this structure (Barrosa 1999; Spencer, et al. 1998a,b). Assuming the structures respond in the elastic range, transfer functions for the displacement and absolute acceleration at the top of each building from ground acceleration can be determined.
Chapter 3 Maximum Energy Dissipation Algorithm
59
3.3
Numerical Examples
In the previous section, we found that MEDA is very simple to implement. In this
section, we will examine the applicability of the MEDAbased semiactive control system from the viewpoint of the performance and the robustness. Through a series of numerical simulations of benchmark problems, the results are compared with those of other control algorithms: Clippedoptimal controller (CO; Yoshida and Dyke, 2002) and sliding mode controller (SMC; Moon et. al, 2003). Clippedoptimal controller was suggested and experimentally examined by Dyke et al. (1996a,b,c). The clippedoptimal control approach is to design a linear optimal controller that calculates desired control forces based on the measured structural responses and the measured control force applied to the structure. A force feedback loop is incorporated to induce the MR damper to generate approximately the desired optimal control force. Sliding mode controller was developed specifically for robust control of uncertain nonlinear systems. The fundamental idea of SMC is to design a controller to drive the state trajectory onto a sliding surface, where the motion is stable. Thus, SMC is known as a robust controller.
3.3.1 Control Performance
Table 3.2 shows the values of the evaluation criteria for the benchmark cablestayed bridge under various earthquakes. The definition of each evaluation criteria can be found in Dyke et al. (2003) and Ohtori et al. (2000, 2002). Each controller employs 24 MR dampers between the deck and abutment and the deck and tower of the bridge, all oriented to apply forces longitudinally. Four devices are located between each of the following pairs of nodes on bent 1 and pier 3; and, two devices are located between each of the following pairs of nodes on piers 2 and 4. Note that each controller is able to achieve a significant reduction in the base shear force when compared with the uncontrolled system; the base shears in MEDA are
Chapter 3 Maximum Energy Dissipation Algorithm
60
reduced to 33 ~ 59% levels in the peak value (J1) and to 23 ~ 46% levels in the normed values (J7) for the three earthquake excitations. Overturning moments are reduced to 26 ~ 56% levels in the peak values (J3) and to 23 ~ 45% levels in the normed value (J9). It is clear that most of structural responses generated by three earthquakes are controlled well. Further, the numerical results show that MEDA performs slightly better than two other controllers. Table 3.3 shows the evaluation criteria for the nonlinear benchmark building. Total 65 MR dampers are used. Four devices are located on the first eight stories, three devices are located on the next nine stories, and two devices are located on the top three stories. The building was subjected to the four earthquakes specified in the benchmark paper with various intensities. The interstory drifts in MEDA are reduced to 50 ~ 80% levels in the peak value (J1) and to 27 ~ 99% levels in the normed values (J4), which are better performances than those of clippedoptimal controller. MEDA, also, prevent the formation of plastic connections (J9) for halfscale of Northridge and Kobe earthquakes. However, note that MEDA fails to reduce the peak floor acceleration and the peak base shear, giving the magnification of 241% and 144%, respectively, in maximum value over the uncontrolled case, whereas clippedoptimal controller success. To improve the performance of MEDA for the building structure, the location and the number of MR dampers can be changed. An infinite number of configurations of MR dampers may be possible, which may result in choosing an optimal one for purpose. In this study, we chose that one device is installed on every second floor from the first floor, and then a total of 10 MR dampers are used. The building was subjected to the four earthquakes specified in the benchmark paper with various intensities. Table 3.4 shows the evaluation criteria for the nonlinear benchmark building with the modified configuration of MR dampers. Note that the magnifications of the peak floor acceleration and the peak base shear are reduced to 128% and 118% respectively in maximum value, but are still over the uncontrolled case. Furthermore, the number and the location of MR dampers could not be chosen at one
453 J2. Peak deviation of cable tension El Centro Mexico Gebze 0.108 1.331 0.393 0.903 0.020 0. However.395 0.147 1.300 0.018 0.557 0. Peak shear at deck level El Centro Mexico Gebze 1. Normed overturning moment El Centro Mexico Gebze 0.128 J10.011 0.012 0.233 0.558 0.2.578 2. which means MEDA for the building structure lose its advantage of simple method to use.2.185 0. Table 3.488 0. Normed base shear El Centro Mexico Gebze 0.397 0.193 0. Various numbers and locations are tried and then one case is selected.376 1.880 1.006 0.552 1.380 0.464 0.466 0.975 1. which is the most suitable for the purpose.282 2.146 1.355 J5.Chapter 3 Maximum Energy Dissipation Algorithm 61 time.090 1.205 0.953 0.079 0.266 J8.624 0.315 1.391 0.217 0.902 1.271 0.392 0.348 J11.123 J3.007 0.084 1.487 0.472 0.012 0. Normed deviation of cable tension El Centro Mexico Gebze 0.331 0.694 1.382 0.440 0.593 Gebze 0.060 0.315 0.234 0. Normed shear at deck level El Centro Mexico Gebze 0.720 1.234 0.143 J7.453 0.300 0.318 J9.381 0. Peak deck displacement El Centro Mexico Gebze 0.267 0. Peak moment at deck level El Centro Mexico Gebze 0.408 1.068 1.941 0.519 0.709 0.537 0.189 0.656 1.159 0. Comparisons of the evaluation criteria for benchmark cablestayed bridge Controller CO SMC MEDA Controller CO SMC MEDA Controller CO SMC MEDA Controller CO SMC MEDA Controller CO SMC MEDA Controller CO SMC MEDA J1.328 0. This process of MEDA for building structure corresponds to designing phase of other controller.391 0.469 0. MEDA for the cablestayed bridge is still simple and efficient controller as listed in Table 3.453 0.933 1.779 J6.064 1.255 0.286 0.372 0.053 0.415 0.883 1.552 0. Peak base shear El Centro Mexico 0.194 0.020 0.464 0. Normed moment at deck level El Centro Mexico Gebze 0.447 J4.142 0.056 0.577 0.010 . Peak overturning moment El Centro Mexico Gebze 0.179 1.
0.688 0.639 0.078 0.904 0.661 0.622 0.009 0.009 0.752 0.480 0.859 0.662 0.646 0.797 0.849 0.933 0.5 North.839 0.449 0.728 0.944 0.648 0.0 Maximum energy dissipation algorithm El Centro 0.008 0.722 0.603 0.548 0.008 0.883 0.009 0.688 0.122 0.782 0.497 0.652 0.007 0.729 0.401 0.858 0.468 0.5/1.009 0.005 0.849 0.559 0.009 0.579 0.010 0.887 0.978 0.730 0.486 1.613 0 0 0.0/1.003 0.776 0.5/1.334 0.929 0.796 0.550 1.058 0.005 0.009 0.5/1.0/1.925 1.227 0.700 0.818 0.0/1.664 0.748 0.501 0.648 0.3 Comparisons of the evaluation criteria for the nonlinear benchmark building Clippedoptimal controller (Yoshida and Dyke.5/1.828 1.004 0. 2002) Earthquake Intensity J1 Interstory drift ratio J2 Floor acceleration J3 Base shear J4 Normed interstory drift ratio J5 Normed floor acceleration J6 Normed base shear J7 Ductility J8 Dissipated energy J9 Plastic connections J10 Normed ductility J11 Control force El Centro 0.724 0.988 1.492 0.833 0.455 0.595 0.271 0.009 .724 0.5/1.816 0.592 0.0 0.230 0.5/1.884 0.308 0.969 0.025 0.328 0.009 0.711 0.642 0.807 0.670 0.542 0.695 0.5/1. 0.5 Hach.0/1.233 0.658 1.768 2.712 0.220 0.796 0.733 0.735 0.810 0.984 0.847 0.5 North.081 1.372 0.710 0.943 0.323 0.5 Hach.009 0.702 0.410 1.713 0.624 0.848 0.009 0.841 0.789 1.986 0.847 0.748 0.666 0.807 0.977 0.996 1.745 2.442 0.325 0.942 0.713 0.724 0.746 0.002 0.5/1.761 0.648 0.0 Kobe 0.961 0.328 0 0.578 0.223 1.267 0.911 0.258 0.780 0.959 0.840 0.091 0.070 0.796 0.791 0.723 0.Chapter 3 Maximum Energy Dissipation Algorithm 62 Table 3.632 0.691 0.909 0.563 0.809 0 0.714 0.257 1.722 0.728 0.658 0.656 0.636 0.0 Kobe 0.440 1.370 1.772 0.777 0.812 0.002 0.248 0. 0.144 0.474 0.747 0.724 0.743 0.760 0.009 0.273 0.638 0.955 0.560 0.885 0. 0.049 0.653 1.885 0.637 0.907 0.689 0.965 0.906 0.733 0.600 0.663 0.683 0.982 1.890 0.172 0.309 0.903 0.
967 0.859 0.009 0.660 0.576 0.818 0.5/1.463 1.917 0.972 1.566 0.800 0.573 0.5/1.606 0.072 1.833 0.655 0.539 0.055 0.009 0.0 0.050 0.025 0.5/1.867 0.977 0.044 0.540 0.372 0.092 0.009 Hachinohe 0.490 0.183 1.4 Evaluation criteria of modified location and number of MR dampers Maximum energy dissipation algorithm (modified location and number of MR dampers) Earthquake Intensity Interstory drift ratio El Centro 0.929 0.662 0.648 0.794 0.720 0.762 0.281 0.0/1.865 0.894 0.757 0.5 0.292 0.009 0.854 0.895 0.818 0.686 0.493 0.981 0.5/1.823 0.009 0.960 0.001 0.009 0.181 0.180 0.740 0.796 0.263 0.960 0.873 0.221 0.009 J1 J2 Floor acceleration Base shear Normed Interstory drift ratio Normed floor acceleration Normed base shear Ductility Dissipated energy Plastic connections Normed ductility Control force J3 J4 J5 J6 J7 J8 J9 J10 J11 .732 0.009 Kobe 0.571 1.663 0.071 0.0/1.823 0.977 0.961 0.568 0.839 1.632 0.734 0.610 0.740 0.755 0.568 0.764 0.829 0.873 1.921 0.Chapter 3 Maximum Energy Dissipation Algorithm 63 Table 3.540 0.163 0.743 0.746 0.009 0.549 0.818 0.869 0.957 1.009 Northridge 0.542 0.890 0.744 0.487 0.737 0.641 0.301 0.0 0.732 0.756 0.544 0.5 0.
The configuration of MR dampers are followings.5 shows the evaluation criteria for ±7% stiffness perturbed system under El Centro earthquake. Table 3. it dose not necessarily guarantee good performance in the actual system.Chapter 3 Maximum Energy Dissipation Algorithm 64 3. Four devices are located between each of the following pairs of nodes on bent 1 and pier 4. the normed overturning moment J9. and. SMC is known as robust controller (Moon et al. Even if the designed controller were confirmed to have good performance in the evaluation model. the robustness of MEDA is compared to that of SMC and the nominal performance is listed with the perturbed performance.4. Therefore.10) where K = nominal stiffness of the bridge.6 shows the evaluation criteria for ±30% stiffness perturbed system under the three earthquakes. MEDA shows more robust performance than SMC except the peak shear J2.2 Controller Robustness The dynamic characteristics of the real structure may not be identical to those of the evaluation model and can be changed after construction. and Kpert = perturbed stiffness. Perturbations of 7% and 30% were considered. The stiffness matrix is perturbed by some factor ε. the robustness check of MEDA is carried out only for the cablestayed bridge. and the resulting bridge model was simulated using the controller for the nominal system. 2003). . Table 3. MEDA is stable and performs well for ±30% perturbed system showing comparable performances with SMC. two devices are located between each of the following pairs of nodes on piers 2 and 3. MEDA is stable and performs well for ±7% perturbed system. ε = perturbation parameter.3. and the normed deviation of cable tension J11. the controller robustness of the MEDAbased semiactive control system using MR dampers was examined with respect to uncertainties in stiffness for the benchmark cablestayed bridge. The perturbed stiffness was calculated as K pert = K (1 + ε ) (3. Since the performance of MEDA is deteriorated for the building structures. In Table 3. So.
552 MEDA ε = ±7% 0.323 0.213 0.862 0.020 0.464 0.235 0.395 1.883 0.347 0.709 0.215 0.198 0.233 0.547 0.54 0.394 1.02 * Evaluation criteria is same with Table 2 .219 0.331 1.017 ε=0 0.224 0.443 0.692 0.218 0.278 0.887 0.108 0.234 0.560 0.233 0.Chapter 3 Maximum Energy Dissipation Algorithm 65 Table 3.130 0.255 0.551 0.5 Evaluation criteria for ±7% stiffness perturbed system under El Centro earthquake Evaluation Criteria* J1 J2 J3 J4 J5 J6 J7 J8 J9 J10 J11 SMC(Moon et al.432 1.901 0.870 0.016 ε = ±7% 0.335 0.891 0. 2003) ε=0 0.189 0.296 0.556 0.185 0.
061 0.334 0.544 1.Chapter 3 Maximum Energy Dissipation Algorithm 66 Table 3.827 0.69 0.056 1.0169 MEDA 0.021 MEDA 0.626 0.564 0.788 0.131 0.78 0.478 0.227 0.506 1.498 2.334 0.6 Evaluation criteria for ±30% stiffness perturbed system Evaluation Criteria* J1 J2 J3 J4 J5 J6 J7 J8 J9 J10 J11 El Centro Earthquake SMC ** Mexico City Earthquake SMC ** Gebze Earthquake SMC** 0.559 0.56 1.242 0.478 1.176 2.563 0. 2003 .007 Evaluation criteria is same with Table 2 **Data from Moon et al.761 1.744 0.264 0.737 0.481 1.513 0.359 0.384 0.797 0.945 0.59 0.577 0.577 1.266 0.469 0.232 0.177 0.63 0.311 1.191 1.422 1.393 1.557 0.944 0.0967 0.489 0.31 1.497 1.387 1.392 0.494 0.0238 MEDA 0.045 0.431 1.531 0.453 1.36 0.76 0.0127 * 0.019 0.0879 0.
.4 Summary of Results The applicability of the MEDAbased semiactive control system using MR dampers in reducing structural responses for seismic loading conditions has been demonstrated through a series of numerical studies of the benchmark problems. MEDA fails to reduce the peak floor acceleration and the peak base shear. but is not appropriate for the nonlinear benchmark building structure. For the benchmark cablestayed bridge. A comparison of results with two other controllers indicates that MEDA performs slightly better than other two controllers. If the number and the location of MR dampers are changed. the performance may be improved at the cost of losing an advantage of simple method to use. For the nonlinear benchmark building. Even for the ±30% stiffness perturbed. MEDA shows more robust performance than SMC in the most evaluation criteria. MEDA is stable and performs well showing comparable performances with SMC. Finally. MEDA is adopted to improve the design efficiency of controller without the deterioration of the performance and the robustness. The robustness of MEDA is investigated with respect to the uncertainties in stiffness for the benchmark cablestayed bridge. we confirm that the MEDAbased semiactive control system using MR dampers is efficient and robust for the benchmark cablestayed bridge. For the ±7% perturbed system.Chapter 3 Maximum Energy Dissipation Algorithm 67 3. the numerical results show that MEDA can reduce the vibration of the seismically excited cablestayed bridge structures effectively.
The current for the electromagnet is supplied by a power supply such as a battery and regulated by a controller which determines control commands. and the damper has a ±2. the MR damperbased control system becomes more complicated to build up and maintain. the electromagnet.1 Electromagnetic Induction System for MR damper A prototype MR damper has been considered to show the schematic of MR dampers. Figure 4. In that case.1. The damper is 21. and sensors as shown in Fig. many MR dampers are used for civil engineering structures such as cablestayed bridges and highrise buildings. which was obtained for evaluation from the Lord Corporation and was used by Dyke et al.5 cm stroke. The main cylinder houses the piston.Chapter 4 Electromagnetic Induction System 68 CHAPTER 4 SMART PASSIVE CONTROL SYSTEM 4. 4.1 is simple. Thus.5 cm long in its extended position.8 cm diameter.1 Schematic of a MR damperbased control system . the magnetic field produced in the device is generated by a small electromagnet in the piston head. to reduce the structural responses.3. and the main cylinder is 3. 1. As shown in Fig.(1996a). 4. a controller. resulting in changes of damping characteristics of the MR fluid. The MR damper needs a control system that consists of a power supply. an accumulator and 50 ml of the MR fluid. Although the MR damperbased control system in Fig.
. make high current at coil 1. Thus. The EMI system changes kinetic energy of reciprocation motion of the MR damper to the electric energy according to the Faraday’s law of induction (Reitz et. the MR damper with the EMI system is able to reduce the vibrations of structures by itself without any power supply and controller as in Fig.Chapter 4 Electromagnetic Induction System 69 Thus. the MR damper is a semiactive device that needs an external power source to change the damping characteristics of MR fluids. Fast relative motions. Miner 1996) and then the electric energy is used to change the damping characteristics of the MR damper. we can call the MR damper with EMI system as a ‘smart passive control system’ although the MR damper is semiactive device. The characteristics of the MR fluid is affected by magnetic field. al. but selfpowered by electromagnetic induction (EMI) system that is attached to the MR damper. the smart passive control system based on MR dampers is not using external power source. However.3. Slow relative motions. The smart passive control system is based on MR dampers. From the above explanation. Fig.2 shows the MR damper with the EMI system that consists of a permanent magnet and a coil. Marshall and Skitek 1990. 4. between the permanent magnet and coil 2. The magnetic fields at coil 1 solidify the MR fluid resulting increase of damping capacity of the damper. between the permanent magnet and coil 2. an EMI system is newly proposed for MR dampers to replace a control system. this dissertation proposes a ‘smart’ passive control system. In this study. make low current at coil 1. The magnetic field is arisen by induced current of the EMI system (consists of permanent magnet and coil 2). 1993. 4. Of course.
3 Schematic of a MR damper with the EMI system implementation .2 Schematic of a MR damper with the EMI system Figure 4.Chapter 4 Electromagnetic Induction System 70 Diaphragm Coil 1 Bearing & Seal Accumulator MR Fluid Coil 2 EMI system Permanent Magnet Figure 4.
2) where B is magnetic field.Chapter 4 Electromagnetic Induction System 71 Faraday’s law of induction is ε = −N dΦ B dt (4. the coil in the EMI system at the end of the pistonaxle moves back and forth inducing the emf. the higher emf is induced and the more slowly MR damper moves.3. magnet flux can be defined r r dΦ B = B ⋅ dA = BdA ⋅ cos φ (4. A is area of cross section. Using (4.1) where ε is induced electromotive force (emf) that has unit of volt(V). 1. we estimates the maximum induced emf using Faraday’s law for the prototype MR damper in Fig. N is number of turns of coil. External loads such as earthquakes and winds cause the reciprocal motion of the MR damper. In (4. This induced emf is carried to an electromagnet in the piston head and generates magnetic field around the electromagnet that changes the damping characteristics of the MR fluid. Provided that the permanent magnet of 1. and ΦB is magnetic flux.3) Faraday’s law of induction states that the induced emf in a closed loop equals the negative of the time rate of change of magnetic flux through the loop. and φ is the angle between B and r r r r dA . Thus.2 Tesla is used in the EMI system.1).2). then the . and the turns of coil are 900. the faster MR damper moves.1) is the direction of induced current. To verify the feasibility of the proposed EMI system. the lower emf is induced. the maximum velocity of reciprocal motion of the MR damper is 9cm/sec (that is about the maximum velocity of uncontrolled case of following numerical example). In consequence. Faraday’s law can be rewritten ε = −N dΦ B dB = −N A dt dt (4. Negative sign in (4.
Besides.16 Tesla/sec. This is the important benefit of using the smart passive control system. Considering that saturation of the MR effect begins in the prototype device when the applied voltage is 2. The resulting emf induced in the coil is about 2. which means there is no need of a power supply.25V. 2. Also. the amount of induced emf can be regulated by the turns of coil or the intensity of permanent magnet. the power for electromagnet in the piston head is supplied by induced emf of the EMI system.Chapter 4 Electromagnetic Induction System 72 time rate of change of magnetic field during the full stroke of 5cm is 2. The proposed smart passive control system does not need sensors that measure structural responses for a controller. the proposed smart passive control system has potential to replace the conventional MR damperbased semiactive control system.54V. because the damping characteristics of the MR damper is automatically regulated in proportion to the time rate of change of magnetic flux. is enough to change the damping characteristics of the MR fluid. the maximum induced emf. . Therefore.54V.
4.4. A schematic of the MR damper implementation is shown in Fig. Figure 4. 1996a) .1.Chapter 4 Electromagnetic Induction System 73 4. which is the exact one used by Dyke et al. A model of a threestory building configured with a single MR damper is considered here for direct comparisons with the MR damperbased semiactive system controlled by the clippedoptimal controller. Analytical Model The performances of the smart passive control system are now evaluated through the simulations.4 Schematic of a MR damper implementation (Dyke et al. The MR damper is rigidly connected between the ground and the first floor of the structure. (1996a).2.2 Analytical Model and Design 4.
and 0 A= −1 − M K I . Γ = 0 .4) where &&g is a onedimensional ground acceleration.5.3 0 0 175 − 50 kg . f is the measured force generated x between the structure and the MR damper. M Γ 0 E = − Λ 0 98.3 − 50 50 (4.5) 0 − 1 12.84 5 1 Λ = 1 1 The simple mechanical model of the MR damper was shown in Fig.3 0 M = C= m 0 0 0 98. z is the state vector.84 N . − M − 1C 0 B = −1 . 4. The equations governing the force f predicted by this model were given by Spencer et al. 0 − 50 100 − 50 N ⋅ sec 98.6) & y= 1 & {αz + c0 x + k0 ( x − y )} (c0 + c1 ) . K = 10 m 0 0 − 6.7 − 6.0 − 6.84 6. (1997a) as follows: & f = c1 y + k1 ( x − x0 ) & & & & & & & z = − γ  x − y  z  z n −1 − β ( x − y )  z n + A( x − y ) (4.84 − 6.Chapter 4 Electromagnetic Induction System 74 The governing equations of the structure are given by & z = Az + B f + E &&g x (4.84 13.
. which is emf induced by the EMI system. 1997a) To account for the dependence of the force on the voltage applied and the resulting magnetic current.(1997a) have suggested α (u ) = α a + α bu c1(u) = c1a + c1bu (4.8) is necessary to model the dynamics.Chapter 4 Electromagnetic Induction System 75 BoucWen c0 k0 c0 k1 x F Figure 4.7) c0(u) = c0a + c0bu where u is given as the output of a firstorder filter given by & u = −η(u − v) (4. Spencer.8) and v is the commanded voltage sent to the current driver. (4.5 Simple mechanical model of the normal MR damper (Spencer et al. et al.
and an appropriate primary controller for this active device is designed.6(a). In the design of the EMI system for this study.6×104 (turns/m).Chapter 4 Electromagnetic Induction System 76 4. Fig. 4. Fig. the summation of peak accelerations and Si. 4. the amount of emf can be regulated by the turns of the coil with a fixed capacity of the permanent magnet. 4. El Centro. 2.2. which are normalized by uncontrolled responses.6(b) is the envelope of the maximum responses of Fig. a normal MR damper system using the clippedoptimal controller (Dyke et al. designed for Si. Then a secondary bangbang type controller causes the MR fluid damper to generate the desired active control force. 4. an “ideal” active control device is assumed. 1996a. is called EMIdr.6 and 4. 4. which is the minimum point of the envelope denoted by the arrow. . From Fig.7 is similar to Fig. 4. For the comparison of the performance. respectively.16×104 and 2. designed for Sa.2 Design of the EMI System According to the Faraday’s law of induction. Appropriate number of coil turns needs to be determined in the design for better performance of the smart passive control system. First. Hachinohe and Kobe. is designated EMIac and the other EMI system. an EMI system.7 except that it is for Si. 2003) is used for parameters Sa and Si with three earthquakes. we can determine the optimal coil turns. the influence of two parameters is considered: Sa.b) is considered. 4. the induced emf is proportional to the turns of the coil and the time rate of change of magnetic flux. This strategy is as follows. Thus. Finally. the maximum response approach (Park et al.6(b). are determined from the Figs. so long as this force is dissipative.7. To determine the coil turns. the summation of peak interdrift displacements at each floor. Two appropriate coil turns.6(a) shows the variations of Sa for each earthquake and Fig.
Chapter 4 Electromagnetic Induction System
77
2 1.8 1.6 1.4 1.2
2 1.8 1.6 1.4 1.2 1
Sa
1 0.8 0.6 0.4 0.2 0 1 2 3 4 5 6
El Centro Kobe Hachinohe
7 8 9 10 x 10
4
0.8 0.6 0.4 0.2 0 1 2 3 4 5 6 7 8 9 10 x 10
4
Coil turns/m (a) Variations of Sa
Coil turns/m
(b) Envelope of max. responses
Figure 4.6 Design of EMI system with Sa under three earthquakes
2 1.8 1.6 1.4 1.2
2
El Centro Kobe Hachinohe
1.8 1.6 1.4 1.2 1 0.8 0.6 0.4 0.2
Si
1 0.8 0.6 0.4 0.2 0 1 2 3 4 5 6 7 8 9 10 x 10
4
0
1
2
3
4
5
6
7
8
9
10 x 10
4
Coil turns/m (a) Variations of Si
Coil turns/m
(b) Envelope of max. responses
Figure 4.7 Design of EMI system with Si under three earthquakes
Chapter 4 Electromagnetic Induction System
78
In this study, an H2/LQG control design is adopted as the primary controller. The ground excitation is taken to be a stationary white noise, and an infinite horizon performance index is chosen that weights appropriate parameters of the structure, i.e.,
1 τ J = lim E ∫ {z T Qz + f T Rf }dt τ →∞ τ 0
(4.9)
where R is an identity matrix, and Q is the response weighting matrix. The weighting parameters are as in (4.10).
q I Q = a 3× 3 0
0 qi I 3×3
(4.10)
where qa and qi weight the accelerations and interstory drifts at each floor, respectively. To determine the weighting parameters qa and qi, the maximum response approach is used as in the EMI system for three earthquakes. Fig. 4.8(a) shows the variations of Sa for increasing weighting parameters in 3dimensional plot. Each surface corresponds to the variation of Sa for each earthquake. Fig. 4.8(b) is the envelope of the maximum response of Fig. 4.8(a). From Fig. 4.8(b), we can determine the appropriate weighting parameters, which is the minimum point of the envelope denoted by the arrow. Fig. 4.9 is similar to Fig. 4.9 except that it is for Si. Two sets of the appropriate weighting parameter, qa = 5.0×1013, qi = 1.0×105 for Sa and qa=5.0×1015, qi=5.0×106 for Si, are determined from the Figs. 4.8 and 4.9. The clippedoptimal controller, designed for Sa, with weighting parameters qa and qi, is designated COac and the other, designed for Si, is called COdr.
Chapter 4 Electromagnetic Induction System
79
Sa qa
qi
qa
qi
(a) Variations of Sa
(b) Envelope of max. responses
Figure 4.8 Design of the clippedoptimal controller with Sa under three earthquakes
Si
qa
qi
qa
qi
(a) Variations of Si
(b) Envelope of max. responses
Figure 4.9 Design of the clippedoptimal controller with Si under three earthquakes
x max The second evaluation criterion is a measure of the reduction in the interstory drift. are normalized by the peak uncontrolled floor acceleration. given by  && (t )  x J1 = max aimax && t. The Northridge earthquake is not considered in the designing phase. and max d n denotes the normalized peak interstory drift in the uncontrolled response. .i d n (4. The first evaluation criterion is a measure of the normalized peak floor accelerations. Simulation results of the proposed smart passive control system are compared to those of the MR damperbased semiactive control system using the clippedoptimal controller by the evaluation criteria based on those used in the second generation linear control problem for buildings (Spencer et al.Chapter 4 Electromagnetic Induction System 80 4. &&ai (t ) .11) x where the absolute accelerations of the i th floor. but is included here to check the validation of the design of the EMI system and the clippedoptimal controller. a set of simulations is performed for the four historical earthquakes: El Centro. 1997b). and Northridge earthquakes. denoted &&a (t ) . Kobe.i xa (4.. The maximum of the normalized interstory drift is  d (t )  J 2 = max imax t.12) where di (t) is the interstory drift of the above ground floors over the response history.3 Numerical Simulation Results To verify the effectiveness of the proposed smart passive control system. Hachinohe.
5 2 1.5 1 Velocity(cm/sec) 5 0 5 10 15 20 0 0.5 1 1.5 5 Voltage(V) 10 0 0.5 2 2.5 4 4.5 5 Time(sec) (b) Hachinohe earthquake Time(sec) Figure 4.5 1 0.5 4 4.5 2 1.10 Velocities and induced voltages under various earthquakes .5 0 0.5 5 Voltage(V) 10 0 0.5 1 1.5 3 3.5 3 3.5 5 Time(sec) (a) El Centro earthquake Time(sec) 20 15 3 2.5 4 4.Chapter 4 Electromagnetic Induction System 81 20 15 3 2.5 1 Velocity(cm/sec) 5 0 5 10 15 20 0 0.5 1 1.5 2 2.5 2 2.5 3 3.5 3 3.5 1 0.5 0 0.5 2 2.5 4 4.5 1 1.
5 2 2.5 5 Time(sec) (d) Northridge earthquake Time(sec) Figure 4.5 1 10 15 20 0 0.5 3 3.5 1 1.5 4 4.5 0 0.5 0 0.5 1 0.5 4 4.5 1 Velocity(cm/sec) 0 0.5 2 2.5 4 4.5 1 1.5 1 1.10 Velocities and induced voltages under various earthquakes (continued) .5 5 Voltage(V) 0 0.5 Velocity(cm/sec) 5 0 5 Voltage(V) 0 0.Chapter 4 Electromagnetic Induction System 82 20 15 3 2.5 5 10 2 1.5 2 2.5 5 Time(sec) (c) Kobe earthquake Time(sec) 20 15 10 5 0 5 10 15 20 3 2.5 4 4.5 2 1.5 2 2.5 3 3.5 3 3.5 1 0.5 1 1.5 3 3.
Fig. El Centro COdr COac EMIdr EMIac Hachinohe Kobe Northridge 0 0.10 shows the velocities at the first floor where the MR damper is attached and the induced voltages by the EMI system for each earthquake. 0. For moderate earthquakes (El Centro and Hachinohe).9V. it can be seen that the higher voltage is induced for severe earthquakes.25V. the induced voltage is continuously varying whereas the command voltage of the clippedoptimal controller takes on values of either zero or the maximum value. 2. However. the velocity of the first floor is smaller than that of severe earthquakes (Kobe and Northridge) with the consequence that the induced voltage by the EMI system is lower according to the Faraday’s law of induction. The maximum induced voltage is 1. and 2.11 Normalized peak acceleration and interstory drift .25V for El Centro. 4. Besides.6V.5 1 0 0.25V for the capacity of the MR damper. and Northridge earthquakes. it should be noted that the induced voltage is restricted within 2. which is identical condition with the clippedoptimal controller. Kobe.Chapter 4 Electromagnetic Induction System 83 Representative responses of the EMI system to four earthquakes are shown in graphs. Hachinohe. respectively. Also. which is the voltage enough to operate the MR damper.5 1 Peak acceleration Peak drift Figure 4.
e. In the Table 4. it has comparable performance to the clippedoptimal controller without the power source. and sensors. however. The reductions in peak acceleration and drift are comparable to those of the clippedoptimal controller – both giving as much as nearly 50% decreased compared to the uncontrolled responses. controller. both achieve reductions in the peak acceleration and interstory drift for the Northridge earthquake.5% and 24. Though the EMI system fails to achieve more reductions over the clippedoptimal controller for the moderate earthquakes. compared to the better clippedoptimal controller.1% additional maximum decreases in the peak acceleration and interstory drift. EMIdr. This is the important benefit of using the smart passive control system. the clippedoptimal controllers achieve more reductions over the EMI systems for the moderate earthquakes such as El Centro and Hachinohe. respectively. Table 4. EMIac. In Tables 4. and COdr) normalized by each uncontrolled response. For the severe earthquakes such as Kobe and Northridge.2. the colored cells are the minimum value among four cases at each floor. the performances of the EMI system are better than those of the clippedoptimal controller giving up to 35. respectively. Although the EMI system and the clippedoptimal controller are designed under the three earthquakes except the Northridge earthquake..1 shows the accelerations and the interstory drifts at each floor for four cases of two categories (i. COac. .1 and 4. 4. except that the EMI systems give minimum value at the first floor.2 reports the percent response reduction (−) or increase (+) compared to the better clippedoptimal controller.Chapter 4 Electromagnetic Induction System 84 Fig.2.11 shows the values for the peak acceleration and peak interstory drift normalized by uncontrolled responses. Table 4. where it can be seen that the EMI system performs well over the entire suite of earthquakes considered. except the Northridge earthquake.
377 0.8428) COdr EMIac 0.8428) COdr EMIac 0.846 0.340 0.3495)* Hachinohe (0.433 0.492 0.476 0.436 0.800 0.827 0.3495) COdr EMIac 0.295 0.897 0.484 0.168 0.319 0.382 0.499 0.354 0.881 0.494 0.473 0.431 0.348 0.393 COac 0.372 0.367 0.345 0.587 0.473 0.568 0.423 COac 0.473 0.423 COac 0.457 0.293 0.228 0.515 0.512 EMIdr 0.396 0.441 El Centro (0.554 0.725 EMIdr 0.568 0.429 0.442 0.456 0.530 0.800 0.441 0.815 Northridge (0.178 0.586 0.2294) COac COdr EMIac EMIdr COac COdr EMIac 0.387 EMIdr 0.859 0.512 0.243 0.815 Northridge (0.384 Hachinohe (0.Chapter 4 Electromagnetic Induction System 85 Story 1st 2nd 3rd Story 1 2nd 3rd st Table 4.384 0.393 COac 0.8337) COdr EMIac 0.2294) COdr EMIac 0.180 0.492 COac 0.194 0.493 0.289 0.423 0.370 0.448 0.612 0.1 Normalized peak absolute accelerations and interstory drifts Accelerations El Centro (0.465 0.835 0.465 0.738 * ( ) is peak ground acceleration (g) .387 EMIdr 0.404 EMIdr 0.8337) COdr EMIac 0.492 0.355 0.410 Kobe (0.563 0.359 0.520 0.301 0.357 0.551 0.404 EMIdr 0.738 Interstory drifts Story 1st 2nd 3rd Story 1st 2nd 3rd COac 0.485 0.384 0.410 Kobe (0.435 0.384 0.428 0.526 0.355 0.308 0.725 EMIdr 0.212 0.
7 0 20.16 0 1.4 5.4 16.2 0 0 El Centro (0.8 1.4 31.5 0 0 Hachinohe (0.5 1.2 0.7 23.Chapter 4 Electromagnetic Induction System 86 Table 4.8* 0 22.0 22.2 6.2 Percent increment compared to the better clippedoptimal controller case Accelerations Story El Centro (0.5 1.8 2nd 6.0 1.8 3rd Interstory drifts Story 1st 2nd 3rd Story 1 2nd 3rd st COac 7.8 Kobe (0.4 10.1 23.8 0 35.4 EMIdr 19.9 Northridge (0.8337) Northridge (0.5 COac 21.4 23.4 22.3 7.5 35.9 0 9.7 24.5 1st 0.2294) COdr EMIac 0 10.5 5.7 6.3495) Hachinohe (0.4 1 28.0 11.8 3.9 0 4.6 2.1 11.4 21.1 11.2 0.5 5.0 EMIdr 15.8 0 0.8428) Story COdr EMIac EMIdr COac COdr EMIac EMIdr COac 0 14.9 2nd 0 7.9 7.0 3.8 * minus sign means reduction .3 20.3 16.6 12.2294) COdr EMIac EMIdr COac COdr EMIac EMIdr COac st 0 10.4 26.9 2.4 7.4 3rd Kobe (0.5 6.4 EMIdr 20.4 COac 19.5 22.5 0 21.9 EMIdr 2.4 1.3495) COdr EMIac 0 5.6 1.3 8.8 10.8428) COdr EMIac 0 0 0 24.4 COac 12.8337) COdr EMIac 0 0 0 5.2 1.2 9.0 0 10.1 2.
The smart passive control system is based on the MR damper with EMI system. In comparing both systems. the effectiveness of performances are evaluated. respectively. replacing a normal control system such as a power supply. Then. two EMI systems were designed. the EMI system generates induced voltages that can supply electricity and control commands to the MR damper. For the severe earthquakes such as Kobe and Northridge.1% additional maximum decreases in the peak acceleration and interstory drift. controller. and sensors. and sensors. . the smart passive control system shows the better performance giving up to 35. the proposed smart passive control system has potential to be implemented in real civil structures. Therefore.5% and 24. and compared with those of the semiactive MR damper system using clippedoptimal controller. To investigate the achievable capabilities of the smart passive control system. According to the Faraday’s law of induction.Chapter 4 Electromagnetic Induction System 87 4. the proposed smart passive control system has the simple structure without any power supply.4 Summary of Results This study has proposed a smart passive control system for a civil engineering application. a controller. the smart passive control system showed the comparable performance to the MR damperbased system controlled by the clippedoptimal controller. it was observed that for the moderate earthquake such as El Centro and Hachinohe. In addition to the comparable performance. The EMI system consists of a permanent magnet and a coil.
although resulting responses varied greatly depending on the choice of measurements available and weightings. MEDA can reduce the vibration of the seismically excited cablestayed bridge structures effectively. For the benchmark cablestayed bridge. The robustness of MEDA was investigated with respect to the uncertainties in stiffness for the benchmark cablestayed bridge. For the ±7% and 30% perturbed . For the nonlinear benchmark building.Chapter 5 Conclusions 88 CHAPTER 5 CONCLUSIONS The study proposes the implementation of simple and efficient control algorithms for seismically excited structures using MR dampers and a smart passive control system based on MR dampers First. was implemented to seismically excited structures using MR dampers. The motion of the structure was effectively suppressed by merely controlling a few lowest modes. MEDA fails to reduce the peak floor acceleration and the peak base shear. the characteristics of the implemented simple and effective control algorithms are summarized as follows by the analytical and numerical examples: (1) Modal control scheme with a lowpass filter and modal estimator of Kalman filter. resulting in reducing the size of weighting matrix Q. (2) Maximum energy dissipation algorithm (MEDA) was adopted to improve the design efficiency of controller without the deterioration of the performance and robustness.
. And. For the severe earthquakes such as Kobe and Northridge. replacing a normal control system such as a power supply. the characteristics of the proposed smart passive control system are summarized as follows by the analytical and numerical examples: (1) The EMI system that consists of a permanent magnet and a coil. MEDA shows comparable performances to SMC in the most evaluation criteria.1% additional maximum decreases in the peak acceleration and interstory drift.Chapter 5 Conclusions 89 system. and sensors. (2) In comparing with MR damperbased semiactive control system using the clippedoptimal controller. a controller. generates induced voltages to supply electricity and control commands for MR dampers.5% and 24. the smart passive control system shows better performances giving 35. respectively. the smart passive control system shows comparable performances for the moderate earthquake such as El Centro and Hachinohe.
이러한 MR댐퍼를 운용하기 위한 많은 알고리즘들이 제안되었으 나. 따라서 제어기의 크기가 축소되어 설계가 용이하다. 아울러. 또한 Kalman필터를 상태관측기로 사용하는데 있어서. 그 성능을 비교하였다. 속도. Lowpass필터를 사용하여 Spillover문제를 해 결하였다. 또한 현실상황과 적합하도록. 제어기 설계가 복잡한 문제 등이 발생했 다. 제어기 설계가 간단하고. . 변위. 그리 고 가속도 피드백을 각각 사용하여. 모드제어기법과 Maximum Energy Dissipation Algorithm (MEDA) 을 연구하였다. 이에 따른 비상시의 신뢰성. 모드제어가 수년간 연구되었음에도 불구하고 반능동 제어장치. 모드제어를 구현하고 그 성능을 평가하였다. 모드제어는 선택된 소수의 저차 모드만을 제어한다.요약문 90 요 약 문 MR댐퍼를 위한 간편한 알고리즘과 스마트 수동제어 시스템 본 논문은 MR댐퍼를 기반으로 하는 제어시스템을 대형 구조물등에 설치할 수 있도록 간편한 알고리즘을 구현하였다. 각각의 성능이 비슷함에도 불구하고. Kalman필터를 상태관측기로 사 용하여 모드 좌표계를 관측하였으며. 배터리와 같은 소규모 전 력을 사용하고. MR댐퍼는 최근에 각광 받는 반능동 제어장치로써. 기존 제어 알고리즘의 성능과 비교하였다. 따라서 본 논문에서는 MR댐퍼의 반능동 제어장치로써의 비선형성을 고려하는 동시에. 특히 MR댐퍼에 대한 가능성은 아직 검증되지 않았다. 그리고 상대적으로 저렴한 제작단가 등 의 장점들이 있다. 따라서 본 논문에 서는 지진하중을 받는 구조물의 MR댐퍼 제어를 위해서. 그 성능 역시 기존 알고리즘에 뒤지지 않는 제어 알고리즘으로써. 이렇게 구현된 모드제어 를 6자유도 구조물에 적용하여. MR댐퍼를 기반으로 하는 스마 트 수동제어 시스템을 제안하였다.
이러한 MR댐 퍼가 토목구조물과 같은 경우에 대량으로 상용되면. 또한 유도되는 전류의 양이 MR 댐퍼의 왕복운동 강도에 비례하므로. MR댐퍼를 위한 외부전원. 기존 MR댐퍼의 효율적인 운용을 위한 연구와 아울러. 성능 및 강인성을 기존 연구의 결과들과 비교하였다. EMI 시스템에서 Faraday의 전자기 유도법칙에 의해 전기에너지로 전환하고. 센서 및 제어기가 필요하지 않다. 특별한 제어기 설계가 필요하지 않다. 이 를 MR댐퍼의 점성특성을 변화시키는데 사용한다. 따라서 반능 동 제어장치인 MR댐퍼를 기반으로 하지만. Lyapunov직접법에 의해 유도된 MEDA는 제어장치가 설치된 곳의 국부변위 등에 의해 제어력을 결정하므로. 즉. 외부전원 대신에 MR댐퍼에 부착 된 EMI 시스템에서 유도되는 유도전류를 사용한다. 진동의 정도를 측정할 수 있는 센서. 그리고 MR댐퍼의 거동을 결정하는 제어기가 기본적으로 필요하다. 외부전력을 사용하지 않고. 그 제어 시스템은 설치 및 유지 관리가 어렵다. MR댐퍼의 왕복운동에너지 를. MR댐퍼를 사용하는 규모가 큰 토목구조물에 대한 그 적용성 및 가능성이 연구되지 않았다. 이를 위해서 수치 예제로써 사장교 및 20층 짜리 비선형 빌딩의 벤치마크문제를 이용하였으며.요약문 91 또한 제어기의 설계에 Lyapunov직접법을 사용하는 MEDA 역시. MR댐퍼를 기반 으로 하는 스마트 제어시스템을 제안하였다. MR댐퍼에 필요한 전원은. . 따라서 본 논문에서는 MEDA의 대규모 토목구조물에 대한 적용성을 성능과 강인성 면에서 평가하였다. 한편. 따라서 제어기 설 계가 어려운 대형 토목구조물에 적합하다. 기존의 MR댐퍼를 사용하여 구조물을 제어하기 위해서는. 장치 스스 로 외부하중의 강도에 맞추어 점성을 변화시키는 ‘스마트’ 수동제어 시스템이다. 스마트 수동제어 시스템은 MR댐퍼와 EMI (Electromagnetic Induction) 시스템으로 구성한다. 따라서 본 논문에서는 MR댐퍼를 기반으로 하는 스마트 수동제어 시스템을 제안하였다.
Clippedoptimal 제어기로 제 어되는 MR댐퍼의 성능을 수치예제를 통해 비교하였다. .요약문 92 제안된 스마트 수동제어 시스템의 성능평가를 위해서.
. (1996). Sheffield.J.D. Clair. .L.” Machine Design. World Scientific. “The Promise of Controllable Fluids.D. 266–270. Lenz. Karlsson & Sorensen Inc. “Performance evaluation of vibration controlled steel structures under seismic loading”.References 93 REFERENCES 1. Barrosa.D.” J. Singapore. Carlson.F. R. on ER Fluids. 2nd Workshop on Structural Control: Next Generation of Intelligent Structures. Balas. 61–64.” Proc. Borgmann and K.D. Carlson. B. 99–109. K. Modern Control Theory. Hibbitt. 6.. Engrg. 5th Int. 10. “Magnetorheological fluid dampers: scalability and design issues for application to dynamic hazard mitigation. Makris. J.D. K. Conf. MR Suspensions and Associated Tech. 8. Carlson.. (1995). “Commercial MagnetoRheologicalFluid Devices. ABAQUS (1998). thesis of Stanford University. N.D. Mech. J.A. J. U. “A Growing Attraction to Magnetic Fluids. 122. 9. Carlson. Carlson. (1978). S. Eds.D. Brogan. (1999). New Jersey. P. 2. 7. L. (1994).).M. 20–28. of Actuator 94 (H. 4. and St. K. Prentice Hall.A. M. W.. Hong Kong. W. “Commercial MagnetoRheological Fluid Devices.. RI. Englewood Cliffs. Ed. Catanzarite. and St Clair.. (1991). 897–906. MR Fluids and Associated Technology. I.. 5. and Weiss.” Proc.. AXON Technologie Consult GmbH. 23(4).. (1994). Ph. “Modeling the response of ER damper: phenomenology and emulation. and Spencer Jr.” IEEE Transactions on Automatic Control. D.J. J.N. Burton. 3. J. of the 5th International Conference on ER Fluids. Bullough. (1996a). August. Pawtucket. D. Catanzarite.A. “Feedback control of flexible system.” Proc.. and Antsaklis. UK. (1996). China. 674679. Konstantopoulos..” Proc.
Fed. Caicedo.” Smart Mat. S. D. 5. 565–575.D... and Spencer. 857–872. Dyke.” J. J.S... 19. Sain. John Wiley & Sons. J. D. M.F. Plesha. and Struct. New York.. C. S. Turan. S.. and Spencer Jr. B... of Structural engineering.J. J. “An experimental study of MR dampers for seismic protection.K.. 9. Ehrgott. Mech.” Proc... Faravelli. G. (1996a). (2003). Int. (1992). 693–703. 17. M.” Smart. B.C. 35–40..K.. and Carlson. Sain. Chang. Wiley. “Modeling the oscillatory dynamic behavior of electrorheological materials in shear. 129(7). (1989). L... Sensors and Smart Structures Technology. S.J. 907–918. of Automatic Control. Dyke. 1. B. S..K. M. ASCE. R. “Phase 1 Benchmark control problem for seismic response of cablestayed bridges.. R.. (1996b). J. 16. Malkus. and Struct..D. L.. 122(9). and Carlson. and Roschke.F. 275–285.J. and Carlson. 755–764. . P.T. B. Mat. (1996b). 13.C.. 15. Dyke... M. New York. and Hague. Spencer Jr. “Neural network modeling of a magnetorheological damper. IFAC World Cong. Bergman.J. Jr. Chiba. L.K... 14.. F. and Struct. Carlson.F.D.F.” Proc. 3. Kaspari Jr.M.. B..” J. Concepts and Applications of Finite Element Analysis.. “Modeling and control of magnetorheological dampers for seismic response reduction.D.J.. Spencer Jr. J.” J.F. (1998). S.. Quast. “Acceleration Feedback Control of MDOF Structures. 12. Proc. “Seismic response reduction using magnetorheological dampers. Sain.. 3rd International Conference on Motion and Vibration Control.D. 145–150. (2003). Jr. Japan. S. M. P. T.References 94 11..” Smart Mat. and Masri. B. of Engrg. Dyke. Spencer Jr. Dyke. eds. Spencer Jr. 18. 20.Sain.. and Soong. (1996c).. (1998).F..A. Intelligent Material Systems and Structures. Cook.. “Magnetorheological fluid dampers for semiactive seismic control. 5.E..
. 63.” J. 126(8). Dyn. E. Sunakoda. (1997). Iwata. F. China. E. ASCE... eds.F. S.E. CE9013. Fujitani. F. ASME. 27.. on Adv.” J.” J.. K. Hanson. S. present and future. Johnson.. D. H. Mech. part 1: analysis and design. Housner.. B. G. 399425. (1994). Gamoto. (1996a). Baker. Feng.P.E.. S. “Dynamic performance evaluation of magnetorheological damper. “Structural control: Past.. 319–326.. A. U. and Filisko. Workshop on Struct. 123(9).. and Dyke. G. Proc. G. R. G. (2001b). Hanson.. Housner. N.. Johnson. Portland. Control Res. and Filisko. et al.” Proc... USC Publ. in Struc. (1990).” Earthquake Engineering and Structural Dynamics. 28. 676–682. ASCE. H. (2001a).. . 32. and Filisko.” ComputerAided Civil Infrastructure Engineering. 795–803. and Soda. “Electrorheological dampers. H. and Shinozuka.. Mech. Nat. 24. (2000). Int. ASME. Conf.. H. M.D.P. “Semiactive control strategies for MR dampers: comparative study.References 95 21. F. .. 897971..A. 63.J. 669–675. R. Hata..M. Inaudi. Y... F. “Use of a variable Damper for Hybrid Control of Bridge Respnse Under Earthquake. “Semiactive Damping of Stay Cables. 22.. 25. (1991). J. (1996b). 23.. “Dynamic mechanical studies of electrorheological materials: moderate frequencies.” Proc. 26. 26(3). 361. Mech.. part 2: testing and modeling. Rheology. and Chassiakos. 1st World Conf. Masri.D. On Structural Control. W.” Proceedings of the XV ASCE Structures Congress. (2000). 31.” J. Komatsu. v1. B. Eng. Spencer Jr. Oregon. Sodeyama. Hague. “Electrorheological dampers... Christenson R. and Fujino. 30. “Semiactive damping of cables with sag. Engrg. (1997).” J. Engrg.F. S. Applied Mech.. K.E.S. accepted.E. Y.” J. Q.A. and Spencer Jr. (1997). Gavin. 35(3).R. Gavin. Hong Kong. L. “Modulated Homogeneous Friction: A Semiactive Damping Strategy.A.. accepted. Applied Mech. 29.. “Composite Design for Long Span Bridges.A. W. Jansen. No..
N. M. “Past. 2nd World Conf... 201251. Crosby.J. 36.G. 1125–1132. “Vibration Control Using SemiActive Force Generators. “An Introduction to observers.. 42. 914 41.” Proc. 5... 5. G. Leitmann.W..M. Karnopp. “Properties and Applications of Commercial Magnetorheological Fluids. T. Y..” J. Yamada.” IEEE Transactions on Automatic Control.. 34. Iemura. Y. San Diego.. 37. M. Control. (1999).K. A. A. N.” Smart Mat. J. Kijewski. (1971). Wiley. Kobori. Hurt. T. (1998). and Tamura. M. 40. (1996). September. “Mitigation of motions of tall buildings with specific examples of recent application. N.M. Kamath. Seto. 39. (1991). (1994)... “Nonlinear viscoelasticplastic mechanismsbased model of an electrorheological damper.. Symposium on Smart Structures and Materials. (1974).. D. On Structural Control. Kamath..M.A.. R.. H. Koshika.” J. 841–846.. T.. of Intelligent Material Systems and Structures. Part 1: Design. Wiley. New York. Kareem.References 96 33. CA. 38. D.. Dyn. Kobori. 20. N..” Wind Struct. and Harwood. eds. May.” Earthquake Eng.M.. “Seismic response controlled structure with active mass driver system. and Wereley. (2003). Inou. 596602. G. and Dynamics. 35. and Wereley. 576–590.R. On Structural Control. K.” ASME Journal of Engineering for Industry. Kobori.C. 2(3). and Ikeda. SPIE 5th Annual Int. Proc. 20(6). 133139. (1998). G. Bender. Y. Struct. J. and Carlson. 3rd World Conf. Luenberger. and Struct. Jolly.” Proc. AC16(6). and Nishitani.. (1997). . “Semiactive Control for Vibration Attenuation. New York. 96(2). “Analysis and testing of Bingham plastic behavior in semiactive electrorheological fluid dampers.D. present and future in seismic response control in civil engineering structures. Guidance. T.
N. 30. (2001).. (1996). and Voulgaris. Of the 2000 Engineering Mechanics Conf.” Earthquake Eng. Christenson. Analytical Methods in Vibrations. 54.H.F.E. “Summary of benchmark control problems for seismically excited nonlinear buildings. (1995). Y. 50. Spencer. Lines and electromagnetic fields for engineers.. Natick. 269291.” J. and Dyke. Oxford University Press.. 129(1). Burton. and Gavin. G. (1990).. (2003). Massachusetts..J. R.F. 122. 45. Jr.. and Skitek. L.. Texas. 46.. S. Miner. G. L. D. B. Spencer. S. (1996). and Baruh. ASCE. 7178. 1003–1011.” Journal of Engineering Mechanics. 49. Meirovitch. 52.. 47. Nishitani.. P. Ohtori..F. Meirovitch. G. May 2124. and Dyke. S. Christenson. Engrg. “Overview of the application of active/ semiactive control to building structures in Japan. Ohtori. Y. S.J. Inc. Mech. R.P. Inc..” Proc. J. M. Macmillan Publishing. McClamroch.References 97 43. 48. New York.V. (1983) “On the Problem of Observation Spillover in SelfAdjoint DistributedParameter Systems. 39(2). “Analysis and design of ER damper for seismic protection of structures. “Closed Loop Structural Control Using Electrorheological Dampers. Y. 15651574.” Journal . Dyn. A. Seattle.. Electromagnetic concepts and applications. “Sliding Mode Control of CableStayed Bridge Subjected to Seismic Excitation. L.. L. John Wiley and Sons. Struct. N. 53. B. Washington. PrenticeHall.” Journal of Optimization Theory and applications. H. Bergman. Conf. Markis. Hill. N.” Proc. The Math Works. Y. of the Amer. G. H.. 4173–4177. Austin. Dynamics and Control of Structures. (1990). Meirovitch. (1967).. 44. ASCE. MATLAB (1997). 51.. S. “Benchmark control problems for seismically excited nonlinear buildings. Marshall. Jr. and Jordan.A. (2002). Ctrl. Moon. (2000). and Inoue. A.E.
W.L. 4557. Rabinow. W. Matrix Structural Analysis.” Comput.” U. Patent 2. Co.C. H.S. "Hybrid Control Strategy for Seismic Protection of a Benchmark CableStayed Bridge. 55. R.. (1994). H. Soda.N. Johnson. Infrastruct. USC Publication Number CE9311. 417–431. AddisonWesley Pub. U. Sack. (1993). of the Int. Foundations of electromagnetic theory.." Engineering Structures. of the U. (2003).. Rabinow. Sack.575. 2. Wu. on Earth.W. Iwata. Workshop on Structural Control. and Hata.. (1990).W. (1989).D thesis.C.. 18. Aided Civ.J. . Eng. Fifth National Conf. 62. (1994). L. R.... 405417.” Ph. Berkeley. Engrg. Park. ASCE. “The magnetic fluid clutch. F. T.. Longman Scientific. and Spencer Jr. Reitz. E. J. (2001). Active structural control: Theory and practice.K.W. 61. Mech. (2003). Phillips.References 98 of Engineering Mechanics: Special Issue on Structural Control Benchmark Problems. ASCE(submitted). “Engineering applications of fluids with a variable yield stress. Essex. Sodeyama. Jung. Sunakoda. K. J.S.. 1308– 1315.L. (1951). H. 59. 311–320. Boston. H. Sack. Co.R. 56. 25(4). Ramallo. (1969).. Milford. K. “Seismic Motion Control via Semiactive Hydraulic Actuators.. S. C. Kuo. T. “Dynamic tests and simulation of magnetorheological dampers.C. 63. and Patten. 67. 57. 58. Illinois. California.. Eng. 65.F. R.. PWSKent Pub. B.” J.” Proc. N. R. J.360. and Patten.. I. (1948). submitted. and Christy R.” Proc. “Magnetic fluid torque and force transmitting device.S. Chicago. H.L.J and Lee. Liu.” AIEE Transactions. “‘Smart’ base isolation systems... J. University of California. 60. Soong.A. Fujitani.. “Semiactive Hydraulic Structural Control.. 64.
“Magnetorheological dampers: a new approach to seismic protection of structures.A. 27(11).. and Sain. Yang. F.F.S. Sain. “Series solution of a class of nonlinear optimal regulators.References 99 66. Mag. B. 676–681.J. K. M.K. on Struct.. 1935. and Dyke. B. J. S.” J.J. 91. H. (1998)..” Proc.. Control. Dyke. S.. Spencer Jr. Jr. (1997). J. 75.F. and Sain. Kyoto..F... Mech. 417–426.F. A.” Proc. (1998a.b).. M. 321–345. and Deoskar.. T. and Yang.K. Spencer Jr. Engrg. G. “Phenomenological model of a magnetorheological damper. B. 1127–2247 . 2. M. B. Carlson..L. 2nd World Conf. S. “Recent Trends in Vibration Control in the U. 3rd Int. Japan. Control...K..S.. M.” Struct. K1K6. T. Timlin. Dyke. Sain.... Tall Build. Soong.. and Sain. Spencer Jr. II.F. Dyke.” IEEE Control Syst. Sain.J. M. On Motion and Vibr. Spencer Jr. (1993).D..” Earthquake Engneering and Structural Dynamics. S.” J.. B. “Controlling buildings: A newfrontier in feedback. 74. Des. “An overview of active and hybrid structural control research in the U. 73. Optimization Theory and Applications.. Struct. (1996b). 24.D. and Carlson. Carlson. G.. 71.. B. Spencer Jr. “Supplementary energy dissipation: Stateoftheart and stateofthepractice.. “‘Smart’ dampers for seismic protection of structures: a fullscale study.Vol. “Benchmark Problems in Structural Control – Part I: Active Mass Driver System. Conf.F. 17(6). 70. 123.” Eng.” Proc.. Spencer. Jr. Part II: Active Tendon System.. 192209. 68. B. F. M. on Decision and Control. B. Jr. (1997b). 67. Conf.. and Spencer.. Spencer.. (2002)... ASCE. (1997a). J. Spencer Jr.S.J. Soong. (1996a). B. 458–462. T. 243259. 72. “On the current status of magnetorheological dampers: seismic protection of fullscale structures. and Reinhorn.” Proc.D. (1996). 230–238. Chiba. M. Jr.K. T.F.K. 69. Japan. T. American Control Conf..
F. and Gravelle W. 2nd World Conf. Pang. 5th Int. Electrostatics. (1999).G. Spencer. D.. B.M. N. 81. 85.. Eng. 108. Structural Control. Symans. 83. On Vibration Engineering.. 79.. Series C. China. and Constantinou. 20.. and Stevens. 21(6). Conf.. Jr. “Idealized hysteresis modeling of electrorheological and magnetorheological dampers.. E.M. Stanway. B. EM5. “Civil engineering applications of smart damping technology. “Semiactive control systems for seismic protection of structures: A stateoftheart review. 77. Spencer. (2002). N.L. Iwata. 20. Johnson... Intelligent Mat. N. 417426.. Sunakoda.” JSME International Journal: Special Issue on Frontiers of Motion and Vibration Control. Yang.” Earth. C. G. Wereley. M. G. and Sain.A.N.F.. Sodeyama.” Proc. Struct. (2000). 3989.” J.” Proc.. J. B. “State of the art of structural control. Fujitani.. 194–203. 84. “ ‘Smart’ dampers for seismic protection of structures: A fullscale study. “Dynamic characteristics of magnetorheological fluid damper.. “Nonlinear modeling of an electrorheological vibration damper.. 82.” Proc.. and Ramallo. 833849. J. Sproston.. 704–711. 9.. Nanjing. R. Conf. M... Carlson. Systems and Struct. and Mat. Eng. 43(3). 78. and Struct.” Eng. .. 80. (1998).” J. “'Smart' isolation for seismic control. Newport Beach. Wilson. (1991). 707721.C. 845856.. 771782.. M. (2003).. Dyn. L. H. H. 1.. (2000). Spencer. Jr. Wiley. K. California. F... J. and Soda. SPIE Smart Struc. S. J. 469487. 642–649. Jr.References 100 76. and Nagarajaiah. F..” J. (1999). K. 129(7). B. No. J. New York. Spencer Jr. (1982). “Control of Tall Buildings Under Earthquake Excitations. 167–184. Struct. S.” ASCE Journal of Enginering Mechanics Division. D. (1987). and Kamath.. “Modelling of a cableStayed bridge for dynamic analysis. Yang.
Yang.C. “‘Smart’ base isolation strategies employing magnetorheological dampers. F. Spencer Jr. “Dynamic model of fullscale MR dampers for civil engineering applications. (2002). .” Proc... and Spencer Jr.. J.. Struct..C. Conf.. (2000). H. of Engineering Mechanics: Special Issue on Structural Control Benchmark Problems. B. and Sain.. On Structural Control. 87. M. O. Ramallo.. and Sain. Newport Beach. 341–348.. (2002). Yoshioka.. Seattle. CDROM. Ramallo. “Dynamic Performance of largescale MR fluid dampers. “Semiactive hybrid control systems for nonlinear buildings against nearfield earthquakes. (2000a).. 92.” Proc. of American Control Conf. (2002). and Spencer Jr. Wiley.. China. S. B. Yang. N. K. J. Chicago. 128(5). IL. 24(3). 1. Carlson.K. USJapan Workshop on Smart Structures for Improved Seismic Performance in Urban Region. and Agrawal.. Carlson. 93. G. 3rd World Conf. Austin.” Eng.. New York..K.” J.D. G. and Sain.C. B. 90..F. WA. 271280. 14th ASCE Engineering Mechanics Conference. J. Ramallo J.. J. and Dyke. Spencer Jr. J. “Dynamic modeling and performance considerations on fullscale MR fluid dampers.. 89. M.F. 94. Carlson. J. (2000b). Spencer Jr. G. H.. (2003)... on Structural Safety and Reliability. Dyn. Yi. M. and Dyke. B.” J. Jung. “Kobori Panel Discussion: Future perspectives on structural control. Hong Kong.” Proc. Yang. ASCE(submitted).. (2001b).. and Dyke.References 101 86. 88. G.K. Yoshida. Eng.. S.. 8th Int. 540551. Yang. A.J. Yang.” Proc. Texas.J. Int.F. 91. J. “Seismic control of a nonlinear benchmark building using smart dampers.J..F. B. Struct. N.F. J. (2001a). “Largescale MR fluid dampers: dynamic performance considerations.” Proc. CA. Yang.D. S.D.. on Advances in Struc. “Structural control systems: performance assessment. 279286..” Proc. Conf.
공주대학 교 고만기 교수님. 곽효경 교수님. 장동두. 중부대 학교 김춘호 교수님. 아울러 복합재료 를 가르쳐 주신 Korea Composites 김덕현 박사님께도 감사 드립니다. 장모님께 감사 드립니다. 김동현. 장인. 손주 사위를 많이 귀여워 해주신 할머님과 맏아들과 맏사위로서 격려해 주신 아버지. 성균관대학교 박선규 교수님. 같은 뜻을 품고 캠퍼스를 누볐던 목원대학교와 대전IVFer들. 또한 후배지만 어엿한 사 회인으로 살아가는 김기영. 지현우. 93동기들. 마지막으로. 그리고 한국과 학기술원 건설환경공학과 교수님들께도 감사 드립니다. 그리고 항상 따 스함으로 다가오는 늘사랑 교회와 구역 식구들의 애정어린 관심과 사랑을 기억합 니다. 결혼해서 훌쩍 어른이 되어 버린. 김동옥. 문영종. 임현우. 세종대학교 정형조 교수님께 감사 드립니다. 한경대학교 김운학 교수님. 하준식에게도 고마운 마음을 전합니다. 권호철. 이헌재. . 이승우 선배님들께 감사 드립니다. 함께 했던 여러분들께 감사 드리고자 합니 다. 장지은에게도 감사 드립니다. 박정규. 또 다른 삶의 한 장을 마무리하며. 김만철. 탄. 경진에게 고마운 마음과 사랑을 전합니다. 무엇보다 지난 7년 동안 세심한 지도와 올바른 삶의 태도를 깨우쳐 주신 이인원 교수님께 깊이 감사 드립니다.감사의 글 사랑하시는 하나님께 감사 드립니다. 각자의 길을 성실히 가고 있는 석사 동기들에게 감사의 뜻을 전합니다. 정환진. 김진근 교수님. 여 동생 범주와 남편 준용이 그리고 처제 경훈과 남편 성구형. 이성진. 세미나에 참석하셔서 많은 것을 가르쳐주신 한남대학교 오주원 교수님. 나보다 나를 더 사랑하는 젊고 아름다운 아내. 그리고 군 복무중인 처 남 진우에게도 고맙다는 말을 전합니다. 조 지성. 기계공학과 장평훈 교수님께 감사 드립니다. 김현택. 경일대학교 이종헌 교수님. 당당하게 살아가는 멋진 정길호. 박규식. 조홍기. 신혜린. 어머니. 세심한 심사와 조언을 해주신 윤정방 교수님. 장종우. 그리고 여전히 실험실에서 함께 하는 동기 김병완 박사. 최강민.
Korea . Civil and Environmental Engineering Korea Advanced Institute of Science and Technology (KAIST).S. Korea EDUCATION Ph.D. KAIST. Korea March 1993February 1997 EXPERIENCE March 1999 to Present Graduate Research Assistant. Civil and Environmental Engineering Korea Advanced Institute of Science and Technology (KAIST). Korea March 1997February 1999 B. Department of Civil and Environmental Engineering. Civil and Environmental Engineering Korea Advanced Institute of Science and Technology (KAIST).CURRICULUM VITAE Name : SangWon Cho Date of Birth : August 15. Korea March 1999Present M. 1974 Place of Birth : Seoul.S.
(Accepted for Publication) .” M. KangMin Choi. 4. SangWon Cho. 1. Vol. "Semiactive Fuzzy Cotrol for Seismic Response Reduction Using MR Damper. KangMin Choi. HyungJo Jung & InWon Lee. Korea. HyungJo Jung & InWon Lee.D. Korea. "The Implementation of Modal Control for Seismic Structures using MR Damper. ManGi Ko & InWon Lee." Computers and Structures. SangWon Cho. 6369. "Design and Control of Magnetorheological Dampers with the Electromagnetic Induction System for Seismic Response Reduction. Department of Civil and Environmental Engineering. Dec. pp. HyungJo Jung & InWon Lee. Dissertation. 2. SangWon Cho. “Efficient Mode Superposition Methods for Nonclassically Damped Systems. 2003. February 2004." Smart Material and Structures.S. Korea Advanced Institute of Science and Technology. Journal Papers: 1. "Algebraic Method for Sensitivity Analysis of Eigensystems with Repeated Eigenvalues. Department of Civil and Environmental Engineering. ByoungWan Kim. (Accepted for Publication) 3. 82. No. Daejeon." ASCE Journal of Engineering Mechanics.LIST OF PUBLICATIONS Dissertations: 1. Dissertation." Earthquake Engineering and Structural Dynamics. Korea Advanced Institute of Science and Technology. February 1999. (Accepted for Publication) 2. Daejeon. SangWon Cho.” Ph. “Simple Control Algorithms for MR Dampers and Smart Passive Control System.
InWon Lee. 1. 2002. 2000. Sep. KyuSik Park. 3. pp. 2326. Dec. pp. 810. 597602. ChunHo Kim. pp. SangWon Cho. 16." KSCE Journal of Civil Enginieering. (in Korean) Conference Papers: 1. 2. JiSeong Jo. 7. "Efficient Mode Superposition Methods for NonClassically Damped Systems. ManGi Ko & InWon Lee. Rome. InWon Lee." The Sixteenth KKCNN Symposium on Civil Engineering. 2003. 3. "Algebraic Method for Sensitivities Analysis of Eigensystem with Repeated Eigenvalues. HyungJo Jung. 2326. 1." The Fifth KKCNN Symposium on Civil Engineering. 859868. InWon Lee. Vol. "Efficient Mode Superposition Methods for NonClassically Damped Systems. 1999. 20632066." Journal of the Earthquake Engineering Society of Korea." The Second International Conference on Structural and Construction Engineering. Italy. pp. SangWon Cho. KyuSik Park. 1. No. DongOk Kim. Singapore. KangMin Choi. 7984. Dec. 19. SangWon Cho. Korea. JongHeon Lee. 1920. ByoungWan Kim. InWon Lee. Kent Ridge. SangWon Cho. . 4. Vol. SangWon Cho. (in Korean) 7. 2003. pp. InWon Lee. pp. HyungJo Jung . KangMin Choi. SangWon Cho. Italy." Journal of the Korean Society of Civil Engineers. "Simple and Efficient Control Algorithm for Seismic Response Reduction of Large Scale Structures using MR Dampers. pp. "Modal Control using MR Damper. ManGi Ko. 4. Sep. No. SangWon Cho. 1723. 2003. Kyeongju. "Modal Control for Seismic Structures using MR Dampers. No. 6. 8998. 2003. 243248. InWon Lee.5. "Algebraic Method for Sensitivity Analysis of Eigensystems with Repeated Eigenvalues. Vol. Rome." The Second International Conference on Structural and Construction Engineering.
2002. JiSeong Jo & InWon Lee. Korea. Korea. Korea.5." KAISTKyoto Univ. "Modal Control for Seismically Excited Structures using MR Damper. JuWon Oh. pp. DongHyawn Kim. Paper No. Sangwon Cho." The Eighth East AsiaPacific Conference on Structural Engineering & Construction. Aug. 2002. HyungJo Jung & InWon Lee. 6570 11. "A Comparative Study on Aseismic Performances of Base Isolation Systems for MultiSpan Bridge.org). 2002. Hong Kong. 2002." KAISTKyoto Univ. Busan. pp. "Modified Sturm Sequence Property for Damped Systems." The Second ChinaJapanKorea Symposium on Optimization of Structural and Mechanical Systems (CJKOSM 2). WoonHak Kim. Singapore. 9. February 25. 7. Korea. SangWon Cho. 6." ASCE's First Virtual World Congress for Civil Engineering (www. Joint Seminar on Earthquake Engineering. Joint Seminar on Earthquake Engineering. KAIST. "Implementation of Modal Control for Seismically Excited Structures using MR Damper. "Intelligent Control of Structural Vibration using CMAC. SangWon Cho & InWon Lee. 48. "NeuroControl for Seismic Response Reduction using a Semiactive MR Fluid Damper. Busan. SangWon Cho & InWon Lee. SangWon Cho. 2001. February 25." The Second ChinaJapanKorea Symposium on Optimization of Structural and Mechanical Systems (CJKOSM 2). SangWon Cho. HyungJo Jung. Nov. 1401. 125132 10. 2002. 1920. KyuSik Park.ceworld. 8. . ByoungWan Kim." International Conference on Advances and New Challenges in Earthquake Engineering Research. KyuSik Park. SangWon Cho. HeonJae Lee. "Modified BangBang Control of Seismically Excited Structures Using MR Damper. 2002. July 1. KAIST. Nov. InWon Lee. JiSeong Jo. December 57. InWon Lee. 48. Daejeon. InWon Lee. Online Conference.
2003. "Maximum Energy Dissipation Algorithm for Seismic Response Reduction of LargeScale Structures using MR Dampers. Gunsan. Australia. 2001. (In Korean) 16. InWon Lee. & WooHyun Yoon. SangWon Cho. JongHeon Lee. "Optimal Placement of MR Dampers for 20story Nonlinear Benchmark Building. November 2023. In Won Lee. InWon Lee. (In Korean) 18. (In Korean) 19." Conference on Earthquake Engineering Society of Korea. SangWon Cho. 8186 15. JongWoo Jang. August 2022. 3540. SangWon Cho. 2425. & WooHyun Yoon. Kyoto.153160." Conference on Korean Society of Civil Engineers. Oct. pp. Daejeon. pp.12. "Vibration Control for a Benchmark CableStayed Bridge using Maximum Energy Dissipation . 11. JongWoo Jang. Daejeon. SangWon Cho. JongWoo Jang." The Fourteenth KKNN Symposium on Civil Engineering. pp. & WooHyun Yoon "Optimal Placement of MR Dampers for 20story Nonlinear Benchmark Building. November 57. pp. Japan. (In Korean) 17. SangWon Cho & InWon Lee. InWon Lee & JuWon Oh. Sep. SangWon Cho. Oct. 14. InWon Lee. 2001. Daegu." Conference on Computational Structural Engineering Institute of Korea. KyuSik Park. 2425. Korea. SangWon Cho. "Efficient Mode Superposition Methods for NonClassically Damped System. Oct. pp. HyungJo Jung. "A Comparative Study on Aseismic Performances of Base Isolation Systems for Multispan Continuous Bridge. HyungJo Jung. 13. SunKyu Park. 467472. 2003. 2003." The First AsianPacific Congress on Computational Mechanics. Daegu. SangWon Cho. Sydney. 12771282. & InWon Lee. "Optimal Placement of MR Dampers for 20story Nonlinear Benchmark Building" Conference on Korean Society of Civil Engineers. 2003. DongHyawn Kim & InWon Lee. "NeuroControl of Structures using CMAC. 1999." The 12th KKNN Seminar/ Workshop on Civil Engineering. 19.
Jun. Sept. 89. HyungJo Jung. 549555. JuWon Oh & InWon Lee. (In Korean) 25. Apr. 23." Conference on Korean Society of Noise and Vibration Engineering. SangWon Cho. 2002. Nov. Ansan. Busan. pp. June 8. (In Korean) . 28." Conference on Korean Society of Steel Construction. 2003. 193200. 357363. Jun. (In Korean) 23. 2002. Jeju. DongOk Kim & InWon Lee. (In Korean) 22. SangWon Cho. JuWon Oh & InWon Lee. (In Korean) 24. 435441. ByoungWan Kim. "Algebraic Method for Computation of Eigenpair Sensitivities of Damped Systems with Repeated Eigenvalues.Algorithm." Conference on Korean Society of Civil Engineers. KyuSik Park. SangWon Cho & InWon Lee. 23. pp. SangWon Cho." Conference on Earthquake Engineering Society of Korea. Gunsan. pp.” Conference on Korean Society of Steel Construction. (In Korean) 20. 2002. 19. "Modified Decentralized BangBang Control Seismically Excited Structures Using MR Dampers. 2003. Mar. (In Korean) 21." Conference on Earthquake Engineering Society of Korea. pp. 12. 12. WoonHak Kim & InWon Lee." Conference on Earthquake Engineering Society of Korea." Conference on Computational Structural Engineering Institute of Korea. SangWon Cho. "Implementation of Method Control for Seismically Excited Structures Using MR Damper. Sangwon Cho. Sep. "Hybrid Control with a BangBang Type Controller. pp. 1999. 2002. 293300. KangMin Choi. Asan. Seoul. SunKyu Park & InWon Lee. SangWon Cho. Seoul. HyungJo Jung. "Efficient Mode superposition Method for NonClassically Damped Systems. HyungJo Jung. pp. "Efficient Mode Superposition Method for NonClassically Damped Systems. "Modal Control of Vibration using MR Damper. (In Korean) 26. 2000. WoonHak Kim & InWon Lee. Seoul. 141147. JiSeong Cho.
KAIST. 2324.” (Principal Investigator: InWon Lee). KAIST. 5. (In Korean) Research Reports: 1. February 2000. KAIST. September 2002. 471474. “Development of Efficient Dynamic Analysis Method for NonClassically Damped Systems. "Development of an Efficient Mode Superposition Method for Nonclassically Damped System. Hyundai E&C Co. 4.” Advanced Structural Engineering Technology Research Team. pp. Korean Patent. (Principal Investigator: InWon Lee).” (Principal Investigator: InWon Lee).” National Research Laboratory (NRL) Program. Ltd. Ministry of Education. September 2002. (in Korean). (in Korean). . “Development of Integrated Aseismic Control Systems for Structures by Advanced Vibration Control Techniques. (in Korean). “Technology for Improving WindResistant Performance of LargeScaled Bridges. SunKyu Park & InWon Lee. (in Korean). MR damper with electromagnetic induction system to replace power source and sensor. “Development of PreDesign Program for LRB. ManCheol Kim. “Research Report for Brain Korea 21 (BK21) Program. Application no: 200261823. Seoul.. Ministry of Science & Technology. Oct.27. March 2000. 3. Unison Industrial. Korea Institute of Construction Technology in Ministry of Construction & Transportation." Conference on Korean Society of Civil Engineers. December 1999.” (Principal Investigator: InWon Lee). 1998. 2. Patent: 1. SangWon Cho. (Principal Investigator: ChungBang Yun). October 2002. KAIST.