S R C O B S U M - R (MR) D: Eismic Esponse Ontrol F Raced Tructures Sing Agneto Heological Ampers

i
S EISMIC R ESPONSE C ONTROL O F B RACED S TRUCTURES

U SING M AGNETO -R HEOLOGICAL (MR) DAMPERS
A Thesis Submitted
in Partial Fulfilment of the Requirements
for the Degree of
Master of Technology
by
Mariyam
DEPARTMENT OF CIVIL ENGINEERING
INDIAN INSTITUTE OF TECHNOLOGY,KANPUR
July 10, 2015

iii
ABSTRACT
In this study, the effectiveness of scissor jack (amplified) bracing systems comprising of
Magneto-rheological (MR) dampers for seismic response control of shear building is investigated.
At first, comparative study has been conducted between the performance of scissor jack (amplified)
bracing and chevron (unamplified) bracing for passive control of MR dampers. For this purpose,
a three-story shear building with damper installed at the first floor is considered, when subjected
to different seismic excitations. Results of this study revealed that higher control force and better
control performance can be obtained for the scissor jack bracing as compared to the chevron brac-
ing. Further, different semiactive control algorithms including clipped-optimal LQR algorithm,
algorithm based on lyapunov stability, decentralized bang-bang control algorithm, and maximum
energy dissipation algorithm have been used to evaluate their performances. It has been observed
that the performance of different control algorithm depends on the ground motions selected. How-
ever, a better control is always observed for the scissor jack bracing as compared to the chevron
bracing system. It has been observed that clipped-optimal LQR algorithm provides better results
as compared to most of the other control algorithms including active control. Further, a numerical
study has been conducted for obtaining the optimal location of MR damper for eight-storey shear
building using modal controllability factor. Effectiveness of damper location has also been studied
by comparing the response of structure for different locations of MR damper for both chevron and
scissor jack bracings. It is observed that, when the structure is excited to fundamental mode, first
floor location is most effective. Finally, a new algorithm (direct control algorithm) has been devel-
oped for inter-storey drift control using LQR approach for continuous current/volatge variation.
The effectiveness of the proposed algorithm for seismic control has been verified for a 8-storey
structure with damper installed at the first floor in scissor jack bracing configuration. Comparative
study has been conducted between the response obtained by direct control algorithm and clipped-
optimal LQR algorithm. It has been found that the direct controller outperforms clipped-optimal
controller for control of inter-storey drift and acceleration response.

iv
ACKNOWLEDGEMENTS
I would like to express my deepest and sincere gratitude to my thesis supervisor, Dr. Samit
Ray Chaudhuri for his constant encouragement and support during my thesis, which enabled the
successful completion of this work. I am thankful to him not only for his valuable guidance but
also the friendliness, that allowed me to speak my mind in front of him without any hesitation.
Both as a teacher and a guide he has been a constant source of inspiration for me and I will always
cherish the fact that he was my supervisor. The things he taught are surely going to last throughout
the journey of my life.
I would also like to thank my professors at IIT Kanpur, Samit ray Chaudhuri, Vinay K. Gupta,
Sudib K. Mishra, Durgesh C. Rai, Sudhir Mishra, S.K.Chakraborty and Rajesh Sathiyamoorthy,
for broadening and enriching my technical knowledge. I am also thankful to Sanjukta di for
guiding me throughout my work.
At last, I express my gratitude to my family for their love, affection, care and inseparable
support without which I would never have been able to achieve anything.
Mariyam
Contents
Page
Abstract iii
Acknowledgement iv
1 Introduction 10
1.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.2 Control Strategy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.2.1 Passive control systems . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.2.2 Active control systems . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.2.3 Semiactive control systems . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.3 Literature Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
1.3.1 Magneto-rheological dampers . . . . . . . . . . . . . . . . . . . . . . . 14
1.3.2 Amplified Bracings : Toggle Based Theory . . . . . . . . . . . . . . . . 21
1.4 Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
1.5 Organization of thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
2 Passive Control of a 3-storey shear building using Magneto-rheological (MR) damper 28
2.1 Brief Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
2.2 State-Space formulation of n DOF shear building model . . . . . . . . . . . . . 29
2.2.1 Uncontrolled case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

2
2.2.2 Controlled case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
2.3 3-Storey Shear Building Model . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
2.4 Magneto-rheological (MR) damper model . . . . . . . . . . . . . . . . . . . . . 36
2.5 Bracing Description (Amplified Bracing) . . . . . . . . . . . . . . . . . . . . . 40
2.5.1 Upper Toggle System . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
2.5.2 Lower Toggle System . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
2.5.3 Reverse Toggle System . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
2.5.4 Scissor Jack Toggle System . . . . . . . . . . . . . . . . . . . . . . . . 44
2.6 Equation of motion for Passive Case of MR damper . . . . . . . . . . . . . . . . 48
2.7 Ground motions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
2.8 Results and Discussions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
2.9 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
3 Control of a 3-storey shear building using LQR algorithm 61
3.1 Brief Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
3.2 Active Control using LQR algorithm (Full State feedback) . . . . . . . . . . . . 61
3.2.1 Feedback Control laws . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
3.2.2 Controllability and Observability . . . . . . . . . . . . . . . . . . . . . 66
3.2.3 Ricatti Optimal Active Control (ROAC)/Linear Quadratic Regulator (LQR)
Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
3.2.4 Generalized Optimal Active Control (GOAC) Algorithm . . . . . . . . . 69
3.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
4 Semi-active Control of three story structure 80
4.1 Brief Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
4.2 Semi-active control analysis of a 3-storey structure . . . . . . . . . . . . . . . . 80

3
4.3 Control Algorithms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
4.3.1 Clipped Optimal approach . . . . . . . . . . . . . . . . . . . . . . . . . 82
4.3.2 Control Based on Lyapunov Stability theory . . . . . . . . . . . . . . . . 84
4.3.3 Decentralized Bang-Bang Control . . . . . . . . . . . . . . . . . . . . . 85
4.3.4 Maximum Energy Dissipation . . . . . . . . . . . . . . . . . . . . . . . 86
4.5 Effect of capacity of dampers . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
4.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
5 Semi-active Control of eight storey shear Building 102
5.1 Brief Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
5.2 Shear Building Model Description . . . . . . . . . . . . . . . . . . . . . . . . . 103
5.3 Modal Controllability for Location Optimization . . . . . . . . . . . . . . . . . 106
5.4 Analysis of 8 storey structure with different damper locations . . . . . . . . . . 109
5.4.1 Results and Discussions . . . . . . . . . . . . . . . . . . . . . . . . . . 111
5.5 Semiactive Direct Control of Inter-Storey Displacement . . . . . . . . . . . . . . 114
5.5.1 Formulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
5.5.2 Results and Discussions . . . . . . . . . . . . . . . . . . . . . . . . . . 121
5.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
6 Summary and Conclusions 127
6.1 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
6.2 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
6.3 Future scope of the study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129

List of Figures
1.1 Block diagram of structural control systems: (a) Passive control system, (b) Active
control system and (c) Semi-active control system (Symansa and Constantinou,
1999) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
1.2 Schematic of MR Damper (B.F. Spencer Jr, 1996) . . . . . . . . . . . . . . . . . 15
1.3 Activation of MR fluid: (a) no magnetic field applied; (b) magnetic field applied;
(c) ferrous particle chains have formed (Lord Corporation, www.lord.com) . . . . 16
1.4 Working modes of MR fluid: (a) Valve mode; (b) shear mode; (c) squeeze mode
(Carlson and Jolly, 2000) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
1.5 Possible Installation Configurations of Scissor-Jack Damping System (Sigaher and
Constantinou, 2003) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
1.6 Toggle based configuration connected to beam-column joints (Hwang et al., 2005) 23
1.7 Installation configuration of MR damper (Lee et al., 2007) . . . . . . . . . . . . 25
2.1 Flowchart for solving state equation without feedback . . . . . . . . . . . . . . . 33
2.2 Three storey shear building . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
2.3 Behavior of Bingham material as described (a) in the stress-strain rate plane and
(b) in the force-displacement plane (Symansa and Constantinou, 1999) . . . . . . 37
2.4 Bingham Model (B.F. Spencer Jr, 1996) . . . . . . . . . . . . . . . . . . . . . . 38
2.5 Bouc-Wen Model (B.F. Spencer Jr, 1996) . . . . . . . . . . . . . . . . . . . . . 39
2.6 Modified Bouc-Wen Model (B.F. Spencer Jr, 1996) . . . . . . . . . . . . . . . . 40

5
2.7 Movement of upper toggle system (Constantinou et al., 2001) . . . . . . . . . . . 41
2.8 Dependency of magnification factor on geometry (Constantinou et al., 2001) . . . 42
2.9 Movement of lower toggle system (Constantinou et al., 2001) . . . . . . . . . . . 42
2.10 dependency of magnification factor on geometry (Constantinou et al., 2001) . . . 43
2.11 Movement of reverse toggle system (Constantinou et al., 2001) . . . . . . . . . . 43
2.12 dependency of magnification factor on geometry (Constantinou et al., 2001) . . . 44
2.13 Movement of Scissor Jack toggle system (Sigaher and Constantinou, 2003) . . . 45
2.14 dependency of magnification factor on geometry (Sigaher and Constantinou, 2003) 46
2.15 Chevron Bracing Arrangement . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
2.16 Scissor-Jack Bracing Arrangement . . . . . . . . . . . . . . . . . . . . . . . . . 47
2.17 GM1 : (a) Displacement, (b) Inter-story drift and (c) Acceleration response . . . 52
3.1 Schematic of open-loop feedback control . . . . . . . . . . . . . . . . . . . . . 63
3.2 Schematic of Closed-loop feedback control . . . . . . . . . . . . . . . . . . . . 64
3.3 Schematic of Open-Closed-loop feedback control . . . . . . . . . . . . . . . . . 64
3.4 Flowchart for solving state equation with state feedback . . . . . . . . . . . . . . 65
3.5 GM1 : (a) Displacement, (b) Inter-storey Drift and (b) Acceleration response . . 75
3.10 GM1 : Displacement time history of top floor . . . . . . . . . . . . . . . . . . . 78

6
4.1 Block diagram of Semi-Active Control . . . . . . . . . . . . . . . . . . . . . . 82
4.2 GM1 : (a) Acceleration response, and (b) Displacement response . . . . . . . . 89
4.6 GM5 :(a) Acceleration response, and (b) Displacement response . . . . . . . . . 98
5.1 8-story shear building model . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
5.2 Mode shapes of the Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
5.3 Damper Location . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
5.4 Response of structure when it is subjected to excitation frequency ω = ω1 : (a)
Displacement, (b) Inter-story drift, (c) Floor Acceleration response. . . . . . . . 111
5.7 Response of structure when it is subjected to El-centro earthquake : (a) Displace-
ment, (b) Inter-story drift, (c) Floor Acceleration response. . . . . . . . . . . . . 114

List of Tables
1.1 Typical MR fluid properties (Carlson and Jolly, 2000) . . . . . . . . . . . . . . . 15
2.1 Frequencies (in Hz) of the three story structure . . . . . . . . . . . . . . . . . . 35
2.2 Parameters of Bouc-Wen Model of 1000 kN MR damper . . . . . . . . . . . . . 49
2.3 Ground Motion Used . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
2.4 Response of building when excited to GM1 . . . . . . . . . . . . . . . . . . . . 53
3.1 Different R values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
3.2 Uncontrolled Response of structure . . . . . . . . . . . . . . . . . . . . . . . . . 71
3.3 Maximum displacement and acceleration response for GM1 . . . . . . . . . . . 72
3.8 Percentage reduction in displacement and acceleration response . . . . . . . . . 77
4.1 Different Semi-active Control strategies . . . . . . . . . . . . . . . . . . . . . . 87

8
4.2 Maximum response of building for GM1 . . . . . . . . . . . . . . . . . . . . . . 90
4.7 Percentage reduction in displacement, acceleration and interstory drift at roof of
building for GM1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
building for GM2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
building for GM3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
building for GM4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
building for GM5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
4.12 Response of building when excited with GM1 for damper capacity of 2000 kN
(Amplification 2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
5.1 Frequencies (in Hz) of the 8-story structure . . . . . . . . . . . . . . . . . . . . 105
5.2 Mode shapes for the 1st three modes φi . . . . . . . . . . . . . . . . . . . . . . 105
5.3 Location Vector {γ} . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
5.4 Modal Controllability Factors . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
5.5 Different Cases for Anaysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
5.6 Percentage reduction in displacement response . . . . . . . . . . . . . . . . . . 115
5.7 Percentage reduction in interstory drift . . . . . . . . . . . . . . . . . . . . . . . 116
5.8 Percentage reduction in floor acceleration . . . . . . . . . . . . . . . . . . . . . 117
5.9 Maximum response of building for GM1 . . . . . . . . . . . . . . . . . . . . . 122

9

Chapter 1
Introduction
1.1 Background
In conventional seismic design of structures, energy dissipation occurs due to the inelastic action
or plastic deformation of the structural frame. The elements that undergo plastic deformation
generally form a part of gravity-load-resisting system. Any damage to such systems during an
earthquake is costly and may lead to loss of functionality or even collapse of the structure. In
order to reduce such vulnerability or reduce wind-induced vibration in tall structures, supplemental
energy dissipation devices are often installed in structures. These devices can easily be replaced
after a damaging event. These supplemental energy dissipation devices are classified into two
categories. One is a mass-type damper such as a tuned mass damper or a tuned liquid damper,
which utilizes the inertial force of an auxiliary mass. Another is the relative response dependent
damper in which damping force is a function of relative displacement or velocity between two
ends of the damping device. The first category is also referred as vibration absorbers. These are
very effective for response control of structures when one mode dominates in response. Due to
this characteristic they are mainly used to suppress wind excitation of structures. Some of the
structures fitted with this type of device include CN Tower (Toronto), Hancock tower (Boston).
A limitation of these dampers is that they are less suited for seismic response control, as they are
11
effective only for one mode. Also, they occupy large space and are very sensitive to mistuning.
The second category of dampers include viscous fluid dampers, magneto-rheological dampers,
electro-rheological dampers. They require some bracing arrangements for their installation. For
these dampers, damping force is a function of inter-storey displacements. For a typical building
Inter-storey responses are very small as compared to global responses (absolute displacement),
leading to the use of more number of or higher capacity of dampers to achieve a target control
performance. These dampers are generally installed in diagonal or chevron bracings, in which the
damping force obtained in structure becomes less than or equal to the force produced in damping
device. Anathor system of bracings include toggle based bracings (scissor jack bracings) for
their installation, which are also known as amplified bracings. The advantage of this bracing
system is that it amplifies the response between the two ends of damper as compared to the actual
inter-storey response of structure. Also the force delivered to the structure gets magnified and is
higher than the force produced in dampers. Thus amplified bracings encourages the use of smaller
capacity of dampers to achieve higher damping force and better control performance, which means
economical efficiency.
In the present study, Magneto-rheological (MR) dampers, due to their effectiveness, have been
used for seismic response control of shear buildings. Two types of bracings have been considered,
one is chevron and the other is scissor jack bracing for installation of damper. The responses
of two types of bracings are compared to obtain the effectiveness of scissor jack bracing system.
Analytical study has been conducted using passive and semiactive control approach using different
control algorithms and response of structures is evaluated.
1.2 Control Strategy
Structural vibration control is achieved mainly by three methods: passive control, active control,
and semi-active control. In passive control there is no external power requirement. In active
control large amount of energy through some external power source is required. However semi-
12
active control has a very less demand of external power which can be fulfilled by battery.
1.2.1 Passive control systems
Passive control is most widely used control strategy for mitigating the effects of seismic and wind
excitation. Control forces are obtained due to the relative motion within devices which is produced
due to motion of structure. These dampers operates on different principles such as friction sliding
(friction dampers), yielding of metals (hysteretic and metallic dampers), phase transformation in
metals (shape memory alloys), deformation of viscoelastic solids (viscoelastic dampers) and fluid
orificing (fluid dampers). It also includes vibration absorbers like tuned mass dampers (TMD),
tuned liquid dampers (TLD) and tuned liquid column dampers (TLCD). However, they have the
disadvantage of not being able to adapt themselves depending on time and different operating
conditions. The performance of passive devices is not necessarily optimal for all loading condi-
tions, as these devices are unable to incorporate any ongoing structural change and varying usage
patterns. To overcome these limitations, active and semiactive control strategies can be adopted.
1.2.2 Active control systems
An active control system may be defined as a system which typically requires a large power source
for their operation. It consists of sensors, control actuator, control computer. Sensors measure the
response of structure. Control forces are developed based on feedback from sensors using feed-
back control laws. Actuators are required to supply control forces to the structure. The feedback
from the structural response may be measured at locations remote from the location of the ac-
tive control system. Weakness of active control system is that it requires large power source
which is not generally available in the events of earthquake. Also system can become unstable
as external energy is added to the system. Different active control mechanisms include the active
bracing system (Reinhorn et al., 1989), active tendon system (Roorda, 1975; Abdel-Rohman and
Leipholz, 1978), the active tuned mass damper/driver (Abdel-Rohman and Leipholz, 1983; Chang
13
and Soong, 1980), and the active aerodynamic appendage mechanism (Soong and Skinner, 1981;
Abdel-Rohman, 1984).
1.2.3 Semiactive control systems
These are originated from passive devices with some modifications in their mechanical proper-
ties. A semiactive control system may be defined as a system which typically requires a small
external power source for operation (e.g. a battery) and utilizes the motion of the structure to
develop the control forces (similar to passive devices). Similar to active control system, feed back
measurements are monitored to generate the appropriate command signal. These devices include
stiffness control devices (Kobori et al., 1993; Nemir et al., 1994; Yang et al., 1996), friction control
devices (Akbay and Aktan, 1990; Feng et al., 1993), fluid viscous devices (Symans and Constanti-
nou, 1997b,a) TMDs (Abe, 1996; Hrovat et al., 1983)/TLDs (Kareem, 1994; Lou et al., 1994)
and electro-rhelogcal(ER) dampers (Ehrgott and Masri, 1992b,a; Makris et al., 1995) /magneto-
rheological (MR) dampers (Dyke et al., 1996b,a, 1997b,a). The main advantage of this system is
that it offers the reliability of passive control devices, yet maintains the adaptability and versatility
of active systems. It consists of sensors, control actuator, control computer and passive energy
dissipation device. Sensors are required to measure excitation or structural response in case of any
event. Control computer processes the measurement received from sensors and produces a signal
for control actuator. Control actuator modifies the mechanical properties of damper as per signal
received from computer. Unlike in active control, in this approach external energy is not added to
the system, so the system cannot become unstable.
Variety of semiactive control algorithms have been been developed to generate the appropriate
command signal to the damping devices. These include (i) Clipped optimal control, (ii) Lyapunov
stability approach, (iii) Decentralized bang-bang control, (iv) Maximum energy dissipation ap-
proach, (v) Modulated homogeneous friction control, and (vi) Riccati matrix solution technique
(vii) Neural network control and (viii) Fuzzy Logic control etc.
14
The block diagrams for Passive, Active and Semi-active control strategies are shown in Figure
1.1
Figure 1.1: Block diagram of structural control systems: (a) Passive control system, (b) Active
control system and (c) Semi-active control system (Symansa and Constantinou, 1999)
1.3 Literature Review
This section is divided into two parts. In the first part literature review of MR dampers is presented
and in the second section work done on amplified braces is briefly discussed.
1.3.1 Magneto-rheological dampers
In semiactive control strategy Magneto-rheological dampers are most common because of their
mechanical simplicity, large force capacity, low power requirement, and robustness. Magneto-
rheological (MR) dampers are semiactive control devices that utilize MR fluids to obtain con-
trollable damping forces. MR fluids consists of micron sized magnetically polarizable particles
(usually ferric particles) suspended within a non magnetic medium (usually oil). The size of MR
15
particles typically range from 10−7 to 10−5 . Carrier liquids used are typically chosen based upon
petroleum based oils, silicone, polyesters, polyethers, mineral oils, water, synthetic hydrocarbon
oils etc (Carlson and Jolly (2000)). MR fluids can reversibly change their form from liquid to semi-
Figure 1.2: Schematic of MR Damper (B.F. Spencer Jr, 1996)
solid by the application of magnetic field. They can achieve a very high order of yield strength
with the requirement of a very low voltage.
B.F. Spencer Jr (1996) indicated that MR fluids can achieve yield stress higher than its ER
conterpart and they can operate at temperatures from 40 to 150o C with only slight variations in the
yield stress. Moreover, Unlike ER fluids, MR fluids are not sensitive to contaminants and impuri-
ties which are commonly encountered during manufacturing and usage. Further a wider choice of
additives (surfactants, dispersants, friction modifiers, anti-wear agents,etc.) can generally be used
with MR fluids to enhance stability, seal life, bearing life, etc. MR fluid can be controlled with
very low voltage. Some of the typical properties of MR fluids are listed in Table 1.1.
Table 1.1: Typical MR fluid properties (Carlson and Jolly, 2000)

Property Typical Value
Maximum yield strength τy(f ield) 50-100 kPa
Maximum field 250 kA/m
Plastic viscosity,ηp 0.1-1.0Pa s
Operable temperature range -40 to 150o C (limited by carrier fluid)
Contaminants unaffected by most impurities
Response time < milliseconds
Density 3-4 g/cm3
2
ηp /τy(f 10-10 -10-11 s/Pa
ield)
Maximum energy density 0.1 J/cm3
Power supply (typical) 2-25 V @ 1-2 A (2-50 watts)
16
In the absence of magnetic field the MR fluid acts as a freely flowing viscous fluid. When
magnetic field is applied ferrous particles align themselves along the direction of magnetic flux
forming chains. It restricts the fluid movement and increases its yield strength. Mechanism is
shown in Figure 1.3.
Figure 1.3: Activation of MR fluid: (a) no magnetic field applied; (b) magnetic field applied; (c)
ferrous particle chains have formed (Lord Corporation, www.lord.com)
There are three main modes of operation for MR fluids.These are (1) valve mode, (2) shear
mode, (3) squeeze mode (Carlson and Spencer, 1996). In valve mode the end plates containing
MR fluid are stationary. Fluid flow takes place in direction perpendicular to magnetic flux due
to pressure gradient between two ends of plate. This pressure driven flow is commonly referred
to as Poiseuille flow. Devices using this mode of operation include servo-valves, dampers, shock
absorbers and actuators. In shear mode, the fluid is contained between two parallel plates sliding
relative to each other in direction perpendicular to magnetic flux. The drag force applied to the
surface from fluid changes by change in viscosity of fluid which can be controlled by magnetic
field. Devices using this mode of operation include clutches, brakes, locking devices and dampers.
In the squeeze mode, the fluid is contained between two plates moving in the direction of mag-
netic flux. MR fluids are subjected to alternate tension and compression forces. As magnetic
field is applied the particle align themselves and form a chain between plates and becomes rigid.
Displacements are small compared to the other modes (in the order of millimetres) but resistive
forces are high. As for the two other modes, the magnitude of these resistive forces can also be
controlled by modifying the magnetic field between the poles. While less well understood than the
17
other modes, the squeeze mode has been explored for use in small amplitude vibration and impact
dampers.
Figure 1.4: Working modes of MR fluid: (a) Valve mode; (b) shear mode; (c) squeeze mode
(Carlson and Jolly, 2000)
Different phenomenological models have been proposed to capture the highly non-linear be-
havior of MR fluids including Bingham model proposed by Stanway et al. (1985, 1987), extension
of bingham model proposed by Gamota and Filisko (1991), which is based on viscoelastic-plastic
model.
B.F. Spencer Jr (1996) proposed Bouc-Wen model which predicts the hysteresis behavior of
MR fluids very well. These are divided in two categories: simple Bouc-Wen model and modified
Bouc-Wen model. Modified Bouc-Wen model consists of an arrangement of linear springs, linear
viscous dashpots in addition to a Bouc-Wen (Wen, 1976) hysteresis element. The prototype MR
damper used for deriving model parameters was obtained from Lord Corporation (www.lord.com).
The damper had a capacity of 3 kN with a stroke length of ±2.5 cm. Length of damper was 21.5
cm in its extended position, and the diameter of main cylinder was 3.8 cm. A constrained non-
linear optimization was used to obtain the MR damper model parameters (Spencer et al., 1996).
When numerical simulation results were compared with experimental data over different oper-
ating conditions, including step voltage, random displacement/random voltage, and random dis-
placement/constant voltage tests, it was found that the proposed model accurately predicted the
response of the MR damper .
Dyke et al. (1996b,a, 1997b,a) had performed shaking table tests on a small scale three-story
18
steel structure subjected to earthquake ground motion. MR damper was located in the first story.
Damper used for testing was same as described by B.F. Spencer Jr (1996). A clipped-optimal
control algorithm was developed for semiactive control of structure with acceleration feedback.
Two passive cases were also tested by setting the voltage at 0 and 2.25 V. A comparison was made
between all the cases and semiactive approach was found to be superior than both the passive
cases. Spencer et al. (1997) and Carlson and Spencer (1996) have developed a 200kN large scale
MR damper for seismic response control of full scale structures.
Jung et al. (2003b) used MR dampers for seismic response control of cable-stayed bridges. The
ASCE benchmark cable-stayed problem was investigated, which was based on the Cape Girardeau
Bridge in Missouri (Dyke et al., 2003). A clipped optimal control algorithm was used to control
the MR dampers. The controller design employed the use of twenty-four 1000 kN capacity MR
fluid dampers, which were installed at four different locations along the bridge. The bridge was
subjected to three different earthquakes : 1940 El Centro, California, 1985 Mexico City, and 1999
Gebze, Turkey. From the study it was found that the performance of semiactive controller was
nearly same as the active system. 7%-69% reduction in all the responses was observed, except
for peak deck displacement and peak shear at deck level. From the results, it was concluded that
proposed controller was a viable option for controlling the vibration response of the cable-stayed
benchmark bridge.
Jansen and Dyke (2000) used semiactive control algorithms for the control of multiple MR
dampers. A variety of algorithms were used including clipped optimal controller, lyapunov con-
troller, decentralized bang-bang controller, and modulated homogeneous friction control algo-
rithm. Small scale six storey structure was used for numerical study with four MR dampers. Two
dampers each were installed at both first and second storey. Structure was subjected to scaled
El-centro earthquake and it was found that each algorithm resulted in a better performance as
compared to passive case. Two types of clipped-optimal controller were used: One with moderate
weighing on the displacements of each floor and the other was designed with higher weight on the
19
floor displacements. First case was very effective in controlling the inter-storey drift and accelera-
tion. 29% more reduction in acceleration was found over the best passive case, which also resulted
in lowest acceleration among all the cases. Second case resulted with better response control for
displacement and inter-storey drifts. However acceleration response was increased in second case.
Hiemenz et al. (2003) used MR dampers in active bracings to control the seismic response
of a 2D three-story scaled-model frame (60 inch tall). It was observed that the sliding mode
controller (SMC) provided 10 percent more reduction in displacements and acceleration responses
as compared to LQR and skyhook controllers (damping control force is applied only when force
and velocity have same sign).
Liu et al. (2005) used MR fluid dampers for semi-active control of bridges. Shake table tests
were performed on a 1:12 scale highway bridge model. Two MR dampers were installed and
different control algorithms such as lyapunov-based control, energy minimization algorithm, fuzzy
logic control (FLC), and variable structure system fuzzy logic (FLC with addition of a sliding
mode) were used for response control. From the results it was concluded that all the algorithms
were effective in decreasing the RMS deck displacements as compared to the uncontrolled case.
FLC showed best results for response control with the requirement of least amount of power.
Xu et al. (2005) had done experimental investigation to access the effectiveness of semiac-
tive control using MR dampers for seismic response control of buildings with podium structure.
A slender 12-story, 2.4-m tall steel building model was built with a surrounding relatively stiff
three-story, 0.6-m tall podium structure. The structure was subjected to the scaled 1940 El Cen-
tro earthquake ground motions. Experimental response was obtained for four different cases : (i)
Uncontrolled case with no connection between the inner building model and podium, (ii) Uncon-
trolled case with a rigid connection between the inner building model and podium, (iii) a passive-
off (no voltage applied) MR damper connecting the inner building model and podium, and (iv)
a semiactive controlled MR damper connecting the inner building model and podium structure.
Multilevel logic control algorithm was use for semiactive control and it was found that semiactive
20
control was effective in controlling RMS displacement responses upto 70% and acceleration re-
sponses upto 60% as compared to uncontrolled response, and upto 34% and 25% , respectively, as
compared to the passive control system.
Yoshida and Dyke (2005) used MR dampers to control the seismic response of full scale irreg-
ular 3D buildings. Two different cases were considered: (i) a nine-story, 40.25-m high, composite
steel-reinforced concrete office building in Japan with an asymmetric plan due to the placement
of shear walls, (ii) The other was an L-shaped, eight-story, 35.1-m high, steel braced benchmark
building (Narasimhan et al., 2002) in which irregularity was due to setbacks. Two earthquake
motions were used to obtain the response: 1940 El Centro earthquake, and 1995 Kobe earthquake.
For the first case 110 MR dampers with 1000 kN force capacity were installed and simulation was
done for one dimensional ground motion (North-South component). For the second case 146 and
168 MR dampers were installed in x and y-directions respectively and building was subjected two
dimensional ground motion (both North-South and East-West component simultaneously). Ge-
netic Algorithms (GAs) was used for MR damper device placement (Adeli and Cheng, 1994a,b;
Kang et al., 2009; Al-Bazi and Dawood, 2010; Lee and Wei, 2010; Cheng and Yan, 2009; Kim
and Adeli, 2001; Jiang and Adeli, 2008). A clipped-optimal control algorithm with H2/LQG con-
troller was used. For case I, 40 to 45% response reduction was observed for both the maximum
inter-story drift, and the maximum acceleration for the El Centro earthquakes and 35 to 40% for
the Kobe earthquakes. For Case II, reduction in maximum acceleration responses was 40 to 52%
for the El Centro earthquake and 49 to 51% for the Kobe earthquakes, and reduction in the max-
imum interstory drifts was 45 to 48% for the El Centro earthquakes and 46 to 47% for the Kobe
earthquakes as compared to the uncontrolled response. In some of the cases, semiactive system
performance was better than the ideal active system in reducing the interstory drift.
Carneiro et al. (2010) used MR dampers to mitigate the response of a two storey three-
dimensional seismically excited building model. Structure had 4m total height, each floor being
2m and a plan dimension of 3m x 4m (Contento et al., 2006). Two MR dampers (RD-1005-3) were
21
attached between ground and the first floor of structure and modified Bouc-Wen model was used
for damper modelling. Clipped-optimal controller based on Linear quadratic regulator (LQR) al-
gorithm was used for control. It was found that the best semiactive configuration reduced the peak
acceleration upto 70% and peak displacemnts upto 62 % as compared to uncontrolled response.
Also controller was 3% more efficient in reducing peak diaplacement and 7% more efficient for
controlling peak accelerations than passive on case. Also controller performed better than active
control using same maximum control forces.
Rahbari et al. (2012) used a direct semi-active control algorithm to control the seismic re-
sponses of structures using MR dampers. Optimal control force was calculated using LQR algo-
rithm and optimal voltage was obtained so that the damper force was close to the Optimal control
force. Unlike previously adopted semi-active control ON-OFF strategies, this controller allowed
the selection of any voltage between 0 and Vmax (maximum allowable damper voltage). MR
damper Bouc-Wen model was used to obtain damper force. Two types of shear building were
used: (i) three storey building with two 1000 kN damper installed at first floor, and (ii) eleven
storey building with three 1000 kN dampers, one one each at the top three stories of the build-
ing. Structures were excited with three earthquake ground motions including Imperial valley,
Northridge and Cape Mendocino earthquake. Cost function for LQR control was chosen in such
a way that it weighed the third floor absolute acceleration for three storey building and all floor
accelerations with highest weight imparted to the top floor acceleration, for eleven-storey building.
From the results it was concluded that direct controller performed better in controlling absolute
acceleration of stories, however the variation was small as compared to Clipped Optimal approach.
1.3.2 Amplified Bracings : Toggle Based Theory
Number of configurations have been used to magnify the displacement of damper especially in
case of stiff structures where drift is low. Diagonal and chevron based bracing arrangements
utilizing viscous damper are quite ineffective for controlling the response of stiff structures or in
22
the structures where interstorey drift is very low . In that case small damping forces are produced
in damper and also the force transferred to the main frame is equal to (chevron brace) or less than
(diagonal brace) the damper force, leading to the use of increased number of dampers which may
increase cost. Several bracing configurations have been proposed for the installation of dampers
to magnify the displacements and velocities in dampers. Toggle based bracings are very effective
for magnifying the effect of damping devices. They make use of shallow truss configuration
and magnify damper displacement. Force delivered to the structural frame is also amplified and
transferred in form of tension or compression in the braces.
Toggle-brace-damper system was proposed and patented by Taylor and Tonawanda (1999).
Constantinou et al. (2001) investigated the toggle-brace-damper system and verified its ability to
amplify the axial displacements of dampers and the efficiency of energy dissipation through both
cyclic loading tests and shaking table tests with a single degree of freedom steel model. Three
new configurations were presented based on toggle-brace mechanisms to substantially magnify
the effect of damping devices. These configurations include upper, lower and reverse toggle sys-
tem. Shake table testing of a large scale steel model structure with a weight of 143 kN was done.
Theoritical expressions for magnification factor, damping ratio and forces in the toggles were ob-
tained and verified through experimental results to confirm the validity of developed theory. Shake
table testing of a half-length scale model was done equipped with linear fluid viscous dampers.
The dampers used in the testing were linear fluid viscous devices, with c0 = 15.7 N-s/mm. The
devices had a diameter of 45 mm, length of 467 mm, and a stroke of 650 mm. Details on the test
model, including manufacturing drawings were presented in report by Constantinou et al. (1997).
Damping ratio was found to increase from 0.04 (without dampers) to 0.26 (with dampers). Tog-
gle system were found to be very effective for reducing drift and acceleration with a substantial
increase in the damping ratio.
Sigaher and Constantinou (2003, 2004) demonstrated that the scissor-jack-damper system, a
variant of the toggle-brace-damper system, can also enlarge the damper displacements or velocity
23
and enhance the efficiency of energy dissipation without occupying the whole bays in frames,
which may fit for an architectural requirement. Some of the possible installation configurations of
Scissor-Jack Damping System are shown in Figure 1.5.
Figure 1.5: Possible Installation Configurations of Scissor-Jack Damping System (Sigaher and
Constantinou, 2003)
Hwang et al. (2005) proposed the design formulas for both lower and upper toggle systems
with dampers directly installed to the beam-column joints to facilitate their practical applications
as shown in Figure 1.6.
Figure 1.6: Toggle based configuration connected to beam-column joints (Hwang et al., 2005)
Magnification factor was formulated for both upper and lower toggle. Geometric constraints
had been been imposed to facilitate practical implementation. In addition, shaking table tests were
conducted using a three-story steel structure with and without linear viscous dampers. Test was
done for both diagonal and upper toggle braces to investigate the effectiveness of toggle braces
24
in seismic response control. Toggle-brace system was found to be more effective as compared to
diagonal brace system in enhancing the seismic response control on a stiff structure.
Lee et al. (2007) considered non linear velocity amplification factors which was neglected
in earlier studies. Bingham model for MR dampers was used for response evaluation. Numeri-
cal study was done to obtain the control performance of a three-storey frame structure with one
MR damper installed at the first floor using chevron, diagonal, and upper toggle braces. The MR
damper with product No. RD-1097-01 manufactured by Lord Company was used. The El Centro
1940 NS component was used as input earthquake load, with different PGA values so that the
nonlinearity of the upper toggle system could be considered. It was observed that the performance
of non amplifying braces deteriorated with increasing PGA while the toggle braces remained ef-
fective for all the cases. Difference between the performance indices between linear and non linear
toggle was not very significant. Shaking table tests were conducted for a small scale three-story
structure with and without MR damper installed at the first storey. Experimental setup can be seen
in Figure (1.7). The performance of the response amplifying brace system such as the upper and
scissor-jack toggles were compared to that of general non-amplifying installation system. Current
applied to the damper was 0.1 Amp and 0.2 Amp. It was observed that the brace system with larger
amplification factor showed better control performance. They also mentioned that for the struc-
tures showing non linear seismic behavior the effect of the nonlinearity is expected to become
more evident. Accordingly, for inelastic structures study for investigating the seismic response
control effect of the nonlinear toggle system with MR damper is required.

25
Figure 1.7: Installation configuration of MR damper (Lee et al., 2007)
Raju et al. (2013) have done experimental investigation for the seismic response control of
a scaled three-storey steel frame. The structure had a plan dimension of 1120 mm x 960 mm
and a total height of 2250 mm. Two MR dampers were installed using upper toggle mechanism
(magnification factor of 3.16) at the ground floor. Passive control was used for five different current
values (0, 0.25, 0.5, 0.75 and 1 Ampere) and modified frequencies and damping ratios of structure
was obtained. Numerical simulation was done with MR dampers attached in different storeys to
evaluate the equivalent viscous damping using the formulation of nonlinear viscous fluid dampers
(Remirez et al., 2000). Numerical results showed that first floor damper location contributed
more in increasing effective damping ratio of the model as compared to other floors. Also from
numerical results it was observed that as the current was increased from 0 to 1 Ampere the damping
ratio increased. Two seismic excitations having PGA of 0.2g and 0.4g were used to obtain the
response of frame. From experimental results, it was observed that after a certain current signal
the inter-storey response control was deteriorating. Also when compared to numerical damping
ratios it was observed that same trend was not followed in the variation of experimental damping
ratio with current .

26
1.4 Objectives
The objectives of this study can be listed as follows:
1. To obtain the effectiveness of scissor jack (amplified) braces for passive control of MR
dampers for seismic response reduction of shear building.
2. Use semiactive control using MR dampers for scissor jack (amplified) bracing system and
compare its response with chevron bracing. The algorithm used for control are : (a) LQR/Clipped
optimal, (b) Lyapunov stability, (c) Decentralized bang-bang, and (d) Maximum energy dis-
sipation.
3. To obtain the optimal location of MR damper for a eight storey shear building using Modal
controllability factor.
4. To obtain the the response of eight storey structure using direct LQR algorithm for inter-
storey drift control of MR damper and to obtain its effectiveness by comparing the response
obtained to the clipped optimal control.
1.5 Organization of thesis
The thesis is divided into six chapters. The content of chapters are listed below.
1. The first chapter gives the brief introduction about different methods of vibration control,
toggle based theory for amplified bracings, brief literature review about objectives of this
study.
2. In second chapter, effectiveness of scissor jack (amplified) bracing has been evaluated by
passive response of MR damper for the three- story shear building for seismic control
3. In the third chapter, active control LQR algorithm has been introduced and the controlled
response of a three-storey shear building has been evaluated by varying weight matrices for
different ground motions.

27
4. In the fourth chapter, semiactive control using full state feedback is used to obtain the re-
sponse of both chevron and scissor jack bracing system for a three storey structure with
MR damper at the first floor. Different Algorithms like LQR/clipped optimal, decentral-
ized bang-bang , lyapunov stability ,and maximum energy dissipation have been used for
analysis and their responses have been compared.
5. In fifth chapter, firstly optimal location of MR damper has been obtained for eight storey
structure for seismic response control. Further semi-active direct control algorithm has been
used to control the inter-storey drift of structure and the response obtained is compared to
LQR clipped optimal theory for absolute displacement control.
6. The summary and concluding remarks, scope of future work are mentioned in the fifth
chapter.
Chapter 2
Passive Control of a 3-storey shear
building using Magneto-rheological
(MR) damper
2.1 Brief Overview
In this chapter, the effectiveness of toggle based bracing systems for installing a Magneto-rheological
damper is investigated for the three story shear building for seismic response control. One 1000
kN damper has been installed at the first floor and two cases of bracing system are considered:
chevron and amplified bracing. For both the bracing system, three different voltages including
0 Volt, 2 Volts and 4 Volts are passed through damper, and the controlled response of structure
such as displacement, inter-story drift and acceleration has been compared. Five Ground motions
with different frequency contents have been selected from SAC ground motion database for base
excitation. Ground motions are scaled so that the first floor drift does not exceed the stroke of
1000 kN MR damper.
29
2.2 State-Space formulation of n DOF shear building model
The state-space formulation of shear building model is described in this section for both controlled
and uncontrolled consideration.
2.2.1 Uncontrolled case
The equation of motion for n degree-of-freedom (d.o.f.) building under seismic excitation is given
by
[M ] {ẍ(t)} + [C] {ẋ(t)} + [K] {x(t)} = − [M ] {λ}x¨g (t) (2.1)
where [M ], [K], [C] are, respectively, (n × n) mass, stiffness, and damping matrix of the n storey
building. If mi , ki and ci (i=1, 2, . . . , n) denote, respectively, lumped mass, stiffness, and damping
coefficient of the ith floor of the shear building, then mass, stiffness, and damping matrices are
given by
 
 m1 0 0 
 
 
 0 m 0 
 2 
[M ] =   (2.2)
 .. 

 . 

 
 
0 0 . . . mn
(n×n)
 
 (k1 + k2 ) −k2 
 
 

 −k2 (k2 + k3 ) −k3 

 
 
[K] = 
 ... 

 
 
−kn−1 (kn−1 + kn ) −kn 
 

 
 
−kn kn
(n×n)
30
 
 (c1 + c2 ) −c2 
 
 

 −c2 (c2 + c3 ) −c3 

 
 
[C] = 
 ... 

 
 
−cn−1 (cn−1 + cn ) −cn 
 

 
 
−cn cn
(n×n)
{x(t)}, {ẋ(t)}, and {ẍ(t)} denote, respectively, (n × 1) displacement, velocity, and acceleration
vector of the building, defined as
{x(t)} = { x1 (t) x2 (t) . . . xn (t) }T
{ẋ(t)} = { x˙1 (t) x˙2 (t) . . . x˙n (t) }T
{ẍ(t)} = { x¨1 (t) x¨2 (t) . . . x¨n (t) }T
where, xi (t), ẋi (t), and ẍi (t) denote, respectively, the displacement, velocity and acceleration of
the ith floor relative to ground at any time instant t. Also in Equation 2.1, x¨g (t) is absolute ground
acceleration at time instant t. Absolute acceleration of ith floor is give by (ẍi (t) + x¨g (t)) . {λ} is
a unit vector of size (n × 1).
Vector {ẋ(t)} can also be written as
{ẋ(t)} = [0]n×n {x(t)} + [I]n×n {ẋ} + {0}n×1 x¨g (2.3)
Equation 2.1 can be rewritten as
{ẍ(t)} = − [M ]−1 [K] {x(t)} − [M ]−1 [C] {ẋ(t)} − {λ}x¨g (t) (2.4)
31
Combining Equations 2.3 and 2.4, following state-space form is obtained
{ṡ(t)} = [A]{s(t)} + {E}x¨g (t) (2.5)
where,
{s(t)} is a (2n × 1) state vector given by
 
 
 {x(t)}
 
 T
{s(t)} = = x1 (t) x2 (t) . . . xn (t) x˙1 (t) x˙2 (t) . . . x˙n (t)
 
 {ẋ(t)} 
 
(2n×1)
[A] is (2n × 2n) matrix, also called as plant matrix of the system, is given by
 
 [0] [I] 
[A] = 



− [M ]−1 [K] − [M ]−1 [C]
(2n×2n)
{E} is a (2n × 1) matrix which can be written as
 
 
 {0}
 

{E} =
 
 −{λ} 
 
(2n×1)
Equation 2.1 is a second order equation, whereas, Equation 2.5 is a first order equation. Trans-
formation of second order equation to first order (state-space formulation) simplifies the solution
procedure especially for vibration control problems.
2.2.2 Controlled case
For controlled case, the equation of motion can be written as below
[M ] {ẍ(t)} + [C] {ẋ(t)} + [K] {x(t)} = [γ] {Fc (t)} − [M ] {λ}x¨g (t) (2.6)
32
In the above equation an extra term containing {Fc (t)} is added to represent the actuator control
force at any time t. For r actuators (r ≤ n), {Fc (t)} is a (r × 1) vector. In this equation, [γ] is the
(n × r) location matrix of control forces.
 
 


 Fc1 (t) 



 


 

 
 Fc2 (t)
 

{Fc (t)} = (2.7)
 .. 
.

 


 


 


 

 Fcr (t)
 

(r×1)
Location matrix is given by
[γ] = [γa ](n×n) [γb ](n×r) (2.8)
where,
 
 −1 1 
 
 

 −1 1 

 
 
[γa ] = 
 ... 
 (2.9)
 
 
−1 1 
 

 
 
−1
(n×n)
 
 1 0 ... 0 
 
 

 0 1 ... 0  
 

[γb ] =  .. .. . . ..  (2.10)
 . . . . 

 
 
0 0 ... 1 
 

 
 
0 0 ... 0
(n×r)
In matrix [γb ], for each column total number of elements are same as the number of floors. For
ith column of this matrix, all elements are zero except for the floor location where ith actuator is
placed. Element at that floor location is assigned a value of 1.

33
Vector {ẋ(t)} can also be written as
{ẋ(t)} = [0]n×n {x(t)} + [I]n×n {ẋ} + {0}n×r {Fc (t)} + {0}n×1 x¨g (2.11)
Equation 2.1 can be rewritten as
{ẍ(t)} = − [M ]−1 [K] {x(t)} − [M ]−1 [C] {ẋ(t)} + [M ]−1 [γ] {Fc (t)} − {λ}x¨g (t) (2.12)
Combining Equations 2.11 and 2.12, following state-space form for the controlled case is obtained
{ṡ(t)} = [A]{s(t)} + [B]{Fc (t)} + {E}x¨g (t) (2.13)
where, [B] is a (2n × r) matrix, given as below
 
 [0] 
[B] = 



− [M ]−1 [γ]
(2n×r)
Schematic of Equation 2.13 is shown in Figure 2.1
Figure 2.1: Flowchart for solving state equation without feedback

34
2.3 3-Storey Shear Building Model
A three storey shear building model (Rahbari et al., 2013) is considered with uniform mass and
stiffness distribution for different storey as shown in Figure 2.2. Mass and stiffness matrix of the
Figure 2.2: Three storey shear building
3 d.o.f. system is given by  

 m1 0 0 
 
 
[M ] = 
 0 m2 0


 
 
0 0 m3
where, m1 = m2 = m3 = 3 × 345.6 × 103 kg for the shear building.
 
 k1 + k2 −k2 0 
 
 
[K] = 
 −k2 k2 + k3 −k3 

 
 
0 −k3 k3
35
where, k1 = k2 = k3 = 3.8 × 1.2 × 108 N/m for the shear building.
Thus, structural properties of the system can be described as below.
 
 345.6 0 0 
 
[M ] = 3 × 103 
 
 (kg)
 0 345.6 0 
 
 
0 0 345.6
 
 2.4 −1.2 0 
 
8
 
[K] = 3.8 × 10  −1.2 2.4 −1.2 
 (N/m)
 
 
0 −1.2 1.2
Eigenvalue analysis has been performed in order to determine the natural frequencies of the shear
building model. Table 2.1 shows the natural frequencies of the system.
Damping ratio for the first two modes is 1 % (ζ = 0.01). Damping matrix of the building has been
Table 2.1: Frequencies (in Hz) of the three story structure

Mode Analytical
1 1.485
2 4.162
3 6.014
obtained using Rayleigh damping. Rayleigh coefficients α and β are obtained from the equations
as given below.
C = αM + βK (2.14)
2ω1 ω2
α=ζ
ω1 + ω2
2
β=ζ
ω1 + ω2
36
where, ω1 and ω2 are the frequencies corresponding to first and second mode respectively.
Therefore, the damping matrix can be written as below.
 
 6.5666 −2.5701 0 
 
5
 
 (N − s/m)
[C] = 10  −2.5701 6.5666 −2.5701 
 
 
0 −2.5701 3.9964
In state-space Equations 2.5 and 2.13, {s(t)} becomes is (6 × 1) state vector, plant matrix [A] is
(6 × 6) matrix for 3 d.o.f shear building. [E] is also a (6 × 1) vector. Since single actuator is
installed at the first floor (r = 1), from Equation 2.7, control force matrix becomes a scalar given
by
{Fc (t)} = Fc (t) (2.15)
Also from Equations 2.8, 2.9, and 2.10, γ becomes a (3 x 1) location vector given by
[γ] = [γa ](3×3) [γb ](3×1) (2.16)
     
   
 −1 1 0 1 −1

 
 
 

 
 
 
 

  
   
 

 
[γ] = 
 0 −1 0

 0  = 0  (2.17)
  
  
 
  
   
 

   
0 0 −1  0   0 
   
(3×3) (3×1) (3×1)
Thus, [B] matrix in state Equation 2.13 becomes a (6 × 1) vector.
2.4 Magneto-rheological (MR) damper model
Various phenomenological models have been developed by researchers to capture the highly non
linear behavior of MR fluids. These include mainly Bingham model, Bouc-Wen hysteretic model,
Modified Bouc-Wen model. In the present study Bouc-Wen model has been used.
Bingham model of viscoplasticity proposed by (Stanway et al., 1985, 1987), are often used to
37
define stress-strain relation of MR and ER fluids. Figure 2.3(a) depicts the behavior of bingham
material. Under harmonic loading conditions, force-displacement response of a Bingham material
is as shown in Figure 2.3(b) (Symansa and Constantinou, 1999).
Total shear stress is given by
Figure 2.3: Behavior of Bingham material as described (a) in the stress-strain rate plane and (b) in
the force-displacement plane (Symansa and Constantinou, 1999)
τ = τy sgn(γ) + η γ̇ (2.18)
where, τy is yield stress due to applied magnetic (MR fluid) or electric field (ER fluid), η is the
viscosity of fluid, and γ̇ represents rate of change of shear strain with respect to time. The Bingham
model consists of a coulomb friction element placed in parallel with a linear viscous damper as
shown in the Figure 2.4.
Force generated by damper is given by
F = Fc sgn(ẋ) + c0 ẋ + f0 (2.19)
38
Figure 2.4: Bingham Model (B.F. Spencer Jr, 1996)
where Fc is the frictional force, c0 is the viscous damping coefficient and f0 is the offset in the
force due nonzero mean observed in the measured force due to the presence of the accumulator.
This model captures force displacement response of MR fluid in good agreement, however, it fails
to capture the force velocity behavior mainly in the region where the acceleration and velocity
have opposite signs and the magnitude of the velocities are small. This model is not very suitable
for control analysis.
Anathor model that very well predicts the hysteretic behaviour of MR fluids is Bouc-Wen
model. This is the most commonly used and reputable model to simulate MR dampers semiactive
control system. This model was proposed by B.F. Spencer Jr (1996) to portray the highly non-
linear force-displacement and force velocity behavior of MR dampers. Schematic is shown in
Figure 2.5. Force generated by the damper is given by
F = c0 ẋ + k0 (x − x0 ) + αz (2.20)
where c0 and k0 are the damping coefficient and spring constant for the viscoelastic part and the
variable z, which defines the hysteretic behavior, is given by
ż = −γ |ẋ| z |z|(n−1) − β ẋ |z|n + Aẋ (2.21)

39
c0 , k0 and α are voltage dependent parameters given by
α(u) = αa + αb u (2.22)
k0 (u) = k0a + k0b u (2.23)
c0 (u) = c0a + c0b u (2.24)
where u is the obtained voltage.
Relation between obtained voltage u and input voltage v is given by first order filter.
u̇ = −η(u − v) (2.25)
Bouc-Wen model depicts force displacement as well as force velocity behavior very well, how-
Figure 2.5: Bouc-Wen Model (B.F. Spencer Jr, 1996)
ever, it fails to get accurate values of force-velocity response in the region where the acceleration
and velocity have opposite signs and the magnitude of the velocities are small. To better predict
the response in these regions B.F. Spencer Jr (1996) modified the Bouc-Wen model in which they
combined the Bouc-Wen model with a linear spring dashpot. The schematic is shown in the Figure
2.6.
Force generated by damper for modified Bouc-Wen model is given by

40
Figure 2.6: Modified Bouc-Wen Model (B.F. Spencer Jr, 1996)
F = αz + c0 (ẋ − ẏ) + k0 (x − y) + k1 (x − x0 ) (2.26)
or
F = c1 ẏ + k1 (x − x0 ) (2.27)
and variable z is obtained from following relation
ż = −γ |ẋ − ẏ| z |z|(n−1) − β(ẋ − ẏ) |z|n + A(ẋ − ẏ) (2.28)
and
1
ẏ = {αz + c0 ẋ + k0 (x − y)} (2.29)
c0 + c1
c1 , c0 and α are voltage dependent parameters.
2.5 Bracing Description (Amplified Bracing)
A number of bracing arrangements have been developed to amplify the damper force in a system
and one of them is toggle based arrangement. Toggle mechanism amplifies the damper displace-
ment for a given interstory drift which results in reduced damping force requirement and damper
size. Since force generated in MR dampers is dependent on interstorey response, amplified brac-
ings becomes very effective for damper installation. These systems make use of shallow trusses
41
that amplify the effect of the interstory drift on the damper displacement and also amplify the small
damper force and deliver it to the structural frame (Constantinou et al., 2001; Sigaher and Con-
stantinou, 2004). The expressions obtained to obtain damper displacement and damper force for
upper, lower, reverse and scissor-jack toggle system are explained briefly in the following section.
2.5.1 Upper Toggle System
For upper toggle system, if u is the inter-storey drift towards right, damper displacement is given
by
Figure 2.7: Movement of upper toggle system (Constantinou et al., 2001)
uD = |A0 B 0 − AB| (2.30)

 " #1/2 
2
h l1
uD = ±  − l1 tan θ1 − h tan θ1 − u − + l1 cos(θ1 ± φ) + (h − l1 sin(θ1 ± φ))2 
cos θ1 cos θ1
(2.31)
Positive sign is used for drift towards right and negative sign holds for drift towards left. Since
the lengths l1 and l2 of braces are not changing in initial and final configuration, the following
equation had been obtained by conservation of length
l2 2 = h2 + l1 2 + (l + u)2 − 2hl1 sin(θ1 ± φ) − 2(l + u)l1 cos(θ1 ± φ) (2.32)
These are highly nonlinear complex equations. Since rotation φ is very small and drift is low as
compared to the dimensions of structure, above equations can be simplified by ignoring higher
42
order terms as
uD = a0 u (2.33)
Force transferred to the structure is given by
F = a0 FD (2.34)
where, FD is the force produced in damper and a0 is the amplification factor. For upper toggle,
system amplification factor is given by
sin θ2
a0 = + sin θ1 (2.35)
cos(θ1 + θ2 )
Figure 2.8: Dependency of magnification factor on geometry (Constantinou et al., 2001)
2.5.2 Lower Toggle System
Figure 2.9: Movement of lower toggle system (Constantinou et al., 2001)

43
As obtained by Constantinou et al. (2001), for lower toggle system damper displacement is
given by
" 1/2 #
1 2 cos(θ1 ± φ)
uD = ±l1 1+ − − tan θ1 (2.36)
cos2 θ1 cos θ1
Similar to upper toggle system, Equations 2.33 and 2.34 are also valid for obtaining relative dis-
placement between the two ends of damper and for force computation. Amplification factor for
lower toggle is given by

sin θ2
a0 = (2.37)
cos(θ1 + θ2 )
Figure 2.10: dependency of magnification factor on geometry (Constantinou et al., 2001)
2.5.3 Reverse Toggle System
Figure 2.11: Movement of reverse toggle system (Constantinou et al., 2001)

44
Figure 2.12: dependency of magnification factor on geometry (Constantinou et al., 2001)
For reverse toggle system damper displacement can be obtained as
uD = ±[αl2 tan θ2 − {αl2 sin(θ2 ∓ φ) + u]2 + [h − αl2 cos(θ2 ∓ φ)]2 }1/2 ] (2.38)
and
l1 2 = h2 + l2 2 + (l − u)2 − 2hl2 cos(θ2 ∓ φ) − 2(l − u)l2 sin(θ2 ∓ φ) (2.39)
Amplification factor a0 for reverse toggle bracing is given by
αcosθ1
a0 = − cosθ2 (2.40)
cos(θ1 + θ2 )
2.5.4 Scissor Jack Toggle System
Other bracing configurations occupy a lot of open space. In addition to amplifying damper force,
the scissor jack damper system due to its compact geometry gives access to open space and is
architecturally more desirable.

45
Figure 2.13: Movement of Scissor Jack toggle system (Sigaher and Constantinou, 2003)
Damper displacement is expressed as
uD = |A0 B 0 − AB| = ±2l1 [sin(θ ± ∆θ) − sin θ] (2.41)
By preservation of length of braces in initial configuration (CD) and final configuration (C’D)
following equation had been obtained
2l1 cos(θ ± ∆θ) = 2l1 cos θ ∓ u cos ψ (2.42)
From Equations 2.41 and 2.42, damper displacement uD and rotation angle ∆θ can be obtained as

−1 cos ψ
uD = ±l1 sin cos cos θ ∓ u − sin θ (2.43)
2l1

−1 cos ψ
± ∆θ = cos cos θ ∓ u −θ (2.44)
2l1
46
For practical applications, inter-storey drift is small as compared to dimensions and also ∆θ is very
small. Simplifying above equations, amplification factor for scissor jack bracing can be obtained
as
cos ψ
a0 = (2.45)
tan θ
Figure 2.14: dependency of magnification factor on geometry (Sigaher and Constantinou, 2003)
From the variation of magnification factor with geometry as shown in Figures 2.8, 2.10, 2.12
and 2.14, it can be seen that very high value of magnification factor can be achieved by changing
the geometry of bracing system. However, practical values of magnification factor lies in the range
of 2 to 5. Practical values are those values for which slight change in geometry does not cause
very large variation in magnification factor. Magnification factor should not be very sensitive to
geometry change.
In the present study, two types of bracing arrangements have been used and their responses
have been compared. One is chevron (unamplified) bracing while the other is scissor-jack (ampli-
fied) bracing. For chevron bracing, force produced in damper is equal to the force transferred to
the structure, so amplification factor becomes unity. For scissor-jack bracing, amplification fac-
tor of 2 is considered. If u and a0 are interstory drift and amplification factor respectively, then
damper relative displacement is given by
uD = a0 u (2.46)
47
If FD is the force induced in the damper, then force transmitted to structure is given by,
F = a0 FD (2.47)
Amplification factor for chevron bracing (F = FD ) is given by
Figure 2.15: Chevron Bracing Arrangement
Figure 2.16: Scissor-Jack Bracing Arrangement
a0 = 1 (2.48)
Amplification factor for scissor jack bracing is given as below
cos ψ
a0 = =2 (2.49)
tan θ
48
2.6 Equation of motion for Passive Case of MR damper
MR damper has been installed at the first floor using chevron (unamplified) and scissor jack (am-
plified) bracing. Bouc-Wen model has been used to obtain the force obtained by damper. Inter-
storey drift at the first floor at any time instant is given by x1 (t). From Equation 2.46, damper end
displacement is given by
x(t) = a0 x1 (t) (2.50)
Since, amplification factor is not very high and drift is small, it assumed to be time invariant. By
differentiating above equation with time, following expression is obtained.
ẋ(t) = a0 x˙1 (t) (2.51)
The force generated in damper becomes
FD (t) = a0 c0 x˙1 (t) + a0 k0 x1 (t) − k0 x0 + αz(t) (2.52)
z is the hysteric component given by
ż(t) = a0 (−γ |x˙1 (t)| z(t) |z(t)|(n−1) − β x˙1 (t) |z(t)|n + Ax˙1 (t)) (2.53)
From Equation 2.47, damping force transferred to the structure due to amplification factor a0 is
obtained as
F (t) = a0 (a0 c0 x˙1 (t) + a0 k0 x1 (t) − k0 x0 + αz(t)) (2.54)
Damper force is a function of displacement and velocity of structure. So, damper force equation
and equation of motion of structure have to be solved simultaneously, to obtain structure response
at each time instant. The equation of motion for passive case for a three storey structure with
damper located at the first floor becomes

49
 
 [A]{s(t)} + {B}F (t) + {E}x¨g (t) 
{Ṡ(t)} = 



a0 (−γ |x˙1 (t)| z(t) |z(t)|(n−1) − β x˙1 (t) |z(t)|n + Ax˙1 (t))
where, the state vector {S(t)} becomes
T T
{S(t)} = s(t) z(t) = x1 (t) x2 (t) x3 (t) x˙1 (t) x˙2 (t) x˙3 (t) z(t)
The voltage dependent parameters used in Equations 2.22–2.24 are constant for passive cases. One
1000 kN MR damper has been installed at the first floor of the structure. All the eleven parameters
of Bouc-Wen model for 1000 kN damper are listed in Table 2.2 (Jung et al., 2003a). Stroke of
damper is ±0.08 m and its saturation voltage is 10 V.
Table 2.2: Parameters of Bouc-Wen Model of 1000 kN MR damper

Parameters Values
αa 26.0 kN
αb 29.1 kN/m/V
k0a 0
k0b 0
c0a 105.4 kN s/m
c0b 131.6 kN s/m/V
γ 141.1 m-2
β 141.1 m-2
A 2074.5
n 2
η 100 s-1
2.7 Ground motions
Five ground motions have been selected from SAC ground motion database. SAC ground motion
consists of 60 ground motions. Among these ground motion, 20 represents a hazard level of 2%
in 50 years, 20 represents a hazard level of 10% in 50 years ,and remaining 20 represents a hazard
level of 50% in 50 years. These ground motions include both recorded as well as simulated ones
and scaled to match response spectrum of a particular hazard level. They were mainly developed
for the analysis of steel moment resisting frames in Los Angeles area, USA. These motions cover
50
a broad broad range of peak ground accelerations (PGA) between 0.11g and 1.33 g, peak ground
velocity (PGV) between 21.67 cm/sec and 245.41 cm, and peak ground displacements (PGD)
between 5.4 cm and 93.43 cm in addition to a wide band of frequency content and a wide range
of strong motion duration. Some of them even possess strong near fault pulse. The five ground
motions selected are based on their different frequency contents. Also the ground motions are
chosen from all the hazard levels. Scaling is done so that the controlled interstorey response of the
floor having damper did not exceed the stroke of 1000 kN damper. Stroke of 1000 kN damper is ±8
cm, so the controlled drift of the first floor of structure should not exceed ±( a80 ) cm. For bracing
having amplification factor of 2, the controlled drift of the first floor of structure for amplified case
should not be more than ±4 cm. The details of the scaled five ground motion selected from SAC
ground motion database is given in Table 2.3.
2.8 Results and Discussions
Maximum displacement, inter-storey drift and acceleration for each floor for all the passive cases
along with the maximum damping force obtained in structure, are given in Tables 2.4–2.8. Values
in brackets show the percentage reduction in response as compared to uncontrolled case. Con-
trolled response for different voltages and amplification factors can be compared from Figures
4.2–4.6 . From these data a reduction in response can be seen for all the controlled cases.
From Table 2.4 and Figure 4.2, for the case of GM1 it can be observed that 4 Volts-Amp2
case is providing better control as compared to all the other case due to maximum control force
obtained in this case (high voltage and amplified bracing arrangement). 35% reduction in peak
displacement and around 32% reduction in both drift and acceleration response of top floor is
observed. For 2volts- Amp2 case, maximum control force is higher as compared to 4Volts-Amp1,
resulting in a better control performance. For 4Volts-Amp1 case, percentage reduction is 30% for
top floor peak displacement and 27% for drift and acceleration, due to maximum control force
of 574 kN. However, for 2volts- Amp2, percentage reduction has increased to 32% for top floor
Table 2.3: Ground Motion Used
GM SAC Record Component Earthquake Distance Scale PGA Freq
code Name Magnitude (km) (g) (Hz)
GM1 LA02 Imperial Valley, 1940, El Centro, Irr dist Fault Parallel 6.9 10 0.32 0.21632 1.4648
GM2 LA19 North Palm Springs, 1986 Fault Normal 6 6.7 0.4 0.408 6.4697
GM3 LA40 Palos Verdes (simulated) Fault Parallel 7.1 1.5 0.95 0.5947 0.5127
GM4 LA49 Morgan Hill, 1984, Gilroy Fault Normal 6.2 15 0.7 0.2233 1.8311
GM5 LA57 San Fernando, 1971, Hollywood Storage Bldg Fault Normal 6.5 1 1 0.253 4.2114
51
52
peak displacement and 30% for drift and acceleration, due to maximum control force of 887 kN.
From the table, it can also be observed that for amplified cases control force is becoming around
2.5 times as compared to unamplified case for all the three voltages resulting in a better control
performance.
Similar trend can be observed for other ground motions also. 4 Volts-Amp2 case is providing
better control performance as compared to all the other cases, providing 20–44% reduction in
top floor peak displacement and 30–50% reduction in maximum drift and acceleration of the top
floor. Even for other ground motions, performance of 2volts- Amp2 case is better as compared to
4Volts-Amp1 case, as the control force is amplified in the prior case to a much higher value.
Figure 2.17: GM1 : (a) Displacement, (b) Inter-story drift and (c) Acceleration response
Table 2.4: Response of building when excited to GM1
Control Floor Displacement Interstory Accleration Maximum Control
Strategy Number (cm) drift (cm) (g) Force (kN)
First 4.128 4.128 0.489
Uncontrolled Second 7.218 3.341 0.692
Roof 9.178 2.037 0.912
First 3.733 (09.57) 3.733 (09.57) 0.456 (06.75)
0 Volt-Amp1 Second 6.503 (09.91) 2.996 (10.33) 0.635 (08.24) 111.39
Roof 8.197 (10.69) 1.821 (10.60) 0.816 (10.53)
First 3.265 (20.91) 3.265 (20.91) 0.448 (08.38)
0 Volt-Amp2 Second 5.689 (21.18) 2.547 (23.77) 0.557 (19.51) 277.47
Roof 7.158 (22.01) 1.558 (23.51) 0.697 (23.57)
First 3.131 (24.15) 3.131 (24.15) 0.448 (08.38)
2 Volts-Amp1 Second 5.453 (24.45) 2.437 (27.06) 0.543 (21.53) 342.67
Roof 6.826 (25.63) 1.538 (24.50) 0.688 (24.56)
First 2.627 (36.36) 2.627 (36.36) 0.456 (06.75)
2 Volts-Amp2 Second 4.808 (33.39) 2.332 (30.20) 0.497 (28.18) 887.01
Roof 6.196 (32.49) 1.441 (29.26) 0.644 (29.39)
First 2.771 (32.87) 2.771 (32.87) 0.455 (06.95)
4 Volts-Amp1 Second 4.953 (31.38) 2.413 (27.78) 0.495 (28.47) 573.98
Roof 6.377 (30.52) 1.488 (26.95) 0.665 (27.08)
First 2.453 (40.58) 2.453 (40.58) 0.458 (06.34)
4 Volts-Amp2 Second 4.683 (35.12) 2.254 (32.54) 0.515 (25.58) 1466.57
Roof 5.938 (35.30) 1.390 (31.76) 0.621 (31.91)
53
54

First 3.977 3.977 0.670
Roof 7.640 1.919 0.860
First 3.736 (06.06) 3.736 (06.06) 0.627 (06.42)
0 Volt-Amp1 Second 6.223 (04.88) 2.630 (03.27) 0.653 (10.79) 112.70
Roof 7.374 (03.48) 1.780 (07.24) 0.797 (07.33)
First 3.409 (14.28) 3.409 (14.28) 0.538 (19.70)
0 Volt-Amp2 Second 5.774 (11.74) 2.494 (08.28) 0.585 (20.08) 291.79
Roof 6.984 (08.59) 1.562 (18.60) 0.698 (18.84)
First 3.313 (16.70) 3.313 (16.70) 0.544 (18.81)
2 Volts-Amp1 Second 5.559 (15.03) 2.419 (11.03) 0.564 (22.95) 355.87
Roof 6.762 (11.49) 1.522 (20.69) 0.680 (20.93)
First 2.495 (37.26) 2.495 (37.26) 0.416 (37.91)
2 Volts-Amp2 Second 4.379 (33.06) 2.025 (25.52) 0.452 (38.25) 855.17
Roof 5.553 (27.32) 1.383 (27.93) 0.618 (28.14)
First 2.940 (26.07) 2.940 (26.07) 0.503 (24.93)
4 Volts-Amp1 Second 4.976 (23.94) 2.238 (17.69) 0.509 (30.46) 578.46
Roof 6.165 (19.31) 1.468 (23.50) 0.657 (23.60)
First 1.774 (55.39) 1.774 (55.39) 0.365 (45.52)
4 Volts-Amp2 Second 3.258 (50.20) 1.687 (37.96) 0.368 (49.73) 1382.34
Roof 4.353 (43.02) 1.334 (30.48) 0.598 (30.47)
First 4.026 4.026 0.647
Roof 9.887 2.283 1.022
First 3.998 (00.70) 3.998 (00.70) 0.648 -(00.15)
0 Volt-Amp1 Second 7.415 (02.49) 3.494 (03.03) 0.653 (00.00) 117.88
Roof 9.602 (02.88) 2.187 (04.20) 0.979 (04.21)
First 3.964 (01.54) 3.964 (01.54) 0.640 (01.08)
0 Volt-Amp2 Second 7.176 (05.63) 3.353 (06.94) 0.650 (00.46) 319.14
Roof 9.232 (06.62) 2.056 (09.94) 0.921 (09.88)
First 3.940 (02.14) 3.940 (02.14) 0.639 (01.24)
2 Volts-Amp1 Second 7.046 (07.34) 3.282 (08.91) 0.644 (01.38) 379.30
Roof 9.046 (08.51) 2.001 (12.35) 0.896 (12.33)
First 3.866 (03.97) 3.866 (03.97) 0.616 (04.79)
2 Volts-Amp2 Second 6.501 (14.51) 2.931 (18.65) 0.632 (03.22) 971.44
Roof 8.023 (18.85) 1.687 (26.11) 0.756 (26.03)
First 3.895 (03.25) 3.895 (03.25) 0.629 (02.78)
4 Volts-Amp1 Second 6.702 (11.86) 3.110 (13.68) 0.641 (01.84) 624.64
Roof 8.545 (13.57) 1.850 (18.97) 0.828 (18.98)
First 3.767 (06.43) 3.767 (06.43) 0.601 (07.11)
4 Volts-Amp2 Second 6.410 (15.70) 2.768 (23.18) 0.630 (03.52) 1527.52
Roof 7.823 (20.88) 1.492 (34.65) 0.667 (34.74)
55
56

First 4.184 4.184 0.540
Roof 9.227 1.992 0.892
First 3.966 (05.21) 3.966 (05.21) 0.495 (08.33)
0 Volt-Amp1 Second 7.115 (04.70) 3.150 (03.99) 0.645 (05.98) 113.84
Roof 8.282 (10.24) 1.889 (05.17) 0.846 (05.16)
First 3.661 (12.50) 3.661 (12.50) 0.426 (21.11)
0 Volt-Amp2 Second 6.503 (12.90) 2.895 (11.76) 0.590 (13.99) 301.10
Roof 8.101 (12.20) 1.748 (12.25) 0.781 (12.44)
First 3.626 (13.34) 3.626 (13.34) 0.434 (19.63)
2 Volts-Amp1 Second 6.350 (14.95) 2.823 (13.96) 0.592 (13.70) 366.18
Roof 7.874 (14.66) 1.720 (13.65) 0.768 (13.90)
First 3.082 (26.34) 3.082 (26.34) 0.386 (28.52)
2 Volts-Amp2 Second 5.559 (25.54) 2.488 (24.17) 0.531 (22.59) 915.02
Roof 6.893 (25.30) 1.462 (26.61) 0.654 (26.68)
First 3.420 (18.26) 3.420 (18.26) 0.425 (21.30)
4 Volts-Amp1 Second 6.075 (18.63) 2.655 (19.08) 0.574 (16.33) 605.03
Roof 7.459 (19.16) 1.590 (20.18) 0.712 (20.18)
First 2.712 (35.18) 2.712 (35.18) 0.366 (32.22)
4 Volts-Amp2 Second 4.956 (33.62) 2.245 (31.58) 0.493 (28.13) 1424.75
Roof 6.148 (33.37) 1.292 (35.14) 0.578 (35.20)
First 3.964 3.964 0.816
Roof 7.549 2.157 0.968
First 3.705 (06.53) 3.705 (06.53) 0.740 (09.31)
0 Volt-Amp1 Second 6.066 (05.56) 2.707 (05.42) 0.742 (07.37) 112.77
Roof 6.928 (08.23) 2.017 (06.49) 0.905 (06.51)
First 3.313 (16.42) 3.313 (16.42) 0.607 (25.61)
0 Volt-Amp2 Second 5.535 (13.83) 2.548 (10.97) 0.652 (18.60) 290.91
Roof 6.374 (15.56) 1.827 (15.30) 0.821 (15.19)
First 3.130 (21.04) 3.130 (21.04) 0.590 (27.70)
2 Volts-Amp1 Second 5.205 (18.96) 2.484 (13.21) 0.615 (23.22) 350.30
Roof 6.128 (18.82) 1.727 (19.94) 0.775 (19.94)
First 2.198 (44.55) 2.198 (44.55) 0.385 (52.82)
2 Volts-Amp2 Second 3.864 (39.84) 2.050 (28.37) 0.437 (45.44) 842.88
Roof 5.166 (31.57) 1.351 (37.37) 0.607 (37.29)
First 2.613 (34.08) 2.613 (34.08) 0.488 (40.20)
4 Volts-Amp1 Second 4.417 (31.23) 2.278 (20.41) 0.526 (34.33) 566.62
Roof 5.639 (25.30) 1.489 (30.97) 0.669 (30.89)
First 1.737 (56.18) 1.737 (56.18) 0.348 (57.35)
4 Volts-Amp2 Second 3.155 (50.88) 1.694 (40.81) 0.383 (52.18) 1293.34
Roof 4.199 (44.38) 1.077 (50.07) 0.484 (50.00)
57
58
59
2.9 Summary
A numerical study of passive control case has been conducted for a three storey shear building.
The performance of the scissor jack (amplified) brace system for three control voltages is evaluated
and compared with chevron (unamplified) brace system. The study revealed that larger effective
60
control force is generated for scissor jack case due to its response amplification mechanism result-
ing in a better control performance. Also at lower current, better performance can be observed for
amplified case (scissor jack bracing), than higher current in unamplified case (chevron bracing).
Hence, reduced voltage or a lower capacity damper can be used with scissor jack bracing system,
which results in cost effectiveness.

Chapter 3
Control of a 3-storey shear building
using LQR algorithm
3.1 Brief Overview
In this chapter, Linear Quadratic Regulator (LQR) control is used for seismic response control
of a three storey shear building. Full state feedback is used to obtain the maximum controller
force required at the first floor for LQR algorithm. Effect of different weighing matrices has been
evaluated on response reduction and control force. Five scaled ground motions for SAC ground
motion have been used for structure excitation
3.2 Active Control using LQR algorithm (Full State feedback)
As described in previous chapter, for n d.o.f. structure installed with r actuators/dampers, state
space equation is given by
{ṡ(t)}(2n×1) = [A](2n×2n) {s(t)}(2n×1) + [B](2n×r) {Fc (t)}(r×1) + {E}(2n×1) x¨g (t) (3.1)
62
Also for 3 d.o.f. structure in which control force is applied at the first floor level, the above
equation becomes
{ṡ(t)}(6×1) = [A](6×6) {s(t)}(6×1) + {B}(6×1) Fc (t) + {E}(6×1) x¨g (t) (3.2)
In passive control using MR dampers, when the current is constant, the damper forces are known
at each time instant. However, for active cases control forces have to be obtained using feedback
control laws and then they are applied to the structure. Some of feedback laws are described below.
3.2.1 Feedback Control laws
The Equation 3.1 has (2n + r) unknown variables, 2n state variables and r control forces. Total
number of equations are only 2n, so it cannot be solved directly, r more equations are required to
obtain solution for controlled response of structure. These r equations are obtained by feedback
control laws. There are three active control laws that are implemented for seismic response control
of smart structures (Cheng, 1988). These are Open-loop feedback control, Closed-loop feedback
control and Open-Closed-loop feedback control. A brief description of all these schemes is given
below.
3.2.1.1 Open-loop feedback control
In open loop feedback control, control forces are determined by the feedback of external excitation
(earthquake ground motion). Response of structure is not required for obtaining control forces.
An accelerometer is placed at the ground floor to obtain the excitation data. The information from
accelerometer sensor goes to control computer in which some active control algorithm is used to
obtain the control force, which is then applied to the structure. Only one sensor is required in
this control which makes it a simple scheme for implementation, however, control gain required
to obtain control force cannot be obtained until the earthquake excitation data is known a priori
for the entire control duration. This makes the algorithm seldom for seismic response control of
63
structures. The schematic is shown in Figure 3.1.
Figure 3.1: Schematic of open-loop feedback control
3.2.1.2 Closed-loop feedback control
In closed loop control, control forces are determined from the measured controlled responses of
the structure. The response of the structure is represented by state variables, which include floor
displacements and velocities of structure. For full state feedback, measurement of all the states
is required. Hence 2n sensors are required, n for displacement measurement and n for velocity
measurement of all the floors. This data from the sensors is used by control computer to calculate
active control forces based on some control algorithm. Obtained control forces are then applied to
the structure. Number of sensors required for state response measurement can be greatly reduced
by the use of state estimators (observers/Kalman filter) or analog differentiators. Some researchers
also proposed direct acceleration feedback (Dyke et al., 1996d; Spencer et al., 1993) in which
instead of displacement and velocity, acceleration response is required for feedback. Schematic is
shown in Figure 3.2

64
Figure 3.2: Schematic of Closed-loop feedback control
3.2.1.3 Open-Closed-loop feedback control
This is combination of Open-loop feedback and Closed-loop feedback control. It requires mea-
surement of earthquake excitation as well as displacements and velocities of the structure. The
information from the sensors is received by control computer to calculate active control forces
using control laws, which are then applied to the structure. This approach is considered to be
superior to both of the above mentioned control system. However, it cannot be used for seismic
response control, due to the same problem as mentioned in Open-loop feed back control, to obtain
control gain. Schematic is shown in Figure 3.3
Figure 3.3: Schematic of Open-Closed-loop feedback control

65
Closed-loop feedback control is the most commonly used control scheme for the seismic re-
sponse control of smart structures. The other two schemes are not feasible for control due to
earthquake loading. In the present study, full state feedback control have been used. The control
force obtained is a linear combination of all the states (displacements and velocities of all the
floors) of the system. The control law for closed-loop feedback is given by.
{Fc (t)}(r×1) = −[G](r×2n) {s(t)}(2n×1) (3.3)
Schematic is shown in Figure 3.4.
In the above equation [G] is (r × 2n) feedback gain matrix. If [G] is known, Equation 3.3 gives r
Figure 3.4: Flowchart for solving state equation with state feedback
extra equation, Equation 3.1 can be solved to obtain the structure response. Substituting feedback
control force {Fc (t)} from Equation 3.3 to 3.1, following control equation is obtained
{ṡ(t)} = ([A] − [B][G]){s(t)} + {E}x¨g (t) (3.4)
or
{ṡ(t)} = [Ac ]{s(t)} + {E}x¨g (t) (3.5)

66
where,
[Ac ] = [A] − [B][G] (3.6)
[Ac ] is called as closed-loop plant matrix of the system.
Gain matrix is designed in a way that the system behave in a desired manner. Linear Quadratic
Regulator (LQR) control is a very popular control strategy, by which [G] matrix is designed by
optimizing performance index consisting of response and control force. LQR control is explained
in detail in the next section. However, for the design of optimal control algorithm, system should
satisfy the criteria of Controllability and Observability. Brief description is given in the next
section.
3.2.2 Controllability and Observability
In this section the concepts of state controllability and observability are discussed in brief. Also
the criteria for a system to be controllable and observable are mentioned.
3.2.2.1 State Controllability
For a linear system, if there exists an input u(t) that drives the system from initial state x(t0 ) at
t = t0 to any desired final state x(tf ) at t = tf in a finite time interval (tf − t0 ),the the system is
said to be state controllable. If this holds for all the initial states and for all t0 , then the system is
said to be completely state controllable.
Controllability matrix for n d.o.f. system is given by
C = [[B]|[A][B]|[A]2 [B]| . . . |[A]n−1 [B]] (3.7)
System is said to be full state controllable if C has rank n.

67
3.2.2.2 Observability
Output of the system is given by
{y(t)} = [C]{x(t)} (3.8)
Suppose for n d.o.f. system q states are available for measurement (q ≤ r). Then y(t) becomes a
(r x 1) vector having measured states of the system. x(t) is full state vector of size (2n × 1). [C]
is constant matrix of size (r x 2n) relating output vector to state vector.
If the state of a system x(t0 ) at t = t0 can be determined from output y(t), t0 ≤ t ≤ tf , system is
said to be observable. If all the initial states are observable for all t0 , system is called as completely
observable. Observability matrix for a n d.o.f. system is given by
O = [[C]T |[A]T [C]T |([A]T )2 [C]T | . . . |([A]T )n−1 [C]] (3.9)
System is said to be completely observable if O has rank n.
Systems, in which all states are not available for measurement, should satisfy the observability
criteria. Also observer/estimator design is required to estimate the state variables that cannot
be measured. In the present case, full state feedback have been used for three storey structure,
observability check and observer/estimator design is not required.
3.2.3 Ricatti Optimal Active Control (ROAC)/Linear Quadratic Regulator (LQR)
Algorithm
If a system is completely controllable and observable, desired optimal controller force is obtained
by minimizing cost function given by
Z tf
1
J= [{s(t)}T [Q]{s(t)} + {Fc (t)}T [R]{Fc (t)}]dt (3.10)
2 0
68
[Q] and [R] are called weighing matrices. [Q] is a (2n × 2n) symmetric positive semi definite
matrix. [R] is a (r × r) positive-definite symmetric matrix. [Q] give the weight assigned to the
state of system while [R] gives the weight assigned to the control forces. By decreasing the value
of [R], {Fc (t)} can be increased and {s(t)} is reduced. Similarly by decreasing [Q], {s(t)} is
increased and Fc (t) gets reduced.
Solution of Equation 3.10 is obtained by defining a Hamiltonian, given by
1
H = ([{s(t)}T [Q]{s(t)} + Fc (t)T [R]{Fc (t)}) + {λ(t)}T ([A]{s(t)} + [B]{Fc (t)} − {ṡ(t)})
2
(3.11)
In LQR algorithm, earthquake excitation is not used in determining control force. In the above
equation, {λ(t)} is a (2n × 1) vector of Lagrange multipliers. Following Euler equations should
be satisfied for optimality.

∂H d ∂H
− = {0} (3.12)
∂{s(t)} dt ∂{ṡ(t)}

∂H d ∂H
− = {0} (3.13)
∂{Fc (t)} dt ∂{Ḟc (t)}

∂H d ∂H
− = {0} (3.14)
∂{λ(t)} dt ∂{λ̇(t)}
Substituting H from Equation 3.11 in Equations 3.12–3.14, following expressions are obtained
[Q]{s(t)} + [A]T {λ(t)} + {λ̇(t)} = {0} (3.15)
[R]{Fc (t)} + [B]T {λ(t)} = {0} (3.16)
{ṡ(t)} = [A]{s(t)} + [B]{Fc (t)} (3.17)

69
Considering {λ(t)} = [P (t)]{s(t)}, Equations 3.15–3.17 gives the following Matrix Riccati
Equation (MRE).
[Ṗ (t)] + [P (t)][A] + [A]T [P (t)] − [P (t)][B][R]−1 [B]T [P (t)] + [Q] = [0] (3.18)
Where [P (t)] is a Ricatti matrix. For a stable system, Ricatti matrix is constant for entire duration
of earthquake and finally rapidly reaches to zero, when close to tf . Hence [P (t)] and [Ṗ (t)] can
be approximated to [P] and [0] respectively. Equation 3.18, becomes time invariant and can be
re-written as
[P ][A] + [A]T [P ] − [P ][B][R]−1 [B]T [P ] + [Q] = [0] (3.19)
The above equation is referred as Algebraic Ricatti Equation (ARE), solution of which gives Ri-
catti matrix [P ]. Also from Equation 3.16, control force is obtained as
{Fc (t)} = −[R]−1 [B]T [P ]{s(t)} = −[G]{s(t)} (3.20)
Control Gain matrix is given by
[G] = [R]−1 [B]T [P ]{s(t)} (3.21)
In matlab environment, [G] can be obtained by using lqr function
[G] = lqr([A], [B], [Q], [R]) (3.22)
3.2.4 Generalized Optimal Active Control (GOAC) Algorithm
Anathor technique is used by researchers called Generalized Optimal Active Control (GOAC)
Algorithm for seismic response control of structures. Unlike in LQR algorithm, earthquake exci-
tation is used in GOAC algorithm for deriving control force. In this approach total time duration
70
[t0 , tf ] is divided into n segments and state of the system is minimized at the end of each time
step. The cost function is obtained as
Z ti
1 1
Ji = {s(ti )}T [S]{s(ti )} + [{s(t)}T [Q]{s(t)} + Fc (t)T [R]{Fc (t)}]dt (3.23)
2 2 ti−1
Control force for GOAC algorithm is given by
{Fc (t)} = −[R]−1 [B]T [S]{s(t)} = −[G]{s(t)} (3.24)
Control Gain matrix is given by
[G] = [R]−1 [B]T [S]{s(t)} (3.25)
If [S] is chosen to be Ricatti marix [P ], then GOAC algorithm becomes same as LQR or ROAC
algorithm. Thus LQR is a special case of GOAC algorithm.
For the analysis of 3-storey structure, LQR method has been used and [Q] matrix is chosen in
a way so that equal weight is assigned to all the states of the system. [Q] matrix is given by
 
 1 0 0 0 0 0 
 
 
 
 0 1 0 0 0 0 
 
 
 
 0 0 1 0 0 0 
Q=
 

 

 0 0 0 1 0 0 

 
 

 0 0 0 0 1 0 

 
 
0 0 0 0 0 1
Since control force is applied only at the first floor of structure, [R] matrix becomes a scalar R.
Six different values of R are used to obtain the optimum response and corresponding maximum
control force for different ground motions. Different R values used are listed in Table 4.1
71
Table 3.1: Different R values

R
Case 1 10-12
Case 2 5x10-13
Case 3 10-13
Case 4 5x10-14
Case 5 10-14
Case 6 10-15
Five Ground motions from SAC motion database, that are already mentioned in Chapter 2, have
been used for LQR active control. Maximum acceleration, inter-storey and displacement response
of the building for no control case is shown in Table 3.2. Maximum acceleration, inter-storey
Table 3.2: Uncontrolled Response of structure
Floor Displcement Inter-storey Acceleration

(cm) Drift (cm) (g)
First 4.128 4.128 0.489
GM1 Second 7.218 3.341 0.692
Third 9.178 2.037 0.912
First 3.977 3.977 0.670
GM2 Second 6.542 2.719 0.723
Third 7.640 1.919 0.860
First 4.026 4.026 0.647
GM3 Second 7.604 3.603 0.653
Third 9.887 2.283 1.022
First 4.184 4.184 0.540
GM4 Second 7.466 3.281 0.686
Third 9.227 1.992 0.892
First 3.964 3.964 0.816
GM5 Second 6.423 2.862 0.801
Third 7.549 2.157 0.968
drifts and displacement response of all the floors of the building model for the controlled cases
and maximum control force for different R values are shown in Tables 4.2–4.6 for all the ground
motion data. From all these tables it can be observed that by reducing weight factor R, control
force is increasing and better control can be achieved. Variation of maximum response of the
structure by changing R values is also shown in plots.
From Figure 3.5 and Table 4.2 it can be observed that, for Case 1 of GM1 data, a high reduction
72
Table 3.3: Maximum displacement and acceleration response for GM1
Floor Case 1 Case 2 Case 3 Case 4 Case 5 Case 6

First 3.123 2.825 2.195 2.069 1.696 1.257
Displacemnt Second 5.403 4.898 4.278 4.008 3.162 1.913
(cm) Third 6.785 6.192 5.609 5.234 4.184 2.649
First 3.123 2.825 2.195 2.069 1.696 1.257
Inter-storey Second 2.449 2.316 2.128 2.008 1.677 1.253
Drift (cm) Third 1.517 1.459 1.331 1.246 1.023 0.736
First 0.425 0.413 0.370 0.344 0.290 0.257
Acceleration Second 0.535 0.492 0.394 0.376 0.329 0.260
(g) Third 0.678 0.659 0.596 0.557 0.458 0.330
Control
force(kN) 493.24 731.12 1717.43 2342.77 4240.97 6601.27

First 3.292 2.950 1.977 1.547 1.322 1.424
(cm) Third 6.634 6.082 4.362 4.095 3.708 3.149
First 3.292 2.95 1.977 1.547 1.322 1.424
Drift (cm) Third 1.593 1.42 1.3 1.268 1.219 1.224
First 0.518 0.481 0.384 0.331 0.316 0.390
Accelration Second 0.577 0.518 0.375 0.356 0.374 0.356
(g) Third 0.712 0.634 0.581 0.570 0.549 0.551
Control
force(kN) 490.17 754.92 1599.44 2131.38 4397.86 7701.23

First 3.990 3.970 3.911 3.894 4.004 5.413
(cm) Third 8.964 8.451 7.985 7.963 8.095 9.327
First 3.990 3.970 3.911 3.894 4.004 5.413
Drift (cm) Third 2.004 1.867 1.512 1.485 1.443 1.489
First 0.636 0.630 0.617 0.609 0.594 0.596
(g) Third 0.900 0.839 0.677 0.664 0.647 0.668
Control
force(kN) 671.46 1048.39 2367.58 3135.04 6014.21 16005.89
73

First 3.472 3.193 2.451 2.137 1.563 1.296
(cm) Third 7.452 6.954 5.526 4.820 3.636 2.409
First 3.472 3.193 2.451 2.137 1.563 1.296
Drift (cm) Third 1.663 1.515 1.171 1.045 0.805 0.627
First 0.447 0.414 0.332 0.295 0.235 0.259
(g) Third 0.744 0.679 0.523 0.468 0.361 0.282
Control
force(kN) 589.95 905.25 1912.44 2511.55 4326.37 6375.47

First 3.219 2.876 2.122 1.860 1.363 1.071
(cm) Third 6.268 5.929 4.757 4.132 3.074 2.182
First 3.219 2.876 2.122 1.86 1.363 1.071
Drift (cm) Third 1.884 1.739 1.313 1.12 0.788 0.661
First 0.670 0.592 0.405 0.344 0.287 0.323
(g) Third 0.846 0.781 0.589 0.504 0.353 0.298
Control
force(kN) 492.42 750.77 1673.98 2163.06 3326.58 5072.50
74
in displacement, interstorey and acceleration response can be observed i.e. around 26% reduction
in top floor maximum displacement, inter-storey drift as well as acceleration response as compared
to uncontrolled case. Also the control performance is improving for all the responses by increasing
the value of R.
From Table 4.3 for GM2, percentage reduction in top floor displacement is 13 % and floor
acceleration and inter-storey drift is 17% for Case 1. From Figure 3.6, it can be seen that not much
variation has occurred in acceleration and inter-storey drifts after Case 4 .
For GM3, percentage reduction for Case 1 is 9% for top floor peak displacement and 12% for
peak inter-storey drift and acceleration. From Figure 3.7, it can be concluded that Case 3 gives the
optimum control for both inter-storey displacement and acceleration. No further improvement in
control performance can be seen by increasing control force after Case 3. Response increase and
a very high control force can be observed for Case 6.
For GM4 and GM5 good control performance can be observed for both Case 3 and Case 4.
Similar trend in controlled response is observed for GM4 and GM5 as obtained for GM1. For both
the ground motions control performance is improving by decreasing R value, thereby increasing
control force . Best controlled response has been obtained for Case 6, for which the maximum
control force requirement is around 6000 kN for both GM4 and GM5.
75
Figure 3.5: GM1 : (a) Displacement, (b) Inter-storey Drift and (b) Acceleration response
76
77
Table 3.8: Percentage reduction in displacement and acceleration response

GM1 GM2 GM3 GM4 GM5
Floor (Case 4) (Case 4) (Case3) (Case3) (Case 4)
First 49.88 61.10 2.86 41.42 53.08
Displacement Second 44.47 56.56 14.11 40.66 50.94
(cm) Third 42.97 46.40 19.24 40.11 45.26
First 49.88 61.10 2.86 41.42 53.08
Inter-storey Second 39.90 44.40 24.87 38.68 42.84
Drift (cm) Third 38.83 34.00 33.77 41.21 48.08
First 29.65 50.60 4.64 38.52 57.84
Acceleration Second 45.66 50.76 3.52 40.09 48.06
(g) Third 38.93 33.72 33.76 41.37 47.93
Force(kN) 2342.77 2131.38 2367.58 1912.44 2163.06
For MR damper response control of three storey structure, damper of capacity 1000 kN has
been used. For amplified bracing the maximum damping force which can be obtained in a structure
becomes 2000 kN (F = a0 Fd ). Hence, percentage reduction in response of building, when the
maximum control force is around 2000 kN is shown in Table 3.8. From the table it can be observed
that, 19-45% reduction has occurred in maximum displacement response of top floor, 33-48%
reduction can be seen top floor maximum inter-storey drift response, and 33-47% reduction in
the maximum top floor acceleration. In LQR control, control response is reduced over the whole
78
duration of earthquake excitation. The displacement time history of top floor of building is shown
from Figure 3.10 to 3.14. In each plot three responses are shown, first is the uncontrolled response,
second is displacement time history for Case 1 ( R value is highest) for all ground motions, and
third response is for the cases mentioned in Table 3.8 for which maximum control force is around
2000 kN ( Case 4 for GM1, GM2, GM5 and Case 3 for GM3, GM4). From the plots, significant
response reduction can be seen over the whole time duration of earthquake for all the controlled
cases. Also response control is better for Case 3/Case 4 as compared to Case 1, due to lower
weight factor R and higher control force in Case 3/Case 4 as compared to Case 1.
Figure 3.10: GM1 : Displacement time history of top floor

79
3.4 Summary
In this chapter, feedback control laws and Linear Quadratic (LQR) Algorithm are introduced. A
numerical study has been conducted on a three storey shear building in which control force has
been applied at the first floor level. Full state feedback active control using LQR algorithm is
used to obtain optimal control force and controlled response of the building. Different weighing
matrices are used and it has been observed that by decreasing weight factor R, control force is in-
creasing and better control is achieved. However, in some of the ground motions it can be observed
that saturation is achieved after a certain level and no further response control is observed by in-
creasing the control force. Control performance is found to be varying with different earthquake
excitations. From the study it can be concluded that one 1000 kN damper installed in amplified
brace can be very effective for the response control, if same level of response reduction can be
achieved for semiactive control using MR dampers.

Chapter 4
Semi-active Control of three story
structure
4.1 Brief Overview
In this chapter, semi-active response of Magneto-rheological damper is analyzed for the seis-
mic response control of a three story structure for both chevron and amplified bracing. Different
control algorithms like clipped-optimal controller, lyanpunov controller, decentralized bang-bang
controller and maximum energy dissipation have been used. One 1000 kN MR damper is installed
at the first floor of structure. Bouc-Wen model has been used for analysis and all the control algo-
rithms are formulated for use with MR damper. Effect of increased damper capacity on response
control has also been analyzed by installing two MR dampers of 1000 kN capacity at first floor.
Reduction of responses like displacement, inter-storey drift and acceleration have been compared
for chevron and toggle bracing system using different earthquake ground motions.
4.2 Semi-active control analysis of a 3-storey structure
Three storey structure mentioned in Chapter 2 is used for semi-active response control. Also
same SAC ground motions have been used for base excitation, as mentioned in previous two
81
chapters. One 1000 kN MR damper is installed at the first floor and Bouc-Wen model has been
used for obtaining damper force. Two types of bracing arrangements are used, one is chevron
(unamplified) bracing and anathor is scissor-jack (amplified) bracing. Amplification factor of 2
is used for toggle bracing arrangement (a0 = 2). Parameters of Bouc-Wen model for 1000 kN
damper are mentioned in Chapter 2. Equation of motion for the shear building can be written as
[M ] {ẍ(t)} + [C] {ẋ(t)} + [K] {x(t)} = {γ}F (t) − [M ] {λ}x¨g (t) (4.1)
In state-space form equation of motion becomes
{ṡ(t)} = [A]{s(t)} + {B}F (t) + {E}x¨g (t) (4.2)
All parameters used in above equations are explained in Chapter 2. In semiactive control, damper
force also varies with current. By incorporating damper dynamics (Bouc-Wen Model) in Equation
4.2, the control equation becomes
 
 [A]{s(t)} + {B}F (t) + {E}x¨g (t) 
 
 
{Ṡ(t)} = 
 a0 (−γ |x˙1 (t)| z(t) |z(t)|
(n−1)
− β x˙1 (t) |z(t)|n + Ax˙1 (t)) 
 (4.3)
 
 
−η(u(t) − v(t))
where, the state vector is
T T
{S(t)} = s(t) z(t) u(t) = x1 (t) x2 (t) x3 (t) x˙1 (t) x˙2 (t) x˙3 (t) z(t) u(t)
(4.4)
where, v(t) is the applied voltage which is set to be either maximum (Vmax ) or minimum (Vmin =
0).
As obtained in Chapter 2, damping force obtained at the first floor level of the structure due to MR
82
dampers for an amplification factor a0 is given by
For passive control, voltage v(t) becomes time invariant, however, for semi-active control, voltage
varies with time. Different semi-active control algorithms are used for obtaining control voltages
at different time instants.
4.3 Control Algorithms
A variety of algorithms have been proposed for controlling semi-active devices. Some of them are
mentioned here.
4.3.1 Clipped Optimal approach
LQR clipped-optimal control was proposed by (Dyke et al., 1996c,b). In this chapter, full state
feedback control has been used to obtain the optimal control force (Carneiro et al., 2010).
Optimal control force Fc is obtained using LQR algorithm, as mentioned in Chapter 3.
Figure 4.1: Block diagram of Semi-Active Control
Fc (t) = −({B}T [P ]/R){s(t)} = −{G}{s(t)} (4.6)

83
where, [P ] is the ricatti matrix obtained by solving Algebraic Ricatti Equation (ARE).
[P ][A] + [A]T [P ] − [P ]{B}[B]T [P ]/R + [Q] = 0 (4.7)
[Q] and R are the weighing matrices.
Main problem with semi-active control is that, the optimal control force obtained Fc cannot be
directly commanded to structure. Only the voltage signal is sent to damper and force gets modified
according to the signal sent. Once the desired optimal force is calculated and the damper force is
known, both the forces are compared at each time instant and voltage signal is sent to the damper
to produced the desired damper force. Three situations may arise for some applied initial voltage
at each time step.
1. When the MR damper is equal to the optimal force, the voltage applied to the damper
remains at the present level.
2. If the magnitude of the damper force is less than the magnitude of the desired optimal force
and the two forces have the same sign, the voltage is set to the maximum.
3. Otherwise, the voltage becomes zero.
The command signal (voltage v(t)) is selected by following algorithm.
v(t) = Vmax H[{Fc (t) − F (t)}F (t)] (4.8)
where, H is the heaviside step function and Vmax is the maximum voltage capacity of MR damper
(10 Volts for 1000 kN MR damper). Also Fc is the Optimal control force obtained using LQR
algorithm and F is the damping force obtained in structure due to MR damper.

84
4.3.2 Control Based on Lyapunov Stability theory
Lyapunov’s direct approach to stability analysis is also used for the design of feedback controller
(Brogan, 1991). It requires a Lyapunov function which must be a positive definite function of
the states of the system. According to Lyapunov stability theory, for a positive definite Lyapunov
function V (s), if the rate of change of V (s) is negative semi-definite, then the origin is stable in
sense of Lyapunov. In order to achieve this control, input is chosen in such a manner to make V̇ (s)
as negative as possible.
Leitmann (1994) used Lyapunovs direct approach for semiactive controller design. In this ap-
proach, following Lyapunov function is used
1
V (s) = ||s||2P (4.9)
2
where, ||s||P is the P-norm of the states defined by
||s||P = [{s}T [P ]{s}]1/2 (4.10)
where, [P ] is a real, symmetric, positive definite matrix of size (6 × 6) for 3 storey shear building.
{s} is a function of time. From the derivative analysis of V (s) with respect to time and ensuring
that V̇ (s) is negative for stability, following equation is obtained
[A]T [P ] + [P ][A] = −[QP ] (4.11)
where [QP ] is a positive definite matrix of size same as [A] and [P ] matrix. Derivative of Lyapunov
function V̇ (s) is given by
1
V̇ (s) = − {s}T [QP ]{s} + {s}T [P ]{B}F + {s}T [P ]{E}x¨g (4.12)
2
85
In the above equation the middle term which contains the control force is directly affected by the
change of control voltage. So in order to minimize V̇ (s) the following control algorithm is used
v(t) = Vmax H[−{s(t)}T [P ]{B}F ] (4.13)
where H is the Heaviside step function and F is the damping force obtained at the first floor of
structure due to MR damper. In this algorithm, a positive definite matrix [QP ] is selected first.
From Equation 4.11, [P ] matrix is obtained and verified that it should be real, symmetric and
positive definite matrix. The main challenge in this algorithm is the selection of [QP ] matrix. For
different choices of matrix, different set of responses are achieved.
4.3.3 Decentralized Bang-Bang Control
Similar approach was used by McClamroch and Gavin (1995) to develop decentralized bang bang
algorithm for control using Electro-rheological (ER) damper. In this approach, lyapunov function
was chosen to be the total vibratory energy of the system (Kinetic energy + potential energy).
1 1 ˙ ˙ + {λ}x˙g )
V (s) = {x}T [K]{x} + ({x} + {λ}x˙g )T [M ]({x} (4.14)
2 2
Rate of change of Lyapunov function is given by
1 ˙ + {λ}x˙g )T (−[C]{x}
˙ − [K]{x} + {γ}F )
V̇ (s) = {x}T [K]{x} + ({x} (4.15)
2
T T
where, {x} = x1 x2 x3 and {ẋ} = x˙1 x˙2 x˙3
In the above equation the last term carrying control force F is affected by the change in control
voltage. So, in order to make V̇ (s) negative and as large as possible like in the case of Lyapunov
86
Stability theory, the following algorithm is used
v(t) = Vmax H(−({ẋ(t)} + {λ}x˙g (t))T {γ}F (t)) (4.16)
This control law requires the measurement of only floor velocities. If the damper is located on
upper floors (not on first floor), inter-story velocity is needed between floors connecting damper.
However, if the damper is located between ground and first floor, measurement of absolute velocity
of first floor is required, which is not readily available. In that case pseudo velocity can be obtained
by integrating absolute acceleration.
4.3.4 Maximum Energy Dissipation
McClamroch and Gavin proposed this algorithm with a slight variation from Decentralized Bang-
Bang approach. Instead of using total vibratory energy in the Lyapunov function, relative vibratory
energy of the structure is used. Ground velocity is not used in the function in Kinetic energy term.
Lyapunov function is defined as
1 1
V (s) = {x}T [K]{x} + {ẋ}T [M ]{ẋ} (4.17)
2 2
Control law is defined as
v(t) = Vmax H(−{ẋ(t)}T {λ}F (t)) (4.18)
Note, if the damper is located between first floor and ground, measurement of relative velocity of
first floor with respect to ground is required unlike in decentralized bang-bang which requires the
measurement of absolute velocity of first floor. However, if the damper is located on upper floors,
inter-storey velocity measurement is required, which is same as decentralized bang-bang control.
Hence, this algorithm gives same results as decentralized bang-bang approach unless the damper
is located at the first floor.

87
Response of 3-story building has been obtained for SAC ground motion data from GM1 to GM5.
Different semiactive control algorithms along with two passive cases have been used when a 1000
kN MR damper is installed at first floor of building for both chevron (Amplification, a0 = 1)
and amplified/scissor jack braces (Amplification, a0 = 2). In passive off no voltage (0 Volt)
is supplied through MR dampers and similarly in passive on case constant voltage of 10 Volts
(maximum capacity) is supplied through damper. Two cases of semiactive LQR control have been
considered, one with wight factor R = 10−14 and other with R = 10−15 . For both the cases [Q]
matrix is a (6 × 6) identity matrix, which means same weight is assigned to all the states. For
lyapunov control there is no standard method for selecting [QP ] matrix. In the present case [QP ]
matrix used is (6 × 6) identity matrix for lyapunov, decentralized bang-bang and maximum energy
dissipation approach. All the semiactive control strategies used are summarized in Table 4.1
Table 4.1: Different Semi-active Control strategies

SA1 Clipped optimal LQR with R=10-14
SA2 Clipped optimal LQR with R=10-15
SA3 Lyapunov Stability theory
SA4 Decentralized bang-bang control
SA5 Maximum energy Dissipation theory
Tables 4.2–4.6 gives the maximum displacement, inter-storey drift and acceleration response of
all the floors of the building for different ground motion data. Also the maximum damping force
obtained at the first floor of building is mentioned. Tables 4.7–4.11 gives the percentage response
reduction of roof of the building. Also percentage reduction for active LQR algorithm (as obtained
in Chapter 3), for which maximum optimal control force is close to 2000 kN, is included in the
results.
Table 4.2 gives response of structure when it is subjected to GM1 ground motion. From the
table, better response control can be observed for all the cases having scissor jack bracing ar-
rangement (Amplification, a0 = 2) as compered to their respective control cases having chevron

88
bracing arrangement (Amplification, a0 = 1), except for first and second floor acceleration re-
sponse of passive on case due to high stiffening of floor. For chevron bracing performance of both
the clipped optimal strategies (SA1 and SA2) is better than all the other cases. For scissor jack
(amplified) bracing same trend is observed. Also for scissor jack (amplified) bracing, the perfor-
mance of SA1 and SA2 is found to be almost similar. From Table 4.7, 43% reduction in top floor
acceleration and inter-storey drift and 45% reduction in top storey displacement can be observed
for both SA1 and SA2 cases. For the other three cases around 35% response reduction in top floor
acceleration and inter-storey drift and 40% reduction in top storey displacement can be observed.
All the semi-active responses performed better in response control than passive cases.
Table 4.3 gives the response of building when it is subjected to GM2 ground motion. From the
table, better response reduction for all the control cases can be seen for amplified cases (scissor
jack bracing) as compared to unamplified cases (chevron bracing). For amplified case, maximum
acceleration control for all the floors and maximum top storey drift control can be seen for SA4
case. However, displacement response, for SA4, of the bottom two floors is higher than other
semiactive responses and passive on case. Response control of SA2 is better as compared to SA1.
In this case, performance of passive on is comparable to SA2 case.
Similarly, for the other three ground motions, significant response reduction can observed for
all cases of amplified bracing. For GM3, maximum reduction in acceleration and displacement
responses is observed for SA1 case (38% reduction for top floor of amplified case). For amplified
case of GM4, control performance of SA2 is better than other control cases. A very slight variation
can be seen in control performance of SA1 and SA2, however, they outperformed other control
algorithms. For SA2 case, maximum percentage reduction of top floor acceleration and inter-
storey drift is 43% and displacement is 46% as compared to uncontrolled case.
For GM5 , both SA1 and SA2 showed a significant reduction in displacement, acceleration
and inter-storey drift responses for amplified case. Even, SA3 response for inter-storey drift and
acceleration control is quite comparable to SA1 and SA2 cases (60% reduction of both top floor
89
inter-storey drift and acceleration).
From the results, it can be seen that there is a very slight reduction in all the responses for
passive off unamplified case (upto 10% for top floor). Further response reduction is achieved for
passive off amplified case (upto 23% for top floor). However, passive on case and all the other
Semiactive cases for amplified bracings are very effective for response control. Also for all the
ground motions, performance of clipped-optimal control is better than the active control.
Figure 4.2: GM1 : (a) Acceleration response, and (b) Displacement response
90
Table 4.2: Maximum response of building for GM1

Displacemnt Acceleration Inter-storey Control
(cm) (g) Drift(cm) Force(kN)
Uncontrolled 4.128 7.218 9.178 0.489 0.692 0.912 4.128 3.341 2.037
Passive off 3.733 6.503 8.197 0.456 0.635 0.816 3.733 2.996 1.821 111.39
Passive on 2.659 4.884 6.142 0.521 0.570 0.638 2.659 2.387 1.427 1000
SA1 2.459 4.601 5.863 0.454 0.487 0.607 2.459 2.306 1.353 1000
a0 = 1 SA2 2.462 4.625 5.897 0.448 0.484 0.601 2.462 2.314 1.339 1000
SA3 2.578 4.767 6.041 0.497 0.527 0.622 2.578 2.335 1.393 1000
SA4 2.405 4.675 6.136 0.445 0.477 0.660 2.405 2.285 1.476 1000
SA5 2.590 4.796 6.063 0.486 0.518 0.632 2.590 2.370 1.416 1000
Passive off 3.265 5.689 7.158 0.448 0.557 0.697 3.265 2.547 1.558 277.47
Passive on 2.259 4.315 5.334 0.598 0.614 0.630 2.259 2.057 1.411 2000
SA1 1.949 3.940 5.010 0.440 0.457 0.515 1.949 2.012 1.154 2000
a0 = 2 SA2 1.985 3.976 5.043 0.444 0.466 0.516 1.985 2.000 1.156 2000
SA3 2.219 4.413 5.484 0.546 0.565 0.583 2.219 2.194 1.306 2000
SA4 2.248 4.169 5.400 0.398 0.494 0.590 2.248 1.944 1.320 2000
SA5 2.189 4.261 5.323 0.558 0.567 0.605 2.189 2.072 1.356 2000
Uncontrolled 3.977 6.542 7.640 0.670 0.732 0.860 3.977 2.719 1.919
Passive off 3.736 6.223 7.374 0.627 0.653 0.797 3.736 2.630 1.780 112.70
passive on 2.104 3.644 4.731 0.487 0.450 0.627 2.104 1.788 1.397 1000
SA1 2.106 3.647 4.715 0.464 0.455 0.610 2.106 1.751 1.360 1000
a0 = 1 SA2 2.109 3.643 4.703 0.452 0.456 0.603 2.109 1.744 1.343 1000
SA3 2.146 3.849 4.952 0.383 0.392 0.609 2.146 1.831 1.357 1000
SA4 2.076 3.634 4.711 0.442 0.437 0.586 2.076 1.748 1.310 1000
SA5 2.093 3.657 4.730 0.470 0.448 0.630 2.093 1.750 1.402 1000
Passive off 3.409 5.774 6.984 0.538 0.585 0.698 3.409 2.494 1.562 291.79
passive on 1.214 2.443 3.611 0.368 0.392 0.541 1.214 1.501 1.207 2000
SA1 1.267 2.789 4.071 0.351 0.315 0.574 1.267 1.579 1.281 2000
a0 = 2 SA2 1.270 2.723 3.972 0.382 0.343 0.559 1.270 1.537 1.250 2000
SA3 1.250 2.644 3.838 0.383 0.365 0.536 1.250 1.421 1.194 2000
SA4 1.446 2.829 3.881 0.360 0.321 0.472 1.446 1.450 1.053 2000
SA5 1.222 2.470 3.659 0.389 0.379 0.545 1.222 1.468 1.214 2000
91
92

Uncontrolled 4.026 7.604 9.887 0.647 0.653 1.022 4.026 3.603 2.283
Passive off 3.998 7.415 9.602 0.648 0.653 0.979 3.998 3.494 2.187 117.88
Passive on 3.748 6.380 7.790 0.612 0.662 0.699 3.748 2.878 1.559 1000
SA1 3.902 6.507 7.942 0.594 0.659 0.745 3.902 2.891 1.666 1000
a0 = 1 SA2 3.963 6.585 8.070 0.635 0.638 0.744 3.963 2.868 1.657 1000
SA3 3.759 6.382 7.813 0.626 0.669 0.699 3.759 2.887 1.562 1000
SA4 4.036 6.668 8.218 0.680 0.702 0.784 4.036 3.116 1.749 1000
SA5 3.786 6.414 7.826 0.640 0.670 0.701 3.786 2.897 1.567 1000
Passive off 3.964 7.176 9.232 0.640 0.650 0.921 3.964 3.353 2.056 319.14
Passive on 3.471 6.090 7.478 0.573 0.679 0.630 3.471 2.684 1.413 2000
SA1 3.901 6.518 7.863 0.610 0.661 0.625 3.901 2.742 1.398 2000
a0 = 2 SA2 3.939 6.518 7.964 0.683 0.629 0.691 3.939 2.718 1.542 2000
SA3 3.444 5.999 7.405 0.580 0.619 0.637 3.444 2.673 1.420 2000
SA4 3.993 6.673 8.113 0.694 0.780 0.712 3.993 3.204 1.597 2000
SA5 3.506 6.085 7.510 0.599 0.676 0.645 3.506 2.685 1.438 2000
Uncontrolled 4.184 7.466 9.227 0.540 0.686 0.892 4.184 3.281 1.992
Passive off 3.966 7.115 8.282 0.495 0.645 0.846 3.966 3.150 1.889 113.84
Passive on 3.025 5.445 6.684 0.428 0.557 0.635 3.025 2.437 1.426 1000
a0 = 1 SA1 2.942 5.227 6.498 0.342 0.515 0.596 2.942 2.285 1.333 1000
SA2 2.955 5.280 6.547 0.348 0.513 0.606 2.955 2.325 1.354 1000
SA3 3.000 5.473 6.767 0.359 0.556 0.622 3.000 2.473 1.391 1000
SA4 3.045 5.502 6.792 0.378 0.540 0.630 3.045 2.457 1.413 1000
SA5 3.007 5.409 6.675 0.378 0.546 0.626 3.007 2.410 1.397 1000
Passive off 3.661 6.503 8.101 0.426 0.590 0.781 3.661 2.895 1.748 301.093
Passive on 2.228 4.039 5.119 0.419 0.545 0.561 2.228 1.848 1.259 2000
a0 = 2 SA1 2.165 3.989 4.951 0.344 0.481 0.515 2.165 1.824 1.153 2000
SA2 2.138 3.973 4.947 0.316 0.485 0.504 2.138 1.834 1.126 2000
SA3 2.482 4.539 5.705 0.402 0.510 0.589 2.482 2.075 1.321 2000
SA4 2.265 4.187 5.262 0.395 0.520 0.541 2.265 1.970 1.211 2000
SA5 2.298 4.141 5.229 0.410 0.511 0.567 2.298 1.887 1.271 2000
93
94

Uncontrolled 3.964 6.423 7.549 0.816 0.801 0.968 3.964 2.862 2.157
Passive off 3.705 6.066 6.928 0.740 0.742 0.905 3.705 2.707 2.017 112.77
Passive on 1.714 3.236 4.383 0.361 0.436 0.520 1.714 1.803 1.159 1000
SA1 1.873 3.356 4.369 0.376 0.422 0.546 1.873 1.800 1.217 1000
a0 = 1 SA2 1.985 3.360 4.446 0.422 0.434 0.548 1.985 1.807 1.222 1000
SA3 1.912 3.627 4.685 0.316 0.417 0.476 1.912 1.827 1.060 1000
SA4 1.908 3.603 4.938 0.401 0.448 0.612 1.908 1.887 1.361 1000
SA5 1.740 3.287 4.429 0.362 0.415 0.516 1.740 1.833 1.149 1000
Passive off 3.313 5.535 6.374 0.607 0.652 0.821 3.313 2.548 1.827 290.91
Passive on 1.524 2.928 3.612 0.337 0.451 0.422 1.524 1.503 0.948 2000
SA1 1.402 2.607 3.429 0.335 0.383 0.378 1.402 1.317 0.839 2000
a0 = 2 SA2 1.367 2.530 3.365 0.328 0.364 0.384 1.367 1.286 0.853 2000
SA3 1.538 2.884 3.682 0.326 0.407 0.379 1.538 1.434 0.850 2000
SA4 1.636 2.992 3.658 0.369 0.403 0.428 1.636 1.362 0.954 2000
SA5 1.489 2.898 3.603 0.348 0.458 0.434 1.489 1.486 0.973 2000
95
Table 4.7: Percentage reduction in displacement, acceleration and interstory drift at roof of build-
ing for GM1
a0 = 2 a0 = 1
Displace- Accele- Inter-storey Displace- Accele- Inter-storey
-ment -ration Drift -ment -ration Drift
Passive off 22.01 23.57 23.51 10.69 10.53 10.60
Passive on 41.88 30.92 30.73 33.08 30.04 29.95
SA1 45.41 43.53 43.35 36.12 33.44 33.58
SA2 45.05 43.42 43.25 35.75 34.10 34.27
SA3 40.25 36.07 35.89 34.18 31.80 31.62
SA4 41.16 35.31 35.20 33.14 27.63 27.54
SA5 42.00 33.66 33.43 33.94 30.70 30.49
Active 42.97 38.93 38.83 - - -
ing for GM2
a0 = 2 a0 = 1
Passive off 8.59 18.84 18.60 3.48 7.33 7.24
Passive on 52.74 37.09 37.10 38.08 27.09 27.20
SA1 46.71 33.26 33.25 38.29 29.07 29.13
SA2 48.01 35.00 34.86 38.44 29.88 30.02
SA3 49.76 37.67 37.78 35.18 29.19 29.29
SA4 49.20 45.12 45.13 38.34 31.86 31.74
SA5 52.11 36.63 36.74 38.09 26.74 26.94
Active 46.40 33.72 34.00 - - -
ing for GM3
a0 = 2 a0 = 1
Passive off 6.62 9.88 9.94 2.88 4.21 4.20
Passive on 24.37 38.36 38.11 21.21 31.60 31.71
SA1 20.47 38.85 38.76 19.67 27.10 27.03
SA2 19.45 32.39 32.46 18.38 27.20 27.42
SA3 25.10 37.67 37.80 20.98 31.60 31.58
SA4 17.94 30.33 30.05 16.88 23.29 23.39
SA5 24.04 36.89 37.01 20.85 31.41 31.36
Active 19.24 33.76 33.77 - - -
96
ing for GM4
a0 = 2 a0 = 1
Passive off 12.20 12.44 12.25 10.24 5.16 5.17
Passive on 44.52 37.11 36.80 27.56 28.81 28.41
SA1 46.34 42.26 42.12 29.58 33.18 33.08
SA2 46.39 43.50 43.47 29.05 32.06 32.03
SA3 38.17 33.97 33.68 26.66 30.27 30.17
SA4 42.97 39.35 39.21 26.39 29.37 29.07
SA5 43.33 36.43 36.19 27.66 29.82 29.87
Active 40.11 41.37 41.21 - - -
ing for GM5
a0 = 2 a0 = 1
Passive off 15.56 15.19 15.30 8.23 6.51 6.49
Passive on 52.15 56.40 56.05 41.94 46.28 46.27
SA1 54.58 60.95 61.10 42.12 43.60 43.58
SA2 55.42 60.33 60.45 41.10 43.39 43.35
SA3 51.23 60.85 60.59 37.94 50.83 50.86
SA4 51.54 55.79 55.77 34.59 36.78 36.90
SA5 52.27 55.17 54.89 41.33 46.69 46.73
Active 45.26 47.93 48.08 - - -
97
98
Figure 4.6: GM5 :(a) Acceleration response, and (b) Displacement response
99
4.5 Effect of capacity of dampers
In this section, building is analyzed with two 1000 kN capacity dampers installed at first floor.
Ground motion data GM1 is used for analysis which has a dominant frequency close to the fun-
damental frequency of building. All the control algorithms have been used, which has been men-
tioned earlier and amplified/scissor jack bracing (Amplification, a0 = 2) arrangement is used.
Results are shown in Table 4.12. Values in brackets shows the percentage reduction in response
as compared to uncontrolled case. For the case when one damper has been used, performance
of SA1 and SA2 is similar (Table 4.2). However, from Table 4.12, it can be observed that SA2
case is most effective in controlling all the responses for each floor of the building as compared to
other cases. The effect of weighing matrices can be seen in this case as the range of control force
variation has increased. By reducing weight factor R, better control can be achieved.
4.6 Summary
In this chapter, effect of amplification factor of bracings for semiactive control is evaluated for MR
damper. Two cases of bracings have been used , one is chevron with no amplification and other
is scissor jack bracing with amplification factor of 2. Different control algorithms for semiactive
control has been used. Passive off, passive on and active control cases are also included. From the
results it can be observed that, for amplified (scissor jack) bracings the maximum capacity of con-
trol force has become twice as compared to chevron, resulting in better control performance and
response reduction. All the semiactive control algorithms performed better for amplified (scissor
jack) bracings as compared to their unamplified (chevron bracing) counterpart. The performance
of different control algorithm is found to be varying with different ground motions. Effect of in-
creased damper capacity has also been evaluated by installing two dampers of capacity 1000 kN.
It ha sbeen observed that LQR clipped optimal algorithms is providing better results as compared
to most of the other cases. Effect of weighing matrices for LQR control is found to be varying
100
Table 4.12: Response of building when excited with GM1 for damper capacity of 2000 kN (Amplification 2)
Displacement Acceleration Inter-storey Control
Uncontorolled 4.128 7.218 9.178 0.489 0.692 0.912 4.128 3.341 2.037
Passive off 2.855 5.006 6.392 0.452 0.489 0.671 2.855 2.370 1.500 536.67
(3 0.84) (30.65) (30.36) (7.57) (29.34) (26.43) (30.84) (29.06) (26.36)
Passive on 1.625 3.513 4.166 0.596 0.605 0.515 1.625 1.889 1.530 4000
(60.63) (51.33) (54.61) (-21.88) (12.57) (43.53) (60.63) (43.46) (24.89)
SA 1 1.749 3.371 4.325 0.257 0.373 0.435 1.749 1.798 0.968 4000
(57.63) (53.30) (52.88) (47.44) (46.10) (52.30) (57.63) (46.18) (52.48)
SA 2 1.471 2.923 3.753 0.328 0.362 0.377 1.471 1.626 0.837 4000
(64.37) (59.50) (59.11) (32.92) (47.69) (58.66) (64.37) (51.33) (58.91)
SA 3 1.874 3.894 4.851 0.467 0.477 0.437 1.874 2.020 0.981 4000
(54.60) (46.05) (47.15) (4.50) (31.07) (52.08) (54.60) (39.54) (51.84)
SA 4 1.817 3.108 3.909 0.382 0.439 0.406 1.817 1.396 0.909 4000
(55.98) (56.94) (57.41) (21.88) (36.56) (55.48) (55.98) (58.22) (55.38)
SA 5 1.592 3.485 4.223 0.512 0.551 0.478 1.592 1.919 1.061 4000
(61.43) (51.72) (53.99) (-4.70) (20.38) (47.59) (61.43) (42.56) (47.91)
101
with damper capacity and ground motions.

Chapter 5
Semi-active Control of eight storey
shear Building
5.1 Brief Overview
In this chapter, the effectiveness of damper location for eight storey shear building has been evalu-
ated using Modal Controllability method. One 1000 kN MR damper is used and installed at differ-
ent stories to obtain the response of structure for different cases. Semiactive LQR clipped optimal
(C-O) control approach has been used for analysis. Responses such as displacement, inter-storey
drift and floor acceleration of the structure for different cases has been evaluated and compared for
both amplified (scissor jack) and unamplified (chevron) braces configuration. Anathor approach,
direct control method has been used to control inter-storey displacement. In this approach, voltage
applied to the damper can take any value between 0 and Vmax . Optimal control force is calculated
using LQR algorithm. Optimal applied current voltage is chosen in such a manner so that the
MR damper force becomes close to the optimal control force at each time instant. To verify the
effectiveness of this method, seismic evaluation of 8-storey building has been done for the case of
amplified bracing and the response obtained is compared with C-O controller.
103
5.2 Shear Building Model Description
An eight story building structure which has been used as a benchmark problem by a number of
other researchers (Spencer et al., 1994; Yeesock et al., 2008) is investigated here. The mass of
each floor is 345,600 kg; the stiffness of each story is 344,400 kN/m; and the damping coeffi-
cient of each floor 2,937,000 N-s/m. Mass, stiffness, and damping matrix are obtained using the
formulations described in Chapter 2.
m1 = m2 = · · · = m8 = 345, 600kg (5.1)
k1 = k2 = · · · = k8 = 344, 400kN/m (5.2)
c1 = c2 = · · · = c8 = 2, 937, 000N s/m (5.3)
Mass matrix of the model is given by
Figure 5.1: 8-story shear building model

104
 
 345600 0 0 0 0 0 0 0 
 
 
0 345600 0 0 0 0 0 0
 
 
 
 
 
 0 0 345600 0 0 0 0 0 
 
 
 
 0 0 0 345600 0 0 0 0 
[M ] =   (kg)
 
 

 0 0 0 0 345600 0 0 0 

 
 

 0 0 0 0 0 345600 0 0 

 
 

 0 0 0 0 0 0 345600 0 

 
 
0 0 0 0 0 0 0 345600
(5.4)
Stiffness matrix of the model can be obtained as
 
 6888 −3444 0 0 0 0 0 0 
 
 
 −3444 6888 −3444 0 0 0 0 0
 

 
 
−3444 6888 −3444
 
 0 0 0 0 0 
 
 
 
 0 0 −3444 6888 −3444 0 0 0 
[K] =   (×105 N/m)
 
 

 0 0 0 −3444 6888 −3444 0 0 

 
 

 0 0 0 0 −3444 6888 −3444 0 

 
 

 0 0 0 0 0 −3444 6888 −3444 

 
 
0 0 0 0 0 0 −3444 3444
(5.5)
105
Damping matrix is given by
 
 5874 −2937 0 0 0 0 0 0 
 
 
 −2937 5874 −2937 0 0 0 0 0
 

 
 
−2937 5874 −2937
 
 0 0 0 0 0 
 
 
 
 0 0 −2937 5874 −2937 0 0 0 
[C] =   (kN − s/m)
 
 

 0 0 0 −2937 5874 −2937 0 0 

 
 

 0 0 0 0 −2937 5874 −2937 0 

 
 

 0 0 0 0 0 −2937 5874 −2937 

 
 
0 0 0 0 0 0 −2937 2937
(5.6)
Eigenvalue analysis has been performed to determine the natural frequencies and mode shapes of
the numerical model. The natural frequencies and mode shape vectors of the shear building are
given in Tables 5.1 and 5.2. Also plots of mode shapes for the first three modes are shown in
Figure 5.2.
Table 5.1: Frequencies (in Hz) of the 8-story structure

Mode Frequency(Hz)
1 0.927
2 2.750
3 4.479
4 6.056
5 7.426
6 8.543
7 9.370
8 9.877
Table 5.2: Mode shapes for the 1st three modes φi

Mode 1 (φ1 ) Mode 2 (φ2 ) Mode 3 (φ3 )
-0.184 -0.529 -0.801
-0.363 -0.899 -0.966
-0.529 -1 -0.363
-0.677 -0.801 0.529
-0.801 -0.363 1
-0.899 0.184 0.677
-0.966 0.677 -0.184
-1 0.966 -0.899
106
Figure 5.2: Mode shapes of the Model
5.3 Modal Controllability for Location Optimization
This method is very effective for structural response with one dominant mode. Measure for modal
controllability gives how controllable a system is, for different actuator/damper locations. This
method gives the effect of control force for response reduction of a system for a particular mode.
Modal coefficients are calculated for each damper location and the location with highest value is
the optimum location (Lu et al., 1994; Cheng and Pantelides, 1988).
From Chapter 2, equation of motion of structure is given by
[M ] {ẍ(t)} + [C] {ẋ(t)} + [K] {x(t)} = {γ}F (t) − [M ]{λ}x¨g (t) (5.7)
where, {x(t)} is displacement vector. for 8-storey structure, it is given by
{x(t)} = { x1 (t) x2 (t) . . . x8 (t) }T (5.8)

107
{γ} is the damper location vector of size (8 × 1) and {λ} is a (8 × 1) unit vector. Equation can
also be written as
[M ] {ẍ(t)} + [C] {ẋ(t)} + [K] {x(t)} = {γ}F (t) + {δ}x¨g (t) (5.9)
where,
{δ} = −{ m1 m2 . . . m8 }T (5.10)
For modal analysis, displacement vector can be expressed as
{x(t)} = [Φ] [q(t)] (5.11)
where [Φ] is mode shape matrix constructed with mode shape vectors {φi }, and [q(t)] is modal
coordinates vector. For ith mode, equation of motion in the form of modal coordinates, can be
expressed as
q¨i (t) + 2ζi ωi q˙i (t) + ωi 2 qi (t) = γi u(t) + δi x¨g (t) (5.12)
In state space form, equation can be obtained as below
        
       
 q˙i (t)
 
 0 1   qi (t) 
  0
 
  0
 

=  + F (t) + x¨g (t) (5.13)
        
 q¨i (t)  −ωi 2 −2ζi ωi  q˙i (t)  γi   δi 
   
    
where,
γi = {φi }T {γ}/Mi
δi = {φi }T {δ}/Mi
Mi = {φi }T [M ]{φi }
108
Thus the effectiveness of control force is proportional to γi . If magnitude of γi is large, control
force is more effective for controlling response for ith mode. If γi is 0, then ith mode is uncon-
trollable, that means control force has no effect on that mode. Hence, the optimal location of
actuator/damper for ith mode is one for which the magnitude of γi is maximum. Since Mi is same
for a particular mode, γi can be expressed as
γi = | [φi ]T {γ}| (5.14)
Location vector {γ} for all the dampers location is given in Table 5.3 and mode shapes are given
in Table 5.2. Using Equation 5.14, the values of modal controllability factors are obtained, which
are shown in Table 5.4.
From the modal controllability factor, it can be seen that when 1st mode is dominant, damper
Table 5.3: Location Vector {γ}

Damper Location
Floor 1 Floor 2 Floor 3 Floor 4 Floor 5 Floor 6 Floor 7 Floor 8
-1 1 0 0 0 0 0 0
0 -1 1 0 0 0 0 0
0 0 -1 1 0 0 0 0
0 0 0 -1 1 0 0 0
0 0 0 0 -1 1 0 0
0 0 0 0 0 -1 1 0
0 0 0 0 0 0 -1 1
0 0 0 0 0 0 0 -1
Table 5.4: Modal Controllability Factors

Dominant Mode
Damper Location Mode 1 Mode 2 Mode 3
1st Floor 0.184 0.529 0.801
2nd Floor 0.178 0.370 0.164
3rd Floor 0.166 0.101 0.603
4th Floor 0.148 0.199 0.891
5th Floor 0.125 0.439 0.471
6th Floor 0.098 0.548 0.323
7th Floor 0.067 0.492 0.861
8th Floor 0.034 0.289 0.714
should be located at the first floor. When 2nd mode is dominant, 6th floor damper location is most
109
effective and 1st floor location also holds goods. For third dominant mode, 4th floor location is
most optimal.
5.4 Analysis of 8 storey structure with different damper locations
In order to validate results obtained by modal controllability factor, model has been subjected to
different harmonic excitation frequencies. Different damper locations which have been used for
analysis are shown in Figure 5.3. Two different arrangements of bracings have been used, chevron
Figure 5.3: Damper Location
(amplification factor, a0 = 1) and scissor jack bracing (amplification factor, a0 = 2) arrangements,
which have already been described in Chapter 2. Six different cases based on bracing arrangements
and damper location are used for analysis which are listed in Table 5.5.
As mentioned in Chapter 2, equation of motion in state-space form for the 8-storey structure can
be obtained as below.
{ṡ(t)} = [A]{s(t)} + [B]{F (t)} + {E}x¨g (t) (5.15)

110
Table 5.5: Different Cases for Anaysis

Case 1 (a) damper installed at first floor in amplified (scissor-jack) bracing
Case 2 (b) damper installed at first floor in chevron bracing
Case 3 (c) damper installed at fourth floor in amplified (scissor-jack) bracing
Case 4 (d) damper installed at fourth floor in chevron bracing
Case 5 (e) damper installed at sixth floor in amplified (scissor-jack) bracing
Case 6 (f) damper installed at sixth floor in chevron bracing
where,
 
 
 {x(t)}
 
 T
{s(t)} = = x1 (t) x2 (t) . . . x8 (t) x˙1 (t) x˙2 (t) . . . x˙8 (t)
 
 {ẋ(t)} 
 
(16×1)
{x(t)} and {ẋ(t)} corresponds to the floor displacement and velocity vectors relative to ground.
     
 
 [0] [I]   {0}   {0} 
 
[A] = 


 {B} = 


 {E} =
−1 −1 −1
 
− [M ] [K] − [M ] [C] − [M ] {γ}  −{λ} 
 
(16×16) (16×1) (16×1)
{γ} vector is obtained from Table 5.3, for all the three damper locations.
One 1000 kN MR damper has been installed at different floors of the structure. Bouc-Wen model
has been used to obtain the force obtained by damper. When damper is located at first storey,
damper force is given by
z is the hysteric component given by
ż(t) = a0 (−γ |x˙1 (t)| z(t) |z(t)|(n−1) − β x˙1 (t) |z(t)|n + Ax˙1 (t)) (5.17)
For damper located on upper storeys, instead of x1 (t) and x˙1 (t), inter-storey displacements and
inter-storey velocities, respectively, of the floors are used between which damper is installed.
LQR clipped optimal (C-O) approach using full state feedback has been used for analysis.
111
Equal weight is assigned to all the states of structure (8 displacements and 8 velocities). Hence,
[Q] is an identity matrix of size (16 × 16) and R is 10-15 . Same weighing matrices are used for
all the cases. The model with different cases of damper location and bracing arrangements is
subjected to four different base excitations. The first three are harmonic motion with first three
modal frequencies and last is scaled El-centro ground motion.
5.4.1 Results and Discussions
In this section, response of the shear building for various damper locations and brace arrangements
have been compared for different base excitations.
Figure 5.4: Response of structure when it is subjected to excitation frequency ω = ω1 : (a)

Displacement, (b) Inter-story drift, (c) Floor Acceleration response.
Figure 5.4 shows the displacement, inter-story and acceleration response of a the structure
when it is subjected to first fundamental frequency. From the plot it can be seen that maximum
reduction of response is obtained for Case 1. Effect of amplification of braces can be very well
112
seen in these plots. Even better response control is observed for Case 3 as compared to Case 2
due to amplification of force. Response reduction is least for Case 6 as expected from theory of
amplification and least modal controllability factor.
Figure 5.5: Response of structure when it is subjected to excitation frequency ω = ω2 : (a)

Displacement, (b) Inter-story drift, (c) Floor Acceleration response.
Response of structure when second mode is dominant is shown in Figure 5.5. From the plots
it can be seen that Case 1 and Case 5 are most effective for controlling the response. Effect of
amplification can be seen in these plots, however, it is not as evident as compared to the 1st modal
response. 4th floor response is most ineffective for response control which agrees the modal
controllability factor.
When the structure has been subjected to harmonic motion with third modal frequency, not
much response reduction is observed. However, maximum displacement response reduction can be
seen in Figure 5.6(a) for Case 1 and maximum inter-story drift and acceleration response reduction
can be observed for Case 3 as expected.

113
Figure 5.6: Response of structure when it is subjected to excitation frequency ω = ω3 : (a) Dis-
placement, (b) Inter-story drift, (c) Floor Acceleration response.
To get a generalized case of damper location, structure has been subjected to scaled El-centro
earthquake. Response obtained is shown in Figure 5.7. The effect of amplification can be best
seen by comparing Cases 1-2 and Cases 3-4. Cases 5 and 6 are not effective for control, however,
effect of amplification can still be seen in Cases 5 and 6 by comparing their displacement profile.
Case 1 is most effective for controlling the peak response of all the floors.
Tables 5.6, 5.7, and 5.8 shows percentage reduction in maximum displacements, inter-storey
drifts and acceleration of all the floors as compared to uncontrolled case, respectively, for different
base excitations. From Table 5.6, it can be observed that maximum reduction in top floor dis-
placement is observed, when structure is excited in the fundamental mode for all the cases (upto
82%). Similarly, from Tables 5.7 and 5.8, maximum reduction in top floor inter-storey drift and
acceleration can be observed when structure is excited in fundamental mode (upto 72%).
In the acceleration plots for all the ground motions and from Table 5.8, it can be seen that
114
Figure 5.7: Response of structure when it is subjected to El-centro earthquake : (a) Displacement,
(b) Inter-story drift, (c) Floor Acceleration response.
there is an increase of floor accelerations at the locations where damper is installed as compared to
uncontrolled case. It is due to the increase in stiffness of floor at the location of damper leading to
increased acceleration. Higher the control force, higher will be the stiffness. Hence at those floors
the acceleration due to amplified (scissor jack) bracing is more than their respective unamplified
(chevron) bracing configurations.
5.5 Semiactive Direct Control of Inter-Storey Displacement
In this section, instead of state control of structure an attempt has been made to control inter-storey
displacement. Since the control force due to MR damper is a function of relative displacements
between the two floors, cost function for LQR control chosen is a compromise between controller
force and relative displacement as well as relative velocities between the floors. Earlier only on-off
approach has been used to obtain the control voltages, in which the voltages can be set to either
115
Table 5.6: Percentage reduction in displacement response

Case 1 Case 2 Case 3 Case 4 Case 5 case 6
ω = ω1
1st F loor 79.09 47.61 64.12 37.26 44.09 23.13
2nd F loor 81.90 48.57 64.88 37.78 43.92 23.07
3rd F loor 81.21 49.68 66.05 38.73 43.81 23.09
th
4 F loor 82.46 49.97 68.48 38.90 44.34 23.60
5th F loor 82.80 49.62 68.66 39.03 45.03 23.34
th
6 F loor 82.70 49.45 69.04 39.21 44.46 22.97
7th F loor 82.59 49.31 69.15 39.20 44.20 22.90
Roof 82.53 49.23 69.07 39.14 44.26 23.04
ω = ω2
1st F loor 70.52 48.66 21.96 14.54 38.35 27.22
nd
2 F loor 59.15 37.24 20.69 13.17 34.79 25.93
3rd F loor 47.23 31.58 20.71 13.85 27.85 21.94
4th F loor 27.94 21.66 18.89 12.70 10.90 11.64
th
5 F loor 12.17 7.26 6.67 4.19 -3.97 -2.70
6th F loor 8.74 5.07 4.79 3.06 3.10 2.26
th
7 F loor 6.57 3.81 3.73 2.46 3.36 2.35
Roof 7.41 5.27 5.34 4.21 5.16 4.25
ω = ω3
1st F loor 53.20 30.51 7.71 7.93 6.73 7.71
2nd F loor 23.23 11.08 -0.75 0.31 1.69 1.69
rd
3 F loor 11.37 4.16 -6.55 -3.74 -0.05 0.00
4th F loor 8.91 3.82 6.70 3.90 -0.32 -0.12
5th F loor 7.34 4.13 5.22 3.63 -0.49 0.28
6th F loor 3.22 1.46 2.17 1.25 1.93 1.29
7th F loor 1.65 0.73 1.15 0.63 1.02 0.66
Roof 3.43 2.96 3.28 2.93 2.10 1.30
El-centro
1st F loor 37.75 22.27 15.30 7.16 -2.33 -0.98
2nd F loor 40.28 23.26 16.88 7.95 -0.09 -0.18
3rd F loor 44.08 21.72 19.90 9.28 2.95 1.52
th
4 F loor 43.89 21.01 31.23 15.99 6.42 3.33
5th F loor 43.57 21.11 33.58 17.05 11.51 6.15
th
6 F loor 44.26 22.01 35.03 17.53 15.89 -8.83
7th F loor 44.94 22.95 35.59 17.63 17.41 9.18
Roof 45.44 23.40 35.88 17.73 18.23 9.95
116
Table 5.7: Percentage reduction in interstory drift

ω = ω1
1st F loor 79.09 47.61 64.12 37.26 44.09 23.13
2nd F loor 80.21 49.22 65.54 38.24 43.75 23.00
3rd F loor 80.88 48.93 67.26 38.71 43.56 23.13
th
4 F loor 78.80 48.32 66.86 37.12 45.18 24.14
5th F loor 77.48 48.45 65.88 38.77 39.14 21.09
th
6 F loor 76.27 48.10 68.33 38.49 39.29 20.00
7th F loor 74.65 47.11 67.25 36.92 29.86 17.48
Roof 72.73 46.36 66.14 36.14 18.18 13.86
ω = ω2
1st F loor 70.52 48.66 21.96 14.54 38.35 27.22
nd
2 F loor 27.61 22.32 19.29 11.62 29.59 23.51
3rd F loor 0.39 1.96 -8.43 -4.90 -3.73 -2.35
4th F loor 5.59 6.36 9.63 8.29 -0.77 -1.54
th
5 F loor 35.73 25.14 25.41 17.93 35.05 28.80
6th F loor 39.61 27.05 24.40 14.73 39.01 25.12
th
7 F loor 39.28 28.27 19.92 11.28 43.45 32.59
Roof 39.04 28.43 17.83 10.12 48.43 34.22
ω = ω3
1st F loor 53.20 30.51 7.71 7.93 6.73 7.71
2nd F loor -4.74 -2.62 -5.99 -3.37 0.00 0.00
rd
3 F loor 8.02 8.14 -2.43 -1.22 6.68 3.52
4th F loor 10.19 8.65 22.87 19.91 11.26 9.60
5th F loor 1.72 2.73 10.20 6.61 3.30 2.59
6th F loor 1.68 2.90 14.20 7.94 7.79 3.66
7th F loor 1.61 3.04 18.43 11.45 3.58 2.15
Roof 7.92 7.92 29.82 18.47 -2.64 -3.17
El-Centro
1st F loor 37.75 22.27 15.30 7.16 -2.33 -0.98
2nd F loor 38.30 17.67 18.83 8.97 2.33 1.17
3rd F loor 39.14 22.65 26.65 13.23 8.57 4.19
4th F loor 40.42 20.16 30.98 14.80 19.27 10.13
5th F loor 42.43 23.43 30.39 16.80 11.49 7.18
th
6 F loor 38.58 25.77 35.38 20.96 14.02 8.95
7th F loor 34.99 26.15 38.12 21.18 -2.58 1.47
Roof 32.87 25.52 38.11 19.93 -18.18 -5.94
117
Table 5.8: Percentage reduction in floor acceleration

ω = ω1
1st F loor -60.00 -58.82 2.35 2.35 -7.06 14.12
2nd F loor 23.78 5.49 13.41 2.44 13.41 15.85
3rd F loor 52.32 33.76 8.44 -14.35 24.89 13.08
th
4 F loor 68.98 44.22 32.67 8.58 9.57 7.26
5th F loor 77.72 45.13 47.91 22.84 -18.11 -10.03
6th F loor 76.37 48.26 55.97 33.58 1.99 0.25
7th F loor 75.46 47.92 66.20 37.73 22.22 11.57
Roof 72.71 46.09 65.77 36.02 16.78 13.20
ω = ω2
1st F loor -40.96 15.87 8.86 -0.74 26.20 22.51
nd
2 F loor 25.43 26.42 11.36 9.88 39.26 28.89
3rd F loor 36.28 25.62 -10.20 -4.76 55.10 36.05
4th F loor 34.26 28.13 -23.68 4.74 31.48 22.01
th
5 F loor 19.58 22.75 -0.53 6.88 -55.03 -27.51
6th F loor 3.08 6.15 -4.62 -3.85 -80.77 -53.85
7th F loor 38.26 27.65 18.65 9.97 22.51 14.47
Roof 38.97 28.64 17.61 10.09 47.89 33.57
ω = ω3
st
1 F loor -26.58 -7.59 9.49 6.12 9.28 6.75
2nd F loor 0.65 2.61 29.19 15.69 16.99 10.89
3rd F loor -13.38 -8.92 -53.18 -42.04 -19.43 -10.51
4th F loor -8.94 -5.59 -30.45 -3.91 2.23 7.26
th
5 F loor 8.08 6.70 34.64 20.32 6.00 6.93
6th F loor 4.97 6.83 28.26 14.29 -23.91 -5.59
7th F loor 0.40 1.61 22.09 12.85 -5.22 -4.82
Roof 7.87 8.12 30.71 18.78 -3.05 -3.81
El-Centro
1st F loor -90.91 -47.11 -39.67 -8.26 -28.93 -15.70
2nd F loor -5.30 8.61 -2.65 9.93 -13.25 33.77
3rd F loor 7.60 8.77 -61.99 -43.27 -20.47 -5.85
4th F loor 16.85 12.92 -44.94 -12.92 -50.56 -19.10
5th F loor 28.19 6.38 -11.70 5.32 -81.38 -62.77
th
6 F loor 37.27 15.45 15.00 14.09 -41.36 -34.09
7th F loor 37.02 27.48 32.82 21.76 -3.82 -6.49
Roof 32.30 24.74 37.80 19.59 -21.31 -7.90
118
maximum or minimum. In this method, a continuous variation of current between minimum and
maximum value has been used to obtain damper force close to optimal control force (obtained
through LQR algorithm ) at each time step. One 1000 kN MR damper is installed at first floor
of 8-storey shear building and Bouc-Wen model is used to obtain the non-linear damper force.
Amplified ( Scissor jack ) bracing with an amplification factor of 2 (a0 = 2) has been used for
damper installation. Three earthquake ground motion data have been used. First ground motion
is scaled LA02 obtained from SAC ground motion data with a scale factor of 0.4 (GM1). Second
ground motion is El-centro with a scale factor of 0.8 (GM2). Third is also El-centro with a scale
factor of 0.4 (GM3). GM2 and GM3 ground motions are chosen to reflect the effect of PGA on
control performance.
5.5.1 Formulation
The state equation for semiactive control is given by
{ṡ(t)} = [A]{s(t)} + {B}F (t) + {E}x¨g (t) (5.18)
where,
T
{s(t)} = x1 (t) x2 (t) . . . x8 (t) x˙1 (t) x˙2 (t) . . . x˙8 (t) (5.19)
In above equation, all the parameters have been described in previous sections. Control force
obtained through LQR control at each time step is obtained by minimizing the modified cost
function J, which is given by
Z tf
1
J= [{y(t)}T [Q]{y(t)} + Fc T (t)RFc (t)]dt (5.20)
2 0
where, {y(t)} is a vector of size (16 × 1), consisting of inter-storey displacements and velocities.
119
Earlier in C-O approach state control has been used.
 
 
x1 (t)

 


 


 


 

(x2 (t) − x1 (t))

 


 


 

..

 

 


 . 



 


 

 
 (x8 (t) − x7 (t))
 

{y(t)} = (5.21)
 
(x˙1 (t))

 


 


 


 

(x˙2 (t) − x˙1 (t))

 


 


 

..

 

 



 . 




 


 

 (x˙8 (t) − x˙7 (t))
 

(16×1)
By expressing vector {y(t)} (Equation 5.21) in terms of state vector {s(t)} (Equation 5.19), fol-
lowing relation is obtained.
{y(t)} = [Cd ]{s(t)} (5.22)
where [Cd ] is (16 × 16) matrix given by
 
 1 0 0 ··· 0 0 0 ··· 0 
0
 
 

 −1 1 0 0 0 0 0 0 

 
 
0 −1 1 0 0 0 0 0 
 

 
.. . . .. .. . . .. 
 
. . . . 

 . .
 
 
 
 0 0 0 ··· 1 0 0 0 ··· 0 
[Cd ] =  (5.23)
 

 

 0 0 0 ··· 0 1 0 0 ··· 0 
 
 

 0 0 0 0 −1 1 0 0 

 
 

 0 0 0 0 0 −1 1 0 

 
 .. . . .. .. . . .. 

 . . . . . . 

 
 
0 0 0 ··· 0 0 0 0 ··· 1
120
Substituting {y(t)} from Equation 5.22 to 5.20, cost function becomes
Z tf
J= [{s(t)}T ([Cd ]T [Q][Cd ]){s(t)} + Fc T (t)RFc (t)]dt (5.24)
0
[Q] is a weight matrix corresponding to inter-storey drifts and velocities. Control force can be
obtained from the following expression.
Fc (t) = −R−1 {B}T [P ]s(t) (5.25)
where, [P ] is a ricatti matrix obtained by solving the following ARE equation.
[P ][A] + [A]T [P ] − [P ]{B}R−1 {B}T [P ] + [Cd ]T [Q][Cd ] = 0 (5.26)
The equation of motion for semiactive control for this case is same as obtained in Chapter 4.
 
 [A]{s(t)} + {B}F (t) + {E}x¨g (t) 
 
 
{Ṡ(t)} = 
 a0 (−γ |x˙1 (t)| z(t) |z(t)|
(n−1)
− β x˙1 (t) |z(t)|n + Ax˙1 (t)) 
 (5.27)
 
 
−η(u(t) − v(t))
where, the state vector is
T T
{S(t)} = s(t) z(t) u(t) = x1 (t) · · · x8 (t) x˙1 (t) · · · x˙8 (t) z(t) u(t)
(5.28)
However, for direct control, applied voltage v(t) is not set to only maximum or minimum value.
Applied voltage is continuously varying between maximum and minimum range. If u(t) is the
voltage obtained in the damper, then damper force can be written as
F (t) = a0 [a0 (c0a + c0a u(t))x˙1 (t) + a0 (k0a + k0b u(t))x1 (t) − k0 x0 + (αa + αb u(t))z(t)] (5.29)
121
All the parameters of 1000 kN damper Bouc-Wen model are mentioned in Chapter 2. Voltage
produced in MR damper u(t) is obtained by equating the control force obtained from Equation
5.25 to the MR damper force obtained in Equation 5.29.
(Fc (t)/a0 ) − (a0 k0a x1 (t) − k0 x0 + a0 c0a x˙1 (t) + αa z(t))

u(t) = (5.30)
(a0 k0b x1 (t) + a0 c0b x˙1 (t) + αb z(t))
Also mentioned in Chapter 2, relation between input voltage u(t) and output voltage u(t) is given
by
˙ = −η(u(t) − v(t))
u(t) (5.31)
Hence, if u0 is the damper obtained voltage at the beginning of each time step (at t = (t − ∆t))
and assuming ∆t (difference between two time steps) to be small, input voltage can be obtained
by following relation
u(t) − u0 e−η∆t
v(t) = (5.32)
1 − e−η∆t
Since the input voltage should lie in domain [0 Vmax ], optimal voltage applied at each time step is
given by 




 0 if v ≤ 0


v(t) = u(t)−u0 e−η∆t (5.33)
 1−e−η∆t
if 0 ≤ v ≤ Vmax




 Vmax if v ≥ Vmax

5.5.2 Results and Discussions
To obtain the robustness of direct Control, controlled response of 8 storey building obtained from
direct controller is compared to clipped optimal controller (C-O) for all the three ground motions.
For the LQR clipped optimal control [Q] matrix is (16 × 16) identity matrix and R is 10-15 .
For inter-storey Direct control approach, same weight is given to all the inter-storey drifts and
velocities i.e. (16 × 16) identity matrix for [Q]. For direct control weight factor R is chosen in
such a way, so that the maximum optimal control force Fc obtained in this case becomes close to
122

FlOOR 1 2 3 4 5 6 7 8
Dispalcement(cm)
Uncontrolled 2.694 5.353 7.887 10.235 12.278 13.891 15.069 15.674
C-O 1.973 3.826 5.703 7.448 8.925 10.102 10.931 11.382
Reduction(%) (26.76) (28.53) (27.69) (27.23) (27.31) (27.28) (27.46) (27.38)
Direct 2.077 4.132 6.012 7.651 9.058 10.200 11.006 11.425
Reduction(%) (22.90) (22.81) (23.77) (25.25) (26.23) (26.57) (26.96) (27.11)
Acceleration(g)
Uncontrolled 0.224 0.335 0.376 0.402 0.446 0.519 0.601 0.652
C-O 0.425 0.341 0.334 0.365 0.376 0.391 0.461 0.520
the maximum clipped optimal control force.
Reduction(%) -(89.73) -(1.79) (11.17) (9.20) (15.70) (24.66) (23.29) (20.25)

Direct 0.351 0.321 0.281 0.303 0.338 0.353 0.411 0.456
Reduction(%) -(56.70) (4.18) (25.27) (24.63) (24.22) (31.98) (31.61) (30.06)
Inter-Storey Drift (cm)
Uncontrolled 2.694 2.659 2.558 2.379 2.103 1.725 1.227 0.642
C-O 1.973 2.051 1.932 1.754 1.553 1.302 0.958 0.509
Reduction(%) (26.76) (22.87) (24.47) (26.27) (26.15) (24.52) (21.92) (20.72)
Direct 2.077 2.064 1.908 1.706 1.478 1.195 0.853 0.449
Reduction(%) (22.90) (22.38) (25.41) (28.29) (29.72) (30.72) (30.48) (30.06)
Damper force, FM R (kN)
C-O 2000 ‘
Direct 2000
Optimal force, Fc (kN)
C-O 37177.56
Direct 33456.88
reduction of top floor acceleration of C-O controller is 20.5% and for direct controller it is 30.06%
outperforms C-O controller for reducing absolute accelerations and inter-storey drift. Percentage
and Direct controller is 27.11% , which is not a very large variation. However direct controller
to uncontrolled case, percentage reduction of top floor displacement for C-O controller is 27.38%
slightly better as compared to Direct controller in reducing displacement response. As compared
From Table 5.9 (for GM1), it can be observed that, clipped optimal controller performed
Storey 1 2 3 4 5 6 7 8
Dispalcement(cm)
Uncontrolled 2.236 4.455 6.574 8.529 10.249 11.694 12.707 13.240
C-O 1.573 3.069 4.627 6.041 7.186 8.093 8.701 9.047
Reduction(%) (29.65) (31.11) (29.62) (29.17) (29.89) (30.79) (31.53) (31.67)
Direct 1.703 3.375 4.890 6.194 7.268 8.142 8.751 9.064
Reduction(%) (23.84) (24.24) (25.62) (27.38) (29.09) (30.37) (31.13) (31.54)
Acceleration(g)
Uncontrolled 0.242 0.301 0.341 0.356 0.376 0.439 0.525 0.582
C-O 0.441 0.266 0.293 0.319 0.327 0.324 0.370 0.431
Reduction(%) -(82.23) (11.63) (14.08) (10.39) (13.03) (26.2) (29.52) (25.95)
Direct 0.330 0.229 0.271 0.265 0.284 0.283 0.327 0.363
Reduction(%) -(36.36) (23.92) (20.53) (25.56) (24.47) (35.54) (37.71) (37.63)
Uncontrolled 2.236 2.229 2.147 2.014 1.810 1.499 1.085 0.571
C-O 1.573 1.698 1.584 1.418 1.242 1.039 0.779 0.419
Reduction(%) (29.65) (23.82) (26.22) (29.59) (31.38) (30.69) (28.2) (26.62)
Direct 1.703 1.695 1.554 1.369 1.174 0.951 0.675 0.355
Reduction(%) (23.84) (23.96) (27.62) (32.03) (35.14) (36.56) (37.79) (37.83)
C-O 2000
Direct 2000
C-O 35799.95
Direct 32447.18
123
124

Storey 1 2 3 4 5 6 7 8
Dispalcement(cm)
Uncontrolled 1.118 2.227 3.287 4.265 5.125 5.847 6.353 6.620
C-O 0.696 1.331 1.840 2.396 2.895 3.263 3.502 3.616
Reduction(%) (37.75) (40.23) (44.02) (43.82) (43.51) (44.19) (44.88) (45.38)
Direct 0.769 1.537 2.223 2.820 3.309 3.679 3.925 4.046
Reduction(%) (31.22) (30.98) (32.37) (33.88) (35.43) (37.08) (38.22) (38.88)
Acceleration(g)
Uncontrolled 0.121 0.151 0.171 0.178 0.188 0.220 0.262 0.291
C-O 0.231 0.159 0.158 0.148 0.135 0.138 0.165 0.197
Reduction(%) -(90.91) -(5.3) (7.6) (16.85) (28.19) (37.27) (37.02) (32.30)
Direct 0.198 0.147 0.133 0.123 0.126 0.132 0.160 0.185
Reduction(%) -(63.64) (2.65) (22.22) (30.9) (32.98) (40.00) (38.93) (36.43)
Uncontrolled 1.118 1.115 1.073 1.007 0.905 0.749 0.543 0.286
C-O 0.696 0.689 0.654 0.600 0.521 0.460 0.352 0.192
Reduction(%) (37.75) (38.21) (39.05) (40.42) (42.43) (38.58) (35.17) (32.87)
Direct 0.769 0.778 0.705 0.602 0.517 0.449 0.338 0.181
Reduction(%) (31.22) (30.22) (34.3) (40.22) (42.87) (40.05) (37.75) (36.71)
C-O 1920.10
Direct 1896.43
C-O 14219.80
Direct 13944.29
125
( 12.31% more as compared to C-O controller). For inter-storey drift, reduction of top floor re-
sponse for C-O control is 20.72% and for Direct control it is 30.06 (13.48% higher as compared
to C-O controller).
Similarly from Table 5.10 (for GM2), it can be observed that C-O performs better in con-
trolling displacement response as compared to direct controller. However, there is a very slight
percentage variation in the two control algorithms for top floor displacement response (31.67%
for C-O and 31.54% for direct Control as compared to uncontrolled case). Percentage reduction
of top floor acceleration for direct control is 15.8% more than C-O control (25.95% for C-O and
37.63% for direct Control as compared to uncontrolled case). Response reduction for top floor
inter-storey displacement for direct controller case is 15.3% better as compared to C-O (26.62%
for C-O and 37.83% for Direct Control as compared to uncontrolled case).
From Table 5.11 (for GM3), same trend of response has been obtained like the previous two
cases. Percentage reduction in response for direct controller is same as compared to GM2. How-
ever, there is slight improvement in response reduction for C-O control as compared to GM2.
Percentage reduction of top floor inter-storey displacement and acceleration in this case for direct
control is around 6% more as compared to C-O controller. Hence, response control for direct
control is consistent for both GM2 and GM3.
From all these three tables it can concluded that, acceleration response reduction for all the
floors and inter-storey drift response reduction of upper floors is much better for Direct control as
compared to C-O controller.
5.6 Summary
In this chapter, the optimal location of MR damper has been obtained using Modal Controllability
method. From the results it can be observed that performance of damper installed at the first
floor of structure is most effective when the structure is excited in the fundamental mode. Even for
second modal excitations and for El-centro earthquake, first floor location of damper is the optimal
126
location. Also direct LQR controller method has been modified to control inter-storey drifts and
it is observed that direct control outperforms C-O controller for inter-storey drift and acceleration
response control of structure for all the ground motions.

Chapter 6
Summary and Conclusions
6.1 Summary
In this study at first different bracings arrangements for the installation of Magneto-rheological
(MR) are explained. A numerical study has been conducted on a three-storey shear building. A
1000 kN MR damper has been installed at the first floor. Two different bracings arrangements are
used: chevron bracings (unamplified) and scissor jack bracing (amplified bracing) for the installa-
tion of damper. Response of both the bracings has been compared for different seismic excitations
using passive control of MR dampers. Further, the idea of amplification has also been extended to
semi-active control of MR dampers. Numerical formulation is done for various semiactive control
algorithms : clipped-optimal LQR algorithm, Algorithm based on lyapunov stability, decentralized
bang-bang control algorithm, and maximum energy dissipation algorithm. Controlled response of
three-storey structure has been compared using these algorithms for chevron and amplified brac-
ings. Effect of increased damper capacity has also been evaluated by installing two dampers of
capacity 1000 kN. Also, semi-active control response is compared to active control using LQR al-
gorithm, for which maximum optimal control force is close to the maximum capacity of dampers.
For a 8-storey shear building optimal location of MR damper has been obtained using modal con-
trollability method. Also, effectiveness of optimal locations is determined for seismic control, by
128
installing dampers at different locations using Clipped-Optimal LQR approach. Finally, a new
algorithm (direct control algorithm) has been developed for inter-storey drift control using LQR
approach, for continuous current/volatge variation. The effectiveness of proposed algorithm for
seismic control has been verified for a eight-storey shear building with damper installed at the
first floor in amplified bracing configuration. Comparative study has been conducted between the
response obtained by direct controller and clipped-optimal controller.
6.2 Conclusions
The following conclusions can be drawn from this study
1. For Passive control, the study revealed that larger effective control force can be generated
for amplified cases as compared to unamplified due to its response amplification mechanism
resulting in a better control performance. Hence lower voltage can be applied or even a lower
capacity damper can be very effective with scissor jack/amplified bracing configuration,
resulting in cost effectiveness.
2. For active control using LQR algorithm, effect of different weighing matrices has been
evaluated and it has been observed that by decreasing weight factor R, control force is
increasing and better control can be achieved. However, in some of the ground motions it
can be observed that saturation has achieved after a certain level and no further response
control can be observed by increasing the control force. Control performance is found to be
varying with ground motions.
3. For semiactive control using ON-OFF mechanism, it has been observed that the maximum
capacity of control force has become twice due to amplification factor of two for the case
of scissor jack bracings, resulting in better control performance as compared to chevron
bracings. It has been observed that he performance of different control algorithm depends
on ground motions selected. However, better control is observed for all the algorithms
129
for amplified case as compared to their unamplified counterpart. It has been observed that
LQR clipped optimal algorithms is providing better results as compared to most of the other
cases including active control. By increasing the number of dampers, better response control
has been obtained for clipped-optimal control as compared to other algorithms. Effect of
weighing matrices for LQR clipped-optimal control is found to be varying with damper
capacity and ground motions.
4. Optimal location of MR damper has been found using modal controllability method. It
has been observed that performance of damper installed at the first floor of structure is
most effective when the structure is excited in the fundamental mode. Even for the second
modal excitation and for El-centro earthquake, first floor location is optimal location of
damper. Also, performance of amplified bracings is found to be better than its corresponding
unamplified bracings.
5. Direct LQR controller method has been used to control inter-storey drifts. From the results
it can be concluded that the direct control outperforms the clipped-optimal controller for
control of inter-storey drift and acceleration response.
6.3 Future scope of the study
Future research areas include the following points
1. In this study, shear building model remaining in elastic range is used. Further study is
required for frames and for structures going in non-linear range.
2. In this study, linear amplification factor (relation between damper velocity and inter-storey
velocity is linear) has been used, since, it has negligible effect as compared to non-linear
velocity amplification factor (relation between damper velocity and inter-storey velocity is
non-linear) for structures remaining in elastic range. Non-linear velocity amplification can
130
be used as discussed by Lee et al. (2007), since, it is expected to become more evident for
inelastic structures.
3. Experimental study should be conducted to verify the effectiveness and limitations of toggle
bracings for semi-active control. Also, experimental study is required to verify the correct-
ness of proposed algorithm.
4. In this study, bracing is assumed to be very stiff. However, no bracing can be of infinite
stiffness. Hence, effect of bracing stiffness should be included in the expressions of ampli-
fication factors to obtain more accurate results.
5. In this study, only one damper is used. However, more number of dampers are required
for structures subjected to severe lateral loading. Location and number optimization should
be done. Also, control algorithm should be modified for multi-input semi-active control
agorithms.
6. In this study, full state feedback is used for LQR control algorithm. Algorithms can be
modified for acceleration feedback with the use of observers/estimators, as large number of
sensors, which is not economical, are required for measuring all the states .
7. Analysis should be carried out in a system of high precision due to very small weight factor
.
Bibliography
Abdel-Rohman, M. (1984). “Optimal control of tall buildings by appendages.” Journal of the
Structural Engineering, ASCE, 110, 937–947.
Abdel-Rohman, M. and Leipholz, H. (1978). “Active control of flexible structures.” Journal of the
Structural Division, ASCE, 104(8), 1251–1266.
Abdel-Rohman, M. and Leipholz, H. (1983). “Active control of tall buildings.” Journal of the
Structural Division, ASCE, 109(3), 628–645.
Abe, M. (1996). “Semi-active tuned mass dampers for seismic protection of civil structures.”
Earthquake Engineering and Structural Dynamics, 25, 743–749.
Adeli, H. and Cheng, N. (1994a). “Augmented lagrangian genetic algorithm for structural opti-
mization.” Journal of Aerospace Engineering, ASCE, 7(1), 104–118.
Adeli, H. and Cheng, N. (1994b). “Concurrent genetic algorithms for optimization of large struc-
tures.” Journal of Aerospace Engineering, ASCE, 7(3), 276–296.
Akbay, Z. and Aktan, H. (1990). “Intelligent energy dissipation systems.” Palm Springs, CA,
427–435.
Al-Bazi, A. and Dawood, N. (2010). “Developing crew allocation system for precast industry
using genetic algorithms.” Computer-Aided Civil and Infrastructure Engineering, 25(8), 581–
595.
132
B.F. Spencer Jr, S.J. Dyke, M. S. J. C. (1996). “Phenomenological model of a magnetorheological
damper.” ASCE Journal of Engineering Mechanics, 123(3), 230–238.
Brogan, W. (1991).” Modern Control Theory, Prentice Hall, Englewood Cliffs, New Jersey.
Carlson, J. and Jolly, M. (2000). “Mr fluid, foam and elastomer devices.” Mechatronics, 10, 555–
569.
Carlson, J. and Spencer, B. (1996). “Magneto-rheological fluid dampers for semi-active seismic
control.” Proceedings of the Third International Conference on Motion and Vibration Control,
Vol. 3, Chiba, Japan, 35–40.
Carneiro, R., Brito, J., and Avila, S. (2010). “Vibration control of structures using magnetorheo-
logical dampers.” 11th Pan-American Congress of Applied Mechanics.
Chang, J. and Soong, T. (1980). “Structural control using active tuned mass dampers.” Journal of
the Engineering Mechanics Division, ASCE, 106(6), 1091–1098.
Cheng, F. (1988). “Response control based on structural optimization and its combination with
active protection.” Proceedings of the ninth world conference on Earthquake Engineeering,
Vol. 8, Tokyo-Kyoto, Japan, 471–476.
Cheng, F. and Pantelides, C. (1988). “Optimal placement of actuators for structural control.”
NCEER Technical Report 88-0037, National Centre for Earthquake Engineering, Buffalo, New
York.
Cheng, T. and Yan, R. (2009). “Integrating messy genetic algorithms and simulation to optimize
resource utilization.” Computer-Aided Civil and Infrastructure Engineering, 24(6), 401–415.
Constantinou, M. C., Tsopelas, P., Hammel, W., and Sigaher, A. N. (1997). “Testing and modeling
of an improved damper configuration for stiff structural systems.” Technical Report Submitted
to the Center for Industrial Effectiveness and Taylor Devices, Inc.

133
Constantinou, M. C., Tsopelas, P., Hammel, W., and Sigaher, A. N. (2001). “Toggle-brace-damper
seismic energy dissipation systems.” Journal of Structural Engineering, 127(2), 105–112.
Dyke, S., Caicedo, J., Turan, G., Bergman, L., and Hague, S. (2003). “Phase 1 benchmark con-
trol problem for seismic response of cable-stayed bridges.” Journal of Structural Engineering,
129(7), 857–872.
Dyke, S., Spencer, B., Sain, M., and Carlson, J. (1996a). “Experimental verification of semi-active
structural control strategies using acceleration feedback.” Proceedings of Third International
Conference on Motion and Vibration Control, Vol. 3, Chiba, Japan, 291–296.
Dyke, S., Spencer, B., Sain, M., and Carlson, J. (1996b). “Modeling and control of magnetorheo-
logical dampers for seismic response reduction.” Smart Materials and Structures, 5, 576–575.
Dyke, S., Spencer, B., Sain, M., and Carlson, J. (1996c). “Seismic response reduction using
magnetorheological dampers.” Proceedings of the IFAC World Congress.
Dyke, S., Spencer, B., Sain, M., and Carlson, J. (1997a). “An experimental study of magne-
torheological dampers for seismic hazard mitigation.” Proceedings of Structures Congress XV,,
Portland, OR, 1358–1362.
Dyke, S., Spencer, B., Sain, M., and Carlson, J. (1997b). “On the efficacy of magnetorheologi-
cal dampers for seismic response reduction.” Proceedings of 1997 ASME Design Engineering
Technical Conferences, Sacramento, CA, Sacramento, CA.
Dyke, S., Spencer, B., J., Quast, P., Sain, M., J., D., and Soong, T. (1996d). “Acceleration feedback
control of mdof structures.” ASCE journal of Engineering mechanics, 122, 907–918.
Ehrgott, R. and Masri, S. (1992a). “Modeling the oscillatory dynamic behavior of electrorheolog-
ical materials in shear.” Smart Materials and Structures, 1(4), 275–285.
Ehrgott, R. and Masri, S. (1992b). “Use of electro-rheological materials in intelligent systems.”

134
Proceedings of U.S.ItalyJapan Workshop/Symposium on Structural Control and Intelligent Sys-
tems, Italy, 87–100.
Feng, M., Shinozuka, M., and Fujii, S. (1993). “Friction controllable sliding isolation system.”
Journal of Engineering Mechanics, 119(9), 1845–1864.
Gamota, D. and Filisko, F. (1991). “Dynamic mechanical studies of electrorheological materials:
Moderate frequencies.” Journal of Rheology, 35(3), 399–425.
Hiemenz, G., Choi, Y., and Wereley, N. (2003). “Seismic control of civil structures utilizing
semi-active mr braces.” Computer-Aided Civil and Infrastructure Engineering, 18(1), 31–44.
Hrovat, D., Barak, P., and Rabins, M. (1983). “Semi-active versus passive or active tuned mass
dampers for structural control.” Journal of Engineering Mechanics, 109(3), 691–705.
Hwang, J., Huang, Y., and Hung, Y. (2005). “Analytical and experimental study of toggle-brace-
damper systems.” Journal of Structural Engineering, ASCE, 131(7), 1035–1043.
Jansen, L. and Dyke, S. (2000). “Semi-active control strategies for mr dampers: A comparative
study.” ASCE Journal of Engineering Mechanics, 126(8), 795–803.
Jiang, X. and Adeli, H. (2008). “Neuro-genetic algorithm for nonlinear active control of highrise
buildings.” International Journal for Numerical Methods in Engineering, 75(8), 770–786.
Jung, H., Spencer, B., and Lee, I. (2003a). “Control of seismically excited cable-stayed bridge em-
ploying magnetorheological fluid dampers.” Journal of Structural Engineering, ASCE, 129(7),
873–883.
Jung, H., Spencer, B.F., J., and Lee, I. (2003b). “Control of seismically excited cable-stayed bridge
employing magnetorheological fluid dampers.” Journal of Structural Engineering, 129(7), 873–
883.
135
Kang, M., Schonfeld, P., and Yang, N. (2009). “Prescreening and repairing in a genetic algorithm
for highway alignment optimization.” Computer-Aided Civil and Infrastructure Engineering,
24(2), 109–119.
Kareem, A. (1994). “The next generation of tuned liquid dampers.” Proceedings of First World
Conference on Structural Control, Los Angeles, CA.
Kim, H. and Adeli, H. (2001). “Discrete cost optimization of composite floors using a floating
point genetic algorithm.” Engineering Optimization, 33(4), 401–415.
Kobori, T., Takahashi, M., Nasu, T., Niwa, N., and Ogasawara, K. (1993). “Seismic response con-
trolled structure with active variable stiffness system..” Earthquake Engineering and Structural
Dynamics, 22, 925–941.
Lee, S., Min, K., Chung, L., Lee, S., Lee, M., Hwang, J., Choi, S., and Lee, H. (2007). “Bracing
systems for installation of mr dampersin a building structure.” Journal of Intelligent Material
Systems and Structures, 18, 1111–1120.
Lee, Y. and Wei, C. (2010). “A computerized feature selection using genetic algorithms to forecast
freeway accident duration times.” Computer-Aided Civil and Infrastructure Engineering, 25(2),
581–595.
Leitmann, G. (1994). “Semiactive control for vibration attenuation.” Journal of Intelligent Mate-
rial Systems and Structures, 5, 841–846.
Liu, Y., Gordaninejad, F., Evrensel, C., Wang, X., and Hitchcock, G. (2005). “Comparative study
on vibration control of a scaled bridge using fail-safe magneto-rheological fluid dampers.” Jour-
nal of Structural Engineering, 137(5), 743–751.
Lou, J., Lutes, L., and Li, J. (1994). “Active tuned liquid damper for struc-tural control.” Proceed-
ings of First World Conference on Structural Control, Los Angeles, CA.

136
Lu, J., Thorp, J., Aubert, B., and Larson, L. (1994). “Optimal tendon configuration of a tendon
control system for a flexible structure.” AIAA Journal of Guidance, Control, and Dynamics,
17(1), 161–169.
Makris, N., Hill, D., Burton, S., and Jordan, M. (1995). “Electrorheological fluid damper for
seismic protection of structures.” Proceedings of Smart Structures and Materials, Vol. 2443,
San Diego, CA.
McClamroch, N. and Gavin, H. (1995). “Closed loop structural control using electrorheological
dampers.” Proceedings of the American Control Conference, Seattle, Washington, 4173–4177.
Narasimhan, S., Nagarajaiah, S., Gavin, H., and Johnson, E. (2002). “Benchmark problem for
control of base isolated buildings.” Proceedings of 15th Engineering Mechanics Conference,
ASCE, Reston, VA.
Nemir, D., Lin, Y., and Osegueda, R. (1994). “Semiactive motion control using variable stiffness.”
Journal of Structural Engineering, 120(4), 1291–1306.
Rahbari, N., Azar, B., Talatahari, S., and Safari, H. (2012). “Semi-active direct control method for
seismic alleviation of structures using mr dampers.” Structural Control and health Monitoring,
20, 1021–1042.
Rahbari, N., Azar, B., Talatahari, S., and Safari, H. (2013). “Semi-active direct control method for
seismic alleviation of structures using mr dampers.” Structural control and Health Monitoring,
20, 1021–1042.
Raju, K., Prasad, A., Muthumani, K., Gopalakrishnan, N., Iyer, N., and Lakshmanan, N.
(2013). “Experimental studies on use of toggle brace mechanism fitted with magnetorheo-
logical dampers for seismic performance enhancement of three-storey steel moment-resisting
frame model.” Structural Control and Health Monitoring, 20, 373–386.

137
Reinhorn, A., Soong, T., Lin, R., Wang, Y., Fukao, Y., Abe, H., and Nakai, M. (1989). “1:4
scale model studies of active tendon systems and active mass dampers for aseismic protection.”
Technical Report NCEER-89-0026.
Remirez, O., Constantinou, M., Kircher, C., Whittaker, A., Jonson, M., Gomez, J., and Chrysosto-
mou, C. (2000). “Development and evaluation of simplified procedures for analysis and design
of buildings with passive energy dissipation systems.” Technical Report MCEER-00-0010, Mul-
tidisciplinary Center for Earthquake Engineering Research (MCEER), University of New York
at Buffalo, NY.
Roorda, J. (1975). “Tendon control in tall buildings.” Journal of the Structural Division, 101(3),
505–521.
Sigaher, A. N. and Constantinou, M. C. (2003). “Scissor-jack-damper energy dissipation system.”
Earthquake Spectra, 19(1), 133–158.
Sigaher, A. N. and Constantinou, M. C. (2004). “Scissor-jack-damper energy dissipation system.”
Technical Report MCEER-04-0010.
Soong, T. and Skinner, G. (1981). “Experimental study of active structural control.” Journal of
Engineering Mechanics Division, 113(6), 1057–1067.
Spencer, B., Dyke, S., Sain, M., and Quast, P. (1993). “Acceleration feedback control strate-
gies for aseismic protection.” Proceedings of the American Control conference, San Fransisco,
California, 1317–1321.
Spencer, B., Suhardjo, J., and Sain, M. (1994). “Frequency domain optimal control strategies for
aseismic protection.” Journal of Engineering Mechanics, 20(1), 135–158.
Spencer, B.F., J., Carlson, J., Sain, M., and Yang, G. (1997). “On the current status of magnetorhe-
ological dampers: seismic protection of full-scale structures.” Proceedings of 1997 American
Control Conference, Albuquerque, NM, 458–462.

138
Spencer, B.F., J., Dyke, S., Sain, M., and Carlson, J. (1996). “Nonlinear identification of semi-
active control devices.” 11th ASCE Engrg. Mech. Spec. Conf., Ft. Lauderdale, Florida.
Stanway, R., Sproston, J., and N.G., S. (1985). “Non-linear identification of an electro-rheological
vibration damper.” IFAC Identification and System Parameter Estimation, 195–200.
Stanway, R., Sproston, J., and Stevens, N. (1987). “Non-linear modelling of an electro-rheological
vibration damper.” J. Electrostatics, Vol. 20, 20, 167–184.
Symans, M. and Constantinou, M. (1997a). “Experimental testing and analytical modeling of
semi-active fluid dampers for seismic protection.” Journal of Intelligent Material Systems and
Structures, 8(8), 644–657.
Symans, M. and Constantinou, M. (1997b). “Seismic testing of a building structure with a semi-
active fluid damper control system.” Earthquake Engineering and Structural Dynamics, 26(7),
759–777.
Symansa, M. and Constantinou, M. (1999). “Semi-active control systems for seismic protection
of structures: a state-of-the-art review.” Engineering Structures, 21, 469–487.
Taylor, D. and Tonawanda, N. (1999). “Toggle linkage seismic isolation structure.” Patent no.
5934028.
Wen, Y. (1976). “Method of random vibration of hysteretic systems.” Journal of Engineering
Mechanics Division, ASCE, 102(2), 249–263.
Xu, Y., Chen, J., Ng, C., and Qu, W. (2005). “Semi-active seismic response control of buildings
with podium structure.” Journal of Structural Engineering, 131(6), 890–899.
Yang, J., Wu, J., and Z., L. (1996). “Control of seismic excited buildings using active variable
stiffness systems.” Engineering Structures, 18(8), 589–596.

139
Yeesock, K., Reza, L., and Stefan, H. (2008). “Supervisory semiactive nonlinear control of
a building-magnetorheological damper system.” 2008 American Control Conference, Seattle,
WA, 2540–2545.
Yoshida, O. and Dyke, S. (2005). “Response control of full-scale irregular buildings using mag-
netorheological dampers.” Journal of Structural Engineering, 131(5), 734–742.

S R C O B S U M - R (MR) D: Eismic Esponse Ontrol F Raced Tructures Sing Agneto Heological Ampers

Uploaded by

Copyright:

Available Formats

S R C O B S U M - R (MR) D: Eismic Esponse Ontrol F Raced Tructures Sing Agneto Heological Ampers

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

S R C O B S U M - R (MR) D: Eismic Esponse Ontrol F Raced Tructures Sing Agneto Heological Ampers

Uploaded by

Copyright:

Available Formats

i

S EISMIC R ESPONSE C ONTROL O F B RACED S TRUCTURES

in Partial Fulfilment of the Requirements

for the Degree of

DEPARTMENT OF CIVIL ENGINEERING

INDIAN INSTITUTE OF TECHNOLOGY,KANPUR

July 10, 2015

controller for control of inter-storey drift and acceleration response.

the journey of my life.

guiding me throughout my work.

1.2 Control Strategy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

1.2.1 Passive control systems . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

1.2.2 Active control systems . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

1.2.3 Semiactive control systems . . . . . . . . . . . . . . . . . . . . . . . . . 13

1.3 Literature Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

1.3.1 Magneto-rheological dampers . . . . . . . . . . . . . . . . . . . . . . . 14

1.3.2 Amplified Bracings : Toggle Based Theory . . . . . . . . . . . . . . . . 21

1.5 Organization of thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

2 Passive Control of a 3-storey shear building using Magneto-rheological (MR) damper 28

2.1 Brief Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

2.2 State-Space formulation of n DOF shear building model . . . . . . . . . . . . . 29

2.2.1 Uncontrolled case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

2.2.2 Controlled case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

2.3 3-Storey Shear Building Model . . . . . . . . . . . . . . . . . . . . . . . . . . . 34

2.4 Magneto-rheological (MR) damper model . . . . . . . . . . . . . . . . . . . . . 36

2.5 Bracing Description (Amplified Bracing) . . . . . . . . . . . . . . . . . . . . . 40

2.5.1 Upper Toggle System . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

2.5.2 Lower Toggle System . . . . . . . . . . . . . . . . . . . . . . . . . . . 42

2.5.3 Reverse Toggle System . . . . . . . . . . . . . . . . . . . . . . . . . . . 43

2.5.4 Scissor Jack Toggle System . . . . . . . . . . . . . . . . . . . . . . . . 44

2.6 Equation of motion for Passive Case of MR damper . . . . . . . . . . . . . . . . 48

2.7 Ground motions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49

2.8 Results and Discussions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50

3 Control of a 3-storey shear building using LQR algorithm 61

3.1 Brief Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61

3.2 Active Control using LQR algorithm (Full State feedback) . . . . . . . . . . . . 61

3.2.1 Feedback Control laws . . . . . . . . . . . . . . . . . . . . . . . . . . . 62

3.2.2 Controllability and Observability . . . . . . . . . . . . . . . . . . . . . 66

3.2.3 Ricatti Optimal Active Control (ROAC)/Linear Quadratic Regulator (LQR)

3.2.4 Generalized Optimal Active Control (GOAC) Algorithm . . . . . . . . . 69

3.3 Results and Discussions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71

4 Semi-active Control of three story structure 80

4.1 Brief Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80

4.2 Semi-active control analysis of a 3-storey structure . . . . . . . . . . . . . . . . 80

4.3 Control Algorithms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82

4.3.1 Clipped Optimal approach . . . . . . . . . . . . . . . . . . . . . . . . . 82

4.3.2 Control Based on Lyapunov Stability theory . . . . . . . . . . . . . . . . 84

4.3.3 Decentralized Bang-Bang Control . . . . . . . . . . . . . . . . . . . . . 85

4.3.4 Maximum Energy Dissipation . . . . . . . . . . . . . . . . . . . . . . . 86

4.4 Results and Discussions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87

4.5 Effect of capacity of dampers . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99

5 Semi-active Control of eight storey shear Building 102

5.1 Brief Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102

5.2 Shear Building Model Description . . . . . . . . . . . . . . . . . . . . . . . . . 103

5.3 Modal Controllability for Location Optimization . . . . . . . . . . . . . . . . . 106

5.4 Analysis of 8 storey structure with different damper locations . . . . . . . . . . 109

5.4.1 Results and Discussions . . . . . . . . . . . . . . . . . . . . . . . . . . 111

5.5 Semiactive Direct Control of Inter-Storey Displacement . . . . . . . . . . . . . . 114

5.5.1 Formulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118