5 New Trends in Earthquake Engineering

CHAPTER 5

New Trends in Earthquake Engineering.

Passive and Active Control

Structural control for civil engineering appeared as a necessity imposed by special,

longer, or taller constructions spread allover the world. The passive approach was

adopted already in many cases. For active control there are experiments showing

good results. Hybrid solutions are under investigations by many researchers.

However, the field is very large and civil engineer designers do not have a

clear image of this scope.

This chapter is showing the aim of structural control. It follows a short image

of passive control and it is then stressing on active control. Theoretical aspects,

devices used, and practical applications of active control are presented.

A critical comparison between the two types of structural control is intended

to introduce the reader into the complexity of the problems involved by the control

implementations. Computational means helping the study of the field are also

reviewed.

5.1 INTRODUCTION

The Civil Engineering field is now undergoing important changes in philosophy and

practice due to dramatic evolution recorded in other branches of human activity like:

electronics, automatics, computer science, robotics, new materials and technologies,

etc. At the same time, it should be mentioned major differences between Civil

Engineering and other engineering fields.

It is specific for constructions to use materials with high costs, on large

surfaces and volumes, and to need huge energy during construction, especially in the

case of long span bridges and very tall buildings.

Life of many people and vital social activities depend on the well functioning

of civil structures during and after important earth shakings. There are also civil

engineering structures with inestimable material and spiritual values, as historical

buildings.

NATURAL ACTIONS STRUCTURES

PROTECTION

Passive Control

Active Control

Typhoons

Earthquakes

Tall

Buildings

Long Span

Bridges0 10 20 30 40 50 60

-800

-600

-400

-200

0

200

400

600KOBE NS 1995

818 gal

acce

lera

tion

(ga

l)

time (s)

Figure 5.1 Relation between actions, structures, and control

Protection of some structures as those reminded above is a very important problem for

human communities. A solution for avoiding the harming effects of strong

earthquakes or strong winds is the structural control. It is using specific means and

procedures that lead to reduction in intensities for actions and change the way they act

on civil engineering structures. It changes also the structural response to the actions

and lowers the induced energy. In Figure 5.1, the existing relation between natural

actions, structural control, and built constructions is shown. This figure highlights the

reduction in the input due to control means.

5.2 SEISMIC RESPONSE CONTROL

In order to determine the ways to control the structures acted by seismic forces, the

equation of energy balance may be written

E E E E Ek s h d (5.1)

where: E is the energy induced by seismic shaking; Ek is the kinetic energy; Es is the

elastic strain energy; Eh is the energy dissipated by the structural system due to

inelastic behavior or other causes; Ed is the energy dissipated by supplemental

damping devices.

From Equation (5.1) it can be deduced that the control means play the role of

taking a part of the energy induced into the structure. However, it should be observed

that there is a dependency between the structural characteristics, e.g. the natural

period of vibration, and the amount of induced seismic energy. Therefore, the seismic

behavior of civil engineering structures is very complex in its nature and its

complexity is amplified through the presence of control means, which are actually

modifying the structural characteristics.

There are many ways to accomplish a seismic response control. According to

Professor Kobori, these ways are:

1. Cut off the energy transmission of the earthquake ground motion to a

structure.

2. Isolate the natural period of the structure from the predominant frequency

domain of the ground motion.

3. Achieve the non-stationary and non-resonant state by providing nonlinear

characteristics.

4. Apply control force such as mass damper/driver or tendon.

5. Utilize the energy absorption mechanism.

If the first method is achievable, then the second and the fifth method are not needed.

However, the first method is difficult to achieve, therefore it is recommended that at

least two of the above methods to be combined.

Table 5.1 Classification of structural protective systems

Seismic

Isolation

Passive Energy

Dissipation

Active

Control

Elastomeric bearings Metallic dampers Active bracing systems

Lead rubber bearings Friction dampers Active mass dampers

Elastomeric bearings with energy

dissipating devices

Viscoelastic dampers Active variable stiffness or

damping systems

Sliding friction pendulum Viscous dampers Pulse systems

Flat sliding bearings with restoring

force devices

Tuned mass dampers Electro-rheological active

dampers

Lubricated sliding bearings with

energy dissipating devices

Tuned liquid dampers Aerodynamic appendages

Table 5.1 shows a classification of protective systems. It provides a part of the many

control devices existing now. Depending on the location of the control devices other

classification can be stated:

a) protective devices located into the structure

b) protective devices located at the base of the structure

c) combination of a) and b).

Each of the above can be either passive or active.

The passive control is based on the nature of the materials they include

(rubber, lead, steel, viscous materials) and especially on supplementary adding in

damping and ductility. Some of passive devices rely on the use of friction forces.

The active control is using external energy for reducing, even minimizing the

seismic response. The controlled structure becomes active, with different

characteristics and different behavior compared to the initial non-controlled one, even

in case of strong actions.

5.3 PASSIVE CONTROL. DEVICES AND PRACTICAL APPLICATIONS

Passive controlled structures are widely spread all over the world. Table 4.3, at the

end of this chapter presents a list of isolated structures and isolation devices used in

different countries.

Also at the end of the chapter there is the Table 4.4 showing some passive

devices used in Japan for seismic isolation of bridges. These devices are the objects of

specifications in a design guide named “Menshin”. Other developed countries try to

setup similar specifications.

Top steel plate

Steel plates

Rubber

Holes for bolts

Bottom steel plate

Figure 5.2 Elastomeric bearing

A rubber bearing with top and bottom steel plates is presented in Figure 5.2. Other

type is that from Figure 5.3. It contains steel plates integrated in rubber. These plates

have the role to limit lateral displacements. The system includes a lead kernel, which

can assure an increased ductility, but it can determine permanent displacements after

strong earthquakes. However, the lead core can be replaced when needed.

Dowel holes

Lead plug

Rubber

Steel plates

End steel plates

Figure 5.3 Lead-rubber bearing

The passive means devices range is considerable large. One important type, named

Tuned Mass Damper (TMD) is based on moving masses. Figure 5.8 shows an Active

Mass Damper. If the active part (the actuator which generates the force ua) is

removed, a TMD is obtained. Such systems determine changes in structural dynamic

characteristics. Especially the frequency response function is lowered if the dynamic

properties of the TMD are close to that of the main structure and if the mass of the

TMD is enough large (between 1% to 4% from the mass of the main system).

In a similar manner is acting a TLD, Tuned Liquid Damper. One TLD is

shown at the fourth level of the structure from Figure 5.9. The mass value, the natural

period of vibration, and the liquid viscosity are the main elements that make the

system to be efficient.

Articulated slider

Supporting

column

Bearing material

Spherical surface

Cylinder

Seal

Figure 5.4 Friction Pendulum System Bearing

Figure 5.4 presents other passive control system based on a sliding bearing moving on

a spherical surface. An important advantage of this device type is that it can assure a

return to the initial position under the structure’s weight.

5.4 ACTIVE CONTROL. THEORETICAL ASPECTS

Every construction suffers changes during its life. At the same time, the environment

where the structure is placed is changing, too. Therefore one could compare existing

constructions to living beings. However, the most common way a civil engineering

structure overcomes external loads is to resist to them. The living beings not only

resist but also adapt to the environmental aggressiveness, responding in a different

manner to different actions or intensities. Adapting to external loads and to structural

changes is a basic idea in active structural control.

In 1972, Prof. James T.P. Yao, through his paper “Concept of Structural

Control”, is defining the start for this new branch in structural synthesis. Figure 5.5

shows a feedback system as J.P.Yao viewed it. The author describes a structural

controlled structure as an error-activated structural system the behavior which varies

automatically in accordance with unpredictable variations in the loading as well as

environmental conditions and thereby produces desirable responses under all possible

loading conditions.

From the point of view of theoretical studies and application methodologies,

there are two main approaches in structural control:

i. LQG (Linear Quadratic Gain) control, based on time domain

ii. H and -Synthesis control, based on frequency domain.

These two ways are very developed in many sub-methods and versions. Meanwhile,

additional tools are added to the main methods: fuzzy sets analysis and neuronal

networks.

Input, r variable, c

Control

Actuating signal, e

Error or

Forward Path Elements

- error sensing device

- compensating network

- amplifier

- servo-motor-

+

Feedback

Path

Elements

Command or

Reference

Figure 5.5 Closed-Loop Control System

From the first category of control, LQG, very popular are: pole assignment, optimal

control, instantaneous optimal control, modal control, critical modal control, and

sliding modes control method.

Majority of these methods is based on rewriting the structural dynamics

classical and familiar system of equations

M z + C z + K z = fs s s (5.2)

in the form of state equation

x = Ax + Bf

y = Cx + Df (5.3)

In the Equation (5.2), Ms, Cs, and Ks are the mass, damping, and stiffness matrices of

the structure; z is the vector of the generalized displacement vector, and f is the vector

of the external forces.

A

C B+

+ xf x

+

+ y

D

Figure 5.6 System described by Equation (5.3)

In the Equation (5.3), A is the system matrix, B is the load location matrix, C is the

measurement matrix, and D is a matrix showing the influence of the input, f, to the

output, y. Equation (5.3) is described by Figure 5.6.

If all the states, x, are known (measured), a state feedback, u, can be

introduced, and the control problem is reduced to finding a gain matrix, K, so that:

x = Ax + B(f - u)

u = Kx (5.4)

and a graphical representation of the Equation (5.4) is given by Figure 5.7.

A

Bx

u

f x

+

+

-

+

K

Figure 5.7 System described by Equation (5.4)

As an example of control method, the optimal control method is presented next. For

this case, the objective is to determine the gain matrix, K, under the condition that a

quadratic index, J, defined by Equation (5.5) should be minimized.

J dtt f

x Qx u Ru0

(5.5)

In the Equation (5.5), Q and R are weighting matrices representing the importance

given to reducing the structural response and the importance given to use less external

energy for obtaining the control, respectively.

This method leads to a Riccati matrical equation

PA PBR B P A P 2Q 011

2 (5.6)

with the matrix P being unknown. After solving, the gain matrix is obtained from the

next equation:

K R B P11

2 (5.7)

Replacing the control force vector, u, into Equation (5.4) one could observe that the

feedback system is transforming the original uncontrolled system by changing the

system matrix A, such that it will respond to the requirements of Equation (5.5).

m2

k1

c1

ua

k2

c2

m1

x x x 2 2 2

x x x 1 1 1

xg

Figure 5.8 Active Tuned Mass Damper

5.5 ACTIVE CONTROL. DEVICES

In structural active control various devices are used. Between them, Active Tuned

Mass Dampers (ATMDs) are widely used and studied. A system with one degree of

freedom equipped with an ATMD is shown in Figure 5.8. It can be seen from this

figure that an ATMD is formed from an additional mobile mass attached to the system

through an actuator, generating the force ua. The active forces values are generated

from the on-line measurements and employing control algorithms.

In Figure 5.8 additional springs and dampers, tuned to dynamic characteristics

of the structure, link the main system and the secondary one. This control system is

also called Hybrid Mass Damper (HMD) because it can be seen as a combination

from a pure active system Active Mass Damper (AMD), and the elements of a passive

Tuned Mass Damper (TMD).

m Active Tuned Mass

Damper, ATMD

Tuned Liquid

Damper, TLD

Active Braces

System, ABS

Active Variable

Stiffness, AVS

Active Tendon

System, ATS

Active/passive base

isolation systems

Figure 5.9 Use of some structural control systems

In order to have a better look about passive and active control systems, Figure 5.9

shows some of them located on a building. The figure is only a representation, not a

true solution, because employing many device types in the same building is not

common. Usually, real applications involve only one type of control system.

However, for hybrid applications, the base isolation is utilized together with some

active devices.

At the top of the building from Figure 5.9 one can observe an ATMD device

and one floor lower, a Tuned Liquid Damper (TLD). At the third floor, there is an

Active Brace System (ABS) which is principally made from a piston and a servo-

valve acting on the diagonal of that floor.

The system from the second floor of the building from Figure 5.9 is named

Active Variable Stiffness (AVS) and was introduced by Professor Takuji Kobori and

Kajima Corporation of Japan. It is made from two very stiff inclined beams moved to

the left or to the right by the upper active piston and therefore minimizing the relative

floor displacement and changing the floor stiffness.

The first floor from the building in Figure 5.9 is equipped with an Active

Tendon System (ATS). This system is based on active diagonals consisting in tendons

having the role to provide the limitation of relative floor displacements.

At the base of the building from Figure 5.9 passive isolators are installed. In

order to prevent too long displacements of the building’s base, an actuator and,

eventually, a spring and a damper are employed. For the base actuators, the problem

of generating huge forces could be prohibitive.

Table 5.2 Buildings with structural active control in Japan

Active Control System's

Name

Developer Building's Name Year

AMD (Active Mass Driver) Kajima Kyobashi Seiwa Bldg. 1989

AVS (Active Variable System) Kajima KaTRI No.21 BLDG 1990

AMD Takenaka Sendagaya INTS 1992

AMD Takenaka Hankyu Chayamachi Bldg. 1992

HMD (Tuned Active Damper) MHI, Yasui A&E Kansai Airport Control Tower 1992

HMD (Hybrid Mass Damper) Shimizu ORC200 Symbol Tower 1992

HMD (DUOX) Kajima Ando Nishikicho Bldg 1993

HMD MHI Landmark Tower 1993

HMD Nikken, Prof. Fujita Long Term Credit Bank Bldg. 1993

HMD Takenaka KS Project 1993

HMD (TRIGON) Kajima, IHI Shinjuku Park Tower 1994

HMD MHI, Nihon Sekkei ACT Tower 1994

AMD (AVICS-1) Obayashi Riverside Sumida 1994

HMD MHI, Nikken Sekkei Osaka WTC Bldg. 1994

HMD Shimizu Hotel Ocean 45 1994

HMD (DUOX) Kajima, KRC Dowa Kasai Phoenix Tower 1995

HMD MHI, Nikken Sekkei Rinku Gate Tower 1995

HMD Fujita, Prof. Fujita Hirobe Miyake Bldg. 1995

5.6 ACTIVE CONTROL. PRACTICAL APPLICATIONS

Most of the practical active control implementations exist in Japan. Table 5.2 shows,

in chronological order, the Japanese buildings equipped with active devices, the name

of the company that developed them and the devices’ types.

As a practical example, in Figure 5.10 the ATMD used in Landmark Tower

Yokohama, the tallest Japanese building (296 m), can be seen. Two identical devices

with that from Figure 5.10 are installed on top of the building, at floor 70. The moving

mass of such device is 170 tones out from 250 tones of the ensemble (it should be

noted that the whole building mass is evaluated at 223,000 tones). Maximum

displacement of the mass is of 1.70 m. The device has the in-plane dimensions

4.90 4.90 m and a height of 9 m. It is expected an efficiency of 60-70% in

acceleration’s decrease during strong typhoons.

Figure 5.10 Hybrid Mass Damper

For protection against winds of the towers during construction some Japanese bridges

had been equipped with active/passive systems: Akashi-Kaikyo Bridge (the longest

world’s suspension bridge, with 1990 m central span) and Tsurumi Tsubasa Bridge

(the longest world’s one plane cable stayed bridge, with 510 m central span).

Figure 5.11 Active Variable Damper

An active system for bridges is under intensive studies at the Public Work Research

Institute Tsukuba, Japan. This device, shown in Figure 5.11 is to be placed between

the infrastructure and the superstructure of highway bridges for limiting the

displacements that could harm the functionality of those bridges during strong

earthquakes.

5.7 COMPARISONS BETWEEN THE TWO CONTROL CATEGORIES

There is no doubt that the active control systems can provide better results when

compared to the passive control. Following a step by step analysis, the displacements

at the top of a structure equipped with a TMD compared to them for the case of an

ATMD are plotted in Figure 5.12. A comparison in frequency domain for acceleration

for the same two cases may be seen in Figure 5.13.

Figure 5.12 Time-history displacement response

Next follows the main reasons for and against choosing passive or active control for

civil engineering structures:

passive control means are cheaper and easier to maintain. However in some

cases the number of needed devices can become extremely high. For example, in each

building of The World Trade Center in New York, USA, 10,000 viscoelastic dampers

had been installed.

passive devices’ condition is or might be damaged by time, environment

and load intensity.

passive devices’ characteristics cannot be easily adjusted, as is the case with

active devices.

for the time being, active means have poor reliability: they still present

malfunctions and, on the other hand, it is hard to obtain for them high power supply

capable to properly work during strong earthquakes.

energy consumption for active systems, even in stand-by state, makes their

maintenance extremely expensive.

for both passive and, especially, active control systems, response time lag

leads to difficulties in strategy control analysis.

when structural control, either passive or active, is to be applied, a very

precise knowledge of the structure is needed. Therefore intensive and expensive

studies have to be done to identify it.

in order to perform an efficient active control, monitoring of the structural

response might lead to highly costing equipment (sensors, computers, etc.).

there is a danger of spillover phenomena (amplification of the response) in

structural active control as a result of poor structural identification, long time lag,

malfunctions in equipment, or due to extremely high external actions.

To the above aspects it should be added another one: the effectiveness of passive

control is still difficult to prove. An example is the stiffness to be chosen for the case

of isolators used as passive devices. If the isolators are too soft they might lead to very

large displacements. If they are too stiff, they will have reduced effects and will

transmit more energy.

Passive Control

Active Control

0 5 10 15 20 25 30-0.2

-0.15

-0.1

-0.05

0

0.05

0.1

0.15

0.2

Top Displacement, El-Centro NS

Time (s)

Dis

pla

cem

ent

(m)

Figure 5.13 Acceleration frequency response

Other problem, linked with the first one, is that if the passive devices can come back

to the initial position, after an earthquake. If some deformations are permanent, then

the system could become unmanageable.

However, the most worrying aspect is that of the contradictory responses

shown by isolators and passive controlled structures: to some earthquakes they

perform very well but in some other cases their effects are null or even amplifications

of the responses might occur. This can be explained by the narrow frequency limits of

the effectiveness for the passive control. Therefore, only the frequency response of the

isolators is shown to be very good. The non-stationary characteristics of the

earthquakes make, in many cases, the isolators behavior to be unexpected.

5.8 SPECIFIC COMPUTER PROGRAMS

Structural computational and seismic analysis methods for structures employing

control means cannot be the same as for common civil engineering constructions. It is

almost obvious that a simple spectral analysis, as is specified by majority of

Earthquake Engineering national codes, is not satisfactory.

For passive controlled structures especially, and also for many active

controlled structures, it is necessary to adopt models that accept non-linearities for

materials and structures, large displacements, P- effect, soil-structure interaction,

different foundation condition, and asynchronous excitations at construction base.

Even if the step-by-step analysis remains a main analysis procedure, controlled

structures must be analyzed using other approaches as linear or non-linear stochastic

analysis, or frequency domain analysis. It is not possible to state that an analysis

method is superior to other analysis method. They just reveal different aspects of the

control system and complete the image of the structural response.

In order to perform analysis for passive controlled structures, majority of finite

element computer programs (ANSYS, ADINA, NASTRAN, ABAQUS, I-DEAS,

ALGOR, etc.) can be used.

Other category of programs is specialized in dealing with non-linearities for

civil engineering structures. For example, IDARC2D, offers a step-by-step analysis

that manipulates the degrading of structural elements, showing the mechanisms that

affects the structural behavior. Also it provides a push-over analysis. Including

Passive Control

Active Control

0 2 4 6 8 100

2

4

6

8

10

Frequency (Hz)

Res

ponse

(m

/s2)

FRF of acceleration

hysteretic damping devices, viscoelastic elements, and friction devices is possible.

Contribution from nonstructural elements can be taken into consideration. Similar

capabilities are offered by other available computer programs: DRAIN, SARCF,

ANSR, SAP2000, ETABS.

Commercial programs are available for solving active control problems. One

of them, MATLAB, is built around a kernel that offers an integrated environment for

various applications in mathematics, statistics, matrix computation, graphical

simulations, signal processing, neuronal network, nonlinear programming, control

systems, etc. MATLAB needs a relatively hard training to be learned and it is limited

to small structural control applications.

However, to obtain very efficient computation for active structural control,

one should think to use specialized computer programs that would intensively employ

all computer resources and to adapt to civil engineering problems. Though large

construction companies surely possess such kind of computer programs, commercial

software is not yet available.

5.9 CONCLUSIONS

Studying the structural active/passive control responses of civil engineering

structures, one can see that there is an obvious superiority of the active control

compared to passive control. Improvement of the structural response is observed in all

analysis types: time domain, frequency domain, stochastic or spectral.

However, active control is, for the time being, very expensive and unreliable.

In this case there are unsolved aspects linked to time delays and instability due to

control devices. Structural system identification is still a complicated task and

therefore the structural control, based on it, is negatively influenced.

For the moment, practical application of active control looks more likely to

generate more problems than the problems it solves. This does not mean that the

future will reject this idea. It is especially a technological matter that slows the

advances in this field. The need of longer/taller and safer bridges/buildings will surely

accelerate the theoretical and practical works of structural active control.

As a small example, one could remember what was happening only 30 years

ago compared with nowadays. At that time, a computer performing similar tasks that

are performed today by a workstation was maybe thousands times more expensive.

The cost for maintenance (mainly because of the need for special rooms highly

conditioned and because of their poor reliability) was huge. Today’s workstations

have prices close to the first personal computers and sooner will be affordable even

for home use. Their maintenance cost is almost zero and their reliability is

outstanding. It is somehow obvious that automatic systems for other fields of activity,

e.g. structural active control, will become more and more reliable at higher

performance/cost ratio.

Table 5.3 Base isolated structures allover the world

Country Type of Structure Number of Structures Type of Isolation Systems

Former Soviet

Buildings over 200 Sliding Bearings

Rocking Columns

Pile-in-sleeve systems

Union Bridges n.a. n.a.

Other structures n.a. n.a.

Buildings 6 Rubber Bearings

France Bridges n.a. n.a.

Other structures 2 Nuclear Power

Plants

Neoprene Bearings

Buildings 9 + several apartment

houses of the Italian

Navy

High Damping Rubber

Bearings

Neoprene Bearings

Italy

Bridges 156 (total length =

150 km)

Sliding Bearings

Rubber Bearings

Lead-rubber Bearings

Various Hysteretic

Damping Devices

Other structures ... ...

Japan

Buildings 67 Rubber Bearings & Energy

Dissipaters

Lead-rubber Bearings

High Damping

Rubber Bearings

Sliding Bearings

Bridges over 100 partially

isolated

Sliding Bearings

Lead-rubber Bearings

High Damping

Rubber Bearings

Other structures Radar Tower Sliding Bearings

Buildings 6 Lead-rubber Bearings

Pile-in-sleeve systems

Lead-rubber & Sliding

Bearings

New Zealand Bridges 37 Lead-rubber Bearings

Various Energy

Dissipating Devices

Other structures Industrial Chimney Rocking Foundation

Buildings 24 Lead-rubber Bearings

High Damping

Rubber Bearings

Friction Pendulum System

Springs & Viscodampers

United States Bridges 54 (total length = 11

km)

Lead-rubber Bearings

Sliding Bearings

Other structures 2 Tanks

Heavy Equipment

Friction Pendulum System

Lead-rubber Bearings

High Damping Bearings

Low Damping Bearings

Table 5.4 "Menshin" passive control devices

Type Menshin Device

High Damping

Rubber Bearing

Slide Friction

Rubber Bearing

Steel

Damper

Roller Menshin

Bearing

Viscous