Documente Academic
Documente Profesional
Documente Cultură
Structural Engineering
To Dottie, my wife
T. T. Soong
To Andrea, my wife
G. F. Dargush
Passive Energy Dissipation Systems in
Structural Engineering
T. T. Soong
G. F. Dargush
The traditional approach to seismic design, for example, has been based upon providing a
combination of strength and ductility to resist the imposed loads. For major earthquakes, the
structural design engineer relies upon the inherent ductility of conscientiously detailed
buildings to prevent catastrophic failure, while accepting a certain level of structural and
nonstructural damage.
In many ways this traditional approach views the transient seismic disturbance as an
equivalent static lateral load that must be resisted by the structure. However, by discarding
this notion and considering the actual dynamic characteristics of environmental disturbances,
dramatic improvements can be realized. In fact, as a result of this dynamical point of view,
many new and innovative concepts of structural protection have been advanced and are at
various stages of development, one of which is passive energy dissipation. The basic role of
passive energy dissipation devices when incorporated into a structure is to absorb or consume
a portion of the input energy, thereby reducing energy dissipation demand on primary
structural members and minimizing possible structural damage. In recent years, serious
efforts have been undertaken to develop the concept of passive energy dissipation or
supplemental damping into a workable technology, and a number of these devices have been
installed in structures throughout the world.
The present book represents an attempt to provide the first unified treatment for passive
energy dissipation systems by consolidating and synthesizing the available information. The
book is directed primarily toward the professional engineer faced with structural design
decisions and to
Contents
Preface ix
Acknowledgments xi
Chapter 1- Introduction 1
References 4
Chapter 2 - Fundamentals 5
2.1 Introduction 5
2.2 Dynamic Response of Simple Structural Models 6
2.2.1 General Formulations 6
2.2.2 Free Vibration 7
2.2.3 Forced Vibration 8
2.2.4 Transient Response 11
2.2.5 Response with Passive Damper 14
2.3 Dynamic Analysis of Structural Systems 18
2.3.1 General Formulation 18
2.3.2 Modal Superposition Method 20
2.3.3 Damping in Traditional and Passively Damped Structures 23
2.3.4 Direct Time Domain Analysis 24
2.4 EnergyFormulations 25
2.4.1 SDOF Structures 26
2.4.2 Energy-based Design 30
2.4.3 MDOF Structures 31
2.5 Concluding Remarks 32
References 33
The technique of seismic isolation is now widely used in many parts of the world. A
seismic isolation system is typically placed at the foundation of a structure. By means
of its flexibility and energy absorption capability, the isolation system partially reflects
and partially absorbs some of the earthquake input energy before this energy can be
transmitted to the structure. The net effect is a reduction of energy dissipation demand
on the structural system, resulting in an increase in its survivability. A detailed review
of seismic isolation technology is provided in the recent monograph by Skinner et al.
(1993).
On the other end of the spectrum as shown in Table 1.1 are semi-active and active
control systems. Semi-active and active structural control is an area of structural
protection in which the motion of a structure is controlled or modified by means of the
action of a control system through some external energy supply. However, semi-active
systems require only nominal amounts of energy to adjust their mechanical properties
and, unlike fully active systems, they cannot add energy to the structure. Considerable
attention has been paid to both semi-active and active structural control research in
recent years, with particular emphasis on the alleviation of wind and seismic response.
The technology is now at the stage where actual systems have been designed, fabricated
and installed in full-scale structures. The interested reader may consult Soong (1990)
and Soong and Constantinou (1994) for discussions on
many theoretical and practical aspects of semi-active and active structural control.
While all these technologies are likely to have an increasingly important role in
structural design, the focus of the present book is on passive energy dissipation (PED)
systems, displayed in the middle column of Table 1.1 and, to a limited extent, smart
materials. Research and development of passive energy dissipation devices for
structural applications have roughly a 25-year history. Similar to seismic isolation
technology, the basic function of passive energy dissipation devices when incorporated
into a structure is to absorb or consume a portion of the input energy, thereby reducing
energy dissipation demand on primary structural members and minimizing possible
structural damage. Unlike seismic isolation, however, these devices can be effective
against wind induced motions as well as those due to earthquakes. Contrary to semi-
active and active systems, there is no need for an external supply of power.
In recent years, serious efforts have been undertaken to develop the concept of energy
dissipation or supplemental damping into a workable technology, and a number of
these devices have been installed in structures throughout the world. This book is an
attempt to introduce the basic concepts of passive energy dissipation, and to consolidate
the available information concerning these devices, which is now scattered over many
sources including technical journals, conferences proceedings, laboratory reports, and
academic theses.
The book is organized to address a wide range of behavior characteristics associated
with passive energy dissipation devices. Each major device type, listed in the middle
column of Table 1.1, is presented and discussed in an individual chapter based upon the
following outline: (a) historical review; (b) basic principles; (c) mathematical
modeling, theory and experiments; (d) practical considerations, design and
implementation issues; and (e) structural applications. In Chapter 9, a brief account of
current activities in smart materials research as related to civil engineering structural
applications is presented. As shown in Table 1.1, these activities are considered to fall
within the domain of semi-active systems, an extension of passive systems with certain
adaptivity properties.
As mentioned above, in order to fully appreciate and realize the benefits of passive
energy dissipation technology, one must possess a good understanding of structural
dynamics. While a detailed presentation of that subject is certainly beyond the scope of
this book, an introductory account, emphasizing only the most pertinent aspects, is
provided in Chapter 2. Included is a discussion of the energy concept in structural
engineering, which provides a unifying principle for the theory of passive energy
dissipation and a central theme for our presentation in subsequent chapters.
Finally, it is pointed out that, since many references were used in the development of
this book, no attempt was made to unify the units of quantities used in the text. It was
felt that, to leave them in their original units, easier
references to the original publications could be made. For convenience, a conversion
table for English-unit to SI-unit conversion is provided in the Appendix.
References
Skinner, R. I., Robinson, W. H., and McVerry, G. H. (1993), An Introduction to
Seismic Isolation, Wiley, Chichester and New York.
Soong, T. T. (1990), Active Structural Control: Theory and Practice, Longman,
London and Wiley, New York.
Soong. T. T. and Constantinou, M. C., eds. (1994), Passive and Active Vibration
Control in Civil Engineering, Springer-Verlag, Wien and New York.
2
Fundamentals
2.1 Introduction
With the development and subsequent implementation of modern protective systems,
including those involving passive energy dissipation, the entire structural engineering
discipline is now undergoing major change, at least conceptually. The traditional
idealization of a building or bridge as a static entity is no longer adequate. Instead,
structures must be analyzed and designed by considering their dynamic behavior. It is
with this in mind that we present some basic concepts related to the subject of structural
dynamics. In this chapter, we concentrate on topics which are of primary importance in
understanding, analyzing, and designing structures that incorporate passive energy
dissipation systems.
In the following section, a simple single-degree-of-freedom structural model is
discussed. This represents the prototype for dynamic behavior. Particular emphasis is
given to the effect of damping. As we shall see, increased damping can significantly
reduce system response to time-varying disturbances. While this model is useful for
developing an understanding of dynamic behavior, it is not sufficient for representing
real structures. We must include more detail. Consequently, a multi-degree-of-freedom
model is then introduced, and several numerical procedures are outlined for general
dynamic analysis. A discussion comparing typical damping characteristics in traditional
and passively damped structures is also included. Finally, a thorough treatment of
energy formulations is provided. Essentially one can envision an environmental
disturbance as an injection of energy into a structure. Design then focuses on the
management of that energy. These energy concepts are particularly relevant in the
discussion of passively damped structures, and will provide a recurring theme
throughout the remainder of the book.
2.2 Dynamic Response of Simple Structural Models
2.2.1 General Formulation
In this section, we begin with the simplest model that exhibits some of the key response
characteristics of a structure subjected to dynamic loading. It is important to keep in
mind that here we are studying the behavior of the mathematical idealization, not that
of a physical structure. Much more detail is required to properly model an actual
building. However, a number of concepts can be most easily introduced by examining
these simple models. Our treatment was influenced by presentations included in
Newmark and Rosenblueth (1971) and Hanson (1993).
Consider the lateral motion of the basic single-degree-of-freedom (SDOF) model,
shown in Fig. 2.1, consisting of a mass m, supported by springs with total linear elastic
stiffness k, and a damper with linear viscosity c. This SDOF system is then subjected to
a seismic disturbance characterized by a spatially uniform, time-dependent ground
displacement xg(t), and to a wind load represented by p(t). The excited model responds
with a lateral displacement x(t) relative to the ground. Thus, the total displacement of
the idealized structure is simply
xt(t) = xg(t) + x(t) (2.1)
Both xg(t) and x(t) are assumed to be continuous, twice differentiable functions. By
definition, a linear elastic spring produces a force proportional to the displacement,
while the force in a linear viscous damper is proportional
Figure 2.1
SDOF Model
to the velocity. As a result, the equation of motion for this SDOF model can be written
Other quantities of interest, the natural frequency fo and the natural period To are related
via
where WD and represent the damped natural circular frequency and the
nondimensional damping ratio, respectively. More specifically,
Notice that for = 0, Eq. (2.6) reduces directly to Eq. (2.3). Actually, Eq. (2.6) is only
valid for underdamped systems, i.e., for < 1. However, that is exactly
the case of most interest to us. The response defined in Eq. (2.6) is plotted in Fig. 2.2
for several values of the damping ratio in order to illustrate the dramatic effect that
increased viscous damping can have on the response of the SDOF system. At = 0.01
the system oscillates through nearly 37 cycles before reducing its amplitude to 0.10x0,
while for = 0.20 less than 2 cycles are needed. In all cases, the damped systems
asymptotically return to their undeformed state as the energy initially stored in the
springs is dissipated through viscous action.
Figure 2.2
Free Vibration Response
The two terms appearing in Eq. (2.10) correspond to the steady-state and transient
components of the motion, respectively. The magnitude of the transient term
diminishes with time in the same manner as did the response in free vibration.
Consequently, if the harmonic load acts for a sufficiently long time, only the steady-
state term will have a significant contribution. We will now assume that is the case.
Then the displacement can also be written in terms of the complex frequency response
function . That is, at steady-state
where
Further examination of Eq. (2.13) reveals that the magnitude of the displacement
response x1 is not only time-independent, but can be written as the product of the static
response (xgt = p0/k) and a dynamic amplification factor, which depends only on the
frequency and damping ratios. Plots of the amplification factor (x1/xgt) versus at
various values of is provided in Fig. 2.3. Peak response occurs near resonance (i.e.,
= 1) and becomes unbounded for the undamped system. With = 0.01 the peak value
of the dynamic amplification factor is approximately 50, while for = 0.20 it is roughly
2.5. Once again viscous damping is seen to have a dramatic effect on the response
characteristics of the SDOF model. Note, however, that for forcing frequencies
removed from the resonance condition, viscous damping has much less influence on
response. For any level of damping, the static response is approached as 0, while
for large the amplification
Figure 2.3
Amplification Factor for Harmonic Loading
factor becomes negligibly small since the load varies too rapidly to invoke a
displacement response.
It is also instructive to consider the steady-state phase relationships. Figure 2.4
illustrates the dynamic balance of the forces associated with the applied load (p), inertia
(f1), stiffness (fs), and damping (fD) in the complex plane. Of course, the actual force at
any time t is the real component of the complex force shown in this Argand diagram.
For the damped system, the spring force, and consequently the displacement response,
lag behind the applied load by the phase angle 1. Meanwhile, the damping and inertia
forces, which are in turn proportional to the velocity and acceleration, lead the
displacement
Figure 2.4
Damper Force Balance for Harmonic Excitation
response by phase angles of 90 and 180, respectively. At resonance, 1 = 90 and the
applied loading is exactly balanced by the force in the damper.
Additional insight can be gained by examining the force-displacement curves as
illustrated in Fig. 2.5 for the spring and damper. The energy stored in the spring at any
time is equal to the area under the curve in Fig. 2.5a, while the energy dissipated via
viscous action per cycle is equivalent to the area within the elliptical hysteresis loop
shown in Fig. 2.5b. The elliptical shape is a direct result of the 90 phase difference
between the damper force and displacement response, along with the constant
amplitude nature of the motion. For a given structure, the area contained within the
hysteresis loop is a function of the frequency ratio, with a maximum occurring near
resonance. The specific case displayed in Fig. 2.5 has = 0.90 and = 0.05.
Figure 2.5
Force-displacement Response for Harmonic Excitation; a) Spring, b) Damper
where
The unit impulse response function h(t - ) defines the displacement at time t due to a
unit impulse force applied at time . For simple time dependence, the integral in Eq.
(2.15) can often be evaluated analytically. For more complicated loadings, the
evaluation can be performed numerically using, for example, an extended trapezoidal
rule or Romberg integration (Press et al., 1992). Unfortunately, the convolution
appearing in Eq. (2.15) requires increasing computational effort as time progresses. An
alternate approach involves the use of the Laplace transform, since in the transform
domain the convolution in Eq. (2.15) becomes simply a product. Thus, for transform
parameter s, we have
in which (s) and (s) are the Laplace transforms of f(t) and h(t), respectively, with the
latter given in explicit form in Eq. (2.14). The desired time domain solution x(t) is then
obtained by performing an inverse Laplace transform on (s). However, in practice
alternative numerical approaches, such as the Newmark-beta algorithm, are often
employed to solve the governing differential Eq. (2.2) directly (Bathe, 1982).
As an example, consider the SDOF model with period To = 1.0s (wo = 2 rad/s) and
damping ratio = 0.05, subjected to the 1940 El Centro S00E earthquake accelerogram
shown in Fig. 2.6. The relative displacement (x) and total acceleration response are
displayed in Figs. 2.7a and b, respectively. Meanwhile, the force-displacement curve
for the damper is provided in Fig. 2.8. (The force is shown per unit mass.) Notice that
the hysteresis loops that determine energy dissipated in the damper are no longer
simple ellipses, since there is no longer constant amplitude cycling. Eventually,
Figure 2.6
1940 El Centro S00E Accelerogram
Figure 2.7
1940 El Centro SDOF Time History Response; a) Displacement, b) Acceleration
Figure 2.8
1940 El Centro Force-displacement Damper Response
the loops collapse to a point at the origin as the linear structure returns to its
undisturbed configuration. Plots of the maximum relative displacement and total
acceleration response versus structure parameters To and are presented in Figs. 2.9a
and b, respectively. These response spectra clearly illustrate the beneficial effect of
increased viscous damping. At any particular structural period, an increase in viscous
damping tends to diminish the displacement response. Beyond a damping ratio of 50%,
however, very little additional benefit is realized.
Figure 2.9
1940 El Centro Spectral Response; a) Relative Displacement, b) Total Acceleration
Figure 2.10
SDOF Model with Passive Damper Element
passive device is written simply as x. This permits quite general response
characteristics, including displacement, velocity, or acceleration-dependent
contributions, as well as hereditary effects. The equation of motion for the extended
SDOF model then becomes
Figure 2.11
Force-Displacement Model for Elastic- Perfectly Plastic Passive Element
The passively damped SDOF model is first subjected to harmonic loading defined in
Eq. (2.8), and examined under steady-state conditions after all transients have
dissipated. Figure 2.12 presents the amplitude of the displacement response versus
forcing frequency. In order to clearly identify
Figure 2.12
Amplification Factors for Harmonic Excitation with Passive Element
the effect of the added passive damper, the normalization is accomplished in terms of
the natural frequency wo = (k/m)1/2 and static response xgt = p0/k of the base structure.
Thus, the uppermost curve in Fig. 2.12 is merely a reproduction of the = 0.05 curve
from Fig. 2.3. The remaining two curves illustrate the effect of adding a yielding device
to the system. It is evident that the peak response, near resonance, is greatly reduced
due to the energy dissipation characteristics of the yielding devices. It should be noted
that based upon energy considerations at resonance an effective viscous damping ratio
can be calculated for our passively damped SDOF model. However, the usefulness of
that linearization is limited, except for very preliminary performance estimates. In
general, it is best to analyze the actual nonlinear system. Force-displacement plots for
the linear spring, viscous damper, and passive damper are provided in Fig. 2.13 for the
specific case with = 0.90 and = 0.05 to permit comparison with results for the base
structure, which were plotted previously to the same scale in Fig. 2.5.
Figure 2.13
Force-displacement Response for Harmonic Excitation with Passive
Element; a) Spring, b) Viscous Damper, c) Passive Damper
Response is noticeably decreased. The area contained within the loops present in Figs.
2.13b and c measures the energy dissipated per cycle in the viscous damper and passive
damper, respectively. For the example considered, under steady-state conditions, the
passive damper dissipates approximately 37% of the energy input to the system.
Finally, the SDOF passively damped structure is analyzed for response due to the 1940
El Centro S00E ground motion. The initial stiffness of the elastoplastic passive device
is specified as = k, while the yield force y is equal to 20% of the maximum applied
ground force. That is,
The resulting relative displacement and total acceleration time histories are presented in
Fig. 2.14. Again, there is significant reduction in response compared to that of the base
structure without the passive element, as shown in Fig. 2.7. Force-displacement loops
for the viscous and passive dampers are displayed in Fig. 2.15. In this case, the size of
these loops indicates that a significant portion of the energy is dissipated in the passive
device. This tends to reduce the forces and displacements in the primary structural
elements, which of course is the purpose of adding the passive device. A more detailed
examination of elastoplastic devices is included in Chapter 3, which addresses metallic
dampers.
Figure 2.14
1940 El Centro Time History Response for SDOF with
Passive Element; a) Displacement, b) Acceleration
When the passive damper is purely viscous in nature, the response of the SDOF model
is always reduced, since this corresponds precisely to an increase in the damping ratio.
The behavioral trends, illustrated in Figs. 2.2, 2.3, and 2.9 can then be expected. For all
other types of passive devices, careful analysis must be performed to insure that
beneficial effects do indeed
Figure 2.15
1940 El Centro Force-displacement Response for SDOF
with Passive Element; a) Viscous Damper, b) Passive Damper
result. For example, with the addition of our elastoplastic device above, we found
significant improvement in the response due to the El Centro seismic signal. However,
this need not be the case. Under certain conditions, the response can actually increase,
even for the SDOF representation studied here. The situation is further complicated
because location and orientation of passive devices within a structure can have a
significant influence on their effectiveness. As a result, the simplistic treatment
provided in this section is primarily conceptual. The analysis of real building structures
requires a more general methodology.
where M, C, and K represent the mass, damping, and stiffness matrices, respectively,
while symbolizes a matrix of operators which model the passive dampers present in
the structure. Meanwhile, the vector contains the rigid body contribution of the
seismic g round displacement to each degree of freedom, and p includes the forces due
to aerodynamic loading. The matrix represents the mass of the passive dampers.
There are several approaches that can be taken to solve Eq. (2.20). The preferred
approach, in terms of accuracy and efficiency, depends upon the form of the various
terms in that equation. Let us first suppose that the
passive dampers can be modeled as direct linear functions of the acceleration, velocity,
and displacement vectors. That is,
in which
Equation (2.22) is now in the form of the classical matrix structural dynamic analysis
problem. In the simplest case, which we will now assume, all of the matrix coefficients
associated with the primary structure and the passive elements are constant. As a result,
Eq. (2.22) represents a set of N linear second-order ordinary differential equations with
constant coefficients. These equations are, in general, coupled. Thus, depending upon
N, the solution of Eq. (2.22) throughout the time range of interest could become
computationally demanding. This required effort can be reduced considerably if the
equation can be uncoupled via a transformation; that is, if , , and can be
diagonalized. Unfortunately, this is not possible for arbitrary matrices , , and .
However, with certain restrictions on the damping matrix , the transformation to
modal coordinates accomplishes the objective, as shown below.
2.3.2 Modal Superposition Method
Consider the generalized eigenvalue problem associated with the undamped free
vibration of our MDOF structure. That is
and form a complete set spanning the N-dimensional vector space. Consequently, this
set provides the basis for a suitable transformation that can be applied to our original
system defined in Eq. (2.22). In Eqs. (2.25a) and (2.25b), superscript T indicates vector
or matrix transpose.
There are numerous methods available to solve the generalized eigenvalue problem
defined in Eq. (2.24). The choice depends largely upon the size and structure of the
matrices and . Nearly all of the structural analysis codes noted above contain
efficient and robust eigenvalue extraction routines that require little user intervention.
Routines are also available in the public domain through the LAPACK (Anderson, et
al., 1992) implementation.
For notational convenience, the natural frequencies are placed in a diagonal matrix wo.
The corresponding mode shape vectors form the columns of a square matrix , which
functions as the transformation matrix. Thus, any relative displacement vector x can be
represented by
x = y (2.26)
where y is the vector of modal (or normal) coordinates. Utilizing Eq. (2.26), along with
Eq. (2.25), in Eq. (2.22) leads to the following equations of motion expressed in terms
of the modal coordinates,
where
In general, Eq. (2.27) still represents a coupled set of ordinary differential equations.
The equations uncouple only when is also a diagonal matrix. This occurs for the
case of proportional (or Rayleigh) damping, in which
for scalar constants 0 and 1. From Eqs. (2.25) and (2.29), one obtains
which is diagonal. The form of can actually be generalized to the Caughey series
with representing a diagonal matrix containing the i. With this assumed, Eq. (2.27)
becomes
Since the equations are now uncoupled, we can write a scalar equation, for each mode i,
as
in which
Equation (2.34) has the same form as the SDOF system we examined in the previous
section. Consequently, all of the methodology and behavioral patterns discussed in that
section are directly applicable to Eq. (2.34). The solution of our original problem
expressed in Eq. (2.22) is greatly simplified. Once Eq. (2.33) is solved, the relative
displacement vector x can be determined at any time via the transformation Eq. (2.26).
The major computational task in this whole process is the determination of the natural
frequencies and mode shapes. Even this task is not as onerous as it first appears, since
for most physical problems only a small percentage of the N modes actually participate
significantly in the system response. As a result, only the structural modes within a
certain frequency range need be calculated.
The price paid for this simplicity is the initial restriction to system matrices with
constant coefficients, and the further constraint on the damping matrix specified in Eq.
(2.31). If the latter condition is relaxed, it still may be advantageous to use a modal
approach. The governing equations in modal coordinates, Eq. (2.27), remain coupled.
However, it is often still possible to utilize a set much smaller than N, since typically
only a small portion of the undamped modes will be excited. For a more thorough
treatment of the modal superposition method, the textbook by Clough and Penzien
(1975) is recommended.
2.3.3 Damping in Traditional and Passively Damped Structures
In traditional structures, the mass and stiffness properties of various members can
generally be modeled with a reasonable degree of accuracy. The damping properties are
much more difficult to characterize, with significant energy dissipation attributable to
the primary structural materials, to non-ideal joint behavior, as well as to various
nonstructural components. Most experimental results suggest that the damping forces
are nearly independent of frequency, which is contrary to the inherent frequency-
dependence of viscous damping forces. Additionally, it is known that damping
properties of real structures are functions of amplitude. Despite these general
observations, it has not been possible to adequately quantify the energy dissipation. As
a result, it is quite reasonable to assume proportional damping, or its generalized form,
in order to simplify the analysis. Historically, this has been the approach taken.
Structural analysis computer codes and design procedures that exist today most often
assume proportional damping. Modal damping ratios in the range of 1%-5% are
generally employed, depending upon the type of structure. In light of the results for our
SDOF model, considerable care must be exercised not to overestimate the damping
capability of an actual structure, since this could lead to the development of
unconservative designs. An argument can be made for the use of zero damping, unless
sufficient documentation can be provided to justify a certain damping ratio.
The use of passive energy dissipation technology in structures, not only allows one to
reduce structural damage resulting from environmental disturbances, but also permits
characterization of the primary damping mechanisms which are now concentrated in
the passive elements. This characterization is accomplished through the development of
detailed force-displacement models for the passive dampers from physical tests and
analytical studies. These models can subsequently be incorporated into the overall
system equations, as indicated by the term x in Eq. (2.20). Under some circumstances,
the resulting equations can be reduced to the form of Eq. (2.22), or perhaps even to Eq.
(2.33) for proportional damping. In the latter case, analysis and design procedures for
the passively damped structure can closely follow those developed for traditional
structures. An example of such a procedure is provided for viscoelastic dampers in
Chapter 5.
Generally speaking, however, the governing equations for a passively damped structure
are not reducible to an uncoupled set such as Eq. (2.33) that are associated with
proportional damping. In some cases, nonlinearities are present, while in others
hereditary effects appear. In fact, from a physical standpoint, it is not necessarily
optimal to design a structure such that Eq. (2.31) is satisfied. (For example, base
isolated buildings, which can be particularly effective against seismic loads, do not
utilize proportional
damping, but instead feature a flexible, highly damped mechanism at the base.)
Consequently, it is important to develop alternative numerical approaches and design
methodologies applicable to more generic passively damped structural systems
governed by Eq. (2.20). Direct time domain numerical integration algorithms are most
useful in that regard. The Newmark beta algorithm, which has been used extensively in
structural dynamics, is briefly described in the following for a structure modeled by Eq.
(2.22) with matrices and not necessarily constant.
2.3.4 Direct Time Domain Analysis
The implicit Newmark algorithm begins with a discretization of the time axis into
intervals of duration t. At any instant, the equations of motion given in Eq. (2.22) can
be recast in the form:
where the superscripts denote evaluation at time t. It is assumed that this solution is
known, and that the role of the analysis is to determine a solution at time t + t. That is,
we wish to solve
With the Newmark parameters chosen in the usual manner, the trapezoidal rule is then
invoked to relate quantities at time t + t to those at time t. Thus,
where
In general, Eq. (2.39) represents a set of nonlinear equations in the variables x t+t. There
are numerous iterative solution techniques that are applicable, including full, modified,
and quasi-Newton methods. For the full Newton-Raphson method, one solves the
linearized system
at each iteration (n). In Eq. (2.41), the effective tangent stiffness matrix and force
vector can be written
respectively. The solution of Eq. (2.41) is used to update the displacements via
Since Eq. (2.41) requires the formation and factorization of at each iteration, in
practice, the full Newton-Raphson algorithm is seldom employed. Instead, the system
matrix is updated less frequently and acceleration techniques are often utilized. All
major structural analysis codes include algorithms similar to the one outlined above. In
many cases, the program adaptively determines appropriate time step size to maximize
computational efficiency and to reduce the burden placed on the design engineer.
Additional details concerning time domain numerical analysis can be found in
Zienkiewicz and Taylor (1989) and Bathe (1982). It is expected that these methods will
play an increasingly important role in structural design, particularly for systems
employing passive energy dissipation technology.
in which ES is assumed separable into additive contributions ESe and ESp, representing
the fully recoverable elastic strain energy and the dissipative plastic strain energy,
respectively.
Figure 2.16a provides the energy response of a 0.3-scale, six-story concentrically
braced steel structure as measured by Uang and Bertero (1986). The seismic input
consisted of the 1978 Miyagi-Ken-Oki Earthquake signal scaled to produce a peak
shaking table acceleration of 0.33g, which was deemed to represent the damageability
limit state of the model. At this level of loading, a significant portion of the energy
input to the structure is dissipated, with both viscous damping and inelastic hysteretic
mechanisms
having substantial contributions. If the intensity of the signal is elevated, an even
greater share of the energy is dissipated via inelastic deformation. Finally, for the
collapse limit state of this model structure at 0.65g peak table acceleration,
approximately 90% of the energy is consumed by hysteretic phenomena, as shown in
Fig. 2.16b. Evidently, the consumption of this quantity of energy has destroyed the
structure.
Figure 2.16
Energy Response of Traditional Structure; a)
Damageability Limit State, b) Collapse Limit
State (Uang and Bertero, 1986)
From an energy perspective, then, for proper aseismic design, one must attempt to
minimize the amount of hysteretic energy dissipated by the structure. There are
basically two viable approaches available. The first involves designs that result in a
reduction in the amount of energy input to the structure. Base isolation systems, for
example, fall into that category. The second approach, and the one that provides the
central theme for this text, focuses on the introduction of additional energy dissipating
mechanisms
into the structure. These devices are designed to consume a portion of the input energy,
thereby reducing damage to the main structure caused by hysteretic dissipation.
Naturally, for a large earthquake, the devices must dissipate enormous amounts of
energy.
The SDOF system with a passive damper is displayed in Fig. 2.10, while the governing
integrodifferential equation is provided in Eq. (2.18). After integrating with respect to
x, an energy balance equation can be written
EK + ED + ESe + ESp + EP = EI (2.47)
where the energy associated with the passive dampers is
Figure 2.17
Energy Response of Test Structure; a) Without Passive
Devices, b) With Passive Devices (Lobo et al., 1993)
2.4.2 Energy-based Design
While the energy concept, as outlined briefly in the prior section, does not currently
provide the basis for aseismic design codes, there is a considerable body of knowledge
that has been developed from its application to traditional structures. Housner (1956,
1959) was the first to propose an energy-based philosophy for earthquake resistant
design. In particular, he was concerned with limit-design methods aimed toward
preventing collapse of structures in seisinically active regions. Housner assumed that
the energy input calculated for an undamped, elastic idealization of a structure provided
a reasonable upper bound to that for the actual inelastic structure.
Berg and Thomaides (1960) examined the energy consumption in SDOF elastoplastic
structures via numerical computation, and developed energy input spectra for several
strong-motion earthquakes. These spectra indicate that the amount of energy EI
imparted to a structure from a given seismic event is quite dependent upon the structure
itself. The mass, the natural period of vibration, the critical damping ratio, and yield
force level were all found to be important characteristics.
On the other hand, their results did suggest that the establishment of upper bounds for
EI might be possible, and thus provided support for the approach introduced by
Housner. However, the energy approach was largely ignored for a number of years.
Instead, limit-state design methodologies were developed which utilized the concept of
displacement ductility to construct inelastic response spectra as proposed initially by
Veletsos and Newmark (1960).
More recently, there has been a resurgence of interest in energy-based concepts. For
example, Zahrah and Hall (1982) developed a multi-degree of freedom energy
formulation and conducted an extensive parametric study of energy absorption in
simple structural frames. Their numerical work included a comparison between energy-
based and displacement ductility-based assessments of damage, but the authors stopped
short of issuing a general recommendation.
A critical assessment of the energy concept as a basis for design was provided by Uang
and Bertero (1988). The authors initially contrast two alternative definitions of the
seismic input energy. The quantity specified in Eq. (2.45d) is labeled the relative input
energy, while the absolute input energy (EIa) is defined by
In conjunction with this latter quantity, an absolute kinetic energy (EKa) is also
required, where
The absolute energy equation corresponding to Eq. (2.47) then becomes
EKa + ED + ESe + ESp + EP = EIa + EIw (2.50)
Based upon the development of input energy spectra for a SDOF system, the authors
conclude that, while both measures produce approximately equivalent spectra in the
intermediate period range, EIa should be used as a damage index for short period
structures, and EI is more suitable for long period structures. Furthermore, an
investigation revealed that the assumption of Housner to employ the idealized elastic
strain energy, as an estimate of the actual input energy, is not necessarily conservative.
Uang and Bertero also studied a multi-degree of freedom structure, and concluded that
the input energy spectra for a SDOF can be used to predict the input energy demand for
that type of building. In a second portion of the report, an investigation was conducted
on the validity of the assumption that energy dissipation capacity can be used as a
measure of damage. In testing cantilever steel beams, reinforced concrete shear walls,
and composite beams the authors found that damage depends upon the load path.
The last observation should come as no surprise to anyone familiar with classical
failure criteria. However, it does highlight a serious shortcoming for the use of the
energy concept for limit design of traditional structures. As was noted above, in these
structures, a major portion of the input energy must be dissipated via inelastic
deformation, but damage to the structure is not determined simply by the magnitude of
the dissipated energy. On the other hand, in non-traditional structures incorporating
passive damping mechanisms, the energy concept is much more appropriate. The
emphasis in design is directly on energy dissipation. Furthermore, since an attempt is
made to minimize the damage to the primary structure, the selection of a proper failure
criterion is less important.
2.4.3 MDOF Structures
In Section 2.4.1, an energy formulation was presented for a SDOF system which
included passive dampers. While this formulation is of interest for a general
understanding of the overall behavior of a structure during a seismic or aerodynamic
event, local characteristics are also quite important. In particular, the ultimate collapse
of a structure can often be traced to areas of damage concentration. In order to properly
account for this behavior, a more comprehensive approach is needed, which permits a
better spatial representation of the structure. Consequently, a general multi-degree of
freedom energy formulation is developed below, in a manner similar to that presented
by Zahrah and Hall (1982) and Uang and Bertero (1986).
The starting point is the equations of motion for the MDOF structure provided in Eq.
(2.20). An energy representation for the structural response
can be formed by integrating the individual force components in Eq. (2.20) over the
corresponding relative displacements. Once again this produces the scalar equation,
EK + ED + ES + EP = EI (2.51)
where
The terms on the left hand side of Eq. (2.51) quantify the contribution of the various
forms of energy that the structure employs to resist the earthquake or wind load event.
These energies can be evaluated in a straightforward manner after a solution to Eq.
(2.20) is obtained via any of the methods outlined in Section 2.3.
The total energy balance Eq. (2.51) is, of course, important for an overall assessment of
performance. However, in addition, the energy response of each structural member can
be determined by utilizing the individual element submatrices. These latter calculations
permit an investigation of the flow of energy within the structure as, for example, a
seismic event progresses. Potential zones of damage concentration can be easily
identified, and perhaps subsequently eliminated through the use of passive dampers.
With the continuing advancement in computational technology, one can envision an
interactive computer-aided design process, in which the design engineer directs the
development of aseismic structures by considering this energy flow.
References
Anderson, E. et al. (1992), LAPACK Users' Guide, SIAM, Philadelphia.
Bathe, K.-J. (1982), Finite Element Procedures in Engineering Analysis, Prentice-
Hall, Englewood Cliffs, New Jersey.
Berg, G. V. and Thomaides, S. S. (1960), Energy Consumption By Structures in
Strong Motion Earthquakes, Proceedings of the Second World Conference on
Earthquake Engineering, II, 681-697, Tokyo.
Bolotin, V.V. (1984), Random Vibrations of Elastic Systems, Martinus Nijhoff
Publishers, The Hague.
Clough, R. W. and Penzien, J. (1975), Dynamics of Structures, McGraw-Hill, New
York.
Cruse, T. A., Unruh, J. F., Wu, Y.-T., and Harren, S. V. (1990), Probabilistic
Structural Analysis for Advanced Space Propulsion Systems, Journal of
Engineering for Gas Turbines and Power, 112, 251-260.
Hanson, R. D. (1993), Supplemental Damping for Improved Seismic Performance,
Earthquake Spectra, 9(3), 319-334.
Housner, G. W. (1956), Limit Design of Structures to Resist Earthquakes,
Proceedings of the World Conference on Earthquake Engineering, 5-1 - 5-13,
Earthquake Engineering Research Center, Berkeley, California.
Housner, G. W. (1959), Behavior of Structures During Earthquakes, Journal of the
Engineering Mechanics Division, ASCE, 85(EM4), 109-129.
Lobo, R. F., Bracci, J. M., Shen, K. L., Reinhorn, A.M., and Soong, T. T. (1993),
Inelastic Response of R/C Structures with Viscoelastic Braces, Earthquake
Spectra, 9(3), 419-446.
Newmark, N.M. and Rosenblueth, E. (1971), Fundamentals of Earthquake
Engineering, Prentice Hall, Englewood Cliffs, New Jersey.
Press, W. H., Teukolsky, S. A., Vetterling, W. T. and Flannery, B. P. (1992),
Numerical Recipes in Fortran, Cambridge University Press, Cambridge, UK.
Shinozuka, M. (1987), Stochastic Mechanics, I, Department of Civil Engineering
and Engineering Mechanics, Columbia University, New York.
Soong, T. T. and Grigoriu, M. (1993), Random Vibration of Mechanical and
Structural Systems, P T R Prentice-Hall, Englewood Cliffs, New Jersey.
Szab, B. and Babuska *, I. (1991), Finite Element Analysis, John Wiley and Sons,
New York.
Uang, C. M. and Bertero, V.V. (1986), Earthquake Simulation Tests and
Associated Studies of a 0.3 Scale Model of a Six-Story Concentrically Braced Steel
Structure, Report No. UCB/EERC-86/10, Earthquake Engineering Research
Center, Berkeley, California.
Uang, C. M. and Bertero, V.V. (1988), Use of Energy as a Design Criterion in
Earthquake Resistant Design, Report No. UCB/EERC-88/18, Earthquake
Engineering Research Center, Berkeley, California.
Veletsos, A. S. and Newmark, N.M. (1960), Effect of Inelastic Behavior on the
Response of Simple Systems to Earthquake Motions, Proceedings of the Second
World Conference on Earthquake Engineering, II, 895-912, Tokyo.
Zahrah, T. F. and Hall, W. J. (1982), Seismic Energy Absorption in Simple
Structures, Structural Research Series No. 501, University of Illinois, Urbana,
Illinois.
Zienkiewicz, O. C. and Taylor, R. L. (1989), The Finite Element Method, Volumes
1 and 2, Fourth Edition, McGraw-Hill, London.
3
Metallic Dampers
3.1 Introduction
One of the most effective mechanisms available for the dissipation of energy, input to a
structure during an earthquake, is through the inelastic deformation of metallic
substances. In traditional steel structures, aseismic design relies upon the post-yield
ductility of structural members to provide the required dissipation. However, the idea of
utilizing separate metallic hysteretic dampers within a structure to absorb a large
portion of the seismic energy began with the conceptual and experimental work by
Kelly et al. (1972) and Skinner et al. (1975). Several of the devices considered by those
researchers included torsional beam, flexural beam, and U-strip dampers as shown
schematically in Fig. 3.1.
Figure 3.1
Metallic Damper Geometries (Skinner et al, 1975;
a) Torsional Beam, b) Flexural Beam, c) U-strip
In order to effectively include these devices in the design of an actual structure, one
must be able to characterize their expected nonlinear force-displacement behavior
under arbitrary cyclic loads. zdemir (1976) was the first to consider this modeling
problem. He utilized analogies with existing elastoplastic and viscoplastic constitutive
theories to develop appropriate forms for the force-displacement relationships.
Additionally, zdemir detailed efficient numerical algorithms for computing the
response of structures with metallic dampers subjected to general time-dependent
loading, such as that caused by an earthquake. Shortly thereafter, Bhatti et al. (1978)
employed that methodology to study the response of structures that utilized torsion bar
dampers in conjunction with a seismic base isolation system.
During the ensuing years, considerable progress has been made in the development of
metallic dampers. For example, many new designs have been proposed, including the
X-shaped and triangular plate dampers displayed in Fig. 3.2. Alternative materials, such
as lead and shape-memory alloys, have been evaluated. Numerous experimental
investigations have been conducted to determine performance characteristics of
individual devices and laboratory test structures. As a result of this ongoing research
program, several commericial products have been developed and implemented in both
new and retrofit construction projects. In particular, a number of existing structures in
New Zealand, Mexico, Japan, Italy, and the United States now include metallic
dampers as a means for obtaining improved seismic resistance.
a) b)
Figure 3.2
Metallic Dampers; a) X-shaped Plate Damper (Courtesy of
CounterQuake Corp.), b) Triangular Plate Damper (Tsai et al., 1993)
In order to fully appreciate the behavior of metallic energy dissipators, one must first
consider the inelastic response of metals under time-dependent cyclic loading. This is
addressed in some detail in the following section, which deals primarily with the
definition of constitutive models. These models are then used in Section 3.3 to develop
appropriate force-displacement relationships for the metallic dampers. Once the force-
displacement model is established, the design engineer must incorporate that
information in the analysis of the overall structure. This task is discussed in Section 3.4.
As noted above, considerable effort has also been devoted to experimental studies at
both the component and structural level. Results from several of these studies are
examined in Section 3.5. The focus then shifts to design and implementation. An
overview of some of the key aspects of design associated with a decision to include
metallic dampers in a construction project is provided in Section 3.6. The chapter then
concludes with a description of several existing structures that have either been newly
designed or retrofitted with metallic energy dissipators for enhanced seismic protection.
= el + in (3.2)
in which
with E representing the elastic modulus. The energy, or more precisely the energy
density, is measured by the area under the stress-strain curve from O to B. Part of that
energy is recoverable. However, the remainder, associated with the inelastic strain in
and identified by the shaded portion
of Fig. 3.4a, is dissipative. A significant portion of that dissipative energy is converted
into heat. In this strain hardening regime, successive increments of stress produce
correspondingly greater increments of strain, until the point M is reached associated
with the maximum load that the tensile specimen can withstand. Beyond M the
specimen becomes unstable. Localization phenomena commence with the appearance
of a necking region in which three-dimensional states of stress are present. Ultimately
failure occurs at point X.
Figure 3.4
Nominal Stress - Conventional Strain Diagrams
As noted above, this description applies to most metals. However, annealed mild steel
and some other alloys behave as depicted in Fig. 3.4b. In this case, the response is
similar to that previously discussed, except in the region just beyond first yield at YU. In
these materials, there is an abrupt drop in stress from point YU to YL, corresponding to
the upper and lower yield stress, respectively. This phenomenon and the ensuing stress-
strain plateau is caused by the formation and propagation of Lders bands. It should be
noted, however, that strain hardened mild steel does not exhibit this behavior, but
instead follows that depicted in Fig. 3.4a. The final stress-strain curve, shown in Fig.
3.4c, is typical of a brittle metal, such as cast iron. Obviously, this type of material is
not a good candidate for use in metallic dampers, since very little energy dissipation
occurs before fracture.
The definitions employed in Eq. (3.1) are perfectly adequate when the length and area
of the specimen does not change significantly from the initial values. However, for
higher strain levels, the natural strain e and true stress are
more appropriate measures, where
Replotting Fig. 3.4a in these coordinates leads to the curve displayed in Fig. 3.5. Notice
that now the true stress increases with natural strain all the way to rupture. More
importantly, if one were to test the specimen instead in compression, a nearly identical
-e curve would result. On the other hand, a plot of n versus for compression would
not reproduce Fig. 3.4a much beyond the yield stress.
Figure 3.5
True Stress - Natural Strain Diagram
Over the years, numerous mathematical models have been introduced to idealize the
stress-strain curves, including the elastic-perfectly plastic model, the elastic-linear
strain hardening model, and the Ramberg-Osgood model shown in Fig. 3.6. The last of
these models can be written
with material constants E, k, and n. Employing Eqs. (3.2) and (3.3) and assuming that
all nonlinearity is inelastic, one can alternatively write
Consequently, in order to develop models for metallic dampers, we must extend our
discussion to behavior under load reversals that involve excursions into the inelastic
range. For these cases, response is path dependent. Consider the cylindrical specimen
first loaded in tension past yield at Y to a point B in the strain hardening range, and
then gradually unloaded as indicated in Fig. 3.7. The unloading branch of the curve is
parallel to the initial loading curve, indicating purely elastic response. However, if
unloading continues sufficiently into the compression range, yielding will again take
place at the point labeled Y'. The stress associated with Y', i.e. , is dependent upon the
prior amount of strain hardening. This is known as the Bauschinger effect. As useful
simplifications, isotropic hardening theories establish
where J2 is the second invariant of the deviatoric stress. For F < 0, the response is
elastic. Yielding occurs when F = 0 and (F/) > 0. In that case, a flow rule is
needed to determine the incremental inelastic strains (din).
For an associative flow rule, normally valid for metal plasticity,
Figure 3.8
Two Surface Plasticity Model - Uniaxial Case
Table 3.1
Uniaxial Two-Surface Model Definition
in which and b are the relaxation time and back stress, respectively, while the
parameter n is an odd integer. The drag stress d is a material constant. Equation
(3.10a) is written in a form that emphasizes the individual elastic and inelastic
contributions to the total strain rate. For the special case of b = 0 and n = 1, Eq. (3.10a)
reduces to the classical Maxwell model of viscoelasticity. However, here we are more
interested in the viscoplastic models obtained for n > 1. The back stress b is an internal
state variable, while the quantity - b is denoted as an overstress. The inclusion of
these stresses
permits the modeling of kinematic hardening. Consequently, both are used extensively
in viscoplastic theories (Krempl, 1987). However, an additional equation is then needed
to determine the evolution of the internal state variable. For the zdemir rate-
dependent model the evolution equation can be written
with material parameter . The two Eqs. (3.10a) and (3.10b) completely define the
response for a variety of rate-dependent materials subjected to general non-monotonic
time-dependent uniaxial loading. Notice that there is no need to check a yield criterion
nor to consider any special rules for loading or unloading.
This last feature suggests that the viscoplastic models may also be useful for rate-
independent plastic analysis, if a suitable limiting form can be obtained. With that in
mind, zdemir derived the following differential equations:
in which is now a constant controlling the slope of the - curve in the inelastic
range. In constructing the final form of Eqs. (3.11), absolute values were introduced in
the equations given by zdemir to permit n to assume any positive real value. The sgn
function in Eqs. (3.11) returns the sign of the argument. Comparing Eqs. (3.11) with
(3.10), one finds that rate-independence is achieved by selecting a variable relaxation
time, such that
The integration of either Eqs. (3.10) or (3.11) in time is required for determination of
material response. This is most easily accomplished numerically using, for example, a
high order Runge-Kutta formula with adaptive step size (Press et al., 1992). As an
illustration of the behavior predicted by Eq. (3.11), consider the response shown in Fig.
3.10 due to a sinusoidal variation of enforced strain. Notice from Fig. 3.10a that the
stress is indeed independent of the strain rate. Furthermore, it is evident from Fig. 3.10b
that the parameter n controls the sharpness of the knee, while a determines the slope in
the inelastic range as indicated in Fig. 3.10c. For n , the drag stress d becomes
equal to the usual yield stress y, and the response approaches the elastic-linear strain
hardening idealization shown
Figure 3.10
zdemir Rate-independent Model; a) Variation of Strain Rate, b)
Variation of Sharpness of Knee, c) Variation of Strain Hardening
in Fig. 3.6b. Additionally, with = 0 the elastic-perfectly plastic model of Fig. 3.6a is
recovered.
An extension of the zdemir model is needed, however, in order to address situations
involving multiaxial loading which may occur in some metallic dampers. Graesser and
Cozzarelli (1991) developed such an extension and then applied the resulting rate-
independent model to study biaxial cyclic response of structural steel. This extended zdemir
model and the two-surface plasticity model, discussed in the previous subsection, are both
applicable for mild steel near room temperature under moderate strain rates. However, the
structure of the zdemir rate-independent model is closely related to early versions of
endochronic plasticity theory (Valanis, 1971), and therefore this model suffers from the same
theoretical defects present in the initial Valanis formulations (e.g., Rivlin, 1981). Although
stable results are typically obtained, some caution is advised.
In a series of experiments, Chang and Lee (1987) found measureable differences in the
response of annealed structural steel due to strain rates above 10-4/s. Figure 3.11 illustrates the
typical behavior under uniaxial cyclic straining. The effect during monotonic loading was
even more pronounced.
For example, the apparent yield strength increased 27% due to an increase in strain rate
from 10-5/s to 10-2/s. Consequently, for steel dampers involving high strain rates, it may
be appropriate to utilize a rate-dependent model, such as that presented in Eq. (3.10) or
one of the established formulations discussed in Krempl (1987). For lead dampers, the
use of a rate-dependent model is even more important.
Figure 3.11
Experimental Strain Rate Dependence
Figure 3.13
Mathematical Model for Triangular Plate Device;
a) Geometric Definition, b) Beam Idealization
in Fig. 3.13. Note that coordinate axes x, y, z are defined on the undeformed midsurface
of the plate.
The force-displacement relationship for the damper can be readily established for
infinitesimal, elastic response. For that case, the classical Euler-Bernoulli beam theory
is valid. A quasistatic formulation is adopted by ignoring inertia of the plate. Then, at
any cross-section, the moment equilibrium equation can be written:
With the cancellation of the term (L - x) from both sides of Eq. (3.16), it is evident that
the stress is independent of position along the beam axis. After applying the
unidirectional elastic constitutive relationship
= E (3.17)
and the kinematic condition
= ky (3.18a)
one finds that the curvature k is constant along the entire length of the plate, with
Substituting Eqs. (3.17), (3.18a), and (3.18b) in Eq. (3.16) produces the following
force-displacement model:
This simple result is provided in Tsai et al. (1993) for the triangular plate damper, and
has been found to agree reasonably well with experiments.
However, the above model pertains only to the elastic response of the damper. Very
little can be inferred from Eq. (3.19) concerning the energy dissipation characteristics
of the device. Information of that type, which is vital for proper aseismic design,
requires examination of the inelastic response. As a first approximation, the kinematic
assumptions employed in Eqs. (3.18) can be retained, while replacing Eq. (3.17) by a
rate-independent inelastic constitutive model. For the constitutive model defined in
Table 3.1, the governing equations can be written in the following rate form:
Unfortunately the resulting force-displacement relationship can no longer be expressed
analytically as in the elastic case. Instead, the necessary integrals can be evaluated
numerically with standard gaussian quadrature formulae utilized for the spatial
integration specified in Eq. (3.20) and an adaptive Runge-Kutta algorithm invoked to
solve the set of first order differential equations.
Under certain conditions (e.g. major earthquakes), the interstory drift may become
comparable in magnitude to the damper plate length L. In these situations, the effects of
finite deformation on damper response cannot be ignored. Either the present beam
model can be extended to include these effects or a finite element analysis can be
performed using a general purpose computer code, such as ABAQUS. The former
approach, which essentially involves writing the equation of moment equilibrium in the
deformed configuration, is briefly summarized in the following.
First an intrinsic , n coordinate system is introduced, as shown in Fig. 3.13b, which
deforms with the beam axis. Also defined in that diagram is , the angle between the x-
axis and the tangent at . As a simplification, we will continue to ignore the effects of
shear deformation and the axial extension of the midsurface. Then, strictly from
kinematic considerations,
With finite deformation, the curvature k is no longer constant along the length of the
plate. Consequently, at each instant of time, a nonlinear boundary value problem (BVP)
must be solved to determine the unknown curvature function. This BVP requires
satisfaction of the moment equilibrium equation
which now varies with . In addition, the free end deflection condition
y(L) = (3.25)
Figure 3.14
Triangular Plate Metallic Damper; a) Experimental (Tsai et al., 1993), b) Numerical
A more detailed account of this mechanics-based modeling approach for the triangular
plate damper is provided in Dargush and Soong (1995), while Tsai and Tsai (1995) also
address the behavior of this device with a related mechanics-based approach. However,
it should be emphasized that the above model development and subsequent calculations
were intended primarily for illustrative purposes. The methodology presented should be
applicable to a wide variety of the metallic dampers that have been developed,
including in particular the X-shaped plate dampers. A more rigorous investigation of
any of these dampers would involve a finite element analysis including geometric
nonlinearity and a three-dimensional constitutive model, written in terms of a
co-rotational stress rate. Additional material and component tests would also be
required to determine reliability and variability of experimental data. In any case, after
establishing a mechanics-based representation for the damper, numerical experiments
can be conducted to determine parameters in a simplified force-displacement
relationship such as Eqs. (3.15), or the model can be used directly in the analysis of the
overall structure. Both options are
discussed in the following section. Before shifting to that topic, let us remark that the
mechanics-based approach permits one to learn much more about the behavior of a
metallic damper than is contained in the force-displacement hysteresis loops. This
additional information can be used to improve damper design and to obtain a better
understanding of its performance within a structural system.
It is clear from these diagrams that there exist ranges of To in which the elastoplastic
damper is either ineffective or even detrimental under this particular ground motion.
The above results for the simple SDOF system certainly suggest that detailed nonlinear
transient dynamic anaylses are required to properly evaluate the effectiveness of any
real structure employing metallic dampers for enhanced seismic protection. As
mentioned in Chapter 2, the finite element
Figure 3.15
SDOF Response to 1940 El Centro S00E;
a) Relative Displacement, b) Total Acceleration
method (FEM) currently provides the most suitable framework for a multi-degree of
freedom (MDOF) analysis of an overall structure. The modified Newton-Raphson time
domain approach, detailed in Section 2.3.4, is directly applicable for structures that
include metallic dampers. This type of algorithm is typically available in most major
FEM computer programs. Many of these programs also permit user-defined structural
elements. In that case, force-displacement relationships, such as those defined by Eqs.
(3.15) or Eqs. (3.20)-(3.22), can be easily incorporated into the finite element model.
Additionally, major codes include substructuring facilities, which permit separation of
linear and nonlinear degrees of freedom (dof). The effective tangent stiffness matrix for
the linear dof can be formed and eliminated once at the beginning of the analysis. The
solution at each time step then involves the formation and decomposition of a relatively
small system containing only the nonlinear
dof. This type of partitioning was utilized by Ozdemir (1976) for analysis of structures
incorporating torsional beam dampers. Subsequently, the same methodology was used
by Bhatti et al. (1978) to develop an optimal design approach for a base isolation
system, which included the torsional beam damper. The method of feasible directions
was employed in the latter work for the optimization of several simple structural
systems. In light of the considerable advances in computing capabilities and
optimization software that has occurred during the intervening years, it would be
interesting to apply a similar approach for the design of metallic dampers positioned
within the superstructure.
Recent applications of nonlinear structural analysis to building frames incorporating
metallic dampers can be found in Xia and Hanson (1992), Jara et al. (1993), and Tsai et
al. (1993). The first two references consider X-shaped plate dampers, while the last
examines triangular plate dampers. All three works utilize the DRAIN-2D computer
code (Kanaan and Powell, 1973) for the analyses.
As a final note in this section, it should be mentioned that there are several attractive
alternatives to the time-domain algorithm outlined in Section 2.3.4. One of these
alternatives involves rewriting the governing equations in state-space form by
introducing both displacement and velocity as primary unknowns, as is often done in
the control literature (Soong, 1990). This permits the use of very accurate and efficient
first-order differential equation solvers. While such algorithms are typically not
currently available in the major FEM programs, Inaudi and de la Llera (1992) have
developed a research code using this formulation.
Figure 3.16
Lead Extrusion Energy Absorbers (Robinson and Greenbank,
1976); a) Constricted Tube Design, b) Bulged Shaft Design
Figure 3.17
Lead Extrusion Energy Absorbers Force-Displacement
Loops at 1cm/min (Robinson and Greenbank, 1976);
a, b) Typical Constricted Tube Response, c, d) Typical
Bulged Shaft Response
Currently, the flexural plate steel dampers have become the predominant form for
metallic devices, and a number of experimental programs have been conducted at
university laboratories. Bergman and Goel (1987) reported on the cyclic testing of X-
shaped and V-shaped plate dampers manufactured by Bechtel Corporation. Geometric
details for the X-shaped damper are provided in Fig. 3.18, while the experimental setup
using a single story full-scale building test frame is displayed in Fig. 3.19. A total of
three dampers were tested under constant amplitude displacement controlled cycling at
0.33 Hz. For each specimen, seven different levels of displacement amplitude ranging
up to approximately 1.5in. were applied for ten cycles each. Cycling at the highest level
then continued until failure occurred. Force-displacement curves for the X-shaped
damper at displacement amplitudes of 0.42in., 1.04in. and 1.56in. are reproduced in
Fig. 3.20. Energy dissipated per cycle can, of course, be calculated from these
hysteresis loops, and is provided in the report. The fatigue life of the specimen was 44
cycles at an amplitude of 1.56in. Meanwhile, the V-shaped damper proved to be a less
favorable design configuration in terms of both energy dissipation and durability. In
concluding, the authors
mention that axial shortening of the device, which occurs at large horizontal
displacement, must be considered in design of the adjacent structural members and
connections.
Figure 3.18
X-shaped Device Geometry (Bergman and Goel, 1987)
Figure 3.19
X-shaped Device Experimental Setup (Bergman and Goel, 1987)
Figure 3.20
Force-Displacement Response of X-shaped Device (Bergman and
Goel, 1987); Displacement Amplitude a) 0.42in, b) 1.04in, c) 1.56in
Figure 3.22
Force-Displacement Response of ADAS Elements (Whittaker et
al., 1991); Displacement Amplitude a) 0.45in, b) 1.5in, c) 2.2in
Based upon the similarities in design, one would expect that the hysteresis loops
obtained by Bergman and Goel (1987) and Whittaker et al. (1991) should also have
similarities. Comparing Figs. 3.20a and 3.22a, we find that to be the case for small
amplitude cycling. However, at larger amplitudes the response contains significant
differences, as can be seen by comparing Figs. 3.20c and 3.22b. This difference at large
displacement amplitudes is due to axial effects associated with finite deformation. In
Bergman and Goel (1987), axial deformation is resisted by the supporting frame
members, while in the study by Whittaker et al. (1991) there is no vertical constraint.
For ADAS deflections that are likely to occur in severe earthquakes, there is coupling
between lateral and axial response, which has not been considered. Even without axial
constraint, effects of geometric nonlinearity will eventually become evident as the
displacement amplitude is increased. For example, the apparent stiffening of the device
in Fig. 3.22c is due to finite deformation, not strain hardening of the material. (Recall
that the stabilized response of steel under cyclic loading follows the curve displayed in
Fig. 3.9. There is no plateau.) It should be emphasized that this discussion does not
imply that ADAS elements are unsuitable at large deflection, but rather that the effects
of finite deformation should be included in the assessment of a design.
As mentioned above, Whittaker et al. (1991) also examined the performance of ADAS
elements in a test structure under seismic excitation. A schematic of the three story,
single bay Ductile Moment Resisting Space Frame (DMRSF), having plan dimensions
12ft. by 6ft. is shown in Fig. 3.23. With the six ADAS elements in place, the entire
structure is referred to as ADAS-3. The natural periods and damping ratios for DMRSF
and ADAS-3 are detailed in Table 3.3. Notice that there is a significant shift in period
of the first mode due to the added stiffness of the chevron bracing and ADAS elements.
Meanwhile, the increase in damping ratio was attributed solely to the ADAS framing.
For the seismic simulation, three different records were utilized including a soft soil
synthetic denoted THSSR, 1985 Chile (Llolleo) N10E, and 1940 El Centro S00E. Table
3.4 provides a comparison of the response of the structure without (DMRSF) and with
(ADAS-3) the metallic dampers due to the Chilean earthquake with a maximum ground
acceleration of 0.13g. Selected results are displayed graphically in Fig. 3.24, which
clearly indicates the beneficial effects of the ADAS dampers in this case. The passively
damped structure was also subjected to El Centro at 0.33g, THSSR at 0.40g, and Chile
at 0.56g. Hysteresis loops for the ADAS dampers during the El Centro signal are shown
in Fig. 3.25, while Fig. 3.26 contains the corresponding energy plot for the entire
structure. From the latter figure, it is evident that the ADAS dampers dissipate a
significant portion of the input seismic energy.
Figure 3.23
ADAS Test Structure (Whittaker et al., 1991)
Figure 3.24
Response Comparison for 1985 Chile (Llolleo) N10E at 0.13g
[solid = MRF; dashed = ADAS frame] (Whittaker et al., 1991)
Table 3.4 Chile 0.1g Response Envelopes for Test Structure (Whittaker et al., 1991)
a) DMRSF response
Roof/3 3/2 2/1
Floor/Story Max Min Max Min Max Min
b) ADAS-3 response
Roof/3 3/2 2/1
Floor/Story Max Min Max Min Max Min
Figure 3.26
Energy Time Histories for 1940 El Centro
S00E at 0.33g (Whittaker et al., 1991)
for the TADAS device was presented previously in Fig. 3.14a for a loading program
that involved increasing displacement amplitude. Note the similarities in these
hysteresis loops compared to those obtained for the X-shaped dampers. This is, of
course, expected since the mechanism is the same. However, in the TADAS device
there are no axial forces developed at large displacement due to the slotted pin
connection at the apex, except for a portion attributable to friction. Tsai et al. (1993)
also performed pseudodynamic testing of the two story steel frame shown in Fig. 3.12.
Experimental results for the first and second floor displacements are displayed in Fig.
3.27 for a simulation of the 1940 El Centro S00E earthquake. A significant reduction is
obtained with the inclusion of the TADAS elements.
Figure 3.27
Pseudodynamic Response of Test Frame (Tsai et al., 1993)
for the selected values of the parameters SR, and U. Design the TADAS device for
each story based upon y and Kd, where
5. Perform lateral force analysis of the structure with TADAS elements in place for
the service level design earthquake. Repeat steps 3 and 4 as necessary to meet the
design requirements.
6. Perform capacity design checks for all members. Assume that the ultimate force
generated in the TADAS device is 1.5Pu, where
Additional details on the process, along with a design example for a twenty story, three
bay planar frame, can be found in Tsai et al. (1993). An alternative design approach
based upon the concept of equivalent viscous damping has also been outlined in Scholl
(1993).
While the methods proposed in the above references are useful in the near term, all are
based on the notion of design as a manual process, with perhaps some use of elastic
analyses at various steps. However, in light of the complicated nature of the metallic
devices, the superstructure, and the seismic signal, it would seem more appropriate to
view design as a computational process. Effort would then be directed toward
constructing realistic mathematical models of devices and superstructures, and on
developing the necessary software tailored for the solution of the resulting optimal
structural design problems. This could also be tied in with the ongoing efforts to
establish building code requirements for passively damped systems which are discussed
in Whittaker et al. (1993).
Figure 3.28
Force-Displacement Model for
TADAS Frame System (Tsai et al., 1993)
The thirteen story Izazaga #38 - 40 building is shown in Fig. 3.30. This is a reinforced
concrete frame with brick infilled end walls constructed in the late 1970s. The building
sustained moderate damage in the 1985 Mexico City earthquakes. An upgrade
afterward was unsuccessful, since further damage occurred during 1986 and 1989
seismic attacks. Passive energy dissipation technology was selected for a second
retrofit. For this project approximately 250 ADAS dampers were installed in 1990 in
outer frame bays to permit continued building operation during construction. A picture
of a typical installation of a brace-damper subassembly is provided in Fig. 3.31.
Structural analysis indicated that the fundamental periods in the principal directions
were reduced from 3.82s and 2.33s to 2.24s and 2.01s with the addition of the ADAS
elements. Calculations also determined a 40% reduction in interstory drift, while
retaining the same base shear coefficient. Nonlinear, time history DRAIN-2DX
analyses were used to verify the final design.
The second Mexican application involves retrofit of the six story Cardiology Hospital
Building, built in the 1970s and damaged in the 1985 earthquakes. The seismic upgrade
of the reinforced concrete frame structure, completed in 1990, features a series of
eighteen external buttresses connected to the building via a total of ninety ADAS
dampers. A front view with the buttresses in-place is presented in Fig. 3.32. This
particular design permitted minimal interference with hospital operation during retrofit
construction. Nonlinear, time history DRAIN-2D analysis was again employed
extensively in the redesign process. However, in this case, the addition of the buttresses
and ADAS dampers reduced both base shears and interstory drift.
Figure 3.30
Izazaga #38-40 Building (Martinez-Romero, 1993)
Figure 3.31
Brace-Damper Assembly in Izazaga
#38-40 Building (Martinez-Romero, 1993)
Figure 3.32
Cardiology Hospital Building with Exterior Buttresses
and ADAS Dampers (Martinez-Romero, 1993)
In 1992, the Reforrna #476 buildings, which contain key operations of the Mexican
Institute of Social Security (IMSS), underwent major seismic upgrade. This cluster of
three ten-story buildings constructed in 1940 is listed in the National Register of
Classical Buildings. An exterior view is provided in Fig. 3.33. While the buildings
sustained some significant damage during
Figure 3.33
IMSS Reforma Building (Martinez-Romero, 1993)
the 1957 earthquake, the more recent seismic events of 1985 had only minor effect.
Through the years, the foundation has been improved to alleviate problems of
differential settlement. Nevertheless, it was felt that the buildings
remained vulnerable to future severe earthquakes. The most recent retrofit, discussed
by Martinez-Romero (1993), involves the use of ADAS dampers with chevron bracing
in forty frame bays throughout the three structures. An example is shown in Fig. 3.34.
The buildings were analyzed in detail using the nonlinear DRAIN-2D program, along
with four earthquake ground motions. Two-dimensional analysis was performed on
thirteen different cross-sections. In order to assess the suitablity of the revised design,
comparisons were made for the bare frame, the structure with braces and ADAS
dampers, and a design using only the braces. The calculated response at one section of
the central building is enumerated in Table 3.5 for one of the earthquakes. The roof
displacements and interstory drifts are significantly reduced for the ADAS design.
However, base shear, critical column force, and roof acceleration all
Figure 3.34
Brace-Damper IMSS Reforma Building Retrofit
Scheme From Outside (Martinez-Romero, 1993)
increased, although not as much as would occur with the addition of only the bracing.
In the final design, a floor diaphragm and several interior columns were strengthened to
accommodate the elevated force levels. As in the other Mexican projects, the retrofit
was completed while the complex remained in operation. It should be noted that in
addition to describing the details of these three retrofit projects, Martinez-Romero
(1993) relates many worthwhile observations concerning design and construction
issues.
Table 3.5 Calculated Earthquake Response for IMSS Reforma Building
(Martinez-Romero, 1993)
Deformations
Displacement
at Roof Story Drift (cm) Story Drift Ratio
The final implementation of metallic dampers considered in this chapter involves the
retrofit of the Wells Fargo Bank building in San Francisco, California (USA) discussed
in Perry et al. (1993). The building is a two story nonductile concrete frame structure
originally constructed in 1967 and subsequently damaged by the 1989 Loma Prieta
earthquake. The voluntary upgrade by Wells Fargo utilized chevron braces and ADAS
damping elements. More conventional retrofit schemes were rejected due to an inability
to meet the performance objectives while avoiding foundation work. A plan view of the
second floor including upgrade details is provided in Fig. 3.35. A total of seven ADAS
devices were employed, each with a yield force of 150kips. Numerous linear and
nonlinear analyses were used in the retrofit design process. For example, the three-
dimensional model shown in Fig. 3.36 provided elastic modes and response spectrum
analysis results for the original structure from the program SAP90. Further three-
dimensional response spectrum analyses, using an approximate equivalent linear
representation for the ADAS elements, furnished a basis for the redesign effort. The
final design was verified with DRAIN-2D nonlinear time history analyses. A
comparison of computed response before and after the upgrade is contained in Fig.
3.37. The numerical results indicted that the revised design was stable and that all
criteria were
met. In addition to the introduction of the bracing and ADAS dampers, several interior
columns and a shear wall were strengthened. The entire project, including design,
permit approval, and construction were completed within a six month period, in time
for a February 1992 opening of the bank branch. Perry et al. (1993) also describe the
permit approval process, and suggest that the ADAS elements have many potential
applications for both new and retrofit projects involving either concrete or steel framed
structures.
Figure 3.35
Wells Fargo Bank Building Retrofit Details (Perry et al., 1993)
Figure 3.36
Structural Analysis Model for Wells
Fargo Bank Building (Perry et al., 1993)
Figure 3.37
Comparison of Computed Results for Wells
Fargo Bank Building (Perry et al., 1993)
References
Aguirre, M. and Snchez, A. R. (1992), Structural Seismic Damper, J. Structural Engineering, ASCE,
118(5), 1158-1171.
Bergman, D. M. and Goel, S.C. (1987), Evaluation of Cyclic Testing of Steel-Plate Devices for Added
Damping and Stiffness, Report No. UMCE 87-10, The University of Michigan, Ann Arbor, MI.
Bhatti, M. A., Pister, K. S. and Polek, E. (1978), Optimal Design of an Earthquake Isolation System,
Report No. UCB/EERC-78/22, University of California, Berkeley, CA.
Boller, Chr. and Seeger, T. (1987), Material Data for Cyclic Loading, Part A: Unalloyed Steels,
Elsevier, Amsterdam.
Chang, K. C. and Lee, G. C. (1987), Strain Rate Effect on Structural Steel Under Cyclic Loading, J.
Engrg. Mech., ASCE, 113(9), 1292-1301.
Cofie, N. G. and Krawinkler, H. (1985), Uniaxial Cyclic Stress-Strain Behavior of Structural Steel, J.
Engrg. Mech., ASCE, 111(9), 1105-1120.
Dafalias, Y. F. and Popov, E. P. (1975), A Model of Nonlinearly Hardening Materials for Complex
Loading, Acta Mechanica, 21, 173-192.
Dargush, G. F. and Soong, T. T. (1995), Behavior of Metallic Plate Dampers in Seismic Passive
Energy Dissipation Systems, Earthquake Spectra, 11(4), 545-568.
Dowling, N.E. (1972), Fatigue Failure Predictions for Complicated Stress-Strain Histories, J.
Materials, JMLSA, 7(1), 71-87.
Fung, Y. C. (1965), Foundations of Solid Mechanics, Prentice-Hall, Englewood Cliffs, NJ.
Graesser, E. J. and Cozzarelli, F. A. (1991), A Multidimensional Hysteretic Model for Plastically
Deforming Metals in Energy Absorbing Devices, Technical Report NCEER-91-0006, National Center
for Earthquake Engineering Research, Buffalo, NY.
Hertzberg, R. W. (1983), Deformation and Fracture Mechanics of Engineering Materials, John Wiley
& Sons, New York, NY.
Inaudi, J. A. and de la Llera, J.C. (1992), Dynamic Analysis of Nonlinear Structures Using State-Space
Formulation and Partitioned Integration Schemes, Report UCB/EERC-92/18, University of California,
Berkeley, CA.
Jara, J. M., Gmez-Sobern, C., Vargas, E. and Gonzlez, R. (1993), Seismic Performance of
Buildings with Energy Dissipating Systems, Proc. ATC 17-1 on Seismic Isolation, Energy Dissipation,
and Active Control, 2, 663-673.
Kanaan, A. E. and Powell, G. H. (1973), DRAIN-2D - A General Purpose Computer Program for
Dynamic Analysis of Inelastic Plane Sturctures, Report No. UCB/EERC 73-06, University of
California, Berkeley, CA.
Kelly, J. M., Skinner, R. I. and Heine, A. J. (1972), Mechanisms of Energy Absorption in Special
Devices for Use in Earthquake Resistant Structures, Bull. N. Z. Soc. Earthquake Engrg., 5(3), 63-88.
Figure 4.1
Friction Dampers; a) Limited Slip Bolt Joint (Pall et al.,
1980), b) X-braced Friction Damper (Pall and Marsh, 1982)
Figure 4.1
Friction Dampers; c) Sumitomo Friction Damper
(Aiken and Kelly, 1990), d) Energy Dissipating
Restraint (Nims et al., 1993a), e) Slotted Bolted
Connection (FitzGerald et al., 1989)
In order to obtain a better appreciation for the behavior of these and other friction
dampers, we begin in Section 4.2 with a brief presentation of basic principles. These
principles are then used to some extent in Section 4.3 to construct the macroscopic
force-displacement models for the devices. The models are, in turn, required for proper
analysis and design, which inevitably involves the consideration of nonlinearities.
Section 4.4 provides a discussion of procedures applicable for overall structural
analysis. Despite a long history of interest in frictional behavior and its apparent
simplicity, purely analytical predictions are not yet possible. There is a need for
physical experiments at both the component and structural level. A number of
experimental studies employing friction dampers within test structures are described in
Section 4.5, while Section 4.6 contains additional information important for design. As
mentioned above, several full-scale structural implementations of friction dampers have
already occurred. These are discussed in further detail in Section 4.7 at the end of the
chapter.
t = n (4.2)
in terms of the tangential t and normal n, tractions. This form is also useful for
determining the nominal contact stresses that are often required for proper design. Note
that an integration of Eq. (4.2) over a planar contact area recovers Eq. (4.1).
The concept of Coulomb friction, as described above, provides the theoretical basis for
most of the work that has appeared concerning friction dampers. However, it should be
emphasized that frictional processes are seldom all that simple. In practice the Coulomb
theory is only approximately true. Furthermore, although the coefficient of friction ,
which appears in Eqs. (4.1) and (4.2), is often assumed to be a constant for a given pair
of contacting materials, this is not always the case. For example, the value of at any
instant depends not only upon the selection of sliding materials, but also on the present
condition of the sliding interface. Since surfaces are often the site of numerous ongoing
physical and chemical processes, the coefficient of friction associated with an interface
may actually vary considerably over time. Many bimetallic interfaces are particularly
susceptible to this behavior.
Although considerable gaps still remain in our understanding of frictional phenomena,
it is appropriate to briefly consider several key aspects of the more contemporary
approach. In particular, the modern theory of solid dry friction focuses on identification
of the true area of contact, the mechanisms involved in interfacial bonding, and the
localized inelastic deformation that occurs in the contact region.
Upon detailed examination, one finds that natural and engineered surfaces are not
smooth at the microscopic level, but rather contain irregularities, often categorized as
waviness and roughness. These irregularities are typically present over a wide range of
scales. As a result of this initial topography and the deformational characteristics of the
materials, true contact occurs only through the interaction of surface asperities, as
depicted in Fig. 4.2a. The diagram suggests that the true area of contact between two
mating surfaces differs substantially from the apparent contact area. Researchers have
found that a variety of topographical models involving conical, pyramidal, and
hemispherical asperities, undergoing elastic or plastic deformation, produce true
contact areas roughly proportional to the normal force. This, of course, is in general
agreement with the Coulomb theory. In any case, it becomes clear that any detailed
investigation of frictional behavior must account for the actual surface irregularities
instead of the naive idealization of perfectly smooth Euclidean objects.
When true contact does occur directly between metals, adhesive bonds form across the
interface often producing coefficients of friction > 1. However, adhesion provides a
significant contribution primarily for the contact of clean metals in a high vacuum.
Actually, the schematic presented in Fig. 4.2a is representative of this situation. For
more typical engineering applications, surface films and debris particles may also be
present at the interface as indicated in Fig. 4.2b. In particular, oxide layers readily form
under atmospheric conditions. These layers generally prevent the development of
adhesive bonds. However, their presence also greatly complicates the investigation of
frictional processes, since the mechanical characteristics of oxide films is not well
understood.
The final aspect of the modern theory concerns the local deformational processes that
occur in the vicinity of the interface. These processes involve the elastic, elastoplastic,
and possibly viscoplastic response of the asperities, surface films, debris particles, and
substrata. This can be viewed as a continuum mechanics problem. However the
geometry, as depicted in Fig. 4.2, is now quite complex. Thermal processes associated
with inelastic dissipation are activated. Additionally, brittle or ductile failure of oxide
films can significantly affect the frictional strength of the interface through the
establishment of direct metal-to-metal contact. It is obviously difficult to analyze all of
these factors. In order to obtain a more tractable problem, both geometric and
constitutive simplifications are usually introduced. For
example, a typical simulation involves the indentation and sliding of a rigid wedge over
a rigid-perfectly plastic half-space. This, along with other related contact mechanics
solutions, are presented in the recent monograph by Johnson (1987). These solutions
provide a means for estimating the effect of key material and geometric parameters on
the frictional response, but neglect many relevant factors.
Figure 4.2
Schematic of Frictional
Contact; a) Clean Metallic
Surfaces, b) Metallic Surfaces
with Oxide Films and Debris
Particles
This modern mechanistic approach to solid friction has led to an improved qualitative
understanding of the process, however a quantitative assessment of frictional response
from first principles is not yet possible. More importantly, since there is still no theory
for sliding friction comparable to the well-established theory of metal plasticity, there is
a need for much more reliance on physical testing. We will return to this issue later in
the chapter, however now attention shifts briefly to the potential effects of
environmental factors on frictional systems.
4.2.2 Environmental Effects
During slippage, localized heating of the constituent materials will definitely occur as
energy is dissipated along the interface. In some cases, these thermal effects may alter
the frictional response by causing material softening or by promoting oxidation.
However, for the type of sliding systems typically encountered in friction dampers, it is
unlikely that system response will be sensitive to the relatively small variations in
ambient temperature that can be anticipated at any site.
Of more concern are physiochemical processes, often triggered by atmospheric
moisture or contaminents, that may occur in the interfacial region. These processes may
change the physical and chemical character of the surfaces, and consequently produce a
significant impact on the frictional response. The formation of oxide layers or scale on
the exposed surfaces of metals in air has already been discussed.
In more aggressive environments, corrosion may pose a problem. Both crevice
corrosion and bimetallic corrosion are relevant to our present discussion. In the former
case, the geometry of the component is of prime importance. Exposed surfaces become
cathodic due to a surplus of oxygen, while inaccessable regions (e.g., threaded joints,
clamped surfaces) having much less dissolved oxygen become anodic and thus corrode.
On the other hand, bimetallic corrosion refers to situations involving direct electrical
contact between two dissimilar metals within a corrosive atmosphere. For example, the
corrosion rates for mild steel can be greatly enhanced when placed in contact with
copper or brass. While these basic corrosive mechanisms are well understood and
discussed at length in Shreir et al. (1994), quantitative predictions of the kinetics are not
yet possible. One must resort to physical testing to determine the extent of corrosion
expected in a given situation and to ascertain its potential effect on the frictional
characteristics of a sliding system.
Figure 4.4
Hysteresis Loops of Limited Slip
Bolted Joints (Pall et al., 1980)
Of the systems considered by Pall et al. (1980), the maximum static slip load was
obtained for metalized surfaces. However, the cyclic response was quite erratic, with
considerable stick-slip associated with the transition from static to kinetic frictional
response. Similar cyclic behavior was obtained for contact between steel plates with
mill scale or sand blasted surfaces. Zincrich painted surfaces and polyethylene coatings
produced a smoother cyclic response, however degradation of the slip load occurred.
On the other hand, as is apparent from the figures, the systems containing heavy duty
brake lining pads inserted between steel plates did provide a consistent, predictable
response. Experimental results from a number of other frictional materials were
reported more recently by Anagnostides and Hargreaves (1990).
It is perhaps not surprising that brake lining materials would perform well in the
dynamic tests. After all, these materials have been developed over a great number of
years in the automotive industry specifically to provide reliable frictional response.
However, the use of brake lining pads does not guarantee suitable performance. For
example, Tyler (1985) presented test results for a brake lining damper that exhibited
significant stick-slip. Mechanical design of the damper is also important. In particular,
bolt tensions must be maintained and various geometric tolerances must be controlled.
Based upon the behavior obtained by Pall et al. (1980) and illustrated in Fig. 4.4e,
characterization of their simple brake lining frictional system in terms of an elastic-
perfectly plastic model is quite appropriate. The model employed to simulate the
behavior of their Limited Slip Bolted (LSB) joint is shown in Fig. 4.5a, b. Notice that
the model sketched in those diagrams includes a stiff bearing stage for displacements
beyond the slip length. In order to quantify response, a mathematical model must also
be provided. A suitable hysteretic model is defined in Table 4.1, where the parameters
Ps and b represent the slip load and the total displacement at first contact with the
bearing surface, respectively. Additionally, K0 is the elastic stiffness, while the stiffness
in bearing is K2. The model, which assumes zero stiffness
Figure 4.5
Macroscopic Model for Limited Slip Bolted Joints (Pall et al., 1980)
Figure 4.5
Macroscopic Model for Limited Slip
Bolted Joints (continued); c) Parameter
Definition and Sample Result
during slippage (i.e., K1 = 0), is specified in rate form in order to properly address
arbitrary loading-unloading histories. (The superposed dot represents differentiation
with respect to time or a pseudotime loading parameter.) A typical result for
displacement controlled cycling at two different amplitudes is displayed in Fig. 4.5c,
where all of the model parameters are detailed.
Subsequently, Pall and Marsh (1982) proposed a system in which the braces in a
moment resisting frame incorporated frictional devices. These devices utilize the same
heavy duty brake lining pads discussed above. In a typical X-braced system, the braces
are designed to buckle at relatively low compressive loads. As a result, the braces
contribute only when subjected to tension. By installing uniaxial friction elements
within each brace, slippage would only occur in the tensile direction and very little
energy dissipation would result during cyclic loading. However, the special damper
mechanism, devised by Pall and Marsh (1982) and depicted in Fig. 4.1b, permits much
more effective
operation. During cyclic loading, the mechanism tends to straighten buckled braces and
also enforces slippage in both tensile and compressive directions.
Initially Pall and Marsh (1982) used a simple elastoplastic model to represent the
behavior of this X-braced friction damper. However, Filiatrault and Cherry (1987)
determined that this is only valid if the device slips during every cycle, and if that the
slippage is always sufficient to completely straighten any buckled braces. Otherwise,
the Pall-Marsh model overestimates the energy dissipation. To remedy this situation,
Filiatrault and Cherry (1987) proposed a more detailed macroscopic model for the
device. A schematic is provided in Fig. 4.6. Each member of the bracing-damper
system is represented by elements reflecting its individual axial and bending
characteristics. Thus, the structural braces are assumed to yield in tension, but buckle
elastically in compression. The device links are permitted to yield in both tension and
compression, while the sliding brake pads are represented by a hysteretic model
corresponding to the experimental results obtained by Pall et al. (1980). Bending
stiffness is included to maintain stability of the damper mechanism.
Figure 4.6
Refined Model for X-braced Friction
Damper (Filiatrault and Cherry, 1987)
In addition to developing the refined model, Filiatrault and Cherry (1987) conducted
some physical experiments. In particular, they examined the response of the X-braced
damper, with heavy duty asbestos brake lining pads, subjected to cyclic displacement-
controlled loading along one diagonal. Results are shown in Fig. 4.7a. Notice that,
although a consistent response is still maintained over a total of fifty cycles with a
displacement amplitude of 15mm, the hysteresis loops are no longer perfectly
rectangular. The notches
appearing in the two corners of the loops were attributed to the difference between the
diameters of the bolts and bolt holes. In fact, the results shown in the figure were
obtained after inserting steel bushings in the corner holes and slots of the device. From
this investigation, it is clear that seemingly minor fabrication details can affect the
overall performance of the friction damper.
A hysteretic model can also be developed for this imperfect Coulomb friction element,
which essentially involves multilevel bearing stops. The model requires the
specification of three parameters (i.e., P1, 1, K1) beyond those required for the single
level friction element. Additionally, an internal variable , representing position within
the inner slip length, is introduced. Table 4.2 contains an explicit definition of the
model, while typical results are displayed in Fig. 4.7b. (The shape of the virgin loading
curve depends upon the initial value assumed for .) Although the model is somewhat
complicated, it is suitable for examining the response of the X-braced damper friction
element under arbitrary loading.
Figure 4.7
Hysteresis Loops for X-braced Friction Damper;
a) Experimental (Filiatrault and Cherry, 1987), b)
Hysteretic Model
As mentioned previously, several simple friction damper designs have been proposed in
the recent literature. For example, Roik et al. (1988) discuss seismic control of
structures through the use of three-stage friction-grip elements. Both concrete-steel and
steel-steel grip elements are considered. A test specimen for the latter type is detailed in
Fig. 4.8a, while the hysteresis loops obtained during a displacement-controlled testing
procedure are reproduced in Fig. 4.8b. The authors note that a consistent response is
obtained without the complexity or cost of including brake lining materials, provided
proper care is exercised in the joint design. For example, nonlinear disc springs (or
Belleville washers) are employed to maintain bolt prestress. It
should be noted, however, that these results are for displacement amplitudes of less
than 1mm. Additionally, a close comparison reveals cycle-to-cycle variability, and
unlike an ideal Coulomb friction element, the slip load is amplitude dependent. Putting
aside the details of the measured response, Roik et al. (1988) then propose that friction
elements with varying degrees of stiffness, prestress, and allowable slip be combined in
order to obtain a smooth transition from the sticking state to the slipping state as
illustrated in Fig. 4.9. This model subsequently provides the basis for an investigation
of anticipated structural response.
Figure 4.8
Steel-Steel Friction Grip Connection (Roik et
al., 1988); a) Test Specimen, b) Hysteresis Loops
Figure 4.9
Hypothetical Three-Stage Stiffening Element (Roik et al., 1988)
Another simple conceptual design, the Slotted Bolted Connection (SBC), shown in Fig.
4.1e, was investigated by FitzGerald et al. (1989). Experimental results obtained under
sizeable displacement controlled loading are provided in Fig. 4.10. In State 1, the
gusset slips relative to the channel plates, while in State 2 there is also slippage between
the channel and cover plates. The model defined previously in Table 4.2 can be utilized
to simulate this behavior for the purposes of structural analysis. It should be noted that
the consistent experimental response present in Fig. 4.10 is attributed by the authors to
the inclusion of Belleville spring washers. On the other hand, Grigorian et al. (1993)
also employ Belleville washers in an SBC design, but obtain more erratic behavior as
indicated in Fig. 4.11. Clearly this latter type of response from a critical structural
component must be deemed unacceptable. Additionally, as discussed previously,
considerable care must be exercised in selecting materials for use in a bimetallic
interface.
Figure 4.10
Typical Load Deformation Diagram for Slotted
Bolted Connections (FitzGerald et al., 1989)
Figure 4.11
Cyclic Response of Slotted Bolted Connections (Grigorian et al., 1993)
The uniaxial friction damper illustrated in Fig. 4.1c, manufactured by Sumitomo Metal
Industries Ltd., utilizes a slightly more sophisticated design. The pre-compressed
internal spring exerts a force that is converted through the action of inner and outer
wedges into a normal force on the friction pads. These copper alloy friction pads
contain graphite plug inserts, which provides dry lubrication. This helps to maintain a
consistent coefficient of friction between the pads and the inner surface of the steel
casing. Aiken and Kelly (1990) indicate that the response of these dampers is extremely
regular
and repeatable with rectangular hysteresis loops. Furthermore, the effect of loading
frequency and amplitude, number of cycles, or ambient temperature on damper
response was reported to be insignificant. Although the supportive component test data
is not available in the literature, it would seem that a simple elastic-perfectly plastic
hysteretic model, defined above in Table 4.1, is appropriate for structural analysis
involving this device.
The final friction damper to be considered is the Energy Dissipating Restraint (EDR)
manufactured by Fluor Daniel, Inc. and detailed in Fig. 4.1d. Superficially the design is
similar to the Sumitomo concept, since this device also includes an internal spring and
wedges encased in a steel cylinder. However, there are several novel aspects of the
EDR that combine to produce very different response characteristics. A detailed
presentation of the design and its performance is provided in Nims et al. (1993a). As
indicated in Fig. 4.1d, the EDR utilizes steel compression wedges and bronze friction
wedges to transform the axial spring force into a normal pressure acting outward on the
cylinder wall. Thus, the frictional surface is formed by the interface between the bronze
wedges and the steel cylinder. Internal stops are provided within the cylinder in order to
create the tension and compression gaps that are illustrated in Fig. 4.1d. Consequently,
unlike the Sumitomo device, the length of the internal spring can be altered during
operation, providing a variable frictional slip force. Typical experimental hysteretic
behavior is displayed in Fig. 4.12 for three different configurations. Figure 4.12a
represents the response obtained with zero gaps and zero spring preload. Triangular
shaped hysteresis loops result indicating slip force proportional to the device
displacement. With non-zero spring preload and very large gaps, the device acts as a
standard Coulomb damper as indicated in Fig. 4.12b. The model presented previously
in Table 4.1 is obviously applicable for this second case. Finally, with a non-zero
preload, but no initial gaps, the flag-shaped hysteresis loops of Fig. 4.12c are obtained.
Clearly, from Fig. 4.12, the repsonse characteristics of the EDR are quite different from
those of the other friction dampers. Consequently, it is appropriate to briefly discuss the
underlying mechanics. Consider the case associated with Fig. 4.12a having zero gaps
and zero preload. This configuration corresponds to that depicted schematically in Fig.
4.13, except with G1 = 0. In its initial state, due to a zero internal spring force, there
exists no normal contact pressure acting between the wedges and casing. However,
once a force P is applied in either tension or compression, the spring, with stiffness Ks,
is compressed and frictional resistance results. Let the spring displacement be
represented by s, while the overall displacement of the device is , which includes
deformation of the rod and connections r. Thus,
= s + r (4.3)
If the stiffness of the rod and connections equals K3, then
Figure 4.12
Experimental Data for EDR (Richter et al., 1990)
Figure 4.13
EDR Configuration Schematic (Nims et al., 1993a)
P = K3r = K1 (4.4a, b)
where K1 is the effective overall stiffness of the device during initial loading.
Furthermore, the spring force becomes
Ps = Kss (4.5a)
and the corresponding frictional resistance during slippage can be written
Pf = Kss (4.5b)
The positive factor a, which is less than one for practical designs, incorporates the
geometric and Coulomb friction effects involved in transforming the action of the
spring force through the wedges into a frictional resistance. For slippage during
loading, equilibrium requires that
P = Ps + Pf (4.6)
Consequently, from Eqs. (4.3)-(4.6), one obtains the following expression for the
effective stiffness:
Note that in this configuration, the EDR device is self-centering. In the absence of
external force, the internal spring will return to its initial zero preload state.
A detailed description of an appropriate hysteretic model is presented in Table 4.3.
Results obtained from the model for displacement controlled cyclic loading are shown
in Fig. 4.14a. It should be noted that this model also permits the simulation of response
due to arbitrary loadings, which may include partial unloading-reloading cycles. Some
care is required, however, in the numerical implementation, particularly for 1. An
extended version can also be developed to incorporate the effects of initial preload,
indicated in Fig. 4.12c, by including an internal variable corresponding to s. While a
detailed presentation is not included here, typical numerical results obtained from the
model are provided in Fig. 4.14b. The device in this configuration is again self-
centering.
Figure 4.14
Hysteretic Model for EDR; a)
No Gaps, No Prelaod, b)
No Gaps, Finite Preload
Before shifting attention to the overall analysis of structures that incorporate friction
dampers, it is perhaps appropriate to mention a couple of aspects of damper response
that have not been addressed adequately in the literature. In reviewing the available
published component test data, one finds that nearly all dampers have been tested at a
single, relatively low frequency. Classical Coulomb behavior is assumed, but not
demonstrated. In particular, there have been no systematic attempts to investigate the
effects of amplitude and frequency on response. Additionally, an examination of the
consistency of damper response after long periods of inactivity has not been reported
for any of the proposed dampers. Consequently, some care is needed in adopting one of
the models presented above in Tables 4.1-4.3, since all assume Coulomb damping with
a constant coefficient of friction.
Figure 4.15
Numerical Results of LSB Joint for
El Centro Ground Motion, Ratio of
LBS Jointed Wall to Elastically
Jointed Wall (Pall et al., 1980);
a) Normal Stress, b) Deflection
Pall and Marsh (1982) presented similar numerical trends for a tenstory steel frame
supported on a rigid foundation, again subjected to the 1940 El Centro earthquake.
Results for three configurations, consisting of a moment resisting (MR) frame, a cross-
braced moment resisting (BMR) frame, and a friction damped braced (FDB) frame,
were compared. The DRAIN-2D analysis included consideration of tensile yielding and
compressive buckling of primary frame members, along with their simple hysteretic
friction damper model. Zero viscous damping was assumed for the structure in all
three configurations. The resulting deflection and column force envelopes are
reproduced in Fig. 4.16. The addition of friction dampers results in significant
reductions in interstory drifts. Reduced column shear and bending moment are also
apparent. On the other hand, axial forces are greater than those obtained for the moment
resisting frame. For the particular case considered, plastic hinges form in beam
members of the MR and BMR designs, while the primary members in the FDB with
optimal slip loads remain elastic. The authors note, however, that results obtained for a
single seismic record may not be conclusive.
Figure 4.16
Numerical Results of X-braced Friction Damper for
El Centro Ground Motion (Pall and Marsh, 1982)
Filiatrault and Cherry (1987) considered several different earthquake inputs in their
combined numerical and experimental investigation of a three-story, one-third scale
steel FDB frame. In the numerical DRAIN-2D analysis, the model was subjected to
1940 El Centro S00E at 0.52g, 1966 Parkfield N65E at 0.52g, Newmark-Blume-Kapur
artificial signal at 0.30g, and band-limited white noise at 0.50g. Optimal slip loads for
the friction dampers were determined via a parametric study, and found to be
reasonably independent of the seismic signal. Results were then compared with those
obtained for the MR frame and BMR frame configurations, which correspond to the
zero and infinite slip load cases, respectively. Two different friction damper models
were considered in the analyses; the simple Pall-Marsh model of Fig. 4.5b, and the
refined model depicted in Fig. 4.6. Although the latter more realistic model produced a
significantly different response, the FDB frame still performed better than the
traditional MR and BMR frames. For example, typical column shear force envelopes
are illustrated in Fig. 4.17.
Figure 4.17
Numerical Column Shear Force Envelopes for
Three-story Frame (Filiatrault and Cherry, 1987)
Figure 4.18
Predicted Damage in Ten-story
Frame (Filiatrault and Cherry, 1988)
In other related analysis work, Baktash and Marsh (1987) conducted a comparative
study of the performance of friction damped braced frames (FDBF) and eccentric
braced frames (EBF). The authors found that while both systems were adequate, the
FDBF design produced smaller deflections, accelerations, and forces. An innovative
friction damped timber shear wall concept was examined by Filiatrault (1990) and also
found to be effective in reducing seismic response.
More recently, numerical and theoretical investigations on the response of structural
systems incorporating EDR devices have appeared. Nims et al. (1993a) conducted
parametric studies on idealized SDOF systems with added frictional elements that
produce either the flag-shaped or triangular-shaped hysteresis loops discussed in
Section 4.3. The parameters considered include the frequencies of the braced and
unbraced structure, unbraced damping ratio , EDR device initial slip load P1 as a
percentage of the structural weight, and amplitude and frequency content of the seismic
signal. Fig. 4.19 provides a typical set of results for devices with flag-shaped hysteresis
loops. In those plots the acceleration response, due to the El Centro and Zacatula
earthquakes, is plotted versus unbraced structural frequency with
Figure 4.19
Numerical Response of SDOF Structure with EDR (Nims et al., 1993a)
unbraced damping ratio = 0.05. The frequency of the braced structure is assumed, in
all cases, to equal twice the unbraced frequency. With reference to Fig. 4.14b, the
device stiffness ratio K1/K3 = 10. From Figs. 4.19a, c, it is apparent that the addition of
frictional devices reduces response, except for structures in the low frequency range.
(Results for an unbraced structure with = 0.10 are also included in those figures to
illustrate the effects of adding purely viscous damping mechanisms.) The variation of
response with initial slip load is quantified in Figs. 4.19b, d.
The above EDR results are for an SDOF system. However, the response of a six-story,
0.3-scale steel MDOF structure, subjected to the El Centro and Zacatula ground
motions, has been reported in Nims et al. (1993b). Comparisons, made with the
corresponding unbraced and conventionally braced frames, indicate that the friction
dampers effectively reduce displacements, while maintaining comparable acceleration
levels.
Nonlinear analysis methods have been employed out of necessity in all of the efforts
referred to in this section to describe the behavior of structural systems that incorporate
friction dampers. In recent work, Inaudi et al. (1993a) have developed some interesting
and potentially useful methods to approximate the response of systems that include
EDR devices with triangular-shaped hysteresis loops. The authors note that although
these frictional elements are nonlinear and the principle of superposition does not
apply, the response is scale invariant. Thus, if a deformation history is scaled by a
certain factor, the forces in the device are multiplied by that same factor. This permits
the effective use of the techniques of harmonic linearization and statistical linearization
(Inaudi et al., 1993a, b) to develop estimates of the actual nonlinear response. In
general, very good results are obtained, indicating that these methods may be suitable
for preliminary design calculations. It should be noted, however, that the use of
discontinuous memory functions within a force-displacement model can lead to
difficulties under arbitrary excitations which include partial unloading-reloading cycles.
Figure 4.20
Response of Three-Story Test Frame for
Taft 0.60g (Filiatrault and Cherry, 1987)
Figure 4.21
Measured Third Floor Accelerations for
Taft 0.90g Input (Filiatrault and Cherry, 1987)
The experimental program designed by Filiatrault and Cherry (1987), and described
briefly above, was well conceived and quite comprehensive. However, one minor point
must be made concerning the use of full scale cross-braced friction devices in a one-
third scale structure. Even though slip loads were adjusted to appropriate levels, this
scaling imbalance reduces the contribution of any geometric nonlinearities that may be
associated with the device. It is difficult to determine the significance of that
contribution without further testing or detailed modeling.
An additional experimental study on the cross-braced dampers was conducted by Aiken
et al. (1988). A three-bay, nine-story, one-quarter scale steel structure was extensively
tested on an earthquake simulator, in both MR and FDB frame configurations. The
dissipating elements in the FDB design utilized a brake lining pad/stainless steel
frictional interface. Low amplitude natural frequencies and damping ratios were first
determined for both configurations. The fundamental frequencies were 2.0 Hz and 2.23
Hz for the MR and FDB frames, respectively, while the corresponding damping ratios
were calculated at 2.4% and 5.6%. The structure was subjected to a total of ten different
seismic signals with varying magnitude. As expected, for a given earthquake, an
increase in magnitude was found to increase the effectiveness of the friction dampers.
For example, the ratio of roof-to-ground acceleration dropped from 3.1 to 2.0 as the
peak ground acceleration of an El Centro time history was elevated from 0.30g to
0.84g. At that highest intensity, the FDB structure had an estimated equivalent damping
ratio of 32%.
In order to compare performance of the MR and FDB frames, Aiken et al. (1988) used
temporal scaling of the 1985 Mexico City ground motion to achieve a quasi-resonance
response for each configuration. The predominant
signal frequency was adjusted to coincide with the fundamental frequency of the
structure. For the MR frame, a peak ground acceleration of 0.25g then produced a
maximum displacement of 3.1 in., while a 0.65g peak acceleration resulted in a
displacement of only 2.8 in. for the friction-damped frame. Furthermore, in the latter
case, approximately 70% of the input energy was dissipated by the frictional devices.
The primary structural members remained in the elastic range.
More recently, using the same nine-story test frame, Aiken and Kelly (1990) performed
a combined experimental and numerical investigation of uniaxial Sumitomo friction
elements installed in conjuction with chevron bracing. A comprehensive test program
was conducted, involving free vibration, random excitation, and pulse loadings, along
with a total of fourteen different seismic ground motions. Only a brief summary can be
provided here, but all test results are thoroughly documented in the referenced report.
The authors note that the performance of the friction dampers was outstanding. The
hysteresis loops indicated very consistent, nearly ideal Coulomb behavior throughout
the duration of the test program. For example, the response at a number of friction
damper locations, for the El Centro ground motion at 0.712g peak acceleration, are
reproduced in Fig. 4.22. Approximately 60% of the input energy is dissipated in the
dampers. Meanwhile, the corresponding floor
Figure 4.22
Response of Sumitomo Dampers in Nine-Story Frame
for El Centro 0.712g Input (Aiken and Kelly, 1990)
accelerations, displacements, and interstory drifts are presented in Fig. 4.23, along with
those for the conventional MR and BMR frames. It is apparent from these profiles that
the FDB structure nearly attains the benefits of the reduced displacement response of
the BMR frame, combined with the acceleration response of the MR frame. The report
also contains details of a numerical study and comparisons with responses obtained
using viscoelastic dampers in the same structural frame.
Figure 4.23
Overall Response Comparison in Nine-Story Frame
for El Centro 0.712g Input (Aiken and Kelly, 1990)
The final experimental study, to be discussed in this section, concerns the behavior of
the EDR device. Richter et al. (1990) describe results from a series of over 400 shaking
table tests conducted on a small, 6ft. high, three-story steel frame structure. Although a
number of different internal EDR damper configurations were considered, in all cases,
two devices were mounted in each story. Typical hysteretic response of the dampers,
configured with no gaps and an initial slip load of 200lbs., is shown in Fig. 4.24 for a
scaled Zacatula ground motion. The flag-shaped loops are well-defined and quite
consistent. Cumulative energy time histories are provided in Fig. 4.25a and b for
Zacatula and El Centro signals, respectively. In both cases, the frictional devices
dissipate over 90% of the total input energy. However, as noted by the authors,
hysteretic mechanisms do not respond quickly to sudden impulses. This is evident in
these energy response curves. Additionally higher modes were sometimes excited due
to sudden stiffness changes associated with the frictional devices. Despite these
limitations, the EDR device consistently provided reductions in displacements and
interstory drifts, and increased the effective damping ratio of the test structure.
Figure 4.24
EDR Response in Three-story Test Frame for
Zacatula Ground Motion (Richter et al., 1990)
Figure 4.25
Cumulative Energy Time History for EDR
Damped Three-story Test Frame (Richter et al., 1990)
4.6 Design Considerations
Several aspects of the design problem have already been discussed. For example, in
Section 4.4, a number of works were reviewed in which the optimal device slip loads
were determined by direct enumeration, using multiple nonlinear analyses. As the
structural systems become more complicated and the number of unknown parameters
increases, this type of approach becomes less attractive. Furthermore, there is no
guarantee that the resulting design will indeed satisfy all of design criteria or code
requirements. Obviously, there is a need to develop more systematic design
methodologies.
Filiatrault and Cherry (1990) present an approach, intended for practicing engineers,
based upon development of design slip-load spectra. In previous work (Filiatrault and
Cherry, 1989), the authors developed a streamlined friction-damped braced frame
analysis program (FDBFAP) to provide a tool for efficient determination of optimal
slip loads. This program includes a detailed model for cross-braced friction dampers,
but assumes that all remaining structural members remain elastic. Additionally, the
program computes a relative performance index (RPI) that provides an estimate of the
response of the friction damped structure compared to that of the corresponding
unbraced frame. More specifically, the RPI is an average of the peak and time-
integrated strain energy responses. Optimum slip load is then assumed to be associated
with the minimum value of RPI.
The program FDBFAP is used in Filiatrault and Cherry (1990) to perform an extensive
series of analyses to establish the functional relationship between design slip load and
the key parameters that define a particular structure. The proposed relationship can be
written:
Vs = f(Tb/Tu, Tg/Tu, N)mag (4.9)
where Vs is the total slip shear load in all friction dampers in the N-story structure, m is
the total structural mass, while Tb, Tu, and Tg represent the braced, unbraced, and
predominant ground periods, respectively. The parameter ag is the design peak ground
acceleration. A total of 45 different structural configurations, ranging from one to ten
stories, were examined for various values of the parameters Tb/Tu, Tg/Tu and ag. Each
structure was subjected to a series of five different artificially generated earthquakes.
Results were then curve fit to produce bilinear design slip load spectra for each
particular set of values for Tb/Tu and N. A typical spectrum is illustrated in Fig. 4.26.
The design procedure for framed structures proposed by the authors can be summarized
as follows:
Step 1 Design the unbraced frame by considering all loads, except those associated
with significant earthquakes.
Step 2 Determine Tu, and choose cross braces such that Tb< 0.4Tu, if possible.
Estimate ag and Tg for the particular building site.
Step 3 Verify that the parameters Tb/Tu, Tg/Tu, ag/g, and N fall within the range of
the parametric studies. If not, perform detailed dynamic analysis using FDBFAP to
determine the optimal slip loads.
Step 4 Construct the design slip load spectrum for the particular values of Tb/Tu
and N.
Step 5 Utilize the design slip load spectrum to determine Vs and distribute this slip
shear load uniformly to all stories.
Step 6 Distribute the story slip shear load to the friction dampers in each story, and
from geometric considerations, determine the device slip loads.
Step 7 Check to insure that wind loads do not cause devices to slip. If necessary,
modify the unbraced frame design and return to Step 2, or modify slip loads and
perform detailed dynamic analysis.
Step 8 Check to insure that cross braces do not yield in tension before slipping
occurs. If necessary modify design.
A detailed design example is also provided in Filiatrault and Cherry (1990). Although
not stated by the authors, it is necessary to conduct a detailed nonlinear analysis of the
final design in order to verify that all members are properly designed.
Figure 4.26
Typical Design Slip-Load Spectrum
(Filiatrault and Cherry, 1990)
While the above procedure is appropriate for the present traditional design office, with
the widespread availability of engineering workstations, it should be possible to employ
more automated approaches for structural design. In fact, more than one decade ago,
Austin and Pister (1985) developed an optimization-based design approach for friction
damped braced frames. With this approach, the engineer explicitly specifies the criteria
that must be satisfied. For example, member bending moments are constrained to limit
yielding, axial loads are restricted to prevent buckling, and structural drift limitations
are also imposed. The authors consider the response to gravity loads alone and in
conjuction with a moderate or severe earthquake. A method of feasible directions is
used for the optimization method. This ensures that an acceptable design is obtained at
each iteration.
As an example, Austin and Pister (1985) examine the design of a tenstory friction
damped steel frame structure. Included are some preliminary analyses to determine
initial member sizes and a condensed list of potentially active constraints. The latter
was necessary to reduce computing effort. The optimization program
DELIGHT.STRUCT was then used to find an improved design which attempted to
minimize volume, primary member hysteretic energy, and interstory drifts. The result
was not only an improved design, but additionally the sensitivity of various design
parameters was explored. This problem was analyzed on a VAX 11/780 computer.
Each design iteration required over 4hr. of run time. This would appear to be
prohibitive. However, in 1997, the same analysis would require only 2min. of run time
on a modern engineering workstation. From all indications, this trend will continue well
into the next century. This suggests that an optimization-based design approach is quite
feasible, even for the small design office, provided that the proper software is
developed.
Before closing this section, the other design issue that must be raised concerns the
predictability and reliability of friction damper performance. If one considers the design
procedure devised by Filiatrault and Cherry (1990), it is clear that the integrity of the
entire structure during a major earthquake is dependent to a large degree on the
performance of the friction dampers. These dampers function as critical elements in the
aseismic protective system. Consequently, their performance must be established not
only at the time of installation, but also throughout their intended design life. Since
there are no reliable theories upon which to base an assumption of consistent frictional
response, this must be established by physical testing. There is an urgent need to
conduct such tests.
4.7 Structural Implementations
In recent years, there have been several commercial applications of friction dampers
aimed at providing enhanced seismic protection of new and retrofitted structures. This
activity is primarily associated with the use of Pall friction devices in Canada and
Sumitomo friction dampers in Japan. The Pall friction dampers, shown in Fig. 4.27, are
updated versions of the cross-braced devices described previously. As documented in
Pall and Pall (1996), several variations of these devices have been installed in thirteen
buildings; six retrofits and seven new facilities. Several of these applications are
described below in more detail.
Figure 4.27
Pall Friction Dampers (Pall et al., 1993)
Figure 4.29
Exposed Friction Damper in McConnel Library
Galleria (Pall and Pall, 1993)
The application of friction dampers to the main headquarters building of the Canadian
Space Agency Complex near Montreal is described in Vezina et al. (1992) and Pall et
al. (1993). An aerial view of the entire
complex is shown in Fig. 4.30. The headquarters building is a three-story steel framed
structure clad with aluminum panels. Since it contains sensitive equipment and
instrumentation, additional protection is required from the potentially damaging effects
of earthquakes. Three-dimensional nonlinear dynamic analyses were conducted by the
authors using DRAIN-TABS to determine the suitability of employing friction
dampers. Input included several artificially generated seismic signals scaled to produce
a peak ground acceleration of 0.18g. Comparison with traditional braced and unbraced
configurations indicated a superior performance for the friction damped design. Based
upon these results, a total of 58 frictional devices were specified, each with a slip load
of 500kN. The cross-braced frame bays with friction dampers are distributed evenly
throughout the building, as illustrated for the first story in Fig. 4.31. Several dampers
were intentionally exposed for viewing purposes.
Figure 4.30
Aerial View of Canadian Space
Agency Complex (Pall et al., 1993)
Figure 4.31
Ground Floor Plan of Main Building in
Canadian Space Agency (Pall et al., 1993)
The three school buildings at Ecole Polyvalante near Montreal, pictured in Fig. 4.32,
were damaged in the 1988 Saguenay earthquake. The original structure, built in 1967,
consists of precast concrete beams and columns with welded connections. Floor and
roof panels are also precast concrete. The result is a structure with little lateral
resistance. In fact, analysis indicated that the existing structure could not withstand an
earthquake with a peak acceleration of 0.05g. Two different retrofit schemes were
considered; a conventional method using concrete shearwalls, and an approach
employing friction dampers. The latter was chosen because it resulted in a 40%
reduction in retrofitting cost and a 60% reduction in construction time. Based upon
results from nonlinear time domain dynamic analyses, a total of 64 friction dampers
were specified, along with 388 friction-based panel connectors. The project was
completed during the summer break in 1990. Further details are provided in Pall and
Pall (1993).
Figure 4.32
Exterior View of Ecole Polyvalante (Pall and Pall, 1993)
The eight story steel structure, shown in Fig. 4.33, was originally constructed by the
French government for EXPO'67 held in Montreal. After the exhibition, the building
was donated to the city, and in 1992 a decision was made by Lotto-Quebec to
rehabilitate the structure to house 'Casino de Montreal.' Preliminary analysis indicated
that the building would no longer satisfy the seismic code requirements, which had
been updated since the time of the original construction. Both conventional and
friction-damped retrofit designs were considered. However, the former required
considerable pile foundation work, which is expensive and time consuming. The
friction-damped alternative employed 32 devices throughout the structure, based upon
results from nonlinear dynamic analysis. In this case, a three-dimensional analysis was
essential due to the eccentricities of the original structure. Additional information on
this retrofit project can be found in Pasquin et al. (1994).
Figure 4.33
North View of Casino de Montreal (Pasquin et al., 1994)
Three building projects in Japan, involving Sumitomo friction dampers, are briefly
described in the report by Aiken and Kelly (1990). The first is the 31-story steel frame
Sonic Office Building in Omiya City, constructed in 1988. A total of eight 22kip
dampers were utilized on each story, primarily to reduce the effect of ground-borne
vibration and small earthquakes. A similar motivation led to the use of four 22kip
dampers per level in the Asahi Beer Azumabashi Building in Tokyo. This 22-story
braced steel frame structure was completed in 1989. The third project mentioned
involves the use of Sumitomo friction dampers as part of a base isolation system for a
six-story reinforced concrete structure in Tokyo.
References
Aiken, I.D. and Kelly, J. M. (1990), Earthquake Simulator Testing and Analytical
Studies of Two Energy-Absorbing Systems for Multistory Structures, Report No.
UCB/EERC-90/03, University of California, Berkeley, CA.
Aiken, I.D., Kelly, J. M. and Pall, A. S. (1988), Seismic Response of a Nine-Story
Steel Frame with Friction Damped Cross-Bracing, in Report No. UCB/EERC-
88/17, University of California, Berkeley, CA.
Anagnostides, G. and Hargreaves, A. C. (1990), Shake Table Testing on an Energy
Absorption Device for Steel Braced Frames, Soil Dyn. Earthquake Engrg., 9(3),
120-140.
Austin, M. A. and Pister, K. S. (1985), Design of Seismic-Restraint Friction-
Braced Frames, J. Struct. Engrg., ASCE, 111(12), 2751-2769.
Baktash, P. and Marsh, C. (1987), Damped Moment-Resistant Braced Frames: A
Comparative Study, Can. J. Civ. Engrg., 14, 342-346.
Bowden, F. P. and Tabor, D. (1950), The Friction and Lubrication of Solids, Part I,
Clarendon Press, Oxford.
Bowden, F. P. and Tabor, D. (1964), The Friction and Lubrication of Solids, Part
II, Clarendon Press, Oxford.
Filiatrault, A. (1990), Analytical Predictions of the Seismic Response of Friction Damped Timber
Shear Walls, Earthquake Engrg. Struct. Dyn., 19, 259-273.
Filiatrault, A. and Cherry, S. (1987), Performance Evaluation of Friction Damped Braced Frames
Under Simulated Earthquake Loads, Earthquake Spectra, 3(1), 57-78.
Filiatrault, A. and Cherry, S. (1988), Comparative Performance of Friction Damped Systems and
Base Isolation Systems for Earthquake Retrofit and Aseismic Design, Earthquake Engrg. Struct.
Dyn., 16, 389-416.
Filiatrault, A. and Cherry, S. (1989), Efficient Numerical Modelling for Seismic Design of Friction
Damped Braced Steel Plane Frames, Canadian J. Civ. Engrg., 16(3), 211-218.
Filiatrault, A. and Cherry, S. (1990), Seismic Design Spectra for Friction-Damped Structures, J.
Struct. Engrg., 116(5), 1334-1355.
FitzGerald, T. F., Anagnos, T., Goodson, M., and Zsutty, T. (1989), Slotted Bolted Connections in
Aseismic Design for Concentrically Braced Connections, Earthquake Spectra, 5(2), 383-391.
Grigorian, C. E., Yang, T. S. and Popov, E. P. (1993), Slotted Bolted Connection Energy
Dissipators, Earthquake Spectra, 9(3), 491-504.
Guendeman-Israel, R. and Powell, G. H. (1977), DRAIN-TABS- A Computerized Program for
Inelastic Earthquake Response of Three Dimensional Buildings, Report No. UCB/EERC 77-08,
University of California, Berkeley, CA.
Inaudi, J. A., Nims, D. K. and Kelly, J. M. (1993a), On the Analysis of Structures With Energy
Dissipating Restraints, Report No. UCB/EERC 93-13, University of California, Berkeley, CA.
Inaudi, J. A., Kelly, J. M. and To, C. W. S. (1993b), Statistical Linearization Method in the
Preliminary Design of Structures with Energy Dissipating Devices, Proc. ATC 17-1 on Seismic
Isolation, Energy Dissipation, and Active Control, 2, 509-520.
Johnson, K. L. (1987), Contact Mechanics, Cambridge University Press, Cambridge, UK.
Kanaan, A. E. and Powell, G. H. (1973), DRAIN-2D - A General Purpose Computer Program for
Dynamic Analysis of Inelastic Plane Structures, Report No. UCB/EERC 73-06, University of
California, Berkeley, CA.
Larsen-Basse, J. (1992), Basic Theory of Solid Friction, in Friction, Lubrication and Wear
Technology, ASM International Handbook Committee, American Society for Metals, Materials
Park, Ohio.
Nims, D. K., Richter, P. J. and Bachman, R. E. (1993a), The Use of the Energy Dissipating
Restraint for Seismic Hazard Mitigation, Earthquake Spectra, 9(3), 467-489.
Nims, D. K., Inaudi, J. A., Richter, P. J. and Kelly, J. M. (1993b), Application of the Energy
Dissipating Restraint to Buildings, Proc. ATC 17-1 on Seismic Isolation, Energy Dissipation, and
Active Control, 2, 627-638.
Pall, A. S. and Marsh, C. (1981), Friction-Damped Concrete Shearwalls, Amer. Concrete Inst. J.,
78(3), 187-193.
Pall, A. S. and Marsh, C. (1982), Response of Friction Damped Braced Frames, J. Struct. Div.,
ASCE, 108(ST6), 1313-1323.
Pall, A. S., Marsh, C. and Fazio, P. (1980), Friction Joints for Seismic Control of Large Panel
Structures, J. Prestressed Concrete Inst., 25(6), 38-61.
Pall, A. S. and Pall, R. (1993), Friction-Dampers Used for Seismic Control of New and Existing
Buildings in Canada, Proc. ATC 17-1 on Seismic Isolation, Energy Dissipation, and Active Control,
2, 675-686.
Pall, A., Vezina, S., Proulx, P. and Pall, R. (1993), Friction-Dampers for Seismic
Control of Canadian Space Agency Headquarters, Earthquake Spectra, 9(3), 547-
557.
Pall, A. S. and Pall, R. (1996), Friction-Dampers for Seismic Control of Buildings:
A Canadian Experience, Proceedings of the Eleventh World Conference on
Earthquake Engineering, Acapulco, Mexico.
Pasquin, C., Pall, A. and Pall, R. (1994), High-Tech Seismic Rehabilitation of
Casino de Montreal, Structures Congress ASCE, 1994.
Richter, P. J., Nims, D. K., Kelly, J. M. and Kallenbach, R. M. (1990), The EDR -
Energy Dissipating Restraint, A New Device for Mitigation of Seismic Effects,
Proc. 1990 Structural Engineers Assoc. of Calif. (SEAOC) Convention, Lake
Tahoe.
Roik, K., Dorka, U. and Dechent, P. (1988), Vibration Control of Structures Under
Earthquake Loading by Three-Stage Friction-Grip Elements, Earthquake Engrg.
Struct. Dyn., 16, 501-521.
Scholl, R. E. (1993), Design Criteria for Yielding and Friction Energy Dissipators,
Proc. ATC 17-1 on Seismic Isolation, Energy Dissipation, and Active Control, 2,
485-495.
Shreir, L. L., Jarman, R. A., and Burstein, G. T., eds. (1994), Corrosion,
Butterworth-Heineman, Oxford, UK.
Tabor, D. (1981), Friction - The Present State of Our Understanding, J. Lubr.
Tech., ASME, 103, 169-179.
Tyler, R. G. (1985), Test on a Brake Lining Damper for Structures, Bull. N. Z. Soc.
Earthquake Engrg., 18(3), 280-284.
Vezina, S., Proulx, P., Pall, R. and Pall, A. (1992), Friction-Dampers for Aseismic
Design of Canadian Space Agency, Tenth World Conference on Earthquake
Engineering, Madrid, 4123-4128.
5
Viscoelastic Dampers
5.1 Introduction
The application of viscoelastic materials to vibration control can be dated back to the
1950s when it was first used on aircraft as a means of controlling vibration-induced
fatigue in airframes (Ross et al., 1959). Since that time, it has been widely used in
aircrafts and aerospace structures for vibration reduction (Morgenthaler, 1987; Gehling,
1987). Its application to civil engineering structures appears to have begun in 1969
when 10,000 viscoelastic dampers were installed in each of the twin towers of the
World Trade Center in New York to help resist wind loads (Mahrnoodi, 1969
Mahmoodi et al., 1987; Caldwell, 1986). This was followed by several similar
applications to tall buildings against wind, both in the U.S. and abroad. Some of these
applications will be highlighted in Section 5.7.
Seismic applications of viscoelastic dampers have a more recent origin. For seismic
applications, larger damping increases are usually required in comparison with those
required for mitigation of wind-induced vibrations. Furthermore, energy input into the
structure is usually spread over a wider frequency range, requiring more effective use
of the viscoelastic materials. Extensive analytical and experimental studies in the
seismic domain have led to the first seismic retrofit of an existing building using
viscoelastic dampers in the U.S. in 1993, which will also be briefly discussed in Section
5.7.
Under a sinusoidal load with frequency w, the shear strain (t) and the shear stress (t)
of a linear viscoelastic material oscillate at the same frequency w but in general out-of-
phase. They can be expressed by (Zhang et al., 1989)
where, as shown in Fig. 5.2, o, and o are, respectively, the peak shear strain and peak
shear stress, and 6 is the lag angle. For a given o, both o and are functions of w.
Figure 5.2
Stress and Strain under Sinusoidal Load
With sin wt given by Eq. (5.1) and using the identity sin2wt+cos2wt =1, we have the
stress-strain relationship
which defines an ellipse as shown in Fig. 5.3, whose area gives the energy dissipated by
the viscoelastic material per unit volume and per cycle of oscillation. It is given by
Figure 5.3
Plot of Stress Versus Strain
It is seen from Eq. (5.5) that the first term of the shear stress is the in-phase portion
with G'(w) representing the elastic modulus, and the second term or the out-of-phase
portion represents the energy dissipation component. This is seen more clearly if Eq.
(5.2) is rewritten in the form
which is valid under harmonic motion since in that case (t) = owcoswt. The quantity
G"(w)/w is the damping coefficient of the damper material. The equivalent damping
ratio is
Accordingly, G'(w) is defined as the shear storage modulus of the viscoelastic material,
which is a measure of the energy stored and recovered per cycle; and G"(w) is defined
as the shear loss modulus, which gives a measure of the energy dissipated per cycle.
The loss factor, , defined by
Another useful expression for the stress-strain relationship can be derived using
complex notation. Equation (5.1) can be written as
(t) = oeiwt, (t) = oei(wt+) (5.11)
and
(t) = G*(w) (t)
(5.12)
where G*(w) is the complex modulus of the viscoelastic material with
and
It is seen that the two moduli, G'(w) and G"(w), or G'(w) and , determine the dynamic
behavior of the linear viscoelastic material in shear under time harmonic excitation.
These moduli are not only functions of the excitation frequency w, but also functions of
the ambient temperature and, sometimes, the shear strain. Thus, a constitutive modeling
of the viscoelastic material consisting of (t), (t) and their time derivatives must first
be established. Another factor which may be important in evaluating the performance
of VE dampers is the variation of internal temperature within the material during
operation since energy is dissipated in the form of heat, giving rise to temperature rise
in the viscoelastic material.
5.3 Shear Storage and Shear Loss Moduli
As discussed above, the shear storage modulus and shear loss modulus of a viscoelastic
material are generally functions of excitation frequency (w), ambient temperature (T),
shear strain (), and material temperature (). One of the expedient ways of estimating
their dependence on these parameters is to perform experiments on viscoelastic
specimens over representative ranges of these variables.
A number of such tests have been carried out over the last few years. In one series of
these tests (Chang et al., 1993a), three types of viscoelastic dampers with
configurations shown in Fig. 5.1 and distinguished by dimensions and types of the
viscoelastic material were used. They are designated as Type A, B and C, respectively.
Types A and B dampers are made of similar VE materials but different in damper
dimensions. Type C damper is made from a different VE material. Table 5.1 lists the
area, thickness and volume of each type of the dampers. At least three dampers from
each group were studied experimentally.
Type A dampers were first tested under six different ambient temperatures (21C,
24C, 32C, 36C and 40C). At each temperature, six tests were conducted at
frequencies of 0.1 Hz, 1.0 Hz, 2.0 Hz, 3.0 Hz, 3.5 Hz and 4.0 Hz, respectively, for up to
fifty cycles of deformation in three different strain ranges (5%, 20% and 50%). Next,
Type B and Type C dampers were tested at constant 5% strain at frequencies of 3.0 Hz,
3.5 Hz and 4.0 Hz at five different ambient temperatures (25C, 30C, 34C, 38C and
42C). Finally, Type B dampers were tested at three more strains (15%, 25% and 50%)
at 24 C to simulate the effect of damper strain on the energy dissipation capacity of the
VE dampers under medium to strong earthquake ground motions.
The force-deformation curves of the three types of dampers subjected to sinusoidal
excitations with frequency of 3.5 Hz and 5% damper strain at two ambient temperatures
are shown in Figs. 5.4a-5.4f. All of the hysteresis loops are fairly rounded in shape,
indicating that the dampers can effectively dissipate energy. It is seen from these
figures that the damper stiffness and the amount of energy dissipation in one cycle
decrease for all types of dampers with increasing ambient temperature. The loss factors,
however, remain more
or less constant for each type of the dampers regardless of the change in ambient
temperature. Comparisons of damper properties among the three types of dampers are
listed in Table 5.2.
Figure 5.4
Comparison of Force-Deformation
Relationships (5% Strain, 3.5 Hz)
From Table 5.2, it may be concluded that the Type C damper is less sensitive to the
change of ambient temperature. The percent reductions in energy dissipation capacity
due to the change in ambient temperature from 24C to 42C are 73%, 71% and 60%,
respectively, for Types A, B and C dampers. The lower temperature sensitivity of the
Type C damper can also be observed in the reduction rates of the damper stiffness,
which are 70%, 68% and 34% respectively, for Types A, B and C dampers.
Table 5.2 VE Damper Properties at 3.5 Hz and 5% Strain
From the descriptions above, it is clear that one has to take into account the effect of
ambient temperature and excitation frequency for an effective design of viscoelastic
dampers in structural applications. The damper properties are also, to a certain degree,
dependent on the number of loading cycles and the range of deformation, especially
under large strain due to temperature increase within the damper material. Test results
of typical Type B dampers under the excitation frequencies of 1 Hz and 3 Hz, ambient
temperatures of 24 and 36 and damper strains of 5% and 20% are listed in Table 5.3.
It can be seen that the damper properties remain somewhat constant and independent of
strain (below 20%) for each temperature and frequency.
or
if (t) = 0 for t < 0, which is true in most cases. In the above, G(t) is the stress
relaxation modulus, defined as the ratio of stress to strain at constant deformation, and
is usually determined from test data for a given material. For our purposes, one of the
pertinent expressions for G(t) (Williams, 1964) is
where Ge is the rubbery modulus, Gg is the glassy modulus, to is the relaxation time,
and a is a constant giving the slope of the relaxation curve through the transition region
between glassy and rubbery behavior. The stress relaxation modulus G(t) as given in
Eq. (5.18) predicts a bounded modulus for all non-negative time and has been found to
be reasonably accurate for most VE materials if a is not restricted to be an integer. Its
initial value is the glassy modulus, which then smoothly approaches the rubbery
modulus with increasing time.
The relationship between G(t) and G'(w) and G"(w) can be found by taking the Laplace
transform of Eq. (5.17). Since
where () represents the gamma function. Therefore, the Laplace transform of the
relaxation modulus is
Upon substituting Eq. (5.22) into Eq. (5.20), one obtains
If one lets s = iw, G*(iw) is the complex modulus whose real and imaginary parts give,
respectively, G'(w) and G"(w) as indicated in Eq. (5.13). They have the forms
and
Equations (5.18), (5.24) and (5.25) permit the determination of the model parameters
Ge, Gg, to and from a combination of tests in the time domain, such as stress
relaxation tests, and in the frequency domain, such as sinusoidal tests. These
parameters can be first estimated based on the data obtained from the stress relaxation
tests. Their accuracy can then be checked by the data obtained from the sinusoidal tests
(Shen and Soong, 1995).
As an example, Fig 5.5 shows a set of stress relaxation tests performed on a set of
viscoelastic dampers at about 26C in which the maximum strain was 20% and strain
rates were 0.1, 0.2, 0.5, and 1.0 in/sec, respectively. The
Figure 5.5
Stress Relaxation Tests at Different Strain Rates
time axis of all the curves are shifted so that they have the same origin. It is interesting
to note that the curves for different strain rates constitute parts of the same "real" stress
relaxation curve. With this approximation, one can begin with a set of constants, Ge,
Gg, and to, as their initial values in an iterative procedure if the measured curves are
included in the curve generated by these constants.
These dampers were also tested under sinusoidal excitations at prescribed frequencies,
temperatures and strains. Figure 5.6 shows the first twenty cycles of the hysteresis
loops at the frequency of 3.0 Hz and 10% strain at about the same temperature. These
data can be used to check the constants obtained in the stress relaxation tests. In this
case, the parameters are estimated to be, at 26C,
It can be seen in Fig. 5.7 that the predicted moduli are very close to the test results
available within the frequency range from 1.0 to 3.0 Hz.
Figure 5.6
Typical Hysteresis Loop Results from a Sinusoidal Test
Figure 5.7
Comparison of Storage and Loss Moduli Between Simulation and Test
It is noted that Eqs. (5.24) and (5.25), while offering a good estimate of the properties
of the VE material, can be further simplified for engineering purposes. It is seen that
when the relaxation time is very small as in this case, cos(/2+wto) and sin(/2+wto)
remain almost unchanged within a certain frequency range. Therefore, the variations of
G'(w) and G''(w) with respect to frequency fall into straight lines if log-log scale is
used. This observation provides a significant simplification for estimating the
frequency dependence of the VE material properties. Basically, at a given temperature,
only two damper tests at different frequencies are required in order to identify the
straight line relationship.
5.3.2 Dependence on Ambient Temperature
In order to determine the dependence of G'(w) and G"(w) on the ambient temperature,
the method of reduced variables can be used (Ferry, 1980). This method affords a
convenient simplification in separating the two principal variables, frequency and
temperature, on which the VE material properties depend, and expressing these
properties in terms of a single function of each, whose form can be experimentally
determined. By using this method, the temperature dependence of the VE properties
can be obtained by plotting
where T, in K, is the ambient temperature of interest, To is an arbitrarily selected
reference temperature, in K, at which the measurements are made, p is the density of
the VE material, and T is a shift factor of time or frequency, which is determined
experimentally. The factor poTo/pT is very close to unity if T is not significantly
different from To. For the same VE material tested above, Fig. 5.8 shows log T versus
temperature from 21C to 40C. The regression expression of this relation is a straight
line with
log T = -0.0561T + 1.218 (5.28)
for 21C < T <40C.
The dependence of G'(w) and G"(w) on the ambient temperature can be estimated using
Eqs. (5.27). On the log-log scale, Figs. 5.9 and 5.10 show that the temperature effect is
a simple shift of the G'(w) and G"(w) curves over the frequency range from 0.5 Hz to
8.0 Hz for this case.
Figure 5.8
Regression of Scale Factor, T
Figure 5.9
Simplified Expression for Shear Storage Modulus
Figure 5.10
Simplified Expression for Shear Loss Modulus
5.3.3 Dependence on Internal Temperature
The effect of temperature rise within the VE material when it is subjected to shear
deformation can also be taken into account using the method of reduced variables. In
this case, the external work needs to be converted into heat in terms of temperature and
the relaxation modulus in Eq. (5.18) needs to be updated incrementally using Eqs.
(5.27) and (5.28).
The internal temperature, , within the VE material due to the mechanical work done
by the damper can be calculated from the heat transfer equation (Kasai et al., 1993)
where cv is the specific heat of the VE material, p is the mass density, and k is thermal
conductivity. The spatial variation of the temperature is assumed to occur in the z-
direction across the VE layer. Experimental results and finite element analyses indicate,
however, this transient heat conduction term is small (Lai, 1991) and Eq. (5.29) can be
approximated by
where
Thus, unlike metallic or friction dampers, a linear structure with added VE dampers
remains linear with the dampers contributing to increased viscous damping as well as
lateral stiffness. This feature represents a significant simplification in the analysis of
viscoelastically damped structures (Zhang et al., 1989; Zhang and Soong, 1992).
For a single-degree-of-freedom structure such as the one shown in Fig. 2.10 and
oscillating at frequency w, the term x in Eq. (2.18) due to the addition of a viscoelastic
damper with shear area A and thickness h takes the form
Figure 5.11
The Five-story Test Structure
The extension of this analysis procedure to multi-degree-of-freedom structural systems
is straightforward if proportional damping is assumed. The equations of motion for the
modal displacements are decoupled in this case, and modifications to the modal
damping and stiffness due to addition of VE dampers can be obtained by following the
modal strain energy method (Ungar and Kerwin, 1962; Johnson and Kienholz, 1982;
Soong and Lai, 1991). The ith modal damping ratio due to added VE dampers can be
calculated as
where (wi) is the loss factor of the VE material at the modal frequency wi of the
original structure, Ei is the ith modal strain energy of the system with dampers and Eu is
the energy stored in the viscoelastic dampers. These energies are calculated from
where i is the ith mode shape vector associated with wi, K is the original stiffness
matrix without added dampers, and is the stiffness matrix attributed to the added
dampers. Equation (5.34) can be written as
Figure 5.12
2/5-Scale Five-story Steel Frame; a) Measured
Natural Frequency, b) Measured Damping Ratio
Figure 5.13
Roof Displacement Time Histories; (a) No
Damper Added, (b) with Dampers at 25C,
(c) with Dampers at 42C
Table 5.4 Summary of Dynamic Response Reduction Percentage under 0.12g White Noise
Excitation
No
Maximum Floor Damper With Dampers (% Reduction of No-Damper Case)
Response Level Reference T=25C T=30C T=34C T=38C T=42C
Relative 5 0.696 81.9 81.3 76.4 70.7 66.7
Floor 4 0.588 83.3 80.1 75.0 69.9 66.7
Disp. 3 0.484 84.3 80.8 75.6 71.1 68.6
(in) 2 0.328 83.5 79.6 75.6 71.6 69.5
1 0.116 76.7 73.3 70.7 65.5 63.8
Interstory 5 0.152 78.3 82.2 78.3 75.0 73.0
Drift 4 0.164 86.0 84.1 78.7 73.2 70.1
(in) 3 0.178 82.0 82.0 74.2 68.5 65.2
2 0.214 85.0 82.7 77.1 72.4 68.2
1 0.116 76.7 73.3 70.7 65.5 63.8
Maximum 5 1.222 85.7 85.1 81.1 77.0 73.9
Floor 4 0.856 82.5 80.5 76.8 72.5 69.4
Accel. 3 0.988 86.5 85.2 81.5 78.7 77.8
(g) 2 0.948 84.6 84.7 84.7 80.1 78.4
1 0.658 74.6 75.2 77.4 75.2 76.4
Shaking table tests were also carried out on the viscoelastically damped structure at
room temperature (25C) under the time scaled El Centro earthquake with a peak
acceleration of 0.6g. Numerical studies using an inelastic analysis program DRAIN-2D
(Kanaan and Powell, 1973) showed that, without added VE dampers, the model
structure would undergo inelastic deformation under this strong ground motion.
Therefore, under such severe ground motions, only analytical studies were conducted
on the model structure without added dampers. The inelastic analysis results are used to
assess the effectiveness of VE dampers under strong earthquake ground motions.
Figures 5.14a-5.14b show displacement time histories at the roof of the model structure
without and with added VE dampers under the 0.6g El Centro earthquake ground
motion. The natural frequencies of the structure without and with added VE dampers
are about 3.1 Hz and 3.7 Hz, respectively. It can be seen that VE dampers provide
significant extra damping to the structure so that the structure behaved elastically and
the seismic response was greatly reduced. Similar observations can be made for story
drifts and floor accelerations at all floor levels. The effectiveness of VE dampers under
the scaled 0.6g El Centro and Hachinohe earthquake ground motions are summarized in
Table 5.5.
Figure 5.14
Displacement Time History at Roof; (a) No Damper Added (Simulated), (b) with Dampers (Measured)
With Dampers
No Damper (% Reduction of
(Inelastic Analysis) No-Damper Case)
Maximum Floor El Centro Hachinohe El Centro Hachinohe
Response Level (0.60g) (0.60g) (0.60g) (0.60g)
Relative 5 2.150 3.490 0.766 0.823
Floor Disp. (64.4) (76.4)
(in) 4 1.990 3.240 0.665 0.719
(66.6) (77.8)
3 1.650 2.700 0.529 0.579
(67.9) (78.6)
2 1.110 1.630 0.346 0.382
(68.8) (76.6)
1 0.390 0.470 0.143 0.148
(63.3) (68.5)
Interstory 5-4 0.207 0.310 0.104 0.111
Drift (49.8) (64.2)
(in) 4-3 0.365 0.599 0.137 0.146
(62.5) (75.6)
3-2 0.598 1.100 0.187 0.201
(68.7) (81.7)
2-1 0.721 1.185 0.214 0.234
(70.3) (80.3)
1-0 0.394 0.470 0.143 0.148
(63.7) (68.5)
Figures 5.15a-5.15d show the envelope curves of the lateral displacement, interstory
drift, cumulated story shear, and overturning moment of the model structure with and
without added dampers under the 0.6g El Centro earthquake. It can be seen that adding
VE dampers to the structure reduces not only the deformation but also the base shear
and overturning moment even when the structure without added dampers behaves
inelastically. The VE dampers dissipate a significant amount of seismic input energy to
prevent the structure from undergoing inelastic deformation. Similar results were
obtained for the 0.6g Hachinohe earthquake.
Figure 5.15
Response Envelopes; (- - - - -) with Dampers, () without Dampers
The equivalent structural damping of the model structure with added VE dampers can
be estimated by using the modal strain energy method as discussed in Section 5.4.
Figures 5.16a-5.16b show the comparisons of the measured damping ratios and those
predicted by Eqs. (5.36) and (5.38), respectively, at various ambient temperatures. As
can be seen, the modal strain energy method predicts the dynamic characteristics of the
viscoelastically damped structure very well.
Figure 5.16
Measured and Predicted 1st Modal
Damping Ratios; a) Eq. (5.36), b) Eq. (5.38)
Using the predicted damping ratios as the modal damping ratios in the linear dynamic
analysis, one can simulate response time histories of the model structure. Figures 5.17a-
5.17b show good agreement between the experimental and analytical displacement time
histories at the roof of the structure with added VE dampers at 34C under the 0.12g
Hachinohe earthquake. Even under the 0.6g El Centro earthquake, good correlation can
be obtained when the responses are compared as shown in Fig. 5.18 and Fig. 5.14b.
Thus, the seismic response of viscoelastically damped structures can be reliably
estimated using conventional linear dynamic analysis methods as developed in Section
5.4.
Figure 5.17
Displacement Time History at
Roof; (a) Experimental, (b) Analytical
Figure 5.18
Predicted Displacement at Roof
under 0.6g El Centro Earthquake
5.5.2 1/3-Scale Three-story Concrete Frame
In this study, the seismic response of a scaled reinforced concrete structure using VE
dampers as a means for seismic retrofit was investigated. Unlike steel structures, the
seismic response of a reinforced concrete structure is by and large inelastic, which is
often accompanied by permanent deformation and damage. The addition of viscoelastic
dampers in this case can dissipate energy at the early stages of cracking of the concrete
elements and reduce the development of damage. With proper selection of dampers,
this damage can be substantially reduced or even eliminated. The quantification of the
influence of viscous and elastic stiffness properties of dampers during the inelastic
response of reinforced concrete structures was the subject of this investigation.
A one-third scale model of a three-story lightly reinforced concrete framed building
was tested under simulated base motions using a shaking table (Foutch et al., 1993;
Lobo et al., 1993; Chang et al., 1994). The structure was tested using a series of
simulated ground motions obtained from the scaled 1952 Taft earthquake, N21E
component, normalized for peak ground accelerations of 0.05g, 0.20g, and 0.30g,
representing minor, moderate, and severe ground motions. The structure, which was
previously damaged in shaking table tests, was retrofitted by adding viscoelastic
diagonal braces in the interior bay of each frame as shown in Fig 5.19. The viscoelastic
dampers were similarly positioned as in the steel frame case.
Some selected experimental results under the 0.2g Taft earthquake are summarized here
for one type of dampers tested from the viewpoint of energy dissipation. The interstory
drifts and story shears in the columns are substantially reduced at all floors as indicated
in Table 5.6. While the deformations are reduced approximately three times, the shear
forces are reduced only twice. These forces are much smaller than the ultimate strength
of the columns; moreover, they are smaller than their yielding strengths. A sample set
of force-deformations at the first floor as shown in Fig. 5.20 indicates that the column
forces and deformations are substantially reduced, while most of the energy dissipation
(area of hysteretic loops) is transferred from the columns to the viscoelastic dampers.
Although some inelastic deformations are experienced by the columns in the presence
of the viscoelastic braces, the column response is substantially improved.
The VE dampers alter the overall energy balance as shown in Fig. 5.21. For the 0.2g
Taft earthquake used in the experiment (Figs. 5.21a-5.21b), the total input energy is
increased. However, the added VE dampers dissipate a majority of this energy, leaving
only a small amount of hysteretic energy to be dissipated by the structural members.
Similar energy calculations under some other earthquakes are also shown in Fig. 5.21,
showing that the overall energy input may vary depending on the match between the
structural frequencies and the earthquake frequency content.
Figure 5.19
Details of the R/C Model with Viscoelastic Braced Dampers
Table 5.6 Maximum Measured Story Response for 0.2g Taft Excitation
Interstory Drifts (in) Column Story Shears (kips)
First Second Third First Second Third
Without Dampers 0.656 0.388 0.167 20.63 16.20 10.71
With Dampers 0.194 0.147 0.066 7.68 5.71 4.19
Figure 5.20
Force-deformation Curves under 0.2g Taft Earthquake
This series of experiments reveals that retrofitting reinforced concrete frames using
viscoelastic dampers can reduce the overall response, but more importantly, can reduce
the risk of developing a damaging mechanism near collapse. In particular, the hysteretic
energy dissipation is transferred from the load bearing elements, such as the columns or
beams, to non-load bearing devices that are not essentially damaged.
Figure 5.21
Contributions of Energy Terms
5.6 Design Considerations and Implementational Issues
One of the fundamental requirements in structural design is to reliably predict the
designed structural response under specified loading conditions. Current state-of-the-
practice permits the engineers to correctly analyze the structures they design, provided
all the design parameters are properly given. In designing structures with added VE
dampers, the most important design parameter is the damping ratio. By properly
incorporating the modal strain energy method into the design flow chart, design of
structures with added VE dampers can be accomplished with minimum modifications
to the current design practice (Chang et al., 1993b, c).
Like many other design problems, the design of viscoelastically damped structures is in
general an iterative process. First, an analysis of the structure without added dampers
should be carried out. Then the required damping ratio becomes the primary design
parameter for adding VE dampers to the structure. The design will normally contain the
following steps which may continue to update the structural properties after each design
cycle: (a) determine structural properties of the building and perform structural
analysis; (b) determine the desired damping ratio; (c) select desirable and available
damper locations in the building; (d) select damper stiffness and loss factor; (e)
calculate the equivalent damping ratio using the modal strain energy method; and (f)
perform structural analysis using the designed damping ratio. When steps (e) and (f)
satisfy the desired damping ratio and the structural performance criteria, the design is
complete. Otherwise, a new design cycle will proceed which may lead to new structural
properties, damper locations or damper dimensions and properties. A general flow chart
of the design procedure is given in Fig. 5.22.
It can be seen that this design procedure falls into the traditional design procedure
except for the determination of the required damping ratio and the selection of damper
stiffness and loss factor. In general, the required damping ratio can be estimated by
using the response spectra of the design earthquake with various damping ratios. The
selection of damper stiffness and loss factor can be a trial and error procedure. They
can also be determined based on the principle that the added stiffness due to the VE
dampers be proportional to the story stiffness of the structure. This is obtained by
modifying the modal strain energy method for each story as:
where is the target damping ratio and i and ki are, respectively, the damper stiffness
and the structural story stiffness without added dampers at the ith
story. For a VE material with known G' and G'' at the design frequency and
temperature, the area of the damper, A, can be determined from, as seen from the first
of Eqs. (5.32),
The thickness of the VE material, h, can be determined from the maximum allowable
damper deformation to insure that the maximum strain in the VE material is lower than
the ultimate value.
In this design procedure, the structure is assumed to be linear elastic. If inelastic
deformation is allowed in the structure, the demand in VE damping can be reduced and
a modified procedure has to be used.
Figure 5.22
VE Damper Design Flow Chart
As a design example, consider the 2/5-scale steel frame model structure considered in
Section 5.5.1. The design earthquakes are 0.6g El Centro and Hachinohe earthquakes
and the design requirements are (a) story drift is less than 0.5%, (b) structure remains
elastic under the design earthquakes, and (c) operating temperature is 25C.
(a) Analysis of the Structure. Analytical results using DRAIN-2D show that plastic
hinges will form over the structure under the specified design earthquakes. The story
drifts under the scaled 0.6g Hachinohe and El Centro earthquakes are 2.5% and 1.53%,
respectively. In order to achieve the design requirement, VE dampers will be needed.
(b) Determination of the Required Damping Ratio. The required damping ratio in
general can be determined from the response spectra of the design earthquakes. In this
example, it is determined that an equivalent structural damping ratio of 15% will be the
initial goal.
(c) Select Desirable and Available Damper Locations. VE dampers can be placed in
any available locations which allow shear deformations to occur within the dampers. In
this example, dampers will be placed as diagonal braces. The angle between the
bracing members and the floor is 42.1 except for the first floor.
(d) Design the VE Dampers. Equation (5.39) can be used to calculate the damper
stiffness with an assumed value of the loss factor. Since the inclination angle is 42.1,
the damper stiffness is equal to /cos2 in this example.
The typical story stiffness without dampers, ki, is 14.73 kip/in. Assuming = 1.1, the
damper stiffness at a typical story, ki, is 5.02 kip/in based on Eq. (5.39).
The thickness of the VE material, h, can be determined from the maximum allowable
damper deformation to insure that the maximum strain in the VE material is smaller
than the maximum allowable value. In this example, the maximum damper deformation
is 0.005 x 47 x cos = 0.174 in. If the maximum damper strain of 60% is allowed, the
damper thickness h is determined as 0.3 in. The damper properties can be determined
based on one-third of that of the maximum damper strain, or 20%. In this example, G' =
250 psi. Finally, if two VE layers are used, the area of the VE dampers can be obtained
from Eq. (5.40) as A = h/2G' = 3.0 in2.
(e) Estimate the Structural Damping Ratio. Following the modal strain energy method
with the damper properties corresponding to 25C, 20% strain and 3.5 Hz, the damping
ratio of the viscoelastically damped structure is about 15%. If the calculated damping
ratio at this stage is lower than the required value, more dampers or larger dampers may
have to be used.
(f) Perform Dynamic Analysis using the Designed Damping Ratio. In this example, it
shows that the structure behaves elastically and the maximum
story drift is less than 0.5%. Since all the performance criteria are satisfied, the design
is complete.
5.6.1 Placement Strategies for VE Dampers
A topic which has not been touched upon in the development of the design procedure is
one of damper placements. It is clear that physical constraints exist which influence the
choice of damper locations. It is also clear that the way in which the dampers are
distributed in a structure may have significant effect on their effectiveness.
A damper placement strategy is discussed in Ashour and Hanson (1987) for structures
modeled as a uniform shear beam. The optimal locations of the dampers are found to
conform with the pattern of distribution which will result in maximizing the first
mode's damping ratio. In another approach (Zhang and Soong, 1992), a sequential
optimization procedure is proposed in which a performance index is maximized at each
step in order to determine the optimal location of a damper in a sequence. The basic
idea is to place each damper at a location experiencing the largest relative displacement
between attachment points of the damper.
The advantage of adding dampers at optimal locations can be shown by estimating the
number of dampers that can be saved to achieve the same effect as compared with the
case of non-optimal placement. Using a ten-story structure as an example, Fig. 5.23
shows one of such comparisons. These two sets of curves are, respectively, the
maximum displacements and maximum story drifts at the top floor plotted against the
numbers of VE dampers added at optimal and at uniformly distributed locations. It is
seen from Fig. 5.23b, for example, that the same story drift of approximately 0.62 in
can be achieved by using four optimally located dampers as compared with nine
uniformly spaced dampers.
5.6.2 Architectural Considerations
Aside from engineering considerations, architectural concerns may lead to necessary
modifications to the method of installation of VE dampers to structural frames. This is
particularly relevant for damper installation in interior bays. While diagonal
installation, such as that shown in Fig. 5.11, is a logical choice in order to maximize
relative displacements between attachment points of a VE damper, other configurations
are possible in order to conform with architectural requirements. Figure 5.24 shows
several installation options in a case where VE dampers are installed in interior bays of
a slab-column nonductile concrete frame structure. Each arrangement can generate
required shear deformation in the VE material while satisfying certain specified
architectural constraints.
Figure 5.23
Examples of Maximum Response as
Function of Number of Dampers; a)
Maximum Displacement, b) Maximum
Interstory Drift
Figure 5.24
Installation Options for VE Dampers
5.7 Structural Implementation
As described in Section 5.1, the first applications of VE dampers to structures were for
reducing acceleration levels, or increasing human comfort, due to wind. In 1969, VE
dampers were installed in the twin towers of the World Trade Center in New York, NY,
as shown in Fig. 5.25a, as an integral part of the structural system. They were designed
to assist the tubular steel frame in limiting wind-induced building vibrations to levels
below human perception. (Mahmoodi, 1969; Mahmoodi et al., 1987). The selection,
quantity, shape and location of the dampers were chosen based on the dynamics of the
towers and required damping to achieve the performance objectives.
There are about 10,000 VE dampers in each tower, which are evenly distributed
throughout the structure from the 10th to the 110th floor. As shown in Fig. 5.25b, they
are located between the lower chords of the horizontal trusses and the columns of the
outside wall. The towers have experienced a number of moderate to severe wind storms
over the last twenty-five years. The observed performance of the VE dampers has been
found to agree well with theoretical values. After hurricane Gloria in 1978, the total
damping of the building was calculated and found to be in the range of 2.5% to 3% of
critical. The aging characteristics of the VE dampers have also been found to be
excellent.
Figure 5.28
The Santa Clara County Building, San Jose, CA
The final design called for installation of two dampers per building face per floor level
as shown in Fig. 5.29, which would increase the equivalent damping in the fundamental
mode of the building to about 17% of critical, providing substantial reductions to
building response under all levels of ground shaking. A typical damper configuration is
shown in Fig. 5.30.
Figure 5.29
Location of VE Dampers (Courtesy of
the Crosby Group, Redwood City, CA)
Figure 5.30
Damper Configuration (Courtesy of
the Crosby Group, Redwood City, CA)
The structure shown in Fig. 5.31 represents the first application of viscoelastic damper
technology to a reinforced concrete structure for seismic upgrade. Located in San
Diego, the building is a three-story structure with its plan and elevation shown in Fig.
5.32. Its lateral load resisting system consists of 8-inch reinforced concrete perimeter
walls. There is a 4-inch separation joint at the center of the building, in the north/south
direction. The second and third floor systems are 10-1/4 and 9-inch flat slabs,
respectively, both with column capitals. The foundation consists of reinforced concrete
wall footings at the exterior long wall, with all columns resting on the 6-inch slab on
grade, which has 6 6 inch mesh of No. 6 wire for reinforcement. Results of extensive
seismic evaluations show that the structure is not of sufficient quality to withstand the
expected seismic loads. Appropriate reduction of interstory drifts was the primary
objective of seismic upgrade and, based on nonlinear dynamic analysis, a total of 64
dampers were required to meet seismic demands on the structural elements in this case.
As shown in Fig. 5.33 each damper consists of four damper units incorporated into the
structure in a K-brace configuration.
Figure 5.31
Building 116, Naval Supply Facility, San Diego
Figure 5.33
Damper Configuration
References
Aiken, I.D. and Kelly, J. M. (1990), Earthquake Simulator Testing and Analytical Studies
of Two Energy-Absorbing Systems for Multistory Structures, Report No. UCB/EERC-
90/03, University of California at Berkeley, CA.
Aiken, I.D., Nims, D. K., Whittaker, A. S. and Kelly, J. M. (1993), Testing of Passive
Energy Dissipation Systems, Earthquake Spectra, 9(3), 335-370.
Aiken, I.D. and Whittaker, A. S. (1993), Development and Application of Passive Energy
Dissipation Techniques in the U.S.A., Proc. Int. Post-Smirt Conf. Seminar on Isolation,
Energy Dissipation and Control, Capri, Italy.
Ashour, S. A. and Hanson, R. D. (1987), Elastic Seismic Response of Buildings with
Supplemental Damping, Report No. UMCE 87-01, The University of Michigan, Ann
Arbor, MI.
Bagley, R. L. and Torvik, P. J. (1983), Fractional Calculus - A Different Approach of the
Analysis of Viscoelastic Damped Structures, AIAA Journal, 21(5), 741-748.
Bergman, D. M. and Hanson, R. D. (1993), Viscoelastic Mechanical Damping Devices
Tested at Real Earthquake Displacements, Earthquake Spectra, 9(3), 389-418.
Caldwell, D. B. (1986), Viscoelastic Damping Devices Proving Effective in Tall
Buildings, AISC Engineering Journal, 23(4), 148-150.
Chang, K. C., Lai, M. L., Soong, T. T., Hao, D.S. and Yeh, Y. C. (1993a), Seismic
Behavior and Design Guidelines for Steel Frame Structures with Added Viscoelastic
Dampers, NCEER 93-0009, National Center for Earthquake Engineering Research,
Buffalo, NY.
Chang, K. C., Shen, K. L., Soong, T. T. and Lai, M. L. (1994), Seismic Retrofit of A
Concrete Frame with Added Viscoelastic Dampers, 5th National Conference on
Earthquake Engineering, Chicago, IL.
Chang, K. C., Soong, T. T., Lai, M. L. and Nielsen, E. J. (1993b), Development of a
Design Procedure for Structures with Added Viscoelastic Dampers, Proc. ATC-17-1 on
Seismic Isolation, Energy Dissipation and Active Control, 2, 473-484.
Chang, K. C., Soong, T. T., Lai, M. L., Nielsen, E. J. (1993c), Viscoelastic Dampers as
Energy Dissipation Devices for Seismic Applications, Earthquake Spectra, 9(3), 371-388.
Chang, K. C., Soong, T. T., Oh, S-T. and Lai, M. L. (1992), Effect of Ambient
Temperature on a Viscoelastically Damped Structure, ASCE Journal of Structural
Engineering, 118(7), 1955-1973.
Chang, K. C., Soong, T. T., Oh, S-T and Lai, M. L. (1995), Seismic Behavior of Steel
Frame with Added Viscoelastic Dampers, ASCE Journal of Structural Engineering,
121(10), 1418-1426.
Crosby, P., Kelly, J. M. and Singh, J. (1994), Utilizing Viscoelastic Dampers in the
Seismic Retrofit of a Thirteen Story Steel Frame Building, Structures Congress XII,
Atlanta, GA, 1286-1291.
Ferry, J. D. (1980), Viscoelastic Properties of Polymers, John Wiley, New York, NY.
Foutch, D. A., Wood, S. L. and Brady, P.A. (1993), Seismic Retrofit of Nonductile
Reinforced Concrete Frames using Viscoelastic Dampers, ATC-17-1 on Seismic Isolation,
Passive Energy and Active Control, 2, 605-616.
Fujita, S. Fujita, T. Furuya, O., Morikawa, S., Suizu, Y., Teramoto, T. and Kitamura, T.
(1992), Development of High Damping Rubber Damper for Vibration Attenuation of
High-rise Buildings, Proc. 10th World Conf. Earthquake Engrg., 2097-2101, Balkema,
Rotterdam.
Gehling, R.N. (1987), Large Space Structure Damping Treatment Performance: Analytic and
Test Results, Role of Damping in Vibration and Noise Control, ASME, NY, 93-100.
Gemant, A. (1938), On Fractional Differentials, Philosophical Magazine, 25, 540-549.
Johnson, C.D. and Kienholz, D. A. (1982), Finite Element Prediction of Damping in Structures
with Constrained Viscoelastic Layers, AIAA Journal, 20(9), 1284-1290.
Kanaan, A. E. and Powell, G. H. (1973), Drain-2D - A General Purpose Computer Program for
Dynamic Analysis of Inelastic Plane Structures, Report No. UCB/EERC 73-06, University of
California, Berkeley, CA.
Kasai, K., Munshi, J. A., Lai, M. L. and Maison, B. F. (1993), Viscoelastic Damper Hysteretic
Model: Theory, Experiment and Application, Proc. ATC-17-1 on Seismic Isolation, Energy
Dissipation, and Active Control, 2, 521-532.
Keel, C. J. and Mahmoodi, P. (1986), Designing of Viscoelastic Dampers for Columbia Center
Building, Building Motion in Wind, (eds., N. Isyumov and T. Tschanz), ASCE, NY, 66-82.
Kirekawa, A., Ito, Y. and Asano, K. (1992), A Study of Structural Control using Viscoelastic
Material, Proc. 10th World Conf. Earthquake Engrg., 2047-2054, Balkema, Rotterdam.
Koh, C. G. and Kelly, J. M. (1990), Application of Fractional Derivatives to Seismic Analysis of
Base-isolated Models, Earthquake Engineering and Structural Dynamics, 19(2), 229-241.
Lai, M. L. (1991), Private Communication.
Lai, M. L., Chang, K. C., Soong, T. T., Hao, D.S. and Yeh, Y. C. (1995), Full-scale
Viscoelastically Damped Steel Frame, ASCE Journal of Structural Engineering, 121(10), 1443-
1447.
Lin, R. C., Liang, Z., Soong, T. T. and Zhang, R. H. (1991), An Experimental Study on Seismic
Structural Response with Added Viscoelastic Dampers, Engineering Structures, 13, 75-84.
Lobo, R. F., Bracci, J. M., Shen, K. L., Reinhorn, A.M. and Soong, T. T. (1993), Inelastic
Response of R/C Structures with Viscoelastic Braces, Earthquake Spectra, 9(3), 419-446.
Mahmoodi, P. (1969), Structural Dampers, ASCE J. of the Structural Division, 95(8), 1661-1672.
Mahmoodi, P. and Keel, C. J. (1986), Performance of Structural Dampers for the Columbia
Center Building, Building Motion in Wind, (eds., N. Isyumov and T. Tschanz) ASCE, NY, 83-
106.
Mahmoodi, P., Robertson, L. E., Yontar, M., Moy, C. and Feld, I. (1987), Performance of
Viscoelastic Dampers in World Trade Center Towers, Dynamic of Structures, Structures
Congress '87, Orlando, FL.
Figure 6.1
Viscous Liquid Dampers; a) Cylindrical Pot GERB Damper (Makris
and Constantinou, 1991), b) Viscous Damping Wall (Miyazaki and
Mitsusaka, 1992)
Both of the devices discussed above accomplish their objectives through the
deformation of a viscous fluid residing in an open container. In order to maximize the
energy dissipation density of these devices, one must employ materials with large
viscosities. Typically, this leads to the selection of materials that exhibit both frequency
and temperature dependent behavior. Thus, many of the issues that were addressed in
Chapter 5 will again be relevant here.
There is, however, another class of fluid dampers that rely instead upon the flow of
fluids within a closed container. In these designs, the piston acts now,
not simply to deform the fluid locally, but rather, to force the fluid to pass through
small orifices. As a result, extremely high levels of energy dissipation density are
possible. However, a correspondingly high level of sophistication is required for proper
internal design of the damper unit.
A typical Taylor Devices fluid damper for seismic application is illustrated in Fig. 6.2a
(Constantinou et al., 1993; Constantinou and Symans, 1993b). This cylindrical device
contains a compressible silicone oil which is forced to flow via the action of a stainless
steel piston rod with a bronze head. The head includes a state-of-the-art fluidic control
orifice design with a passive bimetallic thermostat to compensate for temperature
changes. In addition, an accumulator is provided to compensate for the change in
volume due to rod positioning. High strength seals are required to maintain closure.
Figure 6.2
Viscous Liquid Dampers; a) Taylor Devices Fluid Damper (Constantinou
et al., 1993), b) Jarret Elastomeric Spring Damper (Pekcan et al., 1995)
These uniaxial devices, which were originally developed for military and harsh
industrial environments, have recently found application in seismic base isolation
systems as well as for supplemental damping during seismic and wind-induced
vibration.
Another fluid damper featuring orifice flow to achieve energy dissipation is shown in
Fig. 6.2b. This device, manufactured by Jarret, utilizes a pressurized compressible
silicone-based elastomer to provide additional structural stiffness and damping. Recent
scale model experiments suggest that the device is suitable for application in aseismic
design (Pekcan et al., 1995).
Although the four dampers illustrated in Figs. 6.1 and 6.2 all depend upon the flow of
fluids to achieve energy dissipation, each device is distinct in terms of geometric design
and material performance. In order to obtain a better understanding of their anticipated
behavior within a structural system, one can examine the internal flow fields from a
continuum fluids dynamics viewpoint. This approach is adopted in Section 6.2, where
some of the relevant basic principles of fluid mechanics are provided. Since the
working fluids in most of these dampers do not typically exhibit classical Newtonian
response, a presentation of appropriate constitutive models is also included.
Unfortunately, this section is a little more mathematical than previous chapters. Those
readers interested primarily in macroscopic performance may wish to proceed directly
to Section 6.3, where overall damper behavior is considered. The objective of this latter
section is to provide meaningful macroscopic models for the dampers. Mechanics-
based approaches are discussed, along with those based directly upon measurements
from component level physical testing. The resulting macroscopic models are utilized
for overall structural analysis in Section 6.4, while structure level experiments are
detailed in Section 6.5. Of course, in order to apply viscous fluid dampers in actual
structures, design procedures must be developed. Progress in that area is reviewed in
Section 6.6. Finally, in Section 6.7, several recent full-scale structural implementations
of fluid dampers are examined.
where Eqs. (6.1) and (6.2) represent the conservation of mass and momentum,
respectively. In writing these equations, standard Cartesian tensor notation has been
used. Summation is implied by repeated indices, a superposed dot represents a partial
time derivative, and indices to the right of a comma indicate partial derivatives with
respect to spatial coordinates. A concise summary of this notational system is provided
in Shames and Dym (1985).
The first term on the left side of these equations represents the time rate of change of
the mass and momentum densities at the fixed point xi, while the second term measures
the convection of those same quantities past that point. The appearance of two separate
terms is a consequence of adopting the Eulerian framework.
In three-dimensional space, the general flow field is described in terms of 13 unknowns
(p, vi, ij). Equations (6.1) and (6.2) represent a set of four partial diffential equations.
Conservation of angular momentum furnishes three more equations that reduce to the
requirement for stress tensor symmetry (i.e., ij = ij). Thus, six additional relationships
must be provided to close this system. Various constitutive models that accomplish that
task by relating stresses to velocities will be discussed in the following subsections.
Before examining those models, it should be noted that for dampers involving
significant changes in temperature during operation, the flow becomes non-isothermal
and consideration should also be given to the conservation of energy. In this chapter,
we will limit the theoretical formulations to the isothermal case, which is generally a
reasonable approximation for fluid dampers. However, the effects of changes in
ambient temperatures will be considered.
6.2.2 Inviscid Fluids
This subsection will address the simplest of models, which involves the flow of an
inviscid fluid. In this case, the stress tensor can be completely defined in terms of a
scalar pressure p. Thus,
where ij represents the Kronecker delta. Substituting Eq. (6.3) into Eq. (6.2) produces
This is Euler's equation, which was derived by Euler in the mid-18th century. Equation
(6.4), together with the equation of continuity Eq. (6.1), govern the flow field.
When a fluid can be idealized as incompressible, the density is constant in both space
and time. Consequently, Eq. (6.1) reduces to a requirement for a divergence-free
velocity field. That is,
vjj = 0 (6.5)
Equations (6.4) and (6.5) govern the flow of an inviscid, incompressible fluid. If in
addition the flow field is irrotational, then a velocity potential can be defined such
that
vi = ,i (6.6)
,jj = 0 (6.7)
and the determination of the flow field is greatly simplified.
6.2.3 Newtonian Fluids
Real fluids, of course, have at least some viscosity, and the above idealizations are only
valid under a very limited set of circumstances. More generally, the stress tensor can be
written in terms of the pressure plus a fluid stress tensor ij that is a function of the
deformation field. Thus,
ij = -ijp + ij (6.8)
The most primitive constitutive model incorporating viscous effects is that for a
Newtonian fluid. In this case,
assuming the usual Stokes' hypothesis. For an incompressible fluid from Eq. (6.5), this
constitutive relationship simplifies to the following form:
with velocity amplitude and circular frequency w. Then in this case, the nonlinear
convective term in Eq. (6.11) is negligible, and the following equation written in terms
of velocity and pressure amplitudes obtains:
For sufficiently small frequencies, the remaining inertial term is also negligible. As a
result, the response becomes quasistatic with equilibrium satisfied at each instant of
time. Furthermore, one finds that this quasistatic response is frequency-independent as
a consequence of adopting the classical Newtonian model. Returning to the definitions
introduced for viscoelastic materials in Section 5.2, the complex shear modulus G* for
a Newtonian fluid is simply
G*(w) = iw (6.14)
Thus, the storage and loss shear moduli become G' = 0 and G'' = w, respectively.
6.2.4 Temperature-dependent Non-Newtonian Fluids
While this behavior is quite appropriate for many fluids, including water at moderate
frequencies, the liquids that are generally employed in fluid dampers exhibit frequency
dependent viscosities. Over the years, numerous constitutive models have been
proposed for such viscoelastic fluids. A particularly effective class is based upon the
generalization of classical models to incorporate fractional derivative operators. This
was first suggested by Gemant (1936). Subsequently, Bagley and Torvik (1983)
employed fractional calculus models to analyze the response of viscoelastic structures.
More recently, Makris and Constantinou (1991) applied fractional derivative Maxwell
models to represent the behavior of viscous fluid dampers, and then extended those
models by incorporating complex order derivatives (Makris and Constantinou, 1993).
Consider, for example, the complex-derivative Maxwell constitutive model that has
been used to characterize a particular polybutane fluid over a broad frequency and
temperature range (Makris et al., 1995). At some reference
temperature T0, under the assumption of infinitesimal incompressible deformation, the
proposed model can be written:
one obtains
with
In Eq. (6.17), and extract the real and imaginary part of their argument,
respectively.
The material parameters are then obtained from experimental data by employing a
nonlinear regression algorithm in the complex space, with an additional causality
constraint to ensure non-negative phase lag. At the zero frequency limit for the
complex-derivative Maxwell model, the causality considerations force 2/1 < 0. For
the polybutane fluid, this constraint is active. Consequently, the parameter becomes
real-valued and equal to the zero shear rate viscosity of the fluid. The optimal
parameter values at T0 = 293K are then = 9.30kPa-sec, = 0.0536 + i0.0147sec, and
v = 0.540 - i0.072.
Having established the material parameters in the complex-derivative Maxwell model,
the frequency dependence is completely characterized at the reference temperature To.
It remains to determine the temperature dependence of the response. For a large class of
polymers, the shape of the dynamic moduli versus frequency curves is similar when
evaluated at various temperatures. This led to the development of the method of
reduced variables (Ferry, 1980), which was discussed previously in Section 5.3.2. In
order to
combine data taken at different temperatures into a single master curve, a frequency
shift function T is employed, where
For the polybutane fluid in the temperature range of interest, one finds from
experimental data that the expression
is adequate with c = 0.043K-1 for T0 = 293K. Then a reduced dynamic modulus can be
defined at circular frequency w, such that
A plot of the reduced dynamic modulus versus reduced angular frequency, Tw, is
shown in Fig. 6.3 for the polybutane fluid. This experimental data, obtained at three
different temperatures, clearly delineates a pair of master curves for the storage and loss
moduli, and thus validates the use of the method of reduced variables for this material.
Also displayed in that figure are the dynamic shear moduli obtained from the complex-
derivative Maxwell model. The correlation is quite good over a broad frequency range.
The only noticeable deviation occurs in the storage modulus at low frequency.
However, in that range, the loss modulus is an order of magnitude greater, thus
reducing the significance of the deviation.
Finally, by combining Eqs. (6.18), (6.19) and (6.21), one can obtain an expression for
the dynamic moduli at any frequency and temperature within the bounds of the
experiments. For the complex-derivative Maxwell model, this can be written
with (T) and (T) representing the zero shear rate viscosity and relaxation time at
temperature T, where
(T) = T(T0) (6.23)
In the time domain, the constitutive model can now be expressed in the general form:
The above model has been developed for a particular polybutane fluid that is used
within a cylindrical pot damper similar to that shown in Fig. 6.1a. The same
methodology can no doubt be employed to develop constitutive models for several
other materials that are presently in use in fluid dampers. In some cases, however, the
simple Newtonian model may be sufficient throughout the frequency of interest, while
in other applications a nonlinear constitutive model may be required.
Figure 6.3
Dynamic Modulus for Polybutane Fluid (Makris et al., 1995)
6.3 Damper Behavior and Macroscopic Modeling
In order to properly design structures incorporating any of the fluid dampers illustrated
in Figs. 6.1 and 6.2, macroscopic models of their behavior must be available. As
mentioned above and in previous chapters, two alternative approaches are available.
One approach relies upon physical testing of individual damper units to construct the
macroscopic model. Most of the work on viscous fluid dampers has utilized this
approach. The alternative is to develop mechanics-based models starting from suitable
constitutive relations for the fluid and a representation of the geometric characteristics
of the device. A mixture of both approaches will be provided in the following
presentation, which will address each of the four fluid damper types described in the
introduction.
6.3.1 Cylindrical Pot Fluid Dampers
Consider first the cylindrical pot viscous fluid damper shown in Fig. 6.4, which
contains the polybutane fluid modeled in Section 6.2. The damper consists of inner and
outer stationary cylinders of radii r3 and r2, respectively, with a moving hollow
cylindrical piston of intermediate radius r1. The viscoelastic liquid occupies the space
between the inner stationary cylinder and the piston as well as the space between the
moving piston and the outer cylindrical boundary of the damper. With reference to Fig.
6.4, the geometric dimensions of GERB damper Type RHY-40/V50/H50 are:
r1 = 0.084m, r2 = 0.150m, r3 = 0.018m, H1 = 0.230m, H2 = 0.060m, and H = 0.370m.
Figure 6.4
Cylindrical Pot GERB
Damper (Makris et al., 1995)
For harmonic input, the dynamic stiffness of the damper is the ratio of the amplitude of
the resulting force P0 to the imposed displacement x0. Accordingly, the dynamic
stiffness of the device is K1+iK2 = P0/x0, where K1 and K2 are the storage and loss
stiffness, respectively. A complete description of K1 and K2 as a function of the
frequency can be used to characterize damper behavior under general time-dependent
excitation.
A series of dynamic tests on the viscoelastic fluid damper have been conducted to
measure this response. In each test, the damper piston was subjected to either vertical
(axial) or horizontal (lateral) motion of specific amplitude and frequency. The tests
were completed one after the other at eight different frequencies with an ambient
temperature T0 = 20C (293K). The storage and loss stiffnesses of the damper were
then determined from records of the force needed to maintain the imposed motion by
assuming linear steady-state behavior. Typical force-displacement loops obtained from
this device are shown in Fig. 6.5. Meanwhile, the corresponding dynamic stiffnesses
are plotted in Figs. 6.6a and 6.6b for the axial and lateral response, respectively.
Figure 6.5
GERB Damper Force-Displacement Hysteresis Loops
This is adequate for design purposes, however, each geometric configuration must be
tested. Alternatively, one can develop a mechanics-based approach to determine the
dynamic stiffness. The linearity of the governing equations, presented in Eqs. (6.25)
and (6.26), makes the boundary element method (e.g., Banerjee, 1994) an attractive
approach for the numerical solution of this problem. The development of the method
for this application is beyond the scope of the present chapter, however details can be
found in Makris et al. (1993b, 1995). Essentially one uses the governing equations
defined by Eqs. (6.25)-(6.27), along with an appropriate reciprocal theorem, to derive a
Figure 6.6
GERB Damper Dynamic Stiffness (Makris et al., 1995); a) Axial, b) Lateral
where the integration is performed over the bounding surface S of the damper fluid. In
the above equation, v is the velocity, while t represents the surface traction. The point y
is the location at which the velocity is determined, whereas x identifies the integration
point which, according to Eq. (6.28), varies all along the surface of the body. The
kernel functions G and F are derived from the infinite space fundamental solutions of
the governing equations. It should be noted Eq. (6.28) involves surface velocities and
tractions only. Volume integration has been completely eliminated. Consequently, a
boundary element analysis requires a mesh only on the surface S.
Numerical results obtained from the developed boundary element method for the
viscous fluid damper are also displayed in Fig. 6.6. The correlation between the
experimental data and numerical predictions is quite good throughout most of the
frequency range 0-30Hz. At the higher end of that range, the loss stiffness is
overpredicted by approximately 20%. While errors of
this magnitude may not be significant for design calculations, it is interesting to identify
their origin. It appears that the as-built dampers do not dissipate as much energy as
theoretically possible under idealized linear behavior. This may be due to imperfect
contact between the fluid and damper, to nonisothermal straining, or to localized finite
deformation effects for which the present method does not account.
As noted above, experiments were conducted for both axial and lateral motion of the
piston. However, the damper construction permits movement in all six degrees of
freedom, including torsional and rocking motion. The boundary element method was
also applied to these additional modes to complete the characterization of the damper.
Numerical results are presented in Fig. 6.7.
Figure 6.7
GERB Damper Torsional and Rocking
Dynamic Stiffness (Makris et al., 1995)
Both physical experiments and mechanics-based models provide values of the dynamic
stiffnesses at discrete frequencies. A macroscopic model is still needed that is more
suitable for inclusion in an overall structural analysis. This problem is addressed in
Makris and Constantinou (1991) and Makris et al. (1993a), where the following
fractional derivative Maxwell force-displacement model is proposed:
with P as the force applied to the piston and x as the resulting piston displacement.
Furthermore, the damper parameters C0, , and v represent the zero-frequency damping
coefficient, the relaxation time, and the order of fractional derivative, respectively. In
Makris and Constantinou (1991) the damper parameters are determined directly from
experimental data, while in Makris et al. (1993a) the parameters are estimated from
material data and a simplified analytical model. A typical result is shown in Fig. 6.8. It
should
Figure 6.8
Macroscopic Model for GERB Damper
Response (Makris and Constantinou, 1991)
be noted that for sufficiently low frequencies, inertia effects are unimportant, and one
finds that the macroscopic parameters and v coincide with those obtained from
constitutive modeling. Thus, only Co is a function of the damper geometry.
6.3.2 Viscous Damping Walls
A similar methodology could also be developed to model viscous damping walls
(Arima et al., 1988), shown in Fig. 6.1b. Assuming that the fluid is incompressible and
frequency-dependent, the response of the VDW to harmonic excitation is governed by
Eqs. (6.25) and (6.26), with the form of * established on the basis of viscometric flow
experiments. However, in this case, the fluid is confined to flow within a small gap
between two large parallel plates. Consequently, the flow is predominantly one-
dimensional as indicated in Fig. 6.9, and governing equations reduce to the single
equation:
where is now the fluid velocity in the x-direction and the excitation V exp(iwt)
represents the relative velocity between the inner and outer plates.
Figure 6.9
Viscous Damping Wall Flow Schematic
As indicated in Fig. 6.1b, spacers are used to maintain a uniform gap h throughout the
motion. Furthermore, assuming that the pressure gradients are negligible for this shear
dominated flow, and that the usual no-slip boundary conditions apply, an analytical
solution can be easily obtained. The velocity profile through the gap becomes
Consequently, for a VDW panel of length L and height H, the frequency-domain force-
displacement response can be written:
where x0 = -iV/w. Then the storage and loss stiffness are simply
respectively. Alternatively, the response can be defined in terms of K1 and the damping
coefficient C = K2/w.
The above formulation is based upon an assumption of isothermal flow involving
infinitesimal deformation. Under seismic loads, since the gap h is typically on the order
of 1cm, relatively large shear strains may occur. In such cases, it may be appropriate to
extend the formulation to include the effects of finite deformation by using convective
stress and strain measures within the constitutive model (e.g., Bird et al., 1987).
While the presentation above has concentrated on a mechanics-based formulation, most
of the VDW development work that has appeared in the literature has involved physical
testing. Typical hysteresis loops plotting force versus relative displacement from Arima
et al. (1988) are shown in Fig. 6.10 for harmonic excitation at 0.98Hz. Notice that
significant energy dissipation occurs. Close examination reveals a dependency of
response on amplitude. This is more clearly seen in Fig. 6.11, which presents the
variation of an equivalent damping constant as a function of the displacement
amplitude. Based upon these experimental results, and those briefly described in
Miyazaki and Mitsusaka (1992), an empirical non-Newtonian macroscopic model has
been developed for the VDW. This model is presented in detail in Reinhorn and Li
(1995). This most recent report also presents data from an extensive set of component
tests conducted on two different panels, approximately 1.0m x 1.0m and 1.0m x 0.8m.
Typical hysteresis loops obtained at room temperature (22C) are shown in Fig. 6.12,
while results for the larger panel are also summarized in Table 6.1. Notice from the
tabular results at 4.0Hz that the response is somewhat amplitude dependent.
Figure 6.10
VDW Hysteresis Loops at 0.98Hz (Arima et al., 1988)
Figure 6.11
VDW Damping Versus Amplitude (Arima et al., 1988)
Figure 6.12
VDW Force-Displacement Response (Reinhorn and Li, 1995)
Table 6.1 Summary of Component Tests on 1m x 1m
Viscous Damping Walls (Reinhorn and Li, 1995)
Storage Damping
Frequency Amplitude Stiffness Coefficient
(Hz) (in) (k/in) (k-sec/in)
where v represents the axial velocity of the fluid within the annular region. (Again the
indices following commas represent differentiation with respect to spatial coordinates.
However, these indices do not vary, and hence summation is no longer implied by
repeated subscripts.)
Let us consider two limiting cases. The first applies for low viscosity fluids ,
relatively large gaps h, short flow passages Lp or high velocity flow. For this case, an
inviscid fluid idealization is appropriate, and Eq. (6.36) becomes simply
For low frequencies, the time-dependent term also becomes negligible, and one is left
with
pvv,z = -p,z (6.38)
which is the one-dimensional steady version of Euler's equation. Integrating both sides
of Eq. (6.38) along the flow length Lp, one obtains Bernoulli's equation
where p and v1 represent the pressure differential across the piston head and the exit
velocity, respectively. In Eq. (6.39), the entrance velocity has been assumed to be zero.
After invoking the conservation of mass, one obtains the following relationship
between overall damper force PI and velocity V:
PI = CIV2 (6.40)
where
Notice from Eq. (6.40) that the damper force is proportional to the velocity squared,
and that in this limiting case there is no energy dissipation. The damper force is inertial,
generated exclusively by accelerating the fluid through the annular passage. This
velocity squared relationship is generally not desirable for structural applications
because of the dramatic increase in forces at higher velocities.
At the other end of the spectrum, let us examine the case resulting from the use of high
viscosity fluids, small gaps, long flow passages or low velocity flows. In this situation,
the convective terms in Eq. (6.36) can be ignored, along with the variations of v with z.
Then, for low frequency excitation, Eq. (6.36) reduces to the form
v,rr = p,z (6.42)
After applying no-slip boundary conditions and invoking the conservation of mass, the
damper response can be written:
PV = CVV (6.43)
where
Thus, one obtains the case of a linear viscous damper, with energy dissipation
occurring due to strictly viscous action within the annular passage.
Taking the ratio of the inertial and viscous limiting cases produces
through specially designed passages located in and around the piston head. A schematic
of the fluidic control orifice design is provided in Fig. 6.14. By changing the
configuration of that orifice, nearly power law force-velocity behavior is obtained for
the damper response (Constantinou et al., 1993). That is,
with real parameters and C0 representing the relaxation time and zero frequency
damping coefficient, respectively. Comparisons of the model with the experimentally
determined storage stiffness, damping coefficient and phase angle are displayed in Fig.
6.16. The relaxation time A = 0.006s is quite small. This indicates that, below a cut-off
frequency of approximately 4Hz, Eq. (6.47) can be simplified further by neglecting the
second term on the left hand side which becomes insignificant. One is then left with the
linear, purely viscous dashpot model
In this case, the subsequent structural analysis becomes greatly simplified and standard
analysis packages are directly applicable.
Figure 6.15
Orificed Fluid Damper Force-Displacement
Response (Constantinou and Symans, 1993b)
Figure 6.16
Orificed Fluid Damper Response (Constantinou and Symans, 1993b)
Next, consider the fluid device shown in Fig. 6.2b that utilizes the orificed flow of a
pressurized compressible silicone-based elastomer in order to enhance stiffness and
damping (Pekcan et al., 1995). Figure 6.17 presents typical force-displacement loops
obtained experimentally for the device. Until the internal preload is overcome, the
device remains quite stiff. However, beyond that level of axial force, the piston
compresses the elastomer and forces it to flow through the orifice provided between the
piston head and the inner casing. Thus, one obtains both a stiffness and damping
component in the response. Upon removal of the extenal loading, the device tends
toward its undeformed state
Figure 6.17
Force-Displacement Response of a
Pressurized Fluid Damper (Pekcan et al., 1995)
due to the initial internal elastomer pressure. The following mathematical model was
developed in Pekcan et al. (1995) to describe the response under general conditions:
where Py, K1 and K2 represent the damper static prestress force, the initial damper
stiffness and the elastomeric damper stiffness, respectively. Meanwhile, the remaining
model parameters C, xmax and define the
damping constant, the maximum device stroke and the velocity exponent, respectively.
A typical result obtained with Eq. (6.49), provided in Fig. 6.18, is in general agreement
with the experimental response.
Figure 6.18
Response of a Pressurized Fluid Damper Based
Upon Analytical Model (Pekcan et al., 1995)
A similar fluid restoring device was employed by Tsopelas and Constantinou (1994) to
provide damping and re-centering capability for a base isolation system. This double-
acting preloaded device is illustrated in Fig. 6.19. The resistance is provided by a
combination of effects, including the preload due to initial presssurization, the device
stiffness associated with the compressibility of the silicone oil, seal friction, and
damping due to the passage of fluid through the orifices. The following mathematical
model was developed to represent the
Figure 6.19
Preloaded Fluid Device (Tsopelas and Constantinou, 1994)
macroscopic response:
with the four individual terms on the right-hand-side corresponding, in order, to the
mechanisms defined above. In this equation, P0 is the preload, K0 is the stiffness, Pmin is
the seal friction at zero displacement, and Pd is the fluid damping force. Additionally,
one needs the evolution equation for the internal variable Z,
Model parameters include P0, K0, Pmin, Pdp, Pdm, xy, , , , 0, 1, p and m. Some
comparisons between analytical and experimental response, obtained by Tsopelas and
Constantinou (1994), using this model are shown in Fig. 6.20. The correlation is quite
good during both static and dynamic testing.
Several additional fluid dampers have been examined in the recent literature. Of
particular note is the analytical and experimental work detailed in Niwa et al. (1995) on
a mineral oil-based damping device. The classical Maxwell model of Eq. (6.47) was
employed to represent the behavior of the damper under sinusoidal excitation.
Figure 6.20
Force-Displacement Response of Preloaded
Fluid Device (Tsopelas and Constantinou, 1994)
For building models that employ one degree of freedom per floor, the passive damper
force vector fr contains the horizontal components of damper forces acting on the
individual floors. That is,
where nj is the number of identical dampers installed in the jth story and Pj is the
corresponding horizontal force from a single damper. For the classical Maxwell model
under consideration,
with j representing the placement angle from the horizontal for the jth floor dampers.
After performing a Fourier transform on Eqs. (6.54)-(6.56), one obtains
where the tilde symbolizes a Fourier transformed quantity, and
represents the dynamic stiffness of the structure. The frequency dependent coefficient
matrix in Eq. (6.57) can be readily obtained from
with
The time history response can then be obtained by utilizing the discrete Fourier
transform method (e.g., Press et al., 1992). In essence, one solves Eq. (6.57) for
independently at a number of frequencies, and an inverse fast Fourier transform
algorithm is subsequently employed to reconstruct the time domain response x(t). The
cylindrical pot fluid damper and viscous damping wall models can also be
accommodated with this approach. One only needs to replace Eq. (6.60) with the
appropriate frequency domain macroscopic model.
A time history analysis can also be conducted by solving the set of first-order
differential equations (Constantinou and Symans, 1993b):
directly in the time domain. From Eq. (6.56), the various matrices and vectors in Eq.
(6.61) are defined as follows:
Furthermore, complex eigenvalues and eigenvectors can be determined by solving the
generalized eigenvalue problem
(A + B)z = 0 (6.63)
The natural frequencies and damping ratios are then determined from
Clearly from the above brief discussion, there are many different approaches that can
be adopted for the analysis of structures with passive fluid dampers, although in some
cases typical structural analysis software packages do not presently provide these
capabilities, except through user-defined formulations. On the other hand, for the linear
viscous orificed damper, the structural analysis is straightforward. Even in that case,
however, it must be remembered that generally bracing is added to the structure in
conjuction with the passive fluid dampers. It is naturally quite important to include the
stiffness associated with this bracing in the overall structural model.
Selected experimental results for the structure in the single-story configuration are
presented in Table 6.3, while those for the full three-story structure are displayed in
Table 6.4. Peak accelerations, story shear forces and interstory drifts obtained for
scaled El Centro and Taft signals are shown in Fig. 6.22. Notice that all three quantities
are reduced substantially with the introduction of the passive fluid dampers. These
reductions are primarily due to an increased ability to dissipate seismic energy, and not
to a shift in frequency. Figure 6.23 provides energy time histories for the structure in
the single-story configuration subjected to the Taft motion. Notice that fluid dampers,
when present, dissipate a vast majority of the absolute input energy. Comparisons were
also made between the experimental data and results from a numerical analysis that
employed the classical Maxwell model of Eq. (6.56) to represent the behavior of the
fluid dampers. Typical results
Table 6.2 Dynamic Properties of Steel Test Structure under Elastic
Conditions (Constantinou and Symans, 1993b)
Single-story Configuration:
Number Frequency Damping
of Dampers Mode (Hz) Ratio
(%)
0 1 2.00 0.55
2 1 2.04 28.4
4 1 2.10 57.7
Three-story Configuration:
Number Frequency Damping
of Dampers Mode (Hz) Ratio
(%)
0 1 2.00 1.7
2 6.60 0.8
3 12.2 0.3
2 1 2.03 9.9
2 6.88 14.7
3 12.34 5.0
4 1 2.11 17.7
2 7.52 31.9
3 12.16 11.3
6 1 2.03 19.4
2 7.64 44.7
3 16.99 38.0
Figure 6.22
Experimental Results for Three-story Steel
Test Structure (Constantinou and Symans, 1993b)
obtained for the first story total damper force are presented in Fig. 6.24 for three
different earthquake motions. The correlation is quite good, suggesting that a
reasonable model has been developed for both the dampers and the primary structure.
Furthermore, Constantinou and Symans (1993b) conclude that the inclusion of viscous
fluid dampers generally reduced drifts by 30% to 70%, while also reducing story shear
forces by 40% to 70%.
Figure 6.23
Energy Time History Response for Steel Test
Structure in Single-story Configuration (Constantinou
and Symans, 1993b)
Figure 6.24
Comparison of Experimental and Analytical Results
for Three-story Steel Test Structure (Constantinou and
Symans, 1993b)
A second more recent study involved the application of this same type of orificed fluid
dampers to a 1:3 scale three-story lightly reinforced concrete frame, tested under the
action of a series of scaled earthquake motions (Reinhorn et al., 1995). The fluid
dampers were positioned within the diagonal braces in the middle bay of each of the
three stories as shown in Fig. 6.25. Dynamic characteristics of the structure with and
without the dampers are provided in Table 6.5. Notice that the addition of the fluid
dampers significantly increased the effective damping ratio of the structure from 3% to
16% under low amplitude excitations. Top story displacement and acceleration
response for the El Centro PGA 0.3g test are provided in Fig. 6.26, while peak
deformations and forces obtained from several different tests are presented in Table 6.6.
Typically, displacements are significantly reduced. However, accelerations and total
forces transmitted to the foundation are only minimally affected, and in some cases
may increase slightly. Figure 6.27 illustrates the beneficial effect of the fluid dampers
in reducing both interstory drift and column shear. The seismic energy input to the
structure increases due to the stiffening associated with the added bracing as indicated
in Fig. 6.28. This is more than compensated for by the ability of the fluid dampers to
dissipate a large portion of the overall energy, thus minimizing damage to the primary
structural members. A comparison of the experimental results with those obtained from
a nonlinear time history analysis using the computer code IDARC was also made, and
excellent correlation was obtained using the Maxwell model for the viscous fluid
dampers.
Figure 6.25
Orificed Fluid Damper Locations for Three-story
Reinforced Concrete Test Structure (Reinhorn et al., 1995)
Table 6.5 Dynamic Characteristics of Reinforced Concrete Test Structure
(Reinhorn et al., 1995)
Figure 6.26
Experimental Response of Three-story Reinforced Concrete Test Structure
Under El Centro 0.3g Motion; a) Top Story Displacement, b) Top Story
Acceleration (Reinhorn et al., 1995)
Table 6.6 Summary of Experimental Results for Three-story
Reinforced Concrete Test Structure (Reinhorn et al., 1995)
Orificed
Fluid First Story Base Shear
Excitation Dampers Drift (in) (kips)
Figure 6.27
Experimental Response of Three-story Reinforced Concrete Test Structure
Under El Centro 0.3g Motion; a, b) Without Dampers; c, d, e) With
Orificed Fluid Dampers (Reinhorn et al., 1995)
Figure 6.28
Energy Response of Three-story Reinforced Concrete
Test Structure Under El Centro 0.3g Motion (Reinhorn
et al., 1995)
A significant amount of experimental work has been performed on the viscous damping
wall concept. Arima et al. (1988) studied the response of a scaled five-story steel
building model and a full-scale four-story test frame on an earthquake simulator. A
typical result for the former structure is
shown in Fig. 6.29, which illustrates the reduction in acceleration levels with the
inclusion of the passive damping walls. More recently, Reinhorn and Li (1995) have
conducted an investigation of viscous damping walls within the same 1:3 scale three-
story reinforced concrete frame utilized for the orificed fluid dampers. Figure 6.30
provides a schematic of the various structural configurations examined in that study.
Details on the dynamic characteristics of this three-story structure before and after
installation of the damping walls are included in Table 6.5. Notice that both the
stiffness and damping are significantly increased. Top story displacement and
acceleration response for the El Centro PGA 0.3g test are displayed in Fig. 6.31.
Generally deformations are significantly reduced due to the enhancement of both
damping and stiffness. However, the latter contributes to an increase in accelerations
and to the forces transmitted to the foundation. Selected results are summarized in
Table 6.7. Once again the passive dampers consume a significant proportion of the
input seismic energy, as indicated in Fig. 6.32.
Figure 6.29
Experimental Response of Five-story Building Model With
and Without Viscous Damping Walls (Arima et al., 1988)
Figure 6.30
Viscous Damping Wall Locations for Three-story Reinforced
Concrete Test Structure (Reinhorn and Li, 1995)
Figure 6.31
Experimental Response of Three-story Reinforced Concrete Test
Structure Under El Centro 0.3g Motion; a) Top Story Displacement,
b) Top Story Acceleration (Reinhorn and Li, 1995)
The last experimental program to be discussed in this section concerns the preloaded
orificed damper shown in Fig. 6.2b. This work, which is detailed in Pekcan et al.
(1995), again examined the response of the aforementioned three-story reinforced
concrete frame building. Dampers were incorporated in diagonal bracing added to the
central bay of the structure. Several different configurations were examined, including
cases in which dampers were placed in all stories, the first two stories, and only the first
story. Dynamic properties of the structure are summarized in Table 6.8. A total of 17
shaking table tests were conducted using four different scaled earthquake motions.
Typical maximum response values are provided in Table 6.9. Significant reductions
in interstory drifts and moderate reductions in base shears result from introduction of
the pressurized fluid dampers. Fig. 6.33 provides a comparison of the response of the
structure in various configurations to the Taft 0.2g motion. Also shown in that figure
are analytical predictions obtained by using the damper model from Eq. (6.49).
Excellent correlation is obtained. The energy response for that same motion is
displayed in Fig. 6.34. Once again, although the input energy increases with the
addition of fluid dampers, the damage to the primary structural members is
significantly reduced as indicated by the decrease in hysteretic energy dissipation.
Figure 6.33
First Story Displacements for Three-story Reinforced Concrete
Test Structure Under Taft 0.2g Motion With Pressurized Fluid
Dampers (Pekcan et al., 1995)
Table 6.9 Maximum Responses for Three-story Reinforced Concrete Test
Structure with Pressurized Fluid Dampers (Pekcan et al., 1995)
Number of
Pressurized First Story Drift/ First Story Shear/
Excitation Fluid Dampers Story Height Total Weight
Figure 6.34
Energy Response of Three-story Reinforced Concrete Test
Structure Under Taft 0.2g Motion (Pekcan et al., 1995)
6.6 Design Considerations
The design of a structure that utilizes viscous fluid dampers for enhanced energy
dissipation, of course, relies heavily on the ability to perform the appropriate structural
analyses. A number of different approaches were discussed previously in Section 6.4.
This analysis is considerably simplified for cases involving the linear orificed fluid
damper, which can usually be modeled as a purely viscous damper. If the dampers are
reasonably well distributed throughout the structure, design reduces primarily to the
selection of the desired level of the critical damping ratio. However, since bracing is
generally required, structural stiffness is also affected. As a result, design
methodologies akin to those described in Section 5.6 are relevant when the response of
the primary structural members and dampers remain in the linear range.
Different methodologies are needed for design in the presence of nonlinearities. Several
rational, simplified nonlinear design procedures are currently under development. For
example, preliminary ideas are presented in Reinhorn et al. (1995), based upon the
construction of a composite demand spectrum relating peak inertia forces (or
accelerations) to peak displacements. With this approach, the structural capacity is
estimated from a monotonic nonlinear pushover analysis.
In detailed design, consideration must be given to the actual load transfer paths. For
example, Reinhorn and Li (1995) discuss the additional column axial forces that
develop due to the insertion of viscous damping walls. As illustrated in Fig. 6.35,
bending moments generated by the damping walls are balanced by axial forces that
accumulate for in-line arrangements. Staggered positioning of the walls can be
beneficial in minimizing this effect. Similar consideration should be given to the
column axial loads associated with other types of passive dampers. Constantinou and
Symans (1993a) discuss the potential benefits of purely viscous devices, which produce
additional column axial forces that are out-of-phase with the column bending moments.
This provides improved margins of safety as determined from a column axial force-
moment interaction analysis. In general, however, existing structures that are retrofitted
with passive dampers may also require strengthening of their columns.
Reliability is, of course, of paramount importance. The mechanical simplicity of
cylindrical pot dampers and viscous damping walls is attractive in this regard.
However, the physical properties of the thick viscoelastic fluids, utilized in these
devices, are often quite sensitive to temperature changes. In the more sophisticated
orificed devices, the working fluids are typically less thermally-sensitive. Furthermore,
some include passive controls to compensate for temperature changes. A critical issue
in the orificed damper designs concerns the durability of the high strength seals,
particularly in those devices that must maintain pressurization throughout their design
life.
Figure 6.35
Load Transfer Paths With Viscous
Damping Walls (Reinhorn and Li, 1995)
Figure 6.38
Damper Installation in the Pacific Bell North Area
Operations Center (Courtesy of Taylor Devices, Inc.,
N. Tonawanda, NY)
Figure 6.40
28 State Street, Boston, MA (Courtesy of
Taylor Devices, Inc., North Tonawanda, NY)
As mentioned previously, these fluid devices are also appropriate for application as
components in base isolation systems. The newly-constructed San Bernardino County
Medical Center in California uses 186 nonlinear Taylor Fluid Dampers, each with a
capacity of 1460kN and a stroke of 600mm, to enhance energy dissipation in the
rubber bearing base isolation system. An overall sketch of the five building complex is
provided in Fig. 6.41. Meanwhile, Fig. 6.42 shows a layout of the damper-base
isolation system assembly, along with a more detailed schematic of the individual
dampers.
Figure 6.41
Architectural Drawing of San Bernardino County Medical
Center (Courtesy of Taylor Devices, Inc., North Tonawanda, NY)
Figure 6.42
Damper-Base Isolation System Assembly for San Bernardino
County Medical Center (Courtesy of Taylor Devices, Inc.,
North Tonawanda, NY)
Viscous damping walls were recently used for aseismic protection in the newly
constructed SUT-Building in Shizuka City, Japan, as described in Miyazaki and
Mitsusaka (1992). A perspective view of the structure is shown in Fig. 6.43. The
building consists of a 78m 14-story above-ground steel frame structure, along with a
two-story reinforced concrete basement. The interior design includes a music hall,
video theater, and banquet facilities. A total of 170 damping walls are employed within
the steel frame to elevate the effective damping ratio of the building to approximately
27% at 20C. A schematic illustrating the locations of the walls is provided in Fig.
6.44. Notice the staggered arrangement. Based upon time history dynamic analysis, the
damping walls reduced response by 70-80%. Consequently, no damage is expected for
level-2 earthquakes having a maximum input velocity of 50cm/s.
Figure 6.43
Perspective View of SUT-Building in
Shizuka City, Japan (Courtesy of
Sumitomo Construction Co., Ltd.,
Tokyo, Japan)
Figure 6.44
Viscous Damping Wall Locations
in SUT-Building (Miyazaki and
Mitsusaka, 1992)
References
Arima, F., Miyazaki, M., Tanaka, H. and Yamazaki, Y. (1988), A Study on Buildings with Large
Damping Using Viscous Damping Walls, Ninth World Conf. Earthquake Engrg., Tokyo, V, 821-
826.
Bagley, R. L. and Torvik, P. J. (1983), Fractional Calculus - A Different Approach to the Analysis
of Viscoelastically Damped Structures, AIAA Journal, 21(5), 742-748.
Banerjee, P. K. (1994), The Boundary Element Methods in Engineering, McGraw-Hill, London,
UK.
Batchelor, G. K. (1967), An Introduction to Fluid Dynamics, Cambridge University Press,
Cambridge.
Bird, R. B., Armstrong, R. C. and Hassager, O. (1987), Dynamics of Polymeric Liquids, John
Wiley, New York.
Chiarugi, A., Melegari, A. and Spinelli, P. (1986), On the Dynamic Behavior of Bridge Dampers
Under Seismic Excitation, Proc. Eighth European Conf. on Earthquake Engrg., 5, Lisbon, 73-80.
Constantinou, M. C., Symans, M.D., Tsopelas, P. and Taylor, D. P. (1993), Fluid Viscous Dampers
in Applications of Seismic Energy Dissipation and Seismic Isolation, Proc. ATC 17-1 on Seismic
Isolation, Energy Dissipation and Active Control, 2, 581-591.
Constantinou, M. C. and Symans, M.D. (1993a), Seismic Response of Structures with
Supplemental Damping, Struct. Design Tall Bldgs., 2, 77-92.
Constantinou, M. C. and Symans, M.D. (1993b), Experimental Study of Seismic Response of
Buildings with Supplemental Fluid Dampers, Struct. Design Tall Bldgs., 2, 93-132.
Ferry, J. D. (1980), Viscoelastic Properties of Polymers, John Wiley, New York.
Gemant, A. (1936), A Method of Analyzing Experimental Results Obtained from Elastoviscous
Bodies, Physics, 7(8), 311-317.
Grenier, A. (1991), Seismic Protection of Bridges by the Viscoelastic Technique, Proc. Third
World Congress Joints and Bearings, Toronto, Canada, 1205-1223.
Harris, C. M. and Crede, C. E. (1976), Shock and Vibration Handbook, 2nd Ed., McGraw-Hill,
New York.
Huffmann, G. K. (1985), Full Base Isolation for Earthquake Protection by Helical Springs and
Viscodampers, Nuclear Engrg. Design, 84(2), 331-338.
Landau, L. D. and Lifshitz, E. M. (1959), Fluid Mechanics, Pergamon Press, Oxford.
Makris, N. (1992), Theoretical and Experimental Investigation of Viscous Dampers in Applications
of Seismic and Vibration Isolation, Ph.D. Dissertation, State University of New York at Buffalo,
NY.
Makris, N. and Constantinou, M. C. (1990), Viscous Dampers: Testing, Modeling and Application
in Vibration and Seismic Isolation, Technical Report NCEER-90-0028, National Center for
Earthquake Engineering Research, Buffalo, NY.
Makris, N. and Constantinou, M. C. (1991), Fractional Derivative Model for Viscous Dampers, J.
Struct. Engrg., ASCE, 117, 2708-2724.
Makris, N. and Constantinou, M. C. (1993), Models of Viscoelasticity with Complex Order
Derivatives, J. Engrg. Mech., ASCE, 119(7), 1453-1464.
Makris, N., Constantinou, M. C. and Dargush, G. F. (1993a), Analytical Model of Viscoelastic
Fluid Dampers, J. Struct. Engrg., ASCE, 119(11), 3310-3325.
Makris, N., Dargush, G. F. and Constantinou, M. C. (1993b), Dynamic Analysis of Generalized
Viscoelastic Fluids, J. Engrg. Mech., ASCE, 119(8), 1663-1679.
Makris, N., Dargush, G. F. and Constantinou, M. C. (1995), Dynamic Analysis of
Viscoelastic Fluid Dampers, J. Engrg. Mech., ASCE, 121(10), 1114-1121.
Miyazaki, M. and Mitsusaka, Y. (1992), Design of a Building with 20% or Greater
Damping, Tenth World Conf. Earthquake Engrg., Madrid, 4143-4148.
Niwa, N., Kobori, T., Takahashi, M., Hatada, T., Kurino, H. and Tagami, J. (1995),
Passive Seismic Response Controlled High-Rise Building With High Damping
Device, Earthquake Engrg. and Struct. Dynamics, 24, 655-671.
Oldham, K. B. and Spanier, J. (1974), The Fractional Calculus, Academic Press,
New York.
Pekcan, G., Mander, J. B. and Chen, S. S. (1995), The Seismic Response of a 1:3
Scale Model R. C. Structure with Elastomeric Spring Dampers, Earthquake
Spectra, 11(2), 249-267.
Press, W. H., Teukolsky, S. A., Vetterling, W. T. and Flannery, B. P. (1992),
Numerical Recipes in FORTRAN, Cambridge University Press, Cambridge, UK.
Reinhorn, A.M. and Li, C. (1995), Experimental and Analytical Investigation of
Seismic Retrofit of Structures with Supplemental Damping, Part III: Viscous
Damping Walls, Technical Report NCEER-95-0013, National Center for
Earthquake Engineering Research, Buffalo, NY.
Reinhorn, A.M., Li, C. and Constantinou, M. C. (1995), Experimental and
Analytical Investigation of Seismic Retrofit of Structures with Supplemental
Damping, Part I: Fluid Viscous Damping Devices, Technical Report NCEER-95-
0001, National Center for Earthquake Engineering Research, Buffalo, NY.
Schwahn, K. J. and Delinic, K. (1988), Verification of the Reduction of Structural
Vibrations by Means of Viscous Dampers, Seismic Engineering, ASME, Pressure
Vessel and Piping Conf., Pittsburgh, PA, 144, 87-95.
Shames, I. H. and Dym, C. L. (1985), Energy and Finite Element Methods in
Sturctural Mechanics, Hemisphere Publishing, New York.
Taylor, D. P. and Constantinou, M. C. (1996), Fluid Dampers for Applications of
Seismic Energy Dissipation and Seismic Isolation, Proceedings of the Eleventh
World Conference on Earthquake Engineering, Acapulco, Mexico.
Tezcan, S. and Civi, A. (1979), Reduction in Earthquake Response of Structures
by Means of Vibration Isolators, Proc. Second U.S. National Conf. on Earthquake
Engrg., Earthquake Engrg. Research Institute, Stanford Univ., California, 433-442.
Tsopelas, P. and Constantinou, M. C. (1994), NCEER-Taisei Corporation
Research Program on Sliding Seismic Isolation Systems for Bridges: Experimental
and Analytical Study of a System Consisting of Sliding Bearings and Fluid
Restoring Force/Damping Devices, Technical Report NCEER-94-0014, National
Center for Earthquake Engineering Research, Buffalo, NY.
7
Tuned Mass Dampers
7.1 Introduction
The objective of incorporating a tuned mass damper into a structure is basically the
same as those associated with metallic dampers and other energy dissipation devices
discussed in the preceding chapters, namely, to reduce energy dissipation demand on
the primary structural members under the action of external forces. This reduction, in
this case, is accomplished by transferring some of the structural vibrational energy to
the tuned mass damper (TMD) which, in its simplest form, consists of an auxiliary
mass-spring-dashpot system anchored or attached to the main structure.
The modern concept of tuned mass dampers for structural applications has its roots in
dynamic vibration absorbers studied as early as 1909 by Frahm (Frahm, 1909; Den
Hartog, 1956). A schematic representation of Frahm's absorber is shown in Fig. 7.1,
which consists of a small mass m and a spring with spring stiffness k attached to the
main mass M with spring stiffness K. Under a simple harmonic load, one can show that
the main mass M can be kept completely stationary when the natural frequency (k/m) of
the attached absorber is chosen to be (or tuned to) the excitation frequency.
Figure 7.1
Undamped Absorber
and Main Mass
Subject to Harmonic
Excitation (Frahm's
Absorber)
Den Hartog (Ormondroyd and Den Hartog, 1928) first studied the theory of undamped
and damped dynamic vibration absorbers in the absence of damping in the main
system, who developed their basic principles and the procedure for proper selection of
absorber parameters. The damping in the main system was included in the analysis of
dynamic vibration absorbers developed by Bishop and Welbourn (1952). Subsequent to
the above work, an optimization procedure was devised by Falcon et al. (1967) to
obtain minimum peak response and maximum effective damping in the main system.
Jennige and Frohrib (1977) numerically evaluated a translatory-rotary absorber system
to control both bending and torsional modes in a building structure. Ioi and Ikeda
(1978) developed empirical formulas for correction factors of these optimum absorber
parameters in the case of light damping in the main system. Randall et al. (1981)
presented design charts for these parameters when the effect of damping in the main
system is taken into account. Warburton and Ayorinde (1980) further tabulated
optimum values of the maximum dynamic amplification factor, tuning frequency ratio,
and absorber damping ratio for specified values of the mass ratio and the main system
damping ratio.
In order to increase the effectiveness of an absorber to reduce maximum dynamic
response of the main system, investigators have attempted to widen the tuning
frequency range by introducing nonlinear absorber springs. Roberson (1952)
considered the dynamic response of a main system supporting a dynamic absorber on a
linear plus cubic spring (i.e., Duffing type spring) without damping. He defined a
'suppression band' as a frequency band between the resonance peaks over which the
normalized main system amplitude is less than unity. This band for the nonlinear
absorber was clearly demonstrated to be much wider than that for a linear absorber.
Pipes (1953) considered a hardening spring with a hyperbolic sine characteristic,
concluding that the effect of nonlinearity in the spring is to prevent the existence of
sharp resonant peaks and to introduce odd harmonic components of relatively small
amplitude in the motion of the absorber and main system.
To improve the performance of dynamic vibration absorbers, Snowdon (1960)
investigated the behavior of a solid-type absorber in reducing the main system
response. It was demonstrated that the dynamic absorber using a material with a
stiffness proportional to frequency and a constant damping factor can reduce the
resonant vibration of the main system considerably. Its performance is clearly superior
to that of the spring-dashpot type absorber. Srinivasan (1969) analyzed parallel damped
dynamic vibration absorbers, i.e., a second undamped tuned mass is added in parallel
with an absorber. The main system would remain stationary in this case when the
damper frequency is exactly tuned to the excitation frequency. However, the
suppression band is decreased in this case. Snowdon (1974) also studied alternative
absorbers such as a triple-element absorber showing that, if a third element, namely a
second spring, is added in series with the damper, a 15% to 30% reduction in the main
system amplitude can be achieved. However, this reduction is quite sensitive to the
tuning frequency, which will affect the absorber performance in practice.
7.1.1 Tuned Mass Dampers
Much of the early development as described above has been limited to the use of
dynamic absorbers in mechanical engineering systems in which one operating
frequency is in resonance with a machine's fundamental frequency. Building structures,
however, are subjected to environmental loads, such as wind and earthquakes, which
possess many frequency components. The performance of a dynamic vibration
absorber, frequently referred to as tuned mass damper (TMD), in complex multi-
degree-of-freedom and usually damped building structures is expected to be different.
Much research and development work in the past two decades has thus been directed to
the study of TMD effectiveness when operating in such a vibrational environment. An
account of the theory and practice of TMDs in structural applications is given in the
following sections.
where y1(t) is the displacement of the structural system relative to its base and z(t) is the
relative displacement of the added mass with respect to the structure. The damping
coefficients and stiffnesses are denoted by c and k for the added mass and C and K for
the structural system, respectively. The external force on the structure is represented by
f(t), while g(t) equals zero for wind excitation and equals f(t) for earthquake loading,
= m/M being the mass ratio.
Summation of Eqs. (7.1) and (7.2) leads to
Figure 7.2
Models of SDOF Structure and TMD
It is seen that the net effect of the added small mass (m) on the structure, aside from a
slight decrease in natural frequency and a slight increase in external force from f(t) to
f(t) + g(t), is the addition of a 'force term' . When f(t) is considered as a
harmonic force or a stationary random input, Eq. (7.3) can be rewritten in the form of
energy or power balance as
in which < > is the mathematical expectation for the case of stochastic input and the
time average in one cycle for the case of harmonic excitation. When the steady-state
response y1(t) is concerned, . Equation (7.4) is thus changed into
the simple power balance equation:
in which C < > is the dissipated power due to structural damping and < > is
the input power from external force which is always positive. As indicated in Section
7.1, the power flow (m < >) from the structural system to the secondary mass plays a
basic role in the use of TMDs in structures and thus is an appropriate measure of TMD
effectiveness, i.e., the larger the power flow, the smaller the mean-square velocity
response of the structural system. Maximum power flow is obtained when the relative
displacement of the secondary mass lags that of the structural system by a phase angle
of 90. In this case, the relative acceleration of the secondary mass is in phase with the
velocity response in the structural system and the power flow is equivalent to an
effective dissipative power which increases the total effective damping (Ceq) in the
structural system as shown in Fig. 7.2c, where
It is worth noting here that input power from the external force varies with TMD
parameters. However, the variation is small compared to the power flow and, in fact,
the magnitude of the input power may decrease to produce a change which is favorable
to TMD efficiency.
7.2.1 Den Hartog's Solution
An appreciation of TMD efficiency in reducing structural response can be gained by
following the basic development of Den Hartog for the simple case when the structural
system is undamped (C = 0) and is subject to a sinusoidal excitation with frequency w
(f(t) = Posinwt and g(t) = 0). In this procedure, the dynamic effect of a TMD is
measured in comparison with the static deflection produced by the maximum force
applied statically to the structure. For a sinusoidal excitation with frequency w, the
static deflection is yst = Po/K, while the dynamic amplification factor for an undamped
structural system, R, is
where
= w/ws = forced frequency ratio
Figure 7.3
Amplification Factor as Function of ( = 0.05, = 1)
An objective in adding the TMD is to bring the resonant peak of the amplitude down to
its lowest possible value so that smaller amplifications over a wider frequency
bandwidth, with close to unity, can be achieved. There are two points (P and Q) on
the R vs. plots in Fig. 7.3 at which R is independent of damping ratio a and the
minimum peak amplitude can be obtained by first properly choosing to adjust these
two fixed points to reach equal heights. The optimum frequency ratio following this
procedure is determined as
A good estimate for opt can be determined as the average of two values
which make the fixed points P and Q maxima on an R vs. plot, respectively (Brock,
1946), giving
From Eq. (7.9) one can easily observe that an increase in damper mass always reduces
the maximum amplification factor for the optimally-designed TMD. This maximum
amplification and optimum absorber parameters are summarized in Table 7.1 for a
variety of excitations and response quantities of interest. In Cases 5 and 6 in Table 7.1,
white noise excitation of spectral density So is assumed. Minimizing mean-square
displacement < > with respect to and a by setting < > / = 0 and < >
/a = 0 led to the results presented in Table 7.1 for these two cases.
7.2.2 The Case of Damped Structures
Equation (7.7) for R can also be established when damping is present in the structural
system; however, invariant points P and Q as in the undamped structure case no longer
exist. Hence, one must resort to numerical means in order to determine the optimum
values of a and a. An approach taken by Randall et al. (1981) is to optimize a and a by
minimizing the higher of the two peaks in the amplification curve (Fig. 7.3) using
numerical search procedures. The optimum values of and a for small values of s
can also be formulated empirically (Ioi and Ikeda, 1978). The results are
where s, is the damping ratio of the primary mass. Tolerance or permissible error
ranges attached to the above equations are found to be less than 1% for 0.03 < < 0.40
and 0.0 < s < 0.15, which are the ranges of practical interest.
A frequency locus method proposed by Thompson (1981) also describes optimization
of and a for a damped primary structure. In this procedure, is optimized by
numerical methods and a can be analytically determined once a is known.
A detailed analysis was carried out by Warburton (1982) to determine optimum damper
parameters for both harmonic and random excitations, with random excitation being
applied as a force to the structure (as in the case of wind) or as base acceleration (as in
the case of earthquake). Low damping of the main structure was again assumed in this
analysis.
For harmonic excitations, the optimization criterion was to minimize response
amplitude R, given by
where j is case number referring to a particular case as given in Table 7.2 on the
previous page, Aj and Bj are given in the same table, and
The optimum values of a and (a can again be found by setting Rj/ and Rj/,
equal to zero and solving the resulting equations. However, these equations are
generally nonlinear in and a, and require numerical solutions. As an example,
optimum absorber parameters for a lightly damped SDOF structure are given in Table
7.3 for several values of and s for the case in which random white noise acceleration
with spectral density So is applied to
These empirical equations have errors of less than 1% in the same range of and s, as
Eqs. (7.11) and (7.12).
For building design where both displacement and acceleration are restricted to certain
levels, Eqs. (7.11) and (7.17) and Eqs.(7.12) and (7.18) are at odds. In order to evaluate
their inconsistencies, the values for and are compared in Fig. 7.4 and and
are compared in Fig. 7.5 when p varies for a specific structural damping ratio s. It
can be observed that the optimum damping ratios for displacement and acceleration
reduction are consistent while the frequency ratios show a large discrepancy.
Brock (1946), in his note on Den Hartog's original work, calculated optimum damping
for the case when the damper mass is exactly tuned (which is not optimal) to the
excitation frequency. For this case, the optimum TMD damping is given by
Also, if the damper has only damping and mass and no stiffness (Lanchester type
damper), then optimal damping is given as
Another optimization criterion can be the maximization of effective damping of the
combined structure/TMD system. A closed form expression for this effective damping
of an undamped structure is given by (Luft, 1979)
Maximizing eff with respect to and a by taking partial derivatives, setting them to
zero, and neglecting higher order terms of , we obtain
and
By mixing optimization criteria we can arrive at even simpler expressions. Suppose one
optimizes a under the minimum displacement criterion (Eq. 7.8) and a under the
minimum effective damping criterion, the optimum a is given by (Luft, 1979)
Another optimization criterion can be travel of the damper mass relative to the
structure. Since root-mean-square displacement of the damper mass is proportional to
, one may select a value for a larger than what is optimum from other points of
view, i.e.,
Figure 7.5
Comparison of and (Ioi and Ikeda, 1978)
7.2.4 Wind vs. Earthquake Loads
Under earthquake-type excitations, the effective forces acting on the structure and
TMD are proportional to their masses. In such a case, the external force on the damper
mass, g(t), is equal to f(t). The performance of TMD can still be improved by
increasing the mass ratio as presented in Fig. 7.6. However, this extra force always
reduces the TMD efficiency for any mass ratio because g(t) is in phase with f(t) in this
case as shown in Fig. 7.7. The optimum frequency and damping ratio for the case
without structural damping, given in Eqs. (7.8) and (7.10), are now shifted from opt
and opt to new values as expressed in Table 7.1 (Case 3).
It is to be noted that wind and earthquake loads are stochastic in nature. While these
loads are nonstationary, an analysis of a structure/TMD system under white noise
excitation will shed more light on the effectiveness of TMD with regard to its
application in wind and earthquake engineering.
When f(t) is a white noise, the steady-state displacement y1(t) is stationary and its
variance can be obtained by integrating its power spectral density over an infinite
frequency range (Soong and Grigoriu, 1993). The optimum damper parameters of an
undamped structure for minimizing the displacement variance are given in Table 7.1
(Cases 5 and 6) for both wind and earthquake type excitations. As has been pointed out
earlier, for a damped structure, it is not easy to obtain a closed-form solution for the
optimum damper parameters, but it can be studied numerically (Wiesner, 1979;
Wirsching and Campbell, 1974; Warburton, 1982).
Figure 7.6
Amplification Factor as Function of and
Figure 7.7
Amplification Factor as Function of g(t) and
As another illustration, the optimum TMD frequency and damping ratio for a damped
structure are presented as functions of the mass ratio in Figs. 7.8 and 7.9, respectively.
Included here are the results from the Den Hartog optimum tuning based on the
sinusoidal solution of an undamped structure along with the Wiesner tuning and
Wirsching and Campbell tuning for wind and earthquake white noise solutions of
damped structures, respectively. Since these curves represent optimum tuning under
different conditions, one can expect some differences among these results. However,
for small mass ratios, these differences are tolerable in practical design (Sladek and
Klingner, 1983).
Figure 7.8
aopt as Function of (Sladek
and Klingner, 1983)
Figure 7.9
opt as Function of (Sladek
and Klingner, 1983)
Figure 7.10
eq as Function of and B (Tanaka and Mak, 1983)
Figure 7.11
opt as Function of and B (Tanaka and Mak, 1983)
Figure 7.12
eq as Function of Forced Frequency Ratio
and B ( = 0.98) (Tanaka and Mak, 1983)
coordinate yN (the displacement at the floor where the damper is attached) and a mode
shape 1, i.e.,
y = 1yN (7.28)
Substituting Eq. (7.28) into Eq. (7.26) and premultiplying Eq. (7.26) by , the
equation of motion (7.26) becomes
By comparing Eqs. (7.27) and (7.29) with Eqs. (7.2) and (7.1), one can find that first
modal representation of a MDOF structure is exactly the same as a SDOF structure,
except that the modal mass, stiffness and damping are employed instead of physical
parameters in the SDOF case. The mass ratio () in this case is defined as .
Under wind-type loading, g(t) = 0 and TMD effectiveness when attached to one floor of
an MDOF structure is exactly the same as in the SDOF system discussed before. For
earthquake-type excitations, however, the relation between g(t) and f(t) is
which is different from that in the SDOF case. However, this has a minor influence on
the behavior of a TMD. When the TMD is situated at the top floor of a building
structure, every element of the modal vector 1 is less than or equal to unity. The
participation factor 1 is thus always greater than unity which improves the TMD
effectiveness in the sense that the effect of g(t) on the damper is decreased. On the
other hand, if the damper is installed at a lower floor, the mass ratio is reduced at a
greater rate than is the participation factor, and therefore TMD effectiveness is
expected to diminish.
7.3.1 Elastic Structures
When the first mode of a MDOF structure dominates the structural response, a response
reduction of the MDOF structure using TMD under earthquake-type loading can be
easily achieved. Wirsching and Campbell (1974) investigated the minimum first-mode
response as well as optimum absorber parameters for one-story, five-story and ten-story
structures. Results show that dimensionless response ratio yN/o, the ratio of root-
mean-square (RMS) top-floor displacement with and without TMD, increases little as
the number of stories increases. The optimum TMD stiffness becomes less sensitive to
structural damping and mass ratio when the number of stories increases and the
optimum TMD damping is insensitive to structural damping even for one-story
buildings. In their paper, mass ratio was defined as the ratio of the added mass and
story mass of a uniform building. If this mass ratio is defined as the modal mass ratio,
the above conclusions become even more obvious, i.e., all the previously discussed
parameters are practically independent of the number of stories. This confirms the
effectiveness of TMD in reducing the first-mode response of multi-story structures and
does not preclude the use of TMD design when the structural damping cannot be
identified accurately. The optimum RMS response (yN/o) of the structure under
random excitation at the base and the corresponding relative displacement RMS (z/o)
of TMD for various structural damping ratios are shown in Fig. 7.14 from which one
can observe that yN/o decreases at a slow rate but z/o decreases substantially as p
increases within the practical range. Additionally, yN/o and z/o both increase as
structural damping increases.
Figure 7.14
yN/o and z/o as Functions of
Figure 7.15
Tower and TMD Displacements (Isyumov et al., 1975)
Figure 7.16
Average Hourly Peak Deflections for
Different TMD Configurations (McNamara, 1977)
Figure 7.17
Average Hourly Peak Accelerations for
Different TMD Configurations (McNamara, 1977)
The leaf spring was designed to be adjustable to measure the model response in X and Y
directions separately. As a TMD model, a simple pendulum consisting of an aluminum
block and a thin wire was secured to the top cover of the building model. The proper
stiffness and damping was provided to the TMD by adhering a foam tape to the wire.
The schematic drawing of the prototype CAARC building is shown in Fig. 7.18b. The
model was tested once without, then with, the TMD system. The effectiveness of the
damper in suppressing dynamic response of the building is clearly demonstrated in
Figs. 7.19 and 7.20. The reduction in magnitude was significant, generally in the range
of 30% to 60%. The reduced velocity Vr is defined as VH/wD (wX = wY = 0.2 Hz) in the
X and Y directions, respectively, in which VH is the horizontal wind speed at a particular
elevation.
Figure 7.18
Model and Schematic of CAARC Building (Tanaka and Mak, 1983)
Figure 7.19
Typical Responses With and Without TMD (Tanaka and Mak, 1983)
Figure 7.20
RMS Responses With and Without TMD (Tanaka and Mak, 1983)
More recently, Xu et al. (1992) studied torsional response of 1:400 scaled models, with
and without a TMD, in a series of wind tunnel tests. A wooden rectangular prism model
represented a building of 193m in height, with plan dimensions of 76m by 36m. The
TMD was mounted in a transparent plastic enclosure on the top of the model. When
properly tuned, the TMD was found to be effective in reducing torsional vibration
caused by eccentricities between the centers of mass, elasticity, and instantaneous
aerodynamic pressure. However, the importance of properly characterizing wind
excitation mechanisms, building size and shape, and surrounding terrain was
emphasized.
7.4.2 Full-scale Test
Under a temporary control arrangement, the TMD system installed in the Citicorp
Center, New York, was employed to initiate the building motion and then respond to it
upon the cut-off of control power. The building frequency and damping in each
direction were identified at varying amplitudes of building motion up to 0.01g. The
TMD performance was then assessed by damping out the self-excited building motion
in each direction. A portion of the building acceleration and TMD mass relative
displacement time histories were recorded during the latter test and shown in Fig. 7.21
(Wiesner, 1979). It is evident that the relative displacement lags the building
acceleration by a phase angle of 90. The effective building damping described in Eq.
(7.6) was increased to 4% of critical as compared to 1% measured without TMD action.
Figure 7.21
Building Acceleration and TMD
Displacement Time Histories (Wiesner, 1979)
The performance of TMDs in many other buildings as well as in the Citicorp Center are
being monitored to provide a wealth of TMD performance data. Some of these findings
are described in Section 7.6 under structural implementations.
7.5 Design Considerations And Implementational Issues
Based upon the development presented above, a design procedure for an optimum
TMD for a building structure oscillating at a predominant, say, the first fundamental
frequency can be developed. As the first step, the first modal representation of the
structure (ws, s) and the maximum acceptable response level (ymax or max) need to be
specified as described in Fig. 7.22. Based upon the design response spectrum
prescribed in existing codes and provisions, effective damping of the representative
SDOF system can be estimated, i.e., the effective damping ratio which satisfies
whichever is larger, can be used for the determination of optimum TMD parameters.
Here, Sy(ws,eq) and S(ws,eq) are, respectively, the response spectra of displacement
and acceleration at natural frequency w,and damping ratio eq. The required mass ratio
p can be then estimated by (Luft, 1979)
The optimum tuning frequency ratio and damping of the TMD can be estimated from
Figs. 7.8 and 7.9. The final step in designing the TMD is to check that the selected
TMD parameters give the building response which is in the range of predetermined
response threshold. Otherwise, the preliminary design has to be refined using an
iterative procedure.
Figure 7.22
Design Flow Chart
A number of practical restraints, however, must be observed in the engineering design
of a TMD system. First and foremost is the amount of added mass which can be
practically placed on the top story of a building. Once the mass is specified, the TMD
effectiveness is determined and the percentage of response reduction is set. TMD travel
relative to the building is another important design parameter, which can be appreciably
reduced by the selection of proper TMD damping or by choosing a proper damper
configuration. However, large TMD movements often need to be accommodated for a
reasonable response reduction of the building. Another major engineering problem
associated with a sliding mass arrangement is to provide a very low friction bearing
surface so that the mass can respond to the building movement at low levels of
excitation. This becomes more critical when TMD functions as an additional damper to
improve occupant comfort.
Finally, cost is an issue which must be addressed in the evaluation of a TMD for a
specific application. Involved in cost estimation are:
Design and system engineering
Mass damper supporting system and accessories, e.g., pressure balanced support
bearings for a sliding mass damper
Mass damper restraint system usually including anti-yaw torque box or
equivalent hardware, over-travel snubber system with reaction guides and forces
Building preparation and strengthening to support a large amount of actuator and
spring force
Enclosure of space occupied by the entire TMD system
Roof elevator access and hoisting capability during installation
Maintenance and operational cost
Figure 7.23
Various Forms of TMD
7.6.1 Centerpoint Tower, Sydney, Australia
It appears that the first structure in which a TMD was installed is the Centerpoint
Tower in Sydney, Australia (ENR, 1971; Kwok and MacDonald, 1987). The structure
consists of a 150-ft-high office building topped with a 700-ft-high tower as shown in
Fig. 7.24a. Serving as the building water and fire protection supply, the tower's water
tank in conjunction with hydraulic shock absorbers was incorporated into the design of
the TMD to reduce wind-induced motions. The tank, 7-ft deep and 7-ft wide with a
capacity of 35,000 gal hangs from the top radial members of the turret as shown in Fig.
7.24b.
A 40-ton secondary mass was later installed on the intermediate anchorage ring to
further increase damping in the second mode, which resulted in increases in the
damping level from 1.0% to 1.2% and from 0.4% to 1.5% in the first and second mode,
respectively. Results of acceleration measurements showed that the wind-induced
acceleration response was reduced by 40% to 50% (Kwok, 1984; Kwok and
MacDonald, 1990). Other examples of structures having pendulum-type TMDs include
the 102m steel antenna mast atop the 553m CN Tower in Toronto, Canada, which has
two doughnut-shaped TMDs to reduce the second and fourth modes of vibration, and
the 157m Crystal Tower in Osaka, Japan, which also makes use of water storage tanks
at the top of the structure as pendulum TMDs.
Figure 7.24
TMD in Centerpoint Tower (Kwok and MacDonald, 1987)
7.6.2 Citicorp Center, New York and John Hancock Tower, Boston
One of only two buildings in the U.S. equipped with a TMD, the 960-ft Citicorp Center
(Fig. 7.25) has a distributed mass of about 60,000 tons, a first bending mode lateral
natural frequency of about 0.16 Hz, and a damping ratio of 1% in its resonant response
to wind gust force.
The Citicorp TMD, as shown in Fig. 7.26, is installed on the 63rd floor of the building.
At this elevation, the building can be represented by a simple modal mass of
approximately 20,000 tons, to which the TMD is attached to form a 2-DOF system as
represented by Fig. 7.2b. The TMD system specifications are given in Table 7.4. Tests
and actual observations have shown that the TMD produces an approximate effective
damping of 4% compared to the 1% original structural damping, which can reduce the
building acceleration level by about 50%.
Figure 7.25
Citicorp Center, New York
Figure 7.26 TMD in Citicorp Center (Petersen, 1980; 1981)
Table 7.4 Tuned Mass Dampers in John Hancock Tower, Boston and Citicorp
Center, New York
John Hancock Citicorp Center
Boston, MA New York, NY
Typical floor size (ft) 343 105 160 160
Floor area (sq ft) 36,015 25,600
Building height (ft) 800 920
Building modal weight (tons) 47,000 20,000
Building period 1st mode (sec) 7.00 6.25
Design wind storm (years) 100 30
Mass block weight (tons) 2300 373
Mass block size (ft) 18183 30308
Mass block material (type) lead/steel concrete
TMD/AMD stroke (ft) 6.75* 44.50*
Max spring force (kips) 135 170
Max actuator force (kips) 50 50
Max hydraulic supply (gms) 145 190
Max operating pressure (psi) 900 900
Operating trigger - acceleration (g) .002 .003
Max power (HP) 120 160
Equivalent damping (%) 4.0% 4.0%
* Including overtravel
Involved in the design of the Citicorp TMD system are the linear gas spring, pressure
balance supporting system, control actuator, power supply and electronic control
(Petersen, 1980; 1981). The passive spring was provided by a pair of opposed trunnion
mounted and pneumatically precharged cylinders with connected piston rods. The 373-
ton concrete block was supported on twelve pressure balanced bearings, each with a 22-
in diameter. The individual bearings were hydraulically coupled together into three
groups, each group containing its own height control value. Fluid can flow between
individual bearings to accommodate floor surface irregularities. The overall friction
coefficient achieved due to the pressure balanced bearing installation was about 0.003,
requiring about 2400 lbs drive force to compensate for friction losses.
The biaxial motion of the mass block was controlled by two double-acting hydraulic
actuators. Each control and spring actuator subassembly was mounted in a single linear
guide assembly for each axis of the TMD. The mass block was restrained from
rotational motion about its vertical axis (yaw) by a torsion box assembly mounted on
the floor and connected to the mass block by two radius rods.
The same design principles were followed in the development of the TMD for
installation in the John Hancock Tower, Boston (ENR, 1975). In this case, however, the
TMD consisted of two 300-ton mass blocks as shown schematically in Fig. 7.27. They
move in phase to provide lateral response control and out-of-phase for torsional control.
A comparison of the TMD and structural parameters of this system with those of the
Citicorp TMD system is given in Table 7.4.
Figure 7.27
Dual TMD System in John Hancock Tower
7.6.3 Chiba Port Tower, Tokyo Bay, Japan
In Japan, the first TMD was installed in Chiba Port Tower, a 125m high steel structure.
It has a rhombus plan with a side length of 15m. The TMD, as shown in Fig. 7.28,
consists of a mass and two frames overlapping at right angle. The mass can move in X
or Y direction but cannot rotate. Each frame has coil springs and a damping device.
Pertinent values of the TMD and structural parameters are given in Table 7.5 (Kitamura
et al., 1988; Mataki et al., 1989; Ohtake et al., 1992).
Table 7.5 TMD and Structural Parameters, Chiba Port Tower (Kitamura, et al., 1988)
X-direction Y-direction
Structure: 1st mode effective weight (tons) 1200
Period:
1st mode (sec) 2.25 2.70
2nd mode (sec) 0.51 0.57
Damping ratio (%) 0.005
TMD: Weight(tons) 10.0 15.4
Period (sec) 2.24 2.72
Spring constant (tons/cm) 0.080 0.084
Friction force (tons) 0.045 0.045
Damping ratio (%) 0.15
Figure 7.28
TMD in Chiba Port Tower (Ohtake et al., 1992)
Figure 7.29 shows the wind induced root-mean-square acceleration response under
Typhoon 8719 (before tuned) and under strong wind Run 880205 (after tuned). The
response due to Run 880205 is seen to have been reduced by 40% in the X-direction
and by 50% in the Y-direction approximately.
Figure 7.29
RMS Accelerations Without and
With TMD (Mataki et al., 1989)
Figure 7.33
Ring-shaped TMD on Stack A (Ueda et al., 1992)
Figure 7.34
Resonant Response Curves
of Stack A (Ueda et al., 1992)
Some of the damped free vibration test results are shown in Fig. 7.35, where the
logarithmic decrement increases from 0.01 to 0.25. Later, both stacks were observed
simultaneously. In four months, the largest vibration amplitude of the stack without
TMD was 22cm in contrast to 2.7cm of the stack with TMD. The effectiveness of TMD
was thus confirmed and a TMD was also installed in the other stack. Figure 7.36 shows
that, in the small amplitude range less than 10mm, the motions are the same, and the
TMD begins functioning when the stack amplitude exceeds 10mm. The ratio of the
TMD and stack motion amplitudes is about 3 with a phase lag of about 150 degrees.
Figure 7.35
Damped Free Vibrations of Stack A (Ueda et al., 1992)
Figure 7.36
Comparisons of Stack and TMD Displacements (Ueda et al., 1992)
Figure 7.37
Impact Vibration Damper
Figure 7.38
Sendagaya INTES Building (Courtesy of Takenaka Corp., Japan)
Figure 7.39
Top View of AMD
Figure 7.40
Elevation of AMD
Sufficient data were obtained for evaluation of the AMD performance when the
building was subjected to strong wind on March 29, 1993, with peak instantaneous
wind speed of 30.6 in/sec. An example of the recorded time histories is shown in Fig.
7.41 giving both the uncontrolled and controlled states. Their Fourier spectra using
samples of 30-second durations are shown in Fig. 7.42, again showing good
performance in the low frequency range but not in the region exceeding 3 Hz. The
response at the fundamental mode was reduced by 18% and 28% for translation and
torsion, respectively. Similar performance characteristics were observed during a series
of earthquakes recorded between May 1992 and February 1993.
Figure 7.41
Response Time Histories (March 29, 1993)
Figure 7.42
Response Fourier Spectra (March 29, 1993)
The 160m 34-story Hankyu Chayamachi building (Fig. 7.43) in Osaka, Japan, also uses
an active TMD system. In this case, the heliport at the roof top is utilized as the moving
mass of the AMD, which weighs 480 tons and is about 3.5% of the weight of the tower
portion. The layout of AMD devices is shown in Fig. 7.44. The heliport is supported by
six multi-stage rubber bearings. The natural period of the rubber and heliport system
was set to 3.6 seconds, slightly lower than that of the building (3.8 seconds). The AMD
mechanism used here has the same architecture as that of Sendagaya INTES, namely,
scheme of the digital controller, servomechanism and the hydraulic design, except that
two actuators of 5-ton thrusts each are attached in horizontal orthogonal directions.
Torsional control is not considered here.
Figure 7.43
Hankyu Chayamachi Building (Courtesy of Takenaka Corp., Japan)
Acceleration time histories during a recent typhoon are shown in Fig. 7.45 with
corresponding Fourier spectra given in Fig. 7.46. Since the building in this case
oscillated primarily in its fundamental mode, significant reductions in acceleration
levels were observed.
Figure 7.44
Layout of AMD
Figure 7.45
Acceleration Time Histories
Figure 7.46
Acceleration Fourier Spectra; a) x-direction
Figure 7.46 Acceleration Fourier Spectra; b) y-direction
References
Abe, M. and Fujino, Y (1994), Dynamic Characteristics of Multiple Tuned Mass
Dampers and Some Design Formulas, Earthquake Eng. Struct. Dyn., 23(8), 813-
836.
Bishop, R. E. D. and Welbourn, D. B. (1952), The Problem of the Dynamic
Vibration Absorber, Engineering, Lond., 174, 769.
Brock, J. E. (1946), A Note on the Damped Vibration Absorber, J. Appl. Mech.,
13(4), A-284.
Carter, W. J. and Lin, F. C. (1961), Steady-state Behavior of Nonlinear Dynamic
Absorber, J. Appl. Mech., 60-WA-14, 67-70.
Chowdhury, A. H., Iwuchukwu, M.D. and Garske, J. J. (1987), The Past and
Future of Seismic Effectiveness of Tuned Mass Dampers, Structural Control,
(Leipholz, H. H. E., ed.), Martinus Nijhoff Publishers, 105-127.
Clark, A. J. (1988), Multiple Passive Tuned Mass Damper for Reducing
Earthquake Induced Building Motion, Proc. 9th Wld. Conf. Earthquake Eng., 5,
Japan, 779-784.
Den Hartog, J. P. (1956), Mechanical Vibrations, 4th Edition, McGraw-Hill, NY.
ENR (1971), Tower Cables Handle Wind, Water Tank Dampens It, Engineering
News-Record, Dec. 9, 23.
ENR (1975), Hancock Tower Now to Get Dampers, Engineering News-Record,
Oct. 30, 11.
Falcon, K. C., Stone, B. J., Simcock, W. D. and Andrew, C. (1967), Optimization
of Vibration Absorbers: A Graphical Method for Use on Idealized Systems with
Restricted Damping, J. Mech. Eng. Science, 9, 374-381.
Frahm, H. (1909), Device for Damped Vibrations of Bodies, U.S. Patent No.
989958, Oct. 30, 1909.
Fujino, Y. and Abe, M. (1993), Design Formulas for Tuned Mass Dampers based on a Perturbation
Technique, Earthquake Eng. Struct. Dyn., 22, 833-854.
Grossman, J. S. (1990), Slender Concrete Structures - The New Edge, A CI Structural Journal,
87(1), 39-52.
Higashino, M. and Aizawa, S. (1993), Application of Active Mass Damper System in Actual
Buildings, Proc. Int. Workshop on Structural Control, Hawaii, 194-205.
Hrovat, D., Barak, P. and Rabins, M. (1983), Semi-active versus Passive or Active Tuned Mass
Dampers for Structural Control, J. Eng. Mech., ASCE, 109(3), 691-705.
Hunt, J. B. (1979), Dynamic Vibration Absorbers, Mechanical Engineering Publications Ltd.,
London.
Ioi, T. and Ikeda, K. (1978), On the Dynamic Vibration Damped Absorber of the Vibration System,
Bulletin of Japanese Society of Mechanical Engineering, 21(151), 64-71.
Isyumov, N., Holmes, J. and Davenport, A. G. (1975), A Study of Wind Effects for the First
National City Corporation Project - New York, U.S.A., Research Report BLWT-551-75, University
of Western Ontario, London, Ontario, Canada.
Jacquot, R. G. and Hoppe, D. L. (1973), Optimal Random Vibration Absorbers, J. Eng. Mech.,
ASCE, 99, 612-616.
Jennige, R. L. and Frohrib, D. A. (1977), Alternative Tuned Absorbers for Steady State Vibration
Control of Tall Structures, J. Mech. Des., ASME, Paper No. 77-DET-84, 1-7.
Kawaguchi, A., Temamura, A. and Omote, Y. (1992), Time History Response of a Tall Building
with a Tuned Mass Damper under Wind Force, 8th Int. Conf. Wind Eng., ed. A. G. Davenport et al.,
London, Ontario, Canada, 1949-1960.
Kaynia, A.M., Veneziano, D. and Biggs, J. M. (1981), Seismic Effectiveness of Tuned Mass
Dampers, J. Struct. Div., ASCE, 107, 1465-1484.
Kitamura, H., Fujita, T., Teramoto, T. and Kihara, H. (1988), Design and Analysis of a Tower
Structure with a Tuned Mass Damper, Proc. 9th World Conf. Earthquake Eng., Japan, 8, 415-420.
Kumagai Gumi Co., Ltd. (1991), Private Communication, Tokyo, Japan.
Kwok, K. C. S. (1984), Damping Increase in Building with Tuned Mass Damper, J. Eng. Mech.,
ASCE, 110(11), 1645-1649.
Kwok, K. C. S. and MacDonald, P.A. (1987), Wind-induced Response of Sydney Tower, Proc.
First National Struct. Eng. Conf., 19-24.
Kwok, K. C. S. and MacDonald, P.A. (1990), Full-scale Measurements of Acceleration Response
of Sydney Tower, Engineering Structures, 12, 153-162.
Lee, J. (1990), Optimal Weight Absorber Designs for Vibration Structures Exposed to Random
Excitations, Earthquake Eng. Struct. Dyn., 19, 1209-1218.
Luft, R. W. (1979), Optimal Tuned Mass Dampers for Building, J. Struct. Div., ASCE, 105(12),
2766-2772.
Mataki, Y., Ohkuma, T., Kanda, J., Kitamura, H., Kawabata, S. and Ohtake, K. (1989), Full-scale
Measurement of Wind Actions on Chiba Port Tower, Takenaka Technical Research Report No. 42,
Takenaka Corporation, Tokyo, Japan.
McNamara, R. J. (1977), Tuned Mass Dampers for Buildings, J. Stuct. Div., ASCE, 103(9), 1785-
1789.
Mitsubishi Heavy Industries, Ltd. (1991), Tuned Active Damper for High-rise Buildings, Seismic
Isolation and Response Control for Nuclear and Non-nuclear Structures, SMiRT 11, Tokyo, Japan,
93-96.
Ohtake, K., Mataki, Y., Ohkuma, T., Kanda, J. and Kitamura, H. (1992), Full-scale Measurements
of Wind Actions on Chiba Port Tower, J. Wind Engineering and Industrial Aerodynamics, 41-44,
2225-2236.
Okumura Corporation (1991), R&D on Base Isolation System and Vibration Control System,
Seismic Isolation and Response Control for Nuclear and Non-nuclear Structures, SMiRT 11,
Tokyo, Japan, 125-132.
Ormondroyd, J. and Den Hartog, J.P. (1928), The Theory of the Dynamic Vibration Absorber,
Trans. ASME, APM-50-7, 9-22.
Petersen, N. R. (1980), Design of Large Scale TMD, Structural Control, (Leipholz, H. H. E., ed.)
North Holland, 581-596.
Petersen, N. R. (1981), Using Servohydraulics to Control High-rise Building Motion, Proceeding of
National Convention on Fluid Power, Chicago, 209-213.
Pipes, L. A. (1953), Applied Mathematics for Engineers and Physicists, Chapter 8, McGraw-Hill,
NY.
Randall, S. E., Halsted, D. M. and Taylor, D. L. (1981), Optimum Vibration Absorbers for Linear
Damped Systems, J. Mech. Des., ASME, 103, 908-913.
Roberson, R. E. (1952),Synthesis of a Non-linear Dynamic Vibration Absorber, J. Franklin Inst.,
254, 205-220.
Setareh, M. (1994), Use of the Doubly-Tuned Mass Dampers for Passive Vibration Control,
Proceedings of the First World Conference on Structural Control, Los Angeles, CA, 1, WP4-12 -
WP4-21.
Sladek, J. R. and Klingner, R. E. (1983), Effect of Tuned Mass Dampers on Seismic Response, J.
Struct. Eng., ASCE, 109(8), 2004-2009.
Snowdon, J.C. (1960), Steady-state Behavior of the Dynamic Absorber, J. Acoust. Soc. Am., 31(8),
1096-1103.
Snowdon, J.C. (1974), Dynamic Vibration Absorbers That Have Increased Effectiveness, J. Eng.
for Ind., ASME, Paper No. 74-DE-J, 940-945.
Soong, T. T. and Grigoriu, M. (1993), Random Vibration of Mechanical and Structural Systems,
Prentice Hall, Englewood Cliffs, NJ.
Soong, T. T., Reinhorn, A.M., Aizawa, S. and Higashino, M. (1994), Recent Structural
Applications of Active Control Technology, J. Structural Control, 1(2), 5-21.
Srinivasan, A. V. (1969), Analysis of Parallel Damped Dynamic Vibration Absorbers, J. of Eng. for
Ind., Trans. ASME, 91, 282-287.
Tanaka, H. and Mak, C. Y. (1983), Effect of TMD on Wind Induced Response of Tall Buildings,
Journal of Wind Engineering and Industrial Aerodynamics, 11, 357-368.
Thompson, A. G. (1981), Optimum Tuning and Damping of A Dynamic Vibration Absorber
Applied to a Force Excited and Damped Primary System, J. Sound Vib., 77, 403-415.
Ueda, T., Nakagaki, R. and Koshida, K. (1992), Suppression of Wind Induced Vibration by
Dynamic Dampers in Tower-like Structures, J. of Wind Engineering and Industrial Aerodynamics,
41-44, 1907-1918.
Warburton, G. B. and Ayorinde, E. O. (1980), Optimum Absorber Parameters for Simple Systems,
Earthquake Eng. Struct. Dyn., 8, 197-217.
Warburton, G. B. (1982), Optimal Absorber Parameters for Various Combinations of Response and
Excitation Parameters, Earthquake Eng. Struct. Dyn., 10, 381-401.
Wiesner, K. B. (1979), TMD to Reduce Building Wind Motion, ASCE Spring Convention, Boston,
MA.
Wirsching, P. H and Yao, J. T. P. (1970), Modal Response of Structures, J. Struct. Div., ASCE,
96(4), 879-883.
Wirsching, P. H. and Campbell, G. W. (1974), Minimal Structural Response under
Random Excitation Using the Vibration Absorbers, Earthquake Eng. Struct. Dyn.,
2, 303-312.
Xu, K. and Igusa, T. (1992), Dynamic Characteristics of Multiple Substructures
with Closely Spaced Frequencies, Earthquake Eng. Struct. Dyn., 21, 1059-1070.
Xu, Y. L. and Kwok, K. C. S. (1992), Wind Induced Response of Soil-Structure-
Damper System, 8th Int. Conf. Wind Eng., ed. A. G. Davenport et al., London,
Ontario, Canada, 2057-2068.
Xu, Y. L., Kwok, K. C. S. and Samali, B. (1992), Torsion Response and Vibration
Suppression of Wind Excited Buildings, 8th Int. Conf. Wind Eng., ed. A. G.
Davenport et al., London, Ontario, Canada, 1997-2008.
Yamaguchi, H. and Harnpornchai, N. (1993), Fundamental Characteristics of
Multiple Tuned Mass Dampers for Suppressing Harmonically Forced Oscillations,
Earthquake Eng. Struct. Dyn., 22, 51-62.
8
Tuned Liquid Dampers
8.1 Introduction
The concept of the dynamic vibration absorber for passive control of structures was
presented in the previous chapter. In tuned mass dampers (TMD), typically a solid
concrete or metal block acts as the secondary mass, although in some cases a deep tank
filled with water serves the same purpose. Additional springs and dampers are used to
attach this secondary mass to the primary structure, and to provide the restoring and
dissipative mechanisms needed to tune the system for near-optimal response under
various types of dynamic excitations. Figure 8.1a is a schematic of the prototype TMD
attached to an SDOF structure. Many examples of structures with TMDs were
discussed in Chapter 7.
Figure 8.1
Dynamic Vibration Absorbers; (a) General, (b) Tuned Sloshing Damper
In the present chapter, attention shifts to another class of dynamic vibration absorber,
namely tuned liquid dampers (TLD), in which liquids are used to provide all of the
necessary characteristics of the secondary system. One particular type of TLD, the
tuned sloshing damper is illustrated in Fig. 8.1b. In this case, the liquid not only
supplies the required secondary mass, but also the damping through viscous action
primarily in the boundary layers. Meanwhile gravity provides the necessary restoring
mechanism. As a result, the secondary system has characteristic frequencies that can be
tuned for optimal performance, exactly as with TMDs.
In fact, the prototype dynamic vibration absorber, shown in Fig. 8.2, that was proposed
by Frahm in the early 1900s was actually the forerunner of the modern tuned liquid
damper. Frahm tuned the frequency of motion of the water in the two interconnected
tanks to the fundamental rolling frequency of a ship in order to successfully reduce this
component of motion. These Frahm anti-rolling tanks were installed on several large
German ocean liners (Den Hartog, 1956). More recently, the same concept was used in
nutation dampers for space satellites to reduce long period libration motion (Carrier and
Miles, 1960; Alfriend, 1974). Figure 8.3 provides a schematic of a typical design using
a ring partially filled with liquid.
Figure 8.2
Frahm Anti-Rolling Tanks for Ship
Applications (Den Hartog, 1956)
The idea of applying tuned liquid dampers to reduce vibrations in civil engineering
structures began in the mid-1980s. Bauer (1984a, b) suggested the use of a rectangular
container completely filled with two immiscible fluids to dampen response through the
motion of the interface. This concept is depicted in Fig. 8.4a for reduction of wind-
induced motion. Welt and Modi (1989a, b) were also among the first to suggest the use
of a TLD in buildings to reduce overall response during strong wind or earthquakes. A
few of the proposed devices studied by Welt and Modi are shown in Fig. 8.4b. All of
these operate on the principle of liquid sloshing. Due to the geometric similarities, these
devices are sometimes referred to as nutation dampers. Several other types of tuned
liquid dampers have also been proposed. The simplest of these, known
Figure 8.3
Nutation Damper for Satellite
Applications (Fujino et al., 1988)
as the tuned liquid column damper is shown in Fig. 8.4c (Xu et al., 1992). This device
consists of a partially-filled tube containing several internal orifices. The fundamental
frequency of the system is dependent only on the length of the column of water, while
damping is provided by flow through the orifices. A number of recent structural
applications, employing dampers similar to those illustrated in Fig. 8.4, will be
discussed later in this chapter.
Figure 8.4
Tuned Liquid Dampers for Structural
Applications; (a) Immiscible Fluids
Damper (Bauer, 1984b)
Figure 8.4
Tuned Liquid Dampers for Structural Applications;
(b) Nutation Dampers (Modi et al., 1995), (c)
Liquid Column Damper (Xu et al., 1992)
Much of the theoretical basis for TLDs has already been presented. A brief overview of
the relevant fundamentals of fluid mechanics was provided in Section 6.2, while the
formulations needed for properly tuning dynamic vibration absorbers have been
discussed in Section 7.2 in connection with tuned mass dampers. However, unlike
tuned mass dampers, which often respond in a purely linear manner, the behavior of
TLDs is generally highly nonlinear due to the fluid motion. Consequently significant
research effort has been directed toward both the analytical and experimental
characterization of their behavior.
The following two sections provide a review of this work. Section 8.2 concentrates on
the generic problem involving the response of a shallow fluid layer subjected to
horizontal excitation. Then, in Section 8.3, extensions of these models suitable for
direct application to TLDs will be examined, along with experimental studies of
damper response. After obtaining a better appreciation for the behavior of tuned liquid
dampers, attention shifts to the impact of TLDs on overall structural performance.
Section 8.4 and 8.5 present results obtained from numerical and experimental studies,
respectively. Although the mathematical description of TLD response is difficult,
structural implementation is often quite simple. Some additional implementation issues
are noted in Section 8.6. The final section provides details of several recent full-scale
implementations in Japan.
Figure 8.5
Progressive Waves in Water of Constant Depth
where subscripts following commas denote partial differentiation with respect to the
spatial coordinate x and a superposed dot indicates a partial time derivative.
Additionally, in Eq. (8.1), v, w, g and represent the kinematic viscosity of the fluid, a
characteristic frequency of the fluid motion, the
Figure 8.6
Tuned Liquid Damper Geometric Definition
gravitational acceleration and the horizontal acceleration of the tank base, respectively.
Notice that nonlinearities appear in the second terms in both equations. Meanwhile, the
fourth term in Eq. (8.1b) involving provides dispersion, and the fifth term models
dissipation associated with the boundary layer on the tank bottom and side walls.
Dissipation due to free surface contamination is also included via the parameter C,
which lies in the range from 0 to 2 and is usually assigned a unit value (Miles, 1985).
The relative contributions of nonlinearity and dispersion can be determined by
considering the dimensionless Stokes parameter defined by Us = /h3, where
represents the wave amplitude and is the horizontal wavelength. For small values of
Us, dispersion dominates, while for large values of the Stokes parameter the response
becomes significantly nonlinear. The latter regime is characterized by the appearance of
progressive solitary waves and frequency bifurcations near resonance of the liquid.
These resonant frequencies can be determined approximately by considering the
dispersion relations for linearized response without dissipation. The resulting natural
frequency for the k-th mode can be written:
Lepelletier and Raichlen (1988) solved Eqs. (8.1a, b) numerically using a finite element
approach, and also conducted physical experiments to better understand the response
under harmonic base excitation at frequency w. Figure 8.7 shows the time evolution of
surface waves computed for the sloshing mode with = 2L for three different values of
the Stokes parameter. It should
Figure 8.7
Computed Free Surface Profiles for Sloshing Mode = 2L; (a) Us
= 10, (b) Us = 100, (c) Us = 1000 (Lepelletier and Raichlen, 1988)
be noted that a linearized theory, corresponding to Us = 0, would produce a standing
wave with a single node at x/L = 0.5. The nonlinear result for Us = 10 is evidently quite
similar. However, as the Stokes parameter increases to Us = 100, the effects of
nonlinearity become apparent. A progressive wave now forms having asymmetry about
the nominal fluid level. With Us increased further to 1000, this solitary wave is clearly
defined as it progresses across the length of the tank.
The variation of relative wave extrema at the right-hand edge of the tank are presented
in Fig. 8.8 for excitation frequencies near the first resonance w1 defined by Eq. (8.2). Of
course, a linear model would predict only a single peak. However, in both the
experiments and with the nonlinear theory, a primary peak occurs at w/w1 1.06, while
a secondary peak appears at w/w1 0.97. Notice that the positive peaks near resonance
greatly exceed the negative extrema in concurrence with the surface profiles of Fig. 8.7
that correspond to the propagation of solitary waves. Typical time histories of the free
surface response at x = L are shown in Fig. 8.9 for frequencies near w1 with Us 360. At
w/w1 = 1.04, a single solitary wave appears traveling with a speed given approximately
by its shallow water limiting value of (gh)1/2.
Figure 8.8
Variation of Relative Wave Extrema at
x = L (Lepelletier and Raichlen, 1988)
Figure 8.9
Free Surface Time Histories Near Lowest Resonance; (a)
Experiments, (b) Nonlinear Theory (Lepelletier and Raichlen, 1988)
At somewhat lower frequencies such as w/w1 = 0.96, the single solitary wave becomes
unstable early in the transient process. A bifurcation then occurs
and a second solitary wave appears. The amplitude of this second wave grows with
time until the response corresponds to the second natural mode in which two solitary
waves propagate along the surface with the same speed. Thus, the second mode is
excited at a driving frequency that is approximately one-half of its natural frequency.
This behavior is apparently a consequence of the nonlinearities combined with low
dispersion. These nonlinearities also cause the beating phenomenon that is noticeable in
the figure. For larger values of Us, additional higher modes are excited. In all cases,
Lepelletier and Raichlen obtained a good correlation between the physical experiments
and the finite element solutions, as can be seen from the comparisons in Figs. 8.8 and
8.9.
Similar numerical and experimental results have been obtained by other researchers.
For example, Shimizu and Hayama (1987) developed a formulation from first
principles by assuming incompressible, inviscid and irrotational flow with constant
pressure on the free surface. As a result, for two dimensional flow, the governing
equations of continuity and motion can be written from Eqs. (6.5) and (6.4),
respectively, in the form
where u and w are the horizontal and vertical fluid velocities, respectively, p is the
density and p is the pressure. Again the horizontal base acceleration is , while g
represents the vertical gravitational acceleration. A velocity potential 1 is then
introduced in analogy with the linear theory, such that
where k is a wave number. By utilizing these relationships, Eqs. (8.3) and (8.4) can be
integrated through the fluid layer from the bottom z = -h to the surface z = . After
some manipulation, the results can be written
where
= tanh[kh]/(kh) (8.8a)
= tanh[k(h + )]/tanh[kh] (8.8b)
TH = tanh[k(h + )] (8.8c)
and us is the horizontal velocity on the surface. The above equations are valid for
arbitrary wave number k. Shimizu and Hayama (1987) establish k based upon the first
mode response of the corresponding linear problem. Unfortunately, this destroys the
dispersive character of the solutions. In order to restore some dispersion, the authors
employ a finite difference method with the grid spacing selected such that the
numerical dispersion matches the physical dispersion that is expected under
infinitesimal conditions. The authors compare results from this finite difference
approach with those obtained via physical experiments on a rectangular tank. Response
time histories of surface elevation near first resonance are provided in Fig. 8.10, while
plots of the relative wave extrema are presented in Fig. 8.11. It should be noted that a
damping term was also included in the finite difference formulation, presumably to
enhance the correlation. As a result, the numerical algorithm of Shimizu and Hayama
(1987) provides quite accurate simulations in the range of behavior considered in Figs.
8.10 and 8.11. Furthermore, the response is consistent with that obtained by Lepelletier
and Raichlen (1988).
Figure 8.10
Free Surface Time Histories Near Lowest Resonance; (a) Experiment
With w/w1 = 1.00, (b) Calculation With w/w1 = 1.00, (c) Experiment
With w/w1 = 1.05, (d) Calculation With w/w1 = 1.05 (Shimizu and
Hayama, 1987)
Figure 8.11
Variation of Relative Wave Extrema
(Shimizu and Hayama, 1987)
with
and kinematic viscosity v. The first term on the right side of Eq. (8.9b) obviously now
represents the dissipative contribution. This damping has a significant effect on the
response near resonance and, therefore, must be included. By considering the nature of
the flow within the boundary layer and the energy dissipation per cycle, the form of
can be established as
where w is the frequency of the liquid motion. This term includes the contribution from
viscous dissipation along the side walls and the free surface contamination. There is no
consensus on the actual value to be assumed for the constant Cd in Eq. (8.11). Sun et al.
(1989) set Cd = 8/(37), Fujino et al. (1992) determine that , while Koh et al.
(1994) argue that Cd = 1. Notice that the formulation of Lepelletier and Raichlen (1988)
has a similar damping term in Eq. (8.lb). In any case many assumptions are involved in
these derivations and correlations with experimental data are ultimately needed to
establish the appropriate value.
Figure 8.12
Profile of Horizontal Liquid Velocity
Through the Depth (Sun et al., 1989)
In order to validate the above formulation, Sun et al. (1989) conducted a series of
shaking table experiments on rectangular tank TLDs undergoing harmonic horizontal
base excitation. The typical unit had a length L = 59cm, width b = 33.5cm and still
water level h = 3cm. From the linear theory, the TLD consequently had a first sloshing
mode frequency fw = wl/(2) = 0.46Hz. In addition to examining the free surface profile
as a function of time, Sun et al. (1989) also measured the base shear force and
computed the energy dissipated per cycle. Time histories of the normalized base shear
are shown in Fig. 8.13 for various frequencies near the first resonance, while the
corresponding force-displacement hysteresis loops are displayed in Fig. 8.14 for a base
displacement amplitude of A = 0.25cm. A good correlation is
Figure 8.13
Base Shear Force Response (Fujino et al., 1992)
obtained between the experiments and simulated results. The steady response is
characterized in Fig. 8.15, which presents plots of the normalized wave extrema, shear
force amplitudes and energy dissipation per cycle versus frequency near the first
resonance. Once again good correlation is obtained for base amplitudes of A = 0.5cm
and below. However, significant deviations occur for A = 1.0cm. At that level of
excitation, the wave height often exceeds the still water level, and wave breaking
begins to occur. Since the models do not consider the additional dissipation associated
with breaking waves, the simulations tend to significantly overpredict the wave heights
and base shear forces, as is evident from Fig. 8.15d. A completely analogous
investigation was presented more recently in Sun et al. (1995a) for pitching motions.
Figure 8.14
Typical Damper Force-Displacement Response (Fujino et al., 1992)
Koh et al. (1994) considered the response of rectangular tank TLD units subjected to
earthquake ground motions. Physical and numerical experiments were conducted for
two different tank lengths (59cm, 90cm), several liquid depths from 3cm to 8cm, and
three different earthquake records scaled to various amplitudes. For the simulations, the
model defined in Eqs. (8.9) and (8.11) was employed with Cd = 1 and the frequency w
established from the instantaneous up-crossing rate of the surface waves. The
amplitudes of the earthquake signals were limited to prevent wave breaking, which is
not accounted for in the numerical model. A typical result is shown in Fig. 8.16 for
response due to the El Centro motion. Generally the correlation between the numerical
solutions and the physical experiments was good. The theory tended to predict
somewhat larger free surface displacements and a greater
effect from higher modes. Perhaps the level of damping used in the model was not
sufficient.
As noted above, at higher levels of excitation, the surface waves break and the previous
formulations become invalid. However, in some applications, TLDs must operate in
that range, which is characterized by higher levels of damping. Sun et al. (1992)
proposed an extension of the above formulation to empirically model the effects of
wave breaking. The authors introduce two additional coefficients Cda and Cfr in Eq.
(8.9b) such that the governing equations become
Figure 8.15
Steady Frequency Response in Terms of Relative Wave Extrema,
Shear Forces and Energy Dissipation Per Cycle (Fujino et al., 1992)
Figure 8.15
Steady Frequency Response in Terms of Relative Wave Extrema,
Shear Forces and Energy Dissipation Per Cycle (Fujino et al., 1992)
For small amplitude waves (i.e., < h), these two coefficients are unity, while for
higher amplitudes Cfr models the shift in natural sloshing frequency and Cda
compensates for the additional damping that is associated with breaking. Based upon a
series of parametric experiments, Sun et al. (1992) determined that the following values
are appropriate for these dimensionless coefficients
Figure 8.16
Time Histories of Free Surface Elevation for Scaled El Centro Motion (Koh et al., 1994)
Generally this correction was found to enhance the correlation of the model for higher
amplitude response. An illustration of the performance of this modified formulation
will be presented later in the chapter.
All of the above studies concerned rectangular tank TLDs. On the other hand, Welt and
Modi (1989a, b) examined the behavior of torus-shaped sloshing dampers by using
both analytical and experimental techniques. For the analytical investigation, a
perturbation method is applied to solve the nonlinear free surface equation within a
potential flow formulation. These formulations are, of course, quite involved due to the
problem geometry and consequently will not be presented here. However, Welt and
Modi (1989a) did obtain approximate solutions under both nonresonant and resonant
conditions that suggest the use of long and slender tank geometries with shallow water
heights for efficient damper performance.
As noted earlier, several other types of TLDs have been proposed, including the tuned
liquid column damper (TLCD) shown in Fig. 8.4c. The following simple macroscopic
model of this damper was developed by Saokaet al. (1988)
where, referring to Fig. 8.4c, x is the elevation of the liquid in the tube, y
is the horizontal displacement of the tube, A is the cross-sectional area, B is the
horizontal dimension of the tube and L is its total length. Additionally, p, and g are
the liquid mass density, the coefficient of head loss determined by flow through the
orifices and the gravitational acceleration, respectively. Notice that Eq. (8.14) is
nonlinear due to the presence of the damping term. Under infinitesimal excitation, the
natural frequency of this simplified TLCD model can be written
The model was later verified by Sakai et al. (1989) through a series of experiments.
along with the TLD model of Eq. (8.9). In Eq. (8.9), ws, s and ms represent the natural
frequency, critical damping ratio and mass of the structure, respectively. Furthermore,
xs is the structural displacement, F is the TLD base shear force and Fe is the external
excitation force with frequency f. Numerical solutions were compared with
experimental data obtained for a variety of design configurations and excitation
amplitudes. The specific structure had mass ms = 168kg, damping s = 0.0032 and a
natural frequency fs = 0.91Hz. Two different TLD configurations were considered with
L = 25cm and L = 32cm. For each TLD tank design, the water height was selected to
minimize the steady-state response at f/fs = 1 for an excitation level that caused a bare
structure response amplitude x0 = 1cm. The water levels were thus set at h = 2.1cm and
h = 3.6cm, respectively. The corresponding mass ratios p were 0.01 and 0.087. Typical
results obtained from a frequency sweep in the neighborhood of first resonance are
shown in Fig. 8.17
from the more recent paper by Fujino et al. (1992). Significant reduction in the
displacement response is obtained for this lightly damped structure. For this type and
level of excitation, the TLD performs as an effective dynamic vibration absorber.
Additionally, reasonably consistent results are produced by the mathematical model.
For higher amplitude vibration, wave breaking occurs and the TLD model of Eq. (8.9)
is no longer valid.
Figure 8.17
Frequency Response of Structure With TLD (Fujino et al., 1992)
As noted in Section 8.3, Sun et al. (1992) extended the TLD to empirically account for
wave breaking by introducing two additional coefficients. The resulting TLD model
expressed in Eq. (8.12) can also be combined with Eq. (8.16) to approximate the
response of the structural system at higher levels of excitation. Fig. 8.18 provides the
results obtained from both the model and physical experiments. Once again the
excitation is a simple time harmonic force with frequency around that associated with
the first structural resonance. The experiment indicated the presence of breaking waves,
and clearly from Fig. 8.18 the modified model provides a significantly improved
correlation for this case.
More recently, the performance of TLDs for the reduction of seismic response has been
considered. Koh et al. (1994) examined TLD-structure interaction under earthquake
excitation. Equations (8.9) and (8.16) were again employed, and a tower or tall building
was idealized as an SDOF structure with fs = 0.2Hz and s = 0.03. A synthetic seismic
signal, generated from a Clough-Penzien spectrum, was employed in the analysis and
the amplitude was chosen such that it would cause a bare structure displacement of
10cm. TLDs with mass ratio p. = 0.04, length L = 400cm and water height h = 26.5cm
were included such that fw = fs. Numerical
results are shown in Fig. 8.19. The response of an MDOF structure with multiple TLDs
was also investigated in Koh et al. (1995). However, the modal approach adopted is
strictly valid only in the linear response range for both the structure and the dampers.
More generally, direct time domain methods are needed to assess the performance of
structures utilizing TLDs for seismic response reduction. Furthermore, the amplitude
dependence of the response of structures incorporating tuned liquid dampers must be
carefully examined when considering TLDs for seismic applications.
Figure 8.18
Frequency Response of Structure With TLD (L = 39cm,
h = 3cm, = 1.05%, x0 = 5cm) for Large Amplitude
Excitation (Sun et al., 1992)
Figure 8.19
Time Histories of Structural Displacement Under Seismic
Excitation With and Without TLDs (Koh et al., 1994)
For design purposes, there is a strong desire to simplify the nonlinear equations through
some type of linearization procedure. Xu et al. (1992) and Balendra et al. (1995) take
this approach for tuned liquid column dampers by modifying the nonlinear damping
term in Eq. (8.14) for the damper. Both papers focus on the vibration control of towers
and tall buildings under wind excitation. An examination of the stochastic response of
structures that incorporate generic linearized dynamic vibration absorbers is addressed
in Kareem (1983) and Kareem and Sun (1987).
It should be noted that if the damper behavior can be linearized as a reasonable
approximation, then the optimal design criteria of Chapter 7 becomes applicable. Sun et
al. (1995b) detailed a TMD analogy for the nonlinear sloshing tuned liquid dampers.
The interface force between the damper and the structure was conceptualized as the
force exerted by a virtual mass m, and a virtual dashpot c,. For a linear TMD, the
frequency-dependent values of m, and c, can be obtained from the following analytical
expressions assuming simple harmonic motion with frequency w:
where
However, for TLDs, the virtual mass and damping quantities are also amplitude-
dependent. Consequently, Sun et al. (1995b) performed a series of experiments to
determine effective values for m, a and wa that could be used in Eqs. (8.17) and (8.18)
for sloshing dampers. Typical results for rectangular and circular tanks are shown in
Fig. 8.20 for three different levels of time harmonic excitation. Good fits are obtained
in some, but not all, cases. No attempt was made to use the effective values for the
prediction of the response under more general time-dependent loading.
Figure 8.20
Virtual Mass mv and Damping cv of TLD for
Linear TMD Analogy (Sun et al., 1995b)
Figure 8.22
Damped Free Oscillations of Nagasaki Airport Tower; (a) Without TLD,
(b) With 7 TLD Vessels, (c) With 25 TLD Vessels (Tamura et al., 1995)
Figure 8.23
Across-Wind Response of Nagasaki
Airport Tower (Tamura et al., 1995)
A similar TLD was installed that same year in the landmark Yokohama Marine Tower
shown in Fig. 8.24. The tower is a 100m high, steel trussed structure having a
fundamental frequency of 0.55Hz along with a damping ratio of 0.6%. In this
application, detailed in Tamura et al. (1995), 39 ten-layer sloshing dampers were
positioned in the lighthouse on the top of the tower, and installed as shown in Fig. 8.25.
The individual dampers have a radius of 0.25m and each layer is filled with water to a
height of 0.021m. This produces an overall mass ratio of approximately 1% in the
fundamental mode. Free vibration tests indicated that the damping ratio of the tower
increased to 4.5% with the addition of the TLD. Figure 8.26 presents the root-mean-
square (RMS) acceleration response with and without the TLD in both the along-wind
and across-wind directions, while typical time series comparisons are shown in Fig.
8.27. A clear improvement in response is observed with the addition of the TLD.
Figure 8.24
Yokohama Marine Tower (Tamura et al., 1995)
Figure 8.25
TLD Vessels Installed on the Yokohama
Marine Tower (Tamura et al., 1995)
Figure 8.26
RMS Acceleration Response of Yokohama Marine Tower;
(a) Along Wind, (b) Across-Wind (Tamura et al., 1995)
Figure 8.27
Time Histories of Across-Wind Acceleration Response of Yokohama
Marine Tower; (a) Without TLD, (b) With TLD (Tamura et al., 1995)
In 1992, a TLD was installed in the new 150m high Shin-Yokohama Prince Hotel. This
project, involving 30 multilevel vessels, is described in Wakahara et al. (1992) and
Tamura et al. (1995). However in this case a different style damper was employed with
internal protrusions to enhance energy dissipation. An individual damper, illustrated in
Fig. 8.28, is 2m in diameter with nine water layers, each having a depth of 0.124m.
This provides a mass ratio of nearly 1% and tunes the sloshing resonant frequency to
0.31Hz, which is within a few percent of the fundamental frequency of the hotel. The
damper units are positioned around the periphery of the upper floor as indicated in Fig.
8.29.
Figure 8.28
TLD Vessel for Installation on the Shin-
Yokohama Prince Hotel (Tamura et al., 1995)
Figure 8.29
TLD Vessels Installed on the Shin-
Yokohama Prince Hotel (Tamura et al., 1995)
Figure 8.30
RMS Acceleration Response of the Shin-Yokohama Prince
Hotel; (a) Along Wind, (b) Across-Wind (Tamura et al., 1995)
Two recent applications of tuned liquid dampers to airport towers are detailed in
Tamura et al. (1992, 1994, 1995). The TLD installed in the 78m air-traffic control
tower at Haneda Airport consists of about 1400 vessels containing water, floating
particles and a small amount of preservatives. The vessels, pictured in Fig. 8.31, are
shallow circular cylinders 0.6m in diameter and 0.125m in height. These individual
vessels are then stacked in six layers on steel-framed shelves as shown in Fig. 8.32. The
total mass of the TLD
Figure 8.31
TLD Vessel for Installation on the Haneda
Airport Control Tower (Courtesy of Y. Tanura,
Tokyo Institute of Polytechnics, Japan)
Figure 8.32
TLD Vessels Installed on the
Haneda Airport Control Tower
(Tamura et al., 1995)
is approximately 3.5% of the first-mode generalized mass of the tower and its sloshing
frequency is optimized at 0.743Hz. Floating hollow cylindrical polyethylene particles
were added in order to optimize energy dissipation through an increase in surface area
together with collisions between particles. The performance of the TLD has been
observed during several storm episodes. In one of such episode with a maximum
instantaneous wind speed of 25m/s, the observed temporal variation of the RMS
acceleration and damping ratio of the tower are given in Fig. 8.33, showing that the
damping ratio in the cross-wind direction (x-direction) reached the maximum value of
7.6% before the RMS acceleration recorded its maximum. Figure 8.34 shows a
comparison of the x-direction RMS accelerations of the tower with and without TLD,
indicating that the TLD reduced the response to about 60% of the RMS acceleration
response of the tower without the TLD.
Figure 8.33
Time Variation of RMS Response of the Haneda Airport Control
Tower; (a) Acceleration, (b) Damping Ratio (Tamura et al., 1995)
Tuned liquid dampers can also be beneficial during construction to stabilize various
portions of a structure. Consider the application of TLDs to the Ikuchi Bridge tower
described in Ueda et al. (1992). Based upon wind tunnel tests, it was decided to
suppress vortex-induced oscillation of the 120m tall freestanding tower by using
sloshing dampers. The free standing towers have a natural frequency of 0.25Hz and a
very low damping ratio estimated from free vibration tests to equal 0.2%. In an attempt
to reduce tower displacements from 400mm to approximately 20mm, two 5m by 1.3m
rectangular tank dampers were employed. The tanks contain a number of internal cross
poles in order to increase energy dissipation. Free vibration tests were used to establish
an optimal water height of 0.80m. This provided an effective damping ratio of
approximately 3.7%. Thus, in this case, the tuned sloshing damper approach met the
design objectives.
Figure 8.34
RMS Acceleration Response of the Haneda Airport Control Tower for Near Across-
Wind Excitation (Tamura et al., 1995)
References
Alfriend, K. T. (1974), Partially Filled Viscous Ring Nutation Damper, J.
Spacecraft, 11(7), 456-462.
Balendra, T., Wang, C. M. and Cheong, H.F. (1995), Effectiveness of Tuned
Liquid Column Dampers for Vibration Control of Towers, Engineering Structures,
17(9), 668-675.
Bauer, H.F. (1984a), Oscillations of Immiscible Liquids in a Rectangular
Container: A New Damper for Excited Structures, Journal of Sound and Vibration,
93(1), 117-133.
Bauer, H.F. (1984b), New Proposed Dynamic Vibration Absorbers for Excited
Structures, Vibration Damping Workshop Proceedings, Lynn Rogers (ed.), DD1 -
DD27.
Carrier, G. F. and Miles, J. W. (1960), On the Annular Damper for a Freely
Precessing Gyroscope, J. Appl. Mech., 27, 237-240.
Chaiseri, P., Fujino, Y., Pacheco, B. M. and Sun, L. M. (1989), Interaction of
Tuned Liquid Damper and Structure: Theory, Experimental Verification and
Application, Structural Engineering/Earthquake Engineering, JSCE, 6(2), 273s-
282s.
Den Hartog, J.P. (1956), Mechanical Vibrations, McGraw-Hill, New York.
Fujino, Y., Pacheco, B. M., Chaiseri, P. and Sun, L. M. (1988), Parametric Studies
on Tuned Liquid Damper (TLD) Using Circular Containers by Free Oscillation
Experiments, Structural Engineering/Earthquake Engineering, JSCE, 5(2), 381s-
391s.
Fujino, Y., Sun, L., Pacheco, B. M. and Chaiseri, P. (1992), Tuned Liquid Damper (TLD) for
Suppressing Horizontal Motion of Structures, J. Engrg. Mech., 118(10), 2017-2030.
Hirt, C. W. and Nichols, B. D. (1981), Volume of Fluid (VOF) Method for the Dynamics of Free
Boundaries, J. Comp. Phys., 39,201-225.
Housner, G. W. (1957), Dynamic Pressures on Accelerated Fluid Containers, Bull. Seism. Soc. Am.,
47, 15-35.
Housner, G. W. (1963), The Dynamic Behavior of Water Tanks, Bull. Seism. Soc. Am., 53, 381-
387.
Huerta, A. and Liu, W. K. (1988), Viscous Flow with Large Free Surface Motion, Comp. Meth.
Appl. Mech. Engrg., 69, 277-324.
Kareem, A. (1983), Mitigation of Wind Induced Motion of Tall Buildings, J. Wind Eng. Ind.
Aerodyn., 11, 273-284.
Kareem, A. and Sun, W.-J. (1987), Stochastic Response of Structures with Fluid-Containing
Appendages, J. Sound Vibr., 119, 389-408.
Koh, C. G., Mahatma, S. and Wang, C. M. (1994), Theoretical and Experimental Studies on
Rectangular Liquid Dampers Under Arbitrary Excitations, Earthquake Engrg. Struct. Dynamics,
23, 17-31.
Koh, C. G., Mahatma, S. and Wang, C. M. (1995), Reduction of Structural Vibrations by Multiple-
Mode Liquid Dampers, Engineering Structures, 17(2), 122-128.
Lepelletier, T. G. and Raichlen, F. (1988), Nonlinear Oscillations in Rectangular Tanks, J. Engrg.
Mech., ASCE, 114, 1-23.
Mei, C.C. (1983), The Applied Dynamics of Ocean Waves, John Wiley & Sons, New York.
Miles, J. W. (1985), Resonantly Forced, Nonlinear Gravity Waves in a Shallow Rectangular Tank,
Wave Motion, 7, 291-297.
Modi, V. J. and Welt, F. (1987), Vibration Control Using Nutation Dampers, International
Conference on Flow Induced Vibrations, R. King (ed.), BHRA, London, 369-376.
Modi, V. J., Welt, F. and Seto, M. L. (1995), Control of Wind-Induced Instabilities Through
Application of Nutation Dampers: A Brief Overview, Engineering Structures, 17(9), 626-638.
Okamoto, T. and Kawahara, M. (1990), Two-dimensional Sloshing Analysis by Lagrangian Finite
Element Method, Int. J. Num. Meth. Fluids, 11, 453-477.
Sakai, F., Takaeda, S. and Tamaki, T. (1989), Tuned Liquid Column Damper - New Type Device
for Suppression of Building Vibrations, Proc. Int. Conf. on Highrise Buildings, Nanjing, China,
926-931.
Saoka, Y., Sakai, F., Takaeda, S. and Tamaki, T. (1988), On the Suppression of Vibrations by
Tuned Liquid Column Dampers, Annual Meeting of JSCE, JSCE, Tokyo.
Shimizu, T. and Hayama, S. (1987), Nonlinear Response of Sloshing Based on the Shallow Water
Wave Theory, JSME Int. J., 30, 806-813.
Su, T. C., Lou, Y. K., Flipse, J. E. and Bridges, T. J. (1982), A Nonlinear Analysis of Liquid
Sloshing in Rigid Containers, Department of Transportation, DOT/RSPA/DMA-50/82/1.
Sun, L. M., Fujino, Y., Pacheco, B. M. and Isobe, M. (1989), Nonlinear Waves and Dynamic
Pressures in Rectangular Tuned Liquid Damper: Simulation and Experimental Verification, Struct.
Engrg./Earthquake Engrg., JSCE, 6, 251s-262s.
Sun, L. M., Fujino, Y., Pacheco, B. M. and Chaiseri, P. (1992), Modelling of Tuned Liquid Damper
(TLD), J. Wind Eng. Ind. Aerodyn., 41-44, 1883-1894.
Sun, L. M., Fujino, Y. and Koga, K. (1995a), A Model of Tuned Liquid Damper
for Suppressing Pitching Motions of Structures, Earthquake Engrg. Struct.
Dynamics, 24, 625-636.
Sun, L. M., Fujino, Y., Chaiseri, P. and Pacheco, B. M. (1995b), The Properties of
Tuned Liquid Dampers Using a TMD Analogy, Earthquake Engrg. Struct.
Dynamics, 24, 967-976.
Tamura, Y., Koshaka, R. and Modi, V. J. (1992), Practical Application of Nutation
Damper for Suppressing Wind-Induced Vibrations of Airport Towers, J. Wind
Eng. Ind. Aerodyn., 41-44, 1919-1930.
Tamura, Y., Shimada, K., Sasaki, A., Koshaka, R. and Fujii, K. (1994), Variation
of Structural Damping Ratios and Natural Frequencies of Tall Buildings During
Strong Winds, Proc. 9th Int. Conf. on Wind Engineering, New Delhi, India, 3,
1396-1407.
Tamura, Y., Fujii, K., Ohtsuki, T., Wakahara, T. and Koshaka, R. (1995),
Effectiveness of Tuned Liquid Dampers Under Wind Excitations, Engineering
Structures, 17(9), 609-621.
Ueda, T., Nakagaki, R. and Koshida, K. (1992), Suppression of Wind-Induced
Vibration by Dynamic Dampers in Tower-like Structures, J. Wind Eng. Ind.
Aerodyn., 41-44, 1907-1918.
Wakahara, T., Ohyama, T. and Fujii, K. (1992), Suppression of Wind-Induced
Vibration of a Tall Building Using Tuned Liquid Damper, J. Wind Eng. Ind.
Aerodyn., 41-44, 1895-1906.
Welt, F. and Modi, V. J. (1989a), Vibration Damping Through Liquid Sloshing:
Part I - A Nonlinear Analysis, Proc. Diagnostics, Vehicle Dyn. and Special Topics,
ASME, Design Engineering Division (DE), 18-5, 149-156.
Welt, F. and Modi, V. J. (1989b), Vibration Damping Through Liquid Sloshing:
Part II - Experimental Results, Proc. Diagnostics, Vehicle Dyn. and Special
Topics, ASME, Design Engineering Division (DE), 18-5, 157-165.
Xu, Y. L., Samali, B. and Kwok, K. C. S. (1992), Control of Along Wind
Response of Structures by Mass and Liquid Dampers, J. of Engrg. Mech., ASCE,
118(1), 20-39.
9
Smart Materials
9.1 Introduction
As indicated in Chapter 1, passive energy dissipation constitutes only one of the
possible structural control techniques; considerable research and development efforts
have also been expanded on active control technology involving sensing and actuation
based upon a feedback scheme. Nestled between passive and active structural control
technologies is an emerging area of research addressing the possible use of innovative
or ''smart" materials for sensing and control purposes. This class of smart materials can
be incorporated into structural members or system components as embedded sensors
and actuation elements, capable of modifying structural behavior in response to
external stimuli. Since some of these smart materials provide an alternative, and
potentially attractive, means of energy dissipation in structural systems, they are briefly
introduced here as potential members of a broader family of energy dissipation devices.
The most prominent materials that have been examined as actuation devices in recent
years are shape memory alloys, piezoelectric elements, electrorheological fluids and,
more recently, magnetorheological fluids. These topics are briefly discussed in this
chapter in the context of civil engineering structural control. Let us note that, while
research in these areas has been active in the direction of finding applications in
aerospace structures and mechanical systems such as land vehicles, weapon systems
and robotics, their application potential to the motion control of civil engineering
structures against environmental loads remains to be realistically assessed from the
standpoint of cost as well as technical feasibility.
where all mathematical symbols are defined in Chapter 3. One notes that, on dividing
Eq. (9.la) by , the slope of the - curve is given by
It is seen that the slope is a constant during purely linear elastic loading and unloading
(i.e., when ( - b)/d << 1).
The zdemir model can be modified to simulate SMA behavior by altering the
expression for backstress b as given by Eq. (9.1b) (Graesser and Cozzarelli, 1991).
They have shown that it is possible to describe various aspects of the SMA behavior by
modifying Eq. (9.1b) to the form
where fT, a and c are material constants, u(x) is the unit step function, and erf(x) is the
error function defined by
The second term is added to the zdemir model in order to contribute to the backstress
on the descending branch of the hysteresis in a way that will allow for SMA stress-
strain descriptions. Calculations based upon the SMA constitutive law as described by
Eqs. (9.1a) and (9.3) show that symmetric hysteresis loops can be generated for
martensitic hysteresis (T < TM) and superelastic (T > TA) types of behavior. Also, the
physical constants in the model, which represent the elastic modulus, initial axial yield
and inelastic modulus, were reproduced in the computations. Extensions of the one-
dimensional SMA model to three dimensions have also been carried out (Graesser and
Cozzarelli, 1994).
Experimental characterization testing was also undertaken to evaluate the mechanical
characteristics of SMA under cyclic conditions (Graesser and Cozzarelli, 1991; Aiken
et al., 1992). The work reported in Graesser and Cozzarelli (1991) involved strain
controlled cyclic loading tests of Nitinol specimens, a binary alloy of titanium and
nickel. Varying levels of strain and strain rate were used in the experiments, and a heat
treatment was applied such that a hysteretic material response close to that of
superelasticity was attained. In comparison with the SMA constitutive model developed
above, experimental results show that the SMA model was able to predict the energy
absorbing capacity of the Nitinol specimens to within 35% over the range of strain and
strain rate values tested.
9.2.2 Structural Applications
Some analyses and small-scale experiments have been carried out with respect to the
application of shape memory alloys to civil engineering structural control. In Aiken et
al. (1992), the possible use of Nitinol as a passive energy dissipation device for
structures was studied experimentally by incorporating
small loops of Nitinol wire into diagonal braces in a three-story model structure. The
structural model was six feet in height and weighed 3000 lb. The results of the testing
included two types of behavior of special interest: large strain behavior and cyclic
superelastic behavior. Figure 9.2 shows the large strain behavior of Nitinol, which has
three distinct phases. Initially, the material was stiff and elastic, then at 2% strain the
crystalline structure changed. Here the Nitinol became softer while still remained
elastic, this corresponds to the nearly horizontal region of the hysteresis loop. This
plateau continued until 6% strain was reached, then as the deformation continued, the
additional deformation caused dislocation and permanent deformation in the crystalline
structure, causing an increase in the stiffness. The advantage of this behavior is that, for
low levels of seismic excitation, the structure behaves elastically; for moderate
earthquakes, the Nitinol will dissipate large amounts of energy while remaining elastic
and, for large earthquakes, the structure will stiffen and again dissipate large amounts
of energy.
Figure 9.2
Large Strain Behavior of Nitinol (Aiken et al., 1992)
Figure 9.3 shows the behavior that can be achieved during a moderate earthquake. The
Nitinol wire loops were preloaded so that the hysteretic behavior was confined to the
flat region of the hysteretic curve. During cyclic stretching and relaxation in the
diagonal braces, the increase in the tensile force caused a change in the crystalline
structure of the Nitinol and, when the tensile force decreased, the Nitinol reverted to its
original crystalline structure. This cyclical change in crystalline structure dissipated a
substantial amount of earthquake input energy. Significantly, the Nitinol did not sustain
any dislocations in its crystalline structure to dissipate this energy and thus a Nitinol
energy dissipator could perform in this controlled elastic manner for a
large number of seismic cycles. This behavior is in contrast to that of normal metals
which must be permanently deformed to dissipate large amounts of energy. We also
note that this superelastic material is capable of sustaining strains in the range of 5% to
7%, while normal steels can only sustain strains of less than 0.2% without permanent
deformation.
Figure 9.3
Superelastic Behavior (Aiken et al., 1992)
Figure 9.4
Torsional Bar Design (Witting and Cozzarelli, 1992)
Figure 9.5
Maximum Floor Response for 0.06g El
Centro (Witting and Cozzarelli, 1992)
Figure 9.6
Maximum Floor Response for 0.06g
Quebec (Witting and Cozzarelli, 1992)
No large-scale structural tests involving the use of SMA energy dissipation devices
have been carried out to date and, as indicated in Section 9.1, cost will also be an
important factor in assessing potential applicability of SMA materials to civil
engineering structural control. Nitinol, for example, can be priced as high as $150/lb,
which would be prohibitive in structural applications.
= (d31/hp)V (9.5)
where hp is the thickness of each piezoelectric layer and d31 is the extension
piezoelectric constant. The strain results in a longitudinal stress given by
= (d31/hp)EpV (9.6)
where Ep is the Young's modulus of the piezoelectric material. For a cantilever beam
with width b and depth hm as shown in Fig. 9.8a, the actuator acts as
a concentrated force couple with force magnitude
F = bhp (9.7)
and a bending moment
M(x) = d31Epb(hp + hm)V (9.8)
as shown in Fig. 9.8b.
These basic relationships can be extended to the dynamic analysis of a structural
element under the action of the converse piezoelectric effect.
Figure 9.7
General Layout of a Piezoelectric Actuator
investigations, a set of guidelines was proposed for the design of nonuniform spatial
layer distributions. Later, interests were directed to the study of plate and shell
structures with piezoelectric layers acting as actuators or sensors. Laminated
piezoelectric plate and shell theories were developed based on Love's theory and
Hamilton's principle (Tzou and Gadre, 1989; Lee and Moon, 1989,1990; Tzou, 1992),
although experimental verification was limited to beam-type structures.
The second type of piezoelectric actuators is of segmented design where the actuators
are discontinuously bonded on or embedded in structural elements. Crawley and Luis
(1987) conducted one of the first studies on segmented piezoelectric actuators. Based
on a uniform strain assumption, they derived the static and dynamic analytical models
and carried out three different experiments for cantilever beams with either bonded or
embedded segmented actuators. Crawley and Anderson (1990) further developed more
accurate models of the interaction between induced strain actuators and one-
dimensional structures, including two analytical models and one numerical model. The
actuation strain produced by piezoelectric materials was characterized in greater depth.
The third type consists of discrete actuators for controlling discretized-parameter
structures, such as trusses with node masses. Descriptions of typical discrete actuator
and sensor design and control experiments can be found in
Shibata et al. (1992), Edberg et al. (1992), and Preumont et al. (1992).
Recent research in the development of piezoelectric control technology may be grouped
into the following areas: First, because of complexity associated with piezoelectric
actuator/sensor-structure theory, many investigators have utilized numerical models,
especially finite element approach, to implement control formulations of various
laminated piezoelectric elements or structures. Baz and Poh (1988), Crawley and
Anderson (1990) and Hanagud et al. (1992) developed finite element formulas for
evaluating single beam control using segmented actuators. Tzou and Tseng (1990), Ha
et al. (1992), and Chandrashekhara and Agarwal (1993) carried out finite element
analyses for laminated or composite plates and shells. Second, an important issue in the
design of segmented actuators is optimal placement and sizing. Devasia et al. (1993)
proposed three performance measures to find simultaneous placement and sizing of
distributed piezoelectric actuators in simply supported beams and discussed their
relative effectiveness. Main et al. (1994) derived different cost functions for
determining the optimal placement and thickness of embedded and surface-mounted
piezoelectric actuators in beams as well as plates. Chattopadhyay and Seeley (1994)
established a multiobjective design optimization procedure for control design using
surface-boned piezoelectric actuators and applied the procedure to multistory frame
control. Third, as shown in Fig. 9.7, piezoelectric layers can be used as either actuators
or sensors or both. Recent developments have focused on combining both actuations
into a single piece of a layer segment. Typical theoretical derivations and experimental
verifications on self-sensing piezoactuators can be found in Dosch et al. (1992) and
Garcia et al. (1992). Fourth, an improvement in damper design by replacing the cover
of traditional constrained passive damping treatments by a piezoelectric layer has
shown an enhanced control performance. This new treatment has been referred to as
intelligent constrained layer (ICL). Agnes and Napolitano (1993), for example, added a
piezoelectric layer to a viscoelastic damping layer to provide an ICL damper for
vibration reduction. Baz (1993) proposed a mathematical model resulting in a sixth-
order ordinary differential equation governing bending vibration control of an Euler-
Bernoulli beam through ICL. In addition, Baz and Ro (1993) studied vibration control
of beams with ICL partially covering the beams. At the same time, Shen (1993,1994)
investigated Euler-Bernoulli beams and isotropic plates controlled through the ICL
treatment.
Although considerable progress has been made in research and applications of
piezoelectric control technology, its implementability to large-scale civil engineering
structures remains to be examined. A major limitation is that a very high voltage, of the
order of several hundred or even thousand volts, may be needed in order to generate an
effective control action for large-scale structures. Consider, for example, horizontal
vibration control of a column with top mass 1000 kg under a horizontal earthquake
excitation at
the base by using a piezoelectric distributed actuator as shown in Fig. 9.9. Let
piezoelectric layers cover continuously both side surfaces of the column. By means of a
generalized single-degree-of-freedom approximation, one can derive the piezoelectric
actuator control force acting on the top mass as
in which u(t) is the control force, b and L are the width and length of the column, and hc
is the thickness of the column. Under a 1/3-scaled El Centro earthquake input, it is
estimated that, in order to reduce the uncontrolled top mass displacement by 30%, a
maximum control voltage of 1275 volts is needed; if the required reduction is 50%, the
maximum control voltage will increase to 2340 volts.
Figure 9.9
Column with
Piezoelectric Layers
9.4 Electrorheological Fluids
Electrorheological (ER) materials, which are mostly fluids, are characterized by their
ability to undergo dramatic reversible increases in resistance to flow when subjected to
an electric field. While one can characterize piezoelectric materials discussed in the
preceding section as controllable solids through the application of an electric field, one
can consider ER materials as controllable fluids through the same means. The potential
of ER fluids in applications as control devices was first recognized by Winslow (1949)
because of their ability to provide simple, efficient, and rapid-response to electronic-
mechanical interfaces. Until recently, the most common ER fluids consisted of
nonducting particles suspended in nonconducting liquids with significant amounts of an
activator, such as water, adsorbed onto the particles (Block and Kelly, 1988). Referred
to as ''wet" or hydrous ER fluids, these fluids were limited in applications due to their
inherent need of adsorbed water, causing possible instabilities. The discovery of
anhydrous ER materials in the late 1980s, however, alleviated some of these concerns
and led to a significant increase in ER fluid research (Filisko and Radzilowski, 1990)
with attendant development of ER devices. Early ER devices include clutches and
brakes (Winslow, 1949) and valves (Phillips, 1969). More recent examples of ER
devices are engine mounts (Duclos, 1987; Petek et al., 1988), shock absorbers (Duclos,
1988), robotic devices (Gandhi et al., 1989), and structural vibration dampers (Stevens
et al., 1984; Morishita and Mitsui, 1992; Morishita and Ura, 1993; Carlson et al., 1995;
Makris et al., 1996). In most of these applications, the ER fluid is used either to adjust a
pressure gradient (flow mode) or to provide momentum transfer (shear mode).
Examples of flow mode devices include shock absorbers and dampers. Brakes and
clutches are examples of shear mode devices.
9.4.1 Basic Principles
The exact causes of the ER effect as described above are not completely understood,
but it has been attributed to induced polarization of the dispersed particles of a colloid
or the dissolved phase of a solution (Block and Kelly, 1988). This polarization process
is manifested when yield stresses are developed in the fluid, which are strongly
dependent on the electrical field. Fig. 9.10a illustrates qualitatively the stress-strain
behavior, giving ultimate values of yield stress y and yield strain y (Gavin and
Hanson, 1994). At pre-yield strains, ER materials are basically linearly viscoelastic.
Yield strains typically decrease with the electric field and yield stresses increase
roughly as squares of the electric field.
Figure 9.10
Stress-strain Behavior of ER Materials (Gavin and Hanson, 1994)
The behavior of ER fluids under steady-state flow as described above motivated the use
of the Bingham constitutive model for ER fluids (Stanway et al., 1985,1987; Gavin and
Hanson, 1994), i.e.,
where is the viscosity of the fluid. The exponent of power laws relating the yield
stress y to the electrical field E ranges from 1.2 to 2.5 for typical ER fluids. Typical
values of y range from 2 to 5 kPa for electric fields on the order ot 3 to 5 kV/mm at
25C. Other properties of ER fluids will be discussed in the next section when they are
compared with magnetorheological fluids.
Based on the electrorheological model as described by Eq. (9.10), the hysteretic model
relating the force generated by a typical ER damper, F, to the piston velocity, , is given
by (Stanway et al., 1985, 1987).
where co is the damping coefficient, fc is the frictional force related to the fluid yield
stress, and fo is an offset in F to account for the non-zero mean observed in the
measured force due to the presence of the accumulator. Thus, the Bingham model
consists of a Coulomb friction element placed in parallel with a viscous element as
shown in Fig. 9.11.
Figure 9.11
Bingham Model for ER Fluids
(Stanway et al., 1985,1987)
Several other hysteretic models have also been proposed. Gamota and Filisko (1991),
for example, proposed an extension of the Bingham model, consisting of the Bingham
model in series with a linear solid model. Another, proposed by Spencer et al. (1996a),
makes use of the versatility of the Bouc-Wen model (Wen, 1976) in its ability to exhibit
a wide variety of hysteretic behavior. It is schematically shown in Fig. 9.12. The force
predicted by this model is given by
In the above, , , A and n are the Bouc-Wen model parameters whose values control
the shape of the hysteresis loops for the yielding element.
Several small scale ER devices (Ehrgott and Masri, 1992; Gavin and Hanson, 1994;
Makris et al., 1996) and one large scale ER device (Gavin
Figure 9.12
Mechanical Model with Bouc-Wen
Element (Spencer et al., 1996a, b)
and Hanson, 1994) were designed, constructed, and tested to explore their structural
control potential. For the ER device tested by Gavin and Hanson (1994) as shown in
Fig. 9.13, Fig. 9.14 illustrates the range of the hysteresis loops obtained from constant
E tests under a sinusoidal input. The sinusoidal period in this case is 1.2 sec and the gap
between the device plunger and the box is 0.96 mm. Figure 9.15 shows the ER damper
tested by Makris et al. (1996), which consists of a main cylinder and a piston rod that
pushes an ER fluid through a stationary annular duct. The proposed damper is compact
and can potentially produce relatively large forces.
Figure 9.13
An ER Test Device
(Gavin and Hanson, 1994)
Figure 9.14
Experimental ER Hysteresis Loops at Constant
Voltages [E = 0.0 kV/mm () and E =
2.50 kV/mm (- - -)] (Gavin and Hanson, 1994)
Figure 9.15
ER Damper with Annular Duct (Makris et al., 1996)
Figure 9.17
Calibration Test of Digital and High Voltage Response to
Acceleration Signal; 5-Hz Steady State Acceleration Signal
(), 5-Volt Digital Control Pulse and the Power
Supply High Voltage Response in kV (- - -) (Ehrgott
and Masri, 1992)
Figure 9.18
Structure Displacement Response; ER Damper Off (),
ER Damper On (- - -) (Ehrgott and Masri, 1992)
Table 9.1 Comparison of Properties of Typical ER and MR Fluids (Carlson, et al., 1995)
Property ER Fluid MR Fluid
Yield Strength 2-5 kPa 50-100kPa
(Field) (3-5 kV/mm) (150-250 kA/m)
Field limited by breakdown Field limited by saturation
Viscosity 0.2-0.3 Pa-s 0.2-0.3 Pa-s
(no field) @ 25C @ 25C
Operating +10 to + 90C (ionic, DC) - 40 to + 150C
Temperature -25 to + 125C (non-ionic, AC) (limited by carrier fluid)
Current Density 2-15 mA/cm2 (4kV/mm, 25C) can energize with
(x 10 -x100 @ 90C) permanent magnets
Specific Gravity 1-2.5 3-4
Ancillary Materials Any (conductive surfaces) Iron/Steel
Figure 9.19
Schematic of an MR Damper (Spencer et al., 1996a, b)
Figure 9.20
Experimentally Measured Force for 2.5 Hz Sinusoidal
Excitation with 1.5 cm Amplitude (Spencer et al., 1996a, b)
Figure 9.21
Comparison Between Predicted and Experimentally Obtained
Responses for Proposed Model (Spencer et al., 1996a, b)
Simulation studies using simple structural models have been carried out which
demonstrate that MR devices can be effective in reducing structural responses under
simulated earthquake inputs. These results and their implications to full-scale structural
control using MR dampers await experimental verification and further research and
development.
References
Agnes, G. S. and Napolitano, K. (1993), Active Constrained Layer Viscoelastic Damping,
Proc. 34th SDM Conf., Lajolla, CA, 3499-3506.
Aiken, I.D., Nims, D. K. and Kelly, J. M. (1992), Comparative Study of Four Passive
Energy Dissipation Systems, Bull. N. Z. Nat. Soc. for Earthquake Engrg., 25(3), 175-186.
Bailey, T. and Hubbard, J. E. (1985), Distributed Piezoelectric-polymer Active Vibration
Control of a Cantilever Beam, J. Guidance, Control and Dynamics, 8(5), 605-611.
Baz, A. (1993), Active Constrained Layer Damping, Proc. Damping 93, San Francisco,
CA.
Baz, A. and Poh, S. (1988), Performance of an Active Control System with Piezoelectric
Actuators, J. Sound and Vibration, 126(2), 327-343.
Baz, A. and Ro, J. (1993), Partial Treatment of Flexible Beams with Active Constrained
Layer Damping Recent Developments in Stability, Vibration and Control of Structural
Systems, ASME/AMD, 61-80.
Block, H. and Kelly, J.P. (1988), Electrorheology, J. Applied Physics, 21, 1661-1677.
Burke, S. E. and Hubbard, J. E. (1987), Active Vibration Control of a Simply Supported
Beam using a Spatially Distributed Actuator, IEEE Control System Magazine, 8, 25-30.
Burke, S. E. and Hubbard, J. E. (1988), Distributed Actuator Control Design for Flexible
Beams, Automatica, 24(5), 619-627.
Cady, W. G. (1964), Piezoelectricity, Dover, New York.
Carlson, J. D., Catanzarite, D. M. and St. Clair, K. A. (1995), Commercial Magneto-
Rheological Fluid Devices, Proceedings of the 5th International Conference on ER Fluids,
MR Fluids and Associated Technology, U. Sheffield, UK.
Chandrashekhara, K. and Agarwal, A. (1993), Active Vibration Control of Laminated
Composite Plates using Piezoelectric Devices - A Finite Element Approach, J. Intel. Mat.
Sys. and Struct., 4(3), 496-508.
Chattopadhyay, A. and Seeley, C. S. (1994), A Multiobjective Design Optimization
Procedure for Control of Structures using Piezoelectric Materials, J. Intel. Mat. Sys. and
Struct., 5(3), 403-409.
Crawley, E. F. and Anderson, E. H. (1990), Detailed Models of Piezoelectric Actuation of
Beams, J. Intel. Mat. Sys. and Struct., 1(1), 4-25.
Crawley, E. F. and Luis, J. (1987), Use of Piezoelectric Actuators as Elements of
Intelligent Structures, AIAA J., 25(10), 1373-1385.
Devasia, S., Meressi, T., Paden, B. and Bayo, E. (1993), Piezoelectric Actuator Design for
Vibration Suppression: Placement and Sizing, J. Guidance, Control and Dynamics, 16(5),
859-864.
Dosch, J. J., Inman, D. J. and Garcia, E. (1992), A Self-sensing Piezoelectric Actuator for
Collocated Control, J. Intel. Mat. Sys. and Struct., 3(1), 166-185.
Duclos, T. G. (1987), An Externally Tunable Hydraulic Mount which Uses Electro-
Rheological Fluid, SAE Paper 870963, 121-137.
Duclos, T. G. (1988), Electrorheological Fluids and Devices, Automotive Engineering,
96(12), 45-48.
Duerig, T. W., Melton, K. N., Stockel, D. and Wayman, C. M. (1990), Engineering
Aspects of Shape Memory Alloys, Butterworth-Heinemann, London.
Dyke, S. J., Spencer, B. F., Jr., Sain, M. K. and Carlson, J. D. (1996), Seismic Response
Reduction using Magnetorheological Dampers, Proc of the IFAC World Congress, San
Francisco, CA.
Edberg, D. L., Bicos, A. S., Fuller, C. M., Tracey, J. J. and Fecher, J. S. (1992),
Theoretical and Experimental Studies of a Truss Incorporating Active Members, J. Intel.
Mat. Sys. and Struct., 3(2), 333-347.
Ehrgott, R. C. and Masri, S.F. (1992), Use of Electro-rheological Materials in Intelligent
Systems, Proc. US/Italy/Japan Workshop on Structural Control and Intelligent Systems,
Naples, Italy, 87-100.
Filisko, F. E. and Radzilowski, L. H. (1990), An Intrinsic Mechanism for the Activity of
Alumino-Silicate Based Electrorheological Materials, J. Rheology, 34(4), 539-552.
Gamota, D. R. and Filisko, F. E. (1991), Dynamic Mechanical Studies of
Electrorheological Materials: Moderate Frequencies, J. Rheology, 35, 399-425.
Gandhi, M. V., Thompson, B. S. and Choi, S. B. (1989), A New Generation of Innovative
Ultra-Advanced Intelligent Composite Materials Featuring Electro-Rheological Fluids: An
Experimental Investigation, J. Composite Materials, 23, 1232-1255.
Garcia, E., Dosch, J. J. and Inman, D. J. (1992), The Application of Smart Structures to
the Vibration Suppression Problem, J. Intel. Mat. Sys. and Struct., 3(4), 659-667.
Gavin, H. P. and Hanson, R. D. (1994), Electrorheological Dampers for Structural
Vibration Suppression, Report No. UMCEE 94-35, Department of Civil and
Environmental Engineering, University of Michigan, Ann Arbor, MI.
Graesser, E. J. and Cozzarelli, F. A. (1991), Shape-memory Alloys as New Materials for
Aseismic Isolation, ASCE J. of Engineering Mech., 117(11), 2590-2608.
Graesser, E. J. and Cozzarelli, F. A. (1994), A Proposed Three-dimensional Constitutive
Model for Shape Memory Alloys, J. Intelligent Material Systems and Struct., 5(1), 78-89.
Ha, S. K., Keilers, C. and Chang, F. K. (1992), Finite Element Analysis of Composite
Structures Containing Distributed Piezoelectric Sensors and Actuators, AIAA J., 30(3),
772-780.
Hanagud, S., Obal, M. W. and Calise, A. J. (1992), Optimal Vibration Control by the Use
of Piezoceramic Sensors and Actuators, J. Guidance, Control and Dynamics, 15(5), 1199-
1206.
Ikeda, T. (1990), Fundamentals of Piezoelectricity, Oxford University Press, New York.
Inaudi, J. A., Kelly, J. M., Taniwangsa, W. and Krumme, R. (1993), Analytical and
Experimental Study of a Mass Damper using Shape Memory Alloys, Proc. Damping '93,
San Francisco, CA, Vol. 3.
Lee, C. K. and Moon, F. C. (1989), Laminated Piezoelectric Plates for Torsion and
Bending Sensors and Actuators, J. Acoust. Soc. Am., 85, 2432-2439.
Lee, C. K. and Moon, F. C. (1990), Modal Sensors/Actuators, ASME J. Applied
Mechanics, 57(2), 434-441.
Main, J. A., Garcia, E. and Howard, D. (1994), Optimal Placement and Sizing of Paired
Piezoactuators in Beams and Plates, Smart Materials and Structures, 3(2), 373-381.
Makris, N., Burton, S., Hill, D. and Jordan, M. (1996), Analysis and Design of an
Electrorheological Damper for Seismic Protection of Structures, J. Engineering
Mechanics, ASCE, in press.
Morishita, S. and Mitsui, J. (1992), Controllable Squeeze Film Damper (An Application of
Electro-Rheological Fluid), Trans. ASME, J. Vibration and Acoustics, 114, 354-357.
Morishita, S. and Ura, T. (1993), ER Fluid Applications of Vibration Control Devices and
an Adaptive Neural-Net Controller, J. Intelligent Material Systems and Structures, 4.
lander, A. (1932), An Electrochemical Investigation of Solid Cadmium-Gold Alloys, J.
Am. Chem. Soc., 54, 3819-3833.
Petek, N. K., Goudie, R. J. and Boyle, F. P. (1988), Actively Controlled Damping in
Rheological Fluid-filled Engine Mounts, SAE Paper 881785, Soc. Auto. Eng., Detroit, MI.
Phillips, R. W. (1969), Engineering Applications of Fluids with Variable Yield Stress,
Ph.D. Dissertation, University of California, Berkeley, CA.
Preumont, A., Dufour, J.P. and Malekian, C. (1992), Active Damping by a Local Force
Feedback with Piezoelectric Actuators, J. Guidance, Control and Dynamics, 15(2), 390-
395.
Rabinow, J. (1948), The Magnetic Fluid Clutch, AIEE Transactions, 67, 1308-1315.
Shen, I. Y. (1993), Intelligent Constrained Layers: An Innovative Approach, Intelligent
Structures, Materials and Vibrations, ASME/DE, 75-82.
Shen, I. Y. (1994), Bending-vibration Control of Composite and Isotropic Plates Through
Intelligent Constrained-layer Treatments, Smart Materials and Structures, 3(1), 59-70.
Shibata, S., Morino, Y., Shibayama, Y. and Sekine, K. (1992), Adaptive Control of Space
Truss Structures by Piezoelectric Actuator, J. Intel. Mat. Sys. and Struct., 3(4), 697-718.
Spencer, B.F., Jr. Dyke, S. J., Sain, M. K. and Carlson, J. D. (1996a), Idealized Model of a
Magnetorheological Damper, Analysis and Computation, (F. K. Cheng, ed), ASCE, New
York, NY, 361-370.
Spencer, B.F., Jr., Dyke, S. J., Sain, M. K. and Carlson, J. D. (1996b), Phenomenological
Model of a Magnetorheological Damper, Journal of Engineering Mechanics, ASCE, in
press.
Stanway, R., Sproston, J. L. and Stevens, N. G. (1985), Non-linear Identification of an
Electrorheological Vibration Damper, IFA C Identification and System Parameter
Estimation, 195-200.
*Exact value
To convert from To Multiply by
Velocity
foot/minute (ft/min) metre/second (m/sec) 5.0810-3*
foot/second (ft/sec) metre/second (m/sec) 3.04810-1*
Viscosity
foot2/second (ft2/sec) metre2/second (m2/sec) 9.290310-2
pound-mass/foot-second
(Ibm/ft-sec) pascal-second (Pa-sec) 1.4882
pound-force-second/foot2
(lbf-sec/ft2) pascal-second (Pa-sec) 4.788 10
*Exact value
Author Index
A
Abe, M., 237, 270
Agarwal, A., 330
Agnes, G. S., 330
Aguirre, M., 60
Aiken, I.D., 83, 85, 99, 112, 113, 114, 123, 144, 322, 323, 324
Aizawa, S., 271
Alfriend, K. T., 282
Anagnostides, G., 92
Anderson, E., 21
Anderson, E. H., 329, 330
Arima, F., 171, 186, 187, 188, 211, 212
Ashour, S. A., 144, 159
Austin, M. A., 118
Ayorinde, E. O., 228
B
Babuska *, I., 19
Bagley, R. L., 134, 177
Bailey, T., 327, 328
Baktash, P., 109
Balendra, T., 302
Banerjee, P. K., 182
Batchelor, G. K., 175
Bathe, K-J., 12, 25, 33
Bauer, H.F., 282, 283
Baz, A., 330
Berg, G. V., 30
Bergman, D. M., 61, 62, 63, 65, 144
Bertero, V.V., 27, 28, 30, 31
Bhatti, M. A., 36, 59
Bird, R. B., 175, 187
Bishop, R. E. D., 228
Block, H., 332
Boller, Chr., 49, 56
Bolotin, V.V., 33
Bowden, F. P., 86
Brock, J. E., 233, 238
Burke, S. E., 328
C
Cady, W. G., 327
Caldwell, D. B., 127
Campbell, G. W., 241, 242, 247
Carlson, J. D., 332, 338, 339
Carrier, G. F., 282
Carter, W. J., 269
Chaiseri, P., 299
Chandrashekhara, K., 330
Chang, K. C., 47, 131, 143, 144, 152, 156
Chattopadhyay, A., 330
Cherry, S., 94, 95, 107, 108, 110, 111, 112, 116, 117, 118
Chiarugi, A., 219
Chowdhury, A.H., 248
Civi, A., 219
Clark, A. J., 270
Clough, R. W., 22, 33
Cofie, N. G., 44, 55
Constantinou, M. C., 2, 134, 171, 172, 173, 177, 185, 193, 194, 195, 197, 198,
199, 200, 201, 202, 203, 204, 205, 206, 207, 218, 220
Cozzarelli, F. A., 37, 47, 321, 322, 324, 325, 326
D
Dafalias, Y. F., 43
Dargush, G. F., 56
de la Llera, J.C., 59
Delinic, K., 171
Den Hartog, J.P., 227, 228, 231, 237, 238, 242, 282
Devasia, S., 330
Dosch, J. J., 330
Dowling, N.E., 49
Duclos, T. G., 332
Duerig, T. W., 320
Dyke, S. J., 338
Dym, C. L., 175
E
Edberg, D. L., 330
Ehrgott, R. C., 334, 336, 337, 338
F
Falcon, K. C., 228, 237
Ferry, J. D., 135, 138, 178
Filiatrault, A., 94, 95, 107, 108, 109, 110, 111, 112, 113, 116, 117, 118
Filisko, F. E., 332, 334
FitzGerald, T. F., 84, 85, 98, 99
Foutch, D. A., 144, 152
Frahm, H., 227
Frohrib, D. A., 228
Fujino, Y., 237, 270, 282, 294, 295, 296, 297, 300, 303, 304
Fujita, S., 144
Fung, Y. C., 37
G
Gadre, M., 329
Gamota, D. R., 334
Gandhi, M. V., 332
Garcia, E., 330
Gavin, H. P., 332, 333, 334, 335
Gehling, R.N., 127
Gemant, A., 134, 177
Goel, S.C., 61, 62, 63, 65
Graesser, E. J., 47, 321, 322
Greenbank, L. R., 60, 61
Grenier, A., 219
Grigorian, C. E., 84, 98, 99
Grigoriu, M., 33, 241
Grossman, J. S., 257
Guendeman-Israel, R., 119
H
Ha, S. K., 330
Hall, W. J., 30, 31
Hanagud, S., 330
Hanson, R. D., 6, 59, 69, 144, 159, 332, 333, 334, 335
Hargreaves, A. C., 92
Harnpornchai, N., 270
Harris, C. M., 171
Hayama, S., 290, 291, 292
Hertzberg, R. W., 49, 50
Higashino, M., 271
Hirt, C. W., 292
Hoppe, D. L., 237
Housner, G. W., 26, 30, 285
Hrovat, D., 271
Hubbard, J. E., 327, 328
Huerta, A., 292
Huffmann, G. K., 171, 202, 220
Hunt, J. B., 270
I
Igusa, T., 270
Ikeda, K., 228, 233, 237, 238, 240
Inaudi, J. A., 59, 110, 324
Ioi, T., 228, 233, 237, 238, 240
Isyumov, N., 249, 250
J
Jacquot, R. G., 237
Jara, J. M., 59
Jennige, R. L., 228
Johnson, C.D., 143
Johnson, K. L., 89
K
Kanaan, A. E., 59, 104, 147
L
Lai, M. L., 141, 143, 144
Landau, L. D., 175
Larson-Basse, J., 86
Lee, C. K., 329
Lee, G. C., 47
Lee, H. H., 134
Lee, J., 239
Lepelletier, T. G., 286, 287, 288, 289, 290, 291, 293
Li, C., 187, 189, 190, 212, 213, 214, 218, 219
Lifshitz, E. M., 175
Lin, F. C., 269
Lin, R. C., 144
Liu, W. K., 292
Lobo, R. F., 29, 144, 152
Luft, R. W., 237, 239, 256
Luis, J., 329
M
MacDonald, P.A., 259
Mahmoodi, P., 127, 161, 162
Main, J. A., 330
Mak, C. Y., 243, 244, 245, 250, 253, 254
Makris, N., 134, 171, 172, 177, 178, 180, 181, 182, 183, 184, 185, 332, 334, 335,
336
Marsh, C., 83, 84, 93, 94, 105, 106, 109
Martinez-Romero, E., 71, 73, 74, 75, 76
Masri, S.F., 334, 336, 337, 338
Mataki, Y., 263, 264
Maugin, G., 50
McNamara, R. J., 250, 251, 252
Mei, C.C., 286
Mendelson, A., 37
Miles, J. W., 282, 287
Miller, A. K., 45
Mitsui, J., 332
Mitsusaka, Y., 172, 187, 223, 224
Miyazaki, M., 172, 187, 223, 224
Modi, V. J., 282, 284, 298, 304, 305
Monti, M.D., 60
Moon, F. C., 329
Morgenthaler, D. R., 127
Morishita, S., 332
Mrz, Z., 43
N
Napolitano, K., 330
Newmark, N.M., 6, 30
Nichols, B. D., 292
Nims, D. K., 84, 85, 100, 101, 109, 110
Niwa, N., 198
O
Ohtake, K., 263
Okamoto, T., 292
lander, A., 319
Oldham, K. B., 178
Ormondroyd, J., 228
zdemir, H., 36, 45, 50, 51, 57, 59, 321, 322
P
Pall, A. S., 83, 84, 90, 91, 92, 93, 94, 104, 105, 106, 119, 120, 121, 122
Pall, R., 119, 120, 122
Pasquin, C., 122
Pekcan, G., 173, 174, 196, 197, 215, 216, 217
Penzien, J., 22, 33
Perry, C. L., 76, 77, 78
Petek., N. K., 332
Petersen, N. R., 261, 262
Phillips, R. W., 332
Pipes, L. A., 228
Pister, K. S., 118
Poh, S., 330
Popov, E. P., 43
Powell, G. H., 59, 104, 119, 147
Press, W. H., 12, 46, 201
Preumont, A., 330
R
Rabinow, J., 338
Radzilowski, L. H., 332
Raichlen, F., 286, 287, 288, 289, 290, 291, 293
Randall, S. E., 228, 233
Reinhorn, A.M., 187, 189, 190, 208, 209, 210, 211, 212, 213, 214, 215, 218, 219
Richter, P. J., 101, 114, 115
Rivlin, R. S., 47
Ro, J., 330
Roberson, R. E., 228, 269
Robinson, W. H., 60, 61
Roik, K., 96, 97, 98
Rosen, S. L., 135
Rosenblueth, E., 6
Ross, D., 127
S
Sakai, F., 299
Snchez, A. R., 60
Saoka, Y., 298
Scholl, R. E., 70, 104
Schwahn, K. J., 171
Seeger, T., 49, 56
Seeley, C. S., 330
Setareh, M., 270
Shames, I. H., 37, 175
Shen, I. Y., 330
Shen, K. L., 136, 144
Shibata, S., 330
Shimizu, T., 290, 291, 292
Shinozuka, M., 33
Shreir, L. L., 90
Skinner, R. I., 2, 35, 59, 71, 72
Sladek, J. R., 242, 243
Snowdon, J.C., 228, 270
Soong, T. T., 2, 33, 56, 59, 136, 142, 143, 159, 241, 271
Spanier, J., 178
Spencer, B.F., Jr., 334, 338, 339, 340
Srinivasan, A. V., 228
Stanway, R., 333
Stevens, N. G., 332
Su, T. C., 292
Su, Y-F., 144
Sun, L. M., 292, 293, 294, 295, 296, 297, 300, 301, 302, 303
Sun, W-J., 302
Symans, M.D., 173, 193, 194, 195, 200, 201, 202, 203, 204, 205, 206, 207, 218
Szab, B., 19
T
Tabor, D., 86
Tamura, Y., 306, 307, 308, 309, 310, 311, 312, 313, 314, 315
Tanaka, H., 243, 244, 245, 250, 253, 254
Taylor, D. P., 220
Taylor, R. L., 18, 25, 44
Tezcan, S., 219
Thomaides, S. S., 30
Thompson, A. G., 233, 237
Torvik, P. J., 134, 177
Tsai, C. S., 56, 134
Tsai, K. C., 36, 55, 56, 59, 67, 68, 69, 70, 71
Tseng, C.I., 330
Tsopelas, P., 197, 198, 199, 202
Tyler, R. G., 92
Tzou, H. S., 327, 329, 330
U
Uang, C. M., 27, 28, 30, 31
Ueda, T., 264, 265, 266, 267, 268, 314
Ungar, E. E., 143
Ura, T., 332
V
Valanis, K. C., 47
Veletsos, A. S., 30
Vezina, S., 120
W
Wakahara, T., 311
Warburton, G. B., 228, 233, 234, 235, 236, 237, 241
Welbourn, D. B., 228
Welt, F., 282, 298, 304, 305
Wen, Y. K., 334
Whittaker, A., 63, 64, 65, 67, 68, 70
Wiesner, K. B., 241, 242, 255
Williams, M. L., 135
Winslow, W. M., 332
Wirsching, P. H., 241, 242, 247, 248
Witting, P. R., 324, 325, 326
X
Xia, C., 59, 69
Xu, K., 270
Xu, Y. L., 249, 255, 283, 284, 302
Y
Yamaguchi, H., 270
Yao, J. T. P., 248
Z
Zahrah, T. F., 30, 31
Zhang, R. H., 128, 142, 159
Zienkiewicz, O. C., 18, 25, 44
Subject Index
A
ABAQUS, 19, 54, 200
Active Control, 2
ADAS Damper, 63, 71, 72, 75, 76
ADINA, 19, 200
Ambient temperature, 138
ANSYS, 19
Asahi Beer Azumabashi Building, 123
Austenite, 320
B
Base isolation, 2, 123, 202, 219, 222
Bauschinger effect, 41, 43
Bernoulli's equation, 191
Bimetallic corrosion, 90
Bimetallic interface, 90, 98
Bingham constitutive model, 333
Boltzmann superposition principle, 135
Boundary element method, 182
Brake lining materials, 92
C
CAARC Building, 253
Canadian Space Agency Complex, 120
Cardiology Hospital Building, 72
Casino de Montreal, 122
Centerpoint Tower, 259
Chiba Port Tower, 263
Chile (Llolleo) 1985 earthquake, 65
Citicorp Center, 255, 257, 260
CN Tower, 259
Coefficient of friction, 87
Coffin-Manson relationship, 49, 56
Columbia SeaFirst Building, 162
Complex frequency response function, 9
Complex modulus, 130, 136, 177
Complex shear modulus, 130, 136, 177
Corrosion, 90
Coulomb friction, 86
Cross-braced friction damper, 83, 93, 94, 110, 116
Crystal Tower, 259
Cylindrical pot damper, 171, 181, 219
D
Damping ratio, 231
Damping ratio, 7, 202
Direct time domain analysis, 24
Dispersion, 287, 291
Doubly-tuned mass damper, 270
DRAIN, 19, 59, 69, 72, 75, 76, 104, 105, 107, 119, 121, 147, 200
Dry friction, 86
Dynamic amplification factor, 9, 10, 231
Dynamic stiffness, 201
E
Ecole Polyvalante, 122
Eigenvalue problem, 20, 202
El Centro 1940 earthquake, 12, 65, 68, 104, 105, 107, 109, 112, 113, 114, 147,
150, 203, 208, 212, 295, 325, 331
Elastic modulus, 38, 129
Elastoplastic model, 15
Electrorheological (ER) fluid, 332
Energy Dissipating Restraint, 84, 100, 114
Energy formulation, 25
ETABS, 19
Euler equations, 176, 191
F
Fatigue, 49, 59
Finite difference method, 291, 292
Finite element method, 19, 33, 54, 56, 57, 287, 292
Flow rule, 43
Forced frequency ratio, 9, 231
Fractional derivative, 134, 177
Frahm's absorber, 227, 282
Frequency ratio, 9, 231
Friction damper models,
elastic perfectly plastic, 92
with bearing stops, 93, 96
Friction damper, 83
Funade Bridge Tower, 264
G
GERB Fluid Damper, 171, 181, 219
Glassy modulus, 135
H
Hachinohe earthquake, 145, 147, 150
Haneda Airport Control Tower, 313
Hankyu Chayamachi Building, 273
Hardening, 41
Hybrid mass damper, 270
I
IDARC, 19, 208
Ikuchi Bridge 'rower, 314
Impact vibration absorber, 269
Impulse response function, 11, 12
Inelastic strain, 38
Intelligent constrained layer (ICL), 330
Internal temperature, 141
Isotropic hardening, 41
Izazaga #38-40 Building, 72
J
Jarret Elastomeric Spring Damper, 173, 196, 215
John Hancock Tower, 260
K
Kelvin-Voight model, 134
Kinematic hardening, 41
L
Lanchester damper, 238
LAPACK, 21
Lead extrusion damper, 60
Limited slip bolted joint, 83, 91, 104
Loma Prieta 1989 earthquake, 76
Loss factor, 130
Loss modulus, 130, 178
Loss stiffness, 182, 187
M
Mach number, 192
Magnetorheological (MR) fluid, 338
Martensite, 320
Mass ratio, 229
Maxwell model,
classical, 45, 134, 193, 198, 200, 202, 208
complex derivative, 177
fractional derivative, 177, 185
McConnel Library, 119
Metallic dampers, 35
Mexican Institute of Social Security, 74
Mexico City 1957 earthquake, 75
Mexico City 1985 earthquakes, 72, 75, 112
Miyagi-Ken-Oki 1978 earthquake, 27
Modal damping ratio, 143
Modal strain energy, 143
Modal superposition method, 20
Mode shapes, 20
MSC/NASTRAN, 19
Multi degree of freedom (MDOF)
model,
analysis, 58, 143, 200, 245
definition, 18
energy formulation, 31
N
Nagasaki Airport Tower, 306
Natural frequency,
damped, 7
tuned liquid damper, 287, 299
undamped, 7, 20
with Maxwell elements, 202
Natural period, 7, 57
Naval Supply Facility, 166
Navier-Stokes equations, 177, 190
Newmark algorithm, 24
Newton-Raphson method, 25, 58
Newtonian fluid, 176
Nitonol, 322
Non-Newtonian fluid, 177
Nutation damper, 283
O
Orificed fluid damper, 173, 190, 192, 199, 202, 208, 215, 219
Orthogonality conditions, 21
Ozdemir model, 45, 50
P
Pacific Bell North Area Operations Center, 220
Pall Friction Damper, 119, 120, 122
Palmgren-Minor rule, 49
Parkfield 1966 earthquake, 107
Passive Energy Dissipation, 2
Piezoelectric effect,
converse, 327
direct, 327
Piezoelectric material, 327
Plasticity models,
elastic perfectly plastic, 15, 41, 51, 92
Ramberg-Osgood, 41, 51
Two-surface, 43
Plasticity theory, 37
Progressive waves, 286
Ramberg-Osgood, 40, 51
Rangitikei Bridge, 71
Rayleigh damping, 21
Re-centering, 102, 197, 320
Reduced variables, method of, 138, 141, 178
Reforma #476 Buildings, 74
Relaxation time, 135, 179, 193
Reynolds number, 192
Rich Stadium, 220
Rubbery modulus, 135
Runge-Kutta formula, 46
S
SAP, 76, 200
Saguenay (Quebec) 1988 earthquake, 122, 325
San Bernardino County Medical Center, 222
Santa Clara County Building, 163
Seismic Isolation, 2, 123, 202, 219, 222
Self-centering, 102, 197, 320
Semi-active Control, 2
Sendagaya INTES Building, 271
Shallow water gravity waves, 285
Shape memory alloy, 319
Shape memory effect (SME), 319
Shear loss modulus, 130, 131
Shear storage modulus, 130, 131
Shin-Yokohama Prince Hotel, 311
Single degree of freedom (SDOF)
model,
definition, 6
energy formulation, 26
forced vibration, 8
free vibration, 7
transient response, 11
with passive damper, 14, 57
Sloshing damper, 284, 292, 302, 306, 308, 311, 313, 314
Slotted bolted connection, 84, 98
Smart material, 319
Solitary wave, 288
T
TADAS, 67, 69
Taft 1952 earthquake, 29, 111, 152, 203, 210, 214, 216
Taylor Devices Fluid Damper, 173, 192, 220, 221, 222
Time domain analysis, 24
Triangular plate damper, 36, 51, 67
Tuned liquid column damper, 283, 298
Tuned liquid damper, 281
Tuned mass damper (TMD), 227, 229, 281, 285, 302, 305, 324
Tuned roller pendulum damper (TRD), 258
Two Union Square Building, 162
U
Uniaxial friction damper, 83, 99, 113, 119, 123
Units, conversion, 345
V
Viscoelastic (VE) damper, 127
Viscoplasticity, 44
Viscous Damping Wall, 171, 186, 211, 218, 223
Viscous boundary layer, 293
Viscous fluid damper, 171
W
Water waves, 285
Wave breaking, 295, 296
Wells Fargo Bank Building, 76
Woodland Hotel, 221
World Trade Center, 127, 161
X
X-braced friction damper, 83, 93, 94, 110, 116
X-shaped plate damper, 36, 56, 61, 63, 68, 71, 72, 75
Y
Yield surface, 42
Yokohama Marine Tower, 308
Z
Zacatula earthquake, 109, 114
Zero shear rate viscosity, 179
#
28 State Street, 221
3M Viscoelastic Damper, 127, 161, 162, 163, 166