Use of VE Dampers in Reducing Wind and Earthquake Induced Motion of Buildings-ES
Uploaded by
kostas formulagrUse of VE Dampers in Reducing Wind and Earthquake Induced Motion of Buildings-ES
Uploaded by
kostas formulagrEngineeringStructures,Vol. 17, No. 9, pp.
639 654, 1995
I•UTTERWORTH Elsevier Science Ltd
I'~IE I N E M A N N
Printed in Great Britain
0141-0296(95)00034-8 01414)296/95 $10.00 + 0.00
Use of viscoelastic dampers in
reducing wind- and earthquake-
induced motion of building
structures
B. Samali
Graduate School of Engineering, Universi~ of Technology, Sydney, PO Box 123, Broadway,
NSW 2007, Australia
K. C. S. Kwok
School of Civil and Mining Engineering, Universi~ of Sydney, Sydney, NSW 2006, Australia
Viscoelastic (VE) dampers have been successfully incorporated in
a number of tall buildings as a viable energy dissipating system to
suppress wind- and earthquake-induced motion of building struc-
tures. This type of damper dissipates the building's mechanical
energy by converting it into heat. Several factors such as ambient
temperature and the loading frequency will affect the performance
and hence the effectiveness of the damper system. VE dampers
have been able to increase the overall damping of the structure
significantly, hence improving the overall performance of dynami-
cally sensitive structures. The effectiveness of VE dampers has
been demonstrated both experimentally and analytically by many
researchers over the past 25 years. The twin towers of the World
Trade Center Buildings in New York City and the Columbia SeaFirst
Building in Seattle, Washington, are among the first buildings
which benefited from the installation of VE dampers. In seismic
applications, the VE dampers can be incorporated either into new
construction or as a viable candidate for the retrofit of existing
buildings which adds to the versatility of VE dampers.
Keywords:viscoelastic damper, viscoelastic material, tall building,
wind-excited motion, earthquake-excited motion
1. Introduction retrofit of existing buildings. For either wind- or earth-
quake-related applications, a successful damper system
Viscoelastic dampers are among the earliest types of pass- involves the installation of viscoelastic dampers at various
ive dampers which have been successfully installed in a points in a structure having relative displacement during
number of tall buildings and other structures to reduce the excitation. In this summary paper the special characteristics
motion amplitude and the acceleration occurring due to of viscoelastic dampers, a summary of the earlier research,
wind and earthquake forces. Properly employed, these dam- and the design and special features of a few successful
pers are capable of reducing the building motion by con- installations in tall buildings and other structures are pre-
verting a portion of the mechanical energy of wind or earth- sented.
quake to heat. In the case of wind-induced motion, the
additional damping provided by the viscoelastic damper is
primarily for the comfort of building occupants and is a
2. Earlier research
viable solution for tall and slender buildings where motion Following the successful installation of viscoelastic dam-
is likely to be noticeable and possibly objectionable. In pers in the World Trade Center buildings in New York
seismic situations, this technology can be incorporated City, the interest and curiosity of the engineering com-
either into new construction or as a viable candidate for the munity led to extensive research programs initiated by
639
640 VE dampers for building motion control: B. Samali and K. C. S. Kwok
many researchers to study the behaviour and effectiveness base for utilizing the extensive data generated from the
of viscoelastic dampers as a viable energy dissipating sys- reduced-scale model tests for incorporating VE dampers
tem for structural applications. These research initiatives into full-scale structures.
which have spanned over a quarter of a century to date, Chang et al. 32 summarized some previous experimental
have made significant contributions to the advancement and and analytical studies on viscoelastic dampers as energy
further application of viscoelastic (VE) materials as struc- dissipation devices for structural applications, and an over-
tural dampers. One of the earliest and more significant view of viscoelastic dampers for wind and seismic appli-
works on VE dampers has been the work of Mahmoodi ~ cations is also given. A comprehensive but simple method-
who studied and established the properties of VE materials ology for the design of viscoelastic dampers in earthquake-
following an extensive experimental program. Mahmoodi resistant structures has been given by Abbas and Kelly 15.
studied the dynamic properties of VE materials, both mech- Inaudi et al. 2° performed a detailed analysis of structures
anical and thermal. His findings are referred to later in with viscoelastic dampers. They addressed modelling
this paper. aspects for the constitutive relations of linear viscoelastic
Earlier research into viscoelastic dampers and their appli- materials and investigated the concept of modal equations
cations focused on the suppression of wind-induced sway for viscoelastic structures. They developed an iterative
of high-rise buildings. Mahmoodi, Keel and others ~ 8 technique to compute the response of viscoelastic struc-
showed that the wind-induced sway of high-rise buildings tures, and described an approximate technique for the
can be significantly reduced by adding viscoelastic dampers analysis of structures with frequency-dependent storage and
to the structure. loss moduli, and evaluated its accuracy using tools with
The feasibility of using viscoelastic dampers to mitigate random vibrations. They studied the energy dissipation in
earthquake-induced structural response has also been stud- linear structures with viscoelastic dampers subjected to
ied by various authors 9-18. They studied the seismic transient excitation; bounds were obtained for the energy
response of steel and concrete frame structures equipped dissipation of structures subjected to transient loading; and
with viscoelastic dampers, and the general application of the mean and standard deviation of energy dissipation of
viscoelastic dampers to seismically resistant structures. The linear structures during a finite-duration segment of station-
study of the behaviour of VE dampers included analytical ary response were computed. They also gave some con-
investigations, some employing finite elements 19, and some sideration to the design of structures with viscoelastic dam-
random vibration techniques 2°, and some employing frac- pers and directions for future research.
tional calculus 2~. Extensive experimental works to assess Finally, design procedures for structures with added vis-
the performance of model structures with added VE dam- coelastic dampers have been developed by Chang et a/. 33"34.
pers under seismic loading has been underway since the The design procedures, which are based on extensive ana-
mid 1980s 22-25. Soong and Lai 26 looked at the correlation lytical and experimental investigations, can easily be incor-
of experimental results with the analytical predictions. porated into a conventional design process. Numerical
Mahmoodi and Keel 27 proposed an analytical approach to simulations on equivalent structural damping and structural
calculate the damped energy for a multilayer viscoelastic response confirm that the dynamic behaviour of structures
damper. They further proposed simple methods for the with added viscoelastic dampers can be satisfactorily pre-
analysis and the design of multilayer VE dampers for tall dicted by conventional linear analytical tools.
structures 28. Chang et a l . 29 studied the effects of ambient
temperature on the performance of viscoelastically damped
3. Damper configuration and damping
structures by conducting a number of experiments on a 2/5-
scale five-storey steel model and a full-sized prototype
mechanism
structure, under a variety of precisely controlled ambient Viscoelastic dampers are non-load-carrying elements and
temperatures and recorded ground motions. The results of are designed such that part of the mechanical energy of the
their study showed that viscoelastic dampers are very effec- building motion is transferred into heat, which results in a
tive in reducing seismic structural response at all levels of reduction of the amplitude of the vibratory motion. The
earthquake ground motions, and that their energy dissi- medium in which this transfer of energy takes place is a
pation capacity decreases as the temperature increases. viscoelastic material.
Tsai 3° who also looked into the temperature effect of visco- There are basically three methods of employing a visco-
elastic dampers during earthquakes reached similar con- elastic material as a damping medium. One is the direct
clusions. Other works in relation to VE materials properties application of a viscoelastic layer to the vibrating part such
includes the work of Kasai et al. 31 who proposed a VE as plates and beams (Figure la) where damping is
damper hysteretic model. accomplished by extensional deformation of the viscoelas-
Research is also underway to look into the dynamic tic layer. The second type (Figure l b ) is an extension of
behaviour of a full-scale viscoelastically damped steel the first, but by adding another layer of a rigid material on
frame. The experimental and analytical studies of the full- top of the viscoelastic part, a constraint layer is formed.
scale five-storey steel frame structure identified dynamic Thus, the viscoelastic material will experience both exten-
scaling relations between the full-scale and 2/5-scale mod- sional and shear deformation. The damping achieved is
els with VE dampers. Test results confirmed that damping mostly due to shear deformation rather than extensional dis-
in the full-scale structure can be significantly increased by placement. There are a number of variations of these types
incorporating relatively small VE dampers. The authors of dampers to increase the deformation and consequently
found that the method used to estimate the structural damp- obtain higher damping capacity. The third type of damper
ing of the 2/5-scale model is also applicable to the full- is the one where nearly all of the deformation is in shear
scale structure, and that the damper design procedure (Figure lc). Each of these configurations has some merits
developed from the 2/5-scale model could also be used for and disadvantages, but generally, for a given amount of
the full-scale structure. Their study provided an important viscoelastic material, the third type is more efficient and is
VE dampers for building motion control: B. Samali and K. C. S. Kwok 641
VISCOELASTICLAYER
[E!!iiiiiiiiii!ii!iiiiiiiiiiiii!!i!iiiiiiii[iiiiiv,BRATORY
!!!iiiii!ii!iii!!PART
Single Layer V.E. Damper
(a)
~NBTRAINTLAYER Io ~ EXTENSION
. LAYER
VIBRATORYPART
Constrained LayerDamper
(b) x~__.,.~J_
___] Fo
VIBRATORYPART
• LAYER Figure3 Typical hysteresis l o o p 7
VIBRATORYPART
i. LAYER ing relations according to the linear theory of viscoelastic-
VIBRATORYPART ity.
Double Sandwich Damper The stress in a VE material under sinusoidal loading is
(e) proportional to the strain and the phase angle & Thus if the
strain has the form
Figure 1 Different VE damper configurations 8
Y = "7o sin wt ( 1)
more suitable where large amounts of energies are to be
the stress can be calculated from
damped out. Hence a typical viscoelastic damper is of the
third type.
o" = ~ro sin(~ot + 6) (2)
A typical structural damper is therefore constructed of
two viscoelastic layers bonded between three parallel rigid
The relationship between stress and strain can be
surfaces (Figure 2). The position of the damper, with
expressed as
respect to application of the load, is such that the viscoelas-
tic material undergoes virtually pure shear deformation. A
o" = G'To sin (~ot + 6)
typical load-deformation curve for a viscoelastic material
is in the form of a hysteresis loop shown in Figure 3. = ~'o (G' sin wt + G" cos wt) (3)
where G* = ( G ' 2 + G"2) 1/2 is the complex shear modulus.
4. Properties of viscoelastic materials The quantities G' = G* cos 6 and G" = G* sin 6 are referred
Viscoelastic damping is quite efficient where high damping to as the storage shear modulus and loss shear modulus,
at low frequency is desired. The mechanical properties of respectively.
viscoelastic materials are rather complex and may vary with If the applied load is of the form F = F~) sin wt, the
environmental temperature and excitation frequency. The relationship between the amplitudes of force, stress and
best method of evaluating the properties of the damper is strain are
to generate the hysteresis loop by subjecting the centre part
of the damper (Figure 2) to a periodic displacement then Fo = ~r(A = 7oG* A (4)
plotting this and the corresponding shear force on an x-y
recorder (Figure 3) for one cycle. The area of the hysteresis The damper's maximum displacement is
loop represents the actual energy lost or damped. This
energy is related to other VE properties, through the follow-. Ao = 7ot (51
where A and t are, respectively, damper area and thickness.
FI2 FI2
Since Fo = kyot, one has
ToG* A = kyot (6)
and therefore G* = kt/A.
The stiffness k can be calculated by dividing the force
magnitude Fo by the maximum displacement. The energy
loss per cycle, W, is proportional to the volume of viscoel-
astic material, V, through the following expression
CENTERPLATE V E MATERIAL w=~a"v (7)
The following relationships also exist between various
parameters
F
Figure2 Typical VE damper 7 G' = W/(~r ~o V) = (0-0/%) sin6 (8)
642 VE dampers for building motion control: B. Samali and K. C. S. Kwok
G' = (~ro/7o) cos6 (9) G" 30.0
tan 6 = G"IG' ( 1O) 201/1 - 5; amb. 22.2 - 22.9"C
Following a series of tests performed by Mahmoodi ~'7,
it became apparent that the material's loss shear modulus, E 26.7
G", and hence energy dissipation per cycle varies inversely ID
with temperature (Figure 4). The relationship seems to be E
approximately linear for the first 100 cycles. After about U')
400 cycles (Figure 5) a steady state of heat loss to the 23.1
I 1 I I
250 500 750 1ooo
surroundings is approached, and it is safe to assume that
G" or the energy loss per cycle remains constant beyond Cycles
this point. The results of Mahmoodi's experiments further Figure 5 Variation of sample temperature with load cycles 7
indicated that only a small percentage of the damped energy
is actually stored in the viscoelastic material (2-4%). This
is a significant result, since the main function of a structural
damper is to transform mechanical energy into heat. Since
the behaviour of any viscoelastic material will be dependent quent property damage. An optimum damper design for a
upon temperature, it is imperative to dissipate the generated given application must take into account the various modes
heat as fast as possible to avoid temperature rise and conse- of heat transfer available to the system.
102
4137 90.4 4137 " ~ -- 90.4
3447 - 79.1 "5 34471'~ 79.1 ~
O
--j O
_¢
2758 67.8 ~
ii1
C
~20c~ 56.5 ~
CA
O O
Sample No. 200/1 - 3
1379 _ _ SampleNo. 200/1 - 1
amb. temp. 24.2 - 25.0~C
45.2 ~ 1379
amb. temp. 24.2~C
45.2 ~
• Cycle No. 1 ® • CycleNo. 1
C C
• Cycle No. 107 • Cycle No. 100
689 • Cycle No. 111 33.9 -"' 689-- • Cycle No. 101 33.9 __.
I / I I
22.2 23.3 24.4 25.6 26.7 27.8 22.2 23.3 24.4 25.6 26.7 27.8
Sample Temperature (°C) Sample Temperature (°C)
4137 90,4
3447 79.t "5
0
2758 '67.8
~" 2068 56.5
o_:
~d
b 45.2 3
1379 -- S a m p l e No. 201/1 - 5
arab. temp. 22.2 - 22.9=C
• C y c l e No. 1
689 • C y c l e No. 1000 33.9
• C y c l e No. 1001
t I 1 t
22.2 23.3 24.4 25.6 26.7 27.8
Sample Temperature (°C)
Figure 4 Variation of loss shear modulus, G" and work done in first cycle versus temperature7
VE dampers for building motion control: B. Samali and K. C. S. Kwok 643
5. Design of viscoelastic dampers motion. Figures 6 and 7 show the variation of VE proper-
ties with temperature, strain and the number of load cycles
Damper design requires certain pertinent data. These data for two different types of VE dampers. The first consider-
are generated, to a large extent, by the calculation and esti- ation is normally spelt out in the design specification and
mation of basic vibrational characteristics of the structure. the temperature variation is the usual design thermal spec-
They include resonance behaviour (frequency and trum. The second requirement, i.e., temperature rise due to
amplitude), mode of vibration and total energy induced by internal generation of heat, is quite important and should
vibration in the structure. The source of vibration is another be studied experimentally. One should also determine the
relevant factor, particularly in relation to the frequency and effect of humidity and other environmental conditions on
duration of the oscillatory forces responsible for the the overall fatigue and aging characteristic of the VE
vibration. The data thus generated enables the designer to material. When all these factors are considered, the selec-
decide on the amount of damping required and conse- tion of a suitable VE material with proper G" will narrow,
quently the proper design, size and number of dampers. and one can then decide on the most appropriate VE
This in turn necessitates the selection of a damper location, material to be used.
where the damper stiffness (defined as force per unit shear To maximize the effectiveness of VE dampers by maxi-
deformation) is compatible with its immediate structural mizing the shear strain ~/o, the designer must place the
members and where there is sufficient relative shear dis- damper in a place in the structure where large relative dis-
placement in the viscoelastic material for dissipation of placement with proper section stiffness is available, or
energy. decrease the thickness of the VE material as the strain is
In designing dampers, several other factors should be related to the thickness of the VE material and to the shear
considered. These include the dynamic behaviour of dam- displacement (equation (5)). In most applications there are
pers under cyclic load and also the temperature rise of the only a few suitable locations for the damper in the structure.
dampers. In some cases, damper stiffness can be a signifi- This would limit the displacement. It is also not beneficial
cant factor when considering their effect on the overall to reduce the thickness of the VE material beyond a certain
stiffness of the structure. limit. Designers should also be aware of the fact that above
The stiffness and damping efficiency of the viscoelastic a certain strain, VE material will no longer behave linearly.
materials are dependent on the loss and complex shear In relation to the volume of the VE material, any volume
moduli, respectively. These properties have an inverse increase for the sake of increasing the damping energy of
exponential functional relationship with temperature. the damper should be considered along with the stiffness
Therefore, it is essential to consider in detail the effect of requirement and the thermal characteristics of the system.
temperature on the overall system. In a structure, the Since in most cases the maximum relative displacement of
damper is an element whose stiffness will affect the overall the damper is dictated by the structure and is generally a
dynamic response of the system. Generally the stiffness of fixed design factor, any attempt to increase the volume of
the damper will affect the shear displacement of the visco- VE material by an increase in thickness, width or shear area
elastic material and since the energy loss is related to dis- will affect the stiffness of the damper. Dissipation of
placement, any loss of stiffness (due to temperature etc.) damped energy in the form of heat through the VE material
should be compensated by an increase in the displacement. to the outside will become a function of the available heat
Some of these factors are now discussed in more detail. conduction area and the overall thermal conductivity of the
In order to increase the damping energy W (according system. This transfer of heat also depends on the thickness
to equation (7)), one must increase either the loss shear and diffusivity of the VE material.
modulus, G", maximum shear strain, 3~o, or the volume of The results presented in Figures 6 and 7 clearly demon-
the VE material. The variation of some of these parameters strate the problems which could arise when the volume of
is rather limited, and the designer has to compromise VE material is increased in a certain way. In the case of
between the optimum condition of each. The shear loss damper A1, Figure 6a shows a high temperature rise due
modulus, G", is an inherent property of VE material and to a 100 cycle motion. Most of this temperature rise occurs
represents to a certain extent the overall stiffness of the within the first few cycles. Although the VE volume of
viscoelastic material. The stiffness of the damper is an damper A2 is half of that of damper A1, the temperature
important factor in design. This is due to the requirement rise (Figure 6b) is quite similar to A1. The temperature
of the stiffness compatibility of the damper with its sur- rise in these dampers is due to the increase of total energy
rounding structural member. If the damper is stiffer than damped and the inability of the VE material to dissipate all
the surrounding, most of the displacement will be taken up the energy to the outside. The dependence of the total
by other members and consequently undesirable stresses energy loss in one cycle on the temperature and the level
will be produced in other parts with very little damping. of strain is shown in Figure 88.
Therefore, G" has a design limitation and since it is gener-
ally a function of frequency and temperature, the behaviour
6. Effectiveness of VE dampers in dissipating
of the damper will be affected by the variation of both.
The selection of VE material with a proper G" within the
seismic energy
design range of temperature, frequency and stiffness is of After the successful application of VE dampers to suppress
paramount importance. Fortunately, in most applications, wind-induced motions, the effectiveness of VE dampers
the design frequency remains relatively constant and hence was also confirmed by a series of shake table tests on model
the selection of VE material will be confined to a narrow as well as full-scale structures. The work of Lin et al. 22 on
frequency range. The effect of temperature should be con- a three-storey model frame as a multi-degree-of-freedom
sidered in two respects, first, the environmental condition (MDOF) system (also as a single-degree-of-freedom
of the structure in which the damper is installed; and second (SDOF) system with proper bracing) using 18 different
the temperature rise of the VE material due to cyclic dampers confirmed the effectiveness of VE dampers in dis-
644 VE dampers for building motion control: B. Samali and K. C. S. Kwok
31.1 F .,d~// 31.1
~' ~ "~ DAMPER A~I
+.oL /Y/,,+" +.o "' I-
1. / / .+.o:'-
-+-'- ++ | / / ~ > + = " ~ , o _+-- ~ 2s+
26.7 26.7
F~/" / , ~' I 1M x 1(pro=Volume tP. 24.4
I /// I.,,o+.,v.,-- " 5,¢. ¢',','-
I-/,'Z I,.="-"- 2s.8 23+ ." I
~/ ~-/
K
~ 11.27mmThick
24.4 24.4 23.3 / ,, ~;o Hz
23 3 l' I I I I I 23.3 r I I I I I 22.8 J I I I l
" 20 40 60 80 100 0 20 40 60 80 100 20 40 60 80 100
No. of Cycles No. of Cycles No. of Cycles
Figure 6 Variation of VE properties with temperature, strain and number of load cycles for damper type A ~
DAMPER B2
3.28 x 10~ m~Volume
27.8 1.27mm Th ck
A 26.7
~26.7
60*/05"TP ' ~
f ~
/ (#'k,~ . . . . ,1 +,.., ¢• 25.6 -
..=
25.6 / ~0"/o~S.~ -'~ .
/ // :~J2~~' ~
; ~d t~."~" " - __ - - - - --_--__
~ 24.4 ~' I l~.7mm Thick
E
24.4
23.3 I I [ [ I 23.3 ~'-'''"
," J l -1 1 I
20 40 60 80 100 20 40 60 80 100
No. of Cycles No. of Cycles
Figure 7 Variation of VE properties with temperature, strain and number of load cycles for damper type B8
DAMPERA2
734
~ X I 1"64x 104m' V°lume
A
= 653
I 1958 -- DAMPERA1
U 569
1632 - - " ~ p . . 13.~ x 10"m' Volume ,5" DAMPERA3
~ 4~ _ ~-\\ 228 +
- ~ ~ ' ..RX X ~ Hz
.c_
so% STFIAIN
! 9 7 9 . - ,~_ "%"~ ¢
o
\\
.¢
~ 326 o
f "-+-,.'+,,- \ \
326 [- '~''0'~%'-..~ ,... "~ ~ 245 ~,.'\ \ ", "' 98
#- / ,e
ol I J I J I | i J "-D, i t 65
24.4 26.1 27.8 82.5 31.1 32.8 25 26.7 283 30.0 31.7 23.3 23.9 24.4 25 25.6 2~.1
Temperature ('C) Temper'alum('C)
Temperature (°C)
Figure 8 Variation of total energy loss in one cycle with temperature and strain8
sipating seismic energy. In the SDOF case, a reduction in carefully studied by the same researchers. The experiments
response as high as 87% for the relative displacement and were carried out at temperatures ranging from 22°C to 35°C
60% for the absolute acceleration were observed. The and it was confirmed that the damper properties and their
results for the MDOF case revealed that, with the most efficiency are strongly temperature dependent. Attention
favourable damper configuration, the average reductions in was also paid to damper configurations. Diagonal place-
structural response of up to 80% for relative displacement, ment of the dampers was the best in the case of the SDOF
70% for storey drifts and about 50% for absolute acceler- system since the relative displacement is the greatest in this
ations are possible. position and so is the energy dissipated by the damper.
The temperature dependency of the VE dampers was Similar considerations were given to experiments involving
VE dampers for building motion control: B. Samali and K. C. S. Kwok 645
MDOF systems and the advantage of optimum damper mode, W~is the energy dissipated in one cycle by the dam-
placement was also demonstrated. They also concluded that pers for the ith vibration mode, and E ~ is the strain energy
while damper additions are accompanied by damping of the structure of the ith vibration mode. The above equ-
increases in the structure, their effectiveness in structural ation can also be expressed in terms of modal strain
response reduction diminishes as damping increases beyond energy as
a certain limit.
Chang et al. 2~ studied the seismic response of a 2/5-scale (14)
steel structure with added viscoelastic dampers. They inves-
tigated the effect of ambient temperature on VE dampers where qbi = ith modal shape vector; Kj = structural stiffness
and on the dynamic response of the model scale five-storey matrix due to the contribution of dampers alone;
steel-frame structure. Based on the test results of individual K, = structural stiffness matrix including the contribution of
VE dampers, empirical formulae on the dynamic damper dampers, and rl, = loss factor for the viscoelastic dampers.
properties as functions of excitation frequency, ambient The total strain energy of the building with a deflection
temperature and strain range were derived. The properties slope /3 is given by 3
of viscoelastic materials to be used in the dynamic tests
were determined by a series of experiments ranging in fre- E = 0.5 1,,,(2/3~,:D2 (15)
quency between 0.1 and 4.0Hz. Three strain levels,
namely, 5%, 20% and 50% along with a temperature range w h e r e f is the natural frequency of the structure and/,,, the
of 21-40°C were studied. Based on the load-deformation mass moment of inertia of the structure. Utilizing equations
relationships obtained from the damper tests, important (7) and (13), one can define the damping ratio in terms of
dynamic parameters of the viscoelastic dampers such as damper and structural properties as
energy dissipation per cycle, damper stiffness, storage and
loss shear moduli etc. under each ambient temperature and ~=~G"V/4E=~G"At/4F
excitation frequency were calculated. For an effective
design of viscoelastic dampers, it is necessary to take into = ~ G " A t/{2 I,,,(2/3r/,{) 2} (16)
account the effect of ambient temperature, excitation fre-
quency, maximum expected strain range, and maximum The damping ratio given by equation (16) is the contri-
expected number of loading cycles, However, these factors bution of a single VE damper to the damping ratio. The
are all interrelated. As expected, G" is dependent upon the corresponding damping ratio for N viscoelastic dampers
ambient temperature and the number of loading cycles. The will be
change in G" can be as much as 30% from cycle one to 20
at 4 Hz and 20% strain. For typical earthquake excitations, = N ~ G " A t/{2 I,,, (2/3T/,~02} (17)
the number of significant loading cycles will generally be
less than 20. It was also observed that the loss modulus The dynamic properties of the test frame equipped with
decreases with increase in temperature. viscoelastic dampers were observed to be very much tem-
In order to consider all factors affecting the properties perature dependent. Figure 9 clearly shows this depen-
of viscoelastic dampers in the design, Chang et al. 23 pro- dency.
posed the following empirical formulae for the damper The effectiveness of VE dampers in controlling response
stiffness and the loss factor based on regression analysis quantities such as maximum floor displacement, maximum
using the data obtained from damper tests storey drift and maximum floor acceleration are shown in
Figures 10 and 11. From these figures one can also observe
Ka = e 26"8~ (/-)o.69 (T)-2,26 (11) the deterioration of the level of response control with tem-
0," = el"s5 ~) 0.27 (T) ii.l~ perature rise.
(12)
The results obtained by Chang et al. 23 show that, in gen-
eral, viscoelastic dampers are very effective in reducing
where excessive vibration of the test structure due to seismic exci-
tations. At 25°C, the dampers were able to achieve a
Kj stiffness of damper (N/m) reduction of about 80% of the maximum response quan-
f vibration frequency (Hz) tities. With increasing ambient temperatures, however, the
T ambient temperature (°C) VE material softens and the effectiveness of the dampers
r/,. loss factor of damper is decreased. However, at the temperature of 42°C, the
dampers were still able to achieve more than 40% reduction
The above formulae were derived based on the average in response.
of the first 20 cycles of damper deformation with an aver- Foutch et al. 35 verified the effectiveness of VE dampers
age strain of 5%, which is considered to be reasonable dur- for seismic retrofit of nonductile reinforced concrete frames
ing a typical earthquake excitation. The loss factor, 71,, is by testing four subassemblage specimens under simulated
defined as v/v = W/27rE, where as before W is the energy strong ground shaking. Specimens without dampers
dissipated by the dampers in one cycle, and E is the total behaved in a very brittle manner as expected. The strength
strain energy of the structure. The loss factor is directly and stiffness degraded very quickly. The specimens with
related to the damping ratio through ~ = r/fl2. The damping viscoelastic dampers performed very well. The force-
ratio for the ith mode of the structure can be expressed as displacement response was observed to be stable, and a
significant amount of energy was dissipated by the dam-
~, = W'/(47rE ~) (13) pers. The maximum relative displacements of tested speci-
mens were held to levels that prevented brittle failure of
where ~:~is the structural damping ratio for the ith vibration the structural elements.
646 VE dampers for building motion control: B. Samali and K. C. S. Kwok
20.00" c 1.2-
18.00 t-
15.00"
~- i ¢.oo- 0.8-
.£ I 2.00 0.6-
~z 10.00 0,4 ~
'E 8,00
E 0,2-
6.00
4.00 5 4 3 2 1
2.00 [.o oom.. ! Floor
0.00 0.3-
20.0 25'.0 30.0 35.0 46.0 45.0
Temp.(°C)
4.00 0.2-
0.25- 1
3.g0
0.15-
3.80
~5.70
~ 3.60
C 3.50
-h 3.40
0.05-
0
5 4
!
3 2
W 1
Floor
3.20 I -~ 1.2
v
No Dampers I E
o 1
20.U 25'.o 3o'.o 3~o 4o'o 45.0 --~ 0 . 8
Temp.(°C) 8o
< 0.6
Figure9 Temperature dependence of structural dynamic o
properties 23 U_ 0.4
E
~ 0.2
Chang et al. 36 have also looked at the seismic retrofit of
~ 0
reinforced concrete frames with added viscoelastic dam- 5 4 3 2 1
pers. They studied a previously damaged l/3 scale Floor
reinforced concrete frame using viscoelastic dampers. Two
sets of VE dampers were first designed to provide extra 25°C ~ 30°C E 34°C
damping ratios of 12% and 18%, respectively, to the struc- 38°C ~ 42"C ~ No Dampers
ture to minimize damage due to future earthquakes. Exper-
imental studies carried out on the test frame with added VE Figure 10 Response of five-storey test frame with added dam-
dampers under scaled 0.2 g Taft earthquake confirmed the pers to Hachenoe earthquake (0.12 g)2S
effectiveness of the retrofit. Test results showed that the
seismic response of the structure was greatly reduced due
to the added damping provided by the VE dampers. The strain for a VE damper have been defined, the shear storage
inelastic deformation of the frame was correspondingly and loss moduli can be established from experimental data
reduced. provided by the manufacturer or by empirical equations
The methodology for the design of viscoelastic dampers developed from such data. Such empirical equations may be
in earthquake-resistant structures given by Abbas and expressed as functions of ambient temperature, excitation
Kelly ~5 is briefly described here. Abbas and Kelly proposed frequency and strain amplitude as
simplified analytical models consisting of a SDOF model
using a rigid brace and a two-DOF model with a deform- G' = f ( T , Y , f ) , G" = f ( T , 7,3") (18)
able brace to capture the response of viscoelastically
damped structures (Figure 12). They developed simplified Accordingly, the viscoelastic damper is modelled as an
design equations for the two models. A parametric analysis elastic spring and a dashpot acting in parallel as shown in
of the design equations has shown a significant influence Figure 13. The force in the damper is the sum of the elastic
of brace flexibility on the vibration period ratio and the and the viscous forces and both elements are subjected to
damping provided to the structure. A simplified model was the same deformations. The stiffness of the elastic spring
used to represent the viscoelastic damper. Based on this and the damping coefficient of the dashpot are defined,
model the viscoelastic material is assumed to have linear respectively, as
stiffness for a shear strain amplitude of 100% to 150%.
As mentioned earlier, the mechanical characteristics of VE Kd = G' A / t , Ca = G"A/(Wd t) (19)
materials are strongly dependent on temperature and load-
ing frequency. Once the temperature, frequency and shear where A is the area and t the thickness of the viscoelastic
VE dampers for building motion control: B. Samali and K. C. S. Kwok 647
(a) No Damper
1.20
0.60 .,,,ill,,,,. 1
o,oo ..,lllUIUlilllt ~,dlIIIIItIUtLIUIIU~ imm,qi,,A,
-0.60 "Vllll lll!lll.,
"IT[l'll'" i, i
!
I
1
<
!
-1.20 . . . . ; . . . . i , I
0 5 10 15 20 25 3O
Time (sec)
(b) T = 25°C
1.20
I
I
-ff
0.60-
l ...... t i I
o.oo A,A=~h._A. I I
-.-vVvr,TV v -~ -,, ,,,r,- ,
I
I ' I
•.o.,o . . . . !. . . . i.
! I i
!
!
i
i
.,.20 ! I I
0 5 10 15 20 25 30
Time (see)
(f) T = 42°C
t.20 I
Ii
!
!
0"601 t
i
I
~ 0.~
!
t5
it
"I .........................
!
!
120
° , J
0 5 10 15 20 25 3o
Time (see)
Figure 11 Damper effectiveness on 5th floor relative displacement of test frame 23
layer, and wd is the natural frequency of the viscoelastically the primary structure. The addition of viscoelastic dampers
damped structure. consistently reduces the displacement demands and thus
Extensive parametric analyses show that the acceleration, decreases or eliminates the nonlinear response in the pri-
displacement, ductility and energy dissipation response of mary structure in rad/s.
viscoelastically-damped structures can be significantly Individual VE dampers and associated braces in the mul-
influenced by: the period of primary structure, earthquake tistorey structure are designed using a drift-related pro-
ground motion characteristics, brace-to-structure stiffness cedure. A mode shape was assumed for the fundamental
ratio, loss factor of viscoelastic material; and the strength of mode of vibration of the viscoelastically-damped structure.
648 VE dampers for building motion control: B. Samali and K. C. S. Kwok
O6-- ...~ ,.A
riper
(a) Rigid Brace Model for ViscoelasUcally-Damped Structure
(a) Primary Structure
Brace
Bracing
Viscoelastic Damper
A
A
(b) FlexibleBrace Model for Viscoelastically-Damped Structure A
/%
Figure 12 Simplified analytical models for preliminary design
of viscoelastically-damped structures TM /%
II
(b) Viscoelastic,ally-Damped Structure
Figure 14 Analytical models for primary structure and viscoel-
astically damped structure ~s
C~k d
VisooelastJc Layer
/
Figure 13 Simplified analytical model for viscoelastic
/
/
damper15 /
0 0
0 o
The resulting interstorey drift demand is related to the vol-
ume of the viscoelastic material and the stiffness of the Ste~ Plete
bracing to be provided at each storey. The braces may be
provided in any configuration and are designed to remain
elastic during a design earthquake. The design process was
0 0
illustrated with the design of viscoelastic dampers and 0 0
associated braces for a nine-storey moment resisting steel
frame (Figure 14). A typical viscoelastic damper used with Figure 15 Typical viscoelastic damper used with chevron brac-
chevron bracing is shown in Figure 15. Extensive nonlinear ing ~
time-history analyses of a viscoelastically-damped frame
subjected to different earthquake ground motions showed World Trade Center in New York City in 1969. They are
significant reduction in the storey shear force and inter- an integral and essential part of the structural system, being
storey drift maxima. Approximately 85% of input energy designed to assist the tubular steel frame in suppressing
is dissipated through viscoelastic devices as shown in Table wind-induced building oscillations to a level below human
1 and Figures 16-20. perception. The selection, quantity, shape and location of
the dampers was based on the dynamic analysis of the tow-
ers and the damping required to achieve performance
7. First installation of VE dampers
requirements.
The viscoelastic damper was conceived and developed as Each of the two towers employ approximately 10 000
part of the structural design for the twin towers of the viscoelastic dampers. A typical damper configuration used
VE dampers for building motion control: B. Samali and K. C. S. Kwok 649
35.0,
Table 1 Energy dissipation indices (EDI) for the viscoelast-
ically-damped moment resisting steel frame (MRSF) for differ-
30.0 Input Energy
ent earthquake ground motions
--...
A 25.0
Earthquake EDI I¢
~ 20.0
El Centro 0.84
San Fernando 0.84 Vi=~oelm~ie~ly-Oit~il~ted Enet~/
w 15.0
Mexico City 0.85
Loma Prieta 0.84
10.0
Miyagi-Ken-Oki 0.84
80.0
5.0
0'00.0
J 10.0
I I
20.0
I
30.0
, I
40.0
TIME (Nconds)
Figure 19 Energy-time histories for viscoelastically-damped
60.0 MRSF (Loma Prieta earthquake, 1989)TM
nput Energy
120.0 ,
40.0
,=,,
j ~ Vi~-Dis~p~m:l Energy 100.0
20.0
80.0
>-
80.0
0.0 .,V ., I i I I I , I
0.0 10.0 20.0 30.0 40.0
TIME (seconds} 40.0
Figure 76 Energy-time histories for viscoelastically-damped 20.0
MRSF (El Centro earthquake, 1940)TM
70.0 0.0
0.0 5.0 10.0 150 20.0 25.0
J TIME ( l ~ )
60.0
Figure20 Energy-time histories for viscoelastically-damped
50.0 MRSF (Miyagi-Ken-Oki earthquake, 1978)TM
Input Energy
J
40.0 ViSCoelutiCally-Dil~ pated Energy
gg
in the World Trade Center buildings is shown in Figure 21.
30.0
The dampers are distributed evenly throughout the building
from the 10th to the 110th floor. They are located between
S
20.0
the lower chords of the horizontal trusses (point A in Fig-
10.0 ure 22) and the columns of the outside wall (point B in
Figure 22). Due to the relative motion between the lower
0.0 I I
0.0 5.0 10.0 15,0 20.0 chord of the truss and the column on the building perimeter
TIME (oeooncls) in moderate to severe winds, part of the energy of oscil-
lation is dissipated in the form of heat due to the shear
Figure 17 Energy-time histories for viscoelastically-damped
MRSF (San Fernando earthquake, 1971)1s
deformation generated in the viscoelastic element of the
damper.
Since their installation in 1969, the dampers have been
25.0
20.0
Input E ~
A¢
.~ 15.0
t0
• " Ily-Di~wipated Energy
10.0
5,0
0.0
0.0 20.0 40.0 60.0 80.0 100,0
TIME ( s i m o n )
Figure 18 Energy-time histories for viscoelastically-damped
MRSF (Mexico City earthquake, 1985) TM Figure 21 Typical construction of World Trade Center damper 6
650 VE d a m p e r s for building m o t i o n control: B. Samali a n d K. C. S. K w o k
measured period of vibration of structure in fun-
damental mode
The determination of the energy absorbed by the viscoel-
astic dampers is also complex, but can be significantly sim-
plified by assuming that the energy absorbed by a given
damper is associated directly with the slope of the funda-
mental mode shape of the building at the location of that
damper (Figure 23). While some loss of precision is associ-
ated with this simplification, the loss was not considered
significant. The expression used by Mahmoodi for the total
energy dissipation by the dampers was based on the local
movement of dampers at the 105th floor. Since it is desir-
Figure 22 World Trade Center damper in places able to compare the energy dissipation of the dampers with
the energy of building oscillation, the data from the damper
periodically checked for aging characteristics and they have and building acceleration were collected simultaneously. In
not shown any deterioration of their basic properties. To addition, the formulation used assumes that laboratory data
assess the effectiveness of the dampers, a program of is available which relates measured damper movement to
dynamic measurement was initiated 6. Four dampers on the damper energy dissipation. The expression for the energy
105th floor were instrumented with LVDTs to measure the dissipation by the dampers is
shear deformation in the viscoelastic element of the dam-
DE,,,,1 = M DEH~ (21)
pers. The wind-induced response of the building was meas-
ured by accelerometers installed on the 108th floor. An where DE,,,,,~ is the total energy dissipated by all dampers
anemometer was also placed above the roof to measure the for building motion in a given direction; DEto5 is the energy
wind speed and direction. The performance of the dampers dissipated by a single damper at floor 105 for the same
was tested during Hurricane Gloria (wind gusts up to direction; and M is a constant relating to the total number
85 kin/h) for which the total damping of the building was of dampers in a given direction and their relative motion
calculated and found to be in the range of 2.5-3.0% of with respect to each other.
critical damping, following an estimation of the total oscil- After the energy dissipated by the dampers was determ-
lation of the building. ined, the contribution to overall damping by the viscoelastic
Determination of the total energy of the oscillation is dampers was calculated from
complex for any structure and is dependent on several para-
meters such as loading, stiffness, mass, damping, mode = DE,,,,,J4rr E,,,,,,z (22)
shape etc. Mahmoodi 6 estimated the total energy by adopt-
ing several simplifying assumptions, the most important The performance of the dampers over the last 25 years has
being been good. The New York Port Authority has projected that
the dampers will remain a viable component of the building
(1 The energy of oscillation is based on the mode shape until at least the year 2015.
for the fundamental mode of vibration of the building
(i.e., higher modes of vibration are neglected) 8. Installation in the Columbia SeaFirst building
(2) Variations between the estimated and actual density in Seattle
and stiffness of the building will have a negligible In 1982, a nonstructural passive viscoelastic damper system
effect on the mode shape was designed and installed in the Columbia SeaFirst Build-
The method of calculation gives the energy of oscillation C,=y
at an instant in time as a function of the peak acceleration
at that instant. The equation used to calculate the total
energy stored by the structure in oscillation assumes that
the recorded data for the peak lateral accelerations of the
105.
building is available in the fundamental mode of the struc-
ture. Two additional parameters in the equation allow for
the modification of the result for the actual building weight
and the actual period of vibration. The total energy is
given by
e,o,al = K W (ama x T) 2 (20)
where (exaggeratedfor clarity)
ftotal total energy of oscillation for movement in same
direction as a ....
K a constant for a given structure
amax peak acceleration measured at 105th floor Lateral Drift
W total weight of building and contents within Figure 23 Deflected shape of building for first mode of
boundary of lateral oscillation vibration 6
VE dampers for building motion control: B. Samali and K. C. S. Kwok 651
ing in Seattle, Washington. The building is a 76-storey 9. Other installations
office tower rising to a height of 286 m above ground level.
The building has a triangular plan with the appearance of In 1988, a unique passive damping system was installed in
three interconnected stair stepping towers with concave, the Two Union Square Building in Seattle, Washington.
convex and flat surfaces. The lateral force resisting system This is the fourth major US high-rise building to incorpor-
of the building consists of three massive composite con- ate this effective energy absorbing system to control
crete columns interconnected with steel braced moment excessive accelerations. Viscoelastic dampers have also
frames 37 (Figure 24). Wind tunnel tests performed on a been used in a number of tall buildings in Japan. Seavans
scale model at various flow conditions confirmed the need S Tower, a 100m tall office building in Tokyo was
to reduce acceleration levels. To reduce wind-induced equipped with VE dampers in 1991. The VE material is a
motions and to ensure occupant comfort, the design called bitumen rubber compound. B Office Building, a 29 storey
for 260 dampers to be installed alongside the main diagonal steel structure is also equipped with 72 VE dampers. The
bracing members in the core of the building as shown in dampers are installed in one direction only. Four units per
Figure 25. As the bracing member elongates or shortens in floor are installed at floors 3 to 20 (inclusive). The dampers
the elastic range due to dynamic motion, the viscoelastic are of brace type with an acrylic VE material.
element within the damper is deformed in shear thus dissi- Although VE dampers have been proposed and installed
pating energy in either direction. The temperature depen- mainly in tall buildings 38,39, other structures have also bene-
dence of the properties of a typical VE damper installed in fited from their installation. Chiba Sogo Gymnasium, a
Columbia SeaFirst Building is shown in Figure 26. The 122 m tall steel structure in Chiba is also equipped with a
units are very cost effective, representing roughly 0.5% of similar damper system, i.e., brace type using acrylic VE
the total construction cost, and require no maintenance. The materials. The VE dampers were installed in 1993. Viscoel-
addition of this system to the Columbia Centre Building is astic dampers have been installed in the Chientan railroad
calculated to increase the critical damping ratio from 0.8 station roof in Taipei, Taiwan 4°. The station which is in the
to 6.4% for frequent storms and to 3.2% at design wind. heavy traditional Chinese architectural style has a curved
Figure 27 shows the effectiveness of VE dampers in reduc- roof shell 187.6 m in length, 16.6 m in width and weighs
ing acceleration levels for different storms. It is observed approximately 2800 t. The roof shell is not supported by
that the acceleration levels with VE dampers in place are columns, rather, it is suspended below two main cables by
below the acceptance levels. 26 pairs of suspenders. Wind tunnel studies of the roof
/
BRACED MOMENT FRAMES
Sl LEVEL ,~' PLAN VIEW
LEVE
35
COMPOSITE
CONCRETE
30 COLUMN
25
VISCOELASTIC
OAMPERS
20 ON BRACES
SHOWN i
SCHEMATICS OF DIFFERENT STRUCTURAL ACTIONS OF MOMENT FRAMES
Figure 24 Plan view and details of wind resisting system for Columbia SeaFirst Building 37
652 VE dampers for building motion control: B. Samali and K. C. S. Kwok
t
Strul
Damper
I
Damper
I
Strut
-+i L
I
I
[
---b
Figure 25 Damper locations in Columbia SeaFirst Building 3
structure revealed that without additional damping, aerody- response data indicated large and long-duration motion,
namic instability was likely to commence suddenly at wind including torsional coupling, during even moderate earth-
speeds of about 113 km/h. At this wind speed the roof will quakes. Initially, three different types of devices were con-
vibrate violently in both the vertical and torsional modes. sidered. They were a steel-yielding damping device, a fric-
This dynamic response could not satisfy the satisfactory tion-slip energy dissipation restraint, and a viscoelastic
performance criteria at 146 km/h hourly wind speed set for damper. The 3M VE dampers were chosen primarily
Tamshui, Taipei. Following further wind tunnel investi- because they provided the structure with significantly
gations incorporating VE dampers, it was concluded that increased damping for frequent low-level ground shaking,
damping ratios of 8% and 4% of critical are adequate and as well as for larger seismic events. The final design called
practical values (for vertical and torsional modes, for the installation of two dampers per building face per
respectively) to ensure aerodynamic stability at all wind floor, which would increase the equivalent damping in the
speeds up to the design wind speed. Based on the assumed fundamental mode of the building to about 17% of critical,
structural inherent damping ratios of 0.6% and 1% for the providing substantial reductions to building response under
vertical and torsional modes, respectively, the required all levels of ground shaking 41'42.
damper stiffnesses were calculated from the added damping
ratio requirement of 7.4% for the vertical mode and 3% for
the torsional mode. The addition of dampers shifted the
natural frequencies of the roof structure. The VE materials I0. Conclusions
will work in shear. The required thickness for VE material
was 84 mm with a required shear area of The effectiveness of viscoelastic dampers in suppressing
150 mm x 1020 mm. The damper consists of two subdam- wind- or earthquake-induced motion of building structures
pers connected in series to fit into the light post. 40 alu- has been demonstrated both experimentally and by numeri-
minium fins each 1.6 mm in thickness are used to dissipate cal simulation by many researchers. Viscoelastic dampers
the heat generated in the viscoelastic material during have been successfully incorporated in a number of major
strong wind. tall buildings in USA and Japan, including the twin towers
Viscoelastic dampers are currently being installed in the of the World Trade Center Buildings in New York City
Santa Clara Country building in San Jose, California to and the Columbia SeaFirst Building in Seattle, Washington.
address a seismic need. Situated in a high seismic risk The authors have summarized the research work in this area
region, the SCC building was built in 1976. It is approxi- and identified the factors affecting the performance and
mately 6 4 m in height and nearly square in plan design of VE dampers. The properties of viscoelastic dam-
(51 m × 51 m on typical upper floors). The exterior clad- pers, possible damper configurations and damping mechan-
ding consists of full-height glazing on two sides and metal ism were also discussed. Some guidelines for the design of
siding on the other two sides. The exterior cladding, how- dampers were offered and the application of VE dampers in
ever, provides little resistance to structural drift. The equiv- the aseismic design of structures was emphasized. Finally, a
alent viscous damping in the fundamental mode is less than few examples of earlier installations were given in some
1% of critical. The building has been extensively instru- detail to clearly demonstrate the effectiveness and potential
mented, and useful data has been obtained during a number of VE dampers as a viable, cost effective and maintenance
of recent small to moderate earthquakes. A plan for seismic free damper system to control the motion of dynamically
upgrade of the building was developed, in part, when the sensitive structures subjected to environmental loads.
VE dampers for building motion control: B. Samali and K. C. S. Kwok 653
5516 No Dampers. Resultant at
i
. y " ExtremeCornerofBuilding
4964 "°o. / With Dampers,Resultantat
40%Strain 30- ~ . . . ExtremeCornerof Building
4413 • "*%o
"~/~ .. ~ Acceptance Level
3881 25-
/ -".. ~" "'.. j No Dampers, at Center
| 3310 ff
o 20 '% / "'~.. "./. of Building
,, "o% "°,
2758
15, %%% "~~ %%,
,,,, ~ "%'. "-...
22O6 <o
•".,,. "o..°
With Dampers *~,
1855 10. a Centerof ~ ' ~ , ~ , , "~"~ "". "..
Building ~,~,~, "%~,~ "".....".
1103
5 Perception Threshold "*"~" " ~ ,~~
552
0
i I I t I I I I I 0 i i J i
15,5 18.3 21.1 23.9 26.7 29.4 32.2 35.0 37.8 40.5 43.3 Once Every OneeEvery Once 10Times 100Times
100Years 10 Years a Year a Year a Year
Temperature (C *)
G' vs. Temperature (Damper #1) Figure27 Effectiveness of VE dampers in reducing acceler-
ation levels for different storms 3
6894
8205
40%Strain References
5516 1 Mahmoodi, P. 'Structural dampers', J. Struct. Div., ASCE 1972, 95
(ST8), 1661-1672
4826
2 Mahmoodi, P. "Design and analysis of viscoelastic vibration dampers
4137 for structures', Proe. INOVA-73 World Innovative Week Cop
