You are on page 1of 11

Engineering Structures 30 (2008) 930940

www.elsevier.com/locate/engstruct

Allocation and slip load of friction dampers for a seismically excited


building structure based on storey shear force distribution
Sang-Hyun Lee a, , Ji-Hun Park b , Sung-Kyung Lee c , Kyung-Won Min a
a Department of Architectural Engineering, Dankook University, Republic of Korea
b Department of Architectural Engineering, University of Incheon, Republic of Korea
c Team for Education of Remodelling Elementary Technologies and Training of High Quality Researchers, Dankook University, Republic of Korea

Received 25 August 2006; received in revised form 1 March 2007; accepted 20 March 2007
Available online 24 July 2007

Abstract
In this study, a seismic design methodology of the friction dampers based on the storey shear force distribution of an elastic building structure
is proposed. First, using two normalization methods for the slip-load of a friction damper, numerical analysis of various single degree-of-freedom
(SDOF) systems is performed. From this analysis, the effect of the slip-load and brace stiffness is investigated and optimal stiffness ratios of the
brace versus primary structure are found. Second, from the numerical analysis for five multistorey building structures of different natural periods
and numbers of story, reasonable decision method for the total number of installation floors, location of installation and distribution of slip-loads
are drawn. In addition, an empirical equation on the optimal number of installation floor is proposed. Finally, the superiority of the proposed
method compared to the existing design method is verified from numerical analysis.
c 2007 Elsevier Ltd. All rights reserved.

Keywords: Friction dampers; Storey shear force distribution; Slip-load; Brace stiffness; Optimal design; Empirical equation

1. Introduction
There have been many researches to reduce the seismic
response of large-scale building structures using the frictiontype damping device, because friction is effective in dissipating
a large amount of energy. Pall and Marsh proposed a friction
damper installed at the crossing joint of the X-brace to avoid
the compression in the brace member [1]. Constantinou et al.
proposed a friction damper composed of a sliding steel shaft
and two friction pads clamped by adjustable bolts [2]. Li and
Reinhorn verified the seismic performance of a building model
with friction dampers both analytically and experimentally [3].
Grigorian et al. examined the energy dissipation effect of a
joint with slotted bolt holes [4]. Mualla and Belev proposed
a rotational friction damper with adjustable slip-moment [5].
Cho and Kwon conducted numerical modelling and analysis
of a wall-type friction damper in order to improve the seismic
performance of the reinforced concrete structures [6]. In
Corresponding author. Tel.: +82 2 799 1439; fax: +82 2 749 8653.

E-mail address: lshyun00@dankook.ac.kr (S.-H. Lee).


c 2007 Elsevier Ltd. All rights reserved.
0141-0296/$ - see front matter
doi:10.1016/j.engstruct.2007.03.020

addition, magneto-rheological (MR) fluid damper, which has


been applied successfully in the field of structural control due to
its large capacity of the control force and adjustability through
electric signals, has the characteristics of friction damping [7].
Equivalent lateral force and modal analysis procedures were
developed and incorporated in the 2000 NEHRP provisions for
the design of the yielding buildings with linear and nonlinear
viscous damping systems [8]. In the design of the friction
damper for seismic structural control, it is the most important
factor to determine the quantity and slip-load of the damper
and the brace stiffness systematically. Filiatrault and Cherry
proposed a design procedure for equally-distributed friction
dampers minimizing the sum of normalized displacements
and dissipated energy through parametric study on the natural
period, the frequency content of an earthquake and the slip-load
of the friction damper [9]. Fu and Cherry proposed a design
procedure of the friction dampers using a force modification
factor [10]. Ciampi et al. developed a simple approach for
determining the distribution of stiffness and strengths within the
elastic and inelastic structures [11]. Kim and Choi calculated
the yield load of the buckling-resistant-brace system using

S.-H. Lee et al. / Engineering Structures 30 (2008) 930940

energy spectrum [12]. A common assumption in the design


procedure of those previous studies is that dampers have equal
or linear stiffness and strength and they are installed on the
whole floors of the building structure. However, it may be
more effective to assign optimal slip-load to each friction
damper. Under this premise, Moreschi and Singh pointed out
that gradient-based optimization technique is difficult to apply
to the design of nonlinear dampers such as friction dampers and
proposed an optimization procedure using a genetic algorithm
(GA) as an alternative [13]. Garcia and Soong proposed a
simplified sequential search algorithm (SSSA), which gives
optimal floor distribution of the viscous damping by repeated
installation of unit viscous dampers on the floor with the largest
controllability index defined by interstorey drift or relative
velocity [14]. Also, Shulka and Data applied the SSSA to
the design of a visco-elastic damper (VED) [15]. However,
those studies do not consider the economic profit obtained by
reducing the number of installation floors. Furthermore, the GA
and SSSA have a disadvantage that time-consuming nonlinear
time history analysis needs to be performed repeatedly.
The purpose of this study is to propose a design
methodology of friction damperbrace (FDB) systems, to
determine the quantity and slip-load of the friction damper
and the brace stiffness systematically for an elastic multistorey
building structure based on the story shear forces. First, for
two normalization methods of the slip-load, the variation of
three response quantities, the displacement, storey shear force
and dissipated energy, with respect to the slip-load and brace
stiffness are investigated through the numerical analysis of
various SDOF systems. Then, optimal stiffness ratios of the
brace over the primary structure are found for a range of
the natural periods. Second, seismic performance variation of
multistorey building structures with respect to the quantity,
total slip load and slip-load distribution of the friction dampers
is investigated and an empirical equation to determine the
optimal quantity of the friction dampers is proposed by
regression analysis. Finally, the proposed design methodology
is compared with the SSSA through numerical analyses.

931

Fig. 1. Friction damperbracestructure system.

Fig. 2. Forcedisplacement relation of the FDBS system.

where SR is the stiffness ratio between the brace and the unit
frame given by
SR =

kb
.
kf

(2)

In this study, the numerical analyses were performed using


the well-known Newmarks method for obtaining the dynamic
responses of the bilinear system.
3. Design of the friction damper for a SDOF system

2. Friction damperbracestructure system


In general, additional bracing members are required to install
a damping device between two adjacent floors of the building
structure. The assemblage of the friction damper, brace and unit
frame of the primary structure can be represented as in Fig. 1,
where k f , m f , and c f , denote the lateral stiffness, mass and
damping of the unit frame, and kb , f s , x and f denote the
lateral stiffness of the brace, the slip-load of the friction damper,
the displacement of the unit frame and the external load,
respectively. The primary structure and the brace are assumed to
behave elastically. Therefore, the forcedisplacement relation
of the friction damperbracestructure (FDBS) system is
modelled as a bilinear system presented in Fig. 2, of which
equivalent yield strength, f y can be represented by the
following equation [13].


1
f y = fs 1 +
(1)
SR

Performance of the friction damper changes with respect


to the magnitude and frequency content of the load due to its
strong nonlinearity. In particular, the influence of the brace
should be taken into account in the seismic design of the
friction damper, because the brace changes the natural period of
the structure on which the seismic response depends strongly.
Therefore, the performance variation of the friction damper
with respect to the brace stiffness and slip-load of the friction
damper is investigated in this section.
Numerical analysis is conducted for mass-normalized SDOF
structures with a damping ratio of 5%. The sample ground
acceleration is generated using EQmaker, a software generating
ground acceleration data from the design spectrum represented
in Fig. 3 [16], where seismic coefficients Ca and Cv are
assumed to be 0.09 and the period Ts and To are determined as
Cv /(2.5Ca ), and 0.2Ts , respectively. The applied stiffness ratio,
SR, is 1, 2, 5, 10 and 50. Performance indices for evaluation are

932

S.-H. Lee et al. / Engineering Structures 30 (2008) 930940

where m, c, x and x g denote the mass, the viscous damping


coefficient, the displacement and the ground acceleration,
respectively. The slip-load normalized by the peak storey shear
force shared by the brace, which is obtained from the time
history analysis of Eq. (5), is expressed as follows:
=

fs
fb

(5)

where is the normalized slip-load based on the normalization


method A and f b is given by
f b = kb |x|max .

Fig. 3. Pseudo-acceleration spectrum for the generation of the artificial


earthquake.

defined as
Rd =

|x(t)|max
,
|xo (t)|max

Rf =

|Fs (t)|max
,
|Fso (t)|max

Re = 1

(3)
where xo , x, Fso , Fs , E h and E I are the displacement before
and after damper installation, the story shear force before
and after damper installation, the energy dissipated by the
friction damper and the earthquake input energy, respectively.
Therefore, Rd and R f represent the response reduction due
to the friction damper and Re represents the fraction of the
earthquake input energy taken by the primary structure. Since
the absolute acceleration response has a trend similar to the
story shear force [17], only the latter is considered in the
performance index.
3.1. Normalization of the slip-load
In the design of the friction damper, the ratio of the slip-load
over the external load is more important than the magnitude
of the slip-load. Therefore, it is useful to normalize the slipload of the friction damper with respect to the magnitude of
the external load. We can consider two normalization methods.
One is normalization to the storey shear force shared by the
brace when the slip of the friction damper is restrained, and the
other is normalization to the storey shear force of the original
primary structure before the installation of the friction damper
and brace. In this section, the strength and weakness of those
two normalization methods are compared by application to the
design of the FDB system for various SDOF systems.
Normalization method A
Assuming that the slip of the friction damper is restrained
by a very high slip-load, the FDBS system is the same as a
structure with only brace, of which equation of motion is given
as
m x + cx + (k f + kb )x = m x g

In the last equation, |x|max is the peak displacement obtained


by the time history analysis of Eq. (4). equal to 1.0 means that
slip does not occur and the FDB system behaves like a brace
alone. Therefore, is designed to be lower than 1.0.
Normalization method B
The SDOF system without the FDB system is given as
m x + cx + k f x = m x g .

Eh
EI

(4)

(6)

(7)

The slip-load normalized by the peak storey shear force,


which is obtained from the time history analysis of Eq. (7), is
expressed as follows:
=

fs
ff

(8)

where is the normalized slip-load based on the normalization


method B and f f is given by
f f = k f |x|max .

(9)

In the last equation, |x|max is the peak displacement obtained


by the time history analysis of Eq. (7). Using the normalized
slip-load, , has an advantage in that a unique slip-load is
determined for a primary structure irrelevantly to the brace
stiffness. However, the use of the normalized slip-load gives
different slip-loads corresponding to the change of the brace
stiffness.
3.2. Design using the normalization method A
The result of numerical analysis for the FDB system
designed using the normalization method A is presented
in Figs. 46. First, Fig. 4 represents the variation of the
performance indices with respect to for the SDOF systems
of which natural period is 1.0 s and stiffness ratios are different.
In Fig. 4(a), Rd decreases as increases in the range of < 0.3.
On the other hand, Rd decreases at slower rate or turns to
increase as increases in the range of > 0.3. Also, it
can be observed that the brace stiffness improves displacement
reduction effect, since Rd becomes lower for larger SR in the
entire range of . In Fig. 4(b), R f changes little for < 0.3, but
decreases as SR increases. However, R f increases rapidly, as
increases in the range of > 0.3. This is because infrequent
occurrence of the slip in the friction damper shortens the natural
period and increase the storey shear force, accordingly. In

S.-H. Lee et al. / Engineering Structures 30 (2008) 930940

933

Fig. 4. Performance index variation with respect to (Tn = 1.0 s).

Fig. 5. Performance index variation with respect to SR ( = 0.2).

Fig. 6. Mean optimal SR for each performance index (0.1 0.3).

Fig. 4(c), Re reaches its minimum in the range of 0.1


0.3. Also, convergence of Re to 1.0 at of 1.0 means that the
input earthquake energy is not dissipated at all because the FDB
system behaves just like a brace.
Fig. 5 shows performance index variation with respect to
SR for of 0.2 and different natural periods of the original
structure without FDB, denoted as Tn . For each performance

index, mean optimal SR with respect to the natural period is


represented in Fig. 6, where optimal SRs are calculated for
0.1 < < 0.3 and averaged to obtain the mean value. The
range of for averaged optimal SRs is determined considering
that all three performance indices improve little or deteriorate
out of the chosen range of as can be observed in Fig. 4. In
Fig. 5(a), the extremal point of Rd is observed at SR < 10 for

934

S.-H. Lee et al. / Engineering Structures 30 (2008) 930940

Fig. 7. Performance index variation with respect to (Tn = 1.0 s).

Fig. 8. Performance index variation with respect to SR ( = 0.2).

the short period SDOF systems of Tn = 0.1 and 0.2 second.


However, Rd decreases monotonically as SR increases for the
long period SDOF systems of Tn 0.5. In Fig. 6(a), for
Tn < 0.5, mean optimal SR increases from 1 to 50 almost
linearly, and for Tn 0.5 the optimal SR is constant and equal
to 50, which is the upper limit of SR used in the analysis. In
Fig. 5(b), for the SDOF systems of Tn 1.0 s, R f tends to
decrease as SR increase, and those of Tn = 2.0 and 5.0 s have
minimum R f at SR 1. This supports the Fig. 6(b) where
optimal S R for Tn 2.0 is much lower than that for Tn < 2.0.
In Fig. 5(c), Re for Tn = 0.1 and 0.2 s is minimum for SR = 10,
Re for Tn = 0.5, 1.0 and 2.0 s decreases as SR increases and
this tendency becomes strong as Tn increases. In Fig. 6(c), the
mean optimal SR for the SODF systems of Tn < 1.0 varies
almost linearly and that for those of Tn > 1.0 is close or equal
to 50.
The design result using the normalization method A is
summarized as follows. From Fig. 4, the friction damper is
observed to be effective for the reduction of the displacement
and non-dissipative energy for 0.1 0.3. From Figs. 5 and
6, it should be noted that the optimal SR is different depending
on which performance index and natural period are to be
considered in the design. This means that the brace stiffness
should be chosen appropriately considering trade-off between

different performance indices. In addition, the normalization


method A yields different slip-load even for a single value of
, because the story shear force shared by the brace changes by
SR. As a result, the brace stiffness and slip-load of the friction
damper should be determined simultaneously, which makes it
difficult to apply the normalization method A to design practice.
3.3. Design using the normalization method B
The result of numerical analysis for the FDB system
designed using the normalization method B is presented in
Figs. 710. Fig. 7 represents the variation of each performance
index for the SDOF system of Tn = 1.0 with respect to and
tendencies similar to Fig. 4 is observed. In Fig. 7, it is noted that
the reasonable value of is lower than 0.3 because increasing
higher than 0.3 causes R f and Re to increase considerably
and Rd to increase or change little except for SR = 50. This
means that the reasonable slip-load of the friction damper is
30% of the storey shear force in the original structure before
the installation of the FDB system.
Fig. 8 represents the variation of the performance indices
with respect to Tn , for = 0.2. In Fig. 8(a) and (b), the
displacement and storey shear force tend to decrease as SR
increases except for short period structures. The mean optimal
SRs for each performance index is presented in Fig. 9. In

S.-H. Lee et al. / Engineering Structures 30 (2008) 930940

935

Fig. 9. Mean optimal SR for each performance index ( < 0.3).

Fig. 10. Performance index variation with respect to the natural period and .

Fig. 9(a) and (b), the optimal SRs are calculated for < 0.3
and averaged to obtain the mean optimal SR. In Fig. 9(c),
the optimal SRs are calculated for three sub-divided ranges
of < 0.3. The range of for averaged optimal SRs is
chosen considering that all three performance indices do not
improve by increasing higher than 0.3 and, in particular, Re
has different variation tendency for three sub-divided ranges of
. From Fig. 9(a) and (b), it is observed that the mean optimal
SRs for Rd and R f are very similar. For Tn 0.5, SR equal to
50, which is the largest SR used in the analysis, minimizes Rd
and R f , while for Tn < 0.5, the mean optimal SR is between
10 and 50. This means that, if the normalization method B
is applied, both Rd and R f can be minimized simultaneously
by the same SR, differently from the normalization method A.
Although the mean optimal SR for Re has a tendency different
from those for Rd and R f , the nondissipative energy itself is not
a main design objective and can be excluded from the principal
factors in determination of the brace stiffness. Therefore, the
normalization method B makes it possible to design the FDB
system using a unique SR so that the displacement and story
shear force are minimized simultaneously. In addition, using
the normalization method B, a unique slip-load is defined by a
single value of , so that can be determined independently of
the brace stiffness. Due to these advantages, the normalization

method B is adopted in the design of the FDB system for


MDOF structures which is discussed in the next section.
Fig. 10 represents the variation of the three performance
indices, which are achieved by the corresponding mean optimal
SRs given in Fig. 9, with respect to and Tn . From Fig. 10, it
can be observed that one specific value of ensures a roughly
uniform performance level for the whole Tn s in consideration.
For example, = 0.1 gives performance indices such as
0.2 < Rd < 0.4, 0.3 < R f < 0.5, and 0.2 < Re < 0.5.
Therefore, when a target performance index is given, can be
determined using Fig. 10, and then the corresponding slip-load
is calculated regardless of the brace stiffness.
4. Design of friction dampers for multistorey building
structures
In this section, a methodology to determine the damper
location and the distribution of the slip-load is proposed for
the seismic response reduction of the multistorey building
structures from numerical analysis of various example
structures. In this procedure, the relation between the total slipload of the friction dampers and the number of the installation
story is investigated. The ground acceleration is the same one
as applied to the SDOF system. Five example structures are
chosen among 25 structures to which Garcia and Soong applied

936

S.-H. Lee et al. / Engineering Structures 30 (2008) 930940

Fig. 11. Comparison of inter-storey drift for two slip-load distribution methods.

the SSSA [14]. Their 25 example structures consist of five


groups corresponding to 4, 8, 12, 16 and 20-storey structures
respectively, and each group has five natural periods equal to
0.4, 0.8, 1.2, 1.6 and 2.0 s. However, time history analysis
result shows that the structures that belong to the same group
have little difference in the relative distribution of the peak
inter-story drifts and peak storey shear forces. Accordingly,
0.4, 0.8, 1.2, 1.6 and 2.0 s are chosen as representative natural
periods corresponding to 4, 8, 12, 16 and 20-storey structures,
respectively, and the analysis is conducted only for these five
structures. Modal damping ratio assumed to be 0.05 for every
example structure.
The friction damper becomes more effective for larger
displacement response, since it belongs to a displacementdependent type damper. Therefore, floors to install the FDB
system are chosen in the order of the RMS interstorey drift of
the original structure without FDB, as many as a predefined
target quantity. The brace is assumed to be rigid, because the
brace stiffness is useful for the reduction of the displacement
and storey shear force except for the short period structures
as illustrated in Fig. 9. Performance indices are defined as
follows:
max {|xi (t)|max }
,

Rd =
max xo,i (t) max
i=1,...,N


max xa,i (t) max
i=1,...,N
,

Ra =
max xoa,i (t) max
i=1,...,N


max Fs,i (t) max
i=1,...,N


Rf =
max Fos,i (t)
i=1,...,N

i=1,...,N

since those two responses of the multistorey building structure


may have different tendency unlike those of the SDOF system.
4.1. Distribution of slip-load
Based on the normalization method B, the total slip-load is
normalized with respect to the sum of all the storey shear forces.
Accordingly, the total slip-load, denoted by Ftotal , is expressed
by the following equation:
Ftotal =

Nf
X

fs j =

j=1

N
X

k f i Sdi

(11)

i=1

where f s j denotes the slip-load of the friction damper installed


on the j-th floor, k f i , Sdi , N and N f denote the lateral stiffness
and peak interstorey drift of the i-th floor, the total number
of floors, and total number of installation floors, respectively.
Ftotal given by Eq. (11) is distributed to the slip-load of the
each friction damper equally or in proportion to the storey shear
force of the corresponding floor as follows:
f s j = Ftotal /N f
f s j = Ftotal

(12)

k f j Sd j
Nf
P

(13)

k f n Sdn

n=1

(10)

max

where xo,i , xi , xoa,i , xa,i , Fos,i and Fs,i are i-th interstorey drift
before and after damper installation, i-th absolute acceleration
before and after damper installation, i-th story shear force
before and after damper installation, respectively, and N is the
total number of floors. Both the absolute accelerations and the
storey shear forces are included in the performance indices,

where k f n and Sdn denote the lateral stiffness and peak


interstorey drift of the n-th floor on which the friction damper
is installed. For the comparison of the two slip-load distribution
methods, the ratio of the performance indices on the interstory drift, denoted by Rd1 for the Eq. (13) and Rd2 for the
Eq. (12), is represented in Fig. 11. From this figure, it can be
observed that Rd1 /Rd2 is lower than 1.0 mostly. Therefore, it
can be concluded that slip-load distribution proportional to the
storey shear forces is more effective than equal distribution for
interstorey drift reduction.
4.2. Relation between the performance and the total number of
installation floors
The total number of installation floors is an important
factor related to the construction cost directly, but has been

S.-H. Lee et al. / Engineering Structures 30 (2008) 930940

937

Fig. 12. Performance index with respect to the number of installation floors (N = 4, Tn = 0.4 s).

Fig. 13. Performance index with respect to the number of installation floors (N = 12, Tn = 1.2 s).

Fig. 14. Performance index with respect to the number of installation floors (N = 20, Tn = 2.0 s).

overlooked in many existing researches. In this section, the


relation between the performance and the number of installation
floors is investigated in order to achieve design objectives more
economically and a reasonable guideline to determine the total
number of installation floors is proposed.
Figs. 1214 represent the variation of Rd , Ra and R f with
respect to the number of installation floors, N f , for three
example structures of which number of stories are 4, 12 and

20 and corresponding natural periods are 0.4, 1.2 and 2.0 s,


respectively. The distribution of the slip-load is determined
using Eq. (13). From Fig. 12(a), it is observed that increasing
N f over a specific number does not reduce Rd and, for a certain
value of , even increases Rd . For = 0.01, N f = 1 yields
the lowest Rd and this means that concentrative installation of
the friction damper on a single storey is more effective than
distributive installation for this value of . In the same way,

938

S.-H. Lee et al. / Engineering Structures 30 (2008) 930940

Fig. 15. Performance index J .

for = 0.1, N f = 2 yields the lowest Rd . From such a


tendency observed in the Fig. 12(a), it can be inferred that
concentrative installation is effective for small amount of total
slip-load, while distributive installation is effective for large
amount of total slip-load. Similar tendency can be found for
Ra and R f in Fig. 12(b) and (c). In addition, from Figs. 1214,
it should be noted that increasing slip-load over a certain value
does not improve but deteriorate Rd , Ra and R f . For example,
increasing over 0.05 do not produce additional reduction of
Rd for N f = 1 in Fig. 12(a), and there exists an optimal value
of for N f = 1. These facts imply that there exists a specific
relationship between N f and optimal .
For economical design, a new integrated performance index,
J , in which the total number of installation floors normalized
to the total number of floors is included besides structural
response quantities, is defined as follows:
J = w1 Rd + w2 Ra + w3 R f + w4

Nf
N

(14)

where w1 , w2 , w3 and w4 are weightings on Rd , Ra , R f


and N f /N , respectively. This performance index has the same
design concept as the linear quadratic regulator is designed
by trade-off between the controlled response and control
force [18]. Fig. 15 represents the variation of J with respect
to N f for w1 = w2 = w3 = 2/7 and w4 = 1/7. Existence
of the optimal N f minimizing J can be identified for all three
structures represented in Fig. 15. This is because the friction
dampers added over an optimal quantity do not contribute to the
response reduction efficiently and just increase the construction
cost. Inspired from this result, an empirical equation on the
optimal number of installation floors is proposed. First, the
optimal number of installation floors for each 4, 8, 12, 16
and 20-storey example structure is calculated for various
and normalized to the total number of floors, N . Then,
these normalized optimal N f s are averaged and fitted to the
following equation, using least square criteria:
Nf
= 1.5652 0.4451 .
N

(15)

Fig. 16. Normalized optimal N f minimizing J .

This equation is plotted in Fig. 16 with the normalized


optimal N f s calculated for example structures. As shown
in this figure, the normalized optimal N f increases as
increases. Using the proposed Eq. (15), the optimal number of
installation floors, to which the total slip-load is distributed, can
be determined simply.
5. Comparison with an existing design method
In this section, friction dampers are designed using both the
proposed design method and the SSSA developed by Garcia
and Soong [14] with the same total slip-load for comparison.
The proposed design method is supposed to install friction
dampers on the half of the whole floors corresponding to the
upper 50% of the peak storey shear forces and distribute the
slip-load using Eq. (13). Accordingly, the normalized number
of installation floor, N f /N , is 0.5 and the corresponding
from Eq. (15) is 0.08, which is applied to both the proposed
design method and the SSSA. In the SSSA, the total slipload is subdivided into 16 equal slip-loads and assigned

939

S.-H. Lee et al. / Engineering Structures 30 (2008) 930940


Table 1
Comparison of the performance indices by the SSSA and the proposed design method
Earthquake

Structure

Method

Rd

Ra

Rf

N = 4, T = 0.4 s

SSSA
Proposed

0.514
0.505

0.610
0.622

0.708
0.631

2
2

0.595
0.574

SSSA

0.627

0.951

0.721

0.764

Proposed
SSSA
Proposed
SSSA
Proposed
SSSA

0.651
0.609
0.629
0.576
0.592
0.617

0.876
0.795
0.820
0.795
0.838
0.843

0.674
0.875
0.756
0.705
0.695
0.767

4
4
6
8
8
10

0.700
0.699
0.701
0.665
0.679
0.708

Proposed

0.611

0.754

0.770

10

0.682

SSSA

0.309

0.476

0.443

0.422

Proposed

0.289

0.437

0.410

SSSA
Proposed

0.470
0.503

0.706
0.637

0.626
0.628

5
4

0.396
0.604

SSSA

0.548

0.898

0.645

0.693

Proposed
SSSA

0.581
0.443

0.876
0.672

0.644
0.570

6
11

0.672
0.580

Proposed
SSSA
Proposed

0.440
0.465
0.522

0.646
0.660
0.686

0.601
0.556
0.591

8
13
10

0.553
0.573
0.586

SSSA
Proposed
SSSA

0.445
0.484
0.491

0.672
0.712
0.871

0.596
0.604
0.624

2
2
5

0.561
0.586
0.657
0.649
0.594

N = 8, T = 0.8 s
Northridge

N = 12, T = 1.2 s
N = 16, T = 1.6 s
N = 20, T = 2.0 s

N = 4, T = 0.4 s
N = 8, T = 0.8 s
Kobe

N = 12, T = 1.2 s
N = 16, T = 1.6 s
N = 20, T = 2.0 s
N = 4, T = 0.4 s
N = 8, T = 0.8 s

El Centro

N = 12, T = 1.2 s
N = 16, T = 1.6 s
N = 20, T = 2.0 s
N = 4, T = 0.4 s
N = 8, T = 0.8 s

Loma Prieta

N = 12, T = 1.2 s
N = 16, T = 1.6 s
N = 20, T = 2.0 s

Nf

0.577

Proposed

0.520

0.869

0.631

SSSA
Proposed
SSSA
Proposed
SSSA
Proposed

0.460
0.479
0.431
0.471
0.551
0.563

0.700
0.665
0.581
0.678
0.535
0.557

0.543
0.548
0.521
0.541
0.676
0.674

9
6
11
8
11
10

SSSA
Proposed
SSSA

0.276
0.316
0.426

0.395
0.451
0.737

0.430
0.440
0.551

2
2
6

0.386
0.416
0.597

Proposed
SSSA

0.457
0.217

0.669
0.328

0.580
0.365

4
6

0.559
0.331

Proposed
SSSA
Proposed
SSSA
Proposed

0.234
0.392
0.433
0.474
0.482

0.317
0.599
0.711
0.556
0.479

0.356
0.560
0.551
0.605
0.615

6
11
8
9
10

0.331
0.542
0.556
0.531

sequentially to the floor with the maximum peak story shear


force obtained from repeated nonlinear time history analysis.
Ground accelerations of Northridge, Kobe, El Centro, and
Loma Prieta earthquake are applied and the brace is assumed
to be rigid for the same reason as mentioned in Section 4.
Table 1 compares the performance indices obtained from
the SSSA and the proposed design method. It can be seen
that the response levels obtained by both design methods
are almost similar. However, the proposed design method
conducted nonlinear time history analysis only two times while
the SSSA 17 times. In particular, the proposed design method
installs the friction dampers on fewer or equal number of

0.555
0.536
0.554
0.582
0.584

0.522

floors than the SSSA except for the 12-storey structure excited
by Northridge earthquake and 20-storey structure excited by
Loma Prieta earthquake. This is because the proposed design
method minimizes the performance index taking the number of
installation floor into account directly. This can be confirmed
by the fact that the proposed design method yields smaller J
than the SSSA for 12 cases among 20 represented in Table 1.
6. Conclusion
In this study, a seismic design methodology of the friction
dampers based on the story shear forces of an elastic building

940

S.-H. Lee et al. / Engineering Structures 30 (2008) 930940

structure is proposed. To examine the effect of the brace


stiffness, optimal stiffness ratios between the brace and primary
structure are found for SDOF systems of various natural
periods. In this procedure, it is observed that, if the slipload of the friction damper is normalized with respect to the
storey shear force of the original primary structure, optimal
stiffness ratios for the displacement and storey shear force
become similar and, hence, design practice is facilitated. Also,
it is found that the reasonable value of normalized slip load
is lower than 0.3 because increasing slip load higher than 0.3
causes the shear force and the fraction of the earthquake input
energy taken by the primary structure to increase considerably.
Regarding the slip-load of the friction damper, a small amount
of total slip-load works effectively when concentrated to a
single story, while large amount of total slip-load works
effectively when distributed to different stories. For the purpose
of economical design, an empirical equation providing the
optimal number of installation floors for a given total slipload is proposed by regression analysis. Finally, the proposed
design procedure achieves similar level of response reduction
with much fewer nonlinear time history analyses and occupies
smaller number of installation floor compared to the existing
design method, SSSA. However, it should be mentioned that
a designer should allocate the friction damper with caution
under the current design code which restricts the vertically
incomplete distribution possible both in the proposed method
and the SSSA.
Acknowledgement
The work presented in this paper was partially supported by
both the National Research Laboratory Programme (Project No.
M1-0203-00-0068) from the Ministry of Science.
References
[1] Pall AS, Marsh C. Response of friction damped braced frames. Journal of
Structural Engineering 1982;108(9):131323.
[2] Constantinou MC, Mokha A, Reinhorn AM. Teflon bearings in base
isolation. II: Modeling. Journal of Structural Engineering 1990;116(2):
45574.

[3] Li C, Reinhorn AM. Experimental and analytical investigation of seismic


retrofit of structures with supplemental damping: Part II-Friction devices.
Technical report NCEER-95-0009. Buffalo (NY): State University of New
York at Buffalo; 1995.
[4] Grigorian CE, Yang TS, Popov EP. Slotted bolted connection energy
dissipaters. Report of national science foundation. Berkeley: University
of California; 1992.
[5] Mualla IH, Belev B. Performance of steel frames with a new friction
damper device under earthquake excitation. Engineering Structures 2002;
24(3):36571.
[6] Cho CG, Kwon M. Development and modeling of a frictional wall damper
and its applications in reinforced concrete frame structures. Earthquake
Engineering Structures Dynamics 2004;33(7):82138.
[7] Dyke SJ, Spencer Jr BF, Sain MK, Carlson JD. Modeling and control
of magnetorheological dampers for seismic response reduction. Smart
Materials and Structures 1996;5(5):56575.
[8] Whittaker AS, Constantinou MC, Ramirez OM, Johnson MW,
Chrysostomou CZ. Equivalent lateral force and modal analysis procedure
of the 2000 NEHRP provisions for buildings with damping systems.
Earthquake Spectra 2003;19(4):95980.
[9] Filiatrault A, Cherry S. Seismic design spectra for friction-damped
structures. Journal of Structural Engineering 1990;116(5):33455.
[10] Fu Y, Cherry S. Design of friction damped structures using lateral force
procedure. Earthquake Engineering and Structural Dynamics 2000;29(7):
9891010.
[11] Ciampi V, De Angelis M, Paolacci F. Design of yielding or frictionbased dissipative bracing for seismic protection of buildings. Engineering
Structures 1995;17(5):38191.
[12] Kim JK, Choi H. Response modification factors of chevron-braced
frames. Engineering Structures 2005;27(2):285300.
[13] Moreschi LM, Singh MP. Design of yielding metallic and friction
dampers for optimal seismic performance. Earthquake Engineering &
Structural Dynamics 2003;32(8):1291311.
[14] Garcia DL, Soong TT. Efficiency of a simple approach to damper
allocation in MDOF structures. Journal of Structures Control 2002;9(1):
1930.
[15] Shukla AK, Datta TK. Optimal use of viscoelastic dampers in building
frames for seismic force. Journal of Structural Engineering 1999;125(4):
4019.
[16] Uniform Building Code. International conference of building officials.
Whittier, California; 1997.
[17] Chopra AK. Dynamics of structures: Theory and applications to
earthquake engineering. Upper Saddle River (NJ, USA): Prentice-Hall;
1995.
[18] Soong TT, Chen WF. Active structural control: Theory and practice.
Longman Scientific & Technical; 1990.

You might also like