Engineering Structures 23 (2001) 745–755

www.elsevier.com/locate/engstruct

Rehabilitation of a reinforced concrete frame using eccentric steel

bracing
a,* b
A. Ghobarah , H. Abou Elfath
a
Department of Civil Engineering, McMaster University, Hamilton, Ontario, Canada L8S 4L7
b
Department of Civil Engineering, Alexandria University, Alexandria, Egypt

Received 3 March 2000; received in revised form 20 September 2000; accepted 26 September 2000

Abstract

The seismic performance of a low-rise nonductile reinforced concrete (RC) building rehabilitated using eccentric steel bracing
is investigated. A three-story office building was analyzed using various ground motion records. The effectiveness of the eccentric
steel bracing in rehabilitating the building was examined. The effect of distributing the steel bracing over the height of the RC
frame on the seismic performance of the rehabilitated building was studied. The behavior of the nonductile RC frame members is
represented by a beam–column element capable of modeling the strength deterioration and the effect of the axial force on the yield
moment and the deformation capacities at peak strength of the members. The link behavior is modeled using tri-linear moment
and shear force representations. The performance of the building is evaluated in terms of story drifts and damage indices.  2001
Elsevier Science Ltd. All rights reserved.

Keywords: Reinforced concrete; Frame; Nonductile; Rehabilitation; Eccentric steel brace; Link model; Seismic response

1. Introduction
Existing reinforced concrete (RC) frame buildings
with nonductile detailing represent a considerable hazard
during earthquakes. This type of building suffered severe
damage and was responsible for most of the loss of life
during recent seismic events such as the 1985 Mexico
and 1999 Kocaeli (Turkey) earthquakes. The nonductile
behavior of RC frames is due to inadequate transverse
reinforcement in beams, columns and joints; bond slip
of beam bottom reinforcement at the joint; and lack of
confinement of the column lap splice area. It is important
to develop effective and economic seismic rehabilitation
systems for nonductile RC buildings before an earth-
quake occurs, since the potential for damage and loss of
life during future seismic events is unacceptably high.
Different rehabilitation systems have been developed
to upgrade the seismic performance of existing undam-
aged structures before being subjected to an earthquake.
* Corresponding author. Tel.: +1-905-525-9140; fax: +1-905-529-
9688.
E-mail address: ghobara@mcmaster.ca (A. Ghobarah).
The two main approaches are: (a) add new structural
elements such as structural walls or steel bracing; or (b)
the selective strengthening of deficient structural
elements such as the use of concrete and steel jackets
and fiber reinforced polymers, FRP wrapping.
The use of steel bracing systems for seismic rehabili-
tation of RC frames offers some advantages such as: (a)
the ability to accommodate openings; (b) minimal added
weight to the structure; and (c) external steel systems
have been constructed with minimum disruption to the
function of the building and its occupants. Concentric
steel bracing systems are used for the rehabilitation of
nonductile RC buildings in Mexico and Japan [1,2]. The
use of eccentric steel bracing in the rehabilitation of RC
structures has lagged behind concentric steel bracing
applications due to the lack of sufficient research and
information about the design, modeling and behavior of
the combined concrete and steel system. To facilitate the
application of eccentric bracing in rehabilitation, further
research is needed in several areas such as testing of the
RC beam–steel link connection details and design as
well as the development and implementation of link
element models in analysis software.
In eccentrically braced frames (EBFs), forces are

0141-0296/01/$ - see front matter  2001 Elsevier Science Ltd. All rights reserved.
PII: S 0 1 4 1 - 0 2 9 6 ( 0 0 ) 0 0 1 0 0 - 0
746 A. Ghobarah, H. Abou Elfath / Engineering Structures 23 (2001) 745–755
transferred to the brace members through bending and
shear forces developed in the ductile steel link. The link
is designed to act as a fuse by yielding and dissipating
energy while preventing buckling of the brace members.
Well-designed links provide a stable source of energy
dissipation. Different brace patterns are used in eccentri-
cally braced steel frames. Examples of these patterns
include V-bracing, K-bracing, X-bracing and Y-bracing
as shown in Fig. 1. Most of these patterns utilize short
beam segments as active links. The analysis, design and
performance of eccentrically braced steel frames have
been studied by several researchers [3–9]. In RC frames,
the concrete beams are incapable of performing as a duc-
tile link for the steel bracing system that is inserted in
the frame bays. A vertical steel shear link may be intro-
duced by the Y-bracing pattern shown in Fig. 1(d) [10].
In this case, the vertical shear link can be attached to
the beam of the RC frame. Special consideration should
be given to the connection between the vertical shear
link and the RC beam. This connection should have suf-
ficient capacity to ensure effective transmission of forces
when subjected to seismic loads. Fig. 2 shows details of
a proposed link connection of an eccentric bracing sys-
tem inserted in the bays of an RC frame. The link con-
nection is located at mid-span of the RC beam. The link
is connected to steel plates which are anchored to the
RC beam. The force in the brace is transmitted to the
shear link using an end plate. The proposed link connec-
tion has not been subjected to extensive testing but is
expected to provide the fixation to the link end as has
been modeled in the analysis.
The objective of this study is to evaluate the effective-
Fig. 1. Various types of eccentrically braced steel frames. (a) V-bracing; (b) K-bracing; (c) X-bracing and (d) Y-bracing.
ness of rehabilitating a three-story RC office building
using an eccentric steel bracing system inserted in the
bays of the exterior frames. The seismic performance of
the rehabilitated RC frames is investigated using nonlin-
ear static pushover analysis and dynamic time–history
analysis. Ground motion records with different charac-
teristics were used in the dynamic analysis. The effect
of the eccentric steel brace distribution over the height
of the RC frames on the seismic performance of the
rehabilitated building, is evaluated.
2. Design of shear links and brace members
The vertical shear link shown in Fig. 1(d) is assumed
to act as a cantilever. The brace members are assumed
to be pin-connected to the vertical link while the link
itself is considered fixed to the RC beam. The brace
members will provide negligible constraint to the link
end against rotation. The critical length of a vertical can-
tilever shear link is half the length of a steel link that
has fixed connections at both ends with reverse curvature
and equal end moments.
Mu
ecrit⫽ (1)
Vu
where ecrit is the maximum cantilever shear link length,
Mu and Vu are the ultimate end moment and shear force
for a link with a well stiffened web. Based on experi-
mental data [4], the values of Mu and Vu are approxi-
mated as:
Mu⫽1.2Mp, Vu⫽1.5Vp (2)
 A. Ghobarah, H. Abou Elfath / Engineering Structures 23 (2001) 745–755 747

where bf and tf are the width and thickness of the flange

and tw is the web thickness of a wide flange section link.
The maximum link deformation angle gmax that can be
achieved by the shear link is dependent on the link
detailing. The link deformation angle is defined as the
link lateral displacement over the link length for both
the single and double curvature cases. Shear links with
closely spaced web stiffeners exhibit relatively large ulti-
mate deformation angles under the effect of cyclic load-
ing. The ultimate link deformation angle is defined as
the maximum deformation angle developed by the link
before the occurrence of considerable strength deterio-
ration due to severe flange and web buckling of the link.
It was found that the ultimate link deformation angle for
well stiffened shear links may approach 0.1 rad [11].
Steel brace members in EBFs should be designed to
ensure that they will behave elastically when subjected
to an earthquake loading. The brace member should be
designed as a compression member with its axial load
capacity depending on the plastic strength of the steel
link. An additional factor of safety of at least 1.5 should
be applied to this axial load capacity to ensure that the
brace does not buckle due to the link strain hardening
[12].

3. Deformation capacity of EBFs

An important aspect in the design of EBFs is the con-

sideration of the deformation demands in the steel links.
Fig. 3 shows a plastic mechanism of an eccentrically
braced reinforced concrete frame provided with a verti-
cal steel link. In the figure, small black circles indicate
plastic flexural hinges and the cross-hatched line indi-
cates a plastic shear hinge. The story displacement is
denoted ⌬, H is the story height, L is the frame span,
e is the link length, q is the story drift angle (q=⌬/H)
and g is the link deformation angle. For the RC frame
shown in Fig. 3, the deformation angle of the link can
be estimated approximately as g=q(H/e). In EBFs, e is
Fig. 2. Connection details of a vertical steel link. (a) Elevation, and normally much smaller than L and H, and therefore sev-
(b) Section S-S. ere deformation demands are placed on the link.
For the case of the reinforced concrete frame shown
where Mp and Vp are the plastic moment and shear in Fig. 3, the maximum story drift angle, qmax, is calcu-
capacities of the link section. Substituting into Eq. (1) lated as:
yields: ecritgall
qmax⫽ (5)
Mp H
ecrit⫽0.8 (3)
Vp where gall is the allowable link deformation angle and
ecrit is the maximum shear link length calculated using
Based on experimental data [3], the formula for calcu- equations (3) or (4). For example, if gall=0.1 rad,
lating the length of a cantilever link to ensure that the H=4.0 m and e=0.50 m, then the maximum allowable
link yields primarily in shear is: story drift angle q, is equal to 0.0125 rad. Fig. 3(b) indi-
cates that both the steel bracing system and the existing
2bftf RC frame have to deform laterally to the same story dis-
ecrit⫽ (4)
tw placement level. The deformation of the steel bracing
748 A. Ghobarah, H. Abou Elfath / Engineering Structures 23 (2001) 745–755

Fig. 3. Plastic hinge mechanism of a reinforced concrete frame rehabilitated using a vertical steel link. (a) Original frame, and (b) Plastic mech-
anism.

system results mainly from the link yielding while the to represent the inelastic shear behavior of the link web
deformation of the RC frame is developed mainly by the represented by the multilinear function shown in Fig. 5.
formation of the plastic hinges in the frame members. In the current study, the vertical steel link was rep-
The inelastic hinging system shown in Fig. 3(b) rep- resented by a linear cantilever element with only two
resents one possible failure mechanism. Alternatively, inelastic rotational and translational springs at the fixed
the hinges shown at the top of the column may occur in link end. Each of the springs was designed to produce
the RC beam near the beam–column joint depending on directly the relationships shown in Figs. 4 and 5. One
the relative flexural strengths of the beams and columns. of the developed springs represents the moment–rotation
Other failure mechanisms may be developed in the RC relationship while the other represents the shear force–
frame. shear displacement relationship. The values of My and
Vy are considered equal to Mp and 0.9Vp, respectively
[13]. The moment–rotation and the shear force–lateral
4. The link model displacement relationships of the steel link shown in
Figs. 4 and 5 are given as [5,13]:
Steel links are subjected to high levels of shear forces
Vy1=Vy My1=My
and bending moments in the active link regions. In the
analysis of the performance of links, elastic and inelastic Vy2=1.06Vy My2=1.03My (6)
deformations of both the shear and flexural behaviors Vy3=1.12Vy My3=1.06My
have to be taken into consideration. Few researchers
attempted to develop link models for the dynamic inelas-
tic analysis of EBFs [5,13]. Ramadan and Ghobarah [13] The values of the stiffness are:
modeled the link as a linear beam element with six non- K2v=0.03K1v K2M=0.03K1M
linear rotational and translational springs at each end.
Three rotational bilinear springs were used to represent K3v=0.015K1v K3M=0.015K1M (7)
the flexural inelastic behavior of the plastic hinge at the K4v=0.002K1v K4M=0.002K1M
link end represented by the multilinear function shown
in Fig. 4. Three translational bilinear springs were used
The values of K1M and K1V can be calculated as:

Fig. 4. Moment–rotation relationship of a steel link. Fig. 5. Shear force–shear deformation relationship of a steel link.
 A. Ghobarah, H. Abou Elfath / Engineering Structures 23 (2001) 745–755
K1M⫽3EI/e K1v⫽GAweb/e
where, E is Young’s modulus of steel, I is the moment
of inertia of the link cross section, G is the modulus of
rigidity of steel, and Aweb is the area of the web of the
link section. Under the effect of cyclic loading, moment
yielding obeys the kinematic hardening rule while shear
yielding follows a combination of both isotropic and
kinematic hardening [13]. For the shear spring, a special
function was used to account for the upper bound of the
shear capacity [13]. The function determines the
maximum attainable shear force capacity after a certain
amount of plastic action. This function has the shape:
Vyx⫽Vy[1⫹0.8(1⫺e−10S)]
where Vy is the initial shear yield strength and S is the
accumulated strain in the shear spring.
The shear link model was implemented in the
DRAIN-2DX computer program [14]. The force–defor-
mation relationships of the springs were determined
using the forces and stiffness levels presented in equa-
tions (6) and (7).
5. Rehabilitation cases
The selected structure is a three-story office building
that was designed for gravity loads according to the 1963
ACI code [15]. The concrete strength is 21 MPa and the
steel yield strength is 300 MPa. The design live load for
the building was taken as 2.4 kN/m2. Typical floor plan
and elevation of the building are shown in Fig. 6. The
exterior columns are 300×300 mm reinforced using 4–
19 mm diameter bars and the interior columns are
400×400 mm reinforced using 8–19 mm diameter bars.
The exterior and interior typical beam sections are as
shown in Fig. 7. The three-story building was modeled
as a series of planar frames connected at each floor level
by rigid diaphragms. Therefore, only two-dimensional
analysis was performed. The beams and columns were
modeled using a beam–column model that is capable of
representing the nonductile characteristics of the frame
members. Stiffness degradation, pinching and strength
softening, were explicitly taken into account in the
model. The model includes the effect of the axial forces
on the yield moments and the deformation capacities at
peak strength in RC members. The ultimate base shear
ratio of the existing building determined using the push-
over analysis is 0.15. For simplicity, the rehabilitation
of the frame will be aimed at increasing the lateral load
resistance capacity to a level of 0.23W, where W is the
weight of the structure.
Three rehabilitation cases were designed for the three-
story building described. In the first rehabilitation case
V1, a concentric inverted-V-brace was provided to only
three bays of the exterior frames of the building as
749
(8)
(9)
Fig. 6. Dimensions of the three-story office building: (a) typical
frame elevation; and (b) typical floor plan.
shown in Fig. 8(a). All the steel brace members have the
same length, cross section and material properties. The
brace members were selected as round hollow sections
(HSS 114×8). The effective length factors for concentric
inverted-V-bracing were assumed to be 1.0 for in-plane
and out-of-plane buckling. The Young’s modulus is
E=200,000 MPa and the steel yield stress is fy=350 MPa.
The calculated brace properties are: radius of gyration,
r=37.7 mm, KL/r=109, Pc/Py=2.27, Pr/Pc=0.33, where
Py is the brace yield load, Pc is the initial buckling load,
and Pr is the residual buckling load. The initial buckling
load Pc is calculated using the code formula [16]. The
residual buckling Pr is estimated using the formula pro-
posed by Lee and Goel [17] for calculating the residual
buckling load for tubular sections. In this rehabilitation
case, the lateral load capacity of the rehabilitated build-
ing is 1.7 times the load carrying capacity of the existing
three-story building. The brace was modeled using the
Jain and Goel [18] brace model which was implemented
into the DRAIN-2DX computer program. In the analysis,
the loading of the brace members was prevented during
the application of gravity loads.
In the second rehabilitation case, E1, vertical steel
links were provided to the inverted-V-bracing systems
as shown in Fig. 8(b). The properties and plastic
capacities of the vertical steel links are summarized in
Table 1. The stiffness and the plastic capacities of the
750 A. Ghobarah, H. Abou Elfath / Engineering Structures 23 (2001) 745–755

Fig. 7. Details of the reinforced concrete beams and columns: (a) exterior column; (b) interior column; (c) exterior beam; and (d) interior beam.

steel links in shear and flexure were calculated using

Young’s modulus, E=200,000 MPa and steel yield
stress, fy=300 MPa. The vertical steel link section
(W200×46) and its length of 0.60 m were selected to
satisfy Eq. (4) and to obtain a link shear strength lower
than 2/3 the force level that causes brace buckling. The
links are assumed to be properly stiffened to produce
maximum levels of ultimate rotations. For the eccentric
bracing rehabilitation case, the lateral load capacity of
the rehabilitated building is 1.6 times the load carrying
capacity of the existing three-story building.
An eccentric brace rehabilitation case E2 shown in
Fig. 8(c), was designed in order to investigate the effect
of the distribution of the brace strengths over the build-
ing height. The brace members and shear link properties
are exactly the same as those used before in the rehabili-
tation case E1. For this case, the lateral load capacity of
the rehabilitated building is 1.9 times the load carrying
capacity of the existing three-story building.
The periods of free vibration of the first two modes
of the existing and rehabilitated frames V1, E1 and E2
are listed in Table 2. The rehabilitation cases with eccen-
tric bracing have relatively lower stiffness and therefore
longer fundamental period than the rehabilitation case
with concentric bracing. The rehabilitation case E2 has
slightly shorter fundamental period than the rehabili-
tation case E1.

6. Pushover analysis

A pushover analysis was conducted using an inverted

triangular lateral load distribution to evaluate the stiff-
ness and strength characteristics of the three rehabili-
Fig. 8. Rehabilitation cases for the nonductile three-story building: tation cases. Fig. 9 shows the lateral load–roof drift of
(a) case V1; (b) case E1; and (c) case E2. the existing and the rehabilitated buildings. The ratios
 A. Ghobarah, H. Abou Elfath / Engineering Structures 23 (2001) 745–755 751

Table 1
Properties of the vertical links used in rehabilitating the three story building

Section My=Mp (kN·m) Vy=0.9Vp (kN) e (m) K1M (kN·m) K1v (kN/m)

W200×46 148.8 203 0.6 4.55×104 167,100

Table 2
Periods of free vibration of existing frame and the rehabilitation cases

Case First period T1 (s) Second period T2 (s)

Existing 0.99 0.32

V1 0.46 0.16
E1 0.59 0.20
E2 0.57 0.22

Fig. 10. Variation of link deformation angle in radians with roof drift
for rehabilitation case E1.

7. Seismic response

The seismic behavior of the existing frame and the

Fig. 9. Performance curves of the various rehabilitation cases from
the pushover analysis.
between the initial stiffness of the rehabilitated cases V1,
E1 and E2 to that of the existing building are 4.6, 2.8 and
3.0, respectively. Eccentric bracing rehabilitation caused
less increase in the building stiffness in comparison to
the concentric bracing. The rehabilitation case E2
resulted in higher lateral load carrying capacity than the
concentric case V1 and the eccentric bracing case E1.
In Fig. 9, the lateral load–roof drift relationship of the
rehabilitation case V1 exhibited a drop in strength at
about 0.35% roof drift ratio due to buckling of the brace
members. Buckling of a brace member will lead to a
corresponding drop in the brace load capacity in com-
pression.
The variation of the link deformation angle with the
roof drift for the rehabilitation case E1 is shown in Fig.
10. The links in the first story reach the maximum allow-
able deformation angle, g=0.1 rad at relatively low roof
drift ratio of approximately 0.80%. This represents a
limitation for the use of the selected eccentric steel brac-
ing system for the seismic rehabilitation of the RC
frame.
rehabilitation cases were studied when subjected to
earthquake ground motion. The purpose of the study is
to evaluate the effect of the change in the stiffness and
in the energy dissipation capacity of the rehabilitation
case on the seismic performance of structure. The frames
were subjected to scaled versions of the twelve ground
motion records of different characteristics listed in Table
3. The records were selected to represent wide ranges of
duration and frequency content because of the wide
range of frequencies of free vibration from low funda-
mental frequency of the existing moment resisting frame
to the higher frequencies of the braced frame cases, as
given in Table 2. The peak ground acceleration (PGA)
as a ratio of g is denoted A and the peak ground velocity
in m/s is V. The duration of the ground motion was cal-
culated using the method proposed by Trifunac and
Brady [19]. The acceleration time history records were
scaled to various PGA values. The scaling procedure of
the earthquake records is expected to affect the results
of the analysis. However, this effect is expected to be
well within the levels of standard deviation of the results.
The seismic performance of the existing building and
the two rehabilitation cases were compared in terms of
deformations and damage indices. The damage index
used in this investigation is based on the change in initial
stiffness between two pushover analyses conducted on
the structure before and after the application of the
ground motion. Details of the damage index calculations
are presented elsewhere [20]. The dynamic analysis of
the building was performed using a time step increment
752 A. Ghobarah, H. Abou Elfath / Engineering Structures 23 (2001) 745–755

Table 3
Selected ground motions

V Dur.
No. Earthquake Date Site Comp. Soil type A(g) A/V
(m/s) (s)

1 Parkfield, California 27 June 1966 Temblor, No.2 N65W Rock 0.27 0.15 1.86 5.62
2 Nahanni, Canada 23 Dec. 1985 Site 1, Iverson N10E Rock 1.10 0.46 2.38 7.92
3 Imperial Valley, California 18 May 1940 El Centro S00E Stiff 0.35 0.33 1.04 24.42
4 Kern County, California 21 July 1952 Taft Lincoln School Tunnel S69E Rock 0.18 0.18 1.01 28.86
5 San Fernando, California 9 Feb. 1971 Hollywood Storage P.E. Lot N90E Stiff 0.21 0.21 1.00 13.22
6 San Fernando, California 9 Feb. 1971 234 Figueroa St, L.A. N37E Stiff 0.20 0.17 1.19 11.28
7 Monte Negro, Yugoslavia 15 Apr. 1979 Albatros Hotel, Ulcinj N00E Rock 0.17 0.19 0.88 12.22
8 Long Beach, California 10 Mar. 1933 Subway Terminal, L.A. N51W Rock 0.10 0.24 0.41 23.98
9 Lower California 30 Dec. 1934 El Centro S00W Stiff 0.16 0.21 0.77 21.10
10 San Fernando, California 9 Feb. 1971 2500 Wilshire Blvd., L.A. N61W Stiff 0.10 0.19 0.52 11.66
11 Near E. coast of Honshu, Japan 16 May 1968 Muroran Harbor N00E Stiff 0.23 0.33 0.68 33.38
12 Mexico 19 Sep. 1985 Zihuatenejo, Guerrero Array S00E Rock 0.10 0.16 0.65 19.74

of 0.005 s and Rayleigh damping which was defined to Table 5

achieve 2.0% viscous damping in the first two modes Link deformation angle g (radians) in the rehabilitation case E2
of vibration. Earthquake
Tables 4 and 5 summarize the maximum link defor- PGA (g) 1st story 2nd story 3rd story gmax/qmax
No.a
mation angles of the eccentric bracing cases E1 and E2,
reached due to the application of the ground motion rec- 1 1.80 0.094 0.048 0.080 5.61
ords. The levels of PGA summarized in the table cause 2 2.20 0.076 0.093 0.094 5.65
3 0.72 0.058 0.085 0.035 5.67
the maximum link deformation angles to be close to 0.1 4 0.73 0.092 0.100 0.035 5.67
rad. These selected PGA levels are the highest levels 5 0.68 0.099 0.091 0.053 5.71
considered in this study. The ratio between the 6 1.00 0.078 0.086 0.080 5.67
maximum link deformation angle g and the story drift 7 0.60 0.093 0.091 0.070 5.71
angle q is calculated and is presented in the table. This 8 0.35 0.105 0.058 0.028 5.74
9 0.54 0.082 0.077 0.036 5.69
ratio was found to be close to the ratio calculated from 10 0.42 0.094 0.079 0.026 5.69
Eq. (5) (g/q=H/e=6.0). This indicates that this ratio can 11 0.50 0.051 0.094 0.040 5.69
be approximately considered independent of the loading 12 0.40 0.092 0.099 0.047 5.7
condition. The difference between the g/q ratio calcu- a
lated from Eq. (5) and the values obtained from the The earthquakes corresponding to these numbers are listed in
Table 3.
dynamic analysis is because the axial deformations of
the brace members and the RC frame members were
neglected in the derivation of Eq. (5). The mean of the seismic performance parameters of
the existing frame and the rehabilitation cases are calcu-
Table 4 lated at multiples of 0.10g increments in the PGA for
Link deformation angle g (radians) in the rehabilitation case E1 all the earthquake records. The seismic performance
parameters are related to the PGA level as shown in Figs.
Earthquake 11 and 12. The figures represent the relationships of the
PGA (g) 1st story 2nd story 3rd story gmax/qmax
No.a PGA level versus the mean story drift and the damage
1 1.80 0.094 0.066 0.025 5.53 index Dg, respectively. The maximum standard devi-
2 1.75 0.101 0.051 0.018 5.63 ations in story drift ratios and damage index values of
3 0.48 0.107 0.035 0.009 5.73 the braced frames at PGA level of 0.30g due to the appli-
4 0.50 0.088 0.044 0.009 5.70 cation of 12 selected earthquakes, are 0.70 and 0.22,
5 0.50 0.1 0.034 0.012 5.71 respectively. The results shown in the figures indicate
6 0.65 0.094 0.032 0.017 5.69
7 0.48 0.11 0.036 0.009 5.74 that on average, the eccentric brace rehabilitation
8 0.28 0.098 0.03 0.01 5.71 resulted in a significant improvement in the seismic per-
9 0.40 0.106 0.033 0.008 5.72 formance by comparison to the response of the existing
10 0.32 0.106 0.039 0.007 5.75 building. The mean levels of deformation and damage
11 0.50 0.113 0.05 0.009 5.74 indices were significantly lower in the eccentric brace
12 0.28 0.105 0.032 0.01 5.72
cases E1 and E2 than in the concentric bracing case V1.
a
The earthquakes corresponding to these numbers are listed in For example, at PGA=0.50g, the ratios between the per-
Table 3. formance parameters of case V1 to those of case E1 were
 A. Ghobarah, H. Abou Elfath / Engineering Structures 23 (2001) 745–755
Fig. 11. Relationship between the PGA and the mean story drift for
the rehabilitation cases.
Fig. 12. Relationship between the PGA and the mean damage index
for the rehabilitation cases.
1.23 and 1.20 for the story drift and the damage index
Dg, respectively. The improved performance of the
eccentric bracing case E1 can be attributed to several
reasons among which are the enhancement in the energy
dissipation capacity of the eccentric bracing as a result
of the stable inelastic behavior of the steel links.
Fig. 13 shows examples of the distribution of damage
due to the inelastic response of the existing frame and
the rehabilitation cases. The damage distribution shown
in Fig. 13 was due to the application of the arbitrarily
selected Monte Negro ground motion. The numbers in
the figures show the formation sequence of some selec-
ted plastic hinges. Numbers in italic font are for the col-
umn hinges while numbers in regular font are for the
beam hinges. In the rehabilitation case V1
(PGA=0.475g), hinging starts in the beams of the first
and second floors due to pullout of the bottom reinforce-
ment and yielding of the top reinforcement. With the
increase in the earthquake intensity, damage spreads to
the columns of the first and second stories. At the final
stage of response, all the first story columns experienced
splice failure at the bottom ends and yielding of the
reinforcement at the top ends. No damage was observed
in the columns and beams of the third story. Brace buck-
ling was observed in all the steel bracing members of
753
Fig. 13. Hinge type and location near failure of the existing building
and rehabilitation cases due to the application of scaled Monte Negro
earthquake: (a) existing building, PGA=0.45g; (b) case V1,
PGA=0.475g; (c) case E1, PGA=0.475g; and (d) case E2,
PGA=0.60g.
the first and second stories and in three brace members
of the third story.
Damage to the RC members of the rehabilitation case
E1 due to the application of the Monte Negro ground
motion (PGA=0.475g) was similar to that of the rehabili-
tation case V1. Hinging starts in the beams of the first
and second floors and spreads to the columns of the first
and second stories with the increase in the earthquake
intensity. The columns and beams of the third story
remained undamaged. All the steel brace members of the
rehabilitation case E1 behaved elastically. Yielding of
steel shear links occurred in all the building stories.
8. Effect of brace distribution over the height
The effect of the steel link distribution along the
height of the RC building was investigated by comparing
the seismic response of the two rehabilitation cases E1
and E2 with eccentric bracing. The two rehabilitation
cases have different distributions of eccentric bracing
over the RC building height. Figs. 11 and 12 indicate
that in general, the eccentric bracing case E2 achieved
significantly lower deformation and damage than those
of the rehabilitation case E1. The enhanced seismic per-
754 A. Ghobarah, H. Abou Elfath / Engineering Structures 23 (2001) 745–755
formance of the bracing case E2 is mainly because of
the improvement in the plastic mechanism of the build-
ing. The lower deformation of the rehabilitation case E2
is because of the uniformity of the story displacement
over the building height.
The levels of axial and shear forces in the building
members of rehabilitation cases E1 and E2 due to the
application of the Monte Negro earthquake indicate that
the maximum axial force levels developed in the
rehabilitation case E2 at PGA=0.6g are close to the lev-
els developed in the rehabilitation case E1 at
PGA=0.475g. No tensile axial forces were developed in
the columns. Axial compression and tension forces were
developed in the beams of the nonductile frame. All the
column axial forces are below the axial load at the bal-
anced condition of the column cross section. The
maximum shear force levels developed in the frame col-
umns and beams of the rehabilitation cases E1 and E2
were found to be close. The variation in the maximum
shear forces calculated for each of the building stories
of the rehabilitation case E2 was small due to the uni-
formity of the story displacement along the building
height. The shear capacities of the concrete columns and
beams in the existing building were found to be higher
than the shear demands developed in the frame members
of the rehabilitation cases E1 and E2 due to the appli-
cation of the Monte Negro earthquake record scaled to
PGA of 0.475g and 0.6g, respectively.
The damage distribution for case E2 shown in Fig.
13(d) indicates that hinging starts in the beams of the
second floor and in the columns of the third story. With
the increase in the earthquake intensity, damage spread
to the beams of the first floor and to the second story
columns. At the final stage of response damage spread
to the columns of the first story. All the first and third
story columns experienced splice failure at the bottom
ends and yielding of the reinforcement at the top ends.
No damage was observed in the beams of the third floor.
All the steel brace members of the rehabilitation case E2
behaved elastically and all the steel shear links experi-
enced shear yielding.
The number of plastic hinges in RC members of the
rehabilitation case E2 was larger than that of the
rehabilitation case E1. This can be attributed to the
improvement in the plastic mechanism of the rehabili-
tation case E2 which led to the participation of more RC
elements in the building inelastic response.
9. Conclusions
A model for the vertical shear link in EBFs was
implemented in the DRAIN-2DX computer program and
used to study the performance of a nonductile building
rehabilitated using eccentric steel bracing. One concen-
tric and two eccentric steel bracing rehabilitation cases
were analyzed. Eccentric bracing cases exhibited lower
deformation and damage when subjected to earthquake
ground motions as compared to the behavior of the con-
centric bracing case.
The analysis performed on the eccentric brace
rehabilitation cases indicated that the link deformation
angle is an important parameter. In order to limit the
link deformation angle below the allowable level gall, the
story drift angle q (story displacement/story height) of
the rehabilitated building should not exceed
gall×(link length/story height). This represents a limi-
tation on the deformation of the existing building when
eccentric steel bracing is used in its seismic rehabili-
tation.
The improvement in the seismic performance of the
nonductile building when using well designed eccentric
bracing rehabilitation is expected to be greater than that
of the concentric bracing as long as the building defor-
mation remains below the limits that cause the maximum
allowable link deformation angle. The work involved in
the installation of the eccentric brace is less than that
involved in concentric bracing because: (a) the eccentric
brace requires three connections with the concrete mem-
bers (Fig. 3) as compared to four connections in the con-
centric brace case, and (b) the beam–link connection is
easily accessible.
The distribution of the eccentric brace over the build-
ing height was found to have a significant effect on the
characteristics of the developed plastic mechanism under
the effect of lateral seismic load. It is suggested that the
distribution of brace strength over the height of the
building should be selected to obtain a uniform distri-
bution of story drift.
References
[1] Kawamata S, Masaki Q. Strengthening effect of eccentric steel
braces to existing RC frames. In: Proceedings of the 7th World
Conference on Earthquake Engineering; Istanbul, Turkey,
1980:513–20.
[2] Canales MD, Briseno de la Vega R. Retrofitting techniques used
in telephone buildings in Mexico. In: The 10th World Conference
on Earthquake Engineering, Madrid, Spain, vol. 9, 1992:5143–7.
[3] Popov EP, Malley JO. Design of links and beam-to-column con-
nections for eccentrically braced steel frames. Report No. EERC
83-03. Berkeley (CA): Earthquake Engineering Research Center,
University of California, 1983.
[4] Kasai K, Popov E. A study of seismically resistant eccentrically
braced steel frame systems. Report No. UCB/EERC-86/01.
Berkeley (CA): Earthquake Engineering Research Center, Uni-
versity of California, 1986.
[5] Ricles JM, Popov EP. Dynamic analysis of seismically resistant
eccentrically braced frames. Report No. UCB/EERC-87/07.
Berkeley (CA): Earthquake Engineering Research Center, Uni-
versity of California, 1987.
[6] Popov E, Engelhardt M, Ricles JM. Eccentrically braced frames:
US practice. Engng J, AISC 1989;Second quarter:66–80.
[7] Ghobarah A, Ramadan T. Seismic analysis of links of various
 A. Ghobarah, H. Abou Elfath / Engineering Structures 23 (2001) 745–755
lengths in eccentrically braced frames. Can J Civil Engng
1991;18(1):140–8.
[8] Ghobarah A, Ramadan T. Bolted link–column joints in eccentri-
cally braced frames. Engng Struct 1994;16(1):33–41.
[9] Ghobarah A, Ramadan T. Dynamic analysis of twenty story EBFs
with semirigid joints. Eur Earthquake Engng 1995;IX(3):23–32.
[10] Fehling E, Pauli W, Bouwkamp JG. Use of vertical shear links
in eccentrically braced frames. In: The 10th World Conference
on Earthquake Engineering, vol. 8, 1992:4475–80.
[11] Michael D, Popov EP. Behaviour of long links in eccentrically
braced frames. Report No. UCB/EERC-89/01. Berkeley (CA):
Earthquake Engineering Research Center, University of Califor-
nia, 1989.
[12] Roeder CW, Popov EP. Inelastic behaviour of eccentrically
braced frames under cyclic loading. Report No. EERC-77/18.
Berkeley (CA): Earthquake Engineering Research Center, Uni-
versity of California, 1977.
[13] Ramadan T, Ghobarah A. Analytical model for shear–link
behavior. J Struct Engng, ASCE 1995;121(11):1574–80.
755
[14] Prakash V, Powell GH. DRAIN-2DX — version 1.02 — user
guide. Report No. UCB/SEMM-93/17. Berkeley (CA): Civil
Engineering Department, University of California, 1993.
[15] ACI Committee 318. Building code requirements for reinforced
concrete (ACI 318- 63) and commentary. Detroit (MI): American
Concrete Institute, 1963.
[16] CSA-S16.1. Limit state design of steel structures. Rexdale
(Ontario, Canada): Canadian Standards Association, 1994.
[17] Lee S, Goel SC. Seismic behaviour of hollow and concrete filled
square tubular bracing members. Report No. UMCE 87-11. Ann
Arbor, MI: University of Michigan, 1987.
[18] Jain AK, Goel SC. Hysteresis models for steel members subjected
to cyclic buckling. Report No. UMCE 78R6. Ann Arbor, MI:
University of Michigan, 1978.
[19] Trifunac MD, Brady AG. A study on the duration of strong earth-
quake ground motion. Bull Seismol Soc Am 1975;65(3):581–626.
[20] Ghobarah A, Abou Elfath H, Biddah A. Response-based damage
assessment of structures. Earthquake Engng Struct Dynamics
1999;28:79–104.