th

The 14 World Conference on Earthquake Engineering

October 12-17, 2008, Beijing, China

SEISMIC DESIGN BASIS FOR INTERNALLY-BRACED RC FRAMES

1

M.R. Maheri and H. Ghaffarzadeh

Professor, Dept. of Civil Engineering, Shiraz University, Shiraz. Iran

Assistant Prof, Dept. of Civil Engineering, University of Tabriz, Tabriz, Iran
Email: mmaheri@hotmail.com, ghaffar@tabrizu.ac.ir

ABSTRACT :
Internal steel bracing of RC frames with direct connections has received some attention in recent years, both as
a retrofitting measure to increase the shear capacity of the existing RC buildings and as a shear resisting element
in the seismic design of new buildings. Although its successful use to upgrade the lateral load capacity of
existing Reinforced Concrete frames has been the subject of a number of studies, guidelines for its use in newly
constructed RC frames need to be further developed. An important consideration in the design of steel-braced
RC frames is the level of interaction between the strength capacities of the RC frame and the bracing system. In
this paper, results of experimental and numerical investigations aimed at evaluating the level of interaction
between the bracing system and the RC frame are discussed. For these investigations, cyclic loading tests were
conducted on scaled moment resisting frames with and without bracing. The experimental results are used to
calibrate full-scale numerical models. A parametric numerical investigation on the effects of the main problem
variables is then conducted and the influence of each parameter on the level of interaction is determined. Based
on these findings, guidelines for the seismic design of the internally-braced RC frames with direct connections
are provided.
KEYWORDS: Steel bracing, reinforced concrete, braced RC frames, cyclic load testing
1. INTRODUCTION
Observations from recent earthquakes resulted in updating of some seismic codes. As a result, currently
upgrading of the existing RC frames to achieve the requirements of new seismic codes is of particular interest.
Using steel bracing to upgrade the seismic capacity of existing RC frames has been the subject of several
research investigations over the past three decades. Two bracing systems are typically considered, external
bracing and internal bracing. In external bracing, steel trusses are attached to the building exterior. Bush et al.
(1991) conducted cyclic loading tests on 2/3 scaled models of a number of structures retrofitted using external
bracing. They reported on the efficiency of such a method in retrofitting existing RC buildings. Badoux and
Jirsa (1990) investigated numerically the behaviour of RC frames retrofitted with external bracing. They
recommended using cables instead of steel sections for the brace elements to avoid buckling of the brace
members, and thus increasing the ductility of frames.
In internal bracing, steel trusses or bracing members are inserted in the empty space enclosed by columns and
beams of RC frames. Higashi et al. (1981), Rodriguez and Park (1991), Masri and Goel (1996) studied the
effectiveness of using internal steel trusses to retrofit existing RC frames. They reported that such a method
allows upgrading the seismic capacity of existing structures. Maheri and Sahebi (1997), Maheri et al (2003),
Maheri and Hadjipour (2003) recommended the use of internal brace members directly attached to the RC
members over the internal steel trusses.
Recent works by Ghaffarzadeh and Maheri (2006) have shown further that the directly-connected internal
bracing systems can be used effectively in retrofitting of existing concrete frames as well as shear resisting
elements for construction of new RC structures. In this study, the use of X-shaped concentric internal steel
bracing for new construction was investigated experimentally. An important consideration in the design of steelbraced RC frames is the level of interaction between the strength capacities of the RC frame and the bracing

th

The 14 World Conference on Earthquake Engineering

October 12-17, 2008, Beijing, China

system. In this paper, results of experimental and numerical investigations aimed at evaluating the level of
interaction between the bracing system and the RC frame are discussed. Three specimens representing an RC
moment frame with moderate ductility and two braced RC frame were designed. Current seismic codes were
used to design the moment frame. For the braced frames, a rational design methodology is proposed. The model
frames were subjected to cyclic loads. Their test results are compared and discussed. These results are also used
as a basis for developing and calibrating full-scale numerical models of full-scale frames. Using the numerical
models, a parametric investigation is carried out to determine the role of the main variable parameters affecting
the level of capacity interaction between the RC frames and bracing systems.

2. EXPERIMENTAL PROGRAM
Unit frames were modelled using a mid-span panel measuring 4.0 m by 3.0 m from the third floor of a fourstorey building with dimensions of 12.0 m by 12.0 m. It was assumed that the building is located in a highly
seismic area. Two lateral load resisting systems, namely; RC moment frames and braced RC frames, were
considered. The gravity and earthquake forces acting on these panels were determined in accordance with
Iranian seismic code (1999) using the seismic force reduction factor for moment frames with moderate ductility.
The size of the test specimens was limited to the available laboratory space and equipment limits. A 2/5 scaled
model measuring 1.76 m by 1.36 m was found satisfactory. The forces acting on the models were also scaled
down by a factor of (2/5)2. This factor was chosen to keep the stresses in the scaled model similar to the fullscale panel. The boundary conditions for the tested specimens were chosen such that the internal forces
developed in them are similar to those developed in reality. Two hinged supports were thus used to support the
specimens. The dimensions of the beams and columns were chosen to be 140 mm by 160 mm.
Three specimens were designed and constructed, one moment frame namely F1 and two braced frames, namely
BF1 and BF2. The moment frame was designed according to ACI 318-02 (2002) and its detailing was done in
accordance with the ACI special provisions for seismic design. AISC-LRFD (2001) was used to design the brace
members and their welded connections to the guest plates. Their design was also checked using the AISC
seismic provisions for steel structures (2002).
The specimens were tested using the setup presented in Ghaffarzadeh and Maheri (2006) (Fig. 1). An actuator
was used to apply several cycles of in-plane shear load using a displacement-controlled approach. In each cycle,
the actuator was first pulled to a displacement d1 of 5 mm (drift of 0.417%) then pushed to the same
displacement. The value of d1 was increased in the following cycles by increments of 5 mm. Strain gauges were
used to monitor strains in the beam-column joint, the transverse reinforcement of the columns, and the
longitudinal reinforcement of the beams.

Figure 1. Setup for cyclic testing of specimens

th

The 14 World Conference on Earthquake Engineering

October 12-17, 2008, Beijing, China

200

200

150

150

100

100

Lateral Load (kN)

Lateral Load (KN)

The lateral load-drift hystresis for the moment frame F1 and braced frame BF1 are shown in Fig. 2. The initial
stiffness of the braced frames was higher than that of the moment frame. The yield and failure drifts of the
specimen F1 were 1.67% and 5.00%, respectively and those of the specimens BF1 and BF2 were 2.08%, 4.0%,
and 2.5%, 4.3% respectively. This shows that the ductility of the specimen F1 was 3.0 and the specimens BF1
and BF2 were 1.9 and 1.7, respectively. It is clear from the hysteretic behaviour that the pinching was less
significant in the braced frame, indicating an overall better seismic performance.

50
0
-50
-100

50
0
-50
-100

-150

-150

-200
-6

-5

-4

-3

-2

-1

Lateral Drift (%)

-200
-6

-5

-4

-3

-2

-1

Lateral Drift (%)

Figure 2. Lateral load-drift hystresis of frames F1 and BF1.

The behaviour of the tested specimens was significantly different. For the specimen F1, the first observed crack
occurred at a load of 30.0 kN and was a flexural crack in the bottom beam at the face of the column. At a load of
37.5 kN, yielding of the bottom bars of the lower beam initiated the plastic response. Failure occurred by plastic
hinging at the ends of the top and bottom beams at a load of 55 kN. The observed cracking load for the
specimen BF1 was 90.0 kN. Cracks observed in this specimen were less in number and lower in width than that
for the moment frame. At a load of 105.0 kN, yielding of the brace member initiated the plastic response.
Failure resulted due to buckling of the compressive brace, which was directly followed by plastic hinging of the
ends of the bottom and top beams. The failure load for this specimen was 140 kN. It should be noted that the
brace member connections, including welds and headed studs, behaved adequately. The specimen BF2 exhibited
almost linear behaviour because of the amount of bracing in comparison to specimen BF1. First cracks observed
in this specimen at a load of 140 kN. By increasing lateral drift, failure happened at a load of 200 kN.

3. EXPERIMENTAL BRACE-FRAME CAPACITY INTERACTION

In design of steel braced RC frames the level of interaction between the strength capacities of the RC frame and
the bracing system should be established. To investigate this interaction in the tested model frames, the
corresponding forces in the bracing systems alone were evaluated by considering the relevant test displacements
on the diagonals. A simple bilinear model for steel which accounts for cyclic effects was assumed and used to
represent the force-deflection envelop curve of steel bracing system alone. The envelop curve of the calculated
force-drift relationship for the FX1 bracing system alone (marked as No. 2 in the figure) is plotted in Fig. 3.a.
Also plotted in this figure, for comparison, are the experimental envelop of the force-drift relationship of the
moment frame alone, F1, (marked as No. 1 in the figure) and the experimental envelop of the force-drift
relationship of the FX1 braced frame. To be able to gain an insight into the level of capacity interaction between
different elements, the envelop curves of the bracing system alone (2) and the moment RC frame (1) are added
together to obtain the sum strength capacity of the two elements as also are presented in Fig. 3.a ((1) + (2)). By
comparing the sum strength capacity of the two constituent elements with the actual strength capacity of the
braced frame, it is evident that the actual braced frame exhibits a larger capacity than the sum of the capacities
of the two elements. This means that by adding a bracing system to an RC frame, the strength capacity of the
RC frame is increased beyond the capacity of the bracing system. The positive interaction is evidently due to the

th

The 14 World Conference on Earthquake Engineering

October 12-17, 2008, Beijing, China

(a)

(b)
200

200

F1 (1)
150

Bracing system (2)

(1)+(2)
FX1

Lateral Load (kN)

Lateral Load (kN)

100
50
0
-50

150

F1 (1)
Bracing system (2)

100

(1)+(2)
FX2

50
0
-50

-100

-100

-150

-150

-200

-200

-6

-4

-2

Drift (%)

-6

-4

-2

Drift (%)

Figure 3. Comparison between experimental lateral loaddrift envelop curves of the moment frame, F1, bracing
system and the braced RC frames ((a) FX1& (b) FX2)
stiffening effects of the connections between the RC frame and the bracing system. The capacity interaction for
the frame FX1 is measured, as the minimum of all the evaluated values, at 8.5 percent.
Similarly, the calculated strength capacity of the bracing system of frame FX2 and the experimental strength
capacities of the moment frame, F1, and the braced frame FX2 are plotted in Fig. 3.b. Also plotted in this figure
is the sum of the strength capacities of the bracing system and the RC frame alone. Similar added increases in
the strength can be seen in this case. The capacity interaction for the frame FX2 is measured at 7.0 percent.
Considering these experimental results it is evident that the capacity interaction is in fact an overstrength which
can be attributed mainly to the effects of brace-frame connections in reducing the effective lengths of the RC
beams and columns hence increasing the stiffness and strength of the frame.

4. NUMERICAL EVALUATION OF INTERACTION

To investigate the level of capacity interaction between steel bracing and RC frame in full scale buildings, nonlinear pushover numerical analyses of the moment frame, braced frames and the bracing systems were carried
out. The OpenSEES (Open System for Earthquake Engineering Simulation) program was utilised to numerically
model the frames. The beams and columns of the frames were modeled using nonlinear beam-column element
of the program. To model the steel bracing members, the truss element was considered. Initially the numerical
models were calibrated and their accuracy ascertained by comparing the results of the non-linear cyclic analysis
of the moment frame F1 with the results obtained from the cyclic tests on this frame.
After calibrating the numerical models, a series of nonlinear pushover analyses were conducted on full scale
frames of different heights and widths with different bracing configurations. These included frames, 4, 8 and 12
storeys high and 3, 6 and 9 bays wide. The number of braced bays in each frame was also made a function of the
number of bays such that the three, six and nine-bay frames had, respectively, one, two and three bays braced.
All frames consisted of 3m high and 5m wide unit frames. Another variable parameter in this investigation is the
apportioned share of bracing system from the applied loading. Load shares of 30%, 50%, 80% and 100% for
bracing system are considered. As it was mentioned earlier, the main factor contributing to the interaction in the
form of overstrength is the effect of connections on reducing the effective lengths of beams and columns.
Therefore, a fourth variable parameter considered is the ratio of the length of connection plates (Le) and the
length of their associated beam (Lb). Practical values of (Le)/ (Lb) = 0.05, 0.075, 0.10, 0.125, 0.15, 0.175 and
0.20 are considered. Two hundred and fifty two braced frames were thus designed. Details of the loading
considered and design of each frame and bracing system are given in Ghaffarzadeh (2006).

th

The 14 World Conference on Earthquake Engineering

October 12-17, 2008, Beijing, China

For each frame, three non-linear pushover analyses are carried out. These include; i) pushover analysis of the
RC frame without the bracing system, ii) pushover analysis of the bracing system alone and iii) pushover
analysis of the braced frame. Typical pushover curves are shown in Fig. 4. In this figure the sum of the
individual response of the RC frame and the bracing system are also plotted so that the level of overstrength in
braced frame can be seen.
2400

Lateral Load (kN)

2000
1600
1200
800
400
0
0

50

100

150

200

250

300

Displacement (mm)
Bracing system(2)

RC Frame(1)

(1)+(2)

Braced Frame

Figure 4. Pushover curves for 4-storey, 3-bay frame with (Le)/ (Lb) = 0.1
The overstrength for each frame was then quantified by considering a lower bound value at displacements
corresponding to 2% drift. Typical overstrength values, designated R, for 4-storey, 3-bay frames with different
number of (Le)/ (Lb) and bracing load share are listed in Table 1. The overstrength values, R, for the 4-storey, 3bay frames are also plotted in Fig. 5. A closer look at table 1 and Fig. 5 shows that the effects of load share for
bracing system on the overstrength is negligible compared with the parameter (Le)/ (Lb). This was found to be
true for all the other frames analysed. This was expected as the relative size of the brace-frame connections is
considered to be the main contributor to the overstrength. As a result the effects of load share of bracing (or the
cross-sectional area of the braces) on the overstrength is neglected and for further investigations the load share
of bracing was assumed to be constant at 100% which is on the safe side.
No.
of
Bays
3

Table 1. The overstrength values for the 4-storey frames

(Le)/ (Lb)
Share
of
0.05 0.075 0.10 0.12 0.15 0.175
brace
5
(%)
30
6.00
8.00 11.0 12.6 14.5
15.6
50
5.75
7.70 10.5 12.0 14.0
15.0
80
5.65
7.50 10.3 11.8 13.5
14.7
100
5.40
7.20 9.90 11.3 13.0
14.0

0.20
18.0
17.4
16.9
16.3

20
%30 Bracing
%50 Bracing
%80 Bracing

15

R (%)

%100 Bracing

10

0
0

10

15

20

25

Le/Lb (%)

Figure 5. The overstrength values, R, for the 4-storey, 3-bay frames

th

The 14 World Conference on Earthquake Engineering

October 12-17, 2008, Beijing, China

In the above analyses, the parameter (Le)/ (Lb) was assumed to loosely represent the effect of connections on the
overstrength. However, this parameter does not take into consideration the influence of connections on the
stiffness of columns. Therefore, considering the nature of the interaction, a more representing parameter can be
introduced as the ratio of the effective stiffness of the RC frame with connections (Kr) to the stiffness of the RC
frame without the connections (Ki), designated . Considering that the connections reduce the effective lengths
of RC beams and columns, the effective stiffness of the frame with connections corresponds to the stiffness of a
reduced frame as shown in Fig. 6. For simplicity and conservatively, the reduced frame is assumed to have
beams and columns of lengths equal to the distances between the centroids of the four gusset plates as seen in
Fig. 6. Also, for practical purposes, the parameter is calculated as the ratio of the stiffness of the reduced RC
frame of a central floor (Kr) and the stiffness of the initial RC frame of the same central floor (Ki) shown in Fig.
6. The lateral stiffness of such one-storey frames with upper and lower beams can be calculated analytically
using the well known relation Ghaffarzadeh (2006);
K=

24 E
2
1
1
Lc (
+
+
)
K
K
c bb K bt

(4.1)

where, Kc, Kbb and Kbt, are I/L for columns, lower beams and upper beams, respectively and Lc, is the effective
height of the frame.

K r

Ki

Figure 6. The reduced frame dimensions for calculation of effective stiffness

The stiffness ratio, as described above is calculated for all the frames analysed and the overstrength factors, R
previously determined are plotted against the stiffness ratio for different frame geometries considered. Typical
plots for the 4-storey and 12-storey frames are shown in Fig.7. A near linear relation between the two parameters
can be seen. This enables us to draw linear relations between the two parameters as presented in Fig.7
(a)

(b)
30

20
Computational Data

Computational Data

25

Fitting Curve

Trend Curve

15

R (%)

R (%)

20

10

15

10

5
5

0
1

1.1

1.2

1.3

1.4

1.5

1.1

1.2

1.3

1.4

1.5

1.6

1.7

Figure 7. The overstrength, R as a function of typically for (a) 4-storey, 3-bay and (b) 12-storey, 9-bay frames
To condense the results of the 9 relations thus obtained, the linear relation for the 4-storey, 3-bay frame will be
considered as the base overstrength, Rb, and the effects of the two main variable parameters including the

th

The 14 World Conference on Earthquake Engineering

October 12-17, 2008, Beijing, China

number of braced bays (number of bays in the frame) and the number of storeys will be considered respectively
as correction factors a and b . Therefore;
R = abRb (%)
Rb = 32 27

Where,

(4.2)

The factors a and b are evaluated for different conditions and are plotted against in Fig. 8.a and Fig. 8.b,
respectively. Noting the near linear variation of a against the following relations can be presented for this
correction factor;
a = 0.16m + 0.84
a = 0.09m + 0.91
a = 0.06m + 0.94

for
for
for

0.0 < 1.0

1. 0 < 1.25
1.25 < 1.40

(a)

(4.3)

(b)
1.5

1.4

1.45

Fitting Line on 2 Span Braced

Data
Fitting Line on 3 Span Braced
Data

1.35
1.3

1.4
1.35

1.25

1.3

1.2

1.25
1.2

1.15

1.15
1.1

1.1
1.05

Fitting Line: 8 Storey Frames

Fitting Linre: 12 Storey Frames

1.05
1

1
1

1.05

1.1

1.15

1.2

1.25

1.3

1.35

1.4

1.45

1.05

1.1

1.15

1.2

1.25

1.3

1.35

1.4

1.45

Figure 8. Values of correction factors a and b for all frames

Also, as the variation of b with is small, this correction factor can be presented independent of the stiffness
ratio as;
b = 0.0425n + 0.84

(4.4)

In equations (4.3) and (4.4), m and n are the number of braced bays and the number of storeys, respectively.

5. CONCLUSIONS
The conclusions drawn from the experimental and numerical investigations on the nature and level of capacity
interaction between the bracing system and RC frame may be summarised as followings
1. The overstrength in a braced RC frame is due to the added strength of the brace system as well as an
added strength in the RC frame due to stiffening effects of connections. The later overstrength is termed
the capacity interaction. It is significant and needs to be considered in design
2. The important parameters affecting the capacity interaction are recognised as the number of braced bays
and the number of frame storeys. The third important parameter is the stiffening effects of the
connections taken into consideration as a stiffness ratio.
3. The overstrength, R, for different 2-D frames are determined and presented in simple forms for use in
design of internally braced RC frames.

th

The 14 World Conference on Earthquake Engineering

October 12-17, 2008, Beijing, China

REFERENCES
ACI Committee 318. Building code requirements for reinforced concrete (ACI 318-02). (2002). American
Concrete Institute, Detroit, MI.
AISC Manual of steel construction: load and resistance factor design. (2001). 3rd ed. Chicago (IL): American
Institute of Steel Construction.
AISC. Seismic provisions for structural steel buildings. (2002). Chicago (IL): American Institute of Steel
Construction.
Bush, TD., Jones, EA. and Jirsa, JO. (1991). Behavior of RC Frame Strengthened Using Structural-Steel
Bracing. Journal of Structural engineering, ASCE. 117:4, 1115-1126.
Badoux M, Jirsa JO. (1990). Steel bracing of RC frames for seismic retrofitting. Journal of Structural
engineering, ASCE. 116:1, 55-74.
Ghaffarzadeh, H. and Maheri, MR. (2006). Mechanical compression release device in steel bracing system for
retrofitting RC frames. Earthquake Engineering and Engineering Vibration 5:1.
Ghaffarzadeh, H. Maheri, MR. (2006). Cyclic tests on the internally braced RC frames. Journal of Seismology
and Earthquake Engineering 8:3.
Ghaffarzadeh, H. Design basis for internally-braced RC frames. (2006). PhD Thesis, Shiraz University, Shiraz,
Iran.
Higashi, Y., Endo, T. and Shimizu, Y. (1981). Experimental studies on retrofitting of reinforced concrete
structural members. Proceedings of the Second Seminar on Repair and Retrofit of Structures, Ann Arbor, MI:
National Science Foundation, 126-155.
Iranian code of practice for seismic resistance design of buildings. (1999). Standard No. 2800, 2nd ed.
Masri, AC. and Goel, SC. (1996). Seismic design and testing of an RC slab-column frame strengthened by steel
bracing. Earthq. Spectra 12:4, 645-666.
Maheri, MR. and Sahebi, A. (1997). Use of steel bracing in reinforced concrete frames. Journal of Engineering
Structures 19:12, 1018-1024.
Maheri, MR., Kousari, R. and Razazan, M. (2003). Pushover tests on steel X-braced and knee-braced RC
frames. Journal of Engineering Structures 25, 1697-1705.
Maheri, MR. and Hadjipour, A.. (2003). Experimental investigation and design of steel brace connection to RC
frame. Journal of Engineering Structures 25, 1707-1714.
Rodriguez, M. and Park, R. (1991). Repair and strengthening of reinforced concrete buildings for seismic
resistance. Earthq. Spectra 7:3, 439-459.