AN INNOVATIVE BRACING CONFIGURATION

FOR IMPROVED SEISMIC RESPONSE

Tremblay, R.
1
, Ben Ftima, M.
2
, and Sabelli, R.
3
ABSTRACT

The paper describes an innovative braced steel frame system, the Flexural Braced
Frame (FBF) system, in which buckling restrained braces (BRBs) are used for columns in the
bracing bent to dissipate energy through overall flexural response of the frame. Other
members in the bracing bent are designed to remain essentially elastic. BRB members are
used at the bottom levels, but additional BRBs can be introduced at building mid-height of
taller structures to reduce the storey shear demand from higher mode response. The girders
connecting to the bracing bent from adjacent bays are also used to develop back-up stiffness
for the system, upon yielding of the BRB elements. A design method is proposed for the system
and is applied to 4-, 8- and 16-storey example buildings. Linear and nonlinear dynamic time
history analyses of the structures show that the proposed system exhibits satisfactory seismic
performance, with low and uniform storey drift demand.

Key Words: Steel, seismic design, ductility, buckling restrained members, braces
1. INTRODUCTION

In the aftermath of the Northridge and Kobe earthquakes, the popularity of
concentrically braced steel frames has increased significantly for providing lateral resistance
to building structures. Although stringent seismic design provisions were implemented in
codes, the inelastic demand in multi-storey braced steel frames still tends to concentrate in a
few storeys due to the degrading hysteretic response of the bracing members and the limited
capacity of the system to redistribute the demand over the building height [1-6].
Several approaches have been proposed to achieve a more uniform distribution of the
inelastic deformations in these structures. For instance, zipper columns can be added in

1
Professor,
2
Research Assistant. Dept. of Civil geological and Mining Engineering, Ecole Polytechnique, P.O.
Box 6079, Station Centre-ville, Montreal, Quebec, Canada H3C 3A7
3
Director, Technical Development, DASSE Design, Inc., 33 New Montgomery St., San Francisco, CA 94105-
4525, USA

chevron braced steel frames to mobilise inelastic action in adjacent storeys after brace
buckling has occurred at a given floor [7, 4, 8]. The use of bracing members with enhanced
hysteretic behaviour such as buckling restrained braces (BRBs) [9, 10] has been shown to lead
to satisfactory response for braced steel frames [4, 11, 12]. The robustness of such braced
frames can be improved further by parallel systems providing back-up stiffness [6].
Connecting all link beams with vertical ties in eccentrically braced steel frames has been
shown to be a very efficient means of ensuring uniform demand in the link elements over the
height of multi-storey structures [13, 14].
In this paper, an innovative braced frame configuration, the Flexural Braced Frame
(FBF) system, is proposed that combines several of these beneficial features. It consists of a
chevron bracing in which a central column is added and vertically mounted buckling
restrained members are introduced along the edge columns. The system is illustrated in Fig. 1
with BRB elements anchored to the foundation and extending up to the Level 2 beams.
Capacity design principles are applied in order to limit plasticity to the BRB members. Such a
vertical arrangement of BRBs has been proposed for energy dissipation in high-rise rack
warehouse structures [15]. The system behaves very similarly to a flexure-controlled
conventional reinforced concrete shear wall: upon yielding of the BRB elements, rotation of
the upper bracing bent occurs about the point at which four braces intersect the column. When
compared to other applications of the buckling restrained member technology, one of the
advantages of the system is that it requires a limited number of BRB devices. In addition, the
BRB units can be introduced in a two-storey X-bracing configuration in the bottom two
floors, as shown in Fig. 1, so that they have sufficient length to accommodate large plastic
deformations.

P P
P
P
V / 2R V / 2R
P
2
d d e e
2
1
g
g
1
L
L
L
L

o
bc
L
s
,
1
-
2
h
Steel Core
Stiffener
(typ.)
Splice
Plate
(typ.)
Mortar Fill
Section A
Section B
Steel Tube
bt
L
L
A
B
2
bt L
bj
L
Foundation
2
2
bj 2
Steel
core
A
h
s,i
Fig. 1 Flexural Braced Frame system.
In tall buildings, it may be advantageous to introduce additional vertical BRB
members in the upper levels to dissipate the seismic energy associated with higher mode
response. Another important feature of the braced frame system is that the beams span from
the central column to the column lines adjacent to the bracing bent, thus acting as a series of
outriggers providing rotational restraint to the bracing bent at every level and, thereby,
additional overturning moment capacity. These beams are also designed to remain elastic up
to the maximum anticipated lateral deformation and, hence, provide the structure with positive
stiffness upon yielding of the BRBs as well as a potential re-centring capability. In the system,
the edge columns of the bracing bent are interrupted at every level and act as vertical ties
linking all floors together, such that all beams are forced to respond as a whole when large
inelastic deformation develop in the BRBs.
In the paper, a tentative design procedure for the proposed FBF system is presented. It
is then applied to 4-, 8- and 16-storey structures and the seismic response of the structures
under a suite of earthquake records is examined and discussed, with particular emphasis on
the benefits of introducing additional vertical BRB members for higher mode response for
taller structures.

2. DESIGN PROCEDURE

The design is carried out in accordance with recent building code procedures,
including capacity design principles. Design seismic loads in codes are typically obtained by
dividing expected elastic design forces by prescribed force modification factors to account for
the capacity of structural systems to withstand earthquakes in the inelastic range. For the FBF
system, the anticipated inelastic response mimics that of a ductile reinforced concrete
cantilevered wall, i.e. with a rotational plastic hinge forming at the base of the structure.
Figure 2a shows the idealised lateral load-deformation behaviour of the system under first
mode response, with the overturning moment resistance at the base being provided by the
BRB elements and the girders acting in parallel. In the figure, V
e
is the design elastic seismic
force and the lateral load V
e
/R
dl
corresponds to yielding of the BRB elements, at a deflection
A
yl
= A
e
/R
dl
. For this system, it is proposed that the local ductility-related force modification
factor, R
dl
, be determined for each particular structure such that the maximum anticipated
plastic deformation in the BRB elements remains within acceptable limits. The factor must
also be adjusted to keep the maximum anticipated deflection, A
e
, within code specified limits.
A preliminary design of the girders and the BRB elements is first performed assuming a trial
value for the R
dl
factor. The columns and the bracing members are then designed to ensure
elastic response at the deflection A
e
. The design is then adjusted until the expected response is
deemed acceptable. Static linear elastic analysis is sufficient for the optimisation process but
nonlinear push-over analysis can also be utilised as an alternative, as discussed later.

2.1 Buckling restrained brace members

In order to adequately predict the plastic deformation demand on the BRBs, key
geometrical properties of these elements must be considered in design. In Fig. 1, the total
length of the brace corresponds to the height of the first two storeys, h
s,1-2
= h
s,1
+ h
s,2
. Plastic
deformation, c
pc
, is assumed to develop in the core (narrow) portion of the brace, L
bc
.
Therefore, the total inelastic brace axial deformation, o
b
, is given by o
b
= (o
by
+o
bp
), with o
by
=
c
yb
h
s,1-2
(A
bc
/A
be
) and o
bp
= c
bp
L
bc
. In these expressions, c
yb
is the steel yield strain of the brace
core, A
bc
is the cross-sectional area of the core, and A
be
is the cross-sectional area of an
equivalent BRB member with uniform cross-section [16]:

bj
bc
bj
bt
bc
bt bc
2 1 , s bc
bj
bj
bt
bt
bc
bc
2 1 , s
be
A
A
L
A
A
L L
h A
A
L
A
L
A
L
h
A
+ +
=
+ +
=

(1)

Parameters L
bt
and L
bj
are illustrated in Fig. 1 and the cross-sectional areas A
bt
and A
bj

are those corresponding to the transition zones and joint portion of the brace. These
parameters will vary from one brace system to another, the capacity of the brace, and
connection details. For preliminary design purposes, typical values can be used. For instance,
the ratios A
bc
/A
bt
and A
bc
/A
bj
generally vary from 0.4 to 0.6 and from 0.2 to 0.4, respectively.
The length L
bt
ranges between 200 and 500 mm (total of two ends) and the length of both
joints, L
bj
, would vary from 500 mm to 700 mm for butt splice connections with splice plates,
as shown in the figure. To account for deformations in connecting plates and structural
members beyond the length of the brace, the joint properties can be extended up to the beam
center line at the top end and down to the surface of the foundations at the lower end. For
structural analysis, the equivalent BRB member with cross-sectional area A
be
and total length
h
s,1-2
can be conveniently used in modelling. If nonlinear analysis is performed, the core
plastic strain can be established from the total deformation o
be
of the equivalent brace, c
pc
=
(o
be
-o
by
)/L
bc
. For BRBs located in the upper levels, the same procedure applies except that h
s,i

is used instead of h
s,1-2
.
V
V
Girders
BRBs
Total
Column Axial Load
Additionnal.
BRBs
Exterior
Columns
H
e
i
g
h
t
1
1
K
K
A A A
A
V
V
Additionnal
BRBs
P P
R
R
e
g
e
yl y
e
e
dl
d
by, max by
a) b)
Fig. 2 Design assumptions: a) Idealised lateral load-deformation response;
b) Axial loads in columns.
2.2 Preliminary design of the girders and of the BRB elements at the base of the frame

The girders are designed such that they will remain essentially elastic at maximum
anticipated deflection, A
e
, thus providing the desirable positive stiffness upon BRB yielding
and a restoring force that can potentially be utilized to eliminate residual drift. For simplicity
for the first trial design, the same girder size can be used at all levels and an approximate
estimate of the average vertical girder deflection over the building height can be used to
determine the maximum flexural demand, M
g,max
. At A = A
e
, the BRB elements are expected
to reach their maximum allowable axial deformation, o
b,max
, determined with the core plastic
deformation c
bp
equal to the maximum allowable value, typically between 0.015 and 0.020.
Assuming the average axial strain in the columns above Level 2 is equal to o
c
c
y
, the average
girder deflection o
g,max
, is given by: o
g,max
= o
b,max
+ o
c
c
y
(h
n
- h
s,1-2
)/2. The parameter o
c
generally varies between 0.3 and 0.4, and the required factored resistance of the girders, M
r
, is
obtained from:

max , g grav , g p r
M M M M + > | = , with:
( )
2
max , g g
max , g
L 1
EI 3
M

o
= (2)
, and where | is the material resistance factor (= 0.9 for steel), M
g,grav
is the girder moment due
to gravity loading, M
p
is the plastic moment of the beams, L is the total length of the beams
and L corresponds to half the width of the bracing bent (see Fig. 1). The moment M
g,grav
is
determined neglecting the support provided by the bracing bent columns. Compact sections
should be used for the girders so that they can maintain their flexural capacity in case yielding
develops under the design earthquake and, hence, the resistance can be based on the plastic
moment. The steel shape must be selected among those having an I
g
-M
r
combination that
satisfies Equation (2), typically stubby members with depth of up to approximately 400 mm.
An initial value of R
dl
is chosen, typically between 3.0 (tall frames) and 5.0 (lower
frames), and the overturning moment at the base at yielding of the BRB, M
By
, is computed
using the seismic loads corresponding to V
e
/R
dl
. As illustrated in Fig. 1, this moment is
resisted by the vertical reactions P
1
and P
2
, and M
By
= 2L(P
1y
+P
2y
), where P
1y
is the sum of
the axial force in the BRB element at yield, P
by
, and the vertical component of the axial load
in the first storey braces, and P
2y
is the vertical load induced by the beam shear forces:

dl
1 , s e
by y 1
R L 2
h V
P P

+ = (3)

3 2
gy g
gy y 2
L ) 1 (
EI 3
) 1 n ( P ) 1 n ( P

o
= = (4)

In the equation for P
2y
, o
gy
is the average girder deflection when yielding starts in the
BRB elements, which can be assumed equal to: o
gy
= o
by
+ o
c
c
y
(h
n
- h
s,1-2
)/2. Knowing P
2y
, the
reaction P
1y
can be determined from the expression for M
By
, P
by
can be obtained from
Equation (3), and, finally, the cross-section area of the BRB core, A
bc
, is obtained from A
bc
F
y
= P
by
. In this expression, the yield strength can be replaced by R
y
F
y
, where R
y
is the ratio of
the actual to the minimum specified material strength (=1.1 for steel), recognising that the
expected material strength can be used in design for extreme loading conditions such as
earthquake effects determined for long return periods. If that was the case, the expected yield
strength should also be used for the calculation of o
by
.
Using the above equations, a combination of beams and BRB elements can be selected
to achieve the desired response. Preference should be given to stiffer girders as they will
attract more loads, resulting in smaller BRB elements and exterior columns. Higher elastic
restoring force will also be developed upon yielding of the BRBs. However, the cost of the
girders can become prohibitive if too much beam response is involved, and the columns
adjacent to the bracing bent may need to be strengthened to resist the load P
2
acting in
conjunction with gravity loads. The frame geometry can also be modified at this point to
optimise the response. Typically, values of between 0.30 and 0.45 would lead to the best
solutions in terms of maximising P
2
while keeping low the bending moments in the beams.
The assumed R
dl
value and deflection criteria must then be verified. For R
dl
, the value
used in the calculations must not exceed the available global displacement ductility at roof
level, R
dl,max
, as limited by the plastic deformation capacity of the BRB core: R
dl,max
=
1+A
p,max
/A
yl
, where A
p,max
is the plastic roof displacement corresponding to allowable
maximum plastic strain in the BRB elements and A
yl
is the roof displacement at onset of
yielding in the BRBs. The former is obtained from: A
p,max
= o
bp
(h
n
-L
bc
)/L, with o
bp
calculated
with the maximum c
bp
. The deflection A
yl
is determined from linear elastic analysis of the
frame under the seismic loads corresponding to V
e
/R
dl
. For the first design iteration, braces
and columns have not been sized yet and A
yl
can be determined from the frame geometry and
assuming realistic strain values in the members: 1.0 c
y
for the BRBs and 0.3-0.4 c
y
for the
columns and the braces. For the deflection check, the maximum anticipated inter-storey drift
angle, , is compared to prescribed code limit,
lim
. The angle is the storey drift angle
computed with the loads V
e
/R
dl
and then multiplied by R
dl
. The design of the girders and the
BRBs is adjusted until both the ductility and deflection criteria are met. Alternatively, this
validation process can be done only after the braces and the columns are designed, so that a
better estimate of deflections under the load V
e
/R
dl
is obtained.

2.3 Design of the bracing members and columns and validation

The columns and the bracing members must remain elastic during a strong earthquake
to maintain a gravity load path and ensure that the system will behave as expected. Hence, the
bracing bent columns above level 2 and the braces must resist the forces that will develop at
the maximum lateral deformation, when the axial load in the BRB elements reach their
maximum expected values. That maximum BRB load, P
by,max
, is obtained from |R
sh
R
y
P
yb
,
where | accounts for the friction that can develop between the core and the fill material and
R
sh
accounts for strain hardening in the core. Typical values for these two factors are 1.10 and
1.25, respectively, and R
y
= 1.10, leading to P
by,max
= 1.5 P
yb
. As shown in Fig. 2b, the design
axial load for the columns above Level 2 is assumed to decrease linearly from 1.5 P
yb
at Level
2 to zero at the roof. At the top floor, that design load is small and the columns are also
checked to resist the compression force corresponding to plastic hinging in the roof girders,
thus preventing premature column buckling upon severe flexural response of the girders.
For the bracing members, the axial loads determined under the seismic loads V
e
/R
dl
are
amplified by the ratio R
dl
/R
d,min
, where R
d,min
is the R
d
factor that needs to be applied to V
e
to
reach P
by,max
in the BRB elements, thus reflecting the forces that exist at maximum lateral
deformation, A
e
(see Fig. 2a). R
d,min
can be back-calculated using Equations (3) and (4) as well
as the expression derived earlier for M
B
. In (3), R
dl
and P
by
are respectively replaced by R
d,min

and P
by,max
, and o
gy
is replaced by o
g,max
in (4). In reinforced concrete shear walls forming
plastic hinges at their base, overturning moments are limited by the flexural capacity of the
walls but storey shear forces well in excess of those associated with the bending moments are
observed because higher-mode excitation can cause storey shears larger than that
corresponding to first mode flexural response. The same behaviour is expected in the FBF
system as elastic braces can transfer high shear forces even when the BRB elements are
yielding. Brace design forces are thus amplified further by applying to the forces
corresponding to P
by,max
the dynamic amplification factor proposed for concrete walls, F
DA
=
0.9+n/10 for buildings up to 6 storeys and F
DA
= 1.3+n/20 < 1.8 for taller frames, where n is
the number of storeys [17].
The central column of the braced frame is designed to carry the girder tributary gravity
loads, assuming no vertical support is provided by the exterior columns of the bracing bent.
The columns adjacent to the bracing bents are designed for gravity loads acting together with
the load P
2
determined from Equation (4) with o
gy
replaced by o
g,max
. Once all members are
selected, linear analysis of the frame can be carried out to determine A
yl
and then check the
assumed R
dl
value and drift limits. If modifications are needed, the optimum avenue is
generally to reduce A
y
such that the full plastic deformation capability of the BRB element can
be mobilised and that a high R
dl
factor is used, which will result in lower seismic loads for the
structure and its foundations. This can be achieved by increasing the column or brace sizes, by
reducing the steel yield strength or the length L
bc
of the BRB core plate, or by modifying the
geometry of the frame (e.g., L, L). Alternatively, a lower R
dl
factor can be adopted and the
expected plastic deformation in the BRB core will be reduced accordingly. As part of the
validation process, a linear elastic analysis using a secant stiffness for the BRB element, i.e.
the stiffness based on P
by,max
and the maximum BRB deformation, or a nonlinear pushover
analysis can be performed, which will provide a better distribution of the axial loads in the
columns and the flexural demand in the girders at maximum anticipated deformation, A
e
.
3. BUILDING EXAMPLES

The FBF system is used for the office building shown in Fig. 3. Three building heights
are studied: 4, 8, and 16 storeys. The geometry of the frame is as shown in Fig 1, with a storey
height of 3.8 m for all floors and L = 4.5 m for all three structures. The BRB elements at the
base have L
bc
= 5.8 m and A
be
/A
bc
= 1.18. Steel with F
y
= 345 MPa was used throughout. The
buildings were assumed to be located on a firm ground site in Vancouver, on the Pacific west
coast of Canada, and the seismic loads were computed according to the provisions of the
upcoming 2005 National Building Code of Canada (NBCC) [18] and the CSA-S16 Standard
for the design of steel structures [19]. The elastic seismic load V
e
= S(T
a
) W, where S(T
a
) is
the design spectral response acceleration at the design period, T
a
, and W is the seismic weight.
Values of S correspond to the 2% in 50 years Uniform Hazard Spectrum for the site, as
modified for local soil conditions. The seismic load was distributed over the building height
using the inverted triangle shape plus a concentrated load at the top for higher mode effects. In
NBCC, the storey drift angle limit,
lim
, is 2.5% for buildings of normal importance.

Gravity loads:
Roof: Dead = 3.0 kPa
Live = 1.8 kPa
Floor: Dead = 3.5 kPa
Partitions = 1.0 kPa
Live = 2.4 kPa
Exterior walls = 1.5 kPa
Steel: Fy = 345 MPa
6 @9000 = 54 000
Braced
Frame
Studied
Central
Column
Exterior
Column (typ.)
(typ.)
250
3
@
9
0
0
0
=
2
7
0
0
0
Fig. 3 Plan view of the buildings studied.
Table 1 gives the period T
a
, the corresponding value of S, and the final R
dl
value
adopted for each structure. That factor was modified until the plastic deformation in the BRB
cores, c
bp
, was less than 0.015 and the inelastic drift angle at the top floor,
n
, met the code
limit. These values are given in Table 1. The first criteria governed for the 4- and 8-storey
buildings, whereas the design of the 16-storey structure was controlled by the drift limit. The
design of the 8- and 16-storey frames could have been refined further to match more closely
the target criteria but it was deemed adequate to assess the performance of the system. Table 1
also presents the force P
by
, the ratio R
dl
/R
d,min
, the dynamic amplification factor F
DA
used in
the design of the braces, and the first two periods of vibration of the structures. In design, the
R
y
factor was set equal to 1.0 when sizing the BRB core elements, thus using nominal yield
strength, and the same assumption was adopted in the analysis model.

Table 1 Building properties
n T
a
(s)
S
(g)
R
dl
A
yl
/h
n
(%)
c
bp

( )

n
(%)
P
by

(kN)
R
dl
/R
d,min
F
DA
T
1
(s)
T
2
(s)
4
8
16
0.76
1.52
3.04
0.498
0.257
0.180
4.5
4.0
3.5
0.34
0.43
0.58
0.0150
0.0123
0.0125
2.0
2.2
2.8
1160
3030
10000
1.64
1.60
1.55
1.3
1.5
1.8
0.98
2.03
3.59
0.28
0.54
0.89
Nonlinear time history (NLTH) dynamic analysis of the structures was performed
using a suite of four simulated and six historical ground motion time histories produced by
intra-plate seismic events matching the two dominant magnitude-hypocentral distance
scenarios for the Vancouver region: M6.5 at 30 km and M7.2 at 70 km. The amplitude of the
ground motions was scaled to match the design spectra. The BRB elements were modelled
using a bilinear hysteretic response with 2% strain hardening, resulting in approximately 20%
force increase at c
bp
= 0.020. Bar and frame elements were used for the braces and for the
beams and columns, respectively. All these members were assumed to respond elastically in
the model so that member forces exceeding member capacity could be computed to evaluate
the appropriateness of the capacity design procedure that was adopted. P-delta effects were
considered in the calculations with 100% of the dead load, 50% of the floor live load, and 25
% of the roof snow load applied to the structure. Rayleigh damping was used with 5% of
critical damping in the first two vibration modes.
Figure 4 shows for the three buildings the mean + one standard deviation (mean+SD)
values of the peak axial loads computed in the columns and the braces, as well as the
mean+SD value of the maximum peak storey drifts at any storey in the structures. As
expected, the NLTH analyses of the systems with BRB elements at the base only (1 BRB in
the figure) confirmed that the member force demand at the levels just above the BRBs was
reduced significantly when compared to that from linear time history (LTH) analysis. In the
upper half of the frames, however, the peak columns and brace loads approached and even
exceeded the predicted values used in design. This is attributed to the dominance of higher
mode response in that part of the structure, because the behaviour of the structures in these
modes is not influenced much by the flexural plastic mechanism at the base and, thereby,
remains essentially elastic.
Several solutions can be proposed to solve this problem, the first one being to
strengthen the columns and braces in the upper half of the structure, likely on the basis of
member forces associated to elastic response in second and higher modes. A second approach
is to enforce ductile detailing rules for the braces and the columns, such that these members
can withstand some degree of inelastic response. Note that this would not impair the gravity
loading system as the central column and the girders are designed to support all tributary
gravity loads neglecting the support provided by the exterior columns. The structures would
however suffer more extensive damage and permanent deformations. A third option consists
of introducing additional BRB elements at mid-height of the structure to damp higher mode
response, in the same way as the BRBs at the base act in first mode response. To examine the
potential of this solution, one-storey long BRB members were added at the 5
th
and 9
th
floor of
the 8- and 16-storey structures, respectively, similar to the arrangement illustrated in Fig. 2b.
The core plates of these BRBs were sized such that yielding would develop under the same
lateral load as the BRB members at the base, as illustrated in Fig. 2b, thus preventing
overloading of the adjacent columns upon strain hardening of the additional BRB elements.

0 5000 10000 15000 20000
Axial Column Load (kN)
2
4
6
8
10
12
14
16
Design
NLTH (1 BRB)
NLTH (2 BRBs)
LTH
0 1000 2000 3000 4000 5000
2
4
L
e
v
e
l
0 2000 4000 6000 8000 10000
2
4
6
8
L
e
v
e
l
0 2000 4000 6000
Axial Brace Load (kN)
2
4
6
8
10
12
14
16
0 1000 2000 3000 4000
2
4
0 1000 2000 3000 4000 5000
2
4
6
8
0.0 0.4 0.8 1.2 1.6 2.0
Storey Drift Angle (%)
2
4
6
8
10
12
14
16
0.0 0.4 0.8 1.2 1.6 2.0
2
4
0.0 0.4 0.8 1.2 1.6 2.0
2
4
6
8
Fig. 4 Mean+SD value of the peak column axial loads, brace axial loads, and storey drifts
under nonlinear time history analyses (NLTH) and linear time history analyses (LTH).
The results for this alternative design are also plotted in Fig. 4 (2 BRBs), and Table 2
presents the mean+SD of the peak values of key response parameters for all designs: the drift
at the roof level normalised to A
yl
, the maximum inter-storey drift angle (), the number of
records producing yielding in the BRB elements at the base, the plastic deformation and peak
loads for these BRBs, and, lastly, the plastic demand in the additional BRBs. Figure 4 shows
that that the second set of BRBs at mid building height controlled well the column and brace
axial loads in the upper half of the structures. For the braces, the results in the upper levels
seem to confirm the need for and the adequacy of the dynamic amplification factor F
DA
that
was considered in design. More efficient brace design could probably be achieved in the lower
floors if the storey shears were obtained from modal response spectrum analysis instead of the
inverted triangle seismic load distribution that was used herein.
For both designs (1 or 2 BRBs), the computed storey drift angles at all floors are well
below the code limit of 2.5%. For the single BRB configuration, the storey drifts gradually
increase towards the roof level, without significant localised variations, typical of a system
with flexural hinge mechanism at the base, thus confirming the good performance of the
proposed system. The addition of the second set of BRBs allowed the formation of a plastic
hinge at building mid-height, which shifted the drift demand towards the upper floors. As
expected, this behaviour is more pronounced for the 16-storey structure. Nevertheless, the
storey drift angles in the upper portion of these structures remained comparable to those
observed for the other design, still well below the code limit. Of interest is the fact that
maximum deformations are very much the same, regardless of the building height.
In Table 2, all values of A/A
yl
are lower than R
dl
used in design and all plastic
deformations in the BRBs at the base are less than 0.015, which confirms that the extent of
inelastic response was less than expected on the basis of the equal displacement principle
applied to first mode response. The inelastic demand diminishes when the building height
increases, as anticipated in design, but the reduction is more significant for 16-storey frames,
again indicating the significance of higher mode behaviour in taller structures. As a result of
the low plastic demand in the BRBs at the base, limited strain hardening developed, which
resulted in lower force demand in the braces and the columns. It must also be noted that
friction in the BRB elements and allowance for expected yield strength exceeding nominal
value were not included either in the model. Hence, brace and column forces presented in Fig.
4 can be seen as lower bound estimates. Although bending was ignored in the design of the
central column, no occurrence of yielding was observed for that column at BRB locations.
This is mainly due to the fact that only part of the gravity loads were redirected to the central
column upon yielding of the BRBs, resulting in lower axial loads than expected in design and
providing some reserve capacity for the flexural demand.
For the 8- and 16-storey structures, the addition of BRB elements at mid building
height had negligible effect on the peak roof displacements but the storey drift angles
increased in the upper part of the structures, as noted previously. For both frames, the plastic
demand in the base BRBs reduced slightly when the upper BRBs were added. The demand in
these additional BRBs was much higher, partly because their core length L
bc
was shorter (2.0
m vs 5.8 m) and, especially for the 16-storey frame, because large flexural demand developed
at the level where these elements were introduced. In both cases, however, the plastic demand
still remained within acceptable limits.

Table 2 Mean+SD values of peak response parameters
BRBs at the base
BRBs at
upper level n BRB

A/A
yl

(%)

(%) n yield
( )
c
bp

( )
P
b
/P
by
( )
n yield
( )
c
bp

( )
4
8
16
1
1
2
1
2
2.99
2.41
2.34
1.40
1.45
1.46
1.45
1.44
1.22
1.40
10
10
10
8
8
0.0078
0.0066
0.0064
0.0036
0.0031
1.08
1.07
1.07
1.04
1.03
-
-
10
-
10
-
-
0.013
-
0.015
4. CONCLUSIONS

In this paper, an innovative Flexural Braced Frame system was introduced for
enhanced seismic response. The system includes buckling restrained brace elements that
dissipate energy through overall flexure of the bracing bent at its base. Tentative design
guidelines were proposed and validated through nonlinear time history analysis of three
buildings having different heights. The results indicate that the proposed bracing configuration
can exhibit adequate seismic performance, with stable inelastic response and maximum storey
drifts well below the prescribed code limits. Concentration of storey drift demand as observed
in conventional braced steel frames was eliminated, which was the main motivation for
proposing the system. Design forces that are specified for the design of the columns and the
braces in the upper half portion of the structures should be increased to avoid damage to these
elements. These amplified design forces could be based on elastic higher mode response but
this aspect needs to be addressed in future studies. Alternatively, additional BRB elements can
be introduced in the upper floors to dissipate energy induced by higher mode response. This
study showed that such an approach can result in satisfactory performance, although the drift
demand in the upper levels increased due to the formation of a second plastic hinge at the
building mid-height.
It is recommended that further research be conducted to develop a better knowledge
on the behaviour of this structural system and to optimise the design procedure for achieving
robust and cost efficient structures. Particular attention should be given to the need for
designing the central column for flexure at BRB locations and to achieving uniform plastic
strain demand in the BRBs at the base and at building mid-height when two sets of BRB
elements are used. Parameters to be examined in future studies should be the geometry of the
structures and the characteristics of the seismic ground motions. Robustness of the system
should also be assessed using more extensive analysis procedures such as the incremental
dynamic analysis method.

5. ACKNOWLEDGEMENTS

Financial support for this project was provided by the Natural Sciences and
Engineering Research Council of Canada and the Fonds de recherche sur la nature et les
technologies of the Government of Quebec.

6. REFERENCES

[1] Redwood, R.G., Lu, F., Bouchard, G., and Paultre, P. 1991. Seismic Response of
Concentrically Braced Steel Frames. Can. J. of Civ. Eng., 18, 1062-1077.
[2] Tremblay, R. 2000. Influence of Brace Slenderness on the Seismic Response of
Concentrically Braced Steel Frames. In F. Mazzolani and R. Tremblay (ed.),
Behaviour of Steel Structures in Seismic Area; Proc. STESSA 2000 Conf., 527-534,
Montreal, Canada, August 2000. Rotterdam: Balkema.
[3] Tremblay, R. and Robert N. 2001. Seismic Performance of Low- and Medium-Rise
Chevron Braced Steel Frames. Can. J. of Civ. Eng., 28, 4, 699-714.
[4] Sabelli, R. 2001. Research on Improving the Design and Analysis of Earthquake-
Resistant Steel Braced Frames. NEHRP Fellowship Report No. PF2000-9, Earthquake
Engineering Research Institute, Oakland, California.
[5] Sabelli, R., Mahin, S. and Chang, C. 2003. Seismic Demands on Steel Braced Frame
Buildings with Buckling Restrained Braces. Engineering Structures, 25, 655666.
[6] Tremblay, R. 2003. Achieving a Stable Inelastic Seismic Response for Concentrically
Braced Steel Frames. Eng. J., AISC, 40, 2, 111-129.
[7] Khatib, I.F., Mahin, S.A., and Pister, K.S. 1988. Seismic Behavior of Concentrically
Braced Steel Frames. Report UCB/EERC-88/01, Earthquake Eng. Research Center,
University of California, Berkeley, CA.
[8] Tremblay, R. and Tirca, L. 2003. Behaviour and Design of Multi-Storey Zipper
Concentrically Braced Steel Frames for the Mitigation of Soft-Storey Response. In F.
Mazzolani (ed.), Behaviour of Steel Structures in Seismic Area; Proc. STESSA 2003
Conf., 471-478, Naples, Italy, June 2003. Lisse: Balkema.
[9] Watanabe, A., Hitomi, Y., Saeki, E., Wada, A., and Fujimoto, M. 1988. Properties of
Brace Encased in Buckling-Restraining Concrete and Steel Tube. Proc. of 9th World
Conf. on Earthquake Eng., Tokyo, Japan, IV, 719-724.
[10] Iwata, M., Kato, T., and Wada, A. 2000. Buckling-Restrained Braces as Hysteretic
Dampers. In F. Mazzolani and R. Tremblay (eds.), Behaviour of Steel Structures in
Seismic Area; Proc. STESSA 2000 Conf., 33-38, Montreal, Canada, August 2000.
Rotterdam: Balkema.
[11] Tremblay, R., Degrange, G., and Blouin, J. 1999. Seismic Rehabilitation of a Four-
Storey Building with a Stiffened Bracing System. Proc. 8th Can. Conf. on Earthquake
Eng., Vancouver, B.C., 549-554.
[12] Ko, E., Mole, A., Aiken, I., Tajirian, F., Rubel, Z., and Kimura, I. 2002. Application of
the Unbonded Brace in Medical Facilities. Proc. of the 7th U.S National Conf. on
Earthquake Eng., Boston, MA, Paper No. 16.
[13] Ghersi, A., Neri, F., and Rossi, P.P. 2000. Seismic Response of Tied and Trussed
Eccentrically Braced Frames. In F. Mazzolani and R. Tremblay (ed.), Behaviour of
Steel Structures in Seismic Area; Proc. STESSA 2000 Conf., 495-502, Montreal,
Canada, August 2000. Rotterdam: Balkema.
[14] Ghersi, A., Pantano, S., and Rossi. P.P. 2003. On the Design of Tied Braced Frames.
In F. Mazzolani (ed.), Behaviour of Steel Structures in Seismic Area; Proc. STESSA
2003 Conf., 413-419, Naples, Italy, June 2003. Lisse: Balkema.
[15] Takeuchi, T. and Suzuki, K. 2003. Performance-Based Design for Truss-Frame
Structures Using Energy Disspation Devices. In F. Mazzolani (ed.), Behaviour of Steel
Structures in Seismic Area; Proc. STESSA 2003 Conf., 471-478, Naples, Italy, June
2003. Lisse: Balkema.
[16] Tsai, K.C. and Huang, Y.-C. 2002. Experimental Responses of Large Scale Buckling
Restrained Brace Frames. Report No. R91-03, Center for Earthquake Engineering
Research, National Taiwan University, Tai Pei, Taiwan.
[17] Paulay, T. and Priestley, M.J.N. 1992. Seismic Design of Reinforced Concrete and
Masonry Buildings. John Wiley & Sons, Inc. New York, NY, 744 p.
[18] Heidebrecht, A.C. 2003. Overview of NBCC 2005 Seismic Provisions. Canadian
Journal of Civil Engineering, 30, 2, 241-254.
[19] CSA. 2001. CSA Standard S16-01, Limit States Design of Steel Structures. Canadian
Standard Association, Toronto, ON.