This article was published in an Elsevier journal.

The attached copy

is furnished to the author for non-commercial research and
education use, including for instruction at the authors institution,
sharing with colleagues and providing to institution administration.
Other uses, including reproduction and distribution, or selling or
licensing copies, or posting to personal, institutional or third party
websites are prohibited.
In most cases authors are permitted to post their version of the
article (e.g. in Word or Tex form) to their personal website or
institutional repository. Authors requiring further information
regarding Elseviers archiving and manuscript policies are
encouraged to visit:

http://www.elsevier.com/copyright
 Author's personal copy

Engineering Structures 30 (2008) 10921100

www.elsevier.com/locate/engstruct

Design and behavior of zipper-braced frames

Chuang-Sheng Yang , Roberto T. Leon, Reginald DesRoches
School of Civil and Env. Engineering, Georgia Tech, Atlanta, GA 30332-0355, USA

Received 2 March 2007; received in revised form 30 May 2007; accepted 18 June 2007
Available online 15 August 2007

Abstract

This paper proposes a design methodology for zipper-braced frames aimed at achieving ductile behavior. Three zipper-braced models were
designed on the basis of the proposed design procedure to carry the same masses as the 3-, 9-, 20-story SAC model buildings with moment-resisting
frames designed for the Los Angeles area. Pushover analyses of the models were performed to estimate the overstrength, inelastic strength and
deformation capacities for the entire structures, and assess the sequence of yielding and buckling in the members. The performance of the models
was also evaluated using nonlinear dynamic analyses under an ensemble of 2%-in-50-year pulse-type near-fault ground motions. The analyses
indicate that the design procedure produces safe designs, with the design becoming more conservative as the number of stories increases. The
distribution of interstory drifts demonstrates the efficiency of the zipper struts in achieving uniform damage over the height of the structure, and
generally satisfies allowable interstory drift ratio limits.
c 2007 Elsevier Ltd. All rights reserved.

Keywords: Steel frames; Seismic design; Steel; Struts; Nonlinear analysis; Earthquake resistance structures; Structural models

1. Introduction
Braced frames are economical and efficient structures for
resisting lateral loads. A typical braced frame configuration is
the so-called inverted chevron-braced (inverted-V-braced) one
shown in Fig. 1(a), which provides an opening in the middle
of the story and is thus preferred by architects and owners over
the concentrically braced options. In general, the performance
of these systems is governed by the buckling behavior of the
inclined members in compression. For wind design, it is typical
to assume a tension-only behavior and the compressive strength
Fig. 1. (a) Inverted-V-braced configuration; (b) formation of a soft story
of these braces is ignored. For seismic design, where the design
mechanism.
is controlled by large cyclic drifts and the need for energy
dissipation, the alternating buckling and yielding of the braces
To counteract the tendency of chevron-braced frames to
leads to poor hysteretic behavior, the formation of a soft story
form soft story mechanisms in the first floor, Khatib et al. [1]
mechanism [Fig. 1(b)] and associated potential collapse. Thus
proposed the addition of zipper columns (zipper struts) between
braced frames have traditionally not been considered a suitable
the brace locations at the midspan of floor beams. These
system in high seismic areas unless the buckling and yielding
zipper struts transfer the unbalanced vertical forces at this
are controlled in the design through the use of both bracing
members with low slenderness both at the local and global location induced by buckling of the braces into the stories
levels, and large beams to sustain the unbalanced vertical forces above. The result is the formation of a full-height collapse
resulting from brace buckling. mechanism that provides substantial additional strength and
ductility to an otherwise brittle system. Although this system
Corresponding author. Tel.: +1 404 395 9214. has been mentioned in the AISC Seismic provisions for several
E-mail address: cs.walter.yang@gmail.com (C.-S. Yang). editions [2], a comprehensive design procedure has not been

c 2007 Elsevier Ltd. All rights reserved.

0141-0296/$ - see front matter

doi:10.1016/j.engstruct.2007.06.010
 Author's personal copy
C.-S. Yang et al. / Engineering Structures 30 (2008) 10921100
Fig. 2. Expected behavior and performance of zipper frames.
available. Khatib et al. [1] proposed that the design force for the
zipper column at any story be taken as the minimum of either
a square root of the sum of the squares (SRSS) approximation
of the unbalanced vertical forces that can be transmitted from
the stories below or a SRSS approximation of the unbalanced
vertical forces at the stories above and the story under
consideration. Tremblay and Tirca [3,4] recently proposed
another design methodology for zipper struts based on a full-
height zipper mechanism by considering different scenarios of
brace buckling sequences and subsequent force redistribution.
However, their analytical studies showed that instability and
collapse occurred when the frames were subjected to severe
near-fault earthquake motions. These failures followed the
buckling of all compression braces, as the large resulting
unbalanced vertical forces were reapplied to the floor beams,
which were not originally designed to resist these forces.
To overcome this behavior, Leon and Yang [5] proposed a
design procedure using a partial-height zipper mechanism and
a hat truss system in the top story. The latter was used to prevent
the formation of a full collapse mechanism and to redirect
the unbalanced vertical forces into the exterior columns.
Several experiments on reduced-scale zipper frame specimens
designed according to the preliminary design procedure have
been performed by a variety of testing methods [612]. The
experimental results showed excellent strength and ductility
behavior for the zipper-braced frames. These results were
consistent with those of refined 2D and 3D numerical
simulations conducted using the Open System for Earthquake
Engineering Simulation (OpenSEES) [13], an open-source
platform widely used in the USA for this type of studies.
This paper proposes a refined design methodology for
zipper-braced frames with partial-height zipper mechanisms in
view of the experimental results from this collaborative project.
Three zipper-braced models proportioned according to the new
design methodology are studied to validate the adequacy of the
design based on the performance of both individual members
(zipper struts and top-story braces) in terms of strength and of
the overall frame in terms of interstory drifts.
2. Design methodology for zipper-braced frames
2.1. Design philosophy
The basic design objective for a zipper-braced frame is
to mitigate the typical soft-story mechanism associated with
braced frames by distributing more uniformly both story drift
and energy dissipation over the height of the building. There
1093
are three main components of this innovative system. The
first component is the zipper strut which forces simultaneous
buckling of all stories except the top one, and leads to tension
yielding of all braces [Fig. 2(a)]. The second component is
the hat truss, which prevents the formation of a full plastic
mechanism and thus provides large deformation capacity.
The third component consists of the exterior columns, which
transmit the forces back to the foundation.
From the standpoint of base shear versus roof drift
response, a properly designed zipper-braced frame should
exhibit trilinear response with large ductility, as shown in
Fig. 2(b). As presented in Fig. 2(a), simultaneous buckling
in the compression braces significantly reduces the initial
stiffness of the frame. However, the strength of the whole frame
continues to increase, first reaching its yielding strength and
then entering a hardening range due to the yielding in the
tension braces. The proposed design provisions are presented
next, including a short commentary.
2.2. Design procedure
The design of zipper-braced frames shall consist of a two-
step procedure. The first is a strength design phase for the
braces in which the presence of the zipper elements is ignored.
The second is a capacity design phase in which the zipper struts
are added and other structural elements are redesigned except
for the braces below the top-story level.
2.2.1. Phase I (strength design)
The braces shall be designed, to resist the effects of
earthquake and vertical loadings, from the load combinations
stipulated by the Applicable Building Code without the aid of
the zipper struts.
Commentary: This phase follows the conventional Special
Inverted-V-Braced Frame (SIVBF) design procedure. The
braces are assumed to resist all the lateral loads, with the
critical compression braces designed to carry a force equal to
c A g Fcr . This phase fixes the sizes of the braces in all stories
except the top story. Preliminary design for the other elements
should be carried out, but the design of the beams shall ignore
the shear force at the centerline resulting from the unbalanced
forces induced by the braces.
2.2.2. Phase II (capacity design)
In this phase, the zipper struts are added and other structural
elements are redesigned except for the braces below the
top-story level. The frame designed in Phase I is further
 Author's personal copy
1094 C.-S. Yang et al. / Engineering Structures 30 (2008) 10921100
modified according to the following sequence, which follows
the redistribution of the unbalanced vertical forces.
1. Zipper struts
Zipper struts shall be designed to resist the vertical
unbalanced forces generated by the braces and zipper
element located on the level below. The brace forces shall
be taken as R y Fy A g for the brace in tension and 0.3 times
Pn for the brace in compression.
Commentary: The zipper strut is designed to resist all
the vertical forces from the members framing from below.
In order to channel most of the unbalanced vertical force
from the braces into the story above, the beam needs to be
flexible. As a result, the contribution of the shear capacity of
the beams to overcoming the unbalanced vertical forces is
ignored in this design step.
2. Top-story braces
The top-story braces from Phase I shall be redesigned
to resist the forces corresponding to (1) the smaller of all
the vertical unbalanced forces collected in the top story
zipper element or its yield strength, and (2) the overstrength
equivalent lateral earthquake force (0 Q E ) at the roof level,
with 0 equal to 2.0.
Commentary: The top story is intended to remain elastic
and thus needs to be designed for the maximum vertical
forces that can be delivered by the zipper elements from
below and a conservative estimate of the horizontal inertial
forces from the roof. The overstrength factor, o , is taken as
2 for the zipper-braced system (same as the code-stipulated
overstrength factor for special CBFs).
3. Columns
The required axial compressive and tensile strengths
are determined using the maximum load transferred to the
column considering the capacities of the adjacent braces
(using the required strengths instead of capacities for the
top story braces) and the effect of the unfactored vertical
loading.
Commentary: In the process of design, each story column
is assumed to be pinned at its ends, ignoring the flexural
capacity provided in real continuous columns. This process
will result in fairly uniform column sizes over the height of
the structure. The resulting sizes of the first-story columns
are similar to those of conventional special inverted-V-
braced frames.
4. Beams
Beams shall comply with beamcolumn provisions as
stipulated in Chapter H of the AISC LRFD Specification.
The required axial strength shall be determined using the
maximum load transferred to the beam considering the
capacities of the adjacent braces and the effect of the
unfactored vertical loading. The required bending strength
shall be determined using the maximum bending moment
by the effect of the unfactored vertical loading.
Commentary: (use the required strengths instead of
capacities for the top-story braces).
5. Bracing connections
Table 1
Member sizes for the 3-story zipper-braced bay
Story Brace Column Beam Zipper strut
3 W14 132 W12 96 W8 58 W12 96
2 HSS8 8 1/2 W12 96 W10 88 W12 45
1 HSS8 8 5/8 W12 96 W10 88
The design for the bracebeamcolumn connections shall
follow the Uniform Force method described in the AISC
LRFD manual.
Commentary: In order to make all the braces, except the
top-story ones, buckle out of plane, the gusset plates at the
ends of braces should be designed with a free length of twice
the gusset plate thickness, as recommended in the AISC
seismic provisions. However, the yield line does not need to
be perpendicular to the axial line of the bracing member. A
tapered gusset plate is recommended for the brace-to-beam
connection to provide both flexibility for out-of-plane brace
buckling and larger ductility for in-plane brace yielding.
3. Analytical models
SAC was a joint venture of SEAoC (Structural Engineers
Association of California), ATC (Applied Technology Coun-
cil), and CUREE (California Universities for Research in Earth-
quake Engineering) that studied steel connections and structural
systems in the aftermath of the Northridge and Kobe earth-
quakes. Theme structures designed as part of SAC are com-
monly used as references when assessing performance of new
or improved structural systems. Three office buildings with
zipper-braced systems were designed to carry the same masses
and use the same number of seismic-resisting bays as the 3-,
9-, and 20-story SAC moment-resisting frames (6, 6, and 10
seismic-resisting bays for the 3-, 9-, and 20-story models in the
NS direction, respectively) [14] designed for downtown Los
Angeles. In the Los Angeles area, for the 2% probability of ex-
ceedance in 50 years, the mapped spectral accelerations for the
short period and the 1 s period are 2.16g and 0.72g, respec-
tively, with a PGA of 0.90g. The design code documents used
were the 2005 ASCE-7 for loads and the AISC LRFD 2005 for
member and frame design. The buildings were designed as if
located on stiff soil (site class D as per ASCE 7-05 [15] defini-
tions). An importance factor of 1.5 was assigned to the build-
ings in accordance of Occupancy Category IV. This is a sig-
nificant departure from the SAC buildings, and was intended
to determine if zipper-braced frames could be applicable even
to critical structures. The response modification coefficient, R,
for this type of seismic force-resisting system was taken as 6,
consistent with other ductile braced systems (special steel con-
centrically braced frames). The members selected for the 3-, 9-,
and 20-story zipper-braced bays are shown in Tables 1, 2, 3,
respectively. The elevations of the three models are shown in
Fig. 3.
4. Analytical study
The models described above were used to investigate the
behavior and performance of the zipper-braced models by
 Author's personal copy

C.-S. Yang et al. / Engineering Structures 30 (2008) 10921100 1095

Fig. 4. Effect of initial imperfection.

For the zipper-braced models under a set of pushover loads,

the pushover curve can be approximated by a trilinear curve that
provides three base shear levels in addition to the design seismic
base shear: a buckling base shear, a yielding base shear, and an
ultimate base shear. These values are used to distinguish the
Fig. 3. Zipper-braced models: (a) 3-story bay; (b) 9-story bay; (c) 20-story bay. structural strengths at which the braces begin to buckle, yield,
and approach the ultimate strength, respectively.
Table 2 The nonlinear dynamic analyses consisted of a suite of 20
Member sizes for the 9-story zipper-braced bay ground motions (LA21LA40) [16] representative of a 2%-in-
Story Brace Column Beam Zipper strut 50-year probability of exceedence for the Los Angeles area.
9 W14 398 W14 257 W10 68 W14 398
4.1. Brace model
8 HSS7 7 1/2 W14 257 W14 257 W14 370
7 HSS8 8 1/2 W14 257 W10 88 W14 342
The original brace model developed for the OpenSEES
6 HSS8 8 5/8 W14 283 W10 88 W14 283
5 HSS9 9 5/8 W14 283 W12 96 W14 233 program by Uriz and Mahin [17] at UC Berkeley had an
4 HSS9 9 5/8 W14 311 W12 96 W14 193 additional node at midspan and included a small initial
3 HSS10 10 5/8 W14 311 W12 96 W14 132 imperfection (L w /2000) to simulate the buckling behavior of
2 HSS10 10 5/8 W14 342 W12 96 W14 82 a brace under compression. This brace model predicts the
1 HSS12 12 5/8 W14 342 W12 96
monotonic response of the zipper frame model to pushover
loading as an approximate bilinear curve rather than a trilinear
Table 3 skeleton curve observed from tests [7,8]. In order to obtain a
Member sizes for the 20-story zipper-braced bay more accurate simulation of the trilinear behavior, the model
was modified in two ways. First, a larger initial imperfection
Story Brace Column Beam Zipper strut
ratio for the brace was adopted to lower the maximum
20 W14 550 W14 68 W10 30 W36 800 compression strength of the brace. Second, two rotational
19 HSS5 5 5/16 W14 426 W24 279 W14 730
18 HSS5.5 5.5 5/16 W14 426 W10 30 W14 730
springs were added at the end nodes of the brace to increase its
17 HSS6 6x3/8 W14 426 W10 30 W14 665 maximum compression and minimum post-buckling strengths.
16 HSS6 6x1/2 W14 426 W10 39 W14 665
15 HSS6 6x1/2 W14 426 W10 39 W14 605 4.1.1. Effect of initial imperfection on the brace buckling
14 HSS6 6x5/8 W14 455 W10 45 W14 550 Fig. 4 shows monotonic curves for a compression brace with
13 HSS6 6 5/8 W14 455 W10 45 W14 550 four different initial imperfection ratios (L w /100, L w /150, L w /
12 HSS7 7 1/2 W14 455 W10 45 W14 500
1000, and L w /2000) and a fixed stiffness of rotational springs
11 HSS7 7 1/2 W14 455 W10 45 W14 455
10 HSS7 7 1/2 W14 455 W10 45 W14 398 [79 m kN/rad (700 in. kips/rad)]. The curves show the ax-
9 HSS7 7 1/2 W14 455 W10 45 W14 370 ial strength (Pmodel ) and deformation (model ) predicted by
8 HSS8 8 1/2 W14 500 W12 50 W14 342 the OpenSEES model normalized to the nominal compression
7 HSS8 8 1/2 W14 500 W12 45 W14 283 strength (Pn ) given by the AISC LRFD manual and to the yield-
6 HSS8 8 1/2 W14 500 W12 45 W14 233
ing displacement (yield = y L w /E). The case of L w /150 has
5 HSS8 8 1/2 W14 550 W12 45 W14 193
4 HSS8 8 1/2 W14 550 W12 45 W14 159 the peak strength ratio closest to 1 among the four different
3 HSS8 8 1/2 W14 550 W12 45 W14 109 ratios. It is also important to note that all the four cases generate
2 HSS8 8 1/2 W14 605 W12 45 W14 68 the same post-buckling strength at a displacement ratio of 20.
1 HSS9 9 5/8 W14 605 W12 50
4.1.2. Effect of rotational stiffness on the brace buckling
means of both nonlinear static analyses (pushover analyses) and Fig. 5 shows the monotonic curves for a compression
nonlinear dynamic analyses using the OpenSEES program. brace with four different values of rotational stiffness
 Author's personal copy
1096 C.-S. Yang et al. / Engineering Structures 30 (2008) 10921100
Fig. 5. Effect of rotational stiffness.
Table 4
Stages defined for the zipper-braced models
Model Stage 1 Stage 2 Stage 3
0b (%) b y (%) y o (%)
3-story 00.20 0.200.68 0.6811.05
9-story 00.41 0.411.53 1.536.77
20-story 00.83 0.833.41 1.538.47
[79 m kN/rad (700 in. kips/rad), 475 m kN/rad (4200 in. kips/
rad), 712 m kN/rad (6300 in. kips/rad), 1130 m kN/rad
(10 000in. kips/rad)] and a fixed initial imperfection ratio
(L w /150). In the peak portion of the curves, the maximum
compression strength increases slightly with increasing
stiffness of the rotational spring. However, there is a significant
increase in the post-buckling strength when a larger value
of rotational stiffness is used. For the nominal compression
strength (Pn ) and minimum post-buckling strength (0.3Pn
at 1020 times the yielding displacement, AISC Seismic
Provisions), the curves with the rotational stiffness of
475 m kN/rad provided the best match to the behavior
envisioned by current provisions.
4.1.3. Iterative approach for the brace model
The proposed brace model is semi-empirical due to two un-
known parameters: (1) the initial imperfection of a brace, and
(2) the rotational stiffness of the springs that simulate the re-
straint provided by the gusset plates. To match the behavior ob-
served in the brace tests at several of the NEES sites involved
in this project, a two-step process was followed. First, a large
value for the initial imperfection ratio (say L w /500) was se-
lected. Next, a reasonable value for the stiffness of the rotational
springs based on the out-of-plane stiffness for the gusset plate
was chosen. Then the OpenSEES program was used to deter-
mine the hysteretic curve for the brace. Subsequently, the rota-
tional stiffness was adjusted for the next run and iterations were
continued until the numerical maximum compression strength
was in agreement with the compressive strength (Pn ) of a brace
which was computed by using the expected yielding strength
and k = 1.0, where k is the effective length factor. Finally, the
minimum post-buckling strength was compared with the speci-
fied minimum post-buckling strength of 0.3Pn . If a large differ-
ence existed, then a new value for the initial imperfection was
chosen and the subsequent iterations repeated.
4.2. Performance assessment
To assess the performance of the frames, three measures
were used. First, the demand forces in the zipper struts and top-
story braces were compared to the required forces computed
in the course of design. This represents a measure of the
design procedures ability to handle member demands. Second,
the interstory drift demand ratios were evaluated statistically.
Finally, the PGAs for the ground motion ensemble were further
scaled so as to match the design spectrum at the fundamental
period of each structure, and the nonlinear analyses were
repeated. The statistical values of the 84th percentile [14] of
the peak interstory drift ratios for the modified ground motion
ensemble were then compared to the allowable interstory drift
ratio limit specified in ASCE 7-05.
4.2.1. Pushover analysesglobal lateral load vs. drift behav-
ior
A 2D pushover analysis was performed using OpenSEES
to evaluate the maximum strength and deformation capacities
for each zipper-braced bay designed according to the proposed
design procedure. The resulting pushover curve for the 3-story
zipper-braced model, plotted as the roof drift ratio vs. the
base shear coefficient C (defined as the base shear divided by
the seismic weight per one braced bay), is shown in Fig. 6
at two different scales of the drift value. Fig. 6(a) indicates
a very ductile structure and can be well approximated by a
trilinear curve with three stages as shown in Table 4. In stage 1,
the structure behaves as linearly elastic until the first buckling
(Point B) occurs in a compression brace [Fig. 6(b)]. The
maximum base shear in this stage is called the buckling base
shear, corresponding to Cb = 0.55 and 2643 kN (594 kips).
This is larger than the design seismic base shear (Point S) of
1736 kN (390 kips, Cs = 0.36) by a buckling overstrength
factor, b , of 1.52. The buckling overstrength factor, b , is
defined as the buckling base shear divided by the design seismic
base shear. Note that the design seismic base shear was the
summation of the equivalent lateral forces used in the course
of design. In stage 2, the compression braces buckle and their
capacities tend towards the minimum post-buckling strength,
while the tension braces continue to attract load until their
tension yield capacities are reached (point Y). The yielding base
shear is 3916 kN (880 kips, C y = 0.81) and the corresponding
overstrength y is 2.26. After yielding in the tension braces, the
base shear enters a hardening range and approaches the ultimate
base shear of 4223 kN (949 kips, Co = 0.88) at the roof drift
ratio o = 11.05%. The structural ultimate overstrength o is
2.43.
According to Table 12.2-1 in ASCE 7-05, the response
modification factor R for a ductile braced frame is 6. The elastic
base shear is the multiplication of the design base shear by a
factor of R/I , leading to 6942 kN (1560 kips, Ceu = 1.44) at an
elastic roof drift ratio of e = 0.53% (Point E). The triangular
area under the dashed line shown in Fig. 6(b) corresponds to the
total elastic input energy, or energy demand (E d ) of 218 m kN
(1929 in. kips). In order to provide a conservative structure, the
energy capacity (E c ), which can be viewed as the area under
the pushover curve (roof drift vs. base shear), should be greater
 Author's personal copy

C.-S. Yang et al. / Engineering Structures 30 (2008) 10921100 1097

Fig. 6. Pushover curves for the zipper-braced models.

than the energy demand. This requires that the structure should
remain stable until the roof drift ratio reaches 0.68% defined
as m [E c = 219 m kN (1940 in. kips)], which is almost
the same as the yielding roof drift ratio y = 0.68% and far
away from the ultimate roof drift ratio o . If the area under the
pushover curve up to the ultimate drift ratio is considered, E c
is 4456 m kN (39 423 in. kips) or about 20.4 times that from an
equivalent elastic system. This demonstrates the considerable
overstrength and deformation capacity of the zipper system.
The resulting pushover curves for the 9- and 20-story
zipper-braced bays are shown in Fig. 5(c)(e). The relevant
information, such as the different base shear levels, and
corresponding roof drift ratios and overstrength factors, is listed
in Table 5.
4.2.2. Nonlinear dynamic analyses
Three 3-, 9-, and 20-story zipper-braced models studied
for nonlinear dynamic analyses were the same as those used
 Author's personal copy

1098 C.-S. Yang et al. / Engineering Structures 30 (2008) 10921100

Table 5
Member sizes for the prototype zipper frame

Model Point S Point B Point Y (m ) Point O Point E Seismic weight (kN)

Cs (s ) Cb (b ) C y ( y ) Co (o ) Ceu (e )
b y o
3-story 0.36 (0.13%) 0.55 (0.20%) 0.81 (0.68%) (0.68%) 0.88 (11.05%) 1.44 (0.53%) 4821
1.52 2.26 2.43
9-story 0.20 (0.30%) 0.27 (0.41%) 0.35 (1.53%) (1.53%) 0.37 (6.77%) 0.74 (1.10%) 14 712
1.32 1.71 1.81
20-story 0.14 (0.65%) 0.17 (0.83%) 0.21 (3.41%) (3.43%) 0.24 (8.47%) 0.49 (2.34%) 10 648
1.27 1.56 1.75

Table 6 Table 8
Periods for the zipper-braced models Strengths for the top-story braces

Model First mode (s) Second mode (s) Third mode (s) Model Demand compression force Required compression force(kN)
(kN)
3-story 0.34 0.11 0.07
9-story 1.09 0.35 0.18 3-story 4 883 4 895
20-story 2.57 0.68 0.34 9-story 15 219 18 628
20-story 10 311 26 535

Table 7
Factors of the effective model masses for the zipper-braced models unbalanced vertical forces as expected in the design procedure.
Model First mode Second mode Third mode
3-story 0.88 0.08 0.03
4.2.4. Verification of adequacy of the top-story brace design
9-story 0.76 0.17 0.04
20-story 0.67 0.21 0.06 The demand compression forces, or the maximum peak
compression forces, for the top-story braces are shown in
Table 8, together with the required compression forces obtained
in the pushover analyses. In order to provide a conservative
from the design process. All the demand forces for the top-story
assessment of structural performance, 5% Rayleigh damping
braces are less than the corresponding required forces. The
was specified in the first and last modes of vibration, which
demand forces, particularly for the 3- and 9-story zipper-braced
ensures that the damping ratios between the two extreme modes
models, are close to the required ones. This demonstrates that
are less than 5%. The results from eigenvalue analyses are
the design approach for the top-story braces is adequate for the
shown in Table 6. The effective modal masses for the first three
3- and 9-story models and conservative for the 20-story model.
vibration modes are shown in Table 7.

4.2.3. Verification of adequacy of the zipper strut design
In order to verify the appropriateness of the zipper strut
forces obtained from the design process, the envelopes of
forces in the zipper struts were examined. The normalized peak
tension and compression forces acting in the zipper struts for
each zipper-braced model under an ensemble of ground motion
recorders, LA21 to LA40, are shown in Fig. 7. The peak tension
forces are normalized to the required tension forces, Treq0 d ,
(critical forces for selecting sizes of the zipper struts) obtained
from the design process, whereas the peak compression forces
are normalized to the design compression forces, Pn , of the
given zipper struts. For the 3- and 9-story zipper-braced models,
the ratios in the tension side are close to and less than 1, and
the ratios in the compression side are greater than 1. This
demonstrates that the design procedure provides zipper struts
with conservative and well-predicted required strengths. For
the 20-story zipper-braced model, all ratios are between 1
and 1, still providing zipper struts with conservative required
strengths. Generally, the first mode will not dominate the
behavior of a high-rise building. This implies that not all the
compression braces will buckle at the same time. As a result, the
zipper struts in the 20-story structure did not sustain as much the
4.2.5. Performance of interstory drifts
The allowable interstory drift ratio limits from Section
16.2.4.3 in ASCE 7-05 are 1.875%, 1.250%, 1.250% (or 125%
of the allowable interstory drift ratios specified in Section
12.12.1 in ASCE 7-05) for the 3-, 9-, and 20-story zipper-
braced models, respectively. For comparison to these values,
the ensemble of the ground motions (LA21LA40) was further
scaled to match the design spectral response acceleration at the
structure fundamental period (1.44g and 0.66g for the 3- and
9-story zipper-braced models, respectively). The resulting peak
interstory drift ratios for the modified ensemble of accelerations
are presented as circular symbols in Fig. 8. Characteristic
values (i.e, median and 84th percentile) as described in FEMA
355C [14] are also shown. The median is defined as the
geometric mean of the data points, and 84th percentile as the
median times the exponent of the standard deviation of the
natural log of the data points.
For the 3-story zipper-braced model, as shown in Fig. 8(a),
both the median and 84th percentile lines show a fairly uniform
distribution of peak interstory drifts over the height and are
within the specified interstory drift ratio limit of 1.875%. The
performance of the interstory drift ratio for the 3-story zipper
 Author's personal copy
C.-S. Yang et al. / Engineering Structures 30 (2008) 10921100
Fig. 7. Normalized peak tension and compression forces in the zipper struts
(LA21LA40).
frame is satisfactory, with the 84th percentile values within the
limit of 1.875%.
For the 9-story zipper-braced model, as shown in Fig. 8(b),
the median line also shows a uniform distribution of interstory
drifts over the height and is within the specified interstory drift
ratio limit of 1.25%. The 84th percentile line presents a similar
pattern as the median one, but the statistical value appearing
in the second story is about 1.6%, significantly larger than the
1099
Fig. 8. Peak interstory drift ratios for the 3-, and 9-story zipper-braced models
under the scaled ground motion suite.
code limit of 1.25%. This is because all the damping ratios with
not more than 5% resulted in overestimation of the demands of
the peak interstory drift ratios.
The peak interstory drift ratios for the 20-story zipper-
braced model under the unmodified or original ensemble
of accelerations are presented also as circular symbols in
Fig. 9. Both the median and 84th percentile lines exhibit fairly
uniform distribution over the height of the structure. Even
though the model with the damping ratios of all modes not
greater than 5% was subjected to the severe ground motion
ensemble, the maximum peak interstory drift ratio was less
than 4%. The zipper struts mitigated the concentration of
deformation in some stories and efficiently completed uniform
story distribution over the height. The ground motion recorders
were not further scaled because the design spectral response
acceleration at the fundamental period for this model is small
and 0.28g.
5. Summary and conclusions
Three zipper-braced models ranging from low- to high-rise
buildings were analyzed using both pushover and nonlinear
dynamic analyses in an effort to evaluate the adequacy
of the proposed design procedure for zipper frames and
 Author's personal copy
1100 C.-S. Yang et al. / Engineering Structures 30 (2008) 10921100
Fig. 9. Peak interstory drift ratios when the 20-story zipper frame was under
the original ground motion suite.
to understand better the performance of such frames. The
following conclusions are drawn based on the results and
observations presented herein.
1. The design procedure results in zipper struts and top-story
braces with reasonable strengths in low- and moderate-
rise buildings, and with conservative strengths in high-rise
buildings. The effects of higher modes, which become more
important as the number of stories increases, appear to
account for the increased conservatism of the procedure as
the number of stories increases.
2. The statistical values of the peak interstory drift ratios show
a uniform distribution over the height in low-, moderate-,
and high-rise buildings, demonstrating the efficiency of the
zipper struts in achieving great structural efficiency.
In general, the interstory drifts satisfy the allowable interstory
drift limits specified in 2005 ASCE 7.
Acknowledgments
This work was carried out under the NSF Grant 0324542 to
Georgia Tech. The opinions given and results described herein
are the exclusive responsibility of the senior author and neither
reflect the views of any of the universities involved, nor that
of the National Science Foundation. This project is part of a
NEESR collaborative project with the University at Buffalo
(Dr. A. Reinhorn and M. Schacter), the University of California
at Berkeley (Dr. Stojadvonivic and T. Yang), the University
of Colorado at Boulder (Dr. B. Shing and A. Stavridis), and
Florida A&M (Dr. A. Makollah).
References
[1] Khatib IF, Mahin SA, Pister KS. Seismic behavior of concentrically
braced steel frames. Report no. UCB/EERC-88/01. Berkeley: Earthquake
Engineering Research Center, University of California; 1988.
[2] American Institute of Steel Construction (AISC), AISC Seismic. Seismic
provisions for structural steel buildings. Chicago; 2005.
[3] Tremblay R, Tirca L. Behavior and design of multi-story zipper
concentrically braced steel frames for the mitigation of soft-story
response. In: Proceedings of the conference on behaviour of steel
structures in seismic areas. 2003. p. 4717.
[4] Tremblay R, Tirca L. Influence of building height and ground motion
type on the seismic behavior of zipper concentrically braced steel frames.
Proceedings of the 13th world conference on earthquake engineering,
2004.
[5] Leon RT, Yang CS. Special inverted-v-braced frames with suspended
zipper struts. In: International workshop on steel and concrete
composite construction. Taipei (Taiwan) National Center for Research on
Earthquake, 2003.
[6] Yang CS, Leon RT, DesRoches R. On the development of zipper frames
by quasi-static testing and pushover analyses. In: Proceedings of the 8th
U.S. national conference on earthquake engineering. 2006.
[7] Yang CS, Leon RT, DesRoches R. On the development of zipper frames
by pushover testing. In: The fifth behavior of steel structures in seismic
areas conference. 2006.
[8] Yang CS, Leon RT, DesRoches R. Pushover test and analysis of a braced
frame with zipper struts. ASCE Journal of Structural Engineering; 2006
[submitted for publication].
[9] Yang CS, Leon RT, DesRoches R. Cyclic behavior of zipper-braced
frames. ASCE Journal of Structural Engineering; 2007 [submitted for
publication].
[10] Schachter M, Reinhorn A. On the importance of 3D analysis for zipper
frames structures. In: Proceedings of the 8th U.S. national conference on
earthquake engineering. 2006.
[11] Yang TY, Stojadinovic B, Moehle J. Hybrid simulation evaluation of
innovative steel braced frame system. In: Proceedings of the 8th U.S.
national conference on earthquake engineering. 2006.
[12] Stavridis A, Shing PB. Validation of a fast hybrid test system with
substructure tests. In: 17th analysis & computation specialty conference.
2006.
[13] Mazzoni S, McKenna F, Scott MH, Fenves GL. OpenSEES command
language manual. Pacific Earthquake Engineering Research Center; 2006.
< http://opensees.berkeley.edu >.
[14] FEMA-355C. State of the art report on systems performance of
steel moment frames subjected to earthquake ground shaking. FEMA
355C/September 2000. Washington (DC): Building Seismic Safety
Council; 2000.
[15] Minimum design loads for buildings and other structures. American
Society of Civil Engineers, ASCE/SEI 7-05; 2006.
[16] Somerville PG, Smith N, Punyamurthula S, Sun J. Development of ground
motion time histories for phase 2 of the FEMA/SAC steel project. SAC
background document, report no. SAC/BD 97/04; 1997.
[17] Uriz P, Mahin S. Summary of test results for UC Berkeley special con-
centric braced frame specimen No. 1 (SCBF-1), http://www.ce.berkeley.
edu/patxi/SCBF/publications/PrelimSCBFtestResults.pdf; 2004.