You are on page 1of 37



Doctoral Thesis Proposal

Davide Giannuzzi

Prof. Roberto Ballarini

Table of Contents

Abstract ....................................................................................................................................................... 1

1 Background and Motivation .............................................................................................................. 2

2 Proposed System ................................................................................................................................. 6

2.1 Description ..................................................................................................................................... 6

2.2 Literature Review.......................................................................................................................... 8

3 Preliminary Analysis......................................................................................................................... 13

3.1 Conceptual Design ...................................................................................................................... 13

3.1.1 Reduced model.................................................................................................................... 13

3.1.2 Finite Element Models ....................................................................................................... 14

3.1.3 Elastic buckling ................................................................................................................... 15

3.1.4 Cyclic nonlinear response .................................................................................................. 17

3.1.5 Plastic dissipation ............................................................................................................... 21

3.2 Evaluation of Performance ........................................................................................................ 21

4 Conclusion and Future Work .......................................................................................................... 28

4.1 Experimental Program ............................................................................................................... 28

4.2 FE Model Validation and Refinement...................................................................................... 32

4.3 Performance evaluation ............................................................................................................. 33

References ................................................................................................................................................. 34
The conceptual design of an innovative seismic resistant steel framing system capable of
providing stiffness and ductility to new or existing structures is presented. The proposed
bracing system consists of concentric X-braces connected in series with rectangular sacrificial
shear panels. The braces are designed to remain elastic during seismic events, while the shear
panels are sized and configured to dissipate ample energy through plastic deformation-induced
stable hysteretic behavior.
Detailed three-dimensional nonlinear finite element analyses are performed to
characterize and quantify the effects of the design parameters on the local response of the
bracing system, and to adjust the design so that potential buckling of the elements is mitigated.
The finite element predicted force-displacement curves of bracing systems that achieve the
desired local behavior when subjected to a specified inter-story drift are in turn translated into a
SAP2000 nonlinear link element. Embedment of the link element in a two-dimensional steel
frame model enables the assessment of the performance of the bracing system as applied to a
seven-story steel frame subjected to different intensity levels of seismic excitation. The results
demonstrate that the braced ductile shear panel framing system offers promise for decreasing
the lateral displacements of structures subjected to earthquakes, while minimizing damage to
all structural elements other than the sacrificial panels.
The experimental program aimed at collecting data to calibrate and validate the detailed
finite element model is outlined at the end, together with a description of further analyses and
investigations that will be carried out once more reliable and robust data had been made
available. The data obtained will also be used to include a damage and failure criterion in the
finite element model to predict the actual ductility and energy dissipation capacity of different
geometries. Future investigations will evaluate the performance of the bracing system in two-
and three-dimensional frame structures following current code provisions, and they will be
employed to estimate the over-strength factor, the ductility demands and response modification
1 Background and Motivation
Structures in areas associated with high seismic activity are required to resist strong
lateral forces resulting from earthquake triggered ground accelerations. In a typical scenario a
building or any other structure rising above ground must resist smaller, frequent
earthquakes as well as large, less frequent earthquake. The structure must have enough strength
and stiffness to resist frequent events with little or no damage, but it also should possess
adequate ductility or another form of damping to economically and efficiently ensure
stability and safety during major events. In a system that undergoes large inelastic
deformations, the energy dissipation provided by the ductile mechanism dampens the dynamic
response of the structure, effectively limiting structural demands. The next-generation seismic
lateral force resisting systems (LFRS) should require performance beyond ensuring life-safety
and collapse prevention. They should also protect the primary gravity load resisting system by
limiting damage to easily replaceable elements so that the structure can be repaired following a
major earthquake and remain operational following a moderate earthquake. Limiting or
preventing damage to its non-structural elements and equipment is also desirable from an
operational and monetary standpoint.
In todays steel buildings the most frequently used steel systems are the moment resisting
frame (MRF) and the concentrically braced frame (CBF). Many others lateral force resisting
systems which rely on plastic deformations for energy dissipation have been studied and
successfully used in the past. In particular, the eccentrically braced frames (EBF), the steel plate
shear walls (SPSW) and special concentrically braced frames (SCBF) are included in the current
Seismic Provisions for Structural Steel Buildings (AISC 2010) and allowed for use in steel buildings.
MRFs are flexible, economical structures that get their ductility capacity from beam flexure and

Figure 1 - Concetrically Braced Frames.

(a) V-bracing, X-bracing, and Diagonal Bracing configurations; (b) Yielding mechanism in

panel zone yielding. The relatively large flexibility of these systems results in larger inter-story
drifts, which could lead to severe P- effects. Yielding of the beams and the column panel zones
in MRFs cause extensive yielding and damage to the primary gravity-load system, and make
replacement of damaged area expensive and complicated (Gupta and Krawinkler 1999). CBFs
and SCBFs are a stiff, strong and economical alternative for bracing a steel frame. However their
ductility and energy dissipation capacities are limited by the buckling of the braces in
compression, even when installed in pair with a tension brace (Figure 1). After the compression
brace buckles, the system needs to undergo a large drift in the opposite direction to cancel the
P- effects on the brace and recover the full axial tension stiffness of the element (Lumpkin et al.
2012). EBFs are characterized by a stable hysteretic response under cyclic loading and provide a
good amount of energy dissipation (Hjelmstad and Popov 1983). However this category of
dissipative systems relies on yielding of a segment of the floor beam (link) for energy
dissipation, therefore a strong seismic loading will directly affect the primary gravity load
system and likely result in floor damage. The design and detailing of the link is also
complicated by the kinematics of the dissipative mechanism, which induces both shear and
flexural forces into the element (Hjelmstad and Popov 1984). Moreover since the link is part of
the gravity load system, replacement can be costly and complicated. Steel plate shear walls,
while providing adequate stiffness to resist lateral forces, are governed by the tension field
action. Since this mechanism develops after buckling along the compressed diagonals, the force-
displacement behavior of these bracing systems is pinched. This limits their ability to dissipate
energy. Moreover the SPSWs interact significantly with the boundary elements to which they
are connected. Local buckling of the column can occur if not properly detailed, and the

Figure 2 - Eccentrically Braced Frames.

(a) Typical configurations and mechanisms (Hjelmstad and Popov 1984); (b) Design detail (AISC 2010)

relatively large drifts required to develop the tension field action often lead to formation of
plastic hinges in the gravity load frame. Buckling-restrained braced frames (BRBFs) have similar
elastic behavior to CBFs with improved plastic performance. The special braces in BRBFs
prevent compressive buckling such that the diagonal braces exhibit nearly equal stiffness and
strength in tension and compression. In a BRBF, the diagonal braces are encased in a component
that restrains buckling of the brace core, but does not develop additional force resistance due to
friction between the brace core and encasing component. BRBFs were originally developed in
Japan after several decades of research, and have become a popular structural system
worldwide since the late 1990s. BRBFs overcome much of the shortcomings of CBFs, albeit
with substantial cost premium. Additionally, the design parameters to control the systems
lateral stiffness and strength (cross-sectional area and brace length) are somewhat constrained
by available bracing configurations. Meanwhile, shear panel systems such as those illustrated in
Figure 4 (different configurations are possible) have been used for high-rise buildings in Japan
since the 1990s. In many cases, shear panel systems in Japan use low-yield-point-steel (Saeki et
al. 1998) to ensure that plastic deformation is strictly contained within the shear panels, and the

Figure 3 - Steel Plate Shear Wall.

(a) Braced frame; (b) Strip-model for tension field action; (c) Plastic mechanism (Tsai et al. 2010)

systems are designed to supplement the underlying moment resisting frame. Among currently
available systems and systems studied in the past, the shear panel system most closely
resembles the system studied in this work. However, this Japanese system has not been adopted
widely in the US practice, perhaps due to the small energy dissipation capacity or the
substantial cost premium.

Figure 4 - Low-yield shear panel.
(a) Prototype building including shear panels; (b) Shear panel detail (Nakashima 1995)

Past research has also investigated various steel devices that undergo yielding in different
modes and can be designed to provide either supplemental passive energy dissipation or the
primary lateral force resistance. The added damping and stiffness (ADAS) concept (Bergman
and Goel 1987; Whittaker et al. 1991) consists of X-shaped steel plates that undergo flexural
yielding in double curvature (Figure 5). The similar triangular added damping and stiffness
device (TADAS) investigate (Tsai et al. 1993) consists of triangular shaped steel plates that also
undergo flexural yielding but in single curvature. It is often designed to be implemented to the
underside of a building beam with a vertical slotted hole to prevent axial forces developing on
the triangular plates from gravity loads. Such devices have been shown to provide very ductile
hysteretic behavior if proper fabrication tolerances and weld details are achieved. In addition,
these devices rely on out-of-plane bracing at the plate end that is not connected to the beam.

Figure 5 - ADAS Device.

(a) ADAS X-shaped element; (b) Device installation example (Whittaker et al. 1991)

While existing LFRS have been developed to satisfy a life safety performance objective,
they likely would be difficult to design to achieve desirable seismic performance while

providing replaceable ductile components. Seismic steel lateral force resisting systems capable
of reduced drift, controllable force capacity, stable hysteretic energy dissipation, and
replaceable structural details are needed for the next-generation seismic systems.

2 Proposed System

2.1 Description
The proposed bracing system consists of concentric X-braces, placed in series with a
yielding rectangular ductile shear panel as shown in Figure 6. Since this system is considered a
hybrid of a buckling-restrained braced frame and a steel shear panel system, it is referred to
hereafter as the braced ductile shear panel (BDSP) system. The four short I-shaped braces
transfer the lateral displacements arising from the lateral load on the building to the shear
panel. The ductile shear panel will be comprised of non-slender, in-plane plate elements,
stiffened around the perimeter by a boundary flange, and capable of achieving high levels of
ductility when experiencing plastic shear strains. Unlike the Japanese shear panel systems (the
system shown in Figure 1(d) develops moment at the top and bottom ends which must be
transferred into the existing framing), the braced ductile shear panel proposed here is subjected
to pure shear. Therefore, the behavior of the shear panel is expected to be more stable and
reliable (similar to a short EBF link). The braces need not be designed for significant moment,
and consequently, capacity design can be more easily and reliably implemented. In addition,
the connection between the shear panel and diagonal brace is simplified in the BDSP system.
The series configuration ensures that the strength of the ductile shear panel will define the
limiting seismic strength demand on the bracing system. The slenderness of the shear panel will
be limited such that a stable hysteretic behavior can be ensured for a substantial number of load
cycles, even at high levels of ductility demand. It is well known from extensive experimental
investigations of eccentrically braced frames, particularly those with shorter length shear links,
that cyclic shear yielding can be a stable and dependable mechanism for dissipating seismic
energy in a structural system.

Figure 6 - Scheme of the Braced Ductile Shear Panel

Compared to existing steel LFRS, the BDSP is expected to provide benefits of conventional
braced frames (reduced drift with reasonable architectural flexibility) but with enhanced energy
dissipation capacity and ductility compared to CBF and SPSW, details to allow replacement,
and a configuration that provides greater design flexibility to enhance seismic performance.
It is envisioned that the system could be used both in new buildings and as a retrofit
measure. The device is also designed so that it can be considered sacrificial, meaning that it
could be replaced after a severe loading. This implies that the braces and all other elements
composing the bracing system other than the BDSP are designed to remain elastic. For the same
reason a bolted connection is required between the braces and the BDSP. Additionally, the
connections between the BDSP bracing members and existing framing will primarily be
designed for axial force and not require the detailing required for CBF braces (needed to form
the buckling brace mechanism). The simplified connection detail would also be beneficial in a
seismic rehabilitation by limiting the additional demands placed on the existing framing to axial
forces. After an earthquake loading the plastic deformation in the panel will potentially produce
large forces in the braces. The question arises how to replace the panel in the presence of these
forces. The preliminary thought is that the bolted connection will most likely be designed as
slip-critical to prevent cyclic loading-induced damage to the bolt holes on the beams, and will
allow free expansion of the braces upon loosening of the bolts in long slotted holes.

2.2 Literature Review
A lot of research has been done in the past on steel devices of various shapes designed to
dissipate energy by yielding under shear deformations. Most of the studies were done by
Japanese authors using low-yield steel panels. A selection of the papers that are more relevant
to the development of the BDSP is summarized hereafter. Although limited, this selection is
believed by the writer to present a comprehensive variety of geometries and designs.
One of the earliest study on the use of rectangular shear panels as energy dissipation
devices was done by Nakashima (Nakashima 1995). The author tested six unstiffened shear
panels with variable width-to-thickness ratios, ranging from 58 to 28 (Figure 7a). The specimens
were made of low-yield stress steel (0.2% offset yield stress equal to 120MPa or 17ksi). The tests
performed were performed on the shear panel directly attached to the loading and reaction
frames at each end (Figure 7b). The author notes that all the specimens exhibited a stable
hysteretic behavior under cyclic loading, with the most slender specimen (width-to-thickness
ratio of 58) buckling at early stages of loading. However all the specimen still showed a stable
post-buckling behavior, with a only a minor degradation in stiffness.

(a) (b)
Figure 7 - (a) Specimens geometry; (b) Test setup (Nakashima 1995)

A different type of shear links made of aluminum were tested by Rai and Wallace (Rai
and Wallace 1998), envisioned as rectangular links installed at the top of chevron braces (Figure
8a). The authors performed tests on isolated, 1:4 scaled down specimens made of two different
aluminum alloys (0.2% yield stress equal to 35MPa and 53Mpa). The scaling factor resulted in
small panels 152mm long and 45mm deep. Both stiffened and unstiffened specimens were
tested (Figure 8b). The tests, performed at different cycling frequency of 5, 10 and 17Hz, showed
a good energy dissipation capacity. A stable force-displacement response was also obtained up
until buckling, which then affected the stiffness of the panels. However the behavior of the
devices was found to be insensitive to the range of loading rates used. The authors remark that

bolt-slip on the connection was proven critical given the smoothness of the aluminum surfaces
when compared to steel, and that special coatings or surface preparations should be mandated.
The problem of galvanic corrosion due to the steel-aluminum coupling is also raised by the

(a) (b)
Figure 8 - (a) Proposed system; (b) Test specimen (Rai and Wallace 1998)

An extensive experimental investigation on full-scale shear panels made of ultra-low-

yield steel (yield stress 100MPa or 15ksi) was conducted by Tanaka and Sasaki (Tanaka and
Sasaki 2000). Square and rectangular panels, both stiffened and unstiffened, with varying
width, depth and thickness were tested, for a total of sixteen specimens (Figure 9a). Three
different testing apparatuses were used (Figure 9c); in two of them the panels were directly
bolted to rigid loading blocks, one of them had the panels connected to two elastic supporting
elements the way they would be installed in a building frame following the pillar-type
configuration (Figure 9b). The response of all the specimens was stable under cyclic loading,
with moderate pinching of the most slender panels. Failure was dominate by fracture at welds
in the stockiest panels, while fracture initiated in the center of the more slender panels due to
cyclic buckling and reversed out-of-plane displacements.

(a) (b)

Figure 9 - (a) Typical specimen geometry; (b) Pillar type shear panel configuration;
(c) Experimental testing apparatuses (Tanaka and Sasaki 2000)

Built-up shear links have been used for the new San Francisco-Oakland Bay Bridge east
span, designed by the joint venture of T.Y. Lin International and Moffatt & Nichol. The
structure is a self-anchored suspension bridge, with the cables attached to a 565m long single
tower. The tower consists of four steel shafts inter-connected in the two directions by
replaceable bolted steel shear links, see Figure 10a. Given the location of the structure between
the San Andreas and Hayward faults and the designation by the California Department of
Transportation as a lifeline route, cyclic testing of the sacrificial shear links was conducted
(McDaniel et al. 2003). Two shear links were tested, one for each direction longitudinal and
transverse. The web of both links were made of ASTM A709 Grade 50 steel (measured yield
stress of 354MPa or 51.3ksi), while the connection regions used high-performance Grade
HPS70W steel (Figure 10b). The depth-to-thickness ratio of the specimen was 30.8. The test
setup was composed of two steel columns loaded horizontally at one end and connected by a
shear link (Figure 10c). The tests showed a stable response and a good energy dissipation
capacity of the two links, which were able to sustain shear strains up to 20 times the yield shear
strain and a total shear deformation of 0.06rad. Failure was the same in both specimens, with
fracture initiating at the end of the vertical fillet weld connecting the transverse stiffener to the
web of the link. Subsequent finite element analyses by the authors showed that stopping the
fillet weld at a distance of at least 3 leads to a reduction of 40% in hydrostatic stress and 60%
in cumulative plastic strain, when the links are subjected to a 0.07rad shear deformation. The
design of the links was improved afterwards following this finding.

(a) (b) (c)
Figure 10 - New San Francisco-Oakland Bay Bridge East Span. (McDaniel et al. 2003)
(a) Tower cross-section; (b) Shear link specimen; (c) Test setup

Chen et al. did a numerical investigation of stiffened shear panel made of JIS SS400 steel
(equivalent to ASTM A36 steel grade) under cyclic loading (Chen et al. 2006). The model
included the effect of residual stresses in the web and stiffeners, as well as initial imperfections.
The various simulations tested the sensitivity of the panels to different parameters. The authors
suggest that for cyclically loaded panels the web slenderness should be limited to 0.6, and
the ratio of stiffeners rigidity to optimum / be greater or equal to 3.0 (see below for
parameters explanation). Simulations also showed that the residual stress pattern only affects
the behavior of the panel prior to yielding. The parameters mentioned above are calculated
using the following expressions:

12(1 2 ) 4
= ; = 5.35 +
2 2
2( +1)1
23.1 1.35 1 + 3( +1)0.3
= ; =
( + 1)2.5 ( + 1)0.5 1 + 5.30.6( +1)3 +1
( and = web (or stiffened subpanel) depth and thickness; = shear yield stress; = Youngs modulus; =aspect ratio; =stiffener second

moment of inertia; =flexural rigidity per unit width of the web; =number of stiffeners).

Williams and Albermani (Williams and Albermani 2003) performed cyclic tests on a
chevron braced frame equipped with square 100mm shear panels, made by welding a thin

diaphragm inside a square hollow section (Figure 11). The devices proved easy and cheap to
manufacture, fit, remove and replace. All the devices tested yielded at quite low deformations
and exhibited significant ductility without failure. The load carried by the device continued to
increase after yield, with a ratio of maximum force carried to yield force of around 1.7 in most
tests. The authors found that diaphragm 2 mm thick (width-to-thickness ratio equal to 50) could
dissipate the most energy, although they were prone to buckling and to fracture under repeated
large-amplitude cycling. They recommended diaphragm 3 mm thick (width-to-thickness ratio
33) for the best combination of dissipation and robustness. After the tests the authors observed
significant slip had occurred at bolted connections, which resulted in elongation of the bolt

Figure 11 - Test frame configuration (Williams and Albermani 2003)

From the literature review it can be recognized like most of the studies done in the past
envisioned shear links attached to braces in a chevron or reversed-chevron configuration.
Although some authors do mention that panels could be used in a cross-braced frame, these
very few researchers do not provide any more details on how the system would be realized in
practice. For example, the connections between the frame, the braces and the shear panels are
not illustrated.
Furthermore many of the experimental tests (not cited in this section for brevity) were
conducted on the shear links alone, directly connecting them to rigid reaction blocks, therefore
relying solely on analytical models for the design of the connecting elements. At the same time,
no insights were acquired concerning the stability of the connecting elements, including the
braces. However the data obtained from past experiments is extremely valuable for

understanding the local behavior of panels when subjected to shear, including but not limited to
energy dissipation, ductility capacity, failure modes, stress concentrations, strain-rate
dependency and slenderness limitations. Nevertheless extrapolation of results from reduced
tests to estimate the response of a more complicated assembly would require the assumption
that the behavior of the system will be the same when connected to more flexible members.

3 Preliminary Analysis

3.1 Conceptual Design

3.1.1 Reduced model

The first step in the design of the proposed system was to create a reduced model to
determine simple analytical equations to estimate the behavior of the bracing system when
subjected to lateral forces as a function of the main parameters defining the panel geometry. The
model chosen is composed by four concentric beam elements connected to a deformable
rectangular panel.
A first analysis confirms that in order to have pure shear in the panel, the aspect ratio of
the shear panel / has to match the aspect ratio of the structural frame /. Using a rigid-
plastic analysis the inter-story drift index can be related to the plastic shear strain in the
panel, by imposing the balance between the amount of external work and the work dissipated
by shear deformation in the panel:

= [1]

where is the frame width, the panel width, the inter-story drift index (see Figure 12).
A relationship between the applied lateral force and the shear stress in the panel can be
found by assuming uniform shear stress and strain distribution in the panel. Under this
assumption, at any point in the response of the system the following holds:

= = [2]

where is the uniform shear stress, is the thickness of the panel and is the steel shear
modulus. By substituting the yielding shear stress in the last equation, the yielding lateral force
can be immediately calculated as = , where = 3 using Von Mises yield

Figure 12 Reduced model

3.1.2 Finite Element Models

Improved understanding of the local and global structural response is obtained using a
three-dimensional geometrically nonlinear elastic-plastic (ABAQUS) finite element model. The
braces and the BDSP are discretized using shell elements with different elastic-plastic material
models used to define the properties of the differing steel grades used in the device and braces.
Braces are modeled as built with steel with yield stress equal to 345 MPa (50 ksi). Since
these elements are not expected to experience plastic strains, their constitutive law is assumed
to be elastic-perfectly plastic.
The shear panel, however, will undergo large and reversed plastic strains. The simplified
elastic-plastic relationship used for the braces would predict inaccurate results by neglecting the
strain hardening and the subsequent evolution of the yield surface. In addition,
underestimating the maximum stress in the shear panel would lead to a lower capacity demand
in the braces, which instead must remain elastic and guarantee enough strength so that the
damage is localized in the dissipative device.
It follows that the cyclic behavior of the panel in shear is a fundamental factor in order to
obtain an accurate prediction of the system response. Therefore an elastic-plastic model with
nonlinear kinematic and isotropic hardening is assigned to the dissipating elements
(Nakashima 1995).
The parameters defining the model are fitted to a cyclic stress-strain curve, which models
the behavior of structural steel under repeated cyclic loading (Cofie and Krawinkler 1985). The
resulting yield stress is approximately 240 MPa (35 ksi).






-16 -12 -8 -4 0 4 8 12 16

(a) (b)
Figure 13 Calibration of plasticity model: (a) Experimental results from Cofie and Krawinkler (1985); (b) Finite
Element Analysis

The loading, which is applied in the horizontal direction at the upper braces, consists of a
unit normal force in the elastic buckling analyses, and a prescribed displacement in the
nonlinear cyclic analyses.

3.1.3 Elastic buckling

An estimate of the loads associated with elastic buckling is made using a linearized
version of the finite element model. The finite element analyses showed that the most critical
buckling mode for the geometries and dimensions considered is either the global out-of-plane
buckling of two aligned braces about their weak axis or local shear buckling of the BDSP web.
Illustrative results for the displacements associated with the latter buckled shape are plotted as
contours in Figure 14, and show that the only part of the structure undergoing significant
displacements in this configuration is the shear panel. Simulations of numerous model
geometries suggested that systems that exhibit out-of-plane buckling are associated with
decreased stiffness and unsatisfactory performance under cyclic loading. In Figure 15 the
history of the horizontal reaction force at one of the supports (end of brace) is shown for two
systems with different dominating buckling mode. Both configurations have an equally wide
and thick shear panel (b=1000 mm, tw=7.5 mm), and therefore are expected to yield at the same
load. They also share the same bay width and story height (L=4.50 m, h=3.00 m), and in turn
equal elastic lateral stiffness. The only difference between the models is the section of the braces.
In this plot a positive value of the horizontal reaction causes the attaching brace to be tensioned.

Figure 14 - Shear buckling of the stiffened BDSP web (contour plot of normalized out-of-plane displacements)

The brace-buckling dominated system has worse performance in carrying load through the
compression brace, and the stiffness degrades with cyclic loading, up to the point where the
load transferred to the compressive brace is reduced by more than 50%. The total capacity (sum
of compression and tension contributions on two symmetric braces) will then also degrade with
increasing number of cycles. Note that while AISC suggests a different loading protocol
involving cyclic loading at numerous amplitude magnitudes, for the purpose of this conceptual
design study a constant amplitude cyclic loading protocol is used as to increase the rate of the
stiffness degradation. Therefore the design of the elements in the bracing system should limit
out of plane buckling. The only stable post-buckling behavior under cyclic loading is the shear
buckling of the shear panel. The critical load for shear buckling of a rectangular plate is equal to:

= 2 [3]
= [4]
12(1 2 )
where is a function of boundary conditions:
5.34 + 4.0
= 2
8.98 + 5.6
and , and are the width, height and thickness of an unstiffened plate (or the in plane
dimensions of the regions that the plate is divided into by stiffeners) respectively, and and
are the steels Youngs modulus and Poisson ratio.


Horizontal Reaction (kN) 1000

Braces buckling
Web shear buckling


0 2 4 6 8 10 12

Figure 15 - Comparison of performance for different dominating buckling modes

The panel flanges, as well as the braces, provide partial flexural restraints at the edges of the
plate. The critical load is then expected to lie within the buckling load solutions for simply
supported and fully restrained plates. In Figure 16 the buckling loads calculated using [3] are
plotted together with the buckling load from finite elements analyses for 30 different models.
The plots shows that the shear load calculated for the simply supported plate can therefore be
used as a lower bound for the critical load.

Simply supported
Fully restrained
Buckling Load (kN)


0 5 10 15 20 25 30
Model ID

Figure 16 - Buckling loads comparison

3.1.4 Cyclic nonlinear response

The behavior of the bracing system prior to and after the yielding of the BDSP has been
studied using parametric finite elements analyses. All the analyses included geometric

nonlinearities and initial imperfections. In each analysis the model was subjected to twelve fully
reversed drift cycles of constant amplitude. Parametric studies have been carried out on the
following variables:
BDSP panel thickness (tw);
BDSP in-plane size (b,a) with a/b=h/L;
Bay width (L) and story height (h);
Braces cross-section dimensions;
Imperfection amplitude (0.25 - 0.75 - 2.5 cm);
Boundary conditions (braces-to-frame connection: pinned; continuous moment).
The most significant result obtained from the analyses is the curve that relates the applied
lateral load to the displacement imposed at the top of the braces. The data showed that
equation [2] gives an accurate estimate of the yielding load, and equation [1] is a good
approximation of the average plastic shear strain in the panel.
The linear relationship between tw and Vy can be seen in Figure 17, where five plots of
force-displacement curves for different panel thicknesses are normalized by the load that
initiates yielding calculated using [2]; all the normalized curves share the same initial linear
segment and yielding always starts at / = 1, which proves the accuracy of the equation
used. The force-displacement data collected from all the analyses allowed definition of a simple
equation to estimate the ultimate load of the bracing system: while neglecting the contribution
coming from the flanges surrounding the BDSP web, one can assume that at the ultimate load
the whole shear panel is yielded, and reached its maximum shear stress. Therefore the ultimate
lateral load will be = .
These results suggest a bilinear approximation of the force-displacement behavior of a
system with given dimensions. The first segment, corresponding to the elastic region, starts at
the origin and ends at the point identified by the drift and the lateral load to cause first yielding;
the second segment continues from that point to the ultimate drift and load coordinates. The
equations used to calculate these points can be solved for the BDSP dimensions, thus allowing
the selection of geometries that will result in force-displacement behaviors satisfying practical

3mm 4mm 6mm 8mm 10mm

Normalized lateral force V/Vy




-6.00 -4.00 -2.00 0.00 2.00 4.00 6.00
Inter-story drift (cm)

Figure 17 - Normalized force-displacement plot for different panel thicknesses

The effect of initial imperfections on the initial and cyclic response of the system has also
been studied using finite element analyses. It has been found that the system behavior is not
sensitive to imperfection amplitudes as large as 25mm, which represent extremely large values
that significantly exceed acceptable fabrication and erection tolerances. Larger imperfections
develop its steady cyclic behavior faster. In fact, under these extreme initial imperfections and
cyclic loading, the system tends to displace as the first buckling shape. The amplitude of this
deflection initially increases with each load cycle; however after a few cycles (3 to 6 cycles) this
trend ends and the amplitude stays mostly constant for all subsequent cycles. The imperfections
are shown to affect the number of cycles needed to stabilize, but not the final response. This
behavior can be seen in the plot in Figure 18, where the axial force in one of the braces under
cyclic loading is plotted for different initial imperfection amplitudes (compression if positive).
The braces in the model with no imperfections are able to carry more compression in the initial
cycles compared to the system with imperfections. However, after a small amount of cycles the
responses of the different systems collapse to the same curve.

0.02 in 0.10 in 0.50 in No Imperfection


Brace axial force (kN)





0 1 2 3 4 5 6 7 8 9 10 11 12 13 14

Figure 18 Axial force in one of the braces under cyclic loading for different initial imperfection amplitudes

The force-displacement plots drawn from the analyses showed that the bracing system
could achieve a very stable hysteretic behavior. To obtain a stable response, however, special
attention must be paid to the slenderness of some key elements. The slenderness of the panel
has to be limited so that the out-of-plane deformation, caused by the combined effect of
imperfections and load reversal, will not affect the in-plane stiffness of the web. This can be
achieved (without increasing the thickness of the web and in turn the lateral stiffness) with the
addition of transverse stiffeners. The transverse stiffeners shall be tapered and stop at a distance
of at least 3 from the flanges, as suggested by McDaniel et al. (McDaniel et al. 2003).
In addition the connections and the braces should provide enough flexural stiffness
against the out-of-plane buckling in order to contain the deflection caused by imperfections. The
required stiffness for both pinned and fixed end connections would have to be determined for
each system. A critical improvement to the stability of the system is obtained by providing
transverse stiffeners at the end of the I-beam sections where their flanges are welded to the
shear panel flanges.
If these issues are not accounted for in the design of the bracing system, the system will
undergo cyclic stiffness and strength degradation as a consequence of increasing out-of-plane

3.1.5 Plastic dissipation
Another result obtained from the finite element analyses is the proof that the choice of
braces section for coupling with a specific panel leads to the concentration of plastic strains only
in the BDSP; the braces remain elastic and are thus reusable. Figure 19 illustrates the evolution
of the PEEQ index (Equivalent Plastic Strain) and that no plastic strains are present in the braces
throughout the response. The PEEQ index is used as a measure of plastic strains throughout the
analysis at each integration point. It is defined as the integral of the absolute value of plastic
strain rate; therefore its value is always greater or equal to zero.

Figure 19 Cumulative plastic strain evolution

(PEEQ contour plot after 1.25-4.25-8.25-11.25 cycles)

3.2 Evaluation of Performance

The effects of the installation of the BDSP device in a steel framed structure have been
studied in order to estimate the actual benefits provided by the new bracing system. The steel
structure under study is a five-bay seven-storey frame. The inter-storey height is equal to 3.5m
and the bay length is equal to 6m. The steel frame belongs to a building with rectangular plan of
30m x 36m dimensions consisting of five 6m-bays in the X-direction and six 6m-bays in the Y-
direction. Different IPE steel profiles are used for the beams while columns are made of HE320B
profiles. The building presents welded connections and all the elements are made of S355 steel.
Figure 20 shows the plan of the steel building and the elevation of the analyzed frame with the
section profiles of beams and columns. The building was designed according to the provisions
of Eurocodes 3 and 8, considering a response spectrum type 1, soil type C and peak ground
acceleration = 0.35. The resulting design spectrum for a return period of 475 years and

behavior coefficient q=6.0 is shown in Figure 21. Dead loads consist of the weights of structural
components and partitions, and live loads were considered to be equal to 4 kN/m2. Story
masses include dead loads and a percentage of live loads (30% according to Eurocode 8 for
common residential and office buildings).

Figure 20 Building plan and frame elevation (dimensions in meters)



Sd (g)




0 0.5 1 1.5 2 2.5 3 3.5 4
T (s)

Figure 21 - Design Spectrum

The steel frame has been numerically modeled using the computer code SAP2000 by
means of beam and nonlinear link elements. At each beam and column end, a nonlinear link
element has been inserted in the model due to the typical concentration of plasticity at the
extremities of the structural elements of MRFs. A schematic view of the insertion of nonlinear
link elements at the beam and column ends is shown in Figure 22.

An elastic-plastic constitutive law is associated to the in-plane rotational degrees of
freedom of each link. The generalized force-displacement law used is the an elastic-plastic Bouc-
Wen model with kinematic hardening (Wen, 1976):
= + (1 )
where , are respectively the generalized force and displacement for each degree of freedom,
is the linear elastic stiffness, is the ratio of post-elastic stiffness, is the yielding force and
is an internal hysteretic variable.

Figure 22 - Plastic hinges model

In order to efficiently run time-history analyses to evaluate the effects of the device on the
seismic response of the steel frame an equivalent model of the BDSP system has been created in
SAP2000. Figure 23 shows the schematic representation of the BDSP model consisting of two
link elements crossing the braced bay. The axial degree-of-freedom in each link is associated
with an elastic-plastic constitutive law similar to the one used at the extremities of the beam and
column elements. However in this case the generalized force-displacement curve represents the
axial response of each link and the parameters are calibrated using the expressions for the
yielding lateral load and stiffness worked out for the simplified model and verified with the
finite element analyses.
The yielding axial force and the elastic stiffness in each link are given by the following

= [5]


= [6]
2 2
where is the angle between the link and the beam. The other parameters which control the
post-yield slope and the sharpness of the curve are fitted to the results obtained from the
previous finite element analyses.
The BDSP system was inserted in the lateral bays of all the stories of the external frames,
as shown in Figure 12.

Figure 23 - BDSP-equivalent links model

This numerical model allowed to accurately reproduce the seismic response of the frame
and to estimate the amount of hysteretic energy dissipated during the seismic excitation by the
different inelastic mechanisms.
The El Centro NS component earthquake record (Imperial Valley, 1940) with peak
ground acceleration scaled to 0.35g and 0.6g has been used to perform nonlinear dynamic
simulations of the structural response of the steel frame, with modal damping set to 2%. The
results obtained from the numerical analyses are reported in terms of top displacement time
history, base shear time-history and amount of energy dissipated by different mechanisms.
First, a numerical simulation of the seismic response of the MRF has been carried out and
the results obtained in terms of base shear have been used to determine the BDSP parameters
(stiffness, yielding load, ultimate load) needed for the design of the device. Then the frame has
been provided with the BDSP devices in the lateral bays. In this case the mass pertinent to the
frame has been increased since all the seismic force is assumed to be resisted only by the two
braced XZ-frames, while in the previous case all the XZ-parallel frames were assumed to
participate equally. The sizes of the devices at different levels have been optimized such that

stiffness and strength decrease with the height of the frame, following the shear distribution
along the frame. The main objective of the design is to preserve the original structural elements
from damage. The characteristics of the BDSP devices inserted at the different levels of the
frame are reported in Table 1. The four short braces in the BDSP system consists of HEB280A
section profile with S355 steel.

Level b x a [mm] tw [mm] Vy [kN]

1 1000 x 583 4 554
2 900 x 525 4 498
3 900 x 525 4 498
4 900 x 525 4 498
5 900 x 525 4 498
6 900 x 525 3 374
7 900 x 525 3 374
Table 1 - Characteristics of the BDSP device at the different levels of the frame

The results of the dynamic analyses showed that the presence of the BDSP devices
decreased the lateral displacements at the top of the frame. The maximum drift obtained for the
BDSP frame is equal to 8.8 cm, while the maximum drift computed for the MRF is equal to 28.3
cm, as shown in Figure 24. The plot in Figure 25 indicates that the maximum base shear
registered for the BDSP-braced frame (about 1700 kN) is greater than the one obtained for the
MRF (about 1000 kN). It should be considered that the tributary masses to each BDSP-equipped
frame are three times than the mass tributary to each MRF.
The most relevant effect of the BDSP system on the structural response of the steel frame
is in terms of energy dissipation. Figure 26 shows the amount of energy (normalized with
respect to the total input energy) dissipated by the two frames for two different seismic
intensity levels. For peak ground acceleration equal to 0.35g, most of the energy in the MRF is
dissipated by modal damping, while only few beams dissipate a small amount of energy. When
the peak ground acceleration increases to 0.6g, the amount of hysteretic energy dissipated at the
extremities of the beams is comparable to the one dissipated by modal damping, therefore equal
to about 50% of the total input energy. Therefore according to the model extensive damage is
predicted in the floor beam of the MRF structure. Energy dissipation in the BDSP-equipped
frame proved to be much more efficient. For peak ground acceleration equal to 0.35g, the

devices are able to dissipate a significant amount of energy. At the same time smaller inter-story
drifts prevent the formation of plastic hinges in the beams, therefore making the bracing system
the only dissipating mechanism besides modal damping. A similar response is obtained in case
of excitation with peak ground acceleration of 0.6g, where most of the energy is dissipated in
the bracing system and no plastic hinges activate at the extremities of the beams. In this case the
model predicts a totally elastic response of the primary gravity load system.

Figure 24 Top displacement time-history for the two frames under seismic intensity levels equal to ag=0.35g

Figure 25 Base shear time-history for the two frames under seismic intensity levels equal to ag=0.35g.

Figure 26 - Amount of energy (normalized with respect to the total dissipated energy) dissipated by different
mechanisms in the two frames for different seismic intensity levels

Finally, in order to validate the two links model used to embed the response of the BDSP
system in SAP2000, a detailed model of the system has been created reflecting the dimensions of
the one modeled in the frame. The inter-story drift time history from the dynamic analysis with
peak ground acceleration set to 0.60g is used to prescribe the displacement in the detailed
model, and the resulting reaction forces are compared to the shear force predicted in the frame.
The plots of the total lateral force against time for both the detailed and reduced model are
shown in Figure 27. The reduced model gives a good approximation of the system response,
and can therefore be used in a frame analysis to represent the effect of the studied system.

Figure 27 Lateral force in the bracing system for detailed and reduced model

4 Conclusion and Future Work
The conceptual design process of the proposed system has been detailed in the previous
sections, along with an initial evaluation of the system capabilities of dissipating energy. The
computational study on the new bracing system allowed understanding of the behavior of the
ductile panel under cyclic loading. Suggestions have been defined for its design, in particular
the slenderness of the elements required to prevent buckling. The system is shown to
potentially be a valid alternative to currently used lateral load resisting systems and to have the
ability of overcoming some of their shortcomings. In particular the BDSP should be capable of
undergoing significant plastic deformations without causing damage to surrounding framing
and can be replaced following and earthquake.
The results obtained at this moment are mostly computational. While previous researches
found in the literature give important insights into the cyclic behavior of rectangular flanged
steel plates in shear which are very similar to the shear panel in the proposed system they
arent thought to be sufficient for predicting the behavior of the BDSP, even when coupled with
the numerical analyses. In addition, the behavior of the proposed system when installed in a
frame structure could be affected by the detailing of the system, including but not limited to the
type of connections between the elements, the type of welds used and low-cycle fatigue. Finally
constructability and production tolerances should also be investigated before the system could
be used in real structures. All these factors could modify the response of the system and in turn
affect the performances in terms of ductility and energy dissipation of the panel. In the next
paragraphs, future work planned in the context of this study and aimed at addressing these
issues is presented.

4.1 Experimental Program

The uniqueness of the BDSP, compared to previously published shear panel systems,
limits the amount of information that can be reused from literature. Therefore an extensive
study on the effect that different geometries have on the behavior of the device is needed.
The design of the new system also involves the design of short cross braces, stiff gusset
plates, and specific demands on the strength of beams and columns adjoining the bracing.
Although some of these elements are similar to what is currently used in professional practice
for other bracing systems (e.g. EBFs, BRBFs), there are still substantial differences that make the
design of the BDSP require an assessment of the adequacy of such elements, when used in

conjunction with the new system. While results from finite element analyses are satisfactory and
give useful insights in the system performance, they cannot properly recreate complex local
behaviors (low-cycle fatigue, cracking and fracturing) that can adversely affect the system
In order to obtain data that can be used to address these subjects, an experimental
program has been designed, aimed at acquiring data for calibration of the aforementioned finite
element model, as well as identifying critical details and failure types which might not have
been caught by numerical simulations. The tests will be conducted at the University of
Minnesota Galambos Structural Engineering Laboratory.
The testing setup consists of a half-scale model of a single-bay one-story braced steel
frame (bay width 10, story height 68), which will be tested under cyclic lateral loading.
During loading, the BDSP installed in the frame is expected to undergo large plastic strains
while the beams, columns, and connecting elements remain in the elastic range, allowing the
frame to be reused for testing of a total of ten BDSP specimens (Table 2). The ten panel
specimens that will be tested are an incomplete permutation of three parameters:
1. panel width = {20", 24"}
2. web thickness = {18", 3/16 ", 1/4"}
3. web perforation ratio (perforation diameter/sub-panel width) = {0.25 , 0.40}.

b t D/b ds ts
Specimen ID (in) (in) (in) (in)
BDSP-20-1/8-0.40 20 1/8 0.40 1 3/16
BDSP-24-1/8-0.40 24 1/8 0.40 1 3/16
BDSP-20-1/8-0.25 20 1/8 0.25 1 3/16
BDSP-20-3/16-0.40 20 3/16 0.40 1 1/2 3/16
BDSP-24-1/8-0.25 24 1/8 0.25 1 3/16
BDSP-20-1/8-0.00 20 1/8 0.00 1 3/16
BDSP-24-3/16-0.40 24 3/16 0.40 1 1/2 3/16
BDSP-20-3/16-0.25 20 3/16 0.25 1 1/2 3/16
BDSP-20-1/4-0.40 20 1/4 0.40 1 3/4 1/4
BDSP-24-1/8-0.00 24 1/8 0.00 1 3/16
Table 2 Specimen geometries
(b = clear panel width, t = web thickness, D=perforation diameter, ds=stiffeners depth, ts=stiffeners thickness)

Loading will be applied at one corner of the frame using a 220kips actuator and will
follow the ATC-24 loading protocol.
A drawing of the test setup as it will be constructed in the laboratory is shown in Figure
28. The picture includes the actuator, the loading frame that supports actuator and provides the
necessary reaction, the foundation blocks that connect the frame to the strong floor and the
lateral bracing. A detailed drawing of the testing frame is shown in Figure 29, and a typical
specimen is detailed in Figure 30.
The specimens and the frame will be instrumented with a combination of LVDTs, strain-
pots and strain gages to record data that will be used to calibrate the numerical model. The
initiation and evolution of cracking, as well as failure modes will be monitored throughout the
testing phases.

Figure 28 Test setup overview

Figure 29 - Testing frame

Figure 30 - Specimen #1

4.2 FE Model Validation and Refinement
The data obtained from the experimental program will be used to calibrate and validate
the finite element model of the testing setup (Figure 31).
The properties of the steel used to fabricate the panel will be tested and used to define a
more precise stress-strain relationship for the material used. The in-plane and out-of-plane
displacements of the panel will be compared between the experimental and numerical model,
and their agreement will be evaluated. The same will be done for strain fields in the panel and
internal action in the other elements of the frame, with special attention to the braces and their
internal forces. If deemed necessary, the shear panel in the numerical model could include a
damage and failure criterion due to cyclic inelastic deformations, so as to add the possibility of
evaluating the actual ductility and energy dissipation capacity of different geometries and

Figure 31 - Detailed Finite Element model of the experimental testing frame

Once the model had been calibrated, a numerical investigation of practical sized bracing
systems, with a large variety of parameters will be investigated, allowing optimization of the
design and gathering of further insights into the system behavior and reliability.

4.3 Performance evaluation
After the finite element model has been calibrated, the results will be used to better define
the ductility capacity of the shear panel, and define a failure criterion. Using this data, the
response of different two- and three-dimensional steel frame structures equipped with the
proposed system can be further investigated. In fact, the availability of more reliable and
accurate data on the system behavior will justify a more rigorous and broader study, which has
not yet been carried out due to the lack of experimental confirmation of the numerical analyses.
This investigation will account for the provisions found in the current codes (ANSI/AISC 341-
10, ASCE-7, etc.) and will consider DBE and MCE hazard levels as defined in the codes. The
performance for buildings will be analyzed using seven consistent ground motions
appropriately scaled.
The goal of this part of the research project will be aimed at calculating the over-strength
factor for the proposed system, as well appropriate response modification factor R of the
proposed system. A design procedure will also be determined based on these results.


AISC. (2010). Seismic Provisions for Structural Steel Buildings. ANSI/AISC 341-10, Chicago, IL.

Bergman, D. M., and Goel, S. C. (1987). Evaluation of cyclic testing of steel-plate devices for added
damping and stiffness. Ann Arbor, MI.

Chen, Z., Ge, H., and Usami, T. (2006). Hysteretic Model of Stiffened Shear Panel Dampers.
Journal of Structural Engineering, ASCE, 132(3), 478483.

Cofie, N. G., and Krawinkler, H. (1985). Uniaxial Cyclic Stress-Strain Behavior of Structural
Steel. Journal of Engineering Mechanics, 111(9), 11051120.

Gupta, A., and Krawinkler, H. (1999). Seismic Demands for Performance Evaluation of Steel Moment
Resisting Frame Structures. Stanford.

Hjelmstad, K. D., and Popov, E. P. (1983). Cyclic Behavior and Design of Link Beams. Journal
of Structural Engineering, 109(10), 23872403.

Hjelmstad, K. D., and Popov, E. P. (1984). Characteristics of Eccentrically Braced Frames.

Journal of Structural Engineering, American Society of Civil Engineers, 110(2), 340353.

Lumpkin, E. J., Hsiao, P.-C., Roeder, C. W., Lehman, D. E., Tsai, C.-Y., Wu, A.-C., Wei, C.-Y.,
and Tsai, K.-C. (2012). Investigation of the seismic response of three-story special
concentrically braced frames. Journal of Constructional Steel Research, 77, 131144.

McDaniel, C. C., Uang, C.-M., and Seible, F. (2003). Cyclic Testing of Built-Up Steel Shear Links
for the New Bay Bridge. Journal of Structural Engineering, American Society of Civil
Engineers, 129(6), 801809.

Nakashima, M. (1995). Strain-Hardening Behavior of Shear Panels Made of Low-Yield Steel. I:

Test. Journal of Structural Engineering, American Society of Civil Engineers, 121(12), 1742

Rai, D. C., and Wallace, B. J. (1998). Aluminium Shear-Links For Enhanced Seismic
Resistance. Earthquake Engineering and Structural Dynamics, 27, 315342.

Saeki, E., Sugisawa, M., Yamaguchi, T., and Wada, A. (1998). Mechanical Properties of Low
Yield Point Steels. Journal of Materials in Civil Engineering, 10(3), 143152.

Tanaka, K., and Sasaki, Y. (2000). Hysteretic Performance of Shear Panel Dampers of Ulta Low-
Yield-Strength Steel for Seismic Response Control of Buildings. 18.

Tsai, K.-C., Chen, H.-W., Hong, C.-P., and Su, Y.-F. (1993). Design of Steel Triangular Plate
Energy Absorbers for Seismic-Resistant Construction. Earthquake Spectra, 9(3), 505528.

Tsai, K.-C., Li, C., Lin, C., Tsai, C., and Yu, Y.-J. (2010). Cyclic tests of four two-story narrow
steel plate shear walls. Part 1: Analytical studies and specimen design. Earthquake
Engineering & Structural Dynamics, 39(7), 775799.

Whittaker, A. S., Bertero, V. V., Thompson, C. L., and Alonso, L. J. (1991). Seismic Testing of
Steel Plate Energy Dissipation Devices. Earthquake Spectra, 7(4), 563604.

Williams, M. S., and Albermani, F. (2003). Monotonic and Cyclic Tests on Shear Diaphragm
Dissipators for Steel Frames. Brisbane, Australia, 134.