An Experimental Study On Steel Encased Buckling Restrained Brace Hysteretic Dampers PDF

EARTHQUAKE ENGINEERING AND STRUCTURAL DYNAMICS
Earthquake Engng Struct. Dyn. 2010; 39:561–581

Published online 18 September 2009 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/eqe.959
An experimental study on steel-encased buckling-restrained

brace hysteretic dampers
Mehmet Emrah Eryaşar and Cem Topkaya∗, †

Department of Civil Engineering, Middle East Technical University, Ankara 06531, Turkey
SUMMARY
This paper describes a seismic test program performed on 12 steel-encased buckling-restrained braces
(BRBs). The use of rolled or built-up buckling-restraining mechanisms with welded or bolted attachments
was examined. In addition, the effects of bolt pretension, core-to-encasing attachment details, aspect ratio
of core segment and imperfections due to manufacturing on the brace response were investigated. All
specimens were subjected to a stepwise incremental quasi-static testing protocol with a maximum axial
strain amplitude of 2%. All specimens except one showed satisfactory performance with stable hysteretic
response and sustained cumulative inelastic deformations in excess of 200 times the yield deformation.
Based on the experimental results, the compression strength adjustment factor, and the strain hardening
adjustment factor for each excursion were quantified. Test results revealed that these two factors are
significantly influenced by the parameters investigated in the experimental program. BRB details were
also found to influence the buckling and the yielding patterns of the core segment. Copyright q 2009
John Wiley & Sons, Ltd.
Received 28 May 2009; Revised 10 August 2009; Accepted 17 August 2009
KEY WORDS: buckling restrained brace; hysteretic damper; energy dissipation; friction; yielding; seismic
INTRODUCTION AND BACKGROUND
Frames with buckling-restrained braces (BRBs) can be used as a seismic load resistance system.
A typical BRB consists of a core segment and a buckling restraining mechanism. When compared
to a conventional brace, BRBs provide nearly equal axial yield force in tension and compression.
In addition, BRBs exhibit stable and predictable hysteretic behavior, provide significant energy
∗ Correspondence to: Cem Topkaya, Department of Civil Engineering, Middle East Technical University, Ankara
06531, Turkey.
†
E-mail: ctopkaya@metu.edu.tr
Contract/grant sponsor: College of Engineering, Middle East Technical University; contract/grant number: BAP
2008-03-03-03
Contract/grant sponsor: Scientific and Technological Research Council of Turkey; contract/grant number: TUBITAK
105M242
Copyright q 2009 John Wiley & Sons, Ltd.

562 M. E. ERYAŞAR AND C. TOPKAYA
dissipation capacity and ductility. These braces plasticize during a moderate to severe earthquake
and can be considered as hysteretic dampers.
A significant amount of research work has been performed in Japan and elsewhere in Asia
over the last few decades for the development of BRBs [1]. A detailed summary of findings are
summarized in a report by Uang and Nakashima [2]. A number of different restraining mechanisms
have been developed and these can be grouped into two main categories. In the majority of the
braces produced, buckling is inhibited by a concrete or mortar filled steel encasing member, which
is usually a hollow structural section. The other alternative is to restrain the core brace utilizing
steel sections. Each type has certain advantages and disadvantages. Mortar filled tubes were found
to have problems concerning the quality control in manufacturing process and flexibility in the
design details at both ends of the core plate [3]. On the other hand, steel encasing alone were found
to provide inadequate hysteresis under high strains [4]. Most of the BRBs developed to date are
proprietary. A survey of existing buildings [5] revealed that most of the BRBs used so far utilize
mortar filled tubes as encasing members.
Proprietary BRBs that have been developed in Japan are treated as hysteretic dampers in design,
and no design provisions are available [2]. In the United States, however, design recommenda-
tions have recently been incorporated into AISC 341-05 Seismic Provisions for Structural Steel
Buildings [6]. This provision requires qualifying cyclic tests to be performed on a subassemblage
and a uniaxial test specimen. For the design of bracing connections and adjoining members, AISC
341-05 specification requires the use of adjusted brace strength (Pabs ), which is defined as follows:
Pabs = Pysc in compression
Pabs = Pysc in tension (1)
Pysc = Fysc Asc
where, Fysc : is the actual yield stress of the steel core as determined from a coupon test; Asc is the
net area of steel core; is the compression strength adjustment factor; and the strain hardening
adjustment factor.
The adjustment factors and , which are solely dependent on BRB details, are determined
by testing. When subjected to strong ground shaking, BRB members can experience axial strains
that are 10–20 times their yield strain. During any inelastic excursion, cyclic hardening of the
core material takes place, which in turn increases the brace force beyond the yield force, Pysc .
Furthermore, due to manufacturing details, certain amount of friction that develops between the
core segment and the buckling-restraining mechanism is inevitable. Transfer of frictional forces
also results in an increase in the brace force. The strain hardening adjustment factor, , is calculated
as the ratio of the maximum tension force measured from the qualification tests to the yield force,
Pysc , of the test specimen. When a BRB is subjected to compression, lateral expansion of the steel
core takes place due to Poisson’s effect. The area of the steel core and the frictional resistance due to
contact increase as a result of lateral expansion. Because of this Poisson’s effect compressive force
level attained is higher than the tensile load level for the same amount of axial displacement. The
compression strength adjustment factor, , is calculated as the ratio of the maximum compression
force to the maximum tension force of the test specimen. AISC Seismic Provisions mandate that
the compression strength adjustment factor, , be less than 1.3 for each displacement excursion
greater than the yield displacement.
While a large body of knowledge exists on BRBs, the behavior of steel-encased BRBs has
not been studied in detail. In a recent study by Tremblay et al. [7] authors concluded that
Copyright q 2009 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2010; 39:561–581
DOI: 10.1002/eqe
STEEL ENCASED BUCKLING RESTRAINED BRACE HYSTERETIC DAMPERS 563
steel-encased BRBs has a potential for adequate ductile seismic response. Steel encasing was
adequate to prevent global buckling of the core segment and provided stable hysteretic behavior.
However, their test results indicated the necessity to control local core buckling response to mini-
mize frictional response between the core and the buckling-restraining mechanism and develop
uniform strain demand in the core segment. Moreover, authors concluded that further research is
needed to study the effects of unbonding material on the global response. Based on this discussion
it is apparent that the performance of steel-encased BRBs is limited due to the problems associated
with the core–encasing interaction.
A two-phase experimental study has been undertaken at the Middle East Technical University
to address the problems related with steel BRBs. In the first phase, the core–encasing interaction
was studied. The objectives of the first phase were to investigate welded or bolted attachments for
buckling restraining mechanisms, examine the effects of bolt pretension, and study core-to-encasing
attachment details. A set of design recommendations on detailing for adequate core–encasing
interaction was developed based on the findings of the first phase. These details were applied
further in the second phase of the experimental program where specimens with built-up encasing
members were experimented. In the second phase, the buckling restraining mechanisms were
optimized and the performance of core segments with different aspect ratios was investigated. A
total of 12 tests were performed on small-scale BRB specimens. The details of the experimental
program and the results are presented herein.
TEST SPECIMEN SELECTION AND DESIGN
Details of specimen properties are given in Table I. In addition, longitudinal and cross-sectional
views are provided in Figures 1 and 2, respectively. The first six specimens belong to the first
phase of the study, while the others belong to the second phase. The thought process behind the
design of each element and adopted detail is explained in the following sections.
Core segment
The total length of BRB specimens was 1140 mm. This dimension was determined by the dimen-
sions of the test setup. All BRB specimens had cruciform type nonyielding segments at both ends
and a 900-mm-long restrained yielding segment (Figure 1).
All specimens had a rectangular yielding core segment (Figure 1). Because of the limited screw
jack capacity, the thickness of the core material used was limited to 5 mm. Depending on the
aspect ratio (width divided by thickness), 40, 60, and 80 mm wide core segments were tested.
The 40 mm by 5 mm core segments were obtained as flat bars. These were made of European
S355 grade [8] steel and the measured yield stress (Fy ) and the ultimate stress (Fu ) were 355 and
510 MPa, respectively. All flat bars were from the same batch. Because no flat bars were available
for larger widths, other core segments were produced from a plate. A 5 mm plate was cut using
a precision controlled CNC machine to 60 and 80 mm widths. The plate material was made of
European S275 grade [8] steel and the measured yield stress (Fy ) and the ultimate stress (Fu )
were 280 and 420 MPa, respectively.
Two 120-mm long 40 by 5 mm cross-section stiffening plates were fillet welded on both ends of
the core segment to produce a cruciform nonyielding segment (Figure 1). Later the core segment
cruciform ends were fillet welded to 10-mm-thick plates, which were used to fasten the specimen
DOI: 10.1002/eqe
564
Copyright q
Table I. Properties of specimens.
Core plate Encasing

Specimen Dimension Aspect Fy Weld Bolt Encasing
no. (mm) ratio (MPa) Type Section Connection type pretension Attachment Imperfection Pcr/Py weight (kg)
2009 John Wiley & Sons, Ltd.

1 40×5 8:1 355 R UPN 65 W INT — F N 21.85 15.44
2 40×5 8:1 355 R UPN 65 W CONT — F N 21.85 15.44
3 40×5 8:1 355 R UPN 120 B — S F N 56.37 28.06
4 40×5 8:1 355 R UPN 120 B — H F N 56.37 28.06
5 40×5 8:1 355 R UPN 120 B — H T Y 56.37 28.06
6 40×5 8:1 355 R UPN 120 B — H T N 56.37 28.06
7 40×5 8:1 355 BU 60×6−25×25×2 W INT — T N 6.08 9.44
8 40×5 8:1 355 BU 80×5−50×5 B — H T N 4.71 10.84
9 60×5 12:1 280 BU 80×5−25×25×2 W INT — T N 4.74 10.06
10 80×5 16:1 280 BU 100×5−25×25×2 W INT — T N 3.16 11.62
M. E. ERYAŞAR AND C. TOPKAYA
11 60×5 12:1 280 BU 100×5−50×5 B — H T N 4.86 12.40

12 80×5 16:1 280 BU 120×5−50×5 B — H T N 3.29 13.98
R: Rolled; BU: Built up; W: Welded; B: Bolted; INT: Intermittent; CONT: Continuous; S: Snug Tight; H: Hand Tight; F: Friction; T: Tack Weld;
Y: Yes; N: No; Pcr: Critical Buckling Load of Encasing; Py: Yield Load of Core Brace.
DOI: 10.1002/eqe
Figure 1. Longitudinal views of specimens.
DOI: 10.1002/eqe
Figure 2. Cross-sectional views of specimens.
DOI: 10.1002/eqe
to the crossheads of the test setup. Four 20-mm-diameter high strength European grade 8.8 bolts [9]
were used to connect the specimen to any one of the crossheads. This kind of an attachment detail
provides full fixity against rotation at both ends.
First eight specimens had 40 mm by 5 mm core segments. Specimens 9 and 11 had 60 mm by
5 mm and specimens 10 and 12 had 80 mm by 5 mm core segments, respectively.
Buckling restraining mechanism—phase 1

When properly designed and detailed, steel encasing should not resist any significant axial load [2].
To avoid global buckling of BRBs, Watanabe et al. [10] suggested that the steel encasing be
designed for sufficient flexural stiffness such that
Pcr
1.5 (2)
Pysc
where Pcr is the elastic buckling strength of steel encasing.
Equation (2) has a theoretical basis [5] and has been verified through experimental testing [3].
Steel encased BRBs designed to meet the criterion given in Equation (2) have been experimented
in the past. Representative Pcr /Pysc ratios used in studies by Tremblay et al. [7] and Iwata
et al. [11] were 3.5 and 12, respectively. In these experiments, it was observed that the encasings
were adequate to prevent global buckling of the brace. However, as mentioned before, significant
problems were reported at high strains due to the formation of local buckles and nonuniform
strains.
The present study does not aim at evaluating the validity of Equation (2) for steel-encased
braces. Based on the findings of the previous studies, it was assumed that the criterion given in
Equation (2) is adequate for the design of the buckling restraining mechanism.
The focus of the first phase was to develop details for core–encasing interaction so that the
problems associated with the steel-encased BRBs can be eliminated. Therefore, the design of
buckling restraining mechanism was not considered as a parameter in this phase. Throughout this
research, the core segment was sandwiched between steel encasing members as shown in Figure 2
to prevent weak axis buckling (x-axis). In the first phase, it was decided to use stiff rolled channel
shapes as encasing members. The selection of the channel type was dependent on the required
depth of the section. The first two specimens (Figure 2) had welded attachments; therefore, a UPN
65 channel (depth = 65 mm) was adequate. For specimens 3–6 (Figure 2) bolted attachments were
used and a UPN 120 channel (depth = 120 mm) was selected based on the depth requirement.
The Pcr /Pysc ratios were 22 and 56 for UPN 65 and UPN 120 encasing members, respectively.
In finding the buckling load of the restraining mechanisms with bolted connections, it was assumed
that both sections bend independently. Strictly speaking, the encasings for the first six specimens
were intentionally over designed. As shown in Table I, the encasing weight was 15 and 28 kg for
specimens having UPN 65 and UPN 120 buckling restraining mechanisms.
Buckling restraining mechanism—phase 2

Buckling restraining mechanisms of the second phase specimens were designed after phase 1
experiments had been conducted. Basically, the response of core segments with different aspect
ratios were studied in phase 2. In addition, the buckling restraining mechanisms were optimized
to come up with lighter and inexpensive designs. The criterion given in Equation (2) was used
in the design of encasings together with the constraints on dimensions and material availability.
DOI: 10.1002/eqe
In a recent study by Iwata and Murai [3] on mortar filled planks, researchers concluded that the
cumulative ductility of specimens increase with an increase in Pcr /Pysc ratio. In their study, the
specimens had Pcr /Pysc ratios that range between 0.9 and 4.4.
Typical buckling restraining mechanisms in phase 2 include built-up shapes that consist of a
guide plate welded to either a box section or a flat bar as shown in Figure 2. The size of the guide
plate was determined by the width of the core plate. For specimens 7, 9, and 10 a box section
with 25 mm side dimensions and 2 mm thickness was welded to the guide plate. For specimens 8,
11, and 12 a flat bar with 50 mm by 5 mm dimensions was welded to the guide plate. Selection
of the box section and the flat bar was based on the market availability. The resulting buckling
restraining mechanisms had Pcr /Pysc ratios that range between 3.16 and 6.08. Specimens 7, 9,
and 10 had welded while the others had bolted connections in this phase. The weight of encasing
varied between 9 and 14 kg for phase 2 specimens. When compared with phase 1 specimens, these
have equal or higher yield loads and lighter encasings.
Detailing for core–encasing interaction—phase 1

In the first phase of the experimental program several details were tested to overcome the problems
reported by earlier researchers [7, 11]. The primary concern related with steel-encased BRBs was
the formation of local core buckling that result in the friction between the core and the encasing
and lead to a nonuniform strain demand [7]. As mentioned before, the Poisson’s effect has to be
considered during the design of BRBs. For example, in the study by Tremblay et al. [7] researchers
provided a gap between the core and the encasing in both transverse directions (through thickness
and through width). In addition, no debonding material was utilized. In the present study, a different
approach was adopted by the authors. Fundamentally a gap was provided in the through width
direction. No gap was provided in the through thickness direction, leading to direct contact between
the core and the encasing.
In order to prevent strong axis buckling (y-axis), filler plates (Figure 2) having cross-sectional
dimensions of 20 mm by 5 mm were placed on both sides of the core segment. Filler plates were
continuous along the length of the member and there was a gap of 1 mm between the filler plate
and the core segment. This gap allowed for the lateral expansion of the core in the through width
direction. For all specimens a 80 mm by 25 mm portion (Figure 1) at both ends of the encasing
members were removed to allow for the free elongation and shortening of the core segment. When
the core segment yields, it buckles in the strong axis due to a significant loss of stiffness. This
kind of a buckling pattern is allowed to a certain degree because of the gap present between the
core segment and the filler plates. The load carrying capacity of the BRB does not reduce because
of the limited amount of strong axis bending. Essentially, the core segment bends with respect to
the strong axis and comes into contact with the filler plates that restrain the strong axis buckling.
For all specimens the 900-mm-long yielding core segment was wrapped with four layers of
0.05-mm-thick polyethylene film, which was secured by a tape. The polyethylene film functioned
as a debonding material. For specimens 5–12, grease was also applied to the surface of the outer
polyethylene film to help reduce the frictional forces. Width and thickness of the core segment
were measured at five locations using a digital caliper, which is accurate to 0.01 mm before the
film was wrapped. Same measurements were conducted after testing to monitor the strain levels
along the core segment.
Previous research [7, 11] mostly focused on bolted attachments for encasing members. In this
research both welded and bolted attachments were examined. In addition, specimens were tested
DOI: 10.1002/eqe
horizontally in the past [7]. In order to investigate potential core segment to buckling restraining
mechanism attachment details, all specimens were tested vertically in the present research program.
The first two specimens had welded connections between the steel encasing members. In the
first specimen, filler plates and encasing members were welded to each other at large intervals
in order to reduce the amount of friction produced by the compressive stresses formed during
the cooling of welds. The staggered–intermittent welding detail is given in Figure 1. In general,
60-mm-long welds were applied at 80 mm spacing. Contrary to specimen 1, continuous fillet welds
were utilized in specimen 2. For both specimens the vertical movement of the buckling restraining
mechanism due to gravity was prevented by the frictional forces formed between the core and the
encasing. Stiffening plates (Figure 1) on both sides and at both ends of the specimen were welded
to the channel sections to prevent local buckling of the core in these regions.
Specimens 3–6 had bolted connections between the steel encasing members. Iwata et al. [11]
used a similar type of attachment but connected the channel members through their flanges. This
kind of an attachment detail produces shear forces on the bolts, leading to separation between the
encasing members. In the present research, the channel members were connected through their webs
(Figure 2) so that bolts remained under the action of tensile forces resulting in minimized separation
between the encasing members. Similar type of attachment was also adopted by Tremblay et al.
[7]. Bolts were closely spaced to minimize the local bending of encasing members. A total of 40
grade 8.8 high strength bolts were utilized to connect the filler plates and the channel sections.
As shown in Figure 2 bolts ran through the webs of the channels and the filler plates. Bolt holes
were drilled to 10 mm diameter and the spacing between them was 50 mm center-to-center. For
specimens 3 and 4, 10 mm diameter (M10) bolts and for specimens 5 and 6, 8 mm diameter (M8)
bolts were used.
The difference between specimen 3 and 4 was the amount of pretension applied to the bolts.
In specimen 3, bolts were hand tightened first and then they were brought to snug-tight condition
by applying three quarter turns using a spud wrench. In specimen 4, only one row of bolts at the
center of the specimen was tightened to snug-tight condition while all others were hand tightened.
For both of these specimens, frictional forces were relied upon to prevent vertical movement of
the encasing.
In specimens 5 and 6 the effect of core-to-encasing attachment was studied. Contrary to other
four specimens, these do not rely on the frictional resistance to prevent slipping of encasing. In
order to prevent slip, core segment was tack welded to one of the channel sections at the midspan.
Tack welds were deposited between the filler plates and the core segment. All bolts were hand
tightened. The primary difference between specimens 5 and 6 was the imperfection present in
specimen 5 due to manufacturing. It was observed that the encasing members did not perfectly
fit to the core segment after the encasing member was fastened (Figure 2). Basically due to the
uneven cooling of the tack welds there was a gap between the core plate and the channel section.
When both channel sections were fastened to the core segment there was a 1 mm side opening at
the mid-length between the channels. At this point it was realized that this kind of an imperfection
may as well be present in a BRB that will be constructed for a real application in the future.
Therefore, in order to observe the consequences of having such imperfections, authors decided to
test this specimen without any modifications.
For specimen 6 a different type of tack weld detail was used to reduce the amount of imperfection
due to cooling of welds. Basically, cross-sectional areas of filler plates were reduced at the mid-
span as shown in Figure 1. This reduction enabled to deposit the tack welds with more precision.
In the welding process the core plate was first secured to the channel encasing by making use
DOI: 10.1002/eqe
of C-clips. Then tack welds were deposited to both sides of the core plate and allowed to cool.
After cooling of the welds the C-clips were removed and the buckling restraining mechanism was
fastened using high strength bolts. A photo of the tack weld detail can be found in Eryaşar [12].
Detailing for core–encasing interaction—phase 2

Detailing for phase 2 specimens were conducted after phase 1 testing had been completed. Basically,
the details that showed the best performance were selected and applied to phase 2 specimens. The
same type of unbonding scheme was applied. For all specimens the core brace was tack welded
to one of the encasing members with a detail that was adopted for specimen 6.
Specimens 7 and 8 had an aspect ratio of 8, similar to the phase 1 specimens. The difference
between these specimens and the phase 1 specimens is the use of built-up shapes as encasing
members. Specimen 7 had an encasing with welded connections while specimen 8 had a bolted
one. For the bolted specimens, same number of bolts with the same spacing used in phase 1 study
was adopted. All bolts were hand tightened and had a diameter of 8 mm. Contrary to continuous
welds used in phase 1, intermittent welds at every 50 mm were used to reduce the amount of
clamping force produced during the cooling of welds.
Specimens 9 and 10 utilized built-up encasings that were intermittently welded to each other. The
aspect ratio of the core plate for specimens 9 and 10 were 12 and 16, respectively. Specimens 11
and 12 were similar to specimens 9 and 10. Contrary to an intermittent welding detail, these two
specimens utilized bolted attachments. The aspect ratio of the core plate for specimens 11 and 12
were 12 and 16, respectively.
TEST SETUP AND LOADING PROTOCOL
A self-contained test setup was used in the experimental program. The test setup consists of a
frame, which was specifically built for this project. Tensile and compressive loads are applied to
the specimen by making use of a 250 kN screw jack. The screw jack is driven by a motor, which
is controlled through a frequency inverter. A 200 kN capacity load cell is mounted on the screw
jack to measure the level of axial loads. The details of the test setup can be found in Eryaşar [12].
During a typical experiment axial displacements of the restrained yielding segment were moni-
tored by making use of two 50 mm stroke string potentiometers with 0.01 mm accuracy. These
potentiometers were placed on both sides of the specimen by welding attachment rods to the
nonyielding cruciform segment. Axial load and displacement measurements were collected at
every 2 s using a digital data acquisition system. The average of the two displacement readings
was used to monitor the axial displacement.
All specimens were subjected to a stepwise incremental quasi-static loading protocol. AISC
Seismic Provisions [6] presents a loading protocol to be used in qualifying cyclic BRB tests.
The AISC procedure is based on the deformation quantity at first significant yield (by ) and the
deformation quantity corresponding to design story drift (bm ). According to the AISC procedure
specimens are cycled at increasing axial displacements up to two times the deformation corre-
sponding to the design story drift (bm ). A displacement value at design level needs to be assumed
to construct a loading protocol. In a paper by Tremblay et al. [7] researchers studied the anticipated
strain demand on the brace core. These researchers demonstrated that the strain demand is influ-
enced by several geometric, material factors and seismic design characteristics of the structure.
DOI: 10.1002/eqe
Nonetheless, Tremblay et al. [7] have shown that the strain demand generally remains within the
range of 1–2% unless the brace core is made significantly shorter, in which core strain values up
to 3–5% can be expected.
In the present study, the strain demand at the design level was taken as 1%. A loading protocol
similar yet different than the AISC loading protocol was adopted. In this loading protocol speci-
mens were cycled at eight different axial displacement levels, which are 13 of the yield displace-
ment, 23 of the yield displacement, the yield displacement, and displacements corresponding
to 0.33bm (0.33%), 0.5bm (0.5%), 1.0bm (1%), 1.5bm (1.5%), and 2bm (2%). Two cycles of
loading were applied at each displacement level. For the yield strength values used in the present
study this loading protocol produces cumulative axial deformation levels in excess of 200 times
the yield displacement as required by the AISC Seismic Provisions [6]. In the AISC protocol, the
specimens are directly loaded up to the yield displacement and 50% of the displacement corre-
sponding to the design story drift for the first and second displacement levels, respectively. In
the present study, three additional cycles were added to the loading protocol in order to monitor
the behavior of the specimens at low strain levels. These early cycles correspond to 13 and 23 of the
yield displacement and 33% of the displacement corresponding to the design story drift. Cycles at
low strain levels were needed to gradually increase the amount of axial strain on the specimens.
By following this procedure, the authors had a chance to stop testing and apply necessary changes
if anything goes wrong due to the manufacturing of the specimens. For specimens 11 and 12,
additional cycles at 2% strain were applied after the original sequence was completed. The aim
of these additional cycles was to determine the maximum cumulative inelastic deformations like
it was studied by earlier researchers [5].
Loading was applied slowly due to the limited motor speed and the need to closely monitor the
behavior of specimens. The total time to complete one test ranged between 3 and 4 h. The strain
rate in the core segment of the brace was 55 ε/s.
EXPERIMENTAL RESULTS AND OBSERVATIONS
The overall load-deformation responses for all specimens are given in Figures 3 and 4. The
compression strength adjustment factor, , and the strain hardening adjustment factor, , were
calculated for each strain cycle beyond the yield and the values are presented in Table III. The
stiffness values were calculated at one-third and two-thirds of the yield displacement and were
normalized by the theoretical stiffness values. The normalized secant stiffness at one-third and
two-thirds of the yield displacement are also tabulated in Table III for tension and compression.
In order to understand the variation of strains along the length, width and thickness measurements
were taken before and after each experiment. The change in cross-sectional dimensions was
measured at five locations that are shown in Figure 1. Basically, measurements were taken at the
ends, at the center, and at quarter points. The change in width and thickness at these five locations
are tabulated in Table II.
Buckling patterns of specimens were also investigated after each test was performed. Some
representative buckled configurations are given in Figure 5. In general, all specimens experienced
strong axis buckling due to the gap between the core segment and filler plates (Figure 5). Local
plastic deformations were observed for some specimens with bolted attachments (Figure 5). These
local deformations were usually located close to the ends of the specimens.
DOI: 10.1002/eqe
Figure 3. Normalized axial load vs percentage axial strain for specimens 1–6.
In general, all specimens except the first one showed satisfactory stable hysteretic behavior.
The specific observations and findings from each phase are explained in the following
sections.
Observations and findings from phase 1 specimens

Comparison of performances of specimens 1 and 2 revealed that largely spaced welds result
in an undesirable performance. Specimen 1 behaved well up to 1% strain. Minor load drops
were observed during the first compressive excursion of the 1.5% strain cycle. These load drops
were recovered as the strain was increased. Although the decreases in loads were recovered, the
global load-displacement response had a negative slope after compressive strain reached to 1%.
A separation between the channel sections due to the bending of the filler plates was observed
during the second compressive excursion of the 1.5% compressive strain. When the testing was
DOI: 10.1002/eqe
Figure 4. Normalized axial load vs percentage axial strain for specimens 7–12.
stopped, the mid-span opening between the two channel sections was 13 mm. A photo of specimen 1
after testing can be found in Eryaşar [12]. Contrary to specimen 1, the second specimen showed
desirable performance without any vertical movement of the encasing.
When performances of specimens 3 and 4 are compared, it can be concluded that snug-tight
bolts result in excessive amount of friction. The values for specimen 3 tend to be higher than
the ones of specimen 4. However, specimen 4 showed poor performance due to slipping. During
the second cycle of the 2% strain, the encasing completely slipped and came into contact with the
cruciform end of the core segment. This resulted in the transfer of forces to the buckling restraining
segment.
The use of tack welding the core plate to one of the channels (specimens 5 and 6) prevented
the vertical slip of the encasing and did not have an adverse effect on the global behavior.
When performances of specimens 5 and 6 are compared, it can be concluded that the presence
of imperfections had a negligible effect on the global performance. Local plastic deformations
DOI: 10.1002/eqe
Table II. Percentage strain values for width and thickness of specimens.
Specimen 1 Specimen 2 Specimen 3 Specimen 4 Specimen 5 Specimen 6
Point number b (%) t (%) b (%) t (%) b (%) t (%) b (%) t (%) b (%) t (%) b (%) t (%)
1 NA NA 3.09 3.05 4.76 4.31 4.63 8.91 1.69 2.12 1.41 1.16
2 NA NA −0.15 0.00 0.00 −1.75 1.37 3.70 0.70 1.75 1.20 0.20
3 NA NA 0.00 0.78 0.18 1.56 −0.20 −0.38 0.70 2.30 0.70 1.17
4 NA NA 0.52 2.91 1.10 1.16 −1.02 0.19 0.08 2.10 0.45 0.20
5 NA NA 3.71 4.85 5.15 5.25 0.95 1.54 1.35 1.73 1.35 0.78
Specimen 7 Specimen 8 Specimen 9 Specimen 10 Specimen 11 Specimen 12
Point number b (%) t (%) b (%) t (%) b (%) t (%) b (%) t (%) b (%) t (%) b (%) t (%)
1 0.88 0.20 0.93 0.75 1.47 1.18 0.83 −0.39 NA NA −0.98 −0.20
2 0.85 −0.58 0.63 1.33 1.09 1.19 0.74 −0.78 −0.42 0.99 −1.05 −1.79
3 0.43 0.97 0.67 1.14 0.42 0.40 0.30 0.20 −0.70 −0.79 −1.00 −0.40
4 1.00 1.94 0.80 1.72 0.89 0.80 0.59 0.79 −0.02 1.40 NA NA
5 1.18 1.54 1.12 0.95 1.24 2.21 NA NA 1.68 2.80 NA NA
b: Width; t: thickness.
and local buckling of the core segment in the weak direction were observed due to the presence
of imperfections. This led to load drops as shown in Figure 3. Although load values decreased
due to local plastic deformations and buckles, increase in strains resulted in the recovery of these
loads.
The tension stiffness at the first cycle is usually higher than the stiffness at later cycles. The
amount of friction is more influential on the initial cycles when compared with the later cycles. All
cases except a few had normalized stiffness values close to unity. For Specimens 2 and 3 excessive
stiffness values were measured. These are because of the continuous welding and significant
amount of bolt tightening.
Values in Table II reveal that for specimens 2 and 3 the strain distribution along the length is
nonuniform. Strains are more localized at the ends. These specimens had continuous welds or snug-
tight bolted attachments. Nonuniform straining was also observed for Specimen 4. Measurements
for this specimen are not reliable to draw firm conclusions because of the encasing slip that took
place during loading. Specimens 5 and 6 showed more uniform straining along the length.
Phase 1 testing reveal that carefully detailed core–encasing interaction can result in a desirable
performance. Among the six, specimen 6 showed the best overall performance. The details adopted
for this specimen was used throughout phase 2 testing.
Based on the observations, it can be concluded that providing a gap only in the through width
direction is sufficient to prevent local core buckling. Friction can be minimized by making use of
a debonding material and hand tightened bolted attachments. The use of grease together with a
polyethylene film helps reduce the friction further. However, films without grease did also perform
well. Core segment should be tack welded to the encasing at the midspan to prevent slipping. The
difference between maximum tensile and compressive loads reached to 9.4% for specimen 6, which
was well below the AISC limit. Encasing attachments should either be continuous or closely spaced
to prevent local bending of the buckling restraining mechanism. Continuous welds, however, lead
to the formation of nonuniform strains.
DOI: 10.1002/eqe
Copyright q
Table III. and factors and normalized secant stiffness values.
and factors for post yield strain amplitudes Tension and compression stiffness values
0.33% 0.50% 1.00% 1.50% 2.00% Tension stiffness Compression stiffness
2009 John Wiley & Sons, Ltd.

Specimen no. y/3 Cycle 2∗ y/3 Cycle y/3 Cycle 2∗ y/3 Cycle
1 1.059 0.979 1.048 0.982 1.059 1.048 1.013 1.073 0.909 1.098 1.043 1.084 1.044 1.052
2 1.032 1.008 1.089 1.015 1.139 1.096 1.163 1.163 1.217 1.218 1.548 1.275 1.063 1.053
3 1.076 1.113 1.090 1.172 1.149 1.266 1.202 1.324 1.257 1.368 1.769 1.199 1.167 1.130
4 1.042 1.000 1.054 1.011 1.153 1.089 1.237 1.176 1.352 1.248 0.923 1.110 0.901 1.071
5 1.006 0.935 1.054 0.934 1.070 1.008 1.077 1.075 1.139 1.131 0.891 1.032 0.895 0.991
6 1.011 1.000 1.038 1.002 1.064 1.082 1.075 1.152 1.094 1.210 1.138 0.931 1.024 0.948
7 1.088 1.046 1.084 1.053 1.131 1.120 1.166 1.180 1.201 1.232 1.060 1.112 0.958 1.096
8 1.000 1.042 1.042 1.035 1.059 1.096 1.067 1.169 1.089 1.225 1.064 1.070 0.983 1.024
9 1.024 1.032 1.041 1.018 1.071 1.072 1.090 1.155 1.111 1.229 0.993 1.042 0.933 1.020
10 1.021 1.016 1.021 1.019 1.056 1.054 1.070 1.134 1.085 1.204 1.021 1.053 0.941 0.989
11 1.028 1.011 1.020 0.998 1.047 1.051 1.068 1.133 1.081 1.207 1.074 1.034 0.935 0.959
12 1.045 0.987 1.024 0.998 1.043 1.041 1.056 1.115 1.046 1.187 0.939 1.008 0.868 0.902
STEEL ENCASED BUCKLING RESTRAINED BRACE HYSTERETIC DAMPERS
DOI: 10.1002/eqe
575
Figure 5. Buckling and local deformation patterns of specimens.
Observations and findings from phase 2 specimens

In order to help reduce the frictional forces and produce a more uniform strain demand along
the core segment, closely spaced intermittent welding was utilized in this phase. Performances of
DOI: 10.1002/eqe
specimens 7 and 8 reveal that built-up encasing members with closely spaced connections provide
adequate hysteresis. In addition, performances of other specimens showed that core plates with an
aspect ratio up to 16 can also perform satisfactory by adopting the details presented earlier. Aspect
ratios on this order have not been significantly studied in the past.
Additional cycles at 2% strain resulted in a cumulative ductility of 600 times the yield deforma-
tion for specimen 11. At the 7th additional cycle minor load drops under compressive loading was
observed. These drops, however, did not cause any detriment in the overall behavior. These load
drops were indications of local plastic deformations due to high amount of cumulative inelastic
strains.
Like Specimen 11 additional cycles at 2% strain were applied to specimen 12. During the
compressive excursion of the first additional cycle minor load drops were observed. In the following
compressive excursion the load drops were even more significant. It was observed that the core
brace experienced significant local plastic deformations at one end. A total of three additional
cycles were applied to this specimen and in all these cycles the maximum amount of compressive
force reached decreased. Testing had to be stopped due to the local buckle forming at the core
segment. The encasing members should have stiffeners at the ends due to the large aspect ratio
of the core plate. For this specimen longitudinal stiffeners were placed at the ends as shown in
Figure 1. However, there were no transverse stiffeners to prevent local bending of the encasing.
Placing transverse stiffeners can be considered an option in future studies to increase the amount of
cumulative ductility that high aspect ratio core plates sustain. This specimen endured a cumulative
ductility of 450 times the yield deformation.
In general, the normalized stiffness values for phase 2 specimens were close to unity. An
examination of the strain distributions reveal that most specimens were strained uniformly.
EVALUATION OF ADJUSTMENT FACTORS
Detailed evaluation of and factors, which are given in Table III, are explained in the following
sections. Readers should be cautioned that the curves generated in this section cannot be generalized.
Proper and factors provided by the BRB manufacturers should be used in design.
The strain hardening adjustment factor

The strain hardening adjustment factor, , depends on the steel properties, the loading history, and
the BRB details. The cyclic stress–strain properties of steel differ significantly from the monotonic
stress–strain behavior [13]. Cyclic hardening, cyclic softening, and mean stress relaxation takes
place during repeated loading [13]. Cyclic stress–strain curve for a particular steel can be obtained
by conducting multi-step tests. In such a test procedure, a steel coupon is cycled at various strain
levels beyond the yield strain until a saturation stress is reached. Cofie and Krawinkler [13]
conducted a multi-step test on an A36 specimen and defined the inelastic portion of the cyclic
stress–strain curve as follows:
¯ = K (ε̄p )n (3)
where ¯ is the normalized stress (note that ¯ is equivalent to ), ε̄p is the normalized plastic
strain, K is the cyclic stress–strain curve strength coefficient, and n is the cyclic stress–strain curve
hardening factor.
DOI: 10.1002/eqe
The K and n vales obtained from their experiments were 0.9 and 0.19, respectively. A plot of
the cyclic stress–strain curve is given in Figure 6. A fair assessment of factors can be performed
by grouping the specimens with the same material properties. Specimens 1–8 were from the
same batch and the remaining ones were from another batch. Specimen 1 was excluded from the
evaluation because of its poor performance. In addition, a careful examination of the data reveals
that the yield strength value of specimen 5 is a little lower than the rest of the specimens from the
same batch. The values for this specimen were below unity for strains less than 0.5%. Therefore,
this specimen was also excluded from the evaluation.
The experimental results for the factors are plotted in Figure 6. It is obvious from this figure
and Table III that the factors for specimen 3 were a lot higher compared to its counterparts.
This is because of the frictional forces produced due to the tightening of the bolts. The rest of
the specimens can be grouped into two according to their material properties. Cyclic stress–strain
curves were fit to the experimental data. For S355 steel the K and n constants are 1 and 0.09,
respectively. For S275 steel the K and n constants are 0.95 and 0.09, respectively.
It was observed that the test data falls below the stress–strain curve obtained by Cofie and
Krawinkler [13]. The differences can be attributed to the differences in the materials used and
the loading history adopted. In order to obtain a cyclic stress–strain curve using multi-step tests,
10 to 20 excursions may be performed at each strain amplitude. In the BRB tests, however, only
two cycles of loading were applied at each displacement level. Large number of excursions led to
cyclic hardening of the material, and therefore, a higher stress level attained.
The experimental results reveal that attachments with snug-tight bolting can produce significant
amount of friction that results in an increase in the level of axial forces. All other attachment
details did not have a significant influence on the factors.
The compression strength adjustment factor

The compression strength adjustment factor, , is also influenced by the amount of friction and
with the degree of lateral restraint. Specimen 4 is the only case where the factor is larger than
the AISC limit of 1.3. In this case, however, slipping of the encasing at later stages caused it to
carry some axial load, which in turn modified that balance between load levels in tension and
compression.
Encasing attachment details have a direct impact on the factor. Attachment details in general
influence the amount of frictional resistance formed between the core and the encasing. Specimens
were grouped into three and the variation of values with the normalized plastic strain are given
in Figure 6. First, second, and third groups belong to the hand-tight bolted specimens, the welded
specimens, and the snug-tight bolted specimens, respectively. Note that specimen 4 was included
into group three because there was a single row of bolts that were snug-tightened. For all groups,
a trendline was fitted to the data points. The values can be represented as follows:
= C(ε̄p )+1 (4)
where C is the slope of the trendline, which is equal to 0.008, 0.0132, and 0.0298 for groups 1,
2, and 3, respectively.
It is obvious from the above discussion that the values are the lowest for specimens utilizing
hand-tight bolted attachments. Welded attachments in general produce higher factors compared
to hand-tight bolted ones. The factors significantly increase if snug-tight bolting is utilized.
DOI: 10.1002/eqe
Figure 6. Evaluation of adjustment factors.
DOI: 10.1002/eqe
Data points that belong to the same group are generally scattered. This can best be observed
for the welded specimens. A careful examination of the data reveals that the axial yield load is
influential on the amount of scatter. When welded specimens 7, 9, and 10 are compared, it is
observed that for a particular strain value, factors tend to decrease with an increase in the axial
yield load. Specimen 10, which has the highest axial yield load, has the lowest factors among the
three specimens. If the welding process produces nearly equal frictional resistance for specimens
having the same thickness but different axial yield loads, then this frictional resistance is more
influential for a specimen with a lower yield load.
RECOMMENDATIONS FOR ACTUAL DESIGNS AND FUTURE RESEARCH NEEDS
Test results reveal that local core buckling can be minimized by providing a direct contact between
the core and the encasing. It is recommended that a debonding material be used at the contacting
surface. The application of grease on the debonding surface is desirable but not critical. It is
worthwhile to note that steel BRBs with no debonding material can also be used. The authors
performed a set of pilot tests on welded BRBs without a debonding material where the core plate
was in direct contact with the encasing members. Excessive friction did not develop in these tests,
indicating the potential use of BRBs without debonding. In addition, similar conclusions can be
drawn by examining the test results of Tremblay et al. [7]. In these experiments one specimen
had a small gap at the interface and no debonding material was used. This particular specimen
performed well until the formation of local buckles at the core plate ends. Because the cost of
using a debonding material is less when compared to the total cost of manufacturing a steel BRB,
it is recommended that such a material is used at the interface.
In actual design, a gap in the through width direction should be provided to allow for the free
expansion of the core under compressive forces. It is recommended that the core plate be tack
welded to the encasing at the midspan. This kind of an attachment detail will prevent slipping
of the encasing. Friction should not be relied upon in preventing the slipping of the buckling
restraining mechanism.
Continuously welded and snug-tight bolted attachments for encasing should be avoided. These
kinds of details produce nonuniform strains and high frictional resistances.
Imperfections due to manufacturing should be kept to a minimum. Imperfections are not detri-
mental in terms of global load deformation response; however, these can lead to large local plastic
deformations resulting in early local buckling.
Based on the test results, core segments with aspect ratios up to 16 can be used in actual designs.
The authors consider that the encasing should be designed such that Pcr /Pysc ratio is greater than
1.5 as recommended by Watanabe et al. [10]. In addition, it is preferred that this ratio be greater
than 3.0, which was the lower limit experimented in this research program. Built-up sections that
utilize a guide plate can be used to form a buckling restraining mechanism. Both welded and bolted
attachments can be used for the encasing. These attachments should be closely spaced to prevent
local bending of the restraining mechanism. It is recommended that the bolts be hand tightened
and the welds be deposited intermittently to minimize the amount of friction. BRBs with welded
encasings are lighter when compared with their counterparts.
Good quality control is needed during manufacturing to alleviate the randomness and uncertain-
ties associated with bolt tensioning. In addition, if bolt loosening is considered to be a problem
then special type of bolts that cannot be unfastened can be used in manufacturing bolted BRBs.
DOI: 10.1002/eqe
This study was limited to small-scale component testing. Future research should consider
applying the findings of this study to full scale specimens. Specimens with Pcr /Pysc ratios in
between 1.5 and 3.0 should be tested to investigate their performance. In addition, spacing of
welds and bolts should be studied to minimize the amount of fasteners. Component testing should
be extended to subassemblage testing to investigate the end connection behavior of the type of
steel-encased BRBs presented in this study.
CONCLUSIONS
An experimental study on 12 steel-encased BRBs were presented. Based on the experimental

results, core–encasing details that minimize local core buckling were developed. Test results reveal
that providing a direct contact between the core and the encasing is adequate for stable hysteresis
at high strains. Continuously welded and snug-tight bolted attachments for encasing were found
to be detrimental. Tack welding the core to the encasing was found to be a considerable option
for preventing slipping of the encasing. A set of design recommendations was presented based on
the experimental findings.
The adjustment factors were evaluated based on the experimental results. A set of expressions
was developed to quantify the adjustment factors. Continuous welding and snug-tight bolting were
found to increase these factors when compared with intermittent welds and hand-tight bolting.
Furthermore, axial yield load was found to be influential on the compressive strength adjustment
factor. Properly detailed steel-encased BRBs having core plate aspect ratios up to 16 were found
to provide adequate hysteresis.
REFERENCES
1. Xie Q. State of the art of buckling-restrained braces in Asia. Journal of Constructional Steel Research 2005;
61:727–748.
2. Uang C, Nakashima M. Steel buckling-restrained braced frames. Earthquake Engineering from Engineering
Seismology to Performance Based Engineering. CRC Press: Boca Raton, 2004.
3. Iwata M, Murai M. Buckling-restrained brace using steel mortar planks; performance evaluation as a hysteretic
damper. Earthquake Engineering and Structural Dynamics 2006; 35:1807–1826.
4. Iwata M, Kato T, Wada A. Performance evaluation of buckling-restrained braces in damage controlled structures.
Proceedings STESSA, Naples, 2003; 37–43.
5. Black CJ, Makris N, Aiken ID. Component testing, stability analysis and characterization of buckling restrained
braces. Report No. PEER 2002/08, University of California, Berkeley, 2002.
6. American Institute of Steel Construction (AISC). Seismic Provisions for Structural Steel Buildings, 2005.
7. Tremblay R, Bolduc P, Neville R, Devall R. Seismic testing and performance of buckling-restrained bracing
systems. Canadian Journal of Civil Engineering 2006; 33:183–198.
8. EN 10025. Hot rolled products of non-alloy structural steel. European Committee for Standardization. CEN:
Brussels, 1994.
9. BS 3692. Specification for Iso-Metric Precision Hexagon Bolts, Scres and Nuts. British Standards Institute:
London, 2001.
10. Watanabe A, Hitomi Y, Saeki E, Wada A, Fujimoto M. Properties of brace encased in buckling-restraining
concrete and steel tube. Proceedings of the 9th World Conference on Earthquake Engineering, Tokyo, vol. IV,
1988.
11. Iwata M, Kato T, Wada A. Buckling-restrained braces as hysteretic dampers. Proceedings STESSA, Montreal,
2000; 33–38.
12. Eryaşar E. Experimental and numerical investigation of buckling restrained braces. Thesis, Middle East Technical
University, Ankara, Turkey, 2009.
13. Cofie NG, Krawinkler H. Uniaxial cyclic stress-strain behavior of structural steel. Journal of Engineering
Mechanics 1985; 111:1105–1120.
DOI: 10.1002/eqe

An Experimental Study On Steel Encased Buckling Restrained Brace Hysteretic Dampers PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

An Experimental Study On Steel Encased Buckling Restrained Brace Hysteretic Dampers PDF

Uploaded by

Copyright:

Available Formats

EARTHQUAKE ENGINEERING AND STRUCTURAL DYNAMICS

Earthquake Engng Struct. Dyn. 2010; 39:561–581

An experimental study on steel-encased buckling-restrained

Mehmet Emrah Eryaşar and Cem Topkaya∗, †

Received 28 May 2009; Revised 10 August 2009; Accepted 17 August 2009

INTRODUCTION AND BACKGROUND

Copyright q 2009 John Wiley & Sons, Ltd.

TEST SPECIMEN SELECTION AND DESIGN

Core plate Encasing

2009 John Wiley & Sons, Ltd.

11 60×5 12:1 280 BU 100×5−50×5 B — H T N 4.86 12.40

Figure 1. Longitudinal views of specimens.

Figure 2. Cross-sectional views of specimens.

Buckling restraining mechanism—phase 1

Buckling restraining mechanism—phase 2

Detailing for core–encasing interaction—phase 1

Detailing for core–encasing interaction—phase 2

TEST SETUP AND LOADING PROTOCOL

EXPERIMENTAL RESULTS AND OBSERVATIONS

Observations and findings from phase 1 specimens

2009 John Wiley & Sons, Ltd.

Figure 5. Buckling and local deformation patterns of specimens.

Observations and findings from phase 2 specimens

EVALUATION OF ADJUSTMENT FACTORS

The strain hardening adjustment factor

The compression strength adjustment factor

= C(ε̄p )+1 (4)

Figure 6. Evaluation of adjustment factors.

RECOMMENDATIONS FOR ACTUAL DESIGNS AND FUTURE RESEARCH NEEDS

An experimental study on 12 steel-encased BRBs were presented. Based on the experimental

You might also like