See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/273370264

Experimental Validation of Replaceable Shear Links for Eccentrically Braced

Steel Frames

Article  in  Journal of Structural Engineering · October 2011

DOI: 10.1061/(ASCE)ST.1943-541X.0000350

CITATIONS READS
174 1,247

3 authors, including:

Nabil Mansour Robert Tremblay

Badr University Cairo Polytechnique Montréal
5 PUBLICATIONS   294 CITATIONS    301 PUBLICATIONS   7,012 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Nonlinear analysis and seismic design of steel friction braced frames View project

NEESR: Reserve Capacity in New and Existing Low-Ductility Steel Braced Frames View project

All content following this page was uploaded by Nabil Mansour on 04 May 2021.

The user has requested enhancement of the downloaded file.

 Experimental Validation of Replaceable Shear Links for
Eccentrically Braced Steel Frames
Nabil Mansour1; Constantin Christopoulos, M.ASCE2; and Robert Tremblay3

Abstract: In the current design of steel eccentrically braced frames (EBFs), the yielding link is coupled with the floor beam. This often
results in oversized link elements, which leads to overdesigned structures and foundations. In addition, the beams are expected to sustain
Downloaded from ascelibrary.org by University of Toronto on 03/03/20. Copyright ASCE. For personal use only; all rights reserved.

significant damage through repeated inelastic deformations under design-level earthquakes, and thus the structure may require extensive
repair or replacement. These drawbacks can be mitigated by designing EBFs with replaceable shear links. Two different replaceable link
types with alternate section profiles, connection configurations, welding details, and intermediate stiffener spacings were tested. A total of 13
cyclic quasi-static full-scale cyclic tests were performed, including tests on EBFs with replaceable shear links, to study their inelastic seismic
performance. The links exhibited a very good ductile behavior, developing stable and repeatable yielding. Additional inelastic rotation
capacity can be achieved with bolted replaceable links when bolt bearing deformations are allowed to occur. The on-site replaceability of
the link sections is confirmed even in the presence of residual deformations of 0.5% drift. DOI: 10.1061/(ASCE)ST.1943-541X.0000350.
© 2011 American Society of Civil Engineers.
CE Database subject headings: Steel frames; Bolted connections; Inelasticity; Rotation; Beams.
Author keywords: Aftershocks; Eccentrically braced steel frames; Shear link; Bolted connection; Replaceable link; Inelastic rotations;
Bearing.

Introduction thus the structure’s ability to provide an adequate level of safety for
any subsequent loading is uncertain. Furthermore, repair of the link
Eccentrically braced frames (EBFs) are seismic lateral load- can be very difficult, disruptive, and costly.
resisting systems that include a ductile, energy dissipating portion The design of EBFs with replaceable links addresses these
in the beam elements, referred to as the link. The design of an drawbacks and optimizes the design by decoupling the yielding
EBF depends on defining the strength, stiffness, and ductility of link from the rest of the structure. Using the replaceable link con-
the system, which are closely related to the link size, length, and cept, the designer has greater flexibility to choose a section for the
detailing. In current design [e.g., AISC 2005; Canadian Standards yielding link that best meets the required strength without automati-
Association (CSA) 2009], the link is part of the floor beam, and cally changing the floor beam section. The stiffness and fundamen-
therefore, the floor beam must be designed not only to yield in tal period of vibration of EBFs are largely controlled by the link
the link region, but also to resist forces due to the strain hardening length, and EBFs with short links, preferred for their more stable
of the link region in the portions of the floor beam that are located shear yielding behavior (Engelhardt and Popov 1989), result in
outside the link. Balancing these requirements can be an iterative stiffer structures, which attract larger base shear forces. By decou-
and exacting process often resulting in oversized link elements, pling the replaceable link from the floor beam, the link length is no
which leads to larger force demands on all the other EBF members, longer defined by the distance between the eccentric braces, and
including the braces, columns, floor diaphragms, connections, and thus EBFs with larger door or corridor openings can be designed,
foundations. Ultimately this results in overdesigned structures and while still ensuring that a short shear critical link behavior is
increased overall costs. In addition, the beam can sustain significant achieved. Because the inelastic deformation will be concentrated
damage due to repeated inelastic deformation and localized buck- only within the link, damaged links can be quickly inspected and
ling in the link region during a design-level earthquake. Because replaced following a major earthquake, significantly minimizing
the cumulative inelastic deformations imposed on the structure dur- the disruption time of the structure and extending its life span.
ing an earthquake are unknown, it is difficult to assess the extent of Furthermore, the use of replaceable links allows for the welding
damage, especially when no visible signs of this are apparent, and of critical elements, such as the brace connections, to be performed
in the shop, considerably improving construction quality and
1
Senior Structural Engineer, Dialog, Toronto, ON, Canada; formerly, reducing erection time. Fig. 1 illustrates an EBF designed accord-
Ph.D. Candidate, Univ. of Toronto. ing to this replaceable link concept. The shaded portions of the
2
Associate Professor, Dept. of Civil Engineering, Univ. of Toronto, ON, structure represent elements spanning multiple stories that can
Canada (corresponding author). E-mail: [email protected] be fabricated in the shop, shipped, and assembled on site through
3
Professor, Dept. of Civil, Geological and Mining Engineering, École the bolting of the replaceable links. Multiple modules fabricated in
Polytechnique, Montréal, Canada. the shop can be spliced, as illustrated in the figure, to construct
Note. This manuscript was submitted on April 18, 2010; approved on
taller buildings. In addition, replaceable links made from built-up
November 30, 2010; published online on September 15, 2011. Discussion
period open until March 1, 2012; separate discussions must be submitted sections with relatively thin webs and thick flanges can be used to
for individual papers. This paper is part of the Journal of Structural En- further optimize the design. Links made of steel grades with lower
gineering, Vol. 137, No. 10, October 1, 2011. ©ASCE, ISSN 0733-9445/ yielding strengths than the surrounding frame members can also
2011/10-1141–1152/$25.00. represent a desirable and practical option.

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / OCTOBER 2011 / 1141

J. Struct. Eng., 2011, 137(10): 1141-1152

 Design of Replaceable Link Specimens

Two types of replaceable link configurations with alternate link-to-

beam connections were tested. The first type is fabricated from a
W-section welded to unstiffened end plates, which are bolted to the
floor beam end plates, as shown in Fig. 2(a). This link type is sim-
ilar to those tested by Stratan and Dubina (2004). The second re-
placeable link type, similar to that suggested by Balut and Gioncu
(2003) as an improved reduced beam section solution for moment
resisting frames, has not been tested before. It consists of two chan-
nel sections, back-to-back, connected to the web of the floor beam
through an eccentrically loaded bolted or welded web connection,
as shown in Fig. 2(b). The specifications of each specimen, includ-
ing the primary features of the link-to-beam connection, link
Downloaded from ascelibrary.org by University of Toronto on 03/03/20. Copyright ASCE. For personal use only; all rights reserved.

length, and link stiffener spacing are summarized in Table 1. The

specimens were designed as shear critical links according to the
Canadian Standard CAN/CSA-S16-09 (CSA 2009). The specimens
Fig. 1. Replaceable link concept frame that were studied are representative of the links at the second floor
of a five-story EBF frame, designed according to the 2005 National
Building Code of Canada (NBCC) [National Research Council of
Past studies by Popov and Engelhardt (1988) and Kasai and Canada (NRCC) 2005] specified seismic loads, representative of a
Popov (1986) have demonstrated the excellent seismic performance high seismic area in British Columbia, Canada. As indicated in the
of EBFs and provided invaluable data essential to establishing a table, the 13 specimens are categorized into three groups: (1) links
cohesive design procedure. Ghobarah and Ramadan (1994) exam- with end-plate connections, (2) back-to-back double channel links
with bolted web connections, and (3) a back-to-back double chan-
ined end-plate connections of shear links connected to columns in
nel link with welded web connections.
EBFs under cyclic loading. Okazaki and Engelhardt (2007) further
All the links met Class 1 section requirements
pffiffiffiffiffi for the web and
pffiffiffiffiffi
examined link-to-column connections in EBFs and the behavior of
flanges, h=w ≤ ð1;100= F y Þ and b=t ≤ ð200= F y Þ, respectively
link beams constructed of ASTM A992 steel now commonly used (in metric units). All link specimens, with exception of UT-1, were
in North America. In quasi-cyclic tests carried out in Europe on fabricated from CSA G40.21-350W grade steel, which met the re-
bare steel links built from W-shapes with flush end-plate bolted quirements of ASTM A992 steel. Specimen UT-1 was fabricated of
connections, Stratan and Dubina (2004) and Dubina et al. (2008) CSA G40.21-300W grade steel. All welding on the test specimens
showed that well-proportioned replaceable links can exhibit high was done using the flux cored arc welding (FCAW) process with
ductility capacity, but further attention to the connection design E491T-CSA/CWB series (E71T-AWS) electrodes, which are com-
was needed. Full-scale testing of built-up shear links as energy parable in toughness of weld metal to E7018 stick electrodes used
dissipators for the seismic protection of the San Francisco in shielded metal arc welding (SMAW). All bolted connections
Oakland Bay Bridge with bolted end connections was presented by used 25.4 mm (1 in.) A490 bolts.
McDaniel et al. (2003). The link is first selected by calculating the required shear area to
This paper presents an experimental research program that was resist the imposed shear in the link due to story shears and frame
conducted at the University of Toronto (UT) and École Polytech- geometry. The link plastic shear resistance, V p , given in Clause
nique of Montreal (EPM) to develop and assess the design of 27.7.2 of CSA-S16 (2009) is
replaceable links for EBFs built with steels used in North American V p ¼ 0:55wdF y ð1Þ
practice. Two link designs were examined: W-shapes with bolted
end-plate connections and back-to-back C-shapes with bolted or where w = web thickness; d = overall depth; and F y = yield stress
welded web connections. The test program included cyclic tests of the link web. The steel sections are selected primarily based on
on both individual link specimens and full-scale eccentrically optimizing the required shear area, i.e., minimizing wd. Stocky and
braced frame subassemblages. Subassemblage specimens with a wide sections typically used for columns with compact flanges and
composite floor slab were studied, and axial load effects on link sufficient bending capacity are preferred to develop shear yielding
response could be examined. The design of the replaceable link of the section under ultimate loads.
test specimens is presented in this paper, highlighting elements Defining the link length is the second step. In traditional design
in the design procedure that differ from traditional EBF design. of EBFs, in which the link is part of the floor beam, the link length
The testing procedures are discussed, followed by an overview is the distance between the eccentric braces and is often chosen to
of the test results. Key observations are presented that pertain to accommodate the size of the corridor opening required by the
architect based on building usage. Shorter links that yield in shear
the shear yielding behavior of the replaceable links and the influ-
(where e < 1:6M p =V p ) are preferred because these have a more
ence of the connection behavior on the response of the double
stable energy dissipation mechanism and a more predictable
channel web-connected links, as well as the connection rigidity, postyield behavior than the longer links that yield in flexure
bolt slippage, and bearing deformations on the overall link hyster- (Engelhardt and Popov 1989). In the case of EBFs designed with
etic response. The experimental performance of the full-scale EBF replaceable links, the opening size, e0 , is decoupled from the link
with replaceable links is then presented to evaluate the global length, e, as shown in Fig. 1; and thus the length of the link that
response of the frame and assess the influence of a composite floor governs its response is no longer dependent on the required open-
slab on the link behavior. The ability to replace the damaged link by ing size. However, the floor beam segment between the link ends
means of either a bolted web connection or a welded web connec- and the brace connections must be designed to resist the required
tion is also investigated. strength based on the link ultimate strength.

1142 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / OCTOBER 2011

J. Struct. Eng., 2011, 137(10): 1141-1152

Downloaded from ascelibrary.org by University of Toronto on 03/03/20. Copyright ASCE. For personal use only; all rights reserved.

Fig. 2. Replaceable link specimens: (a) end-plate connected; (b) web connected

Table 1. Test Specimens

Link length
Intermediate Web connection Normalized bolt
Group Specimen Section e (mm) e=ðM p =V p Þ stiffeners reinforcement plate bearing stress
End-plate connection UT-3A W360 × 101 900 1.16 3 at 200 mm None —
UT-3B W360 × 101 900 1.16 3 at 200 mm None —
EPM-11A W360 × 72 800 1.60 3 at 200 mm None —
EPM-11B W360 × 72 800 1.60 3 at 200 mm None —
Bolted web connection UT-1A C310 × 31 680 1.04 4 at 200 mm 6.35 mm 1:66F u
UT-1B C310 × 31 680 1.04 4 at 200 mm 6.35 mm 1:70F u
UT-2A Cut W310 × 39 680 1.03 4 at 200 mm None 2:95F u
UT-2B Cut W310 × 39 680 1.03 4 at 200 mm 6.35 mm 1:65F u
EPM-12 Cut W250 × 24 680 1.40 5 at 150 mm 6.35 mm 1:41F u
EPM-13 Cut W250 × 24 680 1.40 7 at 99 mm 6.35 mm 1:51F u
EPM-14 Cut W250 × 24 680 1.40 7 at 99 mm 6.35 mm —
EPM-16 Cut W250 × 24 680 1.40 5 at 150 mm 4.76 mm 1:61F u
Welded web connection EPM-15 Cut W250 × 24 590 1.23 5 at 150 mm 6.35 mm —

was connected to 30-mm-thick end plates using complete joint

End-Plate-Connected Links penetration (CJP) welds, where weld access holes were provided.
Link specimens UT-3 and EPM-11 were fabricated from W- Specimen EPM-11 was welded to 28.6-mm-thick end plates using
sections welded to end plates and bolted to the floor beam end partial joint penetration (PJP) groove welds with reinforcement
plates, as shown in Figs. 3(a) and 3(b). The end plates had lengths fillet welds to connect the link flanges (without weld access holes)
and widths equal to the floor beam dimensions. Link UT-3 and fillet welds to connect the web. The stiffener welds were

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / OCTOBER 2011 / 1143

J. Struct. Eng., 2011, 137(10): 1141-1152

Downloaded from ascelibrary.org by University of Toronto on 03/03/20. Copyright ASCE. For personal use only; all rights reserved.

Fig. 3. End-plate-connected replaceable links: (a) UT-3; (b) EPM-11

terminated at a distance of five times the link web thickness from M f
beam ¼ 0:9 × ð1:3Ry V p Þe0 =2; V f
beam ¼ 1:3Ry V p ð3Þ
the k-line of the link cross section, as recommended by Okazaki
and Engelhardt (2005). Sets of two identical specimens were tested
M f
beam may be larger than M f
link end because it is calculated based
to assess the replaceability of the links and the performance of the
on the distance between the eccentric braces, e0 , not the link shear
replaced link in comparison to the original link.
length e.
For the replaceable links to achieve satisfactory performance,
the link-to-beam connections must be able to transfer the forces Web-Connected Links
generated by the fully yielded and strain-hardened links. The depth
of the floor beam must also account for erection tolerances to ac- The remaining nine link specimens were all double channel back-
commodate the bolting operation in the field. The probable moment to-back sections. With the exception of UT-1, the link specimens
and shear demand on the connection of shear yielding links can be were fabricated from W-sections that had the flange on one side
calculated as follows: saw-cut flush with the web to create channel sections. This provides
a greater range of channel sections for the design of the links. To
M f
conn ¼ M f
link end ¼ 1:3Ry V p e=2; V f
conn ¼ 1:3Ry V p optimize the shear area, sections with relatively thin webs, corre-
sponding to thin flanges, were selected. It was thus necessary to add
ð2Þ top and bottom flange reinforcement plates to increase the flexural
resistance of the double channel links to ensure a shear yielding
where Ry accounts for the probable yield stress exceeding the mini- dominant behavior. This is shown in Table 1, in which the values
mum specified value, and 1.3 accounts for the increase in stresses of the ratio e=ðM p =V p Þ are all less than 1.6.
attributed to strain hardening of the link yielding in shear. To allow Based on the observed experimental results and analyses of the
for the placement of the floor deck on top of the floor beam, the strains along the link length, it was concluded that the link shear
link-to-beam connections are sized such that the end plates are length of the web bolted channel links would be defined as the
flush with the floor beam section. Furthermore, because the link distance between the first row of bolts on either end, as shown in
section depth is less than the floor beam depth, the link is not Fig. 4. These replaceable links are connected to the web of the floor
connected to the floor slab, allowing for replacement of the link. beam through eccentrically loaded bolted connections. The depth
The end-plate link connections were designed as welded end-plate of the floor beam must allow for the link to fit into the beam. The
moment connections based on the recommendations provided by eccentric connections of the bolted web links were designed using
the Canadian Institute of Steel Construction (CISC) (2005). the method of instantaneous center (IC) of rotation with the load-
To ensure that the frame can be repaired with minimal effort, deformation relationship developed by Kulak et al. (1987), assum-
the floor beams, braces, columns, and their connections must be ing a shear force equivalent to 1:3Ry V p applied with an eccentricity
designed to remain essentially linear elastic when the yielding links equal to the distance between the center of connection rotation and
achieve their overstrength capacities. The code does not require a the link midspan.
moment connection between the brace and the floor beam. How- To investigate the bolt bearing deformation in the link connec-
ever, in practice, the connection between the brace and the floor tion and its effect on the total link rotation, the allowable bearing
beam is typically capable of transferring bending moments. From stress considered in the design was varied between 1.41 up to 2.95
elastic computer modeling analysis, it was determined that 10–15% times F u . Because the expected link shear force for each set of
of the link end moment is transferred to the brace. Hence, the floor identical link specimens is the same, the bolt bearing stress was
beam outside the link may be designed for 90% of the moment modified by changing the thickness of the added web reinforce-
at the beam-to-brace connection. The probable moment and shear ment plate. Values of the bearing stress for each specimen are listed
demand on the floor beam can be calculated as follows: in Table 1. These are based on the load at the most stressed corner

1144 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / OCTOBER 2011

J. Struct. Eng., 2011, 137(10): 1141-1152

Downloaded from ascelibrary.org by University of Toronto on 03/03/20. Copyright ASCE. For personal use only; all rights reserved.

Fig. 4. Bolted web-connected replaceable links, EPM-12

bolt, determined using the IC method and considering the maxi- Loading Protocol
mum measured link shear, V max and the link axial load. The bearing
stress is calculated by dividing the maximum bolt load by the bear- The loading protocol specified in Appendix S6.3 of the 2005 AISC
ing area (bolt diameter times total link thickness at connection) seismic provisions (AISC 2005) was used in all the tests. Accep-
and is normalized with respect to the material nominal tensile stress tance criteria for links are based on inelastic link rotation as defined
F u , equal to 450 MPa. This investigation was initiated based in both CSA-S16 (2009) and AISC (2005). The inelastic link
on preliminary observations in the test program (Mansour et al. rotation, γp , was evaluated by removing the contributions of the
2006), in which specimen UT-2A, which had no web reinforce- elastic response of the link rotation. Both codes specify that a shear
ment, failed at the connection due to excessive bolt-hole ovalization yielding link (e < 1:6M p =V p ) should be capable of developing an
(the bolt-hole size increased to 1.5 times its initial diameter), at inelastic rotation of 0.08 rad. The inelastic rotation capacity of the
γp ¼ 0:066 rad. Specimen UT-2A represents the minimum thick- link specimens was defined by AISC (2005) as the maximum level
ness at the connection of a web bolted link. To ensure a linear elas- of inelastic rotation sustained for at least one full cycle of loading
tic response of the floor beam web at the connection, the allowable prior to the link shear strength dropping below the nominal link
bearing stress at the most stressed bolt location should be limited to shear strength. Here, the nominal strength was evaluated based
0:80F u . This is achieved by welding a reinforcement plate to the on nominal section dimensions and nominal yield strength of
floor beam web. 350 MPa.
Furthermore, a critical section was defined across the first row
of bolts, where the link shear resistance is reduced due to the bolt
holes. Thus the resistance of the net shear area should also be Test Results
checked to ensure that it is capable of resisting the link shear:
Table 2 summarizes the test results, listing the maximum measured
V r; conn ¼ 0:60ϕAnv F u > 1:3Ry V p ð4Þ inelastic link rotations, γp ; maximum link shear force and over-
strength value, the maximum measured total frame story drift, θ;
where Anv = net shear connection area; F u = steel tensile stress; and the inelastic story drift, θp ; the connection contribution to the total
ϕ = 0.9 for steel. To meet these two limit state requirements, a link rotation, Γconn ; and a brief description of the controlling failure
reinforcement plate was welded onto the link web at the connec- mode for each specimen. All the specimens, with exception of
tion, as shown in Fig. 4, for all the specimens except UT-2A (see UT-2A, exceeded the inelastic link rotation angle of 0.08 rad (the
Table 1). full hysteretic responses of all the specimens are shown in Fig. 6).
The web-connected links were also provided with angles
welded to the top and bottom flanges. As illustrated in Fig. 4, the UT Test Setup and Results
angles were positioned such that they could be bolted back-to-back
upon assembling the links to ensure that the two channel sections, The first six link specimens, tested at UT, were designed and tested
with a gap between them equal to the floor beam web thickness, as a proof of concept and to verify that the replaceable links de-
would act as one unit, thus reducing the possibility of lateral tor- signed according to the procedure outlined can achieve the required
sional buckling. Guidance for determining the spacing of these ductility capacity of 0.08 rad plastic link rotation, as specified by
angles was obtained from the research performed by Kim et al. CSA-S16 (2009). A test setup, shown in Fig. 5, was devised to
(2003), which suggested using a clear spacing not exceeding twice subject the link to a constant shear force along its length, equal
the channel flange width. reverse curvature bending moment at its ends, and no axial force,
Although the web bolted channel replaceable links require similar to that experienced by the link in the example building, as
additional fabrication steps to those required for the end-plate- shown in Fig. 1. The measured link shear force is equal to the load
connected links, they are easier and faster to erect onto the floor applied by the MTS Systems Corp. test machine. Connection per-
beams. Furthermore, it is foreseeable to use web-connected links formance of the web bolted channel links was studied to identify
to upgrade or retrofit existing EBF systems because minimal site the primary factors that contribute to the link rotation capacity and
preparation and modification of the existing steel is required. to assess the effects of bolt slippage on the overall performance.

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / OCTOBER 2011 / 1145

J. Struct. Eng., 2011, 137(10): 1141-1152

 Table 2. Summary of Test Results
Frame drift (percentage) Link shear (kN)
Group Specimen γp θ θp Γconn V max V max =V n Failure mechanism
End-plate connection UT-3A 0.104 N/A N/A None 1,274 1.39 Fracture of web initiating at weld access hole
UT-3B 0.105 N/A N/A None 1,279 1.40 Fracture of web initiating at weld access hole
EPM-11A 0.095 1.52 1.12 None 1,120 N/Ab Fracture of top flange near weld
EPM-11B 0.112 1.71 1.40 None 1,170 N/Ab Fracture of web at stiffener weld
Bolted web connection UT-1A 0.125 N/A N/A 0.18 1,438 1.48 Cracks in link web at intermediate stiffener;
test terminated before link failure
UT-1B 0.166 N/A N/A 0.27 1,478 1.52 Fracture of web at stiffener weld
UT-2A 0.066 N/A N/A 0.60 1,132 1.21 Fracture of link at web connection
UT-2B 0.102 N/A N/A N/Aa 1,311 1.40 Exceed test setup deformation limits
Downloaded from ascelibrary.org by University of Toronto on 03/03/20. Copyright ASCE. For personal use only; all rights reserved.

EPM-12 0.106 1.32 1.12 0.04 916 1.58 Fracture of web at stiffener weld
EPM-13 0.104 1.18 0.85 0.09 976 1.68 Fracture of link at web connection
EPM-14 0.103 1.37 1.06 0.05 1108 N/Ac Fracture of web at stiffener weld
EPM-16 0.125 1.54 1.31 0.17 890 1.53 Fracture of web at stiffener weld
Welded web connection EPM-15 0.087 N/A N/A None 855 1.47 Fracture of web at stiffener weld
a
Shear rotation data acquisition was stopped prior to test end.
b
No coupon results.
c
Shear force transferred in link and slab.

severing of the web vertically along the end-plate interface.

Although the use of CJP welds prevented fracture in the welds,
the access holes reduced the link shear area to 90% of the total
web area, which likely caused higher localized stresses and initial
cracking. However, even with this reduction in the web area, the
maximum link shear obtained was 1.4 times the plastic shear
strength determined using the actual web yield stress.
The response of the web-connected links was different from that
of the end-plate-connected links. The web link connection is ini-
tially rigid because of the pretensioned bolts, until it slips and the
bolts bear onto the link web. As the link shear is increased, the bolts
bear more, and localized yielding and ovalization of the bolt holes
occurs. When the loading is reversed, the connection behaves in a
rigid way again until it reaches the connection slip resistance. The
bolts then slip again and move in the now-deformed and ovalized
bolt holes with no increase in the link shear beyond that provided
by the connection slip resistance. Once bearing of the bolts is
reactivated on the opposite side of the bolt hole, the load rapidly
increases up to the levels corresponding to the yielding of the link
web. This explains the pinching in the hysteresis of the web-
connected links, as shown in Figs. 6(c) and 6(d). Furthermore,
because of these increased inelastic connection rotations, the
web-connected links were able to achieve larger link rotations
Fig. 5. Test setup at University of Toronto than the end-plate-connected links. As shown in Fig. 6(c), link
Specimen UT-1B completed a full cycle at 0.17 rad inelastic link
The capacity of the welded end plates to sustain and transfer high rotation, more than double the maximum link rotation required by
end moments and shear was also assessed. CSA-S16.
The behavior of the end-plate-connected links closely re-
EPM Test Setup and Results
sembled that of traditional chevron EBFs. The response of the
replacement link UT-3B was identical to the original link UT-3A, Having proved that the replaceable links achieve the required
as shown in Fig. 6(a). Both specimens displayed full hysteretic ductility capacity, the findings from the six specimens tested at
curves up to and exceeding γp ¼ 0:08 rad. Specimens UT-3A UT were incorporated to enhance the design of the following seven
and 3B, which had CJP welds for the flanges and webs with an specimens tested at EPM. The intent of this second phase of testing
access-hole detail, failed by fracture of the web initiating at the was to evaluate the global performance of EBFs equipped with
weld access hole, as shown in Fig. 7(a). Initial cracks were ob- replaceable links within a frame and to assess the proposed re-
served at the corner of the web at the peak of the negative excursion placement processes in conditions similar to those that would be
of Cycle 36 (0.07 rad link rotation). These cracks propagated hori- encountered in a real building. The behavior of the EBF with a
zontally, running parallel to the flange, and they were accompanied composite concrete floor deck was also evaluated in Specimens
by pronounced inelastic web buckling. This ultimately led to the EPM-14 and 16.

1146 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / OCTOBER 2011

J. Struct. Eng., 2011, 137(10): 1141-1152

Downloaded from ascelibrary.org by University of Toronto on 03/03/20. Copyright ASCE. For personal use only; all rights reserved.

Fig. 6. Plastic link rotation versus link shear for all specimens: (a) 3A and 3B; (b) 11A and 11B; (c) 1A and 1B; (d) 2A and 2B; (e) 12; (f) 13; (g) 14;
(h) 16; (i) 15

The test setup is shown in Fig. 8. The test frame was designed to base such that the loading system subjected the link to constant
represent the first story of a five-story EBF frame designed accord- shear along its length, V link , and reverse curvature bending, where
ing to the 2005 NBCC (NRCC 2005) specified seismic loads,
representative of a high seismic area in British Columbia, Canada. V link ¼ F actuator ðh=LÞ ð5Þ
The test frame was 7.5 m wide by 4.5 m high and had e0 = 1.10 m to where h = frame height of 4.5 m; and L = frame width of 7.5 m. The
accommodate a large corridor opening measuring 2:44 × 2:74 m test was controlled by the link rotation, measured by the tangential
(8 × 9 ft). The opening would have been restricted to 0.8 m had displacement between the ends of the floor beams nearest to the
a standard EBF design been used. Simple shear beam-to-column link specimens, which were instrumented.
connections were used for the floor beams. The total lateral load Past test programs have focused on links subjected to pure shear
was applied onto the floor beam by 2 × 1;000 kN MTS actuators forces, with no or limited axial force imposed on the link. In this
reacting onto the strong wall. The actuators pushed and pulled on test program the two actuators were connected to the floor beam at
the EBF frame, imposing predetermined cyclic shear deformations a point located on one side of the link, and half of the applied
on the link. The frame columns were truly pin-connected at their actuator load was therefore transferred axially through the link.

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / OCTOBER 2011 / 1147

J. Struct. Eng., 2011, 137(10): 1141-1152

 Fig. 7. End-plate-connected link Specimens UT-3B and EPM-11A at failure (images by the authors)
Downloaded from ascelibrary.org by University of Toronto on 03/03/20. Copyright ASCE. For personal use only; all rights reserved.

Fig. 8. Test setup at École Polytechnique of Montreal (image by the authors)

This represents the worst-case axial loading scenario that can be of termination of the fillet welds connecting the stiffeners to the link
imposed on an EBF link, such as would be encountered at the ends web. Past the 0.11 rad cycle, the crack was observed to propagate
of braced column lines in buildings. For the frame aspect ratio vertically through the web, along the length of the intermediate
h=L ¼ 0:6, the axial load was equal to 83% of the link shear stiffener weld. The load dropped by 22% before the test was ended.
V link . The maximum axial tensile load imposed in the tests reached The web-connected link specimens achieved rotations greater
26% of the link axial yield strength for Specimens EPM-11A and than the code requirement, as listed in Table 2. However, the
11B. Lateral bracing was provided at the top and bottom flanges of EPM specimens were not able to achieve the large 0.17 rad link
the floor beams, at the two ends of the link. The columns were also rotations that were achieved by Specimen UT-1B. This is in part
laterally braced near the beam-to-column joints, as shown in Fig. 8. because of a relatively stiffer reinforcement web connection plate
End-plate-connected link Specimens EPM-11A and 11B had and the reduced link shear capacity due to the applied axial load.
PJP welds connecting the flanges to the end plates without an Although bolt bearing deformation did occur, the localized bolt-
access-hole detail and fillet welds to connect the link webs. Be- hole yielding and ovalization was not as pronounced as what was
cause of the large axial demand on the link flanges from both the observed in the UT tests. This is shown by the reduced pinching of
moment and the additional axial load, signs of early flange yielding the EPM link rotation hysteresis plots when compared to the UT
accompanied web yielding. The links were subjected to tension specimens, as seen in Fig. 6.
when deforming toward peak negative rotations. As Specimen The values measured for the inelastic story drifts, θp , listed in
EPM-11A was loaded toward the tension (negative) excursion of Table 2, approximately match those calculated from the measured
Cycle 39 (0.11 rad link rotation), it fractured at the top flange near inelastic link rotation, assuming a rigid plastic frame mechanism,
the end of the groove weld, completely severing the top flange, as using the following relation:
shown in Fig. 7(b). Factors that contributed to this are the large
plastic strain demands at that location, the large axial stresses de- θp ¼ γp ðe=LÞ ð6Þ
veloped, the high degree of constraint due to the presence of the end
plate and the welds, and the heat-affected zone brittleness near the where e = length of the shear link listed in Table 1. These drift
flange-to-end-plate weld. Specimen EPM-11B performed better, values compare well with the peak drifts ranging from 1.0–1.5%
exhibiting a more stable failure mechanism in the web accompa- computed for 3- and 10-story EBFs under design level (10% in
nied with more visible inelastic web buckling. Web fracture initi- 50 years) ground motions in the United States (Chao and Goel
ated at the bottom end of the link web middle stiffener, at the point 2006) and exceed those computed for three-story EBFs designed

1148 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / OCTOBER 2011

J. Struct. Eng., 2011, 137(10): 1141-1152

 according to current Canadian practice, which were found to be less This increases the slip distance and produces more connection ro-
than 0.9% under seismic motions, matching the 2% in 50 years tation, and thereby more pronounced pinching. This is also evident
hazard in western Canada (Rozon et al. 2008). when comparing the results of Specimens EPM-12 and 16. The two
specimens are identical with exception of these two differences: the
first was varying the bolt bearing stress by changing the thickness
Discussion of Test Results of the web reinforcement plates, which were equal to 6.35 and
4.76 mm for EPM-12 and 16 respectively; and the second was that
Based on the test results alone, it is difficult to assess which weld- because EPM-16 was the fourth specimen tested, the floor beam
ing procedure is more suitable for connecting the link to the end bolt-holes of the link connections were slightly elongated due to
plates because all four links achieved link rotation values greater prior bearing deformations, which in turn increased the bolt slip
than 0.08 rad. Although the specimens were designed as shear criti- range. As seen in Fig. 10 (which only illustrates the last few cycles
cal, the flexural demands on the link ends were greater than pre-
for clarity), EPM-16 was capable of sustaining a maximum value of
dicted because the shear overstrength values were larger than that
γp of 0.125 rad compared to 0.104 rad for EPM-12 because of the
specified by the code. This is most evident in Specimens EPM-11A
Downloaded from ascelibrary.org by University of Toronto on 03/03/20. Copyright ASCE. For personal use only; all rights reserved.

increased connection contribution. Measured at the 0.09 rad cycle,

and 11B, where the ratio e=ðM p =V p Þ is 1.6, at the edge of the
Γconn was 0.17 compared to 0.05 for Speciments EPM-16 and 12,
intermediate link length range where both shear and flexural behav-
respectively, showing a three-fold increase in the connection rota-
iors govern the response. Considering the reduction and saving in
tion. These observations indicate that the designer has the ability to
the fabrication time and cost involved in the preparation required
for the PJP welds when compared to CJP welds, it is suggested that choose between a more flexible connection, with thinner web
using PJP welds to connect the link flanges and fillet welds for reinforcement plates and possible oversizing of the link bolt holes,
the web would be adequate to ensure the desired shear critical thus attaining larger link rotations, versus a stiffer connection with
response. It is further recommended that when using PJP welds, lower maximum link rotations but fuller hysteretic response.
which are not designed to transfer M p, that link lengths be limited A notable feature of the web-connected links was that the
to 1:4M p =V p to reduce the flexural demand on the connections. majority of the specimens exhibited web fracture as the controlling
Nonetheless, CJP weld may be used in the connections of the failure mode. The link web fractures initiated at top and bottom
end-plate links when lengths of 1:6M p =V p are required. ends of the link web middle stiffeners (nearer to the link midspan),
As noted previously, the rotation of the web-bolted connection at the point of termination of the fillet welds connecting the stiff-
was significant in the overall link performance. Fig. 9 illustrates eners to the link web. Ultimately, the growth of these fractures led
how the connection rotation and the shear panel rotation contribute to a reduction of the link shear resistance. Specimen UT-1B pro-
to the total link rotation. Instrumentation was placed to measure the vides an example of a specimen that failed because of this type of
total and shear panel link rotations and the connection rotations. fracture. Fig. 11 shows the specimen after testing, and pronounced
The web connection contributions to the total link rotations, inelastic web buckling and fractures running diagonally across the
Γconn , listed in Table 2, were calculated at the 0.09 rad cycle as web are visible.
follows: Specimens UT-1 and 2 and EPM-12, 15, and 16 had stiffeners
spaced at 1.5 times the spacing required by both CSA-S16 and
Γconn ¼ 1
ðγshear; cycle =γcycle Þ ð7Þ AISC (2005). These specimens developed inelastic link rotations
where γcycle = total link rotation completed in one cycle; and greater than the required 0.08 rad and failed due to fracture
γshear; cycle = shear panel rotation in one cycle. The component of of the link web. These test results suggest that the shear link stiff-
the connection rotation was on average 16% of the link rotation. ener spacing requirements in CSA-S16 and AISC (2005) are
Specimen UT-2A, which did not have a web reinforcement plate, conservative. This observation was also noted by Okazaki and
experienced excessive bolt-hole ovalization that resulted in 60% of Engelhardt (2007).
the total link rotation taking place in the connection. This indicates The link overstrength values obtained for each specimen are
the direct correlation between the bolt bearing stress, governed presented in Table 2. This table lists the ratio V max =V n , where
by the plate thickness of the web connection, and the degree to
which the connection rotation contributes to the overall link rota-
tion. For thinner web connection plates, the bolt bearing stress is
greater, which results in more localized yielding in the bolt-holes.

Fig. 9. Contribution of connection rotation to total link rotation Fig. 10. Effect of web connection thickness on hysteretic response

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / OCTOBER 2011 / 1149

J. Struct. Eng., 2011, 137(10): 1141-1152

 Fig. 11. Web bolted link Specimen UT-1B at failure (images by the authors)
Downloaded from ascelibrary.org by University of Toronto on 03/03/20. Copyright ASCE. For personal use only; all rights reserved.

V max = maximum shear force measured in a given test; and V n =

plastic strength of the link. It is calculated per CSA-S16 and AISC
(2005) as the smaller of V p or 2M p =e, where V p and M p are com-
puted by using the actual measured dimensions and actual mea-
sured yield strengths of the test sections. The UT links were
subjected to pure shear forces, with no axial force imposed on
the link. The average overstrength for Specimens UT-1, 2, and 3
of 1.43 is comparable the average of 1.41 measured in links made
of A992 steel that were tested by Okazaki and Engelhardt (2007).
For the EPM web-connected specimens, EPM-12, 13, 15, and 16,
the average overstrength of the peak shear forces under tension was
1.57 and 1.34 under compression; note the asymmetry in link
shears shown in Figs. 6(b), 6(e), and 6(f). This difference can
be attributed to the influence of the axial loading on the shear buck-
ling resistance. The overall average of the EPM links was 1.45,
which is similar to the recorded value of 1.43 for the UT links.
The web-connected link peak shear forces averaged 5% higher than
those measured for the end-plate-connected links. The measured Fig. 12. Main transverse cracks on the top of the concrete floor slab
web yield strength, when compared to the nominal yield strength
of 350 MPa, was 1.2 times greater, on average, which exceeds the
assumed value of Ry ¼ 1:1 specified in the CSA-S16 and 2005
AISC design provisions.

Effect of Composite Floor on Link and Frame Behavior

Link Specimens EPM-14 and 16 were tested as part of the eccentri-
cally braced steel frame with a composite steel deck on top of the
floor beams. The 2.5-m-wide cast-in-place composite floor slab,
representative of a typical floor slab, was 76 mm thick and poured
over a Canam P-2436 76-mm-deep steel deck with trapezoidal
flutes spaced 300 mm on center running perpendicular to the beam.
The deck sheet was 0.91 mm thick (20 gauge) and made of galvan-
ized (Z275) ASTM A653 steel. The composite action was achieved
through 70 mm (2.75 in.) headed studs welded onto the top flange
of the floor beams at every flute.
Initial transverse cracks in the slab were observed at the end of
Cycle 18 (0.0075 rad link rotation) at the first line of studs on either
side of the link. New cracks appeared as the load was continued, Fig. 13. Web bolted replacement link EPM-16 at a frame drift of
and widening of the existing cracks and concrete spalling were also 1.62% and γp ¼ 0:151 rad (images by the authors)
observed. Fig. 12 shows the crack pattern at the end of Cycle 40
(0.13 rad link rotation). During testing, it was observed that the
concrete slab was pulling out from the steel deck, as shown in able to sustain link rotations with a reduced link shear up to
Fig. 13. Upon further inspection, it was evident that the slab pulled 0.148 rad. This indicates that the slab did not restrain the link from
out from its connection to the top of the welded studs, leaving a achieving large shear deformations.
cone of concrete still intact and connected to the base of the studs Before replacing the link it was necessary to repair the slab. The
and the floor beam. The shear transferred across the link Specimen floor beams were not designed considering the increase in moment
EPM-14 and concrete slab was greater than the shear recorded for capacity that is achieved due to composite action, and the replace-
the identical link Specimen EPM-12, which was tested with the able links were not connected to the floor slab. Thus the primary
bare steel frame only. As shown in Fig. 6(g), Speciment EPM-14 concern for repair was not the strength of the slab in flexure but to
displayed a full hysteretic response up to γp ¼ 0:103 rad and was ensure that the mechanism by which inertia forces were transmitted

1150 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / OCTOBER 2011

J. Struct. Eng., 2011, 137(10): 1141-1152

 from the slab to the lateral load-resisting EBF system was rehabili- level are 1.1% for short BRB frames (4–6 stories) and 0.9% for
tated. Based on recommendations in the ACI Repair Application taller BRB frames (10–12 stories).
Procedures (ACI 2005a, b), the most severely damaged concrete Two replacement procedures were investigated: bolted web con-
was removed, exposing the studs. Repair was then done using a nections and welded web connections. Replacement link Specimen
self-compacting, flowable, cement-based concrete mix, Sikacrete- EPM-16 had bolted web connections, for which all the bolt holes
08 SCC. Repair of the main transverse cracks that were present on except the central one were postdrilled to match the geometric con-
the top surface of the concrete slab was performed using a superlow figuration of the floor beam bolt holes that corresponded to the
viscosity injection resin, Sikadur 52. frame’s permanent residual story drift of 0.5% that was present
The concrete failed in the new repaired concrete and not the at the end of the EPM-14 test. The replacement link performed
substrate. From these observations, it can be concluded that the well, as shown in Fig. 13 in which the deformed eccentrically
repair methods were sufficient to retrofit the slab-to-shear stud braced frame system (EPM-16) is shown at 1.62% story drift,
connection. In the tests, it was also observed that less bending which corresponds to 0.151 rad of plastic link rotation. The second
in the slab occurred at the beam ends closer to the columns, and replacement Specimen EPM-15 was connected by welding the link
subsequently no cracking developed. It is thus recommended to web and flanges to the floor beam web, as shown in Fig. 14(a). This
Downloaded from ascelibrary.org by University of Toronto on 03/03/20. Copyright ASCE. For personal use only; all rights reserved.

place enough studs to transfer the floor shear force at the beam was investigated as an alternative to the bolted connection because
ends, so that the slab diaphragm action is guaranteed even after a no alignment of the bolt holes of the link-to-beam is required.
large seismic event, when the slab connection to the studs closer to Welding was done in the laboratory, with the steel deck kept in
the link may have been damaged. place over the floor beams, to simulate realistic constraints that
would be expected in the field with respect to the space available
Replaceability of Shear Links for the welder to work in. As shown in the hysteretic response of
Sets of two identical specimens were tested for each link type Specimen EPM-15 in Fig. 6(i), there was no pinching because of
to assess the replaceability of the links and the performance of the rigid welded detailing, which did not allow for rotations to take
the replaced link in comparison to the original link. In particular, place in the connections. No signs of yielding were observed in the
Specimens EPM-15 and 16 were designed and tested to assess the welded link-to-floor beam connections, as shown in Fig. 14(b),
performance of an EBF that is repaired by replacing the yielded indicating that the design and application of the welds were more
deformed link EPM-14 after experiencing significant seismic than adequate to transfer the connection loads. The reduced link
events. Before replacing the links, the frame was purposely held length and fixed-end connections lead to early shear yielding of
at a residual drift of 0.5%. Studies by McCormick et al. (2008) the link, and a reduced final inelastic link rotation capacity of
indicated that a 0.5% residual drift represents a limit beyond which 0.087 rad. Although both procedures proved feasible, the welded
it is more economical to rebuild a structure than it is to repair it. technique provided greater flexibility and ease to position the
Although not verified in this study and not reported in any other replacement link. Fillet welding on-site was more than adequate
study, it is expected that the residual drifts for the EBF systems will to create a full-strength fully rigid end connection.
not differ significantly from the reported values for buckling re- In the test program, replacement of end-plate-connected links
strained brace (BRB) systems with similar hysteretic responses. required realignment of the test frame to its plumb position before
Kiggins and Uang (2006) report residual drifts equal to 0.39 and the link specimens could be fitted between the floor beams. This
0.29% for three- and six-story BRB frames, respectively, subjected could prove very difficult, if not impossible, to achieve in a real
to a set of design-basis earthquake (DBE) ground motion records. building. However, sizing the link to be shorter by a few milli-
More recently, Erochko et al. (2011) developed a relation between meters and filling the gap between the link end plate and floor beam
peak and residual drifts. Results from their numerical models in- end plate with shims is possible. Tests have been conducted by
dicated that peak drifts corresponding to a 0.5% mean residual drift Sumner (2003) using beams fabricated 4.8 mm (3=16 in:) short

Fig. 14. Welded replacement link EPM-15: (a) welding procedure; (b) link at γp ¼ 0:087 rad (image by the authors)

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / OCTOBER 2011 / 1151

J. Struct. Eng., 2011, 137(10): 1141-1152

 with the gaps filled with finger shims, and no adverse consequences References
or differences in connection behavior were observed.
AISC. (2005). “Seismic provisions for structural steel buildings, including
Supplement No. 1.” ANSI/AISC 341-05, Chicago.
Conclusion American Concrete Institute (ACI). (2005a). “Crack repair by gravity feed
with resin.” ACI Repair Application Procedures Bulletin 2, Farmington
A method for designing EBFs with replaceable shear links has been Hills, MI.
developed and experimentally validated. The method allows for the American Concrete Institute (ACI). (2005b). “Spall repair of horizontal
uncoupling of the yielding element from the main floor beams to concrete surfaces.” ACI Repair Application Procedures Bulletin 7,
Farmington Hills, MI.
optimize the design of the links and the EBF. The concept also
Balut, N., and Gioncu, V. (2003). “Suggestion for an improved ‘dog-bone’
allows for wider clearance for openings or circulation while still
solution.” STESSA 2003: Proc., Conf. on Behaviour of Steel Structures
achieving a desirable link shear yielding inelastic response. Results in Seismic Areas, Balkema, Netherlands, 129–134.
of the test program summarized in this paper show that replaceable Canadian Institute of Steel Construction (CISC). (2005). “Moment connec-
links achieve and exceed the ductility acceptance criteria of tions for seismic applications.” Willowdale, Ontario, Canada.
Downloaded from ascelibrary.org by University of Toronto on 03/03/20. Copyright ASCE. For personal use only; all rights reserved.

0.08 rad inelastic link rotation prescribed by both the Canadian Canadian Standards Association (CSA). (2009). “Design of steel struc-
Standard CAN/CSA-S16-09 (CSA 2009) and 2005 AISC seismic tures.” CAN/CSA-S16, Mississauga, Ontario, Canada.
provisions (AISC 2005). Chao, S.-H., and Goel, S. C. (2006). “Performance-based seismic design of
End-plate shear replaceable links display a consistent full hys- eccentrically braced frames using target drift and yield mechanism.”
teretic response up to γp ¼ 0:104 rad. The link shear overstrength AISC Eng. J., 3rd quarter, 173–200.
is 1.40 on average. Under a combined shear and axial tensile load, Dubina, D., Stratan, A., and Dinu, F. (2008). “Dual high-strength steel
eccentrically braced frames with removable links.” Earthquake Eng.
the link peak shear forces increase by up to 12% when compared
Struct. Dyn., 37(15), 1703–1720.
with those subjected to a similar compressive axial load. Neither
Engelhardt, M. D., and Popov, E. P. (1989). “On design of eccentrically
the connection configuration nor the presence of axial load has braced frames.” Earthquake Spectra, 5(3), 495–511.
an adverse effect on the link rotation capacity. It is suggested to Erochko, J., Christopoulos, C., Tremblay, R., and Choi, H. (2011).
use PJP welds to connect the link flanges to the end plates and limit “Residual drift response of SMRFs and BRB frames in steel buildings
the link length to 1:4M p =V p . Replaceable links with web-bolted designed according to ASCE 7-05.” J. Struct. Eng., 137(5), 589–599.
connections exhibit a more pinched response, but because of Ghobarah, A., and Ramadan, T. (1994). “Bolted link-column joints in
the connection rotations, they are able to sustain larger inelastic eccentrically braced frames.” Eng. Struct., 16(1), 33–41.
link rotations, on average γp ¼ 0:119 rad and a maximum of Kasai, K., and Popov, E. P. (1986). “Cyclic web buckling control for shear
γp ¼ 0:166 rad. The average overstrength of the peak shear forces link beams.” J. Struct. Eng., 112(3), 505–523.
is 1.57 under axial tension, 1.34 under compression, and 1.47 under Kiggins, S., and Uang, C.-M. (2006). “Reducing residual drift of buckling-
restrained braced frames as a dual system.” Eng. Struct., 28(11),
pure shear. The connection rotation is achieved by allowing for
1525–1532.
localized bearing deformations around the bolt holes in the web
Kim, K., Parra-Montesinos, G., Goel, S. C., and Nuttal, B. (2003). “Behav-
connection. However, this should be well controlled to limit the ior of double channel built-up members subjected to reversed cyclic
pinching in the link hysteretic response and to avoid failure at bending.” Proc., Annual Technical Session and Meeting, Structural
the connection. Web reinforcement plates, welded all around at the Stability Research Council, Baltimore, 659–669.
connection, should be used to control the bolt bearing stresses. Top Kulak, G. L., Fisher, J. W., and Struik, J. H. A. (1987). Guide to design
and bottom flange reinforcement plates should be added to ensure a criteria for bolted and riveted joints, Wiley, New York.
shear yielding dominant behavior. Mansour, N., Christopoulos, C., and Tremblay, R. (2006). “Seismic design
The shear transferred across the link and concrete slab is greater of EBF steel frames using replaceable nonlinear links.” Proc., STESSA,
by 14% than the shear resisted by the bare steel link specimen. The Taylor & Francis, London.
presence of the composite concrete slab does not restrain the link McCormick, D., Aburano, H., Ikenaga, M., and Nakashima, M. (2008).
“Permissible residual deformation level for building structures consid-
from achieving large shear deformations. The on-site replaceability
ering both safety and human elements.” Proc., 14th World Conf. on
of the link sections is confirmed even in the presence of residual Earthquake Engineering, Beijing.
deformations of 0.5% drift with bolted and welded web connec- McDaniel, C., Uang, C. M., and Seible, F. (2003). “Cyclic testing of built-
tions. The welded technique provides greater flexibility and ease up steel shear links for the new bay bridge.” J. Struct. Eng., 129(6),
to position the replacement link. The web-connected C-section 801–809.
links are easier to handle and erect. National Research Council of Canada (NRCC). (2005). National building
code of Canada, 12th Ed., Ottawa, Canada.
Okazaki, T., Arce, G., Ryu, H.-C., and Engelhardt, M. D. (2005). “Exper-
Acknowledgments imental study of local buckling, overstrength, and fracture of links in
eccentrically braced frames.” J. Struct. Eng., 131(10), 1526–1535.
Financial support for this study was provided by the National Okazaki, T., and Engelhardt, M. D. (2007). “Cyclic loading behavior of
Sciences and Engineering Research Council of Canada (NSERC) EBF links constructed of ASTM A992 steel.” J. Constr. Steel Res.,
under the Collaborative Research and Development Grant Pro- 63(6), 751–765.
gram, the Steel Structures Education Foundation (SSEF), and Popov, E. P., and Engelhardt, M. D. (1988). “Seismic eccentrically braced
the Canadian Institute of Steel Construction (CISC). Test speci- frames.” J. Constr. Steel Res., 10, 321–354.
mens were fabricated and donated by Walters Inc., Hamilton, Rozon, J., Koboevic, S., and Tremblay, R. (2008). “Study of global behav-
iour of eccentrically braced frames in response to seismic loads.” Proc.,
Ontario, ADF Group Inc., Terrebonne, Quebec, and Constructions
14th World Conf. on Earthquake Engineering, Beijing.
Beauce-Atlas, Quebec, Canada. This in-kind support is kindly ap-
Stratan, A., and Dubina, D. (2004). “Bolted links for eccentrically braced
preciated and gratefully acknowledged. The authors also express steel frames.” Proc., 5th Int. Workshop on Connections in Steel Struc-
their appreciation to the technical staff of the Structural Testing tures V, Delft Univ. of Technology, Netherlands, 223–232.
Facility at the University of Toronto and the Hydro-Quebec Sumner (2003). “Unified design of extended end-plate moment connec-
Structures Laboratories at École Polytechnique in Montreal for tions subject to cyclic loading.” Ph.D. thesis, Virginia Tech, Blacksburg,
their invaluable assistance in the test program. VA.

1152 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / OCTOBER 2011

View publication stats J. Struct. Eng., 2011, 137(10): 1141-1152