Download as pdf or txt
Download as pdf or txt
You are on page 1of 13

Development of Replaceable Cast Steel Links for

Eccentrically Braced Frames


Kyla G. Tan 1 and Constantin Christopoulos, Ph.D., P.Eng., M.ASCE 2

Abstract: Replaceable links for eccentrically braced steel frames (EBFs) decouple the yielding links from the floor beams to facilitate rapid
repair or replacement after a seismic event. Replaceable links that are composed of rolled or built-up steel sections are, however, susceptible to
the same failure modes as links in conventional EBFs. These failure modes include fracture in the base metal or welds, local buckling, and
lateral-torsional buckling, all of which limit the ductility of the link and consequently, the overall ductility of the frame. An extensive numeri-
Downloaded from ascelibrary.org by Kyla Tan on 11/08/16. Copyright ASCE. For personal use only; all rights reserved.

cal analysis study was completed to develop replaceable cast steel link concepts, with practical connection details, that combine the ad-
vantages of replaceable links with the geometric freedom and material properties inherent to castings. Results indicate that the unique
geometry of the proposed concepts achieves a larger rotation capacity and longer low-cycle fatigue life by evenly distributing flexural yielding
over the entire length of the link and minimizing stress concentrations. The replaceable cast steel links have the potential to significantly
improve the performance, reliability, robustness, and resilience of EBFs. DOI: 10.1061/(ASCE)ST.1943-541X.0001550. © 2016 American
Society of Civil Engineers.
Author keywords: Seismic effects.

Introduction increased (Prinz and Richards 2009; Berman et al. 2010; Bruneau
et al. 2011).
Eccentrically braced frames (EBFs) are widely used as lateral force Mansour (2010) proposed a method for designing EBFs with
resisting systems because they exhibit high stiffness through their replaceable shear links and found that more efficient and economi-
diagonal braces and high ductility through the yielding of the ec- cal EBF designs can be achieved by decoupling the link from the
centrically loaded links. The link must be designed as the weakest beam, which also expedites the design process. Decoupling the link
component of the frame in order to act as the ductile fuse in the provides the added benefits of its rapid inspection and replacement
structural system and dissipate energy. The maximum inelastic link following a seismic event, allowing the structure to return to service
rotation capacity, γ pc , is limited by various possible failure modes. with reduced delays. Since replaceable shear links are composed of
Although delayed by stiffeners, shear links exhibit web buckling rolled or built-up steel sections, they are however susceptible to the
which can lead to tension fields forming in the webs between same failure modes as links of conventional eccentrically braced
the stiffeners. Stress concentrations develop because of these ten- frames and thus still only achieve a γ pc of 0.08 rad. The majority
sion fields, stiffener welds, and the variable material properties at of the replaceable shear links tested by Mansour (2010) failed by
the k-line of the link and eventually lead to web fracture (Kasai web fracture initiating at stiffener welds.
and Popov 1986; Okazaki et al. 2005; Kavinde et al. 2014). The The use of steel castings as a method of providing improved
design provisions for steel seismic force-resisting systems in Can- seismic performance in various steel lateral force-resisting systems
ada (CSA 2009) and the United States (AISC 2010) expect γ pc to has recently gained increasing interest. Because of the material and
be at least 0.08 rad for properly designed shear links. For flexural geometric flexibility that castings offer, stress concentrations can be
links, because of the occurrence of flange buckling or lateral- minimized and the element can be shaped for an optimal flow of
torsional buckling, as well as flange fracture as a result of the forces. Examples include the cast steel high-strength connector (de
high flexural strains at the link ends (Engelhardt and Popov Oliveira et al. 2008); the yielding brace system (Gray 2012) for
1992), γ pc is only 0.02 rad. Several improved link designs have
concentrically braced frames; and yielding connectors for mo-
been proposed in an effort to increase the link rotation capacity by
ment-resisting frames, including the panel zone dissipator modular
mitigating the various failure modes. However, each of these
node (Fleischman et al. 2007) and a modular connector (Sumer et al.
designs only addresses one of the many possible failure modes.
2007). The controlled environment in which castings are produced
Consequently, the value of γ pc has not been significantly
can prevent link failure as a result of fabrication errors as found
in the forensic analysis of links composed of rolled sections by
1
Structural Designer, Read Jones Christofferson Ltd., 144 Front St. Kavinde et al. (2014).
West, Suite 500, Toronto, ON, Canada M5J 2L7 (corresponding author). This paper presents the development of replaceable cast steel
E-mail: ktan@rjc.ca links for EBFs that integrate the advantages of replaceable yielding
2
Professor and Canada Research Chair in Seismic Resilience of links with the benefits of steel castings to achieve greater ductility
Infrastructure, Dept. of Civil Engineering, Univ. of Toronto, 35 St. George and a longer low-cycle fatigue life. Four replaceable cast link con-
St., Toronto, ON, Canada M5S 1A4.
cepts were developed and evaluated using finite-element analysis
Note. This manuscript was submitted on August 12, 2015; approved on
February 23, 2016; published online on May 3, 2016. Discussion period
results in this study. One concept is selected and further optimized,
open until October 3, 2016; separate discussions must be submitted for in- and the effect of the selected replaceable cast link on the global
dividual papers. This paper is part of the Journal of Structural Engineer- performance of a five-story EBF structure is further investigated
ing, © ASCE, ISSN 0733-9445. through nonlinear time-history analyses.

© ASCE 04016079-1 J. Struct. Eng.

J. Struct. Eng., 2016, 142(10): 04016079


Finite-Element Modeling and Evaluation Criteria
1500
In this study, specimen EPM-11B from the full-scale experimental
program conducted by Mansour (2010) was selected to represent
1000
the performance of replaceable shear links made of rolled or built-

Link Shear Force (kN)


up sections and to provide a benchmark for assessing the perfor-
mance of the proposed cast steel links. Details on the selection of 500
this benchmark test can be found in (Tan 2014).
0

Finite-Element Modeling
−500
The replaceable link benchmark (RLB) was modeled using the
finite-element analysis program ABAQUS/CAE 6.10-1 and its re- −1000
sults were used to evaluate the performance of the replaceable cast
link concepts. The material model which included inelastic material
−1500
properties used by Mansour (2010) was applied to the RLB finite- −0.1 −0.05 0 0.05 0.1
Downloaded from ascelibrary.org by Kyla Tan on 11/08/16. Copyright ASCE. For personal use only; all rights reserved.

element model. G40.21-350W grade steel with a nominal yield Total Link Rotation (rad)
strength of 350 MPa, a Young’s modulus of 200,000 MPa, and EPM−11B Experimental Data (Mansour 2010)
a Poisson’s ratio of 0.3 were considered. The inelastic material RLB Finite Element Model Results
properties were defined by a nonlinear, kinematic hardening model
with the following parameters: 8,000 MPa for the initial kinematic Fig. 1. Comparison of experimental and analytical hystereses
hardening modulus, C, and 75 for the rate factor for kinematic hard-
ening, γ. The isotropic hardening parameters were 110 MPa for the
maximum change in yield stress, Q∞ , and 4 for the parameter b.
overstrength at large deformations that are observed in the exper-
Ten-node tetrahedral elements with second-order shape func-
imental results (e.g., Della Corte et al. 2013).
tions, referred to as C3D10M, were used to model the RLB.
Automatic meshing with a global seed size of 20 and automatic
incrementation with settings of 0.01, 0.001, and 0.1 initial, mini- Evaluation Criteria
mum, and maximum increment sizes respectively were selected and To be directly comparable, the replaceable cast steel link concepts
applied to the final RLB model. Details about the sensitivity analy- were designed to have a similar yield strength and length as the
ses that were carried out while developing this model can be found RLB: 579 kN of nominal shear resistance and 800 mm respectively.
in Tan (2014). The ABAQUS large-displacement effects option was The concepts were intended to exhibit stiffness values that
enabled in the analysis because the web yielding at low rotations were similar to that of the RLB, which was determined to be
provides sufficient imperfections in the model to cause buckling at 175,640 kN=rad (shear force per radian of total link rotation),
larger rotations without explicitly introducing imperfections at the because the lateral stiffness of an EBF is dependent, to a certain
start of the analysis (Prinz and Richards 2009). extent, on the stiffness of the individual links. In addition, an over-
The left end-plate of the RLB was fully restrained against all strength equal to or less than that of the RLB was desired to min-
translations and rotations while transverse displacements were ap- imize the capacity design forces for the adjacent beams, columns,
plied to the right end plate to simulate link rotation. The right end and braces. In this study, the overstrength was considered as the
plate was restrained against all rotations but was permitted to trans- ratio of the maximum link shear obtained when applying a total
late horizontally along the undeformed link’s longitudinal axis and link rotation of 0.16 rad to the nominal strength of the link and
all the nodes of the right end plate were constrained to have the was determined to be 1.54 for the RLB.
same horizontal translation. Gray (2012) conducted low-cycle fatigue tests and found that
To validate the finite-element model parameters, the results of the ASTM A352 LCB cast steel has slightly lower, but comparable,
the numerical analysis carried out on the RLB were compared to ductility to rolled A36 grade steel. To reduce computation time, the
the experimental results reported by Mansour (2010) for specimen low-cycle fatigue life of the concepts that are studied in this paper
EPM-11B, which was subjected to the full AISC 341-10 (AISC were assessed by comparing the maximum equivalent plastic strain
2010) loading protocol. As a result of the full-scale frame setup values following a monotonic excursion to a total link rotation of
used to test EPM-11B, an axial load equal to half the actuator force 0.16 rad to the peak strains obtained in the RLB model. Fracture in
was applied through the link. A modified loading protocol, which the models was assumed to have occurred if strains exceeded the
consisted of one full cycle of each link rotation angle of the AISC ultimate strain of 0.22 which was measured by Mansour (2010)
341-10 loading protocol starting with 0.0075 rad, was implemented through monotonic coupon tests. The RLB finite-element model
in the analysis of the RLB to reduce the computation time. exhibited a maximum equivalent plastic strain of 0.233 which oc-
The axial loading was also simplified such that the applied axial curred at a corner formed by a stiffener and the link web (Fig. 2),
load linearly increased to that of the maximum recorded in the which is consistent with the web fracture observed in the specimen
EPM-11B experimental data for the corresponding magnitude of by Mansour (2010) during physical testing. The occurrence of local
link rotation. The results of the finite-element analysis are com- buckling and lateral-torsional buckling in the proposed link con-
pared to the experimental results in Fig. 1 and show agreement that cepts was identified by visual inspection of the deformed shape
is deemed satisfactory to validate the use of the RLB model as a of the models and sudden declines of strength in their link-shear
basis for comparison for the cast steel concepts that were devel- rotation curves.
oped. Differences between the curves can be attributed to simpli- The proposed cast steel link concepts were required to have a
fication of the axial load and boundary conditions applied to practical design for general castability since not all configurations
the RLB. The axial restraints provided by the braces in the full or shapes are castable or will result in high quality castings (Gray
frame setup in the EPM-11B tests could explain the higher shear 2012). In addition, the selected concept was expected to possess a

© ASCE 04016079-2 J. Struct. Eng.

J. Struct. Eng., 2016, 142(10): 04016079


Maximum equivalent plastic strain = 0.233
1200
Equivalent
Plastic Strain 1000

Link Shear Force (kN)


(Avg: 75%)
0.235 800
0.220
0.202
0.183 600
0.165
0.147
0.128 400
0.110
0.092
0.073 200
0.055
0.037 0
0.018 0 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16
0.000 Total Link Rotation (rad)

Fig. 2. RLB deformed shape and equivalent plastic strain contour at a RLB CWC−1 YRC SFYC−W
RLB−Fillets CWC−2 SFYC−1 SFYC−T
Downloaded from ascelibrary.org by Kyla Tan on 11/08/16. Copyright ASCE. For personal use only; all rights reserved.

total link rotation of 0.16 rad

Fig. 4. Comparison of link shear-rotation curves for the link concepts


practical connection for rapid installation and repair. The link di-
mensions also had to be within reasonable limits and could not ex-
ceed those of the adjacent beams they would be connected to. An (4) simultaneous flexural yielding (Fig. 3). A finite-element model
end-plate connection with the same dimensions used by the RLB was created for each cast steel link concept using the same param-
was assumed for the initial development of the concepts. Finally, eters used to model the RLB and were loaded to a total link rotation
the replaceable cast link concepts were expected to have the ability of 0.16 rad. Fig. 4 compares the resulting link shear-rotation curves
to be designed for other strengths and lengths to form a complete to that of the RLB while Table 1 compares the performance criteria
set of cast links that can be used in the design of entire EBF struc- values. The following section summarizes the main findings from
tures. A predefined set of links of various lengths and capacities the analytical assessment of each proposed concept.
would allow for prequalification and mass production, which
would reduce their cost and enhance their reliability by preventing Smooth Transitions Concept (RLB-Fillets)
fabrication errors that could otherwise occur when rolled or built-
The RLB-fillets concept incorporated 30-mm fillets into the geom-
up sections are used (Kavinde et al. 2014).
etry of the RLB to eliminate the stress concentrations caused by
abrupt geometric discontinuities at the corners formed by inter-
Replaceable Cast Steel Link Concepts mediate stiffeners, k-line, and end plates. However, the addition
of fillets resulted in higher elastic stiffness, overstrength, and strain
Initially, four general concepts were explored and analyzed: values (Table 1). The higher maximum equivalent plastic strain
(1) smooth transitions, (2) corrugated web, (3) yielding ring, and value of 0.268 indicates that web fracture would have likely

Equivalent
Plastic Strain
(Avg: 75%)
0.340
0.220 (a) (b)
0.202
0.183
0.165
0.147
0.128
0.110
0.092
0.073
0.055
0.037
0.018
0.000

(c) (d)

Fig. 3. Deformed shapes of the replaceable cast link concepts and equivalent plastic strain contours at a total link rotation of 0.16 rad: (a) RLB-fillets;
(b) CWC-1; (c) YRC; (d) SFYC

© ASCE 04016079-3 J. Struct. Eng.

J. Struct. Eng., 2016, 142(10): 04016079


Table 1. Comparison of Performance Values for the Link Concepts and end-plate surfaces. Limited plastic strains were expected to
Elastic Maximum form in the hinges as each ring was supposed to contribute to
Yield stiffness equivalent the overall link rotation. However, the analysis results showed
Concept strength (kN) (kN=rad) Overstrength plastic straina that the high link end moments caused plastic hinges to form,
RLB 579 175,640 1.54 0.233
mainly in the first and last rings, with minimal yielding of the center
RLB-fillets 579 207,690 1.78 0.268 four rings [Fig. 3(c)]. The force required to create this undesired
CWC-1 579 140,928 1.37 0.340 yielding mechanism was about half the expected strength.
CWC-2 579 121,564 1.64 0.228 Because of the behavior and results of the model (Table 1), the
YRC 280 18,564 1.87 0.233 YRC was not further developed. The concept would require signifi-
SFYC-1 579 97,156 1.95 0.116 cant design improvements to be utilized in an EBF such as stiff-
SFYC-W 579 108,336 1.60 0.113 ening the rings at the ends of the link. A single, very large ring
SFYC-T 579 109,137 1.59 0.126 could be used to span the required link length. However, the dimen-
a
Measured at 0.16 rad of total link rotation. sions needed to achieve this would likely be so large that it would
affect the floor-to-ceiling height of typical steel frame buildings.
The axial strength of the YRC was also a concern, as an applied
axial load would likely fracture the crest of each yielding ring.
Downloaded from ascelibrary.org by Kyla Tan on 11/08/16. Copyright ASCE. For personal use only; all rights reserved.

occurred at a lower link rotation than for the RLB and a comparison
of Figs. 2 and 3(a) shows that these higher strain values were ob- Although the focus of this study was on links for frames of the
served over larger regions of the link. The fillets reduced the web eccentric chevron brace configuration that do not introduce an axial
panel areas and resulted in a larger deformation demand on each load through the link for typical symmetric building layouts, a min-
element within the web panels, thus increasing the plastic strains. imum level of axial resistance would be required to allow for
The RLB-fillets concept was not further pursued because of the more practical implementations in a wider range of braced frame
poor performance that was observed. configurations.

Simultaneous Flexural Yielding Link Concept


Corrugated Web Concept
The constant cross-sectional properties of a shear link match the
Corrugated webs were implemented in the corrugated web concept constant shear force applied over its length. The shear capacity
(CWC) for self-stabilization, to mitigate web buckling in lieu of of the link is attained simultaneously over its length such that each
stiffeners, and to consequently eliminate the stress concentrations incremental segment is only required to deform a small amount to
caused by welded stiffeners. Two models with different corrugation produce the total link rotation. The proposed concept for the re-
dimensions, CWC-1 and CWC-2, were analyzed. Local web buck- placeable cast steel link employs varying cross-sectional properties
ling can be observed in the deformed shape of CWC-1 [Fig. 3(b)] to achieve simultaneous flexural yielding over its length. The added
and in the negative stiffness of its shear-rotation curve at approx- damping and stiffness device, triangular added damping added
imately 0.1 rad (Fig. 4). The results for CWC-2 show that with stiffness device (Christopoulos and Filiatrault 2006), and the yield-
smaller web panel dimensions, local web buckling can be success- ing brace system (Gray et al. 2014) are examples of experimentally
fully avoided within a total link rotation of 0.16 rad. validated structural fuses that apply this concept of spreading flexu-
Researchers have developed equations to predict the shear ral yielding to maximize both energy dissipation and low-cycle
strength of webs with trapezoidal corrugation for bridge girders fatigue life.
which must resist high shear loads like EBF links (Elgaaly et al. Originally, the simultaneous flexural yielding link concept
1996; Sause and Braxtan 2011). With appropriate dimensioning (SFYC) was imagined with multiple plates with rectangular cross
of the corrugation, buckling can be delayed as observed in sections and a triangular plan such that it would bend in constant
CWC-2. However, the CWC was not further developed because double curvature [Fig. 5(a)], where the width, bðxÞ, of the plates is
of issues related to its castability. Through interaction with steel governed by
foundries it was a determined that a minimum wall thickness of
approximately 20 mm was desired to ensure the quality of the cast- 2b x e e
ing for an application where the steel casting is intended to develop bðxÞ ¼ o ; for − ≤ x ≤ ð1Þ
e 2 2
a ductile response. With a web thickness equal to this minimum
castable thickness, the target shear strength, V p , of 579 kN would where bo = width of the plates at the link ends; e = link length; and
only be possible if a very small link depth was used. This small link x = distance from the link center. The nominal yield strength, V n , of
depth would require very thick and wide flanges to maintain the the link can be expressed in terms of the dimensions of the plate at
flexural resistance required to ensure that the link remains a the ends of the link
shear-yielding link. Although the CWC was relatively successful
at mitigating web instability, it would not have resulted in an effi- nbo d2 Fy
Vn ¼ ð2Þ
cient cross section. 3e
where n = number of yielding elements in the link; d = thickness of
Yielding Ring Concept each plate; and Fy = yield strength of the material. The elastic stiff-
ness of the link in terms of relative link end displacement, kr , can be
Alternative modes of yielding were investigated because of the
determined by
castability issues related to achieving shear yielding. The yielding
ring concept (YRC) applied the concept of moving flexural hinges 2nbo d3 E
of the U-shaped steel strip (Priestley et al. 1999). The YRC model kr ¼ ð3Þ
3e2
implemented six rings to achieve the desired link length. The finite-
element model utilized surface-to-surface contact properties, in- where E = Young’s modulus. Therefore, the thickness of each
cluding frictionless tangential behavior and hard normal contact yielding plate and the length of the link have a significant
behavior for the interaction between the semicircular segments effect on both strength and elastic stiffness. Although it would

© ASCE 04016079-4 J. Struct. Eng.

J. Struct. Eng., 2016, 142(10): 04016079


V

(a) Isometric View

e
Yielding Yielding
region xs region
(a)
b(x)

bo
Downloaded from ascelibrary.org by Kyla Tan on 11/08/16. Copyright ASCE. For personal use only; all rights reserved.

b min
Plan View
(b)
M i (x) Provided flexural
Applied bending
resistance
moment My

e xs
2 2
x
xs e (b)
2 2
Fig. 6. Trimetric views of the (a) SFYC-W; (b) SFYC-T
My
(c)

Fig. 5. SFYC link: (a) initial geometry; (b) modified width; (c) pro-
vided flexural resistance of the cross section as full plastification of the section gradually
occurs results in the high overstrength that was observed in
SFYC-1 (Table 1 and Fig. 4). Overstrength is characterized by
the shape factor, which must therefore be minimized to achieve
be advantageous to have multiple yielding elements for redun- the desired overstrength performance. The typical North American
dancy, a single yielding element was considered in the initial wide-flange section achieves a shape factor that is very close to
SFYC-1 model, in order to achieve the same strength and compa- unity, while that of a rectangular, solid section used for SFYC-1
rable stiffness as the RLB. is 1.5 (Bruneau et al. 2011).
Since the link must carry a constant shear force over its length, Fig. 6 shows two additional link concepts with different cross
the width cannot be zero at the midspan as given by Eq. (1) and sections, a wide-flange section (SFYC-W) and a rectangular, tubu-
shown in the dashed lines in Fig. 5(b). A minimum allowable lar section (SFYC-T), that were developed to achieve simultaneous
width, bmin , must be provided such that the maximum anticipated flexural yielding along the length of the link. To obtain the geom-
shear force can be resisted etry of these new concepts, Eq. (1) was revised to
ωbo d
bmin ¼ ð4Þ 2jxj½bo d3 − ðbo − 2wÞðd − 2tÞ3  þ 2ewðd − 2tÞ3
1.65e bðxÞ ¼ ;
eðd3 − d þ 2tÞ
where ω = expected overstrength factor. The lengths of the yielding e e
regions are reduced by modifying the width of the link to bmin over for − ≤ x ≤ ð5Þ
2 2
the length xs [Fig. 5(b)], which results in an increased flexural re-
sistance in this region [Fig. 5(c)].
which was obtained by modifying the elastic section modulus, S, in
The SFYC-1 model applied a bmin dimension obtained by con-
its derivation. Similarly, Eqs. (2) and (3) become
servatively assuming that ω ¼ 2.5. Fillets were added to smoothly
transition between the end plate and yielding element, as well as to
transition between the constant and tapered width regions to pre- nFy
Vn ¼ ½b d3 − ðbo − 2wÞðd − 2tÞ3  ð6Þ
vent stress concentrations. Without optimizing the dimensions of 3ed o
SFYC-1, a maximum equivalent plastic strain of 0.116 was ob-
tained, a large improvement from the RLB (Table 1). The yielding
was successfully spread to the regions of varying width [Fig. 3(d)].  −1
Fy V e
kr ¼ eV n ð3e2 − x2s Þ þ n ð7Þ
Applying the SFYC to Other Cross Sections 6dE 2dwG
If plasticity is considered at the cross-sectional level, assuming that
plane sections remain plane in pure bending, the additional flexural respectively where w = web thickness; t = flange thickness; G =
strength available beyond the initial yielding of the extreme fibers shear modulus; and xs is given by

© ASCE 04016079-5 J. Struct. Eng.

J. Struct. Eng., 2016, 142(10): 04016079


the additional benefit of its inherent torsional resistance and stabil-
ity. Although the midspan region defined by the length xs consists
of a slender, rectangular, solid section, lateral-torsional buckling of
this segment is not a concern because the applied bending moment
within this region is low.

Variations of the Tubular SFYC


Any combination of dimensions can be varied to achieve the effect
of simultaneous flexural yielding over the link length. Two alter-
natives on the variation of the sectional dimensions were examined:
(1) maintaining a constant width while varying the depth, and
(2) varying both the depth and the width equally such that the cross
(a) section is always a square, tubular cross section. For these varia-
tions, the minimum depth required to resist the shear was found
to adversely increase the length of the region xs . The length of
Downloaded from ascelibrary.org by Kyla Tan on 11/08/16. Copyright ASCE. For personal use only; all rights reserved.

the region xs can be reduced by gradually changing the cross


section from tubular to solid at the midspan of the link (Fig. 7),
which can be readily achieved with the geometric freedom pro-
vided by castings. However, the variation in web depth provides
excess shear resistance at the link ends and within the length xs
(Fig. 8). The constant depth and varying width was deemed pref-
erable because it provides a constant shear resistance more effi-
ciently over the length of the link to match the applied shear force
diagram.

(b)
Connection Concepts
Fig. 7. Trimetric views of SFYC rectangular, tubular links with (a) con-
stant depth and variable width; (b) varying depth and constant width or A web-bolted connection was initially investigated for the proposed
equally varying depth and width replaceable cast steel link because Mansour (2010) found that for
this configuration, the bolts are more accessible than in a bolted
end-plate connection, making the installation and replacement of
links more practical. Such a connection requires the cast steel link
2ewd3 geometry to be cut in half longitudinally with each half of the
xs ¼ ð8Þ
bo d − ðbo − 2wÞðd − 2tÞ3
3
link placed on either side of the adjacent beams (Fig. 9). However,
finite-element analyses showed that each half of the link acted
The webs of the yielding elements must be thick enough to pre- independently as channel sections and exhibited lateral-torsional
vent the webs from yielding or buckling in shear before the element buckling despite bolting the link halves together within the elastic
yields in flexure. To avoid a hybrid shear-flexure yielding response, region xs . Multiple solutions were studied (Tan 2014) to improve
the webs must also resist the maximum anticipated shear force of stability of the link halves, but it was concluded that they were im-
the link, or the shear corresponding to the maximum overstrength practical to achieve the desired performance and most attempted
moment in the flexural plastic hinges solutions introduced stress concentrations, which increased plastic
ωV n strains.
w≥ ð9Þ A bolt end-plate connection requires significantly less material
1.1dFy
than the web-bolted connection shown in Fig. 9. To cast the hollow
features in the proposed SFYC-T geometry, core molds would
Therefore, bmin ¼ 2w.
SFYC-W and SFYC-T (Table 1 and Fig. 4) performed almost have to be removed through openings in the end plates (Fig. 10).
identically to one another in every aspect. SFYC-W and SFYC-T The increased link flexibility attributable to the reduced volume of
exhibited a slightly higher stiffness than SFYC-1 because of the material in the end plates was examined through the analysis of
increased yielding element depth, which was made possible multiple finite-element models with various opening sizes. The
because less material was present near the neutral axis of the cross connection is essentially rigid if a chamfer is added between the
section compared to the rectangular solid section of the SFYC-1. flanges and the opening (Fig. 10). Alternatively, the opening can
By utilizing a cross section with a lower shape factor, the be closed by welding a steel plate around the perimeter of the open-
overstrength of links applying the SFYC is very similar to that ing as suggested by Tan (2014).
of the RLB or conventional links. In contrast, there was no signifi- A link with a bolted end-plate connection can be dropped easily
cant effect of the type of cross section on the maximum plastic into position on the construction site. Shims can be used to fill gaps
strains. between the end plates of the link and adjacent beams to account for
Links with built-up box sections have been used in situations casting and connection tolerances. The dimensions of the end plates
where the possibility of lateral-torsional buckling is of concern can be cast to accommodate a large range of adjacent beam depths
but lateral bracing is difficult or not feasible to provide like in and can be trimmed accordingly. The bolted end-plate connection
bridge piers (Bruneau et al. 2011). Since the performance of was deemed as the most appropriate connection for the replaceable
SFYC-W and SFYC-T was essentially identical, the rectangular, cast steel link concept and was retained for the more detailed study
tubular cross section was selected for further development given that is presented in the following section.

© ASCE 04016079-6 J. Struct. Eng.

J. Struct. Eng., 2016, 142(10): 04016079


e e

Yielding Yielding Yielding Yielding


region xs region region xs region

Cast Link Cast Link


Front View Front View

Vi (x) Vi (x)
Provided shear resistance
Vy
Applied shear Vy
Vy Vy
x x
Downloaded from ascelibrary.org by Kyla Tan on 11/08/16. Copyright ASCE. For personal use only; all rights reserved.

e 0 e e xs 0 xs e
2 2 2 2 2 2
(a) (b)

Fig. 8. Shear resistance of SFYC rectangular, tubular links with (a) constant depth and variable width; (b) varying depth and constant width or equally
varying depth and width

Connection SFYC Link Connection


tabs geometry tabs A

Trimetric View
(a) Opening

Link halves placed on either Inner chamfer to lip


side of adjacent beam webs dh

Section A-A
Shim plate
Fig. 10. SFYC-T with bolted end-plate connection
(b)

Fig. 9. SFYC-T with a web-bolted connection: (a) elevation view in an


eccentric chevron braced frame; (b) trimetric view of assembly imposing a constant axial force with a magnitude of 955.5 kN to
both the SFYC-T and RLB finite-element models, a conservative
30 and 39% of each link’s axial yield capacity respectively.
A modified cyclic loading protocol was applied to reduce the com-
Sensitivity Analysis of the Rectangular, Hollow putational time. The loading protocol consisted of one full cycle of
SFYC 0.01 rad, followed by one full cycle of 0.04 rad, and then a half
cycle to 0.09 rad. Three analyses were conducted for each model:
no axial load (0), a constant axial tensile load (T), and a constant
Effect of Axial Load axial compressive load (C). The results of these analyses confirmed
A concern for the cast link is the small cross-sectional area over the that an applied axial load has a greater effect on the equivalent plas-
region xs . The minimum gross cross-sectional area of the SFYC-T tic strains (Table 2) and link shears at large deformations (Fig. 11)
model is only 7,020 mm2 in comparison to 9,100 mm2 for the of the RLB than the SFYC-T. Even though the applied axial loads
RLB. In the full-scale EBF test setup, Mansour (2010) measured have a compounding effect on the flexure-induced compression and
a maximum axial link load that was 26% of the link’s axial yield tension in the yielding flanges of the SFYC-T, the plastic strains
capacity. The effect of an axial load on the cast link was studied by in the SFYC-T were lower than the RLB without any axial loads

© ASCE 04016079-7 J. Struct. Eng.

J. Struct. Eng., 2016, 142(10): 04016079


Table 2. Effect of Axial Load and Initial Imperfections on the Maximum those in Fig. 11 of the perfect SFYC-T model. However, the initial
Equivalent Plastic Strain of the RLB and. SFYC-T Models Measured at imperfections resulted in some stress concentrations and increased
0.09 rad plastic strains (Table 2). Set L resulted in generally higher strains
SFYC-T than set M, indicating that more stringent casting tolerances must
RLB No
be considered for the tapered yielding regions of the link while the
Axial load (no imperfection) imperfection Set L Set M elastic midspan region can be cast with more lenient tolerances.
With significant imperfections, the SFYC-T models had lower plas-
0 0.3120 0.1589 0.2755 0.1619 tic strains than a perfect RLB model, and no local buckling or any
Tension 0.3001 0.1728 0.2004 0.1746
other forms of instability were observed. Therefore, the peak strains
Compression 0.3249 0.1656 0.3104 0.2378
shown in Table 2 indicate that even with any unintended axial loads
or geometric casting defects, the proposed cast link is expected to
have a longer low-cycle fatigue life and a larger maximum ductility
capacity than replaceable links made of rolled or built-up sections.
1,000

Replaceable Cast Steel Link Optimization


Downloaded from ascelibrary.org by Kyla Tan on 11/08/16. Copyright ASCE. For personal use only; all rights reserved.

Link Shear Force (kN)

500

Effect of Depth and Width of Yielding Elements


0 There are an infinite number of combinations of the dimensions d
and bo that can produce the same link strength, assuming that all
the other parameters are kept constant. The functional combina-
−500 tions of d and bo can be selected by defining practical limits
and understanding the effect of these dimensions on various link
properties. As d increases, bo must decrease to maintain the same
−1,000 strength, resulting in a more slender cross section at the ends of the
−0.06 −0.04 −0.02 0 0.02 0.04 0.06 0.08 0.1 link. The aspect ratio d=bo should be minimized in order to
Total Link Rotation (rad) maximize the resistance to lateral-torsional buckling and is recom-
mended to be no greater than unity. However, a very small value of
RLB_0 SFYC−T_T SFYC−T_C
d=bo results in a very wide link, which may be impractical in trans-
SFYC−T_0 RLB_T RLB_C
ferring forces between the link and the adjacent beams. The elastic
stiffnesses of the SFYC models were approximately 60% of the
Fig. 11. Effect of link axial load on the hysteresis of the RLB and RLB model (Table 1), but they could have been designed to have
SFYC-T models the same stiffness as the RLB by increasing d [Eq. (7)].
A series of pushover analyses of a single-story, single-bay EBF
was conducted where the link stiffness was varied while the frame
dimensions and other member properties were held constant. Re-
applied. Mansour (2010) found that axial tension increases link
sults showed that if the link stiffness is decreased by 40%, then the
shear strength by preventing local web buckling in links composed
frame stiffness only decreases to 77% of its original value. There-
of wide-flange sections, while a compressive load reduces the link
fore a significant reduction in link stiffness does not necessarily
shear strength by increasing local web buckling. This is observed in
have a great effect on the frame stiffness.
the RLB at large deformations (Fig. 11). The SFYC-T does not
The percentage of the cast link length that can deform plasti-
experience any local web buckling; therefore, the link shear
cally, η, can be defined as
strength is less affected by applied axial loads at large deformations
 
(Fig. 11). xs
η ¼ 100 1 − ð10Þ
e
Effect of Geometric Imperfections
To increase the ductility and low-cycle fatigue life of the cast
To study the effect of initial imperfections on the cast link, the lin- link, η should be maximized by minimizing xs , which can be most
ear superposition of buckling eigenmodes was introduced to the efficiently achieved by minimizing d [Eq. (8)].
SFYC-T model before applying the modified cyclic loading proto- The volume of the steel should also be minimized for an effi-
col. Two sets of eigenmodes were found when imposing different cient and more economical cast link design. The volume of steel in
perturbations on the model: the first set, designated by L, mostly the cast link, V cast , can be approximated by
includes deformed shapes of flanges or webs that are out of plane;
the second set, designated by M, consists of deformed shapes in nðe − xs Þ
V cast ¼ ½dðbo þ 2wÞ − ðd − 2tÞðbo − 2wÞ þ 2nwxs d
which the link is misaligned from the link’s longitudinal centerline. 2
The two sets of eigenmodes were analyzed separately using a linear ð11Þ
combination of the first five buckling modes weighted by 10, 7.5,
2.5, 1, and 0.5 respectively. These weightings resulted in initial im- and can be minimized as a function of d.
perfections large enough to be easily observed by visual inspection
and are more severe than what would be acceptable for real appli-
Effect of Depth and Width of Yielding Elements
cation to a structure.
The reduced cyclic loading protocol was applied to the initially Multiple advantages arise when minimizing the flange and web
imperfect models with and without constant axial loads. The hys- thickness of the cast link: a reduction in mass (Fig. 12); a reduction
tereses obtained from these analyses were essentially identical to in overstrength because of the shape factor; and an increase in

© ASCE 04016079-8 J. Struct. Eng.

J. Struct. Eng., 2016, 142(10): 04016079


w = t = 20 mm Insufficient web resistance Potential lateral-torsional
110

Elastic Stiffness of Cast Link (x 103 kN/rad)


w = t = 25 mm buckling (d/b o > 1.0)

ductility and low-cycle fatigue life


175 Increasing ductility and

connection strength;decreasing
100 w = t = 30 mm low-cycle fatigue life;

Increasing volume and req’d


Cast Link Mass (kg)

decreasing overstrength
150 e = 0.5 m
90 = 66%
e = 0.75 m
80 125 = 75% e = 1.0 m
e = 1.25 m
70 e = 1.5 m
100
e = 2.0 m
e = 1.75 m
60
75
Potential flange
50 local buckling
80 100 120 140 160 180 200 220 240 Recommended
50
Link Depth (mm) 0.25 0. 5 0.75 1.0 design space
d/b o
Downloaded from ascelibrary.org by Kyla Tan on 11/08/16. Copyright ASCE. For personal use only; all rights reserved.

Fig. 12. Effect of flange and web thickness on the mass of an 800-mm-
Increasing d, xs , and lateral-torsional
long cast link with a shear strength of 579 kN using Eq. (11) and
buckling potential; decreasing b o
assuming a steel density of 7,850 kg=m3 ; Fy ¼ 350 MPa;
E ¼ 200,000 MPa; G ¼ 77,000 MPa; and n ¼ 1 Fig. 13. Recommended design space for cast links with a shear
strength of 579 kN assuming Fy ¼ 350 MPa; E ¼ 200,000 MPa;
G ¼ 77,000 MPa; n ¼ 1; w ¼ t ¼ 20 mm; and ω ¼ 1.8
ductility and low-cycle fatigue life, as Eq. (8) shows that xs can be
reduced by minimizing w, which will maximize η. However, the
minimization of the web and flange thicknesses is limited by prac- constant frame dimensions. Therefore, longer links are recom-
tical castability limitations, shear resistance according to Eq. (6), mended to increase the ductility and low-cycle fatigue life of
and the need to prevent local buckling; section Class 1 limits pre- the cast links. Using longer links can also provide the architectural
scribed by CSA S16-09 are recommended. Further investigation benefit of larger openings in the EBF. However, for the same link
would be required to relax the limits on the dimensions of the tri- strength, a longer link will proportionally increase its end moments
angular-shaped flanges of cast links. as well as the capacity design forces for the link connections and
the other members of the EBF. In addition, an increase in link
Effect of Link Length length increases the mass of the casting (Table 3).

The effect of the link length, e, on the cast link properties and per-
formance was investigated by developing several finite-element Cast Link Design Space
models of the cast link with varying lengths (Table 3). A stiffer Fig. 13 was constructed by using Eqs. (5)–(10) and by applying the
cast link can be achieved if a shorter length is used. In contrast, design limits discussed above. It demonstrates that even after applying
the overstrength values were not sensitive to link length as they these recommended limits, there are numerous combinations of di-
were consistently within the range of 1.13 to 1.33. However, in- mensions within the recommended design space for a particular
creasing η or decreasing d can reduce the overstrength (Table 3). strength of the cast links. Every point within the design space repre-
The plastic strains decrease as e is increased. Even lower plastic sents a cast link with a unique combination of d and bo , despite having
strains can be expected for longer links since link rotation demand identical aspect ratios d=bo . Fig. 13 summarizes the effect of the cast
decreases as e increases for a given frame drift and assuming link dimensions on the resulting properties and performance values.
The effect of d on the mass of the cast link is not included in
Table 3. Effect of e and η on Cast Link Properties and Performance Fig. 13. The range of dimensions that achieves the minimum vol-
ume or mass overlaps with those associated with η ¼ 66 to 75%
Maximum (Table 3). Given the numerous possibilities, a cast link with the
equivalent dimensions associated with a point near the center of the design
plastic Mass
space is recommended as it balances all of the cast link properties
Model e (mm) d (mm) bo (mm) η Overstrengtha strainb (kg)
and performance values while minimizing the mass of the casting
e500η75 500 125 215 75 1.13 0.163 30 as much as possible.
e500η66 500 144 168 66 1.19 0.174 29 Further investigation into the prescribed limits is recommended
e500η50 500 176 114 50 1.28 0.220 31
in order to expand the design space. Although it is possible to de-
e1000η75 1,000 176 263 75 1.22 0.097 82 sign a cast link outside of the recommended design space, there will
e1000η66 1,000 203 206 66 1.26 0.117 81 be disadvantages in practicality and performance which must be
e1000η50 1,000 249 138 50 1.32 0.154 86 considered. A replaceable cast steel link can be designed to achieve
e1500η75 1,500 216 301 75 1.24 0.080 148 the required performance by developing a design space similar to
e1500η66 1,500 249 236 66 1.28 0.098 148 the one shown in Fig. 13. A detailed description of this design pro-
e1500η50 1,500 305 156 50 1.33 0.136 157 cedure can be found in (Tan 2014).
Note: Assuming V n ¼ 579 kN; Fy ¼ 350 MPa; E ¼ 200,000 MPa;
G ¼ 77,000 MPa; n ¼ 1; w ¼ t ¼ 20 mm. Cast Link Overstrength
a
Measured at 0.09 rad total link rotation with respect to plastic shear
strength given by Eq. (12). Thus far, the overstrength of the cast link has been calculated with
b
Measured at 0.16 rad total link rotation. respect to the nominal strength of the cast link given by Eq. (6),

© ASCE 04016079-9 J. Struct. Eng.

J. Struct. Eng., 2016, 142(10): 04016079


Group of elements to
9.0 m model replaceable link
structural analysis program, OpenSees. The five-story building
M M P was designed for normal occupancy in Victoria, British Columbia,
Canada, assuming no near-fault effects with a Class C soil profile
e type using the 2010 National Building Code of Canada (NBCC)
M M P
e (NRCC 2010) and the design provisions for steel structures given
4 @ 3.7 m

Group of Elements to Model in CSA S16-09. Each story of the building has a floor area of
M M P
Replaceable Shear Link 2,900 m2 . The lateral force–resisting system consists of four EBFs
(Rolled or Built-Up Section) in the east-west direction and MRFs in the north-south direction
M M P and has no irregularities. The response of one of these four EBF
frames was considered in this study (Fig. 14), assuming that the
M M P e total weight for each frame is 14,433 kN. Further details about
the designs of the EBF structures are given in (Tan 2014). The base
4.2 m

Group of Elements to Model


Replaceable Cast Steel Link structure, EBF-R, had a fundamental period of 0.90 s. It was ini-
tially designed by Mansour (2010), with 1,100-mm-long replace-
able shear links composed of rolled wide-flange sections and bolted
Downloaded from ascelibrary.org by Kyla Tan on 11/08/16. Copyright ASCE. For personal use only; all rights reserved.

Fixed nodal connection Elastic beam-column element end-plate connections.


Pinned nodal connection Nonlinear beam-column element A comparable structure, EBF-C, was designed by replacing the
M
Mass assigned to node Truss element shear links in EBF-R with cast links with the same strengths and
P
Node with gravity loads Rotational zero length element lengths. The beams, braces, and columns of EBF-C were identical
Shear zero length element
to those of EBF-R. The cast links in EBF-C had approximately 60
to 75% of the stiffness of the shear links in EBF-R (Table 4) but
Fig. 14. OpenSees model schematic of five-story EBF sample structure resulted in a fundamental period of 0.93 s, which was only 0.03 s
longer than that of the EBF-R. Since EBFs are inherently stiff and
exhibit story drifts that are typically much smaller than building
code limits, designing a slightly more flexible cast link is not nec-
which specifies the shear force at which the cast link is expected to
essarily detrimental to the seismic performance of EBFs. Further-
initially yield. However, overstrength should be taken with respect
more, a more flexible frame is expected to result in a higher
to V p , the plastic modulus of the rectangular, tubular section at the
fundamental period and attract a lower base shear, which would
ends of the cast link’s yielding elements
result in a more economical structure.
nFy A proposed series of predesigned and prequalified links was de-
Vp ¼ ½b d2 − ðbo − 2wÞðd − 2tÞ2  ð12Þ
2e o veloped in strengths ranging from 200 to 1,000 kN in increments of
200 kN and lengths ranging from 500 to 2,500 mm in increments of
Table 3 shows the overstrength of the cast link models with re- 250 mm. To achieve a comparable design to EBF-R and EBF-C,
spect to the plastic strength, given by Eq. (12), at a total link ro- 1,000-mm-long cast links were selected from this series to form a
tation of 0.09 rad, which is approximately equivalent to a plastic third structure, EBF-S, which had a fundamental period of 0.90 s
link rotation of 0.08 rad. In general, the overstrength is less than 1.3 and a total steel tonnage of approximately 90% of the steel tonnage
with the exception of models with η ¼ 50%. Links with η ¼ 50% of the EBF-R and EBF-C.
are outside of the design space shown in Fig. 13 and are not rec- The EBF-F structure was created by redesigning EBF-R with
ommended. Therefore, the same overstrength value of 1.3 that is significantly more flexible 1,000-mm-long cast steel links. The
used for conventional links (CSA S16-09) is recommended as a maximum allowable design period permitted by the building code
target value for the design of cast links. (NRCC 2010) was ignored and a longer period of 1.5 s was initially
assumed. Capacity design resulted in smaller beams, braces, and
Replaceable Cast Link Performance in a Sample columns with a steel tonnage of approximately 77% of the steel
EBF Structure tonnage of the EBF-R and EBF-C and 86% of the steel tonnage
of the EBF-S. EBF-F had a fundamental period of 1.22 s (36%
Four alternate EBF designs of a sample building with different sets greater than the EBF-R) and is an example of the more economical
of replaceable links were analyzed using the open source nonlinear designs that can be achieved when using more flexible links.

Table 4. Summary of Selected Ground Motion Records


NGAa Seismic event Recording station Directionb Mc Scaling factor PGAd (g)
139 Tabas, Iran (1978) Dayhook Normal 7.4 1.872 0.577
189 Imperial Valley (1979) SAHOP Casa Flores Parallel 6.5 2.050 0.637
517 North Palm Springs (1986) Desert Hot Springs Parallel 6.1 2.367 0.688
778 Loma Prieta (1989) Hollister Diff. Array Parallel 6.9 1.333 0.389
986 Northridge (1994) LA—Brentwood VA Hospital Normal 6.7 2.142 0.403
1,177 Kocaeli, Turkey (1999) Zeytinburnu Parallel 7.5 3.142 0.338
1,512 Chi-Chi, Taiwan (1999) TCU 078 Parallel 7.6 1.325 0.395
a
Identification number from the PEER ground motion database.
b
Direction with respect to the fault.
c
Magnitude.
d
Peak ground acceleration.

© ASCE 04016079-10 J. Struct. Eng.

J. Struct. Eng., 2016, 142(10): 04016079


Nonlinear Modeling of the EBF Structures
3.0 Target spectrum
A schematic of the numerical models developed in OpenSees as- Individual spectra
sumed centerline dimensions. The element types used are shown in

Spectral Acceleration (g)


2.5 Mean spectrum
Fig. 14. All of the members of the EBF were assumed to have full
moment connections, except at the column bases, which were 2.0 Period range
pinned. of interest
A group of elements was used to model the replaceable links 1.5
and assumed lumped plasticity (Fig. 14). Each shear link in
EBF-R was modeled using a zero length spring element between 1.0
two elastic beam-column elements. To model the replaceable cast
steel links in EBF-C, EBF-S, and EBF-F, a zero-length spring 0.5
element acting in the rotational degree of freedom was utilized
on either end of the link group to simulate the flexural yielding. 0
0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
A nonlinear beam-column element with a very large moment of Period (s)
inertia was used between the two zero-length rotational elements
Downloaded from ascelibrary.org by Kyla Tan on 11/08/16. Copyright ASCE. For personal use only; all rights reserved.

to capture any shear deformations of the cast link without contrib- Fig. 15. Spectral accelerations of the seven selected ground motion
uting additional flexural deformations to those of the zero-length records
elements. The zero-length elements employed the nonlinear
material referred to as Steel02, which models bilinear plastic
response with a postyield stiffness; Bauschinger effects; and a
cyclic, isotropic strain hardening. The parameters used were set The link rotations of EBF-S vary over the height of the structure
as b ¼ 0.0045, R0 ¼ 19, cR1 ¼ 0.9, cR2 ¼ 0.15, a1 ¼ as a result of specifying the same link size for two consecutive sto-
a3 ¼ 0.12, and a2 ¼ a4 ¼ 7, which were calibrated by Koboevic ries (Table 4): the demand-capacity ratio of the link was larger for
et al. (2012) based on test data of shear yielding EBF links. This the lower of the two stories and therefore deformed more. A wider
was deemed a sufficient material model for all the links since the range of options could be developed by reducing the cast link
objective of this numerical study was to compare the global behav- strength increments. This would allow a designer to select cast links
ior of the four EBF structures. for each story such that more consistent demand-capacity ratios are
Large displacement effects were accounted for in the models by obtained over the height of the structure. Larger link rotations were
using P-delta formulations for the EBF and leaning columns. Grav- observed for the EBF-F structure because of the increased flexibil-
ity loads considering 1.05 times the dead load and a quarter of the ity, but the mean plastic link rotations were well below the esti-
live load were applied to the top nodes of each leaning column, mated limit (Fig. 16).
while story masses were lumped and applied to the top nodes of Significant link rotations are required to achieve interstory drifts
each EBF column. A rigid diaphragm was modeled by requiring near the 2.5% limit (Fig. 16). The mean interstory drifts of both
the column nodes in each story to have equal displacements in EBF-R and EBF-C were in the range of 1%, less than half of
the horizontal degree of freedom. Further details about the nonlin-
the 2.5% limit prescribed by NBCC 2010 (NRCC 2010). Even with
ear modeling of the EBF structures are given in Tan (2014).
extremely flexible links such as those implemented in EBF-F, the
interstory drift ratio did not exceed this limit for a single ground
Seismic Response of EBF Structures motion record. Therefore, it may be beneficial to take advantage
Each of the four EBF structures was subjected to seven selected of the flexible nature of the cast links to achieve more economical
scaled ground motion records obtained from the PEER Ground structural designs despite the increased nonstructural damage that
Motion Database (PEER 2010) and linearly scaled such that the may result from an increase in the interstory drifts. This is reiterated
mean acceleration spectrum was greater than that of the design by the noticeably reduced story shears that were obtained for EBF-
spectrum within the period range of 0.5–1.8 s (Fig. 15 and F in Fig. 16.
Table 4). Thirty seconds of damped free vibrations were included Residual deformations were also monitored in this study. The
in all of the analyses to determine the residual story drifts. The largest interstory residual drift ratios occurred in the first story
OpenSees models considered Rayleigh damping of 5% in the first of each model and decreased at each consecutive story above to
and third modes to match the damping used to develop the spectral nearly zero in the fifth story. They ranged from 0.49 to 0.57%
accelerations. for the first story of EBF-R, EBF-C, and EBF-S. In contrast, the
The response of the EBF-C structure was found to be similar to first story of EBF-F had a mean interstory residual drift ratio of
those of EBF-R despite being more flexible (Fig. 16). Although the 0.86% and ranged from 0.26 to 0.49% for the upper storys. Work
mean peak absolute plastic link rotations of EBF-R were found to by Stratan et al. (2014) confirmed that replaceability of links in
be less than the 0.08-rad limit for shear links, the capacity was structures with moderate residual drifts is possible without major
closely approached or exceeded for several individual records at intervention in recentering the frame because of the springback ac-
different stories. It is expected that with a larger sample of ground tion of members that remain in the elastic range. As such it was
motion records, the plastic rotation capacity of EBF links would be deemed that for this range of residual deformations the structures
exceeded in a larger number of cases. Since the maximum equiv- could be repaired and rehabilitated following an earthquake by
alent plastic strains of the SFYC links are typically at least 50% less replacing the yielded links.
than that of shear yielding links (Tan 2014), it may be assumed that Since the fundamental period of the EBF-R, EBF-C, and EBF-S
the plastic link rotation capacity of the cast links is at least double structures do not differ significantly from one another, similar
that of shear yielding links composed of rolled or built-up sections absolute accelerations were found over the height of the building
and is approximately 0.16 rad. In contrast to EBF-R, the replace- (Fig. 16). The accelerations are reduced, as expected, in the EBF-F
able cast steel links implemented in EBF-C have a much higher structure, demonstrating an additional benefit, which may be
plastic rotation capacity, providing a greater margin of safety. achieved by implementing more flexible cast links.

© ASCE 04016079-11 J. Struct. Eng.

J. Struct. Eng., 2016, 142(10): 04016079


Shear link limit Drift limit Notation
5 Cast link limit 5
The following symbols are used in this paper:
4 4
b = link plate width;
d = link plate depth;
Storey

Storey
3 3
E = Young’s modulus;
2 2 e = link length;
1 1
Fy = yield strength;
G = shear modulus;
kr = elastic stiffness in terms of relative link end displacement;
0 0.1 0.2 0 1.0 2.0
Interstorey Drift Ratio (%)
n = number of yielding elements in the link;
Absolute Plastic Link Rotation (rad)
S = elastic section modulus;
t = flange thickness;
5 5 V cast = volume of steel in the cast link;
V n = nominal yield strength;
4 4 w = web thickness;
Downloaded from ascelibrary.org by Kyla Tan on 11/08/16. Copyright ASCE. For personal use only; all rights reserved.

x = distance from the link center;


Storey
Storey

3 3
xs = length of constant width region;
2 2 γ pc = plastic shear rotation capacity;
η = percentage of the link length that can deform plastically;
1 1 and
ω = overstrength.
0 1000 2000 0 0.2 0.4 0.6
Storey Shear (kN) Absolute Acceleration (g)

EBF-R EBF-S References


EBF-C EBF-F
ABAQUS/CAE 6.10-1 [Computer software]. Systèmes, Vélizy-Villacou-
blay, France.
Fig. 16. Comparison of mean peak time–history responses
AISC. (2010). “Seismic provisions for structural steel buildings.” ANSI/
AISC 341-10, Chicago.
Berman, J. W., Okazaki, T., and Hauksdottir, H. O. (2010). “Reduced link
sections for improving the ductility of eccentrically braced frame link-
to-column connections.” J. Struct. Eng., 10.1061/(ASCE)ST.1943-
Conclusion 541X.0000157, 543–553.
Bruneau, M., Uang, C. M., and Sabelli, R. (2011). Ductile design of steel
Four replaceable cast steel link concepts for EBF frames involving structures, 2nd Ed., McGraw-Hill, New York.
various types of yielding mechanisms were presented in this paper Christopoulos, C., and Filiatrault, A. (2006). Principles of passive supple-
and investigated using finite-element analyses. From this prelimi- mental damping and seismic isolation, IUSS Press, Pavia, Italy.
nary investigation, it was concluded that the best performance was CSA (Canadian Standards Association). (2009). “Limit states design of
achieved with the replaceable cast steel link that employed the con- steel structures.” CAN/CSA-S16-09, Mississauga, ON, Canada.
cept of simultaneous flexural yielding in regions of varying width Della Corte, G., D’Aniello, M., and Landolfo, R. (2013). “Analytical and
that have a rectangular, hollow cross section. A bolted end-plate numerical study of plastic overstrength of shear links.” J. Constr. Steel
Res., 82, 19–32.
connection with chamfered openings was proposed for this cast
de Oliveira, J. C., Packer, J. A., and Christopoulos, C. (2008). “Cast steel
link concept. Its unique geometry prevents the failure modes that connectors for circular hollow section braces under inelastic cyclic
are typically observed in links composed of rolled, built-up loading.” J. Struct. Eng., 10.1061/(ASCE)0733-9445(2008)134:3(374),
sections, or constant section hollow tubes, allowing the cast link 374–383.
to achieve at least double the rotation capacity and, consequently, Elgaaly, M., Hamilton, R. W., and Seshadri, A. (1996). “Shear strength of
have a longer low-cycle fatigue life. Physical testing and experi- beams with corrugated webs.” J. Struct. Eng., 10.1061/(ASCE)0733-
mentation is necessary to confirm the performance values and de- 9445(1996)122:4(390), 390–398.
sign limits as well as to assess its low-cycle fatigue properties more Engelhardt, M. D., and Popov, E. P. (1992). “Experimental performance of
comprehensively. The effect of cast links on the global performance long links in eccentrically braced frames.” J. Struct. Eng., 10.1061/
(ASCE)0733-9445(1992)118:11(3067), 3067–3088.
of a multistory EBF was studied through nonlinear time-history
Fleischman, R. B., Li, X., Pan, Y., and Sumer, A. (2007). “Cast modular
analyses and results indicate that it is possible to design EBF struc- panel zone for steel special moment frames. Part I: Analytical develop-
tures with cast links while abiding by the provisions of NBCC ment.” J. Struct. Eng., 10.1061/(ASCE)0733-9445(2007)133:10(1393),
(NRCC 2010) and CSA S16-09. The cast links can also be de- 1393–1403.
signed for increased flexibility for more economical EBF struc- Gray, M. G. (2012). “Cast steel yielding brace system for concentrically
tures. Similar EBF seismic responses can be obtained with the braced frames.” Ph.D. dissertation, Univ. of Toronto, Toronto.
cast links as conventional or replaceable shear links composed Gray, M. G., Christopoulos, C., and Packer, J. A. (2014). “Cast steel yield-
of rolled or built-up sections, but with a greater margin of safety. ing brace system for concentrically braced frames: Concept develop-
ment and experimental validations.” J. Struct. Eng., 10.1061/(ASCE)
ST.1943-541X.0000910, 04013095.
Kanvinde, A., Marshall, K., Grilli, D., and Bomba, G. (2014).
Acknowledgments “Forensic analysis of link fractures in eccentrically braced frames
during the February 2011 Christchurch earthquake: Testing and simu-
The authors wish to acknowledge the financial support of the lation.” J. Struct. Eng., 10.1061/(ASCE)ST.1943-541X.0001043,
Natural Sciences and Engineering Research Council of Canada. 04014146.

© ASCE 04016079-12 J. Struct. Eng.

J. Struct. Eng., 2016, 142(10): 04016079


Kasai, K., and Popov, E. P. (1986). “Cyclic web buckling control for shear Priestley, M. J. N., Sritharan, S., Conley, J. R., and Pampanin, S. (1999).
link beams.” J. Struct. Eng., 10.1061/(ASCE)0733-9445(1986)112: “Preliminary results and conclusions from the PRESSS five-storey
3(505), 505–523. precast concrete test building.” PCI J., 44(6), 42–67.
Koboevic, S., Rozon, J., and Tremblay, R. (2012). “Seismic performance of Prinz, G. S., and Richards, P. W. (2009). “Eccentrically braced frame
low-to-moderate height eccentrically braced steel frames designed for links with reduced web sections.” J. Constr. Steel Res., 65(10–11),
North American seismic conditions.” J. Struct. Eng., 10.1061/(ASCE) 1971–1978.
ST.1943-541X.0000433, 1465–1476. Sause, R., and Braxtan, T. N. (2011). “Shear strength of trapezoidal cor-
Mansour, N. (2010). “Development of the design of eccentrically braced frames rugated steel webs.” J. Constr. Steel Res., 67(2), 223–236.
with replaceable shear links.” Ph.D. dissertation, Univ. of Toronto, Toronto. Stratan, A., et al. (2014). “Experimental program for large-scale tests
NRCC (National Research Council Canada). (2010). National building
on re-centring dual eccentrically braced frame.” Proc., EUROSTEEL
code of Canada, 13th Ed., Ottawa.
European Conf. on Steel and Composite Structures, Technical Univ.
Okazaki, T., Arce, G., Ryu, H.-C., and Engelhardt, M. D. (2005). “Exper-
of Denmark (DTU), Herlev, Denmark.
imental study of local buckling, overstrength, and fracture of links in
eccentrically braced frames.” J. Struct. Eng., 10.1061/(ASCE)0733- Sumer, A., Fleischman, R. B., and Hoskisson, B. E. (2007). “Development
9445(2005)131:10(1526), 1526–1535. of a cast modular connector for seismic-resistant steel moment frames.
PEER (Pacific Earthquake Engineering Research Center). (2010). “PEER Part I: Prototype development.” Eng. J., 44(3), 195–211.
ground motion database web application, beta version.” Univ. of Tan, K. G. (2014). “Replaceable cast steel links for eccentrically braced
Downloaded from ascelibrary.org by Kyla Tan on 11/08/16. Copyright ASCE. For personal use only; all rights reserved.

California, Berkeley, CA. frames.” M.A.Sc. thesis, Univ. of Toronto, Toronto.

© ASCE 04016079-13 J. Struct. Eng.

J. Struct. Eng., 2016, 142(10): 04016079

You might also like