You are on page 1of 18

Engineering Structures 223 (2020) 111172

Contents lists available at ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

Experimental and numerical study on the seismic behavior of high-strength T


steel framed-tube structures with end-plate-connected replaceable shear
links
Hao Zhanga, Mingzhou Sua,b, Ming Liana,b, , Qianqian Chenga, Binlin Guana, Huanxue Gonga,b

a
School of Civil Engineering, Xi’an University of Architecture and Technology, Xi’an, China
b
Key Lab of Structural Engineering and Earthquake Resistance, Ministry of Education (XAUAT), Xi'an, China

ARTICLE INFO ABSTRACT

Keywords: With the aim of improving the seismic performance and resiliency of steel framed-tube structures (SFTSs), high-
Steel framed-tube structure strength SFTSs with end-plate-connected replaceable shear links (HSS-FTS-RSLs) were developed in this work.
High-strength steel The deep beams and columns used high-strength steel (HSS), while the replaceable links used conventional or
Replaceable shear link low yield point steel. The replaceable shear links acted as ductile fuses at the mid length of the deep beams to
Quasi-static cyclic test
dissipate energy by shear yielding. A series of quasi-static cyclic tests were performed to investigate the seismic
Seismic behavior
Replaceability
behavior and replaceability of HSS-FTS-RSLs through three two-thirds-scale sub-assemblage specimens. The test
Finite element analysis results indicated that all specimens produced a stable and full hysteretic response, resulting in an excellent
inelastic deformation and energy dissipation capacity under cyclic loading. The shear links exhibited a large
inelastic shear deformation capacity in excess of 0.12 rad. The residual interstory drift and residual link shear
deformations of the specimens, which allowed for the easy installation of the new links, exceeded 0.36% and
0.0084 rad, respectively. The replacement of the links had a limited effect on the initial stiffness. The inelastic
deformation and damage to the sub-assemblage specimens were concentrated only within the shear links, while
the other structural components maintained elasticity under cyclic loading. This indicated that the HSS-FTS-RSLs
could achieve a quick rehabilitation after a major earthquake and reduce the retrofit cost. In addition, nonlinear
finite element (FE) models of test specimens were implemented in ABAQUS. The analysis results showed that the
load-carrying capacity, initial stiffness, development of the plasticity, and failure models obtained by the FE
models were in good agreement with the experimental responses.

1. Introduction stiffness, plastic deformations in the partial column ends occur earlier
than in the beam ends, resulting in an increased risk of collapse. The
Steel framed-tube structure (SFTS) systems consist of closely spaced level of damage to external columns and deep beams may be extensive
perimeter columns interconnected with deep beams. Therefore, SFTSs during a design-level earthquake. Therefore, the post-damage repair
are often employed as efficient high-rise structural systems that resist required to provide an occupancy performance level is challenging.
lateral forces and overturning moments owing to their immense lateral Recently, earthquake resilient structure systems (ERSs) have at­
stiffness. In SFTSs, external columns are placed at 3.0–4.0 m intervals, tracted increasing interest as one of the frontiers in earthquake en­
while deep beams with depths varying from 0.6 to 1.0 m are used to gineering. In eccentrically braced frames (EBFs), the shear link exhibits
achieve a framed tubular behavior [1]. The clear span-to-depth ratios, an excellent energy dissipation capacity and stable hysteretic behavior
ln/hb, of the deep beams are between 3 and 5 (see Fig. 1), thus pro­ through shear yielding under cyclic loading [3–6]. As one type of ERS, a
ducing a sharp moment gradient along the deep-beam span that results structure system with replaceable shear links was proposed and in­
in a short flexural plastic hinge length that cannot be developed vestigated in moment resisting frames (MRFs) [7] and EBFs [8].
properly in the beam ends under cyclic loading. This indicates that In general, high-strength steel (HSS) is defined as structural steel
traditional SFTSs have a limited energy dissipation capacity, in­ with a nominal yield strength of 460 MPa or greater per ANSI/AISC
sufficient ductility, and poor seismic behavior [2]. Furthermore, con­ 360-16 [9]. Improvements in mechanical behaviors including the
sidering the composite action of floor slabs and deep beams with great ductility, toughness, and weldability of HSS have gained an increasing


Corresponding author at: School of Civil Engineering, Xi’an University of Architecture and Technology, Xi’an, China.
E-mail address: lianming@xauat.edu.cn (M. Lian).

https://doi.org/10.1016/j.engstruct.2020.111172
Received 6 November 2019; Received in revised form 20 June 2020; Accepted 24 July 2020
0141-0296/ © 2020 Elsevier Ltd. All rights reserved.
H. Zhang, et al. Engineering Structures 223 (2020) 111172

at the mid-span of the deep beams.


Recently, Lian et al. [2] studied the seismic performance of the sub-
structures and overall structure of SFTS-RSLs (all structural members
used Q345 steel) exclusively by a finite element analysis. However, the
seismic behavior of HSS-FTS-RSLs was not investigated. Nikoukalam
and Dolatshahi [13] proposed a new hybrid energy dissipating steel
moment resisting frame (MRF) with shear links and reduced beam
section (RBS) connections. A finite element analysis was used to study
the monotonic and cyclic behaviors of the hybrid frames with a single-
story and single-span compared to the conventional steel MRFs. Rah­
navard et al. [14] and Hassanipour et al. [15] evaluated numerically
the seismic behavior for reduced beam-section moment connections
under cyclic loading. Furthermore, Mahmoudi et al. [16] analyzed the
cyclic behavior of a hybrid frame with a half-scaled one-story and one-
span specimen via a cyclic loading test. However, all members of the
specimen used ST37 steel, and the replaceability of the shear links and
Fig. 1. Shear and moment demand for beams in SFTSs.
hysteretic behavior of the specimen made of HSS were not investigated.
In this study, replaceable shear links designed as sacrificial dis­
acceptance in structural engineering [10,11]. HSS exhibits a higher sipative elements are inserted into the deep spandrel beams to allow the
load-carrying and elastic deformation capacity compared to conven­ activation of ductile hysteretic dissipative mechanisms under severe
tional mild steel (CMS, nominal yield strength, fy, is lower than ground motions. However, the seismic performance of the proposed
345 MPa), and thus reduces the required cross-section dimensions of HSS-FTS-RSL system has not been investigated in previous research.
structural components under identical design conditions. Furthermore, Guan et al. [17] and Cheng et al. [18] investigated experimentally the
HSS has been successfully applied in typical building structures in re­ feasibility and seismic performance of HSS-FTS-RSLs with single span-
cent years [11]. Q460 HSS (nominal fy = 460 MPa) has been adopted single story sub-structures. To this end, this paper presents an experi­
by the Chinese Standard for Design of Steel Structure (GB 50017-2017) ment and finite element analysis that investigated the seismic perfor­
[12]. HSS, therefore, has immense potential to be a key structural mance of HSS-FTS-RSLs. First, three two-thirds-scale sub-assemblage
material used in construction. specimens with one-bay and one-story were designed. Quasi-static
To mitigate the drawbacks of traditional SFTSs, high-strength steel cyclic tests were conducted to study the cyclic behaviors of the sub-
framed-tube structures with replaceable shear links (HSS-FTS-RSLs) structures. The experimental results were then detailed, including
have been proposed, which combine the advantages of HSS and shear failure modes, hysteretic response, load-carrying and energy dissipation
links. As shown in Fig. 2, the shear links use CMS or low yield point capacity, stiffness degradation, strain profile, inelastic shear deforma­
(LYP) steel, while the columns and deep beams are made of HSS. The tion capacity, overstrength factor, and replaceability of the links. Fi­
replaceable shear links act as a ductile fuse at the mid length of the deep nally, nonlinear 3D finite element models created using ABAQUS 6.14
beams to dissipate the seismic input energy through inelastic shear [19] were developed and validated to capture the observed experi­
yielding, which results in a lower requirement for the development of mental responses.
flexural hinges at the beam-ends under seismic loading. Thus, re­
placeable shear links were used and studied as sacrificial components in
traditional SFTSs with low beam span-to-depth ratios. The inelastic 2. Experimental program
deformation was concentrated within the shear link, and the external
columns and deep beams were designed to remain elastic owing to the 2.1. Test specimens
higher yield strength of HSS compared with CMS. In HSS-FTS-RSLs, a
post-earthquake low-residual permanent deformation occurs due to a A 30-story HSS-FTS-RSL building was designed as the prototype
high lateral stiffness. The damaged links could be quickly disassembled structure according to the requirements in ANSI/AISC 360-16 provi­
and replaced to meet the occupancy performance level after a major sions [9], ANSI/AISC 341-16 [20], Chinese Standard GB 50011-2010
earthquake, which would significantly reduce the disruption time of the [21], and JGJ 99-2015 [22]. The site of this prototype building was a
structure. Furthermore, the section dimensions of the replaceable links highly seismic area in Xi’an, China, where the peak ground acceleration
and deep beams were independently designed based on capacity design (PGA) of the design basis earthquake (DBE) with a 10% probability of
principles to achieve the stated performance objectives owing to the exceedance in 50 years was equal to 0.2 g. As shown in Fig. 3, the
decoupling of the strength and stiffness design of the replaceable links building’s plan dimensions were 27 m × 27 m, and the spans in each
principle direction were 3 m. Each story height was 3.3 m, and the total
height was 99 m. The sectional dimensions of the components of the
prototype building and the design results from a dynamic time history
analysis were reported in the study by Guan et al. [17]. The thirteenth
story of the prototype structure was selected as the representative story,
and two-thirds-scale sub-assemblage specimens were fabricated for
quasi-static cyclic tests based on laboratory conditions (see Fig. 3(c)). In
addition, the inflection points on the two side columns were generally
located at the midpoint of the top and bottom columns under lateral
loading. Thus, simple hinge supports were defined for the top and
bottom of the half columns as the boundary condition.
To concentrate the inelastic deformation and damage in the re­
placeable links, while maintaining linear elasticity in the deep beams
and columns to ensure that the sub-assemblage specimens can be re­
paired with minimal effort, the replaceable link sections were designed
Fig. 2. Replaceable link concept in the HSS-FTS-RSLs. in accordance with the following equations:

2
H. Zhang, et al. Engineering Structures 223 (2020) 111172

Fig. 3. Selected prototype structure for the quasi-static tests.

e 1.6Mp/ Vp, (1) hb), where hb denotes the beam depth, of the deep beams was equal to
4.1. Parameters included the link length and steel grade combination
Vp = 0.6Aw fyL , (2) type of the replaceable link and deep beam. The three specimens were
referred to as HFTSL-1, HFTSL-2, and HFTSL-3. In specimens HFTSL-1
Mpb Vp l pb/2, (3) and HFTSL-2, the link lengths were both 400 mm. The former was made
of Q235 steel (fy = 235 MPa) and the latter was LYP225 steel
Vpb Vp, (4) (fy = 225 MPa). For specimen HFTSL-3, the link length was 280 mm,
and it was fabricated from LYP225 steel. The deep beams and columns
where e is the link length, Mp is the plastic flexural strength of the link,
of the specimens all used Q460 steel (fy = 460 MPa), and the end plates
Vp is the plastic shear strength of the link, Aw is the link web area, fyL is
used Q345 steel (fy = 345 MPa). The section of the deep beams was
the yield stress of the link web, and Ω is the overstrength factor of the
H400 × 148 × 10 × 12 (mm), where the four numbers denote the
link, which can be taken as 1.5 (recommended for shear links in ANSI/
section depth (h), flange width (b), web thickness (tw), and flange
AISC 341-16 [20]). Mpb is the plastic flexural strength of the beam, Vpb
thickness (tf), respectively (i.e., h = 400, b = 148, tw = 10, and
is the plastic shear strength of the beam, and lpb is the distance between
tf = 12). The column sections were H360 × 226 × 12 × 16.
the plastic hinges of the deep beam, where the plastic hinge was as­
End-plate bolted connections were used between the deep beams
sumed to occur at one-third of the beam depth out of the end of the
and replaceable links of each specimen, which provided a strength that
cover plate.
exceeded the link’s ultimate strength and ensured that post-damaged
Fig. 4 depicts the detailed dimensions of the specimens. The speci­
links were replaceable. The end plates had a length and width equal to
mens had a one-bay and one-story structure, and the height and span
the deep beam dimensions. The end plate thickness and diameter of the
were 2.2 m and 2.0 m, respectively. The clear span-to-depth ratio (ln/

Fig. 4. Geometric dimensions of the specimens (dimensions in mm).

3
H. Zhang, et al. Engineering Structures 223 (2020) 111172

Table 2
Material properties of the steel elements.
Steel type t (mm) fy (MPa) fu (MPa) fy/fu E (×105 δ (%) εy (10−6)
MPa)

Q460-10 10.17 563 695 0.81 1.99 21.89 2887


Q460-12 12.14 545 670 0.81 1.98 21.45 2781
Q460-16 16.01 527 631 0.84 2.03 20.91 2596
Q345-20 19.94 364 526 0.69 2.06 28.87 1767
Q345-25 24.90 359 507 0.71 2.03 31.29 1768
Q235-8 7.90 302 447 0.67 2.00 31.23 1510
Q235-10 9.89 290 428 0.68 2.01 34.29 1443
LY225-8 7.82 212 301 0.70 2.04 50.60 1039
LY225-10 10.06 204 297 0.69 2.02 52.21 1010

Fig. 5. Arrangement of the high-strength bolts (dimensions in mm).

high-strength bolts were designed in accordance with the AISC design


guide series 4 [23]. The thicknesses of the link and beam end plates
were 20 mm and 25 mm, respectively, and were clamped using eight
M20 high-strength bolts with a strength grade of 10.9 (minimum tensile
strength fu = 1000 MPa; strength ratio fy/fu = 0.9). The arrangement
of the high-strength bolts is shown in Fig. 5.
All the replaceable links had a length ratio, e/(Mp/Vp), smaller than
1.6, and thus were dominated by shear yielding per ANSI/AISC 341-16
[20] and GB 50011-2010 [21]. The design parameters of the shear links
are summarized in Table 1. The web and flanges of the links satisfied
Class 1 section requirements, h0/tw ≤ 65(235/fyL)1/2, and b/
Fig. 6. Nominal stress-strain curves of Q235 and LYP225 steel.
(2tf) ≤ 9(235/fyL)1/2, respectively, according to Chinese Standard
GB50017-2017 [12], where h0 denotes the link web depth. Stiffeners
were set up on both sides of the link web to delay web buckling and Q235 steel.
enhance the cyclic response. The stiffener spacing of the links on spe­
cimens HFTSL-1 and HFTSL-2 was 133 mm, and 140 mm on specimen 2.3. Specimen details
HFTSL-3, which satisfied the limit requirement (30tw − h/5) for the
stiffener spacing of shear links per ANSI/AISC 341-16 [20] and GB Fig. 7(a)–(b) show the welding details of the specimens. The end
50011-2010 [21], where tw is the link web thickness, and h is the link plates utilized complete joint penetration (CJP) groove welds to con­
depth. It should be noted that the shear links made of LYP225 and Q235 nect the link flanges and deep beam flanges. The column and beam
steel had similar plastic shears and moment capacities to facilitate the flanges were welded with full-penetration welds. The flange and web
comparison of their seismic behavior under cyclic loading. welds of the shear links, beams, and columns were all double-sided fillet
welds. To ensure the strength and rotation capacity of the beam-to-
2.2. Material properties column connections, the cover plates were welded to the upper and
bottom flanges of the deep beam-ends. Detailed dimensions of the cover
The material properties of each steel component were measured by plates are shown in Fig. 7(c)–(d). Link stiffeners were welded to both
tensile coupon tests. The results are tabulated in Table 2 and include the link web and flanges using double-sided fillet welds. To avoid
actual steel thickness (t), yield strength (fy), tensile strength (fu), ratio of premature link web fractures in the region where the flange-to-web
yield to ultimate strengths (fy/fu), Young's elastic modulus (E), elon­ fillet welds met the stiffeners, the link web-stiffener welds were ter­
gation (δ), and yield strain (εy), where εy = fy/E. Q460-10 denotes steel minated at a distance of 5tw from the flange-to-web weld (tw denotes
with a nominal thickness of 10 mm and a nominal yield strength of link web thickness) [24].
460 MPa. Furthermore, the average measured tensile strength of the
Grade-10.9 M20 high-strength bolts was 1137 MPa. Fig. 6 depicts the 2.4. Test setup and loading protocol
nominal stress-strain curves of the LYP225 and Q235 steel used in the
links for the tensile coupon tests. It is clear from a comparison of the Fig. 8 illustrates the test setup. The top and bottom of the columns
results that the ultimate strain, εu, of the LYP225 steel was approxi­ were pinned to the loading beam and foundation beam, respectively, to
mately 60% higher than that of the Q235 steel, which indicated that the achieve the hinge constraints. Lateral bracing was designed at the upper
LYP225 steel exhibited an excellent ductility, inelastic deformation and bottom flanges of the deep beams to prevent out-of-plane in­
capacity, and preferable energy dissipation capacity compared to the stability. Meanwhile, the columns were also laterally braced via a brace

Table 1
Primary features of the replaceable shear links.
Specimen Section dimensions Steel grade Link length, e e/(Mp/ Flange width to thickness ratio, b/ Web width to thickness ratio, Stiffener spacing
(mm) Vp) (2tf) h0/tw (mm)

HFTSL-1 H220 × 130 × 8 × 10 Q235 400 1.09 6.10 25.00 133


HFTSL-2 H230 × 130 × 8 × 10 LYP225 400 1.08 6.10 26.25 133
HFTSL-3 H230 × 130 × 8 × 10 LYP225 280 0.75 6.10 26.25 140

4
H. Zhang, et al. Engineering Structures 223 (2020) 111172

Fig. 7. Details of the sub-assemblage specimens (dimensions in mm).

beam with a welded H-shaped section connected to the gantry


mounting to prevent out-of-plane deformation and torsion. Polytetra­
fluoroethylene (PTFE) plates, 1.5-mm thick, were glued onto the region
where the lateral braces and specimens came into contact to reduce the
friction forces under cyclic loading. A low-cyclic horizontal displace­
ment-controlled loading was applied on the loading beam using a 1000
kN hydraulic actuator mounted to the reaction wall. It should be noted
that the loading beam had a much higher stiffness to ensure that two
identical lateral loads were applied to both sides of the test specimens.
Considering the representative value of a gravity load with a 1.0 dead
load and 0.5 live load, the constant vertical axial force on each column
was 430 kN, which was successively applied to the tip of the columns
using a hydraulic jack with a maximum capacity of 1000 kN.
Fig. 9 shows the displacement-controlled loading protocol for the Fig. 9. Loading protocol.
test. The loading history included two phases. In Phase-I, the

Fig. 8. Test setup.

5
H. Zhang, et al. Engineering Structures 223 (2020) 111172

the horizontal displacement of the specimens, and the horizontal dis­


placement of the foundation beam was measured by LVDT D9 to
evaluate the rigid body displacement. To obtain the strain responses of
each specimen, three-gauge rosettes were installed on the link web and
panel zone of the column web, and uniaxial strain gauges were installed
on the flanges and webs of beams and columns and on the link flange.

3. Experimental phenomena and failure mode

During the Phase-I cyclic loading, only rust peeled off near the
welds on the link web. This was due to the Phase-I loading being only
2.5Δy′, which corresponded to an interstory drift of θ = 1.25%. The
shear links were then replaced with new links. Therefore, this section
mainly presents the cyclic behavior and failure mode of the specimens
during Phase-II loading, which are described in the following sections.

3.1. Specimen HFTSL-1


Fig. 10. Displacement meter and strain gauge distributions.
In specimen HFTSL-1, the link length using Q235 steel was 400 mm,
displacement was increased by increments of 0.25Δy′ for one cycle which corresponded to a length ratio, e/(Mp/Vp), of 1.09. An evaluated
before the yielding displacement, Δy′, was evaluated, and then in­ yielding displacement of Δy′ = 15 mm was adopted in the Phase-II
creased in increments of 0.5Δy′ for three cycles until 2.5Δy′ was loading. The horizontal load versus displacement curve of HFTSL-1
reached, which was approximately the maximum interstory drift, increased linearly up to the point when the displacement increased to
θmax = 1.25%, of the prototype structure under the maximum credible Δy′. During the first cycle of 2Δy′, black rust peeled off at the web and
event (MCE). The shear link was then replaced with a new link to in­ welds of the shear link. During the third cycle of 2Δy′, the rust that
vestigate the replaceability of the specimens. In Phase-II, the displace­ peeled off from the link web was more obvious and indicated that the
ment-controlled loading was applied at a rate of 0.1 mm/s until the shear link had entered the elastic-plastic state. During the second cycle
specimens reached failure. The incremental value was 0.25Δy′ in each of 3Δy′ (pull), a minor crack developed in the upper left corner of the
step before Δy′ was reached; thereafter, the incremental value was Δy′ right gird of the link web (see Fig. 11(a)). At the same time, the link
until the complete failure occurred. Three cycles were repeated at each shear deformation, γs, was found to be 0.056 rad by using the measured
loading level after reaching Δy′. The actuator pushed and pulled the values of LVDTs D1 and D2 and Eq. (6). During the third cycle of 4Δy′,
specimens to apply the predetermined cyclic deformation. tearing on the upper right corner of the link web was observed. Sub­
sequently, the web crack length reached one-third of the weld length
(see Fig. 11(b)), corresponding to a link shear deformation, γs, equal to
2.5. Instrumentation 0.093 rad and an interstory drift, θ, of 2.38%. In the second cycle of
5Δy′ (push), the link stiffener ruptured in the area adjacent to the welds
The displacement and strain responses of the sub-assemblage spe­ owing to a severely increased tension (see Fig. 11(c)). Meanwhile, the
cimens were obtained by the displacement meters and strain gauges, compression flanges of the shear links showed a slight local buckling,
respectively. As shown in Fig. 10, the shear deformation of the shear and γL equaled 0.127 rad. During the first cycle of 6Δy′, the link web
links (referred as to “link shear deformation”) for the three specimens crack propagated along the vertical welds and grew to half the weld
was measured by diagonal linear variable displacement transducers length. Tearing through the entire web depth on the left grid of the
(LVDTs) D1 and D2. A total of six LVDTs (D3–D8) were used to measure link’s web was observed in the second cycle of 6Δy′ (see Fig. 11(d)).

Fig. 11. Test phenomena of specimen HFTSL-1 during cyclic loading.

6
H. Zhang, et al. Engineering Structures 223 (2020) 111172

Fig. 12. Test phenomena of specimen HFTSL-2 during cyclic loading.

Ultimately, the test was terminated due to the occurrence of obvious deformation, γs, and interstory drift, θ, reached 0.151 rad and 3.23%,
strength degradation (i.e., a greater than 20% reduction in capacity). respectively during this loading cycle. Under the cycle loading of 8Δy′,
The maximum link shear deformation, γsmax, and interstory drift, θmax, tearing adjacent to the welds occurred in the link flange and end plate,
were 0.161 rad and 3.39%, respectively. A photograph of specimen and the length of the crack reached two-thirds of the link flange (see
HFTSL-1 after testing is shown in Fig. 11(e). Fig. 12(f)). During the 9Δy′ cycles, the interstory drift, θ, reached
4.26%, and no evident strength deterioration in specimen HFTSL-2 was
detected. Fig. 12(g) shows the final failure of specimen HFTSL-2. No
3.2. Specimen HFTSL-2 CJP groove welds or double-sided fillet welds with abnormal damage
were observed at any point during the test.
In specimen HFTSL-2, the link length of the LYP225 steel was
400 mm, which corresponded to a length ratio, e/(Mp/Vp), of 1.08. An
evaluated yielding displacement of Δy′ = 12 mm was applied in the 3.3. Specimen HFTSL-3
Phase-II loading. The oxide layer flaked off at the link web, and the
hysteresis loops gradually filled during the 2Δy′cycles, which indicated In specimen HFTSL-3, the link length of the LYP225 steel was very
that a shear yielding of the link web could be detected. During the 3Δy′ short at only 280 mm, which corresponded to a length ratio, e/(Mp/Vp),
cycles, flaking of the black rust along the web and welds of the shear of 0.74 [25]. Deep beams commonly have shorter spans than conven­
link was more obvious (see Fig. 12(a)) and was accompanied by a slight tional moment-resisting frame (MRF) beams, and the use of shorter
burning sensation when the link web surface was touched. During the shear links might reduce the link weight and be more prone to re­
first cycle of 4Δy′ (push), a slight local buckling occurred at the end of placement after earthquakes. An evaluated yielding displacement of
the link’s compression flange (see Fig. 12(b)). At that time, the link Δy′ = 12 mm was adopted for the Phase-II loading. During the 3Δy′
shear deformation, γs, was equal to 0.071 rad, and the interstory drift of cycles, black rust flaked off excessively in the link web (see Fig. 13(a)),
the specimen, θ, was equal to 1.93%. During the second cycle of 5Δy′ which indicated that the shear link had entered the elastic-plastic state.
(push), a minor crack adjacent to the welds was observed in the lower At that time, the link shear deformation, γs, equaled 0.076 rad. Slight
left corner of the intermediate gird of the link web (see Fig. 12(c)). With local buckling occurred at the end of the link’s compression flange,
an increased deformation during the 6Δy′ cycle loading, the crack accompanied by a minor crack in the bottom left corner of the right gird
gradually extended along the web depth of the link, and the crack of the link web during the first cycle of 4Δy′ (push) (see Fig. 13(b)).
length reached one-third of the weld length (see Fig. 12(d)). The link Meanwhile, the link shear deformation, γs, reached 0.108 rad. With the
shear deformation, γs, was 0.125 rad at this loading cycle. Severe local increased displacement, during the second cycle of 5Δy′ with γL equal to
buckling was observed at the end of the link’s compression flange 0.121 rad, a fracture adjacent to the link’s flange-to-end plate welds was
during the first cycle of 7Δy′, accompanied by a slight out-of-plane observed, and the crack length reached half of the weld length (see
deformation of the link web (see Fig. 12(e)). The link shear Fig. 13(c)). During the 6Δy′ cycle loading, the link stiffener adjacent to

7
H. Zhang, et al. Engineering Structures 223 (2020) 111172

Fig. 13. Test phenomena of specimen HFTSL-3 during cyclic loading.

the link flange-to-stiffener welds ruptured (see Fig. 13(d)). Subse­ stiffener and link web tear through the depth of the web under large
quently, the link plates, including the web, flanges, and stiffeners, had a displacement loadings. No local buckling of the web or flange was de­
significant sense of scalding, which indicated that the shear link ef­ tected. The specimen HFTSL-2 exhibited a high plastic deformation
fectively dissipated seismic energy through inelastic shear deformation capacity before the loss of load-carrying capacity and was characterized
under cyclic loading. During the first cycle of 7Δy′, severe local buckling by severe local buckling in the link flange, weld fractures in the link
at the end of the link flange was noticed. Furthermore, the crack on the flange-to-end plates, and slight web tearing. For specimen HFTSL-3, the
right gird of the link web propagated along the stiffener-to-web welds shear deformation capacity was more significant than that of HFTSL-2
and reached one-third of weld length (see Fig. 13(e)). The corre­ under the same interstory drift, owing to the length ratio, e/(Mp/Vp), of
sponding link shear deformation, γs, equaled 0.223 rad for this loading 0.74 being smaller than 1.0. Thus, the failure mode of HFTSL-3 was
cycle. During the second cycle of 8Δy′ (push), a crack that tore through characterized by a web tear and ruptured link stiffener under large
the entire web depth along the stiffener-to-web welds was observed, deformation demands. In addition, the end plates between the shear
accompanied by flexural deformation of the link flange (see Fig. 13(f)). link and deep beam and the high-strength bolts of each specimen ex­
Due to excessive strength degradation (i.e., a greater than 20% reduc­ perienced no slippage during the test. Therefore, the bolted end-plate
tion in capacity), the test was terminated in the final cycle of 8Δy′. The connections provided a sufficient constraint for the replaceable links to
link shear deformation, γs, and interstory drift, θ, reached 0.261 rad fully develop inelastic deformation and dissipate seismic energy. No
and 3.79%, respectively, at the end of the loading process. The final CJP groove or double-sided fillet welds with abnormal damage were
failure of specimen HFTSL-3 is shown in Fig. 13(g). observed at any point during the test.

3.4. Failure characteristics 4. Discussion and evaluation of test results

During the cyclic loading tests, the columns and deep beams made 4.1. Experimental results for the Phase-I loading
of HSS remained linearly elastic as evaluated by a strain measurement
analysis, and no obvious phenomena, including yielding and weld Fig. 14 shows the hysteretic responses of the column tip load, P,
fractures, appeared. All inelastic deformation and damage remained versus the column tip displacement, Δ, for all specimens during the
concentrated within the shear links. This indicated that the post-da­ Phase-I loading. The column tip load, P, was equal to the force provided
mage links could be quickly inspected and replaced after a major by the hydraulic actuator due to the rigid loading beam having a higher
earthquake, and the disruption time of HSS-FTS-RSLs was minimized. stiffness with which to transfer the horizontal shear force. The hys­
The failure modes of specimen HFTSL-1 included a ruptured link teretic curve of each specimen showed stable hysteretic behavior, and

8
H. Zhang, et al. Engineering Structures 223 (2020) 111172

Fig. 14. Hysteretic curves for the Phase-I loading.

Fig. 16. Calculation of the equivalent viscous damping ratio, he.

Phase-II loading. It is clear from the hysteretic curves that all specimens
Fig. 15. Skeleton curves for the Phase-I loading. showed stable and increasing hysteretic loops, which indicated that the
sub-structural specimens had good inelastic deformations and excellent
no evident strength degradation was observed. The hysteresis loops of energy dissipation capacities. No pinching phenomenon of the hys­
specimens HFTSL-2 and HFTSL-3 were more stable and plumper than teretic curves occurred for any specimen, and no evident strength de­
that of specimen HFTSL-1. gradation of the hysteretic curves within the three cycles of equivalent
Fig. 15 shows the skeleton curve comparison for each specimen displacement loading was observed until the end of the loading phase.
obtained by connecting the peak points of the hysteretic loops at the Furthermore, the hysteretic loops of specimens HFTSL-2 and HFTSL-3
first circle of the various amplitude displacements during the Phase-I were plumper than those of HFTSL-1. This was due to the high ductility
loading. The curves show that the carrying capacity of each specimen and inelastic deformation capacity of the LYP225 steel shear links used
increased incrementally with the displacement. The carrying capacity in specimens HFTSL-2 and HFTSL-3 being fully developed under cyclic
of specimen HFTSL-1 was higher than that of specimen HFTSL-2, owing loading.
to the measured yield strength of Q235 steel being higher than that of
LYP225 steel and resulting in a higher carrying capacity. Furthermore, 4.2.2. Energy dissipation capacity
the carrying capacity of specimen HFTSL-3 was higher than that of The cumulative dissipated energies, E, of the three specimens are
specimen HFTSL-2, which indicated that reducing the link length could compared in Fig. 19 for each displacement amplitude in the Phase-II
improve the load-carrying capacity. loading. The amount of energy dissipated by the specimens increased
For a symmetric hysteretic response, the equivalent viscous steadily during the test. The cumulative dissipated energies, E, of spe­
damping ratio, he, was calculated by the expression cimens HFTSL-1 and HFTSL-3 were relatively higher than that of
he = (SABC + SCDA)/2π (SOBE + SODF), where SABC, SCDA, SOBE, and SODF HFTSL-2, as specimen HFTSL-1 had a larger increment of cyclic dis­
are shown in Fig. 16. placement and specimen HFTSL-3 had a higher strength during each
The energy dissipation capacity of the specimens in the Phase-I cycle than did specimen HFTSL-2. Additionally, the higher ductility and
loading is shown in Fig. 17. The cumulative dissipated energy and plastic deformation of specimen HFTSL-2 indicated that an increased
equivalent viscous damping ratio were increased with the increment of level of energy dissipation was anticipated.
displacement, which indicated that the sub-assemblage specimens had a The equivalent viscous damping ratio, he, was used to evaluate and
steady and increasing energy dissipation capacity. Meanwhile, the compare the energy dissipation capacity of the specimens (see Fig. 20).
equivalent viscous damping ratios of specimens HFTSL-2 and HFTSL-3 The ratio, he, of the specimens ranged from 0.318 to 0.536, which in­
were higher than that of specimen HFTSL-1 during the test. The reason dicated that the specimens exhibited an excellent energy dissipation
for this is that the shear link made of LYP225 steel provided a better capacity under cyclic loading. The ratio of each specimen increased
ductility and inelastic deformation capacity than did the Q235 steel. gradually during the test process. However, the slope of the curves for
all specimens decreased gradually during the larger displacement
4.2. Experimental results for the Phase-II loading loading due to the accumulated effects of damage to the shear links and
fatigue to the specimens subjected to cyclic loading. In addition, it is
4.2.1. Hysteretic response apparent that the links that used LYP225 steel in specimens HFTSL-2
Fig. 18 shows the hysteretic responses of each specimen during the and HFTSL-3 had an improved inelastic deformation and ductility,

9
H. Zhang, et al. Engineering Structures 223 (2020) 111172

Fig. 17. Energy dissipation capacity in the Phase-I loading.

which resulted in a higher he ratio for each cycle, compared to those of


specimen HFTSL-1. Thus, it can be concluded that the sub-assemblage
specimens had excellent energy dissipation capacities, especially spe­
cimen HFTSL-2.

4.2.3. Load-carrying and ductility capacities


The skeleton curves of the specimen’s hysteretic responses are
compared in Fig. 21 at each loading step. The carrying capacity of each
specimen increased, even near the ultimate limit state during the cyclic
loading process, which indicated that the specimens exhibited without
strength degradation. It should also be noted that all specimens had
nearly identical initial stiffnesses (see Fig. 21).
Table 3 summarizes the yield point, peak point, ultimate point and
displacement ductility ratio, μ, of each specimen from the skeleton
curves. The yield displacement, Δy, of the specimens was obtained
Fig. 19. Cumulative dissipated energy.
based on the “farthest point method” [26], as shown in Fig. 22; Δy
corresponded to the yield load Py. The maximum load Pm of the ske­
leton curves corresponded to the peak displacement Δm. The ultimate the height of the specimen.
point of the skeleton curves was defined as the point when the load had For specimen HFTSL-1, the maximum load, Pm, was, on average,
decreased to 85% of the maximum load or the ultimate load of the 26.5% higher than the yield load, Py. The maximum load, Pm, of spe­
skeleton curve when the load-carrying capacity of the specimen did not cimens HFTSL-2 and HFTSL-3 were, on average, 47.5% and 47% higher
show evident degradation until the experiment stopped. The ultimate than the yield load, Py, respectively. This indicated that the specimens
load, Pu, corresponded to the ultimate displacement, Δu. The displace­ demonstrated a reliable hardening behavior and load-carrying capacity.
ment ductility ratio, μ, was used to evaluate the ductility of the speci­ The yield drift ratio, θy, of the specimens ranged from 0.83% to 1%, and
mens, and was calculated as μ = Δu/Δy. The yield drift ratio, θy; was equal to 2.07–2.5 times [θe] = 0.4%, and the ultimate drift ratio,
maximum drift ratio, θm; and ultimate drift ratio, θu were calculated as θu, of the specimens ranged from 3.45% to 4.17%, and was equal to
θy = Δy/H; θm = Δm/H; and θu = Δu/H, respectively, where H denotes 1.72–2.08 times [θp] = 2%, where [θe] and [θp] are the limit values of

Fig. 18. Hysteretic curves for the Phase-II loading.

10
H. Zhang, et al. Engineering Structures 223 (2020) 111172

Fig. 22. Determination of the yield points of the specimens.


Fig. 20. Equivalent viscous damping ratio, he.

Fig. 21. Skeleton curves of hysteretic responses. Fig. 23. Stiffness degradation curves of the specimens.

the elastic and elasto-plastic ultimate interstory drift for multi-story and degradation of the specimens for each displacement amplitude. The
high-rise steel structures, respectively, according to Chinese Standard secant stiffness was obtained by calculating the slope of the line con­
GB 50011-2010 [21]. The displacement ductility ratio, μ, ranged from necting the peak points in the pull and push directions. The stiffness
2.99 to 4.60. This indicated that the sub-structural specimens exhibited degradation curves of each specimen are compared in Fig. 23. The
excellent inelastic deformation and a high ductility capacity. curves show that the Kj of the specimens decreased gradually, and the
The ductility ratio, μ, of specimen HFTSL-1 was 34.9% lower than stiffness degradation rate decreased when the displacement loading
that of HFTSL-2 in the push direction and 34.6% lower in the pull di­ increased. All specimens were shown to have a relatively similar stiff­
rection. This indicated that the LYP225 steel shear link could improve ness degradation rate. Furthermore, the secant stiffness of specimen
the inelastic deformation and ductility. The maximum load, Pm, of HFTSL-2 was lower than that of specimen HFTSL-3, which indicated
specimen HFTSL-3 in the push direction was 9.5% higher than that of that increasing the replaceable link length could reduce the lateral
specimen HFTSL-2 and was 10.6% higher in the pull direction. The stiffness of the specimen. The residual secant stiffnesses of the three
ductility ratio, μ, of specimen HFTSL-3 in the push direction was 13.3% specimens remained relatively similar at the end of the experiment,
lower than that of specimen HFTSL-2 and was 13.4% lower in the pull owing to the fact that the shear links were damaged and only the col­
direction. This indicated that reducing the length of the shear link could umns and deep beams remained to resist the lateral load.
improve the carrying capacity but would decrease the ductility capacity
of the specimen. 4.2.5. Strength and deformation capacity of the shear links
In the HSS-FTS-RSLs, the replaceable links acted as ductile fuses to
4.2.4. Stiffness evaluation sustain the inelastic shear deformation and dissipate the seismic energy
The secant stiffness, Kj, was used to evaluate the stiffness during a major earthquake. Therefore, the cyclic responses of the links

Table 3
Load-carrying and ductility capacities.
Specimen Load directions Yield point Peak point Ultimate point Pm/Py Ductility ratio, μ

Δy (mm) θy (%) Py (kN) Δm (mm) θm (%) Pm (kN) Δu (mm) θu (%) Pu (kN)

HFTSL-1 Push 24.81 1.00 325.84 61.87 3.33 414.29 74.21 3.57 415.29 1.27 2.99
Pull 24.72 0.99 323.84 61.92 3.33 406.79 74.26 3.45 398.05 1.26 3.01

HFTSL-2 Push 19.78 0.83 254.37 90.77 4.17 377.56 90.77 4.17 377.56 1.48 4.59
Pull 19.75 0.83 246.88 90.81 4.17 362.32 90.81 4.17 362.32 1.47 4.60

HFTSL-3 Push 19.89 0.83 278.11 79.13 3.85 413.29 79.13 3.85 413.29 1.49 3.98
Pull 19.85 0.83 276.86 69.33 3.45 400.79 79.18 3.85 389.81 1.45 3.99

11
H. Zhang, et al. Engineering Structures 223 (2020) 111172

were critical to the seismic performance of the overall structure and the Table 4
seismic design of the deep beams. The link shear, Vlink, was calculated Shear strength and deformation capacity of shear links.
from Eq. (5) using the force equilibrium. Specimen Loading Plastic Maximum Maximum Overstrength
directions shear shear plastic shear factor,
Vlink = Pactuator (D / L), (5)
strength, strength, deformation, Ω = Vmax/Vpa
where Pactuator denotes the load applied by the actuator, D denotes the Vpa (kN) Vmax (kN) γspu (rad)

perpendicular distance from the loading point of the actuator to the HFTSL-1 push 313.25 508.16 0.126 1.62
pinned point of the bottom column, and L denotes the horizontal dis­ pull −306.83 −498.26 −0.129 1.62
tance between the pinned points of the columns.
HFTSL-2 push 238.68 455.39 > 0.179 1.91
It should be noted that the P-delta effect caused by the vertical axial pull −234.64 −449.82 < −0.178 1.92
load on the top of the column influenced the Vlink under large dis­
HFTSL-3 push 241.86 505.77 0.223 2.09
placement loadings. However, the P-delta effect was difficult to quan­ pull −237.05 −477.20 −0.236 2.01
tify and could not be measured during the experiment. Thus, the in­
fluence of the P-delta effect on the link shear was not considered in this Note: For specimen HFTSL-2, the maximum plastic shear deformations, γspu, of
study. The link shear deformation, γs, was determined from Eq. (6) 0.179 rad and –0.178 rad in the push and pull directions corresponded to
suggested by McDaniel et al. [27]. Meanwhile, the plastic link shear 8Δy′ = 96 mm. The LVDTs D1 and D2 were moved after the 96 mm displace­
deformation, γsp, was calculated from Eqs. (7) and (8). It should be ment loading because the deformation measured was out of their measurement
range.
noted that the measured link shear deformation, γs, is different from the
link rotation defined by AISC 341-16 [20], which states that this
quantity is the relative vertical deformation of both ends divided by the defined as Ω = Vmax/Vpa.
The average overstrength factor of the shear links using Q235 steel
link length.
in specimen HFTSL-1 was 1.62, which was 8.0% higher than the value
1 (c 1 c2 ) a2 + b 2 of 1.5 recommended for links per ANSI/AISC 341-16 [20]. The average
= ,
s
2 ab (6) overstrength factors of the links using LYP225 steel for specimens
HFTSL-2 and HFTSL-3 were 1.91 and 2.05, respectively, which were
sp = s e, and (7) 27.3% and 36.7% higher than the recommended value of 1.5. This in­
dicated that the shear links made of LYP225 steel had a higher level of
e = Vpa /(e· KL ), (8) overstrength, with an overstrength factor, Ω, of more than 1.9. Similar
where c1 and c2 denote the deformation measured by LVDTs D1 and D2, findings in past tests were observed by Ji et al. [25]. Meanwhile, the
respectively, a and b denote the effective length and depth of the link shorter shear link, with a length ratio of 0.74, in specimen HFTSL-3
deformation region, respectively, γe is the elastic link shear deforma­ developed significantly higher overstrength (more than 2.0) than that
tion, and Vpa is the measured plastic shear strength as obtained from the of the shear link with a length ratio of 1.08 in specimen HFTSL-2. This
measured yield strength of the web material. KL denotes the elastic suggested that the overstrength factor of shear links with a smaller
shear stiffness and was calculated from the hysteretic curve of the link length ratio or made of LYP steel might be maximized appropriately in
shear, Vlink, versus the link shear deformation, γs. the seismic design of deep beams.
Fig. 24 shows the hysteretic responses of the link shear, Vlink, versus The average maximum plastic link shear deformations, γspu, of
the plastic link shear deformation, γsp, of the shear links of the speci­ specimens HFTSL-1, HFTSL-2, and HFTSL-3 were 0.128 rad, in excess of
mens. The shear links, especially in specimens HFTSL-2 and HFTSL-3, 0.178 rad, and 0.227 rad, respectively. It was concluded that the shear
showed steady and full hysteresis loops during the test, indicating that links, especially those that used LYP225 steel, exhibited a superior in­
the LYP225 steel shear link exhibited an excellent inelastic shear de­ elastic shear deformation capacity. Moreover, the γspu of the shear link
formation and hysteretic energy dissipation capacity. Furthermore, the using LYP225 steel was larger than that of the shear link using Q235
shear strength of the links increased gradually until the end of the cyclic steel. It is noteworthy that the shear link with a length ratio of 0.74 had
displacement loading, and the shear links exhibited an obvious over­ a larger inelastic shear deformation than those with a length ratio of
strength, particularly the LYP225 steel links, after the shear yielding in 1.08, which indicated that reducing the link length could relatively
both loading directions. increase the inelastic shear deformation capacity.
Table 4 summarizes the values of the link plastic shear strength, Vpa,
maximum shear strength, Vmax, maximum plastic shear deformation, 4.2.6. Plastic link shear deformation
γspu, and overstrength factor, Ω, of the specimens. The maximum plastic Fig. 25 shows the plastic link shear deformation versus interstory
shear deformation, γspu, was defined as the shear deformation sustained drift curves for each specimen during cyclic loading. It can be seen that
for at least one full cycle of loading prior to failure. The overstrength the plastic link shear deformation, γsp, of the specimens increased
factor, Ω, accounted for the increase in the shear link strength and was gradually with the interstory drift increment. The θ – γsp curves showed

Fig. 24. Hysteretic curves of the shear links in the Phase-II loading.

12
H. Zhang, et al. Engineering Structures 223 (2020) 111172

The strain distribution of the specimen beam’s upper flange is


shown in Fig. 26(e). It was apparent that the normalized strain, ε/εy, of
the beam’s upper flange nearly reached 0.8 until the end of the loading
cycle. This indicated that the beam flange might have entered the in­
elastic stage when the deep beams were made of CMS (e.g., Q345
(fy = 345 MPa)) under the same section dimension conditions and
resulted in weakening the restraint of the deep beam on the replaceable
link. Furthermore, it is noteworthy that the normalized strain, ε/εy, of
the deep beams and columns, and the equivalent strain, ie , of the panel
zones (see Fig. 26(g)) were less than 1.0 during the cyclic loading test,
which indicated that the stresses on the beams, columns, and panel
zones did not exceed their yield strength until the end of the test. It was
Fig. 25. Plastic link shear deformation versus interstory drift for the specimens.
concluded that the replaceable links entered the elastoplastic stage
while the other key components remained elastic under cyclic loading,
which indicated that the HSS-FTS-RSLs could reduce the residual drift,
a double-linear relationship. When the specimens remained in the quickly achieve a seismic rehabilitation after a major earthquake, sig­
elastic stage, the shear deformation of the links was relatively small, nificantly reduce the retrofit cost, and extend the structural life span.
and the slope of the curves was also small. With the increasing inters­
tory drift, the shear links experienced a significantly inelastic shear
deformation, and the slope of the plastic link shear deformation curves 4.3. Replaceability
increased gradually. Therefore, the change in the curve slopes indicated
that the shear links entered the inelastic stage under cyclic loading. To investigate the replaceability of the specimens, the links were
replaced after the Phase-I loading. First, the lateral force from the ac­
4.2.7. Strain responses tuator and the vertical axial force from the jack were unloaded to zero
The strain distributions of the shear links, deep beams, columns, and to remove the original link. It should be noted that the vertical axial
panel zones of the column webs were obtained by strain gauges glued force was removed to ensure safety during the link replacement.
onto the specimens. The strain distribution of specimen HFTSL-2 under Second, the residual displacement of the specimens was gradually de­
cyclic displacement loading is shown in Fig. 26; ε/εy denotes the nor­ creased to identify the acceptable residual interstory drift, θre, and the
malized strain, where ε and εy are the peak and yield strain values, corresponding residual link shear deformation, γre, which allowed for
respectively, according to the actual material properties. The flange an easy reinstallation of the new link. The replaceability of the links
strain distributions were along the upper flange centerline of the beam was evaluated using the replacement time T, θre, and γre. The replace­
and column, and the web strain distributions were along the web ment time, T, included the disassembly of the original link and in­
centerline of the beam and column. The equivalent strain, ie , was used stallation of the new link. The new links were 3 mm shorter than the
to analyze the strain responses of the shear link web and panel zone removed links to facilitate easy installation. A photograph of the link
using the measurements of the three-gauge rosettes, and was calculated replacement is shown in Fig. 27. Table 5 summarizes the T, θre, and γre
using Eqs. (9) and (10) [28]. values of each specimen. As shown in Table 5, the installation time of
the new links was longer than the disassembly time for the original
e 2 2 2 2, links. Specimen HFTSL-1 took 102 min for replacement, and specimens
= (1 2) +( 3) +(3 1) and
i
3
2
(9)
HFTSL-2 and HFTSL-3 took 95 and 85 min, respectively. This increase
1 in replacement efficiency was attributed to increased familiarity and
e 2 2 2,
= 2.78 + 2.78 0.44 0 90 + 1.5(2 45 0 90 ) proficiency with the process with each iteration. This indicated that, in
i
1.3 2
0 90
(10)
practice, proficiency in the disassembly and installation of the links
where ε1, ε2, and ε3 denote the first, second, and third principal strains might affect the efficiency of any post-earthquake retrofit. The residual
obtained by the three-gauge rosettes, respectively; and ε0, ε45, and ε90 interstory drifts, θre, of specimens HFTSL-1, HFTSL-2, and HFTSL-3
denote the 0°, 45°, and 90° strain values, respectively, obtained by were 0.42%, 0.43%, and 0.36%, respectively, and the residual link
three-gauge rosettes on the link web and panel zone. shear deformation, γre, of the specimens were 0.0084 rad, 0.0086 rad,
The strain distributions of the shear link’s web and upper flange for and 0.0108 rad, respectively. The θre of specimen HFTSL-3 was smaller
each specimen are shown in Fig. 26(a)–(b), respectively. The equivalent than that of specimens HFTSL-1 and HFTSL-2, whereas the γre of spe­
strain of the link web increased rapidly with the displacement loading cimen HFTSL-3 was higher than that of specimens HFTSL-1 and HFTSL-
increment. The strain values of the link web were higher than those of 2. This is due to the link length of specimen HFTSL-3 being much
the link flange for each displacement cycle, which indicated that the shorter than the links of specimens HFTSL-1 and HFTSL-2, and thus the
links of the specimens were dominated by shear yielding and exhibited γre of specimen HFTSL-3 was higher than that of specimens HFTSL-1
an excellent inelastic shear deformation capacity. The results of the and HFTSL-2 under the same residual interstory drift.
strain analysis of the shear link section by the strain gauges were
consistent with the design requirements of the replaceable links at the
mid-span of the deep beams. 4.4. Comparison of cyclic behavior in the Phase-I and -II loadings
The strain distributions of the specimen’s beam and column webs
are shown in Fig. 26(c)–(d), respectively. It is clear from the strain The skeleton curves of the specimens during the Phase-I and -II
distributions that the web strain values of the beams and columns re­ loadings are compared in Fig. 28 to investigate the influence of the
tained a linear strain profile, and the strain of the web centerline of the replacement links on the cyclic behavior of the specimens. The skeleton
beams and columns was almost zero during cyclic loading. This in­ curves of the specimens nearly coincided and had similar initial stiff­
dicated that the web strain distributions almost met the assumption of nesses during the Phase-I and -II loadings. In general, this is due to the
the plane section. It should be noted that in the sub-assemblage speci­ theoretical moment gradient under lateral loading being linear along
mens subjected to vertical axial loads, the compressive strain values of the length of the deep beam with a zero moment at the mid-span (see
the beam and column webs were larger than the tensile strains for each Fig. 1), which indicated that the replacement of the links had a limited
loading step. influence on the lateral stiffness of the specimens under cyclic loading.

13
H. Zhang, et al. Engineering Structures 223 (2020) 111172

Fig. 26. Section strain distributions of HFTSL-2.

Table 5
Replaceability of specimens.
Specimen Replacement time, T (min) Residual Residual link
interstory shear
Disassembly Installation Total drift, θre (%) deformation, γre
(rad)

HFTSL-1 34 68 102 0.42 0.0084


HFTSL-2 25 70 95 0.43 0.0086
HFTSL-3 20 65 85 0.36 0.0108

(C3D8R) solid continuum elements were used for all parts to increase
the simulation accuracy of the FE model. The structured meshing
Fig. 27. Replacement of the links. technique was adopted for all parts of the FE model to obtain a proper
element shape, especially for the round bolts. The finite element sizes of
5. Finite element analysis the shear links, end plates, and high-strength bolts were 10 mm,
10 mm, and 3 mm, respectively, owing to the concentration of the in­
5.1. Finite element modeling elastic deformation and damage in the shear links under cyclic loading.
The finite element sizes of the columns and deep beams were 40 mm.
Nonlinear 3D finite element (FE) models of the test specimens were Note that the mesh sizes for the members of the FE model can obtain the
developed by ABAQUS 6.14 [19]. The FE models consisted of columns, accuracy of the simulation results by mesh a refinement study. The
deep beams, shear links, end plates, cover plates, stiffeners, and high- mesh partition of the FE model is shown in Fig. 29. Material fracture
strength bolts. Three-dimensional, eight-node, reduced integration and tearing were not considered in the FE models. The boundary

14
H. Zhang, et al. Engineering Structures 223 (2020) 111172

The surface-to-surface contact interaction relationship between the


end plates and bolt nuts, shafts, and holes, as well as between the beam
and link end plates, were considered in the FE models. Normal “hard”
contact and tangential “Coulomb friction” contact was assigned to the
surface-to-surface interaction relationship. The deep beams and shear
links were rigidly connected to the end plates by a “Tie” constraint. The
deep beams were also connected to the columns using “Tie” constraints
to simulate welding connections. The friction coefficient of the contact
surface between the beam and link end plates was assumed to be 0.35
for the tests. A pre-tension in the middle face of the high-strength bolts
was applied using the “Bolt Load” command in the ABAQUS “Load”
module (see Fig. 30). The pre-tension value used was 155 kN in ac­
cordance with Chinese Standard GB 50017-2017 [12].

5.2. Comparison of hysteretic responses


Fig. 28. Skeleton curves of the specimens during Phase-I and -II loading.

The hysteretic responses of the FE models and test results are


conditions of the FE model were programmed in accordance with the compared in Fig. 31. Fig. 32 shows a comparison of the skeleton curves
actual boundary conditions of the test specimens. As shown in Fig. 29, a by the FE models and experimental results. It is clear from the com­
simple support was defined at each end of the bottom of the columns. parison of hysteretic and skeleton curves that the initial stiffness and
The out-of-plane translational degrees of freedom (DOFs) were con­ yield strength of the FE models match well with the experimental re­
strained at the upper and bottom flanges of the deep beams to prevent sults. However, for specimens HFTSL-1 and HFTSL-3, the numerical
lateral deformation and torsion of the FE models. A vertical con­ carrying capacity of the FE models was slightly higher than the tested
centrated force and lateral displacement-controlled cyclic loading si­ values under the large displacement cyclic loadings due to the material
milar to the Phase-II loading conditions of the test specimens were fracture or tearing that was not considered in the FE model. Overall, the
applied to the top of the columns. numerical simulations of the FE models were in good agreement with
Many of the previous numerical investigations used a multi-linear the experimental results. This indicated that the modeling approach in
mechanical model to define the plastic properties of steel [29–30]. For ABAQUS accurately predicted the cyclic behavior of the sub-structural
simplicity, the multi-linear kinematic hardening rule with the Von- specimens, especially in the elastic stage.
Mises yielding criterion was used for the columns, deep beams, and
high-strength bolts. The yield strength, fy, tensile strength, fu, and
elastic modulus, E, were obtained from the tensile coupon tests. Fur­ 5.3. Comparison of failure modes
thermore, the Chaboche cyclic constitutive model included a nonlinear
isotropic hardening component, and a kinematic hardening component The plastic equivalent strain (PEEQ) index was used to evaluate the
was used for the link steel to capture the inelastic cyclic responses of the local inelastic strain demand and predict the fracture tendency of steel
shear links in the FE models [31]. This plastic constitutive model of [34]. Fig. 33 shows a comparison of the failure modes by the FE models
metal was defined in “Cycle Hardening” of the “Property” module in and the test results. The link web yielding and local buckling of the link
ABAQUS [19]. The parameters of the cyclic constitutive model were flange that were observed in the test could be properly simulated by the
obtained by data fitting. The calibration parameters of the Q235 and FE model. Fig. 33 also presents the distribution of the PEEQ value of the
LYP225 steel are shown in Table 6 according to references [32] and FE models under cyclic loading. As shown in Fig. 33, except for the
[33]. Note that both the kinematic hardening and Chaboche con­ shear links, the other structural members remained elastic, and the
stitutive models can consider the Bauschinger effect of steel material cumulative plastic strain was concentrated only within the links.
under cyclic loading. Therefore, the predicted failure modes from the FE models were in good
agreement with the experimental results.

Fig. 29. FE model of test specimen.

15
H. Zhang, et al. Engineering Structures 223 (2020) 111172

Table 6
Calibration parameters of Q235 and LYP225 steel.
Material σ|0(N/mm2) Q∞(N/mm2) biso C1(N/mm2) γ1 C2(N/mm2) γ2 C3(N/mm2) γ3 C4(N/mm2) γ4

Q235 290 35 2.5 6000 200 5000 130 3000 35 990 30


LYP225 191 115 9 3041 126 1028 170 890 224 260 1

drawn within the limitations of the research:

(1) During the cyclic loading tests, the accumulated inelastic de­
formation and damage to the specimens were concentrated only
within the shear links, while the other structural members remained
elastic. This indicated that the HSS-FTS-RSLs might allow a quick
seismic rehabilitation through the replacement of the damaged
links, thereby reducing the retrofit cost and significantly extending
the structural life span.
(2) The specimens that utilized bolted end-plate connections between
the replaceable link and the deep beam exhibited stable and in­
creasing hysteretic loops, with a displacement ductility ratio, μ, that
ranged from 2.99 to 4.60. This indicated that the sub-assemblage
Fig. 30. Pre-tension of the high-strength bolts. specimens had good inelastic deformation and high ductility ca­
pacity under cyclic loading.
6. Conclusions (3) The shear links had full hysteretic loops, especially the links made
of LYP225 steel. The maximum plastic link shear deformations
A high-strength steel framed-tube structure with replaceable shear exceeded 0.12 rad. This indicated that the links exhibited an ex­
links (HSS-FTS-RSLs) was proposed to improve the seismic performance cellent inelastic shear deformation and ductility capacity under
of SFTSs through the excellent ductility and inelastic deformation ca­ cyclic loading. The overstrength of the shear links ranged from 1.62
pacity of the shear links as ductile fuses at the mid-span of the deep to 2.09. For the very short shear link, which was fabricated from
beams. Quasi-static cyclic tests were conducted to investigate the cyclic LYP225 steel, the overstrength factor was greater than 2.0.
behaviors of three two-thirds-scale sub-assemblage specimens. (4) The acceptable residual interstory drift, θre, and residual link shear
Meanwhile, nonlinear FE models were established to capture and verify deformation, γre, of the specimens exceeded 0.36% and 0.0084 rad,
the observed experimental responses. The following conclusions were respectively. The replacement of the links had a limited effect on

Fig. 31. Comparison of the experimental and simulated hysteretic curves.

Fig. 32. Comparison of the experimental and simulated skeleton curves.

16
H. Zhang, et al. Engineering Structures 223 (2020) 111172

Fig. 33. Comparison of the experimental and simulated failure modes.

the lateral stiffness of the specimens. doi.org/10.1016/j.engstruct.2020.111172.


(5) Reducing the replaceable link length could enhance the load-car­
rying capacity, initial stiffness, inelastic deformation capacity, and References
overstrength of the link, while the ductility and energy dissipation
capacity of the specimen could decrease. The replaceable link, using [1] Taranath BS. Structural analysis and design of tall buildings. steel and composite
various grades of steel, had an obvious effect on the carrying ca­ construction. Boca Raton, FL, USA: CRC Press; 2016.
[2] Lian M, Zhang H, Cheng Q, Su M. Finite element analysis for the seismic perfor­
pacity, ductility, energy dissipation capacity, inelastic deformation mance of steel frame-tube structures with replaceable shear links. Steel Compos
capacity, and overstrength of the link. Struct 2019;30(4):365–82.
(6) Nonlinear FE models of each specimen were developed using [3] Wang F, Su M, Hong M, Guo Y, Li S. Cyclic behaviour of Y-shaped eccentrically
braced frames fabricated with high-strength steel composite. J Constr Steel Res
ABAQUS 6.14. The numerical simulations of the FE models were in 2016;120:176–87.
good agreement with the cyclic responses of the test results. The FE [4] Lian M, Su M. Seismic performance of high-strength steel fabricated eccentrically
model may be used to further investigate the effects of various braced frame with vertical shear link. J Constr Steel Res 2017;137:262–85.
[5] Lian M, Su M. Experimental study and simplified analysis of EBF fabricated with
parameters on the seismic behavior of the sub-structures. high strength steel. J Constr Steel Res 2017;139:6–17.
[6] Rahnavard R, Hassanipour A, Suleiman M, Mokhtari A. Evaluation on eccentrically
CRediT authorship contribution statement braced frame with single and double shear panels. J Build Eng 2017;10:13–25.
[7] Shen Y, Christopoulos C, Mansour N, Tremblay R. Seismic design and performance
of steel moment-resisting frames with nonlinear replaceable links. J Struct Eng
Hao Zhang: Conceptualization, Methodology, Software, Validation. 2011;137(10):1107–17.
Mingzhou Su: Investigation, Methodology, Resources, Writing - review [8] Mansour N, Christopoulos C, Tremblay R. Experimental validation of replaceable
& editing, Writing - original draft. Ming Lian: Funding acquisition, shear links for eccentrically braced steel frames. J Struct Eng
2011;137(10):1141–52.
Resources, Data curation, Project administration. Qianqian Cheng: [9] AISC. Specification for Structural Steel Buildings (ANSI/AISC 360-16). American
Software, Validation. Binlin Guan: Visualization, Software. Huanxue Institute of Steel Construction; Chicago, IL, USA; 2016.
Gong: Investigation. [10] Shi G, Wang M, Bai Y, Wang F, Shi Y, Wang Y. Experimental and modeling study of
high-strength structural steel under cyclic loading. Eng Struct 2012;37:1–13.
[11] Ban H, Shi G. A review of research on high-strength steel structures. P I Civil Eng-Str
Declaration of Competing Interest B 2018;171(8):625–41.
[12] Ministry of Housing and Urban-Rural Development of the People’s Republic of
China. GB 50017-2017 Standard for Design of Steel Structure. Architecture and
The authors declare that they have no known competing financial Building Press, Beijing, China; 2017 [in Chinese].
interests or personal relationships that could have appeared to influ­ [13] Nikoukalam MT, Dolatshahi KM. Development of structural shear fuse in moment
resisting frames. J Constr Steel Res 2015;114:349–61.
ence the work reported in this paper. [14] Rahnavard R, Hassanipour A, Siahpolo N. Analytical study on new types of reduced
beam section moment connections affecting cyclic behavior. Case Stud Struct Eng
Acknowledgements 2015;3:33–51.
[15] Hassanipour A, Rahnavard R, Mokhtari A, Rahnavard N. Numerical investigation on
reduces web beam section moment connections under the effects on cyclic loading.
This work was supported by the National Natural Science J Multidiscip Eng Sci Technol (JMEST) 2015;2(8):2054–61.
Foundation of China (Grant No. 51708444) and Shaanxi Natural [16] Mahmoudi F, Dolatshahi KM, Mahsuli M, Shahmohammadi A, Nikoukalam MT.
Experimental evaluation of steel moment resisting frames with a nonlinear shear
Science Foundation (Grant No. 2018JQ5074).
fuse. Proceedings of the Geotechnical and Structural Engineering Congress,
Phoenix, Arizona, USA; 2016.
Appendix A. Supplementary material [17] Guan B, Su M, Lian M. Seismic behaviour of combined steel framed-tube sub­
structure with replaceable shear links. J Constr Steel Res 2020;167:105968.
[18] Cheng Q, Su M, Lian M, Zhang H, Guan B, Gong H. Cyclic behavior of high-strength
Supplementary data to this article can be found online at https://

17
H. Zhang, et al. Engineering Structures 223 (2020) 111172

steel framed-tube structures with bolted replaceable shear links. Eng Struct Struct 2015;123:312–24.
2020;210:110395. [27] McDaniel CC, Uang CM, Seible F. Cyclic testing of built-up steel shear links for the
[19] ABAQUS. Analysis user’s manual I_V. Version 6.14. USA: ABAQUS, Inc., Dassault new bay bridge. J Struct Eng 2003;129(6):801–9.
Systèmes; 2014. [28] Wang C, Chen Y, Chen X, Chen D. Experimental and numerical research on out-of-
[20] AISC. Seismic Provisions for Structural Steel Buildings (ANSI/AISC 341-16). plane flexural property of plates reinforced SHS X-joints. Thin-Walled Struct
American Institute of Steel Construction, Chicago, IL, USA; 2016. 2015;94:466–77.
[21] Ministry of Housing and Urban-Rural Development of the People’s Republic of [29] Rahnavard R, Naghavi M, Aboudi M, Suleiman M. Investigating modeling ap­
China. GB 50011-2010 Code for Seismic Design of Buildings. Architecture and proaches of buckling-restrained braces under cyclic loads. Case Stud Constr Mater
Building Press, Beijing, China; 2010 [in Chinese]. 2018;8:476–88.
[22] Ministry of Housing and Urban-Rural Development of the People’s Republic of [30] Naghavi M, Rahnavard R, Thomas RJ, Malekinejad M. Numerical evaluation of the
China. JGJ 99-2015 Technical Specification for Steel Structure of Tall Buildings. hysteretic behavior of concentrically braced frames and buckling restrained brace
Architecture and Building Press, Beijing, China; 2015 [in Chinese]. frame systems. J Build Eng 2019;22:415–28.
[23] Murray TM, Summer EA. Extended End-Plate Moment Connection—Seismic and [31] Chaboche JL. Constitutive equations for cyclic plasticity and cyclic viscoplasticity.
Wind Applications, Steel Design Guide Series 4, (2nd Edition), American Institute of Int J Plast 1989;5(3):247–302.
Steel Construction, Chicago, USA; 2003. [32] Shi Y, Wang M, Wang Y. Experimental and constitutive model study of structural
[24] Mcdaniel CC, Uang CM, Seible F. Cyclic testing of built-up steel shear links for the steel under cyclic loading. J Constr Steel Res 2011;67(8):1185–97.
new bay bridge. J Struct Eng 2003;129:801–9. [33] Shi G, Gao Y, Wang X, Zhang Y. Mechanical properties and constitutive models of
[25] Ji X, Wang Y, Ma Q, Okazaki T. Cyclic behavior of very short steel shear links. J low yield point steels. Constr Build Mater 2018;175:570–87.
Struct Eng 2016;142(2):04015114. [34] Wang M, Shi Y, Wang Y, Shi G. Numerical study on seismic behaviors of steel frame
[26] Feng P, Cheng S, Bai Y, Ye L. Mechanical behavior of concrete-filled square steel endplate connections. J Constr Steel Res 2013;90:140–52.
tube with FRP-confined concrete core subjected to axial compression. Compos

18

You might also like