Journal of Constructional Steel Research 143 (2018) 250–263

Contents lists available at ScienceDirect

Journal of Constructional Steel Research

Seismic behavior of K-shaped eccentrically braced frames with

high-strength steel: Shaking table testing and FEM analysis
Xiaohong Tian a,b, Mingzhou Su a,⁎, Ming Lian a, Feng Wang c, Shen Li d
a
School of Civil Engineering, Xi'an University of Architecture and Technology, Xi'an 710055, PR China
b
School of Mechanical and Precision Instrument Engineering, Xi'an University of Technology, Xi'an 710048, PR China
c
School of Civil Engineering, Lanzhou University of Technology, Lanzhou 730050, PR China
d
School of Civil Engineering and Architecture, Xi'an University of Technology, Xi'an 710048, PR China

a r t i c l e i n f o a b s t r a c t

Article history: A shaking table test was carried out on 1:2 scale three-storey K-shaped eccentrically braced frames with high-
Received 30 June 2017 strength steel (K-HSS-EBFs). The acceleration, displacement, and strain responses of the specimen under differ-
Received in revised form 29 December 2017 ent peak acceleration ground motions were measured and analyzed. The natural frequency, acceleration ampli-
Accepted 31 December 2017
fication factor, storey drift, and distribution of horizontal seismic actions for the model structure were
Available online xxxx
investigated. Finite element models were established for elastoplastic time history analysis on the model struc-
Keywords:
ture. Finite element analysis showed that the links were the weakest part of the structure, especially the webs
EBFs of the links and welds near the connections between the links and frame beams. The results showed that K-
Dynamic analysis HSS-EBFs may absorb excessive seismic energy, enter the elastoplastic state, and even fail under strong earth-
High-strength steel quakes owing to shear deformation of the links during vibration. The PBPD method is found to be suitable for
Shake table test use in the design of K-HSS-EBFs. In addition, the K-HSS-EBFs can fulfill the seismic requirements under limiting
Seismic performance state. These findings will be helpful for the future design of K-HSS-EBFs.
© 2018 Elsevier Ltd. All rights reserved.

1. Introduction
High-strength steel (HSS; fy ≥460 MPa) has higher yield and tensile
strength than conventional steel (fyb 345 MPa). In recent years, scholars
and professionals in the steel structure field of many countries have
completed a large number of experiments and a considerable amount
of theoretical research on HSS. Research on HSS has focused on the ma-
terial properties and mechanic behavior of members. Green et al [1].
completed a bearing capacity test for HSS I-beams and discussed the ef-
fect of the cross-section, loading form, and material properties on the
rotation capacity. Azizinamimi and Barth [2] studied the flexural capac-
ity, tensile ductility, and shear capacity of HSS (fy = 485 MPa) in bridge
construction. Some researchers [3–11] have studied the overall buckling
behavior and local buckling response of HSS columns with welded box
and I-beam cross-sections, the residual stress distribution in the HSS
weld section, and the hysteretic behavior of HSS members under cyclic
loading. In recent years, the development of steel structures and im-
provement of production technology have contributed to the use of
HSS in bridge structures [12,13]. HSS has been gradually applied in
building structures. Under the same stress conditions, the cross-
sections of HSS members tend to be smaller than those of conventional
members, which increases the available space of the building while
⁎ Corresponding author.
E-mail address: sumingzhou@xauat.edu.cn (M. Su).
decreasing material costs. Therefore, their use has considerable eco-
nomic benefits [14]. However, the improved strength increases the
ratio of the yield strength to tensile strength of HSS, which reduces
the plastic deformation capacity. The requirements for steel use in seis-
mic zones are clearly listed in the mandatory provisions of Chinese Code
for Seismic Design of Buildings [15]: the ratio of the measured yield
strength to the measured tensile strength should not exceed 0.85, and
the steel should have an obvious yield plateau with an elongation
b20%. Q460 and higher-grade HSS often fail to meet the above seismic
requirements. These provisions limit the application of HSS in construc-
tion field of China. However, the structures can meet the design concept
of seismic codes by using reasonable structural type and structure de-
tails design. Eccentrically braced frames (EBFs) with HSS are proposed
to solve problem.
These structures can satisfactorily solve the inelastic deformation
and energy dissipation problem [16,17], where seismic energy is dissi-
pated through the elasto-plastic deformation of links. Other members
(e.g., beams, columns, etc.) need to be connected through links to pro-
vide sufficient restraints to maintain elasticity and small deformation.
Thus, HSS can be used for these members. EBFs for which the links
(bold parts in Fig. 1) are made from conventional steel and the rest of
the members are made of HSS are called eccentrically braced frames
in combination with high-strength steel (HSS-EBFs) [18].
At present, there have been few studies on the mechanical proper-
ties and seismic performance of HSS-EBFs. Dubina et al. [19] performed

https://doi.org/10.1016/j.jcsr.2017.12.030
0143-974X/© 2018 Elsevier Ltd. All rights reserved.
 X. Tian et al. / Journal of Constructional Steel Research 143 (2018) 250–263 251

numerical analysis were performed on K-HSS-EBFs. Results of this

analysis reveal that K-HSS-EBFs are capable of fulfilling seismic require-
ments under the limiting state and are more economical when com-
pared against traditional EBFs. In addition, the finite-element model
established in this paper is valid for analyzing K-HSS-EBFs, and the
PBPD method is found to be suitable for use in the design of K-HSS-EBFs.

2. Test overview

The prototype structure was designed according to Chinese current

design codes [14,26,27]: it is a three-storey HSS-EBF structure with
shear links. The plan size of the structure is 5.65 m × 5.65 m, and the
height of the structure is 3.6 m.
Fig. 1. Typical HSS-EBFs.
2.1. Similarity relationship
pseudo-static tests on four one-bay and single-storey K-HSS-EBFs. Their
test results showed that, under the effect of cyclic loading, the link made One of the key tasks of model testing is to design a specimen accord-
of conventional steel is completely in the plastic stage, and beam- ing to the requirements of the similarity theory so that the working con-
column frames made by HSS are still in the elastic stage. The authors' re- ditions are similar to the prototype structure. Based on the lifting
search group [20–22] performed monotonic loading tests, cyclic loading capacity of the shaking table and experimental site conditions of the
tests, and finite element analysis on HSS-EBF specimens. The results in- Key Lab of Structure and Earthquake Resistance of Xi'an University of
dicated that HSS-EBFs have good load-carrying, ductility, and energy- Architecture and Technology, the length scaling factor Sl was set to
dissipation capacities. 1/2. The stress scaling factor Sσ was set as 1 because the specimen and
To the authors' knowledge, no relevant information about shaking prototype structure were made of the same material. The acceleration
table tests on K-HSS-EBFs has been reported. Systematic experimental scaling factor was determined to be 1.2 according to the peak ground ac-
research and numerical analysis of the K-HSS-EBFs has theoretical and celeration (PGA) at the maximum level and laboratory conditions. Other
practical significance for the application of HSS in the construction typical scaling factors were calculated according to similarity theory
field [23–25]. In order to further investigate dynamic characteristics [28], as listed in Table 1.
and seismic performance of K-HSS-EBFs, the shaking table test and
2.2. Model design
Table 1
Typical scaling factors and similarity relationships of the test model. As shown in Fig. 2, the total height of the model was 5.4 m, and the
plan of the model was 2.85 m × 2.85 m. The thickness of the two-way
Physical parameters Similarity relation Ratio of similarity
reinforced concrete slab was 80 mm. The grade, diameter, and spacing
Length Sl 0.5 of the reinforcement were HRB335, 8 mm and 200 mm, respectively.
Stress Sσ 1.0
To ensure a reliable connection between the cast-in-place concrete
Acceleration Sa 1.2
Elastic modulus SE = Sσ 1.0 slab and beams, shear studs (Fig. 2(a)) were arranged every 100 mm
Density Sρ = Sσ/(Sa ⋅ Sl) 1.6667 at the center of the top flange of the beams. However, no shear studs
Mass Sm = Sσ ⋅ S2l /Sa 0.2083 were provided between the link and storey. The links and beams were
Period ST = S1/2
/ ⋅ S−1/2
a 0.6455 welded together to ensure that the connection had superior force-
Force SF = Sσ ⋅ S2l 0.25
bearing performance. According to the density similarity relationship,

Fig. 2. Test model (unit: mm).

252 X. Tian et al. / Journal of Constructional Steel Research 143 (2018) 250–263

Table 2 Table 4
The enhancement factors of EBFs. Mechanical properties of steel.

Members Columns Beams Braces Steel Q345B Q345B Q460C Q460C

Enhancement factors 1.2 1.2 1.3 Thickness t (mm) 6 10 8 10

Yield strength fy (MPa) 414.7 363.8 473.5 516.0
Ultimate strength fu (MPa) 542.0 545.8 635.1 692.0
Elastic modulus E (×105 MPa) 2.11 2.01 2.12 2.06
Table 3 Elongation (%) 28.29 28.74 25.36 23.51
Sectional dimensions and steel types of the test model.

Member Section (h × bf × tw × tf) Steel type

Beam H140 × 100 × 8 × 10 Q460C

obtained by using white noise excitation. The shaking table test loading
Column H145 × 145 × 8 × 10 Q460C plan and actual input PGAs of the test program are listed in Table 5.
Brace H100 × 100 × 6 × 10 Q345B
Link H140 × 100 × 6 × 10 Q345B 2.4. Instrumentation

In the test, 19 accelerometers and 8 displacement meters were ar-

gravity compensation and equivalent loads needed to be further applied
to the experimental structure. As shown in Fig. 2(c), the concrete blocks
were poured and then securely connected to the reinforced concrete
slab through steel bars. The bottom of each column was welded to the
steel plate, while the steel plate and table-board were connected with
high-strength bolts (Fig. 2(c)). The total weight of the model was
22.63 t; this included the HSS-EBF structure of 7.52 t, counterweight
of 14.35 t, and column soleplate weight of 0.76 t.
The base shear force of the specimen was calculated using the base
shear method. The length of a link was 350 mm (e = 0.964 Mp/Vp),
which was short link according to the classification put forward by
[29]: short link for e ≤ 1.6 Mp/Vp, where e, Vp and Mp were the length,
plastic shear force and plastic bending moment of the links, respectively.
For the non-dissipative members, the method of enhancement factors of
internal force was used to increase the cross-sections to ensure that the
links yielded firstly. That is to say, when the total cross-section of the
links entered into plastic, the internal forces of the frame beams, col-
umns and braces were multiplied by the enhancement factors. The en-
hancement factors are related to the seismic level, are selected
according to GB50011–2010 [15], and listed in Table 2.
Table 3 gives the sectional dimensions and steel types of the mem-
bers used in the test model, where “H” refers to the welded H-section
(Fig. 3) and the numbers are the dimensions in millimeters. Table 4
lists the mechanical properties of the steel used in the test.
2.3. Loading case
The ground motions inputted to the table were determined according
to the seismic design requirements and similarity theory of the model.
Based on factors such as the site category and dynamic characteristics
of the prototype structure, three seismic waves were chosen as input ex-
citations: the El Centro wave, the Taft wave, and an artificial seismic
wave (Lanzhou wave). Fig. 4 compares the acceleration response spectra
of the three seismic waves with the standard acceleration response spec-
trum put forward by GB 50011-2010 [15]. Ground motion records were
loaded in ascending order according to PGA. The dynamic characteristics
including the natural frequencies and damping ratios of the model were
ranged on the test model as shown in Fig. 5, where the sensor designat-
ed with D and A are the displacement meters and accelerometers,
respectively. X represents the direction of the sensors along the x-axis;
the first number is the storey where the sensors are located, and the sec-
ond number is the serial number of the sensor. In total, 24 strain gauges
were placed on the test model to measure the deformation of the links,
as shown in Fig. 6. The red letters represent the sites of strain gauges,
and the black numbers represent the serial numbers of strain gauges.
3. Test results and analysis
3.1. Test phenomena
During the test, damage to the test model mainly occurred
on the concrete slabs and links. Throughout the testing process,
no obvious signs of damage such as local bending, buckling, or weld
tearing appeared. Table 6 lists the experimental phenomena,
which indicated that the maximum deformation occurred on the
links during the loading process. This was consistent with the design
purpose.
3.2. Dynamic characteristics
The stochastic subspace identification analysis method [30,31],
which is based on the acceleration responses of each storey, was used
to obtain the damping ratios and natural frequencies of the model struc-
ture. Table 7 presents the tested damping ratios and natural frequencies.
The first natural frequency and damping ratio of the specimen were
7.20 Hz and 3.21%, respectively. The natural frequency decreased with
increasing PGA, while the damping ratio increased. This indicates that

Fig. 3. Specimen sectional dimensions. Fig. 4. Acceleration response spectra.

 X. Tian et al. / Journal of Constructional Steel Research 143 (2018) 250–263 253

Table 5 the specimen was damaged. This may be because the cracks in the
Test loading plan. concrete slabs reduced the structural rigidity. The natural frequency de-
pffiffiffiffiffiffiffiffiffi
Case Seismic Input PGA Case Seismic Input PGA creased by 2.6%. According to f ¼ k=m=ð2πÞ the stiffness is propor-
excitation (g) excitation (g) tional to the square of the natural frequency; a change in the natural
WN1 White noise 0.050 WN5 White noise 0.050 frequency reflects a change in stiffness. The stiffness degradation ξ is de-
El084 El Centro 0.084 El264 El Centro 0.264 fined as follows:
Taft084 Taft 0.084 Taft264 Taft 0.264
Lz084 Lanzhou 0.084 Lz264 Lanzhou 0.264
2 2
WN2 White noise 0.050 WN6 White noise 0.050 k−k0 f −f 0
El120 El Centro 0.120 El480 El Centro 0.480 ξ¼ ¼ 2
ð1Þ
k0 f0
Taft120 Taft 0.120 Taft480 Taft 0.480
Lz120 Lanzhou 0.120 Lz480 Lanzhou 0.480
WN3 White noise 0.050 WN7 White noise 0.050 where k represents the initial stiffness of the model, k0 represents the
El168 El Centro 0.168 El744 El Centro 0.744
Taft168 Taft 0.168 Taft744 Taft 0.744
stiffness during the test, f0 represents the initial frequency of the
Lz168 Lanzhou 0.168 Lz744 Lanzhou 0.744 model, and f represents frequency during the test.
WN4 White noise 0.050 WN8 White noise 0.050 Fig. 8 shows the stiffness degradation of the test model calculated
El240 El Centro 0.240 El1000 El Centro 1.000 with Eq. (1). The maximum decrease in the specimen stiffness was
Taft240 Taft 0.240 Taft1000 Taft 1.000
only 4%. A slight change in stiffness indicates that the specimen was
Lz240 Lanzhou 0.240 Lz1000 Lanzhou 1.000
WN9 White noise 0.050 slightly damaged and nearly remained elastic after a strong earthquake.
This is consistent with the experimental observations. The internal force

B
$; $;
B $; $;

< <

; ;
'; $; ';

1
A $; $;
'; ';
$; $; A
1 2 1 2

(a) First storey (b) Second storey

B B
$; $; $; $;

< <

; ;
'; $; '; '; $; ';

$; $; $; $;

A A
1 2 1 2

(c) Third storey (d) Roof

Fig. 5. Layout of accelerometers and displacement meters.
254 X. Tian et al. / Journal of Constructional Steel Research 143 (2018) 250–263

(a) First storey (b) Second storey

(c) Third storey (d) Photo of strain gauges

Fig. 6. Layout of strain gauges.

(a) Peeling of the link beam (b) Crack of the first storey slab

(c) Partial crushing of the storey slab (d) Stress traces on the links (e) Storey slab cracking
Fig. 7. Test phenomena.
 X. Tian et al. / Journal of Constructional Steel Research 143 (2018) 250–263 255

Table 6
Experimental phenomena.

Case PGA (g) Description

El168-Lz168 0.168 There was no obvious phenomenon in this stage.

El240-Lz240 0.240 Peeling occurred at the links on the first storey of the north frame, as shown in Fig. 7 (a).
El480-Lz480 0.480 Cracks in slab occurred on the storey of the first storey at the top of the links, as shown in Fig. 7 (b).
Cracks in slab were also found on the south of the second storey, part of which was crushed, as shown in Fig. 7 (c).
El744-Lz744 0.744 Storey cracks occurred on the upper right of the links of the south frame.
The storey crack on the north frame of the second storey became longer.
The bolts next to the column foot at the west side of the south frame became loosening.
The east web and bottom corner at the inner side of the links on the first storey peeled.
Stress traces appeared on the links on the north side of the first storey, as shown in Fig. 7 (d).
El1000-Lz1000 1.000 The slab of the first storey cracked, as shown in Fig. 7 (e).
The welds of the south frame of the second storey peeled.
The lower flange and brace welds of the north frame peeled.

of the structure is relatively small and the stiffness is found to decrease a

3

Acceleration amplification
little. This may be due to the specimen being oversized.

3.3. Acceleration response 2

factor /β
The acceleration responses under the actions of the Taft wave, El
Centro wave, and Lanzhou wave at point A1X1 (on the top surface of 1
the shaking table and near the column) were used as a reference. Accel-
eration amplification factors of each storey were defined as the ratios of First storey Second storey Third storey
the acceleration response at points A2X1, A3X1, and A4X1 to the refer- 0
ence value. Fig. 9 shows the envelopes of acceleration amplification fac- 0.0 0.2 0.4 0.6 0.8 1.0
tors under different earthquake levels. The acceleration amplification Peak ground acceleration /g
factors were found to increase with the height but decrease with in-
creasing PGA.
(a) El Centro wave

Table 7 3
Acceleration amplification

Damping ratio and natural frequency of the model.

PGA El Centro wave Taft wave

(g) 2
factor /β

Frequency Damping ratio Frequency Damping ratio

(Hz) (%) (Hz) (%)

0.084 7.20 3.51 7.20 2.90

0.120 7.16 3.38 7.17 2.81
1
0.168 7.16 3.48 7.15 2.81
0.240 7.12 3.35 7.13 2.81 First storey Second storey Third storey
0.264 7.11 3.34 7.13 2.85 0
0.480 7.06 3.80 7.10 3.40 0.0 0.2 0.4 0.6 0.8 1.0
0.744 7.05 4.08 7.08 3.43
1.000 7.01 4.10 7.10 3.79 Peak ground acceleration /g

(b) Taft wave

0
Taft wave
St if f n ess d egra d a t io n /%

3
Acceleration amplification

-1
El Centro wave
-2
2
factor /β

-3

-4 1

-5 First storey Second storey Third storey

0
-6 0.0 0.2 0.4 0.6 0.8 1.0
0 0.2 0.4 0.6 0.8 1 Peak ground acceleration /g

PGA /g (c) Lanzhou wave

Fig. 8. Structural stiffness degradation. Fig. 9. Acceleration response.

256 X. Tian et al. / Journal of Constructional Steel Research 143 (2018) 250–263

Strain/με
3.4. Displacement response
900
measure point 1
Fig. 10 shows the maximum horizontal displacement of each storey 800
700 measure point 2
relative to the base under different earthquake levels. The specimen had 600 measure point 3
basically the same displacement response to the three ground accelera- 500 measure point 4
tion records: the displacement of each storey of the specimen increased 400 measure point 5
with the input PGA. By comparing the displacements under each condi- 300
measure point 6
200
tion, it is found that the maximum displacement of the specimen un- measure point 7
100
dergoes a slight change prior to PGA attaining a value of 0.48 g. When 0 measure point 8
the PGA reached 0.480 g, the maximum displacement varied significant- 0.0 0.2 0.4 0.6 0.8 1.0
ly among different stories, and the second storey showed the greatest
Peak ground acceleration /g
change. When the PGA reached 1.000 g under the El Centro wave, the
displacement of the first storey was 7.61 mm and accounted for 69%
of the total displacement of the roof. (a) Strain response in the link of the first storey

Strain /με
600 measure point 9
500 measure point 10
400 measure point 11
3 0.084g 300
measure point 12
0.120g measure point 13
200
0.168g measure point 14
2 100
Storey /n

measure point 15
0.240g 0
measure point 16
0.264g 0.0 0.2 0.4 0.6 0.8 1.0
1 0.480g
Peak ground acceleration /g
0.744g
1.000g (b) Strain response in the link of the second storey
0
0 5 10 15 20 25
Strain /με
Displacement / mm 350 measure point 17
300 measure point 18
250 measure point 19
(a) El Centro wave
200 measure point 20
150 measure point 21
3 0.084g
100 measure point 22
0.120g measure point 23
50
0.168g measure point 24
2 0
Storey / n

0.240g 0.0 0.2 0.4 0.6 0.8 1.0

0.264g
Peak ground acceleration /g
1 0.480g
0.744g
1.000g (c) Strain response in the link of the third storey
0
0 5 10 15 20 Fig. 11. Variation in strain under the Taft wave.

Displacement / mm
3.5. Strain of links
(b) Taft wave
Fig. 11 shows the strain data at all measuring points for the link of
the model test under the excitation of the Taft wave. These results do
3 0.084g not include the strain under the effect of the representative value for
0.120g the gravity loading. The strain clearly increased with the PGA. In
0.168g
2
Storey / n

0.240g
0.264g
1 0.480g
0.744g
1.000g
0
0 5 10 15
Displacement / mm

(c) Lanzhou wave

Fig. 10. Displacement response. Fig. 12. Principal strain.

 X. Tian et al. / Journal of Constructional Steel Research 143 (2018) 250–263 257

Table 8 reached 0.480 g, the strain values at each point steadily increased.
Maximum inter-storey drift angle. After the PGA reached 0.744 g, the increase in strain slowed down.
PGA (g) 0.084 0.120 0.168 0.240 0.264 0.480 0.744 1.000 As shown in Fig. 12, εx, εy, and εuwere the measured values of the
3rd storey 1/3956 1/2723 1/1887 1/1323 1/1198 1/667 1/437 1/352
strains in the web of the link; σmax is the maximum principal stress,
2nd storey 1/3237 1/2990 1/2195 1/1556 1/1433 1/750 1/497 1/362 and α represents the angle between σmax and the x-axis. According to
1st storey 1/2866 1/2410 1/1667 1/1134 1/1007 1/519 1/307 1/237 Eq. (2), the maximum principal stress was 133 MPa. Including the stress
under the representative value of gravity loading, which is calculated by
using the finite element method, the maximum principal stress was
3 about 134.5 MPa, which was less than the yield stress (fy =
414.7 MPa) of the link. This revealed that the specimen was still in the
elastic stage. This phenomenon can be attributed to two factors: the sto-
rey slab still strongly supported the links, and the actual earthquake in-
Story /n

tensity withstood by the three-storey model structure was relatively

2 El Centro wave
small.
LZ wave
2 rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2 i3
Taft wave h 
2 ðεx −ε u Þ2 þ εu −ε y 7
Calculate value E6 ε þ ε
¼ 6 7
x y
σ max þ ð2Þ
1 2 4 1−μ 1þμ 5
0 20 40 60 80
Comparison of horizontal seismic force /kN

Fig. 13. Comparison of horizontal seismic actions.

3.6. Structural storey drift

addition, under the same earthquake action, the links on the first storey
reached the maximum strain; the ones on the second storey had less
strain; the links on the third storey had the least strain. For example,
when the PGA was 0.744 g, the strain values of the links on the first,
second, and third stories were 680, 482,and 252 με, respectively,
which indicates that the links on the first storey were the weakest
parts in the structure. For the same link, the measured strain in the
web was larger compared to that in the flange, which indicates that
the links were mainly subjected to shear deformation. Before the PGA
The storey drift is an important indicator for determining the earth-
quake resistance. Table 8 presents the maximum storey drift calculated
according to the test results. After the minor earthquake, the storey drift
had a maximum value of 1/1667, which is b1/250 of the current seismic
code requirement. This indicates that the structure met the seismic for-
tification goal of “no damage under minor earthquakes.” After the
strong earthquakes, the maximum storey drift was 1/237, which sat-
isfies the requirement in the current seismic code that the storey
elastoplastic drift should not exceed 1/50. This indicates that the

(a) Whole model (b) Local model

Fig. 14. Finite element model.
258 X. Tian et al. / Journal of Constructional Steel Research 143 (2018) 250–263

Fig. 15. Comparison of acceleration responses at tracking point A4X1.

structure meets the seismic fortification goal of “no collapsing under
strong earthquakes.”
3.7. Distribution of structural horizontal seismic action
Centro wave was 13% higher than the value calculated from the base
shear method, whereas the average measured value with the three
ground acceleration records was 6% lower than the average calculated
value. This indicates that the base shear method can be used to calculate
the horizontal seismic action of multilayer K-HSS-EBFs.

The inertia force time history can be calculated by multiplying the

mass focused on the given storey and the acceleration time history at
the corresponding point. The maximum inertia force of each storey rep-
resents the horizontal seismic action of the structure under different
levels of earthquake. Fig. 13 shows the calculation results for the hori-
zontal seismic actions of the structure under frequent earthquakes; an
inverted triangular distribution pattern was observed. When the PGA
increased to 0.084 g, the measured experimental result with the El
4. FEM analysis and verification
4.1. Numerical model
To further study the failure modes and seismic performance of the
structure, a three-dimensional finite element model (FEM) of the
specimen was modeled. In the FEM, the links and concrete slabs were

Fig. 16. Comparison of displacement responses at tracking point D4X1.

Fig. 17. Comparison of strain responses at tracking point S10 (21).

 X. Tian et al. / Journal of Constructional Steel Research 143 (2018) 250–263 259

Table 9
Comparison of maximum acceleration responses.

PGA (g) Point A2X1 (second storey) Point A3x1 (third storey) Point A4x1 (roof)

FEM Experiment Relative error (%) FEM Experiment Relative error (%) FEM Experiment Relative error (%)

0.480 4.99 4.51 1.11 6.31 6.37 0.99 7.50 7.88 0.95
0.744 9.35 7.89 1.18 11.80 10.83 1.09 14.04 15.46 0.91

Table 10
Comparison of maximum displacement responses.

PGA (g) Point D2X1 (second storey) Point D3x1 (third storey) Point D4x1 (roof)

FEM Experiment Relative error (%) FEM Experiment Relative error (%) FEM Experiment Relative error (%)

0.480 2.37 2.84 0.83 3.69 3.99 0.92 4.38 5.28 0.83
0.744 4.07 5.01 0.81 6.33 7.19 0.88 7.51 8.81 0.85

Table 11
Comparison of maximum strain responses.

PGA (g) Point S2 (5) (on the first storey link) Point S6 (13) (on the second storey link) Point S10 (21) (on the third storey link)

FEM Experiment Relative error (%) FEM Experiment Relative error (%) FEM Experiment Relative error (%)

0.480 623 774 0.80 493 596 0.83 251 312 0.80
0.744 1167 1238 0.94 922 854 1.08 470 454 1.04

Note: in the first row of the table, the numbers inside the brackets represent the serial numbers of the strain gauges.

simulated with element shell181 (four-node structural shell), and other
members (column, beam, brace) were simulated with element
beam188 (3-D two-node beam). The beam elements were connected
with the shell elements by MPC184 (multipoint constraint element).
No restraints exist in the links and slabs owing to presence of no shear
studs arranged between them. The grid size of the link was 27.5 mm,
and the other grid size was 400 mm depending on the calculated
duration and calculation accuracy. Nonlinear analysis was performed
with the bilinear kinematic (BKIN) model. The elastic modulus E and
yield strength fy of the numerical model are the same as that of the
test specimen in Table 3, and Poisson's ratio v = 0.3, tangent modulus
Et = 0.01 E. Fig. 14 shows the FEM model.
responses at the same test point: Figs. 15–17 compare the acceleration,
displacement, and strain responses at test points A4X2, D4X2 and S10
(21) under the El Centro wave when the PGA was 0.744 g. The time his-
tory curves of the numerical simulation were consistent with those of
the test. (3) Maximum displacement, acceleration, and strain responses
of the test model under different ground motions: Tables 9–11 indicate
that, in most cases, the numerical results were close to the experimental
measurements under the El Centro wave when the PGA was 0.480 or
0.744 g. For example, the maximum relative errors were 18%, 19%, and
20%, respectively. It can be inferred that the finite element method is a
credible means of analyzing the seismic performance and failure mode
of a test model subjected to earthquakes.

4.2. Verification of numerical simulation 4.3. Failure mode

The acceleration, displacement, and strain responses of the test
model were calculated with the FEM under the same loading scheme
as the shaking table test. By comparison with the experimental
data, the numerical simulation results were verified in three aspects.
(1) Natural frequency: The numerical and experimental natural fre-
quencies of the test structure were 7.51 and 7.20 Hz, respectively,
which is a small difference. (2) Acceleration, displacement, and strain
Fig. 18 shows the equivalent stress distribution of a link under the
Taft wave with a PGA of 0.744 g. The maximum stress was near the
joint between the link and frame beam because these were modeled
with beam and shell elements, respectively, and then connected with
rigid elements, which resulted in stress concentration at the junction.
Regardless of the stress concentration, the maximum stress was on
the web of the link.

ANSYS 14.0
MAR 5 2017
20:19:30
13105.3
.127E+08
.254E+08
.380E+08
.507E+08
.634E+08
.761E+08
.887E+08
.101E+09
.114E+09

Fig. 18. Equivalent stress distribution.

260 X. Tian et al. / Journal of Constructional Steel Research 143 (2018) 250–263

(a) (b) (c) (d)

Fig. 19. Failure mode: (a) the links on the first storey yield; (b) the links on the second storey yield; (c) local details; and (d) the braces yield.

The El Centro wave with larger PGA was adopted to further study the
failure mode of the test model subjected to earthquakes. The links were
deformed and damaged firstly under seismic action because of the rela-
tively low strength and stiffness. Fig. 19 shows the yielding sequence;
the yielded parts are not shown. Just as expected for the structural de-
sign, the links on the first and second stories yielded first. However,
the brace of the first storey yielded earlier than the link of the third
storey, which was unexpected. Therefore, the specimen needs to be op-
timized to make it more reasonable. When PGA attained a value of 2.0 g,
webs in the link on the first floor were found to have yielded. Experi-
mental measurements and numerical results showed that the failure
mode was triggered by the design of HSS-EBFs, where the links are
the weakest parts.
4.4. The influence on shear distribution of slabs to links
In order to study the effect of a concrete slab, FE models of the spec-
imen, with and without concrete, were modeled and analyzed. As
depicted in Fig. 20, the concrete slab resisted about 15% of the total
shear force. The maximum shear force resisted by the composite links
ranged between 0.96 and 1.06 times that resisted by corresponding
steel links, which was consistent with that reported in [32], wherein
Ricles and Popov reported that the maximum shear force resisted
by composite links was 1% to 13% greater compared to corresponding
steel links.
5. Comparison of seismic performance between K-HSS-EBF and
traditional K-EBF
In order to study the link rotation angle and storey drift of HSS-EBFs
under the limiting state, two samples of traditional K-EBF and K-HSS-
EBF under identical conditions were designed and analyzed. The
specimens were numerically designed and optimized using the
performance-based seismic design (PBSD) method [20]. Both models
represented 5-storeyed office buildings, each storey having a height
and column spacing of 3.6 and 7.2 m, respectively. The shape and size

8 40
Shear force of Shear force of
Shear force / 103 kN
Shear force /103 kN

links with slabs links with slabs

6 30

4 Shear force of 20 Shear force of

links without links without
2 slabs 10 slabs

Shear force of Shear force of

0 0
slabs slabs
1 2 3 1 2 3
Storey / n Storey / n

(a) Frequent earthquake (PGA = 0.11 g) (b) Rare earthquake (PGA = 0.51 g)
Fig. 20. Shear-force resistance comparison.
 X. Tian et al. / Journal of Constructional Steel Research 143 (2018) 250–263 261

Fig. 21. Details of the 5-storeyed office model.

of the plane and its elevation are depicted in Fig. 21. Each link (short
link) measured 900 mm in length (e = 1.245–1.445 Mp/Vp). Details of
the cross sections used in numerical examples are listed in Tables 12
and 13, where “H” and “□” refer to welded H and box sections, respec-
tively. The traditional K-EBF was made of Q345 steel (fy = 345 MPa). For
K-HSS-EBF, links were manufactured using Q345 while remaining
members were manufactured using Q460 steel (fy = 460 MPa).
The yield strength of all materials were of nominal value, the elastic
modulus E = 2.06 × 105 MPa, and the tangent modulus Et = 0.01 E.
Dynamic elasto-plastic time-history analyses were performed on the
2 models under the action of 10 different seismic waves selected from
the peer ground motion database. During analyses, values of storey
drift within the structures and rotation angle of the links were recorded.
Values of storey drift for the K-EBF and K-HSS-EBF models, under eight
degree frequent and rare earthquakes, were compared, as depicted in
Fig. 22. As can be seen in the figure, for a 5-storeyed frame structure,
the storey drift corresponding to K-HSS-EBF is larger compared to that
of traditional EBF. It is, however, still less than the specification limit re-
quired by GB50011-2010 [15]. Also, the performance of K-HSS-EBF
seems to fulfill seismic requirements.
Values of link rotation angles for traditional K-EBF and K-HSS-EBF,
under eight degree rare earthquakes (PGA = 0.51 g) were then com-
pared, as depicted in Fig. 23. It can be seen that, both the maximum en-
velope and average values of K-EBF and K-HSS-EBF are less than
the corresponding specified limit in AISC 341-10, thereby limiting
the inelastic rotation of short links to a value within 0.08 rad.
Fig. 23 (c) illustrates that the average rotation angle of the links in tradi-
tional K-EBF are less than or equal to that of K-HSS-EBF for the condi-
tions under which the comparison is made.
6. Conclusions and suggestions
In this study, the earthquake resistance of K-EBF-HSS was evaluated
through shaking table tests and numerical simulations. The following
conclusions were obtained:
(1) As the PGA increased, the structural natural frequency and
equivalent stiffness slowly decreased while the structural damping
ratio increased. When the PGA reached 0.744 g, the natural frequency
of the structure dropped by 2%, and the damping ratio increased by
23%. These results indicate slight structural damage, which is consistent
with the test observations.
(2) The horizontal seismic action of the specimen was the same as
that derived from the base shear method and showed an inverted trian-
gle distribution pattern. The average test value was at least 6% lower
than the calculated value, which indicates that the base shear method
can be used to calculate the horizontal seismic action of a K-EBF-HSS
structure.
(3) The acceleration amplification factor increased with the height
and tended to decrease with increasing PGA. The average amplification
factors of the third, second, and first stories during the elastic stage were
2.2, 1.6, and 1.1, respectively.

Table 12
Cross sections of members of traditional EBF (unit: mm).

Storey beams links columns braces

5 H370 × 150 × 8 × 12 H300 × 140 × 6 × 12 □350 × 350 × 16 H200 × 200 × 10 × 18

4 H400 × 200 × 10 × 16 H390 × 160 × 8 × 14 □420 × 420 × 16 H220 × 220 × 10 × 18
3 H460 × 230 × 10 × 16 H420 × 180 × 10 × 16 □450 × 450 × 16 H250 × 250 × 10 × 18
2 H510 × 230 × 10 × 16 H480 × 180 × 10 × 16 □500 × 500 × 16 H250 × 250 × 10 × 18
1 H540 × 230 × 10 × 16 H510 × 180 × 10 × 16 □500 × 500 × 16 H250 × 250 × 10 × 18

Table 13
Cross sections of members of HSS-EBF (unit: mm).

Storey beams links columns braces

5 H310 × 140 × 6 × 12 H310 × 140 × 6 × 12 □300 × 300 × 16 H200 × 200 × 10 × 18

4 H370 × 160 × 10 × 16 H400 × 160 × 8 × 14 □350 × 350 × 16 H220 × 220 × 10 × 18
3 H430 × 180 × 10 × 16 H430 × 180 × 10 × 16 □400 × 400 × 16 H250 × 250 × 10 × 18
2 H480 × 180 × 10 × 16 H490 × 180 × 10 × 16 □460 × 460 × 16 H250 × 250 × 10 × 18
1 H510 × 180 × 10 × 16 H520 × 180 × 10 × 16 □460 × 460 × 16 H250 × 250 × 10 × 18
262 X. Tian et al. / Journal of Constructional Steel Research 143 (2018) 250–263

5 5

4 4
Story / n

Story / n
3 3

K-EBF K-EBF
2 2 K-HSS-EBF
K-HSS-EBF
specification limit specification limit
1 1
0 0.002 0.004 0.006 0 0.01 0.02 0.03

The storey drift / rad The storey drift / rad

(a) Frequent earthquake (PGA = 0.11 g) (b) Rare earthquake (PGA = 0.51 g)
Fig. 22. Storey-drift comparison between K-EBF and K-HSS-EBF.

(4) The maximum elastic storey drift of the structure was 1/1667,
and the maximum elastoplastic storey drift was 1/237. Both values
satisfy the current seismic code requirements, which indicates that
the model structure met the seismic fortification goals of “no
damage under minor earthquakes” and “no collapsing under strong
earthquakes.”
(5) The deformation of the links increased with the PGA and de-
creased with increasing height. The maximum strain of links occurred
on the first storey, which indicates that the first storey was the weakest
of the structure.
(6) The numerical analysis results fit well with the test results,
which indicates that the finite element method is a reliable means of
analyzing the seismic performance and failure mode of a structure
subjected to earthquakes. Experimental measurements and numerical
results showed that the failure mode was triggered by the design of
HSS-EBFs, where the links are the weakest parts.
(7) Comparison of seismic performance between K-HSS-EBFs and
traditional K-EBFs showed that the average storey drift and link
rotation angle of K-HSS-EBFs were larger than these of the tradition-
al K-EBFs, but were still less than the specification limit. K-HSS-EBFs
fulfill seismic requirements under the limiting state. In addition,
the PBPD method is found to be suitable for use in the design of
K-HSS-EBFs.
In summary, steel with a lower nominal yield strength is used in the
links of the HSS-EBFs so that the links first enter the plastic stage in the
event of an earthquake, while the other members made by HSS remain
in the elastic stage with minor deformation. This structure satisfactorily
solves the problem of inelastic deformation and energy dissipation and
has good seismic performance.
Acknowledgments
The authors are grateful for the financial support from the National
Natural Science Foundation of China (Grant No. 51178382, 51478380).
Thanks go to all who ensured the success of this experiment and proof
read this article.

5 5 5

4 4 4
Story / n
Story / n

Story / n

3 3 3

2 2 K-EBF
2
K-HSS-EBF
The average The average
specification limit
1 1 1
0 0.02 0.04 0.06 0 0.02 0.04 0.06 0 0.04 0.08 0.12
The rotation angle of The rotation angle of
links/rad links/rad The rotation angle of links/rad

(a) Traditional K-EBF (b) K-HSS-EBF (c) Comparison of average values

Fig. 23. The rotation angle of links.

 X. Tian et al. / Journal of Constructional Steel Research 143 (2018) 250–263
References
[1] C. Miki, K. Homma, T. Tominaga, High strength and high performance steels and
their use in bridge structures, J. Constr. Steel Res. 58 (1) (2002) 3–20.
[2] A. Azizinamimi, K. Barth, High performance steel: research front-historical account
of research activities, J. Bridg. Eng. 9 (3) (2004) 212–217.
[3] G. Shi, C.S. Deng, H.Y. Ban, Y.Y. Chen, Y.Q. Wang, Y.J. Shi, Experimental study on hys-
teretic behavior of high strength steel box-section columns, J. Build. Struct. 33 (3)
(2012) 1–7.
[4] H.Y. Ban, G. Shi, Y.J. Shi, Y.Q. Wang, Overall buckling behavior of 460 MPa high
strength steel columns: experimental investigation and design method, J. Constr.
Steel Res. 74 (2012) 140–150.
[5] H.Y. Ban, G. Shi, Y.J. Shi, M.A. Bradford, Experimental investigation of the overall
buckling behavior of 960 MPa high strength steel columns, J. Constr. Steel Res. 88
(2013) 256–266.
[6] C.H. Lee, K.H. Han, C.M. Uang, D.K. Kim, C.H. Park, J.H. Kim, Flexural strength and ro-
tation capacity of I-shaped beams fabricated from 800-MPa steel, J. Struct. Eng. ASCE
139 (6) (2013) 1043–1058.
[7] G. Shi, Z. Liu, H.Y. Ban, Y. Zhang, Y.J. Shi, Y.Q. Wang, Experimental study on the local
buckling of high strength steel equal angles under axial compression, Eng. Mech. 28
(7) (2011) 45–52.
[8] G. Shi, H.Y. Ban, F.S.K. Bijlaard, Tests and numerical study of ultra-high strength steel
columns with end restraints, J. Constr. Steel Res. 70 (2012) 236–247.
[9] G. Shi, C. Deng, H. Ban, Y. Chen, Y. Wang, Y. Shi, Experimental study on hysteretic be-
havior of high strength steel box-section columns, J. Build. Struct. 34 (1) (2013)
13–21.
[10] H.Y. Ban, G. Shi, Y.J. Shi, Y.Q. Wang, Residual stress of 460 MPa high strength steel
welded box section: experimental investigation and modeling, Thin-Walled Struct.
64 (2013) 73–82.
[11] D.K. Kim, C.H. Lee, K.H. Han, J.H. Kim, S.E. Lee, H.B. Sim, Strength and residual stress
evaluation of stub columns fabricated from 800 MPa high-strength steel, J. Constr.
Steel Res. 102 (2014) 111–120.
[12] P.S. Green, R. Sause, J.M. Ricles, Strength and ductility of HPS flexural members,
J. Constr. Steel Res. 58 (2002) 907–941.
[13] X.H. Qiang, X. Jiang, F.S.K. Bijlaard, H. Kolstein, Mechanical properties and design
recommendations of very high strength steel S960 in fire, Eng. Struct. 112 (2016)
60–70.
[14] M. Lian, Su MZ, Y. Guo, Seismic performance of eccentrically braced frames with
high strength steel combination, J. Steel Compos. Struct. 18 (6) (2015) 1517–1539.
[15] GB 50011-2010, Code for Seismic Design of Buildings, China Architecture & Building
Press, Beijing, China, 2010.
263
[16] A.L. Yu, An experimental study on the a seismic properties for EK-brace and Y-brace
in steel constructions, J. Xi'an Inst. Metall. Constr. Eng. 22 (3) (1990) 253–260.
[17] B. Zhao, Q. Gu, Nonlinear finite element analysis for eccentrically braced steel frames
under cyclic load, Civ. Eng. J. 38 (2) (2005) 27–31.
[18] L.S. Duan, M.Z. Su, Q. Hao, P.P. Jiao, Experimental study on seismic behaviour of high
strength steel composite K-type eccentrically braced frames, J. Build. Struct. 35 (7)
(2014) 18–25.
[19] D. Dubina, A. Stratan, F. Dinu, Dual high-strength steel eccentrically braced frames
with removable links, Earthq, Eng. Struct. Dyn. 37 (15) (2008) 1703–1720.
[20] S. Li, M.Z. Su, M.V. Lian, Seismic behavior study of K-typed eccentricially braced steel
frames with high strength steel combination based on performance design method,
Build. Struct 45 (6) (2015) 71–79.
[21] M. Lian, M.Z. Su, S. Li, X.H. Tian, Shake table test of seismic performance of high
strength steel frames with Y-eccentric braces, J. Build. Struct. 36 (8) (2015) 24–31.
[22] F. Wang, Su MZ, M. Hong, Y. Guo, S. Li, Cyclic behavior of Y-shaped eccentrically
braced frames fabricated with high-strength steel composite, J. Constr. Steel Res.
120 (2016) 176–187.
[23] Xu ZD, Horizontal shaking table tests on structures using innovative earthquake
mitigation devices, J. Sound Vib. 325 (2009) 34–48.
[24] Yu QQ, Gu XL, Y. Li, F. Lin, Collapse-resistant performance of super-large cooling
towers subjected to seismic actions, Eng. Struct. 108 (2016) 77–89.
[25] Q.H. Han, Y. Xu, Y. Lu, J. Xu, Q.H. Zhao, Failure mechanism of steel arch trusses:
shaking table testing and FEM analysis, Eng. Struct. 82 (2015) 186–198.
[26] GB 50009-2012, Code for Load of Building Structures, China Architecture & Building
Press, Beijing, China, 2012.
[27] GB 50017-2003, Code for Design of Steel Structures, China Architecture & Building
Press, Beijing, China, 2003.
[28] Y. Zhou, X.L. Lv, Method and Technique for Shaking Table Tests of Building Struc-
tures, Science Press, Beijing, China, 2012.
[29] K. Kasai, E.P. Popov, General behavior of WF steel shear link beams, J. Struc. Eng.
ASCE 112 (2) (1986) 362–382.
[30] F.Y. Xv, A.R. Chen, S.F. Zhu, Identification of modal parameters for bridge wind tun-
nel test by using a stochastic subspace technique, China Civ. Eng. J. 40 (10) (2007)
67–73.
[31] J. Chang, Modal Parameters Identification of Bridge by Stochastic Subspace Identifi-
cation Method, Tongji University, Shanghai, China, 2006.
[32] J.M. Ricles, E.P. Popov, Composite action in eccentrically braced frames, J. Struct. Eng.
ASCE 115 (8) (1989) 2046–2065.