Engineering Structures 211 (2020) 110460

Contents lists available at ScienceDirect

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

Development and testing of hybrid precast steel-reinforced concrete column- T

to- H shape steel beam connections under cyclic loading
⁎
Chenglong Wua, Sujian Yua, Jiming Liua, , Genda Chenb
a
School of Civil Engineering, Qingdao University of Technology, Qingdao 266033, PR China
b
System and Processing Assessment Research Laboratory, Department of Civil, Architectural, and Environmental Engineering, Missouri University of Science and
Technology, Rolla, MO 65409, USA

A R T I C LE I N FO A B S T R A C T

Keywords: This paper proposes a steel-reinforced concrete column-steel beam composite joint (MPCJ) suitable for modular
Prefabricated prefabricated building construction and investigates its seismic performance. The proposed MPCJ has simple
Beam-to-column joint structure, convenient construction, and good bearing capacity. This study performs an experimental investiga-
Seismic performance tion in which three MPCJ specimens with different beam-column connections (bolted connection, welded
Pseudo-static testing
connection, and bolted-welded hybrid connection) are subjected to low-cycle reversed loading and are in-
Finite element analysis
vestigated for the elastic and elastoplastic development trends, failure characteristics, and seismic response.
Each specimen undergoes bending failure at the end of the beam that follows the buckling of the flange con-
necting plate. MPCJs exhibit stable hysteretic curves, reasonable strength and stiffness degradation, as well as
good ductility and energy dissipation performance. The ductility coefficients and the equivalent viscous damping
coefficients of MPCJ specimens are within the seismic performance limits prescribed by structural design
standards. The joint rotation of an MPCJ caused by shear deformation appears different from that of conven-
tional beam-column joints. Finite element models of MPCJ specimens are established by ABAQUS, and the finite
element results are compared with the experimental results, thus verifying the feasibility and accuracy of the
finite element models. Based on the experimental findings and the numerical validation, simplified equations are
proposed for calculating the flexural and shear bearing capacity of an MPCJ. The proposed MPCJ connection will
be useful in the modular construction of concrete-steel composite frame structures.

1. Introduction zones. Existing studies primarily focus on the experimental and theo-
retical aspects of fabricated concrete frame joints [5,6], and on the
The construction industry is increasingly witnessing composite assembly of steel structures [7–10]. For instance, Deng et al. [11] in-
structures with a modular standardized design that often involves troduced a fully prefabricated beam-column joint with composite steel-
production at a factory and assembly of modular units at a construction ultra high performance concrete and friction damper. Through the ex-
site [1,2]. Moreover, steel-concrete composite structures are quite sui- perimental study, good bearing capacity and seismic performance of the
table for ‘green buildings’ that save energy and protect environment joint were confirmed, and a theoretical model was proposed for the
[3]. While modularized prefabricated steel-concrete composite struc- beam-column joint connection. Zhang et al. [12] studied a beam-
tures are widely applied to high-rise buildings in areas with low seis- column joint with prefabricated sinusoidal corrugated web, and estab-
micity, steel-concrete composite structures have also been widely used lished a design theory by considering the bolt binding coefficient, the
in countries with higher seismic requirements, such as China. Pre- overall bending of the binding section and the overall instability of the
fabricated steel-concrete composite structures can be connected using flange cover plate. In addition, seismic performance of prefabricated
high-strength bolts or welds, which makes the steel-concrete composite side joints and middle joints has been studied with different beam-end
structure quite suitable for constructing prefabricated buildings [4]. connections [13,14].
Beam-to-column joints appear to be the most critical members in the Few studies have focused on the design and manufacturing methods
assembly of steel-concrete composite structures. This is due to the sig- for modular composite joints between prefabricated steel reinforced
nificant role they play in controlling the structural behavior in seismic concrete (SRC) columns and steel beams even though most of the

⁎
Corresponding author.
E-mail address: [email protected] (J. Liu).

https://doi.org/10.1016/j.engstruct.2020.110460
Received 25 October 2019; Received in revised form 12 February 2020; Accepted 2 March 2020
Available online 09 March 2020
0141-0296/ © 2020 Published by Elsevier Ltd.
C. Wu, et al. Engineering Structures 211 (2020) 110460

Fig. 1. Diagram showing the parts and construction process of the three modular prefabricated composite joints.

damage of assembled buildings occurs at the beam-column joints. For the floor and the structural measures of pouring concrete in the core
example, improper connections between prefabricated components area of the joint.
caused serious damage of an assembled building and resulted in sig- A pseudo-static experimental study is performed on the modular
nificant loss as experienced during the Northridge earthquake in 1994 prefabricated steel-reinforced concrete column-steel beam composite
and the Kobe earthquake in 1995 [15,16]. These observations indicate joints. The influences of the three beam-column connections, namely,
that connectors in precast structures should have good bearing capa- bolted joints, welded joints, and bolted-welded joints, on the damage
city, ductility, and energy dissipation capacity. It is crucial to develop modes and seismic performance of the components are analysed.
efficient construction methods for steel-concrete composite structures Subsequently, three MPCJ specimens are modeled and analyzed by a
with reliable connections between modular prefabricated SRC column finite element program, ABAQUS, and the accuracy of the finite ele-
and steel beams. In order to study the mechanical properties and energy ment model is verified. Finally, simplified calculation of the joint
consumption capacity of modular composite joints, Wu et al. [17] bearing capacity is presented. This contribution is expected to lay a
conducted quasi-static tests and investigated the seismic performance of theoretical foundation for the development and application of an as-
three SRC column-steel beam interior joints subjected to a low cycle sembled composite joint structure.
reciprocating load. Nzabonimpa et al. [18] proposed a detachable steel-
concrete composite beam-column joint by performing experimental and
numerical study. 2. Experimental procedure
This paper proposes a modular prefabricated SRC column-to-H
shaped steel beam joint (MPCJ), as shown in Fig. 1. MPCJ comprises 2.1. Specimen design and material properties
three components, namely, prefabricated SRC columns, steel beams and
joint core modules (hereafter referred to as ‘joint modules’). The SRC Three MPCJ specimens (MPCJ1, MPCJ2 and MPCJ3) with a scale of
column is connected with the steel beam through the joint module, and 1:2 were designed according to CEN 2004 [19]. All the joints were
the joint module and the SRC column are connected by bolts. The joint designed as plane frame joints. Taking the MPCJ1 specimen as an ex-
module and the steel beam are connected by either bolt-weld hybrid ample, the dimensions and components of the MPCJ specimens are
connections (MPCJ1) or weld connections (MPCJ2) or bolt connections shown in Fig. 2. Three specimens with different joint covers, flange
(MPCJ3). Each SRC column is composed of H-shaped steel, reinforce- connection plates and webs connection plates are shown in the figure.
ment cage, concrete and column connection end plates (column loading The main dimensional parameters and connection modes of the flange
end plates). The second component, beam, is made of a standard hot- connecting plates and webs are shown in Table 1.
rolled H-shaped steel beam. As shown in Fig. 1, the joint module The core square steel pipe constituting the joint module had a sec-
comprises a square steel tube welded with a joint cover plate and two tion size of 160 mm × 160 mm × 10 mm (depth × width × thickness)
stiffening plates 1 and 2. Prefabricated SRC columns, steel beams and and the height of 280 mm. The square steel tube was designed to resist
joint modules are processed in a factory and are connected using compression in the vertical direction and shear deformation in the
bolting or welding at site. MPCJs do not consider the pouring, grouting, horizontal direction. A joint cover plate and a stiffener plate of the joint
and tying of steel bars in the joint core area, which can be used in module were used to connect it with prefabricated steel reinforced
prefabricated high-rise buildings in multi-shock areas. Nevertheless, in concrete column and steel beam. The section size of the H-shaped steel
this study, beam-column joint does not consider the composite effect of beam was 280 mm × 150 mm × 10 mm × 10 mm
(depth × width × web thickness × flange thickness). Three types of

2
C. Wu, et al. Engineering Structures 211 (2020) 110460

Fig. 2. Dimensions and components of the three MPCJ specimens (All dimensions are in mm).

connections between the SRC core and the beam were investigated by 20 mm diameter threaded steel rod, having a length of 80 mm in the
testing three specimens, namely, MPCJ1, MPCJ2 and MPCJ3. The flange and 150 mm in the web. The spacing between the shear con-
connection in MPCJ1, MPCJ2 and MPCJ3, respectively, comprised a nectors was 120 mm and the distance between the center of the shear
bolt-weld hybrid connection, a welded connection and a bolt connec- connector and the edge of the flange was 75 mm.
tion. As shown in Fig. 1, for the MPCJ2 and MPCJ3 specimens, the bolts Tables 2 and 3 show the material test results of Q345B steel,
used between the steel beam and the joint module were S10.9 M24 high HPB300 and HRB400 reinforcement and C40 concrete used in the MPCJ
strength bolts with a pretension of 225 kN. The cross-sectional di- specimens. Q345B steel was used for the steel beam, H-shaped steel
mensions of the upper and lower SRC columns were 350 × 350 (mm) skeleton joint module and all other steel components. Tensile strength
with the height of 780 mm. The section of the H-section steel was of Q345B steel was tested on steel samples with the same thickness as
150 × 150 × 7 × 10 (mm). The column end connecting plate used in the specimens. The average value of three samples was obtained
(350 × 350 × 10 (mm)) and column end loading plate from the test results. The mechanical properties of steel included yield
(350 × 350 × 20 (mm)) were welded at both ends. The longitudinal strength fy, ultimate strength fu, elastic modulus Es and elongation after
reinforcement in the SRC was made from A20 rebars (A is diameter). fracture δ. SRC column adopted C40 grade concrete. MPCJ specimens
The stirrups were made from A8 rebars with a spacing of 100 mm. The were cured for 28 days at the controlled temperature of 20 ± 3 °C and
SRC columns were connected to the joint module through the joint the relative humidity was greater than 90 percent. The test results of
cover plates with S10.9 M20 high strength bolts and an extension concrete are shown in Table 3.
(50 mm length) of longitudinal steel bar. The pretension of M20 bolt
was 155 kN. The extension section of the four longitudinal steel bars in 2.2. Loading scheme
the SRC column was threaded and passed through the joint cover plate,
so that the steel reinforced concrete column was easily fixed on the Each specimen was subjected to a series of cyclic loading tests at the
cover plate of the joint module. The upper ends of the longitudinal bars Structure Engineering Laboratory of Qingdao University of Technology.
were fixed to the top loading end plate using plug welding. In an SRC The orientation of the specimens during testing was rotated 90 degrees
column, the H-shaped steel and concrete were further connected using counterclockwise relative to the actual field condition they would re-
shear connectors that complied with the shear connection design re- present [21]. In other words, the SRC column specimens were loaded
quirements of the Chinese standard [20]. The layout of the shear con- axially in a lateral position using a hydraulic jack (Fig. 3). Before the
nector is shown in Fig. 2. Each shear connector was made by using a test, the ultimate compressive bearing capacity of the SRC columns was

Table 1
Specimens tested in the experiment.
Specimen number Size of beam end connecting plate (mm) Beam-end connection mode

Flange connection plate Web connection plate Flange Web

Length Width Thickness Height Width Thickness

bolted-welded joint 270 130 10 240 210 10 Welded Bolted

welded joint 270 130 10 220 210 10 Welded Welded
bolted joint 350 150 10 240 210 10 Bolted Bolted

3
C. Wu, et al. Engineering Structures 211 (2020) 110460

Table 2
Material properties of steel and reinforcement.
Material sample Sampling position Thickness/radius (mm) fy (MPa) fu (MPa) Es(MPa) Δ(%)

5
Q345B H-steel skeleton web 7 426.94 590.26 2.05 × 10 26.16
H-steel skeleton flange 10 434.17 571.5 2.02 × 105 23.02
H-steel beam flange, web and other steel plate 10 418.42 563.84 2.01 × 105 19.53
HPB300 Stirrup 4 396.52 497.35 2.32 × 105 20.14
HRB400 Reinforcement 10 471.67 591.67 2.02 × 105 24.33

obtained by a theoretical calculation. Based on this theoretical strength, of the beam-column joint basically remained intact and in the elastic
the axial compression ratio to be applied was set at 0.15. At the be- state.
ginning of the load test, a 100 t jack was used to exert an axial load on The stress state of MPCJ members during the entire testing process
the top of the SRC column and to maintain a constant axial load of 506 could be divided into three phases: (1) elastic phase; (2) elastoplastic
kN. phase; and (3) destruction phase. At the beginning of the loading, the
Both ends of an SRC column were constrained for vertical and entire specimen was in the elastic state. During the process of loading,
horizontal displacement. A 500 kN capacity MTS hydraulic actuator unloading, and reloading, no abnormal phenomena occurred in the
was connected to the beam in the upright position. The loading speed of specimens. As the load increased, concrete in the SRC column began to
the actuator was three minutes per cycle. The MTS hydraulic actuator develop tiny cracks due to repeated cycles of tension and compression.
controlled the proposed load-displacement protocol (Fig. 4) [22,23]. At the end of the elastic phase, the crack width in the concrete surface
The yield load of the specimen was determined based on the theoretical was between 0.05–0.1 mm. As the external load increased, the canti-
calculation before the test. A force-controlled mode was adopted before lever end of the joint cover plate gradually reached its yield stress. With
the test piece yielded, and it was divided into five stages with equal further increase in load, specimens entered into the elastic-plastic stage,
increments of 20% of yield load in each stage. Each loading stage and the surface of the upper and lower flanges of the beam showed
comprised one loading cycle. After the specimen was yielded, the peeling and wrinkling. In case of specimen MPCJ3, the buckling de-
loading mode was changed to displacement-controlled. In the dis- formation and slip on the bolted joints were observed and it became
placement-controlled mode, load was applied according to the multiple increasingly obvious with an increase in loading. The width and length
of yield load, and each load was cycled three times until the specimen of cracks in the concrete surface expanded continuously. One or several
was damaged. This approach demonstrated the strength, deformation vertical cracks dominated the length of the column, and many hor-
and energy consumption of the test specimen under the action of a izontal cracks appeared perpendicular to the column axis. Crack width
reciprocating load. The actual yield load Py was obtained by observing was increased to 0.1–0.3 mm. When the load reached its peak value, the
the inflection point of the load displacement real-time curve and the flange connection plate underwent buckling deformation and entered
node deformation during the test. In the analysis of the test data, the the failure stage. In the later stage of loading, the flange connection
yield point was obtained according to the load displacement curve and plate buckled due to the concentrated stress. This reduced the influence
relevant analysis methods. of the external load on the precast concrete column. Hence, concrete
During the loading process, the test was terminated when either of was not markedly cracked, but the steel beam deformed remarkably.
the following conditions occurred in the joint specimens: (1) the spe-
cimen was suddenly destroyed or experienced lateral instability, and 3.2. Determination of characteristic points
could not continue to bear the load; or (2) the bearing capacity of the
specimen reached its peak value, and the load decreased to less than As shown in Table 4, the eigenvalues of the characteristic points of
85% of the maximum load (ultimate Pmax). each specimen during the process of loading were obtained. The char-
acteristic points included yield point (Δy, Py), maximum point (Δmax,
3. Experimental results and analysis Pmax) and failure point (Δu, Pu) where Δy is the yield displacement and
Py is the yield load corresponding to the yield displacement. Pmax is the
3.1. Damage modes maximum load value on the P–Δ skeleton curve and Δmax is the dis-
placement corresponding to the maximum load value. Similarly, Δu is
Fig. 5 shows the damage modes of specimens MPCJ1, MPCJ2, and the ultimate displacement when the load at the end of the beam was
MPCJ3 at the end of the tests. During the test, components of the decreased to 85% of the peak value, and Pu is the ultimate load cor-
specimens were damaged or destroyed to varying degrees. The failure responding to the ultimate displacement. The yield point (Δy, Py) was
mode of all three MPCJ specimens was the bending failure of the beam- calculated according to the initial stiffness method mentioned in the
end, which occurred at the flange connecting plate. This is attributed to literature [24].
the connection structure of the beam-end. When the beam-end was
subjected to the external load, stress was relatively concentrated in the 3.3. Load P – displacement Δ hysteretic curve
flange connecting plate thus causing the plate to first buckle and then
break. It satisfied the design criteria that a plastic hinge should be The P – Δ hysteretic curves of the joints were obtained from pseudo-
formed at the beam-end thus protecting the core area of the joint. Ex- static tests and are shown in Fig. 6. The hysteretic curves were the main
cept for the failure of the flange connection plate, all other components basis for evaluating the seismic performance of the joints [25]. The P –

Table 3
Material properties of C40 concrete.
Material fcu,k (MPa) fc (MPa) ft (MPa) E0 (MPa) σt0 (MPa) εt0 σc0 (MPa) εc0 ρ (ton/mm3)

C40 42.79 20.44 1.74 3.32 × 104 2.39 1 × 10−4 26.8 1.8 × 10−4 2.4 × 10−9

Note: fcu,k is the standard value of cube compressive strength, fc is the design value of axial compressive strength, ft is the design value of axial tensile strength, and ρ
is the density.

4
C. Wu, et al. Engineering Structures 211 (2020) 110460

Fig. 3. Experimental setup.

Δ hysteretic curves of three joint specimens were mostly fully devel- between the positive and negative peak loads of MPCJ3 was approxi-
oped, indicating that the proposed joint had strong energy dissipation mately 25 kN. The main reason is that with an increase in the the re-
capacity under different connection modes (Fig. 6). The load-displace- ciprocating load at the end of the beam, slip occurred between the
ment skeleton curves for the three specimens are shown in Fig. 7, which flange of the steel beam and the flange connecting plate. Due to the
indicated that the stress state of specimens under reciprocating loads asymmetry of the slip of the bolts in the upper and lower part, the
could be roughly divided into three phases, namely, an elastic phase, an buckling deformation of the connecting plate was larger for the upper
elastoplastic phase and a failure phase. flange and its fracture occurred gradually, which reduced the peak
Some differences were observed among the three test specimens. bearing capacity under the positive loading. Unlike the MPCJ1 and
Under the cyclic loading for each displacement level, the difference MPCJ2 specimens, the MPCJ3 specimen was always accompanied by

5
C. Wu, et al. Engineering Structures 211 (2020) 110460

Fig. 4. Cyclic loading protocol for the MPCJ specimens.

bolt slippage during the loading stage from yield to failure. The reason γj in the joint core provided valuable information on the bearing ca-
for the slippage is that the steel beam flange and the flange connecting pacity and seismic characteristics of the specimens. Under the load at
plate were connected by bolts. Consequently, the moment generated at the end of the beam, the horizontal shear force in the joint region
the loading end of the beam was mainly transmitted through the flange primarily agreed to the equilibrium equation in the shear region as
of the steel beam. When the tension (compression) force on the flange described by EC3 code [27]. The horizontal shear force was defined as
of the steel beam exceeded the normal force generated by the pre- in Eq. (1):
tightening force of the bolt, the relative sliding between the two oc-
curred, conforming to the Coulomb’s friction criterion. Because of the
Vj =
Mb V − Vc2
− c1 =
Vb L − ( b
2 )−VL
b
slip, the ductility and energy dissipation capability of the MPCJ3 spe- z 2 z 2H (1)
cimen were improved. However, with an increase in external load, the
where Mb is the beam-end bending moment, Vb is the beam-end shear
bolt slip phenomenon became more pronounced, and the extent of slip
force, Vc1 and Vc2 are the upper and lower column-end shear forces,
gradually increased, thereby lowering the ultimate bearing capacity
respectively. z = hb − tbf is the center to center distance between the
and the stiffness of MPCJ3 around zero drift ratio when compared to
upper and lower flange of the steel beam. For H-beam sections with the
MPCJ1 and MPCJ2. At the same time, an increase in bolt slip distance
same upper and lower flange thickness, the height of steel beam and the
caused extrusion deformation of the bolt holes, shear deformation of
thickness of flange are hb and tbf, respectively. Parameter b is the width
the bolts, and fracture of the flange connecting plate along the cross
of the joint module (b = 350 mm), L is the horizontal distance between
section of bolt holes. This also increased the overall deformation per-
the loading point at the beam-end and the center line of column; and H
formance of the composite structure. In the real construction, the
is the vertical distance between the constraint points at both ends of the
roughness of the contact surface of the bolt connection can be increased
SRC column.
and the friction coefficient of the bolt can be increased from the aspects
As shown in Fig. 3a, the relative rotation between steel beams and
of design, processing and manufacturing [26].
SRC columns as well as the shear deformation in the joint core area
The ultimate bearing capacity of the MPCJ2 specimen was the lar-
were measured using seven Linear Variable Differential Transformers
gest. Its hysteretic curve was slightly pinched, mainly because the
(LVDTs). Among them, shear deformation in the joint core region γj was
flange and web of the steel beam were welded, thus increasing the
measured by using LVDT1 and LVDT2 in the diagonal direction as
connection stiffness between the beam-end and the joint module. The
shown in Fig. 3(a). The shear deformation was defined as in Eqs. (2)
external load was effectively transferred to the joint module, which
and (3) [28]:
increased the bearing capacity of MPCJ2 specimens, increased the shear
deformation of the stiffening plate and the square steel tube, and re- a2 + b2
duced the overall stiffness of the joint module. Meanwhile, due to the γj = α1 + α2 = ·χ
ab (2)
bending deformation at the end of the beam, the concrete at the end of
the SRC column near the joint module experienced local crushing. The |δ1 + δ3| + |δ2 + δ4 |
χ=
hysteretic curves of MPCJ1 and MPCJ2 were similar, and there was no 2 (3)
significant pinch and slip phenomenon. Although the ultimate bearing where a is the height of the joint module (a = 280 mm), χ is the
capacity of the MPCJ1 specimen was approximately 20% lower than average relative displacement of diagonals AC and BD, δ1 + δ3, δ2 + δ4
that of the MPCJ2 specimen, the skeleton curve of the MPCJ1 specimen are the relative diagonal displacements obtained from displacement
had a descending section and a well-defined elastic-plastic deformation. meters LVDT1 and LVDT2, and a1 + a2 is the shear deformation angle.
The plastic hinge in the MPCJ1 specimen moved outward to protect the Fig. 8 shows the Vj − γj curves for the specimens. The Vj − γj curves
core area of the joint, which was helpful to improve the seismic per- for all the specimens were stable, and there was no apparent pinch or
formance of the joint. slip (Fig. 8). In the entire loading process, the shear deformation of the
joint core region was small and remained in the elastic state. Specifi-
3.4. Joint core shear force Vj-shear deformation γj curve cally, the joint core region of specimen MPCJ2 had the largest shear
deformation with γj between −0.0029 and 0.0123. The second largest
The relationship between core shear force Vj and shear deformation shear deformation was observed in MPCJ1, with γj value between

6
C. Wu, et al. Engineering Structures 211 (2020) 110460

Fig. 5. Photographs showing the damage modes and cracks of the MPCJ specimens.

−0.0022 and 0.0014. The shear deformation in specimen MPCJ3 was 3.5. Moment M–rotation θ response
the smallest, with γj value ranging from −0.00025 to 0.00025. Since
MPCJ3 had a bolted connection, the slip between the bolt and the The curve of joint moment M–rotation θ is an important index for
connecting plate influenced the shear deformation of the joint core evaluating the joint mechanical properties and is also a basis for the
area. On the other hand, MPCJ2 had a welded connection and hence its calculation and analysis of a frame structure. The rotation between the
core experienced the largest shear deformation as no contribution of center line of the steel beam and the center line of the column was
slip would be possible. taken as a variable quantity in the beam-end loading process.

Table 4
Primary performance indicators of the MPCJ specimens.
Type of joint specimen Direction Yield point Maximum point Failure point

Δy (mm) Py (kN) Δmax (mm) Pmax (kN) Δu (mm) Pu (kN)

bolt-welded joint positive 14.84 120.87 51.99 185.11 59.80 167.50

negative −14.82 −125.97 −44.93 −185.11 −59.80 −165.00
welded joint positive 15.01 153.35 74.70 238.04 89.74 198.11
negative −17.18 −150.24 −74.80 −235.76 −90.04 −215.45
bolted joint positive 9.90 78.72 109.01 165.71 118.92 127.09
negative −10.20 −80.39 −108.00 −190.40 −119.72 −163.70

Note: Δy is the yielding displacement, Δmax is the maximum displacement, Δu is the ultimate displacement; Py is the yield load.
Pmax is the maximum load, and Pu is the ultimate load.

7
C. Wu, et al. Engineering Structures 211 (2020) 110460

Fig. 6. Hysteretic curves from the tests and from the FEM analysis (FEM analysis to be discussed in Section 4): (a) MPCJ1; (b) MPCJ2; (c) MPCJ3.

where θb is the beam-end bending deformation, θc is the column-end

bending deformation, Δb is the beam-end actuator displacement, Δc3
and Δc5 are the displacements measured by LVDT3 and LVDT5, re-
spectively, and Lc35 is the distance between LVDT3 and LVDT5.
Fig. 9 shows the M–θ hysteretic responses for MPCJ specimens. By
comparing Fig. 6 and Fig. 9, it can be found that the P–Δ hysteresis
curves and M–θ hysteresis curves of specimens MPCJ1, MPCJ 2, and
MPCJ 3 were similar. Results showed that the beam-end displacement
occupied the largest proportion of the total displacement after yielded.
Fig. 9(d) shows the maximum shear angle γj, the maximum joint rota-
tion θ, and the ratio of (γj/θ). The (γj/θ) value for the three specimens
ranged from 0.207% to 1.138%, which quantitatively shows that the
joint rotation caused by shear deformation accounted for a small pro-
portion of the total rotational deformation (Fig. 9d). Hence, the effect of
joint module shear deformation on the structural deformation can be
neglected, which complies with the design requirement of ‘strong
column weak beam’.
Fig. 7. Skeleton curves from the tests and from the FEM results. Positive and negative values of the M–θ skeleton curves were
averaged using interpolation, and joints were classified according to the
European Code for Steel Structure Design EC3 [27]. As shown in
The joint moment M was defined as in Eq. (4) [29]: Fig. 10, the MPCJ2 specimen behaved as a semi-rigid and full-strength
b connection. Although the peak moment of the MPCJ1 specimen ex-
M = P × ⎛L − ⎞
⎝ 2⎠ (4) ceeded the limiting moment criterion of a semi-rigid connection, its
bending capacity was mainly between that of a rigid connection and a
The rotation θ was defined as in Eqs. (5)–(7): hinged connection, and the MPCJ1 specimen behaved as a semi-rigid
θ = γj + θ b + θc (5) connection. Therefore, in this paper, the MPCJ1 specimen with the bolt-
weld hybrid connection was classified as a semi-rigid connection and a
b partial strength connection. The MPCJ3 specimen was also a semi-rigid
θb = Δb / ⎛L − ⎞
⎝ 2⎠ (6) connection and a partial strength connection. When ultimate failure
occurred, the rotation angle corresponding to the bending moment was
θc = (Δc3 + Δc5)/ Lc35 (7)

8
C. Wu, et al. Engineering Structures 211 (2020) 110460

Fig. 8. Vj–γj curve: (a) MPCJ1; (b) MPCJ2; (c) MPCJ3.

Fig. 9. Hysteretic responses for the three specimens: (a) MPCJ1; (b) MPCJ2; and (c) MPCJ3; and (d) comparison of γj and θ for the three specimens.

9
C. Wu, et al. Engineering Structures 211 (2020) 110460

Table 5
Displacement ductility coefficient and inter-story drift of the specimens.
Specimen Direction Δy (mm) Δy/L Δu (mm) Δu/L μ μ̄
number

MPCJ1 Positive 14.84 0.0130 59.80 0.05 4.03 4.03

Negative −14.82 −0.0130 −59.80 −0.05 4.03
MPCJ2 Positive 15.01 0.0132 89.74 0.08 5.98 5.61
Negative −17.18 −0.0151 −90.04 −0.08 5.24
MPCJ3 Positive 9.90 0.0087 118.92 0.10 12.01 11.87
Negative −10.20 −0.0089 −119.72 −0.11 11.74

Note: Δy is the yielding displacement, Δu is the ultimate displacement, L is the

length of steel beam, and μ̄ is the average value of the ductility coefficient (μ).

flange connection plate, was that a certain amount of slip occurred

Fig. 10. Classification of joints based on the M–θ relationship. between the contact surfaces of bolts and steel in the MPCJ3 specimen,
which further enhanced the overall deformation capacity of the joint.

3.7. Energy dissipation capacity

Energy dissipation capacity is another important index of structural

seismic performance. The energy dissipation capacity of the MPCJ
specimens was measured by the area surrounded by the hysteresis
curves of the specimens. The energy consumption in each cycle Ei,
cumulative energy dissipation Etotal, and equivalent viscous damping
coefficient he were introduced to illustrate the energy dissipation of the
joints during loading. Coefficient he was defined by Eq. (9) [32–34] as:

1 S(ABC + CDA)
he = ×
2π S(OBE + ODF) (9)

where S(ABC+CDA) is the area of ABCD, and S(OBE+ODF) is the total area
of triangles OBE and ODF as shown in Fig. 12 (a).
Fig. 11. Comparison of bending moment and rotation of the three specimens. Fig. 12(b)–(d) show the relationships among the equivalent viscous
damping coefficient, energy consumption in each cycle, cumulative
energy dissipation and the number of cycles. Because of the different
taken as the ultimate joint rotation angle. As shown in Fig. 11, the ul-
connection modes of the three specimens, the number of cycles during
timate rotation angles of the specimens were 0.07 rad, 0.08 rad, and
the final loading (i.e., failure) was not uniform. Therefore, the numbers
0.11 rad, respectively, for the MPCJ1, MPCJ2 and MPCJ3 specimens.
of cycles of the three specimens in Fig. 6 are different. The variation in
The ultimate rotation angle of all specimens exceeded 0.04 rad, which
he with loading was slightly different compared to the variation in P–Δ
meets the ductility design requirements of joints as per FEMA-350 code
skeleton curve (Fig. 12(b)). Parameter he reached its peak value before
[30].
the load imposed at the beam-end attained its peak value. In the elastic
stage, the curve of equivalent viscous damping coefficient he vs. cycle
3.6. Ductility numbers n was relatively smooth, indicating that the energy con-
sumption was small, and the specimens were in the elastic state. With
Ductility is a basic indicator for the quantitative analysis of the an increase in the beam-end load, all specimens entered into the elastic-
seismic performance of the MPCJ specimens. In this study, ductility of plastic stage, and the he value increased rapidly. Among them, the
the specimens was described by the ductility ratio and the inter-story MPCJ3 specimen entered the energy consumption state the earliest,
displacement ratio, where the ductility ratio μ was defined by Eq. (8) because the sliding between the bolts and the connecting plate shor-
[25] as: tened the working time of the MPCJ3 specimen in the elastic stage. At
Δu the same time, the cyclic energy dissipation capacity of the specimen
μ= was reduced to a certain extent, but the ductile deformation perfor-
Δy (8)
mance was increased. The he–n curves of the MPCJ1 and MPCJ2 spe-
The inter-story drift ratio was defined as Δ/L, where Δ is the dis- cimens were similar in the elastic and elastic-plastic stages, but there
placement at the beam-end and L is the cantilever length. Table 5 shows were some differences in the failure stages. This was because the beam-
the ductility ratio and inter-story drift ratio of the MPCJ specimens as column flange connections were welded and under the reciprocating
obtained from the pseudo-static tests. The inter-story drift ratio of the action of the beam-end load, the flange connection plate first resisted
three specimens under both positive and negative loading was greater the bending moment load, and only after the buckling deformation of
than 0.04 rad (Table 5). According to AISC 2005 [31], when the inter- the flange, the web gradually shared the shear load from the beam-end.
story drift ratio is greater than 0.04 rad, the specimen is considered to Fig. 12(c) shows the energy dissipation in each cycle Ei. The Ei–n
have sufficient ductility. Therefore, MPCJ specimens met the ductility curve of the specimens showed a stepwise pattern. Even in the failure
requirements of frame joints in seismic areas. stage, the overall energy dissipation capacity of the specimens was
In addition, the ductility ratio of the MPCJ3 specimen was sig- significantly increased as the number of loadings increased. Under the
nificantly higher than that of the MPCJ1 or MPCJ2 specimens (Table 5). same displacement, with increasing n, the energy dissipation capacity
Indeed, the average ductility coefficient of the MPCJ3 specimen was decreased slightly due to the degradation of joint strength and stiffness.
2.95 times that of the MPCJ1 and 2.12 times that of the MPCJ2 spe- The influence of different beam-column connections on the cyclic en-
cimen. The main reason for this, besides the yield deformation of the ergy dissipation capacity of the joint was remarkable particularly after

10
C. Wu, et al. Engineering Structures 211 (2020) 110460

Fig. 12. Energy dissipation capacity. (a) Illustration of the concept of equivalent viscous damping coefficient; (b) Equivalent viscous damping; (c) Energy dissipated
in each cycle; (d) Energy accumulated.

yielded. Nevertheless, the energy consumption of joints with a welded the complete loading period [35]. The strength degradation coefficient
flange (MPCJ1 and MPCJ2) increased largely compared to the joint was defined as λj = Pj/Pmax, where Pj is the peak load when the dis-
with a bolted flange (MPCJ3). This is attributed to the fact that there placement grade is j (j = Δ/Δy) and Pmax is the peak load during the
was no slip in flange connection of the MPCJ1 and MPCJ2 specimens. complete loading period. The overall strength degradation trends for
The energy dissipation in the MPCJ1 and MPCJ2 specimens decreased the three specimens were basically the same and showed clear sym-
rapidly compared to the MPCJ3 specimen at the same displacement metry (Fig. 13). The λj–Δ/Δy curves of the MPCJ1 and MPCJ2 speci-
level, which was mainly caused by the tearing of the flange connection mens were similar, mainly because the flange connection plates were
plate in the MPCJ1 and MPCJ2 specimens. welded. In contrast, specimen MPCJ3 had bolted joints, and its overall
Cumulative energy dissipation Etotal is the envelope area of a hys- strength degradation was relatively gentle. The coefficients λj of the
teretic loop in the P–Δ curve during loading. The trend of cumulative three specimens ranged from 0.6 to 1.0, and no significant reduction
energy dissipation was identical for all specimens. As shown in was observed in Fig. 13, indicating that the proposed joint design is
Fig. 12(d), the trend of cumulative energy dissipation could be ap- relatively reasonable and the ductility performance is quite good.
proximated by an exponential function. The cumulative energy con-
sumption of each specimen increased slowly during the elastic phase. In 1.2
the elastoplastic phase, the cumulative energy dissipation capacity in-
creased with an increase in cycle numbers. The Etotal value at the time of 1.0
failure was lower for the MPCJ1 and MPCJ2 specimens when compared
to the MPCJ3 specimen. Therefore, in the failure stage, with an increase 0.8
in load and cycle number, the buckling deformation of flange connec-
tion plates was larger in theMPCJ1 and MPCJ2 specimens than in the 0.6
MPCJ3 specimen. In the MPCJ1 and MPCJ2 specimens, the bearing
capacity decreased rapidly, and the loading cycle terminated earlier 0.4
due to the fracturing of the connection plates. MPCJ1
0.2 MPCJ2
3.8. Strength ratio MPCJ3
0.0
Degradation in strength of a joint is often characterized through a -15 -10 -5 0 5 10 15
strength degradation coefficient. In this study, a strength degradation Δ /Δ y
coefficient λj was used to express the strength degradation perfor-
mance, which reflected the overall joint strength degradation during Fig. 13. Strength degradation curves of the three specimens.

11
C. Wu, et al. Engineering Structures 211 (2020) 110460

of 0.35 [39]. The friction coefficient between concrete and steel was
0.57 [40]. The “embedded region” command of ABAQUS was used to
determine the interaction between steel skeleton, reinforcement bar,
and concrete. The “tie” command was used to connect the components
of the joint core module, column end plate, H-shaped steel skeleton,
flange connecting plate, joint cover plate, beam flange (for welded
joints and bolted-welded joints), web connection plate, stiffening plate
2 and beam web (for welded joints) (Fig. 2).
The loads and boundary conditions imposed at the ends of the
columns and beams in the FEM were defined to be consistent with test
conditions [41]. That is, in the finite element model, boundary con-
straints were imposed by controlling the displacement and rotation
angle of column end and beam-end in x, y and z directions (Fig. 15). In
the ABAQUS model, the simplified dumbbell-shaped model was
adopted for the high-strength bolts and the pretension was applied at
Fig. 14. Stiffness degradation of the MPCJ specimens from the test.
the central element of the cross-section of the bolt shank. The bolt
pretension in the ABAQUS model was consistent with the test.
3.9. Stiffness degradation ‘Equivalent bolt’ model was used to simulate the end of the long-
itudinal reinforcement connecting the SRC column and the joint
Stiffness degradation means that the stiffness decreases with an module. The shank of the ‘equivalent bolt’ model had the same cross-
increase in loading repetitions for a given displacement. The stiffness sectional dimension, constitutive relationship, and material properties
degradation ratio is expressed by the ratio of secant stiffness of each as the longitudinal reinforcement. The center of the ‘equivalent bolt’
cycle to the first secant stiffness for a given displacement. It is a critical model section corresponded to the center of the longitudinal bar sec-
index to reflect the level of damage of the MPCJ specimens. As shown in tion, and the ‘coupling’ constraint combined the ‘equivalent bolt’ with
Fig. 14, stiffness degradation patterns of the MPCJ1, MPCJ2 and MPCJ3 the longitudinal reinforcement. The pretension of the ‘equivalent bolt’
specimens in the positive and negative directions were similar. In the was 100 kN, which was slightly lower than the design yield strength of
elastic and elastoplastic stages, stiffness degraded relatively faster for the longitudinal bars (113 kN) [42]. This approach not only avoided the
the MPCJ1 and MPCJ2 specimens, primarily due to the welded flange yield deformation of the extensional section of the longitudinal bars
connecting plates. Welded flanges largely influenced the overall joint before loading but also ensured proper connection.
stiffness and deformation. The bolt-weld mixed joint specimen (MPCJ1)
and the welded joint specimen (MPCJ2), both experienced similar 4.3. Material property
stiffness degradation when they were destroyed. As shown in Fig. 14,
stiffness degradation coefficient for the MPCJ3 specimen dropped ra- In the finite element models of the MPCJ specimens, geometric
pidly at the initial stage of loading, and gradually stabilized as the load nonlinearity, material nonlinearity and contact nonlinearity were con-
continued to increase. Results show that bolted flanges had better re- sidered. C40 concrete was modeled using the plastic damage model
sistance against lateral displacement and deformation energy dissipa- provided in ABAQUS. In the plastic damage model, the dilation angle
tion capacity compared to the welded flanges. and viscosity parameter of concrete were 35 deg and 0.005, respec-
tively. The elastic modulus E0 was 33, 200 MPa and Poisson's ratio νc
4. Establishment and verification of the finite element model was 0.2. The constitutive relationship of concrete was based on the
stress-strain curve recommended in Appendix C.2 of the Code for
4.1. Element and mesh Design of Concrete Structures [42]. The stress-strain curve is shown in
Fig. 16 and is defined as:
Three experimental specimens were numerically simulated by a fi- Concrete tension:
nite element software, ABAQUS. Fig. 15 shows the finite element model
(FEM) of a connection system. T3D2 truss element, which is a 2-node σ = (1 − d t ) E0 ε (10a)
linear 3-D truss element, was used for meshing the reinforced bar. 5
Considering the high accuracy of the model required to calculate the ⎧1 − ρt [1.2 − 0.2x ] x ⩽ 1
dt = ρt
displacement and distortion of the mesh element as displacement am- ⎨ 1 − α (x − 1)1.7 + x x > 1
⎩ t (10b)
plitude would increase, C3D8R element (8-node hexahedron linear re-
duced integral element) was used for meshing concrete, steel and bolts ε σ
x= , ρ = t0
[36]. In order to improve the calculation accuracy of the FEM of the εt0 t E0 εt0 (10c)
MPCJ specimens, and reduce the computation time [37,38], finer mesh
Concrete compression:
size was used for the joint core module, the SRC column, and the steel
beam near the connection area. The average mesh size was 5–35 mm. σ = (1 − d c ) E0 ε (11a)
Additionally, the mesh size away from the node position gradually in-
ρc n
creased by means of a single control of the maximum and minimum ⎧1 − n − 1 + xn
x⩽1
mesh sizes, mainly because it does not affect the calculation accuracy, dc = ρc
⎨1 − x>1
⎩ α c (x − 1)2 + x (11b)
as shown in Fig. 15.
ε σ E0 εc0
4.2. Contact interactions and boundary conditions x= , ρ = c0 , n =
εc0 c E0 εc0 E0 εc0 − σc0 (11c)

The contact relationships between steel and steel, steel and con- where dc and dt are damage factors of concrete under uniaxial com-
crete, and steel and bolt were established, and all contact pairs used pression and tension, respectively; d = 0 means no damage, and d = 1
finite sliding surface-to-surface contact. Among them, the normal con- means complete damage. εt0 and εc0 are the maximum tensile and
tact behavior was assumed to be ‘hard’ contact, and the tangential compressive strain, respectively, and σt0 and σc0 are the maximum
behavior was assumed to be friction contact with the friction coefficient tensile and compressive stress, respectively. In Fig. 16, εtu is the failure

12
C. Wu, et al. Engineering Structures 211 (2020) 110460

Fig. 15. Finite element model of the MPCJ specimens.

descending section, and the material was considered as isotropic. Both

the reinforcing bars and high strength bolts followed the bilinear ki-
nematic hardening rule and Von Mises yield criterion [43]. The elastic
modulus Es and Poisson's ratio νs were 206,000 MPa and 0.30, respec-
tively. The stress-strain relationship of high strength bolts was used as
specified by the suppliers. For steel and reinforcing bars, the true stress
σtrue–true strain εtrue was derived from the nominal stress σnom–nominal
strain εnom relationship of the material tensile test results (Table 1). The
reason is to accurately describe the change of cross-section area of steel
and reinforcement under the action of external load when large de-
formation would occur. The true stress σtrue and true strain εtrue were
defined as [43]:
σtrue = σnom (1 + εnom ) (12)

εtrue = ln(1 + εnom ) (13)

σnom = Δl/ l 0 (14)

Fig. 16. σ -ε curve for concrete as used in the finite element model [43]. εnom = F / A0 (15)
where Δl is the change in length of the steel sample, l0 is the initial
tensile strain, εc,e0 and σc,e0 are the elastic ultimate compressive strain length of the sample, F is the load, and A0 is the initial cross-sectional
and stress, respectively; εcu and σcu are the failure strain and stress, area of the sample.
respectively; εc,in is the inelastic (crushing) strain of the undamaged
material; εc,el is the elastic strain; and εc,pl is the compressive equivalent 4.4. Verification of the finite element model
plastic strain.
The constitutive model of Q345B steel was a tri-linear model with a As revealed by the finite element analysis, the damage mechanism
was similar for the three MPCJ specimens with different connection

13
C. Wu, et al. Engineering Structures 211 (2020) 110460

Table 6 flange P1 supported by flange connection plate P11 or welding seams

Comparison of results of MPCJ specimens from the experiment and the FEM P12 was:
analysis.
P11 = Afy = 1300 × 345 = 448.5 kN (16)
Specimen Yield load (kN) Relative Maximum load (kN) Relative
number error (%) error(%)
Test FEM Test FEM
P12 = ff w h e f ∑ l w = 200 × 7 × 130 × 3 = 546 kN (17)

MPCJ1 123.42 118.09 4.32 185.11 180.87 2.29 P1 = {P11, P12}min × h 0 / l

MPCJ2 151.80 144.22 4.99 236.90 244.40 3.17
= 1300 × 345 × 290/935 = 139.11 kN (18)
MPCJ3 79.56 81.67 2.65 178.06 169.72 4.68
b
The design shear capacity of one M24 bolt Nv was:

forms. As shown in Fig. 6, the load-displacement curves calculated by Nvb = αR kn f μP

finite element method were in agreement with the experimental results. = 0.9 × 2 × 0.35 × 225 = 141.75 kN (19)
The results show that the proposed FEM can better reflect the cyclic
Thus, the bearing capacity (of the joint) at the steel beam web V1
performance of the specimens. At the same time, the yield loads cal-
supported by three bolts was:
culated by FEM and test were compared with the limiting values, as
shown in Table 6. The calculated values for MPCJ specimens agree well V1 = 3Nvb = 3 × 141.75 = 425.25 kN (20)
with the experimental values (Table 6). For example, the absolute value
of the relative error of yield load was between 3.27% and 4.96%, and (2) For welded specimens (MPCJ2)
the absolute value of the relative error of ultimate load was between
2.87% and 8.27%. According to Eq. (9), the test value and simulation The bearing capacity at the steel beam flange P2 and web V2 sup-
value of equivalent viscous damping coefficient of MPCJ1, MPCJ2 and ported by welding seams was:
MPCJ3 specimens were calculated. The test value of the equivalent
P2 = P1 (21)
viscous damping coefficient of the three specimens was 2.25, 1.50 and
2.01 respectively, and the simulation value was 2.32, 1.45 and 2.03,
V2 = f fw he f ∑ l w = 200 × 7 × 220 × 2 = 616 kN (22)
respectively. By comparison, the error ratio between the test value and
the simulation value was between 1% and 4%, which further reflects
the good accuracy of the simulation model used in this study in terms of (3) For bolted specimens
energy dissipation.
The failure modes and failure mechanisms of the three specimens in The bearing capacity at the steel beam flange P3 and web V3 sup-
ABAQUS model were consistent with the test results. As illustrated in ported by bolts was:
Fig. 17, the flange connecting plates were ruptured after undergoing P31 = 4Nvb = 4 × 0.9 × 0.35 × 225000 = 283.5 kN (23)
buckling deformation. These failure modes included buckling and
fracture of flange connection plate, local buckling deformation of stif- P32 = A1 fy = (1500 − (22 × 10 × 2)) × 345
fening rib and joint cover plate, and significant slip between the bolt, = 365.7 kN (24)
flange connecting plate, and steel beam flange. By comparing the re-
sults of the finite element calculation with the test, the finite element P3 = {P31, P32}min × h 0 / l1 = 93.43 kN (25)
analysis could accurately simulate the deformation and stress state
during the test process. Hence, the finite element model is verified. V3 = V1 (26)
where A is the cross-sectional area of the flange connection plate; A1 is
5. Simplified calculation of the lateral force resistance the net cross-sectional area of the flange connection plate as in
Fig. 18(c); fc is the yield strength of the material; l is the distance be-
According to the analysis and final damage distribution of the three tween the center line of the flange connection plate and the loading
specimens, the weakest section of the joint was the central part of the point of the beam-end; l1 is the distance between the center line of the
flange connection plate. Therefore, the joint calculation in the numer- vertical bolt hole and the loading point at the end of the beam; ffw is the
ical analysis assumed that the steel beam flange and web, respectively, tensile and shear strength of the weld seams; and hef is the effective
took the tension and shear force caused by the external load thickness of the fillet weld seams. Similarly, Σlw is the sum of the ef-
[27,44–45]. fective length of the fillet weld on one side; h0 is the distance between
The bolt-weld specimen is sketched in Fig. 18(a). The tension gen- the center of upper and lower flange connection plate; and aR is the
erated by the beam-end moment was carried by the welded flange reciprocal of partial resistance coefficient (aR = 0.9) [16,20]. More-
connecting plate and the welding seam, and the shear force generated over, k is the pass coefficient (k = 1) [16,20]; nf is the number of
by the beam-end load was carried by the high-strength bolts. For the friction surfaces on a bolt; μ is the anti-slip coefficient of friction surface
MPCJ2 specimen, the tension generated by the beam-end moment was (μ = 0.35); and P is the pre-tightening force of high strength bolts.
carried by the welded flange connecting plate and the weld, and the The shearing capacities of the MPCJ1, MPCJ2, and MPCJ3 speci-
shear force generated by the beam-end load was carried by the con- mens were 425.25 kN, 616 kN, and 425.25 kN, respectively, as calcu-
necting web and the weld (Fig. 18(b)). For the bolted specimens lated based on Eqs. (20), (22), and (24). The results show that the web
(MPCJ3), all of the tension was carried by the flange connecting plate bolts or weld seams can bear much greater shear load compared to the
or the bolts on the flange connecting plate, and the shear force gener- steel beams. The yield moment capacities of the MPCJ1, MPCJ2, and
ated by the beam-end load was carried by the bolts (Fig. 18(c)). Me- MPCJ3 specimens were 139.11 kN, 139.11 kN, and 93.34 kN, respec-
chanical analysis and finite element calculations showed that the flange tively, based on Eqs. (18), (21), and (25). Compared with the experi-
connection plate had little effect on the connection module, and hence, mental results in Table 6, the relative error between the theoretical
it was ignored in the calculation diagram. calculation results and the experimental results was 12.71%, 8.36% and
17.32%, respectively, and the relative error between the theoretical
(1) For bolted-welded specimen (MPCJ1): calculation results and the FEM results was 17.80%, 3.54% and
14.29%, respectively, for the MPCJ1, MPCJ2 and MPCJ3 specimens.
Based on Fig. 18, the bearing capacity (of the joint) at the steel beam This shows that these calculation results are similar to the experimental

14
C. Wu, et al. Engineering Structures 211 (2020) 110460

Fig. 17. Typical modes of deformation and failure: (a) buckling and fracture of the flange connection plate of MPCJ1; (b) deformation of the stiffening rib and the
joint cover plate of MPCJ2; (c) buckling and slip on contact surface of the flange connection plate of MPCJ3.

and FEM results and the proposed equations could accurately represent The load-bearing state of the MPCJs could be divided into three
the test results. Therefore, the calculated results from the simplified stages, namely, an elastic stage, an elastic-plastic stage, and a
calculation Eqs. (16)–(26) can basically satisfy the test results. failure stage.
(2) The shear deformations of the core regions of the MPCJ1, MPCJ2
6. Conclusion and MPCJ3 specimens were (−2.9 to 12.3) × 10−3 rad, (−2.2 to
1.4) × 10−3 rad and (−0.25 to 0.25) × 10−3 rad, respectively.
Quasi-static tests and finite element analysis were performed to The joint rotation of the MPCJ specimens caused by shear de-
analyse the failure characteristics and seismic behavior of different formation was negligibly small compared to the total rotational
types of beam-to-column connections between a prefabricated SRC deformation.
column and a steel beam. Three types of joints were studied, namely, (3) The ultimate rotation angles of the MPCJ1, MPCJ2, and MPCJ3
MPCJ1 with bolt-weld hybrid connections, MPCJ2 with weld connec- specimens were greater than 0.04 rad, thus confirming that the
tions and MPCJ3 with bolt connections. According to the research re- joints met the ductility design requirements of FEMA-350 code.
sults, the main conclusions are as follows: (4) Different connection methods (bolt, weld or hybrid) had different
effects on the joint stiffness. MPCJ2 specimen with the welded joint
(1) The type of beam-column connection as studied in this study had behaved as a rigid, full strength joint. The bolt-weld hybrid con-
little influence on the ultimate failure mode of modular pre- nection in MPCJ1 specimen and the bolt connection in MPCJ3
fabricated composite joints, and all MPCJs underwent bending specimen behaved as semi-rigid and partially connected joints.
failure at the end of the beam. The type of connection, however, (5) MPCJs had good seismic performance. The displacement ductility
significantly affected the seismic performance, the load bearing coefficients of MPCJs were all greater than 4 and the equivalent
capacity, and the failure characteristics of individual components. viscous damping coefficients were all greater than 0.24.

15
C. Wu, et al. Engineering Structures 211 (2020) 110460

Fig. 18. Analytical calculation diagram of joints: (a) Bolted-welded joint; (b) Welded joint; and (c) Bolted joint. (Unit: mm).

(6) At the later stage of loading, the strength and stiffness degradation University of Technology are sincerely acknowledged for their valuable
performance of the MPCJ specimens were not observed to be help and advice during this study.
greatly reduced, indicating that the proposed joint design is rea-
sonable and the ductility is good. References
(7) The results from finite element analysis agreed well with the ex-
perimental results, which verified the accuracy of the finite element [1] Choi H-K, Choi Y-C, Choi C-S. Development and testing of precast concrete beam-to-
model. It forms a basis for further analysis of the key parameters of column connections. Eng Struct 2013;56:1820–35.
[2] Kim Sunkuk, et al. The development of modularized construction of enhanced
MPCJ specimens. precast composite structural systems (Smart Green frame) and its embedded energy
(8) Simplified equations were proposed in this paper which provided efficiency. Energy Build 2013;66(5):16–21.
results quite consistent with the test results and the finite element [3] GB/T 50378, Assessment Standard for Green Building, China Building Industry
Press, Beijing, China, 2013 [in Chinese].
analysis results. This demonstrated that the equations can be used [4] Park Seon-Chee, Hong Won-Kee, Kim Sunkuk, et al. Mathematical model of hybrid
to calculate the strength of MPCJ specimens under cyclic loading. precast gravity frames for smart construction and engineering. 14 pages Math
(9) In summary, different connection modes of the beam-end had a Problems Eng2014:916951https://doi.org/10.1155/2014/916951.
[5] Eom T, Park H, Hwang H, et al. Plastic hinge relocation methods for emulative PC
significant influence on the overall stress and deformation char- beam-column connections. J Struct Eng 2016;142(2):4015111.
acteristics of the specimens and on the shear deformation of the [6] Parastesh H, Hajirasouliha I, Ramezani R. A new ductile moment-resisting con-
joint module. Therefore, it is necessary to further optimize the nection for precast concrete frames in seismic regions: an experimental investiga-
tion. Eng Struct 2014;70:144–57.
variables of the joint module, and to improve the connection
[7] Deng KL, Wang T, Kurata Masahiro, Zhao CH, et al. Numerical study on a fully-
structure of the beam-end, such as adopting a ‘dog bone’ type or a prefabricated damage-tolerant beam to column connection for an earthquake-re-
rotating hinge. silient frame. Eng Struct 2018;159:320–31.
[8] Mou B, Zhao F, Qiao QY, et al. Flexural behavior of beam to column joints with or
without an overlying Concrete Slab. Eng Struct 2019;199(11):109616.
Declaration of Competing Interest [9] Mou B, Li X, Qiao QY, et al. Seismic behaviour of the corner joints of a frame under
biaxial cyclic loading. Eng Struct 2019;196(10):109316.
The authors declare that they have no known competing financial [10] Qiao QY, Zhang WW, Mou B, et al. Seismic behavior of exposed concrete filled steel
tube column bases with embedded reinforcing bars: experimental investigation.
interests or personal relationships that could have appeared to influ- Thin-walled Struct 2019;136(03):367–81.
ence the work reported in this paper. [11] Deng K, Zheng D, Yang C, et al. Experimental and analytical study of fully pre-
fabricated damage-tolerant beam to column connection for earthquake-resilient
frame. J Struct Eng 2019;145(3):04018264.
Acknowledgements [12] Jiang ZQ, Yang XF, Dou C, et al. Cyclic testing of replaceable damper: earthquake-
resilient prefabricated column-flange beam-column joint. Eng Struct
The writers gratefully acknowledge the financial support for this 2019;183:922–36.
[13] Jiang ZQ, Dou C, Zhang AL, et al. Experimental study on earthquake-resilient
work, which was provided by the National Natural Science Foundation prefabricated cross joints with L-shaped plates. Eng Struct 2019;184:74–84.
of China (51078195) and the Shandong University Scientific Research [14] Zhang AL, Zhang H, Jiang ZQ, et al. Low cycle reciprocating tests of earthquake-
Program (A2018-065). Dr. Zhou Fei and Mr. Tan Wenya from Qingdao resilient prefabricated column-flange beam-column joints with different connection

16
C. Wu, et al. Engineering Structures 211 (2020) 110460

forms. J Constr Steel Res 2020;164:105771. [30] FEMA. Recommended Seisimic Design Criteria For New Steel Moment Frame
[15] Kim T, Kim J. Collapse analysis of steel moment frames with various seismic con- Buildings, FEMA-350. Washington D.C.: Federal Emergency Management Agency;
nection. J Constr Steel Res 2009;65(16):1316–22. 2003.
[16] Miller DK. Lessons learned from the Northridge earthquake. Eng Struct [31] AISC. Seismic provisions for structural steel buildings. Chicago (IL, USA): American
1998;20(4):249–60. Institute of Steel Construction; 2005.
[17] Wu CL, Liu JM, Tan WY, et al. Seismic behavior of composite interior joints of [32] Fang C, Yam MCH, Lam ACC, et al. Cyclic performance of extended end-plate
prefabricated H-shaped steel reinforced concrete column - steel beam. Structures connections equipped with shape memory alloy bolts. J Constr Steel Res
2020;23:558–72. 2014;94:122–36.
[18] Nzabonimpa JD, Hong W-K, Kim J. Experimental and non-linear numerical in- [33] Augusti G. Dynamics of structures: theory and applications to earthquake en-
vestigation of the novel detachable mechanical joints with laminated plates for gineering. Tsinghua University Press; 2005.
composite precast beam-column joint. Compos Struct 2017. https://doi.org/10. [34] ATC-40. Seismic evaluation and retrofit of concrete buildings. California: Applied
1016/j.compstruct.2017.11.024. Technology Council; 1996.
[19] CEN (Comité Européen de Normalisation). Design of composite steel and concrete [35] Zhong ST, Bai GL. Analysis and design of beam-column joints in composite frame
structures - Part 1-1: General rules and rules for buildings. EN 1994-1-1:2004. structures of high-rise buildings. Beijing, China: China Communications Press;
Brussels, Belgium: CEN; 2004. 2006. [in Chinese].
[20] JGJ138-2016. Code for design of composite structures. Beijing, China: China [36] Bursi O, Jaspart JP. Benchmarks for finite element modelling of bolted steel con-
Architecture & Building Press; 2016 [in Chinese]. nections. J Constr Steel Res 1997;43(1–3):17–42.
[21] Hasaballa Mohamed H, El-Ragaby A, El-Salakawy EF. Seismic performance of ex- [37] Song YC, Wang RP, Li J. Local and post-local buckling behavior of welded steel
terior beam-column joints reinforced with glass fibre reinforced polymer bars and shapes in partially encased composite columns. Thin-Walled Struct
stirrups. Can J Civ Eng 2011;38(10):1092–102. 2016;108:93–108.
[22] Yan QS, Chen TY, Xie ZY. Seismic experimental study on a precast concrete beam- [38] Nzabonimpa JD. Won-Kee Hong. Experimental investigation of hybrid mechanical
column connection with grout sleeves. Eng Struct 2018;155:330–44. joints for L-shaped columns replacing conventional grouted sleeve connections. Eng
[23] MOHURD (Ministry of Housing and Urban-Rural Development Republic of China). Struct 2019;185:243–77.
JGJ/T101-2015. Specification for seismic test of buildings. Beijing, China: China [39] Ataei A, Bradford MA, Liu X. Experimental study of flush end plate beam-to-column
Architecture & Building Press; 1996. composite joints with precast slabs and deconstructable bolted shear connectors.
[24] Li W, Han LH. Seismic performance of CFST column to steel beam joints with RC Structures 2016;7:43–58.
slab: analysis. J Constr Steel Res 2011;67(1):127–39. [40] Rabbat BG, Russell HG. Friction coefficient of steel on concrete or grout. J Struct
[25] Stoakes CD, Fahnestock LA. Cyclic flexural testing of concentrically braced frame Eng 1985;111(3):505–15.
beam–column connections. J Struct Eng. 2011;137(7):739–47. [41] Gholami M, Deylami A, Tehranizadeh M. Seismic performance of flange plate
[26] Liu XC, Zhan XX, Pu SH, et al. Seismic performance study on slipping bolted truss- connections between steel beams and box columns. J Constr Steel Res
to-column connections in modularized prefabricated steel structures. Eng Struct 2013;84:36–48.
2018;163:241–54. [42] GB500010-2010, Code for design of concrete structures. Beijing, China: China
[27] CEN (Comité Européen de Normalisation). Design of steel structures - Part 1-8: Architecture & Building Press; 2010 [in Chinese].
Design of joints. EN 1993-1-8:2005. Brussels, Belgium: CEN; 2005. [43] Abaqus V6.14. Analysis User’ s Manual. SIMULIA, Providence, IR, USA: Abaqus Inc.;
[28] Shi G, Shi YJ, Wang YQ. Calculational method of shear deformation for beam-to- 2013.
column connection panel zone in steel frames. Jilin Daxue Xuebao (Gong xue ban)/ [44] GB50017-2017. Standard for design of steel structures. Beijing, China: China
J Jilin Univ (Eng Technol Ed) 2006;36(4):462–6. [in Chinese]. Planning Press; 2017 [in Chinese].
[29] Salvatore W, Bursi OS, Lucchesi D. Design testing and analysis of high ductile [45] JGJ82-2011. Technical specification for high strength bolt connections of steel
partial-strength steel-concrete composite beam-to-column joints. Computer Struct structure. Beijing, China: China Building Industry Press; 2011 [in Chinese].
2005;83:2334–52.

17