Engineering Structures 234 (2021) 111875

Contents lists available at ScienceDirect

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

Seismic study on an innovative fully-bolted beam-column joint in

prefabricated modular steel buildings
Zhihua Chen a, b, Xiangyuan Niu b, Jiadi Liu b, *, Kashan Khan b, Yuecong Liu b, c
a
State Key Laboratory of Hydraulic Engineering Simulation and Safety, Tianjin University, Tianjin, China
b
Department of Civil Engineering, Tianjin University, Tianjin, China
c
Tianjin Yuantai Modular House Manufacturing Co., Ltd, China

A R T I C L E I N F O A B S T R A C T

Keywords: Prefabricated modular steel beam-column joints have a broad developing prospect due to their quick con­
Beam-column joint struction speed and standardization. A new type of site-bolted assembled joint has been developed. Compared to
Fully-bolted connection the traditional extended-endplate joints, it has a horizontal base plate welded to the column and can be inte­
Cyclic tests
grated at floor level. In order to accurately evaluate the seismic performance of the new type joint under quasi-
Seismic behavior
Numerical analysis
static loads, a total of three tests (two beam-separation joints and a beam-through joint) were conducted on the
full-scale interior column joints, obtaining the curve of hysteretic and stiffness degradation. In addition, the
energy dissipation capacity of the joint was recorded and studied herein. It is demonstrated that the new type of
joint has good seismic performance and can be used as a new form of standardized H-steel for prefabricated
modular steel buildings. The extended plate and the strengthening of the panel zone can lead to the plastic hinge
of the beam end to move away from the joint region. The key parameters that affect the performance of the panel
zone are studied through the advanced finite element model (FEM) and the parametric analysis using Abaqus. It
was noted that the load-bearing capacity could be significantly improved by increasing the thickness of the panel
zone web and the beam flange. Finally, the simplified design of the joint was completed based on the experi­
mental and numerical studies and Chinese code GB 51022–2015, to meet the engineering requirements.

1. Introduction better than ordinary joints. The ductility of carbon steel joints is better.
However, this type of joints has unfavorable factors such as the defor­
Fast assembling routine and good seismic performance are two key mation of the endplate leading to the bolt prying force [9], the endplate
factors for beam-column joints in prefabricated modular buildings. The welded at the beam end, which will form a heat-affected zone and affect
joints connected by the cover plate or the endplate with site-bolted as­ performance, etc. [10], The endplate joints will also affect the bearing
sembly can be quickly assembled [1]. Compared with the traditional capacity due to the initial defects of the column [11]. Some scholars
bwwfs (bolted web-welded flange) joint [2–4], no on-site welding is [12,13] improved the endplate joints, using T-shaped connectors to
needed, and it can be applied to multi-story prefabricated buildings. connect the flanges of beams and columns. The T-shaped connectors
The extended endplate joint is currently one of the common H-steel make the appearance of plastic hinges far away from the panel zone,
beam-column joints, which can be quickly assembled, and correspond­ which is more convenient for rapid assembly. The endplate joints have
ing research and application are extensive [5,6]. Lama et al. studied the improved rotation performance and ductility. Gerama showed that the
endplate joints subjected to bending and axial force [7], and applied the arrangement of bolts would affect the joint failure mode [12].
component method recommended by Eurocode 3 to analyze the joints; The cover-plate reinforced joint is different from the extended end­
Shi et al. researched endplate thickness and connected bolts on the plate joint. The horizontal cover-plate is employed, and it plays the role
joints. The parameter analysis of factors such as the diameter of bolts of strengthening the panel zone and connecting the beam and column
under the earthquake load shows that the joint has sufficient rotation [14]. Cover-plate reinforced joints are also the form recommended by
and energy dissipation capacity[8]; Gao et al. analyzed the stainless FEMA-350 [15], the fully-bolted joint is also allowed by the standard
steel endplate beam-column joint [8], which shows that stainless steel is AISC358-05 [16], and related research has been continuously advancing

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

https://doi.org/10.1016/j.engstruct.2021.111875
Received 2 September 2020; Received in revised form 1 January 2021; Accepted 6 January 2021
Available online 15 February 2021
0141-0296/© 2021 Elsevier Ltd. All rights reserved.
Z. Chen et al. Engineering Structures 234 (2021) 111875

in recent years: Jazany studied 8 beam-column joints of welded rein­ 3. Experimental design
forced cover-plates, reinforced fillet welds and plug welds can improve
the unfavorable deformation of the cover-plate and beam flange [17]; 3.1. Design of the specimen
Liu et al. studied the square steel tube column (HSS)-H-steel beam joint
and HSS-truss beam joint [1,18], The connection connects a beam to a In this experiment, three test specimens were designed to study the
column through cover-plates that extend from the flanges as well as seismic performance of a fully-bolted joint by changing the joint
through the vertical connecting plates and the joint flitches, The connection mode, panel zone web thickness, and other parameters. The
warpage of the column endplates is conducive to energy dissipation, and assembly of the specimen is connected by bolts on the site (the assembly
the excellent plastic deformation capacity is very suitable for multi-story in the field is the same as the three parts in Fig. 2a, the splice plate has
modular buildings; Wang Wei et al. [19–21] designed a composite sys­ been welded to connector in the factory) to simulate the assembly in the
tem of tension-only concentrically braced beam through frames construction site. The whole specimen is taken from the interior column
(TOCBBTFs), different from the traditional H-steel frame employed the joints; its scale has been determined by interrupting the frame at the
through column, each layer of the joint is disconnected, and the beam inflection point. The reason to use the section size of H-steel because it
flange is connected by the column base plate. The braces are set up has been extensively used in steel structures in China and is appropriate
between the columns, and the two full-scale 2-story frames are simu­ for standardization. (See the detailed size and structure of the specimen
lated statically. However, in the seismic design, support is the primary in Figs. 2, 3 and Table 1).
energy dissipation part. The lateral seismic resistance of the beam
through the frame is limited, and the hysteresis curve is also obviously 3.2. Material properties test
pinched. The beam-through joint is a hinged joint to some extent.
In summary, apart from the issues as mentioned earlier, the existing The H-steel selected for this test is a hot-rolled H-steel, and the steel
H-shaped steel beam-column joints have disadvantages such as the type is Q345B. To determine the material properties of steel, standard
inability to guarantee on-site welding quality, workability, uninfluential material specimens are made from H-steel flanges and webs. According
to overall structural stability, difficulty in rapid assembly, and bolt to the national specification GB/T228.12010 (tensile test for Metal
prying action. Therefore, the research was purposely conducted in need Materials – Part 1: Room temperature Test Methods.) [22], Yield
to develop an innovative fully-bolted beam-column joint, aiming to strength, tensile strength and modulus of elasticity are measured by the
serve prefabricated multi-story structures without compromising on tensile testing in a UTM (Universal testing machine). The mean value of
their aesthetics, serviceability, strength, stability, and seismic perfor­ the three sample results is shown in Table 2.
mance. In order to facilitate rapid assembly, all components are made of
standardized H-steel; to assure the joints have good seismic performance
and meet the requirements of strong columns and weak beams, the 3.3. Test plan
design joints are connected by horizontal cover plates, which can
strengthen the core area (panel zone). The cyclic test of 3 full-scale The test was carried out in the structural engineering laboratory at
interior joints and corresponding parametric studies with validated Tianjin University, China. The bottom of the column specimens was
numerical technique were carried out to suggest design recommenda­ supported by a one-way hinge pedestal, which was fixed on the rein­
tions for engineering practices. forced concrete ground. The column was horizontally connected by two
pieces of the rigid plate, with one end fixed on the reaction wall, while
2. Structure of the joint the other by four pillars to as a screw connection. To provide the hori­
zontal reaction, forming a one-way hinge support boundary conditions,
Fully-bolted joints are mainly composed of H-steel column, H-steel the rigid chucks with two hinged support holes fixed at the beam ends
beam, connector, column bottom plate(base plate), splice plate, and are connected with the vertical loading device. A 2000 kN oil pressure
stiffener, etc. The assembly diagram of joints is shown in Fig. 1. The jack provides constant axial pressure on the cylinder top; the vertical
connector is a section of H-steel with bolt holes in flanges in a standard cyclic load at the beam end is supplied by 300 kN tension and
size. The column bottom plate and the splice plate are rectangular steel compression jacks on both ends. Two pairs of pull-bolt tripods that can
plates with bolt holes as per the standard size. Both the connector and be fixed on a rigid floor are used to fix the lateral support of the beam
splice plate are welded in the factory. The web of the beam is joined with and prevent the beam ends from twisting during loading. The test device
a splice plate, while flanges with column bottom plate on the site. On the is shown in Fig. 4, Fig. 5.
other hand, the beam-through joint has no connectors, and the rest of
the structure is the same. 3.4. Loading history

The cyclic loading was according to the code for Seismic Test
Methods for Buildings (China JGJ 101-2015) [23]. The load was carried

Fig. 1. Structural diagram of the joint.

2
Z. Chen et al. Engineering Structures 234 (2021) 111875

Fig. 2. Design of the specimen.

out by using a combination of force and displacement control, as shown The difference between specimens JD3 and JD1 is that the thickness
in Fig. 6. The cyclic loading system was divided into multi-stage loading; of the web in the panel zone of the connector changed from 6.5 mm to
stage I by the force control, stage II by displacement control after two 26.5 mm, and two 10 mm vertical stiffeners at each edge of column were
times of the yield displacement loading for the two loops per level. added at the column base plate. When loading to 27.8kN, the corre­
sponding displacement △Y is 14.5 mm. After the cycle from the fourth
• At the beginning of loading, the yield capacity (p)was estimated by level to the fifth level, the bolts at the splice plate have a slight slip, the
numerical simulation. While the inflection point appears in the flange of the beam has small bending deformation. At this point, the
load–displacement curve, the corresponding displacement is defined flange of the beam produces a plastic hinge, and the bolt slip is further
as yield displacement (△Y). apparent. At the later stage of loading, the out-of-plane deformation of
• When the bearing capacity drops to less than 85% of the ultimate the right beam flange is obvious, the column bottom plate and the flange
bearing capacity, the plastic hinge failure occurs, or the joints pro­ of the connector are no longer tightly fitted, resulting in gaps, and the
duce large cracks, the specimen is considered to be failed, and the connector web has no obvious deformation. Finally, the specimen was
loading was stopped. loaded to 150.06 mm. The deformation of the specimen was evident,
• Pre-loading was conducted and lasted five minutes before the formal and the test was stopped.
force loading was carried out. The pre-loading test aiming to test Through the analysis of the experimental phenomena of three
whether the whole system works well. specimens (as shown in Figs. 7–9, Table 3), the failure modes of the
specimens can be summed up in the following two modes; the first form
into connector web shear failure, with the further increase of load, the
3.5. Test phenomena and failure modes connector web start producing out-of-plane deformation, beam flange
edge bending deformation, such as specimen JD1 and JD2. The second
Specimen JD1 is a standard specimen of prefabricated fully-bolted failure mode is the connector web without obvious deformation, and the
connected beam-column joints. It initially applies an axial force to the beam flange first produces slight bending deformation, with the further
top of the column and then carries out beam end loading after holding increase of the load, the beam flange shows drum failure near the col­
the top column end loading. At the initial stage of loading (the first three umn bottom, forming a plastic hinge, such as specimen JD3. Bolt slip­
levels of loading), an inflection point appears in the beam end curve, and page mainly occurs at the joints with connectors, and the significant
at this point, the joints yield corresponding to △Y is 12.5 mm. Then, the slippage is at the joints connected with the beam, which is related to
load–displacement curve changed steadily, the bolts at the beam flange shear force transferred by bolts.
had a slight slip. After the third level loading, the bolt slippage is further
increased, the cracks at the painted surface of the connector web occurs, 4. Test result
followed by the outer and upward drum deformation at the connector
web under shear. At the later stage of loading (after the fifth level), an 4.1. Hysteretic and skeleton curve
obvious plastic hinge was generated at the connection between the beam
flange. It was accompanied by local cracks that appeared in the Fig. 10 is the hysteretic curve of the specimen. It can be seen that
connection, an obvious drum deformation occurred in the panel zone of specimen JD1 and JD2 have the same maximum bearing capacity. Still,
the specimen, and a continuous slight sound was generated. The seventh the hysteretic curve of JD1 and JD3 is shaped like an “arch” with visible
level of the specimen was cyclically loaded to 89.5 mm, and the test was pinching, the possible reason is the bolt slippage, and apparent bolt
stopped. slippage in JD1 accounts for more evident pinching, and the energy
The difference between specimens JD2 and JD1 is that the beam is dissipation capacity decreases compared with JD2. It can be noted that
integrated (throughout) at the joint, and there are no connectors. When the maximum bearing capacity of JD3 is higher than that of JD1.
the load reaches the third level, an inflection point appears in the Similarly, the area surrounded by the hysteresis curve has also signifi­
load–displacement curve at the beam end, and the joint yields, the cantly increased, indicating that strengthening the thickness of the panel
corresponding displacement △Y is 15.9 mm. Subsequently, there was zone contributes to the improvement of the bearing capacity and energy
slight shear deformation in the panel zone, cross cracks in the paint dissipation capacity of the joint. Compared with the test results of
surface of the web appeared, small bending deformation in the column Wangwei et al. [19], the pinch phenomenon in the developed joint has
base plate, and no sliding of bolts. At the middle stage of loading (after been improved.
3△Y), there was an obvious shear deformation observed in the panel By comparing the skeleton curve of the specimen in Fig. 10 and the
zone of the joint, and the web of the panel zone showed out-of-plane joint load–displacement data in Table 4, it can be seen that initial
buckling. After the fifth stage cyclic loading of the specimen (4△Y), stiffness and maximum bearing capacity of JD1 and JD2 are the same.
the shear deformation in the panel zone was further enlarged, and an However, the peak displacement of JD1is slightly larger than that of
obvious plastic hinge was generated at the flange of the beam, and the JD2, because bolt slip will affect the displacement of the joint to the
test was stopped.

3
Z. Chen et al. Engineering Structures 234 (2021) 111875

(a) beam-separation joints (JD1, JD3)

(b) beam-through joints (JD2)

Fig. 3. details of specimens. (panel zone thickness of JD1 is 6.5 mm, JD3 is 26.5 mm).

maximum bearing capacity. The initial stiffness of JD3 was significantly tween the stiffness under the i-th loading level and the initial stiffness of
increased compared with JD1, and the maximum bearing capacity of the the joint [18]. The variation trend of stiffness degradation coefficient
joint was also changed from 23.27kN to 63.76kN, indicating that the with beam end displacement is shown in Fig. 11a. The stiffness of all
strengthening of the panel zone of the joint domain not only improved three specimens decreased uniformly, indicating that it does not incline
the joint stiffness but also increased the maximum bearing capacity of to generate large yield deformation due to stiffness degradation at the
the joint. beam-column joints. The overall stiffness degradation of specimen JD1
and JD3 is greater than that of JD2, and the addition of connectors
4.2. Stiffness and strength degradation makes joint stiffness degradation faster. Compared with JD1, degrada­
tion of JD3 decrease slower at the early stage of loading, but apparent
The stiffness refers to average loop stiffness (Equation 2), and the deformation of beam flange accounts for further degradation.
stiffness degradation coefficient ξj is selected to represent the ratio be­

4
Z. Chen et al. Engineering Structures 234 (2021) 111875

Table 1
parameter of beam-column joint.
No. Column Beam Form of the Web thickness vertical
(mm) (mm) panel zone in the panel stiffeners in
zone(mm) column
bottom plate

JD1 300 × 200 × Connector 6.5 No stiffeners

150 × 6.5 150 ×
×9 6.5 × 9
JD2 300 × 200 × Beam 6.5 No stiffeners
150 × 6.5 150 × through
×9 6.5 × 9
JD3 300 × 200 × Connector 24.5 10 mm
150 × 6.5 150 ×
×9 6.5 × 9

Fig. 6. Loading history.

Table 2
Material properties for specimen.
Kj
Component t fy fu Es (Gpa) ξj = (1)
(mm) (N/mm2) (N/mm2) K0
⃒ ⃒ ⃒ ⃒
Flange 6.5 372.35 541.03 204.9 ⃒ ⃒ ⃒ ⃒
⃒ + Fji ⃒ + ⃒ − Fji ⃒
Web 9 368.73 537.87 204.7
Kji = ⃒⃒ ⃒ ⃒ ⃒ (2)
Column bottom plate 15 385.52 539.36 205.0 ⃒ ⃒ ⃒
⃒ + Uji ⃒ + ⃒ − Uji ⃒

The strength degradation coefficient ηj is selected to represent the

ratio between the load value under the j-th load level and the maximum
load value under the j-th load level, as shown in Fig. 11b. The ultimate
strength degradation coefficient is between 0.89 and 0.93, and the cyclic
loading does not significantly degrade the joints strength.
Remarkably, the strength degradation curve of JD3 showed a trend
of first decreasing and then increasing before and after the third stage
loading. The possible reason was that many areas entered plasticity
during the second cycle of the third stage loading, which had a particular
impact on the structure resistance and resulted in a steep decline in
strength degradation[24]. In the late loading stage, the degradation of
JD3 is slower than JD1 and JD2.

4.3. Ductility and energy dissipation

Ductility coefficient μ is taken as the ratio of the ultimate displace­

ment at failure and the displacement at yield [24]. Yield displacement is
Fig. 4. Sketch of the test device.
determined by the Park method and using 75% of the peak load [25].
The data processing results of the specimens are shown in Fig. 12b. The
ductility coefficient of this test specimen is between 2.08 and 3.07, and
JD3 is the largest. The reason is that the JD3 generated a plastic hinge at
the beam end, and the ductility of the joint was better. The ductility of a
new joint is better than that of the beam through the joint.
The equivalent viscous damping coefficient he was used to measure
the seismic performance of each specimen. The equivalent damping
coefficient of the three specimens is around 0.2, among which JD2, JD3,
is slightly larger than JD1, indicating that strengthening the connector
web thickness can effectively improve the energy dissipation perfor­
mance. The energy dissipation performance of beam-column joints is
reflected by the maximum hysteresis loop area, as shown in Fig. 12a.
Although the equivalent damping coefficient of the three joints is
similar, the hysteresis loop area increases significantly after the
connector web thickness is strengthened. The new joints can meet the
seismic requirements of engineering structures and adjust the seismic
fortification requirements of different buildings by improving the panel
zone and the column base plate.

4.4. Strain analysis

Fig. 5. Test device.
This section takes the specimen JD1 as an example to analyze the
deformation of each member through the strain data of four parts: beam
flange, column base plate, column flange and web in a panel zone

5
Z. Chen et al. Engineering Structures 234 (2021) 111875

Fig. 7. Test Phenomenon of JD1.

Fig. 8. Test Phenomenon of JD2.

Fig. 9. Test Phenomenon of JD3.

(shown in Figs. 14–16). The arrangement of measuring points is shown panel zone was dominant. Under the same displacement, the strain of
in Fig. 13, in which strain rosette is pasted near the web in the panel the middle strain rosette was greater than that of the corner strain
zone, and the rest are strain gauges.*** rosette. In the observation test, the deformation of the central web is the
Two strain rosette in the corner and middle of the web in the panel largest, and the flange of the connector and the vertical stiffener will
zone were selected for analysis. The variation trend of strain rosette in limit the angular strain of the connector web.
the corner and middle of the web was consistent. The principal strain √̅̅̅ √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
was in the direction of 45◦ , indicating that the shear deformation in the ε + ε90 2
εmax = 0 + (ε0 − ε45 )2 + (ε45 − ε90 )2 (3)
2 2

6
Z. Chen et al. Engineering Structures 234 (2021) 111875

Table 3 is much larger than that of strain gauge No. 19 and 20 at the distal end.
Test phenomena. The possible reason is that strain gauge No. 16 and 17 are located at the
JD1 JD2 JD3 intersection of the beam and joint, where the stress of the beam flange is
wholly transferred from column bottom to joint panel zone. The strain of
△Y 12.5 mm 15.9 mm 14.5 mm
Bolt slippage Apparent No slippage Apparent slippage the edge strain gauge is slightly larger than that of the middle strain
slippage gauge, which may be due to the concentrated stress at the bolt hole. A
Drum Obvious drum Occurs No obvious total of 4 strain gauges were arranged at the bottom of the column
deformation in deformation deformation flange. As a whole, the column flange did not enter into the yield state,
panel zone
Plastic hinge Occur in column Occurs in beam Occur in beam
which was consistent with the final shape of the test column flange. The
bottom plate and flange flange, away from strain gauge near the end and the strain gauge in the middle are slightly
beam beam end larger, and the reason is that the stress in the center of the column flange
Ultimate 89.5 mm. 71.42 mm 150.06 mm is immense and diffuses to the edge, while the stress at the edge of the
displacement
flange is smaller.
Failure mode Connector web Connector web Plastic hinge
shear failure shear failure; beam occurs in beam
edge bending flange 5. Finite element analysis
failure
The test specimens were analyzed based on the modeling of finite
Four strain gauges were arranged symmetrically on the upper surface
of the flange of an H-steel girder. From an overall point of view, the Table 4
strain of the strain gauges steadily increased with the improvement of Load and displacement data.
the loading class, but none of them entered into the yield state, which No. Yielding Peak Ultimate
was consistent with the final shape of the flange of the test specimen. Py (kN) Δy (mm) Pmax (kN) Δmax (mm) Pu (kN) Δu (mm)
The strain near the root of the joint is found larger than that far from the
JD-1 11.62 30.83 20.48 76.94 18.03 89.45
root of the joint. Besides this, the strain on both sides of the flange is seen − 12.81 –33.95 –23.27 − 73.91 –23.27 − 73.91
slightly larger than that on the middle part. The reason is that there may JD-2 18.73 28.56 22.89 58.91 19.24 71.42
be local stress concentration in the area near the bolt hole. − 16.27 − 26.92 − 19.65 56.58 − 17.32 − 68.15
A total of 6 strain gauges are arranged on one side of the column base JD-3 41.89 37.48 63.53 115.06 55.38 150.06
− 45.26 − 49.25 − 65.69 − 114.03 − 63.76 − 134.11
plate. Through analysis, it can be found that strain gauge No. 16 and 17

Fig. 10. Hysteretic and skeleton curve.

7
Z. Chen et al. Engineering Structures 234 (2021) 111875

Fig. 11. Stiffness and strength degradation.

Fig. 12. Ductility and Energy dissipation.

Fig. 13. Arrangement of measuring points.

element software Abaqus, and a standard type solver was adopted. The trilinear strengthening steel model. The high strength bolts using the
approximate mesh size is 20 mm, and part of the area is encrypted. The perfectly elastoplastic model, and considering the high strength bolt pre-
component selection (C3D8I) by the 8-node entity unit model uses the tightening force. The boundary condition for the bottom of the column

8
Z. Chen et al. Engineering Structures 234 (2021) 111875

Fig. 14. Strain in connector web.

stage of loading, the initial stiffness is approximately the same. With the
increase of the load, the bearing capacity of the test specimen under the
same displacement is slightly less than that of the finite element analysis.
Because of the partial machining error of the model, part of the sliding
phenomenon occurs during the loading process, resulting in a decrease
of stiffness. The ultimate bearing capacity of the finite element analysis
and the test is the same; the error of the two is controlled below 6%
(Table 5). The variations in capacities between FE and tests can be due to
possible simplification in FE or material variation and flexible boundary
conditions during testing.
Fig. 19 compares the JD3 finite element model with the experimental
phenomena, and the final failure pattern is quite similar. The model well
simulates the bending of the column, the failure of the beam flange, and
the appearance of the plastic hinge at the beam end. Based on the von
Mises Stress contour of the specimens, JD3 generated beam end bending
moment and shear force under the vertical load of the beam end. The
force from the beam flange was transferred to the column base plate
through bolts, then to the core position through the column base plate,
and then to the column flange, so that the force acted on the reverse
Fig. 15. Strain in web.
bending moment of the joints kept the bending moment balance for the
joints. The overall load-transferring route is from the beam flange to the
column base plate to the panel zone and finally to the column flange.
constraints the displacement in all directions. The top of the column
The maximum stress is at the flange of the beam end, and it yields first in
restricts the displacement in the X and Y direction and the rotation in
the stress process. In the limit state, the whole flange end of the beam is
the X and Z direction. Low cyclic reciprocating loads are applied to the
in the yield state, which is consistent with the failure of the plastic hinge
beam ends. The actual model boundary condition diagram is shown in
formed at the beam end under the test condition.
Fig. 17.
As shown in Fig. 18, the test is in good agreement with the
load–displacement curve of the finite element analysis. At the elastic

Fig. 16. Strain in column.

9
Z. Chen et al. Engineering Structures 234 (2021) 111875

Fig. 17. The FE model.

Fig. 18. Hysteretic curve.

6. Parametric analysis on JD3 specimen, the key factors (i.e., the thickness of connector web,
the thickness of the beam flange, column bottom plate, stiffener, and
This section based on the finite element modeling of section 5, based axial compression ratio) on the mechanical properties of full-bolt

10
Z. Chen et al. Engineering Structures 234 (2021) 111875

Table 5 connections are studied. The influence of the above-mentioned param­

Comparison of FEM and test result. eters, and based on this, the analysis of the flexural bearing capacity of
N0. Direction Pm− FEM Pm− test Pm− FEM /Pm− test
the joint and the shear bearing capacity calculation method are
performed.
JD1 21.08 20.48 1.029
+
The skeleton curve of parametric analysis is shown in Fig. 20, and the
– − 24.67 –23.26 1.054
JD2 + 24.12 22.89 1.054 specific parameters and bearing capacity changes are shown in Table 6.
– − 20.75 − 19.65 1.056 In Fig. 20a, the parameter influence effect is normalized, and poly­
JD3 + 67.01 63.53 1.055 nomial fitting and extrapolation are performed. The term p0 is the
– − 67.35 − 65.68 1.025 bearing capacity of JD3 standard component (67.2kN), a0 is the
dimension parameter of JD3.
Note: JD3 parameters are as follows: the thickness of the connector
web is 24.5 mm, the thickness of the beam flange is 12 mm, the thickness

Fig. 19. Comparison of FEM and test result (JD3).

11
Z. Chen et al. Engineering Structures 234 (2021) 111875

Fig. 20. Skeleton curve of parametric analysis.

of the stiffener is 10 mm, and the axial compression ratio is 0.2. joint, and the additional thickness of the beam flange has little influence
It can be seen from the analysis that the thickness of the web in the on the maximum bearing capacity of the joint. In practical engineering,
panel zone of the joint has a significant influence on the bearing capacity it is more reasonable to strengthen the connector web thickness than the
and stiffness of the joint (±65%). Still, when the thickness reaches a beam flange for economic feasibility.
certain degree, the increase of the bearing capacity of the joint is found The effect of stiffeners on the column bottom plate is not significant
limited. It is because a rise in the thickness of the web in the panel zone (±2%). When the thickness of the column base plate meets the re­
enhances the stiffness of the web, the occurrence of shear failure of the quirements, the stiffeners have little effect on the maximum bearing
web is delayed. When the thickness is further increased, the failure mode capacity of joints, but they can partially improve the stiffness of joints.
is changed, and then the beam-end flange is destroyed. However, compared with the previous finite element analysis, it can be
The thickness of the beam flange also significantly influences the seen that when the stiffener is set at the column bottom, the stress
bearing capacity and stiffness (±30%). The maximum bearing capacity concentration at the column flange can be effectively avoided, results in
of the joint increases linearly with the increase of the thickness of the avoiding the brittle failure due to the stress concentration limited at the
beam flange. When the thickness of the beam flange reaches a specific column end.
value, the plastic hinge failure of the beam flange will not occur at the The thickness of the column base plate has a slight influence of

12
Z. Chen et al. Engineering Structures 234 (2021) 111875

Table 6
the result of parametric analysis.
No. Parameter(mm) Initial Peak load in positive direction Peak load in Negative direction Average load Rate of
stiffness (kN) (kN) (kN) change
(kN/mm)

JD3 / 1.73 67.0 − 67.3 67.2 /

FB- thickness of connector web = 18.5 1.56 62.8 − 63.1 63.0 − 6.3%
18.5
FB- thickness of connector web = 12.5 1.21 44.1 − 44.7 44.4 –33.9%
12.5
FB-6.5 thickness of connector web = 6.5 0.77 24.3 − 24.5 24.4 − 63.7%
LW-15 thickness of beam flange = 9 2.12 87.3 − 86.8 87.6 30.4%
LW-12 thickness of beam flange = 12 1.97 81.2 − 82.1 81.7 21.6%
LW-6 thickness of beam flange = 6 1.36 45.8 − 46.3 46.1 − 31.4%
JJL-5 thickness of stiffener = 5 1.73 66.8 − 67.2 67.0 − 0.3%
JJL-0 No stiffener 1.71 66.3 − 66.6 66.5 − 1.1%
DB-18 thickness of column base plate = 1.71 66.7 67.0 66.9 0.6%
18
DB-15 thickness of column base plate = 1.70 66.3 − 66.6 66.5 /
15
DB-12 thickness of column base plate = 1.65 63.7 − 64.0 63.9 − 3.9%
12
DB-9 thickness of column base plate = 9 1.58 59.2 − 59.1 59.2 − 10.9%
ZY-0.3 axial compression ratio = 0.3 1.74 66.5 − 66.5 66.5 − 1.0%

(±10%). As the tensile force of the beam flange transferred by the col­
umn base plate reaches the panel zone of the joint, with the decrease of
thickness, the column base plate will first produce yield failure, reducing
the bearing capacity of the joint. The thickness of the column base plate
should be checked in the design of the joint to ensure that the column
bottom plate will not fail first.
When the axial compression ratio is small(＜0.3), the axial
compression ratio has little influence on the bearing capacity of the joint
(±1%).

7. Design of joints

Based on the analysis in previous sections, it can be seen that the

following failure modes can occur for the fully-bolted joints: (1) plastic
hinge generated by yielding of the beam flange; (2) shear failure of the
bolt; (3) yield failure of the column base plate; (4) shear failure of Fig. 21. Force transfer mechanism of the joint.
connector web. To generate a plastic hinge at the beam end as a primary
failure mode, it is necessary to check the flexural and shear resistance of web in the panel zone is in the state of composite stress (Formula (5).
the joints when designing the size of joints. This paper refers to the Considering the influence of normal stress on the web, the shear force is
domestic code[26] and based on this, a simplified joint design is carried determined by equation (6):
out. √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
The force diagram in the joint is shown in Fig. 21. When the full σ 2 + 3τ2 ≤ fu (5)
section of the beam enters into plasticity, a plastic hinge will be gener­
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
ated. Bending moment is converted into shear force acting at the reverse tw hb f 2u − σ 2
bending point of the beam: V= √̅̅̅ (6)
3
V(L − b)
≤f (4)
γx wn
tw The thickness of the connector web
τ The shear stress on the connector web
V The design value of shear force at beam end hb The width of the connector web
b The distance from the plastic hinge of the beam to the splice plate fu Ultimate tensile strength of connector web
γx Section plastic coefficient (equal to 1.2) σ Compressive stress on the connector web
wn Modulus of the net section
L Distance from inflection point to panel zone.
Based on Formula (5), the thickness of the web in the panel zone can be
determined. For the design of the column bottom plate, the plastic
design method of the overreach endplate in the bolt joint of the cover-
According to Equation (4), the relationship between plastic hinge posi­
plate is referred to as the maximum value of formula (7)-(9) [26]and
tion and beam section parameters can be determined. In order to avoid
is taken for column bottom plate. t = max {t1 , t2, t3 }
the similar shear failure of JD1 and JD2 at the joint, the shear design
√̅̅̅̅̅̅̅̅̅̅̅̅
should be carried out on the connector web of the main transfer shear 6ef N bt
force. Because the influence of the transfer on the column bottom and t1 ≥ (7)
bf
other components is ignored, this design method is relatively safe and
conforms to the design method of strong shear and weak bending. The

13
Z. Chen et al. Engineering Structures 234 (2021) 111875

√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
Table 7
6ef ew N bt
t2 ≥ [ ( )] (8) Evaluation of JD1 and JD3.
ew b + 2ef ef + ew f
Specimen: JD1 V n1 n2 n3
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
3ew N bt Design capacity 18.06 kN 4 3 2
t3 ≥ (9) Test capacity 21.88 kN 4 6 2
(0.5a + ew )f
Specimen: JD3 V n1 n2 n3
Design capacity 66.33 kN 4 3 2
Test capacity 64.61 kN 4 6 2

Nbt The design value of tensile capacity of single friction type high strength bolt
Nbv The design value of shear capacity of single friction type high strength bolt
a smaller overall bolt slip, whereas the overall bearing capacity is
ef Distance from bolt hole center to column web
much smaller than the thickened panel zone reinforced joint (JD3).
ew The distance between the bolt hole center and the column flange
b Endplate width
4. The validity of finite element models was accurately assessed by the
a Center pitch of bolt hole comparison of the load–displacement curve, failure modes, and force
f The design value of the tensile strength of steel transmission behavior with the test results. Based on the reliable
n Number of high strength bolts required calibrated finite element model a parametric analysis revealed that
among thickness of panel zone, beam flange, column base plate and
stiffeners the thickness of the beam flange and web in the panel zone
It is required that the shear strength of the whole bolt group should be are two key parameters in enhancing the load-carrying capacity of
greater than the tensile yield strength of the steel. The number of bolts joints.
needed can be determined by equations (10)-(11). 5. The proposed simplified design method based on the Chinese Code
for the design of steel structure was found well consistent with the
Af = n Nvb (10) test results, which can validate the accuracy of models as a guide for
a subsequent engineering design of fully-bolted beam-column joints
Af = n Ntb (11) in prefabricated modular steel buildings.
Adopting the above design method, JD1 and JD3 have been checked,
and the results are shown in Table 7. The capacity of JD1 is determined CRediT authorship contribution statement
by the shear force, and the corresponding failure mode is a shear failure
in the connector web. The whole design method is conservative. (n1 is Zhihua Chen: Project administration, Funding acquisition, Valida­
number of bolt in beam flange; n2 is number of bolt in column base plate, tion. Xiangyuan Niu: Investigation, Writing - original draft, Software,
by one side; n3 is number of bolt in splice plate) Formal analysis. Jiadi Liu: Methodology, Supervision, Validation.
Kashan Khan: Writing - review & editing, Validation. Yuecong Liu: .
8. Conclusion
Declaration of Competing Interest
The study aimed to develop joints for H-shaped steel beams and
columns, which can assure easy-handling, avoid compromise on aes­ The authors declared that there is no conflict of interest.
thetics by adding web stiffeners, and improve overall structural stability,
aesthetics, serviceability, strength, stability, and seismic performance of Acknowledgements
multi-story modular steel buildings under the influence of both vertical
and severe lateral loadings. The developed joint assembles the columns This research was sponsored by the National Key Research and
and beams at the connector region through on-site bolting of the column Development Program of China (Grant No. 2019YFD1101005), the
base plate and beams splice plate, avoiding direct welding of structural National Science Foundation of Tianjin (Grant No. 19JCQNJC06600)
components to each other. To investigate the seismic performance of and China Postdoctoral Science Foundation (Grant No. 2020M670655).
developed joints, current study carried out. cyclic tests on three full-
scale frames of the interior joint. Experimental findings were then References
calibrated with Abaqus-based finite element numerical technique to
extend analysis on the parameters affecting the performance of joints. [1] Liu XC, Cui FY, Zhan XX, Yu C, Jiang ZQ. Seismic performance of bolted connection
Based on the detailed parametric analysis, a simplified design was pro­ of H-beam to HSS-column with web endplate. J Constr Steel Res 2019;156:167–81.
[2] Miller DK. Lessons learned from the Northridge earthquake. Eng Struct 1998;20:
posed, the accuracy of which was validated with experimental and nu­ 249–60.
merical findings. Following are the main conclusions drawnfrom the [3] Chen C-C, Lin C-C, Lin C-H. Ductile moment connections used in steel column-tree
study: moment-resisting frames. J Constr Steel Res 2006;62:793–801.
[4] Chen C-C, Lin C-C. Seismic performance of steel beam-to-column moment
connections with tapered beam flanges. Eng Struct 2013;48:588–601.
1. The fully-bolted beam-column joints connected by column base plate [5] Francavilla AB, Latour M, Piluso V, Rizzano G. Design of full-strength full-ductility
can meet the requirements of multi-story prefabricated modular steel extended endplate beam-to-column joints. J Constr Steel Res 2018;148:77–96.
[6] Gervásio H, Simões da Silva L, Borges L. Reliability assessment of the post-limit
buildings by strengthening the panel zone significantly and stiffness and ductility of steel joints. J Constr Steel Res 2004;60:635–48.
improving the seismic performance of the joints by evident energy [7] de Lima LRO, Simões da Silva L, Vellasco PCGS, de Andrade SAL. Experimental
dissipation. evaluation of extended endplate beam-to-column joints subjected to bending and
axial force. Eng Struct 2004;26:1333–47.
2. The insertion of the column base plate can lead to the formation of [8] Shi G, Shi Y, Wang Y. Behaviour of endplate moment connections under
the plastic hinge of the beam away from the core region, avoided earthquake loading. Eng Struct 2007;29:703–16.
strain penetration at the joint region, and maintainer strong-column [9] Levente Katula PM. Study The Prying Effect On Bolted Base-Plate Connections.
Periodica Polytechnica Civil Engineering. 2013;57:157.
weak-beam behavior. Moreover, the thickened panel zone of the
[10] Lee J, Goldsworthy HM, Gad EF. Blind bolted moment connection to unfilled
reinforced beam-column joint (JD3) prevented an apparent defor­ hollow section columns using extended T-stub with back face support. Eng Struct
mation of the joint region, resulted in further strengthening the 2011;33:1710–22.
column. [11] Ahmed R. Behavior of End-Plate Moment Connections with bolts in extended
portion. Dhaka: University of Engineering and Technology; 2009.
3. Compared with the beam-separated joints (JD1 and JD3), the beam- [12] Gerami M, Saberi H, Saberi V, Saedi DA. Cyclic behavior of bolted connections
through joint (JD2) possesses an improved hysteretic behavior due to with different arrangement of bolts. J Constr Steel Res 2011;67:690–705.

14
Z. Chen et al. Engineering Structures 234 (2021) 111875

[13] Sağıroğlu M, Maali M, Kılıç M, Aydın AC. A Novel Approach for Bolted T-Stub [20] wang hlw. Development and behavior testing of beam through connections of
Connections. Int J Steel Structures 2018;18:891–909. floor-by-floor assembled steel braced structures. Building Structure(in Chinese).
[14] Sabol MDETA. Reinforcing of steel moment connections with cover plate: benefit 2012;42.
and limitation. Eng Struct 1998;20:510–20. [21] Wei ZQW. Experimental research on seismic performance of floor-by-floor
[15] FEMA-350.Recommended seismic design criteria for newsteel moment-frame assembled steel braced structures for prefabricated buildings. Building Structure.
buildings. Washington D.C2000. 2012;40.
[16] ANSI/AISC358-05,Prequalified Connections for Special and Intermediate Steel [22] GB/T228.12010, tensile test for Metal Materials – Part 1: Room temperature Test
Moment Frames for Seismic Applications. 2006. Methods. Beijing: Ministry of Housing and Urban-Rural Development, PRC; 2010.
[17] Ahmady Jazany R, Ghobadi MS. Seismic evaluation and upgrading details of plate- [23] GB50017, Code for design of steel structure. In: Ministry of Housing and Urban-
reinforced moment-resisting connections. J Constr Steel Res 2018;150:230–48. Rural Development P, editor. Beijing: China Building Industry Press; 2017.
[18] Liu XC, Xu AX, Zhang AL, Ni Z, Wang HX, Wu L. Static and seismic experiment for [24] Genshu FBWyT. Experimental Study On The Seismic Behavior Of CFST Rectangular
welded joints in modularized prefabricated steel structure. J Constr Steel Res 2015; Column To H-Section Steel Beam Connections With External Stiffeners.
112:183–95. Engineering Mechanics (China) 2020;37.
[19] Wei W, et al. Experimental and numerical investigation on full-scale tension-only [25] Oca JCVMMMMd. Procedure For Assessing The Displacement Ductility Based On
concentrically braced steel beam-through frames. J Constr Steel Res 2013;80: Seismic Collapse Threshold And Dissipated Energy Balance. 16th World
369–85. Conference on Earthquake Engineering. Santiago Chile, January2017.
[26] GB50011-2016, Code For Seismic Design Of Buildings. In: Ministry of Housing and
Urban-Rural Development P, editor. Beijing: China Building Industry Press; 2016.

15