International Journal of Steel Structures (2020) 20(3):752–765 Online ISSN 2093-6311

https://doi.org/10.1007/s13296-020-00320-w Print ISSN 1598-2351

Finite Element Analysis on Seismic Behaviour of Novel Joint

in Prefabricated Modular Steel Building
Kashan Khan1 · Jia‑Bao Yan2

Received: 1 September 2019 / Accepted: 29 January 2020 / Published online: 5 February 2020
© Korean Society of Steel Construction 2020

Abstract
In prefabricated modular steel building, joint plays an important role in the structural behaviour and transfer of load among
connected modules. In this paper, a new type of bolted joint with tenon-gusset plate as horizontal connection and long beam
bolts as vertical connection has been proposed. The nonlinear structural behaviour of the novel joint was analysed against
static and quasi-static cyclic loadings by using finite element (FE) software ABAQUS. Numerical study was carried out to
predict the lateral load carrying capacity, bending moment capacity, rotational stiffness, failure modes and seismic behaviour
of the joint. Connection was simplified with nonlinear spring connector and 3-D beam elements. Parametric study was carried
out on the length of column tenon and gap between adjacent modular units. The results revealed that against lateral loads,
the gap was generated between the upper and lower parts of joint which resulted in the bearing failure of floor beam (FB).
The length of column tenon has obvious while gap between modules showed marginal effect on load carrying capacity and
structural behaviour of joint. The simplified joint accurately mimicked load–displacement curve and structural behaviour of
detailed joint. Then, FE analysis was justified by analysing test results of innovative connection in the reference. With these
evidences, the veracity of the FE analysis for studying nonlinear behaviour of novel joint was confirmed.

Keywords Prefabricated modular steel construction · Finite element analysis · Novel joint · Non-linear analysis · Load–
displacement curve · Parametric study
List of Symbols MPa Mega-Pascal = N∕mm2
P Pretension of bolt Pa Pascal = N∕m2
Ae Effective area of bolt 𝜈 Poisson’s ratio
ftv Tensile strength of bolt fy Yield strength
Δyield Lateral yield displacement fu Ultimate strength
Δ Lateral displacement 𝜀 Strain
Lx Distance of neutral axis from top of column kN 103 N
∅ Rotation P Lateral load
Es Modulus of elasticity m meter
𝜇 Friction coefficient mm millimetre
PTest Ultimate lateral load capacity of test specimen
Abbreviations
PFE Ultimate lateral load capacity of FE model
Cov Coefficient of variation
PFMS Prefabricated modular steel
MSB Modular steel building
* Jia‑Bao Yan
[email protected] HSS Hollow structural section
FE Finite element
Kashan Khan
[email protected] CB Ceiling beam
FB Floor beam
1
School of Civil Engineering, Tianjin University, FS Floor stringer
Tianjin 300350, People’s Republic of China CS Ceiling stringer
2
School of Civil Engineering/Key Laboratory of Coast Civil CP Cover plate
Structure Safety of Ministry of Education, Tianjin University,
Tianjin 300350, People’s Republic of China

13
Vol:.(1234567890)
International Journal of Steel Structures (2020) 20(3):752–765 753

GP Gusset plate Hwan Doh et al. (2017) performed experimental works to

AFR Axial force ratio study structural behaviour of steel bracket joint. The experi-
mental study was carried out to study structural behaviour
of proposed exterior (Chen et al. 2017b) and interior joint
1 Introduction (Chen et al. 2017a) in MSB. Numerical approaches were
adopted to study the ductility and energy absorption per-
Prefabricated modular steel (PFMS) construction consists formance of bracing system with steel rings in steel frame
of prefabricated volumetric modular units that are partially system (Andalib et al. 2018; Bazzaz et al. 2015). These
or fully fitted with essential utilities and architectural fin- previous studies were limited to the behaviour of exterior
ishes in a factory controlled environment then assembled joint, joints with lightweight C-section beam members or
together on site. Before the assembling on site, each module simplified frames with no details of connections and very
is completely fitted in the factory with mechanical, elec- limited information was found on development of joint for
trical, plumbing facilities and floors (Annan et al. 2009a). HSS members, connection arrangement and structural per-
PFMS construction results in reductions of construction formance of complex forms of joints followed by its simpli-
material wastage, noise pollution, site visits, potential dam- fication with spring connector. The detailed researches on
age to the adjacent buildings, construction time, repair cost structural behaviour of the exterior and interior joints, their
and improves architectural and structural qualities of con- connecting techniques, design requirements and the effects
struction materials (Kamali and Hewage 2016; Lawson et al. of connected modular units on adjacent modular unit’s per-
2012). Buildings like student dormitories, offices, hospitals, formance are significantly important for understanding the
schools and hotels with repetitive units can preferably be overall behaviour of PFMS buildings.
constructed by PFMS construction technique (Fathieh and In order to make use of superior lateral torsional buck-
Mercan 2016; Lawson et al. 2012). Nowadays, PFMS con- ling resistance of HSS members and avoid usage of diagonal
struction is gaining attention because of construction diffi- stiffeners to avoid cracking at column and beam intersection
culties in conventional construction technique such as, flex- region. In this paper, firstly a novel bolted joint for HSS
ible workspace on site (Lawson et al. 2014). PFMS buildings is proposed. A 3-D detailed FE model is then developed
differ from the conventional steel buildings in terms of con- for both exterior and interior joint in ABAQUS. Simpli-
struction technique and connection design (Annan et al. fied model of joint (both exterior and interior) with spring
2009b). PFMS buildings are of two types, i.e., load-bearing connector and 3-D beam elements is then developed. The
modules and corner-supported modules. The main load accuracies of the proposed FE models are validated with
carrying component in load-bearing module is C-sectioned test results of experiment carried out by Chen et al. (2017b).
compression resisting walls and in corner-supported mod- Followed, parametric studies on length of column tenon and
ules is hollow structural section (HSS) (Lawson et al. 2012). inter-modular gaps are performed to study their influence on
Figure 1 shows details of novel joint and assembled the lateral loading capacity of the joint. All these numerical
PFMS building. At joint region, clusters of columns and studies and discussions on the basis of detailed paramet-
beams need appropriate connection for proper transfer of ric study contribute in better understanding of the lateral
horizontal and vertical loadings (Annan et al. 2007). In moment capacity, working mechanism and structural behav-
PFMS building, the connection between modules essentially iour of the novel joint.
provides alternate path for the transfer of lateral loads and
structural stability in case of damage to one of the struc-
tural components (Lawson and Richards 2010). Accurate 2 Development and Working Mechanism
arrangement of proper lateral force resisting systems and of Novel Joint
inter-modular connections is vital for modular steel building
(MSB) (Ding et al. 2017). For proper arrangement of bracing Figure 1a shows details of exterior corner, exterior middle
systems in steel frames, linear and non-linear studies were and interior middle connections that connect two, four and
numerically carried out on steel frames (Bazzaz et al. 2012). eight modular units made of HSS sections, while Fig. 1b
Seismic performance of modular frame, horizontally con- illustrates the developed novel joint for MSB. The joint can
nected by site bolting clip angles (welded on flanges of FBs) support eight modular units’; vertically through beam bolts
and vertically connected by partially welding column base and slotted-in tenons and horizontally by intermediate gus-
plates, was studied by Annan et al. (2007, 2009a, b, c). Choi set plate (GP).
and Kim (2015) analytically studied structural performance The upper part of novel connection consists of column
of beam-column joint with access holes. Liu et al. (2015, tenon which is kept hollow in all stories of the building for
2017, 2018) studied structural performance of welded and insertion of tenon from foundation or lower part of joint for
bolted beam-column joints in prefabricated steel structures. developing rigid connection. The tenon from lower part of

13
754 International Journal of Steel Structures (2020) 20(3):752–765

Fig. 1  Novel connection in Assembled MSB on site

prefabricated modular steel
building
Modular unit in factory
C
HSS FB

HSS FS

HSS column

HSS CB B
HSS CS

A- Exterior corner connection B- Exterior middle connection C- Interior middle connection

(a) Modular steel building details

Structural components of MSB Assembly of interior connection

Beam bolt HSS column A

HSS beam

A-Upper component B-Gusset Plate C-Lower component

(b) Connection details

13
International Journal of Steel Structures (2020) 20(3):752–765 755

connection is inserted in column tenon of upper part and is lower parts of the connection. As the gap between FB and
bolted with column. Each part of the connection supports GP keeps increasing, the FB keeps bending and faces bear-
ceiling beams (CBs) and FBs with beam tenons and bolts. ing stresses against beam bolts. With the gradual increase of
Except top story, the lower part of joint mainly consists of uplift force, the movement of column against tenon increases
two solid column tenons; upper solid tenon which is inserted which exerts shearing force on the column bolts.
in the column tenon of upper part and lower solid tenon is From literature review, it has found that connection in
inserted in lower column. The GP in exterior corner con- PFMS building is the most critical part in load distribution.
nection only provides sufficient gap between the upper and Therefore, structural behaviour of the modular units depends
lower modules. But in the exterior and interior middle con- on the connection system and overall behaviour of PFMS
nection as shown in the Fig. 1b, the GP has two functions; building.
one is to provide vertical gap between the upper and lower
modules while second is to horizontally connect adjacent
modular units. 3 Experimental Studies on the MSB
As shown in Fig. 2a, when bending moment as uplift Connection
force is applied on the top of the column, the disturbance is
produced and gap starts generating between the upper and Chen et al. (2017b) reported experimental works to study
working mechanism, load carrying capacity and seismic
behaviour of the innovative joint in MSB. Monotonic and
Bending force
cyclic loading analyses were performed on six specimens
Bolts in tension in Structural engineering laboratory of Tianjin University.

3.1 Test Setup (Chen et al. 2017b)

Bolt in shear
The designing of joint and selection of material properties
were based on four-story MSB constructed in Tianjin, China.
The use of lightweight composite boards was practiced for
ceiling portions and partitions whilst prefabricated concrete
floors for floor slabs. Studying structural response, load car-
rying capacity and energy dissipation of joint in MSB was
the main purpose of the experimental study. Table 1 shows
material properties used in the tests which were same with
the prototype materials used on the site. For the purpose of
studying detailed structural behavior of joint, tests on six
specimens were performed. Firstly, monotonic loading tests
on two specimens with and without diagonal stiffeners were
(a) Force transfer mechanism carried out to determine the load carrying capacity and lat-
eral deformation capacity of the joint as shown in Table 2.
Secondly, quasi-static loading tests on remaining four speci-
mens were performed to predict the seismic performance
and failure modes of connection against cyclic loading.
Specimens without diagonal stiffeners were considered base
specimens and specimens with stiffeners were considered

Table 1  Material properties of specimens

Structural com- Thick- Yield Ultimate Elonga-
ponent ness strength fy strength fu tion in %
(mm) (MPa) (MPa) age

Column, beam 8 425 575 30

plate
(b) Stress-strain curve of steel section used in FE
model (Yan 2015) Stiffeners 16 350 510 26
Cast plug-in – 330 350 22.5
device
Fig. 2  Details of FEM and material model

13
756 International Journal of Steel Structures (2020) 20(3):752–765

Table 2  Details of components in the specimen

Sp. no Ceiling beam ­(mm3) Floor beam ­(mm3) Column ­(mm3) Stiffener thick- Bolts (mm) Axial force Loading method
ness (mm) ratio

S1 150 × 150 × 8 150 × 250 × 8 150 × 150 × 8 None 24 0.2 Static

S2 150 × 150 × 8 150 × 250 × 8 150 × 150 × 8 10 24 0.2 Static
QS1 150 × 150 × 8 150 × 250 × 8 150 × 150 × 8 None 24 0.2 QS
QS2 150 × 150 × 8 150 × 150 × 8 150 × 150 × 8 10 24 0.2 QS
QS3 150 × 150 × 8 150 × 250 × 8 150 × 150 × 8 10 24 0.2 QS
QS4 150 × 150 × 8 150 × 250 × 8 150 × 150 × 8 10 24 0.1 QS

reinforced specimens. By column end loading method, upper behaviour, material’s yield strength “ fy ”, ultimate strength
column was supposed free end constrained condition and “ fu ” and strain values “𝜀” for corresponding strengths were
100t lateral displacement was applied on the upper end of defined according to the values used in test as enlisted in
column, while the lower column and beams ends were pin- Table 1 (Chen et al. 2017b).
constrained. Quasi-static loading pattern adopted in the test,
was taken from Chinese standard, “Regulation of seismic
4.3 Formation of FE Model of Novel Joint
test method (JGJ101-96)”. For comparative study, two axial
force ratios (AFRs) “0.2” and “0.1” were used in the test.
As shown in Fig. 1b, simplifications in the FE modelling
were assumed for accurate representation of connection with
3.2 Material Properties
FE techniques. Accordingly, the bolt nut and head was mod-
elled circular and together with bolt shank while the threads
The HSS columns, FBs, CBs, cover plates (CPs) on beams,
on the bolt shank and nuts were neglected. The gap between
GP, long beam bolts and stiffeners according to Chinese
the bolt shank and bolt hole was neglected to avoid mod-
standards were made of Q345B steel whereas, the plug-in
elling complications. The flanges of beams and CPs were
device was made from ZG35 cast steel. The columns and
modelled together to reduce contact surfaces. The chamfered
beams were welded by groove welding whereas stiffeners
edges and cylindrical solid tenon were modelled as regular
by fillet welding with wire ER50-6. The material properties
geometries for the ease of mesh and to make the contact
used in the tests are enlisted in the Table 1.
surfaces coincident.

4 Finite Element Model 4.3.1 Meshing Technique

4.1 General 4.3.1.1 Detailed FE Model For FE analysis of detailed

model of novel joint, hexagonal structured mesh controls
The ABAQUS/standard (ABAQUS/CAE 2013) type were applied on all structural components. All components
of solver is used for FE modelling and analysis of novel such as bolts, columns, beams, connections were meshed
connection. For more reliable FE results, static-general with 8-node linear brick, reduced integration, and hourglass
procedure with automatic stabilization and possible least control element type (C3D8R). The accurate selection of
increment size was chosen for performing both static and mesh density i.e., fine, extra fine or coarse mesh for each
quasi-static analyses on novel joint (both detailed and sim- component of joint was performed by comparing the load–
plified form of exterior and interior joints). displacement curves of both FE (for each mesh density)
and test. As obtained from the results of non-linear static
4.2 Material Model of Steel analysis, load–displacement curves were overestimated
with coarsely meshed joint whilst extra fine mesh resulted
For steel material used in the test, nonlinear isotropic/kine- in underestimating the results as compared to test results.
matic hardening model from ABAQUS/CAE material library The connection region, bolts and bolt holes were modelled
and von Mises yield criterion were used. In stress–strain with fine meshes for better simulation of stress concentrated
curve, bi-linear behaviour with strain hardening was defined region. In bolt-hole contact application, bolts were consid-
for the material model, shown in Fig. 2b (Yan 2015). In the ered as master surfaces. Mesh element size of 30 mm for
elastic material behaviour, poison’s ratio “𝜈 ” and modulus columns and beams and 10 mm for joint and bolts provided
of elasticity “ Es ” were defined while in the plastic material most reasonable results with the experimental findings.

13
International Journal of Steel Structures (2020) 20(3):752–765 757

4.3.1.2 Simplified FE Model In simplified FE models of both other. To find accurate value, detailed study was performed on
exterior and interior joints, columns and beams were replaced three different values of friction coefficient i.e., 0.2, 0.3 and
by 3-D beam elements that were fined meshed (similar with 0.4. The FE analysis with friction coefficient of 0.2 resulted
detailed model) with 2-node linear beam in space element in lowering the load carrying capacity and 0.4 increasing than
type (B31) while connection were replaced by spring connec- experimental findings. Therefore, friction coefficient of 0.3
tor elements. was chosen for contact surfaces except bolts that were mod-
elled frictionless in FE analysis. In the exterior and interior
4.3.2 Interactions middle connections, the adjacent columns and joints of the
adjacent modular units were simulated with hard contact for
In detailed FE models, the contact between columns and con- accurate sharing of pressure.
nection, beams and connection, columns and bolts and beams
and bolts were simulated with surface to surface (standard) 4.3.3 Boundary and Loading Conditions
with finite sliding and “hard contact” in the normal direction,
while in the tangential direction as “penalty friction formula- In the experiment, movements were restrained in all direc-
tion”. The hard contact formulation interaction in ABAQUS tions at bottom end of the lower column, X and Z direction at
states that two interacting surfaces will share pressure if they top end of the upper column, while at CB and FB movement
contact each other, whilst if they are separated no pressure in Y direction and rotation in Z direction was restrained. In
will be transferred between the surfaces. In ABAQUS the ABAQUS/CAE, horizontal displacement controlled loading
penalty friction formulation allows slip between two or more of 100 mm and axial compression force calculated from AFR
contacting surfaces (bolts with beams, plates) relative to each was applied on the top end of the column (Chen et al. 2017b).
other and the frictional coefficient determines the friction Using Eq. 1, pretension force was calculated for bolt load.
force among interacting surfaces. This type of “hard contact
0.9 × 0.9 × 0.9
and penalty friction formulation” allows the relative motion P= Ae ftv (1)
1.2
of the components and avoids them from penetrating in each

160 200 100

150 50
120

0
P(kN)
P(kN)

P(kN)

100
80 -150 -100 -50 0 50 100 150

-50
50
40 S2 Test Results
S1 Test Results -100
S2 FE Results QS1 Test Results
S1 FE Results 0 QS1 FE Results
0
0 50 100 150 -150
0 30 60 90 120
∆(mm) ∆(mm) ∆(mm)
(a) Specimen S1 (b) Specimen S2 (c) Specimen QS1
150 200
220
100 150
160
100
50 100
50
P(kN)

0 40
P(kN)

P(kN)

-200 -100 0 100 200 0

-20
-50 -150 -100 -50 0 50 100 150 -150 -100 -50 0 50 100 150
-50
-80
-100 QS2 Test Results -100
-140 QS4 Test Results
QS2 FE Results QS4 FE Results
-150 -200 QS3 Test Results -150
QS3 FE Results
-200 -260 -200
∆(mm) ∆(mm) ∆(mm)
(d) Specimen QS2 (e) Specimen QS3 (f) Specimen QS4

Fig. 3  Validation of load–deflection curves of FEM and test

13
758 International Journal of Steel Structures (2020) 20(3):752–765

where, P = Pretension force; Ae = Effective area of bolt; 4.4.1 Validations of FE Model Against Uniaxial
ftv = Tensile strength of bolt = 180 MPa. and Quasi‑Static Loading Test

4.4 Validations of FE Model Figure 3 compares lateral load–displacement curves of six

test specimens with the curves obtained from FE analysis
To validate accuracy of FE model, load–displacement curves which shows FE models well predicted structural behav-
of six specimens obtained from FE analysis were compared iour of test specimens in terms of stiffness, ultimate loading
with test results of experiment. The specimens tested against capacity and ductility. However, there are some fluctuations
quasi-static loadings (as discussed in section 3.1), their enve- found in load–displacement curves between the FE predic-
lope curves were validated with the obtained monotonic tion and tests are due to possible variations in steel material
loading curves of FE analysis. properties, steel sections, flexibility in boundary conditions
or some of the numerical modelling simplifications. Fig-
ure 4 shows that FE models well predicted failure modes of
test specimens both in elastic and plastic states. The weld

Gap Gap widened Tear at column

Crack at column flange
(a) S1 (b) QS1 (c) QS2

Local buckling Inward deformation

(d) QS3 (e) QS4

Fig. 4  Validation of failure modes of FEM and test

13
International Journal of Steel Structures (2020) 20(3):752–765 759

Table 3  Comparison of ultimate load of test specimens and FE 5.1.2 Exterior Corner Connection
Sp. no Loading method Ultimate load Ultimate load PTest/PFE
(Test)/PTest (FE)/PFE(kN) Figure 6a shows the lateral load carrying capacity of the
(kN) exterior corner connection. As load started increasing, gap
was generated between FB and GP, due to which FB experi-
S1 Static 112 118 0.95
enced bearing stresses against beam bolts and the yielding of
S2 Static 183 162 1.13
FB started at lateral displacement of 49.5 mm when lateral
QS1 QS +ve 81 79 1.02
load was 88.9 kN. It was found that FB faced bearing failure
QS −ve − 102 − 118 0.86
at ultimate lateral load and displacement of 119.773 kN and
QS1 QS +ve 81 79 1.02
90.3206 mm.
QS −ve − 102 − 118 0.86
QS2 QS +ve 118 121 0.97
5.1.2.1 Simplification of Exterior Corner Connection By
QS −ve − 137 − 127 1.07
using rotational stiffness obtained from FE analysis of
QS3 QS +ve 161 122 1.32
detailed model of joint, the novel connection was simplified
QS −ve − 183 − 162 1.13
with spring connector and 3-D beam elements. Figure 6c
QS4 QS +ve 141 129 1.09
and d compare the von Mises stress and load–displacement
QS −ve − 168 − 169 0.99
curves of detailed and simplified FE models which justify
Mean 1.05
that simplified model accurately predicted the load carrying
Cov 0.12
capacity and structural behaviour of detailed model of joint.

fracture in beams, local buckling and inward deformation 5.1.2.2 Classification of Novel Joint Based on the strength
in columns, tear in columns and gap formations showed by and stiffness calculation techniques provided by Choi and
the FE models are in well accordance with the test findings. Kim (2015) in the classification table enlisted in Euroc-
Table 3 compares the FE predicted ultimate load capaci- ode (EN 1993-1-1 2005), the novel joint was classified as
ties with the test specimens. It can be found that average test semi-rigid/partial strength joint.
to FE ultimate load capacity prediction is 1.05 and coeffi-
cient of variation (Cov) is 0.12 that are in well accordance
with the accuracy of FE analysis. The ratios of ultimate 5.1.3 Exterior Middle Connection
load capacities higher than 1.0 indicates that FE averagely
slightly underestimated whilst lower than 1.0 indicates that As shown in Fig. 7a, at application of horizontal mono-
FE averagely slight overestimated the ultimate load carry- tonic displacement and axial compression loading on top
ing capacities. The above validations assure the accuracy of mid of two columns, the gap was generated between FB
developed FE models to simulate basic structural behaviour and GP at lateral displacement of 53.4614 mm, and lateral
of novel joint. load of 250.925 kN, which resulted in generation of bear-
ing stresses and bearing failure of FB.

5 Numerical Studies on the Novel Joint 5.1.3.1 Simplification of Exterior Middle Connec‑
tion Detailed exterior middle connection was simplified
5.1 Geometry of the Joint by using same technique of spring connector and 3-D
beam elements. And it can be seen from Fig. 7b and c that
Figure 1b shows main components of novel joint that mainly simplified model has the tendency to accurately predict
consists of three components i.e., upper component, lower the load carrying capacity and structural behaviour of the
component and GP. detailed model with capacity ratio of 1.05.

5.1.1 Dimensions of the Connection 5.2 Parametric Study

Figure 5a shows the dimensions of the components of con- 5.2.1 Length of Column Tenon
nection. Three long beam bolts are inserted vertically and
each bolt is inserted in upper and lower column. The small As tenon is main component of connection so studying
scale specimens of approximately 2/3rd of the full scale effect of length of column tenon on the overall load car-
dimensions of the structural components (according to test rying capacity and structural behaviour of connection is
specimens) are modelled. The vertical gap of 50 mm is con- important. For the reason, FE analysis was carried out on
sidered between upper and lower modules for GP. four models with different lengths of column tenon such

13
760 International Journal of Steel Structures (2020) 20(3):752–765

Upper part plan

Lower part elevation Upper part elevation Lower part plan

(a) Dimensions of main connection components

Lower part elevation Upper part elevation

(b) Geometry for parametric study on length of tenon

Fig. 5  Detailing of novel connection (units:mm)

as 100, 200, 300 and 400 mm denoted as L1 and 420, 520, for rigid connection. It has been found that model 1 shows
620 and 720 mm as L2, respectively as shown in Fig. 5b. column failure whereas, models 2, 3, and 4 show bearing
L1 is the length of the column tenon that supports column failure of FB as shown in Fig. 8. The increase in length of
on both upper and lower part, whereas, L2 is the length of column tenon resulted in obvious local web buckling of both
intermediate column tenon that rests in upper part of joint FB and CB but decrease in column damage because of least

13
International Journal of Steel Structures (2020) 20(3):752–765 761

Fig. 6  Structural behaviour of 200

detailed and simplified exterior
corner connection
150
119.773kN

P(kN)
100 88.901kN

FE P-Delta
50 Ultimate
Yield

0
0 50 100 150 200
Δ(mm)
Von Mises stress Load-displacement curve
(a) Behaviour against monotonic loading

(b) Behaviour against quasistatic loading

Detailed model (unit: MPa) Simplified model (unit: Pa)

(c) Comparison of Von Mises stress of detailed and simplified FE models
150

100
P(kN)

50
Detailed FE Model
Simplified FE Model

0
0 50 100 150
∆ (mm)
(d) Comparison of load-displacement curves of detailed and simplified FE models

13
762 International Journal of Steel Structures (2020) 20(3):752–765

Fig. 7  Structural behaviour of 400

detailed and simplified exterior 323.887kN

middle connection 300 250.925kN

P(kN)
200

FE P-Delta
100
Yield
Ultimate

0
0 60 120 180
∆ (mm)

Von Mises stress Load-displacement curve

(a) Behaviour against monotonic loading

Detailed model (unit: MPa) Simplified model (unit: Pa)

(b) Comparison of Von Mises stress of detailed and simplified FE models

400

300
P(kN)

200

100 Detailed FE Model

Simplified FE Model

0
0 50 100 150
∆ (mm)
(c) Comparison of load-displacement curves of detailed and simplified FE models

column rotation. It can be seen that with increase in length 5.2.2 Gap Between Adjacent Modular Units
of column tenon, the gap between the column and tenon is
reduced which minimizes the risk of column failure at bolted In MSB, joint connects eight modular units together. So,
area while FB and CB fails prior to the failure of column as in order to study the effect of modular units on structural
seen in models 2, 3 and 4. In the model 1, the beams were behaviour of adjacent modular units, 31 models with vary-
not failed but the column failed as a result of which the ing gap from no space to 30 mm were studied under lateral
structure failed prematurely. monotonic loading. Although with increase in gap between
Moreover, from Fig. 10, it can be seen that as the length modular units, a slight rise in stresses was observed but the
of column tenon increases, initial stiffness and load carry- failure mode was still remained same due to bearing failure
ing capacity increase obviously but increment in capacity of FB as shown in Fig. 9. Figure 10 shows load–displace-
followed decreasing pattern from 50% (between model 1 ment curves of only four models (with variation of gap of
having length of 100 mm and model 2 having length of 200 10 mm) which showed marginal increase in capacity up to
mm) to 14% (between model 3 having length of 300 mm and 3.4% between model “having no gap” and the model “with
4 having length of 400 mm). gap of 30 mm”.

13
International Journal of Steel Structures (2020) 20(3):752–765 763

Fig. 8  Parametric study on

length of column tenon with
comparison of von Mises stress
(SF: 4)

L1=100mm, L2=420mm L1=200mm, L2=520mm

(a) (b)

L1=300mm, L2=520mm L1=400mm, L2=620mm

(c) (d)

Fig. 9  Parametric study on gap

between columns with compari-
son of von Mises stress

No Gap 10mm Gap

(a) (b)

20mm Gap 30mm Gap

(c) (d)

5.3 Quasi‑Static Analysis connection under cyclic loading loses its strength gradu-
ally and showed good ductile behaviour with proper energy
The quasi-static analysis using ATC-24 loading protocol dissipation (Fig. 6b).
(Krawinkler 1992), was carried out with limited number
of cycles to observe the seismic behaviour of the exterior
corner connection. Figure 6b shows the lateral load–dis-
placement capacity that is in good accordance with the
monotonic loading results. It can be clearly observed that

13
764 International Journal of Steel Structures (2020) 20(3):752–765

Fig. 10  Comparison of load–

displacement curves of paramet-
ric study on novel joint

6 Conclusions (4) Increase in gap between adjacent modular units from

0 to 30 mm showed marginal effect on load carrying
In this paper, a fully bolted novel joint supporting HSS capacity and stress pattern with no change in failure
member is proposed. The connection includes column and mode of joint.
beam tenons while long beam bolts made of Q345B steel
for vertical connection. GP is used for providing access gap
and horizontally connecting the modular units. The detailed Acknowledgements The first author would like to acknowledge the
scholarship provided by China scholarship council for his study at
nonlinear static analysis and cyclic analysis were performed Tianjin University, China.
on the joint. Structural behaviour of novel joint with para-
metric study using length of column tenon and gap between
the adjacent columns was studied and detailed discussions
were made. This paper focuses on the study of load carrying References
capacity and structural behaviour of connection using FE
software ABAQUS. The FE analysis mainly considers mate- ABAQUS (2013). User manual Version 6.13. DS SIMULIA Corp,
rial nonlinearity, sliding and hard contact among contact Providence, RI, USA., DS SIMULIA.
parts, and bolt pretension. On the basis of numerical studies, Andalib, Z., Ali Kafi, M., Bazzaz, M., & Momenzadeh, S. (2018).
Numerical evaluation of ductility and energy absorption of steel
following conclusions are drawn. rings constructed from plates. Engineering Structures, 169,
94–106.
(1) Nonlinear static and quasi-static analyses on the novel Annan, C. D., Youssef, M. A., & El-Naggar, M. H. (2007). Seismic
connection showed that due to bending, FB faced performance of modular steel braced frames. 9th Canadian con-
ference on earthquake engineering.
bearing failure against the beam bolts and the system Annan, C. D., Youssef, M. A., & El Naggar, M. H. (2009a). Experi-
stopped taking further loadings. However, connection mental evaluation of the seismic performance of modular steel-
under cyclic loading showed good energy dissipation, braced frames. Engineering Structures, 31(7), 1435–1446.
ductile behaviour and plastic rotation angle larger than Annan, C. D., Youssef, M. A., & El Naggar, M. H. (2009b). Seismic
vulnerability assessment of modular steel buildings. Journal of
0.03. Earthquake Engineering, 13(8), 1065–1088.
(2) Simplified model of joint with the spring connector and Annan, C. D., Youssef, M. A., & El Naggar, M. H. (2009c). Seismic
3-D beam elements well predicted the load carrying overstrength in braced drames of modular steel buildings. Journal
capacity with the ratio of 1.05 and structural behaviour of Earthquake Engineering, 13(1), 1–21.
Bazzaz, M., Andalib, Z., Kafi, M. A., & Kheyroddin, A. (2015). Evalu-
of the detailed model of joint. ating the performance of OBS-C-O in steel frames under mono-
(3) Increase in length of column tenon of joint showed that tonic load. Earthquake and Structures, 8(3), 699–712.
lateral load capacity of the connection was improved Bazzaz, M., Kheyroddin, A., Kafi, M. A., & Andalib, Z. (2012). Evalu-
up to limit, which decreased from 50% between models ation of the seismic performance of off-centre bracing system with
ductile element in steel frames. Steel and Composite Structures,
with 100 and 200 mm tenon length to 14% between 12(5), 445–464.
models of 300 and 400 mm tenon length.

13
International Journal of Steel Structures (2020) 20(3):752–765 765

Chen, Z., Liu, J., & Yu, Y. (2017a). Experimental study on interior Lawson, M., Ogden, R., & Goodier, C. (2014). Design in modular con-
connections in modular steel buildings. Engineering Structures, struction. Boca Raton: Taylor & Francis Group, ed., CRC Press.
147, 625–638. Lawson, R. M., Ogden, R. G., & Bergin, R. (2012). Application of
Chen, Z., Liu, J., Yu, Y., Zhou, C., & Yan, R. (2017b). Experimental modular construction in high-rise buildings. Journal of Architec-
study of an innovative modular steel building connection. Journal tural Engineering, 18(2), 148–154.
of Constructional Steel Research, 139, 69–82. Lawson, R. M., & Richards, J. (2010). Modular design for high-rise
Choi, K., & Kim, H. (2015). An analytical study on rotational capac- buildings. Proceedings of the Institution of Civil Engineers—
ity of beam-column joints in unit modular frames. International Structures and Buildings, 163(3), 151–164.
Journal of Civil, Structural, Construction and Architectural Engi- Liu, X. C., Pu, S. H., Zhang, A. L., & Zhan, X. X. (2017). Performance
neering, 9(2), 100–103. analysis and design of bolted connections in modularized prefab-
Ding, Y., Deng, E. F., Zong, L., Dai, X. M., Lou, N., & Chen, Y. ricated steel structures. Journal of Constructional Steel Research,
(2017). Cyclic tests on corrugated steel plate shear walls with 133, 360–373.
openings in modularized-constructions. Journal of Construc- Liu, X. C., Xu, A. X., Zhang, A. L., Ni, Z., Wang, H. X., & Wu, L.
tional Steel Research, 138, 675–691. (2015). Static and seismic experiment for welded joints in modu-
EN 1993-1-1. (2005). Eurocode 3: Design of steel structures - Part larized prefabricated steel structure. Journal of Constructional
1-1: General rules and rules for buildings. Eurocode 3. Steel Research, 112, 183–195.
Fathieh, A., & Mercan, O. (2016). Seismic evaluation of modular Liu, X. C., Zhan, X. X., Pu, S. H., Zhang, A. L., & Xu, L. (2018).
steel buildings. Engineering Structures, 122, 83–92. Seismic performance study on slipping bolted truss-to-column
Hwan Doh, J., Ho, N. M., Miller, D., Peters, T., Carlson, D., & Lai, P. connections in modularized prefabricated steel structures. Engi-
(2017). Steel bracket connection on modular buildings. Journal neering Structures, 163, 241–254.
of Steel Structures & Construction, 02(02), 0437–2472. Yan, J.-B. (2015). Finite element analysis on steel–concrete–steel sand-
Kamali, M., & Hewage, K. (2016). Life cycle performance of mod- wich beams. Materials and Structures, 48(6), 1645–1667.
ular buildings: A critical review. Renewable and Sustainable
Energy Reviews, 62, 1171–1183. Publisher’s Note Springer Nature remains neutral with regard to
Krawinkler, H. (1992). ATC-24 Guidelines for cyclic seismic testing jurisdictional claims in published maps and institutional affiliations.
of components of steel structures. California: Redwood City.

13