Structures 46 (2022) 1119–1133

Contents lists available at ScienceDirect

Structures
journal homepage: [Link]/locate/structures

Progressive collapse mechanisms of multi-story steel-framed modular

structures under module removal scenarios
Sidi Shan a, *, Wei Pan b
a
School of Resources and Civil Engineering, Northeastern University, Shenyang 110819, China
b
Department of Civil Engineering, The University of Hong Kong, Pokfulam, Hong Kong, China

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

Keywords: Due to different assembly requirements and novel integration strategies, the modular structures may have un­
Modular building desirable loading mechanisms and distinct structural behaviors. However, there are limited data on the pro­
Progressive collapse gressive collapse performances of modular structures, which impedes their industrial applications. This study
Module removal scenario
numerically investigates the structural robustness of multi-story steel-framed modular structures against pro­
Finite element model
Steel-framed module
gressive collapse by using pushdown analysis. A six-story structure with five modules per floor is designed and
established. The applied load, failure process and load redistribution mechanisms are studied under module
removal scenarios. Parametric analyses are also conducted to investigate influences of cross-sectional dimensions
of frame members, number of stories and bays, arrangements of bracings, and module removal scenarios on the
robustness of the modular structures. The results show that the prototype modular structure designed with Hong
Kong design code can withstand collapse under module removal scenarios. The load redistribution mechanisms
among columns are significantly influenced by the overturning action and bracing effects. The ceiling beam in
the lower module and the floor beam in the upper module form a double beam system. The column section, beam
section, arrangement of bracings and number of stories significantly influence the robustness of the modular
structure. Finally, a design method based on buckling load equation for columns is proposed to estimate the
progressive collapse resisting capacity of modular structures with good accuracy and efficiency.

1. Introduction becoming popular in developed economies and populated urbans like

Hong Kong, Singapore, London, and so on, where labor shortage, aging
Progressive collapse of a structure refers to the situation where workforce, and increasing costs are critical [7]. This type of structure
damage of local structural elements triggered by abnormal loads results utilize steel-framed modules as the basic structural unit and is built with
in a sequence of failure of adjoining structural elements, eventually an innovative construction method, known as modular integrated con­
leading to the collapse of entire or a large part of a structure. The pro­ struction (MiC) method. The MiC method involves off-site
gressive collapse of a building may cause massive loss of lives and manufacturing of each module and on-site assembly, bringing many
properties, e.g., as in the tragic collapse events of the Ronan Point benefits such as shortened on-site construction period, reduced reliance
Apartment Building in 1968 (The United Kingdom), the World Trade on on-site labor, improved site safety and enhanced construction quality
Center in 2000 (The United States), the collapse of Plasco Building in [8,9]. However, due to the characteristics of MiC, modular structures
2017 (Iran), and so on. As a result, many experimental [1,2], numerical behave distinctively under various loading scenarios, compared with the
[3,4] and analytical [5,6] studies have been carried out in the last conventional structures. The floor system of the steel-framed modular
decade, in an effort to enhance the understanding of progressive collapse structure may be not as continuous as those in a conventional structure,
behaviors of structures and improve the structural robustness against and the load distribution between adjacent modules relies primarily on
progressive collapse. Nevertheless, these studies were primarily focused module corners [10]. These characteristics of the modular structure may
on the conventional structures featuring traditional in-situ construction impair their structural robustness against progressive collapse. Howev­
method. er, even though a series of studies have been conducted to investigate
In recent years, the adoption of steel-framed modular structures is the structural behaviors of modular structures under seismic or wind

* Corresponding author.
E-mail address: shansidi@[Link] (S. Shan).

[Link]
Received 10 July 2022; Received in revised form 2 October 2022; Accepted 23 October 2022
Available online 9 November 2022
2352-0124/© 2022 Institution of Structural Engineers. Published by Elsevier Ltd. All rights reserved.
S. Shan and W. Pan Structures 46 (2022) 1119–1133

scenarios [11,12], very few work has been reported focusing on its progressive collapse performance of steel-framed modular structures. A
progressive collapse behaviors. preliminary design method is also proposed to estimate the resisting
He et al. [13] numerically investigated the progressive collapse capacities of the steel-framed modular structure against progressive
mechanisms of a two-story sub-frame extracted from a modular struc­ collapse in this study.
ture. They demonstrated that the inter-module connections significantly
affected the behaviors of sub-frames against progressive collapse. 2. Description of the prototype structure
However, they only studied the behaviors of planer sub-frames, and the
load redistribution patterns in an entire 3-dimensional modular struc­ A hypothesized six-story steel-framed residential modular structure
ture were not discussed in detail. Alembagheri et al. [14,15] discussed was designed in accordance with Hong Kong Design Code [18,19]. The
the anti-collapse mechanisms and failure modes of modular structure modular structure was designed to resist the wind load conditions in
under module removal scenarios. They found that the arching action Hong Kong [20]. The structure height is 18.1 m; the structure length is
was developed in the upper modules in the failure bays [14], and the 15.04 m and the structure width is 9.0 m. As shown in Fig. 1(a), the
dynamic effect of the modular structure was significantly reduced with structure was assembled with a series of steel-framed modules manu­
increasing removal time under progressive collapse scenarios [15]. factured in an off-site factory. The superimposed dead load (SDL) of 3.5
However, the modules were assumed as volumetric rigid bodies in their kN/m2, 2.0 kN/m2 and 2.0 kN/m2 were applied to account for addi­
studies [14,15], while the loading mechanisms within specific frame tional loads on the floor, ceiling and roof, respectively. The live loads
members, such as beams, corner columns, wall bracing, and so on, were (LL) of 5.0 kN/m2, 2.0 kN/m2 and 2.0 kN/m2 were applied for the floor,
not discussed in detail. ceiling, and roof, respectively.
Luo et al. [16] numerically investigated the progressive collapse Steel S355 material was adopted for structural steel in the modular
performances of entire steel-framed modular structures by using dy­ structure, with the modulus of elasticity of 2.05 × 105 MPa, yield
namic time-history analysis method, and found four possible failure strength of 355 MPa and density of 7850 kg/m3. The Poisson’s ratio of
modes during collapse process. However, loading capacities of the the structural steel is 0.3. The ultimate stress and ultimate strain are
modular structures were not identified in their study [16]. Chua et al. assumed to be 510 MPa [19] and 0.26 [3], respectively. The outer di­
[17] investigated the mechanisms of a 10-story steel-framed modular mensions of each module are 9.0 m long, 3.0 m wide and 3.0 m high, as
structure under column removal scenarios, and demonstrated that the indicated in Fig. 1(b). The clearance between adjacent modules is 10
different column removal locations have nearly no influence on the mm. For each module, intermediate posts were constructed to increase
robustness of the modular structure. However, the displacements of the its integrity. The center-to-center distance between the intermediate
modular structures were small (less than 100 mm) during their analyses, posts is approximately 2.93 m. The floor and ceiling systems consist of a
and the resisting capacities under large deformation stage were not series of purlins. The spacing are 0.5 m for the floor purlins and 1.0 m for
identified. Currently, limited study has been performed to investigate the ceiling purlins. To increase the lateral resistance of the modular
the resisting capacities of the modular structures during the whole structure, chevron steel bracings were installed to external modules
progressive collapse process. located on the edge of the building, and X-shape steel bracings were
All in all, due to the novel assembly strategies of MiC, the modular installed in the modules located on the middle of the building (see Fig. 1
structures may have distinct structural behaviors and undesirable (a)). All intra-module connections were welded together, while steel
loading patterns, which may impair their progressive collapse behaviors bracings were pinned to their surrounding frame members. The cross-
and impede their future industrial applications. However, there are sectional dimensions of structural members are presented in Table 1.
limited data on the robustness of modular structures under the entire The cross-section of the horizontal plates is 150 × 10 mm (Width ×
loading process. In this paper, finite element models are established to Thickness), creating a 10 mm clearance between the adjacent modules.
investigate the mechanical mechanisms of multi-story steel-framed
modular structures against whole progressive collapse process. The 3. Finite element models
loading capacity, load redistribution pattern and failure modes under
module removal scenario are studied by using pushdown method. A 3.1. Modeling strategies
series of parametric analyses are conducted to evaluate the influences of
cross-sectional dimensions of frame members, number of stories and The finite element (FE) models of the modular structures are devel­
bays, arrangements of bracings, and module removal scenarios on the oped by using ABAQUS/Explicit program [21]. To achieve strong

Fig. 1. The prototype modular structure.

1120
S. Shan and W. Pan Structures 46 (2022) 1119–1133

Table 1
Cross-sectional dimensions of the frame members in the modular structure. Beams and columns
Structural Components Section dimensions: Horizontal link
(Height × Width × Thickness)
Module A Module B
Corner column 200 × 200 × 5
Floor beam 200 × 100 × 8
Ceiling beam 200 × 100 × 8
Intermediate post 50 × 50 × 3
Purlin 100 × 50 × 3.2
Module C Module D
Steel bracing 50 × 50 × 6.3
Beams and columns FE model
Pinned connections FE model
convergence, non-linear explicit dynamic analysis is utilized to model Horizontal link FE model
the quasi-static loading scheme. The geometric nonlinearity of the FE
model is considered by using large displacement-finite strain Fig. 3. Schematic view for the modeling methods of the inter-
formulation. module connection.
A schematic view of the prototype modular structure is shown in
Fig. 2. The frame members are simulated with 2-node Timoshenko beam performed over two steps. In the first step, the gravity load 1.2DL +
(B31) elements. The mechanical behavior of structural steel is defined 0.5LL specified by GSA [22] is applied to the structure excepting the
by bilinear material model, and the fracture is represented by dropping failure bay, where DL is the dead load including the self-weight and
the stress suddenly to 5 MPa when reaching the ultimate strain 0.26 superimposed dead load; and LL is the live load on the structure. The
(assuming that the elongation of structural steel equals to 26 %). The gravity load is increased in 5 s until reaching 1.2DL +0.5LL and kept
residual stress is set to be 5 MPa, rather than 0 MPa, in order to avoid constant for 5 s to mitigate dynamic effects. In the second step, the
numerical convergence problem. amplified gravity load on the failure bay α × (1.2DL +0.5LL) is incre­
All intra-module connections were assumed to be rigid in the FE mented slowly until the collapse of the structure occurs, where α is the
model. The schematic view of the inter-module connection is shown in amplified load factor due to dynamic effect. The kinematic energy of the
Fig. 3. The lower end of the column in the upper module was assumed to FE model is monitored to ensure its influence remain negligible during
be pinned to the upper end of the column in the lower module, to cap­ the calculation process.
ture the relative rotation between stacking modules. The horizontal
plates connecting the adjacent modules were simulated by using beam 3.2. Validation of the finite element model
element.
The “General Contact” approach in ABAQUS platform [21] is created 3.2.1. Modular structure without steel bracing
to simulate the interactions among the frame members when collapse The results of a two-story modular frame test conducted by Liu et al.
occurs. “Hard Contact” formulation and “Penalty Contact” algorithm are [23] is adopted to validate the numerical modeling methods in this
adopted for the normal contact and the tangential contact, respectively. study. A schematic view of the tested modular frame is shown in Fig. 4
A rigid ground is added at the bottom of the modular structure, to (a). The frame columns were made of SHS200 × 200 × 8 square hollow
consider the interaction between the structure and the ground surface section (nominal dimension, 200 mm; wall thickness, 8 mm). The
when collapse occurs. column-beam connections were made of SHS200 × 200 × 16 square
Alternate load path method specified by by GSA [22] is utilized to hollow section (nominal dimension, 200 mm; wall thickness, 16 mm).
investigate the collapse processes of the modular structure under mod­ The ceiling beams and floor beams were made of H150 × 150 × 7 × 10
ule removal scenarios. Under extreme loading scenarios, such as fire, and H194 × 150 × 6 × 9H-shaped cross-sections, respectively. The
explosion or demolition, the entire module may be failure or removed, numbers represent the height and width of the section, and the thick­
and the subsequent mechanical behaviors of modular structures should nesses of the web and flange (unit: mm), respectively. The yielding
be studied. As illustrated in Fig. 2(b), the external module in the first strengths of the frame columns, connections, ceiling beams and floor
story of the structure is removed, and the pushdown analysis is beams are 363 MPa, 463 MPa, 399 MPa and 359 MPa, respectively. The

Frame members Bay A Bay B Bay C Bay D Bay E

HL6
Bracings
HL5
Grid 2

HL4
HL3

Bracing 2
BA2 BB2
HL2
Grid 1 Bracing 1
HL1
Bay A Bay B Bay C Bay D Bay E
N1 N2

CB-1 CB-2 CD-1 CD-2

CC-1 CC-2 CE-1 CE-2

(a) Plan view (b) Elevation view on Grid 1

Fig. 2. The number of frame members for the prototype modular structure.

1121
S. Shan and W. Pan Structures 46 (2022) 1119–1133

640 kN 640 kN

H150 150 7 10 Lateral load

2734 mm
SHS200 200 8

400 FE model
250 mm
H194 150 6 9 Experimental
200

Lateral load (kN)

H150 150 7 10

0
2734 mm

SHS200 200 8

-200
H194 150 6 9 SHS200 200 16
-400
-400 -200 0 200 400
2712 mm Lateral displacement (mm)

(a) Schematic view of the tested modular (b) Comparison of numerical (FE model) and

frame[23] experimental results[23]

Fig. 4. Validation of the steel-framed modular structure without steel bracing.

ultimate strengths of the frame column, connection, ceiling beam and X-shape steel bracings, and was assembled with loading beam with inter-
floor beam are 518 MPa, 524 MPa, 546 MPa and 531 MPa, respectively module connections. The beams section was adopted as W100 × 19, in
[24]. The constant loads of 640 kN were applied at the top of each which the width, depth, flange thickness and web thickness were 103
column in the upper module, and the lateral load was then applied at the mm, 106 mm, 8.8 mm and 7.1 mm, respectively. The beams were bent
top of upper module. The columns were restrained to prevent the out-of- about their strong axis during the loading process. The columns and
plane movement of the tested modular frame. Details of the experiment bracings were made of SHS 51 × 51 × 5 (height and width, 41 mm;
was reported by Liu et al. [23]. The tested modular frame is simulated by thickness, 4.78 mm) and SHS 32 × 32 × 3 (height and width, 32 mm;
using the same modeling techniques developed in this study, and the thickness, 3.18 mm) steel tubes, respectively. Based on 0.2 % strain
numerical model of the modular frame is shown by red solid line in Fig. 4 offset method, the yield strength of the steel material was determined as
(a). The numerical lateral load versus lateral displacement at the top of 480 MPa [25]. The columns were restrained to prevent out-of-plane
upper module is compared with the experimental skeleton curve in movement, and the loading beam was pushed until reaching the target
Fig. 4(b). As indicated in the figure, the numerical model give a good displacement of 45 mm. The applied load versus displacement curve
prediction on the experimental results. obtained from the FE model is compared with experimental data and
analytical results from Annan et al. [25] in Fig. 5(b). As indicated in the
3.2.2. Modular structure with steel bracings figure, the FE model predicts the experimental and analytical results
To prove the accuracy of capturing load resisting behaviors of braced with good accuracy.
modular structures, the experiment conducted by Annan et al. [25] is
simulated by using the same modeling techniques developed in this
study. As shown in Fig. 5(a), the steel modular frame was installed with

Applied load 1280

1134

SHS 51 51 5
400 0
W100 19

W100 19
W100 19

200
Lateral load (kN)
1370

0 0
FE model
-200 Experimental
Analytical
SHS 51 51 5
-400
-60 -40 -20 0 20 40 60
Lateral displacement (mm)

(a) Schematic view of the tested braced modular (b) Results obtained from FE model and from

Fig. 5. Validation of the steel-framed modular structure with steel bracing.

1122
S. Shan and W. Pan Structures 46 (2022) 1119–1133

4. Numerical results of the prototype modular structure resisting capacity of the modular structure is provided by the Vierendeel
truss in Bays A and B. The deformation pattern of the modular structure
4.1. Applied vertical load versus vertical displacement in failure stage is shown in Fig. 7(c). In this stage, the modules in Bay A
continually move down, and a significantly large deformation is devel­
In this section, pushdown analysis of the prototype steel-framed oped in modules in Bay B. Correspondingly, Column CB-2 are failed
modular structure is carried out to investigate its progressive collapse under combined bending moment and axial compressive force, resulting
behaviors under external module removal scenario. The amplified load in the collapse of modular structure in Bays A and B. The final collapse
factor (α) versus vertical displacements of Nodes N1 and N2 for the mode of the modular structure is presented in Fig. 7(d). It is seen that the
prototype modular structure are shown in Fig. 6. From the Node N1 removal bay drags the structure laterally, and the modular structure
curve, three typical loading stages are identified, including the initial collapses toward the Bay A under the unbalanced gravity load.
stage, plastic stage and failure stage. In the initial stage (see segment OA
in Fig. 6), the modular structure behaves basically elastically with an 4.3. Load redistribution among frame members
initial stiffness. When the vertical displacement of Node N1 reaches
70.6 mm, the modular structure reaches the plastic stage (see segment 4.3.1. Load redistribution within columns
AB in Fig. 6), and the corresponding load factor reaches 1.30. In this Fig. 8 shows the variations of axial forces within steel columns in
stage, the buckling of Column CB-1 occurs, and the plastic hinges are Grid 1 in the modular structure. The negative value denotes the
developed in the steel frame in Bays A and B, resulting in a reduced compressive force, and the positive value denotes the tension force. As
stiffness in the Node N1 curve. The resisting capacities (equal to the indicated in the figure, the axial compressive force within Column CB-1
vertical load) to the Node N1 in the initial and plastic stages are pri­ gradually increases in the initial stage, until reaching 1334.6 kN.
marily provided by the bending resistance of the structure above the Thereafter, the axial compressive force within Column CB-1 sharply
removed module. After the vertical displacement of Node N1 reaches reduces, indicating the buckling of Column CB-1. The axial compressive
348.7 mm, the modular structure reaches the failure stage (see segment forces within Columns CB-2 and CC-1 gradually increase in the initial
BC in Fig. 6) and the collapse of the structure occurs. In this stage, the and early plastic stages. It is obvious that the gravity load sustained by
pushdown curve flattens, and the collapse of the modular structure oc­ Column CB-1 gradually transfers to Columns CB-2 and CC-1 in these
curs. The corresponding load factor of 1.63 can be considered as collapse stages. In the meantime, the axial compressive forces are gradually
load factor of the modular structure under external module removal reduced within Columns CC-2 and CD-1. In the end of the plastic stage,
scenario. According to GSA guidelines [22], the amplified dynamic co­ the axial compressive forces within Columns CB-1, CB-2, CC-1, CC-2,
efficient used for nonlinear static analysis can be adopted as 1.35 in this CD-1, CD-2, CE-1 and CE-2 are 715.5 kN, 884.7 kN, 1266.6 kN, 160.3
study, which is smaller than 1.63 (the resisting capacity of the prototype kN, 358.6 kN, 334.4 kN, 436.8 kN and 203.4 kN, respectively. Under the
structure). Therefore, the modular structure could withstand progres­ external module removal scenario, the unbalanced gravity load in the
sive collapse under external module removal scenario. Bay A drags the modular structure laterally to the Bay A. Therefore, a
When load factor is the same, the vertical displacement of Node N2 is large amount of axial tension forces are developed in the bracings 1 in
smaller than that of Node N1 in the initial and plastic stages. Node N2 is the Bay C (see Fig. 2(b)), resulting in variations of axial forces within
restrained by its adjacent Column CB-1, so the vertical displacement of surrounding columns. This phenomenon is discussed in the following
Node N2 approximately corresponds to the vertical displacement of section in detail. In the failure stage, the axial compressive forces within
Column CB-1. Node N1 is restrained by the beams above the removed Columns CB-1 and CB-2 gradually reduce to zero, indicating that these
module. Bending deformation is developed in the beams, resulting in a columns are gradually damaged by combined bending and axial
larger vertical displacement in Node N1. compressive force. In the meantime, the axial compressive forces within
Columns CC-2 and CD-1 start to increase, while tension force is devel­
4.2. Failure process oped in Column CE-2. In the end of the failure stage (Node N1 at
approximately vertical displacement of 2000 mm), the axial compres­
The collapse process of the prototype modular structure under sive forces within Columns CC-1, CC-2, CD-1, CD-2 and CE-1 are 1232.6
external module removal scenario is illustrated in Fig. 7. As shown in kN, 1172.7 kN, 1231.5 kN, 363.8 kN and 535.0 kN, respectively.
Fig. 7(a), a small deformation is developed in Bay A under gravity load. Moreover, a tension force of 341.7 kN is developed in Column CE-2.
In the plastic stage, the buckling of Column CB-1 occurs, and the mod­ These are due to the overturning moment induced by the unbalanced
ules in Bay A start to move down (see Fig. 7(b)). In this stage, Vierendeel gravity load in the Bay A, which is explained in detail in the following
truss action is developed in modular structures in Bays A and B, and the section. The variations of axial forces within other columns (Columns
CD-2, CE-1) are relatively small during the entire loading process.
Fig. 9 shows the proportion of the gravity load sustained by each
OA B C
2.0 column in the first story. The positive value / negative value indicates
the ratio of axial compressive force / axial tension force in each column
to the total sustained gravity load in the first story, respectively. Loading
1.5 stages 1, 2, 3 and 4 denote the original state before loading, the end of
Load Factor ( )

N1
initial stage, the end of plastic stage, and the end of loading stage (at the
N2 vertical displacement of 2000 mm), respectively. As indicated in the
1.0 figure, the axial forces are distributed evenly among each column in the
original state (each column sustains approximately 12 % gravity load).
Due to the pushdown load in Bay A, in the end of the initial stage, more
0.5 than 30 % gravity load is sustained by the adjoining Column CB-1, and
30 % load is sustained by Columns CB-2 and CC-1. In the end of plastic
stage, 60 % gravity load is sustained by Columns CB-1, CB-2 and CC-1. It
0.0
0 500 1000 1500 2000 is worth noting that the proportions of the axial force within Columns
Vertical Displacement (mm) CB-2 and CC-1 are significantly increased due to the failure of Column
CB-1. In the end of the loading stage, more than 80 % gravity load is
Fig. 6. Load factor versus vertical displacement of columns for the prototype sustained by Columns CC-1, CC-2 and CD-1 due to the failure of Columns
modular structure under external module removal scenario. CB-1 and CB-2. It is seen that most of gravity load is carried by intact

1123
S. Shan and W. Pan Structures 46 (2022) 1119–1133

Fig. 7. Failure process of the prototype modular structure under external module removal scenario Fig. 7 Failure process of the prototype modular structure under
external module removal scenario.

OA B C 120
500 CE-2
Axial force Percentage (%)

100 CE-1
0 0 CD-2
80
Axial Force (kN)

CD-1
-500 60
CC-2
-1000
40 CC-1
20 CB-2
CB-1 CC-2 CB-1
-1500
CB-2 CD-1 CE-1 0
-2000 CC-1 CD-2 CE-2
-20
0 500 1000 1500 2000 1 2 3 4
Vertical Displacement (mm) Loading Stage
Fig. 8. variation of axial forces within columns versus vertical displacement of Fig. 9. The proportion of the gravity load sustained by each column in the
Node N1 curves under external module removal scenario. first story.

columns closed to the removal module during the progressive collapse the second floor in Bay C of the modular structure. As observed in the
process. figure, significant tension force is developed in Bracing 1 in the initial
and plastic stages. This is because the structure is dragged laterally
4.3.2. Load redistribution within bracings under external module removal scenario, and a great deal of lateral load
Fig. 10 shows variations of internal axial forces within bracings in is resisted by the tension force in Bracing 1. When collapse occurs at the

1124
S. Shan and W. Pan Structures 46 (2022) 1119–1133

B 4.3.4. Load redistribution within horizontal links in inter-module

OA C
200 connections
Bracing 1 in 2nd story Fig. 12 illustrates the variations of axial forces within horizontal
Bracing 2 in 2nd story links adjacent to Bay A of the modular structure. In the initial and plastic
100 stages, a large amount of compressive force is developed in the 1st story
Axial Force (kN)

horizontal link (HL1), while a large tension force is induced in the 6th
story horizontal link (HL6). In the end of the plastic stage, the
0 0
compressive force and tension force within the 1st story and 6th story
horizontal links are 86.5 kN and 84.2 kN, respectively. The axial forces
-100 within horizontal links in the other stories, however, are small in the
initial and plastic stages. In the failure stage, tension forces are gradually
developed in the horizontal links in all stories. The variations of axial
-200 forces within adjacent horizontal links are caused by the tilting and
0 500 1000 1500 2000 collapse of the modular structure under external module removal sce­
Vertical Displacement (mm) nario, which is explained in the following section in detail. The yielding
load of each horizontal link (with the section of 150 × 10 mm) is 532.5
Fig. 10. variation of axial forces within bracings versus vertical displacement
kN, which is significantly larger than the internal forces within hori­
of Node N1 curves under external module removal scenario.
zontal links. It is obvious that the horizontal links have enough resis­
tance under progressive collapse scenarios.
vertical displacement of 348.7 mm (in the end of the plastic stage), the
axial tension force and compressive force in Bracing 1 and Bracing 2 are
4.3.5. Discussion on load redistribution mechanisms
106.8 kN and − 46.5 kN, respectively. Thereafter, the collapse of entire
The critical load redistribution mechanisms of the multi-story
modular structure occurs, and structural components in the upper
modular structure during the progressive collapse process are dis­
stories crushed down on lower stories, resulting in compression force
cussed in this section, including overturning action, bracing effect and
within bracings. The variations of internal forces within other bracings
double-beam system.
in Bay C have the similar trends, so they are not discussed in detail.

[Link]. Overturning action. As illustrated in Fig. 13(a) and (b), the

4.3.3. Load redistribution within beams
modules in Bay A move down under external module removal scenario,
The ceiling beam in the lower module and the floor beam in the
and the modular structure tilts due to unbalanced gravity load. Signifi­
upper module form a double-beam system. Fig. 11 presents the internal
cant tension forces are developed in the horizontal links from the second
axial forces within double-beam systems BA2 and BB2. In the plastic
to the sixth stories in the initial and plastic stages, and large compressive
stage, the tension force is developed in the ceiling beam in the lower
force is developed in the horizontal link in the first story (see Fig. 13(a)).
module and the compressive force is developed in the floor beam in the
In the end of the plastic stage, the tension forces within horizontal links
upper module. In the end of the plastic stage, the axial tension force and
HL2, HL3, HL4, HL5 and HL6 are 21.7 kN, 16.6 kN, 11.2 kN, 14.0 kN and
axial compressive force within ceiling beam and floor beam are 71.5 kN
84.2 kN, respectively, and the compressive force within HL1 is 86.5 kN.
and 60.7 kN, respectively, in the double-beam systems BA2. In the
In the failure stage, large deformation is developed in the structure in
double-beam system BB2, the axial tension force within ceiling beam
Bay A, and tension forces are developed in each horizontal link adjacent
and axial compressive force within floor beam are 87.3 kN and 59.2 kN,
to Bay A (see Fig. 13(b)). In the end of the loading stage (approximately
respectively, in the end of the plastic stage. The couple moment pro­
at the vertical displacement of 2000 mm), the axial tension forces within
duced by the internal axial tension and compressive forces provides an
HL1, HL2, HL3, HL4, HL5 and HL6 are 21.8 kN, 221.6 kN, 181.5 kN,
additional bending capacity to the double-beam systems. This phe­
168.2 kN, 149.4 kN, 133.4 kN, respectively. These load pattern cause an
nomenon is explained in detail in the following section. In the failure
overturning action on the entire structure, dragging the structure
stage, the damage bay (Bay A) collapses and drags the other part of the
laterally, as shown in Fig. 12(a) and (b).
structure. As a result, axial tension forces are developed in both the
ceiling beam and floor beam in system BB2 in this stage.
[Link]. Bracing effect. Due to the overturning action, the structure
moves laterally. The lateral displacement of the roof beam-column

OA B C OA B HL1
200 300
HL2
200 HL3
Axial Force (kN)

100
Axial Force (kN)

100
0 0
Ceiling beam in BA2 0 0
-100 Floor beam in BA2 HL4
Ceiling beam in BB2 -100 HL5
Floor beam in BB2 HL6
-200 -200
0 500 1000 1500 2000 0 500 1000 1500 2000
Vertical Displacement (mm) Vertical Displacement (mm)
Fig. 11. variation of axial forces within beams versus vertical displacement of Fig. 12. variation of axial forces within horizontal links versus vertical
Column A1 under external module removal scenario. displacement of Column A1 curves under external module removal scenario.

1125
S. Shan and W. Pan Structures 46 (2022) 1119–1133

Overturning action
Overturning moment
Tension 156.2
Tension
Tension
Node A

Compression

Compression Compression Tension

Tension
Couple moment Couple moment

(a) Overturning action and bracing effect in (b) Overturning action and failure of bracing

initial and plastic stage effect in failure stage

Floor beam
Compression
Couple moment
Tension

Ceiling beam

(c) Double-beam system

Fig. 13. Schematic diagram of Load redistribution mechanisms of the modular structure under external module removal scenario.

connection above Column CE-2 (see Node A in Fig. 13(a)) is 156.2 mm in ceiling and floor beams in each double beam system are restrained by
the end of the plastic stage. To resist the lateral deformation during the the inter-module connection, deforming as a plane section surrounding
initial and plastic stages, tension forces are developed in bracings 1 in the inter-module connection. Taking the double beam system BA2 as the
Bay C, which in turn results in tension forces within surrounding Col­ example, tension force and compression force are developed in the
umns CC-2 and CD-1 (see Fig. 13(a)) and compression forces within ceiling beam and floor beam, respectively. The internal forces within the
surrounding Columns CB-2 and CC-1. This load pattern forms a resultant ceiling and floor beams are transferred through the bolts in the inter-
couple moment to resist the overturning action on the structure. The module connection, resulting in couple moments in the end of the
other columns (CD-2, CE-1 and CE-2), however, are located in the double beam system. It is calculated that about 13.9 kN⋅m couple
modules, in which no bracing is installed. Therefore, the variations of moment is developed in the double beam system BA2 in the end of the
axial forces within these columns are not obvious in initial and plastic plastic stage (multiplying the axial forces by the relative distance). This
stages, as shown in Fig. 8. mechanism is also developed in the other double-beam systems in the
In the failure stage, the structural components in the upper stories modular structure. This mechanism, however, is not developed in col­
crushed down and the bracing effects are impaired. To resist the over­ umns in this modular structure. This is because that the adjacent col­
turning moment, large axial compressive forces are developed in the umns are only connected by horizontal links using a thin steel plate, and
Columns CC-2 and CD-1, and significant tension force is formed in the the restraints between the adjacent columns are too weak.
Column CE-2 in the failure stage (see Fig. 8 and Fig. 13(b)). This load
pattern results in a couple moment to resist the overturning action on the 5. Parametric analysis of the steel-framed modular structure
structure, as illustrated in Fig. 13(b).
A series of parametric analysis is conducted to study the influences of
[Link]. Double-beam system. Fig. 13(c) presents the double-beam sys­ sections of frame members, number of stories, arrangements of wall
tem of the modular structure against progressive collapse. As observed bracings, and module removal scenarios on the progressive collapse
from the figure, the floor beam in the upper module and the ceiling beam behaviors of multi-story modular structure. The design characteristics of
in the lower module work together as a double beam system. In fact, the the modular structure, modeling and analysis methods described above

1126
S. Shan and W. Pan Structures 46 (2022) 1119–1133

are continually adopted in this section. The numerical analyses are OA B C

performed under external module removal scenario. 3

5.1. Cross-sectional dimensions of columns N1-RHS160 80 8

Load Factor ( )
N2-RHS160 80 8
To investigate the influence of cross-sectional dimensions of steel 2
columns on the progressive collapse resistance of modular structures,
the cross-section of steel columns is adopted as SHS 150 × 150 × 5 mm, N1-RHS200 100 8
SHS 200 × 200 × 5 mm and SHS 250 × 250 × 6.3 mm, respectively. The
1 N2-RHS200 100 8
steel column has a higher resisting capacity with larger cross-sectional
dimensions. Fig. 14 compares the load factor versus vertical displace­ N1-RHS200 0
ment curves under external module removal scenario when adopting
N2-RHS200 0
different cross-sections of columns. As observed from the figure, the 0
pushdown curves yield and flatten at higher load factors with increasing 0 500 1000 1500 2000
cross-sectional dimensions of steel columns, indicating that the structure Vertical Displacement (mm)
has a higher resisting capacity with stronger columns. In the end of the
initial stage, the load factors of the structures adopting HSS 150 × 150 × Fig. 15. Load factor versus vertical displacement curves for the prototype
5 mm, HSS 200 × 200 × 5 mm and HSS 250 × 250 × 6.3 mm columns modular structure using different steel beams.
are 0.67, 1.30 and 2.22, respectively. The vertical displacements of N2
are approximately equal to the axial compressive deformation of Col­ the Vierendeel truss action of the modular structure in Bay B increases
umn CB-1. Since the buckling of columns CB-1 does not occur in the with stronger beams. Similarly, Curves N1 yield and flatten at higher
initial stage, the curves N2 are elastically during this stage. When load factors when adopting larger beam sections.
buckling of Column CB-1 occurs in the plastic stage, the curves N2
yielded. The loading capacity of steel columns enhances with increasing 5.3. Cross-sectional dimensions of horizontal link
cross-sectional dimensions, so the yielding of curves N2 occurs at higher
load factors with larger section dimensions of columns. It can be observed from Fig. 16 that the cross-sectional dimensions of
In the initial stage, the vertical displacements of curves N1 are horizontal links has nearly no influence on the load factor of the modular
basically equal to the vertical deformation of the modules in Bay A. The structure under external module removal scenario. The horizontal links
modular structure in Bay A above the removed module forms a Vier­ work as hoist cables under progressive collapse scenarios, and the
endeel truss under progressive collapse scenario. Since stronger columns modules in the failure bay are suspended by the horizontal links when
increase the Vierendeel truss action of the modular structure in Bay A, collapse occurs. The resisting mechanism of the modular structure is not
yielding of curves N1 occurs at higher load factors with increasing cross- altered with increasing capacity of hoist cables, so the load factor of the
sectional dimensions of columns. modular structure is not increased significantly with increasing hori­
zontal link sections.
5.2. Cross-sectional dimensions of beams
5.4. Number of stories
In this section, the collapse mechanisms of modular structure with
three different sections of beams (RHS 160 × 80 × 8, RHS 200 × 100 × 8 In this section, the modular structures with different number of
and RHS 200 × 150 × 10) are investigated. The larger sections of beams stories are investigated against progressive collapse. As shown in Fig. 17,
result in increasing beam-to-column stiffness ratio. Fig. 15 compares the the modules are stacked on four-, six- and eight-story, respectively, and
load factor versus vertical displacement of N1 and N2 curves under their progressive collapse resisting capacities are compared. The other
external module removal scenario. As can been seen from the figure, design parameters, including the sections of structural member,
curves N2 yield at almost the same point. This is because yielding of arrangement of bracings, gravity load, and so on, are identical with the
curves N2 is caused by the buckling of Column CB-1, and the resisting prototype modular structure. The external module is removed and the
capacities of Column CB-1 remains constant when beam sections are applied load versus vertical displacement curves are shown in Fig. 18. As
changed. Beyond the yielding point, the curves of N2 flatten at higher observed in the figure, the pushdown curves flatten at smaller load
load factors with increasing beam section dimensions. This is because factors with increasing number of stories, indicating that the modular

OA B C OA B C
4 3
N1-SHS250 0
N2-SHS250 0 N1-75 10
3
Load Factor ( )

N2-75 10
Load Factor ( )

N1-SHS200 200 5 2

2 N2-SHS200 200 5 N1-150 10

1 N2-150 10
1 N1-225
N1-SHS150 150 5
N2-SHS150 150 5 N2-225
0 0
0 500 1000 1500 2000 0 500 1000 1500 2000
Vertical Displacement (mm) Vertical Displacement (mm)
Fig. 14. Load factor versus vertical displacement curves for the prototype Fig. 16. Load factor versus vertical displacement curves for the prototype
modular structure using different steel columns. modular structure using different horizontal links.

1127
S. Shan and W. Pan Structures 46 (2022) 1119–1133

(a) 4-story (b) 6-story (c) 8-story

Fig. 17. Schematic view of the modular structure with different number of stories.

5.5. Number of bays

OA B C
3 The modular structures assembled with different number of modules
per story are studied in this section. As illustrated in Fig. 19, the modular
N1-4-story structures have 3-, 4- and 5-bay, respectively. The three structures are
Load Factor ( )

2 pushed down, and their resisting capacities against progressive collapse

N2-4-story
are compared. As shown in Fig. 20, the load factor of the modular
structures alters little with increasing number of bays. The more number
of bays provides larger lateral resisting capacity and overturning
1 N1-6-story moment resistance on the structure under the module removal scenario.
N2-6-story However, the progressive collapse of the modular structure in this study
N1-8-story
N2-8-story
0 OA B C
0 500 1000 1500 2000 3
Vertical Displacement (mm)
N1-3-bay
Fig. 18. Load factor versus vertical displacement curves for the prototype N2-3-bay
modular structure with different number of stories.
Load Factor ( )

2
structures have lower progressive collapse resisting capacity with larger
number of stories. When the buckling of Column CB-1 occurs (in the end N1-4-bay
of the initial stage), the load factors of the modular structures with 4- 1 N2-4-bay
story, 6-story and 8-story are 2.31, 1.30 and 0.82, respectively. When
the collapse occurs (in the end of plastic stage), the load factors of the N1-5-bay
modular structures with 4-story, 6-story and 8-story are 2.59, 1.63 and N2-5-bay
1.26, respectively. The amount of gravity load is increased with 0
increasing number of stories. As the sections of frame members are 0 500 1000 1500 2000
identical, the buckling of Column CB-1 and collapse of the modular Vertical Displacement (mm)
structure occur at a lower load factor when the structure has more
Fig. 20. Load factor versus vertical displacement curves for the prototype
number of stories. modular structure with different number of stories.

(a) 3-bay (b) 4-bay (c) 5-bay

Fig. 19. Schematic view of the modular structure with different number of bays.

1128
S. Shan and W. Pan Structures 46 (2022) 1119–1133

is induced by the failure of the column adjoining to the removal module, OA B C

rather than the overturning moment. So, the dominating factor influ­ 4 N1-VBs in Bays B and D
enced the progressive collapse resistance of the modular structure is the N2-VBs in Bays B and D
strength of the adjacent column, rather than the overturning moment N1-VBs in Bay C
resistance. That is the reason why the number of bays has little influence 3 N2-VBs in Bay C

Load Factor ( )
on the progressive collapse resistance of the modular structure in this
study.
2
5.6. Arrangement of bracings
N1-HBs
The bracings may be arranged in different locations in the modular
1 N2-HBs
structures. In this section, three typical arrangements of bracings are N1-Without Bracing
considered, including no bracing, vertical bracings (VBs) in Bays B and
0 N2-Without Bracing
D, vertical bracings (VBs) in Bay C and horizontal bracings (HBs), as 0 500 1000 1500 2000
shown in Fig. 21. The load factor versus vertical displacement curves for Vertical Displacement (mm)
the modular structures with these three typical arrangements of bracings
are presented in Fig. 22. As seen from the figure, the modular structure Fig. 22. Load factor versus vertical displacement curves for the prototype
with the vertical bracings in Bays B and D flattens at the highest load modular structure with different arrangement of bracings.
factor, while the modular structure without bracing and the modular
structure with vertical bracings in Bay C flatten at the lowest load factor.
In the end of the initial stage (when the buckling of Column CB-1 oc­ OA B C
curs), the load factors of the modular structure without bracing, of the 4
modular structure with VBs in Bays B and D, of the modular structure
with VBs in Bay C and of the modular structure with HBs are 1.26, 1.68,
1.30 and 1.42, respectively. The four modular structures collapse at the 3 Middle module removal

Load Factor ( )
load factors of 1.63, 2.64, 1.63 and 1.90, respectively. A portion of Penultimate module removal
gravity load is transferred through the bracings to the adjacent bays.
Therefore, when more bracings are installed in Bay B, the Vierendeel 2
truss action of the modular structure is enhanced under external module
removal scenario. As a result, the modular structure with VBs in Bays B
and D collapses at the highest load factor. Since no bracing is placed in 1
Bay B, the modular structure without bracing and the modular structure
with VBs in Bays C collapses at the lowest load factor.
0
0 500 1000 1500 2000
5.7. Module removal scenarios
Vertical Displacement (mm)
The unexpected extreme event may occur in different locations of a Fig. 23. Load factor versus vertical displacement curves of the inter-module
building, triggering damages in different structural members. Therefore, connections above removal module for the prototype modular structure
the progressive collapse potentials of a structure should be estimated under middle and penultimate module removal scenarios.
under different module removal scenarios. With this regard, the middle
module (the module in the first story in Bay C) removal scenario and the the load factor reaches 2.55 and 3.37, respectively. Under the penulti­
penultimate module (the module in the first story in Bay B) removal mate module removal scenario, the plastic stage and failure stage begin
scenario are considered in this section. Fig. 23 illustrates the relationship when the load factor reaches 2.69 and 3.70, respectively. Considering
between the load factor and the vertical displacement of the inter- the amplified load factor of 1.35 calculated by GSA guideline [22], it is
module connections directly above the removal module under middle obvious that the modular structure can withstand the collapse under
module removal scenario (above Column CC-1) and penultimate module middle module or penultimate module removal scenarios. The load
removal scenario (above Column CB-1), respectively. Under the middle factors of the modular structure under middle or penultimate module
module removal scenario, the plastic stage and failure stage begin when

(a) Without bracing (b) Vertical bracings in (c) Vertical bracings in (d) Horizontal bracings

Bays B and D Bays C

Fig. 21. Modular structures with different arrangement of bracings.

1129
S. Shan and W. Pan Structures 46 (2022) 1119–1133

removal scenarios are almost twice larger than those under external 6. Estimating the resisting capacity of modular structures
module removal scenario (1.30 and 1.63, respectively). This is because against progressive collapse
the removal bay is sustained by surrounding modules on both sides
under middle or penultimate module removal scenario, while the To prevent the modular structures from progressive collapse, a pre­
removal bay is restrained on single side under external module removal liminary design method is proposed for engineering practices in this
scenario. section. The design objective is to evaluate the resisting capacity of a
The failure process of the modular structure under middle module modular structure under progressive collapse scenarios. As indicated
removal scenario are shown in Fig. 24. The structure has a tiny defor­ above, the collapse process of a modular structure begins when buckling
mation in the initial stage, as indicated in Fig. 24(a). After the buckling of the columns adjacent to the removal module occurs, so the buckling
of Columns CC-1 and CC-2 occurs, the modules in Bay C move down (see load of the steel columns is chosen for the design resistance. Although
Fig. 24(b)). In the failure stage, a significantly large downward the modular structures in the plastic stages might have higher resisting
displacement is developed in the modules in Bay C, and the adjacent capacity against progressive collapse, the resistance in the plastic stage
modules are pulled inward under catenary action (see Fig. 24(c)). The is not considered in the design. This is because that the resistance in the
final collapse mode of the modular structure under middle module plastic stage is highly dependent on the Vierendeel truss action of the
removal scenario is shown in Fig. 24(d). As indicated, the modules in modular structure. The Vierendeel truss action is influenced by many
Bay C collapse and drag the adjacent modules inward, resulting in a parameters, such as the connection stiffness between modules, the
complete collapse of the entire structure. Fig. 25 illustrates the failure beam-column stiffness ratio, and so on. Correspondingly, the Vierendeel
process of the modular structure under penultimate module removal truss action may be significantly varied from one typical modular
scenario. It is seen that the modular structure has the similar failure structure to another, and is inconvenient to be calculated for engineer­
process as that under external module removal scenario. The module in ing purpose. In this study, the load factor of the modular structures in the
Bay B moves down and drags the adjacent modules laterally, leading to a end of initial stage when buckling of columns occur is calculated by the
complete collapse of the structure. following equations:
α = (Rc − G)/LG (1)

Fig. 24. Collapse process of the prototype modular structure under middle module removal scenario.

1130
S. Shan and W. Pan Structures 46 (2022) 1119–1133

Fig. 25. Collapse process of the prototype modular structure under penultimate module removal scenario.

G = (1.2DL + 0.5LL)up (2) gives generally good predictions on the load factors of the modular
structures when buckling of columns occurs. Some of the predictions are
LG = (1.2DL + 0.5LL)removal (3) approximately 10 %-15 % smaller than the numerical results. This is
because some secondary load resisting mechanisms were not considered
Rc = nf φA (4) in the proposed design method, including the Vierendeel truss mecha­
nisms in upper stories, and so on. However, conservative predictions are
where G is the distributed gravity load from the modules directly acceptable for engineering practice. To resist the progressive collapse of
above the adjacent columns; LG is the gravity load from the modules in the modular structures, the load factor α should be larger than the dy­
removal bay; n is the number of columns adjacent to the removal namic load amplified factor Ω specified by GSA guideline [22]. The
module; f is the yielding strength of the steel materials; φ is the stability dynamic effect of the structures under progressive collapse scenarios
coefficient of the columns; A is the cross-section area of the columns; and were discussed in detail by Izzuddin et al. [27] and Vlassis et al. [28].
Rc is the buckling load of the columns adjacent to the removal module,
which is derived from Euler’s critical load and specified in Ref. [26].
7. Conclusions
The proposed design method is utilized to estimate the load factor of
the modular structures under module removal scenarios. It is worth
A comprehensive pushdown analysis was performed in this study to
noting that Rc should consider the buckling capacity of the bracings in
investigate the resisting capacity of multi-story steel-framed modular
Bay B participating in resisting the progressive collapse. The buckling
structures against progressive collapse. The structural responses of a
capacity of bracings (Rb ) can be calculated by Rb = nb fb φb Ab , where nb ,
prototype model (a six-story modular structure with five modules per
fb , φb and Ab are the number, yielding strength, stability coefficient and
floor) were investigated under external module removal scenario. Based
cross-section area of the bracings, respectively. Table 2 compares the
on the prototype model, a series of parametric analysis was performed to
load factors obtained from the proposed design method and those from
study the influences of cross-sectional dimensions of frame members,
the finite element models. Various factors are covered in Table 2, such as
numbers of stories and bays, arrangements of bracings and module
different frame member sections, story number and module removal
removal scenarios on the robustness of modular structures. Finally, a
locations. As can be observed in the table, the proposed design method
design method was proposed to estimate the resisting capacity of

1131
S. Shan and W. Pan Structures 46 (2022) 1119–1133

Table 2 (6) Compared with conventional structures, steel-framed modular

Comparison of progressive collapse load factors of modular structures from the structures develop a new double-beam system mechanism. Moreover,
proposed design method and the FE models. the collapse resisting capacity of steel-framed modular structures is
Rc (kN) α(kN) αFE (kN) Ratio of primarily dependent on the buckling load of the column adjacent to the
α/αFE removal module, rather than the bending resistance of beams as the
External module removal Scenario conventional structures.
Prototype 2664.3 1.12 1.30 0.86 The design method which was proposed based on buckling load
structure equation for columns could be adopted to predict the load factor of the
Altered factors
modular structures against progressive collapse. The proposed design
Column: SHS 150 × 150 × 5 1927.2 0.66 0.67 0.99
SHS 250 × 250 × 4247.4 2.11 2.22 0.95 method can be utilized in engineering practice with good accuracy and
6.3 efficiency. To resist the progressive collapse of modular structures, the
Beams: RHS 160 × 80 × 8 2664.3 1.12 1.14 0.98 load factor should be larger than the dynamic load amplified factor
RHS 200 × 150 × 2664.3 1.12 1.30 0.86 specified by GSA guideline. Future research is recommended to verify
10
Horizontal links Section 75 × 10 2664.3 1.12 1.26 0.89
the feasibility of the proposed design method for the modular structures
Section 225 × 10 2664.3 1.12 1.30 0.86 with various assembly strategies and different module removal
Bracing Without Brace 2664.3 1.12 1.26 0.89 scenarios.
arrangement
Vertical braces in 3219.7 1.47 1.68 0.87
Bays B and D Declaration of Competing Interest
Horizontal braces 2756.8 1.18 1.42 0.83
Story number 4-story 2664.3 2.20 2.31 0.95 The authors declare that they have no known competing financial
8-story 2664.3 0.66 0.82 0.80 interests or personal relationships that could have appeared to influence
Bay number 3-bay 2664.3 1.12 1.38 0.81
4-bay 2664.3 1.12 1.42 0.79
the work reported in this paper.
Middle module removal Scenario
Prototype 2664.3 2.24 2.55 0.88 Acknowledgements
structure
Penultimate module removal scenario
This research is jointly supported by the China Postdoctoral Science
Prototype 2664.3 2.24 2.69 0.83
structure Foundation (No. 2018M631807), Natural Science Foundation of
Liaoning (No. 201702281), Fundamental Research Funds for the Central
Universities (No. N160103002). The research funds above are greatly
modular structures against progressive collapse. The key conclusions of appreciated by the authors.
this paper are provided as follows:
(1) The prototype six-story modular structure designed in accor­ References
dance with Hong Kong Design Code can withstand the progressive
collapse under external or middle single module removal scenarios. [1] Song BI, Sezen H. Experimental and analytical progressive collapse assessment of a
(2) The progressive collapse process of a modular structure can be steel frame building. Eng Struct 2013;56:664–72.
[2] Tian L-M, Li D-Y, Li W, Kou Y-F. Progressive collapse resistance of single-layer
classified into three distinct stages: initial stage, plastic stage, and failure latticed domes under different failure conditions at the same joint. J Build Eng
stage. The modular structure behaves elastically in the initial stage. The 2021;36:102132.
plastic stage starts when buckling of columns adjacent to the removal [3] Shan S, Li S, Wang S. Effect of infill walls on mechanisms of steel frames against
progressive collapse. J Constr Steel Res 2019;162:105720.
module occur, and the failure stage begins when the collapse of modular [4] Meng B, Hao J, Zhong W. Numerical study on the anti-progressive collapse
structure occurs. performance of steel frame-steel plate shear wall structures. J Build Eng 2021;35:
(3) Under progressive collapse scenario, the modules in the removal 102049.
[5] Li G-Q, Zhang J-Z, Jiang J. Analytical modeling on collapse resistance of steel
bay move downward and drag the modular structure laterally, causing beam-concrete slab composite substructures subjected to side column loss. Eng
an overturning action on the entire structure. To resist the lateral Struct 2018;169:238–55.
deformation of the structure, tension forces are developed in some [6] Feng D-C, Xie S-C, Xu J, Qian K. Robustness quantification of reinforced concrete
structures subjected to progressive collapse via the probability density evolution
bracings, resulting in large variations of axial forces within surrounding
method. Eng Struct 2020;2020:109877.
columns. A couple moment is developed by the surrounding columns to [7] Pan W, Chen L, Zhan WT. PESTEL analysis of construction productivity
resist the overturning action. When collapse of the structure occurs, enhancement strategies: a case study of three economies. ASCE J Manage Eng
gravity loads are redistributed among all the columns in the first story, 2018;35(1):5018013. [Link]
[8] Goodier C, Gibb A. Future opportunities for offsite in the UK. Construct Manage
generating a couple moment again to resist the overturning action. Econ 2007;25(6):585–95.
(4) Due to the restraints of the inter-module connections, the ceiling [9] Pan W, Gibb AGF, Dainty ARJ. Leading UK housebuilders’ utilization of offsite
beam in the lower module and the floor beam in the upper module form construction methods. Build Res Informat 2008;36(1):56–67.
[10] Lawson RM, Byfield M, Popo-Ola S, Grubb J. Robustness of light steel frames and
a double beam system to resist the deflection under progressive collapse modular construction. Proc Inst Civil Eng 2008;161(1):3–16.
scenarios. [11] Shan S, Pan W. Structural design of high-rise buildings using steel-framed modules:
(5) Under progressive collapse scenario, the load factors of the a case study in Hong Kong. Struct Design Tall Special Build 2020;29(15):e1788.
[12] Lacey AW, Chen W, Hao H, Bi K. Effect of inter-module connection stiffness on
modular structure in the end of initial stage (corresponding to buckling structural response of a modular steel building subjected to wind and earthquake
of column adjacent to removal bay) increases with increasing column load. Eng Struct 2020;213:110628.
sections. The load factors in the failure stage (corresponding to the [13] He X-H-C, Chan T-M, Chuang K-F. Effect of inter-module connections on
progressive collapse behavior of MiC structures. J Constr Steel Res 2021;185:
collapse of the modular structures) increases with increasing column or 106823.
beam sections. More bracings in the bay adjacent to the removal bay [14] Alembagheri M, Sharafi P, Hajirezaei R, Samali B. Collapse capacity of modular
increase the load factors of the modular structures in the failure stage. steel buildings subject to module loss scenarios: The role of inter-module
connections. Eng Struct 2020;210:110373.
More number of stories reduces load factors of the modular structures
[15] Alembagheri M, Sharafi P, Hajirezaei R, Tao Z. Anti-collapse resistance
when buckling or collapse occur. The cross-sectional dimensions of mechanisms in corner-supported modular steel buildings. J Constr Steel Res 2020;
horizontal link and the number of bays, however, have negligible in­ 170:106083.
fluence on the robustness of the modular structure against progressive [16] Luo FJ, Bai Y, Hou J, Huang Y. Progressive collapse analysis and structural
robustness of steel-framed modular buildings. Eng Fail Anal 2019;104:643–56.
collapse. [17] Chua YS, Pang SD, Liew JYR, Dai Z. Robustness of inter-module connections and
steel modular buildings under column loss scenarios. J Build Eng 2022;47:103888.

1132
S. Shan and W. Pan Structures 46 (2022) 1119–1133

[18] BD. Code of Practice for Dead and Imposed Loads. Building Department, Hong [24] Liu Y, Chen Z, Liu J, Zhong X. Experimental study on seismic behavior of two-
Kong 2011. storey modular structure. Steel Compos Struct 2020;37(3):273–89.
[19] BD. Code of Practice for the Structural Use of Steel. Building Department, Hong [25] Annan CD, Youssef MA, Naggar EI, M.h.. Experimental evaluation of the seismic
Kong 2011. performance of modular steel-braced frames. Eng Struct 2009;31:1435–46.
[20] BD. Code of Practice on Wind Effects in Hong Kong. Buildings Department, [26] GB50017, Standard for Design of Steel Structures, Ministry of Housing and Urban
Kowloon, Hong Kong 2019. Rural Development, China, 2017.
[21] ABAQUS V6.14 [0Computer software]. Paris, Dassault Systems (DS), 2014. [27] Izzuddin BA, Vlassis AG, Elghazouli AY, Nethercot DA. Progressive collapse of
[22] GSA. Alternate path analysis and design guidelines for progressive collapse multi-storey buildings due to sudden column loss—Part I: Simplified assessment
resistance. The U.S. General Services Administration, Washington, D.C., 2013. framework. Eng Struct 2008;30(5):1308–18.
[23] Liu J, Chen Z, Liu Y, Bai Y, Zhong X. Full-scale corner-supported modular steel [28] Vlassis AG, Izzuddin BA, Elghazouli AY, Nethercot DA. Progressive collapse of
structures with vertical inter-module connections under cyclic loading. J Build Eng multi-storey buildings due to sudden column loss—Part II: Application. Eng Struct
2021;44:103269. 2008;30(5):1424–38.

1133