Journal of Building Engineering 82 (2024) 108179

Contents lists available at ScienceDirect

Journal of Building Engineering

journal homepage: www.elsevier.com/locate/jobe

Determination of critical collapse resistance of steel

framed-structure: Fast manual calculation
JingZhou Zhang a, YingHua He a, WenJin Zhang b, GuoQiang Li c, Jing Zhang a, *,
ZhiWei Yu a, **
a
School of Civil Engineering, Guangzhou University, Guangzhou, 510006, China
b
Zhejiang Construction Co. Ltd of China Construction Eighth Engineering Division, Hangzhou, 310000, China
c
College of Civil Engineering, Tongji University, Shanghai, 200092, China

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

Keywords: This paper proposes a fast manual calculation method for determining the critical collapse
Steel frame resistance (CCR) of steel frame structures caused by an edge column loss. The critical collapse
Edge column loss deformation (CCD) of the frame structure in an edge column loss is determined prior to calcu­
Critical collapse resistance lating the CCR. Systematic parametric analysis is performed to examine the impacts of floor
Yield resistance aspect ratio, slab rebar, slab thickness, beam depth, and beam span on the CCD of the frame
Critical collapse deformation
structure. It is found that the CCD of the frame structure caused by the edge column loss can be
determined by the beam depth, beam span and floor aspect ratio. When the beam span is
increased by one beam depth, the CCD of the frame structure is increased by 0.2 times of the beam
depth. The slab rebar and slab thickness have minimal effects on the CCD of the frame. When the
rebar diameter and slab thickness are decreased by 50 % and 30 % respectively, the change of the
CCD of the frame structure is within 10 %. A hardening factor is introduced to calculate the CCR
of the frame in an edge column loss, defined as the ratio of the CCR to the yield resistance of the
frame. Through more than 20,000 parametric analyses, it is found that the hardening factor can
be determined by the normalized CCD (NCCD, the ratio of CCD to the yielding deformation) and
the span-to-thickness ratio (STR) of the slab. For the STR of 40 and NCCD of 32, the hardening
factor of the frame structure in an edge column loss is more than 3.0. A detailed process for
calculating the CCR of the frame structure in an edge column loss is given. The results in the case
study demonstrate that the error between the proposed method and the numerical analysis is
within 10 %.

1. Introduction
The main characteristic of disproportionate collapse of structures is that a localized damage results in a large-scale collapse of the
entire structure [1]. Fig. 1 shows a typical process of disproportionate collapse of the building structures. Initially, a vertical
load-bearing member is damaged due to explosion or fire actions. When the upper storeys of the structure fail, which is defined as the
critical collapse status (Fig. 1(b)), it will fall down to impact the lower storeys of the frame structure, and a series of chain reactions of
collapse may occur at the lower storeys, which eventually results in the disproportionate collapse of the whole frame structure.

* Corresponding author.
** Corresponding author.
E-mail addresses: [email protected] (J. Zhang), [email protected] (Z. Yu).

https://doi.org/10.1016/j.jobe.2023.108179
Received 5 July 2023; Received in revised form 19 October 2023; Accepted 17 November 2023
Available online 22 November 2023
2352-7102/© 2023 Elsevier Ltd. All rights reserved.
J. Zhang et al. Journal of Building Engineering 82 (2024) 108179

Therefore, as long as the critical collapse status of the frame structure is not reached, disproportionate collapse of the structure will not
occur.
In recent decades, considerable progress has been made to investigate the anti-collapse mechanism of the frame structure. The
research objectives mainly include: connections [2–8], beam-column sub-structures [9–13], floor systems [14–18] and global frame
structures [19–27]. Alrubaidi et al. [2] proposed an efficient beam-column connection to reduce the collapse risk of steel frames due to
a column loss. Qiao et al. [5] proposed a new beam-column connection, by reducing the radius of the beam flange and increasing
circular openings at the web to improve the ductility of the frame. Meng et al. [9] studied the collapse behavior of steel frame
considering the unequal span of the steel beam. Zhang et al. [13] studied the collapse deformation of steel beam-column substructures
due to an interior column loss. Alashker and ElTawil [14] proposed an analytical method to compute the resistance-deformation
relationship of the composite floor system. Zhang et al. [16] quantified the dynamic effect on the deformation of the composite
floor system in the frame structure due to an edge column loss. Wang et al. [19] tested the collapse performance of the precast concrete
frame with pre-stressed strands. It was found that the infill walls can enhance the stiffness and load-bearing capacity of the frame
structure, while it decreases the self-centering ability of the structure at the same time. Gan et al. [21] proposed a reliability-based
method to assess the collapse possibility of concrete frame structure, considering the uncertainty of the initial failures of the col­
umns. Different enhancing methods were also proposed to reduce the collapse possibility of the frame structure. Roverso et al. [23]
tested a composite asymmetric floor with planar size of about 12 m × 12 m in an interior column loss. The test results of both
asymmetric and symmetric floors were compared. Lu et al. [24] studied the global collapse performance of the Murrah Federal
Building considering the initiation, and propagation of the explosion.
Compared with experiments and finite element analysis, analytical calculation method is the most efficient approach to evaluate
the critical collapse status of frame structures because when the structural parameters change, repeating finite element analysis or
experiments will consume significant financial and computational cost. However, in current analytical studies [14,17,18,26,27], huge
computational cost or even complicated iterations are required to obtain the critical collapse resistance (CCR) of the frame structures,
which are difficult to be used in engineering practice. Therefore, this paper proposes a fast manual calculation method for calculating
the CCR of steel frame structures caused by an edge column loss. As shown in Fig. 2(a), when the frame structure is subjected to a
column loss, the resistance-deformation curve of the structure is shown in Fig. 2(b). Points A is the yielding status, at which the
resistance can be computed by the yield-line theory. Point B is defined as the critical collapse status, after which there is a high
likelihood that disproportionate collapse of the structure may occur. In Fig. 2(b), the deformation of the frame at point B is defined as
the critical collapse deformation (CCD). The resistance of the frame at point B is the CCR. Note that to obtain the CCR of the frame
structure, the CCD should be firstly determined. In Ref. [17], although the CCR of the frame structure has been quantified, there are
still two main defects. On one hand, the CCD of the frame was obtained directly from numerical analysis and no calculation method
was established. Therefore, as long as the parameters change, numerical analysis is necessary to obtain the CCD of the frame structure.
On the other hand, even if the CCD of the frame is known, the calculation of CCR of the frame is relatively complicated. These two
defects of the method in Ref. [17] limits its practical application with certain extent.
This paper mainly deals with two issues, the quantification of the CCD and CCR of the steel frame structures caused by an edge
column loss. Fig. 3 shows the flowchart of the paper process. In Section 2, systematic parametric analyses are launched to quantify the
CCD of the frame structure, in which the influences of different structural parameters, including the floor aspect ratio, rebar diameter,

Fig. 1. Mechanism of disproportionate collapse of building structure.

2
J. Zhang et al. Journal of Building Engineering 82 (2024) 108179

Fig. 2. Resistance-deformation relationship of the structure due to a column loss.

slab thickness, beam span, and beam depth are considered. In Section 3, the analytical method to calculate the CCR of steel frame
structure in Ref. [17] is briefly reviewed and its reliability is re-examined. In Section 4, a simplified calculation method of the CCR of
the steel frame cause by an edge column loss is established. A hardening factor is introduced, defined by the ratio of the CCR to the
yield resistance of the frame. The influences of different structural parameters on the hardening factor of the frame structure in an edge
column loss are studied. An explicit simplified calculation method for the hardening factor is established. In Section 5, a flow for
computing the CCD and CCR of the structure is given. A case study with a step-by-step calculation process is given in Section 6. The
comparison of the CCR between the methods from this study and other researcher is also made in Section 7. Main conclusions are
drawn in Section 8.

2. Quantification of CCD
2.1. Reduced substructure model
The considered scenario in this study is the loss of an edge column at the first storey of the frame structure. To save computational
cost of analyzing the entire frame structure, only the affected area of the substructure is considered. The simplification of the original
structure is illustrated in Fig. 4. For the single-storey substructure, the boundaries of the three interior sides are designated to be fixed
and the one exterior side is free. The beam-column connection in the frame structure is fully rigid. For the semi-rigid and flexible beam-
column joints, the CCD of the floor system will be greater than that with fully rigid joint because the semi-rigid and flexible joints can
develop greater rotation. The flat concrete slab is used for the floor system. Note that the disproportionate collapse of the frame
structure only due to the failure of steel beam or connection is considered. The collapse of the structure governed by the neighboring
columns of the removed column sequentially losing stability is not studied. The column stability can be conveniently evaluated by the
Chinese Code of Standard for Design of Steel Structures [28].

2.2. Numerical modelling

2.2.1. Validation
The reliability of the numerical modelling has been previously validated in Ref. [29]. A collapse test on a composite floor was
selected to conduct the validation, as shown in Fig. 5. The details of the tested floor system, including the geometric size, member
dimensions, material properties, and boundary conditions can be found in Ref. [29]. In the numerical model, shell elements S4R are
adopted to model the beam, column, and slab. There is no specific modelling for the rebar in the slab. Instead, the “Rebar layer”
command is used to consider the contribution of the rebar. The deterioration of concrete material is captured by the damage plasticity
model. Due to the fact that the in-built fracture model for steel in ABAQUS fails to account for the influence of the stress Lode angle. The
fracture of steel is considered by the Bai-Wierbilcki model in this study, which can be written as [30]:

Fig. 3. Flowchart of the paper process.

3
J. Zhang et al. Journal of Building Engineering 82 (2024) 108179

Fig. 4. Establishment of simplified model.

Fig. 5. Collapse test on the floor by authors [29].

( )
εf (η, θ) = εtf − εsf θ2 + εs (1)

wherein η and θ are the stress triaxiality and normalized Lode angle, respectively.
Details of the Bai-Wierbilcki model for Q235 steel can be found in Ref. [13].To model a fully welding connection, the steel beam
and column are merged. The slab and beam are connected by “Tie” command. The mesh size of each structural member are given in
detail in Ref. [29]. The comparisons of the resistance-deformation curve and failure mode are presented in Figs. 6 and 7, respectively.
In Fig. 6, the blue curve is the error of the numerical analysis. It can be seen that the error is the greatest at the beginning, which is
about 25 %. Then the error sharply decreases when the displacement increases. The final error for the numerical analysis in the CCR of
the structure is within 5 %. In Fig. 7, the numerical analysis has obtained elliptically distributed deformation pattern of the floor
system, which is similar to that of the experimental result.

Fig. 6. Comparison of the resistance of the floor.

4
J. Zhang et al. Journal of Building Engineering 82 (2024) 108179

Fig. 7. Comparison of the final deformation.

2.2.2. Parametric analysis

In this study, the finite element model for the reduced substructure is shown in Fig. 8. The beam and column are both made of Q235
steel, as shown in Fig. 9. The concrete compressive strength is 22.9 MPa. Vertical displacement is imposed at the removed edge column
until the critical collapse status is reached (fracture occurs in the steel beam). In the parametric analyses, the considered variables
include the floor aspect ratio, rebar diameter, slab thickness, beam span, and beam depth. The reason for considering these variables is
based on the previous research conclusions by authors in Ref. [29]. Altogether 60 numerical models are established and analyzed. For
the steel beam, four values of the section depth are considered: 600 mm, 500 mm, 400 mm, and 300 mm. For each section dimension,
six beam spans are studied. The beam span-to-depth ratio (SDR) is 18, 16, 14, 12, 10, and 8. Two values of slab thickness are studied,
wherein the slab span-to-thickness ratio (STR) of the floor is kept to be 28 and 40. The reason why only two values of the slab STR is
selected is due to consideration that the slab thickness has minimal influence on the CCD of the floor [29]. The mesh sizes of the beam
and slab have critical effects on the CCD of the floor system. Previous studies by authors [29] indicate that 40 elements through the
beam depth and 72 elements along the slab span are enough to guarantee the accuracy of the CCD of the floor in the numerical analysis.
Table 1 shows the details of the geometric dimensions and mesh sizes of different numerical models. Note that the floor aspect ratio in
Table 1 is 1:1. The rebar diameter and center spacing are 12 mm and 180 mm, respectively. The concrete cover depth is 20 mm.
In practical engineering, the floor aspect ratio of the structure is always not 1:1. To evaluate the floor aspect ratio impact on the
CCD of the frame structure, the substructure with beam depth 400 mm and span 5.6 m is selected for analysis. Five floor aspect ratios
are considered: 1.2, 1.4, 1.6, 1.8 and 2.0. The floor aspect ratio is changed by changing its long span and the short span remains to be
unchanged. Table 2 presents the numerical models for different floor aspect ratios. Moreover, the effect of rebar diameter on the CCD of
the floor system is studied. Substructure with beam depth 400 mm and span 4 m is selected for investigating the effect of the rebar
diameter. Three values of the rebar diameter are considered: 8 mm, 12 mm, and 16 mm. It should be noted that when the rebar
diameter changes, the rebar ratio of the slab remains unchanged. For example, when the rebar diameter is increased from 12 mm to 16
mm, the center spacing of the rebar is increased from 180 mm to 320 mm.

Fig. 8. Details of the floor system in numerical analysis.

5
J. Zhang et al. Journal of Building Engineering 82 (2024) 108179

Fig. 9. True stress-plastic strain curve for Q235 steel.

Table 1
The studied floor systems in numerical analysis with aspect ratio of 1:1.

Column(mm) Span (m) Slab thickness (mm) Mesh size

Beam (mm)
Slab (mm) Beam (mm)

STR: 28 STR: 40 Fine region Coarse region

600 × 600 × 20 × 30 12 429 300 167 15 (global) 60 (axial)

600 × 200 × 11 × 17 9.6 343 240 133 30 (in section)
8.4 300 210 117
7.2 257 180 100
6 214 150 83
4.8 171 120 67
500 × 500 × 15 × 25 9 321 225 125 12.5 (global) 50 (axial)
500 × 200 × 10 × 16 8 286 200 110 25 (in section)
7 250 175 97
6 214 150 83
5 179 125 69
4 143 100 56
400 × 400 × 13 × 21 7.2 257 180 100 10 (global) 40 (axial)
400 × 200 × 8 × 13 6.4 229 160 89 20 (in section)
5.6 200 140 78
4.8 171 120 67
4 143 100 56
3.2 114 80 44
300 × 300 × 11 × 18 5.4 193 135 75 7.5 (global) 30 (axial)
300 × 150 × 6.5 × 9 4.8 171 120 67 15 (in section)
4.2 150 105 58
3.6 129 90 50
3 107 75 42
2.4 86 60 33

Table 2
Details of the substructure with floor aspect ratio not 1:1.

Notation Slab thickness (mm) Floor aspect ratio Slab size (m)

STR of 28 STR of 40

H400-L5.6-R1.2 200 140 1.2 6.72 × 5.6

H400-L5.6-R1.4 1.4 7.84 × 5.6
H400-L5.6-R1.6 1.6 8.96 × 5.6
H400-L5.6-R1.8 1.8 10.08 × 5.6
H400-L5.6-R2.0 2.0 11.2 × 5.6

2.3. Result and discussion

2.3.1. Influence of slab thickness, beam depth and beam span
As shown in Fig. 10, the critical collapse status of the substructure is dominated by the fracture failure at the lower flange of the

6
J. Zhang et al. Journal of Building Engineering 82 (2024) 108179

beam due to great deformation. This collapse pattern is defined as the connection/beam failure-induced collapse. After the critical
collapse status, the resistance of the substructures sharply decreases. The resistances of the floor in the edge column loss with beam
height 300 mm, 400 mm, 500 mm, and 600 mm are presented in Figs. 11–14, respectively. The legends, such as H400-L3.2, represents
that the beam depth is 400 mm and the beam span before the column loss is 3.2 m. When the beam span increases, both the CCD and
CCR of the substructure increase. For example, when the beam span increases from 3.2 m to 7.2 m, the CCD and CCR of the sub­
structure are increased by about 25 % and 50 %, respectively. Moreover, the slab thickness mainly influences the yield resistance of the
substructure, while its influence on the CCD and CCR of the substructure is minimal. It can be concluded that when determining the
CCD of the floor system, the effect of slab thickness can be ignored. For the floor aspect ratio of 1:1, the CCDs of the substructures are
summarized in Fig. 15. When the beam depth is a constant, the CCD of the substructure is strongly linear against the beam SDR.

2.3.2. Influence of floor aspect ratio

As shown in Fig. 16, the floor aspect ratio has relatively significant effect on the CCD of the substructure. However, its effect on the
CCR of the substructure is not obvious. For the floor aspect ratio of 1:1 and 2:1, the difference of the CCD of the substructure is
approximately 15 %. However, the CCR is only increased by 5 %. The CCDs of the substructure H400–5.6 for different floor aspect
ratios is presented in Fig. 17. The CCD of the substructure is found to be generally linear against the floor aspect ratio.

2.3.3. Influence of rebar diameter

The resistance of the substructure H400-L4 with rebar diameters of 8 mm, 12 mm, and 16 mm under a same rebar ratio are given in
Fig. 18. It is found that the CCDs of the floor system of different rebar diameters are almost the same, which indicates that the rebar
diameter basically has no influence on the CCD of the floor. On the other hand, previous research by authors indicated that the rebar
ratio also has minimal influence on the CCD of the floor [17]. Therefore, it can be concluded that when determining the CCD of the
floor system, the influences of rebar diameter and rebar ratio can be neglected. The variables required to be considered are the floor
aspect ratio, beam span, and beam depth.

2.4. Quantification of the CCD

The quantification process of the CCD of the floor system has the following two steps. Firstly, the CCD for the floor of aspect ratio of
1:1 is determined, in which an explicit formula is established between the CCD, beam span and beam depth. Secondly, the effect of the
floor aspect ratio is considered by introducing an aspect ratio factor to revise the formula proposed in. The CCD of the floor system is
firstly converted to a non-dimensional form, divided by the beam depth:
CCD
CDR = (2)
D

wherein CDR is the non-dimensional form of the CCD, calculated by the ratio of the CCD to the beam depth. D is the beam depth.
From Fig. 15, the CDR of the floor can be obtained, as shown in Fig. 19. The CDR of the floor system is found to be linearly
correlated with the beam SDR. The relationship between the CDR and SDR is as follows:
CDR= 0.2×SDR− 0.63 (3)
Based on the results in Fig. 16, the floor aspect ratio impact on the CCD of the structure can be quantified by defining an aspect ratio
factor far, which is the ratio of the CCD of the structure with floor aspect ratio other than 1.0 to that with floor aspect ratio of 1.0. The
result is shown in Fig. 20. A strong linear correlation can be observed between the floor aspect ratio and far, which is given by:
far = 0.16×AR+0.84 (4)

Fig. 10. Failure mode of the floor system at the critical collapse status.

7
J. Zhang et al. Journal of Building Engineering 82 (2024) 108179

Fig. 11. Resistance of the substructure with beam depth of 300 mm.

Fig. 12. Resistance of the substructure with beam depth of 400 mm.

Fig. 13. Resistance of the substructure with beam depth of 500 mm.

wherein AR is the floor aspect ratio.

Therefore, the CCD of the floor system considering the floor aspect ratio, beam span, and beam depth can be calculated by Eqs. (2)–
(4), which is given by:
CCD = (0.16 × AR+0.84)(0.2 × SDR− 0.63)D (5)

8
J. Zhang et al. Journal of Building Engineering 82 (2024) 108179

Fig. 14. Resistance of the substructure with beam depth of 600 mm.

Fig. 15. The CCDs of the substructures for floor aspect ratio of 1:1.

Fig. 16. Resistance of the substructure with different floor aspect ratios.

3. Analytical basis
In Ref. [17], the analytical formulae to calculate the yield resistance and CCR of the floor system due to an edge column loss were
proposed. The yield resistance and CCR of the floor system can be calculated based on the simplified yield-line mode and failure mode
of the floor system, respectively, as shown in Fig. 21. More details can be found in Ref. [17].

9
J. Zhang et al. Journal of Building Engineering 82 (2024) 108179

Fig. 17. The CCDs of the substructures for different floor aspect ratios.

Fig. 18. Resistance for the substructure with different rebar diameters.

Fig. 19. Regression of the SDR and CDR

Based on the yield line theory, the yield resistance of the floor system in an edge column loss Fy is calculated by Ref. [17]:
( ) 2 1 αLMy
Fy = Mx l + Mx∗ l + MLN + MLP + MTN +2 (6)
αL l l

wherein Mx is the positive yielding moment in unit width of the slab section and Mx* is the negative yielding moment in unit width of
the slab section. MLP, MLN, and MTN are the positive plastic moment of the x-direction beam, negative plastic moment of the x-direction

10
J. Zhang et al. Journal of Building Engineering 82 (2024) 108179

Fig. 20. Quantification of the floor aspect ratio impact.

Fig. 21. Theoretical model to quantify the resistance of the floor system in an edge column loss.

beam, and negative plastic moment of the y-direction beam, respectively.

In Ref. [17], the CCR of the floor system Fu due to an edge column loss is given by:
CCD CCD CCD L
Fu = Cxe (Tx l+2FuL )+2Cya Ty Lb +2Cxa Tx lb +2Ful +2Cym My α
2l αL αL l
(7)
l l
+2Cxm Mx +2Mx∗
αL αL

wherein
(√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ )
2CCD (αL)2 + CCD2 + αL − √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅CCD3
(αL)2 +CCD2
Cxe = (√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ )2
(αL)2 + CCD2 + αL

11
J. Zhang et al. Journal of Building Engineering 82 (2024) 108179

α(3k+2) αk 3 3k+2 k3
Cya = − Cxa = −
3(1 + k)2 3(1 + k)2 6(1 + k)2 6(1 + k)2

k− 1 χ 2y b2 ( 2 ) k− 1 χ b2 ( 2 )
Cym = 1 + χ 1y b− k − k+1 Cxm = 1 + χ 1x b − 2x k − k+1
2 3 2 3

In the above equations, Cxe accounts for the influences of the extension of the rebar and beam. Cya and Cxa consider the influences of the
bending moments by the in-plane slab membrane forces due to geometric nonlinearity. Cym and Cxm consider the impacts of membrane
forces on the pure yielding moment of the slab. More details can be found in Ref. [17]. Although the accuracy of the proposed
analytical formula has been verified in Ref. [17], only limited variables were considered. Hence, to more reliably calculate the CCR, the
analytical formula is re-examined by the results of the systematic parametric analyses in Section 2. The analytical and numerical values
of the CCR of the steel frame are shown in Fig. 22. The red dashed line shows that the analytical results are the same to the numerical
results. It can be seen that all scatter points are closely clustered near the red dashed line. The analytical results are generally smaller
than the numerical results, with acceptable errors.
It can be seen from Eq. (7) that to calculate the CCR of the floor system, complicated calculations are required to determine the
parameters Cxe, Cya, Cxa, Cym, Cxm and other associated parameters, which makes it difficult to be used in practical engineering.
Therefore, simplified calculation of the CCR of the floor system is required.

4. Simplification of CCR
4.1. Hardening factor
A hardening factor f is firstly defined, which is calculated by the ratio of the CCR Fu to the yield resistance Fy of the floor system:
Fu
f= (8)
Fy

This is due to the following consideration. By using the simplified calculation method of CCD in Section 2 and the analytical
formulae in Section 3, systematic parametric analyses can be conducted to investigate the effects of certain structural parameters, such
as floor aspect ratio, slab rebar ratio, slab thickness, beam depth, and beam span on the hardening factor of the floor system due to an
edge column loss. Then the relationship between the hardening factor and the structural parameters can be established. Therefore, the
CCR for a certain frame structure can be conveniently calculated by using the hardening factor times the yield resistance of the
structure.

4.2. Parametric analysis

In the parametric analysis, the considered variables include the beam section dimension, beam SDR, slab STR, slab rebar ratio, and
floor aspect ratio. Reasonable ranges of values are assigned to these parameters to ensure the practical significance of the analysis, as
summarized in Table 3. For the beam section, 10 I-shaped sections are selected from the Chinese code of Standard for Design of Steel
Structures [28]. 13 values of the beam SDRs are considered, ranging from 8 to 20. For the slab STR, six values are considered, from 28
to 40. The rebar diameters of 12, 14, 16, and 18 are considered. The floor aspect ratio ranges from 1.0 to 2.0. These variables are
orthogonally combined and altogether 21,840 parametric analyses are conducted.

4.3. Quantification of the hardening factor

In Fig. 23, the results of the parametric analyses are shown. The y- and x-axis are the hardening factor and the normalized CCD
(NCCD) of the frame structure, respectively. The NCCD is a dimensionless quantity, which is given by:
CCD
NCCD =
va
√̅̅̅̅̅̅
̅ (9)
(0.16R+0.84)(0.2 × SDR− 0.63)D 25 E (0.16R+0.84)(0.2 × SDR− 0.63)
= √̅̅̅̅̅̅ =
3L 2fy 3 2fy SDR
50 E

wherein va is the floor deformation at the yielding stage. More descriptions on the deformation va are given in Ref. [17].
Eq. (9) suggests that the NCCD is determined only by the beam SDR and floor aspect ratio. In Fig. 23, the red line is the linear fitting
for the 21,840 scatter points. It can be seen that the R2 is only about 0.86. For the large values of the NCCD, a great proportion of the
scatter points is beyond the 99 % confidence interval. Moreover, the difference of the hardening factor increases with the increase of
the NCCD. When the NCCD reaches about 32, the variation of the hardening factor for the same NCCD is up to more than 50 %. This
phenomenon indicates that it is not reasonable to determine the hardening factor merely by the NCCD of the frame structure.
Therefore, sensitivity analyses are further conducted to see the influences of different structural parameters on the hardening factor of
the frame structure in an edge column loss. The results are shown in Fig. 24. Fig. 24 (a) indicates the effects of the floor aspect ratio,
rebar diameter, beam span, and beam depth. A strong linear relationship between the hardening factor and the NCCD can be observed.
The influence of the slab thickness on the hardening factor is presented in Fig. 24(b). It is found that for different slab thicknesses, the
NCCD remains unchanged while the hardening factor is significantly changed, with variation more than 20 %. Similar phenomenon

12
J. Zhang et al. Journal of Building Engineering 82 (2024) 108179

Fig. 22. Correlation results of the predicted and actual CCR.

Table 3
Values of the structural parameters in the parametric analyses.

Variables Values

Beam section dimension (mm) H300 × 150 × 6.5 × 9, H400 × 200 × 8 × 13,
H450 × 200 × 9 × 14, H500 × 200 × 10 × 16,
H550 × 200 × 10 × 16, H600 × 200 × 11 × 17,
H650 × 300 × 11 × 17, H700 × 300 × 13 × 24,
H750 × 300 × 13 × 24, H800 × 300 × 14 × 26
Beam SDR 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20
Slab STR 28, 30, 32, 34, 36, 38, 40
Rebar diameter (mm) 12, 14, 16, 18
Floor aspect ratio 1.0, 1.2, 1.4, 1.6, 1.8, 2.0

Fig. 23. Results of the hardening factor in the parametric analyses.

can also be found in Ref. [17] that when the slab thickness of the floor systems increases from 75 mm to 150 mm, the CCR of the floor
system are increased by less than 10 %. However, the yield resistance is increased by about 100 %. It suggests that to determine the
hardening factor, the slab thickness should be considered separately as an independent variable.
Based on the above analysis, the hardening factor can be determined by the NCCD and slab thickness. For simplicity, the slab
thickness is converted to the slab span-to-thickness ratio (STR). The relationship between the hardening factor, slab STR, and NCCD is
shown by a three-dimensional scatter plot in Fig. 25. It can be seen from Fig. 25(a) that the hardening factor is positively correlated
with both the slab STR and NCCD. The mathematical expression of the regressed surface is given by:

13
J. Zhang et al. Journal of Building Engineering 82 (2024) 108179

Fig. 24. Effects of different structural parameters on the hardening factor of the frame structure.

f (NCCD,STR)= 1.057 − 0.006×NCCD− 0.071 × STR+0.004×NCCD × STR (10)

2
The R of this regression is more than 0.96, indicating that by taking the slab STR as a separate variable, the performance of the
regression is more satisfactory.

5. Calculation flow
Based on the above analysis, a calculation flow of the CCR of the frame structure due to an edge column loss is given below.
Step 1: Obtain the required structural parameters.
Step 2: Compute the CCD of the structure by Eq. (5). The CCDs of the structure with different beam SDRs, beam depths, and floor
aspect ratios are summarized in Table 4.
Step 3: Calculate the yield resistance of the frame structure due to an edge column loss by Eq. (6).
Step 4: Compute the NCCD of the structure by Eq. (9).
Step 5: Determine the hardening factor of the frame structure by Eq. (10). The hardening factors of the frame structure for different
slab STRs and NCCDs are summarized in Table 5.
Step 6: Determine the CCR of the frame structure by using the yield resistance times the hardening factor.

Fig. 25. 3-D scatter plot of the hardening factor with the slab STR and NCCD.

14
J. Zhang et al. Journal of Building Engineering 82 (2024) 108179

Table 4
Summary of CCDs of the frame structure due to an edge column loss (unit in mm).

Floor aspect ratio Beam depth (mm) Beam SDR

8 10 12 14 16 18

1 300 291 411 531 651 771 891

400 388 548 708 868 1028 1188
500 485 685 885 1085 1285 1485
600 582 822 1062 1302 1542 1782
1.2 300 300 424 548 672 796 920
400 400 566 731 896 1069 1226
500 501 707 913 1120 1326 1533
600 601 848 1096 1344 1591 1839
1.4 300 310 437 565 693 820 948
400 413 583 753 924 1094 1264
500 516 729 942 1154 1367 1580
600 619 875 1130 1385 1641 1896
1.6 300 319 450 582 713 845 977
400 425 601 776 951 1127 1302
500 532 751 990 1189 1408 1628
600 638 901 1164 1427 1690 1953
1.8 300 328 464 599 734 870 1005
400 438 618 799 979 1160 1340
500 547 773 998 1224 1449 1675
600 656 927 1198 1469 1739 2010
2 300 338 477 6166 755 894 1034
400 450 636 821 1007 1192 1378
500 563 795 1027 1259 1491 1723
600 675 954 1232 1510 1789 2067

6. Case study
A detailed calculation procedure for the CCR of the structure in an edge column loss is given. The frame structure is a three-storey
with four-span and three-bay. The spans of the steel beam are 6 m and 4 m in longitudinal and transverse direction, respectively. The
steel beam is H 400 mm × 200 mm × 8 mm × 13 mm. The thickness of the floor slab is 120 mm.The material properties and mesh sizes
are the same to those in Section 2. The bottom surfaces of the steel column at the ground floor are fixed. Displacement loading is
applied at the top of the removed edge column. The final deformation of the structure is presented in Fig. 26. The resistance-
deformation curve of the frame structure is given in Fig. 27. The CCD and CCR of the frame structure are approximately 535 mm
and 3635 kN, respectively. Therefore, the CCR of each storey of the frame structure is about 1212 kN.
The calculation process of the CCR and CCD of the frame structure using the proposed method is given as follows.
Step 1: Obtain the required structural parameters.
The floor aspect ratio is calculated by:
6000
AR = = 1.5 (11)
4000
The beam SDR is calculated by:
L 4000 − 200
SDR = = = 9.5 (12)
D 400
The slab STR is calculated by:
4000 − 200
STR = = 31.67 (13)
120

Table 5
Summary of hardening factor of the frame structure with different slab STRs and NCCDs.

Slab STR NCCD

20 22 24 26 28 30 32

28 1.19 1.40 1.61 1.83 2.04 2.25 2.46

30 1.21 1.44 1.66 1.89 2.12 2.35 2.58
32 1.23 1.47 1.71 1.9 2.20 2.45 2.69
34 1.24 1.50 1.76 2.023 2.28 2.54 2.80
36 1.26 1.54 1.81 2.09 2.37 2.64 2.92
38 1.28 1.57 1.86 2.15 2.45 2.74 3.03
40 1.30 1.61 1.91 2.22 2.53 2.84 3.15

15
J. Zhang et al. Journal of Building Engineering 82 (2024) 108179

Fig. 26. Final deformation of the frame structure.

Fig. 27. Resistance-deformation curve of the frame structure.

It should be noted that when calculating the beam SDR and slab STR, the net span of the steel beam is used, which is 3800 mm.
Step 2: The CCD of the frame structure is calculated by:
CCD = (0.16 × 1.5 + 0.84) × (0.2 × 9.5 − 0.63)×400 = 548.64mm (14)

Step 3: The yield resistance of the frame structure is given by:

fy = 767.95 kN (15)

Step 4: Calculate the NCCD of the frame structure:

CCD
NCCD = = 23.96 (16)
va

Step 5: Determine the hardening factor:

f = 1.057 − 0.006 × 23.96 − 0.071 × 31.67 + 0.004 × 23.96 × 31.67 = 1.7 (17)

Step 6: Calculate the CCR of the frame structure:

CCR= 1.7 × 767.95 = 1305.52kN (18)
The CCD of the frame structure predicted by the analytical method is found to be very close to that of the numerical analysis, with
error less than 2 %. For the CCR of the frame structure, the prediction from the analytical method is about 7.6 % greater than that of the
numerical result. These errors are in acceptable range. The overestimation of the CCR is because of the simplification of the boundary
conditions of the floor system. In the analytical analysis, the model of a single-storey floor system with fixed boundary conditions is
adopted. However, in the numerical analysis, the boundaries of the floor are not exactly fixed. The comparison of the critical collapse

16
J. Zhang et al. Journal of Building Engineering 82 (2024) 108179

status of the structure in an edge column loss in numerical analysis and analytical method is shown in Fig. 28.

7. Comparisons
The CCD and CCR of the frame structure obtained from this study and Ref. [18] are compared. In Ref. [18], the CCD of the floor
system is calculated by:
√̅̅̅̅̅̅̅̅
3fy
CCD = L (19)
16E
The calculation of CCR of the floor system in Ref. [18] is similar to that in this study, which is obtained by multiplying the yield-line
load by a correction factor. The detailed calculation process can be found in Ref. [18]. Based on the information of the frame structure
given in Section 6, the CCR and CCD of the floor system can be calculated by the method in Ref. [18]. The results are shown by the red
and blue dashed lines in Fig. 29. It can be seen that when the displacement increases, the difference of the resistance obtained from
numerical analysis and Ref. [18] decreases. Moreover, the stiffness of the load-displacement curve by Ref. [18] is very close to the
numerical result at large deformation. However, Ref. [18] gives excessive underestimations on both the CCD and CCR of the floor
system.

8. Conclusion
In this study, the critical collapse deformation and critical collapse resistance of the steel frame structures caused by an edge
column loss were quantified. Systematic parametric analyses were launched, in which the influences of different structural parameters,
including the floor aspect ratio, rebar diameter, slab thickness, beam span, and beam depth were studied. A simplified calculation
method of the critical collapse resistance of the steel frame cause by an edge column loss was established. A hardening factor was
introduced, defined by the ratio of the critical collapse resistance to the yield resistance of the frame structure. An explicit simplified
calculation method for the hardening factor was established. A calculation procedure for the critical collapse deformation and critical
collapse resistance of the structure was given. The following discoveries are found from this study:
● The critical collapse deformation of the frame structure in an edge column loss can be calculated by a linear function of the floor
aspect ratio, beam span, and beam depth. The regression result reveals that when the beam span is increased by one beam depth,
the critical collapse deformation of the frame structure is increased by 0.2 times of the beam depth.
● The slab rebar and slab thickness have minimal effects on the critical collapse deformation of the frame structure in an edge column
loss. When the slab rebar diameter is decreased by 50 %, the variation of the critical collapse deformation of the structure is less
than 5 %. When the slab thickness is decreased by 30 %, the change of the critical collapse deformation of the structure is within 10
%.
● The hardening factor of the frame structure can be determined by the normalized critical collapse deformation and the slab span-to-
thickness ratio. The hardening factor is positively correlated with both the slab span-to-thickness ratio and normalized critical
collapse deformation. The hardening factor of the frame structure in an edge column loss is more than 3.0, corresponding to the slab
span-to-thickness ratio of 40 and normalized critical collapse deformation of 32.

Author statement
JingZhou ZHANG: Conception, Investigation, Writing-Original Draft.
YingHua HE: Investigation, Writing-Original Draft.
WenJin ZHANG: Writing-review & editing.
GuoQiang LI: Supervision, Review.

Fig. 28. Numerical and predicted critical collapse point.

17
J. Zhang et al. Journal of Building Engineering 82 (2024) 108179

Fig. 29. Comparisons of the method in this study and Ref. [18].

Jing ZHANG: Investigation, Review.

ZhiWei YU: Supervision, Review.

Declaration of competing interest

We declare that we have no financial and personal relationships with other people or organizations that can inappropriately in­
fluence our work and we do not have any commercial or associative interest that represents a conflict of interest in connection with the
work submitted.

Data availability

Data will be made available on request.

Acknowledgements
Financial supports from the National Natural Science Foundation of China (52208154), Guangzhou Municipal Bureau of Science
and Technology (2023A04J1008), and Guangdong Provincial Department of Science and Technology (2022A1515110256) are
acknowledged.

References
[1] ASCE/SEI 7 Minimum Design Loads for Buildings and Other Structures [S], American Society of Civil Engineers, Reston, 2005.
[2] M. Alrubaidi, H. Abbas, H. Elsanadedy, T. Almusallam, R. Iqbal, Y. Al-Salloum, Experimental and FE study on strengthened steel beam-column joints for
progressive collapse robustness under column-loss event, Eng. Struct. 258 (2022), 114103.
[3] B. Meng, H. Li, D. Kong, W. Zhong, Q. Du, K. You, Development of an enhanced top-and seat-angle connection to mitigate progressive collapse, J. Constr. Steel
Res. 211 (2023), 108116.
[4] K. Chen, K.H. Tan, Composite joints with fin plate connections under a middle column removal scenario, J. Constr. Steel Res. 161 (2019) 258–274.
[5] H. Qiao, Y. Chen, J. Wang, C. Chen, Experimental study on beam-to-column connections with reduced beam section against progressive collapse, J. Constr. Steel
Res. 175 (2020), 106358.
[6] B. Meng, W. Zhong, J. Hao, X. Song, Improving anti-collapse performance of steel frame with RBS connection, J. Constr. Steel Res. 170 (2020), 106119.
[7] H. Wang, J. Huo, Y. Liu, M. Elchalakani, Z. Zhu, Dynamic performance of composite beam-column connections subjected to impact loadings, J. Constr. Steel Res.
178 (2021), 106498.
[8] K. Qian, X. Lan, Z. Li, Y. Li, F. Fu, Progressive collapse resistance of two-storey seismic configured steel sub-frames using welded connections, J. Constr. Steel
Res. 170 (2020), 106117.
[9] B. Meng, L. Li, W. Zhong, Z. Tan, Y. Zheng, Anti-collapse performance analysis of unequal span steel-concrete composite substructures, Steel Compos. Struct. 39
(2021) 383–399.
[10] G.M. Laleh, Y. Tian, S.L. Orton, Collapse resistance of reinforced concrete frame beams under sustained loads, J. Struct. Eng. 149 (2023), 04023092.
[11] B. Jiang, G. Li, L. Li, B.A. Izzuddin, Experimental studies on progressive collapse resistance of steel moment frames under localized furnace loading, J. Struct.
Eng. 144 (2018), 4017190.
[12] G. Li, L. Li, B. Jiang, Y. Lu, Experimental study on progressive collapse resistance of steel frames under a sudden column removal scenario, J. Constr. Steel Res.
147 (2018) 1–15.
[13] J. Zhang, G. Li, Y. Wang, H. Li, Quantification of the critical displacement initiating collapse of steel moment frames due to an interior column loss, J. Build. Eng.
54 (2022), 104664.
[14] Y. Alashker, S. El-Tawil, A design-oriented model for the collapse resistance of composite floors subjected to column loss, J. Constr. Steel Res. 67 (2011) 84–92.
[15] Y. Weng, K. Qian, F. Fu, Q. Fang, Numerical investigation on load redistribution capacity of flat slab substructures to resist progressive collapse, J. Build. Eng. 29
(2020), 101109.
[16] J.Z. Zhang, G.Q. Li, J. Jiang, Dynamic effects on steel frames with concrete slabs under a sudden edge -column removal scenario, J. Struct. Eng. 146 (2020),
04020185.
[17] G. Li, J. Zhang, J. Jiang, Analytical modeling on collapse resistance of steel beam-concrete slab composite substructures subjected to side column loss, Eng.
Struct. 169 (2018) 18.
[18] C.G. Bailey, Membrane action of slab/beam composite floor systems in fire, Eng. Struct. 26 (2004) 1691–1703.

18
J. Zhang et al. Journal of Building Engineering 82 (2024) 108179

[19] H. Wang, S. Li, C. Zhai, Experimental and numerical investigation on progressive collapse of self-centering precast concrete frame with infill walls, J. Build. Eng.
78 (2023), 107472.
[20] S. Shan, S. Li, Collapse performances of 3D post-tensioned RC frame structures considering combined influences of infill walls and floor slabs, J. Build. Eng. 77
(2023), 107566.
[21] Y. Gan, J. Chen, M. Xiang, PDEM-based reliability assessment of RC frames against progressive collapse considering initial local failure, J. Build. Eng. 76 (2023),
107198.
[22] J. Wang, Y. Sun, S. Gao, W. Wang, Anti-collapse performance of concrete-filled steel tubular composite frame with RC shear walls under middle column
removal, J. Build. Eng. 64 (2023), 105611.
[23] G. Roverso, N. Baldassino, R. Zandonini, F. Freddi, Experimental assessment of an asymmetric steel–concrete frame under a column loss scenario, Eng. Struct.
293 (2023), 116610.
[24] J.X. Lu, H. Wu, Q. Fang, Progressive collapse of Murrah federal building: revisited, J. Build. Eng. 57 (2022), 104939.
[25] Z. Yang, Y. Li, H. Guan, M. Diao, B.P. Gilbert, H. Sun, L. Xu, Dynamic response and collapse resistance of RC flat plate structures subjected to instantaneous
removal of an interior column, Eng. Struct. 264 (2022), 114469.
[26] J. Zhang, G. Li, R. Feng, R. Chen, A simplified approach for collapse assessment of multi-Storey steel framed-structures with one column loss, J. Constr. Steel Res.
176 (2021), 106391.
[27] P. Pantidis, S. Gerasimidis, Progressive collapse of 3D steel composite buildings under interior gravity column loss, J. Constr. Steel Res. 150 (2018) 60–75.
[28] Ministry of Housing and Urban-Rural Development of China, Standard for Design of Steel Structures (GB 50017-2017), Beijing, 2017.
[29] Y. He, G. Li, J. Zhang, Critical collapse deformation of 3-D steel frame with composite floor system, J. Constr. Steel Res. 208 (2023), 108034.
[30] Y. Bai, T. Wierzbicki, A new model of metal plasticity and fracture with pressure and Lode dependence, Int. J. Plast. 6 (2008) 1071–1096.

19