Journal of Building Engineering 43 (2021) 102518

Contents lists available at ScienceDirect

Journal of Building Engineering

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

Behavior of T-shaped CFST column to steel beam connection with

U-shaped diaphragm
Binyang Li a, b, Yuanlong Yang a, c, d, *, Jiepeng Liu a, d, Xianggang Liu a, e, Yu Cheng a, d,
Yohchia Frank Chen a, d
a
Key Laboratory of New Technology for Construction of Cities in Mountain Area (Ministry of Education), Chongqing University, Chongqing, 400045, China
b
State Key Laboratory for Track Technology of High-Speed Railway, Beijing, 100081, China
c
Key Lab of Structures Dynamic Behavior and Control (Harbin Institute of Technology), Ministry of Education, Harbin, 150090, China
d
School of Civil Engineering, Chongqing University, Chongqing, 400045, China
e
China Southwest Architectural Design and Research Institute Co., Ltd., Chengdu, 610042, China

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

Keywords: The connection between special-shaped concrete-filled steel tubular (CFST) column and steel beam is a particular
Special-shaped concrete-filled steel tubular concern to engineers. A novel U-shaped diaphragm connection is introduced in this paper to transfer the moment
column at beam ends in the frame with special-shaped CFST columns and steel beams. Five T-shaped CFST column to H-
Connection
section steel beam connections were tested under low-cycle horizontal loading. U-shaped diaphragm was used as
U-shaped diaphragm
Seismic behavior
the connection to transfer loads from beam to joint panel zone. The experimental parameters include the size of
Mechanical model U-shaped diaphragm and the axial load ratio of column. Based on the experimental results, the strength, ductility
and strain distribution of the connections were calculated. Test results show that the connection can develop
plastic bending capacity in the beam, satisfying the design code requirements of America, Great Britain, and
China. Furthermore, a parametric analysis was conducted using the finite element method. The influences of U-
shaped diaphragm size, tube thickness, and axial load ratio of column were analyzed. Based on the parametric
analysis results, a mechanical model is proposed to calculate the yield strength and ultimate strength of U-shaped
diaphragm connections. In conclusion, with a proper design, the T-shaped CFST column to H-section steel beam
connection using the U-shaped diaphragm can be effectively used in practical applications.

1. Introduction 2009, Lin et al. [4] reinforced L-shaped CFST columns by using plate
ribs. In 2010, Zhang et al. [5] investigated the seismic behavior of
The structure with special-shaped concrete-filled steel tubular special-shaped CFST columns with plate ribs in frame structures. The
(CFST) columns is an improved architectural approach to traditional above researches showed that with proper stiffeners, special-shaped
frame structure. Compared with square-shaped column which exposes CFST columns exhibited good static and seismic behaviors.
column corners to indoor room, the smooth joints of special-shaped For framed structures, the beam-column connection is a critical
column provide a larger efficiency of indoor room and more conve­ component as it will significantly affect the structural reliability. How­
nience of furniture placement. However, design standard for the special- ever, only a small number of experiments were conducted to study the
shaped CFST column structure is currently unavailable. To apply the connection between special-shaped CFST column and steel beam. In
special-shaped CFST column in the building industry, Chinese re­ 2012, Xue et al. [6] conducted a cyclic loading test on nine CFST column
searchers [1–3] have carried out extensive study to investigate its static to steel beam joint specimens to investigate the failure modes and
and seismic behaviors. seismic behavior of the interior diaphragm joint. In 2015, Liu et al. [7]
The academic research on special-shaped CFST columns in China used extended the end-plate and steel angles to connect between the
began around 2000. To delay the local buckling of steel tube, various beam and special-shaped CFST column. Pseudo-static experiment results
stiffeners were studied, including plate rib, pulled binding bar et al., In have shown the specimens could be classified as semi-rigid joints. In

* Corresponding author. Key Laboratory of New Technology for Construction of Cities in Mountain Area (Ministry of Education), Chongqing University, Chongqing,
400045, China.
E-mail address: [email protected] (Y. Yang).

https://doi.org/10.1016/j.jobe.2021.102518
Received 24 December 2020; Received in revised form 24 March 2021; Accepted 8 April 2021
Available online 20 April 2021
2352-7102/© 2021 Elsevier Ltd. All rights reserved.
B. Li et al. Journal of Building Engineering 43 (2021) 102518

Fig. 1. Sketch of the specimen.

the column and P is the horizontal load caused by an earthquake.

Table 1
Considering the limit of loading devices and actual structural size, a
Details of specimens.
scale ratio of approximately 1/2 was adopted for the specimens. Details
No. Specimen Protruding Embedded n α Comment of specimens are listed in Table 1 and Fig. 2. The specimens consist of
width B1/mm depth B2/mm (%)
three connections with large size U-shaped diaphragm (UL series) and
1 E-UL-0.6- 15 25 0.6 9.8 Same two connections with small size (US series). In order to propose a
1 parameters
calculation method of load carrying capacity of U-shaped diaphragm
2 E-UL-0.6- 15 25 0.6
2
connection, the dimension of U-shaped diaphragm is designed to ensure
3 E-UL-0.3 15 25 0.3 the failure of specimens is the connection failure, based on the finite
4 E-US-0.6 15 15 0.6 9.8 element analysis prediction before experiment.
5 E-US-0.3 15 15 0.3 Steel plates were cold formed to form square steel tube and U-section
Notes: steel. The T-shaped multi-cell CFST column was welded with one square
(i) α = steel ratio of CFST column; steel tube and four U-section steel, which forms into bundled tube
(ii) n = axial load ratio = N0/(Asfy + Acfck); member with good constraint effect for concrete. Besides, the special-
(iii) As = cross-sectional area of steel tube; shaped multi-cell CFST column has simple fabrication process and less
(iv) Ac = cross-sectional area of concrete section; welding residual deformation, compared to other forms of special-
(v) fy = yield strength of steel tube; shaped CFST columns. All welds are single-sided fillet welds. Then,
(vi) fck = prismatic compressive strength of concrete.
the commercial concrete was poured into the tube. Standard material
tests [9] were made to determine the material properties of concrete and
2017, Yang et al. [8] investigated the exterior diaphragm and vertical steel, as listed in Table 2.
rib connections experimentally and numerically. Their test results
showed that the connection with exterior diaphragm exhibited good 2.2. Experimental set-up
ductility and achieved a rigid joint.
A novel U-shaped diaphragm was introduced as special-shaped CFST In Fig. 3, all the beam ends and column ends were pinned to the
column to steel beam connection. The U-shaped diaphragm combines experimental devices to satisfy the boundary condition of inflection
the advantages of exterior diaphragm and interior diaphragm. It has points in Fig. 1. The constant axial compressive loads and cyclic hori­
similar mechanical behavior with exterior diaphragm and minimize the zontal loads were applied at the column top by servo actuators through
interference to indoor space compared with exterior diaphragm. In this the L-shaped rigid beam (Fig. 3). Lateral supports were provided to
study, the seismic behavior of T-shaped CFST column to steel beam prevent the out-of-plane movement of specimens.
connections with U-shaped diaphragm was investigated experimentally The loading scheme of cyclic horizontal loads is load-displacement
and numerically. The experimental parameters include U-shaped dia­ hybrid control scheme [10], as shown in Fig. 4. When reaching the
phragm size and the axial load ratio of column. The strength, ductility, yield load, the loading scheme is transferred to displacement-controlled
and strain distribution were analyzed based on the experimental results. loading scheme. During the load-controlled loading stage, the loading
Furthermore, a finite element analysis (FEA) was performed to investi­ rate of each level is 5 kN/min, and the loading rate is 5 mm/min of each
gate the load transmission mechanism. Based on the experiment and level during the displacement-controlled loading stage. The direction
FEA results, a mechanical model for U-shaped diaphragm connections away from reaction wall was designated as negative of displacement
was proposed. direction. The loads and displacement in this direction was added the
prefix “-“.
2. Experimental design
2.3. Measurement diagram
2.1. Details of specimens
The configuration of measurements is shown in Fig. 5. The loads on
The U-shaped diaphragm connections of exterior columns were
the column and beam were measured by force sensors. The horizontal
designed within the inflection points of beams and columns in a frame
displacement at the column top was measured by linear variable dif­
structure (Fig. 1). Note that in the figure N0 is the vertical load acting on
ferential transformers (LVDTs). Another two LVDTs (D-4 and D-5) were

2
B. Li et al. Journal of Building Engineering 43 (2021) 102518

Fig. 2. Details of specimens (Note: All dimensions in mm).

arranged diagonally in the joint panel zone to measure the shear Fig. 6.
deformation. Four inclinometers were placed closely to the joint panel
zone to measure the rotation angles of beam and column’s plastic 3. Discussion of experimental results
hinges.
Strain gauges were installed on the steel beam and U-shaped dia­ 3.1. Failure modes
phragm to measure the stress development. Four strain rosettes were
placed in the plastic hinge region of beam web; and eight strain rosettes To better describe the failure phenomena of specimens, the faces and
were placed in the joint panel zone to measure the strain distribution cells of multi-cell CFST tube column are numbered, as shown in Fig. 7.
and shear deformation. The arrangement of strain gauges is shown in

3
B. Li et al. Journal of Building Engineering 43 (2021) 102518

Table 2 axial load ratio of column.

Material properties of steel and concrete.
Steel Concrete 3.1.3. Specimen E-C-US-0.6
The specimens in the US series have lower peak load than those in the
Components Thickness (mm) fy (MPa) fu (MPa) fck (MPa)
UL series due to the smaller size of U-shaped diaphragm. Similar to
Beam flange 7.75 381.5 493.1 32.1 Specimen E-C-UL-0.6-1, the failure phenomena of Specimen E-C-US-0.6
Beam web 5.77 402.8 532.0
Column tube 3.07 462.3 490.0
can be categorized as follows: (1) necking phenomenon of U-shaped
U-shaped diaphragm 11.66 329.3 488.8 diaphragm near column corner between Faces 5 and 6 (inter-story drift
ratio = 2.37%) (Fig. 10a); (2) broken U-shaped diaphragm at the same
Note: fu = steel ultimate tensile strength.
position (inter-story drift ratio = -2.60%) (Fig. 10b); and (3) concrete
local compression failure near the U-shaped diaphragm (Fig. 10c).
3.1.1. Specimen E-C-UL-0.6-1 Generally, the main failure mode can be identified as U-shaped dia­
During the load control stage, when loaded to -40kN, slight local phragm failure.
buckling was detected at on Faces 4, 5, and 7 by knocking on the steel
tube. When loaded to ±60 kN, the flange yielded and the load control 3.1.4. Specimen E-C-US-0.3
was then replaced by the displacement control. The horizontal load at Three characteristic phenomena of Specimen E-C-US-0.3 were
column top reached a peak value at the inter-story drift ratios of 1.07% observed: (1) the crack of weld between U-shaped diaphragm and tube
and -1.16%. Then, the load declined to 85% of the peak load at the inter- (inter-story drift ratio = -2.35%) (Fig. 11a); (2) the necking of U-shaped
story drift ratio of ±1.67%. Meanwhile, a separation between concrete diaphragm at the corner of column (inter-story drift ratio = 2.51%)
and tube was detected on Face 5 (Fig. 8a). To investigate the final failure (Fig. 11b); and (3) the concrete local compression failure near U-shaped
mode, the experiment was continued. When loaded to the inter-story diaphragm.
drift ratio of 2.47%, the U-shaped diaphragm broke suddenly
(Fig. 8b). To investigate failure mode of inner concrete, the steel tube
was removed after the experiment. No concrete crack was found in all
cells except Cell ① where local compression failure was noted in the
concrete close to the U-shaped diaphragm (Fig. 8c). The experimental
phenomena of Specimen E-C-UL-0.6-2 are similar to those of Specimen
E-C-UL-0.6-1, thus not repeated here.

3.1.2. Specimen E-C-UL-0.3

Due to the failure of U-shaped diaphragm, reducing axial load ratio
had less effect on failure phenomena. When loaded to ±80 kN, the load
control was replaced with the displacement control. The antirust paint
was peeled off at the corner of upper flange between Faces 4 and 5
(Fig. 9a) when the inter-story drift ratio reached -1.20%. Beyond the
peak point at inter-story drift ratio of ±1.66%, an inclined weld crack
was observed at the corner between Faces 4 and 5 below the bottom
flange (Fig. 9b). During the load declining stage, the crack developed
with increasing displacements. The experiment was stopped at the inter-
story drift ratio of -2.6% due to the necking phenomenon occurring at
the bottom of U-shaped diaphragm (Fig. 9c). Compared with Specimens
E-C-UL-0.6-1 and E-C-UL-0.6-2, concrete local compression failure near
the U-shaped diaphragm (Fig. 9d) was much less because of the smaller Fig. 4. Horizontal loading history.

Fig. 3. Experimental set-up.

4
B. Li et al. Journal of Building Engineering 43 (2021) 102518

Fig. 5. Measurement diagram.

Fig. 6. Strain measurement program.

3.2. Load-deformation curves

Horizontal load-inter story drift hysteretic curves are shown in

Fig. 12. All five specimens have plump hysteretic loops, indicating that
the connections have good energy dissipation capacity. To quantify the
energy dissipation capacity of specimens, their equivalent viscous
damping coefficients he [21] are calculated based on the hysteretic
curves shown in Fig. 12. The he values range from 0.204 to 0.339 being
close to that for the connection between square CFST column and
H-section steel beam [11].
The skeleton curves are shown in Fig. 13 and Table 3 lists the me­
chanical properties of the specimens based on the skeleton curves. The
yield point is determined by the geometric drawing method [12]
(Fig. 14). When the load reduced to 85% of the peak load, the specimens
are identified to be in the ultimate failure state.
In Table 3, as influenced by the second-order effect, the axial load
ratio n has a significant effect on the peak and ultimate points. The peak
load of the specimens with n = 0.6 is only about 58.5% of that with n =
0.3. The ultimate inter-story drift ratio is about 1/45 (n = 0.3) and 1/65
(n = 0.6). Because of the U-shaped diaphragm failure, the peak loads of
Fig. 7. Numbering of faces and cells in multi-cell CFST tube column.

5
B. Li et al. Journal of Building Engineering 43 (2021) 102518

Fig. 8. Failure mode of Specimen E-C-UL-0.6-1.

US-series specimens are about 6.7% (n = 0.3) and 18.1% (n = 0.6) lower where σij =is the stress tensor; Xij =is the back stress tensor; J2 =is the
than UL-series specimens. Among the different specimen series, the peak second invariants of deviator stress tensor; k0 =is the strengthen con­
loads at the beam ends of UL-series specimens are about 7.6% (n = 0.3) stant; dεpij =is the increment of plastic strain tensor; and dp =is the
and 7.8% (n = 0.6) higher than US-series specimens. increment of equivalent plastic strain, which is determined by
The comparison of beam peak loads between the measured results √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
and the calculated values from AISC 360–2010(LRFD) [13], BS EN 2 p p
dp = dε :dε (3)
1993-1-1:2005 [14], and GB 50017-2017 [15] are listed in Table 4 along 3 ij ij
with the loads calculated based on full plastic criteria (Pcp). In conclu­
In Eqs. (1) and (2), k0, c and γ are the material characteristic pa­
sion, the peak loads of all specimens exceed Pcp and code-determined
rameters. Hu [17] recommended the following values based on experi­
strengths due to steel hardening. All joint specimens meet the re­
mental data of Shi et al. [18]:
quirements of related specifications. The peak loads of UL series are
respectively 26.5%,13.7%, and 23.8% higher than those of standards k0 = 0.85fy ; γ = 150; c = 82.5fy (4)
AISC, BS EN, and GB, while the peak loads of US series are respectively
The stress development in the beam flange, beam web, and U-shaped
18.5%,6.75%, and 16% higher than those of standards AISC, BS EN, and
diaphragm is shown in Fig. 15. In all specimens, the flange in Section A
GB.
(Fig. 15e) reached the yielding state at an average inter-story drift ratio
of 0.93%. Due to the increase of width and thickness, the stress of U-
3.3. Stress analysis
shaped diaphragm in Section B (Fig. 15e) was relatively lower. Under
tensile loads, the stress along the edge of Section B is higher than that in
A non-linear kinematic hardening model based on von Mises yield
the middle region, as the tensile load is mainly transferred to the two
criterion is used to calculate the stresses of steel components. The
side plates of column tube. In contrast, the compressive load is mainly
equation of the yield criterion and back stress can be expressed as [16]:
transferred to the concrete in joint panel zone, the compressive stress
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
)̅
( along the edge of Section B is close to that in the middle region. The
f = 3J2 σ ij − Xij − k0 = 0 (1)
stress development rule of web is similar to that of flange and U-shaped
diaphragm. The stress level in Section A is generally higher than that in
2
dXij = cdεpij − γXij dp (2) Section B.
3

6
B. Li et al. Journal of Building Engineering 43 (2021) 102518

Fig. 9. Failure mode of Specimen E-C-UL-0.3.

4. Finite element model contact in the normal direction and friction in the tangential direction
[19] were assumed. The friction coefficient of 0.6 and the maximum
4.1. Modeling and verification surface bond stress of 0.6 MPa [23] were adopted.

The finite element software Abaqus [19] was used to simulate the 4.1.3. Verification
behavior of U-shaped diaphragm connections. To verify the model, the The horizontal load-inter story drift hysteretic curves of models were
dimension and boundary condition of models are same with experi­ shown in Fig. 17. A good agreement between models and specimens can
mental specimens (Fig. 16). be observed. The modeling approach is feasible.

4.1.1. Materials
The five-stage elastic-plastic model proposed by Han et al. [20] was 4.2. Parametric analysis
used to describe the stress-strain relationship of steel components (beam
flange, beam web, U-shaped diaphragm, and tube). The concrete dam­ To investigate the mechanical model of U-shaped diaphragm, a
age plasticity model available in Abaqus [19] was used for the concrete, parametric analysis using the FE method was conducted. The parame­
which adopts the uniaxial stress-strain relationship proposed and ters included protruding width B1 (0.125Bc ~ 0.250Bc), embedded
modified by Liu [21] and has the following five parameters describing depth B2 (0.125Bc ~ 0.292Bc), axial load ratio n (0.0–0.6), and the
the yield function and plastic flow: dilation angle ψ is = 35◦ ; eccentricity thickness to width ratio of steel tube t/Bt (1/80–1/30). Bc and Bt are the
e= 0.1; σb0/σ c0 (ratio of biaxial to uniaxial compressive strength) = 1.16; widths of beam and column cells ①, respectively. The basic parameters
the Kc = 0.667; and viscosity parameter = 0.0001 [22]. were set as: B1 = 15 mm, B2 = 25 mm, Bc = Bt = 120 mm, n = 0.3, and t
= 3 mm. Note that the thickness of web was increased to 10 mm and the
4.1.2. Element types and structural interaction yield stress of beam flange and web was increased to 462 MPa to ensure
The beam flange, beam web, U-shaped diaphragm, and tube were the U-shaped diaphragm failure mode.
modelled using the 4-nodes shell elements with reduced integration Since there were many calculation models for parameter analysis, in
(S4R); and the concrete was modelled with the 8-nodes brick elements order to appropriately reduce the calculation amount, the focus of
with reduced integration (C3D8R) [19]. models was put on the U-shaped diaphragm. Only the joint zone was
To describe the interaction between the steel tube and concrete, hard modelled and the vertical load was applied at the beam end (Fig. 18) to
investigate the behavior of U-shaped diaphragm. To verification the

7
B. Li et al. Journal of Building Engineering 43 (2021) 102518

Fig. 10. Failure mode of Specimen E-C-US-0.6.

Fig. 11. Failure mode of Specimen E-C-US-0.3.

8
B. Li et al. Journal of Building Engineering 43 (2021) 102518

Fig. 12. Horizontal load-inter story drift hysteretic curves.

9
B. Li et al. Journal of Building Engineering 43 (2021) 102518

Fig. 13. Skeleton curves of specimens.

10
B. Li et al. Journal of Building Engineering 43 (2021) 102518

Table 3
Mechanical properties of the specimens.
Specimen Direction Horizontal load at column top Vertical load at beam end

Yield point Peak point Ultimate point Yield point Peak point

Pyc (kN) γyc (%) Ppc (kN) γpc (%) Puc (kN) γuc (%) Pyb (kN) γyb (%) Ppb (kN) γpb (%)

E-C-UL-0.3 (+) 43.0 1.00 46.8 1.58 42.7* 2.27* 107.5 1.56 118.8* 2.27*
(-) -40.0 -1.04 -45.6 -1.67 -43.3 -2.13 -92.3 -1.42 -112* -2.37*
E-C-UL-0.6-1 (+) 27.4 0.88 28.9 1.07 24.6 1.70 104 1.41 118.2 2.24
(-) -24.9 -0.77 -27.9 -1.16 -23.7 -1.66 -101.6 -1.46 -118.3 -2.56
E-C-UL-0.6-2 (+) 24.3 0.84 25.7 0.90 21.8 1.66 105.7 1.56 120* 2.54*
(-) -27.6 -1.06 -28.1 -1.14 -23.9 -1.60 -104.8 -1.68 -118.8* -2.53*
E-C-US-0.3 (+) 43.1 1.27 45.2 1.58 38.4 2.35 103.5 1.50 110.0* 2.51*
(-) -37.6 -0.99 -41.3 -1.88 -35.1 -2.43 -89.7 -1.39 -104.4 -2.35
E-C-US-0.6 (+) 23.3 0.78 25.5 0.88 21.7 1.72 93.6 1.38 114.8* 2.31*
(-) -19.1 -0.80 -21.2 -1.20 -18.0 -1.67 -90.8 -1.44 -105.7 -2.36

Notes:
(1) (Pyc, γyc), (Ppc, γpc), and (Puc, γuc): the column loads and corresponding inter-story drift ratios when the specimen reaches the yield point, peak point, and ultimate
point, respectively;
(2) (Pyb, γyb) and (Ppb, γpb): the beam loads and corresponding inter-story drift ratios when the specimen reaches the yield point and peak point, respectively;
(3) The vertical load at the beam end of some specimens do not reach the ultimate or peak point when the experiment is ended. In this situation, the load and inter-story
drift ratio at the end of experiment are used to replace the counterparts at the ultimate point or peak point. To distinguish these replacements, bold fonts and “*” are
used.

local model, the yield load Fy and ultimate load Fu of beam from ex­ 4.3.2. Deformation of steel tube
periments and models were compared (Table 5). A good agreement When the U-shaped diaphragm reaches the yield stress, the de­
between the experiment and FE results was observed. formations of steel tube are shown in Fig. 21. The deformation scale
The results of parametric analysis are shown in Fig. 19. As seen, there factor is adjusted to 50 to ease the recognition on deformation distri­
is an obvious positive correlation between the strengths and parameters bution. The deformation in the middle of front tube (Part I) is larger than
B1, B2, and t/Bt, revealing that the strength of U-shaped diaphragm that in the edge region (Part II). The width of middle region is approx­
mainly depends on diaphragm size and tube thickness. While, the in­ imately equal to B3 (Fig. 21). Compared with the middle region, by using
fluence of the axial load ratio n is relatively small. parametric analysis, the deformations at points A and A′ are about 50%
of those at point B and B′ . This deformation distribution is similar to the
4.3. Stress and deformation analysis deformation model proposed by Nie [24].

4.3.1. Stress of U-shaped diaphragm

The Mises stress contours of U-shaped diaphragm are shown in 4.4. Mechanical model
Fig. 20. The maximum stress of U-shaped diaphragm is obviously
located in a region near the tube (Fig. 20a). This phenomenon is Based on the stress analysis and deformation model, a mechanical
consistent to that observed in the experiments. The red line in Fig. 20b model for the U-shaped diaphragm is proposed. In this model, the
indicates the weak section of U-shaped diaphragm under tensile loads. strength is contributed by three components: U-shaped diaphragm, front
Therefore, the strength contributed by U-shaped diaphragm can be face of tube, and side face of tube. The calculation method for each
calculated based on this section.
Table 4
Comparison of beam peak loads between the measured results and the calculated
values from the design codes.
Specimen Direction Ppb/Pcp Ppb/PAISC Ppb/PBS Ppb/PGB

E-C-UL-0.3 (+) 1.15* 1.28* 1.15* 1.25*

(-) -1.08 -1.20 -1.08 -1.18
E-C-UL-0.6-1 (+) 1.14 1.27 1.14 1.24
(-) -1.14 -1.27 -1.14 -1.25
E-C-UL-0.6-2 (+) 1.16* 1.29* 1.16* 1.26*
(-) -1.15* -1.28* -1.15* -1.25*
E-C-US-0.3 (+) 1.08 1.20 1.08 1.17
(-) -1.02 -1.14 -1.02 -1.11
E-C-US-0.6 (+) 1.13* 1.25* 1.13* 1.23*
(-) -1.04 -1.15 -1.04 -1.13

Notes:
(1) Refer to Note 3 in Table 3 for the explanation bold-faced values;
(2) Pcp = the vertical load of beam calculated based on the full plastic criterion;
(3) PAISC = the strength of beam determined from AISC 360–2010 (LRFD);
(4) PBS = the strength of beam determined from BS EN 1993-1-1:2005;
Fig. 14. The geometric drawing method. (5) PGB = the strength of beam determined from GB 50017-2003.

11
B. Li et al. Journal of Building Engineering 43 (2021) 102518

Fig. 15. Stress developments of beam flange and U-shaped diaphragm.

12
B. Li et al. Journal of Building Engineering 43 (2021) 102518

Fig. 16. The finite element model.

component is described below. the axial load, the yield moment per unit length M1 can be expressed by
( ) t2
4.4.1. U-shaped diaphragm M1 = 1 − n2 fy,c c (8)
4
The mechanical model of U-shaped diaphragm is shown in Fig. 22
where the red line represents the weak section containing parts ① and When the yield line is parallel to the axial load, the yield moment per
②. The strength of U-shaped diaphragm Py1 can be expressed as unit length M2 can be found from

Py1 = Py1,1 + Py1,2 (5) t2 2 − 2n2

M2 = kfy,c c ; k = √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ (9)
4 4 − 3n2
{
Py1,1 = 2B
√1̅̅t̅u fy,u (6) where n is the axial load ratio and fy,c and tc are respectively the yield
Py1,2 = 2 2ηB2 tu fy,u
stress and thickness of steel tube.
1 1 For the inclined yield line, Wood [26] proposed a stepped failure line
η = √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ = √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ = 0.707 (7)
1 + 2cos 2 α 1 + 2cos 2 45∘ model for the yield line analysis of reinforced concrete slabs. This model
was also valid for steel plates and adopted in this paper. The original
where tu is the thickness of U-shaped diaphragm; fy,u is the yield stress of inclined failure line is replaced by a stepped line, and the calculation of
U-shaped diaphragm; η is the impact factor of angle available in the yield moment along the inclined line is thus replaced by the calculation
Chinese code GB 50017-2017 [15], α is the angle between the failure of yield moment along the horizontal line and vertical line.
line and the tension axis=45◦ .
(2) Mechanical model for the front face of tube
4.4.2. Steel tube
As described in Section 4.3 (Fig. 21c), the main deformation for the
(1) Effect of axial load front face and side face of tube are respectively caused by bending (yield
line) and tensioning. To make the description clearly, their mechanical
The axial load of column will reduce the contribution of steel tube, models are described separately in following sections.
which should be considered in the mechanical model. Cao et al. [25] The yield mechanism of front face tube is shown in Fig. 23. Based on
proposed an orthotropic model for connections with axial load, in which the FEA results, the deformation at Point 3 is about 50% of that at Point
the yield moments in two orthogonal directions were found to be 5. Using the principle of virtual work with Δ as the virtual displacement,
different. If the Von Misses criterion is applied, then the yield moment the strength provided by front tube can be determined by
per unit length can be calculated. When the yield line is perpendicular to

13
B. Li et al. Journal of Building Engineering 43 (2021) 102518

Fig. 17. Comparison between load-deformation curves from models and specimens.

14
B. Li et al. Journal of Building Engineering 43 (2021) 102518

√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
√ 12M1 Bc B2
√
x = √( √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ) (13)
− n + 4 − 3n2 fy,c tc B2 + 12M2

(4) Yield and ultimate strengths of U-shaped diaphragm connector

The yield strength of U-shaped diaphragm connector can be

expressed as
Py = Py1 + Py2 + Py3 (14)

And the ultimate strength of the connector can be described by

Pu = Pu1 + Pu2 + Pu3 (15)
Based on the equations of Py1, Py2, and Py3, Pu1, Pu2, and Pu3 can be
calculated by replacing the yield strengths (fy,u and fy,c) with the ulti­
mate strengths (fu,u and fu,c).
Fig. 18. The simplified finite element model.
4.5. Verification of the mechanical model

Table 5 When the connection achieves the yield or ultimate strength, the
Comparison of the yield and peak loads at beam end between the experiment vertical load at beam end can be calculated by
and FEM.
Py ⋅h Pu ⋅h
Specimens Fy (kN) Fu (kN) Fy = ; Fu = (16)
L L
Experiment FEM Experiment FEM

E-C-US-0.3 65.6 62.1 107.2 99.6 where h is the distance between the centroids of top and bottom flanges
E-C-US-0.6 60.4 65.2 110.3 103.8 and L is the distance from the beam end to the front of column tube.
E-C-UL-0.3 80.5 80.5 115.4 114 To verify the mechanical model, a comparison of vertical loads at the
E-C-UL-0.6-1 81.1 82.1 118.3 114.2
beam end between the FEM and mechanical model was made. The FE
E-C-UL-0.6-2 80.2 82.1 119.4 114.2
results are from the parametric analysis presented in Section 4.2 and the
mechanical model is from Section 4.4. A good agreement was obtained,
as shown in Fig. 25, with the mechanical model slightly more
conservative.
0.35Δ Δ
W = 2(2W13 + 2W15 + W12 + W56 ) = 8M2 x + 4M1 Bc (10) 5. Conclusions
B2 x

1 1 Based on this study, the following hey findings are offered:

Py2 = 4M2 x + 4M1 Bc (11)
B2 x
(1) The connection between T-shaped CFST column and steel beam
where W is the total internal virtual work; Wij is the internal virtual work with U-shaped diaphragm show good strength and energy dissi­
along line i-j; the unknown factor x is the tension affected height of steel pation. The load-carrying capacity of beam satisfies the re­
tube. quirements of AISC 360–2010(LRFD), BS EN 1993-1-1:2005, and
The unknown factor x in Fig. 23 can be determined by minimizing GB 50017-2017. With a proper design, this relatively new
the total strength provided by the front and side faces of tube. Its for­ connection can be used in practical applications.
mula is presented in the following section. (2) Due to the primary failure from U-shaped diaphragm (connector),
the size of U-shaped diaphragm will decrease the strength of
(3) Mechanical model for the side face of tube specimen significantly. The strength of UL-series specimens is
7.7% higher than US-series specimens. The connector failure can
The mechanism for the side face of tube is shown in Fig. 24. Based on be changed to beam hinge failure by increasing the size of U-
the principle of virtual work, the following formula can be derived: shaped diaphragm.
( √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ) ( √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ) (3) A parametric analysis was conducted using the finite element
∫ x − n + 4 − 3n2 fy,c tc − n + 4 − 3n2
0.5 (FE) method to investigate the mechanical model of connection.
Py3 = 2 z zdz = fy,c tc x
0 2x x 3 Based on the parametric analysis results and previous experi­
(12) mental research, a mechanical model is proposed. The strengths
of U-shaped diaphragm, front tube plate, and side tube plates are
Then, the unknown factor x can be determined by minimizing the
considered in this model. The strengths calculated by the model
solution (Py2+Py3), giving.
are in good agreement with the FE results.

15
B. Li et al. Journal of Building Engineering 43 (2021) 102518

Fig. 19. Parametric analysis.

16
B. Li et al. Journal of Building Engineering 43 (2021) 102518

Fig. 20. The stress analysis of U-shaped diaphragm.

17
B. Li et al. Journal of Building Engineering 43 (2021) 102518

Fig. 21. Deformation analysis of steel tube.

Fig. 22. Mechanical model of U-shaped diaphragm.

Author statement

Binyang Li: Methodology; Formal analysis; Investigation; Writing –

review & editing; Visualization. Yuanlong Yang: Conceptualization;
Resources; Investigation; Writing – review & editing; Supervision; Fig. 23. Deformation model of front tube.
Fundingn, Jiepeng Liu: Supervision; Writing Editing; Verification;
Investigation; Visualization; Xianggang Liu: Data curation; Simulation.
Yu Cheng: Data curation; Simulation. Yohchia Frank Chen: Reviewing
and Editing.

18
B. Li et al. Journal of Building Engineering 43 (2021) 102518

Fig. 24. Mechanical model for the side face of tube.

Science Foundation Project of CQ CSTC (Grant No. cstc2019jcyj-

msxmX0580), Foundation of Key Laboratory of Structures Dynamic
Behavior and Control (Ministry of Education) in Harbin Institute of
Technology (Grant No. 30620180333), and Research on bridge con­
struction risk and control technology under complicated environment of
Sichuan-Tibet Railway (Grant No. 2019YJ036).

References

[1] D. Zhang, X.G. Ye, Non-linear analysis of special-shaped reinforced concrete

columns, J. Hefei Univ. Technol. (Nat. Sci.) 26 (4) (2003) 490–494 [in Chinese].
[2] D.X. Gao, J. Ke, L.H. Wang, Seismic behavior analysis of special-shaped column
frame structure, J. Xi’an Univ. Technol. 21 (3) (2005) 285–288 [in Chinese].
[3] T.C. Wang, X.H. Li, T.Z. Wang, G.Y. Kang, Hysteretic behavior of frame with
specially shaped columns subjected to cyclic loading, J. Jilin Univ. (Eng Technol
Ed). 37 (1) (2007) 224–228 [in Chinese].
[4] Z.Y. Lin, Z.Y. Shen, J.H. Luo, Hysteretic behavior of concrete-filled L-section steel
tubular columns under cyclic loading, Progr. Steel Build. Struct. 11 (2) (2009)
12–17 [in Chinese].
[5] J.C. Zhang, Z.Y. Shen, Z.Y. Lin, J.H. Luo, Experimental research on seismic
behavior of concrete-filled L-shaped steel tubular frames, J. Build. Struct. 31 (8)
(2010) 1–7 [in Chinese].
[6] P. Zhou, J.Y. Xue, X. Chen, et al., Experimental study on seismic performance of
joints between concrete-filled square steel tubular special-shaped columns and
Fig. 25. Comparison of beam loads between the FEM and the mechanical steel beams, J. Build. Struct. 33 (8) (2012) 44–50 (in Chinese).
model. [7] J.X. Liu, S.B. Dai, K.C. Huo, J. Huang, M.S. Lin, Study on earthquake resistance
Note: FFEM = Vertical load from the FEM; FMM = Vertical load from the me­ behavior of top and seat angle joint of special-shaped concrete-filled rectangular
composite tubular column with steel beam, J. Sichuan Univ. (Eng Sci Ed). 47 (1)
chanical model.
(2015) 128–137 [in Chinese].
[8] J.C. Liu, Y.L. Yang, J.P. Liu, X.H. Zhou, Experimental investigation of special-
shaped concrete-filled steel tubular column to steel beam connections under cyclic
loading, Eng. Struct. 151 (2017) 68–84.
[9] GB50010-2010. Code for Design of Concrete Structures, China Planning Press,
Declaration of competing interest Beijing, 2010 [in Chinese].
[10] JGJ/T 101-2015, Specification for Seismic Test of Buildings, 2015 [in Chinese].
The authors declare that they have no known competing financial [11] X.T. Wang, et al., Experimental study on seismic behavior of penetrated interior
diaphragm on one side beam to column connection for concrete filled square steel
interests or personal relationships that could have appeared to influence
tube, J. Chongqing Jianzhu Univ. 29 (2) (2007) 73–77.
the work reported in this paper. [12] Z.H. Guo, X.D. Shi, Reinforced Concrete Theory and Analyse, Tsinghua University
press, Beijing, 2003.
[13] AISC 360-2010 Specification for Structural Steel Buildings, AISC, Chicago, 2010.
Acknowledgments
[14] BS EN 1993-1-1, Eurocode 3: Design of Steel Structures: Part 1-1: General Urles and
Rules for Buildings, BSI, London, 2005.
The authors are very grateful to the support provided by National [15] GB 50017-2017, Code for Design of Steel Structures, China Architecture and
Natural Science Foundation of China (Grant No. 51878098), Natural Building Press, Beijing, 2017, 2017.

19
B. Li et al. Journal of Building Engineering 43 (2021) 102518

[16] J.L. Chaboche, Time-independent constitutive theories for cyclic plasticity, Int. J. [22] X.G. Liu, J.P. Liu, Y.L. Yang, et al., Resistance of special-shaped concrete-filled steel
Plast. 2 (2) (1986) 149–188. tube columns under compression and bending, J. Constr. Steel Res. (2020) 169.
[17] H.S. Hu, Concrete-filled Steel Plate Composite Coupling Beam and its Application [23] J.P. Liu, X.H. Zhou, D. Gan, Effect of friction on axially loaded stub circular tubed
to Shear Wall Structures. Dissertation for the Doctor Degree in Engineering, columns [J], Adv. Struct. Eng. 19 (3) (2016) 546–559.
Tsinghua University, 2014 [in Chinese]. [24] J.G. Nie, K. Qin, C.S. Cai, Seismic behavior of composite connections — flexural
[18] Y.J. Shi, M. Wang, Y. Wang, Experimental and constitutive model study of capacity analysis, J. Constr. Steel Res. 65 (5) (2009) 1112–1120.
structural steel under cyclic loading, J. Constr. Steel Res. 67 (8) (2011) 1185–1197. [25] J.J. Cao, J.A. Packer, G.J. Yang, Yield line analysis of RHS connections with axial
[19] SIMULIA. ABAQUS 6.12, Analysis User’s Manual, 2012. Providence, RI. loads, J. Constr. Steel Res. 48 (1) (1998) 1–25.
[20] L.H. Han, X.L. Zhao, Z. Tao, Tests and mechanics model for concrete filled SHS stub [26] R.H. Wood, Plastic and Elastic Design of Slabs and Plates, Thames & Hudson,
columns, columns and beam-columns, Steel Compos. Struct. 1 (1) (2001) 51–74. London, 1961.
[21] W. Liu, Research on Mechanism of Concrete-Filled Steel Tubes Subjected to Local
Compression. Dissertation for the Doctor Degree in Engineering, Fuzhou
University, 2005 [in Chinese].

20