Thin-Walled Structures 85 (2014) 57–70

Contents lists available at ScienceDirect

Thin-Walled Structures
journal homepage: www.elsevier.com/locate/tws

Behaviour of multi‐cell composite T-shaped concrete-filled steel

tubular columns under axial compression
Y.Q. Tu n, Y.F. Shen, P. Li
Department of Civil Engineering, Beijing University of Aeronautics and Astronautics, Beijing 100191, China

art ic l e i nf o a b s t r a c t

Article history: To improve the behaviour of conventional T-shaped concrete-filled steel tubular (CFST) columns, multi-
Received 21 April 2014 cell composite T-shaped concrete-filled steel tubular (MT-CFST) columns are proposed in this paper.
Received in revised form Experimental study of 25MT-CFST columns, including 13 short specimens with various cross sections
5 August 2014
and material properties and 12 slender specimens with different slenderness ratios, subjected to axial
Accepted 5 August 2014
loads was conducted. The failure modes, axial load–strain curves for short specimens and axial load–
lateral deflection curves for slender specimens were investigated. The test results were compared with
Keywords: design approaches for conventional CFST columns presented in Eurocode 4, AISC specification, Australian
Concrete-filled steel tube standard AS51006, Chinese code CECS159, and Hong Kong steel code, and it was found that all the design
T-shaped cross section
codes underestimate the bearing capacity of both short and slender MT-CFST columns to some extent.
Multi-cell
& 2014 Elsevier Ltd. All rights reserved.
Composite columns
Mechanical behaviour

1. Introduction strength concrete. Based on the experimental data, a numerical

Concrete-filled steel tubular (CFST) columns could make full use
of the properties of steel and concrete. As a kind of composite
structure with higher bearing capacities and better seismic perfor-
mance, they are used in more and more applications in residential
buildings with good economic benefits. The first experimental
research on circular CFST columns was published by Kloppel and
Goder [1] in German. Since then, experimental studies that mainly
focused on the structural behaviour of CFST columns with a variety of
cross-sectional shapes have been reported. To investigate the influ-
ence of the cross-sectional shape and the thickness of steel tube on
the failure load of the short CFST columns, Schneider [2] performed a
comprehensive experimental and analytical study on the behaviour
of composite columns under axial compression. Han [3] tested 24
specimens with constraining factors from 0.5 to 1.3 and width ratios
of rectangular tube (h/b) from 1.0 to 1.75. It was found that the
constraining factor and width ratio had significant effects on both
the failure load and ductility of the composite column. A series of
research in [4–5] discusses the influence of different concrete
strengths and cross-sectional geometries on the behaviour of CFST
columns filled with high strength concrete.
An extensive experimental test that investigated the perfor-
mance of short and slender concrete-filled stainless steel tubular
columns under different loading methods was carried out by Uy
et al. [6]. Mursi and Uy [7] experimentally studied slender square
CFST columns using high strength structural steel and normal
n
Corresponding author.
model was proposed to study the behaviour of slender concrete
filled high strength steel tubular columns with consideration of
material and geometric nonlinearities. An et al. [8] studied the
behaviour of slender circular CFST columns under axial compres-
sion and developed a finite element analysis method considering
the initial out-of-straightness imperfection. Hernandez-Figueirido
et al. [9] investigated the effect of slenderness on square and
rectangular CFST columns subjected to axial load and a non-
constant bending moment distribution through experiments with
large number of specimens. Eighteen specimens were tested by
Han et al. [10] to investigate the effect of cross-sectional shape,
initial deflection, and slenderness ratio on the behaviour of the
curved composite members under axial compression. In addition,
numerical analysis of slender elliptical CFST columns subjected to
axial load was published in [11].
Although significant progress has been made, engineering
practices at present show that the problem of pillars protruding
from walls remains to be solved despite the fact that the section
size of a CFST column has been greatly reduced compared to that
of a reinforced concrete column. To solve this problem, the concept
of T-shaped cross-sectional CFST columns located at the edges of a
structure has been proposed in [12]. However, for a T-shaped
cross-sectional column, the separation between the steel tube and
concrete at inner corners could prevent two materials from fully
functioning and the large width-to-thickness ratio of the steel
plate could result in premature local buckling. Du et al. [13]
developed a method of welding two rectangular tubes together
to form a T-shaped column for preventing inner corners' separa-
tion. Nevertheless, the premature local buckling still existed in the

http://dx.doi.org/10.1016/j.tws.2014.08.003
0263-8231/& 2014 Elsevier Ltd. All rights reserved.
58 Y.Q. Tu et al. / Thin-Walled Structures 85 (2014) 57–70

composite column. Yang et al. [14] conducted an experimental
study of welding battlement-shaped bar stiffeners and tensile
bar stiffeners on tube surfaces to improve the axial compressive
behaviour of T-shaped CFST columns. But it was expensive to weld
bar stiffeners on the steel plates and fold the steel plates into
a T-shaped cross-sectional hollow tube. Instead of two rectangular
steel tubes or stiffeners, this paper proposes the multi-cell
composite T-shaped concrete-filled steel tubular (MT-CFST)
column for targeting these two problems.
The main objective of the present paper is to understand the
behaviour of MT-CFST column subjected to axial load. The experi-
mental campaign is focused on two aspects: (1) short columns
with various cross-sectional geometries and material properties;
(2) slender columns considering different slenderness ratios. In
addition, comparisons between experimental results and available
design codes are discussed in this paper to validate the feasibility
of the design codes.
2. Experimental investigation
columns that were subjected to axial loads so as to investigate
the effect of cross-sectional geometries (wall thickness, c/a and b/a
ratio) and material properties (concrete strength). Group II con-
tained 15 columns that were subjected to axial compression so as
to study the influence of the slenderness ratio. Details of the
specimens are listed in Tables 1 and 2, and the parameters (a, b, c
and t) and axis (x–x and y–y) of the composite T-section are
defined in Fig. 2. To avoid the effects of overall buckling and end
conditions, the physical length (L, not including the cover-plates)
of the short specimen was chosen to be 3 times the width of the
longest side of the T-section [15].
2.2. Material properties
The material properties of steel and concrete were obtained
according to the requirements defined in China National Codes
(GB/T 50081-2002 and GB/T 228-2002). Test coupons were fabri-
cated with the steel material used for constructing the steel tubes
and were tested in tension. The test results suggest that these steel
coupons don't have obvious yielding platform. Fig. 3 shows the

2.1. General Table 1

Test data of specimens in Group I.
A MT-CFST column is formed by first welding three rectangular
Specimen a
b
c t L fy fck N us SI DI
tubes together. After that, a 10 mm thick steel cover-plate used as (mm3) (mm) (mm) (MPa) (MPa) (kN)
a base for pouring the concrete is welded on the bottom of the
tubes (Fig. 1(a)). Then the concrete is vertically poured into the 1 60
100
80 2.4 450 326 55.3 1550.5 1.041 4.50
steel tube (Fig. 1(b)) and vibrated simultaneously by a poker 2 60
120
60 2.4 500 326 55.3 1497.3 1.073 —
3 60
100
120 2.4 700 326 55.3 1961.0 1.043 3.71
vibrator. After two weeks curing, a high-strength epoxy is adopted
4 60
80
80 2.4 450 326 55.3 1466.7 1.053 —
to fill the space generated by the shrinkage of the concrete and 5 60
120
80 2.4 500 326 55.3 1661.3 1.046 4.17
then another 10-mm-thick steel cover-plate is welded on top of 6 60
120
120 2.4 700 326 55.3 2033.0 1.028 2.43
the specimen. 7 60
80
80 2.0 450 330 55.3 1281.0 1.090 2.03
Experiments with 25 specimens under axial compression were 8 60
80
80 3.4 450 315 55.3 1849.3 1.150
9 60
100
80 2.4 450 326 38.7 1314.5 1.051 —
carried out. The specimens were divided into two groups based on 10 60
100
80 2.4 450 326 61.0 1628.6 1.036 4.21
the influence of different factors. Group I contained 10 short

Fig. 1. Multi-cell composite T-shaped concrete-filled steel tubular (MT-CFST) columns.

 Y.Q. Tu et al. / Thin-Walled Structures 85 (2014) 57–70 59

Table 2
Test data of specimens in Group II.

Specimen a
b
c (mm3) t (mm) L (mm) f y (MPa) f ck (MPa) Axis λx λy λyw N ul (kN) χm

1-x-450 60
100
80 2.4 450 326 55.3 x–x 12.2 1564.8

1-x-1300 2.4 1300 326 55.3 x–x 31.2 1349.3 0.862
1-y-1300 2.4 1300 326 55.3 y–y 37.0 37.5 1276.7 0.816
1-x-1800 2.4 1800 326 55.3 x–x 42.3 1207.5 0.772
1-y-1800 2.4 1800 326 55.3 y–y 50.3 50.6 1198.3 0.766
2-x-500 60
120
60 2.4 500 326 55.3 x–x 11.4 1418.6
2-x-1300 2.4 1300 326 55.3 x–x 26.6 1261.1 0.889
2-y-1300 2.4 1300 326 55.3 y–y 51.1 51.9 1209.2 0.852
2-x-1800 2.4 1800 326 55.3 x–x 36.1 1146.5 0.808
2-y-1800 2.4 1800 326 55.3 y–y 69.4 69.9 963.8 0.679
6-x-700 60
120
120 2.4 700 326 55.3 x–x 16.3 1994.7
6-x-1300 2.4 1300 326 55.3 x–x 28.6 1733.6 0.869
6-y-1300 2.4 1300 326 55.3 y–y 24.4 25.1 1818.8 0.912
6-x-1800 2.4 1800 326 55.3 x–x 38.8 1487.8 0.746
6-y-1800 2.4 1800 326 55.3 y–y 33.1 33.5 1576.1 0.790

Table 3
Steel coupon tests results.

Group σ 0:2 Ultimate strength Elastic modulus Poisson's

(MPa) (MPa) (GPa) ratio

t ¼2.0 mm 315 358 208 0.27

t ¼2.4 mm 326 395 208 0.27
t ¼3.4 mm 330 398 209 0.29

Vertical strain Transverse strain

gauge gauge
1 2

9 3

Fig. 2. Cross section of the MT-CFST column. 8 4

450
7 5
400

350

300 6
Stress (MPa)

250 Fig. 4. Surface numbering and layout of strain gauges.

t=2.0mm
200 t=2.4mm
t=3.4mm properties. Specimens were classified into 8 groups in terms of
150 the casting sequence, where a set of 6 concrete prisms were made
for each casting group. Average prism compressive strength (fck) at
100 the 28th day is shown in Tables 1 and 2, and the average elastic
modulus (Ec) for the concrete with the strength of 55.3 MPa is
50
35,500 MPa. According to the study of Teng and Zhu [16], the
0
0 cylinder compressive strength ðf c Þ is equal to 1.16 times the prism
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08
compressive strength (fck).
Strain

Fig. 3. Typical stress–strain curves of steel.

2.3. Test setup

typical stress–strain curves selected from the tension tests. The
average 0.2% proof stress (σ 0:2 ) and average elastic modulus (Es) for
steel coupons with the thickness of 2.4 mm are 326 MPa and
208 GPa, respectively. More details about the material properties
of steel tubes obtained from steel coupon tensile tests are
summarized in Table 3.
Concrete prisms (100
100
300 mm3 ) made at the time of
the concrete casting were tested to determine the concrete
The experiments were conducted at Institute of Building
Structures, China Academy of Building Research (CABR). The
specimens were tested on a 500 t capacity testing machine
30 days after the casting. For the sake of convenience, the faces
of the specimens are numbered from 1 to 9, as depicted in Fig. 4.
Both vertical and transverse strain gauges were arranged on all the
face centres of the specimen to record the axial and transverse
strains, respectively.
60 Y.Q. Tu et al. / Thin-Walled Structures 85 (2014) 57–70

In Group I, as illustrated in Fig. 6(b), short specimens were variable displacement transducers (LVDTs) were employed to
placed on the bottom rigid padding plate and were loaded directly record the vertical axial displacement, in which 4 LVDTs were
by the raising of the bottom rigid plate of the machine. 6 linear arranged symmetrically between the two rigid plates of the
machine and the other 2 were placed symmetrically between
the two cover-plates of the specimens. In order to explain the
20mm installation of LVDTs explicitly, only 3 LVDTs are shown in Fig. 6(a).
For specimens in Group II, two hinges (in Figs. 5 and 6(c)) that had
sharp knives installed at the middle were used at both ends of
40mm each specimen to ensure that the specimen bent about the
unsymmetrical x–x axis or about the symmetric y–y axis. Thus,
the effective length (Le ) of a specimen was calculated from its
physical length plus 100 mm. The vertical axial displacement and
200mm
the lateral deflection of the column were recorded via 4 LVDTs
Fig. 5. Size of the hinge. arranged at the middle of the specimen and 2 LVDTs arranged

Fig. 6. Test setup, (a) Test setup for short specimen (b) Illustrative drawing of loading for short specimen (c) Test setup for slender specimen.
 Y.Q. Tu et al. / Thin-Walled Structures 85 (2014) 57–70
between the two rigid plates of the machine, respectively. The test
setup for slender specimens under axial compression is shown in
Fig. 6(c). A computerized data acquisition system was used for
collecting the load and deformation data.
A pre-load level of 20 kN was applied to make sure that the
plates of the machine and the specimens were attached firmly to
each other. When the axial load reached 80% of the estimated
ultimate load, the load increment changed to half. Each increment
was maintained for 2 min, and all specimens were compressed to
failure.
3. Experimental results and discussions
3.1. Behaviour of short columns under axial compression
3.1.1. Failure phenomenon
In general, there was similarity in the failure procedures of
specimens. No obvious deformation occurred at surfaces of the
specimens in the initial loading process. At higher levels of
loading, sounds were heard clearly due to the cracking of the
concrete. Once local buckling started at one surface of the speci-
men, consequential local buckling appeared at neighbouring
surfaces of the same height or the specimen displayed staggered
buckling pattern.
A general view of short MT-CTST columns after testing is shown
in Fig. 7. It is obvious that the position of local buckling is different
across specimens. The final amount of local buckling in these
specimens is different as a consequence of differences in the
strains after the failure load. As illustrated in Fig. 7(a), typical
Fig. 7. A general view of short specimens after testing (a) Typical short specimen (b) Short specimen buckled near the bottom (c) Short specimen buckled near the top
(d) Specimen 8.
61
failure was observed by the buckling of the tubes between one end
and the mid-height of the specimen, followed by concrete crush-
ing and the outer steel tube separating from the in-filled concrete.
Some specimens locally buckled outward near the bottom or near
the top, and displayed multi-wave bulking mode, as shown in
Fig. 7(b) and (c). Compared with other short MT-CFST columns,
specimen 8 (in Fig. 7(d)) did not exhibit obvious local deformation
due to the small width-to-thickness ratio and the low strain
development after the failure load. For all the tested specimens
in this paper, there was no weld failure occurrence throughout the
loading process.
3.1.2. Ultimate load and ductility of short columns
To evaluate the positive interaction between concrete and steel
tube, the bearing capacity index (SI) is defined as
SI ¼ N us =Nu ð1Þ
where Nus is the short columns' experimental bearing capacity,
and Nu is the nominal bearing capacity which is determined in
Eq.2.
Nu ¼ f y As þAc ð2Þ
where As and Ac are the area of steel and concrete in the composite
0
cross section, respectively; fy is the yield strength of steel and f c is
the cylinder compressive strength of concrete.
The value of SI is listed in Table 1. It can be found that, in
general, SI decreases with the increase in c/a or b/a ratio, ranging
from 1.028 to 1.073. With a SI of 1.150, the bearing capacity is
greatly enhanced for specimen 8 compared with other MT-CFST
columns in Group I. It is also clear that increasing the compressive
62 Y.Q. Tu et al. / Thin-Walled Structures 85 (2014) 57–70

strength of the in-fill concrete has no obvious effect on the
variance of SI. As shown in Table 1, SI varies slightly from 1.036
to 1.051.
One of the most significant characteristics of CFST columns is
their high ductility which is beneficial to the seismic resistance
capability of a structural system. Nevertheless, no unified formula
is available to determine the ductility of varied cross-sectional
CFST columns. Based on the definition of the ductility performance
index (DI) for traditional CFST columns suggested in [3], the DI of
MT-CFST columns can be defined as
DI ¼ ε0:90 =εy
where ε0:90 is the axial strain corresponding to 90% of the ultimate
load and εy is the axial strain when the composite section is at
yield. The yield strain εy is defined as the strain of the intersection
point of two lines: the initial elastic tangent of the axial load–
strain curve and the flat tangent of the ultimate point on the axial
load–strain curve.
The value of DI is presented in Table 1, where symbol ‘-’
indicates that very large axial strain has occurred before the load
falls to 90% of the ultimate load and represents a perfect ductility.
It can be seen that, except the DI of specimen 8 that didn't attain
the demanding strain, the DIs of all other short specimens in
Group I are higher than 2.0, which shows good ductility perfor-
mance of the MT-CFST columns.
3.1.3. Effect of c/a or b/a ratio of the composite section
The axial strain (ε) is defined as the ratio of axial shortening ðΔLÞ
obtained from the LVDTS to the physical length (L) of the specimen.
Fig. 8(a) and (b) shows the relationships between axial load (N) and
axial strain (ε) of specimens (1–6) in Group I. It can be seen that the
ultimate load and axial stiffness of MT-CFST columns increase with an
increase in c/a or b/a ratio of the cross section, whereas the ductility of
MT-CFST columns decreases. On one hand, the area of composite
ð3Þ
cross section increases with c/a or b/a ratio, which means an increase
in the plastic resistance as presented in Eq. 2. On the other hand, with
the extending of web or flange, the steel tubes have little effect on the
confinement to the encased concrete, which results in a reduction in
the ductility performance of the columns.
3.1.4. Effect of the thickness of the steel tube
Fig. 8(c) depicts the effect of the thickness of the steel tube on
the axial load (N) and axial strain (ε) curves of specimens 4, 7 and
8. With the increase of steel tube thickness, the width-to-thickness
ratio of the steel plate for the same cross section decreases. At the
same time, the yield strength of the steel increases, as shown in

2000 2500

2000
1500
Axial load (KN)

Axial load (KN)

1500
1000
1000

500 spcimen1-60×100×80 spcimen2-60×120×60

500
spcimen4-60×80×80 spcimen5-60×120×80
spcimen5-60×120×80 spcimen6-60×120×120
0 0
0 0.02 0.04 0.06 0.08 0 0.01 0.02 0.03 0.04 0.05 0.06
Axial strain Axial strain

2000 2000

1500 1500
Axial load (KN)

Axial load (kN)

1000 1000

500 spcimen4-t=2.4mm 500 spcimen1-fck=55.3MPa

spcimen7-t=2.0mm spcimen9-fck=38.7MPa
spcimen8-t=3.4mm spcimen10-fck=61.0MPa
0 0
0 0.02 0.04 0.06 0.08 0 0.01 0.02 0.03 0.04 0.05
Axial strain Axial strain
Fig. 8. Effect of geometries and material properties on axial load–displacement curves (a) Effect of b/a ratio on N–ε curves (b) Effect of c/a ratio on N–ε curves (c) Effect of the
steel tube thickness on N–ε curves (d) Effect of concrete compressive strength on N–ε curves.
 Y.Q. Tu et al. / Thin-Walled Structures 85 (2014) 57–70 63

Table 1. It was noted that a significant rise in the section's capacity

was associated with the increase of the thickness of the steel tube. 1600
In addition, specimen 7 with thin steel tubes developed local V7
1400
buckling earlier with lower loading. This can be explained that in-
filled concrete cracked and crushed earlier due to lower confine- 1200
ment. Specimen 8 developed less axial strain after the failure load V6

Axial load (KN)

than other specimens. It seems that the loading process was 1000
interrupted before the required strain. Despite that, compared
with specimens 4 and 7, specimen 8 which has thick steel tubes 800 V5
and thus higher confinement is harder to buckling before the V3
ultimate load. 600
V9
V1
400 V8
3.1.5. Effect of concrete compressive strength
The effect of concrete compressive strength on the axial load (N) 200
and axial strain (ε) curves of specimens 1, 9 and 10 is shown in Fig. 8
(d). It can be seen that increasing the compressive strength of in-filled 0
concrete results in an increase in the ultimate load and axial stiffness -20000 -15000 -10000 -5000 0 5000
of MT-CFST columns. Besides, through comparison, it is known that Axial strain (με)
the difference between the two ultimate loads is only 0.05 in the range
of fck ¼ 55.3 MPa and fck ¼61.0 MPa. Since the compressive strength of
1600
in-filled concrete is comparatively high, the axial ductility of the
specimens 1 and 10 varies slightly. In addition, they demonstrated H9
1400
lower ductility compared with specimen 9. H5
1200
3.1.6. Strain analysis
Since the strain development for specimens in Group I is Axial load (KN) 1000
similar, a typical axial load versus axial and lateral strain curve H3
800
of specimen 1 is shown in Fig. 9, where V and H symbolize the
H6
axial and corresponding transverse strain, respectively. V and H are H7
600
sequenced in terms of corresponding surfaces. Both the two H8
strains are recorded by strain gauges located on different surfaces 400 H1
at the mid height of the specimen. It can be seen that the whole
section was under pure axial load during the loading process, 200
but divergence occurs among the lateral strains after the elastic
region. The reason may be attributed to the influence of different 0
0 5000 10000 15000
confinement provided by different steel edges of the T-shaped
composite cross section. Also, the different level of local buckling Lateral strain (με)
in corresponding surfaces contributed to the development of Fig. 9. Axial load versus axial and lateral strain curve for specimen1 (a) Axial load
divergence around the failure load. versus axial strain (b) Axial load versus lateral strain.
Taking the interaction between in-filled concrete and steel tube
into account, the lateral-to-axial strain ratio curve for steel tubes
of specimen 1 is depicted in Fig. 10. It can be found that, the 1600
lateral-to-axial strain ratio of the steel tube is around 0.27 before
40% of the ultimate capacity, which corresponds to the Poisson's 1400
ratio of steel coupons. It can be concluded that the Poisson's ratio
of concrete is smaller than that of steel in the initial loading stage 1200
and thus the two materials behave independently. As the load
Axial load (KN)

increases, the lateral-to-axial strain ratio increases. This is because 1000

after the Poisson’s ratio of concrete approached that of steel, the
steel tube began to provide confinement to the in-filled concrete, 800
which resulted in the increase of the Poisson’s ratio of steel. It is surface1
worth noting that the Poisson's ratio of steel plates with large 600 surface3
width-to-thickness ratios increased faster than those of steel
plates with small width-to-thickness ratios. This is because the
surface5
400
steel plate with a large width-to-thickness ratio buckled earlier surface6
than the steel plate with a small width-to-thickness ratio at higher surface8
200
stresses since the concrete confinement provided by the former is surface9
lower than the latter. Consequently, the lateral strains of steel
0
plates with large width-to-thickness ratios increased faster than 0.2 0.4 0.6 0.8 1 1.2
those of the steel plates with small width-to-thickness ratios.
Poisson's ratio
3.2. Behaviour of slender columns under axial compression Fig. 10. Development of lateral-to-axial strain ratio for steel tubes of specimen1.

3.2.1. Test observations unsymmetrical axis x, while flexural buckling and twist about
For a centrally compressed slender MT-CFST column in Group longitudinal axis that passed through the shear centre of the cross
II, flexural buckling occurred when the column bent about the section simultaneously occurred when the column bent about the
64 Y.Q. Tu et al. / Thin-Walled Structures 85 (2014) 57–70
1-x-450 2-x-500
Fig. 11. A view of slender specimens after testing (a) 1.8 m high specimens after testing (b) 1.3 m high specimens after testing (c) Short specimens after testing.
symmetric axis y. Fig. 11(a) shows a view of the 1.8 m long speci-
mens with three section sizes (60
100
80 mm3, 60
120

60 mm3 and 60
120
120 mm3) after testing. It was found that
the failing of the 1.8 m long MT-CFST columns resulted from
overall buckling with obvious later deflections at the mid-height.
For the 1.3 m long MT-CFST columns shown in Fig. 11(b), their
failure was characterised mainly by global buckling, coupled with
obvious local buckling on the compressive region of the column
near the top or bottom at the end of the testing. The reason might
be load eccentricity or the imperfect quality of in-filled concrete.
Moreover, it has been proved by Yu et al. [17] that local buckling
near the end of CFST columns has little effect on the overall
axial load bearing capacity. In reality, as the bearing capacity of a
slender specimen under axial compression was affected by both
residual stress and geometric imperfections (initial crookedness as
well as initial eccentricity), its behaviour was similar to that of a
beam-column and it belonged to the extreme point instability.
As far as the three stub specimens (450 mm, 500 mm and
700 mm) in Group II were concerned, their failure mode was
similar to the specimens in Group I, which was dominated by the
section capacity under axial compression. However, an obvious
6-x-700
bending deformation, which can be seen from Fig. 11(c), exhibited
at the end of the testing. Moreover, there was only a small diffe-
rence in the bearing capacity between stub specimens in Group II
and corresponding ones in Group I, as shown in Tables 1 and 2.
Compared with the ultimate load capacity of short MT-CFST
columns subjected to axial loads, a significant reduction in the
ultimate load capacity was observed for all slender specimens.
3.2.2. Strain analysis
Two typical axial load versus axial and lateral strain curves in
Group II are shown in Fig. 12, where the strains are named in the
same way as the specimens in Group I. Increasing divergence
occurs after the elastic region for specimen 1-x-450, mainly due
to the influence of slight bending of the column and different
confinement of the concrete offered by different steel edges of the
composite cross section as well. It also can be found that the whole
section was under compression during the loading process. Owing
to the influence of initial deflection, the divergence among the
measured strains for specimen 2-y-1800 takes place earlier than
that for specimen 1-x-450. Furthermore, obvious divergence was
 Y.Q. Tu et al. / Thin-Walled Structures 85 (2014) 57–70 65

Fig. 12. Axial load versus axial and lateral strain curves.

observed in the inelastic region due to the bending of the slender torsional buckling of MT-CFST columns.
composite column after elastic region. It should be noticed that sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 ffi
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
some strains developed towards the opposite direction after the 1 2
λyw ¼ λy þ λ2w þ ðλ2y þ λ2w Þ2  4λ2y λ2w ð1  y20 =i20 Þ ð6Þ
failure load which indicated that part of the cross section reversed 2
from compression to tension followed by global buckling.
where y0 is the distance from the centroid to the shear centre of
the cross section, and the parameters i0 andλw are defined in
3.2.3. Effect of slenderness ratio Eqs. (7) and (8) respectively.
To investigate the effect of the slenderness ratio on the flexural
buckling of MT-CFST columns, the nominal flexural slenderness i20 ¼ y20 þ i2x þ i2y ð7Þ
ratio about the unsymmetrical x–x axis ( λx ) and about the sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
symmetric y–y axis ( λy ) are defined as follows:  ffi
It
λw ¼ Asc i0 =
2
ð8Þ
λx ¼ Le =ix ð4Þ 2π 2 ð1 þ μsc Þ

where It is moment of inertia of torsion; the composite Poisson's

λy ¼ Le =iy ð5Þ
ratio μsc is defined by Eq. (8)
where Le is the effective length of a MT-CFST column; ix and iy are μc Ac þ μs As
the radius of total cross-sectional gyration μsc ¼ ð9Þ
pffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiabout
ffi x–x axis and y–y
Asc
axis, respectively; ix ¼ I x =Asc , iy ¼ I y =Asc , Asc is the total area
of composite cross section, Ix and Iy are the second moment of where μc and μs are the Poisson's ratio of concrete and steel,
total cross-sectional area about the x–x axis and the y–y axis, respectively; Ac and As are the cross-sectional area of concrete and
respectively. steel, respectively. The values of λx , λy and λyw are all listed in
Following the formula proposed in GB 50017-2003[18], the Table 2. It can be found that the values of λyw are very close to
nominal flexural-torsional slenderness ratio about the symmetric those of λy. This indicates that the effect of torsional buckling on
y–y axis (λyw) is defined in Eq. (5) for investigating the flexural- MT-CFST columns is very slight.
66 Y.Q. Tu et al. / Thin-Walled Structures 85 (2014) 57–70

Fig. 13 shows the influence of the physical length of the column

on the member capacity about the x–x and the y–y axis, respec- 1400
λx =31.2
tively. As expected, the higher the physical length, the smaller the
member capacity. In addition, it can be seen that the bearing 1200
capacity rises along with the increase in the composite cross-
sectional area under the condition of similar physical length. 1000

Axial load (KN)

A reason is that the effect of delaying local and global buckling
λyw =50.6
by the in-filled concrete is enforced as cross-section area increases. 800
Also, the increase in steel's cross-section area can raise the λx =42.3 λyw =37.5
member capacity. It should be pointed out that, confinement 600
effects for slender columns are slight since the lateral deflection
resulting from the initial crookedness and initial eccentricity 400 1-X-1300
amplifies the bending moment. According to the previous research 1-Y-1300
reported by Schneider [2], the in-filled concrete and the steel tube 200 1-X-1800
may act independently for very slender columns. 1-Y-1800
Fig. 14 shows the axial load (N) versus mid-height lateral 0
deflection (Um) curves obtained from the test. It can be found 0 5 10 15 20 25 30
that, the typical load–deflection curve consists of an ascending Mid height deflection (mm)
branch and a descending branch. Compared with the columns
with high nominal slenderness ratio, no obvious lateral deflections
occurred at the mid-height of the columns with low nominal 1400
λx =26.6
slenderness ratio before the axial load reached the ultimate load. λx =36.1
This is attributed to the fact that geometric imperfections, like 1200
initial crookedness, have a more obvious effect on columns with
high nominal slenderness ratios than those with low nominal 1000
Axial load (KN)
800 λyw =69.9
λyw =51.9
2500
600
2-X-1300
400
2000 2-Y-1300
200 2-X-1800
Axial load (KN)

1500 2-Y-1800
0
0 10 20 30 40
1000 Mid height deflection (mm)

cross section: 60×100×80

2000
500 cross section: 60×120×60 λyw =25.1
cross section: 60×120×120 λx =28.6
0 1500
500 1000 1500 2000
Axial load (KN)

Physical length of the column (mm)

1000 λyw =33.5

2500 λx =38.8

6-X-1300
2000
500 6-Y-1300
6-X-1800
Axial load (KN)

6-Y-1800
1500 0
0 5 10 15 20 25
Mid height deflection (mm)
1000
Fig. 14. Axial load versus mid-height lateral deflection curves (a) Cross section
1 — 60
100
80 mm3 (b) Cross section 2 — 60
120
60 mm3 (c) Cross section
cross section: 60×100×80 6–60
120
120 mm3.
500
cross section: 60×120×60
slenderness ratios. In general, the ultimate axial load decreases as
cross section: 60×120×120
the nominal slenderness ratio increases. It should be noted that
0
1000 1200 1400 1600 1800 2000 the N–Um curves of some 1.3 m long columns tend to drop more
quickly than those of 1.8 m long columns with the same cross
Physical length of the column (mm)
section after the ultimate load. This is because the failing of the
Fig. 13. Physical length versus axial load (a) About x–x axis (b) About y–y axis. 1.3 m long columns mainly resulted from local buckling whereas
 Y.Q. Tu et al. / Thin-Walled Structures 85 (2014) 57–70 67

overall buckling resulted in the final failing of 1.8 m long columns.
Besides, the flexural buckling loads about the unsymmetrical x–x
axis are larger than the flexural-torsional buckling loads about the
symmetric y–y axis for the cross section 60
100
80 mm3 and
60
120
60 mm3, due to a fact that the nominal slenderness
ratios of the former are less than those of the latter.
4. Comparison between tests and design codes
Although several current design codes such as Eurocode 4
(EC4) [19], AISC specification [20], Australian standard AS5100.6
[21], Chinese code CECS159 [22], and Hong Kong steel code (HKS)
[23] are available for the design of conventional CFST columns,
none of them covers MT-CFST columns. Based on the design rules
provided in these codes for rectangular CFST columns, compar-
isons will be made between the experimental results presented in
where Es is the elastic modulus of steel; Ecm is the secant modulus
of elasticity for concrete given in Eurocode 2 [25]; I s and I c are the
second moments of area of steel tube section and concrete section,
respectively.
4.1.2. AISC specification
The proposed composite section in this paper belongs to the
compact section in AISC specification. The design compressive
strength of doubly symmetric axially loaded composite members
shall be determined for the limit state of flexural buckling based
on member slenderness as follows:
P u ¼ ϕc P n ; ϕc ¼ 0:75 ð15Þ
(1) when P n0 =P e r 2:25
h P n0 i

the previous section and the design codes shown in this section. In P n ¼ P n0 0:658 Pe ð16Þ
order to check the feasibility of these design codes, all limitations
are ignored. For an accurate comparison between the test results
(2) when P n0 =P e 4 2:25
and design codes, material factors and partial safety factors are
applied in the calculation of design values. P n ¼ 0:877P e ð17Þ
where
4.1. Brief review of current design codes P n0 ¼ As f y þ 0:85Ac f c' ; P e ¼ π 2 ðEIÞef f =L2e ;-
 
ðEIÞef f ¼ Es I s þ C 3 Ec I c ;C 3 ¼ 0:6 þ 2 As Aþs Ac r0:9is the coeffi-
The section capacity of a CFST stub column is generally
determined by summing up the strengths of the steel tube and cient for calculation of effective rigidity of filled
the in-filled concrete although it is defined slightly differently in composite compression member.
the current design codes due to the non-unified material factors
and partial safety factors. It should be noted that none of the 4.1.3. Australian standard AS5100.6
design codes takes into account the enhancement of the compres- The predicted ultimate section capacity is
sive strength of the in-filled concrete. In general, all codes predict Nps ¼ ϕs As f y þ ϕc Ac f c
0
ð18Þ
the member capacities of slender CFST columns in a similar
approach to steel columns. Since the scope of these codes is where ϕs ¼ 0:9 is the capacity reduction factor for structural steel
limited to doubly symmetrical composite cross section along the in compression and ϕc ¼ 0:6 the capacity reduction factor for
member length, only flexural buckling is considered in this section concrete in compression, respectively.
and the effect of torsional buckling is ignored. In addition, EC4 and The predicted ultimate member capacity is
HKS adopt a similar method for predicting the ultimate member N pl ¼ χ N ps r N ps
capacity of a CFST column. qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
χ ¼ ξ ½1  1 ð90=ξλÞ2  ð19Þ
4.1.1. Eurocode 4
ðλ=90Þ
2
þ1þη
The predicted ultimate section capacity of a MT-CFST stub where ξ ¼ ðλ=90Þ2
; λ ¼ λη þ αa αb ; λη ¼ 90λr is the modified
column is member slenderness; αa and αb are the coefficients defined in
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
0
N ps ¼ As f y =γ s þ Ac f c =γ c ð10Þ AS5100.6; η ¼ 0:00326ðλ  13:5Þ Z 0; λr ¼ N u =Ncr is the relative
slenderness; Ncr ¼ π 2 ðEIÞef f =L2e ; ðEIÞef f ¼ ϕs Es I s þ ϕc Ec I c ; ϕs ¼ 0:9;
where γ s ¼1.1 is the partial safety factor for structural steel and
ϕc ¼ 0:6.
γ c ¼ 1:5 the partial safety factor for concrete, respectively.
The predicted ultimate member capacity of a slender MT-CFST
column is 4.1.4. Chinese code CECS159
CECS159 uses an equation similar to Eq. (10) to predict the
N pl ¼ χ N ps r N ps ð11Þ ultimate section capacity of a CFST stub column, where the partial
safety factor for concrete γ c is 1.4 and the cylinder compressive
1 0
χ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð12Þ strength f c is replaced by prism compressive strength fck.
2
ϕ þ ϕ2  λ The predicted ultimate member capacity is
N pl ¼ χ N ps r N ps
where χ is the reduction factor; λ is the non-dimensional slender-
2
ness; ϕ ¼ 0:5½1 þ αðλ  0:2Þ þ λ ; α is the imperfection factor for
2
χ ¼ 1  0:65 λ for λ r 0:215 ð20Þ
buckling curve given in Eurocode 3 [24]. The non-dimensional
slenderness λ for the bending plane is expressed as: 1 h 2
 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2 2
i
χ¼ 0:965 þ0:3 λ þ λ  ð0:965 þ 0:3 λ þ λ Þ2  4λ for λ 4 0:215
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2λ
2

λ ¼ Nu =N cr ð13Þ
ð21Þ
where N u is the nominal section capacity given by Eq. (2); where the non-dimensional slenderness λ is defined as:
N cr ¼ π 2 ðEIÞef f =L2e is the elastic critical normal force. The effective sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
flexural stiffness (EI)eff is determined in Eq. (14). le f y As þf ck Ac
λ¼ ð22Þ
ðEIÞef f ¼ Es I s þ 0:6Ecm I c ð14Þ π Es I s þ Ec I c
68 Y.Q. Tu et al. / Thin-Walled Structures 85 (2014) 57–70

4.1.5. Hong Kong code are from HKS (μ ¼0.795, σ ¼ 0.017). The predicted ultimate section
The predicted ultimate section capacity is capacities have slight differences between EC4 and HKS. Because
concrete cube compressive strength is used in HKS, it predicts
N ps ¼ As f y þ 0:53Ac f cu ð23Þ
higher strength than EC4 and gives better predictions. It seems
where f cu is the cube compressive strength of concrete. that a higher concrete compressive strength than the cylinder
The HKS approach is quite similar to the EC4 approach in compressive strength in the design formula is closer to the actual
predicting the member capacity of a CFST column, except that it condition. Therefore, further research should be done in the future
0
replaces 1.0 by 0.8 and f c by f cu in calculating the nominal section to develop an accurate design approach for MT-CFST stub columns.
capacity defined by Eq. (2).

4.3. Member capacity of slender columns

4.2. Resistance of cross section under axial compression
Fig. 16 compares the predicted ultimate member capacities
Comparison between the measured ultimate loads (Nus) of
with the test results of slender MT-CFST columns under axial
these 13 short MT-CFST columns and predicted loads using design
compression. The predicted values are summarised in Table 5. All
codes are shown in Fig. 15, in which NEC4, NAISC, NAS, NHK, and NCECS
codes give more reasonable predictions for the ultimate member
are with EC4, AISC, AS5100.6, CECS159, and HK approach, respec-
capacities than for the ultimate section capacities, although the
tively. Their values are listed in Table 4. Since material factors and
predictions for member capacities are still conservative. There is
partial safety factors are applied in the calculation of design values,
no significant difference among all the predicted member capa-
and the enhancement of the compressive strength of the in-filled
cities, where the prediction differences in μ for slender CFST
concrete is ignored, all codes are conservative in predicting the
columns are within 7%. It is clear that CECS159 gives the closest
ultimate section capacities. The AISC specification gives the most
predictions (μ ¼0.841, σ ¼0.042) whereas AS5100.6 gives the
conservative predictions with the average value (μ) of 0.774 and
maximum discrepancy 15.4%. In terms of AISC, there is a signifi-
the standard deviation (σ) of 0.02. This is because a lower
cant improvement in predicting the accuracy of member capacity
resistance factor ( ϕc ¼ 0:75) is applied in the calculation of design
than section capacity. Compared with AISC and AS5100.6, EC4 and
strength. It is clear that AS5100.6 and CECS159 are less conserva-
HKS give less conservative predictions for both short and slender
tive, where the AS51006 underestimates the ultimate loads by
MT-CFST columns. Although CECS159 gives the best predictions for
24.5–31%. The closest predictions for the MT-CFST stub columns
member capacity, it gives unsatisfactory predictions for section

EC4 2000
2000
EC4
AISC 1800
Predicted ultimate loads (kN)

AISC
Predicted ultimate loads (kN)

1800 AS5100.6
1600 AS5100.6
CECS159
1600 CECS159
HKS 1400 HKS
1400
1200

1200
1000

1000 800

800 600
1000 1500 2000 1000 1500 2000
Measured ultimate loads Nus (kN) Measured ultimate loads N ul (kN)
Fig. 15. Predicted ultimate section capacities versus test results. Fig. 16. Predicted ultimate member capacities versus test results.

Table 4
Comparison between predicted ultimate section capacities and experimental results.

Specimen N us (kN) N EC4 (kN) N EC4 /N us N AISC (kN) N AISC /N us N AS (kN) N AS /N us N CECS (kN) N CECS /N us N HKS (kN) N HKS /N us

1 1550.5 1206.0 0.778 1067.7 0.689 1142.1 0.737 1162.0 0.749 1242.7 0.801
2 1494.3 1129.4 0.756 994.0 0.665 1070.4 0.716 1088.9 0.729 1165.2 0.780
3 1961.0 1512.7 0.771 1345.1 0.686 1428.8 0.729 1454.4 0.742 1552.6 0.792
4 1466.7 1129.4 0.770 998.6 0.681 1070.4 0.730 1088.9 0.742 1165.2 0.794
5 1661.3 1282.7 0.772 1135.9 0.684 1213.8 0.731 1235.1 0.743 1320.2 0.795
6 2033.0 1589.4 0.782 1414.1 0.696 1500.5 0.738 1527.5 0.751 1630.1 0.802
7 1281.0 1013.4 0.791 919.2 0.718 954.1 0.745 971.7 0.759 1035.1 0.808
8 1849.3 1334.3 0.722 1153.7 0.624 1276.3 0.690 1296.4 0.701 1395.4 0.755
9 1314.5 1033.1 0.786 902.9 0.687 986.4 0.750 1002.3 0.762 1038.3 0.790
10 1628.6 1265.4 0.777 1124.1 0.690 1195.6 0.734 1216.9 0.747 1285.2 0.789
1-x-450 1564.8 1206.0 0.771 1067.7 0.682 1142.1 0.730 1162.0 0.743 1242.7 0.794
2-x-500 1418.6 1129.4 0.796 994.0 0.701 1070.4 0.755 1088.9 0.768 1165.2 0.821
6-x-700 1994.7 1589.4 0.797 1414.1 0.709 1500.5 0.752 1527.5 0.766 1630.1 0.817
 Y.Q. Tu et al. / Thin-Walled Structures 85 (2014) 57–70 69

Table 5
Comparison between predicted ultimate member capacities and experimental results.

Specimen N ul (kN) N EC4 (kN) N EC4 /N ul N AISC (kN) N AISC /N ul N AS (kN) N AS /N ul N CECS (kN) N CECS /N ul N HKS (kN) N HKS /N ul

1-2-x 1349.3 1077.6 0.799 1013.9 0.751 1008.1 0.747 1083.9 0.803 1113.4 0.825
1-3-y 1276.7 1035.9 0.811 994.5 0.779 972.5 0.762 1062.2 0.832 1071.1 0.839
1-4-x 1207.5 972.8 0.806 964.1 0.798 917.9 0.760 1031.2 0.854 1007.1 0.834
1-5-y 1198.3 909.2 0.759 929.1 0.775 861.7 0.719 995.8 0.831 942.5 0.787
2-2-x 1261.1 1043.5 0.827 1000.2 0.793 974.0 0.772 1030.0 0.817 1079.0 0.856
2-3-y 1209.2 873.2 0.722 911.3 0.754 827.1 0.684 936.5 0.774 905.8 0.749
2-4-x 1146.5 962.6 0.840 963.1 0.840 905.1 0.789 988.8 0.862 996.7 0.869
2-5-y 963.8 706.0 0.733 805.6 0.836 675.2 0.701 829.8 0.861 735.5 0.763
6-2-x 1733.6 1446.0 0.834 1353.5 0.781 1346.3 0.777 1439.4 0.830 1487.0 0.858
6-3-y 1818.8 1490.6 0.820 1370.6 0.754 1383.8 0.761 1460.7 0.803 1532.0 0.842
6-4-x 1487.8 1321.1 0.888 1296.5 0.871 1239.7 0.833 1378.5 0.927 1360.7 0.915
6-5-y 1576.1 1387.0 0.880 1328.1 0.843 1296.2 0.822 1411.1 0.895 1427.4 0.906

1 5. Conclusions

A series of tests on short and slender MT-CFST columns under

0.8 axial compression have been described and discussed in the
present paper. Comparisons have also been made between pre-
dicted results using existing design methods and experimental
Reduction factor

0.6 results. From the results of this study, the following conclusions
can be drawn within the scope of this study

0.4
0.2
Reduction factor about the x-x axis
Reduction factor about the y-y axis
0
0 0.2 0.4 0.6 0.8
Non-dimensional slenderness
Fig. 17. Comparison of measured reduction factors about the x–x and y–y axes.
capacity. Since none of the codes gives accurate prediction on the
member capacities of MT-CFST slender columns proposed in this
paper, further research should be done in the future to present a
suitable design method for slender MT-CFST columns.
For the purpose of practical design, the measured reduction
factor χm of a slender MT-CFST column is defined as
χm ¼ Nul =Nus
where Nul and Nus are the measured member capacity of a slender
column and section capacity of the corresponding short column,
respectively. The value of χm is listed in Table 2.
Fig. 17 compares measured reduction factors about the unsym-
metrical x–x axis (χ mx ) and about the symmetric y–y axis (χ my ).
The non-dimensional slenderness λ in the figure is computed by
Eq. (13). It should be pointed out that different ways of calculating
the non-dimensional slenderness λ in the design codes have no
effect on the comparison of measured reduction factors χ mx and
χ my . Although torsional buckling is not considered in calculating
the non-dimensional slenderness λ for the bending plane, some
measured reduction factors about the y–y axis are located rela-
tively higher than the reduction factors about the x–x axis. This
indicates that the effect of torsional buckling is very slight and can
be ignored for MT-CFST columns. Furthermore, this is consistent
with the fact that the values of nominal flexural slenderness λy are
very close to those of nominal flexural-torsional slenderness λyw
listed in Table 2. Therefore, it is appropriate to calculate the
reduction factors about the x–x and y–y axes using the same
buckling curve in practical designs.
(1) Compared with the plastic resistance under axial compression,
the bearing capacity of short MT-CFST columns were
enhanced, and all specimens exhibited good ductility.
(2) Increasing c/a or b/a ratio increases the resistance of the cross
section under axial compression but reduces the axial ductility
of MT-CFST columns. The ultimate load of MT-CFST columns
increases along with a rise in the thickness of the steel tube.
1 Also, a MT-CFST column consisting of thick steel tubes is hard
to buckle under axial compression.
(3) Increasing the concrete compressive strength increases the fail-
ure load of MT-CFST columns but generally decreases their axial
ductility. In addition, an increase in the concrete compressive
strength has a slight effect on the variation of axial ductility since
the concrete compressive strength is relatively high.
(4) In general, for slender MT-CFST columns, the failure load decreases
along with an increase of slenderness ratio. The reduction factors
about the x–x and y–y axes for MT-CFST columns can be calculated
using the same buckling curve in practical designs, and the effect
of torsional buckling can be ignored.
ð24Þ
(5) All the design codes used in this paper are conservative in
predicting the bearing capacity of both short and slender MT-
CFST columns, where HKS gives the closest predictions for
short MT-CFST columns and CECS159 gives the best predic-
tions for slender MT-CFST columns. Compared with AISC and
AS5100.6, EC4 and HKS give less conservative predictions for
both short and slender MT-CFST columns.
Acknowledgements
This programme was sponsored by the National Natural
Science Foundation of China under Grant no. 51278019.
References
[1] Kloppel VK, Goder W. An investigation of the load carrying capacity of
concrete-filled steel tubes and development of design formula. Der Stahlbau
1957;26(2):44–50.
[2] Schneider SP. Axially loaded concrete-filled steel tubes. J Struct Eng (ASCE)
1998;124(10):1125–38.
70 Y.Q. Tu et al. / Thin-Walled Structures 85 (2014) 57–70
[3] Han LH. Tests on stub columns of concrete-filled RHS sections. J Constr Steel
Res 2002;58(3):353–72.
[4] Ellobody E, Young B, Lam D. Behaviour of normal and high strength concrete-
filled compact steel tube circular stub columns. J Constr Steel Res 2006;62
(7):706–15.
[5] Liu DL, Gho WM. Axial load behaviour of high-strength rectangular concrete-
filled steel tubular stub columns. Thin-Walled Struct 2005;43(8):1131–42.
[6] Uy B, Tao Z, Han LH. Behaviour of short and slender concrete-filled stainless
steel tubular columns. J Constr Steel Res 2011;67(3):360–78.
[7] Mursi M, Uy B. Strength of slender concrete filled high strength steel box
columns. J Constr Steel Res 2004;60(12):1825–48.
[8] An YF, Han LH, Zhao XL. Behaviour and design calculations on very slender
thin-walled CFST columns. Thin-Walled Struct 2012;53:161–75.
[9] Hernandez-Figueirido D, Romero ML, Bonet JL, Montalva JM. Influence of
slenderness on high strength rectangular concrete-filled tubular columns with
axial load and non-constant bending moment. J Struct Eng ASCE 2012;138
(12):1436–45.
[10] Han LH, Zheng LQ, He SH, Tao Z. Tests on curved concrete filled steel tubular
members subjected to axial compression. J Constr Steel Res 2011;67(6):
965–76.
[11] Dai XH, Lam D, Jamaluddin N, Ye J. Numerical analysis of slender elliptical
concrete filled columns under axial compression. Thin-Walled Struct 2014;77:
26–35.
[12] Lu XL, Li XP, Wang D. Modelling and experimental verification on concrete-
filled steel tubular columns with L or T section. Front Struct Civ Eng 2007;1
(2):163–9.
[13] Du GF, Xu LH, Wen F. Experimental study on normal section compression
bearing capacity of composite T-shaped concrete-filled steel tubular columns.
China Civ Eng J 2009;42(6):14–20 (in Chinese).
[14] Yang YL, Yang H, Zhang SM. Compressive behaviour of T-shaped concrete filled
steel tubular columns. Int J Steel Struct 2010;10(4):419–30.
[15] Tao Z, Han LH, Wang ZB. Experimental behaviour of stiffened concrete-filled
thin-walled hollow steel structural (HSS) stub columns. J Constr Steel Res
2005;61(7):962–83.
[16] Teng ZM, Zhu JS. Reinforced concrete and masonry structure. Beijing: China
Architecture and Building Press; 1992 (in Chinese).
[17] Yu M, Zha XX, Ye JQ, Li YT. A unified formulation for circle and polygon
concrete-filled steel tube columns under axial compression. Eng Struct
2013;49:1–10.
[18] GB50017. Code for design of steel structures. (in Chinese). Beijing: China
Planning Press; 2003.
[19] Eurocode 4. Design of composite steel and concrete structures, part 1.1:
general rules and rules for buildings. London: British Standards Institution;
2004.
[20] AISC 360–10. Specification for structural steel buildings. Chicago, Illinois:
American Institute of Steel Construction; 2010.
[21] Standards Australia. Bridge design, part 6: steel and composite construction.
AS 5100.6-2004. Sydney; 2004.
[22] CECS 159. Technical specification for structures with concrete-filled rectan-
gular steel tube members. Beijing: China Planning Press; 2004 (in Chinese).
[23] Code of Practice for the Structural Use of Steel 2011. Hong Kong: Buildings
Department; 2011.
[24] Eurocode 3. Design of steel structures, Part1-1: General rules and rules for
buildings. London: British Standards Institution; 2003.
[25] Eurocode 2. Design of concrete structures, Part 1: General rules and rules for
buildings. London: British Standards Institution; 2001.