Thin-Walled Structures 107 (2016) 502–513

Contents lists available at ScienceDirect

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

Full length article

Composite action of hexagonal concrete-filled steel tubular stub

columns under axial loading
Fa-xing Ding a, Zhe Li a, Shanshan Cheng b,n, Zhi-wu Yu c
a
School of Civil Engineering, Central South University, Changsha 410075, PR China
b
Department of Civil and Structural Engineering, University of Sheffield, United Kingdom
c
National Engineering Laboratory for High Speed Railway Construction, Changsha 410075, PR China

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

Article history: Four groups of axial compression tests on hexagonal CFT stub columns have been carried out aiming to
Received 20 April 2016 investigate the effects of the concrete strength and steel ratio on the behaviour of hexagonal CFT stub
Accepted 7 July 2016 columns. Studies on parametric analysis and composite action between core concrete and steel tube have
Available online 25 August 2016
been carried out using FE modelling which had been benchmarked using the test data. Based on the
Keywords: essential data obtained in this paper, the ratio of axial stress-yield strength of steel tube was determined
Hexagon at the ultimate state. The stress contour of core concrete was simplified to an unconfined area without
Concrete-filled steel tubular stub columns constraint and a confined area with uniform constraint imposed by hexagonal steel tube. Eventually, a
Ultimate bearing capacity practical design equation of the ultimate bearing capacity of hexagonal CFT stub columns was proposed
Strain ratio
based on the superposition principle. An excellent agreement between the proposed equation and the
Confinement effect
experimental results was observed, with an average ratio of predicted to measured capacity of 1.08 and a
standard deviation of 0.05.
& 2016 Elsevier Ltd. All rights reserved.

1. Introduction rather few in literature, although hexagonal cross-sections are

A concrete-filled steel tube (CFT) column is a composite
member formed by a steel tube filled with concrete. The concrete
core adds stiffness to the steel tube and prevents the occurrence of
inward local buckling, while the steel tube confines the concrete
core and acts as longitudinal and lateral reinforcement. Due to the
benefit of composite action of the two materials, the CFT columns
provide excellent structural properties such as high strength, high
ductility and large energy absorption capacity. Since 1970s, CFT
has been widely used in high-rise, long-span structures [1,2], not
only due to the favourable structural properties, but also the rapid
construction without removing any formwork.
Extensive research on effects of cross-sectional profiles of CFT
columns has been conducted in literature. Experimental and nu-
merical studies of circular [3–11], elliptical [12–14], octagonal [15],
and square [16–20] CFT columns have been carried out in studying
the axial load bearing capacity of CFT columns. A unified formula
for CFT columns circular and polygonal cross-sectional profiles
subjected to axial compression has been obtained by Yu et al. [21].
However, available studies on hexagonal CFT stub columns are
n
Corresponding author.
E-mail address: shanshan.cheng@sheffield.ac.uk (S. Cheng).
commonly demanded in architectural design. Ketema and Taye
[23] studied the moment-axial load interaction for hexagonal and
octagonal CFT columns subjected to uniaxial bending, and a uni-
fied approach has been presented for designing purpose. Circular,
hexagonal, rectangular and square cross-sections have been tested
by Evirgen et al. [24] recently with the focus on the effects of
width-thickness ratio (b/t), the compressive strength of concrete
and geometrical shape of cross section parameters on ultimate
loads, axial stress, ductility and buckling behaviour.
The effects of the concrete strength and steel ratios on the
mechanical behaviour of hexagonal CFT stub columns were in-
vestigated in this paper. Eight axial compression specimens of
hexagonal CFT stub columns have been conducted at the Central
South University in China. The nonlinear finite element analysis
of hexagonal CFT stub columns throughout loading history has
been carried out using ABAQUS. Based on the essential experi-
mental and numerical data, practical formulas for the ultimate
bearing capacity of hexagonal CFT stub columns have been pro-
posed by using superposition principle at the ultimate state,
which follows the same research idea as Ding et al. [22]. Details
of the studies have been explained in the following sections of
this paper.

http://dx.doi.org/10.1016/j.tws.2016.07.005
0263-8231/& 2016 Elsevier Ltd. All rights reserved.
 F.-x. Ding et al. / Thin-Walled Structures 107 (2016) 502–513 503

Nomenclature Nu, Exp Ultimate bearing capacity of hexagonal CFT stub col-
umns from experimental results
Ac Cross-sectional area of core concrete Nu, FE Ultimate bearing capacity of hexagonal CFT stub col-
Ac1 Non-constrained area of core concrete umns from FE results
Ac2 Constrained area of core concrete p Lateral pressure coefficient
As Area of steel tube t Wall thickness of steel tube
Asc Total area of cross-section s Axial stress of concrete
B Edge length of the hexagonal section si Equivalent stress of steel tube
b Edge length of core concrete sL,c Axial compressive stress of core concrete
DI Ductility index sL,s Axial compressive stress of steel tube
Es Elastic modulus of steel tube sr,c Radial concrete stress of the confined area
fb0 Initial equibiaxial compressive yield stress of concrete sθ,s Tensile transverse stress of steel tube
fc Uniaxial compressive strength of concrete ε0.75 Axial strain when the load attains of 75% the ultimate
fc0 Initial uniaxial compressive yield stress of concrete load in the pre-peak stage
fcu Compressive cubic strength of concrete ε0.85 Strain when experimental bearing capacity is de-
fsc Ultimate strength of CFT column creased to 85% of ultimate value
fu Ultimate strength of steel tube ε Axial strain of concrete
fy Yield strength of steel tube εc Strain corresponding with the peak compressive stress
k Ratio of initial tangent modulus to secant modulus at of concrete
peak stress εL Axial strain of columns
L Height of specimens εi Equivalent strain of steel tube
N Axial load εy Yield strain of steel tube
Nu Axial ultimate bearing capacity εst Hardening strain of steel tube
Nu,c Ultimate bearing capacity of hexagonal CFT stub col- εu Ultimate strain of steel tube
umns from calculated results vsc Strain ratio of steel tube
Nu, Eq.14 Ultimate bearing capacity of hexagonal CFT stub col- θ Dilation angle of concrete
umns from Eq. (14). ρ Steel ratio of columns

2. Experimental investigation Table 1

Properties of tested specimens.

2.1. Test set-up Specimens B
t L fcu fy Nu, Steel ratio DI

Exp
number (ρ)
In total 8 specimens of hexagonal CFT were designed in this
study. The nominal dimension of each specimen is 200 (B) mm
4 mm
mm mm MPa MPa kN – –

(6) (t) mm
1200 (L) mm, where B is the outer edge length of the HST1-A 196
3.73 1200 39.3 311 4947 0.044 3.781
hexagonal section, t is the wall thickness of the steel tube and L is HST1-B 198
3.71 4618 0.043 3.288
the height of the specimen. Detailed cross-sectional dimensions HST2-A 196
5.78 321 6001 0.067 7.174
and material properties are shown in Fig. 1 and in Table 1 re- HST2-B 198
5.96 6041 0.068 7.569
HST3-A 197
3.72 57.4 311 6827 0.043 3.241
spectively. Two identical specimens (namely A and B) were made HST3-B 198
3.76 6803 0.043 3.294
for each group and there were eight specimens in total. HST4-A 199
5.89 321 7079 0.067 5.817
The hexagonal steel tubes were moulded by bending Q235 HST4-B 196
5.81 7289 0.067 6.387
steel plates into grooves and then welding the two edges at the
corner. The position at which the butt welds were made was

Fig. 1. Experimental instrumentation for all specimens.

504 F.-x. Ding et al. / Thin-Walled Structures 107 (2016) 502–513
illustrated in Fig. 5(a). Butt welds were used according to the
standard GB/50017-2003 [25] and it was also necessary to ensure
that the butt welds’ ends were smooth and the dimensions of the
steel tube are in accordance with experimental design.
For better observation of deformation and preventing steel
tubes from rusting, red paint was sprayed on the outer surface of
the steel tubes and grids of 50 mm
50 mm were drawn on the
painted surface. First the hexagonal steel tube was welded with
the bottom loading plate together. Being vibrated with a vibrator
at the same time, concrete was poured from the top of the spe-
cimens. Before initial set of concrete, the top surface of concrete
was smoothed for being at the same level with steel tube. Concrete
splashed down the outer surface of steel tube was cleaned in time
during the process of concrete casting. After pouring, the speci-
mens were covered with plastic film on top surfaces to prevent
water loss. They have been cured at atmosphere condition and
watered regularly. Meanwhile, the standard concrete cubes were
prepared and cured under the same condition as the concrete used
in hexagonal CFT column specimens. After three months of curing,
the concrete surface of the column specimens was polished with
grinder, and then daubed a layer of epoxy resin binder. Finally,
steel cover plates were bonded at the top end of the hexagonal CFT
column specimens to ensure that the steel tube and core concrete
share loads from the initial loading stage.
2.2. Experimental arrangement and measurements
Before testing the axial load capacity of the hexagonal CFT, the
mechanical properties of concrete cubes and steel were tested
respectively according to the corresponding standard methods.
150 mm
150 mm
150 mm concrete cubes were tested under
compression according to the standard GB/T50081-2002 [26] for
investigating the cubic strength fcu, which have been summarized
in Table 1. Q235 steel plates with thicknesses of 4 mm and 6 mm
respectively were used to carry out tensile coupon tests by using
standard GB/T228–2002 [27]. The results on material properties of
the two types of steel plates were shown in Table 2.
Compressive experiments on stub column specimens were
conducted using a 2000-ton tri-axial stress testing machine in the
Civil Engineering Safety Science Laboratory of Central South Uni-
versity. To accurately measure the deformation, four strain rosettes
(1# to 4#) were attached at the mid-height of two opposite side
surfaces and two LVDTs (1# and 2#) were installed at the same
height of another two opposite side surfaces, as shown in Fig. 1.
Load-strain curves were acquired by a DH3818 static strain
measurement system and load-deformation curves were acquired
from electronic transducers and RX-24A data acquisition system.
The compressive load was applied from the top of the specimens
using a 2000-ton axially static stress testing system control mode.
The load was applied through increments of force at the step of 1/
20 of the ultimate load in the elastic stage, while the application of
the load was shifted into increments of displacement at the step of
0.2 mm after the load reached over 60% of the ultimate bearing
capacity. When the ultimate load was approached, specimens
were loaded at the step of 0.5 mm and maintained 5 min. The data
was collected until the axial strain reached 0.04 which was the
Table 2
Properties of steel.
Thickness (mm) fy fu Es
MPa MPa MPa
4 311 460 2.09
105
6 321 480 2.02
105
7500
Axial load N (kN)
6000
4500
3000
HST1-A
HST2-B
1500
HST3-A
HST4-A
0
0.00 0.01 0.02 0.03 0.04
Axial strain εL
Fig. 2. Comparison of load-strain curves of specimens.
largest strain of the specimens. Each specimen has taken about
3 h.
2.3. Experimental results and discussion
Fig. 2 shows the load-strain curves of specimens, which reveals
that the compressive process of hexagonal CFT stub columns are
divided into three phases: elastic stage, elastic-plastic stage and
failure stage.
At the initial loading stage, all the specimens were in elasticity,
which has been illustrated by the linear responses of the load-
strain curves and load-displacement ones. The compressive stiff-
ness of the specimens in this stage is larger than that in the other
stages, and the axially elastic displacement is very small.
For all the specimens, when the imposed load reached 60–70%
of the ultimate load, steel tube started to yield and the load-strain
curves demonstrated elastic-plastic behaviour. In this stage, visible
local buckling generally appeared near both ends of the specimens
due to the end effect and then emerged in the middle where local
bucking developed rapidly. When the specimen reached the ulti-
mate loading capacity, an apparent buckling could be seen on the
steel tube.
When the ultimate load was approached, load-bearing capacity
of the specimens decreased rapidly while the displacement in-
creased dramatically because of the crush of core concrete and the
further bucking of steel tube. Fig. 3 shows the failure mode of all
the tested specimens. It is obvious that the load bearing capacity of
specimens HST1 and HST3, which were with smaller ductility,
declined faster than those of specimens HST2 and HST4 (with
bigger ductility).
After the loading test, the steel tubes were cut open in order to
examine the failure performance of the core concrete. It has been
found that the micro-cracks have developed remarkably, and there
were inclined shear rupture zones or even crushes in core con-
crete, as shown in Fig. 4. When shear rupture occurred, a rapid
increase of deformation in the shear rupture zone in the core
concrete was expected, which led to a rise of transverse tensile
strain in the outer steel tube. As the ductility of butt weld was
poorer than steel tube, weld failure along the butt welds was
found in specimens HST3-A and HST2-A (see Fig. 5).
vs
2.4. Bearing capacity

Fig. 6 shows the experimental results of ultimate bearing ca-
0.292
0.256
pacity for all specimens. In comparison with the HST1 columns,
the average ultimate bearing capacity of HST2 specimens is
 F.-x. Ding et al. / Thin-Walled Structures 107 (2016) 502–513 505

(a)HST1-B (b)HST2-A

(c)HST3-B (d)HST4-A
Fig. 3. Typical failure modes of specimens. (a) HST1-B (b) HST2-A (c) HST3-B (d) HST4-A.

improved by 25.9%, with a 55.2% increase of steel ratio. However,
the comparison between HST3 and HST4 reveals that the average
ultimate bearing capacity only improved slightly (5.5%) with the
increase of steel ratio. It is therefore concluded that increasing
steel ratio can yield an improvement of load bearing capacity, al-
though the effect is more significant to HST columns with lower
strength grade of concrete as used in this study.
Comparing the average ultimate bearing capacities of HST1 and
HST3 shows that with the increase of concrete strength, the ulti-
mate bearing capacity of HST3 improved by 42.5%. However, the
comparison between HST2 and HST4 reveals that the average ul-
timate bearing capacity of HST4 specimens is only improved by
19.4% with the increase of concrete strength. It is therefore in-
dicated that an increase of concrete strength can yield an im-
provement of load bearing capacity, and the effect is more sig-
nificant to those with smaller steel ratio as studied in this paper.
From the above analyses, it has been found that increasing
concrete strength is more efficient to increase the ultimate bearing
capacity of hexagonal CFT specimens than by increasing the steel
ratio.
2.5. Ductility
To investigate the effect of concrete strength and steel ratio on
the ductility of specimens, a ductility index (DI), which has been
used in studying the inner constrained square CFT [22], is also
adopted in this paper and the corresponding ductility index is
defined as follows:
ε0.85
DI =
εb (1)
where ε0.85 is the axial strain when the load falls to 85% of the
ultimate load; εb is equal to ε0.75/0.75, and ε0.75 is the axial strain
when the load attains of 75% the ultimate load in the pre-peak
506 F.-x. Ding et al. / Thin-Walled Structures 107 (2016) 502–513

Fig. 4. The failure modes of core concrete.

(a) hexagonal Section (b) butt weld failure

Fig. 5. Butt weld failure of specimens. (a) Hexagonal Section (b) butt weld failure.

8000 8
Ductility index DI

6000
Axial load N (kN)

4000 4
HST3-A
HST2-A

2000
HST1-B

HST2-B

HST3-B

HST4-A
HST1-A

HST4-B

HST3-A
HST2-A

2
HST1-B

HST2-B

HST3-B

HST4-A
HST1-A

HST4-B

0
1 2 3 4 5 6 7 8 0
Fig. 6. Comparison of ultimate bearing capacity for all specimens.
1 2 3 4 5 6 7 8
Fig. 7. Comparison of ductility index DI for all specimens.
 F.-x. Ding et al. / Thin-Walled Structures 107 (2016) 502–513 507

6000 between core concrete and steel tube.

HST1-B At the initial loading stage, the strain ratio was close to Pois-
son's ratio of steel and it increased slowly. The core concrete was
Axial load N (kN)

4500 hardly enhanced by the steel tube. When the load reached about
50% of its ultimate capacity, the strain ratio was larger than the
Poisson's ratio of steel, which indicates an obvious confinement
3000 effect of steel tube on core concrete. At elastic-plastic stage, the
strain ratio revealed a rapid increase, and the confinement effect
endpoint 1# was further enhanced. The analysis result shows that the trans-
1/4 point verse deformation coefficient is more than 0.5 at the later stage of
1500 mid-point 3# the loading because of combined action. Under the same com-
FE 3# pressive load, the largest strain ratio of the hexagonal steel tubes
FE 1# occurred at endpoint 1, and the smallest strain ratio occurred at
0
mid-point 3. It is indicated that the hexagonal steel tube has much
0.0 0.2 0.4 0.6 0.8 1.0 more confinement effect at the corner than that at the centre of
Steel ratio vsc the core concrete.

(a) Specimen HST1-B 3. Finite element (FE) modelling

8000 3.1. FE models

HST2-B
FE models were established using ABAQUS/Standard6.4. In
Axial load N (kN)

6000 these models, 8-node reduced integral format 3D solid elements

4000
2000
0 tube and core concrete is generated by friction and an appropriate
0.2 0.4 0.6
Steel ratio vsc
(b) Specimen HST2-B
Fig. 8. Comparison of load-transverse strain curves. (a) Specimen HST1-B.
(b) Specimen HST2-B.
stage. Table 1 shows the ductility coefficient calculated by Eq. (1)
for all specimens.
Fig. 7 shows the ductility indexes from Eq. (1) for the tested
specimens, where a larger value of DI indicates a slower process of
load reduction after the peak load. When the concrete strength is
increased, the DI value reduces by an average of 7.6% when com-
paring the specimens HST3 with HST1. However, the reduction of
DI value turns to an average of 17.2% when comparing the speci-
mens HST4 with HST2. This result is suggesting that the ductility
of specimens decreases with the increase of concrete strength.
Moreover, when increasing the steel ratio, the DI value increases
109% on average when comparing the specimens HST2 with HST1.
However, the DI value increases 87% on average when comparing
the specimens HST4 with HST3. It can then be concluded that the
ductility of specimens increases with the increase of steel ratio.
2.6. Strain ratio
Fig. 8 shows the relationship of the axial load (N) versus the
strain ratio (vsc) of HST1-B and HST2-B specimens. The strain ratio,
which is defined as the absolute value of the perimeter strain di-
vided by the axial strain of all gauged points, reveals the con-
finement effect of the core concrete offered by steel tube. The
more the strain ratio is, the more the confinement effect is
were used for the hexagonal steel tube, core concrete and loading
plates of all specimens, and the surface of loading plates was de-
fined as rigid. A structured meshing option was adopted, and a
typical mesh was shown in Fig. 9.
The interaction of the normal direction of the two surfaces was
hard contact. A surface-to-surface contact was adopted for the
endpoint 1#
constraint between steel tube and concrete with finite slip for-
1/4 point 2#
mula. In the proposed FE model, the shear stress between steel
mid-point 3#
0.8 1.0 friction coefficient is required. Baltay and Gjelsvik [28] conducted
essential experimental researches in order to determine the fric-
tion coefficient between the steel tube and the core concrete, and
they indicated that this coefficient ranged from 0.3 to 0.6. In this
study, a friction coefficient of 0.5 was adopted, which was iden-
tified with the prior research [29] and fell in the above scope. The
tie constraint was adopted between the concrete and the loading
plate, and also used between the steel tube and the loading plate
in modelling, in order to ensure that the core concrete and the
steel tube can share load together in the whole loading process.
A concrete constitutive model of concrete in CFT columns has
been given in Ref. [30] through modifying the concrete model
under tri-axial compression presented by Ottosen [31]. The stress-
strain relationship is described as:
⎧ kx + (m − 1)x2
⎪ x≤1
⎪ 1 + (k − 2)x + mx2
y=⎨
⎪ x
⎪ α (x − 1)2 + x x>1
⎩ 1 (2)
where y¼s/fc and x ¼ ε/εc are the stress and strain ratios of the
core concrete to the uniaxial compressive concrete respectively. s
and ε are the stress and strain of the core concrete. fc ¼0.4fcu7/6 is
the uniaxial compressive strength of concrete, where fcu is the
compressive cubic strength of concrete. εc is the strain corre-
sponding with the peak compressive stress of concrete, where
εc ¼383fcu7/18
10  6. The parameter k is the ratio of the initial
tangent modulus to the secant modulus at peak stress. m ¼1.6
(k  1)2 is a parameter that controls the decrease in the elastic
modulus along the ascending branch of the axial stress-strain re-
lationship. For a concrete-filled steel tubular stub column, para-
meter α1 is taken as 0.15. More information of the concrete model
could be found in Ref. [30].
508 F.-x. Ding et al. / Thin-Walled Structures 107 (2016) 502–513

Fig. 9. FE models after meshing (a) FE model of half height, (b) cover plate element, (c) steel tube element, (d) core concrete element.

The damage plasticity model provided in ABAQUS has been validated in previous research [9,16,32], and thereby adopted in
verified to be applicable in simulating triaxially compressive this paper:
concrete in CFT columns with circular [9], square [16] and rec-
⎧ Esεi εi ≤ εy
tangular with round ends [32] cross-sections, by using the para- ⎪
meters defined by Ding et al. [30]: the eccentricity is 0.1, the ratio ⎪
⎪ fy εy < εi ≤ εst
of initial equibiaxial compressive yield stress to initial uniaxial σi = ⎨
⎪ fy + ζEs(εi − εst ) εst < εi ≤ εu
compressive yield stress (fb0/fc0) is 1.225, the ratio of the second ⎪
stress invariant on the tensile meridian to that on the compressive ⎪
⎩ fu εi > εu (3)
meridian is 2/3, the viscosity parameter is 0.005, and the dilation
angle (θ) is 40°. This damage plasticity model is thereby adopted in where, si and εi are the equivalent stress and strain of the steel
this paper to study CFT columns with hexagonal cross-sections. tube; fy, fu (¼ 1.5fy) and Es (¼ 2.06
105 MPa) are the yield
Based on a large number of experimental studies on the me- strength, ultimate strength, and elastic modulus of steel respec-
chanical properties of CFT columns [33], an elastic-plastic me- tively; εy, εst and εu are the yield strain, hardening strain, and
chanical model of steel was proposed by Ding et al. [30] using the ultimate strain of steel, which have been defined as εst ¼ 12εy and
Von-Mises yield criteria, and associated with the Prandtl-Reuss εu ¼ εst þ0.5fs/(ζEs)¼ 120εy. The parameter ζ is taken as 1/216.
flow rule and isotropic strain hardening. The model has been Due to the symmetry of cross-sections, only half of a section
 F.-x. Ding et al. / Thin-Walled Structures 107 (2016) 502–513 509

5600 360
HST1 HST1
300

Various stress σ (MPa)

4200
Axial load N (kN)

axial stress
2800
240

180 (0.02,188)
1400 HST1-A
HST1-B 120 endpoint 1#
FE
0 mid-point 3#
0.00 0.01 0.02 0.03 0.04 60 average
Axial strain εL transverse stress
constitutive curve
0
(a) Specimen HST1
0.00 0.01 0.02 0.03
7500
HST2 Various strain ε
6000 Fig. 11. Comparisons of steel stress-strain relationships from calculated results.
Axial load N (kN)

4500 was analysed for the axially-loaded hexagonal CFT stub columns.
To model the descending stage of load-bearing capacity of speci-
3000 mens, displacement control was applied. Both material and geo-
metrical nonlinearities were considered and solved using the in-
1500 HST2-A cremental-interactive method in ABAQUS.
HST2-B
FE
0 3.2. 2. Results and discussion
0.00 0.01 0.02 0.03 0.04
Axial strain εL The load-axial strain curves of hexagonal CFT columns by using
both FE modelling and experimental results have been shown in
(b) Specimen HST2
Fig. 10. It is seen that good agreement between experimental and
8000 FE modelling results was found in the elastic stage. In the elastic-
HST3
plastic stage and failure stage, the steel tube at measured points
had been yielded, which caused deviations between modelling
Axial load N (kN)

6000
and experimental curves.
Fig. 11 shows axial and transverse stress-strain curves of steel
4000 tube for a half of specimens HST1. The axial compressive stress of
steel tube has been decreased and the transverse tensile stress has
been increased, due to strong composition effect between core
2000 HST3-A concrete and steel tube. Meanwhile, the axial compressive stress
HST3-B curve and the transverse tensile stress curve of the steel tube at
FE
0 endpoints for hexagonal CFT column HST1 have intersected in the
0.00 0.01 0.02 0.03 0.04 process of loading, but it is not seen at the mid-point of the steel
Axial strain εL tube. It is illustrated that the confinement effect of hexagonal steel
tube on core concrete at endpoint is greater than that at mid-
(c) Specimen HST3 point. Furthermore, almost 67% of ultimate capacity of steel tube

8000
HST4 60
HST1
Axial load N (kN)

6000
Axial stress σL,c (MPa)

45
4000

2000 HST4-A
30
HST4-B
FE endpoint 1#
0
0.00 0.01 0.02 0.03 0.04 15 mid-point 3#
Axial strain εL average
plain concrete
(d) Specimen HST4
0
Fig. 10. Comparison of load-axial strain curves of specimens between FE modelling 0.000 0.005 0.010 0.015 0.020
and experimental results. (a) Specimen HST1 (b) Specimen HST2 (c) Specimen
HST3 (d) Specimen HST4. Axial strain εL,c
Fig. 12. Comparison of axial stress-axial strain relationship.
510 F.-x. Ding et al. / Thin-Walled Structures 107 (2016) 502–513

8
Radial stress σr,c (MPa)

HST1

4
endpoint 1#
mid-point 3#
2

0
0.000 0.005 0.010 0.015 0.020
Axial strain εL,c
Fig. 13. Comparison of radial stress-axial strain relationship for different measured
point.

Table 3
Properties of numerical specimens.

Number B L t fcu fy fsc Nu, FE

mm mm mm MPa MPa MPa kN

1 200 800 6 40 235 46.60 4594

2 70 235 77.12 7603
3 70 345 83.96 8278
4 90 345 106.29 10,479
5 90 420 110.91 10,934
6 120 420 143.2 14,118
7 11 40 235 61.34 5770
8 70 235 79.43 7471
9 70 345 99.87 9394
10 90 345 123.7 11,636
11 90 420 127.67 12,009
12 120 420 159.36 14,990
13 16 40 235 70.8 6333
14 70 235 97.58 8729
15 70 345 121.67 10,884
16 90 345 134.33 12,016
17 90 420 151.31 13,535
18 120 420 174.28 15,590

Fig. 15. Division on stress region of cross section (a) stress contour of FE resulting;
(b) simplified stress distribution.
Ratio of axial stress at ultimate state

1.2
Table 4
1.0 Comparison between experimental and numerical predictions of ultimate bearing
capacity for specimens adopted in this paper.
to yield strength σL,S /fy

0.8
avg=0.80 Specimens Nu, Exp Nu, FE Nu, Eq.14 Nu,[21] Nu, FE/ Nu, Eq.14/ Nu,[21]/Nu,
Nu, Nu,
0.6 Exp Exp Exp

kN kN kN kN   
0.4
HST1-A 4947 4806 5335 5316 0.971 1.078 1.075
endpoint HST1-B 4618 4902 5424 5404 1.061 1.175 1.170
0.2 mid-point HST2-A 6001 5666 6143 6112 0.944 1.024 1.018
HST2-B 6041 5724 6312 6280 0.948 1.045 1.040
1/4 point HST3-A 6827 6423 7069 7049 0.941 1.035 1.033
0.0 HST3-B 6803 6482 7148 7127 0.953 1.051 1.048
40 80 120 160 200 HST4-A 7079 7280 7988 7956 1.028 1.128 1.124
Ultimate strength fsc (MPa) HST4-B 7289 7067 7752 7720
Avg.
0.970
0.98
1.064
1.08
1.059
1.07
St. dev. 0.05 0.05 0.05
Fig. 14. Average ratio of axial compressive stress to yield stress of steel tube.
 F.-x. Ding et al. / Thin-Walled Structures 107 (2016) 502–513 511

Table 5
Comparison between experimental and numerical predictions of ultimate bearing capacity for specimens from Ref. [24].

Specimens[24] t b fcc fcu Nu, Exp Nu, Eq.14 Nu,[21] Nu, Eq.14/Nu, Exp Nu,[21]/Nu, Exp

mm mm MPa MPa kN kN kN – –

HS1510 1.5 72 13.3 16.8 400 441 439 1.103 1.098

HS1520 26.0 33.0 510 659 657 1.292 1.288
HS1530 35.3 44.7 690 816 814 1.183 1.180
HS3010 3.0 69 13.3 16.8 620 620 615 1.000 0.992
HS3020 26.0 33.3 700 820 815 1.171 1.164
HS3030 35.3 44.7 810 965 960 1.191 1.185
HS5010 5.0 65 13.3 16.8 1080 830 823 0.769 0.762
HS5020 26.0 33.0 1180 1009 1001 0.855 0.848
HS5030 35.3 44.7 1220 1137 1130 0.932 0.926
HS8010 8.0 59 13.3 16.8 1620 1090 1080 0.673 0.667
HS8020 26.0 33.0 1690 1237 1226 0.732 0.725
HS8030 35.3 44.7 1780 1343 1332 0.754 0.748
Avg. 0.97 0.97
St. dev. 0.21 0.21

also sustain the axial load after the steel tube yield.
Fig. 12 shows axial stress-strain curves of core concrete for a
half of specimens HST1. It is evident that the compressive strength
of core concrete has been improved evidentially compared with
plain concrete due to the confinement effect of the steel tube on
core concrete. The load capacity of the endpoint is greater than
that of the mid-point, because of the larger lateral compressive
stress on the endpoint concrete. The comparison of lateral com-
pressive stress distribution at both endpoint and mid-point is gi-
ven in Fig. 13.
4. Bearing capacity calculation
4.1. Model simplification
According to Ding et al. [22], a similar derived way can be adopted
in this section. In the ultimate state, the FE result of the typical
stress distribution at section for the hexagonal CFT stub columns is
shown in Fig. 15(a). It is indicated that the hexagonal steel tube
mainly has confinement effect on the six small corner areas and
the middle area of core concrete, while the length of un-
constrained areas is approximately 0.4b, where b is the edge
length of core concrete. According to the stress contour in Fig. 15
(a), the stress distribution of the hexagonal CFT columns can be
simplified as Fig. 15(b). This is completely based on stress dis-
tribution and the superposition principle of the concrete section at
the ultimate state, where Ac1 is the non-constrained area of core
concrete, Ac2 is the constrained area of core concrete. In this way,
the following relationship can be obtained:
Ac1 = 0.33Ac (6)

A simplified formula for calculating the ultimate bearing ca- Ac2 = 0.67Ac (7)
pacity of the square CFT stub column under axial loading has been
proposed by Ding et al. [22]. The methodology has been in-
corporated into this paper to predict the bearing capacity of the 4.2. Formulation
hexagonal CFT stub column. Effects of concrete strength, steel ratio
and steel yield strength were further investigated using FE mod- As shown in Fig. 15(b), the relationship between radial concrete
elling. Concrete grades ranging from C40 to C120, and steel grades stress (sr,c) in constrained area and the transverse stress of the
ranging from Q235 to Q420 were included. The concrete C40 refers steel tube (sθ,s) can be expressed as:
to a concrete with characteristic 28-day strength of 40 MPa using
standard cubes, according to Chinese standard [26]. The values of 3t
σr, c = σθ, s
these parameters were chosen based on the engineering practice. b (8)
18 groups of optimized numerical specimens have been adopted in and the axial compressive strength of core concrete in constrained
this study. The properties of them are shown in Table 3. Fig. 14 area is:
shows the ratio of the axial stress to yield strength when the
σL, c = fc + pσr, c (9)
specimens reach their ultimate strength (fsc ¼Nu/Asc, where
Asc ¼Ac þAs is the total area of cross-section). The axial stress was where p is the lateral pressure coefficient reported in Ref. [9],
extracted from the numerical results by taking the in-plane ver- p ¼3.4.
tical stress component of the steel tubes. Three points on the steel Based on the condition of static equilibrium, the axial ultimate
tube were investigated: endpoint, 1/4 point and mid-point of bearing capacity Nu can be given as:
middle section. It can be identified from Fig. 14 that when hex-
agonal CFT stub columns reach their ultimate strength, the average Nu = σL, cAc2 + fc Ac1 + σL, sAs (10)
value of the ratio of the steel tube's axial compressive stress to
where As is the cross-sectional area of hexagonal steel tube.
yield stress is:
The following equations can be given by geometric relation-
σL, s = 0.80f y (4) ship:
As = 6bt (11)
Based on Von Mises yield criterion, the tensile transverse stress
(sθ,s) of the steel tube can be obtained as:
3 3 b2
σθ, s = 0.32f y Ac =
(5) 2 (12)
512 F.-x. Ding et al. / Thin-Walled Structures 107 (2016) 502–513
Combining Eqs. (11)–(12), the ratio of steel thickness to con-
crete edge length was determined as:
t 3 As
=
b 4A c
Substituting Eqs. (4)–(9) and (13) into Eq. (10), Nu can be ob-
tained as:
Nu = fc Ac1 + σL, cAc2 + σL, sAs = 0.33fc
⎛ fy A s ⎞
Ac + 0.67Ac ⎜⎜ fc + 0.816 ⎟ + 0.8f As = f Ac + 1.3f As
⎝ A c ⎟⎠ y c y
where the value of 1.3 is the confinement coefficient of hexagonal
CFT stub columns, which falls between the value of square CFT
columns (1.2) [22] and circle CFT column (1.7) [9].
4.3. Verification
Tables 4 and 5 list the comparisons between the results of load-
bearing capacity calculated by Eq. (14), FE modelling, unified cal-
culation from [21], and the experimental results for all hexagonal
CFT specimens from this paper and Ref. [24], respectively. It is seen
from Table 4 that an average ratio of 0.98 with a standard devia-
tion of 0.05 were obtained by comparing FE modelling with ex-
perimental results, while an average ratio of 1.08 with a standard
deviation of 0.05 were achieved when comparing the prediction
by the proposed equation Eq. (14) with experimental results. The
proposed equation Eq. (14) in this paper ensured same accuracy
level as the formula given by [21], while it is much simpler in
application. Same findings have been revealed using experimental
work conducted by Evirgen et al. [24], as shown in Table 5.
5. Conclusion
This paper presented a combined experimental, numerical and
theoretical study on the mechanical behaviour of hexagonal CFT
stub columns subjected to axial compressive loading. Parametric
studies were conducted in order to understand the influence of
some key parameters on the behaviour of the hexagonal CFT stub
columns, such as concrete strength, steel ratio etc. A simplified
formula was proposed to predict the load bearing capacity of the
hexagonal CFT stub columns under compressive loading. Based on
the results, the following conclusions can be drawn:
1) Increasing the strength grade of concrete will increase the ul-
timate bearing capacity of the hexagonal CFT stub column.
However, it will decrease the ductility of the CFT column. Be-
sides, increasing the steel ratio does not improve the ultimate
bearing capacity of the hexagonal CFT stub column significantly,
while it has a remarkable benefit on the ductility of the CFT
column.
2) FE modelling adopted in this paper showed great reliability in
studying the effects of steel ratio and concrete strength on the
load bearing capacity of the hexagonal CFT stubs. The confine-
ment effect of steel tube on core concrete is greater at endpoints
than that at mid-points, and more than 60% of ultimate capacity
of steel tube still sustain the axial load after the steel tube yield.
3) A simplified approach was proposed to predict the ultimate
bearing capacity of the hexagonal CFT stub column. The equa-
tion is simple in application, and the predictions were found
reliable. It is therefore recommended for designing purpose.
Moreover, the confinement coefficient of hexagonal CFT stub
columns is 1.3 as expected, which falls between that of the
square and circular CFT columns.
Acknowledgment
(13)
This research work was financially supported by the National
Natural Science Foundation of China, Grant No. 51578548.
References
(14) [1] S. Morino, K. Tsuda, Design and construction of concrete-filled steel tube
column system in Japan, Earthq. Eng. Eng. Seismol. 4 (1) (2003) 51–73.
[2] X. Chang, L. Fu, H.B. Zhao, Y.B. Zhang, Behaviors of axially loaded circular
concrete-filled steel tube (CFT) stub columns with notch in steel tubes, Thin-
Walled Struct. 73 (12) (2013) 273–280.
[3] X. Chang, X.L. Luo, C.X. Zhu, C. Tang, Analysis of circular concrete-filled steel
tube support in high ground stress conditions, Tunnel. Undergr. Space Technol.
43 (6) (2014) 41–48.
[4] X. Chang, H.X. Lin, Experimental study on shear resistance of self-stressing
concrete filled circular steel tubes, J. Constr. Steel Res. 65 (4) (2009) 801–807.
[5] M.F. Hassanein, O.F. Kharoob, Q.Q. Liang, Circular concrete-filled double skin
tubular short columns with external stainless steel tubes under axial com-
pression, Thin-Walled Struct. 73 (12) (2013) 252–263.
[6] M.F. Hassanein, O.F. Kharoob, Analysis of circular concrete-filled double skin
tubular slender columns with external stainless steel tube, Thin-Walled Struct.
79 (6) (2014) 23–37.
[7] M.F. Hassanein, Numerical modeling of concrete-filled lean duplex slender
stainless steel tubular stub columns, J. Constr. Steel Res. 66 (8–9) (2010)
1057–1568.
[8] S.H. Lee, B. Uy, S.H. Kim, Y.H. Choi, S.M. Choi, Behavior of high-strength circular
concrete-filled steel tubular (CFST) column under eccentric loading, J. Constr.
Steel Res. 67 (1) (2011) 1–13.
[9] F.X. Ding, Z.W. Yu, Y. Bai, Y.Z. Gong, Elasto-plastic analysis of circular concrete-
filled steel tube stub columns, J. Constr. Steel Res. 67 (10) (2011) 1567–1577.
[10] C. Hou, L.H. Han, X.L. Zhao, Concrete-filled circular steel tubes subjected to
local bearing force: Experiments, J. Constr. Steel Res. 83 (4) (2013) 90–104.
[11] Z.W. Yu, F.X. Ding, C.S. Cai, Experimental behavior of circular concrete-filled
steel tube stub columns, J. Constr. Steel Res. 63 (2) (2007) 165–174.
[12] K. Uenaka, H. Tsunokake, Concrete filled elliptical steel tubular members with
large diameter-to-thickness ratio subjected to bending, Structures 5 (2) (2016)
58–66.
[13] K. Uenaka, Experimental study on concrete filled elliptical/oval steel tubular
stub columns under compression, Thin-Walled Struct. 78 (5) (2014) 131–137.
[14] H. Yang, D. Lam, L. Gardner, Testing and analysis of concrete-filled elliptical
hollow sections, Eng. Struct. 30 (12) (2008) 3771–3781.
[15] M. Tomii, K. Yoshimura, Y. Morishita, Experimental studies on concrete-filled
steel tubular stub columns under concentric loading. International Colloquium
On Stability of Structures Under Static and Dynamic Loads, Washington DC,
May 17–19, 1977, pp. 718–741.
[16] F.X. Ding, D.R. Lu, Y. Bai, W.S. Zhou, M. Ni, Z.W. Yu, G.S. Jiang, Comparative
study of square stirrup-confined concrete- filled steel tubular stub columns
under axial loading, Thin-Walled Struct. 98 (1) (2016) 443–453.
[17] Y. Qin, Z.H. Chen, X.D. Wang, Elastoplastic behavior of through-diaphragm
connections to concrete-filled rectangular steel tubular columns, J. Constr.
Steel Res. 93 (2) (2014) 88–96.
[18] J. Cai, Z.Q. He, Axial load behavior of square CFT stub column with binding
bars, J. Constr. Steel Res. 62 (5) (2006) 472–483.
[19] D.L. Liu, W.M. Gho, Axial load behavior of high-strength rectangular concrete-
filled steel tubular stub columns, Thin-Walled Struct. 43 (8) (2005) 1131–1142.
[20] Q.Q. Liang, B. Uy, J.Y.R. Liew, Strength of concrete-filled steel box columns with
load buckling, Aust. Struct. Eng. Conf. 9 (2005) 11–14.
[21] M. Yu, X.X. Zha, J.Q. Ye, Y. Li, A unified formulation for circle and polygon
concrete-filled steel tube columns under axial compression, Eng. Struct. 49 (4)
(2013) 1–10.
[22] F.X. Ding, C.J. Fang, Y. Bai, Y.Z. Gong, Mechanical performance of stirrup-con-
strained concrete-filled steel tubular stub columns under axial loading, J.
Constr. Steel Res. 98 (11) (2014) 146–157.
[23] E. Ketema, S. Taye, Design chart procedures for polygonal concrete-filled steel
columns under uniaxial bending, Ethiop. J. Sci. 29 (2006) 1–16.
[24] B. Evirgen, A. Tuncan, K. Taskin, Structural behavior of concrete filled steel
tubular sections (CFT/CFSt) under axial compression, Thin-Walled Struct. 80
(7) (2014) 46–56.
[25] GB50017-2003, Code for Design of Steel Structures, China Planning Press,,
Beijing, 2003.
[26] GB/T50081-2002, Standard for Method of Mechanical Properties on Ordinary
Concrete, China Building Industry Press,, Beijing, 2002.
[27] GB/T228-2002, Metallic Materials-Tensile Testing at ambient Temperatures,
Standards Press of China,, Beijing, 2002.
[28] P. Baltay, A. Gjelsvik, Coefficient of friction for steel on concrete at high normal
stress, J. Mater. Civ. Eng. 2 (1) (1990) 46–49.
[29] F.X. Ding, J. Liu, X.M. Liu, Z.W. Yu, D.W. Li, Mechanical behavior of circular and
 F.-x. Ding et al. / Thin-Walled Structures 107 (2016) 502–513 513

square concrete filled steel tube stub columns under local compression, Thin-
Walled Struct. 94 (9) (2015) 155–166.
[30] F.X. Ding, X.Y. Ying, L.C. Zhou, Z.W. Yu, Unified calculation method and its
application in determining the uniaxial mechanical properties of concrete,
Front. Archit. Civ. Eng. China 5 (3) (2011) 381–393.
[31] N.S. Ottosen, M. Ristinmaa, 12-Common Plasticity Models The Mechanics of
Constitutive Modeling, 2005, pp. 279–319.
[32] F.X. Ding, L. Fu, Z.W. Yu, G. Li, Mechanical performances of concrete-filled steel
tubular stub columns with round ends under axial loading, Thin-Walled
Struct. 97 (12) (2015) 22–34.
[33] F.X. Ding, Behavior and Design Method of Concrete Filled Circular Steel Tub-
ular Structures, Central South University,, Changsha, 2006.