Thin-Walled Structures 107 (2016) 453–461

Contents lists available at ScienceDirect

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

Full length article

Composite action of octagonal concrete-filled steel tubular stub

columns under axial loading
Fa-xing Ding a, Zhe Li a,n, Shanshan Cheng b, Zhi-wu Yu a,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: This paper presents a combined experimental, numerical and theoretical study on the mechanical per-
Received 3 April 2016 formances of octagonal CFT stub columns subjected to axial compressive loading. Eight specimens of
Received in revised form axial compression tests have been carried out aiming at investigating the effects of the concrete strength
11 July 2016
and steel ratio on the mechanical and deformed behavior of octagonal CFT stub columns. 3D finite
Accepted 15 July 2016
element modelling was established for parametric studies to probe into the composite action between
Available online 25 August 2016
the steel tube and the core concrete of the octagonal CFT stub column. In addition, a practical formula has
Keywords: been proposed to predict the ultimate bearing capacity of the octagonal CFT stub column with the
Octagon confinement coefficient of 1.5, which is an amplification coefficient of the vertical bearing capacity of
Composite action
steel tube considering the composite action between the core concrete and the steel tube.
Ultimate bearing capacity
& 2016 Elsevier Ltd. All rights reserved.
Practical formula
Strain ratio
Confinement coefficient

1. Introduction Tomii [14]. Yu [15] presented a unified formula of the ultimate

load-bearing capacity for circular and polygonal CFT columns
A concrete-filled steel tube (CFT) column is a composite subjected to axial compression.
member which is consisted of a steel tube filled with concrete. The Eight axial compression specimens of octagonal CFT stub col-
concrete core adds stiffness to the steel tube and prevents the umns have been conducted at the Central South University in
occurrence of inward local buckling, while the steel tube improves China to investigate the composite action between the core con-
the compressive strength and deformed performance of the core crete and the steel tube of octagonal CFT stub columns. The non-
concrete. Given the favorable actions such as high strength, high linear finite element analysis (FE) of octagonal CFT stub columns
ductility and large energy absorption capacity, the CFT column has throughout loading history has been carried out using the com-
been widely used in practical structures especially in long-span, mercial software ABAQUS. Based on the above FE results, the strain
high-rise and heavy-sustainment structures [1]. ratio of steel tube and the stress distribution of core concrete and
The large aggregation of experimental and numerical studies steel tube at the corner and mid-points have been illustrated, as
on CFT columns with circular and square cross-sections used in well as the practical formulas for the ultimate bearing capacity of
common under the axial compressive load has been conducted in octagonal CFT stub columns have been proposed by using super-
position principle at the ultimate state and simplification on stress
literature [2–10]. Mechanical behavior of the circular [11], square
contour, which follows the same research idea as Ding et al. [11–
[12] and rectangular-with-round-ends [13] sectional CFT stub
13]. Details of the studies have been explained in the following
columns under axial loading have been investigated by the co-
sections of this paper.
author Ding, and practical formulas of load-bearing capacities have
been proposed with the corresponding confinement coefficient.
However, available studies on octagonal CFT stub columns are
2. Experimental investigation
rather few in literature, although octagonal cross-sections are
demanded in architectural design in some cases. Experimental
2.1. Materials
studies on octagonal CFT stub columns has been conducted by
Two types of concretes were applied with cubic compressive
n
Corresponding author. strengths 39.3 MPa and 57.4 MPa respectively, at a curing age of
E-mail address: [email protected] (Z. Li). 3 months. The cubic compressive strength and elastic module of

http://dx.doi.org/10.1016/j.tws.2016.07.012
0263-8231/& 2016 Elsevier Ltd. All rights reserved.
454 F.-x. Ding et al. / Thin-Walled Structures 107 (2016) 453–461

Nomenclature columns from FE results

Nu,[15] Ultimate bearing capacity of octagonal CFT stub col-
Ac Cross-sectional area of core concrete umns from reference [15]
( ¼Ac1 þAc2 ¼ 2b2tan 67.5°) p Lateral pressure coefficient
Ac1 Unconstrained area of core concrete (¼ 0.3Ac) t Wall thickness of steel tube
Ac2 Constrained area of core concrete ( ¼0.7Ac) s Axial stress of concrete
As Cross-sectional area of steel tube ( ¼8bt) si Equivalent stress of steel tube
Asc Total area of cross-section ( ¼As þ Ac) sL,c Axial compressive stress of core concrete
B Edge length of octagonal section sL,s Axial compressive stress of steel tube
b Edge length of core concrete (¼ B2t tan 22.5°) sr,c Radial compressive stress of constrained concrete
DI Ductility index, given by DI¼ ε0.85/εb, where εb ¼ ε0.75/ sθ,s Tensile transverse stress of steel tube
0.75 ε0.75 Axial strain when the load attains 75% of the ultimate
Es Elastic modulus of steel tube (¼2.06
105 MPa) load in the pre-peak stage
fc Compressive strength of prism volume ε0.85 Axial strain when bearing capacity is decreased to 85%
fcu Compressive strength of cubic concrete of ultimate value
fsc Ultimate strength of CFT columns ε Axial strain of concrete
fu Ultimate strength of steel tube (¼1.5fy) εc Strain corresponding with the peak compressive stress
fy Yield strength of steel tube of concrete
k Ratio of initial tangent modulus to secant modulus at εL Axial strain of CFT columns
peak stress εL,s Axial compressive strain of steel tube
L Height of specimens εi Equivalent strain of steel tube
N Axial load εy Yield strain of steel tube
Nu Ultimate bearing capacity εst Hardening strain of steel tube ( ¼12εy)
Nu, Eq.10 Ultimate bearing capacity of octagonal CFT stub col- εu Ultimate strain of steel tube (¼ 120εy)
umns from Eq. (10). εθ,s Tensile transverse stress of steel tube
Nu, Exp Ultimate bearing capacity of octagonal CFT stub col- vsc Strain ratio of steel tube (¼ εθ,s/εL,s)
umns from experimental results ρ Steel ratio of columns (¼As/Asc)
Nu, FE Ultimate bearing capacity of octagonal CFT stub

Table 1 2.2. Test specimens

Properties of steel.

A total of 4 groups of octagonal CFT were designed in this study

Thickness fy (MPa) fu (MPa) Es (MPa) vs
and the details were shown in Table 2. Each test was repeated
4 mm 311 460 2.09
105 0.292 twice (namely A and B), and thereby eight specimens were made
6 mm 321 480 2.02
105 0.256 in total. The nominal dimensions of each specimen were 200 (B)
mm
4(6) (t) mm
1500 (L) mm, where B is the outer edge length
of the octagonal section, t is the wall thickness of the steel tube
concrete were measured in accordance with the standard GB/
and L is the height of the specimen.
T50081-2002 [16]. Moreover, two types of steel were adopted in
The octagonal steel tubes were molded by bending steel plates
the test, of which the material properties were determined
into grooves and then welding the two edges at the corner. The
through tensile coupon tests. Three coupons were cut and tested positions at which the butt welds were made were labelled in
in accordance with the standard GB/50017-2003 [17]. The yield Fig. 1.
strength, ultimate tensile strength, elastic module and Poisson's For a better observation of deformation and preventing steel
ratio of steel were listed in Table 1. 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

Table 2
Properties of specimens and comparison between calculated results and tested ones.

Reference Specimens B
t (mm) L (mm) fcu (MPa) fy (MPa) Steel DI Nu, Exp Nu, FE (kN) Nu, Eq.10 Nu,[15] (kN) Nu, FE/ Nu, Eq.10/ Nu,[15]/
ratio ρ (kN) (kN) Nu, Exp Nu, Exp Nu, Exp

Present paper OST1-A 201
3.85 1500 39.3 311 0.032 6.806 9297 8942 10,290 8069 0.968 1.107 0.868
OST1-B 199
3.98 0.033 3.052 9311 8850 10,199 8012 0.950 1.095 0.861
OST2-A 200
6.02 321 0.049 7.987 10,502 10,128 11,796 9466 0.964 1.123 0.901
OST2-B 197
5.89 0.049 7.734 10,713 9800 11,418 9160 0.915 1.066 0.855
OST3-A 200
3.92 57.4 311 0.032 3.984 12,362 12,048 13,631 11,055 0.975 1.103 0.894
OST3-B 199
4.02 0.033 4.034 12,357 11,906 13,572 11,015 0.964 1.098 0.891
OST4-A 197
5.88 321 0.049 8.054 12,992 12,582 14,637 12,025 0.968 1.127 0.926
OST4-B 198
5.98 0.049 5.312 13,263 12,787 14,832 12,189 0.964 1.118 0.919

Reference[14] 2HN 62.1
2.0 / 38.06 341.3 0.053 / 1003 968 1173 925 0.965 1.170 0.923
3HN 62.1
3.2 300.2 0.084 1100 1133 1350 1084 1.030 1.227 0.986
4HN 62.1
4.0 294.3 0.104 1273 1241 1489 1210 0.975 1.170 0.950

Avg. 0.967 1.128 0.907

St. dev. 0.005 0.044 0.058

 F.-x. Ding et al. / Thin-Walled Structures 107 (2016) 453–461 455

Fig. 1. Test specimen and instrumentation.

15000 data acquisition system.

The compressive load was applied from the top of the speci-
12000 mens through a 2000-ton axially static stress testing system
Axial load N (kN)

control mode. The test was initially carried out using load control
9000 in the elastic stage, with a force increment of 1/20 of the expected
ultimate load. After the load reached 60% of the expected ultimate
6000 bearing capacity, the test was switched to displacement control
OST1-B with an increment of 0.2 mm. Each loading step took 3–5 min.
3000 OST2-B When the expected ultimate load was achieved, specimens were
OST3-A loaded at a step of 0.5 mm and maintained 5 min. The tests were
OST4-B
0 stopped when the axial strain reached 0.04 which was the largest
0.00 0.01 0.02 0.03 0.04 strain of the specimens. The test of each specimen lasted about
Axial strain εL 3 h.

Fig. 2. Comparison of load-strain curves of specimens. 2.4. Experimental results and discussion

painted surface. Before the placement of concrete, the octagonal Fig. 2 shows the external load against the corresponding axial
steel tube was welded with the bottom loading plate together. strain curves of specimens, which reveals that the compressive
Concrete was poured from the top of the specimens and vibrated process of octagonal CFT stub columns are generally divided into
with a vibrator at the same time to evenly distribute the concrete three stages: elastic stage, elastic-plastic stage and failure stage.
inside the steel tube. Finally, the top surface of concrete was At the initial loading stage, all the specimens were in elasticity,
smoothed for being at the same level with steel tube. In addition, which has been illustrated by the linear responses of the load-
the standard concrete cubes with dimension of 150 mm were strain curves. The compressive stiffness of the specimens in this
made and cured at the same condition as the specimens. Due to stage was larger than that in the other stages, consequently the
regular calibration of the loading system in the lab, all the speci- axially elastic displacement of specimens and axial and transverse
mens were tested after about three months. The compressive cubic strain of steel tube were very small.
strengths of the concrete measured at 3 months were 39.3 MPa When the imposed load reached 60–70% of the ultimate load,
and 57.4 MPa, which were slightly higher than those with all the specimens went into elastic-plastic stage, in which the
35.5 MPa and 54.4 MPa at 28 days, respectively. The concrete visible local buckling of steel tube initially generates near both
surface of the column specimens was polished with grinder, and ends of the specimens due to the end effect and then emerged in
then daubed a layer of epoxy resin binder. Finally, steel cover the middle where local bucking develops rapidly. When the axial
plates were bonded at the top end of the octagonal CFT column load reached the ultimate loading capacity, an apparent buckling
specimens to guarantee that the steel tube and core concrete could be seen on the steel tube.
shared loads from the initial loading stage. After the ultimate load, the load-bearing capacity of the spe-
cimens decreased dramatically while the displacement increased
2.3. Experimental setup and instrumentation due to 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
Axial compressive experiments on CFT stub columns were apparent that the load bearing capacity of specimens OST1 and
conducted using a 2000-ton tri-axial stress testing machine in the OST3 (with smaller ductility) declined faster than those of speci-
National Engineering Laboratory for High Speed Railway Con- mens OST2 and OST4 (with bigger ductility).
struction at the Central South University. To accurately measure After experiments, the steel tubes are cut open in order to
the deformation, four strain rosettes (i.e. 1# to 4# in Fig. 1) were examine the failure mode of the core concrete. Inclined shear
attached at the mid-height of two opposite side surfaces and two rupture zones were seen near the middle height, as shown in
LVDTs (i.e. 1# and 2# in Fig. 1) were installed at the same height of Fig. 4. In this phase, the tensile transverse strain of steel tube in-
other two opposite side surfaces. Load-strain curves were acquired creased rapidly, which could be attributed to the fast growth of the
by a DH3818 static strain measurement system and load-de- lateral expansion of core concrete at shear rupture zone. As the
formation curves were recorded by the electronic transducers and ductility of butt weld is poorer than steel tube, weld failure along
456 F.-x. Ding et al. / Thin-Walled Structures 107 (2016) 453–461

Fig. 5. Butt weld failure of specimens.

15000

Axial load N (kN)

12000

OST1-A ρ =0.032 C39.3

OST3-A ρ =0.032 C57.4

OST2-A ρ =0.049 C39.3
OST1-B ρ =0.033 C39.3

OST4-B ρ =0.049 C57.4

OST4-A ρ =0.049 C57.4
OST2-B ρ =0.049 C39.3

OST3-B ρ =0.033 C57.4

9000

6000

3000

0
1 2 3 4 5 6 7 8
Fig. 6. Comparison of ultimate bearing capacity for all specimens.

effect of steel ratio on mechanical behaviors of octagonal CFT stub

Fig. 3. Typical failure modes of specimens (a) OST1-A (b) OST2-B (c) OST3-A columns is also discussed. The steel ratio is the variable while
(d) OST4-A. other parameters are kept the same as mentioned above. In
comparison to the OST1 columns, the average ultimate bearing
capacity of OST2 specimens was improved by 14.0%, with a 55.2%
increase of steel ratio. However, the comparison between OST4
and OST3 reveals that the average ultimate bearing capacity only
improved slightly (6.2%) with the same increase of steel ratio. It is
then concluded that increasing steel ratio can produce an im-
provement of load bearing capacity, especially for the octagonal
CFT columns consisted with low-grade concrete.

2.5.2. Concrete strength

The concrete strength is another key parameter which influ-
ences the mechanical behaviors of octagonal CFT stub columns.
The concrete strength was the variable while other parameters are
kept the same as mentioned above. Comparison of the average
ultimate bearing capacities of between OST3 and OST1 shows that
with a 46.1% increase of concrete strength, the ultimate bearing
capacity of OST3 improved by 32.8%. However, when comparing
Fig. 4. The failure modes of core concrete. OST4 and OST2, it is revealed that the average ultimate bearing
capacity of OST4 specimens is only improved by 23.8% with the
the butt welds was found from OST1 specimens, as shown in Fig. 5. same increase of concrete strength. It is therefore indicated that an
increase of concrete strength can yield an improvement of load
2.5. Parametric studies bearing capacity, and the effect is more apparent to those columns
with smaller steel ratio.
2.5.1. Steel ratio
Fig. 6 shows the experimental results of ultimate bearing ca- 2.5.3. Ductility
pacity for all specimens. The steel ratio (ρ) is defined as the steel To investigate the effect of steel ratio and concrete strength on
area (As) divided by the total cross-sectional area (Asc). For CFT the ductility of octagonal CFT columns, a ductility index (DI) pro-
stub columns, the steel ratio is a key parameter which has been posed by Han et al. [18] is also adopted in this paper and the
tentatively investigated in previous researches. In this paper, the corresponding value is defined as follows:
 F.-x. Ding et al. / Thin-Walled Structures 107 (2016) 453–461 457

9.0 12000
OST1-A
7.5

Axial load N (kN)

9000
Ductility index DI

OST3-A ρ =0.032 C57.4

OST2-A ρ =0.049 C39.3
OST1-B ρ =0.033 C39.3
OST1-A ρ =0.032 C39.3

OST4-B ρ =0.049 C57.4

OST4-A ρ =0.049 C57.4
OST2-B ρ =0.049 C39.3

OST3-B ρ =0.033 C57.4

6.0
6000
4.5 tested results of 1#
tested results of 2#
3000
3.0 tested results of 3#
FE results of 1#
1.5 FE results of 3#
0
0.0 0.2 0.4 0.6 0.8 1.0
0.0 Transverse strain νsc
1 2 3 4 5 6 7 8 (a) OST1-A
Fig. 7. Comparison of ductility index DI for all specimens.
12000
OST2-A

Axial load N (kN)

ε0.85 9000
DI =
εb (1)
6000
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 3000 endpoint 1#
1/4 point 2#
stage. Table 2 shows the ductility index calculated by Eq. (1) for all
mid-point 3#
specimens. 0
Fig. 7 shows the DI calculated from Eq. (1) for all tested spe- 0.0 0.2 0.4 0.6 0.8 1.0
cimens, where a larger value of DI indicates a slower process of Transverse strain vsc
load reduction after the peak load. When the concrete strength is (b) OST2-A
increased, the DI value reduces by an average of 18.7% when 15000
OST3-A
comparing the specimens OST3 with OST1. However, the reduction
of DI value shows an average of 15.0% when comparing the spe- 12000
Axial load N (kN)

cimens OST4 with OST2. This result is suggesting that a higher

9000
concrete strength leads to a weaker ductility. Moreover, when
increasing the steel ratio, the DI value increases 59.5% on average
6000
when comparing the specimens OST2 with OST1. And the DI value
increases 66.7% on average when comparing the specimens OST4 3000 endpoint 1#
with OST3. It can then be found that a higher steel ratio leads to a 1/4 point 2#
mid-point 3#
higher ductility. 0
0.0 0.2 0.4 0.6 0.8 1.0
2.6. Strain ratio Transverse strain vsc
(c) OST3-A
Fig. 8 shows the relationship of the axial load (N) against the 15000
strain ratio (vsc) of OST1-A to OST4-A specimens. The strain ratio, OST4-A
which is defined as the absolute value of the circumferential strain 12000
Axial load N (kN)

divided by the axial strain of all gauged points, represents the

composite action between the steel tube and the core concrete. 9000
The more the strain ratio is, the more the composite action is
between them. 6000
At the initial loading stage, the strain ratio remained constant endpoint 1#
3000
and in close proximity to Poisson's ratio of steel. It is indicated that 1/4 point 2#
there was nearly no composite action between the steel tube and mid-point 3#
0
core concrete. After the specimens reached 60% of their ultimate 0.0 0.2 0.4 0.6 0.8 1.0
capacities, the strain ratio started to grow remarkably, which Transverse strain vsc
showed an obvious confinement effect of steel tube on core con- (d) OST4-A
crete. At any period of load time, the largest strain ratio of the
octagonal steel tubes occurred at endpoint 1#, and the smallest Fig. 8. Comparison of load-transverse strain curves (a) OST1-A (b) OST2-A
(c) OST3-A (c) OST4-A.
one at mid-point 3#. It is demonstrated that the octagonal steel
tube has much more confinement effect on the core concrete at
the corner than that at the center. established using commercial software ABAQUS/Standard 6.10.
In these models, 8-node reduced integral format 3D solid ele-
ments (C3D8R) were used to model the core concrete, steel tube
3. Finite element (FE) modelling and loading plates for all specimens. A structured meshing option
was adopted, and the mesh result is shown in Fig. 9. A surface-to-
3.1. FE models surface contact was adopted for the constraint between the steel
tube and the core concrete with finite slip formula. The interaction
A nonlinear finite element (FE) analysis model for simulation of of the normal direction of two surfaces was hard contact, while
octagonal CFT stub columns subjected to axial compression is the one of tangential direction was Coulomb friction with the
458 F.-x. Ding et al. / Thin-Walled Structures 107 (2016) 453–461

referred in [19].
The concrete damaged plasticity model defined in ABAQUS was
used. The eccentricity is taken as 0.1 [21]; the ratio of initial
equibiaxial compressive yield stress to initial uniaxial compressive
yield stress is taken as 1.225 [22]; the ratio of the second stress
invariant on the tensile meridian to that on the compressive
meridian is taken as 2/3 [23]; the viscosity parameter is taken as
0.005 [24]; and the dilation angle is taken as 40° [22].
In case of the above parameters available, the adopted concrete
model, in which the composite action between the steel tube and
the core concrete is considered, is suitable for numerical analysis
on triaxial-compressed concrete in CFT columns with octagonal
cross-section.
Based on a large number of experimental studies on the me-
chanical properties of steel, an elasto-plastic model of steel, con-
sidering Von-Mises yielding criteria, Prandtl-Reuss flow rule and
isotropic strain hardening, is used to describe the constitutive
behavior of steel tube in CFT columns. The expression is given as
follows [19]:
⎧ Esεi εi ≤ εy
⎪
⎪
⎪ fy εy < εi ≤ εst
σi = ⎨
⎪ y
f + ζ E (ε –
s i stε ) εst < εi ≤ εu
⎪
⎪
⎩ fu εi > εu (3)

where, si and εi are the equivalent stress and strain of the steel
tube. fy, fu (¼1.5fy) and Es ( ¼2.06
105 MPa) are the yield
strength, ultimate strength, and elastic modulus of steel respec-
tively. εy, εst and εu are the yield strain, hardening strain, and ul-
timate strain of steel, which have been defined as εst ¼ 12εy and
εu ¼ εst þ0.5fy/(ζEs) ¼120εy. The parameter ζ is taken as 1/216.
Note that the Prandtl-Reuss flow rule, which can be applicable
to three-dimensional stress analysis, is used to the constitutive
relationship of the steel tube, because the elastic part is considered
into the increment in strain during the process of plastic
deformation.
Fig. 9. Mesh generation of model (a) FE model, (b) cover plate element, (c) steel To model the decrease of load-bearing capacity of specimens,
tube element, (d) core concrete element. load is applied using displacement control, and both material and
structural nonlinearities are considered and solved using the in-
friction coefficient of 0.5 which is identified with the previous cremental-interactive method in ABAQUS.
research [12]. The tie constraint was adopted between the core
concrete and the loading plate, and also used between the steel 3.2. Validation
tube and the loading plate in modelling.
A tri-axial concrete model of core concrete in CFT columns with The comparison of experimental and numerical ultimate
the circular or polygonal cross-section under axial compression bearing capacity is shown in Table 2. Good agreement was found
proposed in [19], which is a modified model for the one presented in general. Fig. 10 presents the axial load against axial strain curves
by Ottosen [20], was adopted in the model: of octagonal CFT stub columns in the total loading process by using
both FE modelling and experimental results, which also shows a
⎧ kx + (m–1)x2
⎪ good agreement. It can be concluded that the FE model and ma-
x≤1
⎪ 1 + (k –2)x + mx2 terial constitutive adopted in this paper are reasonable.
y=⎨
⎪ x
x>1
⎪ α1(x –1)2 + x 3.3. Composite action analysis
⎩ (2)

where y (¼s/fc) and x (¼ ε/εc) are the stress and strain ratio of the In order to give a clear understanding of the composite action
core concrete respectively. s and ε are the stress and strain of the of octagonal CFT stub columns under axial load, the confinement
core concrete. fc ( ¼0.4fcu7/6) is the uniaxial compressive strength effect of the steel tube on the core concrete is conducted in this
of concrete, where fcu is the compressive strength of standard section.
cubic concrete with 150 mm. εc is the strain corresponding with The axial and transverse stress-strain curves for the steel tube
the peak compressive stress of concrete, where εc ¼383fcu7/ at critical points on the half height of OST1-A (1# and 3# as shown
18

10  6. The parameter k is the ratio of the initial tangent in Fig. 1) are shown in Fig. 11. According to Ding et al. [12], the
modulus to the secant modulus at peak stress. M ( ¼1.6(k  1)2) is stress-strain curve in the steel tube can reflect the strength of
a parameter that controls the decrease in the elastic modulus composite action. If the axial stress-strain curve intersects with
along the ascending branch of the axial stress-strain relationship. the transverse one in the process of loading, the steel tube will
For a concrete-filled steel tubular stub column, parameter α1 is provide a high confinement effect on the core concrete. Therefore,
taken as 0.15. More information of the concrete model could be it is apparent that the confinement effect at endpoint is greater
 F.-x. Ding et al. / Thin-Walled Structures 107 (2016) 453–461 459

10000 350
OST1

Various stress σ (MPa)

OST1-A
280
Axial load N (kN)
7500
axial stress
210
5000
140 (0.02,163)
endpoint 1#
2500 OST1-A mid-point 3#
70 transverse stress
OST1-B average
FE constitutive curve
0 0
0.00 0.01 0.02 0.03 0.04 0.00 0.01 0.02 0.03
Axial strain εL Various strain ε
(a) OST1 Fig. 11. Comparisons of steel stress-strain relationships from calculated results.

12000 60
OST2
OST1-A

Axial stress σL,c (MPa)

Axial load N (kN)

9000
45

6000
30

3000 OST2-A endpoint 1#

OST2-B 15 mid-point 3#
FE average
0 plain concrete
0.00 0.01 0.02 0.03 0.04 0
Axial strain εL 0.00 0.01 0.02 0.03 0.04
(b) OST2 Axial strain εL,c
14000
OST3 Fig. 12. Comparison of axial stress-strain relationship.
Axial load N (kN)

10500 3
Radial stress σr,c (MPa)

OST1-A
7000
2

3500 OST3-A endpoint 1#

OST3-B 1 mid-point 3#
FE
0
0.00 0.01 0.02 0.03 0.04
0
Axial strain εL 0.000 0.005 0.010 0.015 0.020
(c) OST3 Axial strain εL,c
14000 Fig. 13. Comparison of radial stress-axial strain relationship for different measured
OST4 point.
Axial load N (kN)

10500
endpoint and mid-point on the half height of OST1-A. It is obvious
that the compressive strength of core concrete at any points is
7000 much greater than the uniaxial compressive strength of prism
volume without lateral confinement. That is to say that the core
3500 concrete in octagonal CFT columns is subjected to the effective
OST4-A
lateral constraint from the outer steel tube. Moreover, the com-
OST4-B
FE pressive strength at endpoint (1#) is higher than that at mid-point
0 (3#), because of the larger lateral compressive stress on the end-
0.00 0.01 0.02 0.03 0.04 point concrete. The comparison of lateral compressive stress dis-
Axial strain εL tribution at both endpoint and mid-point is given in Fig. 13.
(d) OST4
Fig. 10. Comparison of load-axial strain curves of specimens between FE modelling
and experimental results (a) OST1 (b) OST2 (c) OST3 (d) OST4. 4. Practical formula for load bearing capacity

4.1. Model simplification and formulation

than that at mid-point for octagonal CFT stub columns. Almost 62%
of ultimate capacity of steel tube is also used to sustain the axial
The stress contour at the ultimate state for specimens OST1-A is
load after the steel tube yield. extracted from the FE model and shown in Fig. 14(a). According to
Fig. 12 shows axial stress-strain curves of core concrete at the above figure, the stress distribution of the octagonal CFT
460 F.-x. Ding et al. / Thin-Walled Structures 107 (2016) 453–461

Ratio of axial stress at ultimate state

1.0

to yeild strength σL,S / fy

0.8
avg=0.77
0.6

0.4

0.2
mid-points
end points
0.0
40 80 120 160 200
Ultimate strength f (MPa)
sc

Fig. 15. Average ratio of axial compressive stress to yield stress of steel tube.

steel paired with C90 and C120 concrete; (6) 18 of optimized nu-
merical specimens are chosen in this study. Two points on the
steel tube are investigated: endpoint and mid-point of middle
section.
It can be seen from Fig. 15 that when octagonal CFT stub col-
umns reach their ultimate strength (fsc ¼Nu/Asc, where Asc ¼ Ac þ As
is the total area of cross-section), the average ratio of the axial
compressive stress to the yield stress of steel tube is:
σL, s = 0.77f y (5)

Based on Von Mises yield criterion mentioned in [11], the

tensile transverse stress (sθ,s) of the steel tube can be obtained as:
σθ, s = 0.36f y (6)

As shown in Fig. 14(b), the relationship between the radial

concrete stress (sr,c) of the core concrete and the transverse stress
of the steel tube (sθ,s) can be expressed as:

2t
σr, c = σθ, s
b (7)
Fig. 14. Division on stress region of cross section (a) stress contour of FE resulting
(b) simplified stress distribution. and the axial compressive strength of core concrete can be ex-
pressed as follow:
columns can be simplified as Fig. 14(b). This is completely based on
σL, c = fc + pσr, c (8)
stress distribution and the superposition principle of the concrete
section at the ultimate state, where Ac1 is the unconstrained area where p is the lateral pressure coefficient reported in reference
of core concrete, Ac2 is the constrained area of core concrete and Ac [11], p ¼3.4.
is the cross-sectional area of core concrete. The length of un- The ultimate bearing capacity (Nu) of axially-loaded octagonal
constrained area is approximately 2/3b, where b is the edge length CFT stub columns can therefore be illustrated as:
of core concrete. In this way, the following relations can be ex-
Nu = σL, cAc2 + fc Ac1 + σL, sAs (9)
pressed:
⎧ A c1 + A c2 = A c where As is the cross-sectional area of steel tube, As ¼8bt.
⎪ Substituting (Eq. (4)–8) into Eq. (9), Nu can be obtained as:
⎨ A c1 = 0.3A c
⎪ Nu = fc Ac + 1.5f y As
⎩ A c2 = 0.7A c (4) (10)
2
where Ac ¼ 2b tan 67.5°. where the value of 1.5 is the confinement coefficient of octagonal
The validated FE model is further used for parametric study to CFT stub columns, which falls in the range between 1.2 for square
understand the effect of steel and concrete strength (fy, fcu) on the CFT columns [12] and 1.7 for circular CFT columns [11].
mechanical performance of octagonal CFT stub columns. The
parameters used for the study are detailed as below: (1) The wall 4.2. Validation
thicknesses of steel tube are set as 6 mm, 11 mm, 16 mm; (2) The
edge lengths of cross-section of octagonal CFT columns are The load bearing capacity obtained by using Eq. (10) (Nu, Eq.10),
200 mm; (3) The depth of the columns are selected as 1500 mm; FE results (Nu, FE), unified calculation from [15] (Nu,[15]) and ex-
(4) The adopted compressive strength of cubic concrete is 40, 70, perimental results (Nu, Exp) for all octagonal CFT stub columns from
90 and 120 MPa; (5) Different yield strengths of steel 235, 345 and this paper and reference [14] is all listed in Table 2. It can be seen
420 MPa are used. The following steel and concrete are paired for that an average ratio of 0.967 with a standard deviation of 0.005 is
the octagonal CFT columns: Q235 steel paired with C40 and C70 obtained by comparing FE models with experimental results.
concrete, Q345 steel paired with C70 and C90 concrete, and Q420 Furthermore, average ratios of 0.907 and 1.128, with standard
 F.-x. Ding et al. / Thin-Walled Structures 107 (2016) 453–461 461

deviations of 0.058 and 0.044 are obtained by comparing the ex- References
perimental results with predictions from reference [15] and Eq.
(10) in this paper respectively. It is therefore seen that the pro- [1] S. Morino, K. Tsuda, Design and construction of concrete-filled steel tube
posed equation in this paper is much simpler in use, and ensured column system in Japan, Earthq. Eng. Eng. Seismol. 4 (1) (2003) 51–73.
[2] M.F. Hassanein, O.F. Kharoob, Analysis of circular concrete-filled doubled skin
same accuracy level as the formula given by [15].
tubular slender columns with external stainless steel tube, Thin-Walled Struct.
79 (6) (2014) 23–37.
[3] M.F. Hassanein, O.F. Kharoob, Q.Q. Liang, Circular concrete-filled doubled skin
5. Conclusion tubular short columns with external stainless steel tubes under axial com-
pression, Thin-Walled Struct. 73 (12) (2013) 252–263.
[4] M.F. Hassanein, Numerical modeling of concrete-filled lean duplex slender
This paper presents a combined experimental, numerical and stainless steel tubular stub columns, J. Constr. Steel Res. 66 (8–9) (2010)
theoretical study on the mechanical performances of octagonal 1057–1568.
[5] X. Chang, L. Fu, H.B. Zhao, Y.B. Zhang, Behaviors of axially loaded circular
CFT stub columns subjected to axial compressive loading. Para-
concrete-filled steel tube (CFT) stub columns with notch in steel tubes, Thin-
metric studies have also been conducted in order to understand Walled Struct. 73 (12) (2013) 273–280.
the influence of some key parameters on the composite action [6] X. Chang, C.K. Huang, Y.J. Chen, Mechanical performance of eccentrically loa-
between the steel tube and the core concrete and structural be- ded pre-stressing concrete filled circular steel tube columns by means of ex-
pansive cement, Eng. Struct. 31 (11) (2009) 2588–2597.
havior of the octagonal CFT stub columns. Based on the results, the
[7] X. Chang, H.X. Lin, Experimental study on shear resistance of self-stressing
following conclusions can be drawn: concrete filled circular steel tubes, J. Constr. Steel Res. 65 (4) (2009) 801–807.
[8] K. Uenaka, Experimental study on concrete filled elliptical/oval steel tubular
1) The experimental results indicate that increasing the strength stub columns under compression, Thin-Walled Struct. 78 (5) (2014) 131–137.
[9] K. Uenaka, H. Tsunokake, Concrete filled elliptical steel tubular members with
grade of concrete will increase the load bearing capacity of the large diameter-to-thickness ratio subjected to bending, Structures 5 (2) (2016)
octagonal CFT stub columns. However, it will decrease the 58–66.
ductility. Besides, increasing the steel ratio does not improve the [10] Z. Tao, B. Uy, F.Y. Liao, L.H. Han, Nonlinear analysis of concrete-filled square
stainless steel stub columns under axial compression, J. Constr. Steel Res. 67
load bearing capacity of the octagonal CFT columns significantly,
(11) (2011) 1719–1732.
while it has a remarkable benefit on the ductility of them. As [11] F.X. Ding, Z.W. Yu, Y. Bai, Y.Z. Gong, Elasto-plastic analysis of circular concrete-
well as the distribution of the strain ratio of the steel tube de- filled steel tube stub columns, J. Constr. Steel Res. 67 (10) (2011) 1567–1577.
monstrates that the octagonal steel tube has much more con- [12] 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.
finement effect on the core concrete at the corner than that at
Constr. Steel Res. 98 (11) (2014) 146–157.
the center. [13] F.X. Ding, L. Fu, Z.W. Yu, G. Li, Mechanical performances of concrete-filled steel
2) A 3D finite element model is established to simulate the tubular stub columns with round ends under axial loading, Thin-Walled
structural behavior of octagonal CFT stub columns subjected to Struct. 97 (12) (2015) 22–34.
[14] M. Tomii, K. Yoshimura, Y. Morishita, Experimental studies on concrete-filled
compression loading based on a reasonable concrete con- steel tubular stub columns under concentric loading. International Colloquium
stitutive model in which the constrained effect from the steel on Stability of Structures Under Static and Dynamic Loads, Washington DC,
tube on the core concrete is considered. The numerical results May 17–19, 1977, pp. 718–741.
[15] M. Yu, X.X. Zha, J.Q. Ye, Y. Li, A unified formulation for circle and polygon
are in good agreement with the experimental results.
concrete-filled steel tube columns under axial compression, Eng. Struct. 49 (4)
3) The analytical results on the composite action between the steel (2013) 1–10.
tube and the core concrete show that there exists a significant [16] GB/T50081-2002, Standard for method of mechanical properties on ordinary
confinement effect within the whole section of core concrete, concrete, China Building Industry Press, Beijing, 2002.
[17] GB50017-2003, Code for Design of Steel Structures, China Planning Press,
especially in position at the corner. Beijing, 2003.
4) A simplified approach is proposed to predict the ultimate [18] L.H. Han, Tests on stub columns of concrete-filled RHS sections, J. Constr. Steel
bearing capacity of the octagonal CFT stub column. The pre- Res. 58 (3) (2002) 353–372.
dicted result is reliable, and is therefore recommended for de- [19] 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,
signing purpose. Moreover, the confinement coefficient of oc- Front. Archit. Civ. Eng. China 5 (3) (2011) 381–393.
tagonal CFT stub columns is 1.5 as expected, since it falls be- [20] N.S. Ottosen, M. Ristinmaa, 12-Common Plasticity Models The Mechanics of
tween that of the square and circular CFT columns. Constitutive Modeling, (2005) 279–319.
[21] Hibbitt, Karlsson and Sorensen, Inc., ABAQUS User's Manuals, Version 6.10.
Rhode Island, USA, 2010.
[22] F.X. Ding, Z.W. Yu, Strength criterion for plain concrete under multiaxial stress
Acknowledgment based on damage Poisson's ratio, Acta Mech. Solid. Sin. 19 (4) (2006) 307–315.
[23] T. Jankowaik, T. Lodygowski, Identification of parameters of concrete damage
plasticity constitutive model, Found. Civ. Environ. Eng. 6 (2005) 53–69.
This research work is financially supported by the National [24] J.S. Liu, H.J. Liu, ABAQUS finite element analysis of reinforced concrete, Equip.
Natural Science Foundation of China, Grant No. 51578548. Manuf. Technol. 6 (2009) 69–70 (in Chinese).