You are on page 1of 14

Journal of Constructional Steel Research 66 (2010) 542555

Contents lists available at ScienceDirect

Journal of Constructional Steel Research


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

Analytical behaviour of concrete-filled double skin steel tubular (CFDST)


stub columns
Hong Huang a , Lin-Hai Han b, , Zhong Tao c , Xiao-Ling Zhao d
a

College of Civil Engineering and Architecture, East of China Jiao Tong University, Jiangxi, 330013, PR China

Department of Civil Engineering, Tsinghua University, Beijing 100084, PR China

College of Civil Engineering, Fuzhou University, Fuzhou, Fujian Province 350108, PR China

Department of Civil Engineering, Monash University, Clayton, VIC 3168, Australia

article

info

Article history:
Received 13 June 2009
Accepted 30 September 2009
Keywords:
Concrete-filled double skin tubes (CFDST)
Axial compression
FE modelling
Composite action
Concrete
Hollow steel tubes
Sectional capacity

abstract
This paper reports a finite element analysis of the compressive behaviour of CFDST stub columns with
SHS (square hollow section) or CHS (circular hollow section) outer tube and CHS inner tube. A set of test
data reported by different researchers were used to verify the FE modelling. Typical curves of average
stress versus longitudinal strain, stress distributions of concrete, interaction of concrete and steel tubes,
as well as effects of hollow ratio on the behaviour of CFDST stub columns, were presented. The influences
of important parameters that determine sectional capacities of the composite columns were investigated.
2009 Elsevier Ltd. All rights reserved.

1. Introduction
Concrete-filled double skin steel tubular (CFDST) members are
composite members which consist of an inner and outer steel skin
with the annulus between the skins filled with concrete. This type
of sandwich cross-section was shown to have high bending stiffness that avoids instability under external pressure. Some background information can be found in [1].
In recent years, many studies have been performed on CFDST
stub columns, such as [213]. A state-of-the-art review was given
by Zhao and Han [1]. A summary of research conducted on
CFDST stub columns is presented in Table 1. It can be seen from
Table 1 that the past studies concentrate mainly on experimental
investigations or predicting the load-bearing capacities of stub
columns.
According to Han et al. [3] and Tao et al. [6], hollow ratio is
an important parameter that affects column behaviour. This ratio
is defined as d/(D 2tso ), where d and D are the major dimensions
of the inner and outer tubes, respectively, and tso is the thickness
of the outer tube. If hollow ratio is equal to 0 for a column, the

Corresponding author. Tel.: +86 10 62787067; fax: +86 10 62781488.


E-mail address: lhhan@mail.tsinghua.edu.cn (L.-H. Han).

0143-974X/$ see front matter 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jcsr.2009.09.014

column is actually a conventional concrete-filled steel tube (CFST).


Generally, the CFDST columns have almost all the same advantages
as conventional CFST members.
In this paper, a finite element (FE) modelling was developed
based on the commercial FE package, ABAQUS [14], to study the
compressive behaviour of CFDST stub columns. Several key issues
in the FE modelling are introduced briefly, i.e. the material models
for concrete and steel, interface model to simulate the concrete and
steel interface, element type, mesh, and boundary conditions.
For CFDST columns, there are four possible combinations of
square hollow section (SHS) and circular hollow section (CHS) as
outer or inner tubes. Since a CHS is less susceptible to local buckling
than a SHS, it is good to use CHSs as both inner and outer tubes for
a CFDST in practice. However the beamcolumn joint for a square
column is easier to be fabricated and installed compared with that
of a circular column. For this reason, two types of CFDST columns,
i.e., section with CHS inner and CHS outer, and section with CHS
inner and SHS outer are investigated.
The main objectives of this paper are threefold: first, a set of
test results reported by different researches are used to verify
the FE modelling. Second, typical curves of average stress versus
longitudinal strain, stress distributions of concrete, interaction of
concrete and steel tubes and hollow ratio effect are investigated.
Third, the influence of important parameters that determine the
sectional capacities of the composite columns is identified.

H. Huang et al. / Journal of Constructional Steel Research 66 (2010) 542555

Nomenclature
Ac
Ace
Asco
Asc
Asi
Aso
CFDST
CFST
d
D
fck
fcu
fc0
fsyi
fsyo
N
Nu
Nuc
Nue
p1
p2
tso
tsi

bond

Cross-sectional area of concrete


Nominal cross-sectional area of concrete
Cross-sectional area of the outer steel tube and the
sandwich concrete (=Aso + Ac )
Cross-sectional area of CFDST (=Aso + Ac + Asi )
Cross-sectional area of inner steel tube
Cross-sectional area of outer steel tube
Concrete-filled double skin tube
Concrete-filled steel tube
Outer diameter of inner steel tube
Outer dimension of outer steel tube
Characteristic concrete strength (fck = 0.67fcu for
normal strength concrete)
Characteristic 28-day concrete cube strength
Concrete cylinder strength
Yield strength of inner steel tube
Yield strength of outer steel tube
Axial compressive load
Ultimate strength of CFDST stub column
Predicted ultimate strength of CFDST stub column
by using FE modelling
Experimental ultimate strength of CFDST stub
column
Interaction stress between the concrete and outer
tube
Interaction stress between the concrete and inner
tube
Wall thickness of outer steel tube
Wall thickness of inner steel tube
Nominal steel ratio, given by n = Aso /Ace
Hollow ratio, given by d/(D 2tso )
Strain
Coefficient of friction between the steel tube and
core concrete
Bond strength between the steel tube and core
concrete
Confinement factor (=n fsyo /fck )

2. Finite element modelling


2.1. Material models
(1) Steel
A steel constitutive model for structural steel presented in [15]
is utilised to represent uniaxial stressstrain relation of steel. For

543

carbon steel tubes, an elasticplastic stressstrain relation model,


consisting of five stages (i.e. elastic, elasticplastic, plastic, hardening and fracture) is used. More details of the stressstrain relationship can be found in [15]. Mises yield function with associated
plastic flow is used in the multiaxial stress states.
The steel is assumed to have isotropic hardening behaviour, i.e.,
the yield surface changes uniformly in all directions so that yield
stresses increase or decrease in all stress directions when plastic
straining occurs [14]. Elastic modulus (Es ) and Poissons ratio for
steel are taken as 2 105 (N/mm2 ) and 0.3, respectively.
(2) Concrete
Concrete is a brittle material with different failure mechanism
in compression and tension, i.e., crushing in compression and
cracking in tension. The damage plasticity model defined in
ABAQUS is used in the analysis [14]. The concrete damage plasticity model adopts a unique yield function with non-associated
flow and a DruckerPrager hyperbolic flow potential function to
describe the plasticity of concrete. Therefore, independent uniaxial
stressstrain relations for concrete both in compression and tension are the basic input data due to the difference in strength and
failure mechanism in compression and tension.
It is expected that the inner tube can restrict the inner indent
of the concrete core if the hollow ratio is not too large, so the
sandwich concrete in the gap has the same behaviour with that
in a fully in-filled steel tube without the inner void. It was found
that in this case the failure features of the CFDST specimens were
very similar to those of CFST columns [3,6]. Therefore, uniaxial
stressstrain relation for concrete in CFSTs is used for the analysis
of CFDST members in this paper. The increasing of the plasticity
of core concrete as a result of the passive confinement of the
steel tube depends on the confinement factor [1517]. The
confinement factor for a CFDST can be defined as:

= n

fsyo
fck

(1)

in which, n is the nominal steel ratio of CFDST columns, which is


given by n = Aso /Ace . Ace is the nominal cross-sectional area of
concrete, which is given by Ace = 4 (D 2tso )2 for section with
CHS inner and CHS outer, and Ace = (D 2tso )2 for section with
CHS inner and SHS outer. Aso is the cross-sectional area of outer
steel tube, fsyo is the yield stress of outer steel tube, and fck is the
characteristic compression strength of concrete. The value of fck is
approximately equal to 67% of the compressive strength of cube
blocks (fcu ) for normal strength concrete.
An equivalent stressstrain model presented by Han et al. [17],
which is suitable for the FE analysis using ABAQUS software for
CFSTs, is used in this paper for the analysis of CFDSTs. Fracture
energy versus displacement cross crack relation is used to describe

Table 1
Summary of research conducted on CFDST stub columns.
Researchers

Combinations

Research results

Wei et al. [8,9]


Lin and Tsai [4]
Zhao et al. [10]
Tao et al. [6]

CHS outer and CHS inner

Zhao and Grzebieta [13]


Zhao et al. [11]

SHS outer and SHS inner

Test results; Theoretical models are developed to predict the ultimate strength.
Test results; Plastic mechanism methods are used to predict the unloading behaviour.

Test results; An analytical model is presented, and an empirical formula is presented for the peak strength.
Test results.
Test results; Mechanics models and simplified models are developed.

Elchalakani et al. [2]

CHS outer and SHS inner

Test results; A simplified formula is derived to determine the compressive capacity.

Han et al. [3]


Zhao et al. [12]

SHS outer and CHS inner

Test results; Mechanics models and simplified models are developed.


Test results; Theoretical models are developed to predict the ultimate strength.

Tao et al. [7]


Tao and Han [5]

RHS outer and RHS inner

Test results; Mechanics models are developed.

544

H. Huang et al. / Journal of Constructional Steel Research 66 (2010) 542555

(a) Circular section.

(b) Square section.


Fig. 1. A schematic view of the element divisions.

Fig. 2. A schematic view of boundary conditions.

the tensile behaviour of concrete. More details of the model can be


found in [17].
The initial modulus of elasticity (Ec ) and Poissons ratio (c )
of concrete are determined according to the recommendations
in
p
ACI Committee 318 [18], given as Ec = 4730 fc0 and c = 0.2
respectively.
2.2. Element type, element mesh and boundary conditions
The inner and outer steel tubes of a CFDST are modelled by
reduced-integration shell elements (S4R), while the concrete core,
as well as the end plates, are modelled by 8-node brick elements
(C3D8R). The finite element meshes for typical members with
circular and square sections are shown in Fig. 1.
Due to symmetry of loading and geometry, only one eighth of
the CFDST columns are modelled in the analysis. Boundary conditions of a model are shown in Fig. 2. The uniform loading in the z
direction is applied to the top surface of the end plate. Load is simulated by applying displacement instead of directly applying load.
The stiffness of the end plate is large enough that its deformation in
the whole loading process is very little. The end plate connects with
the steel tube by SHELL TO SOLID (an interface model in ABAQUS),
which ensures the displacements and rotational angles of the contact elements keep the same in the whole loading process. The
Hard contact relation is selected for the end plate and concrete.

2.3. Steel-tubeconcrete interface


The model to simulate the interaction of steel and concrete
in CFDST is the contact interaction in ABAQUS [14]. The contact
interaction is defined in two aspects, the geometric property and
the mechanical property.
The geometric property of the contact surfaces is defined by selecting appropriate contact discretization, tracking approach and
determination of master and slave surfaces for the contact [14].
The surface-to-surface contact discretization is used in which two
of the contact surfaces are defined as master and slave surfaces
respectively. Some individual nodes in the master surface may
penetrate into the slave surface; however, large, undetected penetrations of master nodes into the slave surface do not occur
with this discretization. Such penetrations can be further reduced
through careful selection of the master surface and finite element
discretization of contact surfaces. A small sliding tracking approach
is selected for the contact. This approach is more efficient in the
calculation since the actual sliding between steel and concrete surfaces in CFDST is relatively small.
The mechanical property of the contact interaction is defined
along normal and tangential directions to the interface respectively. The Hard contact relation is selected as normal mechanical

H. Huang et al. / Journal of Constructional Steel Research 66 (2010) 542555

545

(1) Circular sections.

(2) Square sections.


Fig. 3. Comparison of predicted versus measured axial load NDeformation curves.

property. This property can be described in a pressureoverclosure


relation, i.e. surfaces transmit no contact pressure unless the nodes
of the slave surface contact the master surface. There is no limitation on pressure development when surfaces are in contact. Furthermore, the contact surfaces are allowed to separate each other

after they have contacted. The tangential mechanical property of


the contact interaction is simulated by an isotropic Coulomb friction model [14]. According to the Coulomb friction model, the surfaces can transfer shear stress until the shear stress is greater than
the limit value (crit ). After the relative slip is formed between the

546

H. Huang et al. / Journal of Constructional Steel Research 66 (2010) 542555

(a) Outer steel tubes of circular sections.

(c) Inner steel tubes of circular sections.

(b) Outer steel tubes of square sections.

(d) Inner steel tubes of square sections.

Fig. 4. Comparisons between predicted and observed typical failure modes of specimens.

surfaces, the shear force is taken as a constant (crit ). More details


of the model can be found in [17].
Up to now, there is no research reported regarding the bond
behaviour of CFDST columns. It is expected that, however, the
behaviour of CFDST stub columns is not sensitive to the bond
between the concrete and the inner or outer steel tube since the
three components are loading together. This is also confirmed by
changing the bond value in certain scope by using the FE modelling
in this paper. Therefore, the bond model used for conventional
CFST columns is also used in this paper to simulate CFDST
columns.

Comparisons between predicted and observed typical failure


modes are presented in Fig. 4, where the failure modes for the
outer CHS and SHS are shown in Fig. 4(a) and (b), respectively, and
Fig. 4(c) and (d) show the failure modes of the inner steel tubes. As
can be seen, the failure modes of the outer steel tubes are outward
buckling occurred near the specimen mid-height, while that for the
inner steel tubes is inward buckling since its outward displacement
is restricted by the concrete. In general, the predicted failure modes
of both inner and outer steel tubes agree well with the observed
ones.
3. Mechanism analysis

2.4. Verification of the FE modelling


3.1. Analysis of the loaddeformation relation
The predicted ultimate strengths (Nuc ) by using the FE modelling are compared with the measured ones (Nue ) taken from Tao
et al. [6], Lin and Tsai [4], Han et al. [3] and Zhao et al. [12], as
shown in Table 2. A mean (Nuc /Nue ) of 0.936 and a COV (coefficient
of variation) of 0.045 for members with circular section and a mean
(Nuc /Nue ) of 1.022 and a COV (coefficient of variation) of 0.061 for
members with square section are obtained. Typical predicted axial
load N versus axial strain or axial displacement curves using FE
modelling are compared with the measured curves in Fig. 3. It can
be found that, in general, good agreement is obtained between the
predicted and test results.

Typical calculated curves of average stress sc (=N /Asc , Asc is the


cross-sectional area of CFDST) versus longitudinal strain is shown
in Fig. 5. Four characteristic points are also marked on the curves.
At Point A, the outer steel tube begins to come into elasticplastic
stage. Yielding of the outer steel tube occurs at Point B. At Point C,
ultimate axial load is reached. At Point D, the longitudinal strain
attains the value of 0.02.
Fig. 6 shows the distributions of longitudinal stress (S33 in the
graphs) at these characteristic points for the sandwich concrete in
the cross-section at the mid-height. The basic parameters used in

H. Huang et al. / Journal of Constructional Steel Research 66 (2010) 542555

547

Table 2
Test data of CFDST stub columns under axial compression.
Test
series

Specimen
label

Outer tube
dimensions
D tso (mm)

Inner tube
dimensions
d tsi (mm)

fsyo (MPa)

fsyi (MPa)

fc0 (fcu ) (MPa)

Nue (kN)

Nuc (kN)

Nuc /Nue

CHS
outer

cc2a
cc2b
cc3a
cc3b
cc4a
cc4b
cc5a
cc5b
cc6a
cc6b
cc7a
cc7b
DS-2
DS-6

180 3
180 3
180 3
180 3
180 3
180 3
114 3
114 3
240 3
240 3
300 3
300 3
300 2
300 4

48 3
48 3
88 3
88 3
140 3
140 3
58 3
58 3
114 3
114 3
165 3
165 3
180 2
180 2

0.28
0.28
0.51
0.51
0.80
0.80
0.54
0.54
0.49
0.49
0.56
0.56
0.61
0.61

275.9
275.9
275.9
275.9
275.9
275.9
294.5
294.5
275.9
275.9
275.9
275.9
290
290

396.1
396.1
370.2
370.2
342.0
342.0
374.5
374.5
294.5
294.5
320.5
320.5
290
290

47.4
47.4
47.4
47.4
47.4
47.4
47.4
47.4
47.4
47.4
47.4
47.4
28
28

1790
1791
1648
1650
1435
1358
904
898
2421
2460
3331
3266
2141
2693

1622
1622
1497
1497
1258
1258
807
807
2337
2337
3195
3195
2155
2765

0.906
0.906
0.908
0.907
0.877
0.926
0.893
0.899
0.965
0.950
0.959
0.978
1.007
1.027

scc2-1
scc2-2
scc3-1
scc3-2
scc4-1
scc4-2
scc5-1
scc5-2
S1C1
S2C1

-120 3
-120 3
-120 3
-120 3
-120 3
-120 3
-180 3
-180 3
-100.2 6.12
-100.4 4.13

32 3
32 3
58 3
58 3
88 3
88 3
88 3
88 3
48.5 3.01
48.5 3.01

0.28
0.28
0.51
0.51
0.77
0.77
0.51
0.51
0.55
0.53

275.9
275.9
275.9
275.9
275.9
275.9
275.9
275.9
500
476

422.3
422.3
374.5
374.5
370.2
370.2
370.2
370.2
425
425

46.8
46.8
46.8
46.8
46.8
46.8
46.8
46.8
70
70

1054
1060
990
1000
870
996
1725
1710
1677
1253

993
993
1020
1020
977
977
1835
1835
1651
1337

0.942
0.937
1.030
1.020
1.123
0.981
1.064
1.073
0.984
1.067

SHS outer

Test data
resources

Tao et al. [6]

Lin and Tsai [4]

Han et al. [3]

Zhao et al. [12]

Stage 3: Plastic stage (from Point B to Point C). During this


stage, due to the increasing of the confinement provided by the
outer steel tube, the average longitudinal stress in the concrete
cross-section exceeds the concrete cylinder strength fc0 . But the
longitudinal stress of the concrete distributes unevenly for both
the two types of column. For the member with CHS inner and CHS
outer, the closer the location to the outer steel tube, the larger the
longitudinal stress is, which can be seen from Fig. 6(1)(c). This is attributed to the fact that the confinement is mainly from the outer
steel tube.

Fig. 5. Typical sc versus relations.

the calculations are: D = 400 mm, tso = 9.3 mm, d = 191 mm,
tsi = 3.18 mm, L = 1200 mm, fsyo = fsyi = 345 MPa, fcu = 60 MPa,
= 0.5, where tsi and fsyi are the wall thickness and yield strength
of the inner steel tube respectively, and L is the column height.
It is clear from Fig. 5 that a curve can be generally divided into
four stages, i.e.
Stage 1: Elastic stage (from Point O to Point A). During this stage,
steel and concrete bear axial load independently. It can be seen
from Fig. 6(1)(a) and (2)(a) that the longitudinal stress of concrete
uniformly distributes across the cross-section on the whole.
Stage 2: Elasticplastic stage (from Point A to Point B). During this
stage, with the increasing of the axial load, the concrete cracks
and begins to increase in volume. The confinement provided by
the outer steel tube enhances as the transverse deformation of
concrete increases. It was found from Fig. 6(1)(b) that for the
circular member the longitudinal stress of concrete distributes still
uniformly in the cross-section, but for the square member shown
in 6(2)(b) the maximal longitudinal stress of concrete occurs at the
corner because of the non-uniform confinement provided by the
outer steel tube.

Stage 4: Descending or hardening stage (from Point C to Point D).


During this stage, the average stress sc of CFDST begins to decrease
if < 0 . If > 0 , no descending stage will occur due to the strain
hardening of outer steel tube and its higher confinement on the
concrete. Based on a parametric analysis, the value of 0 is equal
to 1 or so for members with circular section, and is equal to 4.5 or
so for members with square section. It can be seen from Fig. 6(1)
that, for the CFDST column with CHS outer, the distribution of the
concrete stresses at Point D is similar to those at Points B and C,
although the values of concrete stresses decrease obviously. For
the CFDST column with SHS outer, however, a notable change in
stress distribution is found (see Fig. 6(2)) at different points (B, C
and D) with the maximum stress of concrete occurs near the inner
steel tube at Point D.
Fig. 7 shows the loads carried by the outer steel tube, inner steel
tube, sandwich concrete, and overall CFDST versus longitudinal
strain relations. It can be seen that the sandwich concrete bears the
large part of the load, while the inner steel tube contributes little
to the load bearing of the CFDST.
3.2. Interaction of steel and concrete
Fig. 8 shows the confinement to the sandwiched concrete
provided by the steel tubes, in which p1 is the interaction stress
between the concrete and outer steel tube, and p2 is that between
the concrete and inner steel tube.
Due to the influence of the end plate, the interaction stresses p1
and p2 vary along the tube height as shown in Fig. 9, in which H
is the distance to the top end, and L is the length of the composite

548

H. Huang et al. / Journal of Constructional Steel Research 66 (2010) 542555

(1) Circular section.

(2) Square section.


Fig. 6. The distributions of longitudinal stress of concrete (MPa).

(a) Circular section.

(b) Square section.

Fig. 7. The loads (N) carried by outer steel tube, inner steel tube, sandwich concrete, and CFDST respectively versus longitudinal strain ( ).

H. Huang et al. / Journal of Constructional Steel Research 66 (2010) 542555

(a) Circular section.

549

(b) Square section.

Fig. 8. Confinement to the sandwiched concrete provided by the steel tubes.

(a) Circular section.

(b) Square section.


Fig. 9. The variety of interactions p1 and p2 along the stub column height.

(a) Different locations.

(b) Interaction stress p1 .

(c) Interaction stress p2 .

Fig. 10. Interaction stresses p1 and p2 across the cross-section for a member with square section.

member. In Fig. 9, the stresses are shown when the peak loads are
reached. It should also be noted that, the interaction stresses p1 and
p2 shown in Fig. 9(b) are average values around the cross section.
It can be seen that the influence of the end plate on the interaction
is not significant if H > 0.1L.
For members with circular section, the interaction stresses p1
and p2 are almost constant across the cross-section. But this is
not the case for members with square section. To eliminate the
influence of the local buckling of the outer steel tube formed at
the mid-height, the interaction stresses shown in Fig. 10 are taken

from the section with a distance of L/5 away from the mid-height.
It can be seen from Fig. 10(b) that the stress of p1 at the corner
is much higher, which indicates the confinement provided by the
outer steel tube at the corner is the strongest across the crosssection. The interaction stress p1 is almost equal to 0, if the distance
to the corner is larger than D/7, such as those points from 3 to 8
shown in Fig. 10(a). It can be seen from Fig. 10(c) that the values of
p2 do not vary too much around the cross-section compared with
those of p1 .

550

H. Huang et al. / Journal of Constructional Steel Research 66 (2010) 542555

Fig. 11. Interaction stresses p1 and p2 versus longitudinal stain relations.

For convenience of analysis, the interaction stresses of p1 and p2


are taken from the cross-section at the mid-height in the following
analysis, where the basic parameters used in the examples are:
D = 400 mm, tso = 9.3 mm, d = 191 mm, tsi = 3.18 mm,
L = 1200 mm, fsyo = fsyi = 345 MPa, fcu = 60 MPa, = 0.5.
For the circular CFDST column in the above example, Fig. 11
shows the interaction stresses p1 and p2 versus longitudinal stain
relations. It can be found that the lateral deformation of the
outer steel tube is larger than that of the concrete at the initial
loading stage due to the larger Poissons ratio of the outer steel
tube compared with that of the concrete. Therefore there is no
interaction developed between the concrete and outer steel tube in
this stage. However, it should be noted that the bonding strength
between the concrete and steel tube in the normal direction has
been ignored in the FE modelling. This bonding in reality will allow
small tensile stress developed between the concrete and steel tube.
Since the tensile stress has no significant influence on the overall
performance of the double skin composite columns, it is reasonable
to ignore the tensile stress to simplify the FE analysis. With the
increasing of the longitudinal strain, the cracks in concrete develop,
and the lateral deformation rate of the concrete begins to exceed
that of the outer steel tube. Therefore, the confinement provided
by the outer steel tube occurs at this moment. As far as the
interaction stress p2 is concerned, very small p2 develops at the
initial loading stage since the inner steel tube will press against
the concrete outwardly. But this effect is negligible because the
concrete comes into the elasticplastic state soon, and there is no
interaction between the inner steel tube and the concrete core.
After the peak load is reached, compression will develop once again
at the interface of the inner tube and the concrete.
Possible parameters affecting the interaction stress versus longitudinal stain relationship are hollow ratio ( ), nominal steel
ratio (n ), strength of outer steel tube (fsyo ), strength of concrete
(fcu ), strength of inner steel tube (fsyi ) and width to thickness ratio
of inner steel tube (d/tsi ). Since these parameters have no obvious
influence on the interaction stress p2 , only the effects of these parameters on the p1 versus relations are shown in Fig. 12. It can
be found from this figure: (1) With the increasing of hollow ratio
and concrete strength, the interaction stress p1 decreases due to
the decreasing of the confinement provided by the outer steel tube.
(2) With the increasing of nominal steel ratio and strength of outer
steel tube, the interaction stress p1 increases. (3) The strength and
width to thickness ratio of the inner steel tube have little influence
on the interaction stress p1 .

3.3. Effects of hollow ratio


Hollow ratio ( ) is an important parameter affecting the compressive behaviour of CFDST. Fig. 13 shows the distributions of
longitudinal stress of concrete for CFDST members with different
hollow ratio (i.e. = 0, = 0.25, = 0.5, = 0.75) at peak
loads. It can be seen from Fig. 13(1) that with the increasing of hollow ratio, longitudinal stress of concrete decreases obviously for
member with circular section. If is equal to 0, the maximum concrete stress occurs in the centre of the cross-section. If is equal
to 0.25, the maximum concrete stress occurs in the centre of the
sandwich concrete. If is equal to 0.5 or 0.75, the maximum concrete stress appears near the outer steel tube. As hollow ratio increases, apparently, the location of the maximum concrete stress
moves from centre to the periphery of the cross-section. The value
of the maximum concrete stress also decreases with the increasing
of hollow ratio .
Fig. 13(2) shows that the effect of hollow ratio on the stress distribution for members with SHS outer. As can be seen, the maximum concrete stress occurs at the corner and decreases a bit with
the increasing of hollow ratio. Generally, the influence of hollow
ratio on the concrete stresses for members with CHS outer is larger
than that for members with SHS outer.
3.4. Parametric studies
Possible parameters affecting the axial load (N) versus longitudinal strain ( ) relationship of stub columns are hollow ratio ( ),
nominal steel ratio (n ), strength of outer steel tube (fsyo ), strength
of concrete (fcu ), strength of inner steel tube (fsyi ) and width to
thickness ratio of inner steel tube (d/tsi ). Fig. 14 shows the effects
of these parameters on the N versus relations.
It can be found from Fig. 14(a) that axial strength of the composite columns decreases as hollow ratio ( ) increases. The reason
is that the area of concrete decreases as increases, and the concrete carries the majority of load for a CFDST column as demonstrated in Fig. 7. However, the stiffness at the elasticplastic stage
increases with the increasing of because of the increased steel
ratio.
It can be found from Fig. 14(b)(d) that the axial strength of the
composite columns increases obviously as the nominal steel ratio
(n ), strength of outer steel tube (fsyo ) or strength of concrete (fcu )
increases.
It can be found from Fig. 14(e) and (f) that the strength of inner
steel tube (fsyi ) and width to thickness ratio of inner steel tube
(d/tsi ) have no obvious influence on the strength of the composite

H. Huang et al. / Journal of Constructional Steel Research 66 (2010) 542555

551

(a) Hollow ratio.

(b) Nominal steel ratio.

(c) Strength of outer steel tube.

Fig. 12. Effects of different parameters on p1 versus relations.

member and the shape of curves. The reason is that the inner steel
tube contributes comparatively little to the column strength as
shown in Fig. 7.
Comparing the load versus longitudinal strain curves in Fig. 14
for the two different section types shown in Fig. 1 reveals that the
residual strength after experiencing large deformation is higher for
circular sections.
4. Conclusions
The following conclusions can be drawn based on the limited

research reported in this paper.


(1) Finite element method is used in this paper for the analysis of
CFDST stub columns. A comparison of results calculated using
this modelling shows good agreement with those of the test
results.
(2) Typical curves of average stress versus longitudinal strain
are analysed. The stress distributions of concrete at different
characteristic points are investigated. The average stress versus longitudinal strain relations show strain hardening or an
elasticperfectly-plastic behaviour with bigger confinement

552

H. Huang et al. / Journal of Constructional Steel Research 66 (2010) 542555

(d) Concrete strength.

(e) Strength of inner steel tube.

(f) Width to thickness ratio of inner steel tube.


Fig. 12. (continued)

factor ( ), while for composite sections with smaller , the


curves were of a strain-softening type.
(3) Interaction between the concrete and steel tubes in the composite columns is analysed. It is found that the influence of
hollow ratio on the concrete stress for stub columns with circular section is more significant than that on members with
square section.
(4) Important parameters affecting the axial load (N) versus longitudinal strain ( ) relationship of stub columns are investigated.
It is found that the stiffness at elasticplastic stage of N relations increases with the increasing of hollow ratio. The residual
strength after experiencing large deformation is higher for stub

columns with circular section compared to those with square


section.
Based on the finite element modelling presented in this paper,
further efforts can be made in the future to take the effects of
concrete shrinkage and the concrete viscosity under long-term
loading into account.
Acknowledgements
The study of this paper is supported by the Research Foundation
of the Ministry of Railways and Tsinghua University (RFMOR &

H. Huang et al. / Journal of Constructional Steel Research 66 (2010) 542555

(1) Circular section.

(2) Square section.


Fig. 13. The distributions of longitudinal stress of concrete for CFDST members with different hollow ratios (unit: MPa).

(a) Hollow ratio ( ).


Fig. 14. Effects of different parameters on N versus relations.

553

554

H. Huang et al. / Journal of Constructional Steel Research 66 (2010) 542555

(b) Nominal steel ratio (n ).

(c) Strength of outer steel tube (fsyo ).

(d) Concrete strength (fcu ).

(e) Strength of inner steel tube (fsyi ).

Fig. 14. (continued)

H. Huang et al. / Journal of Constructional Steel Research 66 (2010) 542555

555

(f) Width to thickness ratio of inner steel tube (d/tsi ).


Fig. 14. (continued)

THU) (No. J2008G011), the National Basic Research Program of


China (973 Program) (Grant No. 2009CB623200). The financial
support is greatly appreciated.
References
[1] Zhao XL, Han LH. Double skin composite construction. Progress in Structural
Engineering and Materials 2006;8(3):93102.
[2] Elchalakani M, Zhao XL, Grzebieta R. Tests on concrete filled double-skin (CHS
outer and SHS inner) composite short columns under axial compression. ThinWalled Structures 2002;40(5):41541.
[3] Han LH, Tao Z, Huang H, Zhao XL. Concrete-filled double skin (SHS outer and
CHS inner) steel tubular beamcolumns. Thin-Walled Structures 2004;42(9):
132955.
[4] Lin ML, Tsai KC. Behavior of double-skinned composite steel tubular
columns subjected to combined axial and flexural loads. In: Proceedings
of the first international conference on steel & composite structures. 2001.
p. 114552.
[5] Tao Z, Han LH. Behaviour of concrete-filled double skin rectangular steel
tubular beamcolumns. Journal of Constructional Steel Research 2006;62(7):
63146.
[6] Tao Z, Han LH, Zhao XL. Behaviour of concrete-filled double skin (CHS inner
and CHS outer) steel tubular stub columns and beamcolumns. Journal of
Constructional Steel Research 2004;60(8):112958.
[7] Tao Z, Han LH, Zhao XL. Tests on stub columns of concrete filled double
skin rectangular hollow sections. In: Proceedings of the 4th international
conference on thin-walled structures. 2004. p. 88592.

[8] Wei S, Mau ST, Vipulanandan C, Mantrala SK. Performance of new sandwich
tube under axial loading: Experiment. Journal of Structural Engineering ASCE
1995;121(12):180614.
[9] Wei S, Mau ST, Vipulanandan C, Mantrala SK. Performance of new sandwich
tube under axial loading: Analysis. Journal of Structural Engineering ASCE
1995;121(12):181521.
[10] Zhao XL, Grzebieta RH, Elchalakani M. Tests of concrete-filled double skin CHS
composite stub columns. Steel and Composite StructuresAn International
Journal 2002;2(2):12946.
[11] Zhao XL, Han BK, Grzebieta RH. Plastic Mechanism analysis of concrete filled
double skin (SHS inner and SHS outer) stub columns. Thin-Walled Structures
2002;40(10):81533.
[12] Zhao XL, Grzebieta RH, Elchalakani M. Tests of concrete-filled double skin and
(SHS outer and CHS inner) composite stub columns. In: Proceedings of the
third international conference on advances in steel structures. Vol. 1. 2002.
p. 56774.
[13] Zhao XL, Grzebieta RH. Strength and ductility of concrete filled double skin
(SHS inner and SHS outer) tubes. Thin-Walled Structures 2002;40(2):199233.
[14] Hibbitt, Karlson & Sorensen Inc. ABAQUS/standard users manual, version
6.4.1. Pawtucket (RI): Hibbitt, Karlsson, & Sorensen, Inc.; 2003.
[15] Han LH. Concrete-filled steel tubular structures-theory and practice. Beijing
(China): Science Press; 2007 [in Chinese].
[16] Han LH, Yao GH, Zhao XL. Tests and calculations of hollow structural steel
(HSS) stub columns filled with self-consolidating concrete (SCC). Journal of
Constructional Steel Research 2005;61(9):124169.
[17] Han LH, Yao GH, Tao Z. Performance of concrete-filled thin-walled steel tubes
under pure torsion. Thin-Walled Structures 2007;45(1):2436.
[18] ACI 318-02. Building code requirements for reinforced concrete and commentary. Farmington Hills (MI, Detroit, USA): American Concrete Institute; 2002.