Structures 3 (2015) 211–226

Contents lists available at ScienceDirect

Structures

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

Performance of concrete-encased CFST box stub columns under

axial compression
Yu-Feng An a, Lin-Hai Han a,⁎, Charles Roeder b
a
Department of Civil Engineering, Tsinghua University, Beijing, 100084, PR China
b
Department of Civil Engineering, University of Washington, Seattle, WA 98195-2700, USA

a r t i c l e i n f o a b s t r a c t

Article history: A finite element analysis (FEA) model is developed to predict the full range response of concrete-encased CFST
Received 4 March 2015 box stub columns under axial compression. In the model, concrete across the composite section is divided into
Received in revised form 10 May 2015 four regions, i.e. the outer unconfined concrete outside the stirrup, the outer concrete in the web walls, the
Accepted 15 May 2015
outer confined concrete in the corners, and the core concrete in the steel tubes. Different material models are
Available online 3 June 2015
used in each region. The analytical results are compared to past experiments, and generally good agreement
Keywords:
between the predicted and measured results is obtained. Load-axial strain, loading distribution between the
Concrete-filled steel tube (CFST) inner CFST and outer RC components and interface stresses between steel and concrete are analyzed. The influ-
Concrete-encased ence of the web wall slenderness is also investigated. Parameter studies investigate the influence of concrete
Box section and steel strength, steel ratio of CFST, longitudinal bar ratio, stirrup spacing and ratio of the diameter of CFST
Stub column to the sectional width on the ultimate load. A simplified model is proposed to predict the ultimate load of the
FEA model concrete-encased CFST box stub columns under axial compression.
Ultimate load © 2015 The Institution of Structural Engineers. Published by Elsevier Ltd. All rights reserved.

1. Introduction magnification method was proposed to consider second-order effects of

Concrete-encased concrete-filled steel tube (CFST) box members are
a newly developed steel–concrete composite system that has been used
in bridge piers and arches in bridges in China [5]. Fig. 1(a) shows a typ-
ical cross-section of this composite member. The member has an
encased CFST component in each of the four corners of the box as well
as an outer reinforced (RC) box component. The box is constructed in
four stages as illustrated in Fig. 1(b). First, the steel tubes are fixed at
the proper location, and then high strength core concrete is placed
into the steel tubes. The outer reinforcing cage is then placed and tied,
and finally the normal strength outer concrete is poured.
Considerable research has been completed for conventional RC
box member (e.g. [18,19,25]) and concrete-encased CFST member
(e.g. [11,4,2]). This research provides a basic understanding of concrete-
encased CFST box member behavior. An et al. [5,4] experimentally and
analytically studied eccentrically loaded concrete-encased CFST box
columns, since they were mainly subjected to combined axial load and
bending moment as piers or arches. It was found that the crushing and
spalling of the concrete in compression and concrete cracking in tension
influenced the behavior of the eccentrically loaded columns. An increase
in eccentricity and height-to-width ratio led to a decrease in the ultimate
load capacity. The strength and stiffness of the composite columns were
larger than those of the corresponding RC box columns. The moment
⁎ Corresponding author. Tel./fax: +86 10 62797067.
E-mail addresses: lhhan@tsinghua.edu.cn, lhhanqw@gmail.com (L.-H. Han).
the eccentrically boxed columns.
The performance of concrete-encased CFST box stub columns under
axial compression is different than that achieved under eccentric load,
since there is no influence of bending and lateral deflection. The ulti-
mate strength of the stub composite column under axial compression
provides a critical point in the compression–bending interaction curves
for these structural members. Concrete-encased CFST box members are
fundamentally different than pure CFST members, because there is a
much wider range of confinement of the concrete in the box section
as compared to the high degree of confinement in the inner CFST com-
ponent. Four concrete zones (the outer unconfined concrete outside the
stirrup, the outer concrete in the web walls, the outer confined concrete
within the stirrups in the corners, and the core concrete in the steel
tubes) are noted in these members and the stress distribution between
these zones is considered in this research. These differences introduce a
range of different potential failure modes, and information on the stress
distribution and confinement effects in various parts of the box section
are important. Within this framework, the influence of wall slenderness
ratio λw (=bc / tc, where bc and tc are the width and thickness of web
walls as shown in Fig. 1(a)) is a critical parameter for RC box columns
[18] and in this study, for the composite box columns.
This paper analyzes the performance of concrete-encased CFST box
stub columns under axial compression. A 3-dimensional finite element
analysis (FEA) model is developed. The model includes different mate-
rial modeling techniques for the concrete in the four concrete regions
noted earlier. The model is verified by comparison of predicted behavior
and experimental results. The analysis evaluates load–strain relations,

http://dx.doi.org/10.1016/j.istruc.2015.05.001
2352-0124/© 2015 The Institution of Structural Engineers. Published by Elsevier Ltd. All rights reserved.
212 Y.-F. An et al. / Structures 3 (2015) 211–226

performance. A superposition method to predict the ultimate strength

Nomenclature
of the composite stub box columns is proposed.
Acore Cross-sectional area of core concrete in CFST
2. Finite element analysis (FEA) model
Al Cross-sectional area of longitudinal bar
As Cross-sectional area of steel tube
The proposed finite element analysis (FEA) model of concrete-
Asc Cross-sectional area of CFST (=Acore + As)
encased CFST box stub column under axial compression is shown in
Aout Cross-sectional area of outer concrete
Fig. 2. The model utilizes the ABAQUS/Standard module software [14],
B Sectional width of concrete-encased CFST box section
and this model is based on the previous analytical work of concrete-
bc Width of web wall
encased CFST stub columns provided by Han and An [11] and An et al.
D Overall diameter of steel tube
[3]. However, the box section has different concrete regions and con-
Di Diameter of core concrete
finement conditions, which increase the complexity of the analysis of
Ec Concrete modulus of elasticity
the concrete-encased CFST box column.
fck Characteristic concrete strength
fcu Concrete cube strength
2.1. Element type, mesh, boundary conditions and interface model
fc′ Concrete cylinder compressive strength
fyh Yielding strength of stirrup
Four-node conventional shell elements, 2-node truss elements and
fyl Yield strength of longitudinal bar
8-node 3-D solid elements are used for the steel tube, reinforcing
fys Yield strength of steel tube
bar and concrete, respectively. The structured meshing technique is
N Axial load
assigned to have a proper element shape. In order to achieve reliable
Ncfst Strength of CFST component
results with reasonable computation times, a mesh convergence study
Nuc Predicted ultimate strength
is performed to identify an appropriate mesh size. One quarter of
Nue Experimental ultimate strength
the column is modeled due to symmetry of loading and geometry as
Nuc,cfst Predicted ultimate strength of CFST component
shown in Fig. 2. The load is simulated by applying uniform displacement
Nuc,rc Predicted ultimate strength of RC component
to one end plate, while the displacement and rotation of the other end
P1 Contact stress between steel tube and core concrete
plate are fixed. The contact between the steel tube and concrete surfaces
P2 Contact stress between outer concrete and steel tube
is modeled by the contact surface model with “Hard contact” in the
s Stirrup spacing
normal direction and the Mohr–Coulomb friction model in tangential
DI Strength index
directions. When the shear stress is smaller than the shear limitation
t Wall thickness of steel tube
(the bond stress), there is no relative slip between the two surfaces,
tc Thickness of web wall
but the shear force is calculated by a friction coefficient and contact
αs Steel ratio of CFST (=As / Acore)
pressure after initial slip. The frictional coefficient was taken as 0.6
αl Longitudinal bar ratio (=Al / (Al + Aout))
and the bond stress between the steel tube and the concrete was deter-
ε Strain
mined according to Han and An [11]. The bond stress is defined as
σ Stress
follows which is based on the study of Roeder et al. [20].
λw Wall slenderness ratio (=bc / tc)
A f
ξ Confinement factor (¼ Acoresf ys ) of CFST
 
ck;core
Di 
τbond ¼ 2:314−0:0195 N=mm2 ð1Þ
t

the strength distribution between the inner CFST component and outer
RC component, the interaction between steel and concrete, and the
stress distribution in various elements of the box member. The influ-
ence of wall slenderness ratio is examined, and a parametric study
is performed to examine key parameters that affect the box member
where Di and t are diameter of core concrete and thickness of the steel
tube.
Similar to Han and An [11], the embedded element technique is used
between the rebar and the outer concrete, and the translational degrees
of freedom at the rebar node are restricted.

Web wall region

Corner
region

Core concrete

Stirrup
Longitudinal bar
bc

Steel tube tc

(a) Cross-section (b) Constructing process

Fig. 1. A schematic view of the concrete-encased CFST box column.
 Y.-F. An et al. / Structures 3 (2015) 211–226 213

End plate Axial load , dx=dy=0

The outer unconfined The outer concrete in

concrete outside the stirrup the web wall

300
Steel tube 2
The outer confined 1
300
concrete in the corner Core concrete

Stirrup
Longitudinal bar
350 350 300

1000
z End plate
1-Symmetry about the yz plane (x displacement and x rotation restricted) o x
y dx=dy=dz=0
2-Symmetry about the xz plane (y displacement and y rotation restricted)
rx=ry=rz=0
(a) Cross-section (one quarter) (b) Schematic illustration

Fig. 2. FEA model of the concrete-encased CFST box stub column.

2.2. Steel models
The elastic–plastic material model is used for the rebar and steel
tube. A five-stage stress–strain model introduced by Han et al. [13]
is used to describe the uniaxial stress–strain relations of the steel tube.
A bi-linear model considering strain hardening effect and the hardening
stiffness 0.01 Es introduced by Han and An [11] are used to describe the
uniaxial stress–strain relations of the rebar. The elastic modulus and
Poisson's ratio of the rebar and steel tube are taken as 206,000 N/mm2
and 0.3, respectively.
strength fcu,out is 40 N/mm2 and the peak strength of plain concrete (fc′)
is 33 N/mm2 in these RC box column models. The uniaxial compressive
stress–strain relation of unconfined concrete provided by Attard and
Setunge [6] is used in the models as shown in Eq. (2). The model is
appropriate to discuss the relative stress increase of outer concrete
and the confined regions, and the comparison results are used to refine
the material modeling in various parts of the concrete.
AX þ BX 2
Y¼ ð2Þ
1 þ CX þ DX 2

2.3. Concrete division and models where X ¼ ε=ε0 ; Y ¼ σ=σ 0 and σ0 and ε0 are the peak stress and the
corresponding strain;.
The damaged plasticity model is used for the concrete, which is brit- 0
0 4:26 f c
tle material with crushing in compression and cracking in tension. This σ 0 ¼ f c ; ε0 ¼ qffiffiffiffiffi ;
4 0
constitutive model adopts a unique yield function with a non-associated Ec f c
flow rule and a Drucker–Prager hyperbolic flow potential function, and
it is capable of simulating concrete behavior under multiple stress states 0
when ε≤ε0 , A ¼ Ec ε0 =f c ; B ¼ ðA−1Þ2 =0:55−1; C ¼ A−2; D ¼ B þ 1;
when the confining stress is less than four to five times the magnitude of f i ðεi −ε0 Þ
when ε N ε0 ; A ¼ ;B
¼ 0; C ¼ A−2; D ¼ 1;
the compressive strength of the concrete (fc′) [14]. The elastic modulus εi ε0 ð f c − f i Þ
0

qffiffiffiffiffi
0 0
and Poisson's ratio of concrete are taken as 4730 f c (N/mm2) and 0.2,
respectively. For the concrete-encased CFST column, three different
concrete models are considered (i.e. outer unconfined concrete outside
the stirrup, concrete confined within the stirrups and the core concrete
in the steel tube) based upon the confinement conditions [11]. The
uniaxial stress–strain relations of these three different kinds of concrete
are different.
However, the division of concrete in the concrete-encased CFST box
column is more complicated than that in concrete-encased CFST column
due to the geometrical shape and rebar placement. The concrete outside
the CFST in the corners of the box and the concrete in the web walls
have different levels of confinement. Thus the division of confined
concrete regions in the box column requires special study. Three RC
box column models (the basic model excluding the CFST core in each
corner of the box) with different stirrup placement are analyzed in
order to study the division of confined concrete regions. The concrete
stress distributions are compared as shown in Fig. 3. The cube concrete
   0  
0 0
f i =f c ¼ 1:41−0:17In f c ; εi =ε0 ¼ 2:5−0:3In f c and f c N=mm2 .
Fig. 3 shows the comparison of longitudinal stress distribution of
concrete (S33) for three different reinforcement placement patterns at
the ultimate load (Nu). The corner concrete is partially confined by an
un-closed stirrup in Case 1, and Case 2 provides a closed stirrup for
the corner concrete but there is no cross-ties in the web walls. The cor-
ner concrete is confined by a closed stirrup and there are cross-ties in
the web walls in Case 3. Each of these models has different levels of
concrete confinement, and the variation in concrete stress distribution
from these different models is illustrated in the figure. Fig. 3 shows
that the confined region, where the longitudinal stress of concrete is
larger than the peak stress of plain concrete stress (33 N/mm2), occurs
in the corners for all three models. The increased stirrups confining
the box web concrete change longitudinal stress distribution of concrete
in the web walls. The confined region is in the ends of the web walls and
is a very small area in Case 1. In Case 2 the confined region increases due
to the added stirrup at the ends of the web walls, but in the middle of
214 Y.-F. An et al. / Structures 3 (2015) 211–226

stirrup unconfined region unconfined region unconfined region

stirrup stirrup

(a) Case 1 (b) Case 2 (c) Case 3

Fig. 3. Comparison of longitudinal stress distribution of the outer concrete (unit: N/mm2).

the web walls the longitudinal stress is no larger than the peak stress with the same volumetric ratio of stirrup reinforcement. For the
of plain concrete stress. Almost all the area of the web walls becomes stress–strain relations in the wall-type columns with aspect ratios of 1
a confined region due to the stirrup in the added cross-ties in the web and 2 (defined as Bcx/Bcy shown in Fig. 4, where Bcx is the distance
walls for Case 3 and better utilization of this concrete is achieved in between the adjacent two cross-ties, and Bcy is the distance between
Case 3. These observations are consistent with prior work of Papanikolaou the stirrups parallel to the web wall), the peak stress, the corresponding
and Kappos [19]. Thus the concrete of the concrete-encased CFST box strain and ductility were higher than those of unconfined concrete.
column is divided into four regions as shown in Fig. 4, i.e. the outer un- However, for the wall-type columns with aspect ratios of 3 and 4, the
confined concrete outside the stirrup, the outer confined concrete in peak stress, the corresponding strain and ductility did not increase
the corner, the outer concrete in the web walls and the core concrete significantly compared with those of unconfined concrete, which indi-
in the steel tubes. The uniaxial compressive stress–strain relations of cated that the confinement provided by the stirrup was limited. There-
different kinds of concrete are defined as follows: fore, the concrete within the closed stirrups of the web walls (shown
in Fig. 4) was modeled as unconfined concrete if the aspect ratio of
a) The outer unconfined concrete outside the stirrup. the stirrups is larger than 2. For these unconfined concrete zones, the
uniaxial stress–strain relation is used as Eq. (2). If the aspect ratio of
The unconfined concrete outside the stirrup is modeled with no con- the web walls is less than or equal to 2, the uniaxial stress–strain rela-
finement provided by the stirrup, and the uniaxial stress–strain relation tion of outer confined concrete in concrete-encased CFST column pro-
in Eq. (2) is used. vided by Han and An [11] is used here as shown in Eq. (3), where the
increase of the plasticity due to the confinement provided by stirrup is
b) The outer concrete in the web walls. considered.
8
<σ kðε=ε0 Þ
If the outer concrete in the web walls is not confined by the closed 0 ε≤ε0
σ¼ k−1 þ ðε=ε0 Þk ð3Þ
stirrup (as shown in Case 1 in Fig. 3), it is modeled as an unconfined con- :
crete where the uniaxial stress–strain relation shown in Eq. (2) is used. σ 0 −Edes ðε−ε0 Þ εNε0
For the outer concrete in the web walls confined by the closed stirrup
0 ρv f yh
(as shown in Cases 2 and 3 in Fig. 3), the following part is discussed in In which σ 0 ¼ f c ; ε0 ¼ 0:00245 þ 0:0122 fc
0 ; ρv ¼ AAhcclhs ;
order to determine the model. Hoshikuma et al. [15] provided an exper-
imental study on wall-type confined RC columns confined by stirrups rffiffiffiffiffi
under axial compression, and determined that the confinement provid- Ec 0:15σ 0 Bc
k¼ ;E ¼ ; ε0:85 ¼ 0:225ρv þ ε0 :
ed by stirrups depends on the aspect ratio of the wall-type columns Ec −ðσ 0 =ε0 Þ des ε0:85 −ε0 s

Corner region Web wall region

Confined concrete
Outer Bc
unconfined Core concrete
concrete Stirrup

Outer confined concrete in the corner

Bcy

Bcx
Outer concrete in the web wall

Fig. 4. A schematic view of the concrete regions in the concrete-encased CFST box section.
 Y.-F. An et al. / Structures 3 (2015) 211–226 215

ρv is the stirrup ratio, Ah is the cross-sectional area of the stirrup, lh is is used. The uniaxial stress–strain relation of core concrete in the
the total length of the stirrup, Acc is the area of confined concrete, fyh concrete-encased CFST box column is shown in Eq. (4) as suggested
is the yield stress of the stirrup, s is the stirrup spacing and Bc is the by Han et al. [12].
sectional width of confined concrete and equal to Bcy considering the 8
decrease in the confinement by the wall-type stirrup. < 2x−x2 ; ðx≤1Þ;
y¼ x ð4Þ
The wall-type columns tested by Hoshikuma et al. [15] and Mander : ; ðx N1Þ;
[17] are used to verify the accuracy of the above method in predicting β0 ðx−1Þ2 þ x
the behavior of concrete in the web wall. Fig. 5 shows the comparison
0
of typically predicted and measured σc–εc relations of the concrete in where x ¼ ε=ε0 , y ¼ σ=σ 0 ; σ 0 ¼ f c ; ε0 ¼ εc þ 800ξ  10−6 ; εc ¼
0:2

the stirrup. Fig. 6 shows the peak stress (σcc) and the corresponding   ½0:25þðξ−0:5Þ7   0:5
0 0
strain (εcc) of the concrete in the stirrup. The mean values of σcc,c/σcc,e 1300 þ 12:5f c  10−6 ; β0 ¼ 2:36  105 fc  0:5≥
and εcc,c/εcc,e are 0.996 and 0.93, and the corresponding standard devia- 0:12 and f c (N/mm2).
0

tions are 0.056 and 0.113, respectively. Generally, good agreement is

For concrete in tension, the cracking strength of concrete σt is 0:3 
obtained between the predicted and measured results.  0 0:67
fc according to Model Code 2010 [9]. The stress–strain relation
c) The outer confined concrete in the corner
introduced in An et al. [3] is used herein to describe the post-failure
behavior of the concrete in tension.
The confinement provided by stirrup should be considered in the
outer confined concrete in the corner. The uniaxial stress–strain relation 2.4. Verification of the FEA model
of outer confined concrete in concrete-encased CFST column provided
by Han and An [11] is used here as shown in Eq. (3), where the sectional Four concrete-encased CFST box stub columns and one correspond-
width Bc is shown in Fig. 4. ing RC column were tested to verify the above FEA model.
Fig. 7(a) shows the cross-section of the specimens, the geometric
d) Core concrete in the steel tubes dimension, the rebar and arrangement of the encased CFST. The length
of the specimens is 900 mm. Two end plates made of steel 10 mm thick
steel lay at the ends of the specimens.
For the core concrete in the steel tube of concrete-encased CFST box
A f
Table 1 summarizes the material properties (Specimen CE-1-1 to CE-
columns, the confinement factor ξ (ξ ¼ Acoresf ys , where As and Acore 2-2, and RC), where fys, fyl and fyh are the yield strength of the steel tube,
ck;core

are the cross sectional area of steel tube and core concrete, fys is yield longitudinal bar and stirrup, respectively, and fcu,out and fcu,core are the
stress of steel tube, fck,core is the characteristic strength of the concrete) cube strength of the outer concrete and inner concrete, respectively.

40 40

Measured
30 Predicted 30
σc (N/mm2)
σc (N/mm2)

20 20

Measured
10 10
Predicted

0 0
0 10000 20000 30000 40000 0 10000 20000 30000 40000
εc (με) εc (με)
(a) LW1 (b) LW4
80 80

Measured
60 60
Predicted
σc (N/mm2)
σc (N/mm2)

40 40
,

Measured
20 20
Predicted

0 0
0 10000 20000 30000 40000 0 10000 20000 30000 40000
εc (με) εc (με)

(c) Wall 1 (d) Wall 10

Fig. 5. Comparison of predicted and measured σc–εc relations of wall-type columns.
216 Y.-F. An et al. / Structures 3 (2015) 211–226

80 30000

60

σcc,c (N/mm2)
20000
σcc,e=σcc,c εcc,e=εcc,c

εcc,c (με)
40

10000
20

0 0
0 20 40 60 80 0 10000 20000 30000
σcc,e (N/mm ) 2 εcc,e (με)

(a) σcc (b) εcc

Fig. 6. Comparison of predicted and measured σcc and εcc of wall-type columns.

300

Strain gauges
300

80 47 46 47 80
Displacement
transducer Steel fixture

Specimen
RC
300

Strain gauges
Strain gauges Displacement
transducer

Strain gauges 5000 kN testing

300

80 47 46 47 80
machine
61
Steel fixture
61

CFST
Concrete-encased CFST

(a) Cross-section(units: mm) (b) Layout of test

Fig. 7. Cross-section and the layout of the test.

Table 1
Specimen information, test results and predicted ultimate strength.

No. Specimen D×t fys Stirrup fyh Outer long. bar Inner long. bar fcu,out fcu,core DI Measured Predicted Nuc/Nue
label (mm) (N/mm2) Dia.-space (mm) (N/mm2) Dia.- fyl (N/mm2) Dia.- fyl (N/mm2) (N/mm2) (N/mm2) Nue (kN) Nuc (kN)

1 CE-1-1 60 × 3 355 ϕ6.5–100 372 ϕ11.8–406 ϕ10.1–444 35 62.5 2.03 3740 3550 0.949
2 CE-1-2 60 × 3 355 ϕ6.5–100 372 ϕ11.8–406 ϕ10.1–444 35 62.5 2.19 3760 3550 0.944
3 CE-2-1 60 × 3 355 ϕ6.5–50 372 ϕ11.8–406 ϕ10.1–444 35 62.5 2.5 3840 3550 0.924
4 CE-2-2 60 × 3 355 ϕ6.5–50 372 ϕ11.8–406 ϕ10.1–444 35 62.5 2.26 3942 3550 0.901
5 RC – 355 ϕ6.5–100 372 ϕ11.8–406 ϕ10.1–444 35 62.5 1.72 2895 – –
6 H1-1 60 × 1.89 325 ϕ6.5–100 376 ϕ11.4–380 ϕ11.4–380 55.5 55.5 – 4511 4020 0.891
7 H1-2 60 × 1.89 325 ϕ6.5–100 376 ϕ11.4–380 ϕ11.4–380 55.5 55.5 – 4482 4020 0.897
8 H2-1 60 × 1.89 325 ϕ6.5–100 376 ϕ11.4–380 ϕ11.4–380 55.5 73.3 – 4559 4200 0.921
9 H2-2 60 × 1.89 325 ϕ6.5–100 376 ϕ11.4–380 ϕ11.4–380 55.5 73.3 – 4754 4200 0.883
Mean value (μ) 0.914
Standard deviation 0.026
 Y.-F. An et al. / Structures 3 (2015) 211–226 217

Maximum principal
plastic strain
strain
Outward bulging Concrete crushing and
Concrete crushing and
outward bulging
outward bulging

Observed Predicted

(a) CE-2-1 (b) RC

Fig. 8. Failure modes of the outer concrete.

The tests were completed on a 5000 kN capacity universal testing
machine and the test data were collected by an IMP data acquisition
system as shown in Fig. 7(b). Displacement transducers were employed
at the end of the specimens to monitor the axial shortening. Strain
gauges were used to measure the longitudinal strain of the outer con-
crete, longitudinal bars and steel tubes at mid-length of the specimen
as shown in Fig. 7. A load interval of less than one tenth of the estimated
load capacity was used. Each load interval was maintained for about
2 min. When the load came close to the estimated load capacity, dis-
placement control was used until the failure of the specimens.
Fig. 8 gives the typical failure modes of the outer concrete in the
concrete-encased CFST box column (CE-2-1) and the corresponding
RC specimens. The failure modes of the outer concrete in the concrete-
encased CFST box and the corresponding RC specimens are the same,
since concrete crushing and outward bulging occur in the middle of
the specimen for both column types. Fig. 9 shows the failure modes of
the inner CFST component and core concrete within concrete-encased
CFST (CE-2-1). The inner CFST component sustains bending deflection
because the confinement of the outer concrete does not exist in the
middle after the outer concrete is crushed. However the core concrete
within the CFST remained intact due to the confinement of the steel
tube. Figs. 8 and 9 also compare the predicted and measured failure
modes of the outer concrete, the inner CFST component and the core
concrete. In general, good agreement between the predicted and mea-
sured results is obtained.
Fig. 10 shows load (N) versus measured longitudinal strain (εl) at
mid-length in the outer concrete, longitudinal bar and steel tube, and
the three curves are almost the same, which indicates that the outer
concrete, longitudinal bar and steel tube work together well for devel-
opment of composite behavior. Load (N) versus the average measured
axial strain (ε) of the 5 specimens are given in Fig. 11. The Ductility
Index DI defined in Tao et al. [24] for the 5 specimens is provided in
Table 1. DI is equal to ε85%/εy, where ε85% is the axial strain when the
load capacity deteriorates to 85% of the ultimate capacity Nu, and εy is
equal to ε75%/0.75, where ε75% is the axial strain when the load attains
75% of Nu in the pre-peak stage. The Nu and DI of the concrete-encased
CFST box columns are at least 29% and 18% larger than those of the
corresponding RC specimen due to the contribution of the inner CFST
component.
Fig. 11 also shows the calculated results based on the confined and
un-confined models of the outer confined concrete. It can be seen that
the predicted ultimate capacities of the situations are quite close, how-
ever, the predicted εu (stain in Nu) and ε85% with confined effects con-
sidered are relatively significantly greater and more close to the tested
results.
Liu [16] tested four concrete-encased CFST box stub columns with
the same cross-section as shown in Fig. 7(a) under axial compression,
and Table 1 summarizes the properties and results for these tests
(Specimen H1-1 to H2-2). Fig. 12 compares the predicted and measured
Nu and εu at Nu for the 8 specimens shown in Table 1. The mean value
and the standard deviation of Nuc/Nue (where Nuc and Nue are the pre-
dicted and measured ultimate loads, respectively) are 0.960 and 0.018,
respectively. The mean value and the standard deviation of εuc/εue

5000
Concrete
Longitudinal bar
4000 Steel tube
N (kN)

3000

2000

1000

Observed Predicted Observed Predicted 0

0 2000 4000 6000 8000
(a) CFST component (b) Core concrete in CFST ε1(με)

Fig. 9. Failure mode of the CFST component and the core concrete. Fig. 10. N–εl of different materials in the middle.
218 Y.-F. An et al. / Structures 3 (2015) 211–226

5000 5000
RC
4000 Predicted(confined) 4000
Predicted(unconfined)

N (kN)

N (kN)
3000 3000

2000 2000
CE-1-1
1000 1000 CE-1-2
Predicted(confined)
Predicted (unconfined)
0 0
0 2000 4000 6000 8000 0 2000 4000 6000 8000
ε (με) ε (με)

(a) (b)
5000

4000
N (kN)

3000

2000
CE-2-1
1000 CE-2-2
Predicted (confined)
Predicted (unconfined
0
0 2000 4000 6000 8000
ε (με)

(c)
Fig. 11. Comparison of predicted and measured N–ε relations.

(where εuc and εue are the predicted and measured strain in the ulti-
mate load, respectively) are 0.910 and 0.036, respectively. In general,
good agreement between the predicted and measured behavior is
obtained.
3. Analytical behavior
3.1. Analysis of complete load–deformation curves
A typical concrete-encased CFST box stub column with the cross-
section shown in Fig. 2(a) is analyzed. The sectional width B, and length
L are 2000 mm and 6000 mm, respectively. The dimensions of one
quarter of the hollow part are shown in Fig. 2(a). The diameter of
the longitudinal bar is 22 mm, and longitudinal ratio, αl equals Al /
(Aout + Al) and is 0.011, where Al is the total area of longitudinal bar,
and Aout is the area of the outer concrete. The diameter of the stirrup
is 12 mm, and the spacing s is 150 mm. The diameter of the steel tube
D is 400 mm, and the tube wall thickness t is 9.3 mm. The steel ratio
αs (defined as As / Acore) is 0.1. The material properties are: fcu,out =
40 N/mm2 (fc′ = 33 N/mm2), fcu,core = 60 N/mm2 (fc′ = 51 N/mm2),
fys = 345 N/mm2, fyl = 335 N/mm2 and fyh = 335 N/mm2. The aspect
ratio of the concrete in the stirrups of the web wall is larger than 2,
thus the confinement in the concrete of the web wall is not considered
as noted earlier.

5000 4000

4000
3000
εue= εuc
Nuc (kN)

3000
εuc (με)

Nue= Nuc
2000
2000

1000
1000

0 0
0 1000 2000 3000 4000 5000 0 1000 2000 3000 4000
Nue (kN) εue (με)

(a) Nu (b) ε u
Fig. 12. Comparison of predicted and experimental Nu and εu.
 Y.-F. An et al. / Structures 3 (2015) 211–226 219

150000 150000
Whole section
C Point C
120000 B 120000 CFST component
A D
RC component
90000 90000
N (kN)

N (kN)
εscy
60000 60000

30000 30000

0 0
0 2000 4000 6000 8000 0 2000 4000 6000 8000
ε (με) ε (με)

(a) Typical N-ε relation (b) Contributions of the two components

Fig. 13. Full range analysis on the load versus deformation relation.

Fig. 13(a) gives the typical N–ε relation obtained from analysis of
the concrete-encased CFST box stub column. Four characteristic points
are marked in the curve. At Point A, the steel tubes start to yield; at
Point B, the unconfined concrete reaches its peak stress; at Point C, the
column reaches the ultimate load (Nu); and at Point D, the load falls to
85% of Nu.
Fig. 13(b) shows the distribution of resistance between the inner
CFST component and outer RC component of the concrete-encased
CFST box stub column. The outer RC component has reached its ultimate
load at Nu (Point C). However, the load of the inner CFST component
continues to increase throughout the whole loading stage, and as a re-
sult the CFST component contributes an increasing portion of the axial
resistance at increased inelastic deformation. The strain (εscy) is defined
as the strain at the ultimate strength of inner CFST component in
concrete-encased CFST stub column under axial compression due to
the increase in its strength throughout the whole loading stage [11],
and εscy can be calculated as:

0 0  0:2
εscy ¼ 1300 þ 12:5f c þ 600 þ 33:5f c ξ ðμεÞ: ð5Þ
The above strain (εscy) can be defined as the strain at the ultimate
strength of the inner CFST component in the concrete-encased CFST
box stub column under axial compression because the corners of the
concrete-encased CFST box column become concrete-encased CFST

(a) Point A (b) Point B

(c) Point C (d) Point D

Fig. 14. The distribution of longitudinal stress of concrete (unit: N/mm2).
220 Y.-F. An et al. / Structures 3 (2015) 211–226

400 Fig. 14 shows the distribution of longitudinal stress (S33) of the

concrete in the middle for each of these 4 points. The longitudinal stress
in the web walls is larger than that in the corner at Point A as shown
300 in Fig. 14(a). The unconfined concrete and confined concrete in the
σh (N/mm 2)

corners are supporting 90% and 82% of their peak loads, respectively.
Point C The load in the core concrete of the CFST supports 60% of its peak
Point 1 load. At Point B the unconfined concrete reaches its peak resistance,
200 and the longitudinal stress of the confined concrete in the corners
is a little smaller than that of the unconfined concrete as shown in
Fig. 14(b). The load of confined concrete in the corners and core con-
100 Stirrup crete in the steel tubes reach 95% and 72% of their peak loads at Point
B. At Point C the resistance provided by the unconfined concrete de-
Point A creases, however, the resistance of the CFST core continues to increase,
0 and the confined concrete in the corners increases slightly at Point C as
0 2000 4000 6000 8000 shown in Fig. 14(c). The longitudinal stress of the confined concrete in
ε (µε) the corners is larger than that of the unconfined concrete at Point C.
The maximum stress of the core concrete in steel tubes is 14.3% higher
Fig. 15. Stress of hooping (σh) versus axial strain (ε) relation. than the peak stress of plain concrete (fc′ = 51 N/mm2) due to the con-
finement provided by the steel tubes. After Point C, the resistance of the
confined concrete in the corners begins to decrease, and the smallest
stress in the specimen occurs outside of the stirrups in the corner as
areas. Fig. 13(b) shows that the inner CFST component approaches its shown in Fig. 14(d). The resistance of the core concrete in the steel
ultimate strength corresponding to Eq. (5) when the box column tubes continues to increase slightly after Point D. The maximum stress
reaches Nu, but that slight increases in CFST component resistance is of core concrete in steel tubes is 29% higher than the peak stress of
noted at strains larger than εscy. plain concrete (fc′ = 51 N/mm2) as shown in Fig. 14(d).

10 6
Point 1
Point 1
Point2
Point 1 Point2
8 Point 4 Point 2 Point 4
Point 3 Point 3 Point 3
4 Point 4
6
p (N/mm2)

p (N/mm2)

Point C
4 Point C
2
Point 1
2 Point 2
Point 3
Point 4
0 0
0 2000 4000 6000 8000 0 2000 4000 6000 8000
ε (με) ε (με)

(a) P1 (b) P2
Fig. 16. Interaction stresses between steel and concrete.

1.6 1.6

λw=12
1.4 1.4
λw=12
Nuc/Nue

Nuc/Nue

1.2 1.2

1 1

0.8 0.8
0 10 20 30 40 0 10 20 30 40
λw λw
(a) No initial imperfection (b) Initial imperfection (0.24tc)

Fig. 17. Influence of the initial imperfection on Nu.

 Y.-F. An et al. / Structures 3 (2015) 211–226 221

40 Points A and C, σh increases more rapidly with increasing axial strain,

and after Point C, the strain increases more rapidly with small increases
in stress because the stirrup reinforcement reaches the yield stress at
30 approximately Point C.
Fig. 16 shows the contact stresses between the steel tube and con-
crete. When ε is smaller than 1000 με, there is virtually no contact stress
σc (N/mm2)

between the steel tube and the core concrete in the inner CFST compo-
20
nent (P1 = 0) as shown in Fig. 16(a), because in the elastic stage,
λw =1.71 Poisson's ratio of the core concrete is smaller than that of the steel
w =4 tube and the expansion of the core concrete is smaller than that of the
10 λw =8 steel tube. For strains larger than 1000 με, the expansion of the core con-
λw =12 crete is larger than that of the steel tube, which leads to the existence of
λw =15 the stress P1. The value of P1 increases with increasing ε, and P1 provides
λw =20
0 confining stress that increases the longitudinal stress of the core con-
0 2000 4000 6000 crete to a value larger than the peak stress of plain concrete at Point C.
ε (με) When ε is smaller than 1000 με, P2 exists at the elastic stage, because
of the differences in Poisson's ratio as shown in Fig. 16(b). However P2
Fig. 18. Influence of λw on σc–ε of the web wall. does not exist for strains larger than 1000 με, because the expansion
of the outer concrete is larger than that of the steel tube. When ε is
about 6000 με, P2 increases significantly for Points 3 and 4, which indi-
cates that the lateral deformation at Points 3 and 4 of the steel tube is
restricted by the outer concrete.

3.3. Effect of λw

For the RC box columns, the wall slenderness ratio λw is a critical

Local indent
(a) λw =4 (b) λw =15
Fig. 19. Influence of λw on failure mode.
3.2. Interactions between steel and concrete
The confinement provided by the stirrup to the outer concrete, the
contact stress between the steel tube and core concrete (P1) and the
contact stress between the steel tube and outer concrete (P2) are
discussed. Fig. 15 shows the computed stirrup stress (σh)–ε relation
(at Point 1 as identified in Fig. 15), and σh relates proportionally to the
confinement provided by the stirrup to the corner concrete. Before
Point A, σh increases proportionally with the increase of ε. Between
parameter [18]. As λw increases, the ratio of flexural strength and stiff-
ness to self-weight increases, and the material usage efficiency increases.
However, when λw becomes too large local buckling of the web walls
occurs and leads to deterioration in strength and increased deformation.
The AASHTO Specification [1] recommends that this decrease in strength
be considered for the RC box columns when λw is larger than 15, and λw
is limited to no larger than 8 in the earthquake design of the RC box
columns in EC 8 [8]. Maria [18] suggested that the initial imperfections
of the web wall must be considered when λw is larger than 12.
The FEA model is used to evaluate the effect of initial imperfections
on the capacity of axially loaded members. The initial imperfection in
the FEA model is taken as the first buckling mode shape determined
by an eigenvalue analysis multiplied by a specific amplitude. The pre-
dicted results are compared to thin RC wall test results conducted by
Saheb and Desayi [21] and Swartz et al. [23]. Fig. 17 shows the influence
of λw and initial imperfection on Nu (Nuc and Nue are the predicted and
measured ultimate load). Fig. 17(a) shows the results for no initial
imperfection and Fig. 17(b) shows the effect of the maximum value
of initial imperfection equal to 0.24tc suggested by Maria [18]. It
can be seen that when λw ≤ 12, there is no need to consider the initial
imperfection since there is no change in the resistance ratio as shown
in Fig. 17(a). However, when λw is larger than 12, the predicted Nuc is

50 50
λw=12 λw=12
40 40
σcc (N/mm2)

σcc (N/mm2)

30 30

20 20
fcu,out =30N/mm 2 α l=0.6%
fcu,out =40 N/mm 2 αl=1.1%
10 10 αl=1.6%
fcu,out =50 N/mm 2

0 0
0 5 10 15 20 0 5 10 15 20
λw λw
(a) fcu,out (b) αl

Fig. 20. Influence of different parameters on σc–λw relations.

222 Y.-F. An et al. / Structures 3 (2015) 211–226

larger than the measured value Nue, and the average value of Nuc/Nue
is 1.08 with no initial imperfection. When the initial imperfection,
0.24tc is included into the model of walls with λw N 12 as shown in
Fig. 17(b), the predicted Nuc is smaller than the measured value Nue,
and the average value of Nuc/Nue is 0.93. Thus in the analysis on the be-
havior of concrete-encased CFST box stub column, it is recommended
that the initial imperfection of web wall be considered when λw is larger
than 12, and that the imperfection be taken as the first buckling mode
shape with an amplitude of 0.24tc.
Six concrete-encased CFST box stub columns are designed with
different λw of 1.71 to 20 to analyze the influence of λw. The column
described in Fig. 2 has λw of 1.71. The web wall thickness tc is held
constant for all analysis, but bc is varied in the six composite box
columns to obtain the target value. Fig. 18 shows the influence of λw
on the σc–ε relations of the web wall, where σc is the average concrete
stress in the web wall. It can be seen that there is no significant influence
of λw on the σc–ε relations when λw ≤ 12. However, when λw is larger
than 12, the peak stress, the corresponding strain and the ductility

150000 150000

100000 100000

N (kN)
N (kN)

αl=0.6%
50000 fcu,out =30N/mm 2 50000
fcu,out =40 N/mm 2 αl=1.1%
fcu,out =50 N/mm 2 αl=1.6%
0 0
0 2000 4000 6000 8000 0 2000 4000 6000 8000
ε (με) ε (με)
(a) fcu,out (b) αl

150000 150000

100000 100000
N (kN)
N (kN)

50000 fyl=235N/mm 2 50000 s=100mm

fyl=335N/mm 2 s=150mm
fyl=400N/mm 2 s=200mm

0 0
0 2000 4000 6000 8000 0 2000 4000 6000 8000
ε (με) ε (με)
(c) fyl (d) s
150000 150000

100000 100000
N (kN)

N (kN)

D/B=0.1
50000 50000 fcu,core=40 N/mm2
D/B=0.15
fcu,core=60 N/mm2
D/B=0.2
fcu,core=80 N/mm2
D/B=0.25
0 0
0 2000 4000 6000 8000 0 2000 4000 6000 8000
ε (με) ε (με)
(e) D/B (f) fcu,core
Fig. 21. Effects of different parameters on N–ε relations.
 Y.-F. An et al. / Structures 3 (2015) 211–226
150000
100000
N (kN)
50000 αs=0.05
αs=0.10
αs=0.15
0 0
0 2000 4000 6000 8000
ε (με)
(g) αs
Fig. 21 (continued).
after peak stress are smaller than those with λw ≤ 12. The peak stress of
σc (σcc) with λw of 15 and 20 are 0.820 and 0.787 and that with λw of
1.71, respectively. An increase in λw leads to a decrease in σcc. The influ-
ence of λw on predicted failure modes is shown in Fig. 19, which shows
that local buckling deformation occurs in the middle of the web wall
with λw = 15, and no such web wall deformation with λw = 4. Fig. 20
shows the influence of concrete strength fcu,out and the longitudinal
ratio αl of the web wall on σcc. It can be seen that there is no significant
influence of λw on σcc with the same concrete strength when λw ≤ 12.
An increase in λw leads to a decrease in σcc with the same concrete
strength when λw N 12. The σcc values for λw of 15 with the concrete
strength of 30 N/mm2, 40 N/mm2 and 50 N/mm2 are 0.811, 0.820 and
0.840 of those for λw of 1.71, respectively. There is no significant effect
of αl on σcc. For the concrete-encased CFST box stub column, the deteri-
oration in strength and deformation should be considered due to the
thin web walls when λw is larger than 12. However, in the engineering
practice in China the λw of the RC box columns are usually smaller than
12 [22]. Thus this paper focuses on the concrete-encased CFST box col-
umn with λw ≤ 12, and the deterioration in strength and deformation
due to the thin web wall is not considered. The influence of the thin
web wall with λw N 12 will be verified in a further study.
4. Parametric analysis and column strength calculations
4.1. Parametric analysis
A parameter study is performed to determine the effect of various
parameters on the N–ε relations of the concrete-encased CFST box
stub column under axial compression. The variations in parameters
considered in the analysis are: 1) for the outer RC component: strength
of the outer concrete fcu,out = 30–50 N/mm2, longitudinal bar ratio αl =
0.6%–1.6%, yield stress of longitudinal bar fyl = 235–400 N/mm2 and
stirrup spacing s = 100–200 mm; 2) ratio of the tube diameter to the
width of the box section D/B = 0.1–0.25; 3) for the inner CFST compo-
nent: strength of the core concrete fcu,core = 40–80 N/mm2, steel ratio of
steel tube αs = 0.05–0.15 and yield stress of the steel tube fys = 235–
420 N/mm2. Fig. 21 shows the influence of different parameters on
N–ε relations, and Fig. 22 shows the influence of different parameters
on Ncfst/N–ε relations (where Ncfst is the strength of the inner CFST com-
ponent). The influence of the various parameters can be summarized as
follows:
(1) Strength of the outer concrete (fcu,out): As fcu,out increases, Nu
increases, but the ductility decreases due to the decrease in
confinement provided by the stirrup. Ncfst/N decreases because
223
150000
100000
N (kN)
50000 fys=235N/mm 2
fys=345N/mm 2
fys=420N/mm 2
0 2000 4000 6000 8000
ε (με)
(h) fys
the strength of the outer RC component increases, but that of
the inner CFST component is unchanged.
(2) Longitudinal bar ratio (αl): As αl increases, Nu increases a little,
and the ductility is not changed due to the unchanged confine-
ment provided by the stirrup. Ncfst/N decreases a little because
the strength of the outer RC component increases, but that of
the inner CFST component is unchanged.
(3) Yield stress of the longitudinal bar (fyl): As fyl increases, Nu
increases a little, and the ductility is not changed due to the un-
changed confinement provided by the stirrup. Ncfst/N decreases
a little because the strength of the outer RC component increases,
but that of the inner CFST component is unchanged.
(4) Stirrups spacing (s): As s increases, the ductility decreases due
to the decrease in the confinement provided by the stirrup, and
Nu is unchanged. Prior to developing Nu, Ncfst/N is not changed
because the strengths of the inner CFST and outer RC compo-
nents are unchanged. After developing Nu, Ncfst/N increases as s
increases because the resistance of the outer RC component
decreases while the inner CFST component is unchanged.
(5) D/B: As D/B increases, Nu and ductility increase because the
strength of the inner CFST component increases. Ncfst/N increases
significantly due to the increase in the inner CFST component.
This parameter is viewed as one of the key factors affecting
the loading distribution between the outer RC and inner CFST
components [2,4].
(6) Strength of the core concrete (fcu,core): As fcu,core increases, Nu
and Ncfst/N increase due to the increase in the strength of the
inner CFST component. The ductility decreases a little due to
the decrease in the confinement provided by the steel tube to
the higher strength core concrete.
(7) The steel ratio of the steel tube (αs): As αs increases, Nu and
Ncfst/N increase because the strength of the inner CFST compo-
nent increases. The ductility increases a little due to the increased
contribution of the resistance of the steel.
(8) Yield stress of the steel tube (fys): As fys increases, Nu and Ncfst/N
increase due to the increase in the strength of the inner CFST
component. The ductility increases a little due to the increased
contribution of the resistance of the steel.
In the design of the concrete-encased CFST box stub column, the
steel tubes are filled with high strength concrete, because of the con-
finement provided by the steel tubes, whereas normal-strength con-
crete is used in the outer RC component. The use of high-strength core
concrete is expected to permit smaller and lighter sections. The thin-
224 Y.-F. An et al. / Structures 3 (2015) 211–226

0.8 0.8

0.6 0.6

N cfst/N

N cfst/N
0.4 0.4

fcu,out=30N/mm2 αl=0.6%
0.2 fcu,out=40 N/mm2 0.2 αl=1.1%
fcu,out=50 N/mm2 αl=1.6%

0 0
0 2000 4000 6000 8000 0 2000 4000 6000 8000
ε (με) ε (με)

(a) fcu,out (b) αl

0.8 0.8

0.6 0.6

N cfst/N
N cfst/N

0.4 0.4

fyl=235N/mm2 s=100mm
0.2 fyl=335N/mm2 0.2 s=150mm
fyl=400N/mm2 s=200mm

0 0
0 2000 4000 6000 0 2000 4000 6000 8000
ε (με) ε (με)

(c) fyl (d) s

0.8 0.8
D/B=0.1
D/B=0.15
0.6 D/B=0.2 0.6
D/B=0.25
N cfst/N
Ncfst/N

0.4 0.4

fcu,core=40 N/mm2
0.2 0.2
fcu,core=60 N/mm2
fcu,core=80 N/mm2
0 0
0 2000 4000 6000 8000 0 2000 4000 6000 8000
ε (με) ε (με)

(e) D/B (f) fcu,core

0.8 0.8

0.6 0.6
N cfst/N

N cfst/N

0.4 0.4

αs=0.05 fys=235N/mm2
0.2 αs=0.10 0.2 fys=345N/mm2
αs=0.15 fys=420N/mm2
0 0
0 2000 4000 6000 8000 0 2000 4000 6000 8000
ε (με) ε (με)

(g) αs (h) fys

Fig. 22. Effects of different parameters on Ncfst/N–ε relations.
 Y.-F. An et al. / Structures 3 (2015) 211–226 225

walled steel tube with high yield stress can be used in the inner CFST 80000
component, because of the added confinement provided by the outer
concrete which delays buckling and damage to the steel and enhances
the confinement provided to core concrete. Fig. 23 compares the behav- 60000

Ncfst,u (kN)
ior of concrete-encased CFST box columns with various D/B ratios
and the corresponding RC box stub columns, where there is no inner
CFST in the corresponding RC box column. Nu and DI of the concrete-
encased CFST box column with D/B of 0.1 are 6% and 2.7% larger than
40000
those of the corresponding RC columns, while Nu and DI are larger by
13% and 6% for the concrete-encased CFST box column with D/B of
Ncfst,s= Ncfst,u
0.15. The strength contributions of the inner CFST component are 0.09 20000
and 0.20 for the concrete-encased CFST box columns with D/B of 0.1
and 0.15 at Nu as shown in Fig. 22(e). Thus D/B of 0.15 is recommended
to make significant contributions to the loading and the ductility. How- 0
ever, D/B is limited by the dimensional constraints of the box section, 0 20000 40000 60000 80000
because an increase in D/B leads to a decrease in the hollow part of Ncfst,s (kN)
the composite column.
Fig. 24. Comparison of Ncfst,s and Ncfst,u.
4.2. Calculations on the ultimate strength

Han and An [11] introduced a simplified superposition method to
predict the ultimate load of the concrete-encased CFST stub column
under axial compression, where the ultimate loads of the inner CFST
and the outer RC components are added together. This superposition
method is conservative compared with the tested results [11]. In this
method, contact stress between the steel tube and outer concrete is
not considered, but the confinement provided by the steel tube to
core concrete is considered. The outer RC component has reached its
ultimate load at Nu. Fig. 24 compares the load supported by the CFST
component at Nu (Ncfst,s) to the ultimate load of the CFST component
(Ncfst,u) based on the strain of Eq. 5. The figure shows that Ncfst,s ap-
proaches Ncfst,u and the average value of the average value of the
Ncfst,s/Ncfst,u ratio is 0.95. The decrease in the ultimate load of the web
wall due to λw is not considered in this paper. Thus the proposed super-
position method is limited to λw less than 12, and the superposition
method for predicting Nu can be expressed as:
Nuc ¼ Nuc;rc þ N uc;cfst
where Nu,rc and Nu,cfst are the ultimate load of the outer RC component
and the inner CFST component, respectively. Nu,rc and Nu,cfst can be
calculated by Eqs. (7) and (8) according to GB50010-2010 [10] and
DBJ/T13-51-2010 [7], respectively.
where fck,out is the characteristic strength of the outer concrete, and is
calculated as 0.67fcu,out for normal strength concrete.
Nuc;cfst ¼ 4  ð1:14 þ 1:02ξÞ  Asc f ck;core ð8Þ
where fck,core is the characteristic strength of the core concrete and Asc is
the cross-sectional area of CFST (Asc = As + Acore).
Fig. 25 compares the predicted (Nuc) and measured (Nuc) ultimate
loads of the concrete-encased CFST box stub column under axial com-
pression from the tests described in this paper. The mean and standard
deviation of Nuc/Nue are 0.914 and 0.026.
5. Conclusions
This paper presents the results of a limited experimental study on
concrete-encased CFST box stub columns subjected to axial load with-
out slenderness effects, and a more comprehensive analytical study of
the same elements. The following conclusions can be drawn from this
ð6Þ
work.
(1) A FEA model on the analysis of the concrete-encased CFST box
stub column under axial compression is developed. Four differ-
ent concrete regions including the outer un-confined concrete
outside the stirrup, the outer confined concrete in the corners,
the outer concrete in the web wall and the core concrete in the

N ¼ f ck;out Aout þ f yl Al ð7Þ

5000
150000
4000
Nuc (kN)

100000
3000 Nue= Nuc
N (kN)

2000
D/B=0.1
50000 D/B=0.15
D/B=0.2
1000
D/B=0.25
RC 0
0
0 2000 4000 6000 8000
0 1000 2000 3000 4000 5000
ε (με) Nue (kN)
Fig. 23. Comparisons between the composite and the corresponding RC columns. Fig. 25. Comparison of Nu between the predicted and test results.
226 Y.-F. An et al. / Structures 3 (2015) 211–226
steel tube are defined and evaluated based upon the effects of
confinement.
(2) The failure modes from the experiments of concrete-encased
CFST box stub columns are outward bulging and crushing of
the outer concrete at mid-height of the column. Bending deflec-
tion of the inner CFST component at mid-height of the column
is noted and the core concrete remained intact due to the con-
finement of the steel tube. The ultimate load and ductility of
concrete-encased CFST box columns are larger than those ob-
tained from the corresponding RC box column due to the contri-
bution of the inner CFST component. Generally good agreement
is obtained between the predicted and measured results.
(3) When the column reaches its ultimate load Nu, the outer RC
component has reached its ultimate load, and the inner CFST
component is approaching its ultimate strength. There is no
contact stress between the steel tube and outer concrete at Nu,
but the confinement provided by the steel tube to the core con-
crete must be considered. A superposition model for predicting
Nu is presented which is applicable when λw ≤ 12. When
λw N 12, the resistance provided by the web wall is reduced,
and the proposed superposition model will overestimate the
axial resistance.
(4) Nu increases with increasing values of fcu,out, αl, fyl, D/B, fcu,core, αs,
and fys, but there is no significant influence of s on Nu.
Acknowledgements
The research reported in the paper is part of Project supported by
the National Natural Science Foundation of China (51378290) and
the Tsinghua University Initiative Scientific Research Program (No.
20131089347). These financial supports are highly appreciated.
References
[1] American Association of State Highway and Transportation Officials (AASHTO).
AASHTO LRFD Bridge Design Specifications [Washington, D.C.] ; 2012.
[2] An YF, Han LH. Behaviour of concrete-encased CFST columns under combined com-
pression and bending. J Constr Steel Res 2014;101:314–30.
[3] An YF, Han LH, Roeder C. Flexural performance of concrete-encased concrete-filled
steel tubes. Mag Concr Res 2014;66(5):249–67.
[4] An YF, Han LH, Zhao XL. Analytical behaviour of eccentrically-loaded concrete-
encased CFST box columns. Mag Concr Res 2014;66(15):789–808.
[5] An YF, Han LH, Zhao XL. Experimental behaviour of box concrete-encased CFST
eccentrically loaded column. Mag Concr Res 2013;65(20):1219–35.
[6] Attard MM, Setunge S. Stress–strain relationship of confined and unconfined
concrete. ACI Mater J 1996;93(5):432–42.
[7] DBJ/T13-51-2010. Technical specification for concrete filled steel tubular structures.
Fuzhou (China): The Construction Department of Fujian Province; 2010 [in Chinese].
[8] EC 8. Design of structures for earthquake resistance—part 2: bridges. European
Committee for Standardization; 2005.
[9] Fédération Internationale du Béton. Model code 2010: first complete draft.
Switzerland: Lausanne; 2010.
[10] GB50010-2010. Code for design of concrete structures. Beijing, China: China Building
Industry Press; 2010 [in Chinese].
[11] Han LH, An YF. Performance of concrete-encased CFST stub columns under axial
compression. J Constr Steel Res 2014;93:62–76.
[12] Han LH, Yao GH, Tao Z. Performance of concrete-filled thin-walled steel tubes under
pure torsion. Thin-Walled Struct 2007;45(1):24–36.
[13] Han LH, Zhao XL, Tao Z. Tests and mechanics model for concrete-filled SHS stub
columns, columns and beam-columns. Steel Compos 2001;1(1):51–74.
[14] Hibbitt, Karlson, Sorenson Inc. ABAQUS Version 6.5: theory manual, users' manual,
verification manual and example problems manual; 2005.
[15] Hoshikuma J, Kawashima K, Nagaya K, Taylor AW. Stress–strain model for confined
reinforced concrete in bridge piers. J Struct Eng ASCE 1997;123(5):624–33.
[16] Liu LY. Study on behavior of a new concrete-filled steel tube reinforced concrete
column under axial compression. (Master Dissertations) Fuzhou, China: Fuzhou
University; 2013 [in Chinese].
[17] Mander JB. Seismic design of bridge piers. New Zealand: University of Canterbury;
1983.
[18] Maria RHS. Behavior of hollow, rectangular concrete piers subjected to biaxial bend-
ing. [PhD dissertation] Austin: Department of Civil Engineering, University of Texas
at Austin; 2001.
[19] Papanikolaou VK, Kappos AJ. Numerical study of confinement effectiveness in solid
and hollow reinforced concrete bridge piers: analysis results and discussion. Comput
Struct 2009;87(21–22):1440–50.
[20] Roeder CW, Cameron B, Brown CB. Composite action in concrete filled tubes. J Struct
Eng 1999;125(5):477–84.
[21] Saheb SM, Desayi P. Ultimate strength of RC wall panels in two-way in-plane action.
J Struct Eng ASCE 1990;116(5):1384–402.
[22] Sun ZG, Wang DS, Li HN, Du XL. Application of RC hollow bridge pier and review of
seismic behavior research. J Traffic Transp Eng 2013;13(3):22–32 [in Chinese].
[23] Swartz SE, Rosebraugh VH, Berman MY. Buckling tests on rectangular concrete
panels. J Am Concr Inst 1974:33–9.
[24] Tao Z, Han LH, Wang DY. Experimental behaviour of concrete-filled stiffened thin-
walled steel tubular columns. Thin-Walled Struct 2007;45(5):517–27.
[25] Zhang YY, Harries KA, Yuan WC. Experimental and numerical investigation of
the seismic performance of hollow rectangular bridge piers constructed with and
without steel fiber reinforced concrete. Eng Struct 2013;48:255–65.