ARTICLE IN PRESS

Thin-Walled Structures 45 (2007) 517527

www.elsevier.com/locate/tws

Experimental behaviour of concrete-lled stiffened thin-walled

steel tubular columns
Zhong Taoa,, Lin-Hai Hanb, Dong-Ye Wanga
a

College of Civil Engineering, Fuzhou University, Fujian Province 350002, Peoples Republic of China
b
Department of Civil Engineering, Tsinghua University, Beijing 100084, Peoples Republic of China
Received 21 February 2007; received in revised form 3 April 2007; accepted 3 April 2007
Available online 29 May 2007

Abstract
An experimental study on the structural behaviour of concrete-lled stiffened thin-walled steel tubular columns is presented in this
paper. The stiffening was achieved by welding longitudinal stiffeners on the inner surfaces of the steel tubes. Companion tests were also
undertaken on 12 unstiffened concrete-lled steel tubular (CFST) columns, with or without steel bres in the inll concrete. The test
results showed that the local buckling of the tubes was effectively delayed by the stiffeners. The plate buckling initially occurred when the
maximum load had almost reached for stiffened specimens, thus they had higher serviceability benets compared to those of unstiffened
ones. Some of the existing design codes were used to predict the load-carrying capacities of the tested composite columns.
r 2007 Elsevier Ltd. All rights reserved.
Keywords: Concrete-lled steel tube (CFST); Thin walled; Columns; Local buckling; Stiffeners; Steel bre reinforced concrete (SFRC); Strength; Ductility

1. Introduction
For concrete-lled thin-walled steel tubes, it is well
known that local buckling is always a big concern in
column design, especially for square or rectangular steel
tubes [15].
In recent years, in order to improve the performance of
such kind of composite columns, three stiffening measures
have been proposed and studied in the literature: installing
bi-directional binding bars (Fig. 1(a)), welding a set of four
steel tie bars (Fig. 1(b)), and welding longitudinal stiffeners
on the inner or outer surfaces of a steel tube (Fig. 1(c)). The
binding bars or steel tie bars are installed at regular spacing
along the tube axis. Recent test results have revealed these
measures to be promising in increasing ultimate strength or
ductility [611]. Reviews of these studies are given in [3,6],
which indicates that almost all studies in the literature were
performed on stiffened stub columns, while research on

Corresponding author. Tel.: +86 591 87892459;

fax: +86 591 83737442.
E-mail address: taozhong@fzu.edu.cn (Z. Tao).

0263-8231/$ - see front matter r 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tws.2007.04.003

stiffened columns was very rare. The lack of information

indicates a need for further research in this area.
As far as the welding stiffener method is concerned, the
effectiveness of the stiffeners in delaying local buckling of
the steel tubes has been demonstrated by stub column tests
presented in [810]. However, test results reported in [10]
have also shown that, the load versus axial strain curves
tended to drop quickly after peak loads had been achieved
since thin-walled tubes were used, and no obvious ductility
improvement was observed with the increase of stiffener
rigidity. A further experimental investigation was carried
out recently by Tao et al. [11] with an aim to improve the
ductile behaviour of stiffened concrete-lled thin-walled
steel tubular stub columns with various methods. It was
found that welding anchor bars on stiffeners and adding
steel bres in core concrete can effectively enhance the
ductility capacity. The research results reported in this
paper are part of a wider study on stiffened columns with
cross-sections shown in Fig. 1(c).
An experimental investigation of the structural behaviour of concrete-lled stiffened thin-walled steel tubular
columns is presented in this paper. The stiffening was
achieved by welding longitudinal stiffeners on the inner

ARTICLE IN PRESS
Z. Tao et al. / Thin-Walled Structures 45 (2007) 517527

518

Nomenclature
Asc

overall cross-sectional area of composite member ( B2)

As,s
total cross-sectional area of steel stiffeners
B
width of square steel tube
CFST concrete-lled steel tube
DI
ductility index
e
load eccentricity
e/r
load eccentricity ratio, r B/2
Ec
Concrete modulus of elasticity
Es
steel modulus of elasticity
fcu
characteristic cube strength of concrete
fy
yield strength of steel
fy,s
yield strength of the stiffener
surfaces of the steel tubes. The main experimental
parameters considered were load eccentricity and slenderness
ratio. For comparison, additional tests were also undertaken on 12 unstiffened concrete-lled steel tubular (CFST)
columns. In order to investigate the effect of adding steel
bres in core concrete on specimen behaviour, six of the
unstiffened columns were constructed with steel bre
reinforced concrete (SFRC). Existing design codes with
minor modications, were used to predict the load-carrying
capacities of the tested composite columns.
2. Experimental program
2.1. General
A total of 18 specimens, including six stiffened columns
and 12 unstiffened columns, were tested to failure. In the
case of the unstiffened specimens, half of them were lled
with normal concrete (NC), and the rest of them lled with
steel bre reinforced concrete. The steel tube cross-sections
for stiffened and unstiffened specimens are shown in
Fig. 2(a) and (b), respectively.
All specimens were designed to have a same overall
width (B) of 200 mm, as well as a same tube thickness (t) of
2.5 mm. Therefore, a width-to-thickness ratio (B/t) of 80
was achieved. According to design code AS 4100 [12], all
specimens were considered thin walled since the calculated

hs
i
Isc
L
N
Nue
SFRC
t
ts
um
D
e
eL
l

height of the steel stiffener

p
radius of gyration of CFST I sc =Asc
second moment of area for CFST cross-section
length of column
axial load
experimental ultimate strength
Steel bre reinforced concrete
wall thickness of the steel tube
thickness of the steel stiffener
deection at mid-height
axial shortening
strain
longitudinal strain at extreme bres in compression at the time of local buckling
slenderness ratio ( L/i)

slenderness limit is 53.3 [10,12]. In determining the

slenderness classication of the selected column crosssection, a local buckling coefcient k 10 and an
imperfection parameter a 0.651 were incorporated.
Details of slenderness calculation of a cross-section can
be found in [10,12].
The stiffeners were chosen to have a height (hs) of 35 mm
and a thickness (ts) of 2.5 mm. The sizes of the stiffeners
were so determined to meet the rigidity requirement for
stiffeners presented in [10].
The main features of the test specimens are listed in
Table 1, where the slenderness ratio (l) and load
eccentricity (e) for each specimen are given. In Table 1,
specimen designations starting with a U refer to columns

Fig. 2. Cross-sections of test specimens.

Fig. 1. Three stiffening measures to improve the performance of thin-walled CFST columns with square cross-sections.

ARTICLE IN PRESS
Z. Tao et al. / Thin-Walled Structures 45 (2007) 517527

519

Table 1
Specimen labels and member capacities
No.

Specimen
label

L (mm)

e (mm)

e
r

fcu (N/
mm2)

DI

Nue
(kN)

eL (  106)

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18

UCFT1-1
UFRC1-1
SCFT1-1
UCFT1-2
UFRC1-2
SCFT1-2
UCFT1-3
UFRC1-3
SCFT1-3
UCFT2-1
UFRC2-1
SCFT2-1
UCFT2-2
UFRC2-2
SCFT2-2
UCFT2-3
UFRC2-3
SCFT2-3

1190
1190
1190
1190
1190
1190
1190
1190
1190
2340
2340
2340
2340
2340
2340
2340
2340
2340

20.6
20.6
20.6
20.6
20.6
20.6
20.6
20.6
20.6
40.5
40.5
40.5
40.5
40.5
40.5
40.5
40.5
40.5

0
0
0
30
30
30
60
60
60
0
0
0
30
30
30
60
60
60

0
0
0
0.3
0.3
0.3
0.6
0.6
0.6
0
0
0
0.3
0.3
0.3
0.6
0.6
0.6

58.3
59.2
58.3
58.3
59.2
58.3
58.3
59.2
58.3
58.3
59.2
58.3
58.3
59.2
58.3
58.3
59.2
58.3

1.625
2.556
1.845
2.344
3.022
2.473
2.409
3.326
2.510
1.418
2.143
1.818
3.041
3.483
3.325
2.115
2.969
2.939

2260
2275
2640
1760
1740
1770
1130
1189
1220
2305
2285
2455
1660
1555
1660
1068
1150
1200

446
940
1331
1289
1102
1325
2215
1206
3150
662
1052
1933
1300
1670
1950
1213
1048
2190

without stiffeners, and those starting with an S refer to

specimens with stiffeners. Follow the U or S, the CFT in
the labels represents specimens lled with normal concrete,
and FRC represents specimens lled with steel bre
reinforced concrete. The last two numbers in the labels
are used to distinguish specimens with different slenderness
ratio or load eccentricity. Two kinds of slenderness ratio
(l), with values of 20.6 and 40.5 were selected. The
slenderness ratio herein is dened as
L
,
(1)
i
where L is the effective length of a column; which is the
same as the physical p
length
of the column (L) with pin
ended supports; i I sc =Asc , is the section radius of
gyration; Isc and Asc are the second moment of area and
area of a composite cross-section, respectively.
The load eccentricity (e) ranges from 0 to 60 mm. Table 1
lists the load eccentricity ratios (e/r), where r is given by B/2.

2.2. Material properties

All tubes and stiffeners were fabricated with mild steel
sheet with a measured thickness of 2.5 mm. Standard
tensile coupon tests were conducted to measure the steel
properties. The tested steel properties are: elastic modulus
Es 203 kN/mm2; Poissons ratio ns 0.303; yield
strength fy 270 N/mm2; yiled stran ey 1360 me; ultimate
strength fu 346 N/mm2; and ultimate strain eu 0.306.
The concrete used in the test program had a water/
cement ratio of 0.42, made with ordinary Portland cement.
The mix proportions for NC were as follows: cement:
523 kg/m3; water: 220 kg/m3; sand: 581 kg/m3; and coarse
aggregate: 1077 kg/m3. The coarse aggregate was well
graded with a maximum size of 15 mm. For steel bre

reinforced concrete, hooked-end steel bres were used at a

volume percentage of 2%. The bres had an average length
of 30 mm and norminal diameter of 0.5 mm. The mix
proportions for the steel bre reinforce concrete were as
follows: cement: 540 kg/m3; water: 227 kg/m3; sand: 536 kg/m3;
coarse aggregate: 1049 kg/m3; steel bres: 158 kg/m3.
To determine the compressive strength of concrete, three
150 mm cubes were cast for each batch of concrete and
cured in conditions similar to the related specimens. The
measured average cube strengths (fcu) at the time of tests
for NC and SFRC were found to be 58.3 and 59.2 N/mm2,
respectively; and the average values of modulus of elasticity
(Ec) were 30,600 and 33,300 N/mm2, respectively.
2.3. Specimen preparation and test procedure
In preparing the tube of a composite specimen, four steel
plates were cut. If stiffening was specied, longitudinal
stiffeners were welded on the plates with llet welds. After
that the four plates were tack welded into a square shape, and
then welded with a single bevel butt weld along the corners to
form the tube. One end of each tube was tack welded by a
steel bottom plate with a thickness of 12 mm before lling the
tube with concrete. The initial out-of-straightness of all
fabricated tubes were measured along the whole length. It
was found that the maximum values were about L/1000.
Concrete mix was lled in layers for all specimens and
was vibrated by a poker vibrator. These specimens were
then naturally cured in the indoor climate of laboratory.
During curing, a very small amount of longitudinal
shrinkage occurred at the top of the columns. A highstrength epoxy was used to ll this longitudinal gap so that
the concrete surface was ush with the steel tube at the top.
Prior to testing, the top surfaces of the CFST specimens
were ground smooth and at using a grinding wheel with

ARTICLE IN PRESS
520

Z. Tao et al. / Thin-Walled Structures 45 (2007) 517527

Fig. 3. A general view of typical specimens after testing.

diamond cutters. A steel end plate with a thickness of

12 mm was then welded to the top of each of those
specimens. For stiffened specimens, it should be noted that
all stiffeners have been welded to both the top and bottom
end plates. It was thus expected that the stiffeners could
share the axial load.
The columns were tested as pin-ended supported and
subjected to single-curvature bending. Axial loading was
applied through V-shape edges to each specimen, which were
installed with steel caps at both ends prior to loading.
Grooves of 6 mm in depth were machined on each steel cap
to receive the V-shape loading edge so that the load
eccentricity (e) could be precisely controlled [13]. The
effective lengths of all columns (L) shown in Table 1, were
dened as the distance between the tips of the V-shape edges.
For each specimen, eight strain gauges mounted on the
specimen surface were installed to measure the axial and
transverse strains at the mid-height of the column.
Displacement transducers were used along the specimen
span to monitor the deections. Two other transducers, at
the end of the specimen, were used to monitor the axial
shortening (shown in Fig. 3(b)).
Load intervals of less than one-tenth of the estimated load
capacity were used. Each load interval was maintained for
about 23 min. At each load increment, the strain readings,
deection and axial shortening measurements were recorded.
To study the softening response of the specimens, the loads
were applied at closer intervals near the ultimate load.
3. Experimental results and discussions
3.1. Test observations and failure modes
During the tests, it was observed that welding stiffeners
or adding bres to concrete had no apparent inuence on
the failure modes.

For all axially loaded specimens, a local buckling failure

mode was observed (see Fig. 3(a)). Generally, the local
buckling occurred near the top ends due to the effect of end
conditions. Concrete crushing can be found at the locations
where severe tube buckling developed, as shown in
Fig. 4(a). It seems that the overall imperfection had
moderate inuence on such kinds of axially loaded
specimens since little lateral deection was observed during
the tests.
Comparatively, all eccentrically loaded columns showed
an overall buckling failure mode, as shown in Fig. 3(b).
Similarly, concrete crushing can be found at the mid-height
of test specimens where local buckling of steel tubes
occurred (Fig. 4(b)). During the tests, the deection curves
of these specimens were approximately in the shape of halfsine wave. Specimens SCFT1-3 and UFRC2-3 are selected
to illustrate the lateral deection development of the
composite columns during the loading process, shown in
Fig. 5. The half-sine waves are also shown in Fig. 5 in
dashed lines. As can be seen, the agreement between the
test curves and the half-sine waves is better for longer
specimens due to the slenderness effect.
During the tests, the local buckling of steel tubes
occurred generally near the top ends for axially loaded
specimens, while that of eccentrically loaded specimens
occurred near the specimen mid-height. For unstiffened
specimens, the tubes buckled generally at a load level of
about 3050% of the ultimate loads. In contrast, the tube
buckling initially occurred when the maximum load had
almost reached for those stiffened specimens. The dot
points in Figs. 6 and 7 show the approximate locations
of observed local buckling of the steel tubes. It seems
that the local buckling of the tubes was effectively delayed
by the stiffeners. This demonstrates the fact that the
stiffened specimens had higher serviceability benets
compared to those unstiffened ones. In order to ensure

ARTICLE IN PRESS
Z. Tao et al. / Thin-Walled Structures 45 (2007) 517527

such serviceability requirement for local buckling, future

research should be carried out to offer design suggestions
for stiffened thin-walled CFST members.
The effect of adding bres to concrete on tube buckling
was also observed in the test program. It seems the
existence of steel bres can slightly delay the occurrence of

521

tube buckling. A similar phenomenon was also reported in

[14,15]. The main reason is attributed to the fact that the
addition of bres produces increases in compressive
toughness and in the strain corresponding to the peak
stress. Due to the strengthened structural integrity of
concrete after the bres were added, the occurrence of
concrete crushing was postponed, thus made the steel tubes
less susceptible to local buckling.
3.2. Test results and discussions

Fig. 4. Typical failure modes of the tubes and concrete core.

The maximum loads (Nue) obtained in the test are

summarized in Table 1. Fig. 6 shows axial load (N) versus
axial shortening relations (D/L) for the axially loaded
specimens, where the specimen axial shortening (D) is
normalized by the undeformed height of the specimens (L).
Axial load (N) versus mid-height lateral deection (um)
curves for those eccentrically loaded specimens are
presented in Fig. 7. As expected, the values of Nue and
exural stiffness decrease with the increasing of slenderness
ratio or load eccentricity, while the change of ductility is
generally just the opposite.
It can also be found from Figs. 6 and 7 that, welding
stiffeners or adding bres to concrete have only moderate
inuence on specimen elastic stiffness, but they do affect
the load-bearing capacity or ductility of the thin-walled
specimens to some extent.
Fig. 8 shows the effects of stiffeners and steel bres on
the ultimate strength. As can be seen, the ultimate strength
increases generally after inner-welded stiffeners were
provided except specimen SCFT2-2, which has a same
ultimate strength as the unstiffened specimen UCFT2-2.
Compared to the corresponding specimens in series UCFT,
the strength increase for stiffened specimens ranges from
1% to 17% with an average increase of 8.8%. The strength
increase is attributed to the fact that local buckling of steel
tubes was delayed when stiffeners were provided, thus the
steel tubes could carry larger axial load and provide more
connement on the concrete core. At the same time, the
stiffeners could carry a small part of load directly, thus

Fig. 5. Lateral deection along the column at different load levels.

ARTICLE IN PRESS
522

Z. Tao et al. / Thin-Walled Structures 45 (2007) 517527

Fig. 6. Axial load (N) versus normalized axial shortening (D/L) curves for axially loaded specimens.

Fig. 7. Axial load (N) versus mid-span lateral deection (um) curves for eccentrically loaded specimens.

increasing the ultimate strength [10,11]. It can also be seen

from Fig. 8 that, only moderate inuence on ultimate
strength is observed when steel bres were added to the
core concrete of slender columns, while an apparent
increase of ultimate strength was reported in [11] for stub
columns. Similar ndings were also reported in [15].
Although the concrete cube strength increased slightly

due to bre addition, the behaviour of slender columns was

strongly inuenced by steel characteristics and the presence
of SFRC did not substantively alter the ultimate strength
of columns with respect to those members lled with plain
concrete.
To quantify the effect of stiffeners and steel bres on
member ductility, a ductility index (DI) dened in [16] is

ARTICLE IN PRESS
Z. Tao et al. / Thin-Walled Structures 45 (2007) 517527

523

Fig. 8. Effect of stiffeners and steel bres on the ultimate strength.

used herein
u85%
DI
,
uy

(2)

where u85% is the mid-height deection when the load falls

to 85% of the ultimate load, uy is equal to u75%/0.75, and
u75% is the mid-height deection when the load attains
75% of the ultimate load in the pre-peak stage.
Since no apparent deection was detected for those axially
loaded specimens, a ductility index (DI) dened in [17] is used
DI

85%
,
y

(3)

where e85% is the nominal axial shortening (D/L) when the

load falls to 85% of the ultimate load, ey is equal to e75%/0.75,
and e75% is the nominal axial shortening when the load

attains 75% of the ultimate load in the pre-peak stage. The

values of e85% and ey can be determined from axial load
versus axial shortening curves shown in Fig. 6.
The ductility indexes (DI) so determined are given in
Table 1. Fig. 9 shows the effects of stiffeners and steel bres
on the ductility. As can be seen, both measures of welding
stiffeners on tube faces and adding bres to concrete can
increase the member ductility. The ductility increase ranges
from 5.5% to 39.0% for stiffened specimens and
14.557.2% for those lled with SFRC, with average
increases of 16.6% and 38.4%, respectively. It seems
adding bres to concrete is more effective in increasing the
member ductility. This is attributed to the fact that the
SFRC had higher compressive ductility and post-cracking
tensile strength due to the presence of steel bres in the
concrete.

ARTICLE IN PRESS
524

Z. Tao et al. / Thin-Walled Structures 45 (2007) 517527

Fig. 9. Effect of stiffeners and steel bres on the ductility.

Fig. 10 shows the axial load (N) versus extreme bre

strains (e) for all tested columns. It can be seen that the test
specimens exhibit more gradual strain behaviour with the
increase of the slender ratio or load eccentricity. It seems
that the inuence of welding stiffeners on tube faces and
adding bres to concrete on strain behaviour is not
considered signicant for columns generally. However,
they do affect the longitudinal strains at extreme bres in
compression at the time of local buckling (eL). The values
of eL for all specimens are given in Table 1. As can be seen
from Table 1, the values of eL for stiffened specimens are
larger than the steel yield strain and the corresponding
strains of those unstiffened specimens.

Fig. 11 shows the axial load (N) versus axial shortening

(D/L) relations for eccentrically loaded columns. As can
be seen from Fig 11, the non-linearity exhibited by
axial shortening is of the same nature as that exhibited
by mid-height deections.
4. Load-carrying capacity prediction
To evaluate the feasibility of currently available design
codes in predicting the load-carrying capacity of stiffened
thin-walled CFST columns, design equations recommended in AISC [18], BS5400 [19], DBJ1351-2003 [20]
and EC4 [21] are applied to calculate member capacities of

ARTICLE IN PRESS
Z. Tao et al. / Thin-Walled Structures 45 (2007) 517527

525

Fig. 10. Axial load versus extreme bre strains.

the test specimens in this section. It should be noted that all

partial safety factors have been taken as unity. In
calculations, some necessary modications should be made
to take the contributions of stiffeners into account. Tao
et al. [10] has presented the revised methods to calculate
ultimate compressive strength. Generally, an additional
term of As,sfy,s is included, where As,s is the total crosssectional area of the stiffeners, and fy,s is the yield strength
of the steel stiffeners. As far as the ultimate bending
moment capacity is concerned, another term of
hstsfy,s(Bhs2ts) is added to reect the contributions from
the two stiffeners welded on the tube anges. The

contributions from the remaining two stiffeners welded

on the webs are ignored since they are quite close to the
neutral axis for a stiffened composite beam. At the same
time, the contributions of stiffeners to the exural stiffness
are also taken into account.
Predicted member capacities using the different methods
are compared with test results (Nue) in Table 2, where
NAISC, NBS5400, NDBJ and NEC4 are the predicted results
using AISC [18], BS5400 [19], DBJ1351-2003 [20] and EC4
[21], respectively. The comparison results in Table 2 clearly
show that AISC [18] and BS5400 [19] are quite conservative
for predicting the member capacities of the stiffened

ARTICLE IN PRESS
Z. Tao et al. / Thin-Walled Structures 45 (2007) 517527

526

Fig. 11. Axial load (N) versus normalized axial shortening (D/L) curves for eccentrically loaded specimens.

Table 2
Comparisons between predicted ultimate strengths and test results
Specimen label

L
(mm)

e
(mm)

Nue
(kN)

NAISC
(kN)

NAISC/
Nue

NBS5400
(kN)

NBS5400/
Nue

NDBJ
(kN)

NDBJ/
Nue

NEC4
(kN)

NEC4/
Nue

UCFT1-1
UCFT1-2
UCFT1-3
UCFT2-1
UCFT2-2
UCFT2-3
Average
Standard deviation

1190
1190
1190
2340
2340
2340

0
30
60
0
30
60

2260
1760
1130
2305
1660
1068

2079
865
546
1950
842
537

0.920
0.491
0.483
0.846
0.507
0.503
0.625
0.201

1789
1251
905
1480
878
582

0.792
0.711
0.801
0.642
0.529
0.545
0.670
0.118

2182
1533
1121
1904
1223
905

0.965
0.871
0.992
0.826
0.737
0.847
0.873
0.094

2378
1706
1192
2223
1479
1062

1.052
0.969
1.055
0.964
0.891
0.994
0.988
0.062

UFRC1-1
UFRC1-2
UFRC1-3
UFRC2-1
UFRC2-2
UFRC2-3
Average
Standard deviation

1190
1190
1190
2340
2340
2340

0
30
60
0
30
60

2275
1740
1189
2285
1555
1150

2107
870
548
1975
846
539

0.926
0.500
0.461
0.864
0.544
0.469
0.627
0.210

1809
1263
912
1494
883
584

0.795
0.726
0.767
0.654
0.568
0.508
0.670
0.114

2209
1551
1133
1925
1234
913

0.971
0.891
0.953
0.842
0.794
0.794
0.874
0.077

2411
1725
1206
2252
1493
1073

1.060
0.991
1.014
0.986
0.960
0.933
0.991
0.044

SCFT1-1
SCFT1-2
SCFT1-3
SCFT2-1
SCFT2-2
SCFT2-3
Average
Standard deviation

1190
1190
1190
2340
2340
2340

0
30
60
0
30
60

2640
1770
1220
2455
1660
1200

2158
826
590
2025
901
579

0.817
0.467
0.484
0.825
0.543
0.483
0.603
0.171

1844
1261
903
1536
908
601

0.698
0.712
0.740
0.626
0.547
0.501
0.637
0.097

2255
1602
1188
1967
1283
956

0.854
0.905
0.974
0.801
0.773
0.797
0.851
0.077

2455
1763
1229
2296
1540
1102

0.930
0.996
1.007
0.935
0.928
0.918
0.952
0.039

ARTICLE IN PRESS
Z. Tao et al. / Thin-Walled Structures 45 (2007) 517527

columns. Overall, AISC [18] and BS5400 [19] give a

member capacity about 40% and 36% lower than the
results obtained in the tests. DBJ1351-2003 [20] gives a
member capacity about 15% lower than the test values.
EC4 [21] gives a slightly (within 5%) lower capacity than
the test results.
It can also be found from Table 2 that similar
comparison results can be found for those unstiffened
specimens. Since the strengthening effect of bres has been
included in the concrete strength, no obvious difference is
observed in code comparison for columns with or without
steel bres in the inll concrete. Compared to the
comparison results for stiffened specimens, it should also
be noted that the predicted results using all codes are
generally less conservative for unstiffened columns without
taken into account the local buckling effect.
5. Conclusions
An experimental study on concrete-lled stiffened thinwalled steel tubular columns was described in this paper.
Existing design codes with minor modications were used
to predict the ultimate strengths of the tested composite
columns. The ndings may be summarized as follows based
on the experimental results of the study:
1. Local buckling of steel tubes was effectively delayed by
the stiffeners, thus the stiffened columns have higher
serviceability benets compared to those unstiffened
ones.
2. The load-carrying capacity of the composite columns
can be increased when stiffeners were provided, and the
ductility can also be increased slightly.
3. Adding bres to concrete is effective in increasing the
member ductility, but its effect on ultimate strength is
moderate.
4. Existing design codes with minor modications, such as
AISC [18], BS5400 [19], DBJ1351-2003 [20] and EC4
[21], were used to predict the load-carrying capacities of
the tested composite columns. The codes are found
conservative in general, where DBJ1351-2003 [20] and
EC4 [21] give the best results.

Acknowledgements
The research reported in the paper is part of Projects
50425823 and 50608019 supported by National Natural
Science Foundation of China, and the projects supported by
Fujian Province Science and Technology (No. 2005H033), a

527

key Grant of Chinese Ministry of Education (No. 205083).

Their nancial support is highly appreciated.
References
[1] Bridge RQ, OShea MD. Behaviour of thin-walled steel box sections
with or without internal restraint. J Constr Steel Res 1998;47(12):
7391.
[2] Han LH. Concrete-lled steel tubular columnstheory and practice.
2nd ed. Beijing: Science Press; 2007 [in Chinese].
[3] Tao Z, Yu Q. New types of composite columnsexperiments, theory
and methodology. Beijing: Science Press; 2006 [in Chinese].
[4] Uy B. Strength of concrete-lled steel box columns incorporating
local buckling. J Struct Eng ASCE 2000;126(3):34152.
[5] Uy B. Novel composite steel-concrete systems incorporating highperformance steels: applications, behaviour and design. In: Proceedings of the ninth international symposium on structural engineering
for young experts, Fuzhou & Xiamen, China, 2006. p. 1120.
[6] Cai J, He ZQ. Axial load behavior of square CFST stub column with
binding bars. J Constr Steel Res 2006;62(5):47283.
[7] Huang CS, Yeh YK, Liu GY, Hu HT, Tsai KC, Weng YT, et al.
Axial load behavior of stiffened concrete-lled steel columns. J Struct
Eng ASCE 2002;128(9):122230.
[8] Ge HB, Usami T. Strength of concrete-lled thin-walled steel box
column: experiment. J Struct Eng ASCE 1992;118(11):303654.
[9] Kwon YB, Song JY, Kon KS. The structural behaviour of concretelled steel piers. In: Proceedings of 16th congress of IABSE, Iucerne,
Switzerland, 2000.
[10] Tao Z, Han LH, Wang ZB. Experimental behaviour of stiffened
concrete-lled thin-walled hollow steel structural (HSS) stub
columns. J Constr Steel Res 2005;61(7):96283.
[11] Tao Z, Han LH, Wang DY. Strength and ductility of stiffened thinwalled hollow steel structural stub columns lled with concrete. ThinWalled Struct 2007 [under review].
[12] Australia Standard. AS4100-1998 steel structures. Sydney, 1998.
[13] Tao Z, Han LH. Behaviour of concrete-lled double skin rectangular
steel tubular beam-columns. J Constr Steel Res 2006;62(7):63146.
[14] Campione G, Mendola LL, Sanpaolesi L, Scibilia N, Zingone G.
Behavior of ber reinforced concrete-lled tubular columns in
compression. Mater Struct 2002;35:3327.
[15] Campione G, Scibilia N. Beam-column behavior of concrete-lled
steel tubes. Steel Compos Struct 2002;2(4):25976.
[16] Tao Z, Han LH. Behaviour of re-exposed concrete-lled steel
tubular beam columns repaired with CFRP wraps. Thin-Walled
Struct 2007;45(1):6376.
[17] Lin ML, Tsai KC. Behaviour of double-skinned composite steel
tubular columns subjected to combined axial and exural loads. In:
Proceedings of the rst international conference on the steel &
composite structures, Pssan, Korea, 2001. p. 114552.
[18] ANSI/AISC 360-05. Specication for structural steel buildings.
Chicago, IL, USA: American Institute of Steel Construction; 2005.
[19] BS5400. Steel, concrete and composite bridges. Part 5, Code of
practice for the design of composite bridges, London, UK, 2005.
[20] DBJ1351-2003. Technical specication for concrete-lled steel
tubular structures. Fuzhou, China: The Construction Department
of Fujian Province; 2003 [in Chinese].
[21] Eurocode 4. Design of composite steel and concrete structures, Part1.
1, general rules and rules for Building. BS EN 1994-1-1: 2004.
London: British Standards Institution.