Journal of Constructional Steel Research 61 (2005) 902911

www.elsevier.com/locate/jcsr

Tests on high-strength rectangular concrete-filled

steel hollow section stub columns
Dalin Liu
School of Civil and Environmental Engineering, Nanyang Technological University,
Singapore 639798, Singapore

Received 11 August 2004; accepted 4 January 2005

Abstract

An experimental programme on high-strength rectangular concrete-filled steel hollow section

columns is reported. A total of 22 specimens were tested under concentric loading. The test variables
include the material strengths ( f y = 495 MPa; fc = 60 and 89 MPa), cross-sectional aspect ratio
(1.02.0) and volumetric steel-to-concrete ratio (0.130.17). Favourable ductility performance was
observed for the specimens in the tests. The experimental results are employed to calibrate the
specifications in the codes (EC4, ACI and AISC) for the design of high-strength composite columns.
Comparison indicates that the specifications in ACI and AISC conservatively estimate the failure
loads of the specimens by 9 and 11%, respectively. The EC4 method gives a close and conservative
estimate of the ultimate capacities with a difference of 1%.
2005 Elsevier Ltd. All rights reserved.
Keywords: Column; Concentric loading; Concrete; High strength; Steel

1. Introduction

Concrete-filled steel hollow section (CFSHS) columns have become increasingly

popular in a variety of structural applications, such as building structures, bridges and
subway platforms. This can be attributed to their several advantages over steel hollow

Tel.: +65 67905086; fax: +65 67921650.

E-mail address: [email protected].

0143-974X/$ - see front matter 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jcsr.2005.01.001
 D. Liu / Journal of Constructional Steel Research 61 (2005) 902911 903

Nomenclature

Ac Cross-sectional area of concrete core

As Cross-sectional area of steel hollow section
B Breadth of column
Ec Modulus of elasticity of concrete
Es Modulus of elasticity of steel
fc Cylinder strength of concrete
fy Yield stress of steel
H Height of column
L Length of column
Nc Calculated axial load capacity of column
Ne Experimental axial load capacity of column
t Wall thickness of steel hollow section

section and reinforced concrete columns, such as high axial load capacity and favourable
seismic performance. From a viewpoint of construction, economy can be achieved since
the steel hollow section of CFSHS columns can serve as formwork for the concrete infill
during construction.
With the advent of high-strength steel and the production of high-strength concrete
using conventional materials with careful quality control, high-strength CFSHS columns
are both technically and economically feasible. However, they are scarcely adopted in
the construction industry, mainly due to the lack of understanding of their structural
behaviour and reliable design recommendations. The present design codes, such as
Eurocode 4 (EC4) [1], American Concrete Institute (ACI) [2] and American Institute
of Steel Construction (AISC) [3] are only applicable to the design of CFSHS columns
fabricated from normal-strength materials. There is uncertainty whether they can be used
for high-strength composite columns. A research need exists to examine the behaviour
of high-strength CFSHS columns under various types of loading and to develop design
recommendations.

2. Previous research

It is well known that the strength and ductility of CFSHS columns are considerably
higher than those of reinforced concrete columns of similar dimensions. The increase in
strength and ductility is affected by several parameters, such as slenderness, strength of
materials, shape of cross-section and volumetric steel-to-concrete ratio. Many experimental
and analytical studies have been reported since the 1960s to investigate the effects of these
parameters on the behaviour of CFSHS columns.
904 D. Liu / Journal of Constructional Steel Research 61 (2005) 902911

Schneider [4] carried out concentric loading tests on 14 circular and rectangular CFSHS
columns with steel yield stresses ( f y ) varying from 312 to 430 MPa and concrete cylinder
strength ( f c ) between 24 and 31 MPa. Results suggested that the axial load behaviour
was significantly affected by the shape of cross-section and breadth-to-thickness ratio.
Uy [5] performed experiments on 19 square CFSHS columns with f y = 750 MPa and
fc = 2832 MPa. Test results indicated that the EC4 approach was unsafe for the
specimens subjected to pure compression or pure bending. The reason was attributed to
the rigid plastic analysis adopted in EC4, which assumed fully crushed concrete and fully
yielded steel. From the axial loading tests on 24 rectangular CFSHS stub columns with
f y = 194228 MPa and f c = 49 MPa, Han [6] concluded that the structural behaviour
of rectangular CFSHS columns was influenced by the cross-sectional aspect ratio, material
properties and confinement factor.
In recent years, much attention has been paid to the behaviour of CFSHS columns
constructed from high-strength steel and concrete. Based on the test results of 41 circular
CFSHS columns with f y = 410435 MPa and f c = 5896 MPa, Kilpatrick and
Rangan [7] confirmed that the slenderness and load eccentricity significantly affected the
strength of columns. Varma et al. [8] investigated the behaviour of eight square CFSHS
columns with f y = 269660 MPa and f c = 110 MPa. Results showed that the ACI
and AISC provisions conservatively predicted the failure loads of the specimens by 6 and
65%, respectively. The EC4 method, on the other hand overestimated the failure loads
by 6%. Liu et al. [9] reported a set of tests on 22 rectangular CFSHS columns with
f y = 550 MPa and f c = 61 and 72 MPa. Results suggested that the provisions in EC4,
ACI and AISC conservatively predicted the failure loads of the specimens by 6, 14 and
16%, respectively. Furthermore, Gho and Liu [10] studied the flexural behaviour of 12
rectangular CFSHS columns with f y = 409495 MPa and f c = 5691 MPa. Results
indicated that the provisions in EC4, ACI and AISC conservatively predicted the bending
resistance by 11, 15 and 18%, respectively. Sakino et al. [11] carried out concentric loading
tests on 36 circular and 48 square CFSHS stub columns. The test parameters were material
strengths ( f y = 262853 MPa; f c = 2591 MPa) and volumetric steel-to-concrete ratio
(0.090.26). Test results showed a strength increase for square CFSHS columns. However,
the strength improvement was attributed to the strain hardening of the steel section instead
of the confinement effect. Fujimoto et al. [12] reported eccentric loading tests on 33
circular and 32 square CFSHS stub columns. The test variables were material strengths
( f y = 262853 MPa; f c = 2585 MPa), axial load level (0.150.80) and volumetric steel-
to-concrete ratio (0.030.29). Ductility improvement due to the confinement by the steel
section was observed in the tests. For square CFSHS columns, the enhancement of bending
resistance due to the confinement effect was found to be minimal.
Apart from the experimental investigations, analytical studies with varying levels of
complexity have been reported in the literature to analyse and design CFSHS columns. For
example, Susantha et al. [13] and OShea and Bridge [14] proposed fibre-based models
for numerical analyses on CFSHS columns. The fibre-based models were based on strain
compatibility and uniaxial constitutive relationships. On the other hand, finite element
programmes were also employed by a number of researchers (Schneider [4]; Ge and
Usami [15]) to analyse the structural behaviour of CFSHS columns.
 D. Liu / Journal of Constructional Steel Research 61 (2005) 902911 905

Fig. 1. Cross-sectional view of CFSHS column.

After the review of the previous research projects pertaining to CFSHS columns, it
can be concluded that the experimental investigation on rectangular CFSHS columns,
especially high-strength rectangular CFSHS columns is still insufficient. There is a
research need to provide a set of test data and to calibrate the current design codes for
the design of high-strength rectangular CFSHS columns. The current study is thus carried
out to examine the behaviour of high-strength rectangular CFSHS columns subjected to
concentric loading.

3. Experimental investigation

A total of 22 specimens in four series were fabricated and tested in the research
programme. Specimens in each series had uniform material strengths and volumetric steel-
to-concrete ratio ( As /Ac , As and Ac are cross-sectional areas of steel section and concrete
core, respectively), but different cross-sectional aspect ratio (H /B, H and B are height and
breadth, respectively). The material properties of steel and concrete were obtained from
respective tensile and crushing tests. The ultimate capacities and axial loadshortening
relationships of the specimens were recorded for analysis.

3.1. Material properties

A steel hollow section was formed by welding four component flat plates together, as
shown in Fig. 1. All the flat plates were cut from a steel sheet, which had a nominal yield
stress of 450 MPa. Three coupons were prepared from the steel sheet according to the
requirements in ASTM [16] and were tested at a loading rate of 0.5 mm/min before the
yield load and 2 mm/min after the yield load. The average yield stress, tensile strength and
modulus of elasticity were determined as 495, 581 and 206,000 MPa, respectively.
Concrete was mixed in two batches. From each batch, four concrete cubes (150 150
150 mm) and four cylinders (diameter of 150 mm and height of 300 mm) were prepared.
In order to simulate the actual curing condition of the concrete inside the steel hollow
sections, the concrete samples were cured in the moulds covered by wet hessian. On the
same day as the CFSHS specimens were tested, the moulds were stripped from the concrete
samples. The two end surfaces of the concrete cylinders were ground flat before crushing
906 D. Liu / Journal of Constructional Steel Research 61 (2005) 902911

Table 1
Details of high-strength concrete

Batch no. A B

Cement (kg/m3 ) 550 520

Silica fume (kg/m3 ) 0 90
Fly ash (kg/m3 ) 70 0
Water (kg/m3 ) 158 140
Maximum aggregate size (mm) 10 10
Coarse aggregate (kg/m3 ) 1170 1150
Sand (kg/m3 ) 530 550
Modulus of elasticity (MPa) 38000 41000
Average cylinder strength (MPa) 60 89
Average cube strength (MPa) 73 97

tests. Table 1 summarises the details of the mix proportions and the measured material
properties of the two batches of concrete.

3.2. Fabrication of specimens

After the fabrication of the steel hollow section had been completed, a 10-mm-thick flat
plate was welded to its base for holding the wet concrete during construction. Concrete
was then vertically cast into the steel hollow section in three layers. Each layer of concrete
was compacted by a poker vibrator. Then the concrete was cured inside the steel hollow
section with top open to the air for two weeks until a 10-mm-thick flat plate was welded
to the top to form a complete specimen. The geometric and the material properties of the
specimens are summarised in Table 2.

3.3. Test set-up

The specimens were tested 32 days after concrete was cast. The axial load was
applied by a 5000-kN-capacity universal testing machine in the Structures Laboratory,
Fuzhou University. As illustrated in Fig. 2, four displacement transducers were placed
symmetrically around the specimen between the two plattens of the testing machine to
measure the overall deformation. Four strain gauges were bonded on the exterior surfaces
of the steel section at mid-height to record the longitudinal strain.

3.4. Test procedure and results

Prior to the actual tests, a load level of 25 kN was applied so that the plattens of
the testing machine were firmly attached to both ends of the specimen. The axial load
was then applied slowly by careful manipulation of the loading and unloading valves.
The readings of the applied load, axial shortening and longitudinal strain were recorded
at a load increment of 100 kN. The experimental failure load (Ne ) for each specimen
was recorded, as summarised in Table 2. Fig. 3 gives typical axial loadshortening curves
measured in the tests. Favourable ductility performance was noticed for most specimens
except Specimens R8-R10. Crushing of the concrete core occurred for these six specimens
 D. Liu / Journal of Constructional Steel Research 61 (2005) 902911 907

Table 2
Summary of specimens

Series No. H B t L H /B f c fy As / Ac Ne Ne /Nc

(mm) (mm) (MPa) (MPa) (kN) EC4 ACI AISC

A R1-1 120 120 4 360 1.0 60 495 0.15 1701 1.02 1.09 1.10
R1-2 120 120 4 360 1.0 60 495 0.15 1657 0.99 1.06 1.07
R2-1 150 100 4 450 1.5 60 495 0.15 1735 1.00 1.07 1.08
R2-2 150 100 4 450 1.5 60 495 0.15 1778 1.02 1.09 1.11
R3-1 180 90 4 540 2.0 60 495 0.15 1773 0.94 1.01 1.03
R3-2 180 90 4 540 2.0 60 495 0.15 1795 0.95 1.02 1.05
B R4-1 130 130 4 390 1.0 60 495 0.14 2020 1.07 1.15 1.16
R4-2 130 130 4 390 1.0 60 495 0.14 2018 1.07 1.15 1.16
R5-1 160 110 4 480 1.5 60 495 0.14 1982 1.01 1.08 1.10
R5-2 160 110 4 480 1.5 60 495 0.14 1923 0.98 1.05 1.07
R6-1 190 100 4 570 1.9 60 495 0.14 2049 0.97 1.04 1.06
R6-2 190 100 4 570 1.9 60 495 0.14 2124 1.00 1.08 1.10
C R7-1 106 106 4 320 1.0 89 495 0.17 1749 1.05 1.14 1.15
R7-2 106 106 4 320 1.0 89 495 0.17 1824 1.10 1.19 1.20
R8-1 130 90 4 390 1.4 89 495 0.17 1752 1.01 1.10 1.11
R8-2 130 90 4 390 1.4 89 495 0.17 1806 1.04 1.13 1.15
R9-1 160 80 4 480 2.0 89 495 0.17 1878 0.99 1.08 1.11
R9-2 160 80 4 480 2.0 89 495 0.17 1858 0.98 1.06 1.09
D R10-1 140 140 4 420 1.0 89 495 0.13 2752 1.05 1.15 1.16
R10-2 140 140 4 420 1.0 89 495 0.13 2828 1.08 1.18 1.19
R11-1 160 125 4 480 1.3 89 495 0.13 2580 0.96 1.06 1.07
R11-2 160 125 4 480 1.3 89 495 0.13 2674 1.00 1.09 1.11
Average 1.01 1.09 1.11
SD 0.04 0.05 0.05
COV 0.04 0.04 0.04

Fig. 2. Test set-up: (a) specimen in the testing machine; (b) strain gauge locations.

shortly after the critical loads. This indicates that CFSHS columns with concrete cylinder
strength as high as 89 MPa can exhibit poor ductility performance when tested in a
908 D. Liu / Journal of Constructional Steel Research 61 (2005) 902911

Fig. 3. Typical axial loadshortening curves: (a) specimens with f c = 60 MPa; (b) specimens with f c = 89 MPa.

Fig. 4. Typical failure modes of specimens: (a) steel section; (b) concrete core.

load-control testing machine. For these six specimens, only a small descending branch
of the axial loadshortening curve was recorded. For the other specimens, the applied
load decreased slowly and steadily in the post-peak region. This suggests the significant
ductility enhancement of the high-strength concrete core due to the confinement by the
steel section. As shown in Fig. 4(a), buckling occurred almost simultaneously at every face
for specimens with H /B = 1.0. For specimens with H /B = 1.32.0, buckling was more
extensive and occurred earlier at the broader faces. Fig. 4(b) gives a photograph of the
failure modes of the concrete core. For Specimens R4-2, R5-2 and R6-2, slight crushing
of the concrete core was observed close to where the steel section locally buckled, while
severe crushing of the concrete core was noticed for Specimen R8-1.

3.5. Effect of cross-sectional aspect ratio on the strength increase of concrete

Rectangular steel hollow sections are generally considered to be less effective than
circular ones in terms of offering confinement to the concrete core. This is because the
rectangular steel section confines the concrete core by plate bending while the circular
section by hoop stress. With the increase of H /B ratio, the strength increase of the concrete
core due to confinement effect is inclined to decrease. Fig. 5 gives the correlation of the
 D. Liu / Journal of Constructional Steel Research 61 (2005) 902911 909

Fig. 5. Effect of cross-sectional aspect ratio on strength index.

H /B ratio and the strength index (SI) for the specimens tested in the study. SI is defined
as in Eq. (1).
SI = (Ne As f y )/(0.85 Ac f c ). (1)
In Fig. 5, the specimens in each series have identical material strengths and volumetric
steel-to-concrete ratio, but their H /B ratios are different. Results illustrate that the
strength increase of the concrete is adversely influenced by the H /B ratio. Therefore,
CFSHS columns with H /B > 2.0 are not recommended for design, since little strength
improvement is expected for these composite columns with high cross-sectional aspect
ratios.

4. Comparison of results between tests and the design codes

In this section, the experimental results are employed to calibrate the provisions in the
design codes (EC4, ACI and AISC) for the design of high-strength composite columns.
Since the material properties of the specimens are known, all the material safety factors
specified in the codes are set to unity in the calculation.

4.1. Code provisions

According to EC4, the axial load capacity (Nc ) of a CFSHS column can be determined
by summing up the yield load of the steel section and that of the concrete core. The
coefficient of 0.85, which applies to reinforced concrete columns is set to 1.0 to incorporate
the confinement effect, as shown in Eq. (2). The application of EC4 is restricted to
composite columns with steel yield stress and concrete cylinder strength lower than 355
and 50 MPa, respectively.
Nc = As f y + Ac f c . (2)
Similar to EC4, ACI predicts the axial load capacity of CFSHS columns by summing
up the strengths of the steel section and the concrete core. However, a coefficient of 0.85
is included in the concrete cylinder strength to account for long-term and size effects, as
shown in Eq. (3). The strength increase of the concrete core resulting from the confinement
effect is not considered.
Nc = As f y + 0.85 Ac f c . (3)
910 D. Liu / Journal of Constructional Steel Research 61 (2005) 902911

The AISC approach for the design of CFSHS columns is essentially identical to that of
steel columns, as shown in Eq. (4). The use of AISC specifications is limited to composite
columns with steel yield stress and concrete cylinder strength not higher than 415 and
55 MPa, respectively.

Nc = As fcr (4)

in which,

f cr = (0.658c ) f my
2
for c 1.5
f cr = (0.877/2c ) f my for c > 1.5
L

c = f my /(E s + 0.4E c Ac /As )
rs
f my = f y + 0.85 f c Ac /As
rs = radius of gyration of steel hollow section
L = length of column
E c = modulus of elasticity of concrete
E s = modulus of elasticity of steel.

4.2. Comparison of failure loads between tests and the design codes

The experimental failure loads of the 22 specimens are compared with the predicted
values using the provisions in EC4, ACI and AISC. As summarised in Table 2, all three
design codes are conservative to predict the ultimate capacities. Among the codes, ACI
and AISC give safe calculation of the failure loads by 9 and 11%, respectively. The EC4
method, on the other hand is more accurate than ACI and AISC with an average Ne /Nc
ratio of 1.01 and a coefficient of variation (denoted as COV in Table 2) of 0.04. Results
indicate that the EC4 approach can be extended to the design of high-strength rectangular
CFSHS columns subjected to concentric loading.

5. Conclusions

An experimental programme on 22 high-strength rectangular CFSHS stub columns

subjected to concentric loading has been described. Test results manifest the favourable
ductility performance of the high-strength composite columns. The strength increase of
the concrete resulting from confinement by the steel section is observed. The strength
improvement is adversely affected by the cross-sectional aspect ratio. Comparison of
failure loads between the tests and the design codes has been presented. Results show that
the provisions in EC4, ACI and AISC conservatively estimate the ultimate capacities of
the specimens by 1, 9 and 11%, respectively. The EC4 approach can be used for estimating
the ultimate capacity of high-strength rectangular CFSHS columns subjected to concentric
loading.
 D. Liu / Journal of Constructional Steel Research 61 (2005) 902911 911

References
[1] British Standards Institution. Eurocode 4, ENV 1994-1-1. Design of composite steel and concrete structures.
Part 1.1: General rules and rules for buildings. London; 1994.
[2] American Concrete Institute. Building code requirements for reinforced concrete and commentary. ACI
318-02/ACI 318R-02, ACI Committee 318. Detroit; 2002.
[3] American Institute of Steel Construction. Load and resistance factor design specification for structural steel
buildings. Chicago; 1999.
[4] Schneider SP. Axially loaded concrete-filled steel tubes. J Struct Eng 1998;124(10):112538.
[5] Uy B. Strength of short concrete filled high strength steel box columns. J Construct Steel Res 2001;57:
11334.
[6] Han L-H. Tests on stub columns of concrete-filled RHS sections. J Construct Steel Res 2002;58:35372.
[7] Kilpatrick AE, Rangan BV. Tests on high-strength concrete-filled steel tubular columns. ACI Struct J 1999;
96(2):26875.
[8] Varma AH, Ricles JM, Sause R, Lu L-W. Experimental behaviour of high strength square concrete-filled
steel tube beam-columns. J Struct Eng 2002;128(3):30918.
[9] Liu D, Gho W-M, Yuan J. Ultimate capacity of high-strength rectangular concrete-filled steel hollow section
stub columns. J Construct Steel Res 2003;59:1499515.
[10] Gho W-M, Liu D. Flexural behaviour of high-strength rectangular concrete-filled steel hollow sections.
J Construct Steel Res 2004;60:168196.
[11] Sakino K, Nakahara H, Morino S, Nishiyama I. Behaviour of centrally loaded concrete-filled steel-tube
short columns. J Struct Eng 2004;130(2):1808.
[12] Fujimoto T, Mukai A, Nishiyama I, Sakino K. Behaviour of eccentrically loaded concrete-filled steel tubular
columns. J Struct Eng 2004;130(2):20312.
[13] Susantha KAS, Ge H, Usami T. Uniaxial stressstrain relationship of concrete confined by various shaped
steel tubes. Eng Struct 2001;23:133147.
[14] OShea MD, Bridge RQ. Design of circular thin-walled concrete filled steel tubes. J Struct Eng 2000;
126(11):1295303.
[15] Ge H, Usami T. Strength analysis of concrete-filled thin-walled steel box columns. J Construct Steel Res
1994;30:25981.
[16] American Society for Testing and Materials. Test methods for tension testing of metallic materials. West
Conshohocken; 1997.