Construction and Building Materials 28 (2012) 88–95

Contents lists available at SciVerse ScienceDirect

Construction and Building Materials

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

Analysis of steel-reinforced concrete-filled steel tubular (SRCFST) columns

under cyclic loading
Xu Chang a,⇑, You-Yi Wei b, Yan-Chun Yun b
a
College of Civil Engineering, Henan Polytechnic University, Jiaozuo City, Henan Province 454000, People’s Republic of China
b
Department of Building and Construction, City University of Hong Kong, Hong Kong, People’s Republic of China

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

Article history: A new form of composite column, steel reinforced-concrete filled-steel tubular column (SRCFST) has been
Received 25 February 2011 proposed to undertake higher loads. This contribution presents a numerical study of cyclically loaded
Received in revised form 29 July 2011 SRCFST columns based on the ABAQUS standard solver. The feasibility and accuracy of the numerical
Accepted 9 August 2011
method was verified by comparing the calculated results with the experimental observations. The lateral
Available online 1 October 2011
displacement–load curves and sectional stress distributions were analyzed. The results indicate that the
SRCFST columns have higher specimen stiffness, peak lateral load and deformability than common con-
Keywords:
crete filled steel tubular (CFST) columns due to the presence of the section steel. A parametric study,
Steel-reinforced concrete-filled steel tube
Cyclic loading
including influence of axial load levels, ratio of section steel, yield strength of section steel, concrete
Lateral displacement strength and thickness of steel tube on peak lateral load was also carried out.
Crown Copyright Ó 2011 Published by Elsevier Ltd. All rights reserved.

1. Introduction
Over the last several decades, the concrete-filled steel tubular
(CFST) structure is widely used in the construction of modern
buildings and bridges [1–3], even in regions of high seismic risk.
This composite construction ideally combines the advantages of
both steel tube and concrete, namely the speed of construction
and high strength. Moreover, they have lighter weight, higher
bending stiffness, and better cyclic performance than the rein-
forced concrete construction. In recent years, a large structure or
group of structures arises in China and some other place. With
the increase of span-length and height, the cross area of a column
is often designed bigger to provide higher bearing capacity. For
example, the diameter of CFST columns in first story of ShenZhen
Saibo Plaza Building reaches 1600 mm. Such a large cross area of
a column results in a reduced useful indoor area.
Steel reinforced concrete (SRC) structural members are com-
posed of a concrete, a section steel, longitudinal steel bars and
transverse steel bars. They are widely used due to their advantages
in term of high sectional strength. However, it is a very compli-
cated process for construction.
A new form of composite column, steel reinforced-concrete
filled-steel tubular column (SRCFST), has been proposed recently
[4,5]. The new column consists of a steel reinforced concrete inside
and a steel tube outside, as shown in Fig. 1. No steel bars are used
in this composite column and the outside steel tube can provide
⇑ Corresponding author. Tel./fax: +86 411 84707740.
E-mail address: changxu815@163.com (X. Chang).
confining effect to keep the concrete and section steel work to-
gether. The new column is an attempt to combine the advantages
of the SRC column and the concrete filled steel tubular column, so
as to achieve a high performance structural member.
Most of existing studies focused on mechanical performances of
SRCFST columns under axial load or bending both in theoretical
analyses and tests [4–7]. Compared with the studies made in the
fields mentioned above, fewer experimental studies have been car-
ried out on the behaviors of SRCFST columns under cyclic loading
[8].
In order to give a further understanding of such a new column,
in this contribution, the ABAQUS/standard solver is employed to
investigate and predict the resistances of SRCFST columns under
cyclic loading. Validation of this numerical method is carried out
by comparing the computed results with the experimental obser-
vation of five tested specimens. A parametric study, including the
thickness of steel tube, steel ratio of section steel, yield strength
of section steel and strength of concrete, is also carried out.
2. Numerical modeling
2.1. Materials properties
The steel is assumed to behave as an elastic–plastic material
with strain hardening after yield strength. The Mises yield surface
is used to define isotropic yielding for steel material and the model
assumes associated plastic flow. To describe the behaviors of steel
under cyclic loading, linear kinematic hardening model is adopted
to ease the modeling of the composite columns. Steel properties

0950-0618/$ - see front matter Crown Copyright Ó 2011 Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.conbuildmat.2011.08.033
 X. Chang et al. / Construction and Building Materials 28 (2012) 88–95 89

Axial Load
Steel tube

Concrete core

+ steel

Cyclic load
Fig. 1. Cross section for SRCFST column.

specified in ABAQUS include a Young’s modulus of steel (Es), a Pois-

son’s ratio ls and a yield strength. A hardening rigidity of 1% Es
after yield has been adopted.
The damaged plasticity model defined in Standard ABAQUS [9]
is used in the analysis. By using the finite element method,
strength improvement at the state of triaxial loading can be
achieved by the definition of the yielding surface, and the descrip- Fig. 2. Sketch of test arrangement for cyclic loaded SRCFST column.
tion of the plastic behavior comes from the equivalent stress–
strain relationship of concrete core. Many stress–strain models
have been proposed [10–13]. The widely accepted equivalent
stress–strain model proposed by Han et al. [14] is used in this pa-
per, which is based on a large amount test results. As the model has
been fully documented by Han et al. [14], so we only briefly de-
scribes as following:
(
2x  x2 ðx 6 1Þ
y¼ x ð1Þ
b0 ðx1Þg þx
ðx > 1Þ

where x = e/e0, y = r/r0. r0 is the compressive strength of concrete

core and e0 is the corresponding strain. b0 and g are parameters
for controlling the shape of the curve. More details about the defi-
nition of parameters can be found in reference [14].
Poisson’s ratio lc in the elastic part of concrete under uniaxial
compression stress ranges from 0.15 to 0.22, with a representative
value of 0.19, according to ASCE [15]. In this numerical modeling, (a) (b) (c)
Poisson’s ratio of concrete is taken as 0.19. The other parameters,
such as dilation angle, eccentricity, ratio of the biaxial compression Axial load
strength to uniaxial compression strength of concrete, the ratio of
End plate
the second stress invariant on the tensile meridian to that on the
compressive meridian and viscosity parameter are default by
ABAQUS [9].
For the concrete under cyclic loading, it is also necessary to
define the tensile behavior in the damaged plasticity model of
concrete. The equivalent stress–strain relationship of concrete core
under tension proposed by Guo [16] is used:
(
1:2  x  0:2  x6 ðx 6 1Þ
y¼ x ð2Þ
0:31r2p ðx1Þþx
ðx > 1Þ

where x = e/ep, y = r/rp. rp and ep rp are the tensile strength of con- Cyclic load
crete and the corresponding strain, which can be calculated as
follows:
Symmetry plane
rp ¼ 0:26  ð1:25  fc Þ2=3 ðMPaÞ ð3Þ
(d)
ep ¼ 43:1  rp ðleÞ ð4Þ
Fig. 3. Element divisions and boundary conditions. (a–c) are element divisions for
where fc is the compressive strength of concrete. concrete core, section steel and steel tube, respectively, (d) is for boundary
conditions. Axial load applied through elastic end plate and lateral load applied on
symmetry plane at mid-height of column.
2.2. Finite element mesh and boundary conditions

Three-dimensional eight-node linear brick and reduced integra- to be used for concrete the most effective element type for concrete
tion with hourglass control solid element (C3D8R) has been proved [17–19] For steel tube, both shell elements and solid elements were
90 X. Chang et al. / Construction and Building Materials 28 (2012) 88–95

Table 1
Geometrical and material parameters for all test specimens.

Specimen ID Steel tube Section steel Concrete compressive strength fcu (MPa) n
D  t (mm) Yield strength, fy1 (MPa) I steel As (mm2) Yield strength, fy2 (MPa)
HC12-1 218  4 269 I12 3570 314 74.3 0.50
HC12-2 218  4 269 I12 3570 314 74.3 0.66
HC14-1 218  4 269 I14 4160 305 74.3 0.50
HC14-2 218  4 269 I14 4160 305 74.3 0.66
HC14-3 218  4 269 I14 4160 305 74.3 0.50

600 600
Test
Numerical
300 300
P (kN)

P (kN)
0
0

-300
-300
Numerical
-600 Test
-40 -20 0 20 40
-600
δ (mm) -20 -10 0 10 20 30
δ (mm)
(a) HC12-1
(b) HC12-2
800 800
Test Test
Numerical
Numerical
400
400
P (kN)

0
P (kN)

-400
-400

-800
-40 -20 0 20 40
δ (mm) -800
-40 -20 0 20 40
(c) HC14-1 δ (mm)
(d) HC14-2

600
Test
Numerical
300
P (kN)

-300

-600
-20 -10 0 10 20
δ (mm)
(e) HC14-3
Fig. 4. Comparison of numerical cyclic load–lateral displacement curves with experimental ones.

adopted in the previous research work [14,17]. In this paper, the ment sizes were built to select the reasonable mesh that provides
C3D8R element was also used for the steel tube and section steel accurate results with lesser computational time. It was found that
to ease the model set-up. A series of FE models with different ele- a mesh size of 20 mm was the appropriate one for concrete core.
 X. Chang et al. / Construction and Building Materials 28 (2012) 88–95 91

The mesh size for both steel tube and section steel was 18 mm. Fig. 2 600
gives the sketch of test arrangement [8]. Fig. 3 shows a schematic B
view of the element divisions for the concrete core (Fig. 3a), section 400
steel (Fig. 3b) steel tube (Fig. 3c), and composite columns (Fig. 3d),
200 A
where only half of a column was modeled due to the symmetry of
the composite columns.

P (kN)
E O
0
A surface-based interaction was used to represent the contact
C
between the steel tube and the concrete. The normal direction of
-200
the two surfaces is hard contact and the tangent contact is simu-
lated by the Coulomb friction model. The hard contact between D
-400
the concrete and the steel tube prevents the separation between
the surfaces. Selection of coefficient of friction is difficult because -600
there is no standard test procedure to determine it. According to -15 -10 -5 0 5 10 15
the existing studies [14–19], coefficient of friction used for CFST δ (mm)
columns under axial compression ranged from 0.3 to 0.5. In this
Fig. 5. Different process for the cyclic load–lateral displacement curve.
paper, different friction coefficients ranging from 0.1 to 0.7 was se-
lected to explore their effects on column behaviors. The results
indicates that friction coefficient plays little role in column behav-
iors. A convergent problem was induced when the friction coeffi- Table 2
Geometrical and material parameters for numerical model.
cient was greater than 0.8. Therefore a friction coefficient of 0.6
was suggested to achieve a quick convergence. Steel tube Section steel fcu (MPa) n
The ‘‘embedded region’’ constraint defined in ABAQUS was used D  t (mm) fy1 (MPa) I steel As (mm ) 2
fy2 (MPa)
to describe the contact between concrete and section steel.
218  4 270 I12 3570 320 75 0.50
The boundary conditions were indicated in Fig. 3d. The axial
force was firstly applied at the middle line of the elastic end plate
with an elastic modulus of 1010 MPa and a Poisson’s ratio of 0.0001
and kept constant during the whole calculated process. The hori- 500
3
zontal displacement of the middle line of the elastic end plate
2 4
was also fixed. The horizontal load was applied at symmetry plane 400
at the column mid-height in an appointed displacement using
boundary condition under different increment steps. 300
P (kN)

2.3. Model validation 200

To validate the accuracy of the numerical model for the com-
posite columns described in the previous section, five previously
published test specimens [8] were used for comparison purpose.
These five specimens can be divided into two groups (HC12 and
HC14) according to the cross section area of section steel. In each
specimen group, axial load level ranges from 0.5 to 0.66. The geo-
metrical and material parameters for all the test specimens are
summarized in Table 1. In this table, fy1 and fy2 are yield strengths
for the steel tube and the section steel, respectively. fcu is cube
strength for concrete.
It should be noted that the ‘+’ steel was made by two ‘I’ steel in
the test. Two types of ‘I’ steel were used in the test. Fig. 4 gives the
comparison of computed cyclic load (P) versus lateral deflection (d)
curves with the tested ones. It can be found that, generally good
agreement is obtained between the predicted and tested results.
This confirms that the present numerical model can be used with
confidence to simulate the behaviors of the SRCFST columns under
cyclic loading.
3. Analysis and discussion
3.1. Complete curve analysis
The computed cyclic load (P) versus lateral deflection (d) curve
for the composite columns can be generally characterized as four
different processes, loading, unloading, reverse loading and reverse
unloading process, as shown in Fig. 5. The calculated conditions are
listed in Table 2. The typical P–d envelope curve is presented in
Fig. 6. It can be seen that the envelope curve consists of three
different stages, elastic stage, elastic–plastic stage and post peak
stage.
100
0
0 10 20 30 40 50
δ (mm)
Fig. 6. Typical computed cyclic load–lateral displacement envelope curve.
(1) Elastic stage: (from point 0 to point 1). In this stage, the load
against later displacement curve shows linear behavior. Steel
tube and section steel are in elastic phase. As the increasing of
lateral displacement, the steel tube yield at point 1.
(2) Elastic–plastic stage: (from point 1 to point 2). The stiffness
of the specimen deviated from its initial value and the
response becomes non-linear. Fig. 7 shows the distributions
of the longitudinal stress (S33 in the graphs) in the cross sec-
tions when the curve reaches point 1 and point 2. It is found
that the steel tube in the compressive zone begins to yield
firstly when later displacement reaches point 1. The stress
of the concrete in compressive zone exceeds the cylinder
strength due to confinement of the steel tube. Both the stres-
ses in tensile zone and compressive zone of the section steel
are lower than those of the steel tube. For the section steel,
no yielding occurs at point 1. With the increasing of lateral
displacement, the yielding areas of the steel tube and the
tensile area of the concrete core increase and the stiffness
of the curve decreases gradually. The stress of the section
steel also increases. When the curve reaches point 2, yield-
ing of the section steel occurs. Point 3 refers to the peak lat-
eral load of the composite columns.
(3) Post peak stage: The curve descends from point 3.
92 X. Chang et al. / Construction and Building Materials 28 (2012) 88–95
Fig. 7. Longitudinal stress distributions: (a) cross section of steel tube for point 1; (b) cross section of steel tube for point 2; (c) cross section of section steel for point 1; (d)
cross section of section steel for point 2.
3.2. Analysis of section steel
The presence of section steel is a special character of SRCFST
column and a marked difference compared with common con-
crete-filled steel tubular (CFST) column. Therefore, the role of sec-
tion steel is discussed in detail in this section. A CFST column is
also modeled for comparing purpose. Its geometrical and material
parameters are the same as those of SRCFST column.
Fig. 8 shows the distributions of the longitudinal stress (S33 in
the graphs) in the cross sections of the concrete core both for CFST
column and SRCFST column when the steel tube in compressive
zone reaches yielding strength (point 1 indicated in Fig. 6). It can
be found from these figures that, generally, the tensile zone in
the concrete section of CFST column is roughly equal to that of
SRCFST column. However, the tensile zone in the concrete section
of SRCFST column is less than that of CFST column when the curve
reaches point 3. This is because the section steel can hold the
external load and prevent the further expanding of tensile zone.
It can also be found that the highest longitudinal stress (S33 in
the graphs) in the cross section of the concrete core is near the cen-
ter for the SRCFST column. For the CFST column, the highest longi-
tudinal stress in the cross section of the concrete core is near the
compressive flange of the column, as indicated in Fig. 8. This con-
tributes that the flanges of the section steel can provide some con-
fining effect for inner concrete. As a result, the SRCFST column has
obviously higher stiffness and yield strength than those of the CFST
column, as shown in Fig. 9.
3.3. Parametric study
The possible parameters affecting behaviors of the SRCFST
columns are axial load level (n), ratio of section steel (a), yield
strength of the section steel (fy2), thickness of steel tube (t) and
concrete strength (fck). In the experiment, only five specimens are
tested. In order to give a further understanding of mechanical per-
formance of such a composite column, the influences of these
parameters are analyzed.
3.3.1. Influences of section steel
In this paper, the ratio of section steel (a) is defined as follows:
Ass
a¼ ð5Þ
Ac þ Ass
where Ass and Ac are the cross section area of section steel and con-
crete core.
The computed P–d envelope curves for the SRCFST columns
with different ratios of section steel is presented in Fig. 9. The value
of a ranges from 0.03 to 0.13 according to the available test data.
Other parameters are the same as those listed in Table 2. In the
numerical models, the value has an enlarged scope from 0.0 to
0.2. When the ratio of section steel is zero, the specimen is the
common CFST column. It can be seen that increasing the value of
a leads to an increase in peak lateral load and specimen stiffness.
The corresponding lateral displacement at peak load also increases
with the increasing of a, as shown in Fig. 10.
Two groups of specimens with different ratios of section steel
(a = 0.01 and a = 0.1) are modeled and in each group, yield strength
of the section steel ranges from 235 MPa to 450 MPa. Other param-
eters are same as those listed in Table 2. Fig. 11 presented the influ-
ence of yield strength of section steel on peak load. It can be found
that higher yield strength of the section steel results in increasing
of peak load. We can see from Fig. 11 that when fy2 is 235 MPa, the
difference of peak loads is 40 kN. The difference is 119 kN when fy2
is MPa. This contributes that the yield strength of section steel can
contribute more to specimen peak load with higher value of a.
 X. Chang et al. / Construction and Building Materials 28 (2012) 88–95 93

Fig. 8. Longitudinal stress distributions: (a) cross section of concrete of CFST column for point 1; (b) cross section of concrete of CFST column for point 3; (c) cross section of
concrete of SRCFST column for point 1; (d) cross section of concrete of SRCFST column for point 3.

600
α = 0.2 16
Lateral displacement (mm)

α = 0.1
12
400 α = 0.05
P (kN)

α = 0.01 8

200 α =0 4

0
0 0.1 0.2 0.3
0
0 20 40 α
δ (mm)
Fig. 10. Influence of ratio of section steel on corresponding lateral displacement at
Fig. 9. Influence of ratio of section steel on P–d envelope curves for columns. peak lateral load.

3.3.2. Influence of axial load level
In order to study the influence of axial load level, we chose five
different values of axial load level, n = 0, 0.3, 0.5, 0.6 and 0.8 to set-
up models, while keeping other parameters listed in Table 2 con-
stant. The computed P–d envelope curves for the SRCFST column
with different axial load level is shown in Fig. 12. It can be found
that these specimens behave similarly in the linear stage. However,
there are obvious stiffness differences among these curves in non-
linear stage. Higher axial load level leads to lower stiffness. The lat-
eral load decreases after yielding of the specimen when the axial
load level exceeds the value of 0.3. The corresponding lateral dis-
placements at peak lateral load for n = 0, 0.3, 0.5, 0.6 and 0.8 are
20.1, 18.4, 15.7 and 10.3 mm, respectively. This contributes that
higher axial load level also decreases the specimen deformability.
3.3.3. Influence of concrete strength
The influence of concrete strength on peak lateral load is indi-
cated in Fig. 13. The concrete strength ranges from 30 MPa to
70 MPa. It can be seen from Fig. 13 that peak load increases with
increasing of the concrete strength.
94 X. Chang et al. / Construction and Building Materials 28 (2012) 88–95

750 600

Peak lateral load (kN)

400
Peak lateral load (kN)

500
200

α = 0.1

α = 0.01 0
0 2 4 6 8
250 Thickness of steel tube (mm)
200 300 400 500
Fig. 14. Influence of thickness of steel tube on peak lateral load.
Yield strength (MPa)

Fig. 11. Influence of yield strength of section steel on peak lateral load.
4. Conclusions

The mechanical performance of SRCFST columns under cyclic

loading is investigated through the ABAQUS/Standard solver. The
750 feasibility and accuracy of the numerical method was verified by
n=0 comparing the calculated results with the experimental observa-
600 tions. The following conclusions are made based on this research
n=0.3 work.
450 n=0.5
P (kN)

(1) The presence of the section steel can carry the lateral load
n=0.6
300 and reduce the tensile zone of the concrete section. As a
n=0.8 result, the SRCFST columns have higher stiffness and peak
lateral load than the common CFST columns even with the
150
same geometrical and material parameters. The section steel
can also enhance the deformation ability of a SRCFST
0
0 20 40 60 80 column.
δ (mm) (2) The flanges of the section steel can also offer some confining
effect on the concrete inner. The highest longitudinal stress
Fig. 12. Influence of axial load level on P–d envelope curves for the SRCFST column. in the concrete section for SRCFST column is in the flanges
of the section steel.
(3) The parametric study is also conducted to analyze the effects
of ratio of section steel, yield strength of section steel, axial
600 load level, thickness of steel tube and concrete strength on
peak lateral load of SRCFST columns.

Acknowledgments
Peak lateral load (kN)

400
The research reported in this paper is supported by the Key Sub-
ject Foundation of Henan Province (No. 504906) and the Doctor
Foundation of Henan Polytechnic University (B2009-2). The finan-
cial support is highly appreciated. We also appreciate the sugges-
200
tions and comments provided by the reviewers for the manuscript
revision.

0
0 20 40
Concrete strength (MPa)
Fig. 13. Influence of concrete strength on peak lateral load.
3.3.4. Influence of thickness of steel tube
In this section, the thickness of the steel tube is the variable and
other parameters are kept the same as those listed in Table 2. The
computed peak load with increasing of tube thickness is presented
in Fig. 14. It indicates that higher thickness of steel tube leads to
higher peak load.
60 80 References
[1] Ehab Ellobody, Ben Young. Design and behaviour of concrete-filled cold-
formed stainless steel tube columns. Eng Struct 2006;28:716–28.
[2] Mohamed Elchalakani, Xiao-Ling Zhao. Concrete-filled cold-formed circular
steel tubes subjected to variable amplitude cyclic pure bending. Eng Struct
2008;30:287–99.
[3] Uy B. Strength of short concrete filled high strength steel box columns. J Constr
Steel Res 2001;57(1):113–34.
[4] Zhao DZ, Wang QX, Guan P. Research on load-bearing capacity of steel tubular
columns filled with steel – reinforced high-strength concrete subjected to
compression and bending. Ind Constr 2005;35(9):84–93.
[5] Zhu MC, Wang QX, Feng XF. Behavior of axially-loaded square steel tube short
columns filled with steel-reinforced self-consolidating high-strength concrete.
China Civil Eng J 2007;39(6):35–41 [in Chinese].
 X. Chang et al. / Construction and Building Materials 28 (2012) 88–95
[6] Wang QX, Zhu MC, Feng XF. Experimental study on axially loaded square steel
tubes filled with steel-reinforced self-consolidating high-strength concrete. J
Build Struct 2005;26(4):14–21 [in Chinese].
[7] Zhu MC, Wang QX, Liu SR, et al. Experimental study of centrally loaded square
steel tubular slender columns filled with steel-reinforced self-consolidating
high-strength concrete. J Dalian Univ Technol 2006;46(6):875–9 [in Chinese].
[8] Guan P, Wang QX, Zhao DZ. Experimental study on ductility of circular steel
tubular columns filled with steel-reinforced concrete. Earthquake Eng Eng Vib
2003;23(1):84–9.
[9] Hibbitt, Karlson & Sorensen Inc. ABAQUS/standard User’s Manual, Version
6.4.1. Pawtucket (RI, USA): Hibbitt, Karlsson, & Sorensen, Inc.; 2003. .
[10] ACI. ACI 318-95. Building code requirements for structural concrete and
commentary. Detroit (USA): American Concrete Institute; 1999.
[11] Saenz LP. Discussion of ‘equation for the stress_strain curve of concrete’ by P.
Desayi and S. Krishnan. J Am Concr Inst 1964;61:1229–35.
[12] Hu HT, Schnobrich WC. Constitutive modeling of concrete by using non-
associated plasticity. J Mater Civ Eng 1989;1(4):199–216.
95
[13] Mander JB, Priestley MJN, Park R. Theoretical stress–strain model for confined
concrete. ASCE J Struct Eng 1988;114(8):1804–26.
[14] Han LH, Yao GH, Zhao XL. Tests and calculations of hollow structural steel
(HSS) stub columns filled with self-consolidating concrete (SCC). J Constr Steel
Res 2005;61(9):1241–69.
[15] ASCE. ASCE task committee on concrete and masonry structure. New York:
State of the Art Report on Finite Element Analysis of Reinforced Concrete,
ASCE; 1982.
[16] Guo ZH. Strength and constitutive relation of concrete. Chinese Building
Industry Press; 2003.
[17] Dai X, Lam D. Numerical modeling of the axial compressive behaviour of short
concrete-filled elloptical steel columns. J Constr Steel Res 2010;66:931–42.
[18] Han LH, Yao GH, Tao Z. Performance of concrete-filled thin-walled steel tubes
under pure torsion. Thin-walled struct 2007;45:24–36.
[19] Hassanein MF. Numerical modeling of concrete-filled lean duplex slender
stainless steel tubular stub columns. J Constr Steel Res 2010;66:1057–68.