Engineering Structures 33 (2011) 31623171

Contents lists available at SciVerse ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

Advanced design for trusses of steel and concrete-filled tubular sections

M. Fong a , S.L. Chan a, , B. Uy b
a

Department of Civil and Structural Engineering, The Hong Kong Polytechnic University, Hong Kong, China

School of Engineering, University of Western Sydney, Sydney, Australia

article

info

Article history:
Received 15 January 2011
Received in revised form
11 March 2011
Accepted 2 August 2011
Available online 23 September 2011
Keywords:
Concrete-filled steel tubes
Design codes
Effective lengths
Second-order analysis
Trusses

abstract
This paper presents an experimental and analytical investigation of buckling behavior of bare steel and
concrete-filled steel (CFS) tubes used as columns and as members of trusses. The member resistances of
the columns and trusses consisting of steel and CFS tubular members are compared to demonstrate the
beneficial effects of the in-filled concrete, with their resistances predicted using the conventional effective
length and second-order analysis methods of design in various international standards such as Eurocode
3 (EC3), Eurocode 4 (EC4), CoPHK, AISC-LRFD and AS5100. Test results are further used to validate the
proposed second-order analysis, which skips the assumption of effective length, for accurate and reliable
design of composite members. The present holistic approach of considering composite members as
constituting elements in a truss represents a piece of original work on testing and design of structures
as a system, rather than designing members in isolation in the traditional member-based design.
2011 Elsevier Ltd. All rights reserved.

1. Introduction
Extensive experiments have been conducted and reported
on composite columns to investigate their structural resistances
under various loading conditions. Furlong [1] tested concrete-filled
steel tubular columns and indicated the local buckling of the steel
tube was delayed by concrete infill. Circular and square concretefilled steel tubes with different slenderness ratios were tested by
Knowles and Park [2] to investigate the confinement effect on
concrete under various slenderness ratios. The results indicated
the benefits on ultimate capacity from the confinement effect of
short circular columns. Tomii et al. [3] carried out tests on circular,
octagonal and square concrete-filled steel tubular columns, and
noted that the concrete core provided confinement effects only
on circular and octagonal sections. Shakir-Khalil and Mouli [4]
performed tests on concrete-filled steel tubular columns and
concluded that higher concrete strengths or larger steel sections
provided a greater structural efficiency for composite columns. The
shape effect of steel tubes on strength and behavior of concretefilled steel tubes were studied by Schneider [5]. Fourteen circular,
square and rectangular tubes were tested and the results indicated
that the circular tubes provided greater ductility than the other
two section shapes and the confinement effect on the concrete
core was observed only on circular tubes when the yield strength
was approached. Kilpatrick and Rangan [6] reported test results

Corresponding author. Tel.: +852 9025 6814; fax: +852 23346389.

E-mail address: ceslchan@polyu.edu.hk (S.L. Chan).

0141-0296/$ see front matter 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.engstruct.2011.08.002

on concrete-filled steel tubular columns, and indicated the effects

on the behavior and strength of composite columns due to the
slenderness and load eccentricity. Li et al. [7] carried out tests
on six high-strength concrete-filled square steel tubular columns
under bi-axial eccentric load and indicated that both the steel area
ratio and slenderness ratio influenced the capacity of the columns
significantly. Chan and Fong [8] tested steel and composite trusses
with high-strength in-filled concrete to compare the beneficial
effect of the in-filled concrete.
From the review on the subject, most previous experiments
focused on the behavior of isolated columns under pinned or fixed
end conditions, and tests on steel and concrete-filled steel (CFS)
tubes acting as members of a frame or truss structures are limited.
In this paper, tests on trusses composed of the bare steel and CFS
tubes as chords and webs in the truss are reported and compared
with the second-order analysis and design method. In order to
compare the behavior of the members in an isolated column and
in the truss system, the columns with same properties of the
members in the truss under pinned and fixed ends are tested. The
reported work in this paper is believed to be novel in providing
the response of the members in the configuration of a structural
system as a truss, which should be a good reference for the future
research and design of composite trusses and frames.
Various international design codes provide different methods
of design for steel and composite members under different loading
conditions such as EC3 [9], EC4 [10], AISC LRFD [11], AS5100 [12]
and CoPHK [13]. The accuracy of these design methods have been
verified by many researchers such as Wang [14], Al-Rodan [15],
Zeghiche and Chaoui [16]. Most comparisons of experimental and
predicted results from these codes were focused on isolated

M. Fong et al. / Engineering Structures 33 (2011) 31623171

3163

Table 1
Material properties of steel tube.
Item

Steel
section

B
(mm)

D
(mm)

t
(mm)

Yield stress (fy )

(N/mm2 )

Ultimate tensile stress (fu )

(N/mm2 )

Youngs modulus (Es )

(kN/mm2 )

fu /fy

Columns
Truss
members

60 60 3
60 60 3

60.40
60.20

60.30
60.20

3.10
3.10

407.98
404.11

480.22
473.55

206.36
205.72

1.18
1.17

Applied force
Applied force
Load Cell

2000

2000

Load Cell

Pinned end boundary condition

Fixed end boundary condition

Fig. 1. The setup of columns.

members with pinned or fixed end boundary conditions. However,

the behaviors of members with more realistic boundary conditions
such as constitutive members in a truss have not been considered
in previous research. The traditional linear analysis method of
design assumes an effective length factor for different boundary
conditions of a chord or a web member in a truss and assesses
various factors to compensate for buckling and second-order
effects in the determination of buckling strength of individual
members. Unfortunately, the accuracy on determination of the
effective length highly affects the reliability of these methods. For
simple end conditions like pinned and fixed end, previous methods
can give an accurate design solution as shown in the following
example. However, for more complicated and practical structures
like trusses, the accurate determination on the effective length is
difficult and sometime impossible. This paper reports the accuracy
of design methods in several international codes for the prediction
of capacities of steel and CFS tubes as composing members in a
truss under pinned and fixed end conditions.
Design codes such as AISC LRFD [11], EC 4 [10] and CoPHK [13]
propose the use of second-order analysis and design methods because of the accuracy, reliability and convenience. The superiorities
of second-order analysis and design method in comparison with
the first-order analysis and design method have been reported
in literature by different researchers such as Chen [17] and Chan
et al. [18]. Using this design method, nonlinear effects such as the
P- and P- moments, member and frame imperfections are included in the analysis process. The new design method not only
enhances the accuracy of the design output, but it also reduces design time and effort. The Pointwise Equilibrium Polynomial (PEP)
element proposed by Chan and Zhou [19] is used here for its simplicity and computational stability and efficiency allowing modeling of one member by one element. The accuracy of using the

PEP element on various forms of steel structures have been verified on various structural forms such as dome structures [20], angle
trusses in both elastic [21] and plastic analysis [22,23], scaffolding
systems [24] and pre-stressed stayed column [25]. In this paper,
the design method of second-order analysis with the PEP element
would be applied to predict the resistance of steel and CFS tubes
used as isolated columns and as members of trusses.
2. Experimental work
2.1. General
Tests for columns and trusses were carried out. For the column
tests, four specimens of two bare steel tubes and two CFS tubes
were tested. Square hollow section steel tubes of 60 60
3 mm cross-section were used and the average width, depth and
thickness are listed in Table 1. Two boundary conditions as pinned
and fixed ends were set up in the test as indicated in Fig. 1. The
specimen length for the fixed end condition is 2 m and for the
pinned end condition is 1.74 m.
Two trusses with members composed of bare steel tubes
(named as steel truss) and CFS tubes (named as composite truss)
were tested with the mean dimensions of truss members are
listed in Table 1. The same cross-sections used for the isolated
columns tests were adopted in the construction of the trusses and
the dimensions of the trusses are provided in Fig. 2. Steel trusses
were composed of steel tubes for all members and the composite
trusses were composed of CFS tubes in the compressive members
and bare steel tubes in the tensile members. Each of these threedimensional trusses consisted of 19 members which included the
14 main members of 2 m length and 5 tie members of 0.8 m length

3164

M. Fong et al. / Engineering Structures 33 (2011) 31623171

Applied force

Applied force

Hydraulic jack

Hydraulic jack

Load Cell
2000

Load Cell

00
20

4000
Plan View

Side View

Fig. 2. The dimensions of the truss.

in-plane direction
out-of-plane direction
Strain gage

Fig. 3. The locations of strain gages and displacement transducers.

Table 2
Material properties of concrete.
Item

Age at testing (days)

Stress (N/mm2 )

E (N/mm2 )

conditions as the specimens until the day of test. The characteristic

compressive cylinder strength and the modulus of elasticity of the
concrete used in the CFS tube are given in Table 2.

Columns
Truss members

34
56

40.96
41.16

22.17
22.73

2.3. Column tests

connecting the two 2-D trusses. All members were connected as

moment connections by 8 mm fillet welds.
2.2. Material properties
The steel plate slenderness ratio (D/t ) was selected according
to the guidelines provided in EC3 [9] and EC4 [10] to avoid
local buckling. Coupon tests were carried out to determine the
stressstrain curve of the steel material. Properties including
the characteristic yield and ultimate tensile stress and Youngs
modulus of elasticity for the columns and trusses members are
listed in Table 1.
Normal strength concrete was used in filling the steel tubes.
The concrete mix included 200 kg/m3 water, 408 kg/m3 Ordinary
Portland Cement, 1074 kg/m3 coarse aggregate and 716 kg/m3
fine aggregate. A series of standard concrete cylinders, 100 mm in
diameter and 200 mm in height was cast and cured under the same

2.3.1. Test setup

Steel and CFS tubular columns with pinned and fixed boundary
conditions were tested and the test setup is shown in Fig. 1. The
column was 2 m in length, measured from the center of rotation at
each end. The test specimen was cast on the surface of a flat steel
plate at both ends to ensure smooth surface for producing uniform
loading. During the test, load was applied at the top of the column
via the loading jack of maximum capacity equal to 400 kN. The load
interval was taken as 1 kN throughout the whole test and the load
cell was located at the top of the column to record the applied load.
Rotation was only allowed in the in-plane direction for the pinned
end condition, whilst no rotation was allowed in both the in-plane
and out-of-plane directions for the fixed end condition.
2.3.2. Strain gages and displacement transducers
Strain gages were mounted on the faces of the specimen as
indicated in Fig. 3 to monitor the changes of strain during the

M. Fong et al. / Engineering Structures 33 (2011) 31623171

3165

Table 3
Column test results.
Specimen

Steel tubular column

CFS tubular column
Steel tubular column
CFS tubular column

End condition

Pin
Pin
Fix
Fix

Applied load (kN)

Ratio

152.45
186.88
249.40
331.60

CFS column/steel column

Fix/Pin

/
1.23
/
1.33

/
/
1.64
1.77

350

CFS tubular column (fixed end)

300

Load (kN)

250

Steel tubular column (fixed end)

200

CFS tubular column (pined end)

150
100

Steel tubular column (pined end)

50
0

10
20
30
40
Mid-length in-plane deflection of columns (mm)

50

Fig. 5. Load against in-plane lateral deflections of columns.

350

CFS tubular column (fixed end)

300

Load (kN)

250
200

CFS tubular column (pined end)

150
100

Steel tubular
column
(pined end)

CFS tubular
column
(pined end)

Steel tubular
column
(fixed end)

CFS tubular
column
(fixed end)

2.3.3. Test results

All the specimens failed in a flexural buckling mode as shown in
Fig. 4 and no local buckling was observed before reaching the maximum applied load. The maximum loads of all specimens are listed
in Table 3. The load ratios of CFS to steel tubular columns were 1.23
and 1.33 for pinned end and fixed end conditions respectively, and
the load ratios of fixed to pinned end conditions were 1.64 and 1.77
respectively for steel and CFS tubular columns.
The axial load versus mid-length in-plane and out-of-plane
deflection curves for both steel and CFS tubular columns are
plotted in Figs. 5 and 6. For the in-plane direction as depicted in
Fig. 5, the deflection increased linearly with applied load at the
initial stage, and the relationship became nonlinear when the load
approached failure load. After attaining the maximum load, the
deflection of the member increased significantly with decreasing
load. The load against out-of-plane deflection curve presented
in Fig. 6 shows that the deflection was small at the maximum
applied load compared with the in-plane deflection in the pinned
end columns. However, for fixed end columns, the mid-length
deflection along the out-of-plane direction was close to the inplane deflection, which indicated that the column deflected in both

Steel tubular column (pined end)

50
5

Fig. 4. The failure model of each column.

test. Eleven displacement transducers were used to measure the

downward and lateral movements at the top, bottom, middle and
quarter points for both the in-plane and out-of-plane directions
of the columns. The locations of these transducers are shown in
Fig. 3.

Steel tubular column (fixed end)

5
15
25
35
45
Mid-length out-of-plane deflection of columns (mm)

Fig. 6. Load against out-of-plane lateral deflection of columns.

directions and similar initial imperfections in both directions were

present.
The load against strain curves at the mid-length of the columns
are also plotted in Fig. 7. These curves were noted to possess a
typical and consistent pattern similar to loaddeflection curves.
Signs of local buckling on the steel plate were not observed from
these curves.
The deformed shape of each specimen is shown in Fig. 8 and
typical deformed shapes for pinned end and fixed end column were
observed.
For pinned end columns, owing to the friction in the pinned
end, the degree of end restraint may be varied and the maximum
applied load could be higher than the theoretically perfect pinned
end case.
2.4. Truss test
2.4.1. Test setup
Two trusses were simply supported at four points and loaded
in pairs by using the hydraulic jacks with a maximum capacity of
400 kN as shown in Fig. 2. The load increment was taken to be
1 kN during the test, and the two load cells were located under
point loads. Sufficient lateral restraints were provided to ensure
out-of-plane buckling at connecting nodes between members was
prevented.

3166

M. Fong et al. / Engineering Structures 33 (2011) 31623171

Steel tubular column (pined end)

200

SG 4

SG 2

150

SG 4

CFS tubular column (pined end)

SG 2

Load (kN)

150
SG 1

SG 3

1
4

50

SG 1

Load (kN)

100

SG 3100
1
4

50

3
0
8000 6000 4000 2000 0
Strain ()

2000 4000

3
4000

0
2000
0
2000
Strain ()
CFS tubular column (fixed end)
350

6000

Steel tubular column (fixed end)

250

SG 3

SG 1

Load (kN)

Load (kN)

100

1
4

1
4

50

20000 15000 10000 5000

Strain ()

200
SG 3
150

SG 1

100
2

3
0

SG 2

250

SG 2 150

SG 4

300

SG 4

200

4000

50

3
0

5000

40000 30000 20000 10000

10000 20000

Strain ()

2000

2000

1500

1500

1500

1500

1000

500

1000

20 40 60
Deflection (mm)

Steel tubular column

(pined end)

80

1000

20 40 60
Deflection (mm)

80

CFS tubular column

(pined end)

1000

500

500

500

Length (mm)

2000

Length (mm)

2000

Length (mm)

Length (mm)

Fig. 7. Load against strain values of the columns.

0
0

20 40 60
Deflection (mm)

Steel tubular column

(fixed end)

80

20 40 60 80 100
Deflection (mm)

CFS tubular column

(fixed end)

Fig. 8. The deformed shape of the columns.

2.4.2. Strain gages and displacement transducers

The movements on the truss, and deflection of the targeted
failure members of in-plane and out-of-plane directions were
recorded by the transducers, and the strain gages were mounted
on the mid-length and quarter points of the target failure members.
The detailed locations of strain gages and displacement transducers are shown in Fig. 9.
2.4.3. Test result
The maximum applied loads on both trusses are presented in
Table 4. Flexural buckling took place in the in-plane direction
on the target failure member and the final deformed shapes
are shown in Figs. 10 and 11. The maximum load resisted by
the member in the composite truss was 29% higher than that
of the steel truss. The applied load against mid-length in-plane
deflection of the failure member is plotted in Fig. 12. Similar load

deflection curves were observed for both the steel and CFS tubular
members that, when the load was small, a linear relationship
between the load and deflection was noted and the relationship
became nonlinear approaching the maximum loads which were
respectively 250.37 kN and 323.09 kN for steel and CFS tubular
members. The load was applied continuously to the truss after
achieving the maximum load in order to study the post-buckling
behavior. The applied load against the out-of-plane deflection at
mid-length of the failure members are plotted in Fig. 13. The curves
showed that the out-of-plane deflection was small compared with
the in-plane deflection at the maximum load in both steel and
composite trusses, and the out-of plane deflection was mainly due
to the initial imperfections.
The applied load against the strain plots at the mid-length
of the failure members are shown in Fig. 14 and the non-linear
relationship and post failure behaviors were observed. Large

M. Fong et al. / Engineering Structures 33 (2011) 31623171

3167

in-plane direction
out-plane direction
strain gage

Fig. 9. The locations of strain gages and displacement transducers.

350
300

Composite truss

Load (kN)

250
200
150
100

Steel truss

50
0

10

50

10
30
50
Mid-length deflection of members (mm)

Fig. 13. Load against out-of-plane lateral deflection of the tr.

Table 4
Truss test results.

Fig. 10. Buckling mode of steel truss.

Specimen

Applied load on truss (kN)

Failure member force (kN)

Steel truss
Composite truss

250.37
323.09

216.57
279.47

slight difference was mainly due to the out-of-straightness imperfections. The result also implied that the out-of-plane deflection was insignificant before reaching the failure load, after which
the out-of-plane deflection increased significantly with decreasing
load.
The member forces against deflections of columns under pinned
and fixed end conditions and for the failure members of the trusses
are plotted in the same figure (Fig. 15) for comparison. The test
results show that the resistances of both the steel and CFS tubular
members were somewhere between the curves for members with
pinned and fixed end conditions which implies that the boundary
condition of the trusses members is semi-rigid.
3. Comparisons between test and design capacities from codes
Fig. 11. Buckling mode of composite truss.

350

Composite truss

Load (kN)

300
250
200
150

Steel truss

100
50
10

0
50

10
30
50
Mid-length deflection of members (mm)

Fig. 12. Load against in-plane lateral deflection of the truss.

compressive strains were developed at the bottom fibers which

gave a consistent result with displacement transducers. The strain
gages SG2 and SG4 recorded on both sides of the member were
nearly identical for both steel and CFS tubular members and the

3.1. Design methods in EC3 and EC4, CoPHK, AISC LRFD and AS 5100
The resistances of steel and CFS tubular columns were predicted
using EC3 [9], EC4 [10], CoPHK [13], AISC LRFD [11] and AS
5100 [12] and compared with test results. The design methods on
the compressive capacity of composite columns based on different
design codes are summarized below.
3.1.1. EC4 and CoPHK
The section capacity of CFS tubular members is determined by
summation of the resistances of the concrete and steel tubes and
the member capacity is reduced by multiplying a buckling reduction factor , which is obtained from the effective slenderness ratio
and section types, as
Pcp = (As fyd + Ac fcd )

(1)

in which As , Ac , fyd and fcd are the cross-sectional area and the design cylinder strength of the steel and concrete respectively.

3168

M. Fong et al. / Engineering Structures 33 (2011) 31623171

300
SG 1

SG 3

SG 1

100

1
4

Load (kN)

Load (kN)

200
SG 4

SG 2

3
10000

5000
Strain ()

100

1
2
3
0

5000

SG 3

200
SG 2 SG 4

15000

300

20000 15000 10000 5000

5000

10000

Strain ()

Steel truss

Composite truss

Fig. 14. Load against strain value of the trusses.

350

300

Load (kN)

Load (kN)

200

Truss

150
100
50
0

Pined end column

20
40
60
80
Mid-length deflection (mm)

50

Fixed end column

300

Fixed end column

250

250
200
150

Truss

100
50

100

Pined end column

20
40
60
80
Mid-length deflection (mm)

Steel column and truss

100

Composite column and truss

Fig. 15. Member force against in-plane lateral deflection of co.

The reduction factor is determined by

1

=
+

The stability of CFS tubular members is considered by a buckling

reduction factor c in calculating the member resistance as follows,
(2)

(9)

and

and

Nuc = c (0.9As fy + 0.6Ac fc )

1
2

2
1 + ( 0.2) +

(3)

in which is the imperfection factor and is the relative slenderness.

3.1.2. AISC LRFD
The compressive strength of the CFS tubular columns is calculated as,

For Pe 0.44Po

Pn = Po 0.658

For Pe < 0.44Po

Pn = 0.877Pe

Po
Pe

(4)
(5)

and
Po = As Fy + 0.85Ac fc

(6)

in which Fy and fc are the yield stress of steel and compressive

strength of concrete respectively and Pe is the elastic buckling load.
The design compressive resistance of CFS tubular columns can
be obtained as,
Pd = 0.75Pn .

(7)

3.1.3. AS 5100
The ultimate section capacity can be determined by summing
the axial capacity of the steel tube and concrete as follows.
Nus = 0.9As fy + 0.6Ac fc

(8)

in which fy and fc are the nominal yield strength of steel and

characteristic cylinder strength of concrete. The values of 0.9 and
0.6 are the capacity factors for steel and concrete respectively.

c = 1

[
1

90

]2

(10)

The terms and depend on the relative slenderness ratio (r )

and section constant (b ).
3.2. Predicted results and comparisons
The material factors used to reduce the strengths and modulus
of elasticities of steel and concrete were taken as unity in
comparison with the test and predicted results according to the
method in design codes and method of second-order analysis, the
properties from material tests were used in predicting the results
and compared with test results. The effective length factors equal
to 1.0 (Le = 2000) and 0.5 (Le = 1000) were assumed for the
pinned and fixed end conditions respectively. Predicted capacity
according to the different design methods are summarized and
compared in Table 5.
Generally speaking, for steel columns under pinned and fixed
end conditions, conservative predictions were obtained from
EC3 [9], AISC LRFD [11] and CoPHK [13] with an under-estimation
around 5%23%, and a close estimation was made by AS 5100 [12]
(under-design by 1%3%). For CFS tubular columns under pinned
and fixed end conditions, highly conservative predictions were
obtained by AISC LRFD [11] (over-design by 29%35%) due to
the conservative resistance factor, and a reasonable prediction
was made by the other codes with the difference is less than
7%. Although the design codes such as the AS5100 [12], among
others, over-estimate the column capacity, material factors have

M. Fong et al. / Engineering Structures 33 (2011) 31623171

3169

Table 5
Predicted result from different design codes.
Specimen

End
condition

Member
force (kN)

Predicted capacity from design code

Steel tubular column

Steel tubular column
Member in steel truss

Pin
Fix
/

152.45
249.40
216.57

CFS tubular column

CFS tubular column
Member in composite truss

Pin
Fix
/

186.88
331.60
279.47

124.23
141.21
226.59
223.63
Pin end assumed
Fix end assumed
184.33
145.35
353.21
246.39
Pin end assumed
Fix end assumed

EC3 EC4

AISC

been adopted in the codes but not considered in this paper for
consistency and therefore the final design in these codes should
still warrant a design with adequate factor of safety.
In the truss tests, end movements of the members were allowed
but restrained partly by other members. Hence, the boundary
condition of the truss members was between the pinned and fixed
end condition cases. Here, both the predicted resistances of the
members under pinned and fixed end conditions are compared
with experimental results are reported in Table 5. As expected, a
very conservative prediction was made with the difference from
40% to 74% for steel tubular members and 44%92% for CFS tubular
members if the pinned end condition was assumed which implies
an uneconomical design. If the fixed end boundary condition was
assumed, the capacity of the members would be over-estimated
for steel tubular members in the range of 3%16%, and CFS tubular
members with 21% exception for AISC LRFD [11] which gives a load
below the tested load.
4. Comparisons between test and second-order analysis and
design method
4.1. Design by second-order analysis
Using the second-order analysis and design method, the assumption of effective length is avoided and the nonlinear effects
are included automatically in analysis with true behavior of the
structure reflected. The detailed formulations of the element, tangent stiffness and secant stiffness matrix for steel and composite
members have been reported by Chan and Zhou [19,26], and Fong
et al. [27], and will not be repeated here.
4.2. Section capacity check equations for second-order analysis and
design method
In the second-order analysis and design method, the section
capacity check equations are used and included the second-order
effects in analysis process.
For steel members, the following section capacity equation is
adopted.
P
Pp

My + P (y + y )
Mpy

Mz + P (z + z )
Mpz

=1

(11)

in which P is the applied force, Pp is the compressive capacity of

steel, My and Mz are the external moments about the y and z axes,
P (y + y ) and P (z + z ) are the P- and P- moments about the y
and z axes, Mpy , and Mpz , are the moment capacities of steel crosssection about the y and z axes. Note that Eq. (11) is symbolic in
that the P- moment need not be evaluated explicitly but included
in the analysis by updating of member curvature.
For composite members, two section capacity equations are
used for two different load conditions. In Eq. (12), the effects of
axial force and moments are taken into account in the section

Ratio

CoPHK

AS 5100

Test/EC3 EC4

Test/AISC

Test/CoPHK

Test/AS 5100

132.54
236.64

154.61
256.75

184.33
353.21

193.96
354.81

1.23
1.10
1.74
0.96
1.01
0.94
1.52
0.79

1.08
1.12
1.53
0.97
1.29
1.35
1.92
1.13

1.15
1.05
1.63
0.92
1.01
0.94
1.52
0.79

0.99
0.97
1.40
0.84
0.96
0.93
1.44
0.79

capacity equation, and it is used when the applied force is larger

than the capacity of concrete section (i.e. P > Ppm ). And in Eq. (13),
only the moment is considered since the axial force does not reduce
the failure load and it is used when the applied force is less than the
capacity of concrete section (i.e. P Ppm ). These two sets of section
capacity equations are given as follows.
For P > Ppm

Pcp Ppm

Mz + P (z + z )
Mcpz

For P Ppm

P Ppm

My + P (y + y )

=1

Mcpy
(12)

My + P (y + y )
Mcpy

Mz + P (z + z )
Mcpz

=1

(13)

in which Ppm , Pcp are the compressive capacities of concrete and

composite cross-section, Mcpy , and Mcpz are the moment capacities
of composite cross-section about the y and z axes.
As can be seen in Eqs. (11)(13), the P- and P- moments
are included in the section-capacity check equations such that the
approximation of effective length factor and computation of elastic
buckling load factor is not needed.
4.3. Analysis results
The average yield stress and Youngs Modulus of steel and
concrete from material tests were used in the computer analysis.
Two initial imperfections equal to L/300 and L/400 were used for
both steel and composite members, where L is the member length.
The imperfection of L/300 is recommended in CoPHK [13] for both
steel and CFS tubular columns for second-order analysis. Another
set of imperfections of L/400 was also used in the prediction the
column resistance to demonstrate the degree of conservative of the
recommended magnitude of imperfections. The load increment
factor 0.01 kN was used in analysis until the section capacity factor
is equal to 1.0.
The predicted results for both columns and trusses are shown
in Table 6. For the columns under pinned and fixed end conditions,
the ratios of test to predicted results are 1.09 and 1.08 for steel
tubular columns, and 1.14 and 1.03 for CFS tubular columns with
the initial imperfection equal to L/300, and a closer prediction was
made with the imperfection of L/400 that the ratios are 1.02 and
1.03 for steel tubular columns, and 1.08 and 0.98 for CFS tubular
columns.
The computer model of the truss is shown in Fig. 16. The center
to center length of the members was used in the model and rigid
connection between members was assumed. The analysis results
with two imperfection values are presented together with test
results in Table 6. A close and conservative prediction was made
with the discrepancy within 5% for both steel and CFS tubular
members when using the imperfection of L/400.

3170

M. Fong et al. / Engineering Structures 33 (2011) 31623171

Table 6
Predicted truss results from second-order analysis and design method.
Specimen

End
condition

Member force
(kN)

Predicted capacity from second-order

analysis and design method
Load (kN)

Steel tubular column

Steel tubular column
Member in steel truss
CFS tubular column
CFS tubular column
Member in composite truss

Pin
Fix
/
Pin
Fix
/

152.45
249.40
216.57
186.88
331.60
279.47

Ratio

L/300

L/400

L/300

L/400

140.43
229.96
190.20
164.11
322.50
254.36

149.28
241.92
205.70
172.30
340.10
272.94

1.09
1.08
1.14
1.14
1.03
1.10

1.02
1.03
1.05
1.08
0.98
1.02

Fig. 16. Analytic truss model.

5. Conclusions
Experimental investigation on resistances of steel and CFS tubes
used as isolated columns and as the members of trusses was
presented in conjunction with the numerical and codified results
in this paper. The beneficial effects of in-filled concrete on steel
tubes for both columns under pinned and fixed end conditions
and members of trusses were reported. The resistances of the
steel and CFS tubular members were determined and compared
with the experimental results according to the design methods in
EC3 [9], EC4 [10], CoPHK [13], AISC LRFD [11] and AS5100 [12].
The comparisons show conservative predictions of steel columns
for the design codes with the exception of AS5100 [12] giving
an aggressive ultimate load prediction and a close prediction for
composite columns was made except with AISC LRFD [11] which
underestimates considerably the resistance of the CFS tubular
columns. The comparisons with the test results of trusses and
predicted results according to the design codes by assuming the
effective length as member length for pinned and half of member
length for fixed end conditions indicated that the pinned end
assumption under-estimated the resistance of the members which
led to an uneconomical design, while the fixed end assumption
over-estimated the capacities of the members and the design
became non-conservative. The main disadvantage of using the

effective length method was demonstrated in this paper that the

idealized pinned and fixed end conditions did not exist in practice
and the results by either assumption were inaccurate. This paper
further determined the resistance of isolated columns and struts in
trusses by using the second-order analysis and design method with
two imperfection values. The results indicated that the secondorder analysis and design method not only provided an accurate
design solution, but also avoided uncertain approximation of
effective lengths which involved complications in sway or nonsway frame classification or determination from inspection of
buckling mode shape.
Acknowledgments
The authors acknowledge the financial support by the Research
Grant Council of the Hong Kong SAR Government on the projects
Collapse Analysis of Steel Tower Cranes and Tower Structures
(PolyU 5119/10E) and Stability and second-order analysis and
design of re-used and new scaffolding systems (PolyU 5116/11E).
References
[1] Furlong RW. Strength of steel-encased concrete-filled beamcolumns. J Struct
Div, ASCE 1967;93(5):11324.

M. Fong et al. / Engineering Structures 33 (2011) 31623171

[2] Knowles RB, Park R. Strength of concrete-filled steel tubular columns. J Struct
Div, ASCE 1969;95(12):256587.
[3] Tomii M, Yoshimura K, Morishita Y. Experimental studies on concrete filled
steel tubular stub columns under concentric loading. In: Proceedings of the
international colloquium on stability of structures under static and dynamic
loads. 1977. p. 71841.
[4] Shakir-Khalil H, Mouli M. Further tests on concrete-filled rectangular hollowsection columns. J Struct Eng 1990;68(20):40513.
[5] Schneider SP. Axially loaded concrete-filled steel tubes. J Struct Eng, ASCE
1998;124(10):112538.
[6] Kilpatrick AE, Rangan BV. Prediction of the behaviour of concrete-filled steel
tubular columns. Aust J Struct Eng Trans 1999;SE2(23):7383.
[7] Li GC, Yang ZJ, Lang Y. Experimental behavior of high strength concrete-filled
square steel tube under bi-axial eccentric loading. Adv Steel Constr 2010;6(4):
96375.
[8] Chan SL, Fong M. Experimental and analytical investigations of steel and
composite truss. Adv Steel Constr 2011;7(1):1726.
[9] BSI. Eurocode 3: design of steel structurespart 11: general rules and rules
for buildings. London: BS EN 1993-1-1, BSI; 2005.
[10] BSI. Eurocode 4: design of composite steel and concrete structurespart 1.1:
general rules and rules for buildings. London: BS EN 1994-1-1, BSI; 2004.
[11] AISC. Load and resistance factor design specification for structural steel
buildings. Chicago: American Institute of Steel Construction, AISC Inc.; 2005.
[12] Standards Australia. AS5100.6-2004 bridge design, part 6: steel and composite
construction. Sydney (Australia); 2004.
[13] Code of practice for structural use of steel 2005. Buildings Department. Hong
Kong SAR Government. 2005.
[14] Wang YC. Tests on slender composite columns. J Constr Steel Res 1999;49(1):
2541.

3171

[15] Al-Rodan A. Comparison between BS5400 and EC4 for concrete-filled steel
tubular columns. Adv Struct Eng 2004;7(2):15968.
[16] Zeghiche J, Chaoui K. An experimental behaviour of concrete-filled steel
tubular columns. J Constr Steel Res 2005;61(1):5366.
[17] Chen WF. Structural stability: from theory to practice. J Eng Struct 2000;22(2):
11622.
[18] Chan SL, Liu YP, Zhou ZH. Limitation of effective length method and
codified second-order analysis and design. Steel Compos Structs 2005;5(23):
18192.
[19] Chan SL, Zhou ZH. Pointwise equilibrating polynomial element for nonlinear
analysis of frames. J Struct Eng, ASCE 1994;120(6):170317.
[20] Fong M, Chan SL. Second-order analysis and experimental tests on shallow
dome. In: Proceedings of the steel concrete composite and hybrid structures.
Research Publishing Services. 2009. p. 70914.
[21] Chan SL, Cho SH. Second-order analysis and design of angle trusses-part I:
elastic analysis and design. Eng Struct 2008;30(3):61625.
[22] Cho SH, Chan SL. Second-order analysis and design of angle trussesart II:
plastic analysis and design. Eng Struct 2008;30(3):62631.
[23] Fong M, Cho SH, Chan SL. Design of angle trusses by codes and secondorder analysis with experimental verification. J Constr Steel Res 2009;65(12):
21407.
[24] Chan SL, Zhou ZH, Chen WF, Peng JL, Pan AD. Stability analysis of semirigid
steel scaffolding. Eng Struct 1995;17(8):56874.
[25] Chan SL, Shu GP, Lu ZT. Stability analysis and parametric study of pre-stressed
stayed columns. Eng Struct 2002;24(1):11524.
[26] Chan SL, Zhou ZH. Second-order elastic analysis of frames using single
imperfect element per member. J Struct Eng, ASCE 1995;121(6):93945.
[27] Fong M, Liu YP, Chan SL. Second-order analysis and design of imperfect
composite beamcolumns. Eng Struct 2010;32(6):168190.