Journal of Constructional Steel Research 53 (2000) 121147

www.elsevier.com/locate/jcsr

Assessment of current methods for the design

of composite columns in buildings
*
H.S. Saw, J.Y. Richard Liew
Department of Civil Engineering, National University of Singapore, 10 Kent Ridge Crescent,
Singapore 119260, Singapore

Abstract

This paper presents the design assessment of encased I-sections and concrete filled com-
posite columns based on the approaches given in Eurocode 4: Part 1.1, BS 5400: Part 5 and
AISC LRFD. This includes studies on the design parameters, comparison of the nominal
strength predicted by the three codes and comparison of the predicted strengths with the avail-
able test results. In some cases, results obtained from the above three codes may vary consider-
ably. This is because of the different design considerations adopted in these codes. However,
the design methods are found to be mostly conservative when compared with the test results.
Eurocode 4: Part 1.1 has major factors in its favour in terms of its comprehensiveness and
wide scope of application. 2000 Elsevier Science Ltd. All rights reserved.

Keywords: Buckling; Composite columns; Concrete filled; Encased section; Limit states; Steel-concrete
composite

1. Introduction

A composite column is defined as a compression member which may either be a

concrete encased steel section or a concrete filled hollow section. Some typical cross-
sections of double-symmetrical composite columns are shown in Fig. 1. Composite
columns have gained acceptance for high-rise buildings as an alternative to pure
reinforced concrete during the past decades. The advantages of using composite col-
umn are: (1) smaller cross-section and higher strength-to-weight ratio than a conven-
tional reinforced concrete member, (2) significant savings in material and construc-

* Corresponding author. Tel.: +65-874-2154; fax: +65-779-1635.

E-mail address: cveljy@nus.edu.sg (J.Y.R. Liew)

0143-974X/00/$ - see front matter 2000 Elsevier Science Ltd. All rights reserved.
PII: S 0 1 4 3 - 9 7 4 X ( 9 9 ) 0 0 0 6 2 - 0
122 H.S. Saw, J.Y.R. Liew / Journal of Constructional Steel Research 53 (2000) 121147

Nomenclature
Aa Area of steel section
Ac Area of concrete
Ar Area of reinforcement
As Area of reinforcement (EC4); Area of steel section (BS 5400)
Aw Web area of steel section
b Overall breadth of the composite column
c1,c2,c3 Numerical coefficients
cr Thickness of concrete cover
D, d Outer diameter of circular hollow steel section
Ea Modulus of elasticity of steel section
Ec Modulus of elasticity of concrete for long term
Ecm Secant modulus of the concrete
Er Modulus of elasticity of reinforcement
Es Modulus of elasticity of reinforcement (EC4); Modulus of elasticity of steel
section (BS 5400)
Fy Characteristic strength of steel section
Fyr Characteristic strength of reinforcement
fcc Characteristic strength of concrete due to confinement effect
fck, f c Characteristic cylinder strength of concrete
fcu Characteristic cube strength of concrete
fry, fsk Nominal yield strength of reinforcement
fy Characteristic strength of steel section
h Overall width of the composite column
h1 Concrete width perpendicular to the plane of bending
h2 Concrete thickness in the plane of bending
Ia Second moment of area of steel section
Ic Second moment of area of concrete
Is Second moment of area of reinforcement (EC4); Second moment of area of
steel section (BS 5400)
le Effective length of column
M Moment capacity of column
Mu Ultimate moment capacity of column
N Axial load of column
Npl Plastic resistance of the composite column
Nu Squash load of column
rm Modified radius of gyration of steel shape, pipe or tubing in composite
columns
ry Radius of gyration of a member about its minor axis
S Plastic section modulus
t Web thickness
ac Strength coefficient of concrete
l Slenderness of column
ga Partial factor of safety for steel section
gc Partial factor of safety for concrete
gs Partial factor of safety for reinforcement
m Moment resistance ratio
w Density of concrete
 H.S. Saw, J.Y.R. Liew / Journal of Constructional Steel Research 53 (2000) 121147 123

Fig. 1. Types of composite columns.

tion time, (3) inherent ductility and toughness for use in repeated and reversal loads,
(4) enhanced fire resistance characteristics when compared to plain steel, (5) higher
load carrying capacity due to the composite action of steel and concrete and con-
finement of concrete in the case of in-filled columns, and (6) higher rigidity for use
in lateral-load resisting systems.
Several commonly used methods, which are now available for designing com-
posite columns, include Eurocode 4: Part 1.1 [1] (to be referred to as EC4), BS
5400: Part 5 [2] (to be referred to as BS 5400), and AISC LRFD [3] (to be referred
to as LRFD). In the United States, design provisions for composite columns are
currently included in both the ACI-318 [4] and LRFD specifications. The concept
of applying LRFD to composite columns was first presented by Furlong in 1976 [5].
The method of EC4 was based on the work of Roik and Bergmann [6] and others,
while the recommendations given in the BS 5400 were developed by Basu and Som-
erville [7] and modified by Virdi and Dowling [8].
The aim of the present investigation is to assess the current methods for evaluating
the ultimate load behaviour of steelconcrete composite columns and to recommend
design guidelines for these columns. Major codes such as EC4, BS 5400 and LRFD
are studied and compared to investigate their differences. Typical designs were car-
ried out to determine the similarities and discrepancies of various codes. Discrep-
ancies in the predicted nominal strengths by the three codes were also highlighted.
Finally, conclusions are given and future research needs are identified.

2. Comparison of EC4, BS 5400 and LRFD methods

A study performed by Liew et al. [9] shows that the use of EC4, BS 5400 and
LRFD for the design of composite columns does not necessarily converge to the
same result. This is because of the differences in load and resistance factors, design
parameters and other design considerations such as creep, eccentricity, etc. The fol-
lowing sub-sections will discuss and compare some of these parameters used in the
three codes.
124 H.S. Saw, J.Y.R. Liew / Journal of Constructional Steel Research 53 (2000) 121147

Table 1
Terms and specifications used in the three codes of practice for composite column design

Items Code Specifications

Steel EC4 Fe 360 (235 N/mm2), Fe 430 (275 N/mm2) and Fe 510 (355
N/mm2) when thickness, t40 mm [3.2.2]
BS 5400 Grade 43 (275 N/mm2) or 50 (355 N/mm2), t16 mm
[11.1.2.1]
LRFD A36 (248 N/mm2), A50 (345 N/mm2), A55(379 N/mm2)
[I2.1(d)]
Concrete EC4 fck50 N/mm2 where fck is the concrete cylinder strength
[3.1.2.2]
BS 5400 fcu20 N/mm2 for concrete filled tubes
fcu25 N/mm2 for concrete encased sectionsa where
fcu=concrete cubic strength [11.1.2.2]
LRFD 20.7 N/mm2f c55.1 N/mm2 for normal weight concrete
f c27.6 N/mm2 for lightweight concrete
where f c=concrete cylinder strength [I2.1(c)]
Steel contribution ratio EC4 0.2d0.9, where d is the ratio of the contribution of the
steel section to total axial capacity. [4.8.3.1.(3)]
BS 5400 Encased steel sections: 0.15ac0.8
Hollow steel sections: 0.1ac0.8
where ac is the ratio of the concrete contribution to total
axial capacity [11.4.1]
LRFD Cross-sectional area of steel section 4% of total cross-
sectional area [I2.1(b)]
Limiting slenderness EC4 l2.0 about any axis of bending [4.8.3.1.(3)]
BS 5400 Ratio of effective length to least lateral dimension of
composite column should not exceed
30 for concrete encased sections
55 for concrete filled circular hollow sections
65 for concrete filled rectangular hollow sections. [11.1.5]
LRFD KL/r200, where KL is the effective length and r is the
radius of gyration of the section.
Steel reinforcement EC4 0.3%As/Ac4% for concrete encased columns
In concrete filled hollow sections normally no reinforcement
is required except for fire resistance [4.8.3.1.(3)]
BS 5400 For concrete encased columns: stirrups to be placed
throughout length of column, spacing not to exceed 200
mm. At least 4 longitudinal bars to be provided [11.3.9]
LRFD Not less than 0.007 in2 (0.178 mm2) of reinforcement per
inch (25.4 mm) of bar spacing. [I2.1b]
Local buckling EC4 d/t90e2for CHS of outside diameter d and thickness t
h/t52e for RHS of greater overall dimension h, thickness
tb/tf44e for partially encased I-sections of width b,
thickness tf
and e=(235/fy)1/2, fy is the yield strength of steel [4.8.2.4]
BS 5400 Minimum wall thickness, t for hollow sections are
tbs(fy/Es)1/2 for each face of width bs in RHS
tDe(fy/8Es)1/2for CHS of outside diameter De [11.1.2.1]
LRFD Minimum wall thickness, t for hollow sections are
tb(fy/Es)1/2 for each face of width b in RHS
tD(fy/8Es)1/2 for CHS of outside diameter D [I2.1(e)]
(continued on next page)
 H.S. Saw, J.Y.R. Liew / Journal of Constructional Steel Research 53 (2000) 121147 125

Table 1
continued

Items Code Specifications

Modulus of elasticity of
concrete
Longitudinal shear limits
at steelconcrete interface
Partial safety factor for
material at ULS
Partial safety factor for
load at ULS
EC4 From EC2, Ecm=9.5(fck+8)0.33, where Ecm is in kN/mm2 and
fck is the 28 day concrete cylinder strength in N/mm2
[3.1.4.1]
BS 5400 450 times fcu [11.3.1]
LRFD w1.5(f c)0.5, where w is the unit weight of concrete
EC4 0.6 N/mm2 for fully encased I-sections
0.4 N/mm2 for concrete filled hollow sections
0.2 N/mm2 for flanges in partially encased I-sections
0 N/mm2 for webs in partially encased I-sections [4.8.2.7]
BS 5400 0.6 N/mm2 for encased sections
0.4 N/mm2 for hollow sections [11.1.3]
LRFD Accounted for with suitable detailing
EC4 Steel section: 1.1
Concrete: 1.5
Steel reinforcement: 1.15
Shear connector: 1.25 [2.3.3.2]
BS 5400 Steel section: 1.1
Concrete: 1.5
Steel reinforcement: 1.15
Shear connector: 1.10 [4.2.1]
LRFD Compression: 1/0.85=1.18
Flexure: 1/0.9=1.11
EC4 1.35 Dead Load+1.5 Imposed Load [2.3.3.1]
BS 5400 (1.051.20) Dead Load+(1.31.5) Imposed Load, Johnson
[12]
LRFD 1.20 Dead Load+1.6 Imposed Load [A4.1]

a
fcu of 20 and 25 N/mm2 will be equivalent to 16 and 20 N/mm2 cylinder strength, respectively,
assuming that concrete cylinder strength is 0.8 fcu.

2.1. Terms and specifications used

Some of the terms and specifications used in the three codes for the design of
composite column are summarised in Table 1. The relevant clause for each item is
given in brackets. The items of comparison include the grades of steel and concrete,
steel contribution ratio, limiting slenderness, steel reinforcement ratio, local buckling,
modulus of elasticity of concrete, longitudinal shear limits and partial factors of
safety for material and load at ultimate limit state.

2.2. Modulus of elasticity of concrete

The modulus of elasticity of concrete, Ec, is the main material property that is
defined rather differently by the three codes when calculating the column slenderness
126 H.S. Saw, J.Y.R. Liew / Journal of Constructional Steel Research 53 (2000) 121147

Table 2
Recommended design values for concrete modulus of elasticity

Codes Modulus of Concrete, Ec Unfactored Valuec

EC4 0.8 Ecm/1.35a 0.8 Ecm/1.35

BS 5400 450 fcu 670 fcu
LRFD w1.5f cb w1.5f c

a
Ec should be modified accordingly if conditions of long term effect are satisfied.
b
w is the unit weight of concrete in lb/ft3, f c is the concrete cylinder strength in ksi and Ec gives
result in ksi.
c
This value is used to evaluate the test results with no factor of safety.

Table 3
Modulus elasticity of concrete (kN/mm2) evaluated using different codes

Codes C20/25 C25/30 C30/37 C35/45 C40/50 C45/55 C50/60

EC4 17.2 18.1 19.0 19.9 20.7 21.3 21.9

BS 5400 11.3 13.5 16.7 20.3 22.5 24.8 27.0
LRFD 21.5 24.1 26.4 28.5 30.5 32.3 34.0
670 fcu 16.8 20.1 24.8 30.2 33.5 36.9 40.2

ratio. This is because Ec has no unique value, and it decreases as the strain increases.
The recommended values for Ec given by the three codes are shown in Table 2.
Table 3 gives the values of Ec obtained for various grades of concrete based on
the recommendation of Table 2. The concrete used is the normal weight concrete
of density 2400 kg/m3. The grade of concrete is referred by either cylinder or cube
strength. For example, C25/30 refers to cylinder strength=25 N/mm2 and cube
strength=30 N/mm2.
In the calculation of the modulus elasticity of concrete, the variation can be due
to the inclusion of a creep factor for Ec and/or use of cracked section, with or without
limitation on the radius of gyration.
The values of Ec evaluated using EC4 (Table 4) are slightly higher since creep
effect is considered separately. For BS 5400, the values of Ec are relatively low for
lower grades of concrete because the creep effect is considered and it is based on
uncracked concrete section in the analysis. However, it is observed that at higher
grades of concrete, the values of Ec obtained using BS 5400 are larger than EC4.
The LRFD gives significantly higher values since a cracked section is used in the
analysis. In EC4, the value of Ec is used to calculate the effective elastic flexural
stiffness of the cross-section and Ecm is a better measure of the modulus of elasticity
of concrete under short term loading.
From the comparison study shown in Table 3, it is not possible to conclude which
code gives a more conservative prediction of Ec. This is because results of a true
comparison would depend on which strut curves are being used and on other details
 H.S. Saw, J.Y.R. Liew / Journal of Constructional Steel Research 53 (2000) 121147 127

Table 4
Slenderness ratio given in the three codes of practice

Codes Non-dimensional slenderness, l

Concrete encased sections Concrete filled sections

EC4

le Aafy+0.85 Acfck+Asfsk
p EaIa+0.8 EcmIc/1.35+EsIs 1/2

le Aafy+Acfck+Asfsk
p EaIa+0.8 EcmIc/1.35+EsIs
1/2

BS 5400

le Asfy/gs+Acfcu/gc+Arfry/gr
p EsIs+EcIc+ErIr 1/2

le Asf y/gs+Acfcc/gc
p EsIs+EcIc+ErIr 1/2
a

LRFD 1/2 1/2
Ar Ac Ar Ac
Fy+0.7Fyr +0.6f c Fy+Fyr +0.85 f c
le As As le As As
rmp

E+0.2 Ec
Ac
As
rmp
E+0.4 Ec
Ac
As

a
This expression applies only for circular hollow sections. As for rectangular hollow sections, the
expression is the same as the concrete encased sections.

of the methods. Complete designs would have to be done for a wide range of situ-
ations before any conclusions could be drawn.

2.3. Slenderness ratio

The definition of slenderness ratio in the three codes of practice is shown in Table
4. It is important to note that the value of Ec (see Table 2) used in evaluating the
column slenderness ratio in EC4 and LRFD does not include a factor of safety.
However, in BS 5400, a partial factor of safety is included in the calculation of the
column slenderness ratio. If the factored value of Ec is used, there will be a reduction
in the column slenderness ratio, which will in turn lead to an increase in the load
carrying capacity.
The slenderness ratio in LRFD is expressed in terms of the modified radius of
gyration, rm. For an encased section, the modified radius of gyration is defined as
the larger of the radius of gyration of the steel section or 30% of the thickness of
the gross composite section (the radius of gyration of a solid rectangular is about
30% of its depth). The steel shape and the concrete portion of the composite sections
both contribute to flexural resistance. If the steel dominates, the radius of gyration
of the steel section should be used in calculating rm. On the other hand, if the
reinforced concrete section dominates the flexural resistance, the radius of gyration
of the concrete section should be used for calculating the slenderness parameter. In
the absence of a more rigorous method, LRFD recommends that the larger of the
128 H.S. Saw, J.Y.R. Liew / Journal of Constructional Steel Research 53 (2000) 121147

radius of gyration values for steel or concrete be used in calculating the slenderness
index l/rm.
Once the member slenderness of a composite column is determined, the com-
pression resistance of the composite column may be evaluated. The column buckling
curves, which are essentially the same in both the ECCS [10] and the British Stan-
dards, are based on an extensive study by Beer and Schulz [11]. Their investigation
highlighted the effect which the shape of the structural steel section and its method
of fabrication have on its ultimate load carrying capacity. This is a result of the
differences in geometry of cross-section, magnitude and distribution of residual
stresses and the member initial out-of-straightness. These curves are identified as a,
b and c and they have been found to apply equally well to composite columns.
As for LRFD, only one strut curve is used and the position of the strut curve is
close to the strut curve a, with a slightly smaller value for slenderness parameter, l
less than 1 and a slightly larger value for l greater than 1 (see Fig. 2).

2.4. Effect of long term loading

Under sustained loading, creep of the concrete causes an increase in the lateral
deflection of a composite column and a reduction in its strength and stiffness.
EC4 is the only code that treats the effect of long term loading separately. The
influence of creep and shrinkage is taken into account, if both the eccentricity of
loading is less than twice the cross-section dimension and the slenderness, l is more
than the limiting values given in Table 4.6 of EC4. A more accurate account on the
effective modulus of elasticity of concrete under long term loading is given in the
code by considering the portion of the design load that is permanent. If half of the
design load is permanent, this will reduce the effective elastic modulus of concrete
to 75% of its short term value.
In BS 5400, design strengths are obtained by dividing the characteristic (or
nominal) strengths by gc. At the ultimate limit state, gc=1.5 is used for concrete,

Fig. 2. Column buckling curves based on EC4 and LRFD provisions.

 H.S. Saw, J.Y.R. Liew / Journal of Constructional Steel Research 53 (2000) 121147 129

which gives design strengths in axial compression and bending of 0.55 fcu and 0.45
fcu, respectively. These values have been further reduced to 0.45 fcu and 0.4 fcu to
account for the influence of creep. No other allowance is made for creep in columns.
In LRFD, influence of creep is considered in the expressions for effective stiffness,
Em, when the column is subjected to sustained compressive loading. The influence
of creep as well as the influence of concrete cracking have been accommodated
adequately by the use of 40 percent of the nominal initial stiffness of confined con-
crete inside steel tubes, and only 20 percent of that stiffness for unconfined concrete.
This is reflected in the value of c3 given in LRFD (Section I2-2). The influence of
creep is usually small if a large area of steel is being used. Hence, every composite
column designed in accordance to LRFD must have a structural steel area not less
than 4 percent of the gross cross-section area of the column.

2.5. Effect of concrete confinement

For composite column with concrete filled circular sections, the confinement effect
of concrete increases the concrete resistance, but at the same time reduces the axial
resistance of the steel section.
In EC4, the reduction of concrete strength by 0.85 may be omitted for concrete
filled composite columns since the development of concrete strength is better achi-
eved due to protection against the environment and against splitting of concrete. The
effect of confinement is considered when both the relative slenderness, l is less than
0.5 and the eccentricity of the normal force does not exceed the value d/10, in which
d is the outer diameter of the circular hollow section. In BS 5400, the limitation of
the confinement effect is less restrictive than EC4. It is not dependent on the eccen-
tricity of the axial loading and the confinement effect is neglected when l is greater
than 1.0. For most practical columns, l=1 corresponds to a length to diameter ratio
between 24 and 29, but for simplicity, BS 5400 set the limit of l/d to 25. In LRFD,
no confinement effect of concrete is taken into account for design of concrete filled
circular hollow sections.

2.6. Flexural resistance of circular columns

In LRFD, an approximate equation for the calculation the moment resistance of

doubly symmetric sections (Section C.I4 of LRFD) is given as follows:
1 h2 AwFy
MuSFy (h22cr)ArFyr
3
A F
2 1.7f ch1 w y (1)

This formula is based on the plastic moment capacity of the steel section, the longi-
tudinal reinforcement, and the concrete that is compressed along the edge of the
cross-section. It is assumed that at least 1/3 of the longitudinal bars in a cross-section
can be considered concentrated in a position located cr from the edges of the cross-
section. In order to obtain the third term of Eq. (1), the web of shapes encased in
concrete is considered to be tension reinforcement for a concrete cross-section with
130 H.S. Saw, J.Y.R. Liew / Journal of Constructional Steel Research 53 (2000) 121147

a flexural depth taken as half the overall thickness in the plane of bending. It is
apparent that when bending occurs about the minor axis, the web of the steel shape
does not contribute to the plastic section modulus used in the first term of Eq. (1).
Even though the web contributes a minor portion of the major axis section modulus,
at ultimate moment about the major axis of encased shapes, the neutral axis is not
at mid depth but is closer to the compression edge as concrete participated in resisting
flexure. The resulting increase in the distance between total internal tension and
compression forces is more than sufficient to offset the apparent double use of web
area as a part of the term S and as a part of the third term in Eq. (1). The side wall
regions for round or rectangular filled tubes may permit a similar term for Aw, but
no recommendation can be proposed at this time. It is conservative to use Aw=0 for
filled pipe. This explains the relatively smaller flexural resistance in LRFD for con-
crete filled circular composite sections as compared to BS 5400 and EC4.
The expression given in BS 5400 for the flexural resistance of a concrete filled
circular composite section is given as:
Mu0.91 Sfy (10.01 m) (2)
where m=(100/S)[t(Dt)2 (q1 sin q1+cos q11)+0.25r (D2t)3 {(cos 3q1)/30.25
sin q1 (psin 2q12q1)}] and other symbols are as defined in BS 5400. The value
of m is represented graphically in Figure 100.5 of BS 5400, Appendix C.
The derivation of Eq. (2) is given in Appendix C of Johnson [12]. This formula
is obtained based on the integration of a small elemental section of the circular cross-
section. The process is rigorous but more exact as compared to EC4 and LRFD.
However, it must be noted that a safety factor has already been incorporated in
deriving this formula.
For EC4, the moment resistance of the concrete filled circular composite section
can be approximately calculated using the formula for rectangular sections.

3. Strength comparsion of EC4, BS 5400 and LRFD

A comparison is made between the nominal strength of columns computed using

EC4, BS 5400 and LRFD methods. All the partial factors of safety for materials
used in EC4, BS 5400 and the resistance factor used in the LRFD are set to unity.
This will give an unbiased comparison of the resistance predicted by the three
methods since each method has its own resistance factors which are used with the
corresponding load factors.
For this study, the column is assumed to be pin supported at both ends and it is
loaded symmetrically, i.e. the top and bottom are subject to the same axial force and
bending moment causing a single curvature bending. The modulus of elasticity of
concrete is evaluated according to each code of practice (refer to Table 2). As for
the calculation of long term modulus of elasticity of concrete in EC4, it is assumed
that 75% of the design load acting on the column is permanent.
The nominal or characteristics material strength used in the comparison study
 H.S. Saw, J.Y.R. Liew / Journal of Constructional Steel Research 53 (2000) 121147 131

are: Concrete: fc=C25/30, r=2400 kg/m3; Steel: fy=355 kN/mm2, E=210 kN/mm2;
Reinforcement: fry=460 kN/mm2, E=210 kN/mm2.
The dimensions of the composite section used in the comparison study are:
Fully encased steel section: Steel section: UC 254254107, Concrete dimension:
400400 mm; Reinforcement: 412 mm, one at each corner, 150 mm from axes
of symmetry.
Partially encased steel section: Steel section: UC 254254107; Reinforcement:
48 mm, one at each corner, 85 mm from axes of symmetry.
Concrete filled rectangular hollow section: Steel section: 200300 mm and 10
mm thick; Reinforcement: nil.
Concrete filled circular hollow section: Steel section: outer diameter 406.4 mm
and 8.8 mm thick; Reinforcement: nil.
The evaluations of the nominal axial capacity and moment interaction curve based
on the three methods are performed. The moment interaction curve of EC4 has been
modified such that the boundary of the imperfection moment has been subtracted
from the outer interaction curve and a factor of 0.9 is multiplied to m. In the figures,
the base resistances, Nu and Mu are values calculated in accordance with EC4. Hence,
the M/Mu value computed from the method of BS 5400 and LRFD does not converge
to unity for N/Nu=0.
3.1. Axial capacity
Figs. 36 compare the nominal axial resistance of the four types of composite
columns evaluated using the three different codes. The resistances predicted by each
method are compared relatively to one another.

Fig. 3. Unfactored axial capacity of a fully encased composite column.

132 H.S. Saw, J.Y.R. Liew / Journal of Constructional Steel Research 53 (2000) 121147

Fig. 4. Unfactored axial capacity of partially encased composite columns.

The comparison studies show that there is a considerable difference in the axial
resistance of the encased composite columns (see Figs. 3 and 4). At low slenderness
ratio, EC4 gives the highest value, followed by LRFD and BS 5400. The level portion
of BS 5400 is due to the reduction factor of 0.85 used for axially loaded columns
with lex/h and ley/b12. For slender columns (lex/h or ley/b12), BS 5400 requires
that the axial load acts at an eccentricity value of 0.03b. It is noticed that at a slender-
ness ratio of about 15, the long term effect becomes significant in the EC4 approach,
thereby reducing the axial capacity of the column. It is interesting to note that the
axial capacity of LRFD becomes larger than EC4 and BS 5400 as the slenderness
ratio increases. This may be due to the strut curves used and also the difference in
formula used for evaluating the slenderness in the LRFD approach.
Among the four types of composite columns, the concrete filled rectangular com-
posite section gives the least discrepancies (see Fig. 5). Both EC4 and BS 5400 use
strut curve a for rectangular hollow sections. For LRFD, the column curve is close
to strut curve a of EC4, with a slightly smaller value for relative slenderness less
than 1 and a slightly larger value for relative slenderness greater than 1. This may
explain why the axial capacity of LRFD becomes larger than EC4 and BS 5400 as
the slenderness increases.
For concrete filled circular composite sections (see Fig. 6), the discrepancies in
the axial capacity are due mainly to the consideration of the confinement effects. It
is important to note that BS 5400 gives the highest increase in strength due to con-
 H.S. Saw, J.Y.R. Liew / Journal of Constructional Steel Research 53 (2000) 121147 133

Fig. 5. Unfactored axial capacity of concrete filled rectangular composite columns.

Fig. 6. Unfactored axial capacity of concrete filled circular composite columns.

134 H.S. Saw, J.Y.R. Liew / Journal of Constructional Steel Research 53 (2000) 121147

Fig. 7. Interaction curves for fully encased composite column bending about the major axis.

Fig. 8. Interaction curves for fully encased composite column bending about the minor axis.
 H.S. Saw, J.Y.R. Liew / Journal of Constructional Steel Research 53 (2000) 121147 135

Fig. 9. Interaction curves for partially encased composite column bending about the major axis.

Fig. 10. Interaction curves for partially encased composite column bending about the minor axis.
136 H.S. Saw, J.Y.R. Liew / Journal of Constructional Steel Research 53 (2000) 121147

Fig. 11. Interaction curves for concrete filled rectangular composite column bending about the major
axis.

finement effect (except for the portion due to the reduction factor of 0.85) and this
effect is ignored at a slenderness ratio much higher than the EC4 method.

3.2. MN interaction curves

For encased sections, the three methods show considerable discrepancies in the
interaction curves (see Figs. 713) and the discrepancies are more distinct for minor
axis bending. At low slenderness ratio, EC4 gives the highest capacity. However,
as the slenderness ratio increases, the capacity decreases due to the effect of long
term loading. The level portion of the BS 5400 is due to the reduction factor that
allows for construction tolerances. It is noticed that when the slenderness ratio
increases, the capacity of LRFD decreases less significantly as compared to BS 5400
and EC4. For bending about the minor axis, the capacity of LRFD is even higher
than BS 5400 and EC4. One possible explanation for this is the use of different strut
curves. For encased sections, both EC4 and BS 5400 use strut curve b for bending
about the major axis and strut curve c for bending about the minor axis. However,
for LRFD, only one strut curve is used and the position of this strut curve is higher
than the strut curves b and c (Fig. 2).
The interaction curves of the concrete filled rectangular composite section are in
reasonably good agreement with one another (Figs. 11 and 12). At low slenderness
ratio, the strength curves of EC4 are higher than BS 5400 and LRFD in both the
 H.S. Saw, J.Y.R. Liew / Journal of Constructional Steel Research 53 (2000) 121147 137

Fig. 12. Interaction curves for concrete filled rectangular composite column bending about the minor
axis.

major and minor axes. This is because of the inclusion of point E in the interaction
curve (see Figure 4.12 of EC4). For the evaluation of relative slenderness, both EC4
and BS 5400 use strut curve a which is close to the strut curve of LRFD (see Fig.
2). This may explain the close proximity of the interaction curve. Moreover, the
effect of long term loading is not considered in EC4 as the limiting conditions are
not satisfied.
For concrete filled circular composite sections (see Fig. 13), the three methods
show substantial discrepancies. The capacity of LRFD is always the lowest because
no confinement effect is considered for the axial capacity and also a simplified for-
mula which is highly conservative for circular sections is used for the evaluation of
the moment capacity. As for BS 5400, a more rigorous approach is used for the
evaluation of moment capacity of the circular sections. This gives rise to a higher
value than EC4 and LRFD.

4. Comparison with test results

The predicted column strength based on EC4, BS 5400 and LRFD are compared
with the test results of SSRC Task Group 20 [13] and Shakir et al. [14]. In this
comparison, both the axially loaded and eccentrically loaded fully encased, concrete
filled circular and rectangular composite columns are considered.
In order to compare the predicted strength with the test results, the material partial
138 H.S. Saw, J.Y.R. Liew / Journal of Constructional Steel Research 53 (2000) 121147

Fig. 13. Interaction curves for concrete filled circular composite columns.

safety factors of both steel and concrete are taken to be equal to unity. It should be
noted that in the method of BS 5400, the maximum compressive strength of concrete
is taken as 0.67 fcu (instead of 0.45 fcu) for evaluating the squash load and 0.6 fcu
(instead of 0.4 fcu) for evaluating the ultimate moment.
The expression for the modulus of elasticity of concrete, Ec used in the prediction
of experimental results is given in the last column of Table 2, with the partial safety
factor of concrete taken as unity. The modulus of elasticity of steel used for the
calculation is 200 kN/mm2 (29000 ksi) as in the test performed by SSRC Task Group
20 [13] and 205 kN/mm2 by Shakir et al. [14].
In this study, the initial out-of-straightness of the column is assumed to be in the
form of a half sine curve. The amplification factor for the moment can be approxi-
mated by 1/(1P/Pe) where P is the applied load and Pe is the Euler buckling load
of the column [15]. This amplification factor is used for the analysis of second order
moment in BS 5400 for eccentrically load columns since guidance for calculating
the second order moment is not given in the code. For EC4 and LRFD, the method
given in each code for the analysis of the second order moment are used. Concrete
reinforcement is not included in the calculation since it is not mentioned in the above
two references.
Table 5
Comparison of predicted and measured axial capacity of a fully encased composite column, SSRC Task Group 20 [13]a

Section (mmmm) hc (mm) bc (mm) fy fc (N/mm2) Le (m) Test EC4 (kN) BS 5400 (kN) LRFD (kN)
(N/mm2) (kN)

127114UB 29.76 177.8 165.1 248 C18/22.5 0.229 1566 1332(1.18) 1127*(1.39) 1215(1.29)
1.168 1370 1228(1.12) 1038*(1.32) 1182(1.16)
2.083 1366 1023(1.34) 1013(1.35) 1111(1.23)
2.997 1281 697(1.84) 779(1.64) 1008(1.27)
3.886 1027 500(2.05) 593(1.73) 886(1.16)
203152 UB 52.09 254.0 203.2 248 C18/22.5 2.134 2544 1964(1.30) 1657*(1.54) 2054(1.24)
304.8 254.0 3229 2453(1.32) 2077*(1.55) 2331(1.39)
355.6 304.8 3807 3010(1.26) 2545*(1.50) 2674(1.42)
Average (1.43) (1.50) (1.27)

a
* Denotes values multiplied by 0.85 according to BS 5400, Clause 11.3.2.1. Figures in brackets represent test/predicted load ratio.
H.S. Saw, J.Y.R. Liew / Journal of Constructional Steel Research 53 (2000) 121147
139
 140

Table 6
Comparison of predicted and measured axial capacity of a concrete filled circular composite column, SSRC Task Group 20 [13]a

D (mm) t (mm) As (cm2) fy (N/mm2) fc (N/mm2) Le (m) Is (cm4) Test (kN) EC4 (kN) BS 5400 LRFD (kN)
(kN)

215.9 4.09 27.2 292 C22.9/27.5 2.22 1526.8 1650 1489(1.11) 1267*(1.30) 1364(1.21)
C30.0/37.0 2264 1697(1.33) 1403*(1.61) 1543(1.47)
6.00 39.6 392 C22.9/27.5 2180.7 2442 2136(1.14) 2010*(1.21) 2025(1.21)
350 C30.0/37.0 2869 2199(1.30) 1981*(1.45) 2055(1.40)
120.9 3.76 13.84 312 C21.1/26.1 1.05 237.6 721 644(1.12) 614*(1.17) 585(1.23)
2.31 636 530(1.20) 493(1.29) 491(1.30)
5.53 20.0 343 C21.1/26.1 1.05 334.3 1010 876(1.15) 881*(1.15) 817(1.24)
C24.2/29.2 1.05 1090 901(1.21) 893*(1.22) 840(1.30)
95.0 3.65 10.5 350 C25.0/30.0 1.37 109.4 667 461(1.45) 459(1.45) 429(1.55)
1.42 583 457(1.28) 449(1.30) 425(1.37)
1.98 529 390(1.36) 371(1.43) 368(1.44)
Average (1.24) (1.33) (1.34)

a
* Denotes values multiplied by 0.85 according to BS 5400, Clause 11.3.2.1. Figures in brackets represent test/predicted load ratio. D is the outer diameter
of the circular sections.
H.S. Saw, J.Y.R. Liew / Journal of Constructional Steel Research 53 (2000) 121147
Table 7
Comparison of predicted and measured axial capacity of short concrete filled rectangular composite column, Shakir et al. [14]a

Section RHS As (cm2) Ac (cm2) fy fc (N/mm2) Le (m) Test EC4 (kN) BS 5400 (kN) LRFD (kN)
(N/mm2) (kN)

120805 18.55 72.60 357.5 C32.9/41.6 0.2 850 902(0.94) 736*(1.15) 863(0.98)
18.27 72.80 341.0 C33.5/42.6 0.2 900 867(1.04) 706*(1.27) 828(1.09)
18.27 72.80 341.0 C36.2/46.2 0.2 920 887(1.04) 721*(1.27) 844(1.09)
18.66 72.40 362.5 C33.4/42.4 0.2 950 918(1.03) 750*(1.27) 879(1.08)
18.66 73.40 362.5 C32.4/40.8 0.2 955 914(1.04) 746*(1.28) 876(1.09)
1501005 22.39 121.40 346.7 C36.0/46.0 0.1 1370 1213(1.13) 978*(1.40) 1147(1.19)
0.2 1210 1213(1.00) 978*(1.24) 1145(1.06)
22.39 120.00 346.7 C36.2/46.2 0.1 1340 1211(1.11) 976*(1.37) 1145(1.17)
0.2 1200 1211(0.99) 976*(1.23) 1143(1.05)
22.60 120.60 340.0 C36.6/46.6 0.1 1300 1210(1.07) 973*(1.34) 1143(1.14)
0.2 1190 1210(0.98) 973*(1.22) 1141(1.04)
22.60 120.10 340.0 C37.2/47.2 0.1 1320 1215(1.09) 976*(1.35) 1148(1.15)
0.2 1200 1215(0.99) 976*(1.23) 1146(1.05)
Average (1.03) (1.28) (1.09)

a
* Denotes values multiplied by 0.85 according to BS 5400, Clause 11.3.2.1. Figures in brackets represent test/predicted load ratio.
H.S. Saw, J.Y.R. Liew / Journal of Constructional Steel Research 53 (2000) 121147
141
142 H.S. Saw, J.Y.R. Liew / Journal of Constructional Steel Research 53 (2000) 121147

4.1. Axially loaded columns

Tables 57 show the comparison of the predicted and measured axial capacity of
a fully encased, concrete filled circular and rectangular composite columns.
For axially loaded columns, when the slenderness ratios (lex/h and ley/b) are smaller
than 12, BS 5400 requires that a reduction factor of 0.85 be applied to the ultimate
load in accordance with Clause 11.3.2.1 of BS 5400.
It should be noted that in EC4, the effect of long term loading (creep and
shrinkage) is taken into account. Therefore at high slenderness ratio, the ratio of the
test/predicted strength becomes larger. However, this long term effect is not con-
sidered in the methods of BS 5400 and LRFD.
The comparison studies show that the predicted results are generally lower than
the test results. For fully encased composite sections (see Table 5), LRFD gives the
closest prediction with an average of 27% lower than the test results and BS 5400
gives the most conservative with an average of 50% lower than the test results.
Table 6 shows that for concrete filled circular composite sections, EC4 gives the
closest prediction to the test results, followed by BS 5400 and LRFD. They are on
the average about 24%, 33% and 34% lower than the test results respectively.
For concrete filled rectangular composite sections (see Table 7), EC4 and LRFD
give very close prediction to the test results with BS 5400 giving a slightly lower
prediction. EC4 and LRFD are about 3% and 9% lower than the test results, respect-
ively, with BS 5400 on the average about 24% lower. The high (test/predicted) ratio
of BS 5400 is because of the reduction factor of 0.85 being applied to the failure load.

4.2. Eccentrically loaded columns

Tables 8 and 9 show the comparison of the predicted and measured strengths
of uniaxially loaded fully encased and concrete filled circular composite columns,
respectively. Table 10 shows the comparison of results for biaxially loaded rectangu-
lar composite columns. In these studies, second order analysis is used to calculate
maximum moment in predicting the strength of eccentrically loaded columns.
Table 8 shows that for fully encased composite sections, LRFD gives the closest
prediction, on the average of 23% lower than the test results, followed by EC4 and
BS 5400 which are on average 28% and 34% lower than the test results, respectively.
For concrete filled circular composite sections (see Table 9), the average ratio of
the test/predicted strength is much higher than unity. BS 5400 gives the closest
prediction to the test results, followed by EC4 and LRFD. They are on average about
16%, 57% and 114% lower than the test results, respectively. The large difference
in the LRFD (test/predicted) values is attributed to the confinement effect of concrete
which is not considered in the design and also to the usage of a more simplified and
conservative formula in the calculation of the moment resistance.
For concrete filled rectangular composite sections (see Table 10), both EC4 and
BS 5400 are on average about 17% lower than the test results, while LRFD is about
32% lower.
Overall, the average test/predicted values from Tables 510 show that the pre-
Table 8
Comparison of predicted and measured eccentrically loaded concrete encased section, SSRC Task Group 20[13]a

Section hc (mm) bc (mm) fy (N/mm2) fc (N/mm2) Le (m) ey (cm) Test (kN) EC4 (kN) BS 5400 LRFD (kN)
(mmmm) (kN)

127114UB 177.8 165.1 232 C19.3/24.1 2.083 1.91 716 585(1.22) 525(1.36) 595(1.20)
29.76
2.03 747 566(1.32) 511(1.46) 580(1.29)
0.726 1.91 898 876(1.03) 801(1.12) 660(1.36)
2.03 1014 856(1.18) 779(1.30) 642(1.58)
2.54 738 740(1.00) 698(1.06) 578(1.28)
1.156 1.27 996 932(1.07) 863(1.15) 753(1.32)
729 701(1.04) 647(1.13) 566(1.29)
2.083 2.54 627 504(1.24) 458(1.37) 524(1.20)
2.997 1.27 716 479(1.49) 464(1.54) 612(1.17)
2.54 529 367(1.44) 364(1.45) 469(1.13)
3.886 2.54 440 285(1.54) 285(1.54) 408(1.08)
3.81 347 236(1.47) 244(1.42) 340(1.02)
5.08 329 205(1.60) 215(1.53) 293(1.12)
Average (1.28) (1.34) (1.23)

a
Figures in brackets represent test/predicted load ratio.
H.S. Saw, J.Y.R. Liew / Journal of Constructional Steel Research 53 (2000) 121147
143
 144

Table 9
Comparison of predicted and measured eccentrically loaded concrete filled hollow circular section, SSRC Task Group 20[13]a

D (mm) t (mm) As (cm2) fy (N/mm2) fc (N/mm2) Le (m) e (cm) Test (kN) EC4 (kN) BS 5400 LRFD (kN)
(kN)

127.0 2.34 9.16 289.6 C35.2/40.2 1.067 1.55 569 458(1.24) 482(1.18) 276(2.06)
2.36 534 376(1.42) 407(1.31) 220(2.42)
3.99 400 224(1.79) 309(1.29) 157(2.55)
4.50 351 200(1.76) 288(1.22) 144(2.44)
5.56 307 165(1.86) 251(1.22) 123(2.50)
114.8 2.4 8.45 331.0 C23.4/28.4 1.067 1.32 374 338(1.11) 377(0.99) 252(1.48)
4.32 240 185(1.30) 223(1.08) 133(1.80)
13.9 89 59(1.51) 99(0.90) 54(1.65)
114.8 3.58 12.5 331 C28.8/35.4 1.067 3.07 436 361(1.38) 370(1.18) 207(2.11)
6.05 302 188(1.61) 254(1.19) 138(2.19)
8.18 262 140(1.87) 208(1.26) 111(2.36)
17.6 129 66(1.95) 117(1.10) 61(2.11)
Average (1.57) (1.16) (2.14)

a
Figures in brackets represent test/predicted load ratio. D is the outer diameter of the circular sections.
H.S. Saw, J.Y.R. Liew / Journal of Constructional Steel Research 53 (2000) 121147
Table 10
Comparison of predicted and measured biaxially loaded concrete filled rectangular composite column, Shakir et al. [14]a

SectionRHS As(cm2) Ac(cm2) fy(N/mm2) fc(N/mm2) ex(mm) ey(mm) Test(kN) EC4(kN) BS LRFD(kN)
5400(kN)

120805 18.55 72.60 357.5 C32.9/41.6 0 8 407.2 311(1.31) 291(1.40) 292(1.39)

18.27 72.80 341.0 C33.5/42.6 12 8 348.0 293(1.19) 272(1.28) 252(1.38)
18.27 72.80 341.0 C36.2/46.2 42 28 198.5 177(1.12) 189(1.05) 158(1.26)
18.66 72.40 362.5 C33.4/42.4 24 40 206.8 173(1.20) 179(1.16) 164(1.26)
18.66 73.40 362.5 C32.4/40.8 60 16 209.8 192(1.09) 196(1.07) 168(1.25)
1501005 22.39 121.40 346.7 C36.0/46.0 0 0 1003.0 789(1.27) 800(1.25) 752(1.33)
22.39 120.00 346.7 C36.2/46.2 15 10 596.2 519(1.15) 465(1.28) 413(1.44)
22.60 120.60 340.0 C36.6/46.6 45 30 329.2 310(1.06) 319(1.03) 752(1.27)
22.60 120.10 340.0 C37.2/47.2 75 50 254.6 215(1.18) 254(1.00) 194(1.31)
Average (1.06) (1.17) (1.32)

a
Figures in brackets represent test/predicted load ratio. The true length of the column is 2.74 m and the effective length of the column in the major and
minor axes are 3.21 m and 2.94 m, respectively.
H.S. Saw, J.Y.R. Liew / Journal of Constructional Steel Research 53 (2000) 121147
145
146 H.S. Saw, J.Y.R. Liew / Journal of Constructional Steel Research 53 (2000) 121147

dicted column strengths using the three methods are on the conservative side and
they are in reasonable agreement with the test results.

5. Conclusions and further research

The comparisons and conclusions given in this paper are based on a limited range
of published experimental data. Within the present scope of work and investigation
carried out, the following conclusions may be made:

1. The approach in BS 5400 gives a higher increase in concrete strength due to the
confinement effect of concrete at low slenderness ratio as compared to EC4.
2. There is a marked difference between EC4 and BS 5400 in the critical slenderness
parameter, l, above which the confinement effects are ignored. In EC4, the con-
finement effects are ignored when l is less than 0.5, whereas in BS 5400, the
confinement effects are ignored when l is less than 1.0.
3. For encased composite sections, the axial capacity obtained by LRFD is higher
than EC4 and BS 5400. This may be due to the strut curve used in LRFD which
is above that of EC4 and BS 5400.
4. The column interaction curves of concrete filled rectangular hollow sections have
the least discrepancies among the three codes of practice. One possible reason is
the close proximity of the strut curves used in the three codes.
5. For concrete encased circular composite sections, the capacity obtained by LRFD
is always the lowest because no confinement effect is considered for the axial
capacity and also a simplified formula is used for the evaluation of the moment
capacity which is highly conservative for circular sections.
6. The predicted column strengths using the three methods are on the conservative
side and are in reasonable agreement with the available test results.
7. The method of EC4 is recommended because it covers a wide scope of the latest
research findings influencing the resistance of composite columns. Moreover, the
development of MN interaction curve is direct and this enables hand calculation
to be done.

The present studies have identified a number of possible areas that could be investi-
gated further: (i) treatment of multi-storey building columns subject to sidesway and
large moments, (ii) effects of lateral loads on the design and detailing of beam-to-
column joints, (iii) design and detailing of shear connectors between steel and con-
crete interface when the moment is the predominant force, (iv) effects of enhance-
ment of triaxially contained high-strength concrete, (v) elastic, creep and shrinkage
shortening of column columns, (vi) ductility of composite columns subject to earth-
quake and reversal loads, (vii) impact and energy absorption characteristic, and (viii)
design of large composite columns with irregular shapes or openings.
One particular area that requires more in-depth investigation is the physical and
mechanical bond between the structural steel shape and the surrounding concrete.
Additional work is needed to quantify the requirement for shear connectors in the
 H.S. Saw, J.Y.R. Liew / Journal of Constructional Steel Research 53 (2000) 121147 147

cases where the column is subject to sidesway and when bending is significant. The
question of shear and moment transfer through a beam-column joint is yet to be
answered adequately. This is particularly important to understand the shear transfer
mechanism due to reversal wind loads, which could lead to serious degradation of the
joint strength. The behaviour of confined concrete in the steel tube requires special
consideration especially when high-strength concrete and high-strength steel are
used. Although some studies have been made on the above, further investigations
are still required with the aim of providing more in-depth studying for design
implementations.

Acknowledgements

The investigation presented in this paper is part of the research programme on

advanced analysis for the design of composite structures carried out in the Depart-
ment of Civil Engineering, National University of Singapore, and is funded by
research grants RP 3981614.

References

[1] Eurocode 4. Design of composite steel and concrete structures, Part 1.1: General rules and rules for
buildings. Brussels: Commission of European Communities, March 1992.
[2] BS 5400. Steel, concrete and composite bridges: Part 5: Code of practice for design of composite
bridges. London: British Standards Institution, 1979.
[3] Load and resistance factor design specification for structural steel buildings. Chicago, IL: American
Institution of Steel Construction (AISC), 1993.
[4] ACI Committee 318. Building code requirements for reinforced concrete. Detroit, MI: American
Concrete Institution, 1992.
[5] Furlong RW. Strength of steel-encased concrete beam columns. Journal of the Structural Division,
ASCE 1976;93(5):11324.
[6] Roik K, Bergmann R. Eurocode 4: Composite columns. Report EC4/6/89, University of Bochum,
June 1989.
[7] Basu AK, Sommerville W. Derivation of formulae for the design of rectangular composite columns.
Proceedings of the ICE, supplementary volume, 1969:23380.
[8] Virdi KS, Dowling PJ. The ultimate strength of composite columns in biaxial bending. Proceedings
of the ICE 1973;55(2):25172.
[9] Liew RJY, Saw HS, Yu CH. Composite column design in buildingsAssessment of current methods
and interim guidance. Research Report No. CE 026/98. National University of Singapore, May 1998.
[10] European Convention for Constructional Steelwork (ECCS). European recommendations for steel
construction. Construction Press, 1981.
[11] Beer H, Schulz G. The theoretical basis of the new column curves of the European Convention for
Constructional Steelwork (in French). Construction Metallique, No 3, September 1970.
[12] Johnson RP, editor. Composite structure of steel and concrete. vol. 2. London: Granada, 1986:50.
[13] SSRC Task Group 20. A Specification for the Design of Steel-Concrete Composite Columns. AISC
Engineering Journal, Fourth Quarter, 1979:10115.
[14] Shakir-Khalil H, Mouli M. Further tests on concrete-filled rectangular hollow-section columns. Struc-
tural Engineer 1990;68(20):40512.
[15] Chen WF, Lui EM. Structural stability: theory and implementation. New York: Elsevier, 1987.