J. Construct.

Steel Research 15 (1990) 191-213

International Assessment of Design Guidance for

Composite Columns

A . S. E l n a s h a i , A . Y. E 1 - G h a z o u l i & P. J. D o w l i n g
Department of Civil Engineering, Imperial College of Science and Technology,
London SW7 2BU, UK

(Received 5 September 1989; revised version received and accepted

2 November 1989)

ABSTRACT

This paper reviews the current state of design code provisions for
composite steel~concrete columns from West Europe, North America,
Japan and the People's Republic of China, in addition to draft proposals
from East Europe and Australia. A general comparison of code recom-
mendations in terms of design basis, slenderness considerations, material
properties and dimensional limitations is presented. This is followed by a
quantitative assessment of capacity calculation for a specific cross-section
under axial loading and combined axial loading and flexure. A range of
slenderness (in terms of equivalent length divided by relevant section
dimension) from zero to 30 is covered. It is concluded that large
discrepancies exist between codes, even those using essentially the same
design methodology, and sometimes the same experimental data base. This
emphasizes the case for harmonization to arrive at a higher degree of
uniformity of code design procedures and more rational safety margins.

NOTATION

AC A r e a of concrete
Ag Gross cross-sectional area
Ar A r e a of longitudinal reinforcing bars
As A r e a of steel section
b B r e a d t h of steel section
bc B r e a d t h of concrete section
191
J. Construct. Steel Research 0143-974X/90/$3-50(~ 1990 Elsevier Science Publishers Ltd
England. Printed in Great Britain
 192 A. S. Elnashai, A. Y. El-Ghazouli, P. J. Dowling

B Overall breadth of composite section

Cm A factor relating actual moment diagram to an equivalent uniform
moment diagram
E Modulus of elasticity
Ec Modulus of elasticity for concrete
Concrete cube strength
ilVcyI Concrete cylinder strength
Fy Steel yield strength
h Height of steel section
h~ Height of concrete section
H Overall depth of composite section in direction of bending
I Second moment of area
L,, Equivalent length
Ley Major axis equivalent length
Lcz Minor axis equivalent length
Mp Flexural plastic capacity
Muir Ultimate moment capacity
Np Axial plastic capacity
Np~ Squash load of concrete section
Np, Squash load of steel section
sPc Factor to account for concrete strain limit
P Axial load
PE Euler buckling load
Y Radius of gyration
R Strength reduction factor
tpl Steel plate thickness
w Unit weight of concrete

Concrete contribution factor

Factors to account for confined concrete strength
6 Moment magnification factor
A Non-dimensional slenderness parameter
/J Strength reduction factor
Steel contribution factor
T Shear stress
4, Strength reduction factor

1 INTRODUCTION

Whereas reinforced concrete can be considered as a composite material,

the more conventional definition of composite steel and concrete struc-
 Design guidance for composite columns 193

tures uses steel sections in conjunction with concrete. Such systems ideally
combine the advantages of both components. In this context, composite
columns offer an attractive solution to problems such as local and overall
buckling for steel columns, and shear failure and deterioration for
concrete columns. Moreover, this form of structural member exhibits
excellent earthquake-resistant properties, namely high stiffness, strength,
ductility and energy absorption capacity.
Composite columns can be broadly classified as either hollow sections
filled with concrete or steel sections encased in concrete. In concrete-filled
columns, there is a mutual enhancement of ductility, as the tube provides
confinement for the concrete which in turn prevents the inward buckling of
the tube. On the other hand, encased columns offer high strength and
ductility, while providing the steel section with fire protection. Partially
encased columns offer additional advantages such as high impact resist-
ance, simplified beam-to-column connections and reduced or omitted
shuttering.
Different methods for the design of composite columns exist in codes of
practice.l-6 A composite column may be treated in some methods as a
steel column strengthened by concrete, whereas other methods may
consider it as a reinforced concrete column with special reinforcement.
Furthermore, the strength of a column may be evaluated as the sum of
strengths of both components.
Existing code differences can be attributed to two main reasons:
difference in design philosophy, and numerical quantification. Whereas
the former covers the fundamental considerations, such as strain distribu-
tion and compatibility, the latter is a consequence of the use of a specific
experimental data base to arrive at actual design expressions. Even when
two codes use the same philosophy and the same experimental results,
some discrepancies are to be expected in estimating the final section
properties for a given load, or the capacity of a pre-defined section. This
may be due to the differences in safety factors, allowable material
properties, limiting dimensions, consideration of long-term loading, etc. It
is therefore not surprising that various codes would yield a wide range of
designs for the same conditions.

2 D E S I G N CODES, A BRIEF DESCRIPTION

2.1 AIJ standards (Architectural Institute of Japan)

Composite construction, termed steel reinforced concrete (SRC) in Japan

is in common use. This is attributed to its superior earthquake-resistant
 194 A. S. Elnashai, A. Y. EI-Ghazouli, P. J. Dowling

behaviour compared with reinforced concrete structures. Hence, extensive

research on this subject has been undertaken by Japanese researchers
during this century. Concrete-encased steel columns are frequently used
for medium- to high-rise buildings. Concrete-filled steel tubes are used for
columns of subway stations and also in buildings.
The design is based on the allowable stress method. However, the
ultimate strength of the m e m b e r has to be evaluated for earthquake
resistance. The cross-sectional strength is calculated by superimposing the
strength of both the steel and reinforced concrete components. Full plastic
distribution is used with a reduction factor in the concrete strength to
account for the strain limit in concrete. This factor is assigned a different
value for filled sections, where the concrete strength is enhanced by
triaxial confinement. The m e t h o d is applicable to asymmetrical sections
and columns under biaxial bending.
As framed columns are usually stocky, because of the design for large
lateral forces under earthquake loading, less attention has been given to
slender columns. Beam-columns are similarly designed by the superim-
position of steel and reinforced concrete beam-columns. The Euler
buckling load is used with a reduced concrete stiffness and factors of safety
for both materials. M o m e n t magnifiers are used for slenderness considera-
tion in the reinforced concrete portion. The code also specifies minimum
axial load eccentricities.
To achieve sufficient ductility, flexural failure should precede shear
failure in a member. Moreover, the axial compression force and the
width/thickness ratios of steel sections are limited to specified values. On
the other hand, loads are transferred between steel and concrete by bond.
If the bond strength is insufficient, direct bearing is used.

2.2 BS 5400 Part 5 (British Standards Institution)

Early tests on encased columns in the UK led to the cased-strut m e t h o d of

design adopted in the British Standards (BS 449 Part 2 and BS 5950 Draft).
This empirical m e t h o d was developed from earlier design procedures for
steel columns, and was not based on fundamental research on composite
columns. Concrete-filled tubes are not applicable to this method, nor
members where the concrete section is heavily reinforced or is significantly
larger than the steel section. Research on improved methods has since
then been carried out mainly at Imperial College. 7"s This work formed the
basis of the current bridge code (BS 5400 Part 5).
Code provisions in BS 5400 are based on limit state design with loading
factors and partial safety factors for materials. The ultimate m o m e n t is
 Design guidancefor composite columns 195

calculated from plastic stress distribution over the cross-section, and an

approximation for the interaction curve for axial load and moment is used.
Reduced concrete properties are used to account for the effects of creep
and the use of the uncracked concrete section in stiffness calculation.
Minimum eccentricities due to construction tolerances are considered. In
the case of filled columns, an enhanced concrete strength due to triaxial
confinement may be taken into account.
Slenderness effects are accounted for by the use of a non-dimensional
slenderness parameter and the European buckling curves used for the
design of steel columns. The choice of the appropriate curve depends on
the geometry of the cross-section. The different curves reflect the effect of
residual stresses and initial imperfections in the member. The uncracked
concrete section is used in stiffness calculation. The use of the slenderness
parameter is consistent with tl~e design of steel columns as the method
reduces to the bare steel column design when the concrete portion is
removed. Consequently, the method is applicable to symmetric sections
only, and is restricted to the range of sections catered for in the European
buckling curves. For uniaxial bending, distinction is made for the presence
or lack of restraint in the direction orthogonal to the bending axis.
Expressions are given for biaxial bending for filled circular hollow sections
as well as encased sections.

2.3 ACI-318-83 (American Concrete Institute)

The building code of the American Concrete Institute uses the limit state
design format with loading factors and capacity reduction factors. The
strength of a composite column is computed as for reinforced concrete
members. Failure is defined in terms of a 0.3% strain limit for any concrete
fibre. This failure strain is used with a set of neutral axis locations to arrive
at an interaction curve for thrust and moment. Slenderness effects are
analysed in terms of moment magnifiers using a reduced Euler load. The
expression for equivalent stiffness includes a creep factor, and a cracked
concrete stiffness is considered. Minimum eccentricities are specified to
cover construction tolerances.

2.4 AISC-LRFD-1986 (American Institute of Steel Construction)

The load and resistance factor design uses the limit state design with
loading factors and capacity reduction factors. The design of composite
columns is based on the design equations for steel columns. However, the
slenderness and area parameters are modified for the presence of
 196 A . S . Elnashai, A. Y. El-Ghazouli, P. J. DoMing

concrete. The code recognizes the confinement effect in filled columns and
uses a simple interaction formula for uniaxial and biaxial bending.
M o m e n t magnifiers are specified to account for slenderness effects except
for the case of uniaxial bending only when a steel buckling curve is used as
no minimum eccentricities are specified. Load transfer should be provided
by direct bearing at the connections.

2.5 Eurocode 4 and DIN 18806

Eurocode 4 was drafted to cover the design of composite steel and

concrete structures. Limit state design is applied with partial safety factors
for materials. Three methods of design are available in the EC4 draft. The
first two methods are derived from the West German practice (DIN
18806) where slenderness effects are analysed in terms of bending
m o m e n t s as in reinforced concrete design. However, as opposed to
m o m e n t magnification, an equivalent strength reduction is derived using
the European Steel Buckling Curves in conjunction with the cross-
sectional interaction curve. Full plastic distribution is used to obtain the
axial to flexural strength interaction curve, whereas a simplified polygon-
al approximation is introduced in the second method. The third m e t h o d is
based on the cased strut design used in the early British Standard BS 449
Part 2, where steel column design is used with an increase in the radius of
gyration to account for the presence of concrete.

2.6 Ministry of Water Conservancy and Electric Power, People's Republic

of China

The code includes specifications for filled tubes only. Design is based on
ultimate limit state, and slenderness is considered using different curves
for strength reduction depending on material, steel ratio and slenderness
ratio assuming an initial imperfection of Length/800.

2.7 Australian and East European Standards (unpublished)

The Australian code provisions do not include design of composite

columns. However, proposals for draft code provisions have been
discussed, '~'1° where a limit state philosophy is adopted. Two methods are
proposed; one relates to reinforced concrete column design, and the other
is akin to steel codes, namely m o m e n t magnifier and strength reduction.
Specifications in East Europe are currently under consideration, where
plastic design approach is allowed.
 Design guidance for composite columns 197

3 CODE DIFFERENCES

3.1 Design basis

3.1.1 Origin of method

This is by and large dictated by whether steel designers were considering
composite construction as an improvement over bare steel design, or
concrete designers were attempting to achieve higher strengths. In the
former case, steel buckling curves were used, and in the latter reinforced
concrete design methods were adopted.
The modified steel design methods are used in AISC-LRFD and BS
5400, both explicitly reducing to the bare steel section. The American
Concrete Institute makes use of reinforced concrete column design,
whereas EC4 and DIN use a combination of both approaches. Distinctly,
the Japanese Code (AIJ) uses a superimposition of the individually
calculated capacities.

3.1.2 Loading and resistance

Various codes use significantly different loading factors and combinations.
On the other hand, capacity reduction is achieved through one of two
approaches. Whereas the North American codes (ACI and AISC) use
capacity reduction factors 4' (of 0.7 and 0.85 respectively), the European
practice is to use partial safety factors on materials. Different values for
these are given for concrete, steel and reinforcing bars (typically 1-5,
1.0-1-1 and 1.15 respectively).
In contrast to the above limit state-based codes, the Japanese code
(AIJ) uses the allowable stress approach. However, owing to the severe
earthquake problem in Japan, the ultimate strength has to be verified.

TABLE 1
Specifications for Concrete Compressive Strength

Code Encased Filled

AIJ (allowable strength) Fcyl (1-15sPc) Fcyl

(ultimate strength) Fcyl (0-85-2-58pc) 0"85Fcyl
BS 5400 (axial compression) 0"675Fcu 0"675Fcu(/~)
(bending) 0.6Fcu 0.6F¢u (~)
EC4 0-85Fcyl 0-85Fcyl(13)
DIN 18806 0-60Fcu 0.70Fcu (~)
ACI 318-83 0-85Fcyl 0-85Fcyl
198 A.S. Elnashai, A. Y. EI-Ghazouli, P. J. Dowling

3.2 Concrete strength

Table 1 gives a brief comparison of r e c o m m e n d e d values for concrete

compressive strength.
It is n o t e w o r t h y that some codes use cube whereas others use cylinder
strength. F u r t h e r m o r e , there is some controversy on the effect of
confinement on the crushing strength. For an assumed ratio of Fcu/Fcy~ of
1.2, the range of compressive strength (normalized by Fcy0 is b e t w e e n 0.7
and 1.0, for unconfined concrete.

3.3 Interaction curves

Discrepancies observed for minor and major axis bending interaction

curves are discussed in detail in Section 4.3, and shown in Figs 3-12. Table
2 gives a general comparison.

TABLE 2
Basis of Interaction Diagrams

Code Specifications

AIJ Plastic distribution (PD) Reduced Fcyl

BS 5400 PD for Muir Approximate surface
EC4, DIN Full PD Reduced moment capacity
Simplified surface in method B
ACI 318-83 Equilibrium and 0.3% Concrete strain limit
compatibility of strains
AISC-LRFD PD for Mult Approximate surface

As discussed below, there are large variations in capacity under

c o m b i n e d axial loading and bending, depending on the code used. The
comparison given herein does not consider biaxial bending cases.

3.4 Equivalent stiffness

For slenderness considerations and deflection calculations, the equivalent

(E/) is required. Various codes use different approaches to arrive at a
value of E,. which takes account, directly or indirectly, of cracked stiffness
and creep. The r e c o m m e n d e d Ec values for the codes considered herein
are p r e s e n t e d in Table 3.
 Design guidance for composite columns 199

TABLE 3
Equivalent Stiffness

Code Ec Comments

BS 5400 450Fcu Low, to account for use of uncracked section

and creep
EC4 600Fcyl Same as above, but creep accounted for in
expression
ACI 57 000(Fcyl) 1/2(psi) Higher, as a cracked section is used, with a
creep factor
AISC wl"S( Fcyl)1/2 High, as a cracked section is used
(w = unit weight of concrete in lb/ft 3 and FcyI
in ksi)

TABLE 4
Slenderness Considerations

Code Specifications

AIJ Strength reduction R = (1 - vP/PE)

BS 5400, DIN, EC4 Strength reduction
AC1318-83 M o m e n t magnification
M i n i m u m eccentricity
AISC-LRFD M o m e n t magnification
for compression and bending
Critical load for axial
compression
where v = factorofsafety
E u r o p e a n steel buckling curves
t5 = Cm/(1 - P/CkPE)
= Cm/( 1 - P/PE)
rcomposit e = rstee 1
(not less than 0.3 depth)

The expression for (E/)equivalen t m a y or may not use a creep coefficient,

depending on the code.

3.5 Slenderness considerations

The effect of slenderness is accounted for in one of two ways, strength

reduction or m o m e n t magnification. E u r o p e a n codes (BS 5400, D I N and
EC4) use a strength reduction, so does the AIJ code. Alternatively, A C I
uses a m o m e n t magnification factor as well as the definition of a m i n i m u m
eccentricity as discussed below. The A I S C r e c o m m e n d a t i o n s use a
m o m e n t magnifier for bending with axial load, and a critical load for pure
compression, with no m i n i m u m eccentricity. Table 4 gives a comparison of
the specifications.
 200 A. S. Elnashai, A. Y. El-Ghazouli, P. J. Dowling

TABLE 5
Minimum Eccentricities

Code Minimum eccentricity

AIJ 5%H (for the concrete portion design)

BS 5400 3% B (in bridges, would be 4% in buildings)
(axial load limited to 85% of the short column minor
axis capacity)
EC4 None (in methods A and B, 4% in method C)
DIN 18806 None
AC1318-83 (0.6 + 0.03H) (axial load limited to 85% of the plastic capacity)
inches
AISC- LRFD None

3.6 Minimum eccentricities

R e c o m m e n d e d values of minimum eccentricity are given hereafter in

Table 5.
It is noteworthy that limiting the axial capacity to a percentage of the
flexural plastic capacity imposes a further limitation on the minimum
eccentricity, hence values quoted above may not be strictly comparable in
some cases.

3.7 Shear transfer

All codes assume full interaction, but some impose restrictions on the
shear stress at the steel/concrete interface. It is customary to use direct
bearing, or provide shear connectors, if and where the specified limiting
shear stress is exceeded. Table 6 lists the various approaches and values
adopted by design codes.
Furthermore, design of shear connectors, if required, is given in detail in
most codes.

3.8 Material properties

Limits on the extreme values of concrete crushing and steel yield strength
are given in most codes, as shown in Table 7. The upper limits for steel
yield stress are considered to ensure that concrete remains stable until
steel reaches yield.
 Design guidance for composite columns 201

TABLE 6
Specifications for Shear Transfer

Code Specifications

AIJ Detailed design for ENC

Direct beating for FIL if bond is insufficient
BS 5400, EC4 Connectors provided if
T> 0-6 N/mm 2 for ENC
or 0-4 N/mm 2 for FIL
DIN18806 Connectors provided if
r > 0.55-0.80 N/mm 2 for ENC
reduced by 30% for FIL
(for Fcu = 25-55 N/mm 2)
ACI 318-83 Direct beating at connections
Concrete strength for beating = 1.7(0.7 a) (Fcy0
AISC-LRFD Direct bearing at connections
Concrete strength for beating = l'7(0"6a)(Fcyl)

aCapacity reduction factor (resistance factor) for beating.

TABLE 7
Specifications for Material Properties

Code Concrete strength Steel yield

(N/mm2) (N/mm2)

AIJ FcyI -> 15 Fy - 360 (for tpl < 40 mm)

Fcyl -->18 (with re-bars) Fy -< 340 (for tpl > 40 mm)
BS 5400 Fcu -> 20 (for encased)
Fcu -> 25 (for filled)
EC4 F~yl - 20 Fy -< 450 (350 in method C)
DIN 18806 F~u -> 25
ACI 318-83 Fcy I --> 17 Fy -< 330
AISC-LRFD 20 ~< Fcy I ~< 50 Fy --<365

3.9 Dimensional limitations

3.9.1 Steel wall thickness

Expressions given for steel wall thickness reduce to the same values in
North American and Western European codes. However, only EC4 and
DIN include values for partially encased sections. On the other hand, AIJ
is the only code that considers limits for fully encased sections due to the
design for earthquake loads.
 202 A. S. Elnashai, A. Y. El-Ghazouli, P. J. Dowling

3.9.2 Steel and concrete contributions

Limits on concrete contribution to the plastic axial load are included in
Western European codes. North American and Japanese codes define
direct limitations on the steel or concrete area. These differences are
summarized in Table 8.

TABLE 8
Steel and Concrete Contributions

Code Specifications

AIJ A~/Ag >- 0-8%

BS 5400 0.15 < ac < 0-80 (for encased) (c~,, = NI~/Nr, )
0.10 < c~c < 0.80 (for filled)
D I N 18806 0.20 < ~ < 0.90 (~ = Np/Np)
(~= l-a~,if
reinforcement is
At~A,. >- 3% disregarded)
ACI A,/A,, < 0.08 (for filled)
0411 < Ar/Ag ~ 0"08 (for encased)
AISC-LRFD A~/Ag > 0-04

3.9.3 Slenderness and section dimensions

Limits on slenderness are included in some codes which correspond to the
range of available experimental data as shown in Table 9.

TABLE 9
Slenderness R a n g e s

Code Specifications

BS 5400 Lc/B <- 3(1 (encased)

Lc/B <- 55 (circular filled)
L,/B <- 60 (rectangular filled)
EC4, D I N ,~ -< 2.0 (comparable to BS 5400)

3.9.4 Reinforcement details

For practical considerations, several codes include limitations on detailing
of reinforcement bars in encased columns as shown in Table 10.
 Design guidance for composite columns 203

TABLE 10
Specificationsfor ReinforcementDetailing

Code Specifications

BS 5400 Stirrups spacing-<200 mm, with4 corner longitudinalbars

Cover to surface of steel shape -> 50 mm
AC1318-83 Stirrups spacing <-B/2
-<16longitudinalbar diameter
-<48stirrup diameter
Stirrup diameter ->B/50
AISC-LRFD Stirrups spacing -<2/3B
Stirrup area ->0.007of the spacingin inches
(similarto ACI in the practicalrange of columns)

4 C O M P A R A T I V E RESISTANCES

4.1 General

Comparison between the resistances given by the methods in Eurocode 4

and the British Standard BS 540011 was extended to cover the two North
American codes AISC-LRFD-1986 and ACI 318-83. The Japanese
standard was not included as the method is based on allowable stress
design and hence would give resistances that cannot be compared directly,
unless loading factors were considered. Only method A of EC4 was
included in the comparison, as method B is an approximation for A, and
method C is a simplification of the method in BS 5400.
The cross-section considered is an Encased Universal Column
356 x 368 x 129 with h/hc and b/be ratios of 0.74 and 0.75, respectively.
The steel is of 340 N/mm 2 yield strength and the casing is of Grade 35
cylinder strength. For the bridge code BS 5400, the partial safety factor for
steel and the minimum eccentricity were modified to 1.0 and 0-04,
respectively. These modifications would be required to use the code for
building design. The Grade 35 cylinder strength was assumed equivalent
to Grade 40 cube strength in lieu of ISO 3893 recommendations. For the
methods of ACI 318-83 and EC4, the maximum live load was assumed
equal to the maximum dead load in the calculation of the effective
stiffness. The value of modulus of elasticity for concrete was estimated in
accordance with the requirements given in each code. For the methods of
ACI and AISC, the moment magnification required for slenderness
 204 A. S. Elnashai, A. Y. El-Ghazouli, P. ,l. Dowling

i 0

! BS 5 4 0 8 I
---~CI 318-83 :
8 -.- A I S C - L ~ ' F D

<< \ .~
0

"%. "~

0 0
2 4 B 8 I~ 12 14 16 18 2 0 2 2 2 4 2 6 2 8

SLENDERNESS ( Lez/b c I

Fig. 1. Unfactored axial capacity.

i O

-q B S 5"#80
---ACI 318-83
8 -- F~ISC-LPFD
0

0 6 "~.

E~ 3 N ""

P- S

Q 1

0 0
2 4 g 8 10 12 14 16 18 2 ~ 22 24 26 28

SLENDERNESS ( Lez / b c )

Fig. 2. Factored axial capacity.

considerations was equivalently included as a reduction in the moment

capacity.
In Figs 1-12 the resistances given by the different methods are shown
with the values of Np and Mp calculated according to EC4. Comparison of
resistances under combined flexure and compression was made assuming
constant eccentricities measured from the origin of the intercepts of the
 Design guidance for composite columns 205

1 0
--EC4 A "l
o 9 .... BS $400

0 8 ...... . &~ -,-RISC-LRFD

.% •
o 7
% ..
%.% '..
0 S •% -..
r~
%.% ".. ~%
~os % ". %
%. .. %
7"04 %. '. %
%.% '.. %

o 3
%'~ i %%
"% ; |
o 2

01

00
0z o4 o6 oo ~o tz
M/Mp

Fig. 3. Major axis bending (unfactored); L e y / h c = O.

1 o

o q - - EC4-~ !

o
%'%.% ~ ' % % " . . . .
0
o_ %.% % ...

%.% %% ' ~

?t
•.
o

o "% % o~

o
\ #i
h :!]
o

o otoze3e4esaea708ogto
M / Mp

Fig. 4. Major axis bending; Ley/h c = O.

curves. The loading factors were not included in the comparison because
of the different load combinations given in each code, and only the factors
on the resistance side were considered.
In the following quantitative comparison, the code is interpreted from
the latest available version, as would be the case in design office practice.
The possibility of inadvertently misinterpreting certain clauses exists.
 206 A . S. Elnashai, A . Y. E l - G h a z o u l i , P. J. D o w l i n g

" Ec;~
k~ . . . . . . . . -- ~qISC-LRFD

z ~:~ -.,.,.,%.\,~.,>)1
.,,.,
~ 1 @ 2 ~ 3 ~ 4 05 O 6 ~ 7 ~ 8 09

M/Mp

Fig. 5. Minor axis bending (unfactored); L,=/b~ = O.

EE4-~
BS 5400
• I--~CI-31S
"~ i__ AISC-LPFD

o
~z
\,
Z

8 %. \

0
\%
I,\
01 02 ~ 3 04 8 S ~ 6 0 ?
I %. 0 S 8

H,'I'Ip

Fig. 6. Minor axis bending: L~z/b~. = O.

However, this may highlight an inherent ambiguity in the code presenta-

tion. Finally, numbers appearing in square brackets indicate the clause
number in the code under consideration.

4.2 Comparison of axial capacities

4.2.1 Unfactored axial capacity

The unfactored axial capacities of the four methods are shown in Fig. 1
normalized to that of EC4. Slenderness is represented in terms of the ratio
 Design guidance for composite columns 207

0.9

0.8

0.7
__.x .__._
0.6

0_0.5
7"
-N0.
"-,,,".-..,., \
0,3

0.2
"-..?->,. "~.~, "~7'
0.1
,,,, j
0.0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.0 0.9 1.0

M/Mp

Fig. 7. Major axis bending; Ley/hc = 12.

0.75

°17° :ZEc4-~

" 20 . 0 , " N~ % ',

o,o "X', I
o.o5 i~,~.\
,i '.1
-'~,,J'...
0.00
0,1 0.2 0.3 0,4 0 I5 0 I 6 0 I 7 0 I 8 0" 9 11 0

M/Mp

Fig. 8. Major axis bending; Ley/h¢ = 20.

of the pin-ended column length to the cross-sectional dimension in the

direction of minor axis bending.
For very low slenderness ratios, BS 5400 gives the lowest capacity: 19%
less than EC4 and about 4% less than ACI and AISC. This is attributed to
(1) the reduced concrete strength of 0-6Fcu (0.45 factored) [11.1.4] to take
account of creep which is not included in the equivalent stiffness, and (2)
the minimum eccentricity requirement for short columns calculated from
the limitation on axial capacity (0.85 of the minor axis axial capacity)
[11.3.2.1]. The ACI code gives values which are 15% less than EC4 as a
2(J8 A. S. Elnashai, A. Y. El-Ghazouli, P. J. Dowling

0.48
.. . ; _~, ~
0.44
". - - ACI--31E:
8.48 " - - A!EE-LFF[

C~.36 \\ "'L.

@.32

~_0.28

xZ 0 . 2 4
~k. x
?- 0 . 2 0

0.16

0.12

0,08
\',%\ \.
0,e4
"', '~:~, \".. I
e.eB
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

M/r,lp

Fig. 9. M a j o r axis b e n d i n g : L,?lh, = 3(I.

e.8

0 •7 I ~'~ $ 4 0 0
I --- ACI-31E;
! - - F~IS'I-LRFD
0.6 ......

0,5

Z 0.4 x,-,
Z

°' :L X',\
0.0
e . l e.2 0.3 0.4 0.5 0.6 o.7 0.e 0.9
M/M F,

Fig. 1O, M i n o r axis b e n d i n g ; Lezlbc = 12.

result of limitation of 0-85 of the axial plastic capacity, in accordance with

the minimum eccentricity requirement [10.3.5.1]. For AISC, the values
are 16% less than EC4 because of the reduced concrete strength of 0-6Fcy 1
in encased columns [I2.2].
For higher slenderness ratios, the difference between EC4 and BS 5400
is reduced because the value of equivalent stiffness is lower as it takes
account of creep [4.8.6]. However, the minimum eccentricity requirement
of 0-04 in BS 5400 causes a reduction in capacity in the range of slenderness
 Design guidance for composite columns 209

0.60

0.55k I--EC4-A ]

0.401 "x..
\~
0.35 \<<..~
=
0.30 x~..
"~%\
7- 0.25 . . . . . . . .. "~"~
0,20 . . . . . . . ~''~ . ~'~X,h._ ~

0.00
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

M/Mp

Fig. 11. Minor axis bending; Lez/bc = 20.

0.34
0.32
0,30 • ,. ....BS 5400
--ACI-.~I8
0.28
0.26 \\ ..
0,24 \ '.
0.22 \ \ "'..
0.20 \\'.,
~ 0.10
",0.16
7"
0.14
0.12
0.10
0.00
0.06
0.04
"'-C--.Z':~.N
0.02
0.00
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

M/Mp

Fig. 12. Minor axis bending; Lez/bc = 30.

used in practice. The capacities of ACI are in agreement with other codes,
but decrease at high slenderness due to the following: (1) use of the
cracked section of concrete, (2) the creep factor included in the expression
for equivalent stiffness [10.14] and (3) the minimum eccentricity require-
ment accompanied by a moment magnification factor [10.11.15].
Although the AISC code uses the radius of gyration of the steel section
only, this is limited to 0.3 of the overall section dimension [12.2] which is
the governing case for the minor axis for the current example. At a
 210 A. S. Elnashai, A. Y. EI-Ghazouli, P. J. Dowling

slenderness of 30, the ACI code gives a capacity 48% less than AISC and
about 41% less than EC4 and BS 5400.

4.2.2 Factored axial capacitv

Figure 2 shows the axial capacities given by the different codes, including
the capacity factors in ACI and AISC and the partial safety factors in EC4
and BS 5400. The values are normalized by that of the factored capacity of
EC4. The same general observations are made as for the case of factored
axial capacity, for all codes but ACI. This is a consequence of the
equivalence of the partial safety factors used by EC4 and the resistance
reduction factor used in AISC. In contrast, ACI curve drops substantially,
as a consequence of the inclusion of a reduction factor of 0.7 (as opposed to
0.85 for AISC). This situation is aggravated for higher slenderness,
admittedly above the practical limits, where the discrepancy between AC1
and all the other codes is in the range of 65%.

4.3 Combined flexure and axial compression

4.3. l Unfactored capacity

Figures 3 and 5 show the unfactored major and minor interaction curves
for short columns. Np and Mp are the plastic capacities calculated
according to EC4.
ACI considers a full interaction curve calculated from conditions of
equilibrium and compatibility of strains with a strain limit in concrete of
0-3%. On the other hand, in EC4 the resistance may be calculated from
first principles using appropriate stress-strain curves. Alternatively, the
Eurocode considers a full interaction curve but with full plastic distribu-
tion using rectangular stress blocks, which leads to an overestimate of the
ultimate resistance. However, the m o m e n t capacity in EC4 is accompa-
nied by a reduction of 90% on the m o m e n t capacity [4.8.8] to account for
two assumptions: (1) use of uncracked stiffness for concrete and (2) use of
rectangular stress blocks without due consideration to concrete strain
restrictions.
For major axis bending at zero axial force, EC4 converges to 0.9 instead
of unity, because of the above-mentioned reduction in m o m e n t capacity
(which overcompensates for the plastic distribution in this case). The
capacities of BS 5400, AISC are almost the same (about 98% of the plastic
capacity of EC4) because both codes use plastic distribution for determin-
ing the pure bending capacity. In this case, the capacities of BS 5400 and
AISC are similar to that given by A C I because in the major direction at
this position of the neutral axis, most of the steel would have yielded, and
hence, the plastic distribution does not significantly overestimate the
 Design guidance for composite columns 211

strength. It is also obvious that the difference in concrete strength is not

significant in pure bending in the major axis direction because the main
contribution to the strength is due to the steel section.
Similarly, for minor axis bending at zero axial force, EC4 gives a value of
0-9, as a result of the reduction factor. However, other codes show slightly
larger differences than in the major axis case, as the concrete strength and
the use of plastic distribution assume a more prominent role, because of
the geometry of the cross-section.
With regard to the shape of the interaction curve, EC4 is comparable to
ACI by virtue of the 90% reduction, which compensates for the
overestimation of the strength. BS 5400 and AISC use approximate
interaction curves which are very simple to construct but with some
sacrifice of strength. In BS 5400 no account is taken of the increasing
moment capacity at low axial loads because its availability depends on the
loading history of the column. Owing to the use of simplified interaction
curves, BS 5400 gives resistances 14 and 11% less than EC4 for major and
minor axis bending, respectively, For AISC, this drop in strength is as high as
32 and 36%, respectively.

4.3.2 Factored capacity

The factored resistances for major and minor axis bending were calculated
for columns of slenderness zero, 12, 20 and 30. Figures 4, 7, 8 and 9 show
the interaction curves for major axis bending at a slenderness of zero, 12,
20 and 30 respectively, and Figs 6, 10, 11 and 12 show the corresponding
comparisons for minor axis bending. The moment magnification used in
ACI and AISC was considered by an equivalent strength reduction.
At low levels of compression, the resistance factor specified by ACI may
be increased linearly to reach 0-9 at zero axial force [9.3.2.2]. For AISC,
when the axial force is less than 0.3 of the factored nominal axial capacity,
the nominal flexural strength is determined by a straight-line transition
between the nominal flexural strength determined from the plastic
distribution on the composite cross-section at 0.3, and the flexural strength
at zero axial load [14]. This explains the two resistances that appear in the
figures at low values of axial force for ACI and AISC. Comparison
between the unfactored and factored resistances for major and minor axis
at zero slenderness shows that the capacity factors are, in general, more
conservative (especially the 0-7 factor given by ACI, as it is based on
reinforced concrete design).
For major axis bending, at a slenderness of zero, EC4 and BS 5400 are
generally in good agreement, whereas BS 5400 gives values within 5%
difference. ACI and AISC give values up to 20 and 32%, respectively, less
than EC4, except at low values of axial loads and at high axial loads where
 212 A. S. Elnashai, A. Y. El-Ghazouli, P. J. Dowling

the minimum eccentricities affect the resistance. At slendernesses of I2

and 20 the methods of ACI and BS show a much better agreement but are
about 25% less than EC4, whereas AISC generally gives resistances up to
36% less than EC4, except at very low and very high levels of compression.
As the slenderness increases to 30, the resistances given by EC4 and BS are
similar within a discrepancy of 10%, whereas ACI and AISC are about
33% less than EC4, except at extreme values of axial compression.
For minor axis bending at zero slenderness, BS, ACI and AISC are up to
9, 18 and 33% less than EC4 except at extreme values of axial force. At
slenderness of 12, ACI, BS and AISC show very good agreement except at
extreme values of normal force, but are up to 24% less than EC4. At
slenderness of 20 and 30 the agreement between AISC and BS is still
observed and the discrepancy with EC4 decreases to a maximum of about
12%. Also, ACI shows very low resistances of up to 40% less than EC4 at
slenderness of 30 except at extreme values of axial forces, where the
discrepancy increases because of the minimum eccentricity requirement.
This low resistance is due to the low value of stiffness, which is less than
half that given by other codes.

5 CONCLUSIONS

Large discrepancies between various design codes exist in terms of

philosophy and geometric and strength parameters, even when the same
design philosophy is adopted. The comparative study presented in Section
4 highlights the impact of code differences on safety of column design. As
shown in the figures, a design which conforms to one code may be more
than 60% unconservative according to a different code, notwithstanding
that both adopt the same general principles. This is particularly true for
slenderness ratios other than the middle range. Based on the observations
above, the case for rationalization of code provisions for composite
columns cannot be overemphasized, so that a higher degree of uniformity
in design for a given load, or capacity assessment for an existing design,
may be achieved.

REFERENCES

1. Building Code Requirements for Reinforced Concrete (ACI 318-83).

American Concrete Institute, Detroit, MI, 1983.
2. BS 5400, Steel, Concrete and Composite Bridges, Part 5, Code of Practice for
Design of Composite Bridges. British Standards Institution, London, 1979.
3. Eurocode 4, Composite Steel and Concrete Structures, First Draft, 1984.
 Design guidance for composite columns 213

4. Roik, K. & Bergmann, R., DIN 18806 Part 1. In Introductory Document,

Stability of Metal Structures, A World View, Chapter 12, Composite
members, Structural Stability Research Council, Bethleham, Pennsylvannia,
USA, 1989, pp. 12/38-12/44.
5. Wakabayashi, M., AIJ Standards for Steel-Reinforced Concrete Structures.
In Introductory Document, Stability of Metal Structures, A World View,
Chapter 12, Composite members, Structural Stability Research Council,
Bethleham, Pennsylvannia, USA, 1989, pp. 12/15-12/26.
6. LRFD Specifications for the Design, Fabrication and Erection of Structural
Steel for Buildings. American Institute of Steel Construction, Chicago, IL,
1986.
7. Basu, A. K. & Sommerville, W., Derivation of formulae for the design of
rectangular composite columns. ICE Proc., 40 (1969) 37-60.
8. Virdi, K. S. & Dowling, P. J., Derivation of design formulae for composite
columns. CESLIC Rep. CC10, Imperial College, 1975.
9. Bridge, R. Q., Ansourian, P., Rotter, J. M., Patrik, M. & Pham, L.,
Australian Standards for Composite Construction, Engineering Foundation
Conference on Composite Construction in Steel and Concrete. Henniker,
NH, 1987, pp. 71-83.
10. Lapos, J. & Streleckij, N., East European Draft Recommendations. In
Introductory Document, Stability of Metal Structures, A World View, Chapter
12, Composite Members, Structural Stability Research Council, Bethleham,
Pennsylvannia, USA, 1989, pp. 12/9-12/13.
11. Smith, D. G. E. & Johnson, R. P., Commentary on the 1985 Draft of
Eurocode 4, Composite Steel and Concrete Structures. BRE Rep., 1986, pp.
91-108.