You are on page 1of 10

169

Buckling in continuous
composite beams
MA Bradford and AR Kemp
 University of New South Wales, Australia
 University of Witwatersrand, South Africa

Summary
In comparison with the well-researched lateral
and local buckling that takes place in bare
steel beams, the buckling that takes place in
composite steel-concrete beams is far more
complex, and is still a grey area in structural
mechanics. Plastic design of continuous composite
beams is very advantageous, but this can only
be achieved if buckling is prevented. This article

reviews the research to date into local and


lateral-distortional buckling in the negative
moment region of continuous composite beams,
and provides design proposals. It is concluded
that there is still a need for focused research
into the buckling phenomenon if full advantage
is to be taken of moment redistribution in plastic
design.

Prog. Struct. Engng Mater . 2000; 2:169}178

MODES OF FAILURE

Introduction

The modes of buckling in a composite beam in


negative bending may be local, as shown in Fig. 1,
or overall, as in Fig. 2. While a proliferation of research
has considered both local and overall buckling in steel
beams, with summaries being given in Allen
& Bulson[2], Trahair[3] and Trahair & Bradford[4],
the buckling in composite beams is special as in
negative bending the slab restrains the tensile region
of the steel and the neutral axis is not located at
the mid-height of the web. Short-wavelength local
buckling is not overly difficult to predict, but
the overall mode, as shown in Fig. 2, must necessarily
be what is termed a distortional buckle[5], since
the bottom flange buckles sideways and twists, and is
restrained only by the stiffness of the web, as
shown. This mode of buckling does not make use of
the usual Vlasov assumption that the cross-section
remains undistorted, which underlies all conventional
lateral-torsional buckling analyses[3], but relaxes
this condition so that the cross-section is free to

Significant advantages may accrue in composite


construction if a T-beam is made continuous over an
internal support. Such an internal support may be
a bridge pier or an internal column connection in
a building, and the advantages of the moment
redistribution that can occur at ultimate loading are
described in Oehlers & Bradford[1]. Continuous
composite beams consist of positive moment regions,
in which the slab is subjected to compression and
the steel component predominantly to tension,
and negative moment regions, in which the concrete
has cracked and the reinforcement carries the
tensile forces, with the steel component being
subjected to a combination of negative bending
and compression. The steel component in the negative
moment region is therefore prone to buckling, and
the buckling of this steel component still represents
a grey area in the design of continuous composite
beams.

Abbreviations

Terminology

AISC-LRFD
AS4100
BS5950
CSA16.1
WF

EI
fy
h
i ZC
K
Li
Mps
Mp
r

American LRFD steel structures standard


Australian limit states steel structures standard
British limit states steel structures standard
Canadian limit states steel structures standard
North American wide flange I-section

Copyright ^ 2000 John Wiley & Sons, Ltd.

flexural rigidity
yield stress
rotation at internal support
radius of gyration of compressive portion of web and flange
buckling factor
length between sections of zero moment and maximum negative moment
plastic moment of resistance of steel section
plastic moment of composite beam in hogging region
rotation capacity parameter

Prog. Struct. Engng Mater. 2000; 2:169}178

170

COMPOSITE CONSTRUCTION
CLASSIFICATION OF DUCTILITY
Designers are given the opportunity to analyse load
effects either by elastic or plastic analysis (with or
without moment redistribution), and to determine the
ultimate moment resistance either by using rigidplastic stress blocks[1] or elastic stress limits.
A classification has been introduced into the
American[6], European[7] and Canadian[8] codes in
order to clarify these options, as shown in Table 1.
This paper reviews the phenomena of local and
distortional buckling considering elastic and inelastic
behaviour separately. Experimental and theoretical
results and design implications are considered in each
case. The paper concludes with recommendations for
further research.

Fig. 1 Local buckling of a composite beam in negative bending

Elastic behaviour of class 3 and 4 members


THEORETICAL RESULTS

Fig. 2 Overall or distortional buckling of a composite beam in


negative bending

distort during buckling. Design codes of practice


attempt to approximate this behaviour, but when
applied particularly to distortional buckling their
accuracy is at best questionable.

Elastic local buckling


Accurate modelling of elastic local buckling requires
recourse to a computer program. The finite element
method is such a tool, and forms the basis of a suite of
advanced commercial software packages that have
recently become available. However, the finite element
method is computationally inefficient, and has not
been applied significantly to the analysis of local
buckling of composite beams. The finite strip method,
in which the I-section component is discretized into
longitudinal strips, is an alternative and attractive
method of numerical analysis. In the semi-analytical
form presented by Przemieniecki[9], Cheung[10],
Hancock[11] and others, it is unable to model supports
other than simple, nor loading which varies in the
longitudinal direction, since the longitudinal variation
of buckling displacements is represented by sine
curves. The latter restriction was overcome by Plank
& Wittrick[12], who utilized complex arithmetic to

Table 1 Methods of analysis in some national standards


Method of analysing
Method of calculating
AISC LRFD
Eurocode 3
CSA 16.1
load effects
moment resistance
[6]
[7]
[8]
............................................................................................................................................................................................................................................................
Plastic analysis, or elastic
Yielded stress blocks
Compact
Class 1
Class1 plastic
analysis with moment
redistribution
............................................................................................................................................................................................................................................................
Elastic analysis with limited
Yielded stress blocks
Class 2
Class 2 compact
redistribution
............................................................................................................................................................................................................................................................
Elastic analysis with very little
Elastic stress distribution
Non-compact
Class 3
Class 3
redistribution
limited to yield stress
non-compact
............................................................................................................................................................................................................................................................
Elastic analysis with no
Elastic stress limit below
Slender
Class 4
Class 4 slender
redistribution
yield
............................................................................................................................................................................................................................................................
Copyright ^ 2000 John Wiley & Sons, Ltd.

Prog. Struct. Engng Mater. 2000; 2:169}178

BUCKLING IN CONTINUOUS COMPOSITE BEAMS


impose a phase change to handle shear, in a similar
way to the handling of capacitance and inductance in
electrical engineering. Their plates, though, were
assumed to have infinite length. Azhari & Bradford[13]
extended the Plank & Wittrick analysis, and included
an elastic buckling study of a composite beam that
included a longitudinal stiffener, which Climenhaga
& Johnson[14] found delayed significantly the onset of
local buckling.
The restrictions of the semi-analytical finite strip
method, as described above, were overcome by Lau
and Hancock[15], who utilized spline functions to
depict the longitudinal variation of buckling
displacements. This method was an extension of the
stiffness analysis of Fan & Cheung[16], and enables
a variety of boundary conditions and loading
configurations to be considered. The spline finite strip
method requires many more degrees of freedom than
the conventional harmonic-based finite strip method,
and this has detracted from its popularity. However,
Azhari et al[17] have recently included the so-called
bubble functions into the expressions for the
transverse buckling displacements, and have
demonstrated vast computational savings in this form
of the spline finite strip method.

Elastic distortional buckling


Distortional buckling, as shown in Fig. 2, takes place at
longer half-wavelengths, and is usually prevented in
bridge girders by the cross-bracing shown in Fig. 3.
Although the concept of distortion of the cross-section
during overall buckling was noted as early as the
1930s and is treated in Bleichs[18] well-known text of
1951, its accurate analysis requires the use of
a computer. One of the first such studies appears to be
that of Goldberg et al[19], and the first study of
distortional buckling in composite beams appears to
be that made by Hamada and Longworth[20] using the
finite element method.
Conventional rectangular finite elements with
a plane stress-bending formulation result in an
extremely inefficient modelling of the distortional
buckling problem. This inefficiency was overcome by
Bradford & Trahair[21], who developed a beam or
line-type element that incorporated six nodal degrees
of buckling freedom, as shown in Fig. 4. This element

Fig. 3 Cross-bracing to prevent distortional buckling


Copyright ^ 2000 John Wiley & Sons, Ltd.

171

Fig. 4 Representation of beam-finite element used by Bradford


and Trahair[21] to analyse distortional buckling

was extended subsequently to handle flanges of


arbitrary shape by Bradford & Trahair[22], and this
extension was applied to analyse continuous
composite bridge girders by Johnson & Bradford[23].
This study questioned the provisions of the British
Bridge Code BS5400, and suggested that cross bracing
in many cases could be eliminated. A design proposal
was suggested. This beam element was later used in
an elastic study by Bradford and Gao[24], and this
study inspired Williams and Jemah[25] to perform
a similar finite element study. Svensson[26] considered
the compression flange as an elastically supported
strut, and obtained numerical solutions for the most
common cases of moment gradient and boundary
conditions.

EXPERIMENTAL RESULTS
Experimental studies of elastic local and distortional
buckling of composite beams have been very rare.
Elastic local buckling experiments on plates with
numerous boundary and loading conditions have
been reported extensively in the literature[27], and have
formed a means of validating a number of theoretical
studies. However, elastic distortional buckling
experiments on I-section beams have received very
little treatment[28] in comparison with other section
profiles[29].

Inelastic behaviour of class 1 and


2 members
THEORETICAL RESULTS
Inelastic local buckling
Inelastic local buckling has been tackled using two
approaches. Firstly, some researchers prefer a yieldline model to predict the local buckling load. On
the other hand, other researchers have tended to treat
inelastic local buckling as an extension of elastic
local buckling, in which smooth buckling
displacements are adopted instead of the yield linetype kinks of the mechanism approach. Numerical
modelling using the extended elastic buckling model
is extremely inefficient computationally, as the
stiffness and stability matrices in a stiffness analysis
constantly change as the buckling load factor is
increased, and so the solution must be incremental
and iterative.
Prog. Struct. Engng Mater. 2000; 2:169}178

172

COMPOSITE CONSTRUCTION

The first significant research into local buckling in


composite beams appears to date back to the work of
Climenhaga and Johnson[14], who considered both
elastic and inelastic local buckling when the slab
restrained the top flange of the steel I-section. The
underlying assumption in this yield line analysis of
local buckling is that longitudinal line junctions
between intersecting plates remain straight, and this
occurs at short wavelengths. Of course, these line
junctions move sideways with longer length
distortional buckling, which was only considered
subsequent to the work in ref[14]. Apart from the
restraint of the top flange, the main parameters that
affect local buckling are the width-to-thickness ratio of
the web and free flange outstand, and the significance
of these was considered in the Climenhaga and
Johnson[14] study.
Dawe & Kulak[30] developed a pseudo-strip method
for handling inelastic local buckling, and compared
the solutions with tests on North American WF
sections in the inelastic range of structural response.
Independently, Bradford[31] developed an inelastic
finite strip method of analysis based on the flow
theory of plasticity. The solution, although
incremental and iterative, could be obtained relatively
rapidly on computers of the time, and the method was
used by Bradford & Johnson[32] to study inelastic local
buckling of composite beams. The solution was
approximate, since the limiting assumptions of the
harmonic-based semi-analytical method were
adopted. Bradfords method[31] was also used to
calibrate the width-to-thickness limits in the British
BS5950 steel standard.

AS4100-1998 steel standard) are inaccurate, albeit


conservative, for composite beams.
In 1987, Weston et al[36] published a study of
inelastic distortional buckling in composite beams in
which they used a non-bifurcative finite element
method for plastic analysis developed elsewhere.
Their results are considered to be accurate, but the
computational times would have been considerable.
The design rule proposed by Weston et al agreed
basically with that of Bradford[33]. Dekker & Kemp[37]
have shown using a spring model how the elastic
warping coefficient, second moment of area in lateral
buckling and the Saint Venant torsion constant should
be adapted to allow for distortional buckling and
inelastic behaviour. The loss in moment resistance
caused by cross-sectional distortion was confirmed as
being small. Gioncu & Peteu[38] have recently reported
on a major extension of the Climenhaga and Johnson
yield line approach, which includes both local and
distortional buckling and the assessment of available
rotation capacity. This publication provides an
extensive literature survey.

Inelastic distortional buckling


For plastic design, it is important to establish
that attainment of a plastic mechanism will precede
inelastic distortional buckling. Studies in the 1980s
by Bradford[33] established a relationship between
the plastic moment at which inelastic buckling will
occur and the elastic buckling moment, and this work
forms the basis of the lateral buckling strength
curves in a number of national steel standards.
However, it appears that the relationship between
the full plastic moment and the elastic buckling
moment at which lateral buckling occurs is different
from that at which distortional buckling occurs[33].
Bradford[34] extended the beam element concept
for distortional buckling developed by Bradford
& Trahair[21] into the inelastic range of structural
response by using the flow theory of plasticity
with Lays shear modulus[35]. The inelastic finite
element buckling model was applied to composite
bridge girders by Bradford & Johnson[32], and a
design rule was recommended that differed somewhat
from these researchers elastic study[23]. A study
of the inelastic distortional buckling of composite
beams by Bradford[33] indicated that the strength
rules in current codes (particularly the Australian
Copyright ^ 2000 John Wiley & Sons, Ltd.

LIMIT STATES CRITERION OF DUCTILITY


Once continuous construction is adopted, the
following reasons exist, particularly in building
structures, for adopting plastic analysis (Class 1 in
Table 1) or at least enabling considerable moment
redistribution to occur prior to strain softening, based
on a rigid plastic stress block[1] analysis of resistance.
1. Elastic distributions of moment generally give
higher moments at internal supports than
midspan regions, whereas resistances are greater
in positive midspan regions.
2. The shape factor for composite beams is high
(1.25}1.35) and this benefit is only utilized if
stress block analysis is used to determine the
resistance.
3. Elastic distributions of moment are unreliable
due to cracking in negative moment regions and
the influence of partial shear connection[1].
Predicting the inelastic behaviour of Class 1 and
2 members therefore involves not only assessing the
ultimate resistance, but also the ductility or rotation
capacity. Johnson & Hope-Gill[39] have demonstrated
the importance of maintaining the moment resistance
at internal supports above the design resistance while
moment is redistributed to the midspan region (Class
2) or while the plastic collapse mechanism is achieved
(Class 1). The required rotation capacity in negative
moment regions to achieve a plastic collapse
mechanism or defined levels of moment redistribution
has been investigated widely[38}45]. Kemp and
Dekker[40] have expressed this requirement in limit
states format as follows:



ra"

ha
hab#hacon#hacr
"
"rab#racon#racr
he
he

(1)

Prog. Struct. Engng Mater. 2000; 2:169}178

BUCKLING IN CONTINUOUS COMPOSITE BEAMS


where he"0.5MpLi / EI in Figs 5 and 6. If the same
proportion of the web depth is in compression, it is
common practice to assume that local flange and web
buckling of a particular steel section in a composite
beam will achieve an inelastic rotation hab at least as
large as in a plain steel beam. Kemp and Dekker[40]
have analysed the components of the inelastic rotation
in eq. (1) and identified in the following expressions
for the rotation capacity of composite beams:
rab"ras

  
Mps
Mp

EI
EIs

EIcr
racr" !1.0
EI

(2)

(3)

where ras , Mps and EIs refer to the plain steel section,
and EIcr refers to the cracked section.
A conceptual analysis[40] of the influence of these
factors on the rotation capacity limit state of eq. (1) has
indicated that no significant difference is required in
local buckling criteria between composite and steel
beams. Inelastic lateral distortional buckling is less
likely in fully loaded composite beams using hotrolled steel sections due to the restraint from the slab
and shorter buckling length, although this is not the
case for lightly loaded spans in pattern loading.
Large areas of longitudinal reinforcement in the slab
relative to the area of the steel section contribute to
a significant reduction in the available rotation
capacity[40] for the following reasons:
1. Increased depth of the web in compression
reduces the web buckling and the level of strain
in the reinforcement;

173

2. Reduced curvature (equal to the critical strain in


the compression flange/height of the plastic
neutral axis above the compression flange)
results directly in reduced inelastic rotation at
maximum moment.

EXPERIMENTAL RESULTS
The experimental results obtained by a number of
researchers are given in the Table 2.
In addition, Li et al[51] have reported the behaviour
of two, two-span, two-storey composite frames with
flush end plate connections. Local flange and web
buckling was observed in both tests, and a local
crushing failure of the concrete beyond the outside
column in one test led to a sharp redistribution of
moments. It was confirmed in these tests that the
quasi-plastic analysis gave the closest prediction of the
test results.
Double cantilever tests, as illustrated in Fig. 6, have
been conducted to represent the regions on either side
of an internal support between the section of
maximum negative moment and the adjacent point of
inflection. These tests on Class 1 and 2 beams are
summarized in Table 3.
The following conclusions can be drawn from the
tests reported in Table 3.
1. The most important modes of failure are local
flange, local web and lateral-distortional
buckling: if these are controlled, large rotation
capacities can be achieved beyond the plastic
moment of resistance.
2. Local flange and web buckling cause less
significant strain softening behaviour, and

Fig. 5 Moment-rotation curves: (a) long-span specimens (Fig. 6c); (b) short-span specimens (Fig. 6b)
Copyright ^ 2000 John Wiley & Sons, Ltd.

Prog. Struct. Engng Mater. 2000; 2:169}178

174

COMPOSITE CONSTRUCTION

Fig. 6 Test representation: (a) two-span beam; (b) bending moment: live load on both spans; (c) bending moment: live load on left span
only; (d) double-cantilever specimen

Table 2 Inelastic experiments on continuous composite beams


Order and
No. of
No. of
modes of
M and Mp
Authors and reference
tests
spans
Loading
failure
achieved?
............................................................................................................................................................................................................................................................
Ansourian [46]

1
2
CPL, 1 span
e
No
1
2
CPL, 1 span
a#b, (c), e
No
2
2
CPL, 2 spans
e, a
Yes
2
2
CPL, 2 spans
a#b, e
Yes
............................................................................................................................................................................................................................................................

Hope-Gill & Johnson [47]

2
3
CPL, 1 span
e, a
Yes
1
3
PL, 3 spans
a
No
............................................................................................................................................................................................................................................................

Hamada and Longworth [48]

1
2
CPL, 2 spans
e, (a)
Yes
2
2
CPL, 2 spans
a
Yes
............................................................................................................................................................................................................................................................

Johnson, et al [49]

1
2
DL, 2 spans
d
2
2
DL, 2 spans
f
3
2
DL, 2 spans
a#b, c, e
Yes
............................................................................................................................................................................................................................................................

Barnard & Johnson [50]

1
3
PL, 1 span
f
No
1
3
PL, 1 span
f, c
No
2
3
PL, 1 span
f, c
Yes
............................................................................................................................................................................................................................................................

Notes: CPL"central point load, PL"point load, DL"distributed load. In negative moment region: a"local flange buckle, b"local web buckle, c"lateral buckle,
d"vertical shear failure. In positive moment region: e"concrete crushing, f"other slab failure

Copyright ^ 2000 John Wiley & Sons, Ltd.

Prog. Struct. Engng Mater. 2000; 2:169}178

BUCKLING IN CONTINUOUS COMPOSITE BEAMS

175

Table 3 Results of double cantilever tests


Order and modes of failure
M achieved?
Authors and reference
No. of tests
(see Table 2)
............................................................................................................................................................................................................................................................
Dekker, et al [52]

1
a, b, c (short span)
Yes
1
c, b (long span)
No
............................................................................................................................................................................................................................................................

Johnson & Fan [53]

1
c (T-beam)
No
2
a#b, c (U-beams)
Yes
............................................................................................................................................................................................................................................................

Hamada & Longworth [20]

5
c (cover plate)
Yes
17
a#b
Yes
............................................................................................................................................................................................................................................................

Climenhaga & Johnson [14]

2
a, b, c
Yes
6
a, b
Yes
6
b, c
No
............................................................................................................................................................................................................................................................

Johnson et al [49]
6
b
Yes
............................................................................................................................................................................................................................................................

3.

4.

5.

6.

therefore larger rotation capacities, than


interactive lateral-distortional and local buckling
in which the local buckle behaves as a partial
hinge in the formation of longer wavelength
lateral-distortional buckling.
It is essential for the longitudinal reinforcement
in the slab to possess adequate ductility and weld
mesh should therefore not be used.
Large areas of longitudinal reinforcement
relative to the area of the steel section produce
lower rotation capacities due to the larger web
depth in compression and the reduced curvature
at which the critical strain occurs in the flange.
Careful detailing of the reinforcement is required
to avoid secondary failures associated with
horizontal and vertical shear.
Longitudinal web stiffeners can improve local
buckling behaviour.

The expression for the available rotation capacity


ra in eq. (1) provides for inclusion of the contribution to
inelastic rotation by the end connection at internal
supports. Increasingly, semi-rigid connections have
been investigated with a moment resistance similar to
or less than that of the adjacent composite beam in
negative bending and with a significant inelastic
rotation capacity.
A review of these composite connections has
recently been provided by Leon[54] in which a range of
appropriate details are illustrated.
If the end connection does not provide for the
transfer of tension force in the upper flange of the steel
section, the ultimate moment is obtained from the
couple of the resultant reinforcement force in tension
and the bottom flange and adjacent web of the steel
section in compression. The efficiency of this approach
in avoiding local and distortional buckling and
achieving excellent ductility has been demonstrated
Copyright ^ 2000 John Wiley & Sons, Ltd.

by Kemp et al[55]. The results are illustrated in Fig. 5 in


which the behaviour of short- and long-span doublecantilever specimens with rigid end connections SR
and LR are compared to similar specimens with
semi-rigid end plate connections of limited depth SSR
and LSR. Specimens SSR and LSR with simple, semirigid end connections, no continuity of the upper
flange force and less longitudinal reinforcement
exhibited considerably enhanced ductility as well as
moment resistances equal to or greater than the
specimens with full-depth connections.

DESIGN
Johnson & Fan[53] compared the observed distortional
buckling capacities in two Class 2 U-frame tests in
negative bending with the theoretical approaches
of Bradford & Johnson[32], Weston et al[36] and
the Eurocode. They established that all four methods
underestimate the average moment resistance
in the tests by a factor of between 0.51 and 0.63.
They observed a complex interaction between local
and distortional buckling, with the transition
from symmetrical local buckling to S-shaped
distortional buckling taking place at or near the
maximum load.
Couchman & Lebet[56] have proposed a design
method for Class 1 and 2 members, in which
they compare the available percentage moment
redistribution (corresponding to the available
rotation capacity ra) with the required percentage
moment redistribution obtained from two generic
diagrams reflecting a range of spans, loading
and moment ratios. They assessed the models
proposed by Spangemacher and Sedlacek[57], Johnson
& Chen[58] and Kemp & Dekker[59] for assessing
available rotation capacity, and concluded that
the latter model provides the best results in terms
of the criteria of accuracy and conservatism. This
Prog. Struct. Engng Mater. 2000; 2:169}178

176

COMPOSITE CONSTRUCTION

model, simplified in ref[60], is based on an effective


slenderness ratio je as follows:

distortional restraint by the slab to combined lateral


and local buckling of the compression flange and
adjacent web, and the interactive nature of this
buckling;
the area of longitudinal slab reinforcement at
internal supports relative to the area of the steel
section;
the relative flexural rigidity of steel and composite
beams and cracked and uncracked regions of the
slab and its influence on ductility assessments;
differences in the buckling behaviour of elastic
(Class 4), inelastic (Class 2 and 3) and plastic
(Class 1) members, and the development of a global
model for all of these conditions.

ras"available rotation capacity of steel section eq. (2)



3 60  
"
2a je

(4)

where a is the ratio of web depth in compression to


total web depth, je the effective lateral slenderness
ratio Kf Kw Kd(Li/iZC )/ef , in which Kf"(b/tf )/10ef is the
flange factor for width b and thickness tf ,
Kw"(ad/tw)/35ew is the web factor for depth d
and thickness tw , Kd the distortional restraint factor
(1 for steel section, 0.71 for composite section iZC the
radius of gyration of portion of the elastic section of
the web and flange in compression Li the length
between sections of zero and maximum negative
moment and e"(250/fy where fy is the yield stress of
the flange or web.
Linked to studies of the Autostress method used in
North America, White and Barth[61] and Barth and
White[62] developed empirical equations for modelling
moment resistance and inelastic rotation of sections
with Class 1 or 2 flanges, Class 3 or 4 webs and closely
spaced lateral restraints. Their proposals compare
favourably with finite element and experimental
results, and appear to be more accurate than the
AASHTO and AISC LRFD specifications. They
confirmed that the yield moment can be achieved for
these sections and note that the strength is apparently
determined by buckling of the compression flange.
However, significant unconservative errors were
found in the code provisions when the depth of the
web in compression exceeds half the total depth, and
in sections with large ratios of overall depth to
compression flange width (reflecting weakness in
lateral buckling).
In a parallel study, Axhag[63] derived empirical
formulae for the inelastic rotation and slope of the
falling branch of the moment-rotation curve based on
test measurements of specimens with Class 1 or
2 flanges and Class 3 or 4 webs. Closely spaced lateral
restraints were again used to avoid premature failure.

Future directions
There are considerable benefits to be achieved by
providing continuity in composite beams, particularly
if the member possesses the necessary ductility to
develop the plastic moment resistances at both the
internal supports and midspan regions. Although the
negative moment regions of such a beam may be
designed successfully using current design codes and
the results of research reviewed in this paper, it is
apparent that significant differences exist and are not
always recognized between the behaviour of
composite and steel beams. Emphasis should be
placed on recognizing more thoroughly the
implications of these differences, particularly
Copyright ^ 2000 John Wiley & Sons, Ltd.

Conclusions
The paper has reviewed theoretical and experimental
studies related to local, distortional and interactive
local and distortional buckling in the negative moment
region of composite T-beams. Design models have also
been reviewed. The avoidance of buckling in order to
obtain sufficient rotation capacity has been
demonstrated, and the importance in attaining
sufficient rotation capacity in the negative moment
region has been highlighted.
Although buckling of plain steel beams in both the
elastic and inelastic ranges of response has been
studied extensively, and is now considered to be fairly
well understood and quantified, buckling in
composite beams still represents a grey area in
structural engineering research. Future directions in
research and its interpretation have been noted to
illustrate the need for even further research to obtain
a global method for modelling the behaviour that will
lead to accurate and uniform design rules.

References and recommended reading


Papers of particular interest have been marked:
* Special interest
** Exceptional interest
** [1] Oehlers DJ & Bradford MA. Composite steel and structural members:
fundamental behaviour. Oxford: Pergamon Press. 1995.
First book on composite structures to provide extensive treatment of local and
distortional buckling in continuous beams.
[2] Allen HG & Bulson PS. Background to buckling. London: McGraw-Hill.
1980.
[3] Trahair NS. Flexural-torsional buckling of structures. London: E&FN Spon.
1993.
[4] Trahair NS & Bradford MA. The behaviour and design of steel
structures to AS4100. 3rd edition London: E&FN Spon. 1998.
* [5] Bradford MA. Lateral-distortional buckling of steel I-section members.
Journal of Constructional Steel Research 1992: 23: 97}116.
Provides a comprehensive review of research into distortional buckling prior to
1993.
[6] American Institute of Steel Construction. Load and resistance factor
design specification for structural steel buildings. Chicago: AISC. 1986.

Prog. Struct. Engng Mater. 2000; 2:169}178

BUCKLING IN CONTINUOUS COMPOSITE BEAMS


[7] Comite European de Normalisation (CEN). Eurocode 3: design of
steel structures. Part 1.1: general rules for steel buildings. ENV-1993-1-1. Brussels:
CEN. 1992.
[8] Canadian Standards Association. Limit states design of steel structures
CAN/CSA S16.1-94. Rexdale, Ontario: Canadian Standards Association. 1994.
[9] Przemieniecki JS. Finite element analysis of local instability. American
Institute of Aeronautics and Astronautics Journal 1973: 11(1): 33d39.
[10] Cheung YK. Finite strip method in structural analysis. Oxford: Pergamon.
1976.
[11] Hancock GJ. Local, distortional and lateral buckling of I-beams. Journal of
the Structural Division (ASCE) 1978: 104(ST11): 1787d1798.
[12] Plank RJ & Wittrick WH. Buckling under combined loading of thin,
flat-walled structures by a complex finite strip method. International Journal for
Numerical Methods in Engineering 1974: 8: 323d339.
[13] Azhari M & Bradford MA. Elastic local buckling of composite tee-beams
with longitudinal stiffeners. Canadian Journal of Civil Engineering 1993: 20(6):
923d930.
[14] Climenhaga JJ & Johnson RP. Local buckling in continuous composite
beams. The Structural Engineer 1972: 50: 367d374.
[15] Lau SCW & Hancock GJ. Buckling under combined loading of thin,
flat-walled structures by a spline finite strip method. Thin-Walled Structures 1986:
4(4): 269d294.
[16] Fan SC & Cheung YK. Analysis of shallow shells by spline finite strip
method. Engineering Structures 1983: 5: 225d262.
[17] Azhari M, Hoshdar & Bradford MA. On the use of bubble functions in
local buckling of plate structures by the spline finite strip method. International Journal
for Numerical Methods in Engineering 2000: 48(4): 583d593.
[18] Bleich F. Buckling strength of metal structures. New York: McGraw-Hill.
1952.
[19] Goldberg JE, Bogdanoff JL & Glauz WD. Lateral and torsional buckling
of thin-walled beams. Publications, IABSE 1964: 24: 92d100.
** [20] Hamada S & Longworth J. Buckling of composite beams in negative
bending. Journal of the Structural Division (ASCE) 1974: 100(ST11): 2205d2219.
Appears to be the first finite element study of distortional buckling in composite
beams.
[21] Bradford MA & Trahair NS. Distortional buckling of I-beams. Journal of
the Structural Division (ASCE) 1981: 107(ST2): 355d370.
[22] Bradford MA & Trahair NS. Distortional buckling of thin-web beamcolumns. Engineering Structures 1982: 4(1): 2d10.
* [23] Johnson RP & Bradford MA. Distortional lateral buckling of continuous
composite bridge girders. In: Morris, LJ (ed) International Conference on Stability and
Plastic Collapse of Steel Structures, Granada. 1983: 569d580.
Presents the first design rule for overall buckling based on a rational buckling
analysis.
[24] Bradford MA & Gao Z. Distortional buckling solutions for continuous
composite beams. Journal of Structural Engineering (ASCE) 1992: 118(1): 73d89.
[25] Williams FW & Jemah AK. Buckling curves for elastically supported
columns with varying axial force, to predict lateral buckling of beams. Journal of
Constructional Steel Research 1987: 7(2): 133d147.
[26] Svensson SE. Lateral buckling of beams analysed as elastically supported
columns subject to varying axial force. Journal of Constructional Steel Research 1985:
5:179d193.
[27] Donald IB (ed). Thin-walled structures: developments in theory and
practice. London: Elsevier Applied Science. 1990.
[28] Woods RF & Watson SDC. The effects of beam support and web
distortion on the flexural-torsional buckling of I-beams. Honours Thesis. University of
Sydney: School of Civil Engineering. 1977.
[29] Zhao X-L, Hancock GJ, Trahair NS & Pi Y-L. Lateral buckling of
cold-formed RHS beams. In: Kitipornchai S, Hancock GJ & Bradford MA (eds)
International Conference on Structural Stability and Design, Balkema. 1995: 55d60.
[30] Dawe JL & Kulak GL. Plate instability of W-shapes. Journal of Structural
Engineering (ASCE). 1984: 110(6): 1278d1291.
[31] Bradford MA. Local buckling analysis of composite beams. Civil
Engineering Transactions, Institution of Engineers, Australia. 1986: CE28(4):
312d317.
** [32] Bradford MA & Johnson RP. Inelastic buckling of composite bridge
girders near internal supports. Proceedings of the Institution of Civil Engineers,
London, Part 2. 1987: 83: 143d159.
This paper presented design rules for both local and distortional buckling based on
inelastic finite element and finite strip analyses.

Copyright ^ 2000 John Wiley & Sons, Ltd.

177

[33] Bradford MA. Buckling strength of partially restrained I-beams. Journal of


Structural Engineering (ASCE) 1989: 115(5): 1272d1276.
[34] Bradford MA. Inelastic distortional buckling of I-beams. Computers and
Structures 1986: 24(6): 922d933.
[35] Lay MG. Flange local buckling in wide flange shapes. Journal of the
Structural Division (ASCE) 1965: 91(ST6): 95d116.
* [36] Weston G, Nethercot DA & Crisfield MA. Lateral buckling in
continuous composite bridge girders. The Structural Engineer 1991: 69(5): 79d87.
Proposed a design rule for distortional buckling based on sophisticated nonlinear
finite element modelling.
[37] Dekker NW & Kemp AR. A simplified distortional buckling model for
doubly symmetrical I-sections. Canadian Journal of Civil Engineering. 1998: 25: 1d10.
[38] Gioncu V & Peteu D. Available rotation capacity of wide-flange beams and
beam-columns. Journal of Constructional Steel Research 1997: 43(1-3): 161d218 and
219d244.
[39] Johnson RP & Hope-Gill MC. Applicability of simple plastic theory to
continuous composite beams. Proceedings of the Institution of Civil Engineers,
London Part 2 1976: 61(3): 127d143.
* [40] Kemp AR & Dekker NW. Available rotation capacity in steel and
composite beams. The Structural Engineer 1991: 69(5): 88d97.
Quantified components that contribute separately to the rotation capacity in
hogging bending.
[41] Nethercot DA, Li TQ & Choo BS. Required rotations and moment
redistribution for composite frames and continuous beams. Journal of Constructional
Steel Research 1995: 35(2): 121d164.
[42] Aribert J-M & Ragneau E. Theoretical investigation of moment
redistribution in composite beams of different classes. Proceedings of the Conference
on Composite Construction in Steel and Concrete III. Engineering Foundation. 1997:
392d405.
[43] Li TQ, Choo BS & Nethercot DA. Determination of rotation capacity
requirements for steel and composite beams. Journal of Constructional Steel
Research 1995: 32(3): 303d332.
* [44] Kemp AR. Quantifying ductility in composite beams. Proceedings of the
Conference on Composite Construction in Steel and Concrete. Engineering
Foundation. 1988: 107d121.
Recent summary of rotation capacities.
* [45] Rotter JM & Ansourian P. Cross-sectional behaviour and ductility in
composite beams. Proceedings of the Institution of Civil Engineers, London. Part 2
1979: 67(3): 127d143.
This paper provided insight and quantification of ductility limits in the positive
moment region where the strength is governed by crushing of the concrete.
** [46] Ansourian P. Experiments on continuous composite beams. Proceedings
of the Institution of Civil Engineers, London, Part 2. 1981: 71(12): 25d51.
Benchmark experimental data on achievable rotation in continuous composite
beams that includes buckling considerations.
[47] Hope-Gill MC & Johnson RP. Tests on three three-span continuous
composite beams. Proceedings of the Institution of Civil Engineers, London, Part 2.
1976: 61(6): 367d381.
[48] Hamada S & Longworth J. Ultimate strengths of continuous composite
beams. Journal of the Structural Division (ASCE) 1976: 102(ST7): 1463d1478.
[49] Johnson RP, van Dalen K & Kemp AR. Ultimate strength of continuous
composite beams. Proceedings of Conference on Structural Steelwork. British
Constructional Steelwork Association: London 1967: 27d36.
* [50] Barnard PR & Johnson RP. Plastic behaviour of continuous composite
beams. Proceedings of the Institution of Civil Engineers, London 1965: 32: 180d197.
This is a much-cited paper on the early research on strength of continuous
composite beams
[51] Li TQ, Choo BS & Nethercot DA. The experimental behaviour of
a full-scale, semi-rigidly connected composite frame. Journal of Constructional Steel
Research 1996: 39(3): 167d220.
[52] Dekker NW, Kemp AR & Trinchero P. Factors influencing the strength
of continuous composite beams in negative bending. Journal of Constructional Steel
Research 1995: 34: 161d185.
[53] Johnson RP & Fan CKR. Distortional lateral buckling of continuous
composite beams. Proceedings of the Institution of Civil Engineers, London, Part
2 1991: 91(3): 131d161.

Prog. Struct. Engng Mater. 2000; 2:169}178

178
** [54] Leon RT. Composite connections. Progress in Structural Engineering and
Materials 1998: 1(2): 159d169.
State-of-the-art paper where rotation considerations are applied at beam-tocolumn connections and not simply over a rigid internal support.
[55] Kemp AR, Trinchero P & Dekker NW. Ductility effects of end details
in composite beams. Journal of Constructional Steelwork 1995: 34.
[56] Couchman G & Lebet J-P. A new design method for continuous
composite beams. Structural Engineering International 1996: 6(2): 96d101.
[57] Spangemacher R & Sedlacek G. Zum nachweis ausreichender
rationsfahigkeit von fliessgelinken bei der anwendung des fliessgelenkverfahrens.
Stahlbau 1992: 61: 329d339.
[58] Johnson RP & Chen S. Local buckling and moment redistribution in Class
2 composite beams. Structural Engineering International 1991: 4: 27d34.
** [59] Kemp AR & Dekker N. Rotation capacity in steel and concrete beams.
The Structural Engineer 1991: 69(5): 88d97.

COMPOSITE CONSTRUCTION
Early work on identifying the importance of rotation capacity in limit states
strength design that formed the basis for much subsequent work on composite
beams.
[60] Kemp AR. Inelastic local and lateral buckling in design codes. Journal of
Structural Engineering (ASCE) 1996: 122(4): 374d382.
* [61] White DW & Barth KE. Strength and ductility of compact-flange I-girders
in negative bending. Journal of Constructional Steel Research 1998: 43(3): 241d280.
Very recent research work on rotation capacity that can be applied to composite
beams.
[62] Barth KE & White DW. Finite element evaluation of pier momentrotation characteristics in continuous-span steel I girders. Engineering Structures
1998: 20(8): 761d778.
[63] Axhag F. Plastic design of composite bridges allowing for local buckling.
Technical Report . Sweden: Lulea University of Technology. 1995: 09T.

MA Bradford PhD DSc


School of Civil and Environmental
Engineering,
University of New South Wales,
Sydney, NSW 2052,
Australia
AR Kemp PhD
Department of Civil Engineering,
University of Witwatersrand,
WITS, Johannesburg 2050, South Africa

Copyright ^ 2000 John Wiley & Sons, Ltd.

Prog. Struct. Engng Mater. 2000; 2:169}178

You might also like