You are on page 1of 11

Journal of Constructional Steel Research 63 (2007) 13171327

www.elsevier.com/locate/jcsr

Assessment of effective slab widths in composite beams


J.M. Castro, A.Y. Elghazouli , B.A. Izzuddin
Department of Civil and Environmental Engineering, Imperial College, London, UK
Received 4 October 2006; accepted 27 November 2006

Abstract
This paper deals with the behaviour of composite beams with particular focus on the effective slab width, which is required for simplified
structural analysis and design. Current design codes propose values for the effective width which are mostly a function of the beam span
ignoring in this way the influence of other important parameters. Several 3D numerical simulations are conducted in this paper in order to
illustrate these parameters and accordingly a new methodology is suggested for evaluating the effective width. The proposed approach is easier to
apply in comparison with other existing methods based on stress integration, and provides effective width values which result in a more reliable
representation of the actual beam state when simplified analysis is carried out. The application of the new method indicates that the effective width
is mostly related to the actual slab width and, in many cases, the values obtained can significantly differ from those proposed in design codes.
Validation of the new approach is carried out through comparison of simplified 2D models with the results obtained from a recent experimental
investigation as well as from more complex 3D numerical simulations.
c 2006 Elsevier Ltd. All rights reserved.

Keywords: Effective width; Shear lag; Composite beams

1. Introduction
The enhanced stiffness, strength and ductility of composite
steelconcrete beams in comparison with RC and steel
counterparts, which result from the synergy between the two
materials, have been recognised for many years [17,30]. The
consideration of this type of member in the design process is
treated in a similar manner to that of steel and RC members,
i.e. through the application of traditional T-section analysis and
employment of simplified 2D models for the structural analysis.
This simplification however involves a number of assumptions,
most notably regarding the definition of the portion of slab
mobilised, referred to as the effective width. When a composite
beam deforms, shear strains develop in the slab and cause a
shear lag effect. This consists of non-uniform distributions of
normal stresses across the slab width and the non-planarity of
the slab cross-section. The effective width allows consideration
of this effect and it is typically used both for the structural
analysis and design stages.
The effective widths prescribed by current design codes
were established many years ago and are based on research
Corresponding author.

E-mail address: a.elghazouli@imperial.ac.uk (A.Y. Elghazouli).


c 2006 Elsevier Ltd. All rights reserved.
0143-974X/$ - see front matter
doi:10.1016/j.jcsr.2006.11.018

that considered mainly elastic behaviour. In recent years,


several research studies (e.g. Amadio and Fragiacomo [9],
Chiewanichakorn et al. [19]) have been performed which
recognised that the effective width is not a constant parameter
and changes with the development of inelasticity on the
composite member. This work resulted in various suggested
modifications for assessing the effective width.
In this paper, the behaviour of several composite beams
is assessed through detailed 3D numerical simulations which
provide an insight into the key parameters influencing the
effective width. A new proposed methodology for evaluating
effective widths is then described. The approach is more
consistent with underlying behavioural principles and is
also easier to apply compared to other existing procedures
but, at present, is limited to composite beams with full
interaction. The method is applied in a parametric study to
examine the influence of the various parameters governing
the effective width and to illustrate that this parameter is
largely related to the actual slab width. Finally, simplified 2D
numerical simulations are conducted to predict the response
obtained from a recent experimental study on a simply
supported composite beam as well as from more complex
3D analyses.

1318

J.M. Castro et al. / Journal of Constructional Steel Research 63 (2007) 13171327

2. Previous studies on effective width


The study of shear lag effects in composite systems started
during the sixties when several researchers extended existing
work on steel plates [31] and applied the concepts to composite
beams. Adekola [2] used the analytical solutions derived by
Allen and Severn [7] and calculated effective widths for
simply supported beams considering the variation of geometric
parameters. Several years later, Ansourian [10] applied the
finite element method to perform elastic analysis of fixed
composite beams. From the results, it was concluded that
accurate slab stresses are obtained when the effective width is
taken as one quarter of the span. On the other hand, to achieve
adequate accuracy in obtaining steel beam stresses, the use of
an effective width equal to the actual slab width was suggested.
Heins and Fan [24] presented an analytical method for
predicting the ultimate load behaviour of simply supported
composite bridge decks. The method involved the application
of the finite difference technique to solve a set of coupled
partial differential equations and elasto-plastic behaviour was
accomplished by a step-by-step incremental load procedure.
With the developed method, effective widths of slabs at ultimate
load were evaluated.
Fahmy and Robinson [23] investigated ten composite
cantilever beams incorporating ribbed metal deck and
representative of positive moment beam-to-column connections
of an unbraced frame. Nonlinear material behaviour and
shear interaction were considered in the analysis. Effective
widths for strength and stiffness calculations were determined.
This investigation proposed a direct relationship between the
effective width and the length-to-width ratio and the column
width-to-slab width ratio. Elkelish and Robinson [22] in a
similar study examined the influence of the type of loading on
the effective width. Results from this work have also indicated
higher effective widths in the inelastic range compared to those
for the elastic stage.
Brosnan and Uang [11] performed numerical studies using
ANSYS of composite L-beams (edge beams) and concluded
that the effective widths of slab in these systems are different
from those proposed in the codes for internal composite
beams (T-beams). More recently, Amadio and Fragiacomo [9]
conducted a series of parametric studies of simply supported
and cantilever composite beams using ABAQUS. Both elastic
and nonlinear analyses were performed and different levels
of shear connection were considered. The results for elastic
behaviour showed that the connection deformability is a very
important parameter on the evaluation of the effective width
for stress analysis. Proposed values were given for crosssections along the beam span. On the other hand, the full width
was proposed for stiffness calculations. The results from the
nonlinear analyses showed the enlargement of the effective
widths obtained for the elastic stage even reaching the full slab
widths in some situations. These conclusions were validated
with results obtained from experimental tests performed on four
composite beams [8].
From a review of the literature on effective width, the
complexity and inconsistencies surrounding this issue become

apparent. Nevertheless, it is evident that the effective width


depends on a number of parameters such as the span, beam
spacing, boundary conditions, degree of shear connection, etc.
Clearly, the effective width is not constant along the span of
a composite beam and changes during the loading process
particularly at the onset of plasticity.
3. Existing definitions and provisions
There is no standard definition for effective width that can be
generally acceptable for all conditions. Two main approaches
for the evaluation of this parameter are often referred to in the
literature. One is related to the stress state of the slab and a
second to the stiffness of the composite beam.
The first definition is based on the stress distributions in the
slab, which is directly related to the shear lag phenomenon. In
this case, the effective width (beff ) is considered as the width
of slab that sustains a force equal to that in the actual slab,
assuming the longitudinal stresses (x ) to be constant across the
effective slab width and equal to the peak stress over the steel
beam centreline, as shown in Fig. 1. In mathematical terms, this
is expressed as follows:
beff =

1
[x ] y=0

+ b2

b2

x dy.

(1)

The level at which the stresses should be obtained is not clearly


established. It can be applied to the top surface stresses in the
slab or at the mid-plane surface. Fahmy and Robinson [23] and
Elkelish and Robinson [22] employed a modified version of the
original definition as the effective width was evaluated from the
ratio of the total force developed in the slab to the integration of
stresses in the centreline of the beam through the slab thickness,
such that
R + tslab
R + b2
2
beff =

tslab
2
t
+ slab
2
tslab
2

x dydz

b2

[x ] y=0 dz

(2)

Recently, Chiewanichakorn et al. [19] calculated effective


widths based on results from 3D finite element analysis. By
recognising that the stress pattern is not constant through the
slab thickness, the effective width was established in such a
way that the total force developing in the slab and its point of
application is the same in both the 3D model and in a simplified
T-section analysis.
The other approach, whereby the effective width is based
on stiffness calculations, involves the determination of the
deflection of a composite beam and then, from analytical
expressions, derivation of the equivalent second moment of
area that will cause the same deformation in the idealised
beam. From the equivalent second moment of area, an average
effective width is then estimated [3,11].
It is important to note that the effective widths derived from
the two approaches can be substantially different, as shown
by Brosnan and Uang [11]. Also, care must be taken when
dealing with an effective width based on stiffness in cases where
considerable shear connection deformation is expected.

J.M. Castro et al. / Journal of Constructional Steel Research 63 (2007) 13171327

Fig. 1. Stress-based effective width.

1319

when compared to their European counterparts. In AISC 36005, no guidance is provided for continuous beams. However,
in the commentary document of the code [6], a simplified
approach is suggested for evaluating stiffness of continuous
composite beams which considers a weighted average of
second moments of area in the positive and negative bending
moment regions of the beam.
The effective width recommendations available in most
code provisions were derived on the basis of gravity loading
conditions. Therefore, different values of the effective width
would be expected for composite beams under lateral loading
conditions such as those due to seismic actions [21,16].
Eurocode 8 [14], specifically addresses this scenario by
proposing different effective widths to those prescribed in
Eurocode 4. In addition, Eurocode 8 also distinguishes between
effective widths for use in analysis and for strength calculations.
The values proposed are mostly a function of the span length
but are also dependent on column dimensions. Due to the
significant differences between the effective widths proposed in
Eurocode 4 and those proposed in Eurocode 8, designers may
encounter some difficulties during the design process.
In addition to the effective width proposals provided for
composite beams in building structures, similar recommendations are available for composite bridges. A detailed summary
and comparison of these proposals can be found elsewhere [4,
18]. It is worth mentioning that most provisions propose the effective width as a function of the span length and limited by the
distance between adjacent beams. However, in the AASHTO
specifications [1], the effective width is also a function of the
slab thickness. Another important observation is that most recommendations for building and bridge structures do not differentiate between effective widths for use in analysis and for
strength evaluation.
4. Detailed numerical assessment

Fig. 2. Equivalent spans (L e ) proposed in Eurocode 4 [13].

In terms of the design guidance for effective width,


codes normally provide simplified provisions. To discuss this,
European, British and North American provisions are briefly
summarised below.
The effective width proposed in design codes is typically
a function of the beam span. For example, Eurocode 4 [13]
recommends that the total effective width (beff ) should be
determined as the summation of the effective widths evaluated
on each side of the beam web (bei ). The value of bei is
equal to L e /8 but should not exceed the geometric width bi
which corresponds to half of the distance between adjacent
beams, where L e is the equivalent span corresponding to the
approximate distance between points of zero bending moment.
For typical continuous composite beams in frames, Eurocode
4 stipulates the equivalent spans illustrated in Fig. 2. It is also
worth noting that Eurocode 4 allows the adoption of a constant
effective width over the whole of each span. These provisions
are similar to those prescribed in the British Standard 5950 [12].
With regard to North American proposals for effective
width, the most recent AISC provisions [5] are relatively simple

The behaviour of a simply supported composite beam under


positive bending moment is now investigated through a 3D
detailed model of a six metre composite beam analysed in
ADAPTIC [25]. The main aspects of the behaviour are pointed
out for both the elastic and inelastic stages. The beam consists
of a 120 mm thick concrete slab and a European IPE 300 steel
section. The total slab width is 2.5 m. Steel is assumed to have a
yield strength of 275 N/mm2 whereas the uniaxial compressive
and tensile strengths of concrete are 30 N/mm2 and 2 N/mm2 ,
respectively.
In terms of the numerical model, the steel section is
represented with 3D cubic elasto-plastic beam elements (cbp3)
[28] incorporating a fibre approach. On the other hand, the
concrete slab is modelled with a recently developed flat
shell element (csl4) [29] which has been validated against
experimental results [20]. This novel element is able to
efficiently reproduce the behaviour of geometrically orthotropic
slabs such as typical ribbed steel-decked composite floor
systems. Both beam and shell elements account for geometric
nonlinearities. Composite action is achieved through inclusion
of special link elements (lnks) [26] with rigid axial and bending

1320

J.M. Castro et al. / Journal of Constructional Steel Research 63 (2007) 13171327

Fig. 3. Simply supported composite beam.

properties which connect the centroids of both the steel section


and the slab. Note that full interaction is assumed, between
the steel beam and the slab, and that a layer of reinforcement
representing a ratio of 0.4% of the slab cross-section is
provided. A representation of the numerical model is shown in
Fig. 3.
With regard to the material models adopted, a typical
bilinear elasto-plastic model with strain hardening (stl1)
is employed for steel whereas for concrete a biaxial
model (con11), which accounts for the combined effects of
compressive nonlinearity and tensile crack opening and closure,
is adopted. In this model, compressive nonlinear response is
dealt with according to plasticity theory in which a biaxial
interaction surface is employed. Account is taken of the
hardening (before pre-crushing) and softening compressive
response. Tensile behaviour is considered through a smeared
crack approach assuming a fixed crack orientation. Softening
response in tension is incorporated to represent tension
stiffening effects. Further details of this model can be found
elsewhere [27,29].
In order to assess the parameters influencing the longitudinal
stress distribution in the slab, various loading conditions as
well as different slab widths and thicknesses are considered and
discussed below.

Fig. 4. Longitudinal stress distributions across slab width at different slab


surfaces.

4.1. Elastic behaviour


Focusing on the elastic range, the longitudinal stress
distribution across the slab is firstly examined. The beam is
loaded with a uniformly distributed pressure. In Fig. 4, the midspan stress distributions at top, middle and bottom surfaces
are depicted. It is clear from the figure that the stress pattern
is not constant across the slab thickness; hence difficulties
arise for the evaluation of effective widths based on stress
calculations. The nearly uniform stress distribution is also noted
for both middle and top surfaces. The shear lag effect is not
as visible as would be expected. The reason for this is related
to the low gradient of bending moment at the mid-span as
the beam is loaded with a uniformly distributed pressure. The
stress distribution at the bottom surface is also of interest as it
observes its tensile peak at the edges rather than at the centreline
of the steel beam. This can be attributed to the localised effects
arising from the presence of a rigid link.
The influence of slab width on longitudinal stress
distribution in the slab is now examined. This parameter is
directly related to the in-plane shear stiffness of the slab. The

Fig. 5. Influence of slab width (b) on stress distribution at mid-surface.

larger the slab width, the lower the shear stiffness. Two new
composite beams are analysed in which the slab widths are
taken equal to 1.5 m and 3.5 m, respectively. The mid-span
stress distributions at the mid-surface are plotted in Fig. 5. As
expected, the shear lag effect is more evident for larger slab
widths.
A second parameter which is directly related to the inplane shear stiffness of the slab is its thickness. For thicker
slabs, the shear stiffness is higher and therefore the presence
of shear lag is attenuated. This trend is illustrated by observing
the stress distributions plotted in Fig. 6. It is clear that when
the slab thickness is smaller, the variation of the longitudinal

J.M. Castro et al. / Journal of Constructional Steel Research 63 (2007) 13171327

Fig. 6. Influence of slab thickness (tslab ) on stress distribution at mid-surface.

1321

Fig. 8. Top surface longitudinal stress distributions across slab width at midspan for increasing levels of deformation.

Fig. 7. Influence of load type on stress distribution at mid-surface.

stress is relatively more evident. However, the influence of this


parameter is not as significant as that observed for the slab
width.
The last parameter investigated is the type of load. The
results presented above were obtained from a beam loaded
with uniform pressure. Two new cases are now analysed. In
the first one, a single point load is applied at the mid-span of
the composite beam whereas in the second case a line load
is applied at the mid-span of the beam, transversely to the
longitudinal direction. The stress distributions at mid-surface
are depicted in Fig. 7, which clearly demonstrate the influence
of the load type on the beam behaviour. When a single point
load is applied, the stress variation is very significant. Evidently,
if effective widths are derived on the basis of stress calculations,
the resulting values can be considerably different depending on
the load type.
4.2. Elasto-plastic behaviour
The inelastic behaviour of the composite beam is examined
in this section. In order to illustrate the change of stress pattern
in the slab for increasing levels of plasticity, a point load is
applied at mid-span to represent a case of significant shear lag.
In Fig. 8, the longitudinal stress distribution in the top surface

Fig. 9. Edge-to-centreline (edge /centre ) stress ratio.

of the slab is plotted for different levels of deformation ().


At 3.25 cm, the distribution corresponds to the point when the
stress at the centreline attains the peak. Similarly, for a vertical
deformation of around 6 cm, the peak stress is reached at the
slab edge.
Inspection of Fig. 8 reveals that compressive longitudinal
stresses in the slab tend to become uniform for increasing
demand levels. When the peak stress is reached at the
centreline, these fibres enter a softening regime (incorporated in
the concrete constitutive model) and stress redistribution occurs
in the slab. At a certain stage, the stresses at the edge are higher
compared to those at the centreline. The ratio between edge and
centreline stresses (edge /centre ) is depicted in Fig. 9.
Examination of Fig. 9 provides useful information regarding
the nonlinear behaviour of the composite beam. It is of interest
to point out that the stress ratio remains constant while the
beam response is fully elastic. However, a reduction in this ratio
is observed from a vertical deformation of about 1 cm. This
corresponds to the occurrence of first yield in the bottom flange
of the steel beam. At around 2.5 cm, the ratio between edge
and centreline stresses starts increasing due to the initiation of
nonlinear concrete response. After 3.25 cm, the peak stress is
attained at the centreline and the redistribution of stresses in the

1322

J.M. Castro et al. / Journal of Constructional Steel Research 63 (2007) 13171327

developing at the mid-span region of the composite beam. As


shown in Fig. 11, significant compressive stresses develop in
the transverse direction at mid-span around the beam centreline.
The numerical observations provided in this section
demonstrate the complexity of the stress state developing in
the slab of a composite beam. It is clear that shear lag effects
occur in the elastic range but, due to material nonlinearity, stress
redistribution develops in the slab. The concept of a single
effective width for use in both analysis and strength calculations
is therefore inaccurate and its use is not supported by either
numerical or experimental observation.
5. Proposed approach
5.1. Description
Fig. 10. Variation of top surface concrete stresses at mid-span with vertical
deformation.

Fig. 11. Top surface transverse stress distribution along the centreline of the
composite beam.

slab starts developing. This is reflected in a steep increase of


the stress ratio. For a vertical deformation of about 6 cm, the
peak stress is reached at the slab edge and another change is
observed in the stress ratio curve.
With regard to stress magnitudes, the high values recorded
for the concrete peak stresses are noteworthy. As indicated in
Fig. 8, concrete stresses reach values well above 30 N/mm2 ,
the uniaxial strength considered for concrete. In Fig. 10,
top surface concrete stresses at mid-span are plotted against
vertical deformation. It becomes clear from the figure that
concrete stresses at both centreline and slab edge reach values
markedly higher compared to the uniaxial strength considered.
This behaviour is intimately related to the confining effects

A new approach is now presented to assess the effective


width of composite beams in the linear elastic range. The
method can be applied to beams under positive and negative
bending moment [15] but is currently limited to full-interaction
cases.
Rather than evaluating the effective width based on the
complex stress patterns developing in the slab, the approach
consists of finding equivalent second moment of areas from
results obtained using a 3D finite element model. This model
is similar to those employed before in the investigation of
composite beam behaviour.
By analysing the 3D model subject to a set of loading and
boundary conditions, the strain profiles () at the steel sections
and the corresponding curvatures () can be readily obtained,
as shown in Fig. 12.
Since the bending moment applied is also a known
parameter, the equivalent second moment of area (Ieq ) of the
cross-section can be easily derived by applying the expression
=

M
M
Ieq =
.
E Ieq
E

(3)

The effective width of the cross-section under consideration can


then be readily derived.
This approach has several advantages when compared
to those based on stress calculations. Firstly, it avoids the
complexity of stress integration. It also provides more accurate
representation of the behaviour as it maintains the location
of the elastic neutral axis for a given bending moment. This
is not however the case for stress-based methods in which
the effective width is evaluated by controlling the level of
internal forces in the slab whilst not ensuring that these are then
mobilised under the same bending moment in the simplified

Fig. 12. Curvature evaluation in the proposed approach.

J.M. Castro et al. / Journal of Constructional Steel Research 63 (2007) 13171327

1323

analysis. An additional advantage of the new approach is


that it can be easily applied for obtaining effective width
distributions along the length of composite beams. It should
also be noted that this approach can be used directly for both
simply supported and continuous composite beams.
The suggested method may seem to imply that a Mbeff
relationship exists for a given composite beam. This is not
necessarily the case as curvatures in the 3D model are sensitive
to bending moment gradients. This observation is further
emphasised by the different stress patterns obtained for various
loading conditions (Fig. 7).
As noted before, despite its advantages, this new approach
does not currently incorporate beams with partial interaction.
This extension could be an interesting topic for future research.
It is also worth mentioning that the influence of the type of
shear connection is not considered in this approach. However,
this aspect is deemed to be more relevant in beams with
partial interaction and is obviously of extreme importance when
ductility and capacity issues are under investigation.
5.2. Illustrative examples
The applicability of the proposed approach is now examined
by applying it to some of the beams already investigated in
the detailed numerical study presented before in this paper.
Note that, unless stated otherwise, the beams are loaded with a
uniformly distributed pressure. The influence of four geometric
parameters is assessed by plotting the effective widths along the
beam.
The first parameter examined is the slab width (b). In Fig. 13
effective widths are plotted for three composite beams with slab
widths of 1.5 m, 2.5 m and 3.5 m.
As expected, a higher percentage of slab is mobilised for
the 1.5 m case. This is consistent with the low shear lag
present in this beam (Fig. 5) and confirms the higher in-plane
shear stiffness associated with this case. On the other hand,
for larger slab widths, the percentage of width mobilised is
smaller; however an increase of effective widths is observed.
The curves shown in Fig. 13(b) contradict the conventional
recommendation of considering the effective width uniquely as
a function of the span length. From the results, and focusing on
the effective width at mid-span rather than its variation over the
span, it is clear that the effective width is strongly dependent
on the actual slab width. It should also be noted that, in this
example, the value of L/4 proposed in most codes is only valid
for the composite beam with 1.5 m of slab width.
The second parameter studied is the slab thickness.
Two new variant beams are considered with 80 mm and
150 mm slabs respectively. The effective width distributions
are depicted in Fig. 14. The results obtained show that thicker
slabs are associated with higher in-plane shear stiffness and
consequently, with larger effective widths. Like the previous
parameter, the effective width is more sensitive to variations in
slab width rather than span length.
The influence of span length on the effective width
distribution is also examined. An additional composite beam of
10 m length is analysed and the results are presented in Fig. 15.

Fig. 13. Influence of slab width (b) on effective width distribution.

Visual inspection of the figure indicates the larger slab width


mobilised in the longer beam. The agreement with the estimate
of Eurocode 4 for this beam is noteworthy, which is not the
case for the 6 m beam. However, a better correlation between
the effective width and the full slab width is again confirmed.
The proposed method for effective width evaluation is now
applied to a cantilever beam loaded upwards at the tip. This
system is intended to represent a composite beam under lateral
loading spanning to an external joint. The slab edge is assumed
to be aligned with the column flange. Therefore, only an
assumed length (bc ) of 0.25 m is restraining the slab in-plane.
The objective is to investigate whether the proposed approach is
able to represent a reduced effective width in the contact region
with the column. The effective widths obtained are represented
in Fig. 16. The figure clearly shows a smaller effective width
in the contact region. The value obtained is around 2bc ,
i.e. 0.50 m. This corresponds to double the restrained length
which appears to be realistic.
The analyses described above illustrate the applicability of
the proposed method for effective width evaluation. Due to
its simplicity and efficiency, it can be used for performing
extensive parametric studies to quantitatively investigate the
parameters affecting the effective width in the elastic range.
It is important to note that the elastic effective widths
obtained at the mid-span of the beam are in the range of

1324

J.M. Castro et al. / Journal of Constructional Steel Research 63 (2007) 13171327

Fig. 14. Influence of slab thickness (tslab ) on effective width distribution.

Fig. 15. Influence of beam span (L) on effective width distribution.

80%100% of the actual slab width. This conclusion, coupled


with previous observations of stress redistribution in the slab in
the inelastic range, suggests that consideration of full widths
for 2D frame analysis and strength evaluation appears to be
an alternative approach to current code proposals. To assess
the validity of this suggestion, two comparative studies are
undertaken and described in the following section.
6. Comparative studies
In order to assess the validity of adopting effective widths in
2D analysis larger than that prescribed in codes, a comparison is
performed hereafter between the results obtained from a recent
experimental study conducted by Amadio et al. [8] on a simply
supported composite beam (B4 specimen) and those obtained
from simplified 2D beam analyses. It should be noted that the
purpose of this section is not to check the accuracy of the
proposed approach described above but to investigate the global
and local response of composite beams, particularly when large
effective widths are adopted.
As illustrated in Fig. 17, the test specimen consists of a
3.8 m composite beam comprising a European HEB 180 section
which supports a 120 mm thick and 1.6 m wide concrete slab.
Full shear connection was achieved by using 56 headed studs
with 16 mm diameter and 100 mm height. The beam was loaded

Fig. 16. Influence of external joint on effective width distribution.

Fig. 17. Specimen B4 tested by Amadio et al. [8].

J.M. Castro et al. / Journal of Constructional Steel Research 63 (2007) 13171327

1325

Fig. 18. 2D idealisation of a composite beam.


Table 1
Steel properties of the B4 specimen
Material

E (N/mm2 )

f y (N/mm2 )

(%)

Structural steel
Reinforcing steel

210 000
210 000

317
554

1
2

Table 2
Concrete properties of the B4 specimen
Property (N/mm2 )

Value

E
fc
ft

36 744
41.60
3.32

with two symmetric concentrated loads applied at 0.5 m from


the mid-span. The mechanical properties of the materials for
specimen B4 are listed in Tables 1 and 2.
Like the 3D models employed before, the simplified 2D
models prepared in ADAPTIC consist of representing the
composite beam by two parallel lines of cubic elasto-plastic
beam elements (cbp2) incorporating a fibre approach. The
lower line corresponds to the steel section whereas the upper
line represents the concrete or composite slab. As shown in
Fig. 18, the two lines of beam elements are positioned at the
centroids of the two constituent parts and composite action is
achieved through inclusion of links (lnk2) with rigid properties.
Joint elements (jel2) are also incorporated at appropriate
locations within the links in order to model the interaction
effects as realistically as possible.
With regard to material modelling, a bilinear elasto-plastic
model (stl1) is adopted for both structural and reinforcing steel.
On the other hand, concrete nonlinear behaviour is accounted
for by adoption of a uniaxial constitutive model (con1)
featuring both compressive and tensile softening. In terms of
shear interaction, trilinear behaviour curves are assigned to the
joint elements based on the properties provided by Amadio
et al. [8].
An additional 3D model, similar to those described before in
this paper, is prepared in ADAPTIC to compare its accuracy in
predicting the response obtained from the test. Shear interaction
is considered using the same approach adopted for the 2D
model.
The global response of the composite beam obtained from
both the experiment and the analyses is shown in Fig. 19.
Inspection of the figure indicates the accuracy provided by the
simplified 2D model when the effective width is considered as
80% of the actual slab width. Both the initial stiffness and the
ultimate capacity are almost coincident with those observed in
the test and provide more accurate representation in comparison

Fig. 19. Global response of the B4 specimen.

with the case when the code effective width (L/4) is employed.
The excellent prediction provided by the 3D model is also
worth noting, although this model is computationally much
more demanding.
The local response is now examined by comparing the
momentcurvature relationships evaluated at the central region
of the composite beam which is under pure bending conditions.
The results from the various 2D models are compared with the
3D model due to absence of test results. In all the models,
the curvatures are obtained from the strain gradients of the
beam element representing the steel section, and the results
are presented in Fig. 20. The curves clearly illustrate the better
estimate provided by the 2D models adopting effective widths
larger than that proposed by the code. However it should be
mentioned that, for the same level of vertical deformation, large
effective widths lead to an overestimation of curvatures.

1326

J.M. Castro et al. / Journal of Constructional Steel Research 63 (2007) 13171327

Fig. 21. Global responses for full interaction.

Fig. 20. Momentcurvature relationships.

The comparisons established above were based on models


incorporating shear interaction effects. However, it is of interest
to check the accuracy of adopting large effective widths in a full
interaction case. The same beam is now analysed assuming full
interaction. The global responses obtained for both 2D and 3D
models are provided in Fig. 21 whereas the momentcurvature
relationships are illustrated in Fig. 22.
Once again, the curves confirm the relative accuracy
provided by 2D models adopting effective widths approaching
the full slab width both in terms of initial stiffness and
strength. The initial stiffness ratios presented in Table 3 clearly
demonstrate that only a nearly full effective width can provide
a good estimate of the initial stiffness. With regard to the local
response, it is worth noting the underestimation of maximum
curvature provided by the 2D models for the same levels of
vertical deformation of the beam.
From the above discussions it becomes clear that
consideration of code effective widths led to an underestimation
of the stiffness and capacity of a composite beam. This effect
is more pronounced for small span beams. As the common
code expression for effective width is a function of the beam
span, relatively low values are expected for such beams. This
is reflected in smaller internal lever arms developing at the
cross-section level with the consequent underestimation of the
beam capacity. However, for the case of beams associated with
large slab widths, consideration of full effective widths leads to
realistic estimates in terms of stiffness but, in some cases, the
capacity can be overestimated as compared to more detailed
3D simulations, particularly when strain-hardening effects are
considered in the 2D analyses. Nevertheless, it is relevant to
point out that only full effective widths can provide a realistic
estimate of the moment capacity when plastic cross-section
analysis is performed.
7. Conclusions
In this paper, a new methodology for assessment of effective
width in composite beams is proposed. The method, which at
present is limited to full-interaction cases, is easier to apply in
comparison with existing procedures based on stress integration
and provides more accurate estimates of the effective widths

Fig. 22. Momentcurvature relationships for full interaction.


Table 3
Initial stiffness ratios between 2D and 3D models
Effective width (beff )

K 2D /K 3D

b
0.8b
L/4 0.6b

1.03
0.97
0.89

to use in simplified 2D analysis. Illustrative examples show


that the effective width is mostly related to the full slab width
but it also depends on a number of parameters such as the
slab thickness, the beam span and on the boundary conditions.
The effective widths obtained at the most stressed regions of
the composite beams considered were always above 80% of
the slab width. Additionally, detailed 3D numerical simulations
reveal that stress redistribution develops in the slab at the onset
of inelasticity. On this basis, comparisons with experimental
results demonstrate that better predictions of the response are
obtained when effective widths approaching the full slab width
are employed in 2D models.
References
[1] AASHTO. LRFD bridge design specifications. Washington (DC):
American Association of State Highway and Transportation Officials;
1998.

J.M. Castro et al. / Journal of Constructional Steel Research 63 (2007) 13171327


[2] Adekola AO. Effective width of composite beams of steel and concrete.
The Structural Engineer 1968;46(9):2859.
[3] Adekola AO. The dependence of shear lag on partial interaction in
composite beams. International Journal of Solids and Structures 1974;10:
389400.
[4] Ahn IS, Chiewanichakorn M, Chen SS, Aref AJ. Effective flange width
provisions for composite steel bridges. Engineering Structures 2004;
26(12):184351.
[5] AISC. AISC 360-05Specification for structural steel buildings. Chicago
(IL): American Institute of Steel Construction; 2005.
[6] AISC. Commentary to the specification for structural steel buildings
(AISC 360-05). Chicago (IL): American Institute of Steel Construction;
2005.
[7] Allen DNdG, Severn RT. Composite action between beams and slabs
under transverse load. The Structural Engineer 1961;39(Part I):14954.
[8] Amadio C, Fedrigo C, Fragiacomo M, Macorini L. Experimental
evaluation of effective width in steelconcrete composite beams. Journal
of Constructional Steel Research 2004;60(2):199220.
[9] Amadio C, Fragiacomo M. Effective width evaluation for steelconcrete
composite beams. Journal of Constructional Steel Research 2002;58(3):
37388.
[10] Ansourian P. An application of the method of finite elements to the
analysis of composite floor systems. Proceedings of the Institution of Civil
Engineers 1975;59:699726.
[11] Brosnan DP, Uang CM. Effective width of composite L-beams in
buildings. Engineering Journal, AISC 1995;32(2):7380.
[12] BSI. BS 5950-3.1, Structural use of steelwork in buildingPart 3: Design
in composite constructionSection 3.1 Code of practice for design
of simple and continuous composite beams. London: British Standards
Institution; 1990.
[13] CEN. EN 1994-1-1, Eurocode 4: Design of composite steel and concrete
structuresPart 1.1: General rules and rules for buildings. Brussels:
European Committee for Standardization; 2004.
[14] CEN. EN 1998-1, Eurocode 8: Design provisions for earthquake
resistance of structures, Part 1: General rules, seismic actions and rules
for buildings. Brussels: European Committee for Standardization; 2005.
[15] Castro JM. Seismic behaviour of composite moment-resisting frames.
Ph.D. thesis. University of London; 2006.
[16] Castro JM, Elghazouli AY. Behaviour of composite beams in
moment-resisting frames. In: 12th European conference on earthquake
engineering. Paper No. 521. 2002.

1327

[17] Chapman JC. Composite construction in steel and concreteThe


behaviour of composite beams. The Structural Engineer 1964;42(4):
11525.
[18] Chiewanichakorn M. Intrinsic method of effective flange width evaluation
for steelconcrete composite bridges. Buffalo (NY): The State University
of New York; 2005.
[19] Chiewanichakorn M, Aref AJ, Chen SS, Ahn IS. Effective flange width
definition for steelconcrete composite bridge girder. Journal of Structural
Engineering 2004;130(12):201631.
[20] Elghazouli AY, Izzuddin BA. Realistic modeling of composite and
reinforced concrete floor slabs under extreme loading. II: Verification and
application. Journal of Structural Engineering 2004;130(12):198596.
[21] Elghazouli AY, Migiakis CE. Effective size of composite beams for
capacity design. In: Int. conf. on steel structures, EuroSteel 99. 1999. p.
58593.
[22] Elkelish S, Robinson H. Effective widths of composite beams with ribbed
metal deck. Canadian Journal of Civil Engineering 1986;13(5):57582.
[23] Fahmy EH, Robinson H. Analyses and tests to determine the effective
widths of composite beams in unbraced multistory frames. Canadian
Journal of Civil Engineering 1986;13(1):6675.
[24] Heins CP, Fan HM. Effective composite beam width at ultimate load.
Journal of the Structural Division, ASCE 1976;102(ST11):216379.
[25] Izzuddin BA. Nonlinear dynamic analysis of framed structures. University
of London; 1991.
[26] Izzuddin BA. Integration of beamcolumn and shell elements in large
displacement structural analysis. In: International symposium on new
perspectives for shell and spatial structures (IASS-APCS). 2003.
[27] Izzuddin BA, Elghazouli AY. An advanced concrete model for R/C and
composite floor slabs subject to extreme loading. In: 9th int. conf. on civil
and structural engineering computing. 2003.
[28] Izzuddin BA, Elnashai AS. Adaptive space frame analysis. 2A
distributed plasticity approach. Proceedings of the Institution of Civil
EngineersStructures and Buildings 1993;99(3):31726.
[29] Izzuddin BA, Tao XY, Elghazouli AY. Realistic modeling of composite
and reinforced concrete floor slabs under extreme loading. I: Analytical
method. Journal of Structural Engineering 2004;130(12):197284.
[30] Johnson RP. Research on steelconcrete composite beams. Journal of the
Structural Division, ASCE 1970;96(3):44559.
[31] Miller AB. The effective width of a plate supported by a beam. Selected
engineering papers. The Institution of Civil Engineers; 1929. Paper
no. 83.

You might also like