Engineering Structures 113 (2016) 121–132

Contents lists available at ScienceDirect

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

Elastic critical moment of continuous composite beams

with a sinusoidal-web steel profile for lateral-torsional buckling
Janaina Pena Soares de Oliveira a,b,c, Adenilcia Fernanda Grobério Calenzani a,⇑, Ricardo Hallal Fakury b,
Walnório Graça Ferreira a
a
Department of Civil Engineering, Universidade Federal do Espírito Santo, Vitória, ES, Brazil
b
Department of Structural Engineering, Universidade Federal de Minas Gerais, Belo Horizonte, MG, Brazil
c
Department of Civil Engineering, Faesa, Vitória, ES, Brazil

a r t i c l e i n f o a b s t r a c t

Article history: Lateral-torsional buckling (LTB) is an ultimate limit state that can occur in the hogging moment regions of
Received 6 May 2015 continuous composite steel and concrete beams. This limit state is characterised by the buckling of the
Revised 11 January 2016 steel profile compressed flange (bottom flange) about the minor axis, together with a distortion of the
Accepted 13 January 2016
steel profile web. The European Standard EN 1994-1-1:2004 provides an approximate procedure for
Available online 11 February 2016
LTB design that is applicable to continuous composite beams, but only those with a plane web steel pro-
file. The most important step of this procedure is the determination of the elastic critical moment. In this
Keywords:
paper, a finite element analysis (FEA) model was developed using the software ANSYS to determine the
Lateral-torsional buckling
Continuous composite beams
elastic critical moments of continuous composite steel and concrete beams with corrugated sinusoidal-
Sinusoidal-web steel profile web steel profiles, which were evaluated against numerical data from the literature. Ultimately, a study
Elastic critical moment involving 45 models was conducted based on FEA modelling, and a procedure for predicting the elastic
critical moment of composite beams with sinusoidal-web steel profiles was proposed.
Ó 2016 Elsevier Ltd. All rights reserved.

1. Introduction by torsion) and has been well described by several researchers,

including Johnson [2] (Fig. 1).
1.1. Lateral-torsional buckling of continuous composite beams
1.2. Procedures for LTB design for plane web steel profiles
Currently, continuous composite steel and concrete beams are
often used in bridges and buildings for larger spans. Due to European Standard EN 1994-1-1:2004 [1] provides an approxi-
moment redistribution, it is possible for these beams to use the mate procedure for LTB design (the Brazilian Standard ABNT NBR
lightest steel profile, which makes them an attractive structural 8800:2008 [3] adopts the same procedure) for the case in which
option. However, continuous beams are subjected to hogging the same slab is attached to one or more parallel supporting steel
moments at the internal supports, and are susceptible to an ulti- members, but the procedure is applicable only to continuous com-
mate limit state called lateral-torsional buckling [1] (in the follow- posite beams with plane webs. The most important step of the pro-
ing LTB). This failure mode is also referred to as lateral distortional cedure is the determination of the elastic critical moment, Mcr,
buckling (LDB) or restrained distortional buckling (RDB) in current based on the behaviour of the inverted U-frame model, which is
literature. formed by two adjacent beams and the supported slab. As shown
Lateral-torsional buckling is characterised by the buckling of in Fig. 2, this model assumes that the value of Mcr depends on
the steel profile compressed flange (bottom flange) about the the flexural stiffness of the cracked concrete slab, k1; the flexural
minor axis, together with a distortion of the steel profile web stiffness of the steel profile web (web distortion stiffness), k2;
(the bottom flange presents a lateral displacement accompanied and the stiffness of the shear connection, k3. To facilitate the
design, the U-frame model can be represented by a steel beam
alone, in which the lateral displacement is restrained and a rota-
⇑ Corresponding author at: Av. Fernando Ferrari, 514, Goiabeiras, 29075-910
Vitoria, ES, Brazil. Tel.: +55 27 40092666, cell: +55 27 999946020.
tional stiffness per unit length, ks, is used at the level of the top
E-mail addresses: [email protected] (J.P.S. de Oliveira), afcalenzani@ flange; this rotational stiffness can be expressed in terms of k1, k2
gmail.com (A.F.G. Calenzani), [email protected] (R.H. Fakury), walnorio@gmail. and k3 as follows:
com (W.G. Ferreira).

http://dx.doi.org/10.1016/j.engstruct.2016.01.021
0141-0296/Ó 2016 Elsevier Ltd. All rights reserved.
122 J.P.S. de Oliveira et al. / Engineering Structures 113 (2016) 121–132

of the steel section; Iafz is the second moment of area of the bottom
flange about the minor axis of the steel member; C4 is a property
associated with the distribution of bending moments; and kc is a
factor related to the geometry of the structural steel section. If the
cross-section of the steel member is doubly symmetric, this factor
is given by

hs Iy =Iay
kc ¼ ð5Þ
h2s =4þðIay þIaz Þ=Aa
e
þ hs

with
AIay
e¼ ð6Þ
Aa zc ðA  Aa Þ
Fig. 1. Lateral-torsional buckling (LTB) of continuous composite beams.

1 where zc is the distance between the centroid of the steel section
1 1 1 and the mid-depth of the slab; A is the area of the equivalent com-
ks ¼ þ þ ð1Þ
k1 k2 k3 posite section, neglecting concrete in tension; Iy is the second
moment of area for major-axis bending of the composite section;
The stiffnesses k1 and k2 are defined as follows:
Iay and Iaz are the second moments of area of the structural steel sec-
aðEIÞ2 tion; and Aa is the area of the structural steel section.
k1 ¼ ð2Þ
a The critical moment is affected by the bending moment dia-
gram and this influence is taken into account by the factor C4. Roik
Ea t 3w et al. [6] proposed the use of tables to determine the values of this
k2 ¼ ð3Þ
4ð1  m2a Þhs factor for spans of continuous composite beams with plane web
steel profiles. Tables 1 and 2 give the values of C4 for spans with
where a is a coefficient equal to 2 for an edge beam, with or without and without transverse loading, respectively.
a cantilever, or 3 for an inner beam (for an inner beam on a floor To obtain the elastic critical moment of continuous composite
with four or more similar parallel beams, a equal to 4 may be used); beams subjected to transverse loading, end moments and torsional
(EI)2 is the cracked flexural stiffness per unit width of the slab; a is moments, Hanswille [7] studied the analogy between the lateral-
the spacing between the parallel beams; Ea is the elasticity modulus torsional buckling and compression of a member on an elastic
of structural steel; tw is the thickness of the web of the structural foundation. Through this analogy, similarly to Roik et al. [6], Han-
steel section; ma is the Poisson’s ratio of structural steel; and hs is swille [7] considered an equivalent system featuring a steel profile
the distance between the centroids of the flanges of the structural with a rigid support that prevented lateral displacement and an
steel section. elastic torsional support with stiffness ks; both were continuous
The shear connection stiffness, k3, is not mentioned by EN 1994- and positioned at the top flange of the steel profile (Fig. 3).
1-1:2004 [1], because its value is very high when compared to k1 Comparing the compression member on an elastic foundation
and k2 stiffnesses and it can, thus, be neglected. According to John- with the lateral-torsional buckling problem, Hanswille [7]
son and Molenstra [4], tests in bridges of composite beams show obtained the following equation:
that the shear connection stiffness affects less than 1% of the total !
rotational stiffness. 1 p2 Ea Jw;D
Although it provides all of the above-described information, the Mcr ¼ þ Ga J T;eff ð7Þ
kz ðbB LÞ2
standard EN 1994-1-1:2004 [1] does not offer an expression for the
calculation of Mcr. However, the old edition of the European Stan- where L is the length of the span; bB is a buckling coefficient; EaJw,D
dard, ENV 1994-1-1:1992 [5], prescribed the formulation of Roik is the warping stiffness; GaJT,eff is the effective St. Venant torsional
et al. [6] to calculate this moment. Based on the energy method stiffness; and kz is a factor related to the geometry of the structural
applied to the steel beam illustrated in Fig. 2, these authors steel section (see Hanswille [7] for more information about these
obtained the following equation: quantities).
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
! Except for the works that will be cited in Section 1.3, the liter-
u
kc C 4 u
t ks L2 ature reveals studies on the determination of Mcr for continuous
M cr ¼ Ga Iat þ 2 Ea Iafz ð4Þ
L p composite beams with plane web steel profiles only. The energy
method of Roik et al. [6] and the equivalent system of Hanswille
where Ga is the shear modulus of structural steel; L is the length of [7] are usually used in studies on the lateral-torsional buckling of
the beam between points at which the bottom flange of the steel these beams (for example, see Chen and Ye [8] and Ye and Chen
member is laterally restrained; Iat is the St. Venant torsion constant [9]).

Fig. 2. Deformations of inverted-U frame model and representation of steel beam alone. Adapted from EN 1994-1-1:2004 [1].
 J.P.S. de Oliveira et al. / Engineering Structures 113 (2016) 121–132 123

Table 1
Values of factor C4 for spans with transverse loading. Source: Roik et al. [6].

Loading and support conditions Bending moment diagrama C4

w = 0.50 w = 0.75 w = 1.00 w = 1.25 w = 1.50 w = 1.75 w = 2.00 w = 2.25 w = 2.50
41.5 30.2 24.5 21.1 19.0 17.5 16.5 15.7 15.2

33.9 22.7 17.3 14.1 13.0 12.0 11.4 10.9 10.6

28.2 18.0 13.7 11.7 10.6 10.0 9.5 9.1 8.9

21.9 13.9 11.0 9.6 8.8 8.3 8.0 7.8 7.6

28.4 21.8 18.6 16.7 15.6 14.8 14.2 13.8 13.5

12.7 9.89 8.6 8.0 7.7 7.4 7.2 7.1 7.0

a
M0 is the mid-length moment of a simply supported span.

Table 2
Values of factor C4 for spans without transverse loading. Source: Roik et al. [6].

Loading and support conditions Bending moment diagrama C4b

w = 0.00 w = 0.25 w = 0.50 w = 0.75 w = 1.00
11.1 9.5 8.2 7.1 6.2

11.1 12.8 14.6 16.3 18.1

a
M is the absolute value of the maximum hogging moment of the span.
b
Values of w greater than 1.00 should be considered equal to 1.00.
tf
hw

bw
tf

bf
Fig. 3. Equivalent system of Hanswille [7].

1.3. Research on LTB for continuous beams with sinusoidal-web

profiles
w w w

Fig. 4. Sinusoidal-web steel profile and web corrugation geometry. Source: Calen-
Sinusoidal-web profiles have a doubly symmetrical I shape zani et al. [10].
composed of plane plate flanges and a sinusoidal thin plate web,
as shown in Fig. 4. Due to the thin plate web, it is effective to – flange width (bf) varying from 125 mm to 300 mm and flange
use these profiles in steel composite beams of bridges and build- thickness (tf) ranging from 6.3 mm to 19 mm.
ings to obtain lightweight steel structures.
Calenzani et al. [10] and Calenzani [11] performed experiments In the experiments, the authors tested four prototypes repre-
and finite element analyses (FEA) to propose a procedure for sentative of the inverted U-frame model, two with solid concrete
obtaining the rotational stiffness, ks, in composite beams with slabs and two with slabs cast on formed steel decks (composite
sinusoidal-web steel profiles with (see Fig. 4) the following slabs). FEA models were developed using software ANSYS 9.0 and
parameters: were evaluated against results from the corresponding experi-
ments. Finally, a parametric study was performed using 68 numer-
– web height (hw) varying from 400 mm to 1200 mm and web ical models.
thickness equal to 2 mm or 3 mm; Calenzani et al. [10] concluded that ks could be calculated by Eq.
– web wavelength (w) equal to 155 mm and total heights (bw) (1), with slab stiffness k1 obtained by Eq. (2). For the sinusoidal-
equal to 40 mm and 43 mm for web thicknesses (tw) of 2 mm web stiffness of the steel profile, the following expression was
and 3 mm, respectively; proposed:
124 J.P.S. de Oliveira et al. / Engineering Structures 113 (2016) 121–132

Table 3
Shear connection stiffness values, k3, for solid concrete slabs.

Flange connected to slab k3,p (kN/rad)

Longitudinal distance between shear connectors
s 6 201.5 mm 201.5 mm < s 6 403 mm
Thickness Width (mm) 1 connector in the cross- 2 or more connectors in the cross- 1 connector in the cross- 2 or more connectors in the cross-
(mm) section section section section
16 6 tf 6 19 bf P 250 6000 15,600 4000 6300
125 6 bf < 250 5500 10,800 3400 4500
8 6 tf < 16 bf P 250 1750 2700 1100 1250
125 6 bf < 250 1700 2100 1100 1050
6.3 6 tf < 8 125 6 bf < 250 1300 1350 850 650

Table 4
Shear connection stiffness values, k3, for composite slabs.

Flange connected to slab k3,p (kN/rad)

Shear connectors in all ribs Shear connectors in alternated ribs
Thickness Width (mm) 1 connector in the cross- 2 or more connectors in the cross- 1 connector in the cross- 2 or more connectors in the cross-
(mm) section section section section
16 6 tf 6 19 bf P 250 4000 8000 2800 3500
125 6 bf < 250 3200 5200 2100 2800
8 6 tf < 16 bf P 250 1200 1800 900 950
125 6 bf < 250 1050 1300 950 900
6.3 6 tf < 8 125 6 bf < 250 800 850 750 600

Fig. 6. Geometry and loading of example of Hanswille [7].

total rotational stiffness. To determine k3, Calenzani et al. [10] pro-

posed the use of Table 3 for composite beams with solid concrete
slabs and Table 4 for beams with composite slabs.

Fig. 5. Detail view of the numerical model of a beam with plane web. 1.4. About this study

2 This paper proposes a procedure for obtaining the elastic critical

0:4Ea bw tw
k2 ¼ ared ð8Þ moment, Mcr, of composite beams with sinusoidal-web steel pro-
hw files for lateral-torsional buckling, taking into account the results
where ared is a reduction coefficient calculated by multivariable lin- of FEA using the inverted U-frame model. The sinusoidal-web steel
ear regression via the least-squares method, given by profile should comply with the geometric constraints mentioned in
Section 1.3.
ð19  t f Þ
ared ¼ 0:55274571  0:128064  0:000212
tw 2. Development, evaluation and adjustment of FEA model
ð1200  hw Þ ð361  tf Þ 2
 þ 0:009255 ð9Þ 2.1. Preliminary considerations
tw t 2w

with hw, tw and tf in millimetres. In this paper, buckling analysis was performed using commer-
In contrast to composite beams with plane web steel profiles, cial software ANSYS 14.0 [12] to obtain the elastic critical moment,
the shear connection stiffness, k3, cannot be neglected in compos- Mcr, of continuous composite beams. First, FEA models of beams
ite beams with sinusoidal-web profiles because of the greater flex- with a plane web were developed and evaluated against numerical
ural stiffness of the steel profile web. According to Calenzani et al. results reported by Roik et al. [6] and Hanswille [7]. The models
[10], the shear connection stiffness contributes up to 24% of the were then adjusted to consider the sinusoidal-web profile.
 J.P.S. de Oliveira et al. / Engineering Structures 113 (2016) 121–132 125

(a) Model B1 (b) Model B2

(c) Model B3
Fig. 7. FE models of beams with plane web.

Table 5
Evaluation results.

Model Mcr (kN m)

Roik et al. [6] (Mcr,Roik M cr;Roiketal: M cr;FEA M cr;FEA
et al.) Hanswille [7] (Mcr,Hanswille) FEA (Mcr,FEA)
M cr;Hansv ille Mcr;Roiketal: M cr;Hansv ille

B1 15,079 7866 9918 1.92 0.66 1.26

B2 8555 7613 10,040 1.12 1.17 1.32
B3 3816 3946 5026 0.97 1.32 1.27

(a) Flange’s mesh of sinusoidal-web models (b) Numerical model of sinusoidal-web profile
Fig. 8. Numerical model of beams with sinusoidal-web profiles.
126 J.P.S. de Oliveira et al. / Engineering Structures 113 (2016) 121–132

to model rotational springs that were uniformly distributed along

the axis of the top flange of the beam span (Fig. 5).
Because the web profile was modelled by shell elements, the
rotational stiffness, ks, was made equal only to the value of the
rotational stiffness of the slab, k1.
The geometric dimensions of the models are equal to those in
the analytical example of Hanswille [7] (details in Fig. 6), who
studied continuous composite beams with two spans of length
equal to 20 m. The steel profile has a double symmetric I shape
(HE 800 A), with the height of the cross-section and thickness of
the web equal to 734 mm and 15 mm, respectively, and the width
and thickness of the flanges equal to 300 mm and 28 mm, respec-
tively. The rotational stiffness of the slab, k1, is equal to 3640 kN m/
m.
Three FEA models with plane web profiles were processed:
Fig. 9. Moment diagram for a two-span continuous composite beam subjected to model B1 (Fig. 7a), which features two spans and exactly repro-
distributed loads.
duces Hanswille’s beam [7]; model B2 (Fig. 7b), which features
three spans to simulate an internal span; and model B3, which fea-
tures only one simply supported span (Fig. 7c). Models B1 and B2
were subjected to a uniformly distributed transverse load on the
top flange. Model B3 was subjected to a uniform hogging moment
via binary forces at the supports.
In Table 5, the FEA results of this paper are compared to the val-
ues of Mcr obtained by Eqs. (4) and (7) of Roik et al. [6] and Han-
swille [7] for models B1, B2 and B3, respectively.
Since it is not possible to obtain experimentally the elastic crit-
ical moment, the approximations of Roik et al. [6] and Hanswille
[7] were adopted in this paper to evaluate the efficiency of the
numerical models. The large difference between the FEA results
and the results of Roik et al. [6] for model B1 is in agreement with
the observation of Johnson and Anderson [13] that the method of
Roik et al. [6] is less accurate for end spans (in this model there
is also a significant difference between the results of Roik et al.
[6] and those of Hanswille [7]). Despite the differences between
FEA and analytical results, it can be assumed that the numerical
models present good results, since the comparison was made with
approximate calculation methods.

Fig. 10. Moment diagram for a three-span continuous composite beam subjected to 2.3. Adjustment of the FEA model for beams with sinusoidal-web
distributed loads. profile

The FEA models of composite beams with sinusoidal-web pro-

2.2. Models of continuous composite beams with a plane web profile files (Fig. 8) were developed in analogy to the models with a plane
web profile, using SHELL181 and COMBIN14 elements. Triangular
FEA models of beams with a plane web profile were developed, shell elements at the flange were also used because of the web
with the action of the U-frame model guaranteed by adding a uni- corrugation.
form rotational elastic support with stiffness ks on the top flange of The rotational stiffness, ks, of the COMBIN14 elements was
the steel profile and preventing lateral displacement of this flange determined by the formulation of Calenzani et al. [10], considering
(Fig. 2). The steel profile was modelled using rectangular SHELL181 the slab stiffness, k1, and shear connection stiffness, k3. The shear
elements with maximum side lengths equal to 30 mm (value connection stiffness significantly affects ks in composite beams
defined after the mesh test). COMBIN14 spring elements were used with sinusoidal-web profile and cannot be neglected. The

Fig. 11. Linear moment diagrams of simply supported beam.

 J.P.S. de Oliveira et al. / Engineering Structures 113 (2016) 121–132 127

Table 6
FEA models used to analyse the effect of factor C4 in the elastic critical moment Mcr.

Model Loading conditions and support moment diagram w Loading

F1 (kN) F2 (kN) q1 (kN/m) q2 (kN/m) q3 (kN/m)
M1 0 0.0018 0 – – –
M2 0.25 0.00045 – – –
M3 0.5 0.00089 – – –
M4 0.75 0.0013 – – –
M5 1 0.0018 – – –

M6 0.25 0.0018 0.00045 – – –

M7 0.5 0.00089 – – –
M8 0.75 0.0013 – – –
M9 1 0.0018 – – –

F1 (kN) F2 (kN) q1 (kN/m) q2 (kN/m) q3 (kN/m)

M10 0.5 – – – 1.0 –

M11 0.75 – – 0.5 –
M12 1.0 – – 1.0 –
M13 1.25 – – 1.5 –
M14 1.5 – – 2 –
M15 1.75 – – 2.5 –
M16 2.0 – – 3 –
M17 2.25 – – 3.5 –
M18 2.5 – – 4.0 –
M19 0.5 – – 0.13 1.0 –0.25
M20 0.75 – – 0.69 0.13
M21 1.0 – – 1.25 0.5
M22 1.25 – – 1.81 0.88
M23 1.5 – – 2.38 1.25
M24 1.75 – – 2.94 1.63
M25 2.0 – – 3.5 2.0
M26 2.25 – – 4.06 2.38
M27 2.5 – – 4.63 2.75
M28 0.5 – – 0.19 1.0 0
M29 0.75 – – 0.78 0.5
M30 1.0 – – 1.38 1.0
M31 1.25 – – 1.97 1.5
M32 1.5 – – 2.56 2.0
M33 1.75 – – 3.16 2.5
M34 2.0 – – 3.75 3.0
M35 2.25 – – 4.34 3.5
M36 2.5 – – 4.94 4.0
M37 0.5 – – 0.25 1.0 0.25
M38 0.75 – – 0.88 0.88
M39 1.0 – – 1.5 1.5
M40 1.25 – – 2.13 2.13
M41 1.5 – – 2.75 2.75
M42 1.75 – – 3.38 3.38
M43 2.0 – – 4.0 4.0
M44 2.25 – – 4.63 4.63
M45 2.5 – – 5.25 5.25

Fig. 12. Numerical model cross-section (dimensions in millimetres).

128 J.P.S. de Oliveira et al. / Engineering Structures 113 (2016) 121–132

erence and assuming the same length for the three spans, it is pos-
sible to obtain the loads q1 and q3 as a function of the constant w
for each bending moment diagram. Thus, case 1 can be expressed
as follows:
q1 ¼ ð2:25w  1Þq2 ð11Þ

q3 ¼ ð1:5w  1Þq2 ð12Þ

case 2 leads to:
q1 ¼ ð2:375w  1Þq2 ð13Þ

q3 ¼ ð2w  1Þq2 ð14Þ

finally, case 3 yields
q1 ¼ q3 ¼ ð2:5w  1Þq2 ð15Þ
Linear moment diagrams for a simply supported beam sub-
jected to end moments, applied as horizontal forces at the nodes
Fig. 13. LTB mode of model 39 from Table 6. of the bottom and top flanges at the two ends of the steel profile,
as shown in Fig. 11. Two cases of the linear bending moment were
considered. In case 1, the applied loads at the flanges’ nodes pro-
sinusoidal-web stiffness, k2, was not included for the same reason vided a single curvature and were equal to:
cited for the plane web steel profile.
1
F1 ¼ ð16Þ
hs nn
3. FEA and results for sinusoidal-web beams
F 2 ¼ wF 1 ð17Þ
To obtain the elastic critical moment for sinusoidal-web beams,
models with two and three spans were processed, simulating end where hs is the distance between the centroids of the flanges of the
and internal spans of continuous composite beams with structural steel section and nn is the number of nodes in each flange
sinusoidal-web steel profiles, respectively (Figs. 9 and 10), chang- of the steel profile.
ing the values of the distributed transverse loads between the
spans. Additionally, models were processed with spans subjected
to linear variations of the bending moment (Fig. 11).
The maximum bending moment in the central section of the
end span is M0, considering the span to be simply supported, and
the hogging bending moment at the internal support is wM0
(Fig. 9). For this situation, with equal spans, L, and taking the dis-
tributed load q2 as a reference, the following relation between
the coefficient w and the distributed loads, q1 and q2, is well
known:

q1 ¼ ð2w  1Þq2 ð10Þ

For an internal span of continuous composite beams, the max-
imum bending moment, M0, was associated with a hogging bend-
ing equal to wM0 at one support and three different moments at
the other support (Fig. 10): 0.5 wM0 (case 1), 0.75 wM0 (case 2) Fig. 15. Mcr versus w for continuous composite beam at end spans with distributed
and wM0 (case 3). Taking the load at the central span, q2, as a ref- load.

(a) Linear hogging moment (b) Linear hogging and sagging moment
Fig. 14. Mcr versus w for composites beams subjected to linear moments.
 J.P.S. de Oliveira et al. / Engineering Structures 113 (2016) 121–132 129

(a) Case 1 (b) Case 2

300

250

200

Load (kN)
150

100

50

0
0 10 20 30 40 50 60
Deflection (mm)

(c) Case 3
Fig. 16. Mcr versus w for continuous composite beam at internal spans with distributed load.

In case 2, the same loads used in case 1 were applied, except the steel profile were obtained considering the web corrugation,
that force F2 was aligned with force F1 in both flanges, providing to check if these formulations are also suitable for sinusoidal-
reverse curvature. Therefore, it follows that: web profile. Fig. 14 shows that the FEA results for the critical
moment of sinusoidal-web models with linear moment diagrams
(models M1 to M9 – Table 6) are less than those obtained by the
F 2 ¼ wF 1 ð18Þ
formulation of Roik et al. [6], except with respect to model M5.
In total, 45 FEA models were processed (Table 6) with the geo- However, the FEA results are higher than the values of the Han-
metric properties of the composite beam with sinusoidal-web pro- swille [7] formulation.
files and rotational stiffness (ks) shown in Fig. 12. In the composite In the sinusoidal-web models relating to end spans (models
slab is used the steel deck with height (h) and thickness (td) equal M10 to M18 – Table 6) there is a significant difference between
to 75 mm and 1.25 mm, respectively. To obtain the value of slab the numerical results and those of Roik et al. [6] (Fig. 15). This dis-
stiffness (k1) Eq. (2) was used, considering an edge beam, i.e., a crepancy can be explained by the aforementioned observation of
equal to 2, spacing between the parallel beams (a) equal to 2 m Johnson and Anderson [13]. A comparison with Hanswille [7]
and the cracked flexural stiffness of the slab (EI2) equal to results could not be made because the formulation of this author
332 kN m, resulting in k1 equal to 332 kN m/m. The shear connec- has a limited range and does not include all of the w values pre-
tion stiffness (k3) is equal to 1050 kN m/m, as suggested by Calen- dicted by Roik et al. [6].
zani et al. [10] (Table 4). In each model, the cross-section has only Fig. 16 shows that the numerical results yielded by FEA models
one shear connector welded to the centre line of the top flange and of continuous composite beams at internal spans (models M19 to
the longitudinal distance between connectors is equal to M45 – Table 6) are in good agreement with Roik et al. [6]
263.5 mm. The span length, L, is equal to 11.16 m (approximately formulation.
22 times the distance between the centroids of the flanges of the When w is equal to 0.5 for continuous composite beams at end
steel section) and the distance between the beams, a, is equal to and internal spans with distributed loads (Figs. 15 and 16), there is
2 m. a greater difference between the FEA results and the formulation
Fig. 13 shows the buckling mode of model 39. It can be observed by Roik et al. [6]. Under this condition, the examined span has a
that the buckling consists of a single half-wave on each side of the smaller length with the hogging moment and the adjacent span
internal supports, where lateral restraint must be provided. is mostly subjected to the hogging moment (Fig. 17a). This situa-
According to Johnson [2], the half-wave extends over most of the tion affected the results because Roik et al. [6] analysed continuous
length of the hogging moment region, and the point of the maxi- composite beams in a simply supported single span and did not
mum lateral displacement is within two or three beam depths of consider the effect of adjacent spans. For example, in the model
the internal support. M10 (Table 6), for the analysed span shown at Fig. 17a, the hogging
The numerical results were compared with Roik et al. [6] and moment in all the left span increase the destabilizing effect in the
Hanswille [7] formulations, in which the geometric properties of
130 J.P.S. de Oliveira et al. / Engineering Structures 113 (2016) 121–132

The new value of factor C4, represented by C4,sw, is obtained by

Eq. (4), except for the value of ks, which can be determined with Eq.
(1), but adopting Eq. (8) for the sinusoidal-web stiffness, as sug-
gested by Calenzani et al. [10], and using the numerical elastic crit-
ical moment (Mcr,FEA). Thus:
(a) Continuous beam at end spans with ψ equal to 0.5
M cr;FEA
C 4;sw ¼ rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
  ð19Þ
2
kc
L
Ga Iat þ kps L2 Ea Iafz

The values of C4,sw for continuous composite beams at end and

internal spans with a distributed load are shown in Table 7. For
continuous composite beams without transverse load, the values
are shown in Table 8.
(b) Continuous beam at end spans with ψ equal to 0.75 Noticeably, the values of C4,sw in Table 7, for w equal to 0.50,
0.75 and 1.00, have different behaviour than C4 in Table 1, pro-
posed by Roik et al. [6]. For the numerical model of continuous
composite beams was observed considerable influence if the com-
posite beam is modelled with their entire span or in a simply sup-
ported single span, as showed in Fig. 17. Through the influence of
adjacent span on the analysed span, the variation of C4 presented
(c) Continuous beam at end spans with ψ equal to 1.00 erratic behaviours, as shown in Table 7, for w equal to 0.50, 0.75
and 1.00. For example, when w is equal to 0.50, the adjacent span
has only hogging moment along its entire length (Fig. 17a), and
this significantly reduces the elastic critical moment of the anal-
ysed span. This behaviour continues until w equal 1.00 (see
Fig. 17a–c). Then, the behaviour of w is similar to the results of Roik
et al. [6], because the sagging moment in adjacent span will pre-
dominate (Fig. 17d).
In order to check if the parameter C4 is influenced only by
(d) Continuous beam at end spans with ψ equal to 1.25 the bending moment diagram, other models were examined.
Figs. 18–20 show graphical outputs of values obtained by the
Fig. 17. Bending moment diagram for continuous beam at end spans with different
values of w.
procedure proposed for determining the elastic critical moment
of continuous composite beams with sinusoidal-web profiles
and the corresponding FEA results. The following parameters
bottom flange, reducing the elastic critical moment. Increasing w, of Eq. (4) were changed: the ratio between the length of the
Fig. 17b–d, there is a reduction of this effect. span (L) and the height of the steel profile (h), adopting L/h
equal to 14 and 30; the slab rotational stiffness (k1) considering
different slab heights (0.1 m and 0.15 m for concrete slab and
0.14 m and 0.20 m for composite slab); the sinusoidal-web rota-
4. Procedure for determining the elastic critical moment of tional stiffness (k2) considering profile heights, hw, equal to
composite beams with sinusoidal-web profiles 600 mm and 1200 mm and web thicknesses, tw, equal to
0.2 mm and 0.3 mm; the shear connection rotational stiffness
Roik et al. [6] formulation, i.e., Eq. (4), is proposed to determine (k3), adopting different longitudinal spacing between shear con-
the elastic critical moment of composite beams with sinusoidal- nectors (0.2635 m and 0.527 m) and flange thicknesses, tf, equal
web profiles; moreover, the value of factor C4 is corrected and to 9.5 mm and 19 mm. Fig. 17 shows the results obtained from
the rotational stiffness is determined according to Calenzani et al. the analysis of a simply supported beam subjected to uniform
[10] (see Section 1.3). In addition, the remaining parameters of hogging moment (Fig. 17a) and reverse curvature (Fig. 17b).
Eq. (4) depend on the geometric properties of the steel profile Fig. 18 shows the results for a continuous composite beam at
and must be obtained by considering the web corrugation. an end span, and Fig. 19 shows the results for a continuous

Table 7
Values of factor C4,sw for spans with distributed load of composite beams with sinusoidal-web profile.

Loading and support conditions Bending moment diagrama C4,sw

w = 0.50 w = 0.75 w = 1.00 w = 1.25 w = 1.50 w = 1.75 w = 2.00 w = 2.25 w = 2.50
9.4 12.0 12.3 12.2 12.0 11.8 11.6 11.4 11.2

9.8 12.4 12.4 12.0 11.7 11.3 11.1 10.8 10.6

11.2 12.5 12.3 11.8 11.4 11.0 10.7 10.4 10.2

11.6 12.6 12.2 11.6 11.0 10.5 10.1 9.7 9.4

a
M0 is the mid-length moment on a simply supported span.
 J.P.S. de Oliveira et al. / Engineering Structures 113 (2016) 121–132 131

Table 8
Values of factor C4,sw for spans without transverse loading of composite beams with sinusoidal-web profile.

Loading and support conditions Bending moment diagrama C4,swb

w = 0.00 w = 0.25 w = 0.50 w = 0.75 w = 1.00
8.1 7.8 7.4 7.0 6.4

8.1 8.4 8.7 9.0 9.3

a
M is the absolute value of the maximum hogging moment of the span.
b
Values of w higher than 1.00 should be taken equal to 1.00.

(a) Uniform hogging moment with ψ = 1 (b) Linear hogging and sagging moment with ψ = 1
Fig. 18. Mcr,FEA versus Mcr,prop and perfect-fit curve for simply supported beam.

(a) Distributed load withψ = 1 (b) Distributed load with ψ = 2

Fig. 19. Mcr,FEA versus Mcr,prop and perfect-fit curve for continuous composite beam at end span.

(a) Distributed load withψ = 1 (b) Distributed load with ψ = 2

Fig. 20. Mcr,FEA versus Mcr,prop and perfect-fit curve for continuous composite beam at internal span.

composite beam at an internal span, both with a distributed is small, which can be attributed to the fact that the ratio
transverse load. It should be noted that there is good agreement between Mcr,prop and Mcr,FEA is always within the range 0.81–
between the values of the elastic critical moment obtained by 1.12. These results showed that C4 slightly depends also on
the proposed formulation (Mcr,prop) and the numerical results the geometric and mechanical properties of the steel profile.
(Mcr,FEA) because the dispersion in relation to a perfect-fit curve Roik et al. [6] also made this same remark (Fig. 20).
132 J.P.S. de Oliveira et al. / Engineering Structures 113 (2016) 121–132

5. Conclusions References

It was demonstrated that, to date, only studies aimed to deter- [1] Eurocode 4. Design of composite steel and concrete structures. Part 1. 1,
General rules and rules for buildings. EN 1994-1-1 Brussels: European
mine the elastic critical moment, Mcr, for continuous composite Committee for Standardization; 2004.
beams with plane web steel profiles have been performed. It was [2] Johnson RP. Composite structures of steel and concrete: beams, slabs, columns
also shown that many results presented by the authors of these and frames for buildings. 3rd ed. UK: Warwick; 2004. p. 122–69.
[3] ABNT NBR 8800. Design of steel and composite steel and concrete structures
studies were not sufficiently precise because they were obtained for buildings. Associação Brasileira de Normas Técnicas (Brazilian Association
without adequately considering the effects of adjacent spans. of Technical Standards). Rio de Janeiro; 2008 [in Portuguese].
In this paper, a procedure for determining Mcr for continuous [4] Johnson RP, Molenstra N. Strength and stiffness of shear connections for
discrete U-frame action in composite plate girders. Struct Eng 1990;68(19/
composite beams with sinusoidal-web profiles is presented. To 2):386–92.
obtain this procedure, buckling analyses were performed for 45 [5] Eurocode 4. Design of composite steel and concrete structures. Part 1. 1,
models using the software ANSYS 14.0 [12]. For more accurate General rules and rules for buildings. ENV 1994-1-1 Brussels: European
Committee for Standardization; 1992.
numerical results were modelled simply supported composite
[6] Roik K, Hanswille G, Kina J. Solution for the lateral torsional buckling problem
beams, continuous composite beams with two spans to represent of composite beams. Stahlbau 1990;59:327–32 [in German].
an end span, and composite beams with three spans to represent [7] Hanswille G. Lateral torsional buckling of composite beams – comparison of
more accurate methods with Eurocode 4. In: Composite construction in steel
an internal span.
and concrete IV; 2002. p. 105–16.
The proposed procedure uses the equation of Roik et al. [6] for [8] Chen W, Ye JH. Elastic lateral and restrained distortional buckling of doubly
composite beams with plane webs, i.e., Eq. (4), altering the values symmetric I-beams. Int J Struct Stab Dyn 2010;10(5):983–1016.
of the factor C4 (represented by C4,sw), using the rotational stiffness [9] Ye JH, Chen W. Elastic restrained distortional buckling of steel-concrete
composite beams based on elastically supported column method. Int J Struct
according to Calenzani et al. [10] and determining the geometric Stab Dyn 2013;13(1). 1350001(1–29).
properties of the steel profile while taking into account the web [10] Calenzani AFG, Fakury RH, Paula FA, Rodrigues FC, Queiroz G, Pimenta RJ.
corrugation. Values of C4,sw are proposed for composite beams at Rotational stiffness of continuous composite beams with sinusoidal-web
profiles for torsional buckling. J Constr Steel Res 2012;79:22–33.
end and internal spans subjected to distributed transverse loads [11] Calenzani AFG. Procedure to determine the rotational stiffness of composite
and beams without transverse loads. beams with sinusoidal-web profiles. PhD Thesis. Federal University of Minas
The results for the elastic critical moment of composite beams Gerais; Brazil; 2008 [in Portuguese].
[12] Ansys, INC. Release 14.0 Documentation for ANSYS. Canonsburg; 2011.
with sinusoidal-web profiles using the proposed procedure (Mcr, [13] Johnson RP, Anderson D. Designer’s guide to EN 1994-1-1 Eurocode 4: design
prop) showed good agreement with those from the FEA (Mcr,FEA), with of composite steel and concrete structures. London; 2004.
the ratio between Mcr,prop and Mcr,FEA ranging from 0.81 to 1.12.

Acknowledgements

The authors would like to acknowledge the support provided by

the Brazilian public agencies CAPES, CNPq, FAPES and FAPEMIG.