Journal of Constructional Steel Research 61 (2005) 1373–1386

www.elsevier.com/locate/jcsr

Bracing requirements for inelastic castellated beams

Amin Mohebkhaha,∗, Hossein Showkatib
a Department of Civil Engineering, Engineering Faculty, Tarbiat Modares University, Tehran 14115-111, Iran
b Department of Civil Engineering, Engineering Faculty, Urmia University, Urmia, Iran

Received 7 December 2004; accepted 15 March 2005

Abstract

Lateral–torsional buckling can be avoided by properly spaced and designed lateral bracing.
Bracings are usually assumed to be elastic, and so may be characterized by their elastic stiffnesses. It
is well known that an elastic lateral brace restricts partially the lateral buckling of slender beams and
increases the elastic buckling moment. However, a full study of the effect of lateral braces on inelastic
buckling has not been made especially for castellated beams, and it is not known whether the limiting
stiffness for elastic buckling can be applied to castellated beams that buckle inelastically. This paper
develops a three dimensional (3-D) finite-element model using a finite-element program and uses it
to investigate the effect of elastic lateral bracing stiffness on the inelastic flexural–torsional buckling
of simply supported castellated beams with an elastic lateral restraint under pure bending. It was
found that for inelastic castellated beams, the effect of bracing initially is increased to some extent as
the lateral unbraced length increases and then decreased until the beam behaves as an elastic beam.
In other words, the effect of bracing depends not only on the stiffness of the restraint but also on the
modified slenderness of the beam. Also, the results show that Winter’s simplified method to determine
full brace requirements cannot be applied to inelastic castellated beams. Therefore, a general equation
is proposed to determine the value of optimum stiffness (kopt ) in terms of the beam’s slenderness,
applicable to all castellated beams under pure bending.
© 2005 Elsevier Ltd. All rights reserved.
Keywords: Inelastic lateral–torsional buckling; Finite element method; Castellated beam; Beam bracing; Bracing
requirements; Optimum stiffness

∗ Corresponding address: Khabgah Bagheri, Royan Ave., Keshavarz Blvd., Tehran, Iran. Tel.: +98 21
8011001x3385; fax: +98 21 8005040.
E-mail address: [email protected] (A. Mohebkhah).

0143-974X/$ - see front matter © 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jcsr.2005.03.003
1374 A. Mohebkhah, H. Showkati / Journal of Constructional Steel Research 61 (2005) 1373–1386

Nomenclature

bf width of flange
Cw warping section constant
d serial depth of original section
E Young’s modulus of elasticity
Et tangent modulus
Fy yield stress of material
G shear modulus of elasticity
h depth of castellated section
Iy second moment of area about y-axis
J torsion section constant
L unbraced length of castellated beam
Mocr elastic buckling uniform moment
Mopt buckling uniform moment for k = kopt
tf thickness of flange
tw thickness of web
Z plastic section modulus
εy steel strain at yield
εs steel strain at the onset of strain-hardening
λ modified flexural–torsional slenderness of castellated beam
k stiffness of elastic lateral restraint
kopt optimum value of k

1. Introduction
I-beams subjected to flexure have much greater strength and stiffness in the plane in
which the loads are applied (major axis) than in the plane of the minor axis. Unless these
members are properly braced against lateral deflection and twisting, they are subjected to
failure by lateral–torsional buckling prior to the attainment of their full in-plane capacity.
Lateral–torsional buckling is a limit state of structural usefulness where the deformation
of a beam changes from predominantly in-plane deflection to a combination of lateral
deflection and twisting while the load capacity remains at first constant, before dropping
off due to large deflection and yielding.
Castellated beams are made by either flame cutting or automatically cutting a rolled
beam’s web in a zigzag pattern along its centerline and then rejoining the two halves by
welding. This opening up of the original rolled beam increases its section modulus of
inertia about its major axis and results in greater vertical bending strength and stiffness
without any change in weight [9]. The presence of large web openings may have a severe
 A. Mohebkhah, H. Showkati / Journal of Constructional Steel Research 61 (2005) 1373–1386 1375

penalty on the load carrying capacities of castellated beams, depending on the shapes, the
sizes, and the location of the openings.
AISC [1] determines the lateral–torsional buckling moment based on an assumption
of constant moment between lateral supports of the beam. The use of the equation for
evaluating critical moment using a uniform moment diagram as the basic case is intended
to ease analytical analysis. Also, a beam with a uniform moment diagram is the most severe
case in the lateral–torsional buckling consideration.
The critical buckling moment of a simply supported I-beam in the case of pure bending
is given by [2]
π


Mocr = E I y G J (1 + W 2 ), (1)
Lb
where

π ECw
W = ; (2)
Lb GJ
L b is the unbraced length which is the span length, L, in this case. Eq. (1) is derived
for constant moment, and so the resulting differential equation describing the equilibrium
conditions of the beam at its slightly deformed state is linear with constant coefficients.
Most structural forms include members such as columns and beams for which elastic
lateral or flexural–torsional buckling is a possible means of failure. The determination of
the load level, which would cause such a failure, is a problem for the designer because
it is one of the limits to overall load capacity. An estimate of this load must be made
by determining an ‘effective length’, or in some other rational manner. Few members
can be treated in isolation. Each compound of a structure both draws support from and
supports other members against buckling action. Bracing members are deliberately placed
to provide such support. These members are usually assumed to be elastic, in which case
they may be characterized by their elastic stiffness.
Bracing systems for beams must prevent the relative displacement of the top and bottom
flanges (i.e. twist of the section). Substantially there are two general types of beam bracing:
lateral and torsional, which can be discrete or continuous. Lateral bracing restrains lateral
displacement, as its name implies. A torsional brace can be differentiated from a lateral
brace in that twist of the cross section is restrained directly [16]. Some bracing systems
such as slabs attached to the top flange with shear studs restrain lateral displacement and
twist simultaneously.
A review of the laboratory investigations on the bracing of beams and columns reveals
that there are two key aspects, which play a significant role in determining full brace
requirements; that is, the bracing stiffness necessary and the bracing strength necessary.
Winter [15] showed that the effective braces require not only adequate stiffness but also
sufficient strength. The strength requirement is directly related to the magnitude of the
initial out-of-straightness of the member to be braced.
The buckling load increases as the brace stiffness increases until full bracing causes
the beam to buckle between braces. This ultimate value of stiffness is called “optimum
stiffness”. In many instances the relationship between bracing stiffness and buckling load
is nonlinear. In most design situations full bracing is assumed or desired; that is, buckling
1376 A. Mohebkhah, H. Showkati / Journal of Constructional Steel Research 61 (2005) 1373–1386

between the brace points is assumed. For simply supported elastic plain-webbed beams
under uniform moment, the Winter optimum lateral brace stiffness required to force
buckling between the braces is
kopt = Ni P f /L b , (3)
where
P f = π 2 E I yc /L 2b ;
I yc is the out-of-plane moment of inertia of the compression flange, which is I y /2 for
doubly symmetric cross-sections.
Ni is a coefficient depending on the number of braces n within the span, as given
approximately by Ni = 4 − (2/n).
L b is the unbraced length, which is equal to L/2 in this case.
Laboratory tests have been conducted to study the lateral–torsional buckling behavior
of castellated beams by Nethercot and Kerdal [6,9]. Other researchers have studied this
problem by means of the finite-element method by Mohebkhah [7,8], and Showkati [10].
The strength and stiffness requirements of plain-webbed beams with central lateral and
torsional restraints have been discussed by Yura [16], Valentino et al. [12,13], Trahair [11],
and Wang et al. [14]. Mohebkhah [7] have studied the stiffness requirements of castellated
beams with elastic lateral restraints using FEM. The results of many studies on bracing
requirements of elastic plain-webbed beams have been summarized in the Guide to
Stability Design Criteria for Metal Structures [5], Trahair [11], and in Yura [16]. However,
it has not been investigated yet whether the given optimum stiffness in the references can
be applied to beams, especially castellated beams that buckle inelastically.
The purpose of this paper is to investigate the stiffness requirements of central lateral
restraints that are intended to restrain inelastic simply supported castellated beams so as
to increase their lateral buckling moment to those of rigidly restrained beams under pure
bending. For this purpose, a finite-element model is developed for the nonlinear inelastic
flexural–torsional analysis of castellated beams with a wide variety of slenderness and
brace stiffness. Then, it is used to investigate the effect of elastic lateral bracing stiffness
on the inelastic flexural–torsional buckling load of simply supported castellated beams with
an elastic lateral restraint under pure bending.

2. Nonlinear finite-element model

To investigate the bracing requirements of standard Iranian castellated beams with

hexagonal openings, a nonlinear inelastic finite-element model has been developed based
on the assumption that the cross section of the beam is doubly symmetric.

2.1. Mesh and material properties

The nonlinear computations were performed using the commercial finite-element

software package COSMOS/M [3]. The program has the ability to consider both geometric
and material nonlinearities in a given model. However, all modeling reported herein only
considers nonlinear material influences. Four side shell elements from the COSMOS/M
 A. Mohebkhah, H. Showkati / Journal of Constructional Steel Research 61 (2005) 1373–1386 1377

Fig. 1. Finite element mesh.

Fig. 2. Uniaxial constitutive model considered.

element library were used to model the web, flange (top and bottom) and the stiffeners. In
order to model the elastic lateral restraints, a spring element was employed. Flanges were
modeled with four elements across the width. Buckling analyses with mesh sizes about
40 × 40 mm showed convergence of the buckling load. The mesh arrangement used for an
analysis is shown in Fig. 1.
The trilinear elastic–plastic strain-hardening stress–strain curve of Fig. 2 is assumed.
Young’ modulus E was set to 210 GPa and Poisson’s ratio was set to 0.3 [8]. Residual
stresses were not considered in this work, though it is relevant in this type of analysis.
1378 A. Mohebkhah, H. Showkati / Journal of Constructional Steel Research 61 (2005) 1373–1386

Fig. 3. Castellated beam and opening geometry.

Fig. 4. Applying the end moments.

2.2. Loads and restraints

Simply supported castellated beams, CPE140 (d = 14, b f = 7.3, t f = 0.69, tw =
0.47 cm), with different spans length and brace sizes under pure bending with an elastic
lateral brace attached to the top flange at midspan are investigated in order to evaluate the
optimum lateral brace stiffness (kopt) for them.
A typical configuration of the expanded beam, and the notation adopted, are shown
in Fig. 3. All beams were provided with bearing stiffeners at support and bracing points.
And also, it should be mentioned that in finite-element analysis of beams subjected to pure
bending, the end moments were applied as couples in order to prevent web local yielding,
as shown in Fig. 4.
A full Newton–Raphson procedure is used in conjunction with an arc length control
iterative strategy [4] and an automatic incrementation strategy to solve the nonlinear
equilibrium equations. The sign of the initial load increment follows the sign of the
determinant of the tangent stiffness matrix.
A convergence criterion based on the maximum norm of the incremental displacements
is adopted. In the incremental–iterative process, each load step consists of the application
of an increment of the external loads and subsequent iterations to restore equilibrium.

3. Validation of the modeling technique

In this part we intend to evaluate the accuracy of the finite-element model of the beams.
There were no experimental test results on bracing requirements of castellated beams.
 A. Mohebkhah, H. Showkati / Journal of Constructional Steel Research 61 (2005) 1373–1386 1379

Fig. 5. W12x14 with an elastic lateral restraint.

Fig. 6. Effect of lateral brace stiffness.

Hence, the comparison was performed using the available results of a test program on
the lateral bracing of plain-webbed beams in Yura [16]. In the test program, the bracing
requirements of a W12X14 section 20-ft (7.31-m) long, with an elastic lateral brace
attached to the top flange at midspan subjected to a concentrated load was examined
experimentally, as shown in Fig. 5. Therefore, the behavior of the beam is studied
using the finite-element model developed in this study and the effect of different brace
sizes (stiffness) on the buckling strength is illustrated in Fig. 6. Fig. 6 also shows the
experimental buckling loads for comparison. As can be seen, the agreement is good
between the experimental buckling loads and those obtained by nonlinear finite-element
analysis.

4. Nonlinear analysis of the castellated beams

After validating the finite-element model, a nonlinear analysis was performed looking
at the factor that affects the optimum value of stiffness on inelastic castellated beams.
That factor is the beam’s slenderness. Nine castellated beams with different lengths and
lateral brace stiffness under pure bending are considered to study the different structural
behavior. These beams are chosen in a manner that after the attainment of optimum lateral
1380 A. Mohebkhah, H. Showkati / Journal of Constructional Steel Research 61 (2005) 1373–1386

Table 1
Critical buckling moments (kN m)

L (m) k=0 k = 10 k = 20 k = 30 k = 40 k = 50 k = 60 k = 70 k = 80
(N/mm)

0.84 31.48 32.00 32.00 32.00 32.00 32.00 32.00 32.00 32.00
1.26 30.16 30.68 30.68 30.68 30.68 30.68 30.68 30.68 30.68
1.68 26.31 26.41 26.49 26.55 26.60 26.61 26.61 26.61 26.61
2.94 17.52 19.12 20.20 21.36 22.49 23.14 24.02 24.02 24.02
4.20 11.10 13.00 14.79 16.50 18.24 20.00 21.42 22.80 23.07
5.46 8.00 10.31 12.79 14.93 16.92 18.08 18.71 18.86 18.93
6.72 6.25 9.741 12.79 14.56 14.56 14.56 14.56 14.56 14.56
7.56 5.49 9.484 12.46 12.46 12.46 12.46 12.46 12.46 12.46
8.40 4.90 9.374 10.89 10.89 10.89 10.89 10.89 10.89 10.89

brace stiffness, they would be placed in three zones (i.e. elastic, inelastic and plastic
zones). Therefore, a general knowledge on the buckling behavior of castellated beams in
each region under the effect of elastic lateral brace stiffness will be obtained. The plastic
moment for the castellated beam, having the properties of the hole cross section, is given
as follows [7]:
M p = Fy × (2Q) = Fy [b f t f (h − t f ) + 36tw (h − 6t f )(5h − 6t f )]
(4)
M p = 29.05 kN m
in which Q is the static moment of the tension or compression area about the neutral axis
of the castellated section. Other parameters are shown in Fig. 3.

4.1. Buckling analysis of the models under pure bending

A nonlinear analysis is carried out for all the selected castellated beams including an
elastic lateral brace attached to the top flange at midspan with different spans length. The
critical buckling moments due to the FEM solution for the beams are shown in Table 1
with different spans length and brace stiffness. Also, the graphs shown in Fig. 7 illustrate
the effect of different brace stiffness on the buckling strength of the laterally braced beams.
As can be seen from Fig. 7, the elastic lateral brace has the greatest effect in the inelastic
region, namely for the intermediate lengths. The results shown in the figure also indicate
that an elastic lateral brace is less effective for plastic beams (short beams). As an overall
conclusion for inelastic beams, the effect of bracing initially is increased to some extent as
the lateral unbraced length increases and then decreased until the beam enters the elastic
zone.

4.2. Determining the optimum stiffness values

The values of optimum stiffness in terms of span length are extracted from Table 1
and shown in Table 2. It should be mentioned that with respect to the lack of a known
method to estimate the optimum stiffness in the references, the value of kopt is determined
approximately using the form of Mcr –k curves based on engineering judgment. The data
 A. Mohebkhah, H. Showkati / Journal of Constructional Steel Research 61 (2005) 1373–1386 1381

Fig. 7. The effect of brace stiffness on the moment capacity curves.

Table 2
Optimum stiffness and buckling moments

L (m) 1.68 2.94 4.20 5.46 6.72 7.56 8.40

λ 0.85 1.32 1.68 1.98 2.23 2.38 2.52

kopt (N/mm) 40 60 70 50 30 20 14
kopt L 3 /E I y 2.02 16.21 55.13 86.5 96.77 91.5 88.2
Mopt (kN m) 26.6 24.02 22.8 18.08 14.56 12.46 9.98
Mopt /M0 1.01 1.37 2.05 2.26 2.33 2.08 2.04
M0 /M p 0.906 0.603 0.382 0.275 0.215 0.189 0.169

Fig. 8. Optimum stiffnesses for the castellated beam under pure bending.

in Table 2 is plotted as a kopt–L curve, which shows the variation of optimum stiffness of
the bracing with the span length L in Fig. 8.
1382 A. Mohebkhah, H. Showkati / Journal of Constructional Steel Research 61 (2005) 1373–1386

Fig. 9. Flexural capacity of the castellated beam for k = 0.

5. Parametric analysis
Now, for more clarification and clear concluding, all the resulting values are converted
to dimensionless values. For this purpose, the dimensionless parameters kopt L 3 /E I y and
Mopt /M0 are introduced for the optimum stiffnesses and buckling moments, respectively.
M0 is the inelastic buckling moment of the beams without lateral restraint (k = 0), which
is shown in Table 1. It should be noted that the optimum buckling moment value is the
moment corresponding to the optimum stiffness value determined previously. And also,
the value of the unbraced length is turned into dimensionless values as follows:

Mp
λ= (5)
Mocr
in which Mocr is the elastic buckling moment of the castellated beam having the properties
of the hole cross section given by Eq. (1); M p is the plastic moment of the castellated beam
given by Eq. (4).
The values of the dimensionless parameters are calculated and shown in Table 2. In
order to introduce a relationship between the inelastic buckling moment M0 (for the
castellated beams with no lateral brace) and the modified dimensionless slenderness
parameter, λ, the variation of these two parameters is shown in Fig. 9.
Fig. 9 also shows a regression equation for the dimensionless inelastic buckling moment
computed using the finite-element method. For any value of the modified slenderness, the
corresponding inelastic buckling moment of a castellated beam with any size can be easily
computed as follows:
M0 

= 0.8 λ4 + 3 − λ2 ≤ 1. (6)
Mp
It can be seen that this equation is quite close to the finite-element results. By conducting
the same procedure as in the above, the variation of the dimensionless optimum buckling
moment and stiffness parameters are plotted against the modified slenderness in Figs. 10
 A. Mohebkhah, H. Showkati / Journal of Constructional Steel Research 61 (2005) 1373–1386 1383

Fig. 10. Optimum buckling moments versus modified slenderness.

Fig. 11. Optimum brace stiffnesses versus modified slenderness.

and 11, respectively. As can be seen from Fig. 11, the value of optimum brace stiffness
initially is increased to some extent as the modified slenderness increases and then
decreased until the beam behaves as an elastic beam. The relevant polynomial regression
equations for these parameters are as follows:
Mopt
= −1.063λ3 + 4.68λ2 − 5.39λ + 2.85 (7)
M0
3
kopt L
= −86.12λ3 + 418.55λ2 − 566.55λ + 233.8. (8)
E Iy

Having determined the value of parameter λ, for a castellated beam of any size, the
parameters M0 , Mopt , and kopt can be easily calculated using Eqs. (6)–(8).
1384 A. Mohebkhah, H. Showkati / Journal of Constructional Steel Research 61 (2005) 1373–1386

Table 3
Buckling moment of an elastically restrained castellated beam

k (N/mm) k/kopt Mk (kN m) Mk /M0 Mk /M0 Difference

FEM FEM Eq. (9) (%)

0 0 8.0 1 1 0
10 0.2 10.31 1.289 1.35 4.7
20 0.4 12.79 1.6 1.626 1.6
30 0.6 14.93 1.87 1.861 −0.5
40 0.8 16.92 2.12 2.07 −2.3
50 1 18.08 2.26 2.26 0

Fig. 12. Effect of restraint stiffness on beam strength.

To predict the buckling moment of an elastically restrained castellated beam, Mk , under

the condition of k ≺ kopt, Eq. (9) might be used [11]:
   
Mk 2 k Mopt 2
≈1+ −1 . (9)
M0 kopt M0

As can be seen from Eq. (9), if k = 0 (i.e., the lateral brace is absent) the right-hand
side of the equation reduces to unity (i.e., Mk ≈ M0 ), and as k approaches kopt, it reaches
(Mopt /M0 )2 (i.e., Mk ≈ Mopt ).
Eq. (9) was suggested to be used for plain-webbed beams in [11]. So, the accuracy of the
equation for the inelastically braced castellated beam CPE140, 5.46 m long, is investigated
here. The approximation of Mk is compared with the accurate predictions obtained using
the finite-element method, as shown in Table 3. The critical buckling moments obtained
by both methods are compared graphically in Fig. 12. As can be seen, there is a good
correlation between them.
Eq. (9) provides a convenient and direct means for designers to estimate the adequacy
of castellated beams under pure bending with an elastic lateral restraint at midspan.
 A. Mohebkhah, H. Showkati / Journal of Constructional Steel Research 61 (2005) 1373–1386 1385

6. Conclusion
A finite-element model for the inelastic nonlinear analysis of castellated beams that
includes the effects of elastic lateral restraints at midspan and large lateral deflections is
developed in this study. The model is used to investigate the effects of central elastic lateral
restraints attached to the top flange on the inelastic flexural–torsional strengths of simply
supported castellated beams under pure bending loading.
It is found that a central elastic lateral restraint generally increases the inelastic strength
of the beam, but that the effect of the restraint depends not only on the stiffness of the
restraint but also on the modified slenderness of the beam. The enhancement in the inelastic
strength increases to some extent as the modified slenderness increases and then decreases
when the beam behaves like an elastic beam. When the modified slenderness of a beam is
very small so that the failure mode of the beam is in-plane at the fully plastic moment, then
the central elastic lateral restraint dose not increase the strength of the beam.
There also exists an optimum value of the stiffness of the central lateral restraint at
which the inelastic strength of the castellated beam is equal to that of the corresponding
beam with a rigid central lateral restraint, which is related to the modified slenderness
of the beam using Eq. (8). The variation of the optimum stiffness values versus modified
slenderness is also similar to that of inelastic strength (optimum buckling moments), as
expressed by Eq. (7). The values of the optimum stiffnesses kopt given by Eq. (8) for both
elastic and inelastic buckling are sufficient to increase the strengths of castellated beams to
the design resistances of beams with rigid lateral restraints. However, owing to the limited
number of castellated beams studied herein, these findings are not yet conclusive, and more
studies should be performed to validate the proposed equations. And also, the other factors
which may affect bracing requirements of the castellated beams are different beam span,
size, opening, loading condition, end restraint, and position of lateral restraint, which must
be taken into consideration for future works.

References
[1] American Institute of Steel Construction (AISC). Load and resistance factor design specification for
structural steel buildings. Chicago (IL): AISC; 1999.
[2] Chen WF, Lui EM. Structural stability: theory and implementation. New York: Elsevier; 1987.
[3] COSMOS/M. User’s guide (volume 1), version 1.5: Structural research and analysis corp. 1st ed. 1995.
[4] Crisfield MA. Non-linear finite element analysis of solids and structures. Chichester (England): John Wiley
& Sons, Ltd; 1995.
[5] Galambos TV. Guide to stability design criteria for metal structures. 5th ed. Structural stability research
council, New York: John Wiley & Sons; 1998.
[6] Kerdal D, Nethercot DA. Failure modes for castellated beams. J Constr Steel Res 1984;4(4):295–315.
[7] Mohebkhah A. Nonlinear lateral–torsional buckling of castellated beams with an elastic lateral restraint
using FEM. MSc thesis, Engineering Faculty, Urmia, Iran: Urmia University; February 2003.
[8] Mohebkhah A. The moment-gradient factor in lateral–torsional buckling on inelastic castellated beams.
J Constr Steel Res 2004;60:1481–94.
[9] Nethercot DA, Kerdal D. Lateral–torsional buckling of castellated beams. Struct Eng, London 1982;60B(3):
53–61.
[10] Showkati H. Theoretical and numerical buckling of study of castellated beams. Industry Committee, Urmia
University; 2002 [in Persian].
[11] Trahair NS. Flexural–torsional buckling of structures. Boca Raton, London: E&FN Spon; 1993.
1386 A. Mohebkhah, H. Showkati / Journal of Constructional Steel Research 61 (2005) 1373–1386

[12] Valentino J, Pi YL, Trahair NS. Inelastic buckling of steel beams with central torsional restraints. J Struct
Eng, ASCE 1997;123(9):1180–6.
[13] Valentino J, Trahair NS. Torsional restraint against elastic lateral buckling. J Struct Eng, ASCE 1998;
124(10):1217–25.
[14] Wang YC, Nethercot DA. Bracing requirements for laterally unrestrained beams. J Constr Steel Res 1990;
17:305–15.
[15] Winter G. Lateral bracing of columns and beams. ASCE Trans 1960;125:809–25.
[16] Yura JA. Fundamentals of beam bracing. Eng J, first quarter 2001;11–26.