University of Plymouth

PEARL https://pearl.plymouth.ac.uk
Faculty of Science and Engineering School of Engineering, Computing and Mathematics

2019-02-25

Bending Analysis of Castellated Beams

Elaiwi, S
http://hdl.handle.net/10026.1/13030

10.30958/ajte.6-1-1
Athens Journal of Technology & Engineering
Athens Institute for Education and Research ATINER

All content in PEARL is protected by copyright law. Author manuscripts are made available in accordance with
publisher policies. Please cite only the published version using the details provided on the item record or
document. In the absence of an open licence (e.g. Creative Commons), permissions for further reuse of content
should be sought from the publisher or author.
To be published in
Athens Journal of Technology & Engineering for Volume 6 Issue 1, March 2019
https://www.athensjournals.gr/ajte/forthcoming
Available online 10 October 2018
https://www.athensjournals.gr/technology/2018-1-X-Y-Elaiwi.pdf

Bending Analysis of Castellated Beams

Sahar .Elaiwi
School of Engineering, Plymouth University,
Plymouth PL4 8AA, UK
[email protected]

Boksun Kim
School of Engineering, Plymouth University,
Plymouth PL4 8AA, UK

Long-yuan Li
School of Engineering, Plymouth University,
Plymouth PL4 8AA, UK

Abstract

Existing studies have shown that the load-carrying capacity of castellated

beams can be influenced by the shear stresses particularly those around web
openings and under the T-section, which could cause the beam to have
different failure modes. This paper investigates the effect of web openings on
the transverse deflection of castellated beams by using both analytical and
numerical methods and evaluates the shear-induced transverse deflection of
castellated beams of different lengths and flange widths subjected to
uniformly distributed transverse load. The purpose of developing analytical
solutions, which adopted the classical principle of minimum potential energy
is for the design and practical use; while the numerical solutions are
developed by using the commercial software ANSYS for the validation of
the analytical solutions.

Keywords: castellated beam; shear effect, deflection; finite element; energy

method.

Introduction
Engineers and researchers have tried various methods to reduce the
material and construction costs to help optimise the use of the steel structural
members. The castellated beam is one of the steel members which uses less
material, but has comparable performance as the I-beam of the same size
(Altifillisch et al., 1957). An example is shown in Figure 1a. The castellated
beam is fabricated from a standard universal I-beam or H-column by cutting
the web on a half hexagonal line down the centre of the beam. The two
halves are moved across by a half unit of spacing and then re-joined by
welding (Harper, 1991). This process increases the depth of the beam and
thus the bending strength and stiffness of the beam about the major axis are
also enhanced without additional materials being added. This allows
castellated beams to be used in long span applications with light or moderate
loading conditions for supporting floors and roofs. In addition, the
fabrication process creates openings on the web, which can be used to
accommodate services. As a result, the designer does not need to increase the
finished floor level. Thus, despite the increase in the beam depth the overall
building height may actually be reduced.
When compared with a solid web solution where services are
provided beneath the beam, the use of castellated beams could lead to
savings in the cladding costs especially in recent years, the steel cost
becomes higher. Owing to the fact that the steel materials have poor fire
resistance, buildings made from steel structures require to use high quality
fireproof materials to protect steel members from fire, which further increase
its cost. Moreover, because of its lightweight the castellated beam is more
convenient in transportation and installation than the normal I-beam.

LITERATURE REVIEW
For many years, the castellated beam have been used in construction
because of its advantages when considering both the safety and serviceability
while considering functional requirements according to the use for which the
construction is intended. Extensive study has been done by researchers who
are working in the construction field to identify the behaviour of castellated
beams when they are loaded with different types of loads. It was found that
the castellated beam could fail in various different modes depending on the
dimensions of the beam and the type of loading as well as the boundary
conditions of the beams. Kerdal and Nethercot (1984) informed the potential
failure modes, which possibly take place in castellated beams. Also, they
explained the reasons for the occurrence of these failure modes. For instance,
shear force and web weld rupture cause a Vierendeel mechanism and web
post-buckling. Additionally, they pointed out that any other failures whether
caused by a flexural mechanism or a lateral-torsional instability is identical
to the equivalent modes for beams without web opening.
The web openings in the castellated beam, however, may reduce the
shear resistance of the beam. The saved evidence, that the method of analysis
and design for the solid beam may not be suitable for the castellated beam
(Boyer, 1964; Kerdal and Nethercot, 1984; Demirdjian, 1999). Design
guidance on the strength and stiffness for castellated beams is available in
some countries. However, again, most of them do not take into account the
shear effect. As far as the bending strength is concerned, neglecting the shear
effect may not cause problems. However, for the buckling and the
calculation of serviceability, the shear weakness due to web openings in
castellated beams could affect the performance of the beams and thus needs
to be carefully considered.
Experimental investigations (Aminian et al., 2012; Maalek, 2004;
Yuan et al., 2014; Yuan et al., 2016; Zaarour & Redwood, 1996 ) were
carried out and finite elements methods (Hosain et al., 1974; Sherbourne and
Van Oostrom, 1972; Soltani et al., 2012; Sonck et al., 2015; Srimani and
Das, 1978; Wang et al., 2014) were also used to predict the deflection of
castellated beams and/or to compare the predictions with the results from the
experiments. The experimental findings (Zaarour and Redwood, 1996)
demonstrated the possibility of the occurrence of the buckling of the web
posts between web openings. The shear deflection of the straight-sided
tapering cantilever of the rectangular cross section (Maalek, 2004) was
calculated by using a theoretical method based on Timoshenko’s beam
theory and virtual work method. Linear genetic programming and integrated
search algorithms (Aminian et al., 2012) showed that the use of the machine
learning system is an active method to validate the failure load of castellated
beams. A numerical computer programme (Sherbourne and Van Oostrom,
1972) was developed for the analysis of castellated beams considering both
elastic and plastic deformations by using practical lower limit relationships
for shear, moment and axial force interaction of plasticity. An analysis on
five experimental groups of castellated beams (Srimani and Das, 1978) was
conducted to determine the deflection of the beam. It was demonstrated
(Hosain et al., 1974) that the finite elements method is a suitable method for
calculating the deflection of symmetrical section castellated beams. The
effect of nonlinearity in material and/or geometry on the failure model
prediction of castellated beams (Soltani et al., 2012) was done by using
MSC/NASTRAN software to find out bending moments and shear load
capacity, which are compared with those published in literature.
Axial compression buckling of castellated columns was investigated
(Yuan et al., 2014), in which an analytical solution for critical load is derived
based on stationary potential energy and considering the effect of the web
shear deformations on the flexural buckling of simply supported castellated
column. Recently, a parametric study on the large deflection analysis of
castellated beams at high temperatures (Wang, et al., 2014) was conducted
by using finite element method to calculate the growth of the end reaction
force, the middle span deflection, and the bending moments at susceptible
sections of castellated beams. More recently, a comprehensive comparison
between the deflection results of cellular and castellated beams obtained
from numerical analysis (Sonck et al., 2015) was presented, which was
obtained from different simplified design codes. The comparison showed
that the design codes are not accurate for short span beams and conservative
for long span beams. The principle of minimum potential energy was
adopted (Yuan et al., 2016) to derive an analytical method to calculate the
deflection of castellated/cellular beams with hexagonal/circular web
openings, subjected to a uniformly distributed transverse load.
The previous research efforts show that there were a few of articles
that dealt with the deflection analysis of castellated beams. Due to the
geometric particulars of the beam, however, it was remarkable to note that
most of the theoretical approximate methods are interested in calculating the
deflection of the castellated beams for long span beams where the shear
effect is negligible. However, the castellated beams/columns are used not
only for long span beams/columns but also for short beams/columns. Owing
to the complex of section profile of the castellated beams, the shear-effect
caused by the web opening on the deflection calculation is not fully
understood. There are no accurate calculation methods available in literature
to perform these analyses. Thus it is important to know how the shear affects
the deflection of the beam and on what kind of spans the shear effect can be
ignored. In addition, researchers have adopted the finite elements method to
predict the deflection of castellated beams by using different software
programs such as MSC/NASTRAN, ABAQUS, and ANSYS. However,
these programs need efficiency in use because any error could lead to
significant distortions in results. European building standards do not have
formulas for the calculation of deflections of castellated beams, which
include shear deformations.
This paper presents the analytical method to calculate the elastic
deflection of castellated beams. The deflection equation is to be developed
based on the principle of minimum potential energy. In order to improve the
accuracy and efficiency of this method, shear rigidity factor is determined by
using suitable numerical techniques. The analytical results were validated by
using the numerical results obtained from the finite element analysis using
ANSYS software.

Analytical Philosophy of Deflection Analysis of Castellated Beams

An approximate method of deflection analysis of castellated beams
under a uniformly distributed transverse load is presented herein. The
method is derived based on the principle of minimum potential energy of the
structural system. Because of the presence of web openings, the cross-section
of the castellated beam is now decomposed into three parts to calculate the
deflection and bending stress, two of which represent the top and bottom T-
sections, one of which represents the mid-part of the web. The analysis
model is illustrated in Figure 1a, in which the flange width and thickness are
bf and tf, the web depth and thickness are hw and tw, and the half depth of
hexagons is a. The half of the distance between the centroids of the two T-
sections is e. In this study, the cross-section of the castellated beam is
assumed to be doubly symmetrical. Under the action of a uniformly
distributed transverse load, the beam section will have axial and transverse
displacement as shown in Figure 1b, where x is the longitudinal coordinate of
the beam, z is the cross-sectional coordinate of the beam, (u1, w) and (u2, w)
are the axial displacements and the transverse displacements of the centroids
of the upper and lower T-sections. All points on the section are assumed to
have the same transverse displacement because of the beam assumption used
in the present approach (Yuan et al., 2014). The corresponding axial strains
1x in the upper T-section and 2x in the lower T-section are linearly
distributed and can be determined by using the strain-displacement relation
as follows:

hw
In the upper T-section： − ( + t f ) ≤ z ≤ −a
2
2
𝑑𝑢1 𝑑 𝑤
𝜀1𝑥 𝑥, 𝑧 = − (𝑧 + 𝑒) (1)
𝑑𝑥 𝑑𝑥 2
hw
In the lower T-section：a ≤ z ≤ ( + tf )
2
𝑑𝑢2 𝑑2 𝑤
𝜀2𝑥 𝑥, 𝑧 = − (𝑧 − 𝑒) (2)
𝑑𝑥 𝑑𝑥 2
The shear strain γxz in the middle part between the two T-sections can also be
determined using the shear strain-displacement relation as follows:
For the middle part between the two T-sections: −a ≤ z ≤ a

𝑑𝑢 𝑑𝑤 𝑢1 − 𝑢2 𝑒 𝑑𝑤
𝛾𝑥𝑧 𝑥, 𝑧 = + =− + (3)
𝑑𝑧 𝑑𝑥 2𝑎 𝑎 𝑑𝑥
ℎ𝑤 +𝑡𝑓 ℎ ℎ + 2𝑎
𝑏𝑓 𝑡𝑓 +𝑡𝑤 ( 𝑤 − 𝑎) ( 𝑤 )
2 2 4
𝑒= (4)
ℎ𝑤
𝑏𝑓 𝑡𝑓 +𝑡𝑤 ( 2 − 𝑎)
Because the upper and lower T-sections behave according to Bernoulli's
theory, the strain energy of the upper T-section U1 and the lower T-section
U2 caused by a transverse load can be expressed as follows:
ℎ
𝑙 − 𝑤 𝑙 −𝑎
2
𝐸𝑏𝑓 2
𝐸𝑡𝑤 2
𝑈1 = 𝜀1𝑥 𝑑𝑧𝑑𝑥 + 𝜀1𝑥 𝑑𝑧𝑑𝑥
2 2
0 −(𝑡 +ℎ 𝑤 ) 0 −(ℎ 𝑤 ) (5)
𝑓 2 2
𝑙 2
2
1 𝑑𝑢1 𝑑2 𝑤
= 𝐸𝐴𝑡𝑒𝑒 + 𝐸𝐼𝑡𝑒𝑒 𝑑𝑥
2 𝑑𝑥 𝑑𝑥 2
0
ℎ𝑤 ℎ𝑤
𝑙 (2 ) 𝑙 (𝑡 𝑓 + 2 )
𝐸𝑡𝑤 2
𝐸𝑏𝑓 2
𝑈2 = 𝜀2𝑥 𝑑𝑧𝑑𝑥 + 𝜀2𝑥 𝑑𝑧𝑑𝑥
2 2
0 𝑎 0 ℎ𝑤
2 (6)
𝑙 2
2 2
1 𝑑𝑢2 𝑑 𝑤
= 𝐸𝐴𝑡𝑒𝑒 + 𝐸𝐼𝑡𝑒𝑒 𝑑𝑥
2 𝑑𝑥 𝑑𝑥 2
0

where E is the Young's modulus of the two T-sections, G is the shear

modulus, Atee and Itee are the area and the second moment of area of the
T- section, which are determined in their own coordinate systems as follows:
ℎ𝑤
𝐴𝑡𝑒𝑒 = 𝑏𝑓 𝑡𝑓 + 𝑡𝑤 −𝑎
2 (7)
 2 3
𝑏𝑓 𝑡𝑓3 ℎ𝑤 +𝑡𝑓 𝑡𝑤 ℎ𝑤
𝐼𝑡𝑒𝑒 = + 𝑏𝑓 𝑡𝑓 −𝑒 + −𝑎
12 2 12 2 (8)
2
ℎ𝑤 ℎ𝑤 + 2𝑎
+ 𝑡𝑤 −𝑎 −𝑒
2 4
The mid-part of the web of the castellated beam, which is illustrated in
Figure 1a, is assumed to behave according to Timoshenko’s theory (Yuan et
al., 2014). Therefore, its strain energy due to the bending and shear can be
expressed as follows:
1
𝑈𝑏 = 𝐾𝑏 ∆2 (9)
2
where ∆ is the relative displacement of the upper and lower T-sections due to
a pair of shear forces and can be expressed as (∆ = 2aγxz). While Kb is the
combined stiffness of the mid part of the web caused by the bending and
shear, and is determined in terms of Timoshenko beam theory as follows,
1 3𝑙𝑏 𝑙𝑏3
= + (10)
𝐾𝑏 2𝐺𝐴𝑏 12𝐸𝐼𝑏
Where Ab=√3atw is the equivalent cross-sectional area of the mid part of the
web, Ib= (√3a)3tw/12 is the second moment of area, and lb = 2a is the length
of the Timoshenko beam; herein representing the web post length. Note that,
the Young's modulus of the two T-sections is E=2(1+ν)G and the Poisson’s
ratio is taken as v =0.3, the value of the combined stiffness of the mid part of
the web caused by the bending and shear can be determined as fallow:
3𝐺𝑡𝑤
𝐾𝑏 = (11)
4
Thus, the shear strain energy of the web, Ush, due to the shear strain γxy can
be calculated as follows:
𝑛 𝑙 𝑙
3 3𝐺𝑡𝑤 𝑎2 𝐺𝑡𝑤 𝑎
𝑈𝑠ℎ = 𝐺𝑡𝑤 𝑎2 2
𝛾𝑥𝑧 ≈ 2
𝛾𝑥𝑧 𝑑𝑥 = 2
𝛾𝑥𝑧 𝑑𝑥 (12)
2 6𝑎 4
𝑘=1 2×
3 0 0

Let the shear rigidity factor ksh = 0.25. Substituting Eqs (3) into (12) gives the
total shear strain energy of the mid-part of the web:
𝑙
2
𝐺𝑡𝑤 𝑒 2 𝑘𝑠ℎ 𝑑𝑤 𝑢𝛽
𝑈𝑠ℎ = − 𝑑𝑥 (13)
𝑎 𝑑𝑥 𝑒
0

Figure 1. (a) Notations used in castellated beams, (b) displacements and (c)
internal forces.
Note that, in the calculation of shear strain energy of Eq.(12) one uses the
concept of smear model, in which the shear strain energy was calculated first
for web without holes. Then by assuming the ratio of the shear strain
energies of the webs with and without holes is proportional to the volume
ratio of the webs with and without holes, the shear strain energy of the web
with holes was evaluated, in which ksh = 0.25 was obtained (Kim et al.,
2016). However, by using a two-dimensional linear finite element analysis
(Yuan et al., 2016) the value of the combined stiffness of the mid part of the
web of the castellated beam caused by the bending and shear was found to be
3𝐺𝑡𝑤
𝐾𝑏 = 0.78 × (14)
4
which is smaller than that above-derived from the smear model. This leads to
the shear rigidity factor ksh = 0.78x0.25. The reason for this is probably due
to the smear model used for the calculation of the shear strain energy for the
mid-part of the web in Eq. (12).
However, it should be mentioned that the factor of 0.78 in Eq. (14) was
obtained for only one specific section of a castellated beam. It is not known
whether this factor can also be applied to other dimensions of the beams. A
finite element analysis model for determining the shear rigidity factor ksh is
therefore developed herein (see Figure 2c), in which the length and depth of
the unit are (4a/√3) and (2a+a/2), respectively. In the unit the relative
displacement ∆ can be calculated numerically when a unit load F is applied
(see Figure 2c). Hence, the combined rigidity Kb=1/∆ is obtained. Note that
in the unit model all displacements and rotation of the bottom line are
assumed to be zero, whereas the line where the unit load is applied is
assumed to have zero vertical displacement. The calibration of the shear
rigidity for beams of different section sizes shows that the use of the
expression below gives the best results and therefore Eq. (15) is used in the
present analytical solutions.
𝑏𝑓 1
𝐾𝑠ℎ = 0.76 − × (15)
𝑙 4
where l is the length of the beam. Thus the total potential energy of the
castellated beam UT is expressed as follows,
𝑈𝑇 = 𝑈1 + 𝑈2 + 𝑈𝑠ℎ (16)
For the simplicity of presentation, the following two new functions are
introduced:
u1  u2
u  (17)
2
u1  u2
u 
2 (18)
By using Eqs. (17) and (18), the total potential energy of the castellated beam
subjected to a uniformly distributed transverse load can be expressed as
follows:
𝑙 2 𝑙 2
𝑑𝑢𝛽 𝑑2 𝑤
∏ = 𝐸𝐴𝑡𝑒𝑒 𝑑𝑥 +𝐸𝐼𝑡𝑒𝑒 𝑑𝑥
𝑑𝑥 𝑑𝑥 2
0 0 (19)
𝑙
2 2
𝐺𝑡𝑤 𝑒 𝑘𝑠ℎ 𝑑𝑤 𝑢𝛽
+ − 𝑑𝑥 − 𝑊
𝑎 𝑑𝑥 𝑒
0
where W is the potential of the uniformly distributed load qmax due to the
transverse displacement, which can be expressed as follows:
𝑙

W = 𝑞𝑚𝑎𝑥 𝑤 𝑑𝑥 (20)
0
where qmax is the uniformly distributed load, which can be expressed in terms
of design stress σy, as follows:

𝜎𝑦 𝐼𝑟𝑒𝑑𝑢𝑐𝑒𝑑
𝑞𝑚𝑎𝑥 = 16 2 (21)
𝑙 (ℎ𝑤 + 2𝑡𝑓 )
3 3
𝑏𝑓 ℎ𝑤 + 2𝑡𝑓 𝑡𝑤 𝑎3 ℎ𝑤 𝑏𝑓 − 𝑡𝑤 (22)
𝐼𝑟𝑒𝑑𝑢𝑐𝑒𝑑 = − −
12 12 12

Figure 2 Shear strain energy calculation model: (a) unit considered, (b) shear
deformation calculation model and (c) finite element model of 4a/√3 length
unit and (2a+a/2) depth, loaded by a unite force F.

Deflection of Simply Supported Castellated Beam with Uniformly

Distributed Transverse Loading

For a simply supported castellated beam uα(x), uβ(x) and w(x) can be
assumed as follows:
𝑚𝜋𝑥
𝑢𝛼 (𝑥) = 𝐴𝑚 cos (23)
𝑙
𝑚 =1,2,..

𝑚𝜋𝑥
𝑢𝛽 (𝑥) = 𝐵𝑚 cos (24)
𝑙
𝑚=1,2,..

𝑚𝜋𝑥
𝑤(𝑥) = 𝐶𝑚 sin (25)
𝑙
𝑚 =1,2,..

where Am, Bm and Cm are the constants to be determined. It is obvious that the
displacement functions assumed in Eqs. (23)-(25) satisfy the simply support
d 2w du du 
boundary conditions, that are w 0 and  0 at x = 0 and x = l,
dx 2 dx dx
and m = 1,2,… is the integral number. Substituting Eqs. (23), (24) and (25)
into (19) and (20) and according to the principle of minimum potential
energy, it yields,

𝛿 𝑈𝑇 + 𝑈𝑠ℎ − 𝑊 = 0 (26)

The variation of Eq.(26) with respect to Am, Bm and Cm results in following

three algebraic equations:

𝑚𝜋𝑥 2 (27)
𝐸𝐴𝑡𝑒𝑒 ( ) 𝐴𝑚 = 0
𝑙
𝑚𝜋 2 𝐺𝑡𝑤 𝑘𝑠ℎ 𝐺𝑡𝑤 𝑒𝑘𝑠ℎ 𝑚𝜋
𝐸𝐴𝑡𝑒𝑒 ( ) + 𝐵𝑚 − ( ) 𝐶𝑚 = 0 (28)
𝑙 𝑎 𝑎 𝑙
𝑚𝜋 4 𝐺𝑡𝑤 𝑒 2 𝑘𝑠ℎ 𝑚𝜋 2 𝐺𝑡𝑤 𝑒𝑘𝑠ℎ 𝑚𝜋
𝐸𝐼𝑡𝑒𝑒 ( ) + ( ) 𝐶𝑚 − ( ) 𝐵𝑚
𝑙 𝑎 𝑙 𝑎 𝑙
1 − −1 𝑞𝑚𝑎𝑥𝑚 (29)
=
𝑚𝜋

Mathematically Eqs (27) -(29) lead to:

𝐴𝑚 = 0 (30)

𝐺𝑡𝑤 𝑒𝑘𝑠ℎ 𝑚𝜋
𝑎 ( )
𝑙
𝐵𝑚 = 𝐶 (31)
𝑚𝜋 2 𝐺𝑡𝑤 𝑘𝑠ℎ 𝑚
𝐸𝐴𝑡𝑒𝑒 ( ) + 𝑎
𝑙

1 − (−1)𝑚 𝑞𝑙 4
𝐶𝑚 =
𝑚𝜋 5 𝑒 2 𝐸𝐴𝑡𝑒𝑒 (32)
𝐸𝐼𝑡𝑒𝑒 + 2
𝐸𝐴𝑡𝑒𝑒 𝑎 𝑚𝜋
1+
𝐺𝑘𝑠ℎ 𝑡𝑤 𝑙 2

Therefore, the deflection of the castellated beam can be expressed as follows:

𝑞𝑙 4 2 𝑒 2 𝐴𝑡𝑒𝑒
𝑤(𝑥) = 1+
𝐸 𝐼𝑡𝑒𝑒 + 𝑒 2 𝐴𝑡𝑒𝑒 𝑚𝜋 5 𝐼𝑡𝑒𝑒 + 𝑒 2 𝐴𝑡𝑒𝑒
𝑚 =1,2,..
2
(33)
𝐸𝐴𝑡𝑒𝑒 𝑎 𝑚𝜋 𝐸𝐼𝑡𝑒𝑒 𝑎 𝑚𝜋 2 𝑚𝜋𝑥
× 1− sin
𝐺𝑘𝑠ℎ 𝑡𝑤 𝑙 2 𝐺𝑘𝑠ℎ 𝑡𝑤 𝑙 2 𝑒 2 𝑙

The maximum deflection of the simply supported beam is at the mid of the
beam, that is x=l/2 and thus it can be expressed as follows:
 𝑞𝑙 4 2 −1 𝑘+1 𝑒 2 𝐴𝑡𝑒𝑒 𝐸𝐴𝑡𝑒𝑒 𝑎
𝑤|𝑥=𝑙/2 = + ×
𝐸 𝐼𝑡𝑒𝑒 + 𝑒 2 𝐴𝑡𝑒𝑒 𝜋 5 2𝑘 − 1 5 𝐼𝑡𝑒𝑒 + 𝑒 2 𝐴𝑡𝑒𝑒 𝐺𝑘𝑠ℎ 𝑡𝑤 𝑙 2
𝑘=1,2,..
(34)
2 −1 𝑘+1 𝐸𝐼𝑡𝑒𝑒 𝑎 2 −1 𝑘+1
× 2 3
−
𝜋 2𝑘 − 1 𝐺𝑘𝑠ℎ 𝑡𝑤 𝑙 2 𝑒 2 𝜋 2𝑘 − 1
𝑘=1,2,.. 𝑘=1,2,..

Note that, mathematically, the following equations hold,

2 −1 𝑘+1 5
= (35)
𝜋 5 2𝑘 − 1 5 2 × 384
𝑘=1,2,..

2 −1 𝑘+1 1
3 3
= (36)
𝜋 2𝑘 − 1 16
𝑘=1,2,..

2 −1 𝑘+1 1
= (37)
𝜋 2𝑘 − 1 2
𝑘=1,2,..

Using Eqs (35), (36) and (37), the maximum deflection of the beam can be
simplified as follows:
2
5𝑞𝑙 4 𝑞𝑙 2 𝑎 𝑒𝐴𝑡𝑒𝑒
𝑤|𝑥=𝑙/2 = + ×
384𝐸 2𝐼𝑡𝑒𝑒 + 2𝑒 2 𝐴𝑡𝑒𝑒 16𝐺𝑘𝑠ℎ 𝑡𝑤 𝐼𝑡𝑒𝑒 + 𝑒 2 𝐴𝑡𝑒𝑒
(38)
2𝐸𝐼𝑡𝑒𝑒 𝑎
× 1−
𝐺𝑘𝑠ℎ 𝑡𝑤 𝑙 2 𝑒 2
It is clear from Eq. (38) that, the first part of Eq.(38) represents the deflection
generated by the bending load, which is deemed as that given by Bernoulli-
Euler beam, while the second part of Eq.(38) provides the deflection
generated by the shear force. Moreover, Eq. (38) shows that the shear-
induced deflection is proportional to the cross-section area of the two T-
sections but inversely proportional to the beam length. This explains why the
shear effect could be ignored for long span beams.
If the calculation does not consider the shear effect of web openings, Eq. (38)
reduces to the following bending deflection equation.
5𝑞𝑙 4
𝑤|𝑥=𝑙/2 = (39)
384𝐸𝐼𝑟𝑒𝑑𝑢𝑐𝑒𝑑

Numerical Study
In order to validate the abovementioned analytical solution numerical
analysis using the finite element method is also carried out. The numerical
computation uses the ANSYS Programming Design Language (APDL). The
FEA modelling of the castellated beams is carried out by using 3D linear
Quadratic 4-Node thin shell elements (SHELL181). This element presents
four nodes with six DOF per node, i.e., translations and rotations on the X,
Y, and Z axis, respectively. Half-length of the castellated beams is used
because of the symmetry in geometry. The lateral and transverse deflections
and rotation are restrained (uy=0, uz=0 and θx=0) at the simply supported end,
while the symmetrical boundary condition is applied at the other end by
constraining the axial displacement and rotations around the two axes within
the cross-section (ux=0, θy=0 and θz=0). The material properties of the
castellated beam are assumed to be linear elastic material with Young’s
modulus E = 210 GPa and Poisson’s ratio v =0.3.
A line load effect is used to model applied uniformly distribution load,
where the load is assumed acting on the junction of the flange and the web.
The equivalent nodal load is calculated by multiply the distribution load with
beam’s half-length and then divided by the number of the nodes on the
junction line of the flange and the web.

Discussions
Figure 3 shows a comparison of the maximum deflations between
analytical solutions using different shear rigidity factors including one with
zero shear factor and FEA numerical solution for four castellated beams of
different flange widths. It can be seen from the figure that, the analytical
solution using the proposed shear factor is closest to the numerical solution,
whereas the analytical solutions using other shear factors is not as good as
the present one. This demonstrates that the shear factor is also affected by the
ratio of the flange width to the beam length. Also, it can be seen from the
figure that, the longer the beam, the closer the analytical solution to the
numerical solution; and the wider the flanges, the closer the analytical
solution to the numerical solution. Figure 4 shows the relative error of each
analytical solution when it is compared with the finite element solution.
From the figure it is evident that the error of the analytical solutions using
the present shear rigidity factor does not exceed 6.0% for all of discussed
four sections in all the beam length range (>3 meter). In contrast, the
analytical solution ignoring the shear effect, or considering the shear effect
by using smear model or by using the length-independent shear rigidity
factor will have large error, particularly when the beam is short.

Conclusion
This study has reported the theoretical and numerical solutions for
calculating the deflection of hexagonal castellated beams with simply
supported boundary condition, subjected to a uniformly distributed
transverse load. The analysis is based on the total potential energy method,
by taking into account the influence of web shear deformations. The main
novelty of the present analytical solution for the calculation of deflection is it
considers the shear effect of web openings more accurately. Both the
analytical and numerical solutions are employed for a wide spectrum of
geometric dimensions of I-shaped castellated beams in order to evaluate the
analytical results.
From the present study, the main conclusions can be summarized as follows:
1- The present analytical results are in excellent agreement with those
obtained from the finite element analysis, which demonstrates the
appropriateness of proposed approach.
2- Shear effect on the deflection of castellated beams is very important,
particularly for short and medium length beams with narrow or wide
section. Ignoring the shear effect could lead to an under-estimation of the
deflection.
3- Divergence between analytical and numerical solutions does not exceed
6.0% even for short span castellated beam with narrow or wide section.
4- The effect of web shear on the deflection reduces when castellated beam
length increases.
5- Despite that the numerical solution based on FEA has been widely used
in the analysis of castellated beams; it is usually time-consuming and
limited to specific geometrical dimensions. Thus, a simplified calculation
solution that is able to deliver reasonable results but requires less
computational effort would be helpful for both researchers and designers.

Figure3. Maximum deflections of simply supported castellated beams with

uniformly distributed load obtained using analytical solution with different
shear rigidity factors (Eqs. (38) and (39)) and FEA numerical solution for
four castellated beams of different flange widths. (a) bf=100mm, (b)
bf=150mm, (c) bf=200mm and (d) bf=250mm (hw=300mm, tf=10mm,
tw=8mm and a=100mm).
Figure.4. Divergence of maximum deflections of simply supported
castellated beams with uniformly distributed load obtained using analytical
solution with different shear rigidity factors (Eqs. (38) and (39)) and FEA
numerical solution for four castellated beams of different flange widths. (a)
bf=100mm, (b) bf=150mm, (c) bf=200mm and (d) bf=250mm (hw=300mm,
tf=10mm, tw=8mm and a=100mm).

Acknowledgments

The first author wishes to thank the Ministry of Higher Education in

Iraq Trust for funding her PhD study in the University of Plymouth.

References

Altifillisch, M. D., Cooke, R. B., & Toprac, A. A. (1957). An investigation of open

web expanded beams. Welding Research Council Bulletin,New York, 47, 307-
320.
Aminian, P., Niroo mand, H., Gandomi, A. H., Alavi, A. H., & Arab Esmaeili, M.
(2012). New design equations for assessment of load carrying capacity of
 castellated steel beams: a machine learning approach. Neural Computing and
Applications, 23(1), 119-131. http://doi:10.1007/s00521-012-1138-4.
Boyer, J. P. (1964). Castellated Beam- A New Development Castellated beams-new
developments. AISC Engineering 1(3), 104-108.
Demirdjian, S. (1999). Stability of castellated beam webs. (PhD), McGill University
Montreal, Canada.
Hosain, M., Cheng, W., & Neis, V. (1974). Deflection analysis of expanded open-
web steel beams. Computers & Structures, 4(2), 327-336.
Kerdal, D., & Nethercot, D. (1984). Failure modes for castellated beams. Journal of
Constructional Steel Research, 4(4), 295-315.
Kim, B., Li, L.-Y., & Edmonds, A. (2016). Analytical Solutions of Lateral–
Torsional Buckling of Castellated Beams. International Journal of Structural
Stability and Dynamics, 1550044. http://doi:10.1142/s0219455415500443.
Maalek, S. (2004). Shear deflections of tapered Timoshenko beams. International
Journal of Mechanical Sciences, 46(5), 783-805. http://doi:10.1016/j.ijmecsci.
2004.05.003.
Sherbourne, A., & Van Oostrom, J. (1972). Plastic analysis of castellated beams—I
interaction of moment, shear and axial force. Computers & Structures, 2(1),
79-109.
Soltani, M. R., Bouchaïr, A., & Mimoune, M. (2012). Nonlinear FE analysis of the
ultimate behavior of steel castellated beams. Journal of Constructional Steel
Research, 70, 101-114. http://doi:10.1016/j.jcsr.2011.10.016.
Sonck, D., Kinget, L., & Belis, J. (2015). Deflections of cellular and castellated
beams. Paper presented at the Future Visions (International Association for
Shell and Spatial Structures)(IASS2015).
Srimani, S. S., & Das, P. (1978). Finite element analysis of castellated beams.
Computers & Structures, 9(2), 169-174.
Wang, P., Wang, X., & Ma, N. (2014). Vertical shear buckling capacity of web-
posts in castellated steel beams with fillet corner hexagonal web openings.
Engineering Structures, 75, 315-326.
http://doi:10.1016/j.engstruct.2014.06.019.
Yuan, W.-b., Kim, B., & Li, L.-y. (2014). Buckling of axially loaded castellated
steel columns. Journal of Constructional Steel Research, 92, 40-45.
http://doi:10.1016/j.jcsr.2013.10.013.
Yuan, W.-b., Yu, N.-t., Bao, Z.-s., & Wu, L.-p. (2016). Deflection of castellated
beams subjected to uniformly distributed transverse loading. International
Journal of Steel Structures, 16(3), 813-821.
Zaarour, W., & Redwood, R. (1996). Web buckling in thin webbed castellated
beams. Journal of Structural Engineering, 122(8), 860-866.