Analysis and Design of Castellated Steel Beam - 2019 Elaiwi PHD

University of Plymouth
PEARL https://pearl.plymouth.ac.uk
04 University of Plymouth Research Theses 01 Research Theses Main Collection
2019
ANALYSIS AND DESIGN OF

CASTELLATED BEAMS
Elaiwi, Sahar Sahib

http://hdl.handle.net/10026.1/15061
University of Plymouth
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.
ANALYSIS AND DESIGN OF CASTELLATED BEAMS
By
SAHAR SAHIB ELAIWI
A thesis submitted to the University of Plymouth

in partial fulfilment for the degree of
DOCTOR OF PHILOSOPHY
School of Engineering
October 2019
Copyright Statement
This copy of the thesis has been supplied on condition that anyone
who consults it is understood to recognise that its copyright rests with its
author and that no quotation from the thesis and no information derived
from it may be published without the author's prior consent.
ANALYSIS AND DESIGN OF CASTELLATED BEAMS
Sahar Sahib Elaiwi, Doctor of Philosophy, 2019
School of Engineering, University of Plymouth
ABSTRACT
The castellated beam is one of the steel members which uses less material, but has equal
performance as the I-beam of the same size. Most of the castellated beams are 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, then these two halves are moved directly across by a half unit of
spacing and re-joined by welding. This process leads to increasing the beam’s depth and thus
the bending strength and stiffness about the major axis, without adding additional materials.
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.
Existing studies have shown that the resistance of the castellated beam is influenced by shear
stresses, particularly those around web openings and under the T-section, which could cause
the beam to have different failure modes. Therefore, web openings may reduce the shear
resistance of the castellated beam. However, previous studies confirm that the method of
analysis and design for the solid beam, may not be suitable for the castellated beam. Design
guidance on the strength and stiffness for castellated beams is available in most countries, but
also, some of them do not take into account the shear effect because as far as the bending
strength is concerned, neglecting the shear effect may not cause direct problems. However,
for the buckling and 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
reconsidered.
The aims of this study are, firstly to investigate the effect of web openings on the transverse
deflection and lateral-torsional buckling of castellated beams, by using both analytical and
numerical methods, whilst also adopting some of the previous studies for validating the
i
analytical results. The second aim, is to focus on the effect of both the geometric nonlinear
and material inelasticity on castellated beams, under uniformly distributed load within
different boundary conditions, through investigating the load-deflection curve and predicting
the value of lateral-torsional buckling moment capacity, by using nonlinear numerical
analysis method. The third aim, is to study the free vibration, static buckling and dynamic
instability of castellated beams, w subjected to transverse periodic loading by developing
analytical solutions.
The purpose of developing analytical solutions, which adopt the classical principle of
minimum potential energy, is for the design and practical use; while the numerical solutions
developed using the commercial software ANSYS are for the validation of the analytical
solutions. By examining the results of analytical and numerical solutions obtained, a number
of important features of the castellated beams are identified. This study has contributed to
enhancing the knowledge of the effect of web openings on the performance of castellated
beams subjected to uniformly distributed transverse loads with/or without being exposed to
elevated temperatures. Data provided in these studies, can essentially be used by structural
designers for providing better, economical and safe structures.
Keywords: castellated beam; shear effect, transvers deflection; lateral torsional buckling;
dynamic analysis; finite element; linear and nonlinear; energy method; uniformly distributed
loads; non-uniform temperature; simply supported; pinned-fixed supported.
ii
ACKNOWLEDEMENTS
Almost at the completion of my Ph. D study, my sincere thanks and gratitude are due to His
Almighty ALLAH, who helped me and blessed my work during the days of my study and
research.
It gives me great pleasure to express my gratitude to everyone who contributed to the

completion of this thesis.
First of all, it was my pleasure to study for four years under my supervisor, Professor Long
Yuan Li. I'm so grateful for his guidance, support, continuous encouragement and the
countless amount of time spent helping me during these years. My thesis was built with his
valuable advice and sharp insights, as well as his patience throughout my Ph.D. I would like
to thank him so much for everything he has done to help me to reach my goal.
I would also like to give my sincere appreciation and thanks to my second supervisor, Dr
Boksun Kim for her progressive thinking and her open mind. Her continuous advice and
significant comments helped me to develop my work successfully.
I would like to thank the Ministry of Higher Education in Iraq for funding my Ph. D study at
University of Plymouth.
To my parents: I want to tell them of my thanks for all times they have supported me and
made me strong enough to reach this stage of my life. May Allah continuously bless them
with good health. To my sisters and brothers, I would like to say thanks for your feelings and
support. To my husband and my lovely children Ahmed and Noor, who gave me power and
patience during the last four years of study, I thank them from the bottom of my heart,
without whom my goal would not have been achieved. I dedicate this work to my family.
iii
I'm so glad to have studied at the University of Plymouth. During my time in Plymouth city,
I have gained a lot of friends, and studying there was like being in my hometown. My sincere
gratitude to everyone for all the encouragement during my study. I want to tell all of them,
thank you so much for everything, you have helped me.
iv
AUTHOR’S DECLARATION
At no time during the registration for the degree of Doctor of Philosophy has the author been
registered for any other University award without prior agreement of the Doctoral College
Quality Sub-Committee.
Work submitted for this research degree at the University of Plymouth has not formed part of
any other degree either at University of Plymouth or at another establishment.
This study was financed with the aid of a studentship form the Ministry of Higher Education
and Scientific Research of Republic of Iraq.
Paper presentations were given at relevant conferences and several papers have been
published.
Word count of main body of thesis: 34,956 words.
Figure count of the thesis: 69 figures.
Table count of the thesis: 16 tables.
Sahar S Elaiwi 7/oct/2019

Signed ………………………………..Date…………………………………
v
PAPERS AND CONFERENCES
This thesis is based on the following papers published in Journals/ conferences, which will be
referred to in the text by their Roman numerals.
I. Elaiwi, S., Kim, B. and Li, L. (2017). "Analysis and Design of Castellated Beams."
The 2nd Annual International Conference on Structural Engineering and Mechanics
19-22 June, Athens, Greece. (Oral Presentation).
II. Elaiwi, S., Kim, B. and Li, L. (2017). "Bending Analysis of Continuous Castellated
Beams." World Congress on Advances in Structural Engineering and Mechanics
(ASEM17) 28Augest-1September, llsan (Seoul), Korea. (Oral Presentation).
III. Elaiwi, S., Kim, B. and Li, L. (2018). "Linear and Nonlinear Buckling Analysis of
Castellated Beams." Eighth International Conference on Engineering Failure Analysis
8th-11th July, Budapest, Hungary. (poster Presentation)
IV. Elaiwi, S., Kim, B. and Li, L. (2018). "Bending Analysis of Continuous Castellated
Beams." Proceedings (pp. 1-14) of the 2017 World Congress on Advances in
Structural Engineering and Mechanics (ASEM17) 28Augest-1September, 2017, llsan
(Seoul), Korea.
V. Elaiwi, S., Kim, B. and Li, L. (2019). "Bending Analysis of Castellated Beams."
Athens Journal of Technology and Engineering, 6(1): 1-16.
DOI: https:// doi=10.30958/ajte.6-1-1
PEARL (OA): http://hdl.handle.net/10026.1/13030
VI. Elaiwi, S., Kim, B. and Li, L. (2019). "Dynamic instability of castellated beams
subjected to transverse periodic loading." Challenge Journal of Structural Mechanics,
5(1): 9-18. ISSN: 2149-8024.
DOI: http://dx.doi.org/10.20528/cjsmec.2019.01.002
vi
VII. Elaiwi, S., Kim, B. and Li, L. (2019). "Linear and Nonlinear Buckling Analysis of
Castellated Beams." International Journal of Structural and Civil Engineering
Research, 8(2): ISSN: 2319-6009.
DOI: http://dx.doi.org/10.18178/ijscer.8.2.83-93
vii
CONTENTS
LIST OF CONTENENTS
ABSTRACT ............................................................................................................................... i
ACKNOWLEDEMENTS ....................................................................................................... iii
AUTHOR’S DECLARATION.................................................................................................v
PAPERS AND CONFERENCES .......................................................................................... vi
LIST OF CONTENENTS .................................................................................................... viii
LIST OF TABLES ..................................................................................................................xv
LIST OF FIGURES ............................................................................................................. xvii
CHAPTER ONE.........................................................................................................................1
1. INTRODUCTION .................................................................................................................1
1.1. Motivation and background ..........................................................................................1
1.2. The problem statement and objectives..........................................................................4
1.3. Methodology .................................................................................................................6
1.4. Organization of thesis ...................................................................................................8
CHAPTER TWO ......................................................................................................................11
2. LITERATURE REVIEW ....................................................................................................11
2.1. History of development of fabrication method ........................................................... 11
2.2. The failure modes of castellated beams ......................................................................12
2.3. Local failure modes ....................................................................................................13
2.3.1. Vierendeel failure mode....................................................................................13
2.3.2. Web post failure ................................................................................................15
2.3.3. Web post buckling ............................................................................................15
viii
CONTENTS
2.3.3.1. Compression buckling of web post ..................................................... 16
2.3.3.2. Bending of web post failure due to moments in web-posts ................ 16
2.3.3.3. Web welds failure ............................................................................... 17
2.3.4. Shear resistance ................................................................................................ 17
2.3.4.1. Shear failure ........................................................................................ 17
2.3.4.2. Web shear buckling ............................................................................. 18
2.3.4.3. Flexural failure.................................................................................... 18
2.4. Global failure modes .................................................................................................. 18
2.4.1. Lateral-torsional buckling ................................................................................ 19
2.4.2. Lateral buckling failure due to moments in web-posts .................................... 20
2.5. The studies about deflection and shear effect ............................................................ 21
2.6. Research gaps ............................................................................................................. 25
CHAPTER THREE ................................................................................................................. 27
3. DEFLECTION CALCULATION OF CASTELLATED BEAMS .................................... 27
3.1. Introduction ................................................................................................................ 27
3.2. Analytical philosophy of deflection analysis of castellated beams ............................ 27
3.2.1. Deflection of simply supported castellated beam with uniformly distributed

loads ............................................................................................................. 35
3.2.2. Deflection of pinned-fixed castellated beam with uniformly distributed loads37
3.2.3. Deflection of non-uniform temperature distribution simply supported

castellated beam with uniformly distributed loads ....................................... 39
3.2.4. Deflection of non-uniform temperature distribution roller- fixed supported

castellated beam with uniformly distributed loads. ...................................... 43
ix
CONTENTS
3.3. Numerical analysis ......................................................................................................45
3.3.1. The modeling of castellated beams ..................................................................45
3.3.2. Meshing consideration and material model ......................................................46
3.3.3. Boundary conditions and loading .....................................................................47
3.3.3.1. Simply supported castellated beam .....................................................47
3.3.3.2. Pinned-fixed supported castellated beam ............................................47
3.3.4. The dimensions of parameter study and the value of load ...............................48
3.3.5. Comparison of results and discussion ..............................................................50
3.3.5.1. Maximum deflections of simply supported beams subjected to

uniformly distributed loads .................................................................50
3.3.5.2. Maximum deflection of pinned-fixed castellated beam due to

uniformly distributed loads .................................................................51
3.3.5.3. Deflection of simply supported castellated beam due to three fire

scenarios with a uniformly distributed load ........................................61
3.3.5.4. Deflection of roller-fixed castellated beam due to three fire scenarios

with a uniformly distributed ................................................................64
3.4. Conclusions .................................................................................................................67
CHAPTER FOUR ....................................................................................................................69
4. LATERAL-TORSIONAL BUCKLING ..............................................................................69
4.1. Introduction .................................................................................................................69
4.2. Factors influencing lateral-torsional buckling ............................................................70
4.2.1. Effective length (the distance of castellated beam between two laterals
supports) .......................................................................................................70
4.2.2. Boundary conditions .........................................................................................71
x
CONTENTS
4.2.3. Loading location .............................................................................................. 71
4.2.4. Loading type .................................................................................................... 71
4.2.5. Imperfections ................................................................................................... 72
4.2.6. Residual stress .................................................................................................. 72
4.2.7. The slenderness ................................................................................................ 72
4.3. Current design philosophy of lateral-torsional buckling resistance of I-beam with

web openings .............................................................................................................. 73
4.4. The calculation methods of lateral-torsional buckling resistance of castellated beam74
4.5. Analytical philosophy of lateral-torsional buckling of castellated Beam .................. 76
4.5.1. Determining the elastic critical buckling loads of simply supported, doubly
symmetric castellated beam subjected to uniformly distributed loads......... 80
4.5.2. Determining the elastic critical buckling loads of pinned-fixed castellated

beam, doubly symmetric castellated beam subjected to uniformly distributed
loads ............................................................................................................. 81
4.6. Numerical analysis lateral-torsional buckling of castellated beams due to uniformly

distributed load ........................................................................................................... 84
4.6.1. Modelling consideration, material model and loading..................................... 84
4.6.2. Boundary conditions ........................................................................................ 85
4.6.2.1. Simply supported castellated beam .................................................... 85
4.6.2.2. Pinned-fixed castellated beam ............................................................ 85
4.6.3. Comparison of results and discussion .............................................................. 85
4.6.3.1. Linear lateral-torsional buckling stress results of simply supported due

to uniformly distributed load act on top flange .................................. 85
4.6.3.2. Linear lateral-torsional buckling stress results of pinned-fixed

castellated beam due to uniformly distributed load act on top flange 91
xi
CONTENTS
4.7. Conclusions .................................................................................................................96
CHAPTER FIVE .....................................................................................................................98
5. NONLINEAR ANALYSIS OF CASTELLATED BEAMS ................................................98
5.1. Introduction .................................................................................................................98
5.2. Nonlinear finite element analysis of castellated beam ................................................98
5.2.1. Modelling, and boundary conditions ................................................................99
5.2.2. Apply loads gradually (incremental solution) ................................................100
5.3. Serviceability limit state ...........................................................................................100
5.4. Investigating load-deflection response curve for castellated beams with different
boundary conditions. .................................................................................................101
5.4.1. Investigating load-deflection response curve of simply supported beams with

uniformly distributed loads .........................................................................101
5.4.2. Investigating load-deflection response curve of pinned-fixed beams subjected

to uniformly distributed loads .....................................................................109
5.5. Nonlinear lateral-torsional buckling of castellated beams due to a uniformly

distributed load ......................................................................................................... 117
5.5.1. Modelling, loading and boundary conditions ................................................. 117
5.5.2. Comparison of critical lateral-torsional bucking moment of simply supported

beams due to uniformly distributed loads................................................... 118
5.5.3. Comparison of critical moments of lateral-torsional bucking of pinned-fixed

beams subjected to uniformly distributed load ...........................................124
5.6. Conclusions ...............................................................................................................131
CHAPTER SIX.......................................................................................................................133
xii
CONTENTS
6. DYNAMIC INSTABILITY OF CASTELLATED BEAMS UNDER TRANSVERSE

PERIODIC LOADING ..................................................................................................... 133
6.1. Introduction .............................................................................................................. 133
6.2. The studies about dynamic instability ...................................................................... 133
6.3. Governing equations for dynamic instability analysis of castellated beams ............ 135
6.4. Simply supported, doubly symmetric castellated beam subjected to periodic loads on
top flange .................................................................................................................. 141
6.4.1. The free vibration analysis ............................................................................. 142
6.4.2. Buckling analysis ........................................................................................... 143
6.4.3. The dynamic instability .................................................................................. 144
6.4.4. Comparison of the dynamic instability of simply supported beams due to

transverse periodic loading ........................................................................ 145
6.5. Pinned–fixed doubly symmetric castellated beam subjected to periodic loads on top
flange 151
6.5.1. The free vibration analysis ............................................................................. 152
6.5.2. Buckling analysis ........................................................................................... 154
6.5.3. The dynamic instability .................................................................................. 154
6.5.4. Comparison of the dynamic instability of pinned-fixed beams due to transverse

periodic loading .......................................................................................... 155
6.6. Conclusions .............................................................................................................. 161
CHAPTER SEVEN ............................................................................................................... 162
7. CONCLUSIONS AND FUTURE STUDIES ................................................................... 162
7.1. Conclusions .............................................................................................................. 162
7.2. Future studies ........................................................................................................... 165
xiii
CONTENTS
APPENDIX A .........................................................................................................................167
A.1. Web combined stiffness...................................................................................................167
BIBLIOGRAPHY ...................................................................................................................170
xiv
LIST OF TABEL
LIST OF TABLES
Table 3-1 Parameters of castellated beams groups considered ...................................... 48
Table 3-2 Comparison of results of maximum deflections of simply supported

castellated beam with a uniformly distributed load obtained using different approaches
........................................................................................................................................ 53
Table 3-3 The relative errors of maximum deflections of simply supported castellated
beam with a uniformly distributed load obtained based on the finite element numerical
solutions .......................................................................................................................... 55
Table 3-4 Comparison of results of maximum deflections of pinned-fixed castellated

beam with a uniformly distributed load obtained using different approaches ................ 57
Table 3-5 The relative errors of maximum deflections of pinned-fixed castellated beam
with a uniformly distributed load obtained based on the finite element numerical
solutions .......................................................................................................................... 59
Table 3-6 Temperatures and Young’s modulus in two T- sections (Eo is the Young’s
modulus at ambient temperature) ................................................................................... 61
Table 3-7 Comparison of results of maximum deflections of simply supported

castellated beam due to three fire scenarios with a uniformly distributed load .............. 62
Table 3-8 Comparison of results of maximum deflections of roller- fixed castellated

beam due to three fire scenarios with a uniformly distributed load................................ 65
Table 4-1 Comparison of results linear critical lateral-torsion buckling load (𝑞𝑐𝑟, Mcr)
of simply supported castellated beams with a uniformly distributed act on top flange .. 87
Table 4-2 Comparison of results linear critical lateral-torsion buckling load (𝑞𝑐𝑟, Mcr)
of pinned- fixed castellated beams with a uniformly distributed act on top flange ........ 92
Table 5-1 Comparison of results (𝑞𝑐𝑟/𝑞𝑦𝑖𝑒𝑙𝑑) between linear analytical solution and
nonlinear 3D finite element analysis of simply supported castellated beams subjected to a
uniformly distributed for groups C, D, E, G, H, and I with different flange widths. ... 107
xv
LIST OF TABEL
Table 5-2 Comparison of results (𝑞𝑐𝑟/𝑞𝑦𝑖𝑒𝑙𝑑) between linear analytical solution and
nonlinear 3D finite element analysis of pinned-fixed castellated beams subjected to a
uniformly distributed load, for groups C, D, E, G, H, and I beams with different flange
widths. ........................................................................................................................... 115
Table 5-3 Comparison of lateral-torsional buckling results of simply supported

castellated beams subjected to a uniformly distributed load obtained from linear and
nonlinear analysis .......................................................................................................... 120
Table 5-4 Comparison of lateral-torsional buckling results of pinned-fixed castellated

beams subjected to a uniformly distributed load obtained from linear and nonlinear
analysis .......................................................................................................................... 127
Table 6-1 Dimensions and properties of four various flange widths (bf =100 mm,
bf =150 mm, bf =200 mm, and bf =250 mm) castellated beams*................................. 146
Table A-1.Web combined stiffness of castellated beams considered .......................... 168
xvi
LIST OF FIGURES
LIST OF FIGURES
Figure 1-1 Images of application examples of castellated beams in building

construction in Exchange House in London ............................................................... 1
Figure 1-2 A typical castellated beam ....................................................................... 4
Figure 1-3 Methodology and research program ......................................................... 8
Figure 1-4 Organisation diagram of thesis............................................................... 10
Figure 2-1 Fabrication of a castellated beam starting from a plain-webbed parent

section.(Sonck,2014) ................................................................................................ 12
Figure 2-2 Vierendeel failure (a) Vierendeel Truss Analogy, (b) Local failure -
Vierendeel effect (Hosain and Spiers 1973) ............................................................. 14
Figure 2-3 Web buckling due to shear (Badke-Neto et al., 2015) ........................... 15
Figure 2-4 Web buckling due to compressive stress (Hosain and Spiers, 1973). .... 16
Figure 2-5 Web welds failure (Erdal and Saka, 2013)............................................. 17
Figure 2-6 Flexure mechanism (Halleux, 1967). .................................................... 18
Figure 2-7 Lateral-torsional buckling (Kerdal, 1982).............................................. 20
Figure 2-8 Lateral buckling (Showkati, et al. 2012). ............................................... 20
Figure 3-1 (a) Notations used in castellated beams, (b) displacements and (c)
internal forces ........................................................................................................... 28
Figure 3-2 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 unit force F. .................................................................. 34
Figure 3-3.Pinned-fixed beam. ................................................................................ 39
Figure 3-4 Definition of temperature, Young’s modulus and displacement

component in the cross-section (Lei et al., 2017). .................................................... 40
Figure 3-5 Shell element ANSYS library. ............................................................... 46
xvii
LIST OF FIGURES
Figure 3-6 A typical mesh configuration ................................................................. 46
Figure 3-7 The typical loading method and boundary conditions in finite element
model of castellated beam. ........................................................................................ 48
Figure 3-8 Maximum deflections of simply supported castellated beam with a

uniformly distributed load between analytical solutions using different shear rigidity
factors obtained by Eq. (3-49), including one with zero shear factor obtained by Eq.
(3-50) and FEA numerical solution for different beam lengths with various flange
widths. (a) bf =100mm, (b) bf =150mm, (c) bf =200mm and (d) bf =250mm.
(hw=300mm, tf=10mm, tw=8mm and a=100mm) .................................................... 54
Figure 3-9 Relative errors of maximum deflections of simply supported castellated

beam with a uniformly distributed load between analytical solutions using different
shear rigidity factors obtained by Eq. (3-49), including one with zero shear factor
obtained by Eq. (3-50) and FEA numerical solution for different beam lengths with
various flange widths. (a) bf =100mm, (b) bf =150mm, (c) bf =200mm and (d) bf
=250mm. (hw=300mm, tf=10mm, tw=8mm and a=100mm) .................................... 56
Figure 3-10 Maximum deflections of pinned-fixed castellated beam with a

uniformly distributed load between analytical solutions using different shear rigidity
factors obtained by Eq. (3-57), including one with zero shear factor obtained by Eq.
(3-58) and FEA numerical solution for different beam lengths with various flange
widths. (a) bf =100mm, (b) bf =150mm, (c) bf =200mm and (d) bf =250mm.
Figure 3-11 Relative errors of maximum deflections of pinned-fixed castellated

beam with a uniformly distributed load between analytical solutions using different
shear rigidity factors obtained by Eq. (3-57), including one with zero shear factor
obtained by Eq. (3-50) and FEA numerical solution for different beam lengths with
various flange widths. (a) bf =100mm, (b) bf =150mm, (c) bf =200mm and (d) bf
=250mm. (hw=300mm, tf=10mm, tw=8mm and a=100mm) .................................... 60
Figure 3-12 Maximum deflections of simply supported castellated beam due to

three fire scenarios with a uniformly distributed load between analytical solutions
obtained by Eq. (3-74), including one without fire scenarios obtained by Eq. (3-49)
for different beam lengths with various flange widths. (a) bf =100mm, (b) bf
xviii
LIST OF FIGURES
=150mm, (c) bf =200mm and (d) bf =250mm. (hw=300mm, tf=10mm, tw=8mm and
a=100mm). ............................................................................................................... 63
Figure 3-13 Maximum deflections of roller-fixed castellated beam due to three fire
scenarios with a uniformly distributed load between analytical solutions obtained by
Eq. (3-80), including one without fire scenarios obtained by Eq. (3-57) for different
beam lengths with various flange widths. (a) bf =100mm, (b) bf =150mm, (c) bf
=200mm and (d) bf =250mm. (hw=300mm, tf=10mm, tw=8mm and a=100mm). .. 66
Figure 4-1 The mechanism of Lateral-Torsional Buckling ..................................... 70
Figure 4-2 (a) Notations used in castellated beams. (b) Loading and displacements
of web and displacement of flanges when lateral–torsional buckling occurred (c)
Section properties of middle-part of web in four different regions. 𝐼𝑦3 = 𝐼𝑦3 ∗
, 𝐼𝑧3 = 𝐼𝑧3 ∗, 𝐽3 = 𝐽3 ∗ in region 2, in region 4, 𝐼𝑦3 = 𝐼𝑧3 = 𝐽3 = 0 , section
properties vary with x in regions 1 and 3. ................................................................. 80
Figure 4-3 Critical elastic lateral-torsional buckling load (𝑞𝑐𝑟) of simply supported
castellated beam subjected to a uniformly distributed load between analytical
solutions and FEA numerical solution for different beam lengths with various flange
widths (a) bf =100mm, (b) bf=150mm, (c) bf=200mm and (d) bf =250mm
Figure 4-4 Critical elastic lateral-torsional buckling moment (Mcr) of simply

supported castellated beam subjected to a uniformly distributed load between
analytical solutions and FEA numerical solution for different beam lengths with
various flange widths (a) bf =100mm, (b) bf=150mm, (c) bf=200mm and (d) bf
=250mm (hw=300mm, tf=10mm, tw=8mm and a=100mm) .................................... 89
Figure 4-5 Critical elastic lateral-torsional buckling moment (Mcr/Myield) of

simply supported castellated beam subjected to a uniformly distributed load between
=250mm (hw=300mm, tf=10mm, tw=8mm and a=100mm) .................................... 90
Figure 4-6 Critical elastic lateral-torsional buckling load (𝑞𝑐𝑟) of pinned-fixed

xix
LIST OF FIGURES
Figure 4-7 Critical elastic lateral-torsional buckling moment (Mcr) of pinned-fixed

Figure 4-8 Critical elastic lateral-torsional buckling moment (Mcr/Myield) of

pinned-fixed castellated beam subjected to a uniformly distributed load between
=250mm (hw=300mm, tf=10mm, tw=8mm and a=100mm) ..................................... 95
Figure 4-9 Combined modes failure of pinned-fixed castellated beam with a

uniformly distributed load act on top flange for beam (C3 and C4) with two various
flange widths bf=200mm and bf=250mm, obtained from the linear lateral-torsional
buckling 3D finite element analysis using ANSYS software (hw=300mm, tf=10mm,
tw=8mm and a=100mm). .......................................................................................... 96
Figure 5-1 Elastic-perfectly plastic material model ................................................. 99
Figure 5-2 Comparison of the curves of load versus deflection of simply supported
castellated beam with a uniformly distributed load obtained from linear analytical
solutions, nonlinear 3D finite element analysis, and deflection limit (l/250) for
beams C1, D1, E1, G1, H1, and I1. ......................................................................... 103
beams C2, D2, E2, G2, H2, and I2. ......................................................................... 104
xx
LIST OF FIGURES
beams C3, D3, E3, G3, H3, and I3. ........................................................................ 105
beams C4, D4, E4, G4, H4, and I4. ........................................................................ 106
Figure 5-6 Critical values of simply supported castellated beams subjected to a

uniformly distributed load, obtained from linear analytical solution and nonlinear
3D finite element analysis for groups C, D, E, G, H, and I with different flange
widths (a) bf =100mm, (b) bf =150mm, (c) bf =200mm and (d) bf =250mm.
Figure 5-7 Comparison of the curves of load versus deflection of pinned-fixed

castellated beams subjected to a uniformly distributed load, obtained from linear
analytical solutions, nonlinear 3D finite element analysis, and deflection limit
(l/250) for beams C1, D1, E1, G1, H1, and I1. ....................................................... 111

castellated beam subjected to a uniformly distributed load, obtained from linear


Figure 5-11 Critical values of pinned-fixed castellated beams subjected to a

uniformly distributed load, obtained from linear analytical solution and nonlinear
3D finite element analysis for groups C, D, E, G, H, and I with different flange
xxi
LIST OF FIGURES
widths (a) bf =100mm, (b) bf =150mm, (c) bf =200mm and (d) bf =250mm.
Figure 5-12 Comparison of critical moments (Mcr/Myield) of lateral-torsional

buckling of simply supported castellated beam subjected to a uniformly distributed
load, obtained from linear analytical solutions, linear 3D finite element analysis
using ANSYS, and nonlinear 3D finite element analysis (geometric nonlinear and
material inelasticity) for groups C, E, G, H, I, and J beams with various flange
widths (a) bf =100 mm , (b) bf =150 mm , (c) bf =200 mm , and (d) bf =250 mm) 121

buckling of simply supported castellated beam subjected to a uniformly distributed
load, obtained from linear analytical solutions, linear 3D finite element analysis
using ANSYS, and nonlinear 3D finite element analysis (geometric nonlinear and
material inelasticity) using ANSYS software for groups C, E, G, H, I, and J beams
with various flange width. ....................................................................................... 122
Figure 5-14 Comparison of the load-deflection curves of simply supported

castellated beams subjected to a uniformly distributed load, obtained from nonlinear
3D finite element analysis and deflection limit (l/250) for groups C, E, G, H, I and J
beams with various flange widths. .......................................................................... 123
Figure 5-15 Failure mode of simply support castellated beam subjected to a

uniformly distributed load (C4) with flange widths bf=250mm obtained from the
nonlinear lateral-torsional buckling 3D finite element analysis using ANSYS
software (hw=300mm, tf=10mm, tw=8mm and a=100mm) ................................... 124

buckling of pinned-fixed castellated beams subjected to a uniformly distributed load,
obtained from linear analytical solutions, linear 3D finite element analysis using
ANSYS, and nonlinear 3D finite element analysis (geometric nonlinear and material
inelasticity) using ANSYS software for groups C, E, G, H, I, and J beams with
various flange widths (a) bf =100 mm ,(b) bf =150 mm ,(c) bf =200 mm , and (d) bf
=250 mm ................................................................................................................. 128

buckling of pinned-fixed castellated beams subjected to a uniformly distributed load,
xxii
LIST OF FIGURES
obtained from linear analytical solutions, linear 3D finite element analysis using
ANSYS, and nonlinear 3D finite element analysis (geometric nonlinear and material
inelasticity) using ANSYS software for groups C, E, G, H, I, and J beams with
various flange widths. ............................................................................................. 129
Figure 5-18 Comparison of the load-deflection curves of pinned-fixed castellated

beams subjected to a uniformly distributed load, obtained from nonlinear 3D finite
element analysis, and deflection limit (l/250) for groups C, E, G, H, I and J beams
with various flange widths. ..................................................................................... 130
Figure 5-19 Combined failure modes of pinned-fixed castellated beams subjected to

a uniformly distributed load (C3 and C4) with two flange widths bf=200mm and
bf=250mm obtained from the nonlinear lateral-torsional buckling 3D finite element
analysis using ANSYS software (hw=300mm, tf=10mm, tw=8mm and a=100mm)
................................................................................................................................ 131
Figure 6-1 (a) Notations is used in castellated beams. (b) Loading and
displacements of web and displacement of flanges when lateral–torsional buckling
occurred (c) Section properties of middle-part of web in four different regions.
𝐼𝑦3 = 𝐼𝑦3 ∗, 𝐼𝑧3 = 𝐼𝑧3 ∗, 𝐽3 = 𝐽3 ∗ in region 2, in region 4,𝐼𝑦3 = 𝐼𝑧3 = 𝐽3 = 0 ,
section properties vary with x in regions 1 and 3. .................................................. 147
Figure 6-2 Comparison of frequencies of simply support castellated beams with

different flange widths (1st mode) ........................................................................... 147

different flange widths (2nd mode) .......................................................................... 148

different flange widths (3rd mode) .......................................................................... 148
Figure 6-5 Comparison of critical buckling moments of simply support castellated

beam with different flange widths .......................................................................... 149
Figure 6-6 Comparison of dynamic instability regions of simply support castellated

beam (l = 4.156 m) (𝑞𝑦𝑖𝑒𝑙𝑑 = 𝑞𝑐𝑟 and 𝜆𝑠=0) ....................................................... 149
xxiii
LIST OF FIGURES

beam (l = 6.235 m) (𝑞𝑦𝑖𝑒𝑙𝑑 = 𝑞𝑐𝑟 and 𝜆𝑠=0) ........................................................ 150

beam (l = 9.006 m) (𝑞𝑦𝑖𝑒𝑙𝑑 = 𝑞𝑐𝑟 and 𝜆𝑠=0) ........................................................ 150

beam (l = 14.549 m) (𝑞𝑦𝑖𝑒𝑙𝑑 = 𝑞𝑐𝑟 and 𝜆𝑠=0) ...................................................... 151
Figure 6-10 Comparison of frequencies of pinned-fixed castellated beams with

different flange widths (1st mode) ........................................................................... 157

different flange widths (2nd mode) .......................................................................... 157

different flange widths (3rd mode) ........................................................................... 158
Figure 6-13 Comparison of critical buckling moments of pinned-fixed castellated

beam with different flange widths ........................................................................... 158
Figure 6-14 Comparison of dynamic instability regions of pinned-fixed castellated


beam (l =6.235 m) (𝑞𝑦𝑖𝑒𝑙𝑑 = 𝑞𝑐𝑟 and 𝜆𝑠=0) ........................................................ 159

xxiv
CHAPTER ONE INTRODUCTION
CHAPTER ONE
1. INTRODUCTION
1.1. Motivation and background
Technological developments in all areas of construction are encouraging researchers

and engineers as well as designers in the field to improve the performance of structural
members of construction. This requires the development of detailed specifications for
various individual structural members to be used in construction. The purpose of
developing such high-quality standards is to improve the durability, safety, and
serviceability of the structure as well as to reduce the materials and construction costs.
Figure 1-1 shows two images of the application examples of castellated beams used in
the construction of Exchange House in London (photos were taken by Sahar Elaiwi).
Figure 1-1 Images of application examples of castellated beams in building construction in

Exchange House in London
1
Steel is widely utilized in the structural field because of its advantages in material and
mechanical properties as well as time saving in construction, which makes it preferable
in most structures, particularly those used in high-rising buildings. However, due to the
rapidly gaining appeal for steel construction 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 the use of high-quality fireproof materials
to protect steel members from fire, which further increases its cost.
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 equal
performance as the I-beam of the same size (Harper, 1991). In some of the publications
(Altifillisch et al., 1957), the castellated beam is also referred to as the open-web
expanded, perforated steel beam with web openings. 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. This process increases the depth of the beam and
hence the bending strength and stiffness about the major axis without adding additional
materials. 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 (see Figure 1-2). 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. Moreover, because it is light-weight, the castellated beam
is more convenient in transportation and installation than the normal I-beam.
The web openings in the castellated beam, however, may reduce the shear resistance of
the beam. Existing studies have shown that the resistance of the castellated beam is
influenced by shear stresses (Redwood and Demirdjian, 1998), particularly those around
web openings and under the T-section, which could cause the beam to have different
failure modes. There is evidence (Boyer, 1964; Kerdal and Nethercot, 1984;
2
Demirdjian, 1999) that the method of analysis and design for the solid beam may not be
suitable for the castellated beam. 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 for
the calculation of serviceability, the shear weakness due to web openings in castellated
beams could affect the performance of the beams and thus need to be carefully
considered.
The previous researches showed that the responsibility of flange is to resist a large part
of the flexural forces. In contrast, according to (Anupriya and Jagadeesan, 2014), most
of the shear forces are carried by the web. The web openings will not only lead to a
decrease of web in carrying shear forces, but also a decrease in torsional stiffness of
castellated beams; the latter is concerned in relation to lateral torsional buckling.
Clause 4.15.4.5 of BS 5950-4 demonstrates some guidance for the determination of the
lateral–torsional buckling moment of members with web openings. It states that the
method used to determine the lateral buckling resistance moment of solid web beams
can be used for beams with web openings using the section properties at the centreline
of an opening, i.e. the reduced cross-section properties. Therefore, the calculations of
the safe design of castellated beams should take into account the influence of the lateral-
torsion buckling which depends on cross-sectional properties and the depth of the beam
such as the consideration of the effect of flexural and shear capacity.
The design of castellated beams requires the consideration of the lateral-torsion

buckling in addition to the flexural and shear capacity. The previous studies explained
that the cross-sectional properties and the depth of the beam can affect the lateral-
torsional buckling. The castellated beam is more sensitive to lateral-torsional buckling
because it is deeper and more slender than traditional I-beam and the presence of web
openings reduces the torsional stiffness of the web. Note that for long span beams, it is
the beam stiffness that dominates the loading capacity of the beam.
3
1.2. The problem statement and objectives
The transverse deflection and lateral-torsional buckling of structural members are

usually caused by dead loads and live loads such as the weight of the building and
human activities. In castellated beams’ structure, the problem associated with transverse
deflection and lateral-torsional buckling is also influenced by the web openings.
Traditionally, the main serviceability criteria are to limit deflection under a design load
as a percentage of the length of the beam (i.e., for example, l/250 where l is the beam
length). Therefore, the designing procedure of beam needs to consider three different
issues. The first issue is the strength, the second is the deflection and the last one is the
lateral torsional buckling.
Figure 1-2 A typical castellated beam
At the present time, three different methods (Pritykin, 2015) are used to determine the
deflections of castellated beams; one is based on the theory of composed bars, one is
using the theory of Vierendeel truss, and one is the finite element method (FEM). The
first two are the simplified empirical methods; whereas the last one is the numerical
method which could provide more accurate results. However, the use of FEM is quite
complicated and needs to have skills and experience. The FEM software are also
expensive, the use of which requires some training. In contrast, the analytical solution is
simple, easy to use and can be used directly and its key for validating FEM model.
4
The aim of this study is firstly to investigate the effect of web openings on the
transverse deflection and lateral-torsional buckling of castellated beams by using both
analytical and numerical methods. The second aim is to focus on the effect of both the
geometric nonlinearity and material inelasticity on castellated beams under uniformly
distributed loads with different boundary conditions, through investigating the load-
deflection curve and predicting the value of lateral-torsional buckling moment capacity
by using nonlinear numerical solution. The third aim is to study the free vibration, static
buckling and dynamic instability of castellated beams subjected to transverse periodic
loading by developing analytical solutions.
The purpose of developing analytical solutions is for the design and practical use, while
the numerical solutions to be developed are for the validation of the analytical solutions.
The analytical solution will be developed using the classical principle of minimum
potential energy, whereas the numerical solutions will be obtained using the commercial
software ANSYS. The detailed objectives of the research are as follows:
 To investigate the effect of web openings on the transverse deflection of

castellated beams subjected to a uniformly distributed transverse load.
 To develop analytical solutions of transverse deflection of castellated beams

subjected to a uniformly distributed transverse load with/without non-uniform
temperature.
 To evaluate the shear-induced transverse deflection of castellated beams subjected

to a uniformly distributed transverse load.
 To perform linear and nonlinear finite element stress analyses of castellated beams
using 3D linear 4-Node thin shell elements (SHELL181) and to validate the
 To enhancing the knowledge of the effect of web openings on the lateral-torsional

buckling of castellated beams subjected to a uniformly distributed transverse load.
 To develop analytical expressions for the critical load of lateral-torsional buckling

of castellated beams subjected to a uniformly distributed transverse load.
5
 To perform linear and nonlinear finite element also inelastic behaviour lateral-
torsional buckling analyses of castellated beams using three-dimensional shell
elements and to validate the analytical expressions of the critical load of lateral-
torsional buckling.
 To develop analytical expressions for the free vibration (frequency of the lateral-
torsional vibration), static buckling and dynamic instability of castellated beams
subjected to transverse periodic loading by developing analytical solutions.
1.3. Methodology
The main focus of the present study is on the castellated beam for providing economical
and safe design to prevent the beams from the shear, flexural and buckling failures. The
outline of the research program was summarized in Figure 1-3.
The first term of the research program is to calculate the maximum deflection of the
castellated beam subjected to a uniformly distributed load by developing the linear
analytical solution, which is derived, based on the principle of minimum potential
energy. Moreover, linear 3D finite element method is performed by adopting program
ANSYS Mechanical 17.2 software for validation the results of analytical solution.
In the second term, the second order of total potential energy principle is presented to
develop the analytical method for calculating the critical load of lateral-torsional
buckling of the castellated beam subject to a uniformly distributed load. In addition,
linear finite element method is used in the numerical analysis where a commercial
program ANSYS Mechanical 17.2 software package is used to model the castellated
beam in 3D for the validation of the analytical solutions.
Nonlinear finite element method is also used in this thesis to enhance different research
objectives and to understand the behaviours of castellated beams subjected to a
uniformly distributed load in linear elastic and nonlinear performance. Nonlinear
analysis, which takes into account the effect of both the geometric nonlinearity and
material inelastic, was conducted by using a commercial program ANSYS Mechanical
17.2 software package.
6
Another aim of the thesis is to investigate the free vibration, static and dynamic
instability of castellated beams subjected to transverse periodic loading. Bolotin’s
method is used to perform the dynamic instability analysis.
Furthermore, this thesis has presented a tool called ‘solution Tools’ that enhances the
efficiency of the solution process during stages of analytical and numerical analysis.
The 'solution Tools’ includes some of the mathematical programming, such as Maple,
and MS Excel spread sheets.
The analytical results are compared and discussed. The details of research methods used
are listed below:
 The principle of minimum potential energy theory is used to develop the analytical
method for determining the maximum deflection of castellated beams subjected to a
uniformly distributed load with/without non-uniform temperature distribution.
 Second order of the total potential energy theory is used to develop the analytical
method for determining the critical load of lateral-torsional buckling of castellated
beams subjected to a uniformly distributed transverse load.
 Bolotin’s method is used to perform the dynamic instability analysis of castellated

beams subjected to transverse periodic loading. By assuming the instability modes,
the mass, stiffness, and geometric stiffness, matrices are derived using the kinetic
energy, the strain energy and the potential of applied loads.
 Maple mathematical program is used for solving and simplifying the equations of
analytical solution and coupled to MS Excel, which is used to collect characters’
analysis input (load cases, lengths, dimensions of cross-sections, moment of inertia
and area of the section) to find the results and show it in tables and figures.
 Linear and nonlinear finite element method (FEM) is used for validating the
7
Figure 1-3 Methodology and research program
1.4. Organization of thesis
The thesis is organized into seven main chapters (see Figure 1-4). Chapter 1 provides a
brief background about the castellated beam, the general assumptions used for this study
and the motivation of the research.
Chapter 2 contains a literature review of castellated beams, which presents a complete

overview of failure modes, analysis and design methods of castellated beams. This
chapter focuses on several previous studies about castellated beams and theoretical
8
analysis methods of castellated beams. Internet search engines (Google scholar, TU

Delft repository, American Institute of Steel Construction (AISC), Science Direct etc.)
are used to find the resources of relevant studies. The knowledge gap and context for
the contributions of this research are also presented at the end of this chapter.
Chapter 3 introduces the approach of the analytical method, which is used in this study.
The calculation of maximum deflection of the castellated beam subjected to uniformly
distributed loads, with/without non-uniform temperature distribution, with different
boundary conditions, takes into account the shear effect of web openings. The
comparison of results obtained from the analytical and numerical methods are also
provided in this chapter.
Chapter 4 covers a literature review of lateral-torsional buckling castellated beams that

discusses a complete overview of factors that influence lateral-torsional buckling, and
current design philosophy of lateral-torsional buckling resistance of I-beam with web
openings. This chapter focuses on several previous studies about the calculation
methods of lateral-torsional buckling resistance of castellated beams. In addition, the
focuses are on the second-order elastic analysis to develop a simple method to predict
the critical load of lateral-torsional buckling of castellated beams subjected to a
uniformly distributed transverse load. The comparison of results obtained from the
analytical and numerical methods are also provided in this chapter.
Chapter 5 presents the nonlinear finite element numerical analyses of castellated beams
subjected to a uniformly distributed transverse load. Moreover, the nonlinear results are
compared with the linear analytical and linear numerical solutions given in Chapter 4
for investigating the load–deflection relation characteristics and lateral-torsional
buckling of castellated beams under uniformly distributed transverse load.
Chapter 6 discusses the dynamic instability of castellated beams. In addition, analytical

solutions for calculating the free vibration, static buckling and dynamic instability of
castellated beams subjected to transverse periodic loading, are developed. Some
important features of the castellated beams in association with dynamic loading are
highlighted in this chapter.
9
Chapter 7 presents the main conclusions that are drawn from this study and provides
the suggestions and recommendations for future study.
Figure 1-4 Organisation diagram of thesis
10
CHAPTER TWO LITERATURE REVIEW
CHAPTER TWO
2. LITERATURE REVIEW
2.1. History of development of fabrication method
It was reported by Seimaini and Das in 1978 (Bake, 2010), that the first use of the
castellated beam was in 1910 in USA by Horton, who was working on the Chicago
Bridge and Iron Works. This process consisted of cutting the web of the beam and
re-joining the two parts by welding to get a beam that has more strength. However,
the more general concept of fabricating a castellated beam was adopted in 1935 by
the engineer Boyd, who was working in the British Structural Steel Company,
Argentina (Knowles, 1991). According to Knowles, Boyd found a solution to
strengthen a beam by using a simple way; that is to cut the beam web and weld
another, which he cut and welded in the same method. As a result, the new beam has
the capacity of flexural resistance more than the original beam by approximately
20-30% because of increasing the depth by about 50% according (Sherbourne and
Van Oostrom, 1972). In 1940, this product was marketed in the UK and known as
the Boyd beam. Later, this beam's name was replaced by ‘castle-like’ or
‘battlemented’, which refers to the pattern of cutting along the centreline of the web
in a zigzag. (Knowles, 1991)
After the Second World War, the castellated beam was used throughout Europe
because the cost of manufacture was not expensive where the cost of labour was
cheap and the price of material was high. In other words, the ratio of labour cost to
material cost is low. The number of sections of beams, which were available in
European factories, was also limited at that time.
11
Increasing demand for castellated beams led to finding a way to make the fabrication
process more accurate and fast. Litzka Stahlbau Boyer (1964) developed a process
called the Litzka process, which is very efficient using equipment that gives the
designers more suitable design requirements such as the depth of beam, the
dimensions of openings, the angle of openings, the shape of the opening and the
spacing between the openings (Bradley, 2003).
Nowadays the fabrication method of castellated beams is done by cutting the web of
hot rolled I-sections according to a regular alternating pattern along the web by using
either oxy cutting or plasma cutting and then welding teeth of the tee-sections of
halves after one of the halves is shifted. The whole process is controlled
automatically by a computer. Figure 2-1 shows a typical example of the process of a
castellated beam.
Figure 2-1 Fabrication of a castellated beam starting from a plain-webbed parent

section.(Sonck,2014)
2.2. The failure modes of castellated beams
For many years, castellated beams have been used in construction because the
responsibility of a construction engineer is to design the structure according to safety
and serviceability considerations, while considering functional requirements
according to the use for which the construction is intended. Extensive study has been
carried out 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
12
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.
2.3. Local failure modes
The aim of this section is to highlight the failure modes of castellated beams
occurred due to web openings, which is normally located near the opening in the
web posts of the beam. These failures are called local failure modes (Verweij, 2010),
which have been observed and documented in existing studies. In a study done by
Sonck (2014) four different failure modes were defined, namely, the Vierendeel
mechanism caused by web opening, the web post failure also caused by web
opening, the shear and flexural failure caused by beam shear force, and the flexural
failure by bending moment.
2.3.1. Vierendeel failure mode
Altifillisch et al. (1957) and Toprac and Cooke (1959) explained the Vierendeel
failure mode of the castellated beam. This failure mode occurs when a castellated
beam is subjected to high shear forces, which generate the primary moment that is
the conventional bending moment and the secondary moment that is called a
Vierendeel moment. These loads are carried by parts of the beam, which are located
between the openings where they behave as an assembly of individual structural
components, consisting of vertical web posts, upper and lower horizontal T-sections.
Halleux (1967); Hosain and Spiers (1973); (Chung et al. (2001); Liu and Chung
(2003); (Durif et al. (2011); Tsavdaridis and D’Mello (2012) and (Durif et al. 2013)
also discussed the Vierendeel failure mechanism of beams with various different
shapes of openings.
13
In the castellated beam, the global bending moment is redistributed as two axial
forces over the horizontal upper and lower T-sections. The global shear force is
divided into two shear forces over the upper and lower T-sections. The location of
these forces in the components is determined by supposing an inflection point in the
middle of each horizontal T-section, where the local bending moment is zero
assuming that the shear forces in both part sections are equal. According to the static
equilibrium equation, the results of local forces in the upper, lower and middle of the
T-section will be a Vierendeel moment at both T-sections. Therefore, the increase of
horizontal opening length causes an increase in the Vierendeel moment.
Because of the arrival of the T-sections around an opening to the maximum

resistance against the local forces and secondary bending, the castellated beam will
fail by the Vierendeel mechanism which is dominant in the section openings due to
the formation of plastic hinges in the four corners of these openings. Figure 2-2
illustrates Vierendeel Truss Analogy and Vierendeel effect.
Kerdal and Nethercot (1984) investigated the Vierendeel failure mechanism by using
both analytical and experimental methods. They identified the parameters that are
related to the Vierendeel failure mode, such as members with short spans (for which
shear dominates), wide openings (with large secondary bending moments) or
shallow tee sections (low plastic resistance of the tees), etc.
Figure 2-2 Vierendeel failure (a) Vierendeel Truss Analogy, (b) Local failure - Vierendeel
effect (Hosain and Spiers 1973)
14
2.3.2. Web post failure
Slender web-posts in association with deep I-sections present instability in regions

between web openings, which are called web posts. Web post failure is the
consequence of the presence of the axial and shear forces due to the global moment
and global shear force as well as the local Vierendeel moments that encourage
compression in the web post.
2.3.3. Web post buckling
The possibility of the buckling of the web post occurring between web openings of
castellated beams (Zaarour and Redwood, 1996) is due to the presence of the couple
of the horizontal shear forces effective at mid-depth of the web-post, which is
combined with double curvature bending over the height of the post, leading to a
complex failure mode, which is called web post buckling. The technique of the web
post buckling failure mode thought to be caused by one of opening edge of the web
will behave under tension stress but the opposite edge in compression stress. In other
words, the buckling will happen across the compression line, as shown in
Figure 2-3. At a result, the buckling will cause a twisting effect of the web-post
across its height. Kerdal and Nethercot (1984) stated that shear force and web weld
rupture can cause the Vierendeel failure mode and also web post-buckling.
Figure 2-3 Web buckling due to shear (Badke-Neto et al., 2015)
Pourbehi and Pirmoz (2015) also reported the web-post buckling is due to the shear
response and presented some important factors used for the design of castellated
beams. Their data was collected from a numerical study on 300 castellated beam
15
models. Besides the work mentioned above, the web post buckling failure mode in
castellated beams has also been investigated by other researchers. Hosain and Spiers
(1973); Redwood and Demirdjian (1998) and Tsavdaridis and D'Mello (2011) also
investigated the web post buckling failure mode occurred in castellated beams.
2.3.3.1. Compression buckling of web post
In 1990 Ward (Sonck, 2014) stated that – in order to reduce the effect of bearing and
buckling, stiffeners should be used at the reaction location and stress concentrated
points, which can be applied directly at the web post of the castellated beams. Hosain
and Spiers (1973) did an investigation on 12 simple castellated beams and reported
that because of the action of concentrated load, web buckling causes premature
failure (see Figure 2-4). This failure can occur in local loading areas or reaction
force region (Kerdal and Nethercot, 1984). While the buckling behaviour of the
beam of solid web is not affected by the loaded region size, the buckling behaviour
of the castellated beam is affected by the concentrated loads (Okubo and Nethercot,
1985).
Figure 2-4 Web buckling due to compressive stress (Hosain and Spiers, 1973).
2.3.3.2. Bending of web post failure due to moments in web-posts
Bending of web posts failure has investigated by a number of researchers and have
obtained that this failure occurs in castellated beams because the in-plane bending
resistance of the web post of castellated beams will be reduced by the present
16
bending moment in the web post (Müller et al, 2006). He found that the web posts
bending failure can be avoided by increasing the web post width.
2.3.3.3. Web welds failure
Hosain and Spiers (1971) investigations are concerned about the failure at weld joint
area between two segments of a web post in castellated beams (Kerdal and
Nethercot, 1984). Practically, the horizontal shear stresses located in the web post
may potentially exceed the yield strength of the welded throat, which can thus cause
the web welds to fail within a web-post. Note that the web weld failure becomes
more vulnerable if the length of weld throat is short (Maalek, 2004). Figure 2-5
shows the web welds failure, which occurred during an experimental test, carried out
by Hosain and Spiers (1973).
Figure 2-5 Web welds failure (Erdal and Saka, 2013).
2.3.4. Shear resistance
2.3.4.1. Shear failure
Reduction of the local carrying capacity of web posts causes shear failure of the
beam (Müller et al., 2006) and (Tsavdaridis and D'Mello, 2011). It has been reported
that the shear resistance of castellated beams can be determined by the sum of the
vertical shear capacity of top and bottom flanges. However, this state of failure is
17
affected by the cross section of web openings and Vierendeel effects (Demirdjian,
1999).
2.3.4.2. Web shear buckling
Aglan and Redwood (1974) discussed the effect of the spacing between web
openings on the appearance of web shear buckling failure. Redwood and Demirdjian
(1998) reported that high shear forces in a castellated beam could cause the web
shear buckling failure. Demirdjian (1999) also carried out similar investigation.
2.3.4.3. Flexural failure
Toprac and Cooke (1959) and Halleux (1967) discussed the flexural failure mode of
castellated beams (see Figure 2-6). It was reported that applying high bending
moment on the castellated beam can lead to flexural failure.
Figure 2-6 Flexure mechanism (Halleux, 1967).
2.4. Global failure modes
The global failure modes are those caused by global buckling such as flexural
buckling, lateral-torsional buckling, and lateral buckling of the beams (Verweij,
2010).
18
2.4.1. Lateral-torsional buckling
It was reported that the castellated beam with long span often leads to buckling
laterally prior to reaching maximum flexural strength, when sufficient lateral support
is not provided (Demirdjian, 1999). Lateral-torsional buckling failure of slender
beams has been studied extensively since 1980 (Kim et al., 2016). Kerdal and
Nethercot (1984); Korrani et al., (2010) and Sonck et al., (2014) explained the
lateral-torsional buckling mechanism. The fact that the castellated beam has a large
bending stiffness around one axis and a small bending stiffness around other axis
leads the castellated beam to be vulnerable to lateral-torsional buckling unless some
restraints are provided.
Additionally, the presence of web openings causes a decrease of the overall sectional
torsional stiffness, which further decreases the resistance of the beam to lateral-
torsional buckling failure. Kerdal (1982) and Demirdjian (1999) reported that the
torsional stiffness of the web is affected by deeper and more slender section
properties. Figure 2-7 illustrates the shape of castellated beams at lateral-torsional
buckling failure. Mohebkhah and Showkati (2005) used finite element method to
investigate the effects of slenderness on the moment-gradient factor and of elastic
lateral bracing stiffness on the flexural torsional buckling. Zirakian and Showkati
(2006) carried out an investigation by using experimental and finite elements method
to examine the lateral torsional or distortional buckling behaviour of castellated
beams. Sonck and Belis (2016) determined the weak-axis flexural buckling
resistance of castellated beams and cellular beams. The modified diagrams of
castellated beams and cellular beams and the modified residual stress pattern effect
are examined. In this study, critical buckling load and the buckling resistance of
simply support castellated beams and cellular beams are studied using a numerical
method.
19
Figure 2-7 Lateral-torsional buckling (Kerdal, 1982).
2.4.2. Lateral buckling failure due to moments in web-posts
A beam that has a large slenderness and significant web openings, like the long span
castellated beam, can easily fail in a form called lateral buckling of web posts
(Radić. and Markulak, 2007). A number of researchers have investigated the lateral
buckling of web posts of castellated beams and obtained corresponding critical
moments to cause the lateral buckling of web posts. Hosain and Spiers (1973);
Pattanayak and Chesson (1974) and Aglan and Redwood (1974) found that, for most
castellated beams the lateral buckling of web posts comes after the yielding starts in
the web, which indicates that the failure is a combined one of material yield and
buckling. Müller et al. (2006) found that the lateral buckling of the web posts could
be avoided by increasing the web post width. Figure 2-8 shows the shape of lateral
buckling failure of castellated beam (Showkati et al., 2012).
Figure 2-8 Lateral buckling (Showkati, et al. 2012).
20
2.5. The studies about deflection and shear effect
Sherbourne and Van Oostrom (1972) developed a numerical computer program 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.
Hosain et al. (1974) demonstrated that the finite element method is a suitable method
for calculating the deflection of symmetrical section castellated beams. They divided
the castellated beams into an assemblage of elements, which are used to get the
stiffness matrix.
Srimani and Das (1978) conducted an analysis on five experimental groups of

castellated beams to determine the deflection by using finite element method.
Because of symmetry, only one half of the beam was modelled. The numerical
results obtained were compared with experimental data obtained by other
researchers.
Knowles (1991) clarified the previous methods, which have been adopted to
determine design ultimate load and deflection by performing elastoplastic analysis.
He also discussed the design curves used for the castellated beam design.
Zaarour and Redwood (1996) carried out experiments on 12 simply supported

castellated beams with a concentrated load applied at the mid of span which gives a
low moment-to-shear ratio. It demonstrated the possibility of the occurrence of the
buckling of the web posts between web openings. Nonlinear finite element analysis,
taking into account plastic deformation, was conducted to calculate the buckling
loads. A comparison was made between the results they obtained and the results of
previous studies.
Maalek (2004) calculated the shear deflection of the straight-sided tapering

cantilever of the rectangular cross section by using a theoretical method. The
formulae he used were based on Timoshenko’s beam theory and virtual work
21
method. His results were compared with the results obtained from finite element
analysis.
El-Sawy et al. (2009) mentioned that the influence of shear deformation on

castellated beams is directly proportional to the web openings depth and inversely
proportional to the length of the castellated columns and web post width. The wide
range of castellated columns with different cross sections, lengths, and restricted
conditions are examined to estimate the resistance of buckling about the major axis
by using a simplified procedure which is presented in his study. The analytical
results were validated with results obtained by finite element analysis.
Gholizadeh (2011) utilized artificial intelligence technologies called back-

propagation (BP) neural network and adaptive neuro fuzzy inference system
(ANFIS) methods to build a program for predicting the critical load of castellated
beams susceptible to web-post buckling. To validate his results, he also carried out
the nonlinear finite element analysis using NASTAN software and examined the
failure load and failure mode of the beams. It was found that the BP network and
ANFIS methods are able to predict accurate results for the web-post buckling of
castellated beams.
Aminian et al. (2012) proposed new design formulas for calculating the load
carrying capacity of castellated beams by using linear genetic programming and
integrated search algorithms. It was found that the use of machine learning system is
an active method to validate the failure load of castellated beams.
Soltani et al. (2012) studied the effect of nonlinearity in material and/or geometry on
the failure model prediction of castellated beams. Their study was done by using
MSC/NASTRAN software to find out bending moments and shear load capacity,
which are compared with those published in literature.
Erdal and Saka (2013) analysed the cellular beams with different numbers of web
openings and spacings by using nonlinear finite element method, which was carried
out by using ANSYS software. It was found that the combined local buckling, web-
22
post buckling, and Vierendeel failure occur on cellular beams due to the externally
applied loads in different positions and the lateral supports. They also conducted the
experimental tests on 12 cellular beams and the experimental results were used to
validate the numerical results.
Based on Vierendeel failure, Panedpojaman and Rongram (2014) conducted a study

to predict the load carrying capacities by using SCI P100’s method and Chung et
al.’s formula. The results were compared with those obtained from the finite element
analysis method. The load carrying capacities were presented in terms of the
moment-shear interaction curve for studying Vierendeel’s effect on the overall
behaviour of the cellular beams. Moreover, the investigations that were carried out
on 120 non-linear finite element models of cellular beams with beam section sizes
and opening ratios of 0.5 and 0.8 are reported. The results showed that the FE
interaction curve’s shape is slightly affected by the sizes of the steel sections. The
study suggested a simple quadratic equation to estimate the interaction based on
Vierendeel failure.
Recently, Wang et al. (2014) conducted a parametric study on the large deflection
analysis of castellated beams at high temperatures. They used 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. The results were
compared with those of corresponding solid beams to examine the effect of web
openings.
Axial compression buckling of castellated columns was investigated by Yuan et al.

(2014), who derived an analytical solution for critical load based on stationary
potential energy and considering the effect of the web shear deformations on the
flexural buckling of simply supported castellated column. The analytical solution
was validated by using the numerical results obtained from finite element analysis.
Menkulasi et al. (2015) analysed 30 models of castellated beams subjected to

concentrated loads with different cross sections and loading positions for evaluating
the resistance of castellated beams to the concentrated load with and without
23
stiffeners. The study was carried out by employing nonlinear finite element analysis
method.
Morkhade and Gupta (2015) discussed the effect of the dimensions and positions of
web holes and locations of stiffeners around the holes on the behaviour of the
perforated steel I beam with rectangular web holes. The study showed that the load
carrying capacity of the beam is affected by the positions of holes. Experimental and
finite element analysis methods were employed in this study. The nonlinear finite
element analysis was carried out by using ANSYS software.
More recently, Sonck et al. (2015) presented a comprehensive comparison between

the deflection results of cellular and castellated beams obtained from numerical
analysis and those obtained from different simplified design codes. The beams
studied involve various web opening dimensions, different parent sections and
different span lengths. The comparison showed that the design codes are not accurate
for short span beams and conservative for long span beams.
Yuan et al. (2016) adopted the principle of minimum potential energy 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 analytical results were validated by using numerical results obtained from
finite element analysis using ANSYS software.
Lei et al. (2017) investigated the buckling behaviour of castellated columns exposed
to a fire when they were also subjected to axial compression. In their study, an
analytical equation was developed based on the web opening shear effect and the
non-uniform cross-section temperature distribution for predicting the critical
buckling load of castellated columns.
Martins et al. (2017) investigated the lateral-torsional buckling of cellular beams.

Nonlinear geometric and material analyses were developed for modelling the
behaviour of cellular beams with different restricted conditions and different loading
cases, which are achieved by using ABAQUS 6.12 software. In their study, the
24
numerical results, which took into account geometric imperfections, were compared
with those obtained by using ABNT NBR 8800:2008 standards and by Abreu et al.
(2010).
2.6. Research gaps
The goal of the literature review is to identify the research fields, which are needed
to further study and investigate. As a result of the previous research efforts, the
general behaviour of castellated beams and understanding different failure modes,
which occur during the construction and/or loading, were discussed. Due to the
geometric particulars of the beam, however, it was remarkable to note that most of
the theoretical approximate methods are interested in design calculations 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 complexity of section profile of the
castellated beams, the shear-effect caused by the web opening on the calculation is
not fully understood.
In addition, the static load still dominates the current designing of structures for
castellated beams, in spite of the significance of the dynamic response to machinery
loading and to extreme environmental loads. Actually, there are no accurate
calculation methods available in the literature to perform these analyses.
Also, it seems that the significant problem of the castellated beam has remained
unresolved. European building standards do not have formulas for design
calculations of castellated beams, which include shear deformations at the static and
dynamic state.
In literature, researchers have adopted the finite element method to predict the
calculations design of castellated beams by using different software programs such
as (MS/NASTRAN, ABAQUS, and ANSYS). However, these programs need to be
used efficiently because any error could lead to significant distortions in results. In
25
spite of the potential programs, only the case of simply supported castellated beams
has been discussed in most of the studies.
The development of the analytical methods to predict the static and dynamic
calculations of the castellated beam took place in the present research. The equations
of the analytical solutions are 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.
26
CHAPTER THREE DEFLECTION CALCULATION
CHAPTER THREE
3. DEFLECTION CALCULATION OF CASTELLATED BEAMS
3.1. Introduction
This chapter firstly puts emphasise on the development of an approximate method of

deflection calculation of castellated beams subjected to a uniformly distributed
transverse load with/without temperature effect. The method is developed based on the
principle of minimum potential energy. The analytical method developed here considers
the shear effect of web openings. To demonstrate the analytical method, comparisons
between the analytical results, previous studies from the literature, and linear FEA
numerical computations are also provided.
3.2. 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 in this chapter. The method is developed based
on the principle of minimum potential energy.
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 stress, two of which
represent the top and bottom T-sections, one of which represents the mid-part of the
web. The analysis model for this study is illustrated in Figure 3-1 (a).
In this study, the cross-section of the castellated beam is assumed to be doubly

symmetric, with the flange width and thickness as bf and tf, the web depth and thickness
as hw and tw, and the half depth of hexagons as a. The half of the distance between the
27
centroids of the two T-sections is e and the hexagonal opening height is 2a. For a 60°
2a
cutting angle the side length of the hexagonal opening is ( ) and the width of one
√3
orignal depth of beam a
sloping edge of hole S0 is calculated from ( ), in which 𝑆0 is ( ).
tan 60 √3
Hence the distance from the centreline to the centreline of adjacent castellation holes
2a 6a
can be found according to this formulae S = 2 ( + S0 ), which equals to ( )
√3 √3
(Demirdjian, 1999).
Figure 3-1 (a) Notations used in castellated beams, (b) displacements and (c) internal forces
Under the action of a uniformly distributed transverse load, the beam section will have
axial and transverse displacements as shown in Figure 3-1 (b), where (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 have the same transverse
28
displacement according to the use of beam model (Yuan et al., 2014). The
corresponding axial strains ε1x in the upper T-section and ε2x in the lower T-section are
assumed to be linearly distributed, and so do the displacements. Thus, the displacements
can be expressed as follows:
hw
In the upper T-section： − ( + t f ) ≤ z ≤ −a
2
𝑑𝑤
𝑢(𝑥, 𝑧) = 𝑢1 (𝑥) − (𝑧 + 𝑒) 3-1
𝑑𝑥
ℎ
In the lower T-section：𝑎 ≤ 𝑧 ≤ ( 𝑤 + 𝑡𝑓 )
2
𝑑𝑤
𝑢(𝑥, 𝑧) = 𝑢2 (𝑥) − (𝑧 − 𝑒) 3-2
𝑑𝑥
For the middle part between the two T-sections: −𝑎 ≤ 𝑧 ≤ 𝑎
𝑢1 (𝑥) + 𝑢2 (𝑥) 𝑧 𝑢1 (𝑥) − 𝑢2 (𝑥) 𝑑𝑤

𝑢(𝑥, 𝑧) = − ( − (𝑒 − 𝑎) ) 3-3
2 𝑎 2 𝑑𝑥
The corresponding axial strains in the two T-sections can be determined by using the
strain-displacement relation as follows:
For the upper T-section:
𝑑𝑢1 𝑑2 𝑤 3-4
𝜀1 (𝑥, 𝑧) = − (𝑧 + 𝑒) 2
𝑑𝑥 𝑑𝑥
For the lower T section,
𝑑𝑢2 𝑑2𝑤 3-5

𝜀2 (𝑥, 𝑧) = − (𝑧 − 𝑒) 2
𝑑𝑥 𝑑𝑥
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:
𝜕𝑢 𝜕𝑤 𝑢1 − 𝑢2 𝑒 𝑑𝑤 3-6
𝛾𝑥𝑧 (𝑥, 𝑧) = + =− +
𝜕𝑧 𝜕𝑥 2𝑎 𝑎 𝑑𝑥
in which,
ℎ𝑤+ 𝑡𝑓 ℎ ℎ + 2𝑎
𝑏𝑓 𝑡𝑓 ( 2 ) +𝑡𝑤 ( 2𝑤 − 𝑎) ( 𝑤 4 )
𝑒= 3-7
ℎ
𝑏𝑓 𝑡𝑓 +𝑡𝑤 ( 2𝑤 − 𝑎)
29
For simplicity of presentation, the following two new functions are used:
𝑢1 + 𝑢2
𝑢𝛽 = 3-8
2
𝑢1 − 𝑢2
𝑢𝛽 = 3-9
2
For the conservative system in the equilibrium state, the effect of a uniformly
distributed transverse load on castellated beams will lead to internal forces. To
determine the distribution of internal forces at an opening, Lawson and Hicks (2011)
reported that each of the T-sections is subjected to combined axial force, bending and
shear. According to Figure 3-1 (b) these internal forces can be obtained as follows:
−ℎ𝑤 /2 −𝑎
𝑑𝑢1
𝑁1 = 𝐸𝑏𝑓 ∫ 𝜀1𝑥 𝑑𝑧 + 𝐸𝑡𝑤 ∫ 𝜀1𝑥 𝑑𝑧 = 𝐸𝐴𝑡𝑒𝑒 3-10
𝑑𝑥
ℎ ℎ
−(𝑡𝑓 + 𝑤 ) − 𝑤
2 2
−ℎ𝑤 /2 −𝑎
𝑑2𝑤
𝑀1 = 𝐸𝑏𝑓 ∫ (𝑧 + 𝑒)𝜀1𝑥 𝑑𝑧 + 𝐸𝑡𝑤 ∫ (𝑧 + 𝑒)𝜀1𝑥 𝑑𝑧 = −𝐸𝐼𝑡𝑒𝑒 3-11
𝑑𝑥 2
ℎ ℎ
−(𝑡𝑓 + 𝑤 ) − 𝑤
2 2
𝑎
𝑑𝑤 𝑢1 − 𝑢2 3-12
𝑄3 = 𝐺𝑡𝑤 ∫ 𝛾𝑥𝑧 𝑑𝑧 = 2𝐺𝑡𝑤 (𝑒 − )
𝑑𝑥 2
−𝑎
ℎ𝑤 /2 𝑡𝑓 +ℎ𝑤 /2
𝑑𝑢2
𝑁2 = 𝐸𝑡𝑤 ∫ 𝜀2𝑥 𝑑𝑧 + 𝐸𝑏𝑓 ∫ 𝜀2𝑥 𝑑𝑧 = 𝐸𝐴𝑡𝑒𝑒 3-13
𝑑𝑥
𝑎 ℎ𝑤
2
ℎ𝑤 /2 𝑡𝑓 +ℎ𝑤 /2
𝑑2 𝑤
𝑀2 = 𝐸𝑡𝑤 ∫ (𝑧 + 𝑒)𝜀2𝑥 𝑑𝑧 + 𝐸𝑏𝑓 ∫ (𝑧 + 𝑒)𝜀2𝑥 𝑑𝑧 = −𝐸𝐼𝑡𝑒𝑒 3-14
𝑑𝑥 2
𝑎 ℎ𝑤
2
where E is the Young's modulus and 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 system as follows:
ℎ𝑤 3-15
𝐴𝑡𝑒𝑒 = 𝑏𝑓 𝑡𝑓 + 𝑡𝑤 ( − 𝑎)
2
30
2
3
𝑏𝑓 𝑡𝑓3 ℎ𝑤+ 𝑡𝑓 𝑡𝑤 ℎ𝑤
𝐼𝑡𝑒𝑒 = + 𝑏𝑓 𝑡𝑓 ( − 𝑒) + ( − 𝑎)
12 2 12 2 3-16
2
ℎ𝑤 ℎ𝑤 + 2𝑎
+ 𝑡𝑤 ( − 𝑎) ( − 𝑒)
2 4
3
𝑏𝑓 (ℎ𝑤 + 2𝑡𝑓 ) (2𝑎)3 𝑡𝑤 (ℎ𝑤 )3 (𝑏𝑓 − 𝑡𝑤 ) 3-17
𝐼𝑟𝑒𝑑𝑢𝑐𝑒𝑑 = − −
12 12 12
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 +ℎ𝑤 )
𝑓 2
0 −(ℎ𝑤 ) 3-18
2
𝑙 2
1 𝑑𝑢1 2 𝑑2𝑤
= ∫ [𝐸𝐴𝑡𝑒𝑒 ( ) + 𝐸𝐼𝑡𝑒𝑒 ( 2 ) ] 𝑑𝑥
2 𝑑𝑥 𝑑𝑥
0
ℎ𝑤 ℎ𝑤
𝑙(2 ) 𝑙 (𝑡𝑓 + 2 )
𝐸𝑡𝑤 2
𝐸𝑏𝑓 2
𝑈2 = ∫ ∫ 𝜀2𝑥 𝑑𝑧𝑑𝑥 + ∫ ∫ 𝜀2𝑥 𝑑𝑧𝑑𝑥
2 2
0 𝑎 0 ℎ𝑤 3-19
2
𝑙 2
1 𝑑𝑢2 2 𝑑2𝑤
= ∫ [𝐸𝐴𝑡𝑒𝑒 ( ) + 𝐸𝐼𝑡𝑒𝑒 ( 2 ) ] 𝑑𝑥
2 𝑑𝑥 𝑑𝑥
0
Hence, the total of strain energy of the upper and lower T-sections can be expressed as
follows:
𝑙 2 𝑙 2
𝐸𝐴𝑡𝑒𝑒 𝑑𝑢1 2 𝑑𝑢2 𝑑2𝑤 3-20
𝑈𝑇 = ∫ [( ) +( ) ] 𝑑𝑥 + 𝐸𝐼𝑡𝑒𝑒 ∫ ( 2 ) 𝑑𝑥
2 𝑑𝑥 𝑑𝑥 𝑑𝑥
0 0
The mid-part of the web of the castellated beam, which is illustrated in Figure 3-1 (a),
is assumed to behave according to Timoshenko’s theory (Yuan et al., 2014). Therefore,
its strain energy due to the bending and shear Ub can be expressed as follows:
1 3-21
𝑈𝑏 = ∑ 𝐾𝑏 ∆2
2
31
where ∆ is the relative displacement of the upper and lower T-sections due to a pair of
shear forces and can be expressed as(∆= 2𝑎𝛾𝑥𝑦 ), 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: (Timoshenko and Gere,1961)
1 3𝑙𝑏 𝑙𝑏3
= + 3-22
𝐾𝑏 2𝐺𝐴𝑏 12𝐸𝐼𝑏
where (𝐴𝑏 = 𝑡𝑤 √3𝑎) is the equivalent cross-sectional area of the mid part of the web,
3
(𝐼𝑏 = 𝑡𝑤 (√3𝑎) /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 + 𝑣)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 follows: (Yuan et al., 2014)
√3𝐺𝑡𝑤 3-23
𝐾𝑏 =
4
As a result, for castellated beam the total shear strain energy of the mid-part 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 ∑ 𝛾𝑥𝑧
2
≈ 2
∫ 𝛾𝑥𝑧 𝑑𝑥 = 2
∫ 𝛾𝑥𝑧 𝑑𝑥 3-24
2 6𝑎 4
𝑘=1 2 × 0 0
√3
1
Letting the shear rigidity factor 𝑘𝑠ℎ = 4 (Yuan et al., 2014) and substituting Eqs. (3-6)
into (3-24), it gives the total shear strain energy of the mid-part of the web:
𝑙
𝐺𝑡𝑤 𝑒 2 𝑘𝑠ℎ 𝑑𝑤 𝑢𝛽 2
𝑈𝑠ℎ = ∫( − ) 𝑑𝑥 3-25
𝑎 𝑑𝑥 𝑒
0
Note that, in the calculation of shear strain energy of Eq. (3-24) 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
1
shear strain energy of the web with holes was evaluated, in which (𝐾𝑠ℎ = 4) was
obtained. However, by using a two-dimensional linear finite element analysis
32
(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 shears was found to be:
√3𝐺𝑡𝑤 3-26
𝐾𝑏 = 0.78 ×
4
which is smaller than that above-derived from the smear model. This leads to the shear
1
rigidity factor(𝐾𝑠ℎ = 0.78 × 4). 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. (3-24).
However, it should be mentioned that the factor of (0.78) in Eq. (3-26) 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 (see Figure 3-2 (c)), 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
1
applied (see Figure 3-2 (c)). Hence, the combined rigidity (𝐾𝑏 = ) 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. (3-27) is used in the present analytical solutions:
𝑏𝑓 1 3-27
𝐾𝑠ℎ = (0.76 − )×
𝑙 4
Hence the value of the combined stiffness of the mid part of the web of the castellated
beam caused by the bending and shear which is adopted in this thesis is:
𝑏𝑓 √3𝐺𝑡𝑤 3-28
𝐾𝑏 = (0.76 − )×
𝑙 4
where l is the length of the beam. Thus, the total potential energy of the castellated
beam U is expressed as follows:
33
𝑈 = 𝑈𝑇 + 𝑈𝑠ℎ 3-29
The potential of the uniformly distributed load due to the transverse displacement can
be expressed as follows:
𝑊 = 𝑞𝑚𝑎𝑥 ∫ 𝑤 𝑑𝑥 3-30
0
where 𝑞𝑚𝑎𝑥 is the uniformly distributed load, is calculated according to the maximum
𝑀𝑦 (ℎ𝑤 +2𝑡𝑓 )
moment, which can be expressed in terms of design stress 𝜎𝑦 = , as follows:
2𝐼𝑟𝑒𝑑𝑢𝑐𝑒𝑑
𝜎𝑦 𝐼𝑟𝑒𝑑𝑢𝑐𝑒𝑑
𝑞𝑚𝑎𝑥 = 16 2 3-31
𝑙 (ℎ𝑤 + 2𝑡𝑓 )
where 𝜎𝑦 is the yield stress of castellated beam.
In summary, the total potential energy of the castellated beam subjected to a uniformly
distributed load can be expressed as follows:
𝑙 2 𝑙 2
𝑑𝑢𝛽 𝑑2 𝑤
∏ = 𝐸𝐴𝑡𝑒𝑒 ∫ ( ) 𝑑𝑥 +𝐸𝐼𝑡𝑒𝑒 ∫ ( 2 ) 𝑑𝑥
𝑑𝑥 𝑑𝑥
0 0 3-32
𝑙
2
𝐺𝑡𝑤 𝑒 𝑘𝑠ℎ 𝑑𝑤 𝑢𝛽 2
+ ∫( − ) 𝑑𝑥 − 𝑊
𝑎 𝑑𝑥 𝑒
0
Figure 3-2 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 unit force F.
34
3.2.1. Deflection of simply supported castellated beam with uniformly distributed

loads
For a simply supported castellated beam: uα(x), uβ(x) and w(x) can be assumed as
follows:
𝑚𝜋𝑥
𝑢𝛼 (𝑥) = ∑ 𝐴𝑚 𝑐𝑜𝑠 3-33
𝑙
𝑚=1,2,..
𝑚𝜋𝑥
𝑢𝛽 (𝑥) = ∑ 𝐵𝑚 𝑐𝑜𝑠 3-34
𝑙
𝑚=1,2,..
𝑚𝜋𝑥 3-35
𝑤(𝑥) = ∑ 𝐶𝑚 𝑠𝑖𝑛
𝑙
𝑚=1,2,..
where 𝐴𝑚 , 𝐵𝑚 , and 𝐶𝑚 (𝑚 = 1,2, … ) are the constants to be determined. It is obvious

that the displacement functions assumed in Eqs. (3-33), (3-34) and (3-35) satisfy the
𝑑2 𝑤 𝑑𝑢𝛼 𝑑𝑢𝛽
simply support boundary conditions, that are 𝑤 = = 0 , and = = 0 at x =0
𝑑𝑥 2 𝑑𝑥 𝑑𝑥
and x = l.
Substituting Eqs. (3-33), (3-34) and (3-35) into (3-32), according to the principle of
minimum potential energy: (Timoshenko, 1961)
𝛿(𝑈𝑇 + 𝑈𝑠ℎ − 𝑊) = 0 3-36
The variation of Eq. (3-36) with respect to 𝐴𝑚 , 𝐵𝑚 and 𝐶𝑚 results in the following
three algebraic equations:
𝑚𝜋𝑥 2 3-37
𝐸𝐴𝑡𝑒𝑒 ( ) 𝐴𝑚 = 0
𝑙
𝑚𝜋 2 𝐺𝑡𝑤 𝑘𝑠ℎ 𝐺𝑡𝑤 𝑒𝑘𝑠ℎ 𝑚𝜋

[𝐸𝐴𝑡𝑒𝑒 ( ) + ] 𝐵𝑚 − [ ( )] 𝐶𝑚 = 0 3-38
𝑙 𝑎 𝑎 𝑙
35
𝑚𝜋 4 𝐺𝑡𝑤 𝑒 2 𝑘𝑠ℎ 𝑚𝜋 2
[𝐸𝐼𝑡𝑒𝑒 ( ) + ( ) ] 𝐶𝑚
𝑙 𝑎 𝑙
3-39
𝐺𝑡𝑤 𝑒𝑘𝑠ℎ 𝑚𝜋 [1 − (−1)𝑚 ]𝑞𝑚𝑎𝑥
−[ ( )] 𝐵𝑚
𝑎 𝑙 𝑚𝜋
Mathematically Eqs. (3-37) - (3-39) lead to:
𝐴𝑚 = 0 3-40
𝐺𝑡𝑤 𝑒𝑘𝑠ℎ 𝑚𝜋
[ ( )]
𝑎 𝑙
𝐵𝑚 = 𝐶 3-41
𝑚𝜋 2 𝐺𝑡𝑤 𝑘𝑠ℎ 𝑚
[𝐸𝐴𝑡𝑒𝑒 ( ) + 𝑎 ]
𝑙
𝑚
1 − (−1) 𝑞𝑙 4
𝐶𝑚 =
(𝑚𝜋)5 𝑒 2 𝐸𝐴𝑡𝑒𝑒 3-42
𝐸𝐼𝑡𝑒𝑒 +
𝐸𝐴𝑡𝑒𝑒 𝑎(𝑚𝜋)2
1+
𝐺𝑘𝑠ℎ 𝑡𝑤 𝑙 2
Cm can be further simplified as:
𝑞𝑙 4 1 − (−1)𝑚 𝑒 2 𝐴𝑡𝑒𝑒 𝐸𝐴𝑡𝑒𝑒 𝑎(𝑚𝜋)2

𝐶𝑚 = [1 + ×
𝐸(𝐼𝑡𝑒𝑒 + 𝑒 2 𝐴𝑡𝑒𝑒 ) (𝑚𝜋)5 𝐼𝑡𝑒𝑒 + 𝑒 2 𝐴𝑡𝑒𝑒 𝐺𝑘𝑠ℎ 𝑡𝑤 𝑙 2
3-43
𝐸𝐼𝑡𝑒𝑒 𝑎(𝑚𝜋)2
× (1 − )]
𝐺𝑘𝑠ℎ 𝑡𝑤 𝑙 2 𝑒 2
Therefore, the expression of castellated beam deflection is 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,.. 3-44
𝐸𝐴𝑡𝑒𝑒 𝑎(𝑚𝜋)2 𝐸𝐼𝑡𝑒𝑒 𝑎(𝑚𝜋)2 𝑚𝜋𝑥
× × (1 − )] sin
𝐺𝑘𝑠ℎ 𝑡𝑤 𝑙 2 𝐺𝑘𝑠ℎ 𝑡𝑤 𝑙 2 𝑒 2 𝑙
𝑙
The maximum deflection of the beam is at 𝑥 = ，and thus can be can be expressed
2
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,..
3-45
2 (−1)𝑘+1 𝐸𝐼𝑡𝑒𝑒 𝑎 2 (−1)𝑘+1
×( ∑ 2 − ∑ )]
𝜋 (2𝑘 − 1)3 𝐺𝑘𝑠ℎ 𝑡𝑤 𝑙 2 𝑒 2 𝜋 (2𝑘 − 1)
𝑘=1,2,.. 𝑘=1,2,..
Note that:
36
2 (−1)𝑘+1 5
∑ 5 = 3-46
𝜋 (2𝑘 − 1)5 2 × 384
𝑘=1,2,..
2 (−1)𝑘+1 1
∑ 3 3
= 3-47
𝜋 (2𝑘 − 1) 16
𝑘=1,2,..
2 (−1)𝑘+1 1
∑ = 3-48
𝜋 (2𝑘 − 1) 2
𝑘=1,2,..
Substituting Eqs. (3-46), (3-47) and (3-48) into (3-45) gives the maximum deflection:
2
5𝑞𝑙 4 𝑞𝑙 2 𝑎 𝑒𝐴𝑡𝑒𝑒
𝑤|𝑥=𝑙/2 = + ×( )
384𝐸(2𝐼𝑡𝑒𝑒 + 2𝑒 2 𝐴𝑡𝑒𝑒 ) 16𝐺𝑘𝑠ℎ 𝑡𝑤 𝐼𝑡𝑒𝑒 + 𝑒 2 𝐴𝑡𝑒𝑒
3-49
2𝐸𝐼𝑡𝑒𝑒 𝑎
× (1 − )
𝐺𝑘𝑠ℎ 𝑡𝑤 𝑙 2 𝑒 2
It is clear from Eq. (3-49) that the first part indicates the deflection generated by the
bending load, which is deemed as that given by Bernoulli-Euler beam, while the second
part provides the deflection generated by the shear. Moreover, Eq. (3-49) 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.
To determine the approximate deflection, the calculation does not consider webs’ shear
effect, and this is shown in Eq. (3-50).
5𝑞𝑙 4
𝑤|𝑥=𝑙/2 = 3-50
384𝐸𝐼𝑟𝑒𝑑𝑢𝑐𝑒𝑑
3.2.2. Deflection of pinned-fixed castellated beam with uniformly distributed

loads
For pinned-fixed castellated beam uα(x), uβ(x) and w(x) can be assumed as follows:
(2𝑘 − 1)𝜋(𝑙 − 𝑥)
𝑢𝛼 (𝑥) = ∑ 𝐹𝑘 [(2𝑘 − 1) sin ( )
2𝑙
𝑚=1,2,.. 3-51
(2𝑘 + 1)𝜋(𝑙 − 𝑥)
− (2𝑘 + 1) sin ( )]
2𝑙
37
(2𝑘 − 1)𝜋(𝑙 − 𝑥)
𝑢𝛽 (𝑥) = ∑ 𝐸𝑘 [(2𝑘 − 1) sin ( )
2𝑙
𝑚=1,2,.. 3-52
(2𝑘 + 1)𝜋(𝑙 − 𝑥)
− (2𝑘 + 1) sin ( )]
2𝑙
𝑘𝜋(𝑙 − 𝑥) 𝜋(𝑙 − 𝑥)
𝑤(𝑥) = ∑ 𝐷𝑘 sin ( ) sin ( )
𝑚=1,2,..
𝑙 2𝑙 3-53
where 𝐹𝑘 , 𝐸𝑘 , and 𝐷𝑘 (𝑘 = 1,2, …) are the coefficients to be determined. It is obvious

that the displacement functions assumed Eqs. (3-51), (3-52) and (3-53) to satisfy the
𝑑2 𝑤 𝑑𝑢𝛽
pinned-fixed support boundary conditions, that is 𝑤 = = 0, and = 0 at x = 0
𝑑𝑥 2 𝑑𝑥
𝑑𝑤
and 𝑤 = = 𝑢𝛼 = 𝑢𝛽 = 0 at x = l.
𝑑𝑥
Substituting Eqs. (3-51), (3-52) and (3-53) into (3-32), and according to the principle of
minimum potential energy at Eq. (3-36) it yields:
128𝑞𝑘𝑙 5 𝐺𝑘𝑠ℎ 𝑡𝑤 e cos(𝑘𝜋) (1 + 4𝑘 2 )

𝐸𝑘 =
𝜋 4 (16𝑘 4 + 24𝑘 2 + 1)(4𝑘 2 − 1)
1 3-54
×
4(4𝑘 2 + 1)𝐸𝐺𝑘𝑠ℎ 𝑡𝑤 𝑙 2 (𝐴𝑡𝑒𝑒 𝑒 2 + 𝐼𝑡𝑒𝑒 )
[ 2 1 2 2
]
( + (2𝑘 + 3 + 8𝑘 2 ) 8𝐸𝐴𝑡𝑒𝑒 𝐸𝐼𝑡𝑒𝑒 𝜋 𝑎𝑘 )
−128𝑞𝑘𝑙 4 cos(𝑘𝜋) 𝑍
𝐷𝑘 = 3-55
𝜋 5 (16𝑘 4 + 24𝑘 2 + 1)(4𝑘 2 − 1)
where:
1
((2𝑘 2 + 3 + 2 2 2
2 )8𝐸𝐴𝑡𝑒𝑒 𝜋 𝑎𝑘 + (4𝑘 + 1)4𝐺𝑘𝑠ℎ 𝑡𝑤 𝑙 )
2
8𝑘
𝑍=
1 3-56
[4(4𝑘 2 + 1)𝐸𝐺𝑘𝑠ℎ 𝑡𝑤 𝑙 2 (𝐴𝑡𝑒𝑒 𝑒 2 + 𝐼𝑡𝑒𝑒 ) + (2𝑘 2 + 3 + 2 ) 8𝐸𝐴𝑡𝑒𝑒 𝐸𝐼𝑡𝑒𝑒 𝜋 2 𝑎𝑘 2 ]
8𝑘
Therefore, the deflection of pinned-fixed castellated beam can be expressed as:
𝑘𝜋(𝑙 − 𝑥) 𝜋(𝑙 − 𝑥)
−128𝑞𝑘𝑙 4 cos(𝑘𝜋) 𝑍 sin ( ) sin ( )
𝑙 2𝑙 3-57
w(𝑥) = ∑
𝜋 5 (16𝑘 4 + 24𝑘 2 + 1)(4𝑘 2 − 1)
𝑘=1,2,3
38
Clearly, the denominator of Eq. (3-57) is very complicated, and the position of the
maximum deflection point is at (x= 0.375l ) away from the pinned support end, which is
obtained by using the zero derivative of the deflection function with respect to
coordinate as described in the bending theory of beams (Timoshenko, 1961).The
Eq. (3-57) is not easy to simplify but it is regarded as Bernoulli-Euler beam with
modifications.
To determine the approximate deflection, the calculation does not consider webs’ shear
effect, but rather can be employed Eq. (3-58):
𝑞𝑙 4
𝑤𝑚𝑎𝑥 = 3-58
185𝐸𝐼𝑟𝑒𝑑𝑢𝑐𝑒𝑑
Figure 3-3.Pinned-fixed beam.
3.2.3. Deflection of non-uniform temperature distribution simply supported

castellated beam with uniformly distributed loads
Consider a castellated beam that is used in the external floor and wall, in which the
beam can be bent about its major axis but is restrained in its lateral direction. When
there is fire inside the building, the castellated beam is exposed to a fire on one side.
During the fire, the heat will be transferred from the fire into the beam, which causes
the beam to have a non-uniform temperature distribution on its cross-section.
Consequently, the material properties that depend on the temperature will also not be
uniform in the cross-section.
Since the heat is transferred from one flange to another through the web, it is reasonable
to assume that the temperature is uniformly distributed in each T-section, although the
two T-sections may have different temperatures at the same time.
39
Note that the thermal expansion induced by temperature was not considered in this
study. This is because when a simply support boundary is involved in the beam the axial
displacement can be freely expanded so there will be no thermal stresses in the beam.
All thermal strains are converted into displacement and thus the thermal expansion does
not cause significant deflection
Figure 3-4 Definition of temperature, Young’s modulus and displacement component in the
cross-section (Lei et al., 2017).
Let (u1, w) and (u2, w) be the axial and transverse displacements at the centroid points of
the two T-sections (see Figure 3-4) The strain energy of the two T-sections of
castellated beams U1 and U2 due to transverse loads can be expressed as follows:
𝑙 2
1 𝑑𝑢1 2 𝑑2 𝑤 3-59
𝑈1 = ∫ [𝐸1 𝐴1 ( ) + 𝐸1 𝐼1 ( 2 ) ] 𝑑𝑥
2 𝑑𝑥 𝑑𝑥
0
𝑙 2
1 𝑑𝑢2 2 𝑑2𝑤 3-60
𝑈2 = ∫ [𝐸2 𝐴2 ( ) + 𝐸2 𝐼2 ( 2 ) ] 𝑑𝑥
2 𝑑𝑥 𝑑𝑥
0
where A1 and A2 are the areas of the two T-sections, E1 and E2 are the Young's modulus
of the two T-sections and I1 and I2 are the second moments of areas of the two
T-sections about their own centroid axes, and l is the length of the beam.
The average shear strain of the mid-part of the web 𝛾𝑥𝑧 is expressed as follows:
𝑒 𝑑𝑤 𝑢1 − 𝑢2 3-61
𝛾𝑥𝑧 (𝑥, 𝑧) = −
𝑎 𝑑𝑥 2𝑎
40
Thus the shear strain energy of the mid-part of the web Ush is expressed as follows:
𝑙
𝑡𝑤 𝑒 2 (𝐸1 + 𝐸2 )𝑘𝑠ℎ 𝑑𝑤 𝑢1 − 𝑢2 2
𝑈𝑠ℎ = ∫( − ) 𝑑𝑥 3-62
4(1 + 𝑣)𝑎 𝑑𝑥 2𝑒
0
𝑙 2
1 𝑑𝑢1 2 𝑑2𝑤
∏ = ∫ [𝐸1 𝐴1 ( ) 𝐸1 𝐼1 ( 2 ) ] 𝑑𝑥
2 𝑑𝑥 𝑑𝑥
0
𝑙 2
1 𝑑𝑢2 2 𝑑2𝑤 3-63
+ ∫ [𝐸2 𝐴2 ( ) 𝐸2 𝐼2 ( 2 ) ] 𝑑𝑥
2 𝑑𝑥 𝑑𝑥
0
𝑙
𝑡𝑤 𝑒 2 (𝐸1 + 𝐸2 )𝑘𝑠ℎ 𝑑𝑤 𝑢1 − 𝑢2 2
× ∫( − ) 𝑑𝑥 − 𝑊
4(1 + 𝑣)𝑎 𝑑𝑥 2𝑒
0
𝛿(𝑈𝑇 + 𝑈𝑠ℎ − 𝑊) = 0 3-64
For a simply supported castellated beam:𝑢α (𝑥) , 𝑢β (𝑥), and w(x) can be assumed as
follows:
𝑚𝜋𝑥 3-65
𝑢α (𝑥) = ∑ 𝐴𝑚 cos
𝑙
𝑚=1,2,..
𝑚𝜋𝑥 3-66
𝑢𝛽 (𝑥) = ∑ 𝐵𝑚 cos
𝑙
𝑚=1,2,..
𝑚𝜋𝑥 3-67
𝑤(𝑥) = ∑ 𝐶𝑚 sin
𝑙
𝑚=1,2,..
where 𝐴𝑚 , 𝐵𝑚 , and 𝐶𝑚 (𝑚 = 1,2, … ) are the constants to be determined, it is obvious

that the displacement functions assumed Eqs. (3-65), (3-66) and (3-67) satisfy the
𝑑2 𝑤 𝑑𝑢𝛼 𝑑𝑢𝛽
simply support boundary conditions, that are 𝑤 = = 0, and = = 0 at x = 0
and x = l.
Substituting Eqs. (3-65)- (3-67) into (3-63) by using Eqs. (3-8) and (3-9), according to
the principle of minimum potential energy, it yields,
𝐸2
𝐴𝑚 = − 𝐵 3-68
𝐸1 𝑚
Note that for castellated beams or columns, A1 = A2 = Atee and, I1 = I2 = Itee . Thus:
41
(𝐸1 + 𝐸2 )𝐼𝑡𝑒𝑒 (𝑚𝜋)5

[ ] 𝐶3 − 2𝑞(1 − (−1)𝑚 )
𝑙4 3-69
𝐵𝑚 =
2𝐸2 𝐴𝑡𝑒𝑒 𝑒(𝑚𝜋)4
𝑙3
2𝑞(1 − (−1)𝑚 )𝑙4 1

Cm =
(𝐸1 + 𝐸2 )𝐼𝑡𝑒𝑒 + 3-70
(𝑚𝜋)5
4𝐸2 𝐴𝑡𝑒𝑒 𝑒2 (𝐸1 + 𝐸2 )𝐸1 𝐸2

( )
8𝑎𝐸1 𝐸2 (1 + 𝑣)(𝑚𝜋)2 𝐴𝑡𝑒𝑒
(𝐸1 + 𝐸2 )2 (1 + )
[ (𝐸1 + 𝐸2 )2 𝑙2 𝑡𝑤 𝑘𝑠ℎ ]
2𝑞𝑙 4 (1 − (−1)𝑚 )
𝐶𝑚 = [1
4𝐴 𝑒 2 𝐸1 𝐸2 (𝑚𝜋)5
(𝐸1 + 𝐸2 ) [𝐼𝑡𝑒𝑒 + 𝑡𝑒𝑒 ]
(𝐸1 + 𝐸2 )2
4𝐴𝑡𝑒𝑒 𝑒 2 𝐸1 𝐸2 8𝑎𝐸1 𝐸2 (1 + 𝑣)(𝑚𝜋)2 𝐴𝑡𝑒𝑒
+ ×
4𝐴 𝑒 2 𝐸1 𝐸2 (𝐸1 + 𝐸2 )2 𝑙 2 𝑡𝑤 𝑘𝑠ℎ
𝐼𝑡𝑒𝑒 + 𝑡𝑒𝑒 2 3-71
(𝐸1 + 𝐸2 )
1
× ]
(𝐸1 + 𝐸2 )2
Eq. (3-70) can be further simplified as:
Therefore, the expression of the castellated beam deflection is expressed as follows:
2𝑞𝑙 4 (1 − (−1)𝑚 )
𝑤(𝑥) = [1
4𝐴 𝑒 2 𝐸1 𝐸2 (𝑚𝜋)5
(𝐸1 + 𝐸2 ) [𝐼𝑡𝑒𝑒 + 𝑡𝑒𝑒 ]
(𝐸1 + 𝐸2 )2
4𝐴𝑡𝑒𝑒 𝑒 2 𝐸1 𝐸2 8𝑎𝐸1 𝐸2 (1 + 𝑣)(𝑚𝜋)2 𝐴𝑡𝑒𝑒
+ × 3-72
4𝐴 𝑒 2 𝐸1 𝐸2 (𝐸1 + 𝐸2 )2 𝑙 2 𝑡𝑤 𝑘𝑠ℎ
𝐼𝑡𝑒𝑒 + 𝑡𝑒𝑒
(𝐸1 + 𝐸2 )2
1 𝑚𝜋𝑥
× ] 𝑠𝑖𝑛
(𝐸1 + 𝐸2 )2 𝑙
As the maximum deflection of uniformly loaded simply supported beam takes place at
𝑙
𝑥 = , the maximum deflection can be expressed as follows:
2
42
2𝑞𝑙 4 2(−1)𝑘+1
𝑤|𝑥=𝑙/2 = [ ∑
4𝐴 𝑒 2 𝐸1 𝐸2 𝜋 5 (2𝑘 − 1)5
(𝐸1 + 𝐸2 ) [𝐼𝑡𝑒𝑒 + 𝑡𝑒𝑒 ] 𝑘=1,2,..
(𝐸1 + 𝐸2 )2
2
4𝐴𝑡𝑒𝑒 𝑒 𝐸1 𝐸2 8𝑎𝐸1 𝐸2 (1 + 𝑣)(𝑚𝜋)2 𝐴𝑡𝑒𝑒
+ × 3-73
4𝐴 𝑒 2 𝐸1 𝐸2 (𝐸1 + 𝐸2 )2 𝑙 2 𝑡𝑤 𝑘𝑠ℎ
𝐼𝑡𝑒𝑒 + 𝑡𝑒𝑒 2
(𝐸1 + 𝐸2 )
1 2(−1)𝑘+1
× × ∑ ]
(𝐸1 + 𝐸2 )2 𝜋 3 (2𝑘 − 1)5
𝑘=1,2,..
Substituting Eqs. (3-46), (3-47) and (3-48) into (3-73) gives the maximum deflection:
5𝑞𝑙 4 𝑞𝑙 2 𝑎(1 + 𝑣)
𝑤|𝑥=𝑙/2 = + ×
4𝐴 𝑒 2 𝐸1 𝐸2 (𝐸1 + 𝐸2 ) 𝑡𝑤 𝑘𝑠ℎ
384(𝐸1 + 𝐸2 ) [𝐼𝑡𝑒𝑒 + 𝑡𝑒𝑒 ]
(𝐸1 + 𝐸2 )2
2 3-74
2𝑒𝐴𝑡𝑒𝑒 𝐸1 𝐸2
×[ ]
4𝐴 𝑒 2 𝐸1 𝐸2
𝐼𝑡𝑒𝑒 + 𝑡𝑒𝑒 (𝐸1 + 𝐸2 ) 2
(𝐸1 + 𝐸2 )2
Eq. (3-74) reflects the influence of non-uniform material properties caused by the non-
uniform temperature on the maximum deflection of the castellated beam when the shear
effect is considered.
It is obvious that, the first part of Eq. (3-74) indicates the deflection generated by the
bending load but affected by temperature, while the second part provides the deflection
generated by the shear load also affected by temperature. Moreover, Eq. (3-74) shows
that rising or reducing shear effect depends on the cross-section area of major axis as
well as the difference in temperatures between the two T- sections.
3.2.4. Deflection of non-uniform temperature distribution roller- fixed supported

castellated beam with uniformly distributed loads.
Note that the thermal expansion induced by temperature was not considered in this
study because the axial displacement can be freely expanded at the side of roller
boundary in the beam; so there will be no thermal stresses in the beam. All thermal
strains are converted into displacement and thus the thermal expansion does not cause
significant deflection.
43
For roller-fixed castellated beam uα(x), uβ(x) and w(x) can be assumed as follows:
(2𝑘 − 1)𝜋(𝑙 − 𝑥)
𝑢𝛼 (𝑥) = ∑ 𝐹𝑘 [(2𝑘 − 1) sin ( )
2𝑙
𝑚=1,2,.. 3-75
(2𝑘 + 1)𝜋(𝑙 − 𝑥)
− (2𝑘 + 1) sin ( )]
2𝑙
(2𝑘 − 1)𝜋(𝑙 − 𝑥)
𝑢𝛽 (𝑥) = ∑ 𝐸𝑘 [(2𝑘 − 1) sin ( )
2𝑙
𝑚=1,2,.. 3-76
(2𝑘 + 1)𝜋(𝑙 − 𝑥)
− (2𝑘 + 1) sin ( )]
2𝑙
𝑘𝜋(𝑙 − 𝑥) 𝜋(𝑙 − 𝑥)
𝑤(𝑥) = ∑ 𝐷𝑘 sin ( ) sin ( ) 3-77
𝑙 2𝑙
𝑚=1,2,..
where Fk, Ek, and Dk (k = 1, 2 …) are the constants to be determined. It is obvious that
the displacement functions assumed satisfy the pinned-fixed support boundary
𝑑2 𝑤 𝑑𝑢𝛽 𝑑𝑤
conditions, that are 𝑤 = = 0, and = 0 at x = 0 and 𝑤 = = 𝑢𝛼 = 𝑢𝛽 = 0
at x = l.
Substituting Eqs. (3-75), (3-76) and (3-77) into (3-63), and according to the principle of
minimum potential energy, it yields:
−256𝑞𝑘𝑙 4 𝑍 cos(𝑘𝜋)
𝐷𝑘 = 5 3-78
𝜋 (16𝑘 4 + 24𝑘 2 + 1)(4𝑘 2 − 1)(𝐸1 +𝐸2 )
where
1
[16𝐴𝑡𝑒𝑒 𝐸1 𝐸2 𝜋 2 𝑎𝑘 2 (1 + 𝑣) (2𝑘 2 + 3 + ) + 𝑘𝑠ℎ 𝑡𝑤 𝑙 2 (𝐸1 +𝐸2 )2 (4𝑘 2 + 1)]
𝑍= 8𝑘 2
1
(16𝐴𝑡𝑒𝑒 𝐸1 𝐸2 𝐼𝑡𝑒𝑒 𝜋 2 𝑎𝑘 2 (1 + 𝑣) (2𝑘 2 + 3 + 2 ) + 𝐼𝑡𝑒𝑒 𝑘𝑠ℎ 𝑡𝑤 𝑙 2 (𝐸1 +𝐸2 )2 (4𝑘 2 + 1)) 3-79
[ 8𝑘 ]
+4(4𝑘 2 + 1)𝐴𝑡𝑒𝑒 𝑘𝑠ℎ 𝑡𝑤 𝑙 2 𝐸1 𝐸2 𝑒 2
𝑘𝜋(𝑙 − 𝑥) 𝜋(𝑙 − 𝑥)
−256𝑞𝑘𝑙 4 cos(𝑘𝜋) 𝑍 sin ( ) sin ( )
𝑙 2𝑙 3-80
w(𝑥) = ∑
𝜋 5 (16𝑘 4 + 24𝑘 2 + 1)(4𝑘 2 − 1)(𝐸1 +𝐸2 )
𝑘=1,2,3
Eq. (3-80) reflects the influence of non-uniform material properties caused by the non-
uniform temperature on the maximum deflection of pinned-fixed castellated beam when
the shear effect is considered.
44
Clearly, the denominator of Eq. (3-80) is very complicated, and the position of the
maximum deflection point is at (x= 0.375l ). The Eq. (3-80) is not easy to simplify but it
is regarded as Bernoulli-Euler beam with modifications.
3.3. Numerical analysis
Finite Element Analysis (FEA) is a numerical method that has been implemented to
simulate any physical phenomenon for finding out new concepts. Essentially, it was
used for solving cases of structural mechanics, sometimes instead of experimental
techniques as well as to validate versus the experimental or analytical solution.
In general, FEA suits for solving the problems such as heat transfer, fluid mechanics,
and manufacturing modelling in engineering fields by solving mathematical equations
numerically and therefore is named as numerical method.
At present, the accuracy and flexibility of this method have led to it becoming
commonplace in the analysis and design of steel structures consisting of thin-walled
members with web openings exposed to various types of loading and different boundary
conditions.
The ANSYS software is the finite element analysis tool that is adopted in this study for
numerical analysis of castellated beams to verify the accuracy of the analytical solutions
of transverse deflection of castellated beams subjected to uniformly distributed
transverse load.
3.3.1. The modeling of castellated beams
ANSYS software library has different types of elements for the analysis of different
types of structures. Previous studies have shown that the use of three-dimensional (3D)
finite element model was successful in idealizing the structural behaviour of castellated
beams (El-Sawy et al., 2009).
45
The modelling of castellated beams is carried out by using 3D linear 4-Node Thin Shell
Elements (SHELL181) depicted in Figure 3-5. It is designed for modelling thin-walled
members because one dimension of the element is very small compared with another
two dimensions; also for linear, large rotation, and/or large strain nonlinear applications.
This element presents four nodes with six DOF per node, i.e., translations and rotations
on the X, Y, and Z-axis, respectively.
Figure 3-5 Shell element ANSYS library.
3.3.2. Meshing consideration and material model
With regard to the mesh size of the model, the short length beams were meshed by
using element size of 5 mm, whereas for the long length beams the element size of 10
mm was used. A typical mesh configuration for castellated beam is shown in
Figure 3-6. The material model of the castellated beam uses linear elastic material with
Young’s modulus E = 2.1 × 105 MPa, Poisson’s ratio v =0.3 and yield stress σy=275
MPa.
Figure 3-6 A typical mesh configuration
46
3.3.3. Boundary conditions and loading
3.3.3.1. Simply supported castellated beam
For simply supported beam, a half-length analysis model is used because of the
symmetry. Therefore, it has led to reducing the number of elements and nodes, also
saving time. The displacement boundary conditions are applied to all nodes at the two
ends, as shown in Figure 3-7. The lateral and transverse deflections and rotation are
restrained (uy=uz=0 and θx=0) at the simply supported end, while the symmetric
boundary condition is applied at the other end by restricting the axial displacement and
rotations about the two axes within the cross-section (ux=0 and θy=θz=0).
A line load is used to model the uniformly distributed transverse load, which acts on the
junction of the flange and the web. The uniformly distributed load (𝑞𝑚𝑎𝑥 ) is considered
after multiplying with beam’s half-length and then divided by the numbers of the nodes
on the line defining the junction of the flange and the web nodes, which gives the load
applying on each node.
3.3.3.2. Pinned-fixed supported castellated beam
The boundaries of one end simply supported and another fixed are applied in the model.
Full length of the castellated beam is used. The displacement boundary conditions are
applied to all nodes at the two ends, as shown in Figure 3-7. The lateral and transverse
deflections and rotation are restrained (uy=uz=0 and θx=0) at the simply supported end,
while the fixed support boundary condition is applied at the other end by restraining the
axial displacement, transverse deflections and rotations around the three axes within the
cross-section (ux=uy=uz=0 and θx=θy=θz=0).
The uniformly distributed load is also applied at the junction between the flange and
web as a line load, which is similar to the simply supported beam as described above.
47
Figure 3-7 The typical loading method and boundary conditions in finite element model of
castellated beam.
3.3.4. The dimensions of parameter study and the value of load
Ten groups of selected beams with different lengths are analysed, each of them having
different cross sections. The (hw/tw) ratio of the selected sections is (37.5) and the (bf /tf)
ratio of the selected sections ranges from 10 to 25, as tabulated in Table (3-1).
Table 3-1 Parameters of castellated beams groups considered
the
Name
opening l l/2 bf tf hw tw a qmax
Group of
panel mm mm mm mm mm mm mm N/mm
beam
(unit)
A1 2 692.82 346.41 100 10 300 8 100 1739.76
A2 2 692.82 346.41 150 10 300 8 100 2428.21
A
A3 2 692.82 346.41 200 10 300 8 100 3116.67
A4 2 692.82 346.41 250 10 300 8 100 3805.12
B1 4 2078.46 1039.23 100 10 300 8 100 193.31

B2 4 2078.46 1039.23 150 10 300 8 100 269.80
B
B3 4 2078.46 1039.23 200 10 300 8 100 346.30
B4 4 2078.46 1039.23 250 10 300 8 100 422.79
C1 10 3464.16 1732.05 100 10 300 8 100 69.59
C2 10 3464.16 1732.05 150 10 300 8 100 97.13

C
C3 10 3464.16 1732.05 200 10 300 8 100 124.67
C4 10 3464.16 1732.05 250 10 300 8 100 152.20
48
the
Name l l/2 bf tf h w tw a qmax
Group opening
of N/mm
panel mm mm mm mm mm mm mm
beam
(unit)
D1 12 4156.92 2078.46 100 10 300 8 100 48.33
D2 12 4156.92 2078.46 150 10 300 8 100 67.45

D
D3 12 4156.92 2078.46 200 10 300 8 100 86.57
D4 12 4156.92 2078.46 250 10 300 8 100 105.69
E1 14 4849.74 2424.87 100 10 300 8 100 35.50
E2 14 4849.74 2424.87 150 10 300 8 100 49.55

E
E3 14 4849.74 2424.87 200 10 300 8 100 63.60
E4 14 4849.74 2424.87 250 10 300 8 100 77.65
F1 16 5542.56 2771.28 100 10 300 8 100 27.18
F2 16 5542.56 2771.28 150 10 300 8 100 37.94

F
F3 16 5542.56 2771.28 200 10 300 8 100 48.69
F4 16 5542.56 2771.28 250 10 300 8 100 59.45
G1 18 6235.38 3117.69 100 10 300 8 100 21.48
G2 18 6235.38 3117.69 150 10 300 8 100 29.97

G
G3 18 6235.38 3117.69 200 10 300 8 100 38.48
G4 18 6235.38 3117.69 250 10 300 8 100 46.98

H1 26 9006.66 4503.33 100 10 300 8 100 10.29
H2 26 9006.66 4503.33 150 10 300 8 100 14.37

H
H3 26 9006.66 4503.33 200 10 300 8 100 18.44
H4 26 9006.66 4503.33 250 10 300 8 100 22.51

I1 36 12470.77 6235.38 100 10 300 8 100 5.37
I2 36 12470.77 6235.38 150 10 300 8 100 7.49
I
I3 36 12470.77 6235.38 200 10 300 8 100 9.62
I4 36 12470.77 6235.38 250 10 300 8 100 11.74
J1 42 14549.23 7274.61 100 10 300 8 100 3.95
J2 42 14549.23 7274.61 150 10 300 8 100 5.51
J
J3 42 14549.23 7274.61 200 10 300 8 100 7.07
J4 42 14549.23 7274.61 250 10 300 8 100 8.63
49
3.3.5. Comparison of results and discussion
The details of comparisons between the analytical results and finite element analysis
results are given in the tables and plotted in the figures. These results indicate the shear
effect of web openings on the transverse deflection. It can be noticed from the
comparisons that the present analytical solution, taking into account shear, agrees well
with the result provided by the finite element analysis.
3.3.5.1. Maximum deflections of simply supported beams subjected to uniformly

distributed loads
Table (3-2) shows the results of maximum deflections of simply supported castellated
beams with a uniformly distributed load for different lengths with various cross-section
dimensions, obtained using Eq. (3-49). In the results, T1 is based on Eq. (3-28); T2
1 1
uses(𝐾𝑠ℎ = 0.78 × 4); T3 uses(𝐾𝑠ℎ = 4); T4 is obtained from Eq. (3-50); and T5 is the
finite element analysis results. The results are also plotted in Figure 3-8.
Table (3-3) shows the relative errors of the maximum deflections of simply supported
castellated beam with a uniformly distributed load, which are obtained based on the
finite element numerical solutions that are described in Table (3-2), between analytical
solutions using different shear rigidity factors obtained by Eq. (3-49), including one
with zero shear factor obtained by Eq. (3-50) for four castellated beams of different
flange widths.
Figure 3-8 shows a comparison of the maximum deflections between analytical

solutions using different shear rigidity factors including one with zero shear factor and
FEA numerical solution for different beam lengths with various 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. In addition, 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
50
numerical solution. Figure 3-9 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
5.0% for all of discussed four sections in all the beam length range (l >3 meters). For
beams less than 3 meters the error is about 11.0% which is quite big. However, for most
practically used beams the length will be longer than 3 meters. 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.
3.3.5.2. Maximum deflection of pinned-fixed castellated beam due to uniformly

distributed loads
Table (3-4) shows the results of maximum deflections of pinned-fixed castellated

beams with a uniformly distributed load for different lengths with various cross-section
dimensions, obtained using Eq. (3-57). In the results, T1 is based on Eq. (3-28); T2
1 1
uses(𝐾𝑠ℎ = 0.78 × 4); T3 uses(𝐾𝑠ℎ = 4); T4 is obtained from Eq. (3-58); and T5 is the
finite element analysis results. The results are also plotted in Figure 3-10.
Table (3-5) shows the relative errors of the maximum deflections of pinned-fixed
castellated beam with a uniformly distributed load, which are obtained based on the
finite element numerical solutions that are described in Table (3-4), between analytical
solutions using different shear rigidity factors obtained by Eq. (3-57), including one
with zero shear factor obtained by Eq. (3-58) for four castellated beams of different
flange widths.
Figure 3-10 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 different beam lengths with various flange widths. It can be
seen from the figure that, the analytical solution using the proposed shear factor again 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
51
affected by the ratio of the flange width to the beam length. In addition, 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 3-11 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 relative error of the analytical solutions using the present shear rigidity factor does
not exceed 23% for all of discussed four sections in all of the beam length range
(l >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. Compared to the
simply supported beams, the error of the analytical solution of the pinned-fixed beams
seems to be larger. This is probably due to the deflection functions assumed for pinned-
fixed beams that have slow convergence.
52
Table 3-2 Comparison of results of maximum deflections of simply supported castellated beam
with a uniformly distributed load obtained using different approaches
Name
bf T1 T2 T3 T4 T5
of
mm mm mm mm mm mm
beam
A1 2.66 2.21 1.86 0.41 3.00

B1 5.74 5.56 5.19 3.68 6.12
C1 12.24 12.12 11.74 10.23 12.35
D1 16.77 16.62 16.24 14.74 16.65
100
E1 22.04 21.89 21.46 20.06 21.71
F1 28.00 28.03 27.70 26.19 27.60
G1 35.13 34.99 34.66 33.15 34.60
H1 70.80 71.00 70.67 69.16 69.00
A2 3.84 2.86 2.39 0.41 4.18
B2 6.71 6.24 5.73 3.68 7.11
C2 13.02 12.79 12.28 10.23 13.40
D2 17.50 17.30 16.78 14.74 17.75
150
E2 22.80 22.62 22.11 20.05 22.88
F2 28.92 28.76 28.24 26.19 28.81
G2 35.87 35.71 35.20 33.15 35.50
H2 70.89 71.73 71.21 69.16 70.00
A3 5.53 3.50 2.85 0.41 5.31
B3 7.74 6.91 6.21 3.68 8.10
C3 13.85 13.47 12.76 10.23 14.43
D3 18.29 17.98 17.26 14.74 18.80
200
E3 23.57 23.30 22.58 20.05 23.91
F3 29.69 29.44 28.72 26.19 29.94
G3 36.63 36.40 35.681 33.15 36.69
H3 72.59 72.41 71.69 69.16 71.74
A4 7.33 4.23 3.43 0.41 6.90
B4 8.50 7.69 6.81 3.68 9.10
C4 14.68 14.25 13.37 10.23 15.42
D4 19.11 18.76 17.87 14.74 19.80
250
E4 24.50 24.08 23.19 20.05 25.20
F4 30.48 30.22 29.33 26.19 31.00
G4 37.40 37.18 36.29 33.15 37.80
H4 73.35 73.29 72.30 69.16 72.99
53
Figure 3-8 Maximum deflections of simply supported castellated beam with a uniformly
distributed load between analytical solutions using different shear rigidity factors obtained by
Eq. (3-49), including one with zero shear factor obtained by Eq. (3-50) and FEA numerical
solution for different beam lengths with various flange widths. (a) bf =100mm, (b) bf =150mm,
(c) bf =200mm and (d) bf =250mm. (hw=300mm, tf=10mm, tw=8mm and a=100mm)
54
Table 3-3 The relative errors of maximum deflections of simply supported castellated beam
with a uniformly distributed load obtained based on the finite element numerical solutions
Name
bf
of T1 T2 T3 T4
mm
beam
A1 0.11 0.26 0.38 0.86

B1 0.06 0.09 0.15 0.40
C1 0.01 0.02 0.05 0.17
D1 -0.01 0.00 0.02 0.11
100
E1 -0.02 -0.01 0.01 0.08
F1 -0.01 -0.02 0.00 0.05
G1 -0.02 -0.01 0.00 0.04
H1 -0.03 -0.03 -0.02 0.00
A2 0.08 0.32 0.43 0.90
B2 0.06 0.12 0.19 0.48
C2 0.03 0.05 0.08 0.24
D2 0.01 0.03 0.05 0.17
150
E2 0.00 0.01 0.03 0.12
F2 0.00 0.00 0.02 0.09
G2 -0.01 -0.01 0.01 0.07
H2 -0.01 -0.02 -0.02 0.01
A3 0.00 0.34 0.46 0.92
B3 0.04 0.15 0.23 0.55
C3 0.04 0.07 0.12 0.29
D3 0.03 0.04 0.08 0.22
200
E3 0.01 0.03 0.06 0.16
F3 0.01 0.02 0.04 0.13
G3 0.00 0.01 0.03 0.10
H3 -0.01 -0.01 0.00 0.04
A4 -0.01 0.39 0.50 0.94
B4 0.07 0.15 0.25 0.60
C4 0.05 0.08 0.13 0.34
D4 0.04 0.05 0.10 0.26
250
E4 0.03 0.04 0.08 0.20
F4 0.02 0.03 0.05 0.16
G4 0.01 0.02 0.04 0.12
H4 0.00 0.00 0.01 0.05
55
Figure 3-9 Relative errors of maximum deflections of simply supported castellated beam with a
uniformly distributed load between analytical solutions using different shear rigidity factors
obtained by Eq. (3-49), including one with zero shear factor obtained by Eq. (3-50) and FEA
numerical solution for different beam lengths with various flange widths. (a) bf =100mm,
(b) bf =150mm, (c) bf =200mm and (d) bf =250mm. (hw=300mm, tf=10mm, tw=8mm and
a=100mm)
56
Table 3-4 Comparison of results of maximum deflections of pinned-fixed castellated beam with
a uniformly distributed load obtained using different approaches
Name
bf T1 T2 T3 T4 T5
of
mm mm mm mm mm mm
beam
B1 3.30 3.05 2.71 1.47 4.36

C1 5.80 5.69 5.35 4.09 6.99
D1 7.60 7.50 7.15 5.88 8.80
E1 9.73 9.64 9.29 8.01 10.91
100
F1 12.18 12.10 11.75 10.46 13.33
G1 14.97 14.89 14.54 13.24 16.08
H1 29.41 29.34 28.99 27.62 30.36
I1 54.86 54.80 54.45 52.95 55.48
B2 4.10 3.62 3.15 1.47 5.39
C2 6.46 6.27 5.79 4.09 8.06
D2 8.24 8.08 7.60 5.88 9.92
E2 10.36 10.21 9.74 8.01 12.06
150
F2 12.82 12.68 12.20 10.46 14.52
G2 15.60 15.47 14.99 13.24 17.30
H2 30.03 29.92 29.44 27.62 31.78
I2 55.46 55.37 54.89 52.95 57.23
B3 4.88 4.19 3.60 1.47 6.42
C3 7.14 6.84 6.24 4.09 9.12
D3 8.91 8.65 8.05 5.88 11.00
E3 11.02 10.79 10.19 8.01 13.18
200
F3 13.46 13.25 12.65 10.46 15.67
G3 16.24 16.04 15.44 13.24 18.46
H3 30.65 30.50 29.89 27.62 33.05
I3 56.08 55.34 55.34 52.95 58.69
B4 5.80 4.76 4.04 1.47 7.50
C4 7.85 7.41 6.69 4.09 10.18
D4 9.60 9.22 8.50 5.88 12.08
E4 11.70 11.36 10.63 8.01 14.28
250
F4 14.13 13.83 13.10 10.46 16.79
G4 16.90 16.62 15.89 13.24 19.60
H4 31.29 31.07 30.34 27.62 34.26
I4 56.70 56.52 55.79 52.95 60.20
57
Figure 3-10 Maximum deflections of pinned-fixed castellated beam with a uniformly

distributed load between analytical solutions using different shear rigidity factors obtained by
Eq. (3-57), including one with zero shear factor obtained by Eq. (3-58) and FEA numerical
solution for different beam lengths with various flange widths. (a) bf =100mm, (b) bf =150mm,
(c) bf =200mm and (d) bf =250mm. (hw=300mm, tf=10mm, tw=8mm and a=100mm)
58
Table 3-5 The relative errors of maximum deflections of pinned-fixed castellated beam with a
uniformly distributed load obtained based on the finite element numerical solutions
Name
bf
of T1 T2 T3 T4
mm
beam
B1 0.24 0.30 0.38 0.66

C1 0.17 0.19 0.24 0.42
D1 0.14 0.15 0.19 0.33
E1 0.11 0.12 0.15 0.27
100
F1 0.09 0.09 0.12 0.22
G1 0.07 0.07 0.10 0.18
H1 0.03 0.03 0.04 0.09
I1 0.01 0.01 0.02 0.05
B2 0.24 0.33 0.41 0.73
C2 0.20 0.22 0.28 0.49
D2 0.17 0.19 0.23 0.41
E2 0.14 0.15 0.19 0.34
150
F2 0.12 0.13 0.16 0.28
G2 0.10 0.11 0.13 0.23
H2 0.06 0.06 0.07 0.13
I2 0.03 0.03 0.04 0.07
B3 0.24 0.35 0.44 0.77
C3 0.22 0.25 0.32 0.55
D3 0.19 0.21 0.27 0.47
E3 0.16 0.18 0.23 0.39
200
F3 0.14 0.15 0.19 0.33
G3 0.12 0.13 0.16 0.28
H3 0.07 0.08 0.10 0.16
I3 0.04 0.06 0.06 0.10
B4 0.23 0.37 0.46 0.80
C4 0.23 0.27 0.34 0.60
D4 0.21 0.24 0.30 0.51
E4 0.18 0.20 0.26 0.44
250
F4 0.16 0.18 0.22 0.38
G4 0.14 0.15 0.19 0.32
H4 0.09 0.09 0.11 0.19
I4 0.06 0.06 0.07 0.12
59
Figure 3-11 Relative errors of maximum deflections of pinned-fixed castellated beam with a
uniformly distributed load between analytical solutions using different shear rigidity factors
obtained by Eq. (3-57), including one with zero shear factor obtained by Eq. (3-50) and FEA
numerical solution for different beam lengths with various flange widths. (a) bf =100mm,
a=100mm)
60
3.3.5.3. Deflection of simply supported castellated beam due to three fire scenarios
with a uniformly distributed load
Table (3-6) shows three fire scenarios with defined temperatures and material
properties that were applied on castellated beams.
Table 3-6 Temperatures and Young’s modulus in two T- sections (Eo is the Young’s modulus at
ambient temperature) (BSEN1993-1-2:2005)
Parameter Case 1 Case 2 Case 3
Temperature, T1 550 oC 450 oC 350 oC

Temperature, T2 100 oC 200 oC 300 oC
Young’s modulus, E1 0.5Eo 0.6Eo 0.7Eo
Young’s modulus, E2 Eo 0.9Eo 0.8Eo
Table (3-7) describes the results of maximum deflections of simply supported

castellated beams due to the three fire scenarios with a uniformly distributed load for
different beam lengths with various flange widths. These results are determined by
Eq. (3-74), including one without fire scenarios obtained by Eq. (3-49) based on
Eq. (3-28) for various 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 3-12 shows the maximum deflections of simply supported castellated beam due
to the three fire scenarios defined in Table (3-6) with a uniformly distributed load
between analytical solutions obtained by Eq. (3-74), including one without fire
scenarios obtained by Eq. (3-49) based on Eq. (3-28) for various flange widths:
(a) bf =100mm, (b) bf =150mm, (c) bf =200mm and (d) bf =250mm. (hw=300mm,
tf=10mm, tw=8mm and a=100mm).
The figure reflects the shear effect due to web opening on the deformation of the beam
in different fire scenarios when subjected to a uniformly distributed load. It can be seen
from the figure that, in each group of flange width, the curves of the maximum
deflection have a similar variation pattern. However, the maximum deflection of the
61
three examined cases is different, in spite of the average temperatures of them being the
same. The increase in the deflection is directly proportional to the amount of difference
in temperatures between the two T-sections. Thus, case one, which has the largest
temperature difference, is the worst case
Table 3-7 Comparison of results of maximum deflections of simply supported castellated beam
due to three fire scenarios with a uniformly distributed load
Name Max. deflection (mm) due to

bf of
mm
beam (case 1+qmax) (case 2+qmax) (case 3+qmax) ( qmax )
C1 15.32 14.20 13.70 12.24

E1 30.04 27.84 26.85 22.04
100
G1 49.65 46.02 44.39 35.13
H1 103.59 96.02 92.62 71.12
C2 15.33 14.20 13.70 13.02
E2 30.04 27.84 26.85 22.80
150
G2 49.67 46.02 44.39 35.87
H2 103.62 96.02 92.62 71.85
C3 15.33 14.21 13.70 13.83
E3 30.05 27.84 26.85 23.57
200
G3 49.68 46.02 44.39 36.63
H3 103.64 96.03 92.62 72.59
C4 15.33 14.21 13.70 14.68
E4 30.05 27.84 26.85 24.38
250
G4 49.68 46.03 44.39 37.40
H4 103.66 96.03 92.62 73.35
62
Figure 3-12 Maximum deflections of simply supported castellated beam due to three fire
scenarios with a uniformly distributed load between analytical solutions obtained by Eq. (3-74),
including one without fire scenarios obtained by Eq. (3-49) for different beam lengths with
various flange widths. (a) bf =100mm, (b) bf =150mm, (c) bf =200mm and (d) bf =250mm.
(hw=300mm, tf=10mm, tw=8mm and a=100mm).
63
3.3.5.4. Deflection of roller-fixed castellated beam due to three fire scenarios with
a uniformly distributed
Table (3-8) gives the results of maximum deflections of roller-fixed castellated beams
in three different fire scenarios with a uniformly distributed load for different beam
lengths with various flange widths. These results are determined by Eq. (3-80),
including one without fire scenarios obtained by Eq. (3-57) based on Eq. (3-28) for
various 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 3-13 plots the maximum deflections of roller-fixed castellated beams due to the
three fire scenarios as given in Table (3-6), with a uniformly distributed load, obtained
from the analytical solutions presented by Eq. (3-80), including one without fire
scenarios obtained by using Eq. (3-57) based on Eq. (3-28) for various flange widths: (a)
bf =100mm, (b) bf =150mm, (c) bf =200mm and (d) bf =250mm. (hw=300mm, tf=10mm,
tw=8mm and a=100mm).
The figure reflects the shear effect due to web opening on the maximum deflection of
castellated beam in different fire scenarios. It can be seen that, in each group of flange
width, the curves of the maximum deflection have a similar variation pattern. Moreover,
it can be seen that, the maximum deflection of the three examined cases is different, in
spite of the average temperatures in them being the same, where the increase in the
deflection of the castellated beam is directly proportional to the amount of difference in
temperatures between the two T- sections. Thus, case one, which has the largest
temperature difference, is the worst case, which reflects that the increase of deflection is
influenced by the reduction of Young's modulus.
64
Table 3-8 Comparison of results of maximum deflections of roller- fixed castellated beam due
to three fire scenarios with a uniformly distributed load
Name Max. deflection (mm) due to

bf of
mm
beam (case 1+qmax) (case 2+qmax) (case 3+qmax) ( qmax )
C1 8.73 8.27 8.06 5.80

E1 14.79 13.89 13.48 9.73
100
G1 22.89 21.39 20.72 14.97
H1 45.19 42.06 40.66 29.41
C2 9.66 9.20 8.99 6.46
E2 15.70 14.79 14.38 10.36
150
G2 23.79 22.28 21.61 15.60
H2 46.07 42.93 41.53 30.03
C3 10.63 10.17 9.96 7.14
E3 16.63 15.72 15.31 11.02
200
G3 24.70 23.19 22.52 16.24
H3 46.96 43.82 42.41 30.65
C4 11.64 11.17 10.97 7.85
E4 17.58 16.68 16.27 11.70
250
G4 25.63 24.12 23.45 16.90
H4 47.87 44.72 43.31 31.29
65
Figure 3-13 Maximum deflections of roller-fixed castellated beam due to three fire scenarios
with a uniformly distributed load between analytical solutions obtained by Eq. (3-80), including
one without fire scenarios obtained by Eq. (3-57) for different beam lengths with various flange
widths. (a) bf =100mm, (b) bf =150mm, (c) bf =200mm and (d) bf =250mm. (hw=300mm,
tf=10mm, tw=8mm and a=100mm).
66
3.4. Conclusions
In this chapter, a number of equations have been presented for determining the
maximum transverse deflection of simply support and pinned–fixed support castellated
beams subjected to a uniformly distributed load with/without non-uniform temperature.
These equations are derived based on the principle of minimum potential energy. In
order to evaluate the analytical results, ANSYS software is used. From these equations
and the evaluation, the main conclusions that can be summarized as follows:
 The shear-induced deflection caused due to web openings is proportional to the

cross-section area of the two T-sections but inversely proportional to the beam
length.
 The present analytical results are in excellent agreement with those obtained from
the finite element analysis, which demonstrates the appropriateness of proposed
approach.
 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.
 The relative errors between analytical calculation and numerical calculation do not
exceed (24%) even for short span castellated beam with narrow or wide section for
pinned-fixed.
 The effect of web shear on the deflection reduces when castellated beam length
increases.
 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.
67
 For the same average temperatures, the maximum deflection of the castellated beam
under non-uniform temperature distribution with transverse distributed load is
directly proportional to the amount of difference in temperatures between the two
T- sections of the beam.
68
CHAPTER FOUR LATERAL-TORSIONAL BUCKLING
CHAPTER FOUR
4. LATERAL-TORSIONAL BUCKLING
4.1. Introduction
In some cases, castellated beams may undergo a lateral-torsional buckling before they
reach to their ultimate limit state. Recent evidence explains that due to applying
transverse loads on the major axis of the castellated beam, the cross section of the beam
is affected by compression and tension stress. Compression stress on the top flange
causes the vertical movement of the beam that is in the same direction as the load,
which causes lateral moment and thus creating the horizontal movement when the
vertical stiffness is bigger than the lateral stiffness. Meanwhile, tension stress that is
created on the bottom flange tries to reduce the vertical deflection, which leads to
twisting around the major axis when the load exceeds the limit state, as well as causing
a torsional moment, in addition to the existing bending moment. The combinations of
these effects are prone to produce an instability state called lateral-torsional buckling.
Kerdal and Nethercot (1984) indicated that the behaviour of castellated beams is similar
to the plain beam but the properties of the cross section should be considered to evaluate
the lateral-torsional buckling (see Figure 4-1).
This chapter will firstly present the factors affecting the lateral-torsional buckling, and
review the current design philosophy that has been adopted for determining the lateral-
torsional buckling resistance of I-beams with web openings under bending loads.
Secondly, the analytical methods will be presented for elastic critical lateral-torsional
buckling loads of castellated beam under a uniformly distributed transverse load on the
top flange for two common boundary conditions. The analytical approach will be
developed based on the principle of minimum potential energy. Moreover, the linear
69
FEA numerical computations will be provided for verifying the accuracy of the
Figure 4-1 The mechanism of Lateral-Torsional Buckling
4.2. Factors influencing lateral-torsional buckling
Numerous experimental, theoretical and numerical investigations have pointed out that
some factors have an impact on castellated beam’s vulnerability to lateral-torsion
buckling (http://www.scirp.org/journal/ojce). The factors include: the distance between
the lateral supports to the compression of the flanges; boundary conditions; loading type
and position, section type; material properties; magnitude and distribution of the
residual stresses and geometric imperfections (Martins et al., 2017).
4.2.1. Effective length (the distance of castellated beam between two laterals
supports)
Effective length indicates the distance of castellated beam between two lateral supports;
(Korrani et al., 2010) noted that the potential of lateral-torsional buckling occurring is
reduced when the effective length is small which leads to prevent the buckling where
this occurred because of the instability of compression flange. Therefore, this distance
should not exceed the limitations.
70
4.2.2. Boundary conditions
Actually, the lateral-torsional buckling failure is reduced if the buckling moment is

increased which leads to reducing the lateral buckling of the beam, owing to the fact
that the lateral buckling consists of three deformation types, namely lateral bending,
warping, and twisting, where these deformations are affected by the beam boundary
conditions. Hence, the kinds of boundary conditions should be considered to increase
the resistance of beam and prevent all these deformations. In BS5950-1:2000 the
effective length of beam takes into account the effect of the boundary conditions on
lateral-torsional buckling (Sehwail, 2013; Ahnlén. and Westlund, 2013 and New steel
construction technical report, 2006).
4.2.3. Loading location
The lateral-torsional buckling is affected by the vertical line extended between the
location of applying load and the shear canter of the cross-section of the beam.
New steel construction technical report (2006) pointed out that the applying load above
the shear centre makes the beam vulnerable to destabilizing load situation, which
indicated that the beam is more sensitive to the lateral-torsional buckling failure. On the
other hand, the applying load at the shear centre or down leads to decrease the
occurrence of this failure which is called non-destabilizing load. The previous research
found that applying loading below or above the shear centre can change the load of
buckling by ± 40% (Sehwail, 2013).
4.2.4. Loading type
The influence of loading type on the occurrence of a lateral-torsional buckling failure

for steel beams is addressed in BS5950-1:2000 by using the equivalent uniform moment
factor. The previous researchers pointed out that, where the cross-section is subjected to
a uniformly bending moment that is distributed on the length of beam, it resists the
buckling less than the same cross section subjected to different bending moment
distribution. For this reason, the designers use the equivalent uniform moment factor to
modify the design procedure.
71
4.2.5. Imperfections
Initial imperfections are the lateral displacements and twists along the beam length.
The previous investigations stated that these imperfections appear in structural steel
sections due to the fabrication process. In addition, the process of loading could be
eccentric which urges continuous increase in the lateral deflection and twist at the first
stage of loading because the elastic critical moment will be less than yield moment.
Unfortunately, another reason for these imperfections occurring, is that the section’s
properties of beam might change because of distortions and cracks during beam
transportation. However, residual stresses may increase because of imperfections
(Sehwail, 2013 and Jovi, 2015).
4.2.6. Residual stress
Some previous studies stated that the process of manufacture of steel sections and
fabricating castellated beams such as cutting, shifting, and welding could cause a
significant lack of beam strength. This phenomenon is called residual stress. It is clear
that these stresses can affect physical and mechanical properties of the beams. In other
words, the large changes in the temperature during the fabrication process, or erection
could affect the residual stresses, and then the section will start to undergo stress at
lower moments. Therefore, the yielding will extend through the cross section of the
beam concurrently with increased moment. According to the experiments of
Sonck et al. (2014), the castellated beams have compressive residual stresses at the
flanges more than the original parent does. As a result, ignoring these stresses possibly
leads to obtaining unsafe lateral-torsional buckling resistance values (Sehwail, 2013
and Sonck, 2014)
4.2.7. The slenderness
New steel construction technical report (2006) indicates that the slenderness of section
is one of the important conditions that should be considered in the steel structural
design check for lateral-torsional buckling to provide safe structures. Some of the
factors could affect the slenderness of section such as the beam length and the stiffness
72
of both flange lateral bending and the section torsional. However, the castellated beam
is more exposed to the lateral-torsional buckling failure than the I-beam of the same
size due to increasing the depth and the web openings.
4.3. Current design philosophy of lateral-torsional buckling resistance of I-beam

with web openings
Currently, two basic design philosophies have been adopted for determining the lateral-
torsional buckling resistance of I-beam with web openings under bending loads.
The first design philosophy indicates that for I-beam with web openings, the design
check of lateral-torsional buckling is decreased to be a lateral flexural buckling check of
the compressed T-section at web opening. ArcelorMittal (2008) reports this design
philosophy. According to Nseir et al. (2012), this philosophy is conservative because
the tension effect of the flange and the stiffness of torsional of the full cross-section are
completely ignored.
Nethercot and Kerdal (1982) elicit the other design philosophy. They performed
experiments on eight castellated beams and noticed that the lateral-torsional buckling
resistance is not affected by the web openings of the beam. Hence the design philosophy
of lateral-torsional buckling for I-beam without web openings could be used to the
I-beam with web openings, taking into consideration that the properties of the cross-
sectional should be calculated at the centre of the castellation.
Now today, the design specifications such as BSEN1993-1-1:2005; BS5950-1:2000;

Australian standards AS4100 and American standard AISC provide methods, which are
derived, based on the above philosophies, to determine the lateral-torsional buckling
resistance for I-beams with web openings. Many articles found in the literature have
used this design specification.
73
4.4. The calculation methods of lateral-torsional buckling resistance of castellated

beam
Mohebkhah (2004) developed a nonlinear finite element method for simulating inelastic
castellated beams with various loading cases to examine lateral–torsional buckling. The
work also discussed the influence of moment gradient on the lateral-torsional buckling
of castellated beams. The FEA results of inelastic castellated beams with different
slenderness were compared with the results obtained according to the design
specifications AISC-LRFD. Mohebkhah reported that the design specifications AISC-
LRFD is unsafe because the values of moment gradient factors for inelastic beams
provided in AISC are bigger than those determined by nonlinear FEA method.
Zirakian and Showkati (2006) carried out an experimental investigation to examine

lateral-distortional buckling mode and discussed the interaction between local buckling
and lateral-torsional buckling. In their work, six tests were performed on simply
supported castellated beams exposed to a concentrated load. They reported that the
interaction of different buckling modes would lead to a distortion of the cross-section of
the castellated beam due to lack of strength as assumed during lateral-torsional
buckling. The experimental results were compared with the analytical results of the
elastic and inelastic lateral buckling loads, which were obtained by applying the South
well, modified, and Massey extrapolation techniques to gain more accurate predictions
of the critical buckling loads.
Showkati (2008) suggested several empirical formulas to calculate the bending

coefficient of unbraced castellated beams. The comparison was made between his
results and published data by previous studies. The results show that the elastic-bending
capacities of castellated beams, which are subjected to the uniform distributed loads on
the top flange, are affected by the section properties.
Kohnehpooshi and Showkati (2009) carried out the numerical investigations using finite
element method for the evaluation of the effective flexural and torsional stiffness’s,
shear and tension effects of castellated beams on the overall failure of the beams when
subjected to pure bending. The finite element method was carried out using ANSYS
software by using 3-D nonlinear Shell Elements (SHELL181).
74
Sweedan (2011) utilized ANSYS software for simulating the lateral-torsional buckling
of simply supported circular web openings beam. This study applied different cases of
loading on simply supported circular web openings beam which associated with wide
variety of parameters such as cross-sectional dimensions, lengths of beams and
arrangement of web openings to find critical moment values and the moment-gradient
factor. According to numerical results, this study reported that the moment-gradient
factor is affected by the beam geometry and slenderness. In addition, a simplified
approach was proposed to enable accurate determination of a moment modification
factor KLB for the cellular beams.
Ellobody (2011) utilized analytical and experimental methods to investigate the

interaction of buckling modes in castellated beams. Nonlinear finite element method
was utilized to simulate 96 models of the castellated beam by using ABAQUS software.
The effects of various characteristics such as cross section dimensions and length of the
beam on the failure mechanisms of the castellated beams were examined. It was
reported that web distortional buckling occurs on castellated beam because of high
strength, but lateral-torsional buckling failure due to the normal strength of castellated
beam.
Nseir et al. (2012) used both the experimental and numerical methods to examine the
lateral-torsional buckling resistance of circular web opening beams. In addition, they
suggested an analytical design method. Three tests were conducted to make a
comparison between the experimental and numerical results. Their study used a wide
variety of parametric factors, including cross-sectional shape, bending moment
distribution, the relative size of the openings, and yield stress.
The design methods provided in the design specifications BS5950-1, 3.1, 4:1985 and
1988 are adopted by Pachpor et al. (2014) to examine the behaviour of circular web
opening beams to predict the lateral-torsional buckling resistance.
Panedpojaman (2015) made efforts to calculate the lateral-torsional buckling resistance

of I-beams with web openings under a constant bending moment. Two methods, namely
General Method and Specific Method, were used. In addition, the section properties for
75
the calculation were adopted according to Nethercot and Kerdal (1982)’s design
principle.
Kim et al. (2016) presented an analytical study with focussing on the web shear effects
on the lateral–torsional buckling of simply supported castellated beams liable to pure
bending and/or a uniformly distributed load. They performed this study by using the
classical principle of minimum potential energy. They also reported that to increase the
accuracy of the critical moments' value and loads, the average torsional constant of the
full and reduced sections should be considered in calculations, instead of simply taking
the average of the critical moments or loads.
Sonck and Belis (2016) presented a nonlinear numerical study to examine the behaviour
of the lateral-torsional buckling of doubly symmetric castellated beams subjected to a
constant bending moment. For calculations, the study took into account the modified
residual stresses and the cross-sectional properties at the centre of the web opening. The
calculations for lateral-torsional buckling were based on the design specification
BSEN1993-1-1: 2005. The results of the numerical study have been compared with
experimental results to assess the effects of geometric imperfections, elastic-plastic
material behaviours, and residual stresses.
Kwani and Wijaya (2017) presented a paper to investigate the lateral-torsional buckling
of castellated beams. AISC specifications have no equation to determine the critical
moment for lateral-torsional buckling for design purposes of castellated beams.
Therefore, they adopted the collapse analysis by using finite element method to modify
the correction factor of AISC formula for determining the critical moment of the
castellated beam.
4.5. Analytical philosophy of lateral-torsional buckling of castellated Beam
The condition of changing the beam from straight stability state to lateral deflection and
twist state occurs at the critical loads. A calculation method of elastic critical loads of
castellated beams when the beam has a lateral–torsional buckling is presented in this
chapter. The method is derived based on the principle of the total potential energy.
76
According to the model illustrated in, Figure 4-2 (a) the beam shear centre will have
lateral and transverse displacements, respectively v(x), w(x). Furthermore, the cross-
section has an angle of twist ϕ(x). In the linear situation, the strain energy stored in the
beam involves two parts; the energy caused by the deflection and the energy caused by
the twist, which can be written as follows:
2 2 2
1 l 𝑑2𝑤 𝑑2𝑣 𝑑2𝜙 𝑑𝜙 2
𝑈𝑠 = ∫ [𝐸𝐼𝑦 ( 2 ) + 𝐸𝐼𝑧 ( 2 ) + 𝐸𝐼𝑤 ( 2 ) + 𝐺𝐽 ( ) ] 𝑑𝑥 4-1
2 0 𝑑𝑥 𝑑𝑥 𝑑𝑥 𝑑𝑥
where 𝑈𝑠 is the strain energy, 𝑙 is the beam length, 𝐸 is the Young's modulus, 𝐺 is the
shear modulus and 𝐽 is the torsional constant; 𝐼𝑦 and 𝐼𝑧 are the second moments of the
cross-sectional area about the y and z axes respectively, 𝐼𝑤 is the warping constant.
Because of web openings 𝐼𝑦 , 𝐼𝑧 , 𝐼𝑤 and 𝐽 are introduced as a function of x.
In order to consider the web shear influence to determine the elastic critical lateral-
torsional buckling loads in castellated beams, it is assumed that the cross-section of the
castellated beam is decomposed into three parts, two of which represent the top
T-section and bottom T-section, one of which represents the middle-part of the web.
The analysis model for this study is illustrated in Figure 4-2 (a). The second
assumption is that the displacements at the shear centres of the top and bottom tee-
sections are small. The third assumption is that the warping constants of the top and
bottom T-sections and the mid-part of the web are so small and therefore can be ignored.
The displacements of the three parts in the castellated beam can thus be expressed as
follows (see Figure 4-2 (b)) (Kim. et al., 2016):
ℎ ℎ𝜙
𝑣1 = 𝑣 + 𝑠𝑖𝑛𝜙 ≈ 𝑣 + 4-2
2 2
ℎ ℎ𝜙
𝑣2 = 𝑣 − 𝑠𝑖𝑛𝜙 ≈ 𝑣 − 4-3
2 2
ℎ 4-4
𝑤1 = 𝑤 + (1 − 𝑐𝑜𝑠𝜙) ≈ 𝑤
2
ℎ
𝑤2 = 𝑤 + (1 − 𝑐𝑜𝑠𝜙) ≈ 𝑤 4-5
2
77
where 𝑣1 and 𝑣2 are the lateral displacements of the shear centre of the top and bottom
T-section, 𝑤1 and 𝑤2 are the transverse displacements of the shear centre of the top and
bottom T-section, (ℎ) is the distance between the shear centres of top and bottom
T-sections. Hence, the strain energy of the castellated beam based on the three parts can
be written as follows:
2 2
1 𝑙 𝑑2 𝑤1 𝑑2 𝑣1 𝑑𝜙 2
𝑈𝑠 = ∫ [𝐸𝐼𝑦1 ( 2 ) + 𝐸𝐼𝑧1 ( 2 ) + 𝐺𝐽1 ( ) ] 𝑑𝑥
2 0 𝑑𝑥 𝑑𝑥 𝑑𝑥
2 2
1 𝑙 𝑑2 𝑤2 𝑑2 𝑣2 𝑑𝜙 2
+ ∫ [𝐸𝐼𝑦2 ( 2 ) + 𝐸𝐼𝑧2 ( 2 ) + 𝐺𝐽2 ( ) ] 𝑑𝑥 4-6
2 2
1 𝑙 𝑑2 𝑤3 𝑑2 𝑣3 𝑑𝜙 2
+ ∫ [𝐸𝐼𝑦3 ( 2 ) + 𝐸𝐼𝑧3 ( 2 ) + 𝐺𝐽3 ( ) ] 𝑑𝑥
where 𝐼𝑦1 = 𝐼𝑦2 and 𝐼𝑧1 = 𝐼𝑧2 are the second moments of the T- sectional area about
the y and z axes. 𝐽1 = 𝐽2 is the torsional constant of the tee-section, 𝐼𝑦3 and 𝐼𝑧3 are the
second moments of the cross-sectional area of the mid-part of the web about the y and z
axes respectively, and 𝐽3 is the torsional constant of the mid-part of the web.
Hence, the formula of the strain energy of castellated beam (top T- section, bottom
T- section and mid-part of the web), which is susceptible to deflection and twist due to
uniformly distributed load at the top T- sections, can be obtained by substituting
Eqs. (4-2)– (4-5) into Eq. (4-6):
2 2
1 𝑙 𝑑2 𝑤1 𝑑 2 𝑣1
𝑈𝑠 = ∫ [2𝐸𝐼𝑦1 ( 2 ) + 2𝐸𝐼𝑧1 ( 2 )
2 0 𝑑𝑥 𝑑𝑥
2
ℎ2 𝑑2𝜙 𝑑𝜙 2
+ 𝐸𝐼𝑧1 ( 2 ) + 2𝐺𝐽1 ( ) ] 𝑑𝑥 4-7
2 𝑑𝑥 𝑑𝑥
2 2
1 𝑙 𝑑 2 𝑤2 𝑑 2 𝑣2 𝑑𝜙 2
+ ∫ [𝐸𝐼𝑦3 ( 2 ) + 𝐸𝐼𝑧3 ( 2 ) + 𝐺𝐽3 ( ) ] 𝑑𝑥
According to Figure 4-2 (c), 𝐼𝑦1 , 𝐼𝑧1 and 𝐽1 are constants,while 𝐼𝑦3 , 𝐼𝑧3 and 𝐽3 are
depending upon the location of the web openings, therefore they are function of x.
Hence, from the comparison between Eqs. (4-7) and (4-1), it can be obtained, that:
𝐼𝑦 = 2𝐼𝑦1 + 𝐼𝑦3 4-8
78
𝐼𝑧 = 2𝐼𝑧1 + 𝐼𝑧3 4-9
ℎ 2 ℎ2
𝐼𝑤 = ( ) 𝐼𝑧 ≈ 𝐼𝑧1 4-10
2 2
𝐽 = 2𝐽1 + 𝐽3 4-11
Note from Eq. (4-10) the warping strain energy cannot be ignored because the
displacement compatibility occurs when the two T-sections assemble.
The potential energy, which is the negative value resulting from the applied loads when
the lateral-torsional buckling occurs, can be written as follows:
𝑙
𝑑2𝑤 𝑑2𝑣 𝑎𝑧 𝑞𝑧 2
𝑊 = − ∫ [𝑀𝑦 ( 2
) + 𝑀𝑦 𝜙 ( 2
)+ 𝜙 ] 𝑑𝑥 4-12
0 𝑑𝑥 𝑑𝑥 2
where az refers to the z-coordinate of the loading point, which is the vertical distance
hw
between the loading point and the shear centre of the beam, in this case, az = + 𝑡𝑓
2
because the uniformly distributed load is applied on the top flange of the beam.
In summary, by using Eqs. (4-7) and (4-12), the equation of the total potential energy of
the castellated beam considering lateral-torsional buckling deflection can be expressed
as follows:
2 2
1 𝑙 𝑑2 𝑤1 𝑑2 𝑣1
∏ = ∫ [2𝐸𝐼𝑦1 ( 2 ) + 2𝐸𝐼𝑧1 ( 2 )
2
ℎ2 𝑑2 𝜙 𝑑𝜙 2
+ 𝐸𝐼𝑧1 ( 2 ) + 2𝐺𝐽1 ( ) ] 𝑑𝑥
2 𝑑𝑥 𝑑𝑥
2 2
1 𝑙 𝑑2 𝑤2 𝑑2 𝑣2 𝑑𝜙 2
+ ∫ [𝐸𝐼𝑦3 ( 2 ) + 𝐸𝐼𝑧3 ( 2 ) + 𝐺𝐽3 ( ) ] 𝑑𝑥 4-13
𝑙
𝑑2 𝑤 𝑑2 𝑣 𝑎𝑧 𝑞𝑧 2
− ∫ [𝑀𝑦 ( 2 ) + 𝑀𝑦 𝜙 ( 2 ) + 𝜙 ] 𝑑𝑥
0 𝑑𝑥 𝑑𝑥 2
79
Figure 4-2 (a) Notations used in castellated beams. (b) Loading and displacements of web and
displacement of flanges when lateral–torsional buckling occurred (c) Section properties of
∗ ∗
middle-part of web in four different regions. 𝐼𝑦3 = 𝐼𝑦3 , 𝐼𝑧3 = 𝐼𝑧3 , 𝐽3 = 𝐽3∗ in region 2, in region
4, 𝐼𝑦3 = 𝐼𝑧3 = 𝐽3 = 0 , section properties vary with x in regions 1 and 3.
4.5.1. Determining the elastic critical buckling loads of simply supported, doubly
symmetric castellated beam subjected to uniformly distributed loads
The critical moment Mcr of a simply supported castellated beam under a uniformly
distributed load applied on the top flange of the beam can be obtained from Eq. (4-14)
as follows (Kim et al., 2016),
2 2
ℎ𝑤 ℎ 𝜋2 1 𝐺(2𝐽1 + 𝑘𝐽3∗ )𝑙 2 1
−( + 𝑡𝑓 ) + √( 𝑤 + 𝑡𝑓 ) + ( + ) [𝐼𝑤 + ] ∗ )
2 2 6 2 𝜋2𝐸 (2𝐼𝑧1 + 𝑘𝐼𝑧3
Mcr =
1 1 2 4-14
( + 2)
3 𝜋
∗ )
𝐸(2𝐼𝑧1 + 𝑘𝐼𝑧3
× 2
𝑙
where
80
𝑞𝑧 𝑙 2
Mcr =( ) 4-15
8 𝑐𝑟
∗ ∗
𝐼𝑧3 is negligible because in most of castellated beams 𝐼𝑧3 << 2𝐼𝑧1 . Thus, Eq. (4-14)
can be simplified as (Kim et al., 2016)
2 2
ℎ𝑤 ℎ 𝜋2 1 𝐼 𝐺(2𝐽1 + 𝑘𝐽3∗ )𝑙 2
−( + 𝑡𝑓 ) + √( 𝑤 + 𝑡𝑓 ) + ( + ) [ 𝑤 + ] 4-16
𝑞𝑧 𝑙 2 2 2 6 2 2𝐼𝑧1 2𝐼𝑧1 𝜋 2 𝐸 2𝐼𝑧1 𝐸
( ) = ×
8 𝑐𝑟 1 1 2 𝑙2
( + 2)
3 𝜋
ℎ𝑤 ℎ 2
𝜋2 1
2
𝐼 𝐺(2𝐽1 + 𝑘𝐽∗3 )𝑙2
−( + 𝑡𝑓 ) + √( 𝑤 + 𝑡𝑓 ) + ( + ) [ 𝑤 + ]
2 2 6 2 2𝐼𝑧1 2𝐼𝑧1 𝜋2 𝐸
(𝑞𝑧 )𝑐𝑟 =
1 1 2 4-17
(3 + )
𝜋2
16𝐼𝑧1 𝐸
×
𝑙4
4.5.2. Determining the elastic critical buckling loads of pinned-fixed castellated

beam, doubly symmetric castellated beam subjected to uniformly
distributed loads
The buckling displacements of a pinned-fixed castellated beam can be assumed as

follows:
𝑘𝜋(𝑙 − 𝑥) 𝜋 ( 𝑙 − 𝑥)
𝑤(𝑥) = 𝐴 sin ( ) sin ( ) 4-18
𝑙 2𝑙
𝑘𝜋(𝑙 − 𝑥) 𝜋(𝑙 − 𝑥)
𝑣(𝑥) = 𝐵 sin ( ) sin ( ) 4-19
𝑙 2𝑙
𝑘𝜋(𝑙 − 𝑥) 𝜋 ( 𝑙 − 𝑥)
𝜙(𝑥) = 𝐶 sin ( ) sin ( ) 4-20
𝑙 2𝑙
where A, B, and C are the constants to be determined. It is obvious that the above
𝑑2 𝑤
displacements functions satisfy the boundary conditions, that are 𝑤 = 𝑣 = 𝜙 = =
𝑑𝑥 2
𝑑2 𝑣 𝑑2 𝜙 𝑑𝑤 𝑑𝑣 𝑑𝜙
= = 0 at x = o and 𝑤 = 𝑣 = 𝜙 = = 𝑑𝑥 = = 0 at x =l .
𝑑𝑥 2 𝑑𝑥 2 𝑑𝑥 𝑑𝑥
81
Substituting Eqs. (4-18), (4-19) and (4-20) into Eq. (4-7) yields
41 𝜋 4 ∗ ∗ )𝐵 2
𝑈𝑠 = [𝐸(2𝐼𝑦1 + 𝑘𝐼𝑦3 )𝐴2 + 𝐸(2𝐼𝑧1 + 𝑘𝐼𝑧3 + 𝐸𝐼𝑤 𝐶 2
128 𝑙 3
4-21
16 1 ∗)
𝑙 2 2
+ (1 + 2 ) 𝐺(2𝐽1 + 𝑘𝐽3 ( ) 𝐶 ]
41 𝜋 𝜋
According to Kim.et al. (2016) k refers to the fraction of the volume of the solid and
holes in the mid-part of the web beam. For most castellated beams, because of matching
of the solid areas and holes in the mid-part of the web have equal area this leads to the
value of k=0.5.
The internal bending moment for a pinned-fixed castellated beam subject to a uniformly
distributed load can be written as follows:
3 1
𝑀𝑦 (𝑥) = 𝑞𝑧 𝑙𝑥 − 𝑞𝑧 𝑥 2 4-22
8 2
Substituting Eq. (4-22) into Eq. (4-12) yields an expression for the potential energy of
the external loads as:
4𝑙𝑞𝑧 𝑞𝑧 𝜋 2 𝑙 1 1 𝑎𝑧 𝑙𝑞𝑧 2
𝑊= 𝐴+ ( + 2 ) 𝐵𝐶 − 𝐶 4-23
3𝜋 20 3 𝜋 8
Combining Eqs. (4-21) and (4-23) yields an expression for the total potential energy:
41 𝜋 4 ∗ ∗ )𝐵 2
4-24
∏= 3
[𝐸(2𝐼𝑦1 + 𝑘𝐼𝑦3 )𝐴2 + 𝐸(2𝐼𝑧1 + 𝑘𝐼𝑧3 + 𝐸𝐼𝑤 𝐶 2
128 𝑙
16 1 ∗)
𝑙 2 2 4𝑙𝑞𝑧
+ (1 + 2 ) 𝐺(2𝐽1 + 𝑘𝐽3 ( ) 𝐶 ] + 𝐴
41 𝜋 𝜋 3𝜋
𝑞𝑧 𝜋 2 𝑙 1 1 𝑎𝑧 𝑙𝑞𝑧 2
+ ( + 2 ) 𝐵𝐶 − 𝐶
20 3 𝜋 8
The variation of Eq. (4-24) with respect to A, B and C results in the following three
algebraic equations:
41𝑙 𝜋 4 4𝑙𝑞𝑧
∗
( ) [2𝐸(2𝐼𝑦1 + 𝑘𝐼𝑦3 )𝐴] + =0 4-25
128 𝑙 3𝜋
41𝑙 𝜋 4 𝑞𝑧 𝜋 2 𝑙 1 1
∗ )𝐵]
( ) [2𝐸(2𝐼𝑧1 + 𝑘𝐼𝑧3 =− ( + )𝐶 4-26
128 𝑙 20 3 𝜋 2
82
41𝑙 𝜋 4 16 1 ∗)
𝑙 2
( ) [2𝐸𝐼𝑤 𝐶 + (1 + 2 ) 2𝐺(2𝐽1 + 𝑘𝐽3 ( ) 𝐶]
128 𝑙 41 𝜋 𝜋 4-27
2
𝑎𝑧 𝑙𝑞𝑧 𝑞𝑧 𝜋 𝑙 1 1
= 𝐶− ( + )𝐵
4 20 3 𝜋 2
The second-order variation of the total potential energy equation, with respect to A, B
and C should be equal to zero, from which the critical moment/load is obtained
𝛿 2 ∏ = 𝛿 2 (𝑈𝑠 + 𝑊) = 0 4-28
Eq. (4-28) leads to:
𝐴=0 4-29
𝜋 2 8𝑞𝑧 𝑙 2 1 1
𝐸(2𝐼𝑧1 + ∗ )
𝑘𝐼𝑧3 ( ) 𝐵= ( + )𝐶 4-30
𝑙 20 3 𝜋 2
𝜋 2 41 2 8𝑞𝑧 𝑙 2 1 1 41𝑎𝑧 𝑙2 𝑞𝑧
[( ) ( ) 𝐸𝐼𝑤 + 𝐺(2𝐽1 + 𝑘𝐽3∗ )] 𝐶 = ( + 2) 𝐵 + 𝐶 4-31
𝑙 8 20 3 𝜋 4𝜋 2
Eliminating B in Eqs. (4-30) and (4-31), it yields
2
8𝑞𝑧 𝑙 2 1 1 41𝑎𝑧 𝑙 2 𝑞𝑧 ∗ )
𝜋 2
[ ( + 2 )] + (𝐸(2𝐼𝑧1 + 𝑘𝐼𝑧3 ( ) )
20 3 𝜋 4𝜋 2 𝑙
4-32
𝜋 2 41 2 𝜋 2
= [( ) ( ) 𝐸𝐼𝑤 + 𝐺(2𝐽1 + 𝑘𝐽3∗ )] (𝐸(2𝐼𝑧1 + 𝑘𝐼𝑧3
∗ )
( ) )
𝑙 8 𝑙
Solving for 𝑞𝑧 from Eq. (4-32), it yields
2
ℎ𝑤 ℎ 64𝜋 2 1 1 2 𝐼𝑤 𝐺(2𝐽1 + 𝑘𝐽3∗ )𝑙 2 1
−( + 𝑡𝑓 ) + √( 𝑤 + 𝑡𝑓 ) + ( + ) [ + ] ∗ )
𝑞𝑧 𝑙 2 2 2 20 𝜋 2 3 64 412 𝜋 2 𝐸 (2𝐼𝑧1 + 𝑘𝐼𝑧3
( ) =
8 𝑐𝑟 64𝜋 2 1 1 2 4-33
( + )
20 𝜋 2 3
∗ )
41𝐸(2𝐼𝑧1 + 𝑘𝐼𝑧3
× 2
𝑙
∗
Again, if 𝐼𝑧3 is neglected, Eq. (4-33) can be simplified as follows: (Kim et al., 2016)
83
2
ℎ𝑤 ℎ 64𝜋 2 1 1 2 𝐼𝑤 𝐺(2𝐽1 + 𝑘𝐽3∗ )𝑙 2 1
−( + 𝑡𝑓 ) + √( 𝑤 + 𝑡𝑓 ) + ( + ) [ + ]
𝑞𝑧 𝑙 2 2 2 20 𝜋 2 3 64 412 𝜋 2 𝐸 (2𝐼𝑧1 )
( ) =
8 𝑐𝑟 64𝜋 2 1 1 2
( + )
20 𝜋 2 3 4-34
41𝐸(2𝐼𝑧1 )
×
𝑙2
2
ℎ𝑤 ℎ 64𝜋 2 1 1 2 𝐼𝑤 𝐺(2𝐽1 + 𝑘𝐽3∗ )𝑙 2 1
−( + 𝑡𝑓 ) + √( 𝑤 + 𝑡𝑓 ) + ( + ) [ + ]
2 2 20 𝜋 2 3 64 412 𝜋 2 𝐸 (2𝐼𝑧1 )
(𝑞𝑧 )𝑐𝑟 =
64𝜋 2 1 1 2 4-35
( + )
20 𝜋 2 3
328𝐸(2𝐼𝑧1 )
×
𝑙4
4.6. Numerical analysis lateral-torsional buckling of castellated beams due to

uniformly distributed load
The main objective herein is to validate the equations of analytical solutions developed
in sec 4.6 for calculating critical moment of lateral-torsional buckling of castellated
beams subjected to uniformly distributed load on top flange. A linear 3D finite element
analysis is carried out by employing ANSYS mechanical (APDL) software for
conducting Eigenvalue analysis to determine the critical moment
4.6.1. Modelling consideration, material model and loading
The FEA modelling of castellated beams is done by using 3D linear 4-Node Thin Shell
Elements (SHELL181) depicted in Figure 3-5, which is suitable for linear and large
rotation. This element presents four nodes with six DOF per node, i.e., translations and
rotations on the X, Y, and Z-axis, respectively. The beams are meshed using element
sizes not exceeding 10 mm. A typical mesh configuration is shown in Figure 3-6. The
mechanical properties used are the elastic material with Young’s modulus E = 2.1 × 105
MPa, Poisson’s ratio v =0.3, and yield stress σy=275 MPa.
𝜎𝑦 𝐼
The external load is the uniformly distributed load (𝑞𝑚𝑎𝑥 = 16 𝑙2 (ℎ𝑟𝑒𝑑𝑢𝑐𝑒𝑑 ) acting on the
+2𝑡 ) 𝑤 𝑓
junction of the upper flange and web. (see Figure 3-7)
84
4.6.2. Boundary conditions
4.6.2.1. Simply supported castellated beam
The displacement boundary conditions are applied to all nodes at the two ends of the
beam, as shown in Figure 3-7. The lateral and transverse deflections and rotation are
restrained (uy=uz=0, and θx=0) at two ends of the simply supported beam. Moreover,
another boundary condition is applied by restricting the axial displacement (ux=0) at one
node of one end of the simply supported beam.
4.6.2.2. Pinned-fixed castellated beam
The displacement boundary conditions are also applied to all nodes at the two ends of
the beam, as shown in Figure 3-7. The lateral and transverse deflections and rotation
are restrained (uy=uz=0 and θx=0) at the simply supported end, while the fixed support
boundary condition is applied at the other end by restricting the axial displacement,
transverse deflections and rotations around the three axes within the cross-section
(ux=uy=uz=0 and θx=θy= θz=0).
4.6.3. Comparison of results and discussion
4.6.3.1. Linear lateral-torsional buckling stress results of simply supported due to

uniformly distributed load act on top flange
Table (4-1) shows a comparison of the critical loads of lateral-torsional buckling of

simply support castellated beam subjected to the uniformly distributed transverse load
applying on its top flange between analytical solution and numerical analysis. The
analytical solution was obtained directly from Eqs. (4-16) and (4-17), whereas the
numerical result was obtained by using linear buckling analysis built in ANSYS
2𝜎𝑦 𝐼𝑟𝑒𝑑𝑢𝑐𝑒𝑑
software (APDL). The yield moment was calculated by using Myield = with
ℎ𝑤 +2𝑡𝑓
𝑁
𝜎𝑦 = 275 and 𝐼𝑟𝑒𝑑𝑢𝑐𝑒𝑑 was calculated using Eq. (3-17). The results of different
𝑚𝑚2
castellated beam lengths with various flange widths (see Section 3.3.4) shown in
Table (4-1) are also plotted in Figures 4-3, 4-4 and 4-5, respectively
85
It can be seen from the figures that, in each group of flange width, the curves of the
analytical solution and numerical analysis have a similar variation pattern. The
analytical solution is in excellent agreement with the numerical analysis. This indicates
that the analytical model developed above for the castellated beam is appropriate for
calculating the critical load of lateral-torsional buckling. Moreover, the latter reflects the
influence of ignoring the value of the second moment of the cross-sectional area of the
∗
mid-part of the web about the z-axis (𝐼𝑧3 ) , and taking account the torsional constant of
the cross-sectional area of the mid-part of the web, this is comparable to that of the two
T- sections(𝑘𝐽3∗ ) (Kim et al., 2016).
86
Table 4-1 Comparison of results linear critical lateral-torsional buckling load (𝑞𝑐𝑟 , Mcr ) of
simply supported castellated beams with a uniformly distributed act on top flange
Linear buckling
Linear buckling
Name Numerical analysis
bf Myield Analytical analysis
(Ansys)
of N.mm
mm Mcr Mcr Mcr Mcr
beam x107 qcr qcr
N.mm N.mm
N/mm
x107 Myield N/mm
x107 Myield
C1 10.44 35.50 5.33 0.51 34.54 5.18 0.50
E1 10.44 12.23 3.59 0.34 12.10 3.56 0.34

G1 10.44 5.68 2.76 0.26 5.40 2.63 0.25
100
H1 10.44 1.90 1.93 0.18 1.88 1.90 0.18
I1 10.44 0.73 1.41 0.14 0.72 1.41 0.13
J1 10.44 0.45 1.20 0.11 0.45 1.18 0.11
C2 14.57 91.28 13.69 0.94 90.56 13.58 0.93
E2 14.57 28.44 8.36 0.57 28.29 8.32 0.57
G2 14.57 12.64 6.14 0.42 12.74 6.19 0.42

150
H2 14.57 3.99 4.05 0.28 3.86 3.92 0.27
I2 14.57 1.51 2.94 0.20 1.58 3.07 0.21
J2 14.57 0.96 2.53 0.17 0.96 2.53 0.17

C3 18.70 192.00 28.80 1.54 185.71 27.86 1.49
E3 18.70 57.11 16.79 0.90 56.04 16.47 0.88
G3 18.70 23.97 11.65 0.62 23.67 11.51 0.62

200
H3 18.70 7.18 7.28 0.39 7.04 7.14 0.38
I3 18.70 2.61 5.07 0.27 2.61 5.07 0.27
J3 18.70 1.64 4.33 0.23 1.61 4.26 0.23

C4 22.83 356.67 53.50 2.34 317.97 47.70 2.09
E4 22.83 101.70 29.90 1.31 100.22 29.46 1.29

G4 22.83 41.15 20.00 0.88 40.79 19.82 0.87
250
H4 22.83 11.70 11.86 0.52 11.59 11.76 0.51
I4 22.83 4.09 79.61 0.35 4.10 7.98 0.35

J4 22.83 2.53 67.06 0.29 2.51 6.64 0.29
87
Figure 4-3 Critical elastic lateral-torsional buckling load (𝑞𝑐𝑟 ) of simply supported castellated
beam subjected to a uniformly distributed load between analytical solutions and FEA numerical
solution for different beam lengths with various flange widths (a) bf =100mm, (b) bf=150mm,
(c) bf=200mm and (d) bf =250mm (hw=300mm, tf=10mm, tw=8mm and a=100mm)
88
Figure 4-4 Critical elastic lateral-torsional buckling moment (Mcr ) of simply supported
castellated beam subjected to a uniformly distributed load between analytical solutions and FEA
numerical solution for different beam lengths with various flange widths (a) bf =100mm,
(b) bf=150mm, (c) bf=200mm and (d) bf =250mm (hw=300mm, tf=10mm, tw=8mm and
a=100mm)
89
Figure 4-5 Critical elastic lateral-torsional buckling moment (Mcr /Myield ) of simply supported
numerical solution for different beam lengths with various flange widths (a) bf =100mm, (b)
bf=150mm, (c) bf=200mm and (d) bf =250mm (hw=300mm, tf=10mm, tw=8mm and a=100mm)
90
4.6.3.2. Linear lateral-torsional buckling stress results of pinned-fixed castellated

beam due to uniformly distributed load act on top flange
Table (4-2) shows a comparison of the critical loads of lateral-torsional buckling of

pinned-fixed castellated beams subjected to a uniformly distributed transverse load
applying on its top flange between analytical solution and numerical analysis. The
analytical solution was obtained directly from Eqs. (4-34) and (4-35), whereas the
numerical result was obtained by using linear buckling analysis built in ANSYS
software (APDL). The yield moment was calculated by using Myield = with
ℎ𝑤 +2𝑡𝑓
𝑁
𝜎𝑦 = 275 𝑚𝑚2 and 𝐼𝑟𝑒𝑑𝑢𝑐𝑒𝑑 was calculated using Eq. (3-17). The results of different
castellated beam lengths with various flange widths (see Section 3.3.4) shown in
Table (4-2) are also plotted in Figures 4-6, 4-7 and 4-8, respectively
It can be seen from the figures that, in each group of flange width, the curves of the
analytical solution and numerical analysis have a similar variation pattern. The
analytical solution is in very good agreement with the numerical analysis for beams
longer than 4.8 m. This indicates that the analytical model developed here for the
castellated beam is appropriate for calculating the critical load of lateral-torsional
buckling. In the case that the beam is shorter, the critical load obtained from the
analytical solution is found to be larger than that obtained from the numerical analysis.
This is because the critical mode of buckling for the shorter beam is not dominated by
the lateral-torsional buckling. This is demonstrated by the critical buckling mode shown
in Figure 4-9 which was obtained from the numerical analysis for the beam of length
(l = 3.5 m) and flange width (bf = 200 mm). It is obvious that the critical buckling mode
of this short beam not only involves the lateral-torsional buckling mode but also
contains the modes of local buckling of compressed flange and shear buckling of web.
The latter disappears with the increase of the beam length or the decreases of beam
flanges (Kerdal and Nethercot, 1984). In the case that the beam is shorter (l < 3.5 m)
and its flanges are wider (bf =250mm), the critical buckling mode is no longer controlled
by the lateral-torsional buckling and thus the difference between the analytical solution
and numerical analysis becomes large (Ellobody, 2011).
91
Table 4-2 Comparison of results linear critical lateral-torsional buckling load (𝑞𝑐𝑟 , Mcr ) of
pinned-fixed castellated beams with a uniformly distributed act on top flange
Linear buckling
Linear buckling
Name Numerical analysis
bf Myield Analytical analysis
(Ansys)
of N.mm
mm Mcr Mcr Mcr Mcr
beam x107 qcr qcr
N.mm N.mm
N/mm
x107 Myield N/mm
x107 Myield
C1 10.44 80.43 12.06 1.16 79.31 11.90 1.14
E1 10.44 25.79 7.58 0.73 26.02 7.65 0.73

G1 10.44 11.55 5.61 0.54 11.78 5.73 0.55
100
H1 10.44 3.80 3.89 0.40 4.12 4.17 0.41
I1 10.44 1.49 2.89 0.28 1.59 3.13 0.30

J1 10.44 0.92 2.43 0.23 0.91 2.40 0.23
C2 14.57 232.88 34.93 2.40 207.30 31.09 2.13
E2 14.57 68.16 20.03 1.38 66.07 19.42 1.33
G2 14.57 28.40 13.80 0.95 28.31 13.76 0.94

150
H2 14.57 8.47 8.60 0.59 8.84 8.96 0.62
I2 14.57 3.10 6.01 0.41 3.15 6.12 0.42
J2 14.57 1.94 5.15 0.35 1.94 5.13 0.35

C3 18.70 523.73 78.55 4.20 372.75 55.91 3.00
E3 18.70 146.39 43.04 2.30 134.80 39.63 2.12

G3 18.70 58.30 28.33 1.52 56.53 27.48 1.47
200
H3 18.70 16.17 16.39 0.88 16.57 16.80 0.90
I3 18.70 5.58 10.84 0.58 5.60 10.88 0.58

J3 18.70 3.44 9.10 0.49 3.43 9.07 0.48
C4 22.83 1000.72 150.11 6.57 391.17 58.67 2.60
E4 22.83 272.99 80.25 3.52 230.52 67.77 2.97

G4 22.83 105.84 51.44 2.25 101.64 49.40 2.16
250
H4 22.83 27.89 28.27 1.24 28.11 28.51 1.25
I4 22.83 9.16 17.80 0.78 9.12 17.73 0.78

J4 22.83 5.53 14.63 0.64 5.75 15.21 0.67
92
Figure 4-6 Critical elastic lateral-torsional buckling load (𝑞𝑐𝑟 ) of pinned-fixed castellated beam
subjected to a uniformly distributed load between analytical solutions and FEA numerical
93
Figure 4-7 Critical elastic lateral-torsional buckling moment (Mcr ) of pinned-fixed castellated
beam subjected to a uniformly distributed load between analytical solutions and FEA numerical
94
Figure 4-8 Critical elastic lateral-torsional buckling moment (Mcr /Myield ) of pinned-fixed
numerical solution for different beam lengths with various flange widths (a) bf =100mm,
(b) bf=150mm, (c) bf=200mm and (d) bf =250mm (hw=300mm, tf=10mm, tw=8mm and
a=100mm)
95
Figure 4-9 Combined modes failure of pinned-fixed castellated beam with a uniformly
distributed load act on top flange for beam (C3 and C4) with two various flange widths
bf=200mm and bf=250mm, obtained from the linear lateral-torsional buckling 3D finite element
analysis using ANSYS software (hw=300mm, tf=10mm, tw=8mm and a=100mm).
4.7. Conclusions
In this chapter, the behaviour of lateral-torsional buckling of castellated beams

subjected to a uniformly distributed load has been investigated using both analytical and
numerical methods. The analytical study provides simplified equations for predicting
the elastic critical load of lateral-torsional buckling. These equations have taken into
account the web shear effect. FEA numerical analysis has been conducted using
ANSYS commercial software to verify the analytical solution. From the comparison it
can be concluded that:
 The torsional constant adopted for determining the critical lateral–torsional

buckling load of castellated beams should be calculated by using the average
torsional constant of the full and reduced section properties.
 The analytical solution and numerical solution have a similar variation pattern with
the beam length. The analytical solution is in excellent agreement with the
numerical analysis for middle and long length beams.
96
 For pinned-fixed castellated beams lateral-torsional buckling occurs only in long

length beams. For beams shorter than 3.5 m with wider flanges, the buckling mode
involves other modes and thus the critical load obtained from the lateral-torsional
buckling analysis cannot represent the critical load of coupled buckling modes.
97
CHAPTER FIVE NONLINEAR ANALYSIS OF CASTELLATEDBEAMS
CHAPTER FIVE
5. NONLINEAR ANALYSIS OF CASTELLATED BEAMS
5.1. Introduction
This chapter is focusing on the effect of both the geometric and material nonlinearities
on the bending and buckling behaviour of castellated beams under uniformly distributed
transverse loads. The analysis is performed using finite element method with the use of
ANSYS (APDL) commercial software.
Moreover, detailed comparisons of results obtained from the linear finite element
analysis shown in Chapter three, the linear buckling analysis shown in chapter four, and
the nonlinear finite element analysis shown in this Chapter are provided.
5.2. Nonlinear finite element analysis of castellated beam
The analysis of using elastic and small deflection theory provides a linear solution, in
which any deflection is proportional to the externally applied load. This kind of analysis
is correct only when the load is very small or the deflection is small. When the beam is
subjected to a large external load, it may not behave linearly, in this case geometric
nonlinearity and/or material nonlinearity may need to be considered. In general, when
the beam has a large deflection, the geometric nonlinearity have to be considered;
whereas when the stress occurred in the beam is near to the yield stress of the beam
material the material nonlinearity has to be considered. In this section, we use the
nonlinear analysis by considering both geometric and material nonlinearities. In the
former large deflection is considered; while in the latter a bi-linear material model is
employed for the constitutive relationship of the steel material.
98
Nonlinear finite element analysis is described in this section on the castellated beams
that were considered in the analytical study in the previous chapters. The aim of this
analysis is to study the large deflection and inelastic behaviour of castellated beams
when subjected to a uniformly distributed load. The results of the nonlinear analysis are
presented by using the load-deflection response curve, from which the failure load of
the beam is also obtained. Moreover, an evaluation will be conducted of the critical
value of load obtained by analytical buckling analysis and the critical load obtained by
linear buckling analysis using finite element method.
ANSYS mechanical (APDL), that has technical potential, is employed to carry out both
geometric and material nonlinear analysis of castellated beams. The geometric
nonlinearity is to take into account the large deflection/displacement of the beams. The
material nonlinearity is to consider the plasticity of steel material in which a bi-linear
isotropic material model is used. The material properties used in the present nonlinear
analysis are the modu1us of elasticity, E=2.1x105 MPa, yield stress, σy= 275 MPa,
Poisson’s ratio, v = 0.3, and the tangent modulus after the yield, Et=0, that is the elastic-
perfectly plastic material model is used (see Figure 5-1).
Figure 5-1 Elastic-perfectly plastic material model
5.2.1. Modelling, and boundary conditions
The modelling of castellated beams is carried out by using 3D 4-Node Thin Shell
Elements (SHELL181) depicted in Figure 3-5. This element presents four nodes with
six DOF per node, i.e., translations and rotations on the X, Y, and Z-axis, respectively.
99
The geometry of castellated beams is meshed by using 4-noded elements with

maximum size not exceeding 10 mm. Two types of boundary conditions are considered
herein; one is the simply support and the other is pinned-fixed, which are the same as
those shown in Section 4.7.2
5.2.2. Apply loads gradually (incremental solution)
Similar to Chapter four, the external load is applied uniformly on the line between web
and upper flange. This is done by applying the load to each node on the line. The nodal
load applied is equal to the loading density multiplied by beam length and divided by
the number of nodes on that line. In the early 1980’s, the arc-length method was
developed to improve Newton-Raphson method to make sure the convergence can be
achieved for various different nonlinear problems (Crisfield, 1981). Note that, the
Newton-Raphson method divides the load into a series of load increments to be applied
over several load steps then evaluates convergence between the internal force and the
applied loads; if the criteria of convergence are not convinced, the program attempts to
resolve with a smaller load increment until the case converges. However, this method is
suitable only for either load controlled, or displacement controlled nonlinear analysis.
For a general nonlinear problem, the arc-length should be used as it combines the load
and displacement increments in each iterative procedure. As a result, in the present
study the equations of nonlinear equilibrium are solved by using the procedure of the
Newton–Raphson method, in conjunction with Arc-Length Method.
5.3. Serviceability limit state
Structural serviceability indicates the limit states that are considered in the design of the
structure. Therefore, to ensure that a building is safe, these conditions should be
followed. The current standard serviceability design has different deflection limits
which depend on the purpose of service as it is intended, and the material of the
structure. The aim of this section is to validate the results of deflection that are
calculated from the analytical linear method, linear, and nonlinear finite element
100
methods. In this study, the value of deflection under the serviceability load that was
considered is l/250 because the structure is steel.
5.4. Investigating load-deflection response curve for castellated beams with

different boundary conditions.
5.4.1. Investigating load-deflection response curve of simply supported beams with

uniformly distributed loads
Figures 5-2, 5-3, 5-4, and 5-5 show a comparison of the curves of load versus
deflection of simply supported castellated beam with a uniformly distributed load. This
comparison involves the results of linear analytical solutions, which are developed in
chapter three, nonlinear 3D finite element analysis using ANSYS software, and the
deflection limit (l/250) to groups C, D, E, G, H, and I with different flange widths (see
Section 3.3.4). The load is presented as the increments of load calculated following
Eq. (3-31).
From these figures it can be noticed that the nonlinear behaviour of castellated beams,
for the same flange width, drops continuously with the increase of beam length, whereas
the behaviour of the beam towards linear behaviour progresses gradually. In contrast,
the nonlinear behaviour of the castellated beams, for the same beam length, increases
continuously with the increase of the flange width, thus the castellated beams will reach
the yield point in the early stages of the loading process.
As expected, the reason for this is that the dimensions of castellated beams and the
value of yield stress were taken account for calculating the value of uniformly
distributed load, which was applied in this work (see Eq. (3-31)), where the yield stress
value is constant for all groups of castellated beams, thus the load capacity of the beams
is affected by changing the beam dimensions. Note that the uniformly distributed load
with the same beam length increases when the flange width increases, which leads to
being more susceptible to yielding in the early stages of the loading process. In contrast,
the uniformly distributed load with the same flange width reduces when the beam length
101
increases, which leads to being less susceptible to yielding in the early stages of the
loading process.
Table (5-1) presents the comparison of the critical value of loads (𝑞𝑐𝑟 /𝑞𝑦𝑖𝑒𝑙𝑑 ) with the
different beam lengths with various flange widths. The results are also plotted in
Figure 5-6. For the purpose of comparison, the results obtained from the linear
analytical solutions and the nonlinear 3D finite element analyses are evidence that the
critical value of load is influenced by geometry, web openings and material beam. As a
result, the designer should consider a nonlinear analysis of short castellated beams with
wide flange width for the design.
102
Figure 5-2 Comparison of the curves of load versus deflection of simply supported castellated
beam with a uniformly distributed load obtained from linear analytical solutions, nonlinear 3D
finite element analysis, and deflection limit (l/250) for beams C1, D1, E1, G1, H1, and I1.
103
104
105
106
Table 5-1 Comparison of results (𝑞𝑐𝑟 /𝑞𝑦𝑖𝑒𝑙𝑑 ) between linear analytical solution and nonlinear
3D finite element analysis of simply supported castellated beams subjected to a uniformly
distributed for groups C, D, E, G, H, and I with different flange widths.
Nonlinear 3D finite
Linear analytical
element analysis by
bf Name 𝑞𝑦𝑖𝑒𝑙𝑑 solution
Ansys
mm of beam N/mm 𝑞𝑐𝑟 𝑞𝑐𝑟
𝑞𝑐𝑟 𝑞𝑐𝑟
N/mm qyield N/mm qyield
C1 69.59 69.59 1.00 59.94 0.86
D1 48.33 48.33 1.00 47.85 0.99

E1 35.51 35.51 1.00 35.51 1.00
100
G1 21.48 21.48 1.00 21.48 1.00
H1 10.29 10.29 1.00 10.29 1.00

I1 5.37 5.37 1.00 5.37 1.00
C2 97.13 97.13 1.00 73.03 0.75
D2 67.45 67.45 1.00 61.74 0.91
E2 49.56 49.56 1.00 48.67 0.98

150
G2 29.98 29.98 1.00 29.98 1.00
H2 14.37 14.37 1.00 14.37 1.00

I2 7.49 7.49 1.00 7.49 1.00
C3 124.67 124.67 1.00 75.99 0.61
D3 86.57 86.57 1.00 66.06 0.76
E3 63.61 63.61 1.00 56.66 0.89

200
G3 38.48 38.48 1.00 38.48 0.99
H3 18.44 18.44 1.00 18.44 1.00
I3 9.62 9.62 1.00 9.62 1.00
C4 152.21 152.21 1.00 73.66 0.50
D4 105.70 105.70 1.00 67.20 0.64

E4 77.66 77.66 1.00 60.40 0.78
250
G4 46.98 46.98 1.00 43.40 0.92
H4 22.52 22.52 1.00 22.52 1.00
I4 11.74 11.74 1.00 11.74 1.00
107
Figure 5-6 Critical values of simply supported castellated beams subjected to a uniformly
distributed load, obtained from linear analytical solution and nonlinear 3D finite element
analysis for groups C, D, E, G, H, and I with different flange widths (a) bf =100mm,
a=100mm)
108
5.4.2. Investigating load-deflection response curve of pinned-fixed beams

subjected to uniformly distributed loads
Figures 5-7, 5-8, 5-9 and 5-10 show a comparison of the curves of load versus
deflection of pinned-fixed castellated beams subjected to a uniformly distributed load.
The comparison involves the results of linear analytical solutions, which is developed in
chapter three, nonlinear 3D finite element analysis using ANSYS software, and
deflection limit (l/250) for groups C, D, E, G, H, and I with different flange widths (see
Section 3.3.4). The load is presented as the increments of load calculated following
Eq. (3-31).
From these figures it can be seen that the nonlinear behaviour of castellated beams, with
the same flange width, drops continuously with the increase of beam length, whereas
the behaviour of the beam towards linear behaviour progresses gradually. In contrast,
the nonlinear behaviour of the castellated beams with the same beam length increases
continuously with the increase of the flange width, thus the castellated beams will reach
the yield point in the early stages of the loading process.
As expected, the reason for this is that the dimensions of castellated beams and the yield
stress were taken account for calculating the load capacity of the beams, which was
applied in this work (see Eq. (3-31)), where the yield stress is identical for all groups of
castellated beams. Thus, the loading capacity of the beams is affected by the change of
the beam dimensions. Note that, the load for the same beam length increases when the
flange width increases, which leads to being more susceptible to yielding in the early
stages of the loading process. In contrast, the load with the same flange width reduces
when the beam length increases, which leads to being less susceptible to yielding in the
early stages of the loading process.
Table (5-2) presents the comparison of the critical value of loads (𝑞𝑐𝑟 /𝑞𝑦𝑖𝑒𝑙𝑑 ) for
beams with the different length for four groups with various flange widths. The results
are also plotted in Figure 5-11. For the purpose of comparison, which include the
results obtained from the linear analytical solutions and from the nonlinear 3D finite
element analyses. It is evidence from the figure that the critical value of load is
influenced by geometry, web openings and material beam. As a result, the designer
109
should consider to use a nonlinear analysis for short castellated beams with wide flange
for design calculations.
For short beams or the beams with wide flanges, the overall deflections of these beams
are rather small. Before the deflection of the beam reaches to the critical value of l /250,
the material has already yielded, and in this case the material nonlinearity must be
considered and the relationship between the deflection and the externally applied load is
NOT linear.
Compare the results of the beams with simply support boundary, it is noted that the
nonlinearity of the beams with pinned-fixed boundary is more sever because the
deflection of simply supported beams is more than the deflection of the pinned-fixed
beam.
110
Figure 5-7 Comparison of the curves of load versus deflection of pinned-fixed castellated
beams subjected to a uniformly distributed load, obtained from linear analytical solutions,
nonlinear 3D finite element analysis, and deflection limit (l/250) for beams C1, D1, E1, G1, H1,
and I1.
111
Figure 5-8 Comparison of the curves of load versus deflection of pinned-fixed castellated beam
subjected to a uniformly distributed load, obtained from linear analytical solutions, nonlinear
3D finite element analysis, and deflection limit (l/250) for beams C2, D2, E2, G2, H2, and I2.
112
Figure 5-9 Comparison of the curves of load versus deflection of pinned-fixed castellated beam
subjected to a uniformly distributed load, obtained from linear analytical solutions, nonlinear
3D finite element analysis, and deflection limit (l/250) for beams C3, D3, E3, G3, H3, and I3.
113
Figure 5-10 Comparison of the curves of load versus deflection of pinned-fixed castellated
beam subjected to a uniformly distributed load, obtained from linear analytical solutions,
nonlinear 3D finite element analysis, and deflection limit (l/250) for beams C4, D4, E4, G4, H4,
and I4.
114
Table 5-2 Comparison of results (𝑞𝑐𝑟 /𝑞𝑦𝑖𝑒𝑙𝑑 ) between linear analytical solution and nonlinear
3D finite element analysis of pinned-fixed castellated beams subjected to a uniformly
distributed load, for groups C, D, E, G, H, and I beams with different flange widths.
Nonlinear 3D finite
Linear analytical
element analysis by
bf Name 𝑞𝑦𝑖𝑒𝑙𝑑 solution
Ansys
mm of beam N/mm 𝑞𝑐𝑟 𝑞𝑐𝑟
𝑞𝑐𝑟 𝑞𝑐𝑟
N/mm qyield N/mm qyield
C1 69.59 69.59 1.00 48.40 0.70
D1 48.33 48.33 1.00 41.49 0.86
E1 35.51 35.51 1.00 33.92 0.96

100
G1 21.48 21.48 1.00 21.48 1.00
H1 10.29 10.29 1.00 10.29 1.00

I1 5.37 5.37 1.00 5.37 1.00
C2 97.13 97.13 1.00 54.09 0.56
D2 67.45 67.45 1.00 48.27 0.72
E2 49.56 49.56 1.00 40.91 0.83

150
G2 29.98 29.98 1.00 28.05 0.94
H2 14.37 14.37 1.00 14.37 1.00
I2 7.49 7.49 1.00 7.49 1.00

C3 124.67 124.67 1.00 54.54 0.44
D3 86.57 86.57 1.00 50.40 0.58
E3 63.61 63.61 1.00 42.51 0.67

200
G3 38.48 38.48 1.00 31.03 0.81
H3 18.44 18.44 1.00 17.96 0.97
I3 9.62 9.62 1.00 9.62 1.00
C4 152.21 152.21 1.00 57.43 0.38
D4 105.70 105.70 1.00 50.40 0.48

E4 77.66 77.66 1.00 44.71 0.57
250
G4 46.98 46.98 1.00 33.90 0.72
H4 22.52 22.52 1.00 21.02 0.93

I4 11.74 11.74 1.00 11.74 1.00
115
Figure 5-11 Critical values of pinned-fixed castellated beams subjected to a uniformly

distributed load, obtained from linear analytical solution and nonlinear 3D finite element
analysis for groups C, D, E, G, H, and I with different flange widths (a) bf =100mm,
a=100mm)
116
5.5. Nonlinear lateral-torsional buckling of castellated beams due to a uniformly

distributed load
A nonlinear inelastic finite-element analysis is provided to examine inelastic lateral-

torsional buckling behaviour of castellated beams subjected to uniformly distributed
load act on the top flange. In this section, the analyses are carried out by using ANSYS
mechanical (APDL) commercial software, which consider the effect of both the
geometric nonlinear and material inelasticity of castellated beams for calculating the
critical lateral-torsional buckling moment.
The analysis procedure is performed throughout two parts. Firstly, the manufacturing
method of castellated beams causes initial geometrical imperfections that lead to the
initial deformation. Therefore, linear lateral-torsional buckling is employed from which
the lateral-torsional buckling mode is used as the initial geometric imperfection of the
beam for the subsequent nonlinear analysis. Secondly, Newton-Raphson method with
the large deflection method, in conjunction with Arc-Length Method is utilized for
nonlinear lateral-torsional buckling analysis. The nonlinear analysis procedure used is
exactly the same as that described in previous sections and thus is not provided here
again.
5.5.1. Modelling, loading and boundary conditions
The modelling of castellated beams is carried out by using 3D 4-Node Thin Shell
Elements (SHELL181) depicted in Figure 3-5. This element presents four nodes with
six DOF per node, i.e. translations and rotations on the X, Y, and Z-axis, respectively.
The geometric models of castellated beams are meshed by using elements with size not
exceeding10 mm for both the pinned-pinned and pinned-fixed castellated beams. The
load is applied at the junction line between the web and upper flange of the beam. The
nodal load is calculated by the load density multiplied by beam length and divided by
the number of nodes on that line. The mechanical properties and boundary conditions
are the same as those described in the nonlinear analysis sections above.
117
5.5.2. Comparison of critical lateral-torsional bucking moment of simply

supported beams due to uniformly distributed loads
Table (5-3) gives the comparison of critical moments of lateral-torsional buckling of

simply supported castellated beams. This comparison involves the results of linear
analytical solutions, which are developed in chapter four, linear buckling 3D finite
element analysis using ANSYS, and nonlinear 3D finite element analysis (geometric
nonlinear and material inelasticity) using ANSYS software for groups C, E, G, H, I, and
J beams with different flange widths (see Section 3.3.4), in which the value of linear
critical lateral-torsional buckling moment (Mcr ) was obtained by Eq. (4-16), tabulated
under term linear lateral-torsional buckling analytical analysis. The results that are
determined by linear finite element analysis are tabulated under term linear lateral-
torsional buckling analysis by ANSYS software and the results that are determined by
nonlinear finite element analysis are tabulated under term nonlinear lateral-torsional
buckling analysis by ANSYS software. Moreover, the yield moment is obtained by
2𝜎𝑦 𝐼𝑟𝑒𝑑𝑢𝑐𝑒𝑑 𝑁
Myield = , (𝜎𝑦 = 275 𝑚𝑚2 , 𝐼𝑟𝑒𝑑𝑢𝑐𝑒𝑑 is obtained by following Eq. (3-17)),
ℎ𝑤 +2𝑡𝑓
where it is tabulated under term.( Myield ) For the purpose of comparison, the results
shown in the table is also plotted in Figures 5-12 and 5-13 respectively.
From these figures, it can be seen that in each group of flange width, the curves of the
analytical solution and numerical analysis of both linear and nonlinear have a similar
variation pattern. In other words, the analytical solution is in excellent agreement with
numerical analysis by ANSYS for all models of castellated beams. However, it is
observed from these figures that in the case of nonlinear analysis of castellated beams,
the critical lateral torsion buckling load (Mcr /Myield ) drops with the increase the flange
width and the decrease of beam length. The previous studies mentioned that the reasons
for this are because lateral-torsional buckling load of castellated beam is influenced by
the lateral flexural and warping rigidities (Mohebkhah, 2011). In this study, the
increasing flange width leads firstly to increase lateral flexural and warping rigidities,
and secondly to increase applying moments on castellated beams. These issues indicate
that the geometric nonlinear and material nonlinear lateral-torsional buckling resistance
of castellated beams with short beam length is limited to the ultimate load carrying
capacity, in which no lateral-torsional buckling occurs. In practice, Mcr ≫ Myield will
118
not happen (it’s not real) because the beam will fail by yield. When the Mcr is close to
the Myield there are some interactions between Myield and Mcr , in which case
Mcr ≫ Myield needs to be considered. This is why in the plots the part of
Mcr ≫ Myield curves is included.
Figure 5-14 shows a comparison of the load-deflection curves of simply supported

castellated beam subjected to a uniformly distributed load. This comparison involves
the results of nonlinear 3D finite element analysis using ANSYS software, and
deflection limit (l/250) to groups C, E, G, H, I and J beams with different flange widths
(see Section 3.3.4). The load is presented as the increments of load calculated following
by Eq. (3-31). This latter reflects that the critical lateral-torsional buckling for short
beam length with wide flange is influenced by geometry, web openings and material
beam. As a result, the designer should consider a nonlinear analysis for short castellated
beam with wide flange width for design calculations.
Figure 5-15 presents the failure mode of the beam length (l = 3.5 m) with wide flange
width (bf =250mm) obtained from the numerical analysis. It can be observed that in the
nonlinear analysis, the failure load is controlled by the plastic/buckling load, while the
linear solution is controlled purely by the elastic buckling which is much higher than the
yield load
119
Table 5-3 Comparison of lateral-torsional buckling results of simply supported castellated

beams subjected to a uniformly distributed load obtained from linear and nonlinear analysis
Linear buckling Linear buckling Nonlinear buckling
Name Myield analytical analysis analysis by Ansys analysis by Ansys
bf
of N.mm Mcr Mcr Mcr
mm beam x107 qcr Mcr qcr Mcr qcr Mcr
N.mm N.mm N.mm
N/mm 7 Myield N/mm 7 Myield N/mm 7 Myield
x10 x10 x10
C1 10.44 35.50 5.33` 0.51 34.54 5.18 0.50 27.50 4.12 0.40
E1 10.44 12.23 3.59 0.34 12.10 3.56 0.34 10.80 3.18 0.30
G1 10.44 5.58 2.76 0.26 5.40 2.63 0.25 5.34 2.60 0.25
100
H1 10.44 1.90 1.93 0.18 1.88 1.90 0.18 1.93 1.96 0.19
I1 10.44 0.73 1.41 0.14 0.72 1.41 0.13 0.77 1.73 0.15
J1 10.44 0.45 1.20 0.11 0.45 1.18 0.11 0.45 1.19 0.11
C2 14.57 91.28 13.69 0.94 90.56 13.58 0.93 58.00 8.70 0.60
E2 14.57 28.44 8.36 0.57 28.29 8.32 0.57 23.22 6.83 0.47
G2 14.57 12.64 6.14 0.42 12.74 6.19 0.42 11.60 5.64 0.39
150
H2 14.57 3.99 4.05 0.28 3.86 3.92 0.27 3.98 4.04 0.28
I2 14.57 1.51 2.94 0.20 1.58 3.07 0.21 1.67 3.24 0.22
J2 14.57 0.96 2.53 0.17 0.96 2.53 0.17 1.04 2.75 0.19
C3 18.70 192.00 28.80 1.54 185.71 27.86 1.49 93.55 14.23 0.76
E3 18.70 57.11 16.79 0.90 56.04 16.47 0.88 40.97 12.05 0.64
G3 18.70 23.97 11.65 0.62 23.67 11.51 0.62 20.19 9.79 0.52
200
H3 18.70 7.18 7.28 0.39 7.04 7.14 0.38 6.96 7.05 0.38
I3 18.70 2.61 5.07 0.27 2.61 5.07 0.27 2.50 4.86 0.26
J3 18.70 1.64 4.33 0.23 1.61 4.26 0.23 1.72 4.54 0.24
C4 22.83 356.67 53.50 2.34 317.97 47.70 2.09 96.35 14.45 0.63
E4 22.83 101.70 29.90 1.31 100.22 29.46 1.29 60.59 17.81 0.78
G4 22.83 41.15 20.00 0.88 40.79 19.82 0.87 31.60 15.36 0.67
250
H4 22.83 11.70 11.86 0.52 11.59 11.76 0.51 11.10 11.26 0.49
I4 22.83 4.09 79.61 0.35 4.10 7.98 0.35 3.94 7.66 0.34
J4 22.83 2.53 67.06 0.29 2.51 6.64 0.29 2.71 7.17 0.31
120
Figure 5-12 Comparison of critical moments (Mcr /Myield ) of lateral-torsional buckling of

simply supported castellated beam subjected to a uniformly distributed load, obtained from
linear analytical solutions, linear 3D finite element analysis using ANSYS, and nonlinear 3D
finite element analysis (geometric nonlinear and material inelasticity) for groups C, E, G, H, I,
and J beams with various flange widths (a) bf =100 mm , (b) bf =150 mm , (c) bf =200 mm , and
(d) bf =250 mm)
121

simply supported castellated beam subjected to a uniformly distributed load, obtained from
linear analytical solutions, linear 3D finite element analysis using ANSYS, and nonlinear 3D
finite element analysis (geometric nonlinear and material inelasticity) using ANSYS software
for groups C, E, G, H, I, and J beams with various flange width.
122
Figure 5-14 Comparison of the load-deflection curves of simply supported castellated beams
subjected to a uniformly distributed load, obtained from nonlinear 3D finite element analysis
and deflection limit (l/250) for groups C, E, G, H, I and J beams with various flange widths.
123
Figure 5-15 Failure mode of simply support castellated beam subjected to a uniformly
distributed load (C4) with flange widths bf=250mm obtained from the nonlinear lateral-torsional
buckling 3D finite element analysis using ANSYS software (hw=300mm, tf=10mm, tw=8mm
and a=100mm)
5.5.3. Comparison of critical moments of lateral-torsional bucking of pinned-fixed

beams subjected to uniformly distributed load
Table (5-4) shows the comparison of critical moments of lateral-torsional buckling of

pinned-fixed castellated beams subjected to uniformly distributed load. This comparison
involves the results of linear analytical solutions, which are developed in chapter four,
linear 3D finite element analysis using ANSYS, and nonlinear 3D finite element
analysis (geometric nonlinear and material inelasticity) using ANSYS software for
groups C, E, G, H, I, and J beams with different flange widths (see Section 3.3.4), in
which the value of linear critical lateral-torsional buckling moment (Mcr ) was obtained
by Eq. (4-34), tabulated under term linear lateral-torsional buckling analytical analysis.
The results that are determined by linear finite element analysis are tabulated under term
linear lateral-torsional buckling analysis by ANSYS software, and the results that are
determined by nonlinear finite element analysis are tabulated under term nonlinear
124
lateral-torsional buckling analysis by ANSYS software. Moreover, the yield moment is

2𝜎𝑦 𝐼𝑟𝑒𝑑𝑢𝑐𝑒𝑑 𝑁
obtained by Myield = , (𝜎𝑦 = 275 𝑚𝑚2 , 𝐼𝑟𝑒𝑑𝑢𝑐𝑒𝑑 is obtained by following
ℎ𝑤 +2𝑡𝑓
Eq. (3-17)), where it is tabulated under term (Myield ).The results shown in table are
plotted in Figures 5-16 and 5-17, respectively.
From these figures, it can be seen that in each group of flange width, the curves of the
analytical solution and numerical analysis of both linear and nonlinear have a similar
variation pattern. In other words, the analytical solution is in excellent agreement with
numerical analysis by ANSYS for all models of castellated beams. However, it is
observed from these figures that in the case of nonlinear analysis of castellated beams,
the critical lateral-torsional buckling load (Mcr /Myield ) drops with the increase of the
flange width and the decrease of beam length. The previous studies mentioned that the
reason for this is that lateral-torsional buckling load is influenced by the lateral flexural
and warping rigidities (Mohebkhah, 2011). In this study, the increasing flange width
leads firstly to increase lateral flexural and warping rigidities, and secondly to increase
the moments on castellated beams. These items indicate that the nonlinear lateral-
torsional buckling resistance of castellated beams with short beam length is limited by
the ultimate load carrying capacity, in which no lateral-torsional buckling occurs. In
addition, the web opening under high loads makes the castellated beam more prone to
compression buckling of web; and failure can occur in local loading areas or reaction
force region (Kerdal and Nethercot, 1984).
Figure 5-18 shows a comparison of the load-deflection curves of pinned-fixed

castellated beams. The comparison involves the results of nonlinear 3D finite element
analysis using ANSYS software, and deflection limit (l/250) for groups C, E, G, H, I
and J beams with different flange widths (see Section 3.3.4). The load is presented as
the increments of load calculated following by Eq. (3-31). This latter reflects that the
critical lateral-torsional buckling for short beam length with wide flange is influenced
by geometry, web openings, boundary conditions, and material properties of the beam.
As a result, the designer should consider a nonlinear analysis for short castellated beams
with wide flange for design calculations. The failure mode of short castellated beams is
dominated by the plastic failure, whereas the failure mode of long castellated beams is
dominated by the lateral-torsional buckling failure mode, that is observed in
125
Figure 5-19 which obtained from the numerical analysis for the beam of length
(l = 3.5 m) with wide flange width (bf = 200 mm and bf =250mm).
The figures show there is a significant difference between the linear solution and
nonlinear solution, particularly for short beams. The reason for this is because for the
short beams the failure load predicted by the linear solution is controlled purely by the
elastic buckling which is much higher than the yield load. While in the nonlinear
analysis, the failure load is controlled by the plastic/buckling load. While the long
beams usually fail by elastic buckling (the stress is lower than the yield stress) and this
is why the linear and nonlinear solutions give almost the same result for long beams but
not for the short beams.
126
Table 5-4 Comparison of lateral-torsional buckling results of pinned-fixed castellated beams

subjected to a uniformly distributed load obtained from linear and nonlinear analysis
Linear buckling Linear buckling Nonlinear buckling
Name Myield analytical analysis analysis by Ansys analysis by Ansys
bf
of N.mm Mcr Mcr Mcr
mm beam x107 qcr Mcr qcr Mcr qcr Mcr
N.mm N.mm N.mm
N/mm 7 Myield N/mm 7 Myield N/mm 7 Myield
x10 x10 x10
C1 10.44 80.43 12.06 1.16 79.31 11.90 1.14 50.85 7.63 0.73
E1 10.44 25.79 7.58 0.73 26.02 7.65 0.73 22.10 6.50 0.62
G1 10.44 11.55 5.61 0.54 11.78 5.73 0.55 11.64 5.66 0.54
100
H1 10.44 3.80 3.89 0.40 4.12 4.17 0.41 4.34 4.40 0.42
I1 10.44 1.49 2.89 0.28 1.59 3.13 0.30 1.99 3.86 0.29
J1 10.44 0.92 2.43 0.23 0.91 2.40 0.23 1.20 3.18 0.25
C2 14.57 233.13 34.97 2.40 207.30 31.09 2.13 81.97 12.30 0.84
E2 14.57 68.29 20.08 1.38 66.07 19.42 1.33 41.66 12.25 0.84
G2 14.57 28.47 13.84 0.95 28.31 13.76 0.94 22.47 10.90 0.75
150
H2 14.57 8.50 8.62 0.59 8.84 8.96 0.62 8.41 8.53 0.59
I2 14.57 3.10 6.03 0.41 3.15 6.12 0.42 3.10 6.03 0.41
J2 14.57 1.95 5.17 0.35 1.94 5.13 0.35 1.98 5.24 0.36
C3 18.70 522.91 78.44 4.19 372.75 55.91 3.00 85.71 12.85 0.69
E3 18.70 146.00 42.92 2.30 134.80 39.63 2.12 58.37 17.16 0.92
G3 18.70 58.07 28.22 1.51 56.53 27.48 1.47 35.16 17.09 0.91
200
H3 18.70 16.07 16.30 0.87 16.57 16.80 0.90 13.98 14.18 0.76
I3 18.70 5.53 10.75 0.57 5.60 10.88 0.58 5.92 11.51 0.62
J3 18.70 3.41 9.02 0.48 3.43 9.07 0.48 3.52 9.32 0.50
C4 22.83 998.80 149.82 6.56 391.17 58.67 2.60 87.40 13.11 0.57
E4 22.83 272.04 79.98 3.50 230.52 67.77 2.97 69.84 20.53 0.90
G4 22.83 105.29 51.17 2.24 101.64 49.40 2.16 48.11 23.38 1.02
250
H4 22.83 27.64 28.03 1.23 28.11 28.51 1.25 22.36 22.67 0.99
I4 22.83 9.04 17.57 0.77 9.12 17.73 0.78 9.29 18.06 0.79
J4 22.83 5.55 14.68 0.64 5.75 15.21 0.67 5.18 13.71 0.60
127

pinned-fixed castellated beams subjected to a uniformly distributed load, obtained from linear
analytical solutions, linear 3D finite element analysis using ANSYS, and nonlinear 3D finite
element analysis (geometric nonlinear and material inelasticity) using ANSYS software for
groups C, E, G, H, I, and J beams with various flange widths (a) bf =100 mm ,(b) bf =150
mm ,(c) bf =200 mm , and (d) bf =250 mm
128

pinned-fixed castellated beams subjected to a uniformly distributed load, obtained from linear
analytical solutions, linear 3D finite element analysis using ANSYS, and nonlinear 3D finite
element analysis (geometric nonlinear and material inelasticity) using ANSYS software for
groups C, E, G, H, I, and J beams with various flange widths.
129
Figure 5-18 Comparison of the load-deflection curves of pinned-fixed castellated beams

subjected to a uniformly distributed load, obtained from nonlinear 3D finite element analysis,
and deflection limit (l/250) for groups C, E, G, H, I and J beams with various flange widths.
130
Beam C3 Beam C4
Figure 5-19 Combined failure modes of pinned-fixed castellated beams subjected to a

uniformly distributed load (C3 and C4) with two flange widths bf=200mm and bf=250mm
obtained from the nonlinear lateral-torsional buckling 3D finite element analysis using ANSYS
software (hw=300mm, tf=10mm, tw=8mm and a=100mm)
5.6. Conclusions
In this chapter, geometric nonlinear and material nonlinear analysis by using finite
element method has been carried out to investigate the behaviour of pinned-pinned and
pinned-fixed castellated beams subjected to uniformly distributed load at the inelastic
range. Comparison has been made between the result of the linear and nonlinear
analyses. The main conclusions can be summarized as follows:
 The load carrying capacity of castellated beams obtained by nonlinear 3D finite

element analysis is influenced by the geometry, web openings, boundary
conditions, and material properties of the beams.
 The load carrying capacity of castellated beams obtained by using the nonlinear 3D
finite element analysis is generally less than that obtained by using the linear
analysis method. This reflects that the elastic range is unsafe for short length beams
with wide flange width.
131
 The critical moments of lateral-torsional buckling of castellated beams obtained by

using nonlinear 3D finite element analysis are influenced by the geometry, web
openings, boundary conditions, and mechanical properties of the beams.
 The critical moment of lateral-torsional buckling of castellated beams obtained by

using nonlinear 3D finite element analysis are generally less than those obtained
using the linear analysis methods. This indicates that the elastic range is unsafe for
short length beams with wide flange.
 The nonlinear lateral-torsional buckling resistance of castellated beam with the

short beam length under high loads is limited by the ultimate load carrying capacity
of the beams, in which case no lateral-torsional buckling occurs.
 In nonlinear analysis, increasing castellated beam length leads to beam behaviour

that will be similar to that in the linear analysis, and the lateral-torsional buckling
mode will dominate the failure mode.
 When the serviceability is also considered, the deflection limit seems to be the
dominant criterion in controlling the load in most of the beam length regions.
132
CHAPTER SIX DYNAMIC INSTABILITY
CHAPTER SIX
6. DYNAMIC INSTABILITY OF CASTELLATED BEAMS UNDER

TRANSVERSE PERIODIC LOADING
6.1. Introduction
In this chapter, an analytical solution is developed to investigate the free vibration, static
buckling and dynamic instability of castellated beams subjected to transverse periodic
loading. Bolotin’s method is used to perform the dynamic instability analysis that is
utilized in this study. By assuming the instability modes, the mass, stiffness, and
geometric stiffness matrices are derived using the kinetic energy, strain energy and
potential of applied loads. Analytical equations for determining the free vibration
frequency, critical buckling moment, and excitation frequency of castellated beams are
derived, In addition, the influences of the flange width of the castellated beam and the
static part of the applied load on the variation of dynamic instability zones are discussed.
6.2. The studies about dynamic instability
Literature survey on structural members shows that little research has been carried out
on the dynamic instability of castellated beams when the applied load varies with time.
In many countries, the static load still dominates the current designing of structures for
castellated beams, in spite of the significance of the dynamic response to machinery
loading and to extreme environmental loads, for example wind and earthquakes, that
have been considered for some time. It is acknowledged that applying static load can
lead to free vibration behaviour of the structural members, which causes a decrease in
the critical load of buckling of the members. For this reason, we should understand the
effect of applying the dynamic load on the structure behaviour to avoid resonance
disasters due to the dynamic instability.
133
As a result, limited research and studies exist on the dynamics, especially the dynamic
instability of the castellated beams subjected to transverse loading. Since 1960 research
has been carried out on the vibration-induced buckling of beams. For instance:
Morris (1965) investigated the nonlinear vibration problem of a two hinged beam-
column subjected to a harmonic load of any space distribution
In 1966, Hsu carried out the investigation of the dynamic stability of the elastic body
with given initial conditions and reported the necessary and sufficient stability criteria
in terms of trajectories in the phase space of finite dimension.
Huang (1980) and Chen et al. (1991) used the Bolotin’s method to examine the
dynamic instability of generally orthotropic beams and thick bi-modulus beams
subjected to periodic axial loads, respectively.
Huang and Hung (1984) used the averaging method and the Routh-Hurwitz stability to
study the dynamic instability of a simply supported beam under periodic axial excitation.
The coupling of the first two modes was considered to investigate the instability regions
and vibration amplitudes.
Gürgöze (1985) conducted an investigation into the instability behaviour of a pre-

twisted beam subjected to a pulsating axial force. They used the Mettler method and
derived the equations describing the instability regions that could be applied with
various different boundary conditions.
A finite element dynamic instability model of Timoshenko beams was introduced by

Park (1987) that adopted the extended Hamilton's principle to build the equation of the
beam transverse motion in the plane.
Kar and Sujata (1991) examined the dynamic instability of rotating beams with various
different boundary conditions, subjected to a pulsating axial excitation. As well, they
also discussed the effects of the boundary conditions, rotational speed on the static
buckling loads and the regions of parametric instability.
134
Numerous of the papers Uang, and. Fan (2001); Yoon, and. Kim (2002) have been
presented to study the influence of various different supporting boundary conditions on
the dynamic instability behaviour of beams
Yeh et al. (2004) used both the finite element method and the harmonic balance method
to present a study of the dynamic instability problem of a sandwich beam with a
constrained layer and an electro rheological fluid core subjected to an axial dynamic
force. Moreover, this study discussed the influences of the natural frequencies and static
buckling loads on the dynamic instability behaviour.
Zhu et al. (2017) conducted an analytical solution to examine the free vibration, static
buckling and dynamic instability of laterally-restrained zed-section purlin beams under
uplift wind loading. They used the classical principle of minimum potential energy
which assumed the instability modes, the kinetic energy and strain energy of the beam
and the loss of the potential energy of the applied load are evaluated, from which the
mass, stiffness and geometric stiffness matrices of the system are derived.
More recently, Zhu et al. (2018) presented a study on the dynamic buckling of cold-
formed steel channel section beams under the action of uniformly distributed loading.
Gao et al. (2019) provided a nondeterministic dynamic stability assessment of Euler–

Bernoulli beams using Chebyshev surrogate model.
6.3. Governing equations for dynamic instability analysis of castellated beams
The analysis model used for this study is illustrated in Figure 6-1 (a). The cross-section
of the castellated beam is assumed to be doubly symmetric, with the flange width and
thickness as bf and tf, the web depth and thickness as hw and tw, and the half depth of
hexagons as a. The half of the distance between the centroids of the two T-sections is e.
2a
The side length of the hexagonal opening is( ), and the hexagonal opening height is
√3
2a, respectively.
135
According to Figure 6-1 (b), the lateral and transverse displacements of the beam are
assumed to be v(x), w(x), respectively, and the angle of twist of the cross-section is ϕ(x).
In the linear situation, the strain energy of the beam involves two parts; the strain
energy generated by the bending and the energy generated by the twist.
In order to consider the warping influence, the cross-section of the castellated beam is
decomposed into three parts, two of which represent the top and bottom T-section, one
of which represents the middle-part of the web. It is assumed that the displacements at
the shear centers of the top and bottom T-sections are small and can be expressed as
follows (see Figure 6-1) :( Kim et al., 2016)
ℎ
𝑣1 (𝑥) = 𝑣(𝑥) + 𝜙(𝑥) 6-1
2
ℎ
𝑣2 (𝑥) = 𝑣(𝑥) − 𝜙(𝑥) 6-2
2
ℎ
𝑤1 (𝑥) = 𝑤 + (1 − 𝑐𝑜𝑠𝜙) ≈ 𝑤 6-3
2
ℎ
𝑤2 (𝑥) = 𝑤 + (1 − 𝑐𝑜𝑠𝜙) ≈ 𝑤 6-4
2
where 𝑣1 and 𝑣2 are the lateral displacements of the shear centre of the top and bottom
T-section, 𝑤1 and 𝑤2 are the transverse displacements of the shear centre of the top and
bottom T-section, h is the distance between the shear centres of top and bottom
T-sections. Hence, the kinetic energy T of castellated beam due to the transverse
displacement, lateral displacement and rotation thus can be expressed as:
The kinetic energy for the top T- section:
𝜌𝐴𝑡𝑒𝑒 𝑙 ̇ 2 𝜌𝐼𝑝𝑡𝑜𝑝 𝑙 2
𝑇𝑡𝑜𝑝 = ∫ ( 𝑣1 + 𝑤1̇ 2 )𝑑𝑥 + ∫ 𝜙̇ 𝑑𝑥 6-5
2 0 2 0
The kinetic energy for the bottom T- section:
𝜌𝐴𝑡𝑒𝑒 𝑙 ̇ 2 ̇
𝜌𝐼𝑝𝑏𝑜𝑡 𝑙 2
𝑇𝑏𝑜𝑡 = 2
∫ ( 𝑣2 + 𝑤2 )𝑑𝑥 + ∫ 𝜙̇ 𝑑𝑥 6-6
2 0 2 0
The kinetic energy for the middle part between the two T- sections
136
𝜌2𝑎𝑡𝑤 𝑙 ̇ 2 ̇
𝜌𝐼𝑝𝑤𝑒𝑏 𝑙 2
𝑇𝑤𝑒𝑏 = 2
∫ ( 𝑣 + 𝑤 )𝑑𝑥 + ∫ 𝜙̇ 𝑑𝑥 6-7
2 0 2 0
Hence, the total kinetic energy of the beam is:

𝑙 𝑙
𝜌
𝑇= [𝐴𝑡𝑒𝑒 [∫ ( 𝑣1̇ 2 + 𝑤1̇ 2 )𝑑𝑥 + ∫ ( 𝑣2̇ 2 + 𝑤2̇ 2 )𝑑𝑥]
2 0 0
𝑙 𝑙
6-8
+ 𝑎𝑡𝑤 ∫ ( 𝑣̇ 2 + 𝑤̇ 2 )𝑑𝑥 + 𝐼𝑝 ∫ 𝜙̇ 2 𝑑𝑥]
0 0
hw
where ρ the density, l is the beam length, Atee= bf t f + t w ( − a) is the cross-section
2
area of the T-section, Ip = Iptop + Ipweb + Ipbot is the polar moment of inertia. Note
that the dot above a symbol in above equations represents the derivative of the symbol
with respect to time.
The strain energy of castellated beam that is determined based on the three parts due to
the transverse displacement, lateral displacement and rotation. It thus can be written as
follows:
2 2
1 𝑙 𝑑2 𝑤1 𝑑2 𝑣1 𝑑𝜙 2
𝑈𝑠 = ∫ [𝐸𝐼𝑦1 ( 2 ) + 𝐸𝐼𝑧1 ( 2 ) + 𝐺𝐽1 ( ) ] 𝑑𝑥
2 2
1 𝑙 𝑑2 𝑤2 𝑑2 𝑣2 𝑑𝜙 2
+ ∫ [𝐸𝐼𝑦2 ( 2 ) + 𝐸𝐼𝑧2 ( 2 ) + 𝐺𝐽2 ( ) ] 𝑑𝑥
2 0 𝑑𝑥 𝑑𝑥 𝑑𝑥 6-9
𝑙 2 2 2 2 2
1 𝑑 𝑤3 𝑑 𝑣3 𝑑𝜙
+ ∫ [𝐸𝐼𝑦3 ( 2 ) + 𝐸𝐼𝑧3 ( 2 ) + 𝐺𝐽3 ( ) ] 𝑑𝑥
where 𝑈𝑠 is the strain energy, E is the Young's modulus, G is the shear modulus.
𝐼𝑦1 = 𝐼𝑦2 and 𝐼𝑧1 = 𝐼𝑧2 are the second moments of the T- sectional area about the y and z
axes. 𝐽1 = 𝐽2 is the torsional constant of the tee-section, 𝐼𝑦3 and 𝐼𝑧3 are the second
moments of the cross-sectional area of the mid-part of the web about the y and z axes
respectively, and 𝐽3 is the torsional constant of the mid-part of the web.
Hence, the formula of the strain energy of castellated beam (top T- section, bottom
T- section and mid-part of the web) can be written as follows:
137
2 2
1 𝑙 𝑑2 𝑤1 𝑑2 𝑣1
𝑈 = ∫ [2𝐸𝐼𝑦1 ( 2 ) + 2𝐸𝐼𝑧1 ( 2 )
2
ℎ2 𝑑2 𝜙 𝑑𝜙 2
+ 𝐸𝐼𝑧1 ( 2 ) + 2𝐺𝐽1 ( ) ] 𝑑𝑥
2 𝑑𝑥 𝑑𝑥 6-10
𝑙 2 2 2 2 2
1 𝑑 𝑤2 𝑑 𝑣2 𝑑𝜙
+ ∫ [𝐸𝐼𝑦3 ( 2 ) + 𝐸𝐼𝑧3 ( 2 ) + 𝐺𝐽3 ( ) ] 𝑑𝑥
According to Figure 6-1, 𝐼𝑦1 , 𝐼𝑧1 and 𝐽1 are constants, whereas 𝐼𝑦3 , 𝐼𝑧3 and 𝐽3 are the
function of x and depending upon the location of the web openings. Hence, from the
comparison between equation of the strain energy of an I-beam without web openings
and Eq. (6-10), the following relations can be obtained:
𝐼𝑦 = 2𝐼𝑦1 + 𝐼𝑦3 6-11
𝐼𝑧 = 2𝐼𝑧1 + 𝐼𝑧3 6-12
ℎ 2 ℎ2
𝐼𝑤 = ( ) 𝐼𝑧 ≈ 𝐼𝑧1 6-13
2 2
𝐽 = 2𝐽1 + 𝐽3 6-14
𝐼𝑝 = 𝐼𝑦 + 𝐼𝑧 = 2𝐼𝑦1 + 2𝐼𝑧1 + 𝑘𝐼𝑦3 + 𝑘𝐼𝑧3 6-15
According to Kim.et al. (2016) k refers to the fraction of the volume of the solid and
holes in the mid-part of the web. In castellated beams, because of matching of the areas
and holes in the mid-part of the web, the value of k=0.5.
Assume that the transverse load is the periodic load applying on the top flange of
castellated beam when the sheeting is fixed with the top flange (e.g. for wind-induced
vibration). In this case, the loss of potential energy 𝑉 of the transverse load 𝑞𝑧 can be
expressed as follows:
𝑙
𝑑2𝑣 𝑎𝑧 𝑞𝑧 2
𝑉 = ∫ [𝑀𝑦 𝜙 ( 2
)+ 𝜙 ] 𝑑𝑥 6-16
0 𝑑𝑥 2
138
where 𝑞𝑧 is the distribution load, 𝑀𝑦 is the pre-buckling internal bending moment, and
a𝑧 refers to the z-coordinate of the loading point, which is equal to the distance between
ℎ𝑤
the loading point and the shear centre of the beam. In the present case, 𝑎𝑧 = + 𝑡𝑓
2
because the uniformly distributed load is applied on the top flange of the beam. The
second term in Eq. (6-16) is attributed to the effect of loading position, which, in the
present case, has a positive effect on the stability of the beam and thus will increase the
critical buckling load.
According to the Lagrange method, the equations of motion describing the lateral-
torsional buckling of the beam can be expressed as follows: (Massachusetts Institute of
Technology, 2003)
𝑑 𝜕𝐿 𝜕𝐿
( )− =0 6-17
𝑑𝑡 𝜕𝑞̇ 𝜕𝑞
where 𝐿 = 𝑇 − (𝑈 − 𝑉) is the Lagrangian function and 𝑞 = {𝑞1 , 𝑞2 , 𝑞3 }𝑇 is the general

displacement vector. Substituting Eqs (6-1)-(6-4), (6-8), (6-10) and (6-16) into (6-17),
the governing equation for the dynamic instability analysis of a castellated beam is
obtained, as follows: (Kratzig and Nawrotzki, 1991; Li, 1991; and Patel et al., 2006).
[𝐌]{𝐪̈ } + [𝐊]{𝐪} − λ[𝐊 𝐠 ]{𝐪} = {𝟎} 6-18
where [M] is the mass matrix, [K] is the elastic stiffness matrix, [Kg] is the geometric
stiffness matrix, {𝐪̈ } is the generalized acceleration vector, is the general displacement
vector, and λ is the loading factor. The mass, stiffness, and geometric stiffness matrices
are expressed as follows:
𝜕 2𝑇 𝜕 2𝑇 𝜕 2𝑇
𝜕𝑞̇ 12 𝜕𝑞1̇ 𝜕𝑞2̇ 𝜕𝑞1̇ 𝜕𝑞3̇
𝜕 2𝑇 𝜕 2𝑇 𝜕 2𝑇
[𝐌] = 6-19
𝜕𝑞2̇ 𝜕𝑞1̇ 𝜕𝑞̇ 22 𝜕𝑞2̇ 𝜕𝑞3̇
𝜕 2𝑇 𝜕 2𝑇 𝜕 2𝑇
[𝜕𝑞3̇ 𝜕𝑞1̇ 𝜕𝑞3̇ 𝜕𝑞2̇ 𝜕𝑞̇ 32 ]
139
𝜕 2𝑈 𝜕 2𝑈 𝜕 2𝑈
𝜕𝑞12 𝜕𝑞1 𝜕𝑞2 𝜕𝑞1 𝜕𝑞3
𝜕 2𝑈 𝜕 2𝑈 𝜕 2𝑈
[𝐊] = 6-20
𝜕𝑞2 𝜕𝑞1 𝜕𝑞22 𝜕𝑞2 𝜕𝑞3
𝜕 2𝑈 𝜕 2𝑈 𝜕 2𝑈
[𝜕𝑞3 𝜕𝑞1 𝜕𝑞3 𝜕𝑞2 𝜕𝑞32 ]
𝜕 2𝑉 𝜕 2𝑉 𝜕 2𝑉
𝜕𝑞12 𝜕𝑞1 𝜕𝑞2 𝜕𝑞1 𝜕𝑞3
𝜕 2𝑉 𝜕 2𝑉 𝜕 2𝑉
[𝐊 𝒈 ] = 6-21
𝜕𝑞2 𝜕𝑞1 𝜕𝑞22 𝜕𝑞2 𝜕𝑞3
𝜕 2𝑉 𝜕 2𝑉 𝜕 2𝑉
[𝜕𝑞3 𝜕𝑞1 𝜕𝑞3 𝜕𝑞2 𝜕𝑞32 ]
Assume that the externally applied load 𝑞𝑧 is periodic, in which case the loading factor
can be divided into two parts as expressed as follows:
𝜆 = 𝜆𝑠 + 𝜆𝑡 𝑐𝑜𝑠Ω𝑡 6-22
Where 𝜆𝑠 and 𝜆𝑡 are the amplitudes of the static and dynamic parts, respectively, Ω is
the excitation frequency of the dynamic part of the load, and 𝑡 is the time.
The dynamic instability regions of the structure described by Eq. (6-18) can be
calculated by investigating periodic solutions with the periods of T=2π/Ω and 2T=4π/Ω.
The solution with the period of 2T is of particular importance, representing the primary
instability region of the structure, which can be expressed using the form of
trigonometric series given by:
𝑘Ω𝑡 𝑘Ω𝑡
{𝒒} = ∑ [{𝒂𝒌 }𝑠𝑖𝑛 + {𝒃𝒌 }𝑐𝑜𝑠 ] 6-23
2 2
𝑘=1,3,…
where {𝒂𝒌 } and {𝒃𝒌 } are the vectors of coefficients of the assumed solution.
Substituting Eqs. (6-22) and (6-23) into (6-18) and letting the coefficients of the series
associated with sin (Ωt/2) and cos (Ωt/2) be zero, it yields:
2λ𝑠 − λ𝑡 Ω2
([𝐊] − [𝐊 𝒈 ] − [𝐌]) {𝒂𝟏 } = {0} 6-24
2 4
140
2λ𝑠 + λ𝑡 Ω2
([𝐊] − [𝐊 𝒈 ] − [𝐌]) {𝒃𝟏 } = {0} 6-25
2 4
For given values of 𝜆𝑠 and 𝜆𝑡 one can calculate the two frequencies of Ω from
Eqs. (6-24) and (6-25), which represent the boundary of dynamic instability region of
the castellated beams under periodic loading.
6.4. Simply supported, doubly symmetric castellated beam subjected to periodic

loads on top flange
For the calculation due to the dynamic lateral-torsional buckling, the displacement
functions w(x), v(x), ϕ(x) and pre-buckling internal bending moment 𝑀𝑦 (𝑥) that satisfy
the boundary conditions of a simply supported beam can be assumed as follows:
𝜋𝑥
𝑣(𝑥) = 𝑞1 (𝑡)𝑠𝑖𝑛 6-26
𝑙
𝜋𝑥
𝑤(𝑥) = 𝑞2 (𝑡)𝑠𝑖𝑛 6-27
𝑙
𝜋𝑥
𝜙(𝑥) = 𝑞3 (𝑡)𝑠𝑖𝑛 6-28
𝑙
𝑞𝑧 𝑥(𝑙 − 𝑥)
𝑀𝑦 (𝑥) = 6-29
2
where 𝑞𝑖 (𝑡) (𝑖 = 1,2,3) are the functions of time.
Therefore, the mass, stiffness, and geometric stiffness matrices for simply supported
beam are obtained from Eqs. (6-19)- (6-21) and are expressed as follows:
𝑚11 0 0
[𝐌] = [ 0 𝑚22 0 ] 6-30
0 0 𝑚33
where:
𝐴𝑡𝑒𝑒 ℎ2 𝐼𝑝
𝑚11 = 𝑚22 = 𝜌𝑙(𝑎𝑡𝑤 + 𝐴𝑡𝑒𝑒 ) , 𝑚33 = 𝜌𝑙 ( + 2)
4
141
𝜅11 0 0
[𝐊] = [ 0 𝜅22 0 ] 6-31
0 0 𝜅33
where:
𝐸𝑙(2𝐼𝑧1 + 𝑘𝐼𝑧3 ) 𝜋 4
𝜅11 = ( )
2 𝑙
𝐸𝑙(2𝐼𝑦1 + 𝑘𝐼𝑦3 ) 𝜋 4
𝜅22 = ( )
2 𝑙
𝐸𝐼𝑤 𝑙 𝜋 4 𝐺𝑙(2𝐽1 + 𝑘𝐽3 ) 𝜋 2
𝜅33 = ( ) + ( )
2 𝑙 2 𝑙
0 0 𝜅𝑔13
[𝐊 𝒈 ] = [ 0 0 0 ] 6-32
𝜅𝑔31 0 𝜅𝑔33
where:
𝑞𝑧 𝑙 𝝅𝟐 𝑎𝑧 𝑞𝑧 𝑙
𝜅𝑔13 = 𝜅𝑔31 = − ( 𝟑 + 1) , 𝜅𝑔33 = −
8 2
6.4.1. The free vibration analysis
The free vibration analysis frequency of the lateral-torsional vibration of the castellated
beam can be determined using Eq. (6-33):
‖[𝐊] − 𝜔2 [𝐌]‖ = 0 6-33
where 𝝎 is the free vibration frequency. Substituting Eqs. (6-30) and (6-31) into (6-33),
the following frequency can be obtained:
𝜋 2 𝐸(2𝐼𝑧1 + 𝑘𝐼𝑧3 )
𝜔1 = ( ) √ 6-34
𝑙 2𝜌(𝑎𝑡𝑤 + 𝐴𝑡𝑒𝑒 )
𝜋 2 𝐸(2𝐼𝑦1 + 𝑘𝐼𝑦3 )
𝜔2 = ( ) √ 6-35
142
𝐺𝑙 2
2 (𝐸𝐼𝑤 + (2𝐽1 + 𝑘𝐽3 ))
𝜋 2√ 𝜋2 6-36
𝜔3 = ( )
𝑙 𝜌(𝐴𝑡𝑒𝑒 ℎ2 + 2𝐼𝑝 )
∗ ∗
𝐼𝑧3 can be negligible because in most of castellated beams 𝐼𝑧3 << 2𝐼𝑧1 , then
Eqs. (6-34), (6-35) and (6-36) can be simplified as follows: (Kim et al., 2016)
𝜋 2 𝐸(2𝐼𝑧1 )
𝜔1 = ( ) √ 6-37
𝜋 2 𝐸(2𝐼𝑦1 )
𝜔2 = ( ) √ 6-38
𝐺𝑙 2
2 (𝐸𝐼𝑤 + (2𝐽1 + 𝑘𝐽3 ))
𝜋 2√ 𝜋2 6-39
𝜔3 = ( )
𝑙 𝜌(𝐴𝑡𝑒𝑒 ℎ2 + 2(𝐼𝑧1 + 𝐼𝑦1 ))
The above formulations (6-37), (6-38) and (6-39) calculate the natural frequencies,
which are well known and can be found from many vibration textbooks. These
equations represent the translational and rotational vibrations of castellated beams.
Moreover, it indicates that the lateral vibration and torsional vibration modes are
influenced by web openings.
6.4.2. Buckling analysis
The critical load of the lateral-torsional buckling of the castellated beam subjected to a
static load can be calculated using Eq. (6-40):
‖[𝐊] − λ𝑐𝑟 [𝐊 𝒈 ]‖ = 0 6-40
where λ𝑐𝑟 is the loading factor and 𝑞𝑐𝑟 = λ𝑐𝑟 𝑞𝑜 is the critical load for static buckling.
Substituting Eqs. (6-31) and (6-32) into (6-40), the following critical load is obtained:
143
𝐾11 𝐾𝑔33 ± √𝐾11 2 𝐾𝑔33 2 + 4𝐾𝑔13 2 𝐾11 𝐾33

6-41
λ𝑐𝑟 = −
2𝐾𝑔13 2
𝟐 𝟐
𝒉𝒘 𝒉 𝟏 𝑮(𝟐𝑱𝟏 + 𝒌𝑱𝟑 ) 𝒍 𝟐 𝝅𝟐 𝟏
(( + 𝒕𝒇 ) + √( 𝒘 + 𝒕𝒇 ) + (𝑰𝒘 + ( ) ) ( + 𝟏) × )
𝟐 𝟐 𝟒 𝑬 𝝅 𝟑 (𝟐𝑰𝒛𝟏 + 𝑰𝒛𝟑 )
𝒒 𝒛 𝒍𝟐
( ) =
𝟖 𝒄𝒓 𝟏 𝟏 𝟐
( + 𝟐) 6-42
𝟑 𝝅
𝟐𝑬(𝟐𝑰𝒛𝟏 + 𝑰𝒛𝟑 )
×
𝒍𝟐
𝟐 𝟐
𝒉𝒘 𝒉 𝟏 𝑮(𝟐𝑱𝟏 + 𝒌𝑱𝟑 ) 𝒍 𝟐 𝝅𝟐 𝟏
(( + 𝒕𝒇 ) + √( 𝒘 + 𝒕𝒇 ) + (𝑰𝒘 + ( ) ) ( + 𝟏) × )
𝟐 𝟐 𝟒 𝑬 𝝅 𝟑 (𝟐𝑰𝒛𝟏 )
𝒒 𝒛 𝒍𝟐 𝟐𝑬(𝟐𝑰𝒛𝟏 ) 6-43
( ) = ×
𝟖 𝒄𝒓 𝟏 𝟏 𝟐 𝒍𝟐
( + 𝟐)
𝟑 𝝅
It can be noticed that Eq. (6-43) is similar to the formulation of the critical load given in
(Kim et al., 2016) for simply supported castellated beams when the load is applied at the
top flange.
6.4.3. The dynamic instability
The dynamic instability region of the castellated beam can be calculated using
Eq. (6-44):
2λ𝑠 ± λ𝑡 Ω2
‖[𝐊] − [𝐊 𝒈 ] − [𝐌]‖ = 0 6-44
2 4
Substituting Eqs. (6-30), (6-31) and (6-32) into (6-44), it yields:
2𝜆 ± 𝜆𝑡 2 6-45
∗
(𝐾33 ∗
𝑚11 + 𝐾11 𝑚33 ) ± √(𝐾11 𝑚33 − 𝐾33 𝑚11 )2 + 4 ( 𝑠 ) 𝐾𝑔13 2 𝑚33 𝑚11
Ω2 2
=
4 2𝑚33 𝑚11
where
∗
2λ𝑠 ± λ𝑡
𝐾33 = 𝐾33 − 𝐾𝑔33
2
144
It can be seen, four different equations given by Eq. (6-45) which represents four
different Ω values can be obtained, where the value of Ω2 is associated with vibration
modes, which were illustrated in the free vibration analysis shown in Section 6.4.1.
6.4.4. Comparison of the dynamic instability of simply supported beams due to

transverse periodic loading
Table (6-1) gives the dimensions and material properties of four various flange width,
(bf =100 mm, bf =150 mm, bf =200 mm, and bf =250 mm) of castellated beams
discussed herein. The analytical solution to determine the natural frequencies was
obtained directly from Eqs. (6-37), (6-38) and (6-39) while the critical loads were
obtained by Eq. (6-43).
Furthermore, Eq. (6-45) calculated the dynamic instability regions of castellated beam
for various span lengths (small, middle, large and very large) (see Table (6-1)) under a
transverse periodic load applied at the top flange of the beam. The results are plotted in
Figures 6-2, 6-3, 6-4, 6-5, 6-6, 6-7, 6-8 and 6-9 respectively.
Figures 6-2, 6-3, and 6-4 present the variation of the frequencies of lateral vibration,
vertical vibration and rotational vibration of the beams of different flange widths versus
the beam length. The three figures correspond to the 1st, 2nd and 3rd vibration modes.
From these figures it can be observed that, for each vibration mode, the frequency
curves have a similar variation pattern. In addition, the beam length and the flange
width influence the frequencies. Increasing the beam length causes reductions in the
frequencies. In contrast, the larger flange width gives the greater frequencies.
Furthermore, it can be seen from the figures that the frequency of the lateral vibration is
slightly higher than the frequency of the rotational vibration of the beam but a little
higher than the frequency of the vertical vibration of the castellated beam.
Figure 6-5 plots the critical load curves of the beams of different flange widths
subjected to the transverse static load applying on the top flange, where the critical
145
moment has been calculated using the yield moment, is obtained by

Myield = , ( 𝐼𝑟𝑒𝑑𝑢𝑐𝑒𝑑 is obtained by following Eq. (3-17)). It can be noticed,
ℎ𝑤 +2𝑡𝑓
as expected, that for each beam the increase of beam length causes the reduction of the
critical moment.
Figures 6-6, 6-7, 6-8, and 6-9 show the dynamic instability zones of the simply support
castellated beam with four different flange widths, subjected to a transverse periodic
load applied at the top flange of the beams, in which the geometric stiffness matrix is
assessed using the static critical load, that is(𝑞yield = 𝑞𝑐𝑟 ), where (𝑞yield = ).
𝑙2 (ℎ𝑤 +2𝑡𝑓 )
The four figures correspond to four different beam lengths as indicated in Table (6-1). It
can be observed from these figures that, the dynamic instability regions of the four
beams all exhibit a “v” shape in despite of having different flange widths. With the
increase of beam length, the dynamic instability zone not only moves towards to higher
frequency side but its width is also expanded. In contrast, where the beam length is the
same, the width of the dynamic instability zone decreases with the increase of the flange
width.
The figures of the variation of the frequencies, curve of the critical load and the
dynamic instability zones of castellated beams have the same patterns with those
computed in previous studies by Plaut (2017); Zhu et al. (2017); and Zhu et al. (2018).
However, a quantitative comparison is not mentioned because the beams are different
between this study (castellated beams) and previous studies (cold-formed steel beams).
Table 6-1 Dimensions and properties of four various flange widths (bf =100 mm, bf =150 mm,
bf =200 mm, and bf =250 mm) castellated beams*
tf hw tw a E ρ σy G
v
mm mm mm mm MPa kg/m3 MPa MPa
10 300 8 100 2.1x105 7800 275 78750 0.3
Note*: Dynamic instability analysis uses four different beam lengths. They are 4.156m;
6.235 m; 9.006 m and 14.549 m
146
Figure 6-1 (a) Notations is used in castellated beams. (b) Loading and displacements of web
and displacement of flanges when lateral–torsional buckling occurred (c) Section properties of
∗ ∗
middle-part of web in four different regions. 𝐼𝑦3 = 𝐼𝑦3 , 𝐼𝑧3 = 𝐼𝑧3 , 𝐽3 = 𝐽3∗ in region 2, in region
4,𝐼𝑦3 = 𝐼𝑧3 = 𝐽3 = 0 , section properties vary with x in regions 1 and 3.
Figure 6-2 Comparison of frequencies of simply support castellated beams with different flange
widths (1st mode)
147
widths (2nd mode)
widths (3rd mode)
148
Figure 6-5 Comparison of critical buckling moments of simply support castellated beam with
different flange widths
Figure 6-6 Comparison of dynamic instability regions of simply support castellated beam
(l = 4.156 m) (𝑞𝑦𝑖𝑒𝑙𝑑 = 𝑞𝑐𝑟 and 𝜆𝑠 =0)
149
150
6.5. Pinned–fixed doubly symmetric castellated beam subjected to periodic loads

on top flange
For the calculation due to the dynamic lateral-torsional buckling, the displacement
functions w(x), v(x), ϕ(x) and pre-buckling internal bending moment 𝑀𝑦 (𝑥) that satisfy
the boundary conditions of a pinned-fixed beam can be assumed as follows:
𝜋(𝑙 − 𝑥) 𝜋(𝑙 − 𝑥)
𝑣(𝑥) = 𝑞1 (𝑡)𝑠𝑖𝑛 𝑠𝑖𝑛 6-46
𝑙 2𝑙
𝜋(𝑙 − 𝑥) 𝜋(𝑙 − 𝑥)
𝑤(𝑥) = 𝑞2 (𝑡)𝑠𝑖𝑛 𝑠𝑖𝑛 6-47
𝑙 2𝑙
𝜋(𝑙 − 𝑥) 𝜋(𝑙 − 𝑥)
𝜙(𝑥) = 𝑞3 (𝑡)𝑠𝑖𝑛 𝑠𝑖𝑛 6-48
𝑙 2𝑙
𝑞𝑧 𝑥 3𝑙
𝑀𝑦 (𝑥) = ( − 𝑥) 6-49
2 4
where 𝑞𝑖 (𝑡) (𝑖 = 1,2,3) are the functions of time.
151
Therefore, the mass, stiffness, and geometric stiffness matrices for pinned-fixed beam
are obtained from Eqs. (6-19)- (6-21) and are expressed as follows:
𝑚11 0 0
[𝐌] = [ 0 𝑚22 0 ] 6-50
0 0 𝑚33
where:
𝜌𝑙(𝑎𝑡𝑤 +𝐴𝑡𝑒𝑒 ) 𝜌𝑙 𝐴𝑡𝑒𝑒 ℎ2

𝑚11 = 𝑚22 = , 𝑚33 = ( + 𝐼𝑝 )
2 2 4
𝜅11 0 0
[𝐊] = [ 0 𝜅22 0 ] 6-51
0 0 𝜅33
where:
41𝐸𝑙(2𝐼𝑧1 + 𝑘𝐼𝑧3 ) 𝜋 4
𝜅11 = ( )
64 𝑙
41𝐸𝑙(2𝐼𝑦1 + 𝑘𝐼𝑦3 ) 𝜋 4
𝜅22 = ( )
64 𝑙
41𝐸𝐼𝑤 𝑙 𝜋 4 5𝐺𝑙(2𝐽1 + 𝑘𝐽3 ) 𝜋 2
𝜅33 = ( ) + ( )
64 𝑙 16 𝑙
0 0 𝜅𝑔13
[𝐊 𝒈 ] = [ 0 0 0 ] 6-52
𝜅𝑔31 0 𝜅𝑔33
where:
𝑞𝑧 𝑙 𝜋 2 1 𝑎𝑧 𝑞𝑧 𝑙
𝜅𝑔13 = 𝜅𝑔31 = − ( 3 + 1 8), 𝜅𝑔33 = −
8 4
6.5.1. The free vibration analysis
The free vibration analysis frequency of the lateral-torsional vibration of the castellated
beam can be determined using Eq. (6-53):
‖[𝐊] − 𝜔2 [𝐌]‖ = 0 6-53
152
where 𝝎 is the free vibration frequency. Substituting Eqs. (6-50) and (6-51) into (6-53),
also, the following frequency can be obtained:
𝜋 2 41𝐸(2𝐼𝑧1 + 𝑘𝐼𝑧3 )
𝜔1 = ( ) √ 6-54
𝜋 2 41𝐸(2𝐼𝑦1 + 𝑘𝐼𝑦3 )
𝜔2 = ( ) √ 6-55
𝐺𝑙2
𝜋 2 41𝐸𝐼𝑤 + 20 (2𝐽1 + 𝑘𝐽3 )
𝜋2 6-56
𝜔3 = ( ) √
𝑙 8𝜌(𝐴𝑡𝑒𝑒 ℎ2 + 2𝐼𝑝 )
∗ ∗
𝐼𝑧3 can be negligible because in most of castellated beams 𝐼𝑧3 << 2𝐼𝑧1 , then Eqs. (6-54),
(6-55) and (6-56) can be simplified as follows:
𝜋 2 41𝐸(2𝐼𝑧1 )
𝜔1 = ( ) √ 6-57
𝜋 2 41𝐸(2𝐼𝑦1 )
𝜔2 = ( ) √ 6-58
𝐺𝑙2
𝜋 2 41𝐸𝐼𝑤 + 20 (2𝐽1 + 𝑘𝐽3 )
𝜋2 6-59
𝜔3 = ( ) √
𝑙 8𝜌(𝐴𝑡𝑒𝑒 ℎ2 + 2𝐼𝑝 )
The above formulations (6-57), (6-58) and (6-59) give the natural frequencies, which
are well known and can be found from many vibration textbooks. These equations
represent the translational and rotational vibrations of castellated beams. Moreover, it
indicates that the lateral vibration and torsional vibration modes are influenced by web
openings.
153
6.5.2. Buckling analysis
The critical load of the lateral-torsional buckling of the castellated beam subjected to a
static load can be calculated using Eq. (6-60):
‖[𝐊] − λ𝑐𝑟 [𝐊 𝒈 ]‖ = 0 6-60
where λcr is the loading factor and 𝑞𝑐𝑟 = 𝜆𝑐𝑟 𝑞𝑜 is the critical moment for static
buckling.
Substituting Eqs. (6-51) and (6-52) into (6-60), the following critical load is obtained:
𝐾11 𝐾𝑔33 ± √𝐾11 2 𝐾𝑔33 2 + 4𝐾𝑔13 2 𝐾11 𝐾33

λ𝑐𝑟 = −
6-61
2
2𝐾𝑔13
𝟐 𝟐
𝒉𝒘 𝒉 𝟓𝝅𝟐 + 𝟏𝟎𝟖 𝟐𝟎𝑮(𝟐𝑱𝟏 + 𝒌𝑱𝟑 ) 𝒍 𝟐 𝟏
(( + 𝒕𝒇 ) + √( 𝒘 + 𝒕𝒇 ) + ( ) (𝑰𝒘 + ( ) )× )
𝟐 𝟐 𝟏𝟗𝟐 𝟒𝟏𝑬 𝝅 (𝟐𝑰𝒛𝟏 + 𝒌𝑰𝒛𝟑 )
𝒒𝒛 𝒍𝟐
( ) =
6-62
𝟐
𝟖 𝒄𝒓 𝟓𝝅𝟐 + 𝟏𝟎𝟖 𝟐
( ) 𝒍
𝟗𝟔
𝟒𝟏𝑬(𝟐𝑰𝒛𝟏 + 𝒌𝑰𝒛𝟑 )
×
𝟑𝟐
𝟐 𝟐
𝒉𝒘 𝒉 𝟓𝝅𝟐 + 𝟏𝟎𝟖 𝟐𝟎𝑮(𝟐𝑱𝟏 + 𝒌𝑱𝟑 ) 𝒍 𝟐 𝟏
(( + 𝒕𝒇 ) + √( 𝒘 + 𝒕𝒇 ) + ( ) (𝑰𝒘 + ( ) )× )
𝟐 𝟐 𝟏𝟗𝟐 𝟒𝟏𝑬 𝝅 (𝟐𝑰𝒛𝟏 )
𝒒 𝒛 𝒍𝟐
( ) =
𝟖 𝒄𝒓
(
𝟓𝝅𝟐 + 𝟏𝟎𝟖 𝟐
𝟐
) 𝒍 6-63
𝟗𝟔
𝟒𝟏𝑬(𝟐𝑰𝒛𝟏 )
×
𝟑𝟐
It can be noticed that Eq. (6-63) is similar to the formulation of the critical load given in
this study Section 4.6.2 for pinned-fixed castellated beams when the load is applied at
the top flange.
6.5.3. The dynamic instability
The dynamic instability region of the castellated beam can be calculated using
Eq. (6-64):
2λs ± λt Ω2
‖[K] − [K g ] − [M]‖ = 0 6-64
2 4
154
Substituting Eqs. (6-50), (6-51) and (6-52) into (6-64), it yields:
2λ ± λ𝑡 2 6-65
∗
(𝐾33 ∗
𝑚11 + 𝐾11 𝑚33 ) ± √(𝐾11 𝑚33 − 𝐾33 𝑚11 )2 + 4 ( 𝑠 ) 𝐾𝑔13 2 𝑚33 𝑚11
Ω2 2
=
4 2𝑚33 𝑚11
where
∗
2λ𝑠 ± λ𝑡
𝐾33 = 𝐾33 − 𝐾𝑔33
2
It can be seen that, there are four different equations given by Eq. (6-65), which
represent four different Ω values, where the value of Ω2 is associated with vibration
modes, which were illustrated in the free vibration analysis shown in Section 6.5.1
6.5.4. Comparison of the dynamic instability of pinned-fixed beams due to

transverse periodic loading
Table (6-1) gives the dimensions and material properties of four various flange width,
(bf =100 mm, bf =150 mm, bf =200 mm, and bf =250 mm) of castellated beams
discussed herein. The analytical solution to determine the natural frequencies was
obtained directly from Eqs. (6-57), (6-58) and (6-59), while the critical loads were
obtained by Eq. (6-63).
Furthermore, Eq. (6-65) calculated the dynamic instability regions of castellated beam
for various span lengths (small, middle, large and very large) (see Table (6-1) under a
transverse periodic load applied at the top flange of the beam. The results are plotted in
Figures 6-10, 6-11, 6-12, 6-13, 6-14, 6-15, 6-16 and 6-17 respectively.
Figures 6-10, 6-11, and 6-12 present the variation of the frequencies of lateral
vibration, vertical vibration and rotational vibration of the beams of different flange
widths versus the beam length of the pinned-fixed castellated beam. The three figures
correspond to the 1st, 2nd and 3rd vibration modes.
155
From these figures it can be observed that, for each vibration mode, the frequency
curves have a similar variation pattern. In addition, the beam length and the flange
width influence the frequencies. Increasing the beam length causes reductions in the
frequencies. In contrast, the larger flange width gives the greater frequencies.
Furthermore, it can be seen from the figures that the frequency of the lateral vibration is
slightly higher than the frequency of the rotational vibration of the beam but a little
higher than the frequency of the vertical vibration of the castellated beam.
Figures 6-13 plots the critical loads curves of the pinned-fixed castellated beam
subjected to the transverse periodic load applying on top flange versus the beam length
for different flange widths, where the critical moment has been normalized using the
yield moment ,Myield = , ( 𝐼𝑟𝑒𝑑𝑢𝑐𝑒𝑑 is obtained by following Eq. (3-17)). It
ℎ𝑤 +2𝑡𝑓
can be noticed, as expected, that for each beam the increase of beam length causes the
reduction of the critical moment.
Figures 6-14, 6-15, 6-16 and 6-17 show the dynamic instability zones of the pinned-
fixed castellated beam with four different flange widths, subjected to a transverse
periodic load applied at the top flange of the beams, in which the geometric stiffness
matrix is assessed using the static critical load that is (𝑞yield = 𝑞𝑐𝑟 ) where
𝜎𝑦 𝐼
(𝑞yield = 16 𝑙2 (ℎ𝑟𝑒𝑑𝑢𝑐𝑒𝑑 ) . The four figures correspond to four different beam lengths as
+2𝑡 )
𝑤 𝑓
indicated in Table (6-1) It can be observed from these figures that, the dynamic
instability regions of the four beams all exhibit a “v” shape in despite of having diﬀerent
flange widths. With the increase of beam length, the dynamic instability zone not only
moves towards to higher frequency side but its width is also expanded. In contrast,
where the beam length is the same, the width of the dynamic instability zone decreases
with the increase the flange width.
The figures of the variation of the frequencies, curve of the critical load and the
dynamic instability zones of castellated beams have the same patterns with those
computed in previous studies by Plaut (2017); Zhu et al. (2017); and Zhu et al. (2018).
However, I did not make a quantitative comparison. This is because the beams are
different between ours (castellated beams) and others (cold-formed steel beams).
156
Figure 6-10 Comparison of frequencies of pinned-fixed castellated beams with different flange
widths (1st mode)
widths (2nd mode)
157
widths (3rd mode)
Figure 6-13 Comparison of critical buckling moments of pinned-fixed castellated beam with
different flange widths
158
Figure 6-14 Comparison of dynamic instability regions of pinned-fixed castellated beam


(l =6.235 m) (𝑞𝑦𝑖𝑒𝑙𝑑 = 𝑞𝑐𝑟 and 𝜆𝑠 =0)
159


(l =14.549 m) (𝑞𝑦𝑖𝑒𝑙𝑑 = 𝑞𝑐𝑟 and 𝜆𝑠 =0)
160
6.6. Conclusions
This chapter has presented an analytical study on the dynamic instability of castellated
beams subjected to transverse periodic loading at top flange. The dynamic instability
analysis employed in the present study uses Bolotin’s method, while the mass, stiffness,
and geometric stiffness matrices are derived using the kinetic energy, the strain energy
and the potential of applied loads, which are used for conducting the analytical of the
frequency of free vibration, the critical load of lateral torsional buckling, and the
excitation frequency of dynamic instability region.
From the obtained results the following conclusions can be drawn:
 The free vibration, static buckling and dynamic instability analyses of castellated
beams subjected to transverse periodic loading at the top flange are influenced by
the coupling between the translational and rotational modes.
 Increasing the flange width of beam leads to increase both of the frequency and
critical buckling moment. However, increasing beam length reduces this effect.
 The dynamic instability zone of the castellated beam will move towards to high-
frequency side and the corresponding width of the dynamic instability zone
decreases when its flanges become wide.
 The effect of lateral-torsional buckling on the dynamic instability zone becomes

more significant in the short beam than in the long beam, and also in the wide
flange beam than in the narrow flange beam.
161
CHAPTER SEVEN CONCLUSIONS AND FUTURE STUDIES
CHAPTER SEVEN
7. CONCLUSIONS AND FUTURE STUDIES
7.1. Conclusions
Despite the widespread use of castellated beams in the structural field, the current
design specifications of steel structure do not provide design provisions for the beam.
The designing procedure of castellated beams needs to consider three different issues.
The first issue is the strength, the second is the deflection and the last one is the
lateral torsional buckling. In the castellated beam, these issues are affected by web
openings.
The aims of this thesis were to investigate the effect of web openings on the
transverse deflection and lateral-torsional buckling of castellated beams and to focus
on the effect of both the geometric nonlinear and material inelasticity on castellated
beams under uniformly distributed load with different boundary conditions. In
addition, the free vibration, static buckling and dynamic instability of castellated
beams subjected to transverse periodic loading has been also discussed by
developing analytical solutions.
Numerous researches have been conducted to investigate the castellated beams

where various methods have been adopted such as experimental, analytical, and
numerical methods to predict the calculations’ design of the beams. However, these
methods differ in terms of safety and accuracy. Some of them need efficiency in
using, and are furthermore very expensive.
Both analytical and numerical methods were used to achieve the aims of this thesis.
The purpose of developing analytical methods is for the design and practical use,
162
while the numerical methods developed are for the validation of the analytical
methods. The analytical solution is developed using the classical principle of
minimum potential energy, whereas the numerical solutions are obtained using the
commercial software ANSYS. Bolotin’s method is used to perform the dynamic
instability analysis.
The conclusions drawn from the analytical and numerical investigations carried out
in this research can be summarized as follows:
 The present analytical results are in excellent agreement with those obtained
from the finite element analysis, which demonstrates the appropriateness of
proposed approach.
 The shear-induced deflection is proportional to the cross-section area of the two

T-sections but inversely proportional to castellated beam length.
 The web shear effect on the deflection of castellated beams is very important,
particularly for short and medium length beams. In contrast, increasing the beam
length reduces the web shear effect on the deflection.
 Non-uniform material properties caused by the non-uniform temperature affect

the maximum deflection of the castellated beam when the shear effect is
considered.
 For the same average temperatures, the maximum deflection of the castellated
beam under non-uniform temperature distribution with transverse distributed
load is directly proportional to the amount of difference in temperatures between
the two T- sections of the beam.
 For lateral-torsional buckling of the castellated beam, one can ignore the lateral
flexural rigidity of the web openings and use the average torsional constant of
the full and reduced section properties in the calculation of the critical moment
of lateral–torsional buckling of castellated beams.
163
 For linear behaviour, the lateral-torsional buckling results’ curves of the

analytical solution and numerical analysis have a similar variation pattern.
 The failure mode of short castellated beams is dominated by the plastic failure,
whereas the failure mode of long castellated beams is dominated by the lateral-
torsional buckling failure mode.
 The non-linear behaviour showed the critical load of lateral-torsional buckling

of castellated beams is influenced by the beam size, web openings, boundary
conditions, and material properties of the beam.
 The critical loads of flexural and lateral-torsional buckling of castellated beams

obtained by using nonlinear 3D finite element analysis are generally less than
those obtained from the linear analyses, particularly for beams with short lengths
and wide flanges. This indicates that the elastic range is unsafe for short length
beam with wide flanges.
 The lateral-torsional buckling resistance of castellated beams with short length

and/or wide flanges is limited by the ultimate load carrying capacity, in which
no lateral-torsional buckling occurs.
 When the serviceability is also considered, the deflection limit seems to be the
dominant criterion in controlling the load in most of the beam length regions.
 The longer the beam, the closer the critical load obtained from the linear lateral-
torsional buckling analysis to the failure load obtained from of the full nonlinear
analysis.
 The longer the beam, the less importance of the nonlinearity need to be
considered.
 The free vibration, static buckling and dynamic instability analyses of

castellated beams subjected to transverse periodic loading at the top flange are
influenced by the coupling between the translational and rotational modes.
164
 The frequency and critical buckling moment load are directly proportional to the
flange width of castellated beams. However, increasing length of the beam
reduces this effect.
 The dynamic instability zone of the castellated beam will move towards to high-
frequency side and the corresponding width of the dynamic instability zone
decreases when its flanges become wide.
 The effect of lateral-torsional buckling on the dynamic instability zone becomes

more significant in the short beam than in the long beam, and also in the wide
flange beam than in the narrow flange beam.
7.2. Future studies
The main conclusions that have been discussed above have presented the primary
aims of this thesis, which clearly confirm the need for additional researches, which
are recommended as follows:
 The analytical method which was developed in this research to calculate the
maximum deflections of castellated beams in three different fire scenarios with a
uniformly distributed load, needs to enhance the understanding of the fire
performance of castellated beams by carrying out an experimental investigation,
also to validate the results.
 Fire safety is an important aspect that should be considered in the design of the
structure because it refers to the protection of the building from the risk of fire.
Therefore, further research should focus on the method of calculation of the fire
resistance of the castellated beam which is influenced by the cross-sectional
geometry, depth and the material of the beam. Additionally, conduct
investigations to identify the most suitable fire protection materials insulate of
the castellated beam due to the effects of the high temperatures.
 In this thesis the material model of the castellated beam is limited to grade
275 MPa, other materials should be studied.
165
 Dynamic instability regions of a castellated beam subjected to non-uniform

periodic bending moment with damping can be investigated in future.
 The effect of live load on floor vibration may also can be studied.
166
APPENDIX A
APPENDIX A
A.1. Web combined stiffness
As for material model castellated beam of steel grade S275, this is assumed to be
constructed of linear elastic material with Young’s modulus E = 2.1 × 105 MPa,
Poisson’s ratio v =0.3 and yield stress σy=275 MPa.
3D linear shell finite element (SHELL181) is adopted to calculate shear rigidity factor
(Ksh), which was formed at Eq. (3-27).Table (A-1) presents the value of the combined
stiffness of the mid part of the web of the castellated beam caused by the bending and
shear for groups A,B,C,D,E,F,G,H,I and J (see Section 3.3.4) are obtained by Eq. (3-28)
is tabulated under term (Kb1); based on: (Yuan et al, 2016) is tabulated under term (Kb2);
(Yuan et al, 2014) is tabulated under term (Kb3).
167
APPENDIX A
Table A-1.Web combined stiffness of castellated beams considered
Name of bf Kb1 Kb2 Kb3

Group
beam mm N/mm N/mm N/mm
A1 100 172257.90 218238.40 279792.80
A2 150 152065.62 218238.40 279792.80
A
A3 200 131873.32 218238.40 279792.80
A4 250 111681.01 218238.40 279792.80
B1 100 199181.01 218238.40 279792.80
B2 150 192450.24 218238.40 279792.80

B
B3 200 185719.47 218238.40 279792.80
B4 250 178988.70 218238.40 279792.80

C1 100 204565.62 218238.40 279792.80
C2 150 200527.16 218238.40 279792.80

C
C3 200 196488.70 218238.40 279792.80
C4 250 192450.24 218238.40 279792.80
D1 100 205911.78 218238.40 279792.80

D2 150 202546.39 218238.40 279792.80
D
D3 200 199181.01 218238.40 279792.80
D4 250 195815.62 218238.40 279792.80
A1 100 172257.90 218238.40 279792.80
A2 150 152065.62 218238.40 279792.80

E
A3 200 131873.32 218238.40 279792.80
A4 250 111681.01 218238.40 279792.80

F1 100 207594.47 218238.40 279792.80
F2 150 205070.43 218238.40 279792.80

F
F3 200 202546.39 218238.40 279792.80
F4 250 200022.35 218238.40 279792.80
G1 100 208155.37 218238.40 279792.80

G2 150 205911.78 218238.40 279792.80
G
G3 200 203668.19 218238.40 279792.80
G4 250 201424.60 218238.40 279792.80
168
APPENDIX A
Name of bf Kb1 Kb2 Kb3

Group
beam mm N/mm N/mm N/mm
H1 100 209536.04 218238.40 279792.80
H2 150 207982.78 218238.40 279792.80

H
H3 200 206429.53 218238.40 279792.80
H4 250 204876.27 218238.40 279792.80
I1 100 210398.96 218238.40 279792.80
I2 150 209277.16 218238.40 279792.80

I
I3 200 208155.37 218238.40 279792.80
I4 250 207033.57 218238.40 279792.80
J1 100 210719.45 218238.40 279792.80

J2 150 209757.93 218238.40 279792.80
J
J3 200 208796.39 218238.40 279792.80
J4 250 207834.985 218238.40 279792.80
169
BIBLIOGRAPHY
BIBLIOGRAPHY
ABNT NBR 8800:2008 (2008)."Design of Steel Structures and Mixed Structures of

Steel and Concrete Buildings." Brazilian Association of Technical
Standards, Rio de Janeiro.
Abreu, L.M.P., Fakury, R.H. and Castro e Silva, A.L.R. (2010). "Determination of
Bending Moment Resistance to Lateral Buckling with Torsion of Cellular
Steel Beams." Associacao Argentina de Mecanica Computacional, Buenos
Aires, XXIX: 7255-7271.
Aglan, A.A. and Redwood, R.G. (1974). "Web buckling in castellated beams."
Proceedings of the Institution of Civil Engineers Part 2, Research and
Theory,( 57): 307-320.
Ahnlén, M. and Westlund, J. (2013). "Lateral Torsional Buckling of I-Beams. A

Parametric Study of Elastic Critical Moments in Structural Design
Software". Master of Science Thesis in the Master’s Programme
Structural Engineering and Building Technology. Department of Civil
and Environmental Engineering, Division of Structural Engineering,
Steel and Timber Structures, Chalmers University of Technology,
Göteborg, Sweden.
Altifillisch, M. D., Cooke, R. B. and Toprac, A. A. (1957). "An investigation of open

web expanded beams." Welding Research Council Bulletin, New York,
47: 307-320.
American Institute of Steel Construction (AISC) (2011). "ANSI/AISC 360–10

specification for structural buildings14th ed.", Chicago, Illinois, USA.
170
BIBLIOGRAPHY
Aminian, P., Niroomand, H., Gandomi, A. H., Alavi, A. H. and Arab Esmaeili, M.
(2013). "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.
ANSYS User’s Manual. (2017). "ANSYS Mechanical APDL Technology

Demonstration Guide." Release 18.1, Canonsburg, USA: ANSYS Inc.
Anupriya, B. and Jagadeesan, D. K. (2014). "Shear strength of castellated beam with

and without stiffeners using FEA (Ansys 14)." International Journal of
Engineering and Technology (IJET), 6(4): 1970-1981.
ArcelorMittal. (2008). ArcelorMittal Europe: Long products sections and merchant

bars ACB. Cellular Beams.
Badke-Neto, A., Calenzani, A.F. G. and Ferreira, W. G. (2015). "Study of methods

for the design of cellular composite steel and concrete beams." Ibracon
structures and material Journal, 8(6): 827-859.
Bake, S. M. (2010)."Cellular beams at ambient and elevated temperatures.” PhD.

dissertation, The University of Manchester, Manchester, UK.
Bolotin, V.V. (1964). "The dynamic stability of elastic systems." San Francisco, CA:
Holden-day, Inc.
Boyer, J. P. (1964). "Castellated Beam- A New Development Castellated beams-new

developments." AISC Engineering, 1(3): 104-108.
Bradley, T. P. (2003). "Stability of Castellated Beams During Erection." Master’s

Thesis, Thesis submitted to the Faculty of the Virginia Polytechnic
Institute and State University, Blacksburg.
British Standards Institution (BSI). (2000). ".BS5950-1 structural use of steelwork in

buildings-Part 1, code of practice for design rolled and welded sections."
,England.
Chang, C-S. and Hodges, D.H. (2009). "Stability studies for curved beams." Journal
of Mechanics of Materials and Structures, Journal of Mechanical
Material and Structures, 4(7, 8):1257-70.
171
BIBLIOGRAPHY
Chen, L. Y., Lin, P. D. and Chen, L. W. (1991). "Dynamic stability of thick

bimodulus beams)." Computers & Structures, 41(2): 257-263.
Chung, K., Liu, T. and Ko, A. (2001). "Investigation on Vierendeel mechanism in

steel beams with circular web openings." Journal of Constructional Steel
Research, 57(5a): 467-490.
Crisfield, M.A. (1981). "A fast incremental/iterative solution procedure that handles
snap-through." Computer & Structures, 13:55-62.
Demirdjian, S. (1999). "Stability of castellated beam webs." PhD, McGill University

Montreal, Canada.
Durif, S., Bouchaïr, A. and Vassart, O. (2013). "Experimental tests and numerical
modeling of cellular beams with sinusoidal openings." Journal of
Constructional Steel Research, 82: 72-87.
Durif, S., Vassart, O., Bouchaïr, A. and Muzeau, J. P. (2011). "Modèle mécanique
pour les poutres à larges ouvertures d'âmes de section variable." In
XXIXe Rencontres Universitaires de Génie Civil, AUGC, Tlemcen,
1(29): 345-355.
El-Sawy, K. M., Sweedan, A. M. and Martini, M. I. (2009). "Major-axis elastic

buckling of axially loaded castellated steel columns." Thin-Walled
Structures, 47(11): 1295-1304.
Ellobody, E. (2011). "Interaction of buckling modes in castellated steel beams."

Journal of Constructional Steel Research, 67(5): 814-825.
Ellobody, E. and Young, B. (2015). "Nonlinear analysis of composite castellated

beams with profiled steel sheeting exposed to different fire conditions."
Journal of Constructional Steel Research, 113: 247-260.
Erdal, F. and Saka, M. P. (2013). "Ultimate load carrying capacity of optimally

designed steel cellular beams." Journal of Constructional Steel Research,
80: 355-368.
172
BIBLIOGRAPHY
Erdal, F., Tunca, O. and Tas, S. (2015). "Nonlinear finite element analysis of
optimally designed steel cellular beams." Research on Engineering
Structures and Materials 2(2): 59-66.
European Committee for Standardization EN1993-1-1 (2005). “Eurocode 3. Design

of steel structures, Part1.1: General rules and rules for building."
Forde, B. D. R. and Stiemer, S. F. (1987) "Improved Arc Length Orthogonality

Methods for Nonlinear Finite Element Analysis." Computers and &
Structures, 27(5): 625-630.
Gandomi, A. H., Tabatabaei, S. M., Moradian, M. H., Radfar, A. and Alavi, A. H.

(2011). "A new prediction model for the load capacity of castellated steel
beams." Journal of Constructional Steel Research, 67(7): 1096-1105.
Gao, K., Gao, W., Wu, BH. and Song, CM. (2019). "Nondeterministic dynamic
stability assessment of Euler–Bernoulli beams using Chebyshev
surrogate model." Applied Mathematical Modelling, 66, 1-25.
Gholizadeh, S., Pirmoz ,A. and Attarnejad, R. (2011). "Assessment of load carrying
capacity of castellated steel beams by neural networks." Journal of
Constructional Steel Research, 67(5): 770-779.
Gürgöze, M. (1985). "On the dynamic stability of a pre-twisted beam subject to a

pulsating axial load." Journal of Sound and Vibration, 102(3): 415-422
(1985).
Halleux, P. (1967). "Limit Analysis of Castellated Beams." Acier-Stahl-Steel, 3:

133-144.
Harper, C. (1991). "Design in steel 4: castellated & cellular beams." British Steel.
Hosain, M., Cheng, W. and Neis, V. (1974). "Deflection analysis of expanded open-
web steel beams." Computers & Structures, 4(2): 327-336.
Hosain, M. and Spiers, W. (1971). "Failure of Castellated Beams due to Rupture of

Welded Joints." Acier-Stahl-Steel, 36(1):34-40
173
BIBLIOGRAPHY
Hosain, M. and Spiers, W. (1973). "Experiments on castellated steel beams."

American Welding Society, Welding Research Supplement, 52(8): 329S-
342.
Hsu, C.S. (1966). "On dynamic stability of elastic bodies with prescribed initial
conditions." International Journal of Engineering Science, 4 (1):1-21.
Huang, C. C. (1980). "Dynamic stability of generally orthotropic beams." Fibre

Science and Technology, 13(3): 187-198.
Huang, J.S. and Hung, L.H. (1984). "Dynamic stability for a simply supported beam
under periodic axial excitation." International Journal of Non-Linear
Mechanical, 19(4):287-301.
Jovic, M. (2015). "Lateral torsional buckling analysis of multiple laterally restrained

I-beams in bending." Master thesis A-2015.102. Eindhoven University of
Technology Department of the Built Environment Structural Design
Kar, R. and Sujata, T. (1991). "Dynamic stability of a rotating beam with various
boundary conditions." Computers & Structures, 40(3): 753-773.
Kerdal, D. (1982). "Lateral-torsional buckling strength of castellated beams." PhD

thesis, Civil and Structural Engineering, Faculty of Engineering
(Sheffield), University of Sheffield, Sheffield, UK.
Kerdal, D. and Nethercot, D. (1984). "Failure modes for castellated beams." Journal
of Constructional Steel Research, 4(4): 295-315.
Kim, B., Li, L. and Edmonds, A. (2016). "Analytical Solutions of Lateral–Torsional

Buckling of Castellated Beams." International Journal of Structural
Stability and Dynamics, 16(8): 1-16 (1550044).
Knowles, P. R. (1991). "Castellated beams." Proceeding of the Institution of Civil

Engineers, ICE, London, UK 90(Part 1): 521-536.
Kohnehpooshi, O. and Showkati, H. (2009). "Numerical Modeling and Structural

Behavior of Elastic Castellated Section", European Journal of Scientific
Research, 31(2):306–18.
174
BIBLIOGRAPHY
Korrani, H. K., Kabir, M. and Molanaei, S. (2010). "Lateral-Torsional Buckling of

Castellated Beams under End Moments." International Journal. of Recent
Trends in Engineering and Technology, 3(5): 16-19.
Kratzig, WB, Li L. and Nawrotzki, P. (1991). "Stability conditions for non-

conservative dynamical systems." Computational Mechanics, 8(3):145-
51.
Kwani,S. and Wijaya, P. K. (2017). "Lateral torsional buckling of castellated beams

analyzed using the collapse analysis." Procardia Engineering, 171: 813-
820.
Lawson, R. M. and Hicks, S. J. (2011). "Design of composite beams with large web
openings." (SCI P355), The Steel Construction Institute Publication.
Lei, J. S., Yuan, W. B. and Li, L. Y. (2017). "Axial compression buckling of

castellated columns at elevated temperatures." International Journal of
Structural Stability and Dynamics, 17(3): 1750034
Li, L. Y. (1991). "Interaction of forced and parametric loading vibrations."

Computers & Structures, 40(3):615-618.
Li, L. Y., Ren, C. and Yang, J. (2012). "Theoretical analysis of partially restrained
zed-purlin beams subjected to up-lift loads." Journal of Constructional
Steel Research, 70: 273-279.
Liu, T. and Chung, K. (2003). "Steel beams with large web openings of various
shapes and sizes: finite element investigation." Journal of Constructional
Steel Research, 59(9): 1159-1176.
Maalek, S. (2004). "Shear deflections of tapered Timoshenko beams." International

Journal of Mechanical Sciences, 46(5): 783-805.
Martins, C.H., Fer-reira, F.P.V., Rossi, A. and Trentini, E.V.W. (2017)." Numerical
Analysis of Physical and Geometrical Imperfections in Cellular Beams."
Open Journal of Civil Engineering, 7: 116-129.
Menkulasi, F., Moen, C., Eatherton, M. and Kuruppuarachchi, D. (2015).

"Investigation of Stiffener Requirements in Castellated Beams."
175
BIBLIOGRAPHY
Structural Stability Research Council Annual Stability Conference

SSRC.
Mohebkhah, A. (2004). "The moment-gradient factor in lateral-torsional buckling on

inelastic castellated beams." Journal of Constructional Steel Research,
60(10): 1481-1494.
Mohebkah, A. (2011), "Lateral buckling resistance of inelastic I-beams under off-

shear center loading." Thin-Walled Structures, 49: 431-436.
Mohebkhah, A. and Showkati, H. (2005). "Bracing requirements for inelastic

castellated beams." Journal of Constructional Steel Research, 61(10):
1373-1386.
Morkhade, S. G. and Gupta, L.M. (2015). "An experimental and parametric study on
steel beams with web openings." International journal of advanced
structural engineering, Springer, 7(3):249-260.
Morris, N. F (1965). "The dynamic stability of beam-columns with a fixed distance

between supports." Journal of the Franklin Institute, 280(2): 163-173.
Müller, C., Hechler, O., Bureau , A., Bitar, D., Joyeux, D., Cajot , L-G., Demarco,
T., Lawson, M., Hicks, S., Devine, P., Lagerqvist, O., Hedman-
Pétursson, E., Unosson, E. and Feldmann, M. (2006). "Large web
openings for service integration in composite floors." Design Guide
(en), Technical steel report.
Nethercot D.A and Kerdal, D. (1982). "Lateral-torsional Buckling of Castellated

Beams." The Structural Engineer, 60B (3):53-61.
New steel construction technical report. (2006). "Lateral torsional buckling and
slenderness." NSC, October: 30-34.
Nseir, J., Lo, M., Sonck, D., Somja, H., Vassart, O. and Boissonnade, N. (2012).
"Lateral torsional buckling of cellular steel beams." Proceedings of the
Annual Stability Conference, Structural Stability Research Council,
Grapevine, Texas, April 18-21.
176
BIBLIOGRAPHY
Okubo, T. and Nethercot, D. (1985). "Web post strength in castellated steel beams."
Proceedings of the Institution of Civil Engineers, Part 2: Research and
Theory, 79: 533-557.
Pachpor, P.D., Gupta, L.M. and Deshpande, N.V. (2014). "Analysis and design of
cellular beam and its verification", International Conference on Applied
Computing, Computer Science and Computer Engineering, IERI
Procardia, 7: 120-127.
Panedpojaman, P. (2015). "Investigation on lateral torsional buckling resistance of

ec3 for cellular beam." International Journal of Advances in Mechanical
and Civil Engineering, 2(4) ISSN: 389-399.
Panedpojaman, P. and Rongram, T. (2014). "Design equations for Vierendeel

bending of steel beams with circular web openings." Proceedings of the
World Congress on Engineering (WCE 2014), vol. II, July 2-4, London,
UK.
Park, Y.P. (1987). "Dynamic stability of a free Timoshenko beam under a controlled
follower force." Journal of Sound and Vibration, 113(3):407-15.
Patel, S.N, Datta, P.K. and Sheikh, A.H. (2006). "Buckling and Dynamic Instability
Analysis of Stiffened Shell Panels." Thin-Walled Struct, 44: 321-333.
Pattanayak, U. and Chesson, E. (1974). "Lateral Instability of Castellated Beams."

AISC Engineering Journal, 11(3): 73-79.
Pourbehi, P. and Pirmoz, A. (2015). "Shear response of castellated steel beams."

International Journal of Steel Structures, 15(2): 389-399.
Radić., I. and Markulak, D. (2007). "Lateral buckling of castellated beams. "

Technical Gazette, 14(1, 2): 25-35.
Redwood, R. and Demirdjian, S. (1998). "Castellated Beam Web Buckling in

Shear." Journal of Structural Engineering (ASCE), 124(10): 1202-1207.
Sehwail, M.M. (2013). "Lateral Torsional Buckling of steel I-section Cellular

Beams." Gazimagusa, Estern Mediterranean University.
177
BIBLIOGRAPHY
Sherbourne, A. and Van Oostrom, J. (1972). "Plastic analysis of castellated beams—

I interaction of moment, shear and axial force." Computers & Structures,
2(1): 79-109.
Showkati, H (2008). "Lateral-Torsional Buckling of Castellated Beams." Iranian

Journal of Science & Technology, Transaction B, Engineering, 32:153-
156.
Showkati, H., Ghazijahani, T. G., Noori, A. and Zirakian, T. (2012). "Experiments

on elastically braced castellated beams." Journal of Constructional Steel
Research, 77: 163-172.
Soltani, M. R., Bouchaïr, A. and Mimoune, M. (2012). "Nonlinear FE analysis of the

ultimate behavior of steel castellated beams." Journal of Constructional
Steel Research, 70: 101-114.
Sonck, D. (2014). "Global buckling of castellated and cellular steel beams and
columns." Ph.D. dissertation, Ghent Univ., Ghent, Belgium.
Sonck, D. and Belis, J. (2016). "Lateral-Torsional Buckling Resistance of Castellated

Beams." Journal of Constructional Steel Research, 105: 119-128.
Sonck, D., Kinget, L. and Belis, J. (2015). "Deflections of cellular and castellated
beams." Future Visions (International Association for Shell and Spatial
Structures), Proceedings (pp. 1–12). Presented at the Future Visions
(International Association for Shell and Spatial Structures) (IASS2015),
Amsterdam, The Netherlands: Koninklijk Instituut van Ingenieurs,
Nederland.
Sonck, D., Van Impe, R. and Belis, J. (2014). "Experimental investigation of residual
stresses in steel cellular and castellated members." Journal of
Construction and Building Materials, 54: 512-519.
Srimani, S. S. and Das, P. (1978). "Finite element analysis of castellated beams."

Computers & structures, 9(2): 169-174.
Standards Australia (SA) (1998). "AS 4100 Steel Structures."Sydney, Australia.
178
BIBLIOGRAPHY
Sweedan, A. (2011). “Elastic lateral stability of I-shaped cellular steel beams."

Journal of Constructional Steel Research, 67 (2): 151-163.
Timoshenko, S. P. and Gere, J. M. (1961). "Theory of elastic stability."

McGrawHill-Kogakusha Ltd, Tokyo.
Toprac, A. and Cooke, B. (1959). "The plastic behavior of castellated beams."

Welding Research Council Bulletin, New York, 47: 1-10.
Tsavdaridis, K. D. and D'Mello,C. (2011). "Web buckling study of the behaviour and
strength of perforated steel beams with different novel web opening
shapes." Journal of Constructional Steel Research, 67(10): 1605-1620.
Tsavdaridis, K. D. and D’Mello, C. (2012). "Vierendeel Bending Study of Perforated

Steel Beams with Various Novel Web Opening Shapes through
Nonlinear Finite-Element Analyses." Journal of Structural Engineering,
138(10): 1214-1230.
Yeh, J. Y., Chen, L. W. and Wang, C. C. (2004). "Dynamic stability of a sandwich

beam with a constrained layer and electro rheological fluid core".
Composite Structures, 64(1): 47-54.
Yoon, S. J.and Kim, J. H. (2002) "A concentrated mass on the spring unconstrained
beam subjected to a thrust." Journal of Sound and Vibration, 254(4):
621-634.
Yuan, W.B., Kim, B. and Li, L.Y. (2014). "Buckling of axially loaded castellated
steel columns." Journal of Constructional Steel Research, 92: 40-45.
Yuan, W. B., Yu, N.T., Bao, Z.S. and Wu, L.P. (2016). "Deflection of castellated
beams subjected to uniformly distributed transverse loading."
International Journal of Steel Structures, 16(3): 813-821.
Uang, C.M. and Fan, C.C. (2001). "Cyclic stability criteria for steel moment
connections with reduced beam section." Journal of Structural
Engineering, 127(9): 1021-1027.
Van Oostrom, J. and Sherbourne, A. (1972). "Plastic analysis of castellated beams—

II Analysis and tests." Computers & Structures, 2(1-2): 111-140.
179
BIBLIOGRAPHY
Wang, P., Ma, Q. and Wang, X. (2014). "Investigation on Vierendeel mechanism

failure of castellated steel beams with fillet corner web openings."
Engineering Structures, 74: 44-51.
Wang, P., Ma, N. and Wang, X. (2014). "Numerical studies on large deflection
behaviors of restrained castellated steel beams in a fire." Journal of
Constructional Steel Research, 100: 136-145.
Wang, P., Wang, X. and 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.
Ward, J. K. (1990)."Design of composite and non-composite cellular beams." The

Steel Construction Institute Publication, SCI publications 100.
Zaarour, W. and Redwood, R. (1996). "Web buckling in thin webbed castellated

beams." Journal of Structural Engineering, 122(8): 860-866.
Zirakian, T. and Showkati, H. (2006). "Distortional buckling of castellated beams."

Journal of Constructional Steel Research, 62(9): 863-871.
Zhu, J., Qian, S. and Li, L.Y. (2017). "Dynamic instability of laterally-restrained
zed-purlin beams under uplift loading." International Journal of
Mechanical Sciences, 131–132: 408–413.
Zhu, J., Qian, S. and Li, L.Y. (2018). “Dynamic instability of channel-section beams
under periodic loading." Mechanics of Advanced Materials and Structures
(in press) (https://doi.org/10.1080/15376494.2018.1501521).
180

Analysis and Design of Castellated Steel Beam - 2019 Elaiwi PHD

Uploaded by

Copyright:

Available Formats

Analysis and Design of Castellated Steel Beam - 2019 Elaiwi PHD

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Analysis and Design of Castellated Steel Beam - 2019 Elaiwi PHD

Uploaded by

Copyright:

Available Formats

University of Plymouth

ANALYSIS AND DESIGN OF

Elaiwi, Sahar Sahib

SAHAR SAHIB ELAIWI

A thesis submitted to the University of Plymouth

Sahar Sahib Elaiwi, Doctor of Philosophy, 2019

School of Engineering, University of Plymouth

It gives me great pleasure to express my gratitude to everyone who contributed to the

Word count of main body of thesis: 34,956 words.

Figure count of the thesis: 69 figures.

Table count of the thesis: 16 tables.

Sahar S Elaiwi 7/oct/2019

PEARL (OA): http://hdl.handle.net/10026.1/13030

PEARL (OA): http://hdl.handle.net/10026.1/13384

PEARL (OA): http://hdl.handle.net/10026.1/14327

ACKNOWLEDEMENTS ....................................................................................................... iii

PAPERS AND CONFERENCES .......................................................................................... vi

LIST OF CONTENENTS .................................................................................................... viii

LIST OF TABLES ..................................................................................................................xv

LIST OF FIGURES ............................................................................................................. xvii

1.1. Motivation and background ..........................................................................................1

1.2. The problem statement and objectives..........................................................................4

1.3. Methodology .................................................................................................................6

1.4. Organization of thesis ...................................................................................................8

CHAPTER TWO ......................................................................................................................11

2. LITERATURE REVIEW ....................................................................................................11

2.1. History of development of fabrication method ........................................................... 11

2.2. The failure modes of castellated beams ......................................................................12

2.3. Local failure modes ....................................................................................................13

2.3.1. Vierendeel failure mode....................................................................................13

2.3.2. Web post failure ................................................................................................15

2.3.3. Web post buckling ............................................................................................15

2.3.3.1. Compression buckling of web post ..................................................... 16

2.3.3.2. Bending of web post failure due to moments in web-posts ................ 16

2.3.3.3. Web welds failure ............................................................................... 17

2.3.4. Shear resistance ................................................................................................ 17

2.3.4.1. Shear failure ........................................................................................ 17

2.3.4.2. Web shear buckling ............................................................................. 18

2.3.4.3. Flexural failure.................................................................................... 18

2.4. Global failure modes .................................................................................................. 18

2.4.1. Lateral-torsional buckling ................................................................................ 19

2.4.2. Lateral buckling failure due to moments in web-posts .................................... 20

2.5. The studies about deflection and shear effect ............................................................ 21

2.6. Research gaps ............................................................................................................. 25

CHAPTER THREE ................................................................................................................. 27

3. DEFLECTION CALCULATION OF CASTELLATED BEAMS .................................... 27

3.1. Introduction ................................................................................................................ 27

3.2. Analytical philosophy of deflection analysis of castellated beams ............................ 27

3.2.1. Deflection of simply supported castellated beam with uniformly distributed

3.2.2. Deflection of pinned-fixed castellated beam with uniformly distributed loads37

3.2.3. Deflection of non-uniform temperature distribution simply supported

3.2.4. Deflection of non-uniform temperature distribution roller- fixed supported

3.3. Numerical analysis ......................................................................................................45

3.3.1. The modeling of castellated beams ..................................................................45

3.3.2. Meshing consideration and material model ......................................................46

3.3.3. Boundary conditions and loading .....................................................................47

3.3.3.1. Simply supported castellated beam .....................................................47

3.3.3.2. Pinned-fixed supported castellated beam ............................................47

3.3.5. Comparison of results and discussion ..............................................................50