Structural Design

Against Deflection
Structural Design
Against Deflection

Tianjian Ji
CRC Press
Taylor & Francis Group
6000 Broken Sound Parkway NW, Suite 300
Boca Raton, FL 33487-2742

© 2020 by Taylor & Francis Group, LLC

CRC Press is an imprint of Taylor & Francis Group, an Informa business
No claim to original U.S. Government works
Printed on acid-free paper
International Standard Book Number-13: 978-1-138-61098-9 (Hardback)
This book contains information obtained from authentic and highly regarded sources.
Reasonable efforts have been made to publish reliable data and information, but the
author and publisher cannot assume responsibility for the validity of all materials or the
consequences of their use. The authors and publishers have attempted to trace the copyright
holders of all material reproduced in this publication and apologize to copyright holders if
permission to publish in this form has not been obtained. If any copyright material has not
been acknowledged, please write and let us know so we may rectify in any future reprint.
Except as permitted under U.S. Copyright Law, no part of this book may be reprinted,
reproduced, transmitted, or utilized in any form by any electronic, mechanical, or other
means, now known or hereafter invented, including photocopying, microfilming, and
recording, or in any information storage or retrieval system, without written permission
from the publishers.
For permission to photocopy or use material electronically from this work, please access
www.copyright.com (www.copyright.com/) or contact the Copyright Clearance Center, Inc.
(CCC), 222 Rosewood Drive, Danvers, MA 01923, 978-750-8400. CCC is a not-for-profit
organization that provides licenses and registration for a variety of users. For organizations
that have been granted a photocopy license by the CCC, a separate system of payment has
been arranged.
Trademark Notice: Product or corporate names may be trademarks or registered
trademarks, and are used only for identification and explanation without intent to infringe.

Library of Congress Cataloging‑in‑Publication Data

Names: Ji, Tianjian, author.
Title: Structural design against deflection / by Tianjian Ji.
Description: Boca Raton, FL : CRC Press/Taylor & Francis Group, [2020]
Identifiers: LCCN 2019049039 (print) | LCCN 2019049040 (ebook) | ISBN
9781138610989 (hardback ; acid-free paper) | ISBN 9780367897932
(paperback) | ISBN 9780429465314 (ebook)
Subjects: LCSH: Structural design. | Deformations (Mechanics)—Mathematical
models. | Tensegrity (Engineering)—Mathematical models. | Rigidity
(Geometry) | Structural engineering—Case studies.
Classification: LCC TA658 .J523 2020 (print) | LCC TA658 (ebook) | DDC
624.1/771—dc23
LC record available at https://lccn.loc.gov/2019049039
LC ebook record available at https://lccn.loc.gov/2019049040
Visit the Taylor & Francis Web site at
www.taylorandfrancis.com
and the CRC Press Web site at
www.crcpress.com
Contents

Preface ix
Acknowledgements xii
Author Bio xiii
1 Introduction 1
1.1 Deflection of Structures  1
1.2 Form, Deflection and Internal Forces  5
1.3 Intuition of Structures  10
1.3.1 Intuitive Knowledge  10
1.3.2 Intuitive Understanding  12
1.3.3 Intuitive Interpretation  13
1.3.3.1 Mathematical Equations  14
1.3.3.2 Observation of Structural Behaviour  14
1.3.3.3 Hand Calculation  17
1.3.3.4 Definition of Structural Concepts  17
1.4 Design against Deflection Based on Beam Theory  18
1.5 Rules of Thumb for Design  20
1.6 Effectiveness, Efficiency and Elegance  21
1.7 Organisation of Contents  22

2 Deflections and Internal Forces 25

2.1 Deflection of a Structure  25
2.2 Internal Forces, Deflections and Energies of Two
Rods 26
2.3 Internal Forces, Deflection and Energy of a Structure  29
2.4 Physical Meaning of ∆ 2,C  32
2.5 Intuitive Interpretation  34
2.6 Deflections due to Bending Moment, Axial and Shear
Forces 36
2.7 Characteristics of the Structural Concepts  39
2.7.1 The Four Structural Concepts  39
2.7.2 Generality 39
vi  Contents

2.7.3 Interchangeability 40
2.7.4 Compatibility 41
2.7.5 Reversibility 43
2.7.6 Relative Performance  44
2.8 Implementation 45
2.9 Summary 49

3 More Direct Internal Force Paths 51

3.1 Routes to Implementation  51
3.2 Hand Calculation Examples  52
3.2.1 Effect of the Four Bracing Criteria  52
3.2.2 The Most and Least Effective Bracing Patterns for a
Simple Frame  57
3.3 Practical Examples  64
3.3.1 Tall Buildings  64
3.3.1.1 John Hancock Center, Chicago  64
3.3.1.2 Leadenhall Building, London  68
3.3.2 Temporary Grandstands  73
3.3.2.1 Collapse of a Temporary Grandstand in
Corsica, France  75
3.3.2.2 A Temporary Grandstand in Eastbourne,
UK 76
3.3.2.3 Two Further Cases  80
3.3.3 Scaffolding Structures  82
3.3.3.1 Collapse of a Scaffolding Structure,
Manchester 82
3.3.3.2 Lack of Direct Internal Force Paths  83
3.4 Further Comments  83

4 Smaller Internal Forces 88

4.1 Routes to Implementation  88
4.2 Hand Calculation Examples  90
4.2.1 A Simply Supported Beam with and without
Overhangs 90
4.2.2 Y Shaped Columns with and without a Horizontal
Tendon 96
4.3 Practical Examples  104
4.3.1 Structures with Overhangs  104
4.3.1.1 HSBC Hong Kong Headquarters,
China 104
4.3.1.2 Roof of the Harbin Airport Lounge,
China 109
 Contents vii

4.3.2 Tree-Like Structures  110

4.3.2.1 Trees and Tree-Like Structures  110
4.3.2.2 Palazzetto dello Sport, Roma  113
4.3.2.3 Hessenring Footbridge, Germany  114
4.3.2.4 Further Examples  115
4.3.3 Self-Balancing 119
4.3.3.1 Madrid Racecourse, Spain  119
4.3.3.2 Salford Quays Lift Bridge, UK  121
4.4 Further Comments  122

5 More Uniform Distribution of Internal Forces 125

5.1 Routes to Implementation  125
5.2 Hand Calculation Examples  125
5.2.1 A Cantilever with and without an External Elastic
Support 125
5.2.2 An Eight Storey, Four Bay Frame with Different
Bracing Arrangements  128
5.3 Practical Examples  135
5.3.1 Structures with External Elastic Supports  135
5.3.1.1 Samuel Beckett Bridge, Dublin  135
5.3.1.2 Serreria Bridge, Valencia  138
5.3.1.3 Katehaki Pedestrian Bridge, Athens  139
5.3.2 Structures with Internal Horizontal Elastic
Supports 139
5.3.2.1 Manchester Central Convention Complex,
UK 139
5.3.2.2 Raleigh Arena, USA  143
5.3.3. Structures Derived from Topology
Optimisation 145
5.3.3.1 Evolutionary Structural Optimisation
(ESO) 145
5.3.3.2 A Bridge with a Flat Deck on the Top  148
5.3.3.3 A Bridge with a Flat Deck at the Middle
Level 149
5.3.3.4 A Long-Span Footbridge with an Overall
Depth Limit  150
5.4 Further Comments  153

6 Converting More Bending Moments Into Axial Forces 154

6.1 Routes to Implementation  154
6.2 Hand Calculation Examples  155
viii  Contents

6.2.1 A Beam with and without a Vertical Internal

Elastic Support  155
6.2.2 Rigid Plates Supported by Vertical and Inclined
Members 163
6.3 Practical Examples  168
6.3.1 Structures with Vertical Internal Elastic
Supports 168
6.3.1.1 Spinningfields Footbridge, Manchester  168
6.3.1.2 The Roof of the Badminton Arena for the
2008 Olympic Games, Beijing  170
6.3.2 Structures Supported by Inclined Members  175
6.3.2.1 Three Types of Support to
Superstructures 175
6.3.2.2 Ontario College of Art and Design,
Toronto 177
6.3.2.3 Roof Supports of Terminal 5 at Heathrow
Airport, London  178
6.3.3 Using Self-Weight of Structural Members—Alamillo
Bridge, Seville  180
6.4 Further Comments  182

7 Concluding Remarks 185

Bibliography188
Index189
Preface

Deflection is a major consideration in the design of structures. Buildings have

become taller, bridges longer and floors wider, and these all involve the control
of the structural deflection. Designing structures against deflection is not only
concerned with serviceability and stiffness, but is also related to safety, such
as failure due to buckling or excessive stress, efficiency and even the elegance
of structures. Therefore, it is necessary to take a holistic approach to examine
this topic.
Different claims are often made about form, deflection and internal forces
in structures. It is normally thought that structural form determines deflection
and internal forces. It has also been said that structural form is determined by
the flow of internal forces. The two statements appear contradictory, but they
have one thing in common in that form, deflection and internal forces are so
closely related that altering any one will change the other two. Choosing the
form of a structure can also be seen as selecting the path and distribution of
internal forces in the structure which extends the way to producing efficient
designs.
Many well-known engineers believe that the appropriate use of structural
principles will lead to efficient and elegant structures. For example, Professor
David Billington at Princeton University wrote “the best engineers followed
certain general principles of design to arrive at fine works, and these general
principles allowed for their own specific and personal vision of structures” in
his book, The Tower and the Bridge. Professor Mike Schlaich, partner of SBP,
wrote “Elegance appears when the challenging task of fusing the principles of
good structures seems to be achieved without much effort.” These statements
pose three questions: What are the general principles or the principles of good
structures? How can such principles be creatively implemented into the design
of structures? And how can the principles and the routes to implementation be
explicitly expressed and passed on to other engineers, especially the engineers
of the next generation. This book tries to answer the three questions by iden-
tifying some of the principles and concepts relating to deflections and internal
forces, demonstrating their effectiveness and efficiency and examining their
creative implementation in existing structures.
In parallel with the previous thoughts, the idea to write this book originated
from my teaching courses on structural engineering to final-year undergraduate
x Preface

and postgraduate students at the University of Manchester. New teaching con-

tents have been developed to help students’ understanding from structural
elements to whole structures and from theory to practical application, which
form the basis of the book and are organized through the common thread of
deflection and internal forces.
The presentation of the book also follows five philosophical criteria:

1. Seeking new connections between theory and practice. It has been said that
there is a gap between theory and practice. How can this gap be bridged?
When crossing a wide river, a bridge may require several intermediate sup-
ports. Similarly, new intermediate connections need to be sought between
theory and practice, such as those which exist between theory and struc-
tural concepts, between the structural concepts and physical measures to
implement them and between the implementation measures and practi-
cal cases. Connections have also been sought between examples with and
without involving implementation measures that are developed based on
one of the structural concepts. This allows illustrating and quantifying the
effectiveness and efficiency of the implementation measures and the cor-
responding structural concepts. Practical cases have also been connected
with simplified hand calculation models to reveal the effect of the embed-
ded structural concepts.
2. Exploring new meanings of structural theory. It is thought that structural
theory is a mature subject. However, it is still possible to explore new
meanings from old theories. New meanings of the virtual work principle
are explored and interpreted leading to a set of four structural concepts.
The structural concepts reveal the relationships between deflection and
internal forces of truss and frame structures. They form the basis of this
book showing that smaller deflection can be achieved through generating
more desirable distribution of internal forces in a structure. It is noted that
more desirable distribution of internal forces can also lead to more effec-
tive, efficient and elegant structures.
3. Being simple. It is a common belief that a basic and simple theory often
has wide application, i.e. simple and universal, such as the Newton’s sec-
ond law. What is a simple and universal theory for structural design? This
question will be examined in Chapter 2 of this book. Another way is to
make the presentation of theory simple allowing many structural engineers
to use it. Four structural concepts, abstracted from basic theory, will be
presented as “rules of thumb” for easy understanding and for practical
use. It is also believed that a problem, an equation or a structural phe-
nomenon can be explained in a simple manner while its physical essence
is captured. This way of explanation is termed as intuitive interpretation
in this book, which is an effective tool and skill and will be demonstrated
using examples.
4. Evolving into intuitive understanding. When the understanding of theory
evolves into an intuitive understanding, it will help to lead to appropriate
and good use of theory. Structural design, including the design against
 Preface xi

deflection, does not start from theory. Instead, it starts from the intuitive
understanding of structural behaviour and structural adequacy. For devel-
oping such intuitive understanding, a number of hand calculation exam-
ples, which are abstracted or simplified from practical cases, are studied
quantitatively and comparatively between with and without involving one
of the four structural concepts.
5. Making wide and wise applications of theory. A number of practical
cases, linking with the hand calculation examples, demonstrate that the
four structural concepts presented have been used widely and have pro-
vided clever solutions to challenging engineering problems. The routes to
implementation of the structural concepts into the design of structures are
explicitly listed and discussed to promote wider and wiser use in practice.
It is hoped that the reader will be stimulated by the examples and cases
presented to make their own creative applications.

This book contains material on three structural concepts relating to stiffness

developed in an earlier book, Understanding and Using Structural Concepts,
by Tianjian Ji, Adrian Bell and Brian Ellis, to which this book is a successor.
Anyone who has used Understanding and Using Structural Concepts will find
the present book substantially more comprehensive on the understanding and
on the application of the three structural concepts.
Structural design against deflection requires a broad knowledge of materials,
analysis, structural behaviour, loading, environment, construction details, etc.
This book focuses on structural concepts and their implementation in practice
to achieve more effective, efficient and perhaps even more elegant structures.
This book integrates teaching, practice and research through the ­common
thread of structural concepts for structural design against deflection. It is hoped
that the book provides an inspirational experience to advanced ­undergraduate
and graduate students studying civil engineering and a­ rchitecture, and enhances
the holistic comprehension of structural engineers and architects.

Tianjian Ji
The University of Manchester, UK
Acknowledgements

I express my sincerest gratitude to Dr. Adrian Bell, my colleague and co-author

of Understanding and Using Structural Concepts, for his help in checking the
manuscript of this book, for his constructive comments and for making the
book more readable. I am also very grateful to Dr. Brian Ellis, my previous col-
league and also co-author of the previously mentioned book, who was able to
read and check the first two chapters of this book on a short notice.
The writing of this book has been constrained by my available photographs
and figures of actual structures. Several individuals have however helped me in
this respect and their contributions are acknowledged next to relevant photo-
graphs in the text. In particular, I am indebted to Mr. Nicolas Janberg, owner
and creator of structurae.net, Germany, for providing several photos with high
quality.
I am very grateful to Mr. Andrea Codolini, a Ph.D. student at the University
of Manchester, for producing some of the drawings in this book.
The assistance of Taylor & Francis in the publication of this book is greatly
appreciated. I would like to thank Mr. Tony Moore, Senior Editor, for his
encouragement in writing this book. Gabriella Williams and Lisa Wilford,
at the UK office, and Jennifer Stair, Production Editor, at the USA office,
­provided assistance at different stages. I am also grateful for the help from
­Denise File and her team at Apex CoVantage in preparing the final version of
the manuscript.
Author Bio

Dr. Tianjian Ji is Reader in Structural Engineering at

the University of Manchester, a Chartered Engineer
and Fellow of the Institution of Structural Engineers
and Higher Education Academy, UK. He graduated in
Civil Engineering and received MSc degree in Structural
Mechanics both from Harbin Institute of Technology,
China, and PhD degree in Civil Engineering at the Uni-
versity of Birmingham, UK. Before joining Manchester
in 1996, he had worked in industry for over ten years,
mainly at Building Research Establishment, UK, and
China Academy of Building Research, Beijing.
His research area is in vibration and structural dynamics using both experi-
mental and theoretical methods, including structural vibration induced by
human rhythmic crowd loads, human-structure interaction, probabilistic seis-
mic risk assessment of nuclear power plants and finite element modelling of
structural dynamic behaviour. He has supervised 14 PhD students and over
70 MSc students, and published over 140 articles.
He has taught at all year levels on Mechanics, Structural Analysis, Structural
Design, Research Methods and Earthquake Engineering. He led the development
of what is called Seeing and Touching Structural Concepts for helping students
to develop an intuitive understanding of structural concepts (www.structural-
concepts.org). A book was published with the same title in 2008 and the second
edition of the book with a revised title, Understanding and Using Structural Con-
cepts, was published in 2016. Both editions were translated into Chinese and pub-
lished in China. He received the Award for Excellence in Structural Engineering
Education from the Institution of Structural Engineers in 2014 and the Teach-
ing Excellence Award from the University of Manchester in 2016.
He has actively taken consultancy work for helping to solve practical vibra-
tion problems, including some most prestigious structures in the UK, such as
vibration measurement of the London Eye, and for providing or reviewing
remedial schemes of building floors subjected to rhythmic crowd loads or
impact loads.
Chapter 1

Introduction

1.1 Deflection of Structures
For the structural design of a building, engineers need to check deflection, vibra-
tion, stability and strength of the structure and its components, and ensure that
they satisfy all requirements, i.e. they have appropriate values smaller or larger
than limiting values. Deflection and vibration are classified as serviceability
problems while stability and strength are considered to be safety problems.
These four issues are normally analysed and checked independently; but are
there any connections between the four of them?
The deflection of structures is a key serviceability consideration and may
often control the design of slender floors, tall buildings and long bridges. As
buildings become taller, bridges longer and floors wider, the associated deflec-
tions of these structures become major design issues.
Deflection limits are applied to structural elements, such as beams and floors,
and to whole structures, such as buildings and bridges. The limits often require
that the possible maximum deflection of a structure or a structural element
should be smaller than a certain value. For example, the limit for the maximum
deflection of a truss structure is 1/180 of its span [1.1]. For a defined structure
and a given loading, the deflection of the structure is calculated using the fol-
lowing equilibrium equation:

[K ]{U } = {P} (1.1)

where [K] is the stiffness matrix that is related to the structural form and the
cross-sectional and material properties of the structural members, {U} is the
deflection vector to be determined and {P} is the given loading.
Structural vibration is another type of serviceability issue, which may cause
discomfort to users of the structure and restrict the functionality of the struc-
ture. Structural vibration is not only related to the dynamic loads applied but
is also related to the dynamic properties of the structure, i.e. natural frequency,
damping ratio and modal mass or modal stiffness. In the design of grandstands
and floors used for rhythmic activities, one design philosophy requests that the
2  Structural Design Against Deflection

fundamental natural frequency of the structure should be larger than a certain

value to avoid possible resonance [1.2, 1.3]. The natural frequencies and the
mode shapes of the structure can be determined by solving the following eigen-
value equations:

([K ]   2 [M ]){v }  {0} (1.2)

where [M ] is the mass matrix, ω is the circular natural frequency and {φv }
is the vibration mode of the structure. The stiffness matrix [K ] is the same as
that in equation 1.1.
Stability of a structure or a structural member is considered as a safety prob-
lem. When a structure is subjected to external loads and self-weight, compres-
sive forces/stresses are induced in the body of the structure. In such a situation,
engineers need to check if the whole structure will lose its stability and if any
individual member will buckle. Quite often the buckling of a compression
member can result in a sudden failure of the member which may lead to the
development of a mechanism and local or even global failure of the structure.
The global stability of a structure is evaluated by a similar eigenvalue equation
to that for natural frequency:

([K ]  [KG ]){s }  {0} (1.3)

where [KG ] is the geometric or initial stress stiffness matrix that is formed
based on the applied loads and the structural form, λ is the buckling load
factor ( λ times the existing loads would cause global instability of the struc-
ture) and {φs } is the bucking mode of the structure which describes the pattern
of instability.
Strength measures the capacity of individual structural members to with-
stand the internal forces applied to them by the external loads on the structure.
Unlike deflection, vibration and stability, strength is considered for individual
members rather than for the whole structure, but the failure of an individual
member may lead to an unsafe structure. Once the internal force in a member
is determined, the corresponding stress is easily calculated and compared with
its allowable stress. If the stress is larger than the allowable stress, the cross-
section of the member may need to be enlarged.
The relationship between deflection and bending moment of a uniform
beam is:

d 2u(x )
EI  M (x )(1.4)
dx 2

where u(x ) and M (x ) are the deflection and bending moment at coordinate x
of the beam, and EI is the rigidity of the cross-section of the beam.
For a plane element in finite element analysis, the relationship between
strain {ε } and nodal displacement {d } is defined as:

{ }  [E]{ }  [E][B]{d } (1.5)

 Introduction 3

Table 1.1 Relationships betw een structural design problems

Type of problem Strength Deflection Free vibration Stability

d 2u(x)
Basic equation EI = − M(x) [K ]{U} = {P} ([K ]   2 [ M]){v } ([K ]  [ KG ]){S }
dx 2
 {0}  { 0}
{σ } =[ E ][ B ]{d }
Relation to The internal force Deflection, natural frequency and buckling
deflection and stress are load factor are all related to stiffness (the
directly related stiffness matrix).
to deflection.

where [B] is the strain-displacement matrix that transfers the nodal deflections
of the element to the strains within the element, and [E] is the material prop-
erty matrix. The nodal displacement {d} of the element is taken from the global
displacement {U} in equation 1.1.
It can be observed from equations 1.1–1.5 that:

• For deflection, vibration and stability problems, equations 1.1–1.3 contain

the stiffness matrix [K] of the structure and show qualitatively that the
stiffer the structure, the smaller the deflection, the higher the natural fre-
quency and the larger the buckling load factor.
• As the deflection vector and stiffness matrix are “reciprocal” of each other
for a unit load vector, the previous statement can be rewritten as: the
smaller the deflection of a structure, the higher the natural frequency and
the larger the buckling load factor.
• For a strength problem, equations 1.4–1.5 show that the internal forces or
internal stresses are directly related to deflection.

The relationships between the four structural design problems are summarised
in Table 1.1.
It can be seen from Table 1.1 that deflection is a physical quantity that is direc­tly
related to internal forces or stresses (equations 1.4–1.5) and is indirectly related
to the natural frequency and the buckling load factor (equations 1.1–1.3).
For explicitly expressing the relationships between deflection and natural
frequency, between deflection and buckling load, and between deflection and
internal forces, consider a simply supported uniform beam with a length of L,
cross-sectional rigidity of EI and a uniformly distributed mass of m.

a) Deflection

The maximum deflections of the beam due to its self-weight, mg, and a concen-
trated load, F, at its centre are respectively:

5mgL4 FL3
q   and  F   (1.6, 1.7)
384EI 48EI

where g is the acceleration due to gravity.

4  Structural Design Against Deflection

b)  Fundamental Natural Frequency and Deflection

The fundamental natural frequency of the uniform beam is:

 EI
f  (1.8)
2 mL4

It can be seen that equations 1.6 and 1.8 both contain mL4/EI that gives the
connection between the fundamental natural frequency and the maximum
deflection. Eliminating mL4/EI in the two equations gives the relationship
between the fundamental natural frequency and the maximum deflection:

17.75
f  (1.9)
q

In this calculation, g is taken as 9810 mm / s 2 and ∆ q is in mm. Equation 1.9

shows that the fundamental natural frequency of a beam is inversely propor-
tional to the square root of the deflection. In general, the smaller the deflec-
tion, the larger the fundamental natural frequency. Equation 1.9 has been
used in several design guides [1.4] to facilitate a quick estimation of the funda-
mental natural frequency without conducting an eigenvalue analysis.

c)  Critical (or Buckling) Load and Deflection

When the beam is subjected to a compressive load P applied at its ends along
its longitudinal axis, the critical load is:

 2 EI
PCR  (1.10)
L2

Substituting equation 1.7 into equation 1.10, by removing EI, leads to:

FL 2
PCR  (1.11)
48 F

Equation 1.11 indicates that the buckling load of a strut is inversely proportional

to the lateral deflection of an equivalent beam caused by a concentrated load act-
ing at its centre. Equation 1.11 also suggests that the buckling load of a strut can
be experimentally determined by conducting a non-destructive bending test [1.5].

d)  Bending Moment and Deflection

The maximum bending moment in the beam due to its self-weight is:

mgL2
Mq = (1.12)
8
 Introduction 5

The relationship between the maximum bending moment M q and the maxi-
mum deflection ∆q can be derived from equations 1.6 and 1.12 as follows:

48EI
M q  1.2mgEI q  q (1.13)
5L2

Equation 1.13 shows qualitatively, at a structural element level, that the smaller

the maximum deflection, the smaller the maximum bending moment.
Seeking the connections between deflection, natural frequency, buckling
load and internal force not only helps gain a better understanding but also
leads to wider and wiser applications, such as estimating the fundamental nat-
ural frequency using a known deflection and determining the buckling load by
conducting a bending test.
A question then arises how to better design structures against deflection by
reducing deflections which also helps to increase the fundamental natural fre-
quency and buckling load capacity of the structures and reduce internal forces.
Therefore, there is a need to return to basics and examine the relationships between
deflection, structural form and internal forces in addition to applied loading.

1.2 Form, Deflection and Internal Forces

It is often thought that structural form determines internal forces in a structure.
This understanding can be based on an input-structure-output model as follows:

Figure 1.1 Relationships between input, structure and output.

where loading is the external forces applied on a structure; structural form

describes the global structural system that also embraces architectural form, and
internal forces are the forces in structural members resulting from the loading on
the structure and the structural form, which normally include axial forces, shear
forces and bending moments. When a structural form is designed or selected
and the structure is subjected to a given load (input), the deflections and internal
forces (output) can then be uniquely determined (Figure 1.1), i.e. the output is the
consequence of the response of the structural form to the given set of loads, i.e.:

[K ]{U } = {P} or {U }  [K ]1 {P} (1.1)

The internal forces in members of the structure are then normally determined
based on the calculated deflections. These deflections and internal forces may
then be used as feedback to revise the geometry of the structure and the dimen-
sions of its members, which leads to a change of the stiffness matrix in equa-
tion 1.1, to achieve an improved design.
6  Structural Design Against Deflection

Figure 1.2 Relationship between form, internal forces and deflections of a structure.

Alternatively, it may be said that the form of a structure is determined by the

flow of internal forces. In fact, structural form, internal forces and deflections
of a structure subjected to a given loading are closely related and interact with
each other. In order to reveal and examine the relationships between structural
form, internal forces and deflections, Figure 1.1 may be revised as shown in
Figure 1.2. This indicates that altering internal forces can change deflections
and structural form and controlling deflection can also revise structural form
and internal forces, in addition to varying structural form can lead to a new
set of internal forces and deflections. For example, if the deflection at a given
point in a particular direction is constrained, this will correspond to a change
of the structural form by requiring an appropriate support at the point and this
alters the stiffness matrix. Altering internal force paths and varying structural
form occur simultaneously although the magnitudes of the internal forces are
determined after the structural form has been confirmed.
It is unlikely that the relationships between structural form, internal forces
and deflections of a structure can be expressed explicitly. However, it is possi-
ble to gain a qualitative understanding of them through examining two similar
plane frames.

Question

Figure 1.3 shows two four-bay and four-storey plane pin-jointed structures

with the same dimensions. All the members are made of the same material
(E) and have the same cross-sectional area (A). The vertical and horizontal
members have the same length (L). The same concentrated loads of 0.5N are
applied anti-symmetrically at the two top corners in the horizontal direction
on each frame. Two bracing members are required to be placed in each storey,
hence the two frames use the same number of bracing members, i.e. the same
amount of material. The only difference between the two frames is the arrange-
ments of bracing members, which can be discussed as follows:
Frame A: Four bracing members are placed symmetrically on each of the
two side bays and are arranged in the same orientation, i.e. the bracing mem-
bers are not directly linked. As all eight bracing members are placed on the
side bays, the two middle bays have no bracing members. This type of bracing
pattern is often seen in practice.
 Introduction 7

(a) (b)
Figure 1.3 Two plane pin-jointed frames with different bracing arrangements. (a)
Frame A: the bracing members are arranged in parallel in the two side
bays. (b) Frame B: the bracing members are arranged across the four bays.

Frame B: The bracing members are arranged across the four bays of the frame
and are linked in straight lines. This bracing pattern can be generated from that
in Frame A by: 1) changing the orientation of the two bracing members on the
ground (first) level; 2) moving the two bracing members in the second level hori-
zontally inward to the next panels and altering their orientation; and 3) moving
the two bracing members in the third level horizontally inward to the next panels.
With the structure form and the loading defined, the internal forces in the
members and the maximum deflections of the two structures can be deter-
mined and the relationships between the form, deflections and internal forces
for these particular structures can then be examined.

Solution

The two structures are statically indeterminate. However, they are both sym-
metric structures subjected to anti-symmetric loads. According to the struc-
tural concept that a symmetric structure subjected to anti-symmetric loading
will result in only anti-symmetric responses (internal forces and deflections),
the four central vertical members will have to be in a zero-force state and the
nodal points along the central vertical members of the two frames will have no
vertical displacements. Therefore, the two frames can be simplified and equiva-
lently represented by their left halves with appropriate boundary conditions as
shown in Figure 1.4. Each half frame has 16 vertical and horizontal members
and four bracing members.
It can be noted that the middle vertical members are removed as there are no
internal forces in these members and the vertical displacements of the points
along the middle members are constrained using roller supports. Now the two
half frames become statically determinate structures and the internal forces of
all the members can be directly and easily calculated by hand.
The internal forces in the members of the two simplified frames can be deter-
mined using the equilibrium conditions at the pinned joints and the calculated
internal forces in the members of the two half frames are as shown in Fig-
ures 1.4 (a) and (b), where positive values indicate the members in tension and
8  Structural Design Against Deflection

negative values indicate the members in compression. In addition, the internal

force paths (non-zero force members) are indicated using dashed lines.
The maximum lateral deflections at the loading positions of the two half-
frames can be determined using a well-known equation, taken from textbooks
[1.6, 1.7], as:
20
Ni2 Li
 A,max  
i 1 EA

L  1  
2 2 2
3  2
    5  12  2     2   2 2   2  4  2
EA  2  2  2 
   
11.75  2 2 29.16L
 2 
EA EA

L  1  
2 2
N 2L
20  2
 B,max  i i     4  
 2  2  4  2
i 1 EA EA  2    
 


1  2 2 L  2  7.656L
EA EA


(a) (b)

Figure 1.4 S implified and equivalent presentation of Frames A and B, the internal
force paths in dashed lines and values of the internal forces. (a) Frame
A. (b) Frame B.
 Introduction 9

The ratio of the two deflections is:

 B,max 7.656L EA
   0.263
 A ,max EA 29.16L

The maximum lateral deflection of Frame B is only 26.3% of that of Frame

A although the same number of members is used in both frames. This demon-
strates the significant effect of structural form on the deflection and the inter-
nal forces in members as the two frames use the same amount of material.
The paths and values of the internal forces in the members of the two frames
and the characteristics of the bracing patterns can be observed and examined
to provide an intuitive understanding of the reasons why Frame B has much
smaller deflections than Frame A:

1. There are more zero-force members in Frame B than in Frame A. There are
12 zero-force members in Frame B compared with six in Frame A.
2. The magnitudes of the internal forces in the members in Frame B are smaller
than those in Frame A. The largest absolute internal force in a member is
0.5 in Frame B compared with 2.0 in a member in Frame A without con-
sidering the constant internal forces in the bracing members, which are the
same in the two frames.
3. The differences between the magnitudes of the internal forces in the
members of Frame B are smaller than those of Frame A. The largest
absolute difference between the internal forces in members is 0.207 in
Frame B compared with 1.5 in Frame A, ignoring the members with
zero-force.
4. The four-bay, four-storey Frame B is braced globally while Frame A is
braced locally in its two side bays.
5. It may be considered that Frame B looks more pleasing and elegant than
Frame A (Figure 1.3)

The first three observations show that the characteristics of the internal force
paths and distributions in the two frames are clearly technical issues. The
fourth observation is about the geometry or pattern of the bracing members,
which is a design issue. The fifth observation concerns the appearance of
the two frames, which is related to a human perception of the quality and
beauty of a structure. It appears that more zero-force members, smaller inter-
nal forces in members and a more uniform distribution of internal forces in
members lead to smaller deflections. These observations from the two frame
examples are interrelated and inspire the thought that the internal force flow
and its distribution can be positively designed to define the structural geom-
etry and topology, and to control deflections. The observations generate the
following three questions:

1. What are the rules or structural concepts embedded in Frame B, which

result in Frame B having much smaller deflections than Frame A without
10  Structural Design Against Deflection

using more material? Are such rules or structural concepts applicable to

the design of other structures?
2. How can the flow and distribution of internal forces be actively considered
to aid the design of structural form?
3. How can internal forces be designed to make a structure more effective
(smaller deflections), more efficient (less materials) and perhaps more
elegant?

Answering these three questions requires a harmonious combination of intui-

tive understanding of structures and a sound technical knowledge of struc-
tures, which will be discussed in Chapter 2.

1.3 Intuition of Structures
According to Mario Salvadori (a structural engineer and professor of both
civil engineering and architecture at Columbia University), who wrote a for-
ward for Torroja’s book [1.8], outstanding engineers, like Eduardo Torroja
(Spanish structural engineer and architect), reached very high levels through
four phases: 1) devoting their early years to a long and thorough study of fun-
damentals; 2) applying the fundamentals to the solutions of original problems
in practice and accumulating experience; 3) slowly synthesising their accumu-
lated experience to reach what is called “intuition”; and 4) bringing them to
higher and higher levels with ever-decreasing effort and ever-increasing enjoy-
ment of their work.
Pier Luigi Nervi (Italian structural engineer and architect) said that the mas-
tering of structural knowledge is the result of a physical understanding of the
complex behaviour of a building, coupled with an intuitive interpretation of
theoretical calculation.
These thoughts from eminent engineers indicate the importance of intuitive
interpretation of theoretical calculation and structural behaviour and the ways
of developing intuition. They have also led to thinking about what intuitive
interpretation means and how intuition could be learned at an earlier stage or
taught at university [1.9].
Intuitive knowledge, intuitive understanding and intuitive interpretation are
related but they have different meanings and characteristics.

1.3.1 Intuitive Knowledge
Such knowledge often comes from experience which is correct but may not
have theoretical support or the theory behind the knowledge is not available or
is not known. For example, many families know that rubber footpads reduce
the vibration generated by washing machines. However, most do not know the
reason why the small pads can effectively reduce the vibration, but they can
still make a good use of the knowledge. This type of knowledge can be gained
from personal experience or learned from the experience of others.
 Introduction 11


(a) (b)

Figure 1.5 V ibration isolation. (a) Tyres used for isolation in practice.(Courtesy of
Professor Biaozhong Zhuang, Zhejiang University, China). (b) Demonstra-
tion of vibration isolation in teaching.

Figure 1.5 illustrates two isolation measures used for vibration reduction.

Figure 1.5a shows tyres placed between the ground and a generator in a rural
area of a developing country. The presence of the tyres led to a lower natural
frequency of the generator-tyre system and moving it away from the operating
frequency of the generator. The operators of the generator had not received a
university education and were not aware of vibration theory, but they knew
from their own experience, or the experience of others, that the presence of the
tyres could reduce vibration. Figure 1.1(b) shows a laboratory demonstration
of the effect of vibration isolation to students at a university. A medical shaker
was used as a shaking table to generate harmonic base movements in three per-
pendicular directions. One glass is fixed directly to the table of the shaker and
a similar glass is glued to a layer of plastic foam that is fixed to the table. The
two glasses are filled with similar amounts of water. When the shaker moves
at a preset frequency, it can be observed that the water in the glass mounted
on the plastic foam moves less significantly than that in the other glass. The
difference in movements is due to the effect of the plastic foam that isolates
the base motion of the glass above. The plastic foam and the glass and water
12  Structural Design Against Deflection

above form a new system that has a much lower natural frequency than that
of the glass with water alone. This demonstration has been shown to students
to enhance their understanding of vibration reduction and to audiences of the
general public to help them gain the intuitive knowledge that isolation can
reduce vibration.
The two examples of vibration isolation indicate that theory can be illus-
trated and practical cases can be simulated using physical models to produce a
broader perspective and gain intuitive knowledge.

1.3.2 Intuitive Understanding
Such understanding of a problem can be gained from observations and from
practical experience or/and from fundamental theories. It often comes without
conscious learning or theoretical derivations. It is observed that a person who
has many years of practical experience and a sound theoretical foundation is
able to gain an intuitive understanding of a problem.
Figure 1.6 shows the North Stand at Twickenham, UK, in which vibra-
tion measurements were taken on the middle cantilever tier when the stand
was empty and when the stand was full of spectators. Figure 1.7 shows the
measured response spectra of the tier, without spectators and with specta-
tors. Comparing the two spectra, three significant phenomena were appar-
ent [1.10]:

Figure 1.6 T he North Stand, Twickenham.

 Introduction 13

 
(a) (b)

Figure 1.7 Response spectra of the North Stand, Twickenham. (a) Without specta-
tors. (b) With spectators.

1. An additional natural frequency was observed in the occupied stand.

2. The damping increased significantly when spectators were present.
3. The natural frequency of the empty stand was between the two natural
frequencies of the occupied stand.

The observations were contrary to the belief that a human body acts as an inert
mass in structural vibration [1.11]. If the spectators acted as inert masses, the
occupied stand would have only one natural frequency that should be smaller
than that of the empty stand, and the inert body masses would not increase the
damping of the occupied stand. The intuitive understanding of the experimental
observations was that the spectators did not act as inert masses on the stand in
the vertical structural vibration. This intuitive understanding was an outcome
of the site observations and some knowledge of fundamental vibration theory
and has led to much research on the new topic of human-structure dynamic
interaction [1.10].
Students at universities learn structural theory but they may not often have
opportunities to observe structural behaviour and conduct experiments. How-
ever, it is possible to produce physical models and to show related practical
examples for students to appreciate.

1.3.3 Intuitive Interpretation
Intuitive interpretation means that an equation, an observation or structural
behaviour can be explained in a simple manner, while the explanation cap-
tures the physical essence of the problem. This often results from a sound
understanding of theoretical fundamentals and from practical experience.
Intuitive interpretation in structural engineering is an effective tool to explore
new meanings, seek new connections, develop new understanding and pro-
mote wide and wise applications. It is best to illustrate intuitive interpretation
using examples.
14  Structural Design Against Deflection

1.3.3.1  Mathematical Equations

Many equations can be used to practice intuitive interpretation and to gain an
improved understanding of theory, leading to practical applications and appre-
ciation of what intuitive interpretation means. For example, second moment of
area of a plane cross-section is expressed as:

I   y 2dA (1.14)

where y is the distance between the neutral axis of the cross-section and dA that
is the area of an infinitely small area. Second moment of area is the geometrical
property of the section which is related to its area and to the distribution of the
area. Students were asked to interpret equation 1.14. One of the answers was
that the second moment of area of a cross-section is the sum of the products
of a small area and the square of the distance between the centre of the area
and the neutral axis of the section. This statement is correct but is actually a
verbal expression of equation 1.14 rather than an intuitive interpretation that
tends to capture the physical essence of the equation. The intuitive interpreta-
tion of equation 1.14 should be: the further (closer) the material is away from
(to) the neutral axis of a section, the larger (smaller) the contribution to the
second moment of area of the section. It is this interpretation, or understand-
ing, that forms a basis for creatively designing the shape of a cross-section of
a beam, such as I-sectioned beams or cellular beams. As tall buildings can be
treated as cantilevers in conceptual designs, shear walls and columns should be
arranged as far away as possible from the neutral axis of the building plane.
Equation 1.14 provides a means to calculate the second moment of area of a
cross-section while the intuitive interpretation of equation 1.14 paves a way
for creative applications.

1.3.3.2  Observation of Structural Behaviour

Figure 1.8a shows a test rig, equipment and the specimen that were used in a
vibration-buckling test. A straight steel strut was placed in the test rig with the
two ends of the strut having pinned supports. Weights were added gradually
to apply compression to the strut until it buckled. In parallel with the buckling
test, the fundamental natural frequency of the loaded strut in the lateral direc-
tion was measured at each loading stage using a small accelerometer placed at
the centre of the strut and linked to a vibration analyser. At each loading stage,
a gentle lateral impact was applied to the strut (a tap from a finger) to gener-
ate lateral vibrations. The weights and the natural frequency at each loading
stage were recorded. Figure 1.8b shows the relationship between the measured
natural frequency squared, on the vertical axis, and the applied vertical load,
on the horizontal axis. The points show the measurements and a straight line
is fitted to the points.
Students were asked to interpret the observations and the results shown in
Figure 1.8. One answer was that there is a linear relationship between the
natural frequency squared and the compressive force. This is an obvious
 Introduction 15

(a)

(b)

Figure 1.8 V ibration-buckling experiment. (a) Vibration-buckling test of a loaded

strut. (b) Relationship between natural frequency squared and the com-
pression force.
16  Structural Design Against Deflection

observation from Figure 1.8(b). However, there is a more important observa-

tion: when the strut reaches its buckling load, its fundamental natural fre-
quency becomes zero, which corresponds to the intersection point of the fitted
line and the horizontal axis in Figure 1.8b. As the lateral stiffness of the strut is
proportional to its fundamental natural frequency squared, the intuitive inter-
pretation of this observation is that: a strut buckles when it loses its lateral
stiffness. This interpretation provides an alternative definition of buckling.
When checking current textbooks, one can note that the current definition of
buckling of a strut is a description of the phenomenon of buckling.
This experiment and observation have generated further discussion as to
whether frequency measurements can provide a non-destructive method for
predicting the buckling load of a real structure. Taking two frequency meas-
urements at different loading magnitudes and drawing a straight line passing
through the associated two points in Figure 1.8b, the intersection point of the
line and the horizontal axis is the buckling load. This requires high quality
measurement of the fundamental natural frequencies of the strut at two differ-
ent loading states for application. Such a linear relationship between P and f 2
may not exist for other types of structures.
This example demonstrates how intuitive interpretation expresses an impor-
tant observation of structural behaviour in a simple manner while capturing its
physical essence of buckling, and can also be used to illustrate the philosophic
criteria embedded in the presentation of the book:

Seeking new connections: Buckling of a strut and free vibration of a beam

are two different problems in textbooks and in engineering design,
and they are normally considered independently. The new connection
between the two problems was established through the experimental
set-up by which the buckling behaviour of a strut and the vibration
behaviour of a simply supported beam could be examined simultane-
ously in some details.
Exploring new meanings: Following the new connection and the experi-
ment, the new meaning of buckling was explored, interpreted and stated
concisely as a strut buckles when it loses its lateral stiffness that is an
alternative definition of buckling to complement with the existing one: a
strut buckles when bending occurs.
Being simple: For conducting a combined buckling and vibration experi-
ment, a strut may be the simplest structural member. The experiment
used an existing rig for buckling tests of struts, which made the experi-
ment simple and straightforward. A simple frame model could be used
for conducting similar tests, but it would require more efforts both exper-
imentally and theoretically.
Evolving into intuitive understanding: The fundamental natural frequency
also indicates the remaining buckling capacity of a structure. When the
fundamental natural frequencies of similar racking systems with differ-
ent loads can be measured, these data would help estimate the remaining
buckling capacity of the structures.
 Introduction 17

Making wide and wise application: This study encouraged further studies to
examine the possibility and conditions for developing a non-destructive
experimental method that natural frequency measurements are used to
predict buckling loads of structures. Following the same route of this
experiment, a new connection between a bending test and a buckling test
was sought, which led to a conclusion that a bending test can be used to
predict the buckling load of the test member [1.5].

1.3.3.3  Hand Calculation

Hand calculation is an effective skill which can facilitate intuitive interpreta-
tion. It may be a necessity for intuitive interpretation that one should be able to
simplify a complex structure into a model that still retains the physical essence
of the structure but is simple enough to be analysed by hand.
Different bracing patterns can be observed in existing structures, such as
tall buildings, temporary grandstands and scaffoldings. Real structures are
three dimensional and are too demanding for hand analysis. For hand analy-
sis, there is a need to create simple structure models abstracted from the real
structures which possess the physical essence of the bracing patterns. This has
been demonstrated in Figure 1.3 in Section 1.2. The hand calculations for the
two simple frames has provided the necessary results for comparison and for
intuitive interpretation, which help to identify new meanings, new connections
and new understanding of the relationships between form, internal forces and
deflection.

1.3.3.4  Definition of Structural Concepts

In an earlier work [1.5], intuitive interpretation is used to define structural
concepts as

A structural concept is an intuitive interpretation and concise representa-

tion of a mathematical relationship between physical quantities, which
captures the essence of the relationship and provides a basis for practical
applications in structural engineering.

This definition clearly states that structural concepts come from the intuitive
interpretation of mathematical equations. Such interpretation can be applied
to observations, structural behaviour and results from hand and computer
calculations. The intuitive interpretation of equation 1.14 and Figure 1.8
are examples in which two structural concepts are identified and presented
concisely.
The illustrations in this subsection indicate that using models, practical
examples, observations and calculation results can create scenarios for effec-
tively helping students to gain intuitive knowledge and intuitive understanding
and to practice intuitive interpretation, which complements the contents of
textbooks.
18  Structural Design Against Deflection

1.4 Design against Deflection Based on Beam

Theory
Structural design against deflection requires the use of equations to calculate
the deflections. There is a simple equation for the central deflection of a uni-
form beam in textbooks [1.6]:

qL4
 max   (1.15)
EI

where ∆ max is the maximum deflection, q is a uniformly distributed load, L

is the span, E modulus of elasticity and I the second moment of area of the
cross-section of a beam. α is a non-dimensional coefficient relating to bound-
ary conditions, for example, 5/384 for a simply supported beam and 1/8 for a
cantilever. This equation is explicit and clearly shows the relationship between
deflection and five other parameters. Implementation of equation 1.15 for
reducing deflection has been used in practice [1.12] as outlined following a
“rule of thumb” format:

1. Reducing span L: As the deflection is proportional to L to the power of

four, reducing span where possible is the most effective way to reduce
deflection, e.g. via the provision of additional supports. Figure 1.9 shows

Figure 1.9 A  footbridge with cable stayed mid-span support, Southampton, UK.
 Introduction 19

Figure 1.10 Increasing I value by adding a plate at the bottom of a beam.

such an example where cable stays to one side of the footbridge act as
additional supports to reduce the deflection of the deck.
2. Increasing second moment of area I: This is normally applicable to indi-
vidual members, such as by using a larger cross-section or adding mate-
rial as far away as possible from the neutral axis of a given cross-section
to enlarge the I value effectively. Figure 1.10 shows the familiar exam-
ple of a long-narrow steel plate welded to the bottom of an I section
steel beam. As the additional material was placed as far away from the
neutral axis of the cross-section as possible, it effectively contributed
to the second moment of area of the cross-section and resolved a pos-
sible vibration problem. Conceptually, a tall building can be seen as a
large cantilever, the second moment of area of its cross-section can be
increased by arranging the positions of columns, shear walls and bracing
members of the building to be as far away as possible from the neutral
axis of the cross-section.
3. Reducing α : This can be achieved by enhancing the boundary conditions,
such as changing pinned supports to fixed supports. Alternatively, adding
elastic supports to a structure can be adopted. For example, the cables of
a cable-stayed bridge provide elastic supports to the bridge deck, allowing
the bridge to span longer distances. In this case the bridge deck can be seen
as a beam on an elastic foundation. The cable support shown in Figure 1.9
can also be explained as an elastic support.

Equation 1.15 is derived from simple beam theory and is applicable to any

problems that can be converted to an equivalent beam, such as bridges or
tall buildings. Actually, the understanding gained from this equation has been
applied to more complex situations, such as plates, floors and roofs, far beyond
20  Structural Design Against Deflection

beams. Many physical measures have been developed based on the three rules
of thumb to design structures and structural members against deflection.
Equation 1.15 may relate to a single member or structural element as it is about
the bending of a beam, but its application can have wider significance to engi-
neering practice. It is considered that a similar equation at a whole-structure level
would have even wider implementation in structural design against deflection.

1.5 Rules of Thumb for Design

There are “rules of thumb” for designing structural elements such as beams,
columns and floors, which are simple and effective [1.1]. Such rules are famil-
iar to most engineers and are widely used to develop quick preliminary designs.
For example, for a given span and loading, they can quickly and sufficiently
accurately indicate the required cross-section of a beam or the thickness of a
reinforced concrete floor without calculation. These rules of thumb help not
only speed up preliminary designs but also avoid mistakes.
The development of such rules of thumb can be illustrated by Figure 1.11.
The rules of thumb are summarised from or can be abstracted from, 1) sound
engineering practice and 2) subsequently based on, or checked by, theory. The
rules can then be 3) used by many engineers in their design of structural ele-
ments such as beams, columns, walls and floors.
The usefulness of these rules of thumb and their development poses a ques-
tion: are there other types of “rules of thumb” that can be used for designing
whole structures to achieve smaller deflections, or for making structures more
effective, efficient and even more elegant? For determining such rules, a logical
way is to identify them through studying highly praised structures and reading
the books written by the most eminent architects and engineers. After doing so
it is possible to gain a higher level of appreciation of these structures, greater
admiration from the heart of the creative designs and a deeper philosophical
thinking of the relationships between form and function, between architecture
and structures, and between art and technology, etc. However, it is very diffi-
cult to find such rules of thumb explicitly expressed by the great engineers, and
some general rules cannot be abstracted from their designs or books, which
could be passed on to others for use in the design of different structures.
An alternative way to find such rules of thumb for whole structures can be
developed from theory as illustrated in Figure 1.12.

Figure 1.11 Development of rules of thumb to design structural elements.

 Introduction 21

Figure 1.12 Development of structural concepts (rules of thumb) to design whole

structures.
The flow chart in Figure 1.12 can be explained as follows:

1. Start to identify the concepts for a whole structure based on theory in cur-
rent textbooks, which are significant for practical application in structural
engineering. Here the words “structural concepts” are used instead of rules
of thumb because they are general rather than specific, and can be used for
designing many structures, and because particular physical measures need
to be developed based on these structural concepts. Four structural con-
cepts based on the relationships between deflection and internal forces of
a structure are identified in Chapter 2. They can be intuitively interpreted
and expressed in a concise and memorable manner as follows:
• The more direct the internal force paths, the smaller the deflection.
• The smaller the internal forces, the smaller the deflection.
• The more uniform the distribution of internal forces, the smaller the
deflection.
• The more the bending moments being converted into axial forces, the
smaller the deflection.
2. A number of hand calculation problems are examined with and without
using one of the structural concepts. The outcomes demonstrate that the
four structural concepts are both effective and efficient. Having used these
structural concepts to examine several structures designed by well-known
engineers and architects, it is fascinating to note that these structural con-
cepts have been actually embedded in their designs! This explains why
these structures are excellent from the structural point of view and it is
observed that a structure is likely to be effective (smaller deflections), effi-
cient (using less material) and elegant (architecturally pleasing) when one
or more of the four structural concepts has been used.
3. It is hoped that these structural concepts, like some widely used rules of
thumb for designing structural elements, can be used by many architects
and engineers for designing structures against deflection and for achieving
more effective, efficient and elegant designs.

1.6 Effectiveness, Efficiency and Elegance

In The Structural Engineer, the journal of the UK Institution of Structural
Engineers, there was a definition of Structural Engineering on the contents
page as follows [1.13].
22  Structural Design Against Deflection

Structural engineering is the science and art of designing and making, with
economy and elegance, buildings, bridges, frameworks, and other similar struc-
tures so that they can safely resist the forces to which they may be subjected.

There are three key factors in the statement: safety, economy and elegance
that can be seen as the objectives to be achieved in design and construction.
The discipline of structural engineering allows structures to be produced with
satisfactory performance at competitive costs. Elegance, which is not particu-
larly related to safety and economy, is normally considered by architects.
For the purpose of this book, which focuses on the relationships between
deflection and internal forces in structures, there is a need to scale down and
revise the three objectives as effectiveness, efficiency and elegance. In general
effectiveness means that a structure should satisfy all the functional require-
ments, such as those for deflection, stress and usage of the structure. Here
effectiveness will be limited to deflection. If smaller maximum deflections are
achieved in a design, it is likely, as discussed in Section 1.1, that the structure
will have better buckling capacity, a higher fundamental natural frequency
and smaller internal forces. Therefore, it can be said that this design is more
effective than a similar design with a larger maximum deflection. Efficiency
indicates the use of material in a design. When a structure is able meet the func-
tional requirements using less material, it is said that this structure or design
is more efficient than a similar one using more material. Elegance describes
the pleasing and graceful visual appearance of the structure, which is perhaps
somewhat subjective. Elegance here is considered to be structural elegance
which results from structural correctness. For such definitions, the relative
effectiveness and efficiency of two or more similar structures can be quantified.
The beauty and inspirational features of the four structural concepts to be
studied in this book lies in that the effectiveness, efficiency and elegance of
a structure are integrated as a whole. When one of the four concepts can be
embedded into a design to make the structure more effective and efficient, it is
likely that the structure will naturally become more elegant without purposely
pursuing these aims. This point is demonstrated through a number of practical
examples in Chapters 3 to 6.

1.7 Organisation of Contents
This book consists of seven chapters. The connections between the seven
­chapters are illustrated in Figure 1.13.
This chapter has provided an overview of the topic and the thoughts used to
develop the contents of the book. Intuitive interpretation is emphasised in this
book as it is an effective tool, and a skill, for reaching a higher level of under-
standing of structures, and this is further demonstrated in the later chapters.
Chapter 2 illustrates the theoretical background of the four structural con-
cepts in an intuitive manner to enable the reader to gain a thorough under-
standing. The new meaning of the virtual work principle is explored and a
basic equation, connecting deflection and internal forces of a whole structure, is
 Introduction 23

Figure 1.13 T he connections between the chapters of this book.

Figure 1.14  Connections between sections in Chapters 3–6.

intuitively interpreted. This leads to a new understanding of the four structural

concepts on the relationships between smaller deflection and desirable distri-
butions of internal forces, which provides a strong theoretical basis for wide
and wise applications to be illustrated in the following chapters.
Chapters 3–6 are presented in a similar format and each of the chapters focuses
on demonstrating the effectiveness and efficiency of one of the four structural
concepts. Each of the chapters consists of three parts: 1) the routes to implemen-
tation of the structural concept in design, which are presented by physical and
conceptual measures; 2) two hand calculation examples are provided which are
abstracted from practical problems. Each example contains at least two similar
cases, one with and one without involving an implementation measure based
on the concept, by which the effect of the measure can be clearly identified and
quantified; 3) several practical examples in which the implementation measures
are effectively used are examined for demonstrating the application of the struc-
tural concept and its effects. The relationship between the three parts is shown in
Figure 1.14. The detailed hand calculations will demonstrate the way of analysis
24  Structural Design Against Deflection

for understanding and quantify the effectiveness and efficiency of the structural
concept, and the findings will also serve for the comprehension of the related
practical examples. The practical example will help the reader to realise how the
concept has been used for the solution of challenging problems and for achiev-
ing more effective, efficient and elegant structures.
Chapter 7 provides concluding remarks and further discusses the use of the
four concepts.

References
1.1 Schollar, T. Structural Sizing: Rules of Thumb, AJ, 1989.
1.2 Institution of Structural Engineers. Dynamic Performance Requirements for Per-
manent Grandstands Subject to Crowd Action: Recommendations for Manage-
ment, Design and Assessment, The Institution of Structural Engineers, London,
2008.
1.3 Ellis, B. R. and Ji, T. BRE Digest 426: The Response of Structures to Dynamic
Crowd Loads, Building Research Establishment Ltd., Watford, 2004.
1.4 Smith, A. L., Hicks, S. J. and Devine, P. J. Design of Floors for Vibration: A New
Approach, The Steel Construction Institute, P354, Ascot, 2007.
1.5 Ji, T., Bell, A. J. and Ellis, B. R. Understanding and Using Structural Concepts,
Second Edition, CRC Press, London, 2016.
1.6 Gere, J. M. Mechanics of Materials, Thomson Books/Cole, Belmont, 2004.
1.7 Hibbeler, R. C. Mechanics of Materials, Sixth Edition, Prentice-Hall Inc., Singa-
pore, 2005.
1.8 Torroja, E. The Structures of Eduardo Torroja: An Autobiography of an Engi-
neering Accomplishment, F W Dodge Corporation, USA, 1958.
1.9 Ji, T. and Bell, A. J. Can Intuitive Interpretation Be Taught in Structural Engineer-
ing Education? IV International Conference on Structural Engineering Education:
Structural Engineering Education Without Borders, 20–22 June 2018, Madrid,
Spain.
1.10 Ellis, B. R. and Ji, T. Human—Structure Interaction in Vertical Vibrations, Struc-
tures and Buildings, the Proceedings of Civil Engineers, 122(1), 1–9, 1997.
1.11 Meriam, J. L. and Kraige, L. G. Engineering Mechanics, Vol. 2: Dynamics, Fourth
Edition, John Wiley & Sons, New York, 1998.
1.12 Ji, T. and Cunningham, L. S. An Insight into Structural Design Against Deflection,
Structures, 15, 349–354, 2018.
1.13 The Institution of Structural Engineers. The Structural Engineer, 72(3), 1994.
Chapter 2

Deflections and Internal Forces

2.1 Deflection of a Structure
Equation 1.15 provides an explicit expression for the deflection of a beam
at the structural element level, but its application extends far beyond struc-
tural elements. It is logical to examine similar equations which apply at the
whole-structure level so that fundamental equations can be harnessed for more
advanced design of structures against deflection.
At the structure level, the maximum deflections of any pin-connected struc-
ture and rigid frame structure with s members are shown in equations below
[2.1, 2.2]:
s
N i N i Li
 max   (2.1)
i 1 Ei Ai

Li

 max  
s

0
M i (x )M i (x )dx
(2.2)
i 1 Ei I i

where N i is the axial force in the ith member induced by the actual loads and
N i is the axial force in the ith member induced by a unit load applied at the
critical point (location and direction) where the maximum deflection is likely
to occur; Mi(x) and M i (x ), similar to N i and N i , are the bending moments
in the ith member induced by the actual loads and by a unit load applied at
the critical point respectively. Li, Ei, Ai and Ii (i = 1, 2, . . ., s) are the length,
elastic modulus, area and second moment of area of the cross-section of the
ith member.
Equations 2.1 and 2.2 provide a method for calculating the deflection of any
framework structure with pinned or rigid connections. Equation 2.1 is suit-
able for trusses, scaffoldings and lattice structures, and has a history of over
150 years [2.3]. However, equation 2.1 has not been emphasised in textbooks
on Mechanics of Materials and Structural Analysis to the same extent. This is
because the use of the equation requires the calculation of the internal forces
N i and N i to determine deflection, and such a calculation may be regarded as
too tedious for structures with many members, or for statically indeterminate
26  Structural Design Against Deflection

structures. Normally very simple statically determinate plane structures are

provided in textbooks to show how equation 2.1 is used to calculate deflection.
Similarly, equation 2.2 is used to calculate deflections of beams and simple
frames.
Unlike equation 1.15, it is not obvious how to interpret equations 2.1 and
2.2 in a simple manner and to identify the physical essence embedded in the
two equations. This is because N i and M i (x ) (i = 1, 2, . . . , s) are functions of
the loading that can have many variations and because equations 2.1 and 2.2
contain many items (i.e. a structure has many members).
In comparison with equation 1.15, the understanding and implementation
of equations 2.1 and 2.2 for reducing the maximum deflection of a whole
structure are not well known. Based on previous work [2.4–2.6], this chapter
provides a theoretical basis to reveal the physical essence between the maxi-
mum deflection and the internal forces of a structure. The intuitive interpreta-
tion of the principle of virtual work will provide four fundamental structural
concepts that are general, simple to understand, and are useful for practical
applications.

2.2 Internal Forces, Deflections and Energies

of Two Rods
The basic understanding of theory can often be established by studying simple
cases. Figure 2.1a shows two linear elastic rods that have the same modulus E
and the same length L but different cross-sectional areas, Aa and Ab where
Ab > Aa with Ab   Aa (   1 ) [2.2]. The thin and thick rods are subjected
to two pairs of forces, Pa and Pb applied at their ends. Examine the relation-
ships between the deflections, internal forces and elastic stain energies stored
in the two rods for two loading conditions: 1) when the internal forces of the
two rods are the same; and 2) when the total elongations of the two rods are
the same.


(a) (b)

Figure 2.1 Two rods with different cross-sections undergoing axial deformation.

(a) Two robs subjected to external axial loads. (b) Free body diagrams to
reveal internal forces.
 Deflections and Internal Forces 27

Using free-body diagrams (Figure 2.1b) and equilibrium equations, it is easy

to determine that the internal forces in the two rods are equal to the external
forces applied on them, N a = Pa and N b = Pb .
The force-deflection relations for the two rods are:

N aL EAa
a  or N a   a  ka a (2.3a)
EAa L

NbL EAb
b  or N b   b  kb  b (2.3b)
EAb L

where ka = EAa / L and kb = EAb / L are the axial stiffnesses of the two rods,
and indicate the structural ability of the rods to resist axial deformation. The
strain energies of the rods are respectively:

1 1 1 1
Ua  ka a2  N a a and Ub  kb  b2  N b  b  (2.4a) and (2.4b)
2 2 2 2

Equation 2.3 indicates that larger internal force will lead to larger deflec-
tion while equation 2.4 shows that larger deflection will lead to larger strain
energy. These statements come from the very simple case of a uniform rod, but
they are applicable to more complex cases, even to whole structures.
The ratios of the two deflections in equation 2.3 and of the two energies in
equation 2.4 are:

 b N b L EAa A N 1 Nb
   a b  (2.5)
 a EAb N a L Ab N a  N a

Ub  2 1 N b2
  b2  (2.6)
Ua  a  N a2

When the internal forces are the same for the two rods, i.e. N a = N b, it can
be observed from equations 2.5 and 2.6 that  b   a and Ub < U a as   1.
When the two rods are subjected to the same internal force, the thick rod has
a smaller deflection and stores less energy than the thin rod.
When the total deflections are the same for the two rods, i.e.  a   b , it can
also be seen from equations 2.5 and 2.6 that N b > N a and Ub > U a . This indi-
cates that when the two rods are made to deflect the same amount, the thick
rod will experience larger internal forces and will store more strain energy
than the thin rod.
To demonstrate an implication of the last statement, the two rods are now
used to support a weightless rigid plate which in turn supports a concentrated
vertical load of P. To create a symmetric problem, the central rod has a cross-
section area Ab and two side rods have cross-section areas of Aa / 2 (replacing
28  Structural Design Against Deflection

the original single rod Aa ) as shown in Figure 2.2a. Using the free-body dia-
gram shown in Figure 2.2b, the equilibrium equation and the force-deflection
equation give:

EAa EAb
P  Na  Nb  (  )  (ka  kb )(2.7)
L L

where ∆ is the vertical deflection of the rods.

Equation 2.7 indicates that the stiffer member shares or attracts a larger
portion of the load for the same deflection, or the internal forces in the mem-
bers are proportional to their axial stiffnesses. This statement is derived from
a simple axial compression problem, but it is applicable to more complex situ-
ations. For example, a weightless rigid plate is supported by four columns and
is subjected to a concentrated lateral load as shown in Figure 2.3a. The four
columns have the same height of L and the same elastic modulus E but differ-
ent second moments of areas, Ia , Ib , Ic and Id . This is a bending problem,
and the four columns experience the same amount of lateral deflection. The

(a) (b)
Figure 2.2 A  compression problem. (a) A structure. (b) Free body diagram.

(a) (b)
Figure 2.3 A  bending problem. (a) A structure. (b) Free body diagram.
 Deflections and Internal Forces 29

free-body diagram of the rigid plate is shown in Figure 2.3b and the equilib-
rium equation for the rigid plate is:

12EI 12EIb 12EIc 12EId

P  Qa  Qb  Qc  Qd  ( 3 a    )
L L3 L3 L3 (2.8)
 (ka  kb  kc  kd )

=
where ki 12 = EI i / L3 (i a, b, c , d ) . Equation 2.8 has a similar pattern to equa-
tion 2.7, and therefore the observation, or conclusion, from equation 2.7 is appli-
cable to bending problems as described by equation 2.8. The force transmission
from loading positions to structural supports can be seen as a force flow through
the structural members to the supports, with the stiffer members attracting a
larger force flow. A particular case is considered that I= a I=
b Ic and Id = 2Ia,
According to equation 2.8, the three left-hand columns attract 0.2P each and the
right-hand column attracts 0.4P. The result indicates that force flows more to the
stiffer parts of the structure and the force flow can thus be guided through design.

2.3 I nternal Forces, Deflection and Energy

of a Structure
It is of practical importance to know the position of the critical point at
which the maximum deflection of a structure is likely to occur. To identify
such a point, a unit load can be placed in the appropriate direction at every
nodal point in turn and its corresponding deflection calculated, leading to an
array of nodal deflections. The nodal point at which the maximum value in
the array corresponds to is the critical point. According to this definition, the
critical point of a structure is independent of the loading on the structure.
Normally, the critical point of a structure can be intuitively identified with-
out calculations. For example, the critical point of a cantilever is at its free
end and the critical point of a simply supported plate is at its centre. For the
particular case in Figure 2.4, node C is the critical point of the truss in the
vertical direction.
Consider a truss structure that consists of s members and n degrees of
freedom which is subjected to two sets of loading as shown in Figure 2.4.
All members of the truss have the same elastic modulus E. Load case 1,
shown in Figure 2.4a, is the actual loading, {P1 }, and Load case 2, shown in
­Figure 2.4(b), is a specific loading case {P2 } in which a unit concentrated load
is applied at the critical point C of the structure.
Analysing the two truss structures leads to two sets of results in which sub-
scripts 1 and 2 respectively relate to the Load case numbers.
Load case 1: There are external and internal forces {P1 } and {N 1 }, the
nodal deflections are {∆1 } and the member elongations are {δ1 }. The rela-
tionship between the internal force and the elongation of the jth member is
1, j  N 1, j Lj / EAj, where Lj and Aj are the length and the cross-sectional area
of the jth member.
30  Structural Design Against Deflection


(a) (b)

Figure 2.4 Two sets of loading on the same truss structure (a) Load case 1; (b) Load
case 2.

Load case 2: There are similar quantities, {P2 } , {N 2 } , {∆ 2 } and {δ 2 } , and the
relationship  2, j  N 2, j Lj / EAj.
For a conservative system, the work-energy principle states that if the stresses
in a body do not exceed the elastic limit, all the work done on a body by exter-
nal forces is equal to the elastic strain energy stored in the body [2.2], which
can be expressed for Case 1 as:

2
1 n 1 s 1 s N 1, j Lj
W1,1   1,i 1,i 2 
2 i 1
P  
j 1
N 
1, j 1, j  
2 j 1 EAj
(2.9)

where W1,1 is the external work done by the loads {P1 } on the deflections {∆1 }
induced by {P1 } and the right-hand side of equation 2.9 is the elastic energy
stored in the s members.
Consider the work done by the loads {P1 } in Case 1 moving through the
deflections {∆ 2 } resulting from the loads in Case 2, and the strain energy cre-
ated by the internal forces {N 1 } in Case 1 acting on the member elongations
{δ 2 } in Case 2. This leads to:

1 n 1 s 1 s N 1, j N 2, j Lj
W1, 2  
2 i 1
P1, i  2, i   N 1, j  2, j  
2 j 1 2 j 1 EAj
(2.10)

Similarly, the work done by the loads {P2 } moving through the deflections {∆1 }
and the strain energy contributed by the internal forces {N 2 } on the member
elongations {δ1 } are:

1 n 1 s 1 s N 2, j N 1, j Lj
W2,1   2,i 1,i 2 
2 i 1
P  
j 1
N 
2 , j 1, j   EA (2.11)
2 j 1 j

It can be observed that the right-hand side items in equations 2.10 and 2.11 are
the same, which leads to:

W1, 2 = W2,1 (2.12)

 Deflections and Internal Forces 31

This is the reciprocal theorem of work. It states that the work done by the
loads in Case 1 moving through the deflections resulting from the loads in
Case 2 is equal to the work done by the loads in Case 2 moving through the
deflections induced by the loads in Case 1.
As only a unit load is applied at node C in Case 2, the external work in equa-
tion 2.11 becomes:

1 n 1
W2,1  
2 i 1
P2, i 1, i  (1  1,C ) (2.13)
2

Substituting equation 2.13 into equation 2.11 and simplifying gives:

s N 2, j N 1, j Lj
1,C   (2.14)
j 1 EAj

Equation 2.14 provides a method for calculating the deflection at node C of the

structure resulting from the loads in Case 1 (Figure 2.4a) in three steps:

1. Calculate the internal forces {N 1 } resulting from {P1 } , which are the actual
loads on the structure.
2. Calculate the internal forces {N 2 } resulting from the unit load {P2 } .
3. Use equation 2.14 to calculate the vertical deflection at node C.

It is noted that the calculations require to determine {N 1 } and {N 2 }, which

can be challenging and tedious if a truss structure is statically indeterminate or
if it has many members. Hence, equation 2.14 is seldom used to calculate the
deflections of actual truss structures.
It is logical to examine W2, 2 after examining W1,1, W1, 2 and W2,1. W2, 2 can
be expressed as:

2
1 1 s N 2, j Lj
W2 , 2  (1   2,C )   (2.15)
2 2 j 1 EAj

This is a similar equation for W1,1, but W2, 2 means the work done by {P2 } mov-
ing through the deflection {∆ 2 } , i.e. by a unit force P2,C = 1 moving through
the deflection ∆ 2,C . Simplifying Equation 2.15 gives:

s N 22, j Lj
 2,C   (2.16)
j 1 EAj

Before discussing the physical meaning of equation 2.16, a beam type of struc-

ture is considered in which bending moments are the major internal forces and
to which similar equations to those for truss structures apply. If the pinned
32  Structural Design Against Deflection

connections of the truss in Figure 2.4 are all changed to rigid connections, it

becomes a frame structure. The two equations for calculating the deflections
due to the actual loads and the deflections due to the unit load, considering
bending moments alone, can be written as:
Lj
1 n 1 1 s  M 2, j (x )M 1, j (x )dx
W21   P2, i 1, i  (1  1,C )   0 (2.17)
2 i 1 2 2 j 1 EI j

Lj
1 s 0 M 2, j (x )dx
2
1 n 1
W22   P2, i  2, i  (1   2,C )   (2.18)
2 i 1 2 2 j 1 EI j

where M 1, j (x ) and M 2, j (x ) are the bending moments along the jth member
induced by Load cases 1 and 2 respectively. Similar to equation 2.14, equa-
tion 2.17 can be used to calculate the deflection of a frame structure. The
Lj
integration ∫ M 22, j (x )dx for the jth member in equation 2.18 means the area
0
under the curve for M 22, j (x ) between 0 and Lj , which can also be represented
by the same area of an equivalent rectangle with a length Lj and a mean height
M 22, i. Equation 2.18 can therefore also be stated as:

s M 22, j Lj
 2,C   (2.19)
j 1 EI j

Equations 2.16 and 2.19 have a similar form involving an internal force

squared with one equation applicable to truss structures and one applicable to
beam and frame structures. The physical meaning of ∆ 2, C in the two equations
will be examined in the next section.

2.4 Physical Meaning of ∆ 2, C
Consider the most unfavorable loading scenario that all the loads on a
structure are lumped at the critical point. This leads to the largest deflec-
tion at the point of those induced by all possible loading distributions. For
example, all the loads acting on the truss in Figure 2.4a are moved to and
lumped at point C, the vertical deflection at C due to the lumped loads
will be larger than those induced by any other loading distributions. If this
lumped load is then normalised to a unit load, which is not a true load-
ing condition, but is the worst load case for the maximum deflection of a
structure, equations 2.16 or 2.19 can be used to calculate the normalised
maximum deflections of different types of truss and frame structures. There-
fore ∆ 2,C means the maximum deflection under the most unfavourable
loading scenario in which all the loads on a structure are lumped at the
critical point and normalised to a unit.
 Deflections and Internal Forces 33

The flexibility coefficient at a point of a structure is defined as the deflection

induced by a unit load in the loading direction. Therefore ∆ 2,C (equation 2.19)
is the flexibility coefficient at the critical point of the structure and has the larg-
est value among all flexibility coefficients for any truss or frame structure. This
interpretation can be demonstrated mathematically.
Considering a structure that is modelled by s elements and n nodes with each
node having d degrees of freedom. The static equilibrium equation, containing
n × d unknowns, is expressed as:

[K ]{U } = {P}  (2.20)

where {U} is the nodal displacement vector to be determined, {P} is the load
vector and [K] is the stiffness matrix that includes the effect of the boundary
conditions. Equation 2.20 is a general equation of equilibrium and is suitable
for any linear elastic structure.
When a single unit load is applied at the critical node C in a given direction,
l, the lth degree of freedom of a node, the load vector is:

{P}  {0, 0, 1, , 0, 0}T  (2.21)

Substituting equation 2.21 into equation 2.20 and solving leads to the deflec-

tions [2.5]:

 u1   1,1 L 1,cl L 1, n  0  1,cl 

 M   M O M N M   M   M 
      
 L  cl ,cl L  cl , n  1   cl ,cl  (2.22)
 ucl   [K ] {P}  [ ]{P }   cl ,1
1

 M   M N M O M   M   M 
  
und   n,1 L  n,cl L  n, n  0  n,cl 

where [δ ] is the flexibility matrix of the structure (the inverse of the stiff-
ness matrix), and δ cl ,cl is the diagonal element at row cl and column cl in the
flexibility matrix, in which cl means the critical degree of freedom and can
be determined by cl  (c  1)  d  l where c is the node number of the critical
point C, d is the degrees of freedom of each node and l is the number of the
concerned degree of freedom. Considering the row cl in equation 2.22 gives:

 2,C  ucl   cl ,cl  (2.23)

Equation 2.23 states that the deflection at the critical point ucl (∆ 2,C ) induced
by a unit load at this degree of freedom is the coefficient δ cl ,cl in the flexibility
matrix of the structure.
The stiffness matrix [K] gives a detailed description of the distribution
of structural members and their contribution to the stiffness matrix. How-
ever, from [K] it is difficult to sense how stiff the structure is. In practice, a
34  Structural Design Against Deflection

unique value is preferred to define the stiffness of a structure. It is common

for the inverse of the deflection at the critical point induced by a unit load is
defined as the static stiffness of the structure [2.6], i.e.:

1
KS = (2.24)
ucl

For example, when a unit vertical downward load is applied at the free end of
a cantilever, the vertical deflection at the free end is L3 / (3EI ) , the static stiff-
ness of the cantilever is then (3EI ) / L3 . Equations 2.23 and 2.24 give the static
stiffness of the structure as:

1
KS   (2.25)
 cl ,cl

Equation 2.25 indicates that the static stiffness of a structure is the inverse of

the largest diagonal element in the flexibility matrix of the structure.
In summary, the physical meaning of ∆ 2,C (the deflection at the critical
point of a structure due to a unit load applied at this point) is the inverse of
the static stiffness of a structure (equation 2.24) and is equal to the largest
flexibility coefficient in the flexibility matrix of the structure (equation 2.23).
∆ 2,C can also be seen as the largest possible deflection when all the loads are
lumped to the critical point of the structure and scaled to a unit load.

2.5 Intuitive Interpretation
After examining the physical meaning of the left sides of equations 2.16 and
2.19, it is possible to interpret the right sides of the two equations. As the
loading has been lumped at the critical point of the structure and normalised
to a unit load, the internal forces in equations 2.16 and 2.19 are independent
of any particular loading but are functions of structural form. For statically
indeterminate structures they are also functions of material and cross-sectional
properties. For the purpose of design, it is ideal to make the deflections of a
structure as small as possible, or the static stiffness of the structure as large as
possible, using the same amount of material or less material, i.e.:
s N 2
1 2 , j Lj
  2,C    min (2.16)
KS j 1 EAj

s M 2
1 2 , j Lj
  2,C    min (2.19)
KS j 1 EI j

Finding the minimum deflection at the critical point or the largest static stiff-
ness of a structure may be considered as a topology optimisation problem. For
one type of topology optimisation [2.7], the geometry of a structure is altered
 Deflections and Internal Forces 35

gradually by removing the element with the smallest stress or adding an ele-
ment where the stress demand is high. This iterative process seeks to make
the distribution of stress as uniform as possible and eventually leads to an
optimum topology design based on the objective function, a stiffer structure.
Equation 2.16 or equation 2.19 forms an incompletely defined optimisation
problem, and therefore standard optimisation techniques may not be directly
applicable at this stage. However, the physical essence of the incomplete opti-
misation problem can still be identified and interpreted.
As internal forces and structural form are closely related, the internal forces
can be examined directly using equations 2.16 and 2.19, instead of considering
the topology of the structure. The physical quantities in equations 2.16 and
2.19 have the following mathematical characteristics:

1. E > 0; Aj > 0; I j > 0 and Lj > 0;

2. N 22, j ≥ 0 and M 22, j ≥ 0

regardless of whether the member is in tension or in compression or whether

the bending moment is positive or negative.
All the items in equations 2.16 and 2.19 are positive or zero, i.e. there are
no negative terms. When Aj / Lj in equation 2.16 and I j / Lj in equation 2.19 do
not change significantly, the internal forces dominate the deflections in the
two equations. The relationships between the smaller deflections and internal
forces embedded in the two equations can be interpreted intuitively as follows:

1. One way to make the deflection as small as possible is to have as many

terms as possible equal to zero on the right sides of equations 2.16 and
2.19. Mathematically, the fewer the positive terms, the smaller the sum
of all the terms. Physically, many zero terms means that these members
are zero-force members. The unit load positioned at the critical point is
transmitted to the supports of the structure without passing through these
zero-force members and takes a shorter internal force path. The greater
the number of zero-force members, the more direct the internal force path.
This physical phenomenon suggests that shorter or more direct internal
force paths from the load to the structural supports lead to smaller
deflection of a structure.
2. It can be directly observed from equations 2.16 and 2.19 that the sums will
be smaller if each of the terms becomes smaller. The corresponding physi-
cal phenomenon is that smaller internal forces lead to smaller deflections
of a structure.
3. Consider three sets of data, each consisting of five numbers as shown in
Table 2.1. The sums of the three sets of data are the same, but the largest
differences between the five numbers in the three data sets are different.
Consequently, the sums of the squares of the data in the three sets are dif-
ferent. It can be observed that the larger the difference of the five numbers,
the larger the sum of the squares.
36  Structural Design Against Deflection

Table 2.1 Comparison of Three Sets of Data

Data set Five data Largest difference 5 5

in the f ive data ∑a

i =1
i ∑a
i =1
5
i

1 1, 2, 3, 4, 5 4 15 55
2 2, 2, 3. 4, 4, 2 15 49
3 3, 3, 3, 3, 3 0 15 45

Due to the similarity between the right-hand side of equation 2.16 or 2.19 and
5

 a , the observation from the simple comparison in Table 2.1 is applicable

i 1
2
i

to equations 2.16 and 2.19. Smaller differences between the internal forces will
lead to a smaller sum of squares than those with larger differences. This can be
interpreted physically as: more uniformly distributed internal forces result in
smaller deflections in a structure.
In summary, there are three ways to achieve smaller deflections by actively
achieving desirable internal forces and force distributions. They can be pre-
sented in a more memorable way as follows:

1. The more direct the internal force paths, the smaller the deflection of a
structure;
2. The smaller the internal forces, the smaller the deflection of a structure;
3. The more uniform the distribution of internal forces, the smaller the
deflection of a structure.

2.6 Deflections due to Bending Moment, Axial and

Shear Forces
The previous interpretation of how to achieve smaller deflections is based on
equations 2.16 and 2.19, which are based on either axial forces or bending
moments. It is however possible that members can be subjected to bending
moment, axial force and shear force simultaneously. For structures containing
such members the deflection of a structure is expressed as:
Li

C  
s

0
M i (x )M i (x )dx s

N i (x )N i (x )Li s
 i
Q (x )Qi (x )Li
(2.26)
i 1 Ei I i i 1 Ei Ai i 1 Gi Ai

The deflections contributed by bending, axial and shear effects can be illus-
trated by an example. Consider a quarter of a circular ring with a radius of
R, one fixed end, and a free end, as shown in Figure 2.5. The curved member
has a uniform rectangular cross-section with width b and height h and mate-
rial properties of E and G = 0.5E. A unit downward load is applied at the free
end of the member. Determine the vertical deflections at the free end of the
member.
 Deflections and Internal Forces 37


(a) (b)
Figure 2.5 A  quarter ring subjected to a concentrated load. (a) The ring structure.
(b) Free body diagram.

Consider a free body diagram of part of the ring as shown in Figure 2.5b,

which shows the internal forces at a typical cross-section B, defined by θ. The
internal forces in the member can be determined using three basic equilibrium
equations:

M  M  R sin ; N  N  sin ; Q  Q  cos 

Substituting the internal forces into equation 2.26 and noting that dl  Rd ,
gives:
Li Li Li


 0
M (x )M (x )dl

 0
N (x )N (x )dl


0
Q(x )Q(x )dl
EI EA GA
R3  /2 2 R  /2 2 R  /2
EI 0 EA 0 GA 0
= sin  d  sin  d  cos2  d

 R3  R R
=  
4EI 4EA 4GA

Now substituting G = 0.5E and A = 12I / h 2 into the previous equation, the
vertical deflection at the free end is:
2 2
 R3 1 h 1 h 
 [1       ](2.27)
4EI 12  R  6  R 

The terms in the square bracket show the relative contributions to the deflection
from bending moment, axial force and shear force. The relative contributions
38  Structural Design Against Deflection

can be quantified for three values of the ratio of the cross-section height to the
radius of the curved member as follow:

h  R3 1 1
For = 10   [1   ]
R 4EI 1200 600

h  R3 1 1
For =5  [1   ]
R 4EI 300 150

h  R3 1 1
For = 2 .5   [1   ]
R 4EI 75 37.5

The contributions from the axial and shear actions are very small in compari-
son to that arising from bending. When the dimensional sizes of a member
are significantly larger than its cross-sectional sizes, the deflections induced
by axial and shear actions in a bending problem are very small and can be
neglected.
For a structure subjected to bending and axial actions with f members sub-
jected to bending and g members subjected to axial force, the deflection of the
structure can be determined from equations 2.16 and 2.19 as:

f M 22, j Lj g
N 22, j Lj
 2,C    (2.28)
j 1 Ej I j j 1 Ej A j

As the deflection induced by bending action is much larger than that induced
by axial force action, equation 2.28 implies another way to reduce the deflec-
tion by converting bending moment actions into axial force actions, for exam-
ple by replacing bending members with axial force members and/or by adding
bar members to reduce bending members in a structure.
This can be presented as the fourth structural concept to achieve smaller
deflection as follows:

4. The more the bending moments are converted into axial forces, the smaller
the deflection of a structure.

It is well understood that structures will become more efficient when loads
are transmitted through axial forces rather than bending moments. One of
the reasons is to achieve the efficiency of materials, which relates to the stress
distributions on the cross-sections of members, i.e. a uniform distribution for
axial forces and a linear distribution for bending moment. The fourth struc-
tural concept is particularly related to deflection of a structure and indicates
the deflection induced by bending moments will be much larger than that by
axial forces.
 Deflections and Internal Forces 39

2.7 Characteristics of the Structural Concepts

2.7.1  The Four Structural Concepts

The four structural concepts intuitively interpreted from equations 2.16,
2.19 and 2.28 are simple, meaningful, fundamental and general, and they are
related to the deflections and internal forces of a whole structure that can be
any type of truss and/or frame structure. These four structural concepts can be
summarised in a more concise and memorable manner and treated as rules of
thumb as follows:

1. The more direct the internal force paths, the smaller the deflection.
2. The smaller the internal forces, the smaller the deflection.
3. The more uniform the distribution of internal forces, the smaller the
deflection.
4. The more the bending moments are converted into axial forces, the smaller
the deflection.

In these statements, the form of a structure is not explicitly stated but is embed-
ded. It has been shown in Section 1.2 that structural form, deflection and inter-
nal forces are closely related so that altering any one of the three will lead to a
change of the other two. The four structural concepts provide a solid basis for
creative applications. They will be examined and discussed further to gain a
sound and thorough understanding.

2.7.2 Generality
Equations 2.16 and 2.19 are derived from the principle of virtual work and are
general and applicable to all types of truss and frame structures and include
the structural concepts derived from equation 1.15 which are based on beam
theory.
The maximum bending moment of a uniform beam subjected to a uniformly
distributed load is:

M max   qL2 (2.29)

For a simply supported beam,   1 / 8, and for a cantilever,   1 / 2. Sub-

stituting equation 2.29 into equation 1.15, the deflection can be alternatively
expressed as:

qL4 M2
 max     2 max (2.30)
EI  qEI

Equation 2.30 states that the maximum deflection is proportional to the

maximum bending moment squared or in more general terms, the smaller
40  Structural Design Against Deflection

the internal forces, the smaller the deflection, which is the second structural
concept. This demonstrates that the four structural concepts derived using the
principle of virtual work for a whole structure include the basics developed
from beam theory.

2.7.3 Interchangeability
The first three structural concepts are abstracted from the same equations
(equations 2.16 and 2.19), which means that these structural concepts are not
independent and are exchangeable, i.e. if a structure reaches a state with a
more direct internal force path, it is likely that the structure will have smaller
internal forces and a more uniform distribution of internal forces. This can be
illustrated using an example.
Figure 2.6 shows two similar 3-bay and 3-storey truss type structures carry-
ing a unit horizontal load at the top right corner. They have the same dimen-
sions, the same material property, E, and cross-sectional area, A. There are 24
members in each frame and the horizontal and vertical members have the same
length of L. The only difference between the two frames is the arrangement of
the three bracing members. For Frame A in Figure 2.6a, the bracing members
are placed in the right bay and for Frame B, the bracing members are arranged
diagonally across all three bays of the structure. The bracing arrangement in
Frame B can be evolved from that in Frame A by moving the middle bracing
member one panel to its left and the bottom bracing member two panels to the
left. The two frames are statically determinate, and their internal forces can be
easily calculated by hand. For the convenience of the comparison, the non-zero
internal forces are indicated next to the corresponding members of the two
frames in Figure 2.6.
The horizontal deflections at the loading positions of Frames A and B can be
calculated using equation 2.16 as follows:


(a) (b)
Figure 2.6 Two 3-bay and 3-storey frames. (a) Frame A. (b) Frame B.
 Deflections and Internal Forces 41

s N j2 Lj L
A    [4  (1)2  2  (2)2  (3)2  3( 2 )2 2 ]
j 1 EA EA
(2.31)
L (21  6 2 )L 29.49L
= [4  8  9  6 2 ]  
EA EA EA

s N j2 Lj L (3  6 2 )L 11.49L
B    [3  (1)2  3( 2 )2 2 ]   (2.32)
j 1 EA EA EA EA

The ratio of the two deflections is:

 B 11.49
  0.39 (2.33)
 A 29.49

The deflection of Frame B is only 39% of that of Frame A with the same
amounts of material used.
The reasons that the deflection of Frame B is much smaller than that of
Frame A can be explained intuitively using the first three structural concepts. It
is observed from Figure 2.6 that:

• Ten members have internal forces in Frame A while six members have inter-
nal forces in Frame B indicating that Frame B creates more direct internal
force paths to transmit the load to its supports than Frame A (Structural
Concept 1), which leads to a smaller deflection with over 60% reduction.
• The largest force has a magnitude of 3 in Frame A while it is 2 in Frame
B, i.e. Frame B has smaller internal forces than Frame A (Structural con-
cept 2).
• The maximum difference between internal forces is 3  1  2 in Frame
A while the difference is 2  1  0.414 in Frame B. This indicates that
Frame B has a more uniform distribution of the internal forces than Frame
A (Structural concept 3).

It can be observed from this example that the first three structural concepts are
exchangeable. Although any of the three structural concepts can be used for
the design of the bracing patterns, for this particular example achieving a more
direct internal force path is easier than creating smaller internal forces or a
more uniform distribution of internal forces. In other cases, using the second or
third structural concepts may be more convenient than using the first structural
concept. This understanding is useful for design as different approaches can be
followed to achieve smaller deflections.

2.7.4 Compatibility
The first three structural concepts may not be fully compatible as they are
stated from different perspectives based on the same equations. A more direct
42  Structural Design Against Deflection

internal force path requires that more members are in a zero-force state which
may lead to larger internal forces in the other members. On the other hand,
the more uniform distribution of internal forces may imply that more members
share internal forces so that there are no large differences between the internal
forces in individual members. This type of incompatibility can also be demon-
strated using an example.
Two similar 3-bay and 4-storey truss type structures with the same dimen-
sions are shown in Figure 2.7. The horizontal and vertical members have the
same length of L and all members have the same material property E and
cross-sectional area A. Each frame has 32 members including 4 bracing mem-
bers. The only difference between the two frames is the arrangement of the
bracing members in the bottom left panels.

Frame A: The bottom bracing member is placed between nodes B and D and
is linked with the bracing member in the upper storey.
Frame B: The bottom bracing member is linked between nodes A and C and
is parallel to that in the upper storey.

The two frames are statically determinate, and their internal forces can be easily
calculated by hand and the non-zero internal forces are indicated next to the
corresponding members in Figure 2.7. It can be observed from Figure 2.7 that
only nine members are in a non-zero force state in Frame A while eleven mem-
bers are in a non-zero force state in Frame B, indicating that Frame A creates a
more direct internal force path than that in Frame B. However, there are smaller
internal forces and smaller differences between the internal forces in Frame B
than that in Frame A, indicating that Frame B creates a more uniform distribu-
tion of internal forces. Which frame has a smaller deflection? Equation 2.16


(a) (b)
Figure 2.7 C omparison of internal forces of the two simple frames. (a) Frame A.
(b) Frame B.
 Deflections and Internal Forces 43

can be used to determine the deflections of the two frames with the internal
forces indicated in Figure 2.7:

s N j2 Lj L
A    [4  (1)2  (2)2  4( 2 )2 2 ]
j 1 EA EA (2.34)
(8  8 2 )L 19.31L 
 
EA EA

s N j2 Lj L (7  8 2 )L 18.31L
B    [7  (1)2  4( 2 )2 2 ]   (2.35)
j 1 EA EA EA EA

It can be seen that Frame B has a smaller deflection than Frame A, although
it has less direct internal force paths than Frame A. Comparing the internal
forces in Frames A and B (Figure 2.7), only three members, AB, BC and CD,
have different internal forces, which makes the difference between the calcu-
lated horizontal deflections. For Frame A, member AB has an internal force
of 2kN and the two other members have zero-force, while for Frame B, the
three members have the same force magnitude of 1kN. Due to the action of
“square,” 22 > 3 ×12 , i.e. the contribution of the internal force in member
AB in Frame A to the deflection is larger than that from the three members in
Frame B. In this example, the structural concepts of smaller internal forces and
a more uniform distribution of internal forces are even more effective than the
structural concept of a direct internal force path.
This example shows that the first three structural concepts are not fully com-
patible and also tells that there are opportunities for creative use of the struc-
tural concepts.

2.7.5 Reversibility
The presentation of the four structural concepts seems to indicate that a smaller
deflection is the consequence of more direct internal force paths, smaller inter-
nal forces, more uniform distribution of internal forces or converting bending
moments to axial forces. As deflection and internal forces occur at the same
time when a structure is loaded, the structural concepts can also be stated in
reverse as:

1. The smaller the deflection, the more direct the internal force path.
2. The smaller the deflection, the smaller the internal forces.
3. The smaller the deflection, the more uniform the distribution of internal
forces.
4. The smaller the deflection, the more the bending moments are converted
into axial forces.

These reverse statements say that internal forces can be reduced or more uni-
formly distributed when the deflection of a structure can be controlled or
44  Structural Design Against Deflection

Figure 2.8 A dditional supports to a floor.

reduced. Controlling or reducing deflection where possible provides a further

route to reduce internal forces or alter bending action to axial force action. The
reverse statements can be illustrated using an example.
Figure 2.8 shows scaffolding members used to form simple trusses to sup-
port a floor above, which effectively reduces the deflection of the floor. Con-
sequently, part of the internal forces in the floor will pass on through the truss
members to the supports of the truss. This can be interpreted as converting
some of the bending moments in the floor into axial forces in the truss mem-
bers. Therefore, the bending moment in the floor reduces and becomes more
uniform.

2.7.6 Relative Performance
The four structural concepts have been presented in the form of “The more . . .
The smaller. . . ”, which is obviously in a comparative sense. In other words,
the four structural concepts provide an effective way to assess the relative per-
formance of two or more similar forms of a structure for which any of the
structural concepts can be used to achieve smaller deflections of the structure.
Comparing the relative performance of two forms of a structure would
be more appropriate than examining their absolute performances. There are
many sources for introducing errors in the analysis of structures such as inac-
curate modelling. For example, connections may be neither pinned nor rigid,
 Deflections and Internal Forces 45

supports may be between fixed or pinned and material properties may not have
their assumed values. As it is the model of a structure that is actually analysed
rather than the structure itself, the accurate prediction of the behaviour of the
structure is unlikely to be achieved in many cases. However, the relative per-
formance of two similar structure models will remove errors from the analy-
sis and modelling of the structures and allow a more reliable assessment of
their different performances. For example, the two frames shown in Figure 2.6
involved the same degrees of error, possibly generated from the assumed pinned
connections and boundary conditions and the estimated values of the elastic
modulus and the cross-sectional areas of members. The calculated deflections
may not be accurate but the ratio of the two calculated deflections would give
a reliable indication of the relative behaviours of the two frames.
Thus, when evaluating the relative performance of two or more similar struc-
tures or forms of a structure the exact input data may not be necessary, i.e.
the modulus of elasticity, the area and the second moment of area of a cross-
section, loading and even the dimensions of the structures. This effectively
simplifies the analysis while still capturing the physical essence of the problem.
For example, the ratio of the deflections of the two frames in equation 2.33 is
non-dimensional and the physical parameters, E, A and L, together with any
other possible errors arising from assumptions made are cancelled out in the
ratio validating the comparison. It is convenient and effective to analyse the
relative performance of two similar structures. In the next four chapters,
the relative performance of structures in pairs, one involving a structural con-
cept and one not involving a structural concept, will be examined quantita-
tively to demonstrate convincingly the effect of using the structural concepts.

2.8 Implementation
The four structural concepts, interpreted intuitively from the principle of vir-
tual work, provide a sound basis for implementation. This requires the develop-
ment of physical measures to incorporate the benefits of considering structural
concepts into practical cases to create more effective and efficient structures as
has been shown in the previous examples in Figures 1.3, 2.6 and 2.7.
Only four structural concepts for whole structures have been discussed, but
many physical measures can be developed based on these concepts. Many such
physical measures are already being used in practice and there will be fur-
ther measures that can be created to deal with particular cases. For example,
providing a support is an effective way to lead to smaller internal forces in a
structure and thus smaller deflections. Figure 2.9 illustrates four cases demon-
strating different physical measures all serving for providing a support.
Figure 2.9a shows a steel prop used to support the deck of a footbridge. Part
of the bridge loads is transmitted through the compression in the prop to its
foundation. The prop effectively reduces the internal forces and the vertical
deflections of the footbridge. As the flexural stiffness of the prop is not con-
cerned and its axial deformation are negligible, it can be considered as a roller
support to the bridge deck.
46  Structural Design Against Deflection

(a)

(b)

Figure 2.9 Examples of providing a support. (a) Providing a prop. (b) Providing two
inverted triangular trusses to form vertical supports at the centre of a
linking structure. (c) Providing tendons and a wooden bar to form a hori-
zontal support (Courtesy of Mr Jiachen Guo, Beijing Jiaotong University,
China). (d) Providing tendons to form elastic restraints to columns.
 Deflections and Internal Forces 47

(c)

(d)

Figure 2.9 (Continued)
48  Structural Design Against Deflection

Figure 2.9b shows a suspended four-storey link structure between two adja-

cent buildings. Vertical supports are provided to the bottom centre of the link
structure by two inverted triangular trusses (one each side of the link struc-
ture). The horizontal components of the tension forces in the two inclined
members are balanced by the compression in the horizontal member at the
height between the third and fourth storeys of the link structure and the verti-
cal components of the tension forces are transmitted to the two adjacent build-
ings. The use of the truss structure effectively provides a vertical support at the
central position of the link structure to achieve smaller internal forces in the
link structure and thus smaller deflections. Effectively, the two inverted trusses
provide elastic vertical supports to the link structure.
Figure 2.9c shows a remedial measure to provide an equivalent roller sup-
port in the lateral direction in a historic building. It can be observed that the
upper wooden part of the structure has moved to its right from its support-
ing profiled wooden column. To prevent further horizontal movement, which
might lead to at least a local collapse, physical measures were taken. Steel
tendons have been attached to the upper part of the structure to limit further
movement between the upper and lower parts of the structure. When limiting
further movement, the steel tendon forces would be transmitted through the
friction between the upper part of the structure and the lower wooden column
to the lower column and then to the column support. The horizontal tendon
forces on the upper part of the structure tend to pull on the lower column
deforming to its left through the friction force. A further measure was devel-
oped to provide a horizontal roller support to the column. This support was
implemented using two sub-measures: 1) a pair of steel tendons were placed
around the lower profiled wooden columns (one of them can be seen in Fig-
ure 2.9c, which would prevent the wooden column from deforming to its right.
However, the tendons only carry tension forces and a larger tension force could
cause the column to deform too much to its left. In other words, the action of
the tendons was in fact different to a roller support; 2) to compensate for this
effect, a thicker wooden bar was placed between two lower columns to provide
a force in the opposite direction to that of the force in the tendon. The com-
bined action of the tendons and the wooden bar is like that of a roller support
in the lateral direction.
Figure 2.9d shows two sets of perpendicular horizontal tendons provided at
the upper ends of columns in the Palace of the Grand Master of the Knights
of Rhodes. The other ends of the tendons were fixed through the walls of the
room. Tension applied in the tendons in opposite directions to the columns
effectively provides restraints to the columns making them more stable and
compensating for aging effects. The two pairs of tendon in the two perpen-
dicular directions act as roller supports in the two horizontal directions. This
physical measure to provide additional supports to the columns is simple and
effective without affecting the use of the room.
In summary, the four cases show the different implementations of a roller or
spring support into structures: a prop as a rigid support, an inverted triangular
truss serving as an elastic support at the bottom of a link structure between
 Deflections and Internal Forces 49

two adjacent buildings, a combination of the tension in steel tendons and the
compression of a wooden bar acting as a pinned support, and tension forces
applied in opposite directions serving as an elastic support. There are also
many other forms of implementation measures that can be used to realise a
roller or spring support to reduce internal forces and hence deflections to suit
different structural situations.

2.9 Summary
Deflections and internal forces of structures are functions of applied loads that
have many variations and different combinations in design. This leads to the
difficulties to consider the general characteristics between deflections and inter-
nal forces of structures. In this chapter, the loading is simplified into a unit load
applied on the critical point of a structure, which represents the worst loading
scenario that all loads are lumped to the critical point and normalised to a unit.
This avoids the investigation of the particular effects of actual loading on struc-
tures and allows revealing the general and qualitative relationships between
smaller deflections and desirable distributions of internal forces of structures.
Four structural concepts have been directly and intuitively interpreted based
on the principle of virtual work. These structural concepts are simple and gen-
eral, and this helps their applications at least to truss and frame types of struc-
ture. Due to their simplicity and effectiveness, it is hoped that they can be used
widely in practice as rules of thumb. Each of the four structural concepts has
its own emphasis and characteristics and these will be discussed in the next
four chapters.
Due to the interchangeability between the four structural concepts, one appli-
cation can be seen as an implementation of more than one of the four struc-
tural concepts. As an example, the case in Figure 2.9b can be further examined.
The provision of the inverted triangular trusses can be seen as the implementa-
tion of the fourth structural concept as part of the bending moments of the link
structure is converted into the axial forces in the members of the trusses. Alter-
natively, it can be seen to be the realisation of the second structural concept
in which the bending moments in the linking structure become smaller due to
the upward force provided by the inclined members of the trusses. Therefore,
the focus in the next four chapters will be on the creative use of the structural
concepts rather than on exact classification of applications.

References
2.1 Gere, J. M. and Timoshenko, S. P. Mechanics of Materials, PWS-KENT Publish-
ing Company, 1990, ISBN:0-534-92174-4.
2.2 Graig, R. R. Mechanics of Materials, John Wiley & Sons, USA, 1996.
2.3 Timoshenko, S. P. History of Strength of Materials, New York: McGraw-Hill
Book Co., 1953.
2.4 Ji, T. Concepts for Designing Stiffer Structures, The Structural Engineer, 81(21),
36–42, 2013.
50  Structural Design Against Deflection

2.5 Yu, X. Improving the Efficiency of Structures Using Structural Concepts, PhD
Thesis, The University of Manchester, 2012.
2.6 Ji, T., Bell, A. J. and Ellis, B. R. Understanding and Using Structural Concepts,
Second Edition, Taylor & Francis, USA, 2016.
2.7 Huang, X. and Xie, Y. M. A Further Review of ESO Type Methods for Topol-
ogy Optimisation, Structural and Multidisciplinary Optimisation, 41, 671–683,
2010.
Chapter 3

More Direct Internal

Force Paths

3.1 Routes to Implementation
The appropriate use of bracing systems in structures is an effective way to
create more direct internal force paths. Bracing systems are normally used for
stabilising structures, transmitting loads and increasing lateral structural stiff-
ness. They are ideal for use in types of structure that are sensitive to lateral
loads, such as tall buildings, temporary grandstands and scaffolding structures.
Bracing systems provide direct structural expressions of internal force
paths or load flow how lateral loads are transmitted through structures to
their foundations. There are many, almost unlimited, options to arrange
bracing members and there are large numbers of possible bracing patterns,
as evidenced in existing structures. What is the most effective way to design
bracing patterns?
An effective way is to follow the structural concept, the more direct the
internal force paths, the smaller the deflection. For the purpose of application,
four criteria have been intuitively developed based on this structural concept
aiming to transmit a load at the critical point to the supports of a structure
more directly [3.1, 3.2]:

Criterion 1: Bracing members should be provided in each storey from the

support (base) to the top of the structure;
Criterion 2: Bracing members in different storeys should be directly
connected;
Criterion 3: Bracing members should be linked in a straight line where
possible;
Criterion 4: Bracing members in the top storey and in adjacent bays should
be directly connected where possible. (This criterion is suitable to the
structures that the number of bays in the horizontal direction is larger
than the number of storeys in the vertical direction.)

The first criterion is obvious since the critical point for a multi-storey structure
is at the top of the structure and the load at the top must be transmitted to
the supports of the structure. Therefore, bracing members should be arranged
in every storey of a structure. If bracing members are missing in one of the
52  Structural Design Against Deflection

storeys, it means that the internal force path is cut off and the force has to flow
along an alternative path to reach the support. In other words, the internal
forces have to pass along a longer or less effective way to the supports. Conse-
quently, the structure is likely to experience larger deflections.
There are a number of possibilities for achieving the first criterion, but the
second and the third criteria suggest a way for using a more direct force path.
Once the bracing members are directly linked, the internal forces can flow
directly through them; once the bracing members are linked in a straight line,
the internal forces can flow through them even more directly.
The first three criteria concern mainly the bracing arrangements in different
storeys of a structure and are suitable for tall buildings for which the number
of storeys is larger than the number of bays. For other types of structures, such
as temporary grandstands, the number of bays is usually larger than the num-
ber of the storeys. To create shorter internal force paths or more zero-force
members in such structures, the fourth criterion gives a means for considering
the relationship of bracing members across the bays of the structure.
Bracing members can also be used to create alternative, and sometimes
longer, internal force paths to help meet functional requirements of a structure
and solve challenging technical problems.

3.2 Hand Calculation Examples

3.2.1  Effect of the Four Bracing Criteria

This example examines the effectiveness and efficiency of each of the four brac-
ing criteria for reducing internal forces and lateral deflections of simple frames.
In order to examine the effectiveness of the four criteria for arranging bracing
members, four pin-jointed plane frames are created. Each consists of four bays
and two storeys and uses four bracing members. There are a total 22 members
including 4 bracing members in each of the first four frames. The bracing mem-
bers in the four frames are arranged in such a way that the effectiveness of each
criterion given in Section 3.1 can be identified, which are shown in Figure 3.1
and their features can be summarised as follows:

Frame A: The bracing members are arranged to satisfy the first criterion.
Frame B: The bracing members are arranged to satisfy the first two criteria.
Frame C: The bracing members are arranged to satisfy the first three criteria.
Frame D: The bracing members are arranged to satisfy the four criteria.

In order to examine the effect of the bracing members that are not arranged
fully following the four criteria, Frame E is created as follows:

Frame E: Two additional bracing members are added to Frame C between

the first levels and ground, which don not follow any of the four criteria.

All frame members have the same elastic modulus, E, and cross-sectional area,
A with EA = 1000kN. The vertical and horizontal members have the same
 More Direct Internal Force Paths 53


(a) (b)


(c) (d)

(e)
Figure 3.1 Five frames with different bracing arrangements and internal force paths
(dashed lines). (a) Frame A with six zero-force members. (b) Frame B
with eight zero-force members. (c) Frame C with ten zero-force mem-
bers. (d) Frame D with 14 zero-force members. (e) Frame E with six
zero-force members.

length of L = 1000mm. A concentrated horizontal load of 0.2kN is applied to

each of the five top nodes of the frames. Calculate the internal forces and the
averaged deflections of the top five nodes of the five frames in the horizontal
direction.
Before determining lateral deflections and internal forces, it is possible to
identify intuitively which members are in a zero-force state. Dashed lines are
drawn next to the members that are not in a zero-force state as shown in
Figure 3.1, which indicate the internal force paths transmitting the applied
loads to the supports of the frames. The fewer the dashed lines means the
more direct the internal force paths and consequently the smaller lateral
deflection.
54  Structural Design Against Deflection

The five frames are statically indeterminate structures and beyond simple hand
calculations. However, using the structural concept of symmetry that symmet-
ric structures subjected to anti-symmetric loads will lead to anti-symmetric
responses, the two vertical members on the centre lines of Frames A-E must be in
a zero-force state and thus can be removed from the frames for analysis and the
nodes on the centre lines have no vertical movements and can be represented by
roller supports. Consequently, only halves of the five frames need to be analysed,
and the first four halved frames are statically determinate, suitable for hand cal-
culations, but the half of Frame E remains statically indeterminate and is analysed
using computer software. Figure 3.2 shows the halves of the five frames equivalent
in which the calculated internal forces are indicated in kN next to their members.


(a) (b)


(c) (d)

(e)

Figure 3.2 Internal forces in kN for members in the halves of Frames A-E. (a) Frame
A equivalent. (b) Frame B equivalent. (c) Frame C equivalent. (d) Frame
D equivalent. (e) Frame E equivalent.
 More Direct Internal Force Paths 55

The deflections of the five frames can be calculated using equation 2.16.

Frame A in Figure 3.2 is selected as an illustration and equation 2.16 gives:

N i2 Li
ave  
EAi
[0.12  0.32  0.52  3  12  (0.707)2  1.41  2]  2  (1000N )(1000mm)

(1000000N )
 6.52mm

Δave is the averaged lateral deflection of the top five nodes of the frame. The
value in the square brackets is doubled due to the contributions of the internal
forces in the right half of the frame. An alternative expression is that on the left
side of the previous equation is 0.5kN ave for the half frame. The deflections
of the other frames can be calculated in a similar manner.
To appreciate the effect of the four criteria for realising more direct internal
force paths, Table 3.1 summarises and compares five sets of results calculated for
the five frames based on Figures 3.1 and 3.2. The five sets of results in rows are:

1. The total numbers of zero-force members which can be counted directly

from Figure 3.1.
2. The largest absolute values of the internal forces in the horizontal and
vertical members which can be found from the internal forces indicated in
Figure 3.2.
3. The averaged horizontal deflections of the five top nodes which are calcu-
lated using equation 2.16 for the unit load, which is distributed equally to
the five top nodes of the frames.
4. The relative horizontal deflections normalised to that of Frame A.
5. The relative horizontal stiffnesses which are the inverses of the relative
deflections in (4).

Observations from Table 3.1 and Figures 3.1 and 3.2 can be discussed further:

• Frame A (satisfying the first criterion): The frame has a conventional

form of bracing and the horizontal loads at the top are transmitted to the

Table 3.1 A  Summary of the Results of the Five Frames (Figures 3.1 and 3.2)

Frame A B C D E

1 Number of zero-force members 6 8 10 14 6

2 The largest internal force (kN) 1.0 1.0 0.5 0.4 0.6
in the vertical and horizontal
members
3 The averaged horizontal deflection 6.52 6.03 4.03 3.23 3.90
of the five top nodes (mm)
4 Relative deflection 1.0 0.925 0.618 0.495 0.598
5 Relative stiffness 1.0 1.08 1.62 2.02 1.68
56  Structural Design Against Deflection

supports through the bracing, vertical and horizontal members. The inter-
nal force paths can be examined more closely (Figure 3.2a). The loads pass
through the side vertical members and the bracing members in the upper
storey and the internal force in the bracing member then passes to the con-
nected vertical and horizontal members at the top of the lower storey. The
internal force in the horizontal member passes to the bracing and vertical
member in the lower storey and then to the supports. The internal forces in
the side vertical members are generated to balance the vertical components
of the internal forces of the bracing members. This relatively long internal
force path leaves only two members with zero force, i.e. a total of six zero-
force members in the full frame.
• Frame B (satisfying the first two criteria): It can be seen from Figure 3.2b
that the internal force in the bracing member in the upper storey passes
directly to the bracing and vertical members in the lower storey without
passing through the horizontal member at the top of the lower storey.
Frame B provides a shorter force path than Frame A with one more zero-
force member in the equivalent half frame and thus has a smaller deflection
than Frame A.
• Frame C (satisfying the first three criteria): Figure 3.2c shows that a
more direct force path is created with one vertical member in the lower
storey, which has the largest force in Frame B, becoming zero-force
member. The shorter force path produces an even smaller deflection, as
expected. The third criterion is particularly efficient for not only creat-
ing a more direct force path but also for removing the largest internal
force, which effectively reduces the deflection in comparison with that of
Frames A and B.
• Frame D (satisfying all four criteria): In Frame C, to transmit the lateral
loads at the top nodes, where bracing members are involved, forces in
vertical members have to be generated to balance the vertical components
of the forces in the bracing members (Figure 3.2c). In Frame D two brac-
ing members with mirror orientations are connected at the top central
node, with one member in compression and the other in tension. From
Figures 3.1d and 3.2d it can be seen that the horizontal components of the
forces in these bracing members balance the external lateral loads while
the vertical components of the forces are self-balancing. Therefore, all ver-
tical members are in a zero-force state and Frame D leads to the lowest
deflection of Frames A to D.
• Frame E (satisfying the first three criteria and having two additional
bracing members that do not follow any criteria): Two more members are
added to Frame C to form Frame E, but comparison between Frames D
and E indicates that bracing members which follow the criteria set out can
lead to a smaller deflection than providing more bracing members which
do not fully follow the criteria. As Frame E has two added members com-
pared to Frame C, it should be stiffer than Frame C as expected. In com-
parison with Frame D, in Frame E one bracing member has a smaller force
 More Direct Internal Force Paths 57

of 0.558kN against 0.707kN while five more members are in a forced

state with a maximum force value of 0.606kN. Therefore, Frame E, which
uses more bracing members than Frame D, has a larger deflection than
frame D.

It can also be observed from Table 3.1 that the structure has a smaller deflec-
tion and is stiffer when the internal forces are smaller and more uniformly dis-
tributed although the first four criteria are derived on the basis of the structural
concept of more direct force paths. These examples are simple, and the varia-
tion of bracing arrangements is limited, but they do demonstrate the effective-
ness and efficiency of the criteria that are based on the structural concept of
more direct internal force paths.

3.2.2 T he Most and Least Effective Bracing Patterns for a

Simple Frame
This example identifies the most and least effective bracing patterns through
examining many thousands of bracing patterns of a four-bay and four-storey
frame.
Figure 3.3 shows a four-bay and four-storey pin-jointed frame structure
that is composed of sixteen horizontal and twenty vertical bar members. In
this example a panel is defined as the empty area enclosed by two horizontal
and two vertical members. Therefore, the frame in Figure 3.3a has 16 panels.
The frame is stabilised and stiffened using eight bracing members following
the rules: 1) at each storey, two of the four panels are braced; 2) in each of the
braced panels, there are two possible bracing orientations (Figure 3.3b) [3.3].
Selecting any two from the four panels in a storey gives six options for bracing
4!
in each storey (Figure 3.3b), i.e. . For each braced panel, there are
2!(4 − 2)!
two possible bracing orientations. Thus, there are 24 bracing options in each
storey, i.e. 6 x 2 x 2 = 24. For all four storeys, a total of 244 =331776 patterns
are possible. When only symmetric bracing arrangements are considered, the
number of possible bracing patterns reduces to 256 cases, i.e. (2 x 2)4 = 256.
To simplify the analysis, it is considered that the vertical and horizontal
members have the same length, b = a = 1000mm, all members have the same
cross-sectional area A and elastic modulus E, and EA = 1000kN. A pair of
horizontal forces, each with a value of 0.5kN, are applied anti-symmetrically
at the two top corner nodes of the frame. The lateral stiffness of the frame is
defined as the inverse of the average of the lateral deflections of the two nodes.
The maximum horizontal deflections and internal forces of the frame with dif-
ferent bracing patterns will be compared.
The ANSYS finite element method package was used to calculate the maxi-
mum lateral deflections of all 331776 cases and all 256 symmetric cases.
The maximum horizontal deflections for all 331776 cases are ranked from
the smallest to the largest and demonstrate that the X braced frame, shown
58  Structural Design Against Deflection

(a)

(b)

Figure 3.3 A  four-bay and four-storey frame: a) Geometry of the frame; (b) Any two
braced panels in each storey and bracing orientations.

in Figure 1.3 and Figure 3.5a, has the smallest deflection for the four-bay
and four-storey square paneled frame with all members being the same
cross-section.
When a symmetric structure is subjected to anti-symmetric loads, the inter-
nal forces and deformations of the structure must be anti-symmetric. Hence
the axial forces in the central vertical members of the frames must be zero and
the nodes in the central lines have no vertical displacements. Thus, each of the
frames can be equivalently simplified into a half frame, which becomes a stati-
cally determinate structure, as shown in Figure 3.4b. This greatly simplifies the
analysis of the symmetrically braced frames and allows a hand calculation to
 More Direct Internal Force Paths 59


(a) (b)
Figure 3.4 Using symmetry to simplify the frame model: (a) A whole structure; (b)
A half equivalent structure.

be conducted for checking and for gaining an insight into the behaviour of the
braced frames.
Figures 3.5 and 3.6 present the six most effective and six least effective brac-
ing patterns defined by the smallest and largest horizontal deflections from 256
symmetric bracing patterns, in which the internal forces are indicated next to
the members and the maximum deflections are given at the bottoms of cor-
responding frames.
The bracing patterns in each of the two groups of frames are similar with
small variations. However, the differences in the bracing patterns between the
two groups are obvious and can be summarised as follows:

• The six bracing patterns with the smallest lateral deflections have diago-
nally braced panels in general and at least two bracing members are linked
in straight lines (Figure 3.5).
• The six bracing patterns with the largest lateral deflections have two inde-
pendent vertically braced panels and the bracing members are mainly
placed in parallel to each other (Figure 3.6).

These observations suggest that frames should be braced diagonally across the
whole width of a structure and bracing members should be linked in a straight
line where possible. Alternatively, it should be avoided that bracing members
are arranged in independent vertical panels and are placed in parallel.
Figure 3.7 shows a pair of physical models that resemble the frame in
Figures 3.6e and the frame in Figure 3.5a for which detailed hand calcula-
tions are given in Section 1.2. The maximum lateral deflections of the two
frames are 29.16mm and 7.65mm respectively, which gives a stiffness ratio of
29.16/7.656 = 3.81 for the two frames. With such a large difference in stiffness,
it is easy to feel the relative stiffness of the two frame models by pushing the
top left corners horizontally.
60  Structural Design Against Deflection


(a) 7.656 (b) 8.675


(c) 8.675 (d) 8.675


(e) 8.675 (f) 9.157
Figure 3.5 T he six most effective bracing patterns, the internal forces (kN) and the
maximum deflections (mm).
 More Direct Internal Force Paths 61


(a) 29.66 (b) 29.16


(c) 29.16 (d) 29.16


(e) 29.16 (f) 28.66
Figure 3.6 T he six least effective bracing patterns, the internal forces (kN) and the
maximum deflections (mm).
62  Structural Design Against Deflection

Figure 3.7 Physical models used to feel their relative stiffness [3.4].

The 46 stiffest and the 50 least stiff frames among the 256 symmetrically
braced cases can also be analysed by hand using equation 2.16. The 46 stiff-
est frames are selected because the 46th to the 60th frames have the same
stiffness. The deflections of the frames can be classified into three groups, i.e.
these contributed by the horizontal members (H for δH), the vertical members
(V for δV) and the diagonal members (D for δD). Equation 2.16 can be written
for each of the 96 cases as [3.3]:

44
 N 2L   16 20 8
2  a
   H  V   D    i i    Hi  Vi
 N 2
 N 2
 2  N Di  EA (3.1)
i 1  EA   Hi 1 Vi 1 Di 1 

The deflections, δ H , δ V , δ D and δ, for the 96 cases have been calculated and are
presented graphically in Figures 3.8a and 3.8b for the 46 stiffest cases and the
50 least stiff cases respectively. It can be observed from Figure 3.8 that:

1. The lateral deflections contributed by the diagonal members (δD) are con-
stant for all cases.
2. The lateral deflections contributed by the horizontal members (δH) are
approximately constant for all cases and are smaller than those of the
diagonal members.
3. For the 46 stiffest cases, the lateral deflections contributed by the vertical
members (δV) vary insignificantly and are smaller than those of the diago-
nal members.
4. For the 50 least stiff cases, the lateral deflections contributed by the verti-
cal members (δV) vary significantly and are much larger than those of the
bracing members.
5. For the 46 stiffest frames (Figure 3.8a), δH and δV have similar magnitudes.
For the 50 least stiff frames (Figure 3.8b), δV is much larger than δH.
 More Direct Internal Force Paths 63

(a)

(b)

Figure 3.8 Lateral deflections contributed by vertical, horizontal and diagonal mem-

bers: (a) For the 46 stiffest frames. (b) For the 50 least stiff frames.
64  Structural Design Against Deflection

Because δD is constant and δH varies insignificantly for the 96 cases (Figure 3.8),

the deflections and the relative deflections (the rank) of the 256 cases are basi-
cally controlled by δV. Therefore, bracing patterns that produce large values of
δV, or large internal forces in the vertical members, should be avoided. In other
words, the four criteria provide a practical guide for designing bracing patterns
to reduce values of δV, the lateral deflection contributed by the internal forces
in the vertical members.

3.3 Practical Examples

3.3.1  Tall Buildings

There are different classifications of tall buildings. It is generally thought that
multi-storey structures between 35m and 100m are considered to be high-rise
buildings. Buildings higher than 100m are termed skyscrapers, buildings 300m
or higher are termed super tall and buildings 600m or taller are termed mega-
tall. The taller the building, the more susceptible it is to wind loads. Conse-
quently, different structural systems have been developed to deal with height,
such as rigid frame systems, shear wall systems, tube systems including framed
tubes, braced tubes and tube-in-tube systems.
The bracing criteria shown in Section 3.1 will be used to examine the effec-
tiveness and efficiency of some tall buildings in which bracing systems have
been used including braced frame systems and braced tube systems.

3.3.1.1  John Hancock Center, Chicago

The John Hancock Center (or Tower), a 100-storey 344m tall building, was
built in Chicago that has the nickname of “the windy city”. The Tower was built
in 1969 when computers were little used in building design. The form of the
building (Figure 3.9) shows that the wide base of the building provides greater
structural stability and the narrow upmost part effectively reduces the lateral
wind forces. The structural engineer Fazlur Khan and his collaborators proposed
an exterior-braced frame tube structure. Five and a half huge X bracings, each
across 18 storeys, were used in each of the four sides of the building. Horizon-
tal members were placed between the connections of the bracing members. An
advance on the usual steel-framed tube, this design added global cross-bracing to
the perimeter frame to increase the lateral stiffness of the structure (Figure 3.9).
It can be observed from Figure 3.9 that the external frame columns, global
cross-bracings and beams form a huge exterior trussed tube which is highly effec-
tive for resisting lateral loads. The structural expression of the tube and the great
effectiveness of its lateral resistance is highly harmonised. Some $15 million was
saved on the conventional steelwork by using the huge cross-braces [3.5, 3.6]. It
was regarded as an extremely economical design which achieved the required stiff-
ness to make the building stable. One of the reasons for this success was that the
required lateral stiffness of the structure was achieved by using the cross-braces.
The structural effectiveness and efficiency of the John Hancock Center can
be explained in alternative ways. For example, “The form is especially efficient
 More Direct Internal Force Paths 65

(a)
Figure 3.9 B racing systems used in the John Hancock Center, Chicago, satisfying
the first three criteria. (a) The building (Courtesy of Mr. Nicolas Janberg,
structurae.net, Germany). (b) A closer look.

in the Hancock Tower because the diagonals tie together the otherwise widely
spaced columns, thus distributing the vertical forces evenly among them”
[3.7]. It was not clear how Fazlur Khan and his collaborators generated the
idea of using the huge cross braces, but this ingenious idea can be explained
using the implementation criteria in Section 3.1. It is observed from Figure 3.9
that the global X bracing of the building ideally meets with the first three
66  Structural Design Against Deflection

(b)
Figure 3.9 (Continued)

criteria (bracing members from the top to the bottom of the building and brac-
ing members linked and linked in a straight line where possible), which is an
implementation of the structural concept of more direct internal force paths.
Therefore, it may be said that the use of the huge cross braces creates more
direct load paths that led to larger lateral stiffness and hence smaller deflec-
tions of the structure when subjected to wind loads.
Similar huge global X braces can be observed in other well-known build-
ings. Resembling the global steel X bracing in the John Hancock Tower, global
concrete X braces were used in the 60-storey Onterie Center also in Chicago.
These global X braces were achieved by creating a series of solid window
 More Direct Internal Force Paths 67

spaces running diagonally along the exterior of the building as shown in Fig-
ure 3.10a. It can be observed from Figure 3.10a that these effective “bracing
members” on one side satisfy the first three criteria, while on the adjacent side
they meet with the first two criteria in Section 3.1. Based on the understand-
ing obtained from the examples in Section 3.2.2, it can be deduced that the

(a)
Figure 3.10 T he X braces without using beams. (a) Concrete bracing in the Onterie
Center, Chicago (Courtesy of Mr. Nicolas Janberg, structurae.net,
­G ermany). (b) Steel bracing in the Bank of China, Hong Kong.
68  Structural Design Against Deflection

(b)

Figure 3.10 (Continued)

stiffness contributed by the effective X bracing would be much larger than that
of the snake-like bracing on the adjacent side of the building.
The use of a similar global X bracing pattern is observed in the Bank of
China building in Hong Kong, which has also been regarded as an efficient
and elegant design. The lights placed along the braces and columns seem to
illuminate the internal force paths in the building as shown in Figure 3.10b.

3.3.1.2  Leadenhall Building, London

The Leadenhall Building, located in the centre of London, is a 224m tall com-
mercial office tower. It is commonly known as the “Cheesegrater” because of
 More Direct Internal Force Paths 69

the unique tilted elevation and steel diagrid structure [3.8]. The main differ-
ence between an X braced structure, such as the John Hancock Center, and the
diagrid structure is that there are no columns or vertical members in diagrid
structures. Figure 3.11 shows the front and side views of the building.
To maximise internal flexibility, a perimeter mega-frame structure is used
to form a closed braced tube around all four sides of the building as shown

(a)
Figure 3.11 Leadenhall Building, London. (a) Front view (Courtesy of Mr. Nicolas
Janberg, structurae.net, Germany). (b) Side view.
70  Structural Design Against Deflection

(b)
Figure 3.11 (Continued)

in Figure 3.12. It can be observed that the mega-frame has a varied geometri-

cal pattern. The South elevation, the front view, in Figures 3.11a and 3.12a,
shows a diagrid system while the frames in the other three sides of the building
are effectively braced frames. The South frame consists of diagonal members
spanning vertically between beams at the mega levels of 28m and horizontally
between the mega nodes at 16m centres. The frames in the East and West ele-
vations are comprised of columns spaced at 10.5m centres joined to diagonal
bracing members and beams at each mega level, resulting in an asymmetrical
geometry as shown in Figures 3.11b and 3.12b.
Diagrid structures are very effective for resisting lateral loads as the diagonal
members provide more direct internal force paths to transmit the lateral loads,
 More Direct Internal Force Paths 71

(a) (b) (c)

Figure 3.12 E levation of the Leadenhall Building: (a) South Frame, (b) East/West
Frame, (c) North Frame [3.9].

mainly through axial forces rather than bending moments, to the supports
of the structures (the efficiency of inclined members to transmit lateral loads
can be seen in the hand calculation examples in Section 6.2.2). However, they
appear to be less effective for transmitting vertical loads than conventional
columns. The horizontal members between the mega nodes compensate for
this weakness.
Examining a typical unit subjected to vertical loads as illustrated in Fig-
ure 3.13, it can be seen that the vertical loads tend to make nodes A and B
deform toward each other while nodes B and C tend to move apart from each
other. However, the horizontal member, CD, in the central position ties nodes
C and D to prevent them from deforming apart from each other, which in
turn prevents A and B from deforming toward each other. This makes the unit
much stiffer in the vertical direction. Due to the action of member CD, the
vertical loads are transmitted to the supports mainly by axial forces rather than
by bending moments through the inclined members.
72  Structural Design Against Deflection

Figure 3.13 A  typical diagrid unit subjected to vertical loads.

It can be seen from Figures 3.11 and 3.12 that the geometries of the Lead-
enhall South and East/West frames do not include load bearing members at
the corners where the two frames meet. When horizontal loads are applied in
the plane of the East/West frames in Figure 3.12b, large portions of the forces
are transmitted to the bracing members then flow to vertical members through
turns rather than along straight lines. Considering structural efficiency alone,
inclined edge members could be added to the frame which would create more
direct internal force paths as shown in Figure 3.14b, leading to a stiffer struc-
ture. This edge member would also serve for the South frame (Figure 3.14a) by
framing the diagrid structure which would lead to smaller internal forces and
a more uniform distribution of internal forces. To check this intuitive under-
standing, finite element models of the South and East/West frames without and
with the edge members (Figures 3.12a, 3.12b, 3.14a and 3.14b) were created
for analysis. A unit concentrated load is applied at the top of the frame models.
The inverse of the lateral deflection at the loading point is considered as the
lateral static stiffness of the frame model. If for this example the efficiency (e)
of a structure is defined as the ratio of the lateral static stiffness (K) to the total
mass (M) of the frame model, the efficiencies of the South and West/East frame
models have been determined as follows:

K
e= (3.2)
M
 More Direct Internal Force Paths 73


(a) (b)
Figure 3.14 T he South and West/East frames with added edge members.

The physical meaning of the efficiency is the lateral static stiffness contributed
by a unit structural mass. To remove the modelling errors involved, the ratio of
the efficiencies of the models with edge members (eW ) to that without the edge
members (eWO) is defined as:

eW
R= (3.3)
eWO

The results show that the ratio of efficiency for the South frame model is 1.72
and for the West/East frame model is 1.24, indicating that the frames with the
added edge members are more efficient. This comparison examines only the
efficiency of the frames without considering any other design requirements.

3.3.2 Temporary Grandstands
Temporary structures are ideal temporary solutions for temporary purposes.
Temporary grandstands are frequently used at indoor and outdoor activities such
as tennis tournaments, where the spectators are usually sedentary, and pop con-
certs, where the audiences may move energetically following music beats. Tem-
porary grandstands are designed to be erected and demounted easily and quickly
and are usually of lightweight construction with temporary supports and are
74  Structural Design Against Deflection

therefore relatively sensitive to dynamic loads. Unlike permanent grandstands,

temporary grandstands are normally supported by many vertical members so
that vertical stiffnesses of the structures are not a design concern. However, tem-
porary grandstands must possess sufficient transverse and longitudinal stiffnesses
to resist horizontal loads induced by wind and by spectators’ movements [3.10].
Bracing members are normally used to stiffen temporary grandstands.
The structural safety of temporary grandstands had been considered to be
an important issue following several incidents, the most serious being the col-
lapse of the rear part of a temporary grandstand in Corsica in May 1992. Sub-
sequently, the Building Research Establishment, UK, tested 50 demountable
stands of fifteen different types over several years [3.11]. The seating capacities
of the grandstands varied from 243 to 3500. Only one stand had a vertical nat-
ural frequency below 8.4 Hz (at 7.9 Hz), indicating that there was no concern
for human induced vibration in the vertical direction. However, the natural fre-
quencies in the two horizontal directions were low. Table 3.2 summarises the
distribution of the natural frequencies in the sway and front-to-back directions.
The relatively low natural frequencies indicated that the structures had rela-
tively low stiffnesses in the horizontal directions. The structural characteristics
of temporary grandstands can be observed from the many structures tested:

1. They were normally assembled using slender circular steel tubes, usually
using the same cross-section with a small second moment of area, and the
links between the vertical and horizontal members were closer to pinned
connections than to rigid connections. Therefore, the frames which were
formed from horizontal and vertical members had very low lateral stiff-
nesses as limited frame action could be developed.
2. The vertical members of these grandstands were footed directly onto the sur-
face of the ground. Such footing conditions are regarded as pinned supports.
3. Temporary grandstands had different sizes and heights.
4. Bracing members were provided in most of the structures with many vari-
ations of bracing patterns.

The first two observations are common for most temporary grandstands and
are not the main factors responsible for the low natural frequencies in the
two horizontal directions which tend to be even lower in taller temporary
grandstands. An intuitive understanding of the low natural frequencies (or

P rincipal Horizontal Natural Frequencies of Temporary Grandstands

Table 3.2 
[3.10]

Frequency (Hz) Number of stands

Sway (longitudinal) Front-to-back

Under 3.0 15 10
3.0–3.9 17 13
4.0–4.9 13 9
5.0 or over 5 18
 More Direct Internal Force Paths 75

stiffnesses) in the horizontal directions were that ineffective bracing arrange-

ments were used. These site experiments and observations had generated the
study on effective bracing systems for temporary grandstands and the develop-
ment of the criteria for arranging bracing members [3.1].

3.3.2.1  Collapse of a Temporary Grandstand in Corsica, France

On 5 May 1992 a temporary grandstand at Furiani Stadium in Bastia, Cor-
sica, France, collapsed, killing 18 people and injuring 2300. On that day SC
Bastia faced Olympique de Marseille for a semi-final football match in the
French Cup that was the premier knockout cup competition in French football
organised by the French Football Federation (FFF). In order to accommodate
a large-capacity crowd, an additional temporary grandstand was erected at the
back of an existing grandstand to increase the number of seats by 50%. Local
authorities approved the project without restrictions. This added rear part of
the grandstand collapsed at 20:20 shortly before the scheduled start of the
match. An investigation of the disaster concluded that there had been several
violations of rules concerning the construction of the temporary grandstand.
Problems were also identified in the management of ticketing and in the atti-
tudes of sporting and municipal executives.
Figure 3.15 shows the cross-section of the grandstand that had the front
part and the back part in the north-south direction. The front part consisted
of six 3m bays making a total width of 18m and the back part had four 3m
bays with a total width of 12m. The maximum height of stand was about 11m.
The internal force paths of the temporary grandstand (back part) can now be
examined using the bracing criteria based on the information in Figure 3.15.

1. There were no bracing members in Bay 9, and the two horizontal members
and the seating unit in the bay simply linked bays 8 and 10. This did not

Figure 3.15 Cross-section of the collapsed temporary grandstand in Corsica, France.

76  Structural Design Against Deflection

contribute the lateral stiffness of the bay to the back part of the grand-
stand. If there had been no links between the seating decks in bays 8 and
10, the lateral stiffness of the rear part of the grandstand would have been
the sum of the stiffnesses of the two independent bays, 8 and 10, rather
than that of the whole of the back part of the grandstand.
2. There was a weak connection at a point between the front and back parts of
the stand, i.e. the connection between bays 6 and 7. It appeared that the front
part was assembled using standard units and was much stiffer than the back
part. However, advantage was not taken of this leaving the taller back part of
the stand with insufficient supports. If bracing members had been arranged
in bay 7 to unify the two parts of the grandstand, the back part would have
been much stiffer and stronger, and the collapse might have been avoided.
3. There were large eccentricities (1.0m) between the ends of the two middle
bracing members and the intersection points of the horizontal and verti-
cal members in bays 8 and 10. The internal forces in the middle bracing
members were transmitted to the vertical members to which they were
connected and the slender vertical members had to bend to transmit the
eccentric forces to the intersection points.

Through these observations and their qualitative interpretation, it can be real-

ised that the back part of the grandstand had a low natural frequency and
stiffness in the lateral direction and was thus susceptible to human induced
dynamic loads. The identification of the weakness effectively suggests ways to
improve the grandstand to have much higher stiffness in the lateral direction:

1. Providing bracing members in bay 9 to allow bays 8, 9 and 10 working as

a whole.
2. Providing bracing and horizontal members in bay 7 to allow the front and
back parts of the grandstand working as a whole.
3. Placing bracing members at the connections of vertical and horizontal
members to allow for more direct internal force paths in bays 8 and 10.

3.3.2.2  A Temporary Grandstand in Eastbourne, UK

Figure 3.16 shows the back and side views of a temporary grandstand erected
for the International Women’s Tennis Championship in June 1992 in East-
bourne, UK. The stand consisted of 38 trusses constructed and slotted together
using a specially made scaffolding system. Eight vertical members carried each
truss to adjustable supports on wooden bases. The grandstand could hold just
under 2700 people accommodated in 28 rows of seats, with up to 100 seats in
each row. The stand was about 23.2m from front to back, rising from 2.5m
at the front to 10.6m at the back. Its length was estimated to be about 60m.
The bracing patterns in the back and side of the grandstand can be identified
from Figure 3.16. The back frame of the grandstand, with 25 bays, is shown
in Figure 3.17a, in which alternative bays were braced from the bottom to the
top. A bracing member was placed at the first storey level in all the other bays.
Among the 50 temporary grandstands tested, this had the best bracing [3.10].
 More Direct Internal Force Paths 77

(a)

(b)

Figure 3.16 A  temporary grandstand in Eastbourne, UK. (a) The back view. (b) The
side view.
78  Structural Design Against Deflection

(a)

(b)
Figure 3.17 Design of bracing patterns. (a) Frame A based on the design shown in
Figure 3.16a. (b) Frame B based on the four bracing criteria.

The vibration of the stand in the sway (longitudinal) direction was how-
ever a major concern. Vibration tests showed that the fundamental natural
frequency of the stand in this direction changed from about 2.7 Hz when the
stand was empty to about 1.7 Hz when the stand was fairly full.
It can be seen from Figure 3.17a that the bracing members were arranged
from the bottom to the top of the grandstand which satisfies Criterion 1 and
some of the bracing members were linked in straight lines across the first three
storeys. However, the linkage in a straight line did not pass throughout the
full height of the grandstand and no bracing members meet at the tops of the
structure. To fulfil all four criteria, the bracing pattern can be redesigned as
shown in Figure 3.17b. The redesign is straightforward if the four criteria are
implemented without considering anything else, such as safety, economy and
elegance of the structure. To compare the effectiveness and efficiency of the
two bracing patterns for the grandstand, computer analyses were conducted.
Table 3.3 compares the static stiffnesses, the fundamental natural frequencies
and the numbers of bracing members used of the two braced frames.
The comparison in Table 3.3 shows that the lateral stiffness of Frame B with
the improved bracing pattern is much larger, being 284% of that of Frame
A with the original bracing pattern. The ratio of the fundamental natural
frequencies is 169% as the stiffness is proportional to the natural frequency
squared. The significant increase of the stiffness is due to the improved brac-
ing pattern providing far more direct internal force paths, as described in
 More Direct Internal Force Paths 79

Table 3.3 Comparison of the Lateral Stiffnesses and Natural Frequencies of Frames

A and B

Lateral stiffness Fundamental natural Number of bracing

frequency members

Frame A: with the 3.16 MN/m 1.96 Hz 64

original bracing
pattern (Figure 3.17a)
Frame B: with the 8.96 MN/m 3.31 Hz 52
improved bracing
pattern (Figure 3.17b)
Ratio (Model B/Model A) 2.84 1.69 0.81

(a)

(b)
Figure 3.18 Comparison of the shapes of the fundamental modes of the two frames.
(a) Frame A. (b) Frame B.

Sections 3.1 and 3.2. In terms of the consumption of material, Frame B uses

19% less of bracing members than Frame A.
Figure 3.18 compares the shapes of the fundamental modes of vibration of
the two frames in which the maximum movements are normalised to the same
value. It shows that Frame B with the improved bracing pattern displays global
deformations while Frame A with the original bracing pattern displays both
global deformations and local deformations at the top nodes, which further
demonstrates the effectiveness and efficiency of the improved bracing pattern.
Table 3.3 shows that Frame B, with the improved bracing pattern based
on the four bracing criteria, is more effective (possessing much larger stiff-
ness to reduce deflections), more efficient (using a smaller number of bracing
members). When examining the appearance of the two frames in Figure 3.17,
it seems that Frame B, following the bracing criteria, is more elegant than the
Frame A, with the original bracing pattern.
80  Structural Design Against Deflection

3.3.2.3  Two Further Cases

Figure 3.19a shows a temporary grandstand used for the British Grand Prix in
Silverstone, UK. It can be observed from the back of the stand that no bracing
members were provided which led to a low stiffness and a low fundamental
natural frequency in the lateral direction. Fortunately, the spectators watching

(a)

(b)
Figure 3.19 Bracing patterns of two temporary grandstands. (a) Without using brac-
ing members. (b) Assembled using standard units.
 More Direct Internal Force Paths 81

the motor racing were sedentary and the temporary grandstand survived its
use for motor racing events. Such a temporary grandstand, however, could not
be used for pop concerts or for football events in which human dynamic loads
would be experienced, as resonance might occur in either the lateral or the
front-to-back direction of the temporary grandstand.
Figure 3.19b shows a temporary grandstand that was assembled using
standard units in which the units were only connected by the relatively heavy
and stiff seating decks at the top of the grandstand. The advantage of using
this type of grandstand is that it is quick and easy to erect. However, the draw-
back is that it has low lateral natural frequencies. For an easy understanding,
the temporary grandstand is resembled as a simple plane model as shown in
Figure 3.20. The plane model consisting of four equally spaced plane units
that are linked at their tops through a rigid plate. If a new unit is added to the
model, the mass on two bays will also be added. Consider that each unit has a
lateral stiffness of k and a lumped mass of m at its top, the lateral stiffness of
the plane model is 4k and the mass at the top of the model is (2  4  1)m  7 m,
i.e. the sum of the stiffness of the four individual units and the sum of the
mass of the seven bays. If the temporary grandstand (Figure 3.19b) consists
of n units and each unit has its lateral stiffness of k and a deck mass of m, the
lateral stiffness of the stand would be nk and the total mass on the top would
be (2n − 1)m . The natural frequency of the whole grandstand in the lateral
direction ( fw ) would be close to that of a typical unit with the mass for two
bays ( fu ), i.e.:

1 nk 1 k
fw    fu (3.4)
2 (2n  1)m 2 2m

Equation 3.4 indicates that the temporary grandstand takes more units will not
increase its lateral natural frequency as the mass of the grandstand increases
proportionally to the stiffness.

Figure 3.20 A  plane model to resemble the grandstand in Figure 3.19b.

82  Structural Design Against Deflection

3.3.3 Scaffolding Structures
Scaffolding structures are temporary structures that are used primarily to pro-
vide temporary access during the construction or renovation of buildings and
other structures. The design of scaffolding imposes some restrictions that can
be ignored in the design of other structures. For example, scaffolding struc-
tures must be easily assembled and taken apart and the components should
also be relatively light to permit manhandling. Scaffolding structures are often
erected using simple units and slender members and the connections, bracing
patterns and load paths are not always designed appropriately. Many projects
require very large scaffolding structures which must possess sufficient lateral
stiffness to ensure that all the loads acting on them can be transmitted safely to
their supports. Although scaffolding structures are light and temporary, their
design should be taken seriously. The concept of direct force paths and the four
criteria are applicable to scaffolding structures.

3.3.3.1  Collapse of a Scaffolding Structure, Manchester

The scaffolding structure shown in Figure 3.21 collapsed in 1993 [3.12], though
no specific explanation was given. Using the structural concept of direct internal
force paths and the understanding gained from the examples in Section 3.2, the
cause of the incident may be suggested. No diagonal (bracing) members were
provided in the scaffolding structure, i.e. no direct internal force paths were pro-
vided. The scaffolding structure worked as an unbraced frame structure and the
lateral loads, such as wind loads, on the structure had to be transmitted to its sup-
ports through bending of the slender vertical scaffolding members. The structure
did not have enough lateral stiffness and collapsed under wind loads only.

Figure 3.21 Collapse of a scaffolding structure (Courtesy of Mr John Anderson).

 More Direct Internal Force Paths 83

3.3.3.2  Lack of Direct Internal Force Paths

For the convenience of erection of the scaffolding structures, standard pro-
prietary units were used. The unit used in the scaffolding structure shown
in Figure 3.22a consisted of two horizontal members and two short diago-
nal bracing members supported by two vertical members. The unit is useful
for transmitting vertical loads applied to the top horizontal member to the
vertical members. It is equivalent to a thick beam in the overall scaffolding
structure, which is effectively a deep beam and slender column structure.
The diagonal members used in the structure do not provide the force paths
to transmit the lateral loads on the structure from the top to the bottom of
the structure and do not follow the basic criteria for arranging bracing mem-
bers. Therefore, the scaffolding structure can be judged to have low lateral
stiffness.
Figure 3.22b shows another example where the scaffolding structure works
as a frame structure with strong beams, the trusses, and weak columns. It
would be inconvenient to place vertical members in the entrance area and
the lower truss over the entrance supports two vertical members above. The
scaffolding structure resists lateral loads mainly through bending rather than
through axial forces in the vertical members which provides less effective lat-
eral resistance. In addition, the slender vertical members are not ideal for trans-
mitting bending. Judging by the four bracing criteria, this scaffolding structure
has a lack of internal force paths from the top to the bottom of the structure,
and no direct internal force paths have been created. It can be concluded that
the scaffolding structure has a low lateral stiffness.

3.4 Further Comments
The structural concept of more direct internal force paths has been imple-
mented by using appropriate bracing patterns that can be applied to tall
buildings, temporary grandstands and scaffolding structures. The hand
calculation examples, and the practical cases, demonstrate that the use of
the structural concept or the implementation criteria can make structures
stiffer (experiencing smaller deformation), more efficient and perhaps more
elegant. There are other implementation measures to be explored and these
may be observed from existing structures or developed from the structural
concept itself.
Internal force paths or load paths can be designed to solve other practi-
cal and challenging structural problems as can be observed from practical
examples. One such example is seen at the entrance of the Cannon Street
Underground Station in London, the upper eight storeys of the building
cantilever a distance from the building supports as shown in Figure 3.23.
How do the loads of the cantilever building transmit to the supports of the
building? To understand the load paths, a simple diagram may be drawn for
qualitative analysis which aims to capture the physical essence of the load
paths but omits some less important details. Figure 3.24 shows a model of
the side façade of the building based on the photos in Figure 3.23 and this
acts like a truss structure.
84  Structural Design Against Deflection

(a)

(b)
S caffolding structures assembled from proprietary units but lacking
Figure 3.22 
direct internal force paths.
 More Direct Internal Force Paths 85

(a)

(b)

Figure 3.23 A building over the Cannon Street Underground Station, London. (a) The
cantilevered upper eight-storey are supported by a huge bracing system.
(b) The bracing members show the force paths to the foundation.
86  Structural Design Against Deflection

Figure 3.24 Paths and estimation of the internal forces in the truss system.

To estimate the behaviour of the bracing system, assume that the height
to width ratio of the unit ABDC is 4/3. Approximate and replace the load-
ing and self-weight of the cantilever part, ABDC, by two vertical loads, 4P,
applied at nodes A and B (Figure 3.24). The values of internal forces are
shown adjacent to members with positive signs for tension and negative sign
for compression. In this case, the bracing members provide clear and desired
internal force paths for transmitting the loads of the cantilever building to
the supports.

References
3.1 Ji, T. and Ellis, B. R. Effective Bracing Systems for Temporary Grandstands, The
Structural Engineer, 75(6), 95–100, 1997.
3.2 Ji, T. Concepts for Designing Stiffer Structures, The Structural Engineer, 81(21),
36–42, 2003.
3.3 Yu, X., Ji, T. and Zheng, T. Relationships Between Internal Forces, Bracing Pat-
terns and Lateral Stiffness of a Simple Frame, Engineering Structures, 89, 147–
161, 2015.
3.4 Roohi, R. Analysis, Testing and Model Demonstration of Efficiency of Different
Bracing Arrangements, Investigative Project Report, UMIST, 1998.
3.5 Parkyn, N. Super Structures: The World’s Greatest Modern Structures, Merrell,
2004.
3.6 Bennett, D. Skyscrapers—Form & Function, Simon & Schuster, 1995.
3.7 Billington, D. P. The Tower and the Bridge, Princeton University Press, Princeton,
1985.
 More Direct Internal Force Paths 87

3.8 Eley, D. and Annereau, N. The Structural Engineering of the Leadenhall Building,
London, The Structural Engineer, 96(4), 10–20, 2018.
3.9 Saeed, M. Parametric Study on the Diagrid Frame of the Leadenhall Building &
Topology Optimisation of Bracing Systems, MSc Dissertation, The University of
Manchester, 2018.
3.10 Institution of Structural Engineers. Temporary Demountable Structures: Guid-
ance on Procurement, Design and Use, Third Edition, Institution of Structural
Engineers, London, 2007.
3.11 Ellis, B. R., Ji, T. and Littler, J. The Response of Grandstands to Dynamic Crowd
Loads, Structures and Buildings, the Proceedings of Civil Engineers, 140(4), 355–
365, 2000.
3.12 Anderson, J. Teaching Health and Safety at University, Proceedings of the Institu-
tion of Civil Engineers, Journal of Civil Engineering, 114(2), 98–99, 1996.
Chapter 4

Smaller Internal Forces

4.1 Routes to Implementation
There are several routes to create smaller internal forces which are apparent
and intuitive. These provide a basis for developing particular implementation
measures to allow the realisation of smaller internal forces in structures

1.  Reducing Spans

As the deflection is proportional to the span to the power of four, reducing

span whenever possible is the most effective way to achieve smaller deflections.
For example, if the span of a beam is halved, the maximum deflection will be
one sixteenth of that of the original beam. The maximum bending moment will
also be reduced, to one quarter of that of the original beam.

2.  Partially Self-Balancing Internal Forces

While it may not be possible to achieve complete self-balancing of internal

forces, it may be possible to achieve partial self-balancing. The reduction of
large bending moments can be realised by creating partially self-balanced sys-
tems in which a newly generated positive (negative) bending moment offsets
part of an existing negative (positive) bending moment. To do this a designer
needs to sense where the large bending moment would occur and its direction
and then, more importantly, needs to develop an appropriate physical measure
to introduce the new bending moment. Some measures which can be used to
achieve partial self-balancing are:

1. Using pre-stressing or post-stressing techniques to produce a bending

moment or deflection that is in the opposite direction to that induced by
loads.
2. Adding structural elements into a structure which can constrain some
deflections and/or create bending moments in the opposite direction to
that induced by loads.
3. Redistributing internal forces which helps to reduce large bending moments.
 Smaller Internal Forces 89

3.  Providing Elastic Supports

Providing rigid supports to reduce deflections may be difficult to be realised in

practice due to functional, structural or aesthetic requirements. Using elastic
supports may however be a feasible solution. There are two types of elastic
supports: external elastic supports and internal elastic supports.

1. External elastic supports: When an elastic support is cut, a pair of action

and reaction forces are revealed which have equal magnitudes but in oppo-
site directions. If one force acts on the structure while the other does not
act on the structure, this is an external elastic support. The ring shown in
Figure 4.1a has a pair of horizontal springs to restrain its lateral deflection.
The spring forces act on the ring and on solid supports that are not part
of the ring. Therefore, these two springs act as external elastic supports to
the ring. Typical structures making use of external elastic supports are the
many cable stayed bridges in which the cables effectively act as elastic sup-
ports to the bridge decks and allow the bridge span longer distances and
experience smaller bending moments. The other ends of the cables locate
on the pylons that are supported by their foundations.
2. Internal elastic supports: If the action and reaction forces from an elastic
support are all applied directly to the structure, it is regarded as an inter-
nal elastic support. The ring shown in Figure 4.1b has a tendon across its
centre which acts as two springs to restrain the deflections in the lateral
directions of the ring due to the applied load. The actions of the tendon on
the ring are similar to those of the two external springs in Figure 4.1a but
the forces in the tendon act on the two sides of the ring. The internal elastic
supports provided by the tendon may also be considered as the realisation
of partial self-balancing.


(a) (b)
Figure 4.1 Rings with lateral elastic supports. (a) Springs acting as external elastic
supports. (b) Tendon acting as internal elastic supports.
90  Structural Design Against Deflection

4.2 Hand Calculation Examples

4.2.1  A Simply Supported Beam with and without Overhangs

This example demonstrates and quantifies the effectiveness and efficiency of
span reduction and self-balancing of internal forces through using overhangs.
Figure 4.2 shows three simple beams that have the same rigidity of EI. Beam 1 in
Figure 4.2a is a normal, simply supported beam with a span of L and is subjected
to a uniformly distributed load of q. The other two beams are evolved from Beam
1. When the two supports of Beam 1 are moved inward symmetrically with a
distance of µ L , it becomes Beam 2 (Figure 4.2b) that is called a beam with over-
hangs. When Beam 1 increases its overall length by α L at each of its two ends and
a concentrated load of P is applied at each of its two free ends, it becomes Beam
3 as shown in Figure 4.2c, in which α and P can be variables to be determined
for achieving a more efficient design. Determine the maximum bending moments

(a)

 
(b) (d)

(c)

Figure 4.2 A simply supported beam and its two variations. (a) Beam 1: A simply sup-
ported beam. (b) Beam 2: A simply supported beam with overhangs and
the same length as Beam 1. (c) A simply supported beam with overhangs
for a total length of (1 + 2 α ) L. (d) Free-body diagram used to determine
the bending moment at the centre of Beam 2.
 Smaller Internal Forces 91

and the deflections at the centre C of the three beams and examine the efficiency
of Beams 2 and 3 against Beam 1. Some basic formulae for calculating the bending
moments and deflections of such beams can be found in related textbooks [4.1].

Solution:

Beam 1: A simply supported beam (Figure 4.2a).

The maximum bending moment and deflection at the centre of the simply
supported beam are respectively:

1 5qL4
M 1,C = qL2 ; 1,C   (4.1a, 4.1b)
8 384EI

Beam 2: A simply supported beam with overhangs and with the same length
as Beam 1 (Figure 4.2b).
The bending moments at supports A and B are:

1
M 2, A  M 2, B   q  2 L2(4.2a)
2

The bending moment at mid-span C of the beam can be determined using the
free body diagram shown in Figure 4.2d as follows:
2
1 1  1 L 1 1
M 2,C  qL     L  q    qL2  q  L2 (4.2b)
2 2  2 2 8 2

µ (< ½) is a variable in equations 4.2a and 4.2b and can be adjusted to achieve

smaller bending moments. Consider the particular case when the magnitudes
of moments at location A and C, (M2,A and M2,C) are the same. Equating the
magnitudes in equations 4.2a and 4.2b gives:

1 1 1
q  2 L2  qL2  q  L2 or 4 2  4  1  0 (4.3)
2 8 2

The valid solution of the quadratic equation in equation 4.3 is µ = 0.207. Sub-

stituting µ = 0.207 into the expressions for M 2, A and M 2,C leads to:

1 1
M 2,C  M 2, A  q( L)2  q(0.207) 2 L2  0.0214qL2 (4.4a)
2 2

Alternatively, the bending moment at the centre of the beam can be determined
as half of the maximum bending moment of the simply supported beam with
the span of: (1  2  )L  0.586L

11 1
M 2,C  M 2, B  q[(1  2  )L]2  q(0.586) 2 L2
28 16 (4.4b)
 0.0214qL2  17.1%M 1,C
92  Structural Design Against Deflection

The superposition method can be used to calculate the deflection at mid-

span C. The loading in Figure 4.2(b) can be decomposed into two simple cases
as shown in Figure 4.3a and Figure 4.3b. The deflection at C due to the dis-
tributed loads on the overhangs (Figure 4.3b) is the same as that due to two
couples acting at the supports A and B (Figure 4.3c), which is q(µ L)2 / 2 . This
loading generates an upward deflection of the middle span, which can be cal-
culated using a textbook equation [4.1] as:

5q[(1  2  )L]4 M [(1  2  )L]2

 2,C  
384EI 8EI
5q(0.586L)4 q(0.207 L)2 (0.586L)2
= 
384EI 2 8EI (4.5)
4 4 4
5qL 5qL 5qL
= 0.1179   0.0706   0.0473 
384EI 384EI 384EI
 4.73%1,C

(a)

(b)

(c)

Figure 4.3 U sing the superposition method for calculation. (a) Distributed loads
applied between the two supports. (b) Distributed loads applied on the
overhangs. (c) Equivalent loads to (b) for calculating the deflection at
the mid-span.
 Smaller Internal Forces 93

The results show that the maximum bending moment for Beam 2 is about 17%
of that for Beam 1 and that the maximum deflection for Beam 2 is less than 5%
of that for Beam 1. Such significant reductions are due to the use of the first
two physical measures in Section 4.1:

1. The reduced span between the two supports: Bending moment is propor-
tional to the span squared and deflection is proportional to the span to
the power four. Hence the shortened span effectively reduces the bending
moment and deflection.
2. The reduction of bending moment through part self-balancing: The nega-
tive bending moments over the supports due to the use of the overhangs
offset part of the positive bending moment due to the loads at the middle
span. This can also be explained as a redistribution of bending moments.
The reduced bending moment will also lead to smaller deflection.

The effects of the span reduction for Beam 2 can be seen when the distributed
load is only applied on the middle span between the supports of Beam 2, for
which the bending moment at the centre C is

1 qL2
M 2,C  q[(1  2  )L]2  0.5862  34.3%M 1,C (4.6)
8 8

Equation 4.6 indicates that the span reduction leads to a reduced maximum

bending moment of 34.3% of M 1,C . If applied, the loading on the two over-
hangs (Figure 4.3b) further reduces the bending moment of 34.3% M 1,C by a
half to 17.1% (equation 4.4b). The first term in Equation 4.5 shows that the
span reduction results in a reduction of 88.2% of ∆1,C while the loading on the
overhangs causes a further reduction of 7.06% of ∆1,C . Table 4.1 summarises
the efficiency of the two measures on the reduction of both maximum bending
moment and maximum deflection.
It can be seen from Table 4.1 that the measure of span reduction plays the
dominant role in reducing the structural responses, in particular the deflection,
and the self-balancing measure reduces the bending moment more significantly
than deflection in this example.
Figure 4.4 shows the roof structure for the new entrance of the Fountains
Abbey in North Yorkshire, UK. The roof is supported by a series of paral-
lel curved beams that are in turn supported directly by individual columns

Table 4.1 S ummary of the Efficiency on the Reduction of Bending Moment and
Deflection

Maximum bending moment Maximum deflection

Span reduction 65.7% 88.2%

Self-balancing action 17.1% 7.06%
Total 82.8% 95.3%
94  Structural Design Against Deflection

Figure 4.4 Simply supported curved beams with overhangs.

at their lower ends and by a beam spanning between columns toward their
upper ends. Examining the supports of the curved beams, this is an implemen-
tation of the simply supported beam with overhangs shown in Figure 4.2b.
The structural behaviour of the curved beams in the vertical direction is the
same as that of a straight beam with overhangs, but the curved roof surface
and unequal heights of the supports are aesthetically pleasing to the eyes of
visitors.
In engineering practice µ = 0.2 is used rather than the exact solution of
µ = 0.207 for a simply supported overhanging beam for ease of design.

Beam 3: A simply supported beam with overhangs with a total length of

(1+ 2 α ) L (Figure 4.2c).
Similar to the solution of Beam 2, the superposition method is used again
to calculate the bending moments at A(B) and C and the deflections at C and
D(E) of the beam.
The bending moments at B and C are respectively:

M 3, B  P L (4.7)

1 1 8P
M 3,C  qL2  P L  qL2 (1  ) (4.8)
8 8 qL
 Smaller Internal Forces 95

where α is a design parameter and P can be either a given load or a design

force parameter. The downward deflection at C due to the distributed load q
alone is:

5qL4
 3,C 1 
384EI

The upward deflection at C due to the two concentrated loads of P alone (cal-
culated on the basis of a simply supported beam subjected to couples at A and
B of M  P   L   PL (see Figure 4.3(c)) is:

ML2  PL3
 3,C 2  
8EI 8EI

Therefore, the downward deflection at C due to the full loading on the span of
the beam is the sum of the two sub-loading case deflections:

5qL4  PL3 (5qL  48 P)L3

 3,C   3,C 1   3,C 2    (4.9)
384EI 8EI 384EI

To determine the deflection at D or E, it is necessary to know the slopes of

deformation induced by the two sub-loading cases. Due to the action of the
distributed load alone, there is an upward deflection at D caused by the rota-
tion of member DA, which can be calculated using an existing formula [4.1] as:

qL3  qL4
 3, D1  3, A1   L  L 
24EI 24EI

The downward deflection at D due to the concentrated load is the sum of two
deflections: the end deflection of a cantilever of length of α L due to P at its
free end and the end deflection due to the rotation of the overhang DA:

PL3 PL3 (P L)L PL3 (2  3 2 )

 3, D 2   3, A 2 L   L 
3EI 3EI 2EI 6EI

The total downward deflection at D is the sum of the deflections due to the two
sub-loading cases, i.e.:

PL3 (2  3 2 )  qL4 [4P(2  3 2 )   qL]L3

 3, D   3, D 2   3, D1    (4.10)
6EI 24EI 24EI

Equations 4.7 to 4.10 contain two variables, P and α , and these variables can
be used to actively adjust the bending moment and deflection of the beam with
overhangs for the loading condition shown in Figure 4.2c.
Figure 4.5 shows a steel-framed two-storey car park building, which embodies
the study of Beam 3. The vertical loads from floors are transmitted to the cellular
96  Structural Design Against Deflection

Figure 4.5 Overhangs and tendon forces are used to reduce bending moments and
deflections of cellular beams (Courtesy of Mr John Calverley, UK).

beams and then from the cellular beams to the supporting columns. Overhangs
are purposely designed in the structure to reduce the bending moments and
deflections of the cellular beams. Examining the first overhang, two steel cables
link the free end of the overhang and a concrete support. A downward force at
the free end of the overhang is provided by tensions induced in the cables. This
force, similar to P in Figure 4.2c, will generate a negative bending moment in
the beam over the column support which will partly offset the positive moments
in the beam induced by the floor loading. The length of overhangs and the force
in the steel cables could be the design parameters actively selected to reduce the
bending moments and deflections of the cellular beam.
The overhang is subjected to a concentrated force at its free end and therefore
the bending moment varies linearly along the overhang from zero at its free end to
a maximum at the column support. Reflecting the shape of the bending moment
diagram, the overhang is tapered toward the column. This makes the overhang
appear lighter and more elegant than would be the case if a constant cross-section
was used throughout its length. A prop is provided between the concrete sup-
port and the column end of the overhang which stiffens the overhang to prevent
its rotational deformation due to the action of the cables, contributes additional
lateral resistance to the structure and provides anchoring positions for the cables.

4.2.2 Y Shaped Columns with and without

a Horizontal Tendon
This example demonstrates and quantifies the effectiveness and efficiency of
self-balancing through the use of a bar member in a Y shaped column.
 Smaller Internal Forces 97

Figure 4.5 shows two Y shaped columns, one without and one with a
horizontal bar linking the two top ends of the column, which are subjected
to the same pair of symmetric vertical loads. The dimensions of the Y shaped
column can be described by three parameters: the column height, h, the span,
a, and height, b, of the two symmetric inclined members. The length of the
inclined member is s  a 2  b 2 . Assume that the Y shaped columns have a
uniform cross-section with a rigidity of EI and that the horizontal bar has
a sectional area of A and elastic modulus of Eb . Conduct the following
analyses:

1. Determine the bending moments in the two Y shaped columns.

2. Determine the vertical deflection at point A and the relative horizontal
deflections between points A and B of the two inclined members.
3. Examine the effect of the horizontal bar on the reduction of the bending
moment and lateral and vertical deflections of the Y shaped columns.

Solution:

Column 1 (Figure 4.6a):
The Y shaped column is a statically determinate structure and its bending
moment diagram can be drawn easily as shown in Figure 4.7a. There is no
bending moment in the vertical column as the moments induced by the pair
of symmetric vertical loading are self-balancing at the connection point C.


(a) (b)
Figure 4.6 Two Y shaped columns. (a) Column 1, without a horizontal bar. (b) Col-
umn 2, with a horizontal bar.
98  Structural Design Against Deflection

      
(a) (b) (c)

  
(d) (e)

Figure 4.7 Bending moment diagrams of the Y column due to different loads. (a) Due
to a pair of vertical loads. (b) Due to a unit downward load at point A. (c)
Due to a pair of unit horizontal inward forces. (d) Releasing the bar force
in the tied Y shaped column to make it a statically determinate structure.
(e) Due to a pair of horizontal inward forces F.

The maximum bending moment occurs at point C and has a magnitude of

M 1,max = Pa.
The vertical and horizontal deflections can be calculated using the unit load
method and the moment-area method in conjunction with Figures 4.7a, 4.7b
and 4.7c, which show the bending moment diagrams for the Y shaped column
 Smaller Internal Forces 99

due to the pair of vertical loads P, a unit downward vertical load acting at
point A and due to a pair of inward horizontal unit forces at points A and B.
The vertical downward deflection of point A induced by the pair of loads P
is (Figure 4.7a and 4.7b):

1 Pas 2a Pa 2 s
1,V  ( )( )  (4.11)
EI 2 3 3EI

The relative horizontal deflection between points A and B is:

1 Pas 2b 2Pabs
1, H  ( )( )  2   (4.12)
EI 2 3 3EI

In equations 4.11 and 4.12, the first subscript specifies the Y shaped column
(Figure 4.6) and the second indicates the direction of the deflection. The nega-
tive sign in equation 4.12 means that the relative horizontal deflection between
points A and B is opposite to the direction of the unit horizontal forces shown
in Figure 4.7c, i.e. points A and B deform outwards.

Column 2 (Figure 4.6b):
The Y shaped column with a horizontal bar is a statically indeterminate struc-
ture as the internal force in the bar is unknown. The moment-area method can
be used to determine the internal force F in the bar. When the bar is replaced by a
pair of forces F as shown in Figure 4.7d, the Y shaped column becomes statically
determinate. Figure 4.7e shows the bending moment diagram due to the pair of
horizontal forces, F. The force, F, can be determined using the deflection compat-
ibility condition between points A and B of the tied Y shaped column as follows.

The horizontal deflection due to the pair of vertical loads P (Figures 4.7a and
4.7c) is:

1 Pas 2b 2Pabs
 H ,P  ( )( )  2   (4.13)
EI 2 3 3EI

The horizontal deflection due to a pair of horizontal forces F (Figures 4.7e and

4.7c) is:

1 Fbs 2b 2Fb 2 s
 H ,F  ( )( )  2  (4.14)
EI 2 3 3EI

The elongation of the horizontal bar is:

F  2a
b  (4.15)
Eb A

Compatibility of the deflections in equations 4.13, 4.14 and 4.15 requires:

 H , P   H , F  b  0(4.16)
100  Structural Design Against Deflection

Substituting equations 4.13 to 4.15 into equation 4.16 gives:

2Pabs 2Fb 2 s 2Fa

    0 (4.17)
3EI 3EI Eb A

The signs in the three deflections in equations 4.13 to 4.15 may be a little con-
fusing, but they can be judged from an understanding of the physical nature
of the deflections of the Y shaped column without the horizontal bar. The
action of the pair of vertical loads shown in Figure 4.7a alone causes points
A and B to deflect outwards, while the deflection due to the bar forces, F, alone
(Figure 4.7e) would be inwards and smaller than that due to P. The difference
between two deflections is the elongation of the bar, i.e.:

 H , P   H , F  b

which is effectively what equation 4.17 states.

There is only one unknown, F, in equation 4.17 and solving for F gives:

Pa 1 Pa
F   k (4.18)
b (1  3EIa ) b
b 2 sEb A

1
k (4.19)
3EIa
1 2
b sEb A

where k is smaller than 1 and is a non-dimensional coefficient that is related

to the geometry and cross-sectional properties of the inclined member of the
Y shaped column and the properties of the horizontal bar. If the rigidity of the
bar, Eb A, becomes infinite, then k = 1 and the tension force in the bar becomes
F = Pa / b . For this scenario, there are no bending moments in any of the mem-
bers making up the tied Y shaped column when subjected to the symmetric
vertical loads, i.e. the bending moments induced by P are balanced by the
bending moments induced by the horizontal force F in the rigid bar. This can
be demonstrated by calculating the bending moment at any point D of the right
aim of the tied Y column (Figures 4.6b and 4.7d) as follows:

Pa b
M D  P  x  F  y  Px   x  0 (4.20)
b a

While it is impractical that the bar rigidity Eb A could be infinite, this cor-
responds to the equivalent situation that the lateral deflections of the two top
ends of the Y shaped column are constrained by roller supports in the horizon-
tal direction.
 Smaller Internal Forces 101

Once the internal force of the bar, F, has been determined, the Y shaped col-
umn with a horizontal bar becomes a statically determinate structure as shown
in Figure 4.7d and the bending moment and deflections at the key positions
can be easily calculated.
The maximum bending moment in an inclined member is:

M 2,max  Pa  Fb  Pa  Pak  Pa(1  k)  M 1,max (1  k) (4.21)

The vertical downward deflection at A due to P and F is:

1 Pas 2a 1 Fbs 2a Pa 2 s Fabs

 2,V  ( )( )  ( )( )  
EI 2 3 EI 2 3 3EI 3EI
(4.22)
Pa 2 s Pa 2 ks Pa 2 s
   (1  k)  1,V (1  k)
3EI 3EI 3EI

The horizontal outward deflection between A and B due to the action of P and
F is:

1 Pas 2b 1 Fbs 2b 2Pabs 2Fb 2 s

 2, H  ( )( )  2  ( )( )  2  
EI 2 3 EI 2 3 3EI 3EI (4.23)
2Pabs 2Pabks 2Pabs
   (1  k)  1, H (1  k)
3EI 3EI 3EI

Equations 4.18 and 4.19 show that F is smaller than Pa/b as the coefficient
k is less than 1. When the cross-sectional properties, I and A, are measured
in meters, the value for area A would be much larger than that of I; a, b and
s are geometric dimensions of the cantilever arms and s > a. Therefore, the
ratio 3EIa / (b 2 sEb A) is likely to be much smaller than 1.0 for most practical
cases. Consequently, the coefficient k in equation 4.19 would not be much
smaller than 1. Equations 4.21 to 4.23 show that the maximum bending
moment at C and the vertical and horizontal deflections at node A of the
tied Y shaped column are (1-k) times of that of the same Y shaped column
without the horizontal bar.
In order to gain a feel for the effect of the horizontal bar on the maximum
bending moment and deflections of a Y shaped column, a particular case with
the following data is examined.
The inclined member of the Y shaped column has dimensions of a = 2.0m ,
b = 1.5m and s = 2.5m and uses an I section steel beam, UB254 x 102 x
25, with a second moment of area of I = 3415 cm 4 = 3.415  105 m 4. The
steel bar has a radius of 1.0 cm, i.e. a cross-sectional area of A = 3.14 cm2 =
3.14  104 m2 . The elastic modulus for both inclined members and the bar
are the same with E  Eb  200  109 N / m2. Vertical loads of 100kN act on
points A and B (Figure 4.6b).
102  Structural Design Against Deflection

Coefficient k is thus:

1 1 1
k    0.896
3Ia 3  3 . 415  10 5
 2 1  0 .1
1 16
(1  2 ) (1  )
b sA 1.52  2.5  3.14  104

The horizontal force in the steel bar can be determined using equation 4.18 as:

Pa 100, 000  2
F  k  0.896  119, 467 N
b 1 .5

The bending moment at C is:

M 2,max  Pa(1  k)  100, 000  2.0  (1  0.896)

 200, 000  0.104  20, 800Nm
m

The vertical deflection at A, from equation 4.22 is:

Pa 2 s 100, 000  22  2.5

 2,V  (1  k)  (1  0.896)
3EI 3  200  109  3415  108
 0.0488  0.104  0.0051m

The relative deflection between A and B based equation 4.23 is:

2Pabs 2  100, 000  2  1.5  2.5

 2, H  (1  k)  (1  0.896)
3EI 3  200  109  3415  108
 0.0732  0.104  0.0076m

It can be observed that the use of the horizontal bar effectively controls both
horizontal and vertical deflections of the two inclined members of the Y
shaped column, which leads to much smaller internal forces and deflections.
For this particular case, the reductions are significant, up to about 90% of
similar values for the normal Y shaped column. Therefore, it can be said that
the horizontal bar creates internal elastic supports to the tops of the Y shaped
column which leads to smaller deflections and internal forces. Alternatively,
it can be explained as the bending moments induced by the horizontal bar
partly balance those induced by the vertical loads, which results in much
smaller internal forces and consequently smaller deflections. These explana-
tions indicate that the physical measure of using a horizontal bar to tie the
two top ends of the Y shaped column can be generated from different ways
of thinking.
Figure 4.8 shows two practical examples in which tied Y shaped columns
have been used in past and present times. The tied Y shaped columns in Fig-
ure 4.8a are in the railway station in Knaresbough, North Yorkshire, UK. The
station was built in 1890, and the Y shaped columns were made of cast iron.
 Smaller Internal Forces 103

(a)

(b)
Figure 4.8 T he use of Y shaped columns with a horizontal bar. (a) At a train station.
(b) At an airport terminal.
104  Structural Design Against Deflection

It can be seen in Figure 4.8a that the tie member is in fact the lower chord of
a roof truss. This has a thicker section than that of the two arms which effec-
tively prevents the ends of the curved arms from deforming horizontally and
vertically. The use of curved arms instead of the conventional straight arms is
more aesthetically pleasing.
Figure 4.8b shows a straightforward implementation of the Y shaped col-
umn with a horizontal bar. This steel tied Y shaped column is in Terminal 5
at the Heathrow Airport, London, which was opened in 2008. It can be noted
that the horizontal bar has a small cross-section in comparison with that of the
two inclined members. The dimensions and properties of the Y shaped column
were estimated and used in the hand calculations. The use of the horizontal
bar was seen to reduce about 90% of the deflections and bending moments in
a similar Y shaped column without a horizontal bar.
The main differences between the tied Y shaped columns shown in Figure 4.8
are the materials used and the technology involved. In spite of the differences
in locations, construction times and materials, the structural concept embed-
ded in the two designs is essentially the same indicating that the implementa-
tion of structural concepts is not dependent on time or material.

4.3 Practical Examples

4.3.1  Structures with Overhangs

4.3.1.1  HSBC Hong Kong Headquarters, China

The beam with overhangs discussed in Section 4.2.1 is simple and efficient, and
the embedded physical measures of reducing span and partly self-balancing
bending moments can be applied to more advanced structures. Figure 4.9(a)
shows the tower of the HSBC Hong Kong headquarters, which has 47 storeys,
stands 179m above ground and was built between 1979 and 1986 [4.2, 4.3].
Figure 4.9b is a model of the building.
The main structure of the building is exposed allowing for direct apprecia-
tion. The building structure is supported by eight masts, arranged in two rows
of four (two masts can be seen in Figure 4.9a). Each mast consists of four
tubular steel columns which are rigidly connected by rectangular beams and
supported on foundations driven into bedrock over 30m below ground level.
Bracing members are provided between the masts which effectively increase
the lateral stiffness of the building structure. The masts support five discrete,
double, two-storey height, steel trusses which span 33.5m between the masts
and cantilever 10.7m beyond them. This mast and truss system carries all the
structural loads and creates a spectacular column-free area at ground floor
level. Each truss supports several lower floors on hangers at the centre and at
the two ends of the truss. Figure 4.10 shows clearly the top ends of the central
and side hangers, indicating that the truss supports the floors below. One of
the discrete truss systems is now chosen for a closer examination of the load
(internal force) paths.
 Smaller Internal Forces 105

(a)

Figure 4.9 T he tower of HSBC Hong Kong headquarters. (a) A front view. (b) The
model.

Due to the symmetry in elevation and for easy understanding, Figure 4.11

shows half of the elevation of the structure consisting of two columns with
link beams and the five storeys of floor beams supported from a truss by
two hangers (CG and FH). Two roller supports are provided to reflect the
symmetry and to prevent lateral deflections. The main floor beams are pin-
connected to the hangers and to the columns. The vertical loads acting on
the main beams at floor levels are transmitted to the columns and hangers
which generate compression forces in the columns and tension forces in the
hangers. The tension forces in the hangers are then transmitted to nodes C
106  Structural Design Against Deflection

(b)
Figure 4.9 (Continued)

and F, and then to members BF and AC in tension and CD and EF in com-

pression. It can be appreciated that the horizontal component of the force
acting on node B from member BF is partly self-balanced by the force acting
on node A from member AC. Similarly, the horizontal force acting on node
E from member EF is partly balanced by the force acting on node D from
member CD. The effect of the self-balancing effectively reduces the lateral
forces on the masts which consequently reduces the bending moments in
the masts.
Figure 4.11 indicates that the floor beams can be analysed as individual,
simply supported beams. Half of the loads on the beams are transmitted to
the supporting hanger and the other half to the mast. For a better under-
standing of the structural behaviour and the effect of self-balancing due to
 Smaller Internal Forces 107

Figure 4.10 T he truss supports the floors underneath rather than those above.

Figure 4.11 Illustration of a half of the elevation of the structure supported by one

truss [4.3].
108  Structural Design Against Deflection


(a) (b)
Figure 4.12 Models of the mast and truss system considering three trusses for illus-
tration. (a) The full model in which the action of hangers is represented
by point loads. (b) Simplified model based on symmetry.

the vertical loads, the structure can be further simplified to capture its physi-
cal essence for a hand calculation. Three levels of trusses and masts are con-
sidered as shown in Figure 4.12(a) with vertical loading applied through the
hangers, 2P from the central hangers and F from the side hangers. When a
symmetric structure is subjected to symmetric loads, the responses of the
structure will be symmetric and hence in this case a half of the structure
shown in Figure 4.12(b) can be considered, in which the central points of the
trusses are constrained to prevent from any horizontal movements, reflecting
the symmetry of deflection.
When the horizontal forces generated from the horizontal constraints in
Figure 4.12b are not considered for estimation, the analysis of the model in
Figure 4.12b becomes straightforward as the structure is statically deter-
minate. Examining the lateral forces acting on the mast from the truss
members (Figure 4.13a), there are six pairs of forces and that each pair of
forces acts at the same level but in opposite directions. After partial self-
balancing of the forces, six parallel forces are left and form three equal
pairs of forces at different levels (Figure 4.12b). The corresponding shear
 Smaller Internal Forces 109

     
(a) (b) (c) (d)
Figure 4.13 Loading and internal forces on the mast. (a) Loading from truss mem-
bers. (b) Forces after self-balancing. (c) Shear force diagram. (d) Bending
moment diagram.

force diagram and bending moment diagram for the mast are shown in
Figures 4.13c and 4.13d.
The magnitudes of the horizontal forces acting on the column are (Pa − Fb)/c
and the self-balancing of forces is reflected in the term Pa − Fb , in which force
F can be designed to achieve a more efficient structure. The qualitative struc-
tural model of the HSBC Hong Kong headquarters shown in Figure 4.12a is an
extension of a simply supported beam with overhangs (Figures 4.2b and 4.2c)
discussed in Section 4.2.1. Using overhangs reduces spans and creates partial
self-balancing of internal forces.

4.3.1.2  Roof of the Harbin Airport Lounge, China

Figure 4.14 shows a roof structure with overhangs used in the lounge of the
Harbin Airport terminal in China. The roof is supported by a series of trusses
that are in turn supported by circular columns. The columns are positioned
with a distance from the ends of the trusses and make the trusses to work like
beams with overhangs discussed in Section 4.2.1.
110  Structural Design Against Deflection

Figure 4.14 A  roof structure with overhangs in an airport terminal, Harbin, China.

4.3.2 Tree-Like Structures

4.3.2.1  Trees and Tree-Like Structures

The Y shaped column shown in Figure 4.5a is perhaps the simplest tree-like
structure. Tree-like structures, also called branching structures, are structural
forms developed from Y shaped columns with the addition of further branches.
Trees, exposed to sun, rain, wind and other environment conditions, are so
natural, logical and beautiful. Observing an oak tree in a winter (Figure 4.15)
it can be seen that: 1) there is a structural hierarchy with the trunk thickest at
the root of the tree and branches become thinner further away from the trunk.
2) The tree works effectively as a cantilever, i.e. the trunk is a vertical cantilever
and the many individual branches act as smaller cantilevers. As a cantilever,
it transmits the loads acting on it through bending, i.e. the bending moment
becomes the largest at the end of the cantilever and gets smaller toward its tip.
The thicknesses of the trunk and the branches of the tree basically reflect the
relative magnitudes of the bending moments that are experienced.
The inherent beauty and natural forms of trees have been used and improved
in architectural and structural designs in at least two ways:

1. The ends of branches have been used to support roofs or upper structures.
Due to the supports provided by the branches the roofs or upper structures
are able to span longer.
 Smaller Internal Forces 111

Figure 4.15 A n oak tree in winter showing the structural hierarchy

2. The ends of branches have been linked by structural members in the roofs
or upper structures which they support. Therefore, the branches no longer
act as cantilevers, so with the end deflections of the branches constrained
by the linking members they carry mainly axial forces rather than bend-
ing moments, which improves the efficiency and behaviour of the branch
members. This feature has been illustrated using a Y shaped column with
and without a tie in Section 4.2.2.

Section 4.2.2 demonstrates that the Y shaped column with a horizontal bar at

its two top ends has much smaller bending moments than a Y shaped column
without the bar under symmetric vertical loads, as the outward deflections of
the two top ends of the column are constrained by the bar. The Y shaped col-
umn with a horizontal bar retains the tree like shape but is far more efficient,
leading to the use of smaller cross-sections for the branches.
Consider the behaviour of a structure formed by a series of linked Y shaped
columns as shown in Figure 4.16a, in which rotational and horizontal con-
straints are provided at the two top end nodes of the frame. This is a highly
statically indeterminate structure, and simple hand calculations cannot be
used directly. However, the structural form and the loads are both symmetric,
and the property of symmetry can be utilised to simplify the structure and its
analysis. The structure shown in Figure 4.16a can be represented as an equiva-
lent half structure as shown in Figure 4.16b which is still symmetric. It can
112  Structural Design Against Deflection

(a)


(b) (c)

(d)
Figure 4.16 From a Y shaped frame to an equivalent single Y shaped column based
on symmetry. (a) A series of Y shaped columns forming a frame struc-
ture. (b) Equivalent frame to (a) based on symmetry. (c) Equivalent
frame to (b). (d) Equivalent frame to (c).

be similarly further simplified to a quarter of the original structure as shown

in Figure 4.16c, then to the single Y shaped column shown in Figure 4.16d.
Due to symmetry, this single Y shaped column with the rotational and hori-
zontal constraints has two redundant forces, the lateral force and the bend-
ing moment, at the top end supports. The support forces can be determined
by hand calculation, similar to the solution for the tied Y shaped column in
 Smaller Internal Forces 113

Figure 4.6b. The calculation shows that the pair of horizontal redundant forces
equal Pa/b acting inward to each other while the bending moments at the sup-
ports are zero. Therefore, this constrained Y shaped column (Figure 4.16d)
is equivalent to the tied Y shaped column (Figure 4.6b) when the rigidity
of the horizontal bar has an infinite value. Equation 4.20 shows that the
bending moment at any point in the inclined members become zero. i.e. the
bending moment induced by vertical force P is completely offset by that
due to the horizontal support force Pa/b. Because the structures shown in
Figures 4.16a and 4.16d are the same, i.e. one can be generated from the
other using symmetry, the members of the continuous Y frame structure
(Figure 4.16a) do not experience any bending moments due to the given loads.
This zero-moment scenario is created by constraining the lateral deflections of
the top nodes of the Y shaped column. It is noted that deformations due to
axial forces are negligible.
It is well known that a parabolic arch subjected to a uniformly distributed
vertical load experiences no bending moment. This continuous Y shaped frame
structure to the given loads (Figure 4.16a) is another example of a structure in
which all members have no bending moment. The former case is a single struc-
tural member while the latter is a frame structure consisting of several members.

4.3.2.2  Palazzetto dello Sport, Roma

The use of Y shaped columns can be seen in the structure of the Palazzetto dello
Sport (Small Sport Palace), shown in Figure 4.17, which was built in Rome in
1957 and engineered by Pier Luige Nervi. Figure 4.17 shows the form of the
structure in which the shell roof is supported by a series of circularly arranged
inclined Y shaped columns. Vertical columns are used to provide propping

Figure 4.17 Outlook of the Palazzetto dello Sport (Courtesy of Mr. Nicolas Janberg,
structurae.net, Germany).
114  Structural Design Against Deflection

(a)

(b)
Figure 4.18 FE model of the Palazzetto dello Sport [4.4]. (a) FE model showing the
three main parts of the structure, the roof, the series of Y shaped col-
umns and the tension ring. (b) The isolated circularly enclosed Y frame
structure.

supports to all the inclined Y shaped columns. Figure 4.18 shows a Finite Ele-
ment (FE) model of the structure and the isolated circularly arranged Y shaped
columns. As the exact data for the cross-sections of members of the structure are
not available, the FE model provides an illustration for qualitative understand-
ing of relative performance when several parameters of the structure are altered.
Due to the axisymmetry of the closed form of the series of Y frames to the
vertical line through the centre of the roof and their connections to the roof
shell, there is little lateral deflections occurring at the top ends of the Y shaped
columns when subjected to uniformly distributed vertical loads, such as the
self-weight of the roof structure. Following the qualitative analysis of the series
of Y shaped columns in Figure 4.16 and the quantitative analysis of the single
Y shaped column in Figure 4.6b, it can be inferred that there should be little
bending moments in the planes of the Y shaped columns.

4.3.2.3  Hessenring Footbridge, Germany

The Hessenring Footbridge, designed by Schlaich Bergermann Partner, has a
span of 46m and a width of 6.9m, and is located in Bad Homburg, Germany,
Figure 4.19. The slender bridge deck is suspended by 16 cables that transfer
 Smaller Internal Forces 115

Figure 4.19 H essenring Footbridge, Bad Homberg, Germany (Courtesy of Mr. Per
Waahlin, Sweden).

the loads on the bridge deck to a tree-like mast located at the centre of the
bridge. This mast is not only a loadbearing element of the bridge but a deli-
cate, three-dimensional sculpture as well. The bifurcation of four arms from
the truck, the central column, is an evolution of the Y column in two perpen-
dicular directions. The four horizontal members linking the four ends of the
arms confine the outward deflections of the arms due to the actions of the 16
cables, by which the four arms are mainly subjected to axial forces rather than
bending moments. This deduction can come not only from the analysis of the
Y shaped column with a tendon in section 4.2.2 but also from the observation
that the arms have a similar cross-section along their lengths. If the four hori-
zontal members were removed, the four arms would act like cantilevers with
concentrated cable forces at their free ends. This would generate the bending
moments in a triangular shape along the lengths of the arms with zero at their
top ends and the maximum at their bottom ends.

4.3.2.4  Further Examples

Y shaped columns can be arranged in three dimensions to form tree-like struc-
tures. Figure 4.20 shows the structure of the Gare do Oriente Station in Lisbon,
designed by Santiago Calatrava, where the branches of the trees are curved
rather than straight. This variation would not affect the bending moments
in the members of the columns under uniformly distributed vertical loads. In
116  Structural Design Against Deflection

Figure 4.20 A  tree-like structure as an evolution of a series of Y shaped columns.

addition, the folded roof and many thin elements between branches constrain
the relative deformation between the branches and between the columns, lead-
ing to only small bending moments. Therefore, no thick vertical columns and
branch members are required in this structure.
Tree-like structures have been used creatively, and many variations have
been produced to achieve aesthetic beauty and structural efficiency. As the tree-
like structures have fewer columns but many more branches, they are able to
provide good supports to roofs and are particularly suitable to be used in open
spacious areas. Therefore they are often seen in shopping malls, exhibition cen-
tres and airport terminals. Figure 4.21a shows the Pu Dong Airport Terminal,
Shanghai, in which the roof is supported by a series of Y shaped columns. The
first level of branches of the Y shaped column is further divided into a second
level of branches perpendicular to the ones in the first level creating four point
supports for the roof. The connections between the roof and the tops of the Y
shaped columns restrain the horizontal and vertical deformations of the four
branches due to vertical loads. Thus, the members of the Y shaped columns are
subjected mainly to compressive forces rather than bending moments result-
ing in lighter sections. The appearance of the thick vertical columns in Fig-
ure 4.21a is the use of the additional non-structural materials for architectural
reasons and for the protection to the passengers.
Figure 4.21b shows the huge Y shaped columns used in the Bihai Cultural
Centre in Tianjin, China, to support the roof over a large open area. It can be
noted that eight branch members are developed from the central column to
support the roof structure and that the top ends of the branches are connected
to rigid roof members. The branch members are thus constrained to deform in
the horizontal and vertical directions leading to only low bending moments in
the branch members for vertical loads.
In the Madrid Barajas Airport terminal there are intensive uses of V shaped
struts and Y shaped columns to support its roof structure, which allow the roof
spanning over large areas without intermediate supports. Figure 4.22 shows
the internal and external inclined Y shaped columns. The common feature of
 Smaller Internal Forces 117

(a)

(b)
Figure 4.21 Y shaped columns used for large public buildings. (a) Y shaped columns
with two levels of branch used in Pu Dong Airport Terminal. Shanghai,
China. (b) Large Y shaped columns used in the Bihai Cultural Centre,
Tianjin, China (Courtesy of Mr. Peixuan Xie, UK)
118  Structural Design Against Deflection

(a)

(b)
Figure 4.22 Inclined Y shaped columns in the Madrid Barajas Airport terminal, Spain.
(a) Internal use (Courtesy of Professor Zhaohui Chen, Chongqin Uni-
versity, China). (b) External use (Courtesy of Professor Guy Warzée—
Université Libre de Bruxelles, Belgium).
 Smaller Internal Forces 119

these Y columns is that horizontal members are placed between the top ends of
the Y columns. These horizontal members confine the lateral deflection of the
ends and make the arms to subject mainly the axial forces.

4.3.3 Self-Balancing

4.3.3.1  Madrid Racecourse, Spain

Figure 4.23 shows the stand at the Madrid Racecourse (the Zarzuela Hippo-
drome), and a cross-section drawing and a physical model are shown in Fig-
ure 4.24. Figure 4.24 shows that the stand consists an upper roof or canopy,
a seating area on the left and a betting hall on the right, which is covered by a
lower roof. The upper roof is supported by central columns with pinned con-
nections and rods, CD, between the upper and lower roofs. The left-hand side
of the lower roof is rigidly connected to the central columns and is suspended
at mid span by the rods, CD [4.5].
There are several merits of the structural design of the stand, but the partial
self-balancing system embedded in the structure is of interest here. It is noted
that rod CD (Figure 4.24a) is placed between the upper roof for the stand and
the lower roof for the betting hall. The upper roof is supported by the vertical
upward forces from the central columns and the downward forces from ten-
sions in the rods. The weight of the lower roof is largely carried by the rods due
to their locations at about the mid span of the lower roof. As the rods connect

Figure 4.23 A  front view of the Madrid Racecourse Stand, Spain.

120  Structural Design Against Deflection

(a)

(b)
Figure 4.24 T he design of the Madrid Racecourse Stand. (a) Cross-section drawing.
(b) Physical model.
 Smaller Internal Forces 121

Figure 4.25 A  simplified model of the two roofs and the rod linking them.

the upper and lower roofs, the two roofs and the rods form a partially self-
balanced system. The interpretation and the physical essence of the system can
be illustrated and investigated using the diagram in Figure 4.25 in which the
relationships between the upper and lower roofs and the rod CD are presented.
The upper roof requires downward forces to achieve its equilibrium while the
lower roof needs upward forces to stiffen it and reduce its internal forces and
deflections. The placement of rods, CD, serves the two purposes and makes
that the upper roof FG and the lower roof BE are mutually supported.

4.3.3.2  Salford Quays Lift Bridge, UK

Using self-balancing to solve challenging engineering problems often achieves
efficient designs. Figure 4.26 shows a lifting-up position of the Salford Quays
lift bridge, also known as the Salford Quays Millennium Footbridge or the
Lowry Bridge, that spans the Manchester Ship Canal between Salford and
Trafford in Greater Manchester, England. The 91.2m long vertical lift bridge
has a lift of 18m, allowing large watercraft to pass beneath.
The bridge consists of a pair of inward inclined arches that are connected
at the crown, a bridge deck and a series of cables that are uniformly spaced
along the length of the deck and linking the arch and the deck. Most of the
self-weight of the deck and the traffic loads applied on the deck are transmitted
to the cables then to the arches. An arch is effective to transmit external loads
through mainly compression to its supports. However, it requires strong, sub-
stantial supports to balance the large horizontal forces at the ends of the arch.
For this bridge (Figure 4.26), the ends of the arches and the deck are rigidly
integrated and thus the deck, which is a bending member, is also used to bal-
ance the thrusts from the arches. Since the deck has sufficient axial strength to
carry the tension resulting from the arch thrusts no other supports are needed
to balance the horizontal forces from the arches. The idea for using the bridge
deck to balance the arch thrusts appeared to come out of solving the problem
that no external horizontal supports need to be provided for a lift bridge.
A similar example, shown in Figure 4.27, is an arch supported bridge for trams
in Manchester, England. The integrated arch-cable-deck system achieved self-
balancing of horizontal forces, and was a solution of the problem that the site did
not allow for building supports to balance the horizontal forces from the arches.
122  Structural Design Against Deflection

Figure 4.26 Salford Quays Lift Bridge.

Figure 4.27 T he arch, deck and cables of the bridge form a self-balancing system.

4.4 Further Comments
The efficiencies of a beam with overhangs and a Y shaped column with a tie
at its two top ends have been examined independently in Sections 4.2.1 and
4.2.2, and their implementations have been demonstrated in Sections 4.3.1
and 4.3.2 respectively. However, it is possible and effective to integrate the
two physical measures into one design simultaneously. The Chengdu East
Railway Station, which is one of the largest railway hubs in China and the
largest in the West region of the country, is such an example. The station
building was constructed in 2011 [4.6].
 Smaller Internal Forces 123

Figure 4.28a shows the front view of the railway station in Chengdu in

which the overhang roof and Y shaped columns can be seen. Looking at one of
the steel Y shaped roof supports (Figure 4.28a), the lower part of the Y shaped
column is split into two inclined members that are linked by a metal piece
with three pairs of short, horizontal members. The two branch members of

(a)

(b)
Figure 4.28 Chengdu East Railway Station. (a) Front view showing the overhanging
roof with Y column supports. (b) One of the four roof supports show-
ing the Y column supports in two perpendicular directions (Courtesy of
Professor Yuan Feng, China Southwest Architectural Design & Research
Institute, China).
124  Structural Design Against Deflection

the Y shaped column evolve into pairs of loops. Figure 4.28b shows one of
the four roof supports, which can be seen as a Y column being split into four
looped branches that are linked by horizontally parallel members. The widely
spread branches provide four point supports to the roof structure. As the four
top ends are connected to the roof structure, their deflections in the two hori-
zontal directions are constrained, which also limits the bending moments in the
members of the Y shaped column.
In this chapter, only vertical loads have been considered. In reality, lateral
loads are of the same importance as the vertical loads, and the action of the
lateral loads will be discussed in Chapter 6.

References
4.1 Craig, R. R. Mechanics of Materials, John Wiley & Sons, USA, 1996.
4.2 Bennett, D. Skyscrapers: Form and Function, Simon & Schuster, New York, USA,
1995.
4.3 Parkyn, N. The Seventy Architectural Wonders of Our World, Thames & Hud-
son, London, 2002.
4.4 Xu, L. The Y-Shaped Structures, MSc Dissertation, The University of Manchester,
2015.
4.5 Torroja, E. The Structures of Eduardo Torroja: An Autobiography of an Engi-
neering Accomplishment, F W Dodge Corporation, USA, 1958.
4.6 Feng, Y. et al. Practice on Long-Span Spatial Structures, China Construction
Press, Beijing, China, 2015.
Chapter 5

More Uniform Distribution

of Internal Forces

5.1 Routes to Implementation
Achieving more uniform distribution of internal forces will lead to smaller
internal forces. Therefore, the routes to implementing the structural concept of
smaller internal forces presented in Section 4.1, such as using self-balancing,
internal and external elastic supports, etc., are all applicable to realising more
uniform distributions of internal forces. However, creating more uniform dis-
tributions of internal forces provides an alternative way of thinking and can
lead to a topology optimisation method for achieving more efficient structures.

Topology optimisation of structures: Evolutionary Structural Optimisation

(ESO) and its later development, bi-directional Evolutionary Structural Optimi-
sation (BESO) [5.1, 5.2], are a kind of topology optimisation method in which
a structural concept is embedded. By gradually removing inefficient materi-
als with the lowest stresses from a structure and adding material to the most
stress demanding region, an optimum topology of the structure evolves with
the remaining elements having a smaller difference between the highest and
lowest stresses. The maximum stress difference between remaining elements
of the structure gradually becomes smaller through repeating this process. The
outcome from BESO is an efficient design that wisely uses material. This opti-
misation process corresponds to the structural concept: the more uniform the
distribution of internal forces or stresses, the more efficient the structure.
Using BESO based on the finite element method implemented on a computer
will lead to creative solutions for a wide range of structures, some of which can
be imaginative and even beyond what an experienced engineer could think of.

5.2 Hand Calculation Examples

5.2.1 A Cantilever with and without an External

Elastic Support
This example shows that the provision of an external elastic support reduces
the bending moments or makes more uniform distributions of bending
moment and leads to smaller deflections.
126  Structural Design Against Deflection

Figure 5.1 shows three vertical cantilevers that have the same height of L
and same cross-sectional rigidity of EI and are subjected to a uniformly dis-
tributed lateral load of q. The difference between the first two cantilevers is
that Cantilever 1 is an unrestrained cantilever and Cantilever 2 is a cantilever
with a horizontal spring support at its top end. The spring for Cantilever 2 has
the stiffness of Kx. The differences between the Cantilever 2 and Cantilever 3,
which also has a spring support at its top end, are that the load and the spring
have angles ϕ and θ to the cantilever. Calculate and compare the bending
moments at the bases and the deflections at the tops of Cantilevers 1 and 2.

For Cantilever 1, the maximum bending moment at the base and the maxi-
mum deflection at the free end of the cantilever are respectively [5.3]:

qx L2
M 1b = (5.1)
2
qx L4
1  (5.2)
8EI

For Cantilever 2, which is a statically indeterminate structure, the spring

force needs to be determined before calculating the bending moment and
deflection of the cantilever. The spring action can be replaced by a spring force,
Fx, to be determined, which can be expressed as a product of the stiffness Kx
and the deflection ∆2 of the spring. ∆2 is the summation of two deflections, ∆2q
and ∆2s, induced by the distributed load and by the spring force on the stati-
cally determinate cantilever respectively:

q L4 F L3
Fx  kx  2  kx ( 2q   2s )  kx ( x  x ) (5.3)
8EI 3EI

           
(a) (b) (c)

Figure 5.1 A  cantilever subjected to uniformly distributed loads. (a) Cantilever 1.

(b) Cantilever 2: with an external spring support at its top end. (c) Can-
tilever 3: the load and the spring support are inclined to the cantilever.
 Uniform Distribution of Internal Forces 127

Re-rearranging equation 5.3, the expression of the spring force Fx can be

obtained, step by step, as:

qx L4 kx L3
kx
8EI  3qx L 3EI  3qx L kx 3q L 
Fx   x (5.4)
kx L3 8 kx L3 8 K sc  kx 8 1
1 1
3EI 3EI

in which:

3EI k
k sc = ;α= x  (5.5 a, b)
L3 K sc

where Ksc is the static stiffness of the cantilever and is the inverse of the dis-
placement at the top of the cantilever due to a unit load, and α is the ratio of
the spring stiffness to the static stiffness of the cantilever. It can be observed
from equations 5.4 and 5.5 that:

• When the spring has an infinitive stiffness, kx  , it becomes a roller sup-

port and then Fx = 3qx L / 8. This is just the reaction force of a propped
cantilever at the roller support.
• The spring force Fx depends on the ratio of the spring stiffness to the static
stiffness of the cantilever.

With the spring force known, the cantilever with a spring support becomes
a statically determinate structure and the bending moment and deflection at
any point of the beam can be easily calculated. For illustration, consider the
bending moment at the base and the deflection at the top of the cantilever. The
superposition method can be used for calculation:

1 1 3 
M 2b  qx L2  FL  qx L2  qx L2
2 2 8 1
1 3  1
 qx L [1 
2
]  qx L  fM
2
(5.6)
2 4 1 2

qx L4 3qx L  L3
 2   2q   2 F  [ ]
8EI 8 1   3EI
(5.7)
qx L4  qx L4 1 qx L4
= (1  )   f
8EI 1 8EI 1   8EI

3  3 Kx
fM  1   1  (5.8)
4 1 4 K sc  K x

1 K sc Kx
f    1 (5.9)
1   K sc  K x K sc  K x
128  Structural Design Against Deflection

where ƒM and ƒ∆ are the spring effect factors for the base bending moment
and for the top deflection of the cantilever respectively, which describes what
reductions are achieved due to the spring effect. It can be observed from equa-
tions 5.6–5.9 that:

• When kx  , it becomes a propped cantilever. Thus, M 2b = qx L2 / 8 and

there is no deflection at the propped position.
• When kx = 0, it becomes a cantilever and the bending moment at the base is
=
M 2b M=1b qx L2 / 2 and the deflection at the top is  2  1  qx L4 / (8EI )
• When Kx is between the two extremes, the larger the spring stiffness, the
larger the spring force and thus the smaller the bending moment at the
base and the smaller the deflection at the top of the cantilever.

Equations 5.8 and 5.9 indicate that ƒM and ƒ∆ are functions of the ratio of the
spring stiffness to the static stiffness of the cantilever. To appreciate the effect
of the stiffness ratio on the reduction of the responses, these two functions are
plotted in Figure 5.2.
Figure 5.2 indicates that the spring can effectively reduce the base bending
moment and the top deflection of a cantilever, and the rate of the reduction of
the bending moment at the base becomes small when α is larger than 3.

For Cantilever 3, which is a development of Cantilever 2 in Figure 5.1b

obtained by inclining the spring support with an angle of θ and the distributed
load with angle of ϕ. When θ = 900, and ϕ =900, Cantilever 3 becomes Cantilever 2.
Equations 5.3–5.8 are applicable to Cantilever 3 by introducing kx = k sin2θ,
Fx = F sinθ and qx  q sin  , where kx is the spring stiffness k projected to the
horizontal (x) direction. F is the spring force and Fx is the horizontal projection
of F. Similarly, qx is perpendicular to the cantilever, which is the projection of
the load q to the x direction. The derivation of kx can be seen in the example in
Section 6.2.1, in which inclined strings, similar to the spring, are used.

5.2.2 A n Eight Storey, Four Bay Frame with Different Bracing

Arrangements
This example shows that the computer application of the structural concept
of the uniform distribution of internal forces can generate new bracing pat-
terns that are even more efficient than those based on the structural concept
of direct internal force path.
Figure 5.3 shows an eight storey, four bay frame with five different bracing
arrangements. Frames A-D have the same dimensions, the same numbers of
bracing, vertical and horizontal members. All the members have the same cross-
sectional area and elastic modulus. Two bracing members are placed on each
of the eight storeys in Frames A-D, making a total 16 bracing members. The
only differences between the four frames are the bracing patterns. The brac-
ing pattern in Frame A has been discussed in Chapter 3 and can be formed by
using the first three criteria in Section 3.1. The inverted V pattern in Frame B is
 Uniform Distribution of Internal Forces 129

(a)

(b)
Figure 5.2 Spring effect factors as functions of the stiffness ratio. (a) For the base
bending moment. (b) For the deflection at the top of the cantilever.

developed intuitively based on the structural concept of smaller internal forces.

Frames C and D are developed using the Evolutionary Structural Optimisation
for continuous bodies by applying it to discrete systems. Frame E is initially
fully braced so that each panel has two braces. A pair of anti-symmetric loads is
130  Structural Design Against Deflection


(a) (b)


(c) (d)
Figure 5.3 A  frame with four different bracing arrangements. (a) Frame A: Global
X bracing derived from the structural concept of direct force paths.
(b) Frame B: Inverted V bracing derived from the structural concept of
smaller internal forces. (c) and (d) Frames C and D with bracing patterns
derived from the ESO approach based on the structural concept of more
uniform distribution of internal forces. (e) Frame E: Fully braced frame as
a start for the ESO approach.
 Uniform Distribution of Internal Forces 131

(e)

Figure 5.3 (Continued)

applied to the two top corner nodes as shown in Figure 5.3. Finite element struc-
tural analysis of the pin-connected frame is conducted and the pair of symmetric
bracing members that have the smallest strain energy (or stress) are identified
and removed from the structure. This process is repeated until only two bracing
members are left in each storey. Frames C and D are the outcomes from ESO.
Due to the removal of the bracing members with the lowest stain energy, the two
evolved structures have smaller differences in strain energy between the remain-
ing members, which is equivalent to a realisation of more uniform distribution
of internal forces [5.4]. Using Frames A-D shown in Figure 5.3, the internal
forces and the maximum deflections of the four frames can be examined.
The pin-connected frames are statically indeterminate structures. However,
they can be simplified into equivalent statically determined structures follow-
ing the concept: when a symmetric structure is subjected to anti-symmetric
loads, the responses (deflections and internal forces) of the structure will be
antisymmetric, which has been used in Chapters 2 and 3. As the responses
are anti-symmetric, the members in the central line of the frame must be zero
and there are no vertical deflections of the nodes on this line. Therefore, the
equivalent half frames are shown in Figure 5.4 in which the internal forces of
all members are indicated to appreciate their magnitudes and distributions.
The internal forces of all members of the four halved frames can be calculated
by hand using the equilibrium equations at each of the nodes. As the length
and width of each panel are the same, the hand calculation can be quickly
conducted.
132  Structural Design Against Deflection


(a) (b)
Figure 5.4 T he internal forces for the four equivalent half frames based on those
shown in Figure 5.3. (a) Frame A equivalent. (b) Frame B equivalent.
(c) Frame C equivalent. (d) Frame D equivalent.
 Uniform Distribution of Internal Forces 133


(c) (d)
Figure 5.4 (Continued)
134  Structural Design Against Deflection

To understand the differences of the internal forces between the four

frames, Table 5.1 summarises the distribution and magnitude of the internal
forces in the vertical and horizontal members (columns 2–5) and in the brac-
ing members (column 6) based on Figure 5.4 and their contributions to the
deflection (columns 7 and 8) and the total contributions (column 9) of the
four halved frames.
It can be noted and understood from Figure 5.4 and Table 5.1 that:

• The bracing members have the same magnitude of 2 / 2 in all four

frames. Therefore, the difference between the deflections of the four frames
is controlled by the internal forces in the vertical and horizontal members.
• The bracing arrangement in Frame A, following the concept of more direct
path of internal forces, has the highest number (24) of zero-force members
but gives the largest deflection among the four. This is because there are
four members with the largest internal force of 3/2 (Figure 5.4a), which
makes a significant contribution to the total deflection.
• In comparison with that of Frame A, the inversed V-shape bracing pattern
have six more members with a small internal force of 1/2 and two more
members with a force of 1 but two less members with the largest magni-
tude of internal force of 3/2. Due to the square effect on the forces in equa-
tion 2.16, this bracing pattern results in a smaller deflection than that of
Frame A with a global X bracing pattern.
• In comparison with Frame A and B, Frames C has higher numbers of mem-
bers with non-zero force and with small magnitude of internal force (1/2) but
a smaller number of members with the largest magnitude of internal force
(3/2). Therefore, Frame C results in even smaller deflection than Frame B.
• Frame D has the same deflection as Frame C by reducing four members
with an internal force of 1/2 but increasing one member with an internal
force of 1. The two sets of members have the same contributions to the
deflection.

Table 5.1 The number of members at different magnitudes of internal forces and
their contributions to deflections of the four frames

Internal forces (IF) in IF in the Contributions Contributions Contributions

the vertical (V) and Bracing from V and from bracing From all
horizontal (H) members members H members members members
(1) (2) (3) (4) (5) (6) (7) (8) (9)
V, H B All
0 ½ 1 3/2 2 /2 ∑N 2
L ∑N 2
L ∑N 2
L

Frame A 24 4 0 4 8 10 4 2 15.65
Frame B 18 10 2 2 8 9 4 2 14.65
Frame C 16 13 2 1 8 7.5 4 2 13.16
Frame D 19 9 3 1 8 7.5 4 2 13.16
 Uniform Distribution of Internal Forces 135

When a bracing member ends to a vertical member, the internal force in the
vertical member that is lower than the intersection point will increase by 1/2.
For example, there are two bracing members that end at the intersection point
(S) of the outside vertical members at the mid-height of Frame A in Figures 5.3a
and 5.4a. Therefore, the internal force increases 2  1/2  1 from the upper
member of the intersection point to the lower member. To avoid larger accu-
mulated internal forces in the outside vertical members, two bracing members
end at two inside columns next to the outside ones at the six level of Frame C
in Figure 5.3c. This leads Frame C to more members having smaller internal
forces and fewer members experiencing larger internal forces in comparison to
Frame A (Figures 5.4a and 5.4c).
The four bracing patterns are generated based on different structural con-
cepts, the more direct internal force path (Frame A), the smaller internal forces
(Frame B) and more uniform distribution of internal forces (Frames C and D).
Therefore, the outcomes shown in Table 5.1 encourage to think retrospectively
and conceptually the reasons that the frame with the last three bracing patterns
(Figures 5.4b, c, and d) perform even better than the globally X braced frame
(Frame A), which helps to develop ideas for wiser designs. On the other hand,
ESO would be able to create new structural forms that may be beyond what
one can image for.

5.3 Practical Examples

5.3.1  Structures with External Elastic Supports

5.3.1.1  Samuel Beckett Bridge, Dublin

A simple harp, shown in Figure 5.5a, consists of three relatively thick external
members, the neck with harmonic curve, a sound box and a pillar, which form
a loop, and a number of parallel taut strings with different lengths between
the neck and the sound box. The function of the pillar is to support the neck,
prevent relative deformation between the neck and the sound box due to the
action of the taut strings, and transmit its supporting force from the neck to its
lower end and to the base.
For analysing the forces in a harp, a hand drawing of the essence of a harp
similar to that in Figure 5.5a is shown in Figure 5.5b. If the pillar is removed
from the harp, its action forces on the neck and the sound box and the remain-
ing structure are as shown in Figure 5.5(c). The downward force, F, acting on
the sound box would be transmitted to the base that supports the sound box.
Therefore, supports are qualitatively provided to allow the remaining parts
of the harp in an equilibrium position. The upward compressive force, F, is
necessary to support the neck. From a structural point of view, this upward
compressive force can be replaced by a suitable tension force, T, as shown in
Figure 5.5(d), to resist the downward deformation of the neck.
The idea of the harp as illustrated qualitatively in Figure 5.5(d) has been used
in engineering practice in the Samuel Beckett Bridge, which is a cable-stayed,
136  Structural Design Against Deflection

steel box girder, swing bridge over the River Liffey in Dublin. This bridge,
designed by Santiago Calatrava, and has become a landmark of the city reflect-
ing a harp which is the national symbol of Ireland. The side view of the bridge
shown in Figure 5.6a looks like the equivalent harp shown in Figure 5.5d
in which the cables between the bridge deck and the pylon resemble strings,
the pylon acts like the neck of the harp and the bridge deck is similar to the
music box. The backstay cables provide tension forces to limit the forward
and downward deformations of the pylon due to the action of the main cables.

(a)

     
(b) (c) (d)

Figure 5.5 A  harp. (a) The harp displayed at the Guinness factory, Dublin. (b) Hand
drawing of a harp similar to that in (a). (c) Replacing the pillar with two
forces. (d) The upward compressive force F is replaced by a tension force T.
 Uniform Distribution of Internal Forces 137

(a)

(b)
Figure 5.6 T he Samuel Beckett Bridge, Dublin. (a) Side view. (b) Back view.

The bridge is 123m in length and 28m in width and carries four lanes of traf-
fic with cantilevered pedestrian and cycle tracks. The bridge is able to rotate
through 90 degrees to allow shipping to pass below. Figure 5.6 presents the
side and back views of the bridge, which shows the structure of the bridge and
its use. The steel box girder bridge is lifted by 25 stay cables of 60mm diam-
eter from a steel cantilever that is supported and stabilised by six back cables
138  Structural Design Against Deflection

of 145mm diameter. The positions of the six back cables also strengthen the
lateral stability of the pylon [5.5].
Figure 5.1c also suggests a plane model for the analysis of the pylon. The
pylon is subjected to a series of parallel cable forces that are about perpendicu-
lar to the pylon while the back-stay cables between the top of the pylon and
their anchor points act like spring supports to constrain the downward and
forward deflections of the pylon. In this case the back cables are anchored to
the foundations that are independent from the bridge and thus can be consid-
ered to act as an external spring support to the pylon.

5.3.1.2  Serreria Bridge, Valencia

There are other harp-like bridges which have been designed by Santiago Cala-
trava. The Serreria Bridge, shown in Figure 5.7, is situated within the City of Arts
and Science Complex in Valencia, Spain. It has a span of 180m and a width var-
ying from 33.5m to 39.2m. In addition to three external vertical supports, the
bridge deck is suspended by 29 parallel stay cables with a spacing of 5m from
an inclined curved pylon that rises to a height of 118.6m. It can be observed
from Figure 5.7 that the pylon leans backwards which enables the self-weight of
the pylon to balance some of the applied forces from the stay cables supporting
the bridge deck. The two groups of back-stay cables provide effective external
spring support to the pylon at its top and are placed slightly more apart from
each other on ground to improve the lateral stability of the pylon.

Figure 5.7 T he Serreria Bridge, Valencia, Spain (Courtesy of Mr. Nicolas Janberg,
structurae.net, Germany).
 Uniform Distribution of Internal Forces 139

5.3.1.3  Katehaki Pedestrian Bridge, Athens

The Katehaki pedestrian bridge is 93.7m long and spans 73.5m between two
supports. The width of the deck varies from 3.95m at one end to 5.67m at the
other end. The curved steel-boxed pylon has a height of 50.48m and appears
to lean backwards, while the pylon in the Samuel Beckett Bridge leans forward.
The bridge deck is suspended from the pylon by 14 parallel cables. Two back-
stay cables are able to transmit a significant portion of the parallel cable forces
to their foundations. The bending moments in the pylon caused by the back-
stay cables offset part of the bending moments due to the parallel cable forces.

The three bridges are functionally, geographically and architecturally differ-

ent, but they demonstrate similar technical elegance and lightness, achieved by
providing external elastic supports:

1. In addition to the solid vertical supports, the bridge decks are suspended
by a series of cables that act as external spring supports to the decks.
2. The back-stay cables provide external spring supports to the pylons that
act like cantilevers. The effect of such spring supports on a cantilever has
been demonstrated in Section 5.2.1

For a quick hand analysis at a conceptual design stage, the pylons in the
three bridges can be considered as cantilevers with spring supports at their
free ends (Figure 5.1c). The forces acting on the pylons from the parallel fore-
stay cables linking to the bridge decks may be treated as uniformly distributed
loads, which may not be perpendicular to the pylons and can be described by
an angle ϕ. The back-stay cables can be simplified as an external spring sup-
port at the free end of the pylon, and θ is used to define the angle between the
cables and the pylon.

5.3.2 Structures with Internal Horizontal Elastic Supports

The provision of internal elastic supports is perhaps an effective and simple
way to self-balance internal forces which will result in smaller internal forces
and more uniform distribution of internal forces, and smaller deflections.
There are several creative applications of internal elastic supports in the hori-
zontal direction.

5.3.2.1  Manchester Central Convention Complex, UK

Figure 5.9 shows the Manchester Central Convention Complex that has a dis-
tinctive arched roof with a span of 64m. The Complex was originally designed
in 1880 and subsequently used as the Manchester Central Railway Station. The
roof arches were made of wrought iron. Arches are effective structures as they
transfer applied loads mainly through compression, rather than by bending,
to their supports. However, arches normally generate large horizontal forces
140  Structural Design Against Deflection

Figure 5.8 Katehaki pedestrian bridge, Athens.

Figure 5.9 F ront view of the Manchester Central Convention Complex (MCCC),
Manchester.

at supports, which require large foundations. Normally pinned supports are

provided at the two ends of an arch to resist both vertical and horizontal
forces. It can be observed on the arch shown in Figure 5.9 that there are two
substantial horizontal members, one toward the bottom of the arch and one
around mid-height of the arch. The self-weights of the two horizontal members
 Uniform Distribution of Internal Forces 141

are transmitted through the vertical bars to the arch. The two horizontal mem-
bers have large axial stiffnesses and effectively act as internal spring supports
to the arch in the lateral direction, which restrains lateral deformations of the
arch and balances part of the horizontal component of the internal forces in
the arch. This in turn reduces the internal forces in the arches and reduces the
horizontal thrusts at the arch supports.
To examine the effects of the horizontal members on the reduction of internal
forces and deflections in an arch in both horizontal and vertical directions qualita-
tively, the main characteristics of the arch (Figure 5.9) can be extracted as shown
in the simplified model (Model A) in Figure 5.10a, in which the two horizontal
members and the boundary conditions of the arch are shown. It may not be an
easy task to produce a physical model like that in Figure 5.10a because sup-
ports and connections between the supports and the arch are concerned. Using
the concept of symmetry (when a symmetric structure is subjected to symmetric
loads, the response of the structure will be symmetric), Model A (Figure 5.10a) is
just a half of Model B (Figure 5.10b). The advantage to use Model B to replace
Model A is that the supports required in Model A can be removed for model
making. For examining the effect of the horizontal members on the ring, Model

(a)

 
(b) (c)
Figure 5.10 Evolution of models for intuitive understanding. (a) Model A: the arch-
bar model based on MCCC. (b) Model B: The model is equivalent to
Model A based on symmetry. (c) Model C: This model is a simplification
of Model B, which captures the physical essence of Model B.
142  Structural Design Against Deflection

B can be altered to Model C (Figure 5.10c) in which the physical essences of the

horizontal members and the arch remain. The changes between Models B and C
are the replacement of the four horizontal members by a single member and the
replacement of the pin connections by rigid connections. The behaviour of Model
C is similar to that of either Model A or Model B, and it is easier to make a physi-
cal model of Model C than Models A and B.
Due to the action of the vertical forces, P, the horizontal member in Model C
experiences a tensile force T. The behaviour of the ring without a tendon but
still subjected to forces P and T can be analysed qualitatively using the super-
position method. Considering the pair of vertical forces P alone, the deforma-
tion of the ring is illustrated by the dashed line in Figure 5.11a, in which the
top and bottom points B and D deform toward to each other while the side
points A and C deform outwards from each other. Examining the action of the
horizontal forces T, the ring deforms in the opposite direction to that induced
by P (Figure 5.11b). Therefore, the behaviour of Model C is a combination of
the deformations shown in Figures 5.11a and 5.11b. The action of the hori-
zontal member confines the outward deformation of A and C and reduces
the vertical deflection of the ring due to P. The tied ring is much stiffer than
the corresponding ring without a tie. This interpretation can be demonstrated
using physical models.
Figure 5.12 shows two rubber rings, one with and one without a wire
tied across the centre of the ring. The same weight of 22.3N is placed on the
top of each of the two rings, and the reduced deformation of the tied ring
is apparent. The reduced deflection means an increased stiffness of the ring


(a) (b)
Figure 5.11 Illustration of the deformations of the tied ring. (a) A ring subjected a
pair of vertical forces and its deformations. (b) A ring subjected to a
pair of horizontal forces and its deformations.
 Uniform Distribution of Internal Forces 143

Figure 5.12 D emonstration of the effect of the wire that acts as internal elastic
supports to the ring.

which can be felt easily by pressing down on the two rings. The reduced
deflection also indicates that the tied ring experiences smaller bending
moments. This is because the force in the wire increases as the applied load
increases and produces a bending moment in the ring in the opposite direc-
tion to the bending moment caused by the external load. Thus, the force in
the wire balances part of the bending moments in the ring due to the vertical
load, resulting in smaller and more uniform internal forces. As the tied ring
is doubly symmetric and relatively simple, the expressions for its vertical and
lateral deformations and bending moments can be derived and quantitatively
examined [5.6].

5.3.2.2  Raleigh Arena, USA

The roof structure of the Raleigh Arena shown in Figure 5.13 consists of car-
rying (sagging) cables and stabilising (hogging) cables, which are supported by
a pair of inclined arches. The carrying cables apply large forces to the arches
and some of the vertical components of these forces are transmitted to external
columns. The bending moments, shear forces and compressive forces are trans-
mitted through the inclined arches to their supports. Most of the horizontal
forces at the ends of the arches are balanced by underground ties which reduce
significantly the horizontal forces on the foundations. The underground ties
or tendons have a similar function to the wire tie in the rings demonstrated in
Figure 5.12, acting as internal elastic supports, reducing the internal forces of
the arches and making the structure stiffer.
It is useful to interpret the behaviour of the underground ties further. The
lower part of the arches and the tendon, circled in Figure 5.13b, can be simpli-
fied as a rigid frame with a tendon linking the two ends of the frame, which is
144  Structural Design Against Deflection

(a)

(b)
Figure 5.13 R aleigh Arena. (a) The internal force paths (Reproduced from [5.7]).
(b) The finite element model [5.4].

named Model D as shown in Figure 5.14a. Due to the symmetric nature of the

structure, only a vertical force is applied at the top of the model. This vertical
load is transmitted through the two inclined members in bending, compression
and shear to the tendon and the foundations. Model D can be represented by
Model E based on the principle of symmetry (Figure 5.14b). Effectively, the
tendons in Models E and C have the same functions, constraining the lateral
 Uniform Distribution of Internal Forces 145


   

(a) (b) (c)
Figure 5.14 Illustration models. (a) Model D: A simplified model to represent the
lower part of the two inclined arches for studying the effect of the
tendon. (b) Model E: The equivalent model to Model D. (c) Model C:
An alternative representation of Model E for studying the effect of the
tendon.

relative deflection between nodes A and C and the inward vertical deflection
between B and D and reducing the bending moments in the straight and curved
members. The tendon action in the structure (Figure 5.13b) can be explained
and demonstrated as shown in Figures 5.11 and 5.12.

5.3.3 Structures Derived from Topology Optimisation

5.3.3.1  Evolutionary Structural Optimisation (ESO)

Evolutionary Structural Optimisation (ESO) is a popular and relatively simple
topology optimisation method that can be integrated with many commercial
finite element analysis software packages. This circumvents the difficulties in
solving complicated structural analysis problems. The basic ESO method has
been improved and extended to Bi-directional Evolutionary Structural Optimi-
sation (BESO) which allows adding material to the most demanding parts of
a structure to enhance structural performance and reduce stress intensity, in
parallel with removing material from parts of the structure where it is used to
its least advantage. BESO, now a mature technique, is particularly appealing to
practising structural engineers and architects because it is well suited to build-
ing structures and because a structural concept is embedded into the method.
Adding material to or removing material from a structure in BESO makes
the highest stress smaller and the lowest stress larger, which serves to create
a more uniform distribution of stress or internal forces in a structure. After
removing the material with the lowest stresses from a structure, the lowest
146  Structural Design Against Deflection

stress will become larger while after adding material where stresses are highest,
the highest stress will become smaller. In other words, the difference between
the maximum and minimum stresses will become smaller than in the origi-
nal structure. Repeating this process, differences become even smaller and the
stress distribution in members becomes more uniform. In seeking more uni-
form stress-distribution, BESO can create new structural topology. The struc-
tural concept embedded in the method can be expressed as the more uniform
the distribution of stress, the more efficient the structure, in which efficiency
is measured by the uniformity of the stress distribution in the structure. This
structural concept is similar to the one studied earlier in this chapter, the more
uniform the distribution of internal forces, the smaller the deflection. The
BESO process, which is an automatic computational process, applies to local
areas of a structure following the solution of equilibrium equations and gradu-
ally evolves the original structure into a new structure with a superior geo-
metrical form which is usually structurally efficient and aesthetically pleasing.
The topology optimisation problem in BESO is presented as follows [5.1, 5.2]:

Find X, so that:
1 1
Minimise C = PTU   Pu
i i  (5.1)
2 2

Subiected to KU = P  (5.2)

X T V = V*  (5.3)

where X is the design variable vector in which xi /(the ith element in the X vec-
tor) takes either 0 for the relevant element being absent or 1 for it being pre-
sent; P and U are the external load vector and the nodal displacement vector
respectively, and C is the objective function and is called the mean compliance
that indicates the averaged structural flexibility. In other words, C is the inverse
of the overall stiffness of a structure. C is the same as W11, the work done by
the external loads P on the corresponding deflections U in equation 2.9. Equa-
tion 5.2 is the equation of equilibrium. Equation 5.3 is the constraint condition
that the prescribed volume limit of the whole structure, V*, equals the sum of
the element volumes in which vi is the element volume.
There are similarities and differences between the BESO method and the
method using the structural concepts proposed in this book. Table 5.2 summa-
rises the main features of the two methods for achieving more efficient structures.
Further comments on the comparison of the two methods for design are as
follows:

1. The objective
• The minimum averaged structural flexibility for a given body mass is
searched for BESO while the smaller deflection ∆2,C (equation 2.16), or
 Uniform Distribution of Internal Forces 147

Table 5.2 Comparison between BESO Method and SCM

BESO method [5.1, 5.2] Structural Concepts Method

(SCM)

Objective Minimising the objective Making the deflection at

function, i.e. the averaged the critical point of a
structural flexibility. structure smaller.
Design variables Elements being present or Internal forces.
absent.
Constraints Equation of equilibrium and Not explicitly presented.
the prescribed volume of
the structure.
Intermediate outcome N/A The four structural
concepts.
Solution process Computer process by Using the structural
gradually removing less concepts to improve the
effective material and flow or distribution of
adding material to the internal forces.
most needed region.
Final outcome A new geometry of the A structure with a smaller
structure with the deflection in comparison
minimum averaged with similar structures.
structural deflection for
a given body volume.

the smaller maximum flexibility coefficient in the flexibility matrix of

a structure, is pursued qualitatively for SCM.
• The actual loads acting on the structure are used with BESO while
for SCM the actual loads are lumped and normalised at the critical
point of the structure.
• The external work PT U / 2 is explicitly calculated in BESO while inter-
nal energy is qualitatively interpreted in SCM.
2. Design variables: The design variables are the elements that can be either
present or absent in BESO while the design variables are the internal forces
in the elements with SCM.
3. Constraints: The constraints control how much material is to be removed
from the original structure in BESO while there is no similar explicit con-
straint in SCM.
4. Intermediate outcome: The intermediate outcome is not needed as the
optimum topology is the final outcome in BESO while for SCM, four
structural concepts have been identified from the objective function (equa-
tion 2.16), which need to be implemented through developing particular
physical measures.
4. Solution process: The BESO method is implemented with a finite element
analysis package and processed using a computer. Therefore, the user
should be familiar with the method and the package. Although no com-
puter is necessarily required for SCM, an experienced structural engineer
148  Structural Design Against Deflection

or architect is needed to design the flow or distribution of the internal

forces using any of the four structural concepts.
5. Final outcome: The BESO method creates a new structural geometry that
may be very different from the original geometry, perhaps even a topology
beyond one that could be imagined. The SCM route is likely to achieve a
rational design that has relatively small deflections in comparison to other
similar structures.

Three illustration examples of bridge design carried out using BESO [5.8] are
now considered, in comparison with similar practical examples.

5.3.3.2  A Bridge with a Flat Deck on the Top

Figure 5.15a shows a uniform block that consists of a non-designable layer on
the top surface and a design domain that is to be designed as a bridge. A uni-
formly distributed load of 100N/m2 is applied on the top surface of the block
and the four bottom corners are pin supported. Steel is used with the elastic
modulus E = 210GPa and Poisson’s Ratio v = 0.3. In order to create a clearance
under the bridge, an artificial constraint is added into the design domain by
introducing a void strip under the deck in the middle face.
The constraint of removing 80% of the materials in the design domain
was applied when BESO was carried out and the optimal topology solution

(a)

(b)
Figure 5.15 Comparison between designs from BESO and from structural concepts.
(a) The design domain of a top loaded bridge with pinned supports at
the bottom corners. (b) The BESO design (Courtesy of Professor Yi Min
Xie, RMIT University, Australia).
 Uniform Distribution of Internal Forces 149

Figure 5.16 A  cast iron arch bridge in Manchester.

produced is shown in Figure 5.15b, which is an arch bridge. As the given

boundary conditions can take horizontal forces, the arch design would be an
expected outcome. Considering possible designs for the bridge conceptually and
intuitively, an arch design would be a likely reasonable solution. Figure 5.16
shows such an arch bridge made by cast iron in Manchester, UK, which is
globally similar to the BESO design in Figure 5.16b. The differences between
the two designs are the locations and orientations of the columns between the
arch and the deck. In a structural concepts design, vertical members with equal
spaces are likely to be used, as is usually the case in practice. The inclined deck
support members generated in the BESO design are less likely choices, but it
shows a more ideal design that leads to a more uniform distribution of stress.

5.3.3.3  A Bridge with a Flat Deck at the Middle Level

Figure 5.17a shows a H shaped uniform block with the central horizontal layer
being designated a non-designable layer and four corner pin supports. A uni-
formly distributed load is applied on the horizontal layer. The two vertical
elements of the H section are the design domain and materials can be removed
from this domain. The constraint is set to remove 90% of the material from the
design domain in the BESO process.
Figures 5.17b shows a 3D print-out of the optimum solution produced by the
BESO process, which is an arch bridge with tension members supporting the
central section of the deck and compression members supporting the two ends
of the deck. This optimised topology reflects well good engineering practice
for an efficient structure. One such example is the Tyne Bridge over the River
Tyne, linking Newcastle upon Tyne and Gateshead in the UK, which was com-
pleted in 1928, as shown in Figure 5.18. The vertical cables (tension members)
support the central section of the bridge deck and vertical props (compres-
sion members) support the end sections of the bridge deck. Minor differences
between the topologies of this bridge and the BESO design (Figure 5.17b) are
150  Structural Design Against Deflection

(a)

(b)
Figure 5.17 Topology optimisation of a bridge. (a) Design and non-design domains.
(b) 3D print-out of the BESO outcome (Courtesy of Professor Yi Min Xie,
RMIT University, Australia).

the orientations and cables and the props, which are inclined in the BESO
design but are vertical in the actual bridge. It would be interesting to compare
the structural performances of the two arrangements for tension members!

5.3.3.4  A Long-Span Footbridge with an Overall Depth Limit

This design requires an arch shape bridge with a 72m clear span between two
piers and a maximum arch depth of 1.8m. The BESO process was used to
 Uniform Distribution of Internal Forces 151

Figure 5.18 T he Tyne Bridge in Newcastle upon Tyne, UK.

(a)

(b)
Figure 5.19 T he optimised geometries of the bridges. (a) Bridge with pin and roller
supports. (b) Bridge with pin supports and side roller supports.

create two structurally efficient and aesthetically pleasing forms for the foot-
bridge. 3D brick elements were employed in a finite element model of the
design domain of the bridge and a single material of steel was used. A uni-
formly distributed load is applied on the top of the structure. Two different
boundary conditions were considered: 1) a roller support at the bottom of one
pier and a pin support at the other pier; 2) pin supports at both lower ends of
the bridge as well as horizontal pin supports at the two ends of the bridge.
The two solutions for the different boundary conditions are shown in Fig-
ure 5.19. Since the roller support at one pier allows for horizontal movement,
the bridge in Figure 5.19a acts like a simply supported beam for which the
behaviour is well known. The bending moments are the largest at the centre
of the span where the shear forces are the smallest. The BESO solution shows
152  Structural Design Against Deflection

that materials are only placed at the top and bottom in the central part of the
bridge to resist bending moments and material at the top and bottom gradu-
ally reduces away from the centre of the bridge, reflecting the variation of the
bending moment along the bridge. The inclined members gradually become
thicker from the centre to the two ends of the bridge reflecting the variation of
shear force.
As much stronger boundary conditions were applied to the second design,
the BESO process leads to the more efficient design shown in Figure 5.19b. It is
perhaps unlikely that the BESO solution would be anticipated but the rational-
ity of the design can be explained. Referring to Figure 5.19(b), from the distri-
bution of members, the region between A and B appears to be dominated by
bending with only small bending moments occurring around positions A and
B. The inclined members CE and DF provide vertical support at positions C
and D effectively reducing the span of the bridge which of course leads to
smaller internal forces and deflections.
A similar example from practice is the Kirchheim Overpass, a road bridge
built in 1993 in Germany, which is shown in Figure 5.20a. The rigid frame
bridge has a pair of inclined legs that provide support to the bridge deck and
effectively shorten the bridge span. The inclined legs experience mainly com-
pressive forces rather than bending effects as the deformations of the legs in
both horizontal and vertical directions are confined by the deck and the sym-
metry of the two inclined legs, which is shown in the bending moment diagram
due to a uniformly distributed load in Figure 5.20b. This will also be demon-
strated by a hand calculation example in Section 6.2.2.
The rationality of the BESO design can be appreciated through a compari-
son of the form of the structure and the shape of the bending moment diagram

(a)

(b)
Figure 5.20 T he structure of the Kirchheim Overpass. (a) Structural form. (b) Bend-
ing moment diagram (Reproduced from [5.9]).
 Uniform Distribution of Internal Forces 153

for the Kirchheim Overpass bridge. Relatively small bending moments occur
at Points A and B in Figure 5.19b, which correspond to the smallest sections of
the bridge. Points C and D in Figures 5.19b and 5.20b are where the negative
bending moments are the largest and also where the largest bridge sections are
(Figure 5.19a).

5.4 Further Comments
The concepts of smaller internal forces and more uniform distribution of inter-
nal forces can lead to some similar physical measures for implementation.
However, more uniform distribution of internal forces does not necessarily
mean smaller internal forces. In BESO, a structure is evolved mainly through
removing ineffective materials from the optimisation body. Consequently,
the optimised structure would have a more uniform distribution of stress but
higher values of stress due to the use of less material.
It is fascinating that a similar structural concept, the more uniform the dis-
tribution of stress, the more efficient the structure, has been implemented for
a computer realisation in BESO. In other words, the structures generated from
BESO are likely effective, efficient and possibly elegant. The three BESO bridge
examples demonstrate that the BESO process is able to produce good engi-
neering designs, and the optimum topology designs can provide an excellent
starting point for practical design. It is of interest that the three comparative
practical designs can be evolved from the BESO outcomes with a consideration
of practical aspects.

References
5.1 Huang, X. and Xie, Y. M. A Further Review of ESO Type Methods for Topol-
ogy Optimisation, Structural and Multidisciplinary Optimisation, 41, 671–683,
2010.
5.2 Xie, Y. M. and Steven, G. P. A Simple Evolutionary Procedure for Structural
Optimisation, Computer and Structures, 49, 885–896, 1993.
5.3 Hibbeler, R. C. Mechanics of Materials, Sixth Edition, Prentice-Hall Inc., 2005.
5.4 Yu, X. Improving the Efficiency of Structures Using Mechanics Concepts, PhD
Thesis, The University of Manchester, 2012.
5.5 Olierook, G. Construction of Samuel Beckett Bridge Dublin—Ireland, Hollandia,
2009.
5.6 Ji, T., Bell, A. J. and Ellis, B. R. Understanding and Using Structural Concepts,
CRC Press, Taylor & Francis Group, London, 2016.
5.7 Bobrowski, J. Design Philosophy for Long Spans in Buildings and Bridges, Struc-
tural Engineer, 64A(1), 5–12, 1986.
5.8 Xie, Y. M., Zuo, Z. H., Huang, X., Black, T. and Felicetti, P. Application of
Topology Optimisation Technology to Bridge Design, Structural Engineering
International, 185–191, 2014.
5.9 Holgate, A. The Art of Structural Engineering: The Work of Jorg Schlaich and
His Team, Edition Axel Menges, 1996.
Chapter 6

Converting More Bending

Moments Into Axial Forces

6.1 Routes to Implementation
1. Using bar/string members to create vertical internal elastic supports

This physical measure follows the route of providing internal elastic supports
mentioned in Section 4.1 but makes the route more specific in the vertical
direction. It is understood that shortening a span is the most effective way to
reduce deflections, but it may not always be feasible due to conflicting struc-
tural, architectural or functional requirements. In such cases. Providing verti-
cal internal elastic supports while meeting the other requirements becomes an
attractive solution to dealing with deflection.
Beam-string structures, with a variety of forms, have been used as effi-
cient types of structures. The simplest beam-string structure is illustrated in
Figure 6.1a.
The basic beam-string structure is a simply supported beam AB with a verti-
cal internal elastic support that is provided by a strut, CD, placed under beam
and linked to a profiled string (or tendon), ADB. When the beam deforms
downwards due to the action of the load, the strut CD moves down inducing
tension forces in string, ADB, which effectively provides an upward force to
the beam through the strut, CD. The string and strut act like a spring sup-
port to the beam as shown in Figure 6.1b, which converts part of the bending
moment in the beam into the axial forces in the strut and string. The effect of
the string and strut in the beam-string structure will be examined using a hand
calculation example from Section 2.2.1.

2. Using inclined members to replace vertical members.

Columns, as vertical members, have been widely used as supporting elements

in frame structures and can be seen in almost every building transmitting verti-
cal loads, mainly through compression, and lateral loads, in unbraced frames,
mainly through bending to their foundations.
When a pair of long structural members are inclined to form a triangular
frame, they will carry predominantly axial forces for both vertical loads and
 Converting Bending Into Axial Forces 155

 
(a) (b)
Figure 6.1 Vertical internal spring support. (a) A beam-string structure. (b) An equiv-
alent model for the beam in the beam-string structure.

lateral loads applied to the top of the frame, braced and unbraced. There are a
several ways to achieve the ideal situation, two of them are:

a. If the two inclined members lie in a vertical plane and are connected at
their top ends, they will be able to resist vertical and lateral loads in the
vertical plane and transmit the loads to the supports mainly through axial
forces.
b. If the pair of inclined members lean an angle to the vertical plane and is
supported by other members for achieving equilibrium, they will be able to
resist vertical loads and the lateral loads in the plane and out of the plane
mainly through axial forces

3. Orientation of members.

If a cantilever column is orientated away from the vertical, the self-weight of

the column will be transmitted through bending to its support. It may be pos-
sible to use its self-weight to balance part of the action of the external loads.

6.2 Hand Calculation Examples

6.2.1 A  Beam with and without a Vertical

Internal Elastic Support
This example shows how the provision of a vertical internal elastic support
in a simply supported beam will convert much of the bending moment in the
beam into axial forces in the elastic support system and lead to significantly
reduced deflections and bending moments for the beam.
Figure 6.1 shows two simply supported beams that have the same span of
L and same rigidity of Eb I , subjected to the same uniformly distributed loads
of q. Beam 2 is additionally supported at its centre by a vertical strut and two
inclined strings. The strings have an elastic modulus of Es and an area of A.
To simplify the analysis while still capturing the physical essence of Beam 2,
the axial deformation of the strut CD will not be considered. Beam 2 is called
156  Structural Design Against Deflection

a beam-string structure in literature [6.1]. Calculate and compare the bending

moments and deflections at the centre of the two beams.
Beam 1 (Figure 6.2a): The maximum bending moment and the maximum
deflection occur at the centre of the beam and are respectively:

qL2
M 1C = (6.1)
8

5qL4
1C  (6.2)
384Eb I

Beam 2 (Figure 6.2b): This is a statically indeterminate structure and the

upward force from the strut needs to be determined before calculating the bend-
ing moment and deflection at the centre of the beam. The action of the strut CD
on the beam can be replaced by a force, FCD , that is to be determined. Figure 6.3
shows the geometrical relationship of string BD before and after deflection. The
vertical deflection ∆ 2C at the centre of the beam and the elongation δ of string
BD has the following relationship:

   2C sin  (6.3)

The internal force in the string BD is:

Es A Es A
FBD    2C sin  (6.4)
LBD LBD

(a)

(b)
Figure 6.2 Two simply supported beams. (a) Beam 1: A simply supported beam.
(b) Beam 2: A simply supported beam stiffened by a strut and two strings.
 Converting Bending Into Axial Forces 157

Figure 6.3 Geometrical relationship before (solid line) and after deflection (dashed
line) of string BD.

The projection of the forces in the two symmetric strings to the vertical direc-
tion is equal to the force in the strut CD:

Es A 2Es A
FCD  2FBD sin   2 sin    2C sin2  (6.5)
LBD LBD

The deflection at the centre of the beam ∆ 2C results from the actions of the
downward uniformly distributed load, q, and the upward strut force, FCD, gen-
erated from the two strings, i.e.

5qL4 F L3
 2C   CD (6.6)
384Eb I 48Eb I

Substituting FCD in equation 6.5 into equation 6.6 gives:

5qL4 L3 2Es As
 2C    2C sin2  (6.7)
384Eb I 48Eb I LBD

Rearranging equation 6.7 and noting that L / 2  LBD cos  , gives the maxi-
mum deflection of Beam 2 as:

5qL4 1
 2C   1C  (6.8)
384Eb I E AL sin2  cos 
2
1 s
12Eb I

where:
1 1 1
     (6.9a)
E AL sin  cos 
22
2(Es A / LBD )sin 
2
2K s sin2 
1 s 1 1 
12Eb I (48Eb I / L3 ) Kb
158  Structural Design Against Deflection

48Eb I EA
and: K b = ; K s = s (6.9b)
L3 LBD

K b is the flexural stiffness of the simply supported beam, K s is the axial stiff-
ness of string BD, and ∆1C is the maximum deflection of Beam 1 defined in
equation 6.2. Equation 6.8 indicates that the added strings and strut effec-
tively reduce the maximum deflection of the original simply supported beam
by a reduction factor, β . β in equation 6.9a is related to the ratio of the axial
stiffness of the string to the flexural stiffness of the beam and the angle θ
between the string and the beam. The term, 2K s sin2 θ , in equation 6.9a can be
interpreted as the spring stiffness in the vertical direction produced by the two
inclined strings, K sv . The reduction factor in equation 6.9a and the deflection
in equation 6.8 can then be rewritten as:

1 1 Kb
    (6.10a)
2K s sin2  1  K sv K b  K sv
1
Kb Kb

2Es A 2
K sv  2K s sin2   sin  (6.10b)
LBD

Kb
 2C  1C (6.10.b)
K b  K sv

When K sv = 0, i.e. no strut and strings, Beam 2 reduces to Beam 1. The deflec-
tion ∆ 2C depends on the ratio of the flexural stiffness of the beam to the vertical
stiffness of the strut and strings. For example, if K sv = K b, then   1 / 2 and
 2C  1C / 2 .
After introducing the vertical spring stiffness, K sv  2K s sin2 , of the strut
and strings, Beam 2 in Figure 6.2b can be represented as a simply supported
beam with a spring support at its centre as shown in in Figure 6.4. The strut
and two strings effectively provide an internal support to the beam which
can be converted to an external spring support to the beam to investigate the
response of the beam.

Figure 6.4 A lternative presentation of Beam 2 in Figure 6.2b.

 Converting Bending Into Axial Forces 159

FCD , the spring force in equation 6.5, can be written as a function of ∆1C

2Es A 2 K sv K b
FCD  sin    2C  K sv  2C  1C (6.11)
LBD K b  K sv

The bending moment at the centre of the beam-string structure is the summa-
tion of that induced by the uniformly distributed load and that induced by the
concentrated force from the spring:

qL2 FCD L qL2 5qL4 K sv Kb L

M 2C    ( )
8 4 8 384Eb I K b  K sv 4
(6.12)
qL2 5L3 K sv K b qL2 5 K sv
 (1  ) (1  )
8 4  48Eb I K b  K sv 8 4 K b  K sv

When K sv = 0, the beam-string structure reduces to the simply supported

beam and M 2C = M 1C . When K sv  , the spring in Figure 6.3 becomes a
roller support and the beam-string structure becomes a two-span beam and
M2C  (qL2 / 8)(1  5 / 4)  qL2 / 32 , which is just the bending moment at the
fixed end of a propped cantilever with a span of L/2 subjected to a uniformly
distributed load.
In the beam-string structure, the beam will also resist an axial compres-
sion force to balance the horizontal components of the string forces at the
string-beam connections, which remove the need to create external supports
to balance the string forces. Considering an arch as shown in Figure 6.5a,
pin supports are required at the two ends of the arch to balance the horizon-
tal forces generated by the arch, which tend to push outwards. If an arch-
string structure, similar to the beam-string structure, is considered as shown

(a)

(b)
Figure 6.5 Two arches. (a) An arch structure. (b) An arch-string structure.
160  Structural Design Against Deflection

in Figure 6.5b, the horizontal components of the tension forces in the strings

and compression forces in the arch will be balanced at their connection points.
Thus, a roller support can be used at one of the two ends of the arch.
Figure 6.6 shows an application of adding profiled strings, in the forms of
steel tendons, to an existing beam to increase the fundamental natural fre-
quency of a floor system. The floor in a factory on which machines were
operated on a daily basis experienced severe vibrations, causing significant dis-
comfort for workers. It was found that resonance occurred when the machines
operated. The solution was to avoid the resonance by increasing the stiffness of
the floor and thus its fundamental natural frequency. It has been mentioned in
Chapter 1 that reducing the maximum deflection of a structure means increas-
ing its stiffness and hence the fundamental natural frequency of the structure.
Placing column supports would have reduced the span of the beam and pro-
duced a stiffer structure but this was not feasible due to the use of the area
under the floor. Similar to the beam-string structure shown in Figure 6.2b, the
externally added tendons provided two vertical elastic supports at the points
where two steel bars (acting as struts) reacted against the concrete beams that
support the floor. This produced a stiffer floor altering the fundamental natu-
ral frequency and solving the resonance problem.
Comparing the beam in Figure 6.1b and the real application shown in Fig-
ure 6.6, it can be noted that in the real application the inclined profiles of
the tendons were created by the height of the beam rather than the height of

Figure 6.6 A  floor beam is stiffened to form a beam-string structure to increase its
fundamental natural frequency (Courtesy of Professor Jida Zhao, China
Academy of Building Research, Beijing).
 Converting Bending Into Axial Forces 161

Figure 6.7 A simply supported beam with two equally spaced vertical elastic supports.

the strut, which creates two vertical elastic supports without taking the space
under the beam. The equivalent model for the real application is shown in
Figure 6.7 in which the tendons are converted to two vertical springs with stiff-
nesses of K sv . Following equation 6.10b:
2
2Es A sin2  2Es A  h 
K sv     (6.13)
2  
(L / 3)  h
2 2
(L / 3)  h  (L / 3)2  h 2
2


Tendons were placed on each side of the beam (two tendons in all), and h is the
height of the beam. The simply supported beam with two spring supports is a
statically indeterminate structure. However, if the two spring forces of F can
be determined, it becomes a statically determinate structure and the available
equations for a simply supported beam can be used. The basic equations for
calculating the deflection of a simply supported beam subjected to a uniformly
distributed load q and to two symmetrically applied concentrated loads of F
are respectively [6.2]:

qx
v q (x )  (L3  2Lx 2  x 3 )(6.14)
24Eb I

Fx
v F (x )  [3Lx  3x 2  (L / 3)2 ] L / 3 ≤ x ≤ 2L / 3 (6.15)
6Eb I

At x = L / 3, the deflections due to the loads q and F are respectively:

q(L / 3) 3 11qL4
v q (L / 3)  [L  2L(L / 3)2  (L / 3)3 ]  (6.16)
24Eb I 972Eb I

F (L / 3) 5FL3
v F (L / 3)  [3L(L / 3)  3(L / 3)2  (L / 3)2 ]  (6.17)
6Eb I 162Eb I
162  Structural Design Against Deflection

The condition for the compatibility of deformations at x = L / 3 is:

v q (L / 3)  v F (L / 3)  F / K sv (6.18)

This states that the difference between the deflections induced by q and F at
x = L / 3 is equal to the deflection of the elastic spring. Therefore, substituting
equations 6.16 and 6.17 into equation 6.18 gives:

11qL4 5FL3 F
  (6.19)
972Eb I 162Eb I K sv

The elastic spring force F can be determined from equation 6.19 as follows:

11qL4 K ba K sv 11qL K sv
F   (6.20a)
972Eb I K ba  K sv 30 K ba  K sv

162EI
K ba = (6.20b)
5L3

where K ba is the point flexural stiffness of the beam or the inverse of the deflec-
tion at x = L /3 when the beam is subjected to two unit symmetric vertical
forces at x = L /3 and x = 2L / 3. When Ksv  , it becomes a three, equal-span
beam and each of the two middle roller supports will take 11/30 of the total
loads. The spring force F depends on the ratio of the point flexural stiffness of
the beam, Kba , to the spring stiffness Ksv . The deflections at x = L /3 and at the
centre of beam (x = L /2) are:

F 11qL 1
 L /3   (6.21)
K sv 30 K ba  K sv

5qL4 23L3 11qL K sv

L/ 2  
384Eb I 432Eb I 30 K ba  K sv
(6.22)
5qL4 1012 K sv
= (1  )
384Eb I 675 K ba  K sv

To appreciate the effect of the internal vertical elastic supports, consider that
the structure has the following estimated properties based on Figure 6.6: the
span of the beam is L = 6 m and the cross-section of the beam is b= h= 0.5 m,
leading to a second moment of area of I  0.5  0.53 / 12 = 5.208  103 m 4; the
elastic modulus for the concrete beam and the elastic modulus for the tendons
are respectively Eb  30  109 N / m2 and Es  210  109 N / m2; the tendons
have a diameter 20mm resulting in areas of A  314  106 m2 . The dead load
 Converting Bending Into Axial Forces 163

including self-weight of the beam is q = 100, 000N / m. Using equations 6.20b,

6.13, 6.20a, 6.14, 6.15, 6.21 and 6.22 derived previously:

K ba  2.344  107 N / m; K sv  3.763  106 N / m; F = 30436N ;

v q (L / 2) = 10.8mm ; v F (L / 2)  2.240mm ;

 L / 3  8.09mm ;  L / 2  v q (L / 2)  v F (L / 2)  10.8  2.24  8.56mm

The fundamental natural frequencies before and after using the profiled ten-
dons can be estimated using equation 1.9 as:

1 1
= =
fbf 17 .75 17.75 = 5.40Hz
v q (L / 2) 10.8

1 1
faf  17.75  17.75  6.07 Hz
(L / 2) 8.56

The ratio of the two natural frequencies is:

faf 6.07
= = 1.12
fbf 5.40

It can be noted that the use of the tendons increases the fundamental natural
frequency by 12%, which was sufficient to solve the resonance problem [6.3].

6.2.2 Rigid Plates Supported by Vertical

and Inclined Members
This example demonstrates the effectiveness and efficiency of inclined mem-
bers compared to vertical members in resisting lateral deformation by con-
verting bending moments to axial forces.
Figure 6.8 shows models of three plane structures in which a rigid plate
is supported by four uniform members. The three structures have the same
height, h, and all members have the same elastic modulus, E, and circular tubu-
lar section with a radius R and tube thickness t. They are subjected to the same
lateral force P at the plate level. Model 1 is a typical frame structure, in which
a rigid floor is supported by four vertical members that are rigidly connected
to the plate and fixed to the ground. In Model 2, the rigid plate is supported by
four members with pinned connections inclined at an angle of θ to the vertical.
These members only experience tension or compression forces. Model 3 has
the same geometry as that of Model 2, but with rigid connections. The lateral
deflections of the three models will be compared.
164  Structural Design Against Deflection


(a)  Model 1 (b)  Model 2

(c)  Model 3
Figure 6.8 A  rigid plate supported by four members. (a) Model 1: A rigid plate sup-
ported by four vertical members. (b) Model 2: A rigid plate supported by
four inclined members with pin connections. (c) Model 3: A rigid plate
supported by four inclined members with rigid connections.

The relative stiffness of the three models can be assessed qualitatively using
the structural concept, the more bending moments that are converted to axial
forces, the smaller the deflection. Model 1 (Figure 6.8a) will experience the
largest lateral deflection of the three Models as the load P is transmitted to the
ground through bending and shear in the four vertical members. Due to
the pinned connections, Model 2 (Figure 6.8b) transmits the load P to the
ground by the inclined members through tension and compression alone,
which is far more effective than through bending and, as expected, Model 2
experiences smaller lateral deflection than Model 1. Model 3 transmits the load
P through both axial forces and bending moments. The differences between
Models 2 and 3 are the connections at the ends of the members. As Model 3
(Figure 6.8c) has stronger connections than Model 2, it is expected that Model
3 would experience less deflection than Model 2. Following this qualitative
assessment, detailed analysis can be conducted to quantify the abilities of the
three models to resist lateral deflection.
 Converting Bending Into Axial Forces 165

Model 1: A Rigid Plate Supported by Four Vertical Members

(Figure 6.8a)

As the two ends of the vertical members are rigidly connected with the plate
and the ground, the lateral stiffness of each of the four members is 12EI / h3 .
Therefore, the lateral displacement of Model 1 due to load P is:

Ph3
1  (6.23)
48EI

Model 2: A Rigid Plate Supported by Four Inclined Members with

Pin Connections (Figure 6.8b)

A rigid plate is supported by two identical pairs of inverted V shaped members.

Each pair of inclined members carry a half of the lateral load P, hence only the
two members of one of the inverted V shaped frames need be analysed. The
applied load and the internal forces in one inverted V shaped arrangement of
members are illustrated in Figure 6.9 in which the directions of the internal
forces are shown.
The internal forces in the two members, N A and N B , can be determined
from equilibrium conditions:

N A sin   N B sin   P / 2 and N A cos   N B cos   0(6.24)

Solving the equations leads to:

P
NA  NB  (6.25)
4sin 

Figure 6.9 External and internal forces acting on the top node of the frame.
166  Structural Design Against Deflection

When P/2 is replaced by a unit force, the corresponding forces in the two
members are:

1
NA  NB  (6.26)
2sin 

The lateral deflection of Model 2 can be calculated using Equation 2.14 as

follows:

NiNiL 1 P 1 h Ph 1
2      2  (6.27)
EA EA 4 sin  2 sin  cos  4EA sin2  cos 

Model 3: A Rigid Plate Supported by Four Inclined Members with

Rigid Connections (Figure 6.8c)

Model 3 looks like Model 2, except the connections to the plate and the ground
are rigid. The equilibrium equation in the lateral direction for Model 3 can be
found from the element stiffness matrix of an inclined beam member in finite
element analysis [6.4] as follows, when no rotation and axial deformation are
considered at the top node of the members:

EA 2 12EI
4( sin   3 cos2  ) 3  P (6.28)
L L

where L is the length of the inclined member and can be expressed as h /cos θ ;
the terms in the bracket represent the lateral stiffness of a single inclined uni-
form beam. Solving this equation gives:

P P
3   (6.29)
EA 2 12EI EA 12EI
4( sin   3 cos  ) 4(
2
sin  cos   3 cos  )
2 5

L L h h

When the rigid connections reduce to pinned connections, i.e. the members
are not be able to transmit bending, I = 0, and equation 6.29 reduces to equa-
tion 6.27 for Model 2. When θ = 0o, the inclined members become vertical
members and equation 6.29 reduces to equation 6.23 for Model 1. Comparing
equations 6.27 and 6.29, it can be seen that when 0    900 ,  3   2 , i.e.:

P Ph 1
3  
EA 2 12EI 3I
sin  cos   3 cos5  ) 4EA sin  cos  (1 
2
4( n2 cos2  )
ctan
h h Ah 2
(6.30) 1
 2  2
3I
(1  ctan  cos  )
2 2

Ah 2
 Converting Bending Into Axial Forces 167

The lateral deflections of the three Models are shown in equations 6.23, 6.27
and 6.29. The ratios of the deflections of the three Models can be obtained by
substituting I   R 3t and A  2 Rt for a circular tube section when R >> t, as:

2 Ph 1 48EI 12I 1 6R 2 1
    (6.31)
1 4EA sin  cos  Ph
2 3
Ah sin  cos 
2 2
h sin  cos 
2 2

 3 6R 2 1 1
 2 (6.32)
1 h sin  cos 
2
3R 2
(1  ctan2 cos2  )
2h 2
3 1
 (6.33)
2 3R 2
(1  ctan2 cos2  )
2h 2

Consider the cases of R = 100mm, h = 4000mm and 8000mm and   5 0 , 100 , 15 0 , 30

  5 , 100 , 15 0 , 300 and 450 , and the deflection ratios based on equations 6.31–6.33
0

are listed in Table 6.1.

It can be noted from Table 6.1 for this particular case that:

1. Converting bending moments to axial forces makes the structures much

stiffer and significantly reduces the deflections due to the lateral load-
ing. For the case of h = 4000mm and   150 , the reduction is approxi-
mately 94%.
2. Even a small inclination angle ( 50 from the vertical) can still reduce the
lateral deflection by over 50% for h = 4000mm and by over 87% for
h = 8000mm .
3. When the inclination angle is larger than or equal to 100 , the effect of the
rigid connections of the inclined members is negligible for reducing the
lateral deflection.
4. When the structure becomes higher, the inclined members become even
more effective and efficient to resist lateral deflection than normal vertical

Table 6.1 T he Deflection Ratios for Different Angles of Member Inclination and for
Two Different Heights of Structure

h = 4000 mm   50  10 0   150   30 0   450

∆ 2 / ∆1 0.4956 0.1263 0.05796 0.01732 0.01061
∆ 3 / ∆1 0.4419 0.1226 0.05726 0.01728 0.01060
∆3 / ∆2 0.8916 0.9716 0.9880 0.9979 0.9995

h = 8000 mm   50  10 0  150   30 0   450

∆ 2 / ∆1 0.1239 0.0315 0.01448 0.004330 0.002652
∆ 3 / ∆1 0.1202 0.0313 0.01444 0.004328 0.002651
∆3 / ∆2 0.9705 0.9927 0.9970 0.9994 0.9999
168  Structural Design Against Deflection

columns. For h = 8000mm and   150 , the deflection ratio of Model 2

to Model 1 is only 0.014 and there is little difference between rigid and
pinned connections for Models 2 and 3.

This quantitative analysis of the three models provides a theoretical basis

to explain that inclined members can be effectively used, replacing vertical
columns, to support upper structures against lateral deflection. Therefore, it
can be concluded that it is very effective and efficient, where possible, to use
inclined bar members to replace convention vertical columns in which the
fourth concept is embedded.

6.3 Practical Examples

6.3.1  Structures with Vertical Internal Elastic Supports

6.3.1.1  Spinningfields Footbridge, Manchester

The Spinningfields Footbridge over the River Irwell, linking Spinningfields in
Manchester and New Bailey in Salford, UK, has a single span of 44 meters and
was built in 2012. It can be seen from Figure 6.10a that it has the appearance
of a light and elegant steel footbridge. The footbridge consists of a bridge deck,
a group of beams, a series of struts and a cable or a tendon. The struts provide
links between the cable and the beams that support the deck (Figure 6.10b),
and the different heights of the struts create the profile of the bridge.
The bridge experiences bending with the cable carrying a tension force
while the beam with a circular section beneath the deck carries a compression
force to balance the bending moment induced by vertical loads. The distance
between the cable and the circular section beam is largest at the centre of the
bridge and gradually reduces toward the bridge supports, which reflects the
profile of the bending moment diagram for a simply supported beam subjected
to uniformly distributed loads.
The circular section beam, which directly supports the deck, is sup-
ported by a series of struts that are equivalent to vertical internal elastic
supports, which effectively reduce the bending moments and deflections of
the beam.
To gain a better understanding of the behaviour of the footbridge, consider
two simplified cases which still capture the physical essence of the footbridge,
as shown in Figure 6.11. Model 1 is a normal, simply supported beam sub-
jected to uniformly distributed load, and Model 2 is a beam-string structure.
The basic data used for analysis are the span of L = 40m, the elastic modu-
lus of E  200  109 N / m2 and the distributed load of q = 10kN / m . The cable
has a parabolic shape with sag of 2m. Assume the beam has a solid section
of b  h  800mm  400mm and the struts and the cable have solid circular
sections with diameters of 80mm and 40mm respectively. Seven struts are uni-
formly distributed along the length of the beam at an interval of 5m and follow
a parabola with a sag of 2m.
 Converting Bending Into Axial Forces 169

(a)

(b)
Figure 6.10 The Spinningfields Footbridge, Manchester. (a) Overall view. (b) Detailed
view shown the relationship between the cable, struts, beams and the
deck.
170  Structural Design Against Deflection

(a)

(b)

(c)

(d)
Effectiveness and efficiency of a beam-string structure. (a)
Figure 6.11 
Model 1: A simply supported beam. (b) Model 2: A beam-string structure.
(c) Bending moment diagram for Model 1. (d) Bending moment diagram
for Model 2.

Figure 6.11c and 6.11d show the bending moment diagrams of the sim-
ply supported beam and the beam-string system with the maximum values of
2000kNm and 1139kNm respectively. The corresponding maximum deflec-
tions of the two models are 0.391m and 0.193m respectively. These results
indicate that a beam-string system can be designed much lighter than a cor-
responding beam system.
Beam-string structures are often used for roof structures. Figures 6.12a and
6.12b show that the roof structure of the Shanghai Pudong Airport Terminal 1
consists of a series of beam-string beams in which the strings and the struts can
be easily identified. Figure 6.12b also shows several inclined cables anchored
on a column which provide the structural stiffness in the two horizontal direc-
tions of the roof and increase the vertical resistance of the roof to wind uplift.

6.3.1.2 T he Roof of the Badminton Arena for the 2008

Olympic Games, Beijing
The beam-string structures of the footbridge and the terminal roof described in
Section 6.3.1.1 are plane, 2D, structures. However plane beam-string structures
 Converting Bending Into Axial Forces 171

(a)

(b)
Figure 6.12 T he roof at the Shanghai Pudong Airport Terminal 1. (a) A series of
beam-string beams. (b) Inclined cables that increase the lateral stiffness
of the roof and provide vertical resistance to uplift wind loading.
172  Structural Design Against Deflection

have been developed into three dimensional structures forming so called string
supported spherical shells [6.5] or cable supported domes [6.6]. The roof of
the Badminton Arena (Figure 6.13) for the 2008 Olympic Games in Beijing is
a string supported dome that spans 98m.
To understand the structural components and behaviour of the Arena roof,
a similar but simpler example is illustrated in Figure 6.14. The roof consists
of a single-layer shell with struts and cables in both circumferential and radial
directions. The top ends of the struts are connected to the shell and their lower
ends are linked with both radial and circumferential cables. Figure 6.14a shows
that the roof has three layers of circumferential cable rings.
The cross-section of the string supported shell roof (Figure 6.14a) looks like
an arch supported by struts at three different levels. The load paths or internal
force paths of the roof structure are direct and clear. Most of the external loads
applied on the shell are transmitted to the struts and through the struts to the
cables. At the highest level the action of radial and circumferential cables in
space act as a series of plane beam-string structures (Figure 6.12a). The forces
from the two struts are balanced at their connection points to the radial and
circumferential cables and transmitted by radial cables to the struts in the next
lower level. This type of force transmission continues to the lowest level of
struts. The function of the circumferential cables is to position the struts and
the radial cables and allow the struts to provide vertical elastic supports to the
shell. The lowest radial cables apply tensile forces to the supports and tend to

Figure 6.13 T he roof of the Badminton Arena for the 2008 Olympic Games, Beijing.
 Converting Bending Into Axial Forces 173

(a)

(b)

(c)
Figure 6.14 A  string supported shell roof. (a) Cross-section. (b) Plan. c) Elevation
[6.5] (Courtesy of Professor Zhihua Chen, Tianjin University, China).

pull the supports inward while the shell acts in compression tending to push
the supports outward. Therefore, the two sets of forces are partly self-balanced
and produce smaller reactions on the ring beam. As the struts locate on the
radial and circumferential cables, they act as internal vertical elastic supports
to the shell which leads to smaller internal forces in the shell and hence smaller
deflections. The performance of string supported shell roofs can be further
improved by applying pre-stressing to structural members to produce internal
forces in the members to counteract those induced by external loads which
could lead to even more efficient structures.
The analysis of a string supported shell roof needs the use of a computer but
much of the structural behaviour of the shell roof can be illustrated using the
example of the beam-string structures examined in Section 6.2.1.
Back to the Badminton Arena for the 2008 Olympic Games, Figures 6.15a
and 6.15b show the plan and the cross-section of the roof structure. As
shown in Figure 6.15b, the Badminton Arena has five rings of circumferential
174  Structural Design Against Deflection

(a)

(b)

(c)
Figure 6.15 (a) Plan of the Arena roof. (b) Cross-section of the Arena roof. (c) A con-
nection between a circumferential cable, two radial cables and a strut
(Courtesy of Professor Ailin Zhang, Beijing University of Technology, China).
 Converting Bending Into Axial Forces 175

cables at different levels under the single-layer shell, which are linked to the
shell by radial cables and struts. In construction, stressing the circumferential
cables produced tensile forces in the radial cables and compression forces
in the struts supporting the upper single-layer shell. To make the construc-
tion process more convenient, the circumferential cables were stressed at
four tensioning points provided in each cable to reduce the connection fric-
tion between the struts and cables. Figure 6.15c shows a typical connection
between a strut, a circumferential cable and two radial cables, indicating the
internal force paths.
As the struts positioned and supported by radial and circumferential cables
at five different levels, provide many vertical internal supports to the roof
dome, the roof is able to cover a huge area without using any inner supports.

6.3.2 Structures Supported by Inclined Members

6.3.2.1  Three Types of Support to Superstructures

Figure 6.16a shows the external view of an airport terminal. The roof is sup-
ported by slender, inclined members, which have pin connections at their two
ends and thus carry only axial forces. Roofs and floors are normally supported
by vertical columns and horizontal beams, which form frame structures to
transmit both vertical and lateral loads to their supports. For example, Fig-
ure 6.16b shows a building structure in which columns are the main load bear-
ing members with the externally exposed columns supporting the upper storeys
of the building. The two ends of the columns can be considered to have rigid
connections. The cross-section size of the columns and the distance between
adjacent columns leads to a sense of solidity of the building. A combination of
the supporting systems in Figures 16a and 16b (i.e. pin ended inclined mem-
bers and rigidly connected vertical members) leads to inclined members with
rigid connections. Figure 16c shows a building in which the upper structure is
supported by a series of V or inverted V shaped columns in which their bottom
and top ends are close to rigid connections.

(a)
Figure 6.16 C omparison of three supporting systems. (a) The roof of an airport
terminal supported by inclined pin ended members. (b) Vertical column
members support the upper storeys of a building. (c) Inclined members
support the upper storeys of a building.
176  Structural Design Against Deflection

(b)

(c)

Figure 6.16 (Continued)
 Converting Bending Into Axial Forces 177

The three models (Figure 6.8) qualitatively and quantitatively studied in Sec-

tion 6.2.2 are abstracted from the three structures in Figure 6.16, which have
demonstrated that the use of inclined members is very effective and efficient for
resisting lateral loading. The terminal roof (Figure 6.16a) has a height of over
20m, and the results in Table 6.1 have shown that inclined members become
even more effective and efficient as the height increases.

6.3.2.2  Ontario College of Art and Design, Toronto

Figure 6.17 shows the Sharp Centre for Design, an expansion of the Ontario
College of Art and Design (OCAD) in Toronto. It looks like a huge rectangu-
lar block that is in fact a two-storey building 80m long and 30m wide. When
viewed from a distance, the block appears similar to a tabletop floating over
the ground because it is only supported by a small number of slender members
and cantilevered concrete walls, which support stairs between the ground and
the block.
The block is actually supported by twelve 29m long steel legs each with a
diameter of 914mm and a wall thickness of 25mm, which seem to be ran-
domly arranged [6.7]. The 12 legs form six pairs of triangular leg arrange-
ments to achieve better stability and lateral resistance. The steel columns are

Figure 6.17 T he Sharp Centre for Design, Toronto (Courtesy by Mr. Nicolas Janberg,
structurae.net, Germany).
178  Structural Design Against Deflection

tapered at the upper and lower ends indicating that they act as pinned con-
nections to the ground and to the block and act as compression members
rather than bending members. The planes of the two front pairs of legs orient
± 45 degrees from the central axis in the longitudinal (longer) direction and
are perpendicular to each other, which provide the lateral stiffness in the
two horizontal directions, in addition to their vertical supports to the block.
The two middle pairs of triangular legs are only orientated in the transverse
(shorter) direction, providing lateral stiffness in this direction. Due to the
asymmetric position of the concrete stair-core to the block, the two other
pairs of legs are arranged opposite to the concrete core, along the longitudi-
nal direction and leaning inward in the transverse direction to provide lateral
stiffness in both transverse and longitudinal directions. These two pairs of
legs also compensate for the effect of the asymmetric location of the large,
stiff, concrete stair-core that also contributes lateral stiffness in both trans-
verse and longitudinal directions.
At first sight, it is puzzling how twelve slender inclined (in the longitudinal
direction) members can safely support the large structure of the Sharp Centre
for Design. However, the results presented in Table 6.1 effectively explain the
technical feasibility of using slender inclined, pin ended members to replace
more conventional columns.

6.3.2.3  Roof Supports of Terminal 5 at Heathrow Airport, London

Another good example of using long inclined pin ended members to replace ver-
tical columns can be seen in the Heathrow Terminal 5 as shown in Figure 6.18a.
A series of pairs of long inclined members are placed along a window wall to
support the upper structure systems, including long span beams and the roof
of the terminal. Examining one typical unit of six connected inclined members,
Figure 6.18a shows a pair of long steel tubes forming a triangular shape are
placed next to the window wall along the longitudinal (longer) direction and
inclined inward in the transverse direction. The bottom ends of the two tubes
are pinned to their foundations while their top ends are pin connected together
with two upper pairs of inclined steel tube members (Figure 6.18a). The top
ends of the shorter pair of inclined members on the top right (Figure 6.18a)
provide the supports at the ends of two adjacent roof beams and are linked by a
horizontal bar member (Figure 6.18b). The longer pair of inclined tubes acts as
internal props to the two roof beams and provide the end supports to the cables
that are used to stiffen the roof beams. A horizontal bar connects the top ends
of the two inclined props which not only positions the props in the longitudinal
direction but also provides lateral supports to the two roof beams.
The three pairs of inclined members, two horizontal members and two roof
beams form a stable, equilibrium and mutual-supporting system. The long steel
members are used to support the two upper pairs of the inclined members that
in turn support the two roof beams. It is also true that the top ends of the long
steel members are positioned and supported by the lower ends of the two pairs
of inclined upper members, while the top ends of these inclined members are
positioned and supported by the two roof beams and the two horizontal bars.
 Converting Bending Into Axial Forces 179

(a)

(b)
Figure 6.18 Inclined pin ended members supporting the long span roof at Terminal
5 Heathrow Airport. (a) A series of pairs of pin ended inclined steel
tubes with two pairs of upper inclined members supporting the roof
structure. (b) The two upper pairs of inclined members provide four
supports to two adjacent roof beams.
180  Structural Design Against Deflection

Considering the large dimensions of the terminal building (Figure 6.18), the

long inclined support members appear slender and sparsely spaced, which is
due to the effectiveness and efficiency of the inclined members that are sub-
jected to axial forces rather than bending action demonstrated in Section 6.2.2
and due to the self-supporting of the two roof beams, six incline members and
the two horizontal members.

6.3.3 Using Self-Weight of Structural Members—Alamillo

Bridge, Seville
The Alamillo Bridge is a well-known example in which the self-weight of the
pylon is used to balance the self-weight of the bridge deck and part of the live
loads on the bridge. The Alamillo Bridge, shown in Figure 6.19, is a cable-
stayed bridge that was one of the six bridges built to improve infrastructure for
the Expo 1992 on the island of La Cartuja, just outside the city of Seville [6.8,
6.9]. The bridge has a span of 200m and is supported at the two ends and by
13 pairs of cables with a uniform spacing of 12m.
The original idea of the design came from Santiago Calatrava arranging for
the forces in the cables supporting the bridge deck to be balanced by the con-
siderable self-weight of a massive reinforced concrete pylon with a backward
inclination of 58 degrees from the ground, rather than the traditionally used
back-stay cables [6.8]. This idea of the design is illustrated in Figure 6.20.
Normally, a pylon for a cable stayed bridge is a vertical member and cables
are arranged at both sides of the pylon that transfers the compression forces

Figure 6.19 T he Alamillo Bridge, Spain (Courtesy of Mr Per Waahlin, Sweden).

 Converting Bending Into Axial Forces 181

Figure 6.20 Illustration of the forces acting on the Alamillo Bridge (this figure is
produced based on Figure 2 in [6.9]).

and bending moments generated from the cables to its foundation. The lateral
force components generated by the two sets of cables on the pylon are in oppo-
site directions and partly self-balancing, leading to smaller bending moments
in the pylon. The unique design of the inclined pylon (Figure 6.20) has two
effects on internal force paths:

1. Globally, the weight of the pylon would be designed to balance part of the
loads from the bridge deck, self-weight and live loads.
2. Locally, the resultant forces of the self-weight of the pylon and the cable
forces resulting from the bridge deck loads pass through the central axis
of the inclined pylon, leading to that the pylon experiences mainly com-
pressive forces rather than bending moments. In other words, the bending
moments in the pylon induced by the cable forces are partly balanced by
the bending moments generated by the self-weight of the pylon, thus con-
verting bending moments to compressive forces in the pylon.

To achieve this fully would require a pylon of extreme size and mass to
balance the deck forces and create the ideal compression only force in the
pylon. Theoretically, it might be possible to achieve self-balancing and an
ideal state of internal forces for one particular loading case. However, for a
bridge subjected to a wide range of loading scenarios, which need to be con-
sidered in the design of civil engineering structures, some bending moments
would always need to be allowed in the design of the pylon. The idea of an
inclined pylon nevertheless does help to reduce the bending moments in the
pylon.
182  Structural Design Against Deflection

6.4 Further Comments
The Y shaped column subjected to vertical loads and inclined members sub-
jected to lateral loads are examined quantitatively and independently in Sec-
tions 4.2.2 and 6.2.2. The understanding gained from these two sections can
be used to judge the behaviour of existing structures. Large V or Y shaped
supports can be seen at 1 Spinningfields, Manchester, as shown in Figure 6.21.
The cross-section of the inclined members gradually becomes smaller with an
increase in height, which leaves an impression that the inclined members
experience their smallest bending moments at their upper ends and the larg-
est bending moments at their lower ends, i.e. the variable cross-section of
the members seems to suggest that they are subjected to bending. The top
ends of the members support and link to floor beams that in turn restrain
the relative lateral deflection between the two top ends of the members for
vertical loading. The example of the tied Y column in Section 4.2.2 indicates
that the arms are subjected mainly to compression forces rather than bending
moments due to vertical loading. Considering the actions of lateral loads, the
Y columns are similar to that in Figure 6.8c, when the floor supported by the
Y columns is considered as a rigid plate. Table 6.1 shows that at an inclina-
tion angle of the members of 45 degrees to the vertical there would be little
bending moments in the members, and thus, from a structural point of view
alone, the arms of the Y shaped columns could be designed with a constant
cross-section.

Figure 6.21 Variable cross-section of the arms of Y shaped columns in a building, UK.

 Converting Bending Into Axial Forces 183

(a)

(b)

Figure 6.22 T he roof supports of the Xi’an North Railway Station, China. (a) Over-
view of the roof and its supports. (b) A typical Y branch support and
overhangs.

The physical measures mentioned in the earlier chapters can be jointly used
to achieve more efficient structures. Figure 6.22 shows the waiting hall of the
Xi’an North Railway Station, China. The roof of the hall is a large-span light
steel folding plate grid structure that is hidden by the suspended celling, but
the supports to the roof can be seen. Figure 6.22b gives a close look at one of
the supports to the roof structure. Four inclined members, pin connected at the
top of a column, provide four external pin supports to the roof grid structure,
allowing a larger distance between columns.
The physical measures used in the roof structure include overhangs to reduce
the span between supports, the Y shaped columns for providing more supports
to the roof and inclined bar members.

References
6.1 Saitoh, M. and Okada, A. The Role of String in Hybrid String Structure, Engi-
neering Structures, 21, 756–769, 1999.
6.2 Gere, J. M. and Timoshenko, S. P. Mechanics of Materials, PWS-KENT Publish-
ing Company, USA, 1990.
6.3 Ji, T., Bell, A. J. and Ellis, B. R. Understanding and Using Structural Concepts,
Second Edition, Taylor & Francis, USA, 2016.
6.4 Cook, R. D., Malkus, D. S. and Plesha, M. E. Concepts and Applications of Finite
Element Analysis, John Wiley & Sons, USA, 1989.
184  Structural Design Against Deflection

6.5 Chen, Z. Cable Supported Domes, Science Press, Beijing, China, 2010.
6.6 Zhang, A. Olympic Badminton Area: Cable Suspended Dome, The Structural
Engineer, 85(22), 23–24, 2017.
6.7 Silver, P., Mclean, W. and Evans, P. Structural Engineering for Architects: A Hand-
book, Laurence King Publishing Ltd., London, 2013.
6.8 Aparicio, A. C. and Casas, J. R. The Alamillo Cable-Stayed Bridge: Special Issues
Faced in the Analysis and Construction, Structures and Buildings, the Proceedings
of Civil Engineers, 122, 432–450, 1998.
6.9 Guest, J. K., Draper, P. and Billington, D. P. Santiago Calatrava’s Alamillo Bridge
and the Idea of the Structural Engineer as Artist, ASCE, Journal of Bridge Engi-
neering, 18(10), 936–954, 2013.
Chapter 7

Concluding Remarks

The contents of Chapters 2 to 6 can be summarised in the hierarchical rela-

tionships shown in Figure 7.1 in which some of the practical examples in these
chapters are listed.
Figure 7.1 shows paths and connections from theory (the Principle of Virtual
Work) to four structural concepts, then on to several routes to implementa-
tion and finally to a large number of practical application cases indicating that
the four structural concepts have a range of applications for structural design
against deflection. It is noted that this observation is applicable to other struc-
tural concepts, as it is natural that good structural concepts can lead to wide
and wise applications.
Figure 7.1 also illustrates one type of relationship between theory and practice,
i.e. moving from theory downward to practice applications. However, the arrows
in Figure 7.1 can also be reversed and presented in another type of relationship,
i.e. moving from practical cases upward to theory. In other words, implemen-
tation measures can be identified from practical cases and structural concepts
can then be abstracted from the implementation measures. The downward and
upward relationships between theory and practice are complementary to each
other and can enrich both practical applications and theoretical studies. The pres-
entation in this book takes the downward approach from theory to practice, and
the implementation measures are developed based on the structural concepts.
Some implementation measures, however, were created intuitively to solve
problems encountered in practice. For example, to reduce the horizontal
thrusts on the foundations from the two inclined arches of the Rayleigh Arena
discussed in section 5.3.2.2, (Figure 5.13), tendons were provided between
the ends of the two arches to balance part of the thrusts and thus the founda-
tions experienced much smaller horizontal forces from the arches. Further
study of the action of the tendons also led to the implementation routes of
self-balancing of internal forces and the provision of internal elastic supports
and then to the structural concept of smaller internal forces leading to smaller
deflections.
The four structural concepts provide a basis for creative structural design
against deflections for tall buildings, long-span bridge/roofs and other struc-
tures that are sensitive to deflections. Some implementation routes and physi-
cal measures have been explicitly explained and provided through illustrated
Figure 7.1 Hierarchical relationships between theory and practice and between book contents.
 Concluding Remarks 187

hand calculations and practical examples. According to Figure 7.1, there are

four structural concepts, more routes to implementation and even more physi-
cal measures implemented into practical cases, which is not an exclusive list.
In other words, there are opportunities to generate new routes to implementa-
tion, create new implementation measures or use existing ones for achieving
smaller deflections of structures and for designing more effective and efficient
structures that are likely to appear more elegantly.
Bibliography

Allan, E. et al. Form and Forces: Designing Efficient, Expressive Structures, John
Wiley & Sons, USA, 2009.
Balmond, C. Informal, Prestel, Germany, 2002.
Charleson, A. Structure as Architecture—A Source Book for Architects and Structural
Engineers, Architectural Press, UK, 2005.
Frei, O. and Bodo, R. Finding Form: Towards an Architecture of the Minimal, Deutscher
Werkbund Bayern, Edition Axel Menges, Third Edition, 1996.
Heyman, J. Structural Analysis: A Historical Approach, Cambridge University Press,
UK, 1998.
Jennings, A. Structures—From Theory to Practice, Spon Press, London, 2004.
Macdonald, A. J. Structure & Architecture, Second Edition, Architecture Press, Oxford,
2003.
Margolius, I. Architects + Engineers = Structures, Wiley-Academy, UK, 2002.
Parkyn, N. The Seventy Architectural Wonders of Our World, Thames & Hudson,
London, 2002.
Rappaport, N. Support and Resist: Structural Engineers and Design Innovation, The
Monacelli Press, USA, 2007.
Robinson, D. N. Consciousness and Its Implications, The Teaching Company, USA,
2007.
Rosenthal, H. W. Structural Decision, Chapman & Hall Ltd., London, 1962.
Salvadori, M. and Heller, R. Structures in Architecture: The Building of Buildings,
Prentice-Hall, NJ, USA, 1986.
Sandarker, B. N. On Span and Space: Exploring Structures in Architecture, Routledge,
London, 2008.
Sandaker, B. N., Eggen, A. P. and Cruvellier, M. R. The Structural Basis of Architecture,
Second Edition, Routledge, London, 2011.
Schlaich, J. and Bergermann, R. Light Structures, Prestel, Germany, 2004.
Schlaich, M. Elegant Structures, The Structural Engineer, 2015.
Silver, P., Mclean, W. and Wvans, P. Structural Engineering for Architects: A Hand-
book, Laurence King Press, London, 2013.
Sprott, J. C. Physics Demonstrations—A Sourcebook for Teachers Physics, The Univer-
sity of Wisconsin Press, USA, 2006.
Uffelen, C. V. Bridge: Architecture + Design, Braun Publishing AG, Switzerland, 2009.
Young, J. W. A Technique for Producing Ideas, McGraw-Hill, USA, 2003.
Index

Page numbers in italics indicate figures; page numbers in bold indicate tables.

A roof of Badminton Arena (Beijing

airport terminals Olympic Games), 170 – 173, 174, 175
Harbin Airport Lounge root (China), roof of Shanghai Pudong Airport
109, 110 Terminal, 170, 171
Heathrow Terminal 5 supports, 178, stiffening floor beam for increasing
179, 180, 186 natural frequency, 160
Heathrow Y shaped column with beam theory
horizontal bar, 103, 104 designing against deflection, 18 – 20
Madrid Barajas Airport, 116, 118, 119 second moment of area I, 18, 19, 19
Pudong Airport Terminal (Shanghai), bending moments
116, 117 deflection and, 4 – 5, 36 – 38
Pudong Airport Terminal roof, overhangs and tendon force
170, 171 reducing, 96
Alamillo Bridge (Seville, Spain), 180, simply supported beam, 91, 93 – 95
180 – 181, 186 Bi-directional Evolutionary Structural
ANSYS finite element method, 57 Optimisation (BESO), 125, 145 – 148
arches, structure, 159 bridge design examples, 148 – 153
axial and shear forces, deflection due to, comparison to structural concept
36 – 38 method (SCM), 146 – 148, 147, 148
see also bridge design
B Bihai Cultural Centre (Tianjin, China),
Badminton Arena (Beijing Olympic 116, 117
Games) bracing systems
connection between cables, 174 building over Cannon Street
cross-section of roof, 174 Underground Station, 83, 85, 86
plan of roof, 174 calculating deflections of five
roof of, 170 – 173, 175 frames, 55
string supported shell roof, 173 creating more direct internal force
Bank of China, Hong Kong, 67, 68 paths, 51 – 52
beam, see internal elastic supports; effect of four bracing criteria, 52 – 57
simply supported beam five frames with varying bracing
beam-string structures, 154, 155 arrangements and internal force
arch and arch-string structures, 159 paths, 53
flexural stiffness, 158 implementation of, 51 – 52
forces in two symmetric strings, 157 internal forces in halves of five
internal force in string, 156 frames, 54
maximum bending moment, 156 John Hancock Center (Chicago),
maximum deflection, 156, 157 64 – 68
190 Index

in Leadenhall Building (London), Chengdu East Railway Station,

68 – 73 122 – 124, 123
most and least effective bracing columns
patterns for simple frame, 57 – 64 as supporting elements in structures,
scaffolding structures, 82 – 83, 84 154 – 155
summary of results for five frames, 55, see also Y shaped columns
55 – 57 compatibility, structural concept of,
in tall buildings, 64 – 73 41 – 43
temporary grandstands, 73 – 81 Corsica, France, temporary grandstand
see also simple frame collapse in, 75 – 76
branching structures, 110, see also critical load, deflection and, 4
tree-like structures critical point, 29, 32 – 34
bridge design
Alamillo Bridge (Seville, Spain), 180, D
180 – 181 deflection(s), 3
cast iron arch bridge (Manchester, bending, axial and shear forces, 36 – 38
UK), 149, 149 bending moment and, 4 – 5
flat deck at middle level, 149 – 150, calculating for Y shaped columns,
150, 151 98 – 102
flat deck on top, 148 – 149, 149 critical (buckling) load and, 4
Katehaki pedestrian bridge (Athens), designing against, based on beam
139, 140 theory, 18 – 20
Kirchheim Overpass (Germany), 152, equation for calculating for frames, 55
152 – 153 flexibility coefficient and stiffness
long-span footbridge with overall matrix, 32 – 34
depth limit, 150 – 153, 151 form, and internal forces, 5 – 10
Samuel Beckett Bridge (Dublin), 135, lateral, by vertical, horizontal and
136 – 138, 137 diagonal members, 62, 63
Serreria Bridge (Valencia, Spain), maximum, of beam, 3
138, 138 natural frequency and, 4
Tyne Bridge (Newcastle, UK), 149, 151 overhangs and tendon force
see also footbridge reducing, 96
British Grand Prix, Silverstone, UK, 80 proportional to span, 88
buckling (critical) load, deflection and, 4 simply supported beam, 91, 93 – 95
Building Research Establishment of structures, 1 – 5, 25 – 26
(UK), 74 superposition method for calculating,
buildings, see tall buildings 92, 93
design, rules of thumb for, 20 – 21
C designs, see bridge designs
Calatrava, Santiago, 115, 136, 138, 180
Cannon Street Underground Station E
(London), 83, 85, 86 Eastbourne, UK. temporary grandstand
cantilevers in, 76, 77, 78 – 79
spring effect factors as function of economy, 22, 78
stiffness ratio, 129, 129 efficiency of structure, 72 – 73
spring force Fx, 126 – 127 eight storey, four bay frame
superposition method for calculation, bracing arrangements for, 130, 131
127 – 128 computer application of uniform
uniformly distributed loads, 126 distribution of internal forces, 128,
with and without external elastic 131 – 135
support, 125 – 128 distribution and magnitude of internal
cast iron arch bridge (Manchester, UK), forces in members, 134, 134 – 135
flat deck on top, 148 – 149, 149 internal forces for four equivalent half
Cheesegrater, 68, see also Leadenhall frames, 132, 133
Building (London) elastic spring force, 162
 Index 191

elastic supports, see external elastic least effective bracing patterns, 61

supports; internal elastic supports most effective bracing patterns, 59, 60
elegance, 22, 78 physical model for relative stiffness,
equilibrium equation, 1, 27 – 29, 33, 37, 59, 62
131, 146, 166 symmetry simplifying model, 59
Evolutionary Structural Optimisation frame, see simple frame
(ESO), 125, 129 free-body diagrams, showing internal
bi-directional ESO (BESO), 145 – 148 forces, 26, 27 – 29, 28
comparison between BESO and free vibration, 3, 16
structural concept method (SCM), French Football Federation (FFF), 75
146 – 148, 147 fundamental natural frequency, 2
eight storey, four bay frame, 128, 130, buckling load capacity and, 5, 14,
131, 131 16, 22
topology optimisation method, deflection and, 4
145 – 148 floor system, 160, 160, 163
external elastic supports, 89 temporary grandstands, 78 – 79
cantilever with and without, 125 – 128 Furiani Stadium (Corsica, France), 75
forces in harp, 135, 136
Katehaki pedestrian bridge (Athens), G
139, 140 Gare do Oriente Station (Lisbon), 115,
Samuel Beckett Bridge (Dublin), 135, 116, 186
136 – 138 generality, structural concept of, 39 – 40
Serreria Bridge (Valencia, Spain), grandstands, see temporary grandstands
138, 138 Guinness harp, 136
structures with, 135, 136 – 139
H
F hand calculations, 17, 21
finite element (FE) method, 72, 131, beam with and without vertical
151, 166 internal elastic support, 155 – 163
ANSYS package, 57 cantilever with and without external
BESO and, 125, 147 elastic support, 125 – 128
Palazzetto dello Sport, 114, 114 effect of four bracing criteria, 52 – 57
Raleigh Arena (USA), 144 eight storey, four bay frame with
flexibility coefficient, stiffness matrix different bracing arrangements,
and, 32 – 34 128 – 135
footbridge most and least effective bracing
cable stayed mid-span support for, 18 patterns for simple frame, 57 – 64
Hessenring Footbridge (Germany), rigid plates supported by vertical and
114 – 115, 115 inclined members, 163 – 168
long-span, with overall depth limit, simply supported beam with and
150 – 153, 151 without overhangs, 90 – 96
Spinningfields Footbridge Y shaped columns with/without
(Manchester), 168 – 170 horizontal tendon, 96 – 104
steel prop supporting deck of, 45, 46 Harbin Airport Lounge (China), roof of,
see also bridge design 109, 110
Fountains Abbey (North Yorkshire, UK), harp, analysing forces in, 135, 136
92, 93 Heathrow Airport (London)
four-bay and four-storey frame roof supports of Terminal 5, 178, 179,
ANSYS finite element method 180, 186
package, 57 Y shaped column with horizontal bar,
braced panels in each storey, 58 103, 104
bracing orientations, 58 Hessenring Footbridge (Germany),
bracing patterns for, 57 – 64 114 – 115, 115
geometry of, 58 HSBC Hong Kong headquarters
lateral deflections, 62, 63, 64 (China), 186
192 Index

front view of, 105 intuitive interpretation of, 34 – 36

loading and internal forces on two 3-bay and 3-storey frames, 40
mast, 109 see also more direct internal force
model of, 106 paths; more uniform distribution
models of mast and truss system, 108 of internal forces; smaller internal
structures with overhangs, 104 – 109 forces
truss supporting, 107 International Women’s Tennis
Championship, 76
I interpretation
inclined members replacing vertical hand calculation, 17
members, 175 – 180 intuitive, of internal forces, 34 – 36
Alamillo Bridge (Seville, Spain), 180, intuitive, of structures, 13 – 17
180 – 181 mathematical equations, 14
Ontario College of Art and Design observation of structural behavior, 14,
(Toronto), 177 – 178 16 – 17
rigid plates supported, 163 – 168 vibration-buckling experiment, 14,
roof supports of Terminal 5 at 15, 16
Heathrow Airport (London), 178, intuition of structures
179, 180 intuitive interpretation, 13 – 17
Sharp Centre for Design (Toronto), intuitive knowledge, 10 – 12
177, 177 – 178 intuitive understanding, 12 – 13, 
three types of support to 141
superstructures, 175, 175, 176, 177
interchangeability, structural concept of, J
40 – 41 John Hancock Center (Chicago),
internal elastic supports, 89 64 – 68, 69
beam with and without a vertical, bracing systems, 65, 66
155 – 163 global X braces of, 65 – 66
demonstration of wire acting as, 143 structural effectiveness and efficiency,
evolution of models for intuitive 64 – 66
understanding, 141
illustration of deformations of tied K
ring, 142 Katehaki pedestrian bridge (Athens),
Manchester Central Convention 139, 140, 186
Complex (UK), 139 – 143 Khan, Fazlur, 64, 65
Raleigh Arena (USA), 143 – 145 Kirchheim Overpass (Germany), 152,
roof of Badminton Arena (Beijing 152 – 153
Olympic Games), 170 – 173, Knaresbough, North Yorkshire, railway
174, 175 station, 102, 103
roof of Shanghai Pudong Airport
terminal, 170, 171 L
Spinningfields Footbridge Leadenhall Building (London), 68 – 73
(Manchester), 168 – 170 “Cheesegrater”, 68
structures with, 168 – 175 diagrid structures of, 70 – 72
using bar/string members to create, East/West frame, 71, 73
154, 155 efficiency of structure, 72 – 73
internal forces elevation of, 71
3-bay and 4-storey truss structures, 42 front view of, 69
comparing, of two simple frames, 42 North frame, 71
deflection and energy of structure, side view of, 70
29 – 32 South frame, 71, 73
deflections and energies of two roads, Lowry Bridge, 121
26 – 29
form, deflection and, 5 – 10 M
free body diagrams showing, 26, Madrid Barajas Airport (Spain), 116,
27 – 29, 28 118, 119
 Index 193

Madrid Racecourse (Spain), 119 – 121, structures with external elastic

186 supports, 135, 136 – 139
design of, 120 structures with internal horizontal
front view of, 119 elastic supports, 139 – 145
physical model of, 120 see also topology optimisation
simplified model of two roofs and rod structures
linking, 121
Manchester Central Convention N
Complex (UK), 139 – 143 natural frequency
evolution of models for intuitive comparing lateral stiffness and, of
understanding, 141 frames, 79
front view of, 140 deflection and, 4
illustration of deformations of tied principle horizontal, of temporary
ring, 142 grandstands, 74
internal horizontal elastic supports, stiffening floor beam for
139 – 143 increasing, 160
Manchester Central Railway Station, 139 see also fundamental natural frequency
maximum bending moment, 4 – 5 Nervi, Pier Luigi, 10, 113
more bending moments converted to nodal displacement, 2, 3
axial forces nodal displacement vector, 33, 146
beam with/without vertical internal North Stand at Twickenham, 12, 13
elastic support, 155 – 163
hand calculation examples, 155 – 168 O
practical examples, 168 – 181 Ontario College of Art and Design
rigid plates supported by vertical and (OCAD), 177 – 178
inclined members, 163 – 168 Onterie Center (Chicago), 66, 67
routes to implementation, overhangs
154 – 155, 186 curved beams with, 94
self-weight of structured members, HSBC Hong Kong headquarters,
180 – 181 104 – 109
structural concept of, 154 – 155,  reducing bending moments and
186 deflections of cellular beams, 96
structures supported by inclined roof of Harbin Airport Lounge
members, 175 – 180 (China), 109, 110
structures with vertical internal elastic simply supported beam and
supports, 168 – 175 variations, 90
see also beam-string structures; simply supported beam with and
inclined members replacing vertical without, 90 – 96
members structures with, 104 – 109
more direct internal force paths superposition method for calculating
structural concept of, 51 – 52, 83, deflection, 92, 93
86, 186
see also bracing systems P
more uniform distribution of internal Palace of the Grand Master of Knights of
forces Rhodes, 47, 48
cantilever with/without external elastic Palazzetto dello Sport (Roma),
support, 125 – 128 113 – 114, 186
eight storey, four bay frame with finite element (FE) model of, 114
different bracing arrangements, outlook of, 113
128 – 135 panel, definition, 57
hand calculation examples, 125 – 135 pin-connected structure, maximum
practical examples, 135 – 153 deflection of, 25
routes to implementation of, 125 plane pin-jointed structures
structural concept of, 125, 186 bracing arrangements, 7
structures derived from topology four-bay and four-storey, 6 – 10
optimisation, 145 – 153 internal forces, 8
194 Index

Pudong Airport Terminal (Shanghai), Serreria Bridge (Valencia, Spain), 138,

116, 117 138, 186
roof of, 170, 171 serviceability, 1 – 2
Shanghai Pudong Airport Terminal 1
R roof, beam-string structures, 170,
railway station, Y shaped columns with 171, 186
horizontal bar, 102, 103, 104 Sharp Centre for Design (Toronto), 177,
Raleigh Arena (USA), 186 177 – 178, 186
finite element model, 144 simple frame
illustration models, 145 example of four-bay and four-storey
internal force paths, 144 frame, 57 – 64
roof structure of, 143, 144, 185 lateral deflections for frames, 62,
reciprocal theorem of work, 31 – 32 63, 64
relative performance, structural concept least effective bracing patterns, 61
of, 44 – 45 most effective bracing patterns, 59, 60
reversibility, structural concept of, 43 – 44 physical models for relative stiffness of,
rigid frame structure, maximum 60, 62
deflection of, 25 symmetry simplifying frame model, 59
rigid plates see also bracing systems
supported by four inclined members simply supported beam
with pin connections, 164, 165 – 166 bending moment and deflection, 91,
supported by four inclined members 93 – 95
with rigid connections, 164, curved beams with overhangs, 94
166 – 168 effectiveness and efficiency of, 170
supported by four vertical members, example with and without overhangs,
164, 165 90 – 96
vertical and inclined members hand calculations, 91, 93 – 95
supporting, 163 – 168 reduction of bending moment and
rods, internal forces, deflections and deflection, 93
energies of two, 26 – 29 roof of Shanghai Pudong Airport
rules of thumb, 21, 39 terminal, 170, 171
for design, 20 – 21 steel-framed two-storey car park,
development of, 20, 21 95, 96
see also structural concepts structures of, 156
superposition method for calculation,
S 92, 93
safety, 2, 22, 74, 78 variations of, 90
Salford Quays Lift Bridge (UK), 121, 122 smaller internal forces, 186
Salvadori, Mario, 10 hand calculation examples, 90 – 104
Samuel Beckett Bridge (Dublin), 135, partially self-balancing, 88, 93
136 – 138, 137, 186 practical examples, 104 – 122
scaffolding structures providing elastic supports for, 89
collapse of, in Manchester, 82, 82 reducing spans for, 88, 93
lacking direct internal force paths, routes to implementation of, 88 – 89
83, 84 self-balancing examples, 119 – 122
Schlaich Bergerman Partner, 114 simply supported beam with/without
self-balancing overhangs, 90 – 96
Madrid Racecourse (Spain), 119 – 121 structures with overhangs, 104 – 109
partially, internal forces, 88 tree-like structures, 110 – 119
reducing bending moment through Y shaped columns with/without
part, 93 – 94 horizontal tendon, 96 – 104
Salford Quays Lift Bridge (UK), see also Y shaped columns
121, 122 Spinningfields Footbridge (Manchester,
using Y shaped columns, 96 – 104 UK), 168 – 170, 169, 182, 186
self-weight of structural members, spring force Fx, calculation of, 126 – 127
180 – 181 stability, 2, 3
 Index 195

steel tendons, support for structure, 47, T

48 – 49 tall buildings, 186
stiffness matrix, 2, 32 – 34 Bank of China (Hong Kong), 67, 68
strain, 2 John Hancock Center (Chicago),
strain-displacement matrix, 3 64 – 68
strength, 3 Leadenhall Building (London), 68 – 73
structural concepts, 185, 186 Onterie Center (Chicago), 66, 67
characteristics of, 39 – 45, 49 temporary grandstands
compatibility, 41 – 43 activities utilizing, 73 – 74
definition of, 17 bracing patterns of two, 80, 81
generality, 39 – 40 British Grand Prix (Silverstone,
implementation of, 45, 48 – 49 UK), 80
interchangeability, 40 – 41 collapse of, in Corsica, France, 75,
more direct internal force paths, 83, 86 75 – 76
more uniform distribution of internal comparing lateral stiffness and natural
forces, 125 frequencies of frames, 79
relative performance, 44 – 45 comparing shapes of fundamental
reversibility, 43 – 44 modes of frames, 79
routes to implement, 23, 88, 125, cross-section of collapsed, 75
154 – 155 design of bracing systems in, 78
rules of thumb, 21, 39 in Eastbourne, UK, 76, 77, 78 – 79
structural design problems, relationships internal force paths of, 75
between, 3 plane model for, 81, 81
Structural Engineer (journal), 21 principle horizontal natural frequencies
structural engineering, definition, 21 – 22 of, 74
structural form structural characteristics of, 74
deflection and internal forces, 5 – 10 see also bracing systems
internal forces determining, 6 tendons
structural vibration, serviceability, 1 – 2 reducing bending moments and
structures deflections of cellular beams, 96
deflection of, 1 – 5 stiffening floor beam with, 160,
with external elastic supports, 160 – 161
135 – 139 support for structure, 47, 48 – 49
internal forces, deflection and energy Y shaped columns with and without
of, 29 – 32 horizontal, 96 – 104
with internal horizontal elastic topology optimisation structures
supports, 139 – 145 bridge with flat deck at middle level,
intuition of, 10 – 17 149 – 150, 150
intuitive interpretation of, 13 – 17 bridge with flat deck on top, 148 – 149
intuitive knowledge, 10 – 12 Evolutionary Structural Optimisation
intuitive understanding of, 12 – 13 (ESO) method, 145 – 148
with overhangs, 104 – 109 long-span footbridge with overall
supported by inclined members, depth limit, 150 – 153
175 – 180 Torroja, Eduardo, 10
tree-like, 110 – 119 tree(s)
using self-weight of structural in architectural and structural designs,
members, 180 – 181 110 – 111
with vertical internal elastic supports, see also Y shaped columns
168 – 175 tree-like structures, 110 – 119
see also scaffolding structures; tall Hessenring Footbridge (Germany),
buildings; topology optimisation 114 – 115
structures Palazzetto dello Sport (Roma),
supports 113 – 114
examples of, 45, 46, 47, 48 – 49 structural hierarchy of oak tree, 111
providing elastic, 89 Y shaped column as, 110 – 115
reducing spans between, 93 see also Y shaped columns
196 Index

truss structure calculating deflections, 98 – 102

additional supports to floor for, 44 Chengdu East Railway Station,
internal forces, deflection and energy 122 – 124, 123
of, 29 – 32 dimensions of, 97
two sets of loading on, 30 Gare do Oriente Station (Lisbon),
Tyne Bridge (Newcastle, UK), 149, 151 115, 116
Hessenring Footbridge (Germany),
U 114 – 115, 115
UK Institution of Structural Engineers, 21 HSBC Hong Kong headquarters,
104 – 109
V Madrid Barajas Airport, 116, 118, 
vertical supports, suspended four-storey 119
link structure, 46, 48 Madrid Racecourse (Spain),
vibration 119 – 121
isolation measures, 11 Palazzetto dello Sport (Roma),
response spectra of North Stand at 113 – 114
Twickenham, 13 Pudong Airport Terminal (Shanghai),
understanding, 12 – 13 116, 117
vibration-buckling experiment, observing roof of Harbin Airport Lounge
structural behavior, 14, 15, 16 – 17 (China), 109, 110
Salford Quays Lift Bridge (UK),
W 121, 122
work, reciprocal theorem of, 31 – 32 structures with overhangs, 104 – 109
work-energy principle, 30 tree-like structures, 110 – 119
use with horizontal bar, 102, 103, 
X 104
Xi’an North Railway Station (China), variable cross-section of arms of,
183, 183 182, 182
vertical loads and inclined members,
Y 182 – 183
Y shaped columns, 186 with and without horizontal tendon,
bending moment diagrams of, 98 96 – 104
Bihai Cultural Centre (Tianjin, China), Xi’an North Railway Station (China),
116, 117 183, 183