UNDERSTANDING STRUCTURAL ANALYSIS

E‐Training CourseT
1MFBTF
HP
UP
XXXEBWJECSPIODPN
GPS
GVSUIFS
JOGPSNBUJPO
PO
POMJOF
USBJOJOH

DPVSTFT

h6OEFSTUBOEJOH
4USVDUVSBM
#FIBWJPVSh
JT
the first of the three that go to make up

the suite of e‐courses:
Understanding Structural Modelling.

%BWJE
#SPIO
IBT
CFFO
BXBSEFE
UIF
.#&
JO
UIF
+VCJMFF
)POPVST


David Brohn has been awarded the

Institution of Structural Engineers President’s Award

“The President’s Award is to be presented to Dr David Brohn for his visionary

approach to the education of students and graduates in structural engineering.
For several decades, David’s ‘Brohn Test’ has been used by educators and
employers to measure and embed an understanding of structural behaviour.
David’s extraordinary efforts over a prolonged period of time have underpinned the
Institution’s educational initiative in developing the Structural Behaviour Course.
It is for this leadership and vision that David is to receive the President’s Award.”

Professor Tim Ibell, Bath University

President, Institution of Structural Engineers, 2015

www.linkedin.com/in/drdavidbrohn

David Brohn
[email protected]
Understanding
Structural
Analysis
,JOEMF
WFSTJPO
DAVID BROHN
.#&
PhD, CEng, FIStructE
Managing Director,
New Paradigm Solutions Ltd

Foreword to First Edition by Sir Ove Arup

New Paradigm Solutions Ltd

www.newparadigms.co.uk
Copyright © David Brohn 1984 – 2007 Published by New Paradigm Solutions Ltd
5 BMBDL
/PSF
1PJOU
All rights reserved. No part of this publication Portishead
may be reproduced, stored in a retrieval BS20 8(#
system, or transmitted, in any form or by any United Kingdom
means, electronic, mechanical, photocopying,
recording or otherwise without the prior [email protected]
permission of the publisher.

First Edition published 1984

Second Edition published 1990
Reprinted 1991
Third Edition published 2005
Reprinted 2007 – 2019
,JOEMF
7FSTJPO

British Library Cataloguing in Publication Data

Brohn, David
Understanding structural analysis. – 3rd ed.
1. Structures. Analysis
I. Title
624.1'71
This edition designed and produced by StroudPrint.
ISBN 978-0-9556311-0-8 Printed in Great Britain
Contents

Foreword to the First Edition

Foreword to the Third Edition
Preface

Part I
1 The Analysis of Statically Determinate Structures 3

2 Statical Indeterminacy 21

3 The Qualitative Analysis of Beams 37

4 The Qualitative Analysis of Frames 57

Part II
5 The Theorems of Virtual Work 75

6 The Flexibility Method 97

7 The Stiffness Method – Frames 115

8 The Stiffness Method – Grids 137

9 Moment Distribution 149

10 Plastic Analysis of Plane Frames 179

11 The Yield Line Analysis of Reinforced Concrete Slabs 203

12 Influence Lines 223

Appendix: Solutions to Practice Problems

Foreword to the First Edition

A force is not just a straight line with an arrow head at one end. That is just a convenient
abstraction or shorthand for what in real life turns out to be a bundle of particles under stress and
strain, always changing and moving under the slightest provocation from changing circumstances.
The theory of structures and, in fact, our whole scientific apparatus is founded on such
abstractions. They have enabled us to impose some order on the chaos with which we are faced
when we look at the unending and overwhelming wonders of nature that far exceed our powers of
comprehension. We have even found that, if we assume that this imaginary world of science is a
true picture of reality and act accordingly, we can influence and change the world we live in to such
an extent that we can abolish want and drudgery and, in fact, do almost anything we like, including
destroying the planet we depend on, together with its fauna and flora, in a few weeks – if we only
could agree where to start.

This whole mechanistic world-picture, the Cartesian or Newtonian world of science, is under
heavy strain now; we find in all disciplines that it does not work any more, nature simply does not
collaborate. I have no time to elaborate on that now and there are, anyhow, thousands of books
and pamphlets written about that theme. But how does this affect the theory of structures and the
whole business of structural engineering? Our structures are all the time getting better, bigger,
lighter and safer, our machines are more efficient, smoother, using less energy. We are all the time
learning to do more with less, so what is wrong with that?

As David Brohn points out in his book, there is a critically important stage to be reached before
we can even apply our numerical analysis to a structure, namely that we must have a structure to
apply it to. When we have that our analysis will tell us whether the structure is capable of doing
what it is supposed to do. Further, that the skills required to choose this preliminary structure
are of an entirely different nature from those we may gradually attain when we have mastered our
structural techniques.

Brohn confines himself to suggesting that the basis of these skills is the recognition of the
relationship between the load and the resulting behaviour of the structure, in other words that we
gain an intuitive understanding of how a structure will behave under load. This will, when we have
acquired the necessary experience, enable us to choose the at least approximately right structure
for a given task just by looking at its shape and proportions on a drawing. This is of course most
important, for the increasing use of computers has to a great extent killed this understanding
which is so essential for rescuing the art of structural design.

The computer has come to stay, we must live with it, and this book teaches us how to do so whilst
remaining masters of the proceedings. We must be able to check that the output from the computer
is correct, and where and when to use it, and what its limitations are. If this book did nothing
else it would still be the most important contribution to structural design which has appeared for
a long time and should be compulsive reading for anyone interested in the subject. I have been
extremely worried by the fact that gifted graduates from our universities enter the profession
with the idea that it is below their dignity to put pen to paper - the computer does it all. Here
we have the necessary remedy for such conceit, and it is high time. Unfortunately I have neither
the time nor the knowledge to do justice to the achievements of Brohn’s book, I have only a few
hours left for the printers’ deadline, which is entirely my own fault. But he has seemingly gone
through every known method of structural analysis and shows by clear diagrams and explanations
how the structure is affected by the loading, thus at each stage giving the reader this essential
understanding of structural behaviour. I hope this book will give rise to a lively discussion.

I cannot resist adding a comment of my own. Whilst recognising the importance of what Brohn
has done, I do not think he has gone far enough. An understanding of structural behaviour is very
necessary, but are there not many more things that are equally or even more necessary?

Every structural designer of repute has declared that structural design is an art as well as just an
application of science and technique to a given problem. You could also put it the other way round
and say that only if it is a work of art as well will it be admired and add to the reputation of the
designer. It is unfortunately impossible to define what art implies, but it has in any case nothing to
do with numerical analysis. There are many other matters to consider as well. The whole purpose
of a structural design is to help us to make the things we need, or fancy we need, or just fancy.
So we must make it very clear to ourselves what we want to achieve with our design, which will
obviously affect its shape, the materials we use and all kinds of other things. If we want to build
something is this the right place for it, could not our purpose be better achieved in a different way
altogether? Only when we have sorted out all these matters to our and our clients’ satisfaction will
a structural analysis become relevant. Obviously, what I would call design is much more important
than structural analysis, for that determines what we are going to get for our efforts. And moreover
what we decide to do is much more important than how to do it, and that opens the sluice valve for
a whole flood of questions, social, political, ethical, which threaten us all with confusion, or worse,
because we are not able to agree on what to do.

How to live in peace with our neighbours on this planet without destroying it is the ultimate and
now pressing problem and I wish I knew the answer.

Ove Arup
Foreword to the Third Edition

When the second edition was published in 1984, the desktop computer revolution had only just
begun and it was not until Microsoft established a universally accepted operating system that
software developers could see the commercial sense of developing complex systems for structural
analysis and design.

I had just set up QSE Ltd and started to develop a suite of software with a highly intuitive graphics
user interface. Later as a partner in the CIMsteel Project, the system was rewritten under Windows
to become QSE Plane and QSE Space.

In the 20 years since the second edition, the engineering profession, especially structural
engineering, has seen a paradigm shift from a time when all calculations in the design office were
carried out by hand to the present time when virtually all detailed analysis and design is carried out
on a computer.

This has placed a particular emphasis on the modelling of the structure and in particular the
creation of the analysis/design model that will be the basis of the computer analysis.

In that process, the physical reality must be simplified, support conditions rationalised and
so on and the result must be that the final analysis/design model is weaker than the real
structure.

As this is written, there have been three recent roof failures: the Siemens Arena in Copenhagen,
the swimming pool roof in Moscow and the most dramatic of all, the collapse of a part of the roof to
the newly-built terminal at the Charles De Gaulle airport in France.

Although the detailed explanation of the failure will be different in each case, one feature will be
the same: the analysis/design model will be found to be stronger than the real structure. Therefore
the failure is in the process of conceptual design that created the analysis/design model and in that
process the computer is of little help.

The most familiar term for the skills required to be effective in this process is ‘engineering
intuition’ and this book sets out the starting point for the development of this skill. If engineering
intuition is a language, then a sound understanding of structural behaviour is the alphabet and
grammar.

How that understanding is used to solve the real, complex modelling of structure will be dealt with
in the new format of E-Training developed for the Institution of Structural Engineers and providing
approved Continuing Professional Development over the Internet. For a link to E-Training, go to
www.newparadigms.co.uk.

The third edition now includes full versions of QSE Plane and QSE Space limited to 39 nodes, but
with full functionality.

David Brohn PhD, CEng, FIStructE

Preface

This book is aimed at the identification of the fundamental principles of structural analysis together
with the development of a sound understanding of structural behaviour. This combination leads to
the ability to arrive at a numerical solution.

Using a series of structural diagrams as a visual language of structural behaviour that can
be understood with the minimum of textual comments, the book aims to develop a qualitative
understanding of the response of the structure to load. It is ideally suited to undergraduates
studying indeterminate framed structures as part of a core course in civil or structural engineering,
but it is also suitable, because of its qualitative approach, for students of architecture and building
technology.

The first four chapters deal with the development of qualitative skills; that is, the ability to produce
a non-numerical solution to the loaded line-diagram of a structure. It is considered that the ability
to arrive at the qualitative solution to framed structures is a significantly important component of
the overall understanding of structural behaviour.

The remaining chapters deal with methods of structural analysis using the diagrammatic format to
which the student has become accustomed.

The need for the development of qualitative skills increases with the increasing use of the computer
in design offices. The computer has already replaced the majority of analysis and structural design
calculations. Unfortunately, this will also have the effect of eliminating much of the experience and
consequent understanding gained by the student and trainee engineer. The importance of a sound
and reliable understanding of structural behaviour is now of the highest priority.

In the third edition a fully functional copy of QSE Plane and QSE Space, limited to a student
version number of nodes, is included. Instructions on how to download the software are included.

The computer software included with this book is available from:

New Paradigm Solutions

For download details, see the last page of the book.
Email me with the reference number and password on the label.

David Brohn
Part I
1 The Analysis of Statically
Determinate Structures

The subject of this book is the behaviour and analysis of statically indeterminate
structures. However, this first chapter reviews the behaviour of determinate structures,
a thorough understanding of which is essential before the topic of indeterminacy can be
tackled. The text assumes a basic knowledge of mechanics including an understanding of
the principles of overall equilibrium, bending moments, shear and axial forces.

It is possible to analyse determinate structures by consideration of equilibrium – in

general terms, the application of force and moment equations V = 0, X = 0 and M = 0.

With most real structures, this is not possible as the presence of redundant members
(secondary load paths) makes it necessary to consider relative member deformation before
a solution of the structure can be attained. The number of unknowns which cannot be
found from equilibrium considerations is known as the degree of statical indeterminacy.

The design of engineering structures usually starts from a need to sustain loads. Initially
though, it requires an understanding of the way in which a proposed system of members
can provide the required support, and how it will deform.

It is, however, clear that an understanding of the behaviour of statically indeterminate

systems is based upon a thorough appreciation of determinate systems.

This chapter develops the relationship between load and deformation for a range of
structures that are amenable to solution by the application of equilibrium alone.

Once we have analysed the behaviour of the proposed structures we are then able to start
an appropriate process of numerical analysis to find out how much of each of the various
parameters is involved, for example, the values of the loads carried by each member and,
as a consequence, the size each will have to be to carry its load safely. We can go on to
find the values of deformations that will result from the loading.
4 Understanding Structural Analysis

1 1 Thus we see that structural analysis must have two

components. The first is a qualitative understanding
and the second the numerical procedures. It must be
understood that qualitative analysis is not in any way a
substitute for numerical analysis but should be regarded as
a necessary complement to it so that the two approaches
constitute a complete whole giving an understanding of,
and an ability to evaluate, the structural performance.

2 2 The basic principles of our structural analysis lie in the

laws of statics. Most structures are required to be stable
in a static state. This is not to say that we cannot analyse
dynamic behaviour such as may be caused by earthquakes,
wind gusts or moving loads but initially, at least, we
shall concern ourselves with statics. The tower crane
demonstrates the kinds of equilibrium conditions that we
have to satisfy. The vertical reaction at the ground must
balance the total downwards force of counterbalance and
load.

3 3 Secondly, we can see that wind forces will tend to make the
whole crane structure slide sideways and this too, must be
resisted, by horizontal support forces at the base. This we
call horizontal equilibrium.

4 4 Lastly, it is clear that the counterbalance weight cannot

balance all conditions of loading on the jib so that any out-
of-balance will tend to topple the crane. If we add to this
the additional toppling effect of the wind forces we see
that the base of the crane will have to provide resistance
to these out-of-balance toppling moments. This we call
moment equilibrium.
 The Analysis of Statically Determinate Structures 5

5 In our three-dimensional world we can express these 5

equilibrium requirements in the following way. Firstly, we
must ensure that in each of three directions, at right angles
to each other, which we will label X, Y and Z, the resultant
of all forces acting on the structure must be zero. In other
words, reactions must balance loads. Secondly, as we have
seen in the case of the toppling effect, the tendency for
the structure to rotate about any of these three axes must
be resisted. We say that the resultant moment about each
of the three axes X, Y and Z must also be zero. Thus the
moment of reactions must balance the moment of loads.
This gives us all together six conditions of equilibrium.

6 In much of the following explanation of behaviour, and in 6

many real life engineering situations, we find it is possible
to be sure that the forces in the direction Y are zero and
that there are no moments of forces about the X and Z
axes. If, indeed, this is the case, then our problem can
be reduced to the consideration of three conditions of
equilibrium only. Such simplification is described as a plane
problem because all forces lie in one plane.

7 We shall find it convenient to label the forces in the Y 7

direction with symbol H, to denote horizontal, and those in
the Z direction V, to denote vertical. Also we shall use the
symbol M for moments in the plane, about the Y axis.

8 With these symbols we can write down the three equations 8

of equilibrium - all the horizontal forces must sum to zero,
all the vertical forces must sum to zero and the sum of the
moments must also be zero.
6 Understanding Structural Analysis

9 9 Now it will be remembered that a pure moment, called

a couple, can be represented by two equal and opposite
parallel forces at a distance apart. In this diagram we have
the forces at a distance d giving a moment M = F × d.

10 10 Let us now consider the moment of this couple about

two points, A and B in the plane. Considering first the
point A we see that as the downward force F passes
through A it will have no moment about A and all that
remains is the anticlockwise moment of the upward force
F at a distance d. The moment about A is MA = F × d
anticlockwise. Considering the point B, the upward forces
will pass through the point and the only moment is caused
by the downward force F at distance d. This also gives an
anticlockwise moment MB = F × d.

11 11 Let us now consider the moment about a point C distance

e from the line of action of the upward force. We can also
find the total moment of our forces F about this point. For
the downward force F there is an anticlockwise moment
due to the lever arm, d plus e, and for the upward force F
a clockwise moment due to the lever arm e. The resultant
moment MC is the force F multiplied by the separation d, still,
of course, anticlockwise. What we have shown is one of the
most powerful aids to structural analysis, namely that, as a
couple has the same moment about any point in the plane, the
equilibrium of a structure will require that the total moment of
all forces must be zero about any point in the plane.

12 12 Let us now consider an actual structure. The beam ABCD

is simply supported at A and B and loaded at C. We shall
ignore the effects of any self-weight and only study those
due to the applied load W.
 The Analysis of Statically Determinate Structures 7

13 We can start the qualitative analysis of the structure by 13

observing that since there must be equilibrium of moments
about the support A the only force which can balance the
clockwise moment of the load W at C will be an upward
support reaction at B, giving the necessary anticlockwise
moment. Furthermore, since the support B is on rollers,
only a vertical force VB can act at this point.

14 As the force VB has to produce a moment about A 14

balancing that of the load W, we can observe that it will
have to have a magnitude greater than W because its lever
arm about A is less. Thus if there is to be equilibrium of
vertical forces, the vertical support reaction VA at A will
necessarily be a downward force to balance the excess of
VB over W.

15 We can also use the requirement that moments sum to 15

zero about any point in space to confirm this result. Taking
moments about B we observe that there is a clockwise
moment due to the load W at C. A balancing anticlockwise
moment is required. Thus VA must act downwards.

16 Lastly, we should observe that as neither the force W 16

nor the reaction VB has a horizontal component the
requirement of horizontal equilibrium ensures that the
horizontal support reaction at A is zero.
8 Understanding Structural Analysis

17 17 The concept of equilibrium also applies to any part of the

structure. If we consider an imaginary cut separating one
part of the structure from the rest, it is clear that internal
forces at the cut will be required to maintain equilibrium.
Let us suppose that we separate part of the structure to the
right of B by introducing a cut at K.

18 18 If the portion of the beam KCD is to be in overall vertical

equilibrium then there will have to be an internal vertical
force S at the cut point K to balance the load W. It must
act from below upwards on the part KCD and have a value
equal to the load. This internal force is called the shear
force.

19 19 The position at which we choose to place the cut K

between the support point B and the load point C has
no effect upon the value of the shear force, which will
be constant from B to C. To the right of C there are no
external forces and so the shear force reduces to zero in
the portion of the beam CD. We can plot the distribution
of S as a shear force diagram, drawing it underneath the
base line, that being where we have shown the shear force
arrow.

20 20 The diagrammatic convention for shear force is that the

value of the shear force is plotted on one side of the base
line, above or below; that is corresponding to the view an
observer would have sitting behind the cut K. The load
W would cause the part of the beam KCD to fall below
the beam, thus the diagram is plotted below the base line.
The arrow on the base line is required to confirm this
convention.
 The Analysis of Statically Determinate Structures 9

21 For a cut K in the beam between A and B we see that 21

the greater magnitude of VB compared with W will lead to
vertical equilibrium requiring a shearing force S from above
acting downwards at K, being the difference between VB
and W.

22 Again the choice of the position of K between A and B 22

does not affect the equilibrium equation and thus S will
also have a constant value of shear force between these
points.

23 Suppose we had considered the vertical equilibrium of the 23

left-hand portion of the beam from K to A. You will see
that this will lead to an upward shear forces to balance the
only vertical force on this portion of the beam, the support
reaction VA.

24 This is a necessary result because, when we close the cut 24

by putting the two sides of the beam together, the two
shear forces must cancel each other leaving no external
out-of-balance vertical force at the cut section.
10 Understanding Structural Analysis

25 25 It only remains to be clear about the plotting convention

for the diagram of shear force distribution. We have shown
a horizontal arrow pointing to the right to indicate that
we were looking at the force acting on the right-hand side
of the cut. We could equally well make the arrow point
to the left and look at the force acting on the left-hand
side of the cut when clearly the shear force diagram in
this case becomes a vertical reflection of the first case.
Either diagram is correct but it is important to include the
horizontal arrow to show which way we are looking.

26 26 Let us now return to the moment equilibrium at the

imaginary cut point K between B and C. We see that the
load W produces a clockwise moment about K, shown
as MW(K) and for equilibrium, this must be resisted by an
internal anticlockwise moment within the beam at the cut
K itself. As the shear force S passes through K it can
make no contribution to moment equilibrium. This internal
moment is called the bending moment.

27 27 As we move the cut point K towards the load at C the

lever arm of the load W will reduce so that the balancing
internal bending moment will decrease proportionately.
If K coincides with C the bending moment will be zero.
Thus we can plot, on the structure as base, a linear graph
representing the distribution of the bending moment from
zero at C to a maximum value at B. Clearly, as there are
no external forces beyond C, the bending moment in the
portion of the beam CD must be zero.
 The Analysis of Statically Determinate Structures 11

28 Suppose now that we place our imaginary cut point K 28

within the span length AB. In considering equilibrium of
the whole structure we have already observed that the
upward support reaction VB, whilst being larger than the
load W, will only produce equal moments about A. Thus at
an internal point K, less far away than A, the clockwise
moment of W will still exceed the anticlockwise moment of
VB and the additional anticlockwise moment of resistance
will have to be supplied by the internal bending moment.

29 It should be clear that this internal bending moment will 29

be equal to that caused by the load W alone when we place
the cut point at the support B. The effect of the larger
support force VB will reduce the required balancing internal
moment as we move to the left from B until, in order to
satisfy overall moment equilibrium, it will necessarily
disappear altogether when we arrive at A. We can thus plot
the full distribution of bending moment along the beam.

30 Had we considered moment equilibrium of the left-hand 30

portion of the beam with the cut K in the span AB, we
should have found the requirement for a clockwise internal
bending moment to balance the moment effect of the
downward support reaction VA. We can also note that the
value of this bending moment decreases to zero as the cut
point approaches the support point A.

31 If again we consider the imaginary cut being closed up, the 31

two bending moments on the two portions of beam, one
clockwise and one anticlockwise, will cancel and leave no
eternal moment acting at the section.
12 Understanding Structural Analysis

32 32 It is to be noted that the bending moments we have shown

on both sides of the cut will cause bending tensile stresses,
and therefore tensile strains, on the upper fibres of the
beam. We have already adopted the convention that we
draw the bending moment diagram on the tensile side of
the beam that we may indicate with a ‘T’. We always use
the axis of the structure as the baseline for the bending
moment diagram, the ordinates of which are drawn normal
to the baseline.

33 33 Tensile strains in the upper fibres and corresponding

compressive strains in the lower fibres of the beam section
cause it to bend into a curve which is called a hogging
curvature. The centre of curvature, in such a case, lies
below the beam axis.

34 34 Returning to our particular beam problem we can see that

we can also give a qualitative solution to the deflected
shape, as in this diagram. We know that it will hog and
that there will be no deflection at the support points A
and B. We will expect there to be a deflection at the load
point in the direction of the load. Note, in addition, that as
there is no bending moment in portion CD, this will not
bend. However, this portion will deflect and rotate as it is
attached to the rest of the structure ABC which does bend.
You should particularly notice a coherence between the
deflected shape and the bending moment diagram.
 The Analysis of Statically Determinate Structures 13

35 We will end consideration of this very simple problem by 35

observing the various characteristics of the solution which
show consistency and lead us to be satisfied that our
qualitative solution is correct:

(i) The bending moment diagram shows that there are no

moments at the end simple support.

(ii) It also shows that the slope of the diagram changes

abruptly at each point where there is a concentrated
external force and that it is linear between, i.e. where
there are no other external forces.

(iii) We have already observed that there is a general

coherence between the shapes of the bending moment
diagram and the deflections, with hogging curvature
wherever the bending moment diagram is drawn
above the beam axis.

(iv) The shear force diagram is constant where there

are no external forces on the beam and it changes
abruptly at the load points.

36 We shall now look at a slightly more complicated problem 36

in which our beam is extended by having another portion
hinged to it and supported at the right-hand end. We note
that the support system has been arranged so that there
can be no horizontal support forces at B and E and the
applied load W at D is in the vertical direction.
14 Understanding Structural Analysis

37 37 We can now proceed more rapidly to use the various

equilibrium conditions to determine the directions of the
support forces. Firstly, as the hinge provides no moment
of resistance, equilibrium of moments of CDE about
C require an upward force VE to balance the clockwise
moment caused by W. However, because the lever arm of
VE is greater that that of W its value will be less. Thus
vertical equilibrium of the portion CDE requires an upward
shear force SC to act at the hinge point C. An upward shear
on CDE at C will imply an equal downward shear at C on
the part of the beam CBA. Moments about B will require
downward support force VA to produce an anticlockwise
moment to balance the clockwise moment due to SC.
Lastly, vertical equilibrium of ABC requires an upward
support reaction VB to balance the two downward forces
VA and SC. This is just one possible sequence of reasoning
which leads us to establish the full set of support reactions
shown.

38 38 We are now in a position to start drawing the distribution of

the internal bending moment, that is the bending moment
diagram. From A to B the reaction VA must cause tension
on top of the beam, consequently the bending moment
diagram is plotted above the beam axis. From E to D, the
reaction VE causes bending tension on the underside of the
beam axis.

39 39 In the portion of the beam BCD we know that at the

position of the hinge C the internal moment must be zero
whilst at B and at D it must reach the same values as that
in the neighbouring spans. In addition to this, however, we
realise that the moment developed from C to B and that
from C to D is caused by the same shear force SC. Thus
the slope of the bending moment must be the same on
each side of the hinge. The bending moment diagram will,
therefore, be given by a straight line from B to D passing
through zero, that is the beam axis, at C.
 The Analysis of Statically Determinate Structures 15

40 At this stage we can draw the shear force diagram. 40

Remember that we must indicate whether we are looking
at the force on the right- or left-hand side of a cut. Here
we are looking at the right-hand side. Notice that a hinge
carries a shear force, and if there is no external load at the
hinge, there will be no change in the shear force value at
the point C. This is consistent with the constant slope of
the bending moment diagram across the hinge. We shall
not often need to draw out the shear force distribution to
obtain the primary information of our qualitative solution.
In future, therefore, we will concentrate upon the other
three diagrams, namely the support reactions, the bending
moment diagram and the deflected shape.
41

41 In this case we see that the whole of the beam is hogging

from A to C and sagging from C to E. We have indicated
this by placing a T on the tensile side of the beam.

42 The other factors which we know about the deflected shape 42

are firstly, that there can be no vertical deflection at the
supports A, B and E. Secondly, that there can be deflection
and a discontinuity of slope at the hinge point C. Thirdly,
the beam will deflect in the direction of the load W at point
D. This leads us to the deflected shape. Again we can note
a coherence between the deflected shape and the bending
moment diagram. In the lower diagram we have shown the
support reactions, which we have already found, so that
we then have a full three-part qualitative solution to our
problem.
16 Understanding Structural Analysis

43 43 Let us now look at a somewhat more complex problem

that introduces a number of new ideas. The structure to
be considered is a symmetrical portal frame with three
hinges at A, C and F as shown in this diagram. It is often
called a three-pinned portal. The portal is loaded by a
vertical force W placed asymmetrically at the point D. We
note that the corners of the frame at B and E provide rigid
connections between the horizontal and vertical members,
thus forces and moments are fully transferred across these
connections. On the other hand, the hinge at C behaves, as
before, transmitting forces but not a moment.

44 44 As before, we can use the principle of equilibrium of

moments on the whole structure, about any point. If we
choose the point F then it is clear that the anticlockwise
moment MW caused by the load W must be balanced by
an equal and opposite moment MR caused by the only
A
available reaction RA . At this stage we cannot define the
value or direction of the support reaction RA except to note
that, as it has a moment about the point F, it cannot lie on
the line AF.

45 45 However, if we now look at the equilibrium of the part of

the structure ABC we realise that the line of the reaction
RA must pass through the hinge point C since there can
be no resultant moment about this hinge and there are no
other external forces on this part of the structure.

46 46 It is usually more convenient to separate the reaction RA

into its two components: VA in the vertical direction and HA
in the horizontal direction.
 The Analysis of Statically Determinate Structures 17

47 We are now in a position to complete the identification 47

of the remaining support reactions; namely those at F.
If we take moments about the support point A, overall
equilibrium demands that there shall be an upward vertical
reaction VF at F to provide an anticlockwise moment to
balance the clockwise moment of the load W about the
support hinge A. Notice that the line of the horizontal force
HF passes through A and produces no moment about that
point.

48 We may now start to draw our bending moment diagram. 48

Clearly a horizontal force, such as HA acting at a distance
from a point on a vertical member, such as AB, produces
a bending moment in the same way as does a vertical
force on a horizontal member. We start by drawing the
bending moment diagram on the tensile side of the vertical
members. Since the horizontal forces HA and HF are
numerically equal and heights from A to B and from F to
E are equal then the bending moments caused at B and E
will be of equal value.

49 We have noted that the joints at B and E form rigid 49

connections between the members at these corner points.
As no external moments are applied here the bending
moments in the horizontal members at the corners must
have the same values as the bending moments in the
vertical members at these points to maintain moment
equilibrium.
18 Understanding Structural Analysis

50 50 Our next consideration is with the internal hinge at C. At

this point, as at all unloaded hinges, the bending moment
has to be zero. Furthermore, because there is no external
loading within the length BC the bending moment can only
change linearly from its value at B to a zero value at the
hinge C. Indeed, the slope of the diagram will continue
at a constant value until we reach the external load point
D. We saw a similar effect when we studied the case of a
beam with an internal hinge. You will note that this effect is
consistent with our known values of the support reactions.
The bending moment at B is clockwise balancing the
anticlockwise moment caused by the horizontal support
reaction HA. However, as we move from B towards C,
whilst the moment due to HA remains constant, the vertical
support reaction VA begins to have a moment effect in
the reverse direction. The effect of VA is to diminish the
moment due to HA as we move from B towards C until it
reduces to zero at the hinge point C.

51 51 The same reasoning may be applied to the right-hand

portion of our frame from the corner at E to the load point
D. At D, of course, the bending moments from the left
and from the right must come to the same value. We can
therefore complete our bending moment diagram for the
frame by drawing a straight line from the value at E to the
value at D.

52 52 We must now determine the form of the deflected shape

of the frame. We start by indicating, with the letter T, the
sides of the members which will be in tension and therefore
on the outer side of the curvature.
 The Analysis of Statically Determinate Structures 19

53 We note that the support points A and F are restrained 53

both horizontally and vertically but, as they are hinged, the
slopes of the members may change at these points. The
corners B and E may rotate and move horizontally but may
not deflect vertically. Also we observe that the hinge C
may deflect vertically but it will have the same horizontal
displacement as the corners B and E due to the assumed
axial rigidity of the horizontal member BCD. However, the
members connected at a hinge can take up different slopes
on either side.

54 Let us first look separately at the two halves of the frame; 54

the part ABC and the part CDEF. For the moment we will
assume that the corners B and E remain in their original
positions vertically above A and E. All bending on the left-
hand side ABC takes place with tension on the outside. We
can expect that part of the frame to deflect as we see on the
left-hand side diagram, B. On the right-hand portion CDEF
there is equal bending on the vertical member but the
outside tension changes rapidly to inside tension before we
reach D and continues to C. This side of the frame would
deform as shown. What we particularly notice is that the
downward deflection of C on the left-hand side is greater
than that of C on the right-hand side.
55

55 However, our frame is, of course, in reality continuous

and the two points C have to be co-incident since they are
shared by both the left and right parts of the frame. This
can be achieved if we allow the points B and E to move
horizontally, rotating the left part ABC in an anticlockwise
direction about the support A causing point C to rise and
move to the left. The amount of rotation will be the same
for each part of the frame and must be just sufficient to
make the two points C co-incident.
20 Understanding Structural Analysis

56 56 We can now reconnect the two parts of the frame and we

have the correct deflected shape of our un-symmetrically
loaded three-pinned portal frame. The horizontal deflection
at the top of the frame, which has been caused by a load
with a vertical component only, is called the sway of the
frame. It always arises in a frame-type structure when
either the loading is applied un-symmetrically or the
structure itself is un-symmetrical. It is specially to be
noted that the loaded point of this structure deflects both
in the direction of the load and at right angles to it, that is,
horizontally.

57 57 It is worth repeating that in this example we have

introduced a number of new ideas. Firstly, that bending
moments occur in vertical members due to horizontal forces
in a similar way to that in which vertical forces cause
bending moments in horizontal members. Secondly, that in
this simple frame analysis we are assuming that there are
no distortions of the members in the axial direction which
would place secondary restraints upon the deflections of
certain points in the structure. Thirdly, that a deflection
need not occur only in the direction of the loading. The full
three-part solution of the three pinned frame is shown.

58 58 Practice Problems
Remember that the solution consists of three parts; namely
the distribution of bending moments, the support reactions
and the deflected shape. You may not necessarily be able
to start with any one of the three parts and may have to
move from one component of the solution to another until
all three component parts are complete and are consistent
with each other. Test your qualitative solution against
the output from the QSE Plane program and explain any
differences.
2 Statical Indeterminacy

So far in your study of the analysis of structures, the range of structures has been
limited to those which may be solved by the application of the three equations of statical
equilibrium:

all horizontal forces must balance,

all vertical forces must balance,

all moments must balance.

Consequently for a solution to be found there can only be three unknowns.

However, virtually all real structures have more than three unknowns, they are ‘statically
indeterminate’ and cannot be solved by the three equations of equilibrium alone.

That condition is known as ‘indeterminacy’ and there are two ways of identifying the
degree to which a structure is indeterminate: statical and kinematical indeterminacy.
This chapter deals with the former. Kinematical indeterminacy is specifically related to a
particular method of analysis, Stiffness, which is discussed in Chapter 7.

Although this chapter will ultimately show how to find the number of times a structure is
statically indeterminate, this number or ‘degree’, is only of importance in the flexibility
method of analysis. The degree is the number of equations of compatibility which must be
identified to add to the three equations of equilibrium for a solution of all the unknowns.

However, as you will discover, it is the identification that the structure is or is not
indeterminate which is of primary importance, in other words, it is the qualitative
appreciation of this condition which should be uppermost in your mind. The calculation
of the degree of indeterminacy can only be safely carried out once this qualitative
understanding has been soundly established.
22 Understanding Structural Analysis

1 1 Definition of the state of statical indeterminacy

Virtually all real structures do not satisfy this criterion
and are known as statically indeterminate structures. The
primary objective of this text is the development of an
understanding of the relationship between determinate and
indeterminate structures and the solution, in particular, of
the latter.

2 2 To begin with we will define the axes. Most of our

structures and most real structural problems are ultimately
resolved into two dimensions set in the X – Z plane. All
reactions and forces will be referred to as:

H horizontal in the X direction

V vertical in the Z direction

M moment about the Y axis, set normal to the X – Z

plane.

Remember that all these forces must be in equilibrium at

any point on the structure.
3

3 These three equations of equilibrium are sufficient to

analyse statically determinate structures.

4 4 We will look first at a simply supported beam with a

diagonal load W. There are only three unknowns, the
reactions HA, VA and VB. The reaction in the horizontal
direction at B has been removed by putting restraint B on
rollers to allow horizontal movement. The reactions are
shown as:

The action of the joint on the member

Note that HA would be equal to the horizontal component of

the load.
 Statical Indeterminancy 23

5 This second example of a cantilever is also statically 5

determinate. It has the three reactions characteristic of
a built-in support. There are three unknowns, therefore
this structure may be solved by the application of the laws
of equilibrium. The direction of the wall moment reaction
is negative, anticlockwise, the action of the wall on the
member.

6 It is critical to an understanding of the qualitative analysis 6

of structures that the support conditions are properly
understood. This first is generally known as a pinned or
simple support. It allows rotation and has no moment
restraint. The support will provide horizontal and vertical
force reactions.

7 The roller release is a relatively rare occurrence in real 7

structures; however, partial releases are common and the
basic understanding of this full release will help you to
understand the partial release. This support condition is
capable of only one reaction, normal to the direction of
movement. Note particularly that this support will not lift
off the support as it is capable of a downward reaction as
well as an upward reaction.

8 The fully fixed support, sometimes called encastre or 8

‘built in’, will provide three external restraints. It is worth
emphasising here that these restraints are potential and
depend upon the loading.
24 Understanding Structural Analysis

9 9 It is possible to provide an internal release to the structure.

Some internal releases are difficult to visualise and are
introduced here to prepare you for their use later in the text
as analytical devices.

This is an internal moment release, a hinge. It is capable of

transferring axial and shear forces. The bending moment
is always zero at a hinge. The joint is not externally
restrained in position and both members are free to adopt
slope independently of one another.

10 10 The moment release, the hinge, is fairly common in

structures; however, the shear release and the axial
release shown here, are only likely to appear as part of a
theoretical, analytical procedure. In each case two stress
resultants are transmitted and one released.

11 11 Two-dimensional structures are almost always

indeterminate. The exception in its many forms is the
three-hinged arch. There are four external reactions,
one more than may be solved by the three equations of
equilibrium. However, the hinge provides another equation
of equilibrium and if moment equilibrium is considered at
B the unknown reactions HC and VC provide the fourth
equation of equilibrium.

12 12 A particular feature of the qualitative understanding of the

structural analysis and behaviour is the concept of the free
body diagram. Each element or member in a structure must
be in equilibrium.
 Statical Indeterminancy 25

13 It is convenient to separate the concept of indeterminacy 13

into external and internal indeterminacy and we will
look first at external indeterminacy. A third support has
been added to the simply supported beam. This means
that the beam cannot be solved by the three equations of
equilibrium. The structure is then said to be:

statically indeterminate.

The subsequent study of the analysis of structures will

provide us with the means to determine the value of that
fourth reaction, but for the time being in this chapter we
are concerned with the concept and later the degree of
indeterminacy.
14

14 Similarly, the cantilever may be made indeterminate, here

by the introduction of a vertical reaction at B turning it into
a propped cantilever.

15 The three-hinged arch has been fixed at the crown. 15

Although this diagram shows a load on the arch it should
be emphasised that the degree of indeterminacy is a
property of the structure and not of the loading.

16 Because any loading will cause the ribs of the arch to try 16
to spread, inducing a horizontal reaction, there will always
be four unknown reactions in a rigid arch, the vertical and
horizontal reactions at each support.
26 Understanding Structural Analysis

17 17 The particular importance of the condition of indeterminacy

is that the structural properties of an indeterminate
structure will affect the distribution of reactions. This is
not so with a statically determinate structure.

18 18 The simply supported beam, supporting a uniformly

distributed load will produce the same bending moment
diagram, regardless of the varying dimensions of the beam.
The distribution of forces and reactions in a statically
determinate structure are unaffected by the member
properties of the structure.

19 19 Now we can start to see the importance of indeterminacy.

The two-span beam shown has a greater structural depth at
the centre support. Imagine the beam at the support being
very deep, so deep that virtually no load was carried by the
end supports. The hogging moment at B would be greater
than the sagging moment in the spans, because BA and
BC would be acting almost as cantilevers.

20 If the stiffness is transferred to the ends of the beam

virtually no load would be carried by the centre support.
20 In these circumstances the hogging bending moment at
B would be small compared with the sagging moment in
spans AB and BC.

These simple examples illustrate the dilemma for the

structural designer. He must know the size of the members
before the analysis is carried out, yet the size of the
members is the object of the analysis. The answer is that
the designer must always make an intuitive, qualitative
analysis of the structure before the quantitative, numerical
analysis can be carried out.
 Statical Indeterminancy 27

21

21 Pin-jointed structures respond to the condition of

indeterminacy in the same way. The presence of additional
members, here members PQ and PR, will affect the
distribution of forces in the remaining members. If those
members have a large cross-sectional area then the force in
them will be correspondingly large.

22 To summarise then, indeterminacy may be the result of: 22

1. additional external reactions

2. internal rigidity

3. the addition of members

The last two are classified as internal indeterminacy.

23 These three terms are used in other texts. They have the 23
same meaning. The term indeterminacy will be used in
this text and the degree of indeterminacy referred to as ‘n’
times indeterminate.

24 It is important to appreciate that structures may be 24

converted back to a statically determinate condition from
the statically indeterminate, by a variety of strategies. This
technique is important because it models the decisions
made by the structural designer when he reduces the highly
complex, real structure to a simpler form for the purposes
of analysis. The portal frame is 1 times indeterminate
because it has four external reactions. The frame has
been released to a statically determinate structure by the
introduction of a hinge in the beam BC.
28 Understanding Structural Analysis

25 25 As an alternative means of reducing the structure to a

determinate form, a roller release has replaced the pin
joint at D. This particular release should be noted since it
is frequently employed in analytical procedures. Thus no
horizontal reaction can be taken at support D. Because
the horizontal forces must be in equilibrium, the horizontal
reaction at A will be equal to the horizontal component of
the load.

26 26 This last example of reducing the structure to a

determinate form is the axial release in the column. The
bending moment capacity at this release is unaffected.
Such releases are rarely employed either in practice or in
analytical procedures.

27 27 A moment reaction at A has been introduced to the portal

frame turning it into a fully fixed support, and the structure
is now 2 times indeterminate.

28 28 We may reduce this structure to a statically determinate

‘tree’ by releasing the two reactions at D, VD and HD. This
is a particularly important release system because it is the
easiest to recognise qualitatively.
 Statical Indeterminancy 29

29 The concept of a ‘tree’ as a statically determinate structure 29

is a particularly powerful one in the qualitative analysis of
structures because it is natural and easily recognised in
other structures.

30 The effect of a load on a tree is always that of a cantilever. 30

The effect of the load goes straight to the ground. For
this reason it is often preferable for us to convert our
indeterminate structures into an analytical ‘tree’. The
resulting load effects are the simpler to identify and
analyse.

31 We will now look at more complex frames, which represent 31

the complexity of structures met by the structural designer.
The two-bay, two-storey frame is fixed at A and J and
pinned at H. All other joints are rigidly connected.

32 This two-bay frame has eight external reactions at A, H, 32

and J. Therefore the external degree of indeterminacy is:

8 − 3 (equations of equilibrium)
= 5 times indeterminate.
30 Understanding Structural Analysis

33 33 To convert the whole structure into a ‘tree’ we must ‘cut’

the top frame BCDF. A cut will release three unknown
internal forces, VF, XF and MF. Thus the total degree of
indeterminacy is:

5 + 3 = 8.

34 34 If all the supports are pinned and the structure cannot

be reduced to a tree because of the absence of a fully
fixed support, the supports may be reduced to a statically
determinate condition. This two-bay frame is pinned at A,
G and H.

35 35 The structure may be reduced to a statically determinate

form by the removal of the vertical and horizontal
restraints at H and the horizontal restraint at G. The
sub-frame ABCEDG is then statically determinate, with
the determinate cantilever frame EFH fixed to it. The
structure is 3 times indeterminate because it is necessary
to release three reactions to reduce the structure to a
statically determinate form.

36 36 Two-storey frames may be solved in a similar way by

creating a series of statically determinate sub-frames. This
two-storey frame has a hinge at D and has pinned supports
at A and H.
 Statical Indeterminancy 31

37 The top frame BCDEF is reduced to a statically 37

determinate form by the removal of the moments at B
and F, the lower frame by the removal of the horizontal
restraint at H. The structure is therefore 3 times
indeterminate.

38 This next example of a two-storey frame, has an 38

indeterminate sub-frame ABEFG, supporting an
indeterminate sub-frame BCDE. We will now employ a
system of hinges to release the sub-structures to statically
determinate forms.

39 We will introduce hinges at B, C and E. Note that the 39

introduction of a hinge between the three members at B, is,
in fact, two separate hinge releases. Each of the sub-frames
is now a determinate three-hinged portal frame arid the
structure is 4 times indeterminate.

40 If there are less than three external restraints, no matter 40

how many internal restraints, the structure will fail. In this
example, with horizontal roller releases at both supports,
the structure will roll, i.e. this is a state of unstable
equilibrium. Beware of the sophistry of the notion that the
structure could be stable under an exactly vertical load.
Indeterminacy or stability is a property of the structure not
the loading arrangement.
32 Understanding Structural Analysis

41 41 The portal frame ABCD with pinned supports at A and

D also has hinge releases at B and C. Therefore there is
no restraint against sway and the structure will collapse
sideways.

242 42 By calculation this structure is determinate with six

external restraints and 3 hinge releases. However, the
beam would collapse if the structure were assumed to
behave according to the small deflection theory, i.e. when
deflections are small compared with the overall dimensions.
Thus in the context of elastic analysis, this frame is a
mechanism and cannot be solved by linear analysis. It is
known as a ‘pseudo -mechanism’.

Pinned structures

43 43 The pin-jointed frame is recognised as statically

determinate by the three members in a triangle, resulting
in the three unknown axial forces that can be solved by the
three equations of equilibrium.

44 44 The structure will remain determinate no matter how

many triangles are added. The system can be extended
indefinitely.
 Statical Indeterminancy 33

45 We will look first at the Warren girder truss, which is 45

statically determinate internally as it consists of a series
of triangular pin-jointed frames. The truss has statically
determinate external reactions.

46 If we now add an extra vertical reaction at the centre 46

then the structure is statically indeterminate, n = 1. To
emphasise the fact that this is an external indeterminacy
the actual frame may be ignored.

47 If we now introduce additional members BE and EH in the 47

truss the structure is 3 times indeterminate overall. It will
always be simpler to separate internal and external degrees
of indeterminacy, for pin-jointed frames.

48 Although this text is written to emphasise the qualitative

understanding of structural analysis, the actual quantitive
value is usefully examined in order to show how unreliable
it is unless accompanied by an intuitive grasp of behaviour.

For pin-jointed frames the relationship defining

determinacy is stated thus: 48

2 × j = m + 3,

where j = number of joints

m = number of members
3 = three equations of equilibrium.

The three-member frame is thus calculated to be

determinate.
34 Understanding Structural Analysis

49 49 If 2 × j is greater than m + 3 the structure is a

mechanism.

If 2 × j is less than m + 3 then the structure is

indeterminate.

50 50 Examine the structure shown. The calculation shows that

it is statically determinate, but it should be clear that
this structure will collapse because the first panel lacks
diagonal bracing. This demonstrates that the quantitative
solution must be judged qualitatively.

51 51 Three-dimensional indeterminacy may be dealt with in the

same way, although it should be appreciated that most
building structures are reduced to a series of linked two-
dimensional frames for ease of analysis.

There are six equations of equilibrium in three dimensions:

forces and moments relative to the three axes.

52 52 This cantilever ABCD is set in three dimensions. CD is

parallel to the Y axis. Six reactions are required for the
equilibrium of the frame at the ‘built-in’ support at A.
 Statical Indeterminancy 35

53 Consequently, the release of a fully fixed support frees six 53

reactions. This bent frame, fully fixed at A and D is 6 times
statically indeterminate. The ‘tree’ method of identifying
the degree of indeterminacy is even more effective with
three-dimensional structures.

54 The effect of the introduction of a pinned support at D, 54

previously fully fixed, is to release three stress resultants,
the moments about each axis. This frame is therefore 3
times indeterminate. The structure could be released to
a statically determinate ‘tree’ by the removal of the three
force reactions at D.

55 If we now return to the frame fully fixed at A and D 55

an internal hinge would release three internal stress
resultants, the bending moments about each axis. This
frame is now 3 times indeterminate.
36 Understanding Structural Analysis

56 56, 57, 58 Practice Problems

Determine the degree of indeterminacy of each of the
structures shown distinguishing between internal and
external conditions. Sketch the structure reduced to a
statically determinate form and identify the releases that
have been made.

57

58
3 The Qualitative Analysis of Beams

In the next two chapters, you will study an approach to the analysis of structures that is
likely to be unfamiliar to you. The traditional approach to the study of structural analysis
has been based, almost exclusively, on quantitative (i.e. numerical) methods. On one hand,
this is entirely logical. The structural designer does, eventually, need to put values to
loads and dimensions and determine the numerical value of reactions, bending moments
and so on.

There is, however, a critically important stage before that numerical analysis; the
preliminary analysis required to size the structural members. The nature of the analysis of
statically indeterminate structures is such that the designer must know the size of all the
structural members before an analysis is carried out. Consequently, it should be clear that
the detailed, numerical analysis of the structure, nowadays carried out by computer, is a
check on that preliminary analysis.

This places great importance on that stage of the design procedure. If the structure is
incorrectly sized, repeated analysis will need to be carried out. This is true regardless of
whether the analysis is being carried out by hand or by computer.

It is generally assumed that the skills required for this preliminary analysis are the
inevitable consequence of studying numerical techniques. The results of extensive
research into the development of an understanding of structural behaviour (See Appendix:
DMB papers) in undergraduates and trainee engineers suggest very strongly that this is
not the case.

That preliminary analysis requires a quite different set of skills that are referred to here
as ‘qualitative’. Essentially non-numerical, the basis of these skills is the recognition of
the relationship between the load and the resulting behaviour of the structure. In simple
structures, that is the relationship between the load, deflected shape and the resulting
reactions. These skills are distinctly different from quantitative skills; they rely on sets
of coherent diagrams rather than mathematical models, for example. However, the most
obvious difference between the qualitative and the quantitative approach to the analysis of
38 Understanding Structural Analysis

structures is that there is no obvious sequence for the steps in the qualitative approach.

Once a particular numerically based analytical technique has been learned, then the
sequence of the solution will almost always be the same. However, faced with a qualitative
analysis it is not apparent from which point the solution will emerge most effectively. It
may be best to start with the bending moment diagram in one case or the deflected shape
in another.

In many structural problems it is necessary to start with all parts of the full solution at
the same time, which is potentially very confusing. Students studying this approach say
that they understand each of the steps in the examples then, as soon as they try one of
the practice problems they are stuck. This suggests that the teaching method is, in some
way, deficient. This is not necessarily so. The explanation is that qualitative problems are
inherently difficult. This new approach requires a shift of attitude and this takes time and
practice.

However, you will discover as you begin to grasp this approach, that you find a
significantly increased confidence in your overall understanding of structural behaviour
that will contribute to your understanding of problems of structural design and the
methods of analysis studied in later chapters.

Perhaps the greatest value of this qualitative approach will only be apparent to you once
you start your office training. The use of computers is now universal. How can you check
that the output is correct? The checking system must be independent and robust. You will
discover that a sound understanding of structural behaviour will play a significant part in
the overall checking procedures that must be constructed to ensure the correct use of the
computer and the elimination of data and modelling errors.
 The Qualitative Analysis of Beams 39

1 We will start by restating the assumptions that underlie 1

the elastic analysis of structures. The first is that, for all
loading cases, the structure is within the elastic range of
material behaviour. Stress is directly proportional to strain
and deflections directly proportional to load.

2 The second assumption is that deformations due to the 2

loading do not create secondary bending moments. The
cantilever bent ABC is subject to two point loads W1 and
W2. Here the load W2 will cause bending in the cantilever
BC that will result in a horizontal deflection at B, ∆Bh. The
secondary moment W1 × AB is ignored. In the Engineer’s
h
Theory of Bending it is assumed that deformations due
to loading do not result in significant changes in the
structural geometry.

3 In the qualitative analysis of structures and most manual 3

numerical analysis, axial loading, which would result in
axial strain in the member, is ignored in the determination
of bending moments. The frame, ABC is loaded in the
vertical direction at B.

4 Although the load W would cause the column AB to 4

shorten, it is assumed in the qualitative analysis that there
is no secondary moment in beam BC. Generally, in the
elastic analysis of structures, deformations are considered
to be small compared with the dimensions of the structure
so that the equations of equilibrium of the deformed
structure are consistent with those of the un-deformed
structure. It should be noted however, that this deformation
AB is included in the computer analysis of structures.
v
40 Understanding Structural Analysis

5 5 It is important to restate the diagrammatic convention for

the support conditions and the resulting reactions. The
roller support has one reaction, normal to the direction of
movement. The pinned support has two force reactions and
the fully fixed support two force and one moment reaction.

6 6 The principle of superposition is assumed to apply as the

structures are within the elastic range. We will illustrate
this principle with this example. The load W on the two-
span beam ABC produces a hogging moment over support
B with a downward reaction VC, to balance this bending
moment at C.

7 7 If we remove the vertical reaction at C, the beam becomes

a simply supported beam with cantilever. If we now apply
the reaction at C, VC as a load, the combination of these
two conditions will produce the solution to the original two-
span beam. This simple example illustrates one of the most
powerful design tools in the armoury of the designer, since
it allows the reduction of complex structures into simpler
forms and the examination of the load effects separately.

8 8 The deflections may be treated in the same way. The

vertical deflection at C due to the load W must be equal to
the downward deflection at C due to the force reaction VC
since the real deflection at C is zero.
 The Qualitative Analysis of Beams 41

9 This is the first example of the qualitative analysis of 9

structures, the simply supported beam ABC, with a
cantilever CD. The full solution of a qualitative analysis
must always consist of:

the deflected shape

reactions

bending moment diagram

You must become accustomed to the idea that they are of

equal importance and your qualitative analysis will not be
correct until each of them has been determined and found
to check against the other two.

In the qualitative analysis of beams, the best place to

start is likely to be the deflected shape. We will start the
deflected shape by identifying the points through which we
are certain that the deflected beam must pass, A and C and
a downwards deflection beneath the load.

10 We can now draw the smooth elastic curve through these 10

three points. Note that CD is straight. The curve must be
smooth and with practice a very accurate deflected shape
may be drawn. Try practising the identification of the
deflected shape with a flexible plastic ruler.

11 This simple example of the bending moment diagram 11

illustrates the need to distinguish between the structure,
which is drawn as the base line of the bending moment
diagram, and the diagram itself. The point here is that the
value of the bending moment over the portion of the span
CD is zero.

N.B. The bending moment diagram is always drawn on the

bending tension side of the structure.
42 Understanding Structural Analysis

12 12 There are certain simple rules that should be borne in mind

when carrying out the qualitative analysis. Experience
suggests that because this approach is unfamiliar, students
tend to become confused, reach a solution containing
obvious incompatibilities and are not able to spot them.
Having found a solution consisting of the deflected shape,
reactions and bending moment diagram you should take a
cool look at each part and see how it relates to the other.
Is there a bending moment at the pinned support? Unlikely.
Is the bending moment at an internal support zero? Very
unlikely. Does the tension identified by the deflected shape
agree with the bending moment diagram drawn on the
tension side of the structure? We will call them golden rules.
13
Here is the first.

13 And here is another.

Bending moments resulting from point loads are linear

expressions, therefore the bending moment diagram
consists of straight lines. In addition the diagram can only
change direction at a load or reaction.

14 14 And another.

That is, where the hinge occurs at the end of a structure,

or at an internal hinge which we will study later in this
chapter.

15 15 We will now carry out a qualitative analysis on the propped

cantilever AB. We will apply the principle of superposition
to examine the effect of the fully fixed support at A, by the
notional removal of the moment reaction at A, MA .
 The Qualitative Analysis of Beams 43

16 As a simply supported beam there will be a clockwise 16

rotation at A. The propped cantilever however, will have a
zero rotation at A and to return to our original structure we
must apply an anticlockwise moment at A.

17 It is now apparent that there will be a change of curvature 17

in the deflected shape, which is known as the point of
contraflexure. This provides another point of reference
between the deflected shape and the bending moment
diagram because the bending moment is zero at this point.
The member is unstressed at the change of curvature.

18 We can now attempt to draw the bending moment diagram. 18

We know that there will be a hogging bending moment
at the fixed support at A. This is recognised by the
anticlockwise reaction and the identification of tension
on the top of the beam from the deflected shape. At the
other end of the beam the upward reaction VB will cause a
bending moment on the underside of the beam. The bending
moment diagram will be a straight line from B until it
meets the line of action of the point load. The diagram is
completed with a straight line between A and the load.
This line must pass through the point of contraflexure.

19 Another golden rule. 19

44 Understanding Structural Analysis

20 20 The next example is the two-span beam ABC. The

deflected shape must pass through the supports and the
downward deflection under the load. Note that a structure
will always deflect in the direction of the load provided that
it is the only load on the structure.

21 21 The smooth curve shows the change of curvature,

characteristic of all multi-span beams. The direction of the
vertical reaction at C may be identified by imagining that
the reaction is removed. The beam would deflect upwards
and the direction of the reaction necessary to bring the
beam back to the correct position would be downwards.

22 22 Both VA and VB act upwards and the first lines of

the bending moment diagram may be drawn from a
consideration of the effect of the upward reaction at A
producing tension on the underside of span AB and the
downward reaction at C, producing tension on the top of
span BC.

23 23 The bending moment diagram is completed with a straight

line between the load and point B. The coincidence of the
zero moment and the point of contraflexure may now be
recognised and checked.
 The Qualitative Analysis of Beams 45

24 The analysis of a three-span beam would proceed in a 24

similar way.

25 The first step is to identify the points of certainty on the 25

deflected shape. The beam must pass over each of the
supports and deflect below the load. A smooth curve may
then be drawn between these points.

26 The direction of the vertical reactions is identified by their 26

notional removal. The reaction at D must act upwards and
the reaction at C downwards.

27 The first, most obvious, points on the bending moment 27

diagram are usually those at each end of the beam. Here
the upward reaction at A and D both produce bending
tension on the underside of spans AB and CD respectively.
46 Understanding Structural Analysis

28 28 To complete the bending moment diagram you must be

able to identify the fact that the beam is hogging over the
support at B.

This beam has two points of contraflexure, in span BC and

between point B and the point load. Note the identification
of areas of bending tension on the deflected shape and the
visual check against the bending moment diagram.

29 29 This next example is a three-span beam with a fixed

support at A loaded on the central span with a point load
W. Note that the beam must enter the fixed support with
zero deformation compared with the unloaded structure.
Failure to draw this clearly is a common source of error in
drawing the deflected shape.

30 30 Because the bending moment diagram is drawn on the

tension side of the structure, a check should always be
made between the deflected shape and the bending moment
diagram. To facilitate this check, identify the areas of
bending tension (T). The reactions are determined from the
deflected shape by their notional removal.

31 The most obvious lines of the bending moment diagram

31 may now be drawn: the sagging moment at A, the hogging
moments at B and C and the sagging bending moment
below the load. Note that it is known that the line of the
bending moment diagram from A will move to cross the
baseline of the structure because the vertical reaction
VA is causing anticlockwise moments which will reduce
the clockwise moment MA that is moving along the beam
from A to B. The bending moment diagram may then be
completed with straight lines as there are no other loads
between these identified points on the bending moment
diagram.
 The Qualitative Analysis of Beams 47

32 Practice problems 32
Always complete the full solution by drawing the three
parts, however confident you feel about any one of them.
This is particularly true of the direction of the reactions.
There is a temptation to miss this step and attempt to
construct the bending moment diagram from the deflected
shape.

Experience shows this to be most unreliable and the

intermediate step of producing the diagram of the reactions
is usually critical in ensuring coherence between each of
the three diagrams:

deflected shape,

reactions,

bending moment diagrams.

33 Externally applied moments are rare in real structures. 33

This example is introduced here to prepare the student for
the use of external moments in the analytical procedures to
follow. You may find them difficult to visualise and this is
probably because they are unfamiliar.

The moment applied to the end of a cantilever will cause a

constant bending moment and is one of the occasions when
there is a bending moment at the free end of a structure.

34 The deflected shape is derived from the bending moment 34

diagram that shows that the deflected shape must be of
single curvature, and since it is the top fibres which are
extending in tension, the deflection must be downwards.

There is a balancing anticlockwise reaction moment, MA.

Note that there is no vertical reaction as there are no loads
to balance a reaction at A.
48 Understanding Structural Analysis

35 35 When the load moment is applied in the span of the simply

supported beam AB, the first point to recognise is that
the beam must rotate in the direction of the moment. Note
that the beam will almost always deflect away, vertically
or horizontally, from the un-deflected position at the point
of application of the moment. The points of certainty on
the deflected shape are the zero deflection at the supports
at A and B and the rotation of the beam at the point of
application of the load moment.

36 36 Because the vertical equilibrium is only maintained by the

reactions, the vertical reactions are equal and opposite
and their value is the applied moment divided by the span.
Taking moments at A, the applied moment M produces a
clockwise moment on the structure. The position of the
moment, relative to A, has no effect on the value of the
external moment on the structure. This clockwise moment
must be balanced by the anticlockwise moment produced by
VB.

37 37 The bending moment diagram resulting from the

application of an external moment is unusual in that there
is a discontinuity at the point of application of the moment.
The moment is on both sides of the structure at the same
point. The sum of the hogging and sagging moments at the
point of application of the external moment is equal to the
value of the external moment. The slopes of the lines of the
bending moment diagram are equal, because they result
from the same value of the vertical reaction.

38 38 We will now apply the moment to the propped cantilever

beam AB. Note the tension on the underside of the beam at
the fixed support at A.
 The Qualitative Analysis of Beams 49

39 Here are the resulting diagrams of the reactions and 39

bending moments. Note that the slope of the bending
moment diagram either side of the point of application of
the moment is the same, resulting from the constant shear
force across the beam.

40 We will now apply the moment to a three-span beam, 40

ABCD. Try to draw this deflected shape. You will find
it difficult to recognise the anticlockwise rotation at B.
However, if you ensure that the curve is smooth through
the points of certainty, that rotation will be apparent.

41 The direction of reactions may be found from the deflected 41

shape. Perhaps the most difficult part of this solution is the
recognition of tension on the underside of the beam at B.
Try treating BC as a beam, ‘built-in’ at B and C. This will
reveal more clearly the tension on the underside at B.

42 The plotting of the bending moment diagram from a partial 42

uniformly distributed load will now be explained. The
reaction at A will be greater than that at B.
50 Understanding Structural Analysis

43 43 The characteristic parabola of the bending moment diagram

resulting from a uniformly distributed load (UDL) is
plotted on a new baseline between the plotted values of
the bending moment at the start and finish of the uniformly
distributed load. The maximum height of the parabola is
W × l² ⁄ 8 plotted normal to the structure baseline where
l is the length of the uniformly distributed load. The
parabolic curve may now be drawn to complete the bending
moment diagram.

44 44 If the diagram is drawn to a large scale there are a number

of additional geometrical properties that will aid the
accurate drawing of the parabola.

1 and 2. Note that the tangents to the start and finish

of the parabola are the slopes of the bending moment
diagram from the reactions to the UDL. There will be no
discontinuity in the diagram unless there is a point load at
the end of the UDL.

3 and 4. Height at ¼ span = ¾ moment at mid-span of

UDL.

45 45 The tangent to the curve for the maximum value of the

bending moment in the beam is parallel to the structure and
will not coincide with the maximum ordinate for the UDL
unless the load is symmetrically placed on the beam.

46 46 One obvious limitation of the qualitative approach is

that there may be more than one qualitative solution for
structures with more than one load. The simply supported
beam AB is extended by a cantilever from B to C. The
beam is only partially loaded with a uniformly distributed
load; the cantilever is fully loaded. There are two possible
solutions. In this first case it is assumed that the load on
the beam is sufficient to create an upward reaction at A.
 The Qualitative Analysis of Beams 51

47 Consequently there will be a sagging bending moment 47

from A to the start of the uniformly distributed load. Note
that the plotting of the parabolic distribution of bending
moments due to the uniformly distributed load may be
applied to the cantilever.

48 Another golden rule. 48

49 If we now increase the span of the cantilever to induce a 49

downward reaction at A, there will be a single curvature in
the deflected shape, because the effect of the load on span
AB is insufficient to counteract the effect of the hogging
moment from the cantilever.

50 The bending moment diagram will be hogging over the span 50

AB, despite the sagging effect of the partial UDL.
52 Understanding Structural Analysis

51 51 This next example of a propped cantilever beam has a hinge

in the span at B. Remember that one of the most important
features of a hinge is that the bending moment is always
zero.

Because the support at A is simple, allowing full rotation,

as the load causes the beam to deflect downwards, AB will
curve and rotate. The shear induced at C will turn BC into
a cantilever. There will be discontinuity in the deflected
shape at the hinge.

52 52 This diagram shows the reactions and bending moment

diagram. Note that the resulting distribution of bending
moments is similar to that for a propped cantilever without
a hinge. The explanation is that in this example, the hinge
has been positioned at the point of contraflexure.

53 53 We may inspect the equilibrium of each part of the beam.

The equilibrium of part AB is sustained by the upward
vertical shear force SB. This must be balanced by the
downward shear force on BC. It is this shear force that
creates the hogging, clockwise moment reaction at C.

54 54 Our next example is the three-span beam ABCDE, with

a hinge in span CE at D. The point load W is to the left of
the hinge in span CE. As CD deflects downwards under
the load, DE will rotate without bending. Consequently,
there is no shear force across the hinge. The beam will be
un-deformed from the load to the hinge.
 The Qualitative Analysis of Beams 53

55 This diagram shows the reactions and bending moment 55

diagram.

56 This next beam has two spans and is fully fixed at the 56
support A. There is a hinge at B and a point load in span
CD. We can draw the points of certainty on the deflected
shape. Note that the beam will have zero rotation at A.

57 It is sometimes helpful, when there is a hinge in a 57

structure, to imagine it replaced by a short member. The
force in the member, in this case compression, helps to
clarify the resulting bending either side of the hinge. In
this case, AB will act as a cantilever and BC as a beam
supported at B.

58 This diagram shows the reactions and bending moment 58

diagram. Note that the final solution for the bending
moment diagram is similar to that for a beam without a
hinge. In this example the hinge has been placed at or near
to the natural point of contraflexure.
54 Understanding Structural Analysis

59 59 This next example is almost impossible to visualise. It is

the application of equal and opposite external moments
either side of the hinge. This is a device used later in one of
the analytical methods.

This deflected shape is difficult to arrive at without an

attempt at the reactions and bending moment diagram.

60 60 This diagram shows the reactions and bending moment

diagram. In this example, because of the curious loading,
we appear to have broken one of the golden rules. In fact
the moment is still zero at the hinge, but has a value, the
imposed moments, either side. The reaction at A may be
determined by considering the equilibrium of AB. That
downward reaction will cause a hogging moment from A
to C. Taking moments about C of span AC, the applied
moments cancel out.

61 61 We will now study another example of alternative solutions.

The cantilever ABC has two point loads. Here the
downward load at C is greater than the upward load at B.

62 62 This results in an anti-clockwise moment at A and a single

direction of curvature. The dotted lines on the bending
moment diagram indicate the ‘negative’ effect of the
upward load at B.
 The Qualitative Analysis of Beams 55

63 We will now examine the effect of making the load at B 63

greater than the load at C. There is now a change in the
curvature in the deflected shape.

64 This diagram shows the reactions and bending moment 64

diagram. There is a point of contraflexure between A and B
and a clockwise moment reaction at A.

All real structures have a complex loading arrangement.

However, with a sound understanding of the qualitative
method, you will learn to understand how to model real
loadings and the response of real structures by reducing
the complex structures to a series of simpler and sub-
structures.

65 Practice Problems 65
Produce the full three-part solution for each of these
structures.

Test your qualitative solution against the output from the

QSE Plane program and explain any differences.
56 Understanding Structural Analysis
4 The Qualitative Analysis of Frames

The introduction of the second dimension in the plane, the vertical direction, required
for two-dimensional structures, introduces the complication of horizontal loads or forces
acting normal to vertical members. The effect of loads or forces normal to members will
always create internal bending moments. The fixity at the joints will allow the transfer of
bending moments in vertical members into horizontal members. You will find it useful to
refer back to the last part of Chapter 1 which studied the distribution of bending moments
in statically determinate two-dimensional frames. Often the solution to a statically
indeterminate frame is revealed if you try to see the effect of the load on the determinate
frames within the indeterminate system. Thus the qualitative approach to determinate
frames may be seen to link with the analysis of the degree of indeterminacy.

When you reach the end of this chapter and attempt to solve the practice problems you
will find them difficult at first. There are a number of reasons for this. The first and most
important aspect of this approach to understanding is that solving problems qualitatively
is difficult in itself. Secondly, there is no immediate route into the solution. The problems
seem open-ended. In fact they are internally coherent and that is the clue to the solution.
You must check all the loads and forces against the effects – apply the golden rules
and always produce the full three-part solution of bending moment, deflected shape and
reactions. If your qualitative solution is correct then each of these parts to the solution
will check against the other two.
58 Understanding Structural Analysis

1 1 We will start by examining the simplest two-dimensional

structure, the cantilever bent ABCDE, fully fixed at A.

2 2 The bent is loaded with two point loads, W1 vertically at B

and W2 horizontally at D. Cantilever structures are always
statically determinate and it is often simpler to start the
three-part qualitative solution by drawing the bending
moment diagram and using it to help to draw the deflected
shape. Note that the bending moment is constant in that
part of the structure parallel to the load W2, over the
member BC.

3 3 We know that the beam ABC will deflect downward under

load W1 and that load W2 will create a tension in the top
surface of BC. From this we are able to recognise that
joint C will deflect downwards. Note that when drawing
the deflected shape in two-dimensional structures the
real shortening or extension of members under axial
load must not be represented in the diagram otherwise a
misleading deflected shape could result. Such strains are
insignificantly small compared with the bending deflections
in the qualitative analysis. Consequently the deflected
position of joint C is drawn on the line CE of the original
structure.

4 4 Another golden rule.

Virtually all structures with a fixed support are being
deformed by the loading system just before the support.
However, you must identify the zero rotation at the support
as one of the points of certainty on the deflected shape.
 The Qualitative Analysis of Frames 59

5 The final part of the solution is the diagram of the 5

reactions, drawn, as usual, as the action of the joint on the
member. Thus MA is anticlockwise. If you have difficulty
visualising the support conditions, try re-creating the
support conditions and the structure with a piece of flexible
plastic.

6 This is a reminder that the full solution must always

include:
6
1. the deflected shape

2. bending moment diagram

3. reactions

You should start to identify the internal relationship

between each part of the full solution; bending tension and
deflected shape, moment reactions and bending moment
diagram for example.

7 There is a plotting routine for the bending moment at a 7

joint or node that is particularly helpful. We know from the
deflected shape that there will be a tension on the top of
the member at A and from the reaction diagram, that the
reaction moment MA is anticlockwise. Remember that we
have adopted an overall sign convention where clockwise
moments are positive.

8 Follow this routine very carefully: 8

Go from the joint under consideration along the member

and rotate in the direction of the sign of the moment. The
value of the bending moment is plotted on the ordinate
normal to the member. Fortunately this is the same output
convention used by (almost!) all computer programs for the
analysis of plane frames.
60 Understanding Structural Analysis

9 9 This next example is the same cantilever with an imposed

anticlockwise load moment at D.

10 10 The bending moment resulting from the application of a

moment on a cantilever is a constant value throughout the
structure regardless of the shape of the structure. In this
example, the moment produces bending tension on the
underside of the cantilever. Once again, if in doubt, try re-
creating the structure with a piece of plastic.

11 11 The deflected shapes that result from the application of

moments are difficult to produce intuitively. Consequently
it is the relationship between the deflected shape and the
bending moment diagram and the convention of drawing
the bending moment on the tension side of the structure to
which we must turn. Imagine the fibres on the underside
of the frame from A to D extending and it will be apparent
that point C will deflect upwards.

12 12 There is only one reaction at A, the clockwise moment

balancing the anticlockwise load moment. Note particularly
that there are no load reactions.
 The Qualitative Analysis of Frames 61

13 We will now reverse the two-member frame and extend it 13

by a horizontal member CD, hinged at the junction C and
supported on a horizontal roller support at D.

14 We have generally ignored the significance of axial stress 14

resultants in the members. However, it is helpful here to
note that there is an internal compression in member BC.

15 To begin with we identify the points of certainty on the

deflected shape:

1. The member must enter the fully fixed support at A 15

without rotation from its original, unstrained position

2. Point B will be pushed downwards by the

compression in BC

3. Because the right-angle will be preserved at B, point

C will deflect horizontally,

4. the horizontal deflection of C will roll support D to

the right.

16 We can now complete the deflected shape. Member BC 16

will remain straight because the structural configuration
eliminates bending moments in that member and there is
no horizontal reaction at D.
62 Understanding Structural Analysis

17 17 This figure shows the final bending moment diagram. The

slope of the bending moment diagram from C to the line of
the load will be the same as the slope from B to A because
it is the compression in member BC which is creating these
bending moments.

18 18 Our original cantilever bent is now converted into a frame

with two supports, with the addition of a roller support at
C. We will examine the effect of sway by subjecting the
frame to a horizontal load at B.

19 19 To identify the direction of the reaction at C we will

notionally remove it. Point C would deflect downwards as
a result of the horizontal point load at B and the resulting
reaction at C would be upwards.

20 We can now draw the deflected shape, being careful to

maintain the right-angle at B. The reactions are derived as
20 follows:

1. The horizontal reaction HA will balance the applied

load

2. The vertical reaction, VA will be equal and opposite to

the reaction VC

3. There will be an anticlockwise moment at A, MA, the

sign of which is confirmed by the tension on the left-
hand face of the column AB, at A.
 The Qualitative Analysis of Frames 63

21 The bending moment diagram may be derived from the 21

reactions and confirmed with the deflected shape. The
upward reaction at C will cause sagging moments in the
beam BC. We know that there is a tension on the left-hand
side of the column at A, which completes the bending
moment diagram. Note particularly the confirmation of the
point of contra-flexure in member AB.

22 If we change the sign of the load by reversing the direction 22

then all the stress resultants and deformations will have
opposite signs.

23 As with the deflected shape above so the reactions and 23

bending moment diagram are reversed. Thus we have the
effect of sway, either to the right or the left.

24 We will now remove the horizontal load and apply a vertical 24

load on member BC. Which way then, will the support at C
deflect? To the right or the left?
64 Understanding Structural Analysis

25 25 There are a number of ways in which we can examine

the horizontal deflection at C, i.e. sway. This particular
approach of introducing an artificial reaction at C to
prevent horizontal movement, is the same strategy we will
employ later in the study of one of the major analytical
methods, Moment Distribution.

26 26 In drawing the deflected shape, joint B will not move

horizontally but will rotate, because of the artificial
horizontal restraint at C. If we imagine a notional removal
of the horizontal restraint at A (not shown) then it is
apparent that the restraining horizontal reaction HA will be
positive, to the right. This must be balanced by the artificial
reaction at C, HC , i.e. to the left.

27 27 To remove the effect of the artificial horizontal restraint

at C, we apply an equal and opposite force to the
original structure. This is an example of the principle
of superposition. Thus the combination of the reaction
diagrams for the artificially restrained structure and the
application of the value of the artificial restraint at C as a
load, will produce the real, original structural conditions of
a roller support at C.

28 28 This is the final bending moment diagram, the combination

of restrained and ‘sway’ moment diagrams. Note that the
application of the principle of superposition, that is the
combination of these diagrams, cannot predict the constant
bending moment in member AB. It is only because we
recognise the absence of the horizontal reaction at A.
The slope in the bending moment diagram in the columns
of frames is always due to the presence of a horizontal
reaction.
 The Qualitative Analysis of Frames 65

29 The frame will deflect horizontally to the right, the effect of 29

the ‘load’ HC, balancing the artificial reaction. Notice that
there is single curvature in the column AB. For there to be
a change in curvature in the column of a frame there must
be a horizontal reaction to create the zero moment and the
consequent point of contra-flexure. The recognition of the
effect of the horizontal reaction is particularly important as
it is a frequent source of errors in the qualitative analysis
of frames.

30 We will now solve the pin-supported portal frame ABCD. 30

To determine the direction of the horizontal reaction HD,
we will notionally remove it. Support D will roll to the right
because the beam BC will bend and as the right-angle at B
must be maintained, B must deflect to the right. This alone
would induce a horizontal deflection at D. However, the
right-angle at C must also be maintained which will rotate
CD, increasing the horizontal deflection at D.

31 To return support D to its original position the horizontal 31

reaction at D must be negative, i.e. to the left. To satisfy
horizontal equilibrium, the horizontal reaction at A must
be equal to HD. Because HA and HD are equal, and the
columns AB and CD are the same height, the bending
moments at B and C will be equal.

32 Because of the later application to the method of Moment 32

Distribution, and because it is a good example of the
principle of superposition, we will examine the problem of
sway with the introduction of an artificial restraint in the
horizontal direction at C. This qualitative solution depends
upon the relative values of the bending moments induced
at B and C for the artificially restrained frame. You must
accept at this stage that the fixed-end moment at B is
greater than the moment at C, for the case of the point
load closer to B. This is analogous to the distribution of
moments in a fixed-end beam.
66 Understanding Structural Analysis

33 33 This will result in a greater bending moment at B than at

C for the restrained frame. Consequently the horizontal
reaction at A is greater than the horizontal reaction at
D. To balance the system in the horizontal direction, the
artificial reaction, HC must be negative, to the left.

34 34 To remove the artificial reaction HC we must apply an

equal force in the opposite direction. We now have the two
basic sets of diagrams for the application of the principle
of superposition to find the final distribution of bending
moments.

35 35 The final bending moment diagram for the original,

unrestrained frame is the combination of the restrained
and sway solutions. As with the previous example, the
combination of the qualitative bending moment diagrams
will not identify the equality of the bending moments at
B and C. This can only be recognised from the horizontal
equilibrium of HA and HD.

36 36 We will now examine the qualitative solution of a structure

with three members at a joint. Firstly, we will identify the
points of certainty on the deflected shape;

1. Vertical restraint at A

2. Rotation of joint B with the preservation of the right-

angles between the members at joint B

3. Must pass through the supports at C and D

 The Qualitative Analysis of Frames 67

37 The completed deflected form reveals that there will be 37

tension on the top surface of AB and bending tension on
the left-hand side of the column BD. The direction of the
reactions shown in this figure is determined by the notional
‘removal and replacement’ technique described above.

38 The bending moment diagram can now be plotted, starting

with the obvious points:

1. Hogging bending in AB resulting from the downward 38

reaction at A

2. The upward reaction at C will produce sagging in

span BC

3. The horizontal reaction at D, will produce tension on

the left-hand side of the column BD

Note the discontinuity of the bending moments in member

ABC.

39 Let us examine the moments at joint B. These have been 39

drawn here as the action of the joint on the member. Thus
the joint will exert a clockwise reaction on member BA.
Note particularly the relationship between the bending
moment diagram and the direction of the moments shown
here. They are related via the plotting routine described in
Figure 8. Thus the sum of bending moments MBA and MBD
must equal MBC since the joint must be in overall moment
equilibrium.
68 Understanding Structural Analysis

40 40 Remember that bending moments are plotted:

1. From the reference point

2. Along the member

3. Rotating in the direction of the sign of the bending

moment

4. Plotting on the ordinate drawn normal to the member.

41 41 We will now study the qualitative analysis of the frame

ABCDE. The point load W is located horizontally in
column AB. Support A is fully fixed, E pinned and D is on
a horizontal roller.

42 42 We may apply the principle of superposition to the point

load by assuming that it is first applied to a simply
supported ‘beam’, which in turn transfers the support
‘reactions’ to A and B.

43 43 We will start the solution by identifying the points of

certainty on the deflected shape. With practice, and
the study of the deflected shapes of two-dimensional
structures, the deflected shape shown may be drawn.
 The Qualitative Analysis of Frames 69

44 The direction of the reactions is identified from the 44

deflected shape. We can start to draw the bending moment
diagram with the most obvious points:

1. Bending tension on the left of the fixed support at A,

reduced in value by the horizontal reaction, HA

2. Hogging bending in member CD

3. Bending tension on the right-hand face of the column

EC.

45 The bending moment diagram is completed and compared 45

with the reactions. Note that the bending moment at C in
member CB, is equal to the sum of the bending moments in
CD and CE at C.

46 We can now add the effect of the point load in column AB. 46
This ‘simply supported’ effect is ‘added’ to the moment at
A and B.

47 Note particularly that the slope in the bending moment 47

diagram in column AB is related to the horizontal reaction
at A. Similarly, the change in slope in BC is related to the
downward reaction at A.
70 Understanding Structural Analysis

48 48 This golden rule will remind you to carry out this very
important check on the qualitative solution.

49 49 Hinges in frames may produce unusual solutions. This

orthogonal frame ABCD has a hinge at C. The structure
sways to the right to release the moment at B. The
solution is unusual in that there is a zero bending moment
at the internal, rigidly jointed connection at B. The
horizontal reaction at D is zero because moments about C,
considering member CD, must be zero. Thus the horizontal
reaction at A is zero and the bending moment at B, due
only to the horizontal reaction at A is zero.

50 50 We will now study the frame ABCDE, loaded with a point

load at D exactly over the reaction at E. There is a hinge at
B.

51 51 This structure has an unusual equilibrium in that the

vertical reaction at A is zero. Taking moments about E,
there is no out-of-balance moment due to the load since
the line of action passes through E. Also the horizontal
reaction at A is zero because of the hinge at B. Since there
is neither horizontal nor vertical reaction at A there are
no bending moments in either AB or BC. The bending
moment due to the load W in member CD, is resisted only
by member CE.
 The Qualitative Analysis of Frames 71

52 But what happens to the deformation in BC? We can 52

see that joint C must rotate, which appears to require a
downwards reaction at A, through a shear force at B, for
equilibrium.

53 The explanation is that the structure sways to the left and 53

joint C sways and moves downwards as member CE sways
about E. This allows CD and CE to bend, but leaves BC
and AB straight.

54 Clearly there is a compressive force in member CE. But 54

there is only one vertical reaction at E. We need to inspect
the force equilibrium of this support.

55 The explanation is that there is an internal shear force 55

in CE. It is shown here as the action of the joint on the
member.
72 Understanding Structural Analysis

56 56 The resultant of the external reaction to the shear and axial

force is equal to the vertical reaction at A.

57 57,58 Practice Problems

Produce the full three-part solution to each of these
problems. Test the qualitative solution against the output
from QSE Plane supplied with this text. If there is a
difference, then you need to be able to explain it.

58
Part II
5 The Theorems of Virtual Work

The key to the solution of statically indeterminate structures is the ability to determine
structural deformations. In one method of analysis, the deformations provide the basis of
equations of compatibility that, in addition to the three equations of statical equilibrium,
allow the solution of the unknown effects and the full solution to the distribution of
internal forces and moments upon which the subsequent structural design is based.

In addition however, the deflection of a structure may well be a design criterion. All
structures must satisfy two basic states of loading:

1. Serviceability when the structure is subject to its working load. It is at this state
that the deflection of the structure is checked.

2. Ultimate load, where the failure strength of a structure is compared with the
serviceability load multiplied by a load factor.

There are two basic approaches to the analysis of structural deformations, strain energy
and virtual work. The latter has the advantage of being able to deal with conditions other
than those within the elastic range - a limitation of strain energy. Only virtual work will be
studied here as it is generally agreed to be the more powerful of the two.

The student should bear in mind that although virtual work calculations may be the key
to the numerical solution of a problem in structural analysis, the mathematics content is
relatively trivial. The difficulty encountered is the application of the concepts of virtual
work to the problem in hand. Once that can be understood qualitatively, the subsequent
calculations although time-consuming, are straightforward. Without this qualitative grasp,
methods like virtual work may quickly become an exercise in mathematical manipulation
and the overall sense of the method and the behaviour of the structure, lost.
76 Understanding Structural Analysis

1 1 In a real structure two conditions must always be satisfied:

1. The system of forces acting on the structure and the

internal forces must be in equilibrium.

2. At the same time the structural deformations must be

compatible with the system of supports.

Our first example is a propped cantilever ABC, loaded with

a point load at B.

2 2 This figure shows the load and reaction equilibrium.

The built-in support at A will produce an anticlockwise
moment MA and a vertical reaction upwards, VA. There
will be a vertical reaction upwards at C. Since there are no
horizontal forces on the structure, the horizontal reaction
at A is zero.

3 3 The deflected shape of the structure must be compatible

with the support conditions. Since support A is built-in, the
deflection and rotation at that point in the deflected shape
will both be zero. In addition, there will be a zero vertical
deflection at C.

4 In order to solve certain problems, particularly those of

4 deformations, we will introduce the concept of virtual
states. These virtual states need only satisfy one of the two
conditions:

either equilibrium or compatibility.

We can have a virtual force state in which equilibrium

of forces exists, as shown here. The load is supported
by the moment and the vertical reactions at A only, but
the deflection at C is no longer compatible with the rigid
support at that point.
 The Theorems of Virtual Work 77

5 Similarly, we can create a state of virtual displacement by 5

the introduction of a hinge at A and B. This arrangement is
compatible with the support conditions, but the structure is
no longer capable of supporting the load. The introduction
of two hinges has created a mechanism. Consequently this
system satisfies compatibility but not equilibrium.

The virtual state is an analytical device. Because only

one state needs to be satisfied it is more easily evaluated.
When we combine real and virtual states we can develop
theorems of great power and application in the solution of
structural problems.

6 This chapter is concerned with the proof of the theorems of 6

virtual work. The two forms of the theorem discussed above
are of particular interest:

Virtual Displacements, and

Virtual Forces.

7 We will use as a basis for the development of the theorems 7

the deformable, pinned structure ABCD that is subject to
two loads, W1 and W2 acting at node D, which is a hinge.

8 The effect of these two loads is to cause a deflection ∆ at 8

D. The structure is in equilibrium and we have a deformed
shape compatible with the support conditions. This is a real
state.
78 Understanding Structural Analysis

9 9 We will consider first the Theorem of Virtual Displacements.

10 10 The real structural system has imposed upon it a virtual

displacement. The direction of this virtual displacement is
unrelated to the direction of the real deformation. This is a
virtual or notional displacement in the sense that, despite
the introduction of the virtual displacement, the internal
forces, and the external loads W1 and W2 are unchanged by
this virtual deformation.

11 11 In order to make the diagrammatic presentation as clear as

possible we will adhere to the convention shown here.

12 12 As we have assumed that the imposition of the virtual

deflection has no effect upon the external loads, equally
it has no effect upon the balancing internal forces, N1, N2
and N3 in members AD, BD and CD respectively. Note
that because all members are pinned at each end and the
loading restricted to node loading the resistance of the
structure is limited to axial forces.
 The Theorems of Virtual Work 79

13 In the Theorem of Virtual Displacements the expression 13

for the virtual work symbol W is equal to the value of the
real force in the direction of the virtual displacement (i.e. the
component of the real force) multiplied by the value of the
virtual displacement.

14 The external virtual work is determined by resolving 14

the external forces W in the direction of the virtual
displacement. The component of force W1 in the direction
of the virtual displacement is equal to W1 × cos θ1, where
θ1 is the angle between the line of action of force W1 and
the direction of the virtual displacement.

15 For more than one external load acting at D we can 15

determine the resultant force RW that is in the direction of
the virtual displacement.

16 The external virtual work done by the real loads is equal to 16

the resultant component of the real loads in the direction
of the virtual displacement multiplied by the virtual
displacement.
80 Understanding Structural Analysis

17 17 In a similar way, we may find the resultant of the internal

axial forces RN in the direction of the virtual displacement.
The angle is that between the direction of the internal force
and the direction of the virtual displacement.

18 18 Since in the real state the structure is in equilibrium the

sum of RW and RN, the resultants of applied loads and
internal forces in the direction of the virtual displacement,
must be equal to zero.

19 19 The internal virtual work, the symbol for which is u is

equal to the resultant of the internal forces in the direction
of the virtual displacement multiplied by the virtual
displacement.

20 20 However since RW and RN are in equilibrium, the actual

values of the resultant of the external force, and internal
forces will be equal. Therefore, the internal virtual work
must equal the external virtual work, since both RW and RN
are multiplied by the same virtual deformation.
 The Theorems of Virtual Work 81

21 We can now state the Theorem of Virtual Displacements. 21

22 We will examine the application of the Theorem of Virtual 22

Displacements to the ultimate load analysis of the fixed-
end beam ABC loaded at the centre with the load W, such
as to cause collapse. Note that the ultimate resistance
moment at B is twice that at A and C.

23 We will consider a virtual displacement system which is 23

compatible with the supports, by the introduction of hinges
at A, B and C. Point B is assumed to undergo a virtual
displacement of δ and joints A, B and C undergo a virtual
rotation which may be expressed in terms of the virtual
displacement and the span dimension L. Because the
angular rotations are small we can assume that the arc of
the circle radius L/2 is the same as the tangent. Angle θ1
and θ2 are expressed in radians.

24 This is the bending moment diagram at ultimate load, due 24

to the action of the ultimate load W. This system is in force
equilibrium. That is the ultimate moments at A, B and C
are in force equilibrium with the applied, ultimate load W.
82 Understanding Structural Analysis

25 25 To find the relationship between the ultimate load W and

the moments of resistance Mu, we determine the internal
and external virtual work. The internal virtual work due
to a moment is the moment multiplied by rotation that
is the ‘virtual displacement acting against the internal
forces’. The external virtual work is equal to the real force
W multiplied by the virtual displacement. Note that the
proof above is for the general condition. In this example
the direction of the force and virtual displacement are
coincidental and therefore there is no need to determine
the component of the forces in the direction of the virtual
displacement.

26 26 The real ultimate load relationship between the ultimate

load W and moment of resistance for this particular beam
is:

W = 12 × Mu/L

Either W or Mu may be the unknown in this problem.

27 27 We will now develop the equally powerful Theorem of

Virtual Forces.

28 28 We will examine the same pin-jointed frame ABCD with

the same real forces and the resulting real deflection ∆.
 The Theorems of Virtual Work 83

29 We now impose a virtual force p and, at the same time, 29

disconnect the support at C, i.e. there will be internal
virtual forces n, balancing the externally applied virtual
force, p. Note that the internal virtual force in member
CD is zero. Such a system is a virtual force system
because, although it is in equilibrium, it no longer satisfies
the condition of zero deflection at support C. It is not
compatible with the real support systems. Note that the
real deflection ∆ is still identified because we are going to
consider the virtual work condition in the direction of that
real deflection.

30 The external virtual work, w, is equal to the virtual load p 30

multiplied by the component of the real deformation ∆ in
the direction of the virtual load p.

31 The internal virtual work, u, is equal to the resultant of 31

the internal forces, resulting from the application of the
virtual force, in the direction of the virtual force multiplied
by the component of the real deflection ∆ in the direction
of the virtual force. The angle θ is between the line of
action of the virtual force and the real deflection ∆. Angle
α is between the resultant of the internal forces rn and the
direction of the real deflection ∆.

32 Since this is a virtual force state, it is the condition of 32

force equilibrium that is satisfied by the virtual system.
Therefore the virtual force p will be balanced by the
resultant of the virtual internal forces. The external virtual
work w is equal to the internal virtual work u.
84 Understanding Structural Analysis

33 33 The Theorem of Virtual Forces.

34 34 Of the two theorems discussed above, it is the Theorem of

Virtual Forces that is the most used in the development of
classical methods of structural analysis. The Theorem of
Virtual Forces will now be applied to the simply supported
beam ABCD. The vertical deflection at B is the object of
analysis.

35 35 We will apply a virtual force p, a unit load this time,

to the same structure. The structure is sufficiently
straightforward for it to be unnecessary to change the
support conditions, to ease the calculation of the effects
of the virtual force. The unit virtual force is applied in the
direction of the required deformation.

36 36 We must now determine an expression for the internal

virtual work due to bending. In the earlier example, the
internal virtual work on the pin-jointed frame was related
only to the resulting real axial displacement. Here the
deflection at C is due to bending.

We will examine a short length of the member dx, a

distance x, from the left-hand support. Under the real load
w, dx has undergone a rotation βx.
 The Theorems of Virtual Work 85

37 At this section, the value of the real moment is Mx and the 37

virtual moment mx . The virtual moment is that due to the
effect of the virtual action. In this example that is the unit
virtual load at B. That action may be due to a virtual load
or a virtual moment. This will be explained later.

38 The internal virtual work for the elemental length of the 38

member dx is the virtual moment multiplied by the real
rotation βx.

u = βx × mx

39 Under the action of the real loads, the stress in the fibres, 39
distance y from the neutral axis is:

σy = Mx y / I

40 The material of which the beam is made is assumed to 40

satisfy Hooke’s Law, where stress is directly proportional
to strain. Therefore the strain at level y is equal to the
stress at level y divided by E, the modulus of elasticity.
Substituting for stress from above, we obtain an expression
for the strain at level y in terms of the moment at x, the
distance y, E and I.
86 Understanding Structural Analysis

41 41 The total strain over the length dx at level y from the

neutral axis, ∆dxy , is equal to the unit strain multiplied
by the length dx. Substituting for the expression for unit
strain from above, the total strain over the length dx at
level y from the neutral axis is equal to the value of the
bending moment at point x, Mx multiplied by the distance
y and dx all divided by the product of the modulus of
elasticity and the second moment of area of the section.

42 42 Since the actual rotation is small, the rotation may be

assumed to be equal to the strain at level y divided by the
distance to that fibre, y. βx is expressed in radians.

βx = ∆dxy / y

43 43 Substituting for ∆dxy above we obtain an expression for the

rotation at x.

βx = (real moment at x) dx / EI

44 44 We now substitute the expression for βx (Figure 43) in the

equation for internal virtual work in the segment length dx
(Figure 38). Thus the internal virtual work is equal to the
bending moment due to the virtual action, multiplied by the
real bending moment by dx divided by EI.
 The Theorems of Virtual Work 87

45 The expression for the external virtual work is equal to the 45

unit virtual load multiplied by the real deflection ∆B .
v

46 The total internal virtual work in the beam is the 46

integration of the expression for the internal virtual work
u, for the length dx, for the full span of the beam. The total
internal bending strain in the full span contributes to the
deflection at the point of consideration, B. We now have
both sides of the expression for virtual work. The deflection
∆ is the unknown.

47 Virtual forces - bars 47

A similar expression may be developed for the internal
virtual work in pin-jointed frames with loading restricted to
point loads at nodes.

The three-member frame ABC is loaded diagonally at B,

producing internal axial forces, NBA, NBC and NAC.

If the vertical deflection at B is required we apply our unit

virtual load in that direction.

48 The internal virtual work in such systems is equal to the 48

value of the virtual force n for each member multiplied by
the real member displacement. The displacement by simple
consideration of stress and strain is equal to the real axial
force multiplied by the length of the member divided by the
product of the area, A and the modulus of elasticity, E.
88 Understanding Structural Analysis

49 49 This could be expressed as an integral since the length of

the member z is equal to ∫0L x.dx. However, since the axial
force is almost always constant in such structures the
expression in the upper part of this diagram is the simplest
to use.

50 Torsional rotation

50 The bar of constant cross-section is loaded with the

torsional moment, T. The bar undergoes a total rotation of
θ at the free end. There is a balancing internal torsional
reaction equal to the applied moment, T. The rotation θ is
the object of the analysis.

51 51 Applying the Theorem of Virtual Forces, a unit virtual

torsion is applied in the direction of the unknown torsional
rotation θ.

52 52 It may be shown that the internal virtual work is equal

to the integration of the real torsion T, multiplied by the
virtual torsion t, divided by the torsional constant J and the
modulus of rigidity, G.
 The Theorems of Virtual Work 89

53 The subsequent chapters on methods of analysis of 53

statically indeterminate structures depend to a large degree
on the ability to determine structural deformations. To
extend the illustration of the Theorem of Virtual Forces we
will now apply it to two typical problems.

The first is the pin-jointed frame ABCDEF that is subject

to two vertical loads at C and E, 12 and 6 load units
respectively.

54 The vertical deflection at C is required. 54

55 The first step is to determine the distribution of the real 55

forces in the members, N.

56 We now apply a unit virtual load in the direction of the 56

required deformation, i.e. in the vertical direction at C.
90 Understanding Structural Analysis

57 57 The next step is the analysis of the distribution of virtual

forces. The characteristic of the application of the Theorem
of Virtual Forces is that the structure must be stable. If
the structure is statically determinate then releases such
as were applied in the original explanation (Figure 29)
cannot be made, otherwise the structure would become
a mechanism. We cannot simplify this structure for this
particular virtual load, i.e. by the omission of members
AB, BD and BC for example without creating an unstable
structure. Consequently it is necessary to determine the
value of the virtual force in each member.

58 58 The internal virtual work done on member AB, is the value

of the virtual axial force multiplied by the real change
in length. This calculation is to be carried out for each
member in the frame.

59 59 The total internal virtual work is the algebraic sum of these

terms for all members in the frame. In the expression for
the equality of external and internal virtual work, only the
deflection ∆CV is unknown.

60 60 The second example is the simply supported beam ABCD

loaded at B with a vertical point load of 18 load units. The
vertical deflection is required at C.
 The Theorems of Virtual Work 91

61 The first step is to determine the distribution of the real 61

bending moments, M.

62 Using the Theorem of Virtual Forces we apply a unit virtual 62

load in the direction of the required deformation. We then
determine the virtual load bending moments, m.

63 The most reliable approach to this numerical analysis is 63

a tabular method. Firstly, the bending moment diagrams
are split into convenient sections in anticipation of the
expressions for the integral.

64 There is no particular significance in the selection of the 64

first and last point for the integration since the evaluation
of internal work is not related to the sign of the x direction
along the beam axis and so the most convenient is chosen.
The diagram indicates the zero point for the integration for
each section of the beam. The symbol s is chosen for the
length along the member to anticipate the application of
the theorem to two-dimensional structures. The use of s for
the length along the member will avoid confusion with the
horizontal global axis, symbol x.
92 Understanding Structural Analysis

65 65 The integral expression for the internal virtual work is best

created from a graphical summary. This is for the portion
of the beam AB. The total internal virtual work, u for part
AB is 60 units.

66 66 The diagram shows the expression for the real and virtual
bending moments in the portion BC. Note that sagging
moments have been taken as positive. Thus the negative
sign in the expression for the real moments in BC
recognises that the moment is reducing from B to C. The
completion of the integration expression for portion BC
gives a value of internal virtual work for this portion of the
beam of 63.9 units.

67 67 An expression for bending moments and integration for

portion CD is shown. Internal virtual work equals 111.1
units.

68 68 The value of the total internal virtual work is the sum of

the internal virtual work in the parts AB, BC and CD.
Equating internal with the external virtual work, the real
deflection at C is equal to 235.0 divided by EI. The value of
the deflection may be found by substituting for E and I.
 The Theorems of Virtual Work 93

69 The Theorem of Virtual Forces may be used to determine 69

rotations. Using the same beam example, the object now is
to find the rotation of the beam at A, θA .

70 We apply a unit virtual moment at A in the direction of 70

the required deformation. The resulting virtual bending
moment diagram, m is shown.

71 The external virtual work is equal to the real rotation 71

multiplied by the unit virtual moment. The internal virtual
work is determined as above, by integrating the product of
the expression for the real and virtual moments.

72 We will now study the application of virtual work to two- 72

dimensional structures. The frame ABC is loaded at C,
normal to member BC, with the point load w.
94 Understanding Structural Analysis

73 73 The object of the analysis is the horizontal deflection at C,

∆C .
h

74 74 We will now examine the effect of both axial forces N and

bending moments M. Because the real load w is normal
to member BC there is no axial force in the member BC.
However, there is a real compressive force in AB, equal to
the horizontal component of w.

75 75 We now apply a unit virtual load in the direction of the

required deflection that is horizontally at C. The resulting
virtual bending moment and virtual axial load diagrams
are shown. Note that the virtual bending moment in AB is
constant because the virtual load is parallel to AB. Also
the virtual axial force in BC is the component of the unit
virtual load in the direction of BC. Consequently this will
be less than the value of the virtual axial force in member
AB that is equal to 1.

76 76 This figure shows graphical summary of the integration for

the bending effects ... to which must be added ...
 The Theorems of Virtual Work 95

77 ... the axial load effects. 77

78 We will now study the application of the Theorem of 78

Virtual Forces to a statically indeterminate structure, the
continuous beam ABCD loaded with a point load W in
span BD. The vertical deflection at C is required.

79 This figure shows the qualitative distribution of real 79

reactions and bending moments.

This structure is indeterminate, consequently until we are

conversant with analytical methods that follow this chapter
we cannot determine the numerical distribution of reactions
and bending moments. Thus the theorem can only be
applied to a structure that has already been solved because
it is necessary to know the distribution of real load bending
moments, M.

80 We will now be able to see the power of the Theorem of 80

Virtual Forces. We can release the fixed support at A
reducing the structure to a statically determinate form.
This will allow us to determine the distribution of bending
moments due to the application of the unit virtual load,
applied in the direction of the required deformation.
96 Understanding Structural Analysis

81 81 Note that because the values of virtual moments are zero

in span AB, the product of the expression for the real and
virtual moments will be zero. Consequently the integration
will be limited to span BD only.

82 82 Practice Problems 1
These structures are to be analysed for deformations
at X and Y. The subscript α indicates a rotation (i.e. Xα
rotation at X) and δv vertical deflection. δh is a horizontal
deflection. Draw the appropriate real M and virtual mδ or
mα, diagrams.

Practice Problems 2

You should now practise the application of the Theorem of

Virtual Forces to numeric problems.

Use the problems in Chapter 1 for which you should have

a computer solution. Since each of these problems is
statically determinate, they may be used as problems in the
application of the Theorem of Virtual Forces. You should
choose a particular deflection or rotation and find the
value. Since the real moments are given in the computer
solution it is only necessary to identify the direction of the
appropriate virtual action and determine the distribution of
virtual moments. Remember, you should always carry out a
qualitative analysis of the effect of the virtual action before
commencing any numerical analysis.

The solution to the beam problems should check exactly

with the computer solution. However, you must remember
that the computer solution to two-dimensional structures
includes the effects of axial displacement.

You may practise the application of this theorem to

statically indeterminate structures using the problems in
Chapter 6.
6 The Flexibility Method

The Flexibility Method is one of the two main approaches to the analysis of redundant
structures, the other being the Stiffness Method. Both methods are based on identical
theories of material behaviour and differ only in the mathematical treatment of the
basic structural data. Neither method is particularly suited to the hand analysis of real
structures since the conclusion to both methods of analysis is a set of simultaneous
equations, dependent upon the degree of indeterminacy of the structure. In a real
structural frame, the degree of indeterminacy is likely to be large and although
theoretically possible, this hand analysis is too lengthy and error-prone to be a feasible
design office procedure. If such an analysis is required, a computer program based on
the stiffness method is the most likely course of action with the ready availability of such
proven software. Alternatively, in real structures the symmetrical shape of the frame may
allow an approximate analysis of forces and moments or the use of an iterative method
such as moment distribution. This is based on the stiffness approach and will be discussed
in Chapter 9.

The analyst must always bear in mind, however, that he will, at best, be approximating
to the behaviour of the real structure. He will be creating an analytical model that is a
theoretical model of the complex, real structure. The advantage of the flexibility method
is that it contributes significantly to an understanding of the real structure because the
procedure of the method; that of reducing a complex structure to a statically determinate
form, is closely related to the fundamental design decisions of rationalising the real,
complex and highly redundant structure into a simpler form suitable for analysis. That
simplifying procedure is likely to be adopted even when a computer is used. Three-
dimensional structures are usually reduced to a series of linked two-dimensional frames,
for example. These frames may be reduced to a series of sub-frames or further simplified
to continuous beams. The simplification of the real structure which allows the creation of
these analytical forms must not yield a solution less safe than the behaviour of the real
structure and ultimately it is towards this understanding that you must see your study of
the flexibility method directed. The flexibility method is based on structural deformations
for which the Theorem of Virtual Forces is used.
98 Understanding Structural Analysis
 The Flexibility Method 99

1 The two principles on which the method of flexibility rests

1
will be illustrated with the propped cantilever AB. The first
step in the analytical procedure is to determine the degree
of indeterminacy. There is a potential horizontal reaction
at A; however, for this particular loading arrangement it
is zero. There being four notional external reactions the
propped cantilever is 4 − 3 equations of equilibrium
= 1 x statically indeterminate.

2 The second step is to release the structure to a statically

2
determinate form. The reason for this is that the
flexibility method depends upon the ability to determine
the deformations of the released determinate structure.
The Theorem of Virtual Forces is used to determine
the deformations and this method can only be applied
to a statically determinate structure if the real bending
moments and other load effects are known. The vertical
reaction at B is removed reducing the structure to a
cantilever beam. The structure deflects downwards with a
deflection of ∆B at B.
v

3 Now imagine that we know the value of the reaction that

3
has been removed, VB, and we apply it at B. The reaction
VB will return the structure to the original position at B,
exactly, because that is what defines the value of that
reaction, the fact that it holds the beam in that position
with a zero vertical deflection at B.
100 Understanding Structural Analysis

4 4 However, we do not know the value of VB. It is the object

of the analytical procedure. If, however, we apply a unit
load or action in the direction of the removed reaction the
rate of deflection for the application of a unit load applied
at this point, i.e. δB B per unit load, may be found. We also
v v
determine the deformation due to the applied load in the
direction of the removed restraint. These are the second
and third stage in the analytical procedure:

2. Determine deformation due to the application of

the design load ∆B in the direction of the removed
V
restraint.

3. Determine deformation(s) resulting from the

application of unit load(s) and/or reactions, δB B .
v v
Note particularly that the unit loads are applied in the
direction of the removed restraint.

5 To complete the analytical procedure, we apply the

5
principle of superposition. This principle allows us to add
separately identified load effects together, the final solution
being the sum of the parts. The separate effects are the
removal and replacement of VB. One effect is the release
producing the deflection at B, ∆B . The other effect is the
v
deflection at B due to the unit load at B. If we multiply this
rate of deflection by the unknown reaction VB it is clear
that these two deflections ∆B and VB. δB B must have the
v v v
same value.

6 The last stage in the analytical procedure is the setting up

6
and solving of the equation(s) of compatibility (there may
be more than one degree of indeterminacy). The equation of
compatibility is equated to zero because the true deflection
at B of the original structure is zero.
 The Flexibility Method 101

7 The deformations that result from the analytical procedure

7
are given a sign appropriate to the overall convention. In
this case ∆B would be negative and δB B positive. The
v v v
choice of direction for the unit load and thus the direction
of the unknown reaction VB is arbitrary because in a more
complex structure the true direction will not be as apparent
as it is in this simple example. For this propped cantilever
however, it is obvious that VB will act upwards.

8 The solution of the equation of compatibility is that VB is 8

positive. This means that the choice of direction for the unit
load (Figure 4) is correct confirming that the direction of
the reaction VB is upwards. If the sign had been negative
then the direction of the reaction would have been opposite
to that chosen for the unit load.

9 We must remember that the evaluation of the unknown

9
reaction is not, normally, the only object of the analysis.
Usually, the distribution of bending moments is required.
The principle of superposition may be applied here too. The
bending moment diagrams that result from the application
of the point load and reaction at B are combined to produce
the final bending moment diagram.

10 Most real structures are highly redundant and we will now

10
examine the application of the flexibility method to the
beam ABC that is 2 x indeterminate.

Step 1. Reduce the structure to a statically determinate

form: remove VB and VC .
102 Understanding Structural Analysis

11 Step 2. Determine deformations due to the application of

11
the real load(s) in the direction of the removed restraints:
∆B and ∆C . Try to see the flexibility method as collecting
v v
deformations due to the real and unit loads at a particular
point, in a particular direction.

12 Step 3. Apply unit loads and determine deformations in the

12
direction of the removed reactions. Now we can see the
reason for the rather cumbersome system of subscripts. It
allows a clear definition of a particular deformation. Note
the complication of more than one degree of indeterminacy.
The application of the unit load at one point contributes to
the deformation at another.

13 Interpretation of subscripts. The capital Greek symbol ∆

13
is used for deflections due to the real loads and the lower
case δ for deflections resulting from the application of the
unit loads.

14 Step 4. Set up and solve the equations of compatibility. This

14
step is confusing at first. Remember that you are collecting
all the deformations at one point in one direction to make
up one equation of compatibility.
 The Flexibility Method 103

15 To begin with, check each equation. The first subscript of

15
the unit load deformation must agree with the real load
deformation. The second unit load deformation subscript
must relate to the unknown reaction.

16 Having found the unknown reactions VB and VC, the beam 16

may also be solved for the remaining reactions at A, MA
and VA, by applying the three equations of equilibrium.
Note that we can identify the direction of the unknown
reactions VB and VC by a consideration of the qualitative
analysis of the beam. You should always carry out such a
check. Numerical errors are easy to make in this method.

17 For example, a minor error in the calculations of a

17
deformation could yield values of reactions that produce
a net hogging moment at A. Only a sound understanding
of the qualitative analysis of structures would help you to
spot such a mistake.

18 We will now examine an alternative method of reducing the

18
structure to a statically determinate form by the removal of
a moment instead of a load reaction.

Step 1. The removal of the moment reaction at A and

the vertical reaction at B. The beam is now statically
determinate, simply supported at A and C.
104 Understanding Structural Analysis

19 Step 2. Deformations due to the application of the design

19
load. Now instead of a deflection there is a rotation at A
due to the removal of the moment reaction MA. In addition
we must determine the vertical deflection at B, ∆B .
v

20 20 Step 3. The application of the unit loads. At A, instead of a

load, a unit moment is applied. Note the new subscript ‘m’.
Used as a first subscript it refers to the fact that the
deformation is a rotation. As a second subscript it identifies
the fact that the particular deformation is due to a moment
and not a force. ‘δB A ’ is interpreted as a deformation due
v m
to a unit action (δ) at B in the vertical direction (Bv), due to
a unit moment at A (Am).

21 Step 4. Equations of compatibility. It is inevitable that

21
you will experience some difficulty in deciding which
deformation is appropriate in which equation. You should
keep reminding yourself that you are collecting together
bits of the same deformation in the same direction.

22 Having solved the equations of compatibility we can check

22 the direction of the reactions. We know the direction of
the unknowns, MA and VB from a consideration of the
qualitative analysis. MA is clockwise and VB acts upwards.
The unit moment at A was applied in the anticlockwise
direction (figure 20), consequently the sign for MA in
the solution of the equations of compatibility would be
negative indicating that the direction originally chosen
was incorrect. The choice of direction for the unit action is
arbitrary. However, the sign of the resulting displacements
must be in accord with the overall convention, i.e. upwards
and to the right, and clockwise positive.
 The Flexibility Method 105

23 The next problem is the two-dimensional frame ABCD, 23

fixed at A and pinned at D. A question often raised by
students is the problem of sway. This is because the effect
of sway is considered as a separate loading case in the
method of moment distribution. In the flexibility method
however, it is automatically accounted for in the analytical
procedure.

24 We are now going to adopt the symbol convention for the

24
deformations that is used in most current texts. It provides
less information about the particular deformation, but it
matches conventional matrix notation. The structure is
twice times indeterminate and we will choose to release the
structure to a statically determinate form by the removal
of the two reactions at D, creating the determinate ‘tree’
discussed in the Indeterminacy chapter. These are defined
as X with a numerical subscript that does not define the
axis of application. The choice could equally well have been
X2 in the horizontal direction.

25 Step 2. Deformations due to the application of the design

25
load. The subscripts now relate to the direction of the
unknown force (or moment) as identified above.

26 Step 3. Deformations due to the application of the unit

26
load in direction 1. Note that the unit load in direction 1,
horizontal, produces a deflection in direction 2, δ21.
106 Understanding Structural Analysis

27 This figure shows a clarification of the interpretation of the

27
subscripts for the unit load deformations.

28 The application of the second unit load, and the resulting

28
deformations are shown.

29 The final step in the analytical procedure is the setting up

29
of the equations of compatibility.

30 The equations of compatibility may be rewritten in the

30
form of a matrix of deformations due to the unit loads the
unknown forces, X and the design load deformations, ∆.
 The Flexibility Method 107

31 Note that the unit load deformation matrix is dependent

31
only on the properties of the structure, and is known as the
flexibility matrix. The values of ∆1 and ∆2 are dependent
upon the particular loading arrangement.

32 Practice problems 1. Beams

32
A number of structures are shown. The degree of
indeterminacy has been determined and the choice of
release made. Draw the following diagrams for each
problem:

The full three-part solution for the effect of the application

of the design loads and each of the unit loads:

1. deflected shape

2. bending moment diagram

3. reactions
33
Indicate using the first of the subscript systems, the
deformations appropriate to the solution.

33 Practice problems 2. Frames

Adopt the same procedure.

34 There is a potential confusion between the unit load applied

34
for the flexibility solution and the unit load applied as
part of the analysis of a particular deformation using the
Theorem of Virtual Forces. We will examine the qualitative
solution for the deformation due to the applied load for the
frame studied previously. Refer back to diagram 6.25.
108 Understanding Structural Analysis

35 This is the bending moment diagram due to the application

35
of the vertical point load W in the member BC. These are
the ‘real’ bending moments M of the virtual work solution.

36 We will now treat the determination of the vertical

36
deflection ∆2 as a quite separate problem for the Theorem
of Virtual Forces. In order to determine the vertical
deformation at D we apply a unit virtual load in the
direction of the required deformation.

37 This diagram shows the resulting distribution of bending

37
moments due to the application of the unit virtual load.

38 The true vertical deflection at D should include the effects

of axial shortening in member AB. However, these are
usually small compared with deflections due to bending
38 and we are going to ignore them. This diagram shows a
graphical summary of the integration of the internal virtual
work, u required for the evaluation of the vertical deflection
at D, due to the design load W.

Note that the real bending moments M, beyond the point

of application of the load W are zero. Consequently it
is unnecessary to calculate the value of the virtual load
moments for this part of the structure, since when they are
multiplied in the expression for the internal virtual work
integration, the result will be zero.
 The Flexibility Method 109

39 We will now study the full qualitative analysis for the

39
design load deformations ∆1 and ∆2 using the Theorem of
Virtual Forces.

40 These two deformations ∆1 and ∆2 require three bending 40

moment diagrams to be evaluated; The real moments M,
due to the design load W and the bending moment due to
each of the unit virtual actions.

41 Clerk-Maxwell’s Theorem of Reciprocal Deflections

41
The flexibility matrix is always square because the number
of equations (rows) is always equal to the number of
unknown forces or reactions (columns). Clerk-Maxwell’s
Theorem of Reciprocal Deflections allows us to reduce the
number of elements we need to calculate those below and
including the leading diagonal of the matrix.

Deformations on either side of the leading diagonal, δ12/

δ21, δ13/δ31, etc. are of equal value according to Clerk-
Maxwell’s theorem. We will now prove it by studying the
two deflections δ12 and δ21.

42 We will apply the real, horizontal, (flexibility) unit load and

42
examine the resulting unit load deformation, δ21 that is in
the vertical direction. This is now the object of the analysis.
110 Understanding Structural Analysis

43 The virtual forces solution would require the application

43
of a unit virtual load in the vertical direction, that is, in the
direction of the required deformation.

44 We can now draw the graphical summary of the virtual

44
work solution for δ21.The real moment M diagram is due
to the application of the real (flexibility) unit load, in the
horizontal direction at D.

45 We will now look at the application of the real unit load

45
at D in the vertical direction and the resulting horizontal
deflection, δ12.

46 46 The appropriate unit virtual load for the deformation δ12,

is applied in the direction of the required deformation,
horizontally at D.
 The Flexibility Method 111

47 This figure shows the graphical summary of the virtual

47
work solution for deformation δ12.

48 We will now compare these two graphical solutions.

48
It is apparent that the bending moment diagrams are
interchangeable. In one equation the diagram is the result
of the application of a real unit load and in the other a
virtual unit load. However, because the diagrams are both
for unit loads, the actual values of bending moments and
consequently the deformations δ21 and δ12 have the same
value.

49 This may be expressed in this general form and means that

49
only the leading diagonal of the matrix elements need to be
calculated and those below this diagonal, in order to set up
the equations of compatibility.

50 Statically indeterminate pin-jointed frames may be solved

50
by the flexibility method. In this example the frame is
internally and externally indeterminate.
112 Understanding Structural Analysis

51 51 We will release the structure by cutting the member AC,

and identifying the axial forces as X1 and removing the
vertical reaction at E, X2.

52 Having determined the degree of indeterminacy and an

52
appropriate release system that is the first step in the
flexibility procedure, the second step is to apply the real
design load and determine the deformation in the direction
of the releases. That is the displacement of the ‘cut’ in
member AC, ∆1 and the vertical deflection at E, ∆2.

53 Third step. The application of the unit loads and the

53
identification of the unit load deformations. In direction
1, the cut in member AC, an equal and opposite unit load
is applied because the effect of the cut releases the bar
on either side of the cut. This produces a deformation
δ11, which is the extent to which the cut closes (or opens
depending on the direction and application of loads) as
a result of the unit axial load. This load also causes a
deflection at E, δ21.

54 54 The application of the unit load in direction 2. This

produces a vertical deflection at 2, δ22 and a displacement
of the cut in member AC, δ12. Note that the Theorem of
Virtual Forces would be used to determine the deformations
in this pin-jointed frame. The procedure was illustrated in
the previous chapter.
 The Flexibility Method 113

55, 56 Practice Problems

55
Determine the reactions and full distribution of axial forces
in each of the beams and frames shown, substituting values
for dimensions, loads and section properties, and check
against a computer solution.

You should practise full numerical solutions for the

problems shown above in Figures 6.32 and 6.33. Use the
QSE software to generate the deformations and produce the
check solution.

56
114 Understanding Structural Analysis
7 The Stiffness Method – Frames

The Stiffness Method of analysis is similar in its approach to the flexibility method of
analysis, the major difference being that the unknowns in the procedure in the stiffness
method are the node displacements, whereas in the flexibility method they are the internal
forces, moments and reactions. The other major difference between the two methods is
that the stiffness method requires no judgement to be exercised by the analyst, whereas
the flexibility method requires a choice of releases. This makes the stiffness method
particularly suited to automatic computation and it is the basis of virtually all ‘Plane
Frames’ and ‘Space Frames’ computer programs.

The strength of the stiffness method, i.e. its suitability as a basis of automatic
computation, is its weakness as a method of hand analysis. Because the method requires
little or no judgement to be exercised, practice in stiffness problems promotes little
understanding of structural behaviour. On the other hand, the flexibility method requires a
sound understanding of structural behaviour and statics in the first steps of the analytical
procedure.

The reader should be aware that neither the flexibility nor stiffness method are practical
manual methods of analysis in the design office because real structures are too highly
redundant and must be checked for a number of different load cases. The analytical
procedure of both methods, requiring the setting up and solving of large numbers of
simultaneous equations is cumbersome, time-consuming and error-prone. It is also illusory
in terms of accuracy. It may be mathematically accurate but no more accurate in terms of
the relationship of the analytical model to the real structure than the method of moment
distribution. The latter is sometimes described as ‘approximate’ as if it were approximate
to the behaviour of the real structure. It is no more or less close to the real structure than
the so-called ‘exact’ methods of stiffness and flexibility, which are based on idealised
values of material and section properties.
116 Understanding Structural Analysis

1 1 Before starting to explain the stiffness approach to the

analysis of statically indeterminate structures, we need
to define certain relationships between deformations and
member properties. It is assumed that the structural
members are straight and of the same cross-sectional
area throughout. This is not a serious restriction as most
structural members come into this category.

We will start by determining the value of the moment that

will cause a rotation of 1 radian at the pinned end of the
propped cantilever beam AB.

2 2 To distinguish this moment from normal load moments

it is defined as MAb, the absolute moment, that moment
which produces a rotation of 1 radian. The diagram of
the deflected form above, reveals that there is a hogging
moment at the fixed support at A. We refer to the ‘near’
end as that which has the absolute moment applied,
consequently, the hogging moment is at the ‘far’ end A.

3 3 In order to solve this structure, which is 1 x statically

indeterminate, we will determine the reaction, VB .
Consideration of the overall moment equilibrium at A will
show that the vertical reaction at B is downwards, to be
balanced by an upwards reaction at A. Note that there will
be an anticlockwise moment reaction at A, which we can
identify from the deflected shape.

4 4 Using the flexibility method, we reduce the structure to a

statically determinate form by the removal of the vertical
reaction at B, VB. The effect of the absolute moment will be
to produce an upward vertical deflection at B, ∆B .
v

Note that MAb, produces tension on the underside of this

statically determinate cantilever beam.
 The Stiffness Method – Frames 117

5 The flexibility unit load is applied in the direction of the 5

removed reaction. Here in order to illustrate that the choice
of the direction of the unit load is arbitrary, it has been
applied in the opposite sense of the real reaction VB, that is
in the positive z (vertical) direction. The unit load produces
a bending moment of L kNm at A.

6 Using the Theorem of Virtual Forces to find the deflections 6

we must apply a unit virtual load in the direction of the
required deformations ∆B and δB . Note that the value of
v v
the virtual bending moment at A is L kNm. Remember that
we distinguish virtual bending moment diagrams from the
effect of real loads by leaving the virtual diagram without
hatching lines.

7 This figure is a graphical summary of the integration 7

expressions for the internal virtual work for ∆B and δB
v v
using the Theorem of Virtual Forces. Note that for δB the
v
integration is the effect of a unit load ‘squared’.

8 We will now carry out the quantitative solution for ∆B 8

v
assuming that sagging bending moments are positive. The
load moment is constant for the ‘M’ real bending moment
in the expression and consequently does not contain an ‘x‘
term. The origin is taken at B so the expression for the m
moments is 1 times x.
118 Understanding Structural Analysis

9 9 The figure shows the quantitative solution for δB . We

v
could have found this directly as this loading case on a
cantilever is one of the standard cases. The deflection at
the end of a cantilever loaded with a point load = WL3/3EI.

10 10 The final stage in the solution is the equation of

compatibility for the deflection at B. Note that the negative
sign of the solution for VB indicates that the choice of
direction for the real unit load (Figure 5) was incorrect.

11 11 Thus the value of the vertical reaction at B for the propped

cantilever, loaded with the anticlockwise moment MAb is
VB = (3/2) x MAb / L, that is acting downwards.

12 12 If we now consider the overall moment equilibrium at A:

MA − MAb + L.(3/2).MAb / L = 0

MA = −MAb / 2

We find that the value of the fixing moment at A is MAb / 2.

 The Stiffness Method – Frames 119

13 The relationship between the moment applied at the pinned 13

end, the ‘near end’ of a prismatic beam, and the moment
created at the ‘far end’, end A, is known as the carry-over
factor, value ½.

14 We now need to know the relationship between the rotation 14

at B and the applied moment MAb. We will determine the
value of θB by the Theorem of Virtual Forces. This theorem
may be applied to a redundant structure, as was discussed
in Chapter 5. The bending moment diagram is that for the
redundant structure, the values of which have already been
determined. In the subsequent analysis this will be the ‘M’
diagram, the effect of the real loads.

15 From the Theorem of Virtual Forces (see Figures 5.78 15

– 5.81) the structure sustaining the virtual load need
only be able to satisfy equilibrium. The virtual structure
does not have to be compatible with the original support
conditions. The propped cantilever may be reduced to a
statically determinate form by the removal of the support
at B to make AB a cantilever. The unit virtual action is
then applied to the cantilever with the resulting constant
bending moment diagram shown. Note that the unit virtual
moment is applied to the beam in the direction of the
required deformation θB.

16 This diagram shows the graphical summary for the 16

integration for the internal virtual work that results from
the application of the unit virtual moment at B.
120 Understanding Structural Analysis

17 17 The figure shows the quantitative solution of the

internal virtual work expression for θB. The origin for
the expression for the real and virtual bending moments
has been taken at A, clockwise bending moments, that
is bending moments producing sagging in the beam, as
positive. From this solution we find that the value of the
moment to produce a unit rotation at the near, pinned end
of the propped cantilever = 4EI/L.

18 18 This relationship between the absolute moment and the

member properties is most important and is fundamental
to the application of both the stiffness and moment
distribution methods of analysis. Note that this is an
analytical device. No real structure could sustain a rotation
as large as 1 radian.

Note the units: F/L2.L4/L = F. L.

(E) (I/L)

19 19 Similarly, if the end of the beam is pinned, it may be shown

that the moment required to produce a rotation of 1 radian
at the near end is 3E.I/L. The carry-over factor is clearly
zero, since A is simply supported.

20 20 We need to establish one more relationship, that of the

effect of the settlement of one end of a fixed-ended beam,
and the resulting bending moment. Note that, in this case,
both ends are fixed. End B settles downwards relative to
end A. This implies a vertical force (but not a moment)
release at B.
 The Stiffness Method – Frames 121

21 We will assume that the bending moment caused by the 21

settlement of AB is M. There is a balancing vertical
v
reaction at A, equal and opposite to the force required to
produce the settlement. If we consider overall moment
equilibrium about A:

−M −M + VB.L = 0

Therefore VB = 2M/L

22 By applying the Theorem of Virtual Forces to the evaluation 22

of the deflection ∆B we can reduce the fixed-ended beam
v
to a cantilever for the application of the unit virtual force
at B. This is applied in the direction of the unknown
deflection, ∆B , that is in the vertical direction at B.
v

23 This figure shows the resulting virtual bending moment 23

diagram, m. Note that the sign of the bending moment is
negative at A; the action of the joint on the member.

24 This diagram shows the graphical summary of the 24

integration for the internal virtual work to determine the
value of ∆B . Note again the particular advantage of the
v
Theorem of Virtual Forces in that the original structure
may be simplified to a statically determinate form, allowing
the evaluation of the expression for m, which is now based
on a cantilever.
122 Understanding Structural Analysis

25 25 The quantitative solution for the internal virtual work

is shown. The origin has been taken at A. Thus the
expression for the bending moment due to the virtual unit
load is the bending moment at A, − L plus the effect of the
reaction, I upwards at A.

26 26 Thus for a settlement of A the resulting bending moment is

6 E I ∆ / L2.

Note that this negative (i.e. downwards) deflection will

create a negative (anticlockwise) bending moments at A
and B. However, the sign of moments due to settlement
should always be checked against the resulting qualitative
bending moment diagram and the deflected shape.

27 27 If one end of the beam is pinned and there is a relative

vertical deflection or settlement then it may be shown
that...

28 28 ...the relationship between the settlement deformation and

the bending moment induced at the fixed end A is:

M = 3 E I ∆ / L2
 The Stiffness Method – Frames 123

29 The stiffness analysis procedure will now be illustrated 29

with the frame ABC, loaded vertically in the beam BC
with the point load W that is central in the span. The
height of the column AB is equal to the span BC.

30 We will show that the stiffness analysis method for 30

this structure has only one unknown, the rotation at B,
ignoring the effect of the axial strains. The significance
of this assumption will be explained later in the chapter.
If this rotation is determined, the full distribution of
bending moments may be found using the relationships
between bending moment and rotation, bending moment
and deflection developed above. This unknown rotation is
known as a ‘degree of freedom’ that defines the number
of unknowns to be found in the stiffness solution. This is
known as ‘kinematic’ indeterminacy as compared with
‘statical’ indeterminacy used in the flexibility method of
analysis.
31 Axial deformations will give further ‘degrees of freedom’. 31
However, for the time being axial deformations are ignored.
We will start by studying the column AB. Point B is fixed
in space, but free to rotate. For member AB, we know the
relationship between the bending moments at A and B and
their values in terms of the rotation θB. Bending moments
are induced at each end, 4 E Z θB/ L at the ‘near’ end, end
B, and half is ‘carried over’ to the far end, end A.

32 Now we can look at the effect of the rotation θ on member 32

BC. Similar to the column, bending moments are induced
at the ‘near’ end B and the far end C.
124 Understanding Structural Analysis

33 33 To consider the effect of the load, we will start by assuming

that joint B is prevented from rotating. There will,
therefore, be a fixed-end moment at this artificial restraint
due to the application of the applied load W. The bending
moment at B is equal to this artificial restraint, just as if
end B were fully fixed.

34 34 The final bending moments in the beam BC will be the

sum of the effect of the restrained joint B and the effect
of the rotation θB. This is another example of the principle
of superposition. The application of the real rotation θB
releases the artificial restraint at B. This will produce the
true condition of equilibrium at B, i.e. no external restraint
and a rotation of θB.

35 35 This figure shows the distribution of the final bending

moments. Note that the bending moment at B must be
the same for both members. The only unknown in this
analytical procedure is the rotation at B. This then is the
object of the stiffness method of analysis.

36 36 We will now carry out a quantitative analysis of the

frame ABC, previously studied qualitatively. We start by
artificially restraining the rotation at joint B. The moment
required to prevent the rotation at B, R, is W L / 8 ,
assuming that the point load is in the centre of the beam
BC. Because both ends of beam BC are fully fixed against
rotation, the moments are known as ‘ fixed-end ‘ moments.
This distribution of bending moments for a fixed-end beam
is a standard solution, which may be proven by this and the
subsequent analytical methods.
 The Stiffness Method – Frames 125

37 The next step is to apply a moment ‘r’ at B sufficient to 37

cause a rotation of 1 radian at joint B. Note particularly
that this means that ends B of both members AB and BC
have undergone a rotation of 1 radian.

38 This diagram shows the distribution of bending moments 38

in the frame due to the application of ‘r’ and the rotation
of 1 radian at B. You will note that the bending moments
at B are on different sides of the member. They are in
equilibrium with the applied moment r; that is they are
shown as the action of the joint on the member, i.e. in a
clockwise direction. The externally applied moment r is
balanced by the distribution into the column AB and the
beam BC. Because the rotation at B is one radian the
resulting induced bending moment in each member is
4EI/L.

39 To simplify this calculation it is assumed that the spans 39

are the same, and that the E I value is constant for both
members.

‘r total’ is known as the (absolute) stiffness of the joint

and is the sum of the stiffness (remember that means the
moment required to produce a rotation of 1 radian) of each
of the members framing into the joint.

40 In fact the artificial restraint at B does not exist. Joint B 40

is free to rotate to moment equilibrium consistent with the
stiffness of the members. Therefore the artificial external
restraint at B induced by the applied load W is equal to the
moments induced by the rotation of 1 radian multiplied by
the actual rotation θB. This is the equation of equilibrium of
the moments at the joint.
126 Understanding Structural Analysis

41 41 If we adopt the normal sign convention of clockwise

moments positive, (for bending moments, that is the action
of the joint on the member), the artificial restraint
R = −W L / 8, that is in an anticlockwise direction.

The total absolute (i.e. producing a rotation of 1 radian)

moment for both members BA and BC = 8 E I / L . Thus
the rotation at B may be determined:

θB = W L 2/ 64 E I

This is a positive value, indicating that the direction

assumed for the action of the absolute moment r at B
correctly defines the direction of the real rotation θB.
42

42 The actual value of the rotation at B may now be

substituted to determine the final distribution of moments
in the structure, first the column AB. The bending moment
at B is equal to the rotation at B multiplied by 4 E I / L
which equals W L / 1 6 . Half of this bending moment is
carried over to the ‘far’ end, end A.

43 43 The final bending moments in the beam are the algebraic

sum of the distribution of bending moment resulting from
joint B being artificially restrained against rotation and...

44 44 …the bending moments due to the rotation of W L2 / 64 E I

at B.
 The Stiffness Method – Frames 127

45 The final bending moments in the beam BC are the sum 45

of the restrained and ‘released’ conditions. This is another
example of the principle of superposition.

46 This figure shows the final frame bending moments. Note 46

that the sagging moment in span BC may be found by
superimposing the ‘free’ bending moment diagram over the
restraint bending moments at B and C.

47 Had this problem been solved by the flexibility method 47

of analysis, the first step would have been to determine
the degree of statical indeterminacy. The frame is 3
x indeterminate. We could reduce it to a statically
determinate form by the removal of one support, three
reactions, at C. Remember that the degree of indeterminacy
is the number of unknowns in the flexibility method of
analysis.

48 Applying the stiffness method there is apparently only one 48

unknown, the rotation at B. However, we have ignored
the axial deformation in both the horizontal and vertical
directions. There would be a large axial force in member
AB that would cause it to shorten. Thus point B would
‘settle’ relative to point C, inducing the bending moments
shown. In addition member BC is also in compression,
which would cause it to shorten. These are, however, of
a second order when compared with the primary bending
moments and may usually be ignored in hand calculations.
128 Understanding Structural Analysis

49 49 Now we can see that there are three unknowns at B; one

rotation and two deflections.

Indeterminacy in the stiffness method

This is identified by the degree of freedom at all
unrestrained joints and is known as kinematical
(movement) indeterminacy. As has been demonstrated
above, there are always three degrees of kinematical
indeterminacy at any joint. Generally, in rigid-jointed
frames, however, the axial deformations are ignored in the
manual analysis of ‘plane frames’ leaving only one, the
rotation. All commercially available computer programs,
however, include the effects of the axial shortening.
50
50 The effects of movement at a joint ( i.e. relative to a
support) are always included. If we now introduce a
horizontal and a moment release at C we create a further
degree of kinematical indeterminacy, the horizontal
deformation, which will change the distribution of bending
moments. This unknown deformation ∆C must be found to
h
solve the frame. We would still ignore the axial shortening
of AB, however the sway of the frame at C will produce
bending moments of the same order as those due to the
rotation of joint B and must be included.

51 51 In order to illustrate the application of the stiffness method

to more complex structures, we will study this three-bay
frame. The joints are numbered 1–7.

52 52 Step 1. The first part of the analysis is the identification

of the degrees of freedom, the degree of kinematical
indeterminacy. Ignoring axial strains, they are the rotations
at 1, 2 and 3 and the horizontal deflection of member
1–4, identified as the horizontal deflection at 4. Because
we are ignoring axial strains, it is assumed that there is the
same horizontal deflection at joints 1, 2 and 3.
 The Stiffness Method – Frames 129

53 Step 2. Each of the degrees of freedom is now notionally 53

restrained. Thus the rotation at joint 1, 2 and 3 is ‘fixed’
and the horizontal deflection of joint 4 prevented by the
introduction of an artificial horizontal reaction. R1 to R4
are the ‘clamp’ forces required to prevent rotation and
deflection. R1 to R3 are moments, R4 is a force.

54 Because of the artificial restraints, there are unbalanced 54

bending moments at each of the joints. The bending
moments diagram shows the characteristic discontinuity at
joints 1, 2 and 3. The difference between these moments
must be balanced by the artificial restraining moments,
R1, R2 and R3.

55 This out-of-balance or restraining moment (or force) is 55

taken by the clamp. If, as shown in this example, M21 is
less than M23 then the clamp moment to hold joint 2 in
equilibrium would be in an anticlockwise direction, in order
to balance the tendency to rotate this joint in a clockwise
direction due to the effect of the larger bending moment
M23.

56 Step 3. This next step in the analytical procedure is the

release of each of the restrained joints in turn and the
application of the moment required to induce the rotation of 56
1 radian. R11 is the moment required at joint 1. Note that
this rotation at joint 1 will introduce a horizontal reaction,
r41, as well as a carry-over moment at 2. The subscripts
are interpreted as

r11, moment induced at joint 1 due to a rotation at joint 1,

r21, moment induced at joint 2 due to a rotation at joint 1,

r41, force induced at joint 4 due to a rotation at joint 1.

130 Understanding Structural Analysis

57 57 This diagram shows the distribution of bending moments

due to the application of a rotation of 1 radian at joint 1.
The horizontal reaction at 4, r41, is induced by the rotation
of 1 radian at joint 1 and is equal to the horizontal reaction
at joint 5. The general expression for the value of these
balancing reactions was determined above in Figure 18.

58 58 Joint 2 is released and a rotation of 1 radian induced.

Bending moments are carried over to joint 1: r12, joint 3:
r32 and a reaction, r42 is created at joint 4.

59 59 This figure shows the distribution of bending moments due

to a rotation of 1 radian at 2. Note that because the far
end of member 2 is pinned, the bending moment at 2 in
member 26 is 3 E I2 / L 2 . Similarly, the horizontal reaction
r42, equal to the horizontal reaction at joint 6, is that for a
member with the far end pinned.

60 60 This figure shows the release of joint 3 and a rotation of 1

radian.
 The Stiffness Method – Frames 131

61 Bending moments from the rotation of joint 3 and the force 61

induced at 4, r43, are shown. Because the far end of joint
4 is pinned, the bending moment at 3 in member 34 is
3 E I6 / L6 .

62 The last release is the artificial horizontal restraint at 4. 62

r44 and is that force which will induce a unit horizontal
displacement at 4.

63 This figure shows the bending moments induced in columns 63

15, 26 and 37 that are similar to ‘settlement’ moments.
See Figures 20–28 above. Note again the effect of the pin at
the ‘far’ end of member 26.

64 We may now establish the equations of equilibrium, for 64

the displacement at joints 1, 2, 3 and 4. Clerk-Maxwell’s
Theorem of Reciprocal Deflection applies as it did in the
evaluation of matrix elements for the flexibility method.
r13 = r31 and so on.
132 Understanding Structural Analysis

65 65 This may be expressed in a matrix form where [r] is the

stiffness matrix. It will be noted that the elements of the
stiffness matrix are the forces or moments induced at the
various artificial restraints (or degrees of freedom) by each
unit displacement applied in turn. These moments or forces
are functions of the structural dimensions and material
properties only.

66 66 At this point in the explanation, it is useful to summarise

and compare the analytical procedure for both the stiffness
and flexibility methods. The output of the flexibility
method is the unknown forces, whilst the stiffness method
produces deformations. You should remember that the
analytical results from each are identical. Neither method
of analysis is preferable on theoretical grounds.

67 67 We will now carry out a quantitative analysis of the

previous example, and, for simplicity, we will assume that
I values are constant throughout and the lengths of beams
and columns are the same.

68 68 We may now draw the artificial restraining moments

at joints 1, 2 and 3. The fixed-end moments due to a
uniformly distributed load are W L2 / 12. Note that for this
particular loading arrangement the restraining or clamping
moment at joint 2 is zero since the fixed-end moments from
members either side of the joint have the same value.
 The Stiffness Method – Frames 133

69 The artificial horizontal reaction at 4 is equal to half of 69

the horizontally applied uniformly distributed load. At
this stage in the analytical procedure joint 2 is clamped.
Consequently the distribution of reaction forces and
bending moments is that for a fixed-end member.

70 The next step in the analytical procedure is the application 70

of the absolute moments in turn. The bending moment
required to induce a deformation of 1 radian at joint 1 is
the sum of moments required to produce a rotation of 1
radian in members 15 and 12. There is a carry-over to
joint 2, r21. No bending moments are induced at joint 3,
consequently r31 = 0. r41, as above, is 6 E I / L2 .

71 This figure shows the bending moments and the horizontal 71

restraint at 4 due to the application of a rotation of 1
radian at joint 2. By applying the Theorem of Reciprocal
Deflections, we can equate r12 and r21.

72 Bending moments and horizontal restraint at 4 due to the 72

unit rotation at 3 are illustrated.
134 Understanding Structural Analysis

73 73 Bending moments due to the unit deflection at joint 4 are

shown.

The value of r44 is that force required to produce a unit

displacement at the end of members 15, 26 and 37. The
expression for the value of that force was developed for
Figures 20–26. The reaction resulting from the settlement
force equals 2 M / L (Figure 21) which substituting for M
from Figure 26 equals 12 E I / L3 .

74 74 These values of the bending moments and forces due to

unit actions are the elements of the stiffness matrix. As
this is a particularly crucial stage in the analysis, check
each element carefully against Figure 64 above.

75 75 The elements in the column, the right-hand side of the

expression are the restraining moments and reaction due to
the applied loads. This set of four simultaneous equations
would be solved for θ1, θ2, θ3 and θ4. These deformations
would then be substituted back into the expressions for
the bending moments at each of the points of kinematic
indeterminacy in the structure.

76 76 At this stage in the explanation of the stiffness method

it will be simpler if we use a numeric solution. We will
substitute loading and dimensions for a three-bay frame.
The EI value is assumed to be 50 MNm2 units.
 The Stiffness Method – Frames 135

77 We are now able to set up the stiffness matrix by reference 77

to Figure 74 above. The units are MNm.

78 The solution of the four simultaneous equations gives 78

us the values of rotations at nodes 1, 2 and 3 and the
horizontal deformation of node 4. Note particularly that
the sign of the solution to each of the displacements is
a confirmation or denial of the arbitrary choice of the
direction of the unit rotations at 1, 2 and 3 and the unit
displacement at 4. Thus the rotation at joint 2 is actually
in the anti-clockwise direction, consequently the solution is
negative.

79 It is interesting to compare these results with those 79

obtained from a computer analysis. The output of nodal
joint displacements shown is taken from a computer
program based on the stiffness method that takes into
account the axial strains. You will notice that there is a
difference between the hand and the computer analysis
about the second or third significant figure of the solution
for the deformations. This difference is due to the fact
that the axial strains have been ignored in the hand
calculations.

80 We can virtually eliminate the difference between the 80

hand and the computer analysis by inputting an artificially
large value of the area of all the members. This computer
solution is based upon a member area of 150000 x 103
mm2 compared with those shown in Figure 79 above which
is based upon an area of 150 x 103 mm2. As you will see,
apart from the difference of 1 in the fourth significant figure
for the horizontal deformation at computer node 4, the
manual and computer solutions now agree.
136 Understanding Structural Analysis

81 81 The final distribution of bending moments is a result

of combining the clamp moments with those from the
application of the real deformations. We will look at the
distribution of bending moments in member 15 to clarify
the procedure. The final distribution of bending moments is
the algebraic sum of

1. restrained moments,

2. clockwise rotation at node 1,

3. horizontal displacement.

82 82 This figure shows the final distribution of bending moments

in member 15 with the effect of the free bending moments
resulting from the application of the uniformly distributed
load, superimposed on the ‘restraint’ bending moments.

Practice Problems
You should now practise the application of the stiffness
method of analysis to numerical problems. As with the
previous chapters you should check your solution against a
computer solution.
8 The Stiffness Method – Grids

So far in this text, we have been almost exclusively concerned with analysis of two-
dimensional frames, set in the vertical XZ plane. All real structures are, of course,
three-dimensional. However, most structural frames, particularly those for multi-storey
structures, are resolved into a series of sub-frames that are set in this two-dimensional
XZ plane. Consequently, this academic emphasis is a genuine reflection of the day-to-
day activities of the structural engineer. Occasionally though, it is necessary to design a
structure capable of carrying a load normal to the structural plane. This is defined as a
grid. The same basic theoretical approach is employed, the Stiffness Method.
138 Understanding Structural Analysis

1 1 We have studied the analysis of this cantilever bent ABC,

loaded at C from a number of different aspects.

2 2 If we now introduce an additional member, CD, in the

third dimension Y, a new stress resultant is introduced,
the torsion in member BC. That is the most significant
difference between plane frame and grid, the torsional
stiffness of the member is normally included in the
resistance of the grid structure. Consequently, there is a
rotation about the longitudinal (x) axis of the member due
to the torsional effects.

3 3 The stress resultants of axial, shear force, bending moment

and torsion are identified in relation to the member, rather
than the global axes. The member axes will be identified
with lower case symbols. There is the bending moment
about the major axis My, This is the same as the bending
moment in the analysis of plane frames. There is a shear
force in the z direction and a new load effect, i.e. the
torsion about the x axis, Mx.

4 4 We will now study the qualitative analysis of this cantilever

bent frame, ABCD. The vertical load at D produces a
bending moment at C, My, that is about the member y axis.
In addition, there must be an upward shear force at C in
the member z axis resisting the downward load at D.
 The Stiffness Method – Grids 139

5 The shear force in CD at C is equal to the load W. That 5

shear force is transmitted to member BC as a point load
at C that in turn produces a bending moment about the
Y axis of member BC. The shear force is constant in
members CD and BC. The bending moment in CD at C,
My, is resisted by the torsion in BC, which is constant
throughout the member.

6 At joint B, the moment about the y axis in member BC, 6

My is transmitted into the column about its y axis. The
shear force at B, is resisted by the axial force in AB. The
torsion in BC, Mx, is resisted by the bending moment in the
column AB about the z axis of that member.

7 This figure shows the final shear force and axial load 7
diagrams. There are no axial forces in either BC or CD
since the loading is normal to these members.

8 The moment reactions at A resist the My and Mz moments 8

in column AB. There is no torsion moment in AB since the
only load on the structure is parallel to the axis of AB. For
there to be a torsion in AB, there would have to be a load
at an angle to the global z axis.
140 Understanding Structural Analysis

9 9 The deflected shape may be built up by a consideration of

the resulting extension of the fibres in bending tension.
From a vertical start at A the bending tension will deflect
the column AB in the positive direction of both X and Y
axes. Since there is no bending about the z axis of BC or
rotation of AB, point C will deflect directly downwards
in a plane parallel to the XZ plane. Similarly, there is no
horizontal deflection of point D, other than that induced by
the horizontal deflection of point B.

10 10 Practice Problems
The example above and these problems are true three-
dimensional structures and not simple grids that are the
subject of this chapter. However, it is this new torsional
stress resultant which will present problems in developing
an understanding of the behaviour of grids. Practice in the
qualitative analysis of those three-dimensional cantilevers
will help to prepare you for the analytical procedure that
follows. Draw the qualitative diagrams for each of the
stress resultants.

11 11 The stiffness method may be applied with equal facility to

the analysis of grids. We will develop the stiffness approach
with the frame 12345, set in the horizontal XY plane.
This grid is fully fixed at supports 2 and 3 and pinned at 5.
Member 14 is a cantilever loaded vertically, that is parallel
to the Z axis at 1.

12 12 The presentation will be clearer if the structure is reduced

to a line diagram, however you should always bear in mind
that the structure has a three-dimensional reality. This is
particularly helpful when considering torsion effects.
 The Stiffness Method – Grids 141

13 There are three degrees of freedom, three degrees of 13

kinematical indeterminacy. The rotation of node 4 in the x
and y direction and the vertical deflection of node 4. If we
look at member 245 in elevation there will be a vertical
deflection at node 4, ∆4 , and a rotation at node 4, θ4
z y
about the y axis. This is likely to be in the clockwise (i.e.
positive) direction. The axes shown are the global axes and
the deformation will be determined relative to these axes
and not member axes.

13 Taking an elevation of member 143 the rotation at node 14

4 about the x axis, θ4x, is certain to be in an anticlockwise
direction.

15 The first step in the stiffness analysis of grids is the 15

same as that for plane frames, the notional restraint and
identification of the degrees of freedom. The kinematic
indeterminacies are given a numerical reference.

1 is the rotation about the Y axis,

2 is the rotation about the x axis,

3 is the vertical displacement.

The symbol ‘R’ is the restraint required at the clamp to

prevent. rotation and deflection at node 4.

16

16 With node 4 clamped against rotation and vertical

displacement, the point load W causes a clamp moment R2
and a clamp force R3. This particular loading arrangement
does not induce a clamp moment in direction 1 about the y
axis.
142 Understanding Structural Analysis

17 17 The next stage in the analytical procedure is the

application of moments and forces in the direction of the
degrees of kinematic indeterminacy to induce either a unit
rotation or unit deflection, as appropriate. This ‘unlocks’
the clamp in direction 1, but leaves the clamp in directions
2 and 3. r11 is the moment required to cause a rotation of 1
radian in direction 1, rotation about the y axis.

18 18 For the first time in the application of this method of

analysis we must take into account the torsional resistance
of a member, here member 34. This is the diagram for the
bending moment and torsion combined. Note that although
member 14 rotates under the action of r11, it has no
stiffness as it is a cantilever. Consequently, this member
cannot sustain a torsional moment. The application of
moment r11, induces a reaction in direction 3, r31, but
has no effect in direction 2, rotation about the x axis.
Consequently r21 is zero.

19 19 This figure shows r22, the moment required to induce a

rotation of 1 radian at node 4 in direction 2, that is about
the x axis.

20 20 Member 14 will rotate freely and since the far end of

member 45 is pinned (i.e. free to rotate), this member has
no torsional stiffness. r22 will be the sum of the moments
required to rotate member 24 about its longitudinal axis,
that is in torsion and rotate member 43 in bending. r12 is
zero.
 The Stiffness Method – Grids 143

21 r33 is the force required to cause a unit displacement at 21

node 4 in the z direction. This produces bending moments
in members 24, 45 and 34. Remember that only the
vertical restraint has been removed leaving the rotational
restraints, the notional ‘clamps’ in directions 1 and 2. This
upward movement does not create torsion in any member
since the members of the grid are at right-angles to one
another.

22 We will now produce a quantitative solution by providing a 22

value for the point load and dimensions for the structural
members in the grid. EI is constant for all members at 40
MNm2 units. The product of the torsional constant J and
the modulus of rigidity G is 50 MNm2 units.

23 The first step in the analytical procedure is to clamp the 23

free node, node 4. The resulting clamp resistance will be:

R2, about the x axis = −250 kNm,

and

R3, the vertical reaction = +50 kN.

Note that the signs are in relation to the overall global

axes. Note also that the effect of the clamp on member
143 at 4 is to isolate 43, as if 14 were a cantilever, fully
fixed at 4.
144 Understanding Structural Analysis

24 24 We now apply the unit actions in the directions of the

degree of kinematical indeterminacy, the degrees of
freedom. Remember that directions 2 and 3 are clamped
when r11 is applied.

r11 is the moment required to induce a rotation of 1 radian

about the y axis. Consequently, each member at joint 4
must rotate 1 radian. r11 will now include the torsional
resistance of member 43. r11 is applied in the positive
direction about the y axis. The torsional rotation of a
straight bar = J G / L where J is the polar moment of
inertia of the section and G is the modulus of rigidity. Refer
to any standard text on strength of materials for the proof
of this relationship.
25

r21 would be the result in direction 2 of the application of

r11. r11 has no effect in direction 2. Consequently r21 is
zero.

25 The effect of this rotation of 1 radian in direction 1 has an

effect on the vertical ‘clamp’, r31, This reaction is positive
for the bending moments induced in member 42 and
negative for the bending moments in 45. Refer back to
Figure 18, Chapter 7, for proof of this relationship.

36 26 The moment r22 will produce a rotation of 1 radian in

direction 2.

Member 45 has no torsional resistance as the far end 5 is

pinned. There is, however, a torsional resistance in 24. r22
has no effect in direction 1 therefore r12 is zero.

27 27 This figure shows the effect of the unit rotation in direction

2 which is to produce an upward reaction in direction 3,
r32.
 The Stiffness Method – Grids 145

28 The force r33 is that required to induce a vertical deflection 28

of unity. Remember that joint 4 is clamped against rotation
in directions 1 and 2. r33 is a ‘settlement force’. See
Chapter 7, Figures 20–26 for proof of the relationship
between the deflection, bending moments induced and the
value of the force required. For a member with both ends
fixed, the force to induce a relative settlement of unity is
12 E I / L3. If the far end is pinned, the force is 3 E I / L3.
The value of r33 is 27.08 MN. The associated effects of
inducing an upward deformation of unity in the two other
directions do not need to be calculated. Clerk-Maxwell’s
Theorem of Reciprocal Deflections provides them from
previously determined values.
29
29 We may now set up the grid stiffness matrix, unknown
rotations and clamp forces.

30 The values of the elements of the stiffness matrix are 30

now substituted from the calculations above. Note that
the calculated clamp forces are in kN units whereas the
stiffness elements are in MN units. Therefore the clamp
forces have been divided by 1000.

31 The solutions to this set of simultaneous equations are 31

the three unknown deformations, θ1, θ2 and ∆3. Note
the significance of the sign of the deformations. The
directions of application of the actions r11, r12 and r33
are in the positive direction in each case. The positive
sign in directions 1 and 2 means that these are actual
directions of the deformations. Clearly the deflection at 4 is
downwards, consequently the sign is negative.
146 Understanding Structural Analysis

32 32 Always draw the final solution for the deformed structure.

It will be apparent if there is a gross error in the sign or
value of the deformations. You will find when you practise
the solution of similar problems that the solution of the
simultaneous equations seems to be a frequent source
of error. A sketch of the solution at this stage may help
to avoid unnecessary calculations for the distribution of
bending moment and torsions.

33 33 The solution is completed with the substitution of the

deformations into the expression for the bending moments
and torsions for each member. We will look first at member
245. The effect of rotation and deformation have been
separated. Remember that there is a carry-over factor of ½
for the rotation and 1 for the differential displacement.

34 34 This figure shows the effect of rotation and settlement in

member 45.

35 35 This is the final distribution of bending moments in member

245. The difference between the bending moments at 4 is
accounted for by the torsion in member 43. The difference
of 1 kNm is due to errors in rounding up.
 The Stiffness Method – Grids 147

36 This figure shows deformations in member 143. Bending 36

moments in member 14 are unaffected by the deformations
of the grid, as it is a cantilever.

37 The figure shows the separate effects of rotation 37

and deflection in member 43. The plotting of these
diagrams should always be related back to the structural
deformations to check that they are drawn on the tension
side of the structure. For example, the clockwise rotation
of 0.006 radians will produce bending tension on the top of
member 43 at node 4. Similarly, the downward deflection
at node 4 will produce bending tension on the underside of
member 43 at node 4.

38 The final bending moments in member 143 and the 38

balancing torsion in member 24.

39 The complete diagram of bending moments, that is bending 39

moments about the y member axis, of each member.
148 Understanding Structural Analysis

40 40 Torsion moments.

41 41 Practice Problems
Carry out a full qualitative analysis for the diagrams that
are relevant to each stage of the analytical procedure. Then
complete the numerical analysis and compare it with a
computer solution.
9 Moment Distribution

The method of Moment Distribution was developed by the famous American engineer
Hardy Cross and was known for some time as the Hardy Cross Method.

The impetus for the development of this analytical method was the increasing use of
multi-storey reinforced concrete frames in the 1930s. Prior to that time, large, complex
structures had been built in structural steelwork. The nature of the joints in steelwork
and the ease with which effective and reliable hinges may be introduced meant that the
real structure could be resolved into a series of statically determinate sub-systems with
relative confidence. In reinforced concrete, however, the continuity which creates the
problem for the analyst is inherent in the system and to destroy it, as designers using
precast systems tend to do, is to reduce substantially the real, uncalculated strength of
the structure. Consequently, the development of modern structural analysis was born out
of the practical need of the structural designer.

The method of Moment Distribution is based on the relationship between deformation and
moments that was developed in the earlier chapter on the stiffness method. Indeed the
method of moment distribution is the iterative version of the stiffness method. They are
quite different, though, in practical use, in that moment distribution is a realistic design
office method of analysis whereas the structural designer used the Stiffness Method via
the medium of the computer.
150 Understanding Structural Analysis

1 1 In the stiffness method of analysis explained in the

previous chapter the Theorems of Virtual Work were used
to develop the relationship between the effects of a unit
rotation or displacement and the resulting distribution of
bending moments and reactions. In this chapter we will
employ an alternative approach using Mohr’s Moment-Area
method. The proof of this method may be found in any book
on structural mechanics. There are two theorems - the first
is that the change of rotation between any two tangents
to the curve of a deflected beam is the area of the M / E I
diagram between the two points under consideration. In
this diagram the rotation θAB is the change in rotation
between the tangents at points A and B. The bending
moment diagram is shown below and the relevant portion
of the M / E I diagram in the lower part of this figure. We
will distinguish between the bending moment and M / E I
diagrams by using chain line for the latter.

2 2 The second theorem is that the deflection between one

point, here point B and the tangent to another point, here
the tangent passing through point A, is equal to the first
moment of the area of the M / E I diagram between these
two points, taking point B as the point of reference for the
first moment. It is convenient to reduce the M / E I diagram
to simple geometric shapes, here two triangles, and the
lower diagram shows the dimensions relevant to the first
moment of area of the M / E I diagram.
 Moment Distribution 151

3 Unless you particularly want to study the development 3

of these relationships using the moment-area method,
it is unnecessary for those students who have studied
the stiffness method above to repeat the proof again.
The moment-area proof has been included here for those
students studying structural analysis on certain courses
who would not normally need to study the Theorems of
Virtual Work. We are going to find the relationship between
the moment applied at the pinned end of this propped
cantilever AB, which produces a rotation of 1 radian at
point B. This moment is defined as MAb, i.e. the ‘absolute’
moment, to distinguish it from the normal load moments. It
is apparent from the deflected shape of the structure that
the application of the moment at B results in a hogging
4
moment at A. Note that the rotation of 1 radian is an
analytical device. No real structure could sustain so large a
rotation.

4 We will use the flexibility method of analysis, sometimes

known as the unit force method, to solve this structure and
the first step in that method is to release the structure to
reduce it to a statically determinate form. This we have
done by removing the vertical reaction at B. The effect
of the application of the absolute moment is to cause an
upward deflection at B. The bending moment diagram is
5
shown below.

5 We now apply a unit load in the same direction as the

removed restraint. The resulting deflection is δB. Note the
value of the bending moment at A is L kNm.

6 Now if we return to our original problem of the propped 6

cantilever we can see that the application of the absolute
moment produces a downward reaction at B which must
be balanced by an upward reaction at A. The reaction at B,
VB, is the object of this analytical procedure.
152 Understanding Structural Analysis

7 7 We can set up an equation of compatibility for the

displacement at B. The deflection at B due to the applied
absolute moment, ∆B + the real value of the reaction, VB
multiplied by the deflection due to the unit load must be
equal to zero because the actual deflection at B of the real
propped cantilever is zero.

8 8 Mohr’s methods will now be used to find the values of

the displacements in the equation of compatibility above.
Using Mohr II, the displacement between the tangents
at A and the tangent at B of the released cantilever AB
is in fact the vertical deflection at B. This is a particular
case for the application of this method since the tangent at
A is horizontal, i.e. coincident with the original unloaded
position of the cantilever. Mohr II is the first moment of
area of the M / E I diagram about B. The deflection at B is
equal to the absolute moment multiplied by L2 / 2 E I .

9 9 Similarly, we can find the vertical deflection at B due to

the application of a unit load at B. The result is −L3 / 3 E I .
The signs for the deflections are relative to the global
convention previously established.

10 10 Setting up and solving the equation of compatibility gives

us the value for the vertical reaction at B which is equal to
3 MAb / 2 L . A positive sign for the solution of this equation
indicates that the choice of the direction of the unit load in
Figures 5 and 9 above was correct.
 Moment Distribution 153

11 We now substitute the value for the vertical reaction at 11

B and determine the value of the bending moment at A
which equals −MAb / 2. The negative sign indicates that it
is acting in the anticlockwise direction, i.e. the action of the
joint on the member.

12 The result of this analysis is to prove the first of a series 12

of important relationships which is that the carry-over of
moment to the far end of a propped cantilever for a moment
applied at the pinned (near) end is equal to a half.

13 This distribution of moments on the propped cantilever 13

is now slightly redrawn for convenience for solution by
the moment-area method, by separating the hogging and
sagging moments. The rotation at B, which we know to
be 1 radian, is equal to the area of the M / E I diagram
between AB. This rotation at B is equal to
MAb x L / 4 E I .

14 This provides the second and most important relationship 14

which is that the value of the bending moment required to
induce a rotation of 1 radian at the pinned end of a propped
cantilever is equal to 4 E I / L .
154 Understanding Structural Analysis

15 15 Similarly, it may be shown that the absolute moment

required to induce a rotation of 1 radian in a simply
supported beam is equal to 3 E I / L . Because the far end of
the beam is pinned, clearly the carry-over factor is zero.

16 16 One last relationship needs to be determined. That is the

relationship between the bending moment induced in a
beam, both ends of which are fixed, and the displacement
deflection at one end relative to the other. From the
deflected shape it is apparent that there will be a ‘hogging’
bending moment at A and a ‘sagging’ bending moment at
B.

17 17 Redrawing the diagram to produce the M / E I diagram,

the deflection at B is equal to the distance between the
tangents of the two relative points. Here the tangent at
B would be horizontal to the tangent at A. Nevertheless,
Mohr’s theorem still holds, and the deflection, which would
be an application of Mohr II, is equal to the first moment of
the M / E I diagram, about B. The resulting relationship is
that the deflection is equal to M L2 / 6 E I .

18 18 This gives us the last and most important fundamental

relationship between section properties and displacement
which is that for a relative displacement of ∆ at the end of
the fixed beam, the moment induced is equal to 6 E I ∆ / L2 .
 Moment Distribution 155

19 If we now analyse the beam, knowing the value of the 19

fixed-end moments, we find that the reactions are equal to
12 E I ∆ / L3.

20 We will now develop the method of moment distribution 20

using the relationships between rotation, displacement and
induced bending moments shown above. We will consider
first the four-member frame ABCDE with a clockwise
moment applied at the centre node B.

21 This moment would cause node B, and the four members 21

connected to it, to rotate in a clockwise direction, θB. Note
that, because the connection between the four members is
considered to be rigid, the rotation of each of the members
at B is θB.

22 The figure shows the resulting qualitative bending moment 22

diagram. From relationships developed above we know that
the carry-over of bending moment to the fully-fixed end of
each member is half that at the end B.
156 Understanding Structural Analysis

23 23 We now assume that the moment at B is that which causes

a rotation of 1 radian, that is the moment, MAb as defined
above.

24 24 The value of MAb is the sum of the absolute stiffnesses

of each of the members framing into joint B. The actual
rotation θB will be equal to the actual moment MB, divided
by the total absolute stiffness at the joint.

25 25 The proportion of the moment MB, induced into the end of

a particular member is the actual rotation θB multiplied
by the stiffness, i.e. 4 E I / L for that particular member.
Substituting for θB in the expression above...

26 26 ...we find that the distribution of bending moment into any

member is the total moment MB multiplied by the absolute
stiffness of the particular member divided by the total
absolute stiffness of all members at that joint.
 Moment Distribution 157

27 The fraction ( 4 E I / L ) / ( Σ 4 E I / L ) member is known as 27

the distribution factor. This may be simplified by cancelling
‘4 E ’ from the expression. Thus it is only necessary to
determine I / L for each member. This relationship is known
as the ‘relative stiffness’. Since, in the method of moment
distribution it is only ever used to determine distribution
factors, the I value too may be relative.

28 It is customary for theoretical problems to be set in terms 28

of the relative I values or relative stiffness. In real problems
of course, the I value must be calculated. We will extend
our qualitative approach by a numerical solution. The
circled values are the relative I values.

29 The first step in the moment distribution procedure is 29

to determine the value of the relative stiffness of each
member. The symbol K is used here. The reader should note
that there is no generally accepted symbol for stiffness.
Sometimes ‘S’ is used. The total relative value of the
stiffnesses of all the members joining into joint B is 6.63.

30 We will use the symbol ‘a’ to define the distribution 30

factor for any member. This factor is the proportion of
any moment applied to the joint that will be taken by the
particular member. It is the stiffness, or relative stiffness,
of the particular member at a joint divided by the total
of the stiffness of all members at that joint. For member
AB, the distribution factor is 0.18. Therefore the bending
moment at end B of member AB is 0.18 x the applied
moment M.
158 Understanding Structural Analysis

31 31 The analytical procedure is directly analogous to the

stiffness method in that the first step in the actual moment
distribution procedure is the ‘clamping’ of the joints.
Therefore we need to calculate the fixed-end moments for
each member. For the time being, these two standard cases
will suffice.

32 32 We will apply the method of moment distribution to the

two-span beam ABC. The I value is the same for both
spans. The method proceeds by examining the difference in
bending moments at supported joints and distributing the
difference in moments through the structure by a series of
iterations.

33 33 The joints that are free to rotate are notionally fixed with
an analytical clamp, including the pinned joint at C. The
real loads are applied and the resulting fixed-end bending
moments evaluated. Note that the artificial clamp maintains
the difference between the bending moments at joint B. We
must now correct the fixed-end moments to match the real
support conditions of the beam.

34 34 Node C is actually pinned, i.e. zero moment. We can release

this joint by applying an equal and opposite moment,
here the anticlockwise moment of 32 kNm. This moment
applied at C is ‘carried-over’ to the ‘far’ end at B. Note the
direction of the bending moment in the diagram. Both are
drawn as anticlockwise, that is ‘the action of the joint on
the member’.
 Moment Distribution 159

35 The final distribution of bending moments in BC is the 35

sum of the ‘clamped’ condition and the ‘release’ at C. This
results in a hogging moment of 48 kNm at B. At C, of
course, the moment is zero.

36 This roundabout procedure avoids the necessity of 36

remembering the expression for fixed-end moments for a
propped cantilever. Had we known this relationship, we
could have found the ‘clamp’ moment at B for member BC
directly.

37 The routine above is particularly useful for members with 37

unsymmetrical point loading. The fixed-end or ‘clamped’
moments are determined as shown.

38 Then the release of the pin-joint is made to carry-over half 38

of the moment to the far end.
160 Understanding Structural Analysis

39 39 The final fixed-end moment at the ‘clamped’ support is

the sum of the ‘clamped’ and ‘released’ sets of bending
moments.

40 40 To return to the original problem, we now have the

distribution of bending moments assuming only joint B is
clamped. The clamp moment at B required to maintain this
condition is 29.2 kNm anticlockwise.

41 41 If the far end of a member is pinned the absolute stiffness

is 3 E I / L. Therefore, since ‘4E’ is cancelled out to
determine the ‘relative’ stiffness, the relative stiffness of a
member, the far end of which is pinned, is ¾ I / L.

42 42 The relative stiffness of each member is calculated as

follows:

KAB = l/6

KBC = I/8 x 3/4

because the far end (C) is pinned. Since I is the same

for each span in this particular problem, the distribution
factors are 0.64 and 0.36 for BA and BC respectively.
 Moment Distribution 161

43 The next step in the procedure is to eliminate the clamp 43

moment by applying an equal and opposite moment at
B. This moment is distributed in accordance with the
distribution factors.

44 The final distribution of bending moments is the sums of 44

the clamped and distributed conditions. The reader should
note that:

1. The approximate nature of all structural calculations

means that it is unnecessary to express the
distribution factors to more than two decimal places.

2. The distribution factors must always sum to 1.00 just

as the distributed moments 10.5 and 18.7 must be
numerically equal to the balancing moment.

45 Moment distribution is a tabular method. The diagrams 45

above would be summarised in this form. Study this
table carefully and relate it to the diagrams. Note that it
is convenient to release the pinned support at C before
starting the full moment distribution procedure, since this
‘matches’ the fixed-end moments with the true structural
support system.

46 If we now compare the final bending moments with the 46

deflected shape, the points of contraflexure on the deflected
shape may be checked against zero bending moments and
the overall shape of the tension zones checked visually.
162 Understanding Structural Analysis

47 47 Generally, moment distribution is not completed with

only one distribution of the out-of-balance moments. This
three-span beam will be used to illustrate the more usual
procedure. Span CD has the same section properties
as spans AB and BC. The relative stiffnesses, K and
distribution factors, a are determined.

48 48 The next step is to identify the clamp moments. Remember

that these are the difference between the fixed-end
moments at each support. The correct identification of the
direction is crucial. Always carry out a qualitative check.

49 49 Each clamped joint is released, one at a time, with the

balancing moment. The joint is released by the application
of a moment of equal value and in the opposite direction
to the clamp moment. Note particularly the carry-over to
support C of 2.8 kNm, half of the value of the balancing
moment distributed into span BC at B. Both the
distributed and the carry-over moments should be rounded
off to one decimal place. Remember though, that the sum of
the distributed moments must equal the balancing moment.

50 50 This next figure shows the release of joint C by the

application of the balancing moment of 32 kNm in an
anticlockwise direction. Support D is pinned; consequently
there is no carry-over moment to D. Note again, the carry-
over from joint C of 8 kNm to joint B.
 Moment Distribution 163

51 The carry-over of moments to B and C from the release of 51

these joints has unbalanced the system again and requires
external clamp moments to sustain equilibrium.

52 Here is the procedure so far, shown in the more usual 52

moment distribution table. The clamp moments at B and
C must now be released and distributed. Because joint A
is fully fixed in the original structure it ‘receives’ moments
but does not release or redistribute them.

53 It is now apparent that the procedure is theoretically 53

endless, as there will always be a carry-over moment to
unbalance joints B and C. Practically, it is usual to stop
the distribution where the largest value being distributed
is about 5% of the original fixed-end moments, but there
are no hard and fast rules. Note that the values either side
of the support reference lines are the algebraic sum of all
the moments to that side of the support, both clamp and
distributed moments. The bracketed figures that are the
balancing moments are not included.

54 The final bending moment diagram is drawn, first 54

by plotting the support bending moment and then
superimposing the ‘free’ span bending moments. The
next step in the design office would be to determine the
shear forces. This is dealt with here by the method most
often used by the practising engineer. The first step is the
notional release of the moments at each joint, turning the
continuous beam into a series of simply supported spans,
with imposed end moments.
164 Understanding Structural Analysis

55 55 Firstly, we treat the reactions to the loading as that

resulting from the analysis of a series of simply supported
spans.

56 56 We will look at the span AB. The simply supported

span has moments applied at each end that modify
the distribution of shear forces and reactions. Taking
moments about A there is an out-of-balance moment, the
difference between 12.4 and 31.5, of 19.1 kNm which
is anticlockwise. This must be balanced, for moment
equilibrium, by a reaction ‘couple’. This couple is the shear
due to the effect of the restraint moments and is known as
the ‘elastic shear’. The value of the elastic shear is the out-
of-balance moment divided by the span.

57 57 We then determine the value of the elastic shears on each

span. Note particularly that the direction of the elastic
shears must be checked qualitatively.

58 58 The final true shears are the sum of the simply supported
shears and the elastic shears. The algebraic sum of these
shears at each support is the final support reaction.
The diagram of reactions is completed with the moment
reaction of 12.4 kNm at A. Always check that the applied
loads are exactly balanced by the reactions.
 Moment Distribution 165

59 The settlement of beams may be included in the moment 59

distribution procedure. Here there has been a vertical
settlement at C downwards.

The qualitative distribution of bending moments is

shown below. This is always a good idea in any analytical
procedure as it serves as a check on the final numerical
solution. The bending moments caused by the settlement
are dealt with in the moment distribution procedure in
exactly the same way as the bending moments resulting
from normal vertical loads.

60 This relationship was developed in the chapter on the 60

stiffness method, using the Theorem of Virtual Work and
earlier in this chapter using Mohr’s moment-area methods.
The bending moment created at the supports is equal to
6 E I ∆ / L2.

61 This figure shows the qualitative fixed-end moments due to 61

settlement. Remember that the joints are prevented from
rotating by the external ‘notional’ clamps. If settlement is
combined with vertical loading, these moments are added
algebraically to the fixed-end moments resulting from the
vertical loads prior to the start of the distribution of out-of-
balance moments.

62 We will now return to the problem of the four-member 62

frame we started with and look at a more usual loading
arrangement. Span AB has a uniformly distributed load
and BC an unsymmetrically placed point load.
166 Understanding Structural Analysis

63 63 The distribution factors were determined previously. The

next step in the procedure is to determine the fixed-end
moments and from them, the ‘clamp’ force at B required to
maintain equilibrium. This is 60.5 kNm in an anticlockwise
direction, the difference between the fixed-end moments in
members AB and BC at B.

64 64 The tabular layout is a little more complicated for two-

dimensional structures. It is usual to set out the horizontal
members on the left-hand part of the table and other
members on the right. Since there is only one clamped
joint, the distribution of the balancing moment completes
the moment distribution procedure. Always check that the
algebraic sum of moments at a joint is zero.

65 65 The figure shows the final distribution of bending moments.

The sagging moments may be determined graphically by
drawing this diagram to scale and superimposing the ‘free’
span bending moments onto the support moments.

66 66 Alternatively, we can isolate a particular member and

find the point of zero shear and calculate the value of the
maximum sagging moment.
 Moment Distribution 167

67 We will now examine the application of the method to two- 67

dimensional structures that are free to sway. To begin with
we will determine the qualitative solution. Because the
horizontal reactions must be in equilibrium, HA = HD.

68 This figure shows the qualitative solution for the bending 68

moment diagram and the deflected shape. The value of the
bending moments at B and C must be the same, because
for this particular structure HA = HD and the heights of the
columns are the same. Note that the structure will sway to
the left. For a more detailed discussion of this phenomenon
see Chapter 4, Figure 30 onwards.

69 In the moment distribution procedure, the sway condition 69

must be examined separately and so we must prevent that
sway from occurring by the introduction of an artificial
restraint, here a horizontal reaction at C. The joints are
prevented from rotating by notional clamps and the fixed-
end moments are calculated.

70 The columns are equivalent to supports and the moment 70

distribution is conducted as if the frame were a three-span
beam.
168 Understanding Structural Analysis

71 71 It is particularly important to understand the physical

response of the structure to the application of the balancing
moment at B and C, here applied separately. Note the
unusual bending moment diagram that results from the
application of a load moment at a joint.

72 72 This figure shows the final distribution of bending moments

and deflected shape for the restrained frame.

73 73 The value of the horizontal reactions at A and D is equal

to the moments at B and C divided by the height of the
columns. The difference between these reactions is the
balancing horizontal artificial restraint at C, here 0.15 kN,
to the right. This artificial reaction is identified at Q0 .

74 To eliminate this restraint we must apply an equal and

74 opposite force, Q1. However, we cannot determine the
value of the distribution of fixed-end moments directly, so
we use our knowledge of what are, in effect, settlement
moments. These moments are calculated in terms of an
arbitrary horizontal deflection ∆. We have no other way of
finding the set of fixed-end moments with which to start
the moment distribution procedure. Joints C and D are
notionally clamped against rotation. An arbitrary value of
the fixed-end moment is assumed of the same order as the
original fixed end moments. This is expressed in terms of the
arbitrary deflection ∆.
 Moment Distribution 169

75 Note carefully the sign of the sway moments in the 75

distribution table. This is a frequent source of error.
Always check the sign very carefully against the diagram
of the deflected shape of the clamped structure. Since the
structure is symmetrical the moment distribution for only
one column has been completed.

76 The result of this analysis is a distribution of bending 76

moments that are a function of the arbitrary deflection ∆.

77 From the bending moments at B and C the horizontal 77

reactions at A and D may be found and the value, still
in terms of the arbitrary deflection ∆, of the force at C,
Q1. We know that the value of the force to correct the
restrained frame is 0.15 kN. The correcting factor to
be applied to the sway moments is the real value of the
artificial restraint divided by the equivalent horizontal force
resulting from the arbitrary horizontal displacement. This
factor is then applied to the arbitrary sway moments, thus
eliminating ∆.

78 The final solution to the distribution of bending moments 78

is the sum of the restrained and sway distributions. The
value of 0.67 kNm for the sway moments at B and C is the
result of multiplying the arbitrary sway moment of 4.2 ∆ by
the correcting factor of 0.16 / ∆.
170 Understanding Structural Analysis

79 79 This figure shows the final distribution of bending

moments. Note the difference between the moments at B
and C. These should be the same value. The difference
is due to the rounding off in the moment distribution
procedure.

80 80 It is perhaps worth explaining the way in which

the arbitrary sway moments are determined for an
unsymmetrical frame. Frame ABCD is pinned at A and
fully fixed at D. Under the imposition of a sway of ∆, the
moment induced at B is 3 E I1 ∆ / L12 a function of the
properties of column AB. Similarly, column CD fixed at
D has a distribution of moments which are a function of
the properties of CD. The fixed-end moments used in the
distribution table must be in these proportions.

81 81 Frames with sloping members present a further

complication. When the arbitrary sway is applied the
effective ‘settlement’ of the sloping member is different
from the value of the horizontal deformation, because of the
frame geometry.

82 82 In addition, the horizontal deflection at C imposes a

bending moment in the beam BC. The effective deflection
for member AB is ∆BA, i.e. normal to member AB at B, a
deflection that is greater than the horizontal deflection ∆ at
C.
 Moment Distribution 171

83 We will study the analysis of this frame by examining 83

the analysis of the sway moments since the restrained
moments are distributed as if the beam BC were supported
at B. The first step is to prevent the rotation of the joints
for the calculation of the fixed-end moments due to the
sway that will be a function of the arbitrary deformation ∆.
This figure shows the deflected shape for sway to the right.
Remember that we are assuming that any shortening of
member BC is insignificantly small and so we can assume
that the horizontal deflection at C is the same as the
horizontal deflection at B.

84 It is clear now that the deflection ∆AB is a function of 84

the geometry of the frame. We assume that the leg AB
rotates about A and that the deflection is sufficiently small
for the overall geometry to be unaffected. The triangle
of deformations ∆, ∆BC, ∆AB is similar to the frame
dimensions ZAB, XAB and LAB, where the latter is the actual
length of member AB.

85 The fixed-end moments in member AB are based on ∆AB 85

and the length of the member, LAB.

86 The other major difference in the analysis of frames with 86

sloping members is that horizontal sway induces vertical
deflection. The horizontal member BC, unaffected by sway
in the right-angle frame studied earlier, now has fixed-end
sway moments, induced by the horizontal deflection at B.
172 Understanding Structural Analysis

87 87 The fixed-end moments in the vertical member CD are

straightforward and are simply related to the arbitrary
horizontal deflection ∆.

88 88 We will now carry out a quantitative analysis of the slightly

more complicated structure shown here. The only load on
the frame is the horizontal load of 4 kN at B.

89 89 There is no loading normal to any member, consequently

there are no fixed-end moments for the horizontally
restrained condition. The applied horizontal load of 4 kN at
B is taken through member BD to the artificial restraint at
D.

90 90 We now clamp joints B and D against rotation and impose

the arbitrary horizontal deflection ∆. The settlement
deflection, of member DE is √2∆, member DE being set at
45°.
 Moment Distribution 173

91 The fixed-end moments for the member BC are equal to 91

3 E I ∆ / L2 since the far end C is pinned.

92 The diagram shows fixed-end moments for column AB. 92

93 The diagram shows fixed-end moments for beam BD... 93

94 ...and the sloping member DE. Note particularly that 94

the ‘L’ in the expression for the moment induced by
‘settlement’ is the true length of the member.
174 Understanding Structural Analysis

95 95 We now need to determine the relative stiffness for each

member, in preparation for calculating the distribution
factors.

96 96 Distribution factors. Remember the advice about two

decimal places for the distribution factors and the check
that they must always sum to unity.

97 97 It is always advisable to draw the diagram of the fixed-

end moments as a check for the sign of these moments
in the moment distribution table. Bending moments are
always plotted as the action of the joint on the member,
and it is this sign that appears in the moment distribution
table. This was discussed in the earlier chapters on the
qualitative approach. So, for example, the sign of the
moment in BC at C is positive, that is clockwise, since it
is plotted by going along the member from B and rotating
clockwise to plot the value of 8.3 ∆ on the ordinate normal
to member BC at B.

98 98 Only the first line of the distribution table has been shown
to give you the opportunity to practise the remainder of the
operation. The final moments are shown on the bottom line.
 Moment Distribution 175

99 The distribution of final sway moments. The bending 99

moments are, of course, in terms of the arbitrary deflection
∆.

100 In order to find the horizontal external reactions at A, 100

C and E, from which the artificial restraint at B may be
found, each member must be analysed for shear. Shear in
member BC.

101 Shear in member BD. 101

102 Shear in member AB. 102

176 Understanding Structural Analysis

103 103 The key to this particular solution is the axial force in
member BD that balances the shears at B in members
AB and BC. If we consider member DE in isolation this is
balanced by the horizontal reaction at E.

104 104 At this point in the procedure a check may be made on this
value for the horizontal reaction at E. We may consider the
equilibrium of DE as a free body diagram. Taking moments
about D, there is a small error due to the approximate
procedures of moment distribution.

105 105 The arbitrary sway force at B, 3.36 ∆ is equal to the sum
of the reactions at A, C and E.

106 106 This factor is then applied to the distribution of arbitrary

moments, as shown in Figure 99 above, to produce the final
distribution of bending moments.
 Moment Distribution 177

107 It is useful to check any manual solutions with that 107

from the computer program. As you see, there are minor
differences between moment distribution solution and the
computer analysis, due in part to the approximate nature of
the former and due to the inclusion of axial strains in the
latter.

Practice Problems
Choose similar structures and carry out a hand calculation
and computer check.

Do not leave your hand calculations until they agree closely

with the computer analysis!
178 Understanding Structural Analysis
10 Plastic Analysis of Plane Frames

The development of structural theories so far in this book has been based upon an
important assumption, i.e. that the response of the structure to load is such as to keep
stresses within the elastic range where stress is proportional to strain. Thus the effect
of the load, such as stress and deformation, is directly proportional to the loading. Such
structures are described as being within their elastic limit. There is, however, another
approach to the design of structures developed from research work at the University of
Cambridge known as the Plastic Method of Analysis. Instead of considering the response
of the structure at working load and comparing the effects of that loading with a stress
limit normally derived by applying a factor of safety to the elastic limit of that material,
structures designed by the plastic method of analysis are considered at ultimate load.

The structural material for which this method of analysis was first developed was
structural steelwork and this led to the development of an analytical procedure which is
quite different from the elastic analysis we have considered so far. That is the subject of
this and the following chapter on the design of reinforced concrete slabs using the Yield
Line method.

It should be noted that the Plastic Method analysis does not take into account the real
deformations of the structure and that these, in any event, have to be checked at working
load, known in the modern codes of practice as the ‘serviceability’ state since it is the
deflection at working load which is a design criterion insofar as the prevention of damage
to finishes is concerned.
180 Understanding Structural Analysis

1 1 All the analytical procedures considered so far have been

based on the assumption that the structures are within
the elastic limit of behaviour for all conditions of loading.
We idealise the stress-strain curve of the material from
which the structures are made to this familiar straight-line
relationship; the elastic and plastic zones.

2 2 We are now going to examine the relationship between load

and stress resultants when the loading causes the structure
to exceed the yield strain. This analytical procedure is
known as plastic collapse, or ultimate load analysis.

3 3 To begin with, we will examine a simply supported beam.

The cross-section is rectangular. In this first stage of the
loading, the strain at the extreme fibres is within the elastic
limit and the stress is linearly proportional to the strain.

4 4 At a certain load, the strain in the extreme fibres reaches

that appropriate to the yield stress. The bending moment
at which this yielding first takes place is identified as the
yield moment, My .
 Plastic Analysis of Plane Frames 181

5 The section is still capable of taking load. However, as 5

the maximum stress, that is the yield stress fy, cannot be
exceeded, more of the section yields as the strain across
the section increases, bringing fibres which are closer to
the neutral axis, to yield stress.

6 Eventually the whole section is fully stressed at the yield 6

stress fy, except for the fibres very close to the neutral
axis. At this stage in the loading, the beam cannot sustain
any more load and a plastic hinge is formed. This loading
condition is known as the ultimate or collapse load and is
given the subscript c. The bending moment, identified on
the diagram Mp is shown with a bar.

7 The plastic hinge has certain properties that need to be 7

clearly understood:

1. The plastic hinge created by Mp has an effect on the

structure similar to that of a true hinge, as discussed
in the chapter on Indeterminacy.

2. However, the value of the moment at the hinge is not

zero, it is the fully plastic moment Mp.

8 The deflections at the ultimate load are of a considerably 8

larger order than those due to elastic bending so that
although the parts of the beam between the supports and
the plastic hinge, AC and CB, are curved with elastic
bending, they are normally drawn as straight lines.
182 Understanding Structural Analysis

9 9 We will now define two important relationships:

1. The Yield Moment.

2. The Shape Factor.

10 10 The yield moment My, is that which causes the stress in

the extreme fibre just to reach the yield stress. My may be
determined by substituting the value of the yield stress into
the elastic formula:

My / I = fy / y

11 11 The shape factor. At the fully plastic moment Mp, both

tension and compression faces are fully stressed at the
yield stress, because the increase in strain does not
increase the stress beyond the value of the yield stress.
Taking a rectangular section, the centroids of the fully
yielded stress blocks are at a distance d / 4 from the
neutral axis. The lever arm, therefore, is d / 2. The fully
plastic moment, the internal moment of resistance, for a
rectangular section of homogeneous material is:

Mp = fy x b d2 / 4

12 12 The shape factor is equal to the fully plastic moment

divided by the yield moment. The shape factor for a
rectangular section is 1.5.
 Plastic Analysis of Plane Frames 183

13 Generally sections are ‘I’ shaped and not rectangular, 13

structural steelwork for example. The shape factor for ‘I’
sections is around 1.15.

14 The qualitative approach to the analysis of structures is 14

particularly appropriate to plastic analysis because the final
solution may only be determined by the study of alternative
collapse mechanisms, that is the arrangement of plastic
hinges at collapse of the structural system.

To begin with we will see how the collapse mechanism and

the relationship between the collapse moment is arrived at
for straightforward structures. The first step, as always, is
to determine the qualitative solution; our first structure is a
propped cantilever; structure and deflected form.

15 A quantitative analysis of this beam will show that the 15

value of the elastic bending moment at A is greater than
that below the point load P.

16 Let us now study the effect of increasing the load P 16

until the structure collapses at the load PC As the load
increases, the yield stress is reached in the outer fibres at
A. It will occur at A because bending moments within the
elastic limit are in direct proportion to the loading and MA
is greater than that below the load.

In other words the relationship between the moment at A,

3 P L / 16 and under the load at C, 5 P L / 32 will remain
constant until the yield stress is reached at A. That
moment at A is then defined as My, the yield moment.
184 Understanding Structural Analysis

17 17 As the load is increased, more bending moment will

be absorbed at A until a fully plastic state is reached.
However, this is not the condition at C. The span AB
can still sustain an increase in load. Point C may still be
within the elastic limit depending on the relative values of
the bending moments at A and C. However, the bending
moment at C, beneath the point load, is shown here in what
is known as the ‘elasto-plastic’ state. A proportion of the
section is at the yield stress but the section is not yet full
plastic.

18 18 More load is applied until a plastic hinge forms at the next

largest bending moment after MA, that is under the load.
We now have the collapse load Pc . The structure cannot
sustain any further load. We are able to identify three
particularly important features of structures at the collapse
load:

1. The structure is a mechanism, that is the condition of

statical determinacy + 1 release.

2. The value of the fully plastic moments will be the

same at A and C – if the section of the beam is the
same at these points.

19
3. The elastic relationship of 3 P L / 16 at A and
5 P L / 32 at C no longer holds.

19 The value of Mp may be determined from a consideration of

statical equilibrium.

20 20 We may separate the free and restrained moments. The free

moment is that due to the application of a central point load
to a simply supported beam, BM = W L / 4, in this case
Pc L / 4.
 Plastic Analysis of Plane Frames 185

21 The restrained moment is the fully plastic moment at A, 21

Mp.

22 If we now add these together by taking the base line for 22

the free moment as the hogging line of Mp it is clear that
there is an easily identifiable relationship between the net
moment at C and the restraining moment at A.

Mc = Mp = Pc L / 4 − Mp / 2

∴ Mp = Pc L / 6

23 We now see why the plastic method of analysis may 23

produce (theoretical?) economies, because smaller
structural sections may be employed. We have effectively
reduced the maximum moment in the beam from
3 P L / 16, 0.1875 P L, to P L / 6, 0.166 P L, by
‘redistributing’ moments to the mid-span.

24 This calculation will illustrate the potential economy of the 24

plastic method of analysis. We will assume a working load
of 100 kN, a span of 8 m and a factor of safety on the yield
stress of 2 to give the safe working stress:

The elastic section modulus required = 0.652 x 106 mm3.

186 Understanding Structural Analysis

25 25 From the analysis at ultimate load, assuming a shape factor

of 1.15 the section modulus required is

= 0.502 x 106 mm3.

This would give a substantial saving.

Note that this simple design is an illustration and not

precisely correct in detail because of the partial factors of
safety actually employed in practical structural design.

26 26 The plastic analysis of this two-span beam ABCDE will

illustrate the judgement required by the analyst in the
application of the plastic method of analysis. Each span is
subject to a point load P, at mid-span. Clearly hinges will
most likely occur at D, B and E, at the points of maximum
bending moments.

27 27 If we now increase both loads to collapse, Pc, the plastic

hinges will coincide with the largest values of elastic
bending moments; i.e. MB and ME, and they will occur
in that order, because with equal central point loads and
unequal spans the elastic value of MB will be greater than
ME.

Note that the bending moments at collapse loading in span

AB are still within the elastic or elasto-plastic limit.

28 28 The value of Mp is determined, as before, by a simple

consideration of statical equilibrium.

1.5 Mp = Pc L2 / 4

Mp = Pc L2 / 6
 Plastic Analysis of Plane Frames 187

29 Note that the degree of indeterminacy is 1, therefore a 29

minimum of two hinges are required to create a collapse
mechanism.

30 So far it has been assumed that the sizes of the structural 30

members in each span have the same value. If they differ,
then the collapse moment will differ. Here you must be able
to identify that at the change of size, the support B, the
collapse moment will be based on the weaker of the two
sections.

31 The Theorem of Virtual Displacements is used to determine

the numerical relationship between the collapse load
PC and the plastic moment of resistance Mp for more
31
complex structures. We will illustrate the application of
this theorem to the beams studied so far. At the same time
we will introduce the concept of ‘upper’ and ‘lower bound’
solutions.

We will impose a virtual displacement state such that

there is a hinge at C that has undergone a virtual, vertical
displacement of unity.

32 The external virtual work is the collapse load x virtual

deflection of unity. The internal virtual work is the moment
Mp x rotation at C, 2θ. We ignore the bending energy 32
absorbed in the deformation of the beam from A to C and
from C to B. The angle θ is expressed in terms of the span
and virtual deflection. Since θ is small we may assume that

θ = 1 / 0.5 L = 2 / L

Equating external and internal virtual work, the solution

for the collapse load is

Pc = 4 Mp / L
188 Understanding Structural Analysis

33 33 This is, of course, the same solution we would have

achieved by studying the collapse bending moment
diagram:

P = 4 Mp / L

34 34 Let us now assume that we understood so little about

collapse mechanisms that we assumed that the hinge
would be at ¼ span for a collapse load at the centre.

35 35 The result of this analysis is that the ultimate load is

found to be 8 Mp / L, twice as great as the solution above.
It is important that you appreciate the significance of this
result. Firstly, the mechanism is possible, consequently
you may imagine that in more complex structures, you
may have collapse mechanisms less obviously incorrect.
However, it is clearly not the correct mechanism because
another mechanism, the one above with the plastic hinge at
the centre, yields a collapse load less than the mechanism
with a hinge at L / 4 span. Clearly the beam would collapse
at 4 Mp / L before it could attain 8 Mp / L.

36 36 The value of Pc for the hinge at ¼ span, 8 Mp / L is known

as an ‘upper bound’ value. If we plot the value of the
collapse load against the position of the hinge in the span,
the concept of ‘upper’ and ‘lower’ bounds will start to
emerge.

An upper bound solution is defined thus:

‘The prescribed loads are equal to or greater than
the collapse loading, if calculated from an assumed
collapse mechanism.’
 Plastic Analysis of Plane Frames 189

37 There are two other relevant definitions. The lower bound 37

may be stated thus:

‘The prescribed external loads are equal to or less than

the collapse loading provided that the bending moments
throughout the structure are equal to or less than the
ultimate moments and satisfy the equations of static
equilibrium.’

38 We may illustrate this in another way. In the virtual work 38

solution above, with a hinge incorrectly placed at ¼ span,
Mp the plastic moment is evaluated at Pc L / 8. There is
however an ultimate bending moment at the centre which is
greater than Pc L / 8.

39 These two concepts of the ‘upper’ and ‘lower’ bound are

combined and result in the uniqueness theorem: 39

‘The prescribed external loads are equal to the collapse

loading when a collapse mechanism is produced, provided
that the bending moments throughout the structure do not
exceed the ultimate moment and satisfy the equations of
equilibrium.’

In the simple example above of the centrally loaded beam,

the unique solution is that with the plastic hinge beneath
the load.

40 We will now return to the two-span beam and carry out an 40

analysis using the theorem of virtual displacements. We
can identify two possible collapse mechanisms:

Mechanism 1 where plastic hinges occur at B and E.

Mechanism 2 where plastic hinges occur at D and B.

190 Understanding Structural Analysis

41 41 Virtual work calculations

The expressions for the internal and external work are
similar for each mechanism. Clearly, if L1 is less than L2
then Pc in mechanism 1 is less than Pc in mechanism 2,
since the bending moments would be less in span AB
because of the shorter span. This means that span BC
would collapse before the collapse load Pc in mechanism 2
is reached.

42 42 Mechanism 1 would be chosen as the lower of these two

upper bound solutions. If we then compare mechanism I
with the equivalent ultimate load bending moment diagram
it is apparent that it satisfies the uniqueness theorem.

1. Bending moments do not exceed the ultimate

moments.

2. Equilibrium is satisfied because any value of moment

at B will satisfy equilibrium.

43 43 Load/deflection relationship
Although the deflection of ultimate load systems is not a
part of the solution, collapse being essentially a strength
relationship, the load/deflection curve offers some insight
into the nature of the collapse of a system. In the case
of the simply supported beam the deflection is directly
proportional to the load until the yield stress is reached at
My.

44 44 Beyond this load, the second moment of area is effectively

reduced as the stress/strain relationship ceases to be a
constant across the section. As the effective E I value
reduces so the deflection increases until the whole section
is at yield stress. Beyond this point, the structure cannot
sustain any increase in load, but will continue to deflect.
 Plastic Analysis of Plane Frames 191

45 The collapse of frames 45

Although there is a graphical method for the solution of
frames, based on a manipulation of the ultimate bending
moment diagram, it is cumbersome and conceals the first,
and most important decision in the collapse analysis,
the choice of mechanism. Let us examine alternative
mechanisms for the pinned-base portal frame ABCDE.

46 We will now separate the effect of each load on the 46

portal, assumed to be within the elastic limit. These are
the qualitative solutions for the distribution of bending
moments, resulting from the application of the horizontal
and vertical loads.

47 Depending on the relative value of the horizontal and 47

vertical loads, there are two possible elastic solutions. In
both, it is obvious that the combination of the effect of the
horizontal and vertical load will produce a maximum value
of the bending moment at D, since the bending moments in
both load cases at D are on the same side of the structure.
The alternative arrangement of the second hinge required
for the creation of the collapse mechanism is then identified
by the position of the maximum bending moments resulting
from a combination of these two load cases.

Mechanism 1. Hinges at C and D.

Mechanism 2. Hinges at B and D.

48

48 At this point in the procedure, you should be able to draw

a collapse mechanism, i.e. the deflected shape, which
satisfies these conclusions, from a study of the elastic
distribution of bending moments.
192 Understanding Structural Analysis

49 49 We will now solve each of these collapse mechanisms using

the Theorem of Virtual Displacements. We will look first at
mechanism 1. If we assume a virtual displacement of unity
at B in the horizontal direction, we may determine the
virtual displacement at C below the load of 3Pc.

The angle of rotation of AB is θ, i.e. 1 / L. The rotation

of member BC is also 1 / L, hence the vertical (virtual)
deflection at C = 1/2.

50 50 The external virtual work is the value of the load x

deflection in the direction of the load.

The internal virtual work is the value of the plastic moment

of resistance x virtual rotation. Both plastic hinges at C
and D undergo a virtual rotation of 2θ.

The resulting external virtual work is 2.5Pc and the

internal virtual work 4 Mp / L.

51 51 The value of the collapse load Pc, for this arrangement of

the hinges is:

Pc = 8 Mp / 5 L

52 52 We will now study the second mechanism, with hinges at B

and D. Note that this mechanism does not produce a virtual
displacement below the load of 3Pc .
 Plastic Analysis of Plane Frames 193

53 Equating external and internal virtual work: 53

Pc = 2 Mp / L

54 We now have two ‘upper bound’ solutions. There may be 54

another solution which could yield a lower value of Pc. In
this fairly straightforward structure we can be satisfied
that the lower of these two solutions, mechanism l, is the
unique solution since the distribution of bending moments
at collapse does not exceed the ultimate moment as
defined by the hinges and the equilibrium is satisfied by an
inspection of the qualitative solutions for the effect of each
load, shown in Figure 46 above.

55 Clearly, confidence in the choice of the mechanism is 55

related to the elastic distribution of bending moments since
it is this distribution that is the equilibrium condition. A
typical numerical distribution is shown for an arbitrary set
of loads in the correct value relationship.

56 It is clear from the distribution of elastic bending moments 56

that the first hinge will occur at D. If we assume a value of
2820 cm3 for the plastic modulus, the fully plastic moment
is equal to 1297 kNm. From the lower bound solution
above, Pc = 8 Mp / 5 L therefore the collapse load Pc is
equal to 415 kN.
194 Understanding Structural Analysis

57 57 In more complex structures, there will be a number of

alternative collapse mechanisms and the risk for the
analyst is that the unique solution will simply be missed.
Nothing in the calculations subsequent to the decision you
will take on alternative collapse mechanisms will reveal
that the unique solution is there, but that you have missed
it. Only your perception and practice will ensure that the
correct solution has been found.

We will now study the collapse mechanisms for this

two-bay portal frame, loaded with three point loads, one
horizontally at B, and centrally in spans BCD and DFG.

58 58 We can separate the effects of the horizontal and vertical

loads.

59 59 Mechanism 1
The structure is 3 x indeterminate, therefore 4 hinges are
required for a mechanism. The combination of the two
bending moments could produce a mechanism with hinges
at C, D in the beam and G.

The geometry of deformation will require a further hinge at

D in the column DE. This is confirmed by inspection of the
bending moment diagrams above.

60 60 Mechanism 2
If the bending moments due to the horizontal loading are
greater than those due to the vertical loading, a possible
mechanism is a simple sway. Note that the hinge at D is
in the column and that there are only three hinges in this
mechanism. Nevertheless the structure has collapsed,
regardless of the fact that beams BD and DG could still
sustain the load.
 Plastic Analysis of Plane Frames 195

61 Mechanism 3 61
We can identify another mechanism in the collapse of the
beam BCD. Only three hinges are required for this partial
mechanism, nevertheless it is a collapse of the frame.
However, such mechanisms are usually one of the upper
bound solutions.

62 We will now determine the quantitative solution for each 62

of these three mechanisms. The relative values of collapse
loads are shown in this figure. We will assume that the
plastic moment of resistance for the beams is 2Mp and the
columns, Mp.

63 Mechanism 1 63
Assuming a virtual horizontal deflection of unity at B, the
vertical virtual deflection at C = 1/2. The plastic hinges at
D in the beam and column and G each undergo a virtual
rotation of θ. The hinge at C undergoes a rotation of 2θ.

64 Equating the internal and external virtual work, the 64

collapse load Pc is equal to 3.2 Mp / L.
196 Understanding Structural Analysis

65 65 Mechanism 2
Hinges are introduced at B and D in the column DE, but
the beam is continuous across joint D. The last hinge is at
G.

66 66 The collapse load Pc is equal to 3 Mp / L.

67 67 Mechanism 3
There are assumed to be hinges at B, C and D in the beam
BCD. Clearly the external virtual work will be greater in
the beam BCD when compared with DFG because the
central point load and the span are greater. Note that the
plastic hinge at B forms in the column, since the plastic
moment of resistance of the columns is half that of the
beams.

68 68 From the equation of virtual work, Pc = 4.7 Mp / L.

This mechanism is an upper bound solution. The

mechanism which produced the lowest value for Pc is
mechanism 2.
 Plastic Analysis of Plane Frames 197

69 The elastic analysis is a valuable asset in gaining 69

confidence in the final choice of ‘unique’ solution. If we
determine the elastic analysis it appears that since hinges
must occur in descending order of the elastic bending
moments, if N + 1 hinges may be identified, a ‘unique’
solution has been found in mechanism 1 since the three
greatest elastic bending moments are at D in the column
and beam, C and G. However, mechanism 2 gives the
lowest value of Pc because of the partial sway failure
resulting from the lower resistance of the columns.

70 We will now study a slightly more complicated structure, 70

that of the pitched portal frame ABCDE loaded
horizontally at B and vertically at C. It is assumed that
the portal is of constant plastic moment of resistance
throughout.

71 It is always useful to separate the effect of the vertical 71

and horizontal loads and to study the resulting collapse
mechanisms. With the horizontal load at B, the frame will
collapse with hinges at B and D. The collapse mechanism
due to the vertical load at C will produce a change in
geometry as hinges form at B, C and D. It is assumed
that both the columns have been rotated outwards. Since
the structure is assumed to have the same plastic moment
of resistance throughout, hinges at B and D will occur
simultaneously. However, the mechanism requires only 2
hinges, i.e. a degree of statical indeterminacy of 1 plus 1
release.

72 Mechanism 1 72
One possible mechanism is a combination of the sway and
vertical load. The problem now is to relate the two virtual
deformations required for the solution, i.e. the horizontal
deflection at B and the vertical deflection at C.
198 Understanding Structural Analysis

73 73 We can use a device known as the ‘Instantaneous Centre’

(Horne, Plastic Theory of Structures, 1971). This point is the
junction of the lines taken from the points about which the
collapsed parts of the structure rotate, through the hinges.
We will use the symbol I for the instantaneous centre.

74 74 If we now consider this collapse mechanism and in

particular, member CD, the angle of rotation of C to C' and
D to D' about I must be the same because the distance
between C and D is fixed. We will identify this angle as θ.

75 75 If we assume that the horizontal virtual deformation at B

is unity, the angle φ is 1 divided by L. The angle between
the line AC and AC' is also φ because the bent ABC has
rotated about A. Since the value of the dimension AC is
the same as CI (this is a coincident of the symmetrical
geometry of this particular portal frame) θ has the same
value as φ.

76 76 Now imagine a line from C meeting the baseline AE at

F. As ABC rotates, the point F would drop to F', having
rotated through the angle φ. The vertical (that is, virtual)
deformation of point C is 1.5L x φ since the horizontal
dimension AF is 1.5L. Therefore the vertical virtual
deformation at C is 1.5 units.
 Plastic Analysis of Plane Frames 199

77 We now need to be able to describe the rotations in terms 77

of the frame dimensions and the virtual displacements.
The dimension of DD' = 2, since θ = 1 / L and I D = 2L,
therefore the angle γ = 2 / L.

78 The total virtual rotation at C is 2θ and at D, θ + γ. 78

79 The total internal virtual work is equal to 79

Mp x 2θ + Mp (θ + γ). Substituting appropriate values for
the angles, the total internal virtual work = 5 Mp / L.

80 The virtual displacements in the direction of the loads at B 80

and C are 1 and 1½ units respectively.
200 Understanding Structural Analysis

81 81 The total external virtual work is equal to

Pc x 1 + 3Pc x 1½ . Thus the relationship between the
collapse load Pc and the plastic moment resistance of the
structure is:

Pc = 10/11 x Mp / L

82 82 Mechanism 2
The pitched portal has a type of mechanism that could not
occur in a right-angle portal frame. If three hinges occur
at B, C and D, members AB and DE rotate outwards
because of the change in geometry of the pitched part of the
portal, BCD.

83 83 There is a relationship between the vertical deflection at

C, CC', and the horizontal deflection at D, DD', which
may be proved by using Pythagoras. Provided that these
displacements are considered to be small compared with
the overall frame dimensions, the relationship of x (CC') to
y (DD') is equal to the relationship of sides ‘a’ the pitch of
the portal, L / 2 and ‘b’, half the span 3 L / 2.

84 84 If we assume that the virtual horizontal displacement at B

and D is 1 unit, the vertical virtual displacement at C is 3
units. The hinge at C has undergone a total rotation of 2θ
and the hinges at B and C (θ + γ) each.
 Plastic Analysis of Plane Frames 201

85 To determine the internal virtual work we need to know 85

the value of the hinge rotations, θ and γ in terms of the
virtual displacements and the frame dimensions. θ is equal
to the vertical displacement at C divided by the appropriate
horizontal dimension FE.

86 The calculation for the external virtual work is unusual in 86

that there is negative work done against the horizontal load
at B.

87 The result for mechanism 2 is that Pc = 1.25 Mp / L. 87

88 Mechanism 3 88
This last mechanism is a simple sway with hinges at the
top of the columns at B and D. We assume a horizontal
virtual deformation of unity at B.
202 Understanding Structural Analysis

89 89 The solution to the equation of the internal and external

virtual work gives a value of Pc = 2 Mp / L.

90 90 Thus mechanism 1, a combination of sway and ridge beam

collapse, with plastic hinges at B and C, gives the lowest
value of Pc in terms of the fully plastic moment Mp. We can
check that the hinge will occur at these points by carrying
out an elastic analysis.

This distribution of elastic bending moments is based on a

horizontal load of 100 kN, a vertical load of 300 kN and a
span of 3 m. Clearly, the hinges will occur at D and C, in
that order.

91 91 Since only two hinges are required for collapse, mechanism

1 may be assumed to be the ‘unique’ solution.

This subject of the plastic theory of structures, or ultimate

load theory, is one to which full texts are devoted. The
coverage in this and the following chapter on the yield
line method of analysis for reinforced concrete slabs is,
therefore, necessarily limited. The reader should see the
explanations here as an attempt to show how a qualitative
understanding relates to the concepts of analysis at
ultimate load and turn to other texts for a full explanation
and practice problems.

Further reading
Horne, M R (1971) Plastic Theory of Structures, Nelson,
London.
Moy, S S J (1981) Plastic Methods for Steel and Concrete
Structures, Macmillan, London.
11 The Yield Line Analysis of
Reinforced Concrete Slabs

The Danish engineer Johansen developed the method of Yield Line Analysis for the
design of reinforced concrete slabs. The method began to appear in design offices in the
late 1960s, as concepts of ultimate load analysis were applied to reinforced concrete
structures.

Before the advent of ultimate load analysis, reinforced concrete slabs were designed
on the assumption that they were elastic plates. The analysis of such elements, taking
load normal to the member, and spanning in two directions, is complex. The structural
designer normally circumvented this difficulty by the use of standard tables that gave the
distribution of bending moments for orthogonal plates with varying support conditions.

The current codes of practice for reinforced concrete use tables for the design of slabs
spanning in two dimensions, based on the yield line method. However, these slab
tables are limited to rectangular panels. Admittedly, the majority of slabs come into
this category. However, one of the major advantages of reinforced concrete over other
structural materials is the complete flexibility of form. This may produce a structure with
non-orthogonal slabs, or orthogonal slabs with openings. In these circumstances, the code
tables cannot be applied. One of the hand methods open to the designer is the yield line
method.

To begin with however, the Yield Line method will be developed for rectangular slabs
before demonstrating its application to non-rectangular slabs.
204 Understanding Structural Analysis

1 1 We will develop the yield line method by studying first

a rectangular slab, simply supported along each edge,
carrying a uniformly distributed load throughout. The slab
is assumed to be of constant depth and reinforced in each
direction parallel to a support. The moment of resistance in
each direction may vary.

2 2 If that slab is loaded to failure, it will eventually fail along

a series of nearly straight lines, known as yield lines. The
resulting pattern of yield lines and panels is known as the
collapse mechanism. Note that the yield lines and panels
have a geometric compatibility. It is assumed that each of
the panels formed by the supports and yield lines is a flat
plate. The bending deformation within the panels is ignored
because the overall deformation due to the yielding of the
slab is of an order that reduces the deformation of the
panels between the supports and yield lines to a second-
order effect.

3 3 Here is a three-dimensional view of the same slab at

failure. There will be a maximum deflection at the centre
yield line. Because the panels are assumed to be flat plates,
and the slab is rectangular, the deflection of this centre
yield line is constant. This point will be clarified later as we
study the geometry of the yielded slab.

4 4 Generally, however, we draw slabs in plan. These

diagrammatic conventions are used to identify the different
support conditions.
 The Yield Line Analysis of Reinforced Concrete Slabs 205

5 We will call the area bounded by the yield lines and 5

supports a panel.

6 The significance of the assumption that panels are plane 6

surfaces is that it defines the geometrical relationships
between all parts of the panel and its boundary of yield
lines and supports. If we know the angle of rotation of the
panel and its plane dimensions, we can determine the slope
of any line within the panel. Similarly, the junction of any
two flat plates must be a straight line.

7 We are going to apply a concept studied earlier, the 7

Theorem of Virtual Displacements. This allows us to
impose a virtual pattern of deformations on the structure
and compare the equilibrium of the real loads and the
real resistance of the structure. Note the symbol for the
ultimate moment of resistance of the slab in a particular
direction based on the reinforcement, ‘ —m’. The
numerical suffix will be used to identify different values
of moments of resistance. Since the slabs are of constant
depth throughout, this difference will be almost exclusively
due to the amount of reinforcement provided. There is
a minor difference due to the fact that reinforcement in
different directions must be at different levels, one mat
lying on the other; consequently the effective depth will be
8
different in each direction.

8 This ‘bar’ symbol is to be interpreted that the reinforcement

is set at right angles to the bar. Bear that in mind. When
you see the symbol imagine the reinforcement at right
angles to it.
206 Understanding Structural Analysis

9 9 We will inspect the equilibrium of one panel to begin with,

panel A, bounded by cefd, and look at the equilibrium of
that panel in isolation.

10 10 If we assume that the uniformly distributed load w is that

which causes a full pattern of yield lines to be produced,
i.e. a collapse mechanism, we have an ultimate load
condition. At ultimate load, the internal virtual work, based
on the reinforcement, will be equal to the external virtual
work based on the deflection of the ultimate load.

We will assume that the virtual deflection of the yield line

ef is 1 unit. Due to this deflection the panel cefd has
rotated θA from its original horizontal position.

11 11 The figure shows a three-dimensional view of the panel

cefd at failure. Note that the deflection at e and f is unity
because of the plane geometry of the panel.

12 12 We will now imagine a reinforcing bar in a typical position

across yield line ef. The symbol —m does, of course,
represent reinforcement distributed across the full width of
the slab.
 The Yield Line Analysis of Reinforced Concrete Slabs 207

13 From the ultimate load theory of reinforced concrete 13

we know that the ultimate moment, based on the
reinforcement, is the area of tensile reinforcement x yield
stress x lever arm divided by the partial factor of safety on
the steel, 1.15.

14 From the chapter on virtual work we know that the internal 14

virtual work is equal to the moment of resistance multiplied
by the virtual rotation. Consequently the internal virtual
work for yield line of is equal to the moment of resistance
per metre multiplied by the angle of rotation.

15 The total internal virtual work for the whole length of the 15
yield line of is found by multiplying the internal virtual
work per unit length by the total length of the yield line. In
this expression

ylef = length of yield line ef,

m1 = ultimate moment of resistance of reinforcement

crossing this yield line at right-angles,

θA = rotation of the yield line of with respect to panel

A only.

16 We will now look at the internal virtual work of the 16

reinforcement which crosses yield lines ce and fd. A
section is taken parallel to support ac. Remember that we
are still only concerned with the internal virtual work with
respect to the rotation of panel A, bounded by yield lines
cefd, the equilibrium of which we are studying in isolation.
208 Understanding Structural Analysis

17 17 Because the yield line ce is one of the boundaries of

the panel cefd, the rotation of the yield line parallel to
reinforcement m1 is still θA.

18 18 Let us study the equilibrium of the bar crossed by a yield

line at an angle. The forces in compression could be
resolved normal and parallel to the yield line. However, this
leads to a complicated analytical approach where yield
moments are referred to the direction of the yield line.

19 19 If, however, the external and internal virtual work is

related to the direction of the reinforcement, a more
straightforward approach is revealed. We may simplify the
ultimate resistance of the slab by considering the yield line
as a series of steps, parallel and normal to the direction
of reinforcement. Only the ‘step’ of the yield line normal to
the reinforcement is assumed to contribute towards the
bending strength. The ‘strength’ of the step parallel to
the reinforcement is ignored. (Jones, L.L., Ultimate Load
Analysis of Reinforced and Pre-stressed Concrete Structures,
1968.)

20 20 Thus we may assume that the effective length of the yield

line is that dimension parallel to the direction of the
reinforcement as indicated by the ‘—m’ symbol.
 The Yield Line Analysis of Reinforced Concrete Slabs 209

21 The total internal virtual work of the yield line at an 21

angle to the reinforcement is equal to the moment of
resistance of the reinforcement per metre x the effective
length of the yield line x the angle of rotation of the
reinforcement. Therefore the total internal virtual work
for panel A, the work done by the reinforcement, is
equal to the effective length of yield lines ce and fd
plus ef which is the length L1 x the ultimate moment of
resistance per metre length m1 x θA.

22 We will now look at panel B. The reinforcement in direction 22

m1 is not bent by plate B. Therefore no internal virtual
work results from the rotation of panel B for reinforcement
in direction providing the resistance moment m1.

23 We will now introduce a new layer of reinforcement,

parallel to the long side of the slab that provides a 23
resistance moment m2. We will examine the internal virtual
work developed in this reinforcement, by rotation of the
triangular panel B. We will study a section taken just above
yield line ef. Since this reinforcement is parallel to the
line of rotation of panel A, that is the simple support cd,
the reinforcement is not ‘bent’ by the rotation of panel A.
Consequently m2 does not contribute to the equilibrium of
panel A. The position of the junction of yield lines at e and
f is a guess. The dimension that will give us the rotation of
this panel is identified as ‘l’.

24 If we now study section 2–2 we can see that because panel 24

B is a flat plate, the angle of rotation of the reinforcement
in m2 will be the unit deflection at point e divided by the
dimension at right-angles from the line of rotation ac to e.
We will define this as the radius of rotation ‘l’.
210 Understanding Structural Analysis

25 25 Simplifying the behaviour of yield lines ae and ce into a

series of ‘steps’, the effective length of these two yield lines
is equal to L2.

26 26 The total internal virtual work, created by the rotation of

panel B, relative to reinforcement m2 is the moment of
resistance m2 x the effective length of yield lines ae and
ec, the dimension L2 x rotation of panel B.

27 27 One important point in the procedure above needs to be

emphasised. We are relating the length of the yield line and
the rotation of the panel to the line of the reinforcement.
Because the slab is rectangular and the reinforcement
placed parallel to the sides, the rotation of a panel, always
taken as that within the plane lying normal to the line
of rotation, is the same as the angle of rotation of the
reinforcement.

In the general case, studied later, this does not necessarily

occur.

28 28 Now we will turn our attention to the external virtual work

that is created by the virtual deflection of the centroid of
the total load acting on a panel. This would, of course,
include the self-weight of the slab. In panel B, we may
assume that the total load on the panel acts at the centroid.
The virtual deformation of this point is 1/3. The external
virtual work created by the deformation of the panel is
equal to area of panel B x total ultimate load per unit
area, w x virtual deflection of the centroid.
 The Yield Line Analysis of Reinforced Concrete Slabs 211

29 The external virtual work of the trapezoidal panel cefd is 29

best dealt with by dividing it into easily calculated shapes,
triangles and rectangles. This allows the identification of
the position of the centroid from which the value of the
virtual deflection of the centroid may be found.

30 The object of the yield line analysis is to determine the 30

relationship between the ultimate load w and the ultimate
moment of resistance m of the band(s) of reinforcement.
The internal virtual work is a function of m, the external
virtual work of w. These are equated:

internal virtual work = external virtual work

We then solve for the ultimate moment of resistance m in

terms of the ultimate load w.

31 The position of e and f and consequently the value of the 31

dimension l is the analyst’s choice. The value of m, for
a given value of w will vary as a function of the angle of
rotation of the triangular panel.

32 For a given loading the designer is trying to find the worst 32

case, that is the greatest value of m. This is the ‘unique’
value, as discussed in the previous chapter. It is possible
to set up a general expression with the variable l and find
the maximum value of m which can be satisfied by the
reinforcement and the chosen set of yield lines by using
calculus. However, this approach is cumbersome when
applied to non-orthogonal slabs. We will develop the more
generally useful trial and error approach.
212 Understanding Structural Analysis

33 33 We will now examine a slab that allows us to study the

solution in a more general form. The triangular slab abcd
is simply supported along sides ab and ad and free along
bcd. The assumed yield line is ac. There are two bands of
reinforcement, represented by —m1 and —m2, neither of
which is parallel to the supported edges of the panel, nor
are they at right-angles to each other. We will identify the
two panels as A, bounded by acd and B, bounded by abc.
It is assumed that the virtual deflection of point c is unity.

34 34 To determine the internal virtual work for reinforcement

m1 that relates to the rotation of the yield lines bounding
panel A, we need to identify two significant dimensions:

1. The ‘radius of rotation’ of reinforcement in band m1,

dimension l(m ).
1

2. The effective length of the yield line in the direction

normal to the line of the actual reinforcement.

35 35 Now we can see the distinction between the rotation of

the panel, and the rotation of the reinforcement. The angle
α is in the plane at right angles to the line of rotation,
ad. Angle θ is in the plane containing the direction of the
reinforcement.

36 36 Taking a section parallel to the reinforcement, the angle of

rotation of the reinforcement for panel A, band m1, is the
virtual deflection of 1 divided by the radius of rotation l(m ).
1
 The Yield Line Analysis of Reinforced Concrete Slabs 213

37 We may now find the relevant dimensions for reinforcement 37

band m2. Note that the radius of rotation l(m ) is the line
2
ce, where e is an extension of the line of rotation of panel
A, ad. We will take a section parallel to the reinforcement
in band m2.

38 The angle of rotation of the reinforcement in m2 is θA(m ). 38

2
This is equal to the virtual deformation of l at c divided by
the radius of rotation l(m ).
2

39 Now for the significant dimensions for the internal virtual 39

work for panel B. Firstly, for reinforcement m1. The radius
of rotation of panel B is the dimension cf, where f the
coincidence of the extended line of rotation ab and a line
drawn parallel to the direction of the reinforcement in m1,
which passes through c. A section is taken parallel to the
reinforcement.

40 To understand this section you must be able to recognise 40

the rotational continuation of the plane of panel B, abc.
The panel is notionally extended to the area bcf. Thus the
angle of rotation of reinforcement in band m1, crossing
yield line ac, is equal to the unit deflection at c divided
by the radius of rotation l(m ) for this panel, that is the
1
dimension cf.
214 Understanding Structural Analysis

41 41 The figure shows panel B and the significant dimensions

for the reinforcement in band m2. The radius of rotation is
the dimension l(m ).
2

The analysis would be completed with a calculation of the

external work, from which the relationship between m and
w can be found.

42 42 The analysis of simply supported triangular slabs presents

important geometrical relationships. We will assume that
the yield line is at an angle α to the support ad and that
there are two bands of reinforcement m1 and m2. The angle
between the simple supports ab and ad is β.

43 43 If we calculate the relationship between the ultimate load

w, and ultimate moments of resistance m, we find that
there is a position for the yield line that gives a minimum
value of w in relation to m. The result of such an analysis
would show that the yield line bisects the angle dab.
This is, of course, the maximum value of w that the slab is
capable of supporting at ultimate load.

(Jones, L. L. ibid)

44 44 So far in this study of the behaviour of yielded slabs, all

the reinforced yield lines have been sagging. If the slab
is restrained along a support, there will be an additional
hogging yield line, usually shown dashed, with its own
separate band of reinforcement and contribution to the
internal virtual work.
 The Yield Line Analysis of Reinforced Concrete Slabs 215

45 To bring all these ideas together, we will now look at the 45

full analysis of the slab abcd.

The slab is simply supported along sides ad and be and

fully restrained along ab. The edge cd is free. The co-
ordinates of each of these points, in metres, are given. You
will find it useful to draw the slab to scale so that you will
be able to check certain critical dimensions.

46 There are three bands of reinforcement: 46

Band 1. Over the fixed support ab, in the top layer of

the slab, shown dashed. The reinforcement is
normal to ab.

Band 2. Reinforcement parallel to side bc, bottom of the

slab.

Band 3. Reinforcement parallel to side cd, bottom of the

slab.

47 We will assume that each of these bands has a different 47

value for the ultimate moment of resistance. The main
reinforcement is m3 and will be given the greatest value
1.3 units as this will support the majority of the load.
Reinforced concrete slabs will almost always resist loading
by taking the shortest path. In a rectangular slab, for
example, the majority of the bending resistance is taken by
the short span. The remaining values are m1 = m and
m2 = 0.7m.

48 In a real design problem we would have to choose a range 48

of patterns of yield lines. This is our first choice. There
is a hogging yield line at ab and we will assume point e
to form a triangle of yield lines abe. We will assume a
notional virtual deflection of unity for point f. The panels
are given references, A for the triangular panel, B and C
for the trapezoidal panels. Note the co-ordinates of points e
and f.
216 Understanding Structural Analysis

49 49 Because the panels surrounded by the yield lines are

considered to be flat plates, consistent geometry requires
the extension of yield line of to pass through the junction
of the continuation of the lines of rotation be and ad. This
is identified as point o. If both supports be and ad are
simply supported, the point f will be the mid-point between
c and d because the pattern of yield lines is similar to that
for the triangular slab discussed above.

50 50 The virtual deflection of point f is assumed to be unity.

Because of the geometry of the slab, the deflection of point
e is less than unity. The angle of rotation of panel bcfe,
panel B, is 1 / 2.4, i.e. in the plane normal to the line of
rotation bc. From this the virtual deflection at e may be
found to be 0.79.

51 51 We will start our yield line analysis by looking at the

hogging yield line ab. The appropriate rotation is in the
plane parallel to the reinforcement m1, which in this
particular case, is the plane normal to support ab. The
relevant radius of rotation is 1.7 m. The rotation of the
reinforcement in m1 is equal to the virtual deflection 0.79
divided by the radius of rotation, 1.7 m.

52 52 The appropriate length of the yield line ab is normal

to reinforcement, m1, that is side length ab. The total
internal virtual work is equal to the length of the yield line
x value of ultimate moment of resistance per unit length x
angle.
 The Yield Line Analysis of Reinforced Concrete Slabs 217

53 Yield lines ae and be, panel A, are sagging yield lines. 53

We will study the internal virtual work that derives from
the reinforcement in band m2. The radius of rotation for
this band of reinforcement, for panel A is the dimension
ge, 1.6 m. The effective length of the yield line, normal to
reinforcement m2 is 3.4 m.

54 Thus we may determine the internal virtual work for yield 54

line aeb, reinforcement band, m2.

55 Since the reinforcement in m3 is not parallel to the line of 55

rotation of support ab, yield lines ae and be deform the
reinforcement in this band.

56 The angle through which reinforcement in band m3 rotates 56

is equal to the deflection at e, 0.79, divided by the radius of
rotation dimension he, 7.0 m.
218 Understanding Structural Analysis

57 57 The effective lengths of yield lines be and ae, normal to

the reinforcement in m3 are 1.2 m and 2.0 m respectively.
However, the typical bar crossing both yield lines will be
bent in opposite directions, sagging (i.e. positive) over yield
line ae and hogging over line be. This is illustrated in
figure 58.

58 58 From this data we may determine the total internal

virtual work due to yield lines ae and eb for panel A,
reinforcement m3.

59 59 We will now turn our attention to panel B and the bounding

yield lines be and ef. Reinforcement band m1 is in the top
of the slab and since these are sagging yield lines, creates
no internal virtual work in these yield lines. Reinforcement
band m2 also makes no contribution towards the internal
virtual work. The reinforcement must be bent by the
rotation of a particular panel, with respect to a particular
yield line. Since the reinforcement in m2 is parallel to the
line of rotation of panel B, it will not be bent by the rotation
of this panel.

60 60 Now to the evaluation of the internal virtual work created

by yield line bef, reinforcement band m3. The effective
length is that normal to the reinforcement, side be and the
radius of rotation is cf, 2.4 m.
 The Yield Line Analysis of Reinforced Concrete Slabs 219

61 Thus we may determine the internal virtual work, yield line 61

bef, for the reinforcement in band m3, for panel B.

62 The reinforcement in band m2 is almost parallel to yield 62

line ef. Consequently the rotation of panel C about the line
of rotation ad causes the reinforcement in m2 to undergo
only a small rotation. The diagram shows the radius of
rotation, line fk, where k is the junction of the line passing
through f, parallel to the reinforcement and the extension
of the line of rotation of panel C, ad. Note that the effective
lengths of yield lines ae and ef.

63 The bar crossing both yield lines is bent in opposite 63

directions, similar to the situation in panel A, i.e. hogging
over yield line ef and sagging over ae. From this data we
may determine the internal virtual work for yield line aef,
for the reinforcement in band m2 for panel C.

64 The reinforcement in m3 will take the major part of the 64

load and will therefore contribute the most internal virtual
work for this panel. The radius of rotation is the line df
and the effective length of yield line aef is the dimension
normal to the reinforcement.
220 Understanding Structural Analysis

65 65 From these dimensions we may determine the internal

virtual work for yield line aef for panel C, reinforcement
m3.

66 66 The total internal virtual work is the sum of the resistance

of each yield line for each panel. The total for this choice of
yield lines is 10.76m kN.m.

67 67 The external virtual work requires the identification of

the centroids of easily calculated areas for the effective
proportion of the unit deflection at f. Calculation will show
that the external virtual work is equal to 11.0 w kN.m,
where w is the total ultimate load on the structure, per unit
area, inclusive of the self-weight of the slab.

68 68 If we assume that the total ultimate load (1.4 x dead load +

1.6 x imposed load) is 14 kN.m2, then the ultimate moment
of resistance, m for this particular pattern of yield lines is
14.9 kN.m/m run.
 The Yield Line Analysis of Reinforced Concrete Slabs 221

69, 70 Practice problems 69

In each of the four practice problems shown below, the
yield line pattern has been chosen. You should consider
the equilibrium of panel A. Then identify, for each band
of reinforcement, the radius of rotation and the effective
length of the yield line as shown in this example.

70
222 Understanding Structural Analysis
12 Influence Lines

The subject of Influence Lines has been left until last because the basic approach to the
analysis of statically indeterminate structures and influence lines is completely different
and it is important for the student to separate the two approaches quite distinctly.

Normally, the loads on a structure are static. The subsequent analysis determines the
distribution of load effects; bending moments, shear forces, etc. at points along the
structural members.

The influence line is the result of a procedure which is almost exactly opposite. The
influence line is a diagram, using the structure as the base, and is a measure of the load
effect at one point for a moving load.

Clearly it is particularly useful in the analysis of structures that have to be designed for
moving loads; crane gantries and bridges being the most obvious examples. The technique
of influence lines allows the designer to examine the effect on one member in a bridge, for
example, under the effect of a load or a combination of loads moving across the bridge.
This may reveal that a particular member has to act in both tension and compression for a
certain combination of loads.
224 Understanding Structural Analysis

1 1 We will start by examining the qualitative behaviour of this

continuous beam ABCD, loaded in the span AB with a
point load. The resulting deflected shape is shown.

2 2 This diagram shows the resulting reaction and shear force

diagrams. The size of the reaction arrows has been drawn
to show the approximate relative value of the reactions.
The reaction at B will have the greatest value. Note that
we are ‘looking’ along the beam from left to right.

3 3 We now construct the qualitative bending moment diagram.

We are going to inspect the effect of moving a point load
along the beam, from A to D, at a particular point and
effect; the bending moment at B. The bending moment at B
for a point load in span AB is hogging.

4 4 We will now look at the effect of moving the load into span
BC. Note that the direction of all but one of the reactions,
VB is reversed from that resulting from the application of
the load in span AB.
 Influence Lines 225

5 This diagram shows the resulting qualitative shear force 5

and bending moment diagrams. If the spans of the beam
were equal, the bending moment at B for the point load in
span BC would be less than the bending moment which
results from the application of the point load in span AB.
AB is similar to a propped cantilever whilst BC is more
like a fully fixed beam.

6 We will now move the load into span CD. The diagrams are 6
the mirror image of those due to the load in span AB.

7 Reactions and shear force diagrams for the load in span 7

CD.

8 The distribution of bending moments for the load in span 8

CD. Notice that the bending moment at B resulting from
the application of the load in span CD is causing tension
on the underside of the member.
226 Understanding Structural Analysis

9 9 We will start to plot the influence line for the bending

moment at B, MB . The ordinates drawn are the values
of MB when the load is at that point along the beam. MB
is zero when the point load is over the supports at A, B,
C and D. You should confirm these first points on the
influence line diagram by referring back to the bending
moment diagrams above.

10 10 We can now complete the influence line with a smooth

curve, although we will go on to prove that it is a smooth
curve, and not a series of straight lines.

11 11 Now let us look at the influence line for the vertical

reaction at B, VB. When the load is over the support at B,
the reaction is equal to the load.

12 12 We may now draw the first points of reference on the

influence line for VB. Refer back to the qualitative diagrams
above which indicate the relative values of the reactions.
The ordinate in the mid-span of AB is shown in Figure 2;
the ordinate at B is equal to the applied load; ordinates
in spans BC and CD are shown in Figures 4 and 7
respectively. When the load is positioned directly over the
support at C, the vertical reaction at B is zero.
 Influence Lines 227

13 The full influence line for the reaction VB is completed with 13

a smooth curve.

14 We will now go on to prove that the influence line for the 14

particular effect on a structure is the deflected shape
that would result from the application of the object of the
influence line, as a released action. Here the support at B
has been released and the reaction applied as a load. This
is identical in shape to the diagram above in Figure 13.

15 We will now prove this relationship by studying the 15

analysis of the two-span beam ABC. Note that the vertical
reaction at B, VB, is a function of the distance of the point
of application of the load from point A, distance x.

16 We will adopt a flexibility approach to the solution of this 16

two-span beam, which is 1 x statically indeterminate. VB
is removed and the deflection due to the real load w is
determined at B, ∆B . The deflection under the load a
w
distance x from A is ∆x . We now apply a unit load at B
w
and the resulting vertical deflection is δB B at B. That
v v
is the deflection due to a unit load, at B in the vertical
direction, resulting from the application of the load at B in
the vertical direction.
228 Understanding Structural Analysis

17 17 The actual deflection at B is zero. Therefore we can set

up the equation of compatibility at that point. From the
solution of the equation the vertical reaction at B, VB
is equal to the deflection at B of the released structure
divided by the deflection at B, resulting from the
application of a unit load at B. Note the negative sign. If
upward deflections are assumed to be positive, ∆B would
w
be negative. Thus VB would be positive indicating that the
choice of direction, upwards, is correct.

18 18 We will now use a slightly different set of symbols. The

deflection at B due to the load w at x is defined as ∆B .
x
If we place a unit load at x from A, the deflection at B is
identified as δB , that is the x deflection due to the unit load
x
(δ) measured at point B (δB) due to a load at x (δB ).
x

19 19 We can now set up a slightly different equation of

compatibility. ∆B is equal in value to ∆B for the real
x
structure since the actual vertical deflection at point B is
zero. We can substitute wδB for ∆B since ∆B is equal to
x x x
w x δB from the above figure.
x

20 20 We now apply a theorem developed earlier in the chapter

on the Flexibility method: Clerk-Maxwell’s Theorem of
Reciprocal Deflections. A unit load applied at point x will
produce a deflection at B, δB . A unit load applied at B
x
will produce a deflection at x, δx .These deflections are of
B
equal value.
 Influence Lines 229

21 Substituting the deflection δx for the deflection δB in the 21

B x
equation of compatibility shown in Figure 19 we find that
the reaction VB is equal to the deflection at x due to a unit
load at B multiplied by the load w divided by the deflection
at B resulting from the application of a unit load at B.

22 δx is the deflection at any point on the structure due to the 22

B
application of a unit load at B, therefore the influence line
for VB is the deflected shape resulting from the application
of a unit load at B, drawn to a suitable scale.

23 We may summarise this procedure thus: To find the 23

influence line for a reaction, we release the structure
at that point and apply a unit point load. The resulting
deflected shape is the influence line, drawn to a suitable
scale. The ordinate at point B is 1.

24 The influence line for the vertical reaction at A. Note that 24

the influence line curve is negative in span BC. This is as
you would expect since a load applied in span BC would
produce a hogging moment at B, inducing a downward
reaction at A.
230 Understanding Structural Analysis

25 25 We will now return to our original problem of the three-

span beam ABCD. The influence line for VB is the
deflected shape of the structure resulting from the removal
of the vertical reaction at B and the application of a unit
load.

26 26 To determine the actual value of VB, the ordinate at the

point of application of the load is multiplied by the value of the
load.

27 27 This is an appropriate point in the development of these

ideas to summarise the procedure for creating an influence
line.

28 28 It may be shown that the same principles apply to the

influence line for a bending moment at any point in a
structure. The two-span beam ABC is fully fixed at A. If
we wish to draw the influence line for the reaction moment
at A, MA, we must release the fixed part of the reaction at
A and apply a unit moment.
 Influence Lines 231

29 The resulting deflected shape is the influence line, drawn to 29

a suitable scale.

30 A similar procedure is adopted for a shear force influence 30

line. The structure is released at the object point of the
influence line, here a point in the span AB.

31 Note that the moment connection is retained at this 31

vertical release between A and B. The influence line is
the deflected shape resulting from this release. The ‘cut’
must be replaced by equal and opposite loads to create the
correct deflected shape. Note that the tangent to the curve
of the influence line is horizontal either side of the ‘cut’.

32 We will now study the application of this method to 32

influence lines for two-dimensional structures. Care must
be taken in two-dimensional structures not to confuse
the influence line deflection with the normal deflection.
In beams the problem does not arise. We will study the
influence lines for this two-dimensional frame ABC. First
though, we need to determine the full qualitative solution
for the deflected shape, reactions and bending moment
diagram.
232 Understanding Structural Analysis

33 33 Let us assume that we wish to find the influence line for

the horizontal reaction at A, HA. The loading that is used,
implicitly, to create the influence line can only be for loads
in either the horizontal or vertical direction. The released
deflected shape is the place to start to create the influence
line but it has to be carefully interpreted. We must
remember that the value of the horizontal reaction A is the
design load multiplied by the influence line ordinate in the
direction of the design load.

34 34 This diagram shows the influence line for horizontal loads

acting at any point on the frame ABC. We are ignoring
axial strain, consequently the horizontal deflection of point
B is zero. A horizontal load at B would be transferred
directly through the axial resistance of BC to the reaction
of C. Consequently no deformation would occur at B. AB
would remain un-deformed and the horizontal reaction at
A resulting from the application of a horizontal load at B
would be zero. This would be true for any horizontal load
applied anywhere along BC.

35 35 The influence line for vertical loads may be similarly

interpreted. A vertical load applied at B or anywhere along
AB would not produce a horizontal reaction at A.

36 36 Statically determinate structures present a special problem

because the structure will become a mechanism if a release
is applied. The resulting deflected shape of the mechanism
is in the form of a notional rigid body displacement and the
key unit ordinate must be determined by inspection. We
will study the simply supported beam ABC. The influence
line for VC is required. Removing the vertical reaction
at C and applying a unit load, the beam will rotate as a
straight member about A. Consequently, the influence line
is straight from A to a value of unity at C.
 Influence Lines 233

37 Similarly, if the object of the analysis is the influence line 37

for the bending moment at B, we release the structure at
this point by the introduction of a hinge and apply equal
and opposite unit moments, either side of the hinge.

38 This two-span beam ABCD is statically determinate 38

because of the hinge release at B. We want to find the
influence line for the bending moment at C. We therefore
introduce a hinge at C and apply the equal and opposite
unit moments at C. At first it would appear that there
would be bending in CD, due to the application of the
right-hand moment, but this is not the case. These internal
moments react against each other. The structure to the left
of C is now a mechanism. Consequently it has no bending
stiffness. The two balancing moments react against the
stable portion of the beam CD, and the displacement takes
place in ABC.

39 Always check the sense of your influence line. If we place 39

a load on the original structure in span CD, the bending
moment at C is zero.

40 We may confirm this influence line by considering the

application of a point load in span AB which will develop
a hogging moment at C, which will diminish as the load 40
approaches either point A or point C. Consequently it is
correct that the influence line is a maximum value at B.

Practice problems
Choose one of the structures in the solutions for the
practice problems included in the earlier chapters. Choose
a point and load effect to be the object of the influence line.
Determine the deflected shape, and from it the influence
line. Check that the influence line is correct for the
particular loading in the solution.
234 Understanding Structural Analysis
Appendix: Solutions to Practice Problems

Chapter 2: Statical Indeterminancy

Graphical solutions to practice problems shown in figs. 56 to 58

Solution 2.1

Solution 2.2

Solution 2.3

Solution 2.4
236 Understanding Structural Analysis

Solution 2.5

Solution 2.6

Solution 2.7
Solutions to Practice Problems 237

Solution 2.8

Solution 2.9
238 Understanding Structural Analysis

Chapter 5: The Theorems of Virtual Work

Graphical solutions to practice problems shown in figs. 82

Solution 5.1 Solution 5.2

Solution 5.3

Solution 5.4
Solutions to Practice Problems 239

Solution 5.5

Solution 5.6
240 Understanding Structural Analysis

Chapter 6: The Flexibility Method

Graphical solutions to practice problems shown in figs. 32 and 33

Solution 6.1

Solution 6.2

Solution 6.3
Solutions to Practice Problems 241

Solution 6.4

Solution 6.5

Solution 6.6a
242 Understanding Structural Analysis

Solution 6.6b
 Solutions to Practice Problems 243

Chapter 11: The Yield Line Analysis of Reinforced Concrete Slabs

Graphical solutions to practice problems shown in fig. 70

Solution 11.1

Solution 11.2

Solution 11.3
244 Understanding Structural Analysis