es second edition?

A GEA,
Mechanics of
Engineering Materials
Second edition
 Digitized by the Internet Archive
in 2022 with funding from
Kahle/Austin Foundation

https://archive.org/details/mechanicsofenginO00Obenh
Mechanics of
Engineering Materials
Second edition

P. P. BENHAM
Professor Emeritus and formerly Professor of Aeronautical Engineering,
The Queen’s University of Belfast

R. J. CRAWFORD
Professor of Mechanical and Manufacturing Engineering
and Director of the School of Mechanical and Process Engineering,
The Queen’s University of Belfast

C. G. ARMSTRONG
Reader in the Department of Mechanical and Manufacturing Engineering,
The Queen’s University of Belfast

EY LONGMAN
Addison Wesley Longman Limited
Edinburgh Gate
Harlow,
Essex CM20 2JE, England.
and Associated Companies throughout the world

© P. P. Benham and R. J. Crawford 1987

This edition © Longman Group Limited 1996

All rights reserved; no part of this publication may be

reproduced, stored in any retrieval system, or transmitted in
any form or by any means, electronic, mechanical,
photocopying, recording, or otherwise, without either the prior
written permission of the Publishers or a licence permitting
restricted copying in the United Kingdom issued by the
Copyright Licensing Agency Ltd, 90 Tottenham Court Road,
London, W1P 9HE

First Published 1987

Second Edition 1996. Reprinted 1996 and 1997
Reprinted 1998

British Library Cataloguing in Publication Data

A catalogue entry for this title is available from the British Library
ISBN 0-582-25164-8

Library of Congress Cataloging-in-Publication Data

A catalog entry for this title is available from the Library of
Congress

Set by 6 in Monotype Ehrhardt Roman

Produced by Longman Singapore Publishers (Pte) Ltd
Printed in Singapore
 Contents

_ Preface to the second edition 1X AA eves of a thin circular tube

2.10) | Joints
Preface to the first edition x ~ Summary
Problems
Notation xil
3 Stress—Strain Relations
3.1, Deformation
Strain
1 Statically Determinate Force Systems ] Elastic load—deformation behaviour of materials
Revision of statics ] Elastic stress—strain behaviour of materials
Resultant force and moment 5 Lateral strain and Poisson’s ratio
Types of structural and solid-body components 6 Thermal strain
Types of support and connection for structural General stress—strain relationships
components Strains in a statically determinate problem
Statical determinacy Elastic strain energy
Free-body diagrams Strain energy from normal stress
Determinacy criteria for structures 1 Strain energy from shear stress
Determination of axial forces by equilibrium Plastic stress—strain behaviour of materials
statements 12 Viscoelastic stress—strain behaviour of materials
Forces by the method of tension coefficients 18 Summary
Bending of slender members | Problems
Force and moment equilibrium during bending 23
Sign convention for bending 23 4 Statically Indeterminate Stress Systems
Shear-force and bending-moment diagrams 2
Cantilever carrying a concentrated load 24 /4.1\ Interaction between components of different
| stiffness
Cantilever carrying a uniformly distributed load 25
Restraint of thermal strain
Simply supported beam carrying a uniformly
Volume changes
distributed load 26
Constrained material
1.17 Point of contraflexure 28
Maximum stress due to a suddenly applied
Torsion of members 28
load
1.19 Members subjected to axial force, bending
Maximum stress due to impact
moment and torque 29
Summary
G20) Torque diagram 30
Problems
1.21 The principle of superposition 31
Summary 5
Problems 33 5 Torsion
5.1) Torsion of a thin-walled cylinder
Torsion of a solid circular shaft
2 Statically Determinate Stress Systems 43 Relation between stress, strain and angle of
G2 |Stress: normal, shear and hydrostatic 43 twist
2.2 A statically determinate stress system 45 Relation between torque and shear stress
3 ‘Assumptions and approximations 46 The torsion relationship
|2.4 Tie bar and strut (or column) 46 Torsion of a hollow circular shaft
| 2.5 Suspension-bridge cables 50 Torsion of non-uniform and composite shafts
Thin ring or cylinder rotating 51 Torsion of a thin tube of non-circular section
Thin shells under pressure Torsion of thin rectangular strip
Shear in a coupling Effect of warping during torsion
vi CONTENTS

eli Torsion of solid rectangular and square Elastic strain energy: normal stress and
cross-sections 1) shear stress
Summary 119 Strain energy in torsion
References 120 Strain energy in bending
Problems 120 Helical springs
Shear deflection of beams
6 Bending: Stress 125 Virtual work
Displacements by the virtual work method
6.1 Shear-force and bending-moment Strain energy solutions for forces
distributions 125 Complementary energy solution for
6.2 Relationships between loading, shear force deflections
and bending moment 130 Bending deflection of beams
Stress distribution in pure bending 132 The reciprocal theorem
Deformation in pure bending 133 Summary
Stress—strain relationship 134 Problems
Equilibrium of forces and moments 135
The bending relationship 136
The general case of bending 136 10 Buckling Instability
Section modulus 137 10.1 Stability of equilibrium
Beams made of dissimilar materials 139 10.2 Buckling characteristics for real struts
Combined bending and end loading 147 10.3 Eccentric loading of slender columns
Eccentric end loading 148 10.4 Struts having initial curvature
Asymmetrical or skew bending 151 10.5 Empirical formulae for design
Shear stresses in bending 156 10.6 Buckling under combined compression
Bending and shear stresses in I-section and transverse loading
beams 160 10.7 Pin-ended strut carrying a uniformly
Shear stress in thin-walled open sections distributed lateral load
and shear centre 164 10.8 Other examples of instability
Bending of initially curved bars 168 10.9 Local buckling
Beams with a small radius of curvature 196 Summary
Summary 173 References
Reference 173 Problems
Problems 173
11 Stress and Strain Transformations
7 Bending: Slope and Deflection 185
1 Symbols, signs and elements
7.1 The curvature—bending-moment relationship 185 11.2 Stresses on a plane inclined to the direction
Te Slope and deflection by the double-integration of loading
method 186 3 Element subjected to normal stresses
Discontinuous loading: Macaulay’s method 193 11-4 Element subjected to shear stresses
Superposition method 202 1B) Element subjected to general two-dimensional
Deflections due to asymmetrical bending 204 stress system
Beams of uniform strength 205 11.6 Mohr’s stress circle
Summary 206 11.7 Principal stresses and planes
Problems 207 11.8 Maximum shear stress in terms of principal
stresses
8 Statically Indeterminate Beams 211 9 General two-dimensional state of stress at
8.1 Double-integration method 211 a point
8.2 Superposition method 216 11.10 Maximum shear stress in three dimensions
8.3 Moment-area method 220 11.11 States of strain
Summary Ze, 11.12 Normal strain in terms of co-ordinate strains
Problems 232 leit Shear strain in terms of co-ordinate strains
11.14 Mohr’s strain circle
EIS Principal strain and maximum shear strain
9 Energy Methods 235 11.16 Experimental stress analysis
Onl Work done by a single load 235 Mey Rosette strain computation and circle
OP Work done by a single moment 235 construction 314
 CONTENTS vii

11.18 Spreadsheet solution for strain gauge rosette 317 14.11 Stresses in a rotor of varying thickness with
11.19 Relationships between the elastic constants 319 rim loading 403
11.20 Stress and strain transformations in composites 321 Summary 407
11.21 Analysis of a lamina 322 Problems 407
11.22 Analysis of a laminate 329
11.23 In-plane behaviour of a symmetric laminate 330
11.24 Flexural behaviour of a symmetric laminate 334 15 Elementary Plasticity 413
Summary 334
References 335 15.1 Plastic bending of beams: plastic moment 413
Problems 335 15.2 Plastic collapse of beams 418
15.3. Plastic torsion of shafts: plastic torque 423
12 Yield Criteria and Stress Concentration 340 15.4 Plasticity in a pressurized thick-walled cylinder 425
15.5 Plasticity in a rotating thin disc 429
12.1 Yield criteria: ductile materials 340 15.6 Residual stress distributions after plastic
12.2 Fracture criteria: brittle materials 348 deformation 43]
12.3 Strength of laminates 350 Summary 436
12.4 Concepts of stress concentration 351 Problems 436
12.5 Concentrated loads and contact stresses 353
12.6 Geometrical discontinuities 356
12.7 Yield and plastic stress concentration factors 362
Summary 363
16 Thin Plates and Shells 439
References 363 16.1. Assumptions for small deflection of thin
Problems 363 plates 439
16.2 Relationships between moments and
13 Variation of Stress and Strain 367 curvatures for pure bending 439
16.3 Relationships between bending moment
13.1 Equilibrium equations: plane stress Cartesian
and bending stress 444
co-ordinates 367
16.4 Relationships between load, shear force and
13.2 Equilibrium equations: plane stress
bending moment 444
cylindrical co-ordinates 368
16.5 Relationships between deflection, slope and
13.3. Strain in terms of displacement: Cartesian
loading 445
co-ordinates 370
16.6 Plate subjected to uniform pressure 446
13.4 Strain in terms of displacement: cylindrical
16.7 Plate with central circular hole 448
co-ordinates 371
16.8 Solid plate with central concentrated force 450
13.5 Compatibility equations: Cartesian
16.9 Other forms of loading and boundary
co-ordinates 373
condition 451
13.6 Compatibility equations: cylindrical
16.10 Axi-symmetrical thin shells 453
co-ordinates 374
16.11 Local bending stresses in thin shells 454
13.7. Equilibrium in terms of displacement:
16.12 Bending in a cylindrical storage tank 458
plane stress Cartesian co-ordinates SH)
Summary 459
Summary BH,
Bibliography 460
Problems 378
Problems 460
14 Applications of the Equilibrium and
Strain—Displacement Relationships 379
17 Finite Element Method 463
14.1 Boundary conditions 379
14.2 Shear stress in a beam 380 17.1 Principle of finite element method 463
14.3. Transverse normal stress in a beam 381 17.2 Analysis of uniaxial bars 464
14.4 Stress distribution in a pressurized 17.3 Analysis of frameworks 469
thick-walled cylinder 382 17.4 Analysis of beam elements 476
14.5 Methods of containing high pressure 390 17.5 Analysis of continua 483
14.6 Stresses set up by a shrink-fit assembly 391 17.6 Stiffness matrix for a triangular element 484
14.7 Stress distribution in a pressurized 17.7 Effect of mesh density 488
compound cylinder 392 17.8 Other continuum elements 489
14.8 Spreadsheet solution for stress in a Summary 490
compound cylinder 395 Reference 490)
14.9 Stress distribution in a thin rotating disc 397 Bibliography 490
14.10 Rotational speed for initial yielding 401 Problems 49]
viii CONTENTS

18 Tension, Compression, Torsion and 20.5 Fracture mechanics for fatigue 554
Hardness 493 20.6 Influential factors
20.7 Cumulative damage
Stress—strain response in a uniaxial tension
20.8 Failure under multi-axial cyclic stresses
test 493
20.9 Fatigue of plastics and composites
Stress—strain response in a compression test 497
Summary
Stress—strain response in a torsion test 498
References
Plastic overstrain and hysteresis 500
Bibliography
Hardness measurement 500
Problems
Hardness of viscoelastic materials 506
Factors influencing strength, ductility and
hardness 506 21 Creep and Viscoelasticity
Summary 510 21.1 Stress—strain—time—temperature relationships
Reference 511 21.2 Empirical representations of creep behaviour
Bibliography 511 21.3 Creep-rupture testing
Problems 512 21.4 ‘Tension creep test equipment
21.5 Creep during pure bending of a beam
19 Fracture Mechanics 514 21.6 Creep under multi-axial stresses
21.7 Stress relaxation
19.1 Fracture concepts Sle: 21.8 Stress relaxation in a bolt
19.2 Linear-elastic fracture mechanics 515 21.9 Creep during variable load or temperature
19.3. Strain energy release rate 516 21.10 Creep-fatigue interaction
19.4 Stress intensity factor 518 21.11 Viscoelasticity
19.5 Modes of crack tip deformation 521 21.12 Creep behaviour of plastics
19.6 Experimental determination of critical 21.13 Designing for creep in plastics
stress intensity factor 521 21.14 Creep rupture of plastics
19.7 Fracture mechanics for ductile materials D2 Summary
19.8 Toughness measurement by impact testing O34 References
19.9 Relationship between impact testing and Bibliography
fracture mechanics 535 Problems
Summary 537
References 537
Bibliography 538 Appendix A: Properties of Areas
Problems 538
Appendix B: Introduction to Matrix Algebra
20 Fatigue 544 Appendix C: Table of Mechanical Properties of
20.1 Forms of stress cycle 544 Engineering Materials
20.2 Test methods 546
20.3 Fatigue data 548 Appendix D:; Answers to Problems
20.4 Micromechanisms of fatigue: initiation and
propagation 552 Index 624
Preface to the Second Edition
The historical development of this text was outlined chapters and omitted the superficial and to some
in the Preface to the previous edition. During the extent outdated solution methods based on scale
nine years since its publication there have been no drawings.
major changes in the theory of the subject but there The remaining chapters have benefited from a
have been advances in some of the manipulative tools complete overhaul and the chapter on Finite
which can be used to solve problems related to the Elements has been almost totally re-written.
strength of materials. In addition, with experience Throughout the text the reader is introduced to the
and hindsight one can always improve the presenta- concept of creating a spreadsheet representation of a
tion of material to make it more digestible to the problem in strength of materials. Once such a
reader. solution has been formulated it is then a relatively
In this new edition we have engaged the services of simple matter to study the effects of changes in
a new author Dr Cecil Armstrong to assist in a variables, to plot graphs of key variables and to
complete update of the text. New features of the book optimize solutions. This is shown to be a very
include the use of versatile solution techniques based powerful solution methodology which provides a
on spreadsheets, the use of colour to enhance completely new insight to each problem and open up
diagrams and provide emphasis of key concepts, an a whole new approach to problem solving in many
introduction to matrix methods as a powerful tool in subject areas.
engineering analysis, and the incorporation of The authors derived a great deal of pleasure in
additional practical problems to illustrate the working on this new version of the text. However,
application of the theory. although he participated fully in the revision of the
The essential structure and content of Chapters 1— book, regrettably Peter Benham died before he was
9 has not been changed although regular users of the able to see the new book in print. We know that the
previous text will see subtle changes in the mode of many young people who studied under him, and the
presentation to make this important fundamental readers all over the world who benefited from his
information more meaningful to students. Chapter 10 books, were greatly saddened by his death. It is our
from the original text has been removed because it hope that this new edition, about which he was very
was felt that modern computer programs provide excited and proud, will be a fitting tribute to his
powerful tools for the solutions of problems in memory.
structures. In a general text of this type it 1s
impossible to do justice to the wide range of
methods available to analyse structures. Hence, we R. J. CRAWFORD
have moved the important fundamental information C. G. ARMSTRONG
on tension coefficients, energy methods, etc. to other 10 November 1995
Preface to the First Edition
The S.I. edition of Mechanics ofSolids and Structures syllabus of a university or polytechnic degree course in
by P. P. Benham and F. V. Warnock was first engineering, or the examinations of the engineering
published in 1973. It appears to have been very well Council, C.N.A.A. etc.
received over the ensuing period. This preface is Although there is a fairly natural ordering of the
therefore written both for those who are familiar with material there is some scope for variation and lecturers
the past text and also those who are approaching this will have their own particular detailed preferences.
subject for the first time. Although the subject matter As in the previous text, the first eleven chapters are
is still basically the same today as it has been for concerned with forces and displacements in statically-
decades, there are a few developing topics which have determinate and indeterminate components and struc-
been introduced into undergraduate courses such as tures, and the analysis of uniaxial stress and strain due to
finite element analysis, fracture mechanics and fibre various forms of loading such as bending, torsion,
composite materials. In addition, style of presentation pressure and temperature change. The basic concepts of
and illustrations in engineering texts have changed strain energy (Ch. 9) and elastic stability (Ch. 11) are
for the better and certain limitations of the previous also introduced. In Chapter 12 a study is made of two-
edition, e.g. the number of problems and worked dimensional states of stress and strain with special
examples, needed to be rectified. Professor Warnock emphasis on principal stresses and the analysis of strain
died in 1976 after a period of happy retirement and measurements using strain gauges. Chapter 13 combines
so it was left to the other author and the publisher to two chapters of the previous book and brings together
take the initiative to construct a new textbook. the topics of yield prediction and stress concentration
In order to provide fresh thinking and reduce the which are of such importance in design.
time of rewriting Dr Roy Crawford, Reader in Also included in these two chapters is an elementary
Mechanical Engineering at the Queen’s University of introduction to the stress analysis and failure of fibre
Belfast, was invited, and kindly agreed, to join the composite materials. These relatively new advanced
project as a co-author. Although the present text might structural materials are becoming increasingly used,
be regarded as a further edition of the original book the particularly in the aerospace industry, and it is essential
new authorship team preferred to make a completely for engineers to receive a basic introduction to them.
fresh start. This is reflected in the change of title which These thirteen chapters constitute the bulk of the
is widely used as an alternative to Mechanics of Solids. syllabuses covered in first and second-year courses.
The dropping of the reference to structures does not Four of the next five chapters appeared in the
imply any reduction in that topic as will be seen in the previous text and deal with more advanced or
contents. specialized topics such as thick-walled pressure
In order that the book should not become any larger vessels, rotors, thin plates and shells and post-yield or
with the proposed expansion of material in some plastic behaviour, which will probably occupy part of
chapters, it was decided that the previous three final-year courses.
chapters on experimental stress analysis should be One essential new addition is an introductory chapter
omitted as there are several excellent texts in this field. on finite element analysis. It may seem presumptuous
The retention of that part of the book dealing with even to attempt an introduction to such a broad subject
mechanical properties of materials for design was in one chapter, but it is an attempt to provide initial
regarded as important even though there are also encouragement and confidence to proceed to the
specialized texts in this area. complete texts on finite elements.
The main part (Ch. 1-18) of this new book of course Chapters 19 to 22 cover much the same ground as in
still deals with the basic subject of Solid Mechanics, or the previous text, but have been brought up to date
Mechanics of Materials, whichever title one may prefer, particularly in relation to fracture mechanics. Since
being the study of equilibrium and displacement these chapters have such importance in relation to
systems in engineering components and structures to design, a number of worked examples have been
enable designs to be effected in terms of stress and strain introduced, together with problems at the end of each
and the selection of materials. These eighteen chapters chapter. Bibliographies have still been included for
cover virtually all that is required in the three-year further reading as required.
 PREPACESTOVHEEIRST EDITION xi

The first Appendix covers the essential material on development in the relevant chapter. It is most
properties of areas. The second deals with the simple important not to approach solutions on the basis of
principles of matrix algebra. A useful table of plucking the ‘appropriate formula’ out of the text,
mechanical properties is provided in the third inserting the numbers, and manipulating a calculator!
Appendix. Every effort has been made by the authors to ensure
One of the recommendations of the Finniston Report accuracy of text and solutions, but lengthy experience
to higher education was that theory should be backed up demonstrates human fallibility in this respect. When
by more practical industrial applications. In this context errors subsequently come to light they will be corrected
the authors have attempted to incorporate into the at the next reprinting and readers’ patience and
worked examples and problems at the end of each comments will be appreciated!
chapter realistic engineering situations apart from the Some use has been made of data and diagrams from
conventional examination-type applications of theory. other published literature and, in addition to the
There had been a number of enquiries for a solutions individual references, the authors wish to make grateful
manual for the previous text and this can be very helpful acknowledgement to all persons and organizations
to both lecturer and student. Consequently this text is concerned.
accompanied by another volume which contains worked
solutions to nearly 300 problems. The manual should be P. P. BENHAM
used alongside the main text, so that steps in each R. J. CRAWFORD
solution can be referred back to the appropriate 1987
Notation
angle, coefficient of thermal expansion
angle
shear strain, surface energy per unit area
deflection, displacement
direct strain
efficiency, viscosity
angle, angle of twist, co-ordinate
lack of fit
Poisson’s ratio
radius of curvature, density
direct stress
shear stress
angle, co-ordinate, stress function
el
ESS.
Sy
SS
SS
eS
9 angular velocity

area
complementary energy
diameter
Young’s modulus of elasticity
force
shear or rigidity modulus of elasticity, strain energy release rate
force
second moment of area, product moment of area
polar second moment of area
a.
CY
ASS
ayia
per
hoe
Sha
De bulk modulus of elasticity, fatigue strength factor, stress concentration
factor, stress intensity factor
length
bending moment
Zeb number of stress cycles, speed of rotation
force
shear force
force, radius of curvature, stress ratio
cyclic stress
temperature, torque
strain energy
volume
weight, load
CHYRLH'D
MEN body force
body force
ilies body force, section modulus

a area, distance, crack length

b breadth, distance, crack length
‘ distance
d depth, diameter
eccentricity, base of Napierian logarithms
g gravitational constant
h distance
 NOTATION xili

number of joints
diameter ratio of cylinder
length
mass, modular ratio, number of members
number
pressure
shear flow
co-ordinate, radius, radius of gyration
length
thickness, time
displacement in the x- or r-direction
deflection, displacement in the y- or 6-direction, velocity
displacement in the z-direction, load intensity
co-ordinate, distance
co-ordinate, distance
SSeS
Ss
ts
Sass:
§Saas
eS co-ordinate, distance

It should be noted that a number of these symbols have also been used to
denote constants in various equations.
ws — >A
aa _—
oe Gas a a at ey
ee ee :
a ee (as . & 2 Se
yas ¢ —
eum
cy aa | —_— + a)
a
Vy) eaee Sie?
= ae
: : 7 . = ad
+ SS ; a.
NE 2 See 7 —

= =) <a ,
i
 CHAPTER
Statically Determinate Force
Systems

Structural and solid-body mechanics are concerned with analysing the effects of
applied loads. These are external to the material of the structure or body and result in
internal reacting forces, together with deformations and displacements, conforming
to the principles of Newtonian mechanics. Hence a familiarity with the principles of
statics, the cornerstone of which is the concept of equilibrium of forces, is essential.
A force system is said to be statically determinate if the internal forces can be
calculated by considering only the forces acting on the system.
Forces result in four basic forms of deformation or displacement of structures or
solid bodies and these are tension, compression, bending, and twisting.
The equilibrium conditions in these situations are discussed so that the forces
may be determined for simple engineering examples.
eS ee

1.1 Revision of statics

A particle is in a state of equilibrium if the resultant force and moment
acting on it are zero. This hypothesis can be extended to clusters of particles
that interact with each other with equal and opposite forces but have no
overall resultant. Thus it is evident that solid bodies, structures, or any
subdivided part, will be in equilibrium if the resultant of all external forces
and the resultant of all moments are zero. This may be expressed
mathematically in the following six equations which relate to Cartesian co-
ordinate axes x, y and z.

oF, =0
z7,=0| [1.1]
Lie 0

where /’,, Ff, and F, represent the components of force vectors in the co-
ordinate directions.

DM, =0
za =o (1.2]
UM, =0
where M,, M, and M, are components of moment vectors caused by the
external forces acting about the axes wx, y, z.
The above six equations are the necessary and sufficient conditions for
equilibrium of a body.
If the forces all act in one plane, say z = 0, then

DF, =0M, =IM, =0

 MECHANICS OF ENGINEERING MATERIALS

are automatically satisfied and the equilibrium conditions to be satisfied in a

two-dimensional system are

SF. =0
DF, =0 [1.3]
=M, =0
Forces and moments are vector quantities and may be resolved into
components; that is to say, a force or a moment of a certain magnitude and
direction may be replaced and exactly represented by two or more
components of different magnitudes and in different directions.
Considering firstly the two-dimensional case shown in Fig. 1.1, the force
F may be replaced by the two components F’, and F,, provided that

Fe= ee
1.4
Fy, = Fsina |

Fig. 11

Note that throughout this book, externally applied forces will be shown
coloured. Internal forces, that is those within a structural element, will be
shown in black.
If the force F were arbitrarily oriented with respect to three axes x, y, z
as in Fig. 1.2, then it could be replaced or represented by the following
components:
EVE cosa
Per Coan) [1.5]
PSC Gg

Fig. 1.2
 STATICALLY DETERMINATE FORCE SYSTEMS 3

Fig. 13

A couple or moment vector about an axis can similarly be resolved into a

representative system of component vectors about other axes, as shown in
Fig. 1.3 and represented by the following equations:

M, = Mcosa
iby icon | [1.6]
M, = Mcosy

Example 1.1
A set of concurrent forces F; are defined by their components F,, Fy, F, in KN as

F, Fy Fy
Fi 80 -20 -40
F, -40 60 -80
F; 100 -20 30
F, -30 10 40
Calculate the magnitude and direction of the resultant of this force system.

F, = DF, = 80—40+ 100

— 30 = 110kN

oT | M os ] = 20 60i—= 20-210
30kN

F, => F, = —40 — 80 + 30 +40 = —S0kN

The magnitude of the resultant force 1s
PFI eesaleLes + Fe)
[VAG ee = /(110° 2 + 30° 2 + (—50))es= 124 kN

to three significant figures. The angles between the resultant force and the
axes shown in Fig. 1.2 can now be found using

F 110
COS a=Se= 194 (VO! , SO a == 7.

F, 30
cos3 = > = — = 0,241, so B = 76.1°
Po 124
 MECHANICS OF ENGINEERING MATERIALS

F —50
COS 34 ¥ Fpf 124 402 so y = —66.3°
0.402,

Since all the forces are concurrent, none of them have any moment about
the common point through which they all act and there is therefore no
resultant moment about that point.

Spreadsheet solution In engineering practice there are few investigations where the problem is as
clearly defined and straightforward as that given above. There is usually a
degree of uncertainty about the supplied information, some data may be
missing and the numerical computations are significantly more complicated.
Thus laborious calculations are needed and the sensitivity of the results to
variations in the given or assumed information has to be evaluated. This has
lead many engineers to develop computer programs in languages such as
Fortran, Basic or C to automate the required calculations.
In this book, selected numerical problems will be solved using
commercial programming and modelling tools called spreadsheets.
Probably the best known of these are Lotus 1—2—3, Borland Quattro and
Microsoft Excel. A computer shreadsheet can be considered as a grid of cells
which can contain text labels, numbers or formuiae which may refer to
numbers or numerical results in other cells. If the cell contains a formula,
and a number is changed in any cell to which that formula refers, then the
formula is automatically recalculated and the updated result is displayed. A
range of alternatives can be quickly evaluated by entering different numbers

Fig. 1.4
i
2
Se F2) mimi simaremainat inner
me pa

_ F3 errr creer emaent veceenfeeder Vneiyomevvsigeanecfineteyyss nytenmneonnonenennnnspmetonnsnent

bec ual F4 ol)
6
ce FR +B2+B3+B4+B5. _ @SUM(D2..D5), __@SQRT(B742+C742+D742)
is
9 Direction Cosines
alpha gamma

ef ___ Magnitude
ee
3
4
5
Bag
Bs
8
9 Direction Cosines
10
14
 STATICALLY DETERMINATE FORCE SYSTEMS 5

in a given cell. Modern packages have sophisticated facilities for the

graphical display of results, ‘What-if evaluation of a range of alternatives,
querying databases and optimization. Although these programs were initially
developed for financial modelling, many engineers and students now use
these packages as sophisticated programmable calculators.
Figure 1.4(a@) shows the data and formulae required to solve this
problem. These were entered in a Quattro spreadsheet but they will also
work with Lotus 1-2-3 or Microsoft Excel. Figure 1.4() shows the
resulting display in the spreadsheet. Normally the formulae will not be
visible and only the result will be displayed. Throughout this book, the cells
in which the user should input data have been highlighted. The other cells
contain either formulae or text labels and should not be overwritten.
As can be seen in Fig. 1.4(a), the total force in the x-direction is found in
cell B7 by adding up the contents of cells B2, B3, B4 and B5. An even more
convenient technique for summing up the y-components is shown in cell
C7, where the built-in function @SUM is used to total the contents of cells
C2 to C5. Within the spreadsheet program the Edit, Copy and Paste
commands can be used to copy the same formula to cell D7, which then
sums the z-components. Individual cells or groups of cells can be identified
with arrow keys or mouse clicks — it is not necessary to type the cell
references. The resultant force is found using the three-dimensional
equivalent of eqn. [1.7] in cell E7, using the built-in function @SQRT to
calculate the square root and ’\ to indicate that a number is to be raised to a
power.
Once these formulae are entered, any change to any component of the
four forces will cause an immediate recalculation of the magnitude of the
resultant force and the directions. If only three forces are to be summed,
deleting all the components of one force will give the correct answer. If a
larger number of forces is to be summed, extra rows can be inserted above
row 6 and the SUM formula in the current row 7 adjusted to the new range
of cells.

1.2 Resultant force and

It is sometimes more convenient to replace a system of applied forces by a
moment
resultant which of course must have the same effect as those forces.
Considering a two-dimensional case as illustrated in Fig. 1.5, then the most
general solution is obtained by choosing any point A through which the
resultant can act. Then the total force components in the co-ordinate
directions are

i
=f [1.7]
Fy=XF,
and the resultant force is given by

Bean (rier Fk) [1.8]

However, this is not sufficient in itself since the moment due to the
forces must be represented. This is done by having a couple acting about A
such that

M="DM, [1.9]
 MECHANICS OF ENGINEERING MATERIALS

Fig. 1.5

(a) (b)

In general, then, any system of forces can be replaced by a resultant force

through, and a couple about, any chosen point.
The equivalent solution for a three-dimensionai system of forces is
similarly a couple and a resultant force whose direction is parallel to the axis
of the couple.
One of the most useful constructions in force analysis is termed the
triangle offorces. If a body is acted on by three forces then, for equilibrium to
exist, these must act through a common point or else there will exist a
couple about the point causing the body to rotate. In addition the magnitude
and direction of the three force vectors must be such as to form a closed
triangle as shown in Fig. 1.6.

Fig. 1.6

1.3 Types of structural

Structures are made up of a series of members of regular shape that have a
and solid-body
particular function for load carrying. The shape and function are, through
components usage, implied in the name attached to the member. The first group is
concerned with carrying loads parallel to a longitudinal axis. Examples are
shown in Fig. 1.7. A member which prevents two parts of a structure from
moving apart is subjected to a pull at each end, or tensile force, and is
termed a fie (a). Conversely a slender member which prevents parts of a
structure moving towards each other is under compressive force and is
termed a strut (b). A vertical member which is perhaps not too slender and
supports some of the mass of the structure is called a column (c). A cable (d)
is a generally recognized term for a flexible string under tension which
connects two bodies. It cannot supply resistance to bending action.
One of the most important of structural members is that which is
frequently supported horizontally and carries transverse loading. This is
 STATICALLY DETERMINATE FORCE SYSTEMS 7

ZA
a> —> GEE
(a) Tie (b) Strut

fone
>»
(d) Cable
(c) Column

(k) Shell (1) Shait

(g) Beam-column
\ (h) Arch

known as a beam (e), a common special case of which is termed a cantilever

(/), where one end is fixed and provides all the necessary support. A beam—
column (g), as the name implies, combines the separate functions already
described. The arch (h) has the same function as the beam or beam—column,
but is curved in shape.
The filling-in and carrying of load over an area or space are achieved by
flat slabs or plates (1), by panels (7) and also by shells (k), which are curved
versions of the former.
The transmission of torque and twist is achieved through a member
which is frequently termed a shaft (/).
The members described above can have a variety of cross-sectional
shapes depending on the particular type of loading to be carried. Some
typical cross-sections are illustrated in Fig. 1.8. Sections of this type are
generally readily available from stock.

fig. 18

Angle Channel I-section T-section Z-section Tube

1.4 Types of support and

The applied loading on a framework, beam or column is transmitted to the
connection for structural supports which will provide the required reacting forces to maintain overall
components equilibrium. Examples of supports of various kinds suitable to react to
loading in a plane (two dimensions) are shown in Fig. 1.9. In the
accompanying table the possible displacement and reacting force
components are indicated.
8 MECHANICS OF ENGINEERING MATERIALS

Fig. 19 Type of support Equivalent force system

(a) Built-in fixed support

\
(b) Pin connection

\\
(c) Roller support

N
(d) Sliding support

Displacement Reacting force

x y 6 R, R, M,

(a) Built-in or fixed support JV JV J

() Pin connection J Jv Jv
(c) Roller support J V Vv
(d) Sliding support Vv Jv Vv
 STATICALLY DETERMINATE FORCE SYSTEMS 9

Fig. 1.10

(a) Bolting (b) Welding

The separate members of a structural framework are joined together by

bolting, riveting or welding, two examples of which are shown in Fig. 1.10.
Now, if these joints were ideally stiff, when the members of the framework
were deformed under load, the angles between the members at the joint
would not change. This would also imply that the joint was capable of
transmitting a couple. However, calculations for a complete structure on this
basis would become rather involved and tedious.
It is found in practice that there is some degree of rotation between
members at a joint owing to the elasticity of the system. Furthermore, it has
been shown that it is not unreasonable, for purposes of calculation, to
assume that these joints may be represented by a simple ball and socket or
pin in a hole. Even with this arrangement, which of course cannot transmit a
couple or bending moment (other than by friction, which is ignored),
deformations of the members are relatively small. Consequently changes in
angle at the joints are also small, which is why this approximation is not
unreasonable when applied to the actual joints. Thus it is common practice
when calculating the forces in the members of a framework to assume that all
joints are pinned.

1.5 Statical determinacy

In general, structural or solid-body mechanics involves determination of
unknown forces within the structure or body. The approach taken in this
analysis depends initially on whether the system under consideration is
‘statically determinate’ or ‘statically indeterminate’.
If the number of equations available from statements of equilibrium is
the same as the number of unknown forces (including reactions) then the
problem is statically determinate. If the number of unknown reactions or
internal forces in the structure or component is greater than the number of
equilibrium equations available, then the problem is said to be statically
indeterminate.
In order to solve a statically indeterminate problem it is necessary to
consider additional equations relating to the displacement or deformation of
the body.
The above statements are quite general and apply throughout this text.

1.6 Free-body diagrams

When commencing to analyse any force system acting on a component or
structure it is essential firstly to have a diagrain showing the forces acting. If
the structure or part of it is separated from its surroundings and the possible
reactions are inserted then a diagram of this system is called a free-body
diagram. Examples of this are shown in Fig. 1.11.
10 MECHANICS OF ENGINEERING MATERIALS

Fig. Ll
YAM B A B

\
*

E D G E D Sit

(a) Plane frame attached (b) Free-body diagram (c) Free-body diagram
to wall at A and E of section ABDE for joint C

1.7 Determinacy criteria

The principles of statical determinacy will now be considered in relation to
for structures
plane and space-frame structures. There are three classes of frame or truss,
in concept, although one is not of practical interest:

(a) Under-stiff: If there are more equilibrium equations than unknown

forces or reactions the system is unstable and is not a structure but a
mechanism.
(b) Just-stiff- This is the statically determinate case for which there are the
same number of equilibrium equations as unknown forces. If any
member is removed then a part or the whole of the frame will collapse.
(c) Over-stiff. This is the statically indeterminate case in which there are
more unknown forces than available equilibrium equations. There is at
least one member more than is required for the frame to be just stiff.

Figure 1.12 shows an example of each of the three conditions.

Remembering that each joint is pinned, then it is clear that in (a) the
central square of members is not stable unless there is one diagonal member
inserted. The number and arrangement of the members in case () is quite
correct for the ‘just-stiff or statically determinate condition. In case (c) in
contrast to (a) the central square has two diagonal braces one or other of
which is unnecessary and hence the frame is ‘over-stiff or statically
indeterminate.
It is useful to express the three cases in the form of mathematical criteria.
Let the number of joints, including support points, in a frame be j, the
number of members m, and the number of reactions r. Now, for a space
frame there are three equilibrium equations applicable to each joint, namely
UF, =0, uF, = 0, & F, = 0; hence there are 37 equations to determine
m +r unknown forces and reactions, and the statically determinate case is

{> PAR
Fig. 1.12

(a) Under-stiff (b) Just-stiff (c) Over-stiff

 STATICALLY DETERMINATE FORCE SYSTEMS 11

represented by

mt [1.10]
When m+ r < 37 the members form a mechanism, and for m+r > 37 the
frame is over-stiff, or redundant, and therefore statically indeterminate.
When determining the reactions there are six equations for overall
equilibrium of a framework:

OF, =DF,=DF,=0

»M, ==M, ==M, =0

Therefore the minimum value for 7, when using the above criteria to allow
for any general loading system, is six.
For frames lying only in one plane there are only two equilibrium
equations at each joint and so the relationships comparable with the above
are
le ee |
ate ek [1.11]
m+r> dy
The minimum value for r in these expressions must be three for general
forms of loading.
The above criterion for a just-stiff frame is a necessary but not a
sufficient condition, since the arrangement of the members might still not
provide the required stiffness.

Bellet cg Examine the plane frames illustrated in Fig. 1.13. State the class of each and where
members should be inserted or removed to make each statically determinate. Also
indicate any redundant reactions.

Fig. 1.13

X ss

(b)

(¢) (d)
 i MECHANICS OF ENGINEERING MATERIALS

(a) m= 18, r = 3,7 = 10: hence m+r > 2; and the frame is over-stiff or
redundant. Any member may be removed from the central hexagon
structure. None of the members may be removed from the apexes.
(6) m=14,r=4, 7 = 9: hence m +r = 2) and the frame is just stiff and
statically determinate.
(c) m= 8,r= 3,7 = 6: hence m+ r < 27, which constitutes a mechanism.
To make statically determinate insert a member between any pair of
unconnected joints.
(4d) m=15, r=4, 7 =8: hence m+r > 2j, which is redundant both in
members and reactions. Hence remove either the left or right support
and a diagonal member from each square.

1.8 Determination of
The members of pin-jointed plane or space frameworks can only carry axial
axial forces by equilibrium forces, i.e. tension or compression (Fig. 1.14). These may be determined by
statements considering the equilibrium of various parts of the structure as ‘free bodies’.

Plane pin-jointed Equilibrium at joints

frames At any joint in a plane frame two equilibrium equations apply, namely

uF,=0 and »F, = 0 (for a frame lying in the z-plane)

SS —————————

Tension Hence only two unknown forces can be determined. The method therefore
entails making a free-body diagram centred on each joint at which there are
eC
TO only two unknowns. The forces are then resolved in the x- and y-directions
Compression so that the above two equations can be applied. It will probably be necessary
at first to determine support reactions by considering equilibrium of the
Fig. 114
whole frame. This generally gives at least one joint where there is a known
reaction and only two members having unknown forces from which to start
the analysis. The method will be illustrated by the following example.

Example 1.3
Determine the magnitude and the type of force in each member of the plane pin-
jointed frame shown in Fig. 1.15.

Fig. 1.15

In Fig. 1.15, in order to save the space and having to draw the complete
frame twice, once on its supports and a second time as a free body away from
the supports with the reactions in place, the possible reactions have been
placed directly onto the configuration diagram.
 STATICALLY DETERMINATE FORCE SYSTEMS 13
aN EEE

Considering the equilibrium of the whuie frame and taking moments

about joint F,
» M, = (4 x 5000) — 6R4 — (2 x 1000) =0 so that
R4 = 3000 N
For horizontal equilibrium,
Lf. = 0— 1000—H, sothat H-—1000N
For vertical equilibrium,
UF,=0 gives R4+Rp —5000=0
Hence
Rr = 2000N
It may be seen from Fig. 1.14 that members which are in tension have
internal forces which act away from the joint. It is recommended that it is
assumed initially that each member 1s subjected to tension. If the analysis shows
that the internal force is negative this simply means that the member is in
compression.
The only joints at which there are two unknowns are A and F. Start, say,
at A and proceed as follows:
Jot A The free-body diagram for this joint is as shown in Fig. 1.16(a).
uF, =0 gives Fy4gsin45° + 3000 = 0

Internai force
Fig. 1.16 in AB
Y Fag= 4243

|
| Compression
Fag
| /
fs | /
ae | Fas Internal force
A Pinel op Fag
AC in AC
Use ene A
Fac at A Fac= 3000
Tension

3000 3000

(a) Assumed forces (b) Actual forces on pin at A and members AB and AC
on pin at A

Hence
F yp = —4243N

The negative sign shows that AB is in compression and not tension.

~F,=0 gives Fy4gcos45° + Fyc =0

Hence
Fc = 4243 cos 45° = 3000N
The next step as to be at joint B (rather than C).
14 MECHANICS OF ENGINEERING MATERIALS

Joint B- (Fig. 1.17)

LF, = Fra cos45” + Fac = 0

Fc = 4243 cos 45° = 3000N

Fig. 1.17
See = F gp aie 1000 — Fp sin 45° = 0

F gp = —4243 sin 45° — 1000 = —4000N (compression)

Joint C There are now only two unknowns at this joint, Pcp and Fp.
(Fig. 1.18)

a — For — 5000 + Fon cos 45° = 0

Fop = (—3000 + 5000) \/2 = 2828N

» F, = For — Fos + Fcp cos 45° = 0

Fig. 1.18
For = 3000 — 2828 cos 45° = 1000N
Continuing the above process for joints E and D gives
Fip =O For=1000N” For = —2828N
The final force distribution in the frame is shown in Fig. 1.19.

Fig. 1.19 Bd B
O
4000 D
O

3000 5000 2000

As the force in DE is zero one might ask whether that member is

required. With this particular applied load, in fact, DE could be removed
quite safely, since the two members CE and EF really act as one continuous
member CEF in tension. However, if the applied load at C acted upwards
the loads in all the members would have the same magnitude as before but
would be reversed in sense and then CE and EF would be in compression.
While ED was in position even without any force in it, it would keep CE and
EF in line, whereas if ED were removed the joint E would move vertically
due to the slightest misalignment and result in collapse.

Matrix solutions An alternative and very powerful technique for solving problems of this type
involves the use of matrix methods, which are described in Appendix B.
These techniques are central to computer-aided analysis techniques such as
 STATICALLY DETERMINATE FORCE SYSTEMS 15
Se

Table 1.1

Joint F 4p Fc Fc Fp Fep For For F pp Fer Ry Ar Rr External load

A 0.707 1
0.707 1

B —0.707 1 1000
—0.707 =I

G =| 0.707 1
1 0.707 —5000

D =I —0.707 0.707
—0.707 —] —0.707
18. =! 1
1
Ie S707 =I! =I

the finite element method (Chapter 17), but a brief description of the
solution of Example 1.3 will be given here. Those unfamiliar with matrix
methods may wish to skip this section.
The equations of joint equilibrium in Figs. 1.15 to 1.19 can be
summarized in tabular form in Table 1.1. The twelve rows represent the x
and y equations of: force equilibrium for each of the six joints A-F. The
columns contain the coefficients multiplying the nine unknown member
forces and three reactions, plus one column for the known external loads.
An alternative way of writing these equations is as a matrix equation
where a coefficient matrix [A] is multiplied by a column vector {x}
containing the unknown member forces and reactions. The resulting forces
are summed with the known external loads {y} as

[A]{x} + fy} = {0}

To find {x}, which is a column vector containing the unknowns F'4z, F4c,
etc., the equivalent to a division is required. This is known as the matrix
inverse; it can be found using standard numerical techniques and matrix
multiplication and inversion routines are available in spreadsheets. Given
the inverse matrix [4]~', the unknowns can be found by multiplying the
inverse matrix by the known quantities, 1.e.

{x} = [A {-y}
Members which are not connected to a given pin have zero coefficients
multiplying the relevant member force. Members which are connected to a
given pin have non-zero coefficients multiplying the member force. For
horizontal equilibrium these are the cosine of the angle at which the member
leaves the joint. For vertical equilibrium the coefficient is the sine of the
member angle. The right-hand side of each equation is the negative of the
component of any external force acting at that joint.
16 MECHANICS OF ENGINEERING MATERIALS

i —————————

By this technique the series of equations at each joint described in Figs

1.16 to 1.18 become the matrix equation

0 eae 0 0 0 0 ODO Pap 0

0) 0 0 0) One 00 Fc 0)
SSS 0 0 0 0 Oe 20 a0 Frc —1000
0 0 O 0) OF OO 0 Fp 0
0.707 | 0) 0) OP O20 Fep 0
0.707 O 0 0 Vriae <0 00 Pepi Nabe 5000
O70) 0 CO O07 ae 00 For { 0)
—0.707 O -1 -0.707 0 0 0 O For 0
0 oe 0 TU; 00 Pp 0
0 0 1 0 OF Ore 26 Ry 0
0 O° 0 2=0.707--1. 0) =F 0 Hr 0
Ose,
(SS
Sto
OS aa
GS
S01. 0 0 0 0.707 O) On aget Rr 0

This equation is effectively [A]{x}= —{y} and so the solution for the
unknown forces in column vector {x} is given by

{x} = [A)'{-y}
In expanded form, the solution is

Fp —4243
Fac 3000
Frc 3000
Frp —4000
Fon 2828
Pen 1000
For ( 0
F pp —2828
Fre 1000
R4 3000
Hr 1000
Rr 2000

Equilibrium offrame sections

Returning to the analysis of the framework in Fig. 1.15, the alternative to
considering a joint as a free body in equilibrium is to consider the
equilibrium of a single member or group of members. This may have
advantages in effort compared with analysing the equilibrium at every joint.
In Fig. 1.20 cutting along the line XX provides either a free body of member
AB to the left of the section or a free body of the remainder of the frame to
the right of the section. Section YY splits the frame into two parts either of
which could be regarded as a free body the equilibrium of which would give
the forces in BD, CD and CE. In isolating part of the frame by a section it
must be remembered that there are only three equilibrium equations that
can be written (UF, = U Fy = UM, =0); hence only three unknown
forces can be found for a particular ‘section’ and free body. Considering
 STATICALLY DETERMINATE FORCE SYSTEMS 17

Fig. 1.20

section XX then, as shown in Fig. 1.21, insert a force vector at the cut end of
each member in the tensile sense.
Equilibrium for the free body requires that
1000 Bay, Fep
DF,=0 SF,=0 DM, =0
Hence

Fy = 3000 — Feo = 0

Fac = 3000 N
3000
Moments about A give ) M, = (2 x 1000) + 2Fgp + 2Fac = 0
Fig. 1.21 F’gp = —4000 N_ (compression)
Finally, & F, = Fep + F4c + 1000 =0

F4c = 3000 N
These forces are the same as those obtained by taking equilibrium at each
joint or using the matrix solution. The same results would have been

Fig. 122 Free-body diagrams of

parts of a fork-lift mechanism

(a) (b)
18 MECHANICS OF ENGINEERING MATERIALS

obtained if the three equilibrium equations had been written for the
remainder of the frame to the right of section XX.

Other solid bodies and The principle of sectioning off a ‘free body’ and writing equilibrium
structures statements, which was demonstrated for plane pin-jointed frames, is equally
applicable in other engineering problems. Consider for example the front
end of the fork-lift truck shown in Fig. 1.22(a). The individual elements of
this may be analysed quite easily as illustrated in Fig. 1.22(4) so that each
may be designed to withstand the forces acting on it.

1.9 Forces by the

This method was developed as a convenient way of solving for the forces in
method of tension space frames, and it is of course also applicable to plane frames. It amounts
coefficients to a shorthand notation for writing equilibrium equations of resolved force
components at each joint.
The tension coefficient for a member is defined as the force, F, in the
member divided by its length, 2. Consider the member AB shown in Fig.
1.23; the resolved components of the force /'4g in the co-ordinate directions
x and y will be F4gcosa and Fygsina. Now, cosa and sina can be
expressed as the ratio of the length of the member projected onto the x- and
y-axes divided by the length, 14g, of the member; thus
VAB : JAB
cosa =—— and sina=—
AB LAB
Therefore
XVAB
F4p CoS a = F4p—— = typx ap
Lap
and

Fagsina = F4g > = tapyap

Lap
where f4z 1s the tension coefficient for the member.
Since the majority of configuration diagrams give dimensions which are
related to the co-ordinate directions, it is simpler to express resolved
components of force in terms of the tension coefficient and the projected
length, rather than sines and cosines of angles. All members are assumed to
be in tension initially and hence have a positive coefficient. If in the solution
a tension coefficient turns out to be negative then this shows that the
member is actually in compression. There is no need in this method to work
out reactions at supports in advance as they can simply be included as
unknowns in the equilibrium equations.
Fig. 1.23
 STATICALLY DETERMINATE FORCE SYSTEMS 19

The first step is to choose a set of reference co-ordinate axes and

directions and then to insert all the necessary reactions at the support points
in the co-ordinate directions. The arrowhead direction is quite arbitrary.
Now, commencing at any joint and considering one co-ordinate direction,
write down the equilibrium equation. This will consist of the sum of the
products of the tension coefficient ¢ and the projected length, say x (for that
co-ordinate direction), for each member at that joint plus any reaction or
applied force at that point. It is important to remember that, taking the
origin of the co-ordinate axes referenced at the joint, the projected lengths of
the member must be given a sign appropriate to the positive and negative
directions of the co-ordinate axes. Similarly, reactions and applied loads
must also be given the appropriate sign relative to the axes. In the case of a
space frame the above procedure must be repeated for each of the two other
co-ordinate directions and then in turn for each joint.
Having obtained all the required equations these are solved for all the
unknown tension coefficients and reactions. Each tension coefficient is then
multiplied by the actual length of the member to give the force in the
member. It will perhaps now be evident that, for clarity and ease of
checking, a tabular system of solution is most desirable. The method will
now be illustrated in the following example of a space frame.

Example 1.4
Use the method of tension coefficients to determine the reactions and the forces in
the space frame shown in Fig. 1.24.

Fig. 1.24 20 kN
aay,
D
y

4m Nesx
A E
Ley O
ve u 8m -|
Ay 50 kN
Side view Plan view <

The equilibrium equations for each joint are given in the following table.

Eqn. no. Joint/direction Equilibrium equation

1 A/x A, + 2tap = 0
2 A/y A4tap + Ay = (()
3 A/z An
4 B/x B, — 2tgp + 8tgz = 0
5 B/y B, + 4tpp = 0
6 B/z B,+ 3tgp + 3tpr
=0
7 C/x C, — 2tcp + 8icr = 0
C/y GC; + 4tcp = 0
8
9 C/z Cz = Shon = Ses =W
10 D/x 20 — 2tap + 2ten + 2tcp + 10tpr = 0
11 D/y —4tp4 + 4tpz — 4tnc — 4tprp = 0
12 D/z 3tpc
— 3tpg = 0
13 E/x 8trp 8trc 10tgp = 0
14 E/y Ae 500
15 E/z —3tre
+ 3trc = 0
20 MECHANICS OF ENGINEERING MATERIALS

These equations may then be solved simultaneously to give the following

forces and reactions:

Member Tension Length Force Reaction

coefficient (m) (KN) (kN)

AD +20 4.48 +89.6 A, = —40

BD =16.25 39. —87.5 A, = —30

CD —16.25 5.39 —87.5 et)

BE —7.8 8.55 —66.6 Bers)

Cie —7.8 8.55 —66.6 B, = 65

DE +12.5 10.78 +134.8 Be = IL4

C30

C, = 65

Ca A

Earlier, in the analysis of plane frames, the concept of a matrix approach

was introduced. Problems involving the method of tension coefficients may
also be solved using matrix methods, although, once again, those unfamiliar
with matrices may wish to skip this section. In this case the variables to be
solved include the tension coefficient for each member and the unknown
reactions. For the example above the matrix equation to be solved is

0 On Oo Oe OL Or tap 0
0 Od ~Oot’ LOntOre Oe gE sO. 0 lBp 0
0 On LO Die dee Op Ors Dut Ora Oak O tRE 0
0) Omit O Gale 0 10pm Od tcp 0
0 QO OO Oe OF 8 CE 0
0 OP ROS Or Os Oe Pe One te 0
8 Oo FOO SOO Ores re 0 at. 0
0 OF Or Oe Oe ee re oa ga esI
TS)
ES
SS
SS
SSS = o> oie POr Orr tan On mms. m0 aed VAs 0
0 LOS 30 01 0raeO) 0) Oars Oe ot) By 20
Oi 4 S000.) DLO2 ON re Oat B, 0
SSS.
QS)
|)
SS]
SS 0 O20) 0) FOU esl 0L ko ee C) B, 0
=) 10 0" 02020" 00 UU end Cy 0
0 4°00 (O20 0) 00.080 C, 50
23° NN 0e20) “Us 0D OF 0708 0 C, 0
 STATICALLY DETERMINATE FORCE SYSTEMS 21

This equation has the solution

Lap 20
lgp —16.3
tee —7.81
top —16.3
lop —7.81
Ipe i
Wa
Ay A)
A, $= -80
Ay 0)
B, 30
B, 65
B, 7219
G 30
G 65
Cy —72.2

It should be remembered that this equation must satisfy the requirements

for a statically determine structure, eqn [1.10]. To assemble a matrix of this
size manually is time consuming and error prone, but special purpose
computer programs are available which can do this automatically, given a
specification of the co-ordinates of the joints, the connectivity of the
members, the restraints and the applied loads.

1.10 Bending
of slender
In the force analysis of frameworks the members were only subjected to axial
force, namely tension or compression. The next step is to consider the effect
of transverse loads acting on slender members. The deformation that results
is termed bending and is of course very common in structures and machines
— floor joists, railway axles, aeroplane wings, leaf springs, etc. External
applied loads which cause bending give rise to internal reacting forces and
moments. These have to be determined before it is possible to calculate
stress and deflection.
The transverse externally applied load on a beam or bar can take one of
two forms, concentrated or distributed. The former is illustrated in Fig.
1.25(a) in which the load acts on the surface of the beam along a line
22 MECHANICS OF ENGINEERING MATERIALS

Fig. 1.26

Ry Ro

perpendicular to the longitudinal axis. This, of course, is an idealization, and

in practice a concentrated load will cover a very short length of the beam.
A load which is distributed is shown in Fig. 1.25(4) and occupies a length
of beam surface. The load intensity is always taken as constant across the
beam thickness but may be uniformly or non-uniformly distributed along
part or the whole length of the beam. In practice the particular conditions of
force and displacement at beam supports may vary considerably. Theoretical
solutions of beam problems generally employ two simplified forms of
support. These are termed simply supported and built-in or fixed. The former
is illustrated in Fig. 1.26(a) in which the beam rests on knife-edges or
rollers. When the beam is loaded the support prevents vertical movement
and there is a corresponding reaction, R, but since rotation at the supports is
not prevented there is no restraining moment. Hence the deflection at the
support is zero and the beam is free to take up a slope dictated by the
applied load. The built-in support shown in Fig. 1.26(4) reacts with a
transverse force and a moment. Thus both deflection and slope are fully
restrained. The particular example illustrated of a beam built-in at one end
and free at the other is termed a cantilever.
The number and type of supports also has a further important bearing
on a beam solution by making it either statically determinate or statically
indeterminate (Fig. 1.27). In the former the support reactions can be found
simply from force and moment equilibrium equations. This applies, for
example, to beams on two simple supports or one built-in support and no
other. The two equilibrium equations for /, and M, are insufficient to find
the reactions at the supports of a statically indeterminate beam owing to the -
presence of redundant forces. In this case it is necessary to consider also the

Fig. 127
{|

(a) Statically determinate (b) Statically indeterminate

 STATICALLY DETERMINATE FORCE SYSTEMS 23

deflection of the beam in order to obtain additional equations to solve for the
reactions.

1.11 Force and moment

A slender curved bar is shown in Fig. 1.28(a) subjected to various transverse
equilibrium during bending
loads. In general, to maintain equilibrium a force and a couple will be
required at each end of the bar. The force can be resolved into two
components, one perpendicular and the other parallel to each end cross-
section. The internal forces are obtained by cutting the bar to give two free
bodies and inserting at the cut sections the necessary forces and moments
for equilibrium, as shown in Fig. 1.28(4). The couple, M, is termed a bending
moment, the transverse force Q is called a shear force, and P is a longitudinal
force. The most important stresses and deflections that occur during
bending are due to M, rather than Q or P.

Fig. 1.28 W,

W>
(a) Slender beam subjected to external forces and moments

C M 0 W3 Wa Mz P,

oe P | |
OF Re ; ClaleeO)
(b) Sections of beam showing internal forces and moments

1.12 Sign convention for

It is important that the analysis of mternal forces in bending shall be
bending
consistent within and between various problems. This is best achieved by
adopting a sign convention for loading, bending moment, shear force and
distance/length. There is no standardized convention for bending, although
there tends to be ‘common practice’ in, for example, civil engineering
structural design. There is a fair degree of uniformity amongst textbooks
and the convention used in this text (as in previous editions) follows the
general pattern.
For horizontal beams, positive values are as shown in Fig. 1.29. There
are several points worth noting; for example, loading W and wm are taken
positive downwards since for most horizontal members applied loading is in
that sense. (If loading upmards was taken as positive there would be a
profusion of negative signs.)
For horizontal beams, distance x is a/mays taken positive from left to
right along the beam. Also the choice of y positive downwards is again
largely convenience and practice because most deflections tend to be
downwards.
It is very important to remember that definitions of bending moment and
shear force are illustrated as vector pairs in the positive (or negative) sense.
24 MECHANICS OF ENGINEERING MATERIALS

Fig. 1.29

M Q 9
w
bedded} ‘
ee eae ae a M
Positive loading Positive bending Positive shear Positive directions

For example in the case of shear force, QO, the right-hand vector is taken as
positive because it points in the positive y-direction and acts on a face which
has its normal in the positive x-direction. The corresponding shear force on
the other face is positive because it points in the negative y-direction and
acts on a face whose normal is in the negative x-direction. It is important to
note carefully this sign convention relating the nature of the shear force to
the chosen x- and y-directions because the beam may not always be
horizontal. A similar argument applies to the moment vectors, M.
The most important point to remember is not to change the sign
convention (for some apparent convenience) in the middle of a problem
solution.

1.13 Shear-force and

Both stresses and deflections during bending are directly related to S.F.
bending-moment diagrams
(Shear Force) and B.M. (Bending Moment); it is therefore desirable to
know the distribution along the member and hence where maximum or
minimum values occur. To this end S.F. and B.M. are computed for a
number of cross-sections along the beam and diagrams plotted showing the
distribution and magnitude. A few basic examples now follow and further
illustrations of shear-force and bending-moment distributions appear at the
start of Chapter 6 on bending stresses.

1.14 Cantilever carrying

This example is illustrated in Fig. 1.30. The first step is to choose any
a concentrated load
section AA at a distance x from the left-hand end, cut the beam at this
position and draw the free-body diagram, inserting the required internal

Fig. 130
 STATICALLY DETERMINATE FORCE SYSTEMS 25

forces and moments in the positive sense according to the sign convention.
It is advisable not to balance mentally the free body and put on the forces in
what seems the ‘right’ sense. Note that in this first example, the free-body
diagrams are shown for the cut sections to each side of AA. However, in the
solution only the section to the left of AA is required.

Shear force Vertical equilibrium for the left-hand free body gives

Wy O00
where Q denotes the shearing force at a distance x from the left end; hence

Che Sy
It is fairly obvious that this value of shear force is obtained at whatever
distance the beam is cut between v = 0 and x = L.

Bending moment ‘Taking moments about the mid-point of section AA will

give a moment equilibrium equation for the left-hand free body:
We -+-M=0 or M=—Wx

This is a linear variation from M = 0 at x = 0 to M = —WL at x = L, as

shown in the diagram.
The reactions at the built-in end are, from the above,

Ro=W and My) =—-WL

Figure 1.30 illustrates the S.F. and B.M. diagrams which, in this
example, are in the negative area below each zero base line.
If this problem had been set up so that the beam was supported at the left
and the load applied at the right, then it would have been necessary to begin
by determining Rp and Mo from overall equilibrium of the beam.

1.15 Cantilever carrying

Shear force Considering the cut section AA, the total downward load is wx
a uniformly distributed and vertical equilibrium of the free body in Fig. 1.31 is obtained as
load
pie Oia Oa acoso)! =e

Fig. 131
26 MECHANICS OF ENGINEERING MATERIALS

The shear-force diagram is therefore a linear variation from 0 to —wL as

shown in the figure.

Bending moment The moment equilibrium equation is obtained by taking

moments at AA; thus

wx"
pe-+M=0 of M=—— [1.12]
Z 2
which gives a parabolic shape of B.M. diagram varying from M = 0 at x = 0
OM = =—pl /Zat x= L.
The support reactions are therefore

wl
Ro =wLl and My =— D

1.16 Simply supported

In this case, Fig. 1.32, the reaction at the left-hand end has to be determined
beam carrying a uniformly before the values of S.F. can be expressed. From vertical equilibrium and
distributed load symmetry of the whole beam,
wL
Ry =e ar igs

Shear force For the equilibrium of the free-body section,

L
+0 -="+nr=0

12,
Q = -wr +>

This equation shows that the S.F. is zero at mid-span and equal to the
reactions, wL/2, at x = 0 and L.

Fig. 1.32
 STATICALLY DETERMINATE FORCE SYSTEMS 27

Bending moment Moment equilibrium gives

> eee

wLx wx?
= aN
5 5 [1.13]
etl}

This equation represents a moment distribution which is parabolic, being

zero at each support and having a maximum value of wL?/8 at x = L/2.

Example 1.5
Sketch the S.F. and B.M. diagrams for the simply supported beam shown in
Fig. 1.33.

—R,— Rg + 5000+ 10000 =0 (force equilibrium)

—5R,+ (5000 x 4) — 5000 x 2x 1=0 (moment equilibrium)

R4=2000N Rg =13000N

Fig. 133 | 5000 N/m

_ - : = |

NN
im. 4m PS 22 ii
Ra Re
(a)
+10 000 N

2000 Nm

(c) —10 000 Nm

Shear force
Cran1s +O = 2000==0) Q =+2000N

l<x<5: +40 — 2000+

5000 = 0 Q = —3000N

5<x<7: +0 — 2000+ 5000 — 13000 + 5000(x—5)=0 Q@ = —5000x + 35000N

28 MECHANICS OF ENGINEERING MATERIALS

Bending moment

0O<x<1: —2000r+M=0 M = 2000x

l<x<5: —2000x + 5000(r — 1) +M=0 M = 5000 — 3000x

S<x<7: —2000x + 5000(« — 1) — 13000(x — 5)

+5000(r — 5) - Me M = —2500x? + 35 000x — 122

The S.F. and B.M. diagrams are shown in Fig. 1.33(4) and (c).

1.17 Point of
In the above example it will be seen that the B.M. changes sign through a
contraflexure
zero value and this is termed a point of contraflexure. Its position is
determined by putting the bending-moment expression equal to zero. In
real situations these points are of importance to the designer because, for
example, supports placed at such points will not be subjected to bending
moments.

1.18 Torsion of members

Torsion is the engineering word used to describe the process of twisting a
member about its longitudinal axis as illustrated in Fig. 1.34. The angle of
rotation of one section relative to another at a distance / along the member is
termed the angle of twist 0, measured in radians. The twist per unit length is
thus @//. The forces F required to cause this twist are shown at section B.
The product of the forces times the distance to the axis is 2(F'- 4/2) and this
results in a moment about the axis called a torque T. For equilibrium of the
member the applied torque at each end must be equal in magnitude and
opposite in sense acting about the longitudinal axis.
The situation shown in Fig. 1.34 is described as pure torsion, i.e. the
member is only subjected to torque and no other resultant lateral or
longitudinal forces exist. However, in practice applied forces which set up
torque frequently also induce bending and possibly end load on the
component.
 STATICALLY DETERMINATE FORCE SYSTEMS 29

1.19 Members subjected

This situation was referred to at the end of the previous section and will now
to axial force, bending be considered in more detail.
moment and torque Let us take a practical example with which many people have been
personally involved, that of changing a wheel on a car. The wheel nuts have
to be removed with a wheel brace or wrench. There are, of course, various
designs to achieve the same objective which is to be able to apply enough
torque to the nuts to loosen them and at a later stage to retighten them. A
typical wheel brace is shown in Fig. 1.35 in which both hands apply force,
one in the direction necessary to loosen the nut and the other as a reacting
support. The torque set up on the nut is simply the force F times the
perpendicular distance between the plane in which the force is acting and
the plane containing the nut. In addition, the brace is acting as a beam with a
kink in it, i.e. simply supported at each end with a ‘concentrated’ load at the
centre. In fact each section of the brace will have to transmit some
combination of bending and torsion. In this particular case once the brace
has been pushed on to the nut there will not thereafter be any axial force
along the axis of the brace. To calculate the bending moment, shearing force
and torque in each section of the brace, free-body diagrams must be drawn.
If these are done correctly it is then a simple matter to determine the
equilibrium conditions for each section. The following example will
illustrate the method for slightly more complex internal reactions produced
by the application of a single force.

Fig. 135

Example 1.6
A length of pipe ABCD is connected rigidly to a pump unit at A and is bent through
two right angles at B and C as shown in Fig. 1.36. A force F has to be applied at D in
order to connect the pipe to the next unit which is slightly out of alignment.
Determine the system of forces, moments and torques which are set up in the pipe
in relation to the co-ordinate axes given. AB is in the x-direction, BC in the
z-direction and CD in the y-direction and the force F lies in the xz-plane.

The first step is to resolve the force into the x- and z-directions, giving in
each case F cos 45° or F'/,/2. To avoid confusion we shall temporarily call
these P and R respectively. The next step is to draw the series of free-body
diagrams as shown in Fig. 1.37, which should be self-explanatory,
commencing with CD and successively following on with the segments
CB and BA.
30 MECHANICS OF ENGINEERING MATERIALS

Fig. 1.36

(c)

It is seen that in each free body the transfer of the forces P and R to the
next end of the segment requires a moment and/or torque to maintain
equilibrium about the respective axes. This process is an essential first step
towards the analysis of bending and torsional stresses and deflections which
will be derived in Chapters 5, 6 and 7.

1.20 Torque diagram

In the analysis of torsion problems there is an analogous diagram to the
shear-force and bending-moment diagrams for bending and this is a torque
diagram. An example is shown in Fig. 1.38, which illustrates the variation in
torque which may occur along a shaft depending on the various inputs and
outputs of power transmission.
The values of 74g, Tc, etc., in the diagram may be determined from
the free-body diagrams shown in Fig. 1.39.
A full analysis of torsion in shafts is given in Chapter 5.
 STATICALLY DETERMINATE FORCE SYSTEMS 31

Fig. 1.38

|
|
(Ce | Torque diagram
|
|

1.21 The principle of

Consider the beam problem in Fig. 1.40(a). Although this would not be
superposition
difficult to analyse for B.M. in its present form, imagine it split up into the
two separate cases as in Fig. 1.40() for which the B.M. diagrams for the two
parts are as shown. Now, bending moment is always a first-order function of
applied load, i.e. there are never terms in W? or w’ etc.; therefore at any
section along the beam the sum of the bending moments due to the loads
acting separately would be exactly the same as the bending moment due to
the combined load. Similarly the reactions at the supports due to the
separate loads may be summed to give the reactions caused by the combined
loading. The reader may also like to verify that the S.F. diagram could be
derived in exactly the same manner.
The solution of the present example is obtained by summing the
ordinates of bending moment at every section along the beam. This may also
32 MECHANICS OF ENGINEERING MATERIALS

(c)

be done graphically by placing the two B.M. diagrams together as shown in

Fig. 1.40(c) and the resultant is obtained as the shaded area.
This most important technique is known as the principle of superposition.
It will be demonstrated at regular stages throughout subsequent chapters
when analysing force, stress, strain and displacement systems. It is valid
when the quantity to be determined is a linear function of the applied load.
In effect, the quantity to be determined may be found by analysing each load
or group of loads acting separately and then the results may be
superimposed to obtain the total value due to all the loads acting
simultaneously.

1.22 Summary
This chapter has served several purposes, the primary one being to
emphasize the importance and give demonstrations of the development of
equilibrium situations in solid bodies and structures. To this end it is also vital
to be able to draw representative free-body diagrams with all the required
force components. An appreciation of the meaning of statical determinacy
and how to recognize such situations is also most important. The key
equations for equilibrium are

DF, =UF, =F, =0

OM, =UM,
= UM, =0
Engineering components are subjected largely to tension, compression,
bending and torsion, either separately or in a variety of combinations. It is
essential, therefore, to understand how these various modes of deformation
arise and the associated equilibrium requirements for internal and applied
forces.
 STATICALLY DETERMINATE FORCE SYSTEMS 33

The principle of superposition has been shown to be a most valuable means

of solution, and this will be illustrated further in succeeding chapters.
Finally, the use of spreadsheets has been introduced as a means of
obtaining versatile, general solutions to engineering problems. This
approach will be developed further throughout the book. For problems
with a large number of variables, the convenience of matrix solutions has
also been illustrated.

Problems
1.1 A force F has x and y components F, and Fy respectively. The
distances from a point A to any point on the line of action of F and r,
and r, in the x and y directions, Fig. 1.41. The moment about the z
axis of F is therefore
Vi el ye Pyke

Develop a spreadsheet where, given the v and y components of

force and distance to a point on the line of action on the force, the
resulting moment M, is calculated.
A x Using the formulae developed, find the componnts of the resultant
Fig. 1.41
force and moment of the system of forces in Fig. 1.42. Find the
magnitudes of the resultant force and moment.

Fig. 1.42 20

10
Li? Fekete “80
(10,20)
eo—> 30 F
(-40,10)
a n 1 n 4 1 a J
A 20

ii e— > 100
(30,-10)
60. L

40
<— e (-30,-30)
L

1.2. Modify the spreadsheet in Problem 1.1 so that the magnitude of the
force and the angle it makes with the x axis are entered and the x and y
components are automatically calculated.
1.3 Using the formulae of Problem 1.2, find the resultant force and
moment of the 15kN force in Fig. 1.43. Find the moment due to a
unit force acting at F3. From this find the magnitude of the force
needed at F3 in order to balance the moment due to the 15 kN force.

Fig. 1.43
34 MECHANICS OF ENGINEERING MATERIALS

1.4 A force F has x, y and z components F’,, /, and F’, respectively. The
distances from a point A to any point on the line of action of F are r,,
r, and r, in the x, y and z directions, Fig. 1.44. The moments about
the x, y and z axes are therefore
My = ty, —t2.Fy
My = t2-E —tely
M =f, ly — tks

Develop a spreadsheet where, given the x, y and z components of

force and distance to a point on the line of action on the force, the
resulting moments M,, M, and M, are calculated. Use the
spreadsheet to caculate the moments at point A in Fig. 1.36, when
F = 10EN,@ =1 m,.2 =0.6m and ¢ = 0:5
Note: the sign of the displacements to go from A to a point on the line
of action of F is important.
the moments given in Fig. 1.37 are the reactions to force F and
are therefore opposite in sign to the moments caused by F.
15 Use the 3D moment spreadsheet from the previous question to find
the moment of the applied forces at A and C about point B in Fig.
1.45. Sum the forces and moments to verify that the resultant force
and moment of the forces at A, B and C is zero.

Fig. 1.45

Dimensions in mm.

A maintenance cradle in a factory is supported by the pulley system in

Fig. 1.46. A fitter weighing 82 kg can be raised (a) by pulling on the
rope R himself or (b) by someone else pulling on the rope. Calculate
the pull needed on the rope in each case. Friction and the weight of
the cradle and pulleys may be ignored.
Figure 1.47 shows a tyical design for a vehicle weighbridge. Each of
the levers A, B and C have a:d in the ratio of 1:10. If a balance load of

balance
load

Fig. 1.47
 STATICALLY DETERMINATE FORCE SYSTEMS 35

12N is required on the lever A then calculate the weight of the

vehicle.
1.8 A certain design of brake drum is as shown in Fig. 1.48. If the drum
rotates anticlockwise and generates a torque of 210 Nm, calculate the
force needed in the cylinder C in order to stop the drum. The
coefficient of friction between the drum and showWis 0.35.

Fig. 148

Cylinder
Cc

1.9 A wooden beam (SG = 0.6) is 15cm by 15cm by 4m long and is

hinged at A, as shown in Fig. 1.49. At what angle @ will the beam float
in the water?

Fig. 1.49
nee
t
im
Y
Water

1.10 A 4m wide gate for controlling water levels is shown in Fig. 1.50. For
the levels shown calculate the magnitude and direction of the force in
the hydraulic cylinder A. The density of the water is 1000kg/m?.

Fig. 150

1.11 The door AB retains water in a channel as shown in Fig. 1.51.

Determine the magnitude and direction of the force on the ground at
B due to the pressure of the water. The weight of the door may be
ignored.
36 MECHANICS OF ENGINEERING MATERIALS

Fig. 1.51

IE A special type of bicycle calliper brake is shown in Fig. 1.52. If it is

pin-jointed at A, B, C and D and the cyclist applies a force of 300 N at
A when the bicycle is stationary on a | in 7 slope, calculate whether or
not the bicycle would move down the slope. The combined weight of
the bicycle and cyclist is 102 kg and the coefficient of friction between
the wheel and the brake block is 0.7. The outside diameter of the tyre
is 690 mm and the diameter of the line of action of the brake block is
640 mm.

Fig. 1.52

Bicycle
tyre

Brake block Wheel rim

18) A cylinder weighing 800N is supported between two rigid rods as

shown in Fig. 1.53. Determine the tension in the cable AB and the
reactions at D and E.
1.14 A motor cycle is standing on horizontal ground with its front wheel in
contact with the kerb as shown in Fig. 1.54. When the rider slowly
engages gear the motor cycle starts to rise over the kerb when the rear
wheel torque reaches 0.3 kNm. For the instant when the front wheel
just starts to leave the road calculate (a) the value of the coefficient of
friction between the rear type and the road to avoid wheel slip, and ()
the magnitude and direction of the force on the front wheel. The
wheel diameters are both 600mm and the combined weght of the
 STATICALLY DETERMINATE FORCE SYSTEMS 37

Fig. 153
cable

Fig. 1.54

rider and the motor cycle is 285kg acting through a point midway
between the wheels.
Ieee At the moment of start-up, a power hacksaw is in the configuration
shown in Fig. 1.55. If the torque input at the pulley drive is 150 Nm
calculate the magnitude and direction of the force on the workpiece.
The weight of the blade saddle is 45.9kg and the weight of the
support arm may be ignored. Friction between the blade saddle and
the suppert arm may also be ignored.

Fig. 155 400 pe 100 al

|
Blade saddle | a Nm

200 mm
38 MECHANICS OF ENGINEERING MATERIALS

1.16 When the hydraulically-operated digger shown in Fig. 1.56 1s

removing earth the force on the bucket is horizontal wth a magnitude
of 15kN. For the arm positions indicated, calculate the force in each
of the three rams A, B and C. The weight of the arms of the digger
should be ignored.

Fig. 156

3:5

1.17 Ina fork-life truck of the type shown in Fig. 1.57 the load of 3 KN is
raised by means of the rotating screw AC. If the weight of the front
fork is 102 kg (acting through its centroid G) and the weights of the
arms AB and CD are 51 kg each, calculate the force in the screw AC
and the reactions at A for the configuration shown.

Fig. L57
Dimensions in metres

1.18 The frame work shown in Fig. 1.58 is used to support a steel car body
weighing 200 kg. When the car body is suspended in (a) air and (d)
totally immersed in a plating bath containing a liquid of density
1000kg/m?, calculate the support reactions and the forces in
members AB, BF and FE. Density of steel = 7800 kg/m’.
1.19 A dockyard crane may be considered to be a plane pin-jointed frame
as shown in Fig. 1.59. Determine the support reactions and the forces
in the members of the framework by resolution at joints.
 STATICALLY DETERMINATE FORCE SYSTEMS 39

Fig. 1.58

Fig. 1.59

1.20 A plane pin-jointed frame is shown in Fig. 1.60. Determine the force
in member DE when the frame is subjected to the forces shown.

Fig. 1.60

2m pee 2m 2m 2m 2m

Re
20 kN 8 KN 10 kN

all A crane with the dimensions shown in Fig. 1.61(@) supports a weight
of 1000kg. Use the method of sections to calculate the forces in
members HG and HJ. If an alternative design is used as in (0) in
which the cable is attached to the frame at F, calculate the force which
this would cause in HJ.
122 Figure 1.62 shows a simple pin-jointed truss which must support a
load P. Other known variables are the dimensions ) and h, the
allowable stress in the bars oo and the density of the bar material p.
Program cells in a spreadsheet to calculate the following quantities,
given numerical values for the variables above.
(a) Determine the forces in members AB and BC.
(b) If the members can support an allowable stress (force/unit area)
of a9, determine the minimum cross-sectional area of each bar.
(c) Calculate the total weight of the truss, if both bars have a density
p.
40 MECHANICS OF ENGINEERING MATERIALS

Fig. 1.61
D eS

Dimensions in metres

Fig, 162 (d) Use the spreadsheet to evaluate numerical answers for parts (a)—
(c), given b= 1m,h= /2m, P=1kN, 09 = 100 MN/m? and
p = 7800kg/m °.
(ec) Evaluate the total weight of the structure for values of h in the
range ().5—3 m to determine the minimum weight configuration.
1:23 Repeat problem 1.22 but use the sare cross section for both members.
In other words find the bigger cress sectional area and use that for
both members in the subsequent calculation of weights. Hint, there is
a spreadsheet function @ MAX which finds the largest value in a
block of cells.
124 Calculate the forces in each of the members of the pin-jointed space
frame shown in Fig. 1.63.

20 KN
Fig. 1.63

4 kN 4m Use the method of tension coefficients to calculate the forces in the

pin-jointed space frame shown in plan view in Fig. 1.64. The pinned
E p 2un |
supports A, B and C are at the same level, and DE is horizontal and at
a height of 4m above the supports.
4m Using the method of tension coefficients, determine the forces in each
of the members of the framework illustrated in Fig. 1.65.
A semi-circular beam is subjected to a force of 2kN as shown in Fig.
1.66. Draw diagrams to show the variation of axial force, shear force
and bending moment in the beam and state the maximum values of
each.
 STATICALLY DETERMINATE FORCE SYSTEMS 41
LS

Fig. 1.65

0A = 0B =0.2m
O0G=2m
GE=GD=0.1m
0c =0.2m
GF=0.1m

5 kN
5 kN
10 kN

Fig. 1.66

2 kN 2 kN

1.28 Sketch the shear-force and bending-moment diagrams, inserting the

principal numerical values for the beams illustrated in Fig. 1.67. Also
establish any positions of contraflexure.

Fig. 167 ioe 3 kN | a lhe

AL 8 Cw pale te AB uA —E YF

1.29 A 10m x 6m advertisement board is supported by three posts as in

Fig. 1.68. Construct the shear-force and bending-moment diagrams
for the vertical posts if the wind loading on the board is 981 N/m’.
1.30 The bracket ABC is freely pivoted on the vertical rod shown in Fig.
1.69. Determine the forces, bending moments and torque transmitted
on all parts of the bracket when a 1 kN vertical force is applied at C.
Also calculate the reactions on the wall at A and RB.
42 MECHANICS OF ENGINEERING MATERIALS

Fig. 1.68

Fig. 1.69

Dimensions in mm

1.31 A power transmission system is illustrated in Fig. 1.70. The belt

drives at A, C and D require 15kW, 25kW and 10kW of power
respectively at a shaft speed of 500 rev/min. Draw a diagram to show
the torque in each section of the drive shaft.
Output A Output C Output D
Fig. 1.70 = 15 kW = 25 kW = 10 kw
2m 2m
>|

H! Dia= 35 mm

Diameter = 20 mm

1.32 Use the principle of superposition to determine the B.M. diagram for
the beam loaded as shown in Fig. 1.71.
3 KN 4 kN
Fig. 171
ee | | 10 aa

~ 2m} >
 CHAPTER
Statically Determinate Stress
Systems

The effects of external applied forces can now be measured in terms of the internal
reacting forces in a solid body or the members of a framework, as described in the
previous chapter. However, at that stage no mention was made of the cross-sectional
size and shape of the members. This aspect had no effect on the forces in the
members, but conversely one should be able to describe quantitatively the way in
which two members of different cross-sectional size would react to a particular value
of force. This is done through the concept of stress and this chapter shows how
stresses can be determined for simple engineering situations.

2.1 Stress: normal, shear

Consider the member shown in Fig. 2.1(a) subjected to an external force, F,
and hydrostatic
represented by the arrow at each end parallel to the longitudinal axis. The
arrow simply represents a force resu/tant on the end faces and obviously the
force is not actually applied solely along the line of the arrow. Similarly the
internal force reaction does not act along just a single line but is transmitted
throughout the bulk of material from grain to grain. If part of the member is
cut off to give a free body as in Fig. 2.1(4) then equilibrium will be
maintained by appropriate components of internal reacting force such as AF
acting on elements of area AA.
For equilibrium © AF = F and © AA = A the total cross-sectional area.
The internal force per unit area is called the stress, a, and is given by

oO _AF
~ AA
In the limit,

— im (OF\ 9"
Ceice Ne eet
So,

dF =oadA

PF
| di = | o dA
0 area

If it is assumed that the stress is constant over the area then

i a | dA =oA
area

so that stress o = F'/A.

Note that this equation for direct or normal stress will only apply if the
force acts through the centroid of the cross-section. If the applied force, F,
44 MECHANICS OF ENGINEERING MATERIALS

Fig. 2.1

is offset then bending stresses will be set up along with the normal stress and
the stress will no longer be constant across the material. This type of
situation is considered in Chapter 6.

Normal stress
In the simple case in Fig. 2.1 the average direct stress is F/A and will be
denoted by the symbol o as in Fig. 2.1(c). Normal or direct stress acts
perpendicular to a plane and when acting outwards from the plane is termed
tensile stress and given a positive sign. Stress acting towards a plane is termed
compressive stress and is negative in sign. In order to denote the direction of a
stress with respect to co-ordinate axes, a suffix notation is used, so that o,,
Oy, O; represent the components of normal stress in the x-, y- and z-
directions as shown in Fig. 2.2(a).

Fig. 2.2

(a) V4 (b)

Shear stress
When a man who is running wishes to slow down he applies his ‘brakes’.
This is achieved through mounting pressure between the soles of his shoes
and the ground and thus increased frictional force parallel to the ground.
This concept of a force applied tangential or parallel to a surface is termed a
shear force. If internal reacting shear force is expressed as a force per unit
area then it is termed a shear stress. It also acts parallel to any associated plane
within the material and is denoted by the symbol 7. A double suffix notation
is required to define shear stresses with respect to co-ordinate axes. The first
suffix gives the direction of the normal to the plane on which the stress is
acting, and the second suffix indicates the direction of the shear stress
component. Thus 7,,, is a shear stress acting on the yz-plane (the normal in
the x-direction) and pointing in the y-direction. The sign convention
associated with shear stress is defined as positive when the direction of the
stress vector and the direction of the normal to the plane are both in the
positive sense or both in the negative sense in relation to the directions of
the co-ordinate axes. If the directions of shear stress and the normal to the
 STATICALLY DETERMINATE STRESS SYSTEMS 45

Fig. 2.3

plane are opposite in sign then the shear stress is negative. There are twelve
possible shear stress components in a three-dimensional stress system as
indicated in Fig. 2.2(b) (those on the obscured faces have been omitted for
clarity).
There is a further condition of shear stresses which always exists and is
explained as follows. Consider the element of unit thickness in Fig. 2.3
subjected to shearing stresses along its edges 7,, and 7,,. Vertical and
horizontal force equilibrium shows that 7,, on face AB is equal to 7, on face
CD. Also, 7, on face AD is equal to 7,, on face BC. The shear forces along
the sides AB and CD are (7, x 1x dy) and along AD and BC are (7), x 1x
dx). Taking moments about a z-axis through the centre of the element, for
equilibrium

(tay X 1 x dy) x d — (tye X 1 x de) x S =

or

Tey = Tyx

These are termed complementary shear stresses. Thus a shear stress on one
plane is always accompanied by a complementary shear stress of the same
sign and magnitude on a perpendicular plane.

Hydrostatic stress
This is a special state of direct stress which should be mentioned now
although its importance will become more evident at later stages.
Hydrostatic stress may be represented by the stress set up in a body
immersed at a great depth in a fluid. The external applied pressure p being
equal at all points round the body gives rise to internal reacting compressive
force and hence compressive stress which is equal in all directions, i.e.
0, = 0, =0,=—>.

2.2. A statically
If the stresses within a body can be calculated purely from the conditions of
determinate stress system
equilibrium of the applied loading and the internal forces then the problem
is said to be statically determinate. There are very few examples of this
nature; however, they do give further illustration of the application of
equilibrium, and the remainder of the chapter will be devoted to solutions in
this category.
46 MECHANICS OF ENGINEERING MATERIALS

2.3 Assumptions and

Exact solutions for stress, displacement, etc., in real engineering problems
approximations
are not always mathematically possible and even those that are possible can
involve lengthy computation and advanced mathematical techniques which
are not necessarily justifiable. This is because we seldom know the exact
conditions of applied loading on a component or structure for its expected
working life, and the materials used are not wholly predictable in behaviour.
It therefore becomes necessary and desirable in most engineering problems
to make some simplifying approximations and assumptions which, while not
changing the basic nature of the problem, will allow a simpler solution and
an answer which is not too far from the truth. It is important, however, that
any assumptions or approximations are clearly stated at the start so that the
reader may assess the validity of the answer in respect of what might be the
exact solution.
Some of the problems to follow are in general not statically determinate
but, with some realistic geometrical limitations, they can be solved purely
from equilibrium conditions to give answers which, although not exact, are
reasonably accurate for the purposes of engineering design.

2.4 Tie bar and strut (or

These are the simplest examples of statically determinate stress situations,
column)
since the equilibrium condition is simply that the external force at the ends
of the member must be balanced by the internal force, which is the average
stress multiplied by the cross-sectional area.
The case of the tie bar is illustrated in Fig. 2.4. The bar at (a), subjected
to tension, has been cut perpendicular to the axis into two free bodies to
show the normal stress o,. This, multiplied by the area on which it acts,
must be in equilibrium with the applied force, F; hence
He
CAs Or V6 , (tensile stress)

Fig. 2.4

(b)

In Fig. 2.4(d) it is seen that if the bar is cut into two free bodies at an angle to
the axis then there will be two components of stress; one is normal to the
plane o,,, and the other is parallel to the plane 7,. The significance of these
stress components is dealt with in detail in Chapter 12.
A similar reasoning to that above applies to the compression situation in
the strut (column).
 STATICALLY DETERMINATE STRESS SYSTEMS 47

EXSnni@er, one eer al ee e e o a Pe

Figure 2.5(a) shows the general arrangement of an engine piston and connecting
rod. The top surface of the piston is loaded by gas pressure. The resulting force is
transmitted through the pin-jointed connecting rod to a crankshaft creating power.
A heavy flywheel on the crankshaft smooths the variations in torque due to the
rapidly changing gas pressure so that the angular velocity w of the crank is
effectively constant. The aim is to determine the minimum cross-sectional area of
the connecting rod so that the maximum stress does not exceed a specified value.

Fig. 2.5 Connecting rod

Crankshaft

Angular
velocity @

(a)

(b)

|
The pressure p on the surface of a piston of bore b creates a resultant force
ion
g
prb’
EF, = pA = eres

The gas force is resisted by the force in the connecting rod F, and the side
reaction R from the cylinder wall, Fig. 2.5(6). Horizontal force equilibrium
requires that
prb?
Secs Corsa 0

so the conrod force is

io pr?
‘~~ 4c08 ¢
If a stress of magnitude oo is not to be exceeded, the cross-sectional area
required is
pe magnitude(F’,)
00
 MECHANICS OF ENGINEERING MATERIALS

Fig. 2.6

£
ye (m) (m) (rpm) (rad/s) | (m)
3 0.015 | 0.055 | 15000 +C3*2*@P/60 | 0.04
4 |
ae |

as: | | Gas pressure Gas force Conrod force

7 theta theta | phi p Fg Fe
8 (degrees) _ (radians) | (radians) (Pa) (N) (N)
9
eeu! 0 @RADIANS(A1O) @ASIN($A$3*@SIN(B10/$B$3) _ 1145900 _+D10*@PI*$E$3%2/4 |-E10/@COS(C10)
eras 30°) @RADIANS(A11) @ASIN($A$3*@SIN(B11/$B$3) 1717800 | +D11*@PI*$E$342/4 |-E11/@COS(C11)
12 60 | @RADIANS(A12) @ASIN($A$3*@SIN(B12/$B$3) 660100 | +D12*@PI*$E$3°2/4 -£12/@COS(C12)
13 90 @RADIANS(A13) |@ASIN($A$3*@SIN(B13/$B$3) | 317800 |+D13*@PI*$E$3°2/4 —-E13/@COS(C13)
14 120 @RADIANS(A14) |@ASIN($A$3*@SIN(B14/$B$3) _ 130900 _+D14*@PI*$E$3%2/4 -E14/@COS(C14) _ :
ait 2 450 @RADIANS(A15) |@ASIN($A$3*@SIN(B15/$B$3) } 25900 |+D15*@PI*$E$342/4 -E15/@COS(C15) :
16 —.180)} @RADIANS(A16) |@ASIN($A$3*@SIN(B16/$B$3) | : 0 +D16*@PI*$E$342/4 -E16/@COS(C16) .
be 240 @RADIANS(A17) @ASIN($A$3*@SIN(B17/$B$3) _ 700 | +D17*@PI*$E$342/4 — -E17/@COS(C17) : a
418 _ 240 _ @RADIANS(A18) |@ASIN($A$3*@SIN(B18/$B$3) _ 14000 , +D18*@PI*$E$3%2/4 -F18/@COS(C18) :
WaT: 270) @RADIANS(A19)_ @ASIN($A$3*@SIN(B19/$B$3) 71400 | +D19*@PI*$E$3%2/4 -E19/@COS(C19) _
20 _—- 300 ~—@RADIANS(A20) | @ASIN($A$3*@SIN(B20/$B$3) _ 188900 +D20*@PI*$E$342/4 | -E20/@COS(C20) me
21 _—_—_—-330_~— @RADIANS(A21) _@ASIN($A$3*@SIN(B21/$B$3) 553700 | +D21*@PI*$E$342/4 | -E21/@COS(C21) >
wees 360 _@RADIANS(A22) | @ASIN($A$3*@SIN(B22/$B$3) _ 1145900 +D22*@PI*$E$342/4 -E22/@COS(C22)

_@MAX(F(10 : F22) i ~Maximum Conrod force

_ @MIN(F(40.. F22) __Minimum Conrod force

f~ is . ; is 7 | : _§ 200000000 i = “Max stress

= [ @ABS (F24/F27) “Area required for maxforce

@ABS (F25/F27) Area required for min force

| (rad/s) |
15000 1571 |

| : | Gas pressure , a Gas force. Conrod force

_theta _ theta phi. p a | Fe
(degrees) | (radians) ; (radians) | (Pa) - (N) | ___ =f)

“0 | 0.000 0.000 | 1145900 1440 | 1440

2230) 0.524. 0.137 | 1717800 5 2159 i ae
60. 1.047, 0.238, «660100 ; 830 854
90. 1.574 0.276 | 317800 399 415 |
120 2,094, 0.238. 130900 164 169
150 2.618 0.137 | 25900 33 | ES
180 3.142) 0,000 | WON ; ) 0}
210 | 3.665 ;0.137 700 Ea 1 |
240 4.189) = 0.238 PACOON 18 | te
_ 270 oe 0.276 | 71400 90 | a ie i
= 300: | 5.236 | 0.238, 188900 235 696 1440 _| _-242
330 5.760 0.1371 553700 | 702 |
360 6.283 ___0.000 | 1145900 __ ib ‘ 1440

= is _— ii J | o =a. ~ al | "Maximum Conrod force|

| = i aes >.) | > i ____ 2479 | _ Minimum Conrod force

| 7 i} seiieeserits my a Max stress

pee ee ele 4.43970163E-09° ‘Arearequired ffor. max force
| 1.08950278E-05 Area required for mi x force

(b) Spreadsheet display

 STATICALLY DETERMINATE STRESS SYSTEMS 49

The angle ¢ can be determined as a function of crank angle @ from

r sin @=/ sin @

or

@=sin! (;sin 0)

In a small two-stroke chain-saw engine, the variation of gas pressure with

crank angle is shown in the fourth and first columns respectively of a
spreadsheet, Fig. 2.6(a). The appropriate formulae for the calculation of the
angle @, gas pressure force and conrod force have been programmed into the
first row underneath the appropriate titles. ‘These formulae have then been
copied to the rows beneath to calculate the quantities of interest for other
angles 0.
The only special point to note is that the location of the cell containing
the cylinder bore, cell E3, should be the same in all rows when calculating
the gas force. Normally when a cell is copied the cell addresses are
automatically adjusted, so that ifaformula +D10*@PI*E3%2/4 in cell E10
is copied to cell E11 the result is +D11*@PI*E42/4. In this case the
reference to cell E3 should remain the same in all rows. This is indicated by
marking it as an absolute address by pressing an appropriate function key
when entering the cell, or by typing the address as $E$3.
The resulting spreadsheet display is shown in Fig. 2.6(4). From this data,
it is very straightforward to create an XY graph of the variation of conrod
force with angle 0 as shown in Fig. 2.7. It can be seen that for the stroke r
and control length / chosen here, the angle ¢ remains quite small and the gas
pressure force and the conrod force are of similar magnitude. The maximum
and minimum values of conrod force can be extracted with the appropriate
functions @MIN and @MAX and the required cross-sectional areas
calculated. Using the spreadsheet, the effect of varying any of the
parameters such as cylinder bore, cylinder stroke or conrod length can
immediately be evaluated.
This analysis ignores the inertia forces required to accelerate and
decelerate the piston as it moves up and down the bore. For a chain-saw

Fig. 2.7
Conrod Force F,

+ hecererctreeh + +
O 601120 7.180 8) 240) 300) 360

8 (deg)
50 MECHANICS OF ENGINEERING MATERIALS

engine under no load the crank speed can approach 14000 rev/min and
inertial effects are very important.

2.5 Suspension-bridge
A common form of loading on a cable, for example in suspension bridges, is
cables
shown in Fig. 2.8(a). The loading, m per unit length, is distributed
uniformly on a horizontal base, the weight of the cable being neglected. In
this particular example, the ends A and B are set at different heights above

Fig. 2.8

the lowest point. It is useful for the analysis to cut the cable at O, to insert a
reaction at that point, and to consider the equilibrium of the right-hand part
of the cable. The free-body diagram in Fig. 2.8() shows that equilibrium is
satisfied by the triangle of forces Tg, 7 and wx. The position of the lowest
point O and hence the distance x; is not known. The distance x; can be
determined by equilibrium of moments of the forces for either part of the
cable. Thus, taking moments about B, and noting that the force mx, can be
taken as acting through the mid-point of xv;, then

Toy — wx) 2 = (0) [2.1]

You should now draw the free-body diagram for the left-hand part. Taking
moments about A gives

(/ a x1)
Toy2 — w(l — x1) = [2.2]
2
Inspection of eqn. [2.1] shows that, for any point on the cable having co-
ordinates (x, y) relative to O,

DX z
i OT [2.3]

which is an equation for a parabola.

Returning to eqns. [2.1] and [2.2] and eliminating 7o gives

one
2 (=e
 STATICALLY DETERMINATE STRESS SYSTEMS 51

from which

eae Ion/y2)'” [2.4]

1+ (1/92)?
If the ends A and B are at the same level then y2 = y; and x, = //2.
The maximum tension in the cable will naturally occur at end B since
this side of the cable supports the greater proportion of the load. The force
Tp can be determined from equilibrium of the triangle of forces To, wx; and
Tz. Vertical equilibrium:

Tz sin @— wx; = 0

But, from the Seometry of the triangle,

sin iv neu (xJI [27

Therefore

eee Wins) na 25
JA
Similarly,

je Bespeleaaelllls ele 26
J2

where x; is given by eqn. [2.4].

The minimum tension in the cable is Tg, which is obtained from eqn.
[2.1], by substituting the values of x; and y;. Any required stress values are
determined by dividing the appropriate tension by the cross-sectional area,
a, of the cable, since it is assumed that the stress is uniformly distributed
across the section and thus o = 7T/a.

2.6 Thin ring or cylinder

If a cylinder or ring is rotating at constant velocity, Fig. 2.9(a), then an
rotating
inward radial component of force is required to provide the centripetal
acceleration on each element of material. This :nmard force may be resolved
into the tangential direction at each end of a typical element as shown in Fig.
2.9(b), which is thus subjected to circumferential tensile stress. This may be
determined from a consideration of dynamical equilibrium. Alternatively we
can reduce this dynamical situation to a static one by applying D’Alembert’s
principle, in which forces to accelerate masses are replaced by equal and
opposite static forces. In this case the centripetal force is replaced by what is
often termed a centrifugal force acting radially outwards, and we can then
apply statical equilibrium.
Consider the rim of a wheel (no spokes) or a slice through a thin-walled
cylinder each rotating about a central axis at velocity w. If this problem is to
be statically determinate then the diameter of the ring or cylinder must be
large, say greater than ten times the cross-sectional dimensions of the rim. It
is then possible to assume a uniform distribution of stress over the cross-
section in the circumferential direction, and in the radial and axial directions
the stresses can be taken as zero or negligible.
52 MECHANICS OF ENGINEERING MATERIALS

Fig. 2.9

(a)
O

A small element of arc of the ring of cross-sectional area A (= th)

rotating at uniform velocity w is shown in Fig. 2.9(b). The forces acting on
the element are the radial inertia force F and the circumferential tensile
stress o acting over the area A. The resolved component of the radial forces
inwards is

2oth sin ~ Doth for small values of 60

The mass of element is pthré@ where p is the mass per unit volume.
The radial centrifugal force, F = (pthr 60)w*r. For radial equilibrium,
the resolved inward radial force must balance the outward force F; hence

otha — F = 0

oth 60 — pthr’u"d0 = 0
and

a= purr

= pv [2.7]
where v is the tangential velocity of the central point of the element cross-
section.
It should be noted that the tensile stress is independent of the shape and
area of the cross-section of the rim.
An important practical example of the above effect of centrifugal action is
the stress set up in the blades of gas turbine rotors which rotate at very high
speed.

2.7 Thin shells under

Thin-walled shells are used extensively in engineering in two principal
pressure
forms: (a) made of reinforced concrete to form part of a civil engineering
structure and stressed partly by self-weight and partly by the environment,
namely wind, snow, etc.; (b) made of metal and used generally in
engineering as storage containers for liquid, powder, gas, etc. Stresses will
 STATICALLY DETERMINATE STRESS SYSTEMS 53

arise due to, say, uniform internal liquid or gas pressure, e.g. in a steam
boiler, or pressure due to weight of liquids or solids contained.
In general, a shell of arbitrary wall thickness subjected to pressure
contains a three-dimensional stress system. The stresses are perpendicular
to the thickness and in two principal orthogonal directions tangential to the
surface geometry. Each of these stresses has a non-uniform distribution
through the thickness of material, and the solution is statically
indeterminate. If, however, the wall thickness is less than about one-tenth
of the principal radii of curvature of the shell, the variation of tangential
stresses through the wall thickness is small and the radial stress may be
neglected. The solution can then be treated as statically determinate.
In the present treatment the following additional simplifying assump-
tions will be made: that the shell acts as a membrane and does not provide
bending resistance so that there are only uniform direct stresses present,
which are often called membrane stresses; that the shell is formed by a
surface of revolution and there are no discontinuities or sharp bends.
Consider first a general axi-symmetrical shell, Fig. 2.10(@), from which is
cut an element bounded by two meridional lines and two lines perpendicular
to the meridians as at (4). The notation is as follows:

o, = tensile stress in meridional direction, meridional stress

07 = tensile stress along a parallel circle, hoop stress
r, = meridional radius of curvature
r2 = radius of curvature perpendicular to the meridian
t = wall thickness

The forces on the edges of the element are, in the meridional direction,
ait ds, and, in the -perpendicular direction, o2tds,;. These forces have
components inwards towards their centres of curvature. The radial
components are 20;/ ds) sin(d0,/2) and 20 2tds; sin(d@2/2). For small
values of d@ the total radial force becomes

o1t ds7dO, + ont ds} dO,

Fig. 2.10

(a) (b)
54 MECHANICS OF ENGINEERING MATERIALS

or

ds} ds
oyt ds, — + o2t ds; —
r r2
The radial force due to the pressure p acting over the surface area of the
element is

p ds) ds

Thus, for equilibrium of the element,

d d
oyl fpPipenes + oot ds ae p dsj ds2
r| YZ

and, simplifying,

Sees [2.8]
al 12 t

In general 0; and o» are both different and unknown and so another

equilibrium equation has to be formulated relevant to the particular
problem, in addition to eqn. [2.8], in order to solve for the stresses.

Thin sphere
The simplest cases to which eqn. [2.8] may be applied is the sphere. The
symmetry about any axis implies that 0; =o02=0, and of course
rj =1r2 =r; therefore the circumferential stress in any direction is
ao = pr/2t. This stress may also be obtained by cutting the sphere across
any diameter (Fig. 2.11) and considering the equilibrium of either of the
free-body hemispheres in the same manner as shown in the next section for
the cylinder.

Internal
Fig. 2.11 resisting
forces

Force due
to pressure

Thin cylinder
When a thin-walled cylinder is subjected to internal pressure, there are axial
stresses, 7,, and a hoop (or circumferential) stress, o,. The latter can be
determined using eqn. [2.8] but not the former. This is because r2 = oo.
Using eqn. [2.8]

Oy , Oe_P
iP ee) t
STATICALLY DETERMINATE STRESS SYSTEMS 55

pr
Oy = re [2.9]

As a further illustration of the use of free-body diagrams, this problem

can also be solved from first principles.

Axial equilibrium The force acting on each closed end of the cylinder
owing to the internal pressure p, Fig. 2.12(5), is obtained from the product
of the pressure and the area on which it acts. Thus

Axial force = pr’

(b) (c) (d)

The part of the vessel shown in the free-body diagram, Fig. 2.12(d), is in
axial equilibrium simply under the action of the axial force above and the
axial stress, 0,, in the material. The radial pressure shown has no axial
resultant force. The cross-sectional area of material is approximately 2771,
and therefore the reacting force is 0, x 27rt, and, for equilibrium,

2nrto, = mrp

or

Ox == [2.10]

Circumferential equilibrium
Considering the equilibrium of part of the vessel, if the cylinder is cut across
a diameter as in the free-body diagram in Fig. 2.12(c) the internal pressure
acting outwards must be in equilibrium with the circumferential stress, a,
as shown. Consider the length of cylinder / and the small arc of shell
subtending an angle d@ shown in the diagram. The radial component of
force on the element is p x / x rd@; hence the vertical component is
p x1xrdé@ sin 0. Therefore the total vertical force due to pressure is

|pri sin 0d0 = 2prl

0
It is useful to note that the vertical force can also be found by considering
the pressure acting on the projected area at the diameter, which is p(27/).
The internal force required for equilibrium is obtained from the stress 0,
acting on the two ends of the strip of shell. Hence the internal force is
56 MECHANICS OF ENGINEERING MATERIALS

o,(2t/). For equilibrium, 2t/0, = 2rlp, or

Oy Z [2.11]
l

This may be seen to be the same as the value obtained using eqn. [2.8].
Comparing eqns. [2.10] and [2.11] it is seen that the circumferential
stress is twice the axial stress. Figure 2.12(¢) shows a small element of the
shell subjected to the axial and circumferential (hoop) stresses.

Example 2.2
A concrete dome is 250 mm thick, has a radius of 30m and subtends an angle of
120° at the support ring. Calculate the stresses at the supports due to self-weight.
The density for contrete is 2.3 Mg/m?. The dome is illustrated in Fig. 2.13.

Fig. 2.13

Consider the vertical equilibrium of a circular segment containing an angle

20; then if g is the force per unit area due to weight of concrete,
(2ar sin 8 x to; sin 0) + 2nr(r —r cos 0)q =0

gr (1 — cos @) qr 1
t sn?@ t (1+cos 6)
Since self-weight, unlike applied pressure, does not act radially
everywhere, eqn. [2.8] has to be modified to take account of the radially
resolved component of weight, giving
a GO:
Ss t cos @
1) 12 t

Substituting for 04,

pes 7
et, 1+ cos @ iP

The stresses at the supports are obtained by putting @ = 60° and

q = 2.3 x 9.81 x 250/1000 = 5.65 kN/m?:
Peas
Lee CH. 95 que NI
s

5.65 x 30
7 omer ay = 113kN/m?
 STATICALLY DETERMINATE STRESS SYSTEMS 57

2.8 Shear in a coupling

A pinned coupling for a tow bar is illustrated in Fig. 2.14(a). Tensile force F
is transmitted across the joint being carried by shearing action on sections
XX and YY of the pin.
From the earlier comments regarding complementary shear stresses, it is
apparent that the shear stress at the cut section in Fig. 2.14(4) cannot be
constant across the section. This is because the shear stress at the outside
surface is zero and hence the internal shear stress close to the surface must
be zero. However, this type of problem is only statically determinate if we
assume that the shear stress is uniform over the cross-section of the pin. It is
important to remember therefore that 7 is the average shear stress on the
cross-section.

Total internal shear force = 27A

Fig. 2.14

For equilibrium,

LTA

so the shear stress at the two sections of the pin is

ie
Te tg (2.12]

2.9 Torsion of a thin

In general the twisting or torsion of solid and hollow circular-section
circular tube
members is statically indeterminate and is discussed fully in Chapter 5.
However, as with thin-walled shells, if the radius of a circular tube is large
(say over ten times) compared with the wall thickness then the stress can
reasonably be taken as uniform through the thickness and the problem can
be treated as statically determinate.
In Fig. 2.15 a tube of mean radius 7 and wall thickness ¢ is subjected to
opposing torques at each end. The deformation action is that of twisting,
T
Fig. 2.15
58 MECHANICS OF ENGINEERING MATERIALS

and considering the equilibrium of part of the tube, by cutting it at some

section away from the end effects and perpendicular to the axis, an internal
reacting torque will be required for equilibrium and to maintain the twist.
This reaction torque will take the form of a ‘uniform’ shearing stress acting
on the cut face shown. The shear stress does not vary around the tube as this
would imply a variation in the complementary longitudinal shear stress and
thus a resultant axial force when there is none.
Considering a small element of the tube wall subtending an angle dé, it
has an area of tr d@. The tangential force / shown in Fig. 2.15 is given as the
shear stress acting over the element wall, so that F = tir dé.
This force exerts a moment about the central axis of Fr and the total
reacting torque is obtained by summing all such moments around the
periphery, giving
Vis
| rt dé = 2ntr’r
0
and this is in equilibrium with the applied torque. Hence 7 = 270r*r and
the magnitude of the shear stress is
th
1 SS SS [2.13]
2ntr?

2.10 Joints
All branches of engineering have to be able to connect together pieces of
material which may then have to carry working loads through the
connection or joint. The most common ways of constructing joints are by
holding the segments firmly together with bolts, rivets or welds. The forces
which have to be transmitted through a joint generally subject the
connectors, e.g. bolts, to shear, and the holes in the material through which
the connectors pass to tensile and compressive stresses. The determination
of most of these stresses may be regarded as statically indeterminate, owing
to the complex nature of load transfer through rows of bolts or rivets. Hence
for a thorough treatment the reader is referred to the texts specifically
concerned with engineering design. A few simple mechanical joints can be -
treated as statically determinate and the following example illustrates the
application of the principles of equilibrium to such problems.

Example 2.3
A closed-ended cylindrical steel pressure vessel is 2m in internal diameter. It is
made up from 10mm thick steel sheet formed into two semi-cylinders, which are
riveted along longitudinal single-lap seams (as shown in Fig. 2.16) with 12mm
diameter rivets at 36 mm pitch. The dished ends are also single-lap riveted onto the
cylinder at each end and the circumferential joints are made with 8mm diameter
rivets at 30mm pitch. If the maximum shear stress in any rivet is not to exceed
60 MN/m? calculate the maximum allowable internal pressure and the resulting
longitudinal and circumferential stresses in the wall of the cylinder.

Longitudinal seam Maximum shear force per rivet is

1 2
60 x 10° x %x = 6.8KN
 STATICALLY DETERMINATE STRESS SYSTEMS 59

Fig. 2.16

Be

Oo
°
°
@)
°
°
to}

SY

Longitudinal seam Circumferential seam

Separating force at joints due to pressure over one pitch length is

0.036 x 2 x p
The equilibrium condition is
0.036 x 2x p=2 x 6.8
so that p = 189kN/m‘’.

Circumferential seam Maximum shear force per rivet is

2
60 x 10° x x = S02KN
Number of rivets required is
2000
x —— = 209
30

Total force in shear allowable = 630kN

This must be in equilibrium with the longitudinal force due to pressure,

which is 1/4 x 2? x p.
Hence

mp = 630 and p=201kN/m’

Therefore the maximum allowable pressure is 189 kN/m’.
pr 189 x1
Circumferential stress, 0; = — = 18.9MN/m?
i t 0.01

18.9
Axial stress, 02 = crs 9.45 MN/m?

2.11 Summary
The purpose of this chapter has been to introduce the concepts of normal
and shearing stresses which represent the internal reacting forces per unit
area within the material. Engineering design of components to establish size
and shape depends on allowable values of stress which the material can
tolerate when subjected to applied loads or forces. In a few cases it is
possible to establish the stress for a given size or decide on the required size
60 MECHANICS OF ENGINEERING MATERIALS

for an allowable stress by consideration only of the equilibrium of the system

of external applied forces and internal reacting forces. These design
examples are said to be statically determinate.

Problems 2.1 A steel rod of varying cross-section is loaded as shown in Fig. 2.17.
Determine where the maximum stress occurs.
TD: It is required to make a large concrete foundation block which, when
supporting a comprehensive load together with its self-weight, will
have the same compressive stress at all cross-sections. Determine a
suitable profile.
Des The two parabolic cables of a suspension bridge are subjected to a
horizontal uniformly-distributed load of 80kN/m as shown in Fig.
2.18. Calculate the required area of the cables at each end if their
maximum permissible stress is 200 MN/m?. What is the compressive
load in the vertical columns?
Dee A suspension footbridge spanning a ravine is constructed with twin
cables and carries a horizontal uniformly-distributed loading of
2kN/m of span, which is 300m. The lowest point of the cables is
50m below one cliff support, which is 10m below the higher cliff
support. Determine a suitable cross-sectional area for each cable using
Fig. 2.17
a safety factor of 2 and a tensile stress of 300 MN/m?.

Fig. 2.18

150 m
-
2.5 A cable is freely suspended from two points which are at the same
horizontal level. If the cable is subjected to a uniformly-distributed
loading of m per unit length (self-weight, snow, birds), derive an
expression for the maximum tension in the cable.
2.6 A small boat is anchored as shown in Fig. 2.19. When the tide causes a
horizontal force of 1 kN on the boat the steel rope is tangential at the
anchor point. If the rope diameter is 20 mm calculate (a) the distance
between the boat and the anchor point, and (#) the maximum stress in
the rope. The density of the steel is 7800 kg/m? and the density of the
water is 1000kg/m°.

Fig. 2.19

Zl A chemical reaction process is carried out in a thin-walled steel

cylinder of internal diameter 400 mm with closed ends rotated about a
longitudinal axis at a speed of 5000 rev/min. Whilst it is rotating, it is
subjected to an internal pressure of 4MN/m/?. If the maximum
 STATICALLY DETERMINATE STRESS SYSTEMS 61

allowable tensile stress in any direction in the material is 175 MN/m’,

calculate a suitable shell thickness. Density of steel = 7.83 MN/m’.
2.8 The pipeline reducer shown in Fig. 2.20 has a uniform wall thickness
of 3mm. If the pipeline carries a fluid at a pressure of 0.7 MN/m’,
calculate the axial and hoop stresses in the reducer at a point halfway
along its length. Assuming that the coupling at the large end of the
reducer takes all the axial thrust, calculate the stress in each of the six
10mm diameter retaining bolts.

Fig. 2.20

300 mm

2 A thin spherical steel vessel is made up of two hemispherical portions

bolted together at flanges. The inner diameter of the sphere is 300 mm
and the wall thickness of 6mm. Assuming that the vessel is a
homogeneous sphere, what is the maximum working pressure for an
allowable tensile stress in the shell of 150 MN/m??
If twenty bolts of 16mm diameter are used to hold the flanges
together, what is the tensile stress in the bolts when the sphere is
under full pressure?
210 A thin wall conical shell supports a weight W as shown in Fig. 2.21.
Derive an expression for the stress in the wall of the cone.

Fig. 2.21

ZA A conical storage tank has a wall thickness of 20 mm and an apex angle

of 60°. If the vessel is filled with water to a depth of 3 m, calculate the
maximum meridional and circumferential stresses. ‘The water loading
is 9.81 kKN/m’.
Pigg04 In the mechanical digger shown in Fig. 2.22 the combined weight of
the bucket and its contents is 1000 kg and the centre of gravity is at G.
For the position shown in which the arm NKM< 1s horizontal, calculate
the shear stress in the pins at N and M. The inset sketch shows the
joint arrangement in each case, and the pin diameters at N and M are
10mm and 15 mm respectively.
62 MECHANICS OF ENGINEERING MATERIALS

Fig. 2.22 400 800 400 500 800

|\~<~—> | ~<. > |<—> |< —

2.13 A solid circular rod of 10mm diameter is coupled to a metal tube

using a bonded rubber cylinder as shown in Fig. 2.23. If an axial pull
of 10kN is applied to the rod, calculate (a), the shear stress between
the rod and the rubber, (4) the shear stress between the rubber and
the metal tube, and (c) the axial stress in the tube.

Fig. 2.23

rubber

36 mm ‘|

2.14 The hub of a pulley may be fastened to a 25 mm diameter shaft either

by a square key or by a pin, as shown in Fig. 2.24. Determine the
torque that each connection can transmit if the average shear stress in
the key or pin is not to exceed 70 MN/m?.

Fig. 2.24 6 X 6 X 25 mm key

®
10 mm dia. pin

2.15 A splined shaft connection as shown in Fig. 2.25 is 50 mm long and is

used to permit axial movement of the shaft relative to the hub during
torque transmission. In order to facilitate axial movement in the
connection it is to be desgned so that the side pressure on the splines
does not exceed 7MN/m?. Calculate the power that could be
 SPARICADE YY DEDERMIUNATE SDRESS SYSTEMS 63

Fig. 2.25 D=50mm

La d= 40mm

Shaft Hub

transmitted by the shaft at 2000 rev/min and the shear stress in the
splines at this power.
2.16 Extend the spreadsheet of Example 2.1 to calculate the torque about
the crankshaft due to the gas pressure.
2.17. A thin-walled circular tube of 50mm mean radius is required to
transmit 300 kW at 500 rev/min. Calculate a suitable wall thickness so
that the shear stress does not exceed 80 MN/m?.
2.18 (a) Construct a spreadsheet in which, given the applied torque on a
thin tube, its radius and wall thickness, the resulting shear stress
is calculated.
(6) Solve Problem 2.17 by trial and error. In other words try
different values of wall thickness until the stress under the
applied torque is approximated 80 MN/m’.
2.19 A torque tube consists of two sections which are riveted together, as in
Fig. 2.26, by 50 rivets of 4mm diameter pitched uniformly and the
radius of the mating surface of the tubes is 100mm. If the limiting
shear stress for the rivets is 180 MN/m? determine the maximum
torque that can be transmitted through the joint.

Fig. 2.26

2.20 (a) Construct a spreadsheet to solve Problem 2.19, given any values
for allowable stress, number of rivets, rivet diameter and tube
radius.
(b) Ifa torque of 20kN m must be transmitted, how many rivets are
needed.
CHAPTER
Stress—Strain Relations

i a el
As explained previously a statically indeterminate problem cannot be solved from the
conditions of equilibrium alone; additional equations ave required to find all the
unknowns. These equations are obtained by studying the geometry of deformation of
the component or structure and the /Joad-deformation or stress-strain relationship
for the material. These topics are dealt with in this chapter, but the detailed
application in worked examples is carried out in Chapter 4.
2a Se

3.1 Deformation
Deformations may occur in a material for a number of reasons, such as
external applied loads, change in temperature, tightening of bolts, irradiation
effects, etc. Bending, twisting, compression, torsion and shear or
combinations of these are common modes of deformation. In some
materials, e.g. rubber, plastics, wood, the deformations are quite large for
relatively small loads, and readily observable by eye. In metals, however, the
same loads would produce very small deformations requiring the use of
sensitive instruments for measurement.
Stress values do not always provide the limiting factor in design, for
although a component may be safe and employ material economically with
regard to stress, the deformations accompanying that stress need to be
considered. For example, large deflections are highly desirable in the case of
springs or cushions. On the other hand, too large a deflection of an
aeroplane wing can result, among other things, in a detrimental change in
aerodynamic characteristics. A lathe bed which was not sufficiently rigid
would not permit the required tolerances in machining.
In this and succeeding chapters there will be many problems in which
the analysis of displacements will be considered specifically in addition to
the determination of stress magnitude.

3.2 Strain
As explained in Chapter 2 the effect of a force applied to bodies of different
size can be compared in terms of stress, i.e. the force per unit area. Likewise
the deformation of different bodies subjected to a particular load is a
function of size, and therefore comparisons are made by expressing
deformation as a non-dimensional quantity given by the change in
dimension per unit of original dimension, or in the case of shear as a

Fig. 3.1
 STRESS-STRAIN RELATIONS 65

change in angle between two initially perpendicular planes. The non-

dimensional expression of deformation is termed strain.

Direct or normal strain

Consider the uniform bar shown in Fig. 3.1 subjected to an axial tensile load
F. If the resulting extension of the bar is 6 and its unloaded length is L, then
the direct tensile strain is
6
tea
If two bars identical in material, length L and cross-sectional area were each
subjected to a tensile load F, then the extension 6 of each would be the same
and the strain in both would be 6/L. If the bars are now joined end on end
and the same tensile force F is applied, the overall extension of the
combined bar will be 26. However, since the original length is now 2L, the
strain will be 26/2Z, i.e. the same as for the separate bars.
Direct strain is defined as the increase in length per unit original length.
If the bar had been subjected to a compressive force F causing a reduction in
length of 6, the strain would be
6
SSS
iE
The change in length is —6, so compressive strains are negative.
In the description above, the force F was applied to a bar of constant
cross-sectional area, so the stress was uniform at every point along the bar
and so in fact is the strain. If, however, the stress varies along the bar,
because for example the cross-sectional area varies, then so also will the
strain and it is necessary to define strain at a point by considering a small
element of undeformed length Ax, Fig. 3.2.

Fig. 3.2

When the bar is loaded, the left-hand end of the section will be displaced
to the right by an amount w. Owing to the strain in the element of length
Ax, the right-hand end of the shaded section will be displaced by an amount
u + Au. The increase in length per unit length in the section is therefore
Au
aye
66 MECHANICS OF ENGINEERING MATERIALS

In the limit as Ax tends to zero, the strain at a point is

< du
&
dx
The total increase in length 6 of the bar is given by
L il
5=| au = |edx
0 0
The same suffix notation is used for strains as for stresses. €, is the strain of
a line measured in the x-direction and ¢, is the strain of a line in the y-
direction.

Shear strain
An element which is subjected to shear stress experiences deformation as
shown in Fig. 3.3. The tangent of the angle through which two adjacent
sides rotate relative to their initial position is termed shear strain. In many
cases the angle is very small and the angle itself is used, expressed in radians,
instead of the tangent, so that

+ = LAOB — LA'OB' = ¢
Fig.
3.3

When /A’OB’ < /AOB, then ¥ is defined as a positive shear strain, and
when /A/OB’ > / AOB, 7¥ is termed a negative shear strain.
The shear strains corresponding to the shear stresses on the xy, yz and
Zam DlaMeSvaLe: V5.5) Vey Vers Vyas yee G00 Nene

Volumetric strain
The term ‘hydrostatic stress’ was used in Chapter 2 to describe a state of
tensile or compressive stress equal in all directions within or external to a
body. Hydrostatic stress causes a change in volume of the material which, if
expressed per unit of original volume, gives a volumetric strain, or
dilatation, denoted by e.
Volumetric strain may be expressed in terms of the three co-ordinate
direct strains. Let a cuboid of material have sides initially of length Ax, Ay
and Az. If the material is deformed, the new lengths of the sides will be
Ax(1+ €,), Ay(1+e,) and Az(1 +<,) respectively. The new volume is
therefore
AxAyAaz(1 + €,)(1 + €,)(1 + €2)
 STRESS-STRAIN RELATIONS 67

Neglecting products of small quantities, for small strains

New volume = AxAyAz(1 + €, + € + €2)

Original volume = AvAyAz

Therefore
. ; change in volume
Volumetric strain e =
original volume

e== (Cp Meyetes) 3.1

i.e. volumetric strain is given by the sum of the three direct co-ordinate
strains.

3.3 Elastic load—

Studies of material behaviour made by Robert Hooke in 1678 showed that
deformation behaviour of up to a certain limit the extension 6 of a bar subjected to an axial tensile
materials loading F was often directly proportional to F, as in Fig. 3.4(a). This
behaviour in which 6 « F is known as Hooke’s law. It is similarly found that
for many materials uniaxial compressive load and compressive deformation

Fig. 3.4

Ww Ww we

8
Re}
=)
g
ne}
ol
g]
ao)
a |

Deformation 5 Deformation 6 Deformation 5

(a) Linear elasticity (b) Exceeding the (c) Non-linear elasticity

elastic limit

are proportional up to a certain limit of load. A cylindrical bar which is

twisted about its axis by opposing torques applied at each end is also found
to have a linear torque—twist relationship up to a certain point. The
maximum load up to which Hooke’s law is applied is termed the limit of
proportionality. If in each of the above cases at any particular load the same
deformation exists both with increasing and decreasing load, and if after
completely unloading the body it returns to exactly its original size, then it is
said to exhibit the property of e/asticity. This behaviour exists only over a
certain range of load and deformation, the end point being termed the e/astic
limit. In general, the limit of proportionality is a shade lower than the elastic
limit. If the elastic limit is exceeded it is found that some permanent
deformation remains after removal of the load, as illustrated in Fig. 3.4(d).
The stress in the material at the elastic limit is called the yield stress, o,. In
most practical situations it is important to ensure that the stress in the
component does not exceed o,. The situations which arise when yielding
occurs are dealt with in Chapters 12 and 15.

3.4 Elastic stress-strain

If the load, F, in Fig. 3.4(a) is divided by the original cross-sectional area of
behaviour of materials the bar, A, and the extensions on the abscissa are divided by the original
68 MECHANICS OF ENGINEERING MATERIALS

Table 3.1
Material E G
(GN/M?) (GN/m’)

Steels 190-207 77-83

Copper 110-120 3746
Aluminium 69-70 24-28
Glass 50-80 20-35

length of the bar, Z, a graph of stress against strain is obtained. Since A and
L are constants the stress—strain behaviour is also linear in the elastic range.
The slope of the line is constant and may be expressed as

where £ is a constant for the material, and is called Young’s modulus of

elasticity. Since € is non-dimensional, E has the dimensions of stress, i.e.
force per unit area. Some typical values of E are given for a few materials in
Tavle sot:
A relationship between shear stress and shear strain may be derived from
a torsion test on a cylindrical bar in which applied torque and angular twist
are measured. The connections between torque and shear stress and twist
and shear strain will be derived in Chapter 5. It is sufficient to state here that
shear stress is proportional to shear strain within the elastic limit. Hence
Ty) = constant — G.
The constant of proportionality, G, is known as the modulus of rigidity, or
shear modulus, and has the dimensions of force per unit area. Typical values
of G for various materials are given in Table 3.1.
It can also be demonstrated experimentally that within the elastic range
volumetric strain is proportional to the hydrostatic stress, 7, as defined in
Chapter 2. The constant relating those two quantities is termed the bulk
modulus and is denoted by the symbol K. Thus

Be
e
It will be shown in Chapter 11 that the elastic constants, E, G and K are
related to one another.

Example 3.1
Calculate the overall change in length of the tapered rod shown in Fig, 3.5. It carries
a tensile load of 10kN at the free end, and at the step change in section a
compressive load of 2MN/m evenly distributed around a circle of 60 mm diameter.
E = 208GN/m.

Firstly consider the general case of a tapered rod fixed at one end and
subjected to a tensile load F at the other end as in Fig. 3.6. The mean radius
of any arbitrary slice at a distance x from the upper end is
w
Pet a ie
STRESS-STRAIN RELATIONS 69

So the cross-sectional area of the slice is

% 2
o m(r —(%m- n)>)

If the slice extends an amount dw under load then its strain is

= du e 1 |
dx ALE
The total extension of the rod is
L F
= d
% |AE .

. F IC dx
TE Jo [ro — (rm —n)(x/L))

SL
~ Enron

Returning now to the stepped taper rod in Fig. 3.5, the extension of the
lower part will be
5 10000 x 0.6
= +0.0319
mm
4B 508 x 109 x x x 0.024 x 0.012
The compressive load on the upper part will be treated as an axial
concentrated load of magnitude
2 x m Xx 0.03 = 0.067 MN = 188.5kN

Resultant load acting on part A= —188.5 + 10 = —178.5kN

 70 MECHANICS OF ENGINEERING MATERIALS

178.5 x 10° x 0.6

Compression of A =
208 x 109 x m x 0.07 x 0.035

= —().0669 mm

‘Therefore

Overall deformation of rod = —0.0669 + 0.0319

= —0.035mm

3.5 Lateral strain and

Polsson’s ratio If a bar is subjected to, say, longitudinal tensile stress then it will extend in
the direction of the stress and contract in the transverse or lateral directions,
Fig. 3.7(a). If the member were subjected to uniaxial compressive stress then
an expansion would occur in the lateral directions. It is found that the laterai
strain is proportional to the longitudinal strain, and the constant of
proportionality is termed Pozsson’s ratio denoted by the symbol v. Hence
Lateral strain = —v x direct strain (due to stress)

For most metals v is in the range from 0.28 to 0.32. It is important to

remember that lateral strain can occur without being accompanied by lateral
stress.

Final shape
Initial shape

Initial shape

(a)
(b)
Fig. 3.7

3.6 Thermal strain

The effect of a change of temperature on a piece of material is a small change
in size and hence strain. This can, in some circumstances, induce
considerable stresses. The dependence of size on temperature variation is
measured in terms of the basic quantity known as the coefficient of linear
thermal expansion, denoted by a. This defines the change in length per unit
length for unit increase in temperature.
A rod of length LZ has its temperature changed from 7p to T and the
accompanying change in length is
§ = aL(T — T)
 STRESS-STRAIN RELATIONS 71

This may be expressed as a thermal strain:

CSE 2
ies a(T ip — To)

Increasing temperature causes expansion and thus a positive strain, while

decreasing temperature results in contraction and negative strain. An
important feature about this behaviour is that if there is no restraint on the
material there can be strain unaccompanied by stress. However, if there is
any restriction on free change in size then a thermal stress will result.
The total strain in a body experiencing thermal stress may be divided
into two components, the strain associated with the stress, €,, and the strain
resulting from temperature change, ¢7. Thus

C=
6p 57 Sr

Hence

e=;+a(T— Tr)
which is a more general form of the simple uniaxial stress-strain law.

3.7 General stress—strain

relationships Consider an element of material as in Fig. 3.8(a) subjected to a uniaxial
stress, 0,; the corresponding strain system is shown at (4). In the x-direction
the strain is €, and in the y- and z-directions the strains are —ve, and —vé,,
respectively. These strains may be written in terms of stress as €, = 0,/E
and €, = €, = —vo;,/E, the negative sign indicating contraction.
The element in Fig. 3.9(a) is subjected to triaxial stresses 0,, 0, and o;.
The total strain in the x-direction is therefore composed of a strain due to
o,, a lateral strain due to oy and a further lateral strain due to a. Using the
principle of superposition, the resultant strain in the x-direction is therefore
the sum of the separate strains as shown at (4); hence
Tx UO, “VOx

Fig. 3.8

mio

(a) Stress (b) Strain

72 MECHANICS OF ENGINEERING MATERIALS

or

ce ply
Similarly

Ey = alc ae) [3.2]

and
OG, 2
Ey = lig= pina ans

Fig. 3.9

(a) Stress (b) Strain

Equation [3.2] represents three equations for strain in terms of three stress
components. These can be solved to give stresses in terms of the three
strains as

VE E
2 (ap ae) EP as fal Gamay

VE E
C= ea Kea EO a) rae [3.3]

VE E
O,= (6 6+ oe ee
(1 + v)(1 — 2v)
Equation [3.3] above is sometimes written as

Oy = Ae + Zpe,

Oy = re + Lye,

Oz, = e+ Lye,

where A and yu are the Lamé elastic constants of the material,

A= vE/|(1+v)(1 = 2v)], 24 = E/(1+v) and e = (€, +6, +€,) is the
volumetric strain. It will be demonstrated in Chapter 11 that ju is also the
shear modulus of the material.
 STRESS-STRAIN RELATIONS 73

There is no lateral strain associated with shear strain; hence the shear-
stress/shear-strain relationships are
Vyy = Txy y Tyz Tax
Ky AG J: z war

Plane stress In many practical situations the stress component in the z-direction is zero
and this is referred to as a plane stress condition. The above equations may be
applied with o, = 0, but note that the strain in the z-direction is not zero.
Usually €, is not of interest and the direct strains in the x—y plane can be
found from the stresses using
Oy WUO0y
eas (3.4a]
aed. Oy * VO >»

ys
Alternatively, if the direct strains are known, these equations may be solved
to find the direct stresses, giving
Ji
CE ines (€, + vey)
[3.45]
SN aa aa (€y + x)

Plane strain If the strain in the z-direction is zero, then this condition is referred to as
plane strain and the above equations may be applied using ¢, = 0. However,
zero strain in the z-direction does not imply zero stress in that direction.
This may be confirmed quite simply by considering the sample of material

Fig. 3.10

(a) Unloaded material (b) Plane stress

(c) Plane strain

74 MECHANICS OF ENGINEERING MATERIALS

in Fig. 3.10(4) subjected to tensile stresses in the x- and y-directions. Owing

to the Poisson’s ratio effect there would be a change in dimensions in the
z-direction. To keep €, = 0 it would be necessary to have a stress in the
z-direction, as shown in Fig. 3.10(c).

Thermal strains If in addition to strain due to stress there is also thermal strain due to change
in temperature, then eqn. [3.2] has a thermal strain term added as in eqn.
[3.5]. There is no thermal strain contribution to shear strain.

Ex Soe alte +oa,) +a(T — To)

a = - a(02 Oe) el alg) [3.5]

Ex Seals Oy) aC d= fg)

3.8 Strains in a statically

The stresses in a thin-walled cylinder under internal pressure were found in
determinate problem
Chapter 2 as a statically determinate problem, and now that stress—strain
relationships have been developed the strains in the cylinder can be found.
From eqns. [2.9] and [2.10] the axial and hoop stresses shown in
Fig. 3.11 are given by

O,=— and o,=—

Pe: Pat
It will be shown later that 7, is equal to —p at the inside surface and is zero
at the outside surface. Therefore o, is negligible in comparison with o, and
Oye
From the stress-strain equation [3.4a] the axial strain is
Ree pr vpr__ pr (1
ries os tae [3.6]
and the circumferential or hoop strain is
pr vpr pr
Ey = ( v)
YE Eee 3.7]
Taking a value for v of 0.3 it is found that the ratio of the hoop to the axial
strain is

eee Ale!
2 =—= 45
€, 0.4

whereas the ratio for the hoop to axial stresses was 2.0.
In the thin sphere there is only circumferential stress and strain. In this
case 0, = 0, = 0, so that
 STRESS-STRAIN RELATIONS 75

and since

aah
ay

=pare
Ae ee V) [3.8]

3.9 Elastic strain energy

When a piece of material is deformed in simple tension, compression,
bending or torsion, etc., within its elastic range, work is done by the applied
loading. On removal of the loading the material returns to its undeformed
state owing to the release of stored energy. This is termed elastic strain
energy and has the same magnitude as the external work done. It is the
release of strain energy in a stretched rubber band that enables a pellet to be
projected from a catapult. The release of strain energy in a metal loaded and
then unloaded in the elastic range is less obvious. The best illustration of it is
a metal spring, the performance of which depends on the energy stored
when the wire is bent and twisted during loading. The term resilience often
associated with springs has in fact a more general meaning as the strain
energy stored per unit volume.

3.10 Strain energy from

Consider the load—extension diagram shown in Fig. 3.12.
normal stress
The work done during a small increment of extension du is Fdu.
Therefore the total work done up to point N 1s
6 }
Work done = |Fdu= |oy
0 0

AE
ay iy
2h
but

AE6
1 ae Saye
Ly
0)

Work done = 5 aco

Fig. 3.12
EB
aii 2o

>
Extension
76 MECHANICS OF ENGINEERING MATERIALS

It should be noted that this is the area under the load—extension graph.
This is the stored energy in the specimen. If we divide by AL then the
stored energy per unit volume, U, is given by

SS
1 ——————
=
FraxO =
1 :
— 0
2 BAL, Z
Since
6 Obs
1 2
on =o - per unit volume [3.9]

3.11 Strain energy from

If a piece of material is subject to pure shear then the strain energy stored
shear stress
per unit volume is represented by the area under the shear-stress/shear-
strain curve shown in Fig. 3.13.
Hence U = sTY per unit volume, and since y = 7/G,
72
(GP esa per unit volume [3.10]
Ue
This expression only applies if the shear stress is uniform over the element
of material.

Fig. 3.13

Stress

Strain

Example 3.2
A car bumper is to be covered with a layer of energy-absorbing material so that in a
2 nvs collision the material will absorb the kinetic energy of the car and spring back
without permanent deformation afterwards. Which one of the candidate materials
below would give the cheapest solution?

Material Oy iB Density Relative

(MN/m’) (GN/m’) (kg/m?) cost/kg

Mild steel 200 200 7800 1.0

Polypropylene 50 ee 910 25)
Nylon 75 2 1150 5.0
 STRESS-STRAIN RELATIONS 77

Assume that the kinetic energy of the car at 2 m/s is a fixed quantity, as is
the available area of the bumper. If the material must spring back after the
impact, the largest stress to which it must be subjected is its yield stress o,.
The largest energy/unit volume that can be absorbed is therefore

The total energy that can be absorbed is therefore

les;
Ue =
DE
where A is the bumper area and ¢ the thickness of the energy-absorbing
layer.
The required thickness of the energy-absorbing layer is therefore
2
r= Un i oy

A 1 IB

The cost C of the bumper covering is

Cin Pal
where ¢, is the cost for unit weight of the material and p is its density.
Substituting in for the required thickness
Cp PE
6} mae? Z Or
Py
The best material from a cost point of view is therefore the one in which the
combination of material properties c,pE/ o, is least.
Table 3.2 shows that polypropylene is therefore the best material for this
application. Note that only relative costs per unit weight have been quoted
here. Absolute and relative prices fluctuate significantly over time.

Table 3.2
Material C,pE/ o,

Mild steel 0.039

Polypropylene 0.00055
Nylon 0.00204

3.12 Plastic stress—

With relatively few exceptions in the design of structures and machines,
strain behaviour of stresses and strains are limited to the elastic range of a material. However, it
materials is important to appreciate how a material behaves beyond the elastic range in
what is termed the plastic range of stress and strain. As there is some
elementary analysis involving yielding and plasticity in Chapters 12 and 15,
the basic concept will be briefly introduced here.
When a metal specimen is subjected to uniaxial tension to fracture, the
measurements of stress and strain will result in a diagram typically as shown
78 MECHANICS OF ENGINEERING MATERIALS

Fig. 3.14

Nominal
stress Nominal
stress

Strain Strain

(a) (b)

in Fig. 3.14(a) or (6). When the material has passed through the elastic range
and enters the plastic range it is said to be yielding. Stress continues to
increase with strain, but at a slower rate than in the elastic range, until a
maximum value of nominal stress (load divided by original cross-sectional
area) is reached which is termed the tensile strength. Thereafter the specimen
enters a failure range terminating in complete fracture at one cross-section.
The plastic range of strain can be perhaps 300 times the elastic range of
strain, and demonstrates what is described as ductility, namely the ability to
deform plastically.

3.13 Viscoelastic stress-

Some materials, notably plastics and rubbers, do not exhibit a linear elastic
strain behaviour of
range like metals, but show an interdependence of stress and strain with
materials time. In addition these materials have a ‘memory’ in the sense that current
strain or stress is always dependent on the loading history and after
unloading considerable recovery of residual strain can occur at zero load.
The above behaviour is termed viscoelasticity.
The simplest representation of viscoelasticity is the combined features of
a Hookean solid and a Newtonian liquid. The former provides an elastic
component, and the latter a viscous component (Fig. 3.15(a) and (4)). From
the latter, stress is proportional to strain rate, ¢, and the constant of
proportionality is 1, the coefficient of viscosity. Hence the simplest possible
relationship between stress, strain and time for linear viscoelastic behaviour

Fig. 3.15

Stress Stress

Slope = modulus Slope = viscosity

0 Strain 0) Strain rate

(a) Hookean (b) Newtonian
 STRESS-STRAIN RELATIONS 79

| +, is
o = Ee + é [3.11]
3
Because of the dependence on time and the behaviour known as creep (see
Chapter 21) experimentally determined stress-strain relationships demon-
Stress strate a strain-rate dependence and non-linearity as shown in Fig. 3.16. As a
result, the modulus of a viscoelastic material is not as constant as it is for
metals, but depends on the magnitude and nature of the loading applied to
£1> &)> Es the material. Design analysis for viscoelastic materials, e.g. the thermoplastic
components, although still following the basic steps of equilibrium of forces
and geometry of deformation, which are independent of the nature of the
Strain material, is somewhat more complex owing to the time-dependent stress—
strain relationship. This matter is dealt with further in Chapter 21, which
Fig. 3.16
specifically relates to creep and viscoelasticity.

3.14 Summary
In this chapter the concept of strain was introduced together with the
relationships that exist between stress and strain for engineering materials.
The ‘stage’ is now set for the analysis of most engineering design problems
involving a knowledge of strength and stiffness. The next three chapters apply
these basic concepts to the design of important, widely used engineering
components. An early appreciation of the importance of strain energy is
essential for all engineers, since it has a major part to play through energy
theorems in design analysis and further in relation to the mechanics of
yielding and fracture.

Problems 3.1 Determine the overall change in length for the steel rod shown in
Problem 2.1 (Fig. 2.17). The length of the upper section is 300mm
and the two lower sections are each 400 mm long. E, = 200 GN/m’.
3.2 State whether the components, illustrated in Fig. 3.17 are in a state of
plane stress or plane strain.
(a) A grinding wheel rotating at high speed.
(b) A plate of steel being cold-rolled.
(c) A long thick-walled cylinder containing a fluid pressurized by
two end pistons.

Fig. 3.17

3.3. A 60mm diameter mild-steel sphere has parallel flats machined on it

20 mm each side of the central axis. If a compressive load of 5 MN is
applied perpendicular to the flats, calculate the decrease in length
along the loading axis. The modulus of steel is 207 GN/m’.
3.4 A truncated cone with a base radius, R, and a height, H, is attached to a
horizontal surface by its base and truncated at height h. If the material
80 MECHANICS OF ENGINEERING MATERIALS

LL

of the truncated cone has density p and modulus £, show that its
extension as a result of its own weight is given by

6 = pH’(H + 3h)/6E(H +h)

5:5 A copper band 20 mm wide and 2 mm thick is a snug fit on a 100 mm
diameter steel bar which may be assumed to be rigid. Determine the
stress in the copper if its temperature is lowered by 50°C.
Gi 1890055 EH 105 GN/m?.
3.6 What are the shear strain and angle of twist per unit length for the
tube in Problem 2.17. G = 85 GN/m’.
37 Express the stresses 0,, 0), 7, in terms of the three co-ordinate strains
and the elastic constants. Obtain similar expressions for the cases of
plane stress, 0, = 0 and plane strain, «, = 0.
3.8 Program the formulae of eqns !3.2] and [3.3] into a spreadsheet so
that, given the material properties E and v, and the stresses on an
element of material, all the strains are calculated.
ow For Plane Stress loading, construct a spreadsheet whereby, given the
Young’s modulus and Poisson’s ratio
(a) o, and oy, are entered and ¢, and e, are calculated.
(b) e€, and €, are entered and o, and o, are calculated.
3.10 Show that the volumetric strain, e in an element subjected to triaxial
stresses 0,, 0, and o,, is given by
a) ley
e (G2 On.)
Peay:
Determine the maximum strain and change in diameter of the
cylinder in Problem 2.7 if E= 208 GN/m? and v = 0.3.
3,12 A rectangular steel plate of uniform thickness has a strain gauge
rosette bonded to one surface at the centre as shown in Fig. 3.18. It is
placed in a test rig which can apply a biaxial force system along the
edges of the plate. If the measured strains are +0.0005 and +0.0007 in
the x- and y-directions, determine the corresponding stresses set up in
the plate and the strain through the thickness. E = 208 GN/m? and
y= 035

Fig. 3.18

3:13 A trapeze artist weighs 50 kg and is balanced at the centre of a 3mm

diameter wire tightrope of 20m length. There is an initial stress of
100 MN/m? in the tightrope before the artist balances on it.
Determine the strain energy stored in the wire. E = 208 GN/m?.
3.14 Determine the shear-strain energy stored in the torsion tube of Fig.
2.13. G = 85 GN/m?.
STRESS-STRAIN RELATIONS 81

3.15 A block of material is subjected to strains €,, €, and zy.

(a) Derive a spreadsheet to calculate the new position of the corners
of a unit square of material in the xy plane i.e. points (0,0), (1,0),
(1,1) and (0,1). Assume point (0,0) does not move.
(0) Normally the strains in engineering materials are very small. Add
a ‘strain magnification factor’ which increases the displacements
by a given factor.
(¢) Link the input cells in which strains are specified to the output of
the calculation of strains from stresses in Problem 3.9(a), so that
the deformation in response to an applied stresses can be
calculated.
(d) Create an x-y graph of the original and magnified deformed
shape of the unit square. Observe how the deformed shape
changes in response to different stress values.
 CHAPTER
Statically Indeterminate Stress
Systems

The subject has now been developed sufficiently so that statically indeterminate
problems involving uniform normal tensile or compressive stress can be solved. The
principles of solution are fundamental to the mechanics of solids and structures and
require the writing of:
1. Equations of equilibrium of forces — external (applied); internal (as a function of
stress and area).
2. Equations describing the geometry of deformation or compatibility of
displacements.
3. Relationships between /oad-deformation or stress-strain for the materials.
It is of the utmost importance to remember the above principles and apply them in
a logical manner to all future problems. This chapter is devoted to illustrating the
application of these principles to a variety of problems encountered in design. The
solutions have been deliberately worked in letter symbols rather than numerical
values to enable the steps to be followed more easily.

4.1 Interaction between

A series of examples will now be studied involving different materials and
components of different also different component configurations. It is important to note the
stiffness differences in equilibrium and deformation relationships that arise in these
two component arrangements.

Example 4.1
A bimetallic rod is subjected to a compressive force, F, as shown in Fig. 4.1.
Determine the overall change in length.

Fig. 4.1

The two materials will be denoted by subscripts 1 and 2.

Equilibrium It should be clear that the same force is carried through each
material, so that

Fi =h,=F [4.1]

Geometry of deformation ‘The overall change in length is the sum of the

changes in the two parts of the rod, so that

6=6 +h [4.2]
 STATICALLY INDETERMINATE STRESS SYSTEMS 83

Stress—strain relations Since it is a simple uniaxial stress system

Sg ek [4.3]

cee ES [4.4]
E2

Let the cross-sectional areas be A; and 42. Then eqns. [4.3] and [4.4] can be
re-expressed as
Fy 6] Fy 6)
28 2 phe dee 4.5
Ave Oi gel 2
Substituting the values of 6; and 6) into eqn. [4.2] gives

$= F ih i: F 2 _F h ie h
AE, Ayk, AE, Ayk)

For a series of 1 bars

ge De bi oe
4.6]
,

Example 4.2
Two components of different materials are arranged concentrically and loaded
through rigid end plates as shown in Fig. 4.2. Determinate the force carried by each
component.

Fig. 4.2

(b)

Equilibrium In this case the load is shared in some unknown proportions

between the two parts, so that

Fi +P) =F [4.7]
Geometry of deformation If the unloaded lengths, /, are initially the same,
then they will remain the same under load; hence

OS) == [4.8]

Stress—strain relations Again, for a simple uniaxial stress situation,

oO
ale E, and ee Ey [4.9]
a ay)
84 MECHANICS OF ENGINEERING MATERIALS
ET

From eqns. [4.8] and [4.9]

6 6
F = Ey At and Fy = E) A2- [4.10]
l
Substituting in the equilibrium equation [4.7],
6 6
E\A,-+
£2 A,-=F
i) /
Thus

polvcecieelle mang [4.11]

E\A,
+ E2 A2

FEA, FE, Ap
FF, = —————— Ol Ii, = 4.12
ee ega, Se eae ene be

Example 4. eh Sea Pes ae ee eee

ane a Figure 4.3 shows the connecting rod of an internal combustion engine. At the big
end, where the rod is attached to the crankshaft of the engine, the bearing is split
and the bearing cap is secured by two bolts which engage in threads in the rod as
shown in Fig. 4.4. During assembly, tie bolts will be tightened sufficiently to bring
the mating surfaces into contact and then either a prescribed torque will be applied
with an instrumented torque wrench or the bolt heads will be turned through a
prescribed angle. In either case the tension in the bolt will be close to the yield
stress of the material before any external load is applied. Why should this be done?

Fig. 4.3 Connecting rod

The analysis of Example 4.2 can be extended to components where there is

an initial mismatch in size so that internal stresses are created even when no
external load is applied. This analysis is useful in the design of bolted joints,
especially in those cases where there is a risk of failure due to ‘fatigue’ or
repetitive loading, a phenomenon which is described in Chapter 20.
 STATICALLY INDETERMINATE STRESS SYSTEMS 85

Owing to the torque applied to the bolt heads there will initially be
tension in the securing bolt which is opposed by an equal and opposite
compression in the bearing cap. A general method of dealing with this
situation is to treat the bolt and the surrounding material in the bearing cap
as being subject to uniaxial stress. The two components can be represented
as two springs in parallel as shown in Fig. 4.4(4), but owing to the preload in
the system the bolt (spring 4) is stretched by an extra amount 6.

Fig. 4.4 Bearing cap

(a) Bolting face

Bearing cap
spring ko

Bolt spring k,

(b)

Equilibrium The total load supported is the sum of the load in each
component:

tp we

Geometry of deformation The bolt is stretched by an extra amount 69; hence

6; = 6+69

a)

Stress—strain relations The bolt is in simple uniaxial tension. The bearing

cap has a varying cross-sectional area between the mating face and the
underside of the bolt head. In reality the stress distribution can be quite
complex in this component but by far the largest stress is the direct stress in
the direction of the bolt axis. The simplest model of the loading is to treat
the bearing cap as a simple member loaded in uniaxial compression. Its
modulus and length are known. The cross-sectional area could be integrated
over the length as in Example 3.1, but the simplest approach to obtaining an
86 MECHANICS OF ENGINEERING MATERIALS

answer of the right order of magnitude is to take the area of the equivalent
member as the area of the mating surface. As can be seen in Fig. 4.5, this is
around four times the area of the bolt.

Fig. 4.5 Conrod bearing cap

The approximations introduced above are typical of those necessary in

engineering practice to estimate likely stress levels in components during the
early stages of design or to check the results of other more detailed
calculations. Developing confidence and competence in choosing the best
analytical model of a real engineering component is one of the most difficult
tasks to teach and to learn. A useful principle is that it is better to have
answers which are wrong by a factor of 10 than to have no estimate at all!
Even the simplest model will give an appreciation of how the component
works and the likely constraints on improving performance.
Assuming uniaxial stress in both components
EA
Ro 7 (6 + b0) = ki(6 + 60)

E,A
jie 75 = kb

The total force is therefore

Frp=F,+h,= ki(6 + 60) + kd

Consider first the case when no external load is acting on the joint, i.e.
Fr = 0. The deflection dp,eioaa is found from

ky (preload ti 50) ar R2Oyecloud =0

or
 STATICALLY INDETERMINATE STRESS SYSTEMS 87

From this deflection, the preload force in the joint is

ky ko
Fyveload =F, =—F,= = ky preload =
ky + ko

When the applied load Fr is not zero, then

Fr = k16 + ky 69 + kod
The resulting deflection is
eel 109
ky R2
From this deflection the force in either component of the joint can be found.
For the bolt

Fy = kyd
+ ky69

=e ath thie

Sane , ae
= aor al peiad

The significance of this result is that if k2 >> k; and the externally applied
load Fr is varying, the load F, on the bolt has a large static component
F reload but only a small varying component due to the varying Fr. If the
cross-sectional area of the bearing cap is four times that of the bolt, then the
variation in bolt force is only one-fifth of the variation in total load. This
greatly reduces the possibility of a fatigue failure of the bolt due to repeated
loading, as discussed in Chapter 20.
For the connecting rod material surrounding the bolt the load is

Fy 1B i Toy aa
a ki + ko

This can only be negative, i.e. compressive, since the conrod and the bearing
cap will separate rather than carry a tensile force and a fretting failure due to
relative motion may occur. The limiting case of a zero load occurs when

ky
Poreload = 18
ki +k

In the present example, a preload of at least four-fifths of the maximum total

load is required.

Example 4.4
A rigid member AB, the weight of which can be neglected, is supported horizontally
at the pin joints A and C and by the spring at B as shown in Fig. 4.6(a). The stiffness
of member CD is 2kN/m and of the spring is 5kN/m. Calculate the force in CD,
 MECHANICS OF ENGINEERING MATERIALS

which is initially unstressed, and the reaction at A when a vertical load of 10 KN is

applied at B.

Fig. 4.6
i Fy 10

:
441
zee
{ La 2
A aft
Fy

(b)

Equilibrium Let the reaction at A be R and the forces in CD and in the

spring be F and F» respectively.
Vertical: 10+R—F,—F,=0 [4.13]
Moments about B: 3R—2F,; =0 [4.14]

Geometry of deformation ‘The deformation is as shown in Fig. 4.6(4), so that

from similar triangles
a
mene
——=- [4.15]
4.15

Load—deformation relations

Fy
—
i= = 2kN/ m and
ee
——
i = 5kN/m ;
[4.16]

From eqns. [4.15] and [4.16]

2
Fi 1=T5
== Fy [4.17]7

From eqns. [4.13] and [4.14], eliminating R,

F, +3F, = 30 [4.18]
Hence the force in CD is F, =1.276kN and the reaction at A is
k= oF = 0.851 kN.

4.2 Restraint of thermal

When the temperature of a piece of material is changed, its size will also
strain
change, and when expressed non-dimensionally this is termed thermal strain
(see Chapter 3). If the thermal strain is not restricted in any manner, then at
the new steady temperature a previously unstressed component will remain
unstressed. Clearly if there is any form of restraint to free change in size
then thermal stress will result. The following two examples will illustrate this
situation.
 STATICALLY INDETERMINATE STRESS SYSTEMS 89

Example 4.5
The bimetallic component illustrated in Fig. 4.2 consists of a steel rod of cross-
sectional area 600 mm? coaxially surrounded by a copper tube of cross-sectional
area 1200 mm’. It is not subjected to any external load but its temperature is
changed from 20 °C to 100°C. Determine the axial stresses set up in the copper and
the steel.

E, = 205 GN/m’, E, = 115 GN/m’,

Ome WIS 106) Cn 16077)CG

Equilibrium Since there is no applied external force, the sum of the internal
forces in the copper and steel must be zero. Therefore
F.+F,=0 [4.19]
or

O¢ A, a 0A, = (0 “i [4.20]

Geometry of deformation Since the two materials are initially stress-free and
their ends are fixed together the total strain must be the same for each.
Therefore
p= 6.) OF (En-Pe7), = (es Fer), [4.21]

where €, = strain due to stress, €7 = strain due to temperature change.

Stress—strain relation

== +o(T=T) [4.22]

Os
Es = E. aim a;(T = To) [4.23]

Equating ¢€, and ¢, from eqns. [4.22] and [4.23],

+ a,(T — To) =F +0,(T — To)

¢ s

Using eqn. [4.20] to eliminate o, gives

or

= AEE A(T ae To) (Qs ra Q,)

oe LEA
90 MECHANICS OF ENGINEERING MATERIALS

and

A,E,E;(T — To)(Q@s — Qc)

0; =
A,E, + A,E,
The negative sign for o, does not necessarily indicate a compressive
stress but simply that it is opposite in sign to ,. The type of stress in each
material is determined by the numerical values of the quantities, 7p, 7, a,
and a,. Substituting the numerical values in the above equations gives

o, = —21.7MN/m
and

a Ee :
ae 7 = +43.4MN/
i A, 600 ip a
Thus for an increase in temperature, ~, being greater than a,, the copper
is prevented from expanding as much as if it were free and is put into
compression. The steel is forced to expand more than it would if free and is
therefore in tension.
If there was the situation of both an applied load as in Fig. 4.2 and a
change in temperature then the solution could conveniently be obtained by
using the principle of superposition and adding together the two separate
results.

Example 4.6
A copper ring having an internal diameter of 150mm and external diameter of
154 mm is to be shrunk onto a steel ring, of the same width, having internal and
external diameters of 140mm and 150.05 mm respectively.
What change in temperature is required in the copper ring so that it will just
slide on to the steel ring?
What will be the uniform circumferential stress in each ring and also the
interface pressure when assembled and back at room temperature?
Assume that there is no stress in the width direction, E, — 205GN/m?,
E, = 100GN/m?, a, = 18 x 10-°/°C.

The circumferential length of the copper ring has to be increased by heating

till it is fractionally larger than the circumferential length of the steel ring.
Minimum required change in circumference = 7d, — 7d,

aT 0.05

Change in circumference due to heating = 1d, x a(T — Tp)

=n x1 50x18 KO °C — Ty)
Therefore

eM) fide ane lah0.05 o* = 18.5.-C

°~ 2700 x 10-6
 STATICALLY INDETERMINATE STRESS SYSTEMS 91

When the assembly has returned to ambient temperature assume that the
circumferential stresses in the copper and steel are uniformly distributed
over each cross-section.

Equilibrium Let the width of each ring be w; then, with thicknesses of

2 and 5mm respectively,
(pix Daa io <x 5)oo— 0

O, = —2.50; [4.24]

Geometry of deformation The circumferential strains in the copper and

steel must be the same at the mating surface, so
a= c, [4.25]

Stress—strain relations

e. = 0./ Land, n= 07 //E co a Ad [4.26]

since it is only the copper ring that has the thermal strain component. From
eqns. [4.25] and [4.26]
O;

EE,
O;
+ a,AT 4.27]
Using eqn. [4.24],

le” 325
(7 . =) = 18 x 10° x 18.5

The negative sign is due to AT being a reduction in temperature.

Substituting for FE, and E,,

o;=—11.15 MN/m* and o, = +27.9 MN/m?

The radial pressure at the interface between the two rings may be treated as
a thin cylinder under internal or external pressure, so that

p = o,t/r = 27.9 x 2/75 = 745 kN/m?

4.3 Volume changes

The following problem analyses the change in volume of a vessel subjected
to pressure and makes use of the relationship between hydrostatic stress and
volume strain.

Example 4.7
A thin, spherical, steel shell with a mean diameter of 3m and a wall thickness of
6mm, is just filled with water at 20°C at atmospheric pressure. Find the rise in
gauge pressure if the temperature of the water and shell rises to 50°C, and then
determine the volume of water that would escape if a small leak developed at the
top of the vessel.
92 MECHANICS OF ENGINEERING MATERIALS

Steel: Young’s modulus, E = 200 GN/m?

Coefficient of linear thermal expansion a = 11 x 10°°/°C
Poisson’s ratio= 0.3
Water: Bulk modulus, K = 2.2 GN/m?
Coefficient of volumetric thermal expansion 0, = 0.207 x 10°-3/°C

Equilibrium Let the gauge pressure in the sphere after the rise in
temperature be p; then from Chapter 2 the equilibrium condition is
pr
o =1 Ve p [4.28

Geometry of deformation If there is to be a pressure at all then the water and

sphere must remain in overall contact, and hence
Change in volume of sphere = change in volume of water
or
€sphere — €water [4.29]

since the original volume is the same for each.

Stress—strain relations For the water the total volumetric strain is the sum
of that due to pressure and that due to thermal strain:

Cwater = —(p/K) + (T — To) [4.30]

For the sphere the total volumetric strain is a function of strain due to
stress (from pressure) and thermal strain:

sphere — €stress + thermal [4.31]

From egqn. [4.30],

p
Cater = ~ 2.2 % 109 a= (0.207 x 10 x 30)

= —(0.445p x 10~°) + (6210 x 10°)

The change in the internal capacity or volume of the sphere is

$n(r+ ér)> — Sir

which gives, neglecting products of the small quantity 6r,

+7 x 3r°6r

Expressing this as a volumetric strain,

ta
3 x 3r6r = br
tar r
It will now be shown that 6r/r is the linear or circumferential strain in the
material of the sphere.

Change in circumference = 27(r + 6r) — 2ar = 2n6r

 STATICALLY INDETERMINATE STRESS SYSTEMS 93

Therefore,

; : i 2a
Circumferential strain = = —

so that

Volumetric strain of vessel = 3 x circumferential strain

Now, the hoop strain is given by eqn. (3.5):

(OF VO
epee Cl = 10)

benleseaee 8
Therefore the total volumetric strain is

e sphere
125 x 0.7p
(Fr Z 10° 330 x 10
as )

Hence, using eqn. [4.29],

125 x 0.7p
ay.
(—0.445px [OE
10’) G7T0 x 10-2)
+ (6210 107°) 3 Se
Sacie + (330 x 10-*))

from which

p = 2.97MN/m’
The volume of water which escapes through the leak is simply the
difference of the free thermal expansions of the water and the vessel, since
obviously there is no pressure present to affect the issue.
Volume of water escaping

= [(6210 x 107°) — (3 x 330 x 10~°)] x $m x 15003

— 7.4 x 10?m?

4.4 Constrained material

Examples 4.1 and 4.2 introduced the concept of two materials reacting
against each other in a simple uniaxial load and stress situation. There are
some engineering components in which material is constrained on a two- or
three-dimensional basis, so that more complex stress and strain distributions
result. This type of situation is covered in detail in Chapter 14. The
following example, although somewhat contrived, provides an illustration of
the use of the general stress—strain relationships (see eqn. [3.2]).

Example 4.8
A cylindrical block of concrete is encompassed by a close-fitting thin-walled steel
tube of inside radius r and wall thickness t as shown in Fig. 4.7. If the concrete is
subjected to a uniform pressure p, on its ends, determine the required ratio of r/t so
that the steel walls exert a lateral pressure pz on the concrete of at least 0.2p,.
Assume that there is no friction between the concrete and steel. The moduli and
94 MECHANICS OF ENGINEERING MATERIALS

Poisson’s ratio for concrete and steel are E,, 1, and E,, 1, respectively. Assume that
there is no friction between the steel and the concrete. E, = 15E,, v, = 0.25.

Fig. 4.7

Concrete

The physical nature of the problem is that as the concrete is compressed it

expands laterally on to the tube. The resulting internal pressure in the tube
is resisted by hoop stresses in the steel.
Superscripts c and s will be used to annotate the stresses and strains in
the concrete and steel. The stresses in the concrete are o{. = —p; and
i= —p2. Also, by symmetry, o{ = —p2. In the steel, the hoop stress
0, = por/t. roe a thin-walled tube, r/t > 1, this is the only significant
stress and so a, =o, ~ 0.
Since the tube is close-fitting, both steel and concrete will expand
laterally by the same amount and the circumferential strains are equal i.e.
See BR

oy oy
Using the stress-strain relations, eqn. [3.2], this implies

por 1 p2
rth ee eae
kes = pay ae)
Bes eee

n=nw/ [E+ 0-2) rE,

whenp2 = =)
1 VY;
E.
: Fe+a-)

Taking £,—.15E v7, = 0:25

Talis 1
Se
: py or = Ne] ee vs
5.5

A more general analysis involving interference between thick-walled

cylinders is illustrated in Example 14.3.

4.5 Maximum stress due

In Chapter 3, in the section dealing with the elastic strain energy stored
to a suddenly applied load
under uniaxial stress, the load was applied gradually so that the work done
 STATICALLY INDETERMINATE STRESS SYSTEMS 95

Fig. 4.8

fo) Time

(b)

was the average load times the distance moved at the point where the load
was applied. Now suppose that a bar fixed at the top with a flange at the
lower end, as in Fig. 4.8(a), has a mass, m, suddenly released on to the
flange. Let the momentary maximum extension, strain and stress in the bar
be 6’, e’ and o’ respectively. In addition, if the masses of the bar and flange
are small compared with the mass, m, then a reasonable approximation to the
behaviour is made by neglecting the effect of the former. Because the full
load moves through the extension, 6’,
Work done = W6' = mgé'

The strain energy stored per unit volume momentarily is }0’¢’, so that
Ce so'e'al = 5o'aé!

Since the work done is equal to the strain energy,

mg5' =40' ad’

Thus o/ = 2mg/a; but mg/a = o, the stress due to a gradually applied load,
so that

g =20 [4.39]
or the momentary maximum stress due to a suddenly applied load is twice
the stress for a gradually applied load. The bar will subsequently oscillate
about the statical equilibrium position while the stresses and deformations
rapidly die away, as shown in Fig. 4.8(4), to the value obtained for a
gradually applied load. However, the momentary stress intensification by a
factor of 2 might have serious consequences on a component.

4.6 Maximum stress due

An extension of the above problem is the case where the mass m is dropped
to impact
on to the flange from a height 4, causing a momentary extension of the bar
6'. The total potential energy is mg(h +6’), and the momentarily stored
strain energy is

Ui = toad!

Then neglecting the mass of the bar and flange and assuming no losses of
energy during impact,

30'a5' = mg(h + 6’)

96 MECHANICS OF ENGINEERING MATERIALS
‘sms
A A EL A AL ALL,

or

iP alone [4.40]
a a

Now 6! = (o’/E)I, and mg/a = c is the final steady stress; substituting into
eqn. [4.40],
Eh
(@ loo Je ae 0)

Therefore
Eh 1/2
o =o+ (0+ 20-7) (4.41]

It will be seen from this equation that there are two practical situations to
consider:
(a) as h tends to zero then o’ = 20, which is the result obtained in the
previous section; and
(6) for larger values of h, the term 2cEh// is large relative to 0” and so eqn.
[4.41] becomes

Gee
a= wie

The true situation for the stress during impact of one body on another is
more complicated than is indicated in this approximate analysis. In practice
the deformation and stress imposed on the bar at the point of impact take
time to propagate along the length of the bar. The stress wave, as it is called,
on reaching the fixed end of the bar will be reflected towards the point of
initiation. Nevertheless, the above solution provides a good approximation
in most practical situations.

Example 4.9
The lower part of a child’s pogo stick is illustrated in Fig. 4.9 when the child and
stick are just about to descend to the ground and the spring is undeformed.
Fig. 4.9
Determine the momentary maximum stress in the steel compression tube on impact
with the ground, and compare this with the final steady stress. It may be assumed
that the outer sleeve and supports are rigid and that the ground does not deform.
The weights of the various parts may be neglected. E — 208 GN/m?: spring stiffness
= 18 kN/m.

Let 6 be the compression of the spring, and x the compression of the tube. If
the force in the spring on impact is momentarily F then the strain energy
stored in the spring and tube is

250 mm Fé o&

19 dia. | Tube volume = 2505 (25° — 19?) x 107° = 51.8 x 107° m?

25 dia. 75mm

WSS
 STATICALLY INDETERMINATE STRESS SYSTEMS 97

Potential energy lost on impact = 2 x 180(0.075 + 6 + x)

and

5 F Fl FX 250 = 10°
18000 "AE 208 x 10-6 x 208 x 109
— SSS hal = —

Equating potential and strain energies,

F - 250F x 10~°
360 (0.075ae
18000 208 x 208

wits i FE S651 8510;°

36000 2 x (208 x 10°)? x 208 x 10°

(F? x 0.0278 x 10-°) — (F x 20 x 10-*) —27=0

F’ =720F =(970.X 10°). =0

from which Ff =1410N.

1410
Omax == 508 x 10-6 _ 6.78 MN/m 2

360
Final steady stress, 0 = 703x106 > 1.73 MN/m?

4.7 Summary
The importance of this chapter centres on the three requirements of
principle set out in the introduction. If a problem cannot be solved by
equilibrium statements alone, it is then described as statically indeterminate
and we have to assess the geometry of deformation and also link stress and
strain through the modulus and Poisson’s ratio for the material. The bulk of
the chapter has been devoted to a series of illustrative examples which set
out the above three steps as appropriate to each case. It is therefore very
important to work through each example to achieve a full understanding of
the formulation of the equations, since from that point the solution is merely
manipulative computation. The foregoing principles will be extensively used
in the next two chapters on torsion and bending and thus this chapter must
be thoroughly understood.

Problems 4.1 A composite shaft consists of a brass bar 50mm in diameter and
200 mm long, to each end of which are concentrically friction-welded
steel rods of20mm diameter and 100 mm length. During a tensile test
to check the welds on the composite bar, at a particular stage the
overall extension is measured as 0.15 mm. What are the axial stresses
in the two parts of the bar? Fy,as; = 120 GN/m?, Estee: = 208 GN/m?.
4.2 A rigid bar AB is supported by 2 cables and subjected to a 10 kN force
as shown in Fig. 4.10. Calculate the vertical displacement at the point
of force application. E, = 200 GN/m’.
98 MECHANICS OF ENGINEERING MATERIALS

Fig. 4.10

Area = 80 mm2

0.6m
Area = 100 mm2

10 kN

4,3 A spring-loaded buffer stop is illustrated in Fig. 4.11. The spring,

which has a stiffness of 6kN/mm, is located on the end of a steel tube
of inner and outer diameters 21mm and 29mm respectively and
150mm length. Determine accurately the total axial displacement of
the system under a load of 30kN. What would be the simple
approximate solution? E = 207 GN/m?.

Fig. 4.11

29 mm

44 (a) Create a spreadsheet to calculate the extension due to a load F of a

number of bars in series, given the modulus £, cross sectional
area A and length L of each bar. Calculate also the stress in each
bar.
(b) Use the spreadsheet to find the extension of the bars in Problem
4.1 under a unit load. Use trial and error (or the spreadsheet
‘Solve-for’ facilities) to find the load to cause a total extension of
0.15 x 10-3m.
(c) Use the spreadsheet to solve Problem 4.3 by replacing a bar
stiffness EA/L with the spring stiffness.
4.5 The spring shown in Fig. 4.12 has an unstretched length of 200mm
and a stiffness of 500 kN/m. If it is compressed to 150mm, placed
over the aluminium bar AB and then released, calculate the forces
which will be exerted on each of the two side walls. Before the spring
is released there is a gap of 0.15 mm between the end B and the right
hand wall. E.) = 70 GN/m’.
4.6 The steel bolt shown in Fig. 4.13 has a thread pitch of 1.6mm. If the
nut is initially tightened up by hand so as to cause no stress in the
copper spacing tube calculate the axial stresses in the copper and the
 STATICALLY INDETERMINATE STRESS SYSTEMS 99

Fig. 4.12
Spring stiffness = k

10 mm dia.
Fig. 4.13 steel bolt

| Copper spacer
al 12 mm inside dia.

bolt if a spanner is then used to turn the nut through 90°. E, =

100 GN/m?, E, = 208 GN/m’.
a) Figure 4.14 shows a series of bars of modulus £;, cross-sectional area
A,, length L,; which are loaded in parallel by a force F.
The bars are slightly mismatched in length so that each bar has to
be stretched by an amount d, to match its neighbours.
(a) Derive an expression for how the force in a given bar depends on
d and d,;.
(>) By summing up forces on all bars, determine how the final
deflection d depends on the applied total force F.
(c) Set up a spreadsheet for an assembly of 2 bars. Calculate the
stress in each bar once the deflection d has been found.
(d) Use this to solve Problem 4.6. Note that the external load is zero
in this case.

Fig. 4.14
100 MECHANICS OF ENGINEERING MATERIALS
a

4.8 A hydraulic cylinder of 80mm inside diameter and 4mm _ wall

thickness is welded to rigid end plates as shown in Fig. 4.15. The end
plates are tied together by four rods of 8mm diameter symmetrically
arranged around the cylinder. Calculate the stresses in the rods and
the cylinder at the cylinder design pressure of 20 MN/ m’. The
cylinder and tie rods are made from steel with a Poisson’s ratio value
of 0.3.

Fig. 4.15 4 tie rods

a0 A rigid chute 2m in length is supported horizontally, at a height of

(0.33 m above a hopper, by a spring at one end of stiffness 30 kN/m
and a second spring at mid-length of stiffness 20kN/m (Fig. 4.16).
Determine the position on the chute which a component of weight
2.5kN reaches when the unsupported end of the chute just touches
the edge of the hopper.

Fig. 4.16

30 kKN/m

4.10 An elastic packing piece is bolted between a rigid rectangular plate and
a rigid foundation by two bolts pitched 300mm apart and
symmetrically placed on the long centre-line of the plate, which is
450mm long. The tension in each bolt is initially 120kN, the
extension of each bolt is 0.015 mm and the compression of the packing
piece is 0.6mm. If one bolt is further tightened to a tension of 150 kN,
determine the tension in the other bolt.
4.11 A bimetallic temperature-sensitive component consists of a short steel
tube of outside diameter 70mm and inside diameter 60mm,
surrounding a solid copper rod of 50mm diameter. At 20°C the rod
and cylinder have exactly the same length. If a 100 kN load is placed
on top of the rod and cylinder, calculate the forces in the two materials
if the whole assembly is heated to 60°C. Calculate also the temperature
at which the copper would take all the force. E, = 208 GN/m’, E, =
104 GN/m’, a, = 12 x 107° per deg C, a, = 18.5 x 10° per deg C.
4.12 A steel tube of 150 mm internal diameter and 8 mm wall thickness in a
chemical plant is lined internally with a well-fitting copper sleeve of
 STATICALLY INDETERMINATE STRESS SYSTEM'S 101

2mm wall thickness. If the composite tube is initially unstressed,

calculate the circumferential stresses set up, assumed to be uniform
through the wall thickness, in a unit length of each part of the tube
due to an increase in temperature of 100°C. Neglect any temperature
effect in the axial direction. For steel a = 11 x 107° per deg C, E =
208 GN/m’; for copper a = 18 x 10~° per deg C, E = 104 GN/m?.
teal For a hydraulic test a steel tube of 80 mm internal diameter, 2mm wall
thickness and 1.2m in length is fitted with end plugs and filled with
oil at a pressure of 2MN/m/?. Determine the volume of oil leakage
which would cause the pressure to fall to 1.5 MN/m?. Bulk modulus
for the oil = 2.8GN/m’; for the steel E = 208 GN/m?, v = 0.29.
4.14 A drop-weight shearing device consists of a vertical rod with a cross-
sectional area of 125mm? which has an end collar which supports a
spring of stiffness 150 kN/m, as shown in Fig. 4.17. If the shear tool
of 10 kg mass is dropped through a height of 500 mm on to the spring,
calculate (a) the initial instantaneous extension of the rod, (4) the
maximum stress in the rod, and (c) the initial instantaneous
compression of the spring. E = 208 GN/m?.

Fig. 4.17

mdbs) A drop-hammer used for forging metal is illustrated in Fig. 4.18. If

the hammer is dropped through a height of 1 m on to the workpiece,
calculate the resulting compression of the workpiece. Compare the
force transmitted to the foundation for the system shown with that
transmitted if the workpiece were resting on the foundation before the
hammer was dropped 1m on to it. Press: Mass of hammer =
12000kg, mass of anvil = 5000kg, spring stiffness = 15 MN/m.
Workpiece: 30mm dia. x 30mm tall, modulus, E = 208 GN/m?.

Fig. 4.18

Hammer

Guide
im

Workpiece
30 mm
Anvil
Spring
[
|—
CHAPTER
Torsion

One of the common engineering modes of deformation is that of torsion, in which a

solid or tubular member is subjected to torque about its longitudinal axis resulting in
twisting deformation. A design analysis is required in order to estimate the shear-
stress distribution and angular twist for solid and hollow shafts of circular cross-
section and thin-walled closed and open non-circular cross-sections. Engineering
examples of the above are obtained in shafts transmitting power in machinery and
transport, structural members in aeroplanes, springs, etc.

5.1 Torsion of a thin-

The thin cylinder of mean radius r, thickness / and length L shown in Fig.
walled cylinder
5.1 is subjected to an axial torque 7 at each end which causes the cylinder to
twist about its longitudinal axis. In Chapter 2 it was shown that as a statically
determinate problem only circumferential uniform shear stress in the wall of
the cylinder was set up as a reaction to the torque 7.

Fig. 5.1

Equilibrium Referring to Fig. 5.1, the shear stress 7,9 acting on an element
of wall trd@ gives a shear force
F = jer dd

This will provide a reacting moment about the central axis:

a= igptr’ dé

Hence the total reacting torque will be

20
| twtr’ db
0

which is in equilibrium with the applied torque 7. Therefore

fi Tt’ lr
 TORSION 103

or

T.
Tb Onrtt
[5.1]
It should also be noted that the circumferential shear stress 7,9 is associated
with a complementary shear stress 7%, in the longitudinal direction in the
wall. As there are no other shear stresses present, for simplicity the 24
suffices will be omitted.

Geometry of deformation ‘The rotation of one end of the cylinder relative to

the other through an angle @ results in a change in angle y between a cross-
section and a longitudinal generator on the cylinder as in Fig. 5.1. The angle
7 is the shear strain associated with the shear stress. The displacement of B
to B’ may be expressed both as r@ and y/ and therefore
nh ae)

and

10
sory [5.2]

Stress—strain relationship The stress-strain relationship in shear is

expressed as

aC [5.3]
From egns. [5.1], [5.2] and [5.3] we can define the interrelationships for the
cylinder as

[5.4]

5.2 Torsion of a solid

In the case of the thin-walled cylinder, in the previous section the shear
circular shaft
stress was assumed to be constant throughout the wall, but in the case of a
solid cylinder the shear stress varies over the cross-section. Firstly we shall
require that:

(a) The shaft is straight and of uniform cross-section over its length.
(b) The torque is constant along the length of the shaft.

Further we note that the longitudinal and transverse symmetry of the shaft
in relation to the applied torque enables the following deductions to be
made:

1. Cross-sections which are plane before twisting remain plane during

twisting.
2. Radial lines remain radial during twisting.
3. Deformation is by rotation of one cross-sectional plane relative to the
next and planes remain normal to the axis of the shaft.
104 MECHANICS OF ENGINEERING MATERIALS

di peor pence Geometry of deformation he cylindrical shaft of length Z and outer radius
stress, strain and angle of ro subjected to torque T may be regarded as being built up of a large number
twist of thin-walled tubes just fitting inside each other. They are all twisted
through the same angle of rotation 9; therefore for any arbitrary tubes of
radius r, and r, experiencing shear strain 7 and 7, from [5.2] we may write
ae VoL = “gl

lp 1G
or

aie constant
Tp Vy
which demonstrates that at the centre of the shaft where r is zero, ¥ is zero,
and that at the surface -y is maximum and the variation is linear. Therefore

ae. [5.5]

Stress—strain relation ‘The shear-stress—shear-strain relation in terms of the

shear modulus is
ie
z
—=G [5.6]
5.6

From eqns. [5.5] and [5.6] we have

gee 5.7
Thus shear stress has a linear distribution across the shaft diameter, being
zero at the centre and maximum at the outer surface, as shown in Fig. 5.2.

Fig. 5.2 T
Internal resisting forces

Tmax

5.4 Relation between

torque and shear stress Referring again to any one of the thin tubes of radius r and thickness Sr on
which the shear stress is T.
 TORSION 105

Equilibrium
Force per unit length On,
Torque per unit length of tube about shaft axis = Tr ér
Resisting torque on whole tube TP OrlLar
Resisting torque for whole cross-section = ie Tar dr
This is equal to the applied torque; therefore

ti | Tin dr [5.8]
0
Using eqn. [5.7] to substitute for 7,

GO. Gé (”
Y | ——2ar dr = + | 2ar? dr [5.9]
0 L L Jo
The integral function

te ae;
| 2nr dr = | dA
2 1,
= Ss for solid shaft)
0
is the polar second moment of area of the section (see Appendix A) denoted
by J. Therefore,

T= 5 or 7" [5.10]

Hence, from eqn. [5.7]

Pag te |
ane
which relates the shear stress to the torque and the geometry of the cross-
section.

5.5 The torsion

Combining eqns. [5.7] and [5.10] gives the fundamental relationship
relationship
between shear stress, torque and geometry:

fe [5.11]
oie eee b

The quantities concerned and their units are

T = Torque, Nm
J = Polar second moment of area, m*
7 = Shear stress, N/m? at radius r, m
G = Shear modulus, N/m?
@ = Angle of twist, radians (not degrees), over length LZ, m

Example 5.1
Calculate the size of shaft which will transmit 40 kW at 2 rev/s. The shear stress is
to be limited to 540 MN/m? and the twist of the shaft is not to exceed 1° for each
2m length of shaft. The shear modulus G is 77 GN/m?.
106 MECHANICS OF ENGINEERING MATERIALS
Deen eee ee enema aac eI

Firstly converting the power to be transmitted into a torque,

40000
= SO SSIN
EFS i
From egn. [5.11]
TY, if
Tmax: = ay a
J ait,3

4 2 31831
— 40.6 x 10° mm?
"0~ “7 x 50 x 10°
r, = 34.4mm_ on a stress basis

Considering the twist criterion, from eqn. [5.10],

rs ZEL
° 7Gé

2 253183 2 3
i «x 77x
10° x1

= 302 x 10*mm?*

r, =41.7mm_ on a twist basis

This is the governing criterion, and therefore the shaft diameter is 83.4 mm.

5.6 Torsion of a hollow

clecular chatt The above analysis for the solid shaft is similarly applicable to the hollow
shaft (thick-walled tube). Thus the torsion relationship equation [5.11] also
expresses the conditions of equilibrium and compatibility for a hollow
circular shaft. However, the radial boundaries are now r = r, and r = rp, the
outer and inner radii respectively, and thus the polar second moment of
area is

=| amrar

(r) — 72)
The shear stress varies linearly from 7r2/7 at the bore to Tr, /7 at the
outer surface, as shown in Fig. 5.3.
The hollow shaft is more efficient in its use of stressed material than the
solid shaft because the core of a solid shaft has relatively low stresses as
compared with the outer layers. However, hollow shafts are not used widely
in practice owing to the cost of machining, unless saving of weight is at a
premium or it is necessary to pass services down the centre of the shaft.
 TORSION 107

Example 5.2
Compare the torque that can be transmitted by a hollow shaft with that of a solid
shaft, as shown in Fig. 5.3, of the same material, weight, length and allowable
stress.

Fig. 5.3

(a) Solid (b) Hollow

Let 7; and rz be the outer and inner radii of the hollow shaft and let r be the
radius of the solid shaft. Then, for the same maximum shear stress 7,
Tet
UD OLLOIP ra r, 2 (7-7)

Lota =

Eliminating 7 we have
a
i ehies “ee aloe [5 12]

reed Ae
Since, for shafts of the same weight, mr? = (rj — 75) per unit length, eqn.
[5.12] can be simplified to
2 2
ee park al ae ' = “al (: =) [5.13]
Dog nmr is nN

where n = 7/12
Now 7? = re — ts, and putting 72 in terms of r|/n gives

PA tS Se
Therefore
2
pete) mero Gta
[5.14]
dai ny/(n? aa 1)

It is common practice to take n = 2, which gives

T hollow- 5 ae
Tie - 2 vis

Thus the hollow shaft can carry 44% greater torque than the solid shaft for
the same weight, etc.
108 MECHANICS OF ENGINEERING MATERIALS

Example 5.3
Compare the torsional stiffness of a thin-walled hollow tube with a solid shaft of the
same material, torque capacity and allowable stress.

For a thin tube, when mean radius r is much less than wall thickness 1,
J = 2nr't. When the two shafts are subjected to equal torque, the maximum
stress will be the same if

Ta 3
2a t= —
js
where a is the radius of the solid shaft. Alternatively

Ci (4772) 1/ ;
The ratio of the stiffnesses will then be

/; Daiiee Tia Bs GF hollow = 2rrt is 4t le ie

k
Oe iow Cui GFolid rat /2 at — \4e
The stiffest tube will be thin walled with the largest possible ratio of wall
thickness to radius. However, when the tube becomes very thin it will fail,
not by exceeding the maximum allowable shear stress in the material, but
because the thin walls collapse by buckling or becoming unstable. A rolled
up sheet of paper is very stiff in torsion when very small loads are applied,
but it can not carry a useful amount of torque without buckling. An
introduction to the analysis of buckling is given in Chapter 10.
It is usual that several possible modes of failure have to be considered in
a design — there is a design goal, perhaps to maximize the shaft stiffness,
subject to constraints such as maximum stress or the prevention of buckling.

5.7 Torsion of non-

In certain shaft arrangements, the complete shaft is continuous but not of
uniform and composite uniform diameter (Fig. 5.4), or the arrangement may be such that one shaft
shafts is hollow with another shaft arranged coaxially (Fig. 5.5). In each case it is
necessary to investigate the conditions of both the torque and the angle of
twist in each part of the system in order to obtain a sufficient number of
equations for the solution.

Continuous shaft having In this case (Fig. 5.4) the total torque 7 is transmitted by each portion of the
two different diameters shaft; thus:

Fig. 5.4 Shafts in series

 TORSION 109

Fig. 5.5 Shafts in parallel

Shaft 1, radius 7,

Shaft 2, radius rp

Equilibrium

C= 1a [5.15]

Geometry of deformation ‘The total deformation @ is due to 0; over length

I, plus @2 over length L2, so that

0 = 0, +0, [5.16]

Stress—Strain Relation Substituting for 7; and 7) in egn. [5.15], from

[5.11]
IPIEA TL,
(ra
i ————— a
i aCrA
Ch ===
pa
Dell7

Hence,

gl +blfal(hb
Os 2)
5.18
ae tr ee)
If the shafts are of different materials then G,; and G2 may be used in the
above analysis.
For a series of shafts in series, we can write
n L;

9 = Dare [5.19]

Concentric shafts In Fig. 5.5 the shafts have a common axis and are joined at the ends so that
the total torque 7 is made up of that carried by the hollow shaft and that
carried by the solid shaft, these being 7) and 7) respectively:

Equilibrium
T=T, + T2 [5.20]

Geometry of deformation Both shafts twist through the same angle @ since
their ends are rigidly connected; hence

0=0) =
110 MECHANICS OF ENGINEERING MATERIALS

Using [5.11]

7 ee
G
It
In [5.20]
Gif Grfé
ims Ty ui Ly

0
P= (Gs + G2 Fr) 52]

This may be used to get the angle of twist as a result of any applied torque,
or vice versa.
The maximum shear stresses in each shaft are

Tr, d Tro
™j|7=— and »=—
Sh eae
Hence

ee [5.22]
Gp r
so that the ratio of the maximum shear stresses is the same as the ratio of the
outer diameters.

Example 5.4
A solid alloy shaft of 50 mm diameter is to be friction welded concentrically to the
end of a hollow steel shaft of the same external diameter (Fig. 5.6). Find the internal
diameter of the steel shaft if the angle of twist per unit length is to be 75% of that of
the alloy shaft.
What is the maximum torque that can be transmitted if the limiting shear
stresses in the alloy and the steel are 50MN/m? and 75MN/m?2 respectively?
Gsteet = 2.2Gatoy-

Equilibrium

Laitey = Tyeei = T [5.23]

Geometry of deformation
0, 6
a [5.24]
 TORSION 111

Hence

Since

T.=T, and G,=2.2G,

I, =2.2 x 0.759,

md* T
at
37 Ay eas
x 0.7555 ( } d°)

50 = (2.250075 50°) 22 « 0.75 <a)

4a0.65 x 504 dp 9:
sO OS) ar ee
The torque that can be carried by the alloy is

nd? nx 50° 6
i ia) ee = ee 10° = 1227Nm

The torque that can be carried by the steel is

PR (50* — 39.6)
x 75 x 10° = 1120 Nm
P16” 50510?
Hence the maximum allowable torque is 1120 Nm.

mple 5.5
eine A composite shaft of circular cross-section 0.5 m long is rigidly fixed at each end, as
shown in Fig. 5.7. A 0.3m length of the shaft is 50 mm in diameter and is made of
bronze to which is joined the remaining 0.2m length of 25mm diameter made of
steel. If the limiting shear stress in the steel is 55 MN/m? determine the maximum
torque that can be applied at the joint. What is then the maximum shear stress in
the bronze? Gstec: = 82GN/m7; Ghronze = 41 GN/m?.

Equilibrium ‘The free-body diagram Fig. 5.7(4) shows that the torque T is
shared between the two parts of the shaft. Therefore this is another

Fig. 5.7
Bronze
112 MECHANICS OF ENGINEERING MATERIALS
ee

arrangement of the concentric shaft problem illustrated in the previous

section.

T=T1,+ 7,
The torque that can be carried by the steel is

2x 55x10 2x 0.6254
— x = 169 Nm
0.025 32

Geometry of deformation ‘The angle of twist must be the same for each part
at the joint; therefore

6, = 0,

and the angle of twist for the steel is given by

169 an
Oe 0.0108 rad
(1/32)
x 0.0254 ~~82 x 10°—
Therefore 9, = 0.0108 rad and so

Fe 41 x 10’ x 0.0108 yas 0.05*

= 906 Nm
a 0.3 32
The total torque that can be applied at the joint 1s
T, + T; = 906 + 169 = 1075 Nm
The maximum shear stress in the bronze is

eowal 10’ x 0.0108

= 25
x Ty = 36.8MN/m 2

5.8 Torsion of a thin tube

The torsion of a solid shaft of non-circular section is a complex problem, but
of non-circular section
in the case of a thin hollow shaft, or tube, a simple theory can be developed
even if the tube thickness is not constant.
The thin-walled tube shown in Fig. 5.8 is assumed to be of constant
cross-section throughout its length. The wall thickness is variable but at any
point is taken to be ¢. Assume the applied torque 7 to act about the
longitudinal axis XX and to introduce shearing stresses over the end of the
tube; these stresses have a direction parallel to that of a tangent to the tube at
a given point. A shearing stress of magnitude 7 at any point in the
circumference has a complementary shear stress of the same magnitude
acting in a longitudinal direction.
Consider the small portion ABCD of the tube and assume that the
shearing stress T is constant throughout the wall thickness ¢. The shearing
force along the thin edge AB is 7? per unit length, and for longitudinal
equilibrium of ABCD this force must be equal to that on the thin edge CD.
Now, since ABCD was an arbitrary choice, it follows that Tt is constant
for all parts of the tube. The value T¢ = ¢ is called the shear flow and is an
internal shearing force per unit length of the circumference of the section of
the thin tube.
 TORSION 113

Fig. 5.8

(a) (b)

The force, dF, acting in a tangential direction on an element of the

perimeter of length ds is 7¢ ds, and if r is the mean radius at the point, then
the moment of this force about the axis XX is T¢r ds and the total torque on
the cross-section of the tube is

pee rn ds [5.25]

The integration extends over the whole circumference. Since Tt = q is

constant, we may write

T= riprds = gr [5.26]

Now, rds is twice the shaded area shown, and ¢rds for the whole
circumference is therefore equal to 24, where A is the area enclosed by the
centre-line of the wall of the tube (shown dotted); therefore

T =2Aq [5.27]
Thus at any point the shearing stress is given by

of 5.28
OA ee
Owing to the variation of shear stress around the circumference of the
tube it is not possible to predict the deformations, and hence the angle of
twist, in the simple manner used for the circular shaft or tube. The angle of
twist 6 can be determined, however, from the strain energy stored in the
tube. Considering an axial strip along which the shear stress is constant,
then from eqn. [3.9] the shear-strain energy per unit volume is
114 MECHANICS OF ENGINEERING MATERIALS

If the strip is of length /, thickness ¢ and width ds, then the energy stored in
the strip is
42 2
U, =1G x It
ed ds

Therefore the energy stored in the complete tube is

i2
(UI, =) ;Jn
1G 5

Substituting for 7 from egn. [5.28],

- T? = T?]. ds
[ds
5.29
Us }sane = gE | ; B22
But the stored energy is equal to the work 5T@, since in the elastic range the
torque is proportional to the angle of twist 0; therefore

Ty) ds
1T9= —
g 842G ;t

so that

oo a

If the tube is of constant thickness ¢ around the circumference s, then

TlIs
44°Gt
or from eqn. [5.28] the angle of twist, 6, may be expressed as a function of
the shear stress, T.

Tls
~ IAG
(5.31]

Example 5.6
The light-alloy stabilizing strut of a high-wing monoplane is 2m long and has the
cross-section shown in Fig. 5.9. Determine the torque that can be sustained and the
angle of twist if the maximum shear stress is limited to 28 MN/m2. G — 27 GN/m?.

Fig. 5.9
 TORSION 115

The area enclosed by the median line of the wall thickness is

A = (n x 25") + (50 x 50) = 4460 mm?

The allowable torque is obtained using the minimum wall thickness:
t= 2A

= De x 4400)
210° yy
x =
2
103 <720.x, 10 ope 500 Nm

The angle of twist is obtained from egn. [5.30]:

as by po
ey em Ge
Therefore

ne 500 x 2 100 ve507

~ 4x 44602 x 10-12 x 27 x 109 | 2 3

= 0.0476 rad = 2.73°

5.9 Torsion of a thin

An approximate solution for the torsion of a strip of rectangular cross-
rectangular strip
section whose thickness is small compared with the width, Fig. 5.10(a), can
be obtained by considering the strip to be built up of a series of thin-walled
concentric tubes which all twist by the same amount. One of these tubes is
shown in Fig. 5.10(4), and the area enclosed by the median line is

A = (b—2h)2h+ ah?
If b is large compared with h then the terms in 4” can be ignored and

Ax 2bh
If the tube is subjected to a torque 67 then, from eqn. [5.28], the shear
stress 1s
_ OT
[5.32]
7 AbEGh

Fig. 5.10
116 MECHANICS OF ENGINEERING MATERIALS

As the tube becomes infinitely thin,

ag
—
dh = 4bhr
4bh

and the torque carried by the strip is

1/2
T= | 4bhr dh [5.33]
0

From eqn. [5.31],

ee ee Ceen [5.34]
jy ee Ll 2b
Substituting in eqn. [5.33] for 7,

= = bt’ — [5.35]

The quantity 45/3 is termed the torsion constant, but it is not the polar
second moment of area for the section.
Equation [5.34] shows that the shear stress parallel to the long edge of
the cross-section is proportional to the distance / from the central axis. The
maximum shear stress occurs at the outer surface and is
tGé
Tmax = Sy. [5.36]

Example 5.7
Determine the angle of twist per unit length and the maximum shear stress in the
aluminium channel section shown in Fig. 5.11 when subjected to a pure torque of
20Nm. Shear modulus = 27 GN/m?2.

Fig. 5.11
 TORSION 117

The channel may be analysed as three rectangular strips, the two flanges and
the web, and the above solution will be used for each part. Let the
proportions of the torque carried by the flanges and web be 7) and 7)
respectively; then from eqn. [5.35],

Tie
lids
NS as
ay 0
1 5am (a) * oi)
AN ee

9
=5.4-
/

7)
= 36.5-
l
But 27; + 7T> = 20; therefore

47.35 = 20

and

Y
he 0.422 rad/m = 24.2°/m
:

The maximum shear stress in the flanges, from eqn. [5.36], is

2
Tmax = 793 X27 X 10° x 0.422 = 22.8 MN/m’?
and in the web is

3
Tmax = 793 * 27 x 10° x 0.422 = 34.2 MN/m?

cena tia Compare the torsional strength and stiffness of a thin-walled tube of circular cross-
section of mean radius R and thickness t, with the values for a similar tube having a
thin slit cut along its full length, as shown in Fig. 5.12.

Fig. 5.12
118 MECHANICS OF ENGINEERING MATERIALS
ee

Closed Open
(a) Strength:

T ei Treating open tube as a thin

OR rectangular strip:

oe ee JOD eae
SR R24 pe se Bye

Open _ 3TrR’t wok

Closed aR?T ¢
(b) Stiffness:

T 2 GJ SGiR: LU igee
ay ee @ 3

T 2G TAG
— =— (rR? — =—(17RP
Fea a ar ot
Open ne me:
Closed 7R3t 3R2
It may be seen from this example that a tube with a longitudinal slit has a
very small torsional stiffness compared with that of a complete tube.

5.10 Effect of warping

; : In the previous section and the above example dealing with a thin-walled
during torsion
open section subjected to torsion, the pure torque was supposed to be
applied to each end of the member in such a way that there was no axial
restraint. Owing to the variation in transverse shear stress, for example in
the flanges of the above channel or the I-section shown in Fig. 5.13, there is
also a variation in longitudinal complementary shear stress which results in
the axial movement of one flange with respect to the other. Therefore, cross-
sections which were initially plane do not remain so during torsion and there
is warping of any cross-section. If one or more sections of a member are
constrained in some manner to remain plane during torsion then warping is
restrained.
Fig. 5.13
Warping constrained
ere

¥ |
 TORSION 119

The warping constraint has most significance near the ends of the
section. It decays with distance from the constraint so that after several
section depths the influence of warping is negligible.
Resisting torque in the section is supplied in two ways, by simple torsion
and also by torque set up through the restraint of warping. Thus an applied
torque will cause a smaller angle of twist than when the section is free to
warp, and torsional stiffness may be considerably increased if warping is
restrained. Non-circular closed and solid-section members also exhibit
warping, but the effect is much smaller in comparison with the open section.
Further discussion of warping behaviour can be found in books such as that
by Megson!.

5.11 Torsion of solid

The solution in section 5.9 is only applicable for a rectangular section in
rectangular and square which the longer side is much greater than the other. Unfortunately the
cross-sections solution for torsion of general rectangular and square sections, although of
some engineering interest, is very complex and beyond the scope of this
text. (The reader who is interested in the full solution should consult a text
such as Theory of Elasticity by Timoshenko and Goodier’.) The result in
terms of torque, maximum shear stress at the centre of the long side and
angle of twist may be expressed as

T = abt? Tmax = BbtG 8 [5.37]

l
where b is the longer and ¢ the shorter side, and a, (@ are factors dependent
on the geometry, as given in Table 5.1. It may be seen that as b/t becomes
very large, G — } as shown in eqn. [5.35].

Table 5.1
b/t 1 1S 2 25 3 at 6 10 on)

a O20 300223) 0)24.6m 0256 022.67. 02820299 O12 0333S

B ON ON9 One 0229S 02490263 0230299 SOS 12a OSS

5.12 Summary
The analysis of the thin-walled circular tube highlights the basic steps of
equilibrium of internal and external forces, the geometry of deformation of
the tube and the elastic shear-stress/shear-strain relationship for the
material. A full appreciation of the foregoing makes the understanding of the
solution for the solid and hollow circular shafts relatively simple, leading to
the basic relationship

foe
oer

from which we see that shear stress and strain are linearly distributed across
the section, being maximum at the outer surface. In the case of composite
shafts the additional assessed information to the above is the manner in
which the applied torque or the angle of twist is shared between different
components.
Non-circular thin tubes and other open sections play an important part
in engineering, and, whereas the stresses can be determined from the
120 MECHANICS OF ENGINEERING MATERIALS

equilibrium state, we have to evaluate the shear-strain energy to determine

the amount of twist.
Finally the effect of warping of the cross-section and its restraint have a
considerable effect on the stiffness of short, non-circular torsion members
and this must be taken into account.

References
Vi Megson, T. H. G. (1990) Aircraft Structures for Engineering Students,
2nd edition, Edward Arnold, London.
ys Timoshenko, S. P. and Goodier, J. N. (1970) Theory of Elasticity 3rd
edition, McGraw-Hill, New York.

Problems bil A hollow steel shaft with an external diameter of 150 mm is required
to transmit 1MW at 300rev/min. Calculate a suitable internal
diameter for the shaft if its shear stress is not to exceed 70 MN/m’.
Compare the torque-carrying capacity of this shaft with a solid
steel shaft having the same weight per unit length and limiting shear
stress. G = 80 GN/m?.
52 Given the applied torque on a shaft, its length and the shear modulus
of the material from which it is made, develop a spreadsheet to
calculate:
(a) the angle of twist of the shaft
(6) the maximum shear stress in the shaft
for
(:) a hollow circular tube
(7) a thin rectangular strip.
ees) Two shafts AB and CD are coupled by gear wheels as shown in Fig.
5.14. If the pitch circle diameter (PCD) of the gear wheel B is 180 mm
and the PCD of gear wheel C is 60 mm, calculate the angle of twist of
D relative to A when a torque of 3kNm is applied at D. The shear
modulus of the shaft material is 80 GN/m7’. Shaft AB is 0.15 m long
and 30mm diameter whilst shaft CD is 0.2m long and 25mm
diameter.

Fig. 5.14

5.4 The gearbox in Fig. 5.15 is required to supply an output torque of

300 Nm at a shaft speed of 100 rev/min. The gear ratios are such that
shaft 1 rotates at three times the speed of shaft 2, and shaft 2 rotates at
five times the speed of shaft 3. Calculate the input power required
from the motor and the diameters of shafts 2 and 3 if the allowable
 TORSION 121

Fig. 5.15
Output shaft 3

Gear box
Drive shaft 1

shear stress of the shaft material is 400 MN/m?. Neglect losses in the
system and assume a safety factor of 2.
5.5 A steel shaft has to transmit 1MW at 240rev/min so that the
maximum shear stress does not exceed 55 MN/m/? and there is not
more than 2° twist on a length of 30 diameters. Determine the
required diameter of shaft. G = 80 GN/m’.
5.6 A shaft carries five pulleys, A, B, C, D and E, and details of shaft
diameters, lengths and pulley torques are given in Table 5.2.
Determine in which sections the maximum shear stress and angle of
twist occur. G = 80 GN/m’.

Table 5.2
Torque Length Diameter
(Nm) Direction Shaft (m) (mm)

60 Clockwise AB 1 38
900 Anticlockwise BC 100
300 Clockwise CD 75
640 Clockwise DE —
|
RN
NIN)
7S
100 Anticlockwise

A hollow steel drive shaft with an outside diameter of 50mm and an

inside diameter of 40mm is fastened to a coupling using using six
8 mm diameter bolts on a pitch circle diameter of 150 mm. If the shear
stress in the shaft or bolt material is not to exceed 170 MN/m?
calculate the maximum power which the system could transmit at a
rotational speed of 100 rev/min. What would be the effect of only
using three of the bolts?
5.8 A torsional vibration damper consists of a hollow steel shaft fixed at
one end and to the other end is attached a solid circular steel shaft
which passes concentrically along the inside of the hollow shaft, as
shown in Fig. 5.16. Determine the maximum torque 7 that can be
applied to the free end of the solid shaft so that the angle of twist
where the torque is applied does not exceed 5°. Local effects where
the two parts are connected may be ignored. Shear modulus, G =
80 GN/m?.
519 Two shafts are connected end-to-end by means of a coupling in which
there are twelve bolts on a pitch circle diameter of 250mm. The
122 MECHANICS OF ENGINEERING MATERIALS

Fig. 5.16 600 mm

maximum shear stress is limited to 36 MN/m*? in the shafts and

16 MN/m in the bolts. If one shaft is solid, 50 mm diameter, and the
other hollow, 60 mm external diameter, calculate the internal diameter
of the latter and the bolt diameter so that both shafts and coupling are
equally strong.
5.10 A compound drive shaft consists of a solid circular steel bar B
surrounded for part of its length by an aluminium tube A as shown in
Fig. 5.17. The contacting surfaces are smooth and the collar and the
shaft are both rigidly fixed to a machine at D. Pin C fills a hole drilled
completely through a diameter of the collar and shaft. The shearing
deformation in the pin and the bearing deformation between the pin
and shaft can be neglected. Calculate the maximum torque 7 which
can be applied to the steel shaft as shown, without exceeding an
average shearing stress of 8MN/m/ on the cross sectional area of the
pin at C, the interface between the shaft and collar. G,.; = 80 GN/
m’; Crap = 28 GN/m’.

Fig. 5.17

Sali A hollow rectangular section tube with the dimensions shown in Fig.
5.18 is subjected to a torque of 20 Nm. Calculate the shear stresses in
the walls of the tube and the angle of twist per unit length. G =
27 GN/m?. Compare the results with those of the similar open section
in Example 5.7.
512 An aluminium-alloy strut for a light aircraft having the cross-section
shown in Fig. 5.19 is 3m in length. If the shear stress is not to exceed
30 MN/m? and the applied torque is 134Nm, determine the
required thickness ¢ of metal. What is the angle of twist? G =
28 GN/m’.
A torsional member used for stirring a chemical process is made of a
circular tube to which are welded four rectangular strips as shown in
 TORSION 123

Fig. 5.18

Fig. 5.19

Fig. 5.20

Fig. 5.20. The tube has inner and outer diameters of 94mm and
100 mm respectively, each strip is 50 mm by 18mm, and the stirrer is
3 m in length. If the maximum shearing stress in any part of the cross-
section is limited to 56 MN/m/’, neglecting any stress concentration,
calculate the maximum torque which can be carried by the stirrer and
the resulting angle of twist over the full length. a = 0.264, G = 0.258,
G= 83GN/m.
arl4 An I-beam has a width of 100 mm and depth of 150 mm, as shown in
Fig. 5.21. Calculate the maximum pure torque which could be applied
to this beam if the yield shear stress is 240 MN/m/?. Assume a safety
factor of 3.
DLS Using the spreadsheet formulae developed in Problem 5.2, sum the
contributions from the individual thin rectangular strips to derive the
torsional stiffness of an open section composed of 3 strips. From this
124 MECHANICS OF ENGINEERING MATERIALS

Fig. 5.21

calculate the angle of twist for a shaft of a given length and the
maximum shear stress in each strip. Use the resulting spreadsheet to
check the answers for the channel section of Example 5.7 and the I-
beam of Problem 5.14.
5.16 Using the spreadsheet formulae developed in Problem 5.2, sum the
total angle of twist for two shafts in series, given the shear modulus G,
section polar moment7 and length ZL. Calculate the maximum shear
stress in each bar.
Hint: the spreadsheet will be very similar to that of Problem 4.4 for
the extension of a series of bars.
(a) Use the resulting spreadsheet to confirm the results of Example
5.4. In other words, plug in the values obtained for the shaft
diameters and torques, then check that the twist of the steel shaft
is 75% of that in the alloy shaft and that the stress constraints are
satisfied.
(b) Use the spreadsheet to check the results of Example 5.5. Plug in
the torques obtained for each shaft and confirm that the total
twist is zero and that the maximum shear stresses are correct.
 CHAPTER
Bending: Stress

One of the most common modes of deformation of engineering structures and

components is that of bending. The simplest form is the slender member, often
termed a beam, subjected to transverse loading, which is the subject of this chapter.
In Chapter 16 the more complex states of bending in plates and shells are studied.
Beams are designed both for strength and deflection, and this chapter is
concerned with the former. The stresses that are set up in bending are tension and
compression along the length and shear across the section of the beam. These
stresses derive directly from the internal reactions of bending moment and shear
force introduced in Chapter 1 in the applications of equilibrium. The importance of
being able to establish shear-force and bending-moment distributions cannot be
overemphasized, since the correct determination of a design based on allowable
stresses and deflection depends upon that first step. Because of this, the chapter
commences with a number of illustrations of the method of calculating shear-force
and bending-moment distributions. This is then followed by analyses of bending
stresses and shear stresses for various beam configurations.

6.1 Shear-force and

bending-moment The reader should refer back to Chapter 1, if necessary, to revise the basic
notation, sign conventions and equilibrium principles for establishing shear-
distributions
force and bending-moment distributions. The following worked examples
should help to reinforce the basic understanding of the procedure.

Simply supported beam Referring to the loading shown in Fig. 6.1, the left-hand reaction is first
carrying concentrated required and as the beam is statically determinate the reactions can be found
loads from the equations of force and moment equilibrium.
—R4—Rp+W+W,=

—R4L+W,(L—a)+W2(L —b) =0

from which

W(L—a) + W(L—})
Ry= 7
The two concentrated loads cause mathematical discontinuities and so
separate free-body diagrams and equilibrium equations have to be expressed
for each different section, as shown in Fig. 6.1 and in the following
equations respectively:

Shear force
AB Or ya +Q—R,=0 O= Ry

BC RD +O —R,+M,=0

Oe LG
126 MECHANICS OF ENGINEERING MATERIALS

Fig. 6.1

Fig. 6.2 Q

Shear-force diagram

- |)h—lem
Bending-moment diagram

CD be ea +0—-—R4+W+h,=0

Q=R4,-W,- Wy,

Bending moment

AB DV < VeK M— Ryx =0 M = Ryx

BC a<n<b M — Ryx + Wi(x —a) =0

M = Ryx — W(x
— a)

CD eg fae M — Ryx + W(x — a) + Wo(« — b) =0

Vie R4x =a Wi (x a a) = Wr(x a b)

 BENDING: STRESS 127

The diagrams resulting from the above equations are shown in Fig. 6.2. It
should be noted that
(a) The S.F. diagram changes in value at a support or concentrated load by
the amount of the reaction or load.
(6) The B.M. is ahvays zero at the ends of a beam which are either
unsupported or on a simple support.

Simply supported beam Apart from transverse loads, bending can be caused by a couple applied to
with an applied couple the beam at a cross-section, as shown in Fig. 6.3. The reactions at the
supports are obtained from moment equilibrium of the whole beam, from
which

RyL = RzL — M or

Shear force The shear force is constant along the length of the beam and
of value Q = —M/L.

Fig. 6.3

Bending moment

Mx
Forx=a: M-E Roux — 0; “hence Voters

Ee _ Mx
Forx >a: M+Ry4x—M=0; hence M=M———

M
= ee)
The S.F. and B.M. diagrams shown in Fig. 6.3 were derived from these
equations.
128 MECHANICS OF ENGINEERING MATERIALS

Example 6.1 Sketch a B.M. diagram for the curved member shown in Fig. 6.4(a), giving the
principal numerical values.

Fig. 6.4 1000 Nm %

mS
| \ MF

ae B Q
Free-body
y diagram of AB

(b)

The determination of bending-moment distribution along a slender curved

member follows exactly the same procedure as for straight beams.
It is more convenient to resolve the applied load into vertical and
horizontal components. These are

V = 500 cos 30° = 250,/3 and H = 500 sin 30° = 250

A portion of the bar is cut off to give a free body and the forces and
moment are inserted as shown in Fig. 6.4(d).

Bending moment ‘Taking moments about B

M — 500(1 — cos 6) + 500,/3 sin 6 =0

At
Cat) M=0

= 1/2 M = —365 N/m

C= M = +1000 N/m

Also, if M = 0 = 500(1 — cos 6— ,/3 sin 6), then

cos 9 = —} and C120;

which gives a point of contraflexure, where the bending moment changes

sign. It may be seen that this coincides with the point at which the line of
action of the applied force intersects the beam.
When dM/d0 = 0,
sin 9— ,/3 cos 0=0 0= 60°

May = 500(1 : v3) = 500 Nm

 BENDING: STRESS 129

The B.M. diagram is plotted for convenience along the vertical axis as
shown in Fig. 6.4(c).

Example 6.2 3 ? aes

Sketch the S.F. and B.M. diagrams for the cantilever shown in Fig. 6.5.

Fig. 6.5 5000 N 3000 N

|2000 n/m |
[BREST

ems oom 7

5000 N 3000 N

A BY 2000 N/m aye

66000 Nm

Even though there is no loading on AB and therefore no shear force or

bending moment, the distance x will still be measured from A rather than B.

Shear force
OF C0

2<x<5 QO =—5000
— 2000(x — 2)
Oe fas) Q = --5000 — 6000 — 3000 = —14000N

Bending moment
<< ke Vie)

2<x<5 —2) (x 2)
M=-—5000(x —2) — 2000(x
2

= —1000(% — 2)(x+3)Nm
(parabolic distribution)

5<x<8 M=—5000(x— 2) — 6000(x — 34) — 3000(x — 5)

= —14000x + 46000 Nm _ (linear distribution)

The diagrams and principal values are shown in Fig. 6.5.

130 MECHANICS OF ENGINEERING MATERIALS

6.2 Relationships ~The beam shown in Fig. 6.6(a) carries distributed loading which varies in an
between loading, shear arbitrary manner. Consider the free body, Fig. 6.6(4), of a small slice of
force and bending moment length dx for which the loading may be regarded as uniform, ».

Fig. 6.6 Loading w= f(x)

Vertical equilibrium

—QO +mdx+ (2+ Par) ='()

dQ
he
yd:
Re 8 [6.1]

From eqn. [6.1] it follows that, between any two sections denoted by 1
and 2,
2 2
|dQ =| —w dx
1 1

or

2
Q, -—Q) = |—w dx [6.2]
1
Thus the change in shear force between any two cross-sections may be
obtained from the area under the load distribution curve between those
sections. The value of the shear force at point 2 is given by
2
=O
~| wax 1

The use of this equation is illustrated in Example 6.3.

 BENDING: STRESS 131

Moment equilibrium Taking moments about the mid-point of the right-

hand edge,

MS Odetto da
d 8 Mee
dM de) = 0
y dx

Neglecting the term multiplied by (dx)’,

dM
= Q Je 0
dx

dM
= ae [6.3]

When Q = 0, dM/dx= 0, i.e. the S.F. is zero when the slope of a B.M.
diagram is zero and not simply when M is a maximum, although this also is
implied. It is possible to have M = Ming, when dM/dx ¥ 0 (see Fig. 6.4).
From eqn. [6.3], considering two sections, | and 2,
2 y
|dM = |QO dx
1 1

or
2
M, —M, = |Chal
l
Thus the change in bending moment between any two sections is found from
the area under the shear-force diagram between those sections. The value of
the bending moment at point 2 is given by
2

Me AG |Oar 6.4]
I!

The use of this equation is illustrated in the following example.

Example 6.3
Determine the position of zero shear force and the value of the maximum bending
moment for the simply supported beam shown in Fig. 6.7. It carries a distributed
load which varies linearly in intensity from zero at the left to w at the right-hand end.
Plot the S.F. and B.M. diagrams.

Fig. 6.7
132 MECHANICS OF ENGINEERING MATERIALS

By similar triangles, the load intensity at x from the left end is (w/L)x.
From eqn. [6.2] the shear force is given by

Shear force

0 = - | dr = @ -F
2
DX WX

Bending moment From eqn. [6.4], and substituting for Q,

wx?
M = M+ |Q dx = My + Qpx— 7

Now, M = 0 when x = 0 and L. Hence

Ib,
Mo=0 and OQ) = =

Thus

mn wxt wh
= DE ae G6

Since dM/dx = O = 0 at x = L/,/3 is a possible solution for Minax,

w (L ivewL Lb wl?
Mian ae
6L 3 6 4/3 | 973
The S.F. and B.M. diagrams are as shown in Fig. 6.7.

6.3 Stress distribution in

pure bending The determination of stress distributions in pure bending is a statically
Assumptions indeterminate problem, and hence all three basic principles stated in the
introduction in Chapter 4 will need to be employed. However, the steps will
be considered in a different order since at this stage there is insufficient
information about the stress distribution to enable an equilibrium equation
to be formulated. The approach initially will be to consider a prismatic beam
of symmetrical cross-section subjected to pure bending, from which the
geometry of deformtion can be studied and the strain distribution determined.
The stress-strain relations will give the stress distribution which can be
related to forces and moments through an equilibrium condition.
Before commencing the analysis it is necessary to make some
assumptions and these are as follows:
1. ‘Transverse sections of the beam which are plane before bending will
remain plane during bending.
2. From consideration of symmetry during bending, transverse sections
will be perpendicular to circular arcs having a common centre of
curvature.
 BENDING: STRESS 133

3. The radius of curvature of the beam during bending is large compared

with the transverse dimensions.
4. Longitudinal elements of the beam are subjected only to simple tension
or compression, and there is no lateral stress.
5. Young’s modulus for the beam material has the same value in tension
and compression.

6.4 Deformations in pure

bending Longitudinal deformation
Consideration of the beam subjected to pure bending shown in Fig. 6.8
indicates that the lower surface stretches and is therefore in tension and the
upper surface shortens and thus is in compression. Hence there must be an

(a)

Beam cross-section

Neutral
plane

Ymax

~~ wg

(b) (c)

xz-plane in between in which longitudinal deformation is zero. This is

termed the neutral plane, and a transverse axis lying in the neutral plane is
the neutral axis. Consider the deformations between two sections AC and
BD, a distance dv apart, of an initially straight beam. A longitudinal fibre EF
at distance y below the neutral axis will have initially the same length as the
fibre GH at the neutral axis. During bending EF stretches to become E’F’,
but GH, being at the neutral axis, is unstrained when it becomes G/H’.
Therefore, if R is the radius of curvature of G’H’,
GH =GH= dy ek dé

EF’ = (R+y) dd
134 MECHANICS OF ENGINEERING MATERIALS

and the longitudinal strain in fibre E/F’ is

oe
Ex
ac) IEF
But EF= GH = G'H’ = R dO; therefore
(R + y)d@
— Rdé
sl Rdé
Hence

2 = wy
oe [6.5]
6.5

and since R = dx/d6, therefore also

dé
n= =y—
Ia. [6.6]
6.6

From eqn. [6.5] it will be seen that strain is distributed linearly across the
section, being zero at the neutral surface and having maximum values at the
outer surfaces. It is important to note here that eqn. [6.5] is entirely
independent of the type of material, whether it is in an elastic or plastic state
and linear or non-linear in stress and strain.

Transverse deformation
Regarding deformations in the y- and z-directions, it is apparent that
changes in length of the beam will result in changes in the transverse
dimensions. For example, the fibres in compression will be associated with
an increase in thickness, whereas the region in tension will show a decrease
in beam thickness. The transverse strains will be
VO».
c=
fy =
E
and a beam of initially rectangular cross-section will take up the shape
shown in Fig. 6.9. This can be easily demonstrated by bending an eraser.
The neutral surface, instead of being plane, will be curved. This behaviour
is termed anticlastic curvature. The deformations are extremely small and do
not affect the solution for longitudinal strains.

6.5 Stress—strain
relationship Consider the segmental length of beam illustrated in Fig. 6.10 the cross-
section of which, for simplicity at this stage of analysis, has symmetry about
the vertical axis y, but not about a horizontal axis z.
As it has been assumed that o, =o, =0 then the stress-strain
relationship that is applicable for /inear-elastic bending is
Ox

< J
€& ===
dike Wee
 BENDING: STRESS 135

Fig. 6.10
Neutral plane

Applied
ie =
moment, M
Resisting
moment, M M

(a)

Applied moment, M

y
(b) dA

or

@, E
=F
yy I
[6.7]
Thus bending stress is also distributed in a linear manner over the cross-
section, being zero where y is zero, that is at the neutral plane, and being
maximum tension and compression at the two outer surfaces where y is
maximum, as shown in Fig. 6.10(a).

6.6 Equilibrium of forces

and moments Consider an element of area, dA, at a distance y from some arbitrary location
of the neutral surface, Fig. 6.10(0).
Position of the neutral The force on the element in the x-direction is
plane and axis
di On CA

Therefore the total longitudinal force on the cross-section is

fy, = | a, dA
A
where A is the total area of the section.
Since there is no external axial force in pure bending, the internal force
resultant must be zero; therefore

iia oO, dA —)0

A
Using eqn. [6.7] to substitute for o,,
E
= Cl= (0)
aI,
136 MECHANICS OF ENGINEERING MATERIALS

Since E/R is not zero, the integral must be zero, and as this is the first
moment of area about the neutral axis, it is evident that the centroid of the
section must coincide with the neutral axis. (The first moment of area of a
section about its centroid is zero; see Appendix A.)

Internal resisting Returning to the element in Fig. 6.10(4), the moment of the axial force about
moment the neutral surface is yd/’,. Therefore the total internal resisting moment 1s

|ia = |yo, dA
A A

This must balance the external applied moment M™, so that for equilibrium

|Ve (6.8)
A

or, substituting for o,,

E
m=3| y’ dA
Ra
Now WVy? dA is the second moment of area (see Appendix A) of the cross-
section about the neutral axis and will be denoted by /. Thus

Nieyey BE1 «ape

M
6.9
eae Wea: ae
Using eqn. [6.7]
Mo;
I Page
and hence 0, = My/I which relates the stress to the moment and the
geometry of the beam.

6.7 The bending

Combining eqns. [6.7] and [6.9] gives the fundamental relationship between
relationship
bending stress, moment and geometry:

M oan
[6.10]
a mo|

The quantities concerned and their units are

M = bending moment, Nm
4
I = second moment of area, m
o, = stress, N/m?
y = distance from neutral axis to point in question, m
E = Young’s modulus, N/m?
R = radius of curvature of neutral axis, m

6.8 The general case of

The bending relationship, eqn. [6.10], is only an exact solution for the case
bending
of pure bending; however, in practice many beam problems involve bending
moment and shearing force which vary along the length. In these cases it has
 BENDING: STRESS 137

been shown that eqn. [6.10], even if not exact, provides a solution which is
quite accurate for engineering design, except for cross-sections at support
points and concentrated loads where stress concentration may occur (see
Chapter 12).

6.9 Section modulus

Using eqn. [6.10], for the outer surfaces of the beam, the maximum stresses
will be

My, max M
Ofna a
ae
Mycmax M
0 cman — i hee eeZ, [6.11]
6.11

where the subscripts denote tension and compression. The quantities

T/Yimax 200 I/)¢max are functions of geometry only; they are termed the
section moduli and are denoted by Z, and Z,.
Values of / and Z for standard sections are given in manufacturers’
handbooks.

Example 6.4
Table 6.1 shows an extract from a steel designer’s handbook giving some of the
section properties of universal beams. Select a section to carry a 20 kN load which
is cantilevered out 2 m from the support. The structural steel used has a yield stress
of 250 MN/m2, but the allowable bending stress is usually taken as 0.66 x 250 —
165 MN/m2.

The maximum bending moment occurs at the support and is

M = 20 x 2 = 40kNm. The bending stress due to this load is
MD M
Co = ——
= = Zz < allowable
IB (AS,
where Z is the section modulus of the beam. Note that in handbooks this
will often be called the elastic modulus. This is nothing to do with the
Young’s modulus of the material; it is a geometric property only and relates
stress and internal moment during elastic, or recoverable, bending.

Table 6.1
Second
moment Section
Designation Depth Width of area modulus
Mass of of Thickness Area for for
Serial per section section Web Flange of axis axis
size metre D B t if section X—X X-X
(mm) (kg) (mm) (mm) (mm) (mm) (cm’) (cm*) (cm’)
254 x 102 28 260.4 102.1 6.4 100 S07 4004 307.6
25 PASTA AKIN Caaik 84 32.1 3404 264.9
AVE) SEB) NO ZO See Soeon O15 Don S0) 2880 DUS
25 AVE BBP Bs) Wess BVA 2348 231.1
138 MECHANICS OF ENGINEERING MATERIALS

Given the applied moment and the allowable stress, the necessary section
can be found by rearranging eqn. [6.11] as

M 40 x 10°
Li = Sh 0 i
lG5e< 10° a
allowable

ie. Z> 242 cm?

From the table, a 203133 section of 30kg/m has a section modulus of
278 cm?, which should be sufficient.
Increasing the length of the cantilever, for example to allow more
clearance beneath it, will cause a proportional increase in bending moment
and stress on a given section. Bending moment also increases linearly with
applied load.
An additional property, the plastic modulus of the section, is often given.
This is used to relate the material yield stress and the largest, or ultimate,
moment the section can carry when permanent deformation is allowed.
Permanent, or plastic, deformation is covered in Chapter 15.

Example 6.5
AT-section bar has dimensions as shown in Fig. 6.11(b). The bar is used as a simply
supported beam of span 1.5 m, the flange being horizontal as shown in Fig. 6.11(a).
Calculate the uniformly distributed load which can be applied if the maximum
tensile stress in the material is not to exceed 100 MN/m2. What is then the greatest
bending stress in the flange?

, 150 mm Compression
Fig. 6.11 31.30
A . i raeerene tho
z y Be ix % GASES ——— NS

eT aos 0. 100
10mm Tension
(b) (c)

The first step is to find the position of the centre of area, which will also give
the neutral axis. Taking first moments of area about the top surface,
(240 x 10)y = (1500 x 5) + (900 x 55)
y = 23.8mm

To find the second moment of area we can use the parallel axes theorem
(see Appendix A):

150 x 10°
ae + (150 x 10 x 18.82)

10 x 908
+—>5— + (10 x 90 x 31.2%) = 2028 x 10’ mm*

Mie = aeIES? sien

8
 BENDING: STRESS : 139

The maximum tensile stress will occur on the bottom edge of the web where
Y =Jmax = 76.2mm. Therefore, using the allowable tensile stress of
100 MN/m’,
100 x 10° 0.281»
0.0762 2028 x 10-9
2 we
= 9.5kN/m
” = 0.0762
x 0.281
Since stress is proportional to distance from the neutral axis, as shown in
Fig. 6.11(0).
Oc; OO;
ee VG

Therefore the greatest compressive stress in the flange is

23
o, = 100 x 10° x — = 31.3 MN/m

6.10 Beams made of

In some circumstances it may be necessary or desirable to construct a beam
dissimilar materials
such that the cross-section contains two different materials. Usually the
object is for one material to act as a reinforcement to the other material.
There would be a number of reasons (cost, weight, size, etc.) why the whole
beam could not be made from the stronger material. The positioning of the
reinforcement material might not be symmetrical with respect to the
centroid of the cross-section and it could be embedded within or fixed in
some manner to the outside of the main bulk material.
The arguments which were applied to the analysis of simple bending of a
homogeneous beam also apply to the composite beam since the two materials
constrain each other to deform in the same manner, e.g. to an arc of a circle
for pure bending.

Symmetrical sections Consider a beam cross-section consisting of a central part of, say, plastic or
timber with reinforcing plates firmly bonded (no sliding) to the upper and
lower surfaces along the length of the beam as shown in Fig. 6.12. The
section is symmetrical about the centroid and neutral surface.

Fig. 6.12

Material m

Material r
140 MECHANICS OF ENGINEERING MATERIALS

Equilibrium
(a) Since there is no external end load, longitudinal equilibrium requires
that

| Om babs =f | Oy dA, =) [6.12]

Am JA,

where the subscripts m and r refer to the main and reinforcing

materials, respectively, as shown in Fig. 6.12.
(b) The sum of the internal resisting moments must be in equilibrium with
the external applied moment

QM,+M,=M

Hence, using eqn. [6.8] for each of the materials,

| Onn dAm + | C7, dAn—= M [6.13]

An A,

The above equations are independent of whether the material is in an

elastic or a plastic condition.
Geometry of deformation Both materials deform to the same circular arc
and, for a linear strain distribution, from eqn. [6.5]

ls Es SES
=== 6.14
Keay ys: ete
These relationships are purely a function of geometry and therefore are
independent of the material and its properties.

Stress—strain relations For linear elastic materials under uniaxial stress

Om = f ees

On hee [6.15]

From eqns. [6.14] and [6.15], Om = jmEm/R and o, = y,E,/R. Substitution

in the equilibrium equation [6.13] gives

(i) | ome (B)

E ; E,
— Ve ee ee ? d4,=
[arena
where the integrals are the second moments of area of the main and
reinforcing materials, /,, and J, respectively, about the neutral surface.
Therefore

Endn + EL,
ee eee Mh
R

Substituting for 1/R gives

On OF M
Ee Ey) Eee
 BENDING: STRESS 141

or

On = >
ME nym
i
Siar
ME,y,
gee alas 616
Spa eae a
Example 6.6
A timber beam of depth 100mm and width 50mm is to be reinforced with steel
plates on each side, as shown in Fig. 6.13. The composite section will be subjected
to a maximum bending moment of 6 kN/m. If the maximum stress in the timber is
not to exceed 12 MN/m?, calculate the required thickness for the steel plates, which
are 100 mm in depth. What is the maximum stress in the steel? E.,..) = 205 GN/m?,
Esimber = 15 GN/m?.

50 x 1003
Taber = —— = 4.17 x 10°mm*

[ee = x 1003 = 0.0833 x 10°¢mm*4

Fig. 6.13

From eqn. [6.16] for limiting stress in the timber, at the top or bottom
surface,

6000 x 0.05 x 15 x 10? x 10”

12 x 10° =
i: (15 x 109 x 4.17 x 106) + (205 x 10° x 0.0833 x 106)

from which ¢ =18.3 mm; and since this is the total thickness of steel, the
plates are each 9.15 mm thick.
The maximum stress in steel plates also occurs at the top and bottom
edges,

6000 x 0.05 x 205 x 10° x 10!

of =
(15 x 10° x 4.17 x 10°) + (205 x 109 x 0.0833 x 10° x 18.3)

— 164 MN/m?
142 MECHANICS OF ENGINEERING MATERIALS

Equivalent sections In beam sections which are symmetrical geometrically, but unsymmetrical
with respect to location of the different materials, as, for example, in Fig.
6.14(a), the neutral axis no longer coincides with the centroid of the section.
The problem can still be solved using eqns. [6.12]-{6.15], but the arithmetic
is more tedious since the neutral axis has first to be found.

Fig. 6.14 Stress distribution

oes. Neutral axis

by

A more convenient approach, which is also valid, is to transform the

composite section into an equivalent (from the view of resisting forces and
moments) section of only one of the two (or more) materials. The solution is
then of course a simple routine.
Consider the beam section shown in Fig. 6.14(a), in which the unknown
neutral axis has been placed at a distance h from the interface between the
two materials. Assuming no sliding, the strain at the interface must be the
same for each material; therefore, at the interface,
oO} 02
[6.17]
Ee
Also, rearranging eqns. [6.16],

O71 1) 02 Fj
MS A (n
|i +3 ) = —2 (n+
2) (= ? (6.
6.18]

where /; and /) are the respective second moments of area about the neutral
axis. Now,

Ey Ey (bod3 _9
h= 2+ bodrys | = 0 (2d + hy
ieee (ay a ee
where b) = (£2/E)b2, so exactly the same resisting moment will exist if
material 2 is replaced by material 1 having the same depth d) but a new
width of (E£)/£)by. This forms the equivalent section shown in Fig.
6.15(a), which is entirely made of material 1 and is shown for the case of

Fig. 6.15 “Equivalent” beam sections (E,>E,) Stress

distribution
bo! = (E,/E5)b,

(a) (b) (c)

 BENDING: STRESS 143

E\ > Ep. In the converse manner, material 1 can be replaced by material 2 if

the width is made (£)/E2)b) to give the equivalent section as in Fig. 6.15(d).
The neutral axis can be found quite simply (from either equivalent section)
as it now coincides with the centroid. The stress distribution in either
equivalent section, Fig. 6.15(c), can be determined and transposed to that
actually occurring as in Fig. 6.14(4) by using egn. [6.17] for the condition at
the interface.

Example 6.7
A timber beam of rectangular section 100mm x 50mm and simply supported at
the ends of a 2m span has a 30mm x 10mm steel strip securely fixed to the top
surface as shown in Fig. 6.16(a) to protect the timber from trolley wheels. When the
trolley is exerting a force at mid-span of 2kN, determine the stress distribution at
that section. Estee) = ZOE you:

30 mm
Fig. 6.16
Inzal

(a) (b) (c)

Firstly consider the equivalent section made entirely of timber as in Fig.

6.16(4). The 30mm width of steel becomes 30 x 20 = 600mm width of
timber equivalent. The position of the centroid and hence of the neutral axis
from the top surface is given by
7(6000 + 5000) = (6000 x 5) + (5000 x 60)
y = 30mm

600 x 10° 50 x 100°

[== — +(6000 x 25%)
+(5000 x 307)

= 1246.6 x 10*mm*
Maximum bending moment,

M = WL/4 = (2000 x 2)/4 = 1000 Nm

At lower surface,
1000 x 0.08
— 6.4MN/m
° ~~ 1246.6 x 10-8
At interface in timber, multiplying the stress by the ratio of the respective
y-values,

o = 6.4 x —20/80 = —1.6 MN/m’?

144 MECHANICS OF ENGINEERING MATERIALS
a

Since Ege) = 20E wood, then from eqn. [6.17], steer = 200 p00a- Hence at
interface in steel,

oe =—16.% 20. = — 32 MN/ore

At top surface,

o = 64x — 2x 20 = —48MN/m?
80
The distribution in the equivalent section is shown in Fig. 6.16(4) and the
actual distribution in Fig. 6.16(a).
If the equivalent section is made out of steel then this is represented in
Fig. 6.16(c) where the width of the timber is made into an equivalent width
of steel by dividing 50mm by 20, the modulus ratio, to give 2.5 mm.
The centroid is at

7(300 + 250) = (300 x 5) + (250 x 60)

y = 30mm
which is the same as before since the proportions are the same. As only
width dimensions have been changed by a factor of 20 compared with the
previous equivalent section, there is no need to recalculate /, since

LeeIp _ 1246.6 x 104 t: pelt

=r 0 623.3
x 10° mm

_ 1000 x (—0.03) _ 3
At top surface, ¢ = 330 = 48 MN/m

2
At interface in steel, 0 = —48 x a = —32 MN/m’

At interface in timber, 0 = —32 x 55= —1.6 MN/m*

801 2
At |ower surface,
fi 0 == 48 x —x—=
30 x 0 +46.
6.4 MN/m "

Reinforced concrete Perhaps the most common example of a composite beam is one using steel
sections bars to reinforce concrete. The steel is always embedded in the concrete on
the tension side of the beam owing to the weakness of concrete in tension,
but reinforcement may also be included on the compression side to reduce
the amount of concrete required if it were not reinforced.
Consider the case illustrated in Fig. 6.17 and make the conventional
assumption that the concrete takes all the compression and the reinforcing
bars take all the tension. All the required relationships have been derived
above, and it is only necessary now to solve for the unknown quantities as
required.
 BENDING: STRESS 145

Fig. 6.17

(b) (c) Stress diagram

Let the distance of the neutral axis from the outer surface in compression
be A, Fig. 6.17(6), and the ratio of the elastic moduli

E seek =

Lipepptay

From egns. [6.14] and [6.15],

Dake els
Oo. =— and C=
R R
Substituting the stresses into eqn. [6.12] for longitudinal equilibrium gives
£, IB
—P I dA, as
+ — JodA 0 [6.19]
6.19

For the steel reinforcement the tensile stress is considered constant over
the cross-sectional area A, and concentrated at y = (d —h) as in Fig.
6.17(c), since
h
| y, dA, = (d — h)A, and | y, dA, = — = (bh)
a 2
A,
so that eqn. [6.19] becomes

bh?

and, substituting E/E, = m and rearranging gives

WM mA,d
mA, ret
mA,
Miss A
( b )3 b b eal

which gives the position of the neutral axis.

Substituting 0, and o, for 0, and o, in eqn. [6.13] for equilibrium of
moments,

or
146 MECHANICS OF ENGINEERING MATERIALS

Now,

lop a:
RO yj Ee yee,
so that substituting for 1/R gives

2 My
oO;
1ph3 + mA(d — hy?

= M(d —h)m
Os [6.22]
1ph3 + mA,(d — h)?

Example 6.8
A reinforced concrete T-beam, Fig. 6.18, has a flange 1.5 m wide and 100 mm deep.
The reinforcement is placed in the web 380 mm from the upper edge of the flange.
The beam is designed so that the neutral axis coincides with the lower edge of the
flange. The limits of stress are 110 MN/m? for steel and 4 MN/m? for concrete. The
modulus ratio E.:.¢)/Econcrete is 15. Calculate (a) the area of the reinforcement, (b)
the maximum moment the beam can resist, (c) the actual maximum stress in the
steel and the concrete.

Fig. 6.18 |1500

Neutral axis [SRE

(Dimensions in mm)

(a) The area of steel can be calculated using eqn. [6.20]:

15x 0.1?
== + 15(0.38 — 0.1)A, = 0
whence

A, = 1790 mm?
(b) The denominator in egns. [6.22] is given as

1x 1.5 x 0.19 + [15 x 0.001 79(0.38 — 0.1)’] = 2.61 x 10-3 m4

Assuming the steel reaches the maximum stress, then from eqns. [6.22],

S261 0nte 10be 108

Vi 0.28 x15 = 68.3kN/m

Alternatively, if the concrete reaches the limiting stress,

uaz! x 10-3 x 4x 10°

01 = 104.5kNm
 BENDING: STRESS 147

Therefore the maximum moment the beam can resist is 68.3 kN /m.
(c) At this maximum moment the steel has reached its limiting stress of
110 MN/m’.

68.3 x 10° x 0.1

Actual maximum stress in concrete, 0, =
2.61 x 10-3

— 2.62 MN/m?

6.11 Combined bending

A number of situations arise in practice where a member is subjected to a
and end loading
combination of bending and longitudinal load. Problems of this type can be
most easily dealt with by superposition of the individual components of
stress to give resultant values.
Consider the rectangular-section beam shown in Fig. 6.19, which is
subjected to bending moments M about the z-axis, and an axial load P in the
x-direction.
If the end load acted alone, there would be a longitudinal stress, as shown
in Fig. 6.20(a)

Fig. 6.19

—
A
If the moment acted alone, then, as shown in Fig. 6.20(d), the longitudinal
bending stress would be

My
Ox =

By superposition the resultant stress due to P and M is

a ee ee P 12My
A Gl Bbd=s ba
[6.23]
The resultant distribution of 0, over the cross-section is shown in Fig.
6.20(c) and is obtained by superposition of the bending stresses (a) and the
direct stress (4). An interesting feature is that the neutral surface no longer
passes through the centroid of the cross-section since

| ovat Zo
148 MECHANICS OF ENGINEERING MATERIALS

Fig. 6.20 [2 _ Md P_Md

A 21 A. 21

+b eee

rf
we
P ! Md P Md
A oy A” 2
(a) (b) (c)

as in the case of simple bending.

Remember that y in eqn. [6.23] is measured from the centroidal axis of
the section and not from the new neutral axis.

6.12 Eccentric end : i on

If the end load P does not act at the centroid of the cross-section then it will
Shotlid itself set up bending moments about the principal axes. For example, in Fig.
6.21(a) a bar is subjected to a load P, which is eccentric from the z- and y-
axes by amounts m and respectively. The equivalent equilibrium system is
with the load P acting at the centroid and moments Pm and Pn acting about
the z- and y-axes, as in Fig. 6.21(d).

Fig. 6.21

(a) (b)
The direct stress due to P alone is therefore
hee i:
Cn
ye
that due to bending about the y-axis is
ne Pee
C=
I,
and that due to bending about the z-axis is

Ui Pmyu
Oy = if

Therefore the resultant longitudinal stress is by superposition

/ i Mt
Oma oO», oteoO» ai Oy

_P. Paz Pmy

Fis [6.24]
 BENDING: STRESS 149

Substituting J, = bd?/12 and I, = db*/12 gives

P 12nz 12my
es, (4H 7 Zi ) [6.25]

where z lies between +)/2 and y between +d/2.

A practical example of the use of this analysis is in civil engineering
where short concrete beams are subjected to a compressive load which may
be eccentric. In materials which are strong in compression but weak in
tension, such as concrete, it is necessary to limit the eccentricities m and so
that no tensile stress is set up. The condition for no tension is that the
compressive stress due to P is greater than or equal to the maximum tensile
bending stresses set up by the moments Pm and Pn. Therefore, from eqn.
[6.25],
6n 6m
l me >0 [6.26]

This relationship defines the locus of maximum eccentricity as shown by the

shaded area in Fig. 6.22. When P is applied within the shaded area then no
Fig. 6.22
tensile stress will be set up anywhere in the cross-section. The limits on the
z- and y-axes are +b/6 and +d/6, which has resulted in what is known as
the middle-third rule for no tension. Typical distributions for various
amounts of eccentricity along one principal axis are shown in Fig. 6.23.

+ve
Fig. 6.23

—ve
—ve

Example 6.9
In a tensile test within the elastic range on a specimen of circular cross-section an
extensometer is being used which will only measure deformation on one side of the
specimen. Determine how much eccentricity of loading will give rise to a 5%
difference between the surface stress derived from the extensometer and the
average stress over the cross-section.

Let the average stress be a; then for a 5% error due to non-axial loading the
resultant stress on one edge of the specimen will be 0.950, and at the
opposite end of the diameter 1.050. From eqn. [6.25],

A MY max
0.950 = o(1—
if

Therefore

0.057
Ohecies
150 MECHANICS OF ENGINEERING MATERIALS

For a circular cross-section J = rd*/64. Therefore

0.05md* /64
0.00054
(md2/4)d/2
Thus in a tensile test on a specimen of 10 mm diameter, the eccentricity
of loading must be /ess than 0.063 mm to avoid surface stresses being more
than 5% greater than the average direct stress.

Example 6.10
A slotted machine link 6mm thick, illustrated in Fig. 6.24(a), is subjected to a
tensile load of 40 kN acting along the centre-line of the end faces. Find the stress
distribution for a section through the slot such as AA.

Fig. 6.24

(a) Section AA
(dimensions in mm) (b) Stress distribution

We must first find the centroid C of the section AA and by taking the
moment of area about the bottom:

[(40 x 6) + (10 x 6)]% = (40 x 6 x 20) + (10 x 6 x 55)

so that

<= 27 mm

Hence the load is acting eccentrically with respect to the centroid of the slot
cross-section AA by an amount

e=30—27 = 3mm

This eccentricity gives rise to a moment

M = 40000 x 0.003 = 120N/m

which will cause tensile bending stress along the edge at D and compressive
bending stress along the edge at B.
For bending only, the neutral axis passes through C and the greatest
bending stresses occur at B and D; thus y4, = 27 and 33mm and the
second moment of area is 91.28 x 10° mm*. Therefore

oR =
? ee
91.28 x 10-9 = : Na 4
120 x 0.033
Op=4 eee I
 BENDING: STRESS 151

40000
Direct stress = +133 MN/m?
0.05 x 0.006 —
The resultant stresses will be

AtB + 133 — 35.5 = 97.5 MN/m?

AtE + 133 +17.1 = 150.1 MN/m

AtF + 133 + 30.2 = 163.2 MN/m2

AtD = + 133 + 43.5 = 176.5 MN/m°

The distribution of stress is shown in Fig. 6.24(d).

6.13 Asymmetrical or
The previous analysis has been concerned with bending about an axis of
skew bending
symmetry. However, many occasions arise in practice where bending will
occur either of a section which does not have any axes of symmetry or of a
symmetrical section about an asymmetrical axis. In order to express the
conditions of equilibrium in asymmetrical bending a knowledge of first and
second moments of area about arbitrary axes through the centroid of the
section is required. It is therefore necessary to study certain properties of
areas before embarking on the analysis of stress distribution in asymmetrical
bending. The reader is here referred to Appendix A.

Fig. 6.25

Asymmetrical pure bending of a beam is shown in Fig. 6.25 for positive

external moments M, and M, applied about an arbitrary set of centroidal
axes. Considering first bending in the xy-plane only, the equilibrium
equations are

|a,dA=0 (neutral surface through the centroid)

A

|o,ydA = M, and 0,2dA = M,

A JA

But o, = yE/R,, where R, is the radius of curvature in the xy-plane; hence

M, = ral
le,
yd =Ie
NG A R,
152 MECHANICS OF ENGINEERING MATERIALS

and
E ETL,
M, =— |yzdA =—*
i; Ry JA Rk,

Iz is called the product moment of area (see Appendix A).

For bending only in the xz-plane a procedure similar to the above gives

R. R.
E El,
M, = |vat=>
R,. R.

For simultaneous bending in the xy- and xz-planes the above

relationships may be superimposed to give

M, = be
ot RS Gale
Fen
ee (6.27]
y z

The radii of curvature are obtained from the above equations as

1 _ Maly
—Myly
R, ELI, - chy

OU is Pee se
R.z EIT. ailsye )

Note that eqns. [6.27] can be written in matrix form as

M hi, dilbe WIA he
= $ bs [Re [6.28]
M, El,, ET, | \1/R,
so that the radii of curvature can be determined from the inverse matrix
relationship

Rela def ip aipa

1/R, £E yz I, M,

Hin Dy. M,
= E(t ]= Ts) ae
ae ep |te} [6.29]

The resulting bending stress is therefore the sum of the components for
bending in each of the xy- and «z-planes:
yE r al
fae
0, =—+—
ay
WML, — Mylyz) + a(My I, — MzJjz)
[6.30]
eae
 BENDING: STRESS 153

The neutral surface, where o, = 0, is defined by the plane

WML, oe) M,1,) + 2(M,I, ae Mil.) =0 [6.31]

Establishing the position of this plane enables one to determine the points
which are furthest from the neutral plane and hence subjected to the
maximum stresses. If bending is about the principal axes of the section for
which J, = 0, then

Oy
= Mzy siMyz [6.32]
Ie ds
If either M, or M, is in the opposite sense to that shown in Fig. 6.25, then
the appropriate negative value must be used in eqn. [6.30]. This is illustrated
in the following example.
If there is only a single external applied moment M about an axis ss
inclined at 6 to one of the principal axes, as in Fig. 6.26, then the moment
vector M can be resolved into components M, and M, about the y- and z-
axes, so that

M,=—M sin @ and M,=M cos0

Fig. 6.26

and substituting in eqn. [6.32],

—Mz sin 9 My cos 0
= ale [6.33]
a ae? iy
The neutral plane will no longer be perpendicular to the plane of
bending, as in the symmetrical problem, but can be determined by putting
0, = 0 above; then
—z sin 8 y cos 0
0
if I.
154 MECHANICS OF ENGINEERING MATERIALS

and

Pa
i, an Ga ban o
xnI&

where ¢ is the inclination of nn, the neutral surface, to the z-axis.

The neutral surface is perpendicular to the plane of bending if either
I, = 1, or 0=0.

Example 6.11
The angle section shown in Fig. 6.27 is subjected to a bending moment of 2kKNm
about the z-axis. Determine the bending stress distribution.

Fig. 6.27

100 mm

Considering the first moments of area about vertical and horizontal edges
respectively,

(90 x 10 x 5) + (60 x 10 x 30)

eo (90 + 60)10 ages
_ _ (90 x 10 x 55) + (60 x 10x 5) _
a (90 + 60)10 aes
which gives the co-ordinates of the centroid.

10 x 90° 60 x 10°
I, = — + (10 x 90 x 202) +—— + (60 x 10 x 302)

= 151.2 x 10*mm?*
 BENDING: STRESS 155

90 x 103 10 x 603
+ (90 x 10 x 107) + 17 + (10 x 60 x 153)
D9

= 41.3 x 10+ mm*

Ty = [90 x 10 x 10 x (—20)] + [60 x 10 x (—15) x 30]

= —45 x 10*mm*
The position of the neutral axis may now be found using egn. [6.31],
from which, with M, = 0,

y(2000 x 41.3 x 1078) + z(—2000 x (—45) x 1078) =0

U1 00) and = 47°30)

ZS

The maximum tensile stress will occur at A, and using eqn. [6.30], in
which M,, will be zero,

AT 3 10S ee OD a le)
Fe OOS
ve [(41.3 x 151.2) — 452]10-16

= 101 MN/m?
At B,
5) x 10-8)]
SURBS 10”) os (—0.045 x (—4
ADUUIS

ri
y=
- (4l3 5151.2) —457]10-

Neutral axis = —47MN/m’

The maximum compressive stresses occurs at C, where

_ 2000[(—0.065 x 41.3 x 10-8) — (0.005 x (—45) x 1078)

[(41.3 x 151.2) — 452]107!6
OC
7
= —116.6 MN/m
Fig. 6.28 The distribution is illustrated in Fig. 6.28.
Figure 6.29(b) shows a spreadsheet for this problem. The second and
product moments of area of the section have been derived from the co-
ordinates of the points defining the cross-section as described in Appendix
A. These properties plus the co-ordinates of the points relative to the section
centroid are listed. Given this information, the stress at each point can be
written economically in matrix form as
7 7 ae
—Ly:z
1 IE
Ox [z | CANE = 12) Ea
M, 6.34
s ly |Ld
|
156 MECHANICS OF ENGINEERING MATERIALS

H { af
ES ly Iz _ Iz de
2 1512500 413000 —_ -450000_— (F2*G2-H2%2)
3
4
5 Point z stress Matrix Moments i
6 b
7 35 «+B 7 *$1$7+C7 *$1$8 +F2/12 | -H2/12 [EE +F7*H7+G7*H8
8 35 +B8*$1$7+C8*$1$8 -H2/I2 +G2/!12 __-2000*1000 = +F8*H7+G8*H8
9 25 —«+ BO* $1$7+C9*$1S8
25 +B10*$I$7+C10*$1$8 tan -17/\8
5-65. +B11*$1$7+C11*$1$8 | theta @ATAN(H10)
65 +B12*$1$7+C12*$1$8 @DEGREES(H11

G
ly = ly det
1512000 413000. -450000 4.22E+11

Point z y stress ‘Matrix Moments : a

b
100.51 |_ 3.58E-06 |1.07E-06 Jann 2.13
-27.47 1.07E-06 9.79E-06 eOUCODOs 1.96
—47.04 Hl |
59.60 | tan —1.09E+00 =f
-116.58 | theta —8.28E-0.01 rads
PRB
DEGCwMVNMFHRWANE
5 ~95.25 -4.75E+01
(b) Spreadsheet Display

Fig. 6.29 For given moments on the section this can be simplified to
a
o,=[z |A [6.35]
where a and 4 are determined from above. Note that because the co-
ordinates of the section points are given in mm, the moment has been
entered in Nmm so that the stresses appear in N/mm?. Dimensions,
moments and stresses can be calculated in any consistent set of units.

6.14 Shear stresses in

The presence of shear force indicates that there must be shear stress on
bending
transverse planes in the beam. It is not possible to make use of the
conditions of geometry of deformation and the stress-strain relationships
except in the development of an exact solution. However, from the
assumptions about the validity of the bending-stress distribution, it is
possible to estimate the transverse and longitudinal shear-stress distribu-
tions in the beam by using only the cendition of equilibrium.
Firstly, consider the bending-stress distribution in the short section of
beam of length dv shown in Fig. 6.30. The bending moment increases from
M on GK to M + (dM/dx)dx on HJ; therefore the bending stress on any
arbitrary fibre must increase from

o=— on GK [6.36]
 BENDING: STRESS 157

Fig. 6.30

to

+ eax = (M+ dM
Sar) on HJ [6.37]
x

at each value of y.
Now consider the strip of beam below an «z-plane CDEF, Fig, 6.31, at a
distance y; from the neutral surface. Each fibre in this strip will have an
increase in bending stress do along the length dx, so that taken over the
hatched area A there will be a resultant axial force equal to

[acdi=ep (6.38]

Fig. 6.31

Equilibrium of the strip is maintained by a shear force Q), acting on the

surface of the plane CDEF. Therefore

OF =|A dodd (6.39]

Now subtracting eqn. [6.36] from eqn. [6.37] gives
J
do == GUis7
158 MECHANICS OF ENGINEERING MATERIALS

Hence, from eqn. [6.39],

0.3 |dM~ dA [6.40]

If the width of the strip at the plane y = 7 is 4 and this is small compared
with the depth of the beam, then the shear stress on CDEF is almost
uniformly distributed and can therefcre be expressed as
Qy
Tye = b dx

Note that this is a positive shear stress, being the product of negative
directions of the normal to the plane and the stress direction.
Substituting for Q,, in eqn. [6.40] gives
M 1
ie pee | ydA [6.414]
x A

But, from eqn. [6.3], dM/dx = @ the vertical shear force on the section and
the integral is the first moment of area A about the neutral surface; therefore
eqn. [6.41a] becomes

=
;|
= dA =
Ay
—- [6.415]
ba) et bI
Using the principle of complementary shear stresses, Fig. 6.31(d), it is
then evident that the vertical shear stress, T,,, is also given by

Txy — a [6.42]

Again we note that 7,, is positive and so 1s correctly related to the positive
shear force Q.
This solution is only exact for a constant shear force along the beam;
however, if the cross-section is small compared with the span, the error
introduced by a varying shear force is quite small.

Example 6.12
A beam of rectangular cross-section, depth d, thickness b, is simply supported over
a span of length I, and carries a concentrated load W at mid-span. Determine the
distribution and maximum value of the transverse shear stress.

Although changing sign at the centre of the span, the shear force Q is
constant in magnitude along the whole span and is equal to W/2.
Considering the cross-section shown in Fig. 6.32, the transverse shear
stress on some arbitary line EF at a distance y from the neutral surface is

Fig. 6.32 Beam cross-section Stress

e) distribution
ices

(a)
 BENDING: STRESS 159

given by eqn. [6.42], where 4 is the shaded area below EF and jy is the
distance of the centroid of A from the neutral surface. Therefore

d 1 (ee(d BA i aNe Waar

p=6(§-») <3(F+2) =3|(6) >
Ay = b( —— = =

The vertical shear stress is therefore

(oer == x5(5) |
2 ul(s)| (6.43]
The above expression shows that the distribution of vertical shear stress
down the depth of the section is parabolic. The shear stress is zero at the
outer fibres where y = +d/2, as it must be since the complementary shear
stress in the longitudinal direction must be zero at a free surface. The
maximum value is at the neutral surface where y = 0; therefore

_ We 3W
Cee Ty
If uniformly distributed the shear stress would be given by the shear
force divided by the area, or
_
Txy mean = Phd

Hence the maximum shear stress is 1.5 times the mean value.

Example 6.13 RE Pe,

ee ee La Ee
P A box beam is built up of plate material riveted together as shown by the cross-
section, Fig. 6.33. It is simply supported at each end of a 3m span and carries a
concentrated load of 12 kN at 1m from one end. Estimate a suitable rivet diameter
if the rivets are to be pitched at about 100 mm intervals. The shear stress is not to
exceed 50 MN/m? in each rivet.

Fig. 6.33

150|mm|
160 MECHANICS OF ENGINEERING MATERIALS

The maximum shear force will be equal to the larger of the two reactions,
12 2 /SAK:
Now, the force tending to shear the rivets is due to the variation of
bending stress along the length of the beam and is given by eqn. [6.414] if
slightly rearranged as follows:

Shear force per unit length = 7,,5 = QAy

i
If the rivet pitch is p and the cross-sectional area a, then the allowable
shear force per unit length is Ta/p. Hence

fa QAy
bal
As
Required area, a = QAYp
ap

ve 0.15 x 0.15° 0.13 x 0.117

= ~ (2 x 0.09 x 0.01 x 0.06”)2

~ 0.2132 x 107* mt
At the interface between the outer plates and the side plates where
shearing would occur,

Ay = 150 x 10 x 70 = 105 x 10° mm?

Therefore

_ 8000 x 105 x 10°s< 021

a = 7.89 x 107°m?
~ 50 x 10° x 0.2132 x 10-4

= 78.9 mm?

As there are two rivets at each interface resisting shear, the diameter of
each is

A
78.9 4\2 iil ON
2 T

The best compromise is to use 7 mm rivets at 97 mm pitch, giving 31 pitches

in the length of the beam.

6.15 Bending and shear

stresses in |-section beams
The I-section beam shown in Fig. 6.34(a) is widely used in the construction
of buildings, bridges, etc. The shape is efficient to resist both bending and
shear. The latter is carried almost entirely by the web, and the flanges are
located where the bending stress is highest. The practical I-section is usually
idealized for ease of calculation into the rectangular shapes shown in Fig.
 BENDING: STRESS 161

(a)

6.34(4). The second moment of area of the section is given by

BD> (b/2)d>— (b/2)d3__ BD* — bd?

= =
12 12 12 12
The bending stress distribution is then given by
12My
° = SD BB a
which is illustrated in Fig. 6.35.

Fig. 6.35

It can be shown that for a typical I-section the flanges carry

approximately 80% of the total bending moment on the cross-section. In
designing an I-section against failure, consideration must be given, in
addition to the strength of the tension flange, to the avoidance of buckling
(Chapter 12) in the compression flange.

Shear stresses in web The shear-stress distributions in I-section beams are rather more complex
and flanges than in a rectangular section. Referring to Fig. 6.36, firstly we will examine
the distribution of vertical shear 7,, parallel to the axis yy.
For the web, from eqn. [6.414], the shear stress at a distance y’ from the
neutral axis is given by
oO is
y=
Trey =
Th J dA
y
162 MECHANICS OF ENGINEERING MATERIALS

Because the section has a different width for the flange and the web this
integral has to be expressed in two parts:
O od [2 D/2
i) | hy dy Si | By dy
p Tt Jy! 4/2

“10> ps (De 2) (6.45]

This expression gives a parabolic distribution superimposed on a constant

value as shown in Fig. 6.36. For sections in which the wall thickness is small
relative to the overall dimensions, the shear stress in the web may be
approximated by dividing the shear force by the area of the web.

AB Tz stress across flange

Fig. 6.36
Bly
2

stress along yy

The maximum value occurs at the neutral axis, where y = 0, and is

Tay
ie = Oo
BF ie a ;Pie
( p. ) (6.46)
:

In that part of the flange directly above and below the web the vertical
shear stress can be expressed as

=i OM LDS -tiy
cee Tolle gd yal Mae
However, in those parts of the flange on each side of the web the top and
bottom surfaces are ‘free’ from the load, and therefore the longitudinal and
complementary vertical shear stresses must be zero. Thus the distribution is
parabolic as for a rectangular section.

Area A
Fig. 6.37

(a)
 BENDING: STRESS 163

It is evident that in the flanges the vertical shear stress and _ its
complementary component contribute little to balancing the longitudinal
variation in bending stress. However, this may be achieved, as illustrated in
Fig. 6.37, by means of a shear force Q,,. lying in an xy-plane which cuts off a
segment of the flange. A complementary shear force Q,.. then occurs in a yz-
plane. The net end load, P, is given by

Pe| do dA
A
(from eqn. [6.38]) and is equal to the shear force Q,,, which is the shear
stress T,, multiplied by the area f) dx, 1.e.

jp = OF = Tzyl2 dx

Therefore

Rete = |do, dA
A

Tzx = — Ay [6.47]

where Qis the vertical shear force on the section and y is still measured from
the neutral axis to the centroid of the area.
D—tn
Ay = 2h

From eqn. [6.47] and the complementary (positive) shear-stress

condition,

Ty =e = Oz [6.48]

This is a linear distribution of shear stress in the z-direction, being zero at

the outer edges of the flanges and a maximum at the joint with the web. The
distribution is shown in Fig. 6.36, in which the maximum value is

Tre =
O(D b)\(B
4]
ai) [6.49]

Example 6.14
The vertical steel column of 5m height and rolled |-section shown in Fig. 6.38 is
built-in at the lower end and subjected to a transverse force of 4 kN at the free end.
Calculate the bending- and shear-stress distributions at the fixed-end cross-section.

At the base
Bending moment = 4 x 5 = 20kNm

Shear force = 4kN

164 MECHANICS OF ENGINEERING MATERIALS

1.65 MN/m?

150 A
== 422 MIN/m

ey Ray eee
i ILA
ee Bending
|
Shear
16 218mm 16

(a) (b)

32 x 150° af218 x 16°

Second moment of area =
14 12

— 9.075 x 10° mm*

mar = £0.075 m
Therefore

20 x 10° x 0.075 a
= = +165 MN/
ee 9.075 x 10
In the flanges the shear stress in the y-direction is given by

4000 e
Tt 7x 0.016 x 9.075 x 10-6 i 2 ax0.016) dee
2

_ 440
x 10°
(0.0757 — y’)
a 2
This is a parabolic distribution varying from zero at the outer surfaces to

Ty = 220 x 10°(0.0757 — 0.008") = 1.22 MN/m’

at the section where the flanges join the web.
In the web itself, since the width, which appears in the denominator for
shear stress above, is 250 mm, it is evident that the shear stress in it can be
neglected compared with that in the flanges.
A consideration of the geometry of the section indicates that shear
stresses of the T,, type are also insignificant.
The distributions of bending and shear stresss are shown in Fig. 6.38.

6.16 Shear stress in

; : There are a number of beam sections widely used, particularly in aircraft
thin-walled open sections construction, in which the thickness of material is small compared with the
and shear centre overall geometry and there is only one or no axis of symmetry. These
members are termed thin-walled open sections, and some common shapes are
shown in Fig. 6.39
The arguments applied to the shear-stress distribution in the flanges of
I-sections may also be applied in the above cases; however, there is one
important difference owing to the lack of symmetry in the latter.
If the external applied forces, which set up bending moments and shear
forces, act through the centroid of the section, then in addition to bending,
 BENDING: STRESS 165

Fig. 6.39

Channel Top hat Zed Angle Cee

twisting of the beam will generally occur. To avoid twisting, and cause only
bending, it is necessary for the forces to act through a particular point,
which may not coincide with the centroid. The position of this point is a
function only of the geometry of the beam section; it is termed the shear
centre.
The following examples illustrate how the position of the shear centre
may be found. Referring to Fig. 6.40, in which the channel section is loaded

Fig. 6.40

by a vertical force /’, the y- and z-axes are principal axes and hence J/,, = 0.
For the shear stress in the flanges the analysis is similar to that for the I-
section.

Tie = g dA
an A be |,

where Q is the vertical shear force on the section caused by the applied force
F. At a distance z from the free edge of the flange,
fa)id O tx 6.50)
tae =~]tu, ge=-tdz
== x —
re
The shear stress varies linearly with z from zero at the left to a maximum
at the centre-line of the web:
166 MECHANICS OF ENGINEERING MATERIALS

_ Qbd
Txz max = Cae

The average shear stress is Qbd/4/,, and therefore the horizontal shear force
in the top and bottom flange is

Ob td
Qrs = 4],
The couple about the x-axis of these shear forces which would cause
twisting of the section is

On dt
Qy.d =
4,
Twisting of the section is avoided if there is an opposing couple of equal
magnitude. Let the vertical force F act through a point C, the shear centre,
at a distance e from the middle of the web as shown in Fig. 6.40, so that Fe
balances Q,,d. Now Q must equal F for equilibrium; therefore

Qe = Qyzd
and hence
fogs

a = :
or

abd’t (6.51]
z&

which locates the position of the shear centre and is only a function of the
geometry of the section. The vertical shear stress 7,, in the web may be
found in the same way as was that for the I-section.

Example 6.15
A thin-walled tube of circular cross-section has inner and outer diameters of 50 and
70mm respectively. If it is slit longitudinally on one side, at what position must a

Fig. 6.41

8 KN

J 30 sin@

(Dimensions in mm)
BENDING: STRESS 167

vertical force of 8 kN be applied so that there is only bending and no twisting of the
section? Calculate the maximum shear stress in the section.

Referring to Fig. 6.41, since the slit is narrow, the second moment of area, /,
for the section may be taken as
43
I (70* = 50*) = 872 x 10° mm*
64
For the element shown, the shear stress must be zero on the face of the slit
and increases with distance from this free edge as

L
==
al
tl 7
dA

where A is the area of the segment of thin wall between the slit and the point
of interest. For a thin-walled section y = r sin @ and dA = tr dé, therefore,

=fra Dpsin 0d0Ge

[= = 7 (1—cos ¢)

The maximum shear stress occurs when @ = 7, when

20r° 2 x 8000 x 30°

a = 16.5MN/m?
fi a PAS
The torque set up by the shear-stress distribution is given by
20 4, plo
| rrr do =< (1 — cos @)d0
0 . OAs

= Snr = 46.7 x 10* Nmm

Let the shear centre be at a distance e from the centre of the tube. Then
equilibrium of torques gives

Oe Snr

2nr*
= il Peace

Note that for a thin-walled tube, J ~ mr°¢ and e ~ 2r. For a circular tube
with no slit, it can be shown that the shear centre lies on the axis and the
maximum shear stress is half the value for the slit tube.

The analysis of the slit tube and the channel section described earlier was
relatively simple, since the section had one axis of symmetry about which
bending was made to occur. The more general case is that of an asymmetric
open section subjected to bending which is not a principal axis. The
determination of the shear stresses and shear centre in this case is described
in Megson!. Alternatively if the bending is considered relative to the
principal axes of the section, where /,, = 0, the analysis above can be used
for each principal axis in turn. The shear stress arising from the component
168 MECHANICS OF ENGINEERING MATERIALS

of shear force in each principal direction can then be combined by

superposition.

6.17 Bending of initially

The theory of bending developed so far has been related to initially straight
curved bars
bars and beams. The analysis will now be extended to include members
which are initially curved.
The geometry of curved bars has an important bearing on the bending-
stress distribution. If the depth of the cross-section is small compared with
the radius of curvature, then the stress distribution is linear as for straight
beams. On the other hand, if the depth of section 1s of the same order as the
radius of curvature, then a non-linear stress distribution occurs during
bending.
Similar assumptions are made for curved beams as for straight beams,
plane cross-sections remaining plane, etc., although a few of the assumptions
are not strictly accurate for the case of a bar with a small radius of curvature.

Fig. 6.42

(a) Unloaded (b) Loaded

Consider the curved bar shown unloaded in Fig. 6.42(a) and subjected to
pure bending M (Fig. 6.42(4)) with initial and final radii of the neutral axis
R, and R) respectively. The strain in a small element CD at a distance y
from the neutral axis is derived as for the straight beam and is

(Ro + y)o — (Ri + y)0

CD=
(Ri +9)
[6.52]
but for an element AB at the neutral surface there is no change in length, so
that R|@ = Rd. Therefore

—6
€cp = es [6.53]

Making the substitution ¢@ = R\@/R» gives

‘Sel
_ yl(Ri/Ro) = 1 _ (Ri = Ro)
= (6.54)
3 Rien, Ry(Ri + y)
For the slender beam, y can be neglected compared with R; and

(2 re
 BENDING: STRESS 169

For R; infinite, i.e. a straight beam, the expression reduces to that found
previously.
By using the same concept as for the straight beam it can be shown that
for no applied end load the centroidal axis and the neutral axis coincide, and
for equilibrium of the bending moment with the internal resisting moment

Mo ( \) 6.56]

6.18 Beams with a small

For the beam in which y is not negligible compared with Rj, the strain at
radius of curvature
distance y from the neutral axis is given by eqn. [6.54]. This is no longer a
linear distribution of strain across the section as for the slender beam, and

Fig. 6.43 | M
Neutral axis |

ee

oh

_ \7 Bending-stress
distribution

Centroidal axis

hence the distribution of stress is non-linear, as indicated in Fig. 6.43, and the
centroidal and neutral axes no longer coincide, as will now be shown.
Assuming that there is no applied end load,

| eat=o
A
and from egn. [6.54]

o = Ke =
LEN Rie Ry) 6.57
R,(Ri + y) oe
Therefore

E(R — =| Deep ea
Ry aie.
since the integral must be zero, and this is not the first moment of area about
the centroid; therefore the centroidal and neutral axes do not coincide.
For equilibrium of internal and external moments,

Me |ay dA
A

E(R, | ie
AM = dA 6.58
R; ARi+y ne
170 MECHANICS OF ENGINEERING MATERIALS

The integral term may be re-expressed as

a

:
y a4=| dA—R |
y dA
beeaay bea |e Richy

=| vad
A

since the second integral is zero as shown above. Now let the distance
between the centroidal and neutral axes be n; then, for the cross-section of a
curved beam shown in Fig. 6.44, y = y’ +; hence

|vaa=| yaa+| ndA =nA

A A A

Fig. 6.44

Neutral axis

since J,0/ 4A is zero, being the first moment of area about the centroid.
Thus
2
| eA
aRiy
Substituting into eqn. [6.58],

M _E\Ri— Ro)
nA R>

and further substituting from eqn. [6.57],

= ;(Ri +y)

or
o M
y nA(Ri +y) eo
which is in a form similar to the bending-stress relationship for slender
beams, the second moment of area term, /, being replaced by nA(R) + y).
 BENDING: STRESS 171

In order to determine the magnitude of bending stress it is first necessary

to find the values of n and R; for the particular shape of cross-section. From
the condition that

J
dA=0
les
it follows that

|way
/
War @
dA=0

or

[GS R’ , n Jato
, R'+y’ R+y R’ +)!

|84—ea
| |ey
R’ n
dA — | ——d4+ |— dA=0

Hence

|
eee = R’ — dA
ae
6.60

and

RR —7 [6.61]

For rectangular and circular sections the values of n and R, are obtained
from eqns. [6.60] and [6.61].

Rectangular section
d
(os Sra ayeD) ee

Circular section of radius r

r
»= R’ 6.63
4 2[R’ — (R? — 72)'/?) |

Example 6.16
A crane hook as illustrated in Fig. 6.45(a) is designed to carry a maximum force of
12 KN. Calculate the maximum tensile and compressive stresses set up on the cross-
section AB shown at (b).

The position of the centroid is given by

[48 + (2 x 24)] 54 = 24mm
es 1D 3
172 MECHANICS OF ENGINEERING MATERIALS

Fig. 6.45

Section AB

(Dimensions in mm)

(b)

Hence
dy = 54 — 24 = 30mm

The value of 7 is given as follows:

n=R'

(d + by) d
Nl

{bo oii [(dy _ bz) /d|(R' + dy Ne)} log, [(R’ + dz) /(R’ — d1)] — (b1 — 2)
78 36 x 54 3
== == Shana)
HEROIN Ty,
and
Ri =78—
3 = mm
A = 36 x 54 = 1944 mm?
M = —12000 x 0.064 = —768 Nm
(negative since curvature is reduced)
yp = —24-- 3 = —Z2Z1 mm
92 = 1304.3 = 453m

12 000
Direct stress on section AB =
0.001 944

= +6.2 MN/m?

—168 x-—0.021
Bending stress at B =
0.003 x 0.001 944(0.075 — 0.021)

= +51.1 MN/m?

Total stress at B = +57.3 MN/m?

 BENDING: STRESS 173

—768 x 0.033
Bending stress at A =
0.003 x 0.001 944(0.075 + 0.033)

= —40.1 MN/m

Total stress at A — —33.9 MN/m?

The higher value of stress at B illustrates the reason for having the
trapezoidal cross-section, i.e. more material is available at B.

6.19 Summary
Many important practical problems of members subjected to bending have
been analysed in this chapter and if this summary is to be succinct it must
pick out the most important principles. Firstly, little progress can be made
unless there is a clear understanding of the calculation of shear-force and
bending-moment distributions. Then the properties of areas are always present
in order to find the position of the neutral axis and the second moment of
area of the cross-section. From this point in order to analyse a wide range of
situations involving bending the same basic procedure is required for every
case: (i) to write the equilibrium statement relating applied forces and internal
reactions (as a function of stress), (ii) with appropriate assumptions to decide
on the geometry of deformation of a suitable free body of the beam, and (111) to
link (i) and (ii) through the elastic stress—strain relationship. The final solution
will also entail the use of the specific boundary conditions of the problem.

Reference
1. Megson, T. H. G. (1990) Aircraft Structures for Engineering Students,
2nd edition, Edward Arnold, London.

Problems 6.1 Draw the bending-moment and shear-force diagrams for the beam
loaded as shown in Fig. 6.46 and insert the principal values on each
diagram.

im [sm
Fig. 6.46 =

2m 1m 2m

6.2 A gear wheel is mounted on a 25mm diameter shaft which is driven

through a 600 mm diameter pulley as in Fig. 6.47. During operation
the gear wheel is subjected to a horizontal force of 3 kN and a vertical
force of 5kN. Calculate the value of the maximum bending moment
on the shaft and sketch the bending-moment diagram.
6.3 A trolley of weight W is supported on wheels as shown in Fig. 6.48. It
can move over the full length of the beam AB. Show that the
174 MECHANICS OF ENGINEERING MATERIALS

Fig. 6.47 Pulley weight = 1 kN 2 KN

= Gear wheel

200 mm > |< 400 mm [200mm 5 kN 6 KN

|I~<— > |<

maximum bending moment occurs in the beam when a wheel of the

trolley is at a distance c from either support, where c is given by

15 d
<=F(2-4)

W.
Fig. 6.48

B
ui

6.4 The portal frame shown in Fig. 6.49 is pinned at A, C and E. Sketch
the bending-moment diagrams for the sides and top of the frame and
insert the principal values.

Fie, 6.49 10) 40) 40., 10° 4

TTT TTT om
10m

A, E, C pinned
B,D fixed

6.5 Adam 12m in length retains water at a depth of 3m as shown in Fig.

6.50. Sketch the bending-moment and shear-force diagrams for the
dam. The density of water may be taken as 1000 kg/m?.
6.6 <A load of 50KN is supported over a length of 4m on the rigid beam
AB as shown in Fig. 6.51. Plot the shear-force and bending-moment
diagrams for the beam, stating the values of bending moment and
shear force at key points along the beam. Calculate the magnitude and
position of the maximum bending stress in the beam.
I = 30 x 10°mm‘.
6.7 A floor joist 6m in length is simply-supported at each end and carries
a varying distributed load of grain over the whole span. The loading
 BENDING: STRESS 175

Fig. 6.50

50 kN
Fig. 6.51 Rubber padding

300 mm

distribution is represented by the equation m = ax” + bx + c, where w

is the load intensity in kN/m at a distance x along the beam, and a, b
and ¢ are constants. The load intensity is zero at each end and has its
maximum value of 4kN/m at mid-span.
Apply the differential equations of equilibrium relating load, shear
force and bending moment to obtain the maximum values and
distributions of shear force and bending moment.
6.8 A horizontal beam, the weight of which may be ignored, is loaded as
shown in Fig. 6.52. The distributed load varies linearly from zero at
the left-hand end to 6kN/m at a distance of 3m from the left-hand
end. Sketch the shearing-force and bending-moment diagrams for the
loading shown and insert the principal values of each on these
diagrams.
6.9 A cantilevered balcony which projects 2m out from a wall 1s
constructed of timber joists at 0.33 m spacing supporting a boarded
floor 12 mm thick. The design loading on the floor is 4.5kN/m? and
the self-weight of a joist and its associated boarding may be neglected.

Fig. 6.52 Distributed load 6 kN/m 1 kN

176 MECHANICS OF ENGINEERING MATERIALS

If each joist is to be 120 mm deep, determine the required width so

that the maximum tensile bending stress does not exceed 10 MN/m’.
For the purpose of calculating the second moment of area, it may be
assumed that the neutral axis of the combined joist cross-section and
its associated strip of boarding is at the mid-depth of the whole
section.
Also find where the neutral axis is correctly loaded when the joist
width is determined.
6.10 A channel which carries water is made of sheet metal 3 mm thick and
has a cross-section of 400mm width and 200mm depth. If the
maximum allowable depth of water is 150mm determine the
maximum simply-supported span. The maximum bending stress
(tension or compression) is not to exceed 35 MN/m’, and self-weight
160 kN may be neglected. Loading due to water, 9.81 kN/m’.
6.11 A 203 x 133 Universal Beam with the section properties given in the
Table below is loaded as shown in Fig. 6.53. Determine (i) the largest
bending stress which occurs under this loading and (11) indicate where
on the cross-section this stress occurs.

é' Thickness
Serial Mass per Depth Width SS SS ee
size metre D B Web t Flange 7 Area
(mm) (kg) (mm) (mm) (mm) (mm) cm?

PAU < 1e%3) Bt 206.8 133.8 6.3 O33 38.0

Secon d moment of area Elastic modulus Plastic modulus

Axis 2-z Axis y-y = Axis 2-2 Axis y-y Axis 2-2 Axis y-y
(cm*) (cm*) (cm?) (cm?) (cm?) (cm?)

2880 354 278.5 52.85 312.6 83.7

6.12 A beam with the cross-section shown in Fig. 6.54 is to be bent about
the x—x axis. Determine the optimum value of / in order to minimize
the outer fibre stress for a fixed bending moment and beam width, B.
The second moment of area of a triangle about its base is given by

I = 1/12 x base x (height)*

6.13 Given the applied moment on a beam and the dimensions of its cross-
section as shown in Fig. 6.55, develop a spreadsheet to calculate the
maximum bending stress in the beam for
(i) a hollow circular section
(ii) a rectangular section
 BENDING: STRESS 177

Fig. 6.54

angle constant

Fig. 6.55

6.14 The roof of a petrol station is made up of two main beams and eleven
cross-beams (purlins) as shown in Fig. 6.56. The main beams have an
I-section 200 mm wide, 600mm deep with a web thickness of 10 mm
and a flange thickness of 15mm. The purlins also have an I-section
125 mm wide, 250 mm deep with a web thickness of 6mm and a flange
thickness of 10mm. Calculate the maximum stresses in each type of
beam if there is a snow loading of 120kg/m? on the roof. The density
of the beam steel is 7850 kg/m? and you should allow for the weight of
the beams.
6.15 A tapered shaft of length / is built in at the larger end of diameter d),
and is free at the smaller end of diameter d;. A force W is applied at
the free end perpendicular to the axis of the shaft. Show that the
maximum bending stress at any section distant x from the free end of
the shaft is given by
Wx

(1/32){dy+ (dr — di)(x/2)}?

178 MECHANICS OF ENGINEERING MATERIALS

15m

Hence determine the distance x at which point the greatest value of

the maximum bending stress occurs.
6.16 A small trailer has a suspension system as shown in Fig. 6.57. If the
weight of the trailer is 4kN and its centre of gravity is 0.5 m forward
of the wheels, calculate the bending moments and torques in sections
AB and BC. Calculate also the maximum bending and shear stresses in
these sections. Ignore any effects at corners or changes in section.
6.17 A timber beam 80 mm wide by 160mm deep is to be reinforced with
two steel plates 5mm thick. Compare the resisting moments for the

Fig. 6.57

; Tube 30 mm 0.D.
ae 20 mm I.D.
 BENDING: STRESS 179

same value of the maximum bending stress in the timber when the
plates are: (i) 80 mm wide and fixed to the top and bottom surfaces of
the beam, and (ii) 160 mm deep and fixed to the vertical sides of the
beam. F for steel = 20 x E for timber.
6.18 A composite beam is to be made up of a U-shaped steel sheet with a
wooden board glued on top as in Fig. 6.58. What depth must the
wood be to cause the neutral axis in pure bending to be at the
horizontal diameter of the semicircle? Calculate the maximum
bending moment which may be applied to the beam if the maximum
stresses in the steel and wood are not to exceed 280 MN/m? and
7 MN/m’, respectively. The moduli for steel and wood are 210 GN/
m? and 7GN/m’ respectively.

Fig. 6.58

6.19 Extend the spreadsheet of Problem 6.13 to find the stress in a beam
section made of multiple materials, given the modulus and dimensions
of each part. Assume that the centroid of all the individual pieces lies
on the neutral axis. Use the spreadsheet to check the results of
Example 6.6.
6.20 Extend the spreadsheet above to find the stresses in a beam section of
multiple materials where the distribution of materials is not
symmetric. Use the spreadsheet to check the results of Example 6.7.
6.21 A reinforced-concrete beam has a rectangular cross-section 500 mm
deep and 250mm wide. The area of steel reinforcement is 1100 mm”,
and it is placed at 50 mm above the tension face. Calculate the resisting
moment of the section and the stress in the steel if the compressive
stress in the concrete is not to exceed 4.2 MN/m/? and the modular
ratio is 15.
6.22 A horizontal beam of rectagular cross-section 100mm deep and
50mm wide is simply supported at each end of a 1.5m span. Vertical
loads of 5kN are applied at 0.5m and 1m from one end, and a
horizontal tension of 40 kN is applied at the ends 25mm below the
upper surface. Determine and plot the distribution of longitudinal
stress across the section at mid-span.
What eccentricity of end load is required so that there is just no
resultant compressive stress at the outer surface?
6.23 The cross-section through a concrete dam is illustrated in Fig. 6.59.
Calculate the required width of the base AB so that there is just no
tensile stress at B. What is the resultant compressive stress at A?
Loading due to water = 9.81 kN/m*. Weight of concrete = 22.7 kN/
m>,
6.24 A concrete cooling tower may be assumed to consist of two truncated
cones as in Fig. 6.60. If the estimated maximum horizontal wind
pressure is 1.5kN/m/’, calculate the wall thickness of the tower in
order to avoid tensile stresses in the concrete. The density of the
180 MECHANICS OF ENGINEERING MATERIALS

Fig. 6.59

concrete is 2400kg/m*. The volume of a truncated cone

= (1/3)mh{R? +7 + Rr}.

Fig. 6.60

60 m

6.25 Derive an expression for the shear-stress distribution across a solid

circular section rod of radius R, subjected to bending. Calculate the
ratio of the maximum shear stress to the average shear stress on this
section.
6.26 A channel-section beam is 50 mm wide and 50mm deep with a 5mm>
wall thickness. It is simply-supported over a length of 1 m and carries
a uniformly-distributed load of 50 kN/m over its whole length. It also
has a line load of 50kN at mid-span. Sketch the shear stress
distribution across the beam section 0.25 m from one of the supports
and indicate the important values.
6.27 A circular tube of mean radius r and wall thickness t(< r) is subjected
to a transverse shear force Q during bending. Show, by derivation,
that the maximum shear stress, T, occurs at the neutral axis and is
equal to Q /mrt.
6.28 A wooden beam section can be made up by one of the two methods
shown in Fig. 6.61. If the shear force in the beam is constant at 3 kN,
calculate which design is preferable to keep the shearing forces in the
nails to a minimum. If the nails can withstand a shear force of 400 N,
determine the maximum permissible spacing of the nails along the
beam for the design selected.
6.29 A beam is made up of four 50mm x 100mm pieces of timber glued to
a 25mm xX 500mm web of the same wood as shown in Fig. 6.62.
 BENDING: STRESS 181

Fig. 6.61

100 100

— <> || 3
15 15 15 15
(a) (b)
Calculate the maximum allowable shear force and bending moment
that this section can carry. The maximum shearing stresses in the
wood and glued joist must not exceed 500 and 250kN/m?
respectively, and the maximum permissible direct stress is 1MN/m/?.

Fig. 6.62 SOMO

500

100

(dimensions in mm)

6.30 A metal bar AD is 10mm wide and is bent into the shape shown in
Fig. 6.63. Ifa vertical force of 80 N is applied at D, calculate the depth
of the bar if both bending stresses and shear stresses in the metal must
not exceed 60 MN/m?.
6.31 A thin walled hollow square tube of sidelength 4 is to be used in an
aircraft wing. It is known from design handbooks that the side walls of
the tube will locally buckle if the shear stress exceeds
en ie
is el aoa (3)
where E = 70 GN/m’, v = 0.3, 6 = 100mm and ¢ is the thickness
of the sheet from which the tube is fabricated. Ifa shear force of 10 kN
must be supported: (a) find the shear stress in the side wall of the tube
on the neutral axis when the tube is fabricated from 1.5 mm sheet; (5)
find the minimum sheet thickness to prevent failure by local buckling.
 182 MECHANICS OF ENGINEERING MATERIALS
ae

Fig. 6.63
———— 100 "+
A B |

pe
80 N

6.32 Extend the spreadsheet of Example 6.11 so that the additional stress
due to an axial force of 10kN is included.
6.33 Plot the shear stress distribution in the I-beam of Fig. 6.38 using a
spreadsheet and eqn. [6.45]. Assume the section is loaded with a shear
force of 4kN along the line of the web, i.e. perpendicular to the shear
force of Example 6.14. Compare the results with the average shear
stress calculated by assuming the shear force is carried by the web
only.
6.34 A Z-section beam is 2m long and is supported as a cantilever with a
1kN load at the free end. The direction of the 1kN relative to the
beam section is as shown in Fig. 6.64. Calculate the magnitude and
position of the maximum tensile and compressive stresses on the
section.
6.35 A 100mm x 100mm angle section as shown in Fig. 6.65 is built in at
one end of its 1m length and subjected to a point load of 3kN at the
free end. The point load is applied at an angle of 20° to the vertical
axis as indicated. Calculate the stresses at L, M and N and the
orientation of the neutral axis.

Fig. 6.65 nal =i

} 10} | 1203 4

zZ-i--

100 Perea +3
1 kN 100

{ 10 =|

(dimensions in mm)

L 28.7

(dimensions in mm)
Fig. 6.64
 BENDING: STRESS 183

6.36 A channel-section beam | m long is built in at one end and subjected

to a point load of 1 kN at the free end. If the direction of the load is as
in Fig. 6.66, calculate the direction of the neutral axis and the
maximum values of the tensile and compressive stresses on the
section.

Fig. 6.66

6.37 Determine the location of the shear centre for the beam cross-section
shown in Fig. 6.67. Also calculate maximum values of the horizontal
and vertical shear stresses in a flange and the web respectively for a
vertical load of 500 kN applied at the shear centre.

Fig. 6.67 Vo
10 c

200

10
| 50 150
—< >|~< >
(dimensions in mm)

6.38 (i) Determine the shear centre of the three sections shown in Fig.
6.68. All are formed from plates of a constant thickness t, which is
small relative to the other dimensions of the section. (ii) For the
section of Fig. 6.6(a), calculate the maximum shear stress in the
section when a shear force of 27kN is applied vertically through the
shear centre. Assume 6 = 200mm, d = 250mm and ¢ = 15mm.
6.39 A proving ring, used to calibrate a testing machine, has a mean
diameter of 500mm and a rectangular section 76mm _ wide and
184 MECHANICS OF ENGINEERING MATERIALS

Fig. 6.68
i ees a eae

12.7mm thick. If the maximum permitted stress under diametral

compression is 55 MN/m/?, determine the maximum calibration load.
6.40 A U-shaped bar with a square cross-section (40mm x 40mm) is
subjected to forces as shown in Fig. 6.69. Sketch the stress
distribution at the section AB and insert the principal values.
1 kN 1 kN
Fig. 6.69

200 mm

6.41 A chain coupling is made up of a20mm diameter steel rod bent into
an ‘S’ shape as shown in Fig. 6.70. If the coupling is subjected to a
tensile load of 1 kN as indicated, Ss the maximum stress in the
steel.
 CHAPTER 7
Bending: Slope and Deflection

Having studied the stresses set up in bending, we now turn to the equally important
aspect of beam stiffness. In many structural elements, such as floor joists or aircraft
wings, the limiting constraint on the design is stiffness. Any design which is stiff
enough will be strong enough. It is important that we should be able to calculate the
deflection of a beam of given section, since for given conditions of span and load it
would be possible to adopt a section which would meet a strength criterion but would
give an unacceptable deflection.
The total deflection of a beam is due to a very large extent to the deflection
caused by bending, and to a very much smaller extent to the deflection caused by
shear. Various methods are available for determining the slope and deflection of a
beam due to elastic bending, and examples of the use of each method will be found in
this chapter. That part of the total deflection caused by shear will be discussed in
Chapter 9.

* 7.1 The curvature-

In Fig. 7.1, @ is the angle which the tangent to the curve at C makes with the
bending-moment x-axis, and (9 — d@) that which the tangent at D makes with the same axis.
relationship The normals to the curve at C and D meet at O. The point O is the centre of
curvature and R is the radius of curvature of the small portion CD of the
deflection curve of the neutral axis.
Numerically ds = Rd6 and 1/R = d0/ds. Using the sign convention for
bending described in Chapter 1 it will be seen that positive increments of ds
from left to right are associated with a negative change in d@ . Thus, when
signs are taken into account, the last equation becomes

ee ee [7.1]
Deflections of the neutral axis are denoted by the symbol v, measured
positive downwards, and are assumed to be relatively small, giving a flat
form of deflection curve; therefore no error is introduced in assuming that
ds = dx, that 0 tan0=dv/dx, and hence that d6/ds= d’v/dx?.
Therefore

] dv

Fig. 7.1 0
dj} =a
ee V

/ /
e-de 6 / IR

= Deflected shape
186 MECHANICS OF ENGINEERING MATERIALS

Note that from eqn. [6.5] the strain, €,, will be given by
2
ae Lbs 955 [7.26]
Using egn. [6.9], when elastic bending occurs,
Lic eM
Rey
Therefore
2
me=e a (7.3)
This is the differential equation of the deflection curve. If the variation of M
with x is known then this equation can be integrated twice to give the
deflection, v.

‘+ 7.2 Slope and deflection

The first integration of eqn. [7.3] gives the slope of the beam at a distance x
by the double-integration along its length when the origin is taken at A. Therefore
method
S25
g=F= [Grate fe
dx 1a ta
The second integration gives the deflection of the beam at the above point,
or

dv =i
C= [as = [([Gre)er+ Cy Si C, [7.5]
C and C;, the constants of integration, can be evaluated from the known
conditions of slope and deflection at certain points, usually at the supports.
Equations [7.4] and [7.5] are widely used for determining the slope and
deflection of a beam at a given point. Examples of their use will be found in
the following paragraphs. In each case the sign convention used to obtain
eqn. [7.3] will be adopted.

Beam simply supported This problem is illustrated in Fig. 7.2. From eqn. [1.13], the bending
with distributed loading moment at D is

MewL wx
eo a
and

dv wL wx
El] = 4
er ae [7.6]
and

dv IE, 3
joyemepente et Set
 BENDING: SLOPE AND DEFLECTION 187

Fig.12 w/unit length

£\ £\
(a) L Gi

A D B
(b)
Sw f- X= {awe

A |b )
(c) i ae

es Deflected shape

From symmetry dv/dx = 0, at x = +L; therefore C = wL*/24. Hence

do _ 7) (= x <) 7.7]
dx Nee a As

The slopes at the ends of the beam are given by wL*/24E/ at A, where
x = 0, and —wL}/24EI at B, where x = L.
a 7 Le®
(= -t xt LD: “)+6
2EI \ 6 13. lps
At x = 0\-o = 0s therefore C,.— 0 and
4 3
ee Se
ey, (xyeeeo
abe ) [7.8]

The maximum deflection occurs at mid-span, where x = +L. Therefore

© DW i
bee ohDi Os - 5 wLt
ae Ha ime ko) gi 7 484 ET ee]

Simply supported beam The simply supported beam shown in Fig. 7.3 is acted on by a force system
with uniform bending which sets up the moment, M, at each end. The bending moment at any
* moment point, D, along the beam is M, and with the previous sign convention,

dv s
EI—=—M 7.10
dx? |
Therefore
d a
Ble 2 Ma JG
dx
and
Mx?
Ev = a a
188 MECHANICS OF ENGINEERING MATERIALS

Fig. 7.3 on aE |
(a) ¢ mo
oy es
\~ ie ——+ |

- A D Bike,=3
(b) M=Fd M= Fd
—— ye

A | D B
(c) pea A ee a

| \\_— Deflected shape

At x=0, v=0; therefore C; =0; and at x=L, v=0; therefore

C= 5ML, and the slope dv/dx is given by

ies (-1ir+ 4) [7.11]

from which the slopes at the ends are +ML/2EI at A, -ML/2EI at B and
at mid-span is zero. The deflection at any point is given by

1 Mx? ML
=F ( D + , *) [7.12]

The maximum deflection occurs at mid-span, and hence

Umax
_ ML
‘id SET 7.13
[ ¥ ]

Cantilever with The bending moment at D in Fig. 7.4 is

uniformly distributed é;
loading M = Ryx ~My =>

where Ry=wl and My, =tpl’, the fixing moment at the support.
Therefore

dv wl? wx?
EI 12 = —M = —wLx ++ —3
wLx 7 [7.14]

dv wL wL? wx?
EI—= j
Gs 2a)
At x = 0, dv/dx = 0; therefore C = 0. Hence

de ey mL 2 , whi 4 Be
dP een\e 2° a Boa 26 fo
At the free end, x = L. Therefore
L
Slope at free end = 28 [7.16]
6ET
 BENDING: SLOPE AND DEFLECTION 189

Fig. 7.4 w/unit length

(a)

(c) ales; SON el f

Deflected shape

1 wL Dies wxt\
=F aes aaa + Cy

and ati — 0, 7 =. 0; therefore Cy = 05 Hence

1 Wipe ple ese ua

The deflection at the free end B, where x = J, is given by

si! mL! whi, wht _1 2

UB [7.18]
eo iy 6 4 LO eee |
The reader may wish to check that this value for the deflection at the free
end may also be obtained by analysing in the same way the cantilever in the
configuration shown in Fig. 1.31 (i.e. fixed at the right-hand end rather than
the left).

Cantilever carrying a The bending moment at D in Fig. 7.5 is given by

concentrated load Mea

where M,= WI, the fixing moment at the support, and Ry=W.
‘Therefore

M = —WI1+ Wx

dv
El 3 = —M = WI — Wx [7.19]

and

dv .
Wx?
EI— = Wix -——+C
dx 2 2 zu
190 MECHANICS OF ENGINEERING MATERIALS

eee
Fig. 7.5
Ww
| S

{i >|

D B 7

A |D

heat B i
(c) =

Deflected shape — |
S—=
E

At y= 0, du/dx = 0; therefore C = 0, Hence

dv Whe Wie 7.20)

dx EI 2EI
ANE = Uh

dv __ we 17.21]
dx 2EI

Se Wix? Wx’ fe
Pn Gli
andat «== 0, v = 0; therefore C; = 0; and

W x”
= —— |(1
ROE) le’ -—7) 122
ie
which is the equation for the deflection curve for the beam.
For the deflection under the load we substitute x = / in eqn. [7.22] and

ee [7.23]
Either from eqn. [7.5] or from Fig. 7.5(c) and noting the fact that the
moment and curvature are zero from B to E, it may be seen that

Deflection at free end E = deflection at B + (slope at B) x (LZ — /)

Therefore

1we we
VE
. ED Ory

WP i!
= ORI ) i 3 [7.24]
 BENDING: SLOPE AND DEFLECTION 191

Cantilever with The slope and deflection under the load are obtained by substituting L for /
concentrated load at. 1™ eqns. [7.21] and [7.23], and thus
free end ol WL : : WI mh

ane) ees ~ 2EI

Example 7.1
A horizontal beam AB, simply supported at each end, carries a load which increases,
at a uniform rate, from zero at one end. Determine the position of, and the value of,
the maximum deflection.

xX” WwW
j Lw

A pes ae
(6; tB
x
R, 2 Ro

(a) (b)

Fig. 7.6

Let m be the intensity of loading at unit distance from A; then the intensity
at C, Fig. 7.6, is xm and that at B is Lm.
Taking moments about B,
L

] (=)
Bending moment at C = Rix — = 2

tendeve Lwx x3p

dx? 6 6

dv wx xtp
E[— =— G
A (nome
VO aw
EIv — 36 faa 120 + Cx+C
Cx 1

The boundary conditions are, where x = 0, v = 0; therefore C; = 0. And at

i= Lv == 0; theretore

Dw Ly
sy fade slkorer gt
De aee ag
192 MECHANICS OF ENGINEERING MATERIALS

4
moe end
360
and

Fe = L’p 7 xp i! to
dy fos 4-65
]
=a (—30L? wx’ + 15x4w + 7L*p)

At maximum deflection dv/dx = 0. Therefore

, 3017 + /(900L4 — 420L*) 302? — 21.917

a 30 i 30
= 0.27’
so that

= ae (—10L?x? + 3x° + 7L*x)

Omar a
=en
DW
[—10L?(0.52L)°
2 3}
+ 3(0.52L)*yt 5
+ 4
Ost ]
7L4(0.52L)]

Dw
= 0.00654 sy
——

le 7.2
examiye A beam 4m long is simply supported at its ends and carries a varying distributed
load over the whole span. The equation to the loading curve is w = ax* + bx + c,
where w is the load intensity in kN/m, at a distance x along the beam, measured
from an origin at the left-hand support, and a, b and c are constants. The load
intensity is zero at each end of the beam and reaches a maximum value of 100 kN/m
at the centre of the span. Calculate the slope of the beam at each support and the
deflection at the centre. E = 208 GN/m, | = 405 x 10-6 m‘.

The loading conditions are such that at xv= 0, m = 0; therefore c = 0. Also,

at v4 — 0; cherefore

0 = l6a+ 4b and b = —4a

At x = 2, w = 100; therefore

100 = 4a + 25 and 100 = 4a — 8a

 BENDING: SLOPE AND DEFLECTION 193

Hence

a=-—25 and b = 100

Therefore

Loading distribution, » = —25x” + 100xkN/m

4 4
Total load on beam = |w dx = |(—25x? + 100x)dx = 267kN
0 0

Therefore the support reactions are each 133.5kN, by symmetry.

The shear-force distribution is given by

3 100x?
O= — |max = = |{=2547 + 100x)dx = Hs = oe

Atw05 O = cRip— 135-5: therefore 24 ===" 133.5:

25x3 —100x?
Bending moment, M = [oa = (+ = ~ = + 1335)dx

25x* 100x°
=4 7) 6 pals:
Se ese B

At = 0) vie OS teLeloOlens =".

l I 25a" 100
NOD a al
Slope a i aleee
D 6 133.5x
133 sn)|dx

“( Zoe OU S257 )
+C
Sh een 2
At 2 b> ieretore.Cs==" 21.3.
When x = 0 and 4m, @ = +0.002 53 rad.
The deflection is given by

1 | 25x° 100x* 133.5x2

v=7,| ear = \( 60 — 74 ° +213 a

o( 25x° 100%? 133.5x°

= 21
EI \” 360’ 120 ii 30+ D)
Atx = 0) 0.= 0: therefore D —.0. Atsmid-span, 7 = 2 and v= 3.2mm.

~ 7.3 Discontinuous
When considering the bending-moment distribution for a beam with
loading: Macaulay’s discontinuous loading (e.g. Fig. 1.33 or Fig. 6.1), a separate bending-
method moment expression has to be written for each part of the beam. This means
that in deriving slope and deflection a double integration would have to be
performed on each bending-moment expression and two constants would
result for each section of the beam. A further example of discontinuous
loading is shown in Fig. 7.7(a); in this case there would be three bending-
194 MECHANICS OF ENGINEERING MATERIALS

Fig. 7.7

(a)

moment equations and thus six constants of integration. There are

apparently only two boundary conditions, those of zero deflection at each
end. However, at the points of discontinuity, B and C, both slope and
deflection must be continuous from one section to the next, so that

) d
At B (S:) = (=) and VAB = UBC
dx) 4p dx} pe

AtC =) = (=) and UBC = UCD

dx] ac dx/ cp

The above four conditions together with the two conditions of zero
displacement at each end enable the six constants of integration to be
determined. The derivation of the deflection curve by the above approach is
rather tedious; it is therefore an advantage to use the mathematical technique
termed a step function, commonly known as Macaulay’s method when
applied to beam solutions. This approach requires one bending-moment
expression to be written down for a point close to the right-hand end to
cover the bending-moment conditions for the whole length of beam, and
hence, on integration, only two unknown constants have to be determined.
The step function is a function of x of the form f,(x)=|« — a]” such that
for x < a, f,(x) = 0 and for x > a, f,(x) = (x — a)”. Note the change in the
form of brackets used: the square brackets are particularly chosen to indicate
the use of a step function, the curved brackets representing normal
mathematical procedure. The important features when using the step
function in analysis are that, if on substitution of a value for x the quantity
inside the square brackets becomes negative, it is omitted from further
analysis. Square bracket terms must be integrated in such a way was to
preserve the identity of the bracket, i.e.
J[x —a -de= t [x — a]?
Also, for mathematical continuity, distributed loading which does not
extend to the right-hand end, as in Fig. 7.7(a), must be arranged to continue
to x = /, whether starting from « = 0 or x = a. This may be effected by the
superposition of loadings which cancel each other in the required portions of
the beam as shown in Fig. 7.7(d).
An applied couple Mo must be expressed as a step function in the form
Mo|x — a|® so that the bracket can be integrated correctly.
The three common step functions for bending moment are shown in
Fig. 7.8 and several illustrative examples now follow.

Beam simply supported Taking moments about one end, the reactions at the supports in Fig. 7.9 are
with concentrated load
i ueZ = and R= ug
L
 BENDING: SLOPE AND DEFLECTION 195

Fig. 7.8
Mp M= = Mo)[x- _ a] gj

(a) ean oe ae
X=a
W a]*
M= WIx-

(>) —ERe
ers ras Sey x=a

Fig. 7.9

Vv

D
Deflected shape

When x < a the moment at any distance x from the left end of the beam is
given by M = Rx.
When « > a the moment is given by

M = Rix — W(x —a)

Using the step function concept to write one equation for the whole beam,
then
M = Rix — W(x —a|
Note that when x < a, the quantity in the square brackets would become
negative and so W|x — a] is taken as zero. This leaves M = Rx as above.
Using the step function moment expression in eqn. [7.3]
dv
EITa= M= —R\x
+ W[x -a| [7.26]

Integrating eqn. [7.26],

dv Rix W
EI— ie = 5 elt 2
ayr+C Fee,
[7.27]

and

Big + —[x-aP+Cr+C, [7.28]

196 MECHANICS OF ENGINEERING MATERIALS

If we omit the term inside the square brackets on the right-hand side of
eqns. [7.27] and [7.28] when wx < a, the equations are then of the correct
form for the portion AE of the beam, and since the second term on the
right-hand side of these equations vanishes for a value of x= a, then when
these equations are used for the whole beam, both dv/dx and v will be
continuous at the point E.
When x = 0, v = 0, and since the term inside the square brackets is
omitted, Cy = 0: For x= 1, v= 0; therefore

RD WwW
) = 6 + 6 (|BG ayARele
+ Gy

and

RL Ww :
= it
Ga Se eo
3 W(L mesiiy:
—a) W 7 "a a)
6 6L

Wa
= — Wb 7.29

Substituting the values of C and R in egn. [7.28] and rearranging,

Wx L—a , os)
vb) S i) Peel
+4.

This equation gives the deflection at any point along the beam if the last
term on the right-hand side is rejected when it becomes negative, 1.e. for
x <a. For the particular case when x = a, the deflection under the load is
given by

_ Wa?(L—a)
VE = 3EIL [7.31]

It should be noted that the maximum deflection occurs where dv/dx = 0

which is not, in fact, where the load is applied.
If W is placed at mid-span so that a = +L, the deflection under the load
is also the maximum deflection and is

WL
i
48EI
[7.32]

Beam with distributed As explained above (Fig. 7.7(b)), for mathematical continuity the loading
load on part of the span must be continued to the right-hand end, and to maintain equilibrium,
upward loading must be inserted from D to B as shown in Fig. 7.10.
At point E between D and B,

M = Ryx ae al? + sls (a +6)/? [7.33]

 BENDING: SLOPE AND DEFLECTION 197

Fig. 7.10

Deflected shape ==nih

the second and third terms being rejected when x < a and the third term
when x < (a +4).
dv DW 7)
12
EI—~=-—R (e+ ale a| “ lai (a+)|
+ b) |

and
dv Rix? pw. WD.
oe
1 ee aes a3 eer
1 pp 4 eC 734
7.34

Eien
O= cai ais
74 a PRE,
74" (a+6)|'+Cr+C
a | x 1

[7.35]
The values of C and C are found from the conditions of v= 0 when x = 0
and x = TL, the terms inside the square brackets being rejected when
negative.

Beam simply supported From moment equilibrium in Fig. 7.11,

with an applied moment i
Beak, =
Wi
The bending moment at D, where x > a, is
_ MM E
Mp = Rix —-M=—x -—-M
L
A convenient way of dealing with a couple by Macaulay’s method is to
introduce a term [x — a|? which is in fact unity, but allows for subsequent
integration in the correct manner:
oe M . Mt i
Se a Re
dx? oat
198 MECHANICS OF ENGINEERING MATERIALS

Fig. 7.11

where the second term is integrated with respect to (v — a). Therefore

Elv =
io
= ait al? + Cx+C,
:
[7.36]

When x < a, the square-bracketed term on the right-hand side of the

equation becomes negative and is rejected.
At x =0, 0 = 0; therefore C; = 0. And at x= L, o= 0; therefore
ML? M
0= 6 iI al“2 +C Ve

and

M
C =—(-2L’ + 6aL — 3a’) [7.37]
6L
Hence

_ M M
eer ( ein 3 + at a| 2 ar (2L° 2 — 6aL + 3a 2 ») [7.38]
7

At E, where « = a, the deflection is given by

M M
Elvg UR = 6L a : ral AG: 6aL ++ 3a*\a
3a?

or, putting L = (a+ d),

M
Ly = {a> — a> — 2ab(a — 5)
 BENDING: SLOPE AND DEFLECTION 199

M (a—b)ab
as pales Nae iets 7.39
PT) eit, Ue
EXRMMG 10 56s ther ey tra ac Te cy arte wees). peat anit ee etre erg
P A simply supported beam is subjected to the loading shown in Fig. 7.12. Calculate
the deflection at a section 1.8m from the left-hand end. E — 70 GN/m?, |
= 832 cm‘.
16 kN
Fig. 7.12 1.5 | for kN

16 kKN/m 20 kN

(c)

Deflected shape

This example combines the features of the cases above, and so, to satisfy the
Macaulay conditions, the distributed load must be extended to B and an
equivalent negative load inserted to restore the correct resultant load
distribution. Then

(= 5h
M = Ryx —3[x — 1.5] — 16 a 20[x — 2.4]
Z
eas fl
ee
16——————_

M = 0 when x = 3; therefore R4 =10kN.

dv ‘10x’ l6fx —1.5)° 20[x — 2.4)

i dy eesD + 3[x Pie 5] + G AF D

16[x — 2.4]°
= ; = aE

l6[x—1.5)
[x
4
20[x —2.4 |3
reese
Z 24 6
200 MECHANICS OF ENGINEERING MATERIALS

When x = 0, v = 0; therefore C; = 0. And when x = 3, v = 0; therefore

a 10(3)° 3(1.5)?_ 16(1.5)* 20(0.6)° ~—-16(0.6)*

3G
: ee SEE MI eek Ret gt
from which C = 12.54. When x = 1.8 the third and fourth square-
bracketed terms are omitted, and

10« 1.8 303? 16x03*

Elv ~ ere + (12.54 xx 1. 1.8)

= 13kN/m?

13: <10' x 10°

°= 70 x 10° x 832 x 10-8 eae
Example 7.4 eee
nF ee ; e
Calculate the position and magnitude of the maximum deflection for the beam
shown in Fig. 7.13. El = 1000 kNm7.
5 kN
Fig. 7.13
A |B G 2 kKN/m
rs

{ 2m | 2m | 4m
4 kN

(a)
% M

(ce E 2 kN/m
A B D

esd
4 kN
ae: R

(b)

Deflected shape e: ‘

(c)

a? 2
EI node =Fa Di Soe) be 2}+ 1a — 4;
dx?

d 4x* 5 2
j= ttle tetec

aoe SD 2
Ele
v 6 + Le pA
P+ ale = 4J°+Ay" Ce+C,
BENDING: SLOPE AND DEFLECTION 201

The boundary conditions are dv/dx =0 at x = 8, and v=0 at x = 8;

therefore

0 = ~(2 x 64) + (5x 18) + +€

C= +167

8° 16?
0=- wars + (5 x 36) +55 + (16.7 x 8) + Ci

C; = —6.6
At the left-hand end the deflection is obtained when x = 0; therefore
Elv = —6.6kNm?

6.6 x 10° Ae
i= 1000 = eo mm

This may not be the maximum deflection and we must check elsewhere
in the span. However, it is not sufficient merely to equate dv/dx to zero
since square-bracketed terms would then be included which might not be
appropriate, depending on where v4, occurred. The best way is to make a
sensible guess as to the section where the maximum deflection is likely to
occur and to determine the slope at each end of that section. The slopes will
be of opposite sign if the guess was correct. If not, then an adjacent section
must be treated in the same way. For example, assuming dv/dx = 0 occurs
between B and C, then

dv 22
AtB £E[—= 4x D + 16.7
= +8.7
dx

d 42 5
At Co | Byes ldpee =) See a gy an
dx 2 2

and the assumption was correct. Therefore, using the condition that zero
slope occurs between B and C,

Thus

x” — 20x + 53.4 =0
from which « = 3.17 m.
The deflection at this point is given by

17 5
Elv = (73 )+ (gxtat) +0627 x 3.17) -66

= +26.44kNm?
202 MECHANICS OF ENGINEERING MATERIALS

26.44 x 10°
a
v=t+ 1000 (y
+26.44 mm

Hence the maximum deflection occurs at 3.17 m from A and is downwards.

7.4 Superposition
The principle of superposition which was introduced in Chapter | states
method
that the effect of a given combined loading on a structure may be obtained
by determining separately the effects of the various loads and then
combining the results obtained. This type of superposition is only possible if
each effect is linearly proportional to the load which produces it. Also, the
deformation resulting from any given load should be small enough that it
does not affect the conditions of application of the other loads.
In the case of beam deflections, the principle of superposition can be
applied to give the total deflection of a beam which carries individual loads
W,, W2, W3, etc., or distributed loads m1, m2, 3, etc. Let the bending
moments at a section of the beam caused by each load when acting separately
on the beam be M, M2, M3, etc., and the corresponding deflections be 7),
V2, V3, etc. Then the total bending moment is

Ma Mpa Ma [7.40]

But
2
Me =EIc

Therefore

Ss ~ ay](fata) as
ET

| (fn +My. +Ma+.. )ar)dv

= ZL |({anae)ae+ {({anas)ar[(anas)ae+
=) ap OR Se OR ar aoe [7.41]

Thus the deflection at a section of a beam subjected to complex loading can

be obtained by the summation of the deflections caused at that section by the
individual components of the loading.

Example 7.5
Use the principle of superposition to determine the deflections at the ends and
centre of the beam shown in Fig. 7.14. Ei — 500 kN/m7.

This problem may be split into three components as shown in Fig. 7.15(a),
(4) and (c). It should be noted that when breaking a problem down into
elements, it is essential that the same boundary conditions are employed in
each case. For example, it would be incorrect to utilize a solution for beam
 BENDING: SLOPE AND DEFLECTION 203

Fig. 7.14

(a)

Fig. 7.15 deflection from Table 7.1 (at the end of the chapter) which did not use
simple supports at B and C.
The respective deflections for cases (a), (4) and (c) are:

we _ 2000 x 6210:
o LSHEnE asp s00SaIn 8
a 6 == =e — ]

sit 5 x 1000 x 6¢ x 10°

> 384E1 rs a
(c) This may be treated as a beam subjected to couples at B and C of
8kNm magnitude:

; MP 8000 x 6? x 10° ,
*
—
BET =
=
8 x 500 x 103
= mm

Resultant deflection = +18 + 33.8 — 72 = —20.2 mm

To find the deflection at A or D it is necessary to know the slope in each
case at B or C. Then
wp 2000 <6 <2 10°
(a) 14 = Oiplag = Fea lB 16 x 500 x 103

B 1000 x 63 x 2 x 103
O)b 4 = O r p l i n =
ba = Oaplan =77 ppleSPE lay= 24 x 500 x 103
= —36mm

(c)
¢ 634 == 93plap
O35 + Walp
:
BEI

— MIpclap Walkin
DIET 3EL

— 108 (8000 x 6 x 2, 4000 x 8

~ 500 x 108 2 | 3

= T7smm
204 MECHANICS OF ENGINEERING MATERIALS

Resultant deflection at A or D = —18 — 364 117.3

= +63.3 mm

7.5 Deflections due to

In asymmetrical bending, as in symmetrical bending, the deflection will
asymmetrical bending
occur perpendicular to the neutral plane. At any section along the beam the
deflection may be calculated using the deflection formulae developed in this
chapter. For example, for a cantilever of length L, carrying a load W at the
free end, the maximum deflection of the centroid of the cross-section will be
given by

— Wald
— 3EIna
where Wy,4 is the component of the load perpendicular to the neutral axis
and Jj\4 is the second moment of area about the neutral axis.
The most convenient way to obtain /,;4 is to use the co-ordinates /,, [,.
and J,, —J,, to construct a Mohr’s circle for moments of area as illustrated
in Appendix A. This then allows the second moment of area at any angle to
the y- or z-direction to be determined.

Example 7.6
Calculate the maximum deflection of the centroid of the beam section shown in Fig.
6.27, if the beam is a cantilever of length 1m. Young’s modulus for the beam
material is 210 GN/m?.

The beam deflection at the free end is perpendicular to the neutral axis
(N.A.) shown in Fig. 6.28 and is given by

Re WnaL?
By
The solution to Example 6.11 shows that the direction of the neutral axis is
47.45° anticlockwise from the z-axis.
The vertical end load on the beam is 2kN, so

Wna = (2000 cos 47.45°) N

The second moment of area about the neutral axis is obtained from the
Mohr’s circle construction shown in Fig. 7.16. This gives

Fig. 716 7, and /, are second Wa

(x 10° m4)
moments of area about the
0.45
principal axes (i.e. those were
Iz = 0)
 BENDING: SLOPE AND DEFLECTION 205

Iy4 = 0.46 x 10~° m‘; therefore

(2000 cos 47.45°)(1)°

= 4.57mm
~ 3x 210 x 109 x 0.47 x 10-6
This deflection is downwards to the right perpendicular to the neutral axis.
The vertical and horizontal deflections of the centroid are given by

Vertical deflection = 4.57 cos 47.45° = 3.1mm |

Horizontal deflection = 4.57 sin 47.45° = 3.4mm —

7.6 Beams of uniform

All of the beams analysed so far have had a uniform cross-section. This 1s
strength
convenient for manufacturing purposes and the dimensions of the cross-
section will have been chosen on the basis that the maximum stress in the
critical sections of the beam should equal the design stress for the material.
As a consequence of this, the non-critical sections of the beam will be
overdesigned, i.e. they will have too much material for the stresses they are
carrying. A more economical design would be one in which the stress is the
same at all sections. This would effectively be a beam of uniform strength.
The stress at any section in a beam may be obtained from the bending
equation g = My/TJ. The condition, therefore, of uniform strength is that
My/T shall be constant, 1.e. the section modulus shall be proportional to the
bending moment. The value of //y may be varied, in the case of rectangular
beams, by altering the depth or altering the breadth.

Beam having constant

breadth
pee ey My, M
I I Ipp?
rbD; Ipz2
ibd
Therefore

Dy wien
and
a M
M MD, M a
Ie Ted, M
where D, is the depth at mid-span and d the depth at a point distance L
from the support. In the case of a beam resting on supports, then, as before,
considering one-half of the span,
L/2 M
Co) = | —xd
jie x

MP/? pL/2
= =| Mo"? xdx [7.42]
EN, Jo
For a concentrated load at mid-span, M = 5Wx and M, = iWL. Therefore

_ WE
Tie
206 MECHANICS OF ENGINEERING MATERIALS

Other cases may be solved by substituting the value of M in egn. [7.42]

and integrating. Some examples of beams of uniform strength are shown in
Figs. 17:

Fig. 7.17 Uniform strength beam

(a) (b)

Table 7.1 Principal slope and

deflection for beams with basic Slope Dae
loading #3 MP 3

DET

(6) 12ET
Ug el/2 | Cat G
MI
——atA
Tey ens

(c) ES t A,B ss Cc
a ea * ger™
2 3
(d) + appt B Fonte

WP WwP
+—at A,B
(¢) ich Ti

(/) 0 ahi a, 18}, 1G on

=
ia

w/unit length wD wit

(g) ; : pert
=a +o5ytB
ae

w/unit length wi Swit

Gy oA +——at A, B
: 24ET* 38461" ©
w/unit lengt ee 4
(i) 2 0 at A, B,C Bale ye
 BENDING: SLOPE AND DEFLECTION 207

7.7 Summary
All solutions for the slope and deflection of beams depend on the
relationship between bending-moment distribution and curvature (eqn.
[7.3]). It is therefore vital that the correct bending-moment expression can
be stated. Although the double integration is relatively simple, the constants
of integration can only be found by applying the correct boundary
conditions at the supports. Equally, the successful application of Macaulay’s
method for loading discontinuities does depend on following the simple
rules associated with the use of step functions. The superposition method is
a valuable alternative to the double-integration method and choice depends
on the nature of the problem. For example, it is obviously an advantage to
use superposition if the case can be broken down into simple elements the
solutions for which are readily available and can then be superposed. A final
reminder may be made of the need to recognize the statically determinate
nature of the problem; if this is not the case, then the treatment given in
Chapter 8 is required. The principal values for the slope and deflection for
various basic support and loading conditions are given in Table 7.1.

Problems 7.1. A cantilevered deck is built in at the left end and is supported on a
wall at 8m from the end. The deck extends a further 4m beyond this
wall. The loading on each of the beams supporting the deck including
self-weight is 20kN/m and the flexural rigidity (EJ) of each beam
section is 200 MNm/’. Determine the level of the top of the wall
(assumed rigid) relative to the fixed support so that the bending
moment is zero at that support.
7.2 A horizontal beam is subjected to the loading shown in Fig. 7.18.
Fig. 7.18 Calculate the beam deflection under the 8kN force. The flexural
rigidity (EJ) of the beam is 1 MNm/?.

2 kN 8 kN

| 6 kN/m |

aeletia) imi

7.3. A beam 10m long is simply supported at each end and carries the
loading shown in Fig. 7.19. Calculate the position and the magnitude
of the maximum deflection. The flexural rigidity (£7) of the beam is
100 MNm?.
Fig. 7.19 10 kN

ea 1 o
208 MECHANICS OF ENGINEERING MATERIALS

5 7.4 A cantilever beam is subjected to a point load Q at its tip, as shown in

Fig. 7.20. Create a spreadsheet to calculate the deflection of the beam
at 10 points equally spaced along its length. Provide cells in which the
| : length L, second moment of area J and material modulus E can be
Fig. 7.20 Q entered. Create an x—y graph of the negative of deflection, —v, versus
distance along the beam, x.
Bee! yM 75 Extend the spreadsheet of Problem 7.4 so that a moment M can also
be applied to the tip and the deformed shape for the superimposed
moment and shear force is calculated (Fig. 7.21). Observe how the
Fig. 7.21
deformed shape changes as different bending moments and shear
forces are applied and compare the calculated shapes with the
behaviour of a flexible ruler, (Iry BE =f =Je- = to A0,
AV='6):
7.6 A bookshelf is to be supported by two wall brackets positioned
symmetrically (Fig. 7.22). Where should the brackets be positioned if:
(a) the maximum stress in the shelf is to be minimized; (4) the largest
deflection of the shelf is to be minimized.
Note that the deflection at both the end and the centre will need to
Fig. 7.22 be checked.
A spreadsheet can be used to evaluate moment and deflection for a
range of values of a/b.
ah Calculate the deflection of the roof of the petrol station in Problem
6.14 at points A, B and C.
7.8 A flexible mounting pad 3m in length and of flexural stiffness
0.55 MNm? is simply-supported at one end and rests on a spring of
stiffness 100 N/mm at the other end. Determine the overall
displacement of a load of 10kN placed on the pad at 2m from the
spring-supported end.
79 The plastic clip shown in Fig. 7.23 is injection moulded from a nylon
material with modulus 2000 MN/m?. The maximum allowable stress
in the material is 100 MN/m”. Each half of the clip must support an
axial force P of 250N, which may be considered to act at the lip
shown. The design constraints are:

Q
Fig. 7.23 | 3mm

P< Sakte

Beam model
 BENDING: SLOPE AND DEFLECTION 209

(a) the force Q to open the clip should not exceed 25 N

(6) the allowable stress should not be exceeded when: (i) the
maximum axial force P is applied at the point shown; (ii) the clip
arm is pushed fully open against the stop in the centre of the clip.
Set up a spreadsheet to calculate the opening force and the stresses for
any set of dimensions and load. Confirm that the clip shown in Fig.
7.23 meets the design requirements.
Note that the spreadsheet optimization facilities could be used to
find the values of b and / which give the minimum volume of this
material.
7.10 Part of a bridge is being assembled as shown in Fig. 7.24. The central
section is hanging from a crane and is to be bolted to the left- and
right-hand cantilever sections at A and B. In raising the section into
position and before it is properly aligned it fouls at A and B, causing
an upward force W exerted by the crane, on the beams. Show that the
deflection of the beam at the crane hook, if E and / are the same for

Fig. 7.24 A | B

~< iu | Py | -|- l ~|- 21 >

each of the beam sections, is

Tet A sight-screen for a cricket field is illustrated in Fig. 7.25. The cross-
beams AB and CD can be assumed to be rigid and act as simple
supports to the screen itself. When the screen is subjected to uniform
wind pressure the location of the cross-beams is to be such that the
horizontal deflections of the top and bottom edges are the same as the

Fig. 7.25
210 MECHANICS OF ENGINEERING MATERIALS

horizontal deflection of the centre line. Hence determine the

relationship between L and a.
fale A horizontal shaft 1m in length is simply-supported in bearings at
100mm from each end. It carries loads of 0.3kN at each end and
1.2 kN at the centre. Find the deflection under the loads and the slope
at the bearings. Check the solution by the method of superposition.
EI = 7kNm’.
Calculate the vertical and horizontal components of deflection at the
free end of the Z-section cantilever described in Problem 6.34.
E = 208 GN/m’.
A horizontal beam is simply-supported at each end of a 4m span. It
carries a distributed loading varying from 2kN/m at the left end to
3. kN/m at the right-hand end. Find the position and magnitude of the
maximum deflection. E = 208 GN/m?; J = 2 x 107+ m*.
7.15 A floor beam 6 m in length is simply supported at each end and carries
a varying distributed load of grain over the whole span. The loading
distribution is represented by the equation w = ax’ + bx + c, where w
is the load intensity of the grain in kN/m at a distance x along the
beam, and a, ) and ¢ are constants. The load intensity is zero at each
end and has its maximum value of 4kN/m at mid-span. Calculate
the slope at each end of the beam if E=208GN/m? and
C10 mm
G10 A horizontal beam of rectangular section 25 mm wide by 50mm deep
is 1m long and simply supported on (a) rigid rollers, (4) springs of
stiffness 300 kKN/m. A load of 15 kg falls 50 mm onto the beam at mid-
span. Calculate the instantaneous maximum deflections and bending
stresses. E = 208 GN/m/?.
 CHAPTER Ne
Statically Indeterminate Beams

The design of beams depends initially on the evaluation of shear-force and bending-
moment distributions in order to calculate stresses and deflections. A prerequisite is
the calculation of support reactions, and in the case of statically determinate beam
situations there are only two unknown reactions, which are found from the two
equilibrium equations (~M = 0 and oF = 0). Thus a beam which is supported in
such a way as to produce three or more reaction forces or moments is statically
indeterminate. Some typical examples are shown in Fig. 8.1. The principal methods
which are used for analysis are (i) double integration with Macaulay’s method,
(ii) superposition, (iii) moment—area. The application of these methods will be
illustrated in a number of worked examples.

(a) (b) (c)

Fig. 8.1

8.1 Double-integration
This method was first developed in Chapter 7. To reiterate briefly, the
method
curvature is expressed in terms of bending moment at any point along the
beam, this equation is then integrated twice, and the constants of integration
are found from known boundary conditions of slope and deflection at the
supports, or elsewhere. Whereas in the case of the statically determinate
beam the reactions could be found prior to the above procedure, this is not
possible for the indeterminate beam and the reactions must be carried
through the analysis as unknown quantities. Since there are a/ways enough
boundary conditions to determine all the unknowns in the equations, the
reactions can be evaluated together with the constants of integration. A
number of worked cases now follow.

Beam fixed at each end The problem is illustrated in Fig. 8.2, and it is evident that, owing to
with uniformly symmetry, M,= Mp, and R4=Rz=wL/2. However, we cannot
distributed loading determine M, and Mz from a moment equilibrium equation. The next
step is to write the equation for bending-moment distribution as a function
of x and equate this to EJ(d?v/dx”) (see eqn. [7.3]).
dv wLx wx
Ela MY) = 5 +My, - 5 [8.1]
212 MECHANICS OF ENGINEERING MATERIALS
a
ee

dv wLx? Dx: 2
EI a— = ———4 ++ Myx
ML4x + —6 +A [8.2

whe Myx? wx"

Elv lates Maar

Fig. 8.2

wl?
(c)

The boundary conditions are that when x = 0, dv/dx = 0 and v = 0 from

which, respectively, 4 = 0 and B = 0; and when x = L, dv/dx = 0 and
c—,
Either of these conditions can now be used to solve for M4, which is
found to be My = Mz = wl’ /12. The fact that M4 and Mz have worked
out positive confirms that the directions chosen for them in Fig. 8.2(d) were
correct.
At mid-span

sd wl
24
The deflection is a maximum at mid-span and is

A wLt
“max 384EL
The bending-moment diagram, Fig. 8.2, shows two points of contra-
flexure (where B.M. changes sign), occurring at x = 0.211Z and 0.789L.

Beam fixed at each end There are four unknown reactions illustrated in Fig. 8.3(4). Vertical force
carrying a point load equilibrium gives
 STATICALLY INDETERMINATE BEAMS 213

Taking moments about B gives

—Ryal+ W(L — a) +M,—M, =0 [8.5]

Fig. 8.3
|
|

(a) (b)

In this problem, when determining the moment at x from the left-hand side,
it is necessary to use the step function technique (Macaulay’s) as there is a
discontinuity of bending moment at the load, W.
do :
ee = —Ryx+ W{x-a|+My, 8.6
a

LAS
dv
=e
R yx W
[x-a]? + Myx +A (8.7
dx 2, Z

Ryx W 2
Ely
= — [ea ei eae (8.8
6 6
The boundary conditions are: (i) when x =0, v=0 and dv/dx
= 0;
(ii) when «=, v= Oand dvo/dx= 0.
From (i)

PA) and b=

From (ii) and solving for Ry and My,

73

Wa
My => ae Ce = a)

Using eqn. [8.5],

Wa?
Mp = Fs (Ee = a)

From eqn. [8.8] the deflection under the load is

Wa} (L — a)?
SS
° 3EILSa 8.9
a
For the particular case when a = L/2,
W WL
kK7= R= and My, = Mz = —
2 8

L Wi
[8.10]
214 MECHANICS OF ENGINEERING MATERIALS

Cantilever with a prop _ This situation is illustrated in Fig. 8.4, in which the prop, considered as a
at the free end simple support, is, for generality, assumed to be at a level A above the fixed
end. The unknown reactions are P, Rg and Mz.

Fig. 8.4
Me

7
Rg

Vertical force equilibrium gives

= he aU [8.11]

Taking moments about B gives

WL

Using the step function method to deal with the discontinuous loading as
in Chapter 7,

d’v 1g

dv Px? W Ly
= 5 he 5 E s|-+4 [8.14]

Px W Lie
v
Eb= H+. x— 5|-+Axwe +B [8.15]
:

The boundary conditions are: (1) when x = 0, v = —A; (ii) when x = L,

v = 0 and dv/dx= 0.
For (i), from egn. [8.15]
B=—EIA

(The term in square brackets has to be omitted as it is negative.)

For (i), from egn. [8.14],

— peer F dis aia 8.1

= 2 Ck tS
and from eqn. [8.15],

PLOW
O=—-—— +4, + AL — EIA [8.17]

Solving eqns. [8.16] and [8.17],

5 3EIA
=16 a
[8.18]
 STATICALLY INDETERMINATE BEAMS 215

and

ee WL? , SEA
2 jab
Substituting for P in eqns. [8.11] and [8.12],
11 3EIA
Rp =—W -——
2 6 B
and

3 3EIA
Mz = ib
orate game?
Any required S.F., B.M., slope or deflection can now be determined.

Continuous beam on _A fairly general example is illustrated in Fig. 8.5 in which Ag and Ac are
multiple simple known displacements due to the supports not being at the same level.
supports
R4—Rp-—Ro-—-Rp+W+wml=0 [8.19]

Fig. 8.5

5 LE
—3R4L — 2Rpl — Re +5WL+—-=0 [8.20]

d?v ie
EI Ryx
+ W |x —=|— Ralx — L] — Rc[x — 21]
dx? 2

+S Sey [8.21]

dv R gx? -| Ab Rp 2 Rc 2
jee 6 YE
ae el a pot as ara t)

fe a ely ad [8.22]

Ryx? id s|: Rp SRE ;

Vip Ly aA
fee Oe ee eee
W 4
+—[x—2L]'+Axn+B [8.23]
24
216 MECHANICS OF ENGINEERING MATERIALS

The boundary conditions are: (i) x =0, v=0; (ii) x =L, v= —Agz;
Gil) PSL = Aa wy r= sho HC:
From (i),

B=0
From (ii),

= WL? 7 Ry4l? EIA,

48 6 E
From (ii),

Fike = Rul oe a se | 2EIA,

From (iv),

0 = —4R,1* + 122WL? 8Rel* Rel’ wt 3EIA,

48 6 6 aa
Although perhaps somewhat laborious, these latter two equations
together with eqns. [8.19] and [8.20] can be solved to give values for Ry,
Rpg, Rc and Rp. From this stage any required aspect of this beam problem
can be evaluated.

52 Sipe. 2.
e e
method The principle of superposition can be very useful in finding redundant
reactions, particular if a problem can be split up into ‘standard’ cases
(see Table 7.1).

Cantilever with a prop’ This problem is illustrated in Fig. 8.6 and can be represented by
at the free end superposition of the two parts shown in Fig. 8.7(a) and (0). If there were
no support at A there would be a downward deflection due to the
distribution load. If there were no distributed load and the reaction, P, at the
w/unit length
support was considered as a force which could cause an upward deflection of
the beam, then the necessary boundary condition is that the sum of these
deflections must be zero.
Due to loading, » P

aa Bis

But

Fig. 8.7
 STATICALLY INDETERMINATE BEAMS 217

Therefore

Ab WE
Neosat 5s 8.24
SEL BEL bee
Hence
3

From vertical and moment equilibrium,

5 ee
Rp = Rae and Mp = a

Alternative ‘The case of Fig. 8.6 can also be split up in the manner shown in Fig. 8.8(a)
superposition and (4). The superposition must now satisfy the boundary condition of zero
slope at B. Thus

G, + Oy = 0
Now,

ey wL3 poe ae
nary) Wate eee ay
Therefore

pL? Mel
= ee 8.25
24ET a 3EI pes)

w/unit length

wl
Nay te
Knowing Mz we can find P and Rg from the two equilibrium equations.

Beam fixed horizontally Before considering any particular form of applied loading, we will examine
at each end _ the effect of the fixing moments alone. Taking the general case of, say,
Mz > M, as illustrated in Fig. 8.9(a), this may itself be put into the two
parts in Fig. 8.9(4) and (c), giving slopes at each end of
MyL ML
4 ET we By ay
Met ANL Mp — My)L
O04 = ( z A) and Op = ie
e A)
6ET ero" f
218 MECHANICS OF ENGINEERING MATERIALS

Fig. 8.9 R| We M, i, R Mo-Ms

ae
(
(a)
p= |
a
(b)
d+
eral
(c)
)

The resultant slopes are therefore

(Mp +2Ma)L pcg 2Mzg

A + M,4)L
A)
= — [8.26]
6ET 6ET

Fixed beam with This problem may be represented as in Fig. 8.10. The slopes at the ends for
uniformly distributed the simply supported part are +wL/24EI. Using the condition of zero
loading _ slope at each end and the results in eqn. [8.26],

wl (Mp a= 2M 4)L a
0 [8.27]
24E] 6EI a
and

wl (2Mp Sie My)L

=0 [8.28]
~oRRT ” 6EI

Fig. 8.10

Fixed beam with ends This is an important structural situation since a considerable bending
not at same level moment can be set up by the ends not being at the same level, even without
any applied loading to the beam.
Let v be the difference in level between the ends of the beam, Fig. 8.11.
A point of contraflexure occurs at mid-span, owing to symmetry, and each
half of the beam may be taken as a cantilever, the free end of which is caused
to deflect 5v by a force P at the free end. Then, using the basic solution for a
cantilever carrying a load at the free end,

12EIv
eo El so that P= BR
 STATICALLY INDETERMINATE BEAMS 219

Fig. 8.11

The bending-moment diagram on each half, due to P, is triangular, the

maximum ordinate being
EIo L 6Elv
My = Mp si Zou
‘ “ (oReD ie
as shown in Fig. 8.11.
If, say, a distributed load m is now applied to the beam, then if the ends
were at the same level the bending-moment diagram would be as shown in
Fig. 8.12.

Fig. 8.12

By superposition of the bending moments for the fixing moments and

for the distributed loading, the total bending moment distribution may be
obtained.

wl? i 6EIv A wl? 6EIv

My = a an Se [8.29]
Do PF Pfs tee
The combined diagram of bending moment, due to the distributed load
and to the fixing of the ends, is shown in Fig. 8.13.

Fig. 8.13

ike 2 Gaihy if
2 _Wwe
ne 24 f - wi?
,Sev
We |EB
220 MECHANICS OF ENGINEERING MATERIALS

8.3. Moment-—area
In the previous sections, the slope and deflection of a loaded beam have been
method
obtained using mathematical methods based on a knowledge of the variation
of the bending moment along the beam. It is also possible to obtain the slope
and deflection of a loaded beam by examining the geometric properties of
the elastic curve of the deflected beam.
It will be shown that the change in slope or deflection between two
points on a beam is related to the area under the bending-moment diagram
for that section of the beam. This approach is referred to as the moment—area
method. It may be used to determine the slope and deflection of statically
determinate or statically indeterminate beams.

Slope related to area of In Fig. 8.14 a portion of length AB of a beam has a bending-moment
bending-moment diagram of area A represented by CDEF. The distance of the centre of area
diagram G of the diagram from any chosen reference line HH is x. An exaggerated
view of the deflected beam is shown below the bending-moment diagram.

Fig. 8.14

(a) Loading

Reference
line (b) Bending-moment diagram

(c) Deflected shape

Consider a small piece of the beam of length dx over which the bending
moment may be assumed to be constant and equal to M. The change of slope
over the small piece dw is given by d@, where dé is the small angle included
between tangents drawn at each extremity of dx. Let R be the radius of
curvature of the small length dv when deflected; now dé is also the angle
subtended at the centre of curvature of the element dx. Therefore
Rdéd =dx
 STATICALLY INDETERMINATE BEAMS 221

and substituting for R we get

]
dé = — M dx
Ei:
and integrating between the limits Z and /, the distances from the chosen
reference line HH,
Yb l 1 1m
=| —Mdxv=—] Md 8.29
|Eee 7 3 ee
when the cross-section of the beam is constant.
Now if M dx is the area A of the bending-moment diagram CDEF over
the length AB. Therefore

0 ofPe 8.295
[8.295]

? being the change of slope over the length AB.

Thus we have the important relationship that, over any portion of a
loaded beam, the change of slope is equal to the area of the bending-moment
diagram divided by E/. If the beam varies in cross-section, then / must be
retained within the integral in eqn. [8.29a].

Deflection related to The next stage is to develop an expression for the deflection of the portion
area of bending- of beam AB. The intercept dv on HH can be represented as
moment diagram
M
dom xd = x dx

Therefore

1 Ih

al Mx dx

: z E
where J is constant. Now if Mx dx is the first moment of the area 4 of the
bending-moment diagram on AB about the axis HH; and since ¥ is the
distance of the centre of area of A from HH, then

peso [8.29¢]

Thus the distance between the intercepts, on any chosen reference line,
of the tangents drawn to the ends of any portion of a loaded beam is equal to
the product of the area of the bending-moment diagram, over that portion of
the beam, and the distance of the centre of area of this diagram from the
reference line, divided by EJ.
In Fig. 8.15, ACB is the deflected form of a loaded beam of length /
greatly exaggerated and we wish to find the deflection of any arbitrary point
P. The slopes at A and B are 64 and @3 respectively.
The reference intercept lines are AK and BL at each end of the beam,
and FE is a horizontal tangent at C.
222 MECHANICS OF ENGINEERING MATERIALS
—— ee _____

Fig. 8.15

(a)

Rasen 1%
~ x ~

Dyas Ase ded VAN

ill Seer
Fs area Meigenere
fee Lh —— Deflected shape

es |
E
=] va 4
o) 8 We
5 zs
3
5 yy

Kb

The deflection at P is vp, where vp = @ — >. Now

s _ ApnXp
=
EI
where Apy is the area of the bending moment diagram on PN and
v= PN - A,

ApnXp
ee Up = ———
EI — PN - 6,:

since 64 + 6, = Op. Therefore 0, = Op — 04, and

ApnXp
up = EI (Op —— 04)
— PN(6@p

Now

A
064=0-0;
A =—-—
are 98

and
Ae
ye
EI
therefore
A. Ax A
Q = — —_ x
45 Br Ee
 STATICALLY INDETERMINATE BEAMS 223

Hence

_ Apnk&p ENAPNe A(l

— x)
ers Br Ell
The application of this method will now be illustrated using some
worked examples.

Fixed beam with In the case of a horizontally fixed beam, 0 = 0 for the complete span, and
irregular loading since the product E£/ is not zero, it follows from eqn. [8.29b] that A, the
resultant area of the moment diagram for the beam, must be zero. This
enables the deflection and slope at any point on the beam to be evaluated.
Referring to Fig. 8.16 let 4; be the area of the ‘free’ bending-moment
diagram AEFHB due to the distributed load on simple supports and let x;
be the distance of its centroid from A; also let 47 be the area of the moment
diagram ACDB due to the ‘fixing’ moments only, and %2 the distance of its
centroid from A. Then

(M4 + Mp)L
Ay=
2
It is required that 4; + Az = 0; thus

2A
Misi aie ra ne

Fig. 8.16

(a) Fixed beam with irregular loading

F
H
E Area A;
eG +ve
A B
x
(b) Free moment diagram

M,| eG -ve { Mz
Al.—§ %—
(c) Fixing moment diagram

(d) Combined moment diagram

224 MECHANICS OF ENGINEERING MATERIALS

since there is no relative deflection of each end of the beam, then from eqn.
[8.29c]
Ayx] 4 AnxX?2 =F
EY EI
Taking moments about A, for the areas under the fixing moment diagram

Mel? (M4—Mgz)_L
x2 = is
Et St | 3
Msl*? MyL*
= a [8.31]
8.31

Therefore

Mel? MyL’
~Ayz, = : 2 7
Hence

6A\x
Mice Me L2
(8.32]
From eqns. [8.30] and [8.32],
4A, 6A4)%)
M, A = -—ji za DR 8.33
[8.33]

and

6A\x%, 2A;
M,; B =— [2 r
—Ve 8.34

Fixed beam with By combining the positive free bending-moment diagram ABC and the
concentrated central negative fixing moment diagram, EABD, the resultant diagram AEFCGDB
load is obtained as shown in Fig. 8.17. Since area ABC+EABD must be zero for
no change in slope at A compared with B

WL?
M,L +—=0
8
Hence

WL
M4 =-—-=Mp
and the points of contraflexure will be at G and F, distance x from each end,
where x = iL.
The value of M at mid-span is numerically equal to that of M at each
end, but is positive. Considering half the span, and taking the chosen
reference line through the left-hand support, we can obtain the deflection at
mid-span using eqn. [8.29c]:
_ AB
ea
 STATICALLY INDETERMINATE BEAMS 225

7 WL

A B
(b) Free moment diagram

Area = MyL

(d) Resultant moment diagram (c) Fixing moment diagram

where A and ¥ are the values for the free bending-moment diagram and the
fixed bending-moment diagram from the left end of the beam to the mid-
span. Therefore

LC A. Ci OD Wile. To
= x= X 42 | ==
KS KS
EI % 439 Pe gay |
Umax

VE
EIT \48 64

SS ee [8.35]
It may be noted that this is the same result as was obtained earlier by the
double-integration method (egn. [8.10]).

Continuous beam on A beam resting on more than two supports is said to be continuous. Such a
multiple supports beam is represented by Fig. 8.18(a). Changes of curvature occur in each
span, owing to negative bending moments at the supports. In the case
represented, the supports are assumed to be at different levels, being
displaced vp, vj and v2 from a horizontal line AB. Suppose the loading to be
such that the ‘free’ (positive) and ‘fixing’ (negative) bending-moment
diagrams are as shown at (d) and (c), the resultant diagram being shown at
(d). The area of the resultant diagram on the span /; is 4), and the distance
of its centroid G, from the chosen reference line through the left-hand
support is 1; also the area of the resultant diagram on the span /) is 42, and
%2 is the distance of its centroid G» from the other chosen reference lines
through the right-hand support.

(1) Bending moments at supports

Draw CD, a common tangent at the point of contact of the central support,
and let @ which is actually a small angle, be its inclination to the horizontal.
Taking intercepts, between tangents to the deflected beam, on a vertical line
as positive when measured downwards, and vice versa as negative upwards,
the same convention as for deflections, where the intercepts on the left-hand
and right-hand reference lines are zo and z2 respectively,
226 MECHANICS OF ENGINEERING MATERIALS

(a) Continuous beam (b) Free moment diagram

Fig. 8.18

ee [as ee |Me

(c) Fixing moment diagram

(d) Resultant moment diagram

20 0 = etal
EI =V);—
1 VU 0 aa] 1 a

Ajx)
22 2 = EI = —ha+(v1—v
pe ( 1 2)

and
Aix, v1 — v9 Anx2 v1 — 22
—._ — — — =a= — eae eee
Fie ae in ee
Therefore
A,X, Ax nl cee BO si5) toy)
El, El, 4h hb
and

Aix, Ax Oy oh =
i + —L = ( F Si i EI [8.36]
 STATICALLY INDETERMINATE BEAMS 227

When the supports are all at the same level, v9 = v} ="v2, and

AjXy anArk, _ 0 [8.37]

I 1D)

Let the areas of the ‘free’ bending-moment diagrams be S$; and $2 and
the distance of the centroid of S; from the reference line through the left-
hand support be x;, and the corresponding distance of the centroid of Sz
from the reference line through the right-hand support be x2; then, the sum
of the moments of the ‘free’ and ‘fixing’ moment diagrams are respectively

i M, —-M i
Ayx, = S,x; + |Mol; za 0) 1,2 :
2 y, 3

and

L M, —-M L
Anx, = Sox. + (tat ;ae ( d 5 2) 2 :)

Substituting into eqn. [8.37] and rearranging gives

Syx] S2.%2 Mol My, Moh
h+l = ()
je pe =grrr ache
and

Moh +2My(h +h) + Moh = -6(S - S

2) [8.38]
1 2

This is Clapeyron’s theorem of three moments, and by taking the spans in

pairs, sufficient equations are obtained to solve for the bending moments at
the supports.
Useful forms of 6Sx/L are shown in Fig. 8.19:
(a) for a point load,
6Sx Wa
aye
= L? 2

(6) for a uniformly distributed load,

6Sx wL3
—.

Fig. 8.19 w/unit length

(a) For a point load (b) For a uniformly distributed load

228 MECHANICS OF ENGINEERING MATERIALS

(i) Reactions at supports

In order to calculate the support reactions, consider the beam in Fig. 8.18
and let the reactions at E, F and G be Ro, R;, Rz respectively. In Fig. 8.20
we have split the beam into two separate free bodies which have reactions
Rh, Ri and Rj, RY and bending moments at the supports, Mo, M; and Mp.
Let the loading on cach span be f(W, m1) and f(W2,wz) and their
centroidal distances from E and G be x; and x2 respectively. Then taking
moments about E for span EF

Rih +M, — My —f(M, m1)%1 = 0

Ri = =(FM, )%, — (M; — Mo)| [8.39]

Fig. 8.20

Similarly for span FG take moments about G:

Rh + M, —M, —f(W, w2)X2 =(

|
R= zt (Ma 2)x2 — (M; — M2)! [8.40]
Now

= Ri + Ri [8.41]

Therefore

R eA iae 1) x} FM, w2)X2 M, — Mo (M, — M2)

1= a
h h h Lh

[8.42]
Ro and R may be found from vertical equilibrium for each of the spans EF
and FG and the above procedure can then be repeated for other adjacent
spans to determine all the reactions. It then becomes a simple matter to plot
the shear-force diagram for the beam.

Example 8.1
Draw the bending-moment and shearing-force diagrams for a continuous beam
which is supported at three points at the same level, but free at its extremities. The
spans are 15.2m and 10.6 m; the 5.2m span supports two loads of values 8900 N
and 4450N distance 6m and 12m reespectively from a free end, and the 10.6m
span is loaded uniformly with 1459N per metre run.
 STATICALLY INDETERMINATE BEAMS 229

8900 N 4450 N
Fig. 8.21
6m 6m
1 1459 N/m
Va aNaNaNaNANA
A

(a) SE Zam) (b)

Mg Vi

(c)

10340
4506
C

: -4394 a ; 5210

(e)

The maximum ordinate of the free bending-moment diagram (Fig. 8.21(d))

for span BC is

wL? 1459 x 10.6 x 10.6

= 20500 Nm
Sea 8
Considering the span AB, in order to draw the free bending-moment
diagram we must first determine the reaction at A due to the two
concentrated loads only.
Taking moments about B for the free span AB

—15.2R', + (8900 x 9.2) + (4450 x 3.2) =0

R', = 6324
Bending moment at 8900N load = 6324 x 6

= 37944

Bending moment at 4450N load = (6324 x 12) — (8900 x 6)

= 22 488
Referring to Fig. 8.21 M4 = 0 and Mc = 0, since the ends are free. Then
eqn. [8.38],

MOP On ean Wee 6(aes

a =)
1 2D
230 MECHANICS OF ENGINEERING MATERIALS

becomes

2Mp(h + |) = 6(

The free bending-moment diagram for AB is divided into four areas as

shown in Fig. 8.21(4) in order to calculate S)x1/h:

Six) 1 (a ne
= x + (22488 x 6 x 9)
I 15:2 Ms 3

(es | (28s
9 0)

— 165190 Nm?

Syxy =
1 eC500 x 10.6 x 2 5.3) =72433Nm
ip 10.6 3

2 x 25.8Mpz = —6(165 190 + 72 433)

Mp = —27630Nm

The resultant bending-moment diagram is shown in Fig. 8.21(4). Next we

must find the reactions at A, B and C. Taking moments about B for span AB

—15.2R4 + (8900 x 9.2) + (4450 x 3.2) — 27630 = 0

Ry = 4506N

Taking moments about B for span BC

+10.6Rc — (15465 x 5.3) + 27630 = 0

Rc = 5126N

Vertical equilibrium for the whole beam gives

4500 — Ke — 5126 89004-44507 15 465, 0

Ry = 19183 N

Note that this could also have been obtained by taking moments about A
for span AB to get Rz and taking moments about C for BC to get R and
then Rz = Ry =F ie
The shearing-force diagram can now be determined and is shown in
Fig. 8.21(e).
 STATICALLY INDETERMINATE BEAMS 231

(b)

Example 8.2
A cantilever carries a uniformly distributed load w over its span L. Find the correct
location for a prop which is to carry half the total load and to have the point of
support at the same level as the fixed end as shown in Fig. 8.22.

Let the constraint be A, the prop B and the free end C. Then, imagining a
similar span to the right of A, we can get the value of Mj:

wl?

where / =BC and W is the evenly distributed load. Then, if AB= /,

AC= 1; --7T = L. In eqn. [8:38],

new) i
$1 =S2==' x Fh and n=m=>

Therefore

] ip i
Meph + 2My4(h +h) + Meh = | (5 hiex [2

Substituting for Mg and M},,

Therefore

wi? 1
M,= l ie
2 (3 wt?)
Now Rz is equal to half the total load, m(/; + /), and also, from eqn. [8.42],
wl (Mp = M4) wl (Mp = Mc) w(h se )
Re= + — a
2 h 2 i/ 2
232 MECHANICS OF ENGINEERING MATERIALS

so that My — M,(/ +1) = 0, since Mc = 0. Substituting for M4 and Mz

and simplifying gives

P — 4h —6F =0
Therefore

1, = 5.16/ = 0.840L

8.4 Summary
The first consideration in relation to this chapter is the recognition of a
statically indeterminate beam situation, i.e. the appreciation that the support
reactions cannot be determined from the two equilibrium equations for the
beam. The next step is to choose one of the methods of solution that have
been presented. This will depend on the nature of the problem, the
availability of computation, etc. Double integration can be used for any case,
but, for example, superposition is best suited to a convenient breakdown
into simple separate elements. Moment—area can be quite a convenient
method for some continuous beam situations as an alternative to double
integration. Another important aspect is the correct application of the
boundary conditions at support points. It is essential im practice to realize
that supports which are supposed to be at the same level may not be so, and
hence additional bending stresses can be induced. A so-called fixed or built-
in support may not ensure zero deflection, or zero slope for that matter. It is
also quite useful to remember that if points of contraflexure (zero B.M.)
occur, then it may be more efficient to make a beam which has hinged
connections at these points, e.g. in a multi-span bridge such as the Forth
Bridge.
In structures that contain more than a few beams, the amount of
calculation may become excessive. In these cases, computer packages based
on matrix stiffness or finite element techniques (Chapter 17) are usually
employed.

Problems 8.1 An I-section beam of length 5m is built in horizontally and at the

same level at each end. If the maximum allowable stress is 90 MN/m/?,
determine what could be the maximum uniformly-distributed load.
The depth of section is 400mm, and the second moment of area is
$c (0c tm.
8.2 A horizontal beam of length Z is fixed at one end and simply-
supported at the other. A uniformly-distributed load m extends from
the fixed end to mid-span. Determine all the reactions and the
deflection at mid-span.
8.3 A bar of length Z and flexural stiffness EJ is built in horizontally and
at the same level at each end. A clockwise couple M is applied at mid-
span. Find the slope at this point and the deflection curve for the bar.
8.4 A beam of 20 m length is fixed at the left end and simply-supported at
the right end, where a clockwise couple of 210kN-m is applied. A
uniformly distributed load of 4kN/m extends from the fixed support
to mid-span, and a concentrated load of 60kN is applied at 5m
from the right-hand end. Calculate the maximum deflection. EJ =
140 MNm’.
 STATICALLY INDETERMINATE BEAMS 233

8.5 A beam is fixed horizontally at the left-hand end, A, and is simply-

supported at the same level at B and C, distant 4m and 6m from A. A
uniformly-distributed load of 2.5kN/m is carried between B and C.
Determine the fixing moments and reactions by the double-
integration method.
8.6 A beam in a small bridge deck has been damaged at a particular point
and temporarily a prop is to be placed beneath that point to carry half
the concentrated load occurring at that position. The beam is 4m long
and has both ends built in at the same level as shown in Fig. 8.23 and
the concentrated load F occurs at 3m from the left wall. The prop is
to be a circular bar. Calculate its diameter so that, as stated, the beam
and column each carry half the applied load. The second moment of
area for the beam is 30 x 107 mm* and the modulus of the beam
material is three times the modulus of the column material.

Fig. 8.23

8.7 Use the principle of superposition to find the wall reactions and
moments: (a) in Problem 8.3; (4) in Problem 8.2.
8.8 Draw the bending-moment diagram for the beam loaded as shown in
Fig. 8.24.

Fig. 8.24 8 KN 8 KN

\, cdo
=<—10 m>|\< 20m a ‘om-s0 ll

8.9 The beam AE is supported at four points as in Fig. 8.25. Draw the
bending-moment and shear-force diagrams for the beam (a) if the
beam is pinned at A and (0) if the beam just rests on A.
8.10 A continuous beam having three spans each of 10m has the four
simple supports A, B, C and D at the same level. Span AB carries a
uniform load of 5kN/m, and a concentrated load of 20 kN acts at 4m
to the left of D. Sketch the shear-force and bending-moment
diagrams.
A horizontal beam is built in at one end and supported at three points
as in Fig. 8.26. If it is subjected to point loads W at the middle of the
234 MECHANICS OF ENGINEERING MATERIALS

Fig, 8.25 20 kN 10 KN

three spans AB, BC and CD, calculate the length of each span so that
WL
M4 = Mp = Mc =——.

i dh ee ee
Fig. 8.26 WwW

1 i ie
CHAPTER
Energy Methods

The concept of stored elastic strain energy, introduced in Chapter 3, will now be
developed further. It will be shown that, by considering the balance between internal
strain energy and the external work done on a structure, the deflections or stresses at
a particular point in a structure can be found more conveniently than before. These
ideas will be demonstrated first in some simple structures such as helical springs
where only a single load or moment is acting. The concepts of virtual work, strain
energy and complementary energy will be introduced, since these provide the basis
for a number of advanced techniques in the analysis of both statically determinate
and statically indeterminate structures. Castigliano’s hypothesis, which is especially
useful for structures with zero or a small number of redundancies, will be used to
solve some problems.

9.1 Work done by a

Consider the load—extension diagram in Fig. 3.12: the external work done
single load
during a small increment of deflection du is F du. The total external work
done W when the extension is gradually increased from zero to a maximum
value 6 is
5
Vi |F du (9.1]
0
In other words, the work done is equal to the area under the load—
displacement graph. If the material of the structure being loaded is linear
elastic, the load—displacement relationship is linear and the work done is

W =1F6 (9.2]
where F is the maximum load at deflection 6.

9.2 Work done by a

Beams and shafts can support moments in bending (M) and torsion (7). In
single moment
Chapter 1, it was shown that a moment ™ can be replaced by an equivalent
force couple F.d. The work done by a moment can be derived from the
work done by the equivalent force couple. For a small rotation d@ about the
axis of the moment, the work done by the two couple forces is as shown in
Fig. 9.1.
8
w =2| P50 -
O)
|Mdé 9.3]
0 0

Again, if the structure behaves in a linear-elastic manner, the total work

done when a moment is applied 1s

w =1M0 (9.4]
236 MECHANICS OF ENGINEERING MATERIALS

Fig. 9.1

M
F

where @ is the total angle of rotation about the moment axis. The equivalent
expression for the work done on a shaft in torsion is

= 5TO (9.5]

o e

Fig. 92008 w 2 T
g Yom § 6
5 Elastic S Elastic
= limit 5 limit

0) Direct strain e€ 0 Shear strain y

(a) (b)

oa) pAasily Strain Onese) In Chapter 3 expressions were derived for the elastic strain energy stored in
normal stress and shear, material per unit volume under the action of direct stress or shear stress.
stress These represented the area under the appropriate stress-strain curve, as
shown in Fig. 9.2(a) and (4), and were given by

WR 3
= a pe unit volume for normal stress

and

lr J
Us a G Pe unit volume for shear stress

Conservation of energy requires that the internal elastic strain energy must
equal the work done by all the external forces and moments on the structure.
By integrating the above expressions for strain energy density over the
volume of the structure and equating these to the external work done, we
can often obtain solutions for the deflections and stresses in a structure in a
very economical way.

9.4 Strain energy in . . . At

Farsion For a circular shaft subject to a torque 7, the external work is ; 76. The
shear stress in a circular shaft is, from eqn. [5.11],

T=
si
 ENERGY METHODS 237

where 7 = Jp dA is the polar moment of area of the shaft. The internal

strain energy in a uniform shaft of length Z is then

ve bee:Al = (oats||Pddde = (eas

= [9.6]
v2 G 2GF-4o 4 2GF
Equating the external work and the internal strain energy,

TO heal
jE AGE
From this, the angle of twist of the shaft is
15,
§= — = 9.7
w)7

9.5 Strain energy in

For a beam subjected to bending, the direct stress is
bending

=———
My
df
The internal strain energy in a short length dx of the beam is therefore
] few L NZ Mh Ve
U=| ~—=dV= > dAdx = | ——dx 9.8
|,3 B ||2am ‘ |DET i
This may be used to find, for example, the rotation of a beam due to a
particular moment loading. Consider a cantilever with a moment at the free
end as shown in Fig. 9.3. The external work done is

Fig. 9.3

The internal strain energy is

L yr
(Qi) Seat
|Ta
Since moment is constant along the full length of the beam

a ieil
Tr
Equating external work and internal strain energy, the rotation @ of the end
of the beam where the moment is applied is found to be

9 _siML (9.9)
This method is further illustrated in the following examples.
238 MECHANICS OF ENGINEERING MATERIALS

Beam simply supported Since the bending-moment relationship is discontinuous over the length of
with load at mid-span the beam, it is necessary to split the integral of eqn. [9.8] into two parts.
(Fig. 9.4) Thus for 0<«< 11, M=45Fx, and
L/2 Fe x2
Strain energy up to the load = | dx [9.10]
et
Fig. 9.4 F

||

When 42. < v< J M=5 Fe— Fab) = +F(L —«x), and
Strain energy stored in second portion of beam
is he 2
= | (L— x)" dx
1S Pe 8EI
Therefore the total strain energy is
L/2 prryd 25 pe 2
| ae + | tie
Ia: Lj SET

Bp Eis
U = TonEI ’ 19251

FL
(9.11]
~ 96ET
In this particular problem, since the load is at mid-span, the total strain
energy could have been obtained by doubling the first integral (see eqn.
[9.10}).
Now, the total strain energy is also equal to the external work done, so

Work done, W = 5 F Umax

Therefore, from [9.11], Umax = FL? /A48EI.

Deflection of a beam __Let a load F strike a beam of span L simply supported at its ends, at mid-
under impact loading _ span. If / is the distance fallen by W and 6 is the deflection produced, then
the work done is F(h + 6).
If F, is the equivalent static load applied at mid-span to produce the
deflection 6 then the work done by F; is given by 5F 16; therefore

iF 16 = F(h+6)

But the central deflection is given by the solution to the previous example,

meade.
~ 48ET
 ENERGY METHODS 239

Thus

48 E16
apa = P(A +6)
Pe ow trie
24ET 2A4ET

FR] FI \? Fh
ee aaa + [9.12]
48EI | 2 24EI 6EI

Example 9.1
A beam of 3 m length is simply supported at each end and is subjected to a couple
of 9kNm at a point B, 2m from the left end as shown in Fig. 9.5. Determine the
slope at B. El = 30kNm?2.
The reactions at A and C are M/L = 9000/3 = 3kN.

When 0\ay= 2) 1 —

The strain energy stored is

1 [? (Mex q eat eae z

rales zs ane ) oe
2 3
ee 79000% ) |x? " (x — 3)°
~ 60x 103 \ 3 3 e 3 ;

3 (9000 *
TCO0 es
The work done at B is

Therefore

9000, 3/9000 :
OF eer O0ix 10°
240 MECHANICS
OF ENGINEERING MATERIALS

and

6 = 0.1 rad

9.6 Helical springs

Springs are directly concerned with the theories of torsion, bending and
strain energy and form an excellent example of their application. There are
comparatively few machines which do not incorporate a spring to assist in
their operation. The principal function of a spring is to absorb energy, store
it for a long or short period, and then return it to the surrounding material.
Two extremes of operation are found in a watch and on an engine valve. In
the former case energy is stored for a long period and in the latter case the
process is very rapid. The force required to produce a unit deformation of a
spring is called the stiffness.

Fig. 9.6

Fsina

Foosa

The geometry of a helical spring is shown in Fig. 9.6. The centre-line of

the wire forming the spring is a helix on a cylindrical surface such that the
helix angle is a. Helical springs are designed and manufactured in two
categories: close coiled and open coiled. In the former the helix angle a is
very small and the coils almost touch each other. In the second case the helix
angle is larger and the coils are spaced farther apart.

Axial load on open- The most common form of loading on a spring is a force F acting along the
coiled spring central axis. Since this force acts at a distance R, the coil radius, from the
 ENERGY METHODS 241

axis of the wire, there will be torque and bending moment set up about the
mutually perpendicular axes OX, OY and OZ on the cross-section at O in
Fig. 9.6.
M, = FR cos qa (causing torsion on the wire) = T
M, = FR sin a (causing bending about the y-axis and a
change in R)
M10)
In addition, cross-sections of the wire are subjected to a transverse shear
force / cos a and an axial force F' sin a. The stresses due to these forces are
considerably smaller than those due to torsion and bending and are generally
neglected.
The work done in deflecting the spring 6 by the axial load F is + F6, and
the stored energies due to torsion and bending are ;T@ and 5Mé, where @ is
the angular twist of the wire and ¢ is the change in slope of the wire.
Therefore

=}(FR cos a)0+4(FR sin a)d

hence, using eqns. [9.7] and [9.9] for 6 and ¢ where @ is the rotation due to
the torque and ¢ is the rotation due to the bending moment,

AL, ML
i= R(cosa—
FOMOA © VEL
+ sin om)

sip |Sema
nat wG EI
[9.13]
The length of wire in the spring is 27mRsec a, where n is the number of
complete coils.

md* nd*
= — d J/=—
a, 64
where d is the wire diameter: thus

= 64F Rn sec a (= a 2 sin’ *)

6 7 G E [9.14]

Axial load on close- The helix angle a is very small in the close-coiled spring so that sin a — 0
coiled spring and cos a — 1, and egn. [9.14] reduces to

64F Ron
a [9.15]

Example 9.2
A close-coiled helical spring is to have a stiffness of 1kKN/m compression, a
maximum load of 50N, and a maximum shearing stress of 120 MN/m?. The solid
length of the spring, i.e. when the coils are touching, is to be 45mm. Find the
242 MECHANICS OF ENGINEERING MATERIALS

diameter of the wire, the mean radius of the coils and the number of coils required.
Shear modulus G = 82 GN/m72.

; ie Gda*
Suffness = rae: 1000 = CAR

a 2
Maximum torque 7 = a — 120 x 10° = = 5072

Hence

120 x 10°
R=———1
800”
Closed length of spring = nd = 0.045

Substituting for m and R in the equation for stiffness above.

5 Gd* x d
1000
64 x (15 x 10413)? x 0.045

» 82 x 10°
2 ; =e
64 (15%) x 10% 45
Thus d = 0.00406m = 4.06mm; R = 31.6mm; and n = 11 coils.

Example 9.3
Compare the stiffness of a close-coiled spring with that of an open-coiled spring of
helix angle 30°. The two springs are made of the same steel and have the same coil
radius, number of coils and wire diameter, and are subjected to axial loading E
=25G.

From eqn. [9.15], the stiffness of the close-coiled spring is

Oy eke! ss
“6 64R3n
From eqn. [9.14] the stiffness of the open-coiled spring is

re d*
6 64R3n sec a(cos? a/G +2 sin? a/E)

Rec CG cova 2 sina

—=G seca —
Roc G E

2 2 sin’a
= SCG GT COSm Of-|—
230

a aoa =1.1
CAC eo
 ENERGY METHODS 243

Axial torque on open- The other type of loading on a spring which is of interest is that where the
coiled spring spring is subjected to a torque about the central axis. The resolved
components of the axial torque 79 about any cross-section are To sin @
about the axis of the wire and 7) cos a changing the curvature of the coils.
If one end of the spring moves round the longitudinal axis an amount w
relative to the other end owing to the torque T> then the work done is 5Tow
and the stored energies are }To@ sin a and 5To@ cos a, so that

w=8@sina+¢ cosa

: ML
~ 9G sin Caer COS &

sin? a cos” =)
= TomnR sec a(IC + ET

a2 128TonR sec a (sin*a cosa

zl
d* (2G ss E ) ls)

Axial torque on close- Putting sec @ = cos a = 1| and sin a = 0 in eqn. [9.16] gives
coiled spring
1287ToRn
Nae ar ETS
78 (9.17]

9.7 Shear deflection of

A deflection other than that due to bending moment occurs in beams owing
beams
to the shearing forces on transverse sections. This deflection may be found
approximately from strain energy principles and by making use of the
equation for shear stress at a point in the transverse section of a beam. For
the majority of beams, where the span L is large compared with the cross-
section of the beam, it will be seen that the deflection due to shear is
negligible in comparison with that due to bending.

Cantilever with load at Assume the section to be rectangular, of breadth } and depth d, and the total
free end (Fig. 9.7) length of the beam to be L. If v, is the deflection, due to shear, at the free
end, then

Work done by load = 3 Fv,

Fig. 9.7
244 MECHANICS OF ENGINEERING MATERIALS

The shear stress at a distance y from the neutral axis is obtained from
eqn. [6.42]

where the shear force O = F. Also, if dy is the height of the strip in the
direction of the depth of the beam, and we consider a small portion of the
beam of Jength dx and section b dy, we have
2
Strain energy in strip = Gdrbdy
(ile
13)

and substituting for 7, from above

136 (2 dy 4
Strain energy = IG
— dx776 (;
6 - bytbay

Therefore the total strain energy for the piece of beam of length dy is
18F2d +d/2 d+ a 2
a x | US +y' dy
bd°G J_gj2 \16 2

3 F*dx
=~ oi
5 (bd)G 7-18)

The strain energy for the whole beam of length L is

if oone|fre
Sruphay?
eee
GIFEE
5 bdG Jo 5 bdG

Equating the strain energy to the work done by F,

aa
Sark,
ae ee Sand
Therefore
al gl& 9.19
v= > :
"Subd a
Thus the total deflection at the free end due to bending and shear is
= Oar we

a (abe p ae a&
Ee iG
For many materials, E ~ 3G. With J = phd, this equation may be
rewritten as

FL aNe
v 1+ 0.9| —
mei
 ENERGY METHODS 245

Hence, for a typical beam where (d/L) is less than 0.1, the percentage
error in neglecting the shear effect is less than 0.9%.

9.8 Virtual work

The principle of virtual work is one of the most powerful tools in structural
analysis. The principle may be stated as follows:

if a system of forces acts on a particle which is in statical equilibrium and

the particle is given any virtual displacement then the net work done by
the forces is zero.

A virtual displacement is any arbitrary displacement which is

mathematically conceived and does not actually have to take place, but
must be geometrically possible. The forces not only must be in equilibrium
but are assumed to remain constant and parallel to their original lines of
action.
If a body is subjected to the system of forces shown in Fig. 9.8 and is
then given an arbitrary virtual displacement for which the corresponding
displacements in the directions of the forces are 1, u2, u3 and u4, then static
equilibrium exists for the system of forces if

Fyuy + Fou oF F313 an Fug ==()

This expresses in mathematical form the principle of virtual work for the
system considered, i.e. the sum of the work done by each virtual
displacement is zero.
Let the resultant force of the above system be P and assume a virtual
displacement A in the direction of P; then for static equilibrium PA must
be zero. Since A need not be zero, it follows that P must be. Thus the
resultant of a system of forces in equilibrium is zero.

Application of virtual We can now make use of the above principle in deriving energy solutions for
work method deflections. Consider the simple framework in Fig. 9.9 acted upon by forces
F, and F>. Let the displacements at the joints in the direction of the forces
be uw and u2. The internal reactions and deformations of the members of the
frame are P, P2, etc., and A,, Aj, etc., respectively. Then, from the
principle of virtual work,

Fyu, + Fou. = P\A; + P A> ae co aa Se JPN

or

ee = as [9.20]

Fig. 9.9
 MECHANICS OF ENGINEERING MATERIALS

Thus the work done by external forces at the joints equals the internal work
resulting from the tensions in the members. The summations in eqn. [9.20]
cover all joints ;, and all members n, for static equilibrium.

9.9 Displacements by the

If a plane frame is subjected to one external load only and the displacement
virtual work method of the loaded joint is required in the direction of the load, then eqn. [9.20] can
Actual forces, be used with all real values. This is illustrated in the following example.
extensions and
displacements

Example 9.4
Determine the vertical and horizontal displacements of the joint D in the plane pin-
joined framework in Fig. 9.10. All members have a cross-sectional area of
1000 mm? and modulus of 200 GN/m2.

Fig. 9.10

The first step is to calculate the forces in all the members by one of the
methods described in Chapter 1. Next the change in length of each member
is determined and it is convenient to tabulate the foregoing information:

Force Length Change in

Member F (kN) L (m) length, FL/EA (mm)

BC +20 2 +0.2
CD +10 2 saath
FG —10 jb =0:1
FC —10,/2 2/2 —0.2
GE +10 2 +0.1
GD —10,/2 2/2 —0.2

Let the vertical displacement of joint D, which is where the load is

applied and in the same direction, be 6. Then, using numerical values from
the table,

>| PA= (20 x 0.2) + (10 x 0.1) +(=10 x —0.1)

+(—10,/2 x —0.2) + (10 x 0.1) + (—10,/2 x —0.2)

= 12.66kNmm
 ENERGY METHODS 247

Therefore

106 = 12.66 and 6=1.266mm

The horizontal displacement of D cannot be found by the above method,

since there is no force in the horizontal direction at D to give an external
work term.
The above method of using the real forces and extensions cannot be used
on a frame carrying several external loads, unless all the displacements at the
loaded joints, except one, are known, since there would be several unknown
6 values on the left-hand side of eqn. [9.20].

Actual extensions and The previous section demonstrated the use of eqn. [9.20] in which the work
displacements with terms were composed of the real forces and real displacements. However,
forces hypothetical the equation is equally valid if the displacements are those actually occurring
due to the applied loading, but the load and force terms are completely
fictitious, so long as they form an equilibrium system. This feature enables
one to determine the displacement of any joint in any direction, whether
loaded or not, by the use of a ‘dummy’ load of unit magnitude.
Consider again Example 9.4 and the need to find the horizontal
displacement at D. A dummy unit load is inserted horizontally at D. Then
the equilibrium system of forces in the members due to the unit load only (the
10 kN is removed) can be found; the values are as given in the following
table.

Actual extensions, A, | Hypothetical forces, P’,

Member due to 10 kN load due to dummy load PX

BC +0.2 1 +0.2
INC =) 0 0
FG (0) 0 0
CG +0.1 0 0
CD +0.1 1 +0.1
GD = 0 0

yD = 02

If the real horizontal displacement of D due to the actual load of 10 kN is

6, then the external virtual work is 1 x 6. The internal virtual work is given
by the sum of the products of the actual changes in length, A, of the
members due to the actual load of 10 kN and the hypothetical equilibrium
set of forces, P’, resulting from the unit dummy load. The simple
computation is shown in the right-hand column of the above table. Applying
eqn. [9.20] gives

is oP a or 6—0.3mm

Example 9.5
Determine the vertical displacement of joint G of the frame in Example 9.4.

Following the same procedure as above, a unit dummy load is placed

vertically at G; the corresponding force system due to the unit load only 1s
248 MECHANICS OF ENGINEERING MATERIALS

shown in the table below.

Actual Forces due to

Member _ extensions unit load PA

BC +0.2 +] +0.2
FC —().2 —/2 +0.2,/2
FG —().1 0 0
CG +0.1 ] +0.1
CD +0.1 0 0
GD —0.2 0 0

+ = +0.5828

Thus

16g = 0.5828 or 6¢ = 0.5828 mm

9.10 Strain energy

Energy functions may be very usefully employed in the determination of
solutions for forces
deflections of frameworks, beams, shells, etc. The methods depend on the
principle of virtual work as developed above. Now let us suppose that there
is a small change in displacement at joint A of an amount 6), 62 remaining
constant, which results in changes 6A), 6A, etc., in the members for
compatibility. Then
Fy (uy SF 611) + Fyu2 a Pi (Ay Se 6A) + P,(Az =F 6A2)

ere
4 fees Sa)

=) >P(A+6A) (9.21]
n

Subtracting eqn. [9.20] from eqn. [9.21]

F\6u, = 5) P6A [9.22]

But P 6A is the increment of strain energy stored in a member of the system

due to the increments of deformation 6A caused by the change in
displacement 6u;. Therefore for the system

ih OU

or
F\éu; = 6U [9.23]
Thus, for an infinitely small change in displacement,
OU
=
Ou,
[9.24]
 ENERGY METHODS 249

By a similar argument we have that

OU
fF, =—
Ou

Thus the external force on a member is given by the partial derivative of

the strain energy with respect to the displacement at the point of
application and in the direction of the force.

This result can be illustrated in the solution of Example 4.2, where two
bars were loaded in parallel so that both were subjected to the same
extension uw. The strain energy in each bar is
1 o ih Pe iL, Pe P-L
Cee ee ——dAdx = | —dr= 925
ie E | Sip |2EA ~ OF ee
Since P, the force in each component, is given by
EA
P= = = ku [9.26]
where & is the stiffness of the bar, the strain energy may be written in terms
of the displacement of the end of the bars as

(haven = 5(kyu? ar kyu’) [9.27]

where &; and 2 are the stiffnesses of the two bars loaded in parallel. The
external force required to cause the displacement wu is therefore
OU
Fe =— = (ki + kp)u [9.28]
Ou
or the displacement resulting from the applied force is
ie
= 929
Sky + kp saad
9.11 Complementary
We now return to the original: proposition, and instead of changing the
energy solution for displacement u;, we change the force Ff, by an amount 6/}, keeping F2
deflections constant; then there will be a reaction in the system causing changes 6P},
6P2, etc., in the internal forces in the members. Now, by the principle of
virtual work, we have
(Fy + 6F\)uy + Fou, = (P| + 6P\)Ay + (P2 + 6P,) A

+...4+ (Py + 6Py)An

=> (P+ 6P)A [9.30]

Subtracting eqn. [9.20] from eqn. [9.30]

6F\u; = 5 6PA [9.31]

250 MECHANICS OF ENGINEERING MATERIALS

Fig. 9.11

P
Force PForce

qe PSA

10) B B' Displacement A (0) Displacement A

(a) (b)

Now considering Fig. 9.11(a) or (4), the shaded area PdA is the
increment of strain energy 6U below the load—deformation curve used in
eqn. [9.22]. The shaded area above the load—deformation curve represents
A 6P in egn. [9.31]; this is termed the complementary energy and is denoted
by C; thus

oP AoE

and therefore
6F\u,; = 6C [9.32]

For an infinitely small change in the force,

OC
= 5 9.33
OS OF; a
Thus the deflection at a point on a member in the direction of a force
applied at that point is given by the partial derivative of the
complementary energy with respect to the external force at the point.

The above energy theorems provide a most useful method of attack on

many structural analysis problems. A further point of interest is illustrated
in the load—deformation characteristics of Fig. 9.11(a) and (4). The former
illustrates linear elasticity for a member or system of members, while the
latter represents non-linear elasticity, which can occur in certain structures
and materials. In both cases the sum of the strain energy and complementary
energy is given by the force times the deformation, i.e.

OSG as 1
LAN [9.34]
but in the particular case of linear elasticity,
60 = 0G = POX = Nor
and
CSCS eA [9.35]
ENERGY METHODS 251

Because of this last relationship we can express displacements in terms of

strain energy instead of complementary energy for linear-elastic systems.
One of the earliest theorems of this form was due to Castigliano (1875) in
which it was stated that for linear-elastic structures the partial derivative of
the strain energy with respect to a force gives the displacement
corresponding to that force, or

A (9.36]

This relationship may be proved in the following way. Consider a force P

applied to a body giving a displacement A. Then the work done or the
stored strain energy is equal to OAB which equals +P A in Fig. 9.11(a). If an
additional force 6P is applied giving an additional deformation 6A, then the
extra strain energy is
BAA'B’ = P6A +56P 6A = 6U [9.374]
or 6U/éA = P, neglecting second-order products. Therefore,
Total energy, OA’B’ = +P A+Pé6A+ +6P 6A
If both forces had acted simultaneously, the stored strain energy would have
been OA'B! = 5(P + 6P)(A + 5A). Since work done is independent of the
order of application of the forces, we have
1PA+P5A+56P6A =}(P+ 6P)(A + 6A)
On simplifying, and neglecting small products, we find that
PONS NOP [9.375]
(i.e. 6U = 6C for linear elasticity). Thus, substituting in eqn. [9.374]
6U =A6P+}6P6A
Therefore
6U 22k
cot
6P
neglecting the second-order term on the right. Hence
OU
—=A
OP.
which proves Castighano’s hypothesis.
The simplest applications of eqn. [9.36] are related to the deformations
in tension and in torsion:
(i) in the case of a bar under a simple tensile force F,

EL
2AE
and
OUR EL
—— = —_= A, the extension of the bar;
OF AE
252 MECHANICS OF ENGINEERING MATERIALS

(ii) and in torsion for a torque 7,

vi (eas
2c;
or

ae = ae = 0, the angle of twist.

or Gf
9.12 Bending deflection
The complementary energy function can be used very conveniently to solve
of beams for beam deflections, since

OC
—=6
OF
or, using the notation for beams,
OC
—- = UV

OW
where W is a concentrated load whose displacement (beam deflection) is v.
The complementary energy in bending of a small length of beam dy is
shown in Fig. 9.12(a), which is the moment-slope relationship. The shaded
area 1S

dC =ddmM
or
M
C= | 0dM [9.38]
0

For a linear-elastic beam, using eqn. [7.4]

L
M
ee |—dx
9 El
Therefore, from Fig. 9.12(d),
M pL M L M ag
C=
J,lear" —dxdM=
|,ser
| —dx
ae:
Fig 912 = =
: :
s Ss
Cc

0) Angle 6 0) Angle 6

(a) (b)
 ENERGY METHODS 253

This result could also have been arrived at from the fact that U = C ina
linear-elastic system, and it has already been shown in eqn. [9.8] that the
strain energy is
Lae
M
U=) —-d
|ce
Given this expression for the strain energy, the deflection, v, under a
concentrated load, W, may be determined from
=e SoU
“OW OW
We can solve for the deflection in one of two ways, either

° = AW
ne (|.ie tr)
OW \ Jo 2EI
(9.40)

aU (-M OM
or 0= ay = |Fi ay [9.41]

It is merely a question of whether the bending-moment expression in terms

of W and x is substituted and the integral evaluated, followed by partial
differentiation with respect to W, or the latter is carried out first, substituted
in the integral and then evaluated.

Cantilever with At any distance x from the free end, the bending moment is M = Wx;
concentrated load at therefore
free end
Tye ee We ae:
° OW), ZEIT.” OW \ 6EI

= 9.42
Syoy! a

L Wx L Wx yy,2
|
v= | —x«xdx= | —dsr
| EI 9 El

Us
— 3EI

Beam simply supported

with uniformly
distributed load

and 0/OW of the above is zero, indicating no deflection, which is obviously

not true. To get round this difficulty, we introduce an imaginary
concentrated load W at some point in the span, let us say mid-span for
254 MECHANICS OF ENGINEERING MATERIALS

simplicity. Then, for 0 < x < L/2,

W wl wx?
M = D x + D x; D [9.43 |

and OM/OW = 5x; therefore

2 ee iW. " wL wx? ¥
ie a Ge TG Age tones 2
= —_ | — en 95 me It — CL

WL i SwL*
48EI 384k
Putting W = 0, we obtain

5 pL
= — 9.44
MEPIS ET aan)
If we require the deflection due to the point load only, we put » = 0; then

WL
= 9.45
°* 48ET ae
Example 9.6 eS ge ee ae ee eh a ae et eee
P A simply supported beam (Fig. 9.13) carries a concentrated load at a distance a
from the left-hand support and a distance b from the other support. Determine the
deflection of the beam underneath the load.

R, = Wb/L and the bending moment at C is (Wb/L)x. For the portion of

beam AD, we have the complementary energy

- iWSS
(Wb/L)’x? dx i ee x oe Wa
=
ee
EE
) L-2EI
x=
6EIL?
(9.46]
Similarly, we may write that C for the portion of beam DB is
W?a’b> /6EIL’; therefore
Total value of C for the beam

Wea reWah Wael? Gay

> Khe 6k = GEL

_We(L— a)
6EIL [9.47]

Ww
Fig.9.13 -——.—|»—|
y B
2.

(a) (b)
 ENERGY METHODS 255

and

dC Wa(L—a)
Deflection underneath load = = [9.48]
dw S/OIUL,

Example 9.7 ; a ran as SR ee

Determine the vertical and horizontal displacements of the end of the curved
member shown in Fig. 9.14.

Fig. 9.14

(a) (b)

Considering first the displacement wu in the direction of the load.

1/2 Ne
oy —_R dé
» 2EI

au (7?M AM
“= aw | ion
= ——_ = —_—_ —— |

M, = WR sin 0,
OM,
aw *® sin 6

"/? WR sin 0
u=| aa Sm (R sin 0)Rd0
; EI

_7 WR [9.49]
To find the vertical displacement v, an imaginary vertical force Wo is
applied at B. The bending moment on C will be

M, = WR sin 0 — W)R(1 — cos @)

and
256 MECHANICS OF ENGINEERING MATERIALS

2 _____

Putting W = 0 in the expression for M,,

m2
°=a77| ee a lalemrer ty at
ae le El
from which

WR}
= —-—_— 9.50
OS ET be

Example 9.8
Determine the horizontal deflection of the member shown in Fig. 9.15.

Fig. 9.15

(a) (b)

The strain energy function is scalar; therefore the separate strain energy
quantities for the two parts of the member can be added before proceeding
to use Castigliano’s theorem.
From A to B

M=-+Fx and OF 1%

From B to C

FL OM L
M =e ind He
cy a ae es

O07 1 Oe ealof wD
er al Spe des | ds
OF al a ie | Pg
1 [Fx a 1 [FL2x}”
SE gee bay Lele

13FL3
~ T92ET [9.51]

Example 9.9
Determine the bending moment for any cross-section of the slender ring shown in
Fig. 9.16(a).

In view of the symmetry of the ring, only one quadrant need be considered
as shown in Fig. 9.16(4). Cutting the ring at any section the bending
 ENERGY METHODS 257

Fig. 9.16

moment is given by
WR
M, = Mo — aye (1 — cos @) [9.52]

Mp is unknown but may be obtained from the strain energy.

The strain energy is given by

gi WR :
U | (a1 5 (1 — cos 6) Rdé [9.53]

Owing to the symmetry at B, 6 = 0 = 0OU/0Mp. Therefore,

1 (7? WR
pee
a| ( ie
i— = (1 — cos 6) )Rab = 0
0
from which

1
My = WR (= ~) [9.54]
7

and

ee 1
Mp W RN COR 0a [9.55]
AG

9.13 The reciprocal

Mieaten In a linear structural system (Fig. 9.17), the deflection at point 1 due to
forces F', at point 1 and F) at point 2 is, from the principle of superposition,

A, =A + Ap
and the deflection at point 2 is

Ay = An + Ari
The deflections may be expressed in terms of flexibility coefficients, which are
the displacements per unit force, as follows:

Ay = fii + firk? [9.56]

Ag = faki + fak2 [9.57]

Fig. 9.17
258 MECHANICS OF ENGINEERING MATERIALS

If the strain energy of the system due to the application of these forces is U
then we may write

=35OU
bes SS 9.58]
9.58

and

1= Fe
OU
SS == (9.59
9.59

Partially differentiating eqns. [9.56] and [9.58] with respect to /2,

OA,
OF, = fir

and
fag U
:
AF Fy = fia [9.60]
9.60

Similarly, from eqns. [9.57] and [9.59],

OA?
ais = fr

and

aU
—_——_ = 9.61
OF) OF fai rae
From eqns. [9.60] and [9.61],

iu =fir2 [9.62]

This shows that the deflection at any point | due to a unit force at any
point 2 is equal to the deflection at 2 due to a unit force at 1, providing the
directions of the forces and deflections coincide in each of the two cases.
This is termed the reciprocal theorem.

Example 9.10
Determine the deflection at the tip of a cantilever beam of length / when a load F is
applied at a distance x from the fixed end.

From eqn. [7.22], the deflection of a point at a distance x from the fixed end
and when a load F is applied to the tip of a cantilever of length / is

B Fae

By the reciprocal theorem, the same deflection will result at the tip of the
cantilever when the load is applied at x. It can be verified that, with
appropriate changes of variables, this is the same solution as is obtained for
this problem in eqn. [7.24].
 ENERGY METHODS 259

9.13 Summary
It can be seen from the diversity of applications, e.g. from coil springs to
‘structures, that energy theorems have a very important part to play in
engineering design analysis. Some of the concepts may be a little difficult to
grasp at first, especially in relation to a physical appreciation. The principal
advantage of energy methods is that the force, moment or deflection at a
particular point in a structure (such as the end of a cantilever, or a node in a
framework) can be found without investigating the details of the structural
behaviour of every point.

Problems o1 Two possible designs of hydraulic cylinder are being considered as

shown in Fig. 9.1(a) and (4). One involves the use of long bolts and
the other uses short bolts of the same diameter. On the basis that the
bolts are each required to absorb the same amount of strain energy,
decide which of the two designs would be the best.

Fig. 9.18

(a) (b)

92 Two identical rectangular section steel members are loaded in the

following way: —
(a) one is subjected to an axial tensile force W;
(b) the other is supported as a cantilever and subjected to a load W at
its free end.
If each member is required to absorb the same strain energy, compare
the maximum stresses which this would cause in each case.
9.3 Compare the strain energy stored in a beam simply-supported at each
end and carrying a uniformly-distributed load, with that of the same
beam carrying a concentrated load at mid-span and having the same
value of maximum bending stress.
Bie A solid shaft carries a flywheel of mass 150 kg and has a radius of
gyration of 0.5 m. The shaft is rotating at a steady speed of 60 rev/min
when a brake is applied at a point 4m from the flywheel. Calculate the
shaft diameter if the maximum instantaneous shear stress produced is
150 MN/m?. Assume that the kinetic energy of the flywheel is taken
up as torsional strain energy by the shaft. Neglect the inertia of the
shaft. G = 80 GN/m’.
9:3 A valve is controlled by two concentric close-coiled springs. The outer
spring has twelve coils of 25 mm mean diameter, 3mm wire diameter
and 5mm initial compression when the valve is closed. The free
length of the inner spring is 6mm longer than the outer. If the force
required to open the valve 10mm is 150N, find the stiffness of the
inner spring. If the diameter of the inner spring is 16 mm and the wide
diameter is 2mm, how many coils does it have? G = 81 GN/m’.
260 MECHANICS OF ENGINEERING MATERIALS

9.6 When an open-coiled spring having ten coils is loaded axially the
bending and torsional stresses are 140 MN/m?* and 150MN/ m?
respectively. Calculate the maximum permissible axial load and wire
diameter for a maximum extension of 18 mm if the mean diameter of
the coils is eight times the wire diameter. G = 80 GN/m’,
E = 210 GN/m’.
oF: A bar of rectangular section 1 m in length is simply-supported at each
end and carries a uniformly distributed load. Determine the maximum
depth of section so that the deflection due to shear shall not be greater
than 2% of the total deflection. E = 2.6G.
9.8 A beam of 4m length carrying a concentrated load W at mid-span is
simply-supported at the right-hand end and is pin-jointed at the same
level at the left-hand end to the free end of a horizontal cantilever of
length 2m. Use Castigliano’s theorem to find the deflection under the
load. Both beams have a flexural stiffness E/.
ey An anti-roll bar in a car suspension is mounted in bearings which
allow rotation about the z axis at points B and C but prevent any
bending of section BC, Fig. 9.19. It is attached to opposite wheels at
points A and D. If the bar is 20 mm in diameter, what vertical force on
the bar would be generated at point A when the wheel attached to
point A raises by 1 mm and point D falls by 1mm? The bar is made of
steel, with E = 200 GN/m’ and v = 0.3.

Fig. 9.19

9.10
$ WAN All dimensions in mm

A U-shaped pipe connecting two vessels has a radius R and a leg-

length L. Show that the deflection caused by forces P applied at the
free ends and perpendicular to the legs due to thermal expansion is
Pay ny Yi
biplane Ay ightmR?
mS ET te Saba +aie’}
2

Jala A circular proving ring, used in the calibration of tensile testing

machines, is shown in Fig. 9.20. (a) Show that the bending moment in
the ring at any point ‘C’ is
WR
Mi= MM; yal —cos 6)

(b) Given this moment distrbution, show that the reduction in

diameter across AA’ will be
spn he 7 2
ET \4 «x
 ENERGY METHODS 261

Fig. 9.20

(a) WwW

where E is the Young’s modulus of the ring material and / is the

second moment of area of the ring section.
9.12 A weight W is supported by a semi-circular curved bar as shown in
Fig. 9.21. If the supports at A and B to not move calculate (a) the
reactions at the supports, () the position and magnitude of the
maximum bending moment in the curved bar and (c) the vertical
deflection at the weight W.

Fig. 9.21

9.13 A childs pram uses four semicircular springs for its suspension as
shown in Fig. 9.22. If the design weight of the pram and contents is
W, derive expressions for the vertical and horizontal movement of the
wheel axes relative to the body of the pram. The radius of each sring is
r and its second moment of areas is /. The modulus of the spring
material is E.
9.14 A split ring of radius R is used as a retainer on a machine shaft, and, in
order to install it, it is necessary to apply outward tangential forces F

Fig. 9.22 Rigid Suspension

springs
supports
262 MECHANICS OF ENGINEERING MATERIALS

at the split to open up a gap 6. If the flexural stiffness of the cross-

section is EJ, determine the required value of the forces.
g15 A curved beam is in the form of a semicircle as shown in Fig. 9.23.
The free end is pinned but is restrained to move in the horizontal
direction only. If a horizontal force F is applied at the free end show
that the horizontal restraining force on the end of the beam is given by
4F /3n.
Fig. 9.23

|
|
a
|
|
9.16 A curved beam is subjected to a load of 500 N at A, as shown in Fig.
9.24. If the product of Young’s modulus and second moment of area
for the beam section is given by EJ = 26kN-m’, calculate the vertical
and horizontal deflections at A.

Fig. 9.24 500 N

A 150

<———— 250 mm ees

NF The frame ABCD in Fig. 9.25 is fixed at A and a horizontal force F is

applied at D. B and C are rigidly jointed and the lengths of the
members are AB = CD = hf and BC = L. The relevant second
moments of area are /; for AB and CD and J, for BC. Show that if D
moves under load at an angle of 45° to the horizontal then
Ih {[3L—4h
| PeeTin
Ignore the effects of axial load and shear force on the frame.

[l
Fig. 9.25 hoes eae
 ENERGY METHODS 263

9.18 A circular section rod of diameter 20 mm has the shape of a quadrant

of a circle of radius 300mm (Fig. 9.26). If a vertical force 100N is
applied at the free end, determine the deflection in the direction of the
applied force. The Young’s modulus is 208GN/m/? and the shear
modulus is 80 GN/m/?.

Fig. 9.26 i

Zz Diameter = 20 mm

ong Show that for design situations where a circular section bar is to
absorb an axial impact force, a uniform section rod of area A has better
energy absorbing capacity than a stepped rod in which the cross-
sectional areas are A and 2A. The overall lengths are assumed to be
the same in each case.
9.20 The drive shaft from an electric motor has a diameter of 15 mm and a
length of 100mm. When the system is operating the pulley has a
kinetic energy of 10Nm. If the system comes to a sudden halt,
calculate the angle of twist and the shear stress set up in the shaft. The
shear modulus of the shaft material is 80 GN/m/?.
CHAPTER
Buckling Instability

In earlier chapters basic analytical design procedures have been developed in which
components and structural elements have been subjected to tension and
compression forces, bending moment and torque. In this chapter we shail examine
the specific effect of compressive forces in relation to the geometry and boundary
conditions of members. Examples range from the compression force of combustion
on the connecting rod of an internal combustion engine to the vertical columns used
in structural steelwork to support all the vertical mass and forces in a building. In
addition to axial and eccentrically aligned compression forces, columns or struts may
be subjected to transverse loading which contributes to buckling and these cases will
also be examined.

~, 10.1 Stability of
In previous chapters a fundamental condition in all the problems was the
equilibrium
equilibrium of internal and external forces. Now, if the system of forces is
disturbed owing to a small displacement of a body, two principal situations
are possible: either the body will return to its original configuration owing to
restoring forces during displacement, or the body will accelerate farther
away from its original state owing to displacing forces. The former situation
is termed stable equilibrium and the latter is termed unstable equilibrium.
Consider the simple case in Fig. 10.1(a) of a vertical bar pinned at the
lower end and carrying an axial tensile force at the upper end. If there is a
slight displacement from the vertical, the force will tend to restore the bar to
its original position. In Fig. 10.1(4), however, the same bar subjected to a
compressive load when displaced slightly from the vertical will accelerate
towards a horizontal position, illustrating unstable equilibrium. A slightly
more sophisticated case is shown in Fig. 10.1(c), where the bar is assisted in
remaining vertical by the action of the horizontal springs. When the bar is
displaced by an amount w in either direction, there is a displacing moment
Px about O and a restoring moment 2KxL, where K is the stiffness of a

Fig. 10.1
 BUCKLING INSTABILITY 265

spring; hence we have

Px < 2KxL — stable

Px > 2KxL — unstable

The critical condition is when

| RIS By or (Pa Ka

and P, is termed the critical load, being the borderline between stable and
unstable equilibrium.

Elastic stability of The instability of structural members subjected to compressive loading may
slender members in be regarded as a mode of failure, even though stress may remain elastic,
compression owing to excessive deformation and distortion of the structure. This mode
of failure is termed buckling and is prevalent in members for which the
transverse dimension is small compared with the overall length.
A theory of buckling for slender columns under axial compression was
developed by Leonhard Euler. In this analysis it will be assumed that the
column is of uniform cross-section and either is ideally straight or has some
defined initial curvature.

Column with pinned The column shown in Fig. 10.2 is pin jointed at each end and it will be
ends assumed to be straight when unloaded. An axial compressive load is applied
with increasing magnitude until the column takes up the deformed shape as
shown. It will be noted that although bending is taking place so that
displacements are in the horizontal direction the notation used is the same as
for a horizontal beam by rotating the column anticlockwise through 90°.

Fig. 10.2
266 MECHANICS OF ENGINEERING MATERIALS
eS ST

At a distance y from the top joint the displacement is w and the bending
moment M4 at section A is Pu in the positive sense. Therefore, using the
form of the equation derived for beams (eqn. [7.3]), we have

EI— = —Pu [10.1]

du Peak ie v
dee et
where k = \/(P/EI). Hence
du 2

The solution of this equation is

u = A cos(ky) + B sin(ky) [10.3]
The boundary conditions are that w = 0 at y = 0 and ZL; therefore
A=) and 0 = B sin(kL)
Now the condition B = 0) merely gives the trivial case of the undeflected
strut but the condition sin(kL)= 0 leads to the solution kL = nz. Buckling
first occurs for n = 1, from which the critical load P, is given by

[Pe \F ‘itd
El Le
or

2k]
ae a [10.4]
If u = Umax at y = L/2, from symmetry, then B = u,,, and the deflection
curve is given by

= thygasta ey) [10.5]

Note that the magnitude of w,,,, cannot be determined from the boundary
conditions, and it can in fact become arbitrarily large leading to elastic
instability of the structure.

Other end conditions Three other boundary conditions for the end restraint of columns are shown
in Fig. 10.3.
Case (a) (one end fixed, one end free) may be treated as a pin-ended strut
of equivalent length 2; hence

_ WEL El
[10.6]
OEY ale
c

Case (b) (one end fixed, the other end only free to rotate) does not have a
readily assessed ‘equivalent length’ pin-ended strut and will be solved from
 BUCKLING INSTABILITY 267

Fig. 10.3

(a) (b) (c)

first principles. The free-body diagram is illustrated in Fig. 10.4(a) and the
bending moment resulting at point A is

M = Pu — Ay
from which

du P ~° Hy
he
lye © Wil
=
EI
[10.7]
Let \/(P/EI) =k. The solution of eqn. [10.7] is
H
u = A cos(ky) + B sin(ky) + ee [10.8]
The boundary conditions are u = 0 at y= 0 and du/dy= 0 at y = L, from
which
H
A=0 a sec(RL)

Fig. 10.4 P P
Mo
et eel

|| ||
A Ries
268 MECHANICS OF ENGINEERING MATERIALS

The deflection curve is thus

. A sin(ky) — if
10.9
4 Pk cos(kL) e joes ee
Finally « = 0 at y = L, and so

Qs = tan(kL) tol

Therefore
tne) = BL [10.10]
The smallest value of kL (other than kL = 0) to satisfy eqn. [10.10] is
kL = 4.49
or

P= 2025 ET [10.11]

This case is therefore approximately an ‘equivalent length’ of 0.7 that for the
pin-ended column.
Case (c) (both ends fixed) is illustrated in Fig. 10.3(c) and the free-body
diagram is shown in Fig. 10.4(4). It may be solved in a similar way to the
previous cases using the boundary conditions that uw = 0 at each end and
du/dy= 0 at each end and at the middle. The solution is
2
eee
[2
[10.12]
It may be seen that this situation is equivalent to a pin-ended strut of length
L/2. Equations [10.6] and [10.12] contrast the differences in buckling load
depending on the end conditions.
In general the critical buckling load for columns can be expressed as
P, = B(EI/L’), where 3 has values as above or other values dependent on
the end conditions. In practice these are rarely definable as ‘pinned’ or
‘fixed’ and the above formulae must only be applied with careful assessment.

210.2 Buckling
The load—deflection behaviour for an ideal Euler strut is illustrated in Fig.
characteristics for real
10.5(a). For applied loads up to the critical value P. small transverse
struts displacements u can be maintained under load in a stable-equilibrium state.

or y\ ah
Fig. 10.5 12]
oO
no}
ise)
[o) °
em | poe |

Euler strut
Pe P,

Real strut

> >

0) Deflection u O Deflection u
(a) (b)
 BUCKLING INSTABILITY 269

However, at P, and above the smallest transverse displacement is unstable

and will rapidly grow to ‘failure’ of the strut. Obviously the strut can only
accommodate a certain deflection up to its elastic limit; thereafter yielding
and ‘failure’ would occur by plasticity.
In the case of a real strut, which incorporates some deficiency such as
eccentricity of loading, deflection will occur from the moment when load 1s
applied as shown in Fig. 10.5(4). The curve becomes asymptotic to the Euler
load at large deflection. Again, this situation will probably not be attained
owing to yielding.
In order to appreciate the significance of stress during buckling
behaviour we consider the Euler equation [10.4]

WEI wEAr
ge ie es IZ

where A is the cross-sectional area and r is the minimum radius of gyration of

the section (see Appendix A). Therefore
P 2
pees a [10.13]
A (L/r)
The ratio L/r is termed the s/enderness ratio, and plotting this against o,
gives a curve known as the Euler hyperbola, as shown in Fig. 10.6.

Fig. 10.6 7) nh
n n
2
wy
2
y
D n
iS Euler
6 | hyperbola
(Se) yp ‘ E

<
0) | Slenderness ratio 0) Strain

Short i Long
Intermediate

Buckling becomes the limiting mode of failure when the buckling stress
given by equation [10.13] is less than or equal to the yield strength oy of the
material, i.e.

a WE
OD ae 2 [10.14]
(L/r)
Alternatively, for a given material, the buckling stress will be less than the
yield strength of the material for pin-jointed columns whose slenderness
ratio exceeds a value which depends on the material properties

Ne [10.15]
if Oy

This represents the range of slenderness ratios from B to C in Fig. 10.6. For
a given material, a limiting slenderness ratio can be established below which
270 MECHANICS OF ENGINEERING MATERIALS

the column will support at least the yield strength of the material and the
design process is simplified. For structural steel (ay =275 MN/m’,
E = 200GN/m‘), the critical slenderness ratio is 85. For columns with
different end conditions, the appropriate equivalent length can be used,
though the above expression is conservative. If the slenderness ratio is
greater than the critical value, the limiting stress will be the buckling stress
rather than the yield strength of the material.
For intermediate values of L/r, from A to B, instability will be
accompanied by yielding. At a particular point on the stress—strain curve the
stiffness is given by E,, known as the tangent modulus. It is found that, in this
elastic—plastic range, buckling loads can be predicted from the original Euler
expression if the ordinary modulus F is replaced by the tangent modulus £;,.
For short columns, buckling instability does not occur but the stress may be
sufficient to cause yielding.
Figure 10.6 also brings out the influence of end condition in relation to
slenderness ratio.

Example 10.1
A steel column is to be a fabricated I-section shown in Fig. 10.7. It is 6m in height
and is fixed at its lower end. It is to carry a design compressive load of 92 KN at the
central axis 0 of the section. Determine the minimum dimensions of the column to
resist buckling for (a) the upper end free, and (b) the upper end pinned but
restrained in the horizontal direction by other structural members. What are the
compressive stresses in each case? E = 200 GN/m?.

Fig. 10.7

AXIOX36
= 6.72 x 10-° mt
a” x.200°x 10?
I for the section in Fig. 10.7 is given by

2x10 xd! | (2d ~ 20) x 10°

12 i
Therefore (1.67d* + 167d — 1670) x 10-* = 6.72 x 10-°,
d*° + 100d = 4.03 x 10°
 BUCKLING INSTABILITY 271

from which d = 159mm and 6 = 318mm.

Cross-sectional area = 6160 x 10~° m?

and
92 000
Compressive stress = ——— x 10° = 14.9MN/m?
6160
Case (4). The Euler buckling load is, from eqn. [10.11],
vf
pane?!L2
Therefore J = 0.84 x 10~° m* and, from the cubic equation for d above, the
solution is
d = 79mm and 6b = 158mm

Cross-sectional area = 2960 x 107° m?

and

: _ 92.000 - ,
Compressive stress = 7060. 10° = 31. MN/m

Example 10.2
A connecting rod in a high-speed mechanism must be 150mm long and has to
support a tensile load of 2kN and a compressive load of 1.7kN. The rod is
connected through circular pins at each end. Decide on an appropriate size and
shape of cross-section if the rod material has a yield stress of 275 MN/m? and a
modulus of 200 GN/m?. |

The requirement to support a tensile load of 2kN implies that the cross-
sectional area A is sufficient to prevent failure by yielding, i.e.

P, > Aoy [10.16]

Failure due to buckling can occur by either of two mechanisms: buckling in
a plane containing the axes of the pin joints at either end of the rod, or
buckling in a plane perpendicular to this. The pins prevent rotation of the
ends of the rod perpendicular to the pin axis, so for this mode of failure the
rod effectively has fixed ends. The required buckling loads are therefore

2 EI,
Bae — : [10.17]
and

2 BT.
Pa ye: a : [10.18]
This implies that the optimum resistance to buckling occurs when J, = 4/,.
A common section which can be dimensioned to meet this requirement is
the H-section.
For a given cross-sectional area, the largest /-values will be obtained
from a very large H with thin web and flanges. However, for rods
272 MECHANICS OF ENGINEERING MATERIALS
a

manufactured by casting or forging, very thin sections are difficult to create.

Rods fabricated from thin sheets of material are also liable to failure by local
buckling or wrinkling, which will be described at the end of this chapter.
For the purposes of this exercise it will be assumed that the aspect ratio
of the largest to the smallest dimension of the web or flange must be at most
the given value x, Fig. 10.8

ean ae gE (10.19]
La") Bes

Fig. 10.8

From this constraint, the minor dimensions d and / of the section can be
found from B, D and «x as

payee and ee oe [10.20]

Me ®
Figure 10.9 shows a spreadsheet in which the tensile and compressive loads
which can be supported by given dimensions B and D can be evaluated.
While these equations and constraints are very difficult to solve exactly, a
trial and error variation of the B and D can be used to find a solution which
meets the requirements.

Fig. 10.9
if 275 a sy 275
2 200000 - eye | 200000
3 3
4 150 150
5 2.49 5 B 2.49
6 4.86 60 p 4.86
2 3.20 mm x 3.20
8 8
9 +B5-B10/B7 9 b 1.46
+B6-2*B5/B7 10 d 3.30
14
12 A +B5*B6-B9*B10 noe A Tat
13 ly — (BS*B643-B9*B1043)/12 13 ly 19.38
14 iz (2*(B6-B10)*B5‘3+B10*B943)/12 14 Iz 4.84
ma fe tee Minimum] | 45 Minimum
16 Pt +B12*B1 | 2000] f@eler | 2000 2000
17 Pc_y @PIM2*B2*B13/B442 | 4700) [Pc _y| 1700 1700
18 Pc_z 4*@PIN2*B2*B14/B442 +C17 18 Pc_z 1700 1700
(a) Data and cell formulae (b) Spreadsheet display
 BUCKLING INSTABILITY 273

Since a connecting rod is subjected to large accelerations it is desirable to

minimize its weight, which for a rod of a given length and material is
equivalent to minimizing the cross-sectional area. The problem can
therefore be described as an optimization problem! where
the goal is to minimize the area;
subject to the constraints of tensile and buckling load being greater than
or equal to the specified values;
by varying design variables B and D.
Most modern spreadsheets include facilities for optimization, and in fact the
values of B and D in Fig. 10.8 were found by optimization. It can be verified
by small changes in B and D that the dimensions shown give the minimum
cross-sectional area for which the load constraints are met.
The minimum cross-sectional area which is necessary to satisfy both
yielding and buckling constraints varies with aspect ratio as shown in Fig.
10.10. For large values of aspect ratio v a change in the aspect ratio makes no
difference to the minimum weight, since the required cross-sectional area 1s
the limiting constraint. It is undesirable to use an unnecessarily very thin-
walled section, since the rod will occupy more space and be more difficult to
manufacture.

Fig. 10.10 Minimum cross-sectional

area 8.5

8
<
Teo)

Re ee a

7 + + + thet + +
2 2.5, 3 3:5 4
Aspect ratio

In Fig. 10.11 it can be seen that, for values of flange and web aspect ratio
less than 3.2, Euler buckling of the rod is the limiting constraint and the
tensile load which can be supported is larger than required. Thus the weight
can be reduced if the area is distributed over a section with thinner walls. In
Figs. 10.10 and 10.11 the smallest aspect ratio at which the minimum cross-
sectional area is achieved was found by including aspect ratio as a design
variable in the optimization and starting the search for the minimum from a
small aspect ratio value. .
Internal combustion engine connecting rods such as that shown in Fig.
4.3 are usually made with H-shaped cross-sections, even though their
slenderness ratios are well below the value at which buckling becomes the
critical mode of failure. One possible reason for this is that if the yielding
load is exceeded, the buckling load becomes proportional to the tangent
modulus, which falls rapidly with increasing plastic deformation. Thus it is
274 MECHANICS OF ENGINEERING MATERIALS
a

Fig. 10.11 Maximum yielding and

buckling loads: variation with 3000
aspect ratio
T Required tensile load

2500
a Required compressive load
36
4H
2000

1500 +++ + : +--+ ap

2, D9 2 3 4
Aspect ratio A.R.

ae Pte Pery Pe_z|

still advantageous to have a section where the ratio of the /-values is 4:1 to
maximize the load at which the rod will collapse.
Thus, using an H-section, the rod should have sufficient cross-sectional
area to prevent overstressing of the material under the largest of the tensile
and compressive loads. It should have an /-value about the pin axis which is
four times the /-value about the perpendicular axis. For the particular rod
length and loading quoted above, the rod dimensions which allow a
minimum cross-sectional area and the thickest web and flanges are:
D=49mm: d=3.3mm; 2 =725 mms b= 1.5mm.

10.3 Eccentric loading of

It is seldom in practice that a column or strut can be loaded exactly along its
slender columns
central axis as the Euler analysis implies. A general solution will now be
developed for the case of a long column subjected to a load parallel to, but
eccentric from, the central axis.

Pp
—
e beh
| .

vi

Consider the column illustrated in Fig. 10.12 displaced by the load P.

Bending moment at D, Mp = —P(e + a — u), where e is the eccentricity, a
is the deflection at the free end and uw is the deflection at y from the free
 BW CE UNGeuN Say AB eT iy 275

end. Then

eo8 = Pleta—s)

du P P
es mre = a + a) [10.21]

ety (2/1) =p. hen

u =A cos(ky) + B sin(ky) +e+a [10.22]
The boundary conditions are as follows:
At y =0, w= a, therefore A = —e and

u = —e cos(ky) + Bsin(ky) +e+a [10.23]

At y = L, du/dy = 0; therefore B = —e tan(kL)
At y = L, u = 0; therefore a = e[sec(kL) — 1]
The deflection curve is therefore
u =e sec(kL){1 — cos[k(L — y)]} [10.24]
The maximum bending stress occurs at the fixed end and is given by

aoe Pec oe

where c is the half-depth of section in the plane of bending, r the radius of

gyration and A the cross-section area. The maximum resultant compressive
stress 1s
P Pec sec(kL) 184 ec sec(kL)
= = 14 10.2
reel Ar al p eee
Hence
A
= 1+
fi
[ec sec(RL)]
/r?
[10.26]

Fig. 10.13
Eccentricity
€4<00< 63
o
Stress
»

ofele

Slenderness ratio L/r

276 MECHANICS OF ENGINEERING MATERIALS

Note that & is a function of P and so P exists in both sides of eqn. [10.26].
Therefore, a value for the critical buckling load can only be found by
iteration.
The effect of eccentricity on maximum stress as a function of slenderness
ratio is shown in Fig. 10.13.

Example 10.3
A vertical steel tube having 75mm external and 62mm internal diameters is 3m
long, fixed at the lower end and completely unrestrained at the upper end. The tube
is subjected to a vertical compressive load parallel to, but eccentric by 6 mm from,
the central axis. Determine the limiting value of the load so that there is no tensile
stress at the base of the tube.
If the column had been loaded along its axis what would be the value of the Euler
buckling load? E = 208 GN/m?.

For zero tensile stress at the base

P Pee sec(kL)
a A y I
Hence

I 0.815 x 10-6
phy a
sec(kL) =7 = 090138 x 0.006 x 0.0375
PEAS
and

k = 0.388
P, = 0.388’ET = 25.6kN
For the axially loaded column

P,= WEI be m x 208 x 10’ x 0.815 x 10-°

4D? 4x9
= 46.5 kN

10.4 Struts having initial

After eccentricity the next practical departure from the Euler idealization is
curvature
that in some cases a column or strut may not be perfectly straight before
loading. This will influence the onset of instability. The strut is illustrated in
Fig. 10.14, in which the initial maximum deflection is ag, the value of the
deflection at D distance y from P is uo, and
a OY
ug = ag sin —
E
When the buckling load P is applied, the deflection at y is increased by u and
the bending moment at this point is Mp, where

Mp = P(u+u) = P(u + ao sin)

 BUCKLING INSTABILITY 277

Fig. 10.14 | f

y
D Wee D

uU ‘ Te (U+Uo)

Umax ha

ie y

Hence
ae
iw = -P(u + ay sin”)

Therefore

du
aie (u + ao sin) =) (10.27]
where k* = P/ET; hence
kag sin(ry/L)
u = A cos(ky) + B sin(ky) + (72/2) — 2

The boundary conditions are that at y= 0 and y=L, u = 0; therefore

A = (0) and B sin(kL) = 0; and since k is not zero, it follows that B = 0.
Therefore

_ kag sin(my/L)
“Ce /T) — 2
Pag Ty
= sin —
T(EI/L2)—P) L
P.
= (Z“.) sin [10.28]
where P, = 7° EI/L”. Substituting for uo,
uo
“= ea [10.29]
278 MECHANICS OF ENGINEERING MATERIALS

Thus the effect of the end thrust P is to increase the no-load maximum
deflection aj by the multiplying factor [(P,/P) — 1].
From eqn. [10.29] it will be observed that, as the value of P approaches
that of P,, the basic Euler buckling load, the value of u increases, tending to
become infinite. At y = iL, the increased deflection is given by
=
a a [10.30]

Plotting values of P and u, the curve shown in Fig. 10.15(a) is obtained.

Failure of the strut would occur before P reached the theoretical value P,.
Equation [10.30] may be written in the form

(P./P)u! — ul = ag [10.31]
which shows that there is a linear relation between w’ and w'/P,
Fig. 10.15(4), the intercept on the axis of u’ being equal to —ap.

Fig. 10.15 Load P4 uA

P
R

= = 7 ee ee eee
10) Deflection u | 10) u'
a
(a) (b)

Let o = P/A and o, = P./A, where A is the cross-sectional area of the

strut. Hence, from egn. [10.28],

i= (—2= -)ay sin

olay
[10.32]

and total deflection at y is given by

oO . my STC
u+ uy = a4) SN ——- ap Sin —
O,—O Ib Lf

Ce ies _ Ty
= ay sin= [10.33]
Oo. —O

The maximum deflection occurs at O, where y = +L; therefore

Oe
Umax = ( )a [10.34]
O.—0

Maximum bending moment is ao{o,/(o.—0)|P and the maximum

compressive stress is given by

Paylo./(o.—a)|c P
,= ; a [10.35]
 BUCKLING INSTABILITY 279

where ¢ is the distance from the neutral axis to the point of maximum
compressive stress. If r is the least radius of gyration of the section then,
substituting P/A =o,

on o( i 1) [10.36]
where 7) = aoc/7?. Taking a, equal to the yield stress in compression, cy,
we obtain a quadratic equation

o —oloy +0-(n+1)| + cyo, = 0 [10.37]

from which the limiting value of o and hence P can be determined.

10.5 Empirical formulae

In practice many strut or column designs do not fall in the category of
for design
slenderness ratio relevant to Euler theory. Consequently a number of
empirical formulae have been devised for different classes of materials
(metal, timber, concrete) to cover the range A to B in Fig. 10.6. There is not
space here to discuss these in detail, but some are given in summary and the
reader is referred to British Standards and other design codes as required.

Rankine—Gordon
px oA
or AL) [10.38]

where symbols have their previous meaning and a is a constant dependent

on end condition and material.
The formula applies for very short columns where buckling is not a
factor, as well as for a range of larger slenderness ratios. Typical values for a
for pin-ended struts are 0.0001 for mild steel, 0.0006 for cast iron, and
0.0001 for timber.
For eccentric loading let the permissible load be P’; then for the elastic
limiting condition

Pa Per
oy =-—-—,
er pe
where 7, =radius of gyration in the plane of bending, and from eqn. [10.38]
we can write

[eer alae
ia Bar
Hence

p= [11+ + (4(al? Ir)

/r’)|P [10.39]
1 + (ec/r;)

Straight line This represents the region A to B in Fig. 10.6 by a straight line expressed as
P=cAll —¢(L/p)| [10.40]
280 MECHANICS OF ENGINEERING MATERIALS

where P = allowable load, o = allowable compressive stress and c= a

constant depending on the material and end restraint which is typically
().005 for mild steel and 0.008 for cast iron.

Parabolic This formula, which is intended to agree with the Euler formula for long
columns, is

P = cA{l — e(L/r)’] [10.41]

where ¢ is a constant; the other symbols have the same meanings as above.
With pin ends and L/r < 150, then for mild steel ¢ may be taken as
0.000 023.

sean Becky's laa Since transverse loading of a slender member causes bending, the effect on
combined compression instability under axial compression is similar to that of initial curvature of
and transverse loading the strut, and consequently a more rapid rate of deflection occurs.
This situation is also described as a beam-column and one example has
already been studied. The eccentrically loaded pin-ended column can be
regarded as a beam—column carrying axial load P together with end
moments P,. Before proceeding to specific transverse loading cases it is
appropriate to derive the differential relationships as was done in Chapter 6
for bending without end load.
Pp
Fig. 10.16
M
Q

dy du) |<

Qt dQ dy ~< an
Yi ev M+ oy a

The equilibrium equations for the element shown in Fig. 10.16 are

w dy + (2+ Bay) O30

and

d dM
—M ~ Q dy + mdy> + (m+ Say) Page 0 Vj

These reduce to

wee
TE oes [10.42]
 BUCKLING INSTABILITY 281

and

dM One du a
0 [10.43]
dy dy
When P = 0) these equations are the same as [6.1] and [6.3]. Eliminating the
shear force Q between these equations

cM au

dtu au
E?—,. +-P—_.= DY
Ge dy?
or

du Pdu_ pw
[10.44]
dyt EI dy? EI
Equation [10.44] is the general differential equation for beam—column-type
situations. If the lateral loading is zero then the equation is of the form that
was used for pure column buckling cases.
The standard form of solution for eqn. [10.44] is

u = A sin(ky) + B cos(ky) + Ciwy* + Chwy + C3

where k = \/(P/EI) and A, B, Ci, Cy) and C3 are constants related to the
boundary conditions.

10.7 Pin-ended strut The bending moment at an arbitrary section A along the beam in Fig. 10.17
carrying a uniformly is
distributed lateral load Wb Z
My Pa 2
ye Z [10.45]

Vu wL wy?
EI = —P, Hee
dy? ‘me yi

du 2 wk?y? —wLky
sts = 10.46
dy? se 2P 2P.
The standard solution for this type of equation is
p ib
u= A sin(ky) + B cos(ky) 4 oe ae aa [10.47]

The boundary conditions are that at y = 0, u = 0; hence B = w/ Pk’; and at

y= Ll, a= 0, Therefore
: D DW
0 =A sin(kL) + —> cos(kL) — PE
Pe
282 MECHANICS OF ENGINEERING MATERIALS

Fig. 10.17

The deflection equation [10.47] becomes

Py ONS wy wly pw
a) (tan sin(ky) + cos(ty)) =P oP OP PR

NG yi 24s ths tererore

ry kL wL?
Umax = De (se 5 1) ap [10.48]

Now, “mar — CO when k(L/2) — (nm/2). Hence

=)
nr
Be = ee

for which the lowest value is

WEI
Ye FT

which is the Euler load for a simple strut without transverse loading.
Although theoretically the load—deflection relationship would become
asymptotic to the Euler load, in fact ‘failure’ is governed by yielding at a
lower load. Substituting the value of w,,,, from eqn. [10.48] and y = L/2 in
 BUCKLING INSTABILITY 283

eqn. [10.45] gives

W kL
Moo ZB (sect — 1) [10.49]

from which, with a limiting value of o, either an allowable value of m can be

determined for a particular end load P, or vice versa.

Example 10.4
P A part of a machine mechanism is illustrated in Fig. 10.18(a). The oscillating end
portions AB and DE may be regarded as rigid, but can pivot freely at A, B, D and E.
The central slender strut BCD has a simple restraint at C. What is the maximum load
necessary to drive the mechanism?

Fig. 10.18

(a) (b)

The deflected position of the members is shown in Fig. 10.18(4) with the
appropriate forces acting on BCD. The basic differential equation is

Pé
EI— = P(6—1u) ong ae
284 MECHANICS OF ENGINEERING MATERIALS
nnn

The solution of this equation is

, 6
u = A+B sin(ky) + C cos(ky) + 7

The boundary conditions are (i) y=0, M =0; (ii) y =L, du/dy = 0;
(ili) y = 0, u = 6; (iv) y = L, u = 0. From (iti) and (iv),
6=A+C and 0=A+Bsin(kL)
+ C cos(kL) + 6
From (ii),
du 6
— = Bk cos(ky) — Ck sin(ky) +—
dy L
6
0 = Bk cos(kL) — Ck sin(kL) + Z

From (i)

a = —Bh sin(ky) — Ck’ cos(ky)

0 = —Ck
Hence:G = 0, A= 6; and
R= 6 a 26
kL cos(kL) sin(kL)
from which

tan kl — ZeL,

and

kE = 1.166
Hence the load required to drive the mechanism is

1s66\7
P ey (enn 304EI
\e 1)gw Te

10.8 Other examples of

instability Instability is primarily a function of geometrical proportions, and in the case
of columns this was expressed in terms of slenderness ratio in relation to
compressive forces. Two other examples of members which become
unstable with increasing load are shown in Fig. 10.19(a) and (0).
In the first case, because of the small width-to-depth ratio of the beam it
is not possible to maintain the initial plane of bending, and the cross-section
twists out of plane as shown. In the second case we again have a slenderness
ratio influence in that the length of the bar under torsion is large compared
with the diameter and it is not possible to maintain a straight axis of twist.
Instability occurs through the shaft adopting a spiral axis when the torque
reaches a critical level. There is not space here to analyse these cases but it is
 BUCKLING INSTABILITY 285

Fig. 10.19

Instability
SSSe SS

Instability

(a) (b)

important for the designer to be aware of the possibility of these modes of

deformation.

10.9 Local buckling

In structures formed from thin sheets of material there is an additional
possible mode of instability known as local buckling. When a thin
rectangular sheet is subject to compressive stress it may buckle in a series
of waves as illustrated in Fig. 10.20. The analysis of the critical buckling
stresses is similar in principle to the buckling of a column, though the details
are too complex to describe here.
Tables and formulae for the critical compressive stresses are given for a
range of loading and boundary conditions in reference books such as Roark’s
Formulas for Stress & Strain by W.C. Young’ or Formulas for Stress, Strain,
and Structural Matrices by W. D. Pilkey*. Formulae for some simple cases
where the sheet is very much longer in one direction than the other are
given in Fig. 10.21. Using these, measures similar to the slenderness ratio of
columns can be derived which indicate safe designs where buckling should
not be the critical mode of failure.
For example, in an I-beam loaded in bending it is possible that the flange
on the compression side may fail owing to local buckling before the yield
strength of the material is approached. The critical buckling stress in a
rectangular strip of material is
be fe
2
=K (;)
1—1*\b |
10.50 |
where K is a constant which depends on the dimensions of the plate and the
boundary conditions. Requiring that this stress is greater than the yield
strength of the material implies that

< Gre al [10.51]

Fig. 10.20
Treating half the flange as a long strip which is fixed at one long edge (to the
web) and free at the other, K =1.21. For structural steel with
E =200GN/m? and o, = 275MN/m’ this implies that b/t < 25, ice.
the half-width of the flange should not be more than 25 times its thickness
or the flange will locally buckle before the stress reaches the material yield
strength. In fact local buckling is very sensitive to curvature or geometrical
imperfections in the plate or residual stresses, so that the recommended ratio
in the British steel design code is only 15 for an ‘outstand’ element of a
286 MECHANICS OF ENGINEERING MATERIALS

Rectangular plate, a> b>t Circular plate, r >t

Compressive buckling stress Uniform edge compression

1 ONY
/
=K-—,
r
(>) eos ()
1-1? \r
e 2 WA

(a) fixed on two sides Circular tube, 1 >> \/rt

IK oH

(a) longitudinal compression

(b) fixed on one side f 1 ie t
Keele, 3V1—-1 4

Shear buckling stress

era -« 56)IE iN?

(a) simply supported edges where

K = 1.27+ V9.64 + 0.466H!
KA
and Pp
H=vV1-v—
rt

(b) clamped edges (c) external pressure

Ko 774 pbs!
gent 5a;
PS4T-Pr

Fig. 10.21 Local buckling stresses in section in this material and 39 in an ‘internal’ element which is supported on
a sheet both sides. Similar logic can be used to decide on minimum thicknesses of
any material for any of the cases in Fig. 10.21 to ensure that buckling is not
the limiting mode of failure.
Aircraft structures are perhaps the best example of structures where the
prevention of local buckling is a critical design issue. The surface skins will
be a millimetre or less in thickness, and in areas subjected to tensile loading
there is no problem. However, all parts of the aircraft are subjected to some
kind of bending action in flight and so some skins will be subject to
compressive loading. The buckling tendency of the skin has to be limited by
frequent stiffeners, e.g. ribs, stringers, etc. Even the web of a spar has to
have a series of stiffeners in order to carry shear, since this gives rise to
diagonal tension and compression, the latter causing wrinkling.
Thin sheet members subjected to compression or torsion may be stable
in respect of the overall geometry, but will reveal local buckling instability
characteristics as shown in Fig. 10.20.
 BUCKLING INSTABILITY 287

10.10 Summary
Instability in structural elements has been shown to be an important factor
in design as it is a mode of ‘failure’ dependent largely on compressive
loading and geometrical proportions. The critical compressive load for the
buckling of columns or struts can be expressed as
BEA
~ (L]r?
where (3 is a constant depending on the material and end conditions. The
idealized Euler theory is only applicable at large slenderness ratios and ‘real’
struts will ‘fail’ owing to exceeding the yield stress of the material long
before attaining the Euler load. Eccentricity of loading, initial curvature and
transverse loading can each contribute to a lowering of the allowable
‘buckling’ load as a function of the yield stress. Empirical formulae and
design codes are now established for the design of columns for structural
situations.
Local buckling of thin sheet material is also an important design
consideration for which specialized texts such as Theory of Elastic Stability
by Timoshenko and Gere*, and design data sheets, should be consulted.

References
— Arora, J. S. (1989) Introduction to Optimum Design, McGraw-Hill
International, New York.
2. Young, W. C. (1989) Roark’s Formulas for Stress & Strain, McGraw-
Hill International, New York.
3. Pilkey, W. D. (1994) Formulas for Stress, Strain, and Structural
Matrices, John Wiley, New York.
4. Timoshenko, S. P. and Gere, J. M. (1961) Theory of Elastic Stability,
2nd edition, McGraw-Hill, London.

Problems 10.1. A machine mechanism consists of two rigid members each of length
400 mm connected by a frictionless hinge at B and pinned at A and
D as illustrated in Fig. 10.22. A spring of stiffness 30 N/mm is
attached to the lower member at C as shown. Determine the critical
load, P, for the system.
200 10.2. A vertical mechanism linkage consists of a slender member of length
500 mm and stiffness 500 Nm”, built in at the lower end and pinned
at the upper end to a rigid member of length 250mm. The upper
end of the latter is pinned between rollers which are axially aligned
200 with the whole strut. Determine the critical compressive load, when
applied at the roller bearing, which will cause buckling.
10.3. A straight slender column of height 2.77 m is fixed at the lower end
and is entirely free at the upper end. The design criterion is to limit
200 the maximum compressive strain prior to buckling to 0.0008.
Determine the required least radius of gyration.
10.4 A thin-walled square tube, Fig. 10.23(a) of length L is to be designed
to support a compressive load P. The lower end of the tube is fixed.
200 In terms of the tube dimensions L, 4 and ¢ and its material
properties, write expressions for (i) the load at which the stress in the
tube exceeds the yield stress 0, of the material, (11) the load at which
Euler buckling occurs (take the second moment of area J for the
Fig. 10.22
288 MECHANICS OF ENGINEERING MATERIALS

cross-section as h3t / 2), and (iii) the load at which local buckling of
the tube wall occurs. Assume that the factor K in eqn [10.50] is 4 in
this case.
10.5 For a column 2m high which must carry a load of 4000N of
structural steel (E = 200 GN/m?, oy = 200 MN/m’, v = 0.3), (i)
calculate the minimum cross-sectional area of the column if the
Euler buckling and yielding criteria are to be satisfied simulta-
neously, and (ii) calculate the minimum cross-sectional area if the
local and Euler buckling conditions are to be met simultaneously.
10.6 Use a spreadsheet optimization to minimize cross-sectional area
subject to the constraints that all the failure loads in Problem 10.4(i)—
(iii) are greater than P for the column of Problem 10.5.

|
Local buckling when 0 = oraes iA
E(t’
Fig. 10.23 -v

10.7 A 4m long strut has the cross-section shown in Fig. 10.24. Calculate
the Euler buckling load for the strut if it has fixed ends. The
Young’s modulus for the strut material is 208 GN/m’, ie
43.6 mm, 7, = 36.7 mm, where r is the radius of gyration.
10.8 The column shown in Fig. 10.25 has pinned ends. The upper part,
of length (1 — )L, which is slender and of constant stiffness EJ, is

Fig. 10.24

120 mm
 BUCKLING INSTABILITY 289

Pp
fixed to the lower part, of length nL, which is rigid. Show that, at
instability, tan {A(1 — n)L} = —knL. What is the critical load if the
upper and lower parts are of equal length?
10.9 In a temperature-control device a copper strip measuring 8mm x
4mm and 100mm long is pinned at each end. How much axial
pre-compression is required so that buckling will occur after
a temperature rise of 50°C? E = 100 GN/m’, a = 18 x 10° per
deg C.
10.10 A strut 2m long and pinned at both ends is subjected to an axial
(1_n)L compressive force. If the strut cross-section is that shown in Fig.
10.26, calculate the Euler buckling load. E = 207 GN/m’.
10.11 A circular steel column has a length of 2.44_m, an external diameter
101 mm and an internal diameter of 89mm with its ends position
fixed. Assuming that the centre-line is sinusoidal in shape with a
maximum displacement at mid-length of 4.5mm, determine the
maximum stress due to an axial compressive load of IOKN.
E = 205 GN/m’.
nl 10.12 The compressive stress necessary to cause buckling in a long plate
with one of its unloaded edges fixed so that it cannot rotate is

C=
Eh?
has
Fig. 10.25
where / is the plate thickness and 3 is its width, and k = 1.32 when
the plate is very long.
(a) Using this information, estimate the maximum width to
thickness ratio b/h that can be used in an angle section if the
section must yield rather than locally buckle at maximum load,
Fig. 10.27.
(b) What will this ratio be for (1) aluminium alloy (a, =
420 MN/m’, E=70 GN/m’), (ii) structural steel (oy =
350 MN/m?, E = 200 GN/m’).
(c) Design an angle section of minimum weight in the aluminium
alloy below to carry a compressive load of 10kN which (1) will

=i
Fig. 10.26

10 mm

60 mm
290 MECHANICS OF ENGINEERING MATERIALS

Fig. 10.27

not yield, (ii) will not locally buckle, and (iti) has the largest
second moment of area possible while still satisfying (1) and (11).
10.13 A column is made up of two identical steel angle sections, as shown
in Fig. 10.28 which are fixed at the lower end and free at the upper
end. If the length of the column is 2.5m and it is subjected to a
compressive load of 8kN calculate the safety factor in relation to
buckling if:
(a) the two angle sections are touching along AA but not connected
in any way;
(b) the two angle sections are fastened together along AA over the
full length of the column.
The Young’s modulus for steel is 207 GN/m/? and the radius of
gyration about xx or yy is 9.24mm and about zz is 5.92 mm.

Fig. 10.28

30

10.14 A tubular cast-iron column 5m long has fixed ends and an external
diameter of 250mm. Calculate a suitable tube thickness if the
column supports a load 1 MN. Assume a constant of 1/6400 in the
Rankine formula and a stress of 80 MN/m*.
10.15 A column of length L is pinned at each end and is subjected to an
axial compressive load P. A horizontal force F is now applied at mid-
height to the column. Show that the maximum bending-moment is
 BUCKLING INSTABILITY 291

obtained in the form

FP VE,
Cs — y, tan (=) where be 4) (2 ED)

10.16 A horizontal beam of length 3.6m is simply-supported at each end

and is tubular, having internal and external diameters of 46 and
50 mm respectively. It is subjected to axial compression of 5 kN and
uniformly-distributed transverse loading of 50 N/m. Determine the
maximum surface stress. E = 200 GN/m’?.
10.17 A pin-jointed truss is to support a load W at a horizontal distance }
from a wall, Fig. 10.29. The member BC is a bar of circular section
with radius r. Create a spreadsheet to determine the radius which is
required to prevent bar BC from buckling. Given a material
modulus E£, evaluate the total volume of the bar for values of / in the
range 0.3-0.8, with b = 1, W = 1000, E = 200 GN/ m”. Estimate the
value of h which gives a truss of minimum weight.

Fig. 10.29
CHAPTER
Stress and Strain
Transformations

The preceding chapters have been concerned with ‘one-dimensional’ problems of

stress and strain, i.e. in any particular example consideration has only been given to
one type of stress acting in one direction. However, the majority of engineering
components and structures are subjected to loading conditions, or are of such a
shape that, at any point in the material, a complex state of stress and strain exists
involving normal (tension, compression) and shear components in various directions.
A simple example of this is a shaft which transmits power through a pulley and belt
drive. An element of material in the shaft would be subjected to normal stress and
shear stress due to bending action and additiona! shear stress from the torque
required to transmit power. It is not necessarily sufficient to be able to determine the
individual values of these stresses in order to select a suitable material from which to
make the shaft because, as will be seen later, on certain planes within the element
more severe conditions of stress and strain exist. The identification of these
maximum stresses is an essential step in the design process.
In practice, however, it should be noted that stress cannot be measured directly.
It can only be calculated from a knowledge of the strains in the material. These may
be measured easily using electrical resistance strain gauges and generally a rosette
of such gauges is used to measure strains in precise directions on the surface of the
component. This chapter describes how the information from strain gauges may be
analysed.
Finally the chapter introduces the way in which stresses and strains in fibre
composite materials may be analysed. These materials are different to anything
considered so far in the sense that they are anisotropic, i.e. they have different
properties in different directions. However, it will be seen that the procedures
developed in the chapter for the transformation of stresses and strains to different
planes is directly applicable to fibre composites.

11.1 Symbols, signs and

Since conditions will be studied in which several different stresses occur
elements
simultaneously, it is essential to be consistent in the use of distinctive
symbols, and a sign convention must be established and adhered to. A
subscript notation will be used as follows:
Direct stress: “G3, Oy, Oz

where the subscript denotes the direction of the stress.

Shear stress. Tay, Tyzs) Tye, Teys Taes Tee

where the first subscript denotes the direction of the normal to the plane on
which the shear stress acts, and the second subscript the direction of the
shear stress.
As in the previous work, tensile stress will be taken as positive and
compressive stress negative. A shear stress is defined as positive when the
 STRESS AND STRAIN TRANSFORMATIONS 293

Fig. 11.1

direction of the stress vector and the direction of the normal to the plane are
both in the positive sense or both in the negative sense in relation to the co-
ordinate axes. If the directions of the shear stress and the normal to the
plane are opposed in sign, then the shear stress is negative. Pairs of
complementary shear-stress components are therefore either both positive
or both negative.
The angle between an inclined plane and a co-ordinate axis is positive
when measured in the anticlockwise sense from the co-ordinate axis.
A general three-dimensional stress system is shown in Fig. 11.1.

Plane stress If there are no normal and shear stresses on the two planes perpendicular to
one of the co-ordinate directions, which implies that the complementary
shear stresses are also zero, then this system is known as plane stress (see
Section 3.7). It is a situation found, or approximately so, in a number of
important engineering problems. The analysis of complex stresses which
follows is only concerned with plane stress conditions.

11.2 Stresses on a plane

In preceding chapters the analysis has dealt with stress set up on a plane
inclined to the direction of perpendicular (normal stress) or parallel (shear stress) to the direction of
loading loading. However, if a piece of material is cut along a plane inclined at some
angle @ to the direction of loading, then, in order to maintain equilibrium, a
system of forces would have to be applied to the plane. This implies that
there must be a stress system acting on that plane.
In Fig. 11.2 a bar is shown subjected to an axial tensile force F. The area
of cross-section normal to the axis of the bar is denoted by A. If the bar is
cut along the plane AB, inclined at an angle @ to the y-direction, then the

Fig. 112
294 MECHANICS OF ENGINEERING MATERIALS

force F applied to this portion of the bar can be reacted by component forces
F,, normal and F, tangential to the plane AB. Thus for equilibrium,
FE =F cos. and F.=
F sn 0

The area of the plane AB will be the area of the bar normal to its axis
multiplied by sec 6.
Denoting on the plane AB the positive direct stress by o,, and the
positive shear stress by 7,, then
| ae ee cos@ F
= = ae SOR 2 YG
Be Asec@ Asecd Vs
and

F, —FsnO —F., 6 cos 6

i = ————_ = — sin cos
ae sec@ Asecé A
F/A is the direct stress normal to the axis of the bar and is equal to oy.
Therefore

On = 0, C08"0 [11.1]
and

=], = 0, sin cost = 5x sin 26 [11.2]

Note that, in eqn. [11.1], when 6 = 0, o,, is a maximum and equal to o,,
and when 6 = 90°, a, is zero, indicating that there is no transverse stress in
the bar.
Again, in eqn. [11.2], the magnitude of 7, will be a maximum when sin 20
is a maximum, 1.e. when 26 = 90° and 270°, or 6 = 45° and 135°; the value
of 7, is then 5x on the planes prescribed by 6 = 45° and 135°. This result is
borne out in practice, and for materials whose shear strength is less than half
the tensile strength, direct tensile loading results in failure along planes of
maximum shear stress.

11.3 Element subjected

The rectangular element of material of unit thickness, Fig. 11.3, is subjected
to normal stresses
to tensile stresses in the x- and y-directions as shown.

Fig. 113
 STRESS AND STRAIN TRANSFORMATIONS 295

Considering a corner cut off the element by the plane AB inclined at 4 to

the y-axis, then positive normal and shear stresses will act on the plane as
shown. For equilibrium of ABC, the forces on AB, BC and CA must also be
in equilibrium. As the element is of constant unit thickness, the areas of the
faces are proportional to the lengths of the sides of the triangle. Resolving
forces normal to the plane AB,
o,AB — 0,BC cos 6 — o,ACsin@ = 0 [11.3]
Dividing through by AB,
AC
On — Ox ap 0088 ~ Oy ap sine =

Therefore

On = 0, C08? 6 + Oy sin’ 6 = +(ox + d,) + 5(a, — oy) cos 26

[11.4]
Resolving forces parallel to AB,

T,AB + 0,BCsin 6 — 0,AC cos 6 = 0 [11.5]

Dividing by AB,
BC A
T+ Os ap sing - Oy aR os =—a()

—T, = 0, cos @sin# — 0, sin@cos# =4(o,—o,)sin26 [11.6]

If o, is made zero, eqns. [11.4] and [11.6] reduce to eqns. {11.1] and
[11.2] for the normal and shear stresses on a plane inclined to the direction
of loading.

11.4 Element subjected

The rectangular element of the previous section is now considered with
to shear stresses
shear stresses on the faces instead of normal stresses. The plane AB inclined
at 6 to the y-axis. Fig. 11.4, is subjected to positive normal and shear stresses
as previously. Consider the equilibrium of the triangular portion ABC.
Resolving forces normal to the plane AB,
o,AB — T,,BCsin 6— 7,,AC cos@ = 0 [11.7]

Fig. 11.4

T
T xy
296 MECHANICS OF ENGINEERING MATERIALS
Le

Dividing through by AB as before,

On = Txy COS O sin 0 + 7,, Sin

0cos [11.8]
But, from a consideration of complementary shear stresses,

Txy — Tyx [11 9

Therefore

GC, = 1s, si Ocos 0 = 7,,8in 20 [11.10]

Resolving forces parallel to the plane AB,
T,AB — T,,BC cos 6 + 7,,ACsin 0 = 0 [11.11]

Dividing by AB,
B AC
™ Tay apy C089 - Tyr apn? = 0

Therefore

=A, Sys sin’ 6 — Ty cos’ 6 = —Tzy COS 20 [11.12]

11.5 Element subjected

2 f The general two-dimensional stress system shown in Fig. 11.5 may be
to general two-dimensional obtained by a summation of the conditions of stress in Figs. 11.3 and 11.4.
stress system Hence the equations obtained for o, and 7, under normal stresses and shear
stresses separately may be added together to give values for the normal and
shear stress on the inclined plane AB, Fig. 11.5, in the general stress system.
Therefore, from eqns. [11.4] and [11.10],

On = 5(0% + oy) +4 (oy — oy) cos 20 + 7,, sin 20 [11.13]

and from eqns. [11.6] and [11.12]

T, = —4 (0, — 9,) sin 20 + T,, cos 20 [11.14]

Fig. 115
 STRESS AND STRAIN TRANSFORMATIONS 297

The validity of this method of superposition may be checked by

considering the equilibrium of the triangular element of Fig. 11.5 and
resolving forces perpendicular and parallel to the plane AB as was done in
the previous sections.
It must be remembered that the signs in eqns. [11.13] and [11.14] are
dependent on the directions chosen for the arrows (indicating stress) on the
element in Fig. 11.5. If for any reason the directions of the stresses are
different, e.g. compression instead of tension, then the stress components in
eqns. [11.13] and [11.14] must be changed entered as negative quantities.

Example 11.1
A marine propeller shaft of 200 mm diameter is subjected to a torque of 126kNm
and a pure bending moment of 157 kNm. During inspection, a small surface crack is
observed at 60° to the longitudinal axis of the shaft, Fig. 11.6. Determine the
normal and shear stresses at the crack as it passes through positions A, B and C
during rotation of the shaft.

Fig. 11.6

This is an example to illustrate the common engineering situation of shafts

subjected to combined bending and torsion.
The second moments of area are as follows:

nd* 7(0.2)* ee
P=7 = = 78.54
x 10m

nd*
J= aoe 157.1 x 10°°m‘*

Using the appropriate equations for bending and torsion,

My 157510? 011 ‘
me TT 78.54 x 10-6 a
Ee Sco 3 oO 0.1
a > T157. 120 *
The stress components at A, B and C are

o4=+200MN/m?, = +4,xy = 80 MN/m’”

o®i =0, 7?xy — 80 MN/m’”

o¢ =—200MN/m*, ty, = 80 MN/m’

298 MECHANICS OF ENGINEERING MATERIALS

Using eqns. [11.13] and [11.14] and noting that the correct value of 6 on the
element is 30° as shown in Fig. 11.7,

a! = 100 + 100 cos 60° + 80 sin 60°

= +219.3 MN/m’?

74 = —100 sin 60° + 80 cos 60°

5

71 = —46.6 MN/m’
By a similar analysis we find

o? =+69.3MN/m? =? = +40 MN/m’?

o& = —80.7MN/m? sr = +126.6 MN/m*

The stress components above are illustrated acting on the planes in Fig.
11.7. This example is extended in relation to principal stresses in Example
eS

Fig. 11.7 ————> _*0

200
~<-

DN
80

(a)

11.6 Mohr’s stress circle

Considering eqns. [11.13] and [11.14] once more and rewriting,

On — 5(Ox + Fy) =4(o% — Fy) cos 26 + Ty sin 20 [11.15]

R= = }(@, — @,) sim 20 -F 7, cos 20 [11.16]

Squaring both sides and adding the equations,

[on ca 5(om a o,)] a ie a 1(ox - oy)” ee ies [11.17]

This is the equation of a circle of radius

2
Vice =p) tee)
and whose centre has the co-ordinates

b (oy + Oy )h 0}

The circle represents all possible states of normal and shear stress on any
plane through a stressed point in a material, and was developed by the
German engineer Otto Mohr. The element of Fig. 11.5 and the
corresponding Mohr diagram are shown in Fig. 11.8.
 STRESS AND STRAIN TRANSFORMATIONS 299

Fig. 11.8

6, Cr
(2)

3
3 _ Normal
3 |0 stress
oO
JAS
Yn

4
a \Ohe om) |
< >|

The sign convention used on the circle will be, for normal stress, positive
to the right and negative to the left of the origin. Shear stresses which might
be described as trying to cause a clockwise rotation of an element are plotted
above the abscissa, and shear stresses tending to cause anticlockwise rotation
are plotted below the axis.
It is important to remember that shear stress plotted, say, below the o-
axis, although being regarded as negative in the circle construction, may be
either positive or negative on the physical element according to the shear-
stress convention previously defined. Likewise, positive shear stress on the
circle may be either positive or negative on the element.
The diagram is constructed as follows: using co-ordinate axes of normal
stress and shear stress, both to the same scale, point B (ay, Txy) is plotted
representing the direct and shear stress acting on the plane BC of the
element. Assuming in this case that a < o,, the point A (ay, Ty) is plotted
to represent the stresses on the plane AC of the element. The normal stress
axis bisects the line joining AB at C, and with centre C and radius AC a
circle is drawn.
An angle equal to twice that in the element, i.e. 20, is set off from BC in
the anticlockwise direction (the same sense as in the element), and the line
CD then cuts the circle at the point whose co-ordinates are (0, 7;). These
are then the normal and shearing stresses on the plane AB in the element.
The validity of the diagram is demonstrated thus:

On = OC + CE =3 (0, + oy) +7 cos(2¢ — 26)

= 5(ox + a,) +rcos2@cos2¢ + rsin 26 sin 2¢

But

Q Txy
cos 20 = and sin2¢@ = —;

Therefore

One a (x + oy) +3(o% — oy) cos 20 + 7,, sin 20

7, = DE = r sin(2¢ — 20)

= r(sin2¢ cos 20 — cos 2¢ sin 26)

300 MECHANICS OF ENGINEERING MATERIALS

Substituting for cos 2¢ and sin 2¢,

T, = —} (6% — oy) sin 20 + Ty cos 20

These expressions for o, and 7, are seen to be the same as those derived
from equilibrium of the element. Thus, if at a point in a material the stress
conditions are known on two planes, then the normal and shear stresses on
any other plane through the point can be found using Mohr’s circle.
Certain features of the diagram are worthy of note. The sides of the
element AC and CB, which are 90° apart, are represented on the circle by
AC and CB, 180° apart. A compressive direct stress would be plotted to the
left of the shear-stress axis. The maximum shear stress in an element is
given by the top and bottom points of the circle, i.e.

Tmax = +y/[j (o, iz oy a ipa [11.18]

and the corresponding normal stress is 5(a, + a,). The angle @ to the plane
on which a maximum shear stress acts is obtained from the circle as

tan 26 = tan(90° + 2¢) = —cot2¢

Therefore

tan 20 = -(— [11.19]

Lie

The second plane of maximum shear stress is displaced by 90° from that
above.

Example 11.2
At a point in a complex stress field o, = 40MN/m?, o, = 80MN/m? and 7,,
= —20MN/m?. Use Mohr’s circle solution to find the normal and shear stresses on
a plane at 45° to the y-axis.

Fig. 11.9

Stresses in MN/m?
80
(a) (c)

The stresses on the element are shown in Fig. 11.9(4) noting that the shear
stresses are in the negative direction. The corresponding Mohr’s circle is
shown in Fig. 11.9(a). From Fig. 11.9(a) for 6 = 45° on the element which
is 90° on the circle,

On = 40 MN/m? and T, = 20MN/m

Note that 7, will be upwards to the left as shown in Fig. 11.9(c) because it is
below the axis on the circle (i.e. would turn the element anticlockwise). The
 STRESS AND STRAIN TRANSFORMATIONS 301

values of o,, and 7, may be checked by calculation using eqns. [11.13] and
[11.14].
Example 11.3
Construct a Mohr’s circle for the following point stresses: co, — 60 MN/m?, Oy
— 10MN/m? and Ty = +20 MN/m?, and hence determine the stress components
and planes in which the shear stress is a maximum.

The stresses on the element are as shown in Fig. 11.10(4) and the
corresponding Mohr’s circle is shown in Fig. 11.10(q).

Fig. 11.10

Stresses in MN/m?

(a) (b) (c)

From Fig. 11.10, the normal and maximum shear-stress components are

On =35MN/m? = and = 7, = 32 MN/m”

acting on the planes 6; = 64.5°, 02 = 154.5°. The orientation of the element
experiencing these stresses is shown in Fig. 11.10(c).

% 11.7 Principal stresses

It has been shown that Mohr’s circle represents all possible states of normal
and planes
and shear stress at a point. From Fig. 11.8 it can be seen that there are two
planes, QC and CP, 180° apart on the diagram and therefore 90° apart in the
material, on which the shear stress 7, is zero. These planes are termed
principal planes and the normal stresses acting on them are termed principal
stresses. The latter are denoted by o; and @>2 at P and Q
respectively, and are
the maximum and minimum values of normal stress that can be obtained at a
point in a material. The values of the principal stresses can be found either
from eqn. [11.17] by putting 7, = 0, or directly from the Mohr diagram;
hence

Oo ano
Se. oy) +472] [11.20]

ea)
Opa OF
[11.21]
where o is the maximum and g2 the minimum principal stress. The planes
are specified by

20=26 and 180°+2¢

302 MECHANICS OF ENGINEERING MATERIALS

or
ae and 90° + ¢
But from geometry in Fig. 11.11(a)

Gari eeee fret BE ty

ces Ngee NI |Si oe a
Therefore

We
d= 4 tan! (=) and 90° +5 tan! (= [11.22]
Ox — Oy \Fx = Oy
Thus the magnitude and direction of the principal stresses at any point in a
material depend on ¢,, oy and 7, at that point, Fig. 11.11.

ry

Fig. 11.11

Ee Normal
(e)
Stress

Shear
stress

(a) (b)

11.8 Maximum shear

It was shown earlier, eqn. [11.18], that the maximum shear stress at a point
stress in terms of principal is given by
stresses
Tmax = £5) [(0x — oy)’ + 44
If expressions [11.20] and [11.21] for 7, and o2 are subtracted, then

01 — 0, =+yV|(o, — 0,)’ +477]

xy [11.23]
and therefore

Tae = 3(01 — 2) [11.24]

It should be noted that principal stresses are considered a maximum or
minimum mathematically, e.g. a compressive or negative stress is less than a
positive stress, irrespective of numerical value.
In Mohr’s circle the principal planes PC and QC in Fig. 11.8 are at 90°
to those of maximum shear stress, UC and VC, and therefore in the material
the angles between these two sets of planes become 45°, or the maximum
shear-stress planes bisect the principal planes.

11.9 General two-

In the preceding paragraphs certain specific stress conditions which exist on
dimensional state of stress planes in an element subjected to normal and shear stress have been derived,
at a point and these are shown on elements of the material in Fig. 11.12. Although it
 STRESS AND STRAIN TRANSFORMATIONS 303

Fig. 11.12

has been convenient in the foregoing analyses to draw an element of material

it must be remembered that the stresses actually act at a point in the material.

Example 11.4
P Determine the principal stresses and maximum shear stresses at points A and B by
calculation and at point C by a Mohr’s circle construction for the I-section beam
shown in Fig. 11.13.

120 kN
Fig. 1113 | EERE 2
—

200

The reaction at the left-hand end is 40 kN and hence at the required cross-
section of the beam

M=40x1=40kNm Q=40kN J = 61867x 10'mm4

ya = +120mm, ye — 0, yo = +100mm

The bending stresses are, from eqn. [6.10],

4 Mya a_ 40x 10° x (+0.12) = 77.5 MN/m

<a 61867 x 10-9 a
304 MECHANICS OF ENGINEERING MATERIALS

0)

C
OP se +77.5 5 x tiie
0.12 +64.5 MN/m?

The shear stresses are, from eqn. [6.42]

2
ile QA 40x 103 = (0.2)’ 0.1
0.242 . — 0.2 ;
Ty =r = 6rge7x1o> \ 8 *+8x0.02! )

= 10.34 MN/m?

C = 40 x 10° x 2000 x 10-° x 0.11 = 7.11MN/m 2

Ty 0.02 x 61867 x 10-9 ;

40 x 103 x (0.24 — 0.02)(0.1 — 0.02)

— = 2.84 MN/m*
Try 4 x 61867 x 10-9 ae
Using eqns. [11.20] and [11.21], for the principal stresses
Tis 43 (77.9 +4 x 2.84)? = +77.6MN/m?
of =~

of = Tis
—— — 1(77 5? + 4 2.847)" = 0 1MN/
At B, o, = 0, = 0; hence
oy = +1, = +10.34MN/m’, 07 = —7,, = —10.34 MN/m?
Using eqn. [11.24] for the maximum shear stresses,

16 (S01
Amax = te D)
= 38.8 MN/m?

1034 (= 1034
B=
max : ) - 10.34MN/m?

Fig. 11.14 A Tmax = 33 MN/m?

(64.5, 7.11)
Ms
oO, oO

(0, 7.11)

Trax = 33 MN/m?
 STRESS AND STRAIN TRANSFORMATIONS 305

The Mohr’s circle construction is shown in Fig. 11.14, for point C, from
which the principal stresses and maximum shear stress are

of =+65.3MN/m’, of =-0.77MN/m’, 71°, = 33MN/m?

max

Example 11.5 ; : i : es
Derive expressions, in terms of bending moment M and torque T, for the magnitude
and direction of the principal stresses at points A, B and C on the shaft in Example
11.1.

The maximum bending and shear stresses occur at the outer surface of the
shaft and are given by
NESZAL 16T
Ox an eg ——
* rd
where d is the diameter of the shaft.
The stress components on an element of material at the surface points A,
B and C are shown in Fig. 11.7. The shear stresses are the same in each;
however, the bending stress is maximum tension at A, zero at B (the neutral
plane) and maximum compression at C. The principal stresses at these three
points are therefore

At A 01,02 =h0, th (e+ 472.)

AtB 0] = —02 = Try

AtC 91, 02 =~30x $9 V(9, +475) ay

At B, therefore, there is a state of pure shear.

The principal stresses can be expressed in terms of the bending moment
and torque by substituting for a, and 7,.:

16M 1GAEN” (167 \"

AtA =—_+

16
Sse J(M? + T’)] [11.25]

ae ey malee
Fg de ee
At C the values are the same as at A but are negative.
The inclinations of the principal planes to the z-axis are
If iF
fe ttan | (=) and 90° + Stan | (a)

The maximum shear stress is given by

1
i 1 ae 4) [11.26]
Td
306 MECHANICS OF ENGINEERING MATERIALS

11.10 Maximum shear

So far the analysis of complex stress has been restricted to the two-
stress in three dimensions
dimensional situation. Therefore the value of the maximum shear stress
obtained from the equations or the Mohr diagram represents the maximum
shear stress in the (x, y)-plane. However, even for a state of plane stress this
may not be the maximum shear stress in the material. To obtain the true
maximum shear stress for use in design calculations (see Chapter 12) it is
necessary to consider all three principal planes. The three-dimensional
element subjected to the principal stresses is shown in Fig. 11.15. The
principal stress o3 is zero in this particular case since we have been
concerned only with plane stress situations. For convenience, consider the
axis of the three principal stresses to be labelled 1, 2 and 3. Considering each
of the three planes (1, 2), (2,3) and (3, 1) in turn it is possible to construct a
Mohr diagram for each, as shown in Fig. 11.16. It is then apparent that a
composite Mohr diagram could be constructed, as shown in Fig. 11.17, by
superimposing these three diagrams. This Mohr diagram then enables the
true maximum shear stress in the material to be determined.
It is evident from the composite Mohr diagram that if 0; and a2 are
either both positive or both negative than 7,,,, in the (1, 2)-plane will not be
the maximum shear stress in the material. It should also be noted that in the
above superposition of the Mohr diagrams for the three planes, it is not
essential that 03 = 0, so that in fact Mohr diagrams may be used generally
for three-dimensional stress systems.

Fig. 11.16 3-1 plane

QY

Th
Fig. 11.17 Composite Mohr dia-
gram for three-dimensional sys- G max

tem

me) fee)
QyY
 STRESS AND STRAIN TRANSFORMATIONS 307

11.11 States of strain

In the following analysis all strains are considered to be small in magnitude.
Normal strains are defined as the ratio of change in length to original length
in a particular direction, and a subscript notation similar to that for stresses
will be adopted, €, and €, being the direct strains in the x- and y-directions
respectively, positive for tension and negative for compression. Shear strain
is defined as the change in angle between two planes initially at right angles,
and the symbol and subscripts 7,, will be used for shear referred to the x-
and y-planes.

Plane strain Plane strain is the term used to describe the strain system in which the
normal strain in, say, the z-direction, along with the shear strains +,, and
Yzy, are zero. It should be noted that plane stress is not the stress system
associated with plane strain. It will be evident from the definitions of plane
stress and plane strain in Chapter 3 that plane strain, ie. ¢, = 0, is
associated with a three-dimensional stress system, and likewise plane stress
is related to a three-dimensional strain system.
The stress system in Fig. 11.18(a) will give rise to a strain system
combining direct and shear strains as shown in an exaggerated manner at ().
The object now is to determine the direct strain, €,, and shear strain, ,, for
directions normal and tangential to a plane, inclined at @ to a co-ordinate
direction, in terms of the direct strain, €,, €,, and shear strain, Yr), Yyx,
referred to the co-ordinate planes.

Fig. 11.18

5 Total strain

JV
/ Shear strain
.
only
my
J];K
Direct strain
Unstrained only

(a) (b)

11.12 Normal strain in

Referring to Fig. 11.19 (A’C’—AC)/AC gives the strain normal to the plane
terms of co-ordinate FB related to the normal stress ¢,. Considering the triangles ACD and
strains A’/C’D’ and with dw as the increase in length from AD to A’D’, and dv the
increase in length from CD to C’D’, then

ND= Aboud =AD d

(te
AD | ap dhe.)
/ / dv
(CADY {CID
Sr ald) = (ClD) I+ Gp = GD(l

Similarly

A'C’ = AC(1 + €,)

308 MECHANICS OF ENGINEERING MATERIALS

Fig. 11.19

(a) Unstrained (b) Strained

Now,

+ °3)
(A'C’)?= (A'D’)’+ (C'D’)’ — 2A'D’. CD’ cos(90
or

(AC)’(1 + €,)° = (AD)*(1 + ex)? + (CD)’(1 + ey)”

+2AD(1 + ¢,)CD(1 + €,) sin Ye
Since strains are assumed small, then sin 7, ~ 7, and second-order powers
may be neglected. We have

(AC)?(1 + 2e,) = (AD)’(1 + 2e,) (CD) 2)

2A): Cia
which, with (AC)* = (AD)’ + (CD)’, reduces to
2e,(AC)’ = 2e,(AD)* + 2¢,(CD)? + 2AD. CD. yyy
Dividing through by 2(AC)’ and introducing sin6 and cos @

En = €y COS’ 6 + ey sin? 6 + xy Sin 6 cos 6 [11.27]

Therefore

En = Soa pe cos 20 + 4+, sin 20 [11.28]

AiskS Shear ewals In Referring to Fig. 11.19, the shear strain +, related to the shear stress 7,, Fig
terms of co-ordinate 115 isjs given
oj by the change in angle between EB and AE, or
strains =/AEB—/A’E’B’. Considering the triangles AEB and A’ E’B’, then, as before,

A'B! = AB(1 + ¢,)

A'E! = AE(1 + €,)
E/B’ — EB(1 =e En+90° )
STRESS AND STRAIN TRANSFORMATIONS 309

En+90° 1S the normal strain in a direction at 90° to ¢,. Now

(A'B’)? = (A‘E')’ + (E'B’)? — 2AE’. E'B’ cos(90° + 4,) [11.29]

or

(AB)*(1 + 2,)° = (AE) (1+6,)° + (EB)(1-+ e490)"

—2AE(1 + €,)EB(1 + €+99° )cos(90° + 7%)

But cos(90° + 7,) = —siny, & —7, and, neglecting the second order of
small quantities,

(AB)?(1 + 2e)) = (AE)?(1 + 2e») + (EB)°(1 + 2en490:)

+2AE. EB. 7,

But (AB)’ =(AE)?+(EB)’, and dividing by 2(AB)* gives

Ey = En sin’ 0 + En 490° cos’ 6 + y, sin 8 cos 0

Rewriting in terms of 20,

En+90° oF En En+90° — €
—}4,sin 20 = 5 5 “cos 20 —«, [11.30]

Now,

Ey ar Ey Ex — Ey ° 1 : °
Ent9" = —s 7 608 WO = 90 4-5 Vy sin 20 90)

es Ze 5ese Se ge +Yay Sin 20

and

En = s 5Ba | ak 5y cos 26 + 5 Vey sin 20

Therefore

En+90° air A Cy ar Ey

2 Gm
and

a Bh Sper we a © cos 20 — +Yay Sin 20

Substituting the above expressions in eqn. [11.30],

Sear Sy ep
—37; sin 20 = -” cos” 20 — +Ixy Sin 20 cos 20 — €,
Z.

— £# 7 © (1 — cos? 20) — 1 y,,

sin 20 cos 26
Z
a
310 MECHANICS OF ENGINEERING MATERIALS

Dividing through by —sin 26,

Jy,=— (2 20+ 9 00820

5)
sin [11.31]
11.14 Mohr’s strain
The graphical method of obtaining normal and shear stress on any plane by
circle
Mohr’s circle can also be employed to determine normal and shear strains at
a point. Considering eqns. [11.28] and [11.31] and rewriting, we have
Eguey 6,
— ey ;
&—— D a= D * cos 20 + 5 Yxy sin 20

and
Cy= By
1s = — 5 sin 20 + 3% cos 20

Squaring each and adding produces the expression

le
Ex se

2
y 2

Ca)
Ex

Z x
+r)
a Si , 2

11.32
which is the equation of a circle of radius 5\/|(€, cay Yolo and with
centre at [5(€, + €,),0] relating €, and 7.
The Mohs circle as shown in Fig. 11.20 is constructed in the same
manner as for stresses. The correct position for plotting shear strain on the
circle, i.e. above or below the ¢€-axis, may be found either by relating the
deformation of the element to the corresponding shear-stress system, or by
the convention that a positive shear strain in the element corresponds to the
sides of the deformed element having positive slope in relation to the co-
CRUE axes. On co-ordinate axes of — strain € na semi-shear strain
+7, each to the same scale, the points (Ex, + xy) and (Ey, 53 Yyx) are set up, and
a circle is drawn with the linejoe these two points as diameter. The
normal and semi-shear strain ¢, and!
Ys in. a direction at @ to the x-direction
are obtained from the intersection of a radius with the circle, at 20
(anticlockwise) from AB.

Fig. 11.20 U fa Mere > 7,

NOIR

fo}

Shear
stress

(11.15 Principal strain

The maximum and minimum values of the normal strain at a point are given
and maximum shear strain
by P and Q in Fig. 11.20, whence
 STRESS AND STRAIN TRANSFORMATIONS 311

ey = Ey5a & tives — 6) + %%] (maximum)

.
[11.33]
and

ae ies
a2 = ze 2 2 —} ey i a ae oa (minimum) [11.34]

These are termed the principal strains and may be compared for similarity
with the expressions for principal stresses. The former occur on mutually
perpendicular planes making angles 20 = 2¢ and 180° + 2¢, or 0 = ¢ and
90° + @ with the x-direction. From the diagram, when 20 = 2¢,

= ttan7! (2 ) and 90° + ttan7! (22 )

Ss Ey Ex Ey

[11.35]

Either by substituting for @ in eqn. [11.31] or by reference to the circle, it is

found that the shear strain is zero for the planes of principal strain.
Since T,, = G7, when the shear strain is zero, so also must be the shear
stress. But it was previously shown that the shear stress was zero on the
principal stress planes, and therefore the planes of principal stress and
principal strain must coincide.
The maximum shear strain occurs at the top and_ bottom points of the
circle when €, = }(€, + €y) and 54, =3/[(er — ene a anh, therefore

Ynax = V[(Ex = aye =P ag, (1 1.36]

Alternatively, this may be written in terms of the principal strains as

l el Ne? =
2 Vmax = y) or Ymax = €1 — €2

11.16 Experimental
In many practical situations, the geometry of a structure or part of it may be
stress analysis
complex and yet there is a need to know accurately the levels of stress in the
material. This may necessitate measurements rather than calculations but
unfortunately it is not possible to measure force or stress directly. All that
one can do is measure the displacements or deformations which arise as a
result of the force or stress. The strain can then be related to the stress from
a knowledge of the modulus of the material.
One of the most widely used methods of experimental stress analysis is
based on the strain gauge. The principle of the electrical resistance strain
gauge was discussed by Lord Kelvin when he observed that the electrical
resistance of a wire changes when it is stretched.
Modern strain gauges are of metal foil construction as illustrated in
Fig. 11.21. The gauge is bonded onto the surface where the strain is to be
measured. It is then connected to an electrical circuit to monitor its
resistance. When the material surface is deformed, the gauge also
experiences the same strain and this can be quantified from the
measurement of the change in its resistance and a knowledge of the
calibration constant (gauge factor) for the strain gauge.
312 MECHANICS OF ENGINEERING MATERIALS

It will be seen from Fig. 11.21 that foil gauges come in a variety of
patterns to measure strains in different directions. It should be noted that it
is only possible to measure direct strains, not shear strains. The following
example illustrates how the results from strain gauges may be analysed.

Fig. 1121 ‘Typical foil strain gauges

Pal
“S
(published by courtesy of Mea- N
surement Group, Vishay)

1A

ets eae | yn y

Example 11.6
P Two electrical resistance strain gauges are sited at 45° to the axis of a 75mm
diameter shaft. The shaft is rotating, and in addition to transmitting power, it is
subjected to an unknown bending moment and a direct thrust. The readings of the
gauges are recorded, and it is found that the maximum or minimum values for each
gauge occur at 180° intervals of shaft rotation and are —0.0006 and +0.0003 for
the two gauges at one instant and —0.0005 and +0.0004 for the same gauges
180° of rotation later. Determine the transmitted torque, the applied bending
moments and the end thrust. Assume all the forces and moments are steady, i.e. do
not vary during each rotation of the shaft.
E = 208GN/m2, 1» = 0.29, G = 80GN/m’. The shaft and strain gauges are
illustrated diagrammatically in Fig. 11.22.

Fig. 11.22 +0.0003

The simplest starting point is to find the torque. This may be found from
the shear stress which is in turn related to shear strain. From eqn. [11.31] we
have

xy = —(Ep — €q) Sin 20 + 7%, cos 20

= —(0.0003 + 0.0006) sin 90° + +, cos 90°

= —().0009
STRESS AND STRAIN TRANSFORMATIONS 313

Try = —0.0009 x 80 = —72 MN/m?

Hence the torque is

T= z x 0.0753 x 72 x 10° = 5.97kNm

In order to find the bending moment and end thrust it is necessary to
know the strains in the x-direction for the two specific rotational positions of
the shaft. From eqn. [11.28] we have

el el
Top — 0.0006 = =" — 0.00045
and

el el
+0.0003 = Ss + 0.000 45

Hence

ex te, = —0.0003

eB 4 6B
Bottom — 0.0005 = ara — ().000 45

or

cP + 63
+0.0004 = WE + 0.000 45

Hence

ge ee, = SIL
To eliminate ¢, from the above equations we use the relationship
Ey —= — Vex,

a 5 PO = bal
l= wp

c= il DOOT 2 —(0.000 141

be | ==

These strains are the sum of the bending strain which reverses in sign for
180° rotation and the steady compressive strain (due to end thrust).

Bending strain = +4 (e! — c?) = £0.000 141

Compressive strain = —4(e7 +?) = —0.000 282

Hence

Ors = £0.000 141 x 208 x 10? = +29.3 MN/m?

314 MECHANICS OF ENGINEERING MATERIALS

ee

from which

M= 3 x 0.0753 x 29.3 x 10° = 1.21kNm

Oy. = —0.000 282 x 208 x 10° = —58.6MN/m’

and so the end thrust F' is given by

pS Fx 0.0752 x (—58.6) x 10° = —259kN

11.17 Rosette strain For the complete determination of strain at a point on the surface of a
computation and circle component, it is necessary to measure the strain in three directions at the
construction point. This is achieved by cementing an electrical resistance rosette strain
gauge to the surface.

Fig. 11.23

45° eae

> Ey

Let the three measured strains be €/, €,, and €, and the angle between the
directions / and m, and m and 1, be 45° in each case. Then this arrangement
is known as a 45° rosette as shown in Fig. 11.23. If the angle between ¢, and
the principal strain €, is 0, then from egn. [11.27], the principai strains are
related to the measured strains as follows:

€) = €; cos? 6 + €) sin’ @
Em = €1 cos’(O + 45°) + 2 sin?(O + 45°) [11.37]
&, = € cos-(9 + 90°) +e, sin’(@ + 90°)
These equations may be rewritten as

€ = 4(e1 + €2) +3 (e1 — €2) cos 20

Em = 5(€1 + €2) —3(€1 — €2) sin 20 [11.38]
t= $(e1 + €7) —F(e1 — €2) cos 26

Solving the above equations simultaneously gives

2
€) =$ (er ten) + oe onge (ame
[11.39]
agi v2 2 2
€2 = 4 (er + En) + V/[(Er — Em) + (Em — En)’ |

2Em — El — En
tan 20 = [11.40]
 STRESS AND STRAIN TRANSFORMATIONS 315

To obtain principal stresses from principal strains we use the stress-strain

relationships,
01 VO
ey == - —
ae 2

Bes02 eee
VO|
11.41
2 EE oa
from which
E
Ceara ie nel Ve?)

E
0. = Tape + vé1) [11.42]

An alternative approach to the analysis of the strain rosette is to use the

construction described below based on Mohr’s circle.

Fig. 11.24

(a)

(b)

Consider the general case of three strain gauges, a, 6 and c, having

arbitrary orientation as shown in Fig. 11.24(a) and strain readings €,, €, and
€, in the given directions. As an example to illustrate the method assume
Eq > €, > €y. The procedure is as follows:
1. Set up a vertical axis to represent ¢ = 0 (which will subsequently be the
semi-shear-strain axis).
2. Draw three lines parallel to the above axis at the appropriate distances
representing the values (positive or negative) of €,, €, and €,.
3. On the middle line of these three (representing the middle value of the
three strains €,) mark a point P representing the origin of the rosette.
4. Draw the rosette configuration at the point P but lining up gauge c (in
this particular example) along its vertical ordinate.
5. Project the directions of gauges a and b to cut their respective vertical
ordinates at Q and R.
316 MECHANICS OF ENGINEERING MATERIALS

6. Construct perpendicular bisectors of PQ and PR; where these intersect

is the centre of the strain circle, O.
7. Draw the circle on this centre, which of course should pass through the
points P, Q and R. Insert the horizontal strain abscissa through O.
8. Join O to Q, R and S, where S is the other intersection of the circle
with the middle vertical line.
9. The lines OQ, OR and OS represent the three gauges on the circle
where 2a and 2( are the angles between OR and OQ, and OQ and OS,
respectively.
10. From the circle read off as required the principal strains €;, €2 or the
chosen co-ordinate direction strains €,, €), Yxy-

Example 11.7
At a point on the surface of a component, a 60° rosette strain gauge positioned as
shown in Fig. 11.25(a) measures strains of <=; — 0.00046, <,, = 0.0002 and <,
= —0.000 16. Use Mohr’s strain circle to determine the magnitude and direction of
the principal strains and hence the principal stresses. E — 208 GN/m?, v = 0.29.

NIR

>
—-

£4 = 0.000525
>|~<
(b) Ss2= —0,000 19

To construct the strain circle, Fig. 11.25(b), we use the procedure described
above. The principal strain values are represented by TV and TU, and
therefore

€, = 0.000525 and €2 = —0.000 19

The angle between ¢; and ¢€; on the circle is 34° and between ¢; and E2,
214°. Therefore

= 17h and 0, = 107°

 STRESS AND STRAIN TRANSFORMATIONS 317

The principal stresses are given by eqns. [11.42]; thus

208 aaet
Gh a x10° [0.000 525 + 0.29 x (—0.000 19)] = 107 MN/m?

208 x39e
aye Tp 10° (0.000 19 + 0.29 x 0.000 525) = —9 MN/m?

11.18 Spreadsheet
A general method for the analysis of a three-gauge rosette with the elements
solution for strain gauge at arbitrary orientations 6), 02 and 43 will now be demonstrated. Using egn.
rosette [11.27], the strain in a rosette element at an angle @ is

En = Ey COS’ O + ey sin? 6 + Yxy Sin cos 6

Given the strains €,, €, and €, in each element of the rosette lying at angles
0;, 82 and @3 respectively, three linear equations are obtained for the three
unknowns €,, €y and 7, as

cos?@; sin’6, sin 6; cos 6, Ex Eq

cos’6, sin’?@) sin @2 cos 4) Fe amt 5 ia [11.43]
cos’63_ sin’; sin 63 cos 63 Yxy E,

Here the three equations are written, as described in Appendix B, as a single

matrix equation,

[A]{x} = {7} [11.44]

The solution can be found using the inverse matrix [4]~' as
: = 3
Ex cos” sin’, sin 9; cos 0; Eq
Ey » = |cos?0 sin’6) sin 62 cos 0, Ep [11.45]
Yxy cos*6; sin’; sin 03 cos 3 iS,

The inverse matrix can be found using the ‘macro’ or menu commands
beginning in cell D16 of Fig. 11.26. In some spreadsheets such as Microsoft
Excel, the inverse matrix can be inserted as a special array formula and the
macro is not necessary. Once this matrix is evaluated for a given set of
angles, then as soon as the element strains are entered in the highlighted
cells H7..H9, the co-ordinate strains are calculated and appear in cells
H11..H13. The principal strains, principal angles and maximum shear strain
are then evaluated using the basic formulae of eqns. [11.33]-[11.36] and
displayed in cells H16. .H20.
Once the principal strains have been obtained the corresponding
principal stresses can be found using the stress-strain relations. Since a
strain gauge rosette will almost always lie on a free surface which has no
normal or shear stress acting on it, a state of plane stress exists. For a state of
plane stress, stresses can be calculated from the strains using

dieneearhty na
318 MECHANICS OF ENGINEERING MATERIALS

_ Rosette calculations

| =: a Sr! ee
__
Enter rosette element san and
a ingetseps_x, eps_y and gamma_xy.
| Principal Strains. E, gamma_ max and Principal Angle

an ae ; aes | 9a
Matrix toinvert

@RADIANS(B7), @COS(C7)2 @SINC7)*2 | @sinc7yr@cos(c7)

LOHSSIENSHED) |CIOOSC PS | SINC IRGSIM CE RELORICE):
G@RADIANS(B9)|_@COS(C9)N2
|
HG SIN (C8) 127 7 ee i
ee aN Coco a
| Inverted matrix |
+D11*H7+E11*H8+F11*H9
_ +D12*H7+E12*H8+F12*H9
gamma_xy +D13*H7+E13*H8+F13*H9 _

eps_1 ale 0.5*(H11+H12)+0.5*H20

ste: +- dere
{Invert‘Source A:07. FQ} | i __eps_2 |_+H16-H20 |
aoe =e |
aes | thetap
= —/@DEGREES@(ASIN(H13/H20))/4
i. Oe St |_+H18+90
| gammam | @SQRT(H11-H12)42+H1342)

, gamma. max and PrincipalAngle

0.000526
—0.000193
17.660
107.660
aia m | 0.000719

(b) Spreadsheet Display

and

Bey
eae [11.47]
Equation [11.47] is the same as eqn. [11.49], which will be derived below. A
general purpose spreadsheet for calculating stresses given strains, or strains
given stresses, under plane stress is shown in Fig. 11.27. The material
 STRESS AND STRAIN TRANSFORMATIONS 319

Stress strain relations (Plane Stres: )

Material Prof

|Enter strains, get stresses, - |

| Modulus matrix | Strains Stresses

+B4/(1-B542) | +D9*BS .Eta 0.000525 _ +D9*G9+E9* G10

|+E9 +D9 ta _ 70.00019 _ +D10*G9+E10*G10

| +Bay2*(1+B5) 0. +Fi1*G11

Enter stresses, get strains

Compliance matrix Stresses | Strains
+D15*G15+E15*G16

i, _+D16*G15+E16*G16
| 2*(1+B5)/B4 | +F17*G17

(a) Data and cell formulae

Stress strain relations (Plane Stress)

_ Material Properties
E 2.086414.

|Enter strains, get stresses} 2 - :

Modulus matrix | Strains | Stresses

| 2.276411 : 6.59E+10 | 8 .25E-04 4.07E-08
~ 6.59E+10 227 | ie : :1.90E-0. i -9E+06
| 8.06E+10 i iei 0.00 4 0.00E+00
Bes©m®NOORwWNHR
BRE
esefs
Enter stresses, get strains
1 Fears i
af
RE i | Compliance matrix | Strains

ao | 4,81€-12 1.39612 | | O7E+ : 5.27E-04

Veo>} -1.39E-12 4,81E-12 te | -1.92E-04
far~ 1.24E-41 | ‘ | 0.00E+00

(b) Spreadsheet Display

properties £ and vy are input in cells B4..B5, strains can be input in

G9. .G11 or stresses can be input in G15. .G17.

11,19 Relationships In Chapter 3 four constants of elasticity were defined relating various
between the elastic conditions of stress and strain. These are Young’s modulus, F, shear
constants modulus, G, bulk modulus, K, and Poisson’s ratio, v. It will now be shown
that these constants are not independent of one another.

Relationship between K, _ It was shown in eqn. [3.1] that volumetric strain is given by the sum of the
E and v _ three linear strains along the axes of the element. Hence

€ = Ey + Ey TEx
320 MECHANICS OF ENGINEERING MATERIALS

But from eqns [3.2]

Cp
Ex = or

Also for hydrostatic stress, 0, = 0, = 0, = 0; therefore

exy==z(1-2
(1-2)
or

Similarly

ey = (1-2) and é, = 5(1 - 20)

Therefore, summing the above three strains, we have
3
e= = (1-20)
or

P= (ty)
(a

But a/e = K, the bulk modulus; therefore

E = 3K(1 — 2v) [11.48]

Relationship between FE, ‘The square element of unit thickness shown in Fig. 11.28(a) is acted on by
Gandv_ pure shearing stresses. This system is equivalent to the system of direct
stresses on the element of Fig. 11.28(4), and from equilibrium
Gn, = On =T;. The strain along the diagonal AB in terms of the stresses
is given by

i.e. the extension due to a, plus the lateral expansion in the direction AB
due to the compression o,,. But
OV, =O, = 1

Fig. 11.28

Ne MaA
o, En,

(a) Pure shear stress (b) Equivalent direct stress (c) Strain systems
 STRESS AND STRAIN TRANSFORMATIONS 321

Therefore

En, we

Since for pure shear €, and €, are zero, then from eqn. [11.28],

En =4%8in 20
In this case 0 = 45°; therefore

ayaa
Equating the above expressions,
7.

Ie iret Oe,

or
Die
ma
E=>—(1l+v )
But 7,/7, = G, the shear modulus; therefore

E=2G(1+yv) [11.49]

Relationship between K, It follows from eqns. [11.48] and [11.49] that

G and v
3K(1 —2v) = 2G(1+v)
or

2c)
ce 3(\ = 2p)
[11.50]

Thus if any two of the four constants are known, or can be measured, then
the other two can be determined.

11.20 Stress and strain

In recent years there has been a rapid growth in the use of fibre-reinforced
transformations in composites. The major advantage of such materials is that high strength and
composites stiffness can be achieved at low weight. Products that have benefitted from
the use of composites include aircraft, ships, automobiles, chemical vessels
and sporting goods. In these industries the base material is usually metal or
plastic and the fibres used include glass, carbon, aramid (‘Kevlar’), boron
and asbestos. In some cases short (‘chopped’) fibres are used and this
provides a significant property enhancement over the base resin. However,
by far the greatest improvement in properties is observed if the fibres are
continuous. For example, if unidirectional carbon fibres are added to an
epoxy resin, the modulus of the resulting composite is improved by a factor
of about 60 (see Table 11.1) and the strength by a factor of about 30
compared with the unreinforced base resin. However, the composite is
markedly anisotropic in that in the direction perpendicular to the fibre axis
the modulus is only improved by a factor of about 2 and the strength is
322 MECHANICS OF ENGINEERING MATERIALS

Table 111 ‘Typical properties of

some materials Material Bex Et G KY
Vy Density
(GN/m2) (GN/m?) (GN/m*) (kg/m)
Carbon fibre/epoxy 180 10 Te. 0.28 1600
Kevlar/epoxy 76 DE 2S 0.34 1460
Glass/epoxy 39 8.4 4.2 0.26 1800
Spruce 8.9 DLs 400
Aluminium alloy 70 70 Li 2770

*Parallel to fibres or grain.

{Perpendicular to fibres.

likely to be reduced. Therefore, in the aircraft industry, for example, in

order to get property enhancement in all the required directions within the
component, it is normal practice to build up a laminate structure where each
layer has fibres arranged in the desired direction, as shown in Fig. 11.29.
As it is becoming increasingly likely that engineers and designers will at
some stage have to become involved in the design of components made from
fibre composites it is important that they should have an appreciation of the
Fig. 1129 Arrangement of fibre stages in the design process. In the following sections a brief introduction is
orientation in laminate given to the laminate theory involved in fibre-reinforced composites.

11.21 Analysis of a
In this analysis it is necessary to consider both the local co-ordinates (x, 7)
for the lamina and the global co-ordinates (X, Y) for the applied stress
system.

On-axis properties Consider first of all a single lamina in which the fibres are all aligned in the
global X-direction as shown in Fig. 11.30. The lamina is thin in relation to
its transverse dimensions and therefore it will be in a state of plane stress
when forces are applied to it.

Fig. 11.30

Cx | ee

Recognizing that the lamina is anisotropic with modulus values of £, and

Ey in its x- and y-directions and corresponding Poisson’s ratios vy and v,,
then the strains in the lamina may be written as
3 el GC
ae edEt TUN?
ee, VyOX
eased [11.51]

WOOP
OGY per
STRESS AND STRAIN TRANSFORMATIONS 323

Note that it can be shown that v,/E, = v,/E).

It is convenient for subsequent analysis to write these three equations in
matrix form (a brief introduction to the use of matrices is given in
Appendix B):

Ex Vk es 0 ox
Soy = —V,/ Ex Ly 0 Oy [11.52]
YX Y 0 0 1/ Gy 1DOY

or in abbreviated form

{e} = [S]{o} [11.53]

where [S] is referred to as the compliance matrix for the lamina.

At this point it is worth noting that in order to describe completely an
anisotropic material subject to a triaxial stress system, the compliance matrix
will have 36 terms.

EX Diteeesi7) 21g “S14, 15 O16 Ox

Ey D7)? 9523) 2 5506 oy
éz \ _ |S31 S32 S33, S34 S35 S36 Oz
Vvz Sa S42 S43 S4g S45 S46 TYZ
YZx Dei 52 953 54 2055 G6 TZX
Yxy Sel, 62k 6s Ge 8565. 966 TXY
[11.54]
In practice for isotropic materials the number of constants may be
reduced if we assume the following:

1. Shear stresses do not affect normal strains and normal stresses do not
affect shear strains. Hence

S14 = Si5 = Sig = S24 = S25 = S26 = S34 = S35 = S36 = 0

2. Shear strains are only affected by shear stresses in the same plane.
Hence

S45 = S4o = S54 = S56 = Sos = So4 = 0

3. The effect of gy on €x is the same as the effect of ay on ey, etc. Hence

S11 = Sx = $33
4. The effect of ay on €x is the same as the effect of a7 on €x, etc. Hence

Sy2 = 973 = S21 = $23 = S31 = S32

5. The effect of Txy on yxy is the same as the effect of Tyz on yyz, etc.
Hence

S44 = S55 = So6

Hence for isotropic materials the matrix in eqn. [11.54] reduces to one in
which there are only three constants $};, S12 and S¢, the values of which
324 MECHANICS OF ENGINEERING MATERIALS

are

1
E E G
It may be seen that eqn. [11.54] then reduces to eqns. [3.2] and [3.4].
When introducing the stress analysis of isotropic materials it is generally
more convenient to use these simple equations, but it should be remembered
that they are derived from a much more general situation.
However, in the analysis of laminates, which by their nature are
anisotropic, the basis equations are a little more complex and it is generally
found that their manipulation is simplified by the use of matrix algebra. For
a lamina it may be shown (see Jones', for example) that eqn. [11.54] reduces
to the form

Sit, Sid Si6

[S] = |S21 S22 S26
él 962 66

where, from eqn. [11.52]

1 1 1
Sia)
11 E, Sa22 Ey eS
66 Ce

89 12 = Si21 =
V VY.
E, E

S16 = 6 sl oe 0

If eqn. [11.51] is rearranged to give stresses in terms of strains, then

ox = (Ex + Hey)E,/(1 — ny)

oy = (ey + Hex) E,/(1 — ux) [11.55]

Txy = Gyxy
In matrix form this may be written as

ox Qn Qi Qe Ex
Oy = Qo) Qn Qo ey [11.56]
TXY Q61 Q62 Q66 | \ yxy

where [Q] is called the stiffness matrix. It is symmetrical and the individual
terms are

Qy = £,/(- Vy Vx)

Qo = Ey/(1 — yx)
O66 as Gy
 STRESS AND STRAIN TRANSFORMATIONS 325

Oi 0 Vy By/(1 _ UyVy) me VyE,/ (I a Vy Vx)

Qy6 = O61 = Qs = Qe2 = 0

Note that from eqns. [11.53] and [11.56] we may write

{o} = [Ol{e} = [S]'{e}

Therefore it is apparent that the stiffness matrix is the inverse of the
compliance matrix, so that

(Osi. [11.57]

Off-axis properties Consider now a situation where the x—y co-ordinates of the lamina do not
coincide with the global X—Y co-ordinates. This is shown in Fig. 11.31.

' : Oy if
Fig. 1131 Fibres aligned at an
angle to global X-direction a Ty x

Z
nd,

When stresses are applied to the lamina in the global (X—Y) co-ordinates
these must be transformed to the local (x—y) axes for the lamina since it is in
these directions that its properties are known. The transformation necessary
has already been performed in Section 11.5. Therefore using eqns. [11.13]
and [11.14] we may write

O, = ox cos’
0 + cy sin?
6 + 2Txy sin Ocos 0 [11.58]

Cpe OX. sin’ 6 + oy cos’ 6 — 2Txy sin

8cos @ [11.59]

Tx = —ox sin@ cos @ + ay sin @ cos @ + txy(cos’6 = sin’@)

[11.60]

Note that oy is obtained by letting 6 = (@ + 90°) in eqn. [11.13].

Again using matrix notation and letting s = sin 0, c= cos 6, eqns.
[11.58], [11.59] and [11.60] may be written as

Gi, G s 2s¢ Ox
Oy — re Ce CG Oy [11.61]

Ty) —s¢ sc (¢ —3*) TXY

326 MECHANICS OF ENGINEERING MATERIALS

or, in shorthand form,

1O) = ex [11.62]
where [7] is the transformation matrix which may also be inverted to give
global stress components in terms of local stress components:

{o}xy = Dlviale [11.63]

The inversion of [7] gives cag as

Pe ae —2s¢
[ale Saya wi 2s¢ [11.64]
so. ie (FP — 2H)

From the analysis in Section 11.14 it may be seen that the transformation
of strain is similar to that of stress, so we may write

Ex EX
ey S=(T]< ey [11.65]
+ Vey Sxy

At this stage we have developed methods of transforming the values of

stresses and strains independently from the loading axes to the fibre axis or
vice versa. In practice it is much more important to be able to establish the
contribution to overall stiffness which is made by a lamina in which the
fibres are at an angle to the loading axis. This is because in the construction
of laminates a number of lamina will be arranged at different orientations
and bonded together. The stiffness of the laminate will be the sum of the
contribution made by each of the individual lamina.
In order to determine the stiffness in the global directions for a lamina in
which the fibres are at an angle @ to the global X-direction then the
following three steps are involved:

1. Determine the strains in the local (x, )-directions by transforming the

applied strains through 6° from the global (X, Y)-directions.
2. Calculate stresses in local directions using the on-axis stiffness matrix
[QI.
3. Transform stresses back to the global directions through an angle
of —@°.

Using matrix notation these steps are shown below.

Step 1

Ex ie s SC EX
Ey ras
p= s2 ¢2 —S¢ Ey
Yxy —2sc 2c (c - o) Yxy

Note the modification to the transformation matrix [7] so that we may

write 7, instead of + Yay:
STRESS AND STRAIN TRANSFORMATIONS 327

Step 2

Ox Q, 1 Qy2 0 Ex

Oy = Q») Qo 0 Ey
Txy 0 0 066 Vey

Step 3

Ox @ ete —2s¢ Os
og Se Nie ee 2sc Ga
Txy sc —se (c? —s?) Ty

Hence to express global stresses in terms of global strains we perform the

following matrix multiplication:

ox Ge —2s¢ QO; OQ 0
(ep ep a I Meade 2s¢ Qn, Qy» 0
TXY se —sc (c? —5?) 0 0 O66

C s SC EX
di eal hie: —s¢ ey
—2sc se (c* —s*) Yxy

which provides an overall stiffness matrix [OQ] where

Ox Qn Qy Qh EX
Oy = Qo Qo Qo ays [11.66]

Txy O61 O62 Ose YXY

Ol Eco Oo ein 6-8 (2 p41) cca? sin’

eel
On = Qn = 1 [v,E,(cos*0 + sin‘ 6)

+(E, + E, — 4XG) cos’ @ sin? 6]

7 A l
O61 = Qi = 7 (cos? @ sin 0(E, — Ui Ey LAG)

—cos 6 sin’ 0(E, — v,Ey — 2XG)|

? 1
On = ; [E, cos* 0 + E, sin* 6 + sin? 6 cos? 6(2v,£, + 4AG)|

3 = l
057 = 034 = . [cos @sin*® 0(E, — v,E, — 2XG)

—cos’ Osin 6(E, — v,E, — 2G)]

328 MECHANICS OF ENGINEERING MATERIALS

> ]
Og Z [sin?6 cos” 0(Ey + E, — 2v,E, — 2G)

+AG(cos*@ + sin*6)|
in which \ = (1 — 1%).
Note that for high-performance composites, £, is typically much larger
than Ey or G. As 1, and vy are also relatively small the stiffness values may
be approximated by

0}, ~ E, cos*6 On ~ E, sin*6

On, ~ E, sin? 6 cos’ 6 Ox, ~ E, cos @ sin? 6

01 ~ Ey cos’ #sin 6 Oe ~ E, sin? 6 cos? 6

By a similar analysis it may be shown that, for applied stresses (rather

than applied strains) in the global directions, the overall compliance matrix
[.S] has the form

EX Si Si Sie Ox
ey 7 = |S2 S27 5% oy [11.67]
XY Sle (S62 966 TXY

where

Si, = S11, cost 6 + Sx sin*6 + (2S12 + Soe) cos’ @ sin? 9

Sor = Siz = (Sir + S22 — Soe) cos? @ sin?6 + Si2 (cos*@ + sin* 6)

S61 — S16 = (2.822 —_ 2812 — S66) cos? @sin 0

—(2S22 — 2S}. — S66) sin? 6 cos6

Sx = Si; sin’6+ Sx cos* 6 + (28}2 + Sg) cos” @ sin? 8

S62 = S06 = (2.811 = 2S 42 = S66) cos 6 sin? 7]

— (282) = 2812 = S66) sin 6 cos® 7

S66 = 4(Sy1 = 2812 =F S66) cos” 6 sin? 7 at S66(cos?6 = sin’0)”

These calculations, though laborious to perform by hand, are very

straightforward to perform by computer using matrix methods. From the
moduli and Poisson’s ratios, eqn. [11.52] provides the lamina compliance
[|S]. The lamina stiffness is the inverse of the compliance matrix, i.e.

eee be [11.68]
 STRESS AND STRAIN TRANSFORMATIONS 329

and the off-axis lamina stiffness [Q]in the global (X—Y)-co-ordinate system
is

lo io 7:
|e [11.69]
where [77] and [7-] are the transformation matrices of eqns. [11.61] and
[11.65]. The off-axis compliance is the inverse of the stiffness, i.e.

[S] = (a) [11.70]

Given facilities for matrix inversion and multiplication, the detailed
formulae for each of the terms in eqn. [11.67] are not needed. If the
modulus values are required for a lamina with the fibres running at some
angle @ to the global X-axis, they can be determined from inspection of the
compliance matrix as

11.22 Analysis of a
At this stage we are in a position to describe the stress—strain behaviour in
laminate
any co-ordinate direction for a lamina consisting of unidirectional fibres.
Such a lamina is used in beams and tension/compression members where
the excellent longitudinal properties can be used to good advantage.
However, in many cases the low transverse properties could not be tolerated.
For such applications it is usual to build up a laminate in which laminae are
arranged at different orientations in order to achieve the desired overall
properties in the laminate. The orientations of the individual lamina can be
in any desired combination. Generally there are two broad categories of
laminates — those which are symmetric about the mid-plane and those which
are unsymmetric, as shown in Fig. 11.32.

Fig. 1132 Different types of lami-

nate construction

(a) Symmetric angle ply (b) Non—symmetric angle ply

We will consider only those laminates which have mid-plane symmetry

and the reader should refer to textbooks on laminate theory!~* for the
unsymmetric cases. Laminates which are symmetric can be analysed by a
simple extension of the theory developed in the previous section for
unidirectional composites. In both cases their mechanical behaviour is
described by three sets of elastic constants. Thus a symmetric laminate
behaves like a homogeneous anisotropic plate and under uniaxial loading its
effective modulus is simply the arithmetic average of the constituent
laminae.
330 MECHANICS OF ENGINEERING MATERIALS

11.23 In-plane behaviour

; ; During the manufacture of a laminate the individual plies or laminae are
of a symmetric laminate
bonded securely together so that when loaded they all experience the same
strain. However, because the stiffnesses of the plies are all different, the
stresses will not be the same in each case. This is illustrated in Fig. 11.33.

Fig. 1133 Stresses and strains in

uniaxially loaded symmetrical Laminate
TOLD
laminate ULISASIILODIALLELOA
€,,€) OF Yy,

Stresses Strains

When defining the overall stress-strain behaviour of a laminate it is

necessary to use average stresses. These are defined as
l h/2
ox =;| ax dZ [11.71]
hJ_n/2

1 h/2
qe | oy dZ (11.72]
h —h/2

] h/2
TXY =;| Txy dZ [11.73]
=1/2

oy SdZ [11.74]

AG On Qy Ore EX
Qn Qe Ey pdZ
2 |Sym Q66 YXY
Note that the stiffness matrix [Q] is symmetric about its diagonal.
As the strains are independent of Z they can be taken outside the integral:

Ox ae on Qy (Or EX
Geo u(omty Qo. Qre |dZ4 Ey
Txy Hi? | sym. Q66 YXY

OX EX

oy 7= lA) ev [11.75]
 STRESS AND STRAIN TRANSFORMATIONS 331

where, for example,

1 Af2 ®) Afr _
An =; Qaz =; | Q), dZ
—h/2 h Jo
Within a single lamina, such as the ith, the Q-terms are constant so,
referring to Fig. 11.34, the integral may be replaced by a summation.

Fig. 11.34

ith layer

2—-, [7 27,
A= |, a2 =p ain [11.76]

Aj =a 0, (=) [11.76]

Thus the stiffness matrix for a symmetric laminate may be obtained by

adding, in proportion to the lamina thickness, the corresponding terms in
the stiffness matrix for each of the laminae. Note also that the volume
fraction of material in the 7th lamina is given by

iejh
on

and this is a convenient substitution to use when performing the

summations indicated by eqn. [11.76]. The sum of all the v; terms for the
different laminae will of course be 1.
Having obtained all the terms for the stiffness matrix [A] as indicated
above, this may then be inverted to give the compliance matrix [a] and
hence the modulus values for the laminate.
The approximate analysis for the off-axis lamina stiffness terms such as
Oy, ~ E, cos* @ may be extended to a lamina giving

Ay ed Ihe. Sy VU; cos* 0;

where v; is the volume fraction of laminae at angle 6; in the laminate.

For cross-ply laminates and balanced angle-ply laminates equation
[11.75] will be simplified by the fact that 416 = Ag, = 0; Axo = Agr = 0.
The following laminate properties may then be obtained from inspection
of the compliance matrix (see eqn. [11.52]):

Acer,
a\\ ar
eee.)466
332 MECHANICS OF ENGINEERING MATERIALS

Since diagonal terms in the laminate stiffness matrix, such as 4j1, are usually
large compared with off-diagonal terms, such as 42, the laminate modulus
Ex is approximated by the relation

Ey ~ y v;,Eq cos'0;

where v; is the volume fraction of laminae at angle @,, with longitudinal

modulus £,,.

Example 11.8
A filament-wound composite cylindrical pressure vessel is made up from ten plies of
continuous carbon fibres in an epoxy resin. The arrangement of the plies is as
shown in Fig. 11.35. There are two plies at 69°, two plies at —60° and the
remainder are in the hoop direction. Calculate the maximum permissible pressure in
the cylinder if the hoop strain is not to exceed 1%. At this pressure calculate
the axial strain in the cylinder. The properties of the individual plies are E,
= 180 GN/m?’, E =10 GN/m?, G=7 GN/m?’, v, = 0.28. For the cylinder
¢ 7 300:
Lies

Fig. 1135 Composite cylindrical

pressure vessel

Two ply

The first step in the solution is to get the stiffness matrix terms for each ply
in the global co-ordinate directions. Thus from eqn. [11.66]:
Z 1 |
(Onno = 5 [E, cos* 60° + E, sin* 60°

+(2v,E, + 4G) cos” 60° sin? 60°] = 23.24 GN/m?

Similarly,

(Q11)_¢9: = 23.24GN/m? = and_-— (Qj)9.= 180.78 GN/m?

Then, from eqn. [11.76],

Aj, = 0.2(Qh1)gqs + 0-2(Qi1)_692+ 9.6(Q11)9.

= 117.8 GN/m?
STRESS AND STRAIN TRANSFORMATIONS 333

In a similar way the other terms in the stiffness matrix for the laminate
may be calculated to give

117.8 14.6 0
[4]= | 146 49.5 0
O= 0 188
This may then be inverted to give

ofS ube2640) 10
l= 2602007. 0) alex. 10-2
0 0 53.2
From which

ges = 113.5GN/m
ee
ee aa ee 2 fie

E L 47.7GN/m
pe D IA
= — — 5 m

G : : 18.8 GN/m?
emer La
—— = => A m

—a)\2 2.6
wi)
a a)\ 8.81 U 2

— ai? 2.6
= =—_ = 0,124
ree 07
Then, expressing the axial and hoop strains in terms of ay and ox,

Ex S8ie0 2-6). 0 ox
Ey) = | —2.6..20.97... 0 |10-°<" ay
MSZ 0 0 53.2 TXY

eS A 350
i? Ge 007 aul NOx 175s p
0 Onlesoe 0

where p is the internal pressure.

Hoop strain ¢x = (8.81 x 350 — 2.6 x 175)10-3p = 0.01

p = 3.8MN/m’
Also, at this pressure the axial strain

ey = (—2.6 x 350 + 20.97 x 175)10-3 x 3.8 x 107% = 1.05%

334 MECHANICS OF ENGINEERING MATERIALS

Note that the approximate formulae for the laminate modulus give

Ex & i v;Ey; cos* 6; = 180(0.2 cos* 60 + 0.2 cos*(—60)

+0.6 cos* 0] = 112.5GN/m’?

Ey © )_ v,Excos*(6; + 90) = 180(0.2 cos* 150 + 0.2 cos’ 30

+0.6 cost 90) = 40.5 GN/m?
which are quite close to the exact values calculated above.

11.24 Flexural behaviour

When a laminate is subjected to flexure there will be a uniform strain
of a symmetric laminate gradient across the section but, as shown in Fig. 11.36, the stress variation
will be non-linear owing to the different stiffnesses of the individual laminae.

Fig. 1136 Stresses and strains in

laminate subjected to flexure Laminate
———————————
CUM
NLL LL |
AAD SS SII
ee OE
(\\\IAAAAAAAAAAAM

Stresses Strains

Using an analysis similar to that in the previous section it may be shown

that the terms in the stiffness matrix for the laminate are given by

[11.78]
ae di ae }
where J; and Jcgyp. are the second moments of area for the ith lamina and
the complete laminate respectively. Inversion of the matrix [D] will then
enable moduli values to be obtained as before.

11.25 Summary
The design of engineering components depends on the assessment of the
critical stress levels occurring. In general these are the principal stresses, the
importance of which will be demonstrated in the next chapter. It is therefore
most important through either the analytical derivations or the Mohr’s circle
construction to be able to investigate the state of stress at a point in the
material. It will be appreciated that for a component that already exists we
cannot directly measure stress due to applied forces.
However, we can measure 7 situ displacements and strains, from which
the associated stress system can be derived, using the stress—strain
relationships. Therefore it is equally important to understand two-
dimensional strain analysis for which the Mohr strain circle is most
useful. Finally, although we seldom have to derive one elastic constant from
two of the remainder, it is fundamental to the elastic behaviour of materials
that the four constants are interrelated.
 STRESS AND STRAIN TRANSFORMATIONS 335

This chapter has also introduced the theory of fibre composite materials
which by their nature are markedly anisotropic. It will be seen that the
stresses and strains in a laminate, although apparently complex in nature,
may be determined using the stress and strain transformations developed at
the beginning of the chapter.

plies 1. Jones, R. M. (1975) Mechanics of Composite Materials, Scripta Book Co.,

Washington, D.C.
2. Tsai, S. W. and Hahn, H. T. (1980) Introduction to Composite Materials,
Technomic, Westport, CT.
3. Agarwal, B. D. and Broutman, L. J. (1980) Analysis and Performance of
Fiber Composites, John Wiley, New York.

Problems 11.1 For the engineering components illustrated in Fig. 11.37 it is

necessary to determine the normal and shear stresses at the points
. shown. Sketch an element in each case showing the magnitude and
sense of the stresses on each face.

Fig. 11.37

20 kN
im
50 mm

50 mm
al
_ |L [20mm

2m
st _

11.2 For the elements illustrated in Fig. 11.38 calculate the stress
components on the inclined planes shown.

Fig. 11.38 150

50

60°

200 100 100

336 MECHANICS OF ENGINEERING MATERIALS

11.3 Ata point in a boiler rivet the material of the rivet is undergoing the
action of a shear stress of 50 MN/m? whilst resisting movement
between the boiler plates and a tensile stress of 40 MN/m? due to
the extension of the rivet. Find the magnitude of the tensile stresses
at the same point acting on two planes making an angle of 80° to
the axis of the rivet.
14 Construct Mohr’s circles for the stress systems given in Problem
12.2 and check the solutions for the stresses on the inclined planes.
13 A cube is subjected to a hydrostatic pressure, ie the same pressure in
all directions. Show by both calculation and Mohr’s circle that the
resulting maximum shear stress is zero.
11.6 At a point in the cross-section of a girder there is a tensile stress of
50 MN/m? and a positive shearing stress of 25 MN/m*. Find the
principal planes and stresses, and sketch a diagram showing how
they act.
rie Draw the Mohr stress circles for the states of stress at a point given
in Figs. 11.39 (a) and (4). For (a) determine and show the magnitude
and orientation of the principal stresses. For (4) show the stress
components on the inclined plane.

Fig. 11.39 50 MN/m?2 20 MN/m2

40

30

50
(a) (b)

11.8 Determine the principal stresses, maximum shear stresses and their
orientations for locations A, B and C of Example 11.1.
ate9. The loads applied to a piece of material cause a shear stress of
40 MN/m* together with a normal tensile stress on a certain plane.
Find the value of this tensile stress if it makes an angle of 30° with
the major principal stress. What are the values of the principal
stresses?
Wied 0 The I-section beam shown in Fig. 11.40 is simply-supported over a
length of 6m and subjected to a point load of 12kN at mid-span.
Calculate the values of the principal stresses at a point on the cross-
section 54.5 mm above the neutral axis.

Fig. 11.40

230

150
~ ~
Dimensions in mm
 STRESS AND STRAIN TRANSFORMATIONS 337

Piel A pulley of 250mm diameter is keyed to the unsupported end of a

50 mm diameter shaft which overhangs 200 mm from a bearing. The
pulley belt tension on the tight side is three times that on the slack
side. Determine the largest values for these tensions if the maximum
principal stress in the shaft is not to exceed 150 MN/m’.
5B a For a curved bar loaded as in Fig. 11.41 determine at what position
the maximum principal stress has its greatest value. Calculate the
latter and also the maximum shear stress for a split ring of radius
250mm, bar diameter 50mm and loaded with 5kN at one end
perpendicular to the plane of the ring.

Fig. 11.41

els A 45° rosette is fixed to a short rectangular section pillar as

illustrated in Fig. 11.42. If the gauges read €, = 72 x 107°,
e, = 100 x 10-° and e€, = —204 x 10~° determine the values of
the forces F and W. The cross-sectional area of the pillar is 600 mm?
and it is made from steel for which E = 207 GN/m’ and v = 0.3.
11.14 A state of two-dimensional strain is €, = 0.0007, ¢, = —0.0006;
Yxy = Yyx = 9.0003. Calculate the principal strains in magnitude and
direction and check the results using Mohr’s circle construction.
els At a particular point on the surface of a component the principal
strain directions are known, but it is not convenient to attach
electrical resistance strain gauges in these directions. However, it is
possible to cement gauges at 30° and 60° anticlockwise from the
Fig. 11.42 major principal strain direction, and the readings from these gauges
are +0.0009 and —0.0006 respectively. Construct the Mohr strain
circle and find the value of the principal stresses and maximum shear
stress. E = 208 GN/m’, ¢ = 0.29.
11.16 The principal strains, €; and €2, are measured at a point on the
surface of a shaft which is subjected to bending and torsion. The
values are €; = 0.0011 and €2 = —0.0006, and ¢€, is inclined at 20°
to the axis of the shaft. If the diameter of the shaft is 51 mm and the
rigidity modulus for the material is 83GN/m?, determine
analytically the applied torque and the maximum shear stress in
the material at the point concerned and check graphically.
List? There is a method of experimental stress analysis called the brittle-
coating method where the object to be analysed is coated with a thin
coating of brittle material. When the component is loaded, the
coating will crack perpendicular to the maximum tensile strain when
a critical threshold strain is reached. You can assume both
component and coating are subject to the same strain and the
component behaves in a linear elastic manner.
An aluminium component (E = 70 GN/m’, v = 0.3) is coated.
At a particular point cracks in the coating first appear at half the
338 MECHANICS OF ENGINEERING MATERIALS

design load. When the load is further increased to 0.8 times the
design load a second set of cracks appears perpendicular to the first.
The critical strain for the coating used is 500 x 10~°.
(i) |What are the principal strains in the plane of the component
surface at this point under the design load?
(ii) What are the principal stresses under the design Joad? You can
assume the component surface is in a state of plane stress.
(iii) If the first set of cracks was perpendicular to the x axis, what
would be the normal stress on a plane whose normal is at an
angle of 30° to the x axis under the design loading?
11.18 A 60mm diameter solid shaft has a strain gauge mounted at 65° to
the axis of the shaft. In service a torque is applied to the shaft and
the strain gauge reads 200 x 10~°. Calculate the value of the torque
if the shaft is made from steel with E = 207 GN/m? and v = 0.3.
11-19 Three strain gauges A, B and C are fixed to a point on the surface of
a test plate at 120° intervals, and the strains recorded are
€4 = +0.00108, ¢, = +0.00064, e¢ = +0.00090. Draw Mohr’s
strain circle for this problem and determine the principal strains and
the inclination of gauge A to the direction of the greater principal
strain.
11.20 At a certain point in a steel structural element the directions of the
principal stresses 0; and a2 are known. Measurements by strain
gauges show that there is a tensile strain of 0.00083 in the direction
of a; and a compressive strain of 0.00052 in the direction of a2. Find
the magnitudes of a; and 02, stating whether tensile or compressive,
and the maximum shear stress. v = 0.28, E = 207 GN/m”.
| A thin-walled aluminium alloy pressure vessel of 200mm diameter
and 3mm wall thickness is subjected to an internal pressure of
6 MN/m/?. Strain gauges which are bonded to the outer surface in
the hoop and axial directions give readings of 0.00243 and 0.00057 at
full pressure respectively. Determine the four elastic constants for
ais %
\SERRRRK5 x?Ac |Va the material.
/X x
xXXX
2r
122 A chemical pressure vessel is to be manufactured from glass fibres in
“Filaments an epoxy matrix as illustrated in Fig. 11.43. If the optimum fibre
orientation is that in which the fibres are subjected to tensile stresses
Fig. 11.43 with no transverse or shear stresses, determine the optimum value
of a.
123 A unidirectional fibre reinforced composite has a strength of
1500 MN/m? when loaded in the fibre direction, but only
100 MN/m? when loaded perpendicular to the fibres. When
sheared parallel to the fibres it fails at a stress of 120 MN/m’.
(a) A specimen of the composite with fibres running at 30° to the
loading axis is subjected to a uniaxial stress of oo, Fig. 11.44.
Determine the normal stress parallel to the fibres, the normal
stress perpendicular to the fibres and the shear stress parallel to
the fibres.
(b) At what value of a9 will the composite fail, and will it fail
parallel or perpendicular to the fibres, or by shear? You can
assume the material will fail when any one of the stresses
reaches a critical value.
 STRESS AND STRAIN TRANSFORMATIONS 339

Fig. 11.44
~— Bs

124 A cylindrical pressure vessel is made up of continuous carbon fibres

in an epoxy matrix. The fibres are wound at +45° from the cylinder
axis. Calculate the axial and hoop strains in the cylinder when the
internal pressure is 5MN/m’. The cylinder diameter is 1 m and the
wall thickness is 12mm. Unidirectional carbon fibres in an epoxy
matrix have the following properties: E, = 130GN/m/?, y=
TGN/
mi? 2Ge, =5.6GN/n4,y,, = 0.3.
bde25 Part of the hull of a speedboat is in the form of a flat sheet of fibre-
reinforced polyester. The fibres are continuous glass strands and the
lay-up is such that there are four plies at 45°, four plies at —45° and
two plies at 0°. Calculate the in-plane stiffness of the sheet in the 0°
and 90° directions. A unidirectional glass-fibre composite has the
following properties: 1, = 0.3, E, = 40 GN/m/’, EB, = 9.8 GN/m/’,
Gyy = 2.8GN/m’.
CHAPTER
Yield Criteria and Stress
Concentration

==
All the theoretical analysis of the previous chapters has made use of a linear stress—
strain relationship. This is because Hooke’s law established that metals have a linear-
elastic stress-strain range. however, if a ductile metal is subjected to simple axial
loading, it is found that beyond a certain point, stress is no longer proportional to
strain, which results in there being a permanent deformation when the stress is
removed, as illustrated in Fig. 12.1. The material is then said to have yielded.
Knowing the stress at which yielding behaviour commenced, it would then be a simple
matter to design a component from the same material to withstand a particular axial
load without any yielding occurring. This example is simple as there is only one
principal stress to consider.
The problem of designing a pressure vessel, rotating disc, or some component
containing a complex principal stress system so that the material remains elastic, i.e.
no yielding, when under fuil load is rather more complex. One could adopt a trial and
error method of building a component and testing it to find when the deformations
were no longer recoverable, but this would obviously be very uneconomical. It is
therefore essential to find some criterion based on stresses, or strains, or perhaps
strain energy in the complex system which can be related to the simple axial
conditions mentioned above. If a theoretical criterion can be established which
predicts complex material behaviour, it is then only necessary to establish
experimentally the yield point in a simple tension or compression test.
A number of theoretical criteria for yielding have been proposed over the past
century but only those now currently accepted and used for ductile and brittle
materials will be discussed.

Stress A
Fig. 12.1

X— permanent deformation

>
0 A Strain

12.1 Yield criteria:

Materials which exhibit ‘yielding’ followed by some plastic deformation
ductile materials
prior to fracture as measured under simple tensile or compressive stress are
termed ductile. This is a very important property as it provides a design
reserve for materials if they should exceed the elastic range during service. It
 YIELD CRITERIA AND STRESS CONCENTRATION 341

also has significance in relation to stress concentration as explained later in

this chapter. It is therefore essential to have a method of designing to avoid
the possibility of yielding under complex stress situations.
Plastic deformation is related to ‘slip’, as it is termed, in the grain
structure which is dependent on shearing action. There are two yield criteria
based on concepts of shear which are now accepted and used for ductile
materials. There are known as the maximum shear-stress (Tresca) criterion
and the shear-strain energy (von Mises) criterion. These will now be
developed.

Maximum shear-stress Theories of yielding are generally expressed in terms of principal stresses,
(Tresca) criterion since these completely determine a general state of stress. The element of
material shown in Fig. 12.2 is subjected to three principal stresses and it will
be taken that 0] > 02 > 03.
O41
The French engineer Tresca, who proposed this theory, made the
assumption that yielding is dependent on the maximum shear stress in the
material reaching a critical value. This is taken as the maximum shear stress
at yielding in a uniaxial tensile test. The maximum shear stress in the
complex stress system will depend on the relative values and signs of the
three principal stresses, always being half the difference between the
maximum and the minimum. It should be remembered that the minimum
stress can be zero or compressive, in which case it is negative in value.
For a general three-dimensional stress system, or in the two-dimensional
case with one of the stresses tensile, one compressive and the third zero, the
maximum shear stress is
Fig. 12.2
Tmax = (oO) es a3)/2

Under uniaxial tension there is only one principal stress, 01(02 = 03 = 0),
so that the maximum shear stress is

Tmax — 91 Hb

and at yield this becomes Ty = oy /2.

Therefore the Tresca criterion states that yielding will occur when
(1 —03)/2 = oy /2

or when the maximum principal stress difference equals the yield stress in
simple tension, 1.e.

01 — 03 = Oy [12.1]

For the case when two of the principal stresses are of the same type,
tension or compression, and the third is zero, then we have

Tmax = (oy = 0)/2 = a /2

and yielding occurs when

Fj P= Oy2 or OT = ay [12.2]

Shear-strain energy Huber in 1904 proposed that the total elastic strain energy stored in an
(von Mises) criterion element of material could be considered as consisting of energy stored due to
342 MECHANICS OF ENGINEERING MATERIALS

change in volume and energy stored due to change in shape, i.e. distortion
or shear. It was proposed that the latter contribution of stored strain energy
could provide a viable criterion for complex yield conditions. The same
criterion was also suggested independently by Maxwell, von Mises and
Hencky, but is now generally referred to as the von Mises criterion.

oO
Fig. 12.3

(a)

In order to show that the deformation of a material can be separated into

change in volume and change in shape, consider the element in Fig. 12.3
subjected to the principal stresses 0}, 72 and 03. These may be written in
terms of the ‘average’ stresses in the element as follows:

Tp o,
02 =7+ 0; [12.3]
03 =0+0;
where @ is the average or mean stress defined as

o = (0; +02 +03) /3 [12.4]

and the o’ represent the deviatoric stresses.

Now, when an element as in Fig. 12.3(4) is subjected to @ in all
directions, this hydrostatic stress will produce a change in volume, but no
distortion. Adding together eqns. (12.3) gives

01 +02 +03 =36+0,+0,+0%

but c= }(a; + 02 + 03); hence

o, +0, +0;,=0 [12.5]

But from the stress-strain relationships,

=o
F(a +04)
gy mea [12.6]
é = 3% (64+05)
Hence

epte,te=e’ Sed + 05 +04) [12.7]

YIELD CRITERIA AND STRESS CONCENTRATION 343

and since the sum of the three stresses is zero, eqn. [12.5],
Si epee 0 [12.8]
Thus the deviatoric stress components cause no change in volume but only a
change in shape.
We now turn to the determination of the strain energy quantities in the
expression
Urp=Uy
+ Us
where U7 = total strain energy, Uy = volumetric strain energy and Us =
shear or distortion strain energy.
The total strain energy per unit volume is given by the sum of the energy
components due to the three principal stresses and principal strains so that

Ur = 450 €1 + 50262 + 5.033 [12.9]

Substituting for the principal strains from the stress-strain relationships and
rearranging gives
V
U = OR (a, +o +04) (20102 + 20203 + 20301)
ff IE
per unit volume [12.10]
The volumetric strain energy can now be determined from the
hydrostatic component of stress, o.

Uy = 7 9e

3a
a sO (1 — 2v)

Substituting for o from eqn. [12.4] gives

D2:
Uy = bE a (0; +02 + 03) per unit volume fh
ea

But Us = Ur — Uy; therefore

]
Us= — |e; + 0% + Cn — 2v(0\02 + 0203 + 0301)|
2E
1-2
= ap 4+07 +03)

which reduces to
l+v
Us'= [(o1 a2)" + (a2 03)” (G3 o1)'|
6E
per unit volume [12.12]
or alternatively, using the relationship between £, G and v,
1
Us o2) + (2-03) + (03-01)
rs TYaliGe
per unit volume byes)
344 MECHANICS OF ENGINEERING MATERIALS

ee ee ee ap ae ee

Now, the shear or distortion strain energy theory proposes that yielding
commences when the quantity Us reaches the equivalent value at yielding in
simple tension. In the latter case 07 = 03 = 0 and a) = oy; therefore

Us = = per unit volume [12.14]

and

1 ey
polar +e a3) + (03 —01)'] =i
or

(o, —a1) + (or — on? + (os — a) = 20% [12.15]

In the two-dimensional system, 73 = 0 and

a; aieoa; = 0103 = oy [12.16]

for yielding to occur.

The above analysis has been directly aimed at establishing a yield
criterion on an energy basis. However, from eqn. [12.15] one might equally
well propose that yielding occurs as a function of the differences between
principal stresses. On this hypothesis it is evident that eqn. [12.15] can also
be obtained by considering the root mean square of the principal stress
differences in the complex stress system in relation to simple tension. Thus

{3 [(o1 -— 67) (0) — a3) + (3 an)

| = i (204)
[12.17]

The right-hand side of the equation is obtained for simple tension by

putting 0) = ay and o2 = a; = 0. Equation [12.17] reduces to

(a, — oa) + (a2 a3)" + (03 oO) =? av

which is the same as egn. [12.15].
An alternative presentation of eqn. [12.15] is

Cain sla ie a2)” + (02 — 03)" + (03 —- ate

where o, is the von Mises equivalent stress. The basis of the von Mises yield
criterion is that when o, reaches cy, the yield stress in simple tension, the
material is deemed to have yielded.
Many experiments have been conducted under complex stress
conditions to study the behaviour of metals and it has been shown that
hydrostatic pressure, and by inference hydrostatic tension, does not cause
yielding. Now any complex stress system can be regarded as a combination
of hydrostatic stress and a function of the difference of principal stresses,
and therefore a yield criterion such as that of Tresca or von Mises which is
based on principal stress difference would seem to be the most logical.
 YIELD CRITERIA AND STRESS CONCENTRATION 345

Yield envelope and locus For the case of three principal stresses, all non-zero, the shear-strain energy
criterion, eqn. [12.15], is represented by a circular cylinder whose
longitudinal axis is equally inclined to the three co-ordinate axes 0), 02,
o3 (Fig. 12.4). The surface of the cylinder represents the envelope between
an elastic stress system within the cylinder and a plastic stress state outside.

Maximum shear-stress Shear-strain

Fig. 12.4
criterion energy criterion

02

The maximum shear-stress criterion for three-dimensional states of

stress is represented by a hexagonal cylinder lying within the circular
cylinder as shown shaded in Fig. 12.4. Again, the hexagonal envelope
divides elastic from plastic or yielded stress states.
If we wish to consider two-dimensional stress states in which, say,
a3 = 0, then the yield boundary or locus is given by the intersection of the
0 02-plane with the two cylinders, as shown by the dashed lines in Fig. 12.4.
The yield loci for the above two criteria for ductile materials subjected to
two-dimensional principal stress states are illustrated in Fig. 12.5.

Fig. 12.5

| | | | | | | | >
1.0 0.8 0.6 0.4 0.2 0.20.40.60.8 1.0
0.2 Maximum 6, /Sy
shear stress,
0.4
0.6
0.8
1.0 Shear-strain
energy

A number of experimental studies have been carried out on various

ductile metals to establish the appropriate criterion for yielding under
346 MECHANICS OF ENGINEERING MATERIALS

various combinations of principal stresses. There are several classical test .

methods for studying complex stress behaviour. These include a thin-walled
tube subjected to internal pressure and tensile or compressive axial load, or a
tube under combined torsion and axial load or bending. Suitable
measurements of strain or deformation are recorded as applied loading is
increased to a point where linear elasticity is no longer obtained. The
principal stresses may then be calculated and plotted to represent yielding,
as shown in Fig. 12.5. It is seen that there is close correlation with the von
Mises shear-strain energy criterion and that the Tresca maximum shear-
stress criterion is also satisfactory but somewhat more conservative.

Example 12.1
A mild steel shaft of 50mm diameter is subjected to a bending moment of
1.9kNm. If the yield point of the steel in simple tension is 200 MN/m2, find the
maximum torque that can also be applied according to: (a) the maximum shear
stress; (b) the shear-strain energy theories of yielding.

The maximum bending stress occurs at the surface of the shaft and is given
by
_ 32M 32 . 1900
On = 155 MN/m
Ways ia tr5 1058
The maximum shear stress at the surface is
Lovee 16 if
Tay = md3
=—x
‘ge WAS MY
a TI

(a) Maximum shear-stress theory

iar S210

ee ere 2

on 4472
V(a, + AT) — 200 x 10°
2

155” + [4 x (0.04077)7] = 2007

0.001 6677 = 4000
T = 1.55kNm

(b) Shear-strain energy theory

o} + 05 — 0102 = (200 x 10°)’

Putting o? + ae, =A,

i(ox + VA)’ +4 (oe — VAY —4} (6, — VA)(o4 + VA)

= (200 x 10°)?
 YIELD CRITERIA AND STRESS CONCENTRATION 347

tot +20, JAF A402 -20,/A+A-07 +A]

= (200 x 10°)?
Simplifying gives

a, + 372, = (200 x 10°)’

155? + [3 x (0.001 6677)| = 2007
0.001 6677 = 5330
Therefore

T = 1.79kNm

Example 12.2
P A thin-walled steel cylinder of 2m diameter is subjected to an internal pressure of
2.5MN/m?. Using a safety factor of 2 and a yield stress in simple tension of
400 MN/m2, calculate the wall thickness on the basis of the Tresca and von Mises
yield criteria. It may be assumed that the radial stress in the wall is negligible.

The stress system in the wall of the cylinder consists of three principal
stresses, circumferential, axial and radial, of which the last may be neglected
and will be taken as zero. Hence, using eqns. [2.10] and [2.11] we have

O71 en and O2 Aa
—=
t 2t

(a) Tresca criterion Since both axial and circumferential stresses are
tension the maximum difference between principal stresses gives

ric = va = ui= > at yielding

Hence

Oj = oy

Therefore
2.5 x 1000 400
t 2
t= )7 510m

(b) Von Mises criterion For 03 ~ 0 we have

2 2
O71 + 0, — 0102
2
= dy
348 MECHANICS OF ENGINEERING MATERIALS

3 pr Li
oF

= (3\* pre (BN? 28% 1000

a Ee a 200
t = 10.8mm

The slightly larger plate thickness given by the Tresca criterion

illustrates its more conservative characteristic compared with the von Mises
criterion.

12.2 Fracture criteria:

Brittleness in a material may be defined as an inability to deform plastically.
brittle materials
Materials such as glass, some cast irons, concrete and some plastics, when
subjected to tensile stress, will generally fracture at or just beyond the elastic
limit. We are therefore not much concerned with a yield criterion as a
fracture criterion for brittle materials under complex principal stress states.
Traditionally the most widely used criterion has been that suggested by
Rankine known as the maximum principal stress theory.

Maximum principal This hypothesis, proposed by Rankine, which was also intended for use to
stress (Rankine) predict yielding of a ductile material, states that ‘failure’ (i.e. fracture of a
criterion brittle material or yielding of a ductile material) will occur in a complex
stress state when the maximum principal stress reaches the stress at ‘failure’
in simple tension. The two-dimensional locus for this theory is illustrated in
Fig. 12.6. It will be noticed that in the first and third quadrants the
boundary is the same as for the maximum shear-stress theory.

Fig. 12.6 02
A
OF
Maximum principal
stress theory

® Cast iron

°o
7 Be O71

Mohr fracture criterion Some materials, such as case iron, have much greater strength in
compression than in tension. Mohr proposed that, in the first and third
quadrants of a ‘failure’ locus, a maximum principal stress theory was
appropriate based on the ultimate strength of the material in tension or
compression respectively. In the second and fourth quadrants where the two
 YIELD CRITERIA AND STRESS CONCENTRATION 349

Fig. 12.7 ©» tension

— Oc sabi Out ~

0, compression ~< —> 06, tension

Y
©» compression

principal stresses are of opposite sign the maximum shear-stress theory

should apply.
This results in a diagram as shown in Fig. 12.7.

Fracture mechanics The modern approach to fracture in brittle materials is to recognize that all
real materials contain defects which are capable of initiating failure without
yielding in a brittle material. The likelihood of a specific defect causing
failure under a particular stress system can be assessed using the procedures
illustrated in Chapter 20. This may involve the use of statistical methods due
to the random nature of inherent flaws in real materials.

Example 12.3
In a cast-iron component the maximum principal stress is to be limited to one-third
of the tensile strength. Determine the maximum value of the minimum principal
stress using the Mohr theory. What would be the values of the principal stresses
associated with a maximum shear stress of 390MN/m2? The tensile and
compressive strengths of the cast iron are 360MN/m? and 1410MN/m2
respectively.

Fig. 12.8

Oyo = 1410 yt = 360

350 MECHANICS OF ENGINEERING MATERIALS

Maximum principal stress = 360/3 = 120 MN/ m? (tension). According to

Mohr’s theory, in the second and fourth quadrant

Out Ouc

Therefore

120 02
cnet = and a) = —940 MN/m?
360 Ls—1410
Mohr’s stress circle construction for the second part of this problem is
shown in Fig. 12.8. If the maximum shear stress is 390 MN/m‘’, a circle is
drawn of radius 390 units to touch the two envelope lines. The principal
stresses can then be read off as +200 MN/m? and —580 MN/m’.

12.3 Strength of
In the previous sections we have been dealing with isotropic materials, so
laminates
that it has been possible to develop a failure criterion on the basis of one
limiting parameter (the yield strength). For fibre-reinforced composites,
however, it is not possible to define failure in terms of a single parameter. It
was shown in Chapter 11 that unidirectional composites have a longitudinal
strength which is many times the transverse and shear strengths. Thus when
a multi-axial stress system is applied it is necessary to consider the effect of
this in relation to the various strength components of the composite. For
laminates the picture is complicated still further by the different orientations
of the individual lamina or ply in relation to the applied stress system.
However, since a chain is only as strong as its weakest link, the strength of a
laminate will be determined by the strength of the individual plies within the
laminate.
One of the most popular failure criteria for laminates is the Tsai—Hill!
criterion. This is based on the von Mises failure criterion which was
expanded by Hill to anisotropic bodies and applied to composites by Tsai.
The criterion may be expressed as
2 2 2
Ox OxOy Oy a
@ (ie a ) es ra va
where
a, = the stress parallel to the fibres;
a, = the stress perpendicular to the fibres;
T,. = the tensile strength parallel to the fibres;
i
&y the tensile strength perpendicular to the fibres;
Txy,Syy = the shear stress and shear strength values.

The Tsai—Hill criterion would be used as follows. Consider a laminate

subjected to in-plane stresses which are applied relative to the global X—Y-
directions. The strains may then be calculated in the X— Y-directions using
the compliance matrix [a] (see Chapter 11)

dyer a aoe [12.19]

 YIELD CRITERIA AND STRESS CONCENTRATION 351

For example, if there is only a stress in the global X-direction, then

EX =@10x Ey =a20x Yxy =0

For in-plane loading the strains will be the same in all the plies so that
these strain values may be used to get the stress in each ply. Using the
stiffness matrix [4] for each ply,

Ox Se
Oy = [Aly Ey [12.20]

VO ply YXY ply/ laminate

From the terminology for [A] in Chapter 11, this would give

ox = Quex + Qney + Qierxy

oy = Qnex + Qney + Qeyxy
Txy = Qsiex + Qerey + Qosyxy
These are the stresses in each ply in the global X—Y-directions. They
would then need to be transferred to the local x—y-directions using the
transformation matrix [7].

Ox Ox

pees Val eaen: (12.21]

Txy WOOF

At this point the Tsai—Hill criterion could be used to establish whether

or not failure would be expected in the ply under consideration. To do this
eqn. [12.18] could be applied or the alternative form shown below which is
popular because it gives a safety factor S.F.:

S.F.= Ts [12.22]
Ve Onto (Tal 1a)a, + (17/55)

12.4 Concepts of stress

In previous chapters the problems analysed have had stress distributions
concentration
which were either uniform or varied smoothly and gradually over a
significant area. However, in the vicinity of the point of application of a
concentrated load there is a rapid variation in stress over a small area, in
which the maximum value is considerably higher than the average stress in
the full section of the material. This situation is known as a stress
concentration.
The cause of stress concentration is perhaps most readily understood
from consideration of analogous systems such as the flow of a fluid. In a
simple strut subjected to an axial compressive force as shown in Fig. 12.9(a),
the force is transmitted through the strut via the medium of the stresses
exerted on every small element of material by its neighbouring elements.
The lines of transmission of the stress are similar to the flow lines which
would be observed if a fluid entered (at the point of application of the force)
a channel of the same cross-section as the strut. The densely packed flow
lines at the entry and exit points are representative of the concentration of
stress at those points.
352 MECHANICS OF ENGINEERING MATERIALS

Fig. 12.9

SS. ee
——
(b)

Another way in which a stress concentration can be produced is at a

geometrical discontinuity in a body, such as a hole, keyway, or other sharp
change in sectional dimensions. Figure 12.9(/) again uses the flow analogy to
illustrate the stress concentration effect which occurs at notches and holes.
Two questions come to mind when considering the above effect. If, as 1s
the case, all points of support, load application and geometric discontinuity
disturb the uniformity of stress and cause stress concentration, it 1s
surprising that one can obtain realistic results by, say, the simple bending
and torsion theories considered earlier. This problem was studied
theoretically by St Venant, who stated the following principle.

If the forces acting on a small area of a body are replaced by a statically

equivalent system of forces acting on the same area, there will be
considerable changes in the local stress distribution, but the effect on
the stresses at distances large compared with the area on which the
forces act will be negligible.

For instance, in a bar gripped at each end and subjected to axial tension,
the stress distribution at the ends will vary considerably according to
whether gripping is by screw thread, button head, or wedge jaws. However,
it has been shown that, at a distance of between one and two diameters from
the ends, the stress distribution is quite uniform across the section.
Similarly, it is immaterial how the couples are applied at the ends of a beam
in pure bending. So long as the length is markedly greater than the cross-
sectional dimensions, the assumptions and simple theory of bending will
hold good at a distance of approximately one beam depth away from the
concentrated force.
The second feature of interest is that, although a stress concentration is
only effective locally (St Venant), the peak stress at this point is sometimes
far in excess of the average stress calculated in the body of the component.
Why is it then normal practice under static loading to base design
calculations on the main field of stress, and not on the maximum stress
concentration value where a load is applied? For instance, a point or line
application of load would theoretically cause an infinite elastic stress in the
material under the load. This obviously cannot occur in practice since a
ductile material will reach a yield point and plastic deformation will occur
under the point of application of the load. The effect of the plastic flow is to
cause a local redistribution of stress, which relieves the stress concentration
slightly so that the peak value of stress does not continue to increase with
 YIELD GRITERTA AND STRESS GCONCENDRATION 353

increasing load at the same rate as in the elastic range. Eventually, with still
greater loading, general yielding in the body of the material will tend to
catch up and encompass what was the stress concentration area.
Brittle materials have little or no capacity for plastic deformation and
therefore the stress concentration is maintained up to fracture. Whether or
not there is an accompanying reduction in nominal strength depends largely
on the structure of the material. Those such as glass and some cast irons
which have inherent internal flaws, which themselves set up stress
concentration, show little reduction in strength over the unnotched
condition. Others which have a homogeneous stress-free structure will
show a considerable decrease in static strength for a severe notch.
From all of the foregoing it would appear that stress concentration does
not present too serious a problem for components in service. However, there
are two main aspects of material behaviour in which stress concentration
plays the major part in causing failure. These are fatigue and brittle fracture
(notch brittle reaction of a normally ductile metal), both of which topics are
dealt with at length in later chapters.
The theoretical analysis of stress concentration is generally very complex
by classical mathematics. Many theoretical solutions are due to Neuber, and
a number of individual problems have been solved by other theoreticians
and are available in published handbooks*’. The development of finite
element analysis in recent times has provided many more solutions.
The principal experimental method which has provided many simple
and accurate solutions to problems of stress concentration is the technique
of photoelasticity.

12.5 Concentrated loads

and contact stresses
Concentrated load on The local distribution of stress at the point of application of a concentrated
the edge of a plate load normal to the edge of an infinite plate was first studied in 1891 using
photoelasticity. This led to the theoretical solutions a year later of
Boussinesq and Flamant.
F

|
Fig. 12.10

(a) (b)

Consider the three systems of forces shown in Fig. 12.10 acting on the
edge of an infinitely large plate of thickness ). The resultant force in each
case is the same, and hence the systems are statically equivalent and
therefore satisfy the principle of St Venant. Now, case (a) is the one we wish
to solve, but this will result in practice in a small volume of plastic flow as
explained previously. To overcome this difficulty we replace the point load
on the straight edge by a radial distribution of forces, as in (4) or (c), around
a small semi-circular groove. Experiment has shown that the forces in (c)
give the better representation of the stress distribution due to a concentrated
load on a straight edge. The solution by Flamant on this basis shows that the
354 MECHANICS OF ENGINEERING MATERIALS

stress distribution is a simple radial one involving compression only. Using

polar co-ordinates, and referring to Fig. 12.11, any element distance r from
O at an angle @ to the normal to the edge of the plate is subjected to a simple
radial compression only of magnitude
2F cos @
0,=
Tbr

Fig. 12.11

Concentrated load The cross-section of the beam at which the load is acting is subjected to a
bending a beam complex stress condition composed of the stress due to simple bending plus
the stress due to the concentrated load itself.

Fig. 12.12

(a) (b)

Considering the radial pressure distribution on the small groove in Fig.

12.12(a), then the horizontal components give rise to forces F/7 acting
' parallel to the edge of the beam, so that the system of forces equivalent to
the pressure distribution is as shown in Fig. 12.12(d). In this problem we are
not considering an infinite plate, but a beam of finite depth, and.
consequently the horizontal forces, F/7, set up longitudinal tension and
bending stresses. The former is given simply as load divided by area or
F/m x (1/bd). The latter are determined by considering the bending
moment about the axis of the beam given by (f/m) x }d. The bending
stresses are therefore

The total stress acting across the section OA of the beam is then obtained by
the superposition of the various separate quantities:
Fi sige
Ce = ae
4 Teneae

= 4 1 d\1lFy F
(| =) es aad
 YIELD CRITERIA AND STRESS CONCENTRATION 355

Fig. 12.13 F

Actual bending,
small span Simple bending

Actual bending,
large span

This expression is often referred to as the Wilson—Stokes solution.

The distribution of ¢,,, for long and short spans is compared with the
simple bending distribution in Fig. 12.13, and it is seen that the more
accurate solution gives rise to maximum longitudinal stresses which are Jess
than those from simple bending theory.
Hence in this problem, although the stress concentration causes high
normal compressive stresses, the tensile bending stress which would be
expected to be the cause of failure is in fact reduced by the concentrated
load.

Contact stress Another important problem involving stress concentration is the condition
of contact of two bodies under load. Typical examples may be found in the
mating of gear teeth, in a shaft in a bearing, and in the balls and rollers in
bearings. Solutions are too complex and lengthy to be considered here but a
few useful results will be quoted.
Considering first the situation of a ball under loaded contact with a flat
surface, Fig. 12.14, the point of contact when unloaded develops into a small
spherical surface when under load, which is initially elastic deformation.
The radius of the circular contact area is given as

FD\'/?
r= = (38| =
os8(7)
where F is the contact force, D is the diameter of the ball and FE is the
Fig. 12.14 modulus of the two materials (assumed the same in this instance).
The distribution of pressure over the contact area is such that a
maximum value occurs at the centre of the circle equal to

P)
FE 1/3

Ore = VO? &

From the dimensions and pressure on the contact surface the stress
distribution can be calculated along the axis normal to the contact. If
maximum shear stress is taken as the criterion for yielding, it is found that
the greatest value occurs not at the surface of contact but at a small depth
below the surface in each body. It is generally at this point that failure of the
material would originate if the loading were excessive.
For a roller in contact with a plane surface, Fig. 12.15(a), the contact area
is rectangular of length /, and width , with maximum pressure occurring at
the centre of the rectangle.
356 MECHANICS OF ENGINEERING MATERIALS

Fig. 12.15

(a) (b)

where F is the contact force, D the diameter of the roller and / its length, and
E the modulus of the two materials, as before. The maximum pressure is
given as
FE\!2
Cam OOo (5)

The contact dimensions and peak pressures for contact between two
spheres of different diameter or two rollers of different diameter as shown in
Fig. 12.15(d) can be found from the above formulae if the sphere or roller
diameter D is replaced with twice the relative radius of curvature R. The
relative radius of curvature between two surfaces of radii R; and R> is given
by
[ea Ree
RR, R, R,R2

The contact of a ball of radius R; with a spherical seat is represented by

taking R> as the negative of the seat radius. Contact between a roller and a
cylindrical seat is treated similarly. It should be noted, however, that these
analyses are only valid when the contact dimensions are small relative to the
radii of curvature; they will not model accurately the contact between
heavily loaded, closely conforming components.

12.6 Geometrical
It was previously explained in the introduction that abrupt changes in
discontinuities
geometry of a component give rise to stress concentration in a similar
manner to those described in previous sections for loading. In most cases,
the failure of a component can be attributed to some form of geometrical
stress raiser, from either bad design or misfortune.
Typical examples of stress raisers are oil holes, keyways and splines,
threads, and fillets at changes of section.
Figure 12.16(a) illustrates a bar under tensile loading into which has
been machined two grooves. The uniform stress is o at a distance from the
discontinuity where the cross-sectional area is 4}. The mean or nominal
 YIELD CRITERIA AND STRESS CONCENTRATION 357

Fig. 12.16

Mean stress
© at section
x

Co) o
(a) (b)

stress at section XX, where the area is 42, will be a(4)/A2). However, it is
seen that at the base of the grooves there is a peak value of 0 which is much
higher than the average stress. There is a steep stress gradient to the lower
levels of stress in the central region of the bar.
The area under this stress curve must be the same as the area under the
mean stress lines, since the applied load is the same in each case.

Stress concentration The stress concentration set up by such geometrical discontinuities is a

factor function of the shape and dimensions of the discontinuity, and is expressed
in terms of the e/astic stress concentration factor denoted by K;:
maximum boundary stress at the discontinuity
es average stress at that cross-section of the body
This factor is constant within the elastic range of the material. Since this is
dependent on the geometrical proportions and shape of the notch and the
type of stress system (tension, torsion, etc.), it is readily appreciated that to
cover a wide range of parameters would require a great deal more space than
part of one chapter. Furthermore, the limitations on theoretical solutions for
stress concentration at notches have resulted in many analyses being
obtained experimentally (generally photoelastically) and presented as charts
of K, against geometrical proportions. The subject has been treated very
thoroughly by Neuber’, Peterson’, Frocht* and others. For complex three-
dimensional stress concentrations, finite element techniques are frequently
used nowadays.
A few cases of special interest will now be considered.

Circular hole The distribution of axial and transverse stress at the hole cross-section is
illustrated in Fig. 12.16(b). The peak stress arises on plane XX at the edge of
the hole and, for a small hole, in a thin infinite plate subjected to tension, the
stress concentration factor K, = 3. This is the highest value obtained and,
for a circular hole in a finite width strip under axial loading K,, it lies
between 2 and 3. The relationship between K, and geometrical proportions
is plotted in Fig. 12.17.
In pure bending of a finite width plate with a transverse hole, K, is a
function of plate thickness as well as of radius of hole r and width of plate wm.
For a very thick plate and small hole, the K, against r/m curve is identical
358 MECHANICS OF ENGINEERING MATERIALS
| |!

Fig. 1217 Stress concentration 3.0

factor K, for axial loading of a
finite width plate with a trans- 28
verse hole (Peterson*; by courtesy
of John Wiley & Sons, Inc.)
2.0m

2.4)5—

2.2oa

kK, = 2" based on net section

K, 2.0/-
: Snom

Srom = (w-a)h

1.8

1.6

1.4 fee

1.2 ig

1.0 Lease
ee ee |
0) 0.1 0.2 0.3
a
Ww

with Fig. 12.17 for the tension strip. When the hole is large and the plate
thin, K, varies from 1.85 to 1.1 for decreasing width.
The case of a shaft with a transverse hole subjected to tension, bending
or torsion is a common one. It introduces the feature of a three-dimensional
stress system as against plane stress in the plate case. The stress
concentration factor in a triaxial stress field, although defined in the same
way as for a biaxial stress system, will be denoted by K/. Frocht has studied
photoelastically a circular bar with a transverse hole under tension and has

Fig. 1218 Summary of experi- Soca

mental results for shafts with
transverse holes in tension
(Frocht*; by courtesy of John
Wiley & Sons, Inc.)

—-—-— Three-dimensional stress concentration factor

—— Two-dimensional stress concentration factor
of
K;
Factor
concentration
stress
 YIELD CRITERIA AND STRESS CONCENTRATION 359

obtained a curve of Kj against r/d as shown in Fig. 12.18. Also plotted here
for comparison is Howland’s curve for K, for a plate under tension, and it is
interesting to note that Kj is noticeably higher.

Fillet radius Almost without exception cylindrical components do not have a uniform
diameter from one end to the other. The journal bearings of a crankshaft
have to mate into the web, and a motor shaft or railway axle requires
shoulders to retain bearings or a wheel. At these changes of section a sharp
internal corner would introduce an intolerable stress concentration which
could lead to failure by fatigue. The problem is lessened by the introduction
of a fillet radius, to blend one section smoothly into the next. Even a fillet
radius will give rise to some stress concentration; however, this will be
considerably less than with the sharp corner. The stress concentration at a
fillet radius is not readily amenable to mathematical treatment and solutions
have been obtained by photoelasticity and other experimental means.
Values of K, for various geometrical proportions have been obtained for
shafts in tension, bending and torsion. The last two cases are probably the
most common in practice and therefore only the charts for these, in Figs.
12.19 and 12.20, have been included, for reasons of space.

Fig. 12.19 Stress concentration 3.0

factor K, for the bending of a
shaft with a shoulder fillet 2.8
(Peterson*; by courtesy of John
Wiley & Sons, Inc.)
2.6

2.4

22

K, 2.0

1.8

1.6

1.4

th

1.0 RON se Ve sl elt ee)

00.02 0.06 0.10 014 018 0.22 0.26 0.30
r/d

Keyways and splines in Another design feature which requires careful attention is the keyway or
torsion spline in a shaft subjected to torsion.
The rectangular keyway of standard form having root fillets has been
solved mathematically by Leven, and the curve for K,, against the ratio of
360 MECHANICS OF ENGINEERING MATERIALS

Fig. 12.20 Stress concentration

factor K,, for the torsion of a shaft
with a shoulder fillet (Peterson’;
by courtesy of John Wiley &
Sons, Inc.)

16M
Kis ==
Tmax
Where Thom a= eon:
2.2. Thom

Kis 2.0

1.8

1.6 ie

1.4

Ly

1.0 |
00.02 0.006 010 014 018 0.22 0.26 0.30
r/d

fillet radius to shaft diameter is shown in Fig. 12.20. The maximum shear
stress occurs at a point 15° from the bottom of the keyway on the fillet
radius.

Gear teeth The stress distribution in a loaded gear tooth is a complex problem. Stress
concentration at the point of mating of two teeth due to contact load varies
in position with rotation of the teeth, and further stress concentration occurs
at the root fillets of a tooth, the latter being the more serious. The former
can be analysed with the aid of the expressions given in the section under
contact stresses. The fillet stress concentration is a function of the load
components (causing bending and direct stress) and the gear tooth
geometry.
A very comprehensive investigation was conducted by Jacobson*, who
studied involute spur-gears (20° pressure angle) photoelastically and
produced a series of charts of strength factor against the reciprocal of the

Fig. 12.21
 YIELD CRITERIA AND STRESS CONCENTRATION 361

number of teeth in the gear, which cover the whole possible range of spur-
gear combinations. Figure 12.21 shows the photoelastic stress pattern and
this clearly indicates the areas of stress concentration at the root radius and
point of contact.

Screw threads One of the most common causes of machinery or plant having to be shut
down is the fatigue of bolts or studs. This is principally due to the high
stress concentration at the root of the thread®. For a bolt and nut of
conventional design, the load distribution along the screw (Fig. 12.22) is far
from uniform and reaches a maximum intensity at the plane of the bearing
face of the nut. This is due in part to the unmatched and opposing signs of
the strains in the screw and nut. This can be overcome to a large extent by
altering the design of the nut, principally so that the nut thread is in tension
and so matching the strains more evenly with the screw.

Fig. 12.22 Stress distribution in

Free
ordinary stud and nut (Brown end
and Hickson®; reprinted by per-
mission of the Institution of
12 |
Mechanical Engineers) e

& 10 e Stud t
ie or
S&eEo ° Nut
Geers
Saco fe
;
g = & 8 6 e
Si o= 4.00 Mean value for Oe
oes
Hor 4 eee
stud over
>
length of nut eo. Z
ao e a i 3° = pais
Coe
Pop
ee 'e
os Pr]
53 2 !
| eve Le
\ ‘e.
| ° =

—_— :
feats ° ° ° '

i
|

Example 12.4
A shaft is stepped from 48mm diameter to 40mm diameter through a fillet of
2.4mm radius. A gear wheel is keyed to the larger diameter of the shaft and the
radii at the internal corners of the keyway are each 1.5mm. What is the maximum
torque that can be transmitted if the steel has a shear stress limit of 200 MN/m2?

D/d = 48/40 = 1.2 and r/d =2.4/40

= 0.06
From) Fig. 12.207 K,.= 1.5 at the fillet radius. At the keyway
r/d =1.5/48 = 0.031. From Fig. 12.20, K,, = 2.4. Now
Tmax
Ki, =
Tnom

where

Tmax = 200 MN/n2

_ 167
and Ui = sae
ta?
362 MECHANICS OF ENGINEERING MATERIALS
i

At the fillet radius

161s os LODO:
rx0.048 1.5
1 =i16/6Nm

At the keyway

16T oe 2005C 10°

ax 0.0483 24
T = 1810Nm
The maximum torque that can be transmitted is 1676 Nm which is
governed by the shear stress at the fillet radius.

12.7 Yield and plastic

The usual definition of stress concentration factor, as given in the
stress concentration introduction, is in terms of the maximum stress at the discontinuity.
factors ‘There is of course a biaxial stress condition at the free surface of a notch,
and therefore it might be more realistic to express stress concentration in
terms of the maximum ‘equivalent stress,’ i.e. using one of the yield criteria
discussed earlier. If the biaxial stresses can be determined theoretically at the
free surface of a notch, then considering the shear-strain energy criterion,
which seems the most appropriate for both static and fatigue stress
conditions in a ductile material, we have for 0) > a2 and o3 = 0

Oa) (Op 0107 >)

Then
O equiv
C=
On om

In the elastic range there is a constant ratio between a; and a2 for any
particular point on the surface; therefore

Oo} = kaon

and

x = 2evill = (0/8) + 1/8)

Onom

It is seen from the above that K, is always less than K,, except when k = 1
and then K, = K,. This may partly explain why notched strength reduction
in fatigue (Chapter 20) is nearly always less than would be expected from
considerations of K;,.
If loading is increased to the point where yielding occurs at the notch,
there is a redistribution of stress similar to that at a concentrated load. The
stress concentration factor is still defined in the same way as in the elastic
range but is now denoted by Ky, the plastic stress concentration factor, and
this is now also a function of the degree of plastic deformation that has
occurred. It is found in practice that, for ductile materials, as Oma. > Cult,
K, — 1, 1.e. the stress concentration does not reduce the strength of a
 YIELD CRITERIA AND STRESS CONCENTRATION 363

component loaded statically. However, in fatigue (Chapter 20) stress

concentration only achieves a plastic state in a microscopic locality which is
contained in a larger elastic stress concentration area. This is the cause in all
cases of the initiation of the fatigue crack.

12.8 Summary
The latter part of this chapter demonstrates that although designers basically
work in the elastic range of materials and indeed employ safety factors to
make that more sure, the presence of stress concentration can still result in
some local yielding. It is therefore very important to have acceptable yield
criteria to apply to complex stress states. For ductile materials it is now well
proven that the shear-strain energy and maximum shear-stress theories both
give satisfactory predictions of the onset of yielding. Brittle materials, on the
other hand, fracture rather than yield, and this situation can be adequately
designed for on the basis of the maximum principal stress theory or the
Mohr modified maximum shear-stress criterion.
There are plenty of data nowadays on stress concentration factors owing
to the primary effects in fatigue failure and fracture mechanics which will be
discussed in later chapters. From a design point of view we obviously cannot
eliminate stress raisers entirely, but it is possible to minimize their effect by
careful attention to the detail of points of load application, appropriate
surface hardening and heat treatment and the avoidance of sharp internal
corners and sudden changes of section.

References
1. Tsai, S. W. and Hahn, H. T. (1980) /ntroduction to Composite Materials,
Technomic, Westport, CT.
2. Neuber, N. (1946) Theory of Notch Stress, Edwardes, MI.
3. Peterson, P. E. (1974) Stress Concentration Factors, John Wiley, New
York.
4. Frocht, M. M. Photoelasticity, Vol. I (1941), Vol. II (1948), John Wiley,
New York.
5. Jacobson, M. A. (1955) ‘Bending stresses in spur gear teeth’, Proc.
I.Mech.E., 169, 587-694.
6. Brown, A. F. C. and Hickson, V. M. (1952-1953) ‘A photoelastic study
of stresses in screw threads’, Proc. I.Mech.E., 16, 605-612.
7. Young, W. C. (1989) Roark’s Formulas for Stress & Strain, McGraw-
Hill International, New York.

Problems 12.1 A shaft subjected to pure torsion yields at a torque of 1.2kNm. A

similar shaft is subjected to a torque of 720 Nm and a bending
moment M. Determine the maximum allowable value of M
according to (a) maximum shear stress theory, (4) shear strain
energy theory.
12.2 Show that the maximum shear stress in a helical spring subjected
either to axial force or axial couple is independent of the helix angle
of the coils.
An open-coiled helical spring has ten coils of 50mm pitch and
76mm mean diameter made from steel wire of 12.7 mm diameter. If
the 0.1% proof stress for the steel is 840 MN/m?, determine the
364 MECHANICS OF ENGINEERING MATERIALS

allowable axial load according to the maximum shear stress criterion.

E = 206 GN/m’, v = 0.3.
125 A circular steel cylinder of wall thickness 10mm and _ internal
diameter 200mm is subjected to a constant internal pressure of
15 MN/m/?. Determine how much (a) axial tensile load, and (0) axial
compressive load can be applied to the cylinder before yielding
commences according to the maximum shear stress theory. The
yield stress of the material in simple tension is 240 MN/m?*. Assume
that the radial stress in the wall of the cylinder is zero. Sketch the
plane yield stress locus for the maximum shear stress theory, and
show the two points representing the cases above.
124, A thin walled cylindrical tank has a diameter of 2 m. If it is subjected
to an internal pressure of 1MN/m/?, calculate the required wall
thickness for the tank if yielding is not to occur. The von Mises
theory should be used to check for yielding. The yield stress in
simple tension for the material is 280 MN/m/? and a factor of safety
of 1.5 should be used.
VR) A hollow tube with an external diameter of 100mm and a wall
thickness of 5mm is subjected to an axial force of 100 kN. If the
tensile yield stress for the material is 280 MN/m’, use the Tresca
Criterion to establish whether or not a torque of 6kNm could be
applied without causing the tube to yield. A safety factor of 2 should
be used to allow for stress concentrations.
250 mm| : — 12.6 Part of a supporting bracket for a machine consists of a steel rod of
12mm diameter fixed at its lower end and containing a right-angle
| = 12 mm dia. bend which lies in the xy-plane shown in Fig. 12.23. The force P
applied at the free end lies in the yz-plane inclined at 30° to the z-
axis. The 0.1% proof stress for yielding in simple tension of the steel
is 200 MN/m‘?. Calculate the value of P which will cause yielding at
point A on the outer surface according to the shear-strain energy
criterion.
Fig. 12.23 Wie In the pulley system shown in Fig. 12.24 the pulleys cause an
additional load of 500N each. Calculate a suitable shaft diameter so
as to avoid failure by the maximum shear stress criterion. The tensile
yield strength of the shaft material is 248 MN/m*. The shaft weight
may be neglected and the shaft bearings may be treated as simple
supports.
12.8 A cast-iron tube 50mm and 40mm outside and inside diameters
respectively, is being assembled into a structure. Owing to
misalignment it is subjected to a torque about the longitudinal axis
of 2.5kNm and a tensile force of 50kN and it fractures. It was

Fig. 12.24

500 N | 3000 N 3000 N | 600 N

<0.5 m>|/<—— 1.5m ——><0.5 m

 YIELD CRITERIA AND STRESS CONCENTRATION 365

discovered that the line of application of the tensile force was parallel
to the axis of the tube but offset from it. Calculate the amount of
eccentricity which must have occurred to cause failure of the tube
according to the maximum principal stress theory. The failure stress
of the cast iron in simple tension is 280 MN/m?.
Was) A cast-iron cylinder of 60mm internal diameter and 5mm wall
thickness is to be used to check the Mohr theory of failure. The
tensile and compressive strengths of the material have been
measured as 400 MN/m? and 1200 MN/m* respectively. Deter-
mine (a) the internal pressure to cause failure, and (6) the axial
compressive load to cause failure when combined with an internal
pressure of 50 MN/m?.
12.10 An aircraft fuselage spacing strut is made from carbon-fibre-
reinforced epoxy and has the shape shown in Fig. 12.25. In the
manufacture of the strut, six plies are laid in the loading direction
and then two plies at 60° and two plies at —60° from the loading
axis. In service, the axial load causes a compressive stress of
70 MN/m* in the strut. Use the Tsai-Hill criterion to establish
whether or not the composite would be expected to fail at this stress
and, if not, determine the stress at which it would fail. The
compressive strengths of a unidirectional carbon-fibre composite are
840 MN/m/ in the fibre direction and 42 MN/m’ in the transverse
direction. The shear strength is 56MN/m?. Also for the
unidirectional lamina, E, = 207 GN/m’, 1 aie GN/m’, Gy =
4.9 GN/m’, vy = 0.3.

Fig. 12.25 70 MN/m2

—_—————

Za A filament-wound composite-pressure vessel consists of carbon fibre

in an epoxy matrix. The vessel is cylindrical with a diameter of
600 mm and a wall thickness of 10mm. The fibres are arranged with
two plies at 45°, two plies at —45° and six plies at 0° to the axis of the
cylinder. Use the Tsai-Hill criterion to estimate the internal pressure
which would cause the vessel to fail. For a unidirectional composite
using the fibre-matrix combination the following data applies:
T= 207,GN/m t= 1200 MN/m, Fo =77-GN/in ie
28 MN/m’, G,, = 4.9 GN/m’, Ty = 43 MN/m’, v4 = 0.3.
BastBi A steel beam of width 30mm and of depth 90mm is simply-
|
suported at each end of a 2m span. It is subjected to a concentrated
load of 5kN at mid-span. Determine the approximate depth of
compressive yielding beneath the loading point. Taking into account
the effect of the concentrated load, compare the maximum resultant
366 MECHANICS OF ENGINEERING MATERIALS

tensile stress on the lower surface of the beam with that obtained by
simple bending theory. Compressive yield stress = 400 MN/ m?.
12.13 A railway-wagon wheel is 500mm in diameter and has an
approximate contact width on the rail of 40 mm. If the compressive
yield stress of the rail steel is 600 MN/m? and the modulus is
208 GN/m? determine the working load that can be carried per
when using a load factor of 2.
12.14 A spherical steel pressure vessel is 3m in diameter and contains a
hole of 200mm diameter to accommodate a safety valve. If the
working pressure is 1.4 MN/m/? determine a suitable value for the
shell thickness. The allowable tensile stress for the steel is
350 MN/m”. (Hint: Make a reasonable estimate for K;, using Fig.
ay)
ag bs) A shaft projects through a roller bearing from where it may be
assumed to be cantilevered. It is 50mm diameter for a length of
100 mm and then is stepped down to 25mm diameter for a further
100 mm to the free end. At this point a ioad of 2.68 kN is applied. If
the limiting design bending stress is 280 MN/m? determine a
suitable value for the filet radius at the change of section. What is the
safety factor at the bearing housing? (See Fig. 12.19 for relevant
data.)
—— 3 Variation of Stress and Strain

In Chapter 11 the conditions of stress and strain at a poingt in a material were

considered. It is now necessary to take the analysis a stage further by examining the
variation of stress between adjacent points and deriving suitable expressions for this
variation. As in the earlier work, relationships for stresses may be found by
considering the equilibrium of a small element of material. The solution of these
equations of equilibrium must satisfy the boundary conditions of the problem as
defined by the applied forces. However, it is not possible to obtain the individual
components of stress directly from the above equations, owing to the statically
indeterminate nature of the problem. It is necessary, therefore, to consider the
elastic deformations of the material such that, in a continuous strain field, the
displacements are compatible with the stress distribution. These relationships are
termed the equations of compatibility. From this point it is only required to have a
relationship between stress and strain, e.g. Hooke’s law, to obtain a complete
solution of the stress components in a body.
The equations of equilibrium and compatibility are quite general and may be
derived in terms of various co-ordinate systems. The mathematical solution of a
problem may often be simplified if an appropriate set of co-ordinates is chosen. With
this in mind, and suitable illustrative applications in the following chapter, the
various equations will be derived in two-dimensional Cartesian and cylindrical co-
ordinates.

13.1 Equilibrium
Consider the equilibrium of a small rectangular element of dimensions 6x,
equations: plane stress dy, 6z, Fig. 13.1. Owing to the variation of stress through the material, 0,,,
Cartesian co-ordinates is a little different from o,,,,, and likewise for the other stresses a, and 7,,.
The variation that must occur over any particular face may be neglected, as
it cancels out when the force equilibrium on opposite pairs of faces is
considered. On this occasion body forces arising from gravity, inertia, etc.,
will be taken into account, and these are shown as X and Yper unit volume.

Yee
Fig. 13.1

Yap

For equilibrium in the x-direction,

(xen — FxspOY 5% + (Tyxp — Tynsy)O* 6% + X 6x dy bz = 0
[13.1]
368 MECHANICS OF ENGINEERING MATERIALS
a

Dividing by 6x dy 6z gives
OtXCD e.
VAB
netgen
SL. IXBC
tie,
JX AD a oxG — ()

bx oy
In the limit, as 6v — 0 and dy — 0 and the element becomes smaller and
smaller, the terms become partial differentials with respect to x and y, and
thus

+X=0 [13.2]

Considering the y-direction,

(ypc a Oy sp )OX 6z+ (F565 a, Ty gy JOY 624- Y 6x by 6z = 0

[13.3]

Dividing by 6x dy 6z and for dx — 0 and dy — 0,

ey SU ANT
hs OTe WA [13.4]
Oy m Ox ‘

It is often the case that the only body force is the weight of the
component and that it can be neglected in comparison with the applied
forces. Then

OF Obes
a Oy ih

0p Ca
Oy Wore,

13.2 Equilibrium
As has been previously stated, there are certain cases such as cylinders,
equations: plane stress discs, curved bars, etc., in which it is rather more convenient to use 7, 6, z
cylindrical co-ordinates co-ordinates. Consider the element ABCD, Fig. 13.2, which is bounded by
radial lines OC and OD, subtending an angle 66 at the origin, and circular
arcs AB and CD at radii r and r+6r respectively. The element is of
thickness 6z.
In the preceding section and Fig. 13.1, the stress variation on the
element was represented symbolically by the different letter subscripts in
order that the first analysis could be written down simply. However, it is
quite usual to show stress variation in terms of partial derivatives. The body
forces are shown as R radially and © tangentially per unit volume.
Considering equilibrium along the radial centre-line of the element,
there will be, in addition to forces from the radial stresses, the resolved
components of force from the circumferential stress, og, and the shear
 VARIATION OF STRESS AND STRAIN 369

Fig. 13.2 Se96, or

Ot
Te + ae ér

06

stresses 7,g and T,. Hence

(«.
oF - or)(r + 6r)60 6z — 0,r 60 bz
r

Oo, _ 66 . 60
-- (<r
+ Fy86 brbsin — 09 br dz sin =

60
+(m + Fi) or 6z cos — Toy br 6508 + Rr 60 6r6z = 0

[13.7]
As 60 — 6, sin 5 60 — 560 and cos 560 — 1. Also, neglecting second- and
higher-order terms and dividing by r 6r 66 6z, the equation reduces to
Ge, Ole, oy ll ray
r Or r yr OO
or

Oo, 1 O76, On O89

i R10) 13.8
Or r OO Q r 7 bee

Resolving in the tangential direction,

Oo 60 60
(<r
+ 20 2)6r 6z COs eee J9 Or 62 COS zy

OT6, : 60 : 60
ate(mae 6 00)6r 6zsin oF+ 7, dr 6z sin oF

=[ @ + oeir)
(r + 6r)60 bz — T,pr 60 bz
r

+Or
dr 606z = 0 [13.9]
370 MECHANICS OF ENGINEERING MATERIALS

In the limit, as 69 — 0, neglecting the appropriate terms and dividing by

r Or 60 bz,

PO
1 009
ent eee
Oro
a TorOe ey 13.10
[13.10]

Axial symmetry In certain cases, such as a ring, disc or cylinder, the body is symmetrical
about a central axis, z, through O, Fig. 13.2. Then by symmetry the stress
components depend on r only, and og at any particular radius is constant.
Also, from the consideration of symmetry, the shear-stress components 7%,
must vanish. Equation [13.10] no longer exists and the equilibrium equation
[13.8] becomes

do, , 0, — 09
+R=0 [13.11]
dr r

If the body force can be neglected, then

da, 06,—06
a La
dr r e |

13.3 Strain in terms of

With reference to a set of fixed axes the movement of an elastic body
displacement: Cartesian
consists of displacement and rotation of the body combined with strain in
co-ordinates the material. Consider a continuous strain field and the displacement of
elements OA, of length 6x, and OB, of length dy, referred to the axes Ox and
Oy, Fig. 13.3. The point O moves to O’ having co-ordinates u and v, which
are in general functions of x and y. The rate of change of u with respect to x
will be Ou/Ox. Therefore, since OA is of length 6x, the point A will move to
A’, where the displacement in the x-direction will be u + (Ou/Ox)éx. The
change in length along this axis is thus (Ou/Ox)6x and the strain is, in the
limit as 6x — 0,

= (Ou/Ox)dx = Ou
Es re = [13.13]

Fig. 13.3
 VARIATION OF STRESS AND STRAIN 371

Strain in the y-direction is obtained from a consideration of the

displacement of OB to O/B’. The rate of change of v in the y-direction
with respect to y will be Ov/Oy, and therefore the point B’ will have been
displaced from B in the y-direction an amount v + (Ov/Oy) dy. The strain
occurring will thus be

(Ov/Oy) dy _ Ov
wy [13.14]
éy Oy

The shearing strain in the element AOB will be given by the change from
the original right angle to the new angle A’O’B’. Hence

xy = CCO'A’ + /DO'B’

For small displacements,

CA’ DB’
/CO'A' & and DO'B’ =
O/C O/D
Now, CA’ is the rate of change of v in the x-direction for an amount 6x or
(Ov/Ox)6x. Similarly DB’ is the rate of change of u in the y-direction for a
length dy, giving (Ou/Oy)éy. Therefore

COINS (Ov/Ox) dx fsOv

bx Ox
and

éy Oy

Thus, for small displacements,

Ov Ou
wy =—t+— 13-45
T= By Oy |
In a two-dimensional strain field, the strains in terms of displacements are
therefore

= Ou Ov Ov Ou
a 7 ey ya, fy [13.16]

13.4 Strain in terms of When dealing with the analysis of elements having a circular geometry
displacement: cylindrical (curved bars, discs, etc.) it is often more convenient to consider strain and
co-ordinates displacements in terms of cylindrical co-ordinates as was done for the
equilibrium equations. Consider the element ABCD, Fig. 13.4, subtending
an angle 60, with AB at radius r and CD at r + ér. This element is displaced
to A‘B’C’D’ so that the radial and tangential movements to A are u and v
respectively. The displacement of the point D to D’ in the r-direction will be
u + (Ou/Or)6r, where Ou/Or is the rate of change ofw with respect to r. The
change in length of AD is therefore (Ow/Or)6r, and hence the strain in the
372 MECHANICS OF ENGINEERING MATERIALS

Fig. 13.4

loo
Se|
V+ ov
00 88

radial direction is
(Ou/Or)dr =Ou
SSS SS SS [13.17]
: ér Or
In the tangential direction there are two effects of displacement on strain. As
B moves to B’ there is the change of v with respect to 0, giving (Ov/00)60 as
the increase in length and hence a strain of
(0/80) 50 1 do
r 00 r OO
There is also the tangential strain due to the element moving out to the new
radius, ry + u. This part is
(r+u)60—r60 u
r 60
Therefore, the total tangential strain is

see
or OO Fr
[13.18]
:

The shear strain, ¥,9, is given by the difference between /DAB and /D’A‘B’,
which is

Aa = 2) A Seep Ae
Now, /D’A’F’ is the difference between the tangential displacement of D to
D’, (Ov/Or)6r, which as an angle is 0v/Or, and the rigid-body rotation about
O, DOF’, which is v/r. Therefore
 VARIATION OF STRESS AND STRAIN 373

/B'A'E’ is due to variation of radial displacement wu in the tangential

direction and is therefore Ou/r 00. Hence
Ov
fain SNv 1QOu
per 13719
caer a r 00 Dee
The three strains in terms of displacement are

Ou
ceo
= —_ [13.20}
13.20

1 Ov
Ep 6 ce as
eee [13:21

er [13.22]

Axial symmetry For problems which are symmetrical about a z-axis through O, there will be
no tangential displacement, v, and since u will not vary with @ there will be
no shear strain, 7,9, therefore 7,9 is zero, as stated in Section 13.2. The above
equations then reduce to

Ou. du
ar Sn ea 13723
o Or dr |

ey=-r [13.24]

13.5 Compatibility
Because the three strains of eqn [13.16] are expressed in terms of the two
equations: Cartesian co- displacements, u and v, there must be a relationship between them. This
ordinates may be obtained by differentiating €, twice with respect to y, €, twice with
respect to x, and 7,, with respect to both x and y. Hence

Peg = Ou Poy iO Oy. 10 (00 afOu

Oy? Ox Oy? Ox? — Oy Ax’? Ox Oy Ox dy \Ox Oy
Eliminating uw and v between these three equations provides the relationship

Ge 0s O4,
[13.25]
Oy? Ox® Ox Oy

which is known as a compatibility equation in terms of strains. To obtain a

compatibility relationship in terms of stresses, it 1s only necessary to
substitute for the strains in eqn. [13.25], using the stress—strain relation-
ships.
374 MECHANICS OF ENGINEERING MATERIALS
eee

For the case of plane stress, 7, = 0), the stress-strain relationships are
Ox Voy
oy SSS
tad E

Za Oy VO x
fi E E

py Oe ee 2Ty(1 + v)

Substituting the above in eqn. [13.25],

1h Cpe a 0 Oy. OO, AO ig 113.26}

E dy? E Oy E Ox* E 0x2 EE Ox Oy ;
Considering the equilibrium equations and neglecting the body force, if
it is only the weight of the body, then differentiating eqn. [13.5] with respect
to x and eqn. [13.6] with respect to y and adding we get
2 2 2
mE igi oe [13.27]
On Oy? Ox Oy
Eliminating 7,, between eqns. [13.26] and [13.27],

loo, Oo, 1 Oo, v Oo;

E -Oy* E. Oy hb Ox EB Oe

_ (1+v) (@o, i Poy

= E Or ~ Oy
Simplifying,

Oo; Oa Oo wsOO,
= =0
Ox Ox? Oy" Oy?
or

og

An analysis similar to that above can be used to show that the compatibility
eqn. [13.28] also applies to the case of plane strain.

13.6 Compatibility
As in the case of Cartesian co-ordinates there must be a relationship between
equations: cylindrical co-
the strains in cylindrical co-ordinates. This may be obtained by eliminating
ordinates u and v between the strain—displacement equations [13.20]-[13.22].
Differentiating eqn. [13.21] with respect to r and eqn. [13.22] with
respect to 6, dividing the latter by r and subtracting gives

OE6 1 v6 — Ou Uu ] Oru

Or +r 0 ror r Pr Oo
VARIATION OF STRESS AND STRAIN 375

Multiplying by r’ and substituting for Ou/Or, we have

[13.29]
Or Oe : OO?
Differentiating this equation with respect to r, and eqn. [13.20] with respect
to @ twice, and eliminating wu gives

7 OE4 ce 2 eg Ore O48 at OE, Oe

ar Ope br OR OF
if if po

and simplifying,

Oa = a Va 13.30
vo O,6 Oe 2 O-eEg OE9 OE,

Gent aah Os we esu

This is the compatibility equation in terms of strain. To obtain a similar
relationship for stresses it is necessary to use the stress-strain relationships
and equilibrium equations in cylindrical co-ordinates.
The stress-strain equations for plane stress
Oy VO¢@

a
oy v3;
ee
To 2(1+v)
eat = E 70

Substituting in eqn. [13.30],

2(1 +v) OT,6 2(1 +) OT 6 a l OO; Vy Oa

& 00 ie BhOr er oe EF OR”
r’ Oo 2V Oo, 2r Oop 2rv Oo,
Eye «EOF EB Or for
r 00, Tv O66
[13.31]
E Or E Or
Now the equilibrium equations [13.8] and [13.10] are used to eliminate 7,9.
Multiply through eqn. [13.8] by 7, and differentiate with respect to 7; then

Oo, Oo, O19 Oo, Oa¢

=
ir" @2 "Br00' Or Or
Differentiating eqn. [13.10] with respect to 6,

1 Oa . Ot, 2 Orn _9
r 08? ° Ord0 rr OO
Multiplying each of the above equations by r, adding and simplifying, we get

OT Ota
OTr6
beget aegis = Oo9 2 Oo,
Tar ay
Oo, O06
> lip, tig, 13.32
ee
376 MECHANICS OF ENGINEERING MATERIALS

Substituting for 7,9 in eqn. [13.31] and simplifying,

Oa, O'ag. 1 Gor mes CT a Oo, 1 a9 =;

Or? a Or. | + Orr Or re OF 2 OM
or

ee eo,
We Ses eT =0 :
(Faire ae: =) Slee A Hee

This is the equation of compatibility in terms of stresses. The complete

analysis of stress distribution in a body may now be made using the
equilibrium equations [13.8] and [13.10], the above compatibility equation,
and the boundary conditions appropriate to the applied forces or
displacements.
For cases of axial symmetry, since stress and displacement are
independent of 0, the equation becomes

Oo ee)

Multiplying out eqn. [13.34] gives

OG lee Oe 1 Goo ,
Or? ne Or : Or ne ea [13.35]

Now, from eqn. [13.12],

Oo,
oy =r"Wf +0, [13.36]
Therefore

and

Substituting these expressions for og in eqn. [13.35] and gathering terms

together gives

tae tb =0 [13.37]
which is the general equation for o, in an axially symmetric stress system
with no body force. It can be verified by substitution that one particular
solution of this equation is

es oAoe (13.38)
 VARIATION OF STRESS AND STRAIN 377

and substituting for g, and 0o,/Or in eqn. [13.36] gives

a9 = A+B [13.39]

A and B are constants which are determined from the particular boundary
conditions of the problem.
These expressions for o, and og will be derived and used for specific
examples in the next chapter.

13.7 Equilibrium in
The equations of stress equilibrium can be expressed in terms of the
terms of displacement: components of displacement. Substituting the strain-displacement relations,
plane stress eqn. [13.16], into the plane stress-strain equations, eqn. [3.3b], gives
Cartesian co-ordinates

ae ee
ee a eas Yay

seat e(6, tv peas) Oe Me

pit ELS nd ag 92 Oy "Sg

E E Ou Ov
Pe. a ea (#= *) Meas
Substituting these equations into the plane stress equilibrium equations,
eqns. [13.5] and [13.6], gives the equilibrium equations in terms of
displacements as

Ou Ov Ou
— ] 1 =
Awenkad Nasa hne Cia °
Ov Ou Ov
2— 1 1 = 13.41
Bp Uae \ ”) ae e Poet
Since the equilibrium equations are written in terms of the two displacement
components u and v, the compatibility equation [13.28] is automatically
satisfied and does not need to be considered. A similar approach will be
employed in Chapter 14 to derive the equilibrium equation for a thin
rotating disc in terms of the radial displacement wu in cylindrical co-
ordinates.

13.8 Summary
Many engineering design problems involve complex variations of the stress
and strain fields within the component. However, we still must employ the
three basic tenets of equilibrium of forces, compatibility of strain and
displacements and the stress—strain relationships of elasticity. Now that we
have derived the necessary equations in Cartesian and cylindrical co-
ordinates we can proceed in the next chapter to apply these to some design
examples.
378 MECHANICS OF ENGINEERING MATERIALS

Problems 13.1 Show that the equilibrium equation in the radial direction for a
varying stress field in spherical co-ordinates r, 6, ~ with body force
R is of the following form:

OO, l :
R sin b+ sin w+- (2°,sin w—og sin W—oy sin w
ie 7

Org OT py ;
a Tpgy COSese Ue 20 + — sin »)0)

132 What are the strain-displacement relationships in spherical co-

ordinates for an axi-symmetrical stress field?
Using the equilibrium equation in the previous question,
simplified for an axi-symmetrical stress field without body force,
show that the displacement at radius 7, for a spherical shell, is given
by
du 2du 2u
de> rdr Pr
1373 For a particular problem the strain-displacement equations in
cylindrical co-ordinates are

ga
A
Ou u
ze £9 = a Ez = Yr0 = Yor = Vr = 9
"Or
Show that the compatibility equation in terms of the stresses a, and
Og 1s

er 1 Hyped py Gp —0
Or
What is the problem?
13.4 Derive compatibility equations from the following strain-displace-
ment relationships:
6 _ Ou ceOv Ou _ Ou
Vey = ay Ox Yar = Az Vz > Ds

Ow 1 Ow
z Oke = ge
Oz u r OO
6 25 Commencing with the six strain-displacement relationships in three-
dimensional Cartesian co-ordinates, derive the six compatibility
equations for three-dimensional states of strain.
CHAPTER
Applications of the
Equilibrium and
Strain—Displacement
Relationships

It has been explained in principle in the last chapter how the stress components may
be determined in a body by use of equilibrium, compatibility and the particular
boundary conditions of the problem. In a majority of cases, the solutions are complex
but there are a few problems in beams and axi-symmetrical bodies in which a simpler
analysis is possible using the equilibrium, strain—displacement and stress—strain
relationships. As it is important to understand how to apply these principles, the
present chapter commences with two simple beam-bending situations. These are
followed by important engineering components, namely the thick-walled cylinder
used typically in high-pressure chemical engineering and the rotating disc or rotor
used in steam and gas turbines.

14.1 Boundary
The first and most critical step in any stress analysis is specification of the
conditions
correct boundary conditions. With the development of modern numerical
and computer-aided techniques, the solution of a properly defined analysis
can often be obtained with little or no effort on the part of the engineer. This
is of little value, however, if the wrong problem is solved!
The boundary conditions for any analysis of the stress equilibrium and
compatibility equations will be of two kinds:
1. Specified displacements. Over some part of the object ey sei the
displacement of the material may be known.
2. Specified surface tractions. Over the remainder of the object boundary
the direct stress acting normal to the boundary, and/or the shear stress
acting on the surface, may be known.

Fig. 14.1 Hose I.D.

Rigid bar
Rubber hose
380 MECHANICS OF ENGINEERING MATERIALS

Consider for example a rubber hose which has been forced over a rigid
bar whose outer diameter is larger than the bore of the hose, Fig. 14.1. The
boundary conditions at the inner diameter of the rubber hose are that every
point has been forced radially outward by an amount equal to the difference
in radii. In terms of the radial and circumferential displacements u and v
respectively, this implies

Ati —aF i AO

Note that the pressure exerted on the hose at y = a by forcing it over the
tube must be found from the analysis. The radial displacement and normal
stress cannot both be specified.
At the outer diameter of the hose the resulting displacement is not
known; this must be found from the analysis. What is known, however, is
that on that external surface, no normal or shear stress is acting. This
implies
at .f = 6, C= =O

Note that it is not possible to specify og as a boundary condition in this

problem, since all the surfaces on which gg acts are in the interior of the
hose.

14.2 Shear stress in a

The distribution of transverse shear stress in a beam in terms of the shear
beam
force and the geometry of the cross-section was obtained in Chapter 6, using
simple bending theory. An alternative approach will now be developed.
Consider the beam in Fig. 14.2, which, for simplicity of solution, is
shown simply supported and carrying a uniformly distributed load, m per
unit length. The origin of Cartesian co-ordinates is taken on the neutral axis,
with x positive left to right and y positive downards.
This is treated as a two-dimensional problem with no variation of stress
across the width of the beain, and therefore only two equations of
equilibrium are applicable. Using eqns. [13.5] and [13.6],
OO, Oy
. = By = 0 [14.1]

OO}. OF
oD Ber [ees

ea w/unit length Oy=- 45]

Fig. 14.2
 EQUILIBRIUM AND STRAIN-DISPLACEMENT RELATIONSHIPS 381

Making use of the exact solution for pure bending

OP ul): [14.3]
then eqn. [14.2] is not required, neither is any strain—displacement
relationship. This is because, in the derivation of eqn. [14.3], the geometry
of deformation and the stress-strain relationship were included. Substitut-
ing for 0, in eqn. [14.1] gives

O(My/T) a Oxy rs
0
Ox Oy

Therefore

On =- ==
OM y
¢e Ox I wy
But OM/Ox = Q, the shear force on the section, so that

Oi Sr Oy [14.4]

Integrating gives

ge , == SS
Q7 |ydyd + C = —
Q’
apt [14.5]
1 .

At the top and bottom free surface of the beam the shear stress must be zero;
therefore

d 2
Ty =O at y= 7 from which C= a

and

Q” Od?
meeeTk
PeOe eT
eee] [14.6]
14.6
At the neutral axis y = 0 and the shear stress has its maximum value

Qa?
aot)
a ie, [14.7]
14.7
This agrees with the value obtained in Chapter 6.

14.3 Transverse normal

A further example on the analysis of beams is that of the distribution of
stress in a beam direct stress in the y-direction due to the application of a distributed load ».
For the beam in Fig. 14.2 the equilibrium equation which is applicable is
eqn [14.2]:
O85. 7 Olyea
a 0
OV OX
382 MECHANICS OF ENGINEERING MATERIALS

Now, the shear stress, 7,,, was determined in eqn. [14.6], and substituting
that value in eqn. [14.2] gives

doy, 9 (_OF, ie
iia di pelea
But from eqn [6.1] 0Q /Ox = —m. Therefore

= (0 14.8
Oy 21 8] ee

or
2 2
wy
=
o (2 7)dao C

7 tee 49)
3 2
w(y
= ——(—-— ay eg 14.9

Using the boundary condition that at the upper surface y = —t4d, the
compressive stress is 0, = —w/b, where @ is the beam width; then

Cee ic» a iha

ibe» gia ARS 16
Substituting

ie
b 127
= wd?
247
Therefore
3 I2
ad ieeg ey” 3
= 14.10
ee (% go =) at
The distribution of stress is illustrated in Fig. 14.2.
A check on this solution may be made by considering the condition at the
lower free surface. Here y = +14, from which a, = 0, which is correct.

14.4 Stress distribution

This problem is of considerable practical importance in pressure vessels and
in a pressurized thick- gun barrels. It is an application of the cylindrical co-ordinate system, r, 0, z.
walled cylinder A long hollow cylinder which is subjected to uniformly distributed
internal and external pressure is shown in Fig. 14.3(a) and (4). The two
methods of maintaining the pressure inside the cylinder are either by end
caps which are attached to the cylinder as shown in Fig. 14.3(a) or by
pistons in each end of the cylinder, Fig. 14.3(4). Considering a cross-
sectional slice XX as shown in Fig. 14.4, the deformations produced are
symmetrical about the longitudinal axis of the cylinder, and the small
element of material in the wall supports the stress system shown. This is the
 EQUILIBRIUM AND STRAIN-DISPLACEMENT RELATIONSHIPS 383

SEERARaRzED tity |
Po

o,+ Ont

(b)

Fig. 143 same as in Fig. 13.2 for the general stress system except that for axial
symmetry 7,9 = 0 and og is constant at any particular radius. Hence og and
0, are principal stresses and additionally are quite independent of the
method of end closure of the cylinder. Considering axial stress 0, and axial
strain €, then both of these occur in the case of end cap closures (Fig.
14.3(a)). The axial stress, o,, is constant and the axial strain, €,, is dw/dz.
For closure by pistons (Fig. 14.3(4)) it is evident that 0, = 0 and ¢,
occurs only due to the Poisson’s ratio effect of o, and og. From the
symmetry of the system and for a long cylinder, we come to the conclusion
that plane cross-sections remain plane when subjected to pressure and
therefore axial deformation, m, across the section is independent of r and
dw/dr= 0.

Fig. 14.4

Cylinder with end caps The equations of equilibrium for an element of material are
do, 0,—06
=() (eqn. [13.12]) [14.11]
dr r
and since 0, is constant

cla [14.12]
dz
384 MECHANICS OF ENGINEERING MATERIALS
ee annem

The strain—displacement equations are

pes. O eae(ectal 13.23)) [14.13]

dr

spa | (eqn. [19.24 [14.14]

‘f

dw
ore
= — [14.15]
14.15

The stress-strain relationships are

Ce ie du
tt
— ae Tenet Ge) —eee [14.16]
14.16

09 Vv Uu
wh denne (a, + 0;) =-
=a 14.17 ]

Oe a dw
.=—-=(4,
ea FAG + 0) aag [14.18 ]

Substituting for du/dr and u/r from eqns. [14.16] and [14.17] and
simplifying,
l+v dag do, do,
; (0, — 96) = a as oe [14.19]

Now, since €, = constant, de,/dr = 0 and differentiating eqn. [14.18] gives

do, do, dog
ap = ( ay +7) [14.20]

Substituting into eqn. [14.19] for do,/dr from eqn. [14.20] and (0, — a@)/r
from eqn. [14.11] and simplifying gives
dog do,
(t=A) (S24 Se)= 0 [14.21]
From eqns. [14.21] and [14.20] we see that do, /dr = 0 and therefore a, is
constant through the wall thickness. Integrating eqn. [14.21] shows that
(o9 + o,) = constant = 2A [14.22]
Eliminating og between eqns. [14.22] and [14.11] gives
do, ‘a20, —2A =
dr r
0 [14.23]
from which, multiplying by 7’,

DAr =26,
do,
Sr" g: = ()
dr
 EQUILIBRIUM AND STRAIN-DISPLACEMENT RELATIONSHIPS 385

and
d
2Ar ——(r°o,) = 0
By integration,

Ar = ro, =B

Hence

G=A-— [14.24]

and from egn. [14.22],

06 =A+5 [14.25]

where 4 and B are constants which may be found using the boundary
conditions. If will be noted that these equations are the same as eqns. [13.38]
and [13.39], which confirms that they satisfy the equilibrium and
compatibility conditions.

Cylinder with pistons In this case, Fig. 14.3(4), o, = 0 and there is a condition of plane stress.
This solution has been included to show that we can arrive at the same
expressions for o, and og by deriving a differential equation for
displacement, u.
Putting 0, = 0 in eqns. [14.16] and [14.17] and solving for o, and o¢ in
terms of uw gives ~

du vu E
= 14.2
4 ($+7) 75 pee

duu E
06 = (-+*) @=0) 14.2
[14.27]

From egn. [14.26],

do, (3 du “*) E
[14.28]
dp gl near r

Substituting eqns. [14.26], [14.27] and [14.28] into eqn. [14.11] and
simplifying gives

du ldu u
di [14.29]
de> rdr r
This differential equation expresses radial equilibrium in terms of the
displacement, u, in the cylinder wall.
The general solution of this equation is
/

u = Cr+— [14.30]
r
386 MECHANICS OF ENGINEERING MATERIALS

Substituting for u/r and du/dr in eqns. [14.26] and [14.27],

64 E
i (ca +) -50 De [14.31]

v= =) (c+ +50-))
Gi
a ee aay
Se
E
14.32
14.32

where C and C” are constants.

These equations may be rewritten with different constants as
B
a= A 2

and
B
09 = A + =
if

which are the same as egns. [14.24] and [14.25].

Boundary conditions The next stage is the determination of the constants 4 and B.

I. Internal and external pressure

The boundary conditions of this problem are: at r = r;, ¢, = —p; (pressure
being negative in sign); and atr=1,, 0, = —pj,
B B
eee and at ge

from which, eliminating A, we get

pe
(bi aaa
— po) ir and A=
bit?ae
— Bot,

Therefore the radial and hoop stresses become

nak ead)
_ Dit; pot (2 Spy ier

cr leaLON i Wee
Leetaie
{mo Po rr
[14.33]
These equations were first derived by Lamé and Clapeyron in 1833.
Let the radius ratio r,/r; = k; then eqn. (14.33) may be written as

ee
1 r
(1 = red (-4fl)|
[14.34]
1 ro
a eet n( ree Aay
)
 EQUILIBRIUM AND STRAIN-DISPLACEMENT RELATIONSHIPS 387

It is important to note that the stresses depend on the & ratio rather than on
the absolute dimensions.

2. Internal pressure only

An important special case of the above is when the external pressure is
atmospheric only and can be neglected in relation to the internal pressure.
Then with p, = 0,

eae) pS) 14.35

a re Di re

be r Pi r
oo = Pe5(14%) eG (1+4) [14.36]

At the inner surface, 0, and og each have their maximum magnitude so that
att — 1.
0, =—p; (radial compressive stress)

It is appropriate at this point to note that the radial stress shown on the
element in Fig. 14.4 in the positive sense, i.e. tension, is in fact in the
opposite sense, 1.e. compression.
The circumferential or hoop stress at r = 7; is

oa nee ;
=,
De gee
Re
At the outer surface, where r = 7,,

2p;
0.r= 0 an d ==

Stress distributions for To complete the basic analysis of the elastically deformed thick-walled
og and o, pressure vessel, the variation of the two principal stresses og and a, is shown
plotted through the wall thickness in Fig. 14.5 for internal pressure and a k
ratio of 3.
| Stress
Fig. 14.5

Internal
pressure
= pj
 MECHANICS OF ENGINEERING MATERIALS

Example 14.1
The cylinder of a hydraulic jack has a bore (internal diameter) of 150mm and is
required to operate up to 13.8 MN/m?. Determine the required wall thickness for a
limiting tensile stress in the material of 41.4 MN/m?.

The given boundary conditions are that at r=75x 1073,

a, = —13.8 x 10° and op = 41.4 x 10°, since the maximum tensile hoop
stress occurs at the inner surface. Therefore

B
=a}
13.8 x 10 6 =
A 5600 x 10-6
ee

and

B
“ x 10 6 = A-p
41.4 + S600 x 10-6

Adding the two equations,

TA= T1610
A = 13.8 x 10°N/m? and B= 154.5kN

At the outside surface, 0, = 0; therefore

B
ee eeeif
os 154.5
eee x 103
r

r’ = 0.0112m’
r = 0.106m

Axial stress and strain Now that expressions have been developed for the radial and circumferential
stresses within the cylinder, the next step is to consider what conditions of
stress and strain can exist axially along the cylinder. These will depend on
the boundary conditions at the ends of the cylinder.

I. Cylinder with end caps but free to change in length

In this case there must be equilibrium between the force exerted on the end
cover by the internal pressure and the force of the axial stress integrated
across the wall of the vessel. Therefore, from Fig. 14.3(a)

o.(tr — mr) — parr =0

so that

PS [14.37]
 EQUILIBRIUM AND STRAIN-DISPLACEMENT RELATIONSHIPS 389

2. Pressure retained by piston in each end of cylinder

Since there is no connection between the piston and the cylinder, the axial
force due to pressure is reacted entirely by the pistons, and therefore there
can be no axial stress in the wall of the cylinder. Thus

Gua) [14.38]

and

= (0,+ eee 09)

2p;

e E(k
—1)
a r SS =

3. Cylinder built-in between rigid end supports

For this case €¢, = 0; in other words, plane strain exists.
Therefore

= oe) =)
0, = v(o, + 06) [14.39]
Substituting for ¢, and og from eqns. [14.35] and [14.36],

ae [14.40]
= 2upir? _ 2);

0 1

Maximum shear stress Since the radial and circumferential stresses are principal stresses the
in the cylinder maximum shear stress in the plane of the cross-section is given by
09g
— Or
Tmax ar 2

sswe ee [14.41]
This applies for all three end conditions because in each case o, < a, < 9.

Yielding in the cylinder Yielding will commence at the inner surface and using the Tresca criterion
we have

Oy =09-0, at P=;
Using eqn. [14.41],

= 2k pj
ae
Hence the internal pressure to cause yielding is

pi = Lene [14.42]

This applies for each of the end conditions considered previously.

390 MECHANICS OF ENGINEERING MATERIALS

If the von Mises criterion is employed, using eqn. [12.15] we have

(o9—a,) +(o,— gy ld oa) ey

Substituting the expressions for the three principal stresses and simplifying
gives the internal pressure to cause initial yielding at the bore.
For the three cases considered previously this gives

1. Cylinder with end caps but free to change in length

Pi aie:
al 24/3 14.434
" ]

2. Pressure retained by piston in each end of cylinder

j= aaey. [14.433]
J (3k + 1)

3. Cylinder built-in between rigid end supports

(5 :
poe J (3k —
ees
4v + 407 + 1)
(14.434

14.5 Methods of
From Fig. 14.5 it will be observed that there is a marked variation in the
containing high prsessure
stress in the wall of a thick cylinder subjected to internal pressure, and this
situation gets worse when designing for even higher pressures. In order to
secure a more uniform stress distribution, one method is to build up the
cylinder by ‘shrinking’ one tube on the outside of another (Fig. 14.6(a)).
The inner tube is subjected to hoop compression by the shrink fit of the
external tube, which will therefore be subjected to pressure causing hoop
tension. When the compound tube is subjected to working pressure, the
resultant stresses are the algebraic sum of that due to the shrinking and that
due to the internal pressure. The resultant tensile stress at the inner surface
of the inner tube is not so large as if the cylinder were composed of one thick
tube. The final tensile stress at the inner surface of the outer tube is larger
than if the cylinder consisted of one thick tube. Thus a more even stress
distribution is obtained.
Another technique used in special industrial pressure vessels is to wind
around the outside of a tube a high-tensile-strength ribbon of a rectangular
section with sufficient tension to bring the tube into a state of hoop
compression (Fig. 14.6(4)). Subsequent internal pressure then has to
overcome the hoop compression before tensile stress can be set up in the
tube.
A further method for creating hoop compression at the bore of a cylinder
is known as autofrettage. This consists in applying internal pressure to a
single cylinder until yielding and a prescribed amount of plastic deformation
occurs at the bore (Fig. 14.6(c)). Since strain increases along the radius of
the cylinder, on the release of pressure, the elastic recovery of the material
causes the bore of the cylinder to be subjected to compressive hoop stress.
 EQUILIBRIUM AND STRAIN-DISPLACEMENT RELATIONSHIPS 391

Fig. 14.6

Elastic zone

Plastic zone
(a) Compound cylinder (b) Wire-wound cylinder (c) Autofrettaged cylinder

At the same time tensile hoop stress occurs in the outer material. This
technique is discussed further in Chapter 15, in the section on residual
stresses.

14.6 Stresses set up bya

A shrink fit between two components is a very important and secure method
shrink-fit assembly
of assembly. It consists, in the case of two cylindrical objects, of the inner
diameter of the outer cylinder being slightly less (by a fraction of a
millimetre) than the outer diameter of the inner cylinder. Consequently,
when they are at the same temperature the outer cannot be passed over the
inner. However, if the outer cylinder is heated and the inner cylinder is
cooled then the thermal expansion and contraction can be made sufficient to
allow one cylinder to pass over the other. On returning each to room
temperature there is ‘interference’ at the mating surface since they cannot
regain their original dimensions at the interface. The two components are
locked firmly together and a system of radial and circumferential stresses are
set up at the interface and through the wall of each cylinder. For elastic
conditions the principle of superposition can be used to add the stresses due
to shrink-fit interference to those due to internal pressure or rotation.
For two cylindrical components we require eqns. [14.24] and [14.25] for
each component given by
B
o9=A+ 5, 0,=A- > (inner component)
r r

D D
og=C+35, Cp Oe, (outer component) [14.44]
r r
where the constants A, B, C and D are determined from the boundary
conditions. For shrink-fit stresses only, the boundary conditions are that o,
is zero at the inside of the inner cylinder and outside of the outer cylinder,
and at the mating surface r,, the radial stress in each vessel must be the same;
therefore

[14.45]
Finally, at the mating surface the radial interference 6 is the sum of the
displacement of the inner cylinder inwards, —u’, and the outer cylinder
outwards, +w”; thus
§ = —u! + u" =4_(e% — €h) [14.46]
392 MECHANICS OF ENGINEERING MATERIALS

We next substitute into eqn. [14.46] the expressions for ¢/, and €%,

pole.
ae
/

oe EE a
and

TCo
e
/ I

EV! Eur

and thence the relationships (eqn. [14.44]) for 04, 04, 0) and of at r = fn.
We now have sufficient equations to solve for the constants A, B, C
and D.
The following example illustrates the analytical process.

Example 14.2
A bronze bush of 25 mm wall thickness is te be shrunk onto a steel shaft 100 mm in
diameter. If an interface pressure of 69MN/m? is required, determine the
interference between bush and shaft. Steel: E = 207 GN/m”, 1 = 0.28; bronze: E
= 100GN/m’, 1 = 0.29.
Using constants A and B for the shaft and C and D for the bush, then the
radial stress for the shaft is
B
Cv v2

At the centre of the shaft r = 0 and this might imply that o,. was infinite,
but this cannot be so and therefore B must be zero; hence o,, = A = ag, at
all points in the shaft. The boundary conditions are

At the interface 0, = —69 x 10° = A

D 6
At rz = 50, n= US nagg = 00 % Ae

At r, = 75, D oe
eat ;ppecrede 0
st oT = Gk
From which D = 312 x 10° and C =55.5 x 10°. So at f_ = 50 mm;
0,, = —69 MN/m’, oo, =180 MN/m’.
Now, the interference is

j= = 35 Ui = 1m (Eo, = E6,)

where r,, = 50. Substituting the values for 09,, 0,,, 06,, O,5

6 = 0.05{[180 -- 0.29(—69)|/100 — [—69 — 0.28(—69)] /207}

= 0.112 mm

which is the interference required at the nominal interface radius of 50 mm

between the shaft and the bush.

14.7 Stress distribution

As was explained earlier in this chapter the containment of high internal
in a pressurized compound pressure in, for example, chemical processes can be achieved more
cylinder effectively by shrinking two or more cylinders, one over the other, to give
 EQUILIBRIUM AND STRAIN-DISPLACEMENT RELATIONSHIPS 393

a compound or multi-tube vessel. The analysis simply uses the basic thick
cylinder equations for a, and og together with the shrink fit and other
boundary conditions.
The method and stress distribution is illustrated by the following worked
example.

Example 14.3
A vessel is to be used for internal pressures up to 207 MN/m2. It consists of two
hollow steel cylinders which are shrunk one on the other. The inner tube has an
internal diameter of 200 mm and a nominal external diameter of 300 mm, while the
outer tube is 300mm nominal and 400mm for the inner and outer diameters
respectively. The interference at the mating surface of the two cylinders is 0.1 mm.
Determine the radial and circumferential stresses at the bores and outside surfaces.
The axial stress in the cylinders is to be neglected. E — 207 GN/m’. Compare the
stress distributions with that for a single steel cylinder, having the same overall
dimensions, subjected to the same internal pressure.

The boundary conditions are: inner tube, r = 100mm, 0, = —207 MN/m’;

outer tube, r= 200mm, o, = 0; at the mating surface or interface,
ry= 150mm nominally, and the radial stresses in the inner and outer
tubes are equal, 0,; = 0,). Also the radial displacement, u;, of the inner
cylinder inwards plus the radial displacement, u,, of the outer cylinder
outwards due to the shrink fit must equal the interference value. Therefore
—u; +u, = 0.1mm

Using the above conditions and constants A, B and C, D for the inner
and outer tubes respectively, we have four equations:
B
=) 07 x i
10 001 [14.47 ]

D
WO 14.48
0.0

B D
A =C 14.49
0.0225 0.0225 |
and
U; be Uy 0.0001
O15 005 2 7075
or
cm 0.0001
pices alias
and substituting for the strains in terms of the stresses,
0.0001
=09; + V0,, + 09, —Vo,, = X 207 x 10°

and since 0,, = 0,,,

06, — 96% = 138 x 10°

394 MECHANICS OF ENGINEERING MATERIALS
eT

Table 14.1 =a 3
Radius (mm) o, (MN/m’) oy (MN/m’) :

ie oa iy 4 \inner cylinder

fe ee foe \outer cylinder

Therefore
D
* ——
0.0225 (4+ P38
aa) é 10° 14.50
aoe
The next step is to solve for the constants using eqns. [14.47] to [14.50]
and the following values are obtained:

A =28MN/m’, B = 2.35 MN

C = 98.2 MN/m’, D = 3.93 MN

The required stresses are then computed from the basic equations with the
values of the constants above, Table 14.1
If the cylinder had been made of one thick tube of the same overall
dimensions as the compound vessel then at the bore r =100mm,
o, = —207 MN/m*
B
207 10° =A = 14.51
he 0.01 eee
and at r = 200mn, o, = 0

ae B
eae 14.52
0.04 |
Hence A = 69 MN/m’ and B =2.76MN from which og and o, are as
follows:

Radius (mm) o, (MN/m’) og (MN/m?)

100 —~207 +345

150 545 +191.5
200 0 +138

The values of a7 may be compared from the compound cylinder and the
monobloc cylinder.
Figure 14.7 shows diagrammatically the distribution of radial and hoop
stresses through the wall of the compound and single cylinders, and
illustrates the more efficient use of material in the former case.
The stress distribution due to shrinkage only may be obtained directly as
the difference between the curves for the single and compound cylinders as
plotted above. This approach would only apply if the compound tube was
made of the one type of material throughout.
 EQUILIBRIUM AND STRAIN-DISPLACEMENT RELATIONSHIPS 395

Fig. 14.7
Bore Interface Outside
Compound cylinder ; 4 A

single cylinder — — —;
shrinkage stress, — :—

Tensile stress
(MN/m?)

200
—— Radius

Compressive
stress

Shrinkage stresses alone could have been obtained by calculation from

the four boundary conditions: a, = 0 at r = 100 and 200, and o,, = o,, and
—u; + u, = 0.1 at the interface radius r = 150, as described earlier.
If the two tubes are of different materials then the appropriate elastic
constants have to be used where they occur in the various equations.

14.8 Spreadsheet
The general case for the stress in a compound cylinder of two different
solution for stress in a materials subject to internal and external pressure will now be solved using a
compound cylinder spreadsheet. The modulus and Poisson’s ratio of the inner and outer
cylinder will be taken as E’ and v’, E” and v” respectively. The necessary
equations to solve for the four arbitrary constants are:
iV At the inner radius of the inner cylinder a known pressure is applied Dee

Boe)

o =A-—=-», [14.53]

At the outer radius of the outer cylinder a known pressure is applied,

Woe a We

At the interface, there is continuity of direct stress, i.e. at r =

D
a By EN RES
rz
oh v2
[14.55]
396 MECHANICS OF ENGINEERING MATERIALS

4. At the interface, there is compatibility of deformation, i.e. at r = 5

6 i oe
Be Abe ah
=i ch tewnel
aL Eigt Es be ae
(ce een wie as
el: Pi Opa ep
[14.56]
These equations can be written in matrix form as

1 —1/a? 0) 0)
0 0 1 —1/¢7
1 1/3 —1 1/b?
=(1=7)/E “G4 7)\/FE C=) G47 )//FE

A —Pa

| 7 nace [14.57]
D 6/b

This matrix equation can be solved using a spreadsheet by inverting the

coefficient matrix and then multiplying the resulting inverse by the right-
hand side (see Appendix B). A ‘macro’ (a series of menu commands in
Quattro Pro) to perform this operation is shown in Fig. 14.8 starting at cell
G17. Once the unknowns 4, B, C, D have been found the pressure at the
interface can be found from the radial stress in either component at that
radius, e.g.

o =A-— [14.58]

Figure 14.8 shows the results for the data of Example 14.3. The only inputs
(shown shaded) are the inner, interface and outer radii, the material
properties, the amount of interference and the internal and external
pressures. Once this information is entered and the macro to perform the
matrix solution is executed, the radial and circumferential stresses are
calculated automatically. The results corresponding to those in Table 14.1
are shown in cells Al9..C22. Any consistent set of units can be used. If
dimensions are given in m, then modulus, stress and pressure values must
be given in N/m’.
This spreadsheet can also be used to check the interface pressure
generated by the interference calculated in Example 14.1, if a very small
value (say 1.0E—6) is used for the inner radius to approximate a solid inner
cylinder.
 EQUILIBRIUM AND STRAIN-DISPLACEMENT RELATIONSHIPS 397

uh v" | _ Interference
ae
LORS ete 020008
|
4

5
6 O° 0, _o
7 1 -1/B2A2 - 4
8 -(1£2)/D2_ ——__{14E2)/(B292*D2)_(1+G2)/F2 (1+G2) / (B2*2*F2)
9 |

10 ae eee Sl DAD
44 -0,3333333333333 _1.3333333333333 | _1.1233333333333_ 60375000000 207000000 Pa = _ 28750000
12 _-0.0133333333333_0.01333333333333 | 0.01123333333333__ 603750000 0) Pc ____ 2357500
13 -0.3333333333333_1.3333333333333_0.48333333333333 43125000000 ol 97750000
14 -0.0133333333333.0.01333333333333 |0.01933333333333__ 1725000000 - +H2/B2_ 3910000
45 A Bel | a
16 | Macro to solve matrix equation
ange 4 ee
{Invert.Source a5 ..d8} teeth
18 ; ius' s Sigma_r Sigma_t | | {Invert. Destination aii ..d14}
19 +A2 = $HALH12/A1942 |+H114H12/A1902 Inner Cylinder. i * | {Invert.Go} i
20 +82 —————HI1H12/A2082 |+H114H12/A2002 _ {Multiply.matrix_4 a11 .d14}
21+B2 — — +H13H14/A21A2 | +H13+H14/A21A2 Outer Cylinder |— | {Multiply.matrix,2 14 ..d14}
29 +C2. +H13H14/A2202 |+H13+H14/A2242 | {Multiply.Destination h11 ..h14}
N ies) {Multiply.Go} ee

(a) Data and cell formulae

Eee
cA6 SSE Interference
2.07E+14 | 2.07E+11 |
Matrix toinvert) _
z 23
0)
O° epee
1.00, _—_-44.44 _ 1.00
-3.48E-12; —_ -2.75E-10 3.4812, "2.75E-10° iv 3 :

Inverse AD
6.04E+10 _72,07E+08 Pa _ 28750000
5©CM®YIMR®HMAwWNHE
DE 6.04E+08 | 0,00E+00 |-Pe i 2357500
4.31E£+10 0 | _ 97750000
6.67E-04 __ 3910000

| Macro to solve matrix equation

PRR
PRR
®NROHBRW
Radius, ‘Sigma_r =. | {Invert.Source as .d
0.1 | -2.07E+08 Inner Cylinder. iy - | {Invert.Destination at .d14}
roma)
NE 0.15) -7.60E+07 | 1.34E+08 | | {Invert.Go} ine ;
0.15 -7.60E+07 2.72E+08 | Outer Cylinder. {Multiply.matrix41a11 .d14}
(OD -0.00 | 1.96E+08 {Multiply.matrix_2 f11 ..d14}
ifMultiply. Destination h11 ..h14}
BWNHR
NNNN {Multiply.Go}

(b) Spreadsheet display

Fig. 14.8

14.9 Stress distribution

A simplified model of a component such as a gas turbine rotor is a uniformly
in a thin rotating disc
thin disc which, when rotating at a constant velocity, is subjected to stresses
induced by centripetal acceleration. This is a problem which produces
deformations symmetrical about the rotating axis. If the disc is thin in
section then it is assumed that plane stress exists, so the radial and hoop
stresses are constant through the thickness, and there is no stress in the z-
direction.
398 MECHANICS OF ENGINEERING MATERIALS

Fig. 14.9

The equation of equilibrium of an element, Fig. 14.9, is that derived in

Chapter 13 for the axially symmetric stress system, but in this case a body
force term must be included which is determined from the centripetal
acceleration. Hence, from egn. [13.11],
do, 0,—09
+R=0
dr r
where R is the body force per unit volume.
In order to analyse the rotating disc as a static equilibrium problem,
D’Alembert’s principle is applied whereby the inward force on the element
due to the centripetal acceleration is replaced with an outward centrifugal
force given by

F = (pr 60 6r z)u’r = Rr 60 6r z
where p is the density, w is the steady rotational velocity in radians per
second and z is the thickness of the disc. Therefore

eS purr

and
do, o,-o
dr f
— + purr =0 [14.59]
The strain—displacement equations for axial symmetry are
y du u
r= Saar = =
dr ae
and the stress-strain relationships are
x Oy VO9 06 VO,
eeDe aye
Using these four equations to obtain a, and ag in terms of u,

0, = sceLr
|—+—
dr A iT—
14.60

ae pias E
g= aaa hems [14.61]
 EQUILIBRIUM AND STRAIN-DISPLACEMENT RELATIONSHIPS 399

Substituting in eqn. [14.59] we obtain

d'u 1 du ae lw a,
dr? rdr r E ge Gas

or

vu ildu u Peale 2
ws = 14.62
dr? “ rdr fr (E )pw ‘i | |

This is a linear differential equation of the second order. The general

solution consists of the sum of two separate solutions known as the
complementary function and the particular integral. The former is the
solution of the left-hand side and the latter is obtained by considering the
right-hand side of eqn. [14.62]. Thus the complementary function is
C/
Cr ;

and the particular integral is

197) parr’
z=
E 8
The complete solution is therefore

G 1—-v\ por
= = Cre ; (E ) 3 14.
[14.63]

in which C and C’ are constants to be determined from the boundary

conditions.
Using eqn. [14.63] and substituting for w and (duw/dr) in eqns. [14.60]
and [14.61] and simplifying the various constant terms by inserting new
ones, A and B, we obtain

ee B (3 to)par [14.64]
r 8

Ber ((tes
09 =At—- zs “)pur? [14.65]
r 8

The constants A and B are found from the appropriate boundary conditions
of the problem.

Solid disc with unloaded If the disc is continuous from the centre to some outer radius r = r, then it
boundary is apparent that, unless B = 0, the stresses would become infinite at r = 0.
To, find A it is only necessary to use the condition that
o, = at r=n

from which \

3
Ae = purr,
400 MECHANICS OF ENGINEERING MATERIALS

and

C= : ~ puP(? a) [14.66]

3 P+ 3
his _ ry = = pub?

pur (34+ v)% — (14 3v)r]

ane [14.67]

Fig. 14.10

(b)

The distributions of radial and hoop stresses are shown in Fig. 14.10(a).
The maximum stress occurs at the centre, where r = 0, and then
34+7
Or = 096 = purrs

Disc with a central hole _ In this case the boundary conditions are that the radial stress will be zero at
and unloaded +7 =7 and r = 1, the radius of the hole and outside periphery of the disc
boundaries _ respectively. Therefore

B 3-7
mere 3 purr =0

and also

B 3+07
ee ae 8 tereil

Solving these two equations for the constants gives

3
figs — pri and = : .z pur(r, +r)

Therefore

a+ rr
0, = —— pur (2+1 - 1 - ?) [14.68]
8 r

and

3+ in 14a
09 = sella, (7+ a ae a 7) [14.69]
 EQUILIBRIUM AND STRAIN-DISPLACEMENT RELATIONSHIPS 401

The maximum value of the hoop stress, a9, is at r = 1), and is given by

34+ 7 l-v
Tomax = —y— pur (2* wal [14.70]
o, is a maximum when do,/dr = 0 or r = \/(rirz); therefore
3
a p(n — r)° [14.71]

The stress distributions of og and o, for the disc with the hole are shown in
Fig. 14.10(6).

Disc with a loaded In general, rotor discs will have mounted on the outer boundary a large
boundary number of blades. These will themselves each have a centrifugal force
component which will have to be reacted at the periphery of the disc. Given
the mass of each blade, its effective centre of mass and the number of blades
we can compute the force due to each blade at a particular value of w.
Multiplying by the number of blades gives the total force which may then be
computed as a uniformly distributed load. Dividing this by the thickness of
the outer boundary gives the required value of a, to use as the boundary
condition when evaluating 4 and B.

Disc shrunk onto a shaft The concept of a shrink fit between two components was developed earlier
for the compound cylinder. The same principle is very valuable for locating
a rotating disc onto a shaft and avoiding mechanical attachments, e.g.
bolting, riveting. The stresses set up by the shrink-fit mechanism are quite
independent when the disc is stationary. However, since at the mating
surface the radial stress is compressive, when rotation commences there is a
superposition of radial tensile stress. It is therefore necessary to ensure that
the shrink-fit stress is always in excess of the rotational stress at the mating
surfaces so that the disc does not become ‘free’ on the shaft.

14.10 Rotational speed

The design of rotating discs must take account of the limit of rotational
for initial yielding
speed which would induce initial yielding at some point in the disc.

Solid disc The maximum stress occurs at r = 0) and is

Sere te
peop purr

Using the maximum shear-stress criterion (Tresca)

Oy = 09 (since a, = 0)

Therefore

ay = ey
St) ate
niall Sa (14.72]
402 MECHANICS OF ENGINEERING MATERIALS

Using the shear-strain energy criterion (von Mises)

ay = 94+ 0,— 000,

= oj (since 09 = 0,;)

Oy 165

and
a oo [14.73]
Yn V B+vp
which is the same value as for the Tresca criterion.

Disc with central hole The maximum hoop stress occurs at r = r; and is
ae
= pt (dB)
4

For the Tresca criterion

Sea (1 —v)
Jy =09= 4 pt (d+ 0A)

@§ 4ay oe

= (saa) ge
For the von Mises criterion since 0, = 0, = 0 at r= 17, yield occurs when
ay = 09; hence the rotational speed at yield is the same as given by egn.
[14.74].

Example 14.4
A steel ring has been shrunk onto the outside of a solid steel disc and shaft. The
interface radius is 250mm and the outer radius of the assembly is 356 mm. If the
pressure between the ring and the disc is not to fall below 34.5 MN/m2, and the
circumferential stress at the inside of the ring must not exceed 207 MN/m?,
determine the maximum speed at which the assembly can be rotated. What is then
the stress at the centre of the disc? » = 7.75 Mg/m*, vy = 0.28.

For the ring at 7 =356mm, o, = 0, and at r = 250 mm, o; = —34.5 x 10°:

therefore

B 3 + 0.28
0=A 0.126 ( 3 See715
De 0)
«0.126 :
x 10°)

and

B
345510 == =
Pas 0.0625
3 By
m cus x 7.75u~ x 0.062.5 Xx 10°)
 EQUILIBRIUM AND STRAIN-DISPLACEMENT RELATIONSHIPS 403

from which

B = (4280 + 0.025u")10° = and =A = (34.000 + 0.67)

10°
Also when r = 250mm, og must not exceed 207 MN/m’; therefore
B
O10 —=4
. * 0.0625
(:+ (3 x 0.28)
eof eAVises 0.0625)
8
Substituting for 4 and B,

207 x 10° = 34000 + 0.6w* + 68 500 + 0.4u* — 0.111w”

From which

w = 343 rad/s = 3280 rev/min

For the solid disc, using constants C and D, as shown previously, D must
be zero; therefore

3
,=C— ( +) pat?
8
At r = 250, o, = —34.5 x 10°; therefore
3-028
—34.5 x 10°=C— ( x 7.75 x 10° x 117500 x 0.0625

and

C= (—34.5 x 10°) + (23.3 x 10°)

= 11.2 << 10°N/m-

But at the centre of the disc, ry= 0 and 0, = og = C; therefore

Cg MIN
This type of problem also lends itself to the type of spreadsheet solution
illustrated in section 14.8.

14.11 Stresses in a rotor

The thin uniform discs analysed in the previous section, although
of varying thickness with illustrating equilibrium and compatibility concepts, do not represent a
rim loading very realistic design configuration. Because the hoop stress og is highest at
the bore and reduces towards the periphery it is more economical to vary the
thickness in similar proportions. A varying cross-section for a turbine rotor
might be as shown in Fig. 14.11. A simple method of solution was devised
by Donath for steam turbines which consisted of dividing up the cross-
section into a number of constant thickness rings as shown.
The equations [14.64] and [14.65] for o, and og can be applied to each
ring observing the required conditions of equilibrium and compatibility at
404 MECHANICS OF ENGINEERING MATERIALS

0,

Shrink-fit
ie interface at hub

each. Let

Senn Seas :Vere [14.75]

and

D (difference) = 9 — 0, = 2542! : ee [14.76]

Replacing wr by the tangential velocity V at radius r gives

S= BE i V? +) [14.77]

D= o4),(7 4 =) [14.78]

Donath then constructed a chart of a series of S and D curves for various

values of C) and C) and for a particular density and Poisson’s ratio. A typical
Donath chart is illustrated in Fig. 14.12.
If values of S and D are known at any radius then og and o, can be found
since

06 = — and 6, =——— [14.79]

The procedure is as follows having divided the disc into several rings of
constant thickness:
1. Calculate the rim loading due to blades and thence o,, at the rim.
2. Assume a value of 0, at the rim and then calculate S and D at the rim.
3. Using the chart and the tangential velocity at the rim will give starting
positions on the appropriate S and D curves.
4. Proceed along the S and D lines from the rim to the first interface
between rings at the correct tangential velocity for that interface radius.
Use these new values of S and D to calculate o,,.
5. At this interface we must satisfy equilibrium for each side of the
interface; hence

Ces it en) [14.80]

 EQUILIBRIUM AND STRAIN-DISPLACEMENT RELATIONSHIPS 405

Fig. 14.12
300

200

180

160

140

120

100
(MN/m2)
Stress
80

60 a =
e eS ——— =
20 = ee

: SESE
40
‘
ESS:
AN
Tangential velocity (m/s)

where 2 and 22 are the thicknesses of the adjacent rings.

Now Ao; =0,, — 6,,

Mora, (12 <a = oe (1zs=) [14.81]

Oy, eo)

and we must also satisfy compatibility so that circumferential strains

must be equal on each side of the step. Hence, €9, = €9, and
09, — VOrn, = 96, — VO; [14.82]
Aog = vo,

Nor = War. ( _ =) [14.83]

22
Adding and subtracting eqns. [14.81] and [14.83],
Aog + Ao, = AS = (1+ v)Ao, [14.84]

Agog — Ao, = AD = (v — 1)Aa, [14.85]

6. The new values of S and D for the other side of the interface are
calculated using eqns. [14.81], [14.84] and [14.85].
406 MECHANICS OF ENGINEERING MATERIALS

7. Follow these curves on the chart to the next interface velocity V and
calculate a, using the values of S and D. Repeat calculations 5 and 6.
8. Continue for the remaining steps until reaching the centre or the inner
radius. The final values should be o, = og for a solid disc and o, = 0 or
a shrink-fit pressure for a disc with a central hole.
9. The correct values will probably not be obtained on the first run and an
adjustment will then be made to the ag value at the rim and the process
repeated until the conditions at the bore or centre are correct.
10. The distribution of a, and og may now be plotted by calculation from
the values off the S and D curves at the radii for which each ring
thickness equals the disc thickness.
The above analysis may seem rather outdated with the current
availability of computers, but the object here is to demonstrate the
principle of analysis and the Donath chart is merely a graphical aid. The
following simplified worked example will help to clarify the solution.

Example 14.5
A steel rotor disc of 800mm diameter and varying thickness, as shown in Fig.
14.11, rotates at 2860 rev/min. The outer periphery is subjected to radial and
circumferential stresses of 17 and 33 MN/m? respectively. Evaluate the magnitudes
and distributions of radial and circumferential stresses in the disc. The interface
radii are 50, 100, 175 and 275 mm and the ring segments are of width 100, 65, 43
and 30 mm respectively. » — 0.3.

The first step is to calculate the velocities at the periphery and at each ring
interface and these are given in the table below.
At the periphery

S = 33+17=50MN/m?
and

D = 33 —17 = 16MN/m*
which gives the starting points on the chart, Fig. 14.12, at a velocity of
120 m/s. The remainder of the solution is shown in tabulated form. The
final values of o, and og are determined from the values of S and D half-way
between each interface where the ring thickness equals the disc thickness. In
this example further iteration is not required and the negative value of o, is
due to a shrink fit.

%1/29 (1 ~ 21/22) S D 0; Ao, AS AD 09 «OO,

7 74 Liga ses er £ Z es 107-33

 EQUILIBRIUM AND STRAIN-DISPLACEMENT RELATIONSHIPS 407

14.12 Summary
The application of the equilibrium and strain—displacement relationships
has been demonstrated at length in relation to two important engineering
components, namely the thick-walled pressure vessel and the rotating
turbine rotor. Various design problems have been examined such as
containing very high pressure by using a compound vessel. This leads to the
concept of shrink fitting and the associated initial stresses prior to
pressurizing. Numerical constants in the equations for radial and hoop
stresses have to be determined from the boundary conditions which must be
accurately assessed. The axial condition in the cylinder wall depends on the
method of end closure and sealing and this is an important design
consideration.
Stresses set up by rotation are a major design feature of turbine rotors
and compressors. The analysis commenced with the thin uniform disc to
enable a full understanding of the plane stress solution for og and o,
obtained from the equilibrium, compatibility and stress—strain equations,
together with a variety of boundary conditions. It became a fairly
straightforward process then to develop an analysis for the varying-
thickness rotor.
Finally, we must remember that elastic design relies on the application of
yield criteria limits such as those of Tresca and von Mises to enable us to
determine a limiting pressure for a cylinder and limiting speed for a rotor.
However, it will be seen in Chapter 15 that there can be good reasons for
developing limited amounts of plastic deformation in a cylinder or rotor to
induce favourable residual stresses which allow enhanced performance. In
all cases the basic principles of equilibrium of forces, geometry of
deformation and the elastic or plastic stress-strain relationship, together
with the appropriate boundary conditions, must be followed.

Problems 14.1. A beam of depth d and length / is simply-supported at each end and
carries a uniformly distributed load over the whole span. Show that
the maximum vertical direct stress a, is }(d/1 ) times the maximum
bending stress o, at mid-span, and therefore in most cases may be
considered insignificant in relation to the latter.
14.2. Solve Problem 13.3 using the thick-walled cylinder relationships and
compare the difference in axial loads.
14.3 Determine the & ratio for a thick-walled cylinder subjected to an
internal pressure of 80 MN/m* if the circumferential stress is not
to exceed 140 MN/m?. What are the maximum shear stresses at the
inside and outside surfaces?
14.4 In a pressure test on a hydraulic cylinder of 120mm external
diameter and 60mm internal diameter the hoop and longitudinal
strains are measured by means of strain gauges on the outer surface
and found to be 266 x 10~° and 69.6 x 10~° respectively for an
internal pressure of 100 MN/m/?. Determine the actual hoop stress
at the outer surface and compare this result with the calculated
value. Determine also the safety factor for the cylinder according to
the maximum shear stress theory. The properties of the cylinder
material are as follows: o, = 280 MN/m’; E = 208 MN/m’;
Y= 029:
408 MECHANICS OF ENGINEERING MATERIALS
see
iti LE RT LC LALA

14.5 A cylinder of internal radius a and external radius + is sealed at

atmospheric pressure and put into the sea. If the pressure due to the
water is 10 atmospheres, calculate the maximum hoop stress in the
cylinder in units of bars. Assume that the volume of the cylinder
does not change.
14.6 Derive expressions for the radia], circumferential and axial strains at
the inner and outer surfaces of a thick-walled cylinder of radius ratio
k with closed ends subjected to internal pressure p.
14.7 One method of determining Poisson’s ratio for a material is to
subject a cylinder to internal pressure and to measure the axial, €,,
and circumferential, €g, strains on the outer surface. Show that
eg 2€,

. 2&9 — Ez
The axial and circumferential strains on the outer surface of a
closed-ended cylinder of diameter ratio 3 subjected to internal
pressure were found to be 1.02 x 10~* and 4.1 x 10~* respectively.
Calculate the internal pressure and the hoop strain at the bore. It
may be assumed that
_ Oy
+ 09
Ox E = 207 GN/m*
Sea
14.8 Show that the pressure generated at the interface between two
cylinders when they are shrunk together is given by

BONE Ae, Fe
SNe POA ai ote
Pe leek
where 6 is the interference between the outer and inner cylinders
and r is the nominal interface radius. The suffices 1 and 2 refer to
the inner and outer cylinders respectively.
[40 A steel insert of 60 mm ID and 70 mm OD is shrunk with a diametral
interference of 0.1 mm into an aluminium mould for forming plastic.
The mould may be taken as a cylinder with an OD of 100 mm. What
interference pressure will be generated? For steel E = 200 GN/m/?,
v = 0.33; for aluminium E = 70 GN/m?, v = 0.33.

Fig. 14.13 steel

aluminium

14.10 The ram of a hydraulic actuator is 50.00 mm diameter. It is located

in a bearing with an internal diameter of 50.05mm and an outer
diameter of 75mm, Fig. 14.14.
(a) If the ideal conditions for sliding require a clearance on the
diameter of 0.025 mm, what is the optimum compressive load
on the actuator ram? The bearing is unstressed.
 EQUILIBRIUM AND STRAIN-DISPLACEMENT RELATIONSHIPS 409

(b) At what compressive load will the bearing and ram start to
interfere?
(c) What will the interference pressure between ram and bearing be
at a load of 2000 kN?
For both ram and bearing, take E = 200 GN/m’, v = 0.30.

Fig. 14.14

baad Figure 14.15 shows a dowel having an interference fit in a steel plate
which is very large compared to the hole diameter. Both dowel and
plate have the same mechanical properties: E = 200 GN/m’;
vy = 0.3; a, = 400 GN/ m’. What is the maximum interference
between dowel and hole that can be used without causing yielding of
either component? What force would be necessary to pull the dowel
out of the hole, given a coefficient of friction = 0.1? (You can
neglect the effect of any axial force on the interface pressure, i.e.
assume plane stress). What minimum amount of interference would
be necessary to ensure the pull-out force was at least half the
maximum value?
If either component could be heat treated to increase its yield
stress, which should it be?

10 mm 6
Fig. 14.15

20 mm

14a2 A large circular saw, Fig. 14.16 may be considered as a 4mm thick
disc with a central hole bolted to a rigid inner disc rotating at
2500 rpm.
If the disc material is steel (E = 200 GN/m’, v = 0.3,
p = 7800 kg/m*) estimate the shearing force that has to be carried
by each one of the 24 securing bolts.
14.13 Figure 14.17 shows a rigid metal fitting attached to the end of a
thick-walled rubber hydraulic hose. The seal is formed due to
compression of the outer wall of the hose, which has a modulus
E = 100 MN/m’ and v = 0.5. The inner radius of the hose and the
mating diameter of the fitting are both 6mm. Assuming that there is
no axial load on the hose, find the interface pressures between the
410 MECHANICS OF ENGINEERING MATERIALS

Fig. 14.16

24 bolts on 0.4 m P.C.D.

inner and outer radii of the hose and the metal fitting. From your
results, estimate the hydraulic pressure at which the fitting would
start to leak.

0.2 mm compressed |
Fig. 14.17

| Rubber hose

14.14 A gun barrel is formed by shrinking a tube of 224mm external

diameter and 168mm internal diameter upon another tube of
126 mm internal diameter. After shrinking, the radial pressure at the
common surface is 13.8 MN/m/. Determine the hoop stresses at the
inner and outer surfaces of each tube. Plot diagrams to show the
variation of the hoop and radial stresses with radius for both tubes.
1 A steel tube has an internal diameter of 25mm and an external
diameter of 50mm. Another tube, of the same steel, is to be shrunk
over the outside of the first so that the shrinkage stresses just
produce a condition of yield at the inner surface of each tube.
Determine the necessary difference in diameters of the mating
surfaces before shrinking and the required external diameter of
the outer tube. Assume that yielding occurs according to the
maximum shear stress criterion and that no axial stresses are set up
due to shrinking. Yield stress in simple tension or
compression = 414 MN/m’, E = 207 GN/m’.
14.16 A steel bush is to be shrunk on to a steel shaft so that the internal
diameter is extended 0.152 mm above its original size. The inside
EQUILIBRIUM AND STRAIN-DISPLACEMENT RELATIONSHIPS 411

and outside diameters of the sleeve are 203mm and 305mm

respectively. Find (i) the normal pressure intensity between the bush
and the shaft, (ii) the hoop stresses at the inner and outer surfaces of
the bush. E = 207 GN/m’, v = 0.28.
14.17 A solid steel shaft of 0.2 m diameter has a bronze bush of 0.3 m outer
diameter shrunk on to it. In order to remove the bush the whole
assembly is raised in temperature uniformly. After a rise of 100°C
the bush can just be moved along the shaft. Neglecting any effect of
temperature in the axial direction, calculate the original interface
pressure between the bush and the shaft. Fy.) = 208 GN/ m;
Vetee! = 0.29; Osteet = 12 x 107° per deg C; Esonze = 112 GN/m?;
Vbronze = 0.33; bronze = 18 x 10xc per deg C.
14.18 Commencing from the relationship derived in Problem 13.2 show
that the radial and circumferential stresses in a thick-walled spherical
shell may be expressed as

oe and aye
r r°
Determine the maximum shear stress at the inner and outer surfaces
of a spherical shell, having a & ratio of 1.5, for an internal pressure of
7 MN/m’.
14.19 A steel shaft of radius 6 mm has been forced into a steel disc of
thickness 1 mm and outer radius 12 mm. There was an interference
of 0.02 mm on the diameters of disc and shaft. Assuming a state of
plane stress and a coefficient of friction of 0.1, what torque would be
required to turn the disc on the shaft? For steel E = 200 GN/m/?,
Vi 033:
14.20 A thin disc of inner and outer radu 150 and 300mm respectively
rotates at 150rad/sec. Determine the maximum radial and hoop
stresses. v = 0.304, p = 7.7 Mg/m’.
14.21 (a) A solid circular disc of uniform thickness is to be used as a
flywheel to store kinetic energy in a city bus. Derive an
expression for the maximum shear stress in the flywheel.
(6) The energy U stored in a rotating disc is given by

U=35 Iu?

wihekew ie 5pmro't, p is the material density, 7 is the disc

radius, ¢ is its thickness and w is the speed of angular rotation.
Assuming that the thickness ofthe disc is fixed, but the radius is
limited only by the strength of the material, which of the
materials below would allow the disc to store most energy?

Material Oy p v
(MN/m?) (kg/m?)

GFRP 1000 1800 0.3

Titanium Alloy 1200 4500 0.3
Mild Steel 220 7800 0.3

14.22 A thin uniform disc with a central hole is pressed on a shaft in such a
manner that when the whole is rotated at ” revolutions per minute
412 MECHANICS OF ENGINEERING MATERIALS

the pressure at the common surface is p. Derive an expression for

the hoop stress in the disc at the periphery, if the inside radius of the
disc is 7; and the outside radius 72.
14.23 A solid steel disc 457 mm in diameter and of small constant thickness
has a steel ring of outer diameter 610mm and the same thickness
shrunk on to it. If the interference pressure is reduced to zero at a
rotational speed of 3000rev/min, calculate the difference in
diameters of the mating surfaces of the disc and ring before
assembly and the interface pressure. vy= 0.29, p = 7.7 Mg/m’,
E = 207 GN/m’.
14.24 An aircraft window consists of a flat circular sheet of perspex 0.3m
in diameter. If the aircraft is pressurized, i.e. assuming sea level
pressure is maintained even at high altitude, what thickness of
perspex would be required to prevent yielding? Justify your assumed
loading and boundary conditions. For perspex, E = 4 GN/m’,
y = 0.5, o, = 60 MN/m’.
14.25 A steel rotor disc of uniform thickness 50 mm has an outer rim of
diameter 750 mm and a central hole of diameter 150mm. There are
200 blades each of weight 0.22 kg at an effective radius of 430mm
pitched evenly around the periphery. Determine the rotational speed
at which yielding first occurs according to the maximum shear stress
criterion. Yield stress in simple tension for the steel is 700 MN/m?,
vy = 0.29, p = 7.3 Mg/m?*, E = 207 GN/m’.
14.26 A disc is to be designed having uniform strength,that is, the radial
and hoop stresses are the same at any point in the disc. Show that
the required profile of thickness variation is given by
z= ze ee)
pur? /2o
CHAPTER l5
Elementary Plasticity

Engineering design is primarily concerned with maintaining machines and structures

working within their elastic range. The analyses of Chapter 12 were specifically
related to the assessment of the yield boundaries for components and the influence
of stress concentration in possibly causing material to exceed the elastic range
locally. However, it would be imprudent if designers knew nothing of what would
happen to components that were, say, grossly overloaded to the point where marked
yielding and plastic deformation occurred. Another important aspect is that in some
circumstances enhanced performance can be achieved by prior plastic deformation
resulting in favourable residual stresses as, for example, in a thick-walled pressure
vessel or rotor disc. A further application of plasticity relates to forming metals, and
although this is generally a subject outside the scope of this text we shall look at the
simple elements of beams and shafts plastically deformed.
It is evident that the same principles must apply as for elastic deformation,
namely equilibrium of forces, compatibility of deformations and a stress—strain
relationship. It is the nature of this latter material behaviour which particularly
dictates the final solution. Elastic—plastic stress—strain relationships are illustrated in
Fig. 15.1(a), (b) and (c). The first is a typical curve for a real strain-hardening
material and its linear to non-linear development causes some complication in
analysis. Because of this, semi-idealized behaviour is often assumed as in Fig.
15.1(b) in which strain hardening occurs linearly from initial yield, while in Fig.
15.1(c) strain hardening is ignored and we have a linear-elastic non-hardening
plastic relationship.

15.1 Plastic bending of

In considering the behaviour of beams subjected to pure bending which
beams: plastic moment
results in fibres being stressed beyond the limit of proportionality, the
following assumptions will be made:
1. That the fibres are in a condition of simple tension or compression.
That any cross-section of the beam will remain plane during bending as
in elastic bending. This means that the strain distribution will be linear
even if the stress distribution is not.

Stress Stress Stress

fo} oO { fo} 7
6,=yield
stress

Strain € Strain € Strain €

(a) (b) (c)
414 MECHANICS OF ENGINEERING MATERIALS
LL

In elastic bending of a beam there is a linear stress distribution over the

cross-section and when the extreme fibres reach the yield stress, the bending
moment is given by
I
My =oy- [15.1]
ay
where / is the second moment of area of the cross-section about the neutral
axis, and y is the distance from the neutral axis to an extreme fibre. The
value of the yield bending moment My will be found for beams of various
cross-section.

Rectangular section From eqn. [15.1].

bd?

and the stress distribution corresponding to this condition is shown in Fig.

15.2(a), all the fibres of the beam being in the elastic condition.

Fig. 15.2

/
|-o,-| I-o,-| \-5,
(a) Elastic (b) Elastic-plastic (c) Fully plastic

When the bending moment is increased above the value given in eqn.
[15.2], some of the fibres near the top and bottom surfaces of the beam begin
to yield and the appropriate stress diagram for a non-strain hardening
material is given in Fig. 15.2(4). With further increase in bending moment,
plastic deformation penetrates deeper into the beam. The total bending
moment is obtained by consideration of both the plastic stress near the top
and bottom of the beam and the elastic stress in the core of the beam. This
moment is called the e/astic—plastic bending moment.
From Chapter 6, the bending moment is given by

M= |ovat

The elastic component is obtained from eqn. [15.2] in which the depth is
now (d — 2h),so

b(d — 2h)
M = oy
6
The plastic component of the moment, shown shaded heavily in Fig. 15.2, is
given by

Vie oybh(d a h)
 HVE MEIN DARYS PINS hiGiay 415

Hence

— 2)"
M = oybh(d — 8) + oy

5 [!+25(1-3) 3
aybd* h h
= 1 2-—(1 == 155.3)

At a distance (Gd —h) from the neutral axis, the stress in the fibres has
just reached the value oy; then, if R is the radius of curvature, we have

_ BGd2)
oo ane”
or

1 Oy
one ee ee 4:
R_ Ed —h) ies!
The values of M and 1/R calculated from eqns. (15.3) and (15.4) when
plotted give the graph shown in Fig. 15.3. The connection between these
quantities is linear up to a value of M = My. Beyond this point the
relationship is non-linear and the slope decreases with increase in depth, A,
of the plastic state. When h becomes equal to id, the stress distribution
becomes that shown in Fig. 15.2(c) and the highest value of bending
moment is reached.
This fully plastic moment is given by eqn. [15.3], putting 4 = id, as
2 2
M, =foy-e = yy (15.5]

=iMy [15.6]
Fig. 15.3 Moment M

M, | ------7-
722002 eener eeeee

>

1/R

The value of M, is shown in Fig. 15.3 and is the horizontal asymptote of

the curve. Plastic collapse of the beam is shown, by eqn. (15.6), to occur at a
bending moment of one and a half times that at initial yielding of the
extreme fibres of the beam.
The ratio of M,/My is termed the plastic section modulus, Zptastic
(analogous to the elastic section modulus, Z, discussed in Chapter 6). For
416 MECHANICS OF ENGINEERING MATERIALS

the rectangular section it is given by bd”/4 from eqn. [15.5]. It may be noted
that it is simply a function of the geometry of the cross-section and is often
available in standard tables of data. When Zpjgsic and oy are known, the
plastic moment is given by

M, as yZ plastic

I-section When yielding is about to occur at the extreme fibres, the beam 1s still in the
elastic condition and, for the dimensions in Fig. 15.4(a),
3 3 2
My = y(t) 2 [15.7]
d

Fig. 15.4
|-o,-|

0 Fully plastic
stress
distribution

|-o, +
(a) (b)

In the fully plastic condition, the stress diagram is shown in Fig. 15.4(),
and from egn. [15.5] the fully plastic moment is given by

ees) 184
and the ratio M,/My which is termed the shape factor is
M, _ (bd?/4—bid?/4)
My — (bd3/12 —byd3/12) 2
or

M, _ 3 (1—bidt/bd?) | 59)
My 2 (1 — b,d3/bd3)
In an I-beam, 100mm x 300mm, with flanges and web 14mm and
98 mm thick respectively,

M, _ 3 [1~ (91 x 272')/(100 x 300°)] _ |6

My 2 [1 — (91 x 2723)/(100 x 3003)}
This shape factor is fairly representative of standard rolled I-section beams,
and the fully plastic moment is only 16% greater than that at which initial
yielding occurs.

Asymmetrical section In the two previous cases the neutral axis in bending of the section coincided
with an axis of symmetry. If the cross-section is asymmetical about the axis
 BE MEIN
AR Y PIAS
DT Gliny 417

Fig. 15.5

|--o,—-]
(a) (b)

of bending, then the position of the neutral axis must be determined. Figure
15.5(a) shows a T-bar section in which YY is the only axis of symmetry and
ZZ passes through the centroid of the section.
In the fully plastic condition the beam is bent about the neutral axis NN.
If A, and A) are the areas of the cross-section above and below NN
respectively, then since there can be no longitudinal resultant force in the
beam during bending without end load,

Ajay = Aroy
Of At ="47—= +A, where A is the total area of the cross-section. Thus for
the fully plastic state the neutral axis divides the cross-section into two equal
areas and the stress diagram is shown in Fig. 15.5(d).
If C) is the centroid of the area 4;, C; the centroid of the area 42, and h
the distance between C) and C), then the fully plastic moment is given by

My =! Aoyh =1oyAh [15.10]

This equation applies for any shape of cross-section. The reader may wish to
check its use for a rectangular cross-section to obtain eqn. [15.5].
Tables on the properties of areas, such as elastic and plastic section
modulus, moments of area, etc., for standard sections are widely available
from material suppliers and trade organizations.

Example 15.1
The flange and web of the T-bar section in Fig. 15.5 are each 12mm thick, the
flange width is 100mm, and the overall depth of the section is 100mm. The
centroid of the section is at a distance of 70.6 mm from the bottom of the web, and
the second moment of area /,, of the section about a line through the centroid and
parallel to the flange is 2.03 x 10° mm‘. Determine the value of the shape factor.

Let m be the distance of the neutral axis NN from the top of the flange (Fig.
15.5); then
A, =100m and ~—-A) = 100(12 — m) + (88 x 12)
100m = 100(12 — m) + (88 x 12)
m = 11.3mm
418 MECHANICS OF ENGINEERING MATERIALS

If n is the distance of the centroid of area Az from the bottom of the web,
then

(100 x 0.7 « 88.35) + (12 x 88 x 44) = [(100 x 0.7) + (ex 88)|n

n = 46.8mm

Therefore hf, the distance between C; (the centroid of A;) and C) (the
centroid of A2), is
1S
h = 88.7 — 46.8 + 7. a 47.55 mm

so that
|
Me HS: x 47.55 = 53 636ay

ZN3x 10°
My =————- oy = 28 7540y
ssuteueyOibnla oe =
and the shape factor

ag) = e8/
My

15.2 Plastic collapse of

The fully plastic bending moment developed in the preceding section was
beams
due to the application of pure bending. A beam would therefore become
fully plastic at all cross-sections along the whole length once M, was
reached. However, in practice pure bending rarely occurs and the bending-
moment distribution varies depending on the loading conditions. The point
of maximum bending moment along the beam will be the first cross-section
which becomes fully plastic as the load magnitude increases. Cross-sections
adjacent to the fully plastic section will have commenced yielding to various
depths. For a beam simply supported at each end and carrying a central
concentrated load the shape of the plastic zone associated with the central
fully plastic cross-section is illustrated in Fig. 15.6(a). The boundary
between elastic and plastic material is parabolic in shape.
The plastic zones for distributed loading are triangular in shape, as
shown in Fig. 15.6(4). When a cross-section such as those shown reaches the
fully plastic state it cannot carry any higher loading and the beam forms a
hinge at that cross-section. This is termed a plastic hinge about which
rotation of the two halves of the beam occurs, as shown in Fig. 15.7. When

Fig. 15.6
 EVE MENTARY, PIZAS
dT TGT thy 419

Fig. 15.7

S S
we
S Hinge S

(a) Elastic (b) Plastic

one or more plastic hinges occur such that the beam or structure becomes a
mechanism then this situation is described as plastic collapse.
In the example of Fig. 15.7 the maximum bending moment is at the
centre and is WL/4. Therefore plastic collapse occurs for the single hinge
formation and

Mi ce or W,= [15.11]
The next example is a cantilever propped at the free end and carrying a
concentrated load at mid-span, as shown in Fig. 15.8(a). In situations like
this, plastic collapse will only occur if multiple hinges form. The sequence of
events will be as follows:
1. Elastic behaviour occurs in a structural member until a plastic hinge is
formed at a section.
2. If rotation at this hinge results in diffusion of the load to other parts of
the structure or supports then additional load may be carried until
another plastic hinge is formed.
3. As each hinge forms the moment remains constant at the fully plastic
value irrespective of additional load or deformation.
4. When there is no remaining stable portion able to carry additional load
then collapse will occur.
5. The structure as a whole, or in part, will form a simple mechanism at
collapse.
6. The collapse load may be calculated by statical equilibrium if the
locations of the hinges can be identified.

Fig. 15.8 w /unit length

(a) (b)
420 MECHANICS OF ENGINEERING MATERIALS

Considering the loading situation in Fig. 15.8(a), the elastic bending-

moment diagram is shown dashed. At some value of load, Wy, the beam will
yield at A. The moment at A will be My at this point. As the load 1s
increased beyond Wy, yielding will also start to occur at B. As the load is
further increased, a plastic hinge forms at A when the moment there reaches
M,. This will not cause the beam to collapse. As the load is further
increased, the moment at A remains at M, and the beam behaves like a
statically determinate system. Eventually a plastic hinge forms at B and the
bending moment diagram is shown as the solid line in Fig. 15.8(a). At this
point the beam will collapse.
The value of load W, which causes plastic collapse may be determined
from a static equilibrium analysis of the beam because the moments at A and
B are known to be equal.

Pu N|r
roy

Taking moments about A

W,L
—M, + ~— RL = 0

M
cme eetae
RAS [15.12]
where Rc is the vertical reaction at C.
Taking a free-body diagram for section CB, moment equilibrium about
B gives
RL
My — - = 0 [15.13]

Using this in [15.12] gives

6M
ia
Note also that in Chapter 8, it was shown that the elastic moment at A is
given by
3
M=—WL
16
At the yield condition W = Wy and M = My so substituting for L from
[15.13] enables the ratio of the plastic load to yield load to be compared:

W, _98 MyMy
Wy [15.14]
 BELEMENDARY PLASTICITY 421

Next consider the case of a beam fixed at both ends carrying a uniformly
distributed load as shown in Fig. 15.8(4). The elastic bending-moment
distribution is shown dashed. At the collapse condition, plastic hinges will
form at A, B and C.

Letting M = My and w = my at the yield condition, then once again we

may use eqn. [15.15] to get the ratio of the plastic load to yield load. This
will be

[15.16]
This method of obtaining the plastic load is illustrated further in the
following example.

nog ded The beam illustrated in Fig. 15.9 is made of I-section mild steel having a shape
factor of 1.15 and a yield stress of 240 MN/m2. Using a load factor against collapse
of 2 find the required section modulus.

The elastic bending-moment diagram is such that peaks (of different

magnitude) will occur at A, B and C. Plastic hinges will form when these
peaks reach the value of M,. Collapse will occur when the three hinges are
formed as shown in Fig. 15.9().

Fig. 15.9

Plastic collapse condition

(a) (b)
422 MECHANICS OF ENGINEERING MATERIALS

Statical equilibrium may be applied to the collapse condition. Taking a

free-body diagram for AB and applying moment equilibrium about B gives

M, + M, —2R4=0
My, = Ry

Then taking a free-body diagram for AC and applying moment equilibrium

about C gives

(ti 2 |
Ra

M, — M, + (2 x 100)
—4R4 =0
R4= 50 EN

Therefore, from above M, = 50kNm.

Also

M)
—=1./15
My /
Hence

2-50
Y = mG = 86.96
86.96 kN
kN m

having incorporated the load factor of 2.

Now the elastic section modulus
My 86.96
Z =— =——_ = 0.362 m3
a a ea
For the hinge to form at C due to the load at D we have
M, = WE=30
x 1=30kNm

This value would require a section modulus of

& WSC AV
= (0.217m?
~ 1.15
x 240
Therefore the larger section modulus is required.
 PE ME INGAR SYS Pina Sob LG Lays 423

15.3 Plastic torsion of

In the following discussion it will be assumed that we have an ideal stress—
shafts: plastic torque
strain relationship for the material as shown in Fig. 15.10, that a plane cross-
section of the shaft remains plane when in the plastic state, and that a radial
line remains straight. The shearing strain ¥ at a distance r from the axis of
the shaft will be given by y = r6/L.

Fig. 15.10 Shear

stress T

0) Shear strain y

When the shaft has a torque applied in the elastic range, the shear stress,
T, increases from zero at the shaft axis to a maximum value at the surface of
the shaft, and from eqn. [5.11]
TH
T=—r
2
for a solid circular shaft. When the shear stress at the surface of the shaft has
reached the value Ty the torque required to give this stress is
Ta
[= [15.17]
Z
If the torque is increased beyond this value, then plasticity occurs in
fibres at the surface of the shaft and the stress diagram is as shown in Fig.
15.11. The torque carried by the elastic core is

i= ai [15.18]

Fig. 15.11 Stress distribution

Elastic

Plastic
zone
424 MECHANICS OF ENGINEERING MATERIALS
ee

where ry is the interface radius between elastic and plastic material, and that
carried by the plastic zone is
és 2
= | 2nrty dr = sandy = £,) [15.19]
Jry

and the total torque, 7) is

T+ Tr =47ymr, +42 aTy(r, 3 — ry)

3

Therefore
Pe
T =tarty ( _~5) [15.20]

When the fully plastic condition is reached, the shear stress at all points
in the cross-section is Ty, and it follows from eqn. [15.19] that the fully
plastic torque is given by
20 3
ji soam Ye [15.21]
3
and the ratio 7,/Ty is

gE
ie ie
When the fibres at the outer surface of the shaft are about to become
plastic, the angle of twist is given by eqn. [5.11]

[15.22]
ey aTyL
by
~ Gr,
and when the shaft is in the elastic—plastic condition, the angle of twist of the
elastic core is given by

[15.23]
0 ee Gell

Gr,
Since we have assumed that radii remain straight, then the outer plastic
region has the same angle of twist. From eqns. [15.22] and [15.23] it follows
that
by ry
ree [15.24]

It is evident that as the shaft approaches the fully plastic state, the angle of
twist tends to infinity.
Equation [15.20] may be expressed in the form

a 1 (6y\°
T= 3r Ty oer 7h [15.25]

Example 15.3
A mild steel shear coupling in a metal-working process is 40 mm in diameter and
250 mm in length. It is subjected to an overload torque of 1800 Nm which is known
to have caused shear yielding in the shaft.
 ELEMENTARY PLASTICITY 425

Determine the radial depth to which plasticity has penetrated and the angle of
twist. ry = 120 MN/m?, G = 80 GN/m.

Using eqn. [15.20] for the elastic—plastic torque

3
1800 aw?
= 37 x 0.02° : x 120 x 10 j (1= Bom
a
from which ry = 15mm.
Hence the depth of plastic deformation is 5 mm.
The shear strain at ry = 15mm is

A a = 0.0015

but

Very 0.
Therefore
0.0015 x 0.25 = 0.0150
and
6102 5iradi— lets

15.4 Plasticity in a
The problems analysed in the previous sections only involved a uniaxial
pressurized thick-walled
stress condition, and hence a simple tensile or shear yield stress was
cylinder sufficient to define the onset of plastic deformation. In two- or three-
dimensional stress systems it is necessary to use a yield criterion of the type
discussed in Chapter 12 in order to determine the initiation of plastic flow.
A thick-walled cylindrical pressure vessel is a good example of this type of
situation in that there will be radial and hoop stresses and generally also axial
stress. The process of inducing plastic deformation partly through the wall
thickness is known as autofrettage and its purpose was described in Chapter
14.
The following analysis will only consider an ideal elastic—plastic material,
since the problem for a strain-hardening material is beyond the scope of this
text. Furthermore, in order to simplify the mathematics the maximum shear
stress (Tresca) theory of yielding will be adopted.
For a thick cylinder under internal pressure, the maximum shear stress
occurs at the inner surface (see Fig. 14.5) and therefore as the pressure is
increased, plastic deformation will commence first at the bore and penetrate
deeper and deeper into the wall, until the whole vessel reaches the yield
condition.
At a stage when plasticity has penetrated partly through the wall, the
vessel might be regarded as a compound cylinder with the inner tube plastic
and the outer elastic. If the elastic—plastic interface is at a radius a and the
radial pressure there is p,, then from eqns. [14.35] and [14.36],
Zz 2
o, =" aC 5)
Gan GO
[15.26]
426 MECHANICS OF ENGINEERING MATERIALS
a ————————————————

oo a=P (1455 15.27

2 Z rv
pad as 15.27
and

O09 — Or Pat?
Tmax = 2 == ® = v7 [15.28]

But at the interface, yielding has just been reached; therefore

Tmax ae 2

and

Pat? hs Oy

i ay 2

Therefore

Sa a)
Pema ee)
2 [15.29]

From this value of p, the stress conditions in the elastic zone can be
determined using eqns. [14.35] and [14.36]. It is now necessary to consider
the equilibrium of the plastic zone in order to find the internal pressure
required to cause plastic deformation to a depth of r = a. The equilibrium
equation is [14.11].

oe ye Say 0 [15.30]
Hence

r sl 2Tmax = 0
or
do, _ oy
dr i
Integrating this equation gives
o, = oylog,r+C [15.31]
Now, at r = a, o, = —p,; therefore

—p, = ay log,a+ C
or
G=— p77 ovo a
Substituting for C in eqn. [15.31],
0, = oy log, r — p, — oy log,a [15.32]

a
= OY kee
a Pa [15.33]
 ELEMENTARY PLASTICITY 427

Therefore, using eqn. [15.29],

O, = —ay log,“ — ee (7 — a’) [15.34]

babe

which gives the distribution of radial stress in the plastic zone; and at

/ = ih Cp = >=);

Therefore
2
Pes ile a Wo [15.35]
Ti 2, a

where 7; is the internal pressure to cause yielding to a depth of r = a. The

hoop stress is

Op) = OF sO;

Therefore
Z
o9 = oy (1 log, “) = (: “) [15.36]
The internal pressure, Pymax, required to cause yielding right through the
wall is found by putting a = 7, in eqn. [15.35]:

Pe forlogs— ‘;
[15.37]
and from egns. [15.34] and [15.36]

ote cay lone [15.38]

r

o4 = oy(1 — log, “:) if

[15.39]
The stress distribution of o, and og for the cases considered above are
illustrated in Fig. 15.12 in terms of oy for a vessel where 7, = 27;.

Fig. 15.12
9
e 6
5
{2} Ye
5S
25
g
no

©5
o

=
SD
2

°, O
O O
=
2

¢ 3
8
5
5
fs
3
i=) co)
=
fe Co)
D
(?)
gjon
Oo
o shi 16
a
g
je}
= s)
8 —$

Elastic i

Plastic — Elastic Plastic

(a) (b) (c)

428 MECHANICS OF ENGINEERING MATERIALS

Example 15.4
A thick-walled cylindrical pressure vessel has to be autofrettaged prior to its use for
a high-pressure chemical process. The radius ratio is 2.5 and 20% of the wall
thickness has to be brought into the plastic range of the alloy steel which has a yield
stress of 400MN/m?. Calculate the internal pressure required to achieve the
specified plastic deformation. What are the values of hoop stress at the bore,
elastic-plastic interface and the outer surface caused by that internal pressure?
Determine the reserve factor against yielding right through the wall thickness.

To determine a, the radius of the elastic—plastic interface, in terms of 7; we

know that 20% of the wall thickness is in the plastic range so that
b= PUL 1) = Lr,
From eqn. [15.35],

400 1 (134 :
pj = 400 log, 1.3 + ie
2.5%;

= 104.8 + 146 = 250.8 MN/m?

At the bore

o, = —250.8 MN/m?
and

oy = 400 — 250.8 = 149.2 MN/m?

At the interface from eqn. [15.34] putting r = a,

400 3nA\”
0, = t ( | = sensi
y Dots

and

og = 400 — 146 = 254 MN/m2

The value of a, at the interface is also the value of p, in eqn. [15.29].
Therefore at r=,

AI)

and from egn. [15.27]

pe lSe
EEEwes is nie
The shape of the stress distribution is similar to that in eqn. [15.12].
For yielding right through the wall we use eqn. [15.37]

Pmax = 400 log, 2.5 = 366.4 MN/m?

The reserve factor on pressure is
366.4
508 oe
 ELEMENTARY PLASTICITY 429

15.5 Plasticity in a
In a rotating disc with a central hole the maximum hoop stress occurs on the
rotating thin disc
surface of the hole, and both hoop and radial stresses are positive
throughout the disc, the former being the greater of the two. Therefore
maximum shear stress is given by

Tmax =
(Oia) =
)
0
[15.40]
Since 50)
Yielding occurs first at the hole where r= 17, and gradually spreads
outwards with increasing rotational speed. Using the maximum shear-stress
criterion, the speed wy at which yielding first commences is given by
09 Oy
7 max ees
~~— a [15.41]
15.41

From eqn. [14.69] putting r = 7),

2
Ge = 23+) +r(1 —v)| =cy [15.42]
Therefore initial yielding commences at a speed of

4ay )Me
hye [15.43]
fac ea bw)|
Next we determine the rotational speed w at which there is a plastic zone
from r = 7; to r=c. The equilibrium equation is

do,
r 4+ 6,—09+ purr =0
dr
and, since og = oy in the plastic zone,

do,
r 4. CO, =O y— purr [15.44]
dr

Integrating this equation,

10, =?roy — Furr + K

Hence

Pes sor aK [15.45]

Mi ir= A, G: =O, Se

kK
Ceaye eae ae
3 r\

or

KS —=noy + Furr}
430 MECHANICS OF ENGINEERING MATERIALS
SS
ee eee

Substituting for K in eqn. [15.45]

3
or = oy — Ful? —ay 4h gtt [15.46]

0, =—(¢—-1n) oti re) [15.47]

But this value of a, must be the same as a, at r = ¢ for the elastic zone. It is
firstly necessary to determine the constants 4 and B in eqns. [14.64] and
[14.65]. The boundary conditions are: at r=, 0, =0; and at r=c,
og = oy. Therefore

0= 2 ae 3
= Q ; :

and

BASS 5
Oy, = Al aoe 8

from which

ie Coy +hpur[(1 + 3v)c4 + (3 + )r5]

e447

pao ney + gpwrenl(l + sve — B+ v4)

CH

Therefore, for the elastic zone at r = ¢,

_ Coy +3 pu" [(1 + 3v)c4 + (3 + v)r5]

Oy
- ce + re

Koy ae 1 pwr [(1 + 3v)e — (3+ v)r5]

e+ r

— 3+7
(=) pute22 [15.48]

Equating ens. [15.47] and [15.48] and simplifying,

( 12oy [2cr5 —n(e+ r)| Po

Temes Lene anes eeay
[15.49]

where w is the angular speed to cause plasticity to a radial depth of r = c.

 ELEMENTARY PLASTICITY 431

If w, is the speed at which the disc becomes fully plastic, then

substituting ¢ = rz in eqn. [15.49] and simplifying,

by ( acy VA [15.50]
Noh +1 +7)
The stress distributions of o, and og for various degrees of plastic
deformation are shown in Fig. 15.13 for r2 = 107.

Elastic Plastic Elastic Plastic

Fig. 15.13 jt $$$ > }_¢—__—_____ > _
SO

=a ; ee |e 6 ee

Oy oy Pe Sy
gi or

Y I
|

(a) (b) (c)

15.6 Residual stress

distributions after plastic
deformation

Bending or torsion
When a beam is bent or a shaft twisted beyond the elastic limit, permanent
deformation occurs which does not disappear when the load is removed.
Those regions which have suffered permanent deformation prevent those
which are elastically strained from recovering their initial dimension when
the load is removed. Consequently the interaction produces what are termed
residual stresses.
In a beam of rectangular cross-section, we will assume for simplicity that
the beam has been bent to the fully plastic condition such that the stress
distribution diagrams are the two rectangles Ocdb and Oeka, shown in Fig.
15.14(a). Assuming that when the material is stretched beyond the yield
point and then unloaded it will be linear elastic during unloading, then the
bending stresses which are superposed while the beam is being unloaded
follow the linear law represented by the line ajb;. The shaded areas
represent the stresses which remain after the beam is unloaded and are thus
the residual stresses produced in the beam by plastic deformation.
The rectangular and triangular stress distributions represent bending
moments of the same magnitude; hence the moment of the rectangle Ocdb
about the axis eOc is equal to the moment of the triangle Obb; about the
same axis. Since bd represents oy it follows for equal moments about the
axis eOc that the stress represented by bb; is equal to lboy, and thus the
maximum residual tension and compression stresses after loading and
unloading are db,, and ajk = toy. Near the neutral axis the residual stresses
are equal to oy.
432 MECHANICS OF ENGINEERING MATERIALS

Fig. 15.14

(a) (b)

In Fig. 15.14() the residual stresses are replotted on a conventional base

of a cross-section, in order to show these residual tensile and compressive
stresses more clearly. Diagrams showing the residual stresses in a partially
yielded beam are shown in Fig. 15.15. In this case the material has yielded to
such an extent that the stress distribution diagrams are represented by the
areas Ocdb and Oeka in Fig. 15.15(a). Again assuming the material to follow
Hooke’s law during unloading, the bending stress during this operation will
follow the linear law represented by a;b;, such that the moments of Ocdb
and Obb, about the neutral axis are equal, and the shaded areas represent
the residual stresses. These stresses are shown replotted on a conventional
cross-section in Fig. 15.15(d).

Fig. 15.15

(a) (b)

The solution for residual stresses after plastic torsion follows the same
arguments as above (see Example 15.7).
After any plastic deformation there will be some elastic recovery of the
deformation. This ‘springback’ of the material causes difficulties in metal-
forming applications such as sheet pressing and pipe bending, since the tools
forming the shape have to be designed to apply sufficient additional
deformation so that the final shape of the component after recovery is what
is desired.
The amount of the recovered deformation can be estimated by an elastic
analysis of the deformation due to the unloading moment or torque. The
procedure is the same as for the calculation of residual stress above.

Example 15.5
A flat bar 10 mm x 3mm of mild steel (E =200 GN/m?, cy =200 MN/m2) is formed
by bending it round a circular former. What is the maximum change in curvature
which. will occur due to elastic springback?

The maximum moment that could occur due to elastic unloading is the fully
plastic moment, egn. [15.5]
 EVE NUE NAR Ye PLAS Glen 433

bd?
M, = oy [15.51]
The curvature caused by an elastic moment of this magnitude would be
1 M ay3
Ra ChE eg oe
Assuming the bar is bent about the shorter dimension, i.e. d =3 mm, then

T2001 0° aes 5,
Wc =

bs Ue — ]

Gane we3

from which the change in radius of curvature is

Rl [15.54]
The most useful point that this analysis illustrates is that the amount of
elastic recovery is greatest in materials which have a large yield strain
éy =oy/E. The amount of elastic recovery is also proportional to the
amount of elastic deformation before yielding begins. Elastic recovery in
torsion is analysed in Example 15.7.

Axially symmetric If plastic deformation is caused in one part of a body and not in the
components remainder, then, on removal of external load, there still exists a stress system
in the body, owing to the strain gradient, and hence interaction between
parts of the body not being able to return to the unstrained state. Residual
stresses have important implications in engineering practice.
In the pressurized thick-walled cylinder it was assumed that plastic
deformation had partly penetrated the wall. On release of pressure the elastic
outer zone tries to return to its original dimensions, but is partly prevented
by the permanent deformation of the inner plastic material. Hence the latter
is put into hoop compression and the former is in hoop tension, such that
equilibrium exists. The residual stress distributions are then as shown
diagrammatically in Fig. 15.16. They may be calculated by using eqns.
[14.35] and [14.36] to determine the elastic unloading stresses from the
internal pressure p; (eqn. [15.35]) and superposing these onto the loaded
stress distribution. The residual compressive hoop stress at the bore has to

Fig. 15.16

©Tension
Compression

be nullified first on repressurizing, thus allowing a greater elastic range of

hoop stress and therefore a greater internal pressure.
When a rotating disc is stopped after partial plasticity has occurred, a
similar condition of residual stresses is obtained as for the thick cylinder
above. Compressive hoop stress is obtained at the central hole, which may
434 MECHANICS OF ENGINEERING MATERIALS
ee

Fig. 15.17

©Tension
Compression

be calculated using the elastic stress equations and the appropriate rotational
speed, followed by superposition onto the loaded stress pattern. The above
action, known as overspeeding, helps to increase the elastic stress range, and
hence the speed limit available under working conditions. The residual
stress distributions for this case are shown in Fig. 15.17 for the same
conditions as in Fig. 15.13(d).

Example 15.6
After the autofrettage process is completed in Example 15.4 what are the residual
stresses at the bore, interface and outer surface of the vessel assuming elastic
unloading?

The elastic unloading stress range may be found from the elastic equations
for og and o, using p; =250.8 MN/m’.
At = tr

250.8 ;
= (I+ k)
OS
= 250.8 x —— = 346.3 MN/m?
Bs

Therefore the residual stress is

of = +149.2 — 346.3
= —197.1 MN/m?
Atr =a,

250.8
09 = som + 3.69) = 224MN/m?
The residual stress

OC, nae 24
= +30 MN/m’
Atr=~r,, 09 = 95.5 MN/m’, and the residual stress

oF = 3108519515
= +13 MN/m?
 ELEMENDPARY PLASTICITY 435

At r=r, and r =,, the residual radial stresses are zero and at r = a the
radial stress due to unloading elastically from p; = 250.8 MN/m? is
o, = —128.5MN/m’. Therefore the residual stress is

o® = —146 = (—128.5) = —17.5 MN/m?

The shape of the residual stress distribution is similar to that of Fig. 15.16.

Example 15.7
Determine the residual shear-stress distribution after elastic unloading for the
plastically deformed shear coupling in Example 15.3. What is the permanent twist?

The elastic unloading torque of 1800 N m gives rise to a shear-stress range at

the outer surface of
_ 16 x 1800
= 143 MN/m
Te
x 0.043
The residual shear stress at the outer surface of the coupling is the value
of 143 MN/m? less the yield stress value of 120 MN/m? which gives
23 MN/m’?.
The residual shear stress is zero at r= (0). It is also zero where the
unloading stress range equals 120 MN/m’; therefore from Fig. 15.18(a)
120. or
143 20
r= 16.8mm

Fig. 15.18

(a)

As shown in Example 15.3, the elastic-plastic boundary occurs at a radius of

15mm where the elastic shear stress is
15
ip = Lr4350

= 107.3 MN/m
and the residual shear stress is 107.3 — 120 = —12.7 MN/m’.
The complete residual shear-stress diagram is shown in Fig. 15.18().
The elastic unloading twist or ‘spring back’ is given by
__ 1800 x 32 e 0.25
5= 0.0224 rad
— «x 0.44 ~ 80x 10
436 MECHANICS OF ENGINEERING MATERIALS
i
mn

Permanent angle of twist = 0.025 — 0.0224

= ().0026 rad

15.7 Summary
The occurrence of plastic deformation in engineering components is not
common. Nevertheless an understanding of the behaviour is important for
the analysis of modes of failure and in various metal-forming processes.
Since equilibrium and strain—displacement equations are independent of the
type or state of the material, the principal difference from elastic analysis is
the form of the stress-strain relationship. For convenience this is often
assumed to be elastic—ideally plastic, i.e. non-strain hardening, although a
linear plastic strain-hardening relationship is a straightforward extension of
the former.
Plastic bending and torsion represent elementary illustrations of the
principles of analysis. Plasticity in pressurized cylinders and rotating discs
can be seen to have important practical implications for performance
through the influence of residual stresses and strains. The introduction of
this latter concept is an opportunity to emphasize the role, both
advantageous and detrimental, played by residual stresses in engineering
components.

Problems 15.1 Determine the ratio of the fully plastic to the maximum elastic
moment for a beam of elastic-perfectly plastic material subjected to
pure bending for the following shapes of cross-section: (a) solid
circular; (4) solid square about a diagonal axis; (c) thin-walled
circular tube; (d) thin-walled square tube about a centroidal axis
parallel to one of the sides.
15.2. A steel beam of I-section, as shown in Fig. 15.19, and of length 5m
is simply-supported at each end and carries a uniformly-distributed
load of 114kN/m over the full span. Steel reinforcing plates 12 mm
thick are welded to each flange and are made of elastic-ideally plastic
material. Calculate the plate width such that yielding has just spread
through each reinforcing plate at mid-span under the given load.
Determine the positions along the reinforcing plates at which the
outer surfaces have just reached the yield = Yield stress
2 ligt)
= 300 MN/m’, second moment of area = 80 x 10~°m*.
15.3 The flange and areb of a T-section are each 12 mm thick, the flange
width is 100mm and the overall depth is 100mm. The beam is
300 mm
simply-supported over a length of 2 m and it is subjected to a point
load W at mid-span. Calculate the maximum value of W if the beam
is to be is to be designed such that yielding is permitted to penetrate
the web to a depth of 20mm. The yield strength of the beam
material is 300 MN/m?
are
ran mm—>
Daly 15.4 <A short column of 0.05m square cross-section is subjected to a
compressive load of 0.5 MN parallel to but eccentric from the central
axis. The column is made from elastic-perfectly plastic material
Fig. 15.19 which has a yield stress in tension or compression of 300 MN/m’.
Determine the value of the eccentricity which will result in the
cross-section becoming just fully plastic.
 ELEMENTARY PLASTICITY 437

158 Prove that for a beam simply-supported at each end and carrying a
concentrated load at mid-span, which has developed full plasticity,
the shape of the boundary between elastic and plastic material is
parabolic.
15.6 A horizontal cantilever of length L is simply-supported at the same
level at the free end and is subjected to a uniformly-distributed load
m over the full span. Determine the location and magnitude of the
collapse load.
1537 Part of a small bridge deck is represented as shown in Fig. 15.20.
What will be the mode of collapse and the value of the collapse load?

Fig. 15.20

15.8 A steel shaft 100 mm in diameter and | m long is in an elastic—plastic

state under a torque of 30kNm. Determine the diameter of the
elastic core of the shaft and the angle of twist. What is the value of
the fully plastic torque for the shaft? ty = 120 MN/m’,
G = 80 GN/m’.
se) A solid circular shaft is subjected to pure torsion and the material is
elastic—perfectly plastic with a yield stress in shear of 152 MN/m’.
When the shear stress at one-third of the radius from the centre of
the shaft reaches the yield stress, determine the shear strain on the
outer surface.-Also fnd the ratio of the torque carried in the above
conditions to the maxmum elastic torque for the shaft.
G = 83 MN/m’.
15.10 A solid cylindrical composite shaft 1 m long consists of a copper core
of 50mm diameter surrounded by a well-fitting steel sleeve having
an external diameter of 62 mm. If the steel has an elastic—perfectly
plastic stress-strain relationship, determine the torque that can be
applied to the shaft to cause yielding to develop just through to the
inner surface of the sleeve. Neglect any stresses set up by the fit
between the two parts of the shaft and assume there is no slipping at
the copper-steel interface. G (steel) = 83 GN/m’, G (copper) =
45 GN/m‘’. ry (steel) = 124 MN/m’, Ty (copper) = 76 MN/m’.
15-11 A thick-walled cylinder of radius ratio 2:1 and made of an elastic—
perfectly plastic material is subjected to internal pressure. Plot a
diagram of (p¢/Pmax) against a, where p, is the internal pressure to
cause yielding to a depth a through the wall, and p,,,, is the pressure
which results in yielding right through the wall.
15212 A metal disc of uniform thickness is 300mm diameter and has a
central hole of 50 mm diameter. Determine the increase in rotational
speed over that for initial yielding at the hole necessary to cause
plastic deformation throughout the disc. v = 0.3.
15.13 Calculate the residual stress at the outer surfaces of the column in
Problem 15.4 after elastic unloading from the fully plastic condition.
[sal4 A rectangular beam 30mm wide and 50mm deep is simply-
supported over a length of 2m. If it is subjected to a uniformly
438 MECHANICS OF ENGINEERING MATERIALS

distributed load of 8kN/m calculate the depth of penetration of

plastically deformed material in the beam. If the load is removed
calculate the force necessary at mid-span to straighten the beam. The
yield stress of the beam material is 240 MN/m/?.
Lous If the composite shaft in Problem 15.10 has the torque removed,
calculate the residual shear stress at the outer surface of the steel
sleeve and plot the residual stress distrbution.
15.16 A thin circular disc with a central hole has inner and outer diameters
of 51 and 304mm respectively. It is required to have a residual
compressive hoop stress of 77 MN/m? at the hole when the disc is
stationary. Assuming ideal elastic—plastic conditions, yield according
to the maximum shear stress theory and elastic unloading, determine
the rotational speed necessary to effect the required residual stress.
By how much is this speed greater than the speed for initial yielding?
Also find the depth of the plastic zone. oy = 340 MN/m’,
p = 7.83 Mg/m’, v = 0.3.
 Thin Plates and Shells
CHAPTER

The purpose of plates in engineering is to cover, generally, a rectangular or circular

area and to support concentrated or distributed loading normal to the plane of the
plate. A typical example is a pressure diaphragm, as a safety or control device,
supported around its circular periphery and subjected to uniform pressure on one
face and perhaps a central point load on the opposite face.
Some simple examples of thin shells were studied in Chapter 2 to illustrate
statically determinate problems.
The engineering applications of thin shells include storage tanks for liquids or
solids and pressure vessels for a variety of chemical processes, rocket motor casings,
boiler drums, etc.
‘Thin’ is a relative term which indicates that the thickness of the material is small
compared with the overall geometry, a ratio of 10:1 or greater being the usual
criterion.
Solutions which are more accurate for ‘thick’ plates and shells are complex and
may be found in specialized texts.
A further factor which affects the nature of the analysis is the range of
deformation, again in qualitative terms referred to as ‘small’, or ‘large’ relative to the
sheet thickness. This chapter will only consider small elastic deformations of plates
and shells.

16.1 Assumptions for

a) No deformation in the middle plane of the plate, i.e. a neutral surface.
small deflection of thin (2) Points in the plate lying initially on a normal to the middle plane of the
plates plate remain on the normal during bending.
(c) Normal stresses in the direction transverse to the plate can be
disregarded.
Assumption (a) does not of course hold if there are external forces acting
in the middle plane of the plate. Assumption (4) disregards the effect of
shear force on deflection.
The deflection, ™, is a function of the two co-ordinates in the plane of
the plate, the elastic constants of the material and the loading.

16.2 Relationships
The first important step in the analysis of plates is similar to that for beams
between moments and and is to relate the bending moments to curvature from which slope and
curvatures for pure deflection are determined.
bending Consider an element of material as shown in Fig. 16.1 cut from a plate
Rectangular co- subjected to pure bending as in Fig. 16.2. The bending moments M, and
ordinates My, per unit length are positive as drawn acting on the middle of the plate.
This plane is undeformed and constitutes the neutral surface. The material
above it is in a state of biaxial compression, and below it, in biaxial tension.
The curvatures of the mid-plane in sections parallel to the xz- and yz-planes
are denoted by 1/R, and 1/R, respectively. At a depth z below the neutral
surface the strains in the xv- and y-directions of a lamina such as abcd are

&=— ad g=— [16.1]

440 MECHANICS OF ENGINEERING MATERIALS

Fig. 16.1

M
Fig.16.2 pe oe ees ~x

anes Z

using the same approach as for beams in Chapter 6.

The stress-strain relationships are
_ oy Wy
€y = ——— and fy == ;

Combining with eqn. [16.1] and rearranging gives

_ Ez Reade
ceils NAN eal
[16.2]
Ee ie ML
EAN St fea
Equations [16.2] show that bending stresses are a function of plate
curvatures and are proportional to distance from the neutral surface.
Next, we equate the required equilibrium between the internal moments,
due to the bending stresses acting on the sides of the element, and the
applied moments M, and M, per unit length:
h/2
| 0,zdydz = M, dy
—h/2
h/2
(16.3)
| oyzdxdz = M, dx
=/2)

Substituting from eqns. [16.2] for 7, and o, in eqns. [16.3] and integrating
gives

[16.4]
THIN) PADRES AND SHELLS 441

where

Eh
D=———
12(1 — v”)
is termed the flexural rigidity.
Since the principal curvatures are given by

l Ow 1 Ow
—=-—- —~ and —=-———
Re ELE Ry Oy?
where m is the deflection in the z-direction, the relationships between the
applied moments and curvatures are

[16.5]

These equations can be written in matrix form as

{M, b=-o] 1 dies

Ow /Ox?
/ : [16.6]
My, V Ow /Oy

Alternatively, the curvatures arising from given moments on an element are

Ow /Ox?) ~ leah \ = M,
f eae « [16.7]
Ow /Oy? DWH A My,
Two special cases of interest can be identified: one where only a single
moment is acting, the other where the plate is curved in one direction only.

Beam bending
If a total moment M is applied to opposite ends of a rectangular plate of
width b, so that M, = M/b, while the other edges are free, M, = 0, eqn.
[16.7] reduces to

Oy ee) eM Ay M
[16.8]
Ox? D “Eb EI
Therefore plate theory reduces to beam theory when only a single moment
is applied. The other curvature is

Ow me (1—v*) NestM
[16.9]
jens (Comp “ET
This ‘anticlastic’ curvature, mentioned in Chapter 6, arises from the
Poisson’s ratio effect. On the side of the beam in tension a lateral contraction
occurs while on the side in compression lateral expansion occurs, Fig. 16.3.
These opposite changes in width can only be accommodated if a curvature
of the opposite sign occurs in the transverse direction.
442 MECHANICS OF ENGINEERING MATERIALS

Fig. 16.3

Cylindrical bending
If the plate is constrained so that displacement varies in only one direction,
e.g. in a short, wide beam or the axi-symmetric shell described later in
Example 16.3, then a state of cylindrical bending is said to occur, Fig. 16.4.
Assuming that » varies with x only so that 0?/0y? = 0, the moment
curvature relations become

Ow
‘deol
Ow
Mi VD =UM, [16.10]

Fig. 16.4

The plate is stiffer than the equivalent beam by a factor of 1/(1 — v*). Note
that lateral bending moments of vM, are required, so that cylindrical
bending cannot in fact occur near the free edges of a rectangular plate.

Symmetrical bending of |When the loading on the surface of a circular plate is symmetrical about a
circular plates in perpendicular central axis, the deflection surface is also symmetrical about
cylindrical co-ordinates — that axis. Any diametral section may be used to indicate the deflection curve
and the associated slope w and deflection m at any radius r, as shown in Fig.
16.5.
The curvature of the plate in the diametral plane rz is

RAS Se (16.11]
 THIN SReADES ANID SEES 443

Fig. 16.5

The second principal radius of curvature Rg is in the plane perpendicular to

rz and is represented by lines such as PQ which form a conical surface so
that

Ee = u [16.12]
Iy

Now we shall consider an element of the plate subjected to bending

moments along the edges M, per unit length and Mg per unit length
respectively, as shown in Fig. 16.6. This element can be analysed in the
same manner as for rectangular co-ordinates. Thus eqns. [16.4] can be
expressed for the circular plate as follows in terms of slope w:

Moe! 3
eee
12(1 — v*) \ dr f 116.13}

EB (bd -
aS 12(1 — v’) (:184 =)

Fig. 16.6

Mp
444 MECHANICS OF ENGINEERING MATERIALS

These equations can then be interpreted in terms of curvatures or deflection

by further substitution to give

apm v dp 1 tesVy2
RTE Aly OdSteSMacia nee piel 51 ene
ees ce
- o(S “1r =) (+z)
[16.14]

16.3 Rerationsaliis We can determine bending stress as a function of bending moment by

between bending moment — gjiminating the curvatures between eqns. [16.2] and [16.4] so that for
and bending stress _ rectangular plates,
12M ,z 12M,z
OR ae and oy= ai [16.15]
Similar expressions apply for the bending stresses in circular plates as a
function of M, and Mg.

. 16.4 Relationships If an element, Fig. 16.7, is taken from the plate then it must be in
between load, shear force equilibrium under the action of uniformly distributed loading p per unit area
and bending moment and the resulting shear forces Q per unit length and bending moments M
per unit length. Owing to symmetry there are no shear forces on the radial
sides of the element. For vertical equilibrium,
d
Ordé + pr dr dé — (2+ Rar)(r+ dr)dé = 0
r

Fig. 16.7

from which

ce
d
[16.16]
dr r

For moment equilibrium,

dM,
(a,a rp tr)(r + dr)d@ — M,rd@ — 2Mg arsin

+Qr dé dr = 0
 DUN RAE S AN DESHEL IS 445

which may be simplified to

dM,
a pt
r
Mr — Mo + Or= 0 [16.17]

16.5 Relationships
Next we substitute eqns. [16.13] into eqn. [16.17] and after simplifying
between deflection, slope obtain
and loading
fy ldp » @
7ORS dr rr #£D [16.18]

which relates the slope at any radius to the shear force.

If eqns. [16.14] are substituted into eqn. [16.17] then

d’y 1 d'p 1d» Q

[16.19]
di rd r dr D

expresses the variation of deflection with radius.

Equations [16.18] and [16.19] can be expressed in a form which makes a
solution by integration rather more obvious, as follows:

dit 3). 6

and

diid/d\i o

The shear force Q is a function of the applied loading p and may be

related by simple statical equilibrium or by integrating eqn. [16.16].
Multiplying through by rdr gives

rdQ+Qdr=prdr or d(Qr) =prdr

or=| prdr
0

Substituting in eqn. [16.21]

oe a =5| r dr 16.22
"ar|rdr\ dr)| D a he 22

If we know that pf is equal to f(r) or is constant then eqn. [16.22] can be

integrated to find the deflection at any radius.
The constants of integration are evaluated from the boundary conditions
for the particular problem being solved.
Some typical loading and boundary situations will now be considered.
446 MECHANICS OF ENGINEERING MATERIALS

16.6 Plate subjected to

uniform pressure /2, sincep
In this problem the right-hand side of eqn. [16.22] reduces to pr’
is constant; therefore
fl jill el oe UE [16.23]
dr |r dr dr 2D

Integrating,

1 d/ dw pr’
= G
r dr (-
a) 4D al

Multiplying both sides by r and integrating again,

dm pr’ Gr’
+ C
aa tope 2
di pr Cir ©
dr HGi* M22. Bz Vide!
Finally

4
pr Cr’
Hot gt C2 log, r + C3 [16.25]

Clamped periphery For a plate of radius a the boundary conditions are dw/dr = 0 at r = 0 and
r=a, and w= 0 at r = a (Fig. 16.8(a)).

Fig. 16.8

Hence

3 2
C,=0 and ts =0 so that qQ=-%

From eqn. [16.25],

4 4 4
SUS PUES _ pa
Mi meDe feo. aD

[16.26]
 THIN PLATES AND) SHELLS 447

The maximum deflection occurs at the centre, so that

pat
aes [16.27]
Bending stress may now be determined from the moment-slope
relationships, eqns. [16.13] and [16.24]:

M, == (@(1 +7) — 73 +v)]

Masep
caret, (1 + v) — (1 + 3y))
At the periphery r = a and

MM, = es and Myg= ye

At the centre r = 0) and

M, = Mo =! +v) [16.28]

From eqns. [16.15] we have

12M,z 12Mgz
ef. = 7B and og = ae

The maximum stresses occur at r = a and z = +h/2; hence

ee 6M max 3 pa’
as [16.29]

fi ——s Upar /I?

Simply supported For this case the boundary conditions to determine three constants of
periphery Wiicerauon are. 1, — U7 — a dp/dr— 0, rs 0 and» — 07 — a, The
problem will instead be tackled by an alternative method using the principle
of superposition. We can use the solution from the fixed edge case combined
with that for a plate simply supported and subjected to edge moments equal
but opposite in sense to the fixing moment, as illustrated in Fig. 16.8(A).
Since the plate deforms to a spherical surface M, = Mp, R, = Ro then
from eqns. [16.13]
1 M
Rae (1+v)
The deflection at the centre of a spherical surface of radius a and curvature
1/R is
448 MECHANICS OF ENGINEERING MATERIALS

Therefore
2
Sa [16.30]
2D(1 + v)
But from the previous section the fixing moment is M, = —pa?/8; therefore
4
Dee
16D(1 + v)
[16.31]
The resultant deflection at the centre for the simply supported plate is, by
superposition of eqns. [16.27] and [16.31],

Re pat pat 5+V 4

AD = 16D ay = eAl-eoy 16.32
be
The maximum bending moment occurs at the centre, and by superposition
of eqn. [16.28] and M = pa’/8 for the pure bending contribution,
2
Mex Mos = B+v) [16.33]
The maximum stress occurs at the centre and is again obtained using eqn.
[16.15]; thus

3
Ornax a i DEO (3 “ v)

16.7 Plate with central

;
sialon A ring g of moments is applied
hile pp to the inner and outer
outer boundaries
i as shown in
i
Fig. 16.9. Since there is no shear force at the inner boundary, eqn. [16.20]
Edge moments _ reduces to

afla( a] _,
dr |r dr a _ 6.34]

Fig. 16.9 Mz My

Integrating twice and simplifying gives

dv_ CrG
el : [16.35]

and

Cir’
D= rv ar C2 log, r+ C; [16.36]
 THIN RPEATES AND SHELLS 449

To find C, and C; we use the boundary conditions: at r = 6, M, = Mj;

and at r = a, M, = Mp; and by substituting eqn. [16.35] and its differential
into eqns. [16.13] and solving, we obtain

2(a2M, — b°M)
oie (1+v)(@2—B2)D
ab? (M, = M;)

shes (1 —v)(@ —P)D

Since w = (0) at r= a then, from eqn. [16.36],

a’(a?7M,—0°M,) — a?b*(M — M;) log, a

S704 @-P)D* (-v(e-PD
The deflection at any radius may now be determined by substituting the
values of C,, Cz and C; into egn. [16.36].

Edge forces In this case uniformly distributed transverse forces Qp are applied at the
inner and outer edges as shown in Fig. 16.10. The shear force per unit
length at radius r is

Substituting in eqn. [16.20] and integrating gives

dw _ Qo (72 log.r— = 1) + Cir

ores 5 + ; [16.37]

bOor? Cyr
p= - (log, r— 1) + arr C log, r+ C3 [16.38]

Fig. 16.10 fe a

For a simply supported periphery the boundary conditions are: at r = a,

w— 0 and MM, — 07307 — 7M, — 0.
By substituting eqn. [16.37] and its differential into eqns. [16.14] and
solving, we obtain

bQo ( ade 2(a’ log,a— b’ log, =)

DTN iy ae
Cc T

bQ) (Atv) @P b
== ] z
C2 2D (*—v)av—b Taig
450 MECHANICS OF ENGINEERING MATERIALS
I

Then, from eqn. [16.38],

a-bOy ny ee b 2%A1+v)
gape te, “he
= lo
or AD Cares
hb
x 7 po 108 lo

These constants may now be substituted into eqns. [16.37] and [16.38] to
give the slope and deflection at any radius.

16.8 Solid plate with

Considering the problem solved in the previous section and equating the
central concentrated force
total load around the inner periphery to the concentrated force F so that
Simply supported edge
27bQ) = F
and taking the limiting case when 4 is infinitely small, 6’ log,(b/a) tends to
zero and the above constants of integration become

F fl-v
Gi = SS ||
(+= ]05.4), G=0
5

ee Fa’ ra 1/l—-v
~~ 8nD 2\l4+v

Substituting these values into eqn. [16.38],

= F 347 y 2
Apes
9 r
16.39
- leet ase! 08.) i637)
which is the deflection at any radius for a solid plate simply supported and
subjected to a concentrated force at the centre.

Fixed edge For the fixed edge case we find the slope at the edge for the simply
supported plate by differentiating eqn. [16.39]; then this slope has to be
made zero by the superposition of the appropriate ring of edge moments.
The deflection due to these moments may then be superposed onto the
deflection given by eqn. [16.39] to obtain the total deflection.

Example 16.1
A cylinder head valve of diameter 38 mm is subjected to a gas pressure of 1.4 MN/
m2. It may be regarded as a uniform, thin, circular plate simply supported around
the periphery by the seat, as shown in Fig. 16.11. Assuming that the valve stem
applies a concentrated force at the centre of the plate, calculate the movement of

bititts the stem necessary to lift the valve from its seat. The flexural rigidity of the valve is
260 Nm and Poisson’s ratio for the material is 0.3.

We have already derived solutions for a simply supported plate subjected to

uniform loading, p, and a concentrated force, F, at the centre; hence the
deflection at the centre is equal to the sum of the deflections due to the two
separate load components. Therefore
Fig. 16.11
Fa (3+v) pat (5+v)
Pmax v-
16 (1+v) |64D (140)
 CHIN PLADES AND SHELLS 451

but when the valve lifts from its seat F = —za"p; therefore

as (7+3v) pat
m ~~ “(-Ev) 64D

7.9 x 1.4 x 10° x 0.0194

1.3 x 64 x 260

I —(0.0665 mm

:oo ahh The solutions that have been obtained in the previous sections can be used
oading and eee to advantage with the principle of superposition to analyse a number of
condition other plate problems.

Concentric loading A plate which is uniformly loaded transversely along a circle of radius d, as
illustrated in Fig. 16.12(a), can be split into the two components shown at
(d), and the separate solutions which have been obtained previously can be
superposed using the appropriate boundary conditions, namely that there
must be continuity of slope at radius b.

Fig. 16.12

Distributed loading on ‘The situation illustrated in Fig. 16.13(a) can be simulated by taking the
plate with central hole deflection of the solid plate subjected to uniform loading and superposing
that due to the appropriate moment and shear force at radius 0, as in Fig.
16.13(d).

Fig. 16.13
Mo | M

(b)

Fixed inner boundary This edge condition can be associated with several types of loading and
merely entails applying the appropriate moment to give zero slope and shear
force as a function of the applied loading.
A variety of configurations and loadings are illustrated in Fig. 16.14, all
of which can be dealt with easily by superposition of the required
452 MECHANICS OF ENGINEERING MATERIALS

Fig. 16.14 Case 1 Total force = F

Case 4

—— I

components which give the appropriate boundary conditions. However, in

all these cases the maximum deflection can be represented by the following
relations:

Dmax = © OF Dmax = € 3

where ¢’ is a factor involving the ratio a/b and Poisson’s ratio.

The maximum stresses can also be expressed by formulae as follows:
pa “ c'F

where 2” is also a factor as defined above. Values of ¢’ and c’ for v = 0.3 and
a/b in the range iF to 5 for the cases in Fig. 16.14 are given in Table 16.1.

Table 16.1 Coefficients c’ and ¢” for the plate cases shown in Fig. 16.14

a/o= 25 ils jy 3 45 5
Case cl To I c yr c cl c Ui c I c

1 IQ Wiskal E265 OO LAS 0672 ESS O07 34 2 aN ee ON7242134 O04,

2 0.66 0.202 E19 O49 2204902 334 el 22 ees ne OO eonOlmemnte si
3 0.592 0.184 0.976 0.414 1.440 0.664 1.880 0.824 2.08 0.830 2.19 0.813
4 0.194 0.00504 0.320 0.0242 0.454 0.0810 0.673 0.172 1.021 0.217 1.305 0.238
5 0.105 0.00199 0.259 0.0139 0.480 0.0575 0.657 0.130 0.710 0.162 0.730 0.175
6 0.122, 000343 0336 (0.0313 0:74 0.1250 U2) 0.291 1.45; 0417 8590492
7 0.135 0.00231 0.410 0.0183 1.04 0.0938 2.15 0.293 2.99 0448 3.69 0.564
8 0.227 0.00510 0.428 0.0249 0.753 0.0877 1.205 0.209 1.514 0.293 1.745 0.350
9 0.115 0.00129 0.220 0.0064 0.405 0.0237 0.703 0.062 0.933 0.092 1.13 0.114
10 0.090 0.00077 =0:273 «0.0062 0.71 010329 154 O110 2.23 0.179 2:80 0.234
 EUINGPIZAATBS AN DS SIHIEIC IES 453

16.10 Axi-symmetrical
The analysis of thin shells of revolution subjected to uniform pressure was
thin shells
treated as a statically determinate problem in Chapter 2. The fundamental
relationship between the principal membrane stresses in the wall and the
principal curvatures of the shell to the applied pressure and wall thickness
was shown to be

a4 f [16.40]
‘al i)

The simple applications of eqn. [16.40] to the cylinder and sphere under
internal pressure were also dealt with in Chapter 2, together with an
example on the self-weight of a concrete dome. The following example on
liquid storage illustrates a further use of a thin shell.

Example 16.2
The water storage tank illustrated in Fig. 16.15, of 20 mm uniform wall thickness,
consists of a cylindrical section which is supported at the top edge and joined at the
lower end to a spherical portion. An angle-section reinforcing ring of 5000 mm?
cross-sectional area is welded into the lower joint as shown.
Calculate the maximum stresses in the cylindrical and spherical portions of the
tank and the hoop stress in the reinforcing ring when the water is at the level shown.
Density of water is 1000 kg/m’.

Fig. 16.15

The total force due to the weight of water in the tank is

2 3
1 = 98 (nx 6x)+
18| (Bas!) + (EZ )

=(5« 5? x

=35MN
The axial stress in the cylindrical part of the tank is
36
ee 26.95 MN/
Se ae een ae ep
454 MECHANICS OF ENGINEERING MATERIALS

and the hoop stress 1s

6% 9.81% 103 x4
— 11.8 MN/m?
Oh
ro 0.02
The maximum stress in the spherical portion of the tank occurs at the
bottom, where the pressure is 9.81 x 10° x 8 N/m’; therefore

9.80
ahex 5
— 9.81 MN/m?
Ce MEAS
For the reinforcing ring, the tangential force per unit length at the edge
of the spherical part is W/(27 x 4sin d). The inward radial component of
that force is

——————60s ?
2nx4sing |

and therefore the compressive force in the ring 1s

W cot d
x
2a x 4
and the compressive stress is
_Weoote . 1
76 I~ 5000 x 10-8
ean ey: ;
SO ae ke
16.11 Local bending
Whenever there is a change in geometry of the shell, particularly for
stresses in thin shells
discontinuities in the meridian such as in the above example, the membrane
stresses cause displacements which give rise to local bending in the wall.
The resulting bending stresses may be significant in comparison with the
membrane stresses. This was the reason for introducing the reinforcing ring
in the above problem.
To illustrate the method of analysing local bending we will consider the
elementary situation of a cylindrical vessel with hemispherical ends of the
same thickness subjected to internal pressure as illustrated in Fig. 16.16.
The membrane stresses for the cylinder are

Fig. 16.16
 THIN PLATES AND SHELLS 455

and for the hemisphere,

oO Pe!
7
as calculated in Chapter 2.
The corresponding radial displacements for the cylinder and hemi-
sphere, respectively, are

r r2
Ur; = (01 — VO2) =F(2 —v)

2
es AG =e a(1 =
The difference in deformation radially is

C= Pe--0-yjJ=
pr
2 l =
pr
16.4]
16.41

In order to overcome this difference, shear and moment reactions are set up
at the joint as shown in Fig. 16.16.
Since the cylindrical section is symmetrical with respect to its axis we
may consider the reactions Qp and Mo per unit length acting on a strip of
unit width as illustrated in Fig. 16.17. The inward bending of the strip due
to Qp sets up compressive circumferential strain. If the radial displacement is
v then
Fig. 16.17
v : Ev
€g=- and og =—
r r
The circumferential force due to this stress on the edge of the strip per unit
length is Evt/r. The outward radial component of this force is

pit 2 Bvt. i Evt 49

r 2 r

Evt

This force opposes the deflection of the strip and is distributed along the
strip, being proportional to the deflection v at any point. This is a special
case of the bending of a beam on an elastic foundation for which the
deflection curve is

EL ets ae
des
where » is the distributed loading function. For the present problem EJ is
replaced by D, since the strip is restrained from distortion by adjacent
material as for plates; therefore

D v = —4D60 [16.43]
dx* r2
456 MECHANICS OF ENGINEERING MATERIALS

The solution of this equation is

v = exp(Gx)(A cos Bx + Bsin Bx)

-+exp(—x)(C cos Bx + D’ sin Gx) [16.44]

where 4, B, C and D’ are constants determined by the boundary conditions,
and

ee rice ee
Be 4Dr2 i rt?

As x — 00, v> Oand M — 0, which gives A = B=0. Atx = 0, M = Mo

and QO = Qp; therefore

dv
PR or
and

dev
ian
from which

1 Mo
G (Qo —-3My) and D' =
~ 28D 22D
Substituting into eqn. [16.44], the deflection curve for the strip becomes
—Bx

v = ———
20°D
[Op cos Bx — BMp(cos Bx — sin Bx)] [16.45]
This is a rapidly damped oscillatory curve, and thus bending of the cylinder
and head is local to the joint. The bending deflection and stresses decay over
a characteristic length. For most materials,

B= VeRO = YM = 1.25 V0
It can be seen that the bending stresses and deflections decay most rapidly in
thin-walled tubes of small diameter.
When the cylinder and head are of the same material and wall thickness
then the deflections and slopes at the joint produced by Qp are equal and
Mp = 0. Therefore the boundary condition is at x = 0, v = 6/2, and from
eqn. [16.45],

6 Qo
2 263D
or

eH oF D=~ 2tE4Gr2
Pr ge Pe
88 [16.46]
 THIN PLATES AND SHELLS 457

The deflection curve becomes

e—9*pcos Bx e.pr ee:

* cos Bx [16.47]
1634D 4tE
By differentiating this equation twice the bending moment and hence the
bending stress can be calculated for any cross-section. These have to be
added to the membrane stresses to get the resultant stress.
If the wall thickness of the head and cylinder are different then My 4 0
and the boundary conditions would be that (a) the sum of the edge
deflections must be zero, and (4) the rotation of the edges must be the same.
The above solution is equally applicable for other shapes of head.

Example 16.3
Calculate the local bending stresses in the vessel shown in Fig. 16.16 if
p = 1MN/m7, r = 500mm, t = 100mn, v = 0.3.

1/4
p= (Aaa 418.200
O57 a012

pebah er,
tel S le we oie
Since the bending moment in the strip is 41= —D(d*v/dx’), and from
eqn. [16.45] with Mp = 0,

v= a e ** cos Bx

Therefore

i= -* e ** sinBx

This expression takes the largest value for Gx = 1/4, which gives
M7 = 0121 kKNm/m
This gives rise to a maximum bending stress of

6Minax 6X 0.121 x 10°

CE ee a pee oN
The membrane stress is

pr «1K 0.5 ;
ee eel a
The total axial stress is 0, = 7.26 + 25 = 32.26 MN/m’.
The bending of the strip also produces circumferential stresses: (a) since
the strip displacement, v, varies only in the axial direction, the strip is loaded
in cylindrical bending giving circumferential stresses of +6vM/0?; (b)
stresses due to shortening of the circumference of —Ev/r.
458 MECHANICS OF ENGINEERING MATERIALS

Using the above values for v and M and summing gives

Qo exp(—x) (Ae i241)

(hoop) 0% = Be

= 22.6 exp(—Bx)(0.3 sin Bx — 0.55 cos Bx) MN/m/?

The maximum value of this expression is 1.58 MN/m’, which is small

compared with the hoop membrane stress.

1x 0.5
(hoop) Om = = *a = 50 MN/m?

Hence local bending does not have a serious influence in this particular case.

16.12 Bending ina

cylindrical storage tank An upright cylindrical storage tank, of radius r, uniform wall thickness ¢, and
height A, is filled to the top with liquid of density p. The base of the tank is
built into its foundation, Fig. 16.18, and we need to design for the maximum
bending moment due to discontinuity in the shell at the base. Since t < h or
r the shell may be regarded as infinitely long.
The governing equation is basically the same as eqn. [16.43] with an
additional term which defines the variation of pressure loading due to the
liquid:

d‘y a
Dat 4DGB"v = —pg(h — x) [16.48]
dx

fe eenrhe particular integral part of the solution is —pg(h — x)/4DG* and the
complete solution is

v = exp(Gx)(C) cos Bx + C) sin Gx)

+exp(—x)(C3 cos Gx + Cy sin Bx) —

path
ae
— x) [16.49]

The boundary conditions are as follows:

1. The height of the cylinder can be regarded as ‘infinite’, so that M — 0
and v — 0, giving C; = C, = 0.
2; At e=000 = 0'and do/dr— 0: hence

pgh pg 1
ange Me = ane (: |
C = d = ee eee

Putting these values in eqn. [16.49] gives the deflection curve

pgh x eye Lie 1 :

v = ‘DB 1 pe Bx
= yet
Ses Bx—e (1en ai)sin Bx|

[16.50]
 THIN- PLATES AND SHEELS 459

But M = —D(d’v/dx’), so that differentiating eqn. [16.50] twice and

substituting gives

h ee 1 :
ees -e* sin 3x + (1- ale cos 3x] [16.51]

Now, the maximum value of M occurs at the discontinuity, where x = 0;

therefore

pgh l pg

The resultant bending stresses can now be calculated as in Example 16.3,

and as shown in Fig. 16.19. This illustrates the rapid decay in the localized
bending as we move away from the restraint. This plot was generated using
a spreadsheet ‘what if? facility to calculate hoop and axial stress for a range of
x-values.

Fig. 16.19
5X 10° 7

eee Se ge a ee Se SPR See se ae

PM Soa Se ee oe ee

GERM 2 Serie ance enSe SO A On sea aceon oes

Hoop stress
Axial stress
Stress1X 10° +
(MIN/m?) }->4------------------*84,4,---------------

0x 10°

21 10°
0

16.13 Summary
The basic principles for the analysis of plates have been developed involving
equilibrium, geometry of deformation and the stress—strain relationships.
These are essentially the two-dimensional development from the simple
bending theory for beams. Thereafter, each plate solution is dependent on
the particular boundary conditions of loading and support.
The design of thin shells depends principally on the magnitude of the
general system of membrane stresses. However, attention must also be given
to the effect of local bending stresses at regions of discontinuity in the shell.
These stresses are of the same order of magnitude as the membrane stresses,
but they decay rapidly with distance from the discontinuity in thin-walled
shells. Somewhat surprisingly, reinforcement of a shell with rigid stiffeners
increases the stress in a shell due to local bending.
460 MECHANICS OF ENGINEERING MATERIALS

Bibliography
Timoshenko, S, and Woinowsky-Kreiger, S. (1970) Theory of Plates and
Shells, McGraw-Hill, New York.
Young, W. C. [1989] Roark’s Formulas for Stress and Strain, McGraw-Hill
International, New York.

Problems 16.1 Show that for a flat circular steel plate subjected to a uniform
pressure on one surface, the maximum stress when the periphery is
freely supported is 1.65 times that when the periphery is clamped.
Vea:
16.2 Calculate the ratio of (i) the radial stresses, and (11) the tangential
stresses at the edge and centre of a flat circular steel plate clamped at
its periphery and subjected to a uniform pressure p. v = 0.29.
16.3 A circular thin steel diaphragm having an effective diameter of
200mm is clamped around its periphery and is subjected to a
uniform gas pressure of 180 kN/m/?. Calculate a minimum thickness
for the diaphragm if the deflection at the centre is not to exceed
0.5mm. E = 208 GN/m’, v = 0.287.
16.4 A circular aluminium plate 6mm thick has an outer diameter of
250 mm and a concentric hole of 50 mm diameter. The edge of the
hole is subjected to a bending moment of magnitude 900 Nm/m.
Determine the deflection of the inner edge relative to the outer.
E = 70 GN/m’, v = 0.3.
16.5 The end plate of a tube is made from 5 mm thick steel plate as in Fig.
16.20. If a 30 mm diameter rod welded to the end plate is subjected
to a force of 10kN what would be the movement of the rod?
Calculate also the maximum stresses in the end plate.
E = 207 GN/m’, v = 0.29.

Fig. 16.20 —*| = 5mm

5 mm dia.

16.6 Figure 16.21 shows a long hydraulic cylinder and piston. Find an
expression for the radial displacement of the cylinder along its length
in terms of the supply pressure p, and the bending stresses in the
cylinder at the locaton of the piston.

Fig. 16.21
 THIN PEATES AND SHELLS 461

16.7 A circular plate 500mm diameter and 2.5mm thick is clamped

around its edge and is subjected to a concentrated load of 900 N at its
centre. Caculate the radial and tangential bending stresses at the
fixedvedsery = 0:29:
16.8 The initially flat, circular plate shown in Fig. 16.22 is stepped so that
the central disc has twice the flexural rigidity (2D) of the outer
annulus (D). There is no external loading on the disc except at the
edges, where the plate is clamped so that the slope at its edge is a.
Find the moment at the step in the plate. For the plate material take
VI=AN339:

Fig. 16.22

16.9 A pressure transducer uses a probe to measure the deflection of a

thin circular diaphragm, clamped at its periphery and having a rigid
seat at its centre to eliminate curvature in line with the probe as
shown in Fig. 16.23. Calculate the thickness of the diaphragm if the
probe can measure 0-0.25mm and the range of the pressure
transducer is to be 0-0.5 MN/m?. Young’s modulus and Poisson’s
ratio for the diaphragm material are 207GN/m? and 0.3
respectively.

Fig. 16.23

Pressure p

16.10 A 200mm diameter circular plate is clamped around its periphery

and subjected to a uniform pressure of 15 MN/m’. Ifa rod supports
the plate at its centre so that the central deflection of the plate is zero
calculate the force in the rod.
16.11 A circular plate is to be simply-supported at its periphery and
subjected to a load of 15 kN distrbuted around a circle which has the
same centre as the plate. If the ratio of the plate diameter to the
loading circle diameter is 2 calculate the thickness of the plate so that
its maximumn stress does not exceed 200 MN/m’. v = 0.3.
16.12 A circular steel plate of 304mm diameter and 12mm thick 1s
clamped around the edge. A concentric ring of loading of 20 kN 1s
462 MECHANICS OF ENGINEERING MATERIALS

applied uniformly on a circle of 152mm diameter. Calculate the

deflection at the centre of the plate. E = 208 GN/m’, v = 0.29.
16.13 Determine the membrane stresses in a concrete hemisohoncl dome
of radius 4m and thickness 200mm. The concrete has a density of
2.31 Mg/m?.
16.14. A conical water-storage tank has an included angle of 60° as shown
in Fig. 16.24 and a vertical depth of water of 3m. If the wall
thickness is 5mm determine the location and magnitude of the
maximum hoop and meridional stresses. The water loading 1s
9.81 kN/m?.

Fig. 16.24

16.15 A toroidal pressure vessel has an interior diameter of 1m and an

exterior diameter of 2m as shown in Fig. 16.25, with a wall thickness
of 4mm. Derive expressions for the principal membrane stresses and
determine values at the horizontal section of symmetry for an
internal pressure of 150kN/m/?.

Fig. 16.25

2m dia.

16.16 A steel pipe of outer diameter 200 mm and wall thickness 6 mm is to

have a steel ring shrunk on to = outside. The inside diameter of the
ring is 199mm and it is 80 mm” in cross-sectional area. Determine
the maximum local bending stress in the pipe due to the shrink
pressure of the ring. E = 208 GN/m’ and v = 0.3.
16.17 A stainless steel drum of outer diameter 400mm and 6mm wall
thickness is closed at each end by steel discs which house central
bearings. The drum has to rotate in service at 1000 rev/min.
Determine the local bending stress in the drum where it is fixed to
the discs. E = 208 GN/m’, v = 0.3, p = 7700 kN/m’.
16.18 An upright cylindrical steel tank is built in to a concrete base. It is
3m in diameter and has a wall thickness of 12 mm. If the height of
water during a i! is 2m determine the maximum bending stress.
= 1000 kg/m’, v = 0.3.
CHAPTER a Finite Element Method

The finite element method for analysing structural parts has been around since the
1950s. The method was first developed for use in the aerospace and nuclear power
industries. Here, the safety of the structures is critical: they involve large capital
expenditure and the economic consequences of a failure are very severe, so the cost
of the analysis is justified. Today the method is also extensively used in areas such as
the automotive industry, where components are relatively cheap but are
manufactured in large volumes. Furthermore, any small reduction in the safe
weight of a component such as a connecting rod can lead to additional benefits in
areas such as vibration reduction and fuel economy.
The growth in the usage of finite element methods is directly attributable to the
rapid advances in computing technology in recent years. Today there are a number of
large software companies developing and marketing finite element and associated
modelling software. As a result, there exist commercial finite element packages
capable of solving the most sophisticated problems, not just in structural or stress
analysis, but for a wide range of phenomena such as steady and dynamic
temperature distributions, fluid flow, and manufacturing processes such as injection
moulding and metal forming.
Despite the proliferation and power of commercial software, it is still very
important to have an understanding of the principles of the technique, so that an
appropriate analysis model can be selected, correctly defined and interpreted. In this
chapter the basic principles of the technique will be illustrated on some of the
simplest types of finite element. Any solved problems are sufficiently small that the
numerical details do not interfere with the interpretation of the results.
A full appreciation of the power and utility of the technique will require some
experience and experimentation with commercial software. NAFEMS (the National
Agency for Finite Element Methods and Standards) describe a series of benchmark
problems for which answers are known. These are frequently used to develop the
expertise of novices.

17.1 Principle of finite

Glemncat method If a truss of the type shown in Fig. 17.1(a) is being analysed then it is a
straightforward exercise because it is formed from discrete members. The

15 Individual
1
Fig. 17.1 | Deemer

“
le oO O— F
464 MECHANICS OF ENGINEERING MATERIALS

paths of force transmission through the truss to the ground are readily
apparent and the forces may be determined using equilibrium equations as
illustrated in Chapter 1. If the truss had been statically indeterminate then
both equilibrium equations and deformation compatibility would be
necessary, but the method of solution is well established. If, however, a
plate of the same shape as the truss (Fig. 17.1(d) is to be analysed then this is
not so straightforward. The reason is that the plate is an elastic continuum
and since the force transmission paths are not readily apparent the problem
does not lend itself to simple mathematical analysis. Although, for the
purpose of analysis, it might be tempting to consider the truss as being
equivalent to the plate, this would not give accurate results since it ignores
the restraining effect which all points in a continuum will experience and
exert on neighbouring points. However, if the continuum was considered to
be subdivided into a large number of triangular panels (Fig. 17.1(c)) it
should be possible to develop a picture of the stress distribution in the whole
plate by analysing each of the small panels in turn. To do this it would of
course be necessary (a) to analyse the equilibrium of each of the triangular
panels in relation to its neighbours and () to have available equations for the
geometry of deformation and the stress—strain relationships for a triangular
panel. This subdivision of a continuum into a large number of discrete
elements is the basis of the finite element method of stress analysis. The
triangular panels referred to in this example are the ‘elements’, but this is
only one type of element. Others include a spring element (one
dimensional), a plane rectangular element (two dimensional) and solid
elements (three dimensional) as shown in Fig. 17.2.

Fig. 17.2 Types of finite elements

—\\WWWee
Spring Triangle Quadrilateral

Tetrahedron Hexahedron Axi-symmetric

The accuracy of the solution depends on the number of subdivisions

(elements); the more there are the greater the accuracy. However, although
the analysis of each individual element is straightforward the analysis of a
large number of elements becomes extremely tedious. For this reason finite
element solutions to problems are generally carried out on computers and
there are many commercial software packages available.
As an introduction to finite element methods it is convenient to consider
the matrix analysis of skeletal structures using the stiffness method. An
introduction to matrix algebra is given in Appendix B.

17.2 Analysis of uniaxial

When a uniaxial bar such as a pin-jointed tie or strut is part of a structure,
bars
its ends will be able to move due to displacement of the structure and
A single element deformation of the member. This may be modelled by a spring element as
shown in Fig. 17.3.
 FINITE ELEMENT METHOD 465

Fy, U4 ky Fo, Up

The points of attachment of the element to other parts of the structure

are called nodes and are indicated by the points | and 2 in Fig. 17.3. Here, F
and wu are the force and displacement values and their suffix indicates the
node to which they apply. In situations where the spring element is used to
model a tie or strut of length Z and area A, the stiffness of the spring &, will
be given by

4-28 My
(17.1]
where FE is Young’s modulus for the material.
For the simple system illustrated in Fig. 17.3, using the sign convention
that forces and displacements are positive in the x-direction, then the forces
may be related to the displacements by the following equations:

Es = ky (uy = u2) = kyu, = ky u2 [17.2]

ji) = ky (u2 = U1) = —kyu, + kyu [17.3]

Equations [17.2] and [17.3] may be written in matrix form as

feast Neearer elt)

Fy
=
ky
;
—k uy

ia
17.4

In shorthand form this may be written as

{F} = [K‘]{u} [17.5]

where [K‘] is referred to as the stiffness matrix for the spring element. An
important property of the stiffness matrix for an element (and, as will be
seen later, for a complete structure) is that it is symmetrical.

An assembly of bar Consider now a system consisting of two bar elements as shown in Fig. 17.4.
elements Using eqn. [17.4] the force—displacement equation for each element may
be written as

Fig. 17.4 f
F3,U3

4 ky 2 z 3 x
466 MECHANICS OF ENGINEERING MATERIALS
ee...
____ EE

If each of these two equations is expanded so that they are in an

equivalent form

Fy ky —k 0) uy
Fy — —k ky 0 U2 and
FP; 0 0) 0 U3

FE; 0 0 0 uj

F, = |0 ko —k u2
FP; 0 —k, ko U3

The forces in the overall system are obtained by adding all the forces at
each node. This may be obtained by adding the matrices to give

FP, Ry —ky 0 u)

F, = |-k; k +k, —ko u2 [17.6]

FP; 0) —k ko U3

or

{F} = [K]{u}
where [K] is the stiffness matrix for the structure, 1.e. the assembly of two
spring elements.
The influence of each term in the structural stiffness matrix may be
visualized as follows. Imagine all the nodes in the structure except node 7 are
restrained so that all the uw; terms (7 #7) are zero. If then w; is given a unit
value, i.e. u; = 1, the force needed at all the other nodes to hold the
displacements to zero at node 7 is Kj.
Thus in the example above, if “; = u3; = 0 and u2 = 1, the forces are

Fy Kyi Kips 0 Ky —k
Po > =| Koy Bos Ko; lp=< kn Pp=< re
P; Kay hao Kas 0 K32 ky
(17.7
Note that if nodes 7 and 7 are not connected by an element, Kj = 0.

This approach to a solution is the basis of the finite element method and
it is illustrated in the following example of a statically indeterminate force
system.

Example 17.1
Three dissimilar materials are friction welded together and placed between rigid end
supports as shown in Fig. 17.5. If forces of 50KN and 100KN are applied as
indicated calculate the movement of the interfaces between the materials and the
forces exerted on the end supports.
 FINITE ELEMENT METHOD 467

Fig. 17.5

Aluminium

Fs Fy,U2 F3,U3
—>
al k 2 3}
a kp k3
(b)

Table 17.1
For aluminium For brass For steel

Area = 400 mm? Area = 200 mm? Area = 70mm?

Length = 280mm Length = 100mm Length = 100mm
E = 70 GN/m? E = 100 GN/m? E = 200 GN/m?

This system may be representeld by a model consisting of three spring

elements as shown:

_ AE, _ 400 x 70 x 103

k1 iv = 100kN/mm
7 > 280
ky = 200kN/mm

k3 = 140kN/mm

The force—deformation relationships for each element may then be

written as

\{n}
Py ss f 100 —100
io a |—100 100

I(r}
Fy ~ . 200 —200
F; i | —200 200

I(r}
P| eee 140 ales!)
Fy = |-140 140

Using the two rules for the formation of the overall stiffness matrix,

Fy 100 —100 0 u|

Ele. | 100 (100 + 200) 0 u?2

je em —200 —140 U3
Fs 0 0 140 U4
468 MECHANICS OF ENGINEERING MATERIALS
a ——————————————————————————

This yields the equations

Fy= 100,— 100uy [17.8]
Fy = —100u; + 300u2 — 200u3 (17.9]
F; = —200uz + 340u3 — 140ug [17.10]
Fy = —140u3 + 140u4 [17.11]
Using the boundary conditions that uw; =u4—0 and letting
F, = —50kN, F; = 100kN then eqns. [17.9] arid [17.10] may be solved
simultaneously to give uw. = 0.048mm and «3 = 0.323mm. Using eqns.
[17.8] and [17.11] then gives F, = —4.8kN and Fy = —45.2kN. These
forces will both act to the left on the elements and to the right on the
supports.
Note that in any finite element or stiffness analysis, sufficient
displacement restraints must be applied to prevent rigid-body motion of
the structure. This requirement may be illustrated by considering a single
element as shown in Fig. 17.3 with a known force P applied at node 2. The
stiffness matrix gives

ft a]tap tes Yn
which corresponds to the two scalar equations

kuy = ku = Fy
—ku, + ku =P [17.13]

It can be seen that

Le [17.14]

i.e. equilibrium requires an equal and opposite force at node 1. The two
equations are now
kuy me kuz =-—Pp

—ku; + ku, = P [17.15]

The second equation is the same as the first and there is no unique solution
for u,; and uz. The determinant of the stiffness matrix is

k —R
|
= le? [17.16]

and it can be shown that the inverse matrix does not exist. The system
cannot therefore be solved to find the displacements. The physical reason
for this is that if one end of the bar is not restrained the structure can ‘float’
back and forth on the «x-axis. Equilibrium may be satisfied, but unless the
displacement of one end of the bar is fixed, the displacement of the other
end cannot be found.
In a one-dimensional analysis, one restraint is sufficient to prevent rigid-
body motion. In a two-dimensional analysis, three restraints are required:
two to prevent x- and y-translation, one to prevent rotation about the z-axis,
Fig. 17.6. In a three-dimensional analysis, six restraints are required to
 FINITE ELEMENT METHOD 469

Fig. 17.6

prevent rigid-body translation and rotation about each of the three co-
ordinate axes.
Failure to prevent rigid-body motion is a very common cause of
difficulty for inexperienced users of commercial finite element packages.
Error or warning messages using phrases such as ‘zero pivot’ or ‘zero
tangent stiffness’ often arise from this cause.
Some analyses, such as Example 17.1, have more than the minimum
number of restraints. It is also possible to have a sufficient number of
restraints, but fail to prevent rigid-body motion. In Fig. 17.7 the two-
dimensional analysis has three restraints. Rotation about the z-axis and
translation in the y-direction are prevented, but translation in the x-
direction is not restrained.

Fig. 17.7

17.3 Analysis of
Although the bar elements considered so far have been collinear this need
frameworks
not be the case. The spring analogy could, for example, be used to
determine the forces and deformation of a framework as shown in Fig. 17.8.

Fig. 17.8

(a) Framework (b) Model using spring elements

However, as the spring elements in the model are at different angles to

one another it is necessary to express the forces and deformations for each
element (calculated for its own /ocal co-ordinate system x, y) to the
deformation of the whole framework relative to the global co-ordinate
system X, Y. For example, consider member 4—5 in Fig. 17.9.
470 MECHANICS OF ENGINEERING MATERIALS

My x
Fig. 17.9
Local co-ordinate
Yt 5 system

k
Element

4
PKG

poe
Global co-ordinate system x

In the general case it is necessary to consider forces and displacements in

both local co-ordinate directions at each node. For a pin-jointed member
there will only be forces in the x-direction so the F,-components will be
zero. Thus eqn. [17.4] may be expanded to the form

Fs, Pale =F, al el ex,

5 soe eeearlsaOteea leOualal
Ey oe \ 'od,, [17.17]
F en eGns0d) (GyBi
Referring to Fig. 17.9 the forces in the local co-ordinate directions may
be related to the forces in the global co-ordinate directions as follows:

FR Px COs.0 fF ys 0

Fy. = —fy, smd + Fy, cos6

Fo, — le C080 Pye

En Fast) by. cos

Using s = sin @ and c = cos @ as a convenient shorthand, these equations

may be expressed in the matrix form

IBer c Ay 0 0) Pe

Sy AGE oe. 5 Ul ao Py,

lige - 0 0 c S Px
Ea GO. its 4 Fy,
or

Fh, = (THFhe [17.18]

where L refers to the local co-ordinate system and G refers to the global co-
ordinate system. Here [7] is called the transformation matrix. A similar
analysis will show that the local displacements are related to the global
displacements by the following equation:

{6}, = [T]{5}e [17.19]

Now, by combining eqns. [17.5], [17.18] and [17.19] we may write

THF} = [KIT oh
 FINITE ELEMENT METHOD 471

Premultiplying each side by oles this becomes

{F}g = [TK UTH}¢ [17.20]

{F}¢ = [K'l6{o}e [17.21]

where [K‘],, is the stiffness matrix for the element in global co-ordinates.
Performing the matrix multiplication [7]|[K*]{7] gives the value of [K‘],
as

C cs ee == 6S:

Ke 5 cs ° a OS: = 17.22

| le a = ae o cs ;

where k = AE/L.
The use of this method of analysing frameworks using spring elements is
illustrated in the following example.

Example 17.2 ; P : ’
Determine the vertical and horizontal displacements at the loading point in the
framework shown in Fig. 17.10. The value of AE for each of the members is 200 MN.

Fig. 17.10

From egn. [17.22] the stiffness matrix for each of the elements in global co-
ordinates will be

25 14.4 sey UNE ae

14.4. %.33° —144 _—8.33
Kile = ws 4A 25 14.4
EAaA, tits, midds 833

38:50) = 385-0
Digits Wak land i, ee
Kle=|_395 0 38.5 0
Om s0n 0, 20

ae Oieaien(h AO)
Be waliOmrs66,72 10% 366.7
[Kle= 0 0 0 0

0 -=667 0 66.7
472 MECHANICS OF ENGINEERING MATERIALS

Teed 2" Sat hogs ee ree ene 5) ae

: Member Length (m) 0 c(=cos@) s(=sin@) k(MN/m)

1-2 6 30 0.866 0.5 33539

2-3 5.196 0 ] 0 38.5
3-1] 3 90 0 ] 66.7

If these matrices are expanded with rows and columns of zeros so that
they are in an equivalent form, they may then be added to give the stiffness
matrix for the whole structure. This will give the following equation:

1G 25 144 =25° —144 0) 0 6x,

Fy, 14.4 15 —14.4 —8.33 0) —66.7 oy,
ibe —25 -144 63.5 144 —38.5 0 6x, 17.23
jo 244.5833 144 9633... 10 ae We ) Se ee
Fx, 0) 0 —38.5 0 38.5 0 bx;
Ey, 0) —66.7 0 0 0 66.7 by,

Recognizing that 6x, = 6x, = dy, = 0 and also Fy, = Fx, = 0 then this
set of simultaneous equations may be solved for dy,, dx, and dy,.

75 14 48:33 by, 0
4 e042) 14.4 ox 7 = 0 [17.24]
= 9.35 144) 833 by, —10*

Hence

Oy —— 0am (i.e. downwards)

dx, = 0.488 mm (i.e. to the right)

by, = = 2.12 mm (i.e. downwards)

The values of the reactions at the supports may now be determined from
eqn. [17.23]. This gives Fy, =17.3kN, Fy, =10kN and Fy, = —17.3kN
which agrees with the values obtained from a simple equilibrium analysis of
the structure.
The forces in each of the elements may be determined by reverting to
the local co-ordinate system for each element. In the local co-ordinate
system of Fig. 17.9, the force f in the element is

f =AEe = —

= k(x, — bs,) [17.25]

for an element connecting nodes 7 and j and k = EA/L for that member.
Using the displacement transformation, eqn. [17.19],

bs, =[c 5]ea [17.26]

BUNGRES Ee VeN TV EOD 473

and
6
0, = |¢ i a [17.27]
By
so that the strain in the element connecting nodes 7 and 7 is

bx,
1 by,
E se lee —secs] we [17.28]

a
and the force is

Ox,

fail
=k |= GunaSu Cans ewbx, 117.29)
:

oy,
For example, for member 1-2, Fig. 17.10, the force is given by

0
=0.15
fia = 33.33[—0.866 —0.5 0.866 0.5] x
0.488
—2.12

=U KIN
Similarly /23 = 17.3kN, and these values agree with the values obtained
by taking a free-body diagram at the loading point.
The framework in this example was deliberately kept simple in order that
the steps in the solution could be illustrated and the calculation performed
manually. For a large plane or three-dimensional structure the individual
steps in the solution are identical to those illustrated but, although the
calculations are straightforward, they are so numerous that they are best left
to a computer. Even on a computer the time taken to obtain a solution can
be relatively long if a large complex structure is being analysed. This is
where it can be beneficial to have some understanding of the nature of the
calculations being performed so that data input can be rationalized to
streamline the solution procedure. For example, one good way to achieve
this in a structural analysis program is to ensure that the difference between
the node numbers on each element is kept to the minimum. This has the
effect of condensing the data into a band along the main diagonal of the
overall stiffness matrix. This then reduces the subsequent computation and
provides a valuable saving in computing time and disk space requirements.
This method can be used on either statically determine or statically
indeterminate problems. For statically determinate problems, either the
method of joints or the method of tension coefficients described in Chapter
1 is slightly more economical, since only the member forces have to be
solved for. However, with the finite element method, the extra computation
time is usually negligible in all but the largest structures, the displacements
are obtained ‘for free’ and general purpose commercial packages can be
used.
474 MECHANICS OF ENGINEERING MATERIALS

Before extending the analysis of the bar element to other more general
elements, it is convenient to express the steps outlined so far in a more
generalized form.

Geometry of In the one-dimensional bar, the displacement pattern may be expressed as a

deformation linear polynomial
u= Q +. 2x [17.30]

where a; and qa are constants which may be determined from the nodal
displacements and geometry of the element. At node 1, x = 0, so
u= uy = Q [17.31]

whilst at node 2, x = L, so
u=u.=A,+Qa2L [17.32]

From this

Qa, = uy [17.33]

and
un — u
a) =— = [17.34]
The resultant variation in displacement over the element is
u— QQ) 4 OK

fe B

Sle
= [N]{u} [17.35]
Here [N] represents what are called the shape functions of the element, which
specify the form of the variation in displacement within the element, which
in this case is linear.
The shape function associated with a particular displacement takes the
value of one at that node and zero at any other node, see Fig. 17.11. The
displacement at any point within an element can be found by multiplying
the shape function matrix (which varies with position) by the nodal
displacements (which are simply numbers). This applies to one-, two- and
three-dimensional elements.
The only strain of interest in the element is

pelle oAos es
 FINITE ELEMENT METHOD 475

Fig. 17.11

= [B]{u} (17.37
The matrix [B] gives the strain at any point due to unit nodal displacement.
Thus if node 1 is displaced by one unit, ¢, = —1/Z (compressive), whilst if
node 2 is displaced by one unit, ¢, = 1/L (tensile).

Stress—strain The stress-strain relationship for a bar in uniaxial tension is just

relationships
o = Ee = [E\[B]{u}
J 2|-4 ;|ea [17.38]
Here the matrix [£] contains only a single value. For two- and three-
dimensional elements, matrices with more elements will be necessary, but
the procedure is just the same.

Equilibrium of forces The final step in the procedure is to relate the forces on each node of the
element to the stresses within it. For the one-dimensional bar

F, = —Ao, F, = Ao, [17.39]

or

vat Galeba dt
“TLt tlt ce
476 MECHANICS OF ENGINEERING MATERIALS

The resulting element stiffness matrix is exactly the same as eqn. [17.4].
Using virtual work arguments, it can be shown! that for any element, the
stiffness matrix is given by

i is[B]" [E][B] dv [17.41]

where V is the volume of the element. Thus for the one-dimensional bar
element
r f-1/L | ee
[17.42]
m= {inletsa
) eae E\|-= +|dV

Since [B] and [F] are constant within the element then for this element, the

w(t a
matrices are just multiplied by the element volume V = AL to give

Pap Sy
[17.43]
Es eS} 1
which is the same as eqn. [17.4].
For higher-order elements, where the terms of [B] vary within the
element, the integration of eqn. [17.41] is usually performed numerically.
The general equation [17.40] can also be used to determine the stiffness
matrix of the two-dimensional framework element. The matrix [£] relating
stress and strain is the same as for the one-dimensional element, but the
matrix [|B] was shown, eqn. [17.28], to be
1
[B] =—[-c¢ -sc s| [17.44]
IL,

Therefore

—=¢

iK]=| i"eaiay =| | ele -s ¢ sa

JV
|

ta aS awa
EEA ies eae
= ib 2 =< 2 oe [17.45]

ee Ss 2

17.4 Analysis of beam

elements Although the stresses and deformation in beams have been extensively
covered in Chapters 6—8, the amount of manual calculation required for
statically indeterminate structures containing multiple members can become
prohibitive. The development of a finite element representing a single beam
will now be presented. Only the simplest one-dimensional beam which lies
on the x-axis and is loaded in the y-direction will be described. Assemblies
of these elements allow one-dimensional beams with multiple sections or
 FINITE ELEMENT METHOD 477

loadings to be rapidly analysed using a computer, and the extension to two-

or three-dimensional frameworks is straightforward.

Geometry of Figure 17.12 shows a beam bent in the xy-plane. The neutral axis undergoes
deformation no strain and planes normal to the neutral axis remain plane. The bending
strain in the beam at any section is, eqn. [7.2b],

dv
[17.46]
Y dx?

Vo

Ce,

When we assemble a series of the elements together we must ensure that

both displacement and slope are compatible between adjacent elements. In
this case two variables, slope and displacement, are required at each end of
the beam. The assumed form of the deflection of the beam must therefore
contain four unknowns, i.e.

v =a, + agx + 03x? + a4x? = [tw x? x3] [17.47]

3
4

The slope of the beam at any point is then

ay
do =a) + 2a3x + 3a4x° 2 = [0 1 2x 3x']
6= — 2) G2 [17.48]
dx 3

O14
Substituting in the boundary conditions at each end of the beam, Le.
ant i) == {0 =O) = Oy

6=6,=a

Ate v= =A +QL+a3l’
+ aL

Qs 7D) = aQ2 + 203L ae 3021’

478 MECHANICS OF ENGINEERING MATERIALS

we obtain four equations in the four unknowns as

Vi 1 0 0 0 4

Ciitlecne lO beteea Ore 0) a2

ed (ma Sy 2 ae oes a3 17.49}
> Og elie esle 4
These equations can be solved either directly or by matrix inversion to give

Q1 ] 0 0 0 Vv]
(oo ae 0 l 0 0 6;
a3 ( |—3/l? -2/L 3/1 —1/L))\m oe
4 2/15 i Oe VTA aos

Fig. 17.13
 FINITE ELEMENT METHOD 479

The equations for a;...a4 can be substituted back into the original
equation for displacement to give

1 0 0 0 Vi

0 1 0) 0 0;
v=[l «x x x]
=3/1? =2/L “3/1? —1/L))
Die” NIE S 2) NTP ees
x2 ra 2 ra re ia x2 ia
Sy ypee 2 Pg ta 2
(ee see ee eee
Vi

x 0;
U2
= [N]{u} (17.51]
oD)
Here [N] is the shape function matrix and {uw} is a vector of nodal
displacements and rotations. The individual shape functions are shown in
Fig. 17.13. They describe the deformed shape of the beam in response to
unit displacement or rotation of the ends of the beam.

Strain displacement The bending strain in the beam can be derived from the displacement using
relations
Ey) oe = SIN] tu)
dv ae

6 12) 4 6, "6 12, 2 6210,

eA
aE (Oe ae oe \ee

= [B]{u} [17.52]
The individual terms in the matrix [B] describe the variation in strain within
the element in response to unit displacement or rotation of the beam ends.

Stress—strain relations The bending stress is obtained directly from the strain as
o = Ee, = [E\[B\{u} (17.53]

Equilibrium The bending moment at either end of the beam can be obtained from

Ai |oy dA [17.54]
A
From the bending moment, the moments acting at each end of the beam can
be determined. Moment equilibrium can then be used to find the forces
acting at the nodes, so that the nodal forces and moments resulting from a
given set of imposed nodal displacements and rotations can be determined.
This defines the element stiffness matrix.
480 MECHANICS OF ENGINEERING MATERIALS

An alternative technique is to use the general equation for the stiffness

matrix, eqn. [17.41],

—6/L? + 12%/13
= 2 —4/L + 6x/L? #
=|) °%) epee
—2/L+6x/I? |
|
6 12% 4 6% 6 2% 2 36% oF
whe ce UNDE eee Te eo ae ere e
1 “6h ek
ei] \6Le4L =6L: 214 (17.55)
ay SD 60) ele SOR op
bbw hl? Meth
The final stiffness equations are

12. Gla 12 6k 0] 0
EI S6L oes, oe Glee 2 aA| jm 117.56
Fou 2 =6l. 22 61 Ve 0»
6h 2 =o Are 6, M>

Fig. 17.14
5 M,

. M2

Qs Q

The nodal forces and nodal moments corresponding to the nodal

displacements and rotations are shown in Fig. 17.14. Note that these
correspond to a positive displacement in the y-direction or a positive
rotation about the z-axis, so that the force and moment contributions from
adjacent elements can be added during element assembly. This is a different
sign convention from that used for shear force and bending moment derived in
Chapter 1.

Example 17.3
Find the slope and deflection at the free end of a cantilever beam, loaded at the free
end as shown in Fig. 17.15
Ve

The boundary conditions for this single-element problem are

ata) UY} = 0, =1()
Fig. 17.15
aby !, Oe. M,= 0
 FINITE ELEMENT METHOD 481

Using v; = 0; = 0, we can eliminate the first two rows and columns from

“Paine ns
the element stiffness matrix leaving

FLW W286 61.) (e210) (WV

The second equation is

—6Lv, + 4176, = 0 [17.58]

or

pee3u2
yea [17.59]
17.

The first equation is

WL
12v = 6L6, = 3u2 = TETt [17.60]

Therefore

er
guy [17.61]
WL?
0,ae = ——
DET. [17.62]
17.62

These are the solutions listed for these problems in Table 7.1.

Example 17.4
Find the deflection of a centrally loaded beam, built-in at both ends as shown in Fig.
17.16. Determine also the bending moments at the fixed ends.

Fig. 17.16

This is an example of a problem where symmetry can be used to reduce the

v,=0 y size of the finite element problem to be solved, although in this case the
6,=0 | reduction is from two elements to one! The stated problem is equivalent to
we analysing only half of the structure, carrying half the total load and with the
symmetry boundary condition. (See Fig. 17.17.)
Tete) At node 1, the boundary conditions are v; = 0; = 0. At node 2, #2 = 0.
2 6,=0 The full stiffness relationship is

Fig. 17.17 De 6h 12), 6 0 0;

EIN 4. 66 217 0 M,
ae = [17.63]
(oN =6L, 12) =6L\') v, W/2
Gia) 22) .=68 417 0 M,
482 MECHANICS OF ENGINEERING MATERIALS

where L = //2. The third row of this matrix equation gives

EI W
ae 12m, = —,.. 120. = [17.64]
te (1/2) :
Therefore

7 =
wi 17.65
oo ao aEn es
which is the same result as is given in Table 7.1. The second row gives

Heer
EI
de 117.66]
Therefore
WL WI
Mi = SS 17.67
4 8 ee

Example 17.5
Determine the deflection of the free end of a cantilever beam subject to a uniformly
distributed load w per unit length as shown in Fig. 17.18.

w/unit length The basic problem here is that external forces and moments can only be
applied to the nodal points defining the element. The solution is to apply
nodal forces and moments which do the same amount of virtual work as the
distributed load over the element!. These ‘kinematically equivalent loads’
can be derived from

(f}= |(Maa
where [N] are the shape functions, g is the pressure distributed over the
element edge and A is the area of the edge.
All commercial packages have facilities for applying distributed loads and
calculating the correct equivalent nodal loads for internal use. These
facilities should be used if at all possible, since the results are sometimes not
obvious.

Fig. 17.19 Di wo

Ea seat I

For a beam, the correct equivalent nodal forces and moments are shown
in Fig. 17.19. Imposing the conditions v; = 6; = 0, we are left with
EI 12a> —6eA ea 2 wL/2
2 |-6L) 407) \6,) ann? 712 [17.68]
 FINITE ELEMENT METHOD 483

which has the inverse matrix solution

2)" a aM AL? (6L wL/2

Go) 9 WE 6 12:) wi? 12

weeta} [17.69]
These are again the same results as are obtained in Table 7.1. The further
development of this type of element to allow the beam to be loaded while
positioned at an angle @ in the xy-plane follows a similar procedure to that
described for the bar element. Both element types can be further extended
to allow their positioning at any orientation in three dimensions.
The beam and bar elements can be combined and extended to include
torsion so that the behaviour of a three-dimensional assembly of long
slender members subject to any combination of bending and torsional
moments, and axial and shear forces, can be analysed in a straightforward
manner. The analysis package only requires the nodal positions, the element
topology (the nodes at either end of the element), the section properties
(cross-sectional area, moments of inertia and torsional constant), material
properties (£), loads and restraints.

17.5 Analysis of
Although the use of a simple linear spring element is a convenient way to
continua
introduce finite element methods, it is quite limited in its application. The
major advantage of the finite element method is its ability to model complex
two-dimensional and three-dimensional solids. In these cases the elements
used may be of the types shown earlier in Fig. 17.2. However, the approach
to a solution is still similar to the method illustrated for the linear spring
element. In essence the solid continuum is modelled by a mesh of plane or
three-dimensional elements which are joined to each other at their node
points. The system of external loads acting on the actual solid must then be
replaced by an equivalent system of forces acting at the node points. The
type and number of elements used can be decided by the analyst. In general
the accuracy of the solution will be greater if the number of elements is
large. However, computer time (and cost) also increases with the number of
elements chosen so it is generally wise only to use a dense concentration of
elements in the critical areas of the solid which are likely to be of particular
interest. Typical examples are shown in Fig. 17.20.

Fig. 17.20

A computer program is then used to obtain the distribution of forces and

displacements in the solid based on the stiffness matrix for the particular
type of element chosen. Modern commercial packages include procedures
484 MECHANICS OF ENGINEERING MATERIALS

for estimating the error arising from a given density of elements and
adaptively refining the mesh where necessary, so that answers of known
accuracy can be obtained.

17.6 Stiffness matrix for a 4: : ; rT

a triangular element Figure 17.21 shows a two-dimensional triangular element typical of that
used in plane stress problems.

fg 72 2=— |

Geometry of In general, two-dimensional displacement patterns may be expressed in

deformation _ terms of two linear polynomiais

w= ay tar + ay [17.70]
U0) poe Bay
At node 1

Uy = Ay + 2%] + 0371
[17.71]
v1 = Gi + Box1 + Bay
Similar expressions may be written for u2, v2, u3 and v3, and if each set of
equations is written in matrix form so that they may be added then we get

uy ] 1 OV 0 0 ) Q)
V\ 0 0 0 ] v1 Vi Q2
u2 ae ] x2 2 0 0 0 QA3

m({ |0 0 0 1 wm »% Bi ee
U3 | %3 «(V3 0 0 i) By

U3 0 0 0 1 X33 3

or in shorthand

{6} = [C]{a}
This may be rearranged to give the coefficients {a} in terms of the
displacements

{a} = [C]"{u} [17.734]

 FINITE ELEMENT METHOD 485

By matrix manipulation of [C], [Gir! may be written as

a\ 0 a2 0 a3 0

bh 0 bh 0 b3 0
0
ae 1 C7} 0 C2 C3 0
[17.736]
al 0m sate Onken nO. as
Oo bp 0) ber0 OD;
0 C| 0 02 0 C3

where

a = (x23 — 8372) >) = G2 — 973) 1 = (23 = 2)

= (ey — e193) 2, = 63 91) = (4 — 23) [17.73c]
a3 =(xy2- 421) b=Q1-y2) 03 = (x2 — 41)
2a = (a, + a2 + a3) = 2(area of element)

Strain—displacement The next step is to get an expression for the strains in the element as a
relations function of its geometry and nodal displacements. The strains in the element
may be determined from the displacements as described in Chapter 13 as

Ex
ae Bye Z

Ov
6 ih =A,
Oy = 03

Ou Ov
He sa, Bo + 03 (17.74]

ey Om a
fy ¢=|0 0000 110 [17.75]
“5 0; Ost 0-110 :
ie Bo
B3
or in shorthand form

{e} = [A]{o} [17.76]

Then combining egns. [17.73a] and [17.76]
=
{e} = [A][C] {4} [17.77]
So, multiplying the respective matrices (see Appendix B) gives,

{e} = [B]{u}
where

i W th OW B
[B] => — 0 Cy 0 02 0 C3 [17.78]
Ge Py cy 8p 63” OS
486 MECHANICS OF ENGINEERING MATERIALS

It may be seen that the terms in this matrix are known since they are a
function of the co-ordinates of the nodes.
Furthermore all the terms of the matrix are constants, so the strains will
all be constant within the element. This element is therefore known as the
constant strain triangle. If the stresses vary within the part being analysed
the true stresses will be approximated as a series of piecewise constant values
across a given section.

Stress—strain For a two-dimensional plane stress element, the stresses and strains are
relationships related by the following equations:
iE
dn =a] (Ey = Vey)

E
Oy = (=a) (ey + vex)
tree
Key = 2(1 se v) Yay

In matrix form these equations may be expressed as

on E pe 0 Ey
Oa eran ee 1 0 Ey [17.79]
Txy 0 0 ‘(1 = v) Yxy

or in shorthand form

{o} = [D]{e} [17.80]

where [D] expresses the material properties of the element.

Equilibrium of forces The effect of the external stress system on the triangular element is as shown
in Fig. 17.22.

yt
Fig. 17.22

0 x

Assuming the thickness of the element to be /, the force on the left-hand

side of the element is —o,h(y3 — 1). This force is assumed to be shared
equally by nodes 1 and 3, so making the earlier substitution of
b, = (y3 — 1) then
[17.81]
 FINITE ELEMENT METHOD 487

Note, however, that this is just one component of the forces at nodes 1
and 3. There will be other components due to the direct stress on the right-
hand side of the element and the shear stresses on the top and bottom of the
element.
Considering the right-hand side of the element, over nodes 1-2 the force
to be shared is 0,h(y2 — y). Hence

F,, = —4Oshbs [17.82]

The force at node 1 due to the shear stresses will comprise contributions
from face 1—3 and face 1-2.

Py = STO 1 03) — 5Tye [17.83]

Combining eqns. [17.81], [17.82] and [17.83] gives

Fy, = 5A(bi0y + C1 Try) [17.84]

Similar expressions may be obtained for the y-direction at node | and the
x- and y-directions at nodes 2 and 3. The overall interrelationships between
nodal forces and applied stresses may then be written in matrix form.

We by 0 Cl

Ft a| aloe
ips = 2 0
Omen
eS hy
|
ue [17.85]

1a b 0 «6 4
Py, 0 3 b3

Comparing eqn. [17.85] with [17.78], it can be seen that this is equivalent to

{F} = ah[B]" {o} [17.86]

We are now in a position to determine the stiffness matrix for the triangular
element since the matrices |B] and [D] are available from eqns. [17.78] and
[17.79]:
[K*] = ah{B]" (D][B] [17.87]
where V = ah is the volume of the element. Since all the terms in [B] and
[D] are constant within the element, this is equivalent to the general eqn.
17.41]
[K‘] = | er"oeiar [17.88]
The stiffness matrix for a triangular element is thus a 6 x 6 matrix. It can be
appreciated that when a structure is modelled by a large number of
triangular elements, the global stiffness matrix becomes extremely large and
can only be handled by a computer. The following example illustrates the
response of a single element to an imposed displacement.

Example 17.6
A single element of length /, width w as shown in Fig. 17.23 is fixed at nodes 1 and 3
while node 2 is displaced to the right by an amount 6. Find the strain and stress in
the material.
488 MECHANICS OF ENGINEERING MATERIALS

Fig. 17.23 3

The strains can be determined from the nodal displacements, which are all
known in this case, and the matrix [B] defined by egns. [17.73c] and [17.78]
as
0
a ,/-2 9 » 00 Oj],0 6/l
Spe pe Oe le OOOH) eee
50 se eigh oan) gee aa et ae ak 0
. 0
[17.89]

There is a strain of 6// in the horizontal direction. Note that there is no

lateral contraction €, in the material arising from the extension in the x-
direction since nodes | and 3 are fixed. Equally there is no shear strain, since
the element edges 1—3 and 1-2 which were initially perpendicular remain so.
The corresponding stress can be obtained by substituting in eqn. [17.79] as

Ox iB je 22 0) Ex
AD iaaarenray 1 0 Ey
Tey 00 (-v/2} lr

—_ a5 17.90
ery ‘ Be

17.7 Effect of mesh

Since the simple triangular element studied here can only model a constant
density
state of stress, we cannot expect to achieve accurate answers if the true stress
is varying significantly over the area covered by the element. This effect is
illustrated in the analysis of a rectangular region in pure bending in Fig.
17.24. These analyses were performed with a commercial finite element
package. The shading within each element is proportional to the stress
component oy.
In the first analysis, Fig. 17.24(a), only two elements were used through
the thickness of the beam and large differences between the stress in
adjacent elements can be seen. For a three-noded triangle, the ‘stress jump’
between adjacent elements is a measure of the error in the analysis, though
improved estimates of the correct stress at a given node can be determined
by averaging the stress at all the elements connected to it.
 FINITE. ELEMENT METHOD 489

As the number of elements is increased, Fig. 17.24, the size of the stress
discontinuities decreases and a more accurate estimate is obtained, though a
very large number of elements are required for reasonable accuracy.

17.8 Other continuum

Owing to these accuracy limitations, the constant strain triangle element is
elements
not in very common use. In practice, elements which can model a quadratic
variation in displacement or a linear variation in stress are usually preferred.
These may be triangular or quadrilateral in shape for plane stress analyses. A
further advantage of this type of element is that it can have curved sides to
approximate curved boundaries. The results for a coarse mesh with these
elements gives exact answers for the pure bending case, Fig. 17.25.

Fig. 17.25
490 MECHANICS OF ENGINEERING MATERIALS

Plane stress elements such as the constant strain triangle allow the analysis
of loading in the plane of a thin sheet of material. Plane stress elements can
be converted to analyse plane strain or axi-symmetric problems with minimal
effort. If a thin sheet is loaded perpendicular to its plane then p/ate elements
must be used. The development of these is analogous to the beam element
described earlier in the chapter. If the thin sheet of material is curved then
shell elements must be used.
If the region of material to be analysed is long and slender then a one-
dimensional bar or beam element will normally be used, since the stress at
every point can be estimated by solving a small number of equations for only
two nodes, one at either end of the member. This allows a very rapid and
economical computer solution.
If the component to be analysed is not either long and slender or a thin
sheet, then three-dimensional elements such as hexahedron or tetrahedron
must be used, Fig. 17.2. The resulting analysis will be much more time
consuming. A single brick element capable of modelling a quadratic
variation in displacement has 20 nodes with three possible displacements in
x-, y- and z-directions at each node. Thus 60 equations must be solved to
determine the behaviour of a single element. Using three-dimensional
elements for thin sheets or long slender bars is therefore uneconomic and
can result in numerical problems if the element is too distorted. However,
with these elements, accurate predictions can be obtained for the stress
distribution in very complex geometries which would be impossible to
analyse in any other way.

Summary
In the finite element method, the structure to be analysed is subdivided into
a mesh of finite-sized elements of simple shape. Within each element, the
variation of displacement is assumed to be determined by simple polynomial
shape functions and nodal displacements. Equations for the strains and
stresses are then developed in terms of the unknown nodal displacements.
From this, the equations of equilibrium are assembled in a matrix form
which can be easily programmed on a computer. After applying the
appropriate boundary conditions, the nodal displacements are found by
solving the matrix stiffness equation. Once the nodal displacements are
known, element strains and stresses can be calculated.
A range of line, surface and solid element types are available for
efficiently solving problems in long slender members, thin sheets and
chunky solids respectively. Different types of line and surface element are
used for in-plane and bending loads.
The power and generality of this method is demonstrated in excellent
commercial packages, but the analyst or designer needs to have a basic
understanding of the principles of the technique in order to apply it safely
and with confidence.

Reference
Zienckiewicz, O. C. (1989) The Finite Element Vol. 1, Basic formulation
and linear problems, 4th edition, McGraw-Hill, New York.

Bibliography
Bathe, K. J. [1982] Finite element procedures in engineering analysis, Prentice
Hall, London.
 FINITE ELEMENT METHOD 491

Bickford, W. B. [1994] A First Course in the Finite Element Method, 2nd

edition, Irwin, Boston.
Fagan, M. J. (1992) Finite Element Analysis. Theory and Practice, Longman
Scientific & Technical, Harlow.
NAFEMS (1986) 4 Finite Element Primer, Glasgow.

Problems 17.1 (a) Show that a finite element for analysing torsion of a straight bar
(Fig. 17.26) has the stiffness matrix relationship

Chee Ors eel

>) EAs pel wml |B
a. Ty where G = shear modulus, 7 = polar moment of inertia, 1 =
bar length, 7 = torque and 0 = angle of twist.
Ts (b) Form an assembly of two of these bars in series and use this to
check the results of Examples 5.4 and 5.5.
Fig. 1726 17.2 ‘Three dissimilar cylindrical rods are jointed together as indicated in
Fig. 17.27. The brass cylinder is fixed to the rigid support at the top
and the aluminium cylinder may be regarded as simply-supported
with a point load at mid-span. Using the stiffness matrix approach,
calculate the movement at each interface and the forces at the rigid
supports when 25kN and 100kN forces are applied as indicated.
E, = 100 GN/m’, E, = 200 GN/m’, E, = 70 GN/m’.

Fig. 17.27

100 mm
(d= 16 mm)

Steel
(d= 10 mm) 80 mm

Aluminium
(d = 25 mm)

® tas
100 mm

17.3 For the simple pin-jointed framework shown in Fig. 17.28 use the
stiffness matrix approach to calculate the vertical deflection at the
20 kN load and the forces in each of the members. For each member
the product AE = 240 MN.
17.4. Use the method of stiffness matrices to determine the vertical
deflection at the loading point in the plane pin-jointed framework
shown in Fig. 17.29. The product of cross-sectional area and
modulus for each member is 400 MN.
492 MECHANICS OF ENGINEERING MATERIALS
mS

Fig. 17.28

20 KN

1.0m

13m

17.5 (a) Program the shape functions of Eqn. [17.51] into a spreadsheet
so that they are calculated at an arbitrary point x along the
length of a beam. Calculate the displacement at + when given
the nodal displacements and rotations.
(6) Plot a graph of the shape functions and displacement wu at 10
points along the length of the beam.
(c) Insert the nodal displacements and rotations resulting from a
load W on the end of the beam as given in Table 7.1. View the
graph of the deformed shape. Assume W = E=J=L=1.
17.6 Use the stiffness matrix of a single beam finite element to determine
the deflection of a simply-supported beam to (a) a concentrated load
W applied to the centre (Fig. 17.30(a)), (2) a distributed load of m/
unit length (Fig. 17.30(d)).
fig. 1730 17.7 The cantilever beam shown in Fig. 17.31 is to be represented by
three triangular finite elements. Construct the master stiffness
matrix for the beam.

Fig. 17.31
CHAPTER
Tension, Compression,
Torsion and Hardness

For the design engineer it is just as essential to have an understanding of material

behaviour, to aid in the appropriate selection of type and condition, as it is to be able
to calculate the stresses and strains which will have to be withstood by the material
of a component or structure. This textbook is primarily devoted to the latter aspect
and there are many excellent texts which deal fully with the engineering properties of
materials. These remaining four chapters are intended therefore to provide a
sufficient introduction to enable the reader to appreciate the significance of material
response to stress and environmental conditions.
This first chapter concentrates on the principal laboratory tests which are used to
characterize materials from which British Standards (B.S.), other international
standards and manufacturers’ specifications of material properties are derived.

18.1 Stress—strain
The principal concepts of elastic and plastic uniaxial tensile stress—strain
response in a uniaxial behaviour were introduced in Chapter 3 and it is therefore only necessary
tension test briefly to reiterate certain key aspects.
Tensile testing of Figure 18.1(a) shows the typical shape of a flat or round bar specimen for
metals tension testing (B.S. EN10002-1:1990). The enlarged ends are for gripping
in the jaws of a testing machine and the reduced parallel portion contains the
gauge length, across the ends of which is mounted an extensometer. This is a
sensitive instrument which measures the very small longitudinal deforma-
tions that occur in the elastic range.
If a second instrument is used to measure the lateral contraction of the
gauge length as the load is increased then the ratio of lateral to longitudinal
strains can be determined, which is Poisson’s ratio v.
Figure 18.1(4) is a stress-strain graph for the elastic range which is
derived from the loads applied by the testing machine and the extensions of

Fig. 18.1
Stress
Load
eae ae) aan
P

Gauge
length
Slope E

0) Strain
Extension
(original gauge eran)

(a) Tensile specimens (b) Elastic stress-strain curve

494 MECHANICS OF ENGINEERING MATERIALS

Fig. 18.2 Nominal and true tensile 9 2

¢ o ;
stress-strain curves @ / ie
True / 2)
z

ae a Nominal

Uh
Vv 1
F
: | Strain hardening | Necking
(<>
i

0) .
Strain 16) Strain

(a) (b)

the gauge length measured by the extensometer. For metals it is a linear

relationship, obeying Hooke’s law, the slope of which is the Young’s
modulus of elasticity E. The end point of linearity is termed the limit of
proportionality P, and a closely adjacent point is the elastic limit L.
When straining of a ductile metal is continued beyond the elastic limit,
yielding commences and the material is in the plastic range. There are two
characteristically different types of transition from elastic to plastic
behaviour. These are illustrated in Fig. 18.2(a) and (4): the first is
principally found in low- and medium-carbon steels and the second in alloy
steels and non-ferrous metals. In the former, the point Yz,, at which there is
a sudden drop in load with further strain, is termed the upper yield point.
This is followed by Y,, the lower yield point, from which there is a marked
extension at almost constant load.
The upper yield point of low- and medium-carbon steels is a complex
phenomenon, which is a function of strain rate, temperature, type of testing
machine and geometry of the specimen. In some cases the upper yield point
does not appear and the curve departs from the elastic range immediately
into the horizontal lower-yield range. The latter, however, may be regarded
as a material property and is used as the yield point stress for design
purposes. It is this rather unique shape of stress—strain curve to which one
approximates when using an ideal elastic-plastic relationship in plastic
bending or torsion theory as in Chapter 15, since mild steel is widely used
for structural work.
Many materials, particularly the light alloys, do not exhibit a clearly
defined elastic limit, limit of proportionality or yield point, and several
methods of indicating these stresses are in use. The most widely used is that
involving a measure of permanent set and is called the offset method.
Figure 18.3 shows the stress-strain relationship for a material stressed
beyond the limit of proportionality and then unloaded. The slope during the
unloading stage is practically similar to that during the elastic range of
loading. The stress for any given amount of inelastic deformation is easily
obtained from the stress-strain diagram. The amount wx is set off on the
strain or extension axis and a line AB is drawn parallel to the straight line
portion of the loading curve, and the intersection at B gives the stress o, for
a permanent set of +% strain.
 TENSION, COMPRESSION, TORSION AND HARDNESS 495

Fig. 18.3 Determination of proof Stress

stress

0 iA Strain

The elastic limit is taken to be that stress at which there is a permanent

set of 0.02%; this is generally called the 0.02% elastic limit. The proof stress
is that stress at which there is a permanent set of 0.1% and is called the
0.1% proof stress. The yield stress is that stress at which there is a
permanent set of 0.2% of the gauge length.
On a 50mm gauge length, the extensions corresponding to the above
stresses are 0.01, 0.05 and 0.1mm respectively, and thus extensometer
measurements are required for the accurate production of the load—
extension diagram.
Once the initial yield region has been passed in Fig. 18.2(a) and (), the
stress—strain curves have a common form. An increasing stress is required to
cause continued straining, and this behaviour is known as work hardening or
strain hardening and the metal does in fact become harder.
The property of work hardening is quite important and will be discussed
further later. The specimen has a limit to which it can be work hardened
uniformly in a simple tension test, and this is reached when the slope of the
nominal stress-strain curve becomes zero as at T, in Fig. 18.2(a) and (0).
This point is known as the tensile strength (in the past, the ultimate tensile
strength U.T-.S.) and is given by the maximum load carried by the specimen
divided by the original cross-sectional area. This is a very important
quantity as it is sometimes used in design in conjunction with a safety factor,
and is always quoted when comparing metals and describing their
mechanical properties. It is also an approximate guide to hardness and
fatigue strength.
Up to the point T, the parallel gauge length of the specimen has reduced
in cross-section quite uniformly; however, at or close to the tensile strength
T one section is slightly weaker due to inhomogeneity than the rest and
begins to thin more rapidly, forming a waist or neck in the gauge length.
Further extension is now concentrated in the neck and a reducing load is
required for continued straining, and thus the nominal stress—strain curve
also falls off, until fracture occurs at F.
The lower stress—strain curves in Fig. 18.2 were based on nominal values
of stress, i.e. the load at all points was divided by the original cross-sectional
area. This is a convenient arrangement for most practical purposes and is the
generally adopted procedure. However, as the specimen is strained, so the
cross-sectional area reduces, and hence the true stress is higher than the
nominal stress. The true stress is obtained by dividing the load by the
current area of the specimen corresponding to that load. The current area of
496 MECHANICS OF ENGINEERING MATERIALS

the bar may be obtained by direct measurement at various stages during the
test or by calculation.
The shape of the true stress curve is shown in Fig. 18.2. It is only strictly
valid up to the point where necking commences, since the change in
geometry at the neck sets up a complex stress system which cannot be
determined simply from the load divided by the area of the neck.
The foregoing has dealt principally with the property of ‘strength’ or
stress; another property which is of almost equal importance is that of
ductility, or the ability of a material to withstand plastic deformation. In a
tensile test this is expressed in two ways, by the percentage elongation of the
gauge length after fracture and by the percentage reduction in cross-
sectional area referred to the neck or minimum section at fracture.
The second quantity is expressed algebraically as

Ao — Ar
Reduction in area = —— x 100%
0
where Ao and A; are the original and fracture areas respectively.
Elongation is the increase in the gauge length divided by the original
gauge length, or algebraically

Elongation = aa x 100%
Lo
where Lo and Ly are the original and final gauge lengths respectively.
Elongation is only partly a material property since it is also dependent on the
geometrical form of the test piece.
In cases where the specimen necks, the distribution of strain over the
gauge length of a specimen can vary considerably from one metal to another,
as illustrated in Fig. 18.4. This will depend on the grain size and
microstructure.

Strain
Fig. 18.4 Strain distribution along
a tensile test piece
Alloy steel

Mild steel
— Se Cast brass

The elongation at fracture includes the necked region, and this is

relatively independent of original gauge length and is more a function of the
material and shape of cross-section.
Two basic types of fracture can be obtained in tension, depending on the
material, temperature, strain rate, etc., and these are termed brittle and
ductile. The main features of the former are that there is little or no plastic
deformation, the plane of fracture is normal to the tensile stress and
separation of the crystal structure occurs. In the second type, ductile failure
is preceded by a considerable amount of plastic deformation and fracture is
by shear or sliding of the crystal structure on microscopic or general planes
 TENSION, COMPRESSION, TORSION AND HARDNESS 497

at about 45° to the tensile stress. Probably the most well-known type of
failure is called the cup and cone in the cylindrical bar.
A cylindrical bar, even though relatively ductile, may not produce a
sufficient neck to set up marked triaxiality of stress, and failure is found to
occur on a single shear plane right across the specimen. When the two parts
of a fractured flat bar are placed together it is seen that there is a gap in the
middle region, showing the initiation of fracture there, while the outer
regions continued to extend.

18.2 Stress—strain
: ; The mechanical properties of a ductile metal are generally obtained from a
response in a compression
tension test. However, compression behaviour is of interest in the metal-
test forming industry, since most processes, rolling, forging, etc., involve
compressive deformations of the metal, and also often of the forming
equipment.
In compression an elastic range is exhibited as in tension and the elastic
modulus, proportional limit and yield point or proof stresses have closely
corresponding values for the two types of deformation. The real problem
arises in a compression test when the metal enters the plastic range. The test
piece has to be relatively short (D/L > 4) to avoid the possibility of
instability and buckling. The axial compression is accompanied by lateral
expansion, but this is restrained at the ends of the specimen owing to the
friction between the machine platens and the end faces, and consequently on
a short specimen marked barrelling occurs as in Fig. 18.5. This causes a
non-uniformity of stress distribution, and conical sections of material at each
end are strained and hardened to a lesser degree than the central region. The
effect on the load—compression curve, after the smaller values of plastic
strain have been achieved, is a fairly rapid rise in the load required to
overcome friction arid cause further compression.

Fig. 18.5 Deformation during a

compression test

Ne)

Initially Ideal compression Barrelling, showing

less deformed end
cones

Owing to the barrelling effect, only an average stress can be computed

from the load—compression curve, based on an average area determined
from considerations of constant volume.
Various methods have been attempted to overcome the effects of
barrelling, none of which is completely successful. The most satisfactory
appears to be the technique of using several cylinders of the same metal
having different diameter-to-length ratios. Incremental compression tests
are conducted on the set of cylinders at a series of loads of increasing
magnitude, measuring the strain for each of the cylinders at each load.
Extrapolation of curves of D/L against strain with load as parameter to a
value of D/L = 0, representing an infinitely long specimen where barrelling
would be negligible, enables the true compressive stress—strain curve to be
determined, Fig. 18.6. Failure of a ductile metal in compression only occurs
owing to excessive barrelling causing axial splitting around the periphery.
498 MECHANICS OF ENGINEERING MATERIALS

72)
Fig. 18.6 Compressive stress— 1p)

s©
strain curves for various diameter n

length ratios egB aA

re
5
Qa D/L=0
=
[=
je)
rs)

0 Compressive strain

For brittle materials, such as flake cast iron, concrete, etc., which would
not normally be used in tension, the compression test is used to give
quantitative mechanical properties. Although end friction still occurs, which
affects the stress values somewhat, owing to the absence of ductility in these
materials the barrelling condition is barely achieved. The examples of failure
of cylindrical specimens shown in Fig. 18.7 illustrate that fracture takes
place on planes of maximum shear stress.

Fig. 18.7 Modes of compression

failure in various materials

Concrete Flake iron Timber

18.3 Stress—strain
The usual method of obtaining a relationship between shear stress and shear
response in a torsion test
strain for a material is by means of a torsion test. This may be conducted on
a circular-section solid or tubular bar. By applying a torque to each end of
the test piece by a testing machine and measuring the angular twist over a
specified gauge length, a torque—twist diagram can be plotted. This is the
equivalent in torsion to the load—extension diagram in tension. It was shown
theoretically in Chapter 14 that, under elastic conditions in torsion, the
applied torque is proportional to the angle of twist on the assumption that
shear stress is proportional to shear strain. This is found to be true
experimentally, and the linear torque—twist relationship obtained enables the
shear or rigidity modulus to be determined, since

where the symbols are as specified in Chapter 5.

The torsion test is not like the tension or compression tests in which the
stress is uniform across the section of the specimen. In torsion there is a
stress gradient across the cross-section, and hence at the end of the elastic
range yielding commences in the outer fibres first while the core is still
elastic, whereas in the direct stress tests yielding occurs relatively evenly
throughout the material. With continued twisting into the plastic range,
more and more of the cross-section yields until there is penetration to the
 TENSION, COMPRESSION, TORSION AND HARDNESS 499

Fig. 18.8 T

Yield

axis of the bar (see section 15.3). The torque—twist diagram, Fig. 18.8,
appears of much the same form as a load—extension diagram, and work
hardening will occur at a gradually decreasing rate as straining proceeds, but
of course there is no fall-off in the curve, as in tension, since necking cannot
take place. In fact, ductile metals can absorb extremely high values of shear
strain (200%) before failure occurs.
Although the shear-stress/shear-strain relationship can be determined
easily in the elastic range, difficulties are introduced for a solid bar in the
plastic range owing to the stress variation mentioned above. One solution to
this problem is to conduct the torsion test on a thin-walled tubular specimen
in which the shear stress in the plastic range may be assumed to be constant
through the wall thickness, and is given by
iP
A i ==

Qrr2t

where ¢ is the wall thickness and r is the mean radius. Shear strain is
obtained in the plastic range from the same assumptions that apply in the
elastic range; hence

(The tangent of the angle y must be used for large strains.)

The limitation on the torsion test of a thin-walled tube is the possibility
of elastic or plastic instability. If the full shear-stress/shear-strain graph is
required in the plastic range, there are two possible approaches. One is the
construction developed by Nadai. This derives the stress-strain curve from
the torque—twist curve based on a test on a solid bar. The other method
utilizes the relationship between yield stress in simple tension and that in
pure shear as derived from the von Mises criterion, eqn [12.15].
Putting 07 = 0 and o; = —o3 =7 for pure shear we have

or

Hence the yield stress in torsion is 0.577 times the yield stress in simple
tension. It can also be shown that increments of plastic shear strain are equal
500 MECHANICS OF ENGINEERING MATERIALS

to \/3 times the increments of plastic tensile strain for a material so that it is
possible to construct the plastic shear stress-strain curve from the plastic
tensile stress-strain curve.
Fracture in torsion for ductile metals generally occurs in the plane of
maximum shear stress perpendicular to the axis of the bar, whereas for
brittle materials failure occurs along a 45° helix to the axis of the bar owing
to tensile stress across that plane.

18.4 Plastic overstrain

The behaviour of materials in their plastic range is of particular interest in
and hysteresis
relation to the analytical treatment in Chapter 15 of plasticity of engineering
components.
If a metal is taken into its plastic range in tension, compression or
torsion, and at some point the load is removed, then the unloading line,
although having slight curvature, approximates to the slope of the originai
elastic range. On reapplication of load the line diverges only slightly from
that for unloading, and yielding occurs only when the load has reached the
previous point of commencement of unloading. This effect is shown
diagrammatically in Fig. 18.9 and is known as overstraiming. It should be
emphasized that this word does not mean damage to the static properties of
the metal although it may convey that impression.
Overstrain is in fact a very useful means of obtaining a higher yield stress
as shown in Fig. 18.9. Instances of the usefulness of overstraining on
components are mentioned in Chapter 15.

Stress
Fig. 18.9 Hysteresis loops during
Fracture
overstrain
~
Continuous
loading
curve

2 03 Strain

In Fig. 18.9 the ‘loops’ formed by the unloading and reloading lines
during overstrain are caused by mechanical hysteresis, i.e. there is a lag
between stress relaxation and strain recovery, and similarly on reloading. In
unidirectional loading (tension only) hysteresis loops are generally quite
narrow and in some metals, virtually non-existent. Hysteresis is also the
term used to describe the loop obtained when reversed loading is conducted
on a material, 1.e. through yield in tension followed by compression or vice
versa. Figure 18.10 shows the form of a tension—compression loop.

18.5. Hardness
measurement Hardness of materials is a concept which is not directly employed in
engineering design as, for example, yield stress may be. However, it is a
‘property’ which is intimately related to strength and ductility, as discussed
 TENSION, COMPRESSION, TORSION AND HARDNESS 501

Fig. 1810 Hysteresis loop for

reversed loading

Strain

in the next section, and in that context it gives valuable quality-control

information.
The term hardness of a material may be defined in several different ways.
These are principally in relation to the resistance to permanent deformation
such as indentation, abrasion, scratching and machining. Although the last
three are of importance in certain circumstances, they have a limited
application in practice, and therefore discussion will be restricted to
hardness as measured by resistance to indentation. One of the first recorded
tests of this type was made by de Réeaumur (1722) in which a piece of
material was indented by a tool made of the same material and the volume of
the resulting indentation measured. Since then there have been several
variations of the principle of this type of test used and widely adopted in
engineering practice. The most common are the Brinell, Vickers and
Rockwell methods and these are discussed in the following sections.

Brinell method Brinell published the details of the indentation hardness test he devised in
1901. The principle involved is that a hardened steel ball is pressed under a
specified load into the surface of the metal being tested (B.S. 240: 1991).
The hardness, which is quoted as a number, is then defined as
load applied to indenter in kg
Hardness number = - oe A
contact area of indentation in mm
or Brinell hardness number (B.H.N.) = P/A. It is noted that the number
has in fact units of pressure. The contact area A is given by

A= T7Dh

=" p- Vin -#)

where h = depth of impression in mm, D = ball diameter in mm and d =
surface projected diameter of impression in mm.
If the hardness number of a metal is determined at several different
values of load it is found that the number is not constant. The results give all
or part of a curve of the form shown in Fig. 18.11 depending on the
502 MECHANICS OF ENGINEERING MATERIALS
ee

Fig. 1811 Variation of hardness

with load in the Brinell test

hardness
Brinell
Geometric ——~ \
effect

fon Work-hardening
influence

Load

condition of the material. The curve is attributed to two separate effects.

The rising portion with increase in load is caused by the non-proportional
effect of work hardening on the size of the impression. Thus a soft metal will
show a marked apparent rise in hardness while a heavily cold-worked
material will show none.
The falling part of the curve in Fig. 18.11 with increasing indenter load
is caused by geometrical dissimilarity between the spherical areas of
successive impressions. This feature is very important and is worthy of
further analysis. Consideration of Fig. 18.12 shows that similarity can only
be obtained for different loads if different ball sizes are used, since the total
angle subtended by the centre of a ball and the indentation must be equal in
each case. Hence the condition for similarity is that

aq, a
— -+ — = constant
D, Dy,

Fig. 18.12 Indentation geometry in

the Brinell test

For a given angle of indentation the mean pressure is P/ ind’, but since
d = constant x D it follows that for similarity P/D? = constant.
The highest value on the curve of Fig. 18.11 is known as the optimum
hardness number and is the figure quoted for a material when hardness is
 TENSION, COMPRESSION, TORSION AND HARDNESS 503

Table 18.1 ‘Typical hardness values

Tensile
Hardness strength,
Material Condition Brinell Vickers (MN/m/?)

Pure aluminium Annealed 25 (25) 54

Cold rolled (hard) 55 (55) 185
Duralumin alloy (3.5-4.5
Cu, 0.4-0.8 Mg, 0.4-0.7
Mn, 0.7 Si) Solution and
precipitation treated 120 (120) 433
6% Al-Zn, 4% Mg alloy Solution and
precipitation treated 180 (181) 587
Pure copper Annealed a2 (42) 221
Cold rolled (hard) 119 (119) 371
Brass (60-40) Cold drawn 178 (179) 659
Mild steel (0.19 C) Annealed 127 (127) 45]
Cold rolled (192) 595
Si-Mn spring steel Quenched and
tempered (415) 435 1180
ACORN 5 901Crs OSI on
steel Quenched and
tempered (434) 460 1640
Ball-bearing steel - — 700 —
Tungsten carbide Sintered — 1200 -

Note: Equivalent values are in parentheses.

required. Brinell obtained the optimum hardness for steels using a 10 mm

diameter ball at a load of 3000kg, and these conditions have become a
British Standard for hardness testing. Using the above values it is seen that
P/D*’ = 30, and it is therefore possible to obtain comparable hardness
numbers on a metal for different sizes of ball if the load is chosen to satisfy
the above relationship. For softer metals and thin sheet it is found that other
values are required for the constant equal to P/D* in order to give optimum
hardness. Values of P/D* = 10 (non-ferrous alloys), 5 (copper, aluminium)
and | (lead, tin) have been adopted as standard. The peak of the hardness
curve generally occurs for a value of d/D between 0.25 and 0.5 and this fact
can be used to assist in choosing the correct value of P/D?. It is important
when quoting a hardness number to state the conditions used, ball size, load,
etc. It is also essential to space successive impressions adequately and keep
them clear of the edge of the material, owing to the plastic deformation
caused in the area around the indentation (‘ridging’ for hard metals,
‘sinking’ for soft metals).
The size of the Brinell indentation is such that the test is generally
employed for checking raw stock or unmachined components rather than
finished products. Typical hardness values for various materials are given in
Table 18.1.
Another very useful feature of the Brinell method is that an empirical
relationship has been found to exist between hardness number and tensile
504 MECHANICS OF ENGINEERING MATERIALS
Neen
eee. eee eeee eee eee eee eee eee ee

strength for steels. Thus, Kx B.H.N. (kg/ mm’) = tensile strength (MN/
m’), where K lies between 3.4 and 3.9 for the majority of steels. Hence the
Brinell test can be used to get an approximate value for the tensile strength
of a metal. This can be useful as a non-destructive method for checking if
the correct heat treatment has been carried out or to determine the
properties of a failed component.

Vickers method This test was devised about 1920 and employs a square-based diamond
pyramid as the indenting tool (B.S. 427: 1990). The angle between opposite
faces of the pyramid is 136° and this was chosen so that close correlation can
be obtained between Vickers and optimium Brinell hardness numbers. The
angle of 136° corresponds to the geometry of an impression given by a d/D
ratio of 0.375.
In the Vickers test, hardness number is defined in the same way as for
Brinell, i.e. indenter load, kg, divided by the contact area of the impression,
mm. If / is the average length of the diagonal of the impression, Fig. 18.13,
the contact area is given by

UE P
2sin}(136) 1.854

Fig. 18.13

Therefore the Vickers pyramid number

P

VENTS 1.8545

There are two features of this test which are essentially different and
advantageous over the Brinell method. Firstly, there is geometrical similarity
between impressions under different indenter loads, and hardness number is
virtually independent of load as shown in Fig. 18.14, except at very low
loads where there is often a higher hardness owing to a ‘skin’ effect on the
test piece. The standard loads recommended in B.S. 427 are 1, 2.5, 5, 10, 20,
30, 50 and 100kg.
The second advantage of the Vickers test is that the upper limit of
hardness number is controlled by the diamond, therefore allowing values up
to 1500 to be determined, which is far in excess of that possible with the
steel ball in the Brinell test.
The extremely small size of the impression necessitates a very good
surface finish on the test sample, but means that it is advantageous in
 TENSION, COMPRESSION, TORSION AND HARDNESS 505

Fig. 1814 Variation of hardness

with load in the Vickers test

hardness
Vickers

Load

checking the hardness of finished components without leaving a noticeable

mark.

Rockwell method This test was introduced in the U.S.A. at about the same time as the Vickers
test in England. It is quite popular as it has a wide range of versatility, is
rapid and useful for finished parts.
Two types of penetrator are employed for different purposes, a diamond
cone with rounded point for hard metals and a xin (1.6mm) diameter
hardened steel ball for metals of medium and lower hardness values.
In this test hardness is defined in terms of the depth of the impression
rather than the area, and the hardness number is read directly from an

Fig. 1815 Load application in the

Rockwell test

Reference line representing zero hardness

indicator on the machine having three scales, A, B and C.

The procedure for applying load to the specimen is rather different from
the Vickers and Brinell methods and is illustrated in Fig. 18.15. Initially a
minor load of 10 kg is applied, which is followed by a major load, being an
additional 50, 90 or 140 kg depending on the indenter and type of metal.
The major load is now removed, but the minor load is retained while the
hardness number is read, where

Arp=E-e

and Hr = Rockwell number, e = depth of penetration due to the major load

only but while the minor load is operating and E = arbitrary constant which
is dependent on the type of penetrator. The various relevant details of the
test may be found in B.S. 891: 1989.
506 MECHANICS OF ENGINEERING MATERIALS

Comparison of hardness Owing to the wide use of the Brinell, Vickers and Rockwell methods and the
values varying preferences for any one of these tests, there are occasions when the
same material or component is hardness tested by different methods in
different laboratories. This has led to a demand for some correlation
between hardness values determined by the three tests. It has been shown
that there is no general relationship between the hardness scales, and
empirical formulae only hold good for materiais of closely similar
composition and condition.
However, based on experimental results, the British Standards
Institution has issued a table (B.S. 860: 1989) of approximately comparative
values for the three tests, but it is emphasized that it is not intended that the
table shall be used as a conversion system for standard values from one
hardness scale to another.

18.6 Hardness of
Hardness measurements are also necessary for assessing non-metallic
viscoelastic materials
materials such as plastics, rubbers and composites. Indentation methods can
be used, but owing to the much lower levels of hardness compared with
metals very low indenter loads are used. Material thickness must be
adequate in relation to the depth of indentation and a solid mounting plate
must be used. The strain—time dependence of viscoelastic materials also has
to be taken into account. In particular the depth of penetration of the
indenter will increase with time under load and the geometry of the indent
may change due to recovery after load removal. In general it has been found
that conventional Brinnell- and Vickers-type tests are not successful with
plastics and rubbers because these methods require a clear impression on the
indent after the indenter is removed. Using a ball indenter the image of the
indent tends to be imprecise; with a pyramid indenter the indent is sharper,
but in both cases the image is difficult to see owing to the poor reflection of
light from the surface of the plastic or rubber. For plastics the most
successful results are obtained by using a conical or ball indenter (B.S. 2782:
Methods 365, B, D) or a pyramid indenter and measuring the depth of
penetration of the indenter during load application. For rubber, a rigid ball
indenter is used (B.S. 903: Part A26). Hardness testing of rubber is in fact
very common in industry because there is a known relationship between the
hardness number (I.R.H.D. — International Rubber Hardness Degrees) and
Young’s modulus for vulcanized rubber.

18.7 Factors influencing

The chemical composition of a material and the heat treatment to which it
strength, ductility and has been subjected have a great effect on the strength and ductility of the
hardness material. The mechanical properties of steel are very largely influenced by
the amount of carbon in the steel. Its strength increases with increase in
carbon content, but the ductility decreases as illustrated in Fig. 18.16.
A steel may be hardened by heating it to a high temperature and then
rapidly cooling it in a cold liquid. The strength is greatly increased by this
process, but its ductilty is reduced and it is in a brittle state. It may be
brought to the required degree of strength and ductility by further heat
treatment, or tempering. The effect of tempering on a nickel—-chrome steel
for use in highly stressed components in automobiles and aircraft is shown
in Fig. 18.17. The steel was hardened by heating to 820°C, soaked for
i hour and quenched in oil. It was then tempered for 1 hour at a range of
temperatures and cooled in oil.
 TENSION, COMPRESSION, TORSION AND HARDNESS 507

40: == 250 - > 1000

&

SO 200 -—

g 750
= :
20) aa 3 150 f=
7)
179)
o
1s
ss)
LOR aa
Percentage
elongation £ 100}— ae
€
>
500 &
50 LC 2
ee e
0 ee
:
eo
ss in tENSIO
® cystic virnit = A aaa] 200

| FE NOIDA
ttl SSIS | | 0)
fe) 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Fig. 1816 Effect of carbon content
Carbon content
on mechanical properties

Similar influences on strength and ductility are obtained for non-ferrous

alloys by the appropriate types of heat treatment.
The operating temperature of materials can range quite widely under
atmospheric extremes quite apart from special engineering situations such as
internal combustion, or liquid petroleum gas (L.P.G.) storage. It is therefore
necessary for engineers to have data in this context. A few brief comments
follow herewith and a more detailed treatment of high-temperature creep
behaviour is given in Chapter 21.
At low temperatures, for example, the strength of a mild steel is
increased; the material, however, may become very brittle under certain
circumstances, its ductility being almost negligible (see Chapter 19).
With increase in temperature, the elastic limit decreases, and is
accompanied by a corresponding decrease in the modulus of elasticity.
The tensile strength of the material decreases until a temperature of
approximately 150°C is reached. From this temperature the strength
increases, reaching a maximum value in the neighbourhood of 300°C, after
which it decreases as the temperature is further increased. Table 18.2 shows
the effect of high temperature on the tensile strengths of various steels.
The ductility decreases with increase in temperature until a temperature
of about 150°C is reached. After this temperature the ductility increases
with further increase of temperature. The increase is not regular, however,
but takes place in an erratic manner.
Table 18.3 shows the effect of temperature on a cast- and wrought-
aluminium alloy respectively.
One must conclude by emphasizing the importance of the interdepen-
dence of strength, ductility and hardness.
 508 MECHANICS OF ENGINEERING MATERIALS

EE

Maximum stress

1500, |
=——

[eoemue se
90 |-—— oy Brinell
‘ pea | 500

80 (ia -
ne —} 100

<p LOOO Naa 10 |e ae

=
2 3
a 2 60 (ee
g
aA 3
a
eSxy re / \\
mee
g]7
2

50 Reduction in area --— —* / th & 3

— wey \ = e
ae Sali ‘ g be
40 |= coe x \ & S
500 |— a — i
en y—| 300 ieee
/ ata
Izod impact /

va ene x
) be x Nee
Wa <a

Fig. 18.17 Effect

; of Eu a
Elongationon 50mm — 200
tempering tem-
perature on me-
chanical properties es eas ES a GAG
0) 100 200 300 400 500 600 700
Reheating temperatures (°C)

Table 18.2

Tensile strength, hot (°C),

Tensile tests, cold (MN/m7)

Elastic Tensile Elongation Reduction 600 650 700 750 800 850 900
limit strength on50mm_ of area
Steel (MN/m?) (MN/m?) (%) (%)

3% Nickel-chrome 745 874 23.0 62.5 SOD. OO le Sam 4 eee 7, 70

Stainless steel 726 746 24.0 58.1 374 264 158 278 WER Ref TAI
Silicon—chrome 790 Oy 21.0 40.0 Gly osky Swe Waey Ke Nn 60
Chrome steel 650 835 24.5 55.0 560 464 294 201 108 111 113
Cobalt-chrome 650 896 13.0 22.0 ay 2 Sie) — FAN ASS 907125
High-speed steel 711 943 15.0 24.0 0345 4113308 260) 15 91380
High-nickel chrome 650 1050 27.0 45.0 660) 592 523 442 371 300 232
 TENSION, COMPRESSION, TORSION AND HARDNESS 509

Table 18.3
Tensile strength
(MN/m’)
Temperature
(Ge) Wrought Sand cast Die cast

20 433 286 371

200 332 248 340
250 301 216 302
300 201 201 240
350 124 136 139

In broad terms a high hardness value is associated with high tensile

strength and medium to low ductility, while the converse is that a low
hardness number relates to lower tensile strength and higher ductility.
Hence hardness measurement is widely used as a quality-control test for
materials during manufacture.
In applications it is almost always true that a combination of material
properties is required. There will be several possible modes of failure which
must be prevented, or the design must be optimized so that, for example,
cost or weight is minimized.

Material selection A very powerful tool for identifying candidate materials during the
criteria development of new design concepts are the Materials Selection Charts
developed by Ashby’. Figure 18.18 shows a plot of Young’s modulus against
density for a rangeof engineering materials. Materials with high modulus
but low density are in the top-left corner of the chart. Quantitative criteria
for selecting the best material for a bar of minimum weight with a given
stiffness when loaded in tension (E/p) or bending (E2/p) are derived by
Ashby. These criteria can then be uSed to identify candidate materials from
the chart.
Figure 18.19 shows modulus against strength for a similar range of
materials. ‘Springy’ materials are those which can absorb a large amount of
elastic strain energy per unit volume, so /E, Chapter 3. On the log—log
scale shown all materials lying on a given line of slope 2 have the same value
of this parameter and materials below and to the right of the line will be
better than those above and to the left. Thus soft butyl rubber is a good
choice at one extreme, while at the other end of the strength range ceramics
and laminates of engineering composites are also good choices.
The same chart may also be used to choose materials for resistance to
buckling. In Chapter 10, eqn. [10.15], it was shown that buckling will occur
before yielding at a critical slenderness ratio

Z 2G ee [18.1]
if Oy

All materials on the line at 45° in Fig. 18.19 will have the same critical
slenderness ratio. Materials above and to the left of a given 45° line can be
used in longer columns before buckling becomes the limiting mode of
failure.
510 ME CHANICS OF ENGINEERING MATERIALS

1000 f& PARES

wee
i= |
— Modulus-density Engineering
‘— Young's modulus E
Im (G=3E/8;K= 6)
i MFA 88-91 (ai
Steels — ‘Ni alloys

hee : : alloys
100 le
Ee ss n alloys
i so ;
i ra Za Tin alloys
<a Lead ailoys
AE 4 ies Engineering
(5) (M/S), composites
P a
104 Ras Engineering
Fir Pine°* alloys
10) lee Ss
ee Parallel ee
ie j =
Fa Balsa_1° alee a EPOXIES Se
1— a PMMA a
Ve

i ai a PVC e
it Ze al Woods ngineering
is 23X10 ip Polyesters polymers hs

(GN/m2)aa
EYoung’s
modulus IIT. VILIITITTILLA,
= HDPE Eee,
I- Lower E limit : is a
[ for true solids Perpendicular PTFE Be ey
— to grain oO 3 vy,
ie Spruce LDPE ire
ae
MY 7
re Bess
A ee We:
4 /
,-=—— Guidelines
Bea
for
A
10 | y: / minimum weight
0.1 [f& = A A\ design |
as eae a / =
3x102 Elastomers Ai =]
cs Be \ ZE za / 4

ss ana
k
ae / Polymers So ba yf dl
ak ilicone =
Halt Soft | ae Cc
butyl | p
0.01 = joielsal essa fas | lal iaye ae ees
Ox 0.3 1.0 3 10 30

Density p(Mg/m®)
Fig. 18.18

18.8 Summary
The properties discussed in this chapter represent the basic information that
the designer needs to have available in relation to the selection of materials
for ‘steady-load’ components and structures.
The tensile, compression or shear moduli together with the yield (or
pro of) stress and Poisson’s ratio are needed for linear-elastic design. The
plastic properties of tensile or compressive strength and ductility are
required to give a measure of reserve safety in the event of exceeding the
yiel d level. The hardness of the material gives a check on its condition.
However, it is important not to get a false sense of security, since it will
be seen in the next three chapters that although what appears to be a very
suitable material in the required condition has been selected for ‘static’
 TENSION, COMPRESSION, TORSION AND HARDNESS 511

1000 ~-
es | Pa hata ear,
| Modulus-strength va ve
|_ Metals and polymers: yield strength ua
| Ceramics and glasses: compressive strength Engi 5
Elastomers: tear strength ngineering
| Composites: tensile strength alloys

| Min. energy a
storage per we
100 |= unit volume
[= Yield before A fA
I— buckling oo =:) Engineering”.
is y) cae oe v i
vi &
ee v7 Concrete Engineering i I
ie y “fe composites
y A

y,a y \
y 7
Be
10 |-——______ja a 5 nex
Ee v y Porous, wen hee
oe ee Ae Ty, ceramics / VAN oe pu
ar VA 4 vi ne
lee oe ys to grain Zr PMMA ie eer ee =
fe 2B Ae 5
Ed ah _ 49-4 y : Woods aoe Se
E "
CF a =
A
Ze 4 ee
1.0
(GN/m)
Young’s
E
modulus
=
fa
i= es
7 WA 4], fr
Ss Ves4
Vana |
— : a 7 Perpendicula ve) /
ie ve As f° grain Engineers Viera. ad
= ae
rs = 10 o,pele as /” Balsa
107¢°
vy” /
polymersoF eee Design =

y c eae Ve guidelines
(Gyal e V Ti
Polymers, 4 j y Ges i
foams > A
7 iy oe. iy Max. energy al
ze E ve storage per +
Cork es unit volume =
2 :
fies (\ ye Cf Buckling =
Silicone ce Elastomers ae Cc before yield
; Soft He :
0.01 acm fia

0.1 1 10 1.00 1000 10000

Ss 6, (MN/m?
Fig. 18.19 HeDeT SRN Am)

working conditions other factors can have very serious consequences during
the service life of a component or structure.

Reference ; autre
1. Ashby, M. F. (1992) Materials Selection in Mechanical Design,
Pergamon Press, Oxford.

Bibliography
Ashby, M. F. and Jones, D. R. (1980) Engineering Materials I: An
Introduction to their Properties and Applications, Pergamon Press, Oxford.
Ashby, M. F. and Jones, D. R. (1986) Engineering Materials II; An
Introduction to Mucrostructures, Processing and Design, Pergamon Press,
Oxford.
512 MECHANICS OF ENGINEERING MATERIALS
ne aE SSaeRO

Dieter, G. E. (1976) Mechanical Metallurgy, McGraw-Hill, New York.

Jones, D. R. (1993) Engineering Materials IIT: Mechanical Failure Analysis:
Case Studies and Design Implications, Pergamon Press, Oxford.
McClintock, F. A. and Argon, A. S. (1966) Mechanical Behaviour of
Materials, Addison-Wesley, Reading.
Nadai, A. (1950) Theory of Flow and Fracture of Solids, McGraw-Hill, New
York.
Pascoe, K. J. (1980) Introduction to the Properties of Engineering Materials,
Van Nostrand, New York.

Problems 18.1 A tensile test has been carried out on a mild steel specimen 10mm
thick and 50mm wide rectangular cross-section. An extensometer
was attached over a 100mm gauge length and load extension
readings were obtained as follows:

Load (kN) 16 32 64 96 128 136 144 152 158

Extension 0.016 0.032 0.064 0.096 0.128 0.137 0.147 0.173 0.605
(mm)
Load (kN) 154 168 208 222 226 216 192 185.4
Extension 1.81 2.42 Wises 12.0 16.8 22.0 24.0 Fracture
(mm)

Plot load—extension diagrams for the elastic range and the plastic
range and determine: (i) Young’s modulus; (ii) proportional limit
stress; (ii1) yield point stress; (iv) tensile strength; (v) percentage
elongation.
18.2. An aluminium alloy specimen of 1.2 mm thickness and 25 mm width
cross-section and a parallel gauge length of 50mm is tested in
tension giving the following data:

Load (kN) ie. eo6 | as4 C4. FE Vi6ie 80 S4u aS

Extension 0.0443 0.0886 0.133 0.155 0.181 0.198 0.219 0.246 0.281
(mm)
Load (kN) 92-108 12.0 12.4
Extension 0.332 0.645 = 1.05 1.94
(mm)

Determine values for: (i) Young’s modulus; (ii) 0.1% proof stress;
(i1) 0.5% proof stress; (iv) tensile strength.
18.3. The following load—compression data has been obtained on four
copper cylinders of 12mm diameter and lengths 24, 12, 6, 4mm.
Construct the true compressive stress-strain curve for the material.

do/ho Load kN 13.5 Wpess) 29.0 34.0 40.0

3 % Compression 3:3 11.5 19.5 26.7 37
2 % Compression 3.8 12.6 et 29)5 41.5
1 % Compression 4.0 13.8 23.3 B22 45.8
4 % Compression 4.3 14.4 24.2 33.6 SS)
TENSION, COMPRESSION, TORSION AND HARDNESS 513

18.4 A Wallace-type of micro-hardness testing machine measures the

depth of penetration of a standard Vickers indenter. During a test on
polypropylene, using an applied weight of 300g the instrument
records 2.66 x 10? mm. From this information calculate the Vickers
hardness number of the plastic.
18.5 A nickel-chrome alloy steel component fails in service and an
investigation is required. A Brinell hardness test is carried out on the
broken part giving a value of 320 B.H.N. An Izod impact sample cut
from the part gives a fracture energy of 50J.The design maximum
stress for the component was 1075 MN/m?’. What is the main
contributing factor for the failure? (Hint: Examine Fig. 18.17).
CHAPTER
Fracture Mechanics

=
The material properties discussed in the previous chapter principally relate to the
quality control of materials and to initial material selection by a designer. We now
need to recognize that in spite of carefully employing the design stress and strain
analysis procedures to avoid failure by gross deformation, elastic instability or
exceeding the yield stress, there are other factors to be taken into account in design.
These ‘factors’ include straining rate, fluctuating stresses, stress concentration,
metallurgical flaws, high and low temperatures, corrosion and other special effects. |
The designer needs to be aware that these variables can cause an engineering
component to fail by fracture which may lead to a catastrophic disaster and, in the
most serious cases, loss of life.
This chapter and the remaining two will attempt to cover these topics in a manner
which will provide an initial insight which can be built on as required through the
specific texts in these areas.

19.1 Fracture concepts

Fracture is concerned with the imutiation and propagation of a crack or cracks
in the material until the extent of cracking is such that the applied loading
can no longer be sustained by the component or structure. It is generally
accepted that initiation of a crack is relatively difficult to design against and,
in fact, most components or structures will contain some crack-like flaw/
defect by the time manufacturing is completed in spite of rigorous
inspection procedures. We, therefore, must try to design for non-
propagation or, at next best, controlled propagation. In the latter case
normal in-service inspections should enable the presence of growing cracks
to be detected. This tends to be the case in respect of the phenomenon
known as fatigue which is fully discussed in the next chapter. Cracks which
propagate in an uncontrolled manner at very high velocity through materials
is a situation of greater danger in service and is the theme of this chapter.
From a metallurgical viewpoint there are only two paths for a crack
passing through a metal, either transcrystalline or intercrystalline. The latter
only occurs in a few particular circumstances, e.g. creep, stress, corrosion.
The former is the more general mechanism of which there are two types
related to the crystallographic planes known as shear and cleavage. Shear is
the result of certain crystal planes sliding over one another, termed s/ip, and
is associated with a great deal of macroscopic plastic deformation. Cleavage
occurs on different crystallographic planes caused by a normal (tensile)
stress and involves negligible plastic deformation. Shear fractures have a dull
appearance, sometimes described as fibrous, but cleavage fractures reveal
smooth reflecting planes described as bright and crystalline or granular. By
relating these two modes broadly to stress-strain characteristics it is found
that the ductile shear mode tends to be above-yield stress fracture with high
energy absorption and hence toughness. On the other hand, a fully brittle
cleavage mode would be associated with a low toughness, low energy
 BRAG URES ME GHANTIES 515

absorption failure which apparently occurs below the yield stress. Of course
there are the possibilities of mixed-mode fractures, depending on a
combination of a number of factors, between the two extremes above.
One of the more important influences on the mode of fracture is the state
of stress. In engineering components, as compared with simple laboratory
uniaxial stress tests, a complex stress system generally exists. In a triaxial
stress state where 0; > 02 > 03, the maximum shear stress is (a — 93),
but as 03 — 01, T — 0. In the extreme case of hydrostatic tension and
compression, 0; = 02 = 03 and 7 = 0. It is evident that shear cannot occur
and hence cleavage fracture will result. The introduction of a discontinuity
or notch into a piece of material causes a stress concentration and triaxiality
of stress to a degree which depends on notch geometry and _ loading
condition.
Most metals exhibit some temperature dependence of fracture over a
range from, say, —100°C to +100°C. Toughness is reduced by lowering
temperature, and so this aspect of the working environment of engineering
structures and components must be taken into account.
Unstable fracture manifested itself first as a serious engineering problem
of nominally ductile low-carbon steels from the mid-1930s to the mid-1950s.
Large welded structures such as ships, bridges and storage tanks failed in a
catastrophic and apparently brittle manner. From this grew an extensive
research programme into what was then called brittle fracture. Although
factors such as stress concentration at ‘notches’, weld defects and low
temperature contributed to the initiation of a crack, the principal controlling
factors on fast propagation of the crack are the ability of the material to
absorb energy, 1.e. the toughness, and the existence of crack arrest barriers.
In the latter context riveted or bolted plate structures were better than the
‘continuous’ all-welded structure if the welds were not of high quality.
The past 40 years have seen the development of ultra-high-strength
alloys for rocket motors and space vehicles. Some of these low-ductility
materials were found to be susceptible to unstable fracture from small
defects owing to low toughness. This resulted in the development of the
theory of linear-elastic fracture mechanics (L.E.F.M.) which was
accompanied by the establishment of special tests to measure the fracture
resistance of materials. The analytical and experimental techniques of
fracture mechanics now have a major influence on design for crack growth
which is (i) unstable, (ii) intermittent/cyclic (fatigue) and (ii) time
dependent in metallic and non-metallic materials which are either
homogeneous or fibre-matrix composites.

19.2 Linear-elastic
Linear-elastic fracture mechanics (L.E.F.M.) developed from the early work
fracture mechanics
of Griffith! who sought to explain why the observed strength of a material is
considerably less than the theoretical strength based on the forces between
atoms. He concluded that real materials must contain small defects and
cracks which reduce their strength. These cracks cause stress concentrations
but they cannot be allowed for by calculation of a linear-elastic stress
concentration factor K,. This is because an elliptical defect, Fig. 19.1, has its
stress concentration factor defined by the equation

Kiel st AG [19.1]
516 MECHANICS OF ENGINEERING MATERIALS

ce)
As b > 0 the defect becomes a crack, but K, — oo which would suggest
that a material with a crack would not be able to withstand any applied

ae forces. This is contrary to what is observed so Griffith developed a concept

to explain how a stable crack could exist in a material. He postulated that a

batt
crack only becomes unstable if an increment of crack growth results in more
stored energy being released than can be absorbed by the creation of the new
crack surface.
Based on this premise and with subsequent refinements, principally by

|
Irwin’, L.E.F.M. has developed as an analytical approach to fracture. It
relates the stress distribution in the vicinity of a crack tip to other
oO parameters such as the nominal stress applied to the structure and the size,
shape and orientation of the crack. Thus it permits representation of the
Fig. 19.1 Elliptical defect in a material fracture properties, often in terms of a single parameter.
stressed plate There have been two main approaches: (i) energy; (ii) stress intensity
factor.

19.3 Strain energy

For a through crack of length 2a in an infinite body of unit thickness, as
release rate
shown in Fig. 19.2, the surface energy U, stored in the material due to the
formation of the crack is given by
U, = (2a)2y (19.2)
where y =surface energy per unit area. In the context of the fracture of
brittle materials this term is replaced by y=1G, where G is energy
absorbed per unit area of crack (note that G refers to the area of crack which
will be half the new surface area).
Thus eqn. [19.2] may be written as
U, = 2aG [19.3]
From the concept of elastic strain energy introduced in Chapter 3 the
Fig. 19.2 Sharp crack in a stressed elastic energy U, released by the formation of the crack is given by
infinite plane
1
UF = Ala(x). A(x, a) dx [19.4]
Energy
released og 6 6 o where a(x) is the stress distribution in the vicinity of the crack, and A(x, a)
og?
is the vertical opening of the crack.
It can be shown’ that for the through crack of length 2a in an infinite
plate, Fig. 19.3,

To a’
Us=
E
k [19.5]
where k = (1 — v”) for plane strain and 1 for plane stress and v is Poisson’s
ratio.
Thus the surface energy which is developed in the material is
increasing linearly with crack length, whereas the energy released by the
formation of the crack increases with (crack length)’. This is illustrated in
Fig. 19.4.
The net energy in the presence of the crack is thus the mathematical
summation of the surface energy U, and the energy released U,. Griffith
proposed that the threshold between a stable crack and an unstable crack
 FRACTURE MECHANICS 517

occurs when an increment of crack growth causes more energy to be

released than can be absorbed in the material. Thus the critical condition is
U;
dU/da = 0 which occurs at point A in Fig. 19.4 and hence the critical crack
length a, is defined. From eqns. [19.3] and [19.5]

Energy
A
1

0 : >
ge Crack length a

CUE
This then reduces to

(EGc)'? = o(na)'” (plane stress)

Ue EGC, \'2
( 5) = 0(na)!? (plane strain) [19.6]
fig. 194 Variation of energy with —vV
crack length
These equations are an expression of the conditions for fast fracture in a
brittle material. It should be noted that Gc is a material property which is
referred to as the critical strain energy release rate, toughness or crack extension
force, and has the units J/m?.
A high value means that it is hard to propagate cracks in the material as,
for example, in copper for which Gc ~ 10°kJ/m”. This may be compared
with a value for glass of approximately 0.01 kJ/m?.

Example 19.1
Determine what size of defect will cause a fracture of glass before it reaches its
ideal strength. Defect-free thin whiskers of glass can reach strengths of 3600 MN/
m2, but the toughness G, is only 0.01 kJ/m?. The modulus of glass is 69 GN/m2.

The largest defect which can be tolerated before a fracture occurs below the
ideal strength oy is when both modes of failure occur at the same stress, i.e.

(EGc)'? = of (xa)'!” [19.7]

The size of defect a at which this occurs is

E 9 x 10? x 0.01 x 103

pee EE ee ase [19.8]
TOy (3600 x 10°)

Any defect which is larger than this extremely small value will reduce the
apparent strength of the material, since it will fail by fracture long before its
ideal strength is reached.

Example 19.2
The fracture stress of a large sheet of steel with a central crack of 40mm is
480 MN/m2. What is the fracture stress of a similar sheet with a crack of 100 mm?

Fracture occurs when

V(EGc) = 01 (ma1) = 02,/ (a2) [19.9]

518 MECHANICS OF ENGINEERING MATERIALS

Therefore, since the crack length 2a = 40mm in the first sheet and 100 mm
in the second, the stress to cause a fracture in the second sheet is

oi (a1) _ 480,/(20 x 10-%)

= 304 MN/m’* [19.10]
PS Novas iar af(SUG10E2)

19.4 Stress intensity

factor Although Griffith put forward the original concept of L.E.F.M. he
restricted his work to brittle materials (e.g. glass) and it was Irwin who
developed the technique for metals. He examined the equations that had
been developed for the stresses in the vicinity of a sharp crack in a large
plate as illustrated in Fig. 19.5. The equations for the elastic stress
distribution at the crack tip are as follows:
K 0 ahEN | oe
Oy, = Onn? cos D 1 — sin D sin >

K ]
Ci cos e 1 + sin z sin sd [19.11]
; (2nr)'/? 2 2 2 :
ae K Le 6 36
y= (nr) 7 Sin 5 COs 3 cos zy

and

7)
n= 5 COS (5) (plane strain)

or

Fig. 19.5 Stress distribution at

crack tip in an infinite plate
y 4

Variation of o, along
xXaxis (6 = 0)

He observed that the stresses are proportional to (a)'/ >, where a is the
half-length of the crack. On this basis, Irwin defined a stress intensity factor K
FRACTURE MECHANICS 519

as

K =0(na)'” [19.12]
The stress intensity factor is a means of characterizing the elastic stress
distribution near the crack tip but in itself has no physical reality. It has
units of MNm-~*/? and should not be confused with the elastic stress
concentration factor K;,.
Thus the basis of the L.E.F.M. design approach is that:
(a) All materials contain cracks or flaws.
(b) ‘The stress intensity value K may be calculated for the particular loading
and crack configuration.
(c) Failure is predicted if K exceeds the critical value Kc for the material.
The critical stress intensity factor is sometimes referred to as the fracture
toughness and will be designated Kc. By comparing eqns. [19.12] and [19.6]
it may be seen that Kc is related to Gc by the following equations:
(EGE? — Ke (plane stress) [19.13]

Eee!
( =) = Ke (plane strain) [19.14]

In order to extend the applicability of L.E.F.M. beyond the case of a

central crack in an infinite plate, K is usually expressed in the more general
form

K = Yo(na)'”” [19.15]
where Y is a geometry factor and a is the half-length of a central crack or the
full length of an edge crack.
The methods for evaluating stress intensity factors for particular loading
situations are formidable and outside the scope of this book. The main
theoretical methods include: boundary collocation, conformal mapping,
numerical methods and analysis of stress functions. The reader should refer
to other texts** for details of these methods.
Figure 19.6 shows some crack configurations of practical interest and the
expressions for K are as follows:
(a) Central crack of length 2a in a sheet of finite width

W ma\'/?
Ke o(ma) OS
(= an | [19.16 |

(6) Edge cracks in a plate of finite width

2 (W ra 0.2W . na\'?
K=O sie) (an sin
Ta W Ta W

(c) Single edge cracks in a plate of finite width

2 3
K =o(na)!” fw= i es10.6(=) -21.7(<)

430.4 i | [19.18]
520 MECHANICS OF ENGINEERING MATERIALS

Fig. 19.6 —
Typical crack configura- ; ;

ecto
(a) Finite width plate (b) Double-edge crack
eelite Siar
(c) Single-edge crack (d) Internal penny crack

Oo
a/2c ue

af
OFS: pee

oe E 2W a F
0.4 he

(e) Elliptical surface crack (f) Three-point bending

Note: In most cases (a/W) is very smail so Y = 1.12.

(d) Penny-shaped internal crack

K =o0(na)'” (<) [19.19]

TT

assuming a < D.
(e) Semi-elliptical surface flaw

K =o(na)' (Ss
rie [19.20]
(f) Three-point bending

ey3FL a\1/2
93 S| (a\3/2 ue
) H14.+14.53(5
(=)
oe 3.07(5)

Pay) 4x92
=
-25.11(— +25.8(—
25.8(=) [19.21]
or

K = ar ile ) [19.22]
A handbook of stress intensity factors for a range of geometrical
configurations and loadings has been prepared by Rooke and Cartwright°.
 FRACTURE MECHANICS 521

19.5 Modes of crack tip

So far it has been assumed that the loading plane is symmetrical with respect
deformation
to the crack plane. This is probably the most common situation and is
referred to as the opening mode (designated as mode I), Fig. 19.7(a).
Therefore to be strictly correct the stress intensity factor should have the
suffix I. There are in fact three deformation possibilities and the other two
are shown in Fig. 19.6(d) and (c).
(a) Opening mode (mode I) having symmetry about the (x, y)- and (x, z)-
planes.
(b) Sliding or shear mode (mode II) having antisymmetry about the (x, z)-
plane and symmetry about the (x, y)-plane.
(c) Tearing mode (mode II) having antisymmetry about the (x, y)- and
(x, z)-planes.

Fig. 19.7 Cracking modes

(a) Opening mode (1) (b) Shearing mode (II) (c) Tearing mode (III)

Thus the stress intensity factor should be referred to as K;, Kj; or Ky

depending on the deformation mode.

19.6 Experimental
From the previous sections it is apparent that the stress intensity factor K
determination of critical for a particular loading situation may be calculated from a knowledge of the
stress intensity factor nominal stress and the size, shape and orientation of the crack. The basis of
L.E.F.M. is that fast fracture will occur when K reaches the critical value for
the material. This critical value Kc may be determined experimentally from
standardized tests®” on samples of the material.

Effect of size It has been found that the value of the critical stress intensity factor Kc at
which unstable crack growth occurs under static loading depends on the
specimen thickness as illustrated in Fig. 19.8.

Fig. 19.8 Variation of fracture Fracture

toughness Ke !
toughness with plate thickness |
|
(transition) |
'
|
|

Sheer
(ductile) (brittle)
|
|
|

|
|
|
| Kic
|
|
|

0 Increasing plate thickness

522 MECHANICS OF ENGINEERING MATERIALS

The limiting value of Kc is observed in plane strain (which is the

maximum constraint condition). It is designated K7¢ and is usually referred
to as the fracture toughness. It is important not to confuse K; with K7c. As
shown earlier, K; depends on the configuration of the system, but Kyc is a
material property and is independent of the configuration of the system.
Since K7c is the most conservative value of fracture toughness for the
material, it is this value which is used in design calculations. Therefore the
standardized test conditions are carefully chosen to ensure that it is Kyj¢
which is determined and, after testing, checks are made on the validity of the
test conditions.

Test methods The two most commonly used test methods involve bending or tensile
loading of the test specimens shown in Fig. 19.9. The specimens contain a
carefully machined notch which may have a plane front or a chevron profile.
A chevron notch has been found to keep the crack in-plane and so the
machining operation is not quite so critical as with a plane notch.

Fig. 19.9 Details of tensile and

bending fracture toughness spe-
cimen

Root radius oO)

0.1 max. Detail of [iq Section
plane through |} J
w 1 notch chevron :
notch

(a) Compact tension specimen (b) Bend test specimen

In order to get a sharp crack the machined notch is extended by applying

a cyclically varying force to the test piece. This cyclic stressing, in which the
peak value of stress is chosen to ensure that K,,., is less than 70% of the
critical value K7c, causes the crack to grow by a fatigue mechanism (see

Fig. 19.10 Fracture toughness

testing

O Clip gauge displacementA

Slope OQ is exaggerated
for clarity

Clip gauge displacement A

 PRAGHURE VLE CENTES 523

Chapter 20). The sharp crack thus produced should be at least 1.25mm
long. At this point the test specimen is removed from the fatigue machine
and subjected to a static test. Typical forms of test are illustrated in Fig.
1910.
The force is measured directly from the load cell on the testing machine
and this is recorded automatically as the vertical axis on an X—Y plotter.
The horizontal axis is obtained from a displacement (clip) gauge attached to
the test piece as indicated in Fig. 19.10.
From the graph so obtained a value of force, termed Fg, is obtained so
that an interim value of stress intensity factor Kg may be calculated. If the
material is perfectly elastic up to fracture then the peak force is taken as Fg.
Assuming that the material exhibits some non-linearity as in Fig. 19.10, then
the procedure for obtaining Fg is to draw a line OQ with a slope of 95% of
the initial slope of the force—-displacement graph. The value of force at the
point where this line intersects the /—A characteristic is taken as Fg.
The fracture surface is then examined so that the crack length a may be
determined as the average of a), a2 and a3 where a; and a3 are mid-way
between the centre-line of the specimen and its edge as shown in Fig. 19.11.

F :
Fig. 19.11 Crack length measure-
ments (Bend specimen)

ay
ag
a3

End of fatigue crack Machined chevron

Using the values of Fg and a thus obtained the following equations are
used to calculate Kg:
; Fo a
Bending Kg => Afi (eel [19.23]

F Fo a
‘Tension Ko = ayine Ge [19.24]

The values of fi(a/W) and fo(a/W) depend on the particular specimen

geometry. (See Table 19.2, p. 529.)
Having thus obtained an interim value of stress intensity factor Kg,
checks are made that the following conditions have been satisfied:
K 2)

eee 2.5(=2) ise Gs 2055") Por ele

Oy
If these conditions are upheld then the test is a valid one and Kg is taken
to be K7c. Otherwise the result must be discarded. If desired G7¢ may then
be calculated from eqn. [19.22]. Table 19.1 gives typical values of K7¢ and
G,c for a range of materials.

Example 19.3
A pressure vessel is to be fabricated from plate steel which may be either: (i) a
maraging (18% nickel) steel with cy = 1900 MN/m?, Kc = 82MNm_—?/? or (ii) a
524 MECHANICS OF ENGINEERING MATERIALS
SS SSS SSS SSS SS ——eee

Table 19.1 Typical values of Kyc¢

and Gic Kic Gic
Material (MN m~?/2) (kJ/m?)

Mild steel 100-200 50-95

High-strength steel 30-150 5-110
Cast iron 6-20 0.2-3.0
Titanium alloys 30-120 7-120
Aluminium alloys 22-45 7-30
C.F.R.P. (uniaxial fibres)* 20-45 2-30
G-.F.R.P.(uniaxial fibres)* 10-100 3-60
G.F.R.P. (laminate) 10-60 5-100
Wood* 8-13 6-20
Glass 0.3-0.7 0.002-0.01
Acrylic (P.M.M.A.) 1.0-2.0 1.3-1.6
Polycarbonate 1.0-3.5 2.0—5.0
Concrete 0.2-0.4 0.03-1
Epoxy 0.5—0.7 0.08—0.34

* Perpendicular to fibre direction.

medium strength steel with cy = 1000 MN/m7, Kc = 50 MN m °/2. Which of these

two steels has the better tolerance to defects and compare their fracture toughness
if they are to have the same defect tolerance? A factor of safety of 2 should be used
for the design stress.

Assuming that eqn. [19.15] is appropriate for the steel plates then
(i) For the maraging steel the critical defect size is given by

Ke
a = where or = ay
TOY 2

82?
= 5 = 2.4mm
(950)
So the critical crack length is 4.8 mm.
(ii) For the medium-strength steel

50?
a. = —— = 3.18mm
(500)?
So the critical crack length is 6.36mm.
Hence the medium-strength steel is more tolerant of defects in the
material. In order that the maraging steel would have the same tolerance its
Kjc-value would need to be

Krc = o4(na,)'? = 950(m x 3.18 x 10-3)!/

= 95MNm ?/?
This analysis can be extended into a general material selection chart as
described in Ashby.'* On the log—log plot of fracture toughness against
strength of Fig. 19.12, all materials lying on a given 45° line have the same
 FRACTURE MECHANICS 525

1000
f= ] re es eeeSil ee a a RL ie lien heLares
[_ Fracture toughness-strength IC aie OO Z|
| Metals and polymers: yield strength a
| Ceramics and glasses: compressive strength -{
ia Composites: tensile strength 2
Process zone diameter~ Kjg 10, 7
Engineering ay
alloys 1072

100 =kK : we
:
ee Yield before 4
fracture A =
a Ze 1057
y a _ Engineering é
ae y -~ composites 1
ea Guidelines Z oe
ex for safe design Zee
Pl 8 ie Ze 10°3
= 7
va AN0) = |
= BE aay Engineering
< oe
2 2 Mebee De polymers Sjalons : () ,

2
= ite
ae=c,”
fie y
Ges
Md,
3 Za Z~ SiN?
Ze 3Nq
= K Woods SZ Diamond
g . sales a Pottery, }
=] f o brick, etc, we
. 1.0 Z 2 y
: : ic
Be fe We

a Vis
Va Engineering
are iA ceramics ial
me (mm) / Porous =
TO,“ ee ve ceramics
100 / i alsa 4 De
0.1. Fae Polymers SS 7 Cice) S
= foams y ea y
= ve
ve om ve en 2 Fracture
Ss 40 y . Ye y , oy before yeild
Ze & / 4 o
Vi WA Ye
NOT; 10 -2 10°3 107 ee mm

0.01 {ee ee eet [esol atest Sah ee ea cst

01 1 10 100 1000 10 000

Strength O- (MN/m?)
Fig. 19.12
value of a, = K#, (07). As noted in the figure, the strength oy of metals
and polymers can be taken as their yield stress, oy.Materials which can
tolerate a defect larger than the value of a, corresponding to a given line on
the graph lie above and to the left of that line. If a defect of that size exists in
the material it will yield before it fractures. Those materials with smaller
critical defect sizes lie below and to the right of the line corresponding to a
given value of a,. If a defect of size a, exists in a structure made from this
material, failure will occur by fracture before the material yield stress is
reached.

19.7 Fracture mechanics

- : The stress field equations given earlier show that the elastic stresses would
for ductile materials become very large in the vicinity of the crack tip where r < a. In practice
these large stresses do not occur because in a ductile material this region
Plastic zone correction becomes plastically deformed. It would appear therefore that in such
526 MECHANICS OF ENGINEERING MATERIALS

materials the formation of a plastic zone at the crack tip would invalidate the
use of L.E.F.M.
However, Irwin has shown that when yielding occurs at the crack tip,
L.E.F.M. techniques may still be applied if an equivalent crack length is
used, i.e. a physical crack length plus an allowance for the extent of the
plastic zone. This zone is generally represented by a circular boundary of
radius r, at the crack tip and the equivalent crack length then becomes

a’ = (a+r) [19.25]
This effect is illustrated in Fig. 19.13 where the modification to the stress
system due to local yielding is indicated.

Fig. 19.13 Stress distribution at

crack tip due to local yielding \ Elastic stress distribution
Wee
Ni

Stress distibution
Sy
after local yielding

|S nominal
>

Equivalent
crack

For plane stress situations r, has been shown to be given by

ryDp =—(|—
ay
Ia oy 19.2: 6]

Mid-section
Fig. 19.14 Variation of crack tip (plane strain)
plastic zone across thickness of — surface
;
material (plane stress)
l Crack tip rp (Plane strain)

~ Tp (Plane stress)

Thickness Plane
strain
region

Specimen cross-section

For plane strain cases the extent of the plastic zone is less as indicated in
Fig. 19.14. In this case the radius of the plastic zone is given approximately
by the following expression:

| ae
ae
%,
= — —
[19.27]
 FRACTURE MECHANICS 527

Example 19.4
A wide sheet of aluminium alloy has a central crack 25mm long. If the fracture
stress for the sheet is 200 MN/m? and the yield stress of the material is 400 MN/m2,
calculate the fracture toughness of the material (i) using L.E.F.M., and (ii) using the
plastic zone correction.

(i) Using L.E.F.M.

yee
Kic = Omax (ma

1/2
= 200 (=(“S))
Kic = 39.6 MN m7?”
(ii) Using correction for the plastic zone

Kic = Omax |T(a sz r,)/?

=
0.025
y |
1 /200\2] )1”
00 x(2 )) = (a) |}
Kic = 42MNm7?/?
The difference between the values of K7c calculated by the elastic and
plastic zone correction methods will increase as the ratio (Om,/oy)
increases, i.e. when the size of the plastic zone increases. In order to ensure
that the plastic zone dimensions are small compared with other dimensions
of the test piece, it is found that the more ductile the material the larger
must be the test specimen in order to meet the criteria laid down for the
validity of the test. In some cases this can lead to specimen handling
problems as well as difficulty in getting a testing machine with enough
capacity to break the specimen. For these reasons it would be advantageous
to have an alternative fracture test method utilizing smaller specimens of
tough materials.
This need is also borne out by the fact that for materials which exhibit
extensive plasticity before fracture, even the use of the correct crack length
approach becomes inappropriate. As a result of this a number of alternative
test and analysis methods have been developed and the more important of
these will now be described.

Crack opening This method was proposed by Dugdale*® and developed by Wells’. It is
displacement (C.O.D.) based on the fact that owing to plasticity at the crack tip, the crack opens in
the direction of the applied stress as illustrated in Fig. 19.15.
The C.O.D. test is performed using the same type of specimen shown in
Fig. 19.9 and the same procedure as described for determining K. 1c’. The
clip gauge measures the sample opening , which is a measure of the
resistance of the material to fracture initiation. The objective is to determine
528 MECHANICS OF ENGINEERING MATERIALS

Fig. 19.15 Increased crack opening Plastic

due to yielding at crack tip A zone

Pseee - ——
-

the critical crack opening at the onset of crack extension. A typical test
record is shown in Fig. 19.16(a) and this indicates how the plastic
component V, of the clip gauge displacement is obtained.

Force F
Fig. 19.16

(b) Plastic hinge in C.O.D. specimen

=| a Clip gauge displacement
Vp Ves
(a) Load-displacement graph

For this type of characteristic the critical displacement is taken as the

value corresponding to the maximum applied force F. Occasionally other
shapes of characteristics are obtained and the British Standard!” should be
referred to for details of how to obtain V, in these cases. The clip gauge
displacement ’, must be converted to the crack opening displacement 6
using

K(1-v*) ; 0.4(W —a)V,p

O=
aa aT ee [19.28]

X= aon
where

F a
Bw! pp
and z =thickness of knife edge. We obtain fi(a/W) from Table 19.2.
A basic requirement of toughness tests for critical C.O.D. is that they
should be carried out at full thickness. These tests do not have to satisfy
criteria related to plane strain conditions because they seek to determine a
toughness value relevant to the particular thickness of interest. For
application to welded structures it is necessary for tests to be carried out on
material representing different regions of the welded joints.
 FRACTURE MECHANICS 529

Table 19.2
alW 0.45 0.46 0.47- 048 0.49 0.5 0.51 0.52 0.53 0.54 0.55

filaf/W), 91- 9.37 9.66 9.96 10.28 10:61 10.96 11.33 11.71 12:12 12:55
frla/W)' 8.34 8.57 881 9.06 9.32 9.6 9.9 10.21 10.54 10.89 11.26

* Three-point bending (for support span = 41). t Tension.

From elastic—plastic analysis of the crack tip region using several

simplifying assumptions, the critical crack tip opening displacement has
been related to the critical values of fracture toughness’. The most
commonly used relations are

1
lm =—€
Res l
(plane s tress) 4
[19.29]

K?.(1—v”)
ee eee!
Ae l
(plane strain) [19.30]

where X is a constant constraint factor which theoretical analyses have

shown to be in the range 1-2 and which experimental measurements have
shown to be approximately unity for both plane stress and plane strain
situations.

Example 19.5
A pressure vessel has a diameter of 2m and a wall thickness of 10mm. During
routine inspection it is discovered that there is a crack 6.5 mm deep on the outside
surface of the cylinder. The steel used in the vessel is known to be tough and a
C.0.D. test on a sample gives the following results:

Sample width, B= 25mm; V, = 0.35mm

Sample depth, W= 50mm; F- = 65kN
Crack length,a = 25mm; z=2.5mm
Calculate the maximum internal pressure to which the vessel could be subjected. E
= 207 GN/m?, cy = 500 MN/m’, v = 0.3.
From egn. [19.28]

Raa) 0A WW —a)V,
~— -oyE 0.4W
+ 0.6a + z
where
Ve a
R= pwn GF)
From Table 19.2, at (a/W) = 0.5, fi(a/W) = 10.61
0.065 x 10.61
123.4MN m~>/?
~ 0.025(0.05)7
530 MECHANICS OF ENGINEERING MATERIALS

From eqn. 19.28

— (123.4)7(0.91) 0.4 x 25 x 0.35 x 1073

“2 x 500 x 207 x 103 ~—0.4(50) + 0.6(25) + 2.5
= 0:16 x 105° m
Using eqn. [19.29], assuming plane stress conditions and A = 1,

Ke = (Eoy6,)'? = (207 x 103 x 500 x 0.16 x 10-7)!

= 128.7MNm *?
Using eqn. [19.15] with Y = 1.16 for the crack geometry,
Ke 2 128.7
Hoop stress, og =
1.16(ra)\/? — 1.16(4 x 6.5 x 10-3)!/?
= 776 MN/m

_ bd
a;
Itog 29010 X 776 ,
= - = 7.8 MN/
a 2000 L8 =

j-contour integral During the period when the C.O.D. method was being developed in the
U.K. an alternative approach was being developed in the U.S.A.!!. This has
become known as the 7-contour integral approach. It is based on the finding
that for a two-dimensional crack situation, the sum of the strain energy
density and the work terms along a path completely enclosing the crack tip
are independent of the path taken. This is shown in Fig. 19.17.

Fig. 19.17 Crack tip co-ordinate y

system and arbitrary line integral
contour

Crack x

a R

The energy line integral 7 is defined for either the elastic or elastic—
plastic behaviour as follows:

Ou
ip |w dy — r(5*) ds [19.31]

where p = any contour path surrounding the crack tip (note: the integral is
evaluated anticlockwise starting at the lower surface of crack and proceeding
along any path to the top surface); » = strain energy density = i odes i
 FRACTURE MECHANICS 531

traction vector defined according to the outward normal 7 along path p; u =

displacement vector; and s = arc length.
From a more physical viewpoint 7 may be interpreted as the potential
energy difference between two identically loaded bodies having crack sizes
(a) and (a + da). In this context
WX Gs
ea a
where B = material width and U = strain energy or work done (area under
the load—displacement curve).
The j-contour integral provides a means for describing the severity of
conditions at a crack tip in a non-linear elastic material. Although it is a work
absorption rate, # is equivalent to Griffith’s strain energy release rate
concept (note 7 = G for the linear elastic situation) and it is related to the
stress intensity factor used in L.E.F.M. As with these other approaches, the
critical value of # has been found to be dependent on thickness and test
specimen geometry. However, recommended test procedures have been
developed and using the suggested specimen thickness B > 257;c/oy (or
507ic/ay), the specimen thickness of, for example, a tough structural steel
would be only about 10-20mm compared with a required thickness of
approximately 100 mm to conduct a valid L.E.F.M. test on such a material.
The most commonly used test methods for obtaining 77c are based on
compact tension or notch bend specimens as illustrated in Fig. 19.18(a) and
(b). A special test-piece geometry is used to permit load-line displacements
to be measured as shown in Fig. 19.18(c).

Fig. 19.18 ‘Test pieces used to

determine 7

Clip gauge

(b) (c)

From these test specimens, 7 is calculated from

Ze
Vie BIW —a) [19:32]

where U = the total energy under the load—displacement diagram (using the
load-line displacement A), W = the specimen width and a = the crack
depth (so (W — a) is the remaining ligament of the test piece).
When using these tests methods it is important not to overlook the fact
that the definition of 7 as a measure of work in fracturing the specimen is
just a convenient simplification of the original #-contour integral. It so
happens that for the notched bend and compact tension specimens, the total
energy represented by the load—displacement diagram is directly propor-
tional to the rate of release of energy as the crack extends a small amount. It
532 MECHANICS OF ENGINEERING MATERIALS

Fig. 1919 Procedure for 7;¢ mea- Load F

surement: (a) load test pieces to
various displacements; () mea- end
sure crack extension; (c) calculate
J for each test piece and plot 7
against da; (d) construct two (b)

curves for J77¢ measurement J

j= 20;da
Best fit to
data points

Jic ro

da da 3
(c) (d)

is this energy release rate which really justifies the linking of the work done
to J. The steps in a typical test procedure are illustrated in Fig. 19.19 and
summarized below.
(a) Load the test piece to different displacement values with a testing
machine in displacement control.
(b) Unload and mark the extent of crack growth.
(c) Fracture each test piece at a low temperature (e.g.-150°C) and measure
the crack extension.
(d) Calculate 7-values from the load against load-line displacement record
using eqn. [19.32] with U being the area under the graph at the
displacement of interest.
(e) Plot a curve of J against crack extension da.
(/) Draw a straight line 7= 2a, da to intersect the best-fit line through the
data points. Here, of is taken as +(ay + o,), where o,, is the tensile
strength of the material.
(g) The critical value of 7 is at the intersection of the two lines.
The most common method of application of the #-contour integral
approach is to convert experimentally determined values of 77c¢ to equivalent
values of G7c or K7c. As would be expected from the similarity of the tests,
it has also been shown that 7 is related to the crack opening displacement
(C.0.D.). The following equations are frequently used to relate the various
fracture toughness parameters:

Kjae =Gie = fic = roy: [19.33]

where

(aw for plane stress

B= 8 a2) for plane strain

Example 19.6
A bend test piece of the type shown in Fig. 19.9 is made from mild steel. The width B
is 20 mm and the depth Wis 25 mm. At the point of crack growth the crack length a
is 15.3 mm and the area under the load—deflection graph is 14.7 J. If the modulus of
the mild steel is 207 GN/m2, calculate its fracture toughness.
 FRACTURE MECHANICS 533

From eqn. [19.32],

ie 2x 14.7
=151,5 k\ym-
~ 0.02(25 — 15.3)10-3
From egn. [19.33],

Kic = V(EF)
= 4/207 107 151-5 10°) Nim 77
= 177MNm?/2

R curves Another method which has been developed to extend the techniques of
L.E.F.M. into the regime of elastic-plastic fracture involves the use of crack
extension resistance curves or R curves. An R curve is a plot of the crack
growth resistance (R) in a material as a function of the actual or effective
crack extension (Aa). Here, R represents the energy absorbed (dU,) per
increment of crack growth (da). It is therefore given by the value of dU,/da
prior to unstable crack growth at the critical point; R has the same units as
the stress intensity factor K.
It was shown earlier that dU,/da, i.e. R, was independent of the crack
length since U, varied linearly with crack length a in Fig. 19.3. This is
approximately true for cracks under plane strain conditions, but in situations
involving larger proportions of plane stress failure, R is no longer
independent of crack length.
Figure 19.20 shows a typical variation of R (calculated using the effective
crack length) with crack extension. A.S.T.M. E561 provides specific
instructions for the measurement of R. If curves of G (corrected to allow for
the size of the plastic zone) are superimposed on the R curves then the point
of instability occurs at the point of tangency between the two types of
curves. The value of G should be calculated from G = Y?o4,(a + 1y)/E.
The solid curve shows the resistance to crack extension (R curve). The
dashed lines show the driving force for crack extension. Both dashed lines

Fig. 19.20 Typical G—R curves

0 Aa Crack extension Aa
534 MECHANICS OF ENGINEERING MATERIALS

are defined by EG = Y’o%,a, but for curve | the stress is greater than for
curve 2. The higher stress causes unstable crack growth when the crack
length reaches the value a,, shown in Fig. 19.20.

19.8 Toughness
In addition to the more recent methods of fracture toughness testing for
measurement by impact design by fracture mechanics analysis, the traditional quality-control method
testing of measuring the energy absorption of a material during fracture has been by
impact bend tests.
The two forms of the test most widely used are the Charpy and Izod.
The principle of the former is shown in Fig. 19.21. The test piece is a square
bar of material, 10 mmx10mmx55mm, containing a notch cut in the

Additional
“Ax weights
Jaws, , adjustable to a
Safety catch 5SSS Knife-edge
Kni on adj
a slab of steel gap of 40 mm
with pin
Steel pendulum
arm
Scale and
pointer Pivot and catch to
move the pointer

Brake arm e@ Machine Pendulum

Jaws frame knife-edge strikes
Leather strap Specimen test piece here and
to brake swings through
pendulum between the jaws

(a) Charpy-type impact testing machine (b) Location of test piece

Fig. 19.21 middle of one face. The notch is a 45° vee, 2mm deep, with a root radius of
0.25mm. The test piece is simply supported at each end on anvils 40 mm
apart. A heavy pendulum is supported at one end in a bearing on the frame
of the machine, and a striker is situated at the other end. The pendulum in
its initially raised position has an available energy of 300J and on release
swings down to strike the specimen immediately behind the notch, bending
and fracturing it between the supports. A scale and pointer indicate the
energy absorbed during fracture.
In the Izod test, Fig. 19.22, the specimen is of circular section, 11.43 mm
in diameter and 71 mm long, or square section, 10 mmx 10 mmx75 mm, and
the Izod notch 1s a vee, as described above for the Charpy test, 3.33 mm and
2mm deep for the round and square specimens respectively. The specimen

Fig. 19.22 Arrangement of test

piece in the Izod test

Pendulum
Test piece [ff striker
 FRACTURE MECHANICS 535

Table 19.3
Izod
impact
energy at
Tensile Reduction room
strength in area temper-
Material Condition (MN/m?) (%) ature (J)

0.1C, 0.3 Mn steel Annealed 377 65 54.2

0.21 C, 0.82 Mn steel Annealed 505 58 40.6
0.5 C steel Normalized 787 63 29.8
Ni-Cr—Mo steel Quenched 840°C,
tempered 650°C = 895 64 114.0
Quenched 840°C,
tempered 500°C 1472 47 29.8
3.0 Ni, 1.0 Cr steel Quenched 840°C 1668 35 19.0
Quenched and
tempered 550°C —- 865 60 93.5
Stainless steel (18-8) Cold rolled 987 - 46.1

is supported as a vertical cantilever ‘built-in’ to jaws up to the notched cross-

section. A pendulum and striker having an initial energy of 166J is arranged
to swing and strike the free end of the test piece with a velocity of 34 m/s
on the same side as the notch and 22 mm above it. The energy absorbed by
the test piece 1s again recorded on a scale.
It is seen that in both tests the notch is on the tension side of bending,
thus initiating fracture. The Charpy and Izod tests have been adopted as
British Standards and details of alternative types of specimen and other
conditions are given in B.S. 131: 1982. There is no direct correlation
between energy values given in each of these tests; however, experimental
results on a wide range of steels show that there is a linear relationship over
the range from 20 to 95 J.
Some typical values of strength, ductility and toughness for steels are
given in Table 19.3. The most notable feature is the effect of the type of heat
treatment given, whether normalizing, quenching or quenching and
tempering. The reason for this is the difference in metallurgical structure
in each case influencing the ductility of the material, and the ease with which
a crack can propagate through the notched bar. The quenched structure is
hard and brittle and can absorb little energy. In the quenched and tempered
structure, a higher tempering temperature gives greater ductility and
therefore higher impact energy. The importance of temperature effects on
fracture mechanics can be easily demonstrated in the Charpy or Izod impact
tests by prior heating or cooling of the specimen.
If the test temperature is varied from ‘high’ to ‘low’, then a metal such as
mild steel, which exhibits the property of brittle fracture, will have a range
of test temperature in which the mechanism of fracture changes from shear
to cleavage. This is known as the transition temperature range, and within this
range will be determined, according to some criterion, a transition
temperature.
536 MECHANICS OF ENGINEERING MATERIALS

—
Fig. 19.23 Transition curves of
brittle—ductile energy against
5 5
Cc

temperature
= eeee
a) ( is
|
= t | i He
643)
2 \i '\ oe
_o
as I 1 as(2)
© ® \ ! © ®
ionw I 1 ae
wa 1 ! wa
] 1
| !
Poe:ad

Temperature ns Temperature
(a) )

Typical transition curves are shown in Fig. 19.23(a) and (4) in which
plastic deformation at, or energy to, fracture is plotted against test
temperature. The first shows what is known as bimodal behaviour in which
there are two distinct branches, one at high energy and temperature and the
other at low energy and temperature, joined by a narrow region (10-20 °C)
of scattered points at upper and lower energy values. The second diagram
shows continuous behaviour where there is a gradual fall from high to low
energy with decrease in temperature. The transition range in this case can be
as much as 100°C from complete shear to entire cleavage.

19.9 Relationship
The major advantage of the Charpy and Izod types of impact tests is that
between impact testing they are quick and convenient tests to perform when compared with the
and fracture mechanics much more demanding test schedule essential to obtain fracture toughness
parameters. Hence there have been a number of attempts in recent years to
correlate K;7c with the Charpy impact strength C,. No general rule has yet
emerged but the research has met with some success. For high-strength
steels a relationship has been found to exist between K;c and C, for the
upper shelf region, and a variation of this can be applied to the transition
temperature region!”. Typical results are shown in Fig. 19.24 and it is felt
that the type of relationships observed may have more general applicability
to the lower-strength steels, particularly for the upper shelf C,-values.

120 [=eaE ES T aaa

Fig. 19.24 Impact strength against
critical stress intensity for mara-
100 fF
ging steel

onro)

m9?)
(MN 40,
Kic

20/7

= li cane
0) 10 20 30 40
Charpy impact strength (J)
 FRACTURE MECHANICS 537

Thus, although Charpy testing says nothing about stress levels, which
has in the past made it impossible to use Charpy toughness values in design
calculations, the correlation between K7c and C, has encouraged attempts to
provide a criterion for material selection based on C,.
Finally, there has been extensive work done on the use of Charpy tests to
obtain fracture mechanics parameters for polymers. In particular a very
useful relationship has been developed!’ between the energy absorbed in
breaking the precracked specimen C, (in J) and the critical strain energy
release rate Gc (in J/m’). This has the form

C, = GicBDé [19.34]
where B = specimen breadth, D = specimen depth and ¢ = a geometry
factor. This permits G7c, which is a material property, to be determined and
hence K7c¢ (from eqn. [19.14]) although it should be noted that for polymers
the modulus F is not a constant.

19.10 Summary
The development of fracture mechanics on the basis that materials contain
crack-like defects from which fast fracture may occur, has led to the
emergence of new design concepts. In essence these assume that the critical
value of the stress intensity factor K7¢ is a material property which may be
used to calculate the maximum defect size for a given stress or the maximum
permissible stress for a given intrinsic defect size. Linear elastic fracture
mechanics may be applied with confidence in situations where fracture
occurs under essentially elastic conditions. The theory can also be extended
to include materials which exhibit a relatively small amount of yielding prior
to fracture, and the quantity (K;c/ay)° provides a measure of the size of
the plastic zone at the crack tip. However, for very tough materials which
exhibit gross yielding prior to fracture the use of L.E.F.M. is no longer
valid. In such cases recourse may be made to methods such as crack opening
displacement (C.O.D.), 7-contour integral or resistance curves which have
emerged in the wake of the success achieved by L.E.F.M.
The traditional toughness tests such as the Charpy and Izod tests still
have an important role as quality-control tests, but have the disadvantage
that they do not provide information which can be used in design
calculations.

References
1. Griffith, A. A. (1921) ‘The phenomena of rupture and flow in solids’,
Phil. Trans. Ry. Soc., A221, 163-197.
2. Irwin, G. R. (1957) ‘Fracture mechanics’, 7. Appl. Phys., 24, 361.
3. Parker, A. P. (1981) The Mechanics of Fracture and Fatigue, E. & F.N.
Spon, London.
4. Paris, P. C. and Sih, G. C. (1965) ‘Fracture toughness and its
applications’, ASTM STP 381.
5. Rooke, D. P. and Cartwright, D. J. (1976) Compendium of Stress
Intensity factors, HMSO, London.
6. B.S. 5447: 1977. Plane Strain Fracture Toughness of Metallic Materials.
7. A.S.T.M. Test Method E399-74. Standard Method of Test for Plane
Strain Fracture Toughness of Metalic Materials.
8. Dugdale, D. S. (1960) ‘Yielding of steel sheets containing slits’, 7.
Mech. Phys. Solids, 8, 100-108.
538 MECHANICS OF ENGINEERING MATERIALS
i ______ ae

9. Wells, A. A. (1961) ‘Unstable crack propagation in metals’, R.Ae.S.

Symposium on Crack Propagation, Cranfield, p. 210-230.
10. B.S. 5762: 1979. Crack Opening Displacement (C.O.D.) Testing.
11. Rice, J. F. (1962) Fracture, Vol. Il, ed. H. Liebowitz, Academic Press,
New York.
12. Rolfe, S. T. and Barsom, J. M. (1977) Fracture and Fatigue Control in
Structures, Prentice Hall, Englewood Cliffs, NJ.
13. Plati, E. and Williams, J. G. (1975) Polymer, 16, 915.
14. Ashby, M. F. (1992) Materials Selection in Mechanical Design,
Pergamon Press, Oxford.

Bibliography
Anderson, T. L. (1991) Fracture Mechanics. Fundamentals and Applications,
CRC Press, Boca Raton, FL.
Hellan, K. (1984) Introduction to Fracture Mechanics, McGraw-Hill,
London.
Hertzberg, R. W. (1989) Deformation and Fracture Mechanics of Engineering
Materials, 3rd edition, John Wiley, New York.
Knott, J. F. and Withey, P. A. (1993) Fracture Mechanics. Worked examples,
2nd edition, Institute of Materials, London.
Parker, A. P. (1981) The Mechanics of Fracture and Fatigue. An Introduction.
E. & F. N. Spon, London.
Powell, G. W. and Mahmoud, S. E. (1986) Metals Handbook, 9th edition,
Vol. 11, Fracture Analysis and Prevention, American Society for Metals,
Metals Park, OH.
Smith, R. N. L. (1991) BASIC Fracture Mechanics, Butterworth—
Heinemann, Oxford.

Problems 19.1. Describe how the critical strain energy release rate, G7c, might be
obtained by experiment.
19.2 Three fracture toughness samples of an aluminium alloy have
identical external dimensions and a thickness of 25mm. During
three tests on the samples the following information was obtained:

Test Sample crack Applied load Sample elongation

length (mm) (kN) (mm)

1 20 185 Sample fractured

2 19.5 120 0.26
3 20.5 120 0.263

If the Young’s modulus and Poisson’s ratio values for the aluminium
are 70 GN/m? and 0.3 respectively, calculate the fracture toughness
of the material.
19.3. A compact tension fracture mechanics specimen containing a crack
53mm in length fails at a load of 59.1kN. Estimate the fracture
toughness of the material, given that the stress intensity K; in the
specimen was
iP
Ky i BWi/2
iA
FRACTURE MECHANICS 539

where the specimen thickness B= 50mm, effective width

W = 100mm and Y is given in the table below. a is the length of
crack in the specimen.

0.50 0.51 0.52 0.53 0.54

9.60 9.90 10.21 10.54 10.89

If the yield stress of the material is 400 MN/m/?, was this estimate a
valid one?
19°4. Calculate the critical defect size for each of the following steels
assuming they are each to be subjected to a stress of 5ay. Comment
on the results obtained.

Steel Yield strength, vy Fracture toughness

(MN/m?) (MN m7?/2)

Mild steel 207 200

Low-alloy steel 500 160
Medium-carbon steel 1000 280
High-carbon steel 1450 70
18% Ni (maraging) 1900 75
steel
Tool steel 1750 30

19s In a fracture toughness test involving three-point bending, the

following information was recorded: support span = 180mm;
specimen thickness = 22.8 mm; specimen width = 44.72 mm; crack
length = 21.92 mm; fracture force = 19.8kN. Establish whether or
not these data are suitable for measuring K7¢ for the material. The
yield stress of the material is 350 MN/m’?.
19.6 The load—deflection graph from a tensile fracture toughness test on a
50 mm thick steel specimen is shown in Fig. 19.25. Also shown (to
scale) is the fracture surface of the broken specimen. If the yield
stress for the steel is 1650 MN/m‘, establish whether the test is valid
for the determination of K7c.
197, A sheet of glass 0.5m wide and 18mm thick is found to contain a
number of surface cracks 3mm deep and 10 mm long. If the glass is
placed horizontally on two supports, calculate the maximum spacing
of the supports to avoid fracture of the glass due to its own weight.
For glass Kjc = 0.3 MN m~*”” and density = 2600 kg/m’.
19.8 The accident report on a steel pressure vessel which fractured in a
brittle manner when an internal pressure of 19 MN/m? had been
applied to it shows that the vessel had a longitudinal surface crack
8mm long and 3.2mm deep. A subsequent fracture mechanics test
on a sample of the steel showed that it had a Kyc value of
75 MN m ”/”. If the vessel diameter was 1 m and its wall thickness
was 10mm, determine whether the data reported are consistent with
the observed failure.
Ie) A simple lifting crane is illustrated in Fig. 19.26. If the tie bar is
found to have 3mm long cracks at the pin-joint hole, calculate the
maximum vertical force at the pulley to avoid fracture of the tie bar.
540 MECHANICS OF ENGINEERING MATERIALS
a

Fig. 1925 = gg
™ r
Load
(MN)

| 1
0.2

Loading
line

Machined
notch

sg
cael,
Tape
gh
esc
a
0.4

Fracture
surface aed
et
SE
a
|Me

lea
ateiaak
Thal

Displacement

The fracture toughness of the tie-bar steel is 75 MNm-?/2.

Comment on the suitability of the equation used to calculate Kyc.

Fig. 19.26

20
ea
3 mm long
cracks

20 mm radius

(a) Using an energy balance, show that the stress intensity for a
long crack running in the axial direction of a thin walled pipe,
Fig. 19.27 is given by

=P VaR

(b) Estimate the stress intensity for a short crack when L < R.
19.11 A large sheet of aluminium alloy is loaded in tension to a stress of
200 MPa. If its yield stress 0, = 400 MPa and fracture toughness
Kic = 42 MN m~*/?, what is the maximum tolerable defect size
(i) using linear elastic fracture mechanics; (ii) using a plastic zone
correction?
FRACTURE MECHANICS 541

19.12 A vessel is protected from excessive pressure by a bursting disc, a

flat circular plate clamped at its edges with a sharp notch machined
across a diameter in the unpressurized side (Fig. 19.28). What depth
of notch woud be required for a bursting pressure g, of 20 MN/m/?.
The plate has radius r = 40mm and its thickness 4 is 4mm. The
plate material has the properties E = 200 GN/m’,
ay = 1500 MN/m’, v = 0.333 and critical COD = 4 pm. You
can assume that the stress intensity round the notch is given by
K =1.120,/na, and Kc = VEoy6,.
The maximum stress acting perpendicular to the notch can be
derived from Eqn. [16.28]. It occurs at r = 0, when

a 3q.r?(1+ v)
a 8h?

19.13. In a chemical plant, a thick walled cylindrical pressure wheel made

from a titanium alloy is subjected to occasional surges in internal
pressure of up to 100 MN/m‘’. During a routine inspection a 2mm
542 MECHANICS OF ENGINEERING MATERIALS

long crack oriented perpendicular to the hoop direction is found in

the cylinder bore. Should the cylinder be taken out of service
immediately or can this crack be tolerated? The internal radius of the
vessel is 10mm while the outer radius is 15 mm. The hoop stress in
the vessel is given by Eqn. [14.36]. Krc = 106 MN m *”.
TO An aluminium alloy plate with a yield stress of 450 MN/m’ fails in
service at a stress of 110 MN/m’. The conditions are plane stress
and there is some evidence of ductility at the fracture. If a surface
crack 20 mm long is observed at the fracture plane calculate the size
of the plastic zone at the crack tip. Calcuiate also the percentage
error likely if L.E.F.M. was used to obtain the fracture toughness of
this material.
A large medium-carbon steel crane hook is thought to contain
penny-shaped internal cracks. If the non-destructive test equipment
used on the hook is not capable of detecting cracks smaller than
20mm diameter, determine the fracture toughness required from
this steel if the safety factor on stress is to be 2. The yield stress of
the steel is 1050 MN/m?.
19.16 A large plate carrying a tensile load contains a hole with two cracks
propagating from the hole in a direction perpendicular to the applied
load, Fig. 19.29.
(a) Estimate the stress a9 at which the plate will fracture by both
the following methods. (i) Assume the stress intensity in the
cracks is the same as the stress intensity in the edge of an
infinite plate whose average stress is the same as the peak stress
round a hole in an uncracked plate. (11) Assume that the only
effect of the hole is to make the two cracks at opposite sides of
the hole appear as one long continuous crack.
The stress intensity for a crack of length a in the edge of a
plate is K = 1.120\/za. For a central crack of length 2a in the
middle of an infinite plate K = o\/7a.
(6) At what ratio r/L do both approaches give the same estimate?
Which approach is more likely to be correct for r << L?

Fig. 19.29

Loe As a result of a C.O.D. test on a medium-carbon steel, the following

data were recorded: sample width = 25.03 mm; V, = 0.33 mm;
sample depth = 49.98mm; F’, = 80kN; crack length = 24.00 mm;
z = 2.5mm. Calculate (a) the fracture toughness of the material and
 FRACTURE MECHANICS 543

(d) the percentage error in using L.E.F.M. to calculate this value.

For the steel, E = 207 GN/m’, oy = 950 MN/m’, v = 0.3.
19.18 In a three-point bend test on a mild steel sample the load at the point
of crack growth and the final crack length were noted as 70 kN and
26mm respectively. At the point of crack growth the area under the
load—deflection graph was 32.6J. If the specimen width (W) and
thickness (B) were 49.98 mm and 25.05 mm respectively, calculate
the fracture toughness of the mild steel. If a C.O.D. gauge (width
z = 2.5mm) had been attached to this specimen calculate the plastic
component of the gauge reading which you would have expected to
record. For the mild steel, ay = 240 MN/m’, E = 207 GN/m’
and P= 033:
19719 A bench-top impact testing machine for plastics as shown in Fig.
19.30. The pendulum weighs 4kg and its centre of gravity is at A.
When there is no specimen in position it is found that the pendulum
swings through to an angle 0 = 129°. When there is a specimen in
position the pendulum swings through to 6 = 89°. Calculate (a) the
windage and friction losses in the machine and (4) the energy
absorbed in breaking the specimen.

Fig. 19.30
e. ;
x eae ie rae
RS Initial Position
CHAPTER
Fatigue

In the early part of the nineteenth century the failure of some mechanical
components subjected to nominal stresses well below the tensile strength of the
material aroused some interest among a few engineers of that time. The fact that
puzzled these early engineers was that a component such as a bolt or a shaft made
from a ductile material such as mild steel could fracture suddenly in what appeared
to be a brittle manner. There was no obvious defect in workmanship or material, and
the only feature common to these failures was the fact that the stresses imposed
were not steady in magnitude, but varied in a cyclical manner. This phenomenon of
failure of a material when subjected to a number of varying stress cycles became
known as fatigue, since it was thought that fracture occurred owing to the metal
weakening or becoming ‘tired’. The first real attack on this problem was made by the
German engineer Wohler in 1858. Since then a great deal of research has been
conducted on fatigue of metals, and in more recent times other materials also, and
although this work has resulted in an ever-increasing understanding of the problem
there is as yet no complete solution. What has been established is that the early
theories that the metal becomes ‘crystalline’ or brittle under the action of the cyclic
load are erroneous. It is now well known that a fatigue failure starts on a microscopic
scale as a minute crack or defect in the material and this gradually grows under the
action of the stress fluctuations until complete fracture occurs.
It has been estimated that at least 75% of all machine and structural failures have
been caused by some form of fatigue. It is therefore evident that every engineer
should be aware of this phenomenon, and have some idea of its mechanics and what
can be done to minimize or avoid the risk of this type of failure. It is astonishing how
many design engineers ignore warnings about the possibility of fatigue failures and
include in their design geometrical shapes which cause stress concentrations and
initiate fatigue failures. Indeed even when the danger of fatigue failure is recognized
it is not always possible to avoid owing to the many stages of manufacture which a
component may go through and which are outside the direct control of the designer.
There have been reports of situations where the designer specified non-destructive
testing of components to search for flaws that would initiate fatigue failures, and
when none were found the test engineer used a metal stamp to mark the component
as having passed inspection. Then in service fatigue cracks initiated from the stamp
mark!
Figure 20.1(a) shows a typical situation where fatigue failures can arise owing to
geometrical configurations, and Fig. 20.1(6) illustrates the appearance of the
fracture surface in such cases.

20.1 Forms of stress

cycle Throughout the working life of a component subjected to cyclical stress the
magnitude of the upper and lower limits of cycles may vary considerably as
shown in Fig. 20.2. However, it has been general practice when considering
fatigue behaviour to assume or employ a sinusoidal cycle having constant
upper and lower stress limits throughout the life.
 FATIGUE 545

Fig. 20.1 Fatigue cracks in an

engine crankshaft

Initiation

Fatigue crack
growth striations
ey

Fast fracture
area

(b)

Fig. 20.2 Variable stress spectrum Stress | Tension

Compression

Figure 20.3 shows a general type of stress cycle, which is termed

fluctuating, in which an alternating stress is imposed on a mean stress. This
cycle can consist of any combination of upper and lower limits, within the
static strength, which are both positive, or both negative.

Fig. 20.3 Stress parameters 4 steds Oye

Stress amplitude S,

Range of
stress 2S,

Maximum
stress

stress
Minimum
stress
min

(@) Time
546 MECHANICS OF ENGINEERING MATERIALS

Fig. 20.4 ‘Typical stress—time cy- Stress

cles

When one stress limit in a cycle is positive and the other negative as
shown in Fig. 20.4(a) it is known as a reversed cycle. There are two
particular cases of the fluctuating and reversed cycles which arise frequently
in engineering practice. The first is a fluctuating cycle in which the mean
stress is half the maximum, the minimum being zero, as shown in Fig.
20.4(b). The second is a symmetrical reversed cycle as shown in Fig. 20.4(c)
in which the mean stress is zero and the upper and lower limits are equal
positive and negative.
The relationship between the various stress values is of some importance.
The mean stress, S,,,* is half the algebraic sum of the maximum stress, Sina,
and the minimum stress, S,i,. The range of stress, 2S,, is the algebraic
difference between Sy, and Syn. The ratio of the minimum to the
maximum stress is termed the stress raiio, R. Hence

Sy = Simax + Smin (20.1]

, t

2S, = mae 2 ae [20.2]

iv :

Kee 20.3
It should be noted that the foregoing has only considered stresses as being
positive or negative in a general sense. In practice, fatigue can be generated
in direct stress due to axial loading or bending or shear stress due to cyclic
torsion or any combinations of these.
20.2 Test methods
Fatigue failures occur most often in moving machinery parts, e.g. shafts,
axles, connecting rods, valves, springs, etc. However, the wings and fuselage
of an aeroplane or the hull of a submarine are also susceptible to fatigue
failures because in service they are subjected to variations of stress. As it is
not always possible to predict where fatigue failures will occur in service and
because it is essential to avoid premature fractures in articles such as aircraft
components, it is common to do full-scale testing on aircraft wings, fuselage,
engine pods, etc. This involves supporting the particular aircraft section or
submarine hull or car chassis in jigs and applying cyclically varying stresses
using hydraulic cylinders with specially controlled valves.

* The symbol S is being used for stress to conform with B.S. 3518: Part I:
1984; however, @ is also a recommended symbol (I.S.O.).
 BATIG WE 547

Fig. 20.5 Rotating bending fatigue

test
Bearing

Specimen

Belt drive

Laboratory tests are also carried out on particular materials to establish

their fatigue characteristics and to study factors such as their susceptibility
to stress concentrations. The earliest and still widely used method of fatigue
testing of laboratory specimens is by means of rotation bending. A
cylindrical bar is arranged either as a cantilever (Fig. 20.5) or as a beam in
pure bending. It is then rotated while subjected to a bending moment and
hence each fibre of the bar suffers cycles of reversed bending stress. A
bending fatigue test can also be arranged, without rotation, by alternating
bending in one plane. An advantage of this test over the former is that a
mean bending stress can be introduced.
Axial load fatigue tests, although requiring more sophisticated and
expensive equipment, have the significant advantage that they subject a
volume of material to a uniform stress condition as opposed to the stress
gradient in bending.
There are three main principles of operation for axial load fatigue
machines. Firstly, there is the system whereby an electromagnet in series
with the specimen is rapidly energized and de-energized, thus applying
cyclical force. To reduce the amount of power input to the electromagnet to
achieve the required load amplitude, a resonant vibration system is
employed using springs. Another method of achieving a resonant condition
of springs in series with a specimen is by mechanical means using a small
out-of-balance rotating mass. The third arrangement for producing
fluctuating forces on a specimen is by means of hydraulic pulsation.
Frequency is an important factor in fatigue testing, owing to the large
number of cycles it is necessary to achieve at lower stress ranges, and
consequent time factor in obtaining fatigue data. Bending fatigue machines
generally run in the range from 30 to 80 Hz. The electrical excitation push—
pull machines operate between 50 and 300Hz, while the mechanical
excitation is usually restricted to about 50 Hz. The hydraulic machines have
a frequency, often variable,in the range from about 1 to 50Hz. The
frequency of an actual test is generally controlled by the stiffness of the
specimen, i.e. high frequency can only be achieved for high stiffness and low
amplitude of deformation. The testing capacity of axial load fatigue
machines varies typically from 20 to 2000 kN.
548 MECHANICS OF ENGINEERING MATERIALS

Cyclic torsion or combined bending and torsion fatigue machines

generally operate on the principle of direct mechanical displacement of the
specimen by a variable eccentric, crank and connecting-rod system.
Depending on the capacity of the machine, frequencies can vary from 16
to 50 Hz.

20.3 Fatigue data

The most readily obtainable information on fatigue behaviour is the
relationship between the applied cyclic stress S and the number of cycles to
S—N curves
failure N. When plotted in graphical form the result is known as an S—N
curve. Figure 20.6 shows the three ways of plotting the variables: (a) S
against N; (b) S against logy9 N, (c) logio S against logiyg N. The number of
cycles to failure at any stress level is termed the endurance, and this may vary
between a few cycles at high stress and as much as 100 million cycles at low
cyclic stresses, for a complete S—N curve. It is immediately apparent that a
linear scale for endurance N over the whole range is impractical. It is for this
reason that the semi-logarithmic or double-logarithmic plots (4) and (c) are
employed. The former, (4), is the most widely used method of presentation.

=
Fig. 20.6 Typical methods of
presenting fatigue curves n
n
Y )
2 w | Low me alta) oe
2 =}
S
S)

! | L | "eS eS ed
O24 One (Sin LO) 1O°N1.02 A104 10 1.08 10° 102 104 10° 108
Cycles to failure N (millions) Log, NV Log,N

(a) (b) (c)

High-endurance fatigue This relates to endurances from about 10* cycles to ‘infinity’ (or 50 x 10° in
terms of a laboratory test). The S—N curve in Fig. 20.6(d) falls rapidly from
10* cycles followed by a ‘knee’ after which, depending on the material, the
curve either becomes parallel to the N-axis or continues with a steadily
decreasing slope. Most steels and ferrous alleys exhibit the former types of
curve, and the stress range at which the curve becomes horizontal is termed
the fatigue limit. Below this value it appears that the metal cannot be
fractured by fatigue. In general, non-ferrous metals do not show a fatigue
limit and fractures can still be obtained even after several hundred million
cycles of stress. It is usual, therefore, to quote what is termed an endurance
limit for these metals, i.e. the stress range to give a specific large number of
cycles, usually 50 x 10°. Typical S—N curves for an aluminium alloy and a
steel tested in air for a plain condition, i.e. no stress concentration, under
axial loading, are given in Fig. 20.7.

Low-endurance fatigue For many years the low-endurance region (<3000 cycles) of fatigue was
ignored because the bulk of engineering cyclic stress situations had to have a
working life ranging from several millions to infinity before failure. Hence
the important information was contained in the lower stress—higher cycles to
failure part of the curve. However, a low cycle to failure does not necessarily
mean a short lifetime because it is a function of cyclic frequency. Hence an
 FATIGUE 549

Fig. 20.7 Aluminium alloy 24S-T3 350

reversed axial stress fatigue curve
(x); mild steel reversed axial
300
stress fatigue curve (MB)

250

200

150
(MN/m2)
of
Semi-range
stress

Cycles to fatigue fracture

Fig. 20.8 Plastic strain hysteresis

loop

df Strain

aircraft fuselage is only pressurized once every flight and so it may take years
to accumulate 1000 cycles and a pressure vessel may work for 25 years
before achieving 1000 cycles of cleaning and inspection.
Failure at low endurances results from stresses and strains which are
high and will result in marked plastic deformation (hysteresis) in every
cycle, Fig. 20.8. Owing to strain-hardening (or softening) effects it becomes
of some significance whether the cycle is of controlled load (stress)
amplitude or controlled strain amplitude, a problem which does not arise at
long endurance and hence low stress. A good example of strain cycling,
which came into prominence with the design of nuclear plants and aerospace
vehicles, is the effect of repeated thermal changes in a component.
At low endurances the S—N and e—N curves take the form shown in Fig.
20.9. It is usual to consider the S—N curve as starting from a point at a
quarter cycle representing a tension test or single pull to fracture. Likewise
the e—N curve is started at a quarter cycle using the true fracture ductility in
simple tension as the ¢-value. Much work has been concentrated on strain
cycle testing, and it has been found that, if the plastic strain range €,, or
width of the hysteresis loop, is determined during the life of the specimen,
550 MECHANICS OF ENGINEERING MATERIALS

Fig. 20.9 Typical stress cycle and

strain cycle fatigue curves at low
endurance

far)

cyclic
strain
Cyclic
stress
S€,
or
ep)-

107 10° 101 102 103 104 10°

Cycles to fracture, Log,.N

101
Fig. 20.10 Relationship of log €,
with log N
10°
Pie

‘oo
Ss
10-4

10-2

10-3 See ee Be 1 1 —S L

10-4 10° 101 102 103 104 10°

Log N

then a relationship of the form ¢, N° = constant holds for most metals up to

10° cycles, and that on a graph of loge,—logN a straight line results, Fig.
20.10. It has further been demonstrated that a is between —0.5 and —0.6 for
many metals at room temperature, and the constant is broadly related to the
ductility of the material, i.e. the greater the latter, the larger the constant
term.
Thermal strain cycling, 1.e. cyclic strains induced by changes in
temperature, appears to yield the same results as if the specimen or
component were kept at a constant temperature of a value equal to the upper
limit of the previous temperature cycle, and the cyclic strains induced
mechanically.

Statistical nature of Fatigue failure of materials has long been recognized as a random
fatigue phenomenon. Thus, even in carefully controlled experiments, at any
selected stress amplitude there is often a larger scatter in the number of
cycles to failure as seen in Fig. 20.7. The reason for this is that, although on
a macroscopic scale it is convenient to consider a material as being a
homogeneous continuum, on a microscopic scale this is far from being the
case. A metal, for example, is known to consist of a random distribution of
internal defects such as microcracks, dislocations and inclusions contained
within a network of grains which have randomly oriented slip planes and
 FATIGUE 551

grain boundaries. Thus when the same cyclic stress amplitude is applied to
nominally identical specimens it is extremely unlikely that each sample will
fail after the same number of stress cycles. This is because, on a microscopic
scale, the conditions which the fatigue cracks experience as they propagate
through each specimen will be quite different.
Thus although fatigue data is usually presented as a single line on an S—
N curve (Fig. 20.7) it should be realized that this line does not predict the
exact fatigue life at a particular stress amplitude: S—N curves are usually
plotted from a large number of fatigue test results which have been analysed
by some type of statistical method. In the simplest case the line drawn
represents the 50% probability of failure. This means that at stress S$; in
Fig. 20.11, 50% of the samples tested will have failed before cycles Nj. In
some cases other probability lines will also be drawn. Figure 20.11 shows the
1% and 99% probabilities of failure and this type of graph is sometimes
called a P—S—N diagram.

Fig. 20.11 Probability fatigue curve

Stress

50%

Use of fatigue data for In general engineering, components subjected to cyclical stresses will be
design designed on an ‘infinite’ life basis. The design stress range will therefore be
related to a choice of material having a fatigue or endurance limit, with a
suitable safety factor, in excess of that working stress requirement.
There are a number of factors that influence the magnitude of the fatigue
limit which will affect design stresses accordingly and these will be discussed
later.
Limited or finite life fatigue relates to endurances principally in the
range 10°—10’ cycles and it is here that design can benefit from analysis of
fatigue crack propagation. The micromechanisms of fracture and the
fracture mechanics of crack growth in fatigue are introduced in the next two
sections.
Finally low-endurance—high-strain fatigue is a rather specialized area of
design involving cyclical plasticity which is beyond the scope of this short
chapter.

Example 20.1
A mild steel shaft has a pulley mounted at each end and is supported symmetrically
in between two bearing housings. When rotating under load the shaft will be
subjected to a cyclical bending moment of -+-1.2 KN m. Using a safety factor of 2 and
the data in Fig. 20.7, determine a suitable diameter for the shaft for infinite fatigue
life.
552 MECHANICS OF ENGINEERING MATERIALS

From Fig. 20.7 the fatigue limit for infinite life is approximately +190 MN/
m? and using a factor of 2 the design stress will be +95 MN/m’.
The bending relationship is

my
ee
which for a shaft diameter of D gives

ges {jem 10° x 32

= ™D3
Hence D* = 128.6 x 10~° m’, or D = 50.48 mm.

20.4 Micromechanisms
A great deal of research has been devoted to a study of the mechanism of
of fatigue: initiation and fatigue, and yet there is still not a complete understanding of the
propagation phenomenon. It is not an easy problem to handle theoretically or
experimentally, since the process commences within the atomic structure
of the metal crystals and develops from the first few cycles of stress,
extending over thousands or millions of subsequent cycles to eventual
failure.
The fatigue mechanism has two distinct phases, initiation of a crack and
the propagation of this crack to final rupture of the material. One of the
earliest (1910) and classic metallurgical studies of initiation was made by
Ewing and Humfrey, who examined the polished surface of a rotating
bending specimen at intervals during its fatigue life. They observed that
above a certain value of cyclic stress (the fatigue limit) some crystals on the
surface of the specimen developed bands during cycling. These bands are
the result of sliding or shearing of atomic planes within the crystal and are
termed s/ip bands. With continued cyclic action these slip bands broaden and
intensify to the point where separation occurs within one of the slip bands
and a crack is formed as shown in Fig. 20.12(a). In the 1950s Forsyth
discovered that a process of intrusion and extrusion at the surface could
cause a crack to be formed as illustrated in Fig. 20.12(4) and (c). This crack
initially develops along the slip plane of the grain in which it was formed

Fig. 20.122 Stages in the develop-

ment of a crack along a slip plane

Stage II

(c)
 FATIGUE 553

Fig. 20.13 How fatigue cracks grow New surface

Tt5/2
(from Ashby and Jones!)

AO

(a) (b)
(Stage I growth) but eventually propagates across other grains. This Stage II
growth occurs, not as a consequence of any progressive structural damage,
but as a result of the stress concentration effect at the crack tip as it becomes
sharp during unloading.
The mechanism of crack growth is shown in Fig. 20.13. In a material
which is free from defects, the tensile stress produces a plastic zone which
causes the crack tip to stretch open by an amount 6. This creates new
surface at the crack tip. During the compressive part of the cycle, the crack
is squeezed shut and the new surface folds. This causes the crack to advance
by an amount approximately equal to 6. This process is then repeated
during each cycle so that the crack growth rate da/dN is approximately
equal to 6.! This mechanism is illustrated in Fig. 20.13(a).

Fig. 20.14 Stages of crack growth

10+ ie a

x
Grains
Ad
e. 10> fr Ak |
Ss
‘S Inclusions
2
< y
oe A
x —

S)
Precipitates

10
ai xX J

Dislocation
N= number of cycles
N,.= number of cycles to failure
10-8 f i i Y

10-3 10-2 Os 10°

554 MECHANICS OF ENGINEERING MATERIALS
po SS ee —E—E—E—E— Eee eee

In a real engineering material there will always be microscopic defects.

As a result of these, holes will form within the plastic zone. These will link
up with one another and with the crack tip. This is illustrated in Fig.
20.13(b). The crack will thus advance a little faster than before because it is
aided by the presence of the holes.
There have been a number of theories on the initiation and propagation
of fatigue cracks. In recent times it has been suggested that fatigue is
initiated by the movement of dislocations. A dislocation is a fault or
misplacement in the atomic lattice of the metal. Microscopic plastic
deformation allows dislocations or vacancies to ‘move’ through the atomic
lattice, and it is thought that coagulation of dislocations forms the
beginnings of a crack.
Others have suggested that dislocations are much too small to have any
real effect and that it is more likely that cracks develop from intrinsic defects
in the material. These may be of the order of 0.5 wm in size and Fig. 20.14
gives an idea of the scale of this relative to microstructural dimensions. It
also indicates the rate at which a crack might propagate through a material.
Propagation of a fatigue crack is a complex phenomenon depending on the
geometry of the component, the material, the type of stressing, and the
environment. Propagation can occupy as much as 90% of the total
endurance; hence movement is relatively slow. Some experimental evidence
reveals a discontinuous progress of the crack: it is moving for some cycles
and stationary for others. Such investigations have also shown that there is a
fatigue crack growth threshold below which cracks can exist in a material but
will not propagate. This can now be explained in terms of the concept of the
stress intensity factor K which was introduced in Chapter 19, and the
emergence of fracture mechanics has cast new light on fatigue crack growth
phenomena. Fracture mechanics also permits life expectancy to be predicted
for a material or structure containing a crack-like defect of known size—
something not possible using the S—N curve approach since this does not
separate out the initiation and propagation phases.

20.5 Fracture mechanics

Fracture mechanics can only be applicable to fatigue after the crack initiation
for fatigue
phase to enable crack growth to be predicted. In Chapter 19 it was shown
that the state of stress in the vicinity of a crack in an infinite body could be
expressed in terms of the stress intensity factor, K, where

K= Yo./ (aa) [20.4]

In cyclic loading, K varies over a stress intensity range AK where

16 = Nore, = Binuin

INI AY (SS pagrie t anes ais) [20.5]

using S' in place of o.

The simplest and most widely used expression which relates the range of
stress intensity factor to the crack growth rate during cyclic loading is the
Paris-Erdogan equation. This takes the form
da
aN C(AK) [20.6]
 FATIGUE 555

10-3
Fig. 20.15 Crack growth data for an
aluminium alloy
10°

5 ANORL
gz

10°

40-11 : recoeanela anne

Table 20.1 Fracture mechanics

constants for a range of materials Material AKru m (Gs
(MN m~?/) (x10)

Mild steel 3.2-6.6 3.3 0.24

Structural steel 2.0—5.0 3.85-4.2 0.07-0.11
Structural steel in sea water O= IES 35) 1.6
Aluminium 1.02.0 2.9 4.56
Aluminium alloy 1.02.0 2.6-3.9 3-19
Copper 1.8-2.8 3) 0.34
Titanium 2.0-3.0 4.4 68.8

* Units of C will give:'da/dN in m/cycle when AK is in MNm~*/?,

where C and m are material constants although their values will depend on
factors such as the nature of the environment and the level of residual
stresses. Typical values of C and m for a range of materials are given in
Table 20.1. The way in which the crack growth rate varies during cycling is
shown in Fig. 20.15. This illustrates that at the lower end there is a
threshold value of stress intensity range AK 7y below which the crack will
not propagate. It has been found that for most metals AKyy is
approximately proportional to the elastic modulus. At the upper end, the
crack growth rate tends towards an infinitely large value. As the crack grows,
Kynax increases and failure will occur when this exceeds the fracture
toughness of the material K7c or when the remaining ligament of material
ahead of the crack tip fails by plastic collapse.
As eqn. [20.6] expresses the rate at which a crack grows under specified
cyclic stress conditions, it is possible to use this to predict fatigue lifetimes.
From egns. [20.5] and [20.2]
UN GENT
ey Cie)
where Sp is the range of variation of tensile stress.
Substituting in eqn. [20.6],

an
da
= ClY- Sav (na)] m [20.7]
556 MECHANICS OF ENGINEERING MATERIALS
i ________ ae

Letting the initial crack size in the material be ao, then the number of cycles
Ny to cause this crack to grow to a size ar at which failure would occur in
one application of stress can be obtained by integrating eqn. [20.7]

2 1—m/2 1—m/2
Nr = — 20.8

assuming that m # 2 and Y does not vary with a.

In practice it has been found”? that the mean stress intensity will also
influence the crack growth rate. A large mean stress which raises S,,,, and
S‘nin to near the yield stress of the material will increase the crack growth
rate. If S,,i, is compressive the simplest analysis is that the crack will close
when the stress is compressive and so S;nin should be taken as zero.

Example 20.2
A support bracket is welded to a backing plate as shown in Fig. 20.16. A fluctuating
force in the coupling rod causes a stress variation of +50 MN/m? at the weld. Using
the crack growth data in Fig. 20.15 calculate the maximum size of defect which
could be tolerated in the weld.

Fig. 20.16 ai

From Fig. 20.15 the threshold value of the stress intensity factor AK yy is
1.65 MNm-3/?;
AKry = YASS) J (ra)

In this case, assuming an edge crack in the weld, then Y = 1.12 and AS
is taken as 50 MN/m? since it is only the tensile part of the cycle which
causes fatigue crack growth:
1.65 = 1.12(50)/(7a0)
ag = 0.27 mm

Example 20.3
The blades of a turbine rotor are fitted into aluminium alloy discs on the rotor as
shown in Fig. 20.17. If during the assembly of the system a 0.1 mm deep scratch is
made in the surface of the disc as indicated, (a) calculate how many stress cycles
the disc can withstand before fatigue failure occurs and (b) if the rotor is required
to undergo at least 2000 cycles in service, by what factor should the rotation speed
be increased in a test to verify that any cracks in the disc are too small to grow to a
critical size in service? The rotation of the turbine causes a stress of 350 MN/m2 at
 FATIGUE 557

Fig. 20.17

Turbine motor

the plane of the scratch and the crack growth data for the disc material is given in
Fig. 20.15.

(a) From Fig. 20.15 the following information can be obtained:

7da =
es LOD =i AK) 3.54

and

Kic = 35MNm~?/?
From Kjc it is possible to calculate the size of crack which would
cause failure in a single stress application. This will be the value of ar in
eqn. [20.8]:

1 (Kic\’ 1 areen\*
a=— (38)
{=~} =—- |——— ] =
7 Ge x =m) 2.54
mest
It is evident that as the actual crack depth is only 0.1 mm there is no
danger of fast fracture during running of the rotor, even at full speed.
However, repeated ON/OFF cycles for the rotor will cause the crack to
grow. In this case the number of cycles to cause failure can be predicted
from eqn. [20.8]:

—m)(a“S — -EAa”)
Ne =
2 LA
!~ C(VSp)"1/2(2
2 AC SO Ie ales ag
4 x 10-1(1.12 x 350)? **!-77(—1.54)
= 3117 cycles

(b) The critical crack size for failure in service has been found to be
2.54mm. A crack of length ag will grow to this length in 2000 cycles if
2 1—m/2 1—m/2
Nr = : —
My C( vi Sp)" nml2(2 ov m) (a, a )

or

2000 = —2.83[(2.54 x ite ee a0)

558 MECHANICS OF ENGINEERING MATERIALS

from which

ay = 0.168 mm

The stress necessary to cause fracture with a crack of this length is

Kic 35
Si = = = 1361 MN/m’
YJ(na) 1.12\/(m x 1.68 x 10-*) a
Since the stress in a rotating disc is proportional to (angular velocity)” (see
Chapter 14), the required overspeed is a factor x of
SP /et,
ON Gaim
20.6 Influential factors
It is quite common to think of fatigue in terms of the range of cyclic stress;
Mean stress however, the mean stress in the cycle has an important influence on fatigue
behaviour. There are two obvious limiting conditions for the mean and
range of stress. One is for the mean stress equal to the static strength in
tension or compression, whence the range of stress must be zero. The other
condition is for zero mean stress and a stress range equal to twice the fatigue
limit for fully reversed stress. Between these boundaries there is an infinite
number of combinations of mean and range of stress. It is obviously
impossible to study the problem experimentally completely; consequently,
empirical laws have been developed to represent the variation of mean and
range of stress, in terms of static strength values and the fatigue curve for
reversed stress (zero mean). This latter condition is the most widely used for
obtaining experimental data, principally because of the simplicity of the
rotating beam test.

Fig. 20.18 Diagrams for mean

stress against semi-range of stress

of
Semi-range
S,
stress

Mean stress S_,

If a diagram is plotted of the semi-range of stress as ordinate and the

mean stress as abscissa, known as an S,—S,, diagram, as in Fig. 20.18, then
the limiting conditions are S,,=0 and S, = Sp, the fatigue limit for
reversed stress, and S, =0 and S,,=S,, the tensile strength of the
material. Between these limits it is required to have a line which represents
 ReAGG WE: 559

the locus of all combinations of S, and S,,, which result in the same fatigue
endurance. A straight line joining these pairs of co-ordinates represents one
empirical law known as the modified Goodman relationship. Algebraically
this is given by

S, =Sp (1= = (20.9]

u

Another relationship is obtained by joining the limiting co-ordinates by a

parabola known as the Gerber parabola. This is expressed algebraically in
the above symbols as
z

Sas (=) [20.10]

A more conservative line for design purposes was proposed by

Soderberg using a yield stress instead of the tensile strength in the
Goodman relationship:

iS
S, = o(Sp|1—-—
=) [20.11]
20.11

The simplest conclusion that can be reached from the foregoing is that
an increase in tensile mean stress in the cycle reduces the allowable range of
stress for a particular endurance. This applies similarly for direct stress or
shear-stress (torsional) fatigue.
Compressive mean stresses appear to cause little or no reduction in stress
range, and some materials even shown an increase; consequently the S,—S,,
diagram is not symmetrical about the zero mean stress axis.

Geometrical factors
1. Stress concentration
Probably the most serious effect in fatigue is that of stress concentration. It
is virtually impossible to design any component in a machine without some
discontinuity such as a hole, keyway, or change of section. These features
are known as stress raisers or sources of stress concentration. This concept
was introduced in Chapter 12, and it was explained that under static loading
in the elastic range the local peak stress at a notch or discontinuity is raised
in magnitude above the nominal stress on a cross-section away from the
notch. The theoretical elastic stress concentration factor K, for a notch is
defined as the maximum stress at the notch divided by the average stress on
the minimum area of cross-section at the notch.
For ductile metals, static stress concentration does not reduce the
strength owing to redistribution of stress when the material at the notch
enters the plastic range. However, under fatigue loading the position is very
different, since a fatigue limit tends to correspond with the static elastic limit
of the material. Consequently, the fatigue limit for the material having the
peak stress at the discontinuity would correspond to the static elastic limit,
and hence the fatigue limit based on the nominal stress would be reduced by
a factor dependent on the elastic stress concentration factor. In practice this
is generally not so and the notched fatigue strength is rather better than the
‘plain’ fatigue strength divided by K,. This has led to what is termed the
560 MECHANICS OF ENGINEERING MATERIALS

fatigue strength reduction factor, which is defined as

plain fatigue strength at N cycles or fatigue limit
Ky ~ notched fatigue strength at N cycles or fatigue limit
and generally K, > Ky > 1 as shown in Fig. 20.19.

Fig. 20.19 Comparison of plain and 100 ataes Sag T ae.

notched fatigue curves

80 F
Pe
= 60 +
= Un-notched
wn

& 40 + Notched (actual) K, = 1.3

a (theor.) k, =2
20 ;

i et
a = 1 yan #
Sed

Log N

Another way of representing material reaction to stress concentration in

fatigue is by means of a notch sensitivity factor q, which is defined in terms of
the theoretical stress concentration factor and the fatigue strength reduction
factor. Thus
eae
q Rol [20.12]
and for a component or specimen which yields K+ in fatigue equal to K,, the
factor g=1 shows maximum sensitivity. Where there is no strength
reduction, Ky = 1 and q becomes zero showing no sensitivity.
Geometrical discontinuities such as notches, holes, etc., have little or no
strength reduction effect at low endurances for stress cycling conditions.
When the cycles are low, the material fatigue strength is closer to the static
strength. Consequently, if the notched static strength is higher than the
plain tensile strength, the notched fatigue strength at the upper end of an
S—N curve may be higher than the plain fatigue strength as shown in
Fig. 20.19.

Fig. 20.20 Size effect

 AT VG Wi 561

De DERE.
It has been found that for geometrically similar components a large size has a
slightly lower fatigue strength than a small size in cyclic bending or torsion.
As this size effect is not found in uniaxial fatigue, it has been primarily
attributed to the different gradients of stress and strain as shown in
Fig. 20.20.

Example 20.4 Ray peas

P The stepped shaft shown in Fig. 20.21 is subjected to a steady axial pull of 50 kN
and a uniform bending moment M. If the yield strength of the rotor material is
300 MN/m2 and the fatigue limit in reversed bending is 200 MN/m2, calculate the
maximum value of M to avoid fatigue failure in the rotor. K; for the fillet radius is
1.55 and the notch sensitivity factor q— 0.9.

Fig. 20.21 37.5 mm

50 kN ER EG os Ore eS MEN ED, (Rae 50 kN

ESSA a4
MMeanitstresssa as t= LS MN/m?
A (25)
; My Mx 12.5 x 64
Alternating stress, S, = =
i m(25)
= M x 6.52 x 10°* MN/m’
Here, Ky is obtained from eqn. [20.12],

from which Ky = 1.5.

Applying Soderberg’s rule to allow for the mean stress,

200 101.8
6.52 x 10 4M = = (1 a

M = 135100 N mm

Environmental effects
1. Temperature
Many components are subjected to fatigue conditions while working at
temperatures other than ambient. Components in aircraft at high altitude
may experience many degrees of frost, while a steam or gas turbine will be
running at several hundred degrees Celsius.
562 MECHANICS OF ENGINEERING MATERIALS

Tests on a number of aluminium and steel alloys ranging down to

—50°C, and lower in some cases, have shown that the fatigue strength is as
good as and often a little better than at normal temperature (+20 C).
On the other hand, fatigue tests at higher temperatures show little or no
effect up to about 300°C, after which for steels to about 400 °C, there is an
increase in fatigue strength to a maximum value, followed by a rapid fall to
values well below that at +20°C. A further interesting feature is that a_
material having a fatigue limit characteristic at ambient temperature will lose
this at high temperatures, and the S—N curve will continue to fall slightly
even at high endurances, and an endurance limit has to be quoted.
Above a certain temperature there is interaction between fatigue and
creep effects (see Chapter 21), and it is found that up to, say, 700°C for
heat-resistant alloy steels, fatigue is the criterion of fracture, whereas at
higher temperatures, creep becomes the cause of failure. This has led to the
use of combined fatigue—creep diagrams, similar to Fig. 20.18 where the
cyclic stress is plotted against the steady stress which produces failure in a
specified endurance at a particular temperature.

2. Corrosion
Fatigue tests are generally conducted in air as a reference condition, but in
practice many components are subjected to cyclic stress in the presence of a
corrosive environment. Corrosion is essentially a process of oxidation, and
under static conditions a protective oxide film is formed which tends to
retard further corrosion attack. In the presence of cyclic stress the situation
is very different, since the partly protective oxide film is ruptured in every
cycle allowing further attack. A rather simplified explanation of the
corrosion fatigue mechanism is that the microstructure at the surface of the
metal is attacked by the corrosive, causing an easier and more rapid initiation
of cracks. The stress concentration at the tips of fissures breaks the oxide
film and the corrosive in the crack acts as a form of electrolyte with the tip of
the crack becoming an anode from which material is removed, thus assisting
the propagation under fatigue action. It has been shown that the separate
effects of corrosion and fatigue when added do not cause as serious a
reduction in strength as the two conditions acting simultaneously.
One of the important aspects of corrosion fatigue is that a metal having a
fatigue limit in air no longer possesses one in the corrosive environment, and
fractures can be obtained at very low stress after hundreds of millions of
cycles.
A particular form of corrosion fatigue may occur in situations which
involve the relative movement of contacting surfaces under the action of an
alternating load. This is known as fretting corrosion. Some materials are
more susceptible to it than others and hence there are preferred
combinations of materials in situations where it is likely to arise. The use
of antifretting compounds, the reduction of surface stresses and surface
hardening have also been found to alleviate the problem.

Surface finish and Fatigue failures in metals almost always initiate at a free surface so that the
treatments surface condition has a significant effect on fatigue endurance. Immediate
improvement can be effected by polishing a machined surface since this
reduces the mild stress concentration effect of a lathe-turned, milled or
ground surface.
 FATIGUE 563

Manufacturing processes also may introduce residual stresses and strain

hardening. It is important therefore to appreciate the effect which these
operations may have on fatigue endurance.
Surface coating of ferrous metals is often done to improve their wear and
corrosion characteristics. In general it is found that such coating does not
improve the fatigue strength of the substrate and, depending on the plating
conditions, may cause a considerable reduction in fatigue life. Again the
process of anodizing aluminium to improve its wear and corrosion resistance
also has the effect of reducing its fatigue strength.
The introduction of residual compressive stresses on the surface of a
metal has been shown to be a very successful method of improving fatigue
endurance. Conversely, the presence of residual tensile stresses on the
surface has a very detrimental effect. There are a number of metallurgical
methods of introducing compressive residual stresses through hardening of
the surface layers of the material. The three main methods are induction (or
flame) hardening, carburizing and nitriding. It is interesting that the greatest
improvement in fatigue strength is observed when stress concentrations are
present.
There are also a number of methods of physically introducing
compressive residual stresses at the surface. These include shot peening
and skin rolling, both of which have the additional advantage that they
remove any stress concentration marks left by machining operations.

20.7 Cumulative damage

Although most fatigue tests and some components are subjected to a
constant amplitude of cyclic stress during the life to fracture, there are many
instances of machine parts and structures which receive a load spectrum, 1.e.
the load and cyclic stress vary in some way under working conditions. To
establish any difference between fatigue under varying and constant
amplitude conditions, tests are conducted in which a certain number of

Fig. 20.22 Diagrammatic repre-

sentation of cumulative damage

Stress
Nz cycles
n,, cycles

io)Be

Failure curve
S,
amplitude
Stress !
!
T |
Noy No damage
| i
! !
1

Ny, No
Cycles during damage period N
564 MECHANICS OF ENGINEERING MATERIALS
a

cycles are done at one stress level followed by a number at a higher or lower
stress, and this sequence is repeated till failure occurs. It is suggested that
damage by fatigue action accumulates and that a certain total damage line is
represented by the S—N curve. One way of representing this algebraically
was proposed by Miner. If m cycles are conducted at a stress level Sj, at
which the fracture endurance would be Nj, and if this is followed by m2
cycles out of N2 at a second stress level S2, as in Fig. 20.22 and so on, then
for h such blocks
n\ n2 nN3 Nh
ipa ep pe wane Ts
or

h
>t =I [20.13]

Test results often show the sum of the cycle ratios (”/N) differing widely
from the value of unity, generally covering a range from about 0.6 to 2.0
with, in a few cases, extreme values well outside this range.
The value of unity in eqn. [20.13] tends to be an overall average, but
when using this approach in design it is prudent to use a lower value (such
as 0.5) in order to allow for uncertainties in the hypothesis.
Other models for cumulative damage have been proposed in order to
obtain more conservative estimates of the total fatigue life for multi-level
sinusoidal stress histories. However, any slight improvements in accuracy
are outweighed by the increased complexity of the analysis. The problem is
that cumulative damage is dependent on stress history, mean stress, stress
concentration, etc., and no simple model can be developed to predict
accurately the fatigue life with such a wide range of variables. For example,
it is often found that for a two-stress level test, in which one stress is applied
for a number of cycles and then run to failure at a second stress, if S$) < $2,
then U(n/N) > 1, and for S; > S2, &(n/N) < 1. In addition, the variation
from unity is greater for larger differences between S$; and S2. A further
complication is that in many practical situations it is difficult to define
precisely the nature of the loading pattern. For example, in the design of a
vehicle axle it is unreasonable to expect the designer to anticipate all the
extremes of stress which different driving styles will cause in the vehicle.

20.8 Failure under multi-

axial cyclic stresses
In the discussion of fatigue so far, we have considered only the effects of
cyclic uniaxial stresses on materials. However, in practice many components
are subjected to biaxial or triaxial stress systems for which laboratory testing
would be complex and expensive. Consequently, in a similar manner to the
prediction of static yielding under complex stress by the use of a simple
tensile yield stress (Chapter 12), attempts have been made to establish a
criterion for fatigue failure due to complex cyclic stresses, in terms of the
uniaxial stress fatigue limit for the material.
Some success has been achieved in the use of the shear-strain energy
criterion to predict cyclical failure. Hence from eqn. [12.15] the yield stress
in simple tension is replaced by the fatigue limit, Sp, in, say, rotating
 BDATIGUE 565

bending for the material, so that

(j= 02)? Fon = 03) (oj; 0)) = 253

The applicability of this criterion may be related to the static yield
situation in that microyielding will occur at the crack tip as propagation
proceeds. Fatigue cracks generally initiate at a free surface so that a3 will be
zero in the above equation. If mean stresses are present then this must be
allowed for by using the above equation as a function of both alternating
stresses and mean stresses and using the appropriate values of S, and S,,,
from uniaxial data, on the right-hand side. This is illustrated in Example
20.5.
More recent research has shown that the above criterion has a rather
limited range of applicability and that two strain parameters are required
instead of the single equivalent uniaxial cyclic stress. The parameters
proposed are the largest strain circle created in a fatigue cycle and
the position of that strain circle in strain space. These are represented
as $(€) — €3) and }(€1 + €3) respectively from Mohr’s strain circles. The
intermediate strain €2 has been shown to control the direction in which the
crack, once initiated, will grow.

Example 20.5
A 25mm diameter geared shaft, Fig. 20.23, is subjected to a fully reversed bending
moment of +1094 kN m as shown. The gear is a shrink fit onto the shaft and sets up
radial and tangential stresses of 350 MN/m? on the surface of the shaft. If during
operation the gear wheel is subjected to a fluctuating torque of +900 .N m estimate
the fatigue life of the shaft. The shaft material has a yield stress of 925 MN/m2 and
its bending fatigue life may be predicted from
: 16.5
N, = 5.2 x 10° i
Se
where S, is the stress amplitude in MN/m2.

ceeyee kN/m

The bending stress is

My _ +1094 x 10° x 12.5 x 64

o.= ; = £713 MN/m*
I mx (25)
The shear stress is

ee +900 x 10° 3 x 12.5 x 32 — 4793 MN/m?

one nm x (25)'
566 MECHANICS OF ENGINEERING MATERIALS

Fig. 20.24

Tale)

Z 350

Figure 20.24 shows the stress systems and from Mohr’s circle the principal
stresses may be determined for the maximum and minimum limits of cyclic
stress:

Cyclic maximum Cyclic minimum

(MN/m?) (MN/m/?)

a, = 800 oa; = —8/0

i120 or = 190
03, = —390 a; = —350

Amplitudes (MN/m’) Mean (MN/m?)

Onl — 835 Olm = —35

04 = 115 02m = —305

Ox = 0 Oa hall,

Using the shear-strain energy criterion,

28h, = (Ota — 72a)” + (2a — O34)” + (730 — O10)”

= (835 21115)? (115 —.0)" + (0835)
Spa = 784MN/m*
Similarly, Sp, = 295 MN/m?*
Using the Soderberg rule,

784
Sp Ss S51 MN
Mme) eee
From the equation describing the fatigue life of shaft material
l 16.5
Np =O:2% 102 (a) = 1.61 x 10° cycles
 FATIGUE 567

20.9 Fatigue of plastics

Reinforced and unreinforced plastics are susceptible to brittle fatigue
and composites
failures in much the same way as metals, and the fracture mechanics
approach has been used to predict crack growth rates. In addition, however,
the high damping and low thermal conductivity of plastics means that under
cyclic stressing there is also the possibility of short-term thermal-softening
failures if precautions are not taken to dissipate the heat generated.
Therefore, although in metals the fatigue strength is relatively unaffected by
frequencies in the range 3-100 Hz, with thermoplastics the cyclic frequency
is important since it has a pronounced effect on the temperature rise in the
material.
This is illustrated in Fig. 20.25. At a relatively low cyclic frequency, /,,
the material will fail by thermal softening if the stress amplitude is greater
than a4. At cyclic stresses below a4, the material fails by a normal crack
initiation and propagation mechanism similar to that observed in metals.
Thus at a cyclic frequency of f,, the fatigue curve which would be seen is
ABCFG.
At a higher cyclic frequency, /:, a thermoplastic will experience thermal
softening at lower stress amplitudes — down to ag in Fig. 20.25. Below this
stress level, conventional fatigue failures are once again observed. Hence, at
this higher cyclic frequency, the fatigue behaviour which is observed is
described by DEFG.

Fig. 20.25 Fatigue failure beha-

viour of thermoplastics
Cycle frequencies
f, < fp

ON Brittle crack growth failures

Stress
ampltude

Og

Thermal softening failures

Log (Cycles to failure)

With non-reinforced moulded plastics, fatigue cracks usually initiate

within the bulk of the material because the moulding operation tends to
produce a protective skin which inhibits crack growth from the surface.
However, plastics are not immune to the effect of stress concentrations and
many fatigue cracks in moulded articles are initiated at holes, sharp corners,
etc.
Fibre-reinforced plastics (composites) can have a high resistance to
fatigue crack growth. However, the properties of many composites are
568 MECHANICS OF ENGINEERING MATERIALS

(a) (b) (c)

Fig. 20.26 Fatigue failure mechan- anisotropic, i.e. direction dependent, and fatigue crack resistance may be low
isms in composites in certain directions. There are many types of fibre-reinforced composites,
e.g. unidirectional fibres, bidirectional fibres, random short fibres,
thermoplastic matrix, thermosetting matrix, etc. It would be inappropriate
in this book to cover in detail the many types of fatigue behaviour which can
be observed in the various permutations of matrix and fibre type. However,
the following observations will give the reader a feel for the general type of
behaviour to be expected. Those requiring additional information should
refer to specialist texts such as that by Hertzberg’.
In uniaxial fibre composites the material may fail by a number of
mechanisms as illustrated in Fig. 20.26. In the left-hand diagram, the high,
local, cyclic stress has broken a fibre. This causes a high shear-stress
concentration at the fibre tip and this will lead to failure of the fibre—matrix
interface along the broken fibre (‘debonding’).
An alternative mechanism is shown in Fig. 20.26(). In this case, fatigue
cracks have developed in the matrix. Initially the growth of these cracks will
be inhibited by the fibres and in some cases no further propagation will
occur. However, at higher cyclic stress levels, a fibre will break and crack
growth can occur progressively across the section. Such crack growth can
also be along the fibre—matrix interface in neighbouring fibres as shown in
Fig. 20.26(c).
In bidirectional composites and general multi-ply laminates the rate of
progression of cracks can be reduced owing to the constraint provided by
the adjacent plies having fibre orientations along a direction which is
different to that in which the crack is propagating. Thus the strategic
arrangement of fibres can produce materials with a high resistance to fatigue
crack growth.

20.10 Summary
It has been shown that the possibility of premature failure of a material by a
fatigue mechanism is an extremely important design consideration in
situations where fluctuating stresses are either applied directly or
transmitted to the material. The majority of all failures which occur in
practice can be attributed to fatigue and the greatest percentage of these are
caused by bad design, usually ill-considered positioning and/or geometry of
stress raisers such as holes, abrupt changes in section, keyways, etc. Other
factors which affect fatigue endurance are the level of the mean stress,
surface condition and nature of the environment.
Although the crack growth mechanism during fatigue is not completely
understood it is known that when a component is subjected to a variety of
 La IER 8, 569

stress levels the damage incurred is cumulative. It has also been found that
fracture mechanics is a useful tool in predicting fatigue life provided crack
growth data for the material is available.

References
1. Ashby, M. F. and Jones, D. R. H. (1980) Engineering Materials,
Pergamon Press, Oxford.
2. Hellan, K. (1984) Introduction to Fracture Mechanics, McGraw-Hill,
London.
3. Smith, R. N. L. (1991) BASIC Fracture Mechanics, Butterworth-
Heinemann, Oxford.
4. Hertzberg, R. W. (1989) Deformation and Fracture Mechanics of
Engineering Materials, 3rd edition, John Wiley, New York.

Bibliography
Duggan, T. V. and Byrne, J. (1977) Fatigue as a Design Criterion, Macmillan,
London.
Forrest, P. G. (1962) Fatigue of Metals, Pergamon Press, Oxford.
Klesnil, M. and Likas, P. (1989) Fatigue of Metallic Materials, Elsevier,
Amsterdam.
Madayag, A. F. (1969) Metal Fatigue: Theory and Design, John Wiley, New
York.
Miller, K. J. (1991) Metal Fatigue — Past, Current and Future, 27th John
Player Lecture. Preprint 3, Institution of Mechanical Engineers,
London.
Osgood, C. G. (1982) Fatigue Design, Pergamon Press, Oxford.
Pook, L. P. (1983) The Role of Crack Growth in Metal Fatigue, The Metal
Society, London.

Problems 20.1 <A switching device consists of a rectangular cross-section metal

cantilever 200mm in length and 30mm in width. The required
operating displacement at the free end is +2.7mm and the service
life is to be 100000 cycles. To allow for scatter in life performance a
factor of 5 is employed on endurance. Using the fatigue curves given
in Fig. 20.7 determine the required thickness of the cantilever if
made in (a) mild steel, (6) aluminium alloy. Esjee) = 208 GN/m’,
EEilumiatum =79 GN/m’.
20.2 (a) A fatigue fracture produced by cyclic uniaxial tension stresses
exhibits circular striations which have their centre at the point of
crack initiation (usually on the surface). Explain why the striations
have this shape. (d) If a shaft is subjected to cyclic torsional stresses,
on what planes would you expect the fatigue cracks to grow?
20.3. A pressure vessel support bracket is to be designed so that it can
withstand a tensile loading cycle of 0-500 MN/ m’ once every day
for 25 years. Which of the following steels would have the greater
tolerance to intrinsic defects in this application: (i) a maraging steel
(Ki. = 82 MN/m®, C = 0.15 x 107", m = 4.1), or (ii) a medium-
strength steel (K;, = 50 MN/m?, C= 0.24 x 107", m= 3.3)
For the loading situation a geometry factor of 1.12 may be assumed.
20.4 A series of crack growth tests on a moulding grade of polymethyl
methacrylate gave the following results:
570 MECHANICS OF ENGINEERING MATERIALS

da/dN (m/cycle) 225 10 4x 10-7 62 x10= Liat’ ie 29 x 1077

AK (MN m~/*) 0.42 0.53 0.63 0.79 0.94 Un 1N7

If the material has a critical stress intensity factor of 1.8 MN/ m>/? and
it is known that the moulding process produces defects 40 jum long,
estimate the maximum repeated tensile stress which could be applied
to this material for at least 10° cycles without causing fatigue failure.
20.5 As part of the mechanism of a machine a cam is used to cause a metal
beam to oscillate, as shown in Fig. 20.27. If the design of the cam is
such that the beam deflection varies between a maximum of 3mm
and a minimum of | mm relative to its undeflected position, calculate
a suitable beam depth to avoid fatigue failure in the beam material,
using the Gerber, modified Goodman and Soderberg methods to
allow for the effect of mean stress. A fatigue strength reduction
factor of 1.8 should be assumed. The tensile and yield strengths of
the beam material are 350 MN/m? and 200 MN/m* respectively,
and its fatigue strength in fully reversed cycling is 100 MN/m’.
Young’s modulus for the steel is 207 GN/m?.

20 mm
Fig. 20.27 ee |
Deflection varies
ee ah between 1 mm and 3 mm
Steel beam |

||| Push rod

Cam

20.6 <A shaft of circular cross-section is subjected to a steady bending

moment of 1500 Nm and simultaneously to an alternating bending
moment of 1000 Nm in the same place (so that the total moment
fluctuates between 2500 Nm and 500 Nm). Calculate the necessary
diameter of the shaft if the factor of safety is to be 2.5. The yield
stress of the material is 210 MN/m/ and the fatigue limit in reversed
bending is 170 MN/m‘?. Calculate also the diameter of the shaft if
stress concentrations are to be allowed for with a fatigue strength
reduction factor of 2. Assume that the Soderberg rule applies.
20.7. A connecting-rod of circular cross-section with a diameter of 50 mm
is subjected to an eccentric longitudinal load at a distance of 10 mm
from the centre of the cross-section. The load varies from a value of
—F/2 to F. The material used has a tensile strength of 420 MN/m2
and a fatigue limit in fully reversed loading of 175 MN/m‘’.
Determine the limiting value of F to avoid fatigue failure if the
modified Goodman relation applies.
 BARIG WE 571

Fig. 20.28

50 mm

0.25 + 0.25 kN

20.8 Part of the structure of an aircraft is shown in Fig. 20.28 The central
hole in the inverted U-section supports a vertical force of 0.5kN.
During flight the upright section is subjected to a cyclical force
which varies from 0 to 0.5 kN. If all the parts are made from an
aluminium alloy with a yield strength of 392 MN/m*? and fatigue
strength of 270 MN/m/’, estimate where you expect fatigue failure to
occur. Use the Soderberg rule to allow for the effect of mean
stresses. The stress concentration factor at the curved portion of the
U-channel may be taken as 1.85 and K, at the central hole may be
obtained from Chapter 12.
20.9 A series of tensile fatigue tests on stainless steel strips containing a
central through hole gave the following values for the fatigue
endurance of the steel. If the steel strips were 100mm _ wide,
comment on the notch sensitivity of the steel.

Hole diameter (mm) No hole 5 10 20 Ms

Fatigue endurance (MN/m?) 600 25052710320 370

20.10 The fatigue endurances from the S—N curve for a certain steel are:

Stress (MN/m?) Fatigue endurance (cycles)

350 2.000 000

380 500.000
410 125 000

If a component manufactured from this steel is subjected to 600 000

cycles at 350 MN/m? and 150000 cycles at 380 MN/m’, how many
cycles can the material be expected to withstand at 410 MN/m?
before fatigue failure occurs, assuming that Miner’s cumulative
damage theory applies?
20.11 The analysis of the cyclic stresses on part of the landing gear of an
aircraft shows that during each flight it is subjected to the following
stress history:
100.000 cycles at +50 MN/m?
10000 cycles at +100 MN/m?
572 MECHANICS OF ENGINEERING MATERIALS

Fig. 20.29 Fillet radius = 2.25 mm

es Pulley Pulley
sdelale sce uN) oP Tooter R= 400 mm

1000 N} 200N 200 N |1000 N

<0.5 m>|———— 1.5m <0.5m

500 cycles at +150 MN/m?

200 cycles at +180 MN/m?
100 cycles at +185 MN/m?
50 cycles at +200 MN/m?
10 cycles at +210 MN/m?
If the S—N curve, for the part material is given by S = 500 N°,
where S is in MN/m/’, estimate how many flights this component
can withstand before fatigue failure occurs.
20.12 During service a steel cylinder, 320mm diameter, is to be subjected
to an internal pressure which varies from 0 to p. If the wall thickness
of the cylinder is 8 mm estimate the maximum permissible value of p
to avoid fatigue failure in the cylinder. The tensile strength of the
steel is 440 MN/m/? and the fatigue endurance limit in fully reversed
cycling is 210 MN/m’. A fatigue strength reduction factor of 1.8
may be assumed and the modified Goodman relationship should be
used for the mean stress effect.
20.13 Part of a pulley power transmission system is shown in Fig. 20.29.
Assuming a notch sensitivity factor of 1, calculate the required
endurance limit in the shaft steel in order to avoid fatigue failure.
The stress concentration factors for the shaft should be obtained
from Chapter 12. The shaft bearings may be regarded as simple
supports and the pulleys each provide additional load of 300 N.
CHAPTER |
Creep and Viscoelasticity

For the majority of engineering designs the variation of the ambient temperature is
not great and the stiffness and strength of the metal may be regarded as a constant.
However, there are also engineering applications which occur at high temperature in
fields such as steam plant, gas turbines, nuclear and chemical processes, kinetic
heating of supersonic aircraft,! etc.
It was mentioned briefly in Chapter 18 that in general the effect of temperatures
of up to several hundred degrees Celsius on metals is to lower their yield and tensile
strengths by very considerable amounts compared with ambient conditions. Another
factor which did not arise during the development of stress-strain solutions of
engineering problems in earlier chapters was the effect of the length of time under
which a component or structure was subjected to stress. It was assumed that the
application of external loading would develop particular values of stress and strain
and these would remain constant until the applied loading was removed. This is
certainly so for the bulk of engineering alloys in the elastic range at room
temperature. However, with increasing temperature it is possible for a material to
have increasing strain with time even at constant applied load. The time dependence
of strain is termed creep. Another complementary time-dependent response is termed
stress relaxation which occurs when the strain or deformation of a component or
structure is kept constant for a time period during which a reducing applied load
(stress) is required to maintain the strain. These are most important properties in the
design of any high-temperature component.
Nowadays plastics, both reinforced and unreinforced, play a major role as
engineering materials. A brief reference was made in Chapter 3 to their stress—
strain-time behaviour and this is known as viscoelasticity, being the interactive
results of Newtonian viscosity and Hookean elasticity. Unreinforced thermosplastics
are particularly susceptible to creep and room temperature is ‘high’ enough for
plastics to exhibit the phenomenon, the actual rate of creep being a function of the
imposed stress.
The first part of this chapter concentrates on the problem of creep and stress
relaxation as a time-dependent high-temperature phenomenon affecting metals. The
second part of the chapter deals with the basic considerations of viscoelastic data
and creep design for plastics.

21.1 Stress—strain-
Creep manifests itself in metals at temperatures above about 0.37,,, where
time-temperature T is the absolute melting temperature, and around 0.57, creep strain
relationships becomes considerable. Thus Andrade’ commenced studies of creep
behaviour in 1910 using lead, since this metal exhibits creep at room
temperature. The majority of creep experiments are carried out under
uniaxial constant loading conditions for a particular chosen test temperature
which must be very accurately controlled. Measurements of extension are
made at frequent intervals of time until the specimen fractures or the
experiment is stopped after a'sufficiently lengthy period. Typical curves of
creep strain against time are plotted in Fig. 21.1 for various constant load
(nominal stress) levels at a constant specimen temperature.
574 MECHANICS OF ENGINEERING MATERIALS

O7)< O27 = O23 < 04

Fig. 211 ‘Typical creep curves for Temperature T
various stresses at constant tem-
perature

strain
Creep
€
01

\
Primary '

10} Time t

There are four principal aspects of each of the curves shown in Fig. 21.1
as follows:
(a) initial strain, which is elastic but may extend marginally into the plastic
range due to the first application of load;
(b) primary stage, a period of decreasing creep rate during which strain
hardening is occurring more rapidly than softening due to the high
temperature;
(c) secondary stage, in which the creep rate is virtually constant through
equilibrium between strain hardening and thermal softening;
(d) tertiary stage, an increasing rate of strain, due to microstructural
instability from prolonged high temperature and to gradual increase in
stress level and stress concentration at cracks in the grain boundaries,
which leads to complete fracture of the specimen.
It is important to note that elongation to fracture in creep, even for a
ductile metal, is only a fraction of that obtained for continuous loading to
fracture at high temperature. The very small gradient of the secondary stage
at low stresses suggests that there might be a limiting creep stress, below
which de/dt = 0, having similar significance as the fatigue limit under
cyclic stress. However, it has been shown that there is no reliable criterion of
this form for creep, and a ‘limiting creep stress’ is based on a permissible
creep strain after a given time.
The family of curves shown in Fig. 21.1 is based on stress as the
parameter and temperature constant; however, a similar family of curves
would be obtained for a particular constant stress with temperature as the
parameter. Thus a complete picture of the creep behaviour of a metal
necessitates the construction of several families of creep curves for various
stresses and temperatures.

21.2 +Empirical
Typical of the standards of creep strength required for metals are the
representations of creep
stresses to give minimum creep rates of 1% strain in 10000 hours, or 1%
behaviour strain in 100000 hours.
A 10000 hour test occupies approximately 1 year, and it is therefore
evident that, although a few creep tests may be conducted over periods of
this length or longer, it is a very slow and inconvenient process for obtaining
a range of data at different stresses and temperatures. As a result, methods
have been sought whereby long-life data can be extrapolated from short-
term tests.
 GREEPTAND) VISCOELASTICILY 575

Creep strain €,, which is a function of stress, 7, time, ¢, and temperature,

T, can be represented by three functions as follows:

€, =fi(o)-Al)-A(7) (21.1)
(a) Stress function: the most commonly used functions are

fila) =Aio" file) =A sinh(2)

filo)= Arexo(S) [21.2]

where 4), A, A3 are constants and oo and go are reference stresses.
(6) Time function: this is usually expressed as a polynomial, as first
suggested by Andrade’, one reasonably applicable form being

e, = at + Br+ 0 [21.3]
in which a, 3 and + are material constants, but are functions of stress
and temperature, relating to the primary, secondary and tertiary stages
respectively.
(c) Temperature function: the most generally used temperature function is

f(T)(ve=ex —_—
(—92)
AH
214
21.4

where AH is the activation energy, R is the universal gas constant and

T is the absolute temperature. This type of expression is fundamental to
all rate processes.
When we come to designing for creep in components, interest centres
principally on the secondary stage, where at low stresses the creep rate is
constant for very long times producing the major contribution to the total
creep strain to fracture. Consequently, in eqn. [21.3] the tertiary term is
neglected and the primary is replaced by a constant strain €y, being the
intercept of the extrapolated secondary stage back on to the strain axis
(Fig. 21.2). Thus
de
€,-=eo+ (=): [21.5]

Fig. 212 Simplified creep curve

Creep
strain

Time
576 MECHANICS OF ENGINEERING MATERIALS

Fig. 213 Curves of creep strain

against log time, extrapolated to a 0.5 Eee
Stresso

specified strain

strain
Creep

i, tb tg
Log time

where de/dt = €, is the secondary-stage creep rate. This minimum creep

rate has been experimentally related to stress by the empirical expression,

é, = Bo" [21.6]
where B and n are material constants. The dependence on temperature can
then be included by writing

AH
é, = Bo” exp(-a7) [21.7]

A log-log plot of €, against o yields a straight line. However, extrapolation of

the data, particularly at high stress, can be unreliable owing to the
dependence of m and AH on the particular stress/temperature regime.
Alternatively, combining eqns. [21.5] and [21.6], we can express the time to
reach a specified value of total creep strain in the secondary stage as follows:
Ec — €0
Bor
t= ——_ [21.8]
21.8

However, this approach can be uncertain in that tertiary creep may

commence before the predicted value of secondary creep is achieved.

Fig. 214 Curves of temperature

against log time, extrapolated to Creep strain = 0.5

specified time

Temperature

! !
1 i} i}
1 t i}
i}
i | i
[eae
eee
t th ts 100 000
Log time
 CREEP AND VISCOELASTICITY 577

A safer method of predicting the stress and temperature permissible for a

specified creep strain of say 0.5% in 100000 hours is as follows. Creep tests
are carried out at various stresses and temperatures to obtain a significant
part of the secondary-stage creep in each test. Graphs of creep strain against
log time for several temperatures at a particular stress are then extrapolated
to the limiting creep strain of say 0.5% as in Fig. 21.3. Next we plot graphs
of temperature against log time for various stresses at the strain limit of
0.5% as shown in Fig. 21.4. These curves are then extrapolated to a
required life of say 100000 hours. Although several extrapolations are
involved in this method, they tend to be rather more reliable than a simple
extrapolation of a creep—time curve.

21.3 Creep-rupture
As there is as yet no real substitute for a few long-term tests to ensure
testing
reliable creep knowledge, it is useful to have a quick sorting test to enable
the best material from a group to be selected for long term tests.
The creep-rupture test is widely used for the above purpose and also as a
guide to the rupture strength at very long endurances. The principle is to
apply various values of stress, in successive tests at constant temperature, of
a magnitude sufficient to cause rupture in times from a few minutes to
several hundred hours. Plotting log stress against log time as in Fig. 21.5
yields a family of straight lines with temperature as parameter. These tend to
extrapolate back to the respective hot tensile strengths. Extrapolation
forward to longer times is possible, but care has to be exercised in that
oxidation owing to the high temperature can cause a marked increase in the
slope, and hence reduction in stress for a required life.

414
Fig. 2.5 Log rupture/log time oe on
relation for chromium—molybde-
num-silicon steel (adapted from
ee
138 a Cee o—.
Lessels*; by courtesy of John
Wiley & Sons, Inc.) 69 sarge eee e
S 55.20 S ° ee
1.4 °
op)
8 276
Barto 8
ee Apia a
2) *—_, 875, Oxidation —>~ _
C effect nlc
13.8 S = short-time tensile strength

6.9 idee ropes ie Wey ! elec

Of 02504 O:6:20 10 100 1000
Log time to rupture t

One of the physical approaches to the problem of creep suggests that

viscous flow in fluids is analogous to secondary creep in metals and hence a
rate-process theory is applicable, relating creep rate and temperature, using
eqn. (21.4)

e =F = Aexp(-AH/RT) [21.9]

where A is a material constant.

578 MECHANICS OF ENGINEERING MATERIALS
eee _ —_e_een

Larsen and Miller* have analysed the above relationship and put it in the
form

=e = T(log,A+ log, t — log, €) [21.10]

or

(=) - T(a + log,t) [21.11]

for a given value of strain. Here a is a constant for a given strain and
(AH/R), is a function of the stress level o. The right-hand side of eqn.
[21.11] is known as the Larsen—Miller* parameter, 7(a + log, +), and
plotting log, o against the parameter often yields a family of straight lines for
different creep strains, known as master creep curves, which correlate well
over a wide range of times, temperatures and different metals. From the
above curves a general relationship may be written in the form

log,o = C; + C,T(a + log, t) [21.12]

where C) and C) are constants. Combining the strain-rate/stress equation

€ = Bo”
with that above for stress, temperature and time gives

é = Co” exp(—aT) [21.13]

where C, a and 7’ are material creep constants.

21.4 Tension creep test

The most common form of creep test is conducted in simple tension, and
equipment
since constant loading on the specimen is required over very long periods a
dead-weight loading system is usually employed. A typical arrangement is
where a specimen is housed in a furnace. In previous sections it has been
stated that temperature variation has a great influence on the minimum
creep rate; it is therefore essential to have uniform and constant temperature
along the length of the specimen. The relevant British Standard specifies a
maximum variation along the gauge length of 2°C and a variation of mean
temperature of not more than +1 °C up to 600°C and +2 °C from 600 °C to
1000 °C. Ifa resistance furnace is used, it is usual to have three separately
controlled resistance elements to compensate for non-uniform flow of heat
through the furnace. Thermocouples and temperature indicators are used
for measurement and control.

21.5 Creep during pure

In order to demonstrate the application of eqn. [21.6], the case of a beam
bending of a beam
subjected to pure bending at high temperature will be considered.
The assumptions that will be made in this problem are as follows:
(a) plane sections remain plane under creep deformation;
(b) longitudinal fibres experience only simple axial stress;
(c) creep behaviour is the same in tension as in compression.
CREEP AND VISCOELASTICITY 579

If the radius of curvature of the neutral axis is R, then the strain at a

distance y from the axis is

ee
oR
and for creep in the secondary stage,

de a
Bo"
a.
Therefore

dO/R)_
eee. Ba

or

and
1/n
f= =) [21.14]
If the bar is of width 4 and depth d, equilibrium of the external and internal
moments is given by
d/2
M — 2] aby dy
0
Substituting for 7, using eqn. [21.14],
d/2 1/n 2b d/2
= 2] () by dy = —| yitl”) dy
0 \RBt (RBt)/" Jo
Integration gives
va 2b 2+(1/n)

(RB)! 2n+1\2

Substituting for (RBt)'/” from eqn. [21.14] gives

FifeoeCane Ati en lo \ik
“em
— yl/n Wn t+1\2
Inserting the second moment of area J = b bd? and rearranging, we have the
bending stress at distance y from the neutral axis as

= My (2n+1) (2y\°""! 21.1

Piel ae (3 ee)
It has been put in this form to show the difference from the simple linear-
elastic distribution of stress, My/TI.
The stress distribution across the section is shown in Fig. 21.6 for n = 1,
which corresponds to no creep and simple elastic conditions, and n = 10,
580 MECHANICS OF ENGINEERING MATERIALS

Fig. 2L6 Distribution of bending

stress for the cases of zero creep (6)
2 1.0
and secondary-stage creep

where secondary-stage creep is occurring. It is seen that the effect of creep is

to relax the outer fibres’ stress, but to increase the core bending stress. This
is because minimum creep rate must be proportional to the distance from
the neutral axis in order to satisfy eqn. [21.14], and this automatically adjusts
the stresses to the distribution shown.

21.6 Creep under multi-

Under the above heading we find typical engineering components such as
axial stresses
thin-walled tubes subjected to internal pressure and combinations with
bending or torsion, thick-walled cylinders under internal pressure, rotating
discs and plates subjected to bending. The number of experimental
investigations has been limited owing to the intricacies of high-temperature
testing and complex loading. However, there have been sufficiently
comprehensive studies to verify the analytical approach that is generally
adopted.
In order to relate uniaxial creep data to a biaxial problem, some of the
laws of plasticity are invoked, namely that (a) the principal strains and
stresses are coincident in direction, (4) plastic deformation occurs at
constant volume, (c) the maximum shear stresses and shear strains are
proportional.
For constant volume the sum of the three principal strains is zero;
therefore,

€j té2 +63 =0 [21.16]

Expressing maximum shear stress and strain in terms of the difference in

principal stress and strains,
SI = S2 JS) e376]
= = = 3 [21.17]
(Oh) KO 02 = 03 (OR)
= 10}!
 CREEP AND! VISCOELASTICITY 581

Rearranging the above equations to give the individual principal strains in

terms of the principal stresses,

E] =F ho —}(o2 +03)

a2 =F fo 10s + 01) [21.18]

es =F os Hor + 02)
These may be expressed as a constant creep rate by writing

Ey = alo —4}(o ar 03)

é2 = alan —$ (03 + 01) [21.19]

é3 = alo3 — 5(01 + 0)
where qa is a function relating the three principal stresses to the simple
uniaxial stress creep condition.
Using the von Mises yield criterion to obtain the equivalent uniaxial
stress 0,, gives
1
Op =F Ay\(Gg 2G); A Go103) 4 (03 101) [21.20]
J2

From the simple secondary-stage creep law,

é = Baz

and for simple tension, 72 and 03 = 0) and o, = 02; therefore

E= QC,

and hence

Ca tees [21.21]
Therefore the three principal creep rates may be written as

En Bo™ "lo, = 5(a2 = 03)|

é, = Bo" |o, —} (03 +01) [21.22]
é3 = Bo" [03 —} (01 + 2)|

Example 21.1
An Ni-—Cr—Mo alloy steel tube of 100mm, diameter and 3mm wall thickness is to
operate at 400°C with internal pressure for a service life of 100000 hours.
Determine the allowable pressure for a creep strain limit of 0.5%. The constants in
the minimum creep equation at 400°C are n = 3 and B = 1.45 x 10°23 per hour
per MN/m?.

In the thin tube under internal pressure where oj is the hoop stress and a
the axial stress, 0} = 202 and 03 = 0. Hence from eqns. [21.20] and [21.22]

VA)
Cg = oO}
Z
582 MECHANICS OF ENGINEERING MATERIALS
SS
eeeSSS SEE EEE

and
3 n+1

E| = (5) Bo}

é2 =0
3 n+]
3 = -(¥) Bo}

where o; = pr/t (eqn. [2.10]).

It is interesting to observe that there is no creep in the axial direction,
which has also been verified experimentally.
The allowable internal pressure is controlled by the circumferential or
hoop strain rate €,; therefore

E]
Nayn+l
(5
cate
pe yerh

where ¢;, = time in hours.

Substituting the design values,

0.005
f = |(¥
—
ee 1.45
; x 110Ey Vvomeee
: |) fi(G,

from which

p' = 794.4 x 16°

p = 530N/m?
If the problem had combined the internal pressure with axial tension or
torsion, the solution would only entail the use of a different ratio of a; to a2
to give the required expression for o, above.

21.7 Stress relaxation

The chapter has dealt so far with creep in the form of time-dependent
increase in strain at constant stress; an alternative manifestation of creep is a
time-dependent decrease in stress at constant strain. A common example of
this phenomenon is the relaxation of tightening stress in the bolts of flanged
joints in steam and other hot piping, with the resulting possibility of leakage.
Another important case is the component subjected to a cycle of thermal
strain. The effect is illustrated by means of the stress-strain curve in Fig.
21.7, where thermal expansion has set up compressive stress followed by
relaxation with time. On cooling down, a higher residual tensile stress is set
up than if there had been no relaxation of compressive stress. This situation
can lead to failure eventually in a metal such as a flake cast iron which is
weak in tension. Thermal strain concentration around nozzle openings in
pressure vessels is another example where reversal of stress after relaxation
could in time lead to a thermal fatigue failure.
A curve of stress relaxation against time is similar to a mirror image of a
curve of creep strain against time as shown in Fig. 21.8. Just as much care
has to be taken with relaxation testing as with conventional creep testing
 CREEP AND VISCOELASTICITY 583

Fig. 2.7 Hysteresis including

stress relaxation

since results are very sensitive to temperature variation. The procedure

usually adopted is to load the specimen to an initial stress which will give a
specified strain of, say, 0.15%. The stress is then adjusted with time so that
the specified strain is maintained.
In the Barr and Bardgett types of test the decrease in stress is noted after
48 hours for various initial stresses at constant temperature. If the decrease
in stress is plotted against initial stress, the intercept of the curve on the
latter axis gives the initial stress required for ‘zero’ decrease in stress, i.e. no
relaxation. However, this is a very short-term test and can only safely be
used for sorting materials.

Fig. 218 Relaxation of stress with

time at constant strain and Constant strain
Constant temperature
temperature

Stress

Time

21.8 Stress relaxation in

Bolted flanged joints at high temperature represent an example of the
a bolt
problem of stress relaxation. Consider a bolt which is initially tightened to a
stress 09, producing an elastic strain €9, and to simplify the problem it is
assumed that the flange is not deformed by the bolt stress. After a period of
time 7 the effect of creep is to induce plastic deformation or creep strain,
which allows a relaxation of stress o and elastic strain. Now, the total strain
must remain the same if the flange is rigid; therefore,
or
&) = E+ E

or differentiating with respect to time,

de, 1 do
584 MECHANICS OF ENGINEERING MATERIALS

or
1 do
= VARI,
a E dt |
Substituting for €, in terms of stress,
Bo" = 1 do
= Sd
Therefore
1 do
dt = —-—
‘EB —ot 21.24
ae
The time for relaxation of stress, from oo initially to o, at time ¢, is then
obtained by integrating eqn. [21.24], and

Therefore

pen ape er ane [21.25]

ER aah) \ote an
It is found in practice that relaxation of stress is more rapid than that
given above owing to the effects of primary creep, and allowances for this
can only be made by using a more complex creep-rate—stress—time function.

Example 21.2
The bolts holding a flanged joint in steam piping are tightened to an initial stress of
400 MN/m2. Determine the relaxed stress after 10000 hours. E = 200GN/m?, n
= 3 and B= 4.8 x 10-*4 per hour per N/m.

From eqn. [21.25]

1
10000 =
200 x 109 x 4.8 x 10-34 x 2

1 1
* A
(a, x 10°)” (400 x 10°) 2

1 1
192107! =
oe 400
and

o, = 349.8 MN/m?

21.9 Creep during

Experiment and analysis has concentrated principally on constant load and
variable load or
constant temperature conditions during creep. There are some engineering
temperature applications in which the loading conditions change from time to time at
high temperature as illustrated in Fig. 21.9. These are not necessarily
 CREEPSAND VIS GORLAST lly. 585

Fig. 21.9

Stress

Time t

cyclical stress in the fatigue sense, as discussed in the next section, but result
in different creep rates during each different load sequence.
Several hypotheses have been proposed for predicting creep strain
related to changes of load which are constant before the change and constant
after. It is beyond the scope of this chapter to develop these hypotheses, but
two which are commonly quoted are time-hardening and _strain-hardening
theories. These take the following analytical forms:

Time hardening: €, =/(o, ¢) [21.26]

Strain hardening: €, = g(a, €) [21.27]

The former implies that creep rate is a function only of the stress and the
current time. The creep curve after the change of stress from 0; to 02 has
the same shape as the constant stress curve from the time of change, i.e. the
curve, 02, is moved vertically as shown in Fig. 21.10(a).
The latter implies that strain rate depends only on the stress and the
current plastic strain. Again, as above, the same shape of curve is assumed,
but now the appropriate portion of the curve, 02, from the time of change is
moved horizontally as shown in Fig. 21.10(0).

Fig. 21.10
L Creep at o5 Bin

Creep
strain
€, Creep
strain
e,

Creep at oy

HimMenG

(a) Time hardening (b) Strain hardening

586 MECHANICS OF ENGINEERING MATERIALS

These hypotheses can also be applied in a variable strain—stress—

relaxation context.
The predictions of total creep strain after load (stress) change are rather
better in the case of the strain-hardening approach compared with the time-
hardening method both for step-up load change and step-down load change.
Change of temperature in a component or structure, unless extremely
slow, will always induce non-uniform thermal strain gradients and hence
thermal stresses in addition to the stress due to applied load. It is evident,
therefore, that prediction of creep rates or total strain is extremely difficult
under variable temperature conditions.

21.10 Creep-—fatigue
Engineering developments such as gas and steam turbines, rockets and
interaction
supersonic aircraft have involved the use of metals not only at very high
temperatures but also with dynamic fluctuating stresses. In short, the
problem is one in which the mean or steady component of stress can induce
creep, and the alternating component of stress may lead to fatigue failure.
The earliest investigations into this problem were made between 1936 and
1940 by various German investigators, and since then there have been many
interesting studies both in this country and the U.S.A.
The problem of fatigue at high temperature was discussed in Chapter 20,
and this phenomenon can be unaccompanied by creep for fully reversed or
zero mean stress. Therefore, when considering a material for high-
temperature service it is usual to think of the behaviour in terms of a
diagram such as Fig. 21.11, in which fatigue failure is the criterion within
certain stress and temperature limits, and beyond these creep is the
predominant factor. If it is a question not of one or the other phenomenon
acting on its own, but of both influences operating simultaneously, then the
solution becomes rather more involved. Fatigue is essentially a cycle-
dependent mechanism, whereas creep is time dependent. It is therefore both
desirable and convenient to express fatigue behaviour at high temperature
also in terms of time to rupture as suggested by Tapsell°. One of the reasons
for this is because of the greater dependence of fatigue on cyclic frequency at
high temperature.
The most useful way of presenting data for combined creep and fatigue
Strength conditions is in the form of a diagram of alternating stress against steady or
mean stress, which is similar in most respects to the $,—S,, diagram in
normal fatigue (Fig. 20.18). Test results are plotted as the combination of
alternating and mean stress to produce either rupture or a specified creep
strain after a particular number of hours at constant temperature. Points
Temperature
along the abscissa represent creep conditions only and points along the
ordinate are for fatigue only. Some results obtained by Tapsell> on 0.26%
Fig. 2.11 Strength limitations with carbon steel are given in Fig. 21.12 for various amounts of total creep strain
increasing temperature occurring in 100 hours at 400°C under different combinations of cyclic and
steady stress. Theoretically predicted curves are also shown for creep strains
of 0.002 and 0.005.
The influence of alternating stress on the minimum creep rate and time
to rupture varies considerably with temperature, material and length of time.
At higher temperatures or long life, the alternating stress appears to have
little effect on creep rate; in fact, there are cases where creep strengthening
has resulted. On the other hand, at lower temperatures or shorter rupture
times, fatigue appears to play a more detrimental part, giving a higher creep
 CREP AND avis GOEL ANS DiGi mY 587

Fig. 2112 otal creep of 0.26% +300

7 Fatigue fracture
carbon steel occurring in 100 ate a i!

hours at 400°C — Failure curve

+250 2

=e
a Frequency of alternating stress
33 Hz
+200) jae
~
@

+150 fe LC XS
(theoretically n = 4.8) i

0.0026 ® : N:
+100 0.0055 s
®\ 0.0102 .
(MN/m?)
Alternating
stress 0.0034 .
+50 0.002| 0.0022
creep e ce
50.0083 Creep fracture .
(theoretically n = 4.8) 0.0019 ‘
0.00068 \0,0057 dome ot \
UN 0.029 :
0) 50 100 150 200 250 300 #4350
Mean stress (MN/m?)

rate. The rupture strain is also somewhat reduced by the presence of cyclic
stress.

21.11 Viscoelasticity
Because of the increasing use of plastics, both reinforced and unreinforced,
in engineering load-bearing applications, it is important that the response of
these materials to stress and environmental conditions should be
appreciated. A brief mention was made of viscoelastic stress—strain—time
behaviour in Chapter 3 and the discussion will be extended somewhat
further, particularly in relation to creep, in the remainder of this chapter.
In a viscoelastic material the stress is a function of strain and time and so
may be described by an equation of the form

Or= fe; P) [21.28]

This response is known as non-linear viscoelasticity, but as it is not
amenable to simple analysis it is frequently approximated by the following
form:
o=e.f(t) [21.29]
This response is the basis of /mear viscoelasticity and simply indicates
that, in a tensile test for example, for a fixed value of elapsed time the stress
will be directly proportional to the strain. These different stress—strain—time
responses are shown schematically in Fig. 21.13.
Viscoelastic materials invariably exhibit a time-dependent strain response
to a constant strain, which is relaxation. In addition, when the applied stress
is removed the materials have the ability to recover slowly over a period of
time. These effects occur at ambient temperature and, therefore, are a
principal design consideration as compared with metals for which creep and
relaxation only occur in higher-temperature environments.
In Chapter 3 it was explained that tensile test characteristics for plastics
are extremely sensitive to rate of straining. They are equally sensitive to
tensile test temperature and, in some materials, the humidity condition. As a
588 MECHANICS OF ENGINEERING MATERIALS
ee ee

Fig. 2113 Stress—strain behaviour

of elastic and viscoelastic materi- i Elastic — Linear
viscoelastic
als at two values of elapsed time ¢
Non-linear
viscoelastic

Stress

0 Strain

result of these special effects in plastics it is not reasonable to quote

properties such as modulus, yield strength, etc, as a single value without
qualifying these with details of the test condition.

21.12 Creep behaviour

Plastics exhibit a similar shape of creep curve of creep strain against time for
of plastics
constant stress and temperature as for metals (Fig. 21.1). However, one
distinct difference is the ability of plastics to ‘recover’ strain after the
removal of the applied load, and this effect is shown in Fig. 21.14. Plastics
also have ‘memory’ and the current behaviour in creep or relaxation is
dependent on all the past history of stress, strain and time on that sample
(assuming constant temperature).

Fig. 2114. Typical creep and re-

covery behaviour

(%)
Strain
|
| Reo,
‘O.
very
Stress applied Stress removed
Vee
0) Time

Attempts have been made to simulate polymer structure and its creep
and recovery responses by mechanical-type modelling using two principal
elements. These are a coil spring, which represents Hookean behaviour
(linear load deformation), and a dashpot (a piston in an oil-filled container),
which represents viscous Newtonian response (linear load—deformation
rate). These elements may be coupled in two ways: (i) with the spring in
series with the piston of the dashpot, and this is termed a Maxwell model;
and (11) with the spring in a ‘parallel’ location to the dashpot, so that applied
load 1s ‘shared’ between the two elements, and this arrangement is known as
the Kelvin—Voigt model. Because these models are individually quite
inadequate to represent even linear viscoelasticity, more complex assemblies
 CREEP AND VISCOELASTICITY 589

of the above units have been studied and, although somewhat more
representative, they still do not give an adequate prediction of creep
response which could be applied in design. This is principally because of the
non-linear viscoelastic nature of polymers. The alternative to the empirical
methods above is the use of experimental data obtained on the particular
plastics for which design exercises have to be carried out.
Creep data is initially presented in the form of graphs of creep strain
against log time, since linear time is inconvenient to encompass both short-
and long-term tests. A family of creep curves is illustrated in Fig. 21.15(a)
and two commonly used derivative graphs are shown in Fig. 21.15(b) and
(c). The former is constructed by taking a constant strain section through
the curves in Fig. 21.15(a) to give what is termed an isometric curve. A
constant time section through the creep curves gives a stress-strain diagram
as shown in Fig. 21.15(c) which is known as an isochronous curve.

Creep curves o
Fig. 2115 Isometric and isochro- =c so Isometric curve
nous curves from deep curves r=]
g () Ms
o

aay, ae ee Strain = e'

? ee eae LE eee
O05

(b) Log time

wn
wn

=2
a

Time =t'

lsochronous curve

(c) Strain

The isometric curve is often used as a good approximation of stress

relaxation behaviour since this specific experimental method is less common
than creep testing.
Isochronous curves can be developed independently without having to
obtain a family of creep curves as above. The method involves a series of
mini-creep and recovery tests in tension on a material. A stress is applied to
a sample and the strain recorded after a time f¢ (typically 100s), the stress is
then removed and the specimen allowed to recover for a period of four times
the loading time, i.e. 4¢ or 400s. A larger stress is then applied to the same
specimen and, after recording the strain at time ¢, this stress is removed and
the material allowed to recover. This procedure is repeated until sufficient
points have been obtained for the isochronous curve to be plotted.
Obtaining isochronous curves by this direct experimental method is less
time consuming and more economical than through creep experiments. In
fact, one can, of course, derive creep curves from several isochronous
experiments for different time intervals, e.g. 107, 10°, 10*s. Isochronous
curves plotted on linear scales may not show up the slight non-linearity at
low strains so an alternative plot on log—log scales is used so that any non-
590 MECHANICS OF ENGINEERING MATERIALS

Fig. 2116 Typical variation of

modulus with time

2)
aa
oS
ne}

= Strain = €4
Strain = €9
Strain = €3
£4 < €> < £3

pire colfoesara eel De UT

Log time

linearity is demonstrated by the slope of the line being less than unity, Le.
less than 45° on the log—log paper.
Another method of representing long-term creep behaviour is by means
of curves of modulus against time. These are shown in Fig. 21.16 for three
values of constant creep strain. They were derived by taking a constant
strain section through a family of creep curves and dividing the stress values
by the strain to give relaxation moduli which are plotted against the
respective time-value intersections.
The effect of temperature on the creep of plastics principally relates to a
Stress typical range of atmospheric temperatures such as —30°C to +40°C.
However, at the upper end of this range of temperature, marked acceleration
of creep rates will occur compared with the mid-range value and a family of
isochronous curves for one elapsed time and several temperatures will be of
the form shown in Fig. 21.17.
The prediction of creep response to step changes in stress is perhaps
10) Strain
even more complex than for metals. It has been tackled by a variety of
methods of superposition of parts of individual creep and recovery curves.
Fig. 2117 Isochronous curves for The principle that is most frequently quoted is the Boltzmann-type
different temperatures at one superposition but, as with most methods, this only relates to linear
elapsed time viscoelasticity. Solutions for non-linear viscoelastic superposition are too
intractable to be of practical use.

21.13 Designing for

The design of metallic structures and components is generally not
creep in plastics
temperature or time dependent and is based on linear-elastic reversible
stress-strain behaviour and small deformations. However, any load-bearing
component to be made out of plastic has first and foremost to be designed
for time-dependent deformation and, secondly, for time-dependent fracture.
Although the basic tenets of equilibrium of forces and compatibility of
deformations apply in the design of a plastic component, the problem arises
in relation to a suitable stress—strain—time law to link the foregoing. The
more accurate methods that have been proposed have the drawback of being
extremely complex and unattractive to the average designer. Perhaps the
most acceptable approach devised has been called the psewdo-elastic design
method. This involves the use of time-dependent ‘elastic constant’, moduli
and contraction (Poisson’s) ratio substituted into classical equations in place
of the true elastic constants. The time-dependent value of modulus must be
carefully determined to allow for the service life and limiting strain for the
 GREER FAND VIS GCORLAS
DIGI DY 591

plastic component. The limiting strain value for the particular plastic should
generally be decided in consultation with the material manufacturers and,
typically, might be of the order of 1—-2% strain. From this point the use of
published experimental creep data for the material is quite straightforward
in developing the component design. The following examples will illustrate
the technique and the relevant creep data is given in Fig. 21.18.

Fig. 2118 Creep curves for acetal

at 20°C

(%)
Strain

(0)
MO) SOHO? Alor sO ao” aioy’ slo
Time (Ss)

Example 21.3
A solid circular acetal rod, 0.15 m in length, is clamped horizontally at one end and
the free end is subjected to a vertical load of 25N. Determine a suitable diameter
for the rod for a limiting strain of 2% in 1 year. What would be the maximum
deflection at this time?

Using the creep curves in Fig. 21.18, a 1 year isochronous curve is plotted as
shown in Fig. 21.19 from which an allowable stress of 17.1MN/m? is
obtained at the 2% strain limit.

Fig. 2119 A 1 year isochronous 30

® Points taken from Fig. 21.18
curve for acetal at 20°C

‘
Be L eo?
. 20 xy

> ,,, See

S al(fal
mie
|
|
£ |
© l
1
|
1
9.2 |Zeeain ay i}
|
|
|
|

pa
0) Os; EO alts; BXO) “Psy SHO
Strain (%)
592 MECHANICS OF ENGINEERING MATERIALS

The maximum bending moment is 25 x 0.15 = 3.75 Nm. Using the

bending-stress relationship,
_ My _ 32M
ined Pas)
3 _ 32% 3.75 x10"
= er de?
d = 13.07mm
The maximum deflection at the free end is given by

“ WL
= oT
The appropriate value of modulus may be obtained from the isochronous
curve at 2% strain; hence the secant modulus
(ge
= —— = 855 MN/m”
() =F00 _
Therefore,

7 25 x 0.153 x 64x 103

= 23 mm
~ 3 «x 855 x 10° x 7 x 0.013 074
An alternative way of obtaining the design stress would have been to plot
a 2% isometric curve and read off the stress at 1 year.

Example 21.4
A circular acetal diaphragm is 2mm thick and is clamped around its periphery
giving a clear diameter of 100 mm. It is to be subjected to uniform pressure for a
service life of 1 year with a material creep strain limitation of 1% and a maximum
central deflection of 3mm. Determine the allowable working pressure.

The central deflection of a clamped-edge circular plate subjected to uniform

pressure was given in eqn. [16.27] as

_ 1211 —)pa*
— «O4ER
In this problem we shall have two time-dependent functions to consider,
the modulus E(t) and the creep or lateral contraction ratio, v(t). The former
may be determined from the isochronous curve of Fig. 21.19 for a strain
limit of 1%. The secant modulus is given as 920 MN/m*. The data available
for creep contraction ratio (the time-dependent equivalent of Poisson’s ratio)
is rather limited and generally lies between 0.3 and 0.4 but can rise to near
0.5 for ‘rubbery’ materials. For this problem we shall take a value of 0.35.
Rewriting the equation above gives

64 64 X 920 x 23 x 3
P11 =)a* 1211 — 0.352) x 504
p =21.4kN/m?
 CREEP AND VISCOELASTICITY 593

We could equally well have obtained the required modulus value by

taking a 1% strain section through the creep curves and plotting modulus
values against log time and extrapolating to 1 year.

21.14 Creep rupture of

The creep rupture behaviour of metals was described earlier and a similar
plastics
phenomenon also occurs with plastics. Under the sustained action of a
constant load, plastics exhibit a failure mode associated with the creep
deformation of the material. This type of failure is sometimes referred to as
‘static fatigue’, but the preferred engineering term is creep rupture or,
perhaps more generally, creep failure. The more general terminology of
creep failure is necessary for plastics because, although rupture of the
material will eventually occur owing to creep, there may be earlier visible
phenomena such as whitening, necking or crazing (crack-like features in
glassy plastics) which, as far as the user is concerned, terminate the useful
life of the component. Some typical creep failure data for plastics is shown in
Fig. 21.20. Two important points should be noted from this. The first is that
the creep failure data may be correlated with the isometric data obtained
from a constant strain section across the creep curves. A second very
important point is that, although the fracture as a result of creep is generally
ductile in nature, there is a tendency towards embrittlement in some
materials when subjected to constant loads for very long periods of time.
This results in a sharp drop-off in the rupture line in Fig. 21.20. Such
brittle failures can have serious consequences in practice since there is no
prior warning of imminent fracture and there is no ductile tearing of the
material to absorb the energy of the fracture. The possible ‘knee’ in the
fracture line is something to be wary of when extrapolating short- or
medium-term creep failure data to long lifetimes.

Fig. 2L20 Creep rupture beha-

viour ~
ies * Ductile fracture
Donia x
bes an we
4.5 feya .
|

g i ee
Se
n ee .
= Sai Ses
apne ~
Shes Seta eo eee
Ee ee See educa
roy . Uren— ae

ee we Oe 5 gpeateak
Pee *- Onset of whitening,
Isometric curves >i a5> necking or crazing

eerie GAUSS |
10° 101 10? 108 104 10° 10° 10°
Logi time

21.15 Summary
In the case of the design of metal components and structures, creep may be
regarded as not a very common occurrence. However, in the case of the
design of plastics components, it is of fundamental importance from the
start. For either type of material there is no exact analytical method and
heavy reliance must be placed on experimental data.
594 MECHANICS OF ENGINEERING MATERIALS

For metals the secondary stage of creep, which yields a constant

minimum creep rate for very long periods at low stresses, provides the basis
for design using an expression of the form ¢ = Bo”. This relationship can
also be used in the companion time-dependent phenomenon known as stress
relaxation.
While situations of constant stress or strain can be handled reasonably,
those involving changes in stress or strain at regular intervals of time are
very difficult to treat quantitatively. Creep-rupture testing provides very
valuable short-term data for the assessment and sorting of materials.
Creep and stress relaxation of plastics, while having many similarities
with the behaviour of metals, has the added features of memory and
recovery which make in-depth analysis very difficult. However, the
interpretation of basic creep curve data into isometric and isochronous
curves allows a convenient and acceptable approach to design by the
pseudo-elastic design method as illustrated in the worked examples above.

References
1. Pomeroy, C. D. (1978) Creep of Engineering Materials, Ch. 9,
I.Mech.E., London.
2. Andrade, E. N. da C. (1910) “The viscous flow in metals and allied
phenomena’, Proc. R. Soc., A84, 1.
3. Lessels, J. M. (1954) Strength and Resistance of Metals, John Wiley,
New York.
4. Larsen, F. R. and Miller, J. A. (1952) ‘Time—temperature relationship
for rupture and creep stresses’, Trans. ASME, 74, 765.
5. Tapsell, H. J. (1952) Symposium on High Temperature Steels and Alloys
for Gas Turbines, tron Steel Inst., London, p. 43.

Bibliography
Crawford, R. J. (1987) Plastics Engineering, Pergamon Press, Oxford.
Faupel, J. H. (1981) Engineering Design, Ch. 12, John Wiley, New York.
Finnie, I. and Heller, W. R. (1959) Creep of Engineering Materials, McGraw-
Hill, New York.
Metals Handbook, 9th edition, Vol. 8, Mech. Testing, (1985) John R. Newley
(co-ordinator), American Society for Metals, Metals Park, OH.
Penny, R. K. and Marriott, D. L. (1971) Design for Creep, McGraw-Hill,
New York.
Pomeroy, C. D. (1978) Creep of Engineering Materials, 1.Mech.E., London.

Problems 21.1 A series of creep tests on an austenitic high-temperature alloy gave

the following results:

Stress (MN/m’*) €9 (%) Minimum creep rate

(mm/mm/hr)

70 0.041 27-1058
105 0.061 15 510%
140 0.081 21 x 10°
210 0.122 15.8 x 10-3
280 0.162 0.281
350 0.203 2.62
 GREEP AND: VISCOELASTICITY 595

Calculate how long would elapse before a steady stress of

125 MN/m? would cause a strain of 1% in this material.
IND The following table shows the creep data obtained for a metal using a
stress of t100 MN/m/ at a range of temperature. If the constant “a”
in the Larsen—Miller parameter is 20, calculate the time to failure at
a stress of 100 MN/m? when the temperature of the material is
700°C. (R = 8.314J mol~! K).

Temperature (°C) 100 200 250 300

€) (mm/mm/hr) ileaiOseems a7 eel 1 25 Pal fo al0
Temperature 400 500 600
€) (mm/mm/hr) 1725102) .2.065c10- 68: 25510

213 The following creep rupture data was recorded for an alloy steel
when it was tested at a range of stresses and temperatures:

Temperature (°C) Stress (MN/m‘’) Time to failure (hr)

500 300 4724

600 200 570
700 100 MY)
800 60 Ways
1000 30 8.6

If a component made from this material is required to last at least

10000 hours at a stress of 150MN/m/?, what is its maximum
permissible service temperature? The constant “a” in the Larsen—
Miller parameter for the alloy is 20.
alle Figure 21.21 shows a delayed-action contact switch. When the pin is
removed the compressed spring causes a tensile stress in the
previously unstressed lead rod. Due to creep of the lead the gap
between the contact points decreases steadily. Caculate the delay
time if the free length of the spring is 40mm and its stiffness is
10 N/mm. For the lead ¢9 = 5 x 107!°075 mm/mm/hr. ~~

Fig. 21.21

| Contact

Lead
3.5 mm dia.

PAkegs) A sheet of high-temperature alloy is clamped in positon using a

toggle clamp as illustrated in Fig. 21.22. When the lever A is in the
vertical (clamp) position the sheet thickness is reduced by 10 jum. If
the sheet is subjected to a pull of 5kN, calculate how long the clamp
596 MECHANICS OF ENGINEERING MATERIALS

could retain the sheet in position. The coefficient of friction between

the clamp and the alloy is 0.6. Creep data for the alloy is given in
Problem 21.1. E = 207 GN/m’.

Fig. 21.22

21.6 An aerosol container is to be moulded from an acetal copolymer, for

which the creep curves are given in Fig. 21.18. The container
diameter is 50 mm and it has a uniform wall thickness of 2mm. The
base of the container is designed with a “skirt” to prevent rocking
when the bottom deforms under pressure (see Fig. 21.23). Calculate
the depth of the skirt if the container is expected to be subjected to
an internal pressure of 120kN/m/? (absolute) for 1 year. Poisson’s
ratio for acetal may be taken as 0.4.

Fig. 21.23 50 mm

2mm

Skirt

PAM A cylindrical acetal container is subjected to an internal pressure of

0.7 MN/m‘. For aesthetic reasons the strain in the container is not
to exceed 2%. If the diameter and wall thickness of the container are
60 mm and 1 mm respectively calculate how long the container may
be regarded as serviceable. Creep curves for the acetal are given in
Fig. 21.18.
21.8 A plastic snap-fit connection is shown in Fig. 21.24. If the pin will
slip out when the transverse clamping force exerted by the clasp is
33.N calculate (a) the clamping force when the pin is first inserted,
and (b) the elapsed time before the pin would slip out. Use the creep
curves in Fig. 21.18.
 CREEP AND VISCOELASTICITY 597

Fig. 21.24

21.9 A long thin-walled pipe constrained by end fittings made of

polyvinylchloride is subjected to a steady internal pressure of
700kN/m? at 20°C. If a tensile stress of 17.5 MN/m/? is not to be
exceeded and the internal radius is 100mm, determine a suitable
wall thickness. What will be the increase in diameter after 1000
hours? The mean creep contraction ratio v, is 0.45, and tensile creep
curves provide the following values at 1000 hours:

o(MN/m’) 6.9 13.8 20.7 27.6 34.5

€ (%) 0.2 0.48 0.92 1.72 3.38
APPENDIX ; Properties of Areas

The analysis of stresses developed in symmetrical and unsymmetrical bending of

beams (Chapter 6) depends on the shape and area of the cross-section of a beam.
The reason for this is because internal forces and moments derive from stress acting
on elements of area. To obtain the tota/ shear-force or bending-moment effect on a
cross-section we must sum up or integrate all the constituent elements and the final
expressions involve integrals for the total area (shape) in question. [t is these
integrals and these solutions which we refer to as ‘properties’ of areas. There are
direct comparisons with the ‘properties’ of masses which are used in engineering
dynamics or mechanics of machines. The integrals which will need to be evaluated for
any cross-sectional shape are described as the first moment, the second moment and
the product moment of an area. it may seem incongruous to speak of the ‘moment’ of
an area since a moment implies a mass or force multiplied by a distance. However,
the integrals do consist of areas multiplied by distances and that is how the
expression moment of area has become established (this is mot to be confused with
moment of inertia which relates to mass). Co-ordinate axes y and z are used
throughout for beam cross-sections, since the x-axis is along the length of the beam.

A.1 First moment of area Gee

Referring to the plane figure in Fig. A.1, the first moment of the element of
area dA about the z-axis is yd4; therefore the first moment of the whole
figure is [}4Y GA about the z-axis, the suffix 4 indicating summation over the
whole area. The first moment of the whole figure about the y-axis is {4244.

A.2 Position of centre of

é Let the co-ordinates of the centre of ar ea C.A.
A. be
be zZ and y7 as shownee) in Fig.
area, or centroid
A.1. Then the moment of the whole area about an axis is the same as the
sum of the moments of all the elements of area about that axis, or
1
Ay = |ydA - sothat =5| ydA [4.1]
A A Jy
Similarly,

z=5| 204 [4.2

Fig. Al
 PROPERTIES OF AREAS 599

If either or both of the z- and y-axes pass through the centre of area then Z
or y or both are zero, and

|ydA=0 and/or |BOA = 0

A A
If a shape has one axis of symmetry then the centre of area will lie on that
axis. If there are two axes of symmetry then their intersection will be the
centre of area.

Example A.1
Determine the location of the centre of area for the concrete beam cross-section
shown in Fig. A.2.

Fig. A.2

100 bc.

Since there is a vertical axis of symmetry the centre of area will lie
somewhere on that axis as shown by C.
Take as a reference horizontal axis the lower edge of the section AA, and
let the distance of C from AA be y. The cross-section can be divided into
two rectangles by the dashed line. The centre of area of the upper rectangle
is at C) at a distance of 100 mm from AA. The centre of the area of the lower
rectangle is at C; at a distance of 25mm from AA.
The areas of the upper and lower rectangles are 5000 and 7500 mm?
respectively and the total area of the figure is 12 500 mm’.
Referring to eqn. [A.1] above we may write

12 500y = 5000 x 100 + 7500 x 25

from which

2 =i)

A.3 Second moment of

If the first moment of an element d/4 about an axis is multiplied again by its
area
respective co-ordinate we obtain the second moment of area, namely y’ d4,
or z? dA. The second moment of area of the whole figure about the z-axis is

| pas denoted as_ /, [A.3]

A
600 PROPERTIES OF AREAS

and about the y-axis is

|z’dA denoted as J, [A.4]

A
The radius of gyration of an area A with respect to the z-axis is defined by
the quantity 7, which satisfies the relation

= Ar.

from which we can write

Ve =/4 [A.5]

In a similar way

yar 2I, 16
Second moment of area Common structural cross-sectional shapes are composed of rectangles and
for a rectangle the second moment of area of a rectangle is obtained as follows.
Because of the double symetry the centre of area C is at the centre of the
rectangle of width 5 and depth d shown in Fig. A.3.
Consider an element of area b dy as shown.
pela The second moment of this element about the z axis is y*(bdy). To
obtain J, for the whole section we must integrate between the limits of +d/2
so that
le
+d/2 yp +d/2
KS = ein | ybdy = ba
cc 2
| le —d/2 J —d/2

Sst = bd?
Saal
Fig. A.3 By a similar analysis we can obtain

db?
==
12
Example A.2
Determine the second moment of area for a solid circular cross-section of 50mm
diameter about an axis through the centre.

The element of area marked in Fig. A.4 is dd =rd6dr and the second
moment of this element about the z axis is (r sin Q)*r dr. If we now integrate
this expression between the limits of 0 to 27 we shall have the second
moment of an annular element about the z axis, which is
20
| (rsin 0)’rd@dr = rr’ dr
0
and for the solid circle the second moment is

Fig. A.4 R R4 70D)4 wx 50 4 3):

L= |= 3q r ae 4 64 64 4
= 3) 570mm 4
 PROPERTIES OF AREAS 601

A.4 Parallel axes

It is sometimes necessary to determine the second moment of area about
theorem
axes parallel to the centroidal axes.

Fig. A.5

Referring to Fig. A.5, the second moment of the element d4 about the
z'-axis is (y +b)’ dA, and for the whole figure

ly = |, +6)’ dA

=| ypat+2| yaa + |dA

A A A
but |,d4 =0, since it is the first moment about a centroidal axis, so

Ip =1,+0°A [4.7]
and by a similar analysis

ly=1,+@A [4.8]

Example A.3
Determine the second moments of area of the section in Example A.1 about its
centroidal axes.

Any section composed of rectangles can be broken up for analysis into its
separate components. Therefore, in the case of the vertical yy-axis which
passes through the centres of area C; and C2, we do not need the parallel
axes theorem, so

7s 100 x 503 ft50 x 1503

= 15.1 x 10°mm+
ae 12 12
In order to calculate the value of 7, we need to apply the parallel axes
theorem to both the top and bottom rectangles as follows:

_ 150 x 50°
(150 x 50)30* = 8.3 x 10° mm*
ee 12
The first term is the / about a horizontal axis through C; and the second
term is the area of the rectangle multiplied by the square of the distance
between C; and C.

_ 50 x 1003
+ (100 x 50)45* = 14.3 x 10° mm*
Sti?
602 PROPERTIES OF AREAS

The total value of /, is the sum of the two parts above:

I, = 8.3 x 10° + 14.3 x 10° = 22.6 x 10° mm*

A.5 Product moment of

The moment of area requirements for the analysis of symmetrical bending of
area
beams have been covered up to this section. However, for unsymmetrical
sections subjected to bending a further moment of area property is required
termed the product moment of area. It is defined in relation to the plane figure
shown in Fig. A.6 as

[4.9]

Fig. A.6

where the axes pass through the centre of area C.A. of the figure, Two
important differences from the second moments of area /, and J, are that /.,
can have either positive or negative values since z- and y-values can be
positive and negative. Secondly, if either or both of the axes are an axis of
symmetry then /,, = 0.
For the case of the product moment of area related to parallel axes z’y’, it
is straightforward to show that

Ty = Tey + abA [4.10]

Example A.4
Determine the product moment of area of the trianguiar section shown in Fig. A.7 in
relation to the centroidal axes z, y.

ty The equation of the diagonal of the triangle is y = h/3 — hz/k. Hence

2 2k/3 ph(k—3z)/3k
Ty = [oad = | | zy dz dy
—k/3 J—h/3
h ; Z
2 (2/3 :
3 =—, 32° — 2kz") dz = ———
6k? ie ) 22
Faron
3 3
 PROPERTIES OF AREAS 603

A.6 Spreadsheet
The area of the trapezium underneath a straight line segment from (z;, y;)
calculation of area and
to (2:41, Viti) in Fig. A.8 is given by (2; — 2;41)(¥; +3i41)/2. Note that
section properties of a if the order of the points was reversed a negative area would be
polygon computed.
Area under a line
segment

Fig. A8 yt

Area ofa triangle The area of a triangle defined by points (z1, 1), (z2, 9/2) and (z3, 3) shown
in Fig. A.9 is formed by the area under the line from 2 to 3 plus the area
under line 3—1 minus the area under the segment 1-2. Since z; is less than
%2, the area under the segment between points 1 and 2 is negative and
summing up the areas under the three segments will give the right answer,
1.e.

A= 5[(z1 Ss z2)(1 + y2) sea 23) (v2 + y3)

+(23 — 21) (93 +91) [4.11]

Fig. A.9

Area of a polygon For a polygon of n sides

n

A=5) (2 — 241)0%+941) [4.12]

i=1
604 PROPERTIES OF AREAS

We can create a spreadsheet to calculate the area of any polygon by making a

table of the (z,y)-co-ordinates of the polygon vertices, calculating the
contributions of the individual line segments and then summing them.
Enter the formula for one segment and then copy the formula for as many
segments as are required.
There are two points to note in constructing the table of co-ordinates.
1. The first point has to be repeated at the end of the list to account for the
line segment from the last point back to the first, forming a closed loop.
2. The points must be entered in anticlockwise order as you progress
round the boundary of the polygon.
A ZY graph of the vertices can be easily constructed in the spreadsheet to
check that we have correctly entered the data, though the relative scaling of
the z- and y-directions may not be the same.
To calculate the centre of area and the second moments of area of a
polygon about the origin (0,0) the equivalent formulae are!

ao) alee 2 2
Cae 64 20 ae Vi) (2; + SiS + i141) [4.13]

_t1t,
5 = yy ei — 2:41)OF+9011 tI)
1=
4.14]

Fig. A.10a i : Area First loments of Area Second Moments of Area

zz eee be ms _Mz My Z| y Wy
eS Sooo) So OL 0 0 BS 0
a -60. 0 0 0 0 Oo o|- raat
5 =e60) Cl a0 1000 _ ~108000 -15000 -15000 200000 630000
65 S10) te | 0] 0 0 0 8 ee
ee =40| 100 | 2000 -27000 -300000 -300000 4000000 1800000
S | =100 | | 0 0 0 0 0
ee |e |
tT aul | | zbar ybar
an : Totals : @SUM(C3..C9)/2 @SUM(D3..D8)/(6*C11)/ @SUM(E3..E8)/(6*C11) @SUM(F3..F8)/12 @SUM(G3..G8)/12 |@SUM(H3..H8)/72

a2 és p
ales) = ; ; Second Moments of Area about centroid 7
14 : : Ize lye iz : lzyc
| 15 ae = == +F11-$C$11*E112 +G11-$C$11*D1142 4H14-C11*E11*011 |
16 | —_—
aye Coordinates relative to centroid
i}iS, 7 Za y |

20° +A3-$D$11 +B3-$E$11

21 +A4-$D$11 +B3-SES11
22° +A5-$D$11 | +B3-$E$11
23 +A6-$D$11 +B3-$E$11
_-+A7-$D$11 +B3-$E$11
+A8-$D$11 |+B3-$E$11.
(a) Data and cell formulae
Note that the formulae in C3..H8 are too long to display here. They correspond to the individual terms in the summations represented
by eqn. ( A.12) - ( A.20) :
 PROPER S OHGAREAS 605

Fig. A.10b

0 —=45000.-=S«200000 |= 8640000 6300000

27000, i o. 0 0
300000 4000000 _ 360000 18000000
0 | 0 0

3350000 _ 750000 ___ 337500

Second Moments of Area about centroid

zc | lye \zyc
1612500412500, = 450000 |

(b) Spreadsheet display

1 n

Fe = 7a (2 — 2)OFFive +4 + Ii) [4.15]

ial

l : 3 2 2 3
din = Ya) (2j + 2p Zia + Ziq + Seep) [4.16]

Tey a iG mec fein (Ove. + 6yiyit1 + 3y;)

+2; (97 + Oyidi41 + 3y741)] [4.17]

The second moments of area about the centroid can be determined using the
parallel axes theorem from

Me =], = Ay’ [4.18]

Lg Ae A.19]

al iz) [4.20]
The resulting spreadsheet for the section of Example 6.11 is shown in Fig.
A.10.
606 PROPERTIES OF AREAS

Fig. AL

A.7 Transformation of
In unsymmetrical bending of beams it is sometimes necessary to consider
moments of area
the nature of bending about a set of axes rotated through an angle @ with
respect to a reference direction of axes as shown in Fig. A.J]. Let the
moments of area be /,, /, and J, with respect to the reference axes z, y and
Ty, Ty and I, with respect to different axes z’, y’ at an anticlockwise angle
@ to the former.

i |y*dA = |(y cos 0— zsin

0)’dA
A A
= I, cos’ 6 + I, sin’ 6 — 2/,, sin@ cos

=4(1, + 1,) +}(Uz — f,)cos 26 — I. sin 26 [4.21]

Li ee yee [tccos 6 + y sin @)(y cos 6— zsin 8) d4

A
(cos? 0 — sin? 0)J.,, + I, sin 6 cos 6 — I, sin 8 cos 6

=}(I, — I,)sin 20 + I, cos 20 [4.22]

These two equations provide the relationships between moments of area

about two sets of rectangular axes with a common origin. It is interesting to
note the similarity of form between these equations and those for two-
dimensional stress transformation, eqns. [11.13] and [11.14]. It does in fact
suggest the existence of ‘principal’ second moments of area about axes for
which the product moment of area is zero. From eqn. [A.22], putting
Ty = 0 we obtain

an 20 2Ty [4.23]
(Cpe ly)

and this defines the axes about which maximum and minimum principal
second moments of area J, and J, occur.

A.8 Mohr’s circle for

The simplest way of determining the principal second moments of area and
moments of area
indeed the moments of area about any set of axes is to construct a Mohr’s
circle in the same manner as that for stresses or strains.
 PROPERTIES OF AREAS 607

If eqns. [A.21] and [A.22] are squared and added to each other to
eliminate the angle 20 we obtain the following equation:

ey -4(4+5)P +22 =lG,-LY +h ay

Zyl
[4.24]
This represents a circle with axes of product moments of area (/,,), as
ordinate, and second moments of area (/,, J), as abscissa. The centre of the
circle is located at

I+
a)
and the radius of the circle is

Bd. -—1)? + 2)?

Since there are no negative values of second moments of area the circle is
always to the right of the ordinate.
Figure A.12 shows the circle construction for the shape shown shaded
top right and similarly top left. The circle is drawn from known, or
calculated, values of /,, J, and +/,,. For required values about axes z'y’ at 0
to axes ZY we draw the diameter ECF at 20 anticlockwise from ABC. The
values at E and F are those required. The principal second moments of area
I, and I, are the values at U and V, where /,, = 0. The directions of the
principal axces are given by the chords UG and VG which are shown as axes
Ou and Ov on the area (top left).

Fig. A.12

Product
of
moment
area

Example A.5
Use the circle construction to determine the principal second moments of area for
the angle section of Example 6.11 in which I, = 41.3 x 104mm‘,
I, = 151.2 x 10* mm‘ and I, = 45 x 10* mm‘.

The centre of the circle is located at (96.5, 0) and with the values of /, and
I, the circle is drawn as shown in Fig. A.13. The maximum and minimum
608 PROPERTIES OF AREAS

(151.2,45)

>
moment of area (168.5,0)

of
Product
moment
area

principal second moments of area are seen to be 168.5 x 10* and

23.5 x 10* mm‘.

A.9 Polar second

The second moment of area about an axis perpendicular to the plane of an
moment of area
area is termed the polar second moment of area and it is an essential part of the
analysis of the shear stresses in the torsion of circular sections (Chapter 5).
Referring to Fig. A.14 the polar second moment of area of d4 is r* d4 and
for the whole figure

Flor Ly) = | Pad

Fig. A.14

Also, since r? = z2 + y’,

| a4 +| y dA
A A

=I+1, [4.25]
This is known as the perpendicular axes theorem.

References
1. Cope, R. J., Sawko, F. and Tickell, R. G. (1982) Computer Methods for
Civil Engineers, McGraw-Hill, London.
APPENDIX
Introduction to Matrix
Algebra

B.1 Matrix definitions

A matrix is an array of terms as shown below:

411 42 G3 «.-- Aly

421 @22 423 --- Ady
[A] = | 431 432 433 «.-- A3n

Am) Am2 am3 se Amn

If = 1 then we have a matrix consisting of a single column of terms and

this referred to as a column matrix. If m = 1 then the matrix is called a row
matrix.
If in the analysis of a problem there is a set of simultaneous equations
then the use of matrices can be a very convenient shorthand way of
expressing and solving the equations. For example, consider the following
set of equations:

Vi — Aas Clea 1013s) te 1 in

DP ODEN 4 CIPD PIS Ae 88 a pe

Y3 = 431%) + 432%2 + 43343 +... + 23nXn

Vn = Am) HF On Q02 => Ain3X3 1 +. + Any

These may be written in matrix form as follows:

{y} = Alt} [B.1]

where {y} and {x} are column matrices.

B.2 Matrix multiplication

The matrix equation [B.1] involves the multiplication of the matrices [A]
and {x}. To do this one must apply the simple rules of matrix multiplication.
These are:

(a) two matrices may only be multiplied if the number of columns in the
first is equal to the number of rows in the second;
(b) the terms in the product matrix resulting from the multiplication of
matrix [A] with a matrix [B] are given by
610 INTRODUCTION TO MATRIX ALGEBRA

Ci = ‘iyAikpj [B.2]
zl
The use of these rules is illustrated in the following example:

41 a2\}by dp Ay
421 422 || br bor b23

a (a11b11 + 412621) (a11b12 + 412522) (a11b13 + 412623)

(42111 + a22b21) (a21b12 + a22b22) (a21b13 + a22b23)
Suppose

i=[6 sf @=[3 2 7)
Then

= (aei=|5035 10|

B.3 Matrix addition and

Matrix algebra also involves the addition and subtraction of matrices. The
subtraction
rules for this are as follows:

(a) matrices may only be added or subtracted if they are of the same order,
i.e. they each contain the same number of rows and columns;
(6) the terms in the resulting matrix are given by
Cy = ay Se bij [B.3]
The following example illustrates the use of these rules:

ie dy. dy3| |b, din by3

dz, dy do3| |br bar ~—a3

(dz, br) (dx £b22) (dx + 3

Suppose

—2 4 5 eee: 3
p\=| 6 8 Z ae 2 a)
Then

HI=1I+Bl=|~5
1) 4 —l1 6 8
 INTRODUCTION TO MATRIX ALGEBRA 611

B.4 Inversion of a matrix

Referring back to the set of simultaneous equations at the beginning of this
appendix, the objective is usually to solve these for the unknown x terms.
This is where the use of matrices has a major advantage because referring to
eqn. [B.1] we may rewrite this as

{x} = [A] fy} [B.4]

This equation expresses the solution to the set of simultaneous equations
in that each of the unknown » terms is now given by a new matrix [4]7!
multiplied by the known y terms. The new matrix is called the imverse of
matrix [A]. The determination of the terms in the inverse matrix is beyond
the scope of this brief introduction. Suffice to say that it may be obtained
very quickly on a computer and hence the solution to a set of simultaneous
equations is determined quickly using eqn. [B.4].

B.5 Transpose of a
The transpose of a matrix [A] is denoted by [A]’. It is determined by
matrix
exchanging the rows and columns in the original matrix. Thus referring to
the matrix [A] at the beginning of this appendix, then

@\) 421) 4&3) «-- Am)

412 422 432 «+--+ m2
[A]? = |413 423 433 «++ Om

Gin 4n 43n Amn

So if

B.6 Symmetric matrix

A square matrix is one in which the number of columns is equal to the
number of rows. An important type of square matrix which arises quite
often in the finite element method is a symmetric matrix. Such matrices
possess the property that a; = a. An example of such a matrix is given
below:

2 4 7 —3
4 5 1 9 oe ss
7 6 —5 which is often written as

—3 9 —5 4
612 INTRODUCTION TO MATRIX ALGEBRA

2 4 7 =3
:ee 9
sym 6 —5
 “plOjXO ‘syoMsoyNg ‘uonIps wg ‘yoog auasafoy s[vIaWw (9261) syjeusus “[ “D “39 ‘sonsodosd
Joyjo pur sansodosd jo sonjea
[er4ayeW UO syooqpury paystgnd st Jo du0 Jo UONKyIDEds pivpurIg YsHg JUeAspes dy} WYNSUoOD pnoys Jo proi ou) ‘setiayeur
JSOW Ul palivA aq ued INQ eordAy Ayarey are
payiejop os0ur 10, “oI9 ‘ayes ures “ainqesodus} Quounve.s) yoy ‘uorisoduroo se yons s1ojoey Aq A[Qesopisuod AJA Sased
oY} dAIZ 0} papudzUT ATUO SI 9[qQuI SITY], “A707
UOAIS SONA [LOLIowMU sy], ‘TeLIoqet Jo sodA] snorswa WOT o]QRIIVAL saijiodoid jo a8uei opiM ay} JO UORLOIpUl Ue Jopeet
“w YS] UO |
ol 8) ‘uoIssa1d uo’) ,
es
0061 e+ = +-0 OST 09€ OIST 0001 uoqssunJ],
3 _ 70
00St-OLtr 01-8 06r ZI-0l Or 901 0F01-006 O16-0SL Aoye wmrueyf,
<D) ££0
££°0 O0Sr 01-8 7 4 Or OIl 00S 00r (eand) winrueity,
—= ©
S06L CLI 919 6 L8 961 S6ZI OZIT [oa1s ssopureas
fa) e £0-L70
0007-098 00Z-0E1 = 000L-0S1 +00°0 7E-S = (prey) roqqny
; Y o = S0-S+'0
= = FS Al (aa 6S (Arp) Yeo poy
c On = 169
0'I-S'0 = 00-01 5 £01 OL-0E Se Daan
- S Al) OOST-O0ET
096-416 RL I = 0071-001 1-70 Go=8 = auoy jog
To ea Sr 0-40
OSZI-OOIT L0-+0 > O£I-001 i 0'€-07 99-95 - ayeuoqivoAlog
+0
OSIT 01-80 = 00€-09 7 Rie OKC 98-S9 = wo[AN
0 pal +0
TCBL STI Szs 61 LL €0Z S801 +26 [2938 OW-ID-IN
vo S €0-L7'0
OS8L Gil 56S tl $78 LOZ OSZI 000T [2938 [9YSIN
> € 0-270
OS8L ZI 407 SP 18 LOZ 29r 082 [2938 PIL
N £0-L70
Se0 S781 97 c€l ZI SOI Sb ere Sr7 oye umrsouseyy]
cae 0 aA
iS = S€-0Z 08-0 000I-0E = Sse)
Oo} 3 170-70 0087-0062
809 = = = > Il Es (a3 Qem) ay sepsnoq
= or =
oOo Vv D - 09s - ~ - tl SZI 96 (Aap) ay sejZnoq]
9¢ 0-€£'0 0068 LI z 0s 9b-OF 0zI-011 00r 09 (and) 1addo>
Oo oo. om)
00rZ S01 = \sr'0 = S81 er Or ,232.19U07)
om Oo «3 70-10
£0-70 ZSEL ral SI¢ (as = SLI SEL = (az[npou) wos se_
mol
€0-70 ZSEL ZI 611 90 SLI 08Z (aye) wort yse_
— oly =
OS8L ZI L87Z 0€ 78 802 209 OLE [aeqs wogreD
€0-L70
109 CLI OIZ S€ Ly 7 9bS 08Z azuoig
70
OEb8 S81 c€l Or se 101 LU 6S7Z sseig
+0
c£0 06L7-9797 €7 OrI-0Z1 Gees 87-97 ZL-OL OSS-0ZE 0S+-0S7Z Aoyje wnruramyy
c£0 OILZ £7 = 09 97 OL 002 OF (end) wnrunanyy
+0 00ZI 9°0 = 8-7 3 CeaInG 08-05 MAID
S ; b =
= (,wW/NW)
an (D./ 5-01) sapo.Xo (U/W)
uorsuedxo OL 3 yay (,W/ND) (,W/ND) (24/NW) ssa.ns
us jooid o%]0)
(<u / 34) yeutsoyy an3snery 10 wus uO sny~npow snjnpou = yysue.s
. one.
Tvay§ $.8un0X o[tsus 1O PPA [elioqyeyy
Ayisusq] jo “Jg207) sourInpuy uonesuoyy %
<x S UOSSIOd
APPENDIX
Answers to Problems

Chapter 1
Fr = 202, M, = —3500
M,=-—12kNm when F;=1, F;3 = 18.8 kN
M, = —3536, M,=11314, M,=3536 Nm
aSS
mae
We For A, M,=-—250, M,=300, M,=0. For B, M, = 250,
M, = —300, M,=0
(a) 160.9 N_ (4) 201.1 N
1480 kg
809 N
Dasa
-257.2 kN
0.61 MN at 30° to horizontal
Yes
Fxg = 505 N, ReE= Rp = 400 N
(a) w> 0.62 (6) 1.55 kN
2.9 KIN; 27.7"
F4= —24.25 kN; Fp = —48.3kN; Fo = 18.18 kN.
Fie=]05 KN;. By =5-0EN| (Ry = 7-4EN:
me©
oma
ome
ee
Se (a) V4 = 1960 kN;
HBWN
CONAN Ay =2940 kN; Ag = 2940 kN.
F 4p = 2940 kN; F pr = 0; Fre = —3533 kN.
(b) Va =THIOKN; B7=Z505. EN; Ae — Z565EN:
Fygp=2565 kN; Fer =OkN; Fre = —3082 KN.
ial? Ve= 223 EN, y=] Va = F285 EN:
F 4p = 0; Fre = —42.8: Fac aa 16.1; Fup = 36;
Foc = —22.8: For = 48.4.
Fro = —20; Fog = —48.4; Feo = —3.6; Fer = 40.3;
Fro = —72.
Fou = —53.5; Fry = 69KN.
1.20 For = —15.9kN
evAll Fug = —29.4 KN; Fyz = —39.24kKN; New Fyrz = —34.33 kN.
122 (d@) Fag = 1225N, Fpc=—707N, Areaapg = 12.25 x 10-°m?,
Areagc = 7.07 x 10-°m?, Weight = 0.221 kg
(e) Min wt. when / = 1.414 m is 0.221 kg
1.23 Min. wt occurs when h=1.272m and Fag = 1272N,
Fgc = —786N, Area =12.7x 10-® m’, Weight = 0.260 kg
1.24 DFE;.-—8; DE, 8; DA, =20; FE, 3; FA, 11; PC). =283 EC =—42- BA,
—11.3; EB, —9 kN
125 AD, 0; AF, 100.6; BD, 0; BF, 100.6; CF, —200.2; DE, 50;
DF, —70.5; EF, —70.5 kN
ea Max. Axial Porce =>2ZkN, Max SE =2EN, Max BM
= 0.4kNm
1.28 (a) S.F.: A, 0; B, +1; C, +1/ —7; D, —7/ +3; F, +3 KN.
B.M.: A, 0; B;0; C; Fl; D,, —6; F, OkN=m,
(b) S.F.: A, —2; B, —2/ +7; E, —11/ + 8; F, +8kN.
B.M.: A, 0; B, —2; E, —8; F, 0kNm.
(c) S.F.: A, 0; C, —8/ — 24; F, —36 kN.
B.M.: A, 0; C, —8; F, -98 kN-m.
 ANSWERS TO PROBLEMS 615

(d) S.F.: A, 4; B, 4/ — 3; F, —3 kN.

B.M.: A, 0; B, 4; D, -2/6; F, 0kNm.
1229 S.F.: D, —29.44; C, —29.44/ — 22.08; B, —22.08/ — 7.36;
A, —7.36kN.
B.M.: D, —147.2; C, —88.32; B, —22.08; A, 0kNm.
1.30 RUNG? 15 Me 02) At eee Me = 025 Mi
O25, ACRES Ei 2095 hy 2255) Ate 1) Pp 285; FS AC
supporiwA: fp 2.9; = —=—2.5,, FF, = 0; -Atrsuppott)B:
Pee ey eo ee — Ne VM O25, M, =
0.05 kNm
Torque: AB, +286.5; BC, —668.5; CD, —191 Nm.
B.M.: A, —6; B,+2.8; C, —0.4; D, —10kNm.
Chapter 2
7.64 MN/m* in top of rod.

A af olow
or
53037mm?; 10.5 MN.
1720 mm?

wl
Prnax = 2 sin d
27.6m; 3.68 MN/m’.
9mm.
22.4 MN/m?
12MN/m?; 211 MN/m’.
ao = (Wcosf) / (27hx tan B)
09 =736kN/m* 07, = 552 kKN/m’.
70MN/m’*; 138 MN/m?.
(a) 8.3MN/m? (6) 3.4MN/m?_ (c) 61.2 MN/m?
Key: 131Nm; Pin: 138Nm.
49.5kW; 3.15 MN/m?.
Torque varies from 6.5 Nm to —20 Nm
4.56mm.
(b) t = 0.00456 m
11.31 kNm.
(b) 89
Chapter 3
0.014 mm
(a) Plane stress (4) Plane strain (c) Plane stress.
0.413 mm
94.5 MN/m’.
0.000941; 0.0188 rad/m.
E
C= awa) [(1 —v)e, + v(ey + €2)], ete.
E
e= fap + Ve,}, ete:
E
Oy, = [(1 — vey + vey], ete.
(1+v)(1 —2v)
616 ANSWERS TO PROBLEMS

0.00078; 0.31 mm.

162MN/m’; 194MN/m?’; —0.0005.
0.87 MNm/m’.
37.6 kN/m?/m?.

Chapter 4
4.1 122MN/m?; 19.5 MN/m’.
4.2 0.533mm
43 5.069mm; 5mm.
4.4 (b) As Prob 4.1 (F = 38.37 kN)
(c) As Prob 4.3
4.5 Fx = 25.2kN (tensile); Fg = 48.36kN (compressive)
46 386MN/m?’ (steel); 214 MN/m? (copper).

EA; ee Sia
ape
47> (@)tok= iF (d—do); (b) d= —

(d) as Prob4.6 (d=0.214 mm)

48 Cylinder 89.6 MN/m?*; Rods 29.6 MN/m’.
49 1.24m from left end.
410 124.2kN.
4.11 23.9kN; 76.1kN; 95.3°C.
4.12 16.3MN/m?’ (steel); —65.3 MN/m? (copper).
4.13 1634mm’.
4.14 0.151mm; 31.4MN/m?; 26.2 mm.
4.15 6.1MN; 34.1 MN.

Chapter 5
SP RD,= 12 mime Ty 35,
5:3 O:332-rad
5.4 3.14kW; 19.7mm; 11.52 mm.
5.5 163.3 mm.
5.6 CD, 6.52MN/m?; AB, 3.7 x 107? rads.
5.7. 40.3kW; 20.1kW.
5.8 5.33kNm.
5.9 48.34mm; 6.85 mm.
5.10 90.3 Nm.
5.11 + =0.469MN/m? (2mm wall); +t=0.33MN/m? (3mm _ wall):
0.285 x 107% rad
5.12 1mm; 0.142rads.
5.13 2.75kNm; 0.0405 rads.
5.14 324.5Nm.
5.16 (a) Assuming G, = 10? and G, =2.2 x 10°, 6,=0.41 = O:756x
6, = 0.548.
(b) = 55x 10°N/m?, 7, = —37 « 10°N/m?.
Chapter 6
6.1 S.F.: +4, +4/-—1, -1/-6, —6.
B.M.: 0, +8, +7, +12, 0.
6.2 By, 01 Nm JG31355.Nim.
O40 A, Oo BAO er G60. "G02 GE te ya dO DD b)
70; E, 0kNm.
ANSWERS TO PROBLEMS 617

65. O= =58.85%" M19. 624°.

6.6 —Omax = —656 MN/m’ at 4.75m from left support.
O10 = 0 1 2, 3 4 5 6
Q +8 +6.8 +3.8 0 5.05 0.8 =o
M 0 7.6 130, 15 13.0 7.6 0
6.8 9S. 5:5, 3:5, 3.5/4" =4 kN
B.M.: 0, +7.5, +4, OkKNm; Mmax = 8.6 at x = 2.34.
6.9 36.41mm; 91.38mm.
6.10 5.68 m.
6.11 (i) —240MN/m? (ii) at y = —103.4mm, z = —66.9mm
6.12 h=(4B
tan a)/9
6.13 (i) o=3.7MN/m’ (ii) o = E35MN/m*
6.14 27.5MN/m?; 22 MN/m?

aL
6.15 x= ————_
2(d2 — d))

6.16 Mag = 560N-m; Tap = 490Nm; Mpc = 501Nm; ite

123 Nm.
oap = 263MN/m?; tap = 115 MN/m’.
6.17 M,/M) = 1.43.
6.18 27.9mm; 7.93kNm.
6.19 Otimber = 1Z2MN/m?, Ggtee) = 164 MN/m?
6.20 GOtimber = —1.6MN/m? to +6.4MN/m? ete) = —48 MN/m? to
—32 MN/m?
6.21 38kNm; 89.1MN/m’.
6.22 —10 to +26MN/m?; 45.84 mm.
6.23 51m; 1.63-MN/m’.
6.24 66.9mm.
Os 55g Tn ny 3:
6.26 Sides: 0 at top, 104.6 MN/m? at N.A.; 85.2 MN/m?’ at floor; Floor:
17 MN/m’ to 0.
6.28 1.14MN/m?; 1.04MN/m?; 25.6mm.
6.29 4.84kN; 4.3kNm.
6.30 7.5mm.
6.31 (a) 37.5 MN/m’; (b) 1 mm
6.32 o4=108MN/m’?; of = —40MN/m’*; oc = —110 MN/m?
633. Tae 18 MN/i Tren = 115 MN/m*
6.34 +136.9MN/m’.
6.35 L,—119.2MN/m?; M, 70.2MN/m’; N, 15.1 MN/m/?;
16.2° to z axis.
6.36 —13.08° to z axis; 54.7MN/m?’; —33.5 MN/m’?.
6.37 43.4mm; 155MN/m’; 266MN/m’.
6.38 (i) (a) at the intersection of the two plates forming the section;
(b) at the centroid;
3H
GE eras
(ii) Tmax = 10.8 MPa
6.39 2.37KN.
6.40 o4=23.83MN/m’; of = —15MN/m?
6.41 78MN/m? in the larger loop.
618 ANSWERS TO PROBLEMS

Chapter 7
7.1 —8.54mm.
7.2 492mm.
Ta S620 1577 am:
Veiga 0, 0 = 3529
7.6 (a) |= 0.586 (4) § = 0.554
7a. A, 8.66mm; B;2:27 mm; C, 4.98imm.
78 19.2mm.
TAY Biges2s,
7.12 —0.5mm; 1.48mm; 5.14 x 107% rads.
7.13 28.1mm; 32.3° anticlockwise from z axis.
7.14 0.2mm.
7.15 +0.0014 rads.
7.16 (a) 2.44mm; 152.2MN/m? (4) 1.09 mm; 68MN/m’?.

Chapter 8
81° 64 8KN/m.
57 7 9 Vee?
R00) 251s Sa phe —— pl:
Ws Sea is. 6144E1
ML M [ x? Sx? Lx =L?
ee eee "
16EI nl 4. <8. ee 8
8445 mm.
8.5 —250Nm; —187.5; +2938; +2250N.
8.6 63.8mm
8.7 (a) 3M/2L; M/4. (6) 13wL*/6144ET.
8.8 0, 80, 5, 37.5, OkNm.
8.9 S.F.: —4.06; -—4.06/10.62; 10.62/ —9.38; —9.38/ — 2.18;
—2.18/10
B.M.: 0; —40.63; 65.6; —28.13; —50;
S210) SSP 2 27-9 G- B Or Gog |Sait: 11/0
B.M.: 0; 48.1; —28.8; —9.6; 44.8;
Sl Ly = 0.37/50. Wh or

Chapter 9
9.1 Long bolts.
92 Oph — Bo.

93 U7) Ue 16:
94 57.9mm.
95 4kN/m; 10 coils.
96 137N; 4.1mm.
9.7 90.4mm.
9.3... 2W/ET.
99 43.1N.
912) BS Was aN aM = 0.093 5, = 0.019Wr3 /EI.
9.13 6,=aWr/8EI, 6, = —Wr /2EI.
9.14 F=EI6/3nR’.
9.16 0.412 mm vertical; 0.23 mm horizontal.
 ANSWERS TO PROBLEMS 619

9.18 33 mm.
DAL (a) Energy = F*L/2AE (b) Energy = (3)(F°L/2AE).
9.20 0.071 rad; 424 MN/m?.

Chapter 10
2kN.
1.88kN.
50 mm.
wr Eth? ae krEt
(i) P=4hto,; (ii) P= ee eae te i a= eS

iN = ae ano
(i) 20mm? (ii) 90.2 mm?
90.2 mm’.
1026 kN.
1.677 BI / 17.
1.33 kN.
54.65 kN.
—6.72 MN/m.
kE
(a) :=< (6) 148. 5Gir2Z7i.9. (c) k= 089mm,
J

ps Simm

(a) buckling would occur (A) 2.4.

30mm.
37.8 MN/m’.
h = 0.5 for minimum weight.

Chapter 11
(2) 0, = 100; o, = 200 MN/m? (b) Top: 0, = 240 MN/m’;
7, = 0] Neutraliaxis: 7, = 6MN/mr (@)o, = —10.2:
Ty = 122 MNJ
(@) dg, = 199.65 -7-— 5.98 MN/m”” "(b) 0, — 30:
i = NOMN/me. -@)c, = =180:8. t= — 3935 MIN/ a
56, 21.5 MN/m’.
INS for JAE2-
60.4; —10.4 MN/m?.
(a) —47; 67MN/m’; (b) 60, 20, 30, 17 MN/m’.
A: 228, =28,.128-
B: +80, —80, 80;
C: 28, —228, 128 MN/m?
46.3MN/m’*; 69.3, —23.1 MN/m?.
12.5, —0.3 MN/m’?.
2.25: 6.75 kN.
132 MN/m?; 88.13 MN/m/?.
V2 ONKIN ee —— Ulli2IN
TA \0v? = 6.17-x.10;*.
287, —200 MN/m?; 243 MN/m’?.
141MN/m?; 2.36 kNm.
(i) €, = 1000 x 10-°, ey = 625 x 107° (ii) 0) = 91.3 MN/m,
620 ANSWERS TO PROBLEMS

0, =71.2MN/m? (iii) 86.3 MN/m?

11.18 1.13 kNm.
11.19 11.35 x 10-4, 6.15 x 10-*; 19° anticlockwise from gauge A.
11.20 153.7; -—64.6; 109.2MN/m’.
11.21 E=69.9; G=26.9; K=58.3 GN/m’?; v=0.3.
54
11.23 (a) 0.7509, 0.250, 0.43300; (b) 277 MN/m? by shear.
11,246, = —245 «10-4; = 6.85 10 -*
1125 Boiss, E= SGN me

VNR NE To Oso nin et? kan

12.2 4.45KN.
12.3 1.037MN; 1.037MN.
12.4 9.3mm,
12.5. yielding would occur.
12.6 143N.
12.7. 45.4mm.
12.8 15.1mm.
12.9 66.7MN/m’*; 424kN.
12.10 520 MN/m?.
12.11 3.37 MN/m’. ’
12.12 0.265mm; 61.7; 60.55 MN/m‘.
12-13.549.7 KN.
12.14 7.53 mm.
1215-335 mm:; 6.4:

caer 13.3. Cylinder axially constrained.

ay Ven Oye, Oe & oz
13.4 = =; (b = r—
Ona: one 3 "Az

CNeD eS aaOOo LOleMN

14.3. 1.915; 110; 30MN/m’.
14.4 65, 66.7MN/m?; 1.05.
a’ — 190?
14.5 SS
(P= @)
—3up p(2—v) p(1 — 2v)
oo eet Be ee ET
14.6 Out 5 i — i i == ee

Inner:.6, = pate +v)+2v— 1},

Ae=P
a) = ———_ '{kr2 (lt+v)+1-2v},
(1 - & = Ae p +1 - 2v)

14.7 396 MN/m?; 28.7 x 107+.

14.9 18.4MN/m?.
14.10 (a) 654kN (6) 1308kKN_ (c) 330 MN/m?.
14.11 (a) 10m (6) 12.6kN (c)0.5x10->m_ (4d) plate.
 ANSWERS TO PROBLEMS 621

14.12 22 KN.
14.13 (a) 5.93 MN/m’, 4.81 MN/m?_ (6) 4.81 MN/m’.
14.14 Inner: —63, —49.2MN/m?; Outer: 49.3, 35.5 MN/m’.
14.15 1.25mm; 1000 mm.
14.16 (i) 54 (ii) 139.3, 85.6 MN/m?.
14.17 20.3 MN/m’.
14.18 Outside, 2.21; Inside, 7.46 MN/m?.
14.19 2.83 Nm
14.20 1.61, 13.56 MN/m’.
3
14.21 (a) (=) pwr? (6) GFRP.
2pri
14.22 op ae=s"1, 2
ty
+9(Z)
60
(a4 +4 + u(t -A)}
14.23 0.128 mm; 12.7 MN/m’.
14.24 Assume 1 atmosphere = 0.1 MN/m? pressure difference and
clamped edges, thickness = 5.3 mm.
14.25 7331 rev/min.

cmon eT Or ie CCDs
152 475mm; 91-93 m_ from,each end.
1523) eZZ4 KN,
15.4 10.4mm.
15.6 m,=11.7M,/L’; 0.414 from free end.
15.7 F,=M,.
15.8 56.5mm; 3°; 31.4kN-m.
15955108) 137
15.10 5328 N-m.
15.12 44.5%.
15.13 250, 150 MN/m’?.
15.14 10.6mm; 3.47KN.
15.15 —17.4MN/m’.
15.16 15960rev/min; 10%; 6.85 mm.

pPanere Liey ydos) aes

16.3. 3.1mm.
16.4 —1.13mm.
16.5 0.25mm; 166, 48 MN/m’.

16:6" for <0) Be nL cos Bx);

, 2Eh

for x > 0 ee) Atha

= 0) M0)
orcs Eh : oe
16.7 68.8, 19.9 MN/m?.
16.8 M, = —-1.75Da.
16.9 0.840 mm.
16.10 0.47 MN.
16.11 6.45 mm.
622 ANSWERS TO PROBLEMS

16.12 0.113 mm.

16.13 —90.64; +90.64kN/m?.
16.14 1.5m, 2.94MN/m?; 2.25m, 2.2 MN/m.
16.15 8.2, 469 MN/m’.
16.16 487 MN/m?.
Ley, 5.03 MN/m?.
16.18 4.22 MN/m’.

Chapter 17
0, = 0, 02 = 0.0490, 03; = 0.0857 (Example 5.3)
0, = 0; = 0, 62 = 0.0108 (Example 5.4)
—0.15, —0.325 mm; 65.4, 9.6 kN.
—0.37mm; 9.6, —16, —12kN.
—1.64mm.
we 5 wi
2) retary CO) aan pp

Chapter 18
198GN/m*; 270MN/m’; 315MN/m’; 451 MN/m/?;
24.4%.
67.6GN/m?; 290MN/m?*; 335MN/m’; 413 MN/m‘?.
16.
Wrong tempering temperature.

Chapter 19
36.3 MN m~?/2,
Kic = 39.4MN m~*/*, estimate OK.
2380; 26); 200; 5.9: 34.0% 0:75 mm,
Not acceptable.
Test is valid, K7¢ = 149 MN m-?/?.
1.82 m.
Yes.
he 255 kN.

19.10 () K = oy
1 (i) 14mm_ (ii) 12.4mm.
19.12 0.305 mm.
19.13 OK, Ki = 23:< Kie
19.14 0.6mm; 1.5%.
1945 62.8 MN m-3/2.

19.16 (@) )m=— =—

3.36V 0b
Wi) =mr +L)
(0) := 10.3. For r < L, use (i).
L907, 166 MN m-?/2; 14.1%.
19.18 142MN m?/?; 1.07 mm.
1919 0.106 Nm; 5.08 Nm.
 ANSWERS TO PROBLEMS 623

Chapter 20
201 9.73mm; 23.28mm.
20.3 Medium-strength steel.
20.4 2.13 MN/m’.
20.5 Goodman, 5.43; Gerber, 6.6; Soderberg, 4.6mm.
20.6 69, 78.4 mm.
20.7 155 KN.
20.8 At central hole.
20.9 High notch sensitivity.
20.10 50000 cycles.
20.11 11 363 flights.
20.12 10.7 MN/m’.
20.13 518 MN/m?.
Chapter 21
pala 105 hours.
21.2 262 hours.
aS 544°C.
24 7.1 minutes.
215 3.31 hours.
21.6 1.1 mm.
2d 104 days.
21.8 70.8N; 13.9 days.
2A9 4mm; |.] mm.
 Index

Anticlastic curvature, 134 of slender members, 21 laminate analysis, 329-34

Arch, 7 pure, deformations, 133-4 stress and strain transformation, 321
Area relationships, 130-2, 136 Composition, effect of, 506
centre of, 598 section modulus, 137 Compound bars, 82-4
first moment of, 598 sign convention, 23 Compound cylinders, 391-6
polar second moment of, 608 stress, 125 et seq Compression
product moment of, 152, 602 Body force, 368 strain, 64 et seq
properties of, 598-608 Bolt, stress relaxation, 583 stress, 43 et seq
second moment of, 599 Bridge, suspension cables, 50 testing, 497
transformation of moments of, 605 Brinell test, 501 Concentrated load
Area-moment method, 220-32 Brittle fracture, 515 on beam, 354
Asymmetrical bending, 151, 204 Brittle materials, 348-50 on finite plate, 353
Asymmetrical section, plastic bending of, Buckling instability, 264-91 Concentric shafts, 109
416 Bulk modulus, 68 Concrete, reinforced, 144-7
Autofrettage, 390, 425, 434 Concurrent forces, 3
Axial symmetry, 370, 373, 382-406, 442- GOs 527, Connecting rod, (conrod), 47, 49, 84
459 Cables, 7, 50 Connection, types of, 7
Axis, neutral, 133 et seq Cantilevers, 7, 24, 129, 188, 214, 243 Constants, elastic, 68
Car bumper, 76 Constrained material, 93-4
Barrelling in compression, 497 Carbon content, effect of, 507 Contact stresses, 353-6
Bars, curved, 168 Carbon fibre, 322 Continuous beams, 215, 225-9
Beam(s), 7 Castigliano, 250 Contraflexure, point of, 28
bending stress in, 125 et seq Centrifugal force, 51 Corrosion, 562
cantilever, 7, 24, 129, 188, 214, 243 Centroids, 598 Costs, 77
column, 281 Channel section, 7 Couple, 3,5
continuous, 215, 225-9 Charpy test, 534 Crack
curved, 168 Clapeyron, 227, 386 extension force, 517
fixed, 211, 217-19, 223 Close-coiled springs, 241, 243 fatigue, 544
of dissimilar materials, 139-47 Coefficient of linear thermal expansion, 70, growth 553
of I-section, 160 89, 90 initiation, 514
of uniform strength, 205 Columns, 6 opening displacement, 527
propped, 214 buckling of, 264-91 propagation, 514
shear stress in, 160-8, 380 empirical formulae, 279 tip deformation, 521
simply-supported, 26, 125, 127, 186-8, Euler theory for, 265-8 Crane hook, 171
194-6, 253 with lateral loading, 280-4 Creep, 573-97
slope and deflection, 185-210 Combined bending and torsion, 29, 305 empirical relationships, 574
statically indeterminate, 22, 211-34 and end loading, 147 fatigue interaction, 586
with applied couple, 197 Compatibility equations, 373-7 in pure bending, 578
with irregular loading, 223 Complementary energy, 250 of plastics, 588-93
Bearing cap, 85,86 Complementary energy for deflections, 249 rupture tests, 577, 593
Bending, 1 et seq Complementary shear stress, 45 stress relaxation, 582
and direct stress combined, 147 Complex stress test, 344 testing equipment, 578
and torsion combined, 29, 305 Compliance of matrix, 323, 331 under multi-axial stress, 580
asymmetrical, 151, 204 Components, interaction of, 82 variable stress and temperature, 584
moment, 23 et seq Composite shafts, 108 Cumulative damage, 563
moment and shear force diagrams, 24, Composites Curvature, radius of, 185
125 et seq fatigue of, 567 Curved bars, 128, 168
of open sections, 164 lamina analysis, 322-9 slender, 128, 168
 INDEX 625

Cylinders Fatigue, 84, 544-72 Glass fibre, 322

compound, 391-6 corrosion, 562 Griffith, 515
plastic deformation in, 425 cumulative damage, 563
residual stress in, 434 curves Hardness and strength, 506
thick-walled, 382-97 e-N, 549 Hardness measurement, 500-6
thin-walled, 54 S-N, 548 Brinell, 501
Cylindrical bending, 442 limit, 548 number, 502
low-endurance, 548 Rockwell, 505
_ D’Alembert, 51
mean stress, 558-9 Vickers, 504
micromechanism, 552-4 viscoelastic materials, 506
Deflection
multi-axial stresses, 564 Heat treatment, effect of, 506
by strain energy, 249-57
of plastics and composites, 567 Helical spring, 240-3
due to bending, 185-210, 252-7
size effect, 561 Hole in plate or bar, 357
due to shear, 243
statistical nature, 550 Hollow shafts, 106
of beam of uniform strength, 205
stress, concentration, 559 Hooke’s law, 67
Deformation, 64
stress cycles, 544-6 Hoop strains, 333
Deviatoric stresses, 343
surface treatment, 562 Hydrostatic stress, 43, 45, 66, 91
Disc
temperature, 561 Hysteresis, 500, 583
circular solid, 399, 401
test methods, 546
circular with hole, 400, 402
Fillet radius, 359 L-section beams, 160, 416
plastic deformation in, 429-31
Finite element method, 463-92 Impact stress, 95
residual stress in, 433
continua, 483 Impact testing
rotating, 397
frameworks, 469 Charpy, 534
varying thickness, 403
spring elements, 464 Izod, 534
Displacement, strain in terms of, 370-3
triangular element, 484 Indentation tests, 500-6
Dome, 56
Fixed beams, 211, 217-19, 223 Instability buckling, 264-91
Donath solution, 403
Flexible mounting, 87-8 Isochronous stress-strain curves, 589
Double integration method, 186, 211
Flexural rigidity, 441 Isometric stress-time curves, 589
Ductile materials, 340, 525
Fluctuating stress, 545
Ductility, 78, 496
Foil strain gauges, 312 J-contour integral, 530
Force, | et seq Joints, 58
Eccentric end loading, 148-51 Fracture
of slender columns, 274-6 brittle, 515 Kelvin-Voigt, 588
Elastic constants, 68 criteria, 348 Kevlar, 322
relationships between, 319 in compression, 497 Keyway, 359
Elastic limit, 67 in tension, 495 Kinetic energy, 77
Elastic modulus, 68 modes, 521
Elastic range, 78, 493 toughness, 519, 521 Lamé, 72, 386
Elasticity, 67 toughness tests, 522-5 Lamina, 322-9
Elongation, percentage, 496 Fracture mechanics, 349, 514-43 Laminate, 329-34
Empirical formulae for design, 279 for ductile materials, 525 modulus, 332
Endurance limit, 548 for fatigue, 554 strength of, 350
Energy, strain, (see strain energy) Frames Lateral strain, 70
theorems, 235-63 statically determinate, 12-18 Laterally loaded struts, 280-4
Equilibrium equations Frames, forces in Limit of proportionality, 67, 494
Cartesian co-ordinates, 367 method of sections, 16 Linear elastic fracture mechanics, 515
cylindrical co-ordinates, 368 resolution of joints, 12-14 Load, 1 et seq
in bending, 23, 135 Free body diagrams, 9 et seq Local bending stresses, 454
in thick cylinder, 383, 387 Fretting, 87 Local buckling, 285
in torsion, 105 Fully plastic moment, 415
rotating disk, 398, 404 Macaulay’s method, 193-202
Equilibrium, of forces, 1 et seg Gauge Material selection, 509
Equivalent sections, 142 clip or displacement, 523, 528 Materials, interaction of different, 82
Euler hyperbola, 269 electrical resistance strain, 292 Matrix
Euler theory for slender columns, 265-8 length, 493 algebra, 609-12
Experimental stress analysis, 311 Gear teeth, 360 coefficient, 15
Extensometer, 493 Geometrical discontinuities, 356-62 stiffness, 465
626 INDEX

solutions, 14, 323, 396, 465 Plasticity, 413-38 Second moment of area, 136, 599
transformation, 470 Plastics about parallel axes, 601
Maximum principal stress criterion, 348 creep of, 588 about perpendicular axes, 608
Maximum shear stress, 302, 306 creep design for, 590 Section modulus, 137
Maximum shear stress criterion, 341 creep rupture of, 593 Shafts, 7
Maxwell, 588 fatigue of, 567 non-uniform and composite, 108-112
Mean stress, 558-9 Plates, 5, 439-53 torsion of, 103
Mesh density, 488 concentrated load, 450 Shape functions, 474, 479
Middle third rule, 149 deflection equation, 445 Shear
Mild steel, 76, 496, 503, 555 moment-curvature relations, 439-44 centre, 164
Modulus pressure loading, 446 coupling, 57
bulk, 68 shear force and bending moment, 444 deflection of beams, 243
of elasticity, 68 slope and deflection, 445 force and bending moment, 24, 125 et
of rigidity, 68 various cases, 45] seq
of section, 137 with central hole, 448 force in bending, 24
shear, 68 Poisson’s ratio, 70 modulus, 68, 105
Mohr’s circle for moments of area, 606 Polypropylene, 76 strain, 64 et seq
Mohr’s fracture criterion, 348 Principal planes, 301 strain energy, 76, 235
Mohr’s strain circle, 310 Principal strains, 310 strain energy yield criterion, 341
Mohr’s stress circle, 298-306 Principal stress theory, 348 Shear flow, 112
Moment-area method, 220-32 Principle stresses, 301 Shear modulus, 105
Moment(s), 1 e¢ seq Product moment of area, 152, 602 Shear stress, 43 et seq
bending, 23 et seq Proof stress, 495 complementary, 45
bending of three, 227 Properties of areas, 598-608 in beams, 160-8, 380
of area, 598-608 of materials, 613 in open-sections, 164
of resistance, 136 Proportionality, limit of, 67,494 in torsion, 104-5
Multi-axial creep stresses, 580 Pseudo-elastic design, 590 maximum, 302, 306
Multi-axial cyclic stresses, 564 Shells, thin, 7, 52, 453-9
R-curves, 533
cylindrical, 54, 454-9
Radius of gyration, 269, 600
Nadai, 499 local bending stress in, 454-9
Rankine-Gordon formula, 279
Neutral axis, 134, 169 spherical, 52
Reciprocal theorem, 257-9
Neutral plane, 134, 439 under pressure, 52
Reduction of area, 496
Nylon, 76 Shock, stress due to, 95
Reinforced concrete beam, 144-7
Shrink fit, stresses due to, 391, 395,
Reinforced plastics, (see composites)
Offset method, 494 401
Relation between M, Q, and », 130
Open-coiled helical spring, 240, 243 Sign conventions, 23, 292
Relationship between F£, K, G and v, 319
Open sections, bending of, 164-8 Size effect, 521, 561
Relaxation of stress, 582
Optimization, 273 Skew bending, 151
Residual stress distribution, 431-6
Overspeeding, 434 Slenderness ratio, 269
after plastic bending or torsion, 431, 435
Overstrain, 500 Slit tube, 166
in cylinders and discs, 432-3
Slope during bending, 185-210
Resilience, 75
Parabolic formula for struts, 280 Sphere
Resultant forces, 3, 5
Plane strain, 73, 307, 519 thick-walled, 411
Resultant moments, 5
Plane stress, 73, 293, 519 thin-walled, 52
Rigidity modulus, 68
Plastic collapse of beams, 418-22 Spline, 359
Ring
Plastic deformation Spreadsheets, 4, 48, 156, 272, 317, 395,
bending moment in, 256
in beams, 413-8 459, 604
rotation of, 51
in rotating discs, 429-31 Springback, 432
Rockwell test, 505
in thick-walled cylinders, 425-9 Springs, 240-3
Rosette strain, computation, 314-9
in torsion, 423-5 Square shafts, 119
Rotating
Plastic hinge, 418 St. Venant, B., de, 352
disc, thin circular, 397-403
Plastic moment, 413 Stability of equilibrium, 264
ring, 51
Plastic overstrain, 500 Statical determinacy, 9
Rotor of varying thickness, 403-6
Plastic range, 76 Statically determinate
Rupture testing, 577, 593
Plastic stress concentration factor, 362 force systems, 1-42
Plastic torque, 423 Screw thread, 361 stress systems, 43, 63
Plastic zone, 526 Secant modulus, 592 Statically indeterminate beams, 211-34
 INDEX 627

Statically indeterminate stress system, 82- intensity factor, 518-34 Torque-twist curve, 499
101 normal, 43 Torsion, 28, 102-124
Statics, revision of, 1-4 plane, 73, 293 constant, 116
Step function, 194 principal, 301 effect of warping, 118
Stiffness matrix, 324, 327, 465, 484 raisers, 559 Nadai construction, 499
Straight-line formula, 279 relaxation, 582 of circular shaft, 103
Strain, 64 et seq stress-strain curves, 76, 493, 497 of circular tube, 102, 106
circle, 310 stress-strain relations, 65-81 of keyway, 359
cycling, 549 transformation, 292-306 of non-uniform shafts, 108-12
displacement equations, 370-3, 379-407 Structural sections, 7 of solid rectangular section, 119
due to shear, 64 Structural units, 6 of thin non-circular tube, 112-15
lateral, 70 determinacy criteria for, 9 of thin rectangular strip, 115-18
maximum shear, 310 Structures relationship, 105
normal, 65 forces in, 12-18 strain energy in, 236
plane, 73, 307 statically determinate, 12-18 testing, 498
principal, 310 Struts, 7, 46, 114 Toughness, 514
rate, 585 eccentrically loaded, 274-6 Transformation matrix, 470
rosette, 292, 314-9 empirical formulae, 279 Transition temperature, 535
thermal, 70, 74, 88-91 Euler theory, 265-8 Tresca yield criterion, 341
transformation, 292, 307-19 parabolic formula, 280 Triangle of forces, 6
volumetric, 66 Rankine-Gordon formula, 279 Tsai-Hill criterion, 350
Strain energy real, 268 Twist, 1 et seq
elastic, 75 straight-line formulae, 279 angle of, 104
from normal stress, 75, 235 with initial curvature, 276-9
from shear stress, 76, 235 with lateral loading, 280-4
in bending, 235-6 Suddenly-applied loads, 94 Universal beams, 137
in springs, 240 Superposition, method of, 31, 202, 216 Unreinforced plastics, 587
in torsion, 236 Support, types of, 7, 8, 22 Unstable fracture, 516
release rate, 516 Suspension bridge, 50 Unsymmetrical bending, 151, 204
solution for deflection, 249-57
yield criterion, 341 Tangent modulus, 270
Strain gauge, electrical resistance, 292 Tapered rod, extension of, 68 Variation of strength and ductility, 506-9
Strain-hardening, 495, 585 Temperature, effect of, 507, 561 Vickers hardness, 504
Stress, 43 et seq Temperature stress, 71 Virtual work, 245-8
bending, 125 et seg Tempering temperature, effect of, 507 Viscoelasticity, 78, 506, 573, 587
circle, 298-306 Tensile strength, 496 Voigt-Kelvin solid, 588
complementary shear, 45 Tensile stress, 43 et seq Volume modulus, 68
concentration, 351-63, 559 Tension, | et seq Volume strain, 66, 92
concentration factor, 357-63 Tension coefficients, 18-21 Volumetric strain energy, 343
due to rotation, 51 Testing von Mises yield criterion, 341
due to shear, 43 compression, 497
due to shock, 95 creep, 578
due to shrink fit, 391, 395, 401 fatigue, 546 Warping in torsion, 118
due to thermal effects, 70 hardness, 500-6 Wohler, 544
functions, 377 impact, 534-6 Work done, 235
general state of, 296, 307 of plastics, 588, 593 Work-hardening, 495
hydrostatic, 45 tension, 493
in beams, 125 et seq torsion, 498
in compound bars, 82-4 Thermal strain, 69, 88 Yield criteria, 340-8
in curved bars, 128, 168-73 Thermal stress, 70, 88 Yield envelope and locus, 345
in discs, 397-407 Thick cylindrical shell, 382-97 Yield point, 494
in rings, 51 Thin plates and shells, 439-62 Yield stress, 78, 340
in shafts, 104 Thin rotating disc, 397 Yield stress concentration factor, 362
in springs, 240-3 Thin tube, creep in, 581 Yielding, 78, 340
in tapered rod, 68 Tie, 7, 46 in cylinder, 389
in thick shells, 382-97 - Time hardening theory, 585 in disc, 401
in thin shells, 52 Torque, 30, 104 Young’s modulus of elasticity, 68
 . ~,
~\

—
(
ee
’

t ~

‘ _

c a

j : ee :

a
a » ,

\
i
 Mechanics of Engineering Materials is well-established as the definitive
textbook on the mechanics and strength of materials for students of :
engineering principles throughout their degree course. Assuming little or
no prior knowledge, the theory of the subject iis developed from first
' principles and all topics of stress and strain analysis are covered. right up.
to final year level. Mechanical properties such’ asttensile behaviour,
_ fatigue, creep, fracture and impact are discussed and more advanced
materialis also included, particularly on finite element analysis, fracture
_ mechanics and composite materials. . : .

This second edition has been brought fullyup-to-date in finawith today’ ‘Ss
courses. Incorporating new, two-colour illustrations throughout, the book.
reinforces student comprehension of the theory through numerous new -
worked examples and end- -Of-chapter problems involving real
_ engineering situations. An important new feature of this edition is the use |
and illustration of computer sprgauaneat throughout asa a powerful ~ <
problem- solving sie

° Provides the student witha comprehensive, logicaland oratlical

course companion for the duration of their engineering degree.
° Uses worked examples and set proionns: throughout to demonstrate
the application. of theory. ;
-@ Includes highlighted key definitions and‘equations.
- Provides end-of--chapter summaries to clarify key concepts and topics.
Incorporates the use of computer spreadsheets to solve: many
- Practical examples.
=.

Mechanics of Fiigikoating Materials i is an Fantiepiaichle course text for

undergraduate students of Mechanical Engineering, Engineering Science
and Civil Engineering. -It will also bea valuable: reference for those
studying BTEC and GNVa courses.

PP: Benham was formerly Head of the Béperiment ofNerviiautical

Engineering, R J Crawford is Professor of Engineering Materials and
~ Director of the School of Mechanical and Process Engineering, and C G
Armstrong is Reader in Mechanical Engiresring, allat The — $:
University of Belfast.

Cover photograph ByTim Fach,

ee Z

. < Fe RRO Ree oN e T0-582-251bu-

ae Le ae a SN 7 ekerore, a

Be nd
ee,
Z
pt
aes:
So
SS it
a NX
-* : Beets aa
Le

S < ea LONGMAN : %
: : K