Dynamic Models for Structural Plasticity

Springer-Verlag Berlin Heidelberg GmbH

W.J. Stronge and T.X. Yu

Dynamic Models for

Structural Plasticity
With 175 Figures

, Springer
William James Stronge, PhD, MS, BS
Department of Engineering, University of Cambridge, Trumpington Street,
Cambridge CB2 IPZ, UK

Tongxi Yu, PhD, ScD

Department of Mechanical Engineering, Hong Kong University of Science
and Technology, Clear Water Bay, Kowloon, Hong Kong

ISBN 978-3-540-76013-9 ISBN 978-1-4471-0397-4 (eBook)

DOI 10.1007/978-1-4471-0397-4

British Library Cataloguing in Publication Data

A catalogue record for this book is available from the British Library

Apart from any fair dealing for the purposes ofresearch or private study, orcriticism orreview, as
permitted under the Copyright, Designs and Patents Act 1988, this publication may only be
reproduced, stored or transmitted, in any form, or by any means, with the prior permission in writing
of the publishers, or in the case of reprographie reproduetion in aecordanee with the terms of
lieences issued by the Copyright Lieensing Agency. Enquiries coneeming reproduction outside
those terms should be sent to the publishers.

©Springer-Verlag Berlin Heidelberg 1993

Originally published by Springer-Verlag Berlin Heidelberg New York 1993
Correeted 2nd printing, 1995

The publisher makes no representation, express or implied, with regard to the accuracy of the
information contained in this book and cannot accept any legal responsibility or liability for any
errors or omissions that may be made.

Typeset by Fox Design, Bramley, Guildford, Surrey

69/3830-54321 Printed on acid-free paper

Preface

Our topic is irreversible or plastic deformation of structural elements

composed of relatively thin ductile materials. These deformations are
commonly used in sheet metal forming operations to produce
lightweight parts of any particular shape. In another context, this type
of plastic deformation is described as impact damage in the case of
structural components involved in collision. Here we are concerned
with mechanics of both static and dynamic deformation processes.
The purpose is to use typical material properties and structural
characteristics to calculate the deformation for certain types of load;
in particular to find the final deflection and shape of the deformed
structure and to illustrate how the development of this final shape
depends on the constitutive model used to represent the material
behavior. The major issue to be addressed is which structural and
constitutive properties are important for calculating response to either
static or brief but intense dynamic loads. Furthermore, how do the
results of various constitutive models compare with observed behavior.
Static analysis of plastic deformation is concerned with quantitative
assessment of structural safety and function; in practice either live
loads or foundation settlement can cause small sections of a structure
to experience relatively large plastic deformation. The margin between
the initial yielding load and the similar load for fully plastic collapse
is a measure of the ability of a structure to occasionally withstand
deformations that exceed the elastic limit. Static analysis of plastic
deformation is used also for sheet metal forming operations to find
the press forces and die ·shapes necessary to achieve a particular final
shape.
Dynamic plastic deformation of structures is applicable to blast or
impact loading; i.e. intense transient loads where the forces are
substantially more than the plastic collapse force but these forces are
applied for only a very brief period. This type of loading is used in
high-rate forming operations and it is the cause of structural damage
in collisions. Calculations of dynamic structural deformation are useful
for design of soft crash barriers that minimize damage to occupants
of a colliding vehicle; similar calculations are useful for investigating
structural integrity of the vehicle during all stages of a collision.
vi Preface

This book develops calculations of irreversible or plastic deformation

for an elementary structural component on the basis of simple
constitutive models. These models represent various aspects of material
behavior by means of equations developed from a theory; e.g. elasticity,
plasticity or viscoplasticity. The constitutive models are formulated
by combining the material representation with an· approximation for
the distribution of local deformation; this approximation primarily
represents flexure of structures that have a thickness which is small in
comparison with the length. This restriction applies since the
constitutive equations are all based on a deformation approximation
which assumes that plane sections remain plane - the Euler-Bernoulli
approximation. So although the effect of in-plane and shear forces are
considered, our attention is mainly on the static or dynamic transverse
deflection of slender ductile structural components. At this point it is
worth noting that in this structural application, the term constitutive
model generally implies both a mathematical representation of material
properties and the approximation that plane sections remain plane.
Here our approach is to present various constitutive models and
determine the effect of each model on structural response to generic
classes of loading. These models involve material properties that are
representative of different forms of real structural behavior and they
can be used to analyze complex arrangements of loads. The book
begins with elastic behavior of slender members in flexure and finds
limits on flexure that are determined by the onset of plasticity. This
occurs if curvature ~ at a section is larger than the elastic limit ~y so
that the section has an elastic core while near at least one surface the
strain exceeds the elastic limit. Since the stress-strain relation for
plastically deforming material is irreversible, the moment-curvature
relation for a plastically deforming section is also irreversible; i.e. if
~> ~y the moment-curvature relation for loading is different from
that for unloading. To determine structural response, often it is
convenient to assume that the stress resultants on a section are related
to the deformation by a rigid-perfectly plastic constitutive equation.
The rigid-perfectly plastic structural deformation model is the
fundamental constitutive equation considered in this monograph. For
material that is actually elastic-plastic with negligible strain hardening,
the rigid-perfectly plastic constitutive relation is the asymptotic limit
for stress resultants as curvature becomes indefinitely large; conse-
quently it is appropriate for large deformations of rate-independent
materials. We will investigate the accuracy and range of validity of
the rigid-perfectly plastic model by examining differences between
the deformations predicted by this elementary constitutive model and
more complete representations of real materials.
Of course the true test of accuracy for any theory is comparison
with experimental measurements. Some effects such as strain hardening
and strain-rate dependence are due to material behavior such as
dislocation multiplication and dislocation drag, while other effects such
as the influence of large deflections or complex loads are independent
of the material; thus validation of these theories for application to any
particular class of materials requires many tests over a range of load
Preface vii

intensities. Where experimental data are currently available, the

consequences of different constitutive and theoretical models developed
in this book are compared with these data. Some of the data used for
validation are new, but most have been compiled from the literature.
It will become obvious that there is not sufficient data available to
validate all of the different effects; some material effects have not
been tested and some tests lack the necessary documentation to permit
us to calculate the material parameters. Experimental work in dynamic
plasticity is especially difficult but nonetheless should be strongly
encouraged if there is to be continuing progress in adequately modeling
impact damage.
This book compares effects of constitutive modeling by determining
structural response of an elementary structure to a range of different
loads. The slender cantilever is the structural element used as a basis
of comparison. In the earliest development of this subject the cantilever
was a useful element for identifying characteristics of dynamic, as
opposed to static, behavior. For a heavy particle striking the tip of a
uniform cantilever, Parkes calculated the dynamic response of a rigid-
perfectly plastic cantilever and showed that the final deformed shape
depends on the ratio of the mass of the colliding particle to the mass
of the cantilever. The qualitative behavior predicted by this elementary
theory was verified by experiments that used both light and heavy
missiles striking the tip of cantilevers in a transverse direction. The
change in final shape of the cantilever as a function of mass ratio was
impressive and stimulated much subsequent work. While Parkes was
not the first to examine the effect of inertia on impact response of
plastically deforming structures, his analysis was influential because
it demonstrated that complex behavior could be represented by a rigid-
perfectly plastic constitutive model - this simple constitutive
approximation yielded results that appeared to be qualitatively correct.
Subsequent papers on dynamic plasticity have discussed the necessary
conditions where this approximation is valid; e.g. Symonds found that
elastic effects are negligible if the impact (or input) energy is large in
comparison with the limiting strain energy for elastic deformation of
the cantilever. This energy ratio R still provides a guide to applicability
although it must be augmented by other conditions in some cases.
Other investigations by Bodner and Symonds introduced modifications
such as strain-rate effects that brought the analysis and experimental
results for some materials into better quantitative agreement. So
historically, the analysis of a transversely loaded cantilever has been
important for developing understanding of phenomena associated with
dynamic plastic deformation.
The cantilever is a particularly attractive structural element for
comparison of constitutive models because it can undergo large, mainly
flexural deformations wherein the effects of resultant forces and bending
moment at a section are only coupled through the associated plastic
flow law. In the case of the rigid-perfectly plastic constitutive model,
pure bending results in regions of plastic deformation that are localized
at discrete plastic hinges. For any analysis this substantially simplifies
the admissible fields of plastic deformation. The effect of large
viii Preface

deflections on some other structures such as clamped beams or plates

is necessarily linked with stretching that soon dominates the
development of plastic deformation. For a cantilever, however, the
effects of large deflection are introduced only through the equations
of motion. Consequently, any coupling between flexure and stretching
or flexure and shear is obviously a consequence of the plastic flow
law and not of large deflections. The cantilever also has advantages
for any experimental work; it exhibits large, easily measured deflections
for rather modest deformations. Moreover, the specified boundary
conditions are easy to achieve in practice because the degree of
clamping affects neither the stress resultants nor the inertia.
While the fundamental rigid-perfectly plastic structural response
approximation can represent many aspects of deformation for rate-
independent elastic-plastic materials, there are other materials and
deformation conditions where this basic constitutive model is not
adequate. These conditions will be exposed, then methods are presented
to analyze deformation of structural elements where second-order effects
significantly modify the response. This book emphasizes qualitative
rather than quantitative differences that become obvious in comparisons
of deformations calculated with various constitutive models. In almost
all cases the consideration of second-order effects increases the
complexity of the analysis; nonetheless, our aim is to reduce these
complexities and provide the simplest formulation which contains the
essence of a material and deformation effect. There are many numerical
analysis methods which can provide more precise detail on development
of deformation at every point for any particular structure and load.
The codes that implement these methods are necessary to analyze
deformation fields that depend on more than one spatial variable; they
are not suited, however, to identifying common patterns of structural
response. In the wealth of analytical detail provided by these programs
one may lose sight of general phenomena or not identify the most
important parameters for representing a particular class of materials,
loads or structures - we lose sight of the forest because of the trees.
Hence the approach adopted here is to employ analytical methods to
obtain both static and dynamic behavior of a simple structural element
- the cantilever. Using these methods the results of a variety of
constitutive models are examined by comparing the deformations
developed in cantilevers subjected to similar loads. This approach is
useful for segregating effects of constitutive modeling in part because
deformation of the cantilever involves only a single spatial coordinate.
Also, it explicitly identifies the essential parameters for representing
an effect from among the many possibilities presented by dimensional
analysis.
The organization of this monograph has been designed to assist in
learning to recognize patterns of structural behavior associated with
effects of different material properties. Thus we begin with development
of a range of constitutive models using a strength of materials type
approximation for the distribution of deformation at any section. These
models are employed to analyze a series of static loads that result in
increasingly complex deformation fields. Most of the book is concerned,
Preface ix

however, with dynamic response to brief but intense loads that result
from impact. In this case the final displacement field is the culmination
of a deformation process in which the shape of the structure is
continually evolving.
The framework for this monograph has been used for teaching at
the graduate level and many of the details were developed for courses
on Dynamic Plasticity and Impact Response of Structures taught by
the authors in Cambridge and Peking Universities. The book is also
aimed at research workers in nonlinear structural dynamics who can
employ the analytical methods in further developments. For practicing
engineers who are concerned with design of high-rate forming processes,
impact damage prediction or design and evaluation of structurally
crashworthy vehicles, we offer a compilation of data on different
materials that will assist in applying these methods of structural analysis.
This will be handy in using the methods presented here to perform
any particular calculation.
Here we wish to acknowledge our gratitude to our friend, Professor
Bill Johnson, who brought us together and raised many of the questions
addressed in this book. His enthusiasm and insight into problems of
metal forming have been sources of inspiration. Tongxi Yu is grateful
to The Royal Society of London who provided a visiting fellowship
that permitted him to devote full time to this project in the initial
stage of writing. Also, we greatly appreciate the support and
encouragement of our students, former students and colleagues who
have critically read and painstakingly commented on various sections
of the book; in particular we thank Norman Jones, Dongquan Liu,
Steve Reid, Victor Shim and Tieguang Zhang. Their criticisms have
helped to clarify some points and catch some errors. Nevertheless, the
authors are responsible for any remaining blunders; we will appreciate
being informed of any errors that the reader detects. Finally we wish
to thank our wives, Katerina and Shiying, who did most of the typing
and a lot of preliminary editing; their careful attention smoothed some
of our rough edges.

6 March 1993 W.J. Stronge

Cambridge, U.K. T.X. Yu
Contents

List of Symbols ........................................ xvii

1 Elastoplastic and Viscoplastic Constitutive

Relations . ....................................... . 1
1.1 Stress Resultants and Generalized Stress -
Deformations and Generalized Strain. . . . . . . . . . . . . . . . . 1
1.2 Pure Bending of Rate-Independent Bar. . . . . . . . . . . . . . . 2
1.2.1 Kinematics of Deformation .................. 2
1.2.2 Elastic Constitutive Equation. . . . . . . . . . . . . . . . . 4
1.2.3 Stress Resultants (Axial Force and Bending
Moment) .................................. 4
1.2.4 Elastic-Plastic Constitutive Equations. . . . . . . . . 5
1.2.5 Elastic-Power Law Hardening Constitutive
Equations. '. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.3 Pure Bending of Rate-Dependent Bar ................ 9
1.3.1 Strain-Rate Dependent Constitutive Equations. . 9
1.4 Interaction Yield Functions and Associated Plastic
Flow .......................................... 12
1.4.1 Elastic Limit for Bending and Tension. . . . . . . . . 12
1.4.2 Fully Plastic Limit Surface for Bending and
Tension in Elastic-Perfectly Plastic Bar. . . . . . . . 12
1.4.3 Yield and Fully Plastic Stress Condition. . . . . . . 15
1.4.4 Associated Flow Rule for Plastic Deformations. 16
1.4.5 Separated Yield Functions and Separated Plastic
Flow.. ... . .. . .... . . . . . .. . .. . . . .. ........ . . 18
1.5 Interaction Yield Surfaces Including Shear. . . . . . . . . . . . 19
l.5.1 Tension, Shear and Bending in Rectangular
Cross-Section . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
1.5.2 Tension, Torsion and Bending in Circular
Cross-Section . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
1.6 Elastic Springback . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
1.6.1 Pure Bending .............................. 23
1.6.2 Bending and Tension. . . . . . . . . . . . . . . . . . . . . . . . 25
References ............................................. 27
xii Contents

2 Principles of Mechanics . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.1 Kinematics. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.1.1 Inertia Properties of Cross-Section . . . . . . . . . . . . 31
2.2 Balance of Forces ................................. 31
2.2.1 Stress Resultants and Generalized Stresses . . . . . 31
2.2.2 Equations of Motion ., . . . . . . . . . . . . . . . . . . . . . . 32
2.3 Principle of Virtual Velocity. . . . . . . . . . . . . . . . . . . . . . . . 34
2.3.1 Rate of Change for Kinetic Energy of System .. 35
2.3.2 Rate of Change for Kinetic Energy of
Kinematically Admissible Velocity Field W;C . . . 35
2.3.3 Extremal Principles for Complete Solution. . . . . 36
2.4 Bounds For Rigid-Perfectly Plastic Solids and
Structures ................................. 37
2.4.1 Upper and Lower Bounds on Static Collapse
Force.. .. . . . .. . . .. .. . . . . . . . .. .. . ...... .. . . 37
2.4.2 Lower Bound on Dynamic Response Period . . . . 38
2.4.3 Upper Bound on Dynamic Response Period. . . . 39
2.4.4 Lower Bound on Final Displacement . . . . . . . . . . 40
2.4.5 Upper Bound on Final Displacement. . . . . . . . . . 41
2.5 Dynamic Modes Of Deformation. . . . . . . . . . . . . . . . . . . . . 44
2.5.1 Modal Solutions. . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
2.5.2 Properties of Modes. . . . . . . . . . . . . . . . . . . . . . . . . 44
2.5.3 Mode Approximations for Structural Response
to Impulsive Loading. . . . . . . . . . . . . . . . . . . . . . . . 47
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48

3 Static Deflection. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
3.1 Small Elastic-Plastic Deflections .................... 51
3.1.1 Elastic Deflections. . . . . . . . . . . . . . . . . . . . . . . . . . 52
3.1.2 Deflection of Elastic-Perfectly Plastic
Cantilever . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
3.1.3 Deflection of Elastic-Linear Strain Hardening
Cantilever . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
3.1.4 Residual Deflection After Elastic Unloading. . . . 58
3.1.5 Elastoplastic Beam-Columns . . . . . . . . . . . . . . . . . 61
3.2 Large Elastic-Plastic Deflections ............. . . . . . . . 63
3.2.1 Elastica: Large Elastic Deflection. . . . . . . . . . . . . 63
3.2.2 Plastica: Large Plastic Deflection. . . . . . . . . . . . . 67
References ....... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72

4 Dynamic Rigid-Plastic Response . . . . . . . . . . . . . . . . . . . 73

4.1 Step Loading. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
4.1.1 Static and Dynamic Loadings. . . . . . . . . . . . . . . . . 73
4.1.2 Moderate Dynamic Load (~ < F < 3Fc) ....... 75
4.1.3 Intense Dynamic Load (F> 3Fc) ............. 78
4.2 Rectangular Pulse Loading. . . . . . . . . . . . . . . . . . . . . . . . . . 80
4.2.1 Three Phases in Response of Cantilever. . . . . . . . 80
4.2.2 Deformed Shape. . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
4.2.3 Energy Dissipation. . . . . . . . . . . . . . . . . . . . . . . . . . 85
4.2.4 Synopsis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
Contents xiii

4.3 Features of Travelling Hinges. . . . . . . . . . . . . . . . . . . . . . . 88

4.4 General Pulse Loading .............. . . . . . . . . . . . . . . . 90
4.4.1 General Considerations. . . . . . . . . . . . . . . . . . . . . . 90
4.4.2 Example: Linearly Decaying Pulse. . . . . . . . . . . . 93
4.4.3 Equivalent Replacement of Arbitrary Pulse. . . . . 94
4.5 Impact on Cantilever. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
4.5.1 Problem and Assumptions ................... 97
4.5.2 Changing Pattern of Deformation. . . . . . . . . . . . . 98
4.5.3 Acceleration, Force and Bending Moment. . . . . . 101
4.5.4 Deformed Shape. . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
4.5.5 Energy Dissipation. . . . . . . . . . . . . . . . . . . . . . . . . . 105
4.5.6 Modal Approximation. . . . . . . . . . . . . . . . . . . . . . . 106
References ..... -. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110

5 Second-Order Effects on Dynamic Response. . . . . . . . . 111

5.1 Strain-Rate Effect. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
5.1.1 Impulsive Load on Viscoplastic Cantilever. . . . . III
5.1.2 Elementary Estimates of the Effect of Strain-Rate
on Final Deformation ....................... 117
5.2 Strain Hardening (Strain Softening) Effect . . . . . . . . . . . . 120
5.2.1 Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
5.2.2 Elementary Effect of Strain Hardening on Final
Deformation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
5.2.3 Dynamic Analysis of Strain Hardening and
Strain Softening Cantilevers. . . . . . . . . . . . . . . . . . 123
5.3 Effects of Transverse Shear and Rotary Inertia.. . ... ... 132
5.3.1 Interface Conditions for Concentrated Mass . . . . 132
5.3.2 Shear Deformation Adjacent to Colliding
Particle. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
5.3.3 Shear and Rotary Inertia of Finite Size
Colliding Missile. . . . . . . . . . . . . . . . . . . . . . . . . . . 139
5.3.4 Shear Rupture due to Impact. . . . . . . . . . . . . . . . . 144
5.3.5 Measurements of Energy for Shear Rupture .... 146
5.4 Effect of Large Deflection. . . . . . . . . . . . . . . . . . . . . . . . . . 149
5.4.1 General Considerations. . . . . . . . . . . . . . . . . . . . . . 149
5.4.2 Large Deflection of Impulsively Loaded
Cantilever . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150
5.4.3 Methods of Approximating Large Deflection
Effects. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
5.4.4 Effect of Centripetal Acceleration on Bending
Moment Distribution. . . . . . . . . . . . . . . . . . . . . . . . 155
5.5 Effect of Elastic Deformation ............. . . . . . . . . . . 157
5.5.1 General Considerations. . . . .. . . . .. . . . . . . . . . . . 157
5.5.2 Mass-Spring Finite Difference Structural
Model (MS-FD) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158
5.5.3 Timoshenko Beam Finite Element Structural
Model (TB-FE) ..................... , . . . . . . . 163
5.5.4 Dynamic Deformation of Elastic-Plastic
Cantilever from Impact. . . . . . . . . . . . . . . . . . . . . . 164
xi y Contents

5.5.5
Effect of Elastic Deformation at Root of
Cantilever . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
5.5.6 Remarks................................... 181
5.6 Accuracy of Rigid-Plastic Analyses. . . . . . . . . . . . . . . . . . 182
5.6.1 Accuracy of Rigid-Plastic Analysis Estimated
by Single DoF System. . . . . . . . . . . . . . . . . . . . . . . 182
5.6.2 Convergence to Dynamic Plastic Mode Studied
by Two DoF System. . . . . . . . . . . . . . . . . . . . . . . . 184
5.6.3 Remarks. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187
References ..................... . . . . . . . . . . . . . . . . . . . . . . . . 188

6 More Complex Configurations . . . . . . . . . . . . . . . . . . . . . 191

6.1 General Considerations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192
6.1.1 Extremal Properties of Yield Function at Plastic
Hinge. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192
6.1.2 Differentiability of Arbitrary Functions at
Plastic Hinge. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193
6.1.3 Differentiability of Kinematic Variables at
Plastic Hinge. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194
6.1.4 Differentiability of Generalized Stresses at
Plastic Hinge. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196
6.1.5 Differentiability of Yield Function at Plastic
Hinge. . .. ... ..... . .. .... . . . .. ......... .. . . 197
6.2 Straight Cantilevers with Smoothly Varying
Cross-Sections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199
6.2.1 Yield Function and Conditions at Plastic
Hinge.. .. .... ..... .. . . .. . . . .. ..... .... .. . . 199
6.2.2 Suddenly Applied Steady Force at Tip of
Tapered Cantilever: An Example. . . . . . . . . . . . . . 200
6.3 Oblique Impact on Straight Cantilever. . . . . . . . . . . . . . . . 204
6.3.1 Problem and Assumptions ................... 204
6.3.2 Formulation Based on Single Hinge
Mechanism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205
6.3.3 Solution Based on Single-Hinge Mechanism. . . . 212
6.4 Circular Arc Cantilever Subjected to In-plane Step
Force .......................................... 215
6.4.1 Engineering Background and Assumptions ..... 215
6.4.2 Radial Force at Tip. . . . . . . . . . . . . . . . . . . . . . . . . 217
6.4.3 Tangential Force at Tip. . . . . . . . . . . . . . . . . . . . . . 221
6.4.4 Discussion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
6.5 Circular Arc Cantilever Subjected to In-plane Impact. . . 224
6.5.1 Rigid-Plastic Formulation ................... 225
6.5.2 Discussion of Solution ...................... 228
6.5.3 Modal Approximation. . . . . . . . . . . . . . . . . . . . . .. 231
6.6 Circular Arc Cantilever Subjected to Out-of-Plane
Step Force. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232
6.6.1 Equations of Motion .. . . . . . . . . . . . . . . . . . . . . .. 233
6.6.2 Solution. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235
6.6.3 Discussion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 240
6.7 Stepped or Bent Cantilever Subjected to Step Force. . . . 241
Contents xv

6.7.1 General Considerations ...................... . 241

6.7.2 Stepped Cantilever .......................... . 243
6.7.3 Bent Cantilever ............................ . 246
6.704 Discussion ................................ . 249
6.8 Cantilever with an Initial Crack .................... . 250
6.8.1 General Considerations ..................... . 250
6.8.2 Impact on Cantilever with an Initial Crack .... . 251
6.8.3 Crack Stability After Impact ................. . 254
6.804 Numerical Example and Discussion ........... . 256
References ............................................. . 257

7 Impact Experiments .............................. . 259

7.1 Methods of Applying Dynamic Loads ............... . 259
7.1.1 Missile Impact ............................. . 259
7.1.2 Explosive Loading .......................... . 261
7.1.3 Magnetomotive Loading ..................... . 261
7.2 Travelling Hinges - Fiction or Fact? ............... . 262
7.3 Elastic Effects on Plastic Deformation ............... . 265
7A Strain Hardening and Strain-Rate Effects ............. . 266
704.1 Mode Approximations ....................... . 266
704.2 Transient Analysis Including Rate Effects ..... . 270
7.5 Dynamic Rupture .................................. . 270
7.5.1 Location and Mechanism of Rupture ......... . 270
7.5.2 Measurements of Generalized Strain at
Rupture .................................. . 272
References 274

Index . ............................................ ' .. . 277

List of Symbols

A cross-sectional area
a radius of circular cross-section; half side-length
of square block (Sect. 5.3)
b width of rectangular cross-section
C elastic spring coefficient
c wall thickness of thin-walled section
C characteristic flaw size (Sect. 6.8)
D total energy dissipation due to plastic
deformation
energy dissipation due to bending
energy dissipation due to shear (Sect. 5.3)
energy dissipation rate due to plastic deformation
=DIMp, nondimensional energy dissipation
=D,,/Mp, nondimensional energy dissipation due to
bending
=DIMp, nondimensional energy dissipation due to shear
(Sect. 5.3)
E Young's modulus
input energy
tangent modulus for plastic flow
= Ei,/Mp, nondimensional input energy
= Kr:/Mp, impact energy ratio
nondimensional rupture energy (Sect. 5.3)
unit vectors fixed in undeformed configuration
F transverse force
static plastic collapse force
components of external traction
= FIFe = FUMp , nondimensional force
mass of colliding particle; elastic shear modulus
xviii List of Symbols

g distributed force per unit length

h depth of doubly symmetric cross-section in plane of
deformation
I second moment of area about transverse axis through
centroid
lox. loy. loz second moment of area about X-, Y- and Z-axis,
respectively
first and second moments of mass about plastic hinge
(Sect. 5.4)
1 Lee's functional (Sect. 2.5, 5.6); a functional related to
Tamuzh's principle (Sect. 6.3); 1-integral (Sect. 6.7)
=lox, polar moment of inertia per unit length
K kinetic energy
= GV~'f2, impact energy
= KL, non dimensional curvature; Boltzmann's constant
ky = l(yL, nondimensional curvature at yield
k = I(ll(y, curvature ratio
= I(fll(y, curvature ratio after unloading
= 1(.ll(y, largest curvature ratio before unloading
L length of beam or slender bar
effective length of plastic hinge (Sects 5.1, 5.2)
bending moment
fully plastic bending moment
elastic limit moment
m =MIMp, fully plastic bending moment ratio (Chaps 4,5)
= MIMp, fully plastic bending moment ratio (Chaps 1-
3,6)
m =MIM y, elastic bending moment ratio
m. =MJM y, largest bending moment ratio before unloading
N axial force
fully plastic axial force
elastic limit axial force
n = NLIMp, nondimensional axial force; number of
segments (Sect. 5.5)
ii = NINp, fully plastic axial force ratio
n = NIN y = ii, elastic axial force ratio
n. = N.INy, largest axial force ratio before unloading
P impulse of a loading pulse
P eff effecti ve total impulse
Pf total impulse
List of Symbols xix

P = PUMpTo, nondimensional impulse of a loading pulse

PI = PPMpTo, nondimensional total impulse
Po = YVo, nondimensional initial impulse
Q shear force
Qj generalized stresses; nodal shear force (Sect. 5.5)
Qp fully plastic shear force
Qy elastic limit shear force
q = QUMp , nondimensional shear force
qp = QpUMp, nondimensional fully plastic shear force
qj generalized strain-rate vector
q = Q/Qp, fully plastic shear force ratio
q = Q/Qy, elastic shear force ratio
R energy ratio Ein IUJI1 ax (Sects 5.5, 5.6); fracture toughness
(Sect. 5.3); initial radius of curved bar (Sects 6.3, 6.5)
r strain-rate index; radial coordinate
rg radius of gyration of colliding block (Sect. 5.3)
S arc length along axis
s = SIL, nondimensional arc length along axis
T torque
Tp fully plastic torque
Ty elastic limit torque
To characteristic time, L~ pUMp
11 fundamental period of elastic vibration
f tearing modulus (Sect. 6.7)
time
td duration of a pulse
t = TITp, fully plastic torque ratio
t = TIT y, elastic torque ratio
U axial displacement
Ue elastic deformation energy
u~ax maximum elastic flexural deformation energy for structure
V transverse velocity of colliding mass; volume (Chap 2)
Vo initial transverse velocity of colliding mass
Vx axial velocity of colliding mass in oblique impact (Sect.
6.3)
Vz transverse velocity of colliding mass in oblique impact
(Sect. 6.3)
v = VToiL, nondimensional transverse velocity
Vo = VoToiL, nondimensional initial transverse velocity of
colliding mass
xx List of Symbols

= VxTelL, nondimensional axial velocity of colliding mass

= VzTelL, nondimensional transverse velocity of colliding
mass
W transverse deflection of centroid
w = WI L, nondimensional deflection
x axial coordinate in undeformed configuration
x =XI L, nondimensional coordinate
y yield stress; transverse coordinate normal to plane of
loading (and symmetry) in undeformed configuration
z transverse coordinate in plane of loading (and symmetry)
measured from centroid in undeformed configuration
z transverse distance from neutral surface
z distance between centroid and neutral surface

a = £1£, coefficient of strain hardening

coefficient of strain hardening for bending moment (Sect.
5.2)
amUMp (Sect. 5.2)
f3 material or section constant (Chap. 1); coefficient showing
the variation in beam width (Sect. 6.2); angle subtended
by hinge position in circular cantilever (Sects 6.4, 6.5);
bend angle of a bent cantilever (Sect. 6.6)
area where surface tractions applied
= GlpL, mass ratio
transverse shear strain (shear angle)
transverse deflection at the tip
relative transverse displacement due to shearing (Sect.
5.3)
= .1IL, nondimensional transverse deflection at the tip
= .1/L, nondimensional relative transverse displacement
due to shear
e axial strain
material constant
rate dependent material constant
true tensile strain at fracture (Sect. 5.3)
strain
axial yield strain
nondimensional distance between centroid and neutral
List of Symbols xxi

surface (Chap. 1); velocity ratio (Sect. 5.6); ratio 4Uh

(Sect. 6.3); coefficient of imperfection (Sect. 6.7)
1) constant defined by (5.3) in (Sect. 5.1); coordinate with
origin at hinge H (Sect. 5.4); ratio T/Mp (Sect. 6.5)
= L 21L 1, ratio of lengths for stepped cantilever/Sect. 6.6)
= M p21Mp1 ' ratio of fully plastic bending moments for
stepped cantilever
= P2/P 1, ratio of densities per unit length for stepped
cantilever (Sect. 6.6)
=dO/dX
value of 1') at yield
rotation angle, inclination
rotation rate related to bending
rotation of section about centroidal axis
curvature
let residual curvature after unloading
K"y maximum elastic curvature
A coordinate of hinge location H
It = AIL, nondimensional coordinate of hinge
Jl flexibility parameter (Sect, 3.1); ratio of peak force to
yield force (Sect. 5.6); ratio tH/mH (Sect. 6.5)
nondimensional elastic part of half depth (Chap. 1);
coordinate with origin at hinge H (Sect. 5.4)
=pyA, mass per unit length of cantilever
radius of curvature of deformed centroid
density of material
normal stress on cross-section
final residual stress
stress
ultimate stress (Sect. 5.3)
=Y, uniaxial yield stress
=tlTo, nondimensional time; shear stress on cross-section
yield stress in shear
nondimensional group ae o (Sect. 5.2); angle (Sects 6.4,
6.5)
shape factor vector
shape factor for bending
shape factor for axial force
shape factor for shear
shape factor for torsion
xxii List of Symbols

components of mode-shape
yield function
elastic limit surface
fully plastic limit surface
relative rotation angle (Sect. 5.5); angular coordinate for
circular cantilever (Sects 6.3, 6.5)
lfIy relative rotation angle at yield (Sect. 5.5)
il angular velocity
nondimensional angular velocity, ilTo

Superscripts

c kinematically admissible
d dynamically admissible
e elastic
p plastic
* modal
differentiation with respect to coordinate x
differentiation with respect to time variable
+ positive side
negative side

Subscripts

A tip of cantilever
B root of cantilever
C section where cross-section varies or bends
f final
H plastic hinge
in input
m bending
q shearing
torsion
y yield
o initial
phase 1
2 phase 2
Chapter 1

Elastoplastic and Viscoplastic Constitutive

Relations

1.1 Stress Resultants and Generalized Stress -

Deformations and Generalized Strain

In Chapter I we develop constitutive modeling for slender structural elements.

Together with equations of motion, the constitutive equations will be used to
relate nonlinear structural deformations to the applied forces or loads that cause
these deformations. Structural deformations depend on load, structural configuration
and properties of the materials of which the structure is composed.
Material properties are obtained from tests that produce a uniform state of
stress, temperature and rate of deformation in a specimen; e.g. a uniaxial tensile
test. For a homogeneous material, the results of these tests are approximated by
constitutive relations; i.e. mathematical expressions that represent the material
response to stress. For an arbitrary material particle these expressions relate the
current state of stress a to the current state of strain £, the strain ratei:. and the
history of strain. The constitutive relations describe different types of material
behavior such as elastic, elastic-plastic or viscoplastic; for any particular material,
the appropriate constitutive relation can depend on the current strain, history of
strain and temperature of the material particle.
For slender structural members such as bars, beams or plates the normal stress
component through the depth is negligible so it is convenient to express constitutive
relations in terms of generalized measures of stress and deformation on a cross-
section. These stress and deformation variables are related by means of structural
properties for a typical cross-section; e.g. the bending or elongation stiffnesses.
For a bar or beam with axial coordinate X and cross-sectional area A, the
components of stress aij on any cross-section give the following stress resultants
for axial force N, shear force Q, axial torque T and bending moment M:

Q = Lerxz dA
M = - L Zer xx dA
2 Elastoplastic and Viscoplastic Constitutive Relations

where Y, Z are transverse coordinates measured from the axis through the centroid
of every section. These generalized stresses are counterparts of stress; for beams
or plates it is convenient to develop constitutive equations relating generalized
stresses {N, Q, T, M} to deformation variables termed generalized strains.
For slender members, out-of-plane warping of cross-sections is negligible since
the depth is small in comparison with the length; consequently, plane sections
remain plane. This kinematic constraint means that deformations are completely
described by stretching E and shear y, axial twist t'J and curvature IC of the
centroidal axis for the undeformed section. Hence relations between stress, strain
and strain-rate translate to relations between generalized stresses {N, Q, T, M},
generalized strains {E, y, t'J,IC} and the strain-rates {e, y, b,K-} where differentiation
with respect to time t is denoted by (.) == d( )/dt. A convenient notation defines a
vector of generalized stresses Qa == {QI' Q2' Q3' Q4} = {N, Q, T, M} and corres-
ponding vectors of conjugate strains qa == {ql' q2' q3' q4} = {E,y,t'J,IC} and strain-
rates iJa == {iJI' iJ2' iJ3' iJ4} = {e, y, b,K-}. The conjugate strains qa for a set of
generalized stresses Qa give the rate of energy dissipation D in a member of
length L as the sum of contributions from the components of generalized stress
that are incorporated in the yield criterion
. rL
D = JoQaiJa dX

Materials that exhibit reversible deformations during loading and unloading

are termed elastic. In the range of small strain, most structural metals are represented
by a linear elastic relation (Hooke's law). The range of strain for elastic behavior
is limited by yielding and the onset of plastic deformation. Plastic deformation is
irreversible and it can be either rate-independent or rate-dependent depending on
the material, temperature and state of strain. The choice of an appropriate
constitutive model that represents these effects often depends more on the
application of the results and less on a precise material description. In general,
analytical solutions for structural deformation due to impact are attained only for
the simplest material descriptions, so this chapter focuses on elementary constitutive
models.

1.2 Pure Bending of Rate-Independent Bar

1.2.1 Kinematics of Deformation

Finding the relationship between stress resultants (or generalized stresses) and
deformations at any section of a uniform beam is facilitated by starting from a
description of the deformations and then deriving the corresponding stress resultants
for any particular constitutive equation. Here these relationships are found for a
straight and uniform beam that has cross-sections with a common plane of
symmetry. The ends of the beam are loaded by equal but opposed couples that
act in directions perpendicular to the plane of symmetry. Thus the beam bends in
the plane of symmetry and does not twist. Deformation in response to these couples
is known as pure bending; in a uniform segment the axial fibers that were initially
1.2 Pure Bending of Rate-Independent Bar 3

straight are deformed to circular arcs. Moreover, cross-sections perpendicular to

the axis of the undeformed beam are deformed into plane surfaces perpendicular
to the axis of the deformed beam; i.e. plane sections remain plane. This kinematic
condition is exact for pure bending if cross-sections have in-plane deformations
that are negligible; this condition can be shown to follow from considerations of
symmetry and compatibility of displacements for adjacent elements. The result
follows also from the theory of elasticity (e.g. Timoshenko and Goodier [1971],
p 248).
Deformations are defined relative to a set of coordinates X, Z with fixed
orientation in relation to the initial undeformed configuration. The X axis is the
line passing through the centroid of each cross-section of the beam while the Z
axis is transverse to the axis of the beam and in the plane of symmetry. In
addition, an axial coordinate S is measured along the centroid in the deformed
configuration. Axial and transverse displacements are denoted by U(X) and W(X),
respectively. If forces are applied to the beam it deforms as the load increases;
the axis becomes inclined at an angle 8 m (X) that is given by 8 m (X) =
arctan (dW(X)/dX)""dWldX where the approximation is accurate if 18m In l«1.
pure bending the cross-sections rotate with the axis, so every longitudinal fiber is
bent into a circular arc. If the centroid has a radius of curvature Pc(X), the arc
length dS = Pc d8m where Pc depends on the rate of change of inclination along
the beam. Since dS = (1 + (dWldX)2)1I 2 dX, the curvature IC(X) of the centroid is
given by
-I d 2 WI dX 2 d2 W
(Ll)
/C = Pc = [1+(dWldX)2Y'2 "" dX2
Once more, the approximation applies to small inclinations 18m « Notice that I 1.
with the convention shown in Fig. 1.1, positive curvature is associated with
'hogging' .
During pure bending, longitudinal fibers are stretched but not sheared. Near
the convex surface of the beam longitudinal fibers are elongated while near the
concave surface they are compressed. However, somewhere in the cross-section
there is a layer with zero elongation; this is known as the neutral surface. In
planar deformation the neutral surface is represented by the neutral axis. In order

Fig. 1.1 Pure bending of symmetric prismatic bar by momentM acting about axis normal to plane of symmetry.
The transverse deflection of the centroid W(S) z W(X) is measured from the initial undeformed configuration.
4 Elastoplastic and Viscoplastic Constitutive Relations

neutral surface -x

Fig. 1.2 Normal stress distribution on cross-section for elastic flexure.

to satisfy the condition that plane sections remain plane the strain (elongation per
unit length) normal to the cross-section must vary in proportion to the distance
from the neutral axis. Let Z be the distance from the neutral axis (positive in the
same sense as Z). Then the longitudinal strain is e == aU/ax = - ZI(; in any section
the strain has a linear variation with depth.

1.2.2 Elastic Constitutive Equation

For elastic materials Hooke's law provides a unique relation between uniaxial
stress (J and longitudinal strain e for deformation from an unstressed initial
state:
(J = Ee (1.2)

where E is Young's modulus. This elastic modulus is just the slope of the stress-
strain curve in the region of small strain. For pure bending the normal stress
increases linearly with distance from the neutral surface; i.e. (J = - EZI( as shown
in Fig. 1.2.

1.2.3 Stress Resultants (Axial Force and Bending Moment)

The resultant axial force N is the integrated effect of the normal stresses acting
on a cross-section. For pure bending the resultant force vanishes. Thus for cross-
sections with area A, the axial force obtained from the stress distribution for pure
bending is given by

N = L (J dA L
= - EI( Z dA = 0 0.3)

The second integral in this expression is known as the first moment of area about
the neutral surface. Vanishing of the first moment of area implies that the neutral
axis is coincident with the centroid of the section since the distance i from the
neutral axis to the centroid is defined as i = A-I Z dA. 1 f
The bending moment M(X) at a section is the fir£t moment of normal stress on
the cross-section about the centroid,

lPor a beam with substantial curvature in the unstressed state, the neutral axis for pure bending is
not coincident with the centroid of cross-sections. Bending of a curved beam results in plane sections
that remain plane, but the strain distribution is no longer linear since longitudinal fibers in a differential
element have undeformed length that varies across the section.
1.2 Pure Bending of Rate-Independent Bar 5

M == - LZadA = +E1o/( (1.4)

t
where 10 == Z2 dA is the second moment of area about the transverse axis through
the centroid (frequently this is termed the moment of inertia). From equilibrium
we obtain that the moment M is equal in magnitude to the couple at the end of
the beam. Thus Eq. (1.4) describes the moment-curvature relation for pure bending
of elastic beams. Although this expression is exact only for pure bending, the
expression is still useful for a beam that carries both shear force and a bending
moment since the deformation due to shear is usually negligible. This is the case
if the beam is slender, LI h » 1 ; in this case, only a negligibly small part of the
strain energy of deformation is caused by shear.
The previous expressions are valid for bending of elastic bars; i.e. if throughout
the section the strains are less than the yield strain €y. This limiting strain is
related to the uniaxial yield stress Y through €y = Y / E. If the neutral surface is
also a plane of symmetry for cross-sections, the largest strains are at the top and
bottom surface - a distance hl2 from neutral axis. Thus the largest elastic curvature
is /(y = ±2€y / h = ±2Y / Eh, while the corresponding bending moment at yield My
is given by
(1.5)

If the bending moment is larger than this elastic limit, the fibers of the cross-
section furthest from the neutral surface are strained beyond the elastic limit €y.

1.2.4 Elastic-Plastic Constitutive Equations

Plastic deformation is related to dislocation motion on slip planes in crystalline
materials. In both polycrystalline materials and crystals with second phase particles,
dislocation motion only occurs if the shear stress is sufficiently large on slip
planes. The yield strain €y is the manifestation of this limit; in a polycrystalline
solid it is the deformation where slip is activated on a large number of slip
planes. The yield stress Y is the corresponding limit for slip activation. These
parameters are limits for the elastic range; i.e. the range wherein deformations
are reversible. After plastic deformation begins it is assumed that the strain €
can be separated into elastic and plastic parts, € = €e + €p; only the elastic part of
the strain is reversible so after plastic deformation starts to develop, unloading of
the body leaves a state of residual plastic deformation. Ordinarily the unloading
modulus is identical with the elastic modulus E so in the unstressed state the
residual strain is €p.
In many solids the stress required to continue plastic deformation increases
with the plastic strain; i.e. the material strain hardens. This increase in stress that
is required to continue dislocation motion on slip planes has been associated with
the stress to overcome dislocation entanglements. Part of this increase is offset,
however, by an increase in the number of active slip planes as stress increases.
A simple phenomenological model for an elastic-plastic material with strain
hardening presumes that stress increases in proportion to elastic strain if € < €y
while if € > €y the stress increases in proportion to total strain. Thus plastic
deformation develops if € > € y; in this range if strain is increasing, d€ / dt > 0,
the stress and total strain are related by a linear strain hardening modulus Et ·
6 Elastoplastic and Viscoplastic Constitutive Relations

/
o y
(/)
/
(/)

~
/
~I /
r/
V;
'4J

Fig. 1.3 Elastic-plastic material with linear strain hardening and elastic unloading.

This counterpart of the elastic modulus E is shown in Fig. 1.3

Let the strain hardening and elastic moduli have a ratio a = E( 1E , where a < I.
For pure bending of rate-independent materials, the elastic and plastic constitutive
relations result in a unique moment-curvature relationship while the bending
moment increases,
(J' = - EZIC,
(1.6)
(J' = -
-
aEZIC -
-
Y(1- a) sgn(IC) sgn(Z) ,

The second of the expressions applies only if the location of the neutral axis is
independent of curvature; i.e. if the cross-section is doubly symmetric. The bending
moment for the cross-section is obtained by integrating the first moment of the
stress as in Eq. (1.4); the moment is taken about the transverse axis through the
centroid. Thus, for a rectangular cross-section where My = Ybh 2 16,

MIMy
(1.7)
M/My = a(IC/ICy)+0.5(3-IC~/IC2)(l-a)sgn(IC), IICI> ICy, IT> 0
It is often convenient to express this relationship in terms of the curvature resulting
from some applied bending moment. For an elastic-plastic material this results in
a cubic equation that has only one real root,
0= (IC/IC y )3- (m/a)(IC/IC y )2 + 0.5(l-lIa)[1-3(IC/ICy )2], Iml>l (1.8)
where m = MI My. The elastic-plastic relations (1.7) and (1.8) apply also to strain
softening materials (a < 0) if the moment-curvature relation is single valued.
This condition requires sgn(M) = sgn(IC) which is satisfied if IICI/IC y < ~l-l/a.
If the stress-strain idealization is elastic-perfectly plastic the moment-curvature
relation for a rectangular cross-section is obtained directly from Eq. (1.7) by
considering the limiting case as a --t O. Thus
M 1My IC / ICy, IICI :-:; ICy
(1.9)
M/My
or
1.2 Pure Bending of Rate-Independent Bar 7

2·0r------------------,

'·7--
~i~erfec.!.!lastic_ _ .E- - -,:5-
approximation for (0) /,"'" - ~-a---"""
f11f1 y
/..
c
1·0 '·0·

2 3
K/Ky

Fig. 1.4 Elastic-perfectly plastic moment-<:urvature relations for three cross-sections: (a) rectangular, (b)
circular and (e) an ideal I-beam. The asymptotic values for bending moment at large curvature are indicated.

I(/I(y = [3 - 21MI/ My rl/2 sgn(M), IMI> My

The moment-curvature relation for pure bending depends on the shape of the
cross-section. For a circular cross-section with radius a the limiting elastic curvature
is I( y = Y / Ea while the limiting elastic moment My = Ell( y = nYa 3 /4. Thus for a
circular cross-section the elastic-perfectly plastic stress-strain idealization gives
a moment-curvature relation

I( > I(y (l.l 0)

For elastic-perfectly plastic materials, the moment-curvature relations for pure

bending of rectangular and circular cross-sections are compared in Fig. 104. The
fully plastic bending moment Mp is the asymptotic limit of the moment-curvature
relation for a perfectly plastic material as curvature I( becomes indefinitely large.
Thus Mp / My = 1.5 and 1.7 for rectangular and circular cross-sections, respectively.
This ratio of the fully plastic bending moment Mp to the elastic limit My is
known as the shape factor for bending ¢Jm
(1.11)

The shape factor is defined only for perfectly plastic materials since with strain
hardening, there is no asymptotic limit for bending moment as the curvature
becomes large without bound. It depends on the distribution of stress in the cross-
section; i.e. on both the cross-section shape and the stress resultant. Table 1.1
gives this structural property for separate bending, shear and torsion loads acting
on typical cross-sections. Tension N has not been included since in all cases the
elastic and fully plastic limit forces are identical, Ny == YA = N p ; consequently,
the shape factor for axial force is ¢In = 1
 00
Table 1.1 Yield and fully plastic limits for bending, shear and torsion
(Tresca yield k = Y12, von Mises yield k = Y/ Y3 )

2nd
Cross-section Area moment

A 10 My M I/Im Qy I/Iq Ty 1/1,

Yrf -yf kif if KIf -if
bh Ah'/12 hl6 hl4 1.5 2/3 1.5 f3,b' b(3 - blh)' 3 -b/h
1it-~
6 6f32

lUl 2 Aa 2/4 a/4 4al3n 1.7 314 1.33 al2 2al3 1.33

tIl
a* 2lUlC Aa 212 al2 2aln 1.27 1/2 2.0 a a 1.0
0;
'"
8'
'0
:-;e-
A~ [
n'
0>
.,
Q..
A Ah'/4 hl2 hl2 1.0 1.0 h h 1.0
-@
f+- ,.,~
0
'For narrow rectangles, Timoshenko and Goodier [1971] give values of f3,(hIb): f32(1.0) = 0.208, f32 (l.5) = 0.231, f3,<2.0) = 0.246. '0

* Expression for b/h < I.

,.,~
('l
0

a.
g
~.

"
i<'
$!..
!'!
o·
~
1.3 Pure Bending of Rate-Dependent Bar 9

1.2.5 Elastic-Power Law Hardening Constitutive Equations

The linear strain hardening examined previously is not the behavior commonly
observed in materials stretched to a large extent. Over a large range of strain, a
power law hardening relation may be a better fit to measured properties. For
plane stress, the stress-strain relation for elastic-power law hardening materials
with yield stress Y can be expressed as
(J lEI < Ey = YI E
(l.12)
(J IEI>Ey
where /3 is a strain hardening parameter. For a rectangular cross-section deformed
by pure bending this stress-strain idealization results in a moment-curvature
relation,

(1.13)

Again if the hardening coefficient is positive, /3 > 0, the bending moment has no
asymptotic limit as curvature becomes indefinitely large. In a strain hardening
material the moment always increases with curvature; this tends to spread or
diffuse the regions with large curvature. Equation (1.13) has an asymptotic limit
as /3 ~ 0 that is identical with the moment-curvature relation for vanishing strain
hardening obtained from Eq. (1.6); i.e. MIM y =O.5[3-(K'ylKi].
The inverse of (1.13) gives curvature if the bending moment M is specified. In
a power law hardening material this curvature is obtained as a root of the nonlinear
equation f3 2
o= 1~I_l-/3(K'Y) _2+/3 M sgn(K'), (1.14)
K'y 3 K' 3 My

1.3 Pure Bending of Rate-Dependent Bar

1.3.1 Strain-Rate Dependent Constitutive Equations

Material rate-dependence can be of two kinds. Materials such as polymers at

temperatures above the glass transition temperature Bg exhibit viscoplastic behavior
that manifests itself as creep; this behavior is related to viscous flow as polymer
chains slip past one another. In polycrystalline materials (e.g. metals), viscous
drag due to dislocation motion around barriers exhibits similar behavior that has
been termed rate sensitivity of strain hardening. Effects of viscoplastic constitutive
relations on deformation of impulsively loaded structures have been considered
by Ting [1964] and Ting and Symonds [1962]. In many metals, however, rate-
dependence at temperatures substantially less than melting is related to an additional
stress required to generate and accelerate dislocations; i.e. to initiate changes in
the rate of plastic flow. The latter effect is most prevalent in metals that have a
distinct and easily defined yield stress; e.g. Bee metals and precipitate hardened
alloys (Hartley and Duffy, [1984]). Low carbon steel is notoriously rate sensitive
in this respect; at strain-rates that commonly occur during impact of hard bodies
(e.g. 10 3 S-I) the yield stress of mild steel can increase by more than a factor of
two over the static value.
lO Elastoplastic and Viscoplastic Constitutive Relations

a
E> 2Eo
2 y f - - - - - - - - E =2 Eo

y I - - - - - - - E : =0

o
Fig. 1.5 Rate-dependent flow stress yd for nonhardening material.

Microstructural models of rate effects are based on thermal activation of

dislocation motion; thus the effective plastic strain-rate is given by an Arrhenius
equation,
e P leg = exp( -!l.U I kB)
where !l.U is an activation energy which depends on the barriers opposing
dislocation motion, k is Boltzmann's constant, B is absolute temperature and eg
is a characteristic strain-rate for the material. The activation energy decreases
with increasing stress. This model for rate effects results in a stress that depends
on only the current strain-rate and not on strain history. 2 A unidirectional
constitutive relation representing these effects is known as the Cowper-Symonds
[1957] relation.
The Cowper-Symonds relation represents a rigid-perfectly plastic material with
dynamic yield or flow stress (1(£) that depends on the strain-rate e. (In this
model all strain is plastic so it is convenient to omit the superscript p.) Thus the
ratio of dynamic to static yield stress ydly is

( E
-yd = 1 + -)" r e>O (1.15)
Y £0'
where eo and r are material constants. The characteristic rate Eo is the strain-
rate at which yd = 2Y as shown in Fig. 1.5, while the material constant r is a
measure of sensitivity to strain-rate.
For pure bending of beams or other slender members the relationship between
bending moment and rate of curvature is obtained from the condition that plane
sections remain plane. Therefore, at a distance Z from the neutral axis
£ = - ZIC (1.16)
where the rate of change of curvature at a section is IC == dK/dt. Thus for pure
bending of a rigid-perfectly plastic rate-dependent section where (1 = ~ Perzyna
[1962] has given the stress for a linear strain-rate distribution (1.16) as

_'JlIr
-ZIC
= 1+ (~
(1
Y
This stress distribution is illustrated in Fig. 1.6. The normal stresses on a cross-
section of area A yield a resultant couple

2Hartley and Duffy [1984] observed that strain-rate history effects are more pronounced in FCC
metals, whereas strain-rate and temperature sensitivity is stronger in BCC metals.
1.3 Pure Bending of Rate-Dependant Bar 11

Table 1.2 Material parameters for rate effect

. 1
Material £0 (s -1) r £o,(s- ) Reference

Mild steel 40.4 5 65 Forrestal and Wesenberg [1977]

Stainless steel 100 10 160 Forrestal and Sagartz [1978]
Titanium (Ti 50A) 120 9 195
Aluminum 6061-T6 1.70 x 106 4 2.72 X 106 Symonds [1965]
Aluminum 3003-HI4 0.27 x 106 8 0.44 X 106 Bodner and Speirs [1963]

while for a rectangular cross-section with depth h,

M;
Mp
= 1 + ~(h-K)lIr
2r+ 1 2Eo '
(1.17)

where M; = Ybh 2 /4 = 3My12 is the static fully plastic moment for the rectangular
section. The corresponding relation for rate of change of curvature K: is

(1.18)

where Eor == £0(1 + 1I2rr .

Some representative values of material parameters for use in the Cowper-
Symonds relation are given in Table 1.2. These values were obtained for strain
e '" 0.05; they may not be accurate for strains that are either very large or very
small in comparison with the elastic limit.
The methods described thus far have focused on rate effects and ignored strain
hardening. While this is acceptable for small strains, hardening effects can be
more important than strain-rate at large strains. Perrone [1970] has suggested a
constitutive relation that includes both effects; his relation has the advantage of
being separable and thus more easily evaluated.

(1.19)

Fig. 1.6 Strain-rate and stress distributions in a rate--dependent deforming section where Ii: > O.
12 Elastoplastic and Viscoplastic Constitutive Relations

1.4 Interaction Yield Functions and Associated

Plastic Flow

1.4.1 Elastic Limit for Bending and Tension

In general, slender structures are subjected to a combination of simultaneous
loads rather than pure bending; in addition to a couple, the beam may support
axial and shear forces as well as an axial torque. For elastic deformations the
state of stress associated with each of these stress resultants is linearly related to
the strain so the principle of superposition is applicable; each component of stress
and strain is just the sum of contributions from the individual stress resultants.
Thus a beam with axial force N in addition to bending moment M has normal
stresses (1 on a cross-section
N MZ
(1 = - - -
A 10'
1(11<
Y

where 10 is the second moment of area about the transverse axis for a cross-
section of area A and Z is the transverse distance from the centroidal axis. Initial
yield for each of these stress resultants acting independently is denoted by Ny and
My where, for a doubly symmetric section of depth h,
My = EloKy = 2Ylolh, Ny = YA (1.20)

In terms of these separate limits for N, M the elastic range is given by the yield
condition
0.21)

The stress distribution varies linearly through the depth of a section; it has
tension on one side of the neutral surface and compression on the other side. For
pure bending the neutral axis is coincident with the centroid but if there is axial
force N in addition to a couple, the neutral axis is located some distance t;h12
from the centroid. Hence yield occurs first at either the top or bottom fibers of
the beam; thereafter as M and N increase an increasing part of the cross-section is
stressed beyond yield. Around the neutral surface there is a core of elastically
deforming material but elastic-plastic deformation is present beyond some distance
£y I K from the neutral surface. These plastic deformations occur for only a small
range of stress resultants if the material is elastic-perfectly plastic. For any stress
resultant Qa acting independently, the ratio of the fully plastic to the elastic
limit is given by the shape factor ¢a'

1.4.2 Fully Plastic Limit Surface for Bending and Tension in

Elastic-Perfectly Plastic Bar
The stress resultants for tension N, shear Q, torsion T and bending M in the
theory of plasticity for slender structural members are analogous to stress
components in a continuum; hence they are sometimes referred to as generalized
stresses Qa where QJ = N, Q2 = Q, Q3 = T and Q4 = M. For sections that are
partly plastic, in the plastically deforming region the stresses are related by a
yield criterion; e.g. the von Mises or Tresca yield conditions. The yield condition
1.4 Interaction Yield Functions and Associated Plastic Flow 13
-y y
- a I
\.
-
y
a
\.

+
/
I
I
(
I

total an + am tension an bending am

Fig. 1.7 Normal stress distribution for elastic-perfectly plastic bending plus tension. For doubly symmetric
cross-sections the stress distribution separates into a symmetric part (tension) and an anti symmetric part
(bending). The chain line (- - -) is at half the total stress. The chain-dot line is the reflection of the chain line
about the centroidal axis. The sum of tension and bending stresses under the chain-dot line vanishes.

for stress can be used to form a yield function 1f'e (Qa) for stress resultants; e.g.
for a section subjected to simultaneous axial tension N and bending M the yield
function is
1f'
e
=M+M_I
My Ny (l.22)
This function is a measure of the state of generalized stress in the section. The
elastic limit is given by a yield condition 1f'e = 0, that provides an upper bound
on generalized stresses corresponding to elastic (reversible) strains at every point
in the cross-section; thus the stresses are elastic if 1f' < 1f'e while 1f' > 1f'e implies
that some parts of the cross-section are strained beyond the elastic limit. In stress
space the elastic limit 1f'e = 0 is a surface that relates the generalized stresses.
For a material that is elastic-perfectly plastic, there are ultimate values for
stress resultants; these asymptotic limits for large deformations are termed the
fully plastic axial force Np and the fully plastic bending moment Mp. The
independent fully plastic stress resultants for generalized stresses are respectively
Np = Ny = YA (1.23)
A state of stress that results in part of the cross-section being strained beyond the
elastic limit gives 1f'e > O. Further proportional increases in the stress resultants
increase 1f'e and the deformation; the increases in stress asymptotically approach
a limiting or fully plastic state of stress as the deformation becomes indefinitely
large. A fully plastic limit function 1f'p (Qa) can be defined that gives a relationship
between stress resultants in the fully plastic state, 1f'p = O. This can be obtained
from the stress resultants expressed as a function of the location of the neutral
surface. For example, Fig. 1.7 illustrates the normal stress distribution on a section
that carries both bending moment M and axial force N; the cross-section is
rectangular with depth h and width b. The deformation is such that the stress
distribution is elastic-plastic near the top and bottom surfaces; the stresses are
tensile above and compressive below the neutral surface located a distance ~hl2
from the centroid. The perfectly plastic stress distribution for a doubly symmetric
section is readily separated into a part that is anti symmetric with respect to the
centroid and a remainder of one sign. The stress resultants for these separate
parts are the bending moment and axial force respectively

0.24)
 --
14 Elastoplastic and Viscoplastic Constitutive Relations

y y
a
-L--
N I a

total tension bending

Fig. 1.8 Elastoplastic tension N and bending moment M for centroid obtained from alternative separation of
normal stress distribution. This method of separation depends on the cross-section; it divides the section
into a part that is anti symmetric about the neutral axis and the remainder. N / Ny =~, M / My =
~2 +3~(I-n+3[(I-n2 -e]l2.

where the elastic core has half depth ~h/2 = h/(ylK:. These expressions can also
be obtained from the alternative separation of the stress distribution illustrated in
Fig. 1.8; for a rectangular cross-section the tension force N is obtained from the
part of the normal stress field that is not antisymmetric about the neutral axis.
The bending moment M is the moment about the centroid of the normal stresses
on the entire section. For combined forces on a section the curvature /( is obtained
by eliminating , from the expressions above,

[
3_ 21MI _ 3N22]-1/2 (1.25)
My Ny

2
+
M/M
Y

--
t~~
EP

N/,
Ny

Fig. 1.9 Axial force Nand bending moment M for stress states on fully plastic (--) and yield (- - -) surfaces
of beams with circular, rectangular or ideal I-beam cross-sections. The associated flow law for the fully plastic
state relates the rates of stretching and relative rotation.
1.4 Interaction Yield Functions and Associated Plastic Flow 15

For a cross-section composed of an elastic-perfectly plastic material, a fully

plastic state of stress is approached in the limit as curvature becomes indefinitely
large. The fully plastic limit function 'I'p for generalized stresses in a rectangular
cross-section is given by
'I' = 21MI + N2 _ 1
p 3M y N y2
The fully plastic limit is 'I'p = 0 while in partly elastic sections, the stress resultants
satisfy 'I'p < o. Figure l.9 compares the elastic and fully plastic limits of
rectangular and circular cross-sections. In general, the state of stress at yield is
circumscribed by the fully plastic limit.
For application to rigid-perfectly plastic materials it is convenient to express
the fully plastic limit function in terms of independent fully plastic stress resultants
M p ' Np instead of the independent yield limits. Defining nondimensional fully
m,
plastic stress resultants ii,
__ IMI _
m=------
IMI and 0.26)
Mp f/JmMy
gives the plastic limit function 'I'p for stress resultants that correspond to fully
plastic stresses in a rectangular cross-section,
(l.27)
Stress states that give 'I'p = 0 are fully plastic while sections with a stress state
'I'p < 0 are at least partly elastic. (For a rigid-perfectly plastic material there is
no deformation if 'I'p < 0.)
The fully plastic limit function for other cross-sections can also be obtained
by integrating over the cross-section the stresses and the first moment of stresses
about the centroid. Thus for a beam with thin-walled circular cross-section, the
stress resultants for perfectly plastic stresses are bounded by a fully plastic upper
limit 'I'p = 0 where

'I'p = Iml-cos(~) (1.28a)

)r
For a beam with solid circular cross-section the fully plastic limit function is

'I'p = Iml- [cos( ~. - mIl3~I_m2/3 (l.28b)

Also, the fully plastic limit for an ideal sandwich or I-beam is shown in Fig. l.9;
this is an I-beam that carries stress only in thin flanges. The fully plastic limit is
obtained by considering an I-beam with finite thickness flanges and taking the
limit as flange thickness decreases. This gives Mp = My and Np = Ny so the
elastic and fully plastic limits are identical; thus
0.29)
Interaction limit functions for stress resultants in perfectly plastic plates and shells
are discussed by Sawczuk [1989, p 80] and Save and Massonet [1972].

1.4.3 Yield and Fully Plastic Stress Condition

The state of stress required to initiate plastic deformation can be expressed as a
yield condition that relates components of stress to a limiting condition for stress
16 Elastoplastic and Viscoplastic Constitutive Relations

or strain energy; e.g. Tresca's criterion corresponds to a maximum shear stress

while the von Mises criterion corresponds to a maximum deviatoric strain energy.
The yield condition is both an upper bound on the state of stress for elastic
response and the stress at which plastic deformation commences. Usually the
yield condition 'Pe(Qa) = 0 is an equation that relates the largest magnitude of
various components of generalized stress for elastic (reversible) behavior. Elastic
stresses throughout a section give 'Pe(Qa) < 0, while a section that is partly elastic
and partly plastic has 'Pe (Qa) > O. If the material is elastic-perfectly plastic (i.e.
no strain hardening), generalized stresses asymptotically approach a limiting or
fully plastic stress condition, 'Pp(Qa) = O. It is worth noting that yield and fully
plastic stress conditions are almost always expressed in terms of a selected set of
generalized stresses
where 11 ~ 6
Only generalized stresses for the more important sources of energy dissipation
need to be included in the yield condition for a particular problem. 3 The analysis
is simplified by neglecting the effect on yield of small components of stress;
hence, 11 ~ 6.
Yield and fully plastic stress conditions are limiting surfaces in stress space -
they are continuous and piecewise differentiable surfaces surrounding the unstressed
state. For perfectly plastic materials where strain hardening and strain-rate effects
are negligible, these surfaces are upper bounds for elastic and elastic-plastic
behavior, respectively. Yield and fully plastic stress conditions can be used as
functions indicating the state of stress in a section.
elastic stresses
'Pe > 0, 'Pp < 0 elastic-plastic stresses (1.30)
'Pe > 0, 'Pp0 = fully plastic stresses
The fully plastic limit is an upper bound for stress states that satisfy yield in any
part of the cross-section. These limit surfaces are path independent; i.e. independent
of the history of stress in the section. For rigid-perfectly plastic materials, the
yield and fully plastic stress conditions are identical so these terms are synonymous.

1.4.4 Associated Flow Rule for Plastic Deformations

The fully plastic limit is a convex surface in generalized stress space. This surface
is a differentiable function of the stress resultants except at isolated corners such
as m= 0, ii = ±l in Eq. 0.27). Convexity of the locus of fully plastic stresses
implies that if one stress resultant increases the others cannot simultaneously
increase.
Both the yield and fully plastic loci of generalized stresses must be convex if
the rate of work done by stresses is to be positive for any plastic deformation.
Consequently, for any state of stress on the yield (or fully plastic) surface, energy
is dissipated by internal plastic work during an incremental change of stress to

'In developing a constitutive model for an isotropic continuum. we require the yield function '¥e (CJ'ij) to satisfy
a principle of objectivity; i.e. that yield is invarient with respect to orientation of the coordinate system. This
principle does not apply to the yield function '¥e (Qa) for a slender member since the cross-sectional properties
are not isotropic.
1.4 Interaction Yield Functions and Associated Plastic Flow 17

any different state on the yield (fully plastic) surface. Plastic deformation always
dissipates energy. Thus if (Jij is a state of stress on the fully plastic locus and (Jij
is any state of stress inside this locus, an increment of plastic strain deC for (J ij
satisfies
( (J. - (Js.)
If If
> 0
del'I f - or (1.31)

where qg are measures of deformation or generalized strains corresponding to

the generalized stresses included in the fully plastic stress function If'p. The
proposition (1.31) expresses a postulate of nonnegative dissipation by plastic
deformation - this was given independently by Taylor [1947] and Hill [1948].
Here this postulate has also been expressed in terms of the analogous generalized
stresses Qa and the conjugate generalized plastic strains qg. A restricted form
of this proposition is known as Drucker's postulate for a stable, dissipative material
[Drucker 1959]. It immediately gives a positive increment of plastic work
(Jij dEfJ ;;::: 0 when we consider (Jij = O. For a smooth fully plastic locus Drucker's
postulate requires a positive rate of work for plastic stresses. This is satisfied if
the plastic strain increment dEfJ has a direction normal to the locus of fully
plastic stresses. Thus for a state of stress where the convex surface is smooth, the
following plastic strain-rate increments correspond to If'p = 0:
del' dlJ', dlJ',
EfJ == -d'f = v:1 P or dqg = v :1QP (1.32)
t v(Jij v a

where v is a positive scalar constant. This flow rule satisfies the normality condition
relating increments of deformation to the state of stress; consequently, it is termed
an associated flow rule.
In elastic-perfectly plastic beams and slender members the deformations are
stretch and curvature along the centroidal axis. The states of stress for these
deformations are located between the yield and fully plastic loci. For elastic-
perfectly plastic stress resultants that approach the fully plastic limit surface, the
curvature becomes large without bound. An indefinitely large curvature corresponds
to relative rotation at some point along the axis of sections on either side of the
point. Hence for stresses on the fully plastic surface, the correct measure of
flexural rate of deformation is rate of rotation Om (X) rather than rate of change
of curvature. In particular, for static or quasistatic deformation the rigid-perfectly
plastic material idealization always results in discontinuities in inclination at
stationary 'plastic hinges'.
In a rigid-perfectly plastic beam or slender member the increment of
deformation (stretch and rotation) can be related to a fully plastic state of
stress If'p = 0 by an associated flow rule. From the fully plastic limit surface for
a rectangular cross-section, Iml
+ ii 2 -I = 0, one obtains generalized plastic strain-
rates EP = 2vN IN; and Of:. = vi Mp. Hence, in a plastically deforming section the

;t
stretch and relative flexural rotation are related by

!~ 2 for iEPi ~ 21:; Of:.1 (1.33)

This relation applies to states of stress on smooth parts of the fully plastic limit
surface where the limit function is differentiable; it does not apply at corners.
Similar relations can be obtained also between any other generalized plastic strains
that are coupled through the fully plastic limit surface for the conjugate generalized
stresses.
18 Elastoplastic and Viscoplastic Constitutive Relations

1.4.5 Separated Yield Functions and Separated Plastic Flow

A consistent model formulation requires a deformation field with a component of
deformation for each component of generalized stress in the yield function; i.e. if
axial forces significantly influence the bending moment in a plastically deforming
beam, the equations of motion must consider stretching as well as flexural
deformations. Thus interaction fully plastic limit surfaces, such as those illustrated
in Fig. 1.9, introduce several complications for analyses of structural deformations.
These are primarily associated with satisfying the flow rule in plastically deforming
regions while simultaneously satisfying the equations of equilibrium or equations
of motion for the system. These complications are avoided by an additional
constitutive approximation which circumvents the flow rule.
The curved surfaces for fully plastic generalized stress in Fig. 1.9 can be
approximated by a plastic limit function that provides square limit surface Pp = O.
For sections subjected to simultaneous bending m
and axial force ii this square
plastic limit function is given by
m-l

ii-I (1.34)

This fully plastic locus circumscribes the true fully plastic surface for an elastic-
plastic material as shown in Fig. 1.10. Hence it overestimates the stiffness during
plastic deformation; i.e. the reduction in stiffness due to axial force has been
neglected. While the circumscribing square gives an upper bound on fully plastic
generalized stresses, a lower bound can be obtained from an inscribing square.
The main benefits derived from these approximations arise from the associated
flow rule which provides the following relations between generalized strains for
Pp = 0,

£ = Q, for Iml 1,

e > 0, for Iml < 1,

-- EP

Fig. 1.10 Separated limiting stress approximation circumscribing the fully plastic stress locus. The normality
condition provides the ratio between components of deformation rate for any state of stress on the fully plastic
surface.
1.5 Interaction Yield Surfaces Including Shear 19

At corners where Iml = liil = 1, a range of ratios EI emfor deformation rates satisfies
the flow rule; at these points the only thing that can be said is that sgn(e) = sgn(ii)
and sgn(em ) = sgn(m).

1.5 Interaction Yield Surfaces Including Shear

If a bar or beam carries shear force or torsion in addition to either axial force or
bending moment, both normal and shear stress components act on the cross-
sections. In parts of a section that are plastically deforming the yield criterion
involves both these stress components. For thick walled or solid cross-sections
the distribution of individual stress components in the plastically deforming parts
of the cross-section can change with deformation. Here these changes are ignored.
For a perfectly plastic material an upper bound for the yield condition can be
obtained with stresses that are conjugate to a kinematically admissible velocity
field; i.e. a continuous velocity field that satisfies the displacement constraints.
This bound for the fully plastic state will be based on an assumption for the
distribution of normal stress; namely, that in the fully plastic stress state the
normal stress distribution with shear is identical to that without shear. This approach
was used by Hill and Siebel [1953]; the present development uses their method
but obtains yield functions for rectangular cross-sections that carry tension, shear
and bending and for circular cross-sections that carry tension, torsion and bending.
Necessary conditions for application of the bound theorems are described in
Chap. 2.

1.5.1 Tension, Shear and Bending in Rectangular Cross-Section

Tension and Shear Let (J and r be normal and shear stress components on a
section that is plastically deforming. If the stresses are uniformly distributed on a
cross-section of area A they have stress resultants for tension N and shear Q,
N = (JA, Q = rA ( 1.35)
Notice that this stress field corresponds to a uniform shear stress imposed on a
pretensioned bar; thus the shear stresses in the cross-section do not vanish at the
top and bottom surfaces. For boundary conditions of negligible shear stress on
transverse surfaces, this stress distribution is not in equilibrium at the surfaces;
nevertheless, it is useful for obtaining an upper bound of the fully plastic state of
stress.
For a perfectly plastic material the yield limit for this combined state of stress
(1.35) is given by the von Mises yield criterion
(1.36)
In contrast if either tension or shear vanishes the fully plastic stress resultants are
Np = YA and Qp = YAI-fj. After dividing Eq. (1.36) by y2, the yield criterion
can be expressed in terms of nondimensional generalized stresses ii = N I N p and
q = QIQp,
20 Elastoplastic and Viscoplastic Constitutive Relations

(1.37)

Hence, the fully plastic axial stress CTy is reduced by shear

Y
CTy = (1.38)
~1-jl
Tension, Shear and Bending The effect of shear is incorporated in the fully
plastic limit function for tension and bending by reducing the normal stresses by
the factor found in (1.38); i.e. for a rectangular cross-section the limit function
(1.27) with shear gives
-2
'1'= m + n -1 (1.39a)
p ~1-jl l_q2

or alternatively

'Pp = m~l- q2 + ii 2 + q2 -1 (1.39b)

If 'Pp == 0 the cross-section is plastically deforming, while if 'Pp < 0 the section
is rigid or undeforming. Although this surface is an upper bound for the fully
plastic limit surface, it is very slightly smaller than a different upper bound given
by Hodge [1957] for the case of negligible tension.

1.5.2 Tension, Torsion and Bending in Circular Cross-Section

Torsion and Bending For a solid circular cross-section of radius a, the cross-
section area is A == 7ra 2 , the second moment of area about the neutral axis is
10 = 7ra 4/4, while the polar second moment of area for the centroidal axis is
J o = 210 • On an elastic cross-section, the bending moment M gives normal stress
CT in a fiber at distance Z from the neutral axis; this stress varies with curvature
K: at the section. Similarly, a torque T gives a shear stress 'l" in a fiber at distance
r from the centroidal axis so in the elastic range torque varies with axial twist
tJ = dBt 1dX where Ot is the rotation of the section about the longitudinal axis.
Thus for an elastic bar,

-CT = -M
- = -EK: (l.40)
Z 10 '

where G = E12(1 + v) is the shear modulus and v is Poisson's ratio. The largest
stresses occur in the outer fibers at distance a from both the neutral axis for
bending and the centroidal axis for twist. For each stress resultant in isolation,
yield in these outer fibers occurs for a bending moment My = ±YIola or a torque
Ty = ±YJo la.J3, where the yield condition for pure shear 'l"y == YI.J3. Thus for
combined bending and shear the von Mises condition (1.36) gives a criterion
'Pe = 0 for the stress resultants that cause yield in the outer fibers,

(1.41 )
1.5 Interaction Yield Surfaces Including Shear 21

The section is elastic for 'Fe < o. If the stress resultants are on the yield surface
then IJ'e = 0, while if IJ'e > 0 and IJ'p < 0 the plastically deforming region covers
an increasingly large part of the cross-section with increasing IJ'e. Because the
elastic stress distributions due to bending moment and torque are both nonuniform
and different, the boundary between the elastic and plastic regions is a non-
circular surface; plastic deformation is present in a lens shaped region at the top
and bottom of the cross-section. At each point in the plastically deforming segments,
the stress components change in accord with the flow law as deformation increases.
If either one of these stress resultants vanishes, the other resultant for elastoplastic
stresses in a circular cross-section is given by

IMI = -
--
My
2j(
3n
5--- 21(~)n~
1(2
1--
1(2
+ 31(. -1(l(y)] 1(>l(y, T=O
-SIll
I(y
-
I(
,

(1.42)

1'J > 1'J y , M =0

For a combined stress state where both M and T are nonzero, in the limit as
curvature or twist become indefinitely large either the bending moment or torque
approaches the fully plastic state. The shape factors relating fully plastic and
yield stress resultants for the circular cross-section are obtained by taking the
limit of Eqs (1.10) and (1.42) as curvature and twist become indefinitely large;
i.e. MplMy = 16/3n "" l.7 and TplTy = 413. A lower bound on the limit function
IJ'p for fully plastic stress resultants is obtained from Eq. (l.36) with the assumption
that in the elongated and in the compressed halves of the cross-section both (J
and r are uniform. Notice that for any cross-section the combination of the
independent fully plastic stress distributions for torsion and bending also satisfies
the boundary conditions; this is a necessary condition for a lower bound. Hence
if m= M / M p and t = T ITp, the fully plastic limit is given by IJ'p = 0, where

m<l, t <I (1.43 )

Yield and fully plastic limit surfaces for a circular cross-section subjected to
simultaneous bending and torsion are shown in Fig. 1.11.

Tension and Torsion Consider a circular cross-section loaded by simultaneous

tension and torsion. In an elastic section the axial and shear stresses (J, r yield
tension and torque stress resultants, Nand T respectively. The axial stress is
uniform while shear stress increases in proportion with radius;

N
(J = - = Ee (1.44 )
A ' r

where the deformations (i.e. axial strain e and twist 1'J) are related to the stress
resultants through elastic moduli E, G and section properties A, 10 defined in
Sect. 1.2. At the elastic limit the stress components (1.44) in the outer fibers,
r = a, satisfy the yield criterion 0.36); thus at initial yield the tension and torque
are on the yield surface 'Fe = 0 where the yield function is
22 Elastoplastic and Viscoplastic Constitutive Relations

Fig. 1.11 Yield and fully plastic limits for combined bending and torsion in bar with circular cross-section.

( 1.45)

In this expression the tensile yield force is Ny = ±YA and the yield torque is
Ty = ±Y10Ia.J3. (Recall that for a circular or square cross-section, the polar second
moment of area 10 is given by 10 == loy + loz = 210')
A lower bound on the fully plastic limit surface can be obtained with any
statically admissible stress field; i.e. a field that satisfies equilibrium throughout
the cross-section. First suppose the bar is loaded by a tensile force N and torque
T that satisfy P e = O. Consider axial and shear stresses with constant magnitudes
across the entire section; this results in uniform tensile stress (J = N / A and hence
from Eq. (1.36), uniform shear stress r = aT 110 , This stress field is a lower
bound on the actual stresses for this deformation process. Thus for tension and
torque in a circular section the fully plastic limit function is
Pp ~ ii 2 + (2 -1 (1.46)

where ii = N / N p and ( = T ITp .

The actual stresses may not be uniform so this is a lower bound for the fully
plastic limit surface. Note that the shear stress due to torsion reduces the fully
plastic axial stress (J y; i.e.

(1.47)

Tension, Torsion and Bending The effect of shear stress due to torsion is to
reduce the fully plastic axial stresses and hence the stress resultants for tension N
and bending M. Thus the suggestion of Hill and Siebel [1953] given in (1.47),
when substituted into the fully plastic limit function for a circular cross-section
(1.28b), yields

cos
3
21b -[b 11/3 [ [b 12/3]112)
{ 1-
1.6 Elastic Springback 23

For comparison, the yield limit P e = 0 obtained from the von Mises criterion
can be expressed as
Pe = (Inl +l m l)2 + 12 -1
where m= MI My, n = NINy and t = TITy .

1.6 Elastic Springback

When forces are removed from any section of a structure, the section springs
back as the stresses are relieved; the deformation decreases somewhat from the
largest values. For most materials, unloading occurs elastically irrespective of the
current state of strain. Figure 1.12 depicts the uniaxial stress-strain behavior of a
linear elastic material that exhibits linear strain hardening if the strain exceeds
the yield strain ey. This material unloads elastically along the dashed line. If the
material is stretched to some largest strain e. and then the stress is gradually
removed, during unloading the stress is continually in balance with the applied
load. When the applied stress vanishes the material has a residual state of plastic
deformation with strain e P as a consequence of the history of stress for that
location. In this section spring back relations are presented for both linear strain
hardening and perfectly plastic material characterizations; these are developed
for bars in pure bending and combined bending and stretching.

1.6.1 Pure Bending

During bending of a bar, beam or plate, the stresses vary across a section so the
stress history depends on distance from the neutral axis Z. Flexure and stretching
result in a linear distribution of normal strain on a section as a function of Z but
the normal stress distribution is not linear if part of the section is plastically
deforming. For pure bending, the neutral axis is at the centroid Z = Z, so the

a
a b
in =M/My

LK,i-
L -_ _ L-=-__ ~~ ____ y

Fig. 1.12(a) Stress-strain relation for a linear strain hardening material that exhibits the Bauschinger effect
upon unloading; (b) the corresponding moment--<:urvature relation for loading m> 0 and unloading m<O of
an elastic-plastic section.
24 Elastoplastic and Viscoplastic Constitutive Relations

kinematic condition that plane sections remain plane gives

O'(Z) 2Z~ Z
y
-- , - s~y
-
h~y h 2~
(l.48)
O'(Z)
= 1- (1- a) ( 1 - - 2Z~J -
Z ~y
2-
Y h~y h 2~

where a = E( / E, as given in Sect. 1.2.4. After taking the first moment of stress
about the neutral axis and integrating this over the cross-sectional area, the following
nondimensional moment-curvature relations for loading a rectangular cross-section
are obtained

m = k, k s1
(1.49)
_
m = -1 (1- a)(3 - --2-
k ) + ak , k >1
2
where the bending moment iii = M / My and the curvature k = ~ / ~ y. At any secti~n
loading terminates at some maximum bending moment iii. and curvature k•.
Thereafter the stress resultants are removed so the section unloads completely to
the final bending moment iiif = O. During elastic unloading there is a change in
curvature equal to - iii., so finally

(1.50)

Unloading occurs elastically across the entire section as long as there is no reverse
yielding. Reverse yielding first occurs in the outer fibers; for kinematic strain
hardening there can be no reverse yielding if iii. < 2. (Kinematic hardening gives
a yield surface 'Pe (Qa ) = 0 which expands in a self-similar pattern with increasing
plastic strain;. this results in the Bauschinger effect shown in Fig. 1.12.) For
loading followed by unloading of a beam in bending, the distribution of stress in
a rectangular cross-section is illustrated in Fig. 1.13.
In an elastic-plastic material, as the stress resultant is unloaded the stresses
decrease in proportion to their distance from the neutral axis. The residual stresses
that remain after the stress resultants vanish are locked in by the variation in
plastic strain fP across the section.

loading unloading residual

-y
,--

a c

Fig. 1.13 Stress distributions for pure bending in a doubly symmetric linear strain hardening bar during (a)
loading in the elastic-plastic range I < m < rflmm, (b) elastic unloading and (c) the residual stress when finally
"iii = 0 (sum of a plus b).
1.6 Elastic Springback 25

An early analysis of springback was developed by Gardiner [1957] for pure

bending of elastic-perfectly plastic sheet material in plane stress. The analysis
was extended to include the effect of linear strain hardening by Woo and Marshall
[1959]. Elastic springback relations for linear and power law hardening in both
plane stress and plane strain were obtained by Johnson and Yu [1981]; they
calculated final curvatures in rectangular and circular plates subjected to biaxial
bending.

1.6.2 Bending and Tension

In many forming problems tension is applied to a sheet or plate to control wrinkling
and reduce the required bending moment during forming processes; this is termed
stretch-bending. Another source of normal in-plane forces that reduce the fully
plastic bending moment are axial loads induced by impact at an angle of incidence
measured from the transverse direction. The in-plane forces result in a neutral
axis at some distance from the centroid and this also affects springback. For an
elastic-perfectly plastic material in plane stress, Yu and Johnson [1982] obtained
ratios for the change in one dimensional curvature during loading.
Figure l.14 shows the stress distribution through the thickness of an elastic-
perfectly plastic bar loaded in combined tension and bending. The tension has
moved the neutral surface a distance ?;hl2 from the centroid. This bar has an
elastic region that extends a distance ~ hl2 from the neutral surface; the remainder
of the section is fully plastic at tensile yield stress Y.
For combined loading, an elastic-plastic bar has several different patterns of
stress distribution depending on the tension N and bending moment M. These
patterns characteristically appear in specific regions relative to the elastic yield
surface (3) and the fully plastic limit surface (1) illustrated in Fig. l.15. This
figure depicts the yield and fully plastic surfaces in generalized stress space,
'iii = MIM y , Ii = NIN y

Zf
z~
(1~s -S)h/2
/-;T II~( -fj h/2

t h/2 ~. Chh th/2

11-5') {,! th/2

~ h1-t-t) h/2
b c

Fig. 1.14 Stress distributions in elastic-perfectly plastic section with both bending moment M and axial force
N. The curves correspond to (a) elastic stresses (b) primary plastic stress distribution (c) secondary plastic stress
distribution.
26 Elastoplastic and Viscoplastic Constitutive Relations

o.
y
Fig. 1.15 Patterns of stress distribution in

B~:b
elastic-perfectly plastic bar with rectangular
cross-section depend on the magnitudes of
/ -
nondimensional bending moment Iii and
axial force n .
o 0.5 1.0 i'i

where the stress resultants in a bar with rectangular cross-section have independent
elastic limits My = Ybh 2 /6, Ny = Ybh. Also, the curvature IC can be related to
the elastic limit ICy = 2YIEh for a rectangular cross-section; i.e. k = ICIICy . Thus
during loading of a rectangular cross-section, the curvature is related to the stress
resultants by
m,
{ 4(1- n)3 (3 - m - 3n) -2, 1- n ::; m ::; 1 + n - 2n 2
k (1.51)

[3(1- n 2) - 2mr1l2,
These equations refer to the elastic, primary plastic and secondary plastic regions,
respecti vely.
If maximum values of the loads m., n. exceed the yield limit, P(m.,n.) > 0,
after unloading there will be residual stress af in the cross-section and a residual
c~rvature ICf' The residual curvature is related to tEe c~rvature at yield ICy by
kf = ICf I ICy. This residual curvature is given EY _kf = k. - m•. Often the final
curvature is expressed as a springback ratio kf I k.; i.e. the part of the largest
curvature which is recovered on unloading

Figure 1.16 illustrates the springback ratio for a rectangular cross-section; this
1.6 Elastic Springback 27

1.0

~ *
lfi:.*
,
~ 05
"
'->0:*
........
.....
1->0:

Fig. 1.16 Springback ratio in elastic-perfectly plastic beam with rectangular cross-section subjected to
maximum bending moment m* and axial force n* .

indicates that stretch-bending increases the springback ratio in comparison with

that for pure bending.

References

Bodner, S.R. and Speirs, W.G. [1963]. Dynamic plasticity experiments on aluminum cantilever beams
at elevated temperature. 1. Mech. Phys. Solids 11, 65-77.
Cowper, G.R. and Symonds, P.S. [1957]. Strain-hardening and strain-rate effects in the impact loading
of cantilever beams. Brown Univ. Dept. Appl. Math. T.R. 28.
Drucker, D.C. [1959]. A definition of a stable inelastic material. 1. Appl. Mech. 26, 101-106.
Forrestal, M.J. and Sagartz, M.J. [1978]. Elastic-plastic response of 304 stainless steel beams to impulse
loads. 1. Appl. Mech. 45, 685-687.
Forrestal, M.1. and Wesenberg, D.L. [1977]. Elastic-plastic response of simply supported 1018 steel
beams to impulse loads. 1. Appl. Mech. 44, 779-780.
Gardiner, F.J. [1957]. The spring back of metals. Trans. ASME 79, 1-7.
Hartley, K.A. and Duffy, J. [1984]. Strain-rate and temperature history effects during deformation of
FCC and BCC metals. Mechanical Properties at High Rates of Strain (ed. J. Harding). Inst. of
Physics, London, 21-30.
Hill, R. [1948]. A variational principle of maximum plastic work in classical plasticity. Q. 1. Mech. Appl.
Math. 1, 18.
Hill, R. and Siebel, M.P.L. [1953]. On the plastic distortion of solid bars by combined bending and
twisting. 1. Mech. Phys. Solids 1, 207-214.
Hodge, P.G., Jr [1957]. Interaction curves for shear and bending of plastic beams. 1. Appl. Mech. 24, 453.
Johnson, W. and Yu, T.x. [1981]. On spring back after the pure bending of beams and plates of elastic
work-hardening materials - III. Int. 1. Mech. Sci. 23(11), 687-695.
Perrone, N.J. [1970]. Impulsively loaded strain hardening rate-sensitive rings and tubes. Int. 1. Solids
Structures 6, 1119-1132.
Perzyna, P. [1962]. The constitutive equation [orrate sensitive plastic materials. Q. Appl. Math. 20, 321-
332.
Save, M.S. and Massonet, C. [1972]. Plastic Analysis and Design of Plates, Shells and Disks. North-
Holland, Amsterdam.
Sawczuk, A. [1989]. Mechanics and Plasticity of Structures. Ellis Horwood, Chichester.
Symonds, P.S. [1965]. Viscoplastic behavior in response of structures to dynamic loading. Behavior of
Materials under Dynamic Loading (ed. N. Huffington). ASME, 106-124.
28 Elastoplastic and Viscoplastic Constitutive Relations

Taylor, GJ. [1947]. Connection between the criterion of yield and the strain ratio relationship in plastic
solids. Proc. Roy. Soc. Lond. A191, 441.
Timoshenko, S. and Goodier, J.N. [1971]. Theory of Elasticity. McGraw-Hill, New York.
Ting, T.C.T. [1964]. The plastic deformation of a cantilever beam with strain rate sensitivity under
impulsive loading. I. Appl. Mech. 31, 38-42.
Ting, T.C.T. and Symonds, P.S. [1962]. Impact on a cantilever beam with strain rate sensitivity. Proc.
Fourth US Natl. Congo Appl. Meeh. ASME, 1153-1165.
Woo, D.M. and Marshall, J. [1959]. Spring-back and stretch-forming of sheet metal. The Engineer, 135-
136.
Yu, T.X. and Johnson, W. [1982]. Influence of axial force on the elastic-plastic bending and springback
of a beam. I. Meeh. Working Tech. 6,5-21.
Chapter 2

Principles of Mechanics

The response of bodies to slowly varying loads is static or quasistatic. For slowly
varying loads, the sum of all forces acting on any segment of a body are in
balance since there are no accelerations; i.e. for any segment of the body, the
resultant of tractions on the surface of the segment is equal in magnitude but
opposite in direction to any external force acting on the segment. On the other
hand, rapid changes in load cause the body to accelerate; in this case the resultants
of stresses acting on any part of the body are related to the product of acceleration
and inertia by the laws of motion. For these two broad classes of loading the
differential equations for variations in stress (or stress resultant) across an arbitrary
segment of the body are obtained from either equilibrium equations or the laws
of motion. These laws form the basis of several useful principles. In this chapter
special forms of these principles will be developed that apply to small deflection
theory for planar displacements of slender bodies.

2.1 Kinematics

Begin by fixing the directions of a triad of mutually perpendicular unit vectors ej ,

j = I, 2, 3 in the initial undeformed configuration of a body at rest. These
vectors form the basis of a material or Lagrangian reference frame where the
current position of a particle is measured relative to that in the undeformed
configuration using spatial coordinates Xj fixed in the undeformed configuration.
The displacement of each material particle from its initial posi.tion is '!- vector
W(X) = W;(Xj ). In this Lagrangian description the velocity field W(X) = W;(X j ) ==
dW;(X) fdt is the rate of change for the position vector of a particle that was
initially located at Xj. Relative displacement of neighboring points in a body is
the sum of deformation of the material and rotation of the set of unit vectors. The
infinitesimal deformation in any differential element is given by the tensorial
strain tij' while components of rotation about e i x e j are denoted by OJi}: I

lIn this index notation a repeated subscript indicates a sum of components; e.g. £ii '= £" + £22 + £33·
30 Principles of Mechanics

(2.1)

The deformation is related to stress aij by constitutive relations such as Hooke's

law for a linear elastic material.
This monograph focuses on slender prismatic bars or beams where the cross-
section has a plane of symmetry. Applied forces, axial torques and initial velocities
all act in the plane of symmetry so there is no out-of-plane flexure; i.e. the axis
passing through the centroid of every section suffers solely in-plane displacements.
Many of our examples consider uniform cross-sections, but it is only necessary
that any changes in section are gradual. For slender members, let e j be the axial
direction through the centroid of every cross-section and e 2 be the normal to the
plane of loading. Thus if loading is applied in a plane of symmetry for the cross-
section the displacement field W(X) has only axial and transverse displacements
of the centroidal axis, W = ('"'),0, W3 ) as shown in Fig. 2.1. Likewise the rotation
9 of a cross-section has components 9 = (Of' - Om' 0) that are termed axial and
flexural rotation; in general OJ = ejjkO)jk .2 Notice that the flexural component of
rotation Om = -02 has a sign that is adopted from beam convention. For a slender
beam that is simultaneously stretched, twisted and bent, the axial and transverse
displacements (velocities) vary linearly over the cross-section.
For a slender structural member, the equations of motion relate axial variations
of stress resultants for cross-sections to accelerations in the member. In Chap. I
these stress resultants were termed generalized stresses; there is a conjugate set
of generalized strains for the cross-section. If the section has an outward normal,
n = njej in direction ej, the generalized strain vector for the section is obtained
from the strain tensor £ij and rotation vector OJ evaluated at the centroid
{£1l,e21'£31,Oj,02,03}' For planar deflections, however, £21 = 0 and 0)12 = 03 = 0
so that engineering strains and rotations are defined as

where axial twisting Or and in-plane rotation Om have a positive sense as indicated
in Fig. 2.1. Usually these variables are expressed as a reduced set of deformations
or generalized strains qj for deflections in the e j , e 3 plane

qa = {e'Y3,Op Om}

Generalized strains (or strain-rates) exist at every point in the body if the
displacement (velocity) field is at least piecewise continuous so that it is
differentiable. A piecewise continuous displacement field W;C (or velocity field
"-iC) that satisfies all displacement and velocity constraints is said to be
kinematically admissible; i.e. it satisfies compatibility. Kinematically admissible
fields yield generalized strains qf and strain rates 4f.

= =
2The permutation tensor e ijk is defined as eijk I if i, j, k are in cyclic order; eijk 0 if i = j, j = k
or k = i; and e ijk =-1 if i, j, k are in anticyclic order. Thus scalar and vector multiplications of vectors
u=uie i and v=vie i can be expressed as u·v=uie i ·ejv j =UiV i and uxv=eieijkujvk' respectively.
 0/r
2.2 Balance of Forces 31

el

e2~
--
t
~ ~
X
/" T
Y

I. dX
.1
Fig. 2.1 Stress resultants on cross-section of beam that give in-plane deflections of centroidal axis.

2.1.1 Inertia Properties of Cross-Section

For a uniform bar or beam, consider a differential element taken at any point
along the length. The differential element is a cross-sectional lamina with a normal
n = el; this has an area equal to the cross-sectional area A and length
cIS = n· dX = nj dX j . If the bar is composed of a homogeneous material with
density Pv' the differential element has mass pdS and moments of inertia Pv1i cIS
about the centroid:

(2.2)

where P is the mass per unit length and X = X - X . nn lies in the plane of the
lamina. With these definitions, the moments of inertia are directly related to the
second moments of area Ii for the section. Thus for a homogeneous bar with axial
coordinate Xl = X, the cross-section has second moments of area for principal
axes through the centroid,

(2.3)

2.2 Balance of Forces

2.2.1 Stress Resultants and Generalized Stresses

In a homogeneous body the stress a is a continuous function of the spatial
coordinates X. At every point the components of stress CT;j are related to directions
given by the set of mutually perpendicular unit vectors ei ; thus the second order
tensor a can be expressed as a dyadic a = CYij eie j' For a surface that passes
through any point of the body, let n be the unit vector normal to the surface at X.
This normal has components n i which are the direction cosines relative to the
basis vectors ei ; i.e. n = niei' At any point the traction t(n) (i.e. force per unit
area) is a vector that depends on the stress a and the orientation n of the surface
passing through the point
32 Principles of Mechanics

t == O'·n == tjnj

For a slender beam or a prismatic bar, we aIm to find deflections of the centroidal
axis - this is a representative point in the cross-section. The expressions for
these deflections in terms of the applied loads and cross-sectional properties are
obtained from equations of motion for a differential element that lies between
two cross-sections as shown in Fig. 2.1; i.e. an element with length dS along the
member. If the element is oriented such that the normal n has a direction n = e"
then dS == dX and the internal stresses give surface tractions t, == erit e i acting on
the cross-section. Hence a cross-section has stress resultants known as force
N == Njej and a couple M == Mje j that are obtained by integration over the cross-
sectional area A of tractions and the first moment of tractions about the centroid,

These stress resultants have components parallel and transverse to the centroidal
axis. Noting that for in-plane deflections some transverse components vanish
( Ny == M z == 0) while the nonzero stress resultants are given by

N==N x ferxxdA, Q==Nz=ferzxdA

(2.4)

M =- M y == - f Zerxx dA

These stress resultants are the components of the reduced generalized stress vector
Qa == {N, Q, T, M} where the axial force Q, = N, the shear force Q2 = Q, the
twisting moment Q3 = T and the bending moment Q4 = M.

2.2.2 Equations of Motion

Forces acting on any differential beam element with cross-sectional area A and
volume dV== AdX are of two types -long range and short range. The long range
forces, named body forces, are proportional to the mass of the element; typical
examples are gravitational or electromagnetic forces between bodies that are some
distance apart. A body force g has dimensions of force per unit volume. The
other type of force is due to direct contact between bodies. Contact forces or
tractions ten) act at a surface in contact with a neighboring body; the surface has
outward normal n. Tractions have dimensions of force per unit area. In either
case the force acting on one body is equal in magnitude but opposite in direction
from the force acting on the second body of an interacting pair.
Unbalanced forces on any element result in acceleration d 2 W/dt 2 where
W(X, t) is a time dependent displacement field with components Wi. Hence for a
differential element with volume dV, an equation of motion is obtained from the
rate of change of translational momentum
a~j
ax.
a( aw)
ar' '
+ gi = at Pv (2.5)
J
Assuming deformations in the cross-sectional plane are negligible, the acceleration
of the axis through the centroids is representative of translational and flexural
2.2 Balance of Forces 33

acceleration of a beam or a slender bar. For a beam with centroidal axis parallel
to e l and applied (body and contact) forces per unit length g, the stress resultants
acting on cross-sections of area A with normal direction e l give

~ftldA + g = ~(paw)
aX at at
where W is the displacement of the centroidal axis. Likewise, equating the differen-
tial element to the rate of change of moment of momentum about the centroid
gives

a
ax f- Xxt l dA = ata f - (09 -)
Pv Xx Jt"XX dA, o= f (Xxg)dA

where X is the position vector from the centroid in the cross-sectional plane.
Thus the equations of motion for infinitesimal in-plane deflections of a beam or
bar can be expressed in terms of the stress resultants Qa and corresponding
inertia forces that depend on the distribution of momentum in the cross-section,
aN a2 w
ax + gx P at 2 x (2.6a)

dQ+ a2 w
ax gz P at 2Z
(2.6b)

dT a2 () (2.6c)
ax Pv Ix at 2t

aM a2() (2.6d)
ax Pv Iy at 2m

Here Wx(X) and ~(X) are axial and transverse displacements for the centroid
of each section, respectively. The generalized rates of strain in these expressions
can be related to stress resultants Qa by a constitutive relation of the type developed
in Chap. 1. If the stresses satisfy the equations of motion throughout the body
and surface tractions on that part of the boundary where tractions are specified,
the stress field is called dynamically admissible. A dynamically admissible set of
forces in balance with accelerations ~d is denoted as (gf,p~d, Q~). If a
dynamically admissible stress field also satisfies the yield condition, it is a complete
solution only if the velocity field W;d is (1) kinematically admissible, (2) satisfies
initial conditions and (3) gives deformations related to the stresses by the
constitutive equations.
For quasi static or slowly varying loads the accelerations are negligible so that
Eq. (2.5) provides equations of equilibrium

a~j
ax +gi = 0, (2.7)
]

For infinitesimal in-plane deflections of a beam or bar, the equilibrium equations

for stress resultants on cross-sections are as follows:
aN
ax + gx = 0, (2.8)
34 Principles of Mechanics

aT aM
ax = 0, ax
= _Q

'
A stress field (gf, ~J Q~) that satisfies the equilibrium equations (2.7) in the
interior of the body and surface tractions on any part of the boundary where
contact forces are specified is called statically admissible. If a statically admissible
stress field also satisfies the yield condition, the associated stress resultants cannot
exceed the fully plastic stress resultants Q~ for a perfectly plastic solid, Q~ ~ Q~.
This inequality is the cornerstone of a lower bound theorem for forces that induce
quasi static plastic deformation.

2.3 Principle of Virtual Velocity

A principle of virtual velocity (virtual power) is the dynamic counterpart of the

principle of virtual work. For dynamic plasticity, the principle of virtual velocity
and the requirement of positive dissipation by plastic deformation yield useful
extremal principles for identifying dynamic modes and bounding dynamic solutions.
Here the principle of virtual velocity is established for a homogeneous body
with volume V contained withIn a surface r. The body is loaded by prescribed
surface tractions Fj and a body force gi per unit volume. On the part of the
surface r; where surface tractions Fj are specified, the boundary condition is
Fj = ni~j on r;
where ni are direction cosines of the unit outward normal vector n. In the interior
of the body the equation of motion (2.5) for a differential element relates the
a
acceleration components 2 W;ldt 2 == ~ to dynamically admissible stresses at,
- aaij 1axj = - Pv W; + gi
d "d d
(2.9)

A virtual velocity \tic

is any kinematically admissible velocity field. By
integrating the scalar product of \tic
and Eq. (2.9) over the entire body V we
obtain

- f aat
aX -.c WI dV = f[ .-Pv WI d W. I c+d '
g.I WI
c] dV (2.10)
V j V

Notice that the virtual work rate by external tractions Fj is

d . C

fr Fw.·c dr = f r na·w.
J J
d'c
IJ J I
fa(aijWj
dT = V aXi
)
dV (2.11 )

Then, since the stress tensor is symmetric

d·c
aijUYYi UA j = aijt:ij
d~iTC/~

where the strain-rate tensor eij = (a\tic 1aXj + awI

1aXi )/2 is obtained from a
kinematically admissible velocity field. Hence from Eqs (2.10) and (2.11) the
principle of virtual velocities can be expressed as
 r
2.3 Principle of Virtual Velocity 35

LOffeij dV = Q~q~dX = f)-Pv«idltjc +giltjc]dV + tF;ltjc dT (2.12)

This principle relates any kinematically admissible velocity field c with any lti
dynamically admissible stress field that is in balance with the applied forces. It is
helpful to note that the principle does not require a complete solution. Equation
(2.12) is most important for obtaining bounding inequalities for variables measuring
system response to dynamic loads.

2.3.1 Rate of Change for Kinetic Energy of System

For any dynamically admissible stress field we can define a rate of work done on
the system by external forces E~ and a rate of dissipation of energy by internal

J d'W;d JOfjd.d
forces iJ~ These elements are defined as
. d -= f r Fill-!. d dT +
Ein vgi dV and D
. d -= v Cij
r L d .d
dV = Jo Qaqa dX

The dynamically admissible stress field is in balance with the rate of change of
velocity lti d and this velocity field has kinetic energy Kd where

Kd =
-
~J P whvId dV
2 V V I

The rate of change of kinetic energy can be related to the rate of work done by
external and internal forces if the dynamically admissible velocity field ltid is
also kinematically admissible. (This is always true for the complete solution.)
Then with the substitution ltjd = ltic, the principle of virtual velocity (2.12) gives

fv d'd
a IJ.. c"IJ dV = f [--P d (ll-!
'd'd)
ll-!'d
+ g·W
1 +f 'd
(2.13)
V 2 dt I I
dV r FW
I I
dT

Finally, by substituting the deformations above into this expression of virtual

velocity, we find that the dynamic solution satisfies a rate equation,
K
. d = E. d - D. d (2.14)
in

The kinetic energy Kd increases if E~ > iJd and decreases if E~ < iJd.

2.3.2 Rate of Change for Kinetic Energy of Kinematically

Admissible Velocity Field »Ie

In regions that are plastically deforming the stresses are on the yield surface;
elsewhere the state of stress is less than yield. For any kinematically admissible
velocity field (KA VF), Drucker's postulate (1.31) gives aijeij '? aijeij where aij
is a safe state of stress that does not violate the yield criterion. Thus at any time
the stresses at every point satisfy this inequality for the dissipation power or rate
of dissipation. Noting that in plastically deforming regions the stresses aij, Q~
are on the yield surface, we obtain an inequality,
36 Principles of Mechanics

Thus for any KA VF wt, the rate of plastic dissipation is greater than or equal to
the rate of dissipation by a stress field that satisfies the yield condition, if 2 iJ.
It is useful to recall that a statically admissible stress field (J'ij is safe. .
The KA VF has kinetic energy KC and a work rate by external forces Efn,
where
(2.16)

U sing a development similar to that for the principle of virtual velocity, the rate
of work by external forces and 'safe' stresses acting on any KA VF can be defined
as
J = L
(J'iji:ij dV - S/iW;c dT - LgiW;C dV (2.17)

Then, together with definitions (2.16), Drucker's postulate provides an inequality

for rate of change of kinetic energy for any KA VF, i<~ If a rigid-perfectly plastic
body deforms while under the action of statically admissible applied loads, for
any current velocity ~ield .W;(Xj,t), the plastic dissipation exceeds the work rate
of the active forces DC > Ei~ and so the kinetic energy K C continually decreases.
Equations (2.15) to (2.17) give the inequality
(2.18)

The complete solution is kinematically admissible but it has a rate of change of

kinetic energy that is less than or equal to that of any other KA VF with the same
kinetic energy K,
(2.19)
Hence as a rigid-perfectly plastic body decelerates, at every instant the velocity
and deformation fields are evolving such that the rate of dissipation is a minimum
consistent with satisfaction of the yield condition.
This principle is sometimes used to calculate the location or extent of a
deforming region in a rigid-plastic structure. The location of a hinge in a beam
or the orientation of a hinge line in a plate is introduced as a variable; then an
extremum of the global rate of dissipation gives the value of this variable that
minimizes the rate of dissipation in the current configuration. This procedure
should be used with caution, however, unless the state of stress is similar in a
neighborhood of each deforming zone. Calladine [1985] has pointed to several
examples of problems where this procedure does not provide the correct solution;
his examples involve deformation fields with stretching in one part of the structure
and bending in another part. In these problems, the stress approximations used in
different regions are not equally accurate so the approximation 'error' can prejudice
the analytical result.

2.3.3 Extremal Principles for Complete Solution

The principle of virtual velocity is the basis of two theorems for velocities in
dynamically loaded rigid-perfectly plastic bodies. The theorem of Tamuzh [1962]
2.4 Bounds for Rigid-Perfectly Plastic Solids and Structures 37

is a kinematic principle while that of Martin [1972] is a dynamic principle. Each

defines a functional that is an extremum (stationary value) for the actual response
at any stage of motion.

Tamuzh's Principle: For any KAVF Wic

the corresponding accelerations ~c
and rates of change of strain-rate if} can be calculated. Also in those parts of the
body that are deforming the KA VF can be obtained from the flow rule for a
perfectly plastic material. Let T; be any tractions that act on the free surface of
the body during motion. Among all KA VFs ~c (and corresponding stresses aij),
the actual acceleration and corresponding stresses result in an absolute minimum
for the functional

J ( Wi
.. c) 1 f . Wi
= 2" vPv Wi
c .. c
dV -
f"
/i W;c dr + f c ooc
vaijeij dV (2.20)

If the acceleration field ~c contains discontinuities, there are additional terms in

the functional due to work done by stresses acting on the discontinuous surface
(see Tamuzh [1962]). Tamuzh's principle is in fact an application of Gauss's
principle of least constraint to a rigid-perfectly plastic body. This principle is
discussed by Pars [1965, P 42].

Martin's Principle: This theorem applies to any dynamically admissible acceleration

field ~d that is related through the laws of motion to a 'safe' stress field at
(i.e. the stresses satisfy the yield criterion 'P(at) sO). For any current velocity
Wi, the actual acceleration ~ results in a minimum for the functional
OOd) food'
J ( Wi = vPv Wi Wi dV (2.21 )

In rigid-plastic structural dynamics, applications of this principle begin by finding

dynamically admissible accelerations for some KA VF; i.e. the laws of motion are
used to find accelerations ~d for 'safe' stresses in plastically deforming regions.
For this KA VF, these stresses can be obtained from the flow rule.

2.4 Bounds for Rigid-Perfectly Plastic Solids and

Structures

For either static or dynamic loading of more complex elastoplastic structures, a

complete solution may be beyond our grasp. In this case upper and lower bounds
for response variables can be useful. These bounds are available for rigid-perfectly
plastic materials since in this case all energy dissipation occurs in plastically
deforming regions but in these regions there is no direct relationship between
stress and deformation. The bounds are especially valuable if the upper and lower
bounds are not far apart.

2.4.1 Upper and Lower Bounds on Static Collapse Force

For any particular distribution of applied force, the plastic collapse force is the
smallest magnitude of load that results in a plastic collapse mechanism; i.e. the
38 Principles of Mechanics

smallest force magnitude Fi in equilibrium with stresses Qa that are conjugate to

an increment of true plastic strain oqfx. A lower bound for the plastic collapse
force of a perfectly plastic body is obtained from the principle of virtual work.
This principle of statics is just the principle of virtual velocity (2.12) with negligible
acceleration. Thus for an increment of deformation oqfx in a collapse mechanism,
the difference between the work done by the actual and any statically admissible
force gives

where the inequality comes from the condition of positive plastic dissipation or
Drucker's postulate (1.31). Hence

f FioW;
r
dT ~
r
fFisoW; dT (2.23)

During any increment of displacement in a plastic collapse mechanism, the work

done by forces in equilibrium with a statically admissible stress field is a lower
bound for the work done by the collapse force Fi during this same increment of
displacement. Recall that statically admissible stresses satisfy equilibrium, yield
and boundary conditions, and notice that applications of this principle require
statically admissible stresses only in those regions or on surfaces that are plastically
deforming.
The complementary minimum principle considers work done by active forces
during any kinematically admissible increment of displacement OW;C. Any
kinematically admissible displacement field gives stresses Q~ and consequently,
an applied force Fi c that is in equilibrium with these stresses. The difference
between work done by force Fie and that done by the actual force Fi in any
kinematically admissible increment of displacement is given by the principle of
virtual work,

(2.24)

where the inequality comes from Drucker's postulate. Therefore,

fr FicoW;c ctr ~ fr Fi oW;c dT (2.25)

During any increment of displacement in a KA VF, the work done by forces Fi c

in balance with dynamically admissible stresses Q~ is an upper bound for the
work done by the collapse force Fi during this same increment of displacement.
The dynamically admissible stresses Q~ are obtained from the flow law and thus
satisfy yield in plastically deforming sections. Forces Fie are in balance with
these stresses for negligible acceleration but the stresses need not satisfy equations
of equilibrium in the body.
Equations (2.23) and (2.25) are a lower and an upper bound, respectively, for
the plastic collapse force. Additional details on bounds for strain hardening materials
can be found in Martin [1975].

2.4.2 Lower Bound on Dynamic Response Period

For a KA VF W;c (Xj' t), let the rate of dissipation by internal forces be if where
2.4 Bounds for Rigid-Perfectly Plastic Solids and Structures . 39

if = Is Q~q~ dS = LO'ijeij dV ~ LO'ijEij dV

Suppose the body is subjected to both an impulsive velocity field Wo; == W;( X j , 0),
that is suddenly applied at t = 0 and also to statically admissible active forces
t).
F/ (X j , These statically admissible forces prolong the response period but they
are sufficiently small so there is no increase in the kinetic energy of the body.
Thus the total dissipation occurring during the response period tf satisfies an
inequality that comes from the principle of virtual velocity (2.12) and Drucker's

tio
postulate
~ . 0
LSo
~

Soif dt ~ Fi W; dt dr - Pv ~ W;c dt dV (2.26)

Martin [1964, 1965] obtained a lower bound for the period of deformation tj by
(X
con.sidering a kinematically admissible field that is independent of time W;c j , t)
= ~W (Xj ).
It follows that the actual response period tf is greater than the lower
bound
(2.27)

This bound depends on the distribution of the assumed velocity field W;c (Xj )
but
is independent of the velocity magnitude. The interaction requires that the direction
cosines of the actual velocity field remain constant during the deformation period.
(X
The lower bound (2.27) does not require the traction Fi j , t) to be a separable
function of space and time (see Stronge [1983] and Kaliszky (1970]).

2.4.3 Upper Bound on Dynamic Response Period

The upper bound on duration of motion follows from properties of the principal
or fundamental mode for a structure. A mode is a velocity distribution that is an
extremal for the rate of change of kinetic energy of the structure. Among all
KA VFs that have the same kinetic energy K, the principal mode provides a
minimum for the rate of change of kinetic energy -dK/dt. Thus the principal
mode is the compatible velocity distribution that gives the longest duration of
response. An example of the method of obtaining the principal mode is given in
Sect. 2.5.5. Let the velocity in the principal mode be denoted by V*(t) l/>i(X j )
where V*(t) is the speed of some characteristic point in the structure. Then for
deformation in the principal mode of a rigid-perfectly plastic structure, the rate
of dissipation is a linear function of velocity and the rate of change of kinetic
energy is a constant. Consequently, in a mode form solution the velocity decreases
linearly with time,
(2.28)

where tj is the duration of motion for the modal solution. For tractions Fi that
are independent of time, deformation in the principal mode from any initial kinetic
energy K~ = Ko takes place during the period tj

Ko* = it; {i
0 0
L
* *
Qaqa(X, t) dX - f v FiV *(t) I/>i dV } dt (2.29a)
40 Principles of Mechanics

Equation (2.29a) together with (2.28) give

(2.29b)

r
where the negative of the initial rate of change of kinetic energy in the mode is

i; == Q:q:(X,O) dX - t Fj V'(O) (/J;(Xj) dT

The mode response period t;

is an upper bound on response duration t f since at
any kinetic energy K, the complete solution has a rate of change of kinetic energy
liol li;l.
~ The bound on the response period has proved useful in a displacement
bound for rate-dependent materials (see Symonds and Chon [1975]). This bound
together with the lower bound (2.27) bracket the response period; i.e. tj ::; t f ::; t; .
2.4.4 Lower Bound on Final Displacement

Since the true velocity and stress fields W;, Qa are dynamically admissible, they

r
can be used in the principle of virtual velocities (2.12) to give

Qaq~ dX = - fvPv W;W;c dV + I r Fj W;C dT (2.30)

where body forces gi are considered to be negligible. Let the true initial velocity
distribution be Woi == W;(X j , 0). In order to obtain a lower bound for components
of final displacement ..1fi == ~ax[ W;(X j , tf)]' consider a kinematically admissible
velocity distribution W~i == W;c (X j ' 0); i.e .

• C {W~i(l-t/tj), t < tj
W; = (2.31)
0, t > tj
where tj is a lower bound on the response time (2.27). The KA VF W;c gives
generalized strain-rates q~(Xj' t) and initial strain-rates q~a == q~(Xj' 0). With
this assumed velocity field, Eq. (2.30) can be integrated over the estimated response
period tj. After integration by parts and noting that W;c (X j ' tf) = 0 for t > tj
we obtain

tIo! FjW~i(l-
t

tltj) dtdT + fvPv WoiW~i dV

(2.32)

The lower bound results from two types of approximations:

1. In plastically deforming regions of the assumed deformation field where q~ > 0 ,
any difference between the true stress Qa and stress Q~ on the yield surface
introduces an 'error' that separates the bound from the true solution. This
stress difference is more significant in the early phase of deformation because
of the time-dependence of the assumed field, Eq. (2.31).
2. The mean-value theorem for integrals is used to replace the final displacement
field by the maximum displacement at the instant when motion ceases. This is
2.4 Bounds for Rigid-Perfectly Plastic Solids and Structures 41

assumed to be the largest final displacement since elastic unloading effects are
neglected. In general the error associated with this approximation increases
with the number of spatial coordinates required to specify the deformation.
This is called the geometric error.
With these approximations Eq. (2.32) gives a lower bound for the largest magnitude
of each component of final displacement L1 fi

L1fifvPvW~i dV ~ IvPvWiW~i dV ~ tj{frf:! FiW~'i(l-t/tj)dtdT

(2.33)

This bound was obtained by Morales [1972], Morales and Neville [1970] and
clarified by Wierzbicki [1972] who commented on its usefulness with an arbitrary
distribution of initial velocity. In order to obtain a bound for any component of
displacement Wi, that component must be included in the assumed initial velocity
field W~i' The largest lower bound is obtained for an assumed velocity and
coordinate directions that result in a maximum for

Iv Pv W~i Wi (Xj , tf) dV

By using the lower bound on response time tj, Stronge [1983] obtained a
corresponding lower bound for the maximum displacement,

f If
2 Pv W~i d V
1f {i FiW~i[I-2~)dtdr+fvPvWoiW~idV)
r o tf
(2.34)

v
Accuracy of this bound depends on the difference between the yield surface and
the true stresses in those parts of the assumed velocity field that are plastically
deforming. Accuracy of the bound also depends on the dimensionality of the
deformation field. This geometric error generally increases with the number of
coordinates needed to represent the deformation field (Stronge [1985]).

2.4.5 Upper Bound on Final Displacement

An upper bound for the final displacement at a point in section X is obtained
from a statically admissible stress field. Suppose the static collapse force F/
acting at section X of the structure is in equilibrium with the statically admissible
stress distribution Q~(Xj)' Then for the structure deforming in the true velocity
field ~(Xj' t), at any time the difference between the rates of work done by
statically admissible and actual applied forces Fi(Xj,t) results in
t t

foFiS~(X,t)dt = frfoFj~dtdr - Iv~ (W;W;-WoiWoi)dV

42 Principles of Mechanics

In this expression the last term depends on the difference between the dynamic
generalized stresses Qa and the statically admissible stress field. Note that this
term is positive for a rigid-perfectly plastic material. This stress difference'in the
actual plastically deforming regions qa > 0 influences the accuracy of this bound.
By neglecting this difference we obtain an upper bound Aft for the component of
final displacement a~ the point where F/ is applied. Finally, at time tf when the
velocity vanishes, W;(X j , t f) = 0, the final displacement at the point where F/ is

}
applied Aft = W;(X, t f) satisfies
+ _
Aft::;; Aft = - -
1
Fj'(X)
{J r~
rJo
'
FjW; dt dT + J Pv' .
-WoiWoi dV
v 2
(2.35)

The upp~r bound Aft was first obtained by Martin [1964] for impulsive loading
where Woi > 0 and Fj(X, t) = O. For this case the bound states that during the
dynamic response period tf , the work done by the static collapse force Fjs is less
than the initial kinetic energy. Ploch and Wierzbicki [1975] extended the bound
to incorporate large deformations. The complementary upper bound for WOi = 0
and Fj(X, t) > 0 has had limited application since it requires the complete solution
for the period of nonzero surface tractions. Robinson [1970] has obtained an
upper bound on final deflection of elastic-perfectly plastic structures but again,
this requires the complete solution if the load is not impulsive.
As mentioned, the accuracy of this bound on final displacement depends on
the difference between the assumed statically admissible stress distribution and
the actual state of stress in the deforming regions. This difference is largest when
significant energy is dissipated in a transient or 'moving hinge' phase of motion;
on the other hand, if the loading is impulsive and most of the initial kinetic
energy is dissipated in a stationary (modal) configuration, the upper bound on the
largest displacement is fairly accurate.

Example: Bounds for Impulsively Loaded Cantilever For impact or blast loading
the duration of loading (surface tractions) is often brief in comparison with the
time required for deformation waves to redistribute energy in the structure. This
loading can be represented as an initial velocity field imposed over part of the
structure; subsequently the surface tractions are negligible. This is termed impulsive
loading. In this case the energy imparted to the structure by the loading is equal
to the initial kinetic energy Ko =tJpvWoiWOi dV.
Consider a rigid-perfectly plastic cantilever of length L, with a cross-section
that has mass per unit length p and fully plastic moment Mp. There is a concentrated
mass G = rpL that is attached at the tip of the cantilever; the mass G may represent
a colliding body during the contact phase of motion. If the particle at the tip is
given an initial transverse velocity Yo' the kinetic energy is initially Ko = GVo2 12 .
A reasonable KA VF for estimating the dynamic response can be simple in-
plane rotation about a hinge at the root of the cantilever. For coordinates indicated
in Fig. 2.2, this gives
U C = -(1- X/L)Uj~ and if e
= -MP Bc

where 88 (1) = V(t)/ L denotes the rate of rotation at the root. It will be shown that
this velocity distribution also happens to be the primary mode of plastic
deformation, <l>i = (1- X / L). If the initial kinetic energy of this KA VF is equal to
the imposed kinetic energy GVo2 /2, the initial angular speed of the KA VF is
2.4 Bounds for Rigid-Perfectly Plastic Solids and Structures 43

'.0 _ _ _ ~p~bound_ _ _ _ _ _
..
",",' . . ' , .. ....... ', . . . . . . . . . .
" ' ' ,'

mass ratio r =G/pL

Fig.2.2 Complete solution and bounds for final tip displacement of impulsively loaded cantilever as functions
of the ratio of mass of striker at tip to mass of beam r = G / pL.

Consequently, the response period is within the range

1/2
GLVo ~ ti ~ GLVo [ 1+_1_]
A1p A1p 3y

The lower bound on final deflection at the tip AI is obtained from the limit on
i
response time t and the mode of deformation tPf:
_ fiVo GLV;y
L\j = 2+ y-l A1p {1+2Y)

The upper bound on final deflection at the tip is obtained by assuming that all the
initial kinetic energy is absorbed by deformation in the static collapse mechanism
of rotation about the root,
A+ _ GLVo2
Lli - - -
2A1p
The actual response of a rigid-plastic cantilever to impact at the tip involves a
transient phase of motion followed by deformation in the primary mode. The part
of the total kinetic energy dissipated in each of these phases depends on the mass
ratio y = G/pL. The analytical solution for final tip deflection AI can be expressed
as (see Sect. 4.5.4)

A = GLV; [_y_ + 2 Yln(1+ 2Y )]

I 6A1p 1+2y 2y
This solution lies between the bounds, AI::;; AI ~ A;. If the cantilever has
44 Principles of Mechanics

negligible mass in comparison with G, almost all of the energy is dissipated in

modal deformation and both bounds converge to the analytical solution as shown
in Fig. 2.2. For small mass ratios, however, the upper bound is not close to the
actual deflection.

2.5 Dynamic Modes of Deformation

Modal solutions are of fundamental importance for explaining and estimating the
deformation of structures that develop in response to impact or blast loading. For
either rigid-perfectly plastic or rigid-strain hardening structures that are given a
heavy blow over a segment of the structure, the deformation that develops divides
into an initial transient phase where the pattern or location of deformation is
continually changing and a modal phase where the pattern is constant. During the
transient phase the pattern of deformation steadily evolves from the initial velocity
distribution imposed at impact to a mode configuration. After attaining the velocity
distribution of a stable mode configuration, the pattern of deformation remains
constant for some period of time. In most cases a substantial part of the impact
energy is dissipated in a mode configuration during the second phase of deformation.

2.5.1 Modal Solutions

Here we consider a rigid-perfectly plastic structure subjected to an initial velocity
field w
o;( Xj) and time-independent ~urface tractions F;( X j ). Let W;( X j , be t)
the velocity field that develops from Wo; and F; . This velocity satisfies the field
equations; i.e. the laws of motion, compatibility and constitutive relations, in
addition to the initial condition W;(Xj' 0)= Wo;(X j ). In general, W; is not a
separable function of time and spatial coordinates.
A modal solution is a velocity field with separated functions for spatial and
temporal variables
(2.36)

which satisfies the field equations but not necessarily the initial condition. The
function I/J; is called the mode or mode shape; the scalar velocity V* of a
characteristic point is specified such that -1::::; I/J; : : ; 1. Notice that the modal velocity
field W;* and the associated acceleration field W;* have the same shape; i.e.
(2.37)
If no external forces act in the structure, mode shapes can be regarded as natural
modes of dynamic plastic deformation since they are a property of the structure
and independent of the initial velocity field. Modes exist also for structures
subjected to statically admissible time-independent surface tractions F;.

2.5.2 Properties of Modes

Impulsive loading on a rigid-plastic structure results in dynamic deformation
that continually evolves towards a modal solution.
2.5 Dynamic Modes of Deformation 45

Proof for this principle of convergence to a mode was given by Martin and Symonds
[1966]. Before convergence, the part of the initial kinetic energy dissipated in the
transient phase increases with the difference between the initial momentum
distribution and the mode shape. That is, convergence to the modal solution occurs
at an early stage of dynamic response if the initial momentum distribution is not
very different from a mode.
In order to find particular modes for a structure we employ the functional
J* (~c) for the ~ate of change of kinetic energy of a kinematically admissible
velocity field lW; i.e. the dissipation of power J* for the kinetic energy
f
K C = (Pv ~c~c 12) dV is given by,

(2.38)

This is known as Lee's functional. It is used to identify dynamic modes for a

rigid-plastic structure by means of the principle of minimum dissipation of specific
power:3

Among all kinematically admissible velocity fields with the same kinetic energy,
modal solutions will be extrema for the functional 1*.

These extrema can be stationary maxima, minima or saddle points of J*.

Nonstationary extrema, however, also occur in rigid-plastic structures; these extrema
occur at points where J* is nonanalytic so the extrema do not require 8J* = 0
(see Symonds and Wierzbicki [1975] or Palomby and Stronge [1988]). Usually
nonanalytic extrema correspond to nodes (i.e. plastic hinges) located at supports
or other points of displacement constraint where the bending moment is large but
shear force Q:t O.
There are multiple modes (in most cases an infinite number) for any structure
with continuously distributed mass. They can be numbered in the same manner as
elastic modes; for a rigid-plastic structure this number increases with the number
of plastic hinges. Stable modes are local minima for J* while maxima and saddle
points are unstable. If the momentum distribution is very close to that of an
unstable mode, initially the complete solution for the velocity field can converge
to the nearby unstable mode. Nevertheless, the final stage of deformation is always
in a stable mode unless the initial velocity distribution is identical to an unstable
mode (see Martin [1983] and Martin and Lloyd [1983]). Among the stable modes
there is one that gives the minimum specific dissipation of power J*; this basic
mechanism of rigid-plastic deformation is called the primary or fundamental mode.
To sum up, dynamic modes for rigid-plastic structures can be characterized as
follows:

I. Mode shapes are properties of both the structure and any statically admissible
steady load; thus modal velocity distributions ¢;
are kinematically admissible
and independent of time.
2. The velocity and acceleration fields of a modal solution are geometrically
similar.

'The kinematic extremal principle for identifying modes was presented by Lee and Martin [1970] and Lee
[1972]. Later Martin [1981] described a systematic method for finding modes of any rigid-plastic structure.
46 Principles of Mechanics

mode 1

mode 2

Fig. 2.3 First and second dynamic modes of deformation for rigid-perfectly plastic cantilever. The mode
number increases with the number of plastic hinges. For a mode, the location of these hinges Ai and their
relative rotation Pi give an extremum for Lee's functional J* (Ai' Pi)'

3. A modal solution has a rate of energy dissipation that increases linearly with
velocity; hence, during any period wherein time dependent forces are negligible,
the velocities decrease linearly with time.

Example: Dynamic Modes for Rigid-Plastic Cantilever The modes for a rigid-
plastic structure are found by considering a mechanism of deformation with
arbitrarily located plastic hinges; the number of degrees of freedom for the
mechanism are determined by the number of hinges. The location and relative
rates of rotation at these hinges are obtained from Lee's functional (2.38) evaluated
with the velocity field of the mechanism. For the uniform cantilever shown in
Fig. 2.3, the first mode is a mechanism with a single hinge at some distance AIL,
from the tip. The KA VF corresponding to a single hinge mechanism gives a
minimum value for the specific dissipation power 1* if the hinge is at the root of
the cantilever. This result is a non analytic extremal (i.e. dJI dAI *- 0) since the
solution is at the boundary of the admissible region, 0 < Al ~ 1.
The second mode has two hinges. In order to identify the second mode, the
shape parameters AI' A2 and {3 for the mechanism illustrated in Fig. 2.3 are
found that provide an extremal for ]*. The mechanism has hinges located at
X = AIL and X = AzL with rotation rates iJ and {3iJ, respectively. This mechanism
has a transverse velocity distribution Wc where

A]L<X
AzL<X<AI L
X<AzL
Consequently, for a uniform mass per unit length p, the kinetic energy K C of
the mechanism can be expressed as the sum of translational and rotational parts
for each of the two segments; i.e.

KC = PUiJ 2 {(A I -A2)3 +A(A -A +{3A2)2+{32A~}

2 3 2 I 2 2 12

Likewise, for a rigid-plastic beam with yield moment M p' the rate of dissipation
for the two hinge mechanism can be expressed as
2.5 Dynamic Modes of Deformation 47

Table 2.1 Dynamic modes for uniform cantilever

Mode Type of extremum ,1,1 ,1,2 ,1,3 /32 /33

Nonanalytic min. (stable) 1.0

2 Saddle point (unstable) 1.0 0.50 -2.00
3 Saddle point (unstable) 1.0 0.71 0.29 -1.41 2.0

. {/lM,6 fJ > 1
DC = (2 _ fJ)M/J fJ<l
These elements of the specific dissipation power J* = if I~ give an extremum
at a saddle point ..1,\ = 1, ..1,2 = 0.5 and fJ = - 2.0. There are no other analytic
extremes of J* for two hinge mechanisms. Shapes of the first, second and third
modes are listed in Table 2.1.4 The first mode has the smallest specific dissipation
power of any stable mode so it is the primary mode of deformation.

2.5.3 Mode Approximations for Structural Response to Impulsive

Loading
The dynamic deformation of an impulsively loaded rigid-plastic structure generally
involves an initial transient period of deformation where plastic 'hinges' travel
away from points where the initial velocity field is discontinuous; these velocity
discontinuities can be located at supports where displacement is constrained. During
the transient period the hinges move such that the velocity distribution evolves
towards a mode shape. After the velocity distribution converges to a mode, then
deformation continues in the mode configuration. The energy dissipated in the
modal phase of deformation can be a large part of the initial energy imparted to
the structure; this is especially true if the structure is light but it contains small
segments of heavy mass. In this case a modal solution can be a good approximation
for the dynamic deformation of the structure.

'Best' Mode for Approximating Dynamic Deformation A continuous structure

has a large number of dynamic modes of plastic deformation. The structure almost
always finally deforms in the primary mode; therefore it is tempting to use this
mode to approximate the structural response to impulsive loading. Lepik [1980]
showed however, that there are initial velocity distributions that are close to the
shape of higher modes (unstable extrema) which yield more accurate estimates
for final displacements of rigid-plastic beams. The higher mode shape is a better
choice for use in the modal approximation technique when most of the impact
energy is dissipated in this mode before the velocity distribution finally switches
to the primary mode. But what identifies the mode which best represents the
transient velocity field resulting from impulsive loading?

4In obtaining hinge locations for a higher mode, it is helpful to recognize that no shear force is present at
perfectly plastic hinges located away from displacement constraints. If a mode is calculated for no external
force acting on the structure and no shear force is present at the hinges. then there is no acceleration of
the center-of-mass of a rigid segment. In a modal velocity field, each interior segment simply rotates
about it's own center-of-mass.
48 Principles of Mechanics

Symonds [1980] examined several possible criteria for identifying the mode
that gives the best approximation for dynamic deformation of impulsively loaded
rigid-plastic structures. He stated the following conjecture for selecting the mode:

Generally the most accurate modal approximation for final shape of the deformed
structure is given by the mode that gives the largest lower bound for dynamic
response period.

The bound on response period is expressed by Eq. (2.27); for impulsive loading
it depends on the difference between the initial momentum distribution and the
mode shape.
Slutsky et al. [1982] tested this conjecture for impulsively loaded beams and
portal frames. They speculated that the largest lower bound on response time is
also the criterion that gives the best mode for structures composed of rigid-
viscoplastic or homogeneous viscoplastic materials.

Initial Velocity of Modal Solution A measure of the difference between the

initial ~odal velocity field vo*q,;(Xj) al!.d the impulsively applied velocity
Woi == W;(Xj,O) is given by the functional ,10'

Since both the modal approximation and the actual velocity field are comp!ete
solutions (for different but 'equivalent' initial conditions), the difference d(t)
between the modal approximation and the actual velocities is continually decreasing,
d Lil d t ~ O. The initial difference Lio results from the mode shape not being
geometrically similar to the initial velocity distribution. The initial speed of the
modal solution V; that minimizes this difference is given by dLio/dV; = 0; thus

v*o

This initial speed for the modal velocity gives an initial kinetic energy for the
modal solution that is less than that of the imposed velocity field, K: ~ Ko'
Nevertheless, the approximate and actual velocities are often identical during the
final phase of motion if the 'best' mode is the primary mode. This method of
obtaining the initial modal velocity results in a best estimate rather than an upper
or lower bound for the deformation. Alternatively, if the modal field is given the
actual initial kinetic energy K~ = Ko, the approximation gives an upper bound on
the final deflection ,1;.

References

Calladine, C. [1985). Analysis of large plastic deformation in shell structures. Inelastic Behaviour of
Plates and Shells (ed. L. Gevilacqua, R. Feijoo et al.) IUTAM Sym., Rio de Janeiro. 69-101.
References 49

Kaliszky, S. [1970]. Approximate solutions for dynamically loaded inelastic structures and continua. Int.
J. Nonlinear Meeh. 5, 143-158.
Kaliszky, S. [1985]. Dynamic plastic response of structures. Plasticity Today (ed. A. Sawczuk and G.
Bianchi), Elsevier, London. 787-820.
Lee, L.S.S. [1972]. Mode responses of dynamically loaded structures. J. Appl. Meeh. 39, 904-910.
Lee, L.S.S. and Martin, J.B. [1970]. Approximate solutions of impulsively loaded structures of a rate
sensitive material. 1. Appl. Math. Phys. (ZAMP) 21, 1011-1032.
Lepik, U. [1980]. On the dynamic response of rigid-plastic beams. J. Struet. Meeh. 8, 227-235.
Martin, J.B. [1964]. Impulsive loading theorems for rigid-plastic continua. ASCE J. Engng. Meeh. Div.
90,27-42.
Martin, J.B. [1965]. Displacement bound principle for inelastic continua subjected to certain classes of
dynamic loading. J. Appl. Meeh. 31, 1-6.
Martin, J.B. [1972]. Extremum principles for a class of dynamic rigid-plastic problems. Int. J. Solids
Structures 8, 1185-1204.
Martin, J.B. [1975]. Plasticity: Fundamentals and General Problems. M.I.T. Press, 386-403.
Martin, J .B. [1981]. The determination of mode shapes for dynamically loaded rigid-plastic structures.
Meeeaniea 16, 42-46.
Martin, J .B. [1983]. Convergence to mode form solutions in impulsively loaded piecewise linear rigid-
plastic structures. Int. J. Impact Engng. 1, 125-142.
Martin. J.B. and Lloyd, A.R. [1983]. Convergence to higher symmetric modes in impulsively loaded
rigid plastic structures. Int. J. Impact Engng. 1, 143-156.
Martin, J.B. and Symonds, P.S. [1966]. Mode approximations for impulsively loaded rigid-plastic
structures. ASCE J. Engng. Meeh. Div. (EM5), 92, 43-46.
Morales, W.1. [1972]. Displacement bounds for blast loaded structures. ASCE J. Engng. Meeh. Div.
(EM4) 98, 965-974.
Morales, W.1. and Neville, G.E. [1970]. Lower bounds on deformations of dynamically loaded rigid-
plastic continua. AIAA J. 8, 2043-2046.
Palomby, C. and Stronge, W.J. [1988]. Evolutionary modes for large deflections of dynamically loaded
rigid-plastic structures. Meeh. Struet:Maeh. 16(1),53-80.
Pars, A. [1965]. Treatise on Analytical Dynamics Heinemann, London.
Ploch, J. and Wierzbicki, T. [1975]. On an extremum principle for mode form solutions of dynamically
loaded continua and structures. Int. J. Solids Structures 17, 630-640.
Robinson, D.N. [1970]. Displacement bound principle for elastic-plastic structures subjected to blast
loading. J. Meeh. Phys. Solids 18, 65-80.
Slutsky, S., Chon, C.T. and Yeung, K.S. [1982]. On the 'best' mode form in the mode approximation
technique using the finite element method. J. Struet. Meeh. 10, 117-131.
Stronge, W.1. [1983]. Lower bounds to large displacements of impulsively loaded rigid-plastic structures.
Int. 1. Solids Structures 19, 1049-1063.
Stronge, W.1. [1985]. Accuracy of bounds on plastic deformation for dynamically loaded plates and
shells. Int. J. Solids Structures 27, 97-104.
Symonds, P.S. [1980]. The optimal mode in the mode approximation technique. Mech. Res. Comm, 7,
1-6.
Symonds, P.S. [1984]. Twenty years of developments in inelastic dynamics. ASCE Conference.
Symonds, P.S. and Chon, C.T. [1975]. Bounds for finite dellections of impulsively loaded structures with
time dependent plastic behaviour. Int. J. Solids Structures 11, 403-423.
Symonds, P.S. and Wierzbicki, T. [1975]. On an extremum principle for mode form solutions in plastic
structural dynamics. J. Appl. Meeh. 42,630-635.
Tamuzh, V.P. [1962]. On a minimum principle in the dynamics on rigid-plastic bodies. Prikladnaya
Matematika Mekhanika 26, 715-722.
Van, The Vu and Sawczuk, A. [1983]. Lower bounds to large displacements of impulsively loaded
plastically orthotropic structures. Int. J. Solids Struet. 19, 189-205.
Wierzbicki, T. [1972]. Comment on lower bounds on deformations of dynamically loaded rigid-plastic
continua. A.I.A.A. J. 10, 364--365.
Chapter 3

Static Deflection

Flexural deformation in a slender structural element depends on the axial distribution

of the bending moment and the flexural rigidity. Here a basic quasi static loading
on a uniform cantilever will be used to illustrate effects of nonlinear material
behavior in plastically deforming structures. A cantilever with a transverse force
at the tip has a bending moment that increases linearly with distance from the tip.
The cantilever is wholly elastic and has linear behavior if the largest bending
moment (at the root) is less than the yield moment. However, if the force is large
enough there are sections in an elastic cantilever where the moment would be
larger than the yield moment My. At these sections the outer fibers of the cantilever
are stretched or compressed beyond the yield strain ey. In an elastic-perfectly
plastic cantilever the stress in these fibers is limited to the yield stress Y;
consequently, the flexural rigidity of the section decreases with increasing curvature.
Plastic deformation in the outer fibers is irreversible, so sections that are partly
plastic have been permanently deformed; if these sections are unloaded they do
not return to their initial curvature.
This chapter focuses on the development of plasticity in a simple structure
composed of elastic-plastic material when the structure is subjected to slowly
increasing load; the effects of elastic and plastic deformation on deflections of
the structure are described.

3.1 Small Elastic-Plastic Deflections

In this section, small static deflections of a cantilever are analyzed. Changes in

inclination of sections are assumed to remain negligibly small; consequently, the
distance of each section from the line of action of any transverse load remains
constant. Also, the bending moment does not change due to rotation () of the
neutral axis. Mathematically this implies cos(} '" 1 and sin(} '" (). The simplest
loading case is a transverse force F applied at the tip of a uniform cantilever of
length L, as shown in Fig. 3.1. Deflections of sections of the cantilever can be
determined as a function of an axial coordinate X measured from the tip. Transverse
deflection W(X) of the neutral axis depends on the force and properties of the
52 Static Deflection

Fig. 3.1. Transverse force F at tip of cantilever.

cantilever. At any section the shear force Q(X) and the bending moment M(X) are
given by
Q(X) = - F (3.1)

M(X) = FX (3.2)
The positive sense of these variables is indicated in Fig. 3.1. With this load the
shear force is uniform and the bending moment increases linearly with distance
from the tip so the largest moment occurs at the root where X = L.
In order to demonstrate the development of rate-independent plasticity with
increasing load, it is useful to relate the applied force F to the largest force that
can be supported by the structure if the material is perfectly plastic; this largest
force is termed the perfectly plastic collapse force Fe' Plastic collapse of a structure
occurs if there is a sufficient number of fully plastic hinges for the structure to
become a mechanism; at each plastic hinge the magnitude of the bending moment
equals the fully plastic bending moment M p' For a cantilever, a single plastic
hinge at any location results in a mechanism. With a transverse force applied at
the tip of a uniform cantilever, the plastic collapse force Fe is the force magnitude
that results in the moment M p at the root; i.e.
Mp
~. =L (3.3)

Recollect from Table 1.1 that the ratios of fully plastic to yield moments for
rectangular and circular cross-sections are 1.5 and 1.7, respectively. Since this
transverse force results in only a single hinge at the root of the cantilever, the
ratio of plastic collapse force to yield force is Fe / Fy = M p / My = I/Jm, where I/Jm
is the shape factor of the cross-section for bending defined in Eq. (1.11). Hence,
this ratio of forces is a structural characteristic that depends on the shape of
cross-section.

3.1.1 Elastic Deflections

For small inclination of the neutral axis, the curvature is given by I( = I((X) "=

d 2 W / dX2; if 11(1 $; I(y the outer fiber of the section remains elastic and

d 2W M
--"=1(=-- (3.4)
dX 2 E10
where E10 is the elastic flexural rigidity. By using the moment distribution specified
by Eq. (3.2), the differential equation can be written as
3.1 Small Elastic-Plastic Deflections 53

(3.5)

and the boundary conditions of zero displacement and inclination at the root
provide
W(L) = dW(L) = 0 (3.6)
dX
At this point it is convenient to cast the problem in terms of a set of nondimensional
variables and structural parameters which are defined as follows:
X W F F~
x =- w = w(x) = -L' f =- and 11 = _c_ (3.7)
L' Fc E10
In general the flexibility parameter 11 is given by 11 = (Mpl My )L/(y = ¢mL/(y; for
a rectangular cross-section with depth h, 11 = 3L/(yl2 = 3(YI E)(L/h). For forces
that are similar in magnitude to Fe the limitation of this analysis to small
inclinations implies that the flexibility parameter is in the range 0 < 11 < 1; this
parameter approaches unity only if the cantilever is extremely slender.
For compactness, denote differentiation with respect to the spatial variable by
(') == d( )/dx. Thus, using nondimensional variables the differential equation (3.5)
can be written as
w" = fJlX (3.8)
If the cantilever is entirely elastic the elastic region extends to the root at xl L = 1
where displacement and rotation vanish, w(l) = w'(l) = O. Hence we obtain

w' = fl; (x 2 -1), 0:5x:51 (3.9a)

w = fl1 (x 3 - 3x + 2), 0:5x:51 (3.9b)

6
In general, however, the elastic region will extend through only a segment near
the tip of the cantilever. Suppose the cantilever is elastic in a region 0:5 x :5 X
and that at the limit of this region the neutral axis has inclination w'(X) and
deflection w(X). Then the inclination and deflection in the elastic region can be
expressed as

w'(x) = fl1(x 2 _X 2 )+w'(X), O:5x:5X (3. lOa)

2

O:5x:5X (3.10b)

The length X of the elastic region depends on the tip force f and the flexibility of
the section; thus the value of X is determined by the curvature that causes yield
M
X = -y- :5 1 (3.11)
fMp

When f = 1 (i.e. F = Fe)' the length of the elastic region reaches its maximum
value, Xmax = l/¢m, which depends on the cross-section only. For instance,
Xmax = 2/3 and 0.589 for beams of rectangular and circular cross-sections,
respectively.
54 Static Deflection

:t::Cl.
"- 10
:t::
C
'0"
E
E 0·5
Q"J
C
U
C

'"
Ll
O~------____JI____________~I
o 0·5 10
distance from tip, X/L

Fig. 3.2. Bending moment distribution in cantilever with rectangular cross-section for initial
yielding - - - -, and fully plastic collapse - - .

3.1.2 Deflection of Elastic-Perfectly Plastic Cantilever

If the curvature of the loaded cantilever exceeds the yield curvature in a segment
near the root, the moment-curvature relation (3.4) is not valid. In this case, the
appropriate elastic-plastic moment-curvature relation depends on the cross-section
of the cantilever. The analysis in this section is based on a rectangular cross-
section. Given a rectangular cross-section, the curvature is related to the bending
moment by the relation

~
K"y
= (3-2 M
My
)-"2 (3.12)

With a transverse tip force, the bending moment increases linearly with distance
from the tip, M = FX, as shown in Fig. 3.2. For a cantilever of rectangular cross-
section Fe = 3M y l2L and Eq. (3.12) gives
-1/2
d2~ = K"y [ 3(1- FX )]
dX FeL

Since LK"y = 2tL13, this equation for elastoplastic curvature has a nondimensional
form
2
--:'O:x:'O:1 (3.13)
3f
For boundary conditions of vanishing deflection and inclination at the fixed end,
w(1) = w'(1) = 0, integration of (3.13) gives

w'(x) = ~[(I-ft2-(1-fx/,2] (3.14a)

3{3f

w(x) = 9';;f 2 [(3fx - f - 2)(1- f)"2 + 2(1- fx)3/2] (3.14b)

This solution applies to the elastic-plastic region at the root; the region terminates
at x = X where plasticity vanishes. For a rectangular cross-section Eq. (3.11)
gives X = 2/3f·
3.1 Small Elastic-Plastic Deflections 55

Section A-A

A
--;;,-;:1:----="I::-----:::J..,;',---------::-l'L,----L' _ X;,
0·6 0·7 0·8 O·g '.0 L

Fig. 3.3. Plastically deforming region at root of an elastic-perfectly plastic cantilever in case of F f Fe =
1.0 - - ; and Ff Fe =0.96 - ---.

The inclination and deflection of the neutral axis at the interface between
elastic and elastic-plastic regions are obtained from Eqs (3.l4a) and (3.l4b); i.e.

w'(X) = ~[(l- 1)112 __

3-fi1 -fi
1] (3.l5a)

w(X) = 91'/ 2 [31- 1(1- 1)112 ] (3.l5b)

For tip forces F S; 2F;; /3, the bending moment M(X) results in curvatures that are
elastic at every section and deflections that are given by Eq. (3.9b). For values of
force in the range 2F;; /3 < F < F;;, the curvature near the root exceeds the yield
curvature K' y throughout a segment of length (1- X)L. The stress distribution in
this elastic-plastic segment is illustrated in Fig. 3.3. In this segment the outer
fibers of the beam are strained beyond yield while the central core remains elastic.
Figure 3.4a shows the curvature distributions for the yield force 1 = 2/3, and
several larger loads that cause elastic-plastic behavior in a segment near the root.
For the largest force displayed in Fig. 3.4, F = 0.96 Fe' the plastic region has
spread through 65% of the thickness at the root where the moment M(L) =0.96 M p;
only the central core of the cross-section remains elastic. The increasing curvature
in the elastic-plastic region near the root is not especially large until the tip force
is almost equal to the plastic collapse force F;;; consequently after the elastic-
plastic segment develops, changes in deflection with increasing force are only
slightly more rapid; this is shown in Fig. 3.4b.

3.1.3 Deflection of Elastic-Linear Strain Hardening Cantilever

Most materials that yield at a specific state of stress require an increase in stress
to increase plastic strain after yield commences; this material behavior is termed
plastic strain hardening. The stress strain relation for an elastic-linearly strain
hardening material is illustrated in Fig 1.12. Heavily loaded structures composed
of strain hardening materials develop sections where the bending moment exceeds
the yield moment My; in these sections the strains are partly elastic and partly
56 Static Deflection

VI

~
-E 1-0
>
:oJ
U

I a
05 10
distance from tip. X IL
;} 0
~ FIFe = 0,67
B
3:
. 02
c
Q
.----::: FIFe = 083
U
.!"
a; 0,4
~~~FlFe = 0,92
'0 j/ Fife = 096
'/
06 0 0,5 1,0 b
distance from tip, xlL

Fig. 3.4. Elastic-perfectly plastic cantilever: (a) distribution of curvature; (b) transverse deflection for
forces Fy5, F 5, Fe.

plastic. If the material strain hardens, the elastic-plastic sections are somewhat
stiffer than the elastic-perfectly plastic representation in Sect. 3.1.2 would suggest.
So what are the conditions wherein the simpler elastic-perfectly plastic analysis
is sufficiently accurate? In this section we compare curvature and deflection
resulting from this perfectly plastic approximation with results obtained for a
linearly strain hardening material and find that accuracy of the approximation
depends on both the bending moment distribution and the load magnitude in
comparison with the characteristic plastic load F;;. (Note that a linearly hardening
material has no ultimate moment Mp for a section; nevertheless it is convenient
to define a characteristic load F;; = if>mMyl L. For a rectangular cross-section
Fe = 3My I2L.)
In a structure composed of a material with elastic modulus E, yield strain
lOy = Y 1E and plastic strain hardening modulus Et = aE, there is a relationship
between bending moment M and section curvature /( given as

~
My
= ~[3_(~)2]+a~
2 /( /(y
(3.16)

Here the curvature at yield /(y = 2lOy 1h results in a yield moment My = E1o/(Y'
Equati~n (3.16) can be transformed to a nondimensional form by letting iii == M 1My
and k = /(I/(y. For a transverse force at the tip of a cantilever, iii =
(FXif>m)/(FeL) = if>mfx. Hence for a rectangular cross-section where if>m = 3/2,
Eq. (3.16) can be rearranged as
3.1 Small Elastic-Plastic Deflections 57

Fe

(1
~~I~__~~____~I----~____~I ~X/[
0·6 07 0·8 O·g 1·0

Fig. 3.5. Extent of plastically deforming region at root of elastic-linearly strain hardening cantilever for tip
forceF= Fe and strain hardening coefficient a. =0----, a. =O.I----and a. =0.3----.

2(YP+3(I-a-jx)P-(1-a) = 0 (3.17)
The cubic polynomial (3.17) has one real root for curvature k if the section is
within the elastic-plastic segment X < x < 1 and the force j '" 1. This root is
found by letting

Co =~'
(a -1)
Cl =- (Co + 2ajx)2 ' C2 Co _(co + jx )3
2 2a
Then the nondimensional curvature k is given by

k = (-C2+~c~+cir3 + (-c2-~ci+cir3 + Co + :: (3.18)

The bending moment Iii increases linearly with distance from the tip where
force F is applied. For F> 2Fc 13 there is a segment at the root where strain in
the outer fibers exceeds the yield strain; the length of this elastic-plastic segment
(1 - X)L is independent of the hardening coefficient a. On sections within the
elastic-plastic segment, however, the stress distribution does depend on the
hardening coefficient a. Since the stiffness of the section increases with strain
hardening, the height of the elastic core of any section increases with a ; this
corresponds to less curvature in the elastic-plastic segment with increasing
hardening coefficient a. These differences in the extent of the plastically deforming
region are shown in Fig. 3.5.
For any load, a larger strain hardening coefficient a results in larger stiffness
and consequently less curvature in the elastic-plastic region. Figure 3.6a shows
the influence of hardening coefficient a on the curvature distribution in the
cantilever for three force magnitudes j = 2/3, 1.0 and 1.2. The forces considered
are limited because the present analysis is restricted to solutions where rotations
of sections remain small. There is no ultimate force for the small deflection
analysis of a structure composed of linearly strain hardening material.
Inclination and deflection of the neutral axis for the loaded cantilever can be
obtained by integration of the curvature. Hence, for a rectangular section where X
=2/3j, the elastic-plastic segment of length (l - X)L has inclination w' = dwl dx =
dW 1dX as follows
w'(x) = -L/(d~k(X)dX, X:O;x:O;l (3.19)
58 Static Deflection

3-~

,F
~
>-
I
~
~ 2-D
~
::J
c;
>
'-
::J
tJ
1-0

a
I
0-5 1-0

00
>-
~

'8.....
N

0-2
~

c
Q
ti
~
a;
"0 06
b

Fig. 3.6. Elastic-linearly strain hardening cantilever: (a) curvature and (b) transverse deflection for forces f =
FI Fe and hardening coefficient c( = 0.05 - - , c( = 0.1 - - - - and C( = 0.3 - ' - .

(3.20)

The deflections shown in Fig. 3.6b for f = 1.0 and 1.2 were calculated by
numerically integrating the curvature and inclination between the root (x = 1) and
a section located at x. The integrals satisfy boundary conditions at the root,
w(l) = w'(l) = O. At the transition section with the elastic region, (3.19) and (3.20)
provide the interface conditions w'(X) and w(X); these values are used in Eqs
(3.10) to obtain the inclination and deflection for the elastic segment. It can be
seen that the deflections shown in Fig. 3.6b are insensitive to the hardening
coefficient a if F :s: F',; .
The tip deflection W(O)=Wo in Fig. 3.7 also illustrates this point. The tip deflec-
tion begins to vary from the elastic solution only if the applied force F is almost
as large as F',;. This is a consequence of the linearly increasing bending moment
distribution; with this moment distribution the large plastic deformations are
localized to a short segment at the root of the cantilever.

3.1.4 Residual Deflection After Elastic Unloading

If load is slowly (quasistatically) applied to an unstressed structure and then later

slowly removed, as force decreases during unloading the moment at every section
also decreases. The moment-curvature relation for unloading is simply the linear
3.1 Small Elastic-Plastic Deflections 59

... 0·75
~
N
-.J
........
§
~
c- 05
0
:;::
u
~
Q;
"0
a.

0'5 1·0 1·2

tiP force. FIFe = f

Fig. 3.7 Tip deflection Wo of elastic-linearly strain hardening cantilever as a function of transverse force F.

elastic M-K relation unless the change of the moment at a section is larger than
2M y. If there is significant strain hardening and the moment is large, unloading
can cause the outer fibers of the section to be strained into the range of reverse
plastic flow. Reverse plastic flow during unloading can occur only if strain
hardening is kinematic; i.e. in materials that exhibit the Bauschinger effect.
During unloading the bending moment at every section is reduced and finally
vanishes. In elastic segments of the structure this unloading is easily accomplished
by removing the elastic curvature that developed as a result of the load. For an
initially straight cantilever, all sections that remain elastic are thereby restored to
a straight configuration during unloading. In sections where the outer fibers are
plastically deformed during loading however, unloading leaves behind a final or
residual distribution of stress and a residual curvature K j (X). This final curvature
is determined by the curvature K. (X) under largest load and the bending moment
M.(X) that is in equilibrium with the largest magnitude of the applied load. In
order to obtain the residual curvature K j , the elastic curvature due to M. is
subtracted from the curvature under load K.,

(3.21)

In terms of the nondimensional curvature I( = K/Ky and bending moment

m= M / My this gives

(3.22)
For an elastoplastic section where the moment--curvature relation is given by Eq.
(3.16), the bending moments during loading and unloading can be equated to find
the residual curvature; thus from Eqs (3.16) and (3.22),

I(.-I(j = al(.+(I-a)(3-k.-2 )12

Hence throughout the cantilever the residual curvature satisfies
60 Static Deflection

0, (3.23a)

(1- a)(2k. - 3 + k.- 2 )12, (3.23b)

The nondimensional largest curvature k. is obtained from Eq. (3.18). For an
elastic-perfectly plastic cantilever, a = 0 and Eq. (3.12) gives k.- 2 = 3 - 2m. =
3(1 -Ix); so the residual curvature distribution in the elastic-plastic region (x >
2/31) is given by
1
--r====- m. (3.24a)
~3-2m.
or
[3(1- lx)r1l2 - 3fx/2, 2/31<x5,1 (3.24b)
However, outside the elastic-plastic segment near the root, x 5, 2/31 and there is
no residual curvature. Nevertheless, the elastic segment away from the root can
have inclination, and its deflection increases linearly with distance from the root.
The residual inclination and deflection at any section can be obtained by
integrating the residual curvature. For the cantilever with boundary conditions
w(l) = w'(l) = 0, the inclination of a section located at x == XIL is

wj{x) = - L/(y s: kj(x)dX (3.25)

This residual inclination wj only varies in sections of the cantilever where x >
2/31; i.e. w'(x);:: w'(2/3f) for x < 2/3j.
In the plastically deformed segment near the root of an elastic-perfectly plastic
cantilever, using Eqs (3.24b) and (3.25) it can be shown that

(3.26)

At the transition section from elastic to elastic-plastic behavior the inclination or

slope is

wj(2/3f) = L/(y( ~ ~l- 1

-v31
+ 31
4
-~)
1
(3.27)

The nondimensional final deflection of the plastically deformed segment near

the root is given by

Wj(x) =- s: w;(x) dX (3.28)

After using (3.26) as the integrand, Eq. (3.28) gives

Wj(x) = -L/(YL~12 [(1- lx)3/2 -(1- 1)3/2]

__ 2_(1-x)~1-1 - 1(2-3X+X 3 )} (3.29)
-J31 4
At the end of this segment (x = 2/3j) where curvature /( = /( y, the residual deflection
is

271
-J3+ f)2
w j(2/3f) = L/(y [ - -22 (1 + 6-J3 ) - 2(2
3 31
.J1=7 - -(1-
1
2
1)] (3.30)
3.1 Small Elastic-Plastic Deflections 61

~
-E lOr-----
>
~

OJ
U

o
~ 0-5
III
~

°1~.0~~-L--~~---~--~-
2·0
max. bending moment, M/My

Fig, 3,8, Residual curvature of elastic-linearly strain hardening beams of rectangular cross-section depends on
the largest bending moment M(X) at the section.

So every section of the cantilever has flexure that increases as the largest moment
m* at that section increases; the flexure later springs back elastically as the
moment is reduced during unloading. The residual curvature kf for elastic-plastic,
linearly strain hardening beams with rectangular cross-section is shown in Fig.
3.8. The curve shown for perfectly plastic material a = 0 is given by expression
(3.24). For materials with linear strain hardening (a> 0), the results were obtained
by Johnson and Yu [1981]. These authors also examined spring back in elastic-
plastic beams and plates with power law strain hardening.

3.1.5 Elastoplastic Beam-Columns

An oblique force acting on a slender member has both tangential and transverse
components of applied force. The angle between the force and the normal to the
neutral axis in the initial configuration is termed the angle of obliquity. Here it is
considered that the direction of this force does not vary with deflection; the
direction is fixed relative to the initial configuration. This combined loading results
in a beam-column problem; if the angle of obliquity is large this problem enters
the realm of buckling of eccentrically loaded columns.

Uniform Elastic Cantilever The deflection W(X) of a cantilever due to both

vertical force Fz and horizontal force Fx applied at the tip is illustrated in Fig.
3.9. The curvature d 2W/dX2 at any elastic section is directly related to the bending
moment M(X)
2
E10 ~x~ = M = ~x +FAL1- W) (3.31 )

where L1 == W(D). Let Q)2 == Fx / E10 and the ratio r of the vertical to horizontal
force is r == Fz / Fx ; thus
2
d W
--+Q) 2W
dX 2
62 Static Deflection

0'5~ I_ --I

~
"<l
0·4 ~
(z
f
'0 0.3
r::,'

~
~'" 0·2
{;
J
01

0·25 0·5 0·75 1·0

transverse load, FLz/EI

Fig. 3.9. Tip deflection of uniform elastic beam-column with horizontal force Fx and transverse (vertical)
force Fz .

This ordinary differential equation has a solution

W = Ll + rX + AsincoX + BcoscoX
which satisfies the boundary conditions W(L) = dW(L)/ dX = O. Hence the deflec-
tion is
W = LJ+rA X rsincoX (3.32)
- , r>O
co cos coL
where
Ll = ~(tancoL-coL),
co
For small deflections, the deflection of the tip of an elastic cantilever increases
linearly with vertical (transverse) force. A small additional horizontal force
somewhat increases the deflection; however the increase is not large if r > I and
the largest moment M(L) is substantially less than My. An upper bound on the
range of applicability of this elastic theory is provided by the load that results in
the limiting curvature K'y for yield at the root

r2 ] F L2 2LE
[ 1+-(tancoL-coL) _z_ < LK'y = __ v (3.33)
coL E10 h

For Ey = 0.002 we obtain LK'y < 0.004L1 h. Another upper bound is provided by
the Euler buckling load for an elastic column obtained with Fz = 0; i.e.
F L2 Jr2
_x_ < (3.34)
E10 4
It may be noticed that although the above analysis has considered the additional
bending moment which is produced by the axial (tangential) force and the tip
deflection, it still employs the approximate expression K' d 2W/dX 2 for the Z

curvature, which is valid under the condition (dW/dX)2« 1. Hence, this analysis
is applicable only when the changes in geometry of the beam-column are not too
3.2 Large Elastic-Plastic Deflections 63

large. Large geometrical changes will be fully taken into account in the following
section.

3.2 Large Elastic-Plastic Deflections

In deriving equilibrium equations used in the preceding sections of this chapter,

the inclination of the neutral axis for bending has been assumed to remain small.
In this section, effects of large deflections that involve large rotations from the
initial configuration are considered for an elastic-plastic cantilever subjected to a
transverse concentrated force F at the tip. Because the deflection is no longer
required to be small in comparison with the length of the cantilever, the approximate
expression ]( '" d 2W/dX 2 will be replaced by an exact expression for the curvature
of the neutral axis of a deformed cantilever; that is,
dB d 2 W/dX2
]( = - = (3.35)
[I + (dW/dX) 2]
3/2
cIS
The problem of obtaining deformations will be reformulated using this expression
for curvature.

3.2.1 Elastica: Large Elastic Deflection

When M:::; My , then ](:::;](y is satisfied at every cross-section and the cantilever
behaves elastically. With Eq. (3.35) in mind, the moment-curvature relation (3.5)
can be expressed as
dB M
(3.36)
cIS E10
Consider a coordinate system XOZ with the origin 0 located at the tip A of the
undeformed cantilever, as shown in Fig. 3.10. The X axis is along the initial
neutral axis of the cantilever and the Z axis is transverse to the neutral axis and
in the plane of deflection. Let S denote the arc length of the deformed cantilever
measured from the tip A and let e(S) = tan- 1 (dW/dX) denote the inclination angle

-x

w
Fig. 3.10 Large elastic deflection of a uniform cantilever subjected to a concentrated force F at the tip.
64 Static Deflection

measured clockwise from the X axis. Note that if the cantilever is loaded at the
e
tip by a force F in the positive Z direction, is negative at the tip A and gradually
increases to zero at root B.
Introduce nondimensional variables
X W S Fr
W = L'
x =- s = L' Ilf = E1 (3.37)
L' o
where f is nondimensional force and 11 is the flexibility parameter defined in
(3.7). Since the bending moment distribution is M = F(X - XA), the moment-
curvature relation given in Eq. (3.36) results in
dB
ct; = Ilf(x - x A ) (3.38)

Kinematic relationships give

dx dw
- = cosO, = sinO (3.39)
ds ds
Equations (3.38) and (3.39) form a system of first-order ordinary differential
equations for the unknown functions O(s), xes) and w(s). Equations of this type
that describe large elastic deflections of flexible bars were first studied by Euler
[l744] and termed elastica (e.g. see Frisch-Fay [1962] and Wang [1981]).
The differential equation for the elastica is obtained by differentiating (3.38)
with respect to s and then using (3.39) to obtain
dZO
=

r r1
-Z IlfcosO (3.40)
ds

Because ! [(: 1= 2: ( ~~ ). Eq. (3.40) leads to d[( : = 21lf cosOdO.

It follows that
dB
(3.41 )
ds
where 0A denotes the inclination angle at A, (0 A < 0). This expression satisfies
the boundary condition that requires the curvature to vanish at the tip.
Assuming that the total length of the neutral axis does not change during

r
flexural deformation, and noting that 0 = 0 at root B where s = 1, we have

0A
(dO
ds
)-1 dB = fds = 1
0
(3.42)

Combining Eqs (3.41) and (3.42) results in

~21lf = fO 1 liZ dB (3.43)

JOA (sinO-sinOA )
In order to bring the right-hand side of Eq. (3.43) to the standard form of elliptic
integrals, a new variable ¢ is introduced that satisfies the equation
1- sinO = (1- sinO A )sinz ¢ (3.44)
with 0:-::; ¢ :-: ; n12. Furthermore, define a parameter p such that
pZ "" (1-sinOA )/2 (3.45)
3.2 Large Elastic-Plastic Deflections 65

In Eq. (3.43) we change from variable 0 to I{J, eventually to obtain

{iil = f "/2

if'B
(1
-
d¢
2·
P sm
2¢)1I2 = K(p)-F(p'¢B) (3.46)

where the value of I{J at root B (i.e. where 0 = 0) is I{JB sin-I (lI-J2"p) . The
terms on the right-hand side of Eq. (3.46) are an incomplete elliptic integral of
first kind, F(p,l{J) = jt(1- p2 sin 2 ¢)-1I2 d¢ and the complete elliptic integral of
=
first kind, K(p) F(p,n/2). Thus Eq. (3.46) has one unknown variable - the
modulus p; this can be found by trial and error with help of the tables of elliptic
integrals, provided the value of J1f = FL2/E1o is specified.
In order to obtain the deflection of the cantilever for arbitrarily large inclination,
we recall Eqs (3.39b) and (3.41) which give

dW = . Llds sinOdO
Slnu =
~2J1f(sinO - sinOj/2

By changing variable from 0 to ¢, the tip deflection can be expressed as

8 = WA = ~[K(P)-F(p'¢B)-2E(P)+2E(P'¢B)] (3.47)
L -vJ1f
it
where E(p,¢);: (1- p2 sin 2 ¢)1I2 dl{J is the incomplete elliptic integral of the
second kind and E(p) = E(p, n /2) is the complete elliptic integral of the second
kind.
The horizontal displacement at the tip X A , shown in Fig. 3.10, is expressed as
1/2
X = X A = 1- [ 2(2p2 -1) ] (3.48)
A L J1f

Details of the derivation can be found in the book by Frisch-Fay [1962].

Based on numerical results obtained by Bishop and Drucker [1945] and Frisch-
Fay [1962], Fig. 3.11 shows vertical deflection 8 and horizontal displacement of
the tip X A as functions of the external force F. Figure 3.12 sketches the shape of
the cantilever and the inclination angle at the tip eA for various force magnitudes.
The applicability of this elastica solution is restricted by the requirement
that the deformations remain elastic in all cross-sections of the beam; i.e.
F(L-XA ):'5.M y , or
f = ~ :'5. ___M~y___
Fe Fe L (I-xA) ¢m(1-x A )
Using Eqs (3.48) and (3.45), we find

f :'5. J1
2¢;,lsinOA I
or
2
J1f :'5. J1
2¢;,lsinOA I
Consequently, it is necessary that

KyL ~ ~2J1flsineAI (3.49)

66 Static Deflection
1·0 , - - - - - - , - - - r - - - - - - - - - - - - ,

2-/
0·8
!/ c
'8/
!/
..$'
0·5

g
0·4 :Z
t

2 4 5 8 10 f;.;.
Fig.3.11. Vertical deflection 15 and horizontal displacement XA of an elastic cantilever, as functions of force
F at the tip.

Fig. 3.12. Deformed shape and tip inclination of an elastic cantilever due to vertical force at the tip.

Particularly, for cantilevers of rectangular cross-section, ICyL = MyLI E10 =

2YLlEh; hence, the elastica solution is valid only if the force is in the range

flsin8 I ~ -J.l2 (YL)2

A
-Eh (3.50)

or the slenderness of the cantilever is in the range

(3.51 )

Utilizing results shown in Figs 3.11 and 3.12, we summarize the limitations for
the elastica solution of a tip-loaded cantilever in Table 3.1.
This shows that the elastica theory is applicable only for long slender bars.
Since the ratio of E to Y is of the order of 500 for most metals, the last line in
3.2 Large Elastic-Plastic Deflections 67

Table 3.1 Validity of elastica solution of tip-loaded cantilever

-(}A 20° 40° 60° 80° Comment
1) '" tYL 0.21 0.42 0.61 0.79

Jlf 0.73 1.69 3.405 8.678 See Fig. 3.12

1<y L? 0.707 1.474 2.428 4.134
J.l ? 1.06 2.21 3.64 6.20 Rectangular cross-section
Uh? 177 368 607 1034 ElY =500 is assumed

Table 3.1 states that the elastica solution is applicable for tip deflections less
than or equal to L1 = 0.21L only if the slenderness ratio LI h ~ 177. Application of
the elastica solution up to L1 = 0.61L requires LI h ~ 607; this is a slenderness
ratio far beyond the practical range of structural members. Therefore, applicability
of the elastica solution is greatly limited by the fact that in engineering applications
most beams exceed their elastic range before the deflections become large.
Consequently, in the range of large deflections, the effects of nonlinear geometric
relations for beams and struts have been examined in conjunction with nonlinear
elastic constitutive relations. Oden and Childs [1972] performed a finite deflection
analysis of an Euler strut or elastica with moment-curvature relationship given
by a hyperbolic tangent function. Prathap and Varadan [1976] studied the case of
a strut which has a Ramberg-Osgood type constitutive relation. Lo and Das Gupta
[1978] investigated the case in which the stress-strain relationship takes the form
of a logarithmic function when the stress is larger than the elastic limit.

3.2.2 Plastica: Large Plastic Deflection

The discussion in Section 3.2.1 about applicability of the elastica theory is illustrated
by Fig. 3.13. The elastica solution is valid only in the region below the curve; if
the tip deflection (j of a cantilever exceeds the value on the curve, elastic-plastic
deformation occurs in the beam. Therefore, a large deflection analysis is required
for elastic-plastic beams.
In the following analysis, we assume that the material is elastic-perfectly plastic
and the cantilever has a rectangular cross-section. The bending moment in the
cantilever increases as the external force F increases. When F exceeds a value
such that Mmax =Ms =F(L-XA)~My, an elastoplastic region first appears in
the neighborhood of root B; in this region the core of the section is elastic while
the top and bottom layers are perfectly plastic, similar to that shown in Fig. 3.3.
Suppose that for a certain force F, the plastic deformation develops within segment
CB (Sc ~ S ~ L) as shown in Fig. 3.14; then a combination of the M - /( relation

=
(3.12) in the plastic region with the general definition of curvature (3.35) gives
-"2
= /(y [ 3-2 F(X;;.yX A) ] ,

or in nondimensional form,
68 Static Deflection

1- 0

08 large
elastic - plastic
deflection
0-6

0-4 large elastic

deflection

0-2 (elastica)

3
I I
a 200 400 600 800 1000 L/h
(if ElY = 500 )

Fig. 3.13. Validity of the elastica solution of a cantilever subjected to a concentrated force at the tip.

Ll

Fig.3.14. Large elastic-plastic deflection of a uniform cantilever subjected to a concentrated force F at the tip;
point C is the interface between the elastic and plastic deforming segments_

dB =
ds
2/1 [1- f(x-x
313 A'
)r 1l2 (3.52)

where Sc == Sc I L; while definitions of x,f and /1 are given in Eq. (3.7). Combining
(3.52) with the geometric relations (3.39)
dx dw
- = cos8, sin8
ds ds
we obtain a system of first-order ordinary differential equations for the unknown
functions 8(s), xes) and w(s). These equations govern large deflections of elastic-
perfectly plastic bars - a problem that has been termed plastica by Yu and
Johnson [1982].
The boundary conditions for plastic a equations of a cantilever are
w = 8 = 0, at s = I (i.e. at root B) (3.53)
3.2 Large Elastic-Plastic Deflections 69

d8
at (i.e. at interface C) (3.54)
ds
Although the plastica equations appear to be more complex than those for the
elastica, a closed-form analytical solution is obtainable. Starting from (3.52), the
second derivative of s with respect to 8 can be obtained as

~(~) =
de d8
3-vS ~[1- f(x-x
2Jl de A
)t 2 = 3-vS [1- f(x-x
4Jl A
)r 1l2 dx
de
Noting that dx/de = (dx/ds)(ds/d8) = (ds/d8)cos8, gives

!(:) =
27f
---cos8
8Jl2
After integrating this equation and using the boundary condition (3.54) at point
C, we obtain

-ds 27f (.
= - 3 + --2 sm 8e - sm
. 8) (3.55)
de 2Jl 8Jl
The inverse of (3.55) is the nondimensional curvature. The largest nondimensional
curvature is at the root B where

LKB = _de_I = 2Jl--'9~-- > _2Jl_ (3.56)

ds s=1 3 1 + -sin8 3
4Jl e

Notice that sin8e < 0 due to 8e < O. After integrating (3.55) and applying the
boundary conditions of vanishing deflection and rotation at the root of the cantilever,
Eq. (3.53), the intrinsic coordinate s where the bar has rotation 8 is obtained,

s = .2..8 + 27{ (8sin8e + cos8 -1) + 1 (3.57)

2Jl 8Jl
Using dw = cos8ds = cos8(ds/d8)de and Eq. (3.55), the vertical deflection of
the plastic segment CB in the cantilever is found to be

w = .2.. sin 8 + 27{(sinOe sin8-..!.sin 2 0) (3.58)

2Jl 8Jl 2
This satisfies the boundary condition w = 8 ::: 0 at root B. Accordingly, the
deflection at the terminus of the elastoplastic region, point C, is.
3 . 8 27f· 28
we = -sm e + - - 2 sm c (3.59)
2Jl 16Jl
At this stage of the analysis, all quantities in plastic segment CB have been
expressed in terms of parameter 8e . The value of 8 e can be determined by the
continuity conditions for 8 and (de/ds) at point C as follows. By repeating the
elastica analysis presented in Sect. 3.2.1, an equation similar to (3.41) is obtained;
this is a relationship between the arc length Se of the elastic segment and the
inclination 8 e at the elastic-elastoplastic interface C,
1
se = -J]1[K(P)- F(P,<Pe)] (3.60)
with
<Pc = sin-tAp ~1-sin8c ) (3.61)
70 Static Deflection

Substituting (j = (jc in (3.57) and using Eq. (3.60) and s = Sc leads to an equation
for (jc'

k{ ""in-,( ~~p ~1- ,ine, ll}

K(p) -

3 27f .
= -(jc +--2 «(jcsm(jc +cos(jc -1)+1 (3.62)
2f.l 8f.l
By referring to Eqs (3.41), (3.44), (3.45) and (3.53), the continuity of (de/ds) at
point C results in the parameter p that is related to tip rotation,
. (j
= ( 1- sm c + J!:...- )112 (3.63)
p 2 9f
Recall that for a rectangular cross-section the flexibility parameter f.l == 3LK:y /2,
the external force f == F / Fc = FLI M p and the values of (jc and p can be found
for any specified force F by an iteration procedure based on Eqs (3.62) and
(3.63). Then with (3.47) and (3.59), the tip deflection of an elastic-plastic cantilever
is found to be

8 == WA = ~sin(jc + 27~ sin 2 (jc + rk[K(p)-F(p,¢c)-2E(p)+2E(P,¢c)]

L 2f.l 16f.l \j ff.l (3.64)

with ¢c given by (3.61).

Numerical results supporting the plastic a analysis are depicted in Figs 3.15-
3.17. Figure 3.15 shows the tip deflection 8 as a function of the force parameter f
if the flexibility parameter is taken to be f.l = 0.75, 1.5 and 4.5. Recall that small
values of the flexibility parameter (f.l < 1) imply sufficient stiffness so that (jA «
1. The curve f.l = 0 pertains to a rigid-perfectly plastic approximation for E ~ 00.

Fig. 3.15. Tip deflection as a function of the force applied at the tip according to plastica solution. - - ;
elastica solution. - - - - ; elastic small deflection solution. - " - .
3.2 Large Elastic-Plastic Deflections 71

0·'

0·15 f.J. = 0·45

0·3
0'20
0·22

Fig. 3.16. Deformed shapes of an elastic-plastic cantilever with Jl = 0.45.

lp = (1- Sc)
1/3r--------------------------
0·3

0·2

0·1

Fig. 3.17. Variation of the relative length of the plastic segment with force during large elastic-plastic
deflections.

The deformed shape of a cantilever with J.l = 0.45 is shown in Fig. 3.16 where the
evolution of the plastic segment with increasing Be is demonstrated by the locus
of point C.
In studying the complete process of large elastic-plastic deflection of a
cantilever, Wu and Yu [1986] found that with increasing forcefthe length of the
plastic segment will reach a maximum and then decrease as the tip rotation becomes
large. Hence a portion of the cantilever will undergo a loading-unloading process.

lp= 1- S[

~A
0·5 I/J =90·

0·4 C B~ _ - - - - 8 5 . 88 •

0·3

0·2
I: ScL
L
I~L ~
:I
~=====~50°
I/J = O·

f.J. = 015
01

Fig.3.1S. Length of the plastic segment as a function of tip deflection; Jl = 0.15.

72 Static Deflection

The variation of the length of the plastic segment, lp = (1- sc) is shown in Fig.
3.17 for the cases of J1 = 0.75, 1.5 and 4.5. The unloading phenomenon in elastic-
plastic struts and bars that develops with large deflections was also studied by
Monasa [1974, 1980].
The analysis of Wu and Yu [1986] was extended by Liu, Strange and Yu
[1989] by considering large deflections of an e1astoplastic strain hardening
cantilever. Luan and Yu [1991] also studied the effect of an inclined concentrated
force. Figure 3.18 shows the variation in length of the plastic segment with tip
deflection for a fairly large depth J1 = 0.15 and various angles of inclination lJI.
All these analyses of flexure take both geometrical and material nonlinearities
into account but they neglect the influence of the axial component of force on
yielding.

References

Bishop, K.E. and Drucker, D.C. [1945]. Large deflections of cantilever beams. Quart. Appl. Math. 3,
272-275.
Euler, L. [1744]. Methodus Inveniendi Lineas Curvas.
Frisch-Fay, R. [1962]. Flexible Bars. Butterworths, London.
Johnson, W. and Yu, TX. [1981]. On springback after the pure bending of beams and plates of elastic
work-hardening material- III. Int. 1. Meeh. Sci. 23, 687-696.
Liu, J.H., Stronge, W.1. and Yu, T.X. [1989]. Large deflections of an elastoplastic strain hardening
cantilever. ASME 1. Appl. Meeh. 56, 737-743.
Lo, C.C. and Das Gupta, S. [1978]. Bending of a nonlinear rectangular beam in large deflection. ASME
~Appl.Meeh. 45,213-215.
Luan, F. and Yu, TX. [1991]. An analysis of the large deflection of an elastic-plastic cantilever sUbjected
to an inclined concentrated force. Appl. Math. Meeh. 12,547-555.
Monasa, F.E. [1974]. Deflections and stability behaviour of elasto-plastic flexible bars. ASME 1. Appl.
Meeh. 41, 537-538.
Monasa, F.E. [1980]. Deflections of postbuckled unloaded elasto-plastic thin vertical columns. Int. 1.
Solids Struet. 16,757-765.
Oden, J.T. and Childs, S.B. [1972]. Finite deflection of a nonlinearly elastic bar. ASME 1. Appl. Meeh.
37,48-52.
Prathap, G. and Varadan, T.K. [1976]. The inelastic large deformation of beams. ASME 1. Appl. Meeh.
43, 689-690.
Wang, c.Y. [1981]. Folding of elastica: similarity solutions. ASME 1. Appl. Meeh. 48,199-200.
Wu, X. and Yu, T.X. [1986]. The complete process oflarge elastic-plastic deflection of a cantilever. Acta
Mechanica Sinica, 2, 333-347.
Yu, T.X. and Johnson, W. [1982]. The Plastica: the large elastic-plastic deflection of a strut. Int. 1. Non-
Linear Mech. 17, 195-209.
Chapter 4

Dynamic Rigid-Plastic Response

Static deformation of elastoplastic structures applies only if the loading magnitude

is less than the plastic collapse force Fe. With impact or explosive blast loading
however, the structure can be subjected to an intense but short-duration pressure
or force pulse that exceeds the plastic collapse force and initiates structural collapse.
Here we begin with the simplest structural model - a rigid-perfectly plastic
cantilever - and explain how characteristics of intense loads (F >Fc) influence
the structural deformation. The analysis of this deformation uses dynamic equations
of motion rather than the equilibrium equations used in previous chapters. First,
the dynamic behavior of a rigid-perfectly plastic cantilever subjected to a load or
a force pulse at the tip is examined. Then the slightly different response of a
cantilever struck by a colliding body is studied in detail to reveal an additional
transient phase of deformation. For both steady loads and impacts, an elementary
model based on the rigid-perfectly plastic constitutive approximation is used to
calculate the deformed shape as well as the energy dissipation during the process
of deformation.

4.1 Step Loading

4.1.1 Static and Dynamic Loadings

For small deflections, steady forces on a rigid-plastic structure result in accelera-
tions that remain constant. Since the distribution of acceleration does not vary, a
structure that is initially at rest has a velocity distribution that is identical with
the acceleration distribution. At every point the velocity simply increases linearly
with time. The stationary pattern or distribution of acceleration that satisfies the
equations of motion and the yield condition is in fact a dynamic mode for the
structure with these applied forces.
Consider a uniform cantilever that is initially straight; the cantilever is subjected
to a transverse force F(t) at the tip, as shown in Fig. 4.1. In this section it is
assumed that a constant force F is suddenly applied to a stationary cantilever; i.e.
F(t) = F for t > O. This step loading is illustrated in Fig. 4.2.
74 Dynamic Rigid-Plastic Response

r
AI::::::::::::::::::::::::::::::::::::::::~:~
~• • XL _I
z
Fig. 4.1. Uniform cantilever subjected to a transverse force F(t) at the tip.

F(t)

Fig. 4.2. Step force originating at t = O.

The deflection of the cantilever is measured relative to a system of Cartesian

coordinates X, Z with origin at the tip, as shown in Fig. 4.1. Static equilibrium of
a beam element of differential length dX gives a relationship between distributed
force per unit length of beam g(X), shear force Q(X) and bending moment M(X)

dQ = -g(X) (4.1)
dX
dM
-=-Q(X) (4.2)
dX
where the positive sense for g, Q and M is indicated in Fig. 4.3.
If a static transverse force F acts at the tip of a cantilever of length L, then the
shear force and bending moment distributions resulting from this force are
Q(X) = -F and M(X) = FX as discussed in Chap. 3. The plastic collapse load F"
for a transverse force at the tip is related to the structural parameters by
(4.3)
where Mp is the plastic bending moment of the beam. If the tip force is small,
F ~ Fc and there is no distributed load g(X) = 0, the above static solution for
shear force Q and bending moment M does not violate the yield condition anywhere
in the cantilever. Figure 4.4 illustrates the static shear and moment distributions
from an applied force F equal to the collapse force Fc' If F = Fc ' the root of the
beam becomes a plastic hinge since the bending moment M(L) = Mp.

Jlij1j)M ~d: Q+dQ

Fig. 4.3. Forces and moments on a beam element.

4.1 Step Loading 75

1 X~/L

I a

o
.. x
b
Fig. 4.4. Distribution of shear force and bending moment along a stationary cantilever when F = Fe : (a) shear
force; (b) bending moment.

A perfectly plastic cantilever cannot support a static force F larger than the
collapse load Fe' Forces that are larger than the fully plastic collapse force, F >
Fe' cause dynamic deformation; in this case the extra load (F - F;;) generates an
accelerated motion of the beam.

4.1.2 Moderate Dynamic Load (Fe < F <5, 3Fc )

Let a transverse step force F(t) be applied at the tip at time t = 0 and then held
constant. If the magnitude of the force slightly exceeds the fully plastic collapse
force Fe' there will be a plastic hinge located at the root of the cantilever (point
B in Fig. 4.5) where the static bending moment takes its largest value; i.e. when a
load only slightly exceeds the plastic collapse load, the spatial distribution of
deformation is identical with that of static collapse. Thus if F > F;; the cantilever
accelerates as it rotates about this hinge at the root. Let p denote the mass per
unit length of the cantilever and V denote the velocity at the tip; the translational
equation of motion for the entire cantilever rotating about the root is given by
1 dV
-pL- = F+Q (4.4)
2 dt B

where the reaction shear force on the cantilever at the root, QB has a positive
sense downwards (see Fig. 4.5). A second equation of motion for the rotating
cantilever is obtained by taking moments about the plastic hinge at the root B:

(4.5)

Fig. 4.5. Deformation mechanism of cantilever with a plastic hinge at the root.
76 Dynamic Rigid-Plastic Response

This gives a constant rate of acceleration for the tip

dV 3(FL-Mp) 3(F-Fc)
(4.6)
dt = pL2 = pL
Alternatively, Eq. (4.6) can be derived by considering the kinetic energy K of
the rotating cantilever. For a transverse displacement field W(X, t) corresponding
to rotation about a hinge at the root, the transverse velocity of any section is
dWldt = V(l - XIL); thus the kinetic energy is given by

K = i o2
-p (dW)2
Ll - dX
at
= I
-pLV
6
2
(4.7)

The rate of change of kinetic energy is related to the rate of work done by the
external force and the rate of energy dissipation due to plastic deformation; that
is
dK MpV
FV - - - = (F-FJV (4.8)
dt L
where the angular velocity of a cantilever rotating about a plastic hinge at the
root is V/L = d(}ldt. Substituting (4.7) into (4.8) results in Eq. (4.6) as required.
At this point it is convenient to introduce a set of nondimensional variables
and structural parameters to reformulate the present dynamic problem. Using the
nondimensional variables previously introduced for static analysis,
X W F q ;: Q
x == -, w == - , f == - , (4.9)
L L F;; F;;
A nondimensional time is also defined as
t
T;:- (4.10)
To
where To is a characteristic time for the plastic cantilever:

To ;:' L [i = L ~ pL (4.11 )
~F---; Mp
The characteristic time To is related to the fundamental period Tl for elastic
vibration of the cantilever (e.g. see Meirovitch [1986], p 226), where

T.
J
= 1.787L2 ~
fa (4.12)

The ratio of the characteristic time to the fundamental period of vibration is a

structural parameter,

To = 0.323 {Eh = 0.457 ~ 1 (4.13)

TJ fn l( y L

where I( y is the elastic limit curvature of the beam. Consistent with this measure
of time, a nondimensional velocity at the tip v( r) can be defined as

v( T ) =- dw(O, T) =- W. (0 ,T ) (4.14)
dT
where ();: dOl dT. Equations (4.10) and (4.14) relate the nondimensional velocity
to structural and load parameters, v = V ~ pi F,..
4.1 Step Loading 77

To obtain the distribution of the shear force along the beam, note that the
acceleration has the following distribution:
w(x) = v(1-x) (4.15)
Combining (4.6) with (4.4) and using nondimensional quantities, we find the
reaction force at the root,
1
qs = i(f -3) (4.16)
The shear force q(x) at any section can be obtained from the equation of motion
for the tip segment together with the acceleration field (4.15); i.e.

q(x) = -f + foi,(x)dX = -f + {x_x;) (4.17)

From the boundary condition q(l) = qs = (f - 3)/2 (see Eq. (4.16», we obtain an
expression for the tip acceleration,
v= 3(f -1) (4.18)
This is identical with the nondimensional expression for (4.6). Hence (4.17) gives
the shear force distribution in the accelerating segment:
q(x) = t(f - 3) - t(l- x)2(f -1) (4.19)
At the tip, qA = q(O) = - f , which just balances the applied load.
The bending moment distribution is obtained by integrating the shear force
(4.19) since q(x) = -dm(x)/dx; thus, for a plastic hinge at the root,

m(x) = M = -.!..{X(f-3)+[(l-X)3- 1](f-l)} (4.20)

Mp 2
This equation satisfies the boundary condition ms = m(l) = 1. Note that in Chaps
4 and 5 where torque t and axial force ii are not considered, the ratio of bending
moment to fully plastic moment M 1M p is denoted by m instead of ih which is
used in other chapters.
The distributions of shear force q(x) and bending moment m(x) that result
from this applied force and acceleration field are shown in Fig. 4.6, for the case
of Fe < F ~ 3~. In this range of applied force, shear at the root is negative,

q
OI ________________-,l~X

-1
a

:V==l 1
b
Fig. 4.6. Distribution of (a) shear force and (b) bending moment along the cantilever when the step force F is
in range Fe < F::; 3Fc .
78 Dynamic Rigid-Plastic Response

q
x
O~----~~~----~-

a
-1

m
/'
_-t--_
I
I
I

I
m (xl >1

/
~ I
I
I X
o A
b

Fig. 4.7. Distribution of (a) shear force and (b) bending moment along the cantilever for a step force F> 3 Fe .
The broken lines are an inadmissible deformation mechanism with a plastic hinge at the root; the solid lines are
the correct bending moment m(x) and shear force q(x), which result from a plastic hinge at an interior point
X=A<l.

qB < O. If the shear force vanishes at any section, then the bending moment at
that section is an extreme (maximum or minimum) since q = -dm / dx = 0 . If both
qB and qA are negative, the shear force does not change sign in the rotating
segment 0 -::; x -::; 1. Thus the bending moment steadily increases with increasing
distance from the tip x and the largest magnitude of the moment occurs at the
root where M(L) = M p' Hence a suddenly applied force in the range Fe < F -::; 3Fc
results in only a single plastic hinge located at the root where there is a reaction
QB -::; O. When the force magnitude F = 3Fc , the shear force at the root vanishes.

4.1.3 Intense Dynamic Load (F> 3Fc)

Initially, suppose that the same velocity distribution applies for both moderate
and intense forces; i.e. presume that there is a plastic hinge at the root. If F> 3Fc ,
then QB > O. This assumption results in shear force q(x) and bending moment
m(x) distributions that are shown as dashed lines in Fig. 4.7. In this case, shear
force changes sign at an interior point located at

-
x =
1 CEr
- ~3c.1=D (4.21)

and the bending moment has a maximum at this interior section where
q(x) = -dm(x)/dx = O. Since the moment at the root m(l) = 1, this implies that
the bending moment at x exceeds the plastic bending moment; i.e.
m(x) > mel) = 1 (4.22)
This erroneous result is a consequence of assuming that the plastic hinge is located
at the root irrespective of the magnitude of the applied force.
However, Eq. (4.22) and Fig. 4.7 indicate that for F> 3Fc a plastic hinge must
appear at an interior point of the cantilever instead of the root. Assume that the
hinge forms at a point H located a distance A away from the tip, A < L. This
4.1 Step Loading 79

Fig. 4.8. Deformation mechanism of cantilever with plastic hinge at interior point H.

deformation mechanism is shown in Fig. 4.8. For this mechanism the angular
acceleration of segment AH rotating about the hinge at H is (dVl dt)/ A, while
segment HB remains stationary. Since the bending moment distribution has a
local maximum at the hinge, the hinge moment M(A) = M p and shear force
Q(A) = O. Hence by adopting A, = AI L as the nondimensional measure of interior
hinge position, the equations of motion for segment AH are found to be
tA,V = 1 (4.23)

lA,2V - FA,-1 (4.24)

3 - J'

with unknowns v and A,. These equations replace Eqs (4.4) and (4.5). Starting
from Eq. (4.23) the transverse acceleration of any section can be obtained as a
function of hinge position A"

v =
21 (4.25)
A,

tv = v( 1- ~) = 2; (1- ~) (4.26)

Hence, noting that at the hinge the shear force qH = q(A,) = 0 while the bending
moment m H = meA,) = 1, the shear force and bending moment distributions are
given by 2

q(x) =- 1(1- ~) (4.27)

IA,( x)3
m(x) = 1- 3 I-I (4.28)

The distributions of shear force and bending moment, given by (4.27) and
(4.28) are illustrated in Fig. 4.7 as solid lines. It is worthwhile mentioning that
although M(X) = M p throughout segment HB near the root (i.e. A ~ X ~ L), this
segment does not accelerate or deform since there is no transverse force (shear)
in this segment.
The location of the plastic hinge can be identified from (4.28) with the boundary
condition at the tip m(O) = 0 ; thus
A, = ~ (4.29)
1
The acceleration at the tip v can then be obtained using Eq. (4.25),
. 2/2
v =-- (4.30)
3
If I> 3 (or F > 3~) the tip acceleration is proportional to F2 rather than to
(F - Fe) as is the case for moderate loading; that is, if I> 3, v is no longer
80 Dynamic Rigid-Plastic Response

moderate intense
20 dynamic
loading
15

10

o 2 3 5 f

Fig. 4.9 The dependence of the tip acceleration upon the force magnitude of step loads.

proportional to (f -1). For the full range of force magnitudes, this acceleration
can be expressed as

0,
j 3(f -1), (4.31)

2/2/3, 3</

Figure 4.9 shows the dependence of tip acceleration on the magnitude of the
force. The rotation of the hinge as a function of time can be obtained directly by
integration of (4.31).
Several useful remarks may be drawn from the analysis given in this section.
A rigid-perfectly plastic cantilever with a suddenly applied load at the tip does
not deform if F < Fe' Whenever F> F;;, there is a section where the bending
moment equals the plastic bending moment; the segment outside this section begins
to rotate. For a steady force the rate of angular acceleration and the location of
the plastic hinge are both constant. The location of the plastic hinge is at the root
for moderate loads F;; < F ~ 3Fe , but if F> 3Fc the plastic hinge is located between
the ends at a distance A = 3Mp / F from the tip.

4.2 Rectangular Pulse Loading

4.2.1 Three Phases in Response of Cantilever

Impact loads are frequently intense in comparison with the static collapse force;
they are, however, of brief duration. Consequently, the structural deformations
caused by these loads can depend on both the load magnitude and the load duration.
Consider a transverse force at the tip of a uniform cantilever; the force is suddenly
applied at time zero and subsequently suddenly removed at time td' This rectangular
force pulse of magnitude Fo and duration td is shown in Fig. 4.10,
o ~ t ~ td
F(t) = { (4.32)
0,
4.2 Rectangular Pulse Loading 81

F(t)

Fig. 4.10. A rectangula~ force pulse of magnitude Fo and duration td'

During the loading period 0:5 t :5 t d , the solution to this problem is identical
with that of the step-loading case discussed in Sect. 4.1; that is, a stationary
plastic hinge must form at the root if Fe:5 Fo :5 3F;;, or at an interior point Ao =
3M/Fo if Fo > 3Fe. The response in this period can be termed phase I; during
this phase the position of the hinge is denoted by Ho'
The force is suddenly removed at time t = td' Subsequently, the plastic hinge
moves away from the section Ao, where it was located during phase 1. In general
therefore, we may assume a mechanism similar to that shown in Fig. 4.8, in
which segment HB remains stationary and segment AH rotates about hinge H.
Notice, however, that the hinge position A = A(t) varies with time t if the force
is not constant.
Recall that at a purely flexural plastic hinge in the interior of the member,
there is no shear force. Hence, after the loading period, the translational momentum
of the moving segment AH in the direction transverse to the longitudinal axis is
constant,
1 t _
-pAY = F(t) dt =
2 0
_
I {Fat,
Fotd == Pf'
0 :5 t :5 td

td < t
(4.33)

where the total impulse imparted during loading Pf == J~d F(t)dt . A second equation
of motion is obtained by considering the change in the moment of momentum of
segment AH about the initial tip position during the period before the hinge
moves to a location A,

foA P a; XdX = Mpt (4.34)

Substituting the velocity distribution dWldt = (I-XIA)V into (4.34) results in

~PA2V
6 = M pt (4.35)

These equations of motion can be obtained directly in terms of the momentum

because the stresses are uniform and invariable in the rigid segment of the cantilever
ahead of the travelling hinge; this is a consequence of the shear force vanishing
at the hinge.
During the early phase of motion while the force is constant, 0 :5 t :5 td , the two
equations of motion (4.33) and (4.35) provide the hinge location,
3M
A = A = -_P = const (4.36)
o Fo
This stationary hinge for the loading phase has a nondimensional expression
82 Dynamic Rigid-Plastic Response

A == Ao = ~ (4.37)
o L f
with f = Fol Fe' Accordingly, during the loading phase the tip velocity v can be
expressed as
v = -2 f 2 l' (4.38)
3
which is the same as (4.30). Phase I ends at time 1'1 = 1'd == tdlTo' where the
characteristic time To is defined in Eq. (4.11) and subscript 1 pertains to phase I.
The tip velocity at the end of phase I is
2 2
= '3 f
2
vI = v(1' d ) 1'd = '3Pf f (4.39)

where the nondimensional total impulse Pf is defined by

F~
p p
Pf ==
c 0
= L~;cP = f1'd (4.40)

Consequently, the nondimensional deflection of the tip at time 1'1 is

2
81 == .1(1'1) = r'!'~(1') d1' = Pf (4.41)
L Jo 3
Thus at the instant when the loading phase terminates, the tip deflection is
proportional to the square of the applied impulse. At this instant, the rotation
angle at the hinge location Ho is
1 2
91 = -"9 Pff (4.42)
Note that the rotation angle is defined by 9 ==dwldx, so here it has a negative
sign.
When the load is removed the plastic hinge begins to travel away from the
point Ao where it was located during phase I; it moves towards the root. The
period of hinge travel immediately after time td is termed phase II; during this
phase the hinge location is given by Eqs (4.33) and (4.35),

A == A = ~ (4.43)
L Pf
The variation in hinge location with time is shown in Fig. 4.11; after the force is
suddenly removed the hinge travels towards the root at a constant speed during
phase II; i.e.
. dA 3
A == - = - = const (4.44)
d1' Pf
This hinge motion in a uniform member maintains a constant transverse momentum
while the transverse velocity at the tip is slowed. The tip velocity during this
travelling hinge phase is obtained by substituting Eq. (4.43) into (4.35)
2
v = -61'
A2
= -2Pf
31'
(4.45)

while during this phase the tip deflection is given by

8(1') = 8(1'd) +J::<1')d1' = pI [1+21n(1':)] (4.46)

4.2 Rectangular Pulse Loading 83

A(T)

f = 12

" '-.1
;- -

:
_

------- --
V(T) I V,

Phase Phase
II
·1· III
./
Fig. 4.11. The variation of the tip velocity and the hinge position with time for an intense rectangular
pulse, f = Fa/ Fe =12.

Phase II terminates when the travelling hinge reaches the root B (A = 1) at time
r2 where
Pj
r2 = 3" (4.47)

At this instant, the velocity and deflection at the tip are

v2 - v(r 2) = 2pj (4.48)

O2 ~ o(r 2 ) = pl [1+2In(~)] (4.49)

respectively. Note that both r2 and v2 depend only upon the total impulse Pj'
and not upon the magnitude or duration of the load.
After the travelling hinge reaches the root (i.e. for r > r2) the pattern of
deformation again changes and phase III takes place. In this phase of response,
the plastic hinge is fixed at the root while the cantilever rotates about the root as
a rigid body. Taking the change in the moment of momentum about the root for
the period after r2,
vCr) = 3(pj - r) (4.50)

Equations (4.38), (4.45) and (4.50) express the tip velocity during three successive
phases. The variation of the tip velocity with time is shown in Fig. 4.11 as a
broken line, for force f = 12. In this figure the tip velocity is illustrated in
comparison with the velocity at the end of the loading phase, vI'
Motion of the cantilever ceases when the velocity at the tip vanishes,
rj = r3 = Pj = 3r2 (4.51)
From Eq. (4.47) we see that the duration of the final phase with a stationary
hinge at the root is twice the sum of the durations of phases I and II, i.e.
r3 - r2 = 2r2' Integration of (4.50) gives the increase in tip deflection during

r:
phase III. The final nondimensional deflection at the tip is found to be

Of = o(r j ) = O+
2 v(r)dr = pl [1+21n(~)+2]
84 Dynamic Rigid-Plastic Response

where the three terms in the bracket represent the contributions from phases I, II
and III, respectively. The part due to the final phase of rotation at the root is
twice as large as that during the loading phase where the hinge is closer to the tip
(if f > 3). The final tip deflection Df is

Df = PJ[I+~ln(;)l f>3 (4.52)

The final rotation angle at the root at termination of phase III is

(4.53)

4.2.2 Deformed Shape

To find the final deformed shape of the cantilever, first consider the plastic curvature
K(X), caused by the travelling hinge. This curvature is generated during phase II.
Across the travelling hinge there is a difference in angular velocity de/dt; since
there is a one-to-one correlation between hinge position and time during phase II,
the angular velocity can be expressed as a function of hinge position rather than
time,
de = ~ dA = K(A) dA = ~ (4.54 )
dt dA dt dt A

By referring to Eqs (4.43) and (4.44), the nondimensional curvature k(A) that
develops during phase II is found to be
v 2pz
k == KL = - . = _f (4.55)
All. 3Az
In segment AD::; x ::; I, changes in curvature occur only as the plastic hinge
transits any section. Thus, for a cross-section with coordinate x, the curvature is

(4.56)

Although the hinge speed is constant in a uniform member, the curvature generated
by this hinge decreases with increasing distance from the tip since the angular
velocity de/dt is decreasing.
Integrating Eq. (4.56) twice with respect to x and using wz(l) = dwz(1)/Jx = 0
at the root, we find the nondimensional deflection produced in phase II as

2p2 (1 )
w2(x) =-f In-:;-+x-l , (4.57)

The rotational angles produced in phases I and III have already been given in
(4.42) and (4.53), respectively. Summing deflections obtained from each of these
terms, we obtain a final nondimensional deflection

PJ[l-....::...+~li-l
AD 3 lAo
J], (4.58)
Wf(x)
4.2 Rectangular Pulse Loading 85

-8,

2 f= 12

Fig. 4.12. Final deformed shapes of a cantilever subjected to rectangular pulses ( f =6 and f =12).

with AD = 3/ f. Figure 4.12 shows a typical final deformed shape of a cantilever

for force ratios f = Fo/ Fe = 6 and 12.
In the case of impulsive loading, Pf = const although Fo ~ 00 while Ao ~ O.
For an impulsive load the final deflection of the cantilever approaches a logarithmic
curve

(4.59)

It is worth noting that for any force magnitude the final tip deflection L1r of a
uniform cantilever with rectangular cross-section is
p2
L1 oc _f_ (4.60)
f b2h3

where the cross-section has width b and depth h. Thus the final tip deflection is
independent of the beam length L and the slenderness Uh.

4.2.3 Energy Dissipation

While a force acts on a plastically deforming structure, part of the work done on
the structure by the force goes immediately into plastic work of deformation and
part increases the kinetic energy of the structure. Ultimately all the input energy
E in is dissipated by plastic work.
The cantilever has energy dissipation focused at stationary hinges Ho and B;
at each of these sections the dissipation is proportional to the rotational angle.
The remainder of the input energy is dissipated by the plastic curvature (4.56)
produced in phase II. Thus, the nondimensional energy dissipation in each phase
d i can be calculated as follows:

d l == Dl (4.61)
Mp

D2
d2 - f~(X) dx = ~p}(f-3)
9
(4.62)
Mp Ao

D3 2 2
d3 - = 1031 = -Pf
3
(4.63)
Mp
86 Dynamic Rigid-Plastic Response

The input energy, i.e. the work done by the pulse, is

2
PfFo
Ein = Fo t1(t d ) = - - (4.64 )
3pMp
or, in nondimensional form
E 1 2
ein == ~n = 3" P f l (4.65)
p

so that the part of the input energy ein that is dissipated during each phase of
motion has the following proportionality:

(4.66)

It is observed, therefore, that (l) phase I always dissipates 1/3 of the total input
energy, independent of the magnitudes of Fo and td, provided Fo > 3Fe and (2)
the energy dissipated in phases II and III is distributed between these phases
according to the force ratio 1 = Fo / ~.. Hence, as the force ratio 1 increases, the
proportion of total energy dissipated in phase II increases at the expense of phase
III. The distribution of energy dissipation along the cantilever is shown in Fig.
4.13; the dependence of energy dissipation on the magnitude of the force 1 is
illustrated in Fig. 4.14.
If 1 ~ 1 ~ 3 (i.e. Fe ~ F ~ 3Fe) then the hinge will remain at the root B through-
out the response; in other words, in the case of moderate loading, phase II does
not appear and all of the input energy is dissipated at the root. In this case, it can
be shown that the final tip deflection is

3 2( 1)
Of = 2Pf 1- 1 ' 1<1<3 (4.67)

Together with Eq. (4.52), this provides the dependence of the final tip deflection
on the magnitude of the applied force I; this is shown in Fig. 4.15. Since

%
.s 2/3 II ........ J
Lu

---<::l f=5 ),/

)-
.s /
OJ

---'" 1;3
(

I~
I
L-____-L______-L______ ~ ____ ~~~x

o
Fig. 4.13. Accumulation of energy dissipation along a cantilever subjected to rectangular pulses ( f = 6 and
f = 12).
4.2 Rectangular Pulse Loading 87

f-rd = PI' Eqs (4.67) and (4.52) also indicate how the final deflection varies with
the pulse duration td.

4.2.4 Synopsis
1. If f = Fo / Fe > 3, the response of a cantilever to a rectangular pulse consists of
three phases. Phase II is most significant if f» 3. This intermediate phase is
characterized by a travelling hinge triggered by unloading of the external force.
The hinge travels away from the loaded segment at a constant speed.

phase III

phase
~ 2/3 II
'<J
....... phase II
t:::l

c
<lJ 1/3
.......
~
phose
I phose I

I I .. f
o 3 g 12

No
~'I' -I-
[MOderate Loading [Intense Loading
Response Two-Phose Response Three-Phose Response

Fig. 4.14. Dependence of energy dissipation on the force magnitude f of rectangular pulse.

~ ________- L ____- L __~ __~ __~ __~~~f

1_ Moderate loading -I. Intense Loading

Fig. 4.15. Dependence of the final tip deflection on the force magnitude f of rectangular pulse.
88 Dynamic Rigid-Plastic Response

2. Deformation associated with the three phases of response consists of (I) a

rotation about the initial stationary hinge Ho; (II) a continuous change of
curvature in segment HoB; and (III) a rotation about the root B. As a result,
the final deformed shape of the cantilever consists of kinks at Ho and B and a
plastically curved segment HoB.
3. Although time ' 2 , at which the travelling hinge reaches the root, is independent
of the force magnitude Fo, the energy dissipation in phase II depends on the
force. As the force intensity increases, a larger part of the input energy is
dissipated in the curved segment of the cantilever.

The main analytical results obtained in this section can be summarized in Table
4.1.

Table 4.1 Phase transition conditions for response of cantilever to rectangular pulse, Fo 1Fe > 3

Phase Time Hinge position Tip veloci ty Tip deflection

'1'=0 Ao = 3// v=O 0=0

Phase I
--- 'I'I = 'I'd Ao = 3// VI = 2ptfl3 01 = p}l3
Phase II

T2=Pf 13 A=l V2 = 2Pf O2 = ~P;[l + 21n(f)]

f)]
Phase III
Tf= Pf= 3'1'2 A=1 vf= 0 of = P;[l +~ln(

4.3 Features of Travelling Hinges

In the last section, it was shown that a rigid-perfectly plastic cantilever has all
deformation concentrated in a plastic hinge. The hinge is stationary if the applied
force is constant or if the momentum distribution has converged to a mode form;
if the force varies in its magnitude, however, there is a transient phase of motion
wherein the hinge travels along the cantilever. The travelling hinge is an important
concept in rigid-plastic structural dynamics; this concept was first introduced by
Lee and Symonds [1952]. They analyzed flexural deformation in a rigid-plastic
free-free beam subjected to a force pulse at the mid-point. In the present section,
some general features of the travelling hinge will be discussed.
At a stationary plastic hinge the displacement is continuous but the inclination
is discontinuous. Rotation of the hinge forms a kink in the neutral axis at the
hinge location. In contrast, a travelling hinge leaves behind a residual curvature
that is finite at every point. While both the displacement and inclination are
continuous at a travelling hinge, the inclination 8 = d W/dX = d w/dx may have
weak discontinuities; that is the derivative with respect to time d8/dt and that
with respect to a spatial coordinate de/dx (the angular velocity and the curvature)
may be discontinuous although 8 is continuous.
4.3 Features of Travelling Hinges 89

In general, let A(t) denote the location of a plastic hinge. Furthermore, for any
function CP(X) let cP+ and cP- denote the values of the function on either side of
the plastic hinge position; thus,
cP+ == CP(A + 0) == lim CP(A + e) , cP- == CP(A - 0) == lim CP(A - e)
e~O e~O

Accordingly, [CP] == cP+ - cP- denotes the jump or discontinuity in magnitude of

cI>(X) that develops as the hinge transits the point X = A(t).
At a travelling plastic hinge the transverse deflection W(A, r) is continuous so
(4.68)
Furthermore if shear deformation is negligible, then in order to develop continuous
displacements, the velocity w= dwl dr of adjacent material particles must be
continuous; i.e.
(4.69)
Equations (4.68) and (4.69) apply to both stationary and travelling hinges.
At a hinge, continuity of displacements requires that a derivative with respect
to time of (4.68) must also vanish:
d
-[w(A,r)] = 0 (4.70)
dr
where A = A(r) is regarded as a function of nondimensional time r. Thus, (4.68)
gives

That is,
[w] + A[w'J = 0 (4.72)
where velocity wand inclination w' = dwldx are d.erivatives for a material particle
that is being transited by the hinge. Moreover, A is the travelling speed of the
plastic hinge. Hence at a hinge, we obtain from (4.69) and (4.72),
A[w'] = 0 (4.73)
Equation (4.73) indicates that (1) for a stationary hinge A= 0, so [w'] ~ 0, i.e.
the slope is discontinuous at the hinge; and (2) for a travelling hinge A 0 so '*
[w'] = 0, i.e. the slope remains continuous at a travelling hinge.
Similarly, if transverse velocity is continuous, differentiation of Eq. (4.69)
yields
[w] + A[w'] = 0 (4.74)
Hence, (1) for a stationary hinge, [w] = 0 must hold, i.e. the acceleration remains
continuous; and (2) for a travelling hinge, both acceleration wand the angular
velocity w' = iJ can be discontinuous.
90 Dynamic Rigid-Plastic Response

In the case of a travelling hinge, starting from [w'] = 0 and carrying out a
similar differentiation, we obtain

[w'] + i[w"] = 0 (4.75)

Eliminating[w'] from Eqs (4.74) and (4.75) gives

[w] = i 2 [w"] (4.76)

Note that (4.75) and (4.76) apply only to travelling hinges. Expressions (4.73)-
(4.76) provide the features of discontinuities at plastic hinges. These are summarized
in Table 4.2.

Table 4.2 Discontinuities at plastic hinges

S~ationary hinge T,ravelling hinge
)..=0 )...,0
Deflection w C C
Slope or inclination w' = 9 D C
Curvature w" = k D" D
Velocity w C C
Angular velocity w' = iJ D D
Acceleration Iii C D
Nill!;.: C, continuous; D, discontinuous.
" k().. + 0) = k()" - 0) = 0, k()..) ~ co.

A plastic hinge forms at a cross-section of a rigid-perfectly plastic cantilever,

where the yield criterion iM(A)i = M p is satisfied. Since the energy dissipation
rate at the:: plastic hinge is positive (i.e. iJ = M plel>
0), the associated flow rule
'*
requires (J O. Indeed, Table 4.2 indicates that a dIscontinuity in angular velocity
is a common and essential feature of plastic hinges, no matter whether they are
stationary or travelling.
The expressions concerning discontinuities at travelling hinges can be useful
in analysis. For instance, the configuration shown in Fig. 4.8 contains a static and
undeformed segment HB; hence (4.75) leads to
W'(A - 0) + iW"(A - 0) = 0 (4.77)
By substituting W'(A - 0) = -vI A and W"(A - 0) = k(A) into (4.77), one obtains
Eq. (4.55) that was employed to find the final shape of the deformed cantilever.

4.4 General Pulse Loading

4.4.1 General Considerations
Now consider a force pulse F(t) at the tip of a cantilever. Suppose the pulse has
arbitrary shape but is always non-negative, i.e. F(t) ~ O. Since any force that is
smaller than the static collapse load Fe does not deform the initially static rigid-
plastic cantilever, it is convenient to take t = 0 as the instant when F(t) first
reaches Fe' as shown in Fig. 4.16.
4.4 General Pulse Loading 91

F(t)

Fig.4.16. General force pulse, with t = 0 taken as the instant when F(t) first reaches Fe and F(t);:: 0 for the
entire loading duration Os t 5 t d

In order to determine how hinge position A(t) and tip velocity V(t) depend
upon F(t), we assume that at time t a single hinge H is located at spatial coordinate
X = A(t). Then similarly to the analysis in Sect. 4.1, the equations of motion for
segment AH (refer to Fig. 4.8) can be written as

~(.!.PAV)
dt 2
= F(t) (4.78)

~(.!.
dt 6 P
A2V) = M p
(4.79)

By employing the same nondimensional variables defined in Eqs (4.9), (4.10) and
(4.14), the above expressions are recast as

~(.!.Av) = fer) (4.80)

dr 2

~(.!.A2v)
dr 6
= 1 (4.81)

These equations can be solved for hinge location A(r) and tip velocity v(r) in
terms of the impulse that has been applied at any time p( r),
3r
A(r) (4.82)
per)

v(r) ~(p(r»)2 (4.83)

3r
where
per) = ffCf)df (4.84)

The travelling speed of the hinge is found by differentiating Eq. (4.82),

i(r) = dA = 3 P(r)-rf(r) (4.85)

dr (p(r»)2
while the direction of travel is determined by
sgn(i) = sgn[p(r) - r fer)] (4.86)
Hence, the travelling hinge moves towards the root or towards the tip depending
on whether [p(r)-rf(r)] is positive or negative.
If the force fer) is a continuous function of time r, then the mean value
theorem applied at any instant r[ gives
92 Dynamic Rigid-Plastic Response

perl) == ~rrf(r)dr = f(r*), O~r*~r, ( 4.87)

r, r, Jo
The following deductions can be made immediately:
1. For a force that monotonically increase~ with time, j(r) > 0 and fer') < f(r,);
hence (4.86) and (4.87) indicate that A < 0 (i.e. the hinge moves towards the
tip).
2. If fer) is a monotonically decreasing function of time r, then f < 0 and
fer') > fer,) ; consequently, i > 0 (i.e. the hinge moves towards the root).
3. If f( r) = const, then i = 0; i.e. while a constant force acts the hinge is
stationary.
4. If the magnitude of a pulse first increases and later decreases, the travelling
hinge may change direction. In this case the turning point can be determined
as the time when the impulse satisfies the following condition,

f:f(f)df = rf(r) (4.88)

Figure 4.l7a shows that (4.88) is the same as making the area of rectangle OD,DD 2
equal to the area underneath portion OD of the pulse curve. Figure 4.l7b is an
example: for this triangular pulse the turning point of the travelling hinge is at
t, = tdl-Ji = O.707td . The hinge moves from the root towards the tip in the period
0< t < t" and then back towards the root during t > t,.
5. After the termination of any pulse t > td the final impulse p j has already been
applied so p( r) = Pj == S;d
f( r) dr > 0 while r f( r) = O. Thus during this period
the hinge moves away from the loading point, A > 0 according to Eq. (4.86).
Hence, after a pulse of any shape has terminated, the distribution of translational
momentum spreads and becomes more uniform. Equation (4.82) indicates that in
a member with uniformly distributed properties, the hinge reaches the root at
time r = Pj l3 when the tip velocity v = 2pj; thereafter the hinge remains
stationary at the root. It can be easily proved that the total rotation angle at the
root always is equal to
2 2
(}Bj = -3Pj (4.89)
irrespective of the pulse shape if the total impulse is large enough, Pj > 3rd •
6. Since the final rotation angle at the tip A equals the sum of the integral of the
curvatures generated by the travelling hinge along the entire cantilever and the
rotation angles generated at fixed hinges (e.g. at the root), it is concluded that
F(t) F(t)

a b

Fig.4.17. The turning point of the travelling hinge in the F(l) curve: (a) for a general pulse; (b) for a triangular
pulse.
4.4 General Pulse Loading 93

F(f)

Fig. 4.18. A linearly decaying pulse with duration td'

£.
eAf = - ~n == -ein (4.90)
p

where Ein and ein are respectively the dimensional and nondimensional work
done by an applied pulse of any shape.
In related experiments, the final hinge rotations eBf and eAf can be easily
measured on deformed specimens, so according to Eqs (4.89) and (4.90) these
measured angles may provide checks on the final impulse Pf , and the total work
E in done by the pulse. This point is discussed further in Chap. 7.

4.4.2 Example: Linearly Decaying Pulse

Consider a cantilever that is subjected to a linearly decaying force pulse at the

tip, as shown in Fig. 4.18. That is,

t)
F(t) = { F (1-- 0 td ' (4.91)

0,

Taking fo = Fol Fe and Pf = fotd12, a phase-by-phase analysis can be performed

similar to that in Sect. 4.2. It is found that the response is characterized by the
force magnitude fo:
I. If fo = Fo 1F;; < I, the cantilever remains static and undeformed.
2. If I < fo < 2, a stationary plastic hinge appears at the root. The motion begins
at r = 0 and then ceases before the pulse terminates; i.e. r f < rd'
3. If 2 < fo < 3, a stationary plastic hinge appears at the root. The motion continues
until r f > rd'
4. If 3 < fo < 6, a plastic hinge appears and moves towards the root from an
initial position A.o = 31 f o ' This travelling hinge reaches the root before the
pulse terminates. Subsequently, the cantilever simply rotates about the hinge
at the root.
5. If fo > 6, the early response is the same as that in case of 3 < fo < 6, but the
travelling hinge does not get to the root until after the pulse terminates; then
there is a stationary hinge at the root until the motion finally ceases. These
results are summarized in Table 4.3. Figure 4.19 shows the variation of the
hinge position and the tip velocity with time for the case of fo = 12.
94 Dynamic Rigid-Plastic Response

fo = 12

(1\ AITI
I:\
~ I: 1\ I
1
~
\:
I: 1 1

§ I :
-< I 1
'I,
I,
I ,VITI/V,
1 ,
AO '
I "
1 ,
........
o tTd Td 1'2

I Phase n
I.Phose
• -I • "I •
Phase III
·1
Fig. 4.19. Variation of tip velocity and hinge position with time for a linearly decaying pulse of magnitude
10 =F~/Fc =12.

Table 4.3 Phase transition conditions for response of cantilever to linearly decaying pulse with initial
amplitude 10 > 6

Phase Time Hinge position Tip velocity Tip deflection

1: A v 0

1:=0 Ao =3110 v=O 0=0

2 A =_9_ 32 88 2
Phase I 1: = "31:d v max =81 pjlo O=-Pj
2/0 243
(j>O, A>O»
6 1 11 2
1:)=1:d Al = - VI = "3 pjlo 0 1 =-Pj
10 18
Phase II
(j =0, ,i =0)
o2 =Pj2[~+~ln(/o)]
Pj
1: 2 ='"3 A:1 =1 v2 = 2 Pj 18 3 6
Phase III
(j= 0, A =0)

1:j=Pj=31:2 A., = 1 Vj =0 23 + "3

OJ=Pj2[18 '60 )]
2 In(/

4.4.3 Equivalent Replacement of Arbitrary Pulse

Although some general expressions have been given for a pulse of arbitrary shape
(e.g. Eqs (4.82) and (4.83», the analysis is very tedious and error prone. It is also
impossible or impractical to determine or reproduce details of the actual loading
pulse in many engineering problems. It is significant, therefore, to study the effect
of pulse shape on the dynamic plastic deformation of structures. Based on the
rigid-perfectly plastic idealization for dynamically loaded structures, Youngdahl
[1970, 1971] proposed a set of equivalence parameters which effectively eliminate
4.4 General Pulse Loading 95

F(t) F(t)

a b
Fig.4.20. Youngdahl's equivalence parameters for a general pulse: (a) definitions of Peff and t mean ; (b) the
equivalent rectangular pulse.

the effect of pulse shape for the final plastic deformation.

Youngdahl's equivalence parameters are described in Fig. 4.20. An effective
total impulse is defined as

(4.92)

where ty and t1 are the times when plastic deformation begins and ends. Then
an effective load is defined by

F - Peff (4.93)
eff - ~
mean
where t mean is the centroid of the effective pulse. This is given by

(4.94)

In Eqs (4.92) and (4.94), ty can be determined as the time when the force pulse
first equals the plastic collapse force F(ty) = Fe' The termination time t 1 is not
known a priori, however, Youngdahl [1970] proposed that tl can be calculated
from the following approximation:
Peff '" F;;(tl -ty) (4.95)

After examining solutions to four classical problems in dynamic plasticity,

including the simply supported beam, circular plate and cylindrical shell, Youngdahl
concluded that the effect of the pulse shape is virtually eliminated if the equivalent
impulse Peff and effective load Feff are used to characterize the pulse. This suggests
equivalence between pulses, so that an arbitrary pulse can be replaced by an
'equivalent' rectangular pulse, as shown in Fig. 4.20b. The equivalent pulses
result in approximately the same final deflection for plastically deformed structures.
This point has been discussed further by Krajcinovic [1972] and Zhu et al. [1986].
As an example, consider the linearly decaying pulse with initial amplitude
F> 6Fc , which was analyzed in Sect. 4.4.2. It is easily shown that ty = 0, t 1 = t d ,
P"ff = Pf and the mean time for the pulse t mean = td/3. Consequently, the magni-
tude of the effective rectangular pulse Feff is

(4.96)

Substituting Eq. (4.92) into (4.52) and (4.67), the final tip deflection produced by
this equivalent rectangular pulse is found to be
96 Dynamic Rigid-Plastic Response

-<fo$.4
4
3
Dj(equiv. rect.) (4.97)

fo
where is defined by the initial magnitude of the linearly decaying pulse.
Figure 4.21 compares the numerical calculation for final tip deflection obtained
with the linearly decaying and the equivalent rectangular pressure pulses. It shows
that (4.97) provides an excellent approximation for the final tip deflection developed
in response to the linearly decreasing force described in Table 4.3. For > 6, fo
Table 4.3 and Eq. (4.97) give

Dj(lin. decay) = p}[2318 +~ln(fo)]

3 6
= 0.08327 +~ln(fo)
3
(4.98)

Dj(equiv.rect.) = P}[l+~ln(~)] = O.07580+~ln(fo) (4.99)

These equations represent parallel straight lines that are almost coincident as
shown in Fig. 4.21. Table 4.4 compares the accuracy of final displacements obtained
with the equivalent pulse for a range of different force magnitudes.

4.5 Impact on Cantilever

In previous sections, the load applied at the tip of a cantilever is assumed to be a

prescribed force pulse; i.e. the load is a specified function of time. However,
dynamic loads due to impact are not a prescribed function of time; the time
dependence of the applied load in a collision depends on the compliance of the
colliding bodies. Nevertheless, if a high speed missile strikes a structure or if two

'·5

'" '·0
~
...
I()

0·5
- - - linearly decaying pulse
- - - - effective rectangular pulse

Fig. 4.21. Final tip deflection obtained with the linearly decaying and equivalent rectangular pulses.
4.5 Impact on Cantilever 97

Table 4.4 Comparison of tip deflections developed in response to

linearly decaying and equivalent rectangular pulses

fo 8t of Error
(linear decay) (equiv. rect.) (%)

1.0 0.0 0.0 0.0

1.5 0.1975 0.1667 -15.6
2.0 0.50 0.50 0.0
2.5 0.70 0.70 0.0
3.0 0.8333 0.8333 0.0
4.0 1.0139 1.0000 -1.3
6.0 1.2777 1.2703 -0.58
12.0 1.7399 1.7324 -0.43
30.0 2.3507 2.3433 -0.32

vehicles collide, the total impulse imparted to each body may be known from the
change in relative velocities. The impulse is just the integral of the interaction
force during the collision period. The impulse on each body is equal to the change
in momentum of the body during collision. The impulse imparted to a structure
by a high speed collision results in a sudden change in velocity for only the part
of the body that is in contact with the colliding mass.

4.5.1 Problem and Assumptions

Suppose a particle with mass G strikes the tip of a uniform cantilever; the particle
is initially travelling at velocity Va and it remains attached to the tip after impact,
as shown in Fig. 4.22. This problem was first analyzed by Parkes [1955]. The
analysis in the following sections assumes:
1. The constitutive model is rigid-perfectly plastic so effects of elasticity, strain
hardening and strain-rate are all neglected (these effects are discussed in Chap.
5).
2. The collision imparts an initial velocity to a particle of mass G; the size, shape
and deformability of the collision region are all neglected.
3. At impact the cantilever is at rest except for the colliding particle; this has
velocity Va in the direction transverse to the undeformed axis of the cantilever.
4. After impact the colliding mass adheres to the tip of the cantilever.
5. Deflections remain small in comparison with the length of the member.
6. In a plastically deforming region, axial force has negligible effect on the bending
moment.

1--1------1

mp==========t=1
Fig. 4.22. Cantilever with mass G attached at the tip; this mass is given an initial velocity Vo.
98 Dynamic Rigid-Plastic Response

4.5.2 Changing Pattern of Deformation

In order to study the dynamic deformation of the cantilever, let V = Vet) denote
the velocity of the tip (i.e. mass G) at any time t > 0 and F(t) the contact force
applied to the particle by the tip of the cantilever. F(t) is just the shear force on
the section adjacent to the colliding mass. This shear force is equal to the contact
force on the particle; it can be obtained from Newton's second law,

F(t) = - GdV- (4.100)

dt
As in previous sections, it is convenient to introduce nondimensional variables,

( ) = a( )
ar
fer) == L, v - V - VA rP
Fe = LITo - 'V Pc '
where p is the mass per unit length of the cantilever, L is the length of the
cantilever and Fe = M p / L is the static plastic collapse load for a transverse force
at the tip of the cantilever. Let y be the mass ratio between the concentrated
mass at the tip and that of the remainder of the cantilever
G
y=- (4.101)
pL
Then Eq. (4.100) can be recast in nondimensional form as
fer) = -yv (4.102)

Transient Phase of Motion (Phase I) A suddenly applied velocity over part of

the structure is termed impulsive loading. This concept is useful for analyzing
dynamic deformation due to impact. With impulsive loading, the velocity distribu-
tion is discontinuous at the instant of impact. After impact, the velocity distribution
that maintains continuity of displacements consists of a disturbance that propagates
away from points of velocity discontinuity; i.e. away from the moving particle at
the tip. This disturbance is driven by the shear force f( 1') at the interface with the
heavy particle. During the response period, the particle continually decelerates so
the shear force decreases as a function of time. With no axial force, the uniform
rigid-perfectly plastic cantilever has plastic deformation only at an isolated plastic
hinge. This pattern of deformation is similar to that for a pulse load of continuously
decreasing force (Sect. 4.4). Hence the equations for rates of change of transverse
momentum and moment of momentum about the tip can be obtained from Eqs
(4.80) and (4.81) together with (4.102):

~(~AV) = -yv (4.103)

dr 2

~(~A2V) = 1 (4.104)
dr 6
where the tip velocity v = vCr) and the hinge location A = A(r) are both unknown
functions of time r. Integrating (4.103) and (4.104) and using the initial conditions
v(O) = Va and A(O) = 0 gives
4.5 Impact an Cantilever 99

AV = 2y(va - V) (4.105)

.I!?v = 6r (4.106)
Thus, the tip velocity and time when the hinge transits any location .1" are given
by
V = (4.107)

r = 2jq2(1 + ~)-I (4.108)

6 2y
Expressions (4.107) and (4.108) provide the relationships between v, A, and r in
the initial phase of deformation where a hinge travels away from the impact point
at the tip. This stage of response is termed phase I. Differentiating (4.108) with
respect to A, gives

~
dA
= ~.1,(2+~)(1+~)-2
6 2y 2y
(4.109)

so that in phase I the travelling speed of the hinge is found to be

i = dA =
d-r
~(1
va
+ ~)2[.1,(2 + ~)]--I
2y 2y
(4.110)

°
Equations (4.108) and (4.11 0) indicate that at the initial instant 't = 0, A, = but
i = 00. With increasing time A, increases but i decreases; that is, the hinge travels
from the tip towards the root of the cantilever with decreasing speed.
Phase I terminates when the hinge reaches the root of the cantilever. Taking
A, = 1 in (4.108) results in the time of arrival r I and tip velocity vI at this
instant,
~-y- (4.111)
3 1+2y
vI == veri) = ~Vo (4.112)
1+2y
when the hinge is travelling at speed
3 (1 + 2y)2
.1,( r I ) = (4.113)
Vo y(1 + 4y)
The hinge starts moving away from the impact point at an indefinitely large
speed.

Modal Phase (Phase II) For time greater than r I , the motion of the cantilever-
mass system is a rigid body rotation about a hinge at the root B; this period is
termed phase II. The stationary pattern of deformation during this phase is the
same as that of the primary mode of the structure. During phase II the rate-of-
change in the moment of momentum about the root is
. 1. = - 1
yv+-v (4.114)
3
or
3
v= 1+3y
= const (4.115)
This implies a constant rate of deceleration for the particle at the tip. Since
100 Dynamic Rigid-Plastic Response

VI tVa = (1 + 1I2y)-I , the duration of phase II is

2y(l +3y)
" 2 - "1 = (4.116)
3(1 + 2y) V a
The angle of rotation at the root during phase II is
1 2 2 2 (1 + 3y)
() I
B
= --VI("2 - "1)
2
= --V
3 0
y
(1 + 2y)2
(4.117)

The response is completely terminated at

2y(1 + 3y)
"I = "2 = "1 + 3(l+2y) Va = yVo = PI (4.118)

It is interesting to find that the nondimensional response time "I in this problem
is equal to the nondimensional total impulse PI = PIt L~PFc ; this is exactly the
same as that for a cantilever without the tip mass which is subjected to either a
rectangular or a linearly decreasing pulse at its tip (see Eq. (4.51) and Table 4.3).
This direct correspondence between response period and applied impulse occurs
because at the terminal location of the travelling hinge, the bending moment
equals the fully plastic moment throughout the response period. After the applied
force vanishes, the moment of momentum about this terminal location decreases
linearly with time; thus, this part of the response period is directly proportional
to total applied impulse. For these elementary pulse shapes the loading period is
also directly proportional to the total impulse, so the total response period "I is
proportional to the total applied impulse PI' Note that in the analysis above, the
nondimensional response variables are functions of only the mass ratio y defined
by Eq. (4.101).
The variation of the tip velocity and the position of the plastic hinge with time
for mass ratios y = 0.2, 1.0 and 5.0 are shown i.n Fig. 4.23. It is seen that the
hinge speed decreases with increasing distance from the impact point since the
moment of inertia of the moving segment about the hinge is continuously increasing .

• phase transition

,/ f--
II
"'01-
/i~
-----7----_
.......

,/
,/
--- -=~~~--I I
/ Vo
A(,';'"><
...............
" -- __ vi,)
---- I
"0
c:
o
-< /
I
a 0·05 0·10 0·15 020 0·25 0,30

Fig. 4.23. Variation of tip velocity and hinge position with time for cantilevers struck by a rigid mass, for mass
ratio r = GI pL equal to 0.2, 1.0 and 5.0.
4.5 Impact on Cantilever 101

A _d~Jf'= ~
B~
q(xl
/j '::
.. :::
'( H~
Q4
Mp
V(tl A (tl
L

x
a d

A H B m(xl
o
V(t~·
W(x:r),. 0
x
b
x
e

Fig.4.24. Transient phase of dynamic response for cantilever struck by a rigid mass at the tip; (a) deformation
mechanism with a travelling hinge H atx = A(T); (b) velocity distribution along the cantilever; (c) acceleration
distribution; (d) shear force distribution; (e) bending moment distribution.

4.5.3 Acceleration, Force and Bending Moment

Phase I is the transient phase of response. In a rigid-perfectly plastic structure
this phase has a rigid segment AH that rotates about hinge H, while segment HB
remains static and undeformed (Fig. 4.24b). The velocity field is

(4.119)

where A = A(r), and the acceleration field is

. _ '(1 Ax) +-,1\,
w - v
A
-- vX-i
2 '
(4.120)

Note that the acceleration contains. an additional term caused by the increasing
length of the rotating segment, i.e A> O. Differentiating Eq. (4.107) with respect
to time r gives the acceleration of the particle at the tip,

v = 2 '
(4.121)

2r(l+ 2~)
Using Eq. (4.110), expressions (4.121) and (4.120) yield the acceleration at the
tip as well as that at any section of the cantilever,
6
v = x :::; Il (4 122)
1l(1l + 4r) , .
102 Dynamic Rigid-Plastic Response

w= 6 (-1
IL(IL + 4y)
+ 2( IL + y)
IL2
x), (4.123)

Expression (4.122) indicates that the acceleration of the tip mass decreases with
time from infinite at r = 0 to v= -6/(4y + 1) at the end of phase I when r = rl.
Equation (4.123) shows that the transverse acceleration at the tip and that at the
travelling hinge are in opposite directions

W(O) = v= - 6 < 0, w(IL) = 6(IL + 2y) > 0 (4.124)

IL(IL+4y) IL2(IL +4y)
That is, the material at any interior section is first jerked into motion as the hinge
passes; sometime later the sign of acceleration changes from positive to negative.
Hence, as shown in Fig. 4.24c, the acceleration changes sign at an interior point
of segment AH. In fact, taking w(x) = 0 leads to

-
x -
IL2
~
1 IL12, Y~ 0
(4.125)
- 2(IL + y) 0, y~ 00

where the expressions on the right are asymptotic limits for very small or large
mass ratios.
The distribution of shear force within the rotating segment can be found by
integrating the acceleration field in the same manner as in Sect. 4.2; that is,

q(x) -f+ rxw(x)dX = - 6y (l+~_IL;y x 2),X::;IL (4.126)

Jo IL(IL+4y) Y IL y
with
6y
q(O) q(IL) = 0
IL(IL+4y) = - f,

qrnin = q(i) = - 3(IL + 2y)2 < - f

2IL(IL + y)(IL + 4y)
where i is given by Eq. (4.125) as the zero point of the acceleration field. The
shear force distribution q(x) is sketched in Fig. 4.24d.
Integration of the shear force with respect to x and satisfying the boundary
conditions m(O) = 0 and m(IL) == 1 gives the distribution of bending moment within
the rotating segment

m(x) == 6yx (1+~- IL+y x 2 ), x::;IL (4.127)

IL(IL + 4y) 2y 3IL2y
This bending moment distribution is sketched in Fig. 4.24e. The dynamic analysis
for a rigid-perfectly plastic cantilever gives m = 1 in segment HB. Although the
segment ahead of the travelling hinge is at yield it does not deform because there
is no shear force in this segment.

I
The shear reaction between the particle and the tip of the uniform cantilever
can be obtained by combining Eqs (4.102), (4.115) and (4.122):

6y 0::; r::; r 1
IL(IL+ 4y)'
fer) == (4.128)
3y
1+ 3r ' rl ::; r ::; r f

where IL == IL(r) is determined by (4.108). As shown in Fig. 4.25, this reaction

4.5 Impact on Cantilever 103

\
6
\
I
I \
5:
\
\
I \

"-3~
~
~~\O..,
\~, I\
\~ \
2 \U' "-
" I

a
\ "-
I ......... ..........
L ________ ____ "'::--a..~-------------

0L----0~.0-S--~0~.,-0--~OL.,S~--0~.~20~~0~.2~S~~0.~30 '·0 rlrf

Fig. 4.25. Magnitude of shear force acting on mass at tip of cantilever.

force decreases monotonically with time, and it has a discontinuity at the time of
phase transition. Note that for a mass at the tip, this shear force does not vanish
until the entire response terminates; i.e. rd == r i . In this sense the response is
different from the general pulse discussed in Sect. 4.4 for which r d < r I'

4.5.4 Deformed Shape

The curvature generated by the travelling hinge can be determined from the angular
speed of the rigid segment near the tip and the speed of the hinge. At any time,
the nondimensional curvature k along the cantilever is given by Eq. (4.55),

k == 1(L == ~ (4.129)
AA
where k is the curvature developed at the hinge. Substituting the tip velocity
(4.107) and hinge speed (4.110) into (4.129) gives the final curvature

k () wl/(x) == ~vo2y2 (x + 4y) (4.130)

I x == I 3 (x + 2y)3
The final curvature distribution for y == 0.2, 1.0 and 5.0 is plotted in Fig. 4.26. If
the colliding mass is small the final curvature is large throughout the cantilever,
whereas if y» 1, almost all input energy is dissipated by rotation at the root
during the modal phase of response. Integrating (4.130) and making use of the
previously determined rotation angle at the root wf(1) == 8 B (see Eq. (4.117)), one
obtains the variation of the inclination angle along the cantilever when the response
terminates:
8 ( ) '( ) 2 2 2 (x + 3y) (4.131)
I x '" WI x ==-3'v oy (x+2y)2

Hence, the final rotation at the tip 8 AI is found to be

(4.132)
104 Dynamic Rigid-Plastic Response

\
'·2 \
\
,·0 \
\
0·8 \ ¥= 0·2
\.
"-
"- ........
......
0·4 .......
0·2 ¥= 5
-------------------------------
o 0·2 0·4 0·6 08 '·0 x

Fig. 4.26. Curvature distribution along cantilever following transit of travelling hinge.

where Ko = GV02 12 = Eo is the initial kinetic energy and eo is the corresponding

nondimensional energy parameter. Expression (4.132) shows that again the tip
rotation is proportional to eo in agreement with the general statement (4.90).
The final deflection wI can be obtained by integrating Eq. (4.131) with respect
to x and using w(l) = 0 at the root. Hence the final deformed shape of the cantilever
is
wI(X) =!v;y2[_1___X_+2ln(1+2Y)] (4.133)
3 1+2y x+2y x+2y
The final tip deflection, therefore, is

81 :: wI(O) =!v;y2[_1_+2ln(1+_1)] (4.134)

3 1+2y 2y
Figure 4.27 shows the dependence of the final deflection at the tip 81 on the
mass ratio y. This deflection increases in proportion to the input energy eo' In
limiting cases, expression (4.133) gives the final deformed shape of the cantilever:
1. If the colliding mass is much larger than the mass of the cantilever, y» 1
and
(4.135)

Roughly speaking this implies that the cantilever merely rotates about the root
as a rigid bar so the final tip deflection is approximately equal to the non-
dimensional initial kinetic energy
1 2
8 j = Wj(O) '" "2Yvo = eo (4.136)
2. If the colliding mass is much smaller than the mass of the cantilever, y« 1
and
~»l (4.137)
y
In this case the cantilever is deformed into a logarithmic curve with a final tip
deflection
4.5 Impact on Cantilever 105

0·6

complete
- - - solution
0,4 _ _ _ model
solution

0·2

0~.'----~07.2----~0~.5~-O,+.O~---2~----~5~--~10 r

Fig. 4.27. Dependence of final tip deflection on the mass ratio r:

- - complete solution, Eq. (4.134);
- - - modal solution, Eq. (4.154).

Of = wf(O) "" ~r2v;

3
In(_l_)
2r
(4.138)

For a lightweight tip mass, the final tip deflection is proportional to the square
of the initial impulse rather than the impact energy. The final shapes of cantilevers
with mass ratios r = 0.2, 1.0 and 5.0 are plotted in Fig. 4.28. As mass ratio r
increases, a larger part of the impact energy is dissipated in developing plastic
curvature at the root during a modal phase of response.

4.5.5 Energy Dissipation

All the input energy for a rigid-plastic structure is ultimately dissipated by plastic
work (Sect. 4.2.3). For impulsive loading, the input energy is the initial kinetic

02

---- ,(=0·2 _

0·8

,·0

Fig. 4.28. Final deformed shapes of cantilevers with mass ratio r = 0.2, 1.0 and 5.0.
106 Dynamic Rigid-Plastic Response

d2/ eo d1/ eo
, - - - - - - . , - - - - - - 1·0 a -----r--=::::;:::=-~

0·6

0·2 0·8

Fig. 4.29. Input energy dissipated in phases I and II as function of mass ratio r.

energy of the tip mass; i.e. Eo = Ko = GV; /2. Thus the nondimensional energies
dissipated in phases I and II are

d, == .EL
Mp
= leAfl-leBfl (4.139)

d2 - ~2 = leBfl (4.140)
p
where
Di = J;~l M p !8(A,tf )! dt = JLl Mp 1C(X,tf ) dx, i = 1, 2

Hence, the part of the impact energy that is dissipated in phase II is

d2 _ d2 _ leBfl _ 4r(1 + 3r) (4.141)

-;;: - d, + d2 - leAf I - 3(1 + 2r)2
The total energy dissipation can be separated into work done in generating
distributed curvature and the subsequent work done in modal deformation; the
relative size of these parts depends merely on the mass ratio r as shown in Fig.
4.29. The larger the mass ratio, the more energy is dissipated in modal deformation
during phase II. Figure 4.30 shows the part of the input energy that has been
dissipated before the travelling hinge transits a section located a distance x from
the tip. The part of the energy remaining when the hinge reaches the root depends
on the mass ratio r.

4.5.6 Modal Approximation

The above analysis indicates that the complete response of a rigid-perfectly plastic
cantilever that is hit at the tip by a heavy particle consists of two phases: phase I
is a transient motion characterized by a travelling plastic hinge moving from the
tip towards the root of the cantilever; and phase II is a rotation of the cantilever
about the root as a rigid body. In phase I the velocity field changes its shape with
4.5 Impact on Cantilever 107

1·0

0·8

04

0·2

0,2 0·4 0·6 0·8 '·0

Fig.4.30. Accumulation of energy dissipation along cantilever struck by a heavy particle for mass ratio r = 0.2,
1.0 and 5.0.

time; during phase II, however, the deformation occurs in a mode-form or modal
motion where the velocity field does not change in shape with time.
Equation (4.141) and Fig. 4.29.show that most of the input energy is dissipated
either in phase I if the mass ratio y = G/ pL« 1, or in phase II if Y » 1. Equations
(4.111) and (4.118) imply that the relative duration of the transient phase is
1
(4.142)
3(2y + 1)
Thus, when y» 1, phase I constitutes only a small fraction of the response
period. Expression (4.135) confirms that when y» 1, the final deformed shape
of the cantilever is virtually determined by phase II alone. This suggests that a
modal solution could be a good approximation for the final deformed shape provided
y »1.
A modal approximation consists of a kinematically admissible velocity field
that is a separable function of spatial and temporal independent variables,
w*(x,r) = v*(r)(I-x) == v*(r)4>*(x) (4.143)
where the superscript * pertains to the modal solution, and 4> * (x) = 1- x is the
shape function of the modal field for the cantilever. The modal acceleration field
is obtained by differentiating Eq. (4.143) with respect to time r:
w*(x,r) = v*(r)(l-x) == v*(r)t/>*(x) (4.144)
The acceleration and the velocity fields are shown in Fig. 4.31.
Referring to the modal velocity field given in (4.143) and finding the rate of
change in the moment of momentum about the root, one obtains an equation
similar to (4.114). It follows that
3
v*(r) = (4.145)
1+3y
and accordingly,
108 Dynamic Rigid-Plastic Response

o 1 x
V*(rl~--­
W*(x, T) = ~ *(rll1-x)

Fig. 4.31. Modal velocity and acceleration fields for cantilever loaded at tip.

..
w*(x,r)
3 1 -x)
= ---( (4.146)
1+3y
Integration of (4.146) with respect to time yields
3
w*(x,r) v* - - - ( l - x ) r (4.147)
o 1 +3y
3(I-x) r2
W * (x, r) = v*r -
o 1 + 3y
-
2
(4.148)

for initial conditions w* (x, 0) = v~ and W * (x, 0) = O.

In order to entirely determine the modal solution, the initial characteristic velocity
for a mode v~ has to be specified. This is required because the modal velocity
distribution is different from the initial velocity distribution. General considerations
for the modal approximation are given in Chap. 2, where the so-called 'min Ao
technique' for determining the initial velocity of modal solutions is described.
For the present problem the 'min Ao technique' gives an approximate magnitude
v~ for the initial tip velocity in a mode,

_ f>(X)Vol{l *(x)dx 3y vo
v~ -'~l~-------------- (4.149)
1+3y
fo p(x)1{l *(x) I{l *(x) dx

where p(x) is the distribution of mass (including the distributed mass along the
cantilever and the concentrated mass attached at the tip). It is worthwhile to point
out that the initial modal velocity given in (4.149) can be obtained by equating
the initial and the modal moment of momentum about the root B.

GVoL = (G+~PL)V;L (4.150)

Here, conservation of moment of momentum about the plastic hinge has been
used to obtain V;; the equivalence between this method and the 'min Ao technique'
was demonstrated by Zhou et al. [1992].
Combining Eqs (4.147) and (4.149) provides the variation of the modal velocity
as
3
v * (r) = --(yv - r) (4.151)
1 +3y 0

Expression (4.151) indicates that the response time in the modal solution is the
same as that in the complete solution; that is rj = y v0 = Pf = r f (see Eq. (4.118».
As sketched in Fig. 4.32, the variation of the modal velocity given by (4.151)
4.5 Impact on Cantilever 109

vivo
'·0
complete solutIon

0·8 modal solution

',0

Fig. 4.32. Tip velocity variation with time for complete and modal solutions; mass ratio r = 0.2.

exactly coincides with that given by phase II of the complete solution, although
the initial kinetic energy of the modal solution is less than that in the original
problem. In fact,

(4.152)

where Ko and K~ are the initial kinetic energies for the complete solution and
the modal solution, respectively. As the mass ratio y increases, the time when the
transient and modal solutions converge becomes a smaller part of the total response
time.
Furthermore, the final deformed shape predicted by the modal solution can be
found from Eq. (4.148) as
* 3 2 I-x
Wf(x) = w*(x,'lj) =-yvo-- (4.153)
2 1+3y
Hence, the final tip deflection predicted by the modal solution is
* * 3 2 1
8f =wf(O)=W*(O,Tf)=-rvo-- (4.154)
2 I +3y
This approximation of 8; is plotted in Fig. 4.27 for comparison with the complete
solution. As the mass ratio y increases, 8; approaches 8 f' A detailed comparison
of results using the two methods is shown in Table 4.5; this illustrates that the
modal solution is a good approximation for the final tip deflection if the mass of
the colliding missile is larger than the mass of the uniform cantilever.
Table 4.5 Comparison offinal tip det1ections predicted
by complete and modal solutions

r Of leo 0;1 eo Error

(%)

0.05 0.1902 0.1304 - 31.4

0.1 0.2944 0.2308 - 21.6
0.2 0.4293 0.3750 - 11.5
0.5 0.6288 0.6000 - 4.58
I 0.7628 0.7500 - 1.68
2 0.8617 0.8571 -0.53
5 0.9384 0.9375 -0.10
10 0.9680 0.9677 - 0.03
110 Dynamic Rigid-Plastic Response

References

Krajcinovic, D. [1972]. On approximate solutions for rigid-plastic structures subjected to dynamic

loading. Int. J. Non-Linear Mech. 7,571-575.
Lee, E.H. and Symonds, P.S. [1952]. Large deformation of beams under transverse impact. ASME J.
Appl. Mech. 19,308-314.
Meirovitch, L. [1986]. Elements of Vibration Analysis, 2nd Edition. McGraw-Hill.
Parkes, E.W. [1955]. The permanent deformation of a cantilever struck transversely at its tip. Proc. Roy.
Soc. Land. A228,462-476.
Youngdahl, C.K. [1970]. Correlation parameters for eliminating the effect of pulse shape on dynamic
plastic deformation. ASME J. Appl. Mech. 39,744-752.
Youngdahl, C.K. [1971]. Influence of pulse shape on the final plastic deformation of a circular plate. Int.
J. Solids Struct. 7, 1127-1142.
Zhou, Q., Zhang, T.G. and Yu, TX. [1992]. The correlation between the min Llo technique and the
conservation of the moment-of-momentum. ASME J. Appl. Mech. 60,677-679.
Zhu, G., Huang, Y.G., Yu, TX. and Wang, R. [1986]. Estimation ofthe plastic structural response under
impact. Int. J. Impact Engng. 4,271-282.
Chapter 5

Second-Order Effects On Dynamic Response

In the preceding chapter the dynamic response of a rigid-perfectly plastic cantilever

to pulse and impact loads was studied to show how inertia influences the distribution
of plastic deformation. With solely transverse loads applied to a straight beam,
the rigid-perfectly plastic idealization results in a simple two degree-of-freedom
displacement mechanism that clearly describes the main features of dynamic
deformation by impact. However, this analysis can only be regarded as a first
approximation; in some cases, strain-rate, strain hardening, shear force, rotary
inertia, elasticity and geometrical change due to large deformation have significant
effects. These effects modify the displacement field obtained for a rigid-perfectly
plastic beam; in this respect they can be considered to be of second order. In the
present chapter, dynamic analyses of major second-order effects are considered
one-by-one in order to find circumstances wherein each effect is significant.

5.1 Strain-Rate Effect

5.1.1 Impulsive Load on Viscoplastic Cantilever

The rate-independent perfectly plastic behavior described in Chap. 1 and used in
Chap. 4 is representative of materials such as heat-treated aluminum alloys. For
other materials, however, a rate-dependent material characterization is more
representative of observed or measured structural response. Most notably this is
true of mild steel at moderate to large nominal strain-rates, de/dt > 1 s-I. The
effect of strain-rate on dynamic deformation of a cantilever can be incorporated
via the Cowper-Symonds equation described in Sect. 1.4. The rate-dependent
impact response was first analyzed by Bodner and Symonds [1960, 1962] and
Ting [1964]. Their analyses are adopted below with some minor modifications.
The rate-dependent or viscoplastic analysis for flexural response of a slender
member to a transverse impact is based on the following assumptions:
1. A uniform cantilever has a particle with mass G affixed at the tip.
2. Initially the cantilever is at rest while the massive particle has a transverse
component of velocity Vo.
112 Second-Order Effects On Dynamic Response

3. The rotation of every element remains small; i.e. small deflections.

4. Inertia is negligible in the plastically deforming segment adjacent to the root.
5. The quasistatic yield moment Mp is independent of the axial force.
In addition to these assumptions, the rate-dependent bending moment at a plastically
deforming section is obtained from the Cowper-Symonds relation according to
Eq. (l.IS). Hence the rate of change of curvature elKhlt and bending moment M
are related by
elK
2Eor (~-11' (5.1)

.( 1r
dt h Mp
where

tor == £0 1 + 2r (5.2)

In order to maintain consistency with the previous rigid-plastic analysis of

Sect. 4.5, the following nondimensional parameters are introduced:

d
(.) == dr ( ), v -

P P q- Q m== M w W (5.3)
-TMIL' =MIL' M'-L

k == ~, p == 2Eor
1] L~
= 2EoL2 ( pL )P1l2(1 + ~)r
h h Mp 2r
Consequently, Eq. (5.1) can be recast as
. dk r
k == - = 1](m -1) (5.4)
dr
The inclusion of a rate-dependent yield moment into the analysis for the
cantilever impact problem (depicted in Fig. 4.22) completely changes the kinematics
of the system. Instead of two separate phases, the cantilever response becomes a
single continuous motion. This is a consequence of the shear reaction at the root
that develops instantaneously at impact due to the necessary variation of strain-
rate and yield moment along the cantilever. In order to satisfy the equation of
motion, yield condition and flow rule for a rate-dependent cantilever with an
impulse load, the plastic deforming region must initially extend over the full
length of the cantilever. This is in contrast to the rigid-perfectly plastic material
model where plastic deformation was concentrated at a hinge. During the response
of a viscoplastic structure, the plastically deforming region shrinks and an unloading
region extends from the tip where the load is applied. In the unloaded region the
bending moment is decreasing from a previously larger value. Assumption (4)
implies that the bending moment varies linearly in the plastically deforming segment
adjacent to the root.
The deformation mechanism for a viscoplastic relation (5.4) at an intermediate
stage of motion is shown in Fig. S.la. It consists of two regions: (1) a rigid outer
segment AH of length AL where the bending moment is decreasing (m < 1 and
Ie = 0); and (2) a plastically deforming inner segmen~ HB of length (l - A)L
where the bending moment is increasing (m > I and k > 0). These regions are
separated by an interface H located at A(t) = A(To r)L where the bending moment
5.1 Strain-Rate Effect 113

m xi

"
mH me

x
A
rigid b
(unloaded) klx)
a

ks
x
0 .>..
c
Fig.S.l (a) Dynamic deformation mechanism of viscoplastic cantilever; (b) distribution of bending moment;
(c) distribution of curvature-rate, r = 5.

is equal to the plastic bending moment Mp. Hence, the bending moments at the
transition section H and the root Bare
(5.5)
where qB denotes the nondimensional shear force (reaction) at the root B, and A
is an undetermined function of time.
Denoting the nondimensional curvature-rate at the root B as kB = k (1), Eq.
(5.4) leads to
kB = T/(m B _l)r (5.6)
Combining Eqs (5.5) and (5.6), it follows that

mB-I 1 . )11 r
(ry
kB
(5.7)
qB = ~ = I-A
Assume that inertia forces in plastically deforming segment HB are negligible
so that the shear force qB = qH' This results in a linear variation of bending
moment in this segment; i.e.
l-x+(x-A)mB
m()
x = (5.8)
I-A '
This bending moment distribution is shown in Fig. 5.lb. Using Eqs (5.4) and
(5.6), we obtain the rate-of-change of curvature,
.
k(x) = kB - -
. (X_A)r , (5.9)
I-A
This distribution of curvature-rate is shown in Fig. 5.lc for a typical value r = 5.
For r > 3, most deformation is concentra~ed in a small segment near the root.
At any section the angular velocity (J can be obtained by integrating the
curvature-rate in the deforming section. The angular velocity of the rigid segment
adjacent to the tip depends on the rate of curvature throughout the· deforming
segment.
114 Second-Order Effects On Dynamic Response

f l k:(X) dx = f\s(x _It)' dx = k:s I-It (5.10)

A A I-It r+1
The transverse velocity in the deforming segment HB can be expressed as

w(x) = fxl(~- x) de. = flx (~- x)k(~)

. ~ (5.11)

= k:s {(l-X)O-It),+1 _ (1-1t),+2_(X-It),+2}, lt~x~l

(I-It)' r+1 (r+l)(r+2)
Thus the velocity w at the interface is
H

wH = w(lt) = k: (1-1t)2 (5.12)

B r+2
Since segment AH rotates about H as a rigid body, the velocity in segment AH is

w(x) = w +9
H H (It-x) = k:
B
[1-1t(1+~)_1-1t
r+2 r+l
x] O~x~1t
r+l '
(5.13)

and the velocity at the tip is

v = w(O) = w +9 H H It 1-1t(1+~)
= k:Br+2 r+1
(5.14)

With the velocity distribution defined by Eq. (5.11), the transverse momentum
can be calculated as a function of the length It by considering the transverse
momentum and the moment of momentum about the root B. This results in the
following two equations of motion:

Po - f:qB dr = yv + fo~(X) dx (5.15)

1· Po - f:mB dr = l·yv+ fW(X)(I-X)dx (5.16)

where r = G/ pL is the mass ratio and Po = yVo is the nondimensional transverse

impulse applied at the tip at the initial instant. Notice that in Eqs (5.15) and
(5.16) the momentum of segment HB is excluded since it has been assumed to be
negligible. Substituting qB from (5.7), mB from (5.6) and w(x) from (5.13) into
these equations, one obtains

(5.17)

(5.18)

with
T l (lt) :; I-It {y+[I+~]It _ _r_1t2}
r+2 r+1 2(r+l)

T 2 (1t) == I-It {y+[l+~]It- 2r+l 1t2 + 2r+l 1t3}

r+2 r+1 2(r+1) 6(r+1)
5.1 Strain-Rate Effect 115

If Eq. (5.18) is subtracted from (5.17), the resulting equation has a right hand
side proportional to

r (A)-r (A) = 1-,1 ,12 {I-A +~A} (5.19)

1 2 r +2 2 6(r + 1)

Hence the difference between Eqs (5.17) and (5.18) is independent of mass ratio
y and always positive for 0 < A < 1.
Thus, for any specified impulse at the tip of a cantilever, the analysis has been
reduced to solving for two variables kB and A that are both functions of time r.
This solution can be obtained by iterating Eqs (5.17) and (5.18) at each successive
interval of time starting from the initial conditions at r = 0

kB(O) = (r+2)p = (r+2)vo' ,1(0) = 0 (5.20)

y
Having solved these equations for the period before the response terminates, that
is for A::;; 1, the final deformed shape of the cantilever can be calculated by an
integration with respect to time of the velocity given in Eqs (5.11) and (5.13).
Ting [1964] analyzed a numerical example that simulates a mild steel specimen
in the impact experiments of Bodner and Symonds [1960]. The parameter values
were taken as eo
= 40.4 s -1, r = 5, Y = 1.64 and Po = 2.93. In this example, the
nondimensional response time r f was found to be 1.635 and the final angle at
the tip 59.20 or 1.033 radians. The main results are illustrated in Figs 5.2-5.4.
In order to improve their analysis, Bodner and Symonds [1962] examined
methods which bring inertia forces of the plastic segment HB into the equations
of motion. It should be noted, however, that if the momentum in the plastic
region is included in the analysis, the problem will inevitably require a solution
for a fourth -order nonlinear parabolic equation with a floating boundary.
Alternatively, a linear approximation for the velocity distribution results in
additional terms, r;
on the right-hand side of Eq. (5.15) and on the right- r;
hand side of Eq. (5.16). With
. . I-x k' (l-A)(l-x)
w(x) = w --
HI-A
= B r+2
these terms are found to be

J I
,w(x) dx = k
I\.
.

B
(1- ,1)3
2(r+ 2)

r; = JA.W(X)(l- x) dx
i = k
B
(1- ,1)4
3(r+2)
If this modified rate-dependence is incorporated into Eqs (5.15) and (5.16), the
initial values given in (5.20) should also be modified to result in

k = 3(r+2)p = ~(r+2)vo' ,1=0, at r =0 (5.21)

B 1+ 3y 1 + 3y
Numerical examples show, however, that this modification does not significantly
alter the results. In other words, the inertia of the plastically deforming segment
HB does not significantly influence the results in practice and thus it can be
neglected.
116 Second-Order Effects On Dynamic Response

Fig. 5.2 Shear force and bending moment at root B as functions of time.

One notable feature of the viscoplastic response is that the position A of the
interface between the plastic and rigid regions varies very little during most of
the response period. This is illustrated in Fig. S.3. Consequently, most plastic
deformation takes place in a portion of length (1- XL) near the root, where X
denotes the average value of A during the response period.
This example also shows that the bending moment at the root, mB , varies little
with time (see Fig. S.2) and so it can be approximated by its initial value mB(O);
i.e. using kB from (S.21),

m
_ I
- + (i.B)llr
- 3y (r+2)V o ]l!r
-_ 1 + [ ---'---'--"- (S.22)
B 1] 1+ 3y 1]

Assuming that this bending moment is constant during the response, one finds
the response time r f from the moment of momentum equation (S .16) with
v(r!) = w(x, r f) = 0

r f '" (~O)lIr llr (S.23)

1 +kB
- [
1+ ~ (r+2)vo ]
1] 1 + 3y 1]
Similarly whe1!... kB = 0, from (S.IS) one can obtain an estimate for the location of
the boundary A between deformed and undeformed regions at time r f
1
llr (S.24)
[
1+ ~ (r+2)vo ]
1 + 3y 1]

1·0

0·5

a 1·0

Fig. 5.3 Interface location A between plastic and rigid regions, and curvature-rate kB at root as functions of
time.
5.1 Strain-Rate Effect 117

30·

10·

Fig. 5.4 Angular velocity OH at interface, and rotation angle eA at tip as functions of time.

Noting that the angular acceleration ()A is roughly constant, the preceding approxi-
mations provide an estimate for the final angle of rotation at the tip of the cantilever,
3p2I,
() Af "" - 2(1 + 3y) (5.25)

The basic viscoplastic model at the beginning of this section neglects inertia
in the plastically deforming segment adjacent to the root. The significance of this
approximation has been examined by developing a modified model in which inertia
of the entire beam is accounted for, but the bending moment m(x) does not vary
during the response; this modified model invokes a kinematic approximation that
does not satisfy the yield condition at all times. Calculations by Ting [1964] have
shown that the results of these two models are practically the same. For the final
rotation of the tip () Af ' the modified model gives rotations that are less than 10%
smaller in magnitude than those of the first model; this difference is largest for a
light mass at the tip. So unless the tip mass is very light, the inertia of the
plastically deforming segment is negligible.
The theoretical predictions of the viscoplastic analysis above are compared
with relevant experimental measurements in Chap. 7. Before making that com-
parison, however, it is worthwhile to point out that a prominent result from this
viscoplastic analysis is that the final curvature increases from the tip to the root;
consequently, the portion near the tip has only small plastic deformation. This
result for a viscoplastic structure is essentially different from the results obtained
from the elementary rigid-plastic theory in Sect. 4.5, where final curvature
decreased with distance from the tip.

5.1.2 Elementary Estimates of the Effect of Strain-Rate on Final

Deformation
In view of the laborious numerical work involved in the viscoplastic analysis
above, some elementary estimates of the effect of the strain-rate on the final
deformation in the cantilever impact problem could be useful both for comparisons
with experiments and for preliminary calculations of dynamic response. The
elementary rigid-plastic analysis in Sect. 4.5 separated the dynamic response to
impact loading into two phases. Here we take the mean value of the curvature-
118 Second-Order Effects On Dynamic Response

rate for each phase of the rigid-plastic response, and calculate the decrease in the
final curvature for each phase that corresponds to the rate-effect.
For phase I, using Eqs (4.111), (4.117) and (4.132), the mean nondimensional
curvature-rate is found to be

k _ IOAfl-IOBfl _ vo 8y+3
(5.26)
I-'l"l - 2 2y+1
Thus, in phase I the mean curvature-rate shows fairly weak dependence on the
mass ratio y; this curvature-rate only varies by a factor of 4/3 when y increases
from zero to infinity.
In phase II the cantilever rotates about a stationary hinge at the root, so the
curvature-rate can be infinite for a hinge of zero length. Thus the rate-effect for
the model depends on an effective length of a plastic hinge Lh . As discussed by
Nonaka [1967], Jones [1989] and Shu [1990], this effective length is usually
taken to be two to five depths of the beam, i.e.
Lh = (2 - 5)h (5.27)
Hence, the mean curvature-rate in phase II can be calculated as

Vo L
(5.28)
2(1 + 2~ ) Lh

For a cantilever made of rate-sensitive material, the plastic bending moment

increases with curvature-rate. By recalling the Cowper-Symonds relation (5.1), it
is estimated that the plastic bending moment m will increase by factors

1+[~ 8Y +3] (5.29)

21] 2y+1

112 ~ 1+[k2]lIr _ 1+[ Vo L ]lIr (5.30)

- 1] - 21](1 + 112y) Lh

for phase I and phase II, respectively. It follows that the deformation parameters
obtained from the rigid-plastic analysis can be modified accordingly. For instance,
the final rotation at the root 8Bf and that at the tip 8 Af given by Eqs (4.117) and
(4.132) can be modified as
1

r
2 1+-
8Bf _ yVo 3y
= (5.31)
112 112
2~
2
(1 +

1 1

2~ r 2~ r
1+- 1+-
_ yVo2 3y 3y
1- + (5.32)
2 111 112
(1 + (1+
while the following modification incorporates the effect of strain-rate on final
5.1 Strain-Rate Effect 119

deflection at the tip of the cantilever

, 8f1 Of2
8f =-+- (5.33)
111 112
where Of1 = of -012 with of being given by Eq. (4.134) and 012 = lesfl.
For a heavy striker r» 1, the response is dominated by phase II (see Sect.
4.5), so that
(5.34)

On the other hand, for a light striker r« 1, the response is dominated by phase
I, so that
2
rV~ _1_ rvo 1
(5.35)

H~;rl
2 111
2

23
-r 22
voln (IJl
- - 2 2 21 ( J 1
1 (5.36)
2r 111 ,Y VO n 2y H~;rl

These estimates are much simpler to calculate than the involved viscoplastic analysis
presented in Sect. 5.1.1.
Indeed, Parkes [1955] proposed a correction to the plastic bending moment
according to the mean strain-rate in the rigid-plastic analysis; however, he only
accounted for strain-rate in phase II. Hence, his estimates are good for heavy
strikers but they may fail for light strikers. The corrections for rate-effects given
above certainly improve the accuracy of estimates for the final deformation.
The following examples provide a rough idea of the magnitude of the correction
factors. Consider two different specimens subjected to the same impulsive load:
Size of specimens: L = 120 mm, h = 3 mm, b = 10 mm
Materials: mild steel and aluminum alloy, both with static yield stress Y = 300
N/mm 2 and dynamic properties as listed in Table 1.2
Colliding mass: heavy, r = G/ pL = 10, Vo = 5 m s-l or light, r = G/ pL = 0.2,
Vo = 250 m S-l
Length of plastic hinge at root: Lh = 2h
The calculated results for effective bending moments 111 and 112 are listed in
Table 5.1. It is seen that (1) the strain-rate effect is larger for mild steel than for

Table 5.1 Examples of correction factors accounting for strain-rate effect

Material 11 To (s) r Vo J.i] J.i2

Mild steel 40.4 0.0078 10 0.325 1.60

Mild steel 40.4 0.0078 0.2 16.25 1.92
Aluminum alloy 3780 0.0045 10 0.188 1.15
Aluminum aHoy 3780 0.0045 0.2 9.38 1.25
120 Second-Order Effects On Dynamic Response

aluminum alloy; (2) the strain-rate effect is larger for high speed impact; and (3)
for both cases the strain-rate effect is not negligible.

5.2 Strain Hardening (or Strain Softening) Effect

5.2.1 Introduction
Most engineering materials display some strain hardening after initial yielding;
that is, for strain in excess of the yield strain the flow stress increases with
increasing strain. A few materials, such as rocks and concrete that suffer internal
damage as a consequence of deformation, may display strain softening in certain
ranges of strain. In Chap. 3, static analysis of flexural plastic deformation in
beams showed that strain hardening diffuses the stationary plastic hinges, so the
plastically deforming region is spread through a finite volume of material. Although
the effect of strain hardening on static deformation of elastic-plastic beams is
well established, there has been little progress in explaining the influence of
strain hardening on the transient phase of dynamic structural response.
Conroy [1952] was first to consider the dynamic deformation of strain hardening
beams; she commented that 'an infinite number of localized plastic regions form
along the beam'. Later analyses of impulsively loaded beams by Florence and
Firth [1965] and Forrestal and Sagartz [1978] obtained solutions which circumvent
the more difficult transient phase of deformation; i.e. they considered strain harden-
ing during only the modal stage. In the above investigations, an approximation
for the modal phase of dynamic response was obtained by assuming that strain
hardening does not influence the spatial distribution of velocity. Jones [1967]
extended this kinematic approximation to include the transient stage of deformation
by considering a travelling plastic hinge distributed over a predetermined short
length of beam. These approximations for plastic deformation in strain hardening
beams do not satisfy the yield condition at all times; nevertheless, they provide
reasonable estimates for final deflections if the strain hardening modulus is small.
In this section, the effect of strain hardening on the dynamic plastic deformation
of a uniform cantilever is examined. First a rough estimate for this effect is given
in a manner suggested by Jones [1967]; this scheme also takes strain-rate into
account. Then a complete analysis of the dynamic response of strain hardening
cantilevers is presented in Sect. 5.2.3; this analysis also applies to strain softening
cantilevers.
The analysis in this chapter assumes that a section is rigid if the bending
moment is less than yield. If the section is plastically deforming so the rate-of-
curvature IC > 0, the deformation is represented by a linear moment-curvature
relation. The moment-curvature relation for pre-yielding and post-yielding states
can be expressed as

j Pre--Yielding
post-yielding and loading M= Mp + a m /(,
/( =
dK"/ dt > °
OJ
(5.37)
post-yielding and unloading dK"/dt=O
5.2 Strain Hardening (or Strain Softening) Effect 121

Fig.5.5 The moment--curvature relations for rigid-linearly strain hardening and rigid-linearly strain softening
materials.

These relations are illustrated in Fig. 5.5. With strain hardening, a cross-section
remains rigid if the bending moment applied is less than M p' but increases in
curvature if M exceeds M p' Here M p is taken to be the initial plastic bending
moment or the initial yield moment for the beam section. After initial yielding in
a section, loading and unloading are distinguished according to the curvature-
rate dK I dt. The solid lines on Fig. 5.5 represent loading paths where dK I dt > 0,
while the dashed line is an unloading path where dK/dt =0. Coefficient am in
Eq. (5.37) is either a hardening parameter if am >0, or a softening parameter if
am <0.
The post-yield and loading equation from (5.37) can be rewritten in a
nondimensional form as
m;2M=I+ak (5.38)
Mp
with a ;2 amlLMp and k = 1\L

5.2.2 Elementary Effect of Strain Hardening on Final

Deformation
Using amplification factors for stress resultants similar to those in Sect. 5.1.2, the
effect of strain hardening on final deformation can be estimated by using one or
more correction factors to modify the initial yield moment. This factor (or factors)
can be found by substituting the mean curvature calculated without hardening
into the moment-curvature relation.
For the cantilever impact problem analyzed in Sect. 4.5, the mean non-
dimensional curvature produced by the travelling hinge during phase I is equal to

k
I
;2 /(
I
L =Ie I-IeBf 1=
Af
yv~
6
(8y+3)
(2y + 1)2
(5.39)

Introducing the effective length of a plastic hinge Lh as given by Eq. (5.27), the
mean nondimensional curvature for phase II can be estimated as

yv~ (1+*)
-6- (1+ 2~ r L

Lh
(5.40)
122 Second-Order Effects On Dynamic Response

If these mean curvatures are substituted into the constitutive relation (5.37),
the nominal plastic bending moment during phases I and II can be expressed as

-
M 1 = M p + amICI =
M + am yv~ (8y+3) (5.41)
p L 6 (2y + 1)2

Zr
(1+_1 )

M, +~ r;; (1+ (5.42)

Alternatively, one may define correction factors for the nondimensional fully
plastic moment during phases I and II as
a Po
- 2
J11 == MI 1 + amICI 1+ (8y + 3) (5.43)
Mp Mp 6 (2y+l)2

J12 - (5.44)

where a = am / LMp' Then all the deformation quantities found in the rigid-
perfectly plastic analysis (Sect. 4.5) also apply here if they are divided by these
correction factors, as was done in Eqs (5.31)-(5.36). For example, using the
parameters listed at the end of Sect. 5.1.2, and taking the hardening parameter a
to be small (a = 0.01), one finds from Eqs (5.43) and (5.44) that }.l2 = 1.10 for
the heavy striker ( y = 10) and J11 = 1.21 for the light striker (y = 0.2).
Symonds [1965] and Perrone [1966] suggested that this elementary estimate
can represent the effects of both strain hardening and strain-rate. This is possible
if the effects are separable,

~ = fi(: )12(£) (5.45)

In general the material strain hardening and strain-rate effects are not decoupled
but Eq. (5.45) provides a rough estimate of how these effects modify the response.
In (5.45), function fl can take the form of the Cowper-Symonds relation (see
Sect. 1.3) while function f2 can be represented by a linear function of strain.
The notion of separable functions can be extended to the moment-curvature or
rate of change of curvature relation in order to represent the influences of strain
hardening and strain-rate. The ratio between the modified bending moment it
and the fully pla~tic moment M p at any section is

J1 == M = CPI(dIC)CP2(IC) (5.46)
Mp dt
This ratio }.l is a crude correction factor for the perfectly plastic analysis; e.g.
function CPI can represent expression (5.4), while CP2 represents the right-hand
side of Eq. (5.38), leading to

(5.47)
5.2 Strain Hardening (or Strain Softening) Effect 123

where material parameters 1/, r and a

correspond to characteristic rate of change
of curvature, rate index and strain hardening modulus, respectively.

5.2.3 Dynamic Analysis of Strain Hardening and Strain Softening

Cantilevers
Based on an analysis by Stronge and Yu [1989], the present section begins with
equations of motion and shows how a hardening (softening) moment-curvature
relation changes the final distribution of curvature in a basic manner. This is a
constitutive effect that is in contrast to the modification factor J.1. that was developed
in the preceding section for mean curvature based on the perfectly plastic material
model. The following analysis examines the influence of strain hardening and
strain softening on the transient stage of dynamic plastic deformation. For a
transverse collision at the tip of a rigid-strain hardening (softening) cantilever,
the entire beam is suddenly loaded to the initial yield moment M p at the instant
of collision. At this instant deformation begins throughout the cantilever; the
deformations are accompanied by transverse accelerations that are large at sections
near the tip and insignificant near the root. These accelerations soon reduce the
bending moment near the tip to less than the current yield moment so a rigid
(unloaded) segment quickly emerges from the tip. This previously deformed but
currently rigid segment spreads from the tip until it envelops most of the cantilever.
The curvature at any section develops before it becomes a part of the rigid segment;
all sections in the rigid segment are unloading from a previously deformed state.
Hence, at all sections that are not very near the root the bending moment first
increases and then decreases. At each section the curvature developed during this
cycle is uniquely defined for strain hardening beams. The picture is not as clear
for strain softening structural elements. At a deforming section in a strain softening
cantilever, the change in the bending moment with time must be negative for
both loading and unloading paths, see Fig. 5.5. The loading and unloading paths
are associated with increasing curvature and no further change in curvature,
respectively. Hence, the sign of rate of change in bending moment cannot be
used as a criterion to distinguish loading and unloading. The present theory solves
this dilemma by separating the cantilever into two regions: (1) a rigid segment of
increasing length that spreads from the tip, and (2) a deforming segment that
shrinks towards the root. We assume that the root segment is loading and the tip
segment is unloading, i.e. that all deformation develops in the root segment.
Furthermore, it is assumed that inertia forces in the slowly moving root segment
are negligible; this assumption is the same as that employed in Sect. 5.1.2. The
present theory for plastic deformation of both strain hardening and strain softening
cantilevers results in dynamics for the transient stage that converge to Parkes'
rigid-perfectly plastic solution in the limit as the hardening parameter a becomes
negligible.
Let us consider again the problem of transverse impact at the tip of a cantilever
as shown in Fig. 4.22, but now assume that the cantilever is either strain hardening
or softening with a moment-curvature relation of the form of Eq. (5.37). Impulsive
loading of the massive particle at the tip at the initial instant produces a transverse
motion; propagation of this disturbance away from the tip depends on inertia and
the constitutive relation. At any instant of time t the cantilever is composed of
two segments: a rigid segment AH near the tip that is unloading (0 < X < A) and
124 Second-Order Effects On Dynamic Response

.. x

w<0
x
a

c
W
W> 0

q q

0
i-- x 0 x
I
cxm>O cxm<O
d f

m m
cxm>O

,
mH
I
,.--
mH I
r - - - - - - __ _
J I
I I
0 , x 0 A x
A
e g

Fig. 5.6 (a) Dynamic deformation mechanism of a strain hardening (or strain softening) cantilever; (b)
distribution of velocity; (c) distribution of acceleration; (d) distribution of shear force in case of strain hardening;
(e) distribution of bending moment in case of strain hardening; (f) distribution of sbear force in case of strain
softening; (g) distribution of bending moment in case of strain softening.

the remainder of the beam HB that is plastically deforming (A < X < L). These
two segments are shown in Fig. 5.6a. The interface A(t) between the segments
moves from the tip towards the root; a section located at coordinate X begins to
deform on the loading path at t = 0 and ceases deforming when A(t) = X. Since
segment AH is rigid while the velocity in segment HB is assumed to be negligible,
the transverse velocity distribution can be expressed in nondimensional form as

(5.48)

where nondimensional quantities w, w, v and A are all as defined in Eq. (5.3).

The inertia of deforming segment HB is assumed to be negligible since the velocities
in this segment are small in comparison with velocities in segment AH.
Distributions of velocity, acceleration, shear force and bending moment along
the cantilever at two intermediate times are sketched in Fig. 5.6. Figure 5.6d, e
pertains to the case of strain hardening (am> 0) whereas 5.6f, g relates to strain
softening (am < 0). Ahead of the interface the distribution of shear force q(x, or)
is uniform in the plastically deforming segment A < x < 1 when the inertia is
5.2 Strain Hardening (or Strain Softening) Effect 125

m
~gid I deforming

o A* 1 x Fig. 5.7 Distribution of bending moment along

(tip) (root) cantilever and definition of A.'.

neglected. At the interface H the shear force qH = q(A, r) either increases or

decreases monotonously with time depending on the sign of am' Consequently,
depending on the sign of am' the bending moment m(x, r) either increases or
decreases linearly with x. The bending moment at the interface mH is determined
by the curvature kH = k(A, r) = W"(A, r) that develops before A( r) reaches any
location
mH = 1 + iKkH (5.49)
where ix = am / LMp' The linear variation of the bending moment in segment HB
can be expressed as
m(x) = l-qH(X-A*), A:=; x:=; 1 (5.50)
where A*(r) is the intersection of m(x) with m=l. Figure 5.7 shows typical distri-
butions for bending moment m and the residual yield moment my, at an instant
during the transient stage.
A comparison of (5.50) with the moment-curvature relation (5.38) gives the
curvature in segment HB as a function of A*( r)
ixw"(x) = -qH(X-A*), A:=;x:=O;1 (5.51)
By integrating (5.51) with respect to x and applying the boundary condition
w'(I, r) = 0 at the root, the inclination in segment HB is found to be
w'(x, r) = - qH(2A* -1- x)(I- x)/2ix, A:=; x :=; 1 (5.52)
Thus, the angular velocity in segment HB is
w'(x,r) = -[qH(2A* -1-x)+2q):](1-x)l2ix, (5.53)
and at the interface H,
w~ = W'(A, r) = - [qH(2A* -1- A) + 2qHl](l- A)/2ix (5.54)
An additional relationship is required at the interface to ensure that the distri-
bution of final curvature in the rigid segment is an analytic function. Therefore
from (5.51),
o= dm(A, r) = _ (A _ A*) dqH (5.55)
dA* qH dA,*
Substituting qH from (5.55) into Eq. (5.54), this constitutive constraint results in
an expression for the angular velocity at the interface,
w~ = qH (1- A)212ix (5.56)

It is confirmed later that at every rigid section, 0 < x < A, the moment is always
126 Second-Order Effects On Dynamic Response

less than the magnitude of mH when the section entered the rotating segment;
hence, the section is unloading. This rigid segment AH rotates with an angular
velocity d wH I dx so the transverse velocity at the tip is
v = - £1H'4,(I- Ai l2a (5.57)
If the small transverse velocity at the moving interface H is negligible, the velocities
and accelerations in the segment AH are given by Eq. (5.48) and
w(x, r) = v(l- xl A) + vxio}, 0~x~A (5.58)
respectively. These accelerations cause a shear resultant at the interface,
qH = (Y + A12)v + vAI2 (5.59)
Likewise, the moment at the interface is
(5.60)
The continuity of the bending moment at the interface requires that Eqs (5.50)
and (5.60) result in the same moment mH at x = A, so that
1 + ..1,* qH = A(vA + 2vi)/6 (5.61)
If the shear resultant at the interface qH = 0, these relations for the rotating rigid
segment adjacent to the tip are the same as those of Parkes' rigid-perfectly plastic
solution.
Equations (5.55), (5.57), (5.59) and (5.61) are rewritten to provide
nondimensional equations describing the dynamics of the system
£1H(A - ..1,*) = qH..i* (5.62)

£1HA(l- ..1,)2 = - 2av (5.63)

d
-(2yv+Av) = 2qH (5.64)
dr
d 2 *
-(A v) = 6(1 + A qH) (5.65)
dr
These are first-order differential equations containing four unknown functions of
time r; namely A, ..1,*, qH and v. Rearranging the equations in a form that is
convenient for integration we obtain
2av(A - ..1,*)
i* (5.66)
qHA(l- ..1,)2

2{3(2y + A) + qH[3A* (2y + A) - A2]}

A (5.67)
A(4y + A)v
2av
qH (5.68)
..1,(1- ..1,)2

2{3 + qH(3A* - 2.A)}

V (5.69)
..1,( 4y + A)

For compatibility with the moment distribution shown in Fig. 5.6, the initial
conditions for this problem are taken as
5.2 Strain Hardening (or Strain Softening) Effect 127

10 r----,----,-----,-----,...---...,..-----,

'{ = 0·'

a.
.+=
'0 0'4
:>.
--u
0
Ci 02
>

0·0
0000 0-030
a

7f = 1-0
::J:
'-.
::,.,
0·6 key
2-
--- ---- ¢ =iXe o = 0·25 (strain softening)
'0 0·4
- - - - ¢=aeo = 0 (perfectly plastic)
£u
0
'-' - ' ¢ = exeo = 0·25 (strain hardening)
Ci 0·2
> "-"-¢=aeo = , 0 (strain hardening)

006 0·12
b

Fig.5.8 Tip velocity as function of time during transient phase of motion: (a) mass ratio r = 0.1; (b) mass ratio
Y= 10.

A, = t = qH = 0, V == vo ' at r == 0 (5.70)
With these initial conditions, the d!fferential equations for the system are singular.
Nevertheless, the interface speed A, immediately after impact can be approximated
by
for r« 1 (5.71)
that is, the interface moves away from the tip with a speed that decreases with
distance. It follows that

A, == ~6r/vo ' for r« 1 (5.72)

This early time approximation gives

v == vo[l-_l
2r
Kr), Vo
A,* = ~Kr
2 Vo
, 2 aV
qH == - -3 rz.:;::-o ' for r« 1
- o \f6rv (5.73)

Numerical calculations verify that the structural deformations depend on the

impact energy eo and the hardening parameter a; the deformation always depends
on a nondimensional group tP:; Cxeo = arv;
12. For light and moderate mass ratios
128 Second-Order Effects On Dynamic Response

1-0,..----,---,.---,.---,.---,-..--..---,

-<
ci. 0'8
;;:
E
o
.;: 0·6
(ll
u
c
o
~ 04
u
(ll
u
.g
(ll
0·2

0.010 0.020 0.030

1·0r---,------,,------,-----,--c~--,-:;>-_,

r =1-0

¢= cre o =-0·25 (strain softening)

¢ = creo = 0 (perfectly plastic)
._. - . ¢ = creo = 0'25(strain hardening)
._.rp =creo = 1·0 (strain hardening)
O·O~; _ _~_ _~_ _~___~_ _~_ _~
0.00 0.12
r/ Vo
b
Fig.5.9 Interface position as function oftime during transient phase of motIon: ~a) mass ratio r = 0.1 ; (b) mass
ratio r = 1.0 .

r, Fig. 5.8 shows the variation of the tip velocity with time during the transient
phase of motion for impulsively loaded cantilevers; results are plotted for values
of I/> equal to - 0.25, 0, 0.25 and 1.0. The rate of decrease in tip speed during
this phase is primarily influenced by the mass ratio r. If r« 1, most of the
initial velocity disappears and energy is dissipated during phase I; in contrast, if
the mass ratio is moderate to large r ~ 1, strain hardening mostly affects the
length of the deforming segment HB for the modal phase of deformation. This is
shown in Fig. 5.9 where the variation of the interface position with time is
illustrated. It is also seen that hardening only slightly increases the deceleration
of the tip.
For s.train hardening cantilevers, the transient phase of deformation terminates
when A = O. The interface terminus AI is some distance from the root, AI < 1;
this distance can be determined from Eq. (5.67),

Ai
(2r + AI) -
_ 3[A:I + q;:
1 ] (5.74)

with "I being the time at which the transient phase terminates, A~ = A* ("I)'
qHl = qH("I)' The terminal position depends on the hardening parameter and a
5.2 Strain Hardening (or Strain Softening) Effect 129

,·O.-------.------,------r---,-----,

e: 0·6
o
~
a.
." 0·6
'0

oa.
OJ

0'4
iii
'"
'0
>- 0·2 0·25
01
Q; ,·0
C
OJ

2 3 4 5
, I r =,oL /6

Fig. 5.10 Energy dissipated during the transient phase of motion is influenced by mass ratio and strain
hardening (or softening) coefficient.

mass ratio y. The subsequent phase of motion has deformation only in the region
A) < x < 1. This region is more widespread with increases in either the hardening
parameter aor the initial kinetic energy eo'
For a strain softening material, a
< 0, the interface always transits the entire
cantilever. Hence, the transient phase terminates when the interface reaches the
root, A = 1. The subsequent modal phase of motion has all deformation concentrated
in the section at the root of the cantilever.
At the end of phase I, the residual kinetic energy is E2 = E(t)), where subscript
2 indicates the energy dissipated in phase II. The part of the initial kinetic energy
Ko= GVo2/2 that is dissipated during phase II is given by

~
Ko
= (1+~)(~)2
3r Vo
(5.75)

Figure 5.10 shows that the energy dissipated during phase I is remarkably insensitive
to the hardening parameter unless the colliding mass G is very light. For a small
mass ratio r where most of the kinetic energy is dissipated by bending away from
the root, strain hardening increases the part of the total energy dissipated during
phase I. Strain softening, on the other hand, reduces the part of the initial kinetic
energy that is dissipated by distributed bending during phase I.
After the transient phase (phase I), the remaining kinetic energy E2 is dissipated
in a stationary mode of deformation during the period 'l') < 'l' < 'l'2 (phase II). The
final response time 'l'2 is defined as the time when velocity at the tip vanishes,
v('l'2) = O. During phase II deformation continues to develop in the segment
A) < x < 1 adjacent to the root of a strain hardening cantilever. Since in this
region inertia is assumed to be negligible, the shear resultant is uniform and the
bending moment increases linearly with distance from the interface terminus A).
At the end of phase I the moment at the terminus mHI = 1- (A) - A~ )qHl can be
used to express the bending moment m)(x,'l')) within the root segment at the
beginning of phase II, that is
(5.76)
where qHl = qH (A), 'l')). During phase II the moment at any section increases
130 Second-Order Effects On Dynamic Response

monotonously from an initial value ml to the final value ml =ml(x,rl). In order

to satisfy the equations of motion in the rigid segment, the moment meA, r) and
the shear resultant qH (A" r) at the stationary interface must vary proportionately
during phase II. Thus in the deforming segment, when motion finally ceases the
bending moment is
(5.77)
During phase II sections adjacent to the root suffer a change in curvature
k2 - k, = (ml - m,)/ Ex because of the increase in bending moment during this phase.
The energy dissipated by this plastic deformation is equal to the kinetic energy
E2 given in (5.75). Thus, the final shear resultant at the terminus H is found to
be

(5.78)

At the end of phase II, the final bending moment m 2 is obtained from this shear
resultant. Hence the final curvature is completely determined by the initial
conditions for phase II because during this phase the deformation is a separable
function of spatial and temporal variables.
In this analysis the hypotheses that prescribe the initial and boundary conditions
for phase II are consistent with the hypotheses that separate the cantilever into
deforming and rigid segments. A consequence of this approximation is a
discontinuity in bending moment at AI. This discontinuity develops during phase
II with strain hardening. The possibility of a reversal in the direction of interface
travel during phase II was considered by Stronge and Yu [1989]; this reversal,
however, is not possible since it would result in a negative rate of energy dissipation.
A fundamental problem arises in finding the motion of a strain softening
cantilever during phase II, after the deforming region shrinks to a point. The
deforming region has no length so any further deformation has indefinitely large
curvature and no energy dissipation. Othet investigations have circumvented this
problem by specifying a minimum length for the deforming region and defining a
moment-curvature relation that asymptotically approaches a nonzero moment as
curvature increases. Alternatively, a different constitutive relation can be defined
for the localized region, i.e. a moment-rotation rather than a moment-curvature
relation, see Martin [1989]. The latter technique is commonly employed at stationary
hinges in rigid-perfectly plastic members.
The final curvature distributions in strain hardening cantilevers are shown in
Fig. 5.11 for mass ratios r = 0.1 and 1.0. With a heavy mass, most of the initial
kinetic energy is dissipated during phase II while the small part dissipated in the
transient phase is almost independent of the hardening coefficient Ex. During
phase II the length of the deforming segment at the root is influenced by the
hardening coefficient; the length of this segment increases with Ex and this
diminishes the increase in curvature during this phase. Consequently, strain
hardening has a larger effect on final tip deflection if the mass ratio is large as
r
shown in Fig. 5.12. If the tip mass is light, < 0.5, most of the impact energy is
dissipated in distributed curvature that develops during phase I. Figure 5.10 indicates
that strain hardening accentuates this effect - the fraction of impact energy
dissipated in phase II decreases as hardening increases. Strain softening is
exceptional only during the final part of the transient phase when the interface is
5.2 Strain Hardening (or Strain Softening) Effect 131

4.----.----.-----r----,----,
y = 0.1 ,eo = 0.5
.....
~ 3
"
"'~"
:;;J 2
;;
...
>
:;;J
cJ

0
0.0 0.2 0.4 0.6 O.S 1.0
x = XIL a

4.---,.---,.---,.---,.---,
'I =1.0,e.= 0.5
key _
-- - - - 5:= -0.5 (strain softening)
- - a.= 0 (perfectly plastic)
~ . _ . - iX=0.5(strain hardening)
2
- .. - a= 2.0 (strain hardening) ( /
:;;J
;;
>
Co
:;;J
U
.. 1---)
r- - j~"
_,S---
oC::d~~-r~-~~==I=~
0.0 0.2 0.4 0.6 O.S 1.0
x = X IL
b
Fig. 5.11 Final curvature distributions along cantilevers: (a) r =0.1, eo =U.5 ; (b) r =1.0, eo =0.5

almost at the root. Then the shear resultant qHl increases very rapidly and there
is a final large increase in curvature near the root.
The discontinuities in curvature that arise at Al during phase II are not apparent
in the final deformed configurations, see Fig. 5.12. These deformed profiles show
that strain hardening decreases the deformation; the only apparent effect on the
distribution of curvature is the extent of the phase II deforming region if the
r
mass ratio is large, > 1. Changes in the strain hardening coefficient and the a
nondimensional initial kinetic energy eo == rv;
12 have exactly the same effect on
the distribution of deformation for all stages of the dynamic response. Hence, the
final deformation is directly proportional to eo and the effect of increasing eo on
the distribution of curvature is equivalent to a proportional increase in a.
As indicated in Fig. 5.11, the present analysis yields a distribution of curvature
which increases with distance from the tip whereas the rigid-perfectly plastic
theory (Sect. 4.5) indicates that curvature increases with distance from the root.
These distinctly different behaviors result from similar hypotheses; the difference
is a consequence of the assumed moment-curvature relation. With the rigid-plastic
idealization, all deformation occurs at a hinge which travels away from the tip.
The segment ahead of the hinge does not deform because it does not carry any
shear force. On the other hand for an elastic-hardening-softening M-I( relationship,
a pulse-loaded cantilever has finite length regions of hardening and softening that
initiate and develop from an interior point some distance away from the tip, Reid,
Yu and Yang [1995].
132 Second-Order Effects On Dynamic Response

0.0 I-,----r--r--::::::~...-...,
)'=0.1, eo= 0.5 ~~

~
0.1
/?-7
U
c
.2
!!' ~/-
;// key
Q; -:~ - - - ~=O{perfectlY plastic)
'0
- - - . - lX=O.l{stroln hardening)
-,,- a.= 0.5{strain hardening)
- ".- If; 2.0 (strain hardening)
0.4'----'------'-----'----'----'
0.0 0.2 0.4 0.6 0·8 1.0
x = X!L a

0.0

..... 0.1
~
c
0

"n
ClI
0.2
Q;
'0

0.2 0.4 0.6 0.8 1.0

x= X/L b
Fig. 5.12 Final deflections of cantilevers: (a) y = 0.1, eo = 0.5; (b) Y = 1.0, eo = 0.5.

The final distribution of curvature obtained from the present analysis is distinctly
different from that for a perfectly plastic material obtained in Chap. 4. The analysis
of dynamic deformation in a strain hardening cantilever neglects inertia in the
deforming segment whereas in the perfectly plastic analysis, this same segment
of the member is at yield but not deforming. Experiments discussed in Chap. 7
show that in practice the final deformation of strain hardening structures is
somewhere between the predictions of these two models; hence in comparison
with the perfectly plastic analysis, the curvature near the tip is reduced by strain
hardening while in a region near the root, the curvature is increased during the
final or modal phase of deformation.

5.3 Effects of Transverse Shear and Rotary Inertia

5.3.1 Interface Conditions for Concentrated Mass

In Chap. 4, the dynamic response of a rigid-plastic cantilever to a transverse
force F(t) or impact is analyzed by assuming that shear strain and the effect of
shear force on yield are negligible. These assumptions are valid for statically
loaded beams, frames and thin plates since the shear stresses are usually much
5.3 Effects of Transverse Shear and Rotary Inertia 133

smaller than the normal stress. For instance, when a cantilever is loaded by a
static force F at the tip, the largest shear force QB == F is at the root; the largest
bending moment Ma == FL is at the same cross-section. Hence, when the force
reaches the static collapse force F;; == Mpl L, the largest shear force is
Qrnax == M pi L. For a rigid-perfectly plastic material the fully plastic shear force
Qp for a cross-section can be estimated as the product of the shear yield stress
T y and the cross-sectional area A, i.e.

Qp == TyA "= YAI2 (5.79)

where Y is the uniaxial yield stress in tension. Thus
2Mp
(5.80)
YAL
For a rectangular cross-section, Mp == Ybh 2 14 and A == bh, so

Qrnax == ~ « 1 (5.81)
Qp 2L
Thus in a slender member the resultant shear force is much smaller than the fully
plastic shear force unless the applied force is very large, F » F;;.
For rigid-perfectly plastic structures, however, theoretical and experimental
studies (e.g. Symonds [1968], Nonaka [1977], Jones and Song [1986] and Liu
and Jones [1987]) have shown that shear is more important for an impact or blast
load than it is for a static load. This point can be verified by inspecting the
dynamic solutions given in Chapter 4. For example, Sect. 4.1 explains the dynamic
deformation of a cantilever subjected to a suddenly applied steady force F at its
tip. The solution shows that the largest shear force is immediately adjacent to the
loading point where shear force is exactly equal to the applied force, F. The
applied force can be arbitrarily large and consequently, the ratio Qrnax I Qp is no
longer necessarily small. The shear stress in a small region near the tip could be
as large as the yield stress. This situation is significant for the cases of impact or
impulsive loading where the force is indefinitely large. As shown in Sect. 4.5.3,
if a transverse impulse is applied to a particle the shear force at a section of the
beam adjacent to the particle is indefinitely large. Hence, in a small region near
the particle at the tip of the cantilever the deformation must be predominantly
shear rather than flexure. The effect of shear on yield during dynamic deformation
is often significant; consequently, shear deformation and the possibility of shear
failure have to be considered in addition to flexural deformation.
The flexural analysis presented in Chap. 4 also neglects the effect of rotary
inertia. Jones and de Oliveira [1979], de Oliveira and Jones [1979] and de Oliveira
[1982] examined a series of problems to study the effect of rotary inertia on the
dynamic response of uniform beams. Their results indicate that rotary inertia of
the beam is not of practical significance unless the beam is extremely short. In
the following, therefore, rotary inertia is not included in the equations of motion.
However, when the mass attached at the tip of the cantilever is a block rather
than a particle, the effect of rotary inertia of the block can be of great importance.
If an impulse is applied to a block attached to the tip and the moment of inertia
for the block is taken into account, then there can be both a transverse velocity
discontinuity and an angular velocity discontinuity at the junction between the
block and the tip of the cantilever. These two discontinuities cause simultaneous
shear and flexural deformation at the junction. In the following section, shear
l34 Second-Order Effects On Dynamic Response

forces at the interface where a heavy particle is attached to the tip of a rigid-
perfectly plastic cantilever are taken into account. Then in Sect. 5.3.3 the effects
of both shear and rotary inertia of a finite size block at the tip are brought into
the analysis. Thus, the validity of the 'heavy particle' idealization is examined.
Finally, shear failure or separation is discussed in Sect. 5.3.4 based on the preceding
models of rigid-plastic response. All of these modeling considerations are set in
the framework of a cantilever impact problem (Fig. 4.22 in Sect. 4.5).

5.3.2 Shear Deformation Adjacent to Colliding Particle

From the rigid-perfectly plastic analysis for impact of a particle at the tip of a
cantilever (Sect. 4.5), the shear force qA at the tip is given by
QAL 6y
qA(r) == Mp = - f =- A(A+4y) (5.82)

while the bending moment rnA (r) = 0 . Immediately after impact, QA exceeds the
fully plastic shear stress Qp since A ~ 0 when r ~ O. Let the nondimensional
fully plastic shear force qp be defined as
QpL
q
p
== -M (5.83)
p
This limiting shear force depends on the shape of the cross-section and the size
of the beam. For instance, for a beam of rectangular cross-section with depth h
and length L,
2L
qp = (5.84)
h
The limiting shear force represents a ratio of shear to bending strength.

Deformation Mechanism A deformation mechanism with boundary condition

(5.82) and limiting force (5.84) is shown in Fig. 5.13a. At the tip of the cantilever
a limiting shear stress resultant Qp opposes the initial relative motion of the heavy
particle at the tip of the cantilever. Thus, a downward shear force of magnitude
Qp is exerted on the tip of the cantilever; this force is steady during an initial
phase of motion. The steady force Q p accelerates a segment of length AoL that is
rotating about a stationary hinge; this case is examined in Sect. 4.1. Taking the
nondimensional force f = qp for the present case, and requiring that qp = 2L1h >
3, one obtains from (4.29) that the distance from the tip to the stationary hinge is

A =~ =~ (5.85)
o qp 2L
The magnitude of the shear force decreases linearly from IQI = Qp at the tip to
Q=O at Ao =Ao L=3hI2 (Fig. 5.13b). Hence, expression (5.85) implies that
shear deformation only takes place in a localized region near the tip, with the
length of the sheared region being of the order of the depth. It should be noted,
however, that this length depends on the shape of the cross-section. For wide-
flanged I-beams the parameter qp may be quite small even for relatively long
span-to-depth ratios (see Jones and de Oliveira [1979]).
5.3 Effects of Transverse Shear and Rotary Inertia 135

H(xJ/Hp

V(t) x
a c

Ao
~

:r~
x x

-1
b d

Fig. 5.13 (a) Dynamic deformation mechanism in which shear yielding occurs at a section of a cantilever
adjacent to an impulsively loaded particle; (b) distribution of shear force; (c) distribution of bending moment;
(d) distribution of transverse velocity.

Yield Criterion Hodge [1959] obtained an interactive yield criterion that combines
shear force and bending moment at rectangular cross-sections of slender beams.
This can be approximated by Eq. (1.37),

(5.86)

For a suddenly applied force at the tip of a cantilever, the following expressions
for shear force and bending moment in segment AHo are given in Chap. 4.

The shear force and bending moment distributions are shown in Fig. 5.13b and c,
respectively. These results are based on an assumed pattern of deformation. While
this pattern is suggested by the distribution of stress resultants, for a complete
solution we must ensure that the stress resultants finally satisfy the yield condition
throughout the structure. Substitution of (5.87) into (5.86) indicates that the
interactive yield condition (5.86) is not violated in segment 0 < x < ..10' and the
fully plastic yield condition is satisfied at only two locations: (I) the particle
junction at the tip where QA = -Qp' MA = 0 and (2) the interior plastic hinge Ho
where x = ..10' QH = 0, MH = Mp. Although an interaction between shear force
and bending moment is considered, all cross-sections in segment AHo remain
rigid and shear and flexural deformations occur only at these distinct locations as
long as the tip force F = Qp is maintained constant. Hence, the solution is complete
since at every cross-section in the cantilever the proposed deformation field gives
stress resultants that satisfy the yield condition .

Equations of Motion When the transverse shear force at a cross-section equals

the fully plastic shear force, the flow rule for plastically deforming materials
136 Second-Order Effects On Dynamic Response

requires that the associated deformation at this section is pure shear. This
corresponds to a discontinuity in transverse velocity at this section,
[V] == lim{V(O+e)-V(O-e)} == V+-V- (5.88)
£---70

where for any dependent variable, [ ] denotes the change in value at a discontinuous
section. This velocity discontinuity is depicted in Fig. 5.13d. Associated with this
discontinuity, the energy dissipation due to shear Dq has a rate of change

(5.89)

In order to calculate the jump in vertical velocity at the junction between the
particle and the tip of the cantilever, the acceleration at the tip of the uniform
cantilever subjected to a suddenly applied force Qp is calculated using Eq. (4.30),

V·+ = 32 qp2 = const (5.90)

Here v == VTo / L is the nondimensional vertical velocity, T == t/ To and (') == d( )/dT.

Consequently, the transverse velocity at the tip of the uniform cantilever is

v+ -q T
2 2
(5.91)
3 p
On the other hand, the equation of motion for the particle attached at the tip
gives
G dV - =-Q (5.92)
dt p

or in nondimensional form,
yv- = -qp (5.93)
With the initial condition v = Vo at T = 0, this gives the velocity of the particle

(5.94)

At the junction there is a discontinuity in transverse velocity that can be evaluated

from Eqs (5.91) and (5.94)

[v] == v+-v- = -{Vo-qp(~+~qp}} (5.95)

This velocity jump decreases linearly with time T; the shearing terminates at an
instant r = To determined by

(5.96)
To =
q( -+-q
p
Vo
Y 3
I 2
p
)

At the interface where the heavy particle is connected to the uniform cantilever,
a displacement discontinuity develops with time as a consequence of shear; this
occurs if the material is rigid-perfectly plastic. The relative displacement at the
interface can be found by integrating the velocity jump with respect to time; i.e.

[w] == w + -w _ rr q p ( Y+3
= Jo[v]dr = - {vor-T 1 2 qp ) 2}
T (5.97)

so that the discontinuity at T = To is finally

5.3 Effects of Transverse Shear and Rotary Inertia 137

yv 2 3
Oq == [w]r=ro == -----;f- qp(3+2yqp) (S.98)

Shear Rupture Adjacent to Particle Shear rupture can occur if shear deformation
at the interface between the heavy particle and the uniform cantilever exceeds a
critical value. An estimate for conditions that cause rupture can be based on the
magnitude of the displacement discontinuity at this interface. This kind of shear
failure will be discussed in Sect. S.3.S. Here we simply assume that shear rupture
occurs at the particle junction if relative displacement becomes as large as the
depth of the cantilever, i.e. IOql = hi L.

Flexural Deformation after Shear Terminates If IOql < hi L for '~'o' then at
r = ro the plastic shear deformation terminates while the velocity of the particle
and the tip are both equal to

vCr) = 2yqp vA (S.99)

a 3+2yqp
Thereafter shear yielding and relative sliding at the junction disappear, and the
shear force at the tip, q(O), simply depends on acceleration of the particle,
Iq(O)1 = Iyvl < qp (S.100)
The equations of motion for r> '0
are the same as those in Sect. 4.S, that is Eqs
(4.103) and (4.104). Integrating these equations and using (S.8S) and (S.99) as
initial conditions, one obtains the same solutions given in (4.lOS)-( 4.118). This
means that after the initial shear phase there are transient and modal phases of
response that are represented completely by the rigid-plastic flexural theory in
Sect. 4.S. The additional effect of shear on structural response is illustrated in
Fig. S.14 for the case of y = 0.2 and LI h = 6 The flexural theory (Sect. 4.S) is
shown as dashed curves while the solid lines pertain to the present analysis which
takes shear yielding into account. The initial shear phase has no effect on the
subsequent deformation that takes place after r o.

Ii
Vo A
10 r = 0·2
L/h = 6

II
WI
shearing
phase phase I phase II
1"1-
I
.. I • .1
Fig. 5.14 Tip velocity v and position of the plastic hinge A as functions of time for r = 0.2 and L/ h = 6.
138 Second-Order Effects On Dynamic Response

For solutions with and without shear deformation at the junction, the velocity
curves are slightly different only in the short period 0:5: 'I' :5: roo Since the area
underneath the curve of v( r) in Fig. 5.14 represents the final deflection at the tip,
this figure indicates that shear yielding has very little effect on the final tip
deflection. For example, if y = 0.2 and LI h = 6, the final tip deflection is changed
by 1.2% by shear deformation, while if y = 0.2 and LI h =30, this change is less
than 0.02%. Obviously, the effect on final deflection is very small and practically
negligible. This effect may, of course, be more important for other beam support
conditions.

Distribution of Energy Dissipation During the period of increasing deflection,

plastic deformation dissipates all of the kinetic energy initially imparted to the
system. Now let us examine the effect of shear deformation on the distribution of
the energy dissipation. If shear rupture does not occur, i.e. 18q l < hi L, then during
the initial shear phase 0:5: 'I' :5: r o ' the energy dissipation do consists of two parts:
dissipation due to shearing, d q , and dissipation due to bending at H o ' dom '
yv;
dq = 18q lqp = 2 2 (5.101)
1+ 3 YQp

y 2v;qp
(5.102)
(3 + 2yqp)2
Hence, total dissipation do during this first phase of dynamic response to impact
is the sum of these separate parts,

do = dq + d om = --t
yv 2 9 + 8yqp
(3 + 2yqp)2
(5.103)

Since the response during 'I' > ro is exactly the same as that in the rigid-
plastic bending-only theory, the energy dissipation in the sequential phases, I and

1=02, L 1h =6
1·0 -------------------- --r--

Fig. 5.15 The part of the impact energy dissipated before hinge transits any section x, for r =0.2 and LIh =6 .
5.3 Effects of Transverse Shear and Rotary Inertia 139

d z / eo
r-----.,---,--r-r--1·

0·3
0·2

0·1

Fig. 5.16 The part of the impact energy dissipated in various phases of response depends on mass ratio
r=GlpL.

II, is given by
dj eo -do -d2 (5.104)

d2 = Ie I= e 4r(1+3r) (5.105)
sf 0 3(1 + 2r)2
As an example, for a stubby beam LI h = 6 and a relatively light particle r = 0.2,
Fig. 5.15 shows the part of the impact energy dissipated before the travelling
hinge transits any position x. This function has a discontinuity at the tip due to
shear deformation. Again, shear deformation at the junction makes little or no
difference to the final deformation throughout most of the cantilever.
To provide a broader picture of the influence of shear deformation upon the
distribution of energy dissipation, Fig. 5.16 demonstrates the proportions of energies
dissipated in various response phases as functions of mass ratio r = GIpL. It is
clear that the shear deformation is most significant for a light colliding mass and
a stubby cantilever (i.e. small slenderness ratio Llh). These conditions are just
where assumptions of this model are most at odds with the practical situation.
For any value of qp the distribution of energy can be obtained by simply shifting
the original curve in Fig. 5.16 (for which shear yielding is not considered) to the
left along the r -axis a distance of q p 13 .

5.3.3 Shear and Rotary Inertia of Finite Size Colliding Missile

Most previous studies of structural impact have neglected the effect of size or
rotary inertia of a colliding body that embeds in the beam - the analysis is
notably simplified by assuming that the mass of this body is concentrated in a
particle. Shu et al. [1992] took size and rotary inertia of a block at the tip into
account when analyzing oblique impact on a rigid-plastic cantilever. Here a more
complete analysis is given; this considers both transverse shear force at the junction
140 Second-Order Effects On Dynamic Response

MIMp

,
/ ---- -- .......
'\
----- I
(Al
\
-1 0 1
\ I
/
-
\
'-...,. /
.-/
-1

Fig. 5.17 Interactive and independent yield conditions in the shear force-bending moment plane: - --
interactive yield condition (5.86); - - independent yield condition (5.106); (A) refers to Eq. (5.125).

where a block is attached to the tip of a cantilever and also the rotary inertia of
the block.
In this example we employ a separated or independent yield condition rather
than the interactive yield condition (5.86) used previously. With this independent
yield condition a complete solution is obtained that always satisfies yield at every
cross-section. As shown in Fig. 5.17, this yield condition is represented by a
square in the shear force-bending moment plane:

I~I = 1, I~I ~ 1
(5.106)

I~I = 1, I~I ~ 1

Although shear force and bending moment are independent of each other in the
yield condition, the coupling between them still enters the formulation through
the equations of motion.
When rotary inertia of the block at the tip is taken into account, the bending
moment at the junction between the block and the tip of the uniform beam is no
longer equal to zero. Changes in the rate of rotation at the tip require a bending
moment at the junction; this moment depends on the angular acceleration and the
polar moment of inertia for the colliding mass at the tip of the cantilever. However,
the magnitude of the bending moment acting at the junction depends on the size
and rotary inertia of the block. If both shear force and bending moment at the
junction are significant, the consequent deformation associated with these
generalized forces is simultaneous shear and flexure of the plastic hinge at the
junction.

Equations of Motion Consider the deformation mechanism shown in Fig. 5.18,

where both shear and flexural deformation are assumed to be present simultaneously
at the junction between the block and the tip of the cantilever. For an independent
yield condition, the transverse shear force and the bending moment at that cross-
section are equal to Q p and M p' respectively. In the cantilever, the suddenly
applied steady load of Qp and M p at the junction results in a rotating segment of
5.3 Effects of Transverse Shear and Rotary Inertia 141

Fig. 5.18 Dynamic deformation mechanism with both shear and flexural yielding at a section adjacent to
impulsively loaded block.

constant length A == U adjacent to the tip. During an initial period wherein the
shear force at the tip Q(O) == Qp' this segment rotates about a second stationary
plastic hinge Ho that is between the ends. By using previously defined non-
dimensional parameters, the equations of motion for segment AHo are found to
be
1 , .+
-/w == q (5.107)
2 P
1,2.+ _ , 2
- q /l.-
-/l.V (5.108)
3 p

where v+ denotes the transverse acceleration at the tip of the cantilever while the
shear resultant qp == QpLl M p as used in Sect. 5.3.2. These equations give the
segment length A. and tip velocity v+,
A. == ~ == const (5.109)
qp
.+ 1 2
(5.110)
V == ?,qp
The segment of length A. adjacent to the junction at the tip has an angular
acceleration +
.+ v 1 3
OJ == T = 18 qp (5.111)
The associated dimensional angular velocity of this segment is [2+ == OJ+ ITo with
a positive sense indicated in Fig. 5.18.
With the independent yield condition the equations of motion for the block at
the tip are separated from those for the cantilever. The transverse and rotational
equations for this block with a distance a between the center-of-mass and the
junction are
-yvc = qp (5.112)

a
l+q - (5.113)
PL
where ve is the transverse acceleration of the center-of-mass of the rigid block,
tiJ- is the angular acceleration of the block with a positive sense indicated in Fig.
5.18 and rg is the radius of gyration of the rigid block. From Eq. (5.113), the
angular acceleration of the rigid block can be determined,
a
l+q -
PL (5.114)
142 Second-Order Effects On Dynamic Response

Fig.5.19 Square block attached at tip of cantilever; the cantilever cross-section is rectangular while the width
and mass density of block are identical to those of cantilever.

Consequently, the transverse acceleration of the block at the junction is

~(l+q ~)
v- = vc-ro-CZ) = - ~ - Lycrgl:/ (5.115)

Note that at the instant of impact both w+ and w-are zero, and the angular
accelerations ro+ and ro- are both constant. While MA = Mp there is a plastic
hinge at junction A and the flow rule requires relative rotation across this hinge.
Since hinges at A and Ho must have opposite signs, flexural yield at the junction
requires ro+ > ro- or
2
y rg 3 a
- ( - ) q > l+q - (5.116)
18 L P P L

Impulse at Center-of-Mass of Block at Tip of Cantilever The following example

is developed specifically for a cantilever with a rectangular cross-section and a
square block at the tip. The block has side length 2a, while the depth and material
density for cantilever and block are identical, as shown in Fig. 5.19. For this
system, it follows that

rg)2 (rg)2 4Pv ba2 2a 2 4

y (L
G 8 a
= pL L = Pv bhL3 -3- = "3 hL3 (5.1l7)

where Pv is the material density per volume. Substituting (5.1l7) and the shear
force qp = 2L1h from (5.84) into the flexural yield condition (5.116) results in

2.(2a)4
27 h
_2ah -1 > 0 (5.118)

Thus, the deformation mechanism with a combined plastic hinge and shear
deformation at the junction (see Fig. 5.18) requires that

2a > 2.65 (5.119)

h
The transverse velocities on either side of the junction can be obtained by
integrating the accelerations given in Eqs (5.110) and (5.115):
1 2
-q r (5.120)
3 P
5.3 Effects of Transverse Shear and Rotary Inertia 143

v (5.121)

Shear Rupture at Junction with Block Transverse relative displacement at the

junction terminates at the instant "0
given by v+("o) = v-("o); therefore,

r r( 2r
"
o
= vi:!L+ Z(I+ qpZ )
0
+~q21-1 3 P
(5.122)

In particular, for the square block of side length 2a mentioned above,

" =
o
v
0
(!!...)2{~+~~+~(2a)2}-1
L 4 42a 3 h
(5.123)

During the shear phase O:-=;,,:-=; "0' the total sliding distance between two sides of
the interface is

1
0
q
1 == ~v"
2 0 0
== V;(!!...)2{~+~~+~(2a)2}-1
2 L 4 4 2a 3 h
(5.124)
Shear rupture will occur at a critical relative displacement; a detailed discussion
of this point is presented in Sect. 5.3.5. For now suppose that shear rupture
occurs when 10ql = hi L. If the block does not shear off during the shear phase
o:-=; " :-=; "0' the shear force vanishes at time "0' Thereafter solely flexural deforma-
tion of the cantilever continues. After the initial period O:-=;,,:-=; "0
the magnitude
of the shear force at the junction decreases A IQ 1<
Qp, while the remaining
momentum makes the plastic hinge travel from Ho towards the root. The analysis
for the period when" > "0
is omitted here since it is similar to that given by Shu
et al. [1992] and the previous flexural theory in Sect. 4.5.

Junction with Shear Deformation but no Flexure If the polar moment-of-

inertia of the block at the tip is small there can be shear without flexure. A small
polar moment-of-inertia does not satisfy conditions (5.116) and (5.119). In this
case a different model of junction deformation is required instead of the mechanism
shown in Fig. 5.18. At the junction the bending moment MA is smaller than the
yield moment M p' so there is shear deformation but no flexure; i.e. in the shear
force-bending moment plane of a separated or independent yield condition the
resultant 'force' is on the side QA = -Q p and MA < M p' which is marked by (A)
in Fig. 5.17. The 'forces' at the junction that correspond to this state of deformation

I
must satisfy

q A == ~: = - 1, 1m I ==
A I~: < 1 (5.125 )

Since QA is constant during an initial phase of motion, all accelerations and

hence the moment MA are also constant. On either side of the junction the angular
accelerations are found to be
144 Second-Order Effects On Dynamic Response

rnA +aqp/L
(S.126)
r(rg / L)2
where rg is the radius of gyration for the center-of-mass of the block. In this case
there is no flexural plastic hinge at the junction; therefore, there is no relative
angular acceleration across the junction; i.e.
m+ = m- (S.127)
and consequently,

(1 + rnA)2(rnA +qp ~) = %q!r(ir (S.128)

If the block is square with side length 2a and it has the same width and mass
density as the cantilever, then condition (S.128) boils down to a cubic equation
for rnA if the beam cross-section is rectangular.

(1 + rn )2(rn + 2a) = 2..(2a)4 1 :0:; 2a < 2.6S (S.129)

A A h 27 h ' h
At the junction which is shearing, the bending moment rnA is a function of (2a/ h).
This expression is based upon qp = 2L1 h or Qp = Ybh12, which is representative
if 2a > h.
If the side length of the block is less than the depth of the cantilever, 2a < h,
then the cross-sectional area at the junction is reduced to 2ab instead of bh, so it
is reasonable to take Q p = Yab and correspondingly qp = QpLl M p = 4aLlh2 . This
leads to a modified equation for determining rnA as a function of (2a/h),

2a 1
- < (S.130)
h
By combining the solutions ofEqs (S.129) and (S.l30), we obtain the variation of
bending moment rnA for the entire range of relative size for the block, (2a/h).
This is shown in Fig. S.20, where the solid line is the solution of Eq. (S.129) and
the dashed line is the solution of (S.130).
For a very small block, Fig. S.20 shows that fully plastic shearing results in
if ~<1 (S.l31)
h
When the side length of the colliding block is less than twice the depth of the
cantilever, the moment applied at the junction is relatively small so the mass can
be regarded as a particle. Consequently, the boundary condition at the junction is
MA =M(O)""O.

5.3.4 Shear Rupture due to Impact

As mentioned in Sect. S.3.2, if the total sliding distance during the shear phase
reaches or exceeds the depth of the beam, then shear failure or rupture occurs at
the junction where shear deformation develops. Initially we assume rupture occurs
if
(S.132)
If rotary inertia of the colliding mass is negligible, at the junction the deformation
is only shear (see Sect. S.3.2). Assuming that the shear force at this section
remains constant while the relative displacement increases, an estimate can be
5.3 Effects of Transverse Shear and Rotary Inertia 145

05

, \
\
\

"
\
\

··0·5
\ '"
2a<h--+-- 2a>h

2{cCJ:E 2
-:J ~2a l.2fl-l
=n 1CP
-1L-______________________________ __

Fig. 5.20 The bending moment rnA at the junction (section A) varies with the size of the square block.

obtained for the impact energy that results in rupture. By substituting Eqs (5.84)
and (5.98) into the rupture criterion (5.132) we obtain the energy criterion for a
rectangular cross-section,

~rv2
2 0
~ 2+ .i3 rq p (5.133)
or
(5.134)

where Ko = GVo2/2 is the initial kinetic energy or impact energy of the colliding
mass. The rupture and nonrupture regions illustrated in Fig. 5.21 are separated by
a straight line, Eq. (5.134). Rupture depends on the ratio of the initial kinetic

shear
rupture
10

6
rupture

Fig. 5.21 Map showing the region of shear rupture for a particle at the tip.
146 Second-Order Effects On Dynamic Response

energy to the fully plastic bending moment of the cantilever, in comparison with
the product of mass and slenderness ratios G/ ph. Note that the rupture criterion
(5.134) and the map showing conditions which favor shear rupture (Fig. 5.21)
both vary with the shape of the cross-section.
To compare the rupture criterion for pure shear with that due to combined
shear and flexure, consider the special case adopted previously. That is, assume a
square block is attached to the tip; the block has side length 2a and the same
mass density and width as the cantilever. By using the results for shear displacement
obtained in Sects 5.3.2 and 5.3.3, the rupture criterion based on these two models
can be cast in a unified form
K
eo = _ 0 _ > e (5.135)
- Mp - r
where the rupture energy e r is a function of (2a/h) as given below.

1. Pure shear rupture (Fig. 5.13a, Sect. 5.3.2):

er = 2+!(2a)23 h '
if 2a ~ 1
h
(5.136a)

er = ·f -
1
Za< 1 (5. 136b)
h

Z. Shear-flexure rupture (Fig. 5.18, Sect. 5.3.3):

er = 5+3(;J+~e:r, if Za
h
~ Z.65 (5.l37a)

er = 5 + 3m A (Zah) + 3(1+m
-
8 (za)2
-
A ) h
, if Za < Z.65
h
(5.l37b)

where m A IS determined by Eq. (5.129) if 1 ~ Za/ h < 2.65, or by Eq. (5.130) if

o ~ Za/h < 1.
These criteria can be summarized as a map, Fig. 5.ZZ, where the broken line
describes rupture due to pure shear deformation at the junction while the solid
line describes rupture due to shear-flexure deformation. For each model, the curve
divides the space of (Za/h, eo) into a region of shear rupture and a region where
there is no rupture at the tip. If the block is large, more kinetic energy is required
to cause shear rupture. From (5.135)-(5.137) we note that the shear rupture criteria
are independent of length of the cantilever; that is, shear rupture is a local
phenomenon that depends on impact energy and the relative size of the block.

5.3.5 Measurements of Energy for Shear Rupture

The preceding analysis considered a discontinuity in transverse displacement that
developed due to shear between an impulsively loaded block and the uniform
cantilever; the discontinuity at the junction increases with time while shear force
Q(O) at this section remains equal to the fully plastic shear force Qp = Ybh/ Z
Relative motion at the junction stops when the shear force becomes less than that
5.3 Effects of Transverse Shear and Rotary Inertia 147

40
Ko/f1p

30

shear rupture
20

10

Fig.5.22 Map showing the region of shear rupture for simple shear and shear-bending theories if the cantilever
has a square block at the tip.

required for plastic deformation, Q(O) < Qp . This model is used to calculate
conditions that result in rupture where the impulsively loaded block is sheared
from the cantilever. The rupture criterion is that the relative displacement d q
Id j8 IL
equals the depth h; i.e. q 1== q = h. This criterion provides an upper bound on
the impact energy that results III shear failure; the shear force-relative displacement
relation for this model is shown as the solid line in Fig. 5.23.
For ductile materials with small strain hardening, a better model for rupture
energy is one that considers that fully plastic shear force decreases with increasing
relative displacement between the beam and the transversely moving block at the
tip. If relative displacement at rupture is equal to the depth h, then a typical value
for the shear resultant at the junction is

Q =
p
~Ybh(l-'dqll
2 h
(5.138)

This is shown as the broken line in Fig. 5.23.

Both models assume that shearing consists of a transverse component of relative
displacement at one section. In practice shear strain is distributed over a short

Fig. 5.23 Variation of the plastic shear force at junction during shearing.
148 Second-Order Effects On Dynamic Response

Table 5.2 Material Properties and Measured Rupture Energy DA for Impact Cropping
Experiments (Jouri and Jones [1988], Atkins [1990])
Material Y au n £ fr RI,y 2DA IYbh 2
(MNm-') (MNm-') (m)
AI. alloy 115 320 0.223 0.49 5.9 x 10-3 1.54
Mild steel 245 485 0.224 0.90 3.0 x 10-3 1.04

length and shear rupture is the culmination of a fracture process that initiates at a
critical strain. Jouri and Jones [1988] performed impact cropping experiments on
sharply notched aluminum alloy and mild steel specimens of various depths. For
notched specimens the highly deformed shear zone width was 10-15% of the
plate depth h while narrower zones were associated with thicker plates (5 mm < h
< 40 mm). The properties of these materials are listed in Table 5.2; the relevant
properties are yield stress Y, ultimate stress au' work-hardening index n and true
tensile strain at fracture Efr' The ratio of fracture toughness R to yield stress 'fy
for pure shear was obtained by Atkins [1990]. In Table 5.2, DA denotes the
energy dissipated at the junction A, which includes energy dissipation due to
both shear deformation and the initiation and propagation of fracture.
For notched specimens with depth h in the range 5mm < h < 40mm, Jouri and
Jones found the following surface displacement at rupture 8f

_8f 0.16+ 1.19 f or Al . a II oy

h h(mm)
(5.139)
8f 0.27+ 0.94 for mild steel
h h(mm)
The impact energy DA that produced rupture in these tests went into local
shear deformation in a diffuse shear band and fracture energy. For 5 mm thick
specimens the fracture energy was roughly twice the energy dissipated by plastic
deformation in the shear band, but this ratio depends on the extent of shear band
diffusion; i.e. on beam depth, impact speed and displacement constraints that
focus shear band development (see Johnson and Travis [1968]). Because of diffuse
plastic deformation the impact energy for rupture will usually be larger than that
measured with a notched specimen; nevertheless, this energy is not very different
from the work DA done by the fully plastic shear force during rupture; i.e.
DA '" Ybh 2 12.
In these tests the diffuse shear band was well developed and had propagated
through most of the depth h before fracture initiated at the periphery of the contact
region. The fracture at the junction with the colliding block began on the impact
surface where the junction is both sheared and stretched due to flexure of the
beam. After initiation the fracture propagated through the depth while penetration
continued; the propagation was quite rapid in aluminum but more prolonged in
mild steel. Accordingly, the variation of shear force Qp during the fracture process
can be sketched as the chain line in Fig. 5.23.
Despite the crude model of shear rupture, the energy requirement 2DA /Ybh 2
shown in Table 5.2 indicates that this approximation for rupture energy is reasonable
for these tests. Of course if the beam were not notched, the relative extent of the
deforming zone would increase with depth and the nonfracture part of the dissipated
energy at rupture would also somewhat increase.
5.4 Effect of Large Deflection 149

5.4 Effect of Large Deflection

5.4.1 General Considerations

The preceding analysis has been based on the assumption that the deflections of
the structure remain infinitesimally small. This means that the equations of motion
are written once for the initial configuration and that the geometric changes are
small enough that the configuration does not change noticeably. For deflections
that become large, however, the equations of motion must reflect the current
configuration at each instant of motion. When rotations of some sections become
large, it is no longer sufficient merely to consider the transverse momentum
distribution - the axial momentum also comes into play. It should be recognized,
however, that the rigid-perfectly plastic theory is applicable only if the curvatures
are much larger than the elastic curvature. In order to apply the rigid-perfectly
plastic constitutive model to practical engineering problems, it is important to
take the effect of large deflections into account. Some experiments on dynamically
loaded cantilevers are discussed in Chap. 7. In these experiments the final tip
deflection is of the order of the length of the specimen, while the final rotation of
the tip is as large as 60° or more. Notable discrepancies between the experimental
measurements of final tip deflection and rigid-plastic predictions have been reported
for these specimens that have undergone large deflections. The neglect of the
geometrical changes owing to large deformations is partly responsible for these
discrepancies together with the effects of strain-rate and strain hardening.
In Sect. 5.6 it will be seen that the rigid-perfectly plastic idealization is valid
only if the ratio R of input energy to the maximum elastic deformation energy
which can be stored in the structure is much larger than unity. For an impulsively
loaded cantilever this requires
R = !..GV2 2EI » 1 (5.140)
2 ° M2L
p
For cantilevers of rectangular cross-section, the moment-of-inertia 1= bh 3 1l2 and
the fully plastic moment M p = Ybh 2 14. Here, the total elastic capacity is
represented by the bending moment M p. For both small and large deflection
analysis the magnitude of the final rotation of the tip of the cantilever leAfl
equals the ratio eo between the impact energy and the fully plastic bending moment,
i.e.
(5.141)

Consequently, the condition for the rigid-plastic idealization (5.140) requires

R = leAf I( ~:2) = leAf 13:y » 1 (5.142)

where the nondimensional curvature at yield k y == K:yL is given by

MyL 2MpL 2YL
k -------- (5.143)
Y - EI - 3EI - Eh
Typically for a ductile metal cantilever, ElY = 500, so that (5.142) results in
leAfl = 0.003(Llh)R
150 Second-Order Effects On Dynamic Response

where R» 1. For a fairly modest slenderness ratio hi L = 1/30 and an impact

= Ie 1=
energy ratio R 5, this results in a tip rotation AI 0.45 rad "" 26°, which is a
large deflection. This rough estimate indicates that for a slender cantilever, validity
of the rigid-perfectly plastic idealization implies that the final deflections are
large.

5.4.2 Large Deflection of Impulsively Loaded Cantilever

The large deflection analysis in this section follows the method presented by
Ting [1965], but for consistency, the nondimensionalization adopted is that used
in the remainder of the book. In this analysis, all the assumptions in Sect. 4.5 are
applicable other than assumption (5) on small deflection; in particular, the axial
elongation and axial force components are neglected.
Consider a uniform rigid-perfectly plastic cantilever with a particle of mass G
attached at the tip; the particle is instantaneously given velocity Vo at time t = 0
(see Fig. 4.22). Thereafter a typical state of the cantilever at any time t is as
shown in Fig. 5.24a; all deformation is localized in a discrete plastic hinge at a
distance A = AL from the tip. Figure 5.24b shows the nondimensional arc length
s = SI L measured from the tip A towards the root B, and coordinates ~, 1] that
have an origin at hinge H. Since axial elongation is neglected, the hinge H is
located at s = A, and the root B occurs at s = 1. As in Sect. 4.5, the following
nondimensional quantities can be defined
_ ~ _t _ Vo _G
To = L ~ M p , r = To' v° = LI To ' r = pL

The position of the travelling hinge, A, is a function of time r. If A(r) is a

monotonously increasing function then A and r are uniquely related; in this case
the spatial variable A can be taken as the independant variable rather than r.
Therefore, when the hinge is at A the inclination angle e(S,A) at s is (see Fig.
5.24b)

(5.144 )

where k(s) = 1CL is the nondimensional curvature of the cantilever at s. Within the
framework of rigid-perfectly plastic theory, at any instant the deformation in the
cantilever takes place only at a plastic hinge. Therefore, at a given section s,
k(s)=Ofor S>A.
Consider a point in segment AH between the tip and the hinge (0::; s ::; A).
Referring to the Cartesian coordinates ~, 1] shown in Fig. 5.24b, one finds

Vo
14
A
11 = >"L
L
-I t
f)
(S=>..)
H
(S=l)
B

[j

a b

Fig. 5.24 Large deformation mechanism; (a) hinge location; (b) coordinate .;til).
5.4 Effect of Large Deflection 151

~(S,A) = l~os(-e(z'A)) dz (S.14S)

1](S,A) = l~in(-e(z'A)) dz
Then, the first moment of mass for the deformed segment AH about the ~- and 1]-
axes, and the second moment of mass about the plastic hinge H are respectively,

f~(A) == fo~(S'A) ds + }{(O,A) (S.146a)

A
f1)(A) - fo 1](S,A) ds + m(O,A) (S.146b)

f:{~2(S'A)+1]2(S'A)}ds
fH(A) == + r{~2(0,A)+1]2(0,A)} (S.146c)

where the mass ratio r = GI pL.

Using Eqs (S.146), the nondimensional equations of conservation of transverse
momentum and moment of momentum about the original position of the tip for
segment AH are,
-fg(A)8 A (A) (S.147)

-fH (A)8A (A) (S.148)

The angular velocity 8A (A) of the deformed segment about the plastic hinge is
derived from Eq. (S.144)

e. (A)
A ==
de(O,A)
---''-'---=-
dr
de(X,A)
dr
= -k(A)A, 0:::; x:::; A (S.149)

In addition, the current kinetic energy plus the energy that has been dissipated at
the hinge equal the impact energy; that is,
1
-8;(A)fH(A) +
fAk(z)dz = ~
r 2 (S.IS0)
2 0 2
where r v; 12 = eo = Ko 1¥ p is the nondimensional initial or impact energy. To
eliminate one variable, e A , the equation of translational motion (S.147) can be
substituted into the expression for conservation of energy (S.lS0),

rv; fH(A) + r~(z)dz = rv; (S.lS1)

2 fl(A) Jo 2
Direct differentiation of first and second moments of mass about H, Eqs (S.146),
with respect to A gives
dfg(A)/dA A+ r - k(A)I1) (A)
df1)(A)/dA k(A)fg(A) (S.IS2)

dfH(A)/dA = 2I~(A)

Using Eq. (S.lS2), the curvature k(A) at the plastic hinge can be obtained by
differentiating Eq. (S.lS1) with respect to A; i.e.
2 2 (A + r)IH(A) - Il(A)
k(A) r Vo 3 2 2 (S.lS3)
Ig(A) +r V o I H(A)/1)(A)
152 Second-Order Effects On Dynamic Response

Similar to the small deflection rigid-plastic analysis (Sect. 4.S), the transient
or travelling hinge phase is called phase I. During this phase, the relation between
time and the hinge position can be obtained by eliminating the angular velocity
from Eqs (S.147) and (S.148):

r = yv [A, - IH(A,)] (S.lS4)

o I~(A,)

When A, = 1, the travelling plastic hinge reaches the root of the cantilever, and
phase I terminates. According to Eq. (S.IS4) this happens at an instant

r 1 = yv
o
[1_ (I)]
/H
I~ (1)
(S.ISS)

After this instant, the cantilever rotates about a stationary hinge at the root until
the remained kinetic energy is completely dissipated by the fully plastic moment
that does work in resisting this rotation. This stationary hinge or modal phase is
called phase II.
The transition from phase I to phase II occurs when A, = 1; at this instant the
energy balance gives

y v; IH (1) + r1k(z)dz
2
= yv; (S.lS6)
2 Il(l) Jo 2
The remaining kinetic energy at time r] is given by the first term on the left-
hand side. This remaining kinetic energy is dissipated by deformation at the root
during phase II. Hence, when motion ceases, there is a final rotation angle at the
root,
2 2
e - _ y VO IH (1)
(S.lS7)
Bf - 2 Il(I)
According to (S.147), phase II begins with the entire cantilever rotating at an
angular veloci~y
eA(l) = _ yVo (S.IS8)
I~(l)
Hence, the duration of phase II is found to be

r2 - rl = 21eB fl IH(l)
= yvo-- (S.IS9)
yvo/I~(I) 1~(I)
The total response time rf is obtained as the sum of Eqs (S.ISS) and (S.IS9)
rf = r 2 = rvo = Pf (S.160)
where Pf = yVo is the impulse applied at the initial instant. Note that the response
time in this large deflection analysis is exactly the same as that obtained in the
small deflection analysis, Eq. (4.118).
At time r f' all motion terminates and the cantilever reaches its final deformed
state. A further study of the energy equation (S.IS1) confirms that the final inclina-
tion angle at s can be expressed as
22)
e (s) = _y
IH(s Vo (S.16l)
f Il(s) 2
Thus, the final inclination angle at any point s is equal to the nondimensional
5.4 Effect of Large Deflection 153

kinetic energy in the cantilever when the plastic hinge reaches the section s. In
particular, the final angle at the tip is equal to the nondimensional input energy;
i.e.
K
- OAf = eo == MO (5.162)
p
which is exactly the same relation as that obtained in the small deflection analysis.
Finally, note that integration of Eq. (5.161) along the arc length s provides the
final deformed shape of the cantilever for the case of large deflection.
At this point the formulation of the large deflection analysis of the cantilever
subjected to impact load is completed. The solution can be obtained by a numerical
scheme as discussed in detail by Ting [1965]. In brief, Eqs (5.144)-(5.146) and
(5.153) provide a scheme of successive approximation for the calculation of the
curvature k(A). Simultaneously, the values of 0, I~, 11/ and IH are all found for a
sequence of values for A and r. This process is continued until the hinge reaches
the root A = 1. Thereafter energy, inclination angle and final deflection can all be
determined from the expressions above.
For example, final deformed shapes of cantilevers calculated by this numerical
scheme are shown in Fig. 5.25. The solid lines are the results based on the large
deflection analysis, while the broken lines are obtained with the small deflection
analysis. It is seen that the difference between them increases as the nondimensional
input energy increases.

5.4.3 Methods of Approximating Large Deflection Effects

It is worthwhile noticing that in the small deflection analysis each section is
assumed to move only vertically, so that the deformed cantilevers (the broken
lines in Fig. 5.25) appear to be longer than their original length. This causes
significant discrepancy in the final shapes. On the other hand, this fact implies
that the small deflection analysis in Chap. 4 can be improved by taking Eq.

Fig. 5.25 Final deformed shapes of impulsively loaded cantilevers. - - large deflection analysis; - - -
small deflection analysis. (a) y = 0.5, eo =0.5; (b) y =0.2, eo =1.0; (cl y =0.2, eo = 7r / 2.
154 Second-Order Effects On Dynamic Response

18,$(5) lje~7-- -----------

Fig.5.26 Distribution offinal inclination 81 along cantilever. Solid lines result from large deflection analysis,
t
with various specified values of eo; 8 results from an approximation, refer to (5.163).

(4.131) as the distribution of the final inclination angle rather than the function
w/(x) that was obtained in producing the final deformed shape of the cantilever.
Taking arc length s instead of x in the small deflection expression (4.131) results

r
in
8;(s) =
v;
3y+s
(5.163)
-6' (1+ 2Sy
where superscript s designates the small deflection analysis. A numerical example
shown in Fig. 5.25c for the case of y = 0.2 and eo = rc/2, results in the dashed
line that is close to the deformed shape obtained by the large deflection analysis
(the solid line).
Another way of comparing the improved small deflection solution (5.163) to
the exact large deflection solution is shown in Figs 5.26 and 5.27. For small eo'

0·25...-----,------,--,-----,---,

0·20
18f (511
1---
18, (511
0.15 eo = 37T/1.

0·10
008 eo = 7T/2
006
0·04 eo = 1-0
eo = 0-5
2 3 5
sir
t
Fig. 5.27 Error of approximate 8 given by (5.163) compared with large deflection result 81 ,
5.4 Effect of Large Deflection 155

'·5-

,·0
small deflection solution ~

~
0·5 ~
A

Fig. 5.28 Variation of hinge position with time. - - large deflection solution; - - - small deflection
solution.

9i(s) is a good approximation of 9 j (s). The approximation is also better for any
eo if (slY> is either small or sufficiently large. From Fig. 5.27 it is seen that (1)
for the case of large mass ratio y» I where (sly)« 1, the improved small
deflection inclination 9i(s) is an excellent approximation of 9 j(s); but (2) for
the case of small mass ratio, 9i(s) is most inaccurate in the segment 2y < S < 4y.
Figure 5.28 shows the variation of the hinge position with time. Again the
broken line is based on the improved small deflection solution (5.163). For a
small impact energy ratio eo' the large deflection analysis approaches the small
deflection analysis in Sect. 4.5. If an initially straight member has only small
deflections, then
cos 9 '" 1, sin 9 '" 9, s '" x (5.164)
Hence, the coordinates of a point in segment AH become
~ '" A-X, 11 '" l-xlA (5.165)
and the inertia parameters defined by Eqs (5.146) are recast as

(5.166)

Employing Eqs (5.166), the equations of conservation of transverse momentum

(5.147) and moment-of-momentum about the tip (5.148) lead to the same results
as those obtained with the assumption of small deflections (4.105) and (4.106).
The equation for curvature (5.153) asymptotically approaches (4.130) which resulted
from the small deflection analysis. If the deflection is small the final deformed
shape of the cantilever is given by (5.163). To sum up, the large deflection analysis
presented in this section asymptotically approaches the small deflection analysis
in Sect. 4.5, provided the deflections remain small.

5.4.4 Effect of Centripetal Acceleration on Bending Moment

Distribution
The large deflection analysis of Sect. 5.4.2 determines the dynamic response of
the cantilever under impact loading, and calculates the variation of the velocity
156 Second-Order Effects On Dynamic Response

field with time. The original analysis by Ting [1965], however, did not calculate
the distribution of the bending moment in segment AH and so it did not determine
whether this solution satisfies the yield condition in the entire segment AH at all
times.
In the large deflection analysis there is another term - centripetal acceleration
- that must be added to the acceleration field as a result of the rotation of
segment AH about the plastic hinge H. The magnitude of the centripetal acceleration
is determined by
.. _ e' 2 1;:2 2
(5.167)
W, - A 1/'0 + 1]
where sand 1] are the coordinates of a section in segment AH. This is a radial
component of acceleration so it does not affect the bending moment at the plastic
hinge H. Figure 5.29a illustrates the magnitude of the radial acceleration tV,
along the arc length s, while the direction of this term of acceleration is shown in
Fig. 5.29b. Without further calculation, these figures show that the inertia related
to the centripetal acceleration of segment AH (including the tip mass) will diminish
the bending moment along segment AH (i.e. in 0 < S < A), but it does not change
the bending moment at the tip (M = 0) or plastic hinge H where M = Mp. Thus,
the distribution of the bending moment in the cantilever can be sketched as shown
in Fig. 5.29c, where the broken line pertains to the small deflection analysis and
the solid line pertains to the large deflection result. It is clear from this figure
that the appearance of centripetal acceleration during large dynamic deflections
of the cantilever does not cause a violation of yield condition IMI.::; M p in any
section of the cantilever.
A similar problem is related to impact and bending of a rigid-plastic fan blade;
the blade may be regarded as a cantilever that is subjected to continuous acceleration
in the radial direction due to the fan rotation. The blade may also be subjected to
transverse impact at the tip. This problem was analyzed by Stronge and Shioya
[1984] and Shioya and Stronge [1984], who took both transverse and centrifugal
forces into account in determining the moment distribution and the position of
the travelling hinge.

.. 5

a ..
5

c
H B
~ ____ I I
~ __ --;.- (S=A) (5 =1)

A ..
(5 =0) wr
b
Fig. 5.29 Centripetal acceleration and its effect: (a) magnitude distribution of the radial acceleration w,; (b)
direction of Wr ; (c) distribution of bending moment. - - - small deflection analysis and --large deflection
analysis.
5.5 Effect of Elastic Deformation 157

The conclusion that centrifugal force does not alter the basic pattern of dynamic
deformation of a rigid-plastic beam is valid only for a yield condition that is
independent of the axial force. In general the inertia related to centripetal
acceleration of segment AH (including the tip mass) produces an axial force in
segment AH; this force increases from the tip to the hinge section H. Therefore,
the argument that centripetal accelerations during large dynamic deflections of a
cantilever are unlikely to cause a violation of yield condition is correct only if
the effect of axial force on yielding is negligible; i.e. if the axial force is small in
comparison with the fully plastic yield force for the section at hinge H.
The cantilever analyzed here has large flexural deformation but it has been
assumed that the deformation is inextensional. In some other structural members
such as axially constrained beams, circular plates and shells the axial or membrane
forces unavoidably induce stretching as the deflection becomes large. In these
cases, the effects of the axial or membrane forces on the yield condition and the
equations of motion must be considered. This coupling of forces by the yield
condition has been discussed by many authors, e.g. Symonds and Mentel [1958],
Jones [1971] and Yu and Stronge [1990].

5.5 Effect of Elastic Deformation

5.5.1 General Considerations

Most ductile metals exhibit elastic-plastic behavior rather than the rigid-plastic
idealization that has been considered thus far. Although the largest elastic strains
are limited (typically ey '" 0.002 for structural metals) the elastic deformation is
always a precursor to subsequent plastic deformation. If a structure is subjected
to intense dynamic loading, the first phase of response is always elastic. Initially
stresses increase as deformation spreads away from the loaded region and soon
some sections have stresses as large as the yield stress; plastic deformation then
begins at these sections. The regions with large stresses move during dynamic
deformation - sometimes these regions disappear and reappear. After plastic
deformation has dissipated most of the input energy in an elastoplastic structure,
the structure is left in a final state of elastic vibration. Ultimately this is also
damped out and the structure comes to rest in a state of permanent deformation
with an associated field of residual stress. The rigid-plastic idealization considered
earlier in this book glossed over these details. It is possible that there is a synergistic
effect that was overlooked.
Due to the complex intermingling of elastic and plastic behavior in the structural
response, analytical methods have not made any headway in dealing with effects
of inertia combined with distributed plastic deformation. Thus we turn to numerical
schemes such as finite element and finite difference to gain some insight into the
effect of elasticity on subsequent plastic deformations. The aim is to identify
conditions where the rigid-plastic model accurately represents the response of
elastoplastic structures.
The rigid-plastic idealization for structural deformation in response to impulsive
loading is based on the assumption that at every section the plastic deformation
158 Second-Order Effects On Dynamic Response

is much larger than the elastic deformation. If the total input energy imparted to a
structure by dynamic loading is much larger than the maximum elastic strain
energy that the structure is able to store, then most input energy is dissipated by
plastic deformation of the structure. The maximum elastic energy capacity U:,ax
can be compared with the total input energy imparted by dynamic loading E in ;
thus the specific input energy is defined as an input energy ratio, R, where

R == ~ (5.168)
U:,ax
If R» 1 the rigid-plastic structural idealization usually provides a good approxi-
mation for deformation of a structure subjected to a brief but intense loading
pulse. By brief we mean short in comparison with the rigid-plastic structural
response period. Quantitative analysis of the accuracy of the rigid-perfectly plastic
approximation will be discussed in Sect. 5.6.
In the present section the problem of an elastic-plastic cantilever struck at the
tip by a particle with mass G is examined to assess how elastic effects modify
dynamic response to impulsive loading. The reasons for choosing this example
are:
1. the solution can be compared with that of a rigid-perfectly plastic cantilever
(see Sect. 4.5 or Parkes [1955]).
2. a few experimental results are available for comparison (see Chap. 7).
3. no axial force is induced by large deflections.
For a cantilever of length L, the input energy ratio R becomes
2
R - Ko - GVo EI (5 169)
- U max - M2L .
e p
where Vo is the initial velocity at the tip, EI is the elastic stiffness and Mp is the
fully plastic moment.

5.5.2 Mass-Spring Finite Difference Structural Model (MS-FD)

For elastoplastic structures, equations of motion (and of static equilibrium) are
nonlinear because of both the material properties and the effects of large deflection.
Methods of calculating dynamic response of nonlinear continua (e.g. beams, plates,
shells) can be based on either a finite difference or a finite element discretization
of the structure. Here a mass-spring finite difference model (MS-FD) is developed
for transient response of a cantilever to impact; this is used to investigate effects
of elastic deformations on the subsequent plastic deformations. The results are
compared with an alternative finite element model of the same problem. This
comparison elucidates effects due to short wave length flexural elastic waves that
are neglected by the finite difference model formulation.

MS-FD Model and Its Formulation At an arbitrary section of a beam, the

effect of elastic deformation preceding any plastic deformation can be considered
by discretizing the beam into a number of rigid links connected by elastic-plastic
hinges. To calculate elastic-plastic deformation of impulsively loaded cantilevers,
Hou et al. [1995] adopted a finite difference discretization based on equations of
motion for small deflection. The constitutive equation for this model is an elastic
strain hardening relation between bending moment and relative rotation angle.
5.5 Effect of Elastic Deformation 159

[j ~Ode O..MJ} 2 __ _ _ _n -1
..N'vI.

~ ~~~~L~=-fn:!::;~
Lin

~I
a

Fig. 5.30 Mass-spring finite difference (MS-FD) structural model: (a) discretization of a cantilever; (b)
positive sense of deflection, shear force and bending moment.

Strictly speaking, this idealization is correct for an ideal sandwich Euler-Bernoulli

beam, but generally it is a useful approximation for compact cross-sections. This
phenomenological modeling is convenient for a semi-analytical approach because
it naturally brings forth nondimensional variables that represent the system. The
deformation approximations, however, limit the present formulation to small
deflections in response to transverse loads. A more elaborate multilayered beam
discretization by Hashmi et a1. [1972] is more versatile but less easy to interpret
for comparisons with the analytical results in the preceding chapters.
The MS-FD model is sketched in Fig. 5.30a. The present discretization separates
the cantilever into n rigid elements of equal length connected by flexible joints or
node points. The mass of the cantilever pL is concentrated at the nodes; each
node has mass pL/ n. The nodes are numbered consecutively, beginning from the
tip. The colliding mass G is attached to the zeroth node at the tip. Hence, the
model consists of (n+ 1) lumped masses connected by n massless rigid links of
length Un. Since both shear and axial deformations are neglected, only flexure is
admissible; this is represented by relative rotation between adjacent rigid links.
At each node the transverse nodal displacement W;(t) (i = 0,1,,, ·,n) is a function
of time; the tip has deflection Wo(t). If the links are also labelled so the ith link
joins nodes i -1 and i, then under the assumption of small deflection, rotation 8i
of each link is given by
8i = n(W;_I-W;)IL, i=1,2, .. ·,n (5.170)
Thus at the ith joint the relative rotation VIi between the adjacent links i and
(i+ 1) is
li'i == 8i - 8i +1 = n(W;+1 - 2W; + W;-I)/ L, i = 1,2, .. ·,n-l (5.171a)
Vln = 8n (5.171b)
This relative rotation between each pair of links is resisted by elastic-plastic
stiffness at each joint, which reflects the flexural rigidity of the cantilever. This
160 Second-Order Effects On Dynamic Response

mi :; Mi/My

Fig. 5.31 Elastic-linear hardening relation between moment and relative rotation angle.

stiffness is portrayed in Fig. S.30 by an elastic-plastic rotational spring at each

joint. The joint is elastic for small rotations but the elastic range is limited by a
'yield' rotation angle 'I'y. For 'I'i > 'I'y a joint exhibits linear 'strain' hardening.
Thus, as shown in Fig. S.3l, for increasing relative rotation tVi > 0, the moment
Mi at a node is given by
C'I'i' 'I'i ::; 'I'y
M i ('I'i) = { (S.l72a)
My + aC('I'i - 'I'y), 'I'i > 'I' y, tVi > 0
If the largest rotation at a joint '1':
exceeds the yield rotation, > 'I'y, and '1':
subsequently the bending moment is decreased, the joint unloads elastically. The
moment-rotation relation for unloading is
M;('I'i) = M i ('I':)-C('I': -'I'i)''1': > 'I'i > '1'; -2'1'Y (S.l72b)
where the range of applicability is limited to 'I'i > '1'; - 2'1' y by the Bauschinger
effect for kinematic strain hardening.
This discretization represents the inertia by a particle with mass pLI n at each
node as shown in Fig. S.30b; this specifically neglects the effect of rotary inertia
and thereby decouples displacement variables in the equations of motion. The
nodes are connected by rigid links. Equations of motion for this model give nodal
shear force Qi and bending moment Mi

(S.173a)

2
Q. _ Q.
I 1-
1 = pL d W i -
n dt2
l i = 2,3,.· ',n (S.173b)

L
Mi-Mi- l = --Qi' i=1,2,.··,n (S.174)
n
For a cantilever the boundary conditions are
Qo = Mo = 0, Wn = 0 (S.17S)
Impact by a colliding body that strikes the tip transversely at a velocity Vo results
in initial conditions
dWo Vo
Wo = 0, dt ~1 + (pLl2nG)
(S.176)
Wj = dWj = 0, i = 1,2,.· ',n
dt
where at impact the speed of the particle at the tip and the speed of the colliding
5.5 Effect of Elastic Deformation 161

mass G instantaneously change to a common speed Vo I ~1 + (pLi2nG) obtained

from conservation of transverse momentum.
Using matrix notation, the above equations may be recast in condensed form
as
{e} = ~AT{W} (S.170')
L
{1jI} = AT {e} (S.171')
2
A{Q} = pLB~{W} (S.173')
n dt 2
L
A{M} = --{Q} (S.174')
n
where
{Q} == (Qj,Q2,···,Qn)T, {M}

{W} == (Wo,H),··,Wn_j)T (S.I77a)

o ny +t 0
-I 0
A - , B - (S.I77b)
o 0
-1 0

Energy Ratios and Nondimensional Formulation To nondimensionalize the

formulation, define

Ty == L~ My
pL , r ==
Ty '
() - ~(
dr
)

{w} - ~ {W}, {q} == ~{Q},

My
{m} __1_{M}
My

ljIy -
MyL
, IjI -
IjI
() --
e (S.178)
nE/ ljIy ljIy
G Ko GV 2
- ~
_o_
Y - - eo - , R
pL Mp 2Mp max Ue
where R denotes the ratio of the initial kinetic energy to the maximum elastic
deformation energy that can be stored in the cantilever, while eo is the ratio of
initial kinetic energy to fully plastic bending moment M p at any section. These
ratios relate input energy to either the global elastic strain energy capacity U:?ax
or a local elastic capacity of a section My = M pi ¢>m . The latter energy ratio eo is
useful in making comparisons with rigid--perfectly plastic solutions. The
cross-section shape factor for pure bending ¢>m = M pi My and yield curvature
/( y can be used to relate these energy ratios if strain hardening is negligible and
the cross-section is uniform, eo I R = ¢>mL/(y 12 = ¢>mL£y I h. In particular, when an
ideal sandwich cross-section is considered, ¢>m = 1 and My = Mp.
162 Second-Order Effects On Dynamic Response

By using these nondimensional variables, Eqs (5.170'), (5.171'), (5.173') and

(5.174') are recast in nondimensional form:
{8} = ~AT{w} (5.170")
lfIy

{iii} = AT{8} (5.171")

A{q} = !B{w} (5.173")

n
1
A{m} = --{q} (5.174")
n
Combining these equations gives
n3
{iii} = --F{m} (5.179)
lJIy
where
(ny+t)-1+5 -4
-4 6-4
-4 6
F _ (A T)2B-l(A)2 (5.180)
6 -4
-4 5 -2
-2
The constitutive relation between moment and relative rotation angle, (5.172)
can also be recast in nondimensional form (see Fig. 5.31)
Vii'
{ (5.181a)
1+ a(Vii -1) Vii >1, Vii ~o

mi = Vii - (1- a)(Vit -1), (Vii' - 2) -::; Vii -::; Vii' (5.181b)
with i = 1,2, .. ·,n.
The present model consists of n nonlinear springs at the nodes, so when all
springs have rotated through the yield angle the model has a maximum elastic
deformation energy capacity
max n
Ue = -MylJly (5.182)
2
The energy capacity U~ax given by (5.182) can be equated to the maximum
elastic deformation energy which can be stored in an elastic-plastic cantilever,
U~ax = M;Ll2EI. Thus the elastic limit lfIy is related to compliance of a uniform
cantilever by
l/>mMpL
lJIy = nEI (5.183)
where l/>m is the cross-section shape factor for bending. Accordingly, the elastic
spring constant C = My/lfly = nEIl Ll/>;".
The relationship between the energy ratios R and eo obtained from (5.182) is
2U_e _
_
max
= 2l/>m eo
lJIy = (5.184)
nMy nR
5.5 Effect of Elastic Deformation 163

so Eq. (5.179) can be rewritten as

n4 R
{iii} = ---F{m} (5.185)
2tPm eo
Equations (5.185) and (5.181) are a set of2n equations for 2n unknown functions
Vii and mi (i = 1,2,. ··,n). Since a linear strain hardening relationship between
moment and relative rotation angle is employed, if the hardening coefficient a
tends to zero this structural representation approaches that of an elastic-perfectly
plastic ideal sandwich beam. Therefore, the shape factor is taken to be tPm = 1 in
all calculations.

Solution of Initial-Value Problem The nondimensional form of the initial

conditions (5.176) at r = 0 can be written as

Wo = 0,
(5.176')
Wi = Wi = 0, i = 1,2,.· ·,n
Hence the second order differential equation (5.185) has the following initial
conditions:
ViI 0 ViI = n 2 R[ eo (2y+n- l ) tl2
(5.186)
...:....
lfIi 0, lfIi = 0, i = 2,3,·· ·,n
An examination of Eqs (5.181), (5.185) and the initial conditions (5.186)
indicates that there are four nondimensional parameters y,eo,R and a, together
with the number of links, n. These five parameters specify the problem. With
initial conditions, a solution can be obtained by numerical integration of the initial-
value problem using a Runge-Kutta procedure. Calculations confirm that when n
is sufficiently large, say n;::: 20, then the qualitative features of the solution are
insensitive to n.

5.5.3 Timoshenko Beam Finite Element Model (TB-FE)

Model and Parameters In recent years some large general-purpose finite element
numerical codes have been available for calculating the dynamic deformation of
elastic-plastic structures in response to arbitrary loading. Both Symonds and
Fleming [1984] and Reid and Gui [1987] have reported finite element analyses of
impulsively loaded elastic-plastic cantilevers. These analyses employed a finite-
element code ABAQUS and assumed elastic-perfectly plastic material behavior.
The principal approximations of these analyses (designated as the TB-FE model)
are summarized in Table 5.3. This table compares the TB-FE and MS-FD models.
The major differences between these models are:
1. The TB-FE model employs an elastic-perfectly plastic constitutive relation
between stress (J and strain e; while the MS-FD model employs an elastic-
linear hardening constitutive relation between moment M and relative rotation
angle lfI.
164 Second-Order Effects On Dynamic Response

Table 5.3 Comparison of analytical finite element and finite difference models for transverse impact on
cantilever
Rigid-plastic theory TB-FEmodel MS-FDmodel
E-B beam Timoshenko beam E-B beam
Sect. 4.5 Sect. 5.5.3 Sect. 5.5.2
Elasticity (Sects 5.5, 5.6) 0 x x
Strain-rate effect (Sect. 5.1) 0 0' o
Strain hardening (Sect. 5.2) 0 0 x
Shear, elastic 0 x o
Shear, plastic (Sect. 5.3) 0 0 o
Rotary inertia of beam 0 x o
Rotary inertia of tip mass (Sect. 5.3) 0 0 o
Large geometry changes (Sect. 5.4) 0 x o
Impulsive loading x x x
Note: x, present; 0, neglected.
• Except Example 3 in Reid and Gui [1987].

2. The TB-FE model includes the effects of elastic shear deformation and rotary
inertia of the section; i.e. it represents a Timoshenko beam of rectangular cross-
section. The MS-FD model, on the other hand, represents an ideal-sandwich
Euler-Bernoulli beam.
3. The TB-FE model includes the effects of large changes in geometry, while
the MS-FD formulation is limited to small deflections.

5.5.4 Dynamic Deformation of Elastic-Plastic Cantilever from

Impact
Choosing Numerical Examples In order to explore the elastic-plastic dynamic
behavior of cantilevers after impact, we look first at two typical examples; one
with a moderately large mass ratio and the other with a small mass ratio. For
these examples, numerical calculations of the dynamic response obtained from
the TB-FE and MS-FD models will be presented and compared. Then more
systematic results obtained with the MS-FD model will be discussed.
In order to compare with experimental data for a steel cantilever presented by
Bodner and Symonds [1962], Symonds and Fleming [1984] calculated the
deformations with the TB-FE model using the parameter values of Table 5.4.
This impact test had a moderately large mass ratio for the colliding mass r = 1.64 .
Symonds and Fleming calculated deformations resulting from values of R ranging
from 2 to 14.8; for this cantilever, these energy ratios correspond to values of Va
ranging from 4.7 to 12.9 ms- i .
A similar TB-FE model was used by Reid and Gui [1987] for finite element
calculations aimed at explaining how the presence of elastic deformation changes
the response history; in particular the development of plastic deformation along
the cantilever after it is hit at the tip. They calculated the response for three cases
of impact: (1) conditions identical to those considered by Symonds and Fleming
but with more detail on the distribution of bending moment and plastic work; (2)
a high velocity bullet impact test with a small mass ratio r that was performed
by Parkes [1955]; (3) the same high-velocity impact problem but including strain-
rate sensitivity in the plastic constitutive law (Cowper-Symonds). The latter analysis
5.5 Effect of Elastic Deformation 165

Table 5.4 Parameters used to calculate elastic-perfectly plastic response to impact (Symonds
and Fleming [1984], Reid and Gui [1987])
Symonds and Reid and Gui Reid and Gui
Fleming Example 1 Example 2
y=GlpL=Glp,bhL 1.64 1.64 0.0228
eo = Ko / Mp = GV'; I2Mp 0.228-1.69 1.69 2.93
R= 2KoEIIM;L 2.0-14.8 14.8 51.7
R"I 0.5-0.0676 0.0676 0.0193
Number of elements 28 28 24
Mp(N m) 16.5 16.5 24.8
G (kg) 0.336 0.336 0.0023
Vo (ms-I) 4.7-12.9 12.9 251.5

showed that with a small mass ratio the response was not qualitatively altered by
strain-rate effects.

Elastic-Plastic Behavior for Moderately Large Mass Ratio (r = 1.64) For a

small global energy ratio R = 2 the elastic-plastic response at early times shows
little resemblance to the travelling hinge phase of response given by the rigid-
perfectly plastic analysis in Sect. 4.5. Most plastic deformation develops near the
root after a period wherein elastic flexural waves transmit energy away from the
impact point.
For a moderately large energy ratio R = 14.8 and mass ratio r = 1.64 (i.e.
Example I in Table 5.4), Reid and Gui [1987] calculated the bending moment
distribution at various times. They described the elastic-plastic response of the
cantilever as the following sequence of events:
Phase 1: elastic-plastic bending wave. The moment distribution in this phase
shows many of the characteristics of wave propagation in Timoshenko beams.
The moment distribution is oscillatory in nature and the dispersion of the bending
wave is evident. At the tail of the disturbed region there exists a small region
wherein the bending moment is the fully plastic moment M p; this may be regarded
as a plastic hinge. As time progresses, the position of this short plastically deforming
segment moves along the cantilever in a manner reminiscent of the rigid-plastic
solution. In this example, the head of the bending wave reaches the root of the
cantilever after 0.28 ms.
Phase 2: reversed hinge at root - arrest of travelling hinge. After 0.28 ms the
elastic bending wave is reflected at the root; the interaction between the primary
bending wave and the reflected wave arrests the progress of the travelling hinge.
Thereafter the hinge tends to move back towards the tip. Meanwhile at the root
the bending moment oscillates and increases in magnitude until, at approximately
1.43 ms, a hinge with reverse rotation is formed at the root.
Phase 3: initiation of positive root rotation. While a certain amount of plastic
work is consumed in reversed bending at the root, the position of the deforming
segment slowly moves back towards the tip and the moment at the root is reduced
in magnitude. A positive root hinge forms after 4.9 ms. This continually changing
location of plastic deformation is illustrated in Fig. 5.32.
Phase 4: root rotation and elastic vibration. The plastic deforming region rapidly
shrinks to a hinge at the root; after approximately 62.0 ms only elastic vibrations
remain in the cantilever.
166 Second-Order Effects On Dynamic Response

Umsl rigid-plastic 51·52ms

(Parkes)
8
" = 7·27 ms
e M=MP=15'5N.m
(Parkes )
5 TB - FE { "M = 15·4 N·m
oM=-Mp= -15·5 N·m a.,,1
4

2
""" "."/
.1 "
iI
!..-".
/ I
r
o 0·25 0·5

f(ms) 80'35 ms

eo = 1.59
8
'I = 1.54 ',=7·27 ms
e M ~Mp
MS-FD { 0 R = 14.8
5 M~-Mp

4 I
I
2 8o
I
o

Fig. 5.32 Evolution of plastic deformation region along cantilever during dynamic response when
r = 1.64, eo = 1.69 and R = 14.8; (a) obtained from the TB-FE model; (b) obtained from the MS-FD model.

This example can also be calculated by means of the finite difference model
(MS-FD). For comparison with the solutions obtained by the TB-FE model of a
perfectly plastic cantilever, a small value of the strain hardening coefficient is
taken; e.g. a = 0.001. Apart from this, all other parameters (i.e. eo' y and R)
were taken from Example 1 of Table 5.4.
Figure 5.32 shows the time when plastic deformation is present at locations
along the cantilever. Here Fig. 5.32a was obtained from the TB-FE model, while
Fig. 5.32b was calculated with the MS-FD model. The two models show almost
identical behavior. Initially a focused region of plastic deformation travels away
from the tip. This seems similar to a travelling plastic hinge except that this
region disappears midway along the length when the bending moment at the root
reverses. Thereafter plastic deformation occurs mostly near the root in a mode-
like mechanism. Unlike the rigid-plastic moment distribution (Fig. 4.23c) which
5.5 Effect of Elastic Deformation 167

0·30 rigid - plastic theory

(0~X/L<1)
,-----x
===t~
o::c

I• •I
eo = 1-69
r = 1·64

o 0 (. xlL <:
0.964
0·05 II0 (. xlL ~
0·929
o 0 ~ xlL "
0·893
• 0 ~XIL .; 0·857

Fig.5.33 Energy dissipation in a segment adjacent to the root for r =1.64, eo =1.69; - - results obtained
from the TB-FE model; - - - energy dissipation at root according to the rigid-plastic theory.

has a uniform plastic moment M p throughout the region between the travelling
hinge and the root, here at any instant the bending moment is close to the fully
plastic moment for only short segments of the cantilever.
The development of reverse bending at the root begins as a consequence of an
elastic flexural wave that has travelled from the impact point; when this arrives
at the root it is reflected back towards the tip. The reflected elastic wave and the
focused region of plastic deformation meet near mid-length of the cantilever and
this disrupts the travel of the 'plastic hinge'. The subsequent transient development
of plastic deformation depends on the mass ratio r and energy ratio R. In any
case with this moderately large mass ratio r ,= 1.64, the early transient stage of
deformation has substantially less plastic deformation (permanent curvature) than
that given by the rigid-plastic theory. Figure 5.33 shows that the total energy
dissipated by curvature away from the root is substantially less than that predicted
by the rigid-perfectly plastic theory.
For the present example, calculations of the changing distribution of curvature
within the cantilever are shown in Figs. 5.34a, b, for the TB-FE and MS-FD
models, respectively. The elastic-plastic calculations predict that almost 87% of
the available kinetic energy is absorbed in the element at the root compared with
a prediction of only 72% from the rigid-plastic idealization. The peak values of
plastic work are found near the middle of the cantilever as a consequence of the
oscillation in the position of the plastic hinge in phases 2 and 3. Nevertheless,
the vast majority of the energy is dissipated in root rotation during a modal phase
of deformation that dominates the behavior for moderate or large y.

Elastic-Plastic Behavior for Small Mass Ratio (r = 0.0228) Example 2 in

Table 5.4 pertains to bullet impact on a cantilever so it has a very small mass
ratio r = 0.0228 and a high impact speed. Calculations of the response by the
TB-FE model, show roughly the same phases as in Example 1. The location,
duration and extent of plastic deformation for this example are plotted as a function
of time in Fig. 5.35a. Again the rigid-plastic analysis shows good agreement
168 Second-Order Effects On Dynamic Response

0·5 r =1·64 R =14·8

>.
iii
c
0·4

--- -- ---
Qj
"D ............ rigid-plastic 1Parkes')
c .......... ~
0
-§ 0'3 .....
e-
VI
VI
"6
>.
0·2
~
Qj
c
Qj 0·1

o 1 X/L
tip root

882
"(=164 R=14·8 111 )
06
3-92
?:- 0·5 (10)
iii
c

---- --
Qj
"D 0·4 ........ rigid - plastic (Parkes')
c
o ........ ....j
8.
VI
0·3 ........
VI
"D

'"
en 02
L.
Qj
C
Qj

05 1·0 X/L
root

Fig. 5.34 Change of curvature distribution along cantilever during dynamic response when
r = 1.64, eo = 1.69 and R = 14.8; (a) obtained from the TB-FE model; (b) obtained from the MS-FD model.

with the elastic-plastic solution during the period before the travelling hinge
halts midway along the beam. In the elastic-plastic solution the root rotation
initiates slightly later than the 3.lms obtained with the rigid-plastic idealization.
For a light tip mass r = 0.0228, the majority of the energy (91.75%) is absorbed
between the ends of the cantilever. As shown in Fig. 5.35b, the distribution of
plastic work approaches more closely that predicted by the rigid-plastic analysis
(given by the dashed line in the figure); however, a major discrepancy exists
around the central part of the cantilever where a high curvature region (or 'kink'
in the final deformed shape) appears. This is a result of the arrest of the travelling
hinge by reflected flexural waves.
5.5 Effect of Elastic Deformation 169
t (ms)
15·76ms
- - - rigid-plastic IPorkes') ~
4
" H~Hp"24.8Nm
TB-FE { h./'1:::2t.-7N·m
c.,
··i
OH=-Hp ='21.8Nm ":!~~!! t,
IP )
3
j • •" "_6e6..
C>C> c>"" arkes'
: •• I
~ I
2 2,: I
I I
I
o
"" I
I
I
1-0 x = Xli.
I root)

12 21·7J 20·2J
eo = 2·93
'If = O· 0228
:210
c
R = 51·7
o
'E
Q.
8
iii
Vl
is 6
u
.;:; rigid - plastic
Vl

~4 IParkes')
TB-FE
c elastic-plastic
())
E 2 15·27ms
())
a;
°o~--~~--- ~~~~~~~~~--
10 Xi[
Itip) Iroot)
b

Fig. 5.35 (a) Evolution of plastic deformation region along cantilever obtained from the TB-FE model
for y = 0.0228, eo = 2.93 and R = 51.7; (b) distribution of plastic dissipation.

Influence of Mass Ratio r and Energy Ratio R in Strain Hardening Cantilevers

The MS-FD model can be used to examine the influence of nondimensional
parameters such as mass ratio r and energy ratio R, on the dynamic behavior of
impulsively loaded beams. Some results obtained with this model are given here.
The distributions of bending moment along a strain hardening cantilever at
various instants are given in Figs 5.36 and 5.37 for mass ratios r = 5 and 1,
respectively. The case considered has a fairly small energy ratio R = 5; other
parameters are taken to be eo = 1, a = 0.1 and n = 20. Fig. 5.36 shows that for a
large mass ratio (r = 5) the bending moment at the root can reach a value well
beyond the initial plastic bending moment M p; this implies a mode-like response
since the root dissipates most of the input energy. For a smaller mass ratio r = I,
170 Second-Order Effects On Dynamic Response

H/Hp
HS-FD
n = 20
eo = 1·0
1·5 0: = 0·1
R = 5·0
r = 5·0
1·0

0·5

0·6 1·0
x = X/L
.......
-0·5

............... r =171·2 ._.-._.-. l' = 332·2

- - - - - 1'= 404·5 '--'- l' = 549·1
. _ ... _ - 1'=2213·8 - - - - T = 3341·0

Fig.5.36 Distribution of bending moment along strain-hardening cantilever, obtained from the MS-FD model;
r=5, eo =1, R=5 and 0:=0.1.

Fig. 5.37 shows that the largest bending moment in the transient phase is only
Mp; this may be regarded as a travelling plastic hinge that propagates from the
tip towards the middle part of the cantilever. This transient deformation field
becomes unclear after it begins to interact with reflected elastic waves returning
from the fixed end.
For a heavy colliding mass ( r = 5) and a very light colliding mass ( r = 0.025 ),
Figs 5.38a, b shows the change in position of plastic regions during the response.
It is seen from Fig. 5.38a that when r = 5 and R = 5, the plastic region first
appears near the middle of the cantilever for a very short time; then it reappears
at a region close to the root at a time later than that predicted by the rigid-plastic
solution as noted in Fig. 5.36. This behavior is more like a modal solution and
differs significantly from the rigid-plastic solution. When r = 5 but R = 15, the
plastic region also first appears at the middle part of the cantilever but there are
indications that the plastic region continues to move (backwards then forwards)
along the cantilever until it finally reaches the root at the time predicted by the
rigid-plastic solution. It is noted that both reverse yielding at the root and arrest
of the travelling hinge occur only in the case of R = 15; for smaller R, (e.g.
R = 5) the root section directly enters modal-type yielding without prior reverse
yielding.
As shown in Fig. 5.38b, when the mass ratio r is very small (e.g. r = 0.025),
the travelling hinge only reaches about 0.5-0.6 of the cantilever length before it
disappears. This implies that the response mode is quite similar to Pattern lIb in
5.5 Effect of Elastic Defonnation 171

MIMp

IX""""
1·0 r{~(-\
! "' I I r J\ \
0·5 ~!ll;1:: "'X'" '\ r:\
; ';
! \
! . \;
'\' I
,
:
0: ~ <. \ I I

\]
9- ..... \

- 0·5
............. 1: = 6·2 ·_·-·_·1:=16·8:
----1:=65·6 --'--1:=174·0
-1'0 - ' ' ' - - 1 : = 270.8 a

MIMp
n = 20 MS-FD
1·5 eo = 1·0
0: = a.'
R = 5·0
/
/-
r = 1·0
/.,., / - - ....
1·0
i . /
i '. /
j/ )!
0·5
/! / \
.9- . i \
\
0~~~~~~----1-----,·~\----+----
, _ / 0'4 O'S\
'.
\
-0·5 ·_·_·-·-"t=365·1 - - -1:=833.8"
- - . --1:= 5546 ---1:=8159 b

Fig.5.37 Distribution of bending moment along strain hardening cantilever, obtained from the MS-FD model;
.r=l, eo =1, R=5 and a=O.1

Sect. 5.5.5; that is, only a part of the cantilever undergoes plastic deformation
when the colliding mass is very light or the cantilever itself is very long (i.e.
those cases with a very small mass ratio r). Also, if r« 1, most of input
energy must be dissipated in curvature at interior sections and only a small fraction
of energy is dissipated by reverse flexure at the root.
Figure 5.39a, b demonstrates the influence of both mass ratio r and energy
ratio R, on the distribution of plastic dissipation dp in a strain hardening cantilever.
This dissipation is proportional to the final curvature of each section of the
cantilever. It is seen that more energy is dissipatt:d in the outer half of the cantilever
172 Second-Order Effects On Dynamic Response

1'15-FO:

R=5 {/o M':?-l'1p

n = 20
A M ~ -Mp eo='·0
0: = 0·001
0 .15 R=15 {.I'1::-l'1p
o 1'1 ~ -l'1p r = 5·0

_ _ _ rigid - plastic theory

( Parkes') (Parkes')
D·' r, =0·0958
••
••
•
0·05
•

°o~~~~--~--~~--~L---~~­

(tip)
a

••
.... (Parkes')
T, =0·0710

0·6

04

0·2 0' = 0·025

0·4 0·5 08 10 x = Xi/.

(root)
b

Fig.5.38 Evolution of plastic deformation region along the cantilever during dynamic response, obtained from
the MS-FD model, eo=l, a=O.OOI, R=5 and 15;(a) r=5;(b) r=O.025.

(near the tip) when the mass ratio y is smaller or elastic energy ratio R is larger.
The final deflected shape of the cantilever is insensitive to minor changes in
the distribution of energy dissipation, because the deflection is calculated by a
double integral of the curvature. The calculated results confirm that the final
deflected shape obtained by the present elastic-plastic model almost coincides
with that predicted by the rigid-plastic solution, provided mass ratio y is not
very small, see Fig. 5.40a. However, when mass ratio y is very small, e.g.
y = 0.025, a notable difference is seen in the middle part of the cantilever as
shown in Fig. 5.40b. Nevertheless the tip deflections predicted by the elastic-
plastic model and the rigid-plastic solution are almost the same. In Fig. 5.40b
5.5 Effect of Elastic Defonnation 173

-----)(=0·2
. _ - - . )(=1'0

't= 5·0
. I

f/
1/
. I

I
/
..-'" X= X1
0L---~0~'2----J---~L---~----1'~0---

tip root

I
a

4 /'1S-FD
n-20
eo~1'0
- - - - - R=lO
. _ _ _ • R = 50 !
j
ex =0·1
~ r=10 R=10 O /.
~ 3 I
~ ngid-perfectly plastic theory (R= 00)
o ........... eo=lO. ex =0. r=lO
~
CL

~ 2

0·2 06 1·0
root
b

Fig. 5.39 Distribution of energy dissipation along cantilever; (a) influence of mass ratio r ; (b) influence of
energy ratio R.

there is also a clear indication of reverse rotation at the root as a result of reverse
yielding during an early phase of the response.

5.5.5 Effect of Elastic Deformation at Root of Cantilever

Model with Rotational Spring at Root The preceding sections have shown that
elastic deformations near the root influence the development of plastic deformation
throughout the remainder of the structure if the energy ratio R is not too large;
this influence is due to elastic compliance that provides some springback. It seems
that elastic deformation near the root is the source of both the early termination
174 Second-Order Effects On Dynamic Response

0·2 -

0·" n = 20
eo = 1·0
0·6 Q: = 0·001
R=5
Y = 5·0

1-0 R=15
a

0·1
/--,--- rigid - plastic (Parkes

n = 20
J

eo = 1·0
ex = 0·001
Y = 0·025
b

Fig. 5.40 Final deflected shape of cantilever, eo = I, a = 0.001, R = 5 and 15; (a) r =5; (b) r =0.025 .

of the travelling hinge phase for cantilevers with high mass ratio and elastic
springback from the final modal phase of deformation. The single degree-of-
freedom mass-spring system proposed and studied by Symonds et a1. (see Sect.
5.6) refers solely to the modal motion of a structure. For a dynamically loaded
cantilever, this single degree-of-freedom system is in fact equivalent to setting an
elastic-plastic rotational spring at the root while the cantilever itself is regarded
as a rigid body. Combining this elastic-plastic rotational spring at the root and
the rigid-perfectly plastic beam of most previous analyses, Wang and Yu [1991]
proposed a simplified structural model which has some capacity for storing elastic
energy in the modal deformation configuration although the majority of the beam
is rigid~plastic. This model is certainly not as good as the previous numerical
models in representing details of elastic-plastic deformation, but it is much simpler
in analysis and reflects some basic features of the dynamic response of an elastic-
plastic structure.
The simplified structural model proposed by Wang and Yu [1991] is shown in
Fig. 5.41a; this consists of a rigid-perfectly plastic cantilever and an elastic-
perfectly plastic rotational spring at the root. The material is assumed to be rate-
independent, the influence of shear on yielding is neglected, the deflections are
assumed to be small, and the colliding body is treated as a particle with mass G.
5.5 Effect of Elastic Defonnation 175

VOl f. p. Mp ~
G[!]~A=======_==;;:~

I. a
·1
b

Fig. 5.41 (a) Simplified structural model with an elastic-perfectly plastic rotational spring at the root. (b)
Relation between moment and rotation angle at the root.

Under these assumptions, the dynamics of this structural model are almost the
same as those of the cantilever analyzed by Parkes [1955], except the bending
moment at the root is related to the root rotation by a moment-rotation angle
relationship for an elastic-perfectly plastic spring, see Fig. 5.41b. The spring has
an elastic constant k, a root rotation angle composed of an elastic part lfIe and a
plastic part lfI p' During loading the moment at the root MB is elastic for rotations
lfI ~ MplC and fully plastic Mp for lfI2 MpIC. For any cross-section of the
cantilever the fully plastic bending moment is Mp.
Based on the idea that all of the elastic strain energy of the structure is
represented by the elastic capacity of the spring, the spring constant C can be
easily determined. For a cantilever of length L and flexural rigidity EI, the maximum
elastic deformation energy that can be stored is U:,nax == LM; I2EI, while the
maximum elastic energy capacity of the root spring is U:,nax == M; 12C . Equating
the two expressions for U:,nax gives C == Ell L. In this model the energy ratio R
represents the part of the input energy that can be stored elastically by the spring.

Formulation For the model described above, impact at the tip causes plastic
bending of the cantilever as well as elastic-plastic deformation of the rotational
spring at the root. The former implies a travelling hinge, as shown in the rigid-
plastic analysis and Fig. 5.42. At any time t this plastic hinge is at an interior
section H, located a distance A(t) from the tip. The undeformed segment ahead
of the hinge, HB, has rigid-body rotation with angular velocity dlfl I dt about the
root B, where lfI is the angle between HB and the X-axis. In segment HB, the
velocity of a point X is

d; : (L-X), A~X~L (5.1S7a)

w
Fig. 5.42 Deformation mechanism with a travelling hinge at H and rotation at root B.
176 Second-Order Effects On Dynamic Response

If the heavy particle at the tip has velocity V relative to a reference frame fixed
on the rotating segment HB, then segment AH has a rigid-body rotation with
relative angular velocity VIA. From the sum of rotational velocities about the
travelling hinge H and the root B, point X in segment AH has transverse velocity

dW = dlfl (L-X)+V(I-XIA), 0::; X::; A (5.187b)

dt dt
The transverse acceleration throughout the entire length of the cantilever is obtained
by differentiating the velocity,
d 2lf1
2 -2-(L-X),
d W { dt (5.188)
df2 = d 2 lf1 (L-X)+ dV (l-XIA)+ dA VXIA2 ,
dt 2 dt dt
At hinge H the bending moment has a local maximum, so shear force QH = O. By
considering segments AH and HB respectively, the equations of motion are

MB - Mp + f L

A
d2lf1 2
P-2-X dX
dt
= 0 (5.189)

G d2W (0) + rA d 2 W dX =0 (5.190)

dt 2 Jo P dt 2

(5.191)

By substituting (5.188) into (5.189)-(5.191) and integrating, one obtains three

differential equations for lfI(t) , Vet) and A(t). To nondimensionalize these
equations, define
X W G
x == w A == A== r ==-
L' L' L' pL
(5.192)
T ==
oM'
pL L~ v ==
LIT'
V
r ==
T'
(') == ~( )
dr
p o o

Thus, Eqs (5.189)-(5.191) can be recast as

1 1 3 ..
m B - 1 + -(1-
3 /l,) III
't'
= 0 (5.193)

(r +~A + [r + ) A(I- ~)]v; +~vi o (5.194)

1 12 v+
-/l,
. (1
--- A)12 lfI+-V/l,/l,-1
. 1 11
/l, = 0 (5.195)
6 2 3 3
where
elastic or unloading
(5.196)
otherwise
The three unknown functions lfI(r) , v(r) , A(r) can be solved from Eqs (5.193)-
(5.195) with initial conditions
lfI(O) = 0, ljt(O) = 0, v(O) = vO ' A(O) = 0 at r = 0 (5.197)
5.5 Effect of Elastic Deformation 177

v
In the formulation (5.193)-(5.197), some quantities such as and A, are singular
at T = 0; however this can be removed by starting the integration at a very small
initial time To (0 < To « 1) with modified initial conditions. Three nondimensional
parameters govern the solution; i.e.
Y _ G e _ Ko _ 2GV; R == ~ = GVo2C
(5.198)
= pi- 0 = Mp
- Ybh 2 ' eU max
p M2
The first two, y and eo are the only significant parameters for the rigid-perfectly
plastic analysis in Sect. 4.5; while the stored energy ratio R- 1 is the part of the
input energy that can be stored in elastic deformation throughout the structure.

Patterns for Process of Dynamic Deformation If these three nondimensional

parameters are specified, a solution of the present problem can be obtained by
step-by-step numerical integration using a Runge-Kutta procedure. The results
indicate that the introduction of elastic deformation at the root notably changes
the dynamic behavior from that of the rigid-perfectly plastic idealization. The
latter always has a travelling hinge that transits from the impact point to the
support at the root of a cantilever; this transient phase is followed by a modal
phase of motion. In contrast, the present model exhibits two distinctly different
patterns of response depending on the stored energy ratio R- 1 •
Pattern I. A travelling plastic hinge moves from the tip towards the root, but
before it reaches the root, the rotation at the root exceeds the elastic range
lfI > MpIC. In this case the travelling hinge and the hinge at the root occur
simultaneously in the cantilever. With this pattern, the travelling hinge always
finally coalesces with hinge at the root of the cantilever; thereafter, deformation
enters a modal phase where the remaining kinetic energy is partly dissipated and
partly transformed into elastic vibration about the root.
Pattern II. A travelling plastic hinge moves from the tip towards the root, but at
a certain interior section the hinge motion halts and the hinge disappears; hinge
travel ceases while the root rotation is still elastic. After the disappearance of the
travelling hinge the root mayor may not enter into a plastic state depending on
the remaining kinetic energy. Thus Pattern II can be further separated into pattern
lIb and pattern IIa, depending on the largest root rotation.
Numerical examples have established that the stored energy ratio R- 1 has the
largest effect on the pattern of response. A small value of R- 1 results in pattern I
response which is similar to the rigid-perfectly plastic solution. When the stored
energy ratio R- 1 is very small, the root enters the plastic state soon after the
hinge begins to move from the tip. In this case the shear force at the hinge
vanishes so there is no angular acceleration of the segment between the hinge and
the root; this is just Parkes' solution. On the other hand, a large value of R- 1
results in Pattern II. Hence if the energy stored in elastic deformation at the root
is relatively large, only a part of the cantilever undergoes plastic deformation.
The variation of the hinge location with time is plotted in Fig. 5.43. It shows
that the value of stored energy ratio R- I has only a minor effect on the speed of
the hinge before it reaches A, == A/l., '" 0.4; thus the elastic-plastic response of the
present model is not very different from the rigid-perfectly plastic (Parkes') solution
during an initial period. According to the rigid-perfectly plastic solution, the
travelling hinge passes through the entire cantilever. The present results indicate,
however, that if the input energy is relatively small (compared with the elastic
178 Second-Order Effects On Dynamic Response

0·15

0·10

0·05

Fig. 5.43 Variation of hinge location with time for different values of R; r = I, eo =1.

capacity of the cantilever), the travelling hinge passes through only a part of the
cantilever before it disappears (pattern II). Thereafter, the entire cantilever simply
rotates about the root. If the remaining kinetic energy is small at the beginning of
the modal phase, the root deforms in the elastic range (pattern lIa); if this energy
is large, the root deforms plastically (pattern lIb). Figure 5.43 also shows that for
either pattern I or pattern II, the hinge slows down near the middle of the cantilever.
This is similar to the results obtained by finite-element approaches; it results in a
high-curvature region (or a 'kink') around the middle part of the cantilever in its
final configuration.
For eo = 1, Fig. 5.44 shows a map, which illustrates the regions of occurrence
for these patterns in the R-1-r plane. If the stored energy ratio R- 1 > 1, the
elastic capacity is more than the input energy so the response is always pattern
lIa. If R- 1 < 1 but the mass ratio r» 1, the response is always pattern lIb. In
this case, most of the input energy is dissipated plastically at the root in a modal

10

JIb Ha
r incomplete travel I incomplete
of hinge; later I travel of
another hinge I hinge; no
5 for ms at root I hinge at
I root
I
I
I
I
0
10- 3 10 -1 10 1!R
1000 10 0·1 R
Fig. 5.44 Occurrence of various response patterns in the R- 1 - r plane.
5.5 Effect of Elastic Defonnation 179

phase of deformation. Finally, if the stored energy ratio R- 1 is very small while
the mass ratio r is moderate, the response is Pattern I. This is similar to the
response of a rigid-perfectly plastic beam. For a small local energy ratio eo = 1,
the global energy ratio R can be large only if the beam is stubby; i.e. Llh is
small.
It is interesting to compare this conclusion with the previous results of finite
element analyses. For instance, the parameters adopted for Example 1 in Table
S.4 seem to fall in the region of modal response, i.e. pattern lIb.

Energy Dissipation At any instant during dynamic deformation, part of the

input energy has already been dissipated, part remains as kinetic energy and part
is stored as elastic strain energy. The kinetic energy of the system at any instant
is determined by

K = ~GV2 + ~ rL p(dW)2 dX (S.199)

2 2 Jodt
Substituting the velocity field (S .187) into (S .199) gives

~ = ~rv2+~[~2+~A(3-A)+v2A] (S.200)
Mp 2 6
The elastic deformation energy at the root, Ve ' is
1 M2
V -ClJI2 ~ V max = ---.l!.... (S.201)
e 2 e e 2C
where V:-ax denotes the maximum elastic energy that can be stored in the system.
Consequently, the ratio of elastic strain energy to the local bending stiffness is
2 2
Ve Rifle RlJIe (S.202)
Mp = 4e o = 2/'V~
Let Dp denote the total plastic dissipation, while andD; D;
denote the
dissipation at the root hinge and at the travelling hinge, respectively; then the
energy balance leads to
(S.203)
Using the expressions above, Eq. (S.203a) may be recast into a nondimensional
form,
Dp = -r(v
1 2 2 1[ . 2 . 2] RlJI;
- o -v )--lJI +lJf\'A(3-A)+V A - - - 2 (S.204)
Mp 2 6 2/'Vo
To elucidate the energy dissipation pattern for different ranges of input and
structural parameters, some typical examples are given in Fig. S.4S, in which
D DR DT
d p=
- --E.. ,
R = -..E.-.
dp - ,
T = -..E.-. =
dp -
Ko Ko Ko
Figure S.4Sa shows the source of energy dissipation as a function of time for the
rigid-perfectly plastic solution; while Fig. S.4Sb shows the dissipation in pattern
I response. These figures are obviously similar since in both cases the hinge
transits the entire cantilever. Figure S.4Sc, d shows the energy dissipation for
pattern lIb and Pattern IIa, respectively. There is no plastic dissipation in Pattern
lIb during the interval from the disappearance of the travelling hinge 't TD to the
formation of the root hinge 'tRp . If 'tTD tends to zero, the energy dissipation pattern
180 Second-Order Effects On Dynamic Response

1·0...------.,:-----=--.. 1·0.------------=-
rigid-plastic.
eo=1. 1'=1

0·5 0·5

I
I
I
I

O·O'-~-----'I-------'-------
O·O'l'TR 0·5 1·0
'l'RP 'I'
a b

,. O,----,-------:--::-::-c:--------, 1·0 ,-------:-------;-:--=------,

eo=1·6q. R-'=0'0676 eo= 1. 1=1. R-'= 1·0
1 =1·64 Pattern Ha
Pattern IIb

0·5 0·5-
dp d£
,- ,,"'"
-- ~----I

O.O'-~_ _ _. .J_I_ _...J_I____'I'________'

2 1'RD 0·0 0·05 'l'TO 0·1
l'
c d

Fig. 5.45 Accumulation of energy dissipation with time: - - total plastic dissipation (d p ), - - - - -
dissipation at root hinge (d:), and - - - dissipation at travelling hinge (d~). 1'RP is the instant when the root
spring enters into plastic state, 1'TR is the instant when the travelling hinge reaches the root, 1'RD is the instant
when the root hinge disappears, 1'TD is the instant when the travelling hinge disappears; (a) rigid-perfectly
plastic solution (R- 1 =0); (b) pattern I; (c) pattern lib; (d) pattern IIa.

approaches that of a modal solution. For Pattern IIa the root never dissipates any
energy.
It may be noted that the parameters used for calculating the energy dissipation
shown in Fig. S.4Sc are exactly the same as Example I in Table S.4. The finite
element results indicate that in this example the interior plastic deformation develops
first and this is nearly complete before appreciable rotation at the root hinge
occurs. This is quite similar to the sequence of events given here for pattern lIb.
Reid and Gui [1987] pointed out that the final modal response is delayed by an
interaction between a reflected elastic flexural wave and the travelling hinge. For
the current composite model, this feature is clearly present in pattern lIb. This
characteristic feature of elastic-plastic response to impact is captured by this
composite model that incorporates elastic deformation only at the root.
In most of the results shown here, we consider eo =1; i.e. Ko = Mp. Wang
and Yu [1991] reported that the effect of eo on the pattern of response is similar
to that of R-I. For specified values of rand R- 1, the response tends to be pattern
5.5 Effect of Elastic Deformation 181

I or pattern II if eo is very small or very large, respectively. These authors found

that eo has little influence on the total energy dissipated d p'

5.5.6 Remarks

1. In some respects the impact response of an elastic-plastic cantilever differs

from that calculated for a rigid-perfectly plastic idealization of this structure; the
dynamic response differs in distribution of bending moment, position and
development of plastic regions, the deformation process, and the final distribution
of plastic deformation. In particular, the elastic-plastic response of a cantilever
struck at the tip exhibits the following features:
(a) If R '" I or larger, a plastic region always propagates away from the impact
point in an early period of the response. This small region of plastic deformation
is similar to the travelling hinge predicted by the rigid-perfectly plastic theory.
(b) When the small plastically deforming region approaches the middle of the
cantilever it slows and can oscillate back and forth; this causes more plastic
dissipation and curvature in this central segment. Meanwhile, the root section
usually undergoes reverse flexure.
(c) After the plastic region oscillates near midlength, it can resume travelling
towards the root (if R is relatively large), or it can disappear (if R is relatively
small). If the travelling hinge disappears near midlength, thereafter the root
becomes a plastic hinge.
(d) The spatial distribution of energy dissipation in elastic-plastic response is
usually rather different from that calculated with the rigid-perfectly plastic
theory. While the perfectly plastic theory gives large final curvature near the
root when the mass ratio r is large, it also gives large final curvature near
the tip when mass ratio r is small. Only the elastic-plastic response gives
large curvature midway between the ends of the cantilever.
2. From various approaches we conclude that although the input energy ratio R is
important in judging the validity of the rigid-plastic theory, it is not a unique
parameter since it does not characterize all of the approximations of the rigid-
plastic theory. For a cantilever struck at the tip by a heavy rigid body, comparisons
of the finite difference, finite element and the simplified model with an elastic-
plastic root indicate that the energy ratio R, the mass ratio r = G/ pL and the
impact energy ratio eo all play significant roles in determining the validity of the
rigid-plastic theory. The combined effect of input energy ratio R and mass ratio
r on dynamic behavior of impulsively loaded elastic-plastic cantilevers may be
summarized in Table 5.5, which clearly states the essential influence of mass
ratio r on the deformation process.
3. The basic feature of dynamic response of elastic-plastic cantilevers is not
significantly altered by some minor differencf:s in modeling; e.g. the difference
between a Timoshenko beam of rectangular cross-section and an ideal sandwich
Euler-Bernoulli beam. Indeed the finite element, finite difference and rigid-plastic
beam with elastic-plastic root all provide similar patterns for the deformation
process.
182 Second-Order Effects On Dynamic Response

Table 5.5 Dynamic response of impulsively loaded cantilevers: effects of energy ratio Rand
mass ratio r
R Large r Small r
R>IOO Hinge passes through the entire length of the cantilever: similar to rigid-
plastic solution (Sect. 4.5), or Pattern I (Sect. 5.5.5)

Root dissipates most energy Root dissipates little energy

IO<R<IOO after travelling hinge terminates after travelling hinge terminates
at midlength; Pattern lib in Sect. at midlength; Pattern IIa in Sect.
5.5.5 or Ex. I in Table 5.4 5.5.5 or Ex. 2 in Table 5.4

2<R<5 Mode-like response, Hinge travel delayed at

root dissipates almost all energy midlength, Ex. 2 in Table 5.4

0.5< R< 2 Almost entirely elastic response Very localized plastic

with small plasticity at root deformation near tip

5.6 Accuracy of Rigid-Plastic Analyses

Studies in Sects 5.1-5.5 about second-order effects in the dynamic plastic response
of impulsively loaded cantilevers have enlightened us about the accuracy of the
easy to use rigid-perfectly plastic idealization. In the present section, particular
attention will be paid to the following questions: When is the rigid-plastic model
valid? What conditions are necessary for elastic effects to be insignificant? Which
factors control the 'error' of the rigid-plastic prediction in comparison with elastic-
plastic solution for the final deflection?
Before we answer these questions, notice that (1) dynamic behavior of elastic-
plastic structural members (e.g. cantilevers) is rather complex, as shown in Sect.
5.5; and (2) dynamic behavior of rigid-plastic structural members is complicated
by the transient phase of deformation, see Sect. 4.5. It is difficult, therefore, to
answer these questions by directly comparing elastic-plastic and rigid-plastic
solutions. It is possible nevertheless to obtain some preliminary answers from
comparisons with mode-type solutions. The concept of modal approximation has
been explained in Chap. 2.
In the following, a single degree-of-freedom (DoF) mass spring system, which
represents the mode-type motion of structures, is first studied in Sect. 5.6.1 in
order to show what factors, apart from energy ratio R, significantly affect accuracy
of the rigid-plastic prediction. Then the convergence of the response of a two
DoF model of an elastic-plastic cantilever to the primary plastic mode is examined
in Sect. 5.6.2; there the influence of initial momentum distribution is discussed.

5.6.1 Accuracy of Rigid-Plastic Analysis Estimated by Single

DoF System
Previous investigations in Sect. 5.5 have confirmed that rigid-plastic analyses
provide good approximations for dynamic plastic deformation of a structure
subjected to intense short pulse loading if the energy ratio R, defined by Eq.
(5.168), is large. In the limiting case of impulsive loading (i.e. an initial velocity
5.6 Accuracy of Rigid-Plastic Analyses 183

u
Fig.5.46 Single DoF system loaded by pulse
F(/); (a) single DoFmass-spring system; (b)
force-displacement relation for spring.

a b

field), the 'error' in the final deflection caused by the rigid-plastic idealization is
estimated to be of the order of (l/R), see Symonds [1985]. In this case the error
is positive since the initial kinetic energy must all be dissipated by plastic
deformation rather than beirig divided between plastic work and elastic strain
energy.
However, R» 1 is a necessary but not sufficient condition for validity of the
rigid-plastic idealization. When the loading pulse is relatively long, the error of a
rigid-plastic solution can increase in magnitude and may become negative; i.e.
the rigid-plastic solution can underestimate the final deformation of structures.
Symonds [1981] has pointed out that if the duration of a loading pulse td, is not
brief in comparison with the fundamental period for elastic vibration of the structure
1J., the errors caused by the rigid-plastic idealization may be large (e.g. 30-
60%). This occurs although the actual plastic deformations are as large as 10-20
times the elastic limit strain. In general, therefore, rigid-plastic predictions are
applicable only if both of the following conditions are satisfied:
R » 1, td < 1J. (5.205)
Noting that a modal approximation for dynamic rigid-plastic deformation of a
structure is a single DoF motion, Symonds and Frye [1988] examined the dynamic
behavior of a single DoF mass-spring model with a spring made from either
elastic-perfectly plastic or rigid-perfectly plastic materials. (A sketch of this single
DoF system is shown in Fig. 5.46.) They investigated the response of this simple
model to six different shaped pulse loads (rectangular, half-sine and triangular
with four rise times) in order to assess the influence of rise time and pulse duration
on the accuracy of the rigid-plastic predictions. Their calculations confirm the
earlier observation that a large energy ratio R is necessary, but not sufficient to
ensure that the rigid-plastic idealization gives. a small error in comparison with
the elastic-plastic solution. For example, Fig. 5.47 is obtained from this single
DoF model when it is subjected to a linear decreasing pulse; i.e. a triangular
pulse with zero rise time. Figure 5.47a shows the ratio uf Iuy, where uf is the
final displacement given by the elastic-plastic solution and Uy = Ny leis the
displacement at which the spring in the single DoF system enters the plastic
state. Here Ny is the yield force of the spring, Fmax denotes the peak force
applied on the mass and a load parameter is defined as Jl = Fmax I Ny. Figure
5.47b shows the error of the rigid-plastic prediction relative to the elastic-plastic
solution for the final displacement. Error is dl~fined as (ui - u,?)lu,? where ui
is the permanent displacement given by the rigid-plastic solution, as a function
of the pulse duration (t d l1J.) and R is the energy ratio defined by (5.168). For the
present model, R = 2u? I Uy. It is seen that while the error of the rigid-plastic
prediction reduces with increasing R, the error increases with pulse duration td'
184 Second-Order Effects On Dynamic Response

125
~ax~
100 a ~
linearly
75 decreasing
u, pulse
uY50

25

00
a

2 4 6 td/ T1 8
0
~
(s
~ -50

-100
b

Fig.5.47 Behavior of single DoF system subjected to linear decreasing pulse; (a) dependence of displacement
ratio on loading parameters; (b) error of rigid-plastic prediction relative to elastic-plastic solution on final
displacement.

A notable fact explored by Symonds and Frye [1988] is that when the loading
pulse has a nonzero rise time, the error of the rigid-plastic prediction displays a
wavy character over a large range of the ratio (tdl1;), and the discrepancies
between the rigid-plastic and the elastic-plastic analyses may become quite
significant. For example, Fig. 5.48 is obtained from the single DoF model when
it is subjected to a half-sine pulse with duration td and peak force Fmax = J1Ny.
The error of the rigid-plastic prediction is shown in Fig. S.48b, which indicates
that the error is smaller for larger values of R, and the peaks in the error curves
decrease as the pulse duration increases. Triangular pulses with nonzero rise time
result in similar error curves. In cases where R is not very large, no simple
formula based only on energy ratio R could provide an estimate of the error of
the rigid-perfectly plastic approximation.

5.6.2 Convergence to Dynamic Plastic Mode Studied by Two DoF

System
Simple lumped mass models of impulsively loaded beams can greatly simplify
the beam dynamics by limiting the number of DoF; likewise, they limit the number
of dynamic modes. Stronge and Hua [1990] adopted this kind of two DoF model
as a tool to study the transient response of elastic-plastic beams and cantilevers,
in comparsion with rigid-plastic solutions. Results from the two DoF model of a
cantilever are briefly extracted below.
5.6 Accuracy of Rigid-Plastic Analyses 185

50 F

40~
o sine td
30 pulse
U,
u y 20

10

Fig.5.48 Behavior of single DoF system subjected to half-sine pulse; (a) dependence of displacement ratio on
loading parameters; (b) error of rigid-plastic prediction relative to elastic-plastic solution on final displacement.

To model a structure subjected to arbitrarily distributed impulsive loading,

consider a cantilever of length L with two equally spaced masses G, as shown in
Fig. 5.49a. These masses are located at the midpoint and the tip; and they have
initial velocities Vo and 0'0' respectively. Assume that deformations occur only
at the discrete hinges located at the root and the midpoint. The cantilever consists
of two rigid segments connected by these hinges. The hinges are assumed to have
elastic-perfectly plastic behavior; that is, if the relative rotation at hinge i (i = 1,2)
is lfIi' then the moment Mj during loading is given by

i= 1,2 (5.206)
Vti > 0
where parameter lfIy is the yield rotation angle of each hinge. The deformation of
this model is always initially elastic while both hinges have rotations \lfId < lfIy;
later, it can be elastoplastic after one of the hinges has a rotation magnitude
\lfIi \ ~ lfI y; and finally it can become fully plastic if both hinges have rotation
magnitudes \lfId ~ lfIy· The analysis given by Stronge and Hua [1990] neglects
elastic unloading so that the final deformed configuration is unambiguous.
By introducing nondimensional variables
mi - M;fMp' 1fii == lfI;flfly
(5.207)
Wi - 2W;fLlfIy,
186 Second-Order Effects On Dynamic Response

elastic modes dynamic plastic modes

2'4~
primary 2~
- 0·4 L..;t>...::.......-=--~,--_.......,,_
~
2nd mode
-1 ['=-...."
~
b unstable c

Fig.5.49 Two DoF system; (a)two DoF model of impulsively loaded cantilever; (b) elastic modes; (c) dynamic
plastic modes.

the equation of motion of this two DoF model is

(5.208)

where wi is the nondimensional transverse nodal displacement, mi is the moment

at node (i -1), and (.) == d( )/df. For impulsive loading the initial conditions are
(5.209)

where Vo == Vo (C/G) 1I2 V'yl.

The cantilever has primary and secondary modes for fully plastic deformations
as shown in Fig. 5A9c. The primary dynamic plastic mode is a rigid body rotation
about the root. This mode is stable while the secondary plastic modes are unstable.
The last secondary mode shown in Fig. 5.49c has moments of the same sense at
both hinges. This curious mode only exists for WiW2 > 0 and IW2 - > 0 when 2wd
it appears in conjunction with the primary mode - it does not exist independently.
Stable modes are minima for the dissipation rate at any specified kinetic energy.
For impulsively loaded rigid-plastic structures, deformation finally converges to
a stable modal solution unless the initial momentum distribution is identical to an
unstable mode shape.
Stronge and Hua [1990] gave formulations for the elastic, elastoplastic and
fully plastic phases of motion of a structural model with two discrete masses; this
model has only two locations where elastoplastic hinges can form. For the fully
plastic phase, both of these hinges are plastic, and deformation occurs in a
combination of dynamic plastic modes before coalescence with a stable plastic
mode. After coalescence the final period of motion has a velocity distribution
that is identical with a stable plastic mode; the deformation finally terminates
when motion ceases.
The energy ratio for the model is R = (V;
/2)(1 + S-2) where S- is the ratio of
initial transverse speeds for the two masses. For v;
12 = 5 and O:s; S-:S; 3, the
velocity trajectories for elastoplastic deformations of the impulsively loaded
5.6 Accuracy of Rigid-Plastic Analyses 187

a b

Fig. 5.50 Velocity trajectories for two OoF model; (a) elastic-plastic case; (b) rigid-perfectly plastic case.

to an initial momentum distribution, ,=

cantilever are illustrated in Fig. 5.50a. The stable dynamic plastic mode corresponds
O. For initial conditions in the range
1.1 < , < 2.0, the deformation at the central hinge is still elastic when the velocity
distribution is identical to that of a stable plastic mode so the elastoplastic trajectory
crosses this mode line. For an initial momentum distribution near the stable plastic
mode configuration, yield occurs first at the root hinge.
Velocity trajectories corresponding to the rigid-perfectly plastic approximation
for the same model are. shown in Fig. 5.50b. These solutions converge to the
stable plastic mode along paths parallel to the unstable plastic modes; the
discontinuities in these trajectories at it! = 0 are caused by a change in the sense
of the moment at the root. For a specified energy ratio R, in comparison with the
elastoplastic response, the error of the rigid-plastic approximation is largest for
an initial momentum distribution that is not very different from the stable plastic
mode shape.
Stronge and Hua [1990] also studied a two DoF model of a simply supported
elastoplastic beam, which has two stable dynamic plastic modes. Their analysis
shows that elastic effects during an intermediate phase of elastoplastic deformation
can significantly affect subsequent widespread plastic deformation when multiple
stable dynamic plastic modes exist. As a result, the accuracy of the rigid-plastic
approximation is very sensitive to the initial momentum distribution, especially
when the latter is close to the stable plastic mode. In these cases, R will not serve
as the unique criterion in determining the error caused by the rigid-plastic
idealization.

5.6.3 Remarks
1. From various approaches we may conclude that although energy ratio R is of
importance in judging the validity of the rigid-plastic theory, it is not the only
factor determining the 'error' of the rigid-plastic theory. A study of a pulse-
188 Second-Order Effects On Dynamic Response

loaded one DoF model (Symonds and Frye [1988]) indicated that pulse duration
has substantial influence, especially if rise time of the pulse is significant and the
duration is long in comparison with the natural period. An analysis of a two DoF
model (Stronge and Hua [1990]) shows that this 'error' strongly depends on whether
the distribution of initial momentum is close to a mode shape.
2. The value of R is related to the elastic capacity U-: ax of the structure by
(5.168), while U-:ax is usually calculated by assuming that all parts of the structure
enter the yield state simultaneously (e.g. see (5.169». This leads to an overestimate
of U-:ax and a underestimate of R, especially when the structure involved is
large. In fact, for a very large structure subjected to local loading, the rigid-
plastic solution is a good approximation near the impact point and not accurate
further away even for a small value of R.
3. According to the convergence theorem (refer to Chap. 2 or Martin and Symonds
[1966]), the rigid-plastic approximation always converges to a stable plastic mode
configuration; therefore, a plastic modal solution gives a good approximation
provided the correct mode is selected and this is given the proper initial velocity.
However, when elastic deformation is involved, the accuracy of a plastic mode
approximation significantly depends on the initial momentum distribution, especially
if the latter differs greatly from a stable plastic mode-shape.

References

Atkins, A.G. [1990]. Note on scaling in rigid-plastic fracture mechanics. Int. 1. Mech. Sci. 32, 547-548.
Bodner, S.R. and Symonds, P.S. [1960]. Plastic deformation in impact and impulsive loading of beams.
Plasticity, Proceedings of the Second Symposium on Naval Structural Mechanics, (eds E.H. Lee and
P.S. Symonds) Pergamon Press, 488-500.
Bodner, S.R. and Symonds, P.S. [1962]. Experimental and theoretical investigation of the plastic
deformation of cantilever beams subjected to impulsive loading. ASME 1. Appl. Mech. 29, 719-728.
Conroy, M.F. [1952]. Plastic-rigid analysis of long beams under transverse impact loading. ASME 1.
Appl.Mech.19,465-470.
Florence, A.L. and Firth, R.D. [1965]. Rigid-plastic beams under uniformly distributed impulses. ASME
1. Appl. Mech. 32,481-488.
Forrestal, M.J. and Sagartz, MJ. [1978]. Elastic-plastic response of304 stainless steel beams to impulse
loads. ASME 1. Appl. Mech. 45, 685-687.
Hashmi, SJ., AI-Hassani, S.T.S. and Johnson, W. [1972]. Large deflexion elastic-plastic response of
certain structures to impulsive load: Numerical solutions and experimental results. Int. 1. Mech. Sci.
14, 843-860.
Hodge, P. G., Jr. [1959]. Plastic Analysis of Structures. McGraw-HilL
Hou, WJ., Yu, T.X and Su, XY. [1995]. Elastic effect in dynamic response of plastic cantilever beam
to impact, Acta Mechanica Solida Sinica, 16, 13-21.
Johnson, W. and Travis, F.W. [1968]. High-speed blanking of steeL Engineering Plasticity (eds J.
Heyman and EA. Leckie) Cambridge University Press, 385-400.
Jones, N. [1967]. Influence of strain-hardening and strain-rate sensitivity on the permanent deformation
of impulsively loaded rigid-plastic beams. Int. 1. Mech. Sci. 9,777-796.
Jones, N. [1971]. A theoretical study of the dynamic plastic behaviour of beams and plates with finite-
deflections. Int. 1. Solids Struct. 7, 1007-1029.
Jones, N. [1989]. On the dynamic inelastic failure of beams. Structural Failure, (eds T. Wierzbicki and
N. Jones) John Wiley, 133-159.
Jones, N. and de Oliveira, J.G. [1979]. The influence of rotatory inertia and transverse shear on the
dynamic plastic behaviour of beams. ASME 1. Appl. Mech. 46, 303-310.
Jones, N. and Song, B.Q. [1986]. Shear and bending response of a rigid-plastic beam to partly distributed
blast-type loading. 1. Struct. Mech. 14(3), 275-320.
References 189

Jouri, W.S. and Jones, N. [1988]. The impact behaviour of aluminium alloy and mild steel double-shear
specimens. Int. J. Mech. Sci. 30,153-172.
Liu, J. and Jones, N. [1987]. Experimental investigation of clamped beams struck transversely by a mass.
Int. J. Impact Engng. 6, 303-335.
Martin, J .B. [1989]. Dynamic bending collapse of strain-softening cantilever beams. Structural Failure
(eds T. Wierzbicki and N. Jones) John Wiley, 365-388.
Martin, J.B. and Symonds, P.S. [1966]. Mode approximation for impulsively loaded rigid-plastic
structures. Proc. ASCE, J. Engng. Mech. Div. 92(EM5), 43-66.
Nonaka, T. [1967]. Some interaction effects in a problem of plastic beam dynamics, Parts 1-3. ASME J.
Appl. Mech. 34, 623-643.
Nonaka, T. [1977]. Shear and bending response of a rigid-plastic beam to blast-type loading. Ingenieur-
Archiv, 46, 35-52.
de Oliveira, J.G. [1982]. Beams under lateral projectile impact. Proc. ASCE, 1. Eng. Mech. Div.
108(EM1),51-71.
de Oliveira, J.G. and Jones, N. [1979]. A numerical procedure for the dynamic plastic response of beams
with rotatory inertia and transverse shear effects. J. Struct. Mech. 7(2), 193-230.
Parkes, E.W. [1955]. The permanent deformation of a cantilever struck transversely at its tip. Proc. Roy.
Soc. Land. A228, 462-476.
Perrone, N. [1966]. A mathematically tractable model of strain-hardening, rate-sensitive plastic flow.
ASME J. Appl. Mech. 33, 210-211.
Reid, S.R. and Gui, X.G. [1987]. On the elastic-plastic deformation of cantilever beams subjected to tip
impact. Int. J. Impact Engng. 6, 109-127.
Reid, S.R., Yu, T.X. Yang, J.L. [1995]. Response of an elastic, plastic tubular cantilever beam subjected
to force pulse at its tip, - small deflection analysis, Int. J. Solids Struct. 32 (in press).
Shioya, T. and Stronge, WJ. [1984]. Impact on rotating fan blades. Inst. Phys. Can! Ser. No.70, 511-
518.
Shu, D. [1990]. Structural arrangement and geometric effects on plastic deformations in collisions. Ph.D.
thesis, Chap. 4, Cambridge University, U.K.
Shu, D., Stronge, W.J. and Yu, T.X. [1992]. Oblique impact at tip of cantilever. Int. J. Impact Engng. 12,
37-47.
Stronge, W J. and Hua Yunlong [1990]. Elastic effects on deformation of impulsively loaded elastoplastic
beams. Int. J. Impact Engng. 9, 253-262.
Stronge, WJ. and Shioya, T. [1984]. Impact and bending of a rigid-plastic fan blade. ASME J. Appl.
Mech. 51,501-504.
Stronge, WJ. and Yu, T.X. [1989]. Dynamic plastic deformation in strain-hardening and strain-
softening cantilevers. Int. J. Solids Struct. 25, 769-782.
Symonds, P.S. [1965]. Viscoplastic behavior in response of structures to dynamic loading. Behavior oj
Materials Under Dynamic Loading (ed. NJ. Huffington) ASME, 106-124.
Symonds, P.S. [1968]. Plastic shear deformations in dynamic load problems. Engineering Plasticity (eds
J. Heyman and F.A. Leckie) Cambridge University Press, 647-664 ..
Symonds, P.S. [1981]. Elastic-plastic deflections due to pulse loading. Dynamic Response ojStructures
(ed. G. Hart) ASCE, New York, 887-901.
Symonds, P.S. [1985]. A review of elementary approximation techniques for plastic deformation of
pulse-loaded structures. Metal Forming and Impact Mechanics (ed. S.R. Reid) Pergamon Press,
Oxford, 175-194.
Symonds, P.S. and Fleming, W.T. Jr [1984]. Parkes revisited: on rigid-plastic and elastic-plastic
dynamic structural analysis. Int. J. Impact Engng. 2, 1-·36.
Symonds, P.S. and Frye, C.W.G. [1988]. On the relation between rigid-plastic and elastic-plastic
predictions of response to pulse loading. Int. J. Impact Engng. 7, 139-149.
Symonds, P.S. and Mentel, TJ. [1958]. Impulsive loading of plastic beams with axial restraints. J. Mech.
Phys. Solids 6, 186-202.
Ting, T.C.T. [1964]. Plastic deformation of a cantilever beam with strain-rate sensitivity under impUlsive
loading. ASME 1. Appl. Mech. 31, 38-42.
Ting, T.C.T. [1965]. Large deformation of a rigid, ideally plastic cantilever beam. ASME J. Appl. Mech.
32, 295-302.
Wang, X.D. and Yu, T.X. [1991]. Parkes revisited: effect of elastic deformation at the root of a cantilever
beam. Int. J. Impact Engng. 11, 197-209.
Yu, TX. and Stronge, WJ. [1990]. Large deflections of a rigid-plastic beam-on-foundation from impact.
Int. J. Impact Engng. 9, 115-126.
Chapter 6

More Complex Configurations

This chapter considers the dynamic effects resulting from more complex loads
and structural shapes. In some cases the more complicated configurations require
additional degrees-of-freedom or other types of plastic hinges which undergo
either stretching or twisting in addition to bending. For plastic deformations these
combined states of generalized stress result in deformations and deformation rates
that are related by the associated flow rule. Hence, this chapter begins with general
considerations for the extremal properties and differentiability of the yield function
at generalized hinges. Here we consider only rigid-perfectly plastic materials and
hence plastic deformation that is localized at a finite number of discrete generalized
plastic hinges. While continuous plastic regions are possible, they require a special
distribution of applied force if the deformations are related to generalized stress
through an associated flow rule.
Depending on the generalized stresses that are significant for dissipation, we
may separate dynamic plastic problems into two groups: those involving only
flexural deformation and those involving an interaction yield condition. Transversely
loaded straight cantilevers with varying cross-section (Sect. 6.2) belong to the
first group, although more than one flexural plastic hinge may be required if
there is a crack or step in stiffness at a cross-section of a straight member (Sects
6.7 and 6.8). For a slender cantilever of curved or bent plan form, the effects of
the shear and axial force on yielding can be negligible for in-plane loading; if so,
these examples also have ordinary flexural plastic hinges. Thus they also fall into
the first group.
It is important to consider an interaction yield criterion if the total work done
on the structure by external forces is ultimately dissipated by more than one
component of deformation. This occurs for example, in the problem of oblique
impact on a structure (Sect. 6.3), where both normal and tangential components
of initial momentum are brought to rest by the plastic dissipation at a generalized
hinge (or hinges) that combines bending and stretching. The problems of curved
or bent cantilevers subjected to out-of-plane loading also fall into this second
group since here torque and bending moment are both significant in the plastically
deforming section. Modal solutions that involve interactions between bending
and twisting of curved members are dealt with in Sects 6.6 and 6.7.
192 More Complex Configurations

6.1 General Considerations

6.1.1 Extremal Properties of Yield Function at Plastic Hinge

The generalized stresses Qa = Qa(S) (a = 1,2,··,1]') that are functions of a spatial
coordinate S are related by the yield function P(Qa)' The yield condition P(Qa)
is satisfied on at least one side of each plastic hinge S=A j (j=1,2,"',1]H);
however, it is not necessary that the yield condition is continuous at hinges. With
regard to the differentiability of the yield function at a generalized plastic hinge,
we can distinguish the following three cases.
1. If the yield function P(Qa) is a continuously differentiable function of S, in a
neighborhood of an interior generalized plastic hinge, the function has a
maximum at the hinge location S = Aj (0 < Aj < L), (Fig. 6.1a); consequently,
the following conditions are satisfied:

(6.1a)

(6.1 b)

()2 P (Qf(S))1 .~ 0 (6.1c)

()S S=A
J

where (6.1c) requires Qa(S)EC Il (11?2); i.e. the generalized forces Qa must
have continuous second derivatives with respect to the spatial variable S.
2. If the yield function P(Qa) is continuous but not differentiable at an interior
generalized plastic hinge located at S = Aj (0 < Aj < L), condition (6.1a) is
maintained while the extremum conditions (6.1b) and (6.1c) are no longer
valid (Fig. 6.1 b); in this case one has
(6.2a)

()P(Qa(S)) I ? 0, ()P(Qa(S))1 . ~ 0 (6.2b)

()S S=A-O
J
()S S=A J +0
Here the derivatives ahead of and behind the hinge may not be equal to each
other, since Qa(S) e: d.
3. For a generalized plastic hinge at a boundary (S = L) where displacements are
constrained, the yield function P is a nonanalytic extremum; hence conditions
(6.1b) and (6.1c) are not necessary. As shown in Fig. 6.1c, at a displacement
constraint,

P(Qa(S))IS=L = 0 (6.3a)

()(Qa(S))1 ? 0 (6.3b)
()S S=L-O
6.1 General Considerations 193

L S

a b c
Fig. 6.1. Variation of yield function 'P (S) along cantilever: (a) at an interior generalized plastic hinge 'P (S)
is a local extremum so 'P(.II.) = 0; (b) 'P(.II.) = 0 but not differentiable; (c) hinge located at point of displacement
constraint has 'P (L) = 0 but not an analytic extreme.

It must be noted, however, that each set of expressions, (6.1), (6.2) and (6.3),
provides a different set of local properties for the yield function in a neighborhood
of a plastic hinge located at S = Aj (0 < Aj < L, j = 1,"',1]H); they do not guarantee
that the yield criterion is not violated at other sections of the cantilever. Thus
each of the above sets of equations form a set of necessary but not sufficient
conditions for a deformation mechanism at an interior plastic hinge S = A j
( 0 < A j < L) of a rigid-perfectly plastic cantilever. If this mechanism is a complete
solution it satisfies the laws of motion, the yield condition and the flow law
throughout the entire body. In cases 2 and 3, the spatial derivative of 'l'does not
necessarily vanish at a plastic hinge, so the requirements for a complete solution
stated in Chap. 2 must be applied directly to examine whether or not a mechanism
is a complete solution. Conditions (6.2) or (6.3) can be employed as a check.

6.1.2 Differentiability of Arbitrary Function at Plastic Hinge

In order to construct a complete solution for a specified dynamic problem, attention
must be paid to continuity conditions that are associated with the differentiability
of the yield function at generalized plastic hinges. First let us consider arbitrary
dependent variables that are functions of spatial coordinate S and time t, and
examine their differentiability at plastic hinges.
If the Jith order partial derivative of a function F(S,t) with respect to S,
aJ.l F(S,t)J asJ.l , is continuous but not differentiable with respect to S in the range
o S S s L for t ~ 0, then the value of Ji is defined as the index of differentiability
of function F(S,t) in 0 S S s L. This is indicated by the symbol (F); that is,
Ji == (F(S,t») (6.4)
The index of the differentiability of function F(S,t) at any point in a lineal
structure 0 S S s L has the following properties:
1. If F(S,t) E CJ.l while F(S,t) e: CJ.l+!, then F'(S,t) E CJ.l-l and F'(S,t) e: CJ.l ,
where F'(S,t) == aF(S,t)J as. Hence, using the symbol ( ) for index of differen-
tiability
(F'(S,t») = (F(S,t»)-1 (6.5)

2. The second property concerns differentiability of a function F(S,t) in a

neighborhood of a generalized plastic hinge at S = A. If the jump of a function
194 More Complex Configurations

vanishes at any time t, [F(A,t)] == F(A+,t) - F(A-,t) = 0, then (refer to Sect.

4.3)
[F(A,t}]+A[F'(A,t}] = 0 (6.6)

where (') == d( )/ at, ( )' == d( )/ dS and the speed of the travelling plastic hinge
is A. Thus, the following deductions can be made:
(a) For a stationary hinge A = 0; thus Eq. (6.6) gives [F(A,t}] = O. Consequently

(F(A,t)) = (F(A,t)), if A=0 (6.7a)

(b) For a travelling hinge A"# 0, so [F(A,t)] and [F'(A,t)] both are either
zero or nonzero. It follows that
(F(A,t}) = (F'(A,t}) = (F(A,t})-l, if A"# 0 (6.7b)
From Eqs (6.5) and (6.7b), it is seen that the index of differentiability of an
arbitrary function could take a negative value. A negative value of J.l, however,
does not relate to the conventional meaning of the differentiability of a function;
it is simply a result of the mathematical relationship (6.5) or (6.7b). Examples
are given in Sect. 6.1.3.
3. Concerning the differentiability of the sum and the product of two functions
Fi (S,t) and F2 (S,t), it can be proved that in general,
(Fi(S,t) + F2(S,t)) min{(Fi (S,t)),(F2(S,t))} (6.8a)

(Fi (S,t)· F2(S,t)) min{(Fi (S,t)),(F2 (S,t))} (6.8b)

6.1.3 Differentiability of Kinematic Variables at Plastic Hinge

In this section the differentiability of the kinematic variables that participate in a
kinematically admissible velocity field is examined. This field represents a
deformation mechanism with discrete generalized plastic hinges.
flex ural hinge torsional hinge

(.~ t. Q 0
at
~ 3.+
8t
Bf- " 13/
a b

stretching 'hinge'

L
C:=:::::JII,"~CI====::::J
T.Wx _ .. ~Wx
W
x':

c d
Fig. 6.2. Discontinuity at generalized plastic hinge: (a) flexural hinge - discontinuity of angular velocity
related to transverse motion; (b) torsional hinge - discontinuity of angular velocity about the centroidal axis;
(c) extensional hinge - discontinuity of axial velocity; (d) shear hinge - discontinuity of transverse velocity.
6.1 General Considerations 195

For ideal flexural hinges, as already discussed in Sect. 4.3, the discontinuity in
angular velocity is a common and essential feature at a plastic hinge, no matter
whether they are stationary or travelling; e.g. see Fig. 6.2a, where subscripts plus
and minus pertain to the values at the right and left side of the hinge, respectively.
Furthermore, a discontinuity in the axial component of angular velocity develops
if a generalized plastic hinge involves torsional plastic deformation; see Fig. 6.2b.
However, if a generalized plastic hinge involves stretching or shear deformation,
then there is a difference in translational velocity between the two sides of the
hinge. Consequently, a discontinuity in axial or transverse (shearing) velocity
develops at a stationary hinge if the length of the plastically deforming segment
is negligible.
Let "i and Wx denote the transverse and axial components of displacement,
respectively. The following deductions can be made if the displacement field is
continuous at an ideal plastic hinge:
1. For flexural plastic deformation, the transv~rse velocitr ~ust be continuous if
transverse displacement is continuous so [W
z ] = 0 but l W;] i: 0, thus
(~) = 0, (W;) = -1, (6.9a)

("i) {". for stationary hinges (s.h.)

(6.9b)

r-
1, for travelling hinges (t.h.)
(s.h.)
(6.9c)
(W,) " -1, (t.h.)
2. For torsional plastic deformation, the differentiability of rotational deformation
quantities is at least of the same order as those of flexure.
3. For axial plastic deformation, suppose [Wx] "I:- 0 so that
(Wx ) -1 (6. lOa)
-1, (s.h.)

1 -2,
-1,
(t.h.)
(s.h.)
(6. lOb)

"1 0, (t.h.)
(6.lOc)

From a physical point of view the axial displacement Wx has to be continuous

at plastic hinges, so that (6.10c) excludes any deformation mechanism with a
stationary plastic hinge that involves axial plastic deformation; however, a
deformation mechanism with both flexural and axial plastic deformation at a
travelling hinge remains acceptable.
4. For continuous displacement the plastic shear deformation [Wz ] = 0 but [~] i: 0,
so that
(~) = -1 (6. 11 a)

1
-1, (s.h.)
(W,) " -2, (t.h.)
(6.11b)

Similar to the case of axial plastic deformation, plastic shear at a stationary

hinge also results in discontinuous transverse displacement.
196 More Complex Configurations

6.1.4 Differentiability of Generalized Stresses at Plastic Hinge

The differentiability of generalized stresses is related to the differentiability of
kinematic variables that participate in the dynamic equations. In the case of
impulsive loading, typical dynamic equations in the transverse and axial directions
are
(6.12a)

M' -Q (6.12b)

N' (6.12c)
where p = p(S) is the density per unit length of the cantilever, which may vary
with S. Note that a strong or weak discontinuity of p is caused by a discontinuity
of cross-section (e.g. a stepped beam) or by a discontinuity in material density
per unit length (e.g. a composite beam consisting of two beams of different materials
that are butted together).
According to the property (6.8b) the differentiabilities of Q and N depend on
those of Wz and Wx , as well as the differentiability of p. For instance, from
(6.12a),
(6.13)

where (Wz ) is found from (6.9c).

In the case of a homogeneous beam of uniform cross-section, (6.12) leads to
(Q) = (Wz )+ 1 (6.14a)

(M) = (Q)+1 = (Wz )+2 (6.14b)

(N) = (Wx)+ 1 (6.14c)

It should be noted that the rotary inertia of beam differential elements is
disregarded in dynamic equation (6.12b). If this effect is considered, that equation
becomes
M' = -Q+lox8m (6.12b')
where lox is the second moment of mass per unit length about a transverse axis
through the centroid of the cross-section (in general, lox may vary with S if the
cantilever is not uniform); em is the usual rotation angle of the beam center line
due to bending alone. The total inclination of the center line W;and the shear
angle (i.e. transverse shear strain) ji are related by
(6.15)
Hence, when the rotary inertia of beam differential elements is incorporated in
the dynamic equations, the differentia.~ilities of M and Q will be mainly determined
by the angular acceleration term loxem

(M) =(8m)+1 = (W;)+l = (Wz) (6.16a)

(Q) = (M)-1 = (Wz)-l (6.16b)

Comparison of Eq. (6.16) with (6.14) shows that for the same degree of smoothness
in the displacement field, bending moment and shear force are two orders less
differentiable when rotary inertia is considered.
6.1 General Considerations 197

central
aXIs inertia
)orelated to
~\' torsion
inertia
~
\
related to
bending

~
a b c
Fig. 6.3. (a) Curved cantilever subjected to transverse force at tip. The force is normal to the plane of curvature;
(b) rotary inertia of curved cantilever about tangent line through hinge H if all mass is concentrated at center
line; (c) rotary inertia of beam element with respect to longitudinal axis.

In general, if a crooked or curved thin beam is subjected to bending and/or

torsion, as sketched in Fig. 6.3, and a plastic hinge forms at section H, then the
rotary inertia of the beam segment between the tip and the hinge, AH, consists of
two parts. The first part comes from the second moment of mass of segment AH
about the centroid of the hinge section H, as illustrated in (Fig. 6.3b); this is due
to the out-of-straightness of the member. The second part comes from the polar
second moment of mass in a cross-section about the centroid (Fig. 6.3c). The
latter is called the rotary inertia of the beam element about the longitudinal axis
of the beam. If a crooked or curved beam is slender, the second part of the rotary
inertia will be much smaller than the first part. In most problems this second part
is disregarded in the same way that transverse rotary inertia is usually disregarded.
Bending-torsion problems of curved and bent cantilevers have dynamic equations
that are obtained in Sects 6.6 and 6.7. In these sections it is confirmed that
negligible polar second moment of inertia for the cross-section about the central
axis results in better differentiability for torque rather than for bending moment,
(T) = (M) + 1 (6.17)
If the rotary inertia of a beam element about the central axis of the beam is taken
into account, however, the differentiability of torque T will be similar to that of
bending moment M as given in (6.16a). This is because the torque is related to
the polar second moment of mass about the central axis of the element in a way
similar to that of transverse rotary inertia (see Eq. (6.12b')).

6.1.5 Differentiability of Yield Function at Plastic Hinge

If the yield function is an analytic function of the generalized stresses Qu (S, t)
(a = 1,2, .. ·,1]'), i.e. P(Qu) E C il where J1 is an arbitrary positive number, then at
generalized plastic hinges the yield function has the same differentiability as that
of the generalized stresses appearing in '1':
(P(Qu)) = min ,(Qu(S,t)) (6.18)
u~l."·.1J

Some yield surfaces ('I' = 0) have singular points where the yield function
p(Qu) is continuous but not continuously differentiable; that is, P(Qu) E C Jl
with J1 < 1. If the stress state at a hinge is at one of these singular points, then at
the hinge the differentiability of the yield function is
198 More Complex Configurations

Table 6.1 Index of differentiability of various quantities for continuous displacements at a

generalized plastic hinge
Stationary hinge Travelling hinge
A=O A;eO
Bending only
Deflection Wz 0
Velocity ~z 0 0
Acceleration Wz 0 -I
Slope or inclination W;=9 m -I 0
Angular velocity ~;=9m -I -I
Angular acceleration 9m -1 -2
Curvature W;'= J( -2 -I
Shear force Q 0
Bending moment Ma 2
Bending moment Mb 0 -I
Bending + twist
Rotational angle 9, -I 0
Rotational velocity 0., -I -I
Rotational acceleration 9, -I -2
Torque Ta 3 2
Yield function 'P(M, T)" 2
Torque Tb 0 -1
Yield function 'P(M,ryb 0 -I

Bending + stretching
Axial displacement ~x 0
Axial velocity ~x -I
Axial acceleration Wx -2
Axial force N -I
Yield function 'P(M,N) -1

Bending + shear
Shear displacement Wz -I 0
Shear velocity ~z -I -I
Shear acceleration Wz -I -2
Shear force Q 0 -I
Yield function 'P(M, Q)" 0 -I
Shear force Qb -I -2
Yield function 'P(M,Q)b -1 -2
a Effect of rotary inertia of beam elements is disregarded.
b Effect of rotary inertia of beam elements is considered.

(P(Qa)) = 0 (6.19)
and the extremal conditions (6.lb) and (6.1c) for an interior plastic hinge are not
valid. For example, if a maximum normal stress criterion is employed for a beam
of rectangular cross-section that is simultaneously subjected to bending and
stretching, then the yield function is

P(M,N) = IMI
Mp
+ [~)2
Np
_
1 (6.20)

Consequently, at points where the yield function is analytic (i.e. smooth),

6.2 Straight Cantilevers With Smoothly Varying Cross-Sections 199

(P(M,N») = min(IM(S,t)I,N(S,t)) (6.2la)

while at singular points on the yield surface (e.g. M = 0, N = ±Np of (6.20») we
obtain
(P(M,N)) = 0 (6.2lb)
An example of a beam with an interaction yield function P(M,Q) subjected to
simultaneous bending and shear was examined by Symonds [1968]. His conclusions
regarding differentiability of the yield function were similar to (6.21).
In general, at any nonsingular point of the yield surface, the differentiability
of the yield function P(Qa) is determined by Eq. (6.18); that is, it depends on
differentiability of the participating generalized stresses Qa (a = 1, 2,,,,, rn.
For
instance, in the case of bending-stretching interaction, the yield condition has the
index given in (6.2la). Equation (6.lOc) excludes any deformation mechanism
involving axial deformation at a stationary plastic hinge; hence combined flexural
and axial plastic deformations can occur only at a travelling hinge. At a travelling
hinge, the combination of (6.l4b) and (6.9c) gives (M) = 1, while the combination
of (6.l4c) and (6. lOb) leads to (N) = -1. Thus, (P(M,N») = min{(M),(N)} =-1.
Table 6.1 is obtained by this process of deduction; the table summarizes the
differentiability conditions which components of deformation must satisfy in order
to give continuous displacements. The table lists the minimum requirements for
indices of differentiability based on the assumptions: (1) the density per unit
length p = peS) is a smooth function of S; and (2) only nonsingular points on the
yield surface are considered. These requirements are necessary to obtain continuous
displacements if the flow law is associated with the yield condition.
From Table 6.1 it can be seen that the yield function is only an analytic extreme
at a plastic hinge for the cases of pure bending and bending-torsion and then
only if there is negligible rotary inertia of the beam element about a transverse
axis. There is no necessity, however, for the yield function to be an analytic
extreme at a plastic hinge if the yield function P is not continuous at this location.
Discontinuities in stress resultants (and thus the yield function) at a travelling
hinge in a rigid-perfectly plastic beam are considered in Sect. 6.3. Prager [1954]
has termed such discontinuities as 'artificial' rather than natural since acceleration
of mass passing through the travelling hinge is caused by stress jumps that are
not along the yield surface - rather they are from an interior point to the yield
surface. The analysis in Sect. 6.3 assumes that at a travelling hinge, axial strain
develops from these jumps in axial stress and particle velocity.

6.2 Straight Cantilevers With Smoothly Varying

Cross-Sections

6.2.1 Yield Function and Conditions at Plastic Hinge

An analysis of an initially straight rigid-perfectly plastic beam with smoothly
varying cross-sectional properties of mass per unit length p(X), second moment
of area leX) and fully plastic bending moment Mp(X), is a first step towards
200 More Complex Configurations

generalizing the previous results for application to more complex structural

configurations. In this section we consider members with cross-sectional properties
that vary smoothly; i.e. the properties are parameters that can be differentiated
with respect to the spatial coordinate X at least twice. Later in this chapter (Sect.
6.7), examples of members with discontinuous cross-sectional properties are
considered. In this section our aim is to identify consequences of the rigid-perfectly
plastic constitutive approximation in more complex configurations.
Only flexural deformations are considered so the yield function is

P(X t) = iM(X,t)i -1 (6.22)

, Mp(X)
where the fully plastic moment M p (X) is a prescribed positive function of the
spatial coordinate X, and the index of differentiability (defined in Sect. 6.1) is
(M p (X») ~ 2. According to the discussion in Sect. 6.1, if the cross-section properties
have continuous derivatives at every point, then continuity of displacements requires
the dynamic moment distribution M(X,t) to have an index of differentiability
that is no less than (M(X, t)) = 2 or 1 at stationary or travelling hinges, respectively.
At a plastic hinge, therefore, the yield function must satisfy certain limiting
conditions given by Eqs (6.1) or (6.3).
1. At a plastic hinge located at X =A in the interior of the beam (0 < A < L) the
bending moment satisfies
M(A) = Mp (A) sgn(M) (6.23a)

dM(A) dMp(A) sgn(M) (6.23b)

£IX dX
2. At a plastic hinge located at the root A = L where the cross-section properties
are continuous, the bending moment M(L) = lim M(L - E) satisfies
E~O

M(L) = Mp (L) sgn(M) (6.24a)

dM(L) dM (L)
p sgn(M) (6.24b)
£IX dX
In particular, note that at a travelling hinge in the interior of the beam, the shear
*
force Q(A) = dM(A)/ £IX 0 if the fully plastic moment M/X) is not uniform
along the beam. This is in contradiction to an assumption made in some previous
analyses (e.g. Al-Hassani et al. [1973]). The conditions above were pointed out
first by Zhou and Yu [1987] who provided analyses of dynamic response of
tapered cantilevers subjected to either a suddenly applied force or impact by a
rigid body.

6.2.2 Suddenly Applied Steady Force at Tip of Tapered

Cantilever: An Example
Deformation Mechanism and Equations of Motion Consider a cantilever of
variable cross-section subjected to step loading of magnitude F suddenly applied
at the tip, (see Fig. 6.4a). Recalling the deformation mechanism for a cantilever
of uniform cross-section found in Sect. 4.1, we begin by assuming that a stationary
6.2 Straight Cantilevers With Smoothly Varying Cross-Sections 201

F
t .
.
B~ ..
p(X). Mp(X) 0 A L
'

V~
A X X

I.. ..I
a b

Fig.6.4. (a) Cantilever of variable cross-section subjected to suddenly applied transverse step force F at the tip;
(b) velocity field of a single hinge mechanism.

plastic hinge forms at an interior cross-section X = L. The kinematically admissible

velocity field related to this mechanism is shown in Fig. 6.4b, i.e.
. {V(l-X/A), O:S:X:S:A
W= (6.25)
~ A:S:X:S:L
where V is the velocity at the tip of the cantilever. Accordingly, the acceleration
field is
.. {V(l-X/A), O:S:X:S:A
W= (6.26)
0, A:S:X:S:L
The distributions of bending moment and shear force can be found from the
equations of motion for a segment of length X that begins at the tip,

FX - foXp @V(1-';,A)(X -.;) d';, O:S:X:S:A

(6.27)
FX - foAp@V(1-';' A)(X -.;) d';, A:S:X:S:L

(6.28)
A:S:X:S:L

It is seen that both bending moment M(X) and shear force Q(X) are independent
of time t, since the plastic hinge is stationary.

Fig.6.S. Tapered cantilever

with (a) positive {3 ; (b)
a b negative {3.
202 More Complex Configurations

Solution for a Tapered Cantilever Now let us suppose that the cantilever has
uniform density per unit volume, constant thickness h and a width that varies
linearly from bo at the tip to bo(l
+ /3) at the root,

b(X) = bo(l + 13 ~) (6.29a)

A positive width b(X) > 0 at sections X in the range 0:5: X:5: L requires 13 >-1.
Tapered cantilevers with either positive or negative 13 are sketched in Fig. 6.5a,
b, respectively. For this uniformly varying section, the density per unit length
and the fully plastic bending moment Mp(X) can be found in relation to the
respective values at the tip as

p(X) = Po(l + fj~) (6.29b)

Mp(X) = Mo(l+fj~) (6.29c)

From Table 6.1 and Eq. (6.29c) it is seen that the present example has a yield
function (6.22) with an index of differentiability ('P) ~ 2, so the bending moment
at a hinge is governed by Eq. (6.23). Substituting (6.29b) into (6.27) and (6.28)
and using (6.23) results in the following expressions for moment of momentum
about the tip and transverse momentum,

(6.30a)

(6.30b)

By eliminating Po V from these equations and defining

A
A ==- (6.31)
L
one finds that the hinge location A satisfies a quadratic equation
fj(f - /3)A2 + 2(f - 2/3)A - 6 = 0 (6.32)
Consequently, the nondimensional hinge position is

A == A = 1 (_ 1 + 213 + ~ 12 + 2113 _ 213 2) (6.33)

L fj(f - /3)
If a plastic hinge is located at an interior cross-section of a tapered cantilever,
o < A < 1, the diagrams of shear force and bending moment along the cantilever
are similar to those sketched in Fig. 6.6, where (a) and (b) illustrate cases of
13 > 0 and 13 < 0, respectively. The case of 13 = 0 represents the usual cantilever
of uniform cross-section. Figure 6.6 shows for cantilevers of nonuniform cross-
section (13 "# 0), shear force does not vanish at the plastic hinge; correspondingly,
the bending moment curve M(X) is not a maximum at the hinge point X = A, but
it is tangential to the fully plastic moment curve Mp(X). Thus 'zero shear force'
is not a general condition for determining the location of a plastic hinge; this
condition applies only to uniform beams.
6.2 Straight Cantilevers With Smoothly Varying Cross-Sections 203

I
I
I
I
I
I
I
I
I
I
I
I
: p<O
I

o x A
a b
Fig. 6.6. Distributions of shear force and bending moment along cantilever for (a) positive {3; (b) negative {3.

In a tapered cantilever, a fully plastic segment of the beam X> A supports a

limited shear force without accelerating. In this respect, this kinetic solution is
similar to the solution for a structure subjected to a suddenly applied force equal
in magnitude to the static collapse force F;,. In both cases the state of stress
yields fully plastic sections and a mechanism of collapse but the structure does
not accelerate unless F > Fc .

Location of Plastic Hinge as Function of Force Magnitude Expression (6.30a)

indicates that positive tip acceleration V> 0 appears if and only if the force is
larger than a collapse force,
FL
- == f "? fi (/3) == 1 + f3 (6.34)
Mo
If f < fi (f3), the tapered cantilever remains static; if f slightly exceeds fi (/3), a
plastic hinge appears at the root of the cantilever; i.e. at A = 1.
If the applied force is only slightly larger than the plastic collapse force the
hinge remains at the root. The plastic hinge moves inboard of the root if the
value of A given by Eq. (6.33) is less than unity. Thus, the minimum force for an
interior hinge is

2
no hinge

2 3 4 5 P
Fig. 6.7. Occurrence of three possible deformation mechanisms depends on the parameters {3 and f
204 More Complex Configurations

10

~
8 \~
~

\\ ~

I'",
"
6
f = FL ,,-P =1
1'10
4 P=-0.5 "" ' "
,
"
'2
2 f,

0
0.2 0.4 0.6 0.8
)" = II IL

Fig. 6.S. Dependence of plastic hinge location on step force j, for various values of the taper parameter /3.

FL =1 > 12 ({3) =3 + {3(l + {3) (6.35)

Mo 2 + {3
Figure 6.7 gives a map of the ({3,/) plane divided into three regions by curves
iJ({3) and 12({3): (1) if I<N{3)=l+{3 no hinge forms; (2) if iJ({3)~/~fz({3)
a hinge forms at the root; and (3) if I> fz (/3) a hinge forms at an interior cross-
section.
The dependence of the plastic hinge location on the magnitude of the step
force is shown in Fig. 6.8. It is seen that for tapered cantilevers, the taper ratio {3
does not significantly influence the plastic hinge location A(T).

6.3 Oblique Impact on Straight Cantilever

6.3.1 Problem and Assumptions

In many practical cases, impact loading on a structure is not entirely transverse,
i.e. it involves both bending and axial stretching. For a perfectly plastic structure
with a yield condition that couples bending moment and axial tension, a model
with discrete travelling hinges which combine bending and stretching deformations
cannot satisfy both the flow rule and yield condition throughout the deforming
region. A similar problem arises in transversely loaded beams with clamped end
supports due to the axial constraints at the ends of these beams. Axial stretching
from deflection between the fixed ends, however, results in negligibly small axial
accelerations. Hence Symonds and Mentel [1958] assumed that in these beams,
the simultaneous stretching and bending that occur at plastic hinges are only
kinematically related. For oblique impact on a cantilever where axial motion is
not constrained, Shu et al. [1992] reported on experiments which showed that the
axial disturbance decays quickly in an early part of the transient response period.
Subsequently the response is almost entirely flexural and similar to that caused
by the transverse component of the initial momentum. To represent this behaviour,
6.3 Oblique Impact on Straight Cantilever 205

a
l

GQ)::::::X ==:Jl ~ N(X)

~I----"--"'''~ ra (Xl
b c
Fig.6.9. (a) A single 'hinge' mechanism for oblique impact on a cantilever; (b) positive senses of axial force
N, shear force Qand bendingmomentM; (c)free-body diagram of segment HB (X>A) when MH >0, QH > O.

we assume that deformation occurs at a discrete travelling plastic hinge but require
that yield is satisfied on one side only of the (infinitesimal) deforming region.
Let a straight cantilever of uniform rectangular cross-section with length L
and density p per unit length be subjected to oblique impact on a heavy particle
attached to the tip, as shown in Fig. 6.9a. The particle has mass G. After impact
the initial velocity of mass G has a transverse component Vzo and an axial
component Vxo ' Apart from the heavy particle at the tip, the cantilever is initially
at rest. In the present analysis, the following assumptions are made.
1. The structure is composed of a rigid-perfectly plastic, rate-independent material.
2. The effect of shear force on yield is neglected, so the yield criterion involves
only two generalized stresses, i.e. bending moment M and axial force N. The
member has a rectangular cross-section with width b and thickness h, so the

I:,H:. J
yield function at any cross-section X is

'Y = 'Y(M,N) = -1 (6.36)

where Mp = Ybh 2 /4 and Np = Ybh are the fully plastic bending moment and
fully plastic axial force, respectively.
3. Deflections are small so that all geometrical relations are measured relative to
the initial configuration. Furthermore, the additional bending moment (beam-
column effect) caused by the axial force has been disregarded.

6.3.2 Formulation Based on Single-Hinge Mechanism

Deformation Mechanism and Kinematically Admissible Field Let X be an axial
coordinate measured from the tip and Z be a transverse coordinate that is orthogonal
to X and in-plane with the initial velocity (Vxo' Vzo ), as shown in Fig. 6.9a.
Suppose the deformation mechanism has a single hinge. At time t after impact,
the hinge is located a distance X = A from the impact point at the tip; at this
instant the rigid segment AH is rotating about H while segment HB remains at
rest. In addition to rigid body rotation about hinge H, segment AH also has axial
motion in response to the axial momentum imparted initially by the axial velocity
206 More Complex Configurations

N(X)
Wx(X.t)

v·(t)1 I 0 11 X

~
0 11 X
N+
a H
c

Nt"t'

M(X)

.. HH

,,(t,[//Z, ----
0 11 X
QH= 0
~
;j )-ty.,.?;--"
~ X

.1"
0 11
b MH/QH d
I..
Fig. 6.10. Distributions of (a) axial velocity; (b) transverse velocity; (c) axial force; (d) bending moment.

Vxo of G. This axial motion requires axial plastic deformation and this occurs
only at the generalized hinge H; hence, the rotating segment AH has uniform
axial speed Wx ' Therefore, the axial and transverse velocity fields related to this
single-hinge mechanism are
O:S; X:S; A
(6.37)

. { -8m(A-X)=Vz(t)(I-X), O:S;X:S;A
Wz(X,t) = A (6.38)
0,
Here, Wx(X, t) and Wz(X, t) denote the axial and transverse displacements,
respectively; 8H (t) = Vz (t)/ A is the angular velocity of segment AH rotating about
hinge H; Vx(t) and Vz(t) are the axial and transverse velocity components,
respectively, of the concentrated mass at the tip, as illustrated in Fig. 6. lOa and b.
Equation (6.10c) shows that axial plastic deformation can occur only at a
travelling hinge; in other words, if continuity of displacements requires that axial
velocity is not uniform, a~ial stretching occurs only at concentrated travelling
'hinges' moving at speed A(t)"# O. By differentiating (6.37) and (6.38), the axial
and transverse accelerations are

{
Vx(t), O:S;X:S;A
(6.39)
0, A:S;X:S;L

. (I-A
Vz(t) X ) +'-'z(t)A(t)
. AX ' O:S;X:S;A

{
2
(6.40)
0, A:S;X:S;L
6.3 Oblique Impact on Straight Cantilever 207

respectively. Note that both axial and transverse accelerations have discontinuities
at hinge X = A; the transverse acceleration is similar to that shown in Fig. 4.23c.

Equations of Motion We assume that dissipation due to shear is negligible so

that the generalized stresses which are active in the yield condition are only
bending moment M(X) and axial force N(X). The positive senses of these stress
resultants is defined in Fig. 6.9b. Then by making use of the transverse acceleration
given by (6.40), the equations of motion of segment AH in the transverse direction
can be written as

-d(1 dVz
.-pAV ) = -G-+Q (6.41)
dt 2 z dt H

(6.42)

Where QH is the shear force at hinge H. Note that both shear force Q(X) and
bending moment M(X) are continuous at the hinge since shear deformation and
rotary inertia of beam elements are disregarded, see Table 6.1.
Similarly, by making use of the axial acceleration given by (6.39), the equations
of motion of segment AH in the axial direction can be written as

(G+PA)dd~X = N;; (6.43)

~{( G + pA)Vx } = N: (6.44)

dt .
where N;; = N(A - O,t) and N: = N(A + O,t) denote the axial forces on left and
right sides, respectively, of the plastic hinge H. Consequently, at the moving
hinge the axial force is discontinuous,
N H+ = N-H + PVxdA
- (6.45)
dt
At the travelling hinge this difference in axial force accelerates added mass to the
axial speed of the tip segment.
For initial conditions shown in Fig. 6.9a, during the first phase of motion the
axial velocity Vx satisfies

dVx(t) :5:0, dA
->0 (6.46)
dt dt
Consequently, (6.43) and (6.45) provide
N;; < 0 (6.47)
The distribution of axial force along the cantilever for the case of N: < 0 is as
sketched in Fig. 6.l0c. It will become clear that in the early stage of motion
while A« L, the hinge speed dNdt is large enough that the axial force N: just
ahead of the hinge which is given by (6.44) could be positive (i.e. a tensile
force); it can be so large that it exceeds the fully plastic axial force N p' This is
an unacceptable consequence of rigid-perfectly plastic modeling; it can be removed.
by taking the initial position of the travelling hinge a short distance away from
the tip. This small distance is chosen such that the following relation is always
satisfied:
208 More Complex Configurations

(6.48)

Yield Criterion and Flow Rule By substituting MH , N;; and N; into the yield
function 'l' given by (6.36), one finds that function 'l'takes different values on
the left and right sides of the hinge. For segmentAH with length that satisfies
(6.48), however, the value of 'l' is algebraically larger on the side of the hinge
closer to the impact point, H-. Thus, the yield criterion is satisfied behind the

Iz;H ZJ
travelling plastic hinge; this can be written as

Y'(A) ~ -1 ~0 (6.49)

The discontinuities in rate-of-rotation and axial velocity at the travelling hinge

are related by the flow rule associated with yield criterion (6.49)

~x hlN;;1 _ hN;;
, if N-H > -Np (6.S0a)
-8m 2Np 2Np

~x h
;?: if N-H -Np (6.S0b)
-8m 2'
so that
hN;;Vz
Vx = - 2N A' if N-H > -Np (6.S1a)
P

hVz
Vx ;?: - 2A ' if N-H = -Np (6.S1b)

The equations of motion and kinematic conditions (6.41)-(6.44), (6.49) and

(6.51) provide six equations for seven unknown functions of time, i.e. Vx (t),
Yz(t), A(t), N;;(t), N;(t), MH(t) and QH(t). One more condition is needed to
completely determine the dynamic response of the cantilever after oblique impact.

Limitations on Shear Force at Hinge As shown above, the yield function 'l'
given by (6.36) takes its largest value ('l' = 0) at the generalized hinge. However,
this does not ensure that 'l'is an analytic extreme at X = A, because according to
Table 6.1 for (M,N) interaction the yield function 'l'(X) is not continuous at the
hinge. Nevertheless, the yield condition requires either lim d'l'(A - E)/dX;?: 0
£---70
or O;?: lim d'l'(A + E)/dX. For the yield function (6.36) these inequalities lead to
£---70

~'l'(X) sgn(M)_I- dM(X) + 2N(X) dN(X~ ;?: 0, at X = 11-

dX Mp dX N;
dX
(6.S2a)

:: 'l'(X) = sgn(M)_I_dM(X) + 2N(X) dN(X) ~ 0, at X = 11+ (6.S2b)

QA Mp dX N;
dX
The choice between these two conditions depends on whether the yield condition
is satisfied ahead of (11 +) or behind (A -) the plastic hinge. It will be confirmed
that M(X) > 0 holds for most of the cantilever in the present problem so sgn(M)
= 1. From equilibrium of the beam element and continuity of shear force at the
hinge,
6.3 Oblique Impact on Straight Cantilever 209

aM(X)i = aM(x)i - Q(A) == - QH (6.53)

ax X=A- ax X=A+
Combining (6.52), (6.53) and MplN p =h14 gives

~ N;
2N
aN(X)
ax
I ~ Q
H
~ ~ N; ~N(X)I
2N ax (6.54 )
p X=A- P X=A+
Ahead of the travelling hinge, sections of the beam in segment HB (A+ < X ~ L)
remain stationary so the axial force N(X) is constant in this segment; hence
aN(X)i =0
ax X=A+
Consequently, the limitation on shear force at a travelling hinge (6.54) is recast
as
~
N; aN(X)
2 N ax
I
Q
H
0 ~ ~ (6.55)
P X=A-

Choice of Shear Force at Hinge According to inequality (6.55), there is only a

limited range of admissible shear force at a travelling hinge. The range falls
between two extreme cases:

yr(X)

(A)
in =.MIMp
a 1

-1

Fig. 6.11. (a) Variation of yield function 'l'(X) along the cantilever; (b) stress profile in the (m, ii) plane.
210 More Complex Configurations

1. ~
Q = N;; aN(X) I if alf'(A)laX=O
H 2 Np ax X=A-

2. QH =0 if aMI ax = 0
If QH takes a positive value as in case (1), then the bending moment in segment
HB (A::; X::; L) is determined by M(X) = MH - QH(X - A). As a result, the bending
moment distribution is sketched as the broken line in Fig. 6.10d; the moment
changes sign when X> A + MH/QH' On the other hand, if QH = 0 as in case (2),
then the bending moment in segment HB remains constant, i.e. M(X) = MH in
A ::; X ::; L. This bending moment distribution is sketched as the solid line in Fig.
6.lOd.
The corresponding distributions of the yield function 1f'(X) along the cantilever
are sketched in Fig. 6.11a, where the broken line and the solid line refer to cases
of (1) nonvanishing shear force at the generalized hinge QH > 0 and (2) vanishing
shear force at QH = 0, respectively.
In the plane of generalized stresses (N I Np, M IMp ), the stress distributions at
an instant during the transient phase are represented by the broken and solid line,
respectively, in Fig. 6.11b. In the moving segment AH, as distance from the
hinge decreases, the stress state approaches the yield surface at H-. At this point
case (1) requires the stress variation in AH to be tangent to the yield (limit) locus
if a If'I ax = 0; while case (2) requires the stress variation to have a horizontal
slope.
Both Fig. 6.11a and b indicate that when QH > 0, as in case (2), there can be
other sections ahead of the travelling hinge that also satisfy yield if X is sufficiently
large. In other words, this condition implies that in the early part of the transient
phase, there is more than one travelling hinge and only some of these emanate
from the impact point. On the other hand, QH = 0 as in case (2) ensures satisfaction
of the yield criterion ahead of a single travelling hinge. While it is simpler to
analyze only a single travelling hinge, there are no additional theoretical conditions
that determine the shear force at the travelling hinge other than the limits specified
in (6.54); i.e. there is not a unique solution when stretching and flexural motions
are coupled through an interaction yield surface.

Equations of Motion with QH = 0 A complete transient solution with only a

single generalized hinge can be imposed by assuming that shear force vanishes at
a travelling hinge QH = 0; at this hinge the flow rule relates flexure and stretching.
With QH = 0 the previous equations of motion (6.41) and (6.42) are recast as

~(~ AV) = _GdVz (6.41 ')

dt 2 P Z dt

dt 6 PAV) =
~(~ M
2 (6.42')
Z H

These equations, together with (6.43), (6.44), (6.49) and (6.51), constitute six
governing equations for six unknown functions Vx (t), Vz(t), A(t), N;; (t), N: (t)
and MH(t).
A direct integration of Eq. (6.41') results in

v = ~o (6.56)
Z 1+ pAI2G
6.3 Oblique Impact on Straight Cantilever 211

Then, from (6.51),

N;' 2VxA(l + pAI2G)
(6.57)
Np hYzo
where the initial value of A is small enough that the axial force at the hinge
satisfies the yield condition N;'INp ~-1. In accord with (6.43) this axial force
gives an axial acceleration for the tip segment; using (6.43) and (6.49), one finds
dVx 2Np VxA(1 + pA/2G)
-= (6.58)
dt hYzo(G+pA)
while Eqs (6.43) - (6.45) together with the yield condition (6.49) give

dA 6 (1 + pA/2G)2 Mp f
(2VxA(1 + PAI2G))2}
dt 2
= pVzo {2A(1 + pAI2G)-pA I2G} 1 - hVzo 1 (6.59)

Equations (6.58) and (6.59) require N;' > -Np . Note that they are first order
o.d.e. for two unknown functions Vx (t) and A(t); the appropriate initial conditions
for the equations are
Vx(O) = Vxo ' A(O) = 0 (6.60)

Nondimensionalization To develop general understanding of the system behavior,

it is helpful to obtain solutions in terms of the following nondimensional variables:
== Wx == Wz X 1 A ~
Wx L Wz L x="L' /1-=="L' To==L~Mp

Vx v = Vz _ Vzo (6.61)
r == vx =
T'
o
- LIT'
o
z - LIT '
o LITo

y _ G m == M n == NL == t;N
pL' Mp , Mp Np
where the slenderness ratio for beams of rectangular cross-section is defined as
~ == 4L1 h. Then, Eqs (6.56), (6.57), (6.45), (6.49), (6.58) and (6.59) can be recast
in nondimensional form

v = vzo (6.56')
z 1+ AI2y

~2vxA(1 + A/2y)
nH (~ -~) (6.57')
2v zo

nH+ n;' +v x -
dA (~ ~) (6.45')
dr

mH 1-( ~vxA(1 + A/2 y

2vzo
)J (6.49')

dvx = ~2vxA(1 + AJ2y)

(6.58')
dr 2V zo (Y+A)
212 More Complex Configurations

dA, 6 (1+A,/2r)2 {(~VXA,(I+A,l2r))2} (6.59')

d'l" ~ {2A.(1 + A.l 2r) - A,2 12r } 1- 2vzo
The initial conditions related to (6.58') and (6.59') are
vx = Vx 0 ' A, = 0, at 'l" = 0 (6.60')

6.3.3 Solution Based on Single-Hinge Mechanism

Removing Singularity from the Initial Conditions As shown in Eq. (6.60'), the
travelling hinge is initially located at the impact point so A, = 0 at r = 0;
consequently (6.58') and (6.59') indicate that

dvx = 0 dA
-~oo, at 'l"=0 (6.62)
d'l" ' d'l"

loading, Eq. (6.45') leads to n:

But if both transverse and axial momentum are imparted to the tip by impulsive
~ 00 at 'l" = 0, and this violates the yield criterion.
Whenever axial forces are coupled to the bending moment through an interaction
yield criterion, this analytical difficulty arises immediately from discontinuous
impulsive loading. It occurs because of the infinitely fast moving plastic hinge
that emanates from points where momentum is discontinuous.
To avoid the singularity at 'l" = 0, the initial conditions (6.60') can be altered
by assuming that the hinge is initially a short distance away from the impact
point. Let this short distance from the tip be ,10« 1; with this numerical
approximation, the initial values of hinge speed d AI d 'l" and axial force
be estimated from Eqs (6.59'), (6.57') and (6.45'),
n: can

dA, 3
dr vzoAo'

Hence, the requirement of condition (6.48), i.e. In;;-I ~ In:

I can be satisfied if the
hinge is initially located a small distance A,o from the tip,

,10 ~ -J3/~ (6.64)

or in dimensional form,
A > h-J3
0- 4 (6.64')

This shows that as long as Vxo is of the same order as Vzo ' the assumed starting
distance Ao should be the same order as the depth h of the beam and therefore
very small in comparison with the length L of the cantilever. Based on this
discussion, the initial conditions (6.60') are replaced by

(6.65)
The present modeling is based on predominately flexural motion and thus is not
appropriate for primarily axial impact where V xo I Vzo » 1.

Iterative Procedure for Numerical Solution The following procedure is employed

to obtain numerical solutions for this problem.
6.3 Oblique Impact on Straight Cantilever 213

41
01
I:
:.c
41
:£
'0
0'
1: .....

~ ~ 0.5
8.""
a
.~
1/1
I:
41
E
'2g
0.02 0.04 0.06 0.08
time r = tlTo

Fig. 6.12. Movement of plastic hinge along cantilever subjected to oblique impact. Energy ratio
eo " Ko IMp =1.0; mass ratio r" G I pL =0.2 and 1.0; angle of obliquity V' =0°, 30 and 60°.
0

1. Specify the four nondimensional parameters that appear in the governing

equations and initial conditions, i.e. y", vxo and vzo ; or better still,
y, " (v xo !vzo ) and the nondimensional initial kinetic energy
eo =y(v;o +vto)/2.
2. Numerically integrate Eqs (6.58') and (6.59') with initial conditions (6.65) to
find A(r) and vx (r) during period before time 'l' = 'l'u when axial motion
ceases, vx('l'u) =0.
3. Obtain v('l') , n;('l') , n~('l') and mH('l') , for the coupled phase of motion
'l'o ~ 'l' ~ 'l'u with the help of (6.56'), (6.57'), (6.45') and (6.49'), respectively.
4. After axial motion ceases (i.e. for 'l' ~ 'l'u)' transverse motion and flexural
deformation continue. The governing equations during 'l'u ~ 'l' ~ 'l'f' are the
same as those in Sect. 4.5; the solution can be obtained in the previous manner.
The transverse motion finally ceases at 'l' = 'l'f'

'E41 / r\0.0049
E
o M/Mp---
E
0141
I: U N/Np -------
:00
1:-
1l:§
_ >C
01Y
o 0 ~---- ------------ --'\.---------------------------------
5'1:1
.~ :5 :-r=0.00000012 r =0.0049
41 I r. G/pL = 1
E
:0 t' = 0.0012 '" = 3 O·
I:
o ~...... ------~---------------------------------------
I:
-10~--~--~0~.2~~--~0~.4~~--·~0~.6~~---0~.~8--~~
x =X/L
Fig. 6.13. Distributions of bending moment and axial force at various instants after oblique impact,
eo " Ko IMp =1.0, r" G I pL =1.0 and V' =30°.
214 More Complex Configurations
'<;
~ 0.8.----,---,---,---,---,--,--.--,--,----,
~
0'1
C
yr= 60·
~ 0.6
~
Ui
=
y SlpL = 0.2 - - -
>0 Y li S/pL = 1
~ 0.4
~
4> II' = 30·
~ 0'2If~"- - - - - - - - 1
4> I
o
- II
.e-
Ul
:
I
.~ 0 ~/-L-__::l::_--1-__::J..,_-..L---;;_l;:_-...L-----;;:_';;_---'--~
'0 0 0.2 0.4 0.6 0.8
x= XIL
a

QO.8.----,---,---,---,---,--.---r--,--,--~
~
J YSGlpL=0.2--

.So.6
'0
~
.0
>0
.0
>0
0'1
L.
0.4
--- ------
4>
C
~ 0.2
4>
C;
a.
'iii ----- ------ --------
.~
'0 0.2 0.4 0.6 0.8 1
X= XIL
b

Fig. 6.14. Energy dissipation accumulated from tip to cross-section at x =XlL from (a) axial deformation and
(b) bending.

Numerical Examples For mass ratios r

=0.2 and 1.0, and angles of obliquity
for impact lfI = tan-l(VxoIVzo) = 0 0 ,30 0 and 60 0 , numerical calculations have been
carried out; some results are depicted in Figs 6.12-6.15. In all these examples,

slenderness ratio Uk = 25 (i.e. 100). ,=

the nondimensional input energy eo == Ko IMp = 1.0 remains the same, and the

Figure 6.12 shows the variation of the non dimensional position of the travelling
hinge with time. The figure shows that for larger angles of obliquity the average
speed of the hinge during the transient phase is reduced while axial motion exists.
The bending moment and axial force throughout the beam at various instants of
time are shown in Fig. 6.13 for a mass ratio r
= I and impact angle, lfI = 30 0
The axial force increases rapidly in the initial period while bending moment
decreases; subsequently, the axial force decreases before axial motion ceases.
After axial motion ceases, the hinge is still near the tip. Thereafter the bending
moment equals the fully plastic moment. This indicates that an axial component
of impact only affects the early part of the transient phase of dynamic response.
The axial momentum is completely eliminated while the hinge is still close to the
6.4 Circular Arc Cantilever Subjected to In-plane Step Force 215

0r-__0,.2~==0.~4==~0~.6~~0.~8~~
x= X/L

0.2 ~

/
'/~/
,,
// l/I= 0',30',60'
"
~0.6

/~
0.8 l/I = O~30~60'
yEG/pL=O.2
rE G/pL =,

Fig. 6.15. Final deformed shapes of cantilevers subjected to impact. Energy ratio eo" Ko IMp = 1.0; mass
ratio r" G I pL = 0.2 and 1.0; angle of obliquity 'I' = 0°, 30° and 60°.

impact point; thereafter, the response is essentially the same as that of the cantilever
subjected to a transverse impact (see Sect. 4.5).
Figure 6.l4a and b shows the energy dissipated by axial and flexural
deformations before the hinge transits any location x. As shown in Fig. 6.l4a, if
the axial deformation ceases before the travelling hinge reaches the root of the
cantilever, then the total energy dissipated by axial deformation (tension or
compression) is independent of the mass ratio r and depends only on the obliquity
lfI of impact. Figure 6.l4b shows, however, that the energy dissipated by bending
varies significantly with both the mass ratio r and the obliquity lfI of impact.
The sum of energies dissipated at the interior cross-sections (0 ~ X < L) by flexural
and axial deformations is usually less than the input energy Ko; the remaining
energy is dissipated by rotation about the stationary hinge at the root.
The final deformed shape resulting from oblique impact at a nondimensional
initial energy eo == Ko / Mp = 1.0 is depicted in Fig. 6.15. This illustrates that the
axial component of impact has almost no influence on the final shape of the
cantilever.

6.4 Circular Arc Cantilever Subjected to In-plane

Step Force

6.4.1 Engineering Background and Assumptions

Structural members that are not straight are frequently employed in machine parts,
buildings and pipe networks; common examples are curved beams, bent beams,
arches, circular rings and simple frames. In particular, the dynamic behavior of
curved beams is useful for risk analyses of pipework systems.
A sudden or guillotine break in pressurized piping gives a suddenly applied
follower force acting on the broken end of the pipe and this causes pipe-whip.
216 More Complex Configurations

The jet reaction forces exerted on a fractured pipe by escaping fluid can be large
enough so that the broken pipe undergoes large dynamic deflection and rotation.
This is a hazard to other pipes or equipment in the vicinity. Since plastic deforma-
tion in a pipe-whip process is usually much larger than the elastic deformation, a
rigid-plastic model for pipe deformation has been employed in studies of pipe-
whip. According to measurements from experiments, the jet forces exerted on
broken pipes are somewhat similar to rectangular pulses so analyses of straight
cantilevers subjected to step or rectangular pulse loading may be used directly as
first-order approximations for pipe-whip prediction. This type of analysis was
given in Sects 4.1 and 4.2.
When the postulated break is near an elbow, it is necessary to analyze the
dynamic behavior of a curved cantilever. Since the jet reaction force can be in
any direction, the pipe may be subjected not only to a bending moment, but also
to torque and axial force. If the dynamic force or impulse acts in the plane of the
curved beam, then the response is primarily bending. If the load is out-of-plane,
torsional deformations can be equally large.
With pipe-whip of a curved section of tubing in mind, Yu and Johnson [1981]
first examined a quadrantal cantilever beam of circular plan form that is subjected
to in-plane step loading at the tip. They employed an interactive yield criterion
between bending moment and axial force and found that a stationary plastic hinge
at an internal cross-section requires a step force within a certain range of magnitude
and direction. For some directions of the force, any deformation mechanism with
an internal hinge violates the yield criterion.
In Sects 6.4 - 6.6 we examine the dynamic response of circular curved
cantilevers under various dynamic loading conditions. In these analyses, in addition
to the rigid-perfectly plastic idealization, it is assumed that
1. the curved cantilever has uniform cross-section and constant density p per
unit length;
2. the cross-section dimensions are much smaller than the radius of the center
line R, so the influence of shear and axial forces on yielding can be neglected
(however, the interaction between bending and torsion will be considered for
the case of out-of-plane loading, in Sect. 6.6);

a b

Fig.6.16. Circular cantilever subjected to step force at the tip: (a) configuration with plastic hinge H; (b) free-
body diagram of arc element AC within segment AH, showing the sign convention of bending moment M, shear
force Q and axial force N.
6.4 Circular Arc Cantilever Subjected to In-plane Step Force 217

3. the deflections are so small in comparison with R, that the dynamic equations
and geometric relations can all be written on the basis of the initial configuration.
In the present section, we first consider a curved cantilever with a center line
at radius R; the center line subtends an angle a (tr/2::;a::; tr) of a circular arc.
End B of the circular cantilever is clamped while the free end A, is subjected to a
suddenly applied radial force Fr or tangential force F~ that acts in the plane of
the arc, as shown in Fig. 6.16.
If the magnitude of the force is constant, the dynamic deformation mechanism
does not change with time; i.e. the response takes a stationary pattern of
deformation. In other words, the dynamic response to a steady force is a modal
solution, provided the material is assumed to be rigid-perfectly plastic and only
small deflections are considered. Also, the effects of shear and axial forces on
yield are neglected. Under these assumptions, the circular cantilever starts to
deform if F; (radial force Fr or tangential force Frp) is as large as the static
collapse load Fc' The collapse mechanism of the circular cantilever is characterized
by the location of one or more plastic hinges. If F; > Fc, the cantilever deforms
in a dynamic mode which is different from that of static collapse because of
inertia. This dynamic mode has a constant acceleration field; the number of hinges
and their locations depend on both the applied and inertia forces, i.e. they are
functions of F; and the mass per unit length p, the radius of the curved beam R
and the fully plastic bending moment M p of the cross-sections.

6.4.2 Radial Force at Tip

For a circular arc cantilever the spatial coordinate is chosen as the arc length
along the center line S = Rl/>, where l/> is the angular coordinate measured from
tip A. By taking the sign conventions for bending moment M and shear force Q
as shown in Fig. 6.16, M and Q are related by dM I dS =-Q or dM I dl/> =-RQ.
The shear distribution Q(S) can be calculated from the equations of motion of a
generic segment such as AC. Integration of the shear resultant along the cantilever
then gives the bending moment distribution M(S). Using this formulation, both
global variables (i.e. the rotational accelerations and hinge positions) and the
local distribution of moment can be obtained simultaneously. For a cantilever
subjected to a radial force Fr at the tip (Fig. 6.16), there is no deformation if
Fr < Frc = M pi R; if Fr = Frc , a plastic hinge H forms at f3 = tr I 2.

Force Frc < Fr < Fr2 For Fr > Frc , the position angle f3 of the hinge H moves
from f3 = tr 12 towards the loading point A as the force increases. This single-
hinge mode is valid for Frc < Fr < Fr2 . If Fr ~ Fr2 , a second hinge occurs at the
root B. The complete analysis is carried out as follows.
If Frc < Fr < Fr2 , the cantilever deforms with a single hinge H located at angle
f3 from tip A. The equation of motion for acceleration of segment AC in the
direction CO is used to derive the shear force Q at a generic point C on segment
AH (Fig. 6.16); this relation does not involve the axial force at C. The center of
mass of segment AC is designated by D in Fig. 6.16; it is located a distance OD
from the center of the circular arc,
OD = R sin(l/>I2)
l/>12
218 More Complex Configurations

where cp = LAOC . The radial line OD bisects LAOC.

The acceleration of point D has a component in direction CO,

m[ODsin(1/> 12) + Rsin(f3 -1/»] = mR[(l- cosl/»I I/> + sin(f3 -1/»]

m
where is the anticlockwise angular acceleration of segment AH. Using the above
expression, the translational equation of motion for segment AC in direction CO
can be written as
Fr cosl/>- Q = (pRcp)mR[(l- cos 1/»1 I/> + sin(f3 -1/»]
(6.66)
= pmR 2 [1- cos I/> +l/>sin(f3 -1/»]
As the effects of shear and axial forces on yield are neglected, the bending moment
M is a local extremum at hinge H where cp = f3, so at the hinge the shear Q = O.
Hence, from Eq. (6.66),
F,. cosf3 = pmR 2 0- cosf3) (6.67)
In the rotating segment, the moment M(8) is found by integrating the expression
for the shear force with the boundary condition M(O) = 0; this results in
M = FrRsinl/>- pR 3 m[l/>- sin I/> + cpcos(f3 -cp) + sin(f3 -I/» - sinf3] (6.68)

The bending moment at the hinge M(f3) = Mp , which gives

M p = FrRsinf3 - 2pR 3 m(f3 - sinf3) (6.69)

From Eqs (6.67) and (6.69), Fr and mare determined as functions of f3,

F = Mp 1 - cos f3
(6.70)
r R sin f3(l + cos f3) - 2 f3 cos f3

0
30 30 Fr.
Fr

'."./)
~ 24
Cl
+ Q.

Q:: ~ 24
Q::
o.t:
.:: •
09

~
11

~" 18
"
....!:!. 18
.:: "-
.:: F 1. 93< 'r<9.09
"
oo!" "

"I'~
Fr ..!-
CII 12 12

1<1,<~
...
u CII
...u
.E .E
6 6

0 0
0 Jr/G 1113 nl2 2Jr/3 P 0 Jr/2 Jr P
hinge position hinge position
(measured from loading point) (measured from loading point)
a b
Fig. 6.17. Dependence of hinge position f3 on the force magnitude fr '" Fr / Fre for a circular cantilever
subjected to a transverse step force Fr at the tip. in case of (a) a = 210 / 3 and (b) a = 10.
6.4 Circular Arc Cantilever Subjected to In-plane Step Force 219

. Mp cos 13
w = ------------~------- (6.71)
pR sinf3(1 + cosf3) - 2f3cosf3
3

The hinge location 13 is shown in Fig. 6.17a and b for cantilevers with included
angles a = 1200 and 180 0 , respectively. The corresponding moment distributions
are shown in Fig. 6.18a and b ..

Force Fr > Fr2 The single-hinge mode of deformation will only be valid for a
limited range of force. If Fr ~ Fr2 , two hinges are required to satisfy the yield
condition; the location of the second hinge can be determined by considering the
moment distribution in segment HB. This moment depends on the axial force NH
at hinge H. The axial force NH can be obtained by formulating the translational
equation of motion for segment AH in the direction of the tangent at H,
(6.72)

The axial force NH at H is due to both a component of the applied force and a
component of the acceleration of the tip segment; thus the preceding equation can
be rearranged as

N = pR 2 w. (13 - sm
. 13) - F sm
. 13 = - -
Mp sinf3 - 13 cos 13
- ------''----'----'--- (6.73)
H r R sinf3(1 + cos/3) - 2f3cosf3
Using this expression, the moment at the root Me can be written as
Me = NHR[I- cos(a - 13)] + Mp (6.74)

Substituting Eq. (6.73) into Eq. (6.74), the limiting condition Me = -Mp yields
(sin 13 - f3cosf3)[I- cos (a - 13)] = 2sinf3(l + cosf3) - 413 cos 13 (6.75)

The angle 13 for formation of a second plastic hinge can be calculated for a given
value of a from this expression; the minimum force for development of this
second mode is denoted as Fr2 , which is a function of the included angle a. For
a radial force of arbitrarily large magnitude at the tip and a statically admissible
single-hinge mode of deformation, the hinge can be located no further from the
tip than the quarter circle 13 = nl2.
If Fr > Fr2 , the cantilever deforms in a double-hinge mode with the newly
formed hinge located at the root B. The location of the first hinge H moves
towards the loading point as Fr increases (Fig. 6.17). The equations of motion
for this mode are formulated as follows.
1. From the rate of change of moment of momentum of HB about the additional
hinge at B (see Fig. 6.16) we obtain
-NHR[I- cos(a - /3)] - 2Mp = 2pR 3 il[a - 13 - sin(a - 13)] (6.76)

2. The equation of translational motion of segment AH in the radial direction HO

gives

Fr cos 13 = Pf3R[ cO R sin ~~12) - uiR sin( a ;: 13 )cos( a ; 13 J] (6.77)

where the term in the square bracket is the component of acceleration o~ the
center of AH in direction HO. Note that a positive angular acceleration Q of
segment HB is clockwise while positive cO of segment AH is anticlockwise.
220 More Complex Configurations

_ 0.0 ~------------''"-c----'\~----i
c '
QI '
E
o
~
.
" ,
E '. ,
-0.5 .... "
", ,
.....~."
'.,
-1.0 L----L_....l...-_L---1_......l..._-'-_L-----L_......l..._-'-_'--:-.-'>I
.•,
o lC/6 n/3 lC/2 21f/3
angle measured from loading point cp
a

- ..........
""
""
"",
,,
,,
,
,,
,,
C
QI
E
o
E
,
-0.5

angle measured from loading point cp

b

Fig. 6.18. Bending moment distribution in cantilever of circular plan form subjected to transverse step force
F,atthetip,incaseof(a) a=2n/3 and (b) a=n.

3. The equation of translational motion of segment AH in the tangential direction

at H gives
. /3 = p/3'D[
NH + Fr sm "D(1 -
"OJI\. Sin(/3I2)COS(/312») + 2nD'
~.m.sm 2(a
--- /3)] (6.78)
/312 2
4. For segment AH, the rate of change in moment of momentum about the centroid
results in

FR sin2 (/312) -N i1- Sin(/3I2») = /3R 3W[1_(sin(/312»)2]+M (6.79)

r /312 H /312l P /312 p l
When a force Fr is given, the corresponding hing~ location /3, hinge force
N H , and angular accelerations of the segments wand .Q can be determined from
6.4 Circular Arc Cantilever Subjected to In-plane Step Force 221

240
20
210

180

150
u u12
~ c...'&
"- 120 "-
c..."" L..:!'
"
.....'& 90 ~
II 8

60

30

0
0 Tf/6 rr/3 1[/2 21C/3 f3
a b

Fig. 6.19. Dependence of hinge position f3 on the force magnitude it/!" F", / F¢C for a circular cantilever
subjected to a tangent step force F¢ at the tip, in case of (a) a; 211: / 3 and (b) a; 11: •

Eqs (6.76) - (6.79). The Ir vs. b relation is shown in Fig. 6.17. Using the same
procedure as before, the moment distributions are computed and shown in Fig.
6.18 for various ranges of Fr' It is found that when a < 158.0° , this double-hinge
mode is valid for all subsequent values of Fr , while if a > 158.0° , the hinge at
root B moves away from the root as Fr increases.
To summarize, under a radial force Fr , the static collapse force Frc = M p / R
results in deformation with a single hinge located at f3 = n/2. If Frc < Fr < Fr2'
the location of the single hinge moves towards the loading point as Fr increases.
If Fr = P,2' a second hinge develops; this is located at the root. Radial applied
forces Fr > P'2 require a second hinge located at the root for included angles
90° :=; a :=; 158°, while larger angles have a second hinge located inboard of the
root if the force is sufficiently large.

6.4.3 Tangential Force at Tip

Consider the same cantilever as in Sect. 6.4.2 but with a tangential force F,p
instead of the radial force of the previous case. All other conditions are identical.
Using the same procedure we obtain somewhat different results.
The fully plastic collapse force for a tangential force F", applied at the tip
of a curved cantilever with constant radius R that subtends an angle a is
FI/Jc = M p/[R(l- cosa)]. If the applied force F", = FI/Jc' plastic collapse occurs with
a hinge at the root B; in this mechanism the entire cantilever rotates about B. The
equation of motion is written as follows:
F",R(1- cosa) - Mp = 2(a - sina)mR 3 (6.80)

In calculating hinge positions for larger values of FtjJ' it is found that a second
hinge forms at locations that depend on the included angle a. When a < 133.6°,
the second hinge H forms between B and A at a characteristic force FtjJ2 (a)
222 More Complex Configurations

1.0
/~ \
I \
cf},' ,
'V./ ,
"'- 0.5
l: ~/ \
"- III ~ ,
~
.-.,""/ "v,,,/-'" \
cCII / 1/"
,~
'\'
, \
E
0
E
0.0
.,,,
\\, I
-0.5
\\..
I

-1.0
0 TriG lr/3 lr/2 2]"[/3
angle</> measured from loading point
a
0.0

-0.2

"'-
~ -0.4
l:
C
~ -0.6
a
E
-0.8

-1.0
0 ]"[

angle </> measured from loading point

b

Fig. 6.20. Bending moment distribution in cantilever of circular plan form subjected to tangent step force F~
atthetip,incaseof(a) a=21r/3 and (b) a=1r.

(Fig. 6.19a). As F¢ increases, the cantilever deforms with two hinges for
F¢2 ::; F¢ ::; F¢3(a). With an applied force F¢ > F¢3(a) a third hinge forms between
H and the root B; modes for F¢ > F¢3 have three or more hinges.
If a > 133.6°, the location of the hinge originally at B moves towards the tip
as F¢ increases (Fig. 6.19b). If the force F¢ = F!p2(a) the first hinge is located at
H away from the root and a second hinge forms at the root B. If F¢ > F¢2' the
mode has two hinges; the first hinge H continues to approach the loading point as
F¢ increases while the other hinge stays at the root. The hinge at H bifurcates at
a characteristic value of applied force, F¢3'
The modal hinge locations as a function of applied load are shown in Fig.
6.19a and b for a = 120° and 180° respectively. The corresponding moment
distributions are shown in Fig. 6.20a and b.
In summary, the circular cantilever has a static plastic collapse force
F!p = M p /[R(I- cos a)] . For somewhat larger forces, F¢ > F¢2' the beam deforms
with one hinge either at or near the root B. If a < 133.6° , one hinge remains at
6.4 Circular Arc Cantilever Subjected to In-plane Step Force 223
20.---,----------------------,rr/2

«l.
C
o
"0
IT/4 1l
GI
~

'---
t1l
C

-----~---
£.

1.0 0L - . - L - -2L - . - - -4'-----'o-6-----!-a---'J..0----"2 0

FrRIMp

Fig. 6.21. Hinge position and magnitude of axial force at hinge as functions of the magnitude of a transverse
step force F, at tip of circular cantilever, in case of a = If 12 .

the root and when F¢ > F¢2(a) > F4Jc a second hinge forms within the beam.
If a> 133.6° , the first hinge forms near rather than at the root. The location of
this hinge moves away from the root as F¢ increases; a second hinge forms at the
root when the applied force exceeds a limit value F¢2' For still larger force there
is another limiting magnitude F¢3 where the hinge closest to the loading point
bifurcates. If F¢ > F¢3 a three hinge mode appears.

6.4.4 Discussion
With a radial force F" there are two different patterns of response depending on
whether a is larger or smaller than 158.0°; similarly, for a tangential force F¢
two distinctive patterns of modes are separated by an arc of 133.6°. With either a
radial or tangential force, the curved segment immediately adjacent to the force
has an arc length that monotonically decreases towards zero as the magnitude of
the force increases.
There are some differences between the modes for radial and tangential forces.
For the same curved structure, the number of hinges increases more rapidly for a
tangential force because there is a large axial component of force at hinges near
the tip. For a curved cantilever, the number of plastic hinges in a mode increases
without bound as the magnitude of a tangential applied force increases. For the
same structure with a radial force, dynamic modes have only a small number of
plastic hinges although the force may be very large.
This effect in curved beams can be understood by considering a quadrantal
cantilever (a = nI2). With a radial force at the tip the axial resultant force at the
hinge H is NH <2MpIR for all F" as shown in Fig. 6.21. The axial force NH at
hinge H is not large enough to form another hinge at the root B and hence the
modes for this direction of force have only a single hinge irrespective of the
magnitude of Fr' In contrast, with a tangential force F¢ at the hinge closest to
the loading point, the axial force increases without bound as F¢ increases;
correspondingly, the number of hinges for a mode also increases in order to
satisfy yield.
224 More Complex Configurations

Stronge et al. [1990] presented modal analyses that brought out these characteris-
tics for several different arch, ring and curved beam structural elements. 1 They
found that dynamic modes for slender curved structures are sensitive to displacement
constraints that induce axial forces. When a step force acts at an interior point of
a structure, displacement constraints at the loaded point result in an axial force at
the point where the load is applied that increases with the load. This axial resultant
force depends on both the radial and tangential components of the applied force.
In curved structures where axial resultant forces increase without limit as the
load increases, the number of rigid segments in a dynamic mode can also increase
without limit.
These conclusions arise when interactions between axial and bending stress
resultants are neglected; i.e. when the yield condition depends solely on the bending
moment. With a constant yield moment at plastic hinges and an impulsive load, a
closed curved structure such as a ring has no perfectly plastic solution that is
both kinematically and dynamically admissible. For large forces (or localized
impulsive loads) the initial stage of motion is dominated by plastic axial
compression at hinges close to the load and this cannot be neglected. Applied
forces that induce large axial stress resultants require use of a yield condition
that couples axial and bending stress resultants in deforming segments; i.e. an
analysis similar to that in Sect. 6.3. The solutions presented by Stronge et al.
[1990] demonstrate that the effect of the axial stress resultant on yield is not
negligible when large axial resultant forces are induced by load and geometry.
The example of a rigid-plastic ring with a yield polygon that couples axial and
bending stress resultants was analyzed by Cline and Iahsman [1967] for a smoothly
distributed impulsive load. The early time solution exhibited a plastic compression
segment of changing length within the loaded section; this segment finally dissipated
a substantial part of the initial energy. For a force applied at a point on a curved
structure, however, the corresponding plastic compression segment degenerates
to a hinge at the loading point. Although the analysis developed in the present
section neglects the effect of axial stress resultants on yield, it is useful for
developing general understanding of mechanisms for deformation of suddenly
loaded, curved structures.

6.5 Circular Arc Cantilever Subjected to In-plane

Impact
In this section, we consider a quadrantal circular cantilever beam subjected to
radial impact at the tip by a rigid particle mass travelling in the plane of the
curved beam; this problem was first investigated by Yu et al. [1985b, 1986].
The basic assumptions have been declared at the beginning of Sect. 6.4. As
depicted by Fig. 6.22a, suppose that the collision imparts an initial inward velocity
Vo in the radial direction to a particle of mass G at the tip of the curved cantilever.
Here only a small deflection analysis is presented; a further analysis of this problem
based on a large deflection formulation is given by Zhang and Yu [1986].

I Dynamic response of a rigid-plastic ring was previously studied by Owens and Symonds [1955], Cline
and lahsman [1967] and Zhang and Yu [1989]; the large deflection dynamic response of arches was
analyzed by Palomby and Stronge [1988].
6.5 Circular Arc Cantilever Subjected to In-plane Impact 225

Section 6.4 showed the existence of multi-hinge mechanisms, but these are
related to various subtending angles of curved cantilevers that are larger than
a == n12. In the present problem where a == nl2, the quadrantal circular configura-
tion and impulsive loading allow the complete solution to be constructed using a
single hinge mechanism which travels from the tip to the root, provided the effects
of shear and axial force on yielding are neglected.

6.5.1 Rigid-Plastic Formulation

Velocity Field In a single-hinge mechanism, the dynamic deformation can be
expected to have two distinct phases: a transient phase (phase I) and a modal
phase (phase II). In phase I, a plastic hinge travels from the impact point (Le. the
tip of the curved cantilever) towards the root. Suppose that the collision occurs at
time t == 0; and at time t > 0 a plastic hinge forms at an interior cross-section H,
located at angle {3, as shown in Fig. 6.22b. Arc segment AH rotates about the
instantaneous center H as a rigid body, with angular velocity m, while arc segment
HB remains at rest. Hence, magnitudes of radial velocities related to this mechanism
are given by

W(tP) = j 2mRsin ({3 - tP)

0,
2 '
0 $. tP $.

{3$.tP$.nI2
{3
(6.81)

where tP denotes an angle measured from the vertical Z-axis to an arbitrary

cross-section C in the curved cantilever, see Fig. 6.22b. Here angle tP plays the
same role as arc length measured from the tip. In particular, the velocity at the
tip is given by taking tP == 0 in (6.81); that is,

V = W(O) == 2mRsin Ii2 (6.82)

A comparison of (6.81) and (6.82) leads to

W(</1) == j Vsin ({3-tP) Isinli

0,
2 2'
{3 $. tP $. nl2
(6.83)

z z
a b
Fig. 6.22. (a) A quadrantal circular cantilever subjected to in-plane radial impact by heavy particle at the tip;
(b) deformation mechanism with single hinge at interior section H.
226 More Complex Configurations

Fig.6.23. Change in direction of velocity at section C in arc AH when plastic

hinge moves from HI to H 2 ·

which relates to the distribution of the velocity magnitude to the tip velocity Y
and the hinge location 13.
Let i and k denote unit vectors in the directions of coordinates X and Z,
respectively. Note that the direction of velocity W is perpendicular to chord HC,
see Fig. 6.22b, so that the velocity vector at cross-section C within arc segment
AH can be written as
W = - WSillll'i + Wcosljlk
(6.84)
-WSin(f3 ;~} + wcos(f3; ~)k
where a geometric relation is defined as ljI = (13 + ~ )/2.
Acceleration Field To obtain the acceleration field, the differentiation of (6.84)
with respect to time t gives

W = dd~ = -{wsin(P;~)+w~cos(f3;~)}i
(6.85)
+ { Wcos( 13 ; ~ ) - W~ sin( 13 ; ~ ) } k
Accordingly, the magnitude of the acceleration is found to be
(w/ = W· W = W2 + W2 p2 /4 (6.86)
for a cross-section C within arc segment AH (0 ~ ~ ~ 13); and in particular,
(\1)2 = V'V = V2+y2p2/4 (6.87)
for tip A (~= 0). The first term on the right-hand side of (6.86) and (6.87)
indicates that the accelerations at C and A are partly caused by the variation of
velocity magnitude related to the change in the angular velocity about hinge H;
while the second term shows that the accelerations are also partly caused by the
change in the direction of velocities at C and A due to the movement of hinge H.
As shown in Fig. 6.23, ~hen a h.inge moves from HI to H2 the velocity vector at
point C changes from WI to W2 , and this change in direction causes an axial
component of acceleration.
The magnitude of W, which is only the variation of W in the current direction,
can be obtained directly from (6.83):
Vsin(f3 - ~)12) + .!. Yf3' sin( ~12)
(6.88)
W(~) = sin(f3/2) 2 sin 2(f3/2)
6.5 Circular Arc Cantilever Subjected to In-plane Impact 227

To apply Tamuzh's variational principle for rigid-plastic dynamics, Eq. (2.20),

it should be noted that in the present case, the functional related to this principle
is
. .) 1 P .... r 1··
J = J (V,fJ = "2J pW·WRdf/J + ·2GV.V+MpOJ (6.89)
o
where OJ = VI[2Rsin(fJI2)] is the instantaneous angular velocity of segment AH
about point H. Substituting Eqs (6.83), (6.86),. (6.87) and (6.88) into (6.89), one
finds

. fJ'
J -_ J(V, ) 1
= -pR i f3 { . 2 sin 2 (fJ - f/J)I2)
V sin(fJ -f/J)/2)sin(f/JI2)
+ VV'13' --'--------;;,...,..--''-:--'---'-
2 0 sin 2(f3I2) sin 3 (fJl2)

1 2f3'2 sin 2(f/J12) 1 2fJ'2 sin 2(fJ-f/J)/2)}

+ -V. + -V. df/J (6.90)
4 sm (f312)
2 4 sm 2 (fJI2)

1 {'2
V2/3 2 }
+ -G V + - - + -
Mp VSin(fJI2)-0.5V/3cos(f3I2)
2 4 2R sin 2(fJI2)
According to Tamuzh's principle the actual acceleration field makes J a minimum;
that is,
aJ = 0 a~ = 0 (6.91)
ali ' af3
which gives the following pair of equations

V.{fJ - smfJ
. G
+ -(I-cosfJ)}+.VfJ fJ -
{1 - I+COS
-.- fJ } Mp.
= --sm(fJI2) (6.92)
pR 2 smf3 2 pR

V·{I - -fJ 1+ cos fJ} + -v/3 { f3 -2 sin fJ + f3 -sm. f3 +-sm

2G. (f3 12 )} -_ --cos
Mp
2
(fJ I2 )
2
2 sinfJ 4 sin (fJI2) pR 2pR
(6.93)
From the sum of twice (6.93) and the product of cot(fJI2) and (6.92) we obtain

V{I-COSfJ+~sinf3}
pR
+ v/3{2f3-sinf3+~(1-COSf3)}
2 pR
= 0 (6.94)

By defining the following set of nondimensional variables

To=R~~' r=~o' w=~, v='RfTo ' r=p~ (6.95)

the basic equations (6.94) and (6.92) are recast in nondimensional form as

dV[l-cosfJ +rsinfJ] + ~ dfJ[2fJ -sinfJ +r(l-cosf3)] = 0 (6.96a)

dr 2 dr

dV[f3 -sinf3 +r(l-cosf3)] + ~ dfJ[2'_fJl+.COSfJ] = -sin(f3I2)

dr 2 dr smfJ (6.96b)
Solving (6.96) for (dv/dr) and (dfJldr), one obtains
228 More Complex Configurations

dv
= sin(,B12X2,B - sin,B + r(l- cos,B)]l2r
dr
(6.97)
d,B = -sin(,B12)[1-cos,B +ysin,B]/vr
dr
with
r = 1- cos,B + ,Bsin,B - sin2(,B12)
+ y[2sin,B - 2,B + ,Bcos,B + sin,Bcos,B] - y2{1- cos,Bf 12 (6.98)
In view of the singularity at 'l' = 0 the numerical integration of Eqs (6.97) can
be started by choosing slightly modified initial conditions,

v = Vo (6.99)
1+_1_~6'l'o
2r Vo

where 'l'o « 1 is an arbitrary small value and Vo == VoTo IRis the nondimensional
initial velocity of the colliding mass. The initial condition (6.99) is obtained by
taking series expansions of Eqs (6.97) and (6.98) for small ,B and then integrating
these expressions.

6.5.2 Discussion of Solution

Complete Rigid-Plastic Solution When nondimensional parameters for initial
velocity Vo and mass ratio r are specified, a numerical solution can be obtained
by integrating Equation (6.97) using the Runge-Kutta procedure, starting from
the initial condition (6.99).
Like the response of a straight cantilever to impact at the tip (Sect. 4.5), the
dynamic response of the curved cantilever also has two phases - a transient
phase with a travelling hinge and a subsequent modal phase where the cantilever
rotates about a plastic hinge at the root. Figure 6.24 shows the variation of the
hinge position ,B with time (tlt 1). Here tl denotes the time when the travelling
hinge reaches the root of the curved cantilever; i.e. the time when phase I ends.
This indicates that the smaller the mass ratio y = GI pR, the more uniform is the

0.5

0.5 1 1.5
p(rad)

Fig. 6.24. Variation of hinge location f3 with time.

6.5 Circular Arc Cantilever Subjected to In-plane Impact 229

travelling speed of the plastic hinge along the curved cantilever. For phase II (the
modal phase) the rotation angle about the root hinge 90 , can be easily calculated
from the remaining kinetic energy at the end of phase I.
To illustrate the admissibility of this rigid-plastic solution, we should examine
whether the yield condition is satisfied in the entire cantilever. First an expression
for the axial force at the hinge, N H , can be deduced according to the acceleration
field (6.85); then the bending moment at the root, MB , is found to be

;no == Mo = 1 + NHR (1 - sinf3)

Mp Mp

1 + 1- sinf3 ([2f3 - sinf3 + '1'(1- cosJ3)][f3 - sinf3 + '1'(1- cosJ3)]

4T
+ sinf3[I- cosf3 + ysinf3](f3 - sinJ3)/(1- cosf3)}

.K = XIR
t=O·~
o ~~ _______~OT.5~________~
y = 0.5
t =0.25tl • eo = 0.5
1= 11.
t = I, •

0.5

x=XIR
- 0.4 o 0.5 1.0
t=o r = 0.5
t = 0.73ms eo = 0.5
---
t =2.25 ms ."
t =3.75mse
t=7.75ms·
0.5

1.0"'--------·--------"
b

Fig. 6.25. Deformed shapes of circular cantilever in case of eo =0.5, r =0.5: (a) complete rigid-plastic
solution; (b) finite-element elastic-plastic solution.
230 More Complex Configurations

where r is defined by (6.98). For an inward radial force Fr , calculation indicates

that (1) the axial force at the travelling hinge H is always compressive; (2) by
taking f3 -t 0 in the expression above, it is known that immediately after the
impulsive loading (i.e. when r« 1 and f3« 1) N HRI M p '" - 514 and rna '" -114 ;
and (3) during the transient phase, the bending moment at the root increases
monotonically from rna '" - 0.25 at f3 = 0 to rna '" 1.0 at f3 = n/2. During the
entire dynamic response, the yield condition is not violated anywhere in the curved
cantilever; hence the single-hinge rigid-plastic solution is complete.

Deformed Shapes If the tip velocity v0 (r) and the hinge location f3( r) are
known for phase I and the hinge rotation eB is known in phase II, the deflection
of a circular arc cantilever can be calculated at every instant. Figure 6.25a shows
the deformed shapes based on a complete rigid-plastic solution for a typical
example of a light mass, r = 112 and input energy ratio eo = 112. In the figure
the termination times of phases I and II are tl and t f' respectively. For comparison,
Fig. 6.25b gives deformed shapes obtained from a finite-element calculation using
the same energy ratio; the finite element analysis, however, considered an elastic-
perfectly plastic material, see Yu et al. [1986]. Note that in the first part of the
elastic-plastic response period, the segment close to the root undergoes rotation
in a reverse direction (clockwise in the figure) before it begins to rotate in an
anti-clockwise direction. Also it should be noted that the time t == 7.75 ms is the
termination of plastic deformation in the elastic-plastic cantilever, but the deformed
shape is still changing due to elastic vibration.
Like the case of a straight rigid-plastic cantilever, the small deflection solution
for a curved cantilever also gives a final inclination at the tip A that is always
equal to the initial kinetic energy divided by the fully plastic bending moment of
the cantilever,
leAfl = eo == KolMp (6.100)

Energy Dissipation The part of the initial impact energy that is dissipated in
phase I is independent of the initial velocity vo; it is a function of only the mass
ratio r = GlpR. As shown in Fig. 6.26, the value DIIKo decreases with increasing
mass ratio r .
For very large mass ratio, an asymptotic limit can be obtained by taking r -t 00
in expression (6.98) and Eq. (6.97); thus, one finds

0.8

energy dissipated
in phase I

0.2

0.05 0.1 0.2 0.5 1 2 5 10 20

r = l1/pR
Fig. 6.26. Proportion of energy dissipated in phases I and II depends on mass ratio y.
6.S Circular Arc Cantilever Subjected to In-plane Impact 231

r ~ _y2(1-cosf3)2 /2

:: ~ sin(f3/2)y(l- cosf3)l2r = -[2ysin(f3/2)t

:~ ~ -sin(f3I2)y sinf3lvr = cos(jU2)[vysin2(f312)r

It follows that
dv vsin(f3/2 )
= (6.101)
df3 2cos(f3/2 )
The differential equation has a solution
v = Vo cos(f3/2) (6.102)
which also satisfies the initial condition v v0 at 'l' = 0 when the hinge is initially
at the impact point f3 = O. At the end of phase I, the hinge has moved to f3 = 1C 12,
so (6.102) gives a tip speed v = v0 cos( 1C I 4) = vol.fi . The part ofthe impact energy
dissipated in phase I is found to be
(6.103)
This is illustrated by Fig. 6.26, where the curve approaches 112 as mass ratio
y ~ 00. Expression (6.103) indicates that for a curved cantilever of circular plan
form subjected to transverse impact at the tip, the travelling plastic hinge dissipates
at least half the entire energy, even if the colliding mass is very heavy. This
conclusion is notably different from that for a straight rigid-plastic cantilever; in
that case DI I Ko ~ 0 if r ~ 00. In other words, for a straight cantilever all the
impact energy is dissipated in the modal phase when the mass ratio is very large,
see Eq. (4.141) and Fig. 4.29.

6.5.3 Modal Approximation

A modal approximation consists of a kinematically admissible velocity field that
is a separable function of spatial and temporal independent variables (see Chap.
2). For a curved cantilever, a modal velocity field has the form
W(¢,t) = V *(t)4> *(¢) (6.104)
where V *(t) is the tip speed of a modal approximation, and ~*(¢) =
<1>:(¢)·i + <1>;(¢)·k is the shape vector, with i and k being unit vectors shown in
Fig. 6.22b. In the present problem, the principal mode for the curved cantilever is
a rigid-body rotation about the root B; thus at the tip where ¢ = 0, the mode
shape has components <1>;(0)=-1 and <1>;(0):=1. Hence, the components of the
shape function can be written as
(6.105)
The initial value of the tip velocity V * (t) in the modal approximation can be
determined by the min Lio technique (refer to Sect. 2.5). Thus one finds
V* (0) == YeO) (6.106)
2(1 + 0.570Sly)
where y = GI pR is the mass ratio. Accordingly, the initial kinetic energy of the
modal field is
232 More Complex Configurations

(6.107)
The final rotation angle at the root B is then

le* 1- ~
Bf M
-
- 2(1 + 0.57081
p
eo
- r)
(6.108)

from which the final displacement at the tip can be easily found.
For the limiting case r ~ 00, expressions (6.106)-(6.108) give
v* (0) ~ Vo 12, K~ ~ Ko /2, 10=f I ~ eo /2 (6.109)
Note that the modal approximation corresponds to phase II in the complete rigid-
plastic solution; hence (6.109) is consistent with (6.103), and both expressions
indicate that phase I and phase II each dissipate one half of the initial kinetic
energy when the mass ratio is very large.
The fact that only half of the impact energy is dissipated by flexure can be
explained as follows. The initial velocity of the colliding mass at point A can be
resolved into two components; these are perpendicular and parallel to the line
BA, respectively. Only the component that is perpendicular to BA contributes to
the bending moment at the root where the modal hinge is located. Therefore, only
half of the initial kinetic energy is dissipated by the rotation at this hinge in
either phase II or the modal approximation. The remainder of the impact energy
is dissipated at impact by the reaction impulse at the hinge.

6.6 Circular Arc Cantilever Subjected to Out-of-

Plane Step Force
In Sects 6.4 and 6.5, a circular curved cantilever was dynamically loaded in the
plane of initial curvature, so yielding was dominated by the bending moment. In
the problem considered below the curved cantilever is subjected to a step force
suddenly applied in a direction normal to the plane of the bend, so the loading is
a combination of bending moment and torque.

Fig. 6.27. Cantilever of circular plan form subjected to step force F acting normal to the plane; H denotes the
position of a generalized plastic hinge.
6.6 Circular Arc Cantilever Subjected to Out-of-Plane Step Force 233

o
Fig. 6.28. Free-body diagram of small segment of curved cantilever, showing the positive sense of external
force. shear force. bending moment and torque.

6.6.1 Equations of Motion

Problem and Assumptions Consider a uniform curved cantilever of constant
radius R. that subtends a circular arc of 1800 ,. as shown in Fig. 6.27. End B is
built-in while the other end A is free. Suppose that an out-of-plane step force of
magnitude F is applied at the tip A of the curved cantilever. The direction of the
force is normal to the plane of the cantilever, (Fig. 6.27). An analysis of this
problem with the assumptions stated in the beginning of Sect. 6.4 was originally
given by Yu, Hua and Johnson [1985a].
Previous analysis of a straight (or curved) cantilever subjected to a step force
at the tip (see Sects 4.1 and 6.4, respectively) has indicated that a steady pattern
of deformation develops in response to this suddenly applied steady force; this
pattern is a dynamic mode for the structure subjected to this force. For a steady
force, the stationary pattern appears because the accelerations are independent of
the velocities. In the present problem, therefoTl~, it can be expected that the step
force will create a stationary generalized plastic hinge at a cross-section H; this
hinge can take any position along the circular arc. The aim of the present analysis
is to find the modal solution for a curved cantilever. The mode shape (i.e. the
location of the generalized plastic hinge) is a function of the force magnitude F.

Static Equilibrium Let an angular coordinate ifJ measured from the tip be used
to locate any section; then static equilibrium of a differential element along the
length of a curved cantilever such as shown in Fig. 6.28 leads to
dQ =-gR (6.110)
difJ
dT =-M (6.111)
difJ
dM
--T=-QR (6.112)
dlfJ
234 More Complex Configurations

where g (tfJ), Q (tfJ), M (tfJ) and T(tfJ) denote the distributed external force, shear
force, bending moment and torque, respectively. The positive sense of forces and
couples are indicated in Fig. 6.28. A combination of Eqs (6.110)-(6.112) results
in
d 2M
-2-+ M =gR 2 (6.113)
dtfJ
Equations (6.111)-(6.113) can be recast into nondimensional form
di I _
- --m (6.111')
dtfJ TJ
run - TJt- = - q
ctq; (6.112')

d2 m _ gR 2
--+m = (6.113')
dtfJ2
where
M T QR _ Tp
m- Mp
, t -,
Tp
q ;;: - - ,
Mp
TJ = - -
Mp
(6.114)

Parameter TJ depends solely on the shape of the cross-section. For some typical
cross-sections the values of TJ are listed in Table 6.2. The table indicates that
11 < 1 holds for most cross-sections, but TJ ~ 1 is possible if the cross-section is
very thin and wide. For example, TJ ~ 1 for rectangular cross-sections if bl h ~ 2.488
with a = -J3; also TJ ~ 1 for thin-walled rectangular box sections if bl h ~ 3.232 with
a= -J3.Here, the ratio of uniaxial yield stress Y to yield stress in pure shear 't' y
is a;;: Y I 't'y - this ratio depends on the yield criterion. For example, a = 2 for
the Tresca criterion and a = -J3
for the von Mises criterion.

Table 6.2 Ratio of fully plastic torque to fully plastic bending moment '1

Cross-section Mp Tp '1 '" Tp IMp 1) (with a~J3)

~Yb2(3h-b) ~(3-~) 2b (3 b)

Ii
6a 3ai1 h 3{3h --;;
.!...Ybh 2 (b <; h) (b <; h) (b <; h)
4
~ ~Yh2(3b-h)
6a
2-(3-!:)
3a b 3~ (3-~)

11
(b~ h) (b ~h) (b~ h)

m}
~YbJ 4
.!...Yb J 0.770
4 3a 3a

I 4 b 4 b
"2Y(2b + h)he ~Ybhe
~ a -;; (2b+h) .J3 (2b+h)

~ 3
~YaJ 21f Ya J
3a
If
2a
0.907

{} 4Ya 2 e 21f Ya 2 e
3a
If
Til 0.907
6.6 Circular Arc Cantilever Subjected to Out-of-Plane Step Force 235

Yield Function and Flow Rule In Sect. 1.5, it was shown that the interactive
yield function for bending moment and torque for beams of circular or rectangular
cross-section can be approximated by
'P(iiz,t) == iiz2 + -;Z -1 (6.115)

where 'P(iiz, t) = 0 represents a circular yield locus in the plane of generalized

stresses (iiz, t), see Fig. 1.11.
Suppose a generalized plastic hinge forms at a cross-section H. The location
of the hinge subtends an angle f3 from the tip. At the hinge the bending moment,
torque and yield function are as follows:
- _ MH - _ TH
mH = - , tH = -, (6.116)
Mp Tp
Associated with the yield criterion 'P(iiz,t) = 0 that is satisfied at the hinge, the
flow rule gives
(6.117)
where J1 == tH / iizH , while em
and 8r denote the rotation rates in the directions of
the unit vectors er and el/l' respectively, see Fig. 6.27. Thus, arc segment AH
has angular acceleration cO about a stationary hinge where
d· . .. J1
co = dt(eme, + erel/l) = em(e r + -rye 4,) (6.118)

Let the angle subtended by arc HC be lfI = f3 - 4J. Then the acceleration of section
Cis

w = cOxHC = -R8m {sin lfl - ~ (l-COSlfl)}k (6.119)

where k = -e r x el/l is a unit vector normal to the plane of the figure and inward.
Related to this acceleration field, the inertia force per unit length is given by

g(tfJ) = -Ipwl = -PR8m {sin lfF - ~ (l-COS lfl )} (6.120)

Substituting (6.120) into equation of motion (6.113') leads to

2
d iiz + iiz = _ pR 8 {sin lfI - J1 (1-- cos lfI)}
3
(6.121)
dtfJ2 Mp m 1]

Equations (6.111'), (6.112') and (6.121) can be solved subject to the following
boundary conditions at the tip A,
q =- f, iiz = 0, t = 0, at lfI = f3 (i.e. tfJ = 0) (6.122)
where f == FRJ Mp is the nondimensiona1 force.

6.6.2 Solution
Bending Moment and Torque Combining Eqs (6.122) and (6.112') gives
diiz
iiz = 0, = f, di
at iff = t~ (i.e. tfJ = 0) (6.123)
236 More Complex Configurations

The solution of the second-order differential equation (6.121) that satisfies boundary
conditions (6.123), leads to the following expression for bending moment in
3
in = C, sinlfl + C2 coslfl + pR Om{(-.u sinlfl + cos lfl ) lfI + 2.u} (6.124)
2Afp ij ij
with 3
C, = - pR Om(Atsinp+A2 cosP)- jcosp
2Afp
3
C2 =- pR Om (A, cosp - A2 sinp) + jsinp
2Afp

At = (-~ sinp+ cosp )p+ 2;

A2 = - ~ sinp+cos p +(-~ cosp-sinp Jp
Substituting (6.124) into Eq. (6.111') and integrating, an expression for torque i
is obtained that satisfies boundary condition (6.122),
-
t = -111 {CtCOSlfl-C2Stnlfl+j
.

- pR3 Om[-.u (sin lfI-lfIcoslfl) + coslfl + lfIsinlfl + 2.u lfI- A2 - C3 ]} (6.125)

2Afp ij ij

where
C3 = _.u (sinp - pcosP) + cosp + psinp + 2.u p
ij ij

Conditions at Plastic Hinge For mode identification, only a deformation mode

with stationary hinges ne.eds to be considered if the applied force is steady. In the
analysis concerned with differentiability of the yield function (see Table 6.1), we
showed that at a stationary hinge combined bending-torsion gives an interaction
yield function If'E C 2 • Hence, at the plastic hinge H, conditions (6.1) apply; that
is at tP = p, 2
If' = 0, dlf' = 0 dlf'<O (6.126)
2 - ,
diP ' diP
Since
dlf'
-
d
= -(m +t -1)
-2 -2 _din
2m-+ 2t-
-dt
diP diP diP diP
by using (6.111'), the second equation in (6.126) leads to

dlf'\
di _(din - 1 J\
~=fJ = 2m diP - t11 ~=fJ = 0 (6.127)

This implies that a stationary plastic hinge at an interior section H must be one of
two possible types. Either
Type I: section H is under pure torsion so in = 0 at tP = p; or
Type II: section H is subjected to combined bending and torsion, where
6.6 Circular Arc Cantilever Subjected to Out-of-Plane Step Force 237

( elm _1)[
di-t-ry ¢={3 =0
(6.128)

Analysis of Type I Hinge For a hinge of type I, there is pure torsion at H, so

mH = 0 and l.ul ~ 00, while iiI = .uiim fry still holds. By applying conditions (6.123)
and mH = 0, Eq. (6.124) gives

(6.129)
with

2 sin {3
Cs = ----'-......,...
(l-cos{3/
After substituting (6.129) into equation (6.111'), the torque t can be found by
integration and use of the boundary condition t = 0 at lfI = {3,

t = - ~ {C4 coslfl + Cs ( lfI - %sin lfl + ~ cos lfl) + C6 } (6.130)

with
2(1- cos{3 - {3sin{3) + sin 2 {3
C6 =
(1- cos{3)2
The plastic hinge conditions PH:: m~ + t; -
1 = 0 and mH = 0 require tH = ± 1 .
However, tH = + 1 leads to a negative j according to (6.130). Hence for j > 0,
only tH = -1 is possible, and this leads to a relationship between the step force j
and the hinge position {3
j=ryC* (6.131)
with
C* = (1- cos {3)2
4 - 4cos{3 - 3{3sin{3 + sin 2 {3
This expression gives the location of a pure torsion hinge of type I if force j is

r
held steady. Now we examine conditions required for this type of hinge to develop.
Since

~; = {: + 2m ~~ - 2: : -~ :
for a type I hinge, the third condition in (6.126) and tH = -1 result in

~;I¢={3 = 2~H(: -~tH) = 2~:H( ~H +~) ~ 0 (6.132)

It is known from (6.129) that dm H fdl/J=-jC4 <0 sincej> 0 and C4 >0 for
hinge positions in the range 0° ~ {3 ~ 1800 • Hence, (6.132) requires
jC4~1/ry (6.133)
or, by using (6.131),
C C· = 2 - 2cos{3 - {3sin{3 ~; 1 (6.134)
4 4 - 4cos{3 - 3{3sin{3 + sin 2 {3 --:;r
238 More Complex Configurations

Satisfying inequality (6.134) establishes a necessary condition for forming a

generalized plastic hinge of type I at a point H which subtends angle 13. In order
to judge whether this hinge is unique, however, it is also necessary to examine
whether or not the generalized yield condition is satisfied throughout the
undeformed portion of the curved cantilever; i.e. in arc segment HB, where B is
at the root of the cantilever. Uniqueness requires

(6.135)

for an arbitrary section C in arc segment HB. It can be shown (see Yu, Hua and
Johnson [1985a]) that Eq. (6.135) results in an inequality

C4 C• -- 2 - 2 cos 13 - 13 sin 13
<
2(1- cos Yc ) =
-
r (6.136)
4-4cosf3-3f3sinf3+ sinz 13 (1-cosyc)z +T)zsinZyc
where Yc is the angle subtended by the arc HC. When Yc ~ 0, r ~ l/1]z, so
that (6.136) coincides exactly with Eq. (6.134) when point C approaches hinge H.
Consequently, after some calculation the following conclusions can be drawn.
1. If 1] < 1, then r min = 1 at Yc = 180°, so that C 4 C* ~ r min requires
13 ~ 13* == 113°, where 13* = 113° is a root of the equation

1- cos 13 - 13 sin 13 + ~sinz 13 = 0 (6.137)

2
Hence, the hinge is unique if it is located at 13 ~ 113 0 •
2. If 1] ~ 1, then r min = 1/1]z at Yc -? 0, so uniqueness of the hinge H can be
examined by applying inequality (6.134).

Analysis of Type II Hinge For a stationary type II hinge, combining expressions

(6.124) and (6.125) with condition (6.128) leads to

pR 3 jj =L

r[
(6.138)
2Mp m

II(
where

Go = [(1 - :z )GZ - :z (1- Gd - 1 - :z )cos 13 + 1]\

G[ = A2 +C3 = 2(-~(Sinf3-f3)+cosf3)
Gz = A[ sin 13 + A2 cos 13 = - 11 (13 - 2sinf3 + sin 13 cos 13) + cos z 13
1]
In these expressions the value of 11 = tH / mH is the negative root of a quadratic
equation,
BZI12+B[I1+Bo = 0 (6.139)
with
Bo = (1 - cos f3)z

- (1-~)
T)
1]sin 2 f3cos 2 13 + ~(f3
1]
- 2sin 13 + sin 13 cos 13) - 3..(2 sin 13 - f3)cos 13
1]
6.6 Circular Arc Cantilever Subjected to Out-of-Plane Step Force 239

+(1+ 1]\ )1]Sin,B-[(l- ;2 }ost3+ ;2]1](,B-cOS,BSin,B)

B2 (1- 1]12 },B - 2sin,B + sin,8cos,8)sin,8 + 1]22 (sin,8 - ,8)sin,8

+[(1- 1]\ }os,8+ ;2 j(2-2CosJ3-sin 2 ,8)

i;
The yield condition at a plastic hinge 'PH == lii~ + -1 = 0 gives a relationship
for the hinge location ,8 as a function of force f at a type II hinge as

f = q{( ~o (G2-G +1)+ 1- ,o,p

1 J q2( ~: lTI2
+ Hinp (6.140)

with
G3 = - J.l (2 - 2cos,8 - sin2,8) +,8 - sinl3cos,8
1]
Also, at a type II hinge, the third condition in (6.126) requires

~
-d4)2-1"'=J3 =
2 - H (d mH
2
1 -) 2 - 2(1
m dl/>2 + 1]2 m H = mH ry'2 - 1) :5:
0 (6.141)

where Eq. (6.121) has been used together with I/> =,8. Inequality (6.141) holds if
and only if 1] ~ 1. This implies that if 1] ~ 1, the yield function 'P has a local
maximum at the type II hinge H, while in case of 1] < 1 , the yield function 'P has
a local minimum at the type II hinge H. An examination of the yield function at
an arbitrary section in arc segment HB indeed leads to a similar conclusion, see
Yu, Hua and Johnson [1985a] for details. Therefore, a mechanism with a single
type II hinge provides a complete solution if 1] ~ 1 . A single hinge cannot provide

15

'~I ~~ltypel
r--~
o

Fig. 6.29. Relationship between nondimensional step force f and hinge location f3 for cross-sections with
11 '" Tp / M p = 0.5. 1.0 and 2.0.
240 More Complex Configurations

a complete solution if 11 < I, since the yield criterion is violated at some other
sections. Table 6.2 shows that the ratio of fully plastic torque and bending moment
satisfies 11 ~ I for only unusually elongated cross-sections; i.e. for combined
bending and torsion the single hinge solution is exceptional.

Numerical Examples Calculated values relating the magnitude of the

nondimensional step force and the location of the generalized plastic hinge (i.e.
the f-{3 curves) for 11 == T/ Mp = 0.5, 1.0 and 2.0 are shown in Fig. 6.29. These
results lead to the following observations:
I. A cantilever with circular plan form and a cross-section with 11 < I can have a
plastic hinge of type II if f is very large; iff is somewhat smaller it can have a
plastic hinge of type I. Neither provides a complete solution however, since
the yield criterion is violated somewhere in the curved cantilever. Only when f
is small and the hinge is of type I located at {3 ~ 113°, does the single hinge
mechanism provide a complete solution.
2. For a cantilever having circular plan form and a cross-section with 11 ~ I, the
single-hinge mechanism with a hinge located at any {3 always leads to a complete
solution; this hinge is type I if f is small and type II if f is large.

6.6.3 Discussion

I. The deformation mechanism with a single stationary generalized plastic hinge

is kinematically admissible for the out-of-plane step force on a curved cantilever
(Fig. 6.27), whether or not the yield criterion is satisfied in the entire cantilever.
The above analysis indicates that the yield function is a maximum or extreme
at the hinge if the yield condition is satisfied in the entire cantilever. In this
problem, the single-hinge mechanism leads to a complete solution; in some
other problems, however, the single hinge solution is not complete since the

!]
30
A'

HI L
20 ~~
Q. A 8 B'
~
"-
.....
"-

10

11 =1.5

0 0.5
IIIL or PRIL

Fig.6.30. Comparison of hinge locations in straight ( - - ) and curved ( - ) cantilevers subjected to step force
F. Hinge position is given by arc length A = {3R measured from tip.
6.7 Stepped or Bent Cantilever Subjected to Step Force 241

yield criterion is violated somewhere between the hinge and the root. For the
latter problems, a complete solution requires mechanisms with more than one
hinge. Nevertheless, the single-hinge mechanism is a kinematically admissible
field so it gives an upper bound for the final displacement obtained from any
pulse load. .
2. To compare the present results for an out-of-plane step force on a cantilever
of circular plan form with those for a step force on a straight cantilever (Sect.
4.1), let both cantilevers have the same length and cross-sectional properties.
The relationship between force magnitude and plastic hinge location for these
two cantilevers is shown in Fig. 6.30, where factor 1] is chosen to be 1.5. It is
seen that when the plastic hinge is near the tip of these cantilevers, the hinge
locations are almost the same. The effect of beam curvature on hinge location
A = f3R / L increases as the applied force f decreases since the hinge is further
away from the tip.
3. Both Hodge [1959] and Boulton and Boonsukha [1959] observed that the yield
function (6.115) is a lower bound for the interaction yield curve in the Case of
combined bending and torsion. Boulton andl Boonsukha [1959] further pointed
out that the values of (m 2 +i2)112 for the actual yield surface exceed unity by
less than 15% for a square solid or hollow cross-section. This is unlikely to
affect the previous discussion about whether or not a particular solution for
the present problem violates the yield criterion.

6.7 Stepped or Bent Cantilever Subjected to Step

Force

6.7.1 General Considerations

If a cantilever with circular plan form is subjected to a step force in the direction
normal to the plane of its centroid, as shown in Sect. 6.6, a single-hinge deformation
mechanism provides a complete solution only if the parameter 1] == Tp / M p as
well as the position of the hinge along the curved member satisfy limiting
conditions. After examining the dynamic solution for a sharply bent cantilever
subjected to a step force normal to the plane of its centroid, Hua, Yu and Johnson
[1985] reached a conclusion similar to that for a circular arc cantilever; that- is, a
single-hinge deformation mechanism provides a complete solution that satisfies
the yield criterion within the entire bent cantilever only in special cases. Otherwise,
a single-hinge deformation mechanism results in a solution that violates yield in
some sections of a bent cantilever. Here again, completeness of the solution for a
single-hinge mechanism depends on the paramt~ter 1] == I;, / Mp and the magnitude
of the step force. Hence it is worthwhile to examine other possible deformation
mechanisms when the single-hinge mechanism fails.
Before developing a double-hinge deformation mechanism for either stepped
or sharply bent cantilevers subjected to a step force, we first consider a straight
stepped cantilever as an example. This will illustrate why in some cases, a single-
hinge deformation mechanism cannot lead to a complete solution. Figure 6.31 a
shows a straight stepped cantilever subjected Ito a transverse step force F at its
242 More Complex Configurations
Mp l>M p 2 Mpl <M p2
FI FI
, Mpl
A-----H~ll--C----~---B~~
t1p2 ~
:~_M~P~1---1C--~~-2--M~P2---B~~
a c

Mp2

L X o L X
b d

Fig. 6.31. (a) Stepped cantilever with M pJ > M p2 subjected to a step force F at the tip; (b) bending moment
diagram (solid curve) with hinge HI located in segment AC; dashed lines denote the fully plastic bending
moment; (c) stepped cantilever with M pJ < M p2 subjected to a step force F at the tip; (d) bending moment
diagram (solid curve) with hinge H2 located in segment CB; dashed lines denote the fully plastic bending
moment.

tip. Let M pi and M p2 be the fully plastic bending moment in segment AC and
CB, respectively, of cantilever AB. If a single plastic hinge forms to the left of
the discontinuous section C and M pi > M p2' then the bending moment distribution
shown by the solid curve in Fig. 6.31 b must violate the yield criterion (shown by
the dashed lines) in segment CB. On the other hand, if a straight stepped cantilever
satisfies Mpl < Mp2 as shown in Fig. 6.31c, the yield criterion will be violated
in some portion of segment AC when a single plastic hinge H2 forms at any
section within segment CB, as indicated by Fig. 6.31d.
To overcome these difficulties Hua, Yu and Reid [1988] proposed a double-
hinge deformation mechanism, in which two plastic hinges occur simultaneously
in a stepped or bent cantilever. The following is a brief summary of their study
on stepped cantilevers; then an analogy is established between the behavior of
stepped cantilevers and bent cantilevers.
For these analyses, some of the assumptions employed in previous sections of
this chapter are employed. Thus, rigid-perfectly plastic and time-independent
materials are considered. The effect of shear force on yielding is neglected and
the deflections are small in comparison with the length of the cantilever. Also,
each segment of the stepped cantilever is considered to have uniform cross-section
and density.
The dynamic response of a stepped or bent cantilever subjected to a step force
at the tip, can be obtained from a five step analysis:
1. suppose that either one or two hinges form somewhere in the cantilever (for
single or double-hinge mechanisms, respectively);
2. obtain two equations of motion for each rigid segment; these equations relate
the external force to shear force, bending moment and acceleration distributions
in each segment;
3. equations of motion obtained in (2) are nondimensionalized;
4. obtain an expression relating f (nondimensional step force) and A (nondimen-
sional coordinate of hinge position) by eliminating terms containing angular
acceleration in the equations;
6.7 Stepped or Bent Cantilever Subjected to Step Force 243

5. the following restrictions are checked: (a) yield criterion at each segment is
not violated; (b) tip acceleration is in the same direction as that of step force;
and (c) hinge location is inside the segment concerned.
Some of these restrictions limit the range of force f for a particular single or
double hinge mechanism, or they impose a restriction on the geometry of the
member. Hence, combining steps (4) and (5) above leads to a complete solution
for some range of force f.

6.7.2 Stepped Cantilever

Case A: Stepped Cantilever with Mpl > Mp2 Consider a stepped cantilever
with Mpl > M p2 ' see Fig. 6.32a, and let Lt and ~ be the length of segments
AC and CB, respectively, while PI and P2 are the mass per unit length in segments
AC and CB, respectively. By defining
(6.142)

it can be shown that plastic collapse initiates when the bending moment at the
root B is equal to the yield moment. This gives a static collapse load fAo for the
cantilever,
(6.143)
where subscript A pertains to Case A.
If f > fAo, one or two hinges form somewhl~re in the cantilever. Following the
method described in Section 6.7.1, various dynamic deformation mechanisms can
be examined and finally five mechanisms are identified. These are summarized in
Table 6.3.
In Table 6.3, HI denotes a hinge located in segment AC, and H2 a hinge
located in segment CB. The values off, at which the dynamic deformation transfers
from one mechanism to another are given by (6.143) and
3T1M(TlpTlZ + 2T1L + I)
(6.144)
TlpTlL3 + 3T1pTlL2 + 3TIL + 1

(6.145)

3(2 - }.)
(6.146)
}.(3 - 2A)

Table 6.3 Deformation mechanisms in stepped cantilever

where 1jM=M p2 IM pl <1
Mechanism Plastic Hinge Range off

No hinge o 5,f< fAo

2 Single-hinge B fAo 5,f5,fAI
3 Single-hinge H2 fA! <f<fA2
4 Single-hinge C fA2 5, f 5, fA3
5 Double-hinge HI-C fA3 <f
244 More Complex Configurations

Fh
B~
Mpl Mp 2
I

AI H~, .CI.
: :\
L2
L
x
a
-F
c

b d
Fig. 6.32. (a) Case A: stepped cantilever with M pi > M p2 subjected to step force F at tip; a double-hinge
deformation mechanism HI -C (refer to Mechanism 5 in Table 6.3) is shown; (b) free-body diagram for the H I -
C double-hinge mechanism; (c) shear force distribution for the HI -C double-hinge mechanism; (d) bending
moment distribution for the HI -C double-hinge mechanism; the dashed lines denote the fully plastic bending
moment.

where A is the nondimensional coordinate of hinge HI that can be obtained as a

positive root of the cubic equation,
2(l-A)3_(I-1JM)(3-lA)A2 = 0 (6.147)
It is seen from Table 6.3 that a double-hinge mechanism takes place when the
step force is sufficiently large. In this case, plastic hinges form at sections HI
and C that are shown in Fig. 6.32a. Figure 6.32b is the free-body diagram for a
.. ..
two hinge mechanism, with 8 c and 81 denoting the angular accelerations of segment
HI C and of A HI relative to HI C respectively. Using these free-body diagrams
and equations of motion, the distributions of shear force and bending moment
along the cantilever are as shown in Fig. 6.32c and d, respectively. The dashed
lines in Fig. 6.32d represent the fully plastic bending moment while the solid
curve represents the dynamic bending moment distribution, so this figure illustrates
that the solution obtained from the double-hinge mechanism satisfies the yield
condition throughout the member.
The question of whether the yield function takes an extreme at a plastic hinge
was investigated in Sect. 6.1; this depends on differentiability of the yield function
at that hinge. In the present case, the only generalized stress is the bending moment
so the yield function is
P(X) = IM(X)I-Mp(X) (6.148)
where
O~X~~-o
(6.149)
6.7 Stepped or Bent Cantilever Subjected to Step Foree 245

I
I
,,
I

I
,,
,
I
I I --single hinge
,,,
I
\
- ---double hinge
I
\
, ,, ~L = 1.2
:
\

TiM = 0.8
\

1)/> = 0.8

~----
A C

Fig.6.33. Relationship between the nondimensional forcel and hinge location Afor case A, when I) L = 1.2 and
I)M = I)p = 0.8.

This yield function 'P(X) has differentiability that is sufficient to represent a

travelling hinge at all sections except the stepped cross-section C where 'P(X) is
discontinuous. Consequently, if a plastic hinge forms at cross-section C, then the
bending moment M(X) need not be an extreme at C. This implies that if there is
a hinge at C (A = Lt) the shear force at that hinge can be nonzero; if a plastic
hinge forms at any cross-section other than C, then the shear force must vanish at
the hinge position, dM(A)/ dX = Q(A) = 0 at A*, L l . These remarks are illustrated
by Fig, 6,32c and d, where QHI = 0 but Qc O. *'
Based on the analysis given by Hua, Yu and Reid [1988], the dependence of
hinge position(s) on the nondimensional force magnitude f is calculated; a typical
example is given in Fig. 6.33 for Th = 1.2 and TIM = TIp = 0.8. This clearly
demonstrates that with increasing force magnitude, Mechanisms 1 to 5 defined in
Table 6.3 appear one after the other.
For Mechanism 5, i.e. the HI-C double-hinge mechanism, the total energy is
partly dissipated at hinge C and partly at hinge HI' For TIM = TIp = 0.8, Fig. 6.34
shows how the part of the energy dissipated at hinge C varies with force magnitude
f The vertical distance below the solid curve represents the part of the energy
fA3
100·/.,---,.----....- - - - - - - - ,
ryM = O.S
TJp = 0.8
80

rttte of energy
dissipation at
60 hinge H,

40

20
rate of energy
dissipation at
hingeC

4 7

Fig. 6.34. The part of the total energy dissipation rate that occurs at hinge C depends on the force magnitude
I for theHI -C double-hinge mechanism, when I)M= 1/ p =O.8 (,~ase A).
246 More Complex Configurations

F~ H

B~
Hp1 C Hp2 Hz
A I

~ ~
j
L1

11
".I b
x

Fig. 6.35. (a) Case B: stepped cantilever with M pI < M p2 subjected to step-force F at tip. A double-hinge
deformation mechanism C- H 2 is considered. (b) Bending moment distribution for the C- H 2 double-hinge
mechanism; the dashed lines denote the fully plastic bending moment.

Table 6.4 Deformation mechanisms in stepped cantilever

where TIM = Mp2IMpl >1
Mechanism Plastic hinge Range of f
No hinge o$f <fBO
2 Single-hinge B fBO $f$fBI
3 Single-hinge H2 fBI <f<fB2
4 Double-hinge C - H2 fB2 $f <fB3
5 Double-hinge C-B fB3 $f <fB4
6 Single-hinge C fB4 $ f $ f BS
7 Single-hinge HI fBS <f

dissipation rate that occurs at hinge C; while the vertical distance above the
curve represents the remaining part dissipated at hinge HI located somewhere
between A and C. This figure indicates that the energy dissipated at the stepped
cross-section C rapidly decreases at larger forces.

Case B: Stepped Cantilever with Mpl < Mp2 If the cantilever is stepped with
Mpl < Mp2 (i.e. TIM> 1) as shown in Fig. 6.35a, there are seven dynamic
deformation mechanisms for steady forces; these are summarized in Table 6.4.
Figure 6.35b depicts the bending moment distribution for the C- H2 double-hinge
mechanism (i.e. Mechanism 5 in Table 6.4, where H2 denotes a hinge position
between C and B). This figure indicates that the yield criterion is not violated
anywhere in the cantilever, and that the yield function is an extreme at hinge H2
but not at hinge C.
The mechanism transition values of forcef, i.e. fBO,fBl'fB2,fB3,fB4 andfBs' depend
on ratios TIL> TIM' and TIp, see Hua et. al. [1988] for details. In fact for some
combinations of TIL' 11M' and TIp some of the mechanisms are not present. Figure
6.36 gives a typical example of Case B where TIL = 2.0 and TIM = TIp = 1.2. The
distribution of energy dissipation rates at hinges C and H2 for the C- H2 double-
hinge mechanism (i.e. Mechanism 4 in Table 6.4) are shown in Fig. 6.37.

6.7.3 Bent Cantilever

Case C: Bent Cantilever with 1'/ == J;, / Mp < 1 Consider a cantilever which has
been bent through an angle f3 with 0°::;; f3 ::;; 90° . A step force F acts at the tip in a
6.7 Stepped or Bent Cantilever SUbjected to Step Foree 247

5 - - sing l~ hing~

---- cloubl~ hing~

4

I
I

--1
I

lj
I
I
I

A c t1 e B

Fig.6.36. Relationship between the nondimensional force f and hinge location Afor case B, when 1) L = 2.0 and
1)M = l1p = 1.2.

'BS
100·/.

80

60
rat~of ~n~rgy
dissipation at
40 hing~ C

TIL = co
20 TIM = 1.2
TIp = 1. 2

'B2 20 25 30 f

Fig. 6.37. The part of the total energy dissipation rate that occurs at hinge C depends on the force magnitude
ffor the C- Hz double-hinge mechanism, when 11M = TIp = 1.2 (case B).

Fig. 6.38. Bent cantilever subjected to an out-of-plane step force F at tip A.

direction normal to the plane of the member, see Fig. 6.38. In addition to the
assumptions made in Sect. 6.7.1, it is further assumed that the entire cantilever
has a uniform cross-section and density; that is, M pi = M p2 == M p' Tpl = Tp2 == Tp
and PI = P2 == p, where subscripts 1 and 2 pertain to segments AC and CB,
respectively; Mp and Tp denote the fully plastic bending moment and fully plastic
248 More Complex Configurations

torque, respectively. Here we consider rpiii Tp IMp < 1; this case is representative
of almost all practical cross-sections (see Table 6.2).
Within the range of small deformation, segment AC is subjected to bending
while segment CB is subjected to combined bending and torsion. Hence, similar
to Sect. 6.6, the yield function given by (6.115) can be written as 'l'(m,t) where
m and i are the nondimensional bending moment and torque at a cross-section.
Table 6.1 indicates that the differentiable function 'l'(m,t) must take an extreme
value at a generalized plastic hinge located at any cross-section other than the
bend C where torque is discontinuous.
By comparing the yield function for a stepped cantilever with that for a bent
cantilever, an analogy can be established. If Tp<Mp, i.e. 1J==TpIMp<l, the
beam is 'stronger' when it is subjected to merely bending rather than either torsion
or combined bending and torsion. Since segment AC is subjected only to bending,
it seems to be 'stronger' than segment CB, although the cross-section and density
are identical for both segments. Thus, Case C is analogous to Case A that was
discussed before. In the same way, Case D, which pertains to a bent cantilever
with 1J == Tp IMp > 1, is analogous to Case B, below.
Similar to the analysis of Case A, Hua, Yu and Reid [1988] found five dynamic
deformation mechanisms for bent cantilevers with 1J == Tp IMp < 1. These
mechanisms can be summarized in the same way as those in Table 6.3, provided
that the transition values of force f are changed from fAo' fAI' fA2' fA3 to
fco, fCI' fC2' fC3accordingly. The values of fco' fCI' fC2' fC3 depend on both
cross-sectional property 1J == Tp IMp and the bend angle f3; they can be calculated
for any bend angle f3,
fco 1J{sin 2 f3 + 1J2(cosf3 + 1JJ 2 1l2 r (6.150)

fC2 31Jb' 1Jb == {(-\- -1)sin 2 f3 + 1}-112 (6.151)

3(2-),,) 1J
(6.152)
),,(3 - 2A)

,,
,
,
I
3V"C B~
B
3~C l
8 I
, A A
- - single hinge - - single hinge
,
I
I

6 I ----double hinge 6 ----- double hinge

I
I
I
\ 7) = 0.5
\ 1) =2.0
tC3 l1L =1.5 4
ryL = 1.5
'os
tC2
t02
tCI
fco
A Sc C B A
a b
Fig. 6.39. (a) Relationship between the nondimensional force f and hinge location A for bent cantilever in case
C ( 11 ;;;; Tp IMp < 1 ), when f3 = 30°, 11 = 0.5 and 11 L = 1.5 ; (b) Relationship between the nondimensional force
f and hinge location A for bent cantilever in case D ( 11 ;;;; Tp IMp> 1 ), when f3 = 30°, 11 = 2.0 and IlL = 1.5 .
6.7 Stepped or Bent Cantilever Subjected to Step Force 249

where It is the nondimensional coordinate of hinge HI' This hinge location can
be obtained as a root of the equation
(6.153)
The expression of ICI is rather complicated, see Hua, Yu and Johnson (1985], so
it is omitted here. Note that l1b given in (6.151) combines both the cross-sectional
property 11 '= Tp IMp and the bend angle /3 as one single parameter. A comparison
of (6.151) with (6.145) indicates that for bent cantilevers 11b plays the same role
as 11M '= M p2 1Mpi for stepped cantilevers. With the correspondence between 11M
and 11b in mind, the similarities between (6.146) and (6.152) as well as between
(6.147) and (6.153) are evident.
A typical example for the relationship between the magnitude of the step force
I and the hinge position It is shown in Fig. 6.39a for /3 = 30°, 11 = 0.5 and 11L = 1.5.
Case D: Bent Cantilever with Tf '= ~ IMp> 1 This case is similar to Case B
above. A similar analysis results in seven deformation mechanisms, which can
be summarized in the same way as those in Table 6.4, provided
lBO' IBI' IB2' I B3' I B4, IB5 are changed into 100' 101' 102' 10 3' 10 4' 105' The
latter values are functions of parameters /3, J7 and 11L> see Hua, Yu and Reid
(1988] for details. Also, for some combinations of these parameters, some
deformation mechanisms do not occur. The re:lationship between the step force
magnitude I and the hinge position It for a typical example is shown in Fig.
6.39b for /3 = 30°,11 = 2.0 and 11L = 1.5.

6.7.4 Discussion

1. The analyses in this Section show that the dynamic response of stepped or
bent cantilevers result in double-hinge mechanisms if the applied force is
sufficiently large. Generally speaking, a single-hinge mechanism fails to provide
a complete solution if the force is large because the generalized stresses violate
the yield criterion somewhere in the structure. In this case it is worthwhile to
examine a double or multi-hinge mechanism. For example, the analysis in
Sect. 6.5 shows that in some cases a single-hinge mechanism cannot lead to a
complete solution for a cantilever of circular plan form that is subjected to a
step force acting normal to the plane of the structure. This suggests a multi-
hinge mechanism or even a mechanism with a continuous plastic zone. A
solution for a particular in-plane distributed load on a curved beam gave a
continuous plastic deforming region for an interaction yield condition (Cline
and Jahsman [1967]).
2. An analogy between the response of stepped and bent cantilevers has been
established. Application of this analogy yields a rather easy extension of the
results for stepped cantilevers to those for bent ones. This analogy applies if
rotary inertia of beam elements about the centroidal axis is negligible in segment
CB.
3. According to this analogy, a bent cantilever with 11 '= Tpl Mp = 1 is directly
analogous to a straight cantilever of uniform cross-section, so its dynamic
response to step force is merely a single-hinge mechanism. The dynamic response
of a bent cantilever subjected to an impulsive loading at the tip was previously
250 More Complex Configurations

analyzed by Martin [1964]; he found that in the transient phase, a travelling

plastic hinge runs continuously from the tip towards the root. At the hinge the
deformation changes from bending to a combination of bending and torsion
when the hinge transits the bend in the cantilever.
4. In general, if a stepped or bent cantilever with 1)"* 1 is subjected to an impulsive
load applied to a heavy particle at the tip, then the force at the tip can be
regarded as F = - GV , which decreases with time. Referring to the f-A relations
obtained in this section (see Figs 6.33, 6.36, 6.39), one can expect that the
dynamic response of an impulsively loaded cantilever will undergo a series of
deformation mechanisms; for instance, in Case A, Mechanisms 5, 4, 3, 2, 1
(see Table 6.3) will appear in succession during the dynamic response, and in
some phases a travelling hinge will move along segments of the cantilever. If
a stepped or bent cantilever is subjected to general pulse-loading, this transient
phase also occurs. An analysis for this problem was given by Reid, Hua and
Yang [1990]. In these cases, a double-hinge mechanism consists of a stationary
hinge at section C or at the root B, together with a travelling hinge that moves
along segment AC or segment CB. This travelling hinge never transits section
C.
5. Extending the aforementioned studies, Reid et al. [l995a] showed that the
family of deformation modes for the bent beam problem can be completed by
introducing a triple-hinge mechanism. This consists of a hinge at the bend C, a
hinge (HI) in the first segment and a hinge (H 2 ) in the second segment. This
mechanism provides a solution to the problem where the single- and double-
hinge mechanisms become inadequate; these authors have shown that no further
mechanism (with more than three hinges) is required. These authors also
discussed cases of pulse and impact loading (Reid et al. [1995b], Wang et al.
[1995]; this showed the transient behavior of the triple-hinge mechanism and
the partition of the energy dissipated by plastic deformation in a multiple
hinge mode of deformation.

6.8 Cantilever with an Initial Crack

6.8.1 General Considerations

The previous studies of statically and dynamically loaded cantilevers assume that
any variation in section properties with axial distance is monotonic. When an
initial crack or flaw exists at some cross-section this can significantly reduce the
fully plastic bending moment at the cross-section and thereby alter the kinematics
of deformation. If the material is brittle, fracture is likely to occur at the damaged
section with little if any plastic deformation. On the other hand, in ductile materials
the crack is likely to serve as the initiation site for a stationary plastic hinge that
absorbs a significant portion of the input energy. Since the material in many
modem structures is deliberately chosen to be ductile so that defects can be tolerated
without precipitating brittle fracture, the problem of large dynamic plastic
deformation of structures with initial cracks is one of current interest.
6.8 Cantilever with an Initial Crack 251

Petroski [1984a, 1984b] studied impact and the dynamic response of cracked
cantilevers. First he investigated the effect of a crack at the root on dynamic
deformation of a cantilever subjected to impact. An initial velocity was imparted
to an attached mass at the tip. Only the modal· phase motion was considered; that
is, the cracked cantilever merely rotated about the root as a rigid-body. Thus the
final rotation angle at the root was easily found. This analysis served as a basis
to discuss stability of the crack after impact loading. Later Petroski [1984b]
considered a crack at an interior cross-section of a cantilever. The rotation angle
at the cracked section after impact was estimated by employing a mechanism
wherein stationary plastic hinges were located at the root and the cracked cross-
section throughout the entire response. Woodward and Baxter [1986] reported
their experimental study on impact bending of unnotched and notched free-free
steel beams; they commented' that the notched beam problem is an ideal case for
application of the rigid-plastic approach because a notch localizes deformation,
making it more like a hinge.
Inspired by these works and by the idea thalt more than one hinge could travel
in a double-hinge mechanism (refer to Sect. 6.7), Yang and Yu [1991] proposed a
complete analysis to an impact-loaded cantilever with an initial crack at an arbitrary
interior cross-section. Both transient and modal phases were considered in the
analysis. They obtained the partition of dissipated energy and a criterion for stability
of the crack.

6.8.2 Impact on Cantilever with an Initial Crack

Assumptions The following assumptions are adopted in analyzing the dynamic
response of a cracked cantilever:
1. The material is rigid-perfectly plastic and rate-independent.
2. The effect of shear force on yielding is neglected.
3. The deflection is small in comparison with the cantilever length L.
4. The cantilever has uniform section properties other than at an interior section
where there is a crack. The initial momentum due to impact is parallel to an
axis of symmetry for the cross-section. The (:rack spans the entire width and is
located on the tensile side of the beam during deformation.
5. The crack remains stable during the dynamic response of the cantilever to
impact; that is, coupling between the overall dynamic response and local crack
propagation is disregarded.

Fig. 6.40. Cantilever with crack at section C, subjectl~d to impact at tip by a rigid mass.
252 More Complex Configurations

H
~. B~ ~H2 B~
A A
a b

B~ ~
~c B~
~ "
A
A
c d

Fig. 6.41. (a) Deformation mechanism in the first phase with stationary hinge C and travelling hinge HI'
Possible deformation mechanisms in the second phase: (b) double stationary hinge C- H2 (case I); (c) double
stationary hinge C-B (case II); (d) single stationary hinge C (case III).

Transient Solution Suppose an initially straight cantilever of length L and mass

per unit length p is struck transversely at the tip by a particle of mass G. The
particle acquires an initial transverse velocity Vo at time t = O. Let a transverse
crack exist at a cross-section C located a distance ~ away from the tip, as
shown in Fig. 6.40. Let the fully plastic bending moment for noncracked cross-
sections be M p' while that for the cracked cross-section C be i;Mp' where t;, (0
< t; < 1) is a parameter depending on the size and shape of the crack, for example
see Sect. 6.8.3 below.
By applying the previous analysis on stepped cantilevers (Sect. 6.7.2) to the
present configuration, Yang and Yu [1991] obtained the following results.
1. The first phase of dynamic response usually consists of a stationary hinge at
the cracked cross-section C and a travelling plastic hinge HI which moves
away from the impact point at the tip, as shown in Fig. 6.41a. This double-
hinge mechanism has bending moments Mp at the travelling hinge and i;Mp
at the cracked section. This mechanism satisfies the yield criterion in the
entire cantilever and leads to a complete solution if the crack location
T/ L == Lz / ~ satisfies

T/L == Lz < 2(1 + t;) (1- A I L ) (6.154)

~ }(1- t;) I I

where t; is the crack parameter defined above and Al « ~) is the terminal

position of the plastic hinge HI when it stops moving. For very light and very
heavy colliding masses, inequality (6.154) results in

T/L < 2 (.12 - ~1- t; ), YI == G/pLI« 1 (6.155)

3~
2(1 + t;)
T]L < Yl == GI p~» 1
3~ (1 +~1-0'
respectively. In most cases of practical interest, the crack is small compared
with the dimensions of the cross-section, so t; is slightly less than one and
inequality (6.155) is satisfied throughout most of the member. If inequality
(6.155) is violated, then a third plastic hinge may be required at the root B.
6.8 Cantilever with an Initial Crack 253

Fig. 6.42. Sequence of response phases for cracked cantilevers.

2. The first phase of dynamic response ends when the travelling hinge HI halts
at X= AI somewhere between A and C. Thereafter, the subsequent motion
follows one of three possible cases, depending on the range of parameters
YI=G/P~, 1h=~/~ and the crack depth s:
Case I: if
1 S 1}2
o<-y-< 3r(1+~L)-1 (6.156)

where r;: (YI + 1/3)/(Yl + 1/2), then the second phase of the response is
characterized by two stationary plastic hinges located at C and H2 (i.e. C-
H 2 ), where H2 is in segment CB, see Fig. 6.41b.

Case II: if

(6.157)

then the second phase is characterized by two stationary plastic hinges (C-B)
located at the crack and the root, see Fig. 6.41c.

Case III: if
~< l-S (6.158)
r- S
then the second phase is characterized by a single stationary plastic hinge C
located at the cracked section, see Fig. 6.41d.
3. The sequence of response phases for a cracked cantilever can be summarized
by a flow chart as in Fig. 6.42. In this figure the letters designating each phase
denote hinge locations; the arrow above letter HI and H2 represents the direction
of travel for the hinge, and the letters without a superscribed arrow pertain to
stationary hinges.
4. The final rotation angle Be! at the cracked cross-section C, can be calculated
from an analysis of the first and second phases during which a stationary
plastic hinge forms at section C.
254 More Complex Configurations

6.8.3 Crack Stability After Impact

Crack Stability Criteria Only very ductile materials can be accurately represented
by the rigid-perfectly plastic moment-curvature relation used above, and even
these materials cannot maintain stable or nonextending cracks under extreme
conditions. The most useful analysis to determine the stability of cracks under
conditions of large plastic strain employs the concepts of the l-integral and tearing
modulus as described in fracture mechanics, e.g. in Hutchinson and Paris [1979]
or Paris et al. [1979].
For perfectly plastic behavior of the cracked beam analyzed above, the value
of the l-integral can be related to the deformation work done at the cracked
section C. Thus for a rotation angle Be at the crack location,
1 = _ iJ(sMp) B (6.159)
iJAer e

where Acr is the crack area, which depends upon the geometry of the assumed
crack. The stability of the crack is determined by a quantity known as the tearing
modulus; for a perfectly plastic material it is given by
- E iJl
T =- -
y2 ()C
(6.160)

where E is Young's modulus, Y is the yield stress of the material, and c is a

characteristic flaw size.
For a specified material, the material resistance curve for crack propagation
can be determined by experiments. Let 1m and Tm denote the values of the l-
integral and the tearing modulus, respectively, at the critical state of crack
propagation. Then the crack stability criteria can be expressed by
for stable cracks
(6.161)
or for unstable cracks

Equation (6.159) indicates that the value of the l-integral is proportional to

the rotation angle Be at the cracked section C. For the rigid-perfectly plastic
idealization, rotation angle Be increases monotonically during the dynamic response
of the cracked cantilever to impact. Hence, both land T reach their largest
values 1f and ~, respectively, when the rigid-plastic response ceases. For this
final state the crack stability criteria (6.161) can be recast as
for stable cracks
(6.161)'
for unstable cracks
with

(6.159)'

(6.160)'

Indeed, when the elastic contribution is neglected, conditions If < 1 m and

~ < Tm, guarantee a nonpropagating crack during dynamic response of the
6.8 Cantilever with an Initial Crack 255

Fig. 6.43. Typical stress corrosion circumferential crack in cross-section of pipe.

cantilever. On the other hand, 1f ~ 1m or ~ ~ Tm, implies that the crack extends
during some part of the overall response.

Stability of Circumferential Crack in Thin-Walled Pipe Referring to Petroski

[1984a], a typical stress corrosion crack in a boiling water nuclear reactor steam
supply pipe is as shown by the shaded part in Fig. 6.43. Assume that this
circumferential crack is located on the tensile side of the pipe during impact.
This is the worst case for dynamic failure. The fully plastic bending moment for
the undamaged cross-section is M p = 4Yca 2 , with a and c being the mean radius
and the wall thickness of the pipe section, respectively. For the cracked cross-
section with the geometry shown in Fig. 6.43, the fully plastic bending moment
is t;M p' with
S= c sinl/>
coslfI- (6.162)
2c
where c denotes the radial depth of the circumferential crack, I/> is the half angle
subtended by the circumferential crack, and lfI = cl/>l2c is the angle subtended by
the neutral axis. The crack area for this cross-section is
AT = 2acl/> (6.163)
With the aid of (6.162) and (6.163), the 1-integral can be evaluated for cases
where the crack extends in either the radial or the circumferential direction. Finally,
this results in

(6.164)

a(SMp) al/> .
11/>f =- aI/> J
dAcT
(Jef = Ya(sllllfl + cosl/»(Jef (6.165)

where subscripts r and I/> represent the radial and circumferential directions, respec-
tively, and subscript f denotes the final value after the dynamic response.
Correspondingly, the final values of the tearing modulus at the cracked cross-
section in the radial and circumferential directions are found as

(6.166)
256 More Complex Configurations

90
80
70
~ 60
[l' 50
C7>

~ 40
B- 30
20
10

0 0.2 0.4

Fig.6.44. Circumferential extent of crack I/J and axial location LIt L for occurrence of each of three different
deformation mechanisms.

E ()JI/Jf = -
-2 - -
EC {(
- (Jej COSljf - --=-sin<p
2C) - ()(Jej
--(sinljf+ cos<P) 2} (6.167)
Y a()<p 2Yc c ()(
Since, sin<p/<p > COS<p, Eqs (6.164) and (6.165) imply that
Jrj > JI/Jj (6.168)
On the other hand, since ()(Jcj/()( <0, (6.166) and (6.167) imply that
(6.169)
provided alP > c or that the crack has a longer dimension in the circumferential
rather than the radial direction. Since this is the case for most stress corrosion
cracks, the inequalities (6.168) and (6.169) indicate that crack extension is more
likely in the radial direction. This leads to a leak-before-break process that is
desirable from a safety point of view. In this case most of the input energy is
dissipated by crack extension in the radial direction so the pipe is unlikely to
break due to circumferential fracture.

6.8.4 Numerical Example and Discussion

In the complete analysis given by Yang and Yu [1991], various combinations of

parameters led to three cases with different sequences of response phases (see
Fig. 6.42). For the particular crack configuration shown in Fig. 6.43 and c/ c = 0.5,
the occurrence of these three cases can be depicted on a map of crack
circumferential extent <p and axial location ~ / L, as shown in Fig. 6.44. In the
figure, regions I, II and III correspond to Cases I, II and III in Sect. 6.8.2,
respectively. The solid and dashed lines describe the boundaries between these
regions for very large and very small mass ratios (Yl ~ 00 and Yl ~ 0, where
Yl =.G/pLI as defined in (6.155)) respectively. For O<Yl <00, the boundaries
between the regions are located somewhere between the corresponding solid line
and dashed line. Since in Case III all the remaining energy after the first phase
has to be absorbed by a stationary hinge at the cracked cross-section C, a crack
associated with region IlIon the map is most hazardous with respect crack
extension.
Figure 6.45a and b presents the radial and circumferential variations of final
values for the J-integral as functions of the crack angle <p, for a crack located at
6.8 Cantilever with an Initial Crack 257
0.4 r - - - - - - - - - - - , 0.2 r------:;;---;;--,,.----,
L,IL = 0.5
cle =0.5

0.1

o 90 <1>(0)
a b
Fig. 6.45. Final values of J-integral depend on the crack geometry. for L, I L =0.5 and c/ c =0.5; the final
values are (a) Jif; in radial direction, and (b) iH in circumferential direction.

the half-length of the cantilever and crack depth half way through the wall thickness.
Note that eo == Ko IMp is the nondimensional initial kinetic energy as used before,
so lr!IYae o and l,pt'Yaeo are both nondimensional. In this case, both lr! and
l,p! reach their largest magnitudes at a mass ratio Yl "'" 0.7. Figure 6.45a indicates
that the tendency towards crack propagation in the radial direction increases with
the circumferential size of the crack; while Fig. 6.45b shows that the maximum
value of l,p! appears for a circumferential extent of crack in the range
rp = 50° - 70°. This implies that even if l,p! = 1 m' any tendency towards crack
propagation in the circumferential direction is suppressed. The crack tends to
propagate in the radial direction through the wall thickness and cause leakage of
the pipe rather than breakage.

References
AI-Hassani, S.T.S., Johnson, W. and Vickers, G.W. [1973). Dynamically loaded variable thickness
cantilevers using a magnetomotive impulse. Int. J. Mech. Sci. 15, 987-992.
Boulton, N.S. and Boonsukha, B. [1959). Plastic collapse loads for circular-arc bow girders. Proc. Inst.
Civil Engrs. 13, 161-168.
Cline, G.B. and Jahsman, W.E. [1967). Response of a rigid-plastic ring to impulsive loading. ASME J.
Appl. Mech. 34, 329-336.
Hodge, P.G., Jr. [1959). Plastic Analysis of Structures. McGraw-Hill, New York.
Hua, Y.L., Yu, T.X. and Johnson, W. [1985). The plastic hinge position in a bent cantilever struck normal
to its plane by a steady jet applied at its tip. Int. J. Impact Engng. 3, 223-241.
Hua, Y.L., Yu, TX. and Reid, S.R. [1988). Double-hinge modes in the dynamic response of plastic
cantilever beams subjected to step loading. Int. J. Impact Engng. 7,401-413.
Hutchinson, J.W. and Paris, P.C. [1979). Stability analysis of J-controlled crack growth. Elastic-Plastic
Fracture (eds J. D. Landes et al.l, American Society for Testing and Materials, ASTM STP668, 37-
64.
Martin, 1.B. [1964). The plastic deformation of a bent cantilever, Part I: Theory. NSF GP 1115/15.
division of Engineering, Brown University.
Owens, R.H. and Symonds, P.S. [1955). Plastic deformations of a free ring under concentrated dynamic
loading. ASME J. Appl. Meeh. 22,523-529.
258 More Complex Configurations

Palomby, e. and Stronge, W.J. [1988]. Evolutionary modes for large deflections of dynamically loaded
rigid-plastic structures. Mech. Struct. Mach. 16(1), 53-80.
Paris, P.e., Tada, H., Zahoor, A. and Ernst, H. [1979]. The theory of instability of the tearing mode of
elastic-plastic crack growth. Elastic-Plastic Fracture (eds J. D. Landes et al.), American Society for
Testing and Materials, ASTM STP668, 5-36.
Petroski, H.J. [1984a]. Stability of a crack in a cantilever beam undergoing large plastic deformation after
impact. Int. J. Pres. Yes. and Piping 16, 285-298.
Petroski, H.J. [1984b]. The permanent deformation of a cracked cantilever struck transversely at its tip.
J. Appl. Mech. 51, 329-334.
Prager, W. [1954). Discontinuous fields of plastic stress and flow. Second U.S. National Congress of
Applied Mechanics, ASME, NY, 21-32.
Reid, S.R., Hua, Y.L. and Yang, J.L. [1990). Development of double-hinge mechanisms in a bent
cantilever subjected to an out-of-plane force pulse. Int. J. Impact Engng. 9,485-502.
Reid, S.R., Wang, B. and Yu, T.x. [1995a]. Yield mechanism of a bent cantilever beam subjected to a
suddenly applied constant out-of-plane tip force. Int. J. Impact Engng. 16,49-73.
Reid, S.R., Wang. B. and Hua, Y.L. [1995b]. Triple plastic hinge mechanism for a bent cantilever beam
subjected to an out-of-plane tip force pulse of finite duration. Int. J. Impact Engng. 16, 75-93.
Shu, D., Stronge, W.J. and Yu, T.x. [1992]. Oblique impact at tip of cantilever. Int. J. Impact Engng. 12,
37-47.
Stronge, W.J., Shu, D. and Shim, V.P.W. [1990). Dynamic modes of plastic deformation for suddenly
loaded, curved beams. Int. J. Impact Engng. 9, 1-18.
Symonds, P.S. [1968]. Plastic shear deformation in dynamic load problems. Engineering Plasticity (eds
J. Heyman and F.A. Leckie) Cambridge University Press, 647-664.
Symonds, P.S. and Mentel, T.J. [1958]. Impulsive loading of plastic beams with axial restraints. J. Mech.
Phys. Solids 6, 186-202.
Wang, B. Reid, S.R. and Yu, T.x. [1995]. Response of a right-angled bent cantilever subjected to an out-
of-plane impact and force pulse applied at its tip. Submitted to ASME J. Appl. Mech.
Woodward, R.L. and Baxter, B.J. [ 1986]. Experiments on the impact bending of continuous and notched
steel beams. Int. J. Impact Engng. 4, 57-68.
Yang, J.L. and Yu, T.X. [1991]. Complete solutions for dynamic response of a cracked rigid-plastic
cantilever to impact and the crack unstable growth criteria (in Chinese ).Acta Scientiarum N aturalium
Universitatis Pekinensis 27, 576-589.
Yu, T.x., Hua, Y.L. and Johnson, W. [1985a]. The plastic hinge position in a circular cantilever when
struck normal to its plane by a constant jet at its tip. Int. J. Impact Engng. 3, 143-154.
Yu, T.x. and Johnson, W. [1981]. The location of the plastic hinge in a quadrantal circular curved beam
struck in its plane by ajet at its tip. AS ME Structures and Materials Conference, Washington D.C.,
November, 1981. Advances in Aerospace Structures and Materials - 1 (eds S.S. Wang and W.J.
Renton), ASME, Washington D.C., 175-180.
Yu, T.x., Symonds, P.S. and Johnson, W. [1985b]. A quadrantal circular beam subjected to radial impact
in its own plane at its tip by a rigid mass. Proc. Roy. Soc. Lond. A400, 19-36.
Yu, T.x., Symonds, P.S. and Johnson, W. [1986]. A reconsideration and some new results for the circular
beam impact problem. Int. J. Impact Engng. 4, 221-228.
Zhang, T.G. and Yu, T.x. [1986]. The large rigid-plastic deformation of a circular cantilever beam
subjected to impulsive loading. Int. J. Impact Engng. 4, 229-241.
Zhang, T.G. and Yu, T.X. [1989]. Dynamic plastic response of a simply supported ring to an impact on
its top (in Chinese). Acta Mechanica Sinica, special issue, 148-155.
Zhou, Q. and Yu, T.X. [1987]. Plastic hinge in beams of variable cross-section under intense dynamic
loading (in Chinese). Explosive and Shock 7,311-318.
Chapter 7

Impact Experiments

In his seminal paper on dynamic deformation due to impact at the tip of a cantilever,
Parkes [1955] showed sketches of the final profiles of some cantilevers struck by
lightweight, high speed bullets, and others struck by a larger mass that collides at
a relatively slow speed. For the same impact energy, these two sets of tests had
deformations that were qualitatively distinct; the light mass gave a smoothly curved
final shape with the largest curvature near the tip whereas the heavy mass gave
almost all of the deformation concentrated near the root similar to the modal
solution. This distinction on the basis of mass at the impact point in comparison
with mass of. the remainder of the uniform beam was convincing evidence for the
utility of the rigid-perfectly plastic constitutive approximation.

7.1 Methods Of Applying Dynamic Loads

As has been mentioned, there are several methods of applying intense but short
duration loads to structures. These range from impact of a colliding missile to a
blast wave from a nearby explosion. For experiments designed to measure dynamic
response, some methods of loading are better than others because they provide a
more accurately measured input impulse or energy.

7.1.1 Missile Impact

High Speed Bullet Impact Missiles fired from guns at speeds larger than
100 m s-1 easily achieve contact durations of no more than 0.1 ms if both the
missile and the target are composed of hard materials; i.e. if the materials have
relatively large flow stresses. To obtain exp(:rimental measurements of impact
phenomena, it is ordinarily best to employ spherical missiles for impact tests.
This eliminates the need to consider tumbling of the missile both in free flight
from the gun muzzle to the target and during impact. While it is straight- forward
260 Impact Experiments

to measure the missile speed before impact, this method of loading suffers from
difficulty in obtaining both the impulse and energy imparted to the target. The
missile either rebounds from the impact with unknown velocity or it is captured
in a block of soft material at the impact point. In the latter case, some soft
material from the block is usually ejected backwards at high speed; also some
energy is dissipated by deformation around the contact region. In both cases, the
initial momentum and kinetic energy imparted to the target are not simply related
to the missile impact parameters.

Drop Hammer or Powered Sledge For impact speeds of the order of 10 m S-l,
a gravitationally powered drop hammer can provide controlled impact conditions
for heavy masses. The impact speeds of these rigs can be increased somewhat by
using elastic bungee cords to increase the acceleration of the falling mass.
Sometimes a higher terminal speed can be achieved by orienting the guide rails
in a horizontal plane because this permits acceleration over a longer stroke; here,
bungee cords may be the only source of power that drives the hammer. As the
hammer (or sledge) is accelerated towards the anvil, most of these devices
experience a series of small shocks applied to the hammer by the constraining
guide rails and this excites large vibrations in the hammer. These vibrations are
one reason why it is not ordinarily a good idea to mount slender compliant test
specimens on a hammer head that is struck against a fixed anvil; the vibrations
result in substantial uncertainty about initial conditions for the test specimen at
impact. The duration and shape of the force pulse applied by the hammer to the
anvil can be somewhat controlled by shaping the hammer head; e.g. a conical
hammer head composed of a low strength material gives a force that first increases
and subsequently decreases as a function of time. The hammer sketched in Fig.
7.1 illustrates some of these features of drop hammer design.

J;:- missile ~ Vo•

G
I~
.. "",·"." .. ":,,,,'=====j·" ,',', ,':,:,', >: ,'",

a bullet impact

detosheet
explosive drop hammer
buffer

c contact explosive

Fig. 7.1. Various methods of applying impulsive or short duration pulse loads to test specimens,
7.1 Methods Of Applying Dynamic Loads 261

7.1.2 Explosive Loading

Contact Explosives Sheet explosive (Detasheet) or explosive cord (Primacord)
have detonation speeds of 5-8000 m S-1 so they can be used to provide nearly
simultaneous impulsive loading over a surface of a test specimen. Because explosive
pressures behind the shock front are very large, explosions in contact with a test
specimen can result in rear surface spallation. Spallation is prevented by covering
the loaded surface with a thin sheet of neoprene used as a standoff that somewhat
diffuses the pressure pulse. Contact explosions provide a uniform distribution of
initial impulse with a magnitude that is controlled by changing the explosive
thickness. The impulse intensity is calibrated by measuring the post-detonation
velocity of free flying specimens that have the same thickness and material
properties as the test specimens. Although the detonation speed is faster than
elastic wave speeds in metals or plastics, if detonation is initiated at one end, that
end of the structure begins to deform before the explosion pressure wave reaches
the other end of the structure. Thus, it is preferable to initiate detonation at the
center or use a 'wave shaper' to obtain more nearly simultaneous detonation over
the specimen (see Rinehart and Pearson [1963]).

Explosive Blast Pulses A blast wave propagated outward from an explosive

detonation has a pressure that decays exponentially with distance behind the shock
front. The speed of the shock front decreases with increasing radius from the
explosion while simultaneously, the shock pressure at the front decreases. When
the blast wave impinges on the surface of a lest specimen, the wave is at least
partially reflected and this increases the pressure acting on the specimen in
comparison with the pressure behind the free running pulse. Unless the test specimen
is extremely close to the explosion, blast waves have a duration for the pressure
pulse that is not insignificant in comparison with the structural response period
so they should be classified as pulse loads. Blast pressure pulses of rather large
amplitude can be measured directly by Manganin piezoelectric pressure gauges.
Details of the decay of shock pressure and speed with increasing radius from an
explosion are given by Kinney [1962] and Glasstone [1959].

7.1.3 Magnetomotive Loading

Magnetomotive loading is caused by an interaction between two currents; this is
produced by a current passed through two parallel conductors or by the magnetic
field from one conductor that induces eddy currents in a secondary conductor.
These fields give a force that tends to drive the conductors apart; this force is
proportional to the square of the current passing through the conductor. Force
pulses of less than 10 ms are generated by the current from a rapid-discharge
capacitor bank (see Hall et al. [1971]). The impulse imparted to the test specimen
must be measured indirectly; e.g. by mounting the test specimen on a ballistic
pendulum. Then if loading is applied to the test specimen only, the initial momentum
of the system equals the impulse imparted to the specimen. This momentum can
be calculated from the amplitude of swing of the pendulum.
262 Impact Experiments

250
aluminium alloy

N
,
E
E
"-
Z

t 3.25
I/)
I/)

L
~
Iii
.!!!
iii
100 I----r-'------------t~
9
I~~-----1~50~-----.~1 -r
c
2
50

00
200

Fig.7.2. Specimen configuration and stress-strain relation for blast loaded aluminum alloy (HI5) cantilevers;
h = 3.25 mm, b = 6 mm, L = 150 mm .

7.2 Travelling Hinges - Fiction or Fact?

Dynamic analyses of deformation processes in plastically deforming beams, plates

or shells separate into a transient or moving hinge phase and a modal phase of
motion. Discrete hinges that travel away from the impact point are a consequence
of the rigid-perfectly plastic structural idealization. In practice, many materials
show a significant range of elastic deformation and also strain hardening. These
effects tend to diffuse the concentrated bending moment at a 'hinge'. Numerical
investigations of dynamic response of nonlinear structures suggest nevertheless
that a concentrated region of large flexural deformation does travel away from
the impact point in the first phase of structural response, Reid and Gui [1987].
The length of the deforming region is larger than the thickness of the member but
short in comparison with the length of the beam.
To investigate the development of plastic deformation, Shu et al. [1992] reported
strain gauge measurements made on aluminum alloy cantilevers of length L =
150 mm that were hit transversely at the tip by an explosive blast pulse. The
blast strikes a light block at the tip, r = G/pL = 0.39. The specimen dimensions,
strain gauge locations and material properties are described in Fig. 7.2. These
tests showed that in the half-length adjacent to the tip, all of. the plastic curvature
developed in the first millisecond during the initial transient or travelling hinge
phase of motion. The curvature at a section 40 mm from the tip is shown as a
continuous line in Fig. 7.3a. (The curvatures in this figure have been translated
from the difference between strains on top and bottom surfaces of the beam.)
Near the root the curvature developed somewhat later as indicated by the dashed
line in Fig. 7.3a. Moreover, the slower rise time at the latter location showed that
the deforming segment increased in length as it travelled. At location 2 near the
root the pair of gages measured a travelling speed of the deforming segment that
had slowed to roughly 50 m s-1 while the length of this segment had increased to
Lh / h '" 10. A few milliseconds after the travelling pulse passed this second pair
7.2 Travelling Hinges - Fiction or Fact? 263

---
¢ ... .,. .,.-
~1
20

15

-
<lJ
'-
OJ
a
>
'-
:J JC 1f
U

10

20

phase
,.- 15 I
E
QI
'-

o>
:J

'-
:J
U

Ky _ __

o 5 10 15 20
time (ms)

Fig. 7.3. Strain gauge measurements of transient curvature in blast loaded aluminum cantilevers with light
block attip. r =0.39; (a) gauges 40 mm and 125 mm from the tip. Ii;, I Mu =1.95. (b) gauges 110 mm and 125
mm from the tip. Ii;, 1Mu = 1. 65 .

of gauges, location 2 experienced another increase in curvature. This second stage

of bending was more diffuse; it was associated with the region of modal deformation
that spread from the root of this strain hardening, aluminum alloy beam.
For a similar test with somewhat less input energy, Fig. 7.3b shows the
development of curvature at two pairs of strain gauges located a distance 15 mm
apart at sections about 3L14 from the tip. The deforming segment arrived at these
sections approximately 10-3 s after impact; at this time the average speed of the
travelling flexural pulse had slowed to 35 m s-l. Notice. that the first stage of
bending began in the section that is closer to the tip at X = O.73L . A few
milliseconds after the flexural pulse passed each gauge location, a second phase
264 Impact Experiments

Fig.7.4. Photograph of final deflection of aluminum alloy (Hl5) cantilevers with light tip mass r = 0.39 that
is hit transversely at tip by explosive blast pulse. Input energies were E in I Mu = 0.97, 1.65 and 1.95.

of modal bending began, The modal deformation began first at the gauge
located at X = 0.84L, close to the root. In this test with a large input energy
EinlMu = 1.65, most of the final curvature at these sections near the root developed
during the second or modal phase of deformation.
These aluminum cantilevers with substantial strain hardening show that if the
input energy is large and the colliding mass is lightweight, then far away from
points of displacement constraint most plastic deformation develops within a short
bending segment that travels away from the impact point. The final configurations
of the deformed cantilevers are shown in Fig, 7.4. In comparison with these
experiments, rigid-plastic calculations of the final tip rotation underestimate the
rotation if the fully plastic moment Mp is estimated by the ultimate stress au and
overestimate the rotation if the fully plastic moment is obtained from the yield
stress Y; i.e. My < Mp < Mu' These cantilevers with lightweight tip masses r =
0.39 have a final deformed shape with curvature that decreases with distance
from the tip. Bullet impact and explosive loading on an aluminum alloy cantilever
(BS 1470 SIC) by Hall et al. [1971] showed a similar distribution of curvature.
These results are not in agreement with the theory for strain hardening materials.
By requiring the stress distribution to satisfy yield over the full length of the
member ahead of the travelling hinge, that theory neglects limitations introduced
by elastic deformations and thereby exaggerates the effect of strain hardening for
the transient phase of motion. In the initial phase of motion of an elastic-plastic
beam, the effective length of the beam is limited to the span behind the elastic
wavefront that moves outward from the region of load application.
7.3 Elastic Effects on Plastic Deformation 265

c
0
Q.
tJ)
'0
E
'0
c
.~

Vi

0
time (ms I

Fig.7.S. Surface strain at midspan of magnetomotively loaded mild steel beam as a function of time. Impulse
intensity P = 750 Pa s which is five times the minimum intensity to initiate plastic deformation.

7.3 Elastic Effects on Plastic Deformation

Slender members can experience rather large elastic deflections preceding any
plastic deformation. The most obvious effect of elastic behavior at small strains
is springback from the maximum deformation that occurs at each section; this
effect which occurs during unloading, is most noticeable if the energy ratio is not
large; e.g. R < 10. For simply supported beams subjected to uniformly distributed

Fig. 7.6. Photograph of final deflection of aluminum alloy (H 15) cantilevers hit transversely at tip by explosive
blast pulse.
266 Impact Experiments

magnetomotive pulses of short duration, strain gauge measurements of transient

response by Forrestal and Wesenberg [1976, 1977] have shown elastic springback
that is almost independent of the intensity of the applied impulse (see Fig. 7.5).
For slender beams, elastic deflections can exceed the depth of the beam.
In addition to widespread elastic springback, some cantilevers and free-free
(unsupported) beams subjected to transverse impact near an end experience a
bending moment distribution that causes the short, plastically deforming segment
to hesitate and sometimes halt near midlength of the beam. This leaves a large
curvature or kink that is not anticipated by the rigid-plastic theory - some
examples are shown in the photographs of Fig. 7.6 . Final curvature of the cantilever
in the segment between the central kink and the root is smaller than that expected
from the rigid-perfectly plastic theory. The analysis of a cantilever with an elastic
support at the root (Sect. 5.5) and numerical studies (Symonds and Fleming [1984],
Reid and Gui [1987]) indicate that the central kink is due to elastic variations in
the bending moment ahead of the travelling deforming segment. The effect of
these elastic variations gives a central kink if the energy ratio R is moderately
large 5 < R < 50. In this range there is large plastic curvature but if the mass ratio
r is small enough, more energy is dissipated along the beam rather than at the
root. The effect is most prevalent with small mass ratios r < 0.25 as expected
from Fig. 5.44. A similar concentrated region of large curvature appears in
experiments where one end of a free-free beam is subjected to transverse impact.
A numerical analysis by Woodward [1984] has shown that this is also caused by
elastic variations in the bending moment ahead of a short plastically deforming
segment which travels away from the impact point.

7.4 Strain Hardening and Strain-Rate Effects

Experiments on impulsively loaded structures show that calculations which use
the rigid-perfectly plastic approximation exaggerate the magnitude of terminal
deflections. This is due partly to neglect of the elastic strain energy stored in the
deforming structure. Two other possible causes are neglect of strain hardening
and strain-rate effects. Most materials exhibit strain hardening at strains E > Ey
and many have a yield condition that is more or less rate sensitive. These effects
increase the yield stress and thereby decrease the predicted deformation; they
also modify the distribution of deformation. Strain hardening and rate effects
disperse the plastically deforming region so at any instant deformation is no
longer concentrated at a plastic hinge; instead, with these effects, the deformation
develops in a short segment of a. structure and the segment expands in length
with increasing local deformation. These effects cause preliminary deformation
that increases the yield moment ahead of a primary travelling deforming segment.
This tends to shift the distribution of structural deformation towards the modal
configuration which develops naturally as the final stage of dynamic response to
impulsive loads.

7.4.1 Mode Approximations

Approximations for the stress amplification in a short plastically deforming zone
were described in Sect. 5.1.2. The simplest approximation entirely neglects the
transient phase of response and assumes that all of the initial input energy ein is
7.4 Strain Hardening and Strain-Rate Effects 267

0.2

°O~------~------_~I~------~I----
10 20 30
impact velocity Vo(ms-1j

Fig. 7.7. Comparison of strain hardening and strain-rate modal solutions with test results for impact at tip of
eo
mild steel cantilever. (a) - - strain hardening and rate sensitivity; (b) - - - strain-rate sensitivity only, =
40 S-I and T =5; (c) - - - strain hardening only, E, /Y =6 (after Symonds [1965]).

dissipated during phase II in the mode configuration. This is valid if the mass
ratio r of a rigid body attached to the structure is moderately large so deformation
in phase I is negligible. To estimate the strain rate in the stationary pattern of
deformation given by a mode, the deforming segment is assumed to be spread
over a length Lh where 1 < Lhlh < 12. This gives an approximate curvature
/( = 281 Lh and rate of curvature ,c = 2in Lh where the factor 2 is an approximation
that accounts for a linear variation of curvature in the deforming segment.
With these approximations for curvature and rate of curvature, the bending
moment Md at a deforming section can be related to the initial fully plastic moment
Mp and the rate indepen'dent moment M;(/() that depends on strain hardening in
accord with Eq. (1.18),
Md MSp (./( )IIT
---=--+ - (7.1)
Mp Mp ,cor '

,cor =e!o )(2~:IJ

where the rate independent fully plastic moment at the current strain M;(/() is
given by
(7.2)

am = ErIo (7.3)
and the last equation for the hardening coefficient am applies to a beam with
rectangular cross-section.
The separate effects of strain hardening and strain-rate together with the
combination of these two effects on the final deflections of an impulsively loaded
cantilever with a moderately heavy tip mass are illustrated in Figs 7.7 and 7.8.
The graphs compare calculations of the final rotation at the tip 8'fv of a hardening
and rate dependent cantilever with similar calculations for the final tip rotation
8lf obtained using the elementary rate independent and perfectly plastic material
approximation. The material parameters that were used to calculate the curves
268 Impact Experiments

0.4

0.2

o~--------~--------~----------~-----
o
Impact velocity Va (ms-')

Fig.7.S. Comparison of strain hardening and strain rate modal solutions with test results for impact at tip of
6061-T6 aluminum alloy cantilever. (a) --- strain hardening and rate sensitivity; (b) - - - strain-rate
sensitivity only with £0=6500 S-1 and r=4; (c) - - - strain hardening only, E, / Y = 2, (after Symonds [1965]).

represent mild steel and 6061-T6 aluminum alloy, respectively. To distinguish

between rate independent and rate dependent effects, the graphs also contain
experimental points from Bodner and Symonds [1962]; these demonstrate that the
perfectly plastic material approximation gives excessively large deformations. In
these tests, the ratio of incident energy to maximum elastic strain energy was
large R > 10 and the mass at the tip was moderately heavy r = 2 so the mode is
a reasonably good approximation for the distribution of deformation in the beam.
These cantilevers were loaded by a small explosive charge detonated on one side
of the mass at the tip.
Comparisons of the various material approximations with the experimental data
indicate that rate dependence estimated with the Cowper-Symonds parameters is
more important than strain hardening effects in representing these tests. The rate
effect becomes significant at a rather small energy ratio R (or initial rate of
change of curvature ,() and flattens out as strain-rate increases at larger values
of R. The calculated curves show that the length of the plastically deforming
segment Lh has some influence as a result of hardening but very little as a result
of strain-rate.
A different modal approximation method for deformations of beams composed
of rate sensitive materials is due to Forrestal and Wesenberg [1976, 1977]. Rather
than assuming a priori some length for the deforming region, they calculated the
initial elastic response of a rate-insensitive beam and obtain the strain-rate in the
section where curvature is largest. This strain-rate is used in the Cowper-Symonds
relation to calculate a dynamic yield stress Y,j. Once the maximum elastic stress
reaches the dynamic yield stress, it is assumed that the structure is rigid-perfectly
plastic with a fully plastic moment obtained from the yield stress Y,j and an
initial momentum for the mode that is equal to the transition momentum of the
elastic phase of response. Thus the problem has been reduced to a rate-insensitive
analysis that separates into an initial elastic and a subsequent perfectly plastic
phase of response. The transition between these two phases occurs when the
elastic stresses satisfy a rate sensitive yield condition; the subsequent plastic
7.4 Strain Hardening and Strain-Rate Effects 269

Table 7.1 Yield stress amplification due to strain-rate in mild steel beam
(Structural parameters: L = 0.21 m, M p = 107 Nm, To = 14.8 ms)
Impulse Nondim. Input Elastic Nondim. Yield stress
intensity total impulse energy ratio period elastic period amplification
(Pa s) Pf Kol Mp TI (ms) r., ITo YdlY
200 0.283 0.04 2.495 0.169 1.32
300 0.414 0.09 1.448 0.098 1.46
450 0.616 0.20 0.942: 0.064 1.50
600 0.848 0.36 0.710 0.048 1.56
750 1.060 0.57 0.573 0.039 1.60

deformation depends upon the rate sensItIve yield stress ld. The basis of this
approximate method is an observation by Perrone [1965] that rate sensitivity of
the yield stress is largest at small strains so the dynamic yield stress depends
most on the strain-rate when yield initiates. This approximate method of
incorporating rate effects was successfully employed to calculate dynamic response
of stainless steel beams by Forrestal and Sagartz [1978].
Table 7.1 gives the amplification factor for yield stress calculated by Forrestal
and Wesenberg [1977] for some tests on simply supported mild steel beams.
These correction factors for the rate-effect gave a very accurate estimate of the
maximum deflection measured in their experiments on magnetomotively loaded
mild steel beams.

(j
14 /
o
12 I·
---
aluminum
6061 - T6

0.6
~
01.

x
~
aluminium alloy tests
02 o mild steel tests
-1
o =fJL/(j
I I
2.0 2.5
inverse mass ratio

Fig. 7.9. Final tip rotations from bullet impact tests in comparison with elementary rigid-plastic theory
(perfectly plastic theory gives final rotation proportional to input energy). Dashed curves are a best fit to data
for aluminum alloy and mild steel beams. Solid curves are the part of the input energy dissipated in transient
(el) and modal (en) phases according to rate independent rigid-plastic theory (after Bodner and Symonds
[1962]).
270 Impact Experiments

7.4.2 Transient Analysis Including Rate Effects

An analysis for the transient response of an impulsively loaded, strain-rate
dependent cantilever was presented in Sect. 5.1.1; instead of two separate phases,
the deformation develops in a single continuous motion. This analysis assumed
that the plastically deforming region initially extended throughout the entire length
of the cantilever but that a rigid (nondeforming) segment of increasing length
spread outward from the impact point at the tip. This analysis neglected any
inertial effects in the deforming region. According to Ting [1964] the approximation
has little effect on calculations of the final deflection. This form of transient rate-
dependent analysis was used by Bodner and Symonds [1962] to investigate rate
effects in mild steel and 6061-T6 aluminum alloy cantilevers. They performed
explosive loading and bullet impact tests on specimens with a lightweight tip
mass 0.3 < r < 1.8. Almost all the tests on mild steel cantilevers had input
energy ratios R > 10, while the aluminum alloy cantilevers were hit with energy
ratios R> 3. Consequently the cantilevers in these tests exhibited fairly large
final deflections.
Results of these experiments are compared with calculations from a rigid-
perfectly plastic structural analysis in the data points plotted on Fig. 7.9; there
the ratio of experimental to calculated final tip rotation is given as a function of
the mass ratio r. The experimental values given on the graph indicate that the
final tip rotations are in the range 24° ~ 8Af ~ 64~ In comparison with the
predictions of the rate independent, perfectly plastic analysis, the deformation in
most tests is reduced, especially at large mass ratios. This is attributed to strain-
rate dependence of the flow stress; this effect is largest at large mass ratios where
a modal pattern accounts for most deformation. The curves labelled e, and ell
indicate the part of the input energy dissipated in phases I and II of the rate
independent perfectly plastic analysis, respectively.

7.5 Dynamic Rupture

Rupture occurs when a crack penetrates through the full thickness of a member.
In ductile members the deformation processes that culminate in rupture almost
always involve shear or stretching in addition to bending. Thus the elementary
rigid-perfectly plastic approximation is of little help in analysing or predicting
rupture. To consider effects of shear and stretching superposed on bending, it is
necessary to include these other components of deformation and the conjugate
stresses in the structural response analysis.

7.5.1 Location and Mechanism of Rupture

Rupture of a ductile member initiates when either the strain or the density of
dissipated energy at a section exceeds some characteristic value. The pattern of
deformation determines where energy is dissipated at an instant of response; at
any time the distribution of accumulated deformation throughout a structure can
7.5 Dynamic Rupture 271

80

<II
'"
:;
Cl. 20
.~

depth h (mm)

Fig. 7.10. Impulse intensity at rupture of explosively loaded 6061-T6 aluminum alloy beams clamped at the
ends where tap = dyne s cm-2 (after Menkes and Opat [1973]).

be obtained by integrating the rate-of-deformation from the initial to the current

time.

Distributed Dynamic Loading The distribution and relative magnitude of various

components of deformation or generalized strain are most evident for steady pulse
loading on a rate-independent structure. While: a constant pressure is applied to
some part of a structure, the pattern of deformation is stationary and each component
of deformation increases linearly with time. Symonds [1968] used a square yield
surface to approximate the interaction between flexure and shear in an analysis of
deformation in a simply supported beam subje:cted to a uniform pressure pulse.
This showed that more energy can be dissipated by shear adjacent to the supports
rather than flexure if the length Llh is small and the applied force FIFe» 1.
Experiments on impulsively loaded thin aluminum alloy (6061-T6) beams by
Menkes and Opat [1973] have shown that with both ends clamped to prevent any
displacement or rotation, the deformation that culminates in rupture is even more
complicated; this is because clamped ends constrain axial displacement and induce
stretching. These beams were driven transversely by a uniform explosive load
and those that ruptured always broke adjacent to the root. The fractures initiated
on the convex surface of the beam where bendi.ng, stretching and shear combine
to give the largest effective strain. With increasing intensity of the applied impulse,
Menkes and Opat identified three patterns of deformation among which the latter
two mechanisms culminate in rupture:
I. Large flexural deformation
II. Large flexural deformation with tensile tearing at ends
III. Small flexural deformation with shear deformation and shear rupture
at ends
Figure 7.10 illustrates that the behavior changed from one of these mechanisms
of rupture to another at the same impulse intensity for beams of 102 mm and 203
272 Impact Experiments

mm length. At least for shear rupture in mechanism III, this meant that the input
energy was large enough so that shear sliding at the generalized hinges adjacent
to the supports persisted until rupture. For impulsive loading, the shear sliding
mechanism occurs at discontinuities in the initial velocity field.
These experiments were analyzed by Shen and Jones [1992]. They attributed
the initiation of tearing rupture at the transition to mechanism II to a critical
strain on the convex surface that develops when the central deflection is large
enough to add significant stretching to the flexural strain at a clamped end. For
impulses larger than that for transition to mechanism II, shear sliding at the hinge
at each end of the beam reduces the effective thickness and hence decreases the
stretch integrated over the total length that is required to rupture the cross section
of the beam. This integrated stretch or increase in length is directly related to the
central deflection. For the aluminum alloy beams in Fig. 7.10, they calculated
that the mechanism II-III transition occurred when roughly 45% of the energy
dissipated at end hinges was due to shear sliding.

Concentrated Dynamic Loading If a transverse impulsive load is applied to a

concentrated mass on a slender structure or a heavy missile strikes a light beam,
the pattern of deformation and hence the distribution of energy dissipation at any
time are different from those obtained from a uniform impulsive load. Again
shear sliding occurs at the velocity discontinuities but now these are located
adjacent to the colliding missile as described in Sect. 5.3.3. In this case the
plastic hinges are more diffuse than was found with uniform impulsive loading
where the hinges were located adjacent to displacement constraints; consequently
more energy can be dissipated at each hinge before rupture.
Liu and Jones [1987] performed drop weight impact tests on aluminum alloy
and mild steel beams of half-length L = 51 mm with clamped end conditions. The
colliding missile had a nose radius roughly equal to the depth of the beam and
was much heavier than the beam so the mass ratio y > 250. This missile struck
either the center or an off-center position along the clamped beam at a transverse
impact speed in the range 2 < Va < 12 m s-~ Although the initial kinetic energy of
the colliding mass was very large in comparison with the capacity of the beam,
the impact speed was slow so both aluminum and mild steel beams ruptured by
predominantly shear sliding only if the impact point was immediately adjacent to
a clamped end. Otherwise, rupture occurred at the impact point after stretching of
the beam had become significant due to large deflections. These ruptures initiated
on the tensile side of the beam. These results for impact by a heavy striker have
been correlated with results for other loading conditions by Jones [1989a].

7.5.2 Measurements of Generalized Strain at Rupture

Rupture initiates at the section with the largest tensile strain when the largest
principal strain equals a rupture or ultimate strain t: r . In general the rupture
strain depends on triaxiality of the strain state; for unaxial stress however, it is a
material property. Values of the rupture strain measured in uniaxial tension tests
are given in Table 7.2.
The tensile strain in the outer fiber at a plastic hinge is the sum of the largest
flexural strain max( t: m ) and the strain due to stretching t: n . The criterion for
7.5 Dynamic Rupture 273

Table 7.2 Material properties for aluminum alloy and steel specimens
Material Thickness Yield Stress Ultimate Stress Rupture
(mm) (N mm- 2 ) (N mm- 2 ) Strain
AI. alloy plate
(BSI470-HSI5)" 182 318 0.19
AI. alloy bar" 3.81 354 475 0.19
5.08 354 475 0.19
6.35 354 475 0.19
7.62 412 553 0.15
AI. alloy
sheet (HI5)b 3.25 103 269 0.18
AI. alloy
sheet (6061-T6)C <20. 280 336 0.10
Steel plate
(BS4360-43A)" 324 SOl 0.31
Steel bar" 3.81 337 464 0.39
5.08 337 464 0.39
6.35 302 444 0.40
7.62 302 444 0.40

aLiu and Jones [1987]

bShu et al. [1992]
CMil. Hdbk. 5 [1959]

tensile or tearing rupture is that the largest tensile strain equals the rupture strain
cr ; i.e.
(7.4)

In any particular problem, the strains are related to displacements through kinematic
constraints; they also depend on the hinge length Lh (i.e. the extent of a plastically
deforming region).
For example, the fundamental mode of deformation for a rigid-perfectly plastic
uniform beam of length 2L is illustrated in Fig. 7.11. If the ends of the beam are
clamped, there is a plastic hinge at the center and one at either end. These hinges
are short in comparison with the length of the beam, LhlL « 1. Assuming that
stretching due to moderately large deflections is distributed in proportion to the
curvature at each hinge, we can estimate the separate components of strain and
hence the smallest central deflection for tensile rupture at a hinge,

Fig. 7.11. Fundamental dynamic mode of deformation for uniform beam clamped at the ends.
274 Impact Experiments

max(em ) = h1f:12 = hWz l2Uh

en = {(I + Wz2 fL2 (2 -1}(Ll2Lh )

Here it is clear that hinge length Lh is a crucial but ill-defined parameter for
obtaining predictions of rupture. The length Lh is typically specified in the range
1 < Lhlh < 12; this length certainly increases due to dispersion in strain hardening
and strain-rate dependent materials. Most likely, hinge length depends also on the
shear deformation at a hinge. In analysing experimental results for transition from
one mechanism of rupture to another, Shen and Jones [1992] observed that the
hinge length decreased as the dissipation due to shear Dq became a larger part of
the total dissipation D at any particular hinge. Their empirical equation for the
data of Menkes and Opat was as follows:
(7.5)

As hinge length decreases the tensile strain increases where the strain is compatible
with deflection of an axially constrained beam; thus, large impulses that give
significant shear deformation at initial velocity discontinuities in axially constrained
beams also reduce the central deflection at which rupture initiates.

Rupture by Shear Sliding Shear deformation at a generalized hinge causes

transverse relative displacement; the extent of this relative displacement at rupture
is strongly affected by the length of the plastically deforming region; i.e. L h . For
Menkes and Opat's tests of impulsively loaded clamped beams of 606l-T6
aluminum, Shen and Jones [1992] found that at rupture in mechanism III,
,1Jlh "'" 0.6. In contrast, blunt bullet impact against 6.35 mm thick 2024-T3
aluminum alloy plates gave ,1Jlh '" 0.75 for a bullet diameter of 7.82 mm (Liss,
Goldsmith and Kelly [1983] or Jones [1989b]).

Rate Effect on Rupture Strain In rate dependent materials, large strain rates
increase the flow stress for plastic deformation but decrease the ultimate or rupture
strain. The dynamic rupture strain e~ can be related to the static rupture strain
e~ by an expression similar to the Cowper-Symonds equation:

~ ~HU'T (7.6)

where Eo and r are the usual Cowper-Symonds material constants. This expression
gives a fracture strain that is inversely proportional to dynamic stress so the
dissipation density at fracture is independent of strain-rate (if hinge length is rate
independent).

References

Bodner, S.R. and Symonds, P.S. [1962]. Experimental and theoretical investigation of the plastic
deformation of cantilever beams subjected to impulsive loading. J. Appl. Mech. 29,719-727.
References 275

Forrestal, M.J. and Wesenberg, D.L. [1976]. Elastic-plastic response of6061-T6 aluminum beams to
impulse loads. J. Appl. Mech. 43, 259-262.
Forrestal, M.J. and Wesenberg, D.L. [1977]. Elastic-plastic response of simply supported 1018 steel
beams to impulse loads. J. Appl. Mech. 44,779-780.
Forrestal, MJ. and Sagartz, M.J. [1978]. Elastic-plastic response of 304 stainless steel beams to impulse
loads. 1. Appl. Mech. 45, 685-686.
Glasstone, S. [1959]. Effects of Nuclear Weapons. McGraw-Hill, New York.
Hall, R.G., AI-Hassani, S.T.S. and Johnson, W. [1971]. The impulsive loading of cantilevers. Int. J.
Mech. Sci. 13,415-430.
Jones, N. [1989a]. On the dynamic inelastic failure of beams. Structural Failure (ed. T. Wierzbicki and
N. Jones). John Wiley & Sons, 133-159.
Jones, N. [1989b]. Structural Impact. Cambridge University Press, 274.
Kinney, G.F. [1962]. Explosive Shocks in Air. Macmillan, New York.
Liss, J., Goldsmith, W. and Kelly, J.M. [1983]. A phenomenological penetration model of plates. Int. J.
Impact Engng. 1,321-341.
Liu, J. and Jones, N. [1987]. Experimental investigation of clamped beams struck transversely by a mass.
Int. J. Impact Engng. 6,303-335.
MIL-HDBK-5 [1959]. Strength of metal aircraft elements. Armed Forces Support Center, Washington,
D.C.
Menkes, S.B. and Opat, H.J. [1973]. Broken beams - tearing and shear failures in explosively loaded
clamped beams. Experimental Mechanics 13, 480-486.
Parkes, E. W. [1955]. The permanent deformation of a cantilever struck transversely at its tip. Proc. Roy.
Soc. Lond. A228, 462-476.
Perrone, N. [1965]. On a simplified method for solving impulsively loaded structures of rate-sensitive
materials. J. Appl. Mech. 32, 489-492.
Reid, S.R. and Gui, X.G. [1987]. On the elastic-plastic deformation of cantilever beams subjected to tip
impact. Int. J. Impact Engng. 6, 109-127.
Rinehart, J. and Pearson, J. [1963]. Explosive Working of Metals. Macmillan, New York.
Shen, W.Q. and Jones, N. [1992]. A failure criterion for beams under impulsive loading. Int. J. Impact
Engng. 12, 101-121.
Shu, D., Stronge, W J. and Yu, T.x. [1992]. Oblique impact at the tip of a cantilever .Int. J. Impact Engng.
12,37-47.
Symonds, P.S. [1965]. Viscoplastic behavior in response of structures to dynamic loading. Behavior of
Materials Under Dynamic Loading (ed. N.J. Huffington, Jr.). ASME, New York, 106-124.
Symonds, P.S. [1968]. Plastic shear deformations in dynamic load problems. Engineering Plasticity (ed.
J. Heyman and F.A. Leckie), Cambridge University Pn~ss, 647-664.
Symonds, P.S. and Fleming, W. T. [1984]. Parkes revisited; on rigid-plastic and elastic-plastic dynamic
structural analysis. Int. J. Impact Engng. 2, 1-36.
Ting, T.C.T. [1964]. The plastic deformation of a cantilever beam with strain rate sensitivity under
impulsive loading. J. Appl. Mech. 31, 38-42.
Woodward, R.L. [1984]. Transverse projectile impacts on beams. J. Appl. Mech. 51,437-438.
Index

Amplification factor for yield stress 269 Cowper-Symonds relation 10, 112
Associated flow rule 17 Crack stability 254
Axial force 4, 12, 25, 204 Curvature 2, 84, 90, 103

Ballistic pendulum 261 Deflection 51, 84

Beam elastic 52
Euler-Bernoulli 159 elastic-perfectly plastic 54
ideal sandwich 8, 15, 159 large 149
Timoshenko 163 large elastic 63
Bending moment 4 large elastic-plastic 67
at yield 5 residual 58
fully plastic 7 Deformed shape 84, 104, 229
Blast pulse 261 Differentiability 193, 198
Body force 32 Double hinge mechanism 241
Bounds 37 Drucker's postulate 17
dynamic response period 38, 39 Dynamically admissible stress 33
final displacement 40, 41 Dynamic rupture 144, 270
static collapse force 37 location 270
mechanism 271
Cantilever 51, 73, 260, 262 pure shear 144
bent 241, 246 shl~ar 144, 271
circular arc 215, 224, 232 shear-flexure 146
imperfect 250 transition of mechanism 272
stepped 241
subjected to impact 96 Effective load 95
sUbjected to general pulse 90 Elastica 64
subjected to rectangular pulse 80 Elastic deformation 157,265
subjected to step force 73 at root of cantilever 173
tapered 201 Elastic springback 23, 265
with initial crack 251 E1astoplastic beam-column 61
Centripetal acceleration 156 Elliptic integral 65
Characteristic load 56 Energy 35
Circular arc cantilever 215, 224, 232 kinetic 35, 104
subjected to in-plane force 215 she'ar rupture 148
subjected to in-plane impact 225 Energy ratio 149, 158, 266
subjected to out-of-plane force 232 Equation of equilibrium 33
subjected to radial force 217 Equation of motion 33
subjected to tangential force 221 Equivalent replacement of arbitrary pulse 94
Circular cross-section Euler-Bernoulli beam 159
solid 7, 15, 234 Extremal principle 36
thin-walled 15, 234
Compatibility 30 Final configuration 104, 264
Complete solution 33 Finite sized colliding missile 139
Complex configuration 191 Free - free beam 266
Constitutive relation 1
Contact force 32 Hooke's law 4
Convergence theorem 45
Convergence to dynamic mode 184 Ideal sandwich beam 8, 15, 159
Correction factor 122 Impact 96, 259
278 Index

oblique 204 general features of 88

Impulse 81, 91 pure torsion 236
effective total 95 stationary 79, 81, 89, 236
total 81, 92 travelling 82, 89, 99, 205, 225, 262
Impulsive loading 42 Plastic a 68
Initial crack 251 Power law hardening 9
Principle
J-integral 254 maximum plastic dissipation 17
virtual velocity 34
Kinematic variables 194 virtual work 34
Kinematically admissible velocity Pulse duration 80, 91
field (KAVF) 19, 30 Pulse loading 80, 90
Kink 266 blast 261
general 90
Lagrangian reference frame 29 half-sine 184
Lee's functional 45 linear decaying 93, 184
Limit function 13 rectangular 80
elastic 13 Pure bending 2
fully plastic 13
Loading 29 Quasi-static loading 29
impact 96, 134, 139, 204, 259
impulsive 42, 111, 182 Rate dependency 9
general pulse 90, 261 Rate of dissipation of energy 35
quasi-static 29, 51 Ratio of fully plastic torque to
rectangular pulse 80 fully plastic moment 234, 241
step force 73, 200, 215, 232, 241 Rectangular cross-section 6, 14, 234
Reverse plastic flow 59
Martin's principle 37 Rotary inertia 132, 139
Mass ratio 98
Mass-spring finite difference Shape factor 7
(MS-FD) model 158 for bending 7
Material 1 for shearing 8
elastic 4 for tension 7
elastic-perfectly plastic 5 for torsion 8
rigid-perfectly plastic 16 Shear phase 134
strain-hardening 5 Shear sliding 134, 274
power-law hardening 9 Single DoF system 182
visco-plastic 9 Single hinge mechanism 75, 205
Moment-curvature relation 6 Springback 23, 265
Modal shape 44, 47, 99 Static admissible stress field 34
Modal solution 44, 106 Static collapse force 37
ini tial velocity of 48 Step loading 73
Mode 44 Strain 2
best 47 at rupture 272
convergence 45 flexural 272
configuration 44 generalized 2
dynamic 44, 186 stretching 272
elastic 186 Strain-hardening 5, 55
fundamental 45 effect of 120, 266
primary 45 Strain-rate 9
effect of 111, 266
Neutral axis 3, 13 Strain-softening 121
Neutral surface 3 Stress 1
Normality condition 17 generalized 2
resultants 1, 32
Phase transition 82, 99 safe state of 35
Plastic collapse force 37, 52, 56 ultimate 264
Plastic hinge 52, 75
combined bending-torsion 235 Tamuzh's principle 37
double 241 Tearing modulus 254
effective length of 119, 121, 262-3,267 Tensile tearing 271
Index 279

Timoshenko beam 163 Virtual work rate 34

Timoshenko beam finite element
(TB-FE) model 164 Yield condition 5, 15
Traction 32 interacti ve 12
Transient phase 98 separated 18
Transverse shear 132 Yield criterion 15
Two DoF system 184 Tresca 15
Von Mises 16
Virtual velocity 34 Yield function 15, 192
Virtual work 34 differentiability of 197
extremal properties of 192