Solid Mechanics -...
Transcript of Solid Mechanics -...
Clive L. Dym · Irving H. Shames
Solid MechanicsA Variational Approach
Augmented Edition
Solid Mechanics
.
Clive L. Dym • Irving H. Shames
SolidMechanics
A Variational Approach
Augmented Edition
Clive L. DymDepartment of EngineeringHarvey Mudd CollegeClaremont, CAUSA
ISBN 978-1-4614-6033-6 ISBN 978-1-4614-6034-3 (eBook)DOI 10.1007/978-1-4614-6034-3Springer New York Heidelberg Dordrecht London
Library of Congress Control Number: 2013932844
# Springer Science+Business Media New York 2013This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part ofthe material is concerned, specifically the rights of translation, reprinting, reuse of illustrations,recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission orinformation storage and retrieval, electronic adaptation, computer software, or by similar or dissimilarmethodology now known or hereafter developed. Exempted from this legal reservation are brief excerptsin connection with reviews or scholarly analysis or material supplied specifically for the purpose of beingentered and executed on a computer system, for exclusive use by the purchaser of the work. Duplicationof this publication or parts thereof is permitted only under the provisions of the Copyright Law of thePublisher’s location, in its current version, and permission for use must always be obtained fromSpringer. Permissions for use may be obtained through RightsLink at the Copyright Clearance Center.Violations are liable to prosecution under the respective Copyright Law.The use of general descriptive names, registered names, trademarks, service marks, etc. in thispublication does not imply, even in the absence of a specific statement, that such names are exemptfrom the relevant protective laws and regulations and therefore free for general use.While the advice and information in this book are believed to be true and accurate at the date ofpublication, neither the authors nor the editors nor the publisher can accept any legal responsibility forany errors or omissions that may be made. The publisher makes no warranty, express or implied, withrespect to the material contained herein.
Printed on acid-free paper
Springer is part of Springer ScienceþBusiness Media (www.springer.com)
Foreword to the Augmented Edition
The reappearance of this book represents the culmination of a long-standing dream
of mine. The original was published in 1973. A successor volume, built on much of
the foundation of the first book to incorporate an extensive text of the finite element
method, first appeared in 1985. Notwithstanding the passage of time since these
volumes, or the advances in finite element analysis, both theory and application, the
original book on variational methods in mechanics continues to attract attention
from points flung far around the globe. To this day, I get questions about points in
the book and inquiries about the availability of solutions to its vast array of
problems. Consequently, I have long hoped to see our original work reappear.
Happily, due to the foresight, encouragement and effort of my editor, Michael
Luby, Springer agreed, subject to some modification. I proposed adding
two chapters intended to briefly introduce finite element analysis, set very clearly
in its variational context. While accounting for some of the modern theoretical
developments, the two chapters were intended to follow the style and level of the
original Dym and Shames text: state the general principles, and follow them with
detailed illustrative examples. Needless to say, the writing style is now clearly my
own, and I alone am responsible for errors and omissions.
I thank my good friend, colleague and mentor, the late Irv Shames, for encour-
aging me to write my very first book: I told him I would not do it, starting just a year
or two out of graduate school, without his being co-author. I also thank Irv and
Sheila’s children, Bruce and Lisa, for their approval to proceed with this project.
I have also benefitted greatly from suggestions and support from my Harvey Mudd
colleague, Harry E. Williams, while I was writing the (new) Chaps. 10 and 11. And
finally, many thanks to Michael Luby for his perseverance, and to his very able
assistant, Merry Stuber, for helping to keep me on track.
Claremont, CA Clive L. Dym
v
.
Foreword to the First Edition
In this text we shall employ a number of mathematical techniques and methods. We
shall introduce these techniques and methods at places where it is felt maximum
understanding can be achieved. The physical aspects of the concepts will be
stressed where possible. And, although we shall present most of this material
with the purpose of immediate use in solid and structural mechanics, we shall
also “open-end” the discussions where feasible to other fields of study. Such
discussions are of necessity more mathematical in nature and are generally
asterisked, indicating that they can be deleted with no loss in continuity.
We shall employ Cartesian tensor concepts in parts of this text and the
accompanying notation will be used where it is most meaningful. (It will accord-
ingly not be used exclusively.) For those readers not familiar with Cartesian tensors
(or for those wishing a review), we have presented a self-contained treatment of this
subject in Appendix I. This treatment (it includes exercises as well) will more than
suffice the needs of this text.
Clive L. Dym
Irving H. Shames (deceased)
vii
.
Preface to the First Edition
This text is written for senior and first year graduate students wishing to study
variational methods as applied to solid mechanics. These methods are extremely
useful as means of properly formulating boundary-value problems and also as a
means of finding approximate analytical solutions to these boundary-value problems.
We have endeavored to make this text self-contained. Accordingly, virtually all
the solid and structural mechanics needed in the text is developed as part of the
treatment. Furthermore the variational considerations have been set forth in a rather
general manner so that the reader should be able to apply them in fields other than
solid mechanics.
A brief description of the contents of the text will now be given. For those readers
not familiar with cartesian tensor notation or for those wishing a review, we have
presented in Appendix I a development of this notation plus certain ancillary
mathematical considerations. In Chap. 1, we present a self-contained treatment of
the theory of linear elasticity that will serve our needs in this area throughout the text.
Next, in Chap. 2 comes a study of the calculus of variations wherein we consider the
first variation of functionals under a variety of circumstances. The delta operator is
carefully formulated in this discussion. The results of the first two chapters are then
brought together in Chap. 3 where the key variational principles of elasticity are
undertaken. Thus we consider work and energy principles, including the Reissner
principle, as well as the Castigliano theorems. Serving to illustrate these various
theorems and principles, there is set forth a series of truss problems. These truss
problems serve simultaneously as the beginning of our efforts in structural mechan-
ics. In developing the aforementioned energy principles and theorems, we
proceeded by presenting a functional first and then, by considering a null first
variation, arrived at the desired equations. We next reverse this process by
presenting certain classes of differential equations and then finding the appropriate
functional. This sets the stage for examining the Ritz and the closely relatedGalerkin
approximation methods. In later chapters we shall present other approximation
techniques. In Chap. 4 we continue the study of structural mechanics by applying
the principles and theorems of Chap. 3 to beams, frames and rings. These problems
are characterized by the fact that they involve only one independent variable—they
ix
may thus be called one-dimensional structural problems. In Chap. 5 we consider the
elastic and inelastic torsion of shafts. Use is made of earlier methods in the text but
now, because there are two independent variables, new approximation techniques
are presented—namely themethods of Trefftz and Kantorovich. Chapter 6 dwells on
the classical theory of plates. We set forth the equations of equilibrium and the
appropriate boundary conditions, via variational methods, and then we find approx-
imate solutions to various problems via the techniques presented earlier. In Chap. 7,
free vibrations of beams and plates are covered. With time now as a variable we first
present Hamilton’s Principle and then go on to formulate the equations of motion of
beams and plates. The methods of Ritz and Rayleigh–Ritz are then employed for
generating approximate natural frequencies of free vibration as well as mode shapes.
To put these methods on firm foundation we then examine the eigenvalue–eigen-
function problem in general, and this leads to the Rayleigh quotient that will be used
in the study of stability in Chap. 9. Also developed is the maximum–minimum
principle of the calculus of variations—thus providing a continuation of the varia-
tional calculus in Chap. 2. Up to this point only small deformation has been
considered in our undertakings (the non-linear considerations thus far have been in
the constitutive laws), and so in Chap. 9 we consider large deformation theory. In
particular, the principle of virtual work and the method of total potential energy are
presented. The climax of the chapter is the presentation of the von Karman plate
theory. The closing chapter considers the elastic stability of columns and plates.
Various approaches are set forth including the criterion of Trefftz and the asymptotic
postbuckling theory of Koiter.
Note that we have not included finite element applications despite the impor-
tance of variational methods in this field. We have done this because the finite
element approach has become so broad in its approach that a short treatment would
not be worthwhile. We recommend accordingly that this text serve as a precursor to
a study in depth of the finite element approach.
At the end of each chapter there is a series of problems that either call for
applications of the theory in the chapter or augment the material in the chapter.
Particularly long or difficult problems are starred.
We wish to thank Prof. T. A. Cruse of Carnegie-Mellon University for reading
the entire manuscript and giving us a number of helpful comments. Also our thanks
go to Prof. J. T. Oden of the University of Alabama for his valuable suggestions.
Dr. A. Baker and Dr. A. Frankus helped out on calculations and we thank them for
their valuable assistance. One of the authors (C.L.D.) wishes to pay tribute to
three former teachers—Prof. J. Kempner of Brooklyn Polytechnic Institute and
Profs. N. J. Hoff and J. Mayers of Stanford University—who inspired his interest
in variational methods. The other author (I.H.S.) wishes to thank his colleague
Prof. R. Kaul of State University of New York at Buffalo for many stimulating and
useful conversations concerning several topics in this text. Finally we both wish to
thank Mrs. Gail Huck for her expert typing.
Clive L. Dym
Irving H. Shames (deceased)
x Preface to the First Edition
Contents
1 Theory of Linear Elasticity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Part A STRESS
1.2 Force Distributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Stress . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.4 Equations of Motion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.5 Transformation Equations for Stress . . . . . . . . . . . . . . . . . . . . 9
1.6 Principal Stresses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
Part B STRAIN
1.7 Strain Components . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
1.8 Physical Interpretations of Strain Terms . . . . . . . . . . . . . . . . . 25
1.9 The Rotation Tensor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
1.10 Transformation Equations for Strain . . . . . . . . . . . . . . . . . . . . 31
1.11 Compatibility Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
Part C GENERAL CONSIDERATIONS
1.12 Energy Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
1.13 Hooke’s Law . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
1.14 Boundary-value Problems for Linear Elasticity . . . . . . . . . . . . 48
1.15 St. Venant’s Principle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
1.16 Uniqueness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
Part D PLANE STRESS
1.17 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
1.18 Equations for Plane Stress . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
1.19 Problem of the Cantilever Beam . . . . . . . . . . . . . . . . . . . . . . . 57
1.20 Closure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
2 Introduction to the Calculus of Variations . . . . . . . . . . . . . . . . . . . 71
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
2.2 Examples of Simple Functionals . . . . . . . . . . . . . . . . . . . . . . . 73
2.3 The First Variation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
xi
2.4 The Delta Operator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
2.5 First Integrals of the Euler–Lagrange Equation . . . . . . . . . . . . 85
2.6 First Variation with Several Dependent Variables . . . . . . . . . . 89
2.7 The Isoperimetric Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
2.8 Functional Constraints . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
2.9 A Note on Boundary Conditions . . . . . . . . . . . . . . . . . . . . . . . 100
2.10 Functionals Involving Higher-Order Derivatives . . . . . . . . . . . 102
2.11 A Further Extension . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
2.12 Closure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
3 Variational Principles of Elasticity . . . . . . . . . . . . . . . . . . . . . . . . 117
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
Part A KEY VARIATIONAL PRINCIPLES
3.2 Virtual Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
3.3 The Method of Total Potential Energy . . . . . . . . . . . . . . . . . . 124
3.4 Complementary Virtual Work . . . . . . . . . . . . . . . . . . . . . . . . 132
3.5 Principle of Total Complementary Energy . . . . . . . . . . . . . . . 136
3.6 Stationary Principles; Reissner’s Principle . . . . . . . . . . . . . . . 140
Part B THE CASTIGLIANO THEOREMS AND STRUCTURAL
MECHANICS
3.7 Preliminary Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
3.8 The First Castigliano Theorem . . . . . . . . . . . . . . . . . . . . . . . . 143
3.9 The Second Castigliano Theorem . . . . . . . . . . . . . . . . . . . . . . 154
3.10 Summary Contents for Parts A and B . . . . . . . . . . . . . . . . . . . 159
Part C QUADRATIC FUNCTIONALS
3.11 Symmetric and Positive Definite Operators . . . . . . . . . . . . . . . 160
3.12 Quadratic Functionals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
Part D APPROXIMATE METHODS
3.13 Introductory Comment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170
3.14 The Ritz Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171
3.15 Galerkin’s Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176
3.16 Closure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
4 Beams, Frames and Rings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187
Part A BEAMS
4.2 Technical Theory of Beams . . . . . . . . . . . . . . . . . . . . . . . . . . 188
4.3 Deflection Equations for the Technical Theory of Beams . . . . . 191
4.4 Some Justifications for the Technical Theory of Beams . . . . . . 196
4.5 Timoshenko Beam Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . 200
xii Contents
4.6 Comments on the Ritz Method . . . . . . . . . . . . . . . . . . . . . . . . 208
4.7 The Ritz Method for a Series Solution . . . . . . . . . . . . . . . . . . 212
4.8 Use of the Reissner Principle . . . . . . . . . . . . . . . . . . . . . . . . . 218
4.9 Additional Problems in Bending of Beams . . . . . . . . . . . . . . . 221
Part B FRAMES AND RINGS
4.10 Open Frames . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227
4.11 Closed Frames and Rings . . . . . . . . . . . . . . . . . . . . . . . . . . . . 240
4.12 Closure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 248
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 248
5 Torsion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 257
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 257
5.2 Total Potential Energy; Equation for Torsion . . . . . . . . . . . . . . 259
5.3 The Total Complementary Energy Functional . . . . . . . . . . . . . . 261
5.4 Approximate Solutions for Linear Elastic Behavior via the
Ritz Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 266
5.5 Approximate Solutions for a Nonlinear Elastic
Torsion Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269
5.6 The Method of Trefftz; Upper Bound for Torsional Rigidity . . . 276
5.7 The Method of Kantorovich . . . . . . . . . . . . . . . . . . . . . . . . . . . 283
5.8 Extended Kantorovich Method . . . . . . . . . . . . . . . . . . . . . . . . . 289
5.9 Closure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 293
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 295
6 Classical Theory of Plates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 299
6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 299
6.2 Kinematics of the Deformation of Plates . . . . . . . . . . . . . . . . . 299
6.3 Stress Resultant Intensity Functions and the Equations
of Equilibrium . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 303
6.4 Minimum Total Potential Energy Approach . . . . . . . . . . . . . . 309
6.5 Principle of Virtual Work; Ectangular Plates . . . . . . . . . . . . . . 317
6.6 A Note on the Validity of classical Plate Theory . . . . . . . . . . . 321
6.7 Examples From Classical Plate Theory; Simply-supported
Rectangular Plates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 327
6.8 Rectangular plates; Levy’s Method . . . . . . . . . . . . . . . . . . . . . 331
6.9 The Clamped Rectangular Plate; Approximate Solutions . . . . . 336
6.9.1 Ritz Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 337
6.9.2 Galerkin’s Method . . . . . . . . . . . . . . . . . . . . . . . . . . . 341
6.9.3 Kantorovich’s Method . . . . . . . . . . . . . . . . . . . . . . . . 344
6.10 Eliptic and Circular Plates . . . . . . . . . . . . . . . . . . . . . . . . . . . 346
6.11 Skewed Plate Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 351
6.12 Improved Theory—Axisymmetric Circular Plates . . . . . . . . . . 354
6.13 Closure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 363
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 364
Contents xiii
7 Dynamics of Beams and Plates . . . . . . . . . . . . . . . . . . . . . . . . . . . . 373
7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 373
7.2 Hamilton’s Principle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 373
Part A BEAMS
7.3 Equations of Motion for Vibrating Beams . . . . . . . . . . . . . . . . 376
7.4 Free Vibrations of a Simply-Supported Beam . . . . . . . . . . . . . 380
7.5 Rayleigh’s Method for Beams . . . . . . . . . . . . . . . . . . . . . . . . 385
7.6 Rayleigh–Ritz Method for Beams . . . . . . . . . . . . . . . . . . . . . . 389
7.7 The Timoshenko Beam . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 394
Part B PLATES
7.8 Equations of Motion for Plates . . . . . . . . . . . . . . . . . . . . . . . . 403
7.9 Free Vibrations of a Simply-Supported Plate . . . . . . . . . . . . . . 405
7.10 Rayleigh’s Method for Plates . . . . . . . . . . . . . . . . . . . . . . . . . 408
7.11 Rayleigh–Ritz Method for Plates . . . . . . . . . . . . . . . . . . . . . . 411
7.12 Transverse Shear and Rotatory Inertia—Mindlin
Plate Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 417
Part C GENERAL CONSIDERATIONS
7.13 The Eigenfunction–Eigenvalue Problem Restated . . . . . . . . . . 429
7.14 The Rayleigh Quotient in Terms of Operators . . . . . . . . . . . . . 431
7.15 Stationary Values of the Rayleigh Quotient . . . . . . . . . . . . . . . 432
7.16 Rayleigh–Ritz Method Re-Examined . . . . . . . . . . . . . . . . . . . 435
7.17 Maximum–Minimum Principle . . . . . . . . . . . . . . . . . . . . . . . . 439
7.18 Justification of the Estimation from Above Assertion of the
Rayleigh–Ritz Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 441
7.19 Closure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 442
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 443
8 Nonlinear Elasticity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 449
8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 449
8.2 Kinematics of Points and Line Segments . . . . . . . . . . . . . . . . 450
8.3 Interpretation of Strain and Rotation Terms . . . . . . . . . . . . . . . 455
8.4 Volume Change During Deformation . . . . . . . . . . . . . . . . . . . 457
8.5 Changes of Area Elements During Large Deformation . . . . . . . 459
8.6 Simplification of Strain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 462
8.7 Stress and The Equations of Equilibrium . . . . . . . . . . . . . . . . 464
8.8 Simplification of Equations . . . . . . . . . . . . . . . . . . . . . . . . . . 470
8.9 Principle of Virtual Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . 472
8.10 Total Potential Energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 475
8.11 Von Karman Plate Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . 478
8.12 An Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 488
xiv Contents
9 Elastic Stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 493
9.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 493
Part A STABILITY OF RIGID BODY SYSTEMS
9.2 Stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 493
9.3 Rigid Body Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 494
Part B ELASTIC STABILITY OF COLUMNS
9.4 The Euler Load; Equilibrium Method . . . . . . . . . . . . . . . . . . . 500
9.5 Energy Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 507
9.6 Imperfection Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 510
9.7 The Kinetic Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 513
9.8 General Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 515
9.9 The Elastica . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 516
9.10 An Intermediate Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 521
9.11 A Note on Koiter’s Theory of Elastic Stability . . . . . . . . . . . . 523
Part C ELASTIC STABILITY OF PLATES
9.12 The Buckling Equation for Rectangular Plates . . . . . . . . . . . . . 530
9.13 The Equilibrium Method—An Example . . . . . . . . . . . . . . . . . 535
9.14 The Rectangular Plate Via the Energy Method . . . . . . . . . . . . 538
9.15 The Circular Plate Via the Energy Method . . . . . . . . . . . . . . . 538
Part D APPROXIMATION METHODS
9.16 Comment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 543
9.17 The Rayleigh Quotient for Beam-Columns . . . . . . . . . . . . . . . 543
9.18 The Rayleigh and the Rayleigh–Ritz Methods
Applied to Columns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 544
9.19 Rayleigh Quotient for Rectangular Plates . . . . . . . . . . . . . . . . 546
9.20 The Kantorovich Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . 548
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 553
10 Finite Element Analysis: Preliminaries and Overview . . . . . . . . . . 559
10.1 Matrix Forms of the Equations of Elasticity and Key
Variational Principles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 560
10.1.1 Matrix Representations of the Equations of Isotropic
Linear Elasticity . . . . . . . . . . . . . . . . . . . . . . . . . . . . 560
10.1.2 Matrix Representation of the Key Variational
Principles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 563
10.2 Strong and Weak Forms of Elasticity Problems . . . . . . . . . . . . 566
10.2.1 Strong Form to Weak Form . . . . . . . . . . . . . . . . . . . . 568
10.2.2 Weak Form to Strong Form . . . . . . . . . . . . . . . . . . . . 570
10.3 A Formal Statement of Finite Element Analysis
in Elasticity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 571
10.3.1 Finite Element Analysis of Elastic Bodies
and Structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 572
10.3.2 The Direct Stiffness Method . . . . . . . . . . . . . . . . . . . 577
Contents xv
10.4 Finite Elements: Degrees of Freedom, Shapes,
and Interpolation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 578
10.4.1 Polynomial-Based Approximations . . . . . . . . . . . . . . 580
10.4.2 Interpolation Across Finite Elements . . . . . . . . . . . . . 581
10.4.3 Finite Element Shapes . . . . . . . . . . . . . . . . . . . . . . . . 584
10.5 Closure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 586
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 588
11 Finite Element Applications: Trusses and Beams . . . . . . . . . . . . . 591
11.1 Truss Forces and Deflections and the Direct Stiffness
Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 591
11.1.1 Analyzing an Indeterminate Truss with Castigliano’s
Second Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . 592
11.1.2 Element Stiffness Matrix for a Truss Finite
Element . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 596
11.1.3 Truss Element Stiffness in Global Coordinates . . . . . . 599
11.1.4 Assembled Stiffness of a Truss System . . . . . . . . . . . 602
11.1.5 Solving the Assembled (Matrix) Equilibrium
Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 605
11.1.6 Determining the Bar Forces in the Truss . . . . . . . . . . 607
11.1.7 Remarks on the Direct Stiffness Methods
for Trusses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 609
11.2 Applying FEA to Axially Loaded Bars . . . . . . . . . . . . . . . . . . 611
11.2.1 Total Potential Energy and Element Equilibrium
for Bars . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 612
11.2.2 FEA of a Centrally Loaded Bar . . . . . . . . . . . . . . . . . 613
11.2.3 Solving the FEA Assembled Equilibrium
Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 617
11.3 Finite Element Analysis for Beams in Bending . . . . . . . . . . . . 618
11.3.1 Displacements and Shape Functions for Beams
in Bending . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 618
11.3.2 Total Potential Energy and Element Equilibrium for
Beam Elements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 619
11.3.3 Assembling Beam Elements to Model Beams . . . . . . . 622
11.4 Closure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 631
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 632
Appendix I Cartesian Tensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 635
I.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 635
I.2 Vectors and Tensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 635
I.3 Transformation of Coordinates and Introduction to Index
Notation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 637
I.4 Interpretation of the Free Indices: Kronecker Delta . . . . . . . . . 640
I.5 Operations with Indices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 642
I.6 Scalars and Vectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 644
xvi Contents
I.7 Tensors: Symmetry and Skew-symmetry . . . . . . . . . . . . . . . . . 648
I.8 Vector Operations Using Tensor Notation: the Alternating
Tensor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 653
I.9 Gauss’ Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 657
I.10 Green’s Formula . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 660
I.11 Closure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 662
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 662
Appendix II Rotation Tensor for a Deforming Element . . . . . . . . . . . . 667
Appendix III Integration ofR 10xa 1� xg½ �hdx . . . . . . . . . . . . . . . . . . . . 670
Appendix IV To Show that Lagrange Multipliers are Zero for
Development of the Rayleigh–Ritz Method . . . . . . . . . . . 672
Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 675
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 679
Contents xvii
Theory of Linear Elasticity 1
1.1 Introduction
In much of this text we shall be concerned with the study of elastic bodies.
Accordingly, we shall now present a brief treatment of the theory of elasticity.
In developing the theory we shall set forth many concepts that are needed for
understanding the variational techniques soon to be presented.
We will not attempt solutions of problems directly using the full theory of
elasticity; indeed very few such analytical solutions are available. Generally we
will work with special simplifications of the theory wherein a priori assumptions are
made as to
(a) the stress field (plane stress problems for example)
(b) the strain field (plane strain problems for example)
(c) the deformation field (structural mechanics of beams, plates and shells).
The significance of these simplications and the limitations of their use can best be
understood in terms of the general theory.
Specifically in this chapter we will examine the concepts of stress, strain,
constitutive relations, and various forms of energy. This will permit us to present
the equations of linear elasticity and to consider the question of uniqueness of
solutions to these equations. The plane stress simplification will be considered in
this chapter; examples of simplification (c) will be presented in subsequent
chapters.
Although we shall present certain introductory notions pertaining to finite
deformations in this chapter so as to view small deformation in the proper perspec-
tive, we shall defer detailed considerations of finite deformation to Chap. 8 wherein
we consider geometrically nonlinear elasticity.
C.L. Dym, I.H. Shames, Solid Mechanics, DOI 10.1007/978-1-4614-6034-3_1,# Springer Science+Business Media New York 2013
1
Part A
STRESS
1.2 Force Distributions
In the study of continuous media, we are concerned with the manner in which forces
are transmitted through a medium. At this time, we set forth two classes of forces that
will concern us. The first is the so-called body-force distribution distinguished by thefact that it acts directly on the distribution of matter in the domain of specification.
Accordingly, it is represented as a function of position and time and will be denoted
as B(x,y,z,t) or, in index notation, as Bi(x1,x2,x3, t). The body force distribution is an
intensity function and is generally evaluated per unit mass or per unit volume of the
material acted on. (In the study of fluids, the basis of measure is usually per unit mass
while in the study of solids the basis of measure is usually per unit volume.)
In discussing a continuum there may be some apparent physical boundary that
encloses the domain of interest such as, for example, the outer surface of a beam.
On the other hand, we may elect to specify a domain of interest and thereby
generate a “mathematical” boundary. In either case, we will be concerned with
the force distribution that is applied to such boundaries directly from material
outside the domain of interest. We call such force distributions surface tractionsand denote them as T(x,y,z,t) or Ti(x1,x2,x3,t). The surface traction is again an
intensity, given on the basis of per unit area.
Now consider an infinitesimal area element on a boundary (see Fig. 1.1) over
which we have a surface traction distribution T at some time t. The force dfi trans-mitted across this area element can then be given as follows:
dfi ¼ TidA
Note that T need not be normal to the area element and so this vector and the unit
outward normal vector v may have any orientation whatever relative to each other.
We have not brought the unit normal v into consideration thus far, but we will find ituseful, as we proceed, to build into the notation for the surface traction a superscript
referring to the direction of the area element at the point of application of the
surface traction. Thus we will give the traction vector as
TðvÞðx; y; z; tÞ
or as
TðvÞi x1; x2; x3; tð Þ
where (v) is not to be considered as a power. If the area element has the unit
normal in the xj direction, then we would express the traction vector on this element
as T(j) or as TðjÞi .
2 1 Theory of Linear Elasticity
In the following section we shall show how we can use the superscript to good
advantage.
1.3 Stress
Consider now a vanishingly small rectangular parallelepiped taken at some time
t from a continuum. Choose reference x1, x2, x3 so as to be parallel to the edges of
this rectangular parallelepiped as has been shown in Fig. 1.2. We have shown
surface tractions on three rectangular boundary surfaces of the body. Note that we
have employed the superscript to identify the surfaces. The cartesian components of
the vector T(1) are then T1(1), T2
(1), T3(1). We shall now represent these components
by employing τ in place of T and moving the superscript down to be the first
subscript while deleting the enclosing parenthesis. Hence, the components of T(1)
are then given as τ11, τ12, τ13. In general we have
Tð1Þi � τ11; τ12; τ13
Tð2Þi � τ21; τ22; τ23
Tð3Þi � τ31; τ32; τ33
In a more compact manner we have
TðiÞj ¼ τij
where the nine quantities comprising τij are called stresses and are forces per unit
area wherein the first subscript gives the coordinate direction of the normal of the
area element and the second subscript gives the direction of the force intensity
x3n
x1x2
dA
T
Boundary
Fig. 1.1
1.3 Stress 3
itself. These nine force intensities are shown in Fig. 1.3 on three orthogonal faces of
an infinitesimal rectangular parallelepiped with faces parallel to the coordinate
planes of x1, x2, x3.Representing the set τij as an array we have
τ11 τ12 τ13τ21 τ22 τ23τ31 τ32 τ33
0
@
1
A
where the terms in the main diagonal are called normal stresses, since the force
intensities corresponding to these stresses are normal to the surface, while the off-
diagonal terms are the shear stresses.
x3
x1
x2
T(3)
T(1)
T(2)
Fig. 1.2
t33
t31
t32
t23
t22t21t12t11
t13
x1 x2
x3Fig. 1.3
4 1 Theory of Linear Elasticity
We shall employ the following sign convention for stresses. A normal stress
directed outward from the interface is termed a tensile stress and is taken by
definition as positive. A normal stress directed toward the surface is called
a compressive stress (it is exactly the same as pressure on the surface) and is, by
definition, negative. For shear stresses we employ the following convention:
A shear stress is positive if (a) both the stress itself and the unit normal point in positive
coordinate directions (the coordinates need not be the same), or (b) both point in the
negative coordinate directions (again the coordinates need not be the same). A mixture of
signs for coordinate directions corresponding to shear stress and the unit normal indicates a
negative value for this shear stress.1
Note that all the stresses shown in Fig. 1.3 are positive stresses according to the
above sign convention.
Knowing τij for a set of axes, i.e., for three orthogonal interfaces at a point, wecan determine a stress vector T(v) for an interface at the point having any directionwhatever. We shall now demonstrate this. Consider any point P in a continuum
(Fig. 1.4(a)). Form as a free body a tetrahedron with P as a corner and with three
orthogonal faces parallel to the reference planes as has been shown enlarged in
Fig. 1.4(b). The legs of the tetrahedron are given as Δx1, Δx2, and Δx3 as has beenindicated in the diagram, and the inclined face ABC has a normal vector v . The
Stresses on coordinatefaces are average stresses
a b
(T (ν))av(T (ν))av
(T (ν))av
Δx3
Δx2Δx1
21
(T (ν))3
x1
x1
x2
x3
x3
x2
P P
A
B
C
av
Fig. 1.4
1Note that normal stress actually follows this very same convention.
1.3 Stress 5
stresses have been shown for the orthogonal faces as have the stress vector T(v) and
its components for the inclined face. We will denote by h the perpendicular distancefrom P to the inclined face (the “altitude” of the tetrahedron). The stresses and
traction components given on the faces of the tetrahedron are average values over
the surfaces on which they act. Also, the body force vector (B)av (not shown)
represents the average intensity over the tetrahedron. Using these average values as
well as an average density, ρav, we can express Newton’s law for the mass center of
the tetrahedron in the x1 direction as follows
� τ11ð Þav PCB� τ21ð Þav ACP� τ31ð Þav APBþ B1ð Þavρav1
3ABCh
þ �T1ðvÞ�avABC ¼ ρav
1
3ABCha1 (1.1)
where a1 is the acceleration of the mass center in the x1 direction. We can next
replace the average values by values taken at P itself, plus a small increment which
goes to zero as Δxi goes to zero.2 Thus we have
τ11ð Þav ¼ τ11ð ÞPþe1 ðB1Þav ¼ ðB1ÞP þ eB
τ21ð Þav ¼ τ21ð ÞPþe2�TðvÞ1
�av¼ �TðvÞ
1
�Pþ eT
τ31ð Þav ¼ τ31ð ÞPþe3 ρav ¼ ρP þ eρ
(1.2)
where e1, e2, e3, eB, eT and eρ go to zero with Δxi. Thus we have for Eq. (1.1):
�½ðτ11ÞP þ e1� PCB� ½ðτ21ÞP þ e2� ACP� ½ðτ31ÞP þ e3� APB
þ ½ðB1ÞP þ eB�ðρP þ eρÞ 1
3
� �
ðABChÞ þ T1ðvÞ
� �
Pþ eT
h iABC
¼ 1
3ðρP þ eρÞðABCÞha1 (1.3)
Next dividing through by areaABCand noting thatPCB=ABC ¼ v1; ACP=ABC ¼ v2,etc., we have:
� ½ðτ11ÞP þ e1�v1 � ½ðτ21ÞP þ e2�v2 � ½ðτ31ÞP þ e3�v3þ ½ðB1ÞP þ eB�ðρP þ eρÞ h
3þ T1
ðvÞ� �
Pþ eT
h i¼ ðρP þ eρÞ h
3a1
Now go to the limit as Δxi! 0 in such a manner as to keep vi constant. Clearly, thee’s disappear and h! 0. We then get in the limit:
� ðτ11ÞPv1 � ðτ21ÞPv2 � ðτ31ÞPv3 þ T1ðvÞ
� �
P¼ 0
2We are thus tacitly assuming that the quantities in Eq. 1.1 vary in a continuous manner.
6 1 Theory of Linear Elasticity
Since P is any point of the domain we need no longer carry along the subscript.
Rearranging terms we get:
T1ðvÞ ¼ τ11v1 þ τ21v2 þ τ31v3 (1.4)
For any coordinate direction i we accordingly get:
TðvÞi ¼ τ1iv1 þ τ2iv2 þ τ3iv3
And so using the summation convention, the stress vector TðvÞi can be given as
TðvÞi ¼ τjivj
where you will note that vj can be considered to give the direction cosines of the unitnormal of the interface on which the traction force is desired. We will soon show
that the stress terms τij form a symmetric array and, accordingly the above equation
can be put in the following form:
TðvÞi ¼ τijvj (1.5)
Thus, knowing τij we can get the traction vector for any interface at the point.
The above is called Cauchy’s formula and may be used to relate tractions on the
boundary with stresses directly next to the boundary.
1.4 Equations of Motion
Consider an element of the body of mass dm at any point P. Newton’s law requires
that
df ¼ dm _V
where df is the sum of the total traction force on the element and the total body force
on the element. Integrating the above equation over some arbitrary spatial domain
having a volume D and a boundary surface S, we note as a result of Newton’s thirdlaw that only tractions on the bounding surface do not cancel out so that we have,
using indicial notation:
ðð�S
TðvÞi d Aþ
ð ð ð
D
Bi dV ¼ð ð ð
D
_Viρ dV (1.6)
Now employ Eq. (1.5) to replace the stress vector TðvÞi by stresses. Thus:
1.4 Equations of Motion 7
ðð�S
τji vj dV þð ð ð
D
Bi dV ¼ð ð ð
D
_Viρ dV
Next employ Gauss’ theorem for the first integral and collect terms under one
integral. We get:ð ð ð
D
½τji;j þ Bi � _Viρ�dV ¼ 0
Since the domain D is arbitrary we conclude from above that at any point the
following must hold:
τji;j þ Bi ¼ ρ _V (1.7)
This is the desired equation of motion.
Suppose next we consider an integral form of the moment of momentum
equation derivable from Newton’s law, i.e.,M ¼ _H. Thus we may say (considering
Fig. 1.5):
ðð�S
r� TðvÞdAþð ð ð
D
r� B dV ¼ð ð ð
D
r� d
dtðV dmÞ
Considering the continuum to be composed of elements whose mass dm is constant
but whose shape may be changing, we can express the integrand of the last
expression as r � _V dm. Thus we have for the above equation in tensor notation
ðð�S
εijkxjTðvÞk d Aþ
ð ð ð
D
εijkxjBk dV �ð ð ð
D
εijkxj _Vkρ dV ¼ 0
where we replace dm by ρ dv. Now replaceTðvÞk by τlkvl and employ Gauss’ theorem.
Thus we may write the above equation as:
r
r
dA
dV
D
T(ν)(x1,x2,x3)
B(x1,x2,x3)
x3
x1 x2
Fig. 1.5
8 1 Theory of Linear Elasticity
ð ð ð
D
εijk xjτlk� �
;lþ xjBk � ρxj _Vk
h idV ¼ 0
Since the above is true for any domain D, we can set the integrand equal to zero.
Carrying out differentiation of the first expression in the bracket and collecting
terms we then get:
εijkxj½τlk;l þ Bk � ρ _Vk� þ εijkxj; lτlk ¼ 0
Because of Eq. ((1.7) we can set the first expression equal to zero and so we get:
εijkxj; lτlk ¼ 0
Noting that xj,l ¼ δjl we have
εijkδjlτkl ¼ εijkτkj ¼ 0 (1.8)
It should be clear by considering the above equations for each value of the free
index i, that the stresses with reversed indices with respect to each other are equal.
That is:
τkj ¼ τjk (1.9)
The stress components form a symmetric array.3
1.5 Transformation Equations for Stress
We have thus far shown how a stress vector T(v) on an arbitrarily oriented interface
at a point P in a continuum can be related to the set of nine stress components on a
set of orthogonal interfaces at the point. Furthermore, we have found through
consideration of Newton’s law, that the stress terms form a symmetric array. We
shall now show that the stress components for a set of Cartesian axes (i.e., for three
orthogonal interfaces at a point) transform as a second-order tensor.
For this purpose consider Fig. 1.6 showing axes x1, x2, x3 and x01; x02; x
03 rotated
arbitrarily relative to each other. Suppose we know the set of stresses for the
unprimed reference, i.e., for a set of orthogonal interfaces having x1x2x3 as edges,and wish to determine stresses for reference x0y0z0. Accordingly, in Fig. 1.7, we haveshown a vanishingly small rectangular parallelepiped having x01x
02x
03 as edges so as
3We have only considered body force distributions here. If one assumes body-couple distributionsthat may occur as a result possibly of magnetic or electric fields on certain kinds of dielectric and
magnetic materials, we will have in Eq. (1.8) the additional integralÐ Ð Ð
M dv where M is the
couple-moment vector per unit volume. The result is that Eq. (1.8) becomes εijkτkj þ Mi ¼ 0. The
stress tensor is now no longer a symmetric tensor. Such cases are beyond the scope of this text.
1.5 Transformation Equations for Stress 9
to present the set of orthogonal interfaces for the desired stresses. The stress vectors
for these interfaces have been shown as well as the corresponding stresses.
Suppose we wish to evaluate the stress component τ031. We can do this by first
computing the stress vector on the interface corresponding to this stress (this
interface is denoted as (30)). The direction cosines v01; v02 and v03 for this interface
can be given in our familiar notation as a31, a32, and a33 where the first subscript
x'3
x'2
x'1
x3
x2
x1
Fig. 1.6
1´ 2´
3'
t'33
t'31
t'32
x'3
x'2
x'1
x3
x2
x1
T(3')
T(2')
T(1')
Fig. 1.7
10 1 Theory of Linear Elasticity
identifies the normal to interface x03ð Þ and the second subscript identifies the properunprimed axis. Thus we can say, using Eq. (1.5):
Tð30Þi ¼ τija3j (1.10)
You must remember that the stress vector components for surface (30) are given
above as components along the unprimed axes. Hence, to get τ031 we must project
each one of the above components in a direction along the x01 direction. This is
accomplished by taking an inner product using a1i in the above equation.4 Thus
τ031 ¼ Tð30Þi a1i ¼ τija3 j
� �a1i ¼ a3 ja1iτij ¼ a3 ja1iτji
where in we have used the symmetry property of τij in the last expression. For
another component of stress such as τ03k we could change the (1) to free index (k) inthe above equation and we get:
τ03k ¼ a3 jakiτji
And to get a corresponding result on any other interface we need only change the
subscript (3) to free index p. Thus we can say for any stressesτ0pk.
τ0pk ¼ ap jakiτji (1.11)
We have thus shown that stress components transform as a second-order tensor.5
1.6 Principal Stresses
We have shown that given a system of stresses for an orthogonal set of interfaces at
a point, we can associate a stress vector for interfaces having any direction in space
according to the formulation:
TðvÞi ¼ τijvj
4a1i represents the set of direction cosines between the x01 axis and the x1, x2 and x3 axes. It is
therefore a unit vector in the x01 direction.5We could also have arrived at this equation by using one of the quotient rules of Appendix I in
conjunction with Cauchy’s formula
TðvÞi ¼ τijvj
since vj is an arbitrary vector (except for length) and TðvÞi is a vector.
1.6 Principal Stresses 11
We now ask this question. Is there a direction v such that the stress vector is
collinear with v? That is to say, is there an interface having a normal such that there
is only one non-zero stress—the normal stress? We shall call such a stress, if it
exists, a principal stress and we shall denote it as σ(see Fig. 1.8). Thus we can
express the above equation for this case as follows:
σvi ¼ τi jvj (1.12)
Now replace σvi by (σvj)(δij) and rearrange the above equation to form the relation:
ðτij � σδijÞvj ¼ 0 (1.13)
We have here three simultaneous equations with unknowns, σ, v1, v2, and v3.However, we do have a fourth equation involving these direction cosines, namely:
vivi ¼ 1 (1.14)
A non-trivial solution to the set of equations (1.13) requires
τij � σδij��
�� ¼ 0 (1.15)
τ11
σ
τ33
τ32
τ31
τ23
τ22τ21 τ12
τ13
x3
x1
x2
v
Fig. 1.8
12 1 Theory of Linear Elasticity