You are on page 1of 229

SCHAUM'S O U T L I N E O F

THEORY AND PROBLEMS

CONTINUUM MECHANICS

GEORGE E. MASE, Ph.D.


Professor o f Mechanics
Michigan State University

SCHAUM9S OUTLINE SERIES


McGRAW-HILL BOOK COMPANY
Nezo York, St. Louis, San Francisco, London, Sydney, Toronto, Mexico, and Panama
Ctrpyright @ 1!)70 1)s Illv(:raw-Hill. lric*. A11 Rights Rcserved. Printed in the
United States of America. No part of this publication may be reproduced,
stored in a retrieval system, or transmitted, in any form or by any means,
electronic, mechanical, photocopying, recording, or otherwise, without the prior
wrltten permission of the publisher.
ISBN 07-04M63-4
Preface
Because of its emphasis on basic concepts and fundamental principles, Continuum
Mechanics has an important role in modern engineering and technology. Severa1 under-
graduate courses which utilize the continuum concept and its dependent theories in the
training of engineers and scientists are well established in today's curricula and their
number continues to grow. Graduate programs in Mechanics and associated areas have
long recognized the value of a substantial exposure to the subject. This book has been
written in an attempt to assist both undergraduate and first year graduate students in
understanding the fundamental principles of continuum theory. By including a number of
solved problems in each chapter of the book, i t is further hoped that the student will be
able to develop his skill in solving problems in both continuum theory and its related fields
of application.
In the arrangement and development of the subject matter a sufficient degree of con-
tinuity is provided so that the book may be suitable a s a text f o r an introductory course in
Continuum Mechanics. Otherwise, the book should prove especially useful as a supple-
mentary reference for a number of courses f o r which continuum methods provide the basic
structure. Thus courses in the areas of Strength of Materials, Fluid Mechanics, Elasticity,
Plasticity and Viscoelasticity relate closely to the substance of the book and may very well
draw upon its contents.
Throughout most of the book the important equations and fundamental relationships
are presented in both the indicial or "tensor" notation and the classical symbolic or "vector"
notation. This affords the student the opportunity to compare equivalent expressions and
to gain some familiarity with each notation. Only Cartesian tensors are employed in the
text because i t is intended as a n introductory volume and since the essence of much of the
theory can be achieved in this context.
The work is essentially divided into two parts. The first five chapters deal with the
basic continuum theory while the final four chapters cover certain portions of specific
areas of application. Following a n initial chapter on the mathematics relevant to the
study, the theory portion contains additional chapters on the Analysis of Stress, Deforma-
tion and Strain, Motion and Flow, and Fundamental Continuum Laws. Applications are
treated in the final four chapters on Elasticity, Fluids, Plasticity and Viscoelasticity. At
the end of each chapter a collection of solved problems together with severa1 exercises f o r
the student serve to illustrate and reinforce the ideas presented in the text.
The author acknowledges his indebtedness to many persons and wishes to express his
gratitude to a11 for their help. Special thanks are due the following: to my colleagues,
Professors W. A. Bradley, L. E. Malvern, D. H. Y. Yen, J. F. Foss and G. LaPalm each of
whom read various chapters of the text and made valuable suggestions for improvement;
to Professor D. J. Montgomery for his support and assistance in a great many ways; to
Dr. Richard Hartung of the Lockheed Research Laboratory, Palo Alto, California, who
read the preliminary version of the manuscript and gave numerous helpful suggestions; to
Professor M. C. Stippes, University of Illinois, for his invaluable comments and suggestions;
to Mrs. Thelma Liszewski for the care and patience she displayed in typing the manuscript;
to Mr. Daniel Schaum and Mr. Nicola Monti for their continuing interest and guidance
throughout the work. The author also wishes to express thanks to his wife and children
for their encouragement during the writing of the book.
Michigan State University
June 1970
CONTENTS
Page
Chapter 1 MATHEMATICAL FOUNDATIONS
1.1 Tensors and Continuum Mechanics .................................... 1
1.2 .
General Tensors . Cartesian Tensors Tensor Rank ...................... 1
1.3 Vectors and Scalars .................................................. 2
1.4 Vector Addition . Multiplication of a Vector by a Scalar .................. 2
1.5 Dot and Cross Products of Vectors .................................... 3
1.6 Dyads and Dyadics .................................................... 4
1.7 . .
Coordinate Systems Base Vectors Unit Vector Triads .................. G
1.8 .
Linear Vector Functions Dyadics a s Linear Vector Operators .......... 8
1.9 .
Indicial Notation Range and Summation Conventions .................. 8
1.10 Summation Convention Used with Symbolic Convention .................. 10
1.11 Coordinate Transformation . General Tensors .......................... 11
1.12 The Metric Tensor. Cartesian Tensors .................................. 12
1.13 Transformation Laws f o r Cartesian Tensors . The Kronecker Delta .
Orthogonality Conditions ........................................ 13
.
Addition of Cartesian Tensors Multiplication by a Scalar ................ 15
Tensor Multiplication .................................................. 15
Vector Cross Product . Permutation Symbol Dual Vectors ............... 16
.
Matrices Matrix Representation of Cartesian Tensors .................. 17
Symmetry of Dyadics. Matrices and Tensors ............................ 19
Principal Values and Principal Directions of Symmetric
Second-Order Tensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
.
Powers of Second-Order Tensors Hamilton-Cayley Equation ............ 21
.
Tensor Fields Derivatives of Tensors .................................. 22
Line Integrals. Stokes' Theorem ...................................... 23
The Divergence Theorem of Gauss ...................................... 23

Chapter 2 ANALYSIS OF STRESS


The Continuum Concept ................................................ 44
. .
Homogeneity Isotropy Mass-Density .................................. 44
Body Forces . Surface Forces .......................................... 45
Cauchy's Stress Principle . The Stress Vector ............................ 45
State of Stress a t a Point . Stress Tensor ................................ 46
The Stress Tensor-Stress Vector Relationship .......................... 47
Force and Moment. Equilibrium . Stress Tensor Symmetry . . . . . . . . . . . . . . 48
Stress Transformation Laws .......................................... 49
Stress Quadric of Cauchy .............................................. 50
. .
Principal Stresses Stress Invariants Stress Ellipsoid . . . . . . . . . . . . . . . . . . 51
Maximum and Minimum Shear Stress Values . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
Mohr's Circles f o r Stress . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
Plane Stress . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
Deviator and Spherical Stress Tensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
CONTENTS

Page
Chapter 3 DEFORMATION AND STRAIN
3.1 Particles and Points ................................................ 77
3.2 Coritinuum Configuration . Deformation and Flow Concepts . . . . . . . . . . . . . . 77
3.3 Position Vector . Displacement Vector .................................. 77
3.4 Lagrangian and Eulerian Descriptions .................................. 79
3.5 Deformation Gradients. Displacement Gradients ........................ 80
3.6 Deformation Tensors . Finite Strain Tensors ............................ 81
3.7 Small Deformation Theory . Infinitesimal Strain Tensors ................ 82
3.8 Relative Displacements . Linear Rotation Tensor . Rotation Veztor ........ 83
3.9 Interpretation of the Linear Strain Tensors ............................ 84
3.10 Stretch Ratio . Finite Strain Interpretation ............................ 86
3.11 Stretch Tensors . Rotation Tensor ...................................... 87
3.12 Transformation Properties of Strain Tensors ............................ 88
3.13 Principal Strains. Strain Invariants . Cubical Dilatation ................ 89
3.14 Spherical and Deviator Strain Tensors .................................. 91
3.15 Plane Strain . Mohr's Circles f o r Strain ................................ 91
3.16 Compatibility Equations for Linear Strains .............................. 92

Chapter 4 MOTION AND FLOW


4.1 Motion . Flow . Material Derivative .................................... 110
4.2 Velocity. Acceleration . Instantaneous Velocity Field .................... 111
4.3 P a t h Lines. Stream Lines. Steady Motion .............................. 112
4.4 Rate of Deformation . Vorticity . Natural Strain ........................ 112
4.5 Physical Interpretation of Rate of Deformation and Vorticity Tensors . . . . . . 113
4.6 Material Derivatives of Volume. Area and Line Elements .................. 114
4.7 Material Derivatives of Volume, Surface and Line Integrals .............. 116

Chapter 5 FUNDAMENTAL LAWS OF CONTINUUM MECHANICS


5.1 Conservation of Mass . Continuity Equation ............................ 126
5.2 Linear Momentum Principle . Equations of Motion .
Equilibrium Equations .............................................. 127
Moment of Momentum (Angular Momentum) Principle .................... 128
Conservation of Energy . F i r s t Law of Thermodynamics.
Energy Equation .................................................. 128
Equations of State. Entropy . Second Law of Thermodynamics . . . . . . . . . . . . 130
The Clausius-Duhem Inequality . Dissipation Function .................... 131
Constitutive Equations . Thermomechanical and Mechanical Continua . . . . . . 132

Chapter 6 LINEAR ELASTICITY


6.1 Generalized Hooke's Law . Strain Energy Function ...................... 140
6.2 Isotropy . Anisotropy . Elastic Symmetry ................................ 141
6.3 Isotropic Media . Elastic Constants .................................... 142
6.4 Elastostatic Problems . Elastodynamic Problems ........................ 143
6.5 Theorem of Superposition. Uniqueness of Solutions .
S t. Venant's Principle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
6.6 Two-Dimensional Elasticity. Plane Stress and Plane Strain . . . . . . . . . . . . . . 145
CONTENTS

Page
6.7 Airy's Stress Function ................................................ 147
6.8 Two-Dimensional Elastostatic Problems in Polar Coordinates .............. 148
6.9 Hyperelasticity . Hypoelasticity ........................................ 149
6.10 Linear Thermoelasticity ................................................ 149

Chapter 7 FLUIDS
7.1 .
Fluid Pressure . Viscous Stress Tensor Barotropic Flow ................ 160
7.2 . .
Constitutive Equations Stokesian Fluids Newtonian Fluids ............ 161
7.3 .
Basic Equations f o r Newtonian Fluids Navier-Stokes-Duhem Equations .... 162
7.4 . .
Steady Flow Hydrostatics Irrotational Flow .......................... 163
7.5 .
Perfect Fluids Bernoulli Equation . Circulation ........................ 164
7.6 .
Potential Flow Plane Potential Flow .................................. 165

Chapter 8 PLASTICITY
8.1 Basic Concepts and Definitions ........................................ 175
8.2 Idealized Plastic Behavior .............................................. 176
8.3 Yield Conditions . Tresca and von Mises Criteria ........................ 177
8.4 . .
Stress Space The 1I.Plane Yield Surfaces ............................ 173
8.5 .
Post-Yield Behavior Isotropic and Kinematic Hardening ................ 180
8.6 .
Plastic Stress-Strain Equations Plastic Potential Theory ................ 181
8.7 .
Equivalent Stress Equivalent Plastic Strain Increment .................. 182
8.8 .
Plastic Work Strain Hardening Hypotheses ............................ 182
$8.9 Total Deformation Theory .............................................. 183
8.10 Elastoplastic Problems ................................................ 183
8.11 Elementary Slip Line Theory f o r Plane Plastic Strain .................... 184

Chapter 9 VISCOELASTICITY
9.1 Linear Viscoelastic Behavior ............................................ 196
9.2 Siniple Viscoelastic Models ............................................ 196
9.3 .
Generalized Models Linear Differential Operator Equation .............. 198
9.4 Creep and Relaxation .................................................. 199
9.5 Creep Function . Relaxation Function . Hereditary lntegrals . . . . . . . . . . . . . . 200
9.6 Complex Moduli and Compliances ...................................... 202
9.7 Three Dimensional Theory .............................................. 203
9.8 .
Viscoelastic Stress Analysis Correspondence Principle .................. 204
Chapter 1

Mathematical Foundations

1.1 TENSORS AND CONTINUUM MECHANICS


Continuum mechanics deals with physical quantities which a r e independent of any
particular coordinate system that may be used to describe them. A t the same time, these
physicaI quantities a r e very often specified most conveniently by referring to a n appropriate
system of coordinates. Mathematically, such quantities a r e represented by tensors.
As a mathematical entity, a tensor has a n existence independent of any coordinate
system. Yet i t may be specified in a particular coordinate system by a certain set of
quantities, known a s its components. Specifying the components of a tensor in one
coordinate system determines the components in any other system. Indeed, the law o f
transformation of the components of a tensor is used here as a means f o r definicg the
tensor. Precise statements of the definitions of various kinds of tensors a r e given a t the
point of their introduction in the material that follows.
The physical laws of continuum mechanics a r e expressed by tensor equations. Because
tensor transformations a r e linear and homogeneous, such tensor equations, if they are valid
in one coordinate system, a r e valid in any other coordinate system. This invarz'unce of
tensor equations under a coordinate transformation is one of the principal reasons f o r the
usefulness of tensor methods in continuum mechanics.

1.2 GENERAL TENSORS. CARTESIAN TENSORS. TENSOR RANK.


In dealing with general coordinate transformations between arbitrary curvilinear
coordinate systems, the tensors defined a r e known a s general tensors. When attention is
restricted to transformations from one homogeneous coordinate system t o another, the
tensors involved are referred to a s Cartesian tensors. Since much of the theory of con-
tinuum mechanics may be developed in terms of Cartesian tensors, the word "tensor" in
this book means "Cartesian tensor" unless specifically stated otherwise.
Tensors may be classified by r a n k , o r order, according to the particular form of the
transformation law they obey. This same classification is also reflected in the number of
components a given tensor possesses in an n-dimensional space. Thus in a three-dimensional
Euclidean space such as ordinary physical space, the number of components of a tensor is
3N, where N is the order of the tensor. Accordingly a tensor of order zero is specified in
any coordinate system in three-dimensional space by one component. Tensors of order
zero are called scalars. Physical quantities having magnitude only a r e represented by
scalars. Tensors of order one have three coordinate components in physicaI space and a r e
known as vectors. Quantities possessing both magnitude and direction a r e represented by
vectors. Second-order tensors correspond to dyadics. Severa1 important quantities in con-
tinuum mechanics a r e represented by tensors of rank two. Higher order tensors such as
triudics, or tensors of order three, and tetradics, o r tensors of order four a r e also defined
and appear often in the mathematics of continuum mechanics.
2 MATHEMATICAL FOUNDATIONS [CHAP. 1

1.3 VECTORS AND SCALARS


Certain physical quantities, such as force and velocity, which possess both magnitude
and direction, may be represented in a three-dimensional space by directed line segments
that obey the parallelogram law of addition. Such directed line segments are the geometrical
representations of first-order tensors and are called vectors. Pictorially, a vector is simply
an arrow pointing in the appropriate direction and having a length proportional to the mag-
nitude of the vector. Equal vectors have the same direction and equal magnitudes. A unit
vector is a vector of unit length. The nu11 or zero vector is one having zero length and an
unspecified direction. The negative of a vector is that vector having the same magnitude
but opposite direction.
Those physical quantities, such as mass and energy, which possess magnitude only are
represented by tensors of order zero which are called scalars.
In the symbolic, or Gibbs notation, vectors are designated by bold-faced letters such a s
a, b, etc. Scalars are denoted by italic letters such as a, b, A, etc. Unit vectors are further
distinguished by a caret placed over the bold-faced letter. In Fig. 1-1, arbitrary vectors a
and b are shown along with the unit vector ê and the pair of equal vectors c and d.

Fig. 1-1

The magnitude of an arbitrary vector a is written simply as a, or for emphasis it may


be denoted by the vector symbol between vertical bars as Ia].

1.4 VECTOR ADDITION. MULTIPLICATION OF A VECTOR BY A SCALAR


Vector addition obeys the parallelogram law, which defines the vector sum of two vectors
as the diagonal of a parallelogram having the component vectors as adjacent sides. This
law for vector addition is equivalent to the triangle rzde which defines the sum of two vectors
as the vector extending from the tail of the first to the head of the second when the summed
vectors are adjoined head to tail. The graphical construction for the addition of a and b
by the parallelogram law is shown in Fig. 1-2(a). Algebraically, the addition process is
expressed by the vector equation
a+b = b f a = c (1.1)
Vector subtraction is accomplished by addition of the negative vector as shown, for
example, in Fig. 1-2(b) where the triangle rule is used. Thus
a-b = -b+a = d (1-2)
The operations of vector addition and subtraction are commutative and associative as
illustrated in Fig. 1-2(c), for which the appropriate equations are
(a+b) + g = a + (b+g) = h (1.3)

Fig. 1-2
CHAP. 11 MATHEMATICAL FOUNDATIONS 3

Multiplication of a vector by a scalar produces in general a new vector having the same
direction a s the original but a different length. Exceptions are multiplication by zero to
produce the nu11 vector, and multiplication by unity which does not change a vector. Multi-
plication of the vector b by the scalar m results in one of the three possible cases shown ip
Fig. 1-3, depending upon the numerical value of m.

O<m<l
Fig. 1-3

Multiplication of a vector by a scalar is a,ssociative and distributive. Thus


m(nb) = (mn)b = n(mb) (1.4)

In the important case of a vector multiplied by the reciproca1 of its magnitude, the
result is a unit vector in the direction of the original vector. This relationship is expressed
by the equation A

b = blb (1.7)

1.5 DOT AND CROSS PRODUCTS OF VECTORS


The dot or scalar product of two vectors a and b is the scalar
= a - b = boa = a b c o s 8
in which '6 is the smaller angle between the two vectors a s shown in Fig. 1-4(a). The dot
product of a with a unit vector ê gives the projection of a in the direction of ê.

Fig. 1-4

The cross or vector product of a into b is the vector v given by

in which O is the angle less than 180" between the vectors a and b, and ê is a unit vector
perpendicular to their plane such t h a t a right-handed rotation about ê through the angle
0 carries a into b. The magnitude of v is equal to the area of the parallelogram having
a and b as adjacent sides, shown shade'd in Fig. 1-4(b). The cross product is not commutative.
4 MATHEMATICAL FOUNDATIONS [CHAP. 1

The scalar triple product is a dot product of two vectors, one of which is a cross product.
a.(bxc) = ( a x b ) * c= a.(bxc)= h (1.10)
As indicated by (1.10) the dot and cross operation may be interchanged in this product.
Also, since the cross operation must be carried out first, the parentheses are unnecessary
and may be deleted as shown. This product is sometimes written [abc] and called the box
product. The magnitude h of the scalar triple product is equal to the volume of the
parallelepiped having a, b, c as coterminous edges.
The vector triple product is a cross product of two vectors, one of which is itself a
cross product. The following identity is frequently useful in expressing the product of a
crossed into b x c.
a x (b x c) = (a c)b - (a b)c = w (1.11)
From (1.11), the product vector w is observed to lie in the plane of b and c.

1.6 DYADS AND DYADICS


The indeterminate vector product of a and b, defined by writing the vectors in juxtaposi-
tion as ab is called a dyad. The indeterminate product is not in general commutative, i.e.
ab # ba. The first vector in a dyad is known as the antecedent, the second is called the
consequent. A dyadic D corresponds to a tensor of order two and may always be represented
as a finite sum of dyads
D = albi azbz + +
.. a ~ b ~ + (1.12)

which is, however, never unique. In symbolic notation, dyadics are denoted by bold-faced
sans-serif letters as above.
If in each dyad of (1.12) the antecedents and consequents are interchanged, the resulting
dyadic is called the conjugate dyadic of D and is written
Dc = blal + bzaz + +b ~ a ~ (1 .I 3)

If each dyad of D in (1.12) is replaced by the dot product of the two vectors, the result is a
scalar known as the scalar o f t h e dyadic D and is written
Ds = a i - b l + az-bz + +a ~ * b ~ (1.l4)
If each dyad of D in (1.12) is replaced by the cross product of the two vectors, the result is
called the vector o f t h e dyadic D and is written
Dv = a 1 x b l + a z x b z + +a,vXb~ (1.15)
I t can be shown that Dc, D, and Dv are independent of the representation (1.12).
The indeterminate vector product obeys the distributive laws
+
a(b c) = ab + ac (1.16)
(a + b)c = ac + bc (1.l7)
(a + b)(c + d) = ac + ad + bc + bd (1.18)
and if h and p are any scalars,
(h + p)ab = Aab + pab (1.19)
(ha)b = a(hb) = hab (1 .20)
CHAP. 11 MATHEMATICAL FOUNDATIONS 5

If v is any vector, the dot products v D and D v are the vectors defined respectively by
v ' D = (v.a1)bl+(v.az)bz+ + ( v * ~ N ) ~ N= i.i (1.21)

In (1.21) D is called the postfactor, and in (1.22) it is called the prefactor. Two dyadics D
and E are equal if and only if for every vector v, either
v * D = v.E or D.v = E-v (1.23)
The unit dyadic, or idemfactor I, is the dyadic which can be represented as
+
I = êl& i%, êS3 + (1-24)
where 21, ê2, ê3 constitute any orthonormal basis for three-dimensional Euclidean space
(see Section 1.7). The dyadic I is characterized by the property
1.v = v - I = v (1.25)
for a11 vectors v.
The cross products v x D and D x v are the dyadics defined respectively by
+
v x D = (v X a1)bl (vX az)bz + +
(vX a ~ ) =b ~F (1.26)
DXv = al(b1 X v) + az(bz X v) + +a ~ ( Xbv)~ = G (1.27)

The dot product of the dyads ab and cd is the dyad defined by


a b *cd = (b. c)ad (1.28)
From (1.28), the dot product of any two dyadics D and E is the dyadic
D E = (albl+ azbz + . - + aNb~) (cidi + czdz + .- +C N ~ N )
= (bi ci)a1dl + (bi cz)ald~+ . . ( b ~ C N ) ~ N ~ N = G (1-29)
The dyadics D and E are said to be reciproca1 of each other if
E-D = D-E = I (130)
Fór reciprocal dyadics, the notation E = D-I and D = E-' is often used.
Double dot and cross products are also defined for the dyads ab and cd as follows,
ab : cd = (a c)(b d) = A, a scalar (1.31)
ab cd = (a x c)(b d) = h, a vector (1.32)
ab cd = (a. c)(b x d) = g, a vector (1.33)
ab c cd = (a x c)(bx d) = uw, a dyad (1.34)

From these definitions, double dot and cross products of dyadics may be readily developed.
Also, some authors use the double dot product defined by
a b - . c d = (b*c)(a.d) = A, ascalar (1.35)
A dyadic D is said to be self-conjugate, or symmetric, if
D = D,
and anti-self-conjugate, or anti-symmetric, if

Every dyadic may be expressed uniquely as the sum of a symmetric and anti-symmetric
dyadic. For the arbitrary dyadic D the decomposition is
D = ?j(D + Dc) + ?j(D- Dc) = G + H (1.38)
6 MATHEMATICAL FOUNDATIONS [CHAP. 1

for which G, = &(D, + (D,),) = +(D, + D) = G (symmetric) (1.39)

and H, = *(Dc - (D,),) = +(D, - D) = -H (anti-symmetric) (1.40)

Uniqueness is established by assuming a second decomposition, D = G* + H*. Then

and the conjugate of this equation is


G*-H* = G-H
Adding and subtracting (1.41) and (1.42) in turn yields respectively the desired equalities,
G* = G and H* = H.

1.7 COORDINATE SYSTEMS. BASE VECTORS. UNIT VECTOR TRIADS


A vector may be defined with respect to a particular coordinate system by specifying
the components of the vector in that system. The choice of coordinate system is arbitrary,
but in certain situations a particular choice may be advantageous. The reference system
of coordinate axes provides units for measuring vector magnitudes and assigns directions
in space by which the orientation of vectors may be determined.
The well-known rectangular Cartesian coordi-
nate system is often represented by the mutually
perpendicular axes, Oxyz shown in Fig. 1-5. Any
vector v in this system may be expressed as a
linear combination of three arbitrary, nonzero,
noncoplanar vectors of the system, which are
called base vectors. For base vectors a, b, c and
suitably chosen scalar coefficients h, p, V the vector
v is given by
V = ha pb vc + + (1.43)
Base vectors are by hypothesis linearly independ-
ent, i.e. the equation
+ +
ha pb vc = O (1.44)
is satisfied only if h = p = V = O. A set of base
vectors in a given coordinate system is said to
constitute a bash for that system. Fig. 1-5
The most frequent choice of base vectors for the rectangular Cartesian system is the
A A A

set of unit vectors i, j, k along the coordinate axes as shown in Fig. 1-5. These base vectors
constitute a right-handed unit vector triad, for which
$ $ A A A C A A h
i x j = k, j x k = i, kxi = j (1.45)
A A A A A A
and i.; j - j = k.k = 1

Such a set of base vectors is often called an orthonormal bash.


A A A
In terms of the unit triad i, j, k, the vector v shown in Fig. 1-6 below may be expressed by

in which the Cartesian components


CHAP. 11 MATHEMATICAL FOUNDATIONS 7

are the projections of v onto the coordinate axes.


The unit vector in the direction of v is given ac-
cording to (1.7) by
ê" = VIU
= (COSa)? + (COS P ) + (COS y ) k (1.48)
Since v is arbitrary, i t follows that any unit vec-
tor will have the direction cosines of that vector
as its Cartesian components.
In Cartesian component form the dot product
of a and b is given by
A A A

a*b = (a,?+a,~+a,k)*(b,i+b,j+b~)
+ +
= axbx ayby a,b, (1.49) Fig. 1-6

For the same two vectors, the cross product a x b is


A

+
a x b = (ayb,- a,by)i (a,b, - a,b,) j
A

+ (a,b, - aybx)k
A

This result is often presented in the determinant form

in which the elements a r e treated as ordinary numbers. The triple scalar product may also
be represented in component form by the determinant
a, a, a,
[abc] = bx b~ b~
C, Cy c,
I n Cartesian component form, the dyad ab is given by
c
+
ab = (a,i +a,; a,k)(b,? b,; b,k) + +
A A
= a,b,i i+ axbyij + a,b,i k
A A
9"

+ ayb,S$ + a , b y +~ ayb,$k
~
+ a,b,kT + a,bykj + a,b,kk
A A

(1.53)
Because of the nine terms involved, (1.53) is known a s the nonion form of the dyad ab.
It is possible to put any dyadic into nonion form. The nonion form of the idemfactor in
A A A
terms of the unit triad i, j , k is given by
A A A A A A
I = ii+jj+kk (1.54)
In addition to the rectangular Cartesian coordinate system already discussed, curvi-
linear coordinate systems such as the cylindrical (R, O, z ) and spherical ( r ,O, systems C$)

shown in Fig. 1-7 below a r e also widely used. Unit triads ( $ R , ês, ê,) and (ê,, 60,ê+) of base
vectors illustrated in the figure a r e associated with these systems. However, the base
vectors here do not all have fixed directions and a r e therefore, in general, functions of
position.
8 MATHEMATICAL FOUNDATIONS [CHAP. 1

(a) Cylindrical (b) Spherical


Fig. 1-7

1.8 LINEAR VECTOR FUNCTIONS. DYADICS AS LINEAR VECTOR OPERATORS


A vector a is said to be a function of a second vector b if a is determined whenever
b is given. This functional relationship is expressed by the equation
a = f(b) (1.55)
The function f is said to be linear when the conditions
f(b + c) = f(b) + f(c) (1.56)

are satisfied for a11 vectors b and c, and f o r any scalar A.


Writing b in Cartesian component form, equation (1.55) becomes
c
a = f(b,i b,; b,k) + + (1.58)
which, if f is linear, may be written
a = b,f(?) + b,f(;) + b,f(k) (1.59)
A A
.?
I n (1.59) let f ( i ) = u , f ( j ) = v , f ( k ) = w, so that now

which is recognized a s a dyadic-vector dot product and may be written


a = D-b (1.61)
A

+ +
A A
where D = u i v j wk. This demonstrates that any linear vector function f may be
expressed a s a dyadic-vector product. In (1.61) the dyadic D serves a s a linear vector
operator which operates on the argument vector b to produce the image vector a.

1.9 INDICIAL NOTATION. RANGE AND SUMMATION CONVENTIONS


The components of a tensor of any order, and indeed the tensor itself, may be represented
clearly and concisely by the use of the indicial notation. In this notation, letter indices,
either subscripts or superscripts, are appended to the generic o r kernel letter representing
the tensor quantity of interest. Typical examples illustrating use of indices are the tensor
symbols
ai, bj, Tij, pij, ' i j k , R'>q
CHAP. 11 MATHEMATICAL FOUNDATIONS 9

In the "mixed" form, where both subscripts and superscripts appear, the dot shows that j
is the second index.
Under the rules of indicial notation, a letter index may occur either once or twice in a
given term. When an index occurs unrepeated in a term, that index is understood to take
on the values 1,2,. . .,N where N is a specified integer that determines the range of the
index. Unrepeated indices are known as free indices. The tensorial rank of a given term
is equal to the number of free indices appearing in that term. Also, correctly written
tensor equations have the same letters as free indices in every term.
When an index appears twice in a term, that index is understood to take on a11 the
values of its range, and the resulting terms summed. In this so-called summation conven-
tion, repeated indices are often referred to as dummy indices, since their replacement by
any other letter not appearing as a free index does not change the meaning of the term in
which they occur. In general, no index occurs more than twice in a properly written term.
If it is absolutely necessary to use some index more than twice to satisfactorily express a
certain quantity, the summation convention must be suspended.
The number and location of the free indices reveal directly the exact tensorial character
of the quantity expressed in the indicial notation. Tensors of first order are denoted by
kernel letters bearing one free index. Thus the arbitrary vector a is represented by a symbol
having a single subscript or superscript, i.e. in one or the other of the two forms,

The following terms, having only one free index, are also recognized as first-order tensor
quantities:
aijbj, Fikk, rijkUjZ)k

Second-order tensors are denoted by symbols having two free indices. Thus the arbitrary
dyadic D will appear in one of the three possible f o m s

In the "mixed" form, the dot shows that j is the second index. Second-order tensor
quantities may also appear in various f o m s as, for example,
Aijip, B?.jk, 8ifikvk
By a logical continuation of the above scheme, third-order tensors are expressed by symbols
with three free indices. Also, a symbol such as X which has no indices attached, represents
a scalar, or tensor of zero order.
In ordinary physical space a basis is composed of three, noncoplanar vectors, and so
any vector in this space is completely specified by its three components. Therefore the
range on the index of ai, which represents a vector in physical three-space, is 1,2,3.
Accordingly the symbol aiis understood to represent the three components a1,az,a3. Also,
ai is sometimes interpreted to represent the ith component of the vector or indeed to rep-
resent the vector itself. For a range of three on both indices, the symbol Aij represents
nine components (of the second-order tensor (dyadic) A). The tensor Aij is often presented
explicitly by giving the nine components in a square array enclosed by large parentheses as
Ai1 A12 A13

Aij = ( 1.62)
10 MATHEMATICAL FOUNDATIONS [CHAP. 1

I n the same way, the components of a first-order tensor (vector) in three-space may be
displayed explicitly by a row or column arrangement of the form

ai = (ai, az, as) or ai = (1.63)


a3
I n general, for a range of N, a n nth order tensor will have Nn components.
The usefulness of the indicial notation in presenting systems of equations in compact
form is illustrated by the following two typical examples. F o r a range of three on both
i and j the indicial equation
Xi = C i j 2 j (1.64)
represents in expanded form the three equations

F o r a range of two on i and j, the indicial equation .


Aij = BipCjqDpq (1.66)
represents, in expanded form, the four equations
Ali = BiiCiiDii + BliC12D12 + B12CilD21 + B12Ci2D22
Ai2 = BiiC2iDii + BiiC22Di2 + BIZCZID~I + Bi2CmD22 (1.67)
A21 + B ~ I C I ~+DB22CiiD21
= BSICIIDI~ I~ + Bz2C12D22
F o r a range of three on both i and j, (1.66) would represent nine equations, each having
nine terms on the right-hand side.

1.10 SUMMATION CONVENTION USED WITH SYMBOLIC NOTATION


The summation convention is very often em-
ployed in connection with the representation of
vectors and tensors by indexed base vectors
written in the symbolic notation. Thus if the
rectangular Cartesian axes and unit base vectors
of Fig. 1-5 are relabeled a s shown by Fig. 1-8,
the arbitrary vector v may be written
V = viêi + v2ê2 + v3ê3 (1.68)
in which vi, v2, v3 are the rectangular Cartesian
components of v. Applying the summation con-
vention to (1.68), the equation may be written in
the abbreviated form
A
V = Viei (1.69)
where i is a summed index. The notation here is
essentially symbolic, but with the added feature
of the summation convention. In such a "com-
bination" style of notation, tensor character is
not given by the free indices rule as it is in true
indicial notation. Fig. 1-8
CHAP. I] MATHEMATICAL FOUNDATIONS 11

Second-order tensors may also be represented by summation on indexed base vectors.


Accordingly the dyad ab given in nonion form by (1.53) may be written
ab = (aiêi)(b$j) = aibjêiêj (1.70)
I t is essential that the sequence of the base vectors be preserved in this expression. In
similar fashion, the nonion form of the arbitrary dyadic D may be expressed in compact
notation by

1.11 COORDINATE TRANSFORMATIONS. GENERAL TENSORS


Let xi represent the arbitrary system of coordinates xl, x2,x3 in a three-dimensional
Euclidean space, and let Bi represent any other coordinate system B1, 02,O3 in the same
space. Here the numerical superscripts are labels and not exponents. Powers of x may
be expressed by use of parentheses as in ( x ) ~or ( x ) ~ . The letter superscripts are indices
a s already noted. The coordinate transformation equations

assign to any point (xl, x2,x3) in the xi system a new set of coordinates (O1, 02,B3) in the Oi
system. The functions Bi relating the two sets of variables (coordinates) are assumed to
be single-valued, continuous, differentiable functions. The determinant

or, in compact form,

is called the Jacobian of the transformation. If the Jacobian does not vanish, (1.72)
possesses a unique inverse set of the form

The coordinate systems represented by xi and Bi in (1.72) and (1.75) are completely general
and may be any curvilinear or Cartesian systems.
From (l.72), the differential vector dOi is given by

This equation is a prototype of the equation which defines the class of tensors known as
contravariant vectors. I n general, a set of quantities bi associated with a point P are said
to be the components of a c o n t r a v a ~ a n ttensor of order one if they transform, under a
coordinate transformation, according to the equation

where the partia1 derivatives are evaluated at P. In (1.77), bj are the components of the
tensor in the xj coordinate system, while bti are the components in the Oi system. In general
12 MATHEMATICAL FOUNDATIONS [CHAP. 1

tensor theory, contravariant tensors are recognized by the use of superscripts as indices.
I t is for this reason that the coordinates are labeled xi here rather than Xi, but it must be
noted that it is only the differentials dxi, and not the coordinates themselves, which have
tensor character.
By a logical extension of the tensor concept expressed in (l.77), the definition of con-
travariant tensors of order two requires the tensor components to obey the transformation
law

Contravariant tensors of third, fourth and higher orders are defined in a similar manner.
The word contravariant is used above to distinguish that class of tensors from the
class known as covariant tensors. In general tensor theory, covariant tensors are recognized
by the use of subscripts as indices. The prototype of the covariant vector is the partia1
derivative of a scalar function of the coordinates. Thus if + = +(xl, x2,x3) is such a function,

I n general, a set of quantities bi are said to be the components of a covariant ,temor of


order one if they transform according to the equation

In (1.80), b: are the covariant components in the ei system, bi the components in the xi
system. Second-order covariant tensors obey the transformation law

Covariant tensors of higher order and mixed tensors, such as

are defined in the obvious way.

1.12 THE METRIC TENSOR. CARTESIAN TENSORS


Let xi represent a system of rectangular Cartesian coordinates in a Euclidean three-
space, and let 8' represent any system of rectangular or curvilinear coordinates (e.g. cylindri-
cal or spherical coordinates) in the same space. The vector x having Cartesian components
xi is called the position vector of the arbitrary point P(xl, x2,x3) referred to the rectangular
Cartesian axes. The square of the differential element of distance between neighboring
points P(x) and Q(x +dx) is given by

From the coordinate transformation


xi = xi(ol,02,e3)
relating the systems, the distance differential is
CHAP. 11 MATHEMATICAL FOUNDATIONS 13

and therefore (1.83) becomes

where the second-order tensor g,, = (dxi/dOp)(dxi/d$q) is called the metric tensor, or funda-
mental tensor of the space. If Oi represents a rectangular Cartesian system, say the x"
system, then

where S, is the Kronecker delta (see Section 1.13) defined by S,, = O if p # q and S,, = 1
if p = q .
Any system of coordinates for which the squared differential element of distance takes
the form of (1.83) is called a system of homogeneous coordinates. Coordinate transforma-
tions between systems of homogeneous coordinates are orthogonal transformations, and
when attention is restricted to such transformations, the tensors so defined are called
Cartesian tensors. In particular, this is the case for transformation laws between systems
of rectangular Cartesian coordinates with a common origin. For Cartesian tensors there
is no distinction between contravariant and covariant components apd therefore it is cus-
tomary to use subscripts exclusively in expressions representing Cartesian tensors. As
will be shown next, in the transformation laws defining Cartesian tensors, the partia1
derivatives appearing in general tensor definitions, such as (1.80) and (1.81), are replaced
by constants.

1.13. TRANSFORMATION LAWS FOR CARTESIAN TENSORS.


THE KRONECKER DELTA. ORTHOGONALITY CONDITIONS
Let the axes Oxixzxs and Ox;xtx$ represent
two rectangular Cartesian coordinate systems
with a common origin a t an arbitrary point O
as shown in Fig. 1-9. The primed system may be
imagined to be obtained from the unprimed by
a rotation of the axes about the origin, or by a
reflection of axes in one of the coordinate planes,
or by a combination of these. If the symbol aij
denotes the cosine of the angle between the ith
primed and jth unprimed coordinate axes, i.e.
aij = cos (xl, xj), the relative orientation of the
individual axes of each system with respect to the
other is conveniently given by the table

Fig. 1-9

or alternatively by the transformation tensor


14 MATHEMATICAL FOUNDATIONS [CHAP. 1

From this definition of aij, the unit vector along the x: axis is given according to (1.48)
and the summation convention by

An obvious generalization of this equation gives the arbitrary unit base vector as

In component form, the arbitrary vector v shown in Fig. 1-9 may be expresse(; in the
unprimed system by the equation
v = ,ê,.
2). (1.90)
and in the primed system by
V = vi ei (1.91)

Replacing in (1.91) by its equivalent form (1.89) yields the result

Comparing (1.92) with (1.90) reveals that the vector components in the primed and unprimed
systems are related by the equations
v 3. -
- a,.
%,Vi
I
(1.93)

The expression (1.93) is the transformation law for first-order Cartesian tensors, and as
such is seen to be a special case of the general form of first-order tensor transformations,
expressed by (1.80) and (1.77). By interchanging the roles of the primed and unprimed
base vectors in the above development, the inverse of (1.93) is found to be

I t is important to note that in (1.93) the free index on aij appears as the second index. In
(1.94), however, the free index appears as the first index.
By an appropriate choice of dummy indices, (1.93) and (1.94) may be combined to pro-
duce the equation
v j = a..
tjazkVk
. (1.95)

Since the vector v is arbitrary, (1.95) must reduce to the identity vj = vj. Therefore the
coefficient aijaik, whose value depends upon the subscripts j and k , must equal 1 or O
according to whether the numerical values of j and Ic are the same or different. The
Kronecker delta. defined b'r

may be used to represent quantities such as aijaik. Thus with the help of the Kronecker delta
the conditions on the coefficient in (1.95) may be written

In expanded form, (1.97) consists of nine equations which are known as the orthogonality
or orthonormality conditions on the direction cosines aij. Finally, (1.93) and (1.94) may also
be combined to produce vi = &ijakjvk from which the orthogonality conditions appear in the
alternative form
aijakj = s i k (1.98)
A linear transformation such as (1.93) or (1.94), whose coefficients satisfy (1.97) or (1.98),
is said to be an orthogonal transformation. Coordinate axes rotations and reflections of
the axes in a coordinate plane both Iead to orthogonal transformations.
CHAP. 11 MATHEMATICAL FOUNDATIONS 15

The Kronecker delta is sometimes called the substitution operator, since, for example,
8ijbj = 8iibi + 8i2b2 f 8i3b3 = bi
and, likewise,
8ijFik = 8ijFik +8 2 j F 2 k $. 83jF3k = F j k

It is clear from this property that the Kronecker delta is the indicial counterpart to the
symbolic idemfactor I, which is given by (1.54).
According to the transformation law (1.94), the dyad UiVj has components in the primed
coordinate system given by
U I V ~ = ( U i p ~ ) ( ~ j q=~ qUipUjqUpVq
) (I.I 01)
In an obvious generalization of (1.101), any second-order Cartesian tensor Tij obeys the
transf ormation Iaw
TC = aipajqTpq
With the help of the orthogonality conditions i t is a simple calculation to invert (1.102),
thereby giving the transformation rule from primed components to unprimed components:

The transformation laws for first and second-order Cartesian tensors generalize for an
Nth order Cartesian tensor to
T~&.. . = aipajqalcm. . . Tpqm... (IJ0-4)

1.14 ADDITION OF CARTESIAN TENSORS. MULTIPLICATION BY A SCALAR


Cartesian tensors of the same order may be added (or subtracted) component by com-
ponent in accordance with the rule

The sum is a tensor of the same order as those added. Note that like indices appear in the
same sequence in each term.
Multiplication of every component of a tensor by a given scalar produces a new tensor of
the same order. For the scalar multiplier A, typical examples written in both indicial
and symbolic notation are
bi = xai or b = Xa (1.106)

1.15 TENSOR MULTIPLICATION


The outer product of two tensors of arbitrary order is the tensor whose components
are formed by multiplying each component of one of the tensors by every component of the
other. This process produces a tensor having an order which is the sum of the orders of
the factor tensors. Typical examples of outer products are
(a) aibj = Tij (C) DijTkrn = Qiikrn

(b) viFjk = Nijk (d) €ijkVm = @ijkm

As indicated by the above examples, outer products are formed by simply setting down the
factor tensors in juxtaposition. (Note that a dyad is formed from two vectors by this very
procedure.)
16 MATHEMATICAL FOUNDATIONS [CHAP. 1

Contraction of a tensor with respect to two free indices is the operation of assigning
to both indices the same letter subscript, thereby changing these indices to dummy indices.
Contraction produces a tensor having an order two less than the original. Typical examples
of contraction are the following.
( a ) Contractions of Tij and uivj
+ +
Tii = Til TZZ T33
UiVi = UiVi+ + U2V2 U3V3

( b ) Contractions of Eijak
E..
va3. = bi
Eijai = cj
Eiiak = dk
(C) Contractions of EijFkm
EijFim = Gim EijFkk = Pij
EijFki = Hik EijFjm = Qim
EiiFkm = Kkm EijFkj = Rik
An Znner product of two tensors is the result of a contraction, involving one index from
each tensor, performed on the outer product of the two tensors. Severa1 inner products
important to continuum mechanics are listed here for reference, in both the indicial and
symbolic notations.
Outer Product Inner Product
Indicial Notation Symbolic Notation
1. aibj aibi a*b
2. aiEjk aiEik = fk a.E = f
aiEji = hj E-a = h
3. Eij Fkm EijFjm = Gim E.F = G
4. Eij Ekm Eij Ejm = Bim E . E = (E)2
Multiple contractions of fourth-order and higher tensors are sometimes useful. Two
such examples are
1. Eij Fkm contracted to Eij Fij, or E : F
2. EijEkmE,, contracted to Eij Ejm Emq, or ( E ) 3

1.16 VECTOR CROSS PRODUCT. PERMUTATION SYMBOL. DUAL VECTORS


In order to express the cross product a x b in the indicial notation, the third-order tensor
eij,, known a s the permutation symbol or alternating tensor, must be introduced. This
useful tensor is defined by
1 if the values of i , j, k are an even permutation of 1 , 2 , 3 (i.e. if
I-1
they appear in sequence as in the arrangement 1 2 3 1 2 ) .
if the values of i , j , k are an odd permutation of 1 , 2 , 3 (i.e. if
'ijk - they appear in sequence as in the arrangement 3 2 13 2).

I O if the values of i , j, k are not a permutation of 1 , 2 , 3 (i.e. if


two or more of the indices have the same value).
CHAP. 11 MATHEMATICAL FOUNDATIONS 17

From this definition, the cross product a x b = c is written in indicial notation by

Using this relationship, the box product a x b . c = h may be written

Since the same box product is given in the form of a determinant by (1.52), it is not sur-
prising that the permutation symbol is frequently used to express the value of a 3 X 3
determinant.
I t is worthwhile to note that cijk obeys the tensor transformation law for third order
Cartesian tensors as long as the transformation is a proper one (det aij = 1) such as arises
from a rotation of axes. If the transformation is improper (det aij = -I), e.g. a reflection
in one of the coordinate planes whereby a right-handed coordinate system is transformed
into a left-handed one, a minus sign must be inserted into the transformation law for cijk.
Such tensors are called pseudo-tensors.
The dual vector of an arbitrary second-order Cartesian tensor Tij is defined by
2 - c <. . ~ T.
v. - k ik (1.110)
which is observed to be the indicial equivalent of T,, the "vector of the dyadic Tu, as defined
by (1.15).

1.17 MATRICES. MATRIX REPRESENTATION OF CARTESIAN TENSORS


A rectangular array of elements, enclosed by square brackets and subject to certain laws
of combination, is called a matrix. An M x N matrix is one having M (horizontal) rows
and N (vertical) columns of elements. In the symbol Aij, used to represent the typical
element of a matrix, the first subscript denotes the row, the second subscript the column
occupied by the element. The matrix itself is designated by enclosing the typical element
symbol in square brackets, or alternatively, by the kernel letter of the matrix in script.
For example, the M x N matrix 4 ,or [Aij] is the array given by

A matrix for which M = N, is called a square matrix. A 1x N matrix, written [alk],


is called a row matrix. An M x 1matrix, written [akl], is called a column matrix. A matrix
having only zeros as elements is called the zero matrix. A square matrix with zeros every-
where except on the main diagonal (from A11 to ANN)is called a diagonal matrix. If the
nonzero elements of a diagonal matrix are a11 unity, the matrix is called the unit or identity
mat.ix. The N X M matrix CAT, formed by interchanging rows and columns of the M x N
matrix 4 , is called the transpose matrix of 4.
Matrices having the same number of rows and columns may be added (or subtracted)
element by element. Multiplication of the matrix [Aij] by a scalar h results in the matrix
[AAij]. The product of two matrices, cA%, is defined only if the matrices are conformable,
i.e. if the prefactor matrix cA has the same number of columns as the postfactor matrix
23 has rows. The product of an M X P matrix multiplied into a P X N matrix is an M X N
matrix. Matrix multiplication is usually denoted by simply setting down the matrix
symbols in juxtaposition as in
18 MATHEMATICAL FOUNDATIONS [CHAP. 1

Matrix multiplication is not, in general, commutative: cA% z LBd


.
A square matrix cA whose determinant \Aij\is zero is called a singular matriz. The
cofactor of the element Aij of the square matrix 4 , denoted here by AT~,is defined by
* =
Aij (-l)i+jMij (1.113)
in which Mij is the minor of Aij; i.e. the determinant of the square array remaining after the
row and column of Aij are deleted. The adjoint matrix of O4 is obtained by replacing each
element by its cofactor and then interchanging rows and columns. If a square matrix
e4 = [Aij] is non-singular, i t possesses a unique inverse matrix 4-' which is defined as
the adjoint matrix of O4 divided by the determinant of 04. Thus

From the inverse matrix definition (1.114) it may be shown that

where 4 is the identity matrix, having ones on the principal diagonal and zeros elsewhere,
and so named because of the property
4 c A = CAJ = (1.116)
I t is clear, of course, that 4 is the matrix representation of 6,,, the Kronecker delta, and of I,
the unit dyadic. Any matrix cA for which the condition cAT = cA-I is satisfied is called
an orthogonal matrix. Accordingly, if O4 is orthogonal,

As suggested by the fact that any dyadic may be expressed in the nonion form (1.53),
and, equivalently, since the components of a second-order tensor may be displayed in the
square array (1.62), it proves extremely useful to represent second-order tensors (dyadics)
by square, 3 x 3 matrices. A first-order tensor (vector) may be represented by either a
1 x 3 row matrix, or by a 3 x 1 column matrix. Although every Cartesian tensor of order
two or less (dyadics, vectors, scalars) may be represented by a matrix, not every matrix
represents a tensor.
If both matrices in the product 4 % = C are 3 x 3 matrices representing second-order
tensors, the multiplication is equivalent to the inner product expressed in indicial notation by

where the range is three. Expansion of (1.I18) duplicates the "row by column" multiplica-
tion of matrices wherein the elements of the ith row of the prefactor matrix are multiplied
in turn by the elements of the kth column of the postfactor matrix, and these products
summed to give the element in the .ith row and kth column of the product matrix. Severa1
such products occur repeatedly in continuum mechanics and are recorded here in the various
notations for reference and comparison.

(a) Vector dot product


a b = b *a = A [aij][bjl] = [A]
CHAP. 11 MATHEMATICAL FOUNDATIONS 19

( b ) Vector-dyadic dot product


a*E = b a& = CB

(c) Dyadic-vector dot product


E*a = c &a = c
E..
tja3. -
- Ci [Eij][ajl] [cii]

1.18 SYMMETRY OF DYADICS, MATRICES AND TENSORS


According to (1.36) (or (1.37)), a dyadic D is said to be symmetric (anti-symmetric) if it is
equal to (the negative of) its conjugate D,. Similarly the second-order tensor Dij is
symmetm'c if D.. - D..
a3 - 3% (1.I22)
and is anti-symmetric, or skew-symmetric, if

Therefore the decomposition of Dij analogous to (1.38) is

or, in an equivalent abbreviated form often employed,

where parentheses around the indices denote the symmetric part of Dii, and square brackets
on the indices denote the anti-symmetric part.
Since the interchange of indices of a second-order tensor is equivalent to the interchange
of rows and columns in its matrix representation, a square matrix cA is symmetric if it is
equal to its transpose CAT. Consequently a symmetric 3 x 3 matrix has only six independent
components as illustrated by
Ali Ai2
(1.126)
A23 A33

An anti-symmetric matrix is one that equals the negative of its transpose. Consequently
a 3 x 3 anti-symmetric matrix CB has zeros on the main diagonal, and therefore only three
independent components as illustrated by
20 MATHEMATICAL FOUNDATIONS [CHAP. 1

Symmetry properties may be extended to tensors of higher order than two. In general,
an arbitrary tensor is said to be symmetric with respect to a pair of indices if the value of
the typical component is unchanged by interchanging these two indices. A tensor is anti-
symmetric in a pair of indices if an interchange of these indices leads to a change of sign
without a change of absolute value in the component. Examples of symmetry properties
in higher-order tensors are
(a) Rijkm = Rikjm (symmetric in k and j )
( b ) cijk = -c kii (anti-symmetric in k and i)
(c) Gijkm = Gjimk (symmetric in i and j; k and m)
(d) Pijk = Pikj = P k j i = pjik (syrnmetric in a11 indices)

1.19 PRINCIPAL VALUES AND PRINCIPAL DIRECTIONS OF SYMMETRIC


SECOND-ORDER TENSORS
In the following analysis, only symmetric tensors with real components are considered.
This simplifies the mathematics somewhat, and since the important tensors of continuum
mechanics are usually symmetric there is little sacrifice in this restriction.
For every symmetric tensor Tij, defined a t some point in space, there is associated with
each direction (specified by the unit normal ni) a t that point, a vector giveii by the inner
p r oduct

Here Tij may be envisioned as a linear vector operator which produces the vector vi conjugate
to the direction ni. If the direction is one for which vi is parallel to ni, the inner product
may be expressed as a scalar multiple of ni. For this case,

and the direction ni is called a principal direction, or principal axis of Tij. With the help
of the identity ni = Sijnj, (1.129) can be put in the form

which represents a system of three equations for the four unknowns, ni and h, associated
with each principal direction. In expanded form, the system to be solved is

Note first that for every A, the trivial solution ni = O satisfies the equations. The purpose
here, however, is to obtain non-trivial solutions. Also, from the homogeneity of the system
(1.131) it follows that no loss of generality is incurred by restricting attention to solutions
for which nini = 1, and this condition is imposed from now on.
For (1.130) or, equivalently, (1.131) to have a non-trivial solution, the determinant of
coefficients must be zero, that is,
]Tij- ASijl = O (1.132)
Expansion of this determinant leads to a cubic polynomial in h, namely,
A3 - ITX' + IITA- IIIT = O (I .133)
which is known as the characteristic equation of Tij, and for which the scalar coefficients,
CHAP. I] MATHEMATICAL FOUNDATIONS 21

IT = Tii = t r Tij (trace of Tij) (1.134)

IIIT = ITijl = det Tij (1.136)

. are called the first, second and third invariants, respectively, of Tij. The three roots of the
cubic (1.133), labeled X(I),h ( ~A(,),
, are called the principal values of Tij. For a symmetric
tensor with real components, the principal values are real; and if these values are distinct,
the three principal directions are mutually orthogonal. When referred to principal axes,
both the tensor array and its matrix appear in diagonal form. Thus

If X(I) = h(z), the tensor has a diagonal form which is independent of the choice of X ( I )
and htz) axes, once the principal axis associated with XG) has been established. If a11
principal values are equal, any direction is a principal direction. If the principal values are
ordered, it is customary to write thern as h ( , ) ,h(II),h(,,,) and to display the ordering as in
Ao, > A,,,, > ~ ( I I I , .
For principal axes labeled OxTx2x8, the transformation from Oxixzx3 axes is given by
the elements of the table
$1 x2 x3

~1 all = n(l) a12= n(1)


2
al, = ni1)

2;
aZ1= n(2) aZ2= n(2) (L*, = n(32)

a31 = ni3) = a,, = n 3 )


E;

in which nlj) are the direction cosines of the jth principal direction.

1.20 POWERS OF SECOND-ORDER TENSORS. HAMILTON-CAYLEY EQUATION


By direct matrix multiplication, the square of the tensor Tij is given as the inner
product TikTkj; the cube as TikTkrnTmj; etc. Therefore with Tii written in the diagonal form
(1.137), the nth power of the tensor is given by

A comparison of (1.135) and (1.137) indicates that Tij and a11 its integer powers have the
same principal axes.
Since each of the principal values satisfies (1.133), and because of the diagonal matrix
form of T ngiven by (1.138), the tensor itself will satisfy (1.133). Thus

in which ,4 is the identity rnatrix. This equation is called the Hamilton-Cayley equation.
Matrix multiplication of each term in (1.139) by I produces the equation,
22 MATHEMATICAL FOUNDATIONS [CHAP. 1

Combining (1.140) and (1.139) by direct substitution,

Continuation of this procedure yields the positive powers of 7 as linear combinations of


T2, T and 8.

1.21 TENSOR FIELDS. DERIVATIVES OF TENSORS


A tensor field assigns a tensor T(x, t) to every pair (x, t) where the position vector x
varies over a particular region of space and t varies over a particular interval of time.
The tensor field is said to be continuous (or differentiable) if the components of T(x, t) are
continuous (or differentiable) functions of x and t. If the components are functions of x
only, the tensor field is said to be steady.
With respect to a rectangular Cartesian coordinate system, for which the position vector
of an arbitrary point is A
x = xiei (1.142)
tensor fields of various orders are represented in indicial and symbolic notation a s follows,
(a) scalar field: + = +(xi, t) or $J = +(x, t) (1.143)
(b) vector field: vi = vi(x, t) or V = V(X,t) (1,144)
(c) second-order tensor field:
Tij = Tij(x, t) or T = T(x, t) (1.145)
Coordinate differentiation of tensor components with respect to xi is expressed by the
differential operator aldxi, or briefly in indicial form by a;, indicating a n operator of tensor
rank one. In symbolic notation, the corresponding symbol is the well-known differential
vector operator V, pronounced de1 and written explicitly

V = êi a
aXi
= eidi
A
(1.I46)
Frequently, partia1 differentiation with respect to the variable xi is represented by the
comma-subscript convention as illustrated by the following examples.

avi aTij
(b) -
axi = vi,i (e) ã2* = Ti!, k

avi a2Tij -
(c)- = vi,j
ax ( f ) a s r d z ,- km

From these examples i t is seen t h a t the operator ai produces a tensor of order one higher
if i remains a free index ((a)and (c) above), and a tensor of order one lower if i becomes
a dummy index ((b) above) in the derivative.
Severa1 important differential operators appear often in continuum mechanics and are
given here f o r reference.

divv = V - v or divt = visi (1.148)

curl v = V x v Or ~ijkajvk= 'ijkvk,i (1.I49)


CHAP. 11 MATHEMATICAL FOUNDATIONS 23

1.22 LINE INTEGRALS. STOKESy THEOREM


In a given region of space the vector function of position, F = F(x), is defined a t every
point of the piecewise smooth curve C shown in Fig. 1-10. If the differential tangent vector
to the curve a t the arbitrary point P is dx, the integral
XF-~X = L ~ F - ~ X (1.151)

taken along the curve from A to B is known as the line .integral of F aIong C. In the indiciaI
notation, (1.151) becomes (z~)s

S,Fidxi J
(zi)* Fidxi (1.I 52)

Fig. 1-10 Fig. 1-11

Stokes' theorem says that the line integral of F taken around a closed'reducible curve
C, as pictured in Fig. 1-11, may be expressed in terms of an integral over any two-sided
surface S which has C as its boundary. Explicitly,

in which 6 is the unit normal on the positive side of S , and dS is the differential element of
surface as shown by the figure. In the indicial notation, (1.153) is written

1.23 THE DIVERGENCE THEOREM OF GAUSS


The divergence theorem of Gauss relates a volume integral to a surface integral. In
its traditional form the theorem says that for the vector field v = v(x),

where 2 is the outward unit normal to the bounding surface S , of the volume V in which
the vector field is defined. . I n the indicial notation, (1.155) is written
i vi,i dV = L vini dS (1.156)

The divergence theorem of Gauss as expressed by (1.156) may be generalized to incor-


porate a tensor field of any order. Thus for the arbitrary tensor field Tijk... the theorem is
written
Tijk . . . p dV = Tijknp dS (1.157)
24 MATHEMATICAL FOUNDATIONS [CHAP. 1

Solved Problems
ALGEBRA OF VECTORS AND DYADICS (Sec. 1.1-1.8)
Determine in rectangular Cartesian form the
unit vector which is (a) parallel to the vector
+
v = 2? 3 3 - 6g, ( b ) along the line joining
points P(1,0,3) and Q(O,2 , l ) .
+
( a ) Ivl = v = 1 / ( 2 ) ~ (3)2 (-6)2 = 7 +
A
v = VIV = (217)i
A
+ (317)j A
- (617)k
A

( b ) The vector extending from P to Q is


A
u = (O-I)?+ (2-O);+ (1-3)k
A A A
= -i+2j -2k
u = d(-1)2 + (2)2+ (-2)2 = 3 Fig. 1-12
Thus
A
u = -(1/3) i
A
+ (213); - (213)k directed from P to Q
or 6 = (1/3)? - (213)3 + ( 2 / 3 ) k directed from Q to P

1.2. Prove that the vector v = a? b + i+


c k is normal
to the plane whose equation is a x by cz = A. + +
Let P ( x l , yl, zl) and Q(x2,y2, z2) be any two points in
Iz
+ +
the plane. Then a s l byl czl = X and ax2 by2 cz2 = + +
and the vector joining these points is u = (x2- x l ) i
A
+
A
+ A
(v2- y l ) j ( 2 , - 2,) k. The projection of v in the direction
of u is
u *-v - 1
- A A

u u
-[(xz-xI)~+ (v2-y1)j
A

+( 2 , - 2,) k ] [a? +
b3 cc] +
U
1 +
= - ( a s 2 by2+ cz2- axl - byl - cz,) = X- -X
= O
U
Z

Since u is any vector in the plane, v is I to the plane. Fig. 1-15

1.3. +
If r = x? + y 3 z c is the vector extending from the origin to the arbitrary point
+ +
P ( x , y, z) and d = a? b j c k is a constant vector, show that (r - d) r = O is the
vector equation of a sphere.
Expanding the indicated dot product,
A A A A A A
(r-d)*r = [(x-a)i+(y-b)j+(z-c)k]*[xi+yj+zk]
= x2+y2+z2-ax-by-cz = O
Adding d2/4 = (a2 + b2 + c2)/4 to each side of this equation gives the desired equation
( x - a/2)2 + ( y - b/2)2 + ( z - c/2)2 = (d/2)2
which is the equation of the sphere centered a t d/2 with radius d/2.

1.4. Prove that [a b X c]r = ( a -r)b x c + (b r)c x a + (c r)a x b.


Consider the product a X [ ( bX c ) X r]. By direct expansion of the cross product in brackets,
a X [ ( bX c ) X r] = a X [ ( b r)c - ( c r)b] = - ( b r)c X a - (c r)a X b .
Also, setting b X c = v ,
a x [ ( b x c ) x r ] = a x ( v x r ) = ( a W r ) b x c- ( a . b x c ) r
CHAP. 11 MATHEMATICAL FOUNDATIONS 25

Thus -(b r)c x a - ( c r ) a X b = ( a r ) b X c - ( a b X c)r and so


( a * b x c)r = ( a * r ) bx c + (b.r)c X a + ( c * r ) aX b
This identity is useful in specifying the displacement of a rigid body in terms of three arbitrary
points of the body.

1.5. Show that if the vectors a, b and c are linearly dependent, a b x c = O . Check the
linear dependence or independence of the basis
A A A
u = 3i+j-2k
A A A
v = 4 i - j -k
A A A
w = i -2j +k
The vectors a, b and c are linearly dependent if there exist constants h, p and v, not a11 zero,
such that ha pb + + vc = O. The component scalar equations of this vector equation are

ha, + pbY + VC, = O


ha, + pb, + vc, = O
This set has a nonzero solution for h, p and v provided the determinant of coefficients vanishes,
a, b, c,
a y by cY = O
a, b, c,
which is equivalent to a b X c = O. For the proposed basis u, v, w,

Hence the vectors u, v, w are linearly dependent, and indeed v = u + w.

1.6. Show that any dyadic of N terms may be reduced to a dyadic of three terms in a
form having the base vectors êl, ê2,ê3 as (a) antecedents, (b) consequents.
Let D = albl + a2b2+ - - - + aNbN= aibi (i = 1,2, . . .,N ) .
( a ) In terms of base vectors, ai = aliel + azie2 + a3ie3 = a j i e j
A A A A
3 1 3 =
and so D = a-2.b. 3 t t = $.c.
~ ê.(a..b.)
3 1 1

with j = 1,2,3.
( b ) Likewise setting bi = bjiêj i t follows that D = aibjiêj = (bjiai)êj= g j ê j where j = 1 , 2 , 3 .

1.7. For the arbitrary dyadic D and vector v, show that D - v = v . D,.
Let D = albl + a2b2+ . - - + aNbN. Then
D v = al(b, v) + a,(b2 v ) + . . . + a N ( b N v )
= (V. b,)a, + ( v .b2)a2+ . . + ( v . b N ) a N = v . D,

1.8. Prove that ( D , D ) , = D, .D.


From (1.71),ü = D i j ê i ê j and D , = D j i ê i ê j . Therefore
D,.D = D..ê.ê..D
jt I 1
êê A A A A
p q p q = DjiDpq(ej'ep)eieq

and (o, o),


A A
= DjiDpq( e j e,) e,ei
A A A A A A
= Dpqeq( e , e j ) e i D j i = D,,ê$, .~ A A
.
~ = D~, D e ~ e ~
26 MATHEMATICAL FOUNDATIONS [CHAP. 1

1.9. Show that (D x v), = -v X Dc.


D X v = al(bl X V) + a,(b2 X + + aN(bNx V) V) ' ' '

(D X v), = (bl X v)al + (b, x v)a2 + - + (bNX v)aN


= -(V X b,)al - (v x b,)a, - ..- - (v X bN)aN = -v X D,

1.10.
A A
+
If D = a ii b j j + c k k and r is the position vector r = x i
AA AA A
+ y j + xk,
A A
show
that r D r = 1 represents the ellipsoid ax2 by2 cx2 = 1. + +
r - ~ . r= (s?+ y3+z k ) (a??+ b33+ ckk) (s?+ zk)
A A A A A A
= ( z i + y j + z k ) * ( a x i + b y j + c z k ) = as2+by2+cz2 = 1
A A AA AA AA A A
1.11. Forthedyadics D = 3 ? ? + 2 j j - j k + s k k and F = 4 i k - t - 6 3 3 - 3 k j + k k , com-
pute and compare the double dot products D :F and D F. ..
From the definition ab : cd = (a c)(b d) i t is seen that D : F = 12 + 5 = 17. Also, from
ab cd = (b e)(a d) i t follows that D F = 12 3 5 = 20. + +
1.12. Determine the dyadics G = D . F and H = F * D if D and F are the dyadics given in
Problem 1.11.
From the definition ab cd = (b c)ad, .
O = (3?? + 2;: - 3 k + 5 k k ) . (4;: + :6: -3k3 +kk)
AA AA A A A A AA A A
= 12ik+12j j + 3 j j - jk -15kj+5kk
AA AA AA AA AA AA AA
Similarly, H = (4ik+6jj-3kj+kk).(3ii+2jj-jk+5kk)
AA
= 20ik+123!-63k.-6k3+8kk

1.13. Show A A
directly from
A
theAnonion form
A
ofA the dyadic D that D = (Do?)? + (D-?)? +
( D 0 k ) k and also i w D . i = Dz,, i - D - j = D,,, etc.
Writing D in nonion form and regrouping terms,
II = (D,,i + D,,j + D,,k) i + (D,, i + D,,; + Dryk)j + (Dxzi + D,,? + D,$);
A A A A A A A A

= d l i + d 2 j +d,k = (o-?)?+ ( D * 3 ) 3 + ( ~ - 2 ) k
A A A

Also now
A A e A
i - d l = i * ( ~ - i =) i . ( D , , i + D Y , ~ +D,,k)
A 6 A
= D,,
A A
6
j * d l = ; . ~ . i = D,,, j * d 2 = 3 . ~ . j = D,,, etc.

1.14. For an antisymmetric dyadic A and the arbitrary vector b, show that 2 b . A = A, x b.
From Problem 1.6(a), A = &cl + &c2 + $,c3; and because it is antisymmetric, 2A = (A - A,)
or
( elel + e2c2+ e3c3 - elel - c2e2 - c3e3)
A A A A A A
2A =
= (êlcl - elel
A
+ e2c2- c2e2 + esc3 - c3e3)
A A A A

and so .
2b A = [(b $,)cl - (b c,) ê,] . + [(b.ê2)c2- (b . ê2]+ [(b ê3)c3 C,) - (b c3)$31

= [ ( ê , x c 1 ) x b + ( ê z X c 2 ) X b+ ( ê 3 X ~ 3 ) X b ]= (AvXb)

If D = 6; + 3 ij 4 k k and u = 2 i + k , v = 5 j, show by direct calculation


+
AA AA A A A

1.15.
that D . ( u x v ) = ( D x u ) . ~ .
A A A A A
Since u X v = ( 2 i + k ) X 5 j = 1 0 k - 5 i ,
AA AA A A A A
D * ( U X V=
) (6ii+3ij+4kk)-(-5i+lOk) = -30?+40k
A A AA AA AA AA
Next, D ~ =U ( 6 ? ? + 3 ? ; + 4 k k ) ~ ( 2 i + k ) = -6ik+8kj-6ij+3ii
AA AA AA AA A A A
snd (DXu)*v = ( 3 i i - 6 i j - 6 i k + 8 k j ) . 5 j = -30i+40k
CHAP. 11 MATHEMATICAL FOUNDATIONS 27

Considering the dyadic


A A A A A A
D = 31:-4:;+2ji+jj+kk
as a linear vector operator, determine the
vector r' produced when D operates on
r = 4?+23+5k.

Fig. 1-14

1.17. Determine the dyadic D which serves as a 'inea? vec^r operator for the vector
function a = f(b) = b + b x r where r = x i y j xk and b is a constant vector. + +
In accordance with (1.59) and (1.60), construct the vectors
u = f(i) = i
A A
+ iXr
A A
= i -zj+yk
A A

v = f(j) = j
A A
+ j Xr
A
= xi
A
+j A
-xk
A

A A A A A A
w = f(k) = k + k X r = - y i + x j + k
A A A A
Then ü = ui+vj
A A
+ wk A A
= (i -zj+yk)i+(xi
A A A A A A
t j -xk)j +(-yi+xj+k)k
A

and a = D b = (b, + b,z - b,y) i


A
+ (-b,z + b, + b,x) j + (b,y A
- b,x + b,) k A

AS a check the same result may be obtained by direct expansion of the vector function,
a = b +bXr A
= b,i -i-b,3 + b,k + (b,z - b,y) i + (b,x - bxz) j + (bxy- b,x)k
A A A A

1.18. Express the unit triad ê4,ê, ê, in terms of A


A A A
i, j, k and confirm that the curvilinear triad
A
is right-handed by showing that ê, x ê, = e,.
By direct projection from Fig. 1-15,
A
e, = (COS g cos e) i
A
+ (cos g sin e) j A
- (sin g ) k
A

A
e, = (- sin e) i + (cos e) j
A A

(sin g cos e) i + (sin g sin e) j + (cos @) k


A A A
A
e, =
and so Fig. 1-15
e A

1 k
A
e+ x ê, = cos g cos e cos g sin e - sin @

- sin e cos e O
= (sin @ cos e ) i
A
+ (sin g sin e) j
A
+ [(cosze + sin2 e) cos g] k A
= 2,.

1.19. Resolve the dyadic D = 3;: + 4:: + 6 3 1+ 7;: + 10 ko;f 2 k3 into its symmetric
and antisymmetric parts.
Let D = E +F where E = E, and F = -F,. Then
E +
= (1/2)(D D,) = (1/2)(6 i i
A A
+ 4 i k + 4 k i + 6 j i + 6 i j + 14 j j
A A A A A A A A A A

+ ioct + i o t k + 2 2 3 + 2;;)
A A A A A A A A A A A A A A A A
= 3ii+3ij+7ik+3ji+7jj+jk+7ki+kj = E,
A A A A A A A A A A A A A A A A
F = (1/2)(~-o,) = ( 1 / 2 ) ( 4 i k - 4 k i + 6 j i - 6 i j + 1 0 k i - 10ik+2kj-2jk)
A A A A A A A A A A A A
= -3i j - 3 i k + 3 j i - jk+3ki+kj = -F,
28 MATHEMATICAL FOUNDATIONS [CHAP. 1

1.20. With respect to the set of base vectors ai, a2, a3 (not necessarily unit vectors), the set
a', a2, a3 is said to be a reciprocal basis if ai- a3 = aij. Determine the necessary rela-
tionships for constructing the reciprocal base vectors and carry out the calculations
for the basis
bs = ? + $ + %
A A A
bl = 3 i + 4 3 , bz = - i + 2 ? + 2 k ,
By definition, al a1 = 1, a, a1 = O, a3 a1 = O. Hence a1 is perpendicular to both a, and
a,. Therefore i t is parallel to a, X a,, i.e. a1 = X(a2 X a3). Since al a1 = 1, al Xa2 X a, = 1 and
h = l / ( a l a, X a,) = l/[ala2a3]. Thus, in general,

For the basis b l , b2,b3, l / h = bi ba X b3 = 12 and so


A A
bl = (b, x b3)/12 = ( j -k)/4
- i13 + j 14 + k / 1 2
A A A
b2 = (b3x bl)/12 =

INDICIAL NOTATION - CARTESIAN TENSORS (Sec. 1.9-1.16)


1.21. For a range of three on the indices, give the meaning of the following Cartesian tensor
symbols: Aii, Bijj, Rij, ai Tij,aibjSij.
Ai, repi.esents the single sum Aii = All + A,, + A,,.
Bijj represents three sums: (1) For i = 1, Blll + B12, + B133.
(2) For i = 2, BZll + BZz2+ BZ33.
(3) For i = 3, B3,, + B322+ B333.
Rii represents the nine components Rll, R12,Ri3,R,,, Rz2,RZ3? R32>R33.
ai Tij represents three sums: (1) For j = 1, al Tll + a2 T2i + a3 T3i.
(2) For j = 2, a1T12+ a2TZ2f a3 T3,.
(3) For j = 3, a1T13+ a2T23 + a3T33.
aibjSijrepresents a single sum of nine terms. Summing first on i, aibjSij= albjSlj a2bjSzj + +
a3bjS3j. Now summing each of these three terms on j,
aibjSij = alblSll + alb2S12+ alb3S13+ a2b1S2,+ a2b2Sz2
+~ +
2 ~ 3 ~ ~2 33 ~ 1 ~ ~ + +
3 31~ 2 S 3 2~ 3 ~ 3 ~ 3 3

122. Evaluate the following expressions involving the Kronecker delta aij for a range of
three on the indices.
( a ) Sii = S l l + Sz2 + = 3
(b) SijSij = S l j S l j + 62jS2j + S3jS3j = 3
(C) S i j S i k S j k = S l j s l k 8 j k + 82j62k8jk + 83j83k8jk = 3
( d ) Gijsjk = s i l S l k + Si2S2k + 8i383k = Sik
(e) 6ifiik = S1#11, + 82jA2k + 83jA3k = Ajk

1.23. For the permutation symbol Cijk S ~ O Wby direct expansion that (a) Cijk'kij = 6,
( b ) ' i j k a j a k = 0.
(a) First sum on i, i ~ k i..
€.. j
= €ljkCklj + ' 2 j k e k 2 j + '3jkek3j
CHAP. 11 MATHEMATICAL FOUNDATIONS 29

Next sum on j. The nonzero terms are


'ijk'kij = '12kEk12 + '13kCk13 + '21kek21 f '23kek23 + '31kCk31 + '32kEk32
Finally summing on k, the nonzero terms are
e..' ..
~k k21
-
- '123'312 + '132'213 + '213'321 + '231'123 + '312'231 + '321'132
+
= ( l ) ( l ) (-I)(-1) + (-I)(-1) + ( l ) ( l+) ( l ) ( l+) (-I)(-1) = 6
( b ) Summing on j and k in turn,
Eijkajak = ' i l k a l a k + Ei2ka2ak + Ei3ka3ak
= 'i12ala2 + ' i 1 3 ~ 1 ~f3 ' i 2 1 ~ 2 ~+1 'i23a2a3 +
f ' i 3 1 ~ 3 ~ 1 Ei32a3a2

From this expression,


when i = 1, eljkajak = a2a3- a3a2 = O
when i = 2, €2jkajak = a i a 3 - a3a1 = 0
when i = 3, €3jkajak = a1U2 - a2al = 0
Note that eiikajak is the indicial form of the vector a crossed into itself, and so a X a = 0.

1.24. Determine the component f2 for the vector expressions given below.
fi 'ijkTjk

f2 = '2jkTjk = €213~13 + '231T31 = -T13 + T31


( b ) fi = C,, j b j - cj,ibj
f2 = c2,1b1 +~ 2 , 2 +
~ 2~ 2 , 3 ~ 3 - ~ 1 , 2 ~ ~1 2- . 2 ~ 2 ~- 3 , 2 ~ 3
= +
fc2, 1 - ~ l , 2 ) ~ 1 ( ~ 2 . 3- ~ 3 , 2 ) ~ 3

(C) fi = Bijf;

f2 = B21f: + B22f2 + B23f3*

1.25. Expand and simplify where possible the expression Dijxixj for (a) Dij = Dji,
(b) D i j = - D j i .
Expanding, Dijxixj = Dljx1xj + D2jx2xj+ Dsjx3xj
= D11x1x1 + D 1 2 x 1 x 2 + 0 1 3 x 1 5 3 + D 2 1 ~ 2 ~+1D 2 2 ~ 2 ~ 2
+ D23x2x3 + D31x3x1 + D32x3x2 + D33x3x3
(a) Dijxixj = D11(x1)2 + D22(x2)2 + D33(x3)2 + 2D12x1x2+ 2D23x2x3+ 2D13x1x3
( b ) Dijxixj = O since Dll = -Dll, D12 = -D2,, etc.

1.26. Sh0w that c i j k c k p q = - 8 i q s j p for (a)i = 1, j = q = 2, p = 3 and for (b) i = q = 1,


j = p = 2. (It is shown in Problem 1.59 that this identity holds for every choice of
indices.)
( a ) Introduce i = 1, j = 2, p = 3, q =2 and note that since k is a summed index i t takes on
a11 values. Then - -
'ijk'kpq - '12kEk32 - E121E132 + '122'232 + '123'332 = O
and SipSjq - aiqS, = 813S22 - S12S23 = O
(b) Introduce i = 1, j = 2, p = 2, q = 1. Then eijkekpq = ~123'321 = -1 and SipSjq - SiqSjp =
S12S21 - S11S22 = -1.

1.27. Show that the tensor Bik = cijkaj is skew-symmetric.


Since by definition of cijk an interchange o f two indices causes a sign change,
Bik = eijkaj = - ( e k j i a j ) = - ( B k i ) = -Bki
30 M A T H E M A T I C A L FOUNDATIONS [CHAP. 1

1.28. If Bij is a skew-symmetric Cartesian tensor for which the vector bi = ( + ) E ~ ~ ~


i .
show that Bpq= t p qbi
Multiply t h e given equation b y epqi and use t h e identity given i n Problem 1.26.

+
~pqibi = & ~ ~ q i ~ i j k=~ j&k( a p j a q k - SpkSqj)Bjk = & ( B p q- B q p ) = &(Bpq B p q ) = B p q

1.29. Determine directly the components of the metric


tensor f o r spherical polar coordinates as shown
in Fig. 1-7(b).
axi azi
W r i t e (1.87) a s gpq = and label t h e co-
;j~q
ordinates as shown i n Fig. 1-16 (r = e,, @ = e,, e = e,).
Then
x , = 8 , sin 0, cos 8,
x , = e , sin e, sin e,

Fig. 1-16
Hence
ax1
- - ax1 - 8x1 -
- sin O , cos 8 , - - e , cos 0 , cos e3 - - -e, sin e, sin e,
ao1 as2 863

8x2
- - 8x2 - 8x2 -
- sin e2 sin e, - - 0, cos O , sin e, - - 6, sin e2 cos e,
861 302 as3

8x3
- - 8x3
- -el sin e, - = o
as2 863

f r o m which g,,
axi axi
= --
as, asl
-
- sinz e, cos2 8 , + sin2 o, sin2 e, + cos2 e, = 1

axi axi
g33 =
-- -
ao3 ae, - eSsin2 e, sin2 e, + eSsin2 6 , cos2 e , = eSsin2 O ,

Also, gpq = O for p Z q. For example,


axi axi
= (sin e, cos @,)(e,cos e, cos e,)
g12 = aBi. dB.,
e
+ (sin O 2 sin e3)(e1cos e, sin e3) - (COS eZ)(elsin e,)
= o
T h u s for spherical coordinates, (ds)2 = (de1), + (el)2(de2)" (e1 sin de^)'.

1.30. Show that the length of the line element d s


resulting from the curvilinear coordinate incre-
ment doi is given by ds = G d O i (no sum).
Apply this result to the spherical coordinate
system of Problem 1.29.
W r i t e (1.86) as (ds)2 = gpqde, de,. T h u s f o r t h e
line element (de,, O , O ) , i t follows t h a t (ds), = g l l ( d e l ) 2
and d s = \/g,, de,. Similarly for (0, de,, O ) , ds = G d e 2 ;
and for ( 0 , 0 , de3), ds = de3. Therefore (Fig. 1-17),
CHAP. 11 MATHEMATICAL FOUNDATIONS 31

( 1 ) For (del,O,O), ds = del = d r


( 2 ) For ( O , de,, O ) , ds = e, de, = r d+
( 3 ) For (O, O, de,), ds = e , sin 8 , de, = r sin + de

1.31. If the angle between the line elements represented by (doi, O, O) and (O, don,O) is
812
denoted by p,,, show that cosp,, =
66.
Let ds, = 6de, be the length of line element represented by (del, O , 0 ) and dsz = G d 6 2
axi
be that of ( O , de,, O). Write (1.85) a s dxi = -dek, and since (A),= cos plz h,dsz,
aek

ax, a%,
- - -de, de,
ae, ae,
+ as,
-
ax,
ae, ae2
ax, ax3
- de, de2 i- - -de1 de2
ae, as,
= giz de1 dez

del de2 812


Hence using the result of Problem 1.30, tos plz = g,, - -=
ds1ds2 66

1.32. A primed set of Cartesian axes O X ~ X $isXobtained


~ by a rotation through an angle
e about the X , axis. Determine the transformation coefficients aij relating the axes,
and give the primed components of the vector v = ~ $ 1 vzêz v&. + +
From the definition (see Section 1.13) ai; =
cos (si,s j ) and Fig. 1-18, the table of dire8ion
cosiiies is

51 22 x3

X; cos e sin e O

xó -sin e cos 6 O

x3 o o 1

Thus the transformation tensor is

cos 6 sin 6 O
A = -sin e cos 6 O
Fig. 1-18

By the transformation law for vectors (1.94),


v ; = a I j v j = v , cose + v , sin 6
v i = a Z j v j = -vl sin e + v , cos 6

v j = a3jvj = V,

1.33. The table of direction cosines relating


two sets of rectangular Cartesian axes
is partially given as shown on the
right. Determine the entries for the
bottom row of the table so that 0 x i x ; x i
is a right-handed system.
32 MATHEMATICAL FOUNDATIONS [CHAP. 1

The unit vector along the xí axis is given by the first row of the table a s $I
= (315)ê1- (415)ê 2 .
Also from the table 2; = ê3. For a right-handed primed system 23 = Xêi, or Ae t, =
[ ( 3 / 5 ) ê 1- (4/5)$2]x 2, = (-3/5)& - (416)$~and the third row is 1 1
x: -415 1 -3/5 1 O I.
1.34. Let the angles between the primed and
unprimed coordinate directions be given
by the table shown on the right. Deter-
mine the transformation coefficients aij
and show that the orthogonality condi-
tions are satisfied.
The coefficients aij are direction cosines and
may be calculated directly from the table. Thus
- 1 112 -112
aij - o llfi
- 1 112 -112
The orthogonality conditions aiiaik = S i , require:

1. For j = k = 1 that a l l a l , a2,a,, + + a3,a3, = 1 which is seen to be the sum of squares of


the elements in the first column.
2. +
For j = 2, k = 3 that a12a13+aZ2a23 a3,a3, = O which is seen to be the sum of products
of corresponding elements of the second and third columns.
3. Any two columns "multiplied together element by element and summed" to be zero. The
sum of squares of elements of any column to be unity.
For orthogonality conditions in the form ajiaki= Sjk, the rows are multiplied together
instead of the columns. A11 of these conditions are satisfied by the above solution.

1.35. Show that the sum x A +~pBij ~ represents the components of a second-order tensor
if Aij and Bij are known second-order tensors.
By (1.108) and the statement of the problem, Aij = a p i a q j ~ p and
q Bii = apiaqjBiq. Hence
XAij + @Bii = ~ ( ~ ~ ~ ~ ~ = k a p~i a ,)i ( ~ ~+
+ ay(apia,iBg,) k qyBP,)
which demonstrates that the sum transforms a s a second-order Cartesian tensor.

1.36. S ~ O that
W (Pijk + Pjki + Pjik) XiXjXk = 3Pijk xixjxk.
Since a11 indices are dummy indices and the order of the variables xi is unimportant, each
tem of the sum is equivalent to the others. This may be readily shown by introducing new dummy
variables. Thus replacing i, j, k in the second and third terms by p, q , r , the sum becomes
Pijkxixjxk + Pqrpxpxqxr + Pqprxpxqx,
Now change dummy indices in these same terms again so that the form is
Pijkxixixk + Pijk + Pijkxjxixk = 3Pijkxixjxk

1.37. If Bij is skew-symmetric and Aij is symmetric, show that AijBij = 0.


Since A . = A l,t. and B . . = -B..31' A..B..
13 t3
-A..B..
I1 I1
or A..B..
1J 13
+ A..B..
31 31
= A..B..
L3 i 1
+ A P P BPP = O.
Since a11 indices are dummy indices, ApPBpq= AijBij and so 2AijBij= O, or AijBij= 0.
CHAP. 11 MATHEMATICAL FOUNDATIONS 33

1.38. Show t h a t the quadratic form Dijxixjis unchanged if Dij is replaced by its symmetric
part D(ij).
Resolving Dij into its symmetric and anti-symmetric parts,
Dij = D(;j) + DLijl = S ( D i j+ Dji) + &Dij Dji)-

Then D,ijlxixj = #Dij + Dji)xixj = S(Dijxixj+ Dpqxqxp) = Dijxixj

1.39. Use indicial notation to prove the vector identities


(1) a x (b x c) = ( a *c)b - ( a . b)c, .
(2) a x b a = O
(1) Let v = b X c. Then vi = eijkbjck; and if a X v = w, then
w p = ePqiaqeijkbjck
- ( S p j S q k - SpkSqj)aqbjck (see Problem 1.26)
= aqbpcq - aqbqcp
= (aqcq)b, - (aqbq)cp
Transcribing this expression into symbolic notation,
w = a X ( b X c ) = ( a - c ) b- ( a . b ) c
( 2 ) Let a X b = v. Thus vi = eij,ajbk; and if h = v a , then X = eiik(aiajbk). But eijk is skew-
symmetric in i and i, while (aiajbk) is symmetric in i and j . Hence the product eijkaiajbk
vanishes a s may also be shown by direct expansion.

1.40. Show that the determinant

may be expressed in the form cijkA1iA2jA~k.


From (1.52) and (1.109) the box product [abc] rnay be written
a1 a2 a3
x = a.bXc = [abc] = eijkaibjck = bl b2 b3
C1 C2 C3

If now the substitutions ai = A l i , b, = A Z i and ci = A s i are introduced,


X = eijkaibjck = eijkAliAZjA3k

This result may also be obtained by direct expansion of the determinant. An equivalent expression
for the determinant is eijkAilAj2Ak3.

1.41. If the vector vi is given in terms of base vectors a, b, c by vi = aiai + pbi + yCi,
34 MATHEMATICAL FOUNDATIONS [CHAP. 1

By Cramer's rule, a =
I v3 b3 c3 I and by (1.52) and (1.I 0 9 ) , a = eiikvibjck
a1 bl c1 'pqrapbq~r'
a2 b2 c2

EijkaivjCk cijkaibjvk
Likewise /3 =
epqrapbqcr = ~pqrapbqcr'

MATRICES AND MATRIX METHODS (Sec. 1.17-1.20)


1.42. For the vectors a = 3:+42, b = 23-62 and the dyadic D = 3 ? ? + 2 ? k -
4 3 - 5k^5 3,
compute by matrix multiplication the products a D, D b and a . D b. .
Let a D = v; then [ v l ,v2, v3] = [3,0,4] O -
-4 [ ] = 9 , -20,6].

.
Let D b = w; then
O -5 -6 -10

Let a D b = v . b = h; then [h] = [9, -20,6] ( 2 1= [-761.

1.43. Determine the principal directions and principal values of the second-order Cartesian
tensor T whose matrix representation is
3 -1 o
[Tij] = -1 3 O
O 0 1
From (1.132), for principal values h,

which results in the cubic equation h3 - 7h2 + 14h - 8 = ( h - l ) ( h- 2 ) ( h- 4 ) = O whose principal


values are h ( 1 , = 1, = 2, h ( 3 )= 4.
Next let n:l) be the components of the unit normal in the principal direction associated with
Ao, = 1. Then the first two equations of (1.231) give 2n:" - nil' = O and -n:" 24'' = 0, +
from which n:" = nll)
= 0 ; and from nini = 1, nB1)= *I.
For h t 2 ) = 2, (1.131) yields ni2' - nl2'
= 0 , -ny' + nf' = O, and -ny' = O. Thus
ny' = ni2' = %1/\/2, since nini = 1 and ni2' = 0.
For h,,, = 4, (1.131) yields -ni3' - n:' = 0, -ni3' - ns' = O, and 3 n F ) = O. Thus
n33' = 0, n : 3 ) = -nF' = TI/@.
The principal axes x: may be referred to the original axes xi through the table of direction
cosines
CHAP. 11 MATHEMATICAL FOUNDATIONS 35

from which the transformation matrix (tensor) may be written:

1.44. Show that the principal axes determined in Problem 1.43 form a right-handed set of
orthogonal axes.
Orthogonality requires that the conditions aijaik= 6 j k be satisfied. Since the condition
nini = 1 was used in determining the aii, orthogonality is automatically satisfied for j = k.
Multiplying the corresponding elements of á n y row (column) by those of any other row (column)
+
and adding the products demonstrates that the conditions for j k are satisfied by the solution
in Problem 1.43.
Finally for the system to be right-handed,
~ ( 3 =
A
n(2) ) n ( l ) . ~h~~

A A A
e1 e2 e3

i i o = (4+ 4)ê3 A
= e3

-116 1 o
As indicated by the plus and minus values of
aij in Problem 1.43, there are two sets of prin-
cipal axes, x: and x:*. As shown by the sketch
both sets a r e along the principal directions
with x: being a right-handed system, x;" a
left-handed system. Fig. 1-19

1.45. Show that the matrix of the tensor Tij of Problem 1.43 may be put into diagonal
(principal) form by the transformation law T: = a;,ajqTPq, (or in matrix symbols
T* = cATcAT).
36 MATHEMATICAL FOUNDATIONS [CHAP. 1

1.46. Prove that if the principal values h ( l ) , h ( 3 ) of a symmetric second-order tensor


are a11 distinct, the principal directions are mutually orthogonal.
The proof is made for h t 2 ) and h ( 3 ) . For each of these (1.129) is satisfied, so that T i j n j 2 )=
~ ' ~ ~ ~= n~ ~
~ ( , , n ~and j ,~n ': Multiplying
~ ) . the first of these equations by ni3' and the second

Since Tij is symmetric, the dummy indices i and j may be interchanged on the left-hand side of
the second of these equations and that equation subtracted from the first to yield
(A,,, - ~ ( ~ , ) n : ~ ' n=: ~O'
Since h(,, Z h(,,, their difference is not zero. Hence n:2'n:3' = 0 , the condition for the two
directions to be perpendicular.

1.47. Compute the principal values of ( T ) 2 of Problem 1.43 and verify that its principal
axes coincide with those of T.

[Ti. = [-i i !][-i i


3 -1 3 -1
!] [-: =
10 -6
1;

The characteristic equation for this matrix is


10 - h -6 O
-6 10 - h O = ( 1 - h)[(10- h)2 - 361 = ( 1 - X)(h - 4)(h- 16) = O
o O l - h
from which h(,) = 1, = 4, h ( 3 )= 16. Substituting these into (1.131) and using the condition
nini = 1,
gn"' - Gn(l' =
For h ( , ) = 1,
-6n:"
1

$.
2

9n;" = O
= n21' = 0' n(l)
3
=21

6n(2) - Cn(2) = O
1 2
For A ( 2 , = 4, -Gn?' + 6nk2' = 0 = ni2'= k l f i , nF) = 0
-Sn'2' -
3 - 0

For = 16,
-15ni3' = O
which are the same a s the principal directions of 1.

1.48. Use the fact that (T)2 has the same principal directions as the symmetrical tensor T
to obtain fiwhen

First, the principal values and principal directions of T are determined. Following the procedure
of Problem 1.43, the diagonal form of T is given by
CHAP. 11 MATHEMATICAL FOUNDATIONS 37

with the transformation matrix being

[aijl = [ llfi
0
-216
lldz

llfi
llh

11,h

Therefore fi = ;" ó 3 o o
and using [aij; to relate this to the original axes by the

transformation fi = ~ A,~ f i
the matrix equation is

CARTESIAN TENSOR CALCULUS (Sec. 1.21-1.23)


1.49. For the function x = Aijxixj where Aij is constant, show that dhldxi = (Aij Aji)xj +
and d2X/dxidxj = Aij + Aji. Simplify these derivatives for the case Aij = Aji.

Consider - =
ax axi
A.. -X . + A,.%.a%,
- Since
axi
-z Sik, i t i~ seen that -
ax = A k j z j+ Aikzi =
axk %3axk 3 a> 'axk ' Jxk Jxk

(A + A jk)xj. a2A 8%
Continuing the differentiation, -- ( A k j + A . ) -
-
- A k P + A P k .If A 23. . = A3.i 9
ax, axk lk axp
- = 2Akjxj and -- 2APk.
axk as, axk

1.50. Use indicial notation to prove the vector identities (a) V X V+ = O, ( b ) V V X a = 0.


( a ) By (1.147), V + is written $,i and so v = V X V + has components v i = ~ ~ ~= ~~i j k +a, k j .~But
+ , ~
eijk is anti-symmetric in j and k, whereas + , k jis symmetric in j and k; hence the product
E ~ ~ vanishes.
~ + , ~ ~ The same result may be found by computing individually the components of
v. For example, by expansion vl = '123+,23~ ~ 3 ~ = +
, ~ 2-
+ ($,23 32) = 0. +,
( b ) V V x e = h = ( ~ ~ ~= ~eijkak,ji
a ~ =, O since ak,ij ak,ji and ~ i j k= -'jik.

1.51. Determine the derivative of the function X = ( X I ) ~ 2 x 1 - +


~ (~ ~ 3 )in~ the direction
of the unit normal n
= (2/7)&- (3/7)&- (6/7)ê3 or & = (261- 3ê2 - 6ê3)/7.
ax
The required derivative is - = V h n = X,ini.
A
Thus
an
38 MATHEMATICAL FOUNDATIONS [CHAP.1

1.52. If is a second-order Cartesian tensor, show that its derivative with respect to
Aij
xk, namely Aij,k, is a third-order Cartesian tensor.
F o r the Cartesian coordinate systems xi and xl, xi = ajixi and âxildxi = ajt. Hence

which is the transformation law f o r a third-order Cartesian tensor.

1.53. If r2= Xixi and f ( r ) is a n arbitrary function of r, show that ( a ) V ( f ( r )= ) f'(r)xlr,


+
) ) f l r ( r ) Zf'lr, where primes denote derivatives with respect to r.
and ( b ) V 2 ( f ( r =
a f ar
( a ) The components of V f a r e simply f,+ Thus f,i = - - ; and since
ar
= 2r ax,
-- = 2 ~ . . x . i t
ar axi axj 2 ~ a
ar 2. a f ar
follows t h a t - = 3. Thus f,i = - - = f'xi/r.
dxj r ar axi

Use the divergence theorem of Gauss to


that
S ~ O W L Xinj d S = V6ij where nj d S
represents the surface element of S , the
bounding surf ace of the volume V shown
in Fig. 1-20. xi is the position vector of
nj d S , and ni its outward normal.

= SijV Fig. 1-20

1.55. If the vector b = V x v, S ~ O Wthat


a scalar function of the coordinates.
i hbisdS = i h,ibi d V where h = h ( x i ) is

Since b = V X v, bi = eijkvk, and so

i hbini d S = i rijkhvk,jni d s

= & Li bi d ~ since hrijkvk,ii = O


CHAP. 11 MATHEMATICAL FOUNDATIONS 39

MISCELLANEOUS PROBLEMS
1.56. For the arbitrary vectors a and b, show that
A = ( a x b) (a x b) + (a.b)2 = ( ~ b ) ~
Interchange the dot and cross in the first term. Then
h = a - b x (axb) + (a.b)(a0b)
= a [(b. b)a - (b.a)b] + ( a . b)(a0b)
= ( a . a ) ( b - b ) - ( b . a ) ( a W b+) (a.b)(a.b)
= (ab)2

since the second and third terms cancel.

d
1.57. If ii = o x u and V = w x v, show that -
dt
(u X v) = ca X (u x v).
(a) In symbolic notation,
d
z(uxv) = h X v + ux; = (mXu)Xv+ uX(mXv)
= (V. m)u - (V U)OI + (U v)a - (U @)V

= (V' ,)U - (U @)V = a X (U X V)

d
( b ) I n indicial notation, let (u X V) = W. Then

wi =
d
( E ~ ~ ~=u ~eijkGjwk
v ~ ) + eijkuj;k

and since Aj = e j p q w p u q and ck = ekmnamVn,


W i. €y. k. E3pqapUqwk
, + eijk'kmnUjamvn ('ijkEkmn -~ink'kmj)~~jamwn
and using the result of Problem 1.59(a) below,
wi = (SimSjn - S i n S j m - SimGjn + SijSmn)~jamwn
= (SijSmn - 8 i n 8 j m ) u j ~ m ~ nE ~irnk'~j~Ujarnw~

which is the indicial form of m X (u X v).

8mp 8mq Sms

1.58. Establish the identity cPqsrmnr = Snp 8nq 8ns .


8rp 8rq 8rs

A11 -412 A13

Let the determinant of Aij be given by det A = AZl Az2 A23 . An interchange of rows
or columns causes a sign change. Thus
A31 A32 A33

A,, A,, A23 A12 A,, A13

A,, A,, A,, = A,, A,, AZ3 = -detA


A31 A,, A,, A,, A31 A33

and for an arbitrary number of row changes,

Aml Am, Am3

Anl Ana An3


-
- em,, det A
Ar1 A, 4 3
40 MATHEMATICAL FOUNDATIONS [CHAP. 1

or column changes,
A,, A,, A I S
-
-42, -42, -4% - epqS det A
A,, -43, A,,
Hence f o r a n arbitrary row and column interchange sequence,

A,, Am, Ams


-
Anp Anq Ans - 'mnr'pqs det A
A, Ar, Ars
When Aij = Sij, det A = 1 and the identity is established.

1.59. Use the results of Problem 1.58 to prove (a) cpqscsnr = S P n S ~ r- Spraqn, ( b ) ~ P , S ~ S Q=~ - 2 6 ~ ~ .
Expanding the determinant of the identity in Problem 1 . 5 8 ,
-
'pqs'mnr - S m p ( S n q S r s - s n s a r q ) +
Omq(SnsSrp - Onpsrs) + Sms(SnpSrq - SnqSrp)
( a ) Identifying s with m yields,
'pqs'snr = Ssp(8nqsrs - SnsSrq) f S s q ( 8 n s s r p - SnpSrs) + S s s ( S n p ~ r q- SnqSrp)
-
-
srpsnq - SpnSrq + SqnSrp - SnpSqr + 3SnpSrq - 3Snqsrp
-
- S n p a r q - SnqSrp
( b ) Identifying q with n in ( a ) ,
-
~pqs'sqr - Sqparq - aqqsrp = S p r - 3 S p r = -2Spr

1.60. If the dyadic B is skew-symmetric B = -B,, show that B , x a = 2a B.


Writing B = blêl + b2ê2+ b3ê3 (see Problem 1.6),
B, = bl x ê1 + bz X êz + b3 X ê3

and B , X ~ = (blXêl)xa+ (b2xê2)xa+ (b3xê3)xa


= (a bl)ê, - (a.A
el)bl + A
( a . b , ) ê z - ( a e z ) b 2 -t ( a b 3 ) ê 3 - ( a ê 3 ) b 3
= a (blêl + bzêz+ b 3 ê 3 ) - a .( ê l b l + ê2b2+ ê 3 b 3 )
-
- a.0 - a.0, = 2a.B

1.61. Use the Hamilton-Cayley equation to obtain (B)4 f o r the tensor B =


Check the result directly by squaring ( B ) 2 .
The characteristic equation f o r B is given by

By the Hamilton-Cayley theorem the tensor satisfies its own characteristic equation. Hence
+
- 2 ( ~ -) 6~8 91 = 0, and multiplying this equation by B yields = 2(B)3 6 ( 8 ) 2 - 9 8 01 +
+
(B)4 = 1 0 ( B ) z 38 - 181. Hence
CHAP. 11 MATHEMATICAL FOUNDATIONS 41

Checking by direct matrix multiplication of (B)2,

(23)4 =
O 26

1.62. Prove that (a) Aii, (b) AijAij, ( c ) c i j k ~ k j p A t pare invariant under the coordinate trans-
formation represented by (1.103), i.e. show t h a t Aii = ~ : i ,etc.
(a) By (1.103), Aij = aPiaqjAP,. Hence Aii = a,,aqiA;, = S,,A:, = A:, = Aii.
(b) AijAij = a p i a q j A ~ q a m i a n j =
A ~ nB ~ ~ s ~ ~ A= PApqApq
~ A & =~ A!.A!.
21 13

(C) cijk'kjpAip = €ijk'kjpamianp~hn


= ( 8 i j s j p- ~ipsjj)amianpAkn
-
- ( S m n - &mnsjj)Akn= (SmjSnj - ~ m n ~ j j ) ~=k n~mjk%jnALn

1.63. Show that the dual vector of the arbitrary tensor Tij depends only upon T ~ i j ibut that
the product TijSijof Tij with the symmetric tensor Sij is independent of Triji.
) eijk T L j k l since
By (1.110) the dual vector of Tij is vi = cijkTjk, or vi = eijk(TCjk)T c j k l = +
= O (cijk is anti-symmetric in j and k, T ( j k ,symmetric in j and k).
cijkT(jk)
For the product T i j S i j= TCii>Sij TLij1 +
Si? Here TLii,Sii = O and T,Sij = TCij,S,.

1.64 Show that D : E is equal to D 0 . E if E is a symmetric dyadic.


WriteD = lIijêiêj and E = ~,,ê,ê,. By (1.31), D : E = D i j ~ , , ( ê i êp)(êj2,). By (1.35),
D..E - .
= DijEpq( g j $,)(Si 3,) = DijE,, (êj A%)(Aei Ae,) since E,, = E,,. Now interchanging the
A A
role of dummy indices p and q in this last expression, D E = êq)(ei e,).

1.65. Use the indicial notation to prove the vector identity V x (a x b) = b Va - b(V a) +
a ( V * b )- a - V b .
Let V X (a x b ) = v; then v, = ~ ~ , ~ c ~or~ ~ a ~ a ~ b ~
Vp = 'pqi'ijk (ajbk),q = epqi'ijk(aj,qbk ajbk,q)
= (Spjsqk - Spksqj)(aj,q bk + ~ j ~ k , q=) bq - aq,q bp + - aqbp,q
Hence v = b V a - b ( V a) - + a(V b) - a Vb.

1.66. By means of the divergente theorem of Gauss show t h a t & X (a X x) dS = 2aV


where V is the volume enclosed by the surface S having the outward normal n. The
position vector to any point in V is x, and a is an arbitrary constant vector.
In the indicial notation the surface integral is
the volume integral (rqPirijkajxk),,
s, cqPinpcijkajxkdS.By (1.157) this becomes
dV and since a is constant, the last expression becomes

q p j , p d = (Sqjspk - s q k s p ~ a j x k , p dv = (aqxp,p - apxq,p) 'V

: L (aqS p p - a, a,,) dV = (3aq- a,) dV = Za,V


42 MATHEMATICAL FOUNDATIONS [CHAP. 1

1.67. For the reflection of axes shown in Fig. 1-21 show that the transformation is
orthogonal.
From the figure the transformation matrix is

The orthogonality conditions aijaik = S j k or ajiaki= S j k


a r e clearly satisfied. I n matrix form, by (1.117),

i-: :l[u-i 01
L 0 0 1 ] L 0 0 1 J
r: : 01
L 0 0 1 J Fig. 1-21

1.68. Show that (I x v) D = v x D.


I Xv =
A A
+ kk) X (v,i+v,j +v,k)
(i i + j j
A A A A A A A

i(v,k-v,j) + j(-v,k+v,i) + k(v,j-vyi)


A A A A A A A A
=

(vXi)i + (vXj)j + (vXk)k = v X I


A A A A A A
=
Hence ( I X v ) - D = V X I * D = V X D .

Supplementary Problems
1.69.
A
+ A A A A
Show t h a t u = i j - k and v = i - j a r e perpendicular to one another.
h
Determine w so
t h a t u, v, & forms a right-handed triad. Ans. w = ( - l / f i ) ( ? + 2c) 3+
1.70. Determine the transformation rnatrix between the u,v,W axes of Problem 1.69 and the coordinate
directions.

1.71. Use indicial notation to prove ( a ) V x = 3, (b) V X x = O, (c) a . V x = a where x is the position
vector and a is a constant vector.

1.72. Determine the principal values of the symmetric p a r t of the tensor Tij =
Ans. h ( , , = -15, h ( Z )= 5 , h,,) = 10 -3 -18
CHAP. 11 MATHEMATICAL FOUNDATIONS 43

1.73. F o r the symmetric tensor Tij = determine the principal values and the directions
of the principal axes.

Ans. h c l , = 2 , A ( z , = 7, A,,, = 1 2 ,

1.74. Given the arbitrary vector v and a n y unit vector ê, show t h a t v may be resolved into a component
parallel and a component perpendicular to 2, i.e. v = ( v * ê ) ê ê X ( v X 3 ) . +
1.75. If .
V v = O, V x v = 6 and V X w = -c, show t h a t V2v = y.

1.76. Check the result of Problem 1.48 by direct rnultiplication to show t h a t fifi = T.

1.77. Determine the square root of the tensor B =

*(&+I) #&-I) 0

1.78. Using the result of Problem 1.40, det A = E ~ ~ ~ A show


~ ~ t hAa t ~ det A ~= ~det, A det 0.
~ (AB)

1.79. Verify t h a t (a) O3,,vp= v3, ( b ) 8siAji = Aj3, (C) Sijeijk = 0 , ( d ) ai2aj3Aij = A23.

1.80. Let the axes 0 x ~ x be


~ xrelated
~ to Oxlx,x3 by the table

( a ) Show t h a t ' t h e orthogonality conditions aijaik= S j k and apqaSq= a,, a r e satisfied.


( b ) W h a t a r e the primed coordinates of the point having position vector x = 2 ê , - ê3?
( c ) W h a t is the equation of the plane xl - x, + 3x3 = 1 in the primed system?
Ans. ( b ) ( 2 / 5 f i , 1115, - 2 1 5 f i ) ( c ) fiX ; - X; - 2fi X; = 1

h
1.81. Show t h a t the volume V enclosed by the surface S may be given a s V = Q
x ia the ponition vector and n the unit normal to the surfaee. Hint: Write V = (116)
and use (1.157). '
1 S
(xixi),inidS
Analysis of Stress

2.1 THE CONTINUUM CONCEPT


The molecular nature of the structure of matter is well established. --- In numero-
--

-investigations of material behavior,


-- however,
---- the individual molecule is of no conce- _and
-only tKe35havior of &e material as a whole is-deemed
. important. For these cases the
observed macroscopic behavior is usually explained by disregardinfmolecular considerations
and, instead, by assuming the material to be continuously distribuied throughout its volume
and to completely
- -- fi11 the space it occupies. This continuum concept of matter is the
-fundamental postulate of Continuum Mechanics. Within the limitations for which the
continuum assumption is valid, this concept provides a framework for studying the behavior
of solids, liquids and gases alike.

--Adofion
-- ------
of t_hecontinuum viewpoint
-- --
material behavior means that -field
as- the
-- -basis for the mathematical
__ _ _ such as stress
_- quantities and
- . -. -
displacement
description of
are expressed
-- --_-_r _ -
functions of the space coordinates and time.

2.2 HOMOGENEITY. ISOTROPY. MASS-DENSITY


A homogeneoousmaterial is-tne having-id~nticalpropertiesat a11 poi&. With respect
,to some
_- property, a material
____I_---
is isotropic
- --- -- -_ if that
- - - - A
property
- -is the same in a11 - -
directions at a
point. A material
-- is called
-- anisotropic
- -
with gspect to those properties
.
which are directional
-- -
a t a pÒi$t.
---
The concept of density is developed from
the mass-volume ratio in the neighborhood
of a point in the continuum. In Fig. 2-1 the
mass in the small element of volume AV is
denoted by AM. The average density of the
material within AV is therefore
- anl
P(av) - AV (2.1)
The density a t some interior point P of the
volume element AV is given mathematically
in accordance with the continuum concept by
the limit,
= lim -al)I = - dM
-
*"-O AV
-
d!' (2.2)
Mass-density p is a scalar quantity. Fig. 2-1
CHAP. 21 ANALYSIS O F STRESS

2.3 BODY FORCES. SURFACE FORCES


Forces are vector quantities which are best described by intuitive concepts such as push
or pull. Those forces which act on a11 elements of volume of a continuum are known a s
body forces. Examples are gravity and inertia forces. These forces are represented by the
symbol bi (force per unit mass), or as pi (force per unit volume). They are related through
the density by the equation
pbi = pi or pb = p (2.3)
Those forces which act on a surface element, whether it is a portion of the bounding
surface of the continuum or perhaps an arbitrary interna1 surface, are known as surface
forces. These are designated by f i (force per unit area). Contact forces between bodies
are a type of surface forces. .
,
'

2.4 CAUCHY'S STRESS PRINCIPLE. THE STRESS VECTOR


A material continuum occupying the region R of space, and subjected to surface forces
f i and body forces bi, is shown in Fig. 2-2. .As a result of forces being transmitted from
one portion of the continuum to another, the material within an arbitrary volume V
enclosed by the surface S interacts with the material outside of this volume. Taking ni
as the outward unit normal a t point P of a small element of surface AS of S, let Afi be the
resultant force exerted across AS upon the material within V by the material outside of
V. Clearly the force element Afi will depend upon the choice of AS and upon ni. I t should
also be noted that the distribution of force on AS is not necessarily uniform. Indeed the
force distribution is, in general, equipollent to a force and a moment a t P , as shown in
Fig. 2-2 by the vectors Afi and AMi.

Fig. 2-2 Fig. 2-3

The average force per unit area on AS is given by Af,lAS. The Cauchy stress principie
asserts that this ratio Afi/As tends to a definite limit d f i / d S as AS approaches zero a t the
point P-, .while a t the same time the moment of afiabout the point P vanishes
- - in-thgimiting
__h

process. The resulting vector d f i l d S (force per unit area) is called the stress vector ti"'
-and i s shown in Fig. 2-3. If the moment a t P were not to vanish in the limiting process,
a couple-stress vector, shown by the double-headed arrow in Fig. 2-3, would also be defined
a t the point. One branch of the theory of elasticity considers such couple stresses but
they are not considered in this text.
46 ANALYSIS O F STRESS [CHAP. 2

Mathematically the stress vector is defined by


dfi Or
A
f (n) = 1im
. Af
- =
df
dS as-o AS
The notation tln' (or tt;)) is used to emphasize the fact that the stress vector a t a given
point P in the continuum depends explicitly upon the particular surface element ASchosen
there, a s represented by the unit normal ni (or ^n). For some differently oriented surface
element, having a different unit normal, the associated stress vector a t P will also be
different. The stress vector arising from the action across A S a t P of the material within
A

V upon the material outside is the vector -t:"'. Thus by Newton's law of action and
reaction,
= - t' -
- "'
h

(2.5)
The stress v e c t o r is very often referred to as the t r a c t i o n v e c t o r .

2.5 STATE O F STRESS AT A POINT. STRESS TENSOR


At an arbitrary point P in a continuum, Cauchy's stress principie associates a stress
A

vector t:"' with each unit normal vector ni, representing the orientation of a n infinitesimal
surface element having P a s a n interior point. This is illustrated in Fig. 2-3. The
totality of a11 possible pairs of such vectors tiE;'and ni a t P defines the s t a t e of stress a t that
point. Fortunately it is not necessary to specify every pair of stress and normal vectors
to completely describe the state of stress a t a given point. This may be accomplished by
giving the stress vector on each of three mutually perpendicular planes a t P. Coordinate
transformation equations then serve to relate the stress vector on any other plane a t the
point to the given three.
Adopting planes perpendicular to the coordinate axes f o r the purpose of specifying
the state of stress a t a point, the appropriate stress and normal vectors are shown in
Fig. 2-4.

Fig. 2-4

For convenience, the three separate diagrams in Fig. 2-4 are often combined into a
single schematic representation a s shown in Fig. 2-5 below.
Each of the three coordinate-plane stress vectors may be written according to (1.69) in
terms of its Cartesian components as
CHAP. 21 ANALYSIS O F STRESS 47

Fig. 2-5 Fig. 2-6

The nine stress vector components, A


t:ei) E u..

are the components of a second-order Cartesian tensor known as the stress tensor. The
equivalent stress dyadic is designated by Z, so that explicit component and matrix rep-
resentations of the stress tensor, respectively, take the forms

Pictorially, the stress tensor components may be displayed with reference to the
coordinate planes as shown in Fig. 2-6. The components perpendicular to the planes
(all,a2,, a,,) are called normal stresses. Those acting in (tangent to) the planes (al2,a,,, a,,,
a,,, os?,o,,) are called shear stresses. A stress component is positive when it acts in the
pgsitive direction of the coordinate axes, and on a plane whose outer normal points in one
of the positive coordinate directions. The component uii acts in the direction of the jth
coordinate axis and on the plane whose outward normal is parallel to the ith coordinate
axis. The stress components shown in Fig. 2-6 are a11 positive.

2.6 THE STRESS TENSOR - STRESS VECTOR RELATIONSHIP


The relationship between the stress ten- h

sor uij a t a point P and the stress vector t,'"'


on a plane of arbitrary orientation a t that
point may be established through the force
equilibrium or momentum balance of a small
tetrahedron of the continuum, having its
vertex a t P. The base of the tetrahedron
is taken perpendicular to %, and the three
faces are taken perpendicular to the coor-
dinate planes as shown by Fig. 2-7. Des-
ignating the area of the base ABC as dS, the
areas of the faces are the projected areas,
dSl = dS nl for face CPB, dS2 = dS n2 for
face APC, dS3 = dS n3 for face BPA or Fig. 2-7
48 ANALYSIS OF STRESS [CHAP. 2

The average traction vectors -t:'&' on the faces and t:':' on the base, together with the
average body forces (including inertia forces, if present), acting on the tetrahedron are
shown in the figure. Equilibrium of forces on the tetrahedron requires that
tf(.) d S - tf'"' d s l - tf'"' d s 2 - tf'"' d s 3 + pbf d V = 0 (2.10)
If now the linear dimensions of the tetrahedron are reduced in a constant ratio to one
another, the body forces, being an order higher in the small dimensions, tend to zero more
rapidly than the surface forces. At the same time, the average stress vectors approach
the specific values appropriate to the designated directions a t P. Therefore by this limiting
process and the substitution (2.9), equation (2.10) reduces to

Cancelling the common factor d S and using the identity tje" =.u j i , (2.11) becomes

Equation (2.12) is also often expressed in the matrix form


rt$i = Ln1k] [vki]

which is written explicitly


1 '=12 u13
A A A

The matrix form (2.14) is equivalent to the component equations

2.7 FORCE AND MOMENT EQUILIBRIUM. STRESS TENSOR SYMMETRY


Equilibrium of an arbitrary volume V
of a continuum, subjected to a system of
surface forces t:;' and body forces bi (in-
cluding inertia forces, if present) as shown
in Fig. 2-8, requires that the resultant force
and moment acting on the volume be zero.
Summation of surface and body forces
results in the integral relation,
1 ti" d~ + 1 pbi d V = O
or (2.16 )
L t ( ^ ' d ~ + L ~ b =d ~O
A

Replacing tin' here by ujinjand converting


the resulting surface integral to a volume
integral by the divergence theorem o£ Gauss
(1.157). equation (2.16) becomes Fig. 2-8
CHAP. 21 ANALYSIS OF STRESS 49

Since the volume V is arbitrary, the integrand in (2.17) must vanish, so that
+ pbi =
ajijj O or V I + pb = O (2.18)
which are called the equilibrium equations.
In the absence of distributed moments or couple-stresses, the equilibrium of moments
about the origin requires that

in which xi is the position vector of the elements of surface and volume. Again, making
h

the substitution tin' = ujini, applying the theorem of Gauss and using the result expressed
in (2.18), the integrais of (2.19) are combined and reduced to

For the arbitrary volume V, (2.20) requires

Equation (2.21) represents the equations u,, = a,,, a,, -


- a,,, a,, = V,,, or in a11

which shows that the stress tensor is szjmmetric. In view of (2.22), the equilibrium equations
(2.18) are often written
+ ~ O
o ~ ~p b i, =

which appear in expanded form as

2.8 STRESS TRANSFORMATION LAWS


At the point P let the rectangular Cartesian
~ XPx;x:x~
coordinate systems P X I X and ~ of Fig.
2-9 be related to one another by the table of
direction cosines

Fig. 2-9
50 ANALYSIS OF STRESS [CHAP. 2

or by the equivalent alternatives, the transformation matrix [aijl, or the transformation


dyadic

According to the transformation law for Cartesian tensors of order one (1.93), the
A

components of the stress vector tln)referred to the unprimed axes are related to the primed
A

axes components tl'"' by the equation


A A
t;(P) = a..t(n) or t ' ( n ) -
- A. (2.26 )
J

Likewise, by the transformation law (1.102) for second-order Cartesian tensors, the stress
tensor components in the two systems are related by
O; = aiPajqap, or 1' = A . X A, (2.27)

In matrix form, the stress vector transformation is written


A

[ti:")]= [aii][t$] (2.28)


and the stress tensor transformation as
[a,] [apq][aqjI
LaijI (2.29)
Explicitly, the matrix multiplications in (2.28) and (2.29) are given respectively by

and

2.9 STRESS QUADRIC OF CAUCHY


At the point P in a continuum, let the stress
tensor have the values uij when referred to directions
parallel to the local Cartesian axes P61g253shown
in Fig. 2-10. The equation
uijgigj = fk2 (a constant) (2.32)
represents geometrically similar quadric surfaces
having a common center a t P. The plus or minus
choice assures the surfaces are real.
The position vector r of an arbitrary point lying
on the quadric surface has components gi = rni,
where ni is the unit normal in the direction of r.
At the point P the normal component aNni of the
A

stress vector tin' has a magnitude


= t i l ) n i = t ( D ) . n = ~ i j n i n j (2.33)
' J ~ Fig. 2-10
Accordingly if the constant k2 of (2.32) is set equal to uNr2,the resulting quadric
uij[icj= -iaNr2 (2.34)
CHAP. 21 ANALYSIS O F STRESS 51

is called the stress quadric of Cauchy. From this definition i t follows that the magnitude
O , of the normal stress component on the surface element dS perpendicular to the position

vector r of a point on Cauchy's stress quadric, is inversely proportional to r2,i.e. U ,= i-k2/r2.


Furthermore i t may be. shown that the stress vector t$' acting on dS a t P is parallel to the
normal of the tangent plane of the Cauchy quadric a t the point identified by r.

2.10 PRINCIPAL STRESSES. STRESS INVARIANTS. STRESS ELLIPSOID


At the point P for which the stress tensor com-
A

ponents are aij, the equation (2.12), t:"' = uiini, as- A

sociates with each direction ni a stress vector t,'"'.


Those directions f o r which tiA and ni are collinear as
shown in Fig. 2-11 are called principal stress direc-
tions. For a principal stress direction,
A h

tin) = &&i or t'"' -


A
an (2.35)

in which a, the magnitude of the stress vector, is


called a principal stress zjalue. Substituting (2.35)
into (2.12) and making use of the identities ni = Siinj
and uij = ajii results in the equations

(aij- Sija)ni= O or (1- lu) n = 0 . A


(2.36) Fig. 2-11

In the three equations (2.36), there are four unknowns, namely, the three direction cosines
ni and the principal stress value O.

For solutions of (2.36) other than the trivial one nj = 0, the determinant of coefficients,
vanish. Explicitly,
luij - Sijal, must

all - "12 u13

lu..
11
- S..
7 1
I ~
= 0 or a~~ u22 - u23 = 0 (2.,?7)
as 1 a32 O33 -

which upon expansion yields the cubic polynomial in a,

where I, = aii = tr 1 (2.39)

11, = +(aiiujj
- a.21.c..)
li

111, = laijl = det

are known respectively as the first, second and third stress invariants.
The three roots of (2.38), a,,,, a,,,, a(,) are the three principal stress values. Associated
with each principal stress a,,,, there is a principal stress direction for which the direction
cosines n,'k'are solutions of the equations
CHAP. 21 ANALYSIS O F STRESS 53

This resolution is shown in Fig. 2-13 where


the axes are chosen in the principal stress
directions and it is assumed the principal
stresses are ordered according to U, > u,, > u,,,.
Hence from (2.12), the components of t$) are
ti.^' = u1n1
=
'-'II~z (2.48)
h
t(n) =
3 u111n3
and from (2.33), the normal component mag-
nitude is
U, = u,n; +
~ , , n ; ~,,,n,2 +
(2.49)
Substituting (2.48) and (2.49) into (2.47), the
squared magnitude of the shear stress as a
function of the direction cosines ni is given by Fig. 2-13

The maximum and minimum values of U, may be obtained from (2.50) by the method of
Lagrangian rnultipliers. The procedure is to construct the function
F = U: - Anini (2.51)
in which the scalar A is called a Lagrangian multiplier. Equation (2.51) is clearly a function
of the direction cosines ni, so that the conditions for stationary (maximum or minimum)
values of F are given by JFldni = O. Setting these partials equal to zero yields the
equations

+ +
n2{uTI - 2uI1(uIn; ~ , , n ; ~,,,n;) + A} = O (2.523)
n3{uSII- 2 ~ , , , ( ~ , n+; ~ , , n +
; ~,,,n;)+ A} = O (2.52~)
which, together with the condition nini = 1, may be solved for A and the direction cosines
nl, ne, na, conjugate to the extremum values of shear stress.
One set of solutions to (2.52), and the associated shear stresses from (2.50), are
n1=&1, nz=O, n3=0; forwhich~,=O (2.53~)
nl = O, na = 11, n3 = 0; for which a, =O (2.533)
ni = O, nn = O, na = ? I ; for which U, =O (2.53~)
The shear stress values in (2.53) are obviously minimum values. Furthermore, since (2.35)
indicates that shear components vanish on principal planes, the directions given by (2.53)
are recognized as principal stress directions.
A second set of solutions to (2.52) may be verified to be given by
nl = 0, n2 = -
+ I , n3 = +l/fi;
- for which U, = (a,, - uIII)/2 (2.54a)
ni = *l/fi, n z = 0, n3 = *1/fi; for which a, = (oIII- uI)/2 (2.54b)
nl = 11/fi, nz = e l l f i , n3 = 0; for which U, = (aI - uII)/2 (2.54c)
Equation (2.54b) gives the maximum shear stress value, which is equal to half the difference
of the largest and smallest principal stresses. Also from (2.543), the maximum shear stress
component acts in the plane which bisects the right angle between the directions of the
xhaximum and minimum principal stresses.
54 ANALYSIS O F STRESS [CHAP. 2

2.12 MOHR'S CIRCLES FOR STRESS


A convenient two-dimensional graphical
representation of the three-dimensional state
of stress a t a point is provided by the well-
known Mohr's stress circles. In developing
these, the coordinate axes are again chosen in
the principal stress directions a t P as shown
by Fig. 2-14. The principal stresses are as-
sumed to be distinct and ordered according to
> O I I > u~~~ (2.55)
For this arrangement the stress vector tln) has
normal and shear components whose magni-
tudes satisfy the equations Fig. 2-14
U , +
= u1nS uI1n;+ uI1,n; (2.56)

Combining these two expressions with the identity nini = 1 and solving for the direction
cosines ni. results in the eauations

These equations serve as the basis for Mohr's stress circles, shown in the "stress plane" of
Fig. 2-15, for which the U , axis is the abscissa, and the U , axis is the ordinate.
In (2.58a), since U , - u,, > O and U , - u I I 1> O from (2.55), and since ( n ~ )is
' non-negative,
the numerator of the right-hand side satisfies the relationship

which represents stress points in the (V,, os) plane that are on or exterior to the circle

In Fig. 2-15, this circle is labeled CI.

Fig. 2-15
56 ANALYSIS OF STRESS [CHAP. 2

and, on the bounding circle arcs BC, CA and AB,


nl = cosx12 = O on BC, n2 = COSTIZ= O on CA, n3 = ~ 0 ~ x 1=2 O on AB
According to the first of these and the equation (2.58a), stress vectors for Q located on
BC will have components given by stress points on the circle CI in Fig. 2-15. Likewise, CA
in Fig. 2-16 corresponds to the circle C2, and AB to the circle C3 in Fig. 2-15.
The stress vector components O, and O, for an arbitrary location of Q rnay be deter-
mined by the construction shown in Fig. 2-17. Thus point e may be located on C3 by
drawing the radial line from the center of C3 a t the angle 2B. Note that angles in the
physical space of Fig. 2-16 are doubled in the stress space of Fig. 2-17 (arc AB subtends
90" in Fig. 2-16 whereas the conjugate stress points O, and O,, are 180" apart on C3). In
the same way, points g, h and f are located in Fig. 2-17 and the appropriate pairs joined by
circle arcs having their centers on the O, axis. The intersection of circle arcs ge and hf
h

represents the components and O, of the stress vector tin) on the plane having the normal
direction ni a t Q in Fig. 2-16.

Fig. 2-17

2.13 PLANE STRESS


In the case where one and only one of the principal stresses is zero a state of plane
stress is said to exist. Such a situation occurs a t an unloaded point on the free surface
bounding a body. If the principal stresses are ordered, the Mohr's stress circles will have
one of the characterizations appearing in Fig. 2-18.

=O
Fig. 2-18
CHAP. 21 ANALYSIS OF STRESS 57

If the principal stresses are not ordered and the direction of the zero principal stress
is taken as the X Q direction, the state of stress is termed plane stress parallel to the xlxz
plane. For arbitrary choice of orientation of the orthogonal axes X I and xz in this case,
the stress matrix has the form
ull @12 0
(2.65)

The stress quadric for this plane stress is a cylinder with its base lying in the XIXZ plane
and having the equation
u,,xS + +
2 u l z ~ , ~ ,u,,x~ = -Ck2 (2.66)
Frequently in elementary books on Strength of Materials a state of plane stress is rep-
resented by a single Mohr's circle. As seen from Fig. 2-18 this representation is necessarily
incomplete since a11 three circles are required to show the complete stress picture. In
particular, the maximum shear stress value a t a point will not be given if the single circle
presented happens to be one of the inner circles of Fig. 2-18. A single circle Mohr's diagram
is able, however, to display the stress points for a11 those planes a t the point P which include
the zero principal stress axis. For such planes, if the coordinate axes are chosen in
accordance with the stress representation given in (2.65), the single plane stress Mohr's
circle has the equation

The essential features in the construction of this circle are illustrated in Fig. 2-19. The
circle is drawn by locating the center C a t = (all + O,,)/:! and using the radius
+
R = j/[(u1, - uzz)/2]2 given in (2.67). Point A on the circle represents the stress
state on the surface element whose normal is nl (the right-hand face of the rectangular
parallelepiped shown in Fig. 2-19). Point B on the circle represents the stress state on
the top surface of the parallelepiped with normal nz. Principal stress points U , and a,, are
so labeled, and points E and D on the circle are points of maximum shear stress value.

Fig. 2-19

2.14 DEVIATOR AND SPHERICAL STRESS TENSORS


I t is very often useful to split the stress tensor uij into two component tensors, one of
which (the spherical or hydrostatic stress tensor) has the form
58 ANALYSIS O F STRESS [CHAP. 2

where U , = - p = ukk/3 is the mean normal stress, and the second (the deviator stress tensor)
has the form

This decomposition is expressed by the equations


uij = sijUkk/3 + Sij or 1 = UMI + tD
The principal directions of the deviator stress tensor Sij are the same as those of the
stress tensor uij. Thus principal deviator stress values are
-
S(k) - u(k) - a, (2.71)
The characteristic equation for the deviator stress tensor, comparable to (2.38) for the
stress tensor, is the cubic

I t is easily shown that the first invariant of the deviator stress tensor Ir, is identically
zero, which accounts for its absence in (2.72).

Solved Problems
STATE OF STRESS AT A POINT. STRESS VECTOR.
STRESS TENSOR (Sec. 2.1-2.6)
2.1. At the point P the stress vectors tin)
A

and t:"*' act on the respective surface


elements ni AS and n: AS*. Show that
the component of ti.^)in the direction
of n: is equal to the component of
t:^.*)in the direction of ni.
It is required to show that
h

t:^.*'n, =
A

Frorn (2.12) tln*'ni = ojin;ni, and b y (2.22)


oJ.i . = 0 . . so that
A
o..n*n.
31 1 I
= (o..n.)n*
1J 1 1
= t:")n; Fig. 2-20
CHAP. 21 ANALYSIS O F STRESS 59

2.2. The stress tensor values a t a point P are given by the array
7 o -2

-2 o 4
Determine the traction (stress) vector on the plane a t P whose unit normal is
íi = (213)êi - (213)$2(113)$3. +
From (2.12), t';' = n. 1. The multiplication is best carried out in the matrix form of (2.13):
7

7 o -2
-
A A A

[ t p ) ,t?', t!j")] = [ 2 / 3 , -213, 1/31


h
10A
Thus tCn' = 4 e , - 3 e,.

2.3. For the traction vector of Problem 2.2, determine (a) the component perpendicular
to the plane, (b) the magnitude of t i n ) , ( C ) the angle between tin)and G.
, + Qê3) =
A

( a ) tin' a = ( 4 2, - - Ae,) (%2, - ê, 4419

+ 10019
A

( b ) jtin'( = d 1 6 = 5.2
( c ) Since tia' 6 = t : ; ' [ cos O , cos O = ( 4 4 / 9 ) / 5 . 2 = 0.94 and e = 20'.

, , A

2.4. The stress vectors acting on the three coordinate planes are given by t l e l ' , tlez' and t i ê 3 ) .
Show that the sum of the squares of the magnitudes of these vectors is independent
of the orientation of the coordinate planes.
Let S be the sum in question. Then
s A A A
= t;ei)t:ei)+ t(.edt!ez)+ t!edt(ed
t t
A

a
A

i
A

which from (2.7) becomes S uliuli + uziu2i + u3iu3i= ujiuji, an invariant.


2.5. The state of stress a t a point is given by the stress tensor

where a, b, c are constants and u is some stress value. Determine the constants a, b
and c so that the stress vector on the octahedral plane (2 = (11fi)êl ( 1 l f i ) ê ~+ +
(11fl )ê3)vanishes.
A

I n matrix forrn, tin' = uijnj must be zero for the given stress tensor and normal vector.

Solving these equations, a = b = c = - 1 / 2 . Therefore the solution tensor is


-012 -012

-0/2 -012
60 ANALYSIS O F STRESS [CHAP. 2

2.6. The stress tensor a t point P is given by the 53


array

t : =
(1- 5
-
3
i)
1

Determine the stress vector on the plane passing


through P and parallel to the plane ABC shown
in Fig. 2 - 2 1 .
B 52
+
The equation of the plane ABC is 3x1 6x2+ 22, = 12,
and the unit normal to the plane is therefore (see Prob-
lem 1.2)
n = 3ê1 + d7 ê2 + 2s
A

7 3 Fig. 2-21
From (2.14),the stress vector may be determined by matrix multiplication, -

Thus

2.7. The state of stress throughout a continuum is given with respect to the Cartesian
axes 0 ~ 1 ~ 2 by
x 3 the array
3xix2 5x;
I: =
2x3

Determine the stress vector acting a t the point P ( 2 , 1 , fi)of the plane that is tangent
+
to the cylindrical surface x ; x3 = 4 a t P.
A t P the stress components a r e given by

The unit normal to the surface a t P is determined


+
from grad + = V+ = ~ ( 2 ; 23 - 4 ) . Thus V 4 =
+
2x2ê2 223ê3 and so

Therefore the unit normal a t P is =


ê ,h
-ê,.
2 2
This may also be seen in Fig. 2-22. Finally the stress
vector a t P on the plane I to R is given by

or t':' = 5 $,I2 + 3 ê2+ d ê 3 . Fig. 2-22


, CHAP. 21 ANALYSIS O F STRESS 61

EQUILIBRIUM EQUATIONS (Sec. 2.7)


2.8. For the distribution of the state of stress given in Problem 2.7, what form must the
body force components have if the equilibrium equations (2.24) are to be satisfied
everywhere.
Computing (2.24) directly from i given in Problem 2.7,
3x2 + 10x2 -k 0 + pbl = 0
O+O+2+pb2 = O
O+O+O+pb3 = O
These equations are satisfied when bl = -13x21p, b2 = -2Ip, b3 = 0.

2.9. Derive (2.20) from (2.19), page 49.


Starting with equation (2.19),

substitute t:" = oiini in the surface integral and convert the result to a volume integral by
(1.157):
~ ) ~ dV
( E ~ ~ d~ s x =~ o ~(eijkxjopk),p
~

Carrying out the indicated differentiation in this volume integral and combining with the first
volume integral gives
k ~ j ( ~ p k+
E i i k ( ~ j , p o p+ , p pbk)} dV = O

But from equilibrium equations, u ~+ pbk


~ -
=, 0 ;~ and since xi,, - this volume integral
reduces to (2.20),

\
STRESS TRANSFORMATIONS (Sec. 2.8)
2.10. The state of stress a t a point is given with respect to the Cartesian axes 0 ~ ~ byx 2 ~
the array

Determine the stress tensor Z' for the rotated axes OxíxSxl related to the unprimed
axes by the transformation tensor
llfi llfi

1 112
The stress transformation law is given by (2.27) a s ali = aiPaiqopq O r 1' = A A,. The de-
[aqi] given in
tailed calculation is best carried out by the matrix multiplication [ali] = [aip][opq]
(2.29). Thus
l l f i -1lfi
[o$] =
1 112 -112

o
62 ANALYSIS O F STRESS [CHAP. 2

2.11. Show that the stress transformation law may be derived from (2.33), the equation
uN = u i j n i n j expressing the normal stress value on an arbitrary plane having the unit
normal vector ni.
Since U N is a zero-order tensor, i t is given with respect to an arbitrary set of primed or
unprimed axes in the same form as
ohr = o!.n!nj
a3 1 = uijnin j
and since by (1.94) nE = aijnj,

U!.n!n! = o!.a. n a . n =
E3 Z 3 21 IP P 54 q "N = "PqnPnq

where new dummy indices have been introduced in the last term. Therefore

("!.a. a. -
23 IP 3 9 O
and since the directions of the unprimed axes are arbitrary,

2.12. For the unprimed axes in Fig. 2-23, the stress


tensor is given by

Uij = (i ; :) o o

Determine the stress tensor for the primed


axes specified as shown in the figure.
I t is first necessary to determine completely the
transformation matrix A. Since x; makes equal angles
with the xi axes, the first row of the transformation
table together with a,, is known. Thus Fig. 2-23

Using the orthogonality equations aijaik = Si,, the transformation matrix is determined by com-
puting the missing entries in the table. I t is left as an exercise for the student to show that
CHAP. 21 ANALYSIS O F STRESS 63

l/,h l l f i -1lfi
1 1 6 1 1 6 llfi
The result obtained here is not surprising when one considers the Mohr's circles for the state of
stress having three equal principal stress values.

CAUCHY'S STRESS QUADRIC (Sec. 2.9)


2.13. Determine the Cauchy stress quadric a t P for the following states of stress:
(a) uniform tension o,, = o,, = o,, = a ; o,, = u,, = a,, = O
( b ) uniaxial tension o,, = U ; o,, - -
- V~~ -
--
- u13=
(rl2 =O
-
( C ) simple shear o,, = o,, = T ; a,, - a,, = = a,, = o,, = O
(d) plane stress with a,, = u = o,,; -
- o;, - - T; -
- 03, -
- = 0.
From (2.32),the quadric surface is given in symbolic notation by the equation { * I J = *k2.

[i E I!]![
Thus using the matrix form,

a ) [{I, = ~ 1 :+ <+ = *k2

Hence the quadric surface for uniform tension is the sphere 5: + 522 + c3 2 = kk2lcr.

Hence the quadric surface for uniaxial tension i s a circular cylinder along the tension axis.

and so the quadric surface for simple shear is a hyperbolic cylinder parallel to the 5, axis.

(d) [Si?52,531 [ []
O 0 0
= 05; + 2 ~ ( ~ {+; 05: = k 2

and so the quadric surface for plane stress is a general conic cylinder parallel to the zero principal
axis.

ai i
2.14. Show that the Cauchy stress quadric for a state of stress represented by

is an ellipsoid (the stress ellipsoid) when a, b and c are a11 of the same sign.
The equation of the quadric is given by

53 -
Therefore the quadric surface is the ellipsoid bcg +g
a c +-
ab -
+k2
aL
bc '
64 ANALYSIS O F STRESS [CHAP. 2

PRINCIPAL STRESSES (Sec. 2.10-2.11)


2.15. The stress tensor a t a point P is given with respect to the axes 0 x 1 ~ 2 ~ ~ by the values

Determine the principal stress values and the principal stress directions represented
by the axes Oxtx,*x3.
From (9.37) the principal stress values o are given by

3-0 1 1
1 -a 2 = O or, upon expansion, (a + 2)(0- 4 ) ( o- 1 ) = O
1 2 -a

The roots are the principal stress values = -2, = 1, = 4. Let the x: axis be the
direction of o ( l ) , and let n;" be the direction cosines of this axis. Then from (2.42),
(3 + 2 ) n i 1 )+ n k l ) + n"'
3 = 0

Hence n:" = O.' n")


2
= -ni" and since nini = 1, =
(ni1))2 112. Therefore nil)= 0, nil' = l l f i ,
n i l ) = -i/dZ.
Likewise, let x: be associated with a ( 2 ) . Then from (2.42),
Zn:2' + n (22 ) + n ( 2 ) = 0

Finally, let xB be associated with Then from (2.42),


-n(3) + n(3) + = 0
2

so that ni3' = - 2 / f i , ni3)= - 1 1 6 , nF) = -1 / f i

2-16. show that the transformation tensor of direction cosines determined in Problem 2.15
transforms the original stress tensor into the diagonal principal axes stress tensor-
According to (2.29), [Gj]= [ a i p ] [ a P q ] [ a Qwhich
j], for the ~ r o b l e ma t hand becomeS

o -,h -2 o o

O 0 4
CHAP. 21 ANALYSIS O F STRESS 65

2.17. Determine the principal stress values and principal directions for the stress tensor

From (2.37),

Hence = O, <rc2)
~
= 0,
7

7 L
7-a

~
o(3)
a
T = O or (7 - a)[-27a

= 37. For o ( 3 )= 37, (2.42) yield


+ u2] + 27% = [37 - 0102 = 0.

and therefore n:3' = n:' = ni3)= l l f i . For o(,) = = 0, (2.42) yield

which together with nini = 1 a r e insufficient to determine uniquely the first and second principal
directions. Thus any pair of axes perpendicular to the n:3' direction and perpendicular to each
other may serve a s principal axes. For example, consider the axes determined in Problem 2.12, for
which the transformation matrix is
llfi llfi llfi
[aiil

According to the transformation law (2.29), the principal stress matrix [a*,] is given by

Show that the axes O x T x ~ x ~(where,


x,*, x3 and x: are in the same vertical
plane, and xy, xi and x2 are in the same
horizontal plane) a t e also principal axes
for the stress tensor of Problem 2.17.
The transformation matrix [aii] relating
the two sets of axes clearly has the known
elements

[aijl =
11,h 11,h llfi

a s is evident from Fig. 2-24. From the orthog- h 1

onality conditions aijaik= Sjk, the remaining


four elements are determined so that Fig. 2-24
66 ANALYSIS O F STRESS [CHAP. 2

I
-1lfi 11d5
1 1 fi16
116 11a lld3
O
As before,

2.19. Show that the principal stresses and the stress components oij for an arbitrary
set of axes referred to the principal directions through the transformation coefficients
aij are related by oij = 5 ap,apjy.
p=1

From the transformation law for stress (2.27), oij = apiaqjop,;* but since o:, are principal
stresses, there are only three terms on the right side of this equation, and in each p = q. There-
3
fore the right hand side may be written in form oij = 2
p=1
apiapjop.

2.20. Prove that uijuikokjis an invariant of the stress tensor.


By the transformation law (2.27),
f f /
aiPikakj = ai~ajqupqairaksarsakmajnomn

= (aipair)(ajqajn)(aksakm)~pq"rs"mn

= aprsqnssmopqarsamn
= (sprupq)(aqnamn)(8smurs)
- arqoqmarm
-
- aijoikokj

2.21. Evaluate directly the invariants IE,111, 1111for the stress tensor

Determine the principal stress values for this state of stress and show that the
diagonal form of the stress tensor yields the same values for the stress invariants.
From (2.39), IE = oii = 6 + 6 f 8 = 20.
From (2.40), IIE = ( 1 / 2 ) ( ~-~u~ ~0 ~ ~u ~ ~ )
= 01ia22 + O22O33 + 033011 - 412a12 - 023423 - 031a31
= 36+ 48 + 48 - 9 = 123.
From (2.41), IIIz = (oij(= 6(48) + 3(-24) = 216.
The principal stress values of aij are o1 = 3, oI1 = 8, a111= 9. In terms of principal values,
1, = a1 + o11 + = 3 + 8 + 9 = 20
0111

IIE = o1011 + + = 24 + 72 + 27
~11~111 ~ 1 1 1 ~ 1 = 123
1111 = OIO~IOIII = (24)9 = 216
CHAP. 21 ANALYSIS O F STRESS 67

2.22. The octahedral plane is the plane which makes equal angles with the principal stress
directions. Show that the shear stress on this plane, the so-called octahedral shear
stress, is given by
UOCT = 3d(uI
- ~11)' -I- (011 - ~ 1 1 1 ) ' + ( ~ 1 1 1- VI)'

With respect to the principal axes, the normal


to the octahedral plane is given by

Hence from (2.12) the stress vector on the octa-


hedral plane is

1
= -
6
+ 011ê2 + ~ ~ ~ $ 3 )

and its normal component is


,,x = n. t(n) = h(o1+ o ~ I+ o ~ ~ ~ ) Fig. 2-25

Therefore the shear component is

2.23. The stress tensor a t a point is given by uij = . Determine the maxi-

mum shear stress a t the point and show that it acts in the plane which bisects the
maximum and minimum stress planes.
From (2.38) the reader should verify that the principal stresses are o, = 10, o,, = 5,
= -15. From (2.51b) the maximum shear value is os = (oIII- 4 1 2 = -12.5. The principal
axes 0 x ~ x are
~ x related
~ to the axes of maximum shear 0 x ; x ~ xby~ the transformation table
below and are situated as shown in Fig. 2-26.

Fig. 2-26
68 ANALYSIS O F STRESS [CHAP. 2

The stress tensor referred to the primed axes is thus given by

1 o llfi
[a&] =
-12.5 O -2.5
The results here may be further clarified by showing the stresses on infinitesimal cubes a t the
point whose sides are perpendicular to the coordinate axes (see Fig. 2-27).

Fig. 2-27

MOHR'S CIRCLES (Sec. 2.12-2.13)


2.24. Draw the Mohr's circles for the state of stress discussed in Problem 2.23. Label
important points. Relate the axes 0 x 1 ~ (conjugate
~ ~ 3 to uij) to the principal axes
OX?x2 x9 and locate on the Mohr's diagram the points giving the stress states on the
coordinate planes of O X ~ X ~ X ~ .
The upper half of the symmetric Mohr's circles diagram is shown in Fig. 2-28 with the maxi-
mum shear point P and principal stresses labeled. The transformation table of direction cosines is

from which a diagram of the relative orientation of the axes is made a s shown in Fig. 2-29. The
x1 and 2; axes are coincident. x2 and x3 are in the plane of x: x; as shown. From the angles
a = 36.8O and fl = 53.2O shown, the points A ( - 6 , 1 2 ) on the plane L to x2 and B ( 1 , 1 2 ) on the
plane I to x , are located. Point C(5,O) represents the stress state on the plane L to x,.

Fig. 2-28 Fig. 2-29


CHAP. 21 ANALYSIS O F STRESS 69

2.25. The state of stress a t a point referred to axes 0 ~ 1 x 2 is


~ sgiven by
o

Determine analytically the stress vector components on the plane whose unit normal
+ +
is íi = (2/3)ê1 (1/3)ê2 (2/3)&. Check the results by the Mohr's diagram for
this problem.
From (2.13) and the symmetry property of the stress tensor, the stress vector on the plane of
A
n is given by the matrix product

Thus t':) = -10ê1/3 - 1 0 ê 2 - 10ê3/3; and from @.33), UN = t ( : ) hn = -7019. From (2.47),
a,= 70.719.
F o r oij the principal stress values a r e o, = 10, oI1 = -5, CIII = -15; and the principal axes
a r e related to 0 x l x 2 x 3by the table

Thus in the principal axes frame, n; = aijnj or Ili]


[: i53uj
O -315 415 213
= . Accordingly

tlie angles of Fig. 2-16 a r e given by 0 = ,8 = cos-I 213 = 48.2' and $ = cos-1113 = 70.5O, and
the Mohr's diagram comparable to Fig. 2-17 is a s shown in Fig. 2-30.

Fig. 2-30
70 ANALYSIS OF STRESS [CHAP. 2

2.26. Sketch the Mohr's circles for the three cases of plane stress depicted by the stresses
on the small cube oriented along the coordinate axes shown in Fig. 2-31. Determine
the maximum shear stress in each case.

Fig. 2-31

The Mohr's circles are shown in Fig. 2-32.

(b)
Fig. 2-32

SPHERICAL AND DEVIATOR STRESS (Sec. 2.14)


12 4 o
2.27. Split the stress tensor aij = into its spherical and deviator parts and

show that the first invariant of the deviator is zero.


UM = akk/3 (12 + 9 + 3 ) / 3 = 8. Thus

uij = uMSij + ~ i j =
o -2 -5
and sii = 4 + 1 -.5 = 0.

2.28. Show that the deviator stress tensor is equivalent to the superposition of five simple
shear states.
The decomposition is
CHAP. 21 ANALYSIS OF STRESS 71

where the last two tensors are seen to be equivalent to simple shear states by comparison of cases
(a) and (b) in Problem 2.26. Also note that since sii = 0, -sll - s~~= s ~ ~ .

2.29. Determine the principal deviator stress values for the stress tensor
10 -6 O

0 0 1

The deviator of oij is


from the determinant
sij = (--i -
O -6
g) and its principal values may be determined

Thus sI = 9, sII = -3, sIII = -6. The same result is obtained by first calculating the principal
stress values of oij and then using (2.71). For qj,a s the reader should show, a1 = 16, o11 = 4,
oII1 = 1 and hence sI = 16 - 7 = 9, sI1 = 4 - 7 = -3, sII1= 1- 7 = -6.

2.30. Show that the second invariant of the stress deviator is given in terms of its principal
+ +
stress values by 111, = (sIsI1 SIISIII SIIISI), or by the alternative form IIn, =
+ + sh).
sS1

In terms of the principal deviator stresses the characteristic equation of the deviator stress
tensor is given by the determinant
SI-s o o
o SI,-s o = (SI - s)(sII- s)(sIII- s) = o
O 0 SI11 - s

= s3 + (81811 + SII~III+ S I I I ~ I )-~ ~ 1 ~ 1 1 ~ 1 1 1


Hence from (2.72), I I n D = (sIsII + sI1sIII+ sIIIsI). + sII + sIII = 0,
Since sI

+ + 2811181
IInD = *(~SISII 2~11~111 - (SI + + 12) = -*(*I + + s : ~ ~ )
811 8111 SII

MISCELLANEOUS PROBLEMS
2.31. Prove that for any symmetric tensor such as the stress tensor uij, the transformed
tensor in any other coordinate system is also symmetric.
From (2.27), o; = aipajqo,,, = ajqaipaqp= aie
72 ANALYSIS O F STRESS [CHAP. 2

2.32. At the point P the principal stresses are such that 2 ~ , =


, ul u I I I . Determine the unit +
normal ni for the plane upon which U , = u,, and U , = (u, - ~,,,)/4.
2
From (2.33), oN = nloI + 2
+
ni(oI a I I I ) / 2 n,~,,, +
= (o, a I I I ) / 2 ; and since n: + + n: + ni = 1,
these equations may be combined to yield n, = n3. Next from (2.47),
a i = n,2 a 2, + n i ( a , + a I I I ) 2 / 4+ n3ai1, - (oI + o I I I ) 2 / 4= (aI - a I I I)2/16

Substituting n l = n, and ni - 1 = -nf - n; = -2nS into this equation and solving for n,, the direc-
tion cosines are found to be nl = 1 / 2 f i , n2 = f i / 2 , n3 = 1 / 2 f i . The reader should apply these

results to the stress tensor oij =

2.33. Show that the stress tensor uij may be decomposed into a spherical and deviatoric
part in one and only one way.
Assume two decompositions, <rij = Sijh -t sij = Si$* +
sTi with sii = O and sTi = O. Then
aii = 3X = 3h*, so
X = h*; and from ASij + +
sij = hSij s$ i t foilows that sij = s*,.

Prove that the principal stress values are real if 1 is real and symmetric.
For real values of the stress components the stress invariants are real and hence the coefficients
in (2.38) are a11 real. Thus by the theory of equations one root (principal stress) is real. Let a ( , )
be this root and consider a set of primed axes a: of which xi is in the direction of ( r c 3 ) . With respect to
aí1-. cri2 o
such axes the characteristic equation i s given by the determinant -o O = O o& a&
0 a(,) - 0 o
12
or ( a c 3 ) - o) [(ai1- o)(aS2- a ) - a12] = O. Since the discriminant of the quadratic in square brackets
+ +
D = (ail a i 2 ) 2 - 4[a;1a;12- ( o i 2 ) 2 ] = (aíl - o i 2 ) 2 4aii > 0, the remaining roots must be real.

Use the method of Lagrangian multipliers to show that the extrema1 values (maximum
and minimum) of the normal stress U , correspond to principal values.
From (2.33), U N = aijninj with qni = 1. Thus in analogy with (2.51) construct the function
H = oN - A n i n i for which aHlani = O. Then
- - aijni,pnj+ aijninj,p- 2hni,pni
8 % ~
= aijSipnj + oijniS, - 2XSipni
-
- apjnj + a,ni - 2XSiPni = 2(api- XSip)ni = O
which is equivalent t o equation (2.36) defining principal stress directions.

2.36. Assume that the stress components uij are derivable from a symmetric tensor field
+ij by the relationship uij = Q ~ ~ ~ c Show
~ ~ +that
~ in
~ the
, ~ absence
~ . of body forces the
equilibrium equations (2.23) are satisfied.
Using the results of Problem 1.58, the stress components are given by
oij = Sij($qq,pp - $qp,qp) $. $ p i , p j + +jp,pi - $pp, ji - @ji,pp
or explicitly 011 = $33.22 + #22,33 012 = a21 = -@33,21
-
a22 = $11,33+ @33,11 O23 = O32 - -@11,23
-
O33 = $22.11 + $11.22 a31 = "13 - -@22,13
CHAP. 21 ANALYSIS O F STRESS 73

Substituting these values into qj, = 0,


"11.1 + "12.2 + "13.3 = $33,221 + $22,331 - $33,212 - g 2 2 . 1 3 3 = O
"21,l f "22,2 + "23,3 = -@33,211 + g 1 1 . 3 3 2 + $33,112 - $11,233 = O
"31.1 + "32,2 f "33,3 = -@22,131 - g 1 1 . 2 3 2 f g 2 2 . 1 1 3 + +11,223 = O

2.37. Show that, as is asserted in Section 2.9, the normal to the Cauchy stress quadric
a t the point whose position vector is r is parallel to the stress vector tln).
Let the quadric surface be given in the form $ = oijlirj2 k2 = O. The normal a t any point is
then V g or &$/ali =
A
Hence $, = u ~ ~
oijriSjp +
S = 2uPiTi.
~ ~ li = rni, this becomes
~ Now~ since
2uPimi or 2 9 - ( ~=~ 2 ~t ' n~) . )

2.38. At a point P the stress tensor referred to axes 0 ~ 1 ~ 2 x 3 is given by


15 -10
c.. -
o 20
/ / r
If new axes O x l x 2 x 3 are chosen by a rotation about the origin for which the trans-
r315 O -4151
formation matrix is [aii] = I O 1 O I , determine the traction vectors on each
L _1
of the primed coordinate planes by projecting the traction vectors of the original axes
onto the primed directions. In this way determine a;. Check the result by the trans-
formation formula (2.27).
A

From (2.6) and the identity tte" = uij (2.7), the traction vectors on the unprimed coordinate
+
axes are t ' ; l ) = 15ê1 - 10ê,, t ( ê z ) = -10 ê1 5ê2, t ' ; 3 ) = 20 ê3 which correspond to the rows
of the stress tensor. Projecting these vectors onto the primed axes by the vector form of (2.12),
A

t(G) = nlt(el) + A

n2t'ez) + A
n 3 t ( e 3 ) , gives

which by the transformation of the unit vectors becomes

Likewise

and

so that
74 ANALYSIS O F STRESS [CHAP. 2

2.39. Slrow that the second invariant of the deviator stress tensor IIp, is related to the
octahedral shear stress by the equation a,,, = \/-311r,.
From Problem 2.22, <rOCT + +
= $ d ( o I - a I I ) 2 (o11 - o I I I ) 2 ( o I I I- a I ) 2 , and because o1 = U M + SI,
OII = 6111 +
s I 1 , etc.,
~ O C T= SI - ~11)' + ( S I I - SIII)' + ( S I I I - SI)^

Also s , + sII + sII1 = O and so (sl + s I I + s I I I ) 2= O or

Hence

2.40. The state of stress throughout a body is given by t h e stress tensor


Cx3
o.. =
o -cx1
where C is an arbitrary constant. (a) Show that the equilibrium equations are
satisfied if body forces are zero. (b) At the point P(4, -4,7) calculate the stress
+
vector on the plane 2x1 2x2 - X Q = -7, and on the sphere ( ~ 1 ) ( ~ ~ 2 ) ( ~ ~ 3 =) (9)2.
~ + +
( c ) Determine the principal stresses, maximum shear stresses and principal deviator
stresses a t P . (d) Sketch the Mohr's circles for the state of stress a t P.
(a) Substituting directly into (2.24) from aij, the equilibrium equations are satisfied identically.

( b ) From Problem 1.2, the unit normal to the plane 2 2 , 2x2 - x3 = -7 + is 3 1 + 2ê


fi = 22 3 2 - 1"
3e3.
Thus from (2.12) the stress vector on the plane a t P is

A A
The normal to the sphere xixi = (9)2 a t P i s ni = $,i with $ = xixi - 81, or n = 49 e 1 -.
+
$êz $ê3. In the matrix form of (2.14) the stress vector a t P is

( c ) From (2.37), for principal stresses o,


-o 7 o
7 -a -4 = o(a2 - 6 5 ) = 0; hence
O -4 -a
o1 = fi, oI1 = O , a I I 1 = -m.
The
maximum shear stress value is given by -
(2.54b) as os = ( o I I -I a I ) / 2= *&.
Since
the mean normal stress a t P is =
+
(oI4 o,, o I I 1 ) / 3= 0, the principal devia-
tor stresses are the same a s the principal
stresses.

(d) The Mohr's circles are shown in Fig. 2-33. Fig. 2-33
CHAP. 21 ANALYSIS O F STRESS 75

2.41.
Supplementary Problems

A t point P the stress tensor is aij = [r 2 .


.

Determine the stress vector on the

plane a t P parallel to plane ( a ) BGE, ( b ) BGFC of the small parallelepiped shown in Fig. 2-34.

Fig. 2-34

Ans. ( a ) t':' = 11 ê14- 12 ê2+ 9 ê 3 , ( b ) t '" +


= (21 2, f 14 ê2 21 ê , ) l f i

2.42. Determine the normal and shear stress components on the plane BGFC of Problem 2.41.
Ans. U N = 6315, os = 37.715

2.43. The principal stresses a t point P are o, = 12, o,, = 3, o,,, = -6. Determine the stress vector and
its normal component on the octahedral plane a t P.
Ans. t':' = (12êl+3ê2-6ê3)/fi, oN = 3

2.44. Determine the principal stress values for

(a) oij =
(i 1)
1 0 1

and show that both have the same principal directions.


and (b) oij =
(: 1:)
1 2 1

Ans. ( a ) a1 = 2, o l ~
= olI1= -1, ( b ) o, = 4 , 0 1 , = o,,, = 1

3 -10
4 Decompose the stress tensor oij = O 3:) into its spherical and deviator par. and
30 -27
determine the principal deviator stresses. Ans. S I = 31, SI, = 8, 8111 = -39

2.46. Show that the normal component of the stress vector on the octahedral plane is equal to one third
the first invariant of the stress tensor.

2.47. The stress tensor a t a point is .ven a s oij = 1


2 1 o
o
ozz
1
:) with unspecified. Determine 0.2

so that the stress vector on some plane a t the point will be zero. Give the unit normal for this
traction-free plane.
Ans. oz2 = 1, f; = ( S I - 2 ê 2 + ê 3 ) / 6 .
76 ANALYSIS O F STRESS [CHAP.2

2.48. Sketch the Mohr's circles and determine the maximum shear stress for each of the following
stress states:

(a) uij

Ans. ( a ) U , = 7, ( b ) os = 3712
= i:: )
0 0
(b) 0. =
o
-7 :
0 -27

2.49. Use the result given in Problem 1.58, page 39, together with the stress transformation law (2.27),
page 50, to show that ~ ~ ~ is an~invariant.
~ ~ ~ ~ u ~ ~ u ~

2.50. In a continuum, the stress field is given by the tensor

o
Determine ( a ) the body force distribution if the equilibrium equations are to be satisfied throughout
the field, ( b ) the principal stress values a t the point P(a, 0,2&), (c) the maximum shear stress a t
P , (d) the principal deviator stresses a t P .
Ans. ( a ) b, = -4x,, ( b ) a, -a, 8a, ( c ) 24.5~3,( d ) - 1 1 ~ 1 3 ,-5a/3,16a/3
Deformation and Strain

3.1 PARTICLES AND POINTS


In the kinematics of continua, the meaning of the word "point" must be clearly under-
stood since it may be construed to refer either to a "point" in space, or to a "point" of a
continuum. To avoid misunderstanding, the term "point" _will he u e d exclusively t-o
designate alõcation in fixed space. The word "particle" will denote a small vo1umetri.c
'element, or "material point", of a continuum. In brief, a point i s a place in space, a particle
is a small part of a material continuum.

3.2 CONTINUUM CONFIGURATION. DEFORMATION AND FLOW CONCEPTS


At any instant of time t, a continuum having a volume V and bounding surface S will
occupy a certain region R of physical space. The identification of the particles of the
continuum with the points of the space it occupies a t time t by reference to a suitable set of
coordinate axes is said to specify the configuration of the continuum a t that instant.
The term deformation refers to a change in the shape of the continuum between some
ini'tial (undeformed~configuration
--- ãnd a subsequent -(deformed)
- configuration. The emphasis
-
in deformation studies is on the initial and final configurations. No attention is given to
-\intermedia~configurations
---
-_ or
-I
-.------------I-----

_ _
L

.__-_I___
-CI

to the particular sequence


_____-of configurations
- -_ by which the
deformation o c c u r s-X y contrast, the word jlow is used to designate
-A----- - the
- - -continuigg-sme
- - ----
of motion-of a continuum. Indeed, a configuration history is inherent in flow investigations
-for which the specification of a time-dependent velocity field is given.

3.3 POSITION VECTOR. DISPLACEMENT VECTOR


In Fig. 3-1 the undeformed configuratiofi of a material continuum a t time t = O is
shown together with the deformed configuration of the same continuum a t a later time
t = t . For the present development i t is useful to refer the initial and final configurations
to separate coordinate axes as in the figure.

Fig. 3-1
78 DEFORMATION AND STRAIN [CHAP. 3

Accordingly, in the initial configuration a representative particle of the continuum


occupies a point POin space and has the position vector

with respect to the rectangular Cartesian axes OXlX2X3. Upper-case letters are used as
indices in (3.1) and will appear as such in severa1 equations that follow, but their use as
sixmmation indices is restricted to this section. In the remainder of the book upper-case
subscripts or superscripts serve as labels only. Their use here is to emphasize the connection
of certain expressions with the coordinates (XiXaX3), which are called the material coor-
dinates. In the deformed configuration the particle originally a t PO is located a t the point
P and has the position vector

when referred to the rectangular Cartesian axes oxixnx3. Here lower-case letters are used
as subscripts to identify with the coordinates (x1x2x3)which give the current position of
the particle and are frequently called the spatial coordinates.
The relative orientation of the material axes OXlX2X3 and the spatial axes 0 x 1 ~ 2 ~is3
specified through direction cosines a,, and a,,, which are defined by the dot products of
unit vectors as
A A A A -
e, I, = I; e, = a,, - a,, (3.3)
No summ:,tion is implied by the indices in these expressions since k and K are distinct
A h
indices. lnasmuch as Kronecker deltas are designated by the equations I K - I P= SKP and
A A
ekDep= 8kp, the orthoganality conditions between spatial and material axes take the form

Tn Fig. 3-1 the vector u joining the points Po and P (the initial and final' positions,
respectively, of the particle), is known as the displacement vector. This vector may be
expressed as A
u = ukek (3.5)
A

or alternatively as U = U,I,

in which the components UK and uk are interrelated through the direction cosines A
a,,.
From (1.89) the unit vector êk is expressed in terms of the material base vectors IK as
A
A
ek = ~ILKIK
Therefore substituting (3.7) into ( 3 4 ,

from which U~ = (3.9)

Since the direction cosines a,, are constants, the components of the displacement vector are
observed from (3.9) to obey the law of transformation of first-order Cartesian tensors, as
they should.

The vector b in Fig. 3-1 serves to locate the origin o with respect to O. From the
geometry of the figure,
u = b+x-X (3.1O)
CHAP. 31 DEFORMATION AND STRAIN 79

Very often in continuum mechanics it is possible to consider the coordinate systems


~ oxixexs superimposed, with b = O, so that (3.10) becomes
O X I X Z Xand
u = x-X (3.11)
In Cartesian component form this equation is given by the general expression

However, for superimposed axes the unit triads of base vectors for the two systems are
identical, which results in the direction cosine symbols a,, becoming Kronecker deltas.
Accordingly, (3.12) reduces to
U, = X , - X,$ (3.13)
in which only lower-case subscripts appear. In the remainder of this book, unless specifi-
cally stated otherwise, the material and spatial axes are assumed superimposed and hence
only lower-case indices will be used.

3.4 LAGRANGIAN AND EULERIAN DESCRIPTIONS


When a continuum undergoes deformation (or flow), the particles of the continuum
move along various paths in space. This motion may be expressed by equations of the
f orm

which give the present location xi of the particle that occupied the point (XIX2X3) a t time
t = O. Also, (3.14) may be interpreted as a mapping of the initial configuration into the
current configuration. I t is assumed that such a mapping is one-to-one and continuous,
with continuous partial derivatives to whatever order is required. The description of
motion or deformation expressed by (3.14) is known as the Lagrangian formulation.
If, on the other hand, the motion or deformation is given through equations of the form

in which the independent variables are the coordinates xi and t, the description is known
as the Eulerian formulation. This description may be viewed as one which provides a
tracing to its original position of the particle that now occupies the location ( x l ,X Z , ~ 3 ) . If
(3.15) is a continuous one-to-one mapping with continuous partial derivatives, as was also
assumed for (3.14), the two mappings are the unique inverses of one another. A necessary
and sufficient condition for the inverse functions to exist is that the Jacobian

should not vanish.


As a simple example, the Lagrangian description given by the equations

has the inverse Eulerian formulation,


Xl =
-XI + x2(et- 1)
1 - et - e-t
80 DEFORMATION AND STRAIN [CHAP. 3

3.5 DEFORMATION GRADIENTS. DISPLACEMENT GRADIENTS


Partial differentiation of (3.11) with respect to Xj produces the tensor dxildXj which is
called the material d e f o r m a t i o n gradient. In symbolic notation, dxildXj is represented by
the dyadic

a
in which the differential operator V, = --$idXi is applied from the right (as shown
explicitly in the equation). The matrix form of F serves to further clarify this property
of the operator V, when it appears as the consequent of a dyad. Thus

Partial differentiation of (3.15) with respect to x j produces the tensor dXi/dxj which is
called the spatial d e f o r m a t i o n gradient. This tensor is represented by the dyadic

having a matrix form

The material and spatial deformation tensors are interrelated through the well-known
chain rule for partia1 differentiation,

Partial differentiation of the displacement vector ui with respect to the coordinates


produces either the material displacement gradient a~Jax~, or the spatial displacement
gradient duiIdxj. From (3.13), which expresses u, as a difference of coordinates, these tensors
are given in terms of the deformation gradients as the material gradient

and the spatial gradient

In the usual manner, the matrix f o m s of J and K are respectively

and
CHAP. 31 DEFORMATION AND STRAIN
81

3.6 DEFORMATION TENSORS. FINITE STRAIN TENSORS


In Fig. 3-2 the initial (undeformed) and final (deformed) configurations of a continuum
are referred to the superposed rectangular Cartesian coordinate axes OX1X2X3and ox1x2x3.
The neighboring particles which occupy points POand &O before deformation, move to
points P and Q respectively in the deformed configuration.
'*

Fig. 3-2

The square of the differential element of length between POand QOis


( d X ) 2 .= d X * d X = d X i d X i = Gij d X i d X j (3.28)
From (3.15), the distance differential d X i is seen to be
dXi
d X i = =dxj or dX = H-dx
so that the squared length ( d X ) 2in (3.28) may be written

in which the second-order tensor

is known as Cauchy's deformation tensor. -

In the deformed configuration, the sqiare of the differential element of length between
P and Q is

From (3.14) the distance differential here is


axi
d x i = --dXj or d x = F-dX
dXj
so that the squared length ( d ~in) (3.32)
~ may be written
d ~ dk ~ k
( d ~ ) '= ---- dXi dXj = GijdXi d X j or (dx)' = d X . G . d X (3.34)
dXi d X j
in which the second-order tensor

is known as Green's deformation tensor


82 DEFORMATION AND STRAIN [CHAP. 3

The difference (dx)' - (dX)' for two neighboring particles of a continuum is used as
the measure of deformation that occurs in the neighborhood of the particles between the
initial and final configurations. If this difference is identically zero for a11 neighboring
particles of a continuum, a rigid displacement is said to occur. Using (3.34) and (3.28),
this difference may be expressed in the form

01- ( d ~ )- . .
' (dx)' = dX ( F , F - I) dX = dX -2Lc dX (3.36)

in which the second-order tensor

L.. - - or LG = &(Fc F - I)
/
is called the Lagrangian (or Green's) finite strain tensor.
*' /

Using (3.32) and (3.30), the same difference may be expressed in the form

in which the second-order tensor

is called the Eulerian (or Almansi's) finite strain tensor.


An especially useful form of the Lagrangian and Eulerian finite strain tensors is that
in which these tensors appear as functions of the displacement gradients. 'rhus if dxi/dXj
from (3.24) is substituted into (3.37), the result after some simple algebraic manipulations
is the Lagrangian finite strain tensor in the form

In the same manner, if dXi/dxj from (3.25) is substituted into (3.39), the result is the
Eulerian finite strain tensor in the form

The matrix representations of (3.40) and (3.41) may be written directly from (3.26) and
(3.27) respectively.

3.7 SMALL DEFORMATION THEORY. INFINITESIMAL STRAIN TENSORS


The so-called small deformation theory of continuum mechanics has as its basic condi-
tion the requirement that the displacement gradients be small compared to unity. The
fundamental measure of deformation is the difference (dx)' - (dX)', which may be expressed
in terrns of the displacement gradients by inserting (3.40) and (3.41) into (3.36) and
(3.38) respectively. If the displacement gradients are small, the finite strain tensors in
(3.36) and (3.38) reduce to infinitesimal strain tensors, and the resulting equations represent
small deformations.
CHAP. 31 DEFORMATION AND STRAIN 83

In (3.40), if the displacement gradient components duildXj are each small compared to
unity, the product terms are negligible and rnay be dropped. The resulting tensor is the
Lagrangian infinitesimal strain tensor, which is denoted by

or i. = i(uvx+ V, u) = S(J + Jc)


Likewise for duiIdxi < 1 in (3.41), the product terms rnay be dropped to yield the Eulerz'ccn
infinitesimal strain t e m o r , which is denoted by

If both the displacement gradients and the displacements themselves are small, there is
very little difference in the material and spatial coordinates of a continuum particle.
Accordingly the material gradient components duildXj and spatial gradient components
dui/dxj are very nearly equal, so that the Eulerian and Lagrangian infinitesimal strain
tensors rnay be taken as equal. Thus
l i . = r i . or L = E (3.44)
if both the displacements and displacement gradients are sufficiently small.

3.8 RELATIVE DISPLACEMENTS. LINEAR ROTATION TENSOR.


ROTATION VECTOR
In Fig. 3-3 the displacements of two neighboring Q
particles are represented by the vectors u:"o' and uiQO'
(see also Fig. 3-2). The vector

v
dui = u i < Q ~ ) - U ! P ~ ) or
1
du = ~ ( Q o )- u ( P o ) dx

is called the relative displacement vector of the particle


originally a t Qo with respect to the particle originally
a t Po. Assuming suitable continuity conditions on the P
displacement field, a Taylor series expansion for uiPO) U(PO)

rnay be developed in the neighborhood of Po. Neglect- po


ing higher-order terms in this expansion, the relative
displacement vector can be written as Fig. 3-3

Here the parentheses on the partia1 derivatives are to emphasize the requirement that the
derivatives are to be evaluated a t point Po. These derivatives are actually the components
of the material displacement gradient. Equation (3.46) is the Lagrangian form of the
relative displacement vector.
It is also useful to define the u n i t relative displacement vector d ~ l d Xin which d X is
the magnitude of the differential distance vector dXi. Accordingly if vi is a unit vector in
the direction of d X i so that d X i = V , d X , then

Since the material displacement gradient duildXj rnay be decomposed uniquely into a
symmetric and an antisymmetric part, the relative displacement vector dui rnay be written a s
84 DEFORMATION AND STRAIN [CHAP. 3

The first term in the square brackets in (3.48) is recognized as the linear Lagrangian strain
tensor lij. The second term is known as the linear Lagrangian rotation tensor and is
denoted by

In a displacement for which the strain tensor lij is identically zero in the vicinity of point
Po,the relative displacement a t that point will be an infinitesimal rigid body rotation.
This infinitesimal rotation may be represented by the rotation vector

in terms of which the relative displacement is given by the expression

The development of the Lagrangian description of the relative displacement vector, the
linear rotation tensor and the linear rotation vector is paralleled completely by an analogous
development for the Eulerian counterparts of these quantities. Accordingly the Eulerian
description of the relative displacement vector is given by

and the unit relative displacement vector by

Decomposition of the Eulerian displacement gradient auilaxj results in the expression

The first term in the square brackets of (3.54) is the Eulerian linear strain tensor eij. The
second term is the linear Eulerian rotation tensor and is denoted by

From (3.55), the linear Eulerian rotation vector is defined by

in terms of which the relative displacement is given by the expression

3.9 INTERPRETATION OF THE LINEAR STRAIN TENSORS


For small deformation theory, the finite Lagrangian strain tensor Lij in (3.36) may be
replaced by the linear Lagrangian strain tensor lij, and that expression may now be written
CHAP. 31 DEFORMATION A N D STRAIN 85

+
(dx)' - (dX)' = ( d x - d X ) ( d x d X ) = 2Ljd X i d X j
(dx)' - ( d X ) 2 = ( d -~d X ) ( d +
~ dX) = d x 21. d x (3.58)
Since d x - d X for small deformations, this equation may be put into the form

The left-hand side of (3.59) is recognized as the


change in length per unit original length of the
differential element and is called the normal strain
f o r the line element originally having direction
cosines dXildX.
When (3.59) is applied to the differential line
element PoQo, located with respect to the set of
local axes a t POas shown in Fig. 3-4, the result will
be the normal strain f o r that element. Because
POQO here lies along the X Z axis,
d X i l d X = dX3ldX = O, dXzldX = 1
and therefore (3.59) becomes

Fig. 3-4

Thus the normal strain for an element originally along the X z axis is seen to be the com-
ponent 122. Likewise for elements originally situated along the X i and X 3 axes, (3.59) yields
normal strain values l1l and 133 respectively. I n general, therefore, the diagonal terms of
the linear strain tensor represent normal strains in the coordinate directions.

Fig. 3-5

The physical interpretation of the off-diagonal terms of lij may be obtained by a con-
sideration of the line elements originally located along two of the coordinate axes. In
Fig. 3-5 the line elements POQOand POMOoriginally along the X2 and Xs axes, respectively,
become after deformation the line elements P Q and P M with respect to the parallel set of
local axes with origin a t P. The original right angle between the line elements becomes
the angle O. From (3.46) and the assumption of small deformation theory, a first order
approximation gives the unit vector a t P in the direction of Q as
86 DEFORMATION AND STRAIN [CHAP. 3

and, for the unit vector a t P in the direction of M, as

Therefore

or, neglecting the product temi which is of higher order,


duz dua
COS o = - + - = 2JZ3
aX3 aX2
Furthermore, taking the change in the right angle between the elements as y2, = 712 - O,
and remembering that for the linear theory r,, is very small, i t follows that
Y23 - sin yZ3 = sin ( ~ 1 -
2 O ) = cos O = 24, (3.65)
Therefore the off-diagonal terms of the linear strain tensor represent one-half the angle
change between two line elements originally a t right angles to one another. These strain
components are called shearing strains, and because of the factor 2 in (3.65) these tensor
components are equal to one-half the familiar "engineering" shearing strains.
A development, essentially paralleling the one just presented for the interpretation of
the components of lij, may also be made for the linear Eulerian strain tensor eii. The
essential difference in the derivations rests in the choice of line elements, which in the
Eulerian description must be those that lie along the coordinate axes after deformation.
The diagonal terms of cij are the normal strains, and the off-diagonal terms the shearing
strains. For those deformations in which the assumption lij = aii is valid, no distinction
is made between the Eulerian and Lagrangian interpretations.

3.10 STRETCH RATIO. FINITE STRAIN INTERPRETATION


An important measure of the extensional strain of a differential line element is the
ratio dxldX, known as the stretch or stretch ratio. This quantity may be defined a t either
the point POin the undeformed configuration or a t the point P in the deformed configuration.
Thus from (3.34) the squared stretch a t point POfor the line element along the unit vector
i% = dXldX, is given by

Similarly, from (3.30) the reciproca1 of the squared stretch for the line element a t P along
the unit vector íi= dxldx is given by

For an element originally along the local X2 axis shown in Fig. 3-4, = êz and
therefore dXildX = dXJdX = O, dXzldX = 1 so that (3.66) yields for such an element

Similar results may be determined for A:;,, and A:;,).


CHAP. 31 DEFORMATION AND STRAIN 87

For a n element parallel to the xz axis after deformation, (3.67) yields the result

with similar expressions f o r the quantities 1 / ~and


~ 1~1 ~ : ,~ In
~ )general,
. A,&, is not equal
to A,;2, since the element originally along the X2 axis will not likely lie along the x2 axis
after deformation.

The stretch ratio provides a basis for interpretation of the finite strain tensors. Thus
the change of length per unit of original length is

and f o r the element POQOalong the X2 axis (of Fig. 3-4), the unit extension is therefore

This result may also be derived directly from (3.36). For small deformation theory, (3.71)
reduces to (3.60). Also, the unit extensions L ( 1 )and L(,) a r e given by analogous equations
in terms of Ltt and L33 respectively.

F o r the two differential line elements shown in Fig. 3-5, the change in angle y,, = ~ 1-
26
is given in terms of A , ; ~ ,and A , ; ~ ,by

When deformations are small, (3.72) reduces to (3.65).

3.11 STRETCH TENSORS. ROTATION TENSOR


The so-called polar decomposition of a n arbitrary, nonsingular, second-order tensor is
given by the product of a positive symmetric second-order tensor with a n orthogonal second-
order tensor. When such a multiplicative decomposition is applied to the deformation
gradient F, the result may be written

in which R is the orthogonal rotation tensor, and S and T are positive symmetric tensors
known a s the right stretch tensor and left stretch tensor respectively.

The interpretation of (3.73) is provided through the relationship dxi = (dxi/dXi)dXj


given by (3.33). Inserting the inner products of (3.73) into (3.33) results in the equations

From these expressions the deformation of dXi into dxi a s illustrated in Fig. 3-2 may be
given either of two physical interpretations. In the first form of the right hand side of
(3.74), the deformation consists of a sequential stretching (by S ) and rotation to be followed
by a rigid body displacement to the point P. In the second form, a rigid body translation
to P is followed by a rotation and finally the stretching (by T). The translation, of course,
does not alter the vector components relative to the axes Xi and X i .
88 DEFORMATION AND STRAIN [CHAP. 3

3.12 TRANSFORMATION PROPERTIES OF STRAIN TENSORS


The various strain tensors Lij, Eij, lij and cij defined respectively by (3.379, (3.39), (3.42)
and (3.43) are a11 second-order Cartesian tensors as indicated by the two free indices in
each. AccordingIj for a set of rotated axes Xil having the transformation matrix [bij]with
respect to the set of local unprimed axes Xi a t point PO as shown in Fig. 3-6(a), the com-
ponents of L&and 1; are given by

and l!.
11 - b.I P b.34 G,4 or lf=B*L.B, (3.76)

(b)
Fig. 3-6

Likewise, for the rotated axes xi' having the transformation matrix [aij] in Fig. 3-6(b),
the components of E; and are given by

and C!. = a1P. a34. €P 4 or E'=A*EaAC (3.78)

By analogy with the stress quadric described in Section 2.9, page 50, the Lagrangian
and Eulerian linear strain quadrics may be given with reference to local Cartesian coor-
dinates q i and (Si a t the points Po and P respectively as shown in Fig. 3-7. Thus the
equation of the Lagrangian strain quadric is given by
CHAP. 31 DEFORMATION AND STRAIN 89

and the equation of the Eulerian strain quadric is given by

Two important properties of the Lagrangian {Eulerian) linear strain quadric are:
1. The normal strain with respect to the original {final) length of a line element is
inversely proportional to the distance squared from the origin of the quadric PO
{P) to a point on its surface.
2. The relative displacement of the nedghboring particle located a t QO {Q) per unit
original {final) length is parallel to the normal of the quadric surface a t the point
of intersection with the line through POQO{PQ).
Additional insight into the nature of local deformations in the neighborhood of POis
provided by defining the strain ellipsoid a t that point. Thus for the undeformed continuum,
the equation of the bounding surface of an infinitesimal sphere of radius R is given in
terms of local material coordinates by (3.28) as

After deformation, the equation of the surface of the same material particles is given by
(3.30) as
(dx)' = Cijdxidxj = R2 or ( d X ) 2 = d x . C . d x = R2 (3.82)
which describes an ellipsoid, known as the material strain ellipsoid. Therefore a spherical
volume of the continuum in the undeformed state is changed into an ellipsoid a t Po by the
deformation. By comparison, an infinitesimal spherical volume a t P in the deformed
continuum began as an ellipsoidal volume element in the undeformed state. For a sphere
of radius r a t P, the equations for these surfaces in terms of local coordinates are given
by (3.32) for the sphere as

and by (3.34) for the ellipsoid as

The ellipsoid of (3.84) is called the spatial strain ellipsoid. Such strain ellipsoids as
described here are frequently known as Cauchy strain ellipsoids.

3.13 PRINCIPAL STRAINS. STRAIN INVARIANTS. CUBICAL DILATATION


The Lagrangian and Eulerian linear strain tensors are symmetric second-order Cartesian
tensors, and accordingly the determination of their principal directions and principal
strain values follows the standard development presented in Section 1.19, page 20.
Physically, a principal direction of the strain tensor is one for which the orientation of an
element at a given point is not altered by a pure strain deformation. The principal strain
value is simply the unit relative displacement (normal strain) that occurs in the principal
direction.
For the Lagrangian strain tensor lij, the unit relative displacement vector is given by
(3.477, which may be written

n
Calling 1;"' the normal strain in the direction of the unit vector ni, (3.85) yields for pure
strain (Wij 0) the relation A A

1;"' = Ljnj or 1'"' = 1.2 (3.86)


90 DEFORMATION AND STRAIN [CHAP. 3

If the direction n, is a principal direction with a principal strain value 1, then


1:" = lni = /aijnj or 1';) = l$ = $102 (3.87)
Equating the right-hand sides of (3.86) and (3.87) leads to the relationship

which together with the condition nin+= 1 on the unit vectors ni provide the necessary
equations for determining the principal strain value 1 and its direction cosines ni. Nontrivial
solutions of (3.88) exist if and only if the determinant of coefficients vanishes. Therefore

which upon expansion yields the characteristic equation of lij, the cubic
P - IL12 + IIJ - IIIL = O (3.90)

where IL = &i = tr L, 1 = (ZiZj - Zjij), , IIIL = I&jl = det L (3.91)


are the first, second and third Lagrangian strain invar.iants respectively. The roots of
(3.90) are the principal strain values denoted by l(l), and z(31.

The first invariant of the Lagrangian strain tensor may be expressed in t e m s of the
principal strains as
IL= Zii = Z(1) Z(2) Z(3) + + (3.92)
and has an important physical interpretation. To see this, consider a differential rec-
tangular parallelepiped whose edges are parallel to the principal strain directions a s
shown in Fig. 3-8. The change in volume per unit original volume of this element is called
the cubical dilatation and is given by

For small strain theory, the first-order approximation of this ratio is the sum
Do = l(l>+ 1<2>+ Z(3) = IL (3.94)

Fig. 3-8
CHAP. 31 DEFORMATION AND STRAIN 91

With regard to the Eulerian strain tensor eij and its associated unit relative displacement
vector E:"', the principal directions and principal strain values C,,,,E,,,, C,,, are determined
in exactly the same way as their Lagrangian counterparts. The Eulerian strain invariants
may be expressed in terms of the principal strains as
IE = E(i, + + E<3,
IIE = '(1)'<2) '(2)'(3) + '(3)'(1)

IIIE = c(1)E(2, E ( 3 )

The cubical dilatation for the Eulerian description is given by

3.14 SPHERICAL AND DEVIATOR STRAIN TENSORS


The Lagrangian and Eulerian linear strain tensors may each be split into a spherz'cal
and deviator tensor in the same manner in which the stress tensor decomposition was
carried out in Chapter 2. As before, if Lagrangian and Eulerian deviator tensor com-
ponents are denoted by dij and eij respectively, the resolution expressions are

lij = dij+6..-Ikk or L = LD+-


I(tr L)
i3 3 3

and ci. = e..


L3
+ 6.. -
i33
'kk
or E = ED +-
l(tr E)
3
The deviator tensors are associated with shear deformation for which the cubical dilatation
vanishes. Therefore it is not surprising that the first invariants di and eii of the deviator
strain tensors are identically zero.

3.15 PLANE STRAIN. MOHR'S CIRCLES FOR STRAIN


When one and only one of the principal strains a t a point in a continuum is zero, a
state of plane strain is said to exist a t that point. In the Eulerian description (the
Lagrangian description follows exactly the same pattern), if x3 is taken as the direction
of the zero principal strain, a state of plane strain parallel to the x1x2plane exists and the

(i: i: ;)
linear strain tensor is given by

t,
E..
-
- or [c] = [; :i %] (3.99)

When xi and xz are also principal directions, the strain tensor has the form

Cij = k' €;)


o o
01 [Eij] = ['O) i] E.;
o

In many books on "Strength of Materials" and "Elasticity", plane strain is referred


to as plane deformation since the deformation field is identical in a11 planes perpendicular
to the direction of the zero principal strain. For plane strain perpendicular to the x3 axis,
the displacement vector may be taken as a function of xl and xz only. The appropriate
displacement components for this case of plane strain are designated by
92 DEFORMATION AND STRAIN [CHAP. 3

Ul = u1(x1,xz)
Uz = uz(x1,xz)
u3 = C (a constant, usually taken as zero)
Inserting these expressions into the definition of eij given by (3.43) produces the plane
strain tensor in the same form shown in (3.99).

A graphical description of the state of strain a t a point is provided by the Mohr's


circles for strain in a manner exactly like that presented in Chapter 2 for the Mohr's circles
for stress. For this purpose the strain tensor is often displayed in the form

Here the yij (with i # j ) are the so-called "engineering" shear strain components, which are
twice the tensorial shear strain components.
The state of strain a t an unloaded point on the ~ 1 2
'1

bounding surface of a continuum body is locally


plane strain. Frequently in experimental studies
involving strain measurements a t such a surface
point, Mohr's strain circles are useful for reporting
the observed data. Usually three normal strains are
measured a t the given point by means of a strain €1
rosette, and the Mohr's circles diagram constructed
from these. Corresponding to the plane stress
Mohr's circles, a typical case of plane strain diagram
is shown in Fig. 3-9. The principal normal strains
are labeled as such in the diagram, and the maxi-
mum shear strain values are represented by points
D and E. Fig. 3-9

3.16 COMPATIBILITY EQUATIONS FOR LINEAR STRAINS


If the strain components rii are given explicitly as functions of the coordinates, the six
independent equations (3.43)
cii -(-
1 8%
= 2 axj + 2)
may be viewed as a system of six partia1 differential equations for determining the three
displacement components ui. The system is over-deterrnined and will not, in general,
possess a solution for an arbitrary choice of the strain components cii. Therefore if the
displacement components ui are to be single-valued and continuous, some conditions must
be imposed upon the strain components. The necessary and sufficient conditions for such
a displacement field are expressed by the equations

There are eighty-one equations in a11 i n (3.103) but only six are distinct. These six written
in explicit and symbolic form appear as
CHAP. 31 DEFORMATION AND STRAIN

5. -
-
8x3 axi

d (dc23 -
6. - - + - - -'€31 -
~ X S8x1 ax2 axs 8x1 8x2

Compatibility equations in terms of the Lagrangian linear strain tensor lij may also be
written down by an obvious correspondence to the Eulerian form given above. For plane
strain parallel to the ~ 1 x 2plane, the six equations in (3.104) reduce to the single equation

where E is of the form given by (3.99).

Solved Problems
DISPLACEMENT AND DEFORMATION (Sec. 3.1-3.5)
3.1. With respect to superposed material axes Xi and spatial axes xi, the displacement
+
field of a continuum body is given by xl = Xi, 2 2 = X2 AX3, x s = X3 AX2 where +
A is a constant. Determine the displacement vector components in both the material
and spatial forms.
From (3.13) directly, the displacement components in material form are ul = xl - X l = 0,
u2 = x2 - X2 = AX3, u3 = x3 - X3 = AX,. Inverting the given displacement relations to obtain
X, = x,, X2 = (x, - Ax3)l(1- AZ), X3 = (2, - Ax2)l(1- AZ), the spatial components of u are
U, = O, u, = A(x3- Ax2)l(l- A2), ~3 = A(x2 - Ax3)l(1- A2).

From these results i t is noted that the originally straight line of material particles expressed
+ +
by X, = O, X2 X3 = lI(1+ A ) occupies the location xl = 0, x2 x3 = 1 after displacement.
Likewise the particle line X1 = O, X2 = X3 becomes after displacement x1 = 0, x2 = x3. (Inter-
pret the physical meaning of this.)

For the displacement field of Problem 3.1 determine the displaced location of the
material particles which originally comprise (a) the plane circular surface XI = 0,
+
XE X : = 1/(1- A2), (b) the infinitesimal cube with edges along the coordinate axes
of length d X i = d X . Sketch the displaced configurations for (a) and (b) if A = .S.
(a) By the direct substitutions X2 = (x, - Ax3)l(l - A2) and X3 = (x3 - Ax2)l(1- A2), the circular
surface becomes the elliptical surface (1+ A2)xl - 4Ax2x3 + (1+ A2)xi = (1 - A2). For
A = 4, this is bounded by the ellipse 5x22 - 8x223 + 5x3 = 3 which when referred to its
prinzipal axes x i (at 45O with xi, i = 2,3) has the equation xi2 +9 ~ =
; 3.
~ Fig. 3-10 below
shows this displacement pattern.
94 DEFORMATION AND STRAIN [CHAP. 3

Fig. 3-10 Fig. 3-11

( b ) From Problem 3.1, the displacements of the edges of the cube are readily calculated. For the
edge X , = X,, X 2 = X 3 = O, u , = u , = u3 = O. For the edge X 1 = O = X2, X 3 = X3,
u, = u3 = O , u2 = A X 3 and the particles on this edge are displaced in the X , direction propor-
tionally to their distance from the origin. For the edge X , = X 3 = O, X , = X,, u, = u, = 0 ,
u3 = AX,. The initial and displaced positions of the cube are shown in Fig. 3-11.

33. For superposed material and spatial axes, the displacement vector of a body is given
+ +
by u = 4x:ê1 X 2 X i ê 2 Xixiê3. Determine the displaced location of the particle
originally a t ( 1 , 0 , 2 ) .
The original position vector of the particle is X = ê, + 2êg. Its displacement is u = 4ê1 4 ê 3 +
and since x = X + u, its final position vector is x = 5ê1 6ê3. +
3.4. With respect to rectangular Cartesian ma-
terial coordinates Xi, a displacement field is
given by U l = - A X 2 X 3 , U2 = A X l X 3 , U 3 = 0
where A is a constant. Determine the dis-
placement components for cylindrical spatial
coordinates xi if the two systems have a com-
mon origin.
From the geometry of the a h s (Fig. 3-12) the
A
transformation tensor apK = e p IK is

cosx, sin x , O
aPK (-7x2 0)
and from the inverse form of (3.9) up = aPKUK.Thus
since Cartesian and cylindrical coordinates are related
through the equations X , = xl cos x,, X 2 = x1 sin x,,
X 3 = x,, equation (3.9) gives Fig. 3-12
+
ul = (-cos x2)AX,X3 (sin x 2 ) A X l X 3
+
= (-cos x2)Ax3xl sin 2 , (sin x2)Ax3x1cos x2 = O
u, = (sin x,)AX,X3 + (cos x,)AXlX3
+
= ( s i n ~ x 2 ) A x l x 3 (cos2 x2)Ax1x3 = A x 1 x 3
Ug = o
This displacement is that of a circular shaft in torsion.
CHAP. 31 DEFORMATION AND STRAIN 95

3.5. The Lagrangian description of a deformation is given by XI = XI X3(e2- I), +


+
x2 = X2 X3(e2- e-2), x3 = e2X3 where e is a constant. Show that the Jacobian
J does not vanish and determine the Eulerian equations describing this motion.
1 O (e2- 1)
From (.9.16), J = O 1 (e2- e-2) = e2 # 0.
O O (e2)
Inverting the equations, X1 = x1 + x3(e-2 - I), X2 = x2 + x3(e-4 - I), X3 = e-2x3.

3.6. A displacement field is given by u = +


êl x:x2ê2 x2x3ê3.Determine inde- +
pendently the material deformation gradient F and the material displacement gradient
J and verify (3.24), J = F - I.
From the given displacement vector u, J is found to be

+
Since x = u X, the displacement field may also be described by equations xl = X1(l + x:),
+
x2 = X2(1 x:), x3 = X3(l x:)+from which F is readily found to be

Direct substitution of the calculated tensors F and J into (3.26) verifies that the equation is satisfied.

3.7. A continuum body undergoes the displacement u = ( 3 x 2 - 4X3)ê1 (2x1- X3)$2 + +


(4x2- X1)ê3. Determine the displaced position of the vector joining particles A(1,0,3)
and B ( 3 , 6 , 6 ) , assuming superposed material and spatial axes.
+
From (3.13), the spatial coordinates for this displacement are xl = Xl 3X2 - 4X3, x2 = 2X1 +
X2 - X3, s3= -Xl + +
4X2 X3. Thus the displaced position of particle A is given by xl = -11,
x2 = -1, x3 = 2; and of particle B, xl = -3, x2 = 6, x3 = 27. Therefore the displaced position
of the vector joining A and B may be written V = 82, +
7ê2 25ê3. +

3.8. For the displacement field of Problem 3.7 determine the displaced position of the
position vector of particle C ( 2 , 6 , 3 ) which is parallel to the vector joining particles
A and B. Show that the two vectors remain parallel after deformation.
By the analysis of Problem 3.7 the position vector of C becomes U = 82, 7ê2 + + 25ê3 which
is clearly parallel to V. This is an example of so-called homogeneous deformation.

3.9. The general formulation of homogeneous deformation is given by the displacement


i = AijXj where the Aij are constants'or a t most functions of time. Show that
field u
this deformation is such that ( a ) plane sections remain plane, (b) straight lines
remain straight.
(a) From (3.13), xi = Xi + ui = Xi + AijXj = (a..+A..)X.
i.3 13 3
96 DEFORMATION AND STRAIN [CHAP. 3

+
According to (3.16) the inverse equations Xi = (aij B i j ) x j exist provided the determinant
+
Iaij Aijl does not vanish. Assuming this is the case, the material plane piXi cu = O becomes +
+ +
&(aij B i j ) x j cu = O which may be written in standard form a s the plane h j x j a. = O +
where the coefficients = &(aij Bij). +
( b ) A straight line may be considered a s the intersection of two planes. I n the deformed geometry,
planes remain plane a s proven and hence the intersection of two planes remains a straight line.

3.10. An infinitesimal homogeneous deformation ui = AijXj is one f o r which the coeffi-


cients Aij are so small that their products may be neglected in comparison to the
coefficients themselves. Show that the total deformation resulting from two suc-
cessive infinitesimal homogeneous deformations may be considered as the sum of the
two individual deformations, and that the order of applying the displacements does
not alter the final configuration.
+ +
Let xi = (aij A i j ) X j and x l = (aij B i j ) x j be successive infinitesimal homogeneous displace-
+ + + + +
ments. Then xl = (aij Bij)(Sjk A j k ) X k = ( S i k Bik Aik B i j A j k ) X k . Neglecting the higher
+ + +
order product terms B i j A j k , this becomes xl = ( S i k Bik A i k ) X k = (Oilc C i k ) X k which represents
the infinitesimal homogeneous deformation
uE' = XE - X i = CikXk = (Bik +Aik)Xk = (Aik + B i k ) X k = tii + U;

DEFORMATION AND STRAIN TENSORS (Sec. 3.6-3.9)


3.11. A continuum body undergoes the deformation xi = X i , x2 = X Z AX3, x3 = X3 AX2 + +
where A is a constant. Compute the deformation tensor G and use this to determine
the Lagrangian finite strain tensor LG.
From (3.35), G = F,* F and by (3.20) F i s given in matrix form a s

so t h a t Gij] = [i 1l
o
f 2
o

yAj
Therefore from (3.37), L~ = S ( G - I) =
2
O 2A A2

3.12. For the displacement field of Problem 3.11 calculate


the squared length ( d ~of) the
~ edges OA and OB,
and the diagonal OC after deformation for the
small rectangle shown in Fig. 3-13.
Using G a s determined in Problem 3.11 in (3.34), the
squared length of the diagonal OC is given in matrix
form by
o
( d ~ =
) ~ [O, dX2,dX3] O

=
[: l:Aj[il
1 +A2
2A

+ A"(dX2)2 + 4 A d X 2 dX:, + ( 1 + A"(dX:,)"


(1
Fig. 3-13

Similarly for O A , (dx)' = ( 1 + A2)(dX'2)2; and for O B , (dx)z = ( 1 + A2)(dX3)2.


CHAP. 31 DEFORMATION A N D S T R A I N 97

3.13. Calculate the change in squared length of the line elements of Problem 3.12 and check
the result by use of (3.36) and the strain tensor LG found in Problem 3.11.
Directly from the results of Problem 3.12, t h e changes are:
( a ) f o r OC, ( d ~ -) (dX)2
~ = ,(I+A Z ) ( ~ X ; dx;) + + 4 A dX, dX3 - ( d x ; + d x ; )
= A Z ( ~ X : dx;) + + 4 A dX, dX3
( b ) for OB, (dx)2 - (dX)2 = ( 1 + A ' ) d x ; - dx; = A2 d x ;

(c) f o r O A , (dx)2 - (dX)2 = ( 1 + A,) d x i - d ~ =


i A2 d ~ i .

By equation (3.36), f o r OC

(dx). - ( U ) 2 = O , dX2 -4 L:O


o
A2
2A
o'][A;]
A2 dXg
+
= A Z ( ~ X ~d x ; ) + 4 A dX, dX3

The changes f o r O A and O B may also be confirmed in the same way.

3.14. For the displacement field of Problem 3.11 calculate the material displacement
gradient J and use this tensor to determine the Lagrangian finite strain tensor LÇ.
Compare with resillt of Problem 3.11.
F r o m Problem 3.11 the displacement vector components a r e ul = O , u, = A X 3 , z(3 =AX, so
that

O O A2
Thus f r o m (3.40)

~ L G
O A O O A O O O A2 O 2A A2

t h e identical result obtained in Problem 3.11.

+
3.15. A displacement field is given by X I = X I A X 2 , x2 = X Z AX3, x s = X3 A X l where + +
A is a constant. Calculate the Lagrangian linear strain tensor L and the Eulerian
linear strain tensor E. Compare L and E for the case when A is very small.
F r o m (3.42),
/ 0 A O\ 10 O A \ /o A A\

Inverting the displacement equations gives


ul A(A2xl + x2 - A x 3 ) l ( l + A3), u2 = A ( - A x l + A2x2 + x 3 ) / ( 1+ A3),
~3 = A(xl -Ax2 + A2x3)/(1+ A3)
f r o m which by (3.43)

-i2)
A2 1 -A A2 - A
+ K,)
:,)
A
2E = (K =
1 +A3
+ t2
h(-:

When A i s very small, A? and higher powers may be neglected with the result t h a t E reduces to L.
98 DEFORMATION AND STRAIN [CHAP. 3

3.16. A displacement field is specified by u = X? X2ê1 (X2 - x ; ) ê 2 +


X; X3ê3. Determine +
the relative displacement vector du in the direction of the -XZ axis a t P ( 1 , 2 , -1).
Determine the relative displacements uai - UP for Q1(1,1, -I), Q2(1,3/2, -I),
Q3(1,7/4, -1) and Q4(1, 1518, -1) and compare their directions with the direction
of du.
For the given u, the displacement gradient J in matrix form is

~ a u i l ~ x i=

so that from (3.46) a t P in the -X2 direction,


[: 2XlX2 x;
-i]
+
Next by direct calculation from u, up = 2 ê1 ê2- 4 ê3 and UQ, - - êl - ê3. Thus
UQ, - up = -e, - e2 + 3 e,. Likewise, uQ2- up = - ê2(-e1 +
3.5 ê3)/2, ug3 - up = (- e, - e,
A A
+
+
3.75ê3)/4, uQ4- up = (- ê, - ê2 3.875ê3)/8. I t is clear that a s Qi approaches P the direction
of the relative displacement of the two particles approaches the limiting direction of du.

3.17. For the displacement field of Problem 3.16 determine the unit relative displacement
vector a t P ( 1 , 2 , -1) in the direction of Q(4,2,3).
The unit vector a t P in the direction of Q is $ = 3ê1/5 + 4ê3/5, so that from (3.47) and the
matrix of J a s calculated in Problem 3.16,

3.18. Under the restriction of small deformation theory, L = E. Accordingly for a dis-
placement field given by u = (xl - ~ 3 ) ~ ê (x2
l + +
~ ~ ) -~ x1x2ê3,
ê 2 determine the linear
strain tensor, the linear rotation tensor and the rotation vector at the point P(O,2, -1).
Here the displacement gradient is given in matrix form by

[auilaxj] =
2(x1- x3)
O
0
2(x2+ x3)
-a21 - x3)
2(x2 + 23)
1
which a t the point P becomes
r2
Decomposing this matrix into its symmetric and antisymmetric components gives

[eijI + [wijI =
-2 o -1 o
Therefore from (3.56) the rotation vector wi has components wl = -1, w2 = w3 = 0.
CHAP. 31 DEFORMATION AND STRAIN 99

3.19. For the displacement field of Problem 3.18 determine the change in length per unit
+
length (normal strain) in the direction of Y = (8êi - ê2 4ê3)/9 a t point P(O,2, -1).
From (3.59) and the strain tensor a t P a s computed in Problem 3.18, the normal strain a t P in
the direction of Y is the matrix product

3.20. Show that the change in the right angle between two orthogonal unit vectors G and í;
in the undeformed configuration is given by G.2L.fi for small deformation theory.
Assuming small displacement gradients, the unit vectors in the deformed directions of $ and $
+ +
a r e given by (3.47) a s ($ J * c ) and (li J .fi) respectively. (The student should check equations
(3.61) and (3.62) by this method.) Writing J in the equivalent form $ * J c and dotting the two
displaced unit vectors gives (as i n (3.63)), cos e = sin ( ~ 1 2 . -e) = sin y,, = y,, or y,, = [$ $ * J,] +
[ f i + J . C ] = $ . f i + G - ( ~ + ~ , ) . f i + Y . ~ , . ~ . f i . Here J,.J is of higher order f o r small displacement
gradients and since Y Ifi, 4. fi = O so t h a t finally by (3.42), y,, = $ 21. c.

3.21. Use the results of Problem 3.20 to compute the change in the right angle between
+ +
G = (8ê1- êz 4ê3)/9 and i; = (4ê1 4ê2- 7ê3)/9 a t the point P(O,2, -1) f o r the dis-
placement field of Problem 3.18.
Since L = E for small deformation theory, the strain tensor eij = lij and so a t P

y,, = [8/9, -119, 4/91


-719

STRETCH AND ROTATION (Sec. 3.10-3.11)


3.22. For the shear deformation XI = Xi, xz = XZ AX3, +
x 3 = X3 +
AX2 of Problem 3.11 show t h a t the
stretch A,;, is unity (zero normal strain) for line
elements parallel to the XI axis. For the diagonal
directions OC and D B of the infinitesimal square
OBCD (Fig. 3-14), compute A:;, and check the
results by direct calculation from the displacement
field.
From (3.66) and the matrix of G a s determined in
Problem 3.11, the squared stretch f o r & = êl is Fig. 3-14

Likewise for OC, & = ( ê 2 +&)I& and so


100 DEFORMATION AND STRAIN [CHAP. 3

From the displacement equations the deformed location of C is x, = 0, x2 = dL AdL, +


+
x, = dL AdL. Thus (dx)2 = 2(1+ A)2(dL)2 and since dX = fid ~ the
, squared stretch (dx/dX)2
+
is (1 A)2 a s calculated from (3.66).
2
Similarly, f o r DB, & = (-ê2+ ê 3 ) / f i and so A(m, = (1 - A ) 2 .

, h(;, a r e equal only if 2 is the deformed direction of A.


The stretch ratios n , ~ and
F o r the displacement field of Problem 3.22, calculate A(;, f o r & = ( $ 2 + & ) / f iand
show that it agrees with A : ~ , for the diagonal OC in Problem 3.22.
Inverting the displacement equations of Problem 3.22 one obtains
X1 = xl, X2 = (2, - Ax3)/(l- A2), X3 = ($3 - Ax2)l(1- A2)
from which the Cauchy deformation tensor C may be computed. Then using (3.67),

Thus A:;, +
= (1 - A2)2/(1- A)2 = (1 A)2 which is identical with A(m, calculated for OC. The
2

diagonal element OC does not change direction under the given shear deformation.

3.24. By a polar decomposition of the deformation gradient F f o r the shear deformation


+ +
xl = X1, x2 = X2 AX3, x3 = X B AX2, determine the right stretch tensor S together
with the rotation tensor R. Show that the principal values of S a r e the stretch ratios
of the diagonals OC and DB determined in Problem 3.22.
In the polar decomposition of F, the stretch tensor S = 6 ;and from (3.73), R = FS-1. By

r1 O O 1
(3.35), G = F, .
F or here [Gij] = 1 0 1 + A2 2A I. The principal axes of G a r e given by
L0 2A i t ~ 2 J
a 4 5 O rotation about X1 with the tensor in principal form [G~;] =

o o
Therefore [Sij] = [ G ] =

Relative to the coordinate axes Xi, the decomposition is

I n this example the deformation gradient F is its own stretch tensor S and R = i. This is the
result of the coincidence of the principal axes of LG and EA f o r the given shear deformation.

3.25. An infinitesimal rigid body rotation is given by ul = - C X 2 + B X 3 , us = C X l - A X 3 ,


u3 = - B X l +
A X 2 where A , B , C are very small constants. Show that the stretch
is zero (S = I) if terms involving squares and products of the constants are neglected.
F o r this displacement,
rl + c-+ B2 -A R -.c 1
CHAP. 31 DEFORMATION AND STRAIN 101

Neglecting higher order terms, this becomes

[Gijl =
O 0 1

STRAIN TRANSFORMATIONS AND PRINCIPAL STRAINS (Sec. 3.12-3.14)


+
3.26. For the shear deformation X I = XI, xz = X2 f i X 3 , x3 = X3 f i ~ 2 show that the +
principal directions of LG and Ea coincide a s was asserted in Problem 3.24.

From ( 3 . 3 7 ) [Lij] = [! i ]: o
which f o r principaI axes given by the transformation

matrix [aij = [:
o -1lfi
11.
o

11fi
becomes [L: = [i ;fi
1
o
.

Likewise from ( 3 . 3 9 , [Ed = which by the same transformation matrix Iaij]

r 0 O O 1
is converted into the principal-axes form [E:] = . The student should
verify these calculations.
o -l+fi

3.27. Using the definition (3.37), show that the Lagrangian finite strain tensor Lij trans-
forms a s a second order Cartesian tensor under the coordinate transformations
xi = bjixf and X( = bijXj.

By (3.37), Lij = -
2
1
íax,ax,
axk axk
--- Sij which by the stated transformation becomes

ax; axq d(b,,zL) ax;


b .b .S (since bpkbqk = S p q )
m Z n~ p q ax:, ax:, a,; ax,

3.28. A certain homogeneous deformation field results in the finite strain tensor

. Determine the principal strains and their directions f o r this

deformation.
Being a symmetric second order Cartesian tensor, the principal strains a r e the roots of
102 DEFORMATION AND STRAIN [CHAP. 3

Thus L ( , , = -2, L(2, = 2, Lo, = 8. The transformation matrix f o r principal directions is

[aijl =
1 1 216

3.29. For the homogeneous deformation xi = fiX I , x, = 2 x 2 , xj = fiX3 - X2 determine


the material strain ellipsoid resulting from deformation of the spherical surface
+ +
X: X: X: = 1. Show that this ellipsoid has the form xS/A:,, x;/Ak) x i l ~ ~=, 1.
, + +
By (3.82), o r alternatively by inverting the given displacement equations and substituting into
X i x i = 1, the material strain ellipsoid is xS + + +
xi x3 x2x3 = 3. This equation is put into
+
the principal-axes form xS13 x;/6 + x i l 2 = 1 by the transformation

= E[ o
11,h llfi

From the deformation equations, the stretch tensor S = 6 i s given (calculation is similar to
t h a t in Problem 3.24) a s
O 1

which by the transformation [aij] = is put into the principal form

with principal stretches A:~, = 3, A:,, = 6, A&) = 2. Note also t h a t the principal stretches may
be calculated directly from (3.66) using [aij] above.

3.30. For the deformation of Problem 3.29, determine the spatial strain ellipsoid and show
that i t is of the form A:,,x? +
A?,,X; t A:,,X~ = 1.
By (3.84) the sphere xixi = 1 resulted from the ellipsoid X G X = 1, o r

This ellipsoid is put into the principal-axes form 3 ~ : 6x22 + +2 : 1 by the transformation
~ =
CHAP. 31 DEFORMATION AND STRAIN 103

3.31. Verify by direct expansion that the second invariant IILof the strain tensor may be
expressed by
IIL =
I 111

1.1
112

122 1 /+ 111

131
113

133 1 1 122

132
123

133 1
Expansion of the given determinants results in IIL = ll1l2, 1,,13, + + 133111 - ( 1 f 2 + l i 3 + l i l ) .
I n comparison, direct expansion of the second equation of ( 3 . 9 1 ) yields
IIL = +[(111 + lZ2 + 133)ljj - ( l l j l l j + 12j12j + 13j13j)]
= &[(I11 + l22 + 133)(111 + l22 + l33) - ( l l l l l l + 112112 + 113113
+ l2l12l + l22l22 + l23l23 + 131131 + l32l32 + 133133)l
= 111122 + l22l33 f 133111 - (1S2 + 123 + 1321)

3.32. For the finite homogeneous deformation given by ui = AijXj where Aij a r e constants,
determine an expression for the change of volume per unit original volume. If the
Aij a r e very small, show that the result reduces to the cubical dilatation.
Consider a rectangular element of volume having original dimensions d X l , d X 2 , d X 3 along the
+
coordinate axes. F o r the given deformation, xi = ( A i j S i j ) X j . Thus by ( 3 . 8 3 ) the original volume
d V o becomes a skewed parallelepiped having edge lengths d x i = ( A i c n ) Si(,)) d X ( , , , n = 1,2,3. From+
+
( 1 . 1 0 9 ) this deformed element has the volume d V = e i j k ( A i l S i 1 ) ( A j 2 4- S j 2 ) ( A k 3 Sk3) d X 1 d X 2 d X 3 . +
Then

If the Aij a r e very small and their powers neglected,


AV/dVo = ~ ~ ~ ~ + S( i l A Aj 2 S k 3~ + ~S i l S8j 2 A k~3 +~8i18j28k3)
8 ~ -~ 1' = Ali + AZ2+ A33

F o r linear theory the cubical dilatation lii = auilaXi, which f o r ui = A i j X j is lii = A,, + A,, + A33,

3.33. A linear (small strain) deformation is specified by ul = 4x1 - xz 3x3, u z = x1 7x2, + +


+ +
us = -3x1 4x2 4x3. Determine the principal strains and the principal deviator
strains e,,,, for this deformation.
Since cij is the symmetrical p a r t of the displacement gradient a u i l d x j , i t i s given here by

strain deviator is eij =


or in principal-axes form by

t h a t e ( , ) = c(,) - e k k / 3 .
2 -1
i)
=

and its principal-axes form


. Also, €,,I3

e$ =
(1
O -1 u).
= 5 and so the

Note

PLANE STRAIN AND COMPATIBILITY (Sec. 3.15-3.16)


3.34. A 45" strain-rosette measures longitudinal strain along the
axes shown in Fig. 3-15. At a point P, C , , = 5 x lOP4,
C ; , = 4 x 10-4, = 7 x 10-4 inlin. Determine the shear
strain C,, a t the point.
1
By ( 3 . 5 9 ) , with = (êl+ê2)/\/S a s the unit vector in the si
direction, Fig. 3-15
104 DEFORMATION AND STRAIN [CHAP. 3

11. ,./I O] [:
5 x 10-4 e,,

7 .:Op4 ]-[';I
llfi
= 4 x 1Op4

Therefore
12 x 10-4 + 2 ~ , ,= 4 x 10-4 or ,€
, = -2 x 10-4.
2

3.35. Construct the Mohr's circles for the


case of plane strain 1 '"

and determine the maximum shear


strain. Verify the result analytically.
With the given state of strain referred
to the xi axes, the points B ( e Z Z= 5, €23 = & )
and D a r e established a s the diameter of the
larger inner circle in Fig. 3-16. Since
= O is a principal value f o r plane strain,
the other circles a r e drawn a s shown.
A rotation of 30' about the x , axis
(equivalent to 60° in the Mohr's diagram)
results in the principal strain axes with the
principal strain tensor given by Fig. 3-16

Fig. 3-1 7 Fig. 3-18

Next a rotation of 45O about the x: axis (90° in the Mohr diagram) results in the xl axes and
the associated strain tensor e / j given by

the first two rows of which represent the state of strain specified by point F in Fig. 3-16. Note
t h a t a rotation of -45O about xg would correspond to point E in Fig. 3-16.
CHAP. 31 DEFORMATION AND STRAIN 105

3.36. The state of strain throughout a continuum is specified by

Are the compatibility equations for strain satisfied?


Substituting directly into (3.104), a11 equati'ons a r e satisfied identically. The student should
c a r r y out the details.

MISCELLANEOUS PROBLEMS
3.37. Derive the indicial form of the Lagrangian finite strain tensor L, of (3.40) from its
definition (3.37).
From (3.24), dx,/dXi = S i j + aui/dXi. Thus (3.37) may be written

Li$ = [(Ski f -
"k)
ax, ( Skj + axj-
a?1i) -

= Ldui
2 ax,+-+--
au,
axi ax, axj
auk auk
I
3.38. A displacement field is defined by xl = X I - CX2 BX3, xz = C X i + Xz - AX3, +
+
x3 = - B X l + AX2 X3. Show that this displacement represents a rigid body rotation
only if the constants A, B, C a r e very small. Determine the rotation vector w f o r the
infinitesimal rigid body rotation.
1 -C B
F o r the given displacements, F = and from (3.37),

If products of the constants a r e neglected, this strain tensor i s zero and the displacement reduces
to a rigid body rotation. From (3.50), the rotation vector is
A A A
e1 e2 e3
W = -
2
â/dXl a/ax,
alax, = Aê, + sê, + cê3
-CXz + BX3 C X l - AX3 -BXl + AX,

3.39. F o r the rigid body rotation represented by ul = 0.02X3, uz = -0.03X3, u3 =


-0.02X1 + 0.03X2, determine the relative displacement of Q(3,0.1,4) with respect to
P(3, o, 4 ) .
A A A A
From the displacement equations, UQ = .O8 ê1- .12êz - .O57 e, and up = .O8 el - .12 e2- .O6 e3.
Hence du = UQ- UI>= -.003ê3. The same result is obtained by (3.51), with w = .03ê1 + .02ê,:
A A A
e1 e2 e3

du = .O3 .O2 O = -.003ê:,


o .1 o
106 DEFORMATION A N D S T R A I N [CHAP. 3

3.40. For a state o£ plane strain parallel to the ~ 2 x 3axes, determine expressions for the
normal strain r;, and the shear strain r;, when the primed and unprimed axes are
oriented as shown in Fig. 3-19.
By equation (3.59),

"42 = ~o,cosesinel[~
o
€J[:;:]
-
- E, cos2 e + 2e23sin e cos O i- sin2 o

-
- -'22 + €33 + - -2 eos 2e +
'22 €83
sin 20
2

Similarly from (3.65) and Problem 3.20,

4 =

=
[o, cos e, sin e,

-E,, sin O cos O


[.
+
o
em

E23

€23 C O S ~0
€7
E33
[-si!

- €23
COS
e]
0
sin2 e + 9, sin e cos e
"22 - "33
= €23 COS20 - -sin 20
2

Fig. 3-19 Fig. 3-20

3.41. For a homogeneous deformation the small strain tensor is given by

What is the change in the 90" angle ADC depicted by the small tetrahedron OABC
in Fig. 3-20 if OA = OB = OC, and D is the midpoint of A B ?
The unit vectors Y and a t D a r e given by f; = (êl - ê2)/\/S and 2 = (2 ê3- ê2-êl)/& . From
the result of Problem 3.20,
CHAP. 31 DEFORMATION A N D STRAIN 107

3.42. At a point the strain tensor is given by rij - i) and in principal form

. Calculate the strain invariants for each of these tensors

and show their equivalence.


+ +
By (3.95) and Problem 3.31, IE = 5 4 4 = 13, IE*= 6 4 3 = 13. Likewise IIE = + +
19 + +
19 16 = 54, 11,. = 24 + +
18 12 = 54. Finally IIIE = 5(16)- 4 - 4 = 72, 111,. =
(6)(4)(3)= 72. The student should check these calculations.

+
3.43. For the displacement field xl = X I AX3, x z = X z - AX3, x3 = X3 - A X l AX2, +
determine the finite strain tensor LG. Show that if A is very small the displacement
represents a rigid body rotation.
Since u , = AX3, u, = -AX3, u3 = -AXl + AX,, by (3.40),

If A is small so t h a t A2 may be neglected, LG = 0; and by (3.50)the rotation vector w = A ê1 A ê2. +

+
3.44. Show that the displacement field ul = A x l 3x2, u2 = 3x1 - Bxz, u3 = 5 gives a state
of plane strain and determine the relationship between A and B for which the
deformation is isochoric (constant volume deformation).

From the displacement equations, by (3.43), eii = 1 3 B O 1 which is of the form of


L 0 O 01
(3.99). From (3.96),the cubical dilatation is D eii = A - B, which is zero if A = B.

3.45. A so-called delta-rosette for measuring longitudinal


surface strains has the shape of the equilateral
triangle A and records normal strains E ~ ~ , E ; , E, ; ; in
i"/";
the directions shown in Fig. 3-21. If E , , = a, r l l = b,
e;; = c, determine E , , and E,, a t the point. 21

By (3.59)with L = E, f o r the x{ direction, Fig. 3-21

Also f o r the x;' direction

Solving simultaneously f o r e12 and ez2 yields e l z = ( b - c ) / 6 and E,Q = (-a + 2b + 2c)/3.
108 DEFORMATION AND STRAIN [CHAP. 3

3.46. Derive equation (3.72) expressing the change in angle between the coordinate direc-
tions Xp and X3 under a finite deformation. Show that (3.72) reduces to (3.65) when
displacement gradients are small.
Let y , ~= ~ / -2 8 be the angle change a s shown in Fig. 3-5. Then sin yz3 = cos (r12 - 8) =
A A
n, n,, or by (3.33) and (3.34)
sin Yzn =
dx2 -.-
dx3 -
-
dX, F, F dX3 . .
Idx21 ldxzl d d ~ ,G dX2 d d ~ ,G dX3
Now dividing t h e numerator and denominator of this equation by IdX21 and IdX31 and using (3.353
and (3.66) gives
sin yz3 =
A
A
e, G e, -
-
A
e, G ê, .
,I- ,I- A(:,) A(&]

A
Next from (3.37), e,. G -e3= e,A
(21, + I) *e3= ê2.2~~ e3 + e,
A A
l e3
A A A
= 2Lz3 since e2 e3 = 0.
Also from (3.68), A(;2) = d m , etc., and so

sin y?, = 2L23


d - d m

If aui/dX, 4 1 this reduces to sin yz3 = duz/aX3 + du3/ax2 = 2123.


3.47. For the simple shear displacement X I = XI, x2 = XZ, x3 = X3 + 2 ~ z / f i , determine
the direction of the line element in the XzX3plane for which the normal strain is zero.
Let & =im2ê, + m,ê, be the unit normal i n the direction of zero strain. Then from (3.66),
since AZcm, = 1,
o

or 7mE + +
4~ m2m3 3mi = 3. Also mi +
rn; = 1, and solving simultaneously m2 = ffi12,
m, = 7112, o r m2 = O, m3 = f 1. Thus there is zero strain along the X3 axis and f o r the element
a t 60' to the X3 axis.
A
The student should verify this result by using the relation & 2LG m = O derived from (3.36).

Supplementary Problems
3.48. F o r the shear displacement of Problem 3.47, determine the equation of the ellipse into which the
circle X: +X: = 1 is deformed. Ans. x: 9%; = 3+
3.49. Determine the shear angle y,, f o r the deformation of Problem 3.47
(Fig. 3-22). Ans. y,, = sin-12/$?

3.50. +
Given the displacement field x, = X, 2X3, x2 = Xz - 2X,, x, =
+
X3 - 2X1 2X2, determine the Lagrangian and Eulerian finite strain
tensors L, and E*.
2 -2 o 2 -2 o
x,
Fig. 3-22
CHAP. 31 DEFORMATION AND STRAIN 109

3.51. Determine the principal-axes form of the two tensors of Problem 3.50.

Ans. 1; = ( c 4).
4 o
E: = (: :)
419 o

419

' -i),
3.52. For the displacement field of Problem 3.50 determine the deformation gradient i, and by a polar

i:: (-i :-:'


decomposition of F find the rotation tensor R and the right stretch tensor S.
1 2 -1 o 2
Ans. R = ?3 ( 2 s = ,'B F =
-2 2

3.53. Show that the first invariant of 1, may be written in terms of the principal stretches as
ILG
-
- - 1 ) (h:ê2) + +
- 1 ) (h2,, - 1 ) ] / 2 . Hint: See equation (3.68).
(e31

3.54. The strain tensor a t a point is given by

in the direction of = $,I2 -$,I2 ê3/fi +


qj = (i.):- -3 fi

and the shear strain between


Determine the normal strain

and fl = -ê1/2 +
+
$212 $,/\/S. A n s . e ( AV ) = 6 , y,, = 0.
3.55. Determine the principal-axes form of eij given in Problem 3.54 and note that and fl of that
problem are principal directions (hence y,, = 0).

Ans. eij -
o -2

3.56. Draw the Mohr's circle for the state of strain giv'en in Problem 3.54 and determine the maximum
shear strain value. Verify this result analytically. Ans. y, =4

3.57. Using eij of Problem 3.54 and E


: given in Problem 3.55, calculate the three strain invariants from
each and compare the results. Ans. IE= 6 , IIE = -4, IIIE = -24.

3.58. For eij of Problem 3.54, determine the deviator tensor eij and calculate its principal values.

3.59. A displacement field is given by ul = 3 ~ ~ x 2 2u2


, = 2x3xl, u3 = x i - x1x2. Determine the strain
tensor eij and check whether or not the compatibility conditions are satisfied.
32: 3xlx2+ X3 -z2/2
o

A=
2112 2x3

3.60. For a delta-strain-rosette the normal strains


are found to be those shown in Fig. 3-23.
Determine e12 and ez2 a t the location.
Ans. ez2 = 1 X 10-4, e l z = -0.2885 X 10-4 r;l = 1 x 10-4 10-4

3.61. For the displacement field x l = X l AX,, +


x2 = X 2 , x3 = X 3 - A X 1 , calculate the vol-
ume change and show that i t is zero if A .I, = 2 x 10-4
is a very small constant. Fig. 3-23
Chapter 4
Motion and Flow

4.1 MOTION. FLOW. MATERIAL DERIVATIVE


Motion and flow are terms used to describe the instantaneous or continuing change in
configuration of a continuum. Flow sometimes carries the connotation of a motion leading
to a permanent deformation as, for example, in plasticity studies. In fluid flow, however,
the word denotes continuing motion. As indicated by (3.14) and (3.15), the motion
of a continuum may be expressed either in terms of material coordinates (Lagrangian
description) by
~i = x ~ ( X I , X ~ ,t)
X ~=, xi(X, t) or x = x(X, t)
or by the inverse of these equations in terms of the spatial coordinates (Eulerian description)

The necessary and sufficient condition for the inverse functions (4.2) to exist is that the
Jacobian determinant
J = laxiIaXil (4.3)
should not vanish. Physically, the Lagrangian description fixes attention on specific
particles of the continuum, whereas the Eulerian description concerns itself with a particu-
lar region of the space occupied by the continuum.
Since (4.1) and (4.2) are the inverses of one another, any physical property of the con-
tinuum that is expressed with respect to a specific particle (Lagrangian, or material descrip-
tion) may also be expressed with respect to the particular location in space occupied by the
particle (Eulerian, or spatial description). For example, if the material description of the
density p is given by
p = P(Xi,t) or p = p(X,t)

the spatial description is obtained by replacing X in this equation by the function given
in (4.2). Thus the spatial description of the density is
P = p(Xi(x,t), t) = p*(~i,
t) ~r p = p(X(x, t), t) = p*(x>t) (4.5)
where the symbol p* is used here to emphasize that the functional form of the Eulerian
description is not necessarily the same as the Lagrangian form.
The time rate of change of any property of a continuum with respect to specific particles
of the moving continuum is called the material derivative of that property. The material
derivative (also known as the substantial, or comoving, or convective derivative) may be
thought of as the time rate of change that would be measured by an observer traveling
with the specific particles under study. The instantaneous position xi of a particle is itself
a property of the particle. The material derivative of the particle's position is the instan-
taneous velocity of the particle. Therefore adopting the symbol dldt or the superpositioned
dot as representing the operation of material differentiation (some books use DIDt), the
( O )

velocity vector is defined by


vi = dxi/dt = X i or V = dxldt = X (4.6)
CHAP. 41 MOTION AND FLOW 111

. is any scalar, vector or tensor property of a continuum that may be


In general, if Pij,.
expressed as a point function of the coordinates, and if the Lagrangian description is
given by

the material derivative of the property is expressed by

The right-hand side of (4.8) is sometimes written ;jX7t)] X


to emphasize that the
X coordinates are held constant, i.e. the same particles are involved, in taking the derivative.
When the property Pij... is expressed by the spatial description in the form

the material derivative is given by

where the second term on the right arises because the specific particles are changing position
in space. The first term on the right of (4.10) gives the rate of change a t a particular

[dP'.ir9
Iocation and is accordingly called the local rate of change. This term is sometimes written
t)] to emphasize that x is held constant in this differentiation. The second term
X

on the right in (4.10) is called the convective rate of change since it expresses the contribu-
tion due to the motion of the particles in the variable field of the property.

From (4.6), the material derivative (4.10) may be written

which immediately suggests the introduction of the material derivative operator

which is used in taking the material derivati,ves of quantities expressed in spatial coordinates.

4.2 VELOCITY. ACCELERATION. INSTANTANEOUS VELOCITY FIELD


One definition of the velocity vector is given by (4.6) as vi = dxildt (or v = dxldt). An
alternative definition of the same vector may be obtained from (3.11) which gives xi = ui Xi +
+
(or x = u X). Thus the velocity may be defined by

since X is independent of time. In (4.13), if the displacement is expressed in the Lagrangian


form ui = ui(X, t), then
112 MOTION AND FLOW [CHAP. 4

If, on the other hand, the displacement is in the Eulerian form zci = u ~ ( xt),
, then

In (6.15) the velocity is given implicitly since i t appears a s a factor of the second term on
the right. The function
vi = vi(x,t ) or V = V(X, t) (4.16 )
is said to specify the instantaneous velocity field.
The material derivative of the velocity is the acceleration. If the velocity is given in
the Lagrangian form (4.14), then

If the velocity is given in the Eulerian form (4.15), then

4.3 PATH LINES. STREAM LINES. STEADY MOTION


A path 2ine is the curve or path followed by a particle during motion o r flow. A
stream line is the curve whose tangent a t any point is in the direction of the velocity a t that
point. The motion of a continuum is termed steady motion if the velocity field is independent
of time so that av,lat = O . For steady motion, stream lines and path lines coincide.

4.4 RATE OF DEFORMATION. VORTICITY. NATURAL STRAIN INCREMENTS


The spatial gradient of the instantaneous velocity field defines the velocity gradient
tensor, dvildxj (or Yij). This tensor may be decomposed into its symmetric and skew-
symmetric parts according to
y.. -dvi =
aXj

or Y = 9(vvx+ V xV ) + 4(vVx - VxV ) = (D+V) (6.19 )


This decomposition is valid even if vi and avildxj are finite quantities. The symmetric tensor

is called the rate of deformation tensor. Many other names are used for this tensor; among
them rate of strain, stretching, strain rate and velocity strain tensor. The skew-symmetric
tensor

is called the vorticity or spin tensor.


CHAP. 41 MOTION AND FLOW 113

The rate of deformation tensor is easily shown to be the material derivative o'f the
Eulerian linear strain tensor. Thus if in the equation

the differentiations with respect to the coordinates and time are interchanged, i.e. if

d t axj
(e)
is replaced by
dxj dt
, the equation takes the form

By the same procedure the vorticity tensor may be shown to be equal to the material
derivative of the Eulerian linear rotation tensor. The result is expressed by the equation

A rather interesting interpretation may be attached to (4.23) when that equation is


rewritten in the form
deij = Dijdt or dE = D d t
The left hand side of (4.25) represents the components known as the natural strain incre-
ments, widely used in flow problems in the theory of plasticity (see Chapter 8).

4.5 PHYSICAL INTERPRETATION OF RATE OF DEFORMATION


AND VORTICITY TENSORS
In Fig. 4-1 the velocities of the neighboring
particles a t points P and Q in a moving continuum
+
are given by vi and vi dvi respectively. The
relative velocity of the particle a t Q with respect
to the one a t P is therefore
aVi
dvi = -dxj or d v = vVX d x (4.26)
dxj
in which the partia1 derivatives are to be evalu-
ated a t P. In terms of Dij and Vij,(4.26) becomes
dvi = (Dij + Vij)dxj
O?.- dv = (D+V).dx (4.27) Fig. 4-1

If the rate of deformation tensor is identically zero (Dij = O),

and the motion in the neighborhood of P is a rigid body rotation. For this reason a velocity
field is said to be irrotational if the vorticity tensor vanishes everywhere within the field.

Associated with the vorticity tensor, the vector defined by


qi = cijk V k , j or q=v,xv (4.29)
is known as the vorticity vector. The symbolic form of (4.29) shows that the vorticity
vector is the curl of the velocity field. The vector defined as one-half the vorticity vector,
MOTION AND FLOW [CHAP. 4

ai = q = iikvk,j or a = &q = &Vxx v (4.30)


is called the rate of rotation vector. For a rigid body rotation such as that described by
(4.28), the relative velocity of a neighboring particle separated from P by dxi is given as
dvi = ~,,,n,dx, or d v = 0 x dx (4.31)
The components of the rate of deformation tensor have the following physical interpreta-
tions. The diagonal elements of Dii are known as the stretching or rate of eztension com-
ponents. Thus for pure deformation, from (S.27),
dv, = Dij dxj or dv = D d x (4.32)
and, since the rate of change of length of the line element dxi per unit instantaneous length
is given by
dvi
d i ( ~=) - dxi - A

= D..- - Dijvj or d'") = ü a $ (4.33)


dx a3 dx

the rate of deformation in the direction of the unit v,ector V , is


d = di(V'~i
= Dijvjvi or d = 3.D.D (4.34)
From (4.34), if vi is in the direction of a coordinate axis, say $2,
d = d22 or d = ê2 D .e2 = D22
A
(4.35)
The off-diagonal elements of Dij are shear rates, being a measure of the rate of change
between directions a t right angles (See Problem 4.18).
Since Dij is a symmetric, second-order tensor, the concepts of principal axes, principal
values, invariants, a rate of deformation quadric, and a rate of deformation deviator tensor
may be associated with it. Also, equations of compatibility for the components of the rate
of deformation tensor, analogous to those presented in Chapter 3 for the linear strain
tensors may be developed.

4.6 MATERIAL DERIVATIVES OF VOLUME, AREA AND LINE ELEMENTS


In the motion from some initial con-
figuration a t time t = O to the present
configuration a t time t, the continuum
particles which occupied the differential
volume element dVo in the initial state
now occupy the differential volume ele-
ment dV. If the initial volume element is
taken as the rectangular parallelepiped
shown in Fig. 4-2, then by (1.10)
dV = dXiêi X dXzê2 dX3ê3
= dXi dX2dXa (4.36)
Due to the motion, tkis parallelepiped is
moved and distorted, but because the
motion is assumed continuous the volume
element does not break up. In fact,
because of the relationship (3.33), dxi
(axilaXj)dXj between the material and
-
spatial line elements, the "line of particles" Fig. 4-2
CHAP. 41 MOTION AND FLOW 115

that formed dX1 now form the differential line segment dx,'" = (dxilaXi)dX1. Similarly
dX2 becomes d ~ : = ~ (axildxn)
' dX2 and dX3 becomes dx13' = (axiIdX3)dXa. Therefore the
differential volume element dV is a skewed parallelepiped having edges dx?', dx:", dxi3)and
a volume given by the box product
dV = dxC1)x dxC2) dxC3) = Éijk dxi(l)dxj2' dxk' (4.37)
But (4.37) is seen to be equal to
dV = c..
axi
--- dX d X 2d X 3 = J dVo
ax, ax, ax3 (4.38)
where J = laxiIdXjl is the Jacobian defined by (4.3).
Using (4.38), it is now possible to obtain the material derivative of dV as

d
since dVo is time independent, so that - (dVo)= O. The material derivative of the Jacobian
dt
J rnay be shown to be (see Problem 4.28)

and hence (4.39) rnay be put into the form

For the initial configuration of a continuum, a differential element of area having the
magnitude dSOrnay be represented in terms of its unit normal vector ni by the expression
dSoni. For the current configuration of the continuum in motion, the particles initially
making up the area d S O nnow fill an area element represented by the vector dSnt or dSi. It
rnay be shown that
ax.
dSi= J L d X j or dS= JdX*Xvx (4.42)
axi
from which the material derivative of the element of area is

The material derivative of the squared length of the differential line element dxi rnay
be calculated as follows,

However, since dxi = (axildXj)d X j ,

and (4.44) becomes


d dvi d
-(dx2)=2-dxkdxi or z(dx2) = 2dx.vXv.dx (4.46)
dt dxk
The expression on the right-hand side in the indicial form of (4.46) is symmetric in i and k,
and accordingly rnay be written
116 MOTION AND FLOW [CHAP. 4

or, from (4.20),


d d
- (dx2) = 2 Dij dxidxj ~r - (dx2) = 2 d ~D ' d~ (4.48)
dt dt

4.7 MATERIAL DERIVATIVES OF VOLUME, SURFACE AND LINE INTEGRALS


Not a11 properties of a continuum may be defined for a specific particle as functions of
the coordinates such as those given by (4.7) and (4.9). Some properties are defined as
'integrals over a finite portion of the continuum. In particular, let any scalar, vector or
tensor property be represented by the volume integral

P2j... (t) = Jv P < ; . . . ( x , ~ ) ~ v (4.49)

where V is the volume that the considered part of the continuum occupies a t time t. The
material derivative of Pij.. .(t) is

and since the differentiation is with respect to a definite portion of the continuum (i.e. a
specific mass system), the operations of differentiation and integration may be interchanged.
Therefore

which, upon carrying out the differentiation and using (4.411, results in

Since the material derivative operator is given by (4.12) as dldt = d/& + v~aldxp, (4.52)
may be put into the form

dt P;. . .(x, t) dV =
V
+ ~-(aX vPPt
P . . (x, t))] dV

By using Gauss' theorem (1.157), the second term of the right-hand integral of (4.53) may
be converted to a surface integral, and the material derivative then given by

P: ( x , t ) d v = . '(xft, dV
at
+ 1 v,[P:. . .(x, t)] dSp (4.54)

This equation states that the rate of increase of the property Pij...(t) in that portion of the
continuum instantaneously occupying V is equal to the sum of the amount of the property
created within V plus the flux V,[P;. . .(x, t)] through the bounding surface S of V.
The procedure for determining the material derivatives of surface and line integrals is
essentially the same as that used above for the volume integral. Thus for any tensorial
property of a continuum represented by the surface integral

where S is the surface occupied by the considered part of the continuum a t time t, then, as
bef ore,
& 1Q: . . .(x, t) dSp = i& [Q: . . .(x, t) dSp] (4.56)
CHAP. 41 MOTION AND FLOW 117

and, from (4.43), the differentiation in (4.56) yields


dQij ... ( t ) -
- [dQ:i. ( x ,t, + Q* (x,t ) ]d S p - [Q* -"d S p ] (4.57)
dt s dt 8% v . . . s . . dXq
For properties expressed in line integral form such as .

R 13.. .( t ) = IR; C
...(~ , t ) d x , (4.58)
the material derivative is given by

&1 R&. . ( x , t )d x p =12 [R:. . (r,t )d x p ] (4.59)

Differentiating the right hand integral as indicated in (4.59), and making use of (4.45),
results in the material derivative
d
-[R
dt
ij...(t)] = J C
d[R:. . .(x, t)]
dt dx, + 12 [ ~ ~ . . ( x , dt x) ,] (4.60)

Solved Problems
9

MATERIAL DERIVATIVES. VELOCITY. ACCELERATION (Sec. 4.1-4.3)


4.1. The spatial (Eulerian) description of a continuum motion is given by X I = X1et +
+
X3(et - I), x2 = X3(et - e c t ) X Z , x3 = X3. Show that the Jacobian J does not vanish
for this motion and determine the material (Lagrangian) description by inverting the
displacement equations.
By (4.9) the Jacobian determinant is
et O et-1
J = laxi/aXjl = O 1 et-e-t = et
O 0 1
Inverting the motion equations, X 1 = xle-t +
x3(e-t - I), X 2 = x2 - x3(et- e-t), X3 = x3.
Note that in each description when t = 0, xi = Xi.

4.2. A continuum motion is expressed by xl = X I , xz = et(X2 X3)/2 e c t ( X 2- X3)/2, + +


+
x3 = et(X2 X3)/2 - e-t(X2 - X3)/2. Determine the velocity components in both their
material and spatial forms.
+ +
From the second and third equations, X 2 X 3 = e-t(x, x,) and X , - X3 = et(x, - x,).
Solving these simultaneously the inverse equatjons become X 1 = x,, X , = e-t(x2+xj)/2 +
+
et(xz - x3)/2, X3 = e-t(x2 x3)/2 - et(x2- x3)/2. Accordingly, the displacement components ui =
zi - Xi may be written in either the Lagrangian form ul = O, u , = et(X2 X3)/2 e-t(X, - X3)/2 - + +
+
X2, u3 = et(X2 X3)/2 - e-t(X2 - X3)/2 - X3, or in the Eulerian form ul = O, u2 = x2 -
+ +
x3)/2 - et(x2- x3)/2, u3 = x3 - e-t(x2 x3)/2 et(xz- x3)/2. +
By (4.14), vi = aui/dt = üXi/ãt and the velocity components in Lagrangian form are v l = 0,
+ + +
v2 = et(X2 X3)/2 - e - f ( X 2 - X3)/2, v , = et(Xz X3)/2 e-t(X2 - X3)/2. Using the relationships
+
X z 4- X3 = e-t(x2 x3) and X 2 - X3 = et(x2- x3) these components reduce to v , = O, v , = x3,
v3 = 2 2 .
118 MOTION AND FLOW [CHAP. 4

Also, from (4.15), for the Eulerian case,


du~.ldt = v2 = + x3)/2 - et(x, - x3)/2 + v2(2 - e-t - et)/2 + v3(-e-t
C - ~ ( X ~ + et)/2
du$dt = v3 = e-Yxz + x3)12 + et(xz - %,)I2 + v2(-e-t + et)/2 + v3(2 - e-t - et)/2
Solving these equations simultaneously for v , and v,, the result is a s before v, = x,, v 3 = xs.

4.3. t), vn = 2x2/(1+ t), v3 = 3x3/(1+ t). Deter-


A velocity field is described by v1 = x ~ l ( l +
mine the acceleration components for this motion.

4.4. Integrate the velocity equations of Problem 4.3 to obtain the displacement relations
xi = xi(X, t ) and from these determine the acceleration components in Lagrangian
form for the motion.
+ +
By (4.13), v , = d x l / d t = x l l ( l t ) ; separating variables, dxllxl = d t l ( 1 t ) which upon inte-
+ +
gration gives ln x , = ln ( 1 t ) ln C where C is a constant of integration. Since xl = X , when
+ +
t = O, C = X , and so x1 = X l ( l t). Similar integrations yield z2 = X 2 ( l t ) 2 and x, = X3(1 t)3. + '

+
Thus from (4.14) and (4.17), v , = X 1 , v , = 2X2(1 t ) , v , = 3X3(1 + t)2 and a, = O, a, = 2X,,
+
a , = 6X3(1 t).

4.5. +
The motion of a continuum is given by xl = A (e-BAIX)sinA(A w t ) , x2 = -B - +
+
(ecBVX)cos X(A d ) , x3 = X3. Show that the particle paths are circles and that the
velocity magnitude is constant. Also determine the relationship between X l and X Z
and the constants A and B.
By writing x , - A = (e-BVk) sin x ( A ~ t ) x, , + +
B = (-e-BV/h) cos X(A ~ t ) then
, +
squaring
and adding, t is eliminated and the path lines are the circles ( x l -A)2 +
( x , B)2 = e-zBh/X. +
+ +
From (4.6), v l = oe-Bh cos h ( A wt), v , = we-BA sin h ( A wt), v , = O and v2 = v: v; vi = + +
+
~2e-zBA. Finally, when t = O, xi = Xi and so X , = A (e-Bh/X) sin x A , X , = -B - ( e - B ~ x cos
) xA.

4.6. A velocity field is specified by the vector v = xStêl x2t2ê2 x1x3tê3. Determine+ +
the velocity and acceleration of the particle a t P(1,3,2) when t = 1.
By direct substitution, v, = e,
A
+ 3 ê2+ 2 ê3. Using the vector form of (4.18) the accelera-
tion field is given by
a = xSê, + 2x2tê2 + x1x3ê3 + + x2t2ê2 + x1x3tê3)
(xStê,
+ x3tê,ê3 + t2ê$, + x1tê3ê3)
(2x1tê$,
or a = (xS + 2x1t2)ê, + (2x2t + %,e)ê2+ (51x3 + 2x x3t2)ê3

Thus ap = 3 ê l + 9 ê 2 + 6 ê , .

4.7. For the velocity field of Problem 4.3 determine the streamlines and path lines of the
flow and show that they coincide.
A t every point on a streamline the tangent is in the direction of the velocity. Hence for the
differential tangent vector d x along the streamline, v X d x = O and accordingly the differential
equations of the streamlines become d x l l v l = dx2/vz = dx3/v,. For the given flow these equations
are d x l l x l = dx,/2x2 = dx3/3x3. Integrating and using the conditions xi = Xi when t = 0, the
equations of the streamlines are ( x l / X l ) z= x,/X,, (x1/X1)3= x3/X3, (x2/X2)3= (x3/X3)2.
Integration of the velocity expressions dxi/dt = vi a s was carried out in Problem 4.4 yields
+ + +
the displacement equations x l = X l ( l t ) , x2 = X z ( l t)*, x3 = X 3 ( l t)3. Eliminating t from
these equations gives the path lines which are identical with the streamlines presented above.
C H A P . 41 MOTION AND FLOW 119

4.8. The magnetic field strength of an electromagnetic continuum is given by A = ecAtlr


where r2 = X : X; + +
xi and A is a constant. If the velocity field of the continuum
is given by vi = Bx1x3t, v2 = B x t2,
~ v3 = B x ~ x ~determine
, the rate of change of
magnetic intensity for the particle a t P(2, -1,2) when t = 1.
Since a(r-l)/axi = -xi/r3, equation (4.11) gives
i= -Ae-At/r - e-~t(Bx:x,t + B X ~ + Bx$x2)/r3
t2

Thus for P a t t = 1, i,= -e-A(3A + B)/9.

4.9. A velocity field is given by vi = 4x3 - 3x2, v2 = 3x1, v3 = - 4 ~ ~ Determine


. the
acceleration components a t P(b, O, 0 ) and &(O, 4b, 3b) and note that the velocity field
corresponds to a rigid body rotation of angular velocity 5 about the axis along
A
e +
= (4ê2 3ê3)/5.
From (4.18), a , = -25s1, a2 = -9x2 i12x3, a , = 12x2 - 162,. Thus a t P(b, 0 , O ) , a = -25bêl
which is a normal component of acceleration. Also, a t Q(O,4b, 3b) which is on the axis of rotation,
A
a = O . Note that v = w X x = ( 4 ~ 2 + 3 ~ 3 ) ~ ( x l ~ ~ + ~ 2 ê 2 + x 3 ~ 3 ) = ( 4 ~ 3 - 3 x 2 ) ~ 1 + 3 ~ 1 ~ 2 - 4

RATE OF DEFORMATION, VORTICITY (Sec. 4.4-4.5)


4.10. A certain flow is given by vi = O, v2 = A ( X I -
X xi)ecBt,
~ v3 = A ( x - ~ x1x3)ecBtwhere
A and B are constants. Determine the velocity gradient d v i l d x j for this motion and

"
from it compute the rate of deformation tensor D and the spin tensor V for the point
P(1,0,3) when t = 0.

By (4.19) avihxj = (-23


o
x,
22, -2,
which may be evaluated a t P when t = O

and decomposed according to (4.20) and (4.21) a s

4.11. For the motion xi = X l , x2 = X2 +


- I ) , x3 = X 3 X I ( ~-- 1) +
~ compute
~ the
rate of deformation D and the vorticity tensor V. Compare D with d e i j l d t , the rate
of change of the Eulerian small strain tensor E.
Here the displacement components are u , = O, u2 = %,(e-2t- 1), u3 = x,(e-3t- 1 ) and from
(4.14) the velocity components are v, = O, v2 = -2x1e-2t, v3 = -3x,e-3t. Decomposition of the
+
velocity gradient avilaxj gives avi/asj = Dij V i j . Thus

Likewise, decomposition of the displacement gradient gives aui/dxj = cij + uij. Thus
120 MOTION AND FLOW [CHAP. 4

Comparing D with dEldt,

!
-e-% -3e-3t/2
ds,ldt = -e-2t O = Dij
-3e-3tI2
O O o
The student should show that dwijldt = V i j .

4.12. A vortex line is one whose tangent a t every point in a moving continuum is in the
direction of the vorticity vector q. Show that the equations for vortex lines are
dxilqi = dx21q2 = dx31q3.
Let d x be a differential distance vector in the direction of q. Then q X d x E O , or
(9, dx3 - q3 dx,) 2, 3- (93 dxl - d ~ 3 $2
) (91 dxz - 9, d x i ) ê 3 E O
from which dxllql = dx21q2 = dx31q3.

4.13. Show that for the velocity field v = ( A x s - Bx2)êl (Bxl - Cx3)êz + + ( C X Z- Axl)ê3
the vortex lines are straight lines and determine their equations.
+ +
From (4.29), q = V , X V = 2 ( C ê l A 3, sê3),and by Problem 4.12, the d.e. for the vortex
lines are A dx3 = B dx,, B d x , = C dx,, C d x z = A dxl. Integrating these in turn yields the equa-
+ +
tions of the vortex lines x3 = Bx21A K1, x 1 = Cx31B K,, x z = A x J C K , where the Ki are +
constants of integration.

4.14. Show that the velocity field of Problem 4.13 represents a rigid body rotation by
showing that D = 0.
Calculating the velocity gradient avilaxj, i t is found to be antisymmetric. Thus avildxj =

(-; r )
O -B A
-f = V i j and Dij = O.

4.15. For the rigid body rotation v = 3x3ê1- 4x3& + (4x2- 3xl)ê3, determine the rate of
rotation vector O and show that v = O X x.
From (4.30), 2iZ = q, or i2 = 4 ê 1 + 3ê2. This vector is along the axis of rotation. Thus

( 4 ê l f 3 ê 2 )X ( ~ ~ ê ~ f ~ =
~ ê3x3ê1
~ + - 4x3ê2 +
~ ~ ê ( 4~x 2)- 3 x 1 ) ê 3 = V

4.16. A steady velocity field is given by v = ( x ; - xix:)êl (x: $2 xz)ê2. Determine the + +
unit relative velocity with respect to P(1,1,3) of the particles a t Q1(1,0,3),Q2(1,3/4,3),
Q3(1,7/8,3) and show that these values approach the relative velocity given by (4.26).
By direct calculation v , - v n , = -e1 2 e,,
h
+ h
4(vp - vQ2)= -7ê1/4 + 2ê2 and 8 ( v p- V Q3) =
-15$,/8 +
22,. The velocity gradient matrix is
3x1 - xp -2xix2
CHAP. 41 MOTION AND FLOW 121

and a t P(l,1,3) in the negative x2 direction,

Thus ( d v l d x ) ~= -22, + 2ê2 which i s the value approached by the relative unit velocities
vp - Vgi. e2

4.17. For the steady velocity field v = 3x;x2êl 2xix3ê2 ~ 1 x 2 ê3, + +


~ : determine the rate

of extension a t P(1,1,1) in the direction of $ = (3êl- 4ê3)/5.

Here the velocity gradient is [8vi/8xi] = [


6xlx2
O
x,x;
3%;
4x2x3 2::
xlx; 2x1x2x3
] and its symmetric part

Thus from (4.34) for 2 = ( 3ê1- 4ê3)/5,

4.18. For the motion of Problem 4.17 determine the rate of shear a t P between the orthog-
onal directions 2 = (3êi - 4&)/5 and fi = (4&+ 3ê3)/5.
In analogy with the results of Problem 3.20 the shear rate Y, is given by Y, = c *28 -2, or
in matrix form

Y, = [4/5,O, 3/51

4.19. A steady velocity field is given byv1 = 2x3, v2 = 2x3, 213 = O. Determine the principal

E : r]
directions and principal values (rates of extension) of the rate of deformation tensor
for this motion.

Here avi/axiI =
0

O
0
0
0
2

0
=
O
+ [ ;]
-1
o o
o o
-1
1
and for principal values

A of D,,

Thus X 1 = + f i , AII=O, XI,l=-fi

The transformation matrix to principal axis directions is


-112 -112 l/fi
122 MOTION AND FLOW [CHAP. 4

with the rate of deformation matrix in principal form [D;] =


o -fi
4.20. Determine the maximum shear rate :, for the motion of Problem 4.19.
Analogous to principal shear strains of Chapter 3,
the maximum shear rate is, , ,? = ( A 1 - A I I 1 ) / 2 = fi.
This result is also available by observing that the
motion is a simple shearing parallel to the x l x 2 plane in
the direction of the unit vector 2 = (êl+$,)I@. Thus,
a s before,

i., = 7," = [0,0,11

I t is also worth noting that the maximum rate of


extension for this motion occurs in the direction
A
+ +
n = ( ê l ê2 fiê3)/2 a s found in Problem 4.19. Thus Fig. 4-3

MATERIAL DERIVATIVES OF VOLUMES, AREAS, INTEGRALS, ETC. (Sec. 4.6-4.7)


Calculate the second material derivative of the scalar product of two line elements, i.e.
4.21.
determine d2(dx2)ldt2.
d(dxi) avi d(dx2)
From (4.451, 7 = -d x k ; and i t is shown in (4.48) that - = 2Dii dxi dxj. Therefore
83, dt

and by simple manipulation of the dummy indices,

4.22. Determine the material derivative pidSi of the flux of the vector property pi
dt
through the surface S.
BY (4.5%

4.23. Show that the transport theorem derived in Problem 4.22 may be written in symbolic
notation as
$A P * & ~ s = [$ + v(T7.p) + V x ( p x v ) ] . & d s
By a direct transcription into symbolic notation of the result in Problem 4.22,
CHAP. 41 MOTION AND FLOW 123

Now use of the vector identity V X ( p X v) = p(V v) - v(V p) + (v V ) p - ( p V ) v (see Problem


1.65) gives
it p - ~ d s= L ["z+ v(vop) + v x ( p x V ) ]~ ~ n d s

4.24. Express Reynold's transport theorem a s given by equations (4.53) and (4.54) in
symbolic notation.
Let P*(x, t ) be any tensor function of the Eulerian coordinates and time. Then (4.53) is

.and by Gauss' divergence theorem this becomes (4.54),

4.25. If the function P*(x, t ) in Problem 4.24 is the scalar 1, the integral on the left is
simply the instantaneous volume of a portion of the continuum. Determine the
material derivative of this volume.
Using the vector form of (4.58) a s given in Problem 4.24, $i i dV = V . v dV. Here

V v d V represents the rate of change of dV, and so V v is known a s the cubical rate of dilata-
tion. This relationship may also be established by a direct differentiation of (4.38). See Problem 4.43.

MISCELLANEOUS PROBLEMS
4.26. From the definition of the vorticity vector (4.29), q = curl v, show that qi = e i j k V k j
and that 2Vii = ejikqk.
By (4.29), qi = ~ ~= ~ ~ ~ ~ +
~ v (and ~, ~ ~ , =
~v since ~ O ~(see, for example, Problem
= (GriSsk - S r k S s i ) V k j = 2Vsr.
1.50), qi = qjkV[k,jl = eiikVkj. From this result q,,qi = eirsYjkVkj

av
4.27. Show that the acceleration a may be written a s a = -
dt
+ q x v + +Vv2.
avi avi
From (4.18), ai = -
at + v , axk
- and so

which, a s the student should confirm, is the indicial form of the required equation.

4.28. Show that d(ln J)ldt = div v.


Let axi/aXp be written here a s si,, so that J = eP~Rx1,P%2,Q%3,R and J becomes the sum of
the three determinants, +
= f p Q R ( j S l , p ~ 2 , Q ~ 3 ,x1,pjS2,Qx3,R
R +
c ~ ~ , p x ~ , ~ j NOW S ~ , ~ ) . = vl,,x,,p,
ek., and so J = E P Q R ( ~ ~ , ~ X%3,R +
~ , P+~~ZI, .QP ~ ~ , s ~ s , QZ ~~ , .P~X,ZR. Q ~ ~ , ~ ~ S , Of
R ) . the nine 3 x 3
beterminants resulting from summation o? s in this expression, the three non-vanishing ones yield
+ +
J = v,, J v2,2J v3,3J = v,,, J. Thus J = J V v and so d(ln J ) l d t = div v.
124 MOTION AND FLOW [CHAP. 4

4.29. Show that for steady motion (avilat = 0) of a continuum the streamlines and
pathlines coincide.
As shown in Problem 4.7, a t a given instant t streamlines are the solutions of the differential
equations d x l l v l = dx21v2 = dx31v3. Pathlines are solutions of the differential equations dxildt =
vi(x, t). If vi = vi(x), these equations become d t = dx,lv, = dxZ/v2= dx3/v3 which coincide with
the streamline differential equations.

4.30. +
For the steady velocity field v1 = x l x z x i , v2 = -x; - ~ 1 x 2 ,v3 = O determine
expressions for the principal values of the rate of deformation tensor D a t an arbitrary
point P(x1, X Z , x3).

Principal values d(i, are solutions of


2x,x2 - d -2: +xi , O
-x; + z; -2x1x2 - d O = O = -d[-4xix; + d2 - (4- x;)2]
o o -d
Thus d,,, = O, do, = -(x: +xi), do, = ( x ; + 2;). Note here that d I = (x: +xi), d I I = O,
+
d I I I= -(x; 22).

4.31. Prove equation (4.43) by taking the material derivative of d S i in its cross product
form d S i = cijkd x j 2 )d x k 3 ) .
axi axi a%, axk
Using (S.&?), dSi = ~ ~ ~ ~ ( a xdXz(axklaX3)
~ / a X ~ ) d X 3 and axi - dSi = qjk dX2dX3 = ax,
ax ax. dSi = 6, dSi = d S p = -
8x1
J d X z dX3. Thus 2 J d X z d X 3 and by Problem 4.28,
axP ax, a x ~

4.32. Use the results of Problems 4.27 and 4.23 to show that the material rate of change
of the vorticity flux -
d"t L
q *6 dS equals the flux of the curl of the acceleration a.

Taking the curl of the acceleration a s given in Problem 4.27,


V x a = V x at + V x ( q x v) + V x ~ ( ~ 2 1 . 2 )
or V x a = aqlat + V x ( q x v ) = dqldt + q ( V v ) - ( q V ) v
since q = V X v and V X V(v212) = O. Thus if q is substituted for p in Problem 4.23,

$L ,*nd~ = 1[$ + q ( V * v )- ( q * V ) v ] . R d S = L (Vxa)*RdS


CHAP. 41 MOTION AND FLOW 125

4.33. For the vorticity qi show that a


at qi d V = [€,,ak + qjvi- qivi]d s j .
From Problem 4.32 the identity V X a = aqlat + V X (q X v) may be written in indicial form
a s aqi/at = eijkak, - E ~ ~ ~ ~~ v , (. ) , E~Thus
.~ ~ ~

i$ d~ = Sv
[Eijkak,i- (.,srn~rn~r),sl dv

and by the divergence theorem of Gauss (1.157),

Supplementary Problems
A continuum motion is given by x 1 = Xlet + X3(et- 1 ) , x2 = X 2 + X,(et - e-t), x3 = X3. Show
that J does not vanish for this motion and obtain the velocity components.
+
Ans. v , = ( X , X3)et, v 2 = X,(et +
e-t), v3 = O or v , = x1 - x3, v2 = x3(et -e-t),
i v3 = O

A velocity field is specified in Lagrangian form by v , = -X2e-t, v , = -X3, v3 = 2t. Determine


the acceleration components in Eulerian form. Ans. a , = e-t(x2 t x 3 - e), a, = O, a3 = 2 +
+
Show that the velocity field vi = eijkbjxk ci where b, and ci are constant vectors, represents a
rigid body rotation and determine the vorticity vector for this motion.
Ans. qi = bixLi - b, = 2bi

Show that for the flow vi = xil(l +t) the streamlines and path lines coincide.

The electrical field strength in a region containing a fluid flow is given by h = ( A cos 3t)lr where
r2 = x; +
x i and A is a constant. The velocity field of the fluid is v , = 2 2 , v , = -x; - XSX, +
xixi, v , = O. Determine dXldt a t P(x,, x,, x,). Ans. dhldt = (-3A sin 3t)lr

Show that for the velocity field v , = x l x 2 + 22, v , = -x; - xixi, v3 = O the streamlines are
circular.
i

+ +
For the continuum motion x , = X,, x, = et(X2 X3)/2 e-t(X2 - X,)/2, x3 = et(X2 X,)/2 - +
c~-~(-X ,X3)12, show that Dij = deijldt a t t = O. Compare these tensors a t t = 0.5.

For the velocity field v , = s;x, + si, v , = -(xt + x,x;), v3 = 0, determine the principal axes
and principal values of ü a t P ( l , 2,3).

Am. D: = (:5 0

o
o
o);
-5
a = ( 31-

1
O
3
11-

o
o

For the velocity field of Problem 4.41 determine the rate of extension in the direction
V - (3,- 2 3 ,
A-

h
+
2 ê 3 ) / 3 a t P ( l , 2,3). What is the maximum shear rate a t P?
Ans. d"' = -2419, Ymax = 5
Show that d(axilaXj)ldt = v~,,x,,~ and use this to derive (4.41) of the text directly from (4.38).

Prove the identity ~pq,(v,vr,,),q = qPvq,, +-vqqP,, - qqvP,, where vi is the velocity and qi the
vorticity. Also show that v,, jvi,i = Dij Dij - qiqi/2.

Prove that the material derivative of the total vorticity is given by


Chapter 5

Fundamental Laws
of Continuum Mechanics

5.1 CONSERVATION OF MASS. CONTINUITY EQUATION


Associated with every material continuum there is the property known as mass. The
amount of mass in that portion of the continuum occupying the spatial volume V a t time t
is given by the integral
(5.1)
in which p(x, t ) is a continuous function of the coordinates called the m a s density. The law
of conservation of mass requires that the mass of a specific portion of the continuum remain
constant, and hence that the material derivative of (5.1) be zero. Therefore from (4.52)
with P:. . .(x,t ) = p(x, t ) , the rate of change of m in (5.1) is

Since this equation holds for an arbitrary volume V , the integrand must vanish, or

This equation is called the continuity equation; using the material derivative operator it
may be put into the alternative form

For an incompressible continuum the mass density of each particle is independent of


time, so that dpldt = O and (5.3) yields the result
v,,, = O or divv = O (5.5)
The velocity field v ( x , t ) of an incompressible continuum can therefore be expressed by the
equation
vi = c ~ ~ ~ ors ~ V, =~ V xs (5.6)
in which s(x, t ) is called the vector potential of v.
The continuity equation may also be expressed in the Lagrangian, or material form.
The conservation of mass requires that

where the integrals are taken over the same particles, i.e. V is the volume now occupied by
the material which occupied V0 a t time t = O . Using (4.1)and (4.38),the right hand integral
in (5.7) may be converted so that
CHAP. 51 FUNDAMENTAL LAWS OF CONTINUUM MECHANICS 127

Since this relationship must hold for any volume VO,it follows that
Po = PJ (5.9)
which implies that the product pJ is independent of time since V is arbitrary, or that
d
~ ( P J=
) 0 (5.10)
Equation (5.10) is the Lagrangian differentialform of the continuity equation.

5.2 LINEAR MOMENTUM PRINCIPLE. EQUATIONS OF MOTION.


EQUILIBRIUM EQUATIONS
A moving continuum which oc-
cupies the volume V a t time t is
shown in Fig. 5-1. Body forces bi
per unit mass are given. On the dif-
ferential element dS of the bounding
surface, the stress vector is tin'. The
velocity field Vi = duildt is prescribed
throughout the region occupied by
the continuum. For this situation,
the total linear momentum of the
mass system within V is given by

Fig. 5-1

Based upon Newton's second law, the principle o f linear momentum states that the time
rate of change of an arbitrary portion, of a continuum is equal to the resultant force acting
upon the considered portion. Therefore if the interna1 forces between particles of the con-
tinuum in Fig. 5-1 obey Newton's third law of action and reaction, the momentum principle
for this mass system is expressed by

Upon substituting tin' = ajini into the first integral of (5.12) and converting the result-
ing surface integral by the divergence theorem of Gauss, (5.12) becomes

In calculating the material derivative in (5.13), the continuity equation in the form given
by (5.10) may be used. Thus

Replacing the right hand side of (5.13) by the right hand side of (5.14) and collecting terms
results in the linear momentum principle in integral form,
128 FUNDAMENTAL LAWS OF CONTINUUM MECHANICS [CHAP. 5

Since the volume V is arbitrary, the integrand of (5.15) must vanish. The resulting
equations,

are known as the equations of motion.


The important case of static equilibrium, in which the acceleration components vanish,
is given a t once from (5.16) as
+
u ~ ~pbi, =
~ O or V, 1 pb = O + (5.17')
These are the equilibrium equations, used extensively in solid mechanics.

5.3 MOMENT OF MOMENTUM (ANGULAR MOMENTUM) PRINCIPLE


The moment of momentum is, as the name implies, simply the moment of linear
momentum with respect to some point. Thus for the continuum shown in Fig. 5-1, the
total moment of momentum or angular momentum as it is often called, with respect to the
origin, is
J
~ ( t=) v ~,,x,pv, d~ or N = ( X X ~ Vd
) v (5.18)

in which xj is the position vector of the volume element dV. The moment of momentum
principle states that the time rate of change of the angular momentum of any portion of a
continuum with respect to an arbitrary point is equal to the resultant moment (with respect
to that point) of the body and surface forces acting on the considered portion of the con-
tinuum. Accordingly, for the continuum of Fig. 5-1, the moment of momentum principle
is expressed in integral form by

Equation (5.19) is valid for those continua in which the forces between particles are equal,
opposite and collinear, and in which distributed moments are absent.
The moment of momentum principle does not furnish any new differential equation of
h

motion. If the substitution t p ' = uPknpis made in (5.19), and the symmetry of the stress
tensor assumed, the equation is satisfied identically by using the relationship given in (5.16).
If stress symmetry is not assumed, such symmetry may be shown to follow directly from
h

(5.19), which upon substitution of t p ' = upknp, reduces to

Since the volume V is arbitrary,


,,, ,
... .V. = O or 1, = O
which by expansion demonstrates that uik = ukj.

5.4 CONSERVATION OF ENERGY. FIRST LAW OF THERMODYNAMICS.


ENERGY EQUATION
If mechanical quantities only are considered, the principle of conservation of energy
for the continuum of Fig. 5-1 may be derived directly from the equation of motion given
by (5.16). To accomplish this, the scalar product between (5.16) and the velocity vi is first
computed, and the result integrated over the volume V. Thus
CHAP. 51 FUNDAMENTAL LAWS OF CONTINUUM MECHANICS 129

But L pv,;,dV = i vivi


cLdV =
dK
dt
which represents the time rate of change of the kinetic energy K in the continuum. Also,
v i ~ j i ,=j ( v , ~ ~ -
, )vi,j
, ~ uji and by (4.19) +
= Dij V,,, so that (5.22) may be written

since Vijuji O. Finally, converting the first integral on the right hand side of (5.24) to a
surface integral by the divergence theorem of Gauss, and making use of the identity
ti"' = ~31. . n3 . ,the energy equation for a continuum appears in the form

This equation relates the time rate of change of total mechanical energy of the continuum
on the left side to the rate of work done by the surface and body forces on the right hand
side of the equation. The integral on the left side is known as the time rate of change of
internal mechanical energy, and written dUldt. Therefore (5.25) may be written briefly a s

where d W l d t represents the rate of work, and the special symbol d is used to indicate that
this quantity is not an exact differential.
If both mechanical and non-mechanical energies are to be considered, the principle of
conservation of energy in its most general form must be used. In this form the conservation
principle states that the time rate o f change of t h e kinetic plus the internal energy i s equal
t o t h e s u m o f the rate o f work plus a11 other energies supplied to, or removed f r o m the
continuum per unit time. Such energies supplied may include thermal energy, chemical
energy, or electromagnetic energy. In the following, only mechanical and thermal energies
are considered, and the energy principle takes on the form of the well-known first law of
thermodynamics.
For a thermomechanical continuum it is customary to express the time rate of change
of internal energy by the integral expression

where u is called the specijic internal energy. (The symbol u for specific energy is so well
established in the literature that it is used in the energy equations of this chapter since
there appears to be only a negligible chance that it will be mistaken in what follows for the
magnitude of the displacement vector ui.) Also, if the vector s is defined as the heat fEux
per unit area per unit time by conduction, and z is taken a s the radiant heat constant per
unit mass per unit time, the rate of increase of total heat into the continuum is given by

Therefore the energy principle for a thermomechanical continuum is given by


130 FUNDAMENTAL LAWS OF CONTINUUM MECHANICS [CHAP. 5

or, in terms of the energy integrals, as

Converting the surface integrals in (5.30) to volume integrals by the divergence theorem of
Gauss, and again using the fact that V is arbitrary, leads to the local form of the energy
equation:

or

Within the arbitrarily small volume element for which the local energy equation (5.31)
is valid, the balance of momentum given by (5.16) must also hold. Therefore by taking the
scalar product between (5.16) and the velocity p'Uivi = vioji,j+ pvibi and, after some simple
manipulations, subtracting this product from (5.31),, the result is the reduced, but h i g h l ~
useful form of the local energy equation,

This equation expresses the rate of change of internal energy as the sum of the stress power
plus the heat added to the continuum.

5.5 EQUATIONS OF STATE. ENTROPY. SECOND LAW OF THERMODYNAMICS


The complete characterization of a thermodynamic system (here, a continuum) is said
to describe the state of the system. This description is specified, in general, by severa1
thermodynamic and kinematic quantities called state variables. A change with time of the
state variables characterizes a thermodynamic process. The state variables used to describe
a given system are usually not a11 independent. Functional relationships exist among the
state variables and these relationships are expressed by the so-called equations of state.
Any state variable which may be expressed as a single-valued function of a set of other
state variables is known as a state function.
As presented in the previous section, the first law of thermodynamics postulates the
interconvertibility of mechanical and thermal energy. The relationship expressing con-
version of heat and work into kinetic and internal energies during a thermodynamic
process is set forth in the energy equation. The first law, however, leaves unanswered the
question of the extent to which the conversion process is reversible or irreversible. A11
real processes are irreversible, but the reversible process is a very useful hypothesis since
energy dissipation may be assumed negligible in many situations. The basic criterion for
irreversibility is given by the second law of thermodynamics through its statement on the
limitations of entropy production.
The second law of thermodynamics postulates the existence of two distinct state
functions; T the absolute temperature, and S the entropy, with certain following properties.
T is a positive quantity which is a function of the empirical temperature O, only. The
entropy is an extensive property, i.e. the total entropy in the system is the sum of the
entropies of its parts. In continuum mechanics the specific entropy (per unit mass), or
entropy density is denoted by s, so that the total entropy L is given by L = 1
- .
ps dV. The
entropy of a system can change either by interactions that occur with the surroundings, or
by changes that take place within the system. Thus
ds = + ds(i) (5.33)
CHAP. 51 FUNDAMENTAL LAWS OF CONTINUUM MECHANICS 131

where d s is the increase in specific entropy, d s c e )is the increase due to interaction with the
exterior, and ds'" is the internal increase. The change dsCi)is never negative. I t is zero
for a reversible process, and positive for an irreversible process. Therefore
ddi) > O (irreversible process) (5.34)
ds(i) = O (reversible process) (5.35)
In a reversible process, if d q ( ~denotes
) the heat supplied per unit mass to the system, the
change dsCe)is given by
dS(e) = -dq(Ri (reversible process) (5.36)
T

5.6 THE CLAUSIUS-DUHEM INEQUALITY. DISSIPATION FUNCTION


According to the second law, the time rate of change of total entropy L in a continuum
occupying a volume V is never less than the sum of the entropy influx through the con-
tinuum surface plus the entropy produced internally 'by body sources. Mathematically, this
entropy principle may be expressed in integral form as the Clausius-Duhem inequality,

where e is the local entropy source per unit mass. The equality in (5.37) holds for reversible
processes; the inequality applies to irreversible processes.
The Clausius-Duhem inequality is valid for arbitrary choice of volume V so that trans-
forming the surface integral in (5.37) by the divergence theorem of Gauss, the local form
of the internal entropy production rate y, per unit mass, is given by

This inequality must be satisfied for every process and for any assignment of state variables.
For this reason it plays an important role in imposing restrictions upon the so-called
constitutive equations discussed in the following section.
In much of continuum mechanics, it is often assumed (based upon statistical mechanics
of irreversible processes) that the stress tensor may be split into two parts according to
the scheme,
a,. = a!?) + u 'iDj ' (5.39)
where u::) is a conservative stress tensor, and is a dissipative stress tensor. With this
assumption the energy equation (5.32) may be written with the use of (4.25) a s

1
In this equation, - o;?) iij is the rate of energy dissipated per unit mass by the stress, and
0
d q l d t is the rate Af heat influx per unit mass into the continuum. If the continuum under-
goes a reversible process, there will be no energy dissipation, and furthermore, dqldt =
d q ( ~ ) l d t so
, that (5.40) and (5.36)may be combined to yield

Therefore in the irreversible process described by (5.40),the entropy production rate may
be expressed by inserting (5.41). Thus
132 FUNDAMENTAL LAWS O F CONTINUUM MECHANICS [CHAP. 5

The scalar ;ij is called the dissipation function. For an irreversible, adiabatic process
(dq = O), dsldt > O by the second law, so from (5.42) it follows that the dissipation function
is positive definite, since both p and T are always positive.

5.7 CONSTITUTIVE EQUATIONS. THERMOMECHANICAL AND


MECHANICAL CONTINUA
In the preceding sections of this chapter, severa1 equations have been developed that
must hold for every process or motion that a continuum may undergo. For a thermo-
mechanical continuum in which the mechanical and thermal phenomena are coupled, the
basic equations are

(a) the equation of continuity, (5.4)

(b) the equation of motion, (5.16)

(c) the energy equation, (5.32)

Assuming that body forces bi and the distributed heat sources z are prescribed, (5.43),
(5.44) and (5.45) consist of five independent equations involving fourteen unknown functions
of time and position. The unknowns are the density p, the three velocity components vi,
(or, alternatively, the displacement components ui), the six independent stress components
oij, the three components of the heat flux vector ci, and the specific interna1 energy u. In
addition, the Clausius-Duhem inequality (5.38)

which governs entropy production, must hold. This introduces two additional unknowns:
the entropy density s, and T, the absolute temperature. Therefore eleven additional equa-
tions must be supplied to make the system determinate. Of these, six will be in the form
known as constitutive equations, which characterize the particular physical properties of
the continuum under study. Of the remaining five, three will be in the form of temperature-
heat conduction relations, and two will appear as thermodynamic equations of state; for
example, perhaps as the caloric equation of state and the entropic equation of state. Specific
formulation of the thermomechanical continuum problem is given in a subsequent chapter.

I t should be pointed out that the function of the constitutive equations is to establish
a mathematical relationship among the statical, kinematical and thermal variables, which
will describe the behavior of the material when subjected to applied mechanical or thermal
forces. Since real materials respond in an extremely complicated fashion under various
loadings, constitutive equations do not attempt to encompass a11 the observed phenomena
related to a particular material, but, rather, to define certain ideal materials, such as the
ideal elastic solid or the ideal viscous fluid. Such idealizations or material models as they
are sometimes called, are very useful in that they portray reasonably well over a definite
range of loads and temperatures the behavior of real substances.
CHAP. 51 FUNDAMENTAL LAWS O F CONTINUUM MECHANICS 133

In many situations the interaction of mechanical and thermal processes may be


neglected. The resulting analysis is known as the uncoupled thermoelastic theory of con-
tinua. Under this assumption the purely mechanical processes are governed by (5.43) and
(5.44) since the energy equation (5.45) for this case is essentially a first integral of the
equation of motion. The system of equations formed by (5.43) and (5.44) consists of four
equations involving ten unknowns. Six constitutive equations are required to make the
system determinate. In the uncoupled theory, the constitutive equations contain only the
statical (stresses) and kinematic (velocities, displacements, strains) variables and are often
referred to as stress-strain relations. Also, in the uncoupled theory, the temperature field
is usually regarded as known, or a t most, the heat-conduction problem must be solved
separately and independently from the mechanical problem. . In isothermal problems the
temperature is assumed uniform and the problem is purely mechanical.

Solved Problems
CONTINUITY EQUATION (Sec. 5.1)
5.1.' An irrotational motion of a continuum is described in Chapter 4 as one for which the
vorticity vanishes identically. Determine the form of the continuity equation for
such motions.
-
B y (1.29),curl v = O when q 0, and so v becomes the gradient of a scalar field +(xi, t ) (see
+
Problem 1.50). Thus vi = +,i and (5.3)is now dpldt p+,kk = O or dpldt pV2+ = 0 . +

5.2. If P:?, .(x, t) represents any scalar, vector or tensor property per unit mass of a con-
tinuum so that P:. . .(x, t) = pp". .(x, t) show that

since by ( 5J), dpldt + pvk, = 0.

5.3. Show that the material form d(pJ)ldt = O of the continuity equation and the spatial
+
form dpldt p v , , , = O are equivalent.
Differentiating, d ( p J ) l d t = ( d p l d t ) J + p dJ/dt = O and from Problem 4.28, d J l d t = J v k , , so
+
that d ( p J ) l d t = J ( d p l d t P V , , ~ )= 0 .

5.4. Show that the velocity field vi = Axil?, where xixi = r2 and A is an arbitrary con-
stant, satisfies the continuity equation for an incompressible flow.
From (5.5) v k , k = O for incompressible flow. Here

and so u k S k= ( 3 - 3)/$ = O to satisfy the continuity equation.


134 FUNDAMENTAL LAWS O F CONTINUUM MECHANICS [CHAP. 5

5.5. For the velocity field vi = x i l ( l + t), show that pxlx,x3 = p,,X1X2X3.
+
Here vk,k = 3 / ( 1 t ) and integrating (5.3) yields ln p = -1n (1 t)3 ln C where C is a + +
constant of integration. Since p = po when t = 0, this equation becomes p = pd(l t)3. Next +
by integrating the velocity field dxi/xi = d t / ( l + t ) (no sum on 9, xi = X i / ( l t ) and hence +
Pxlx2x3 =~ 0 ~ 1 ~ 2 ~ 3 .

LINEAR AND ANGULAR MOMENTUM. EQUATIONS OF MOTION (Sec. 5.2-5.3)


A
5.6. Show by a direct expansion of each side that the identity eijkujke,= 1, used in (5.20)
and (5.21) is valid.
By (1.15) and (2.8),
L, = alSl x ê1 + olGl x $2 + a l g l x ê3 + . . + u33ê3 x $3
-
- (023 - 0 3 2 ) ê1 + ('731 - 013) $2 + ( ~ 1 -2 ~ 2 123)
Also, expanding eijkajkgives identical results, (az3- a32) for i = 1 , (031 - ul3) for i = 2, (al2- oZ1)
for i = 3.

5.7. If distributed body moments m i per unit volume act throughout a continuum, show
that the equations of motion (5.16) remain valid but the stress tensor can no longer
be assumed symmetric.
Since (5.16) is derived on the basis of force equilibrium, i t is not affected. Now, however, (5.19)
acquires a n additional term so that

which reduces to (see Problem 2.9) ( Y j k a j k+mil dV = O, and because V i s arbitrary, eijkajk +
mi = O for this case.

5.8. The momentum principle in differential form (the so-called local or "in the small"
form) is expressed by the equation d ( p v , ) l d t = pbi + ( u i j - ~ V ~ V ~ Show
) , ~ . that the equa-
tion of motion (5.16) follows from this equation.
Carrying out the indicated differentiation and rearranging the terms in the resulting equation
yields
~,(a~/+
a t p,jvj + pvj, +j) p(avi/Jt + vivi, j) pbi aij, j
The first term on the left is zero by (5.4) and the second term is pai. Thus pai = pbi + aij,j which
is (5.16).

5.9. Show that (5.19) reduces to (5.20).


h

Substituting oPknpfor tp' in (5.19) and applying the divergence theorem (1.157) to the resulting
surface integral gives

Using the results of Problem 5.2, the indicated differentiations here lead to

cijkvjvk = O, so that finally


s,
The term in parentheses is zero by (5.16), also
€ijkUjk .v = o.
zj,, = S j p and
CHAP. 51 FUNDAMENTAL LAWS O F CONTINUUM MECHANICS 135

5.10. For a rigid body rotation about a point, vi = cijkojxk. Show that for this velocity
(5.19) reduces to the well-known momentum principle of rigid body dynamics.
The left hand side of (5.19) is the total moment Mi of a11 surface and body forces relative to
the origin. Thus for vi = eijkwjXk,

=
d
[O. J"~ ( 4 P q x q-
where Zip = p(8ipzqzq- 44 d V is the moment of inertia tensor.

ENERGY. ENTROPY. DISSIPATION FUNCTION (Sec. 5.4-5.6)


5.11. Show that for a rigid body rotation with vi = C~,W~X,, the kinetic energy integral o£
(5.23) reduces to the familiar form given in rigid body dynamics.

From (5.2?)),
K = i p Y d v = i peiikUj%k€ipq~pZq
dv

= ii ~ ~ p w j ( 8 j p-~6ki q~S k p ) x k z q dV

- wiw~
2- i p(6ipzqzq - z P z $ d V wiwpzip
-
2
o.l.0,
In symbolic notation note that K = - 2

5.12. At a certain point in a continuum the rate of deformation and stress tensors are
given by

4 2 5
and

Determine the value h of the stress power Dijuij a t the point.


uii = - -:i
Multiplying each element of Dij by its counterpart jn oij and adding, h = 4 +O -4 +O -6 +
14 -4 + +
14 40 = 58.

5.13. If uij = -pSij where p is a positive constant, show that the stress power may be
-.
p dp
expressed by the equation D i j u i j = -
P dt
By (4.19), Dij = v i B j- Vij; and since V i j u i j= O, i t follows that Dijoij = = -pvi,i.
From the continuity equation (5.3), vi,i = -(llp)(dpldt) and so D i j q j = (plp)(dpldt) for oij = -paij.

5.14. Determine the form of the energy equation if uij = (-p + h* Dkk)8,, 4- 2 , ~D,,
* and the
heat conduction obeys the Fourier law ci = -kTai.
From (5.92),
du
p d t
- (-p+ X*Dkk)SijDij 2p*DijDij
-
+ kTSii + +z
- P
- --do + +
(h* 2 / ~ * ) ( 1 , )-
P dt
~ 4/~*11, kT,ii + +z
where I, and 11, are the first and second invariants respectively of the rate of deformation tensor.
136 FUNDAMENTAL LAWS O F CONTINUUM MECHANICS [CHAP. 5

5.15. If uij = - p s i j , determine an equation for the rate of change of specific entropy during
a reversible thermodynamic process.
Here oij '=
::o
dt
ds
and (5.41) gives T - = -
dt
du
+ --
p dp
p2 dt
upon use of the result in Problem 5.13.

5.16. For the stress having = /3DikDkj, determine the dissipation function in terms of
the invariants of the rate of deformation tensor D.
Here by (4.25), o:;) iij = PDsDkjDij which is the trace of D~ (see page 16) and may be evaluated
by using the principal axis values D ( , , , D ( 2 ) ,DC3). Thus by (1.138) the trace
DiiD,Dki = D:,) i- D:Z) + ~ ( 3 )

= (D(l) + + D(3))3 - 3(D(1) + D(2) + D(3))(D(1)D(2) + D(2)D(3> + D(3)D(1))


+ 3D,l,D(2,D,3)
Therefore o:;' gj = p[1: - 31DIID+ 3111D].

CONSTITUTIVE EQUATIONS (Sec. 5.7)


5.17. For the constitutive equations uij = Ki,,DPq show that because of the symmetry of
the stress and rate of deformation tensors the fourth order tensor Kijp,has a t most 36
distinct components. Display the components in a 6 x 6 array.
Since oij = uji, KiTpq= Kjipq; and since Dij = Dji, KijPq= Kijqp. If Kijpq is considered a s the
outer product of two symmetric tensors AijBpq= Kijpq, i t is clear that since both Aij and Bij have
six independent components, Kijpqwill have a t most 36 distinct components.
The usual arrangement followed in displaying the components of Kib, is

5.18. If the continuum having the constitutive relations


uij = KijpqDpqof Problem 5.17 is assumed isotropic
so that Kijpqhas the same array of components in any
rectangular Cartesian system of axes, show that by
a cyclic labeling of the coordinate axes the 36 com-
ponents may be reduced to 26.
The coordinate directions may be labeled in six different
ways a s shown in Fig. 5-2. Isotropy of Kijpqthen requires x1 or x2
that K,,,, = Kl13, = KZZ3,= K2,,, = K3,,, = K3322 and that
K l q I 2= 1(1313= K2323= K2121= K3131= K3232 which reduces
the 36 components to 26. By suitable reflections and rotations 53
of the coordinate axes these 26 components may be reduced
to 2 for the case of isotropy. Fig. 5-2

5.19. For isotropy Kijpqmay be represented by Kijpq= A*8ij8pq , ~ * ( 8 ~ ~ 8iq8jp).


8~, +
Use +
this to develop the constitutive equation uij = KijpqDpqin terms of A* and P * .
CHAP. 51 FUNDAMENTAL LAWS O F CONTINUUM MECHANICS 137

5.20. Show that the constitutive equation of Problem 5.19 may be split into the equivalent
equations aii = (3A* + 2p*)Dii and sij = 2p*DG where sij and DC are the deviator ten-
sors of stress and rate of deformation, respectively.
+ +
Substituting U i j = S i j Sijukk/3 and Dij = D;j SijDkk/3 int0 uij = h*SijDkk 2@*Dij 0f +
+ +
Problem 5.19 resuits in the equation sij aijukk/3 = h*SijDkk 28*(Dii SijDkk/3). From this +
when i f j, sij = 2@*D& and hence ukk = ('&h* 2@*)Dkk. +

MISCELLANEOUS PROBLEMS
581. Show that +
= ( ~ , , a ~ , ~ qjvi,j)lpwhere p is the density, ai the acceleration and
qi the vorticity vector.
By direct differentiation
dd, (f) = 4 -
-
P p2'
But 4 = rijkak,i + qivi,i - qivjj (see Problem
4.32); and by the continuity equation (5.4, e = -pvci. Thus

5.22. A two dimensional incompressible flow is given by v1 = A(xS - xi)l.P, v2 = A ( 2 x l x z ) l l A ,


v3 = 0, where r2 = X: + x;. Show that the continuity equation is satisfied by this
motion.
By (5.5), vi,i = O for incompressible flow. Here v l , , = A[-4x1(xS - xi)/18 + 2x1/r<l] and
v e j 2 = A [2x1/1A- 8x1xi/r8]. Adding, v,, ,+ v,, ,= 0.

5.23. Show that the flow of Problem 5.22 is irrotational.


By (4.29), curl v = 0 for irrotational flow. Thus
A A A
e1 e2 e3
curlv = alasl ala%, alax,
A ( X ; - xi)/1.4 2Ax1x2/r<l O
+
= A[2x2/r<l - 8 x ~ x 2 / r ~2x2/r<l + 4x2(x: - xg)/r6]ê3 = O

5.24. In a two dimensional incompressible steady flow, v1 = -Ax21r2 where r2= x: + xi .


Determine v2 if v2= O a t xl = O for a11 x2. Show that the motion is irrotational
and that the streamlines are circles.
From (5.5), viBi= O or v,,, = -v2,, = 2Axlx2/.P for this incompressible flow. Integrating
with respect to x2 and imposing the given conditions on v 2 yields v2 = Axllr2.
For a n irrotational motion, curl v = O. Here
curl v = A [(x: - xe)/la + (-E: + x i )I+] ê3 = 0
From Problem 4.7, page 118, the equations for streamlines are dxllvl = dx21vz. Here these
+
equations are xl dxl x2 dx2 = O which integrate directly into the circles x; x i = constant. +
5.25. For a continuum whose constitutive equations are a, = (-p + A* Dkk)Sij+ 2p*Dij,
determine the equations of motion in terms of the velocity vi.
+
From (5.16), p$i = pbi uijSj or here p6i = pbi - p,jSij X*D,,,~S, + +
28*DilBj. By definition,
+
2Dij = v i s jt vjBi so that Dkk = v k j k and 2Diisj = vi,jj viaij. Therefore
FUNDAMENTAL LAWS O F CONTINUUM MECHANICS [CHAP. 5

pCi = pbi - p,i + (X* + p * ) ~ j , +~ j p*vi,jj


In symbolic notation this equation is
p; = pb - V p + (A* + p * ) V ( V V) + p*V2v

5.26. If the continuum of Problem 5.25 is considered incompressible, show that the diver-
gente of the vorticity vanisbes and give the form of the equations of motion for
this case.
By (4.29), qi = e i j k v k , and div q = eijkvk,ji = O since eijk is antisymmetric and vk,ji is sym-
metric in i and j . Thus for V v = O the equations of motion become p'Ui = pbi - p,i p*vi,jj +
or in Gibbs notation p; = p b - V p p*VZv. +

5.27. Determine the material rate of change of the kinetic energy of the continuum which
occupies the volume V and give the meaning of the resulting integrals.
,.
By (5.23). dK/dt = J pvi'UidV. Also the total stress power of the surface forces is
. L viti'' d S which may be written
V
viuijni d S and by the divergence theorem (1.157) and the

equations of motion (5.16) expressed a s the volume integrals


r -
( p(v$i - bivi) d V . Thus

This sum of integrals represents the rate of work done by the body forces, the interna1 stresses
and the surface stresses, respectively.

5.28. A continuum for which = h*Dk,8,, +


2p*Dij undergoes an incompressible irrota-
tional flow with a velocity potential + such that v = grad +. Determine the dissipa-
tion function ;ij.

Here o::) aij = u : ~ ) D


= ~(?,*Dkksij
~ +
2p*Dij)Dij= 2p*DijDij since Dkk = v k , = O for incompres-
sible flow. Also since vi = +,i, the scalar DijDij = +,ij+,ij and so u : y 3 D ' =
~ i j2p*+,ij+,ij.
Because the motion is incompressible and irrotational, +,li = O and 2+,u+,ij = (+,i+,i),jj =
V2(V+)2.I t is also interesting to note that

which for +,ii = O reduces to 4+,ij+,ij. Thus +i:' gij = p * ~ z ( v + ) =


z p*v4(+z)/2.

5.29. For a continuum with uij = - p 8 . .23'. the specific e n t h a l p y h = u p l p . Show that the +
+
energy equation may be written h = +Ip Ti using this definition for the enthalpy.
From (5.41), = -pSijDijlp .+ T S for the given stress; and by the result of Problem 5.13 and
the definition of h, ii = a - +Ip - p;/pZ = -p;/pZ + T B . Canceling and rearranging, i= ;IP + TS.

5.30. If the continuum of Problem 5.25 undergoes an incompressible flow, determine the
equation of motion in terms of the vorticity q in the absence of body forces and
assuming constant density.
For incompressible flow, V v = 0 ; and if b 0, the equation of motion in Problem 5.25
reduces to pCi = -pZi +
P * v i , j j . Taking the cross product V X with this equation for p = constant
+
gives c ~ ~ =~- B~ ~~ , ~ ~ ~( Pp* / P, ) ~~p q~i v i ,l j jBut
q~. ~ ~ = O ~ and ~by (4.29)
p the
, result
~ ~is 4, =
( ~ * / p ) q , ~In~ .symbolic notation, dqldt = (p*lp)V2q.
CHAP. 51 FUNDAMENTAL LAWS O F CONTINUUM MECHANICS 139

Supplementary Problems
5.31. Show that for the rate of rotation vector a, P

5.32. Show that the flow represented by v , = - ~ X , X ~ ~ /v T = , (x: - x~)x,l+, v , = %,/r2 where
, ~
r2= x; +
xg, satisfies conditions for an incompressible flow. 1s this motion irrotational?

5.33. In terms of Cartesian coordinates x , y, z the continuity equation is


aplat + a ( p ~ , ) l a+~ a ( p ~ , ) l a+~ a ( ~ v , ) l a=~ O
Show that in terms of cylindrical coordinates r , 6 , z this equation becomes
r(aplat) + d(rpv,)lar + a(pve)la6+ r(d(pv,)ldz) = O

5.34. +
Show that the flow v, = (1- rz) cos elrz, v e = (1 r,) sin #/r,, v , = O satisfies the continuity
equation in cylindrical coordinates when the density p is a constant.

5.35. I f P i j , .,(x, t ) is a n arbitrary scalar, vector or tensor function, show that

5.36. I f a continuum i s subjected to a body momeht per unit mass h in addition to body force b, and a
couple stress g'"^ in addition to the stress t'"', the angular momentum balance may be written

where m i s distributed angular momentum per unit mass. If G = g'L', show that the local form
of this relation is p dmldt = h V G Z,.+ +
5.37. If a continuum has the constitutive equation oii = -pGij + PDij + aDikDkj, show that uii =
3(-p - 2 ~ ~ 1 1 ~ 1 3Assume
). incompressibility, Dii = 0.

5.38. For a continuum having uij = -pSij, show that du = T ds - p dv where in this problem v = l I p ,
the specific volume.

5.39. I f T dsldt = -vi,ilp and the specific free energy is defined by * = u - Ts, show that the energy
+
equation may be written p dqldt ps dTldt = oiiDij.

5.40. For a thermomechanical continuum having the constitutive equation


q j = hekkSijf 2peii - (3h f zp)a Gij(T- T o )
where T o is a reference temperature, show that ekk = 3a(T - To) when oij = sij = oij - ukkSij/3.
Linear Elasticity

6.1 GENERALIZED HOOKE'S LAW. STRAIN ENERGY FUNCTION


In classical linear elasticity theory it is assumed that displacements and displacement
gradients are sufficiently small that no distinction need be made between the Lagrangian
and Eulerian descriptions. Accordingly in terms of the displacement vector ui, the linear
strain tensor is given by the equivalent expressions

L = E = ~ ( u v+
, Vxu) = . &(uVx+ Vxu) = 4(uV + Vu)
In the following it is further assumed that the deformation processes are adiabati; (no heat
loss or gain) and isothermal (constant temperature) unless specifically stated otherwise.
The constitutive equations for a linear elastic solid relate the stress and strain tensors
through the expression -
uii = Cijkmekm or I: = C : E (6-2)
which is known a s the generalized Hooke's law. In (6.2) the tensor of elastic constants
Cijkm has 81 components. However, due to the symmetry of both the stress and strain
tensors, there are a t most 36 distinct elastic constants. For the purpose of writing Hooke's
law in terms of these 36 components, the double indexed system of stress and strain com-
ponents is often replaced by a single indexed system having a range of 6. Thus in the
notation - - -
'=H - '=I u23 - u32 - O4

and

-
€33 - €3 2c12 = 2~~~= E6

Hooke's law may be written


-
uK - CKMcM ( K ,M = 1 , 2 , 3 , 4 , 5 , 6 )
where CKMrepresents the 36 elastic constants, and where upper case Latin subscripts are
used to emphasize the range of 6 on these indices.
When thermal effects are neglected, the energy balance equation (5.32) may be written
CHAP. 61 LINEAR ELASTICITY 141

The interna1 energy in this case is purely mechanical and is called the strain energy (per
unit mass). From (6.6),
1
du = - u..d,,, (6.7)
P 23

and if u is considered a function of the nine strain components, u = u(c,,), its differential
is given by
au
du = -de..
a,,, 3'

Comparing (6.7) and (6.8), it is observed that

The strain energy density u* (per unit volume) is defined a s


u* = tu'
and since p may be considered a constant in the small strain theory, u* has the property that

Furthermore, the zero state of strain energy may be chosen arbitrarily; and since the stress
must vanish with the strains, the simplest form of strain energy function that leads to a
linear stress-strain relation is the quadratic form
U* = -&Cijkmcijckm (6.12)
From (6.2), this equation may be written
U* = +uijeij or U* = -&Z: E (6.18)
In the single indexed system of symbols, (6.12) becomes
U* = -&CKMcKeM (6.14)
in which CKM= CMK. Because of this symmetry on CKM,the number of independent elastic
constants is a t most 21 if a strain energy function exists.

6 2 ISOTROPY. ANISOTROPY. ELASTIC SYMMETRY


If the elastic properties are independent of the reference system used to describe it, a
material is said to be elasticauy isotropic. A material that is not isotropic is called anko-
tropic. Since the elastic properties of a Hookean solid are expressed through the coefficients
CKM,a general anisotropic body will have an elastic-constant matrix of the form

When a strain energy function exists for the body, CKM= CMK, and the 36 constants in
(6.15) are reduced to 21.
142 LINEAR ELASTICITY [CHAP. 6

A plane of elastic symmetry exists a t a


point where the elastic constants have the same
values for every pair of coordinate systems
which are the reflected images of one another
with respect to the plane. The axes of such
coordinate systems are referred to as "equiva-
lent elastic directions." If the ~ 1 x plane
2 is one
of elastic symmetry, the constants CKMare in-
variant under the coordinate transformation
X; = XI, xá = ~ 2 , ~3 = -x3 (6.16)
a s shown in Fig. 6-1. The transformation
matrix of (6.16) is given by Fig. 6-1
r 1 O 01

Inserting the values of (6.17) into the transformation laws for the linear stress and strain
tensors, (2.27) and (3.78) respectively, the elastic matrix for a material having ~ 1 x 2as a
plane of symmetry is
C11 C12 c13 0 O
C21 C22 C23 0 0

The 20 constants in (6.18) are reduced to 13 when a strain energy function exists.
If a material possesses three mutually perpendicular planes of elastic symmetry, the
material is called orthotropic and its elastic matrix is of the form

having 12 independent constants, or 9 if CKM= CMK.


An axis of elastic symmetry of order N exists a t a point when there are sets of equiva-
lent elastic directions which can be superimposed by a rotation through an angle of 2nlN
about the axis. Certain cases of axial and plane elastic symmetry are equivalent.

6.3 ISOTROPIC MEDIA. ELASTIC CONSTANTS


Bodies which are elastically equivalent in a11 directions possess complete symmetry and
are termed isotropic. Every plane and every axis is one of elastic symmetry in this case.
J
CHAP. 61 LINEAR ELASTICITY 143

For isotropy, the number of independent elastic constants reduces to 2, and the elastic
matrix is symmetric regardless of the existence of a strain energy function. Choosing as
the two independent constants the well-known Lamé constants, A and p, the matrix (6.19)
reduces to the isotropic elastic form

In terms of A and p, Hooke's law (6.2) for an isotropic body is written

where C = c,, = IE. This equation may be readily inverted to express the strains in terms
of the stresses as
C.. =
-A
+
2 ~ ( 3 A 2 ~S i j )u k k
+ 2P
1
-oij ~r E =
-A
2 ~ ( 3 +2P)
h
I@+-E
1
2P
where o = u,, = I,, the symbol traditionally used in elasticity for the first stress ihvariant.
For a simple uniaxial state of stress in the X I direction, engineering constants E and V
may be introduced through the relationships u,, = Ec,, and c,, = C, = -vc,,. The constant
E is known as Young's modulus, and V is called Poisson's ratio. In terms of these elastic
constants Hooke's law for isotropic bodies becomes
E
u..
13
= -
l E
+ v ( ) or E = - 1( + V E +:ic)1 - 2~
or, when inverted,
- l + v E = -
l + v x -Li@
-E 001"
c.. - E U i j Lf3 i j ukk E E
From a consideration of a uniform hydrostatic pressure state of stress, it is possible to
define the bulk modulus,

which relates the pressure to the cubical dilatation of a bpdy so loaded. For a so-called state
of pure shear, the shear modulus G relates the shear components of stress and strain. G
is actually equal to p and the expression
- E
p - G = ----
2(1+ v)
may be proven without difficulty.

6.4 ELASTOSTATIC PROBLEMS. ELASTODYNAMIC PROBLEMS


In an elastostatic problem of a homogeneous isotropic body, certain field equations,
namely,
( a ) Equilibrium equations,
~31.1. . . + ~ b , = or
0 V . t + p b = O (6.27')
144 LINEAR ELASTICITY [CHAP. 6

( b ) Hooke's law,
uij = X a i j c k k + 2peij or 2 = hlc + ~,.LE (6.28)

(c) Strain-displacement relations,


cij = 4 ( ~+ ,u ~~ , ~ or
) E = &(uV + Vu) (6.29)
must be satisfied a t a11 interior points of the body. Also, prescribed conditions on stress
andlor displacements must be satisfied on the bounding surface of the body.
The boundary value problems of elasticity are usually classified according to boundary
conditions into problems for which
(1) displacements are prescribed everywhere on the boundary,
(2) stresses (surface tractions) are prescribed everywhere on the boundary,
(3) displacements are prescribed over a portion of the boundary, stresses are prescribed
over the remaining part.
For a11 three categories the body forces are assumed to be given throughout the continuum.
For those problems in which bound'ary displacement components are given everywhere
by an equation of the form
ui = gi(X) or u = g(X)
the strain-displacement relations (6.29) may be substituted into Hooke's law (6.28) and the
result in turn substituted into (6.27) to produce the governing equations,

which are called the Navier-Cauchy equations. The solution of this type of problem is
therefore given in the form of the displacement vector Ui, satisfying (6.31) throughout the
continuum and fulfilling (6.30) on the boundary.
For those problems in which surface tractions are prescribed everywhere on the
boundary by equations of the form

the equations of compatibility (3.104) may be combined with Hooke's law (6.24) and the
equilibrium equation (6.27) to produce the governing equations,

which are called the Beltrami-Michell equations of compatibility. The solution for this
type of problem is given by specifying the stress tensor which satisfies (6.33) throughout
the continuum and fulfills (6.32) on the boundary.
For those problems having "mixed" boundary conditions, the system of equations (6.27),
(6.28) and (6.29) must be solved. The solution gives the stress and displacement fields
throughout the continuum. The stress components must satisfy (6.32) over some portion
of the boundary, while the displacements satisfy (6.30) over the remainder of the boundary.
In the formulation of elastodynamics problems, the equilibrium equations (6.27) must
be replaced by the equations of motion (5.16)
CHAP. 61 LINEAR ELASTICITY 145

and initial conditions as well as boundary conditions must be specified. In terms of the
displacement field ui, the governing equation here, analogous to (6.31) in the elastostatic
case is ..
p ~ ~ , ~ ~ + ( h + p ) ~ ~ , ~0~ 1+ p~V b 2~ ~=+ p( A
~ +
~ p ) V V * ~ + p b = p Ü(6.35)

Solutions of (6.35) appear in the form ui = u~(x,t) and must satisfy not only initial condi-
tions on the motion, usually expressed by equations such a s

ui = ui(x,O) and Ui = Ui(x,O) (6.36)

but also boundary conditions, either on the displacements,

ui = gi(x, t) or u = g(x, t)
o r on the surface tractions,

6.5 THEOREM OF SUPERPOSITION. UNIQUENESS OF SOLUTIONS.


ST. VENANT PRINCIPLE
Because the equations of linear elasticity are linear equations, the principle of super-
position may be used to obtain additional solutions from those previously established. If,
for example, ,ujl) represent a solution to the system (6.27), (6.28) and (6.29) with body
forces b:" , and o;;), ut2) represent a solution for body forces b12', then <rij = a;:) + d2)
ij 9

ui = u,'" +ui2) represent a solution to the system for body forces bi = bil' + b'2).
i

The uniqueness of a solution to the general elastostatic problem of elasticity may be


established by use of the superposition principle, together with the law of conservation of
energy. A proof of uniqueness is included among the exercises that follow.
St. Venant's principle is a statement regarding the differences that occur in the stresses
and strains a t some interior location of an elastic body, due to two separate but statically
equivalent systems of surface tractions, being applied to some portion of the boundary.
The principle asserts that, for locations sufficiently remote from the area of application
of the loadings, the differences are negligible. This assumption is often of great assistance
in solving practical problems.

6.6 TWO-DIMENSIONAL ELASTICITY. PLANE STRESS AND PLANE STRAIN


Many problems in elasticity may be treated satisfactorily by a two-dimensional, or
plane theory of elasticity. There are two general types of problems involved in this plane
analysis. Although these two types may be defined by setting down certain restrictions
and assumptions on the stress and displacement fields, they are often introduced descrip-
tively in terms of their physical prototypes. In plane stress problems, the geometry of
the body is essentially that of a plate with one dimension much smaller than the others.
The loads are applied uniformly over the thickness of the plate and act in the plane of the
plate as shown in Fig. 6-2(a) below. In plane strain problems, the geometry of the body is
essentially that of a prismatic cylinder with one dimension much larger than the others.
The loads are uniformly distributed with respect to the large dimension and act per-
pendicular to it as shown in Fig. 6-2(b) below.
146 LINEAR ELASTICITY [CHAP. 6

('J)
Fig. 6-2

For the plane stress problem of Fig. 6-2(a) the stress components a,,, a,,, a,, are taken
a s zero everywhere, and the remaining components are taken as functions of xl and 2 2 only,
na8 = (xl>xZ) (a,P =1~2) (6.39)

Accordingly, the field equations for plane stress are

in which
a a
-êl + -$2 and
V = 8x1 ax,

(6.43)

Due to the particular form of the strain tensor in the plane stress case, the six compatibility
equations (3.104) may be reduced with reasonable accuracy for very thin plates to the
single equation

In terms of the displacement components u,, the field equations may be combined to give
the governing equation

where V 2 = -
d2
dx:
+ 8-
d2
~ ;
'

For the plane strain problem of Fig. 6-2(b) the displacement component us is taken as
zero, and the remaining components considered as functions of X I and X n only,
CHAP. 61 LINEAR ELASTICITY 147

In this case, the field equations may be written

(a) u,,,,+pb, = O or ~ . I : + p b= 0

(b) u,,3 - XSa,'uv


- +~ P E ~ , or 2: = Xlr + ZpE
- X
O33 - vu,, = 2(X + /L)

in which I: = [i: i: :) O33


and E = (i: i)
From (6.47), (6.48), (6.49), the appropriate Navier equation for plane strain is

As in the case of plane stress, the compatibility equations for plane strain reduce to the
single equation (6.44).

If the forces applied to the edge of the plate in Fig. 6-2(a) are not uniform across the .
thickness, but are symmetrical with respect to the middle plane of the plate, a state of
generalized plane stress is said to exist. In formulating problems for this case, the field
variables a,,, r,, and u, must be replaced by stress, strain and displacement variables
averaged across the thickness of the plate. In terms of such averaged field variables, the
generalized plane stress formulation is essentially the same as the plane strain case if X is
replaced by
A' = -2XP -- - -
VE (6.52)
X 2/.L +
1 - v2
A case of generalized plane strain is sometimes mentioned in elasticity books when r,,
is taken as a constant other than zero in (6.50).

6.7 AIRY'S STRESS FUNCTION


If body forces are absent or are constant, the solution of plane elastostatic problems
(plane strain or generalized plane stress problems) is often obtained through the use of the
Airy stress function. Even if body forces must be taken into account, the superposition
principie allows for their contribution to the solution to be introduced a s a particular
integral of the linear differential field equations.

For plane elastostatic problems in the absence of body forces, the equilibrium equations
reduce to
ua4.~-
- O or 8-2:= O (6.53)
and the compatibility equation (6.44) may be expressed in terms of stress components as

The stress components are now given as partia1 derivatives of the Airy stress function
= +(xl, XP) in accordance with the equations
- - - (6.55)
ull - 4,229 u12 - -4,129 u22 - +,I1
148 LINEAR ELASTICITY [CHAP. 6

The equilibrium equations (6.53) are satisfied identically, and the compatibility condition
(6.54) becomes the biharmonic equation

Functions which satisfy (6.56) are called biharmonic functions. By considering biharmonic
functions with single-valued second partia1 derivatives, numerous solutions to plane elasto-
static problems may be constructed, which satisfy automatically both equilibrium and
compatibility. Of course these solutions must be tailored to fit whatever boundary
conditions are prescribed.

6.8 TWO-DIMENSIONAL ELASTOSTATIC PROBLEMS IN POLAR COORDINATES


Body geometry often deems it convenient to formulate two-dimensional elastostatic
problems in terms of polar coordinates r and O. Thus for transformation equations
x1 = r cos 8, x2 =. r sin O (6.57)
the stress components shown in Fig. 6-3 are found to lead to equilibrium equations in
the form

in which R and Q represent body forces per unit volume in the directions shown.

'c'
Fig. 6-3

Taking the Airy stress function now as @ = @(r,O), the stress components are given by

The compatibility condition again leads to the biharmonic equation


7 ' ( V 2 @ )= V 4 @= O

but, in polar form, V2 = -


a2 + --
i a i a2
+ r2
--
ar2 r ar ao2 '
CHAP. 61 LINEAR ELASTICITY 149

6.9 HYPERELASTICITY. HYPOELASTICITY


Modern continuum studies have led to constitutive equations which define materials
that are elastic in a special sense. In this regard a material is said to be hyperelastic if it
possesses a strain energy function U such that the material derivative of this function is
equal to the stress power per unit volume. Thus the constitutive equation is of the form

in which Dij is the rate of deformation tensor. In a second classification, a material is said
to be hypoelastic if the stress rate is a homogeneous linear function of the rate of deforma-
tion. In this case the constitutive equation is written

= KijkrnDkm

in which the stress rate U: is defined as


d
u; = (cij)- uiqVqi - ujqV q i

where V i j is the vorticity tensor.

6.10 LINEAR THERMOELASTICITY


If thermal effects are taken into account, the components of the linear strain tensor r,
may be considered to be the sum
c,, = ,(SI
ij
+ €i(j T ) (6.67)
in which r::' is the contribution from the stress field and rr
is the contribution from the
temperature field. Due to a change from some reference temperature T Oto the temperature
T , the strain components of an elementary volume of an unconstrained isotropic body are
given by
€CT)
ij
= a(T - TO)Sij (6.68)
where a denotes the linear coefficient of thermal expansion. Inserting (6.68), together with
Hooke's law (6.22), into (6.67) yields

which is known a s the Duhamel-Neumann relations. Equation (6.69) may be inverted to


give the thermoelastic constitutive equations
uij = XSijrkk + 2pe, - (3X + 2,)aSij(T - T o ) (6.70)

Heat conduction in an isotropic elastic solid is governed by the well-known Fourier law
of heat conduction,
ci = -kTai (6.71)
where the scalar k, the thermal conductivity of the body, must be positive to assure a
positive rate of entropy production. If now the specific heat a t constant deformation c(")is
introduced through the equation
- c ~ ,=
~ p~(")T (6.72)
and the interna1 energy is assumed to be a function of the strain components cij and the
temperature T, the energy equation (5.45) may be expressed in the form
150 LINEAR ELASTICITY [CHAP. 6

kTBii= p~'"' !i' + (3h + 2,.~)aT,,&


which is known as the coupled heat equation.
The system of equations that formulate the general therrnoelastic problem for an
isotropic body consists of

( a ) equations of motion ..
~ , , ~ + p b ~ = oru ~ V o Z + p b = Ü

( b ) thermoelastic constitutive equations


uii = Xaijrkk + BPcij - (3A+ 2 , . ~ ) a 8 ~-~To)
(T

I: = Xlr + ~ P -E (3A + 2 ~ ) a l ( T
- To)

( c ) strain-displacement relations

( d ) coupled heat equation

This system must be solved for the stress, displacement and temperature fields, subject to
appropriate initial and boundary conditions. In addition, the compatibility equations must
be satisfied.
There is a large collection of problems in which both the inertia and coupling effects
may be neglected. For these cases the general thermoelastic problem decomposes into two
separate problems which must be solved consecutively, but independently. Thus for the
uncoupled, quasi-static, thermoelastic problem the basic equations are the

(a) heat conduction equation

( b ) equilibrium equations
a i j , j + p b i =O ~r V0E+pb= O

( c ) thermoelastic stress-strain equations


aij = XSijckk + 2 p i j - (3A + 2 ~ ) a (TS ~-~T O )
E Ale +~ P -
E (3X + 2,.L)aI(T- To)
( d ) strain-displacement relations
CHAP. 61 LINEAR ELASTICITY 151

Solved Problems
HOOKE'S LAW. STRAIN ENERGY. ISOTROPY (Sec. 6.1-6.3)
6.1. Show that the strain energy density u* for an isotropic Hookean solid may be
expressed in terms of the strain tensor by u* = X(tr E)V2 /.LE: E, and in terms of +
+
the stress tensor by u* = [(I v)Z: I: - ~ ( tZ)2]/2E.
r
Inserting (6.21) into (6.13), u* = ( X S i j e k k + 2peij)e i j / = Xeiiyjl2 + peijyj which in symbolic
notation is u* = X(tr E)2/2 pE : E.+
+
Inserting (6.24) into (6.13), u* = o i j [ ( l v)oij - vSijukk]/2E= [(I + v)oijoij- voiiojj]12Ewhich in
+
symbolic notation is u* = [ ( 1 v)L : L - ~ ( tL)2]/2E.
r

6.2. Separating the stress and strain tensors into their spherical and deviator components,
express the strain energy density v,* as the sum of a dilatation energy density u&)
and distortion energy density u&).
Inserting (3.98) and (2.70) into (6.13),

+ +
u* = &(sij okksij/3)(eij e,,Sij/3) + +
= +(sijeij oiiejj/3 siieii/3 + u,ejj/3)
+
and since eii = sii = O this reduceS to U* = u'lS> uTD,= oiiejj/6 sijeij/2. +

6.3. Assuming a state of uniform compressive stress uij = -pai,, develop the formulas
for the bulk modulus (ratio of pressure to volume change) given in (6.25).
With uij = -pSii, (6.24) becomes eij = [ ( I V)(-pSii) + +
vSij(3p)]lE and so yi = [ - 3 p ( l + V) +
9pvIIE. Thus K = -p/eii = E / 3 ( 1 - 2v). Likewise from (6.21), oii = ( 3 1 2p) eii = -3p +
so that
+
K = (3h 2p)/3.

6.4. Express u:,, and u&, of Problem 6.2 in terms of the engineering constants K and G
and the strain components.
From a result in Problem 6.3, uii = 3Keii and so
u : ~ )= uiiejj/6 = Keiiejj/2 = K(IE)2/2
From (6.21) and (2.70), oij = ASij ekk + 2peii = sij + okkSij/3 and since oii = (3X + 2p)eii i t foiiows
that sij = 2p(eij - ekk Sijl3). Thus
u : ~ )= zP(eij - ~ ~ ~ / 3 ) (=~ p ~
~ ~ 8 - ( ~ i~j ~-
i i eiieji/3)
Note that the dilatation energy density u:,) appears a s a function of K only, whereas the distortion
energy u'lD, is in terms of p (or G), the shear modu?us.

6.5. In general, u* may be expressed in the quadratic form u* = c ~ ~ in z which


~ E the~
C&, are not necessarily symmetrical. Show that this equation may be written in the
form of (6.14) and that au*la~,= a,.
Write the quadratic form a s
U* = & c Z M e K c M + - & c K ~ ~=~ E&CKMeKeM
~ ++ ~ : ~ e= +
~ e& (~C z M c M K ) e K e M = *CKMeKeM
where C K M= C M K .
Thus the derivative au*laeR is now
+ + + C K R E K=
au*/aeR = ~ C K M (R~' MK , ~ K ~ M ,=R )& C K M ( ~ K R EEMK ~ M R ) = J ( C R M E M =
) CRM~M "R
152 LINEAR ELASTICITY [CHAP. 6

Show that for an orthotropic elastic continuum


(three orthogonal planes of elastic symmetry)
the elastic coefficient matrix is as given in
(6.19), page 142.
Let the zlx2 (or equivalently, z;xh) plane be a
plane of elastic symmetry (Fig. 6-4). Then UK = CKMcM
and also u a = CKMcM. The transformation matrix
between xi and zl is

Fig. 6-4
and from (2.27) and (J.78), o;( = UK, = cK for K = 1,2,3,6 whereas UL = -UK, eg = -cK
for K = 4,5. Thus, for example, from c; = CIMcM,

These two expressions for U; = o, are equal only if C14 = C15 = O. Likewise, from o; = o,,
03 = u3,U; = -v4, U; = -u5, U; = o, i t is found that C% = C25 = = C35 = C64 = C65 C41 =
= c43 = = = C53 = C5@= 0.
If Z2x3 (or XZX:) is a second plane of elastic symmetry such that a; = CKMcM, the trans-
formation array is

I1 -
and now from (2.27) and (5.78), 02 = UK, cK - -€K for K = 1,2,3,4 whereas O; = -UK,
)I -
c~ - -cK for K = 5,6. Now C16 = C2, = C3, = C45 = C54 = = Cs2 = Cs3 = O and the elastic
coefficient matrix attains the form (6.19). The student should verify that elastic symmetry with
respect to the (third) xlx3 plane is identically satisfied by this array.

6.7. Give the details of the reduction of the orthotropic elastic matrix (6.19) to the
isotropic matrix (6.20).
For isotropy, elastic properties are the same with respect to a11 Cartesian coordinate axes.
In particular, for the rotated zl axes shown in Fig. 6-5, the method of Problem 6.6 results in the
matrix (6.19) being further simplified by the conditions Cll = CZ2= C33, C44 = C55 = and
C12 = = Cl3 = = Cz3 = c32.

Fig. 6-5 Fig. 6-6


CHAP. 61 LINEAR ELASTICITY 153

Finally, for the axes x:'obtained by a 45O rotation about x3 a s in Fig. 6-6 above, the transformation
matrix is

[aijl =

so that o[ = (o, - u1)/2 = (Cll - Clz)(az- 4 1 2 and e; = e, - e,. But o[ = C44~[ and so
2C4, = C,, - C,,. Thus defining p = C44 and X = C1,, (6.20) is obtained.

6.8. Give the details of the inversion of (6.21)to obtain (6.22).


+ 2p)
From (6.21) with i = j, qi = (3X + 2P)Yi and so 2 p i j = uij - XSij~kk/(3X or eij = oij/2p -
+
X8ii~kk/2p(3X 2p).

6.9. Express the engineering constants V and E in terms of the Lamé constants h and p.
From (6.25), E l ( 1 - 2 ~ =
) 3X +
2p; and from (6.26), E l ( l + v ) = 2p. Thus (3X + 2p)(1- 2v) =
2p(l + +
v ) from which v = X/2(X p). Now by (6.26), E = 2p(l + v) = p(3X f 2p)/(X + p).

6.10. Determine the elastic coefficient matrix for a continuum having an axis of elastic
symmetry of order N = 4. Assume CKM= CMK.
Let x3 be the axis of elastic symmetry. A rota-
tion e = 2a/4 = r12 of the axes about x3 produces
equivalent elastic directions for N = 4. The trans-
formation matrix is

[aiil =
o o
and by (2.27) and (3.78), o; = o,, o; = o,, 04 = o,,
o: = -05, o; = u4, u; = and e: = e2, E; = cl, = '3,
1 -
e4 - -e5, 4 = e4, e: = -eg. Thus, for example, from
o3 = 83, C34= C35= = 0, C3, = C3,. Likewise, from
the remaining five stress relations, the elastic matrix
becomes Fig. 6-7

[CKMI =

with seven independent constants.

ELASTOSTATICS. ELASTODYNAMICS (Sec. 6.4-6.5)


6.11. Derive the Navier equations (6.31).
Replacing the strain components in (6.28) by the equivalent expressions in terms of displace-
+ +
ments yields oij = AGijuk,, lr(ui,j u ~ , ~ )Thus
. oij,j = Au,,,, + +
$(ui,i j ujPij).Substituting this
into the equilibrium equations (6.27) and rearranging terms gives p u , j j + +
( X lr)uj,ji +
pbi = 0.
154 LINEAR ELASTICITY [CHAP. 6

6.12. Show that if V4Fi = 0, the displacement ui = ( A 2p)Fi,jiIp(A p ) - Fj,j i l p + + is a


solution of the Navier equation (6.31) for zero body forces.
Differentiating the assumed soiution, the terms @ui,jj= ( A 2 p ) F i , k k j j / ( A P ) - F k , k i j j and +
(A + + +
= ( A 2 p ) F j , k k j i / P - (A p ) F k , k j j i / P are readily calculated. Inserting these into ( 6 . 3 1 )
gives
+ +
( A 2 G ) F i , k k j j / ( A P ) - [ P - ( A + 2 P ) f ( A + P)IFj, jkki = O
provided Fi,kkjj= V 4 F i = 0.

6.13. If body forces may be neglected, show that (6.35) is satisfied by ui = +,i +c ~ ~ ~ $
provided + and $i each satisfies the familiar three dimensional wave equation.
Substituting the assumed ui into ( 6 . 9 5 ) yields
+
~ ( @ , i k k eijk$'k,jqq) + ( A + ~ ) ( @ , j j i + €jpq$'q.pji)
.. + eijk$'k,j)..
= P( @,i
Since ypq\l.q,pji= 0, this equation may be written
( ( A f 2 1 ~ ) @ , k k- P ? ) , i + 'ijk(~$'k,qq
..
- ~ $ ' k ) j, = O
= p?/(A
which is satisfied when V 2 @ +2 ~ and
) V2$'k = pYk/fl.

6.14. Writing c2V2+= where c2 = ( A + 2 p ) / p for the wave equation derived in Problem
6.13, show that + = g(r
ct) r h(r - ct) is a solution with g and h arbitrary func-
+ +

tions of their arguments and r2= X i x i .


Here i t is convenient to use the spherical form V 2 -- (r2:) since @ = + ( r , t ) . ~ h u s ;C
r 2 ( a @ l a r ) = r ( g ' + h ' ) - ( g + h ) where primes denote derivatives with respect to the arguments of
g and h. Then V 2 4 = (g" +
h")/r. Also 4
= ( g ' c - h r c ) / r and ;b' = c2(g1' hl')lr. Therefore +
c 2 V 2 @ = ;6. for the given @.

6.15. Derive the Beltrami-Michell equations (6.33) and determine the form they take when
body forces are conservative, i.e. when p b i = +,i.
Substituting ( 6 . 2 4 ) into ( 3 . 1 0 3 ) yields

where 8 = I= = uii. Only six of the eighty-one equations represented here are independent. Thus
setting m = k and using ( 6 . 2 7 ) gives

from which = -(i + ~ ) p b ~ , ~-/ (v l) . inserting this expression for O , k k into the previous
equation leads to ( 6 . 9 3 ) .
If p b i = @ , i , then p ( b i , j + bj,i) = 2@,ij and p b k , , = @ , k k = V 2 @ so that ( 6 . 3 3 ) becomes

TWO-DIMENSIONAL ELASTICITY (Sec. 6.6-6.8)


6.16. For plane stress parallel to the x1x2plane, develop the stress-strain relations in terms
of A and Show that these equations correspond to those given as (6.41).
Here u~~ = nl3 = uz3 = O so that ( 6 . 2 1 ) yields e13 = ep3 = O and €33 = -A(e11+ e Z 2 ) / ( A 2 ~ ) . +
Thus ( 6 . 2 1 ) reduces to o,@ = 2ApS,Bey,/(X 2 ~ ) + +
with a , P, y = 1 , 2 from which o,, =
+ +
2 ~ ( 3 A 2 p ) ~ , , / ( ~ 2 ~ so
) that the equation may be inverted to give

Also, €33 = -Ae,,/(A +2@) = -Aoyy/2P(3), +2p) = -voyy/E


CHAP. 61 LINEAR ELASTICITY 155

6.17. F o r plane strain parallel to x1x2,develop the stress-strain relations in terms of V and

becomes
inverting,
-
E. Show that these equations correspond to those given a s (6.48).
Here u3 O so that eg3 = O and (6.24) gives 033 = v(oll o,,) = Ao,,/2(A
+ +
+
+
= (1 v)o,plE - v ( l v ) S a p o Y y l E from which e,, = ( 1 v ) ( l - 2v)o,,lE.
+
p). Thus (6.24)
Finally,

o,p = vES,peyyl(l+ v ) ( l - 2") Ee,pl(l +


v) = AS,peyy + 2pe,p +

6.18. Develop the Navier equation for plane stress (6.45) and show that i t is equivalent to
the corresponding equation f o r plane strain (6.51) if A' = 2 A p / ( A 2 p ) is substituted +
f o r A.
Inverting (6.41) and using (6.42) leads to a,p = E ( U , , ~ up,,)/2(1 + + v) + 2vES,puy,y/2(1 - v2).
Differentiating with respect to x p and substituting into (6.40) gives
Eu,, pp/2(1+ v) + Eup,p,/Z(l- v) + pb, = h V 2 u a + p(3A + 2 p ) ~ ~ , p , / (+h 2p) + pb, = O
Thus since p(3X + 2p)/(A+ 2p) = (2Ap/(A+ 2p) + p) = (A' + p), (6.45) and (6.51) have the same form
for the given substitution.

6.19. Determine the necessary relationship between the constants A and B if + =


+
AxSxi Bx2 is to serve a s a n Airy stress function.
By (6.56), + must be biharmonic or +,1111 2+,1122 + + +,2222= O + 2 4 A x 2 + 120Bx2 = 0 , which
is satisfied when A = -5B.

6.20. Show that += xi is suitable f o r use a s an Airy stress function


and determine the stress components in the region x t > O, -c < xz < C.
Since 84+ is identically zero, + is a valid stress function. The stress components as given by
(6.55) are oll = +,22 = -3Fx1x2/2c3 + P/2c, = -3F(c2 - x i ) / 4 c 3 , 0 2 2 = +,il = 0.
uiz =
These stresses are those of a cantilever beam subjected to a transverse end load F and an axial
pull P (Fig. 6-8).

Fig. 6-8

6.21. In Problem 2.36 it was shown that the equilibrium equations were satisfied in the
absence of body forces by uij = c ~ ~ ~ c Show
~ ~ ~ that+ Airy's
~ ~ ,stress
~ ~ function
. is
represented by the case +,,
= +(xl,x,) with = = = = +,, +,, +,, +,, +,,
0.
Since +,,is the only non-vanishing component, oij = eipqejmn+qn,pm becomes q j = eip3ejrn3+33,pm
which may be written oap = E , ~ ~ E ~ y t~. ~ Thus
+ ~ ~ ,since +33 = +, ",B = ( S a ~ S y ( ;- =( ;
SYB)@,~
6,p+,yy - +,,p . The stress components are therefore 0 1 1 = +,li +
+,22 - +,li = +,22, 0 1 2 = -+,12,
" 2 2 = 9.11 + +,22 - +,22 = +,
11.
156 LINEAR ELASTICITY [CHAP. 6

6.22. In polar coordinates (r, O ) the Airy stress function


CJ= BB is used in the solution of a disk of radius
a subjected to a central moment M. Determine the
stress components and the value of the constant B.
From (6.60) and (6.61), o(,,, = o(oe) = O. From (6.62)
o(,,, = Blr2. Equilibrium of moments about the center of

the disk requires M =


Thus B = M 1 2 ~ .
2r
de) =
~(,~ d Jo' 2r
B de = 2nB.
Fig. 6-9

LINEAR THERMOELASTICITY (Sec. 6.10)


6.23. Carry out the inversion of (6.69) to obtain the thermoelastic constitutive equations
(6.70).
From (6.69) with i = j, aii = (3X + - 3a(T - T o ) ) . Solving (6.69) for oij gives

uij = 2peii + X S i j o k k / ( 3+~ 2p) - 2aa8ij(T - T o )


= 2pyj + XSij(ekk - 3a(T - T o ) )- 2paSij(T - T o )
= 2peij + ASijekk - (3X + 2p)c18~~(T - TO)
I

6.24. Develop the thermoelastic energy equation (6.73) by use of the free energy f = u - Ts.
Assuming the free energy to be a fynction of the strains and temperature, f = f(e,, T ) and
substituting into (5.41) p& +
aijS pTs where dots indicate time derivatives, the result is
+
(oij- pôf/dsij)iij - p(s d f l d T ) T = O. Since the terms in parentheses are independent of strain
~ ~ s = -dfldT.
and temperature rates, i t follows that uij = p d f l a ~ and From (5.38) for a reversible

isothermal process, =p ~ =
(a:j
S pT - i.. dTas
+ ">
- T . A t constant deformation, Zij = 0 and

comparing this equation with (6.72) gives c(") = T(ds/dT) or from above, since ds/dT = -a2flaT2,
c(") = - d Z f l d T 2 . = dSjldT and so combining ( 5 3 8 ) with (6.71),
Also, from above, p(d2fld~ijdT)

-ci,i = kTBii = pT -iij


(YT
+ )-
:c
T .
) Finally from (6.70), doil/dT = (3X+ so that
kT,ii = pc(")T + (3X + 2p)aT0eii which is (6.73).
6.25. Use (6.13) and (6.70) to develop the strain energy density for a thermoelastic solid.
Substituting (6.70) directly into (6.13),
+
U* = X8ij~kk€ii/2 peijeij - (3X + 2p)aSij(T- To)eij/2
= Xeiiejj12 + peijeij - (3X + 2p)a(T - To)eii12

MISCELLANEOUS PROBLEMS
6.26. Show that the distortion energy density u:,, may be expressed in terrns of princigal
stress values by the equation uy,, = [ ( a , - (a2 - u J 2 +
( a 3- ~ ~ ) ~ ] / 1 2 G . +
\
From Problem 6.2, u:,) = sijeij/2 = siisij/4G which in terms of stress components becomes
u f D , = (uij-8ijukk/3)(aij- 8ijop,/3)/4G = (oijoij-oiioji/3)/4G

In terms of principal stresses this is


ufD) = [o: + o ; + o: - (a1+ o2 + o3)(u1+ + 03)/3]/4G 42

= [2(u; + a: + o: - olo2 - u 2 ~ 3- a3u1)/3]/4G


= [(vl- u2)2 + (v2- u3)2 + - u1)2]/12G
(03
CHAP. 61 LINEAR ELASTICITY 157

6.27. Use the results of Problem 6.1 to show that for an elastic material d u * l d ~ =
, ~ uij and
~ u * I ~=u É~i j . ,
+
From Problem 6.1, u* = Xeiiejjl2 prijsij and so
+ +
Ju*larPq = X/2[eii(aejjla~pq)ejj(âciilòepq)] Z p ~ ~ ~ ( a e ~ ~ l a $ , )
= X/2[eii8jp8jq + ~ ~ ~+ 82 / ~J ~~ ~ 8=~ 8X/2[€iispq
~ ~ ~~ 8] $
~ ~ € j j s p q ]f 2,Urpq
-
- heiiSpq + 2pePq = upq
Likewise from Problem 6.1, u* = [(I + v)uijuii - ~ u ~ ~ u and
~ ~so] / ~ E
au"la,, = [2(1+ u)uijSipSjq- v(uiiôpq + ujjôW]/2E = [ ( I + v)uPq- u8pquii]lE = epq

6.28. Express the strain energy density u* as a function of the strain invariants.
From Problem 6.1, u* = Xeiiqj/2 + priirij; and since by comparison with (3.91), I, = cii and
11, = (riirjj- 4 j Y j ) / 2 , i t follows that

6.29. When a circular shaft of length L and radius


a is subjected to end couples as shown in Fig.
6-10, the nonzero stress components are
a13 = -G<Yx~,u~~= G a x l where a is the angle
of twist per unit length. Determine expres-
sions for the strain energy density and the
total strain energy in the shaft.
+
From Problem 6.1, u* = [ ( I v ) I :I - v(tr I ) 2 ] / 2 E .
Here t r I O and I : I = 2G2a2r2 where r 2 =
x: +x:. Thus u* = Ga2r2/2. The total strain
energy is given by Fig. 6-10

Note that since T = LPT +


f ~ a ( % : x:)r dr de = Gaa4n-12, U = TaLI2, the externa1 work.

6.30. Show that for a continuum having an axis of elastic symmetry of order N = 2, the
elastic properties (Hooke's law and strain energy density) are of the same form as a
continuum having one plane of elastic symmetry.
Here a rotation of axes e = 2 ~ 1 N= 2 ~ 1 2= n produces equivalent elastic directions. But this
is precisely the same situation a s the reflection about a plane of elastic symmetry.

6.31. Show that (6.19) with C11 = C22 = C33, C44 =


Cs5= C66 and C12 = C13 = C23 may be reduced to
(6.20) by an arbitrary rotation í3 of axes about 2 3
(Fig. 6-11).
The transformation between xi and xf axes is

cos 0 sin 0 0

and from (2.27), Fig. 6-11


158 LINEAR ELASTICITY [CHAP. 6

ai2 = (-sin e cos e)ull + (cos2 e - sin2 e)o12 + (sin 6 cos @)oz2
or i n single index notation,
= (-sin e cos e)ul + (cos2 8 - sin2 e)a6 + (sin e cos e)u2
Likewise from (3.78)and (6.4),
= (-2 sin e cos e)€,
e; +
(cos2 e - sin2 @)e6 ( 2 sin e cos @)e2 +
B u t a; = C44€; for a n isotropic body and so here u2 - ai = 2C44(€2- cl). Finally f r o m (6.19) w i t h
the given conditions, ul = Cllel i- +
C12(e2 e l ) and q = CtlY C12(el e,) +
and so o2 - ul = +
(Cll - C12)(e2- el). Therefore (Cll - C12)= 2C44 and w i t h C* = p, G12 = h, Cll = h 2p as +
given i n (6.20).
A

6.32. For an elastic body in equilibrium under body forces bi and surface forces ti"', show
that the total strain energy is equal to one-half the work done by the externa1 forces
acting through their displacements ui.
I t is required t o show t h a t
A
i pbiui dV +
Jv
t:"ui d S = 2 u* dV. Consider first the surface
integral w i t h ti("' = ujiuj and convert b y Gauss' theorem. T h u s

L uijuinjd S = l, (cijui), dV =
L (aij,jui + uijui, j) dV

B u t a i j u b j= uij(cij+ oij)= uijeij, and f r o m equilibrium oij,j = -pbi Thus

4 t$'ui d S Sv
= - pbiui dV +2 S aiieij/2 dV

and t h e theorem is proved.

6.33. Use the result of Problem 6.32 to establish uniqueness of the elastostatic solution of a
linear elastic body by assuming two solutions O : ) , u:') and O : ; ) , ui2).
For linear elasticity superposition holds, so uij = b:;) - a:;', ui = uil' - ~ 1 would
~ ) also be
a solution for which bi = O. T h u s for this "difference" solution
LA t:")uidS = 2
Problem 6.32. SinceAthe t w o assumed solutions satisfy boundary conditions, t h e l e f t hand integral
Jv
u* dV from

is zero here since tln' = tu'


i - ti2' on the boundary for equation (6.32) and ui = u:l' - u'i2' on
the boundary for equation (6.5'0). Thus rs* dV = O and since u* is positive definite this occurs
Jv
only i f eij = e:!) - e.. i2)
0 , or e::) = e!?). I f the strains are equal for the t w o assumed solu-
tions, t h e stresses are also equal b y Hooke's law and the displacements are equal t o within a rigid
body displacement. T h u s uniqueness is established.

6.34. The Navier equations (6.31) may be p u t in the form pui,s + uj.ji + pbi = O
which for the incompressible case ( V = 4) are clearly indeterminate. Use the equilib-
rium equations for this situation to S ~ O Wthat pUi,jj @,i13 pbi = 0. + +
From equation (6.24), eii = ( 1 - 2v)aiilE; and for v = Q, eii uiBi= O. T h u s f r o m (6.24),
2cij, = u 1.33
. .. u 3,13 + +
. .. = 2(1 v)aij,jlE - 2v8ijukk,j/E
B u t uj,ji= O and E = 3G when v = 3, so t h a t ui,j j = -pbilG - ukkJ3G or pV2ui + e j i / 3+ pbi = 0.

Supplementary Problems
- 6.35. Prove t h a t the principal axes o f the stress and strain tensors coincide for a homogeneous isotropic
elastic body.
6.36. Develop the expression for the strain energy density u* for an orthotropic elastic medium. Use
equations (6.14) and (6.19).
+ +
U* = ( C l l ~fl 2C12e2 2 C 1 3 ~ 5 ) ~f1 / 2(C22e2f 2c23e4)"2/2 C3363 f C44'42 Cs5e: f C66ei. +
CHAP. 61 LINEAR ELASTICITY 159

6.37. Determine the form of the strain energy density for the case of ( a ) plane stress, ( b ) plane strain.
Ans. ( a ) u* = [o;, + + +
- 2volla22 2(1 v ) o y 2 ] 1 2 ~
( b ) u* = (a + ~/2)(e?1+egz) + Xe11'22 + 2aef2
2a
6.38. Determine the value of c for which ul = A sin- ( x l c t ) , u 2 = u3 = O is a solution of equation
I
( 6 3 5 ) when body forces are zero. Ans. c = d w
6.39. Show that the distortion energy density uyD,= (uijoii- oiiojj/3)/4G and the dilatation energy density *
urS, = oiiajjl18K.

6.40. Show that 1 / ( 1 + V ) = 2(X + & ( 3 ~+ 2p) and vl(1- V) = X/(h + 2p).
6.41. For plane strain parallel to x,x,, show that b, O and that b, and b2 are functions of x , and
x, only.

6.42. Use the transformation laws for stress and strain to show that the elastic constants Cijk,,, are the
components of a fourth order Cartesian tensor so that Ci;km = aipaiqakTa,,,sCpqrs.

6.43. +
Show that the Airy stress function + = 2%; 12x;x; - 6x2 satisfies the biharmonic equation
V4+ = O and determine the stress components assuming plane strain.
2: - 3x2 -2x,x,
Ans. <rij = 24
o 2v(x; - x ; )

6.44. Determine the strains associated with the stresses of Problem 6.43 and show that the compatibility
equation (6.44) is satisfied.

Am. eij = 24 (%) ( x; - 32; - 24%; - 2 ; )


-2xlx2
o
x;
-2x1x2
+ x; - 24%: - x ; )
o
O

6.45. For an elastic body having an axis of elastic symmetry of order N = 6, show that C2- = C,,,
C,, = CG, Ce6 = 2(Cll - C I 2 ) and that C13 and C33 are the only remaining nonzero coefficients.

6.46. Show that for an elastic continuum with conservative body forces such that pba = V+ = $,,, the
compatibility condition (6.44) may be written V20aa = Q2+/(1 - V ) for plane strain, or
820aa = (1 +
v ) V 2 + for plane stress.

6.47. If V4Fi = 0, show t h a t ui = 2(1- v)V2Fi/G - FjBji/G is a solution of the Navier equation (6.31)
when bi O (see Problem 6.12). If F = B ( x 2 ê l - x l ê 2 ) / r where r 2 = xixi, determine the stress
components.
Ans. o,, = -oz2 = 6QGx1x2/r5, = 0, u12 = 3QG(x; - 4 ) / ~ 5 ,u13 = -oz3 = 3QGx2x3/r5, where
Q = 4 B ( 1 - v)lG.

6.48. In polar coordinates an Airy stress function is given by @ = Cr2(cos 2e - cos 2a) where C and a
are constants. Determine C if use = O, ora = T when e = a, and = O , a,, = -T when
e = -a. Ans. C = r / ( 2 sin 2a)

6.49. Show that in plane strain thermoelastic problems 0 3 =


~ v(oll +
oZ2)- a E ( T - T o ) and that =
+
hSapeaa 2 p n B - Sap(3X 2p)a(T - To). In plane stress thermoelasticity show that
€33 = -v(ul1 + + +
aZ2)lE a(T - T o ) and eap = ( 1 v)aaBIE- vSaPoauIE Sap ( T - To)a +
6.50. In terms of the Airy stress function + = +(xl,x2), show that for plane strain thermoelasticity the
compatibility equation (6.44) may be expressed as V4+ = - a E V 2 ( T - T o ) / ( l- v ) and that for plane
stress as V 4 + = -aEV2(T - To).
Chapter 7

Fluids

7.1 FLUID PRESSURE. VISCOUS STRESS TENSOR. BAROTROPIC FLOW


In any fluid a t rest the stress vector tin)on an arbitrary surface element is collinear with
the normal íi of the surface and equal in magnitude for every direction a t a given point.
Thus A
A
tin) = ~23. .3 n=. -poni or t(^) = 1-íi = -p,n (r*1)
in which po is the stress magnitude, or hydrostatic pressure. The negative sign indicates a
compressive stress for a positive value of the pressure. Here every direction is a principal
direction, and from (7.1)
a , , = -p o s i j or 1 = -p O I
which represents a spherical state of stress often referred to as hydrostatic pressure.
From (7.2), the shear stress components are observed to be zero in a fluid a t rest.
For a fluid in motion, the shear stress components are usually not zero, and it is cus-
tomary in this case to resolve the stress tensor according to the equation

where rii is called the viscous stress tensor and p is the pressure.
A11 real fluids are both compressible and viscous. However, these characteristics vary
widely in different fluids so that it is often possible to neglect their effects in certain situa-
tions without significant loss of accuracy in calculations based upon such assumptions.
Accordingly, an inviscid, or so-called perfect fluid is one for which rij is taken identically
zero even when motion is present. VZscous fluids on the other hand are those for which
rij must be considered. For a compressible fluid, the pressure p is essentially the same as
the pressure associated with classical thermodynamics. From (7.3), the mean normal stress
is given by

For a fluid a t rest, r , vanishes and p reduces to p, which in this case is equal to the negative
of the mean normal stress. For an incompressible fluid, the thermodynamic pressure is
not defined separately from the mechanical conditions so that p must be considered as an
independent mechanical variable in such fluids.
In a compressible fluid, the pressure p, the density p and the absolute temperature T are
related through a kinetic equation of state having the form

An example of such an equation of state is the well-known ideal gas law .


CHAP. 71 FLUIDS 161

where R is the gas constant. If the changes of state of a fluid obey an equation of state
that does not contain the temperature, i.e. p = P ( ~ ) such
, changes are termed barotropic.
An isothermal process for a perfect gas is an example of a special case which obeys the
barotropic assumption.

7.2 CONSTITUTIVE EQUATIONS. STOKESIAN FLUIDS. NEWTONIAN FLUIDS


The viscous stress components of the stress tensor for a fluid are associated with the
dissipation of energy. In developing constitutive relations for fluids, i t is generally assumed
that the viscous stress tensor T~~ is a function of the rate of deformation tensor D i j . If the
functional relationship is a nonlinear one, as expressed symbolically by

the fluid is called a Stokesian fluid. When the function is a linear one of the form

where the constants Ki,, are called viscosity coeficients, the fluid is known as a Newtonian
fluid. Some authors classify fluids simply as Newtonian and non-Newtonian.
Following a procedure very much the same as that carried out for the generalized
Hooke's law of an elastic media in Chapter 6, the constitutive equations for an isotropic
homogeneous Newtonian fluid may be determined from (7.7) and (7.3). The final form is

where A* and p* are viscosity coefficients of the fluid. From (7.9)' the mean normal stress
is given by

+
where K* = Q(3A* 2p*) is called the coeficient of bulk viscosity. The condition that

is known as Stokes' condition, and guarantees that the pressure p is defined as the average
of the normal stresses for a compressible fluid a t rest. In this way the thermodynamic
pressure is defined in terms of the mechanical stresses.
In terms of the deviator components sij = aii - 6ijokk/3and D i = Dii - 6iiD,k/3, equa-
tion (7.9) above may be rewritten in the form

Therefore in view of the relationship (7.10),equation (7.12) may be expressed by the pair
of equations
sij = 2p*D,; or S = 2p*Df (7.13)

the first of which relates the shear effects in the fluid and the second gives the volumetric
relationship.
162 FLUIDS [CHAP. 7

7.3 BASIC EQUATIONS FOR NEWTONIAN FLUIDS.


NAVIER-STOKES-DUHEM EQUATIONS
I n Eulerian form, the basic equations required to formulate the problem of motion for
a Newtonian fluid are
I

(a) the continuity equation (5.3),


= O or ; + p ( V x 0 v )= O (7.15)

(b) the equations of motion (5:16),

( c ) the energy equation (5.32),

(d) the constitutive equations ( 7 4 ,

( e ) the kinetic equation of state (7.5),


P = ~ ( p T, )
If thermal effects are considered, as they very often must be in fluids problems, the
additional equations

( f ) the Fourier law of heat conduction (6.71),

( g ) the caloric equation of state,


9-4 = ~ ( p T)
,
are required. The system of equations (7.15) through (7.21) represents sixteen equations
in sixteen unknowns and is therefore determinate.

If (7.18) above is substituted into (7.16) and the definition 2 0 , = ( v L j +


is used, the
equations that result from the combination are the Navier-Stokes-Duhem equations o f
motion,
+ (A* + p * ) ~ , ,+, ~p * ~ ~
pGi = pbi - pSi , ~ ~
or
pV = pb - VP + (A* + p * ) v ( v + p*V2v
V)

When the flow is incompressible ( v , = O), (7.22) reduce to the Navier-Stokes equations
for incompressible flow,

If Stokes condition is assumed (A* = -pp*), (7.22) reduce to the Navier-Stokes equations
for compressible flow
PVi = Pbi - P*i + *!J*vj,ji+ p*vcjj
or (7.24)
+
PV = pb - V p +p.*v(v V ) p*V2v +
CHAP. 71 FLUIDS 163

The Navier-Stokes equations (7.23), together with the continuity equation (7.15) form
a complete set of four equations in four unknowns: the pressure p and the three velocity
components vi. In any given problem, the solutions of this set of equations must satisfy
boundary and initial conditions on traction and velocity components. For a viscous fluid,
the appropriate boundary conditions a t a fixed surface require both the normal and
tangentiaI components of velocity to vanish. This condition results from the experimentally
established fact that a fluid adheres to and obtains the velocity of the boundary. For an
inviscid fluid, only the normal velocity component is required to vanish on a fixed surface.
If the Navier-Stokes equations are put into dimensionless form, severa1 ratios of the
normalizing parameters appear. One of the most significant and commonly used ratios is
the Reynolds number N,,, which expresses the ratio of inertia to viscous forces. Thus if
a flow is characterized by a certain length L, velocity V and density p, the Reynolds number
is
N,,, = VL/v (7.25)
where V = p*lPis called the kinematic viscosity. For very large Reynolds numbers, the
,

viscous contribution to the shear stress terms of the momentum equations may be neglected.
In turbulent flow, the apparent stresses act on the time mean flow in a manner similar to
the viscous stress effects in a laminar flow. If turbulence is not present, inertia effects
outweigh viscous effects and the fluid behaves as though it were inviscid. The ability of
a flow to support turbulent motions is related to the Reynolds number. It is only in the case
of laminar flow that the constitutive relations (7.18) apply to real fluids.

7.4 STEADY FLOW. HYDROSTATICS. IRROTATIONAL FLOW


The motion of a fluid is referred to as a steady flow if the velocity components are
independent of time. For this situation, the derivative avilat is zero, and hence the material
derivative of the velocity

reduces to the simple form


i i = V .J v . .
2.3 Or +=VV.VXV
A steady flow in which the velocity is zero everywhere, causes the Navier-Stokes equa-
tions (7.22) to reduce to
pbi = P,, or pb = VxP (7.28)
which describes the hydrostatic equilibrium situation. If the barotropic condition p = p(p)
is assumed, a pressure function
P(p) = spgpa P
(7.29)

may be defined. Furthermore, if the body force may be prescribed by a potential function

equations (7.28) take on the form

A flow in which the spin, or vorticity tensor (4.21),


164 FLUIDS [CHAP. 7

vanishes everywhere is called an irrotational fiow. The vorticity vector qi is related to


the vorticity tensor by the equation
q, = cijkVkj 01: q = V" (7.33)
and therefore also vanishes for irrotational flow. Furthermore,
Qi = or
zijkvk,j q = Vxv (7.34)
and since V x v = O is necessary and sufficient for a velocity potentkl 4 to exist, the
velocity vector for irrotational flow may be expressed by
v. = -4, or v = -V+ (7.35)

7.5 PERFECT FLUIDS. BERNOULLI EQUATION. CIRCULATION


If the viscosity coefficients h* and p* are zero, 'the resulting fluid is called an inviscid
or perfect (frictionless) ftuid and the Navier-Stokes-Duhem equations (7.22) reduce to
the form

which is known as the Euler equation of motion. For a barotropic fluid with conservative
body forces, (7.29) and (7.30) may be introduced so that (7.36) becomes
;i = -(n+P), or V = -v(a+P) (7.37)
For steady flow (7.37) may be written
v3. vt i 3. . = -(n+P),i or v.Vv = -V(n+P) (7.38)
If the Euler equation (7.37) is integrated along a streamline, the result is the well-
known Bernoulli equation in the form (see Problem 7.17)

For steady motion, avilat = O and C(t) becomes the Bernoulli constant C which is, in general,
different along different streamlines. If the flow is irrotational as well, a single constant
C holds everywhere in the field of flow.
When the only body force present is gravity, the potential n = gh where g is the
gravitational constant and h is the elevation above some reference level. Thus with h, = Plg
defined as the pressure head, and v2/2g= hv defined as the velocity head, Bernoulli's equa-
tion requires the total head along any streamline to be constant. For incompressible fluids
(liquids), the equation takes the form
h + h, + h" = h + plpg + v2/2g = constant (7.40)

By definition, the velocitz~circulation around a closed path of fluid particles is given by


the line integral
r, = f v i d x i or r, = vedx f (7.41)

From Stokes theorem (1.153) or (1.154), page 23, the line integral (7.41) may be converted
to the surface integral

where íiis the unit normal to the surface S enclosed by the path. If the flow is irrotational,
V x v = O and the circulation is zero. In this case the integrand of (7.42) is the perfect
differential d+ = -v. dx with the velocity potential.
CHAP. 71 FLUIDS 165

The material derivative drcldt of the circulation may be determined by using (4.60)
which when applied to (7.41) gives

For a barotropic, inviscid fluid with conservative body forces the circulation may be shown
to be a constant. This is known as Kelvin's theorem of constant circulation.

7.6 POTENTIAL FLOW. PLANE POTENTIAL FLOW


The term potential flow is often used to denote an irrotational flow since the condition
of irrotationality, V x v = O, is necessary and sufficient for the existence of the velocity
potential + of (7.35). For a compressible irrotational flow, the Euler equation and the
continuity equation may be linearized and combined as is done in acoustics to yield the
governing wave equation
..
+ = or ;= c2V2+ (7.44)
where c is the velocity of sound in the fluid. For a steady irrotational flow of a compressible
barotropic fluid, the Euler equation and continuity equation may be combined to give

which is the so-called gas dynamical equation.

For incompressible potential flow the continuity equation attains the form

and solutions of this Laplace equation provide the velocity components through the defini-
tion (7.35). Boundary conditions on velocity must also be satisfied. On a fixed boundary,
for example, d+ldn = O. An important feature of this formulation rests in the fact that
the Laplace equation is linear so that superposition of solutions is possible.

In a two-dimensional incompressible flow parallel to the x,x, plane, v, = 0, and the


continuity equation becomes
v,,, = o or B'V = o (7.47)
where, as usual in this book, Greek subscripts have a range of two. By (7.47), regardless
of whether the flow is irrotational or not, i t is possible to introduce the stream function
+ = +(x,,x,) such that
If the plane flow is, indeed, irrotational so that

then from (7.48) and (7.49) the stream function and velocity potential are seen to satisfy
the Cauchy-Riemann conditions
4.1 = q.2 and +,, = -+,i (7.50)
By eliminating + and + in turn from (7.50) i t is easily shown that
166 FLUIDS [CHAP. 7

Thus both + and + are harmonic functions when the flow is irrotational. Furthermore, the
complex potential
@(4 = +(x19 x,) + i+(x1, $2) (7.53)
is an analytic function of the complex variable, z = x , + ix, so that its derivative dqldz
defines the c o m p l e x v e l o c i t y
dqldz = -vl iv, + (7.54)

Solved Problems
FUNDAMENTALS OF FLUIDS. NEWTONIAN FLUIDS (Sec. 7.1-7.3)
7.1. Show that the deviator sij for the stress tensor a , of ( 7 . 3 ) is equal to t,, the deviator
of T i j of ( 7 . 3 ) .
From ( 7 4 , aii = -3p + rii and so here

7.2. Determine the mean normal stress o i i / 3 for an incompressible Stokesian (nonlinear)
+
fluid for which T , = a D i j pDikDkj where a and p are constants.
From ( 7 4 , aij = -pSij + +
aDij /3DikDkj and so ai< = -3p + aDii + /3DikDki. But Dik =
Dki and Dii = = O for an incompressible fluid so that
4 3 = -p + /3DijDij13 = -p - 2/311D/3
where 11, is the second invariant of the rate of deformation tensor.

7.3. Frictionless adiabatic, or isentropic flow of an ideal gas, is a barotropic flow for which
p = cpk where C and k are constants with k = c(*)/c("), the ratio of specific heat a t
constant pressure to that a t constant volume. Determine the temperature-density
and temperature-pressure relationships for such a flow.
Inserting p = Cpk into equation (7.6), the temperature-density relationship is # - l / T = RIC,
a constant. Also, since p = (plC)l/k here, (7.6) yields the temperature-pressure relationship a s
p(k-l)/k/T = R/Cl/k, a constant.

7.4. Determine the constitutive equation for a Newtonian fluid with zero bulk viscosity,
i.e. with K * = 0.
If K* O , A* = - 2 / ~ * / 3 by (7.11) and S O (7.9) becomes aii = -pSij - (2/l*/3)SijDkk+ 2/~*Dij
which is expressed in terms of the rate of deformation deviator by

If the deviator stress sij is introduced, this constitutive relation is given by the two equations
S i j = ~ / L * D ;and aii = -3p.
CHAP. 71 FLUIDS 167

7.5. Determine a n expression for the "stress power" cijDijof a Newtonian fluid having
equation (7.9) a s its constitutive relation.
From (7.9) and the stress power definition,

u 23 = -pSijDij
o..D.. + h*SijDkkDij+ 2p*DijDij = -pDii + h*DiiDjj + 2p*DijDii
In symbolic notation, this expression is written

I : D = -p(tr D) + A*(tr D)2 + 2p*D : D


In terms of D& the expression is

In symbolic notation,
: D = -p(tr D ) + ~ * ( tD)2r + 2p* D' : D'

7.6. Determine the conditions under which the mean normal pressure p,,, = -aii/3 is
equal to the thermodynamic pressure p f o r a Newtonian fluid.
With the constitutive equations in the form (7.13) and (7.14), the latter equation gives
P ( , > - - P = -K *Dii. Thus p(m, = p when K* = O (by (7.11) when A* = -Qa*) or when Dii = 0.

7.7. Verify the Navier-Stokes-Duhem equations of motion (7.22) f o r a Newtonian fluid


and determine the form of the energy equation (7.17) f o r this fluid if the heat conduc-
tion follows the Fourier law (7.20).
Since Dii = vi,i, equation (7.18) may be written aij = -pSii + h*Sijvk,k + p * ( v t j + v j J . Thus

o.. 3. = - p j j S i j
i13 + X * S i j ~ k , k+j p*(vi,jj + vj,ij) = -P,i (h* + p*)vj,ji + p*vi,jj
and with this expression inserted into (7.16) a direct verification of (7.22) is complete.

Substituting the above equation for qjtogether with (7.20) into the energy equation (7.17), the
result is
P; = [-psij h*6ijvk,k + + +
v ~ , ~ ) ] ( v v~j , i ) / 2- kT,ii +
pZ +
which reduces to

7.8. Determine the traction force T,acting on


the closed surface S which surrounds the
volume V of a Newtonian fluid f o r which
the bulk viscosity is zero.
A ds
The element of traction is dTi = t i n ' d S and

the total traction force is T i =


L A tin' dS which
because of the stress principle is T i =
From Problem 7.4, this becomes i ojinjd s .

Ti = (-p~ij + 2p*D;)9dS
for a zero bulk modulus fluid; and upon application
of Gauss' theorem,
/-----
Fig. 7-1
168 FLUIDS [CHAP. 7

7.9. In a n axisymmetric flow along the x3 axis the velocity is taken a s a function of x3 and
r where r2 = X : + x2. If the velocity is expressed by v = qê, v3ê3 where 6, is +
the unit radial vector, determine the form of the continuity equation.
Equation (5.4) gives the continuity equation in symbolic notation a saplat V (pv) = 0. +
1 d(rpg) i a(Pv3)
Here the cylindrical form of the operator V may be used to give V (pv) = --
Inserting this into (5.4) and simplifying, the continuity equation becomes r ar 8x3 '

7.10. In a two-dimensional flow parallel to the X I X Z plane, v3 and dldxa a r e zero. Determine
the Navier-Stokes equations for a n incompressible fluid and the form of the con-
tinuity equation for this case.
From (7.23) with i = 3, pb, = p,, and when i = 1 , 2 , p'U, = pb, - P,, + P * V , , ~ ~ .The con-
tinuity equation (7.15) reduces to v,,, = 0. 4

If body forces were zero and v , = v l ( x l , x,, t), v , = O , p = p ( x l , x,, t ) the necessary equations
would be pGl = -dp/axl + +
a * ( d z ~ l / d ~ : d 2 v 1 / d x i ) and dvl/axl = 0.

HYDROSTATICS. STEADY AND IRROTATIONAL FLOW (Sec. 7.4)


7.11. Assuming a i r is an ideal gas whose temperature varies linearly with altitude a s
T = T O- axg where T Ois ground leve1 temperature and X B measures height above
the earth, determine the a i r pressure in the atmosphere a s a function of x3 under .
hydrostatic conditions.
From (7.6) in this case, p = P R ( T , - CYX,); and from (7.28) with the body force b3 = -g, the
gravitational constant, dp/dx3 = - ~ g= - p g / R ( T o - ax3). Separating variables and integrating
yields ln p = ( g / R a ) ln ( T o- CYX,) +
ln C where C is a constant of integration. Thus p =
C ( T o- ax3)glRff and if p = po when x3 = O, C = po To-g/Rff and so p = po(l - a ~ 3 / T o ) g / R f f .

7.12. A barotropic fluid having the equation of state p = hpk where h and k a r e constants
is a t rest in a gravity field in the x3 direction. Determine the pressure in the fluid
with respect to x3 and po, the pressure a t x3 = 0.
From (7.28), dp/dx3 = -pg, d p / d x l = d p l d x , = O. Note that pressure in x , and x z directions
is constant in the absence of body forces b1 and bz. Since here p = (p/h)Uk, p-llkdp = -gh-1fkdx3
and integration gives ( k / ( k - l))pck-1)lk = + c. But p = po when x3 = O so that
C = ( k / ( k - 1 ) ) ( k - l ) / k . Therefore x3 = ( k p o l ( k- l ) g p o ) ( l- ( p / p 0 ) ( k - l ) f k ) where po = ( p o / ~ ) ~ ~ k .
po

7.13. A large container filled with an incompressible


liquid is accelerated a t a constant rate a = a2êz+
a3ê3 in a gravity field which is parallel t o the x 3
direction. Determine the slope of the free surface
of the liquid.
From (7.28), d p l d x , = 0 , d p l d x , = pa, and dp/dx3 =
-p(g - a 3 ) . Integrating, p = pa,x, +
f(x,) and p =
-p(g - a,)%, +
h ( x 2 ) where f and h are arbitrary functions
of their arguments. In general, therefore, p = pazx2 -
+
p(g - a,)%, po where po is the pressure a t the origin of
coordinates on the free surface. Since p = po everywhere I

on the free surface, the equation of that surface is x 2 / x 3 =


(9- a3)/az. Fig. 7-2
CHAP. 71 FLUIDS 169

7.14. If a fluid motion is very slow so that higher order terms in the velocity are negligible,
a limiting case known a s creeping flow results. For this case show that in a steady
incompressible flow with zero body forces the pressure is a harmonic function, i.e.
v 2 p = o.
For incompressible flow the Navier-Stokes equations (7.23) are
p(avilat + vjvi, +
= pbi - P , ~ p*vi, j j
and for creeping flow these linearize to the form

Hence for steady flow with zero body forces, p,i = Taking the divergence of this equation
yields pJii = P * ~ i , i j jand
; since the continuity equation for incompressible flow i s viai= O, i t follows
that here p., = V 2 p = 0 .

7.15. Express the continuity equation and the Navier-Stokes-Duhem equations in terms of
the velocity potential 4 for an irrotational motion.
By (7.35), v i = - @ , i so that from (7.15) the continuity equation becomes i- pV2@ = O. Also
with v i = -+,i, (7.22) becomes
-Pd,i = Pbi - P,i - (h* p*)+,jji - p*@,iii
or -p(a$,i/at + @,k@,ik) = pbi - P,i - (h* + 2p*)$,jji
In symbolic notation this equation is written
-pv(a+/at + (v+)2/2) = pb - v p - (A*+ z p * ) v ( v 2 + )

7.16. Deterinine the pressure function P(p) for a barotropic fluid having the equation of
state p = X p k where X and k are constants.
From the definition (7.29),

Also since d p = hkpk-1 dp, the same result may be obtained from

PERFECT FLUIDS. BERNOULLI EQUATION. CIRCULATION (Sec. 7.5)


7.17. Derive equation (7.39) by integrating Euler's equation (7.37) along a streamline.
Let dxi be an increment of displacement along a streamline. Taking the scalar product of this
increment with (7.37) and integrating gives

Since dxi = d a and P,i dxi = d P the last two terms integrate a t once. Also, along a stream-
line, dxi = (viIv)d s where ds is the increment of distance. Thus in the second integral,
v j v i , dxi = v j v i , ( v i / v )ds = vivi, ( v j / v )ds = v i v i , d x j = v i dvi

Therefore S v j v i , dxi =
s v . dvi = +vivi = +v2, and (7.39) is aehieved.
170 FLUIDS [CHAP. 7

7.18. The barotropic fluid of Problem 7.16 flows from a large closed tank through a thin
smooth pipe. If the pressure in the tank is N times the atmospheric pressure, deter-
mine the speed of the emerging fluid.
Applying Bernoulli's equation for steady flow between point A , a t rest in fluid of the tank and
+ +
point B, in the emerging free stream, ( 7 3 9 ) assumes the form aA PA & v i = aB PB & v i . + +
But v, = 0, and if gravity is assumed negligible this equation becomes (see Problem 7.16),

Since pBIPA = ( p B l p A ) - l J k= N - l l k the result may be written

7.19. Show that for a barotropic, inviscid fluid with conservative body forces the rate of
change of the circulation is zero (Kelvin's theorem).
From (7.M) :, = § (Gi dxi + vi dvi) and by (7.37), Gi = -(a + P),i for the case a t hand. Thus
i.,= $ (-qidxi - Paidxi + vi dvJ = - f (& + dP - d(vz12)) = - f P - P / 2 ) = 0 , the
d(Q-i- inte-
grand being a perfect differential.

rh
7.20. Determine the circulation around the square xl = k1,
x2 = 1+1, x3 = O (see Fig. 7-3) for the two-dimensional
+
f l o ~v = ( X I x ~ (x; +
) -~x ~~ ) ~ ~ .
Using the symbolic form of (7.42) with ^n = ê3 and
v x v = (2x1- 1 )ê3,

= s:.Sl 1
(2x1 - 1) d x , dx,

The same result is obtained from (7.41) where


= -4

Fig. 7-3
r, = $v-dx
1 -1
= (i-x2)dx2 + ~ - l ( x l + l ) d x l+ (1- x,) dx, + ( x , - 1) dxl = -4

with the integration proceeding counterclockwise from A.

POTENTIAL FLOW. PLANE POTENTIAL FLOW (Sec. 7.6)


7.21. Give the derivation of the gas dynamical equation (7.45) and express this equation in
terms of the velocity potential +.
+
For a steady flow the continuity equation (5.4) becomes pjivi pvi,i = O and the Euler equa-
+
tion (7.36) becomes pvjvi,j pIi = O if body forces are neglected. For a barotropic fluid, p = p(p)
and so dp = ( a p l a ~ ~ ) ( a ~ ~ or
l a rearranging,
~)d~; p,i = ( d ~ l d p ) p=, ~c ~ P where
, ~ c is the local velocity
of sound. Inserting this into the Euler equation and multiplying by vi gives pvivjvi,j +
c2vip,i = 0.
From the continuity equation c ~ v=~-C%. ~ 1,, t. = 21 ~3 .
~ -c2S..v. and so (c2Sij- vivj)vi,j= O. In terms
of + , i = -vi this becomes (c2Sij- @,i@, j)+,ij= 0.

+
7.22. Show that the function + = A(-x: - x2 2 x i ) satisfies the Laplace equation and
determine the resulting velocity components.
Substituting +
into (7.46) gives -2A - 2 A 4 A = O. From (7.3~9, v , = 2Ax,, v , = 2 A x Z ,
v 3 = -4Ax3. Also, by the analysis of Problem 4.7 the streamlines in the x , plane are represented
by x i x 3 = constant; in the x , plane by xSx3 = constant. Thus the flow is in along the x3 axis
against the x,x, plane (fixed wall).
CHAP. 71 FLUIDS 171

7.23. Show that the stream function +(xI,x2) is constant along any streamline.
From (7.48) and the differential equation of a streamline, dxl/vl = dx21v2 (see Problem 4.7),
=
-d~,l+,~ or +,ldxl += d+ = O. Thus + = a constant along any streamline.

7.24. Verify that + = A(xl - x t ) is a valid velocity x2


potential and describe the flow field.
For the given +, (7.46) is satisfied identically by
2 A - 2 A = 0 ; and from (7.49), v l = -2Axl, v2 =
+2Ax2. The streamlines are determined by integrat-
ing dxl/xl = -dx2/x2 to give the rectangular
hyperbolas xlx2 = C (Fig. 7-4). The equipotential
lines A ( X ; - 2;) = C1 form a n orthogonal set of
rectangular hyperbolas with the streamlines. Finally
+
from (7.50), + = -2Az1x2 CO and is seen to be con-
stant along the streamlines a s was asserted in Prob-
lem 7.23. Fig. 7-4

7.25. A velocity potential is given by + = Axl + Bx1Ir2 where r2 = x: + x:. Determine


+
the stream function for this flow.
From (7.50), +
= - G , ~ = 2Bx1z2/+ so that by integrating, + = -Bx21r2 f ( x 2 ) where f ( z 2 )
is an arbitrary function of z,. Differentiating, +,2 = -B(X; - s i ) / + +
f'(x2). But from (7.50),
+,2 = +
= A B(-z; + +
z8)/+. Thus f'(x2) = A and f ( x 2 ) = A x z C. Finally then $ =
+
A x z - Bx2/r2 C.

7.26. Differentiate the complex potential a(z) = Alx to obtain the velocity components.
Here dWdz = -AI22 = -A/(xl + ix2)2 which after some algebra becomes d<P/dz =
-A(%; - x;)/+ +
i2Ax1x2/+. Thus
v1 = A(%;- si)+ and v2. = 2Ax1x2l.P
Note that since @ = A l z = A(%,- iz2)/r2, + = Axllr2 and JI = -Ax21r2. Also note that
v, = = A ( x ; -X:)I+ and v2 = -9.2 = 2 A ~ ~ x ~ 1 7 . 4

MISCELLANEOUS PROBLEMS
7.27. Derive the one-dimensional continuity equation for the flow of an inviscid incompres-
sible fiuid through a stream tube.
Let V be the volume between arbitrary cross sec-
tions A and B of the stream tube shown in Fig. 7-5.
In integral form, for this volume (5.2) becomes
Sv V * v dV = O since p is constant here. Convert-
ing by Gauss' theorem,
LA n o v dS = O where " n s
the outward unit normal to the surface S enclosing V .
ntA
A
n
Since i. v on the lateral surface, the integration re-
duces to " " Fig. 7-5

A
The velocity is assumed unifonn and perpendicular over SA and S,; and since va = - v B n ~ ,
vA iA iB
dS - vB dS = O or v A S A = v B S B = a constant.
172 FLUIDS [CHAP.' 7

7.28. The stress tensor a t a given point for a Newtonian fluid with zero bulk viscosity is

c =
-6
(
2 -9
2 -1

-1 4 -3
Determine ).
From (7.14), for this fluid p = -uii/3 = 6. Then from ( 7 4 ,
.
(4-: :)
-6 2 -1 6 O 2 -1
oi; +6 8 or
-:: )+ =

7.29. Show that aij and T~~ of (7.3) have the same principal axes.
When written out, (7.8) becomes .oll = -p = -p
T ~ I ,u 2 ~ +722, u33 = -p +
T33, u12 = 7 1 2 , +
u23 = u13 = 7 1 ~ .For principal directions x: of qj, uTz = = uT3 = O and by the last three
equations of (7.39, o; = 7; = O for i # j. Thus Z: are principal axes for T~~ also.

7.30. A dissipation potential Q, is often defined for a Newtonian fluid by the relationship
a, = ( K / ~ ) D ; ~p,*D;D;.
D~~ +
Show that d a D l d D i j = r i j .
Here D~,)+
d+,laDpq = ( ~ / 2 ) [ D ~ ~ ( a D ~ ; l a (aDiilaDpq)Djj] +
2 5 [ D , ' , ( a ~ & 3 ~ , , ) ] . But aDiilaDpq =
~ , 6, , aj4 - S i j Spq/3 so that
Sip Si, = S p q and d ~ , ( ~ / a =

+
S ~ ~- 8ijDkk/3)(8ip8jq
d@~laD,, = K D ~ ~ 2p*(Dij - SijSpq/3) = K D + eP*(Dpq
~ ~- SpqDii/3)
~ ~ ~
Finally since K = A* + 2p*/3,
a+,/a~,, = ?,*sPqoii+ 2 p * ~ p q= Tpq

. 7.31. Determine the pressure-density relationship for the ideal gas discussed in Problem 7.11.
A t s3 = O, p = po and p = po. The ideal gas law (7.6) is here p = pR(To - as3) so that
po = poRT0; and from the pressure elevation relationship p = p,(l - a ~ ~ / T ~ ) g /ofR aProblem 7.11,
p l p o = (T/To)(g/Ra-i). Thus writing p = po(l - as31To)g/Ra in the form p/po = (T/To)g/Ra, i t is
seen that T / T o = (plpo) Ralo and so P/Po= (p/po)(l-Ra/g).

7.32. For a barotropic inviscid fluid with conservative body forces show that the material
derivative of the total vorticity,
d
qi dV = viqi dSi. aL
From (4.54) and the results of Problem 4.33, $ i qi dV =.L (ciik ak + qivi) dSj. But here
ak = -(R +P),, from (7.37); and by the divergence theorem (1.157),

since the integrand is zero (product of a symmetric and antisymmetric tensor). Hence

7.33. For an incompressible Newtonian fluid moving inside a closed rigid container a t rest,
show that the time rate of change of kinetic energy of the fluid is -,.L*
ing zero body forces. q is the magnitude of the vorticity vector.
CHAP. 'i] FLUIDS 173

From Problem 5.27, the time rate of change of kinetic energy of a continuum is

In this problem the first and third integrals are zero; and for a Newtonian fluid by (7.18),

But incompressibility means vili = Dii = O and so

7.34. Show that for a perfect fluid with negligible body forces the rate of change of cir-
culation bc may be given by - ~ ~ ~ j~p , ( a ís i/. ~ ) ,

From (1.43). bc = $ Vi d z , + $ v i dvi; and since d d ( f v 2 ) = O, the second integral i s zero.

From (7.36) with bi = 0, Gi = - ~ , ~ l pand so now

bc = - 6 IP) =
( P . ~ ~d.. - i eijk(p,k/p),ini d s

where (7.42) has been used in converting to the surface integral. Differentiating a s indicated,

h = -L +
e i j k [ ( l l ~ ) , j p , k P , L ~ P @i
] = - L eijk(lI~).j
~ , dSi
k

Supplementary Problems
7.35. The constitutive equation for an isotropic fluid is given by aij = -paij +
K i j p q D p q with Kihq con-
stants independent of the coordinates. Show that the principal axes of stress and rate of deformation
coincide.

7.36. Show that ( l l p ) ( d p l d t =


) O i s a condition for ,-uii/3 = p for a Newtonian fluid.

7.37. Show that the constitutive relations for a Newtonian fluid with zero bulk viscosity may be expressed
by the pair of equations sij = 2r*Dii and -oii = 3p.

7.38. Show that in terms of the vorticity vector q the Navier-Stokes equations may be written
\I = b - V p l p - "*V x q where v* = r*lP is the ,kinematic viscosity. Show that for irrotational
motion this equation reduces to (7.36).

7.39. If a fluid moves radially with the velocity v = v(r, t) where r2 = xixi, show that the equation of
continuity is A
a a + 2 (r%) = 0.
+v 2
at ar r2 ar

7.40. A liquid rotates a s a rigid body with constant angular velocity o about the vertical s3 axis. If
gravity is the only body force, show that plp - o2r212 + gz3 = constant.
174 FLUIDS [CHAP. 7

7.41. For an ideal gas under isothermal conditions (constant temperature = To),
show that p l p , = pIpo =
e-(glRTozs) where po and po are the density and pressure a t x3 = 0.

7.42. Show that if body forces are conservative so that bi = -nJi, the Navier-Stokes-Duhem equations
for the irrotational motion of a barotropic fluid may be integrated to yield - p ( a @ l a t +-
(V@)V2) +
+ +
pQ P +
(i* 2@*)V2+= f ( t ) . (See Problem 7.15.)

7.43. Show that the velocity and vorticity for an inviscid flow having conservative body forces and
constant density satisfy the relation Gi - qjvi,j = O. For steady flow of the same fluid, show that
-
vjqi,j - qjvi,j-

7.44. For a barotropic fluid having p = p(p) and P ( p ) defined by (7..29),show that grad P = grad plp.

7.45. Show that the Bernoulli equation (7.39) for steady motion of an ideal gas takes the form
+ +
(a)Q p ln ( p l p ) v2/2 = constant, for isothermal flow, ( b ) + +
(klk - l ) ( p l p ) v212 = constant,
for isentropic flow.

7.46. Show that the velocity field vl = -2x1x2x3/T1, v 2 = (x: - x;)x3/r4, v3 = x2/r2 where r2 = x: +
x;+ $32is a possible flow for an incompressible fluid. 1s the motion irrotational? Ans. Yes

+ + i$ +
7.47. If the velocity potential +(z) =
ix2 = rei0 show that in polar coordinates 3ar
= '9 and --
r ae
1 a6 =
r ae
-*
is an analytic function of the complex variable z = x ,

ar '

7.48. If body forces are zero, show that for irrotational potential flow +,ii = v* = , ~ * l p is the kinematic
viscosity.
Chapter 8

8.1 BASIC CONCEPTS AND DEFINITIONS


Elastic deformations, which were considered in Chapter 6, are characterized by com-
plete recovery to the undeformed configuration upon remova1 of the applied loads. Also,
elastic deformations depend solely upon the stress magnitude and not upon the straining
or loading history. Any deformational response of a continuum to applied loads, or to
environmental changes, that does not obey the constitutive laws of classical elasticity may
be spoken of as an inelastic deformation. In particular, irreversible deformations which
result from the mechanism of slip, or from dislocations a t the atomic level, and which
thereby lead to permanent dimensional changes are known as plastic deformations. Such
deformations occur only a t stress intensities above a certain threshold value known as the
elastic limit, or yield stress, which is denoted here by o,.
In the theory of plasticity, the primary concerns are with the mathematical formulation
of stress-strain relationships suitable for the phenomenological description of plastic defor-
mations, and with the establishment of appropriate yield criteria for predicting the onset
of plastic behavior. By contrast, the study of plastic deformation from the microscopic
point of view resides in the realm of solid state physics.
The phrase plastic flow is used extensively in plasticity to designate an on-going plastic
deformation. However, unlike a fluid flow, such a continuing plastic flow may be related
to the amount of deformation as well as the rate of deformation. Indeed, a solid in the
"plastic" state can sustain shear stresses even when a t rest.
Many of the basic concepts of plas-
ticity may be introduced in an elementary A O B
way by consideration of the stress-strain
diagram for a simple one-dimensional ten-
sion (or compression) test of some hypo-
thetical material as shown by Fig. 8-1.
In this plot, u is the nominal stress (force1
original area), whereas the strain r may ou
--
represent either the conventional (engi-
neering) strain defined here by
e = (L- Lo)/Lo (8.1) -e

where L is the current specimen length


and LOthe original length, or the natural
(logarithmic) strain defined by Fig. 8-1

For small strains, these two measures of strain are very nearly equal as seen by (8.2) and it
is often permissible to neglect the difference.
176 PLASTICITY [CHAP. 8

The yield point P, corresponding to the yield stress o, separates the stress-strain curve
of Fig. 8-1 into an elastic range and a plastic range. Unfortunately, the yield point is not
always well-defined. It is sometimes taken a t the proportional limit, which lies a t the
upper end of the linear portion of the curve. It may also be chosen as the point J, known as
Johnson's apparent elastic limit, and defined as that point where the slope of the curve
attains 50% of its initial value. Various offset methods are also used to define the yield
point, one such being the stress value a t 0.2 per cent permanent strain.
In the initial elastic range, which may be linear or nonlinear, an increase in load causes
the stress-strain-state-point to move upward along the curve, and a decrease in load, or
unloading causes the point to move downward along the same path. Thus a one-to-one
stress-strain relationship exists in the elastic range.
In the plastic range, however, unloading from a point such a s B in Fig. 8-1 results in the
state point following the path BC which is essentially parallel with the linear elastic portion
of the curve. At C, where the stress reaches zero, the permanent plastic strain cP remains.
The recoverable elastic strain from B is labeled rE in Fig. 8-1. A reloading from C back to
B would follow very closely the path BC but with a rounding a t B, and with a small
hysteresis loop resulting from the energy loss in the unloading-reloading cycle. Upon a
return to B a load increase is required to cause further deformation, a condition referred
to as work hardening, or strain hardening. It is clear therefore that in the plastic range
the stress depends upon the entire loading, or strain history of the material.
Although it is recognized that temperature will have a definite influence upon the plastic
behavior of a real material, it is customary in much of plasticity to assume isothermal con-
ditions and consider temperature as a parameter. Likewise, it is common practice in
traditional plasticity to neglect any effect that rate of loading would have upon the stress-
strain curve. Accordingly, plastic deformations are assumed to be time-independent and
separate from such phenomena as creep and relaxation.

8 2 IDEALIZED PLASTIC BEHAVIOR


Much of the three-dimensional theory for analyzing plastic behavior may be looked
upon as a generalization of certain idealizations of the one-dimensional stress-strain curve
of Fig. 8-1. The four most commonly used of these idealized stress-strain diagrams are
shown in Fig. 8-2 below, along with a simple mechanical model of each. In the models
the displacement of the mass depicts the plastic deformation, and the force F plays the role
of stress.
In Fig. 8-2(a), elastic response and work-hardening are missing entirely, whereas in
(b), elastic response prior to yield is included but work-hardening is not. In the absence
of work-hardening the plastic response is termed perfectly plastic. Representations (a) and
(b) are especially useful in studying contained plastic deformation, where large deformations
are prohibited. In Fig. 8-2(c), elastic response is omitted and the work-hardening is assumed
to be linear. This representation, as well a s (a), has been used extensively in analyzing
uncontained plastic flow.
The stress-strai,n curves of Fig. 8-2 appear in the context of tension curves. The
compression curve for a previously unworked specimen (no history of plastic deformation)
is taken a s the reflection with respect to the origin of the tension curve. However, if a
stress reversal (tension to compression, or vice versa) is carried out with a real material
that has been work-hardened, a definite lowering of the yield stress is observed in the
second type of loading. This phenomenon is known as the Bauschinger effect, and will be
neglected in this book.
CHAP. 81

E
PLASTICITY

(a) Rigid- Perfectly Plastic

(b) Elastic - Perfectly Plastic


- Rough

Rough
L

F
177

I
......................................
Rough
(c) Rigid-Linear Work Hardening

Rough
(d) Elastic - Linear Work Hardening
Fig. 8-2

8.3 YIELD CONDITIONS. TRESCA AND VON MISES CRITERIA


A ?jield condition is essentially a generalization to a three-dimensional state of stress of
the yield stress concept in one dimensional loading. Briefly, the yield condition is a
mathematical relationship among the stress components a t a point that must be satisfied
f o r the onset of plastic behavior a t the point. I n general, the yield condition may be
expressed by the equation
f(.,,) = c, (8.3')
where C, is known a s the yield constant, or a s is sometimes done by the equation

in which f,(aij) is called the yield function.


178 PLASTICITY [CHAP. 8

For an isotropic material the yield condition must be independent of direction and may
therefore be expressed as a function of the stress invariants, or alternatively, as a sym-
metric function of the principal stresses. Thus (8.3) may appear as

Furthermore, experiment indicates that yielding is unaffected by moderate hydrostatic stress


so that it is possible to present the yield condition as a function of the stress deviator
invariants in the form
f3(IIxD,1111~)= o (8.6)
Of the numerous yield conditions which have been proposed, two are reasonably simple
mathematically and yet accurate enough to be highly useful for the initial yield of isotropic
materials. These are:

(1) Tresca yield condition (Maximum Shear Theory)


This condition asserts that yielding occurs when the maximum shear stress reaches
the prescribed value C,. Mathematically, the condition is expressed in its simplest form
when given in terms of principal stresses. Thus for U, > a,, > a,,,, the Tresca yield
condition is given from (2.54b) as
+(aI- uIII) = Cy (a constant) (8.7)
To relate the yield constant C, to the yield stress in simple tension u,, the maximum
shear in simple tension a t yielding is observed (by the Mohr's circles of Fig. .8-3(a), for
example) to be ~ ~ 1 2Therefore
. when referred to the yield stress in simple tension,
Tresca's yield condition becomes
u~ - -'I'II = u~ (8.8)
The yield point for a state of stress that is so-called pure shear may also be used as a
referente stress in establishing the yield constant C,. Thus if the pure shear yield
point value is k, the yield constant C, equals k (again the Mohr's circles clearly show
this result, as in Fig. 8-3(b), and the Tresca yield criterion is written in the form

(a) Simple Tension ( 6 ) Pure Shear

Fig. 8-3

(2) von Mises yield condition (Distortion Energy Theory)


This condition asserts that yielding occurs when the second deviator stress invariant
attains a specified value. Mathematically, the von Mises yield condition states
CHAP. 81 PLASTICITY 179

which is usually written in terms of the principal stresses as


(aI - uII)2 + (gII - uIII)2+ (vIII- u1)2 = 6 c y (8.11)
With reference to the yield stress in simple tension, it is easily shown that (8.11) becomes
(oI - uIJ2 + (cII- aIII)2f (vIII- a1)2 = 2~; (8.12)
Also, with respect to the pure shear yield value k, von Mises condition (8.11) appears
in the form
(aI - uIJ2 + (rII
- uIII)2f bIII
- uI)2 = 6k2 (8.13)
There are severa1 variations for presenting (8.12) and (8.13) when stress components
other than the principal stresses are employed.

8.4 STRESS SPACE. THE II-PLANE. YIELD SURFACE


A stress space is established by using
stress magnitude as the measure of dis-
tance along the coordinate axes. In the
Haigh-Westergaard stress space of Fig.
8-4 the coordinate axes are associated
with the principal stresses. Every point
in this space corresponds to a state of
stress, and the position vector of any such
point P(u,, oII,uIII) may be resolved into a
component OA along the line 0 2 , which
makes equal angles with the coordinate
axes, and a component OB in the plane
(known as the n-plane) which is perpendic-
ular to OZ and passes through the origin.
The component along OZ, for which U, =
a,, = uIII, represents hydrostatic stress, so
that the component in the n-plane repre-
sents the deviator portion of the stress
state. I t is easily shown that the equation
of the n-plane is given by
a, + ali + a,,, = 0 (8.14) Fig. 8-4

In stress space, the yield condition (8.5), f2(a,, u,,,olII)= C,, defines a surface, the
so-called yield surface. Since the yield conditions are independent of hydrostatic stress,
such yield surfaces are general cylinders having their generators parallel to 02. Stress
points that lie inside the cylindrical yield surface represent elastic stress states, those
which lie on the yield surface represent incipient plastic stress states. The intersection of
the yield surface with the R-plane is called the yield curve.
In a true view of the n-plane, looking along OZ toward the origin 0, the principal stress
axes appear symmetrically placed 120" apart as shown in Fig. 8-5(a) below. The yield
curves for the Tresca and von Mises yield conditions appear in the n-plane a s shown in
Fig. 8-5!b) and (c) below. In Fig. 8-5(b), these curves are drawn with reference to (8.7) and
(8.11), using the yield stress in simple tension as the basis. For this situation, the von Mises
circle of radius m 3U, is seen to circumscribe the regular Tresca hexagon. In Fig. 8-5(c),
the two yield curves are based upon the yield stress k in pure shear. Here the von Mises
circle is inscribed in the Tresca hexagon.
180 PLASTICITY [CHAP. 8

(b)
Fig. 8-5

The location in the II-plane of the projection of an arbitrary stress point P(u,, a,,, o,,,)
is straightforward since each of the stress space axes makes cos-I @ with the II-plane.
Thus the projected deviatoric components are ( m v I , d2/3uII, moIII). The inverse
problem of determining the stress components for an arbitrary point in the n-plane is not
unique since the hydrostatic stress component may have any value.

8.5 POST-YIELD BEHAVIOR. ISOTROPIC AND KINEMATIC HARDENING


Continued loading after initial yield is reached leads to plastic deformation which may
be accompanied by changes in the yield surface. For an assumed perfectly plastic material
the yield surface does not change during plastic deformation and the initial yield condition
remains valid. This corresponds to the one-dimensional perfectly plastic case depicted by
Fig. 8-2(a). For a strain hardening material, however, plastic deformation is generally
accompanied by changes in the yield surface. To account for such changes it is necessary
that the yield function f,(oij) of (8.4) be generalized to define subsequent yield surfaces
beyond the initial one. A generalization is effected by introduction of the loading function

which depende not only upon the stresses, but also upon the plastic strains r; and the work-
hardening characteristics represented by the parameter K. Equation (8.15) defines a loading
surface in the sense that f: = O is the yield surface, f: < 0 is a surface in the (elastic)
region inside the yield surface and f r > 0, being outside the yield surface, has no meaning.
Differentiating (8.15) by the chain rule of calculus,

Thus with f: = O and (af :Iaoij) doij < 0, unloading is said to occur; with f: = O and
(af :/aoij) doij = 0, neutra1 loading occurs; and with f i = O and (af:la~,,) doij > 0, loading
occurs. The manner in which the plastic strains c; enter into the function (8.15) when
loading occurs is defined by the hardening rules, two especially simple cases of which are
described in what follows.
The assumption of kotropic hardening under loading conditions postulates that the yield
surface simply increases in size and maintains its original shape. Thus in the II-plane the
yield curves for von Mises and Tresca conditions are the concentric circles and regular
hexagons shown in Fig. 8-6 below.
CHAP. 81 PLASTICITY 181

(a) Mises Circles ( b ) Tresca Hexagons


Fig. 8-6

In kinematic hardening, the initial yield surface is translated to a new location in stress
space without change in size or shape. Thus (8.4) defining an initial yield surface is
replaced by
f i (oij- a,) = O (8.17)
where the aij are coordinates of the center of the new
yield surface. If linear hardening is assumed,
Xi. = c;: (8.18)

where c is a constant. In a one-dimensional case,


the Tresca yield curve would be translated as shown
in Fig. 8-7. Fig. 8-7

8.6 PLASTIC STRESS-STRAIN EQUATIONS. PLASTIC POTENTIAL THEORY


Once plastic deformation is initiated, the constitutive equations of elasticity are no
longer valid. Because plastic strains depend upon the entire loading history of the material,
plastic stress-strain relations very often are given in terms of strain increments - the
so-called incremental theories. By neglecting the elastic portion and by assuming that the
principal axes of strain increment coincide with the principal stress axes, the Levy-Mises
equations relate the total strain increments to the deviatoric stress components through
the equations
dcij = sijdA
Here the proportionality factor dX appears in differential form to emphasize that incre-
mental strains are being related to finite stress components. The factor dA may change
during loading and is therefore a scalar multiplier and not a fixed constant. Equations
(8.19) represent the flow rule for a rigid-perfectly plastic material.
If the strain increment is split into elastic and plastic portions according to

and the plastic strain increments related to the stress deviator components by

the resulting equations are known as the Prandtl-Reuss equations. Equations (8.21) rep-
resent the flow rule for an elastic-perfectly plastic material. They provide a relationship
between the plastic strain increments and the current stress deviators but do not specify
the strain increment magnitudes.
182 PLASTICITY [CHAP. 8

The name plastic potential function is given to that function of the stress components
g(u,) for which dg dA
de; = - (8.22)
8%
For a so-called stable plastic material such a function exists and is identical to the yield
function. Moreover when the yield function f l ( u i j )= IIr,, (8.22) produces the Prandtl-
Reuss equations (8.21).

8.7 EQUIVALENT STRESS. EQUIVALENT PLASTIC STRAIN INCREMENT


With regard to the mathematical formulation of strain hardening rules, i t is useful
to define the equivalent or efSective stress a,, as

This expression may be written in compact form as

In a similar fashion, the equivalent or effectiveplastic strain increment de:, is defined by

which may be written compactly in the form

In terms of the equivalent stress and strain increments defined by (8.24) and (8.25)
respectively, dA of (8.21) becomes
dh = -- 3 de;,
2 OEQ

8.8 PLASTIC WORK. STRAIN-HARDENING HYPOTHESES


The rate a t which the stresses do work, or the stress power as i t is called, has been given
in (5.32) as uijDijper unit volume. From (4.25), den = D,dt, so that the work increment
per unit volume may be written
dW = ali dcij (8.2 8)
and using (8.20) this may be split into

For a plastically incompressible material, the plastic work increment becomes

Furthermore, if the same material obeys the Prandtl-Reuss equations (8.21),the plastic work
increment may be expressed as
d W P = a,,dcEQ (8.31)
and (8.21) rewritten in the form
3 dWP
de; = -- S..
2 a;,
CHAP. 81 PLASTICITY 183

There are two widely considered hypotheses proposed for computing the current yield
stress under isotropic strain hardening plastic flow. One, known as the work-hardening
hypothesis, assumes that the current yield surface depends only upon the total plastic work
done. Thus with the total plastic work given as the integral

the yield criterion may be expressed symbolically by the equation

for which the precise functional form must be determined experimentally. A second harden-
ing hypothesis, known a s the strain-hardening hypothesis, assumes that the hardening is a
function of the amount of plastic strain. In terms of the total equivalent strain

this hardening rule is expressed symbolically by the equation


f , (aij) = (8.36)
for which the functional form is determined from a uniaxial stress-strain test of the
material. For the Mises yield criterion, the hardening rules (8.34) and (8.36) may be shown
to be equivalent.

8.9 TOTAL DEFORMATION THEORY


In contrast to the incrernental theory of plastic strain as embodied in the stress-strain
increment equations (8.19) and (8.21), the so-called total deformation theory of Hencky
relates stress and total strain. The equations take the form

I n terms of equivalent stress and strain, the parameter + may be expressed as


(P = 3
--
2 um
where here L& = d Z 5 7 3 so that
€P = 3 cEe
--
23 2 UEQ

8.10 ELASTOPLASTIC PROBLEMS


Situations in which both elastic and plastic strains of approximately the same order
exist in a body under load are usually referred to as elasto-plastic problems. A number of
well-known examples of such problems occur in beam theory, torsion o£ shafts and thick-
walled tubes and spheres subjected to pressure. In general, the governing equations for
the elastic region, the plastic region and the elastic-plastic interface are these:
(a) Elastic region
1. Equilibrium equations (2.23), page 49
2 . Stress-strain relations (6.23) or (6.24), page 143
3. Boundary conditions on stress or displacement
4. Compatibility conditions
184 PLASTICITY [CHAP. 8

( b ) Plastic region
1. Equilibrium equations (2.23), page 49
2. Stress-strain increment relations (8.21)
3. Yield condition (8.8) or (8.11)
4 . Boundary conditions on plastic boundary when such exists
(c) Elastic-plastic interf ace
1. Continuity conditions on stress and displacement

8.11 ELEMENTARY SLIP LINE THEORY FOR PLANE PLASTIC STRAIN


In unrestricted plastic flow such as occurs in metal-forming processes, i t is often pos-
sible to neglect elastic strains and consider the material to be rigid-perfectly plastic. If
the flow may be further assumed to be a case of plane strain, the resulting velocity field
may be studied using slip line theory.
Taking the X l X 2 plane as the plane of flow, the stress tensor is given in the form

and since elastic strains are neglected, the plastic strain-rate tensor applicable to the
situation is
. .
'li '12 0
(8.42)

In (8.41) and (8.42) the variables are functions of XI and xz only, and also

where v,are the velocity components.


For the assumed plane strain condition, de,, = 0; and so from the Prandtl-Reuss equa-
tions (8.21), the stress V , , is given by
u33 = +(all+ nz2) (8.44)

Adopting the standard slip-line notation o,, = -p, and ~((r,, -~ , , ) ~ +


/ 4( u , , ) ~= k, the
principal stress values of (8.41) are found to be

The principal stress directions are given


with respect to the xlx2 axes as shown in
Fig. 8-8, where tan 28 = 2ul2/(al1- u,,).
As was shown in Section 2.11, the maxi-
mum shear directions are a t 45" with respect
to the principal stress directions. In Fig.
8-8, the maximum shear directions are des- 21
ignated as the a and p directions. From the Fig. 8-8
CHAP. 81 PLASTICITY 185

geometry of this diagram, O = r14 ++ so that


1
tan2+ =
.- -
tan28 (8.46)
and for a given stress field in a plastic flow, two families of curves along the directions of
maximum shear a t every point may be established. These curves are called shear lines, or
slip lines.
For a small curvilinear element bounded by the two pairs of slip lines shown in Fig. 8-9,
u,, = -p - k sin2+
qZz = -p + k sin 2+
U12 = FC COS2+
and from the equilibrium equations i t may be shown that
p + 2k+ = C, a constant along a n a line
p - 2k+ = C, a constant along a ,8 line

Fig. 8-9 Fig. 8-10

With respect to the velocity components, Fig. 8-10 shows that relative to the a and
/i' lines,
v, = v, cos + - v, sin +
(8.49)
vz = v, sin + + v, cos +
For a n isotropic material, the principal axes of stress and plastic strain-rate coincide.
Therefore if x, and x, are slip-line directions, ;,, and i,, are zero along the slip-lines so t h a t

(& (v, cos + - v, sin +)S*=. = o

These equations lead to the relationships


dv, - v, d+ = O on a lines (8.52)
dv, + v, d+ = O on /i' lines (8.53)
Finally, for statically determinate problems, the slip line field may be found from (8.48),
and using this slip line field, the velocity field may be determined from (8.52)and (8.53).
CHAP. 81 PLASTICITY 187

8.5. Convert the von Mises yield condition (8.10) to its principal stress form a s given in
(8.11).
From (2.72), - 1 1 ~=~-(slsI1 + SIISIII + sII1s1); and by (2.71), SI = o1 - UM, etc., where
UM + +
= (01 oI1 uIII)/3. Hence

Thus (01 - QII )2 + (011 - 0111)~+ (oIII - = 6Cy

8.6. With the rectangular coordinate system OXYZ oriented so that the XY plane coincides
with the n-plane and the a,,, axis lies in the YOZ plane (see Fig. 8-13 and 8-4), show
that the Mises yield surface intersects the n-plane in the Mises circle of Fig. 8-5(b).

Fig. 8-13
The table of transformation coefficients between the two sets of axes is readily determined to
be a s shown above. Therefore
,I = -x/fi - Y/& + z/fi, o,, = x/fi - Y/& + z / h , QIII = 2 ~ 1+
6 216

and (8.12) becomes


(-fix)z + ( x / h - 3 ~ / f i )+ (x/fi+ 3~/1&)2 =
2 20.;

which simplifies to the Mises yield circle 3X2 + 3Y2 = 20; of Fig. 8-5(b).

8.7. Using the transformation equations of Problem 8.6, show that (8.14), a, + a,, + u,,, = 0,
is the equation of the n-plane.
Substituting into (8.14) the o's of Problem 8.6, a1 + o,, + aII1 = fiZ = O, or Z = O which
is the XY plane @-plane).

8.8. For a biaxial state of stress with a,, = O,


determine the yield loci for the Mises and
Tresca conditions and compare them by a
plot in the two-dimensional u,/u, vs. a I I I / ~ ,
space.
From (8.12) with o,, = 0, the Mises yield condi-
tion becomes
0: - o,o,11 +
o;,, = o;
which is the ellipse
= 1
-) (~Q I U I I I If~ )(UIII/"Y)~
(~I/UY Fig. 8-14
188 PLASTICITY [CHAP. 8

with axes a t 45' in the plot. Likewise, from (8.8) and the companion equations a111 - a,, = <ry,
uI1 - u1 = o,, the Tresca yield condition results in the line segments AB and ED with equations
( o I / o y )- ( o I I I / o y )= 21, DC and F A with equations oI,,/uy = f1, and BC and EF with equations
o I / o y = T I , respectively.

8.9. The von Mises yield condition is referred to in Section 8.3 as the Distortion Energy
Theory. Show that if the distortion energy per unit volume u,dD,is set equal to the
yield constant C, the result is the Mises criterion as given by (8.12).
From Problem 6.26, uTD)is given in terms of the principal stresses by

and for a uniaxial yield situation where o, = o y , oI1 = o I I I = O , uTD)= ~ $ 1 6 ~Thus


. C , = a$/6G
and, as before, the Mises yield condition is expressed by (8.12).

PLASTIC DEFORMATION. STRAIN-HARDENING (Sec. 8.4-8.8)


8.10. Show that the Prandtl-Reuss equations (8.21) imply that principal axes of plastic
strain increments coincide with principal stress axes and express the equations in
terms of the principal stresses.
From the form of (8.21), when referred to a coordinate system in which the shear stresses are
zero, the plastic shear strain increments are seen to be zero also. In the principal axes system,
P P P P
(8.21) becomes de;lsI = d e I I l s I I= d e I I I / ~ I=I I dX. Thus deI = (oI - aM)dX, deII = ( o I I- o M ) d h , etc.,
and by subtracting,
P
de: - de: - de: - dcFII - de:[ - deI
- - = dX
" I - "11 "11 - "111 "111 - " I

8.11. For the case of plastic plane strain with r,, = O, de,, = O and u,, = 0, show that the
Levy-Mises equations (8.19) lead to the conclasion that the Tresca and Mises yield
conditions (when related to pure shear yield stress k) are identical.
Here (8.19) becomes dell = (2õ11- u3,) dh/3, dez2 = -(oll + dh/3, O = 203, - 011. Thus in the
absence of shear stresses, C , = u l l , ull = 03, = u I 1 / 2 , ( r I I 1 = O = az2. Then from (8.9) the Tresca
yield condition is o, - a,,, = o,, = 2k. Also, from (8.19) for this case, Mises condition becomes
+
(ul1/2)2 (-ul1/2)2 +
(-ul1)2 = 6k2 or oS1 = 4k2 and oll = 2k.

8.12. Show that the Prandtl-Reuss equations imply equality of the Lode variable p (see
Problem 8.3)and v = ( 2 d 4 - dc; - d r g I ) l ( d r ; - d&).
From equations (8.21),

8.13. Writing 111, u ~sij.


= sijsij/2, show that ~ I I z , / ~ = ~
Here a I I I D /aopq = ( a ~ , ~ l a ~ ~where
,)~,~ = a(oij - 6 i i ~ k k / 3 ) / a ~=p qOipSjq - ~ ~ ~ Thus
~ ~ ~
a I I x D / a ~ p=q ( S i p S j q - SijSpq/3)sij= spq since sii = I I D = 0.
CHAP. 81 PLASTICITY 189

8.14. Show that when the plastic potential function g(aii) = IIr,, the plastic potential equa-
tions (8.22) become the Prandtl-Reuss equations.
The proof follows directly from the result of Problem 8.13, since aglaa, = sij in this case and
(8.22) reduce to (8.21).

8.15. Expand (8.24) to show that the equivalent stress a,, may be written in the form
of (8.23).
From equation (8.24),
o g Q = 3siisij/2 = 3(aij- Sijupp/3)(aij
- Sijaqq/3)/2 = (3aiiuij - uiioii)/2
Expanding this gives

[3(all + ai2 + 033 + 6(aS2 + 0223 + ail ) - (<rll + a22 + ~ ~ ~ ) 2 ] / 2


- [2(uli 4 2 + o33 - a11cz2 - 0 2 2 ~ 3 3- "33ul1) 6(a:2 + ai3 + )]/2
-
- [(ali- ~ + (o22
2 2 ) ~ - 033)' f (a33 - ~ +
1 1 ) ~6(oS2 + c i 3 + uil ) ] I 2
which confirms (8.23).

8.16. In plastic potential theory the plastic strain increment vector is normal to the loading
(yield) surface a t a regular point. If [N,,N,, N3] are direction numbers of the normal
to the yield surface f , ( a i j ) , show that d r p l s , = d~;ls,, = d ~ ~ under
~ l the
s Mises
~ ~ yield
~
condition and flow law.
The condition of normality is expressed by N = grad f l which requires N l l ( a f l l a ~ I )=
+ +
N 2 / ( d f l / d o I I )= N31(afllauIII) for the Mises case where f l = (aI - aII)2 ( o I I- o I I I ) 2 ( a I I I- o J 2 -
2a; = O. Here af1/aaI = 2(2u1- a,, - a I I I ) = 6s,, etc., and since the plastic strain increment vector
is along the normal i t follows that deF/sI = d ePI I / s I I= d e g I / s I I I .

8.17. Determine the plastic strain increment ratios for (a) simple tension with a,, = a,,
( b ) biaxial stress with V,, = - ~ , / f i ,uZ2= a Y / f i , u~~= a12 = u~~= u13= O, (C) pure
shear with V,, = ~ , / f l .
(a) Here all = a1 = a,, aI1 = ' a I I I= O and s1 = 2ay/3, s I I = s I I I = -oy/3. Thus from Problem
8.16, depl2 = - d e g l l = - d e E I / l .

(b) Here o1 = a y / f i , a I 1 = O , a I I I = - a y / f i and S I = o y / d , S I I = 0, S I I I = - u y / d . Thus


d e F / l = - d e g I / l and the third term is omitted since i t is usually understood in the theory that
if the denominator is zero the numerator will be zero too.

8.18. Determine the plastic. work increment dWP and the equivalent plastic strain incre-
ment de;, for the biaxial stress state a,, = - ~ ~ / fu~~ i =, uy/\/3, = = = (rl2

a,, = O if plastic deformation is controlled so that de: = C, a constant.

In principal-axis form, (8.30) becomes d W P = o, de; aI1de: + + a,,, de;,; and for the stress
state given, Problem 8.17 shows that de; = -deEI, de: = 0 ; hence
dWP = -u,c/\/~ (ay/fi)(-c) = -2cay/d
From (8.25),
= {2[(dep- deg)2 + (de: - d e g I ) 2+ - dep)21}112/3
190 PLASTICITY [CHAP. 8

8.19. Verify (8.32) by showing that for a Prandtl-Reuss material the plastic work increment
is dWP = o,, d ~ ; , as given in (8.31).
From (8.30), d W P = sijsijdh for a Prandtl-Reuss material satisfying (8.21). But from (8.27),
P
dh = 3 deEQ/2UEQ for such a material and so d W P = ( 3 ~ ~ ~ 8 ~ ~ / 2 ) ( dwhich
e ; ~ / ubecause
~ ~ ) of the
&EQ.
definition (8.24) gives dWP = U E Q Thus de:, = dWP/uEQ here and ( 8 3 2 ) follows directly
from (8.21).

8.20. For a material obeying the Mises yield condition, the equivalent stress a,, may be
taken as the yield function in the hardening rules (8.36) and (8.36). Show that in this
case uEQFr= H' where Fr and H r are the derivatives of the hardening functions with
respect to their respective arguments.
Here (8.34) becomes U E Q = F ( W P ) and so doEQ= F' dWP. Likewise (8.36) is given here by
OEQ = H(€;,) and so daEQ= H' de;,. Thus F' dWP = H' de;Q; and since from (8.31) (or Problem
8.19) dWP = U E Q i t follows a t once that uEQF1= H'.

TOTAL DEFORMATION THEORY (Sec. 8.9)


8.21. The Hencky total deformation theory may be represented through the equations
t j = C; + C;
C. with C ; = e; +
6ij~kk/= +
3 (sij/2)G 6,,(1 - 2 v ) o k k / 3 E and C ; = +sij.
Show that these equations are equivalent to (8.37) and (8.38).
The equation e; = QSij implies e: = O so that €5
= e; = +vij, and from eij = E: + €5
i t fol-
lows that here yj = .e: +
From the same equation, eij Sijekk/3 = e: + +
sijefk/3 e; which reduces
to eij = e: + +
e: = ( Q * ~ ) s (8.37).
~ ~ , ~ i s ofrom eii = e:, eii = ( 1 - 2v)Ukk/E, (8.38).

8.22. Verify that the Hencky parameter 4 may be expressed as given in (8.39).
Squaring and adding the components in the equation e: = @sij of Problem 8.21 gives e; €5 =
+2sijsij or Q = 4-luEQ which when multiplied on each side by 213 becomes

ELASTOPLASTIC PROBLEMS (Sec. 8.10)


8.23. An elastic-perfectly plastic rectangular beam is loaded steadily in pure bending.
Using simple beam theory, determine the end moments M for which the remaining
elastic core extends from -a to a as shown in Fig. 8-15.

Fig. 8-15

Here the only nonzero stress is the bending stress ull. In the elastic portion of the beam
(-a < x , < a ) , ull = Ecll = E x d R where R is the radius of curvature and E is Young's modulus.
In the plastic portion, ull = uy. Thus

where uy = E a I R , the stress co?dition a t the elastic-plastic interface, has been used. From the
result obtained, M = 2bc2uy/3 a t first yield (when a = c ) , and M = bczu, for the fully-plastic
beam (when a = 0).
CHAP. 81 PLASTICITY 191

8.24. Determine the moment for a beam loaded as in Problem 8.23 if the material is a
+
piecewise linear hardening material for which a , , = a , A(€,,- u Y / E ) after yield.
The stress distribution for this beam is shown in Fig. 8-16.
Again c l l = x2/R and so

or using oy = E a / R a s in Problem 8.23, 12 3


M e2boy(l- AIE) + 2c3bA/3R + ~ u ; R ~ ( A-/ E1)/3E2 Fig. 8-16

8.25. An elastic-perfectly plastic circular shaft of radius


c is twisted by end torques T as shown in Fig. 8-17.
Determine the torque for which an inner elastic core
of radius a remains.

O 6 r 6 a, and by o,, -
The shear stress u12 is given here by alz = krla for
k for a 5 r 6 e where k is the
yield stress of the material in shear. Thus
1

(k+/a) dr + 2a
S.'kr2 dr = ?qc3-
3
a3 / 4 )

Therefore the torque a t first yield is T 1 = a k ~ 3 / 2when a = c;


and for the fully plastic condition, T , = 2ake3/3 = 4T1/3
when a = 0. Fig. 8-17

A thick spherical shell of the dimensions shown in


Fig. 8-18 is subjected to an increasing pressure po.
Using the Mises yield condition, determine the pres-
sure a t which first yield occurs.
Because of symmetry of loading the principal stresses
a r e the spherical components o(%%) = o1 = U I I , o(,r) = 0111.
Thus the Mises yield condition (8.12) becomes o(%%) -o(,,) = UY.
The elastic stress components may be slíown to be

Therefore ay = 3b3p0/2r3(b3/a3- 1) and po = 2ay(l - a3/b3)/3


a t first yield which occurs a t the inner radius a. Fig. 8-18

SLIP LINE THEORY (Sec. 8 . 1 1 )


8.27. Verify directly the principal stress values (8.45) for the stress tensor (8.41) with
+
= ( a l i a , , ) / 2 as given in ( 8 . 4 4 ) .
The principal stress values a r e found from the determinant equation (2.37) which here becomes

Expanding by the third column,


192 PLASTICITY [CHAP. 8

The roots of this equation a r e clearly o = -p +


and o = &(o11 oZ2)2 d i ( o l l +~ +
2 2 ) ~ = -P k.

8.28. Using the condition that the yield stress in shear k is constant, combine (8.47) with
the equilibrium equations and integrate to prove (8.48).
From the equilibrium equations aall/dxl + ao121axz = O and do12/ax1 + a a 2 2 / a ~ 2= O which
a r e valid here, (8.47) yields
-aplaxl - k ( 2 cos 2g)(aglax1)+ k(-2 sin 2g)(agldx,) = O
and k ( 2 cos 2g)(a+lax2) = O
-k(2 sin 2g)(ag/axl)- aplax, -I

If x 1 is along a n a line and x , along a ,L3 line, @ = O and these equations become -dplax, -
2k(dglaxl) = O along the a line, -aplaxz +
2k(agldx2) = O along the ,L3 line. Integrating directly,
p + 2kg = C1 on the a line, p - 2kg = C z on the line.

8.29. I n the frictionless extrusion through a square die causing a fifty per cent reduction,
the centered f a n region is composed of straight radial ,8 lines and circular a lines a s
shown in Fig. 8-19. Determine the velocity components along these slip lines in
terms of the approach velocity U and the polar coordinates r and O.

Fig. 8-19

Along the straight ,L3 lines, d@= 0; and from (8.53), dv2 = O o r v 2 = constant. From the ,

normal velocity continuity along B C , the constant here must be U cose and so v2 = U cose. Along
the circular a lines, dg = de; and from (8.52),
0

= S-,,, U cos e de = U(sin e + l/fi)

MISCELLANEOUS PROBLEMS
8.30. Show that the vonSMisesyield condition may be expressed in terms of the octahedral
shear stress uoCt (see Problem 2.22) by ao,, = f i ~ , / 3 .
+ +
I n terms of principal stresses 3o,,, = d ( o I - o I I ) 2 ( o I I- u I I I ) 2 ( o I I I- uI)2 (Problem 2.22)
+ +
and so 9oOCt = (oI - o I I ) 2 ( o I I- a I I I ) 2 ( o I I I- ar), = 2u$ i n agreement with (8.12).

8.31. Show that equation (8.13) for Mises yield condition may be written as
sS + s:, + s;,, = 2k2

From (2.71), o1 = SI + UM, etc., and so (8.13) a t once becomes


(SI - SII)' + (SII -~111)' + (SIII -~ 1 =
) ~ 6k2

Expanding and rearranging, this may be written s: sSI sSIr - ( s I + + + s I I + s I I I ) 2 / 3= 21~2. But
sI + +
sl, s,,, = IpD O and the required equation follows.
C H A P . 81 PLASTICITY 193

8.32. At what value of the Lode parameter p = ( 2 u 1 ,- U , - u , , , ) / ( u ,- u I I I ) are the Tresca


and Mises yield conditions identical?
+ +
From the definition of p, o,, = ( O , o I I I ) / 2 p(o, - o I I 1 ) / 2 which when substituted into the
Mises yield condition (8.12) gives after some algebra (see Problem 8.42) o, - U I I , = 2 o y / m .
Tresca's yield condition, equation (8.8), is o, - C,,, = uP Thus when /I = 1 the two are identical.
When a,, = o,, p = 1 which is sometimes called a cylindrical state of stress.

For the state of stress where u and T are constants, determine the

yield condition according to Tresca and von Mises criteria.


The principal stresses here are readily shown to b e a1 = o 7 , o,, = u, + = u - r. Thus
from (8.8), the Tresca condition V , - O,,, = UY gives 27 = uy. From (8.12) the Mises condition
gives 7 = cy/*. Note that in each case yielding depends on 7,not on c , i.e. yielding is independent
of hydrostatic stress.

8.34. Show that the Prandtl-Reuss equations imply incompressible plastic deformation and
write the equations in terms of actual stresses.
P
-
From (8.21), deii = sii dh = O since sii = IXD O and the incompressibility condition dez = O
+
is attained. In terrns of stresses, d e z = (oij- Siiukk/3)d ~ .~ h u sde:, = (2/3)[uli- (oz2 q 3 ) / 2 ]d ~ ,
etc., for the normal components and de:, = o,, dA, etc., for the shear components.

8.35. Using the von Mises yield condition, show that in the U-plane the deviator stress com-
ponents at yield are
S, = [ - 2 ~ ,cos (9 - ~ / 6 ) ] / 3 , s,
' +
= [2uY cos (9 ~/6)]/3, s,,, = ( 2 u y sin $)I3
where 9 = tan-I Y / X in the notation of Problem 8.6.
The radius of the Mises yield circle is &%ry so that by definition X = m o y cose,
Y = m o y sin e a t yield. From the transformation table given in Problem 8.6 together with
o, = s , +
U M , etc., the equations s , - s,, = - f i=~ - ( 2 / f i ) a y cose and S I -i - 2~111=
-6 + +
Y = -20y sin e are obtained. Also, in the n-plane, sI s I I s I I 1= O . Solving these three
equations simultaneously yields the desired expressions, a s the student should verify.

8.36. An elastic-perfectly plastic, incompressible


material is loaded in plane strain between
rigid pIates so that u,, = O and C,, = O (Fig.
8-20). Use Mises yield condition to determine
the loading stress CT,, a t first yield, and the
accompanying strain C , , .
The elastic stress-strain equation
Ec33 = o33 - v(oii+ 0 2 2 )
reduces here to o,, = vu,,. Thus the principal stresses
are o, = O, a,, = -vull, u I I 1= - c l 1 ; and by (8.12)
we have
+
(vo1,)2 ('Tll(l- v ) ) , +
( - u , ~ )=
~ 20;

from whieh a l i = - u v / d ~(cornpressive) a t


yield. Likewise, from Eell = o l l - u3R) we +
A,
see that here = -ay(l - V Z ) / E ~ - a t yield. Fig. 8-20
194 PLASTICITY [CHAP. 8

q-J
8.37. An elastic-perfectly plastic rec-
tangular beam is loaded in pure
bending until fully plastic. Deter-
mine the residual stress in the
beam upon removal of the bending
6 r l . ) _L
k b - 4
m0ment M. Fig. 8-21

For the fully plastic condition, the moment is (see Problem 8.23) M = bc20y. This moment
would cause a n elastic stress having a = Mel1 = 3oy/2 a t the extreme fibers, since I = 26~313.

x
Thus remova1 of M is equivalent to applying a corresponding negative elastic stress which results
in the residual stress shown in Fig. 8-22.

30~12 oy/2

fully
plastic
+ negative
elastic

Fig. 8-22
-

residual
stress

8.38. A thick-walled cylindrical tube of the dimensions shown in Fig. 8-23 is subjected to
an interna1 pressure p , . Determine the value of p, a t first yield if the ends of the tube
are closed. Assume (a) von Mises and ( b ) Tresca's yield conditions.

Fig. 8-23 Fig. 8-24

The cylindrical stress components (Fig. 8-24) are principal stresses and for the elastic analysis
+
) pi(b2/r2 l ) / Q , o(zL)= pilQ where Q = (b2Ia2 - 1).
may be shown to be o(,,) = -pi(b2/r2 - 1 ) / Q , o ( e o =

( a ) Here Mises yield condition is


+
(a(,,) - o ( 8 0 ) ) ~ (o(e,) - ~ +
( z z ) ) ~( O ( , , ) - = 20; or pf b41+ = Q20;/3
The maximum stress is a t r = a , and a t first yield pi = ( o y / f i ) ( l - a2/b2).

(b) For the Tresca yield condition, o(oe)- o(,,) = o y since o,, is the intermediate principal stress.
Thus 2pibz/r2 = Qoy and now a t r = a , pi = ( o y / 2 ) ( 1- a2/b2) a t first yield.
CHAP. 81 PLASTICITY

Supplementary Problems
A one-dimensional stress-strain law is given by o = Ken where K and n are constants and E is
true strain. Show that the maximum load occurs a t e = n.

Rework Problem 8.4 using the yield stress in shear k in place of u y in the Mises and Tresca yield
conditions. +
Ans. Mises: ( o l f i k)2 (rlk)2 = 1; Tresca: ( ~ / 2 k ) 2 ( ~ / k ) =
2 1 +
Making use of the material presented in Problem 8.6, verify the geometry of Fig. 8-5(c). .
From the definition of Lode's parameter p (see Problem 8.3) and the Mises yield condition, show
that o, - u,,, = 2 o y / ~ ~ .

In the 11-plane where s = tan-1 Y I X with X and Y defined in Problem 8.6, show that p = -6tan 0 .
Show that the invariants of the deviator stress IrID = sijsij/2 and IIIED= sijsjkski/3 may be written
= (s; + s;, + s:,,)/2 and III,, = (si + s;, + si,,)/3 respectively.

Show that von Mises yield condition may be written in the form
(011 - ~ 2 2 ) ~(o22 - ~ 3 3 4-) ~
(a33- 011)2 + 6(a:2 + ui3 + mil ) = 6k2

Following the procedure of Problem 8.17, determine the plastic strain increment ratios for
(a)biaxial tension with ul1 = o,, = uy, ( b ) tension-torsion with oll = oy/2, u12 = oY/2.
Ans. (a) deFl = dei, = -der3 12 ( b ) deTl /2 = -dei2 = -der3 = ddeL/3

Verify the following equivalent expressions for the effective plastic strain increment degQ and note
that in each case degQ = deFl for uniaxial tension ull.

( a ) deeQ + (de; + (de; + 2(dcr2 + 2(de& + 2(de&


[(derl)2 )2 )2 )2 )2 )2]1/2

( b ) de:, = (\/Sj3)[(der1- dei2)2 + (de; - ds[3)2 + (dei3 - deTl)2 + 6 ( d ~ f , +


)~ + 6(de&)2]112
A thin-walled elastic-perfectly plastic tube is loaded in combined tension-torsion. An axial stress
o = uy/2 is developed first and maintained constant while the shear stress T is steadily increased
from zero. A t what value of r will yielding first occur according to the Mises condition?
A n s . 7 = uy/2
I x9

-
The beam of triangular cross section shown in Fig. 8-25
is subjected to pure bending. Determine the location of
the neutra1 axis (a distance b from top) of the beam when
fully plastic. Ans. b = h l f i

Show that the stress tensor (8.41) becomes 2h ---4

Fig. 8-25

\ 0 o-P/
when referred to the axes rotated about x3 by an angle O
in Fig. 8-8.

8.51. A centered fan of a circle arcs and P radii includes an


angle of 30° a s shown in Fig. 8-26. The pressure on A B
is k. Determine the pressure on AC.
+
Ans. p = k(1 a / 3 ) Fig. 8-26
Linear Viscoeiasticity

9.1 LINEAR VISCOELASTIC BEHAVIOR


Elastic solids and viscous fluids differ widely in their deformational characteristics.
Elastically deformed bodies return to a natural or undeformed state upon remova1 of
applied loads. Viscous fluids, however, possess no tendency a t a11 f o r deformational
recovery. Also, elastic stress is related directly to deformation whereas stress in a viscous
fluid depends (except for the hydrostatic component) upon rate of deformation.
Material behavior which incorporates a blend of both elastic and viscous characteristics
is referred to a s viscoelastic behavior. The elastic (Hookean) solid and viscous (Newtonian)
fluid represent opposite endpoints of a wide spectrum of viscoelastic behavior. Although
viscoelastic materials are temperature sensitive, the discussion which follows is restricted
to isothermal conditions and temperature enters the equations only a s a parameter.

9.2 SIMPLE VISCOELASTIC MODELS


Linear viscoelasticity may be introduced conveniently from a one-dimensional viewpoint
through a discussion of mechanical models which portray the deformational response of
various viscoelastic materials. The mechanical elements of such models are the massless
linear spring with spring constant G, and the viscous dashpot having a viscosity constant 7.
As shown in Fig. 9-1, the force across the spring u is related to its elongation r by
a = Ge (9.1)
and the analogous equation f o r the dashpot is given by
a = qe
where := deldt. The models are given more generality and dimensional effects removed
by referring to u as stress and r a s strain, thereby putting these quantities on a per unit basis.

(a) Linear Spring ( b ) Viscous Dashpot


Fig. 9-1
CHAP. 91 LINEAR VISCOELASTICITY 197

The Maxwell model in viscoelasticity is the combination of a spring and dashpot in series
as shown by Fig. 9-2(a). The Kelvin or Voigt model is the parallel arrangement shown in
Fig. 9-2(b). The stress-strain relation (actually involving rates also) for the Maxwell
model is
-u+ - u = ; '
(9.3)
G 7
and for the Kelvin model is
u = Ge+rl;

These equations are essentially one-dimensional viscoelastic constitutive equations. I t is


helpful to write them in operator form by use of the linear dijferential t i m e operator
d, = a/dt. Thus (9.3) becomes
{a,/G + l/7)u = {a,), (9.5)
and (9.4) becomes
u = {G + qat)€
with the appropriate operators enclosed by parentheses.

(a) Maxwell ( b ) Kelvin


Fig. 9-2

The simple Maxwell and Kelvin models are not adequate to completely represent the
behavior of real materials. More complicated models afford a greater flexibility in por-
traying the response of actual materials. A three-parameter model constructed from two
springs and one dashpot, and known as the standard linear solid is shown in Fig. 9-3(a). A
three-parameter viscous model consisting of two dashpots and one spring is shown in
Fig. 9-3(b). I t should be remarked that from the point of view of the form of their con-
stitutive equations a Maxwell unit in parallel with a spring is analogous to the standard
linear solid of Fig. 9-3(a), and a Maxwell unit in parallel with a dashpot is analogous to the
viscous model of Fig. 9-3(b).
G9 G,

(a) Standard Linear Solid ( b ) Three-parameter Viscous Model


Fig. 9-3

A four-parameter model consisting of two springs and two dashpots may be regarded
as a Maxwell unit in series with a Kelvin unit as illustrated in Fig. 9-4 below. Severa1
equivalent forms of this model exist. The four-parameter model is capable of a11 three of
the basic viscoelastic response patterns. Thus it incorporates "instantaneous elastic re-
198 LINEAR VISCOELASTICITY [CHAP. 9

sponse" because of the free spring G,, "viscous flow" because of the free dashpot v,, and,
finally, "delayed elastic response" from the Kelvin unit.

Fig. 9-4

The stress-strain equation for any of the three or four parameter models is of the
general form
+ +
p2a pl& pou = q2Y+ ql; Q o C + (9.7)
where the p,'s and qi's are coefficients made up of combinations of the G's and ?'s, and depend
upon the specific arrangement of the elements in the model. In operator form, (9.7) is
written
{p2a: + plat + p0}u = {q2a: + qlat + qo)€ (9-8)

9.3 GENERALIZED MODELS. LINEAR DIFFERENTIAL OPERATOR EQUATION


The generalized Kelvin model consists of a sequence of Kelvin units arranged in series
as depicted by Fig. 9-5. The total strain of this model is equal to the sum of the individual
Kelvin unit strains. Thus in operator form the constitutive equation is, by (9.6),

Fig. 9-5

Similarly, a sequence of Maxwell units in parallel as shown in Fig. 9-6 is called a generalized
Maxwell model. Here the total stress is the resultant of the stresses across each unit; and
so from t9.5),

t 0

Fig. 9-6
CHAP. 91 LINEAR VISCOELASTICITY 199

For specific models, (9.9) and (9.10) result in equations of the form
p O u + p l ~ + p 2 a +. . . = qOc+ql;+q2;.+ ...
which may be expressed compactly by

This linear diflerential operator equation may be written symbolically as

{P>u = {&}c (9.13)


where the operators { P ) and {Q) are defined by
m ai
CO p.-2dta'
{P) = i= i=O

9.4 CREEP AND RELAXATION


The two basic experiments of viscoelasticity are the creep and relaxation tests. These
tests may be performed as one-dimensional tension (compression) tests or as simple shear
tests. The creep experiment consists of instantaneously subjecting a viscoelastic specimen
to a stress U , and maintaining the stress constant thereafter while measuring the strain
(creep response) as a function of time. In the relaxation experiment an instantaneous strain
e0 is imposed and maintained on the specimen while measuring the stress (relaxation) as a
function of time. Mathematically, the creep and relaxation loadings are expressed in terms
of the unit step function [ U ( t- t,)],defined by

and shown in Fig. 9-7. 1


For the creep loading, 1
t t 1

0 = ",[u(t)l ( 9 .I 6 ) Fig. 9-7

where [ U ( t ) ]represents the unit step function applied a t time t, = O. The creep response
of a Kelvin material is determined by solving the differential equation

which results from the introduction of (9.16) into (9.4). Here T = rllG is called the retarda-
tion time. For any continuous function of time f ( t ) ,it may be shown that with t' as the
variable of integration,

f (t')[U(tr- t,)]dt' = Stf(t')


[ U ( t- t,)]
ti
dt' (9.18 )

by means of which (9.17) may be integrated to yield the Kelvin creep response
"o
.(t) = -
G
( 1 - e-t17)[U(t)] (9.19)

The creep loading, together with the creep response for the Kelvin and Maxwell models
(materials) is shown in Fig. 9-8 below.
200 LINEAR VISCOELASTICITY [CHAP. 9

1,,

uo -

*
t t
(a) Creep Loading ( b ) Creep Response
Fig. 9-8

The stress relaxation which occurs in a Maxwell material upon application of the strain

is given by the solution of the differential equation

obtained by inserting the time derivative of (9.20) into (9.3). Here [ S ( t ) ] = d [ U ( t ) ] l d t is a


singularity function called the unit i m p u l s e f u n c t i o n , or Dirac delta f u n c t i o n . By definition,

This function is zero everywhere except a t t = t , where i t is said to have a n indeterminate


spike. For a continuous function f ( t ) ,i t may be shown t h a t when t > t,,

with the help of which (9.21) may be integrated to give the Maxwell stress relaxation

The stress relaxation for a Kelvin material is given directly by inserting ;= c o [ S ( t ) ]


into (9.4) to yield
dt) = GcOIU(t)l+ V E~[~(~)I (9.25)

The delta function in (9.25) indicates that it would require a n infinite stress to produce a n
instantaneous finite strain in a Kelvin body.

9.5 CREEP FUNCTION. RELAXATION FUNCTION. HEREDITARY INTEGRALS


The creep response of any material (model) to the creep loading u = u 0 [ U ( t ) ] may be
written in the form
~ ( t =) * ( t ) u , (9.26)
where q ( t ) is known as the c r e e p f z ~ ~ z c t i o n .For example, the creep function f o r the gen-
eralized Kelvin model of Fig. 9-5 is determined from (9.19) to be
Y
@(t) = J , ( 1 - e t"l)[ú'(t)] (9.27)
1-1
CHAP. 91 LINEAR VISCOELASTICITY 201

where Ji = l/Gi is called the compliance. If the number of Kelvin units increases indefinitely
so that N + m in such a way that the finite set of constants ( r , Ji) may be replaced by the
continuous compliance function J(T),the Kelvin creep function becomes

The function J(T)is referred to as the "distribution of retardation times", or retardation


spectrum.
In analogy with the creep response, the stress relaxation for any model subjected to the
strain c = co[U(t)] may be written in the form

where +(t) is called the relaxation function. For the generalized Maxwell model of Fig. 9-6,
the relaxation function is determined from (9.24) as

Here, as N + .o the function G(T) replaces the set of constants (G, T ~ )and the relaxation
function is defined by
+(t) = G(T)~-"'dr Sm (9.31)

The function G(7) is known as the "distribution of relaxation times", or relaxation spectrum.
In linear viscoelasticity, the superposition principle is valid. Thus the total "effect"
of a sum of "causes" is equal to the sum of the "effects" of each of the "causes". Accord-
ingly, if the stepped stress history of Fig. 9-9(a) is applied to a material for which the creep
function is ~k(t),the creep response will be

Therefore the arbitrary stress history u = ~ ( t )of Fig. 9-9(b) may be analyzed a s an
infinity of step loadings, each of magnitude da and the creep response given by the super-
position integral
-m
~ ( t ) = Jt $p
9 ( t - t') dt'

Such integrals are often referred to as hereditary integrals since the strain a t any time is
seen to depend upon the entire stress history.

"o '
"1

tl
I
t2
"3

---------------
I I
I
t3
- v - - - - - -

Fig. 9-9
/e
" + gdt' -- - -----
" ----

I
I
I
I
t'
(a,
t'
I
I
I
I
+ dt' t -
202 LINEAR VISCOELASTICITY [CHAP. 9

For a material initially "dead", i.e. completely free of stress and strain a t time zero, the
lower limit in (9.33) may be replaced by zero and the creep response expressed as

c(t) = ltd"(t3
dt'
qr(t - t') dt'

Furthermore, if the stress loading involves a step discontinuity of magnitude o, a t t = 0,


(9.34) is usually written in the form

Following similar arguments a s above, the stress as a function of time may be rep-
resented through a superposition integral involving the strain history e(t) and the relaxation
function +(t). In analogy with (9.33) the stress is given by

.(t) = J
m g +(t - t') dtf

and with regard to a material that is "dead" a t t = 0, the integrals comparable to (9.34)
and (9.35) are respectively
.(t) = Xt%+(t-t') dt' (9.37)

and

Since either the creep integral (9.34) or the relaxation integral (9.37) may be used to
specify the viscoelastic characteristics of a given material, it follows that some relationship
must exist between the creep function +(t) and the relaxation function +(t). Such a relation-
ship is not easily determined in general, but using the Laplace transform definition

it is possible to show that the transforms G ( s ) and $(t) are related by the equation

where s is the transform parameter.

9.6 COMPLEX MODULI AND COMPLIANCES


If a linearly viscoelastic test specimen is subjected to a one-dimensional (tensile or shear)
stress loading o = U, sin wt, the resulting steady state strain will be c = E, sin (ot - a), a
sinusoidal response of the same frequency but out of phase with the stress by the lag angle
6. The stress and strain for this situation may be presented graphically by the vertical
projections of the constant magnitude vectors rotating a t a constant angular velocity w as
shown in Fig. 9-10 below.
The ratios of the stress and strain amplitudes define the absolute dynamic modulus
ao/%,andthe absolute dynamic compliance C ~ / U , . In addition, the in-phase and out-of-phase
components of the stress and strain rotating vectors of Fig. 9-10(a) are used to define

(a) the storage modulus


CHAP. 91 LINEAR VISCOELASTICITY

o0 sin S
( b ) the loss modulus G, = -
€0

c, cos S
( c ) the storage compliance J, = -
"o
c, sin 6
(d) the loss compliance J, = -
,o

Fig. 9-10

A generalization of the above description of viscoelastic behavior is achieved by


expressing the stress in complex form as
*, ,oeiwt (9.41)
and the resulting strain also in complex form as
= c ei(wt-6)
o
From (9.41) and (9.42) the complex modulus G*(io) is defined as the complex quantity
V*/€* = G*(iw) = (oO/ r o )eis = G, + iG, (9.43)
whose real part is the storage modulus and whose imaginary part is the loss modulus.
Similarly, the complex compliance is defined as
e*/,* = J*(iw) = (Eoluo)e-i6
(94 4 )
= J, - U,
where the real part is the storage compli-
ance and the imaginary part the negative G1
of the loss compliance. In Fig. 9-11 the
vector diagrams of G* and J* are shown.
Note that G* = 1/J*. Fig. 9-11

9.7 THREE DIMENSIONAL THEORY


In developing the three dimensional theory of linear viscoelasticity, it is customary to
consider separately viscoelastic behavior under conditions of so-called pure shear and pure
dilatation. Thus distortional and volumetric effects are prescribed independently, and
subsequently combined to provide a general theory. Mathematically, this is handled by
resolving the stress and strain tensors into their deviatoric and spherical parts, for each of
which viscoelastic constitutive relations are then written. The stress tensor decomposition
is given by (2.70) as
204 LINEAR VISCOELASTICITY [CHAP. 9

and the small strain tensor by (3.98) as


eij = e,, + Sijrkk/3 (9.46)
Using the notation of these equations, the three dimensional generalization of the viscoelastic
constitutive equation (9.13) in differential operator form is written by the combination

and {M)uii = 3{N)eii (9.47b)


where { P ) , {Q), { M ) and { N ) are operators of the form (9.14) and the numerical factors
are inserted for convenience. Since practically a11 materials respond elastically to moderate
hydrostatic loading, the dilatational operators { M ) and { N ) are usually taken as constants
and (9.47) modified to read
{ P > s i j= 2 { Q ) e i j (9.48~)
uii = 3K9, (9.48b)
where K is the elastic bulk modulus.
Following the same general rule of separation for distortional and volumetric behavior,
the three-dimensional viscoelastic constitutive relations in creep integral form are given by

and in the relaxation integral form by

The extension to three-dimensions of the complex modulus formulation of viscoelastic


behavior requires the introduction of the complex bulk modulus K*. Again, writing shear
and dilatatíon equations separately, the appropriate equations are of the form

9.8 VISCOELASTIC STRESS ANALYSIS.


CORRESPONDENCE PRINCIPLE
The stress analysis problem for an iso-
tropic viscoelastic continuum body which
occupies a volume V and has the bounding
surface S as shown in Fig. 9-12, is formulated
a s follows: Let body forces bi be given
throughout V and let the surface tractions
t$'(xk, t ) be prescribed over the portion SI of
S, and the surface displacements gi(xk,t ) be
prescribed over the portion S, of S. Then
the governing field equations take the form
of: Fig. 9-12
CHAP. 91 LINEAR VISCOELASTICITY 205

1. Equations of motion (or of equilibrium)

2. Strain-displacement equations
2fij = (ui,j + uj,i)
or strain-rate-velocity equations
+
2Gij = (v,,~ V,~)
3. Boundary conditions A

uij (x,, t) n, (x,) = tin)(x,, t) on S ,

4. Initial conditions
ui (x,, 0) = UO

5. Constitutive equations
(a) Linear differential operator form (9.48)

(b) Hereditary integral form (9.49) or (9.50)

(c) CompIex modulus form (9.51)

If the body geometry and loading conditions are sufficiently simple, and if the material
behavior may be represented by one of the simpler models, the field equations above may be
integrated directly (see Problem 9.22). For more general conditions, however, it is con-
ventional to seek a solution through the use of the correspondente pr.inciple. Thjs principle
emerges from the analogous form between the governing field equations of elasticity and the
Laplace transforms with respect to time of the basic viscoelastic field equations given above.
A comparison of the pertinent equations for quasi-static isothermal problems is afforded
by the following table in which barred quantities indicate Laplace transforms in accordance
with the definition
f(xk9S ) = xw f ( ~ ,t)e-st
> dt

Elastic Transformed Viscoelastic


1. + bi
a,j3i = O 1. üij,i + -b, = O

From this table it is observed that when G in the elastic equations is replaced by &lP, the
two sets of equations have the same form. Accordingly, if in the solution of the "correspond-
ing elastic problem" G is replaced by &lP for the viscoelastic material involved, the result
206 LINEAR VISCOELASTICITY [CHAP. 9

is the Laplace transform of the viscoelastic solution. Inversion of the transformed solution
yields the viscoelastic solution.
The correspondence principle may also be stated f o r problems other than quasi-static
. problems. Furihermore, the form of the constitutive equations need not be the linear dif-
ferential operator form but may appear a s in (9.49), (9.50) or (9.51). The particular problem
under study will dictate the appropriate form in which the principle should be used.

Solved Problems
VISCOELASTIC MODELS (Sec. 9.1-9.3)
9.1. Verify the stress-strain relations f o r the Maxwell and Kelvin models given by (9.3)
and (9.4) respectively.
In the Maxwell model of Fig. 9-2(a) the total strain is the sum of the strain in the spring plus
+
the strain of the dashpot. Thus e = es ED and also ;= Zs +
GD. Since the stress across each
element is o, ( 9 . 1 ) and (9.2) may be used to give ;= O/G a/?. +
, In the Kelvin model of Fig. 9-2(b), o = os + a~ and directly from (9.1) and (9.2), o = v ; + Ge.
9.2. Use the operator form of the Kelvin model stress-strain relation to obtain the stress-
strain law for the standard linear solid of Fig. 9-3(a).
Here the total strain is the sum of the strain in the spring plus the strain in the Kelvin unit.
Thus E = es +
e~ or in operator form e = olG1 +
o / { G 2 q2at). From this +
and so GIGze + G l s z ; = ( G 1+ Gg)a + 72;.
9.3. Determine the stress-strain equation for the four parameter model of Fig. 9-4. Let
T,I~ +
and compare with the result of Problem 9.2.
Here the total strain e = e~ + e~ which in operator form is
e = o/{Gp + vzat) + {at + l / ~ l ) o / G l { a t )
Expanding the operators and collecting terms gives

As 7 , -+ m this becomes Ü + ( G 1+ G2)&l'lg= G1'E+ G1G2;Iq2 which is equivalent to the result of


Problem 9.2.

9.4. Treating the model in Fig. 9-13 a s a


special case of the generalized Max-
well model, determine its stress-strain
equation. <I - -..O
Writing (9.10) for N = 2 in the form
o + + +
= G l ; / { a t 1 / ~ ~ G) q ; / { a t 1 1 ~ ~ )
and operating as indicated gives Fig. 9-13
CHAP. 91 LINEAR VISCOELASTICITY 207

{at 4- 1/r2)(6 + "/r1) = Gl{at + 1/.r2); + G2{at+ 1/7,);


which when expanded and rearranged becomes
..
0 (71 72);/r1r2 + +
= (G1 G2)7 + (G1/72+ G2/r1);

9.5. The model shown in Fig. 9-14 may be considered as a


degenerate form of the generalized Maxwell model with
- m for the case N = 3. Using these values in
' 7 1 = '12 -
(9.10) develop the stress-strain equation for this model.
Here (9.10) becomes o = vl; + G2;/{at) + ;/{at/G3 + l / q 3 1 or
{dt/G3 + l/v3)8 = {atlG3 + 1 / 7 3 ) ( ~ 1 7 +G2:) + ; 93
'I 1

Application of the operators gives

which may also be written t


713& + G30 = B ~ T ~(G213+
~ Y + G391 G393):+ G2G3e Fig. 9-14

CREEP AND RELAXATION (Sec. 9.4)


9.6. Determine the Kelvin and Maxwell creep response equations by direct integration of
(9.17') and (9.21) respectively.
Using the integrating factor et/7, (9.17) becomes eetlr = -
(9.18) yields

eet/7 = ( ~ ~ [ U ( t ) ] / ~ ) [ r=
e t '(uO/G)(et/7-
/~]~ l)[U(t)]
n. i'
or
et'/7[U(t1)]dt' which by formula

e = (ao/G)(l- e-t/T)[U(t)]

(9.23),
Use of et/' a s the integrating factor in (9.21) gives oet/7 = GeO

aet17 = Geo[U(t)] or a = Geoe-t/~[U(t)]


i' dt'; and by formula
et1/T[8(t')]

9.7. Determine the creep response of the standard linear solid of Fig. 9-3(a).
Since e = es + EK for this model the creep response from (9.1) and (9.19) is simply
e(t) = [ l / G l + (1/G2)( 1 - e - t / 7 ~ ) ] a o [ U ( t ) ]
The same result may be obtained by setting q2 = m in the generalized Kelvin (N = 2) response
2
c = 2
i=l
J i ( l - e - t / r , ) ~ o [ U ( t ) ]or by integrating directly the standard solid stress-strain law. The
student should carry out the details.

9.8. The creep-recovery experiment consists of a A0


creep loading which is maintained for a period
of time and then instantaneously removed.
Determine the creep-recovery response of the 00 . I
I
standard solid (Fig. 9-3(a)) for the loading I
I
shown in Fig. 9-15. 6
272 t
From Problem 9.7 the response while the load is
on ( t < 272) is
e = oo[l/Gl + ( l / G z ) ( l- e - t / 7 ~ ) ] Fig. 9-15
208 LINEAR VISCOELASTICITY [CHAP. 9

A t t = 2r2 the load is removed and o becomes zero a t the same time that the "elasticJJ deformation
uo/G1 is recovered. For t > 272 the response is governed by the equation +
€/r2 = O which is
the stress-strain law for the model with u = O (see Problem 9.2). The solution of this differential
equation is C = C e - T / r ~where C is a constant and !i'
= t - 2 ~ A~ t . T = 0, C = C = oO(l- e-2)/G2
and so
e = aO(l- e-2)e-T1r2/G2 = uo(e2- l ) e - t / r z/G 2 for t > 2 r 2

9.9. The special model shown in Fig. 9-16 is elongated a t a constant rate ;= ~ , / t , as
indicated in Fig. 9-17. Determine the stress in the model under this straining.

Fig. 9-16 Fig. 9-17

From Problem 9.5 the stress-strain law for the model is + +


017 = q Z + 3G; GE/T and so
+ +
here O U / T = 3Gso/tl Ge0t/7tl. Integrating this + +
yields o = a0(39 G t - ?) Ce-t/r where C
is the constant of integration. When t = 0, o = q ~ o / t land so C = -l ) ~ O l t lThus
. o =
+
eo(2q G t - qe-tlr)/tl. Note that the same result is obtained by integrating

i r+2
3 ~ ~ t 'dt'
tl
Xt dtf

9.10. Determine by a direct integration of the stress-strain law for the standard linear
solid its stress relaxation under the strain E = e , [ U ( t ) ] .
+ + +
Writing the stress-strain law (see Problem 9.2) a s 6 (G1 G2)uIq2= ~ ~ G ~ ( [ 8 (G1G2[U(t)]lT2)
t)]
for the case a t hand and employing the integrating factor e(Gi+Gz)tlnz i t is seen that

Integrating this equation with the help of (9.18) and (9.23),


u +
= eOGl(G2 Gle-(Gi+Gz)tlnz)[U(t)]/(G1 + G2)

CREEP AND RELAXATION FUNCTIONS. HEREDITARY INTEGRALS (Sec. 9.5)


9.11. Determine the relaxation function +(t)for the
three parameter model shown in Fig. 9-18.
The stress-strain relation for this model i s -o-- -u
6 + u/r2 = (G1+ G2); + G1G2c/q2
and so with = eo[U(t)] and ;= eo[6(t)] use of the
integrating factor etlrz gives Fig. 9-18

+
Thus by use of (9.18) and (9.23), o = eo(Gl G2e-t/rz) = eo$(t). Note that this result may also be
obtained by putting q1 + m in (9.30) for the generalized Maxwell model.
CHAP. 91 LINEAR VISCOELA'STICITY 209

9.12. Using the relaxation function +(t)for the model of Problem 9.11, determine the creep
function by means of (9.40).
+
The Laplace transform of $ ( t ) = G1 G2e-t/Tr is 5 ( s ) = G l / s + G z / ( s+ 1 1 ~ (see
~ ) any stand-
ard table of Laplace transforms). Thus from (9.40),

which may be inverted easily by a Laplace transform table to give


li/ = +
l / G l - [G2/G1(G1 G2)]e-Git/(Gi+G2)T8

This result may be readily verified by integration of the model's stress-strain equation under creep
loading.

9.13. If a ramp type stress followed by a sus-


tained constant stress U , (Fig. 9-19) is
applied to a Kelvin material, determine
the resulting strain. Assume ~ , / t=, A.
The stress may be expressed a s
t- I. t
u = h t [ U ( t ) ]- h ( t - t l ) [ U ( t- t,)]
Fig. 9-19
which when introduced into (9.4) leads to

Integrating with the aid of (9.18) gives


E ~ ( e --
= ( h l G ) ( ( ti- ~ l l) )~[ U ( t ) ]- ( ( t- t l ) + ~ ( e ( t i - t ) / ~l )-) [ U ( -t t l ) ] )
which reduces a s t -+ m to E = x ~ , I G= u,/G.

9.14. Using the creep integral (9.34) together with the Kelvin creep function, verify the
result of Problem 9.13.
For the Kelvin body, $ ( t )= (1- e-tI7)/G and (9.34) becomes

.(t) = S-tm$ ([U(tl)] + t f [ s ( t l )-] [u(tf- t l ) ]- ( t f- t l ) [ s ( t-f t l ) ] ) (-l e - ( t - t l ) / T )dt'


which by (9.18) and (9.23) reduces to

=
X
-
G o
St
[ [ ~ ( t ) ] ( 1 - e - ( t - t l ) l ~dt'
) - [ U ( t- tl)]
tl
t
$
(1 - e - ( t - t f ) l r ) d t f
I
A straightforward evaluation of these integrals confirms the result presented in Problem 9.13.

9.15. By a direct application of the superposition principle, determine the response of a


Kelvin material to the stress loading shown in Fig. 9-20.

Fig. 9-20 Fig. 9-21


210 LINEAR VISCOELASTICITY [CHAP. 9

The stress may be represented a s a sequence of ramp loadings a s shown in Fig. 9-21 above.
+
From Problem 9.13, E = ( h / G ) [ t r(e-tlf - I ) ][ U ( t ) ] for this stress loading. In the présent case
therefore
+
) ( i / G ) [ ( t r(e-tlT - l ) ) [ U ( t ) -
~(t= ] ( ( t- t i ) + r ( e - ( t - t l ) l ~ - l ) ) [ U ( t- ti)]
-((t t 2 t l ) ] + ( ( t- 3tl) + ~(e-(t-3t1)I7- l ) ) [ U ( -
- 2tl) + ~ ( e - ( t - z t ~-) /l ~) ) [ U ( - t 3tl)]l
Note that a s t + m, E + 0.

COMPLEX MODULI AND COMPLIANCES (Sec. 9.6)


9.16. Determine the complex modulus G* and the lag angle S for the Maxwell material of
Fig. 9-2.
Writing (9.3) a s 6 + U/T= G ; and inserting (9.41) and (9.42) gives iwooeiwt4- o0eiot/r =
Giw~,ei(wt-6) from which ~ ~ e= G* i ~= G
l i w~ ~ l~( l +i w ~ ) , or in standard form

From Fig. 9-11, tan 8 = G d G 1 = Gor/Gw2r2 = 1 1 ~ ~ .

9.17. Show that the result of Problem 9.16 may also be obtained by simply replacing the
operator d, by .i in equation (9.5)and defining u / e = G*.
After the suggested substitution (9.5) becomes (iolG + l / q ) o = &e from which
U / E = Giol(io + 117) = G ~ w T+ /(~ ~ W T )
a s before.

9.18. Use equation (9.10) for the generalized Maxwell model to illustrate the rule that
"for models in parallel, the complex moduli add".
From Problem 9.17 the complex modulus for the Maxwell model may be written G* = a/€ =
Giwrl(l+ iwr). , Thus writing (9.10) a s

the generalized Maxwell complex modulus becomes

9.19. Verify the relationship Jl = l / G l ( l+ tan28) between the storage modulus and
compliance.
From (9.43) and (9.44), J* = 1/G* and so J1 - iJ2 = l / ( G l iG2) = (G1- ~G,)I(G? 62). + +
Thus
+ +
J1 = G ~ / ( G : 02) = l / G 1 ( l (G2/G1)2) = l / G 1 ( l tan2 6) +
9.20. Show that the energy dissipated per cycle is related directly to the loss compliance
J2 by evaluating the integral
S u d over
~ one cycle.
I

For the stress and strain vectors of Fig. 9-10, the integral
S odr evaluated over one cycle is
ds
dt = l 2 ~ 1 0
(oo sin U cos (ut - 6 ) d t
~ ) E ~ W

- ooEo. J2'I* sin w t (cos u t cos a + sin u t sin a ) d t


o
2nlo
sin 2ot
dt + J, lbiu (sin2 o t ) d t ] = o:sJ2
CHAP. 91 LINEAR VISCOELASTICITY 211

THREE DIMENSIONAL THEORY. VISCOELASTIC


STRESS ANALYSIS (Sec. 9.7-9.8)
9.21. Combine ( 9 . 4 8 ~ )and ( 9 . 4 8 3 ) to obtain the viscoelastic constitutive relation uij =
+
a i j { R } e k k {S)cij and determine the form of the operators { R } and { S I .
Writing ( 9 . 4 8 ~a) s {P)(aii- oijakk/3)= 2{Q)(eij- oijekk/3) and replacing ukk here by the right hand
side of (9.48b), the result after some simple manipulations is

aij oij{(3KP- 2Q)/3p)ekk + {2Q/P)eij

9.22. A bar made of Kelvin material is pulled in tension so that a,, = a o [ U ( t ) ] , a,, = a33 =
-
ai, - a,, = a,, = O where ao is constant. Determine the strain ell for this loading.
From (9.48b), 3cii = o o [ U ( t ) ] l Kfor this case; and from ( 9 . 4 8 ~with
) i = j = 1, { P ) ( u l l- a11/3)=
+
{ 2 Q ) ( ~ l l -eii/Q). But from (9.61, { P ) 1 and { Q ) = {G Tat) for a Kelvin material; so that now
200[U(t)I/3 = 2{G +~ a ~- aO[U(t)]lgK)
l ( ~ ~ ~

Solving this differential equation yields


€11 = aO(3K + + aoe-tlT[U(t)]/9K
e-t1T)[ U ( t ) ] l 9 K G
As t + m, €11 -> (3K + G)aO19KG= a J E .

9.23. A block of Kelvin material is held in a container


with rigid .walls so that e, = e,, = O when the
stress o,, = - a o [ U ( t ) ] is applied. Determine c,,
and the retaining stress components V,, and u3,
for this situation.
Here qi = ell and a,, = so that (9.48b) becomes
ai1 i-2u2, = 3Kell and (9.48a) gives 2(u11- u2,)/3 =
2G{1+ ~ a ~ ) ( 2 ~for
~ ~a/ Kelvin
3) body. Combining these
relations yields the differential equation Fig. 9-22

which upon integration gives

Inserting this result into ( 9 . 4 8 ~ for


) i = j = 2 gives
a,, +
= (+ao/2 - 9Kao(l - e-(4G+3K)t/4G~)/(8GG K ) ) [ U ( t ) ]

9.24. The radial stress component in an elastic half-


space under a concentrated load a t the origin
may be expressed as
a,,,, ) - 2 v ) a ( r , 2 ) - P(r,41
= ( P 1 2 ~[(I
where a and p are known functions. Determine
the radial stress for a Kelvin viscoelastic half-
space by means of the correspondence principle
when P = Po[U(t)]. Fig. 9-23
212 LINEAR VISCOELASTICITY [CHAP. 9

The viscoelastic operator for the term ( 1 - 2 ~ is) { 3 Q ) l { 3 K P + Q ) so that for a Kelvin body
the transformed viscoelastic solution becomes

which may be inverted with help of partia1 fractions and transform tables to give the viscoelastic
stress

9.25. The correspondence principie may be used to obtdin displacements a s well as stresses.
The x displacement of the surface of the elastic half space in Problem 9.24 is given by
W ( z = ~=) P(l- v 2 ) l E T r . Determine the viscoelastic displacement of the surface for
the viscoelastic material of that problem.
The viscoelastic operator corresponding to ( 1 - v2)lE is { 3 K + 4Q)/4Q(3K + Q ) which for the
Kelvin body causes the transformed displacement to be

After considerable manipulation and inverting, the result is

Note that when t = O, w ( , = o ) = O and when t + m, ) P o ( l - u z ) / E ~ r ,the elastic deflection.


w ( , = ~ -+

9.26. A simply supported uniformly loaded beam is


assumed to be made of a Maxwell material. ,
Determine the bending stress a , , and the deflec-
tion w(x,, t) if the load is p = p o [ U ( t ) ] . Fig. 9-24

The bending stress for a simply supported elastic beam does not depend upon material properties,
so the elastic and viscoelastic bending stress here are the same. The elastic deflection of the beam
is w ( x l ) = pon(xI)/24EZ where a ( x J is a known function. For a Maxwell body, {P) = {a, 117) +
and { Q ) = {Ga,), so that the transformed deflection is

which when inverted gives

When t = O, w ( x , , O ) = pw(x1)/24EZ, the elastic deflection.

9.27. Show that as t += .o the stress a,, in Problem 9.23 approaches ao (material behaves as
a fluid) if the material is considered incompressible ( V = 112).
From Problem 9.23,
0221t+m = -oO(9K - (4G + 3K))/2(4G+ 3 K ) = -oo(3K - 2G)l(3K + 4G)
which may be written in terms of v as 02,1t,, = -voo/(l - v). Thus for v 112, o,,lt+, = -o*
CHAP. 91 LINEAR VISCOELASTICITY 213

MISCELLANEOUS PROBLEMS
9.28. Determine the constitutive relation for the 81
Kelvin-Maxwell type model shown in Fig. .
9-25 and deduce from the result the Kelvin
and Maxwell stress-strain laws. c<t-

Here
a = UM + aK = ;/{dt/G1 + l / v l } + {Gz + q2at}r 82

which upon application of the time operators


becomes
I I
6 + a/rl = q2T + (G1 G2 q 2 / ~ 1 ) ( G 2 / ~ 1 ) ~ Fig. 9-25

In this equation if q2 = 0 (spring in parallel with Maxwell), g d r l = (Gl G2); ( G 2 1 ~ 1 ) ~ . + + +


Further, if G, = 0, the Maxwell law 6 U / T= +
~ G1; results. Likewise, if G2 is taken zero first
(dashpot in parallel with Maxwell), 6 a/rl = q2Y+ (Gl i- +
q2/r1);; and when q , = 0, this also
reduces to the Maxwell law.
If the four-parameter constitutive relation is rewritten
1118 + Glu +
= 7118~Y ( G l ~ l +G281+ G172); + GlG2e
and q1 set equal to zero, the result is the Kelvin law a = 8,; Gze. Likewise, if Gl = O the +
reduced equation is 8 = q,T+ G,;, again representing the Kelvin model.

9.29. Use the superposition principie to obtain


the creep recovery response for the stand-
ard linear solid of Fig. 9-3(a) and compare
the result with that obtained in Problem 9.8.
With the stress loading expressed by

U = ao[U(t)]- ao[U(t- 2r2)]


(see Fig. 9-26), the strain may be written a t once
from the result of Problem 9.7 a s Fig. 9-26

E = ao(l/Gl + ( 1 - e - t ' ~ ~ ) / G ~ ) [ U (-t ) laO(l/G1-(i1- e-(t - z ~ z ) ~ ~ ) l G ~-) 2r2)]


[U(t

A t times t > 272 both step functions equal unity and


E
-
- a0(-e-t/7, + e-(t - 2 7 2 ) / ~ ~ ) / =G ~ aO(e2- 1)e-t/r2/G2
which agrees with the result in Problem 9.8.

9.30. Determine the stress in the model of Prob-


lem 9.9 when subjected to the strain history
shown in Fig. 9-27. Show that eventually
the "free" spring in the model carries the
entire stress.
I
From Problem 9.9 and the superposition prin-
ciple the stress is Fig. 9-27
-
- ~ 0 ( 7 (-
2 e-t/7) + G t ) [ U ( t ) ] l t -l eo(7(2- e-(t-ti)/7) + G ( t - t l ) ) [ U (-t t l ) ] / t l
For times t > t 1 the stress is o = cotl(etl/T- 1)e-t/r/tl f Gco, and as t + m this reduces to
o = Gço.
LINEAR VISCOELASTICITY [CHAP. 9

9.31. The "logarithmic retardation spectrum" L is defined in terms of the retardation spec-
trum J by L(ln r ) = r J ( r ) . From this definition determine the creep function +(t)in
terms of L(ln r ) .
Let ln r = h so that eX = T and thus drldh = "e T, or dr = ~ d ( l Tn) . From this, (9.28)
defining g ( t ) becomes q ( t ) =
LW L(ln ~ ) ( le-t")d(ln
- r). In the same way, if
defines the logarithmic relaxation spectrum, $ ( t ) in (9.81) may be written
H(ln r ) = TG(T)

9.32. For the Maxwell model of Fig. 9-2(a) deter-


mine the storage and loss moduli, Gl and
G,, as functions of ln o r and sketch the
shape of these functions.
From Problem 9.16,
+
G* = G ( w ~ TL~r ) / ( 1 + ~ 2 ~ 2 )

for a Maxwell material. Thus


G1 = Gw2~2/(l+~ 2 ~ =
2 ) Ge2X/(l+ e2h)
where 1 = l n w ~ . For A = O, G1 = G/2; for
h = m, G1 = G ; and for h = -a, Gl = O. Like-
wise G, = GeX/(l+ e2X) and for A = O, G, = G/2;
for h = I m , G2 = O. The shape of the curves for
these functions is a s shown in Fig. 9-28. Fig. 9-28

9.33. Determine the viscoelastic operator form of the elastic constant V (Poisson's ratio)
using the constitutive relations (9.48).
Under a uniaxial tension o,, = o,, (9.48b) gives si/3 = o d 9 K so that ( 9 . 4 8 ~ )for i = j = 1
+
yields cll = ( 3 K P Q)ooI{9KQ). In the same way ( 9 . 4 8 ~ )for i = j = 2 yields cz2 =
(2Q - 3PK)oo/{18KQ). Thus in operator form, V = -c2,/c11 = (3PK - 2Q)/{6KP 2Q). +

9.34. A cylindrical viscoelastic body is inserted into a


rigid snug-fitting container (Fig. 9-29) so that
= O (no radial strain). The body is elastic in
dilatation and has the creep function +, = A
+
Bt CeM where A, B, C, A are constants. If
+ T h
;33 = ;OIU(t)],determine a,,(t).

Here oii = 3Keii and by the symmetry of the problem,


2all + o33 = 3Ke3,. Also from ( 9 . 5 0 ~ )with i = j = 1,

011 - 413 = -it


2 - h(t t') dt'. Soiving for o,, from
these two relations we obtain Fig. 9-29
osr = KE,, +3 it2 +,(t - t f )dt'

The relaxation function +, may be found with the help of (9.40). The result is

= [(rl- h)e+it - (r,- h)eTzt]/(rl- r,)


$,

where r,,, = [ A h - B * ~ ( A+AB)' + 4BCh ] / 2 ( A+ C). Thus finally


CHAP. 91 LINEAR VISCOELASTICITY 215

t - ~ ) ~ r l ( t - t '-
) - h)eVz(t-tl)
rs3 = Kbt + i.
(r1 - ~
2
2 )
I IU(tl)]dt'
which upon integration gives
033 + +
= (K;Ot 2Z0[-(r1- X ) ( I - e~it)/3r, (r, - X ) ( I - erzt)/3r2]/(rl
-r , ) ) [ ~ ( t ) ]

9.35. The "creep buckling" of a viscoelastic column may be analyzed within the linear
theory through the correspondence principle. Determine the deflection w(x,, t ) of a
Kelvin pinned-end column by this method.

Fig. 9-30

The elastic column formula is d2wldx: +


Pow/EZ = 0, and for a Kelvin material E may be
+
replaced by the operator {E 78,) so that for the viscoelastic column { E q a t ) ( d 2 ~ l d ~ : ) + +
P,wlZ = O. Assuming the deflection in a product form w(xl, t) = W(xl)e(t), the operator leads to
the differential equation
(Ee + qè)(d2wldx;)+ PoWeIZ = O from which + [l+ P, W / ~ ~ ( d ~ ~ / d z =; ) O] e l ~
where T = 7IE. But the elastic buckling load is P, = - ~ Z ( d 2 ~ l d x : ) land
~ so +
(1 - P,lP,)el~ = O which integrates easily to yield e = e ( p ~ / ~ ~ - l )Finally
t / ~ . then the "creep
buckling" deflection is w = We(Po/p~-l)t/T.

9.36. Formulate the steady-state vibration problem for a viscoelastic beam assuming' the
constitutive relations are those given by (9.48).
,
The free vibrations of an elastic beam are governed by the equation E Z ( ~ ~ W / ~ X ; ) +
+
p A ( a 2 ~ l a t 2=) O. From (9.48) the viscoelastic operator for E is {9KQl(3KP Q ) ) ,and if the deflec-
tion w ( x l ,t ) = W ( x l ) e ( t ) the resulting viscoelastic differential equation may be split into the space
+
equation d 4 W l d x ~- k4W = O and the time equation (3KP Q)(d2eldt2) (k4ZlpA){9KQ)(~) + = 0.
The solution W i of the space equation represents the ith mode shape, and from the time equation for
N
k = ki the solution ei = 2 Aijehilt where N depends upon the degree of the operator. The total
j=1 m N
solution therefore is w ( x l ,t ) = Wi(xl)AijeAijt in which the Aij are complex.
i=1 j=l

Supplementary Problems
9.37. Determine the constitutive equation for the four parameter mode1 shown in Fig. 9-31.
Ans. 8 + (Gilv2 + G2/v2+ Gl/7l); + ( G l G z / ~ l ~ z=) u G l f . + (G1G2/72)'

Fig. 9-31
216 LINEAR VISCOELASTICITY [CHAP. 9

9.38. Determine the creep response of the standard linear solid by direct integration of E +€ 1 =~ ~
+ +
oo[U(t)](Gl G2)/G1v2 uo[8(t)]/Gl. (See Problem 9.7.)

9.39. Deduce the Kelvin and Maxwell stress-strain laws from the results established in Problem 9.5 for
the four parameter model of that problem. (Hint. Let G3 = 0 , etc.)

9.40. Use equation (9.40) to obtain $ ( t ) if +(t) = a(b/t)m


with m < 1. (Hint. Take m = 1 - k ; then +(t) =

abmtk-1.) A m . $(t) =

9.41. Determine the creep and relaxation functions for the


model shown in Fig. 9-32.
Ans. $ ( t ) = 1/G2 - Gle-Czt/(Gl+c2)71/G2(G1 G2) + L"""d '11

+
+(t) = G2 Gle-t/Tl Fig. 9-32

9.42. Determine G* for the model shown in Fig. 9-33.


G1(l 7220,)+ G2a272 + o(G2r2 q3(1 7 ; a2)) + +
Ans. G* =
1 027; + i +
1 ~ ~ 7 2 +

Fig. 9-33 Fig. 9-34

9.43. In the model of Problem 9.42 let G1 = G, = G and q, = q3 = q and determine the stress history
of the resulting model when it is subjeeted to the strain sequence shown in Fig. 9-34.
€0
Ans. u = - (G(2t- t,)
tl
+ ~ ( -4( 1 + eti'r)e-tlT)) for t l < t < 2tl

9.44. - A viscoelastic block having the constitutive equation 8 ao = P ; + +


ya where a,P, y are con-
stants is loaded under conditions such that o,, = -oo[U(t)], 9,= O, q3= O (see Fig. 9-35).
Assuming oii = 3Kqi, determine ~ ~ ~ ~( t~ ) ~, and
( 0~ )~ ~ ( 0 0 ) .

Ans. o33 = 3aK - 2 y


2(3aK y)h + + ( 3 K- 2P
( 23K + P) - 2(3aK
3aK -+2y)hy ) e - ~ t ] where h = (3aK + y)/(3K + p).

wo sin (7x11)
Po[U(t)l P,[U(t)l

Fig. 9-35 Fig. 9-36

9.45. A pinned-end viscoelastic column is a Maxwell material for which 8 olr = E;. The initial +
shape of the column is w = w o sin ( ~ z l l when ) the load Po [ U ( t ) ]is applied (see Fig. 9-36). Deter-
mine the subsequent deflection w ( x l , t ) a s a function of P,, the elastic buckling load.
Ans. w ( x l , t ) = w0 sin ( a ~ ~ / l ) e - t ~ ( ' - P ~ / P o ) ~
INDEX
Absolute dynamic, Conjugate dyadic, 4
compliance, 202 Conservation of,
modulus, 202 energy, 128
Acceleration, 111 . m a s , 126
Addition and subtraction, Constitutive equations, 132
of matrices, 17 Continuity equation, 126
of (Cartesian) tensors, 1 5 Continuum concept, 44
of vectors, 2 Contraction, 1 6
Adiabatic process, 140 Contravariant tensor, 11
Airy stress function, 147 Convective,
Almansi s t r a i n tensor, 82 derivative, 110
Angle change, 86 r a t e of change, 111
Angular momentum, 128 Conventional stress and strain, 175
Anisotropy, 44, 141 Coordinate transformation, 11
Antisymmetric, Correspondence principle, 205
dyadic, 5 Couple-stress vector, 45
matrix, 19 Coupled h e a t equation, 150
tensor, 19 Covariant tensor, 1 2
Axis of elastic symmetry, 142 Creep,
function, 200
Barotropic change, 161 test, 199
Base vectors, 6 Cross product, 3, 5, 1 6
Basis, 6, 9 Cubical dilatation, 90
orthonormal, 6 Curvi1ine:ir coordinates, 7
Bauschinger effect, 176 Cylindrical coordinates, 8
Beltrami-Michell equations, 144
Bernoulli equation, 164 Dashpot, 196
Biharmonic equation, 148 Decomposition,
Body forces, 45 polar, 87
Boundary conditions, 144, 146 velocity gradient, 112
Bulk, Deformation, 77
modulus, 143 gradients, 80
viscosity, 161 inelastic, 175
plane, 91
Caloric equation of state, 132 plastic, 175
Cartesian, tensors, 81
coordinates, 6 total (deformation) theory, 183
tensors, 1, 12, 1 3 De1 operator, 22
Cauchy, Density, 44, 126
deformation tensor, 8 1 entropy, 130
strain ellipsoids, 89 strain energy, 141
stress principle, 45 Derivative,
stress quadric, 50 material, 110, 114
Cauchy-Riemann conditions, 165 of tensors, 22
Circulation, 164 of vectors, 22
Kelvin's theorem of, 165 Deviatoric,
Clausius-Duhem inequality, 131 s t r a i n tensor, 91
Column matrix, 17 stress tensor, 57
Compatibility equations, 92, 114 Diagonal matrix, 17
Complex, Dilatation, 90
niodulus, 202 Direction cosines, 7
potential, 166 Displacement, 78, 83
Compliance, 201 gradient, 80
Coniponent, 1, 7 relative, 83
Compressible fluid, 165 Dissipation function, 132
Configuration, 77 Dissipative stress tensor, 131
Conformable niatrices, 17 Distortion energy theory, 178
INDEX

Divergence theorem (of Gauss), 23 Forces,


Dot product of, body, 45
dyads, 5 surface, 45
vectors, 3 Fourier heat law, 149
Duhamel-Neumann relations, 149 Fundamental metric tensor, 1 3
Dyadics, 1, 4
antisymmetric, 4 Gas,
conjugate of, 4 dynamical equation, 165
symmetric, 4 law, 160
Dyads, 4 Gauss's theorem, 23
nonion form, 7 Generalized,
Dynamic moduli, 203 Booke's law, 140
Kelvin model, 198
Effective, Maxwell model, 198
plastic strain increment, 182 plane strain, 147
stress, 182 plane stress, 147
Elastic, Gibb's notation, 2
constants, 141 Gradient,
limit, 175 deformation, 80
symmetry, 142 displacement, 80
Elasticity, 140 Green's
Elastodynamics, 143 deformation tensor, 81
Elastoplastic problems, 183 finite strain tensor, 82
Elastostatics, 143
Energy, Hamilton-Cayley equation, 21
kinetic, 129 Hardening,
strain, 141 isotropic, 180 ,

thermal, 129 kinematic, 181


Engineering stress and strain, 175 strain, 176, 183
Entropy, 130 work, 176, 183
specific, 130 ' Harmonic functions, 166
e - o identity, 39 Heat,
Equations of conduction law, 149
equilibrium, 48, 128 flux, 129
motion, 128 radiant, 129
state, 130 Hencky equations, 183
Equivalent, Hereditary integrals, 201
plastic strain increment, 182 Homogeneous,
stress, 182 deformation, 95
Euclidean space, 11 material, 44
Eulerian, Hookean solid, 196
coordinates, 78 Hydrostatic pressure, 160
description, 79 Hydrostatics, 163
finite strain tensor, 82 Hyperelastic material, 149
linear strain tensor, 8 3 Hypoelastic material, 149
Hysteresis, 176
Field equations,
Ideal,
elastic, 143, 147
gas, 160
viscoelastic, 205
materiais, 132
Finite strain tensor, 81, 82 Idealized stress-strain curves, 177
First law of thermodynamics, 129 Idemfactor, 5
Flow, 77, 110 Identity matrix, 18
creeping, 169 Incompressible flow, 126
irrotational, 164 Incremental theories, 181
plastic, 175 Indeterminate vector product, 4
potential, 175 Indices, 9
rule, 181 Indicial notation, 8
steady, 163 Inelastic deformation, 175
Fluid, Inertia forces, 48
inviscid, 160 Infinitesimal strain, 83
perfect, 160,164 Initial conditions, 145
pressure, 160 Inner product, 16
INDEX

Integral theorems, 23 Momentum principie, 127


Interna1 energy, 129 Motion, 110
Invariants, 21 steady, 112
of r a t e of deformation, 114 Multiplication of,
of strain, 89, 90 matrices, 1 8
of stress, 51 tensors, 15
Irreversible process, 130 vectors, 3
Irrotational flow, 113, 164
Isothermal process, 140 Navier-Cauchy equations, 144
Isotropy, 44, 142 Navier-Stokes equations, 162
Navier-Stokes-Duhem equations, 162
Jacobian, 11, 79 Newtonian viscous fluid, 161, 196
Johnson's apparent elastic-limit, 176 Normal stress components, 47

Kelvin (material) model, 197 Octahedral,


Kinematic, plane, 59
hardening, 181 shear stress, 192
viscosity, 163 Orthogonal,
Kinetic energy, 129 tensor, 87
Kronecker delta, 1 3 transformation, 1 3
Othogonality conditions, 13, 14
Lagrangian, Orthotropic, 142
description, 79 Outer product, 1 5
finite strain tensor, 82
infinitesimal strain tensor, 83 Parallelogram law of addition, 2
Lamé constants, 143 Particle, 77
Laplace, P a t h lines, 112
equation, 165 Perfectly plastic, 176
transform, 202, 205 Permutation symbol, 16
Left stretch tensor, 87 r-plane, 179
Levy-Mises equations, 181 Plane,
Line integrais, 23 deformation, 91
Linear, elasticity, 145
momentum, 127 strain, 91, 145
rotation tensor, 83 stress, 56-57, 145
therinoelasticity, 149 Plastic,
vector operator, 8 deformation, 175
viscoelasticity, 196 flow, 175
Local rate of change, 111 potential theory, 181
Logarithmic strain, 175 range, 176
strain increment, 182
Mass, 126 Point, 77
Material, Poisson's ratio, 143
coordinates, 78 Polar,
derivative, 110, 114-117 decomposition, 87
description of motion, 79 equilibrium equations, 148
Matrices, 17-19 Position vector, 77
Maximum,
Post-yield behavior, 180
normal stress, 52
Potential,
shearing stress, 52, 53
Maxwell (material) model, 197 flow, 165
plastic, 182
Metric tensor, 1 3
Mises yield condition, 178 Prandtl-Reuss equations, 181
Modulus, Pressure,
bulk, 143 fluid, 160
complex, 203 function, 163
loss, 203 Principal,
shear, 143 axes, 20
storage, 202 strain values, 89
Mohr's circles, stress values, 51, 58
f o r strain, 91 Proper transformation, 17
for stress, 54-56 Proportiona1 limit, 177
Moment of momentum, 128 Pure shear, 178
INDEX

Quadric of, Stream function, 165


strain, 88, 89 Stream lines, 112
stress, 50 Stress,
Quasistatic viscoelastic problem, 205 components, 47
conservative, 131
Range convention, 8 deviator, 57
Rate of deformation tensor, 112 effective, 182
Rate of rotation vector, 114 ellipsoid, 52
Reciproca] basis, 28 function, 147
Rectangular Cartesian coordinates, 6 invariants, 51
Kelative displacement, 83 Mohr's circles for, 54-56
Relaxation, normal, 47
function, 201 plane, 56
spectrum, 201 power, 130
test, 199 principie, 51
Retardation, quadric, 50
spectrum, 201 shear, 47, 52
time, 199 spherical, 57
Reversible process, 130 symmetry, 48
Reynold's transport theorem, 123 tensor, 46
Right stretch tensor, 87 transformation laws, 49
Rigid displacement, 82 vector, 45
Rotation tensor, Stretch,
finite, 87 ratio, 86
infinitesimal, 83, 84 tensor, 87
Rotation vector, 83 Summation convention, 8, 10
Superposition theorem, 145
St. Venant's principle, 145 Surface forces, 45
Scalar, 1
Symbolic notation, 2, 10
field, 22
of a dyadic, 4 Symmetry,
triple product, 4 elastic, 142
Second law of thermodynamics, 130 tensor, 19
Shear,
modulus, 143 Temperature, 130
strain components, 86 Tension test, 175
stress components, 52 Tensor,
Slip line theory, 184 Cartesian, 1, 12, 1 3
Small deformation theory, 82 components of, 1
Spatial coordinates, 78 contravariant, 11
Specific, covariant, 12 ,
entro&, 130 deformation, 81
heat, 149 derivative of, 22
Spherical, fields, 22
coordinates, 8 general, 1, 11
tensor, 57, 91 metric, 12
Spin tensor, 112 multiplication, 1 5
Standard linear solid, 197 powers of, 21
State of stress, 46 rank, 1
Stokes' condition, 161 stretch, 87
Stokes' theorem, 23 transformation laws, 1
Stokesian fluid, 161 Tetrahedron of stress, 47
Strain,
Thermal equation of state, 132
deviator, 91
ellipsoid, 88 Thermodynamic process, 130
energy, 141 Thermoelasticity, 133, 149
hardening, 176 Transformation,
natural, 112, 175 laws, 13, 88
plane, 91 of tensors, 1
rate, 112 orthogonal, 13, 14
shearing, 86 Tresca yield condition, 178
spherical, 91 Triangle rule, 2
transformation laws, 88 Triple vector product, 4
INDEX 221

Tw-o-dimensional elastostatics, Velocity, 110, 111


in polar form, 148 complex, 166
in rectangular form, 145 potential, 164
strain, 112
Uncoupled thermoelasticity, 133, 150 Viscoelastic stress analysis, 204-206
Uniqueness theorem, 145, 158 Viscoelasticity, 196
Unit, Viscous stress tensor, 160
dyadic, 5 Voigt model, 197
relative displacements, 83 Vorticity,
triads, 7 tensor, 112
vector, 3 vector, 113

Work,
Vector, hardening, 176
addition, 2 plastic, 182
displacement, 78
dual, 16 Yield,
field, 22 condition, 177
of a dyadic, 4 curve, 179
position, 12, 77 surface, 179
potential, 126 Young's modulus, 143
products, 3, 4, 16
rotation, 83 Zero,
traction, 46 matrix, 17
transformation law, 14 order tensor, 1
vorticity, 113 vector, 2